diff --git a/Lib/compileall.py b/Lib/compileall.py
new file mode 100644
index 000000000..1c9ceb693
--- /dev/null
+++ b/Lib/compileall.py
@@ -0,0 +1,295 @@
+"""Module/script to byte-compile all .py files to .pyc files.
+
+When called as a script with arguments, this compiles the directories
+given as arguments recursively; the -l option prevents it from
+recursing into directories.
+
+Without arguments, if compiles all modules on sys.path, without
+recursing into subdirectories.  (Even though it should do so for
+packages -- for now, you'll have to deal with packages separately.)
+
+See module py_compile for details of the actual byte-compilation.
+"""
+import os
+import sys
+import importlib.util
+import py_compile
+import struct
+
+try:
+    from concurrent.futures import ProcessPoolExecutor
+except ImportError:
+    ProcessPoolExecutor = None
+from functools import partial
+
+__all__ = ["compile_dir","compile_file","compile_path"]
+
+def _walk_dir(dir, ddir=None, maxlevels=10, quiet=0):
+    if quiet < 2 and isinstance(dir, os.PathLike):
+        dir = os.fspath(dir)
+    if not quiet:
+        print('Listing {!r}...'.format(dir))
+    try:
+        names = os.listdir(dir)
+    except OSError:
+        if quiet < 2:
+            print("Can't list {!r}".format(dir))
+        names = []
+    names.sort()
+    for name in names:
+        if name == '__pycache__':
+            continue
+        fullname = os.path.join(dir, name)
+        if ddir is not None:
+            dfile = os.path.join(ddir, name)
+        else:
+            dfile = None
+        if not os.path.isdir(fullname):
+            yield fullname
+        elif (maxlevels > 0 and name != os.curdir and name != os.pardir and
+              os.path.isdir(fullname) and not os.path.islink(fullname)):
+            yield from _walk_dir(fullname, ddir=dfile,
+                                 maxlevels=maxlevels - 1, quiet=quiet)
+
+def compile_dir(dir, maxlevels=10, ddir=None, force=False, rx=None,
+                quiet=0, legacy=False, optimize=-1, workers=1):
+    """Byte-compile all modules in the given directory tree.
+
+    Arguments (only dir is required):
+
+    dir:       the directory to byte-compile
+    maxlevels: maximum recursion level (default 10)
+    ddir:      the directory that will be prepended to the path to the
+               file as it is compiled into each byte-code file.
+    force:     if True, force compilation, even if timestamps are up-to-date
+    quiet:     full output with False or 0, errors only with 1,
+               no output with 2
+    legacy:    if True, produce legacy pyc paths instead of PEP 3147 paths
+    optimize:  optimization level or -1 for level of the interpreter
+    workers:   maximum number of parallel workers
+    """
+    if workers is not None and workers < 0:
+        raise ValueError('workers must be greater or equal to 0')
+
+    files = _walk_dir(dir, quiet=quiet, maxlevels=maxlevels,
+                      ddir=ddir)
+    success = True
+    if workers is not None and workers != 1 and ProcessPoolExecutor is not None:
+        workers = workers or None
+        with ProcessPoolExecutor(max_workers=workers) as executor:
+            results = executor.map(partial(compile_file,
+                                           ddir=ddir, force=force,
+                                           rx=rx, quiet=quiet,
+                                           legacy=legacy,
+                                           optimize=optimize),
+                                   files)
+            success = min(results, default=True)
+    else:
+        for file in files:
+            if not compile_file(file, ddir, force, rx, quiet,
+                                legacy, optimize):
+                success = False
+    return success
+
+def compile_file(fullname, ddir=None, force=False, rx=None, quiet=0,
+                 legacy=False, optimize=-1):
+    """Byte-compile one file.
+
+    Arguments (only fullname is required):
+
+    fullname:  the file to byte-compile
+    ddir:      if given, the directory name compiled in to the
+               byte-code file.
+    force:     if True, force compilation, even if timestamps are up-to-date
+    quiet:     full output with False or 0, errors only with 1,
+               no output with 2
+    legacy:    if True, produce legacy pyc paths instead of PEP 3147 paths
+    optimize:  optimization level or -1 for level of the interpreter
+    """
+    success = True
+    if quiet < 2 and isinstance(fullname, os.PathLike):
+        fullname = os.fspath(fullname)
+    name = os.path.basename(fullname)
+    if ddir is not None:
+        dfile = os.path.join(ddir, name)
+    else:
+        dfile = None
+    if rx is not None:
+        mo = rx.search(fullname)
+        if mo:
+            return success
+    if os.path.isfile(fullname):
+        if legacy:
+            cfile = fullname + 'c'
+        else:
+            if optimize >= 0:
+                opt = optimize if optimize >= 1 else ''
+                cfile = importlib.util.cache_from_source(
+                                fullname, optimization=opt)
+            else:
+                cfile = importlib.util.cache_from_source(fullname)
+            cache_dir = os.path.dirname(cfile)
+        head, tail = name[:-3], name[-3:]
+        if tail == '.py':
+            if not force:
+                try:
+                    mtime = int(os.stat(fullname).st_mtime)
+                    expect = struct.pack('<4sl', importlib.util.MAGIC_NUMBER,
+                                         mtime)
+                    with open(cfile, 'rb') as chandle:
+                        actual = chandle.read(8)
+                    if expect == actual:
+                        return success
+                except OSError:
+                    pass
+            if not quiet:
+                print('Compiling {!r}...'.format(fullname))
+            try:
+                ok = py_compile.compile(fullname, cfile, dfile, True,
+                                        optimize=optimize)
+            except py_compile.PyCompileError as err:
+                success = False
+                if quiet >= 2:
+                    return success
+                elif quiet:
+                    print('*** Error compiling {!r}...'.format(fullname))
+                else:
+                    print('*** ', end='')
+                # escape non-printable characters in msg
+                msg = err.msg.encode(sys.stdout.encoding,
+                                     errors='backslashreplace')
+                msg = msg.decode(sys.stdout.encoding)
+                print(msg)
+            except (SyntaxError, UnicodeError, OSError) as e:
+                success = False
+                if quiet >= 2:
+                    return success
+                elif quiet:
+                    print('*** Error compiling {!r}...'.format(fullname))
+                else:
+                    print('*** ', end='')
+                print(e.__class__.__name__ + ':', e)
+            else:
+                if ok == 0:
+                    success = False
+    return success
+
+def compile_path(skip_curdir=1, maxlevels=0, force=False, quiet=0,
+                 legacy=False, optimize=-1):
+    """Byte-compile all module on sys.path.
+
+    Arguments (all optional):
+
+    skip_curdir: if true, skip current directory (default True)
+    maxlevels:   max recursion level (default 0)
+    force: as for compile_dir() (default False)
+    quiet: as for compile_dir() (default 0)
+    legacy: as for compile_dir() (default False)
+    optimize: as for compile_dir() (default -1)
+    """
+    success = True
+    for dir in sys.path:
+        if (not dir or dir == os.curdir) and skip_curdir:
+            if quiet < 2:
+                print('Skipping current directory')
+        else:
+            success = success and compile_dir(dir, maxlevels, None,
+                                              force, quiet=quiet,
+                                              legacy=legacy, optimize=optimize)
+    return success
+
+
+def main():
+    """Script main program."""
+    import argparse
+
+    parser = argparse.ArgumentParser(
+        description='Utilities to support installing Python libraries.')
+    parser.add_argument('-l', action='store_const', const=0,
+                        default=10, dest='maxlevels',
+                        help="don't recurse into subdirectories")
+    parser.add_argument('-r', type=int, dest='recursion',
+                        help=('control the maximum recursion level. '
+                              'if `-l` and `-r` options are specified, '
+                              'then `-r` takes precedence.'))
+    parser.add_argument('-f', action='store_true', dest='force',
+                        help='force rebuild even if timestamps are up to date')
+    parser.add_argument('-q', action='count', dest='quiet', default=0,
+                        help='output only error messages; -qq will suppress '
+                             'the error messages as well.')
+    parser.add_argument('-b', action='store_true', dest='legacy',
+                        help='use legacy (pre-PEP3147) compiled file locations')
+    parser.add_argument('-d', metavar='DESTDIR',  dest='ddir', default=None,
+                        help=('directory to prepend to file paths for use in '
+                              'compile-time tracebacks and in runtime '
+                              'tracebacks in cases where the source file is '
+                              'unavailable'))
+    parser.add_argument('-x', metavar='REGEXP', dest='rx', default=None,
+                        help=('skip files matching the regular expression; '
+                              'the regexp is searched for in the full path '
+                              'of each file considered for compilation'))
+    parser.add_argument('-i', metavar='FILE', dest='flist',
+                        help=('add all the files and directories listed in '
+                              'FILE to the list considered for compilation; '
+                              'if "-", names are read from stdin'))
+    parser.add_argument('compile_dest', metavar='FILE|DIR', nargs='*',
+                        help=('zero or more file and directory names '
+                              'to compile; if no arguments given, defaults '
+                              'to the equivalent of -l sys.path'))
+    parser.add_argument('-j', '--workers', default=1,
+                        type=int, help='Run compileall concurrently')
+
+    args = parser.parse_args()
+    compile_dests = args.compile_dest
+
+    if args.rx:
+        import re
+        args.rx = re.compile(args.rx)
+
+
+    if args.recursion is not None:
+        maxlevels = args.recursion
+    else:
+        maxlevels = args.maxlevels
+
+    # if flist is provided then load it
+    if args.flist:
+        try:
+            with (sys.stdin if args.flist=='-' else open(args.flist)) as f:
+                for line in f:
+                    compile_dests.append(line.strip())
+        except OSError:
+            if args.quiet < 2:
+                print("Error reading file list {}".format(args.flist))
+            return False
+
+    if args.workers is not None:
+        args.workers = args.workers or None
+
+    success = True
+    try:
+        if compile_dests:
+            for dest in compile_dests:
+                if os.path.isfile(dest):
+                    if not compile_file(dest, args.ddir, args.force, args.rx,
+                                        args.quiet, args.legacy):
+                        success = False
+                else:
+                    if not compile_dir(dest, maxlevels, args.ddir,
+                                       args.force, args.rx, args.quiet,
+                                       args.legacy, workers=args.workers):
+                        success = False
+            return success
+        else:
+            return compile_path(legacy=args.legacy, force=args.force,
+                                quiet=args.quiet)
+    except KeyboardInterrupt:
+        if args.quiet < 2:
+            print("\n[interrupted]")
+        return False
+    return True
+
+
+if __name__ == '__main__':
+    exit_status = int(not main())
+    sys.exit(exit_status)
diff --git a/Lib/configparser.py b/Lib/configparser.py
new file mode 100644
index 000000000..af5aca1fe
--- /dev/null
+++ b/Lib/configparser.py
@@ -0,0 +1,1339 @@
+"""Configuration file parser.
+
+A configuration file consists of sections, lead by a "[section]" header,
+and followed by "name: value" entries, with continuations and such in
+the style of RFC 822.
+
+Intrinsic defaults can be specified by passing them into the
+ConfigParser constructor as a dictionary.
+
+class:
+
+ConfigParser -- responsible for parsing a list of
+                    configuration files, and managing the parsed database.
+
+    methods:
+
+    __init__(defaults=None, dict_type=_default_dict, allow_no_value=False,
+             delimiters=('=', ':'), comment_prefixes=('#', ';'),
+             inline_comment_prefixes=None, strict=True,
+             empty_lines_in_values=True, default_section='DEFAULT',
+             interpolation=<unset>, converters=<unset>):
+        Create the parser. When `defaults' is given, it is initialized into the
+        dictionary or intrinsic defaults. The keys must be strings, the values
+        must be appropriate for %()s string interpolation.
+
+        When `dict_type' is given, it will be used to create the dictionary
+        objects for the list of sections, for the options within a section, and
+        for the default values.
+
+        When `delimiters' is given, it will be used as the set of substrings
+        that divide keys from values.
+
+        When `comment_prefixes' is given, it will be used as the set of
+        substrings that prefix comments in empty lines. Comments can be
+        indented.
+
+        When `inline_comment_prefixes' is given, it will be used as the set of
+        substrings that prefix comments in non-empty lines.
+
+        When `strict` is True, the parser won't allow for any section or option
+        duplicates while reading from a single source (file, string or
+        dictionary). Default is True.
+
+        When `empty_lines_in_values' is False (default: True), each empty line
+        marks the end of an option. Otherwise, internal empty lines of
+        a multiline option are kept as part of the value.
+
+        When `allow_no_value' is True (default: False), options without
+        values are accepted; the value presented for these is None.
+
+        When `default_section' is given, the name of the special section is
+        named accordingly. By default it is called ``"DEFAULT"`` but this can
+        be customized to point to any other valid section name. Its current
+        value can be retrieved using the ``parser_instance.default_section``
+        attribute and may be modified at runtime.
+
+        When `interpolation` is given, it should be an Interpolation subclass
+        instance. It will be used as the handler for option value
+        pre-processing when using getters. RawConfigParser object s don't do
+        any sort of interpolation, whereas ConfigParser uses an instance of
+        BasicInterpolation. The library also provides a ``zc.buildbot``
+        inspired ExtendedInterpolation implementation.
+
+        When `converters` is given, it should be a dictionary where each key
+        represents the name of a type converter and each value is a callable
+        implementing the conversion from string to the desired datatype. Every
+        converter gets its corresponding get*() method on the parser object and
+        section proxies.
+
+    sections()
+        Return all the configuration section names, sans DEFAULT.
+
+    has_section(section)
+        Return whether the given section exists.
+
+    has_option(section, option)
+        Return whether the given option exists in the given section.
+
+    options(section)
+        Return list of configuration options for the named section.
+
+    read(filenames, encoding=None)
+        Read and parse the list of named configuration files, given by
+        name.  A single filename is also allowed.  Non-existing files
+        are ignored.  Return list of successfully read files.
+
+    read_file(f, filename=None)
+        Read and parse one configuration file, given as a file object.
+        The filename defaults to f.name; it is only used in error
+        messages (if f has no `name' attribute, the string `<???>' is used).
+
+    read_string(string)
+        Read configuration from a given string.
+
+    read_dict(dictionary)
+        Read configuration from a dictionary. Keys are section names,
+        values are dictionaries with keys and values that should be present
+        in the section. If the used dictionary type preserves order, sections
+        and their keys will be added in order. Values are automatically
+        converted to strings.
+
+    get(section, option, raw=False, vars=None, fallback=_UNSET)
+        Return a string value for the named option.  All % interpolations are
+        expanded in the return values, based on the defaults passed into the
+        constructor and the DEFAULT section.  Additional substitutions may be
+        provided using the `vars' argument, which must be a dictionary whose
+        contents override any pre-existing defaults. If `option' is a key in
+        `vars', the value from `vars' is used.
+
+    getint(section, options, raw=False, vars=None, fallback=_UNSET)
+        Like get(), but convert value to an integer.
+
+    getfloat(section, options, raw=False, vars=None, fallback=_UNSET)
+        Like get(), but convert value to a float.
+
+    getboolean(section, options, raw=False, vars=None, fallback=_UNSET)
+        Like get(), but convert value to a boolean (currently case
+        insensitively defined as 0, false, no, off for False, and 1, true,
+        yes, on for True).  Returns False or True.
+
+    items(section=_UNSET, raw=False, vars=None)
+        If section is given, return a list of tuples with (name, value) for
+        each option in the section. Otherwise, return a list of tuples with
+        (section_name, section_proxy) for each section, including DEFAULTSECT.
+
+    remove_section(section)
+        Remove the given file section and all its options.
+
+    remove_option(section, option)
+        Remove the given option from the given section.
+
+    set(section, option, value)
+        Set the given option.
+
+    write(fp, space_around_delimiters=True)
+        Write the configuration state in .ini format. If
+        `space_around_delimiters' is True (the default), delimiters
+        between keys and values are surrounded by spaces.
+"""
+
+from collections.abc import MutableMapping
+from collections import OrderedDict as _default_dict, ChainMap as _ChainMap
+import functools
+import io
+import itertools
+import re
+import sys
+import warnings
+
+__all__ = ["NoSectionError", "DuplicateOptionError", "DuplicateSectionError",
+           "NoOptionError", "InterpolationError", "InterpolationDepthError",
+           "InterpolationMissingOptionError", "InterpolationSyntaxError",
+           "ParsingError", "MissingSectionHeaderError",
+           "ConfigParser", "SafeConfigParser", "RawConfigParser",
+           "Interpolation", "BasicInterpolation",  "ExtendedInterpolation",
+           "LegacyInterpolation", "SectionProxy", "ConverterMapping",
+           "DEFAULTSECT", "MAX_INTERPOLATION_DEPTH"]
+
+DEFAULTSECT = "DEFAULT"
+
+MAX_INTERPOLATION_DEPTH = 10
+
+
+
+# exception classes
+class Error(Exception):
+    """Base class for ConfigParser exceptions."""
+
+    def __init__(self, msg=''):
+        self.message = msg
+        Exception.__init__(self, msg)
+
+    def __repr__(self):
+        return self.message
+
+    __str__ = __repr__
+
+
+class NoSectionError(Error):
+    """Raised when no section matches a requested option."""
+
+    def __init__(self, section):
+        Error.__init__(self, 'No section: %r' % (section,))
+        self.section = section
+        self.args = (section, )
+
+
+class DuplicateSectionError(Error):
+    """Raised when a section is repeated in an input source.
+
+    Possible repetitions that raise this exception are: multiple creation
+    using the API or in strict parsers when a section is found more than once
+    in a single input file, string or dictionary.
+    """
+
+    def __init__(self, section, source=None, lineno=None):
+        msg = [repr(section), " already exists"]
+        if source is not None:
+            message = ["While reading from ", repr(source)]
+            if lineno is not None:
+                message.append(" [line {0:2d}]".format(lineno))
+            message.append(": section ")
+            message.extend(msg)
+            msg = message
+        else:
+            msg.insert(0, "Section ")
+        Error.__init__(self, "".join(msg))
+        self.section = section
+        self.source = source
+        self.lineno = lineno
+        self.args = (section, source, lineno)
+
+
+class DuplicateOptionError(Error):
+    """Raised by strict parsers when an option is repeated in an input source.
+
+    Current implementation raises this exception only when an option is found
+    more than once in a single file, string or dictionary.
+    """
+
+    def __init__(self, section, option, source=None, lineno=None):
+        msg = [repr(option), " in section ", repr(section),
+               " already exists"]
+        if source is not None:
+            message = ["While reading from ", repr(source)]
+            if lineno is not None:
+                message.append(" [line {0:2d}]".format(lineno))
+            message.append(": option ")
+            message.extend(msg)
+            msg = message
+        else:
+            msg.insert(0, "Option ")
+        Error.__init__(self, "".join(msg))
+        self.section = section
+        self.option = option
+        self.source = source
+        self.lineno = lineno
+        self.args = (section, option, source, lineno)
+
+
+class NoOptionError(Error):
+    """A requested option was not found."""
+
+    def __init__(self, option, section):
+        Error.__init__(self, "No option %r in section: %r" %
+                       (option, section))
+        self.option = option
+        self.section = section
+        self.args = (option, section)
+
+
+class InterpolationError(Error):
+    """Base class for interpolation-related exceptions."""
+
+    def __init__(self, option, section, msg):
+        Error.__init__(self, msg)
+        self.option = option
+        self.section = section
+        self.args = (option, section, msg)
+
+
+class InterpolationMissingOptionError(InterpolationError):
+    """A string substitution required a setting which was not available."""
+
+    def __init__(self, option, section, rawval, reference):
+        msg = ("Bad value substitution: option {!r} in section {!r} contains "
+               "an interpolation key {!r} which is not a valid option name. "
+               "Raw value: {!r}".format(option, section, reference, rawval))
+        InterpolationError.__init__(self, option, section, msg)
+        self.reference = reference
+        self.args = (option, section, rawval, reference)
+
+
+class InterpolationSyntaxError(InterpolationError):
+    """Raised when the source text contains invalid syntax.
+
+    Current implementation raises this exception when the source text into
+    which substitutions are made does not conform to the required syntax.
+    """
+
+
+class InterpolationDepthError(InterpolationError):
+    """Raised when substitutions are nested too deeply."""
+
+    def __init__(self, option, section, rawval):
+        msg = ("Recursion limit exceeded in value substitution: option {!r} "
+               "in section {!r} contains an interpolation key which "
+               "cannot be substituted in {} steps. Raw value: {!r}"
+               "".format(option, section, MAX_INTERPOLATION_DEPTH,
+                         rawval))
+        InterpolationError.__init__(self, option, section, msg)
+        self.args = (option, section, rawval)
+
+
+class ParsingError(Error):
+    """Raised when a configuration file does not follow legal syntax."""
+
+    def __init__(self, source=None, filename=None):
+        # Exactly one of `source'/`filename' arguments has to be given.
+        # `filename' kept for compatibility.
+        if filename and source:
+            raise ValueError("Cannot specify both `filename' and `source'. "
+                             "Use `source'.")
+        elif not filename and not source:
+            raise ValueError("Required argument `source' not given.")
+        elif filename:
+            source = filename
+        Error.__init__(self, 'Source contains parsing errors: %r' % source)
+        self.source = source
+        self.errors = []
+        self.args = (source, )
+
+    @property
+    def filename(self):
+        """Deprecated, use `source'."""
+        warnings.warn(
+            "The 'filename' attribute will be removed in future versions.  "
+            "Use 'source' instead.",
+            DeprecationWarning, stacklevel=2
+        )
+        return self.source
+
+    @filename.setter
+    def filename(self, value):
+        """Deprecated, user `source'."""
+        warnings.warn(
+            "The 'filename' attribute will be removed in future versions.  "
+            "Use 'source' instead.",
+            DeprecationWarning, stacklevel=2
+        )
+        self.source = value
+
+    def append(self, lineno, line):
+        self.errors.append((lineno, line))
+        self.message += '\n\t[line %2d]: %s' % (lineno, line)
+
+
+class MissingSectionHeaderError(ParsingError):
+    """Raised when a key-value pair is found before any section header."""
+
+    def __init__(self, filename, lineno, line):
+        Error.__init__(
+            self,
+            'File contains no section headers.\nfile: %r, line: %d\n%r' %
+            (filename, lineno, line))
+        self.source = filename
+        self.lineno = lineno
+        self.line = line
+        self.args = (filename, lineno, line)
+
+
+# Used in parser getters to indicate the default behaviour when a specific
+# option is not found it to raise an exception. Created to enable `None' as
+# a valid fallback value.
+_UNSET = object()
+
+
+class Interpolation:
+    """Dummy interpolation that passes the value through with no changes."""
+
+    def before_get(self, parser, section, option, value, defaults):
+        return value
+
+    def before_set(self, parser, section, option, value):
+        return value
+
+    def before_read(self, parser, section, option, value):
+        return value
+
+    def before_write(self, parser, section, option, value):
+        return value
+
+
+class BasicInterpolation(Interpolation):
+    """Interpolation as implemented in the classic ConfigParser.
+
+    The option values can contain format strings which refer to other values in
+    the same section, or values in the special default section.
+
+    For example:
+
+        something: %(dir)s/whatever
+
+    would resolve the "%(dir)s" to the value of dir.  All reference
+    expansions are done late, on demand. If a user needs to use a bare % in
+    a configuration file, she can escape it by writing %%. Other % usage
+    is considered a user error and raises `InterpolationSyntaxError'."""
+
+    _KEYCRE = re.compile(r"%\(([^)]+)\)s")
+
+    def before_get(self, parser, section, option, value, defaults):
+        L = []
+        self._interpolate_some(parser, option, L, value, section, defaults, 1)
+        return ''.join(L)
+
+    def before_set(self, parser, section, option, value):
+        tmp_value = value.replace('%%', '') # escaped percent signs
+        tmp_value = self._KEYCRE.sub('', tmp_value) # valid syntax
+        if '%' in tmp_value:
+            raise ValueError("invalid interpolation syntax in %r at "
+                             "position %d" % (value, tmp_value.find('%')))
+        return value
+
+    def _interpolate_some(self, parser, option, accum, rest, section, map,
+                          depth):
+        rawval = parser.get(section, option, raw=True, fallback=rest)
+        if depth > MAX_INTERPOLATION_DEPTH:
+            raise InterpolationDepthError(option, section, rawval)
+        while rest:
+            p = rest.find("%")
+            if p < 0:
+                accum.append(rest)
+                return
+            if p > 0:
+                accum.append(rest[:p])
+                rest = rest[p:]
+            # p is no longer used
+            c = rest[1:2]
+            if c == "%":
+                accum.append("%")
+                rest = rest[2:]
+            elif c == "(":
+                m = self._KEYCRE.match(rest)
+                if m is None:
+                    raise InterpolationSyntaxError(option, section,
+                        "bad interpolation variable reference %r" % rest)
+                var = parser.optionxform(m.group(1))
+                rest = rest[m.end():]
+                try:
+                    v = map[var]
+                except KeyError:
+                    raise InterpolationMissingOptionError(
+                        option, section, rawval, var) from None
+                if "%" in v:
+                    self._interpolate_some(parser, option, accum, v,
+                                           section, map, depth + 1)
+                else:
+                    accum.append(v)
+            else:
+                raise InterpolationSyntaxError(
+                    option, section,
+                    "'%%' must be followed by '%%' or '(', "
+                    "found: %r" % (rest,))
+
+
+class ExtendedInterpolation(Interpolation):
+    """Advanced variant of interpolation, supports the syntax used by
+    `zc.buildout'. Enables interpolation between sections."""
+
+    _KEYCRE = re.compile(r"\$\{([^}]+)\}")
+
+    def before_get(self, parser, section, option, value, defaults):
+        L = []
+        self._interpolate_some(parser, option, L, value, section, defaults, 1)
+        return ''.join(L)
+
+    def before_set(self, parser, section, option, value):
+        tmp_value = value.replace('$$', '') # escaped dollar signs
+        tmp_value = self._KEYCRE.sub('', tmp_value) # valid syntax
+        if '$' in tmp_value:
+            raise ValueError("invalid interpolation syntax in %r at "
+                             "position %d" % (value, tmp_value.find('$')))
+        return value
+
+    def _interpolate_some(self, parser, option, accum, rest, section, map,
+                          depth):
+        rawval = parser.get(section, option, raw=True, fallback=rest)
+        if depth > MAX_INTERPOLATION_DEPTH:
+            raise InterpolationDepthError(option, section, rawval)
+        while rest:
+            p = rest.find("$")
+            if p < 0:
+                accum.append(rest)
+                return
+            if p > 0:
+                accum.append(rest[:p])
+                rest = rest[p:]
+            # p is no longer used
+            c = rest[1:2]
+            if c == "$":
+                accum.append("$")
+                rest = rest[2:]
+            elif c == "{":
+                m = self._KEYCRE.match(rest)
+                if m is None:
+                    raise InterpolationSyntaxError(option, section,
+                        "bad interpolation variable reference %r" % rest)
+                path = m.group(1).split(':')
+                rest = rest[m.end():]
+                sect = section
+                opt = option
+                try:
+                    if len(path) == 1:
+                        opt = parser.optionxform(path[0])
+                        v = map[opt]
+                    elif len(path) == 2:
+                        sect = path[0]
+                        opt = parser.optionxform(path[1])
+                        v = parser.get(sect, opt, raw=True)
+                    else:
+                        raise InterpolationSyntaxError(
+                            option, section,
+                            "More than one ':' found: %r" % (rest,))
+                except (KeyError, NoSectionError, NoOptionError):
+                    raise InterpolationMissingOptionError(
+                        option, section, rawval, ":".join(path)) from None
+                if "$" in v:
+                    self._interpolate_some(parser, opt, accum, v, sect,
+                                           dict(parser.items(sect, raw=True)),
+                                           depth + 1)
+                else:
+                    accum.append(v)
+            else:
+                raise InterpolationSyntaxError(
+                    option, section,
+                    "'$' must be followed by '$' or '{', "
+                    "found: %r" % (rest,))
+
+
+class LegacyInterpolation(Interpolation):
+    """Deprecated interpolation used in old versions of ConfigParser.
+    Use BasicInterpolation or ExtendedInterpolation instead."""
+
+    _KEYCRE = re.compile(r"%\(([^)]*)\)s|.")
+
+    def before_get(self, parser, section, option, value, vars):
+        rawval = value
+        depth = MAX_INTERPOLATION_DEPTH
+        while depth:                    # Loop through this until it's done
+            depth -= 1
+            if value and "%(" in value:
+                replace = functools.partial(self._interpolation_replace,
+                                            parser=parser)
+                value = self._KEYCRE.sub(replace, value)
+                try:
+                    value = value % vars
+                except KeyError as e:
+                    raise InterpolationMissingOptionError(
+                        option, section, rawval, e.args[0]) from None
+            else:
+                break
+        if value and "%(" in value:
+            raise InterpolationDepthError(option, section, rawval)
+        return value
+
+    def before_set(self, parser, section, option, value):
+        return value
+
+    @staticmethod
+    def _interpolation_replace(match, parser):
+        s = match.group(1)
+        if s is None:
+            return match.group()
+        else:
+            return "%%(%s)s" % parser.optionxform(s)
+
+
+class RawConfigParser(MutableMapping):
+    """ConfigParser that does not do interpolation."""
+
+    # Regular expressions for parsing section headers and options
+    _SECT_TMPL = r"""
+        \[                                 # [
+        (?P<header>[^]]+)                  # very permissive!
+        \]                                 # ]
+        """
+    _OPT_TMPL = r"""
+        (?P<option>.*?)                    # very permissive!
+        \s*(?P<vi>{delim})\s*              # any number of space/tab,
+                                           # followed by any of the
+                                           # allowed delimiters,
+                                           # followed by any space/tab
+        (?P<value>.*)$                     # everything up to eol
+        """
+    _OPT_NV_TMPL = r"""
+        (?P<option>.*?)                    # very permissive!
+        \s*(?:                             # any number of space/tab,
+        (?P<vi>{delim})\s*                 # optionally followed by
+                                           # any of the allowed
+                                           # delimiters, followed by any
+                                           # space/tab
+        (?P<value>.*))?$                   # everything up to eol
+        """
+    # Interpolation algorithm to be used if the user does not specify another
+    _DEFAULT_INTERPOLATION = Interpolation()
+    # Compiled regular expression for matching sections
+    SECTCRE = re.compile(_SECT_TMPL, re.VERBOSE)
+    # Compiled regular expression for matching options with typical separators
+    OPTCRE = re.compile(_OPT_TMPL.format(delim="=|:"), re.VERBOSE)
+    # Compiled regular expression for matching options with optional values
+    # delimited using typical separators
+    OPTCRE_NV = re.compile(_OPT_NV_TMPL.format(delim="=|:"), re.VERBOSE)
+    # Compiled regular expression for matching leading whitespace in a line
+    NONSPACECRE = re.compile(r"\S")
+    # Possible boolean values in the configuration.
+    BOOLEAN_STATES = {'1': True, 'yes': True, 'true': True, 'on': True,
+                      '0': False, 'no': False, 'false': False, 'off': False}
+
+    def __init__(self, defaults=None, dict_type=_default_dict,
+                 allow_no_value=False, *, delimiters=('=', ':'),
+                 comment_prefixes=('#', ';'), inline_comment_prefixes=None,
+                 strict=True, empty_lines_in_values=True,
+                 default_section=DEFAULTSECT,
+                 interpolation=_UNSET, converters=_UNSET):
+
+        self._dict = dict_type
+        self._sections = self._dict()
+        self._defaults = self._dict()
+        self._converters = ConverterMapping(self)
+        self._proxies = self._dict()
+        self._proxies[default_section] = SectionProxy(self, default_section)
+        if defaults:
+            for key, value in defaults.items():
+                self._defaults[self.optionxform(key)] = value
+        self._delimiters = tuple(delimiters)
+        if delimiters == ('=', ':'):
+            self._optcre = self.OPTCRE_NV if allow_no_value else self.OPTCRE
+        else:
+            d = "|".join(re.escape(d) for d in delimiters)
+            if allow_no_value:
+                self._optcre = re.compile(self._OPT_NV_TMPL.format(delim=d),
+                                          re.VERBOSE)
+            else:
+                self._optcre = re.compile(self._OPT_TMPL.format(delim=d),
+                                          re.VERBOSE)
+        self._comment_prefixes = tuple(comment_prefixes or ())
+        self._inline_comment_prefixes = tuple(inline_comment_prefixes or ())
+        self._strict = strict
+        self._allow_no_value = allow_no_value
+        self._empty_lines_in_values = empty_lines_in_values
+        self.default_section=default_section
+        self._interpolation = interpolation
+        if self._interpolation is _UNSET:
+            self._interpolation = self._DEFAULT_INTERPOLATION
+        if self._interpolation is None:
+            self._interpolation = Interpolation()
+        if converters is not _UNSET:
+            self._converters.update(converters)
+
+    def defaults(self):
+        return self._defaults
+
+    def sections(self):
+        """Return a list of section names, excluding [DEFAULT]"""
+        # self._sections will never have [DEFAULT] in it
+        return list(self._sections.keys())
+
+    def add_section(self, section):
+        """Create a new section in the configuration.
+
+        Raise DuplicateSectionError if a section by the specified name
+        already exists. Raise ValueError if name is DEFAULT.
+        """
+        if section == self.default_section:
+            raise ValueError('Invalid section name: %r' % section)
+
+        if section in self._sections:
+            raise DuplicateSectionError(section)
+        self._sections[section] = self._dict()
+        self._proxies[section] = SectionProxy(self, section)
+
+    def has_section(self, section):
+        """Indicate whether the named section is present in the configuration.
+
+        The DEFAULT section is not acknowledged.
+        """
+        return section in self._sections
+
+    def options(self, section):
+        """Return a list of option names for the given section name."""
+        try:
+            opts = self._sections[section].copy()
+        except KeyError:
+            raise NoSectionError(section) from None
+        opts.update(self._defaults)
+        return list(opts.keys())
+
+    def read(self, filenames, encoding=None):
+        """Read and parse a filename or a list of filenames.
+
+        Files that cannot be opened are silently ignored; this is
+        designed so that you can specify a list of potential
+        configuration file locations (e.g. current directory, user's
+        home directory, systemwide directory), and all existing
+        configuration files in the list will be read.  A single
+        filename may also be given.
+
+        Return list of successfully read files.
+        """
+        if isinstance(filenames, str):
+            filenames = [filenames]
+        read_ok = []
+        for filename in filenames:
+            try:
+                with open(filename, encoding=encoding) as fp:
+                    self._read(fp, filename)
+            except OSError:
+                continue
+            read_ok.append(filename)
+        return read_ok
+
+    def read_file(self, f, source=None):
+        """Like read() but the argument must be a file-like object.
+
+        The `f' argument must be iterable, returning one line at a time.
+        Optional second argument is the `source' specifying the name of the
+        file being read. If not given, it is taken from f.name. If `f' has no
+        `name' attribute, `<???>' is used.
+        """
+        if source is None:
+            try:
+                source = f.name
+            except AttributeError:
+                source = '<???>'
+        self._read(f, source)
+
+    def read_string(self, string, source='<string>'):
+        """Read configuration from a given string."""
+        sfile = io.StringIO(string)
+        self.read_file(sfile, source)
+
+    def read_dict(self, dictionary, source='<dict>'):
+        """Read configuration from a dictionary.
+
+        Keys are section names, values are dictionaries with keys and values
+        that should be present in the section. If the used dictionary type
+        preserves order, sections and their keys will be added in order.
+
+        All types held in the dictionary are converted to strings during
+        reading, including section names, option names and keys.
+
+        Optional second argument is the `source' specifying the name of the
+        dictionary being read.
+        """
+        elements_added = set()
+        for section, keys in dictionary.items():
+            section = str(section)
+            try:
+                self.add_section(section)
+            except (DuplicateSectionError, ValueError):
+                if self._strict and section in elements_added:
+                    raise
+            elements_added.add(section)
+            for key, value in keys.items():
+                key = self.optionxform(str(key))
+                if value is not None:
+                    value = str(value)
+                if self._strict and (section, key) in elements_added:
+                    raise DuplicateOptionError(section, key, source)
+                elements_added.add((section, key))
+                self.set(section, key, value)
+
+    def readfp(self, fp, filename=None):
+        """Deprecated, use read_file instead."""
+        warnings.warn(
+            "This method will be removed in future versions.  "
+            "Use 'parser.read_file()' instead.",
+            DeprecationWarning, stacklevel=2
+        )
+        self.read_file(fp, source=filename)
+
+    def get(self, section, option, *, raw=False, vars=None, fallback=_UNSET):
+        """Get an option value for a given section.
+
+        If `vars' is provided, it must be a dictionary. The option is looked up
+        in `vars' (if provided), `section', and in `DEFAULTSECT' in that order.
+        If the key is not found and `fallback' is provided, it is used as
+        a fallback value. `None' can be provided as a `fallback' value.
+
+        If interpolation is enabled and the optional argument `raw' is False,
+        all interpolations are expanded in the return values.
+
+        Arguments `raw', `vars', and `fallback' are keyword only.
+
+        The section DEFAULT is special.
+        """
+        try:
+            d = self._unify_values(section, vars)
+        except NoSectionError:
+            if fallback is _UNSET:
+                raise
+            else:
+                return fallback
+        option = self.optionxform(option)
+        try:
+            value = d[option]
+        except KeyError:
+            if fallback is _UNSET:
+                raise NoOptionError(option, section)
+            else:
+                return fallback
+
+        if raw or value is None:
+            return value
+        else:
+            return self._interpolation.before_get(self, section, option, value,
+                                                  d)
+
+    def _get(self, section, conv, option, **kwargs):
+        return conv(self.get(section, option, **kwargs))
+
+    def _get_conv(self, section, option, conv, *, raw=False, vars=None,
+                  fallback=_UNSET, **kwargs):
+        try:
+            return self._get(section, conv, option, raw=raw, vars=vars,
+                             **kwargs)
+        except (NoSectionError, NoOptionError):
+            if fallback is _UNSET:
+                raise
+            return fallback
+
+    # getint, getfloat and getboolean provided directly for backwards compat
+    def getint(self, section, option, *, raw=False, vars=None,
+               fallback=_UNSET, **kwargs):
+        return self._get_conv(section, option, int, raw=raw, vars=vars,
+                              fallback=fallback, **kwargs)
+
+    def getfloat(self, section, option, *, raw=False, vars=None,
+                 fallback=_UNSET, **kwargs):
+        return self._get_conv(section, option, float, raw=raw, vars=vars,
+                              fallback=fallback, **kwargs)
+
+    def getboolean(self, section, option, *, raw=False, vars=None,
+                   fallback=_UNSET, **kwargs):
+        return self._get_conv(section, option, self._convert_to_boolean,
+                              raw=raw, vars=vars, fallback=fallback, **kwargs)
+
+    def items(self, section=_UNSET, raw=False, vars=None):
+        """Return a list of (name, value) tuples for each option in a section.
+
+        All % interpolations are expanded in the return values, based on the
+        defaults passed into the constructor, unless the optional argument
+        `raw' is true.  Additional substitutions may be provided using the
+        `vars' argument, which must be a dictionary whose contents overrides
+        any pre-existing defaults.
+
+        The section DEFAULT is special.
+        """
+        if section is _UNSET:
+            return super().items()
+        d = self._defaults.copy()
+        try:
+            d.update(self._sections[section])
+        except KeyError:
+            if section != self.default_section:
+                raise NoSectionError(section)
+        # Update with the entry specific variables
+        if vars:
+            for key, value in vars.items():
+                d[self.optionxform(key)] = value
+        value_getter = lambda option: self._interpolation.before_get(self,
+            section, option, d[option], d)
+        if raw:
+            value_getter = lambda option: d[option]
+        return [(option, value_getter(option)) for option in d.keys()]
+
+    def popitem(self):
+        """Remove a section from the parser and return it as
+        a (section_name, section_proxy) tuple. If no section is present, raise
+        KeyError.
+
+        The section DEFAULT is never returned because it cannot be removed.
+        """
+        for key in self.sections():
+            value = self[key]
+            del self[key]
+            return key, value
+        raise KeyError
+
+    def optionxform(self, optionstr):
+        return optionstr.lower()
+
+    def has_option(self, section, option):
+        """Check for the existence of a given option in a given section.
+        If the specified `section' is None or an empty string, DEFAULT is
+        assumed. If the specified `section' does not exist, returns False."""
+        if not section or section == self.default_section:
+            option = self.optionxform(option)
+            return option in self._defaults
+        elif section not in self._sections:
+            return False
+        else:
+            option = self.optionxform(option)
+            return (option in self._sections[section]
+                    or option in self._defaults)
+
+    def set(self, section, option, value=None):
+        """Set an option."""
+        if value:
+            value = self._interpolation.before_set(self, section, option,
+                                                   value)
+        if not section or section == self.default_section:
+            sectdict = self._defaults
+        else:
+            try:
+                sectdict = self._sections[section]
+            except KeyError:
+                raise NoSectionError(section) from None
+        sectdict[self.optionxform(option)] = value
+
+    def write(self, fp, space_around_delimiters=True):
+        """Write an .ini-format representation of the configuration state.
+
+        If `space_around_delimiters' is True (the default), delimiters
+        between keys and values are surrounded by spaces.
+        """
+        if space_around_delimiters:
+            d = " {} ".format(self._delimiters[0])
+        else:
+            d = self._delimiters[0]
+        if self._defaults:
+            self._write_section(fp, self.default_section,
+                                    self._defaults.items(), d)
+        for section in self._sections:
+            self._write_section(fp, section,
+                                self._sections[section].items(), d)
+
+    def _write_section(self, fp, section_name, section_items, delimiter):
+        """Write a single section to the specified `fp'."""
+        fp.write("[{}]\n".format(section_name))
+        for key, value in section_items:
+            value = self._interpolation.before_write(self, section_name, key,
+                                                     value)
+            if value is not None or not self._allow_no_value:
+                value = delimiter + str(value).replace('\n', '\n\t')
+            else:
+                value = ""
+            fp.write("{}{}\n".format(key, value))
+        fp.write("\n")
+
+    def remove_option(self, section, option):
+        """Remove an option."""
+        if not section or section == self.default_section:
+            sectdict = self._defaults
+        else:
+            try:
+                sectdict = self._sections[section]
+            except KeyError:
+                raise NoSectionError(section) from None
+        option = self.optionxform(option)
+        existed = option in sectdict
+        if existed:
+            del sectdict[option]
+        return existed
+
+    def remove_section(self, section):
+        """Remove a file section."""
+        existed = section in self._sections
+        if existed:
+            del self._sections[section]
+            del self._proxies[section]
+        return existed
+
+    def __getitem__(self, key):
+        if key != self.default_section and not self.has_section(key):
+            raise KeyError(key)
+        return self._proxies[key]
+
+    def __setitem__(self, key, value):
+        # To conform with the mapping protocol, overwrites existing values in
+        # the section.
+
+        # XXX this is not atomic if read_dict fails at any point. Then again,
+        # no update method in configparser is atomic in this implementation.
+        if key == self.default_section:
+            self._defaults.clear()
+        elif key in self._sections:
+            self._sections[key].clear()
+        self.read_dict({key: value})
+
+    def __delitem__(self, key):
+        if key == self.default_section:
+            raise ValueError("Cannot remove the default section.")
+        if not self.has_section(key):
+            raise KeyError(key)
+        self.remove_section(key)
+
+    def __contains__(self, key):
+        return key == self.default_section or self.has_section(key)
+
+    def __len__(self):
+        return len(self._sections) + 1 # the default section
+
+    def __iter__(self):
+        # XXX does it break when underlying container state changed?
+        return itertools.chain((self.default_section,), self._sections.keys())
+
+    def _read(self, fp, fpname):
+        """Parse a sectioned configuration file.
+
+        Each section in a configuration file contains a header, indicated by
+        a name in square brackets (`[]'), plus key/value options, indicated by
+        `name' and `value' delimited with a specific substring (`=' or `:' by
+        default).
+
+        Values can span multiple lines, as long as they are indented deeper
+        than the first line of the value. Depending on the parser's mode, blank
+        lines may be treated as parts of multiline values or ignored.
+
+        Configuration files may include comments, prefixed by specific
+        characters (`#' and `;' by default). Comments may appear on their own
+        in an otherwise empty line or may be entered in lines holding values or
+        section names.
+        """
+        elements_added = set()
+        cursect = None                        # None, or a dictionary
+        sectname = None
+        optname = None
+        lineno = 0
+        indent_level = 0
+        e = None                              # None, or an exception
+        for lineno, line in enumerate(fp, start=1):
+            comment_start = sys.maxsize
+            # strip inline comments
+            inline_prefixes = {p: -1 for p in self._inline_comment_prefixes}
+            while comment_start == sys.maxsize and inline_prefixes:
+                next_prefixes = {}
+                for prefix, index in inline_prefixes.items():
+                    index = line.find(prefix, index+1)
+                    if index == -1:
+                        continue
+                    next_prefixes[prefix] = index
+                    if index == 0 or (index > 0 and line[index-1].isspace()):
+                        comment_start = min(comment_start, index)
+                inline_prefixes = next_prefixes
+            # strip full line comments
+            for prefix in self._comment_prefixes:
+                if line.strip().startswith(prefix):
+                    comment_start = 0
+                    break
+            if comment_start == sys.maxsize:
+                comment_start = None
+            value = line[:comment_start].strip()
+            if not value:
+                if self._empty_lines_in_values:
+                    # add empty line to the value, but only if there was no
+                    # comment on the line
+                    if (comment_start is None and
+                        cursect is not None and
+                        optname and
+                        cursect[optname] is not None):
+                        cursect[optname].append('') # newlines added at join
+                else:
+                    # empty line marks end of value
+                    indent_level = sys.maxsize
+                continue
+            # continuation line?
+            first_nonspace = self.NONSPACECRE.search(line)
+            cur_indent_level = first_nonspace.start() if first_nonspace else 0
+            if (cursect is not None and optname and
+                cur_indent_level > indent_level):
+                cursect[optname].append(value)
+            # a section header or option header?
+            else:
+                indent_level = cur_indent_level
+                # is it a section header?
+                mo = self.SECTCRE.match(value)
+                if mo:
+                    sectname = mo.group('header')
+                    if sectname in self._sections:
+                        if self._strict and sectname in elements_added:
+                            raise DuplicateSectionError(sectname, fpname,
+                                                        lineno)
+                        cursect = self._sections[sectname]
+                        elements_added.add(sectname)
+                    elif sectname == self.default_section:
+                        cursect = self._defaults
+                    else:
+                        cursect = self._dict()
+                        self._sections[sectname] = cursect
+                        self._proxies[sectname] = SectionProxy(self, sectname)
+                        elements_added.add(sectname)
+                    # So sections can't start with a continuation line
+                    optname = None
+                # no section header in the file?
+                elif cursect is None:
+                    raise MissingSectionHeaderError(fpname, lineno, line)
+                # an option line?
+                else:
+                    mo = self._optcre.match(value)
+                    if mo:
+                        optname, vi, optval = mo.group('option', 'vi', 'value')
+                        if not optname:
+                            e = self._handle_error(e, fpname, lineno, line)
+                        optname = self.optionxform(optname.rstrip())
+                        if (self._strict and
+                            (sectname, optname) in elements_added):
+                            raise DuplicateOptionError(sectname, optname,
+                                                       fpname, lineno)
+                        elements_added.add((sectname, optname))
+                        # This check is fine because the OPTCRE cannot
+                        # match if it would set optval to None
+                        if optval is not None:
+                            optval = optval.strip()
+                            cursect[optname] = [optval]
+                        else:
+                            # valueless option handling
+                            cursect[optname] = None
+                    else:
+                        # a non-fatal parsing error occurred. set up the
+                        # exception but keep going. the exception will be
+                        # raised at the end of the file and will contain a
+                        # list of all bogus lines
+                        e = self._handle_error(e, fpname, lineno, line)
+        self._join_multiline_values()
+        # if any parsing errors occurred, raise an exception
+        if e:
+            raise e
+
+    def _join_multiline_values(self):
+        defaults = self.default_section, self._defaults
+        all_sections = itertools.chain((defaults,),
+                                       self._sections.items())
+        for section, options in all_sections:
+            for name, val in options.items():
+                if isinstance(val, list):
+                    val = '\n'.join(val).rstrip()
+                options[name] = self._interpolation.before_read(self,
+                                                                section,
+                                                                name, val)
+
+    def _handle_error(self, exc, fpname, lineno, line):
+        if not exc:
+            exc = ParsingError(fpname)
+        exc.append(lineno, repr(line))
+        return exc
+
+    def _unify_values(self, section, vars):
+        """Create a sequence of lookups with 'vars' taking priority over
+        the 'section' which takes priority over the DEFAULTSECT.
+
+        """
+        sectiondict = {}
+        try:
+            sectiondict = self._sections[section]
+        except KeyError:
+            if section != self.default_section:
+                raise NoSectionError(section)
+        # Update with the entry specific variables
+        vardict = {}
+        if vars:
+            for key, value in vars.items():
+                if value is not None:
+                    value = str(value)
+                vardict[self.optionxform(key)] = value
+        return _ChainMap(vardict, sectiondict, self._defaults)
+
+    def _convert_to_boolean(self, value):
+        """Return a boolean value translating from other types if necessary.
+        """
+        if value.lower() not in self.BOOLEAN_STATES:
+            raise ValueError('Not a boolean: %s' % value)
+        return self.BOOLEAN_STATES[value.lower()]
+
+    def _validate_value_types(self, *, section="", option="", value=""):
+        """Raises a TypeError for non-string values.
+
+        The only legal non-string value if we allow valueless
+        options is None, so we need to check if the value is a
+        string if:
+        - we do not allow valueless options, or
+        - we allow valueless options but the value is not None
+
+        For compatibility reasons this method is not used in classic set()
+        for RawConfigParsers. It is invoked in every case for mapping protocol
+        access and in ConfigParser.set().
+        """
+        if not isinstance(section, str):
+            raise TypeError("section names must be strings")
+        if not isinstance(option, str):
+            raise TypeError("option keys must be strings")
+        if not self._allow_no_value or value:
+            if not isinstance(value, str):
+                raise TypeError("option values must be strings")
+
+    @property
+    def converters(self):
+        return self._converters
+
+
+class ConfigParser(RawConfigParser):
+    """ConfigParser implementing interpolation."""
+
+    _DEFAULT_INTERPOLATION = BasicInterpolation()
+
+    def set(self, section, option, value=None):
+        """Set an option.  Extends RawConfigParser.set by validating type and
+        interpolation syntax on the value."""
+        self._validate_value_types(option=option, value=value)
+        super().set(section, option, value)
+
+    def add_section(self, section):
+        """Create a new section in the configuration.  Extends
+        RawConfigParser.add_section by validating if the section name is
+        a string."""
+        self._validate_value_types(section=section)
+        super().add_section(section)
+
+
+class SafeConfigParser(ConfigParser):
+    """ConfigParser alias for backwards compatibility purposes."""
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        warnings.warn(
+            "The SafeConfigParser class has been renamed to ConfigParser "
+            "in Python 3.2. This alias will be removed in future versions."
+            " Use ConfigParser directly instead.",
+            DeprecationWarning, stacklevel=2
+        )
+
+
+class SectionProxy(MutableMapping):
+    """A proxy for a single section from a parser."""
+
+    def __init__(self, parser, name):
+        """Creates a view on a section of the specified `name` in `parser`."""
+        self._parser = parser
+        self._name = name
+        for conv in parser.converters:
+            key = 'get' + conv
+            getter = functools.partial(self.get, _impl=getattr(parser, key))
+            setattr(self, key, getter)
+
+    def __repr__(self):
+        return '<Section: {}>'.format(self._name)
+
+    def __getitem__(self, key):
+        if not self._parser.has_option(self._name, key):
+            raise KeyError(key)
+        return self._parser.get(self._name, key)
+
+    def __setitem__(self, key, value):
+        self._parser._validate_value_types(option=key, value=value)
+        return self._parser.set(self._name, key, value)
+
+    def __delitem__(self, key):
+        if not (self._parser.has_option(self._name, key) and
+                self._parser.remove_option(self._name, key)):
+            raise KeyError(key)
+
+    def __contains__(self, key):
+        return self._parser.has_option(self._name, key)
+
+    def __len__(self):
+        return len(self._options())
+
+    def __iter__(self):
+        return self._options().__iter__()
+
+    def _options(self):
+        if self._name != self._parser.default_section:
+            return self._parser.options(self._name)
+        else:
+            return self._parser.defaults()
+
+    @property
+    def parser(self):
+        # The parser object of the proxy is read-only.
+        return self._parser
+
+    @property
+    def name(self):
+        # The name of the section on a proxy is read-only.
+        return self._name
+
+    def get(self, option, fallback=None, *, raw=False, vars=None,
+            _impl=None, **kwargs):
+        """Get an option value.
+
+        Unless `fallback` is provided, `None` will be returned if the option
+        is not found.
+
+        """
+        # If `_impl` is provided, it should be a getter method on the parser
+        # object that provides the desired type conversion.
+        if not _impl:
+            _impl = self._parser.get
+        return _impl(self._name, option, raw=raw, vars=vars,
+                     fallback=fallback, **kwargs)
+
+
+class ConverterMapping(MutableMapping):
+    """Enables reuse of get*() methods between the parser and section proxies.
+
+    If a parser class implements a getter directly, the value for the given
+    key will be ``None``. The presence of the converter name here enables
+    section proxies to find and use the implementation on the parser class.
+    """
+
+    GETTERCRE = re.compile(r"^get(?P<name>.+)$")
+
+    def __init__(self, parser):
+        self._parser = parser
+        self._data = {}
+        for getter in dir(self._parser):
+            m = self.GETTERCRE.match(getter)
+            if not m or not callable(getattr(self._parser, getter)):
+                continue
+            self._data[m.group('name')] = None   # See class docstring.
+
+    def __getitem__(self, key):
+        return self._data[key]
+
+    def __setitem__(self, key, value):
+        try:
+            k = 'get' + key
+        except TypeError:
+            raise ValueError('Incompatible key: {} (type: {})'
+                             ''.format(key, type(key)))
+        if k == 'get':
+            raise ValueError('Incompatible key: cannot use "" as a name')
+        self._data[key] = value
+        func = functools.partial(self._parser._get_conv, conv=value)
+        func.converter = value
+        setattr(self._parser, k, func)
+        for proxy in self._parser.values():
+            getter = functools.partial(proxy.get, _impl=func)
+            setattr(proxy, k, getter)
+
+    def __delitem__(self, key):
+        try:
+            k = 'get' + (key or None)
+        except TypeError:
+            raise KeyError(key)
+        del self._data[key]
+        for inst in itertools.chain((self._parser,), self._parser.values()):
+            try:
+                delattr(inst, k)
+            except AttributeError:
+                # don't raise since the entry was present in _data, silently
+                # clean up
+                continue
+
+    def __iter__(self):
+        return iter(self._data)
+
+    def __len__(self):
+        return len(self._data)
diff --git a/Lib/distutils/README b/Lib/distutils/README
new file mode 100644
index 000000000..408a203b8
--- /dev/null
+++ b/Lib/distutils/README
@@ -0,0 +1,13 @@
+This directory contains the Distutils package.
+
+There's a full documentation available at:
+
+    http://docs.python.org/distutils/
+
+The Distutils-SIG web page is also a good starting point:
+
+    http://www.python.org/sigs/distutils-sig/
+
+WARNING : Distutils must remain compatible with 2.3
+
+$Id$
diff --git a/Lib/distutils/__init__.py b/Lib/distutils/__init__.py
new file mode 100644
index 000000000..d823d040a
--- /dev/null
+++ b/Lib/distutils/__init__.py
@@ -0,0 +1,13 @@
+"""distutils
+
+The main package for the Python Module Distribution Utilities.  Normally
+used from a setup script as
+
+   from distutils.core import setup
+
+   setup (...)
+"""
+
+import sys
+
+__version__ = sys.version[:sys.version.index(' ')]
diff --git a/Lib/distutils/_msvccompiler.py b/Lib/distutils/_msvccompiler.py
new file mode 100644
index 000000000..30b3b4739
--- /dev/null
+++ b/Lib/distutils/_msvccompiler.py
@@ -0,0 +1,574 @@
+"""distutils._msvccompiler
+
+Contains MSVCCompiler, an implementation of the abstract CCompiler class
+for Microsoft Visual Studio 2015.
+
+The module is compatible with VS 2015 and later. You can find legacy support
+for older versions in distutils.msvc9compiler and distutils.msvccompiler.
+"""
+
+# Written by Perry Stoll
+# hacked by Robin Becker and Thomas Heller to do a better job of
+#   finding DevStudio (through the registry)
+# ported to VS 2005 and VS 2008 by Christian Heimes
+# ported to VS 2015 by Steve Dower
+
+import os
+import shutil
+import stat
+import subprocess
+import winreg
+
+from distutils.errors import DistutilsExecError, DistutilsPlatformError, \
+                             CompileError, LibError, LinkError
+from distutils.ccompiler import CCompiler, gen_lib_options
+from distutils import log
+from distutils.util import get_platform
+
+from itertools import count
+
+def _find_vc2015():
+    try:
+        key = winreg.OpenKeyEx(
+            winreg.HKEY_LOCAL_MACHINE,
+            r"Software\Microsoft\VisualStudio\SxS\VC7",
+            access=winreg.KEY_READ | winreg.KEY_WOW64_32KEY
+        )
+    except OSError:
+        log.debug("Visual C++ is not registered")
+        return None, None
+
+    best_version = 0
+    best_dir = None
+    with key:
+        for i in count():
+            try:
+                v, vc_dir, vt = winreg.EnumValue(key, i)
+            except OSError:
+                break
+            if v and vt == winreg.REG_SZ and os.path.isdir(vc_dir):
+                try:
+                    version = int(float(v))
+                except (ValueError, TypeError):
+                    continue
+                if version >= 14 and version > best_version:
+                    best_version, best_dir = version, vc_dir
+    return best_version, best_dir
+
+def _find_vc2017():
+    import _distutils_findvs
+    import threading
+
+    best_version = 0,   # tuple for full version comparisons
+    best_dir = None
+
+    # We need to call findall() on its own thread because it will
+    # initialize COM.
+    all_packages = []
+    def _getall():
+        all_packages.extend(_distutils_findvs.findall())
+    t = threading.Thread(target=_getall)
+    t.start()
+    t.join()
+
+    for name, version_str, path, packages in all_packages:
+        if 'Microsoft.VisualStudio.Component.VC.Tools.x86.x64' in packages:
+            vc_dir = os.path.join(path, 'VC', 'Auxiliary', 'Build')
+            if not os.path.isdir(vc_dir):
+                continue
+            try:
+                version = tuple(int(i) for i in version_str.split('.'))
+            except (ValueError, TypeError):
+                continue
+            if version > best_version:
+                best_version, best_dir = version, vc_dir
+    try:
+        best_version = best_version[0]
+    except IndexError:
+        best_version = None
+    return best_version, best_dir
+
+def _find_vcvarsall(plat_spec):
+    best_version, best_dir = _find_vc2017()
+    vcruntime = None
+    vcruntime_plat = 'x64' if 'amd64' in plat_spec else 'x86'
+    if best_version:
+        vcredist = os.path.join(best_dir, "..", "..", "redist", "MSVC", "**",
+            "Microsoft.VC141.CRT", "vcruntime140.dll")
+        try:
+            import glob
+            vcruntime = glob.glob(vcredist, recursive=True)[-1]
+        except (ImportError, OSError, LookupError):
+            vcruntime = None
+
+    if not best_version:
+        best_version, best_dir = _find_vc2015()
+        if best_version:
+            vcruntime = os.path.join(best_dir, 'redist', vcruntime_plat,
+                "Microsoft.VC140.CRT", "vcruntime140.dll")
+
+    if not best_version:
+        log.debug("No suitable Visual C++ version found")
+        return None, None
+
+    vcvarsall = os.path.join(best_dir, "vcvarsall.bat")
+    if not os.path.isfile(vcvarsall):
+        log.debug("%s cannot be found", vcvarsall)
+        return None, None
+
+    if not vcruntime or not os.path.isfile(vcruntime):
+        log.debug("%s cannot be found", vcruntime)
+        vcruntime = None
+
+    return vcvarsall, vcruntime
+
+def _get_vc_env(plat_spec):
+    if os.getenv("DISTUTILS_USE_SDK"):
+        return {
+            key.lower(): value
+            for key, value in os.environ.items()
+        }
+
+    vcvarsall, vcruntime = _find_vcvarsall(plat_spec)
+    if not vcvarsall:
+        raise DistutilsPlatformError("Unable to find vcvarsall.bat")
+
+    try:
+        out = subprocess.check_output(
+            'cmd /u /c "{}" {} && set'.format(vcvarsall, plat_spec),
+            stderr=subprocess.STDOUT,
+        ).decode('utf-16le', errors='replace')
+    except subprocess.CalledProcessError as exc:
+        log.error(exc.output)
+        raise DistutilsPlatformError("Error executing {}"
+                .format(exc.cmd))
+
+    env = {
+        key.lower(): value
+        for key, _, value in
+        (line.partition('=') for line in out.splitlines())
+        if key and value
+    }
+
+    if vcruntime:
+        env['py_vcruntime_redist'] = vcruntime
+    return env
+
+def _find_exe(exe, paths=None):
+    """Return path to an MSVC executable program.
+
+    Tries to find the program in several places: first, one of the
+    MSVC program search paths from the registry; next, the directories
+    in the PATH environment variable.  If any of those work, return an
+    absolute path that is known to exist.  If none of them work, just
+    return the original program name, 'exe'.
+    """
+    if not paths:
+        paths = os.getenv('path').split(os.pathsep)
+    for p in paths:
+        fn = os.path.join(os.path.abspath(p), exe)
+        if os.path.isfile(fn):
+            return fn
+    return exe
+
+# A map keyed by get_platform() return values to values accepted by
+# 'vcvarsall.bat'. Always cross-compile from x86 to work with the
+# lighter-weight MSVC installs that do not include native 64-bit tools.
+PLAT_TO_VCVARS = {
+    'win32' : 'x86',
+    'win-amd64' : 'x86_amd64',
+}
+
+# A set containing the DLLs that are guaranteed to be available for
+# all micro versions of this Python version. Known extension
+# dependencies that are not in this set will be copied to the output
+# path.
+_BUNDLED_DLLS = frozenset(['vcruntime140.dll'])
+
+class MSVCCompiler(CCompiler) :
+    """Concrete class that implements an interface to Microsoft Visual C++,
+       as defined by the CCompiler abstract class."""
+
+    compiler_type = 'msvc'
+
+    # Just set this so CCompiler's constructor doesn't barf.  We currently
+    # don't use the 'set_executables()' bureaucracy provided by CCompiler,
+    # as it really isn't necessary for this sort of single-compiler class.
+    # Would be nice to have a consistent interface with UnixCCompiler,
+    # though, so it's worth thinking about.
+    executables = {}
+
+    # Private class data (need to distinguish C from C++ source for compiler)
+    _c_extensions = ['.c']
+    _cpp_extensions = ['.cc', '.cpp', '.cxx']
+    _rc_extensions = ['.rc']
+    _mc_extensions = ['.mc']
+
+    # Needed for the filename generation methods provided by the
+    # base class, CCompiler.
+    src_extensions = (_c_extensions + _cpp_extensions +
+                      _rc_extensions + _mc_extensions)
+    res_extension = '.res'
+    obj_extension = '.obj'
+    static_lib_extension = '.lib'
+    shared_lib_extension = '.dll'
+    static_lib_format = shared_lib_format = '%s%s'
+    exe_extension = '.exe'
+
+
+    def __init__(self, verbose=0, dry_run=0, force=0):
+        CCompiler.__init__ (self, verbose, dry_run, force)
+        # target platform (.plat_name is consistent with 'bdist')
+        self.plat_name = None
+        self.initialized = False
+
+    def initialize(self, plat_name=None):
+        # multi-init means we would need to check platform same each time...
+        assert not self.initialized, "don't init multiple times"
+        if plat_name is None:
+            plat_name = get_platform()
+        # sanity check for platforms to prevent obscure errors later.
+        if plat_name not in PLAT_TO_VCVARS:
+            raise DistutilsPlatformError("--plat-name must be one of {}"
+                                         .format(tuple(PLAT_TO_VCVARS)))
+
+        # Get the vcvarsall.bat spec for the requested platform.
+        plat_spec = PLAT_TO_VCVARS[plat_name]
+
+        vc_env = _get_vc_env(plat_spec)
+        if not vc_env:
+            raise DistutilsPlatformError("Unable to find a compatible "
+                "Visual Studio installation.")
+
+        self._paths = vc_env.get('path', '')
+        paths = self._paths.split(os.pathsep)
+        self.cc = _find_exe("cl.exe", paths)
+        self.linker = _find_exe("link.exe", paths)
+        self.lib = _find_exe("lib.exe", paths)
+        self.rc = _find_exe("rc.exe", paths)   # resource compiler
+        self.mc = _find_exe("mc.exe", paths)   # message compiler
+        self.mt = _find_exe("mt.exe", paths)   # message compiler
+        self._vcruntime_redist = vc_env.get('py_vcruntime_redist', '')
+
+        for dir in vc_env.get('include', '').split(os.pathsep):
+            if dir:
+                self.add_include_dir(dir.rstrip(os.sep))
+
+        for dir in vc_env.get('lib', '').split(os.pathsep):
+            if dir:
+                self.add_library_dir(dir.rstrip(os.sep))
+
+        self.preprocess_options = None
+        # If vcruntime_redist is available, link against it dynamically. Otherwise,
+        # use /MT[d] to build statically, then switch from libucrt[d].lib to ucrt[d].lib
+        # later to dynamically link to ucrtbase but not vcruntime.
+        self.compile_options = [
+            '/nologo', '/Ox', '/W3', '/GL', '/DNDEBUG'
+        ]
+        self.compile_options.append('/MD' if self._vcruntime_redist else '/MT')
+
+        self.compile_options_debug = [
+            '/nologo', '/Od', '/MDd', '/Zi', '/W3', '/D_DEBUG'
+        ]
+
+        ldflags = [
+            '/nologo', '/INCREMENTAL:NO', '/LTCG'
+        ]
+        if not self._vcruntime_redist:
+            ldflags.extend(('/nodefaultlib:libucrt.lib', 'ucrt.lib'))
+
+        ldflags_debug = [
+            '/nologo', '/INCREMENTAL:NO', '/LTCG', '/DEBUG:FULL'
+        ]
+
+        self.ldflags_exe = [*ldflags, '/MANIFEST:EMBED,ID=1']
+        self.ldflags_exe_debug = [*ldflags_debug, '/MANIFEST:EMBED,ID=1']
+        self.ldflags_shared = [*ldflags, '/DLL', '/MANIFEST:EMBED,ID=2', '/MANIFESTUAC:NO']
+        self.ldflags_shared_debug = [*ldflags_debug, '/DLL', '/MANIFEST:EMBED,ID=2', '/MANIFESTUAC:NO']
+        self.ldflags_static = [*ldflags]
+        self.ldflags_static_debug = [*ldflags_debug]
+
+        self._ldflags = {
+            (CCompiler.EXECUTABLE, None): self.ldflags_exe,
+            (CCompiler.EXECUTABLE, False): self.ldflags_exe,
+            (CCompiler.EXECUTABLE, True): self.ldflags_exe_debug,
+            (CCompiler.SHARED_OBJECT, None): self.ldflags_shared,
+            (CCompiler.SHARED_OBJECT, False): self.ldflags_shared,
+            (CCompiler.SHARED_OBJECT, True): self.ldflags_shared_debug,
+            (CCompiler.SHARED_LIBRARY, None): self.ldflags_static,
+            (CCompiler.SHARED_LIBRARY, False): self.ldflags_static,
+            (CCompiler.SHARED_LIBRARY, True): self.ldflags_static_debug,
+        }
+
+        self.initialized = True
+
+    # -- Worker methods ------------------------------------------------
+
+    def object_filenames(self,
+                         source_filenames,
+                         strip_dir=0,
+                         output_dir=''):
+        ext_map = {
+            **{ext: self.obj_extension for ext in self.src_extensions},
+            **{ext: self.res_extension for ext in self._rc_extensions + self._mc_extensions},
+        }
+
+        output_dir = output_dir or ''
+
+        def make_out_path(p):
+            base, ext = os.path.splitext(p)
+            if strip_dir:
+                base = os.path.basename(base)
+            else:
+                _, base = os.path.splitdrive(base)
+                if base.startswith((os.path.sep, os.path.altsep)):
+                    base = base[1:]
+            try:
+                # XXX: This may produce absurdly long paths. We should check
+                # the length of the result and trim base until we fit within
+                # 260 characters.
+                return os.path.join(output_dir, base + ext_map[ext])
+            except LookupError:
+                # Better to raise an exception instead of silently continuing
+                # and later complain about sources and targets having
+                # different lengths
+                raise CompileError("Don't know how to compile {}".format(p))
+
+        return list(map(make_out_path, source_filenames))
+
+
+    def compile(self, sources,
+                output_dir=None, macros=None, include_dirs=None, debug=0,
+                extra_preargs=None, extra_postargs=None, depends=None):
+
+        if not self.initialized:
+            self.initialize()
+        compile_info = self._setup_compile(output_dir, macros, include_dirs,
+                                           sources, depends, extra_postargs)
+        macros, objects, extra_postargs, pp_opts, build = compile_info
+
+        compile_opts = extra_preargs or []
+        compile_opts.append('/c')
+        if debug:
+            compile_opts.extend(self.compile_options_debug)
+        else:
+            compile_opts.extend(self.compile_options)
+
+
+        add_cpp_opts = False
+
+        for obj in objects:
+            try:
+                src, ext = build[obj]
+            except KeyError:
+                continue
+            if debug:
+                # pass the full pathname to MSVC in debug mode,
+                # this allows the debugger to find the source file
+                # without asking the user to browse for it
+                src = os.path.abspath(src)
+
+            if ext in self._c_extensions:
+                input_opt = "/Tc" + src
+            elif ext in self._cpp_extensions:
+                input_opt = "/Tp" + src
+                add_cpp_opts = True
+            elif ext in self._rc_extensions:
+                # compile .RC to .RES file
+                input_opt = src
+                output_opt = "/fo" + obj
+                try:
+                    self.spawn([self.rc] + pp_opts + [output_opt, input_opt])
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue
+            elif ext in self._mc_extensions:
+                # Compile .MC to .RC file to .RES file.
+                #   * '-h dir' specifies the directory for the
+                #     generated include file
+                #   * '-r dir' specifies the target directory of the
+                #     generated RC file and the binary message resource
+                #     it includes
+                #
+                # For now (since there are no options to change this),
+                # we use the source-directory for the include file and
+                # the build directory for the RC file and message
+                # resources. This works at least for win32all.
+                h_dir = os.path.dirname(src)
+                rc_dir = os.path.dirname(obj)
+                try:
+                    # first compile .MC to .RC and .H file
+                    self.spawn([self.mc, '-h', h_dir, '-r', rc_dir, src])
+                    base, _ = os.path.splitext(os.path.basename (src))
+                    rc_file = os.path.join(rc_dir, base + '.rc')
+                    # then compile .RC to .RES file
+                    self.spawn([self.rc, "/fo" + obj, rc_file])
+
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue
+            else:
+                # how to handle this file?
+                raise CompileError("Don't know how to compile {} to {}"
+                                   .format(src, obj))
+
+            args = [self.cc] + compile_opts + pp_opts
+            if add_cpp_opts:
+                args.append('/EHsc')
+            args.append(input_opt)
+            args.append("/Fo" + obj)
+            args.extend(extra_postargs)
+
+            try:
+                self.spawn(args)
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+
+        return objects
+
+
+    def create_static_lib(self,
+                          objects,
+                          output_libname,
+                          output_dir=None,
+                          debug=0,
+                          target_lang=None):
+
+        if not self.initialized:
+            self.initialize()
+        objects, output_dir = self._fix_object_args(objects, output_dir)
+        output_filename = self.library_filename(output_libname,
+                                                output_dir=output_dir)
+
+        if self._need_link(objects, output_filename):
+            lib_args = objects + ['/OUT:' + output_filename]
+            if debug:
+                pass # XXX what goes here?
+            try:
+                log.debug('Executing "%s" %s', self.lib, ' '.join(lib_args))
+                self.spawn([self.lib] + lib_args)
+            except DistutilsExecError as msg:
+                raise LibError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+
+    def link(self,
+             target_desc,
+             objects,
+             output_filename,
+             output_dir=None,
+             libraries=None,
+             library_dirs=None,
+             runtime_library_dirs=None,
+             export_symbols=None,
+             debug=0,
+             extra_preargs=None,
+             extra_postargs=None,
+             build_temp=None,
+             target_lang=None):
+
+        if not self.initialized:
+            self.initialize()
+        objects, output_dir = self._fix_object_args(objects, output_dir)
+        fixed_args = self._fix_lib_args(libraries, library_dirs,
+                                        runtime_library_dirs)
+        libraries, library_dirs, runtime_library_dirs = fixed_args
+
+        if runtime_library_dirs:
+            self.warn("I don't know what to do with 'runtime_library_dirs': "
+                       + str(runtime_library_dirs))
+
+        lib_opts = gen_lib_options(self,
+                                   library_dirs, runtime_library_dirs,
+                                   libraries)
+        if output_dir is not None:
+            output_filename = os.path.join(output_dir, output_filename)
+
+        if self._need_link(objects, output_filename):
+            ldflags = self._ldflags[target_desc, debug]
+
+            export_opts = ["/EXPORT:" + sym for sym in (export_symbols or [])]
+
+            ld_args = (ldflags + lib_opts + export_opts +
+                       objects + ['/OUT:' + output_filename])
+
+            # The MSVC linker generates .lib and .exp files, which cannot be
+            # suppressed by any linker switches. The .lib files may even be
+            # needed! Make sure they are generated in the temporary build
+            # directory. Since they have different names for debug and release
+            # builds, they can go into the same directory.
+            build_temp = os.path.dirname(objects[0])
+            if export_symbols is not None:
+                (dll_name, dll_ext) = os.path.splitext(
+                    os.path.basename(output_filename))
+                implib_file = os.path.join(
+                    build_temp,
+                    self.library_filename(dll_name))
+                ld_args.append ('/IMPLIB:' + implib_file)
+
+            if extra_preargs:
+                ld_args[:0] = extra_preargs
+            if extra_postargs:
+                ld_args.extend(extra_postargs)
+
+            output_dir = os.path.dirname(os.path.abspath(output_filename))
+            self.mkpath(output_dir)
+            try:
+                log.debug('Executing "%s" %s', self.linker, ' '.join(ld_args))
+                self.spawn([self.linker] + ld_args)
+                self._copy_vcruntime(output_dir)
+            except DistutilsExecError as msg:
+                raise LinkError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+    def _copy_vcruntime(self, output_dir):
+        vcruntime = self._vcruntime_redist
+        if not vcruntime or not os.path.isfile(vcruntime):
+            return
+
+        if os.path.basename(vcruntime).lower() in _BUNDLED_DLLS:
+            return
+
+        log.debug('Copying "%s"', vcruntime)
+        vcruntime = shutil.copy(vcruntime, output_dir)
+        os.chmod(vcruntime, stat.S_IWRITE)
+
+    def spawn(self, cmd):
+        old_path = os.getenv('path')
+        try:
+            os.environ['path'] = self._paths
+            return super().spawn(cmd)
+        finally:
+            os.environ['path'] = old_path
+
+    # -- Miscellaneous methods -----------------------------------------
+    # These are all used by the 'gen_lib_options() function, in
+    # ccompiler.py.
+
+    def library_dir_option(self, dir):
+        return "/LIBPATH:" + dir
+
+    def runtime_library_dir_option(self, dir):
+        raise DistutilsPlatformError(
+              "don't know how to set runtime library search path for MSVC")
+
+    def library_option(self, lib):
+        return self.library_filename(lib)
+
+    def find_library_file(self, dirs, lib, debug=0):
+        # Prefer a debugging library if found (and requested), but deal
+        # with it if we don't have one.
+        if debug:
+            try_names = [lib + "_d", lib]
+        else:
+            try_names = [lib]
+        for dir in dirs:
+            for name in try_names:
+                libfile = os.path.join(dir, self.library_filename(name))
+                if os.path.isfile(libfile):
+                    return libfile
+        else:
+            # Oops, didn't find it in *any* of 'dirs'
+            return None
diff --git a/Lib/distutils/archive_util.py b/Lib/distutils/archive_util.py
new file mode 100644
index 000000000..b002dc3b8
--- /dev/null
+++ b/Lib/distutils/archive_util.py
@@ -0,0 +1,256 @@
+"""distutils.archive_util
+
+Utility functions for creating archive files (tarballs, zip files,
+that sort of thing)."""
+
+import os
+from warnings import warn
+import sys
+
+try:
+    import zipfile
+except ImportError:
+    zipfile = None
+
+
+from distutils.errors import DistutilsExecError
+from distutils.spawn import spawn
+from distutils.dir_util import mkpath
+from distutils import log
+
+try:
+    from pwd import getpwnam
+except ImportError:
+    getpwnam = None
+
+try:
+    from grp import getgrnam
+except ImportError:
+    getgrnam = None
+
+def _get_gid(name):
+    """Returns a gid, given a group name."""
+    if getgrnam is None or name is None:
+        return None
+    try:
+        result = getgrnam(name)
+    except KeyError:
+        result = None
+    if result is not None:
+        return result[2]
+    return None
+
+def _get_uid(name):
+    """Returns an uid, given a user name."""
+    if getpwnam is None or name is None:
+        return None
+    try:
+        result = getpwnam(name)
+    except KeyError:
+        result = None
+    if result is not None:
+        return result[2]
+    return None
+
+def make_tarball(base_name, base_dir, compress="gzip", verbose=0, dry_run=0,
+                 owner=None, group=None):
+    """Create a (possibly compressed) tar file from all the files under
+    'base_dir'.
+
+    'compress' must be "gzip" (the default), "bzip2", "xz", "compress", or
+    None.  ("compress" will be deprecated in Python 3.2)
+
+    'owner' and 'group' can be used to define an owner and a group for the
+    archive that is being built. If not provided, the current owner and group
+    will be used.
+
+    The output tar file will be named 'base_dir' +  ".tar", possibly plus
+    the appropriate compression extension (".gz", ".bz2", ".xz" or ".Z").
+
+    Returns the output filename.
+    """
+    tar_compression = {'gzip': 'gz', 'bzip2': 'bz2', 'xz': 'xz', None: '',
+                       'compress': ''}
+    compress_ext = {'gzip': '.gz', 'bzip2': '.bz2', 'xz': '.xz',
+                    'compress': '.Z'}
+
+    # flags for compression program, each element of list will be an argument
+    if compress is not None and compress not in compress_ext.keys():
+        raise ValueError(
+              "bad value for 'compress': must be None, 'gzip', 'bzip2', "
+              "'xz' or 'compress'")
+
+    archive_name = base_name + '.tar'
+    if compress != 'compress':
+        archive_name += compress_ext.get(compress, '')
+
+    mkpath(os.path.dirname(archive_name), dry_run=dry_run)
+
+    # creating the tarball
+    import tarfile  # late import so Python build itself doesn't break
+
+    log.info('Creating tar archive')
+
+    uid = _get_uid(owner)
+    gid = _get_gid(group)
+
+    def _set_uid_gid(tarinfo):
+        if gid is not None:
+            tarinfo.gid = gid
+            tarinfo.gname = group
+        if uid is not None:
+            tarinfo.uid = uid
+            tarinfo.uname = owner
+        return tarinfo
+
+    if not dry_run:
+        tar = tarfile.open(archive_name, 'w|%s' % tar_compression[compress])
+        try:
+            tar.add(base_dir, filter=_set_uid_gid)
+        finally:
+            tar.close()
+
+    # compression using `compress`
+    if compress == 'compress':
+        warn("'compress' will be deprecated.", PendingDeprecationWarning)
+        # the option varies depending on the platform
+        compressed_name = archive_name + compress_ext[compress]
+        if sys.platform == 'win32':
+            cmd = [compress, archive_name, compressed_name]
+        else:
+            cmd = [compress, '-f', archive_name]
+        spawn(cmd, dry_run=dry_run)
+        return compressed_name
+
+    return archive_name
+
+def make_zipfile(base_name, base_dir, verbose=0, dry_run=0):
+    """Create a zip file from all the files under 'base_dir'.
+
+    The output zip file will be named 'base_name' + ".zip".  Uses either the
+    "zipfile" Python module (if available) or the InfoZIP "zip" utility
+    (if installed and found on the default search path).  If neither tool is
+    available, raises DistutilsExecError.  Returns the name of the output zip
+    file.
+    """
+    zip_filename = base_name + ".zip"
+    mkpath(os.path.dirname(zip_filename), dry_run=dry_run)
+
+    # If zipfile module is not available, try spawning an external
+    # 'zip' command.
+    if zipfile is None:
+        if verbose:
+            zipoptions = "-r"
+        else:
+            zipoptions = "-rq"
+
+        try:
+            spawn(["zip", zipoptions, zip_filename, base_dir],
+                  dry_run=dry_run)
+        except DistutilsExecError:
+            # XXX really should distinguish between "couldn't find
+            # external 'zip' command" and "zip failed".
+            raise DistutilsExecError(("unable to create zip file '%s': "
+                   "could neither import the 'zipfile' module nor "
+                   "find a standalone zip utility") % zip_filename)
+
+    else:
+        log.info("creating '%s' and adding '%s' to it",
+                 zip_filename, base_dir)
+
+        if not dry_run:
+            try:
+                zip = zipfile.ZipFile(zip_filename, "w",
+                                      compression=zipfile.ZIP_DEFLATED)
+            except RuntimeError:
+                zip = zipfile.ZipFile(zip_filename, "w",
+                                      compression=zipfile.ZIP_STORED)
+
+            if base_dir != os.curdir:
+                path = os.path.normpath(os.path.join(base_dir, ''))
+                zip.write(path, path)
+                log.info("adding '%s'", path)
+            for dirpath, dirnames, filenames in os.walk(base_dir):
+                for name in dirnames:
+                    path = os.path.normpath(os.path.join(dirpath, name, ''))
+                    zip.write(path, path)
+                    log.info("adding '%s'", path)
+                for name in filenames:
+                    path = os.path.normpath(os.path.join(dirpath, name))
+                    if os.path.isfile(path):
+                        zip.write(path, path)
+                        log.info("adding '%s'", path)
+            zip.close()
+
+    return zip_filename
+
+ARCHIVE_FORMATS = {
+    'gztar': (make_tarball, [('compress', 'gzip')], "gzip'ed tar-file"),
+    'bztar': (make_tarball, [('compress', 'bzip2')], "bzip2'ed tar-file"),
+    'xztar': (make_tarball, [('compress', 'xz')], "xz'ed tar-file"),
+    'ztar':  (make_tarball, [('compress', 'compress')], "compressed tar file"),
+    'tar':   (make_tarball, [('compress', None)], "uncompressed tar file"),
+    'zip':   (make_zipfile, [],"ZIP file")
+    }
+
+def check_archive_formats(formats):
+    """Returns the first format from the 'format' list that is unknown.
+
+    If all formats are known, returns None
+    """
+    for format in formats:
+        if format not in ARCHIVE_FORMATS:
+            return format
+    return None
+
+def make_archive(base_name, format, root_dir=None, base_dir=None, verbose=0,
+                 dry_run=0, owner=None, group=None):
+    """Create an archive file (eg. zip or tar).
+
+    'base_name' is the name of the file to create, minus any format-specific
+    extension; 'format' is the archive format: one of "zip", "tar", "gztar",
+    "bztar", "xztar", or "ztar".
+
+    'root_dir' is a directory that will be the root directory of the
+    archive; ie. we typically chdir into 'root_dir' before creating the
+    archive.  'base_dir' is the directory where we start archiving from;
+    ie. 'base_dir' will be the common prefix of all files and
+    directories in the archive.  'root_dir' and 'base_dir' both default
+    to the current directory.  Returns the name of the archive file.
+
+    'owner' and 'group' are used when creating a tar archive. By default,
+    uses the current owner and group.
+    """
+    save_cwd = os.getcwd()
+    if root_dir is not None:
+        log.debug("changing into '%s'", root_dir)
+        base_name = os.path.abspath(base_name)
+        if not dry_run:
+            os.chdir(root_dir)
+
+    if base_dir is None:
+        base_dir = os.curdir
+
+    kwargs = {'dry_run': dry_run}
+
+    try:
+        format_info = ARCHIVE_FORMATS[format]
+    except KeyError:
+        raise ValueError("unknown archive format '%s'" % format)
+
+    func = format_info[0]
+    for arg, val in format_info[1]:
+        kwargs[arg] = val
+
+    if format != 'zip':
+        kwargs['owner'] = owner
+        kwargs['group'] = group
+
+    try:
+        filename = func(base_name, base_dir, **kwargs)
+    finally:
+        if root_dir is not None:
+            log.debug("changing back to '%s'", save_cwd)
+            os.chdir(save_cwd)
+
+    return filename
diff --git a/Lib/distutils/bcppcompiler.py b/Lib/distutils/bcppcompiler.py
new file mode 100644
index 000000000..9f4c432d9
--- /dev/null
+++ b/Lib/distutils/bcppcompiler.py
@@ -0,0 +1,393 @@
+"""distutils.bcppcompiler
+
+Contains BorlandCCompiler, an implementation of the abstract CCompiler class
+for the Borland C++ compiler.
+"""
+
+# This implementation by Lyle Johnson, based on the original msvccompiler.py
+# module and using the directions originally published by Gordon Williams.
+
+# XXX looks like there's a LOT of overlap between these two classes:
+# someone should sit down and factor out the common code as
+# WindowsCCompiler!  --GPW
+
+
+import os
+from distutils.errors import \
+     DistutilsExecError, DistutilsPlatformError, \
+     CompileError, LibError, LinkError, UnknownFileError
+from distutils.ccompiler import \
+     CCompiler, gen_preprocess_options, gen_lib_options
+from distutils.file_util import write_file
+from distutils.dep_util import newer
+from distutils import log
+
+class BCPPCompiler(CCompiler) :
+    """Concrete class that implements an interface to the Borland C/C++
+    compiler, as defined by the CCompiler abstract class.
+    """
+
+    compiler_type = 'bcpp'
+
+    # Just set this so CCompiler's constructor doesn't barf.  We currently
+    # don't use the 'set_executables()' bureaucracy provided by CCompiler,
+    # as it really isn't necessary for this sort of single-compiler class.
+    # Would be nice to have a consistent interface with UnixCCompiler,
+    # though, so it's worth thinking about.
+    executables = {}
+
+    # Private class data (need to distinguish C from C++ source for compiler)
+    _c_extensions = ['.c']
+    _cpp_extensions = ['.cc', '.cpp', '.cxx']
+
+    # Needed for the filename generation methods provided by the
+    # base class, CCompiler.
+    src_extensions = _c_extensions + _cpp_extensions
+    obj_extension = '.obj'
+    static_lib_extension = '.lib'
+    shared_lib_extension = '.dll'
+    static_lib_format = shared_lib_format = '%s%s'
+    exe_extension = '.exe'
+
+
+    def __init__ (self,
+                  verbose=0,
+                  dry_run=0,
+                  force=0):
+
+        CCompiler.__init__ (self, verbose, dry_run, force)
+
+        # These executables are assumed to all be in the path.
+        # Borland doesn't seem to use any special registry settings to
+        # indicate their installation locations.
+
+        self.cc = "bcc32.exe"
+        self.linker = "ilink32.exe"
+        self.lib = "tlib.exe"
+
+        self.preprocess_options = None
+        self.compile_options = ['/tWM', '/O2', '/q', '/g0']
+        self.compile_options_debug = ['/tWM', '/Od', '/q', '/g0']
+
+        self.ldflags_shared = ['/Tpd', '/Gn', '/q', '/x']
+        self.ldflags_shared_debug = ['/Tpd', '/Gn', '/q', '/x']
+        self.ldflags_static = []
+        self.ldflags_exe = ['/Gn', '/q', '/x']
+        self.ldflags_exe_debug = ['/Gn', '/q', '/x','/r']
+
+
+    # -- Worker methods ------------------------------------------------
+
+    def compile(self, sources,
+                output_dir=None, macros=None, include_dirs=None, debug=0,
+                extra_preargs=None, extra_postargs=None, depends=None):
+
+        macros, objects, extra_postargs, pp_opts, build = \
+                self._setup_compile(output_dir, macros, include_dirs, sources,
+                                    depends, extra_postargs)
+        compile_opts = extra_preargs or []
+        compile_opts.append ('-c')
+        if debug:
+            compile_opts.extend (self.compile_options_debug)
+        else:
+            compile_opts.extend (self.compile_options)
+
+        for obj in objects:
+            try:
+                src, ext = build[obj]
+            except KeyError:
+                continue
+            # XXX why do the normpath here?
+            src = os.path.normpath(src)
+            obj = os.path.normpath(obj)
+            # XXX _setup_compile() did a mkpath() too but before the normpath.
+            # Is it possible to skip the normpath?
+            self.mkpath(os.path.dirname(obj))
+
+            if ext == '.res':
+                # This is already a binary file -- skip it.
+                continue # the 'for' loop
+            if ext == '.rc':
+                # This needs to be compiled to a .res file -- do it now.
+                try:
+                    self.spawn (["brcc32", "-fo", obj, src])
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue # the 'for' loop
+
+            # The next two are both for the real compiler.
+            if ext in self._c_extensions:
+                input_opt = ""
+            elif ext in self._cpp_extensions:
+                input_opt = "-P"
+            else:
+                # Unknown file type -- no extra options.  The compiler
+                # will probably fail, but let it just in case this is a
+                # file the compiler recognizes even if we don't.
+                input_opt = ""
+
+            output_opt = "-o" + obj
+
+            # Compiler command line syntax is: "bcc32 [options] file(s)".
+            # Note that the source file names must appear at the end of
+            # the command line.
+            try:
+                self.spawn ([self.cc] + compile_opts + pp_opts +
+                            [input_opt, output_opt] +
+                            extra_postargs + [src])
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+
+        return objects
+
+    # compile ()
+
+
+    def create_static_lib (self,
+                           objects,
+                           output_libname,
+                           output_dir=None,
+                           debug=0,
+                           target_lang=None):
+
+        (objects, output_dir) = self._fix_object_args (objects, output_dir)
+        output_filename = \
+            self.library_filename (output_libname, output_dir=output_dir)
+
+        if self._need_link (objects, output_filename):
+            lib_args = [output_filename, '/u'] + objects
+            if debug:
+                pass                    # XXX what goes here?
+            try:
+                self.spawn ([self.lib] + lib_args)
+            except DistutilsExecError as msg:
+                raise LibError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+    # create_static_lib ()
+
+
+    def link (self,
+              target_desc,
+              objects,
+              output_filename,
+              output_dir=None,
+              libraries=None,
+              library_dirs=None,
+              runtime_library_dirs=None,
+              export_symbols=None,
+              debug=0,
+              extra_preargs=None,
+              extra_postargs=None,
+              build_temp=None,
+              target_lang=None):
+
+        # XXX this ignores 'build_temp'!  should follow the lead of
+        # msvccompiler.py
+
+        (objects, output_dir) = self._fix_object_args (objects, output_dir)
+        (libraries, library_dirs, runtime_library_dirs) = \
+            self._fix_lib_args (libraries, library_dirs, runtime_library_dirs)
+
+        if runtime_library_dirs:
+            log.warn("I don't know what to do with 'runtime_library_dirs': %s",
+                     str(runtime_library_dirs))
+
+        if output_dir is not None:
+            output_filename = os.path.join (output_dir, output_filename)
+
+        if self._need_link (objects, output_filename):
+
+            # Figure out linker args based on type of target.
+            if target_desc == CCompiler.EXECUTABLE:
+                startup_obj = 'c0w32'
+                if debug:
+                    ld_args = self.ldflags_exe_debug[:]
+                else:
+                    ld_args = self.ldflags_exe[:]
+            else:
+                startup_obj = 'c0d32'
+                if debug:
+                    ld_args = self.ldflags_shared_debug[:]
+                else:
+                    ld_args = self.ldflags_shared[:]
+
+
+            # Create a temporary exports file for use by the linker
+            if export_symbols is None:
+                def_file = ''
+            else:
+                head, tail = os.path.split (output_filename)
+                modname, ext = os.path.splitext (tail)
+                temp_dir = os.path.dirname(objects[0]) # preserve tree structure
+                def_file = os.path.join (temp_dir, '%s.def' % modname)
+                contents = ['EXPORTS']
+                for sym in (export_symbols or []):
+                    contents.append('  %s=_%s' % (sym, sym))
+                self.execute(write_file, (def_file, contents),
+                             "writing %s" % def_file)
+
+            # Borland C++ has problems with '/' in paths
+            objects2 = map(os.path.normpath, objects)
+            # split objects in .obj and .res files
+            # Borland C++ needs them at different positions in the command line
+            objects = [startup_obj]
+            resources = []
+            for file in objects2:
+                (base, ext) = os.path.splitext(os.path.normcase(file))
+                if ext == '.res':
+                    resources.append(file)
+                else:
+                    objects.append(file)
+
+
+            for l in library_dirs:
+                ld_args.append("/L%s" % os.path.normpath(l))
+            ld_args.append("/L.") # we sometimes use relative paths
+
+            # list of object files
+            ld_args.extend(objects)
+
+            # XXX the command-line syntax for Borland C++ is a bit wonky;
+            # certain filenames are jammed together in one big string, but
+            # comma-delimited.  This doesn't mesh too well with the
+            # Unix-centric attitude (with a DOS/Windows quoting hack) of
+            # 'spawn()', so constructing the argument list is a bit
+            # awkward.  Note that doing the obvious thing and jamming all
+            # the filenames and commas into one argument would be wrong,
+            # because 'spawn()' would quote any filenames with spaces in
+            # them.  Arghghh!.  Apparently it works fine as coded...
+
+            # name of dll/exe file
+            ld_args.extend([',',output_filename])
+            # no map file and start libraries
+            ld_args.append(',,')
+
+            for lib in libraries:
+                # see if we find it and if there is a bcpp specific lib
+                # (xxx_bcpp.lib)
+                libfile = self.find_library_file(library_dirs, lib, debug)
+                if libfile is None:
+                    ld_args.append(lib)
+                    # probably a BCPP internal library -- don't warn
+                else:
+                    # full name which prefers bcpp_xxx.lib over xxx.lib
+                    ld_args.append(libfile)
+
+            # some default libraries
+            ld_args.append ('import32')
+            ld_args.append ('cw32mt')
+
+            # def file for export symbols
+            ld_args.extend([',',def_file])
+            # add resource files
+            ld_args.append(',')
+            ld_args.extend(resources)
+
+
+            if extra_preargs:
+                ld_args[:0] = extra_preargs
+            if extra_postargs:
+                ld_args.extend(extra_postargs)
+
+            self.mkpath (os.path.dirname (output_filename))
+            try:
+                self.spawn ([self.linker] + ld_args)
+            except DistutilsExecError as msg:
+                raise LinkError(msg)
+
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+    # link ()
+
+    # -- Miscellaneous methods -----------------------------------------
+
+
+    def find_library_file (self, dirs, lib, debug=0):
+        # List of effective library names to try, in order of preference:
+        # xxx_bcpp.lib is better than xxx.lib
+        # and xxx_d.lib is better than xxx.lib if debug is set
+        #
+        # The "_bcpp" suffix is to handle a Python installation for people
+        # with multiple compilers (primarily Distutils hackers, I suspect
+        # ;-).  The idea is they'd have one static library for each
+        # compiler they care about, since (almost?) every Windows compiler
+        # seems to have a different format for static libraries.
+        if debug:
+            dlib = (lib + "_d")
+            try_names = (dlib + "_bcpp", lib + "_bcpp", dlib, lib)
+        else:
+            try_names = (lib + "_bcpp", lib)
+
+        for dir in dirs:
+            for name in try_names:
+                libfile = os.path.join(dir, self.library_filename(name))
+                if os.path.exists(libfile):
+                    return libfile
+        else:
+            # Oops, didn't find it in *any* of 'dirs'
+            return None
+
+    # overwrite the one from CCompiler to support rc and res-files
+    def object_filenames (self,
+                          source_filenames,
+                          strip_dir=0,
+                          output_dir=''):
+        if output_dir is None: output_dir = ''
+        obj_names = []
+        for src_name in source_filenames:
+            # use normcase to make sure '.rc' is really '.rc' and not '.RC'
+            (base, ext) = os.path.splitext (os.path.normcase(src_name))
+            if ext not in (self.src_extensions + ['.rc','.res']):
+                raise UnknownFileError("unknown file type '%s' (from '%s')" % \
+                      (ext, src_name))
+            if strip_dir:
+                base = os.path.basename (base)
+            if ext == '.res':
+                # these can go unchanged
+                obj_names.append (os.path.join (output_dir, base + ext))
+            elif ext == '.rc':
+                # these need to be compiled to .res-files
+                obj_names.append (os.path.join (output_dir, base + '.res'))
+            else:
+                obj_names.append (os.path.join (output_dir,
+                                            base + self.obj_extension))
+        return obj_names
+
+    # object_filenames ()
+
+    def preprocess (self,
+                    source,
+                    output_file=None,
+                    macros=None,
+                    include_dirs=None,
+                    extra_preargs=None,
+                    extra_postargs=None):
+
+        (_, macros, include_dirs) = \
+            self._fix_compile_args(None, macros, include_dirs)
+        pp_opts = gen_preprocess_options(macros, include_dirs)
+        pp_args = ['cpp32.exe'] + pp_opts
+        if output_file is not None:
+            pp_args.append('-o' + output_file)
+        if extra_preargs:
+            pp_args[:0] = extra_preargs
+        if extra_postargs:
+            pp_args.extend(extra_postargs)
+        pp_args.append(source)
+
+        # We need to preprocess: either we're being forced to, or the
+        # source file is newer than the target (or the target doesn't
+        # exist).
+        if self.force or output_file is None or newer(source, output_file):
+            if output_file:
+                self.mkpath(os.path.dirname(output_file))
+            try:
+                self.spawn(pp_args)
+            except DistutilsExecError as msg:
+                print(msg)
+                raise CompileError(msg)
+
+    # preprocess()
diff --git a/Lib/distutils/ccompiler.py b/Lib/distutils/ccompiler.py
new file mode 100644
index 000000000..b71d1d39b
--- /dev/null
+++ b/Lib/distutils/ccompiler.py
@@ -0,0 +1,1115 @@
+"""distutils.ccompiler
+
+Contains CCompiler, an abstract base class that defines the interface
+for the Distutils compiler abstraction model."""
+
+import sys, os, re
+from distutils.errors import *
+from distutils.spawn import spawn
+from distutils.file_util import move_file
+from distutils.dir_util import mkpath
+from distutils.dep_util import newer_pairwise, newer_group
+from distutils.util import split_quoted, execute
+from distutils import log
+
+class CCompiler:
+    """Abstract base class to define the interface that must be implemented
+    by real compiler classes.  Also has some utility methods used by
+    several compiler classes.
+
+    The basic idea behind a compiler abstraction class is that each
+    instance can be used for all the compile/link steps in building a
+    single project.  Thus, attributes common to all of those compile and
+    link steps -- include directories, macros to define, libraries to link
+    against, etc. -- are attributes of the compiler instance.  To allow for
+    variability in how individual files are treated, most of those
+    attributes may be varied on a per-compilation or per-link basis.
+    """
+
+    # 'compiler_type' is a class attribute that identifies this class.  It
+    # keeps code that wants to know what kind of compiler it's dealing with
+    # from having to import all possible compiler classes just to do an
+    # 'isinstance'.  In concrete CCompiler subclasses, 'compiler_type'
+    # should really, really be one of the keys of the 'compiler_class'
+    # dictionary (see below -- used by the 'new_compiler()' factory
+    # function) -- authors of new compiler interface classes are
+    # responsible for updating 'compiler_class'!
+    compiler_type = None
+
+    # XXX things not handled by this compiler abstraction model:
+    #   * client can't provide additional options for a compiler,
+    #     e.g. warning, optimization, debugging flags.  Perhaps this
+    #     should be the domain of concrete compiler abstraction classes
+    #     (UnixCCompiler, MSVCCompiler, etc.) -- or perhaps the base
+    #     class should have methods for the common ones.
+    #   * can't completely override the include or library searchg
+    #     path, ie. no "cc -I -Idir1 -Idir2" or "cc -L -Ldir1 -Ldir2".
+    #     I'm not sure how widely supported this is even by Unix
+    #     compilers, much less on other platforms.  And I'm even less
+    #     sure how useful it is; maybe for cross-compiling, but
+    #     support for that is a ways off.  (And anyways, cross
+    #     compilers probably have a dedicated binary with the
+    #     right paths compiled in.  I hope.)
+    #   * can't do really freaky things with the library list/library
+    #     dirs, e.g. "-Ldir1 -lfoo -Ldir2 -lfoo" to link against
+    #     different versions of libfoo.a in different locations.  I
+    #     think this is useless without the ability to null out the
+    #     library search path anyways.
+
+
+    # Subclasses that rely on the standard filename generation methods
+    # implemented below should override these; see the comment near
+    # those methods ('object_filenames()' et. al.) for details:
+    src_extensions = None               # list of strings
+    obj_extension = None                # string
+    static_lib_extension = None
+    shared_lib_extension = None         # string
+    static_lib_format = None            # format string
+    shared_lib_format = None            # prob. same as static_lib_format
+    exe_extension = None                # string
+
+    # Default language settings. language_map is used to detect a source
+    # file or Extension target language, checking source filenames.
+    # language_order is used to detect the language precedence, when deciding
+    # what language to use when mixing source types. For example, if some
+    # extension has two files with ".c" extension, and one with ".cpp", it
+    # is still linked as c++.
+    language_map = {".c"   : "c",
+                    ".cc"  : "c++",
+                    ".cpp" : "c++",
+                    ".cxx" : "c++",
+                    ".m"   : "objc",
+                   }
+    language_order = ["c++", "objc", "c"]
+
+    def __init__(self, verbose=0, dry_run=0, force=0):
+        self.dry_run = dry_run
+        self.force = force
+        self.verbose = verbose
+
+        # 'output_dir': a common output directory for object, library,
+        # shared object, and shared library files
+        self.output_dir = None
+
+        # 'macros': a list of macro definitions (or undefinitions).  A
+        # macro definition is a 2-tuple (name, value), where the value is
+        # either a string or None (no explicit value).  A macro
+        # undefinition is a 1-tuple (name,).
+        self.macros = []
+
+        # 'include_dirs': a list of directories to search for include files
+        self.include_dirs = []
+
+        # 'libraries': a list of libraries to include in any link
+        # (library names, not filenames: eg. "foo" not "libfoo.a")
+        self.libraries = []
+
+        # 'library_dirs': a list of directories to search for libraries
+        self.library_dirs = []
+
+        # 'runtime_library_dirs': a list of directories to search for
+        # shared libraries/objects at runtime
+        self.runtime_library_dirs = []
+
+        # 'objects': a list of object files (or similar, such as explicitly
+        # named library files) to include on any link
+        self.objects = []
+
+        for key in self.executables.keys():
+            self.set_executable(key, self.executables[key])
+
+    def set_executables(self, **kwargs):
+        """Define the executables (and options for them) that will be run
+        to perform the various stages of compilation.  The exact set of
+        executables that may be specified here depends on the compiler
+        class (via the 'executables' class attribute), but most will have:
+          compiler      the C/C++ compiler
+          linker_so     linker used to create shared objects and libraries
+          linker_exe    linker used to create binary executables
+          archiver      static library creator
+
+        On platforms with a command-line (Unix, DOS/Windows), each of these
+        is a string that will be split into executable name and (optional)
+        list of arguments.  (Splitting the string is done similarly to how
+        Unix shells operate: words are delimited by spaces, but quotes and
+        backslashes can override this.  See
+        'distutils.util.split_quoted()'.)
+        """
+
+        # Note that some CCompiler implementation classes will define class
+        # attributes 'cpp', 'cc', etc. with hard-coded executable names;
+        # this is appropriate when a compiler class is for exactly one
+        # compiler/OS combination (eg. MSVCCompiler).  Other compiler
+        # classes (UnixCCompiler, in particular) are driven by information
+        # discovered at run-time, since there are many different ways to do
+        # basically the same things with Unix C compilers.
+
+        for key in kwargs:
+            if key not in self.executables:
+                raise ValueError("unknown executable '%s' for class %s" %
+                      (key, self.__class__.__name__))
+            self.set_executable(key, kwargs[key])
+
+    def set_executable(self, key, value):
+        if isinstance(value, str):
+            setattr(self, key, split_quoted(value))
+        else:
+            setattr(self, key, value)
+
+    def _find_macro(self, name):
+        i = 0
+        for defn in self.macros:
+            if defn[0] == name:
+                return i
+            i += 1
+        return None
+
+    def _check_macro_definitions(self, definitions):
+        """Ensures that every element of 'definitions' is a valid macro
+        definition, ie. either (name,value) 2-tuple or a (name,) tuple.  Do
+        nothing if all definitions are OK, raise TypeError otherwise.
+        """
+        for defn in definitions:
+            if not (isinstance(defn, tuple) and
+                    (len(defn) in (1, 2) and
+                      (isinstance (defn[1], str) or defn[1] is None)) and
+                    isinstance (defn[0], str)):
+                raise TypeError(("invalid macro definition '%s': " % defn) + \
+                      "must be tuple (string,), (string, string), or " + \
+                      "(string, None)")
+
+
+    # -- Bookkeeping methods -------------------------------------------
+
+    def define_macro(self, name, value=None):
+        """Define a preprocessor macro for all compilations driven by this
+        compiler object.  The optional parameter 'value' should be a
+        string; if it is not supplied, then the macro will be defined
+        without an explicit value and the exact outcome depends on the
+        compiler used (XXX true? does ANSI say anything about this?)
+        """
+        # Delete from the list of macro definitions/undefinitions if
+        # already there (so that this one will take precedence).
+        i = self._find_macro (name)
+        if i is not None:
+            del self.macros[i]
+
+        self.macros.append((name, value))
+
+    def undefine_macro(self, name):
+        """Undefine a preprocessor macro for all compilations driven by
+        this compiler object.  If the same macro is defined by
+        'define_macro()' and undefined by 'undefine_macro()' the last call
+        takes precedence (including multiple redefinitions or
+        undefinitions).  If the macro is redefined/undefined on a
+        per-compilation basis (ie. in the call to 'compile()'), then that
+        takes precedence.
+        """
+        # Delete from the list of macro definitions/undefinitions if
+        # already there (so that this one will take precedence).
+        i = self._find_macro (name)
+        if i is not None:
+            del self.macros[i]
+
+        undefn = (name,)
+        self.macros.append(undefn)
+
+    def add_include_dir(self, dir):
+        """Add 'dir' to the list of directories that will be searched for
+        header files.  The compiler is instructed to search directories in
+        the order in which they are supplied by successive calls to
+        'add_include_dir()'.
+        """
+        self.include_dirs.append(dir)
+
+    def set_include_dirs(self, dirs):
+        """Set the list of directories that will be searched to 'dirs' (a
+        list of strings).  Overrides any preceding calls to
+        'add_include_dir()'; subsequence calls to 'add_include_dir()' add
+        to the list passed to 'set_include_dirs()'.  This does not affect
+        any list of standard include directories that the compiler may
+        search by default.
+        """
+        self.include_dirs = dirs[:]
+
+    def add_library(self, libname):
+        """Add 'libname' to the list of libraries that will be included in
+        all links driven by this compiler object.  Note that 'libname'
+        should *not* be the name of a file containing a library, but the
+        name of the library itself: the actual filename will be inferred by
+        the linker, the compiler, or the compiler class (depending on the
+        platform).
+
+        The linker will be instructed to link against libraries in the
+        order they were supplied to 'add_library()' and/or
+        'set_libraries()'.  It is perfectly valid to duplicate library
+        names; the linker will be instructed to link against libraries as
+        many times as they are mentioned.
+        """
+        self.libraries.append(libname)
+
+    def set_libraries(self, libnames):
+        """Set the list of libraries to be included in all links driven by
+        this compiler object to 'libnames' (a list of strings).  This does
+        not affect any standard system libraries that the linker may
+        include by default.
+        """
+        self.libraries = libnames[:]
+
+    def add_library_dir(self, dir):
+        """Add 'dir' to the list of directories that will be searched for
+        libraries specified to 'add_library()' and 'set_libraries()'.  The
+        linker will be instructed to search for libraries in the order they
+        are supplied to 'add_library_dir()' and/or 'set_library_dirs()'.
+        """
+        self.library_dirs.append(dir)
+
+    def set_library_dirs(self, dirs):
+        """Set the list of library search directories to 'dirs' (a list of
+        strings).  This does not affect any standard library search path
+        that the linker may search by default.
+        """
+        self.library_dirs = dirs[:]
+
+    def add_runtime_library_dir(self, dir):
+        """Add 'dir' to the list of directories that will be searched for
+        shared libraries at runtime.
+        """
+        self.runtime_library_dirs.append(dir)
+
+    def set_runtime_library_dirs(self, dirs):
+        """Set the list of directories to search for shared libraries at
+        runtime to 'dirs' (a list of strings).  This does not affect any
+        standard search path that the runtime linker may search by
+        default.
+        """
+        self.runtime_library_dirs = dirs[:]
+
+    def add_link_object(self, object):
+        """Add 'object' to the list of object files (or analogues, such as
+        explicitly named library files or the output of "resource
+        compilers") to be included in every link driven by this compiler
+        object.
+        """
+        self.objects.append(object)
+
+    def set_link_objects(self, objects):
+        """Set the list of object files (or analogues) to be included in
+        every link to 'objects'.  This does not affect any standard object
+        files that the linker may include by default (such as system
+        libraries).
+        """
+        self.objects = objects[:]
+
+
+    # -- Private utility methods --------------------------------------
+    # (here for the convenience of subclasses)
+
+    # Helper method to prep compiler in subclass compile() methods
+
+    def _setup_compile(self, outdir, macros, incdirs, sources, depends,
+                       extra):
+        """Process arguments and decide which source files to compile."""
+        if outdir is None:
+            outdir = self.output_dir
+        elif not isinstance(outdir, str):
+            raise TypeError("'output_dir' must be a string or None")
+
+        if macros is None:
+            macros = self.macros
+        elif isinstance(macros, list):
+            macros = macros + (self.macros or [])
+        else:
+            raise TypeError("'macros' (if supplied) must be a list of tuples")
+
+        if incdirs is None:
+            incdirs = self.include_dirs
+        elif isinstance(incdirs, (list, tuple)):
+            incdirs = list(incdirs) + (self.include_dirs or [])
+        else:
+            raise TypeError(
+                  "'include_dirs' (if supplied) must be a list of strings")
+
+        if extra is None:
+            extra = []
+
+        # Get the list of expected output (object) files
+        objects = self.object_filenames(sources, strip_dir=0,
+                                        output_dir=outdir)
+        assert len(objects) == len(sources)
+
+        pp_opts = gen_preprocess_options(macros, incdirs)
+
+        build = {}
+        for i in range(len(sources)):
+            src = sources[i]
+            obj = objects[i]
+            ext = os.path.splitext(src)[1]
+            self.mkpath(os.path.dirname(obj))
+            build[obj] = (src, ext)
+
+        return macros, objects, extra, pp_opts, build
+
+    def _get_cc_args(self, pp_opts, debug, before):
+        # works for unixccompiler, cygwinccompiler
+        cc_args = pp_opts + ['-c']
+        if debug:
+            cc_args[:0] = ['-g']
+        if before:
+            cc_args[:0] = before
+        return cc_args
+
+    def _fix_compile_args(self, output_dir, macros, include_dirs):
+        """Typecheck and fix-up some of the arguments to the 'compile()'
+        method, and return fixed-up values.  Specifically: if 'output_dir'
+        is None, replaces it with 'self.output_dir'; ensures that 'macros'
+        is a list, and augments it with 'self.macros'; ensures that
+        'include_dirs' is a list, and augments it with 'self.include_dirs'.
+        Guarantees that the returned values are of the correct type,
+        i.e. for 'output_dir' either string or None, and for 'macros' and
+        'include_dirs' either list or None.
+        """
+        if output_dir is None:
+            output_dir = self.output_dir
+        elif not isinstance(output_dir, str):
+            raise TypeError("'output_dir' must be a string or None")
+
+        if macros is None:
+            macros = self.macros
+        elif isinstance(macros, list):
+            macros = macros + (self.macros or [])
+        else:
+            raise TypeError("'macros' (if supplied) must be a list of tuples")
+
+        if include_dirs is None:
+            include_dirs = self.include_dirs
+        elif isinstance(include_dirs, (list, tuple)):
+            include_dirs = list(include_dirs) + (self.include_dirs or [])
+        else:
+            raise TypeError(
+                  "'include_dirs' (if supplied) must be a list of strings")
+
+        return output_dir, macros, include_dirs
+
+    def _prep_compile(self, sources, output_dir, depends=None):
+        """Decide which souce files must be recompiled.
+
+        Determine the list of object files corresponding to 'sources',
+        and figure out which ones really need to be recompiled.
+        Return a list of all object files and a dictionary telling
+        which source files can be skipped.
+        """
+        # Get the list of expected output (object) files
+        objects = self.object_filenames(sources, output_dir=output_dir)
+        assert len(objects) == len(sources)
+
+        # Return an empty dict for the "which source files can be skipped"
+        # return value to preserve API compatibility.
+        return objects, {}
+
+    def _fix_object_args(self, objects, output_dir):
+        """Typecheck and fix up some arguments supplied to various methods.
+        Specifically: ensure that 'objects' is a list; if output_dir is
+        None, replace with self.output_dir.  Return fixed versions of
+        'objects' and 'output_dir'.
+        """
+        if not isinstance(objects, (list, tuple)):
+            raise TypeError("'objects' must be a list or tuple of strings")
+        objects = list(objects)
+
+        if output_dir is None:
+            output_dir = self.output_dir
+        elif not isinstance(output_dir, str):
+            raise TypeError("'output_dir' must be a string or None")
+
+        return (objects, output_dir)
+
+    def _fix_lib_args(self, libraries, library_dirs, runtime_library_dirs):
+        """Typecheck and fix up some of the arguments supplied to the
+        'link_*' methods.  Specifically: ensure that all arguments are
+        lists, and augment them with their permanent versions
+        (eg. 'self.libraries' augments 'libraries').  Return a tuple with
+        fixed versions of all arguments.
+        """
+        if libraries is None:
+            libraries = self.libraries
+        elif isinstance(libraries, (list, tuple)):
+            libraries = list (libraries) + (self.libraries or [])
+        else:
+            raise TypeError(
+                  "'libraries' (if supplied) must be a list of strings")
+
+        if library_dirs is None:
+            library_dirs = self.library_dirs
+        elif isinstance(library_dirs, (list, tuple)):
+            library_dirs = list (library_dirs) + (self.library_dirs or [])
+        else:
+            raise TypeError(
+                  "'library_dirs' (if supplied) must be a list of strings")
+
+        if runtime_library_dirs is None:
+            runtime_library_dirs = self.runtime_library_dirs
+        elif isinstance(runtime_library_dirs, (list, tuple)):
+            runtime_library_dirs = (list(runtime_library_dirs) +
+                                    (self.runtime_library_dirs or []))
+        else:
+            raise TypeError("'runtime_library_dirs' (if supplied) "
+                            "must be a list of strings")
+
+        return (libraries, library_dirs, runtime_library_dirs)
+
+    def _need_link(self, objects, output_file):
+        """Return true if we need to relink the files listed in 'objects'
+        to recreate 'output_file'.
+        """
+        if self.force:
+            return True
+        else:
+            if self.dry_run:
+                newer = newer_group (objects, output_file, missing='newer')
+            else:
+                newer = newer_group (objects, output_file)
+            return newer
+
+    def detect_language(self, sources):
+        """Detect the language of a given file, or list of files. Uses
+        language_map, and language_order to do the job.
+        """
+        if not isinstance(sources, list):
+            sources = [sources]
+        lang = None
+        index = len(self.language_order)
+        for source in sources:
+            base, ext = os.path.splitext(source)
+            extlang = self.language_map.get(ext)
+            try:
+                extindex = self.language_order.index(extlang)
+                if extindex < index:
+                    lang = extlang
+                    index = extindex
+            except ValueError:
+                pass
+        return lang
+
+
+    # -- Worker methods ------------------------------------------------
+    # (must be implemented by subclasses)
+
+    def preprocess(self, source, output_file=None, macros=None,
+                   include_dirs=None, extra_preargs=None, extra_postargs=None):
+        """Preprocess a single C/C++ source file, named in 'source'.
+        Output will be written to file named 'output_file', or stdout if
+        'output_file' not supplied.  'macros' is a list of macro
+        definitions as for 'compile()', which will augment the macros set
+        with 'define_macro()' and 'undefine_macro()'.  'include_dirs' is a
+        list of directory names that will be added to the default list.
+
+        Raises PreprocessError on failure.
+        """
+        pass
+
+    def compile(self, sources, output_dir=None, macros=None,
+                include_dirs=None, debug=0, extra_preargs=None,
+                extra_postargs=None, depends=None):
+        """Compile one or more source files.
+
+        'sources' must be a list of filenames, most likely C/C++
+        files, but in reality anything that can be handled by a
+        particular compiler and compiler class (eg. MSVCCompiler can
+        handle resource files in 'sources').  Return a list of object
+        filenames, one per source filename in 'sources'.  Depending on
+        the implementation, not all source files will necessarily be
+        compiled, but all corresponding object filenames will be
+        returned.
+
+        If 'output_dir' is given, object files will be put under it, while
+        retaining their original path component.  That is, "foo/bar.c"
+        normally compiles to "foo/bar.o" (for a Unix implementation); if
+        'output_dir' is "build", then it would compile to
+        "build/foo/bar.o".
+
+        'macros', if given, must be a list of macro definitions.  A macro
+        definition is either a (name, value) 2-tuple or a (name,) 1-tuple.
+        The former defines a macro; if the value is None, the macro is
+        defined without an explicit value.  The 1-tuple case undefines a
+        macro.  Later definitions/redefinitions/ undefinitions take
+        precedence.
+
+        'include_dirs', if given, must be a list of strings, the
+        directories to add to the default include file search path for this
+        compilation only.
+
+        'debug' is a boolean; if true, the compiler will be instructed to
+        output debug symbols in (or alongside) the object file(s).
+
+        'extra_preargs' and 'extra_postargs' are implementation- dependent.
+        On platforms that have the notion of a command-line (e.g. Unix,
+        DOS/Windows), they are most likely lists of strings: extra
+        command-line arguments to prepand/append to the compiler command
+        line.  On other platforms, consult the implementation class
+        documentation.  In any event, they are intended as an escape hatch
+        for those occasions when the abstract compiler framework doesn't
+        cut the mustard.
+
+        'depends', if given, is a list of filenames that all targets
+        depend on.  If a source file is older than any file in
+        depends, then the source file will be recompiled.  This
+        supports dependency tracking, but only at a coarse
+        granularity.
+
+        Raises CompileError on failure.
+        """
+        # A concrete compiler class can either override this method
+        # entirely or implement _compile().
+        macros, objects, extra_postargs, pp_opts, build = \
+                self._setup_compile(output_dir, macros, include_dirs, sources,
+                                    depends, extra_postargs)
+        cc_args = self._get_cc_args(pp_opts, debug, extra_preargs)
+
+        for obj in objects:
+            try:
+                src, ext = build[obj]
+            except KeyError:
+                continue
+            self._compile(obj, src, ext, cc_args, extra_postargs, pp_opts)
+
+        # Return *all* object filenames, not just the ones we just built.
+        return objects
+
+    def _compile(self, obj, src, ext, cc_args, extra_postargs, pp_opts):
+        """Compile 'src' to product 'obj'."""
+        # A concrete compiler class that does not override compile()
+        # should implement _compile().
+        pass
+
+    def create_static_lib(self, objects, output_libname, output_dir=None,
+                          debug=0, target_lang=None):
+        """Link a bunch of stuff together to create a static library file.
+        The "bunch of stuff" consists of the list of object files supplied
+        as 'objects', the extra object files supplied to
+        'add_link_object()' and/or 'set_link_objects()', the libraries
+        supplied to 'add_library()' and/or 'set_libraries()', and the
+        libraries supplied as 'libraries' (if any).
+
+        'output_libname' should be a library name, not a filename; the
+        filename will be inferred from the library name.  'output_dir' is
+        the directory where the library file will be put.
+
+        'debug' is a boolean; if true, debugging information will be
+        included in the library (note that on most platforms, it is the
+        compile step where this matters: the 'debug' flag is included here
+        just for consistency).
+
+        'target_lang' is the target language for which the given objects
+        are being compiled. This allows specific linkage time treatment of
+        certain languages.
+
+        Raises LibError on failure.
+        """
+        pass
+
+
+    # values for target_desc parameter in link()
+    SHARED_OBJECT = "shared_object"
+    SHARED_LIBRARY = "shared_library"
+    EXECUTABLE = "executable"
+
+    def link(self,
+             target_desc,
+             objects,
+             output_filename,
+             output_dir=None,
+             libraries=None,
+             library_dirs=None,
+             runtime_library_dirs=None,
+             export_symbols=None,
+             debug=0,
+             extra_preargs=None,
+             extra_postargs=None,
+             build_temp=None,
+             target_lang=None):
+        """Link a bunch of stuff together to create an executable or
+        shared library file.
+
+        The "bunch of stuff" consists of the list of object files supplied
+        as 'objects'.  'output_filename' should be a filename.  If
+        'output_dir' is supplied, 'output_filename' is relative to it
+        (i.e. 'output_filename' can provide directory components if
+        needed).
+
+        'libraries' is a list of libraries to link against.  These are
+        library names, not filenames, since they're translated into
+        filenames in a platform-specific way (eg. "foo" becomes "libfoo.a"
+        on Unix and "foo.lib" on DOS/Windows).  However, they can include a
+        directory component, which means the linker will look in that
+        specific directory rather than searching all the normal locations.
+
+        'library_dirs', if supplied, should be a list of directories to
+        search for libraries that were specified as bare library names
+        (ie. no directory component).  These are on top of the system
+        default and those supplied to 'add_library_dir()' and/or
+        'set_library_dirs()'.  'runtime_library_dirs' is a list of
+        directories that will be embedded into the shared library and used
+        to search for other shared libraries that *it* depends on at
+        run-time.  (This may only be relevant on Unix.)
+
+        'export_symbols' is a list of symbols that the shared library will
+        export.  (This appears to be relevant only on Windows.)
+
+        'debug' is as for 'compile()' and 'create_static_lib()', with the
+        slight distinction that it actually matters on most platforms (as
+        opposed to 'create_static_lib()', which includes a 'debug' flag
+        mostly for form's sake).
+
+        'extra_preargs' and 'extra_postargs' are as for 'compile()' (except
+        of course that they supply command-line arguments for the
+        particular linker being used).
+
+        'target_lang' is the target language for which the given objects
+        are being compiled. This allows specific linkage time treatment of
+        certain languages.
+
+        Raises LinkError on failure.
+        """
+        raise NotImplementedError
+
+
+    # Old 'link_*()' methods, rewritten to use the new 'link()' method.
+
+    def link_shared_lib(self,
+                        objects,
+                        output_libname,
+                        output_dir=None,
+                        libraries=None,
+                        library_dirs=None,
+                        runtime_library_dirs=None,
+                        export_symbols=None,
+                        debug=0,
+                        extra_preargs=None,
+                        extra_postargs=None,
+                        build_temp=None,
+                        target_lang=None):
+        self.link(CCompiler.SHARED_LIBRARY, objects,
+                  self.library_filename(output_libname, lib_type='shared'),
+                  output_dir,
+                  libraries, library_dirs, runtime_library_dirs,
+                  export_symbols, debug,
+                  extra_preargs, extra_postargs, build_temp, target_lang)
+
+
+    def link_shared_object(self,
+                           objects,
+                           output_filename,
+                           output_dir=None,
+                           libraries=None,
+                           library_dirs=None,
+                           runtime_library_dirs=None,
+                           export_symbols=None,
+                           debug=0,
+                           extra_preargs=None,
+                           extra_postargs=None,
+                           build_temp=None,
+                           target_lang=None):
+        self.link(CCompiler.SHARED_OBJECT, objects,
+                  output_filename, output_dir,
+                  libraries, library_dirs, runtime_library_dirs,
+                  export_symbols, debug,
+                  extra_preargs, extra_postargs, build_temp, target_lang)
+
+
+    def link_executable(self,
+                        objects,
+                        output_progname,
+                        output_dir=None,
+                        libraries=None,
+                        library_dirs=None,
+                        runtime_library_dirs=None,
+                        debug=0,
+                        extra_preargs=None,
+                        extra_postargs=None,
+                        target_lang=None):
+        self.link(CCompiler.EXECUTABLE, objects,
+                  self.executable_filename(output_progname), output_dir,
+                  libraries, library_dirs, runtime_library_dirs, None,
+                  debug, extra_preargs, extra_postargs, None, target_lang)
+
+
+    # -- Miscellaneous methods -----------------------------------------
+    # These are all used by the 'gen_lib_options() function; there is
+    # no appropriate default implementation so subclasses should
+    # implement all of these.
+
+    def library_dir_option(self, dir):
+        """Return the compiler option to add 'dir' to the list of
+        directories searched for libraries.
+        """
+        raise NotImplementedError
+
+    def runtime_library_dir_option(self, dir):
+        """Return the compiler option to add 'dir' to the list of
+        directories searched for runtime libraries.
+        """
+        raise NotImplementedError
+
+    def library_option(self, lib):
+        """Return the compiler option to add 'lib' to the list of libraries
+        linked into the shared library or executable.
+        """
+        raise NotImplementedError
+
+    def has_function(self, funcname, includes=None, include_dirs=None,
+                     libraries=None, library_dirs=None):
+        """Return a boolean indicating whether funcname is supported on
+        the current platform.  The optional arguments can be used to
+        augment the compilation environment.
+        """
+        # this can't be included at module scope because it tries to
+        # import math which might not be available at that point - maybe
+        # the necessary logic should just be inlined?
+        import tempfile
+        if includes is None:
+            includes = []
+        if include_dirs is None:
+            include_dirs = []
+        if libraries is None:
+            libraries = []
+        if library_dirs is None:
+            library_dirs = []
+        fd, fname = tempfile.mkstemp(".c", funcname, text=True)
+        f = os.fdopen(fd, "w")
+        try:
+            for incl in includes:
+                f.write("""#include "%s"\n""" % incl)
+            f.write("""\
+main (int argc, char **argv) {
+    %s();
+}
+""" % funcname)
+        finally:
+            f.close()
+        try:
+            objects = self.compile([fname], include_dirs=include_dirs)
+        except CompileError:
+            return False
+
+        try:
+            self.link_executable(objects, "a.out",
+                                 libraries=libraries,
+                                 library_dirs=library_dirs)
+        except (LinkError, TypeError):
+            return False
+        return True
+
+    def find_library_file (self, dirs, lib, debug=0):
+        """Search the specified list of directories for a static or shared
+        library file 'lib' and return the full path to that file.  If
+        'debug' true, look for a debugging version (if that makes sense on
+        the current platform).  Return None if 'lib' wasn't found in any of
+        the specified directories.
+        """
+        raise NotImplementedError
+
+    # -- Filename generation methods -----------------------------------
+
+    # The default implementation of the filename generating methods are
+    # prejudiced towards the Unix/DOS/Windows view of the world:
+    #   * object files are named by replacing the source file extension
+    #     (eg. .c/.cpp -> .o/.obj)
+    #   * library files (shared or static) are named by plugging the
+    #     library name and extension into a format string, eg.
+    #     "lib%s.%s" % (lib_name, ".a") for Unix static libraries
+    #   * executables are named by appending an extension (possibly
+    #     empty) to the program name: eg. progname + ".exe" for
+    #     Windows
+    #
+    # To reduce redundant code, these methods expect to find
+    # several attributes in the current object (presumably defined
+    # as class attributes):
+    #   * src_extensions -
+    #     list of C/C++ source file extensions, eg. ['.c', '.cpp']
+    #   * obj_extension -
+    #     object file extension, eg. '.o' or '.obj'
+    #   * static_lib_extension -
+    #     extension for static library files, eg. '.a' or '.lib'
+    #   * shared_lib_extension -
+    #     extension for shared library/object files, eg. '.so', '.dll'
+    #   * static_lib_format -
+    #     format string for generating static library filenames,
+    #     eg. 'lib%s.%s' or '%s.%s'
+    #   * shared_lib_format
+    #     format string for generating shared library filenames
+    #     (probably same as static_lib_format, since the extension
+    #     is one of the intended parameters to the format string)
+    #   * exe_extension -
+    #     extension for executable files, eg. '' or '.exe'
+
+    def object_filenames(self, source_filenames, strip_dir=0, output_dir=''):
+        if output_dir is None:
+            output_dir = ''
+        obj_names = []
+        for src_name in source_filenames:
+            base, ext = os.path.splitext(src_name)
+            base = os.path.splitdrive(base)[1] # Chop off the drive
+            base = base[os.path.isabs(base):]  # If abs, chop off leading /
+            if ext not in self.src_extensions:
+                raise UnknownFileError(
+                      "unknown file type '%s' (from '%s')" % (ext, src_name))
+            if strip_dir:
+                base = os.path.basename(base)
+            obj_names.append(os.path.join(output_dir,
+                                          base + self.obj_extension))
+        return obj_names
+
+    def shared_object_filename(self, basename, strip_dir=0, output_dir=''):
+        assert output_dir is not None
+        if strip_dir:
+            basename = os.path.basename(basename)
+        return os.path.join(output_dir, basename + self.shared_lib_extension)
+
+    def executable_filename(self, basename, strip_dir=0, output_dir=''):
+        assert output_dir is not None
+        if strip_dir:
+            basename = os.path.basename(basename)
+        return os.path.join(output_dir, basename + (self.exe_extension or ''))
+
+    def library_filename(self, libname, lib_type='static',     # or 'shared'
+                         strip_dir=0, output_dir=''):
+        assert output_dir is not None
+        if lib_type not in ("static", "shared", "dylib", "xcode_stub"):
+            raise ValueError(
+                  "'lib_type' must be \"static\", \"shared\", \"dylib\", or \"xcode_stub\"")
+        fmt = getattr(self, lib_type + "_lib_format")
+        ext = getattr(self, lib_type + "_lib_extension")
+
+        dir, base = os.path.split(libname)
+        filename = fmt % (base, ext)
+        if strip_dir:
+            dir = ''
+
+        return os.path.join(output_dir, dir, filename)
+
+
+    # -- Utility methods -----------------------------------------------
+
+    def announce(self, msg, level=1):
+        log.debug(msg)
+
+    def debug_print(self, msg):
+        from distutils.debug import DEBUG
+        if DEBUG:
+            print(msg)
+
+    def warn(self, msg):
+        sys.stderr.write("warning: %s\n" % msg)
+
+    def execute(self, func, args, msg=None, level=1):
+        execute(func, args, msg, self.dry_run)
+
+    def spawn(self, cmd):
+        spawn(cmd, dry_run=self.dry_run)
+
+    def move_file(self, src, dst):
+        return move_file(src, dst, dry_run=self.dry_run)
+
+    def mkpath (self, name, mode=0o777):
+        mkpath(name, mode, dry_run=self.dry_run)
+
+
+# Map a sys.platform/os.name ('posix', 'nt') to the default compiler
+# type for that platform. Keys are interpreted as re match
+# patterns. Order is important; platform mappings are preferred over
+# OS names.
+_default_compilers = (
+
+    # Platform string mappings
+
+    # on a cygwin built python we can use gcc like an ordinary UNIXish
+    # compiler
+    ('cygwin.*', 'unix'),
+
+    # OS name mappings
+    ('posix', 'unix'),
+    ('nt', 'msvc'),
+
+    )
+
+def get_default_compiler(osname=None, platform=None):
+    """Determine the default compiler to use for the given platform.
+
+       osname should be one of the standard Python OS names (i.e. the
+       ones returned by os.name) and platform the common value
+       returned by sys.platform for the platform in question.
+
+       The default values are os.name and sys.platform in case the
+       parameters are not given.
+    """
+    if osname is None:
+        osname = os.name
+    if platform is None:
+        platform = sys.platform
+    for pattern, compiler in _default_compilers:
+        if re.match(pattern, platform) is not None or \
+           re.match(pattern, osname) is not None:
+            return compiler
+    # Default to Unix compiler
+    return 'unix'
+
+# Map compiler types to (module_name, class_name) pairs -- ie. where to
+# find the code that implements an interface to this compiler.  (The module
+# is assumed to be in the 'distutils' package.)
+compiler_class = { 'unix':    ('unixccompiler', 'UnixCCompiler',
+                               "standard UNIX-style compiler"),
+                   'msvc':    ('_msvccompiler', 'MSVCCompiler',
+                               "Microsoft Visual C++"),
+                   'cygwin':  ('cygwinccompiler', 'CygwinCCompiler',
+                               "Cygwin port of GNU C Compiler for Win32"),
+                   'mingw32': ('cygwinccompiler', 'Mingw32CCompiler',
+                               "Mingw32 port of GNU C Compiler for Win32"),
+                   'bcpp':    ('bcppcompiler', 'BCPPCompiler',
+                               "Borland C++ Compiler"),
+                 }
+
+def show_compilers():
+    """Print list of available compilers (used by the "--help-compiler"
+    options to "build", "build_ext", "build_clib").
+    """
+    # XXX this "knows" that the compiler option it's describing is
+    # "--compiler", which just happens to be the case for the three
+    # commands that use it.
+    from distutils.fancy_getopt import FancyGetopt
+    compilers = []
+    for compiler in compiler_class.keys():
+        compilers.append(("compiler="+compiler, None,
+                          compiler_class[compiler][2]))
+    compilers.sort()
+    pretty_printer = FancyGetopt(compilers)
+    pretty_printer.print_help("List of available compilers:")
+
+
+def new_compiler(plat=None, compiler=None, verbose=0, dry_run=0, force=0):
+    """Generate an instance of some CCompiler subclass for the supplied
+    platform/compiler combination.  'plat' defaults to 'os.name'
+    (eg. 'posix', 'nt'), and 'compiler' defaults to the default compiler
+    for that platform.  Currently only 'posix' and 'nt' are supported, and
+    the default compilers are "traditional Unix interface" (UnixCCompiler
+    class) and Visual C++ (MSVCCompiler class).  Note that it's perfectly
+    possible to ask for a Unix compiler object under Windows, and a
+    Microsoft compiler object under Unix -- if you supply a value for
+    'compiler', 'plat' is ignored.
+    """
+    if plat is None:
+        plat = os.name
+
+    try:
+        if compiler is None:
+            compiler = get_default_compiler(plat)
+
+        (module_name, class_name, long_description) = compiler_class[compiler]
+    except KeyError:
+        msg = "don't know how to compile C/C++ code on platform '%s'" % plat
+        if compiler is not None:
+            msg = msg + " with '%s' compiler" % compiler
+        raise DistutilsPlatformError(msg)
+
+    try:
+        module_name = "distutils." + module_name
+        __import__ (module_name)
+        module = sys.modules[module_name]
+        klass = vars(module)[class_name]
+    except ImportError:
+        raise DistutilsModuleError(
+              "can't compile C/C++ code: unable to load module '%s'" % \
+              module_name)
+    except KeyError:
+        raise DistutilsModuleError(
+               "can't compile C/C++ code: unable to find class '%s' "
+               "in module '%s'" % (class_name, module_name))
+
+    # XXX The None is necessary to preserve backwards compatibility
+    # with classes that expect verbose to be the first positional
+    # argument.
+    return klass(None, dry_run, force)
+
+
+def gen_preprocess_options(macros, include_dirs):
+    """Generate C pre-processor options (-D, -U, -I) as used by at least
+    two types of compilers: the typical Unix compiler and Visual C++.
+    'macros' is the usual thing, a list of 1- or 2-tuples, where (name,)
+    means undefine (-U) macro 'name', and (name,value) means define (-D)
+    macro 'name' to 'value'.  'include_dirs' is just a list of directory
+    names to be added to the header file search path (-I).  Returns a list
+    of command-line options suitable for either Unix compilers or Visual
+    C++.
+    """
+    # XXX it would be nice (mainly aesthetic, and so we don't generate
+    # stupid-looking command lines) to go over 'macros' and eliminate
+    # redundant definitions/undefinitions (ie. ensure that only the
+    # latest mention of a particular macro winds up on the command
+    # line).  I don't think it's essential, though, since most (all?)
+    # Unix C compilers only pay attention to the latest -D or -U
+    # mention of a macro on their command line.  Similar situation for
+    # 'include_dirs'.  I'm punting on both for now.  Anyways, weeding out
+    # redundancies like this should probably be the province of
+    # CCompiler, since the data structures used are inherited from it
+    # and therefore common to all CCompiler classes.
+    pp_opts = []
+    for macro in macros:
+        if not (isinstance(macro, tuple) and 1 <= len(macro) <= 2):
+            raise TypeError(
+                  "bad macro definition '%s': "
+                  "each element of 'macros' list must be a 1- or 2-tuple"
+                  % macro)
+
+        if len(macro) == 1:        # undefine this macro
+            pp_opts.append("-U%s" % macro[0])
+        elif len(macro) == 2:
+            if macro[1] is None:    # define with no explicit value
+                pp_opts.append("-D%s" % macro[0])
+            else:
+                # XXX *don't* need to be clever about quoting the
+                # macro value here, because we're going to avoid the
+                # shell at all costs when we spawn the command!
+                pp_opts.append("-D%s=%s" % macro)
+
+    for dir in include_dirs:
+        pp_opts.append("-I%s" % dir)
+    return pp_opts
+
+
+def gen_lib_options (compiler, library_dirs, runtime_library_dirs, libraries):
+    """Generate linker options for searching library directories and
+    linking with specific libraries.  'libraries' and 'library_dirs' are,
+    respectively, lists of library names (not filenames!) and search
+    directories.  Returns a list of command-line options suitable for use
+    with some compiler (depending on the two format strings passed in).
+    """
+    lib_opts = []
+
+    for dir in library_dirs:
+        lib_opts.append(compiler.library_dir_option(dir))
+
+    for dir in runtime_library_dirs:
+        opt = compiler.runtime_library_dir_option(dir)
+        if isinstance(opt, list):
+            lib_opts = lib_opts + opt
+        else:
+            lib_opts.append(opt)
+
+    # XXX it's important that we *not* remove redundant library mentions!
+    # sometimes you really do have to say "-lfoo -lbar -lfoo" in order to
+    # resolve all symbols.  I just hope we never have to say "-lfoo obj.o
+    # -lbar" to get things to work -- that's certainly a possibility, but a
+    # pretty nasty way to arrange your C code.
+
+    for lib in libraries:
+        (lib_dir, lib_name) = os.path.split(lib)
+        if lib_dir:
+            lib_file = compiler.find_library_file([lib_dir], lib_name)
+            if lib_file:
+                lib_opts.append(lib_file)
+            else:
+                compiler.warn("no library file corresponding to "
+                              "'%s' found (skipping)" % lib)
+        else:
+            lib_opts.append(compiler.library_option (lib))
+    return lib_opts
diff --git a/Lib/distutils/cmd.py b/Lib/distutils/cmd.py
new file mode 100644
index 000000000..939f79594
--- /dev/null
+++ b/Lib/distutils/cmd.py
@@ -0,0 +1,434 @@
+"""distutils.cmd
+
+Provides the Command class, the base class for the command classes
+in the distutils.command package.
+"""
+
+import sys, os, re
+from distutils.errors import DistutilsOptionError
+from distutils import util, dir_util, file_util, archive_util, dep_util
+from distutils import log
+
+class Command:
+    """Abstract base class for defining command classes, the "worker bees"
+    of the Distutils.  A useful analogy for command classes is to think of
+    them as subroutines with local variables called "options".  The options
+    are "declared" in 'initialize_options()' and "defined" (given their
+    final values, aka "finalized") in 'finalize_options()', both of which
+    must be defined by every command class.  The distinction between the
+    two is necessary because option values might come from the outside
+    world (command line, config file, ...), and any options dependent on
+    other options must be computed *after* these outside influences have
+    been processed -- hence 'finalize_options()'.  The "body" of the
+    subroutine, where it does all its work based on the values of its
+    options, is the 'run()' method, which must also be implemented by every
+    command class.
+    """
+
+    # 'sub_commands' formalizes the notion of a "family" of commands,
+    # eg. "install" as the parent with sub-commands "install_lib",
+    # "install_headers", etc.  The parent of a family of commands
+    # defines 'sub_commands' as a class attribute; it's a list of
+    #    (command_name : string, predicate : unbound_method | string | None)
+    # tuples, where 'predicate' is a method of the parent command that
+    # determines whether the corresponding command is applicable in the
+    # current situation.  (Eg. we "install_headers" is only applicable if
+    # we have any C header files to install.)  If 'predicate' is None,
+    # that command is always applicable.
+    #
+    # 'sub_commands' is usually defined at the *end* of a class, because
+    # predicates can be unbound methods, so they must already have been
+    # defined.  The canonical example is the "install" command.
+    sub_commands = []
+
+
+    # -- Creation/initialization methods -------------------------------
+
+    def __init__(self, dist):
+        """Create and initialize a new Command object.  Most importantly,
+        invokes the 'initialize_options()' method, which is the real
+        initializer and depends on the actual command being
+        instantiated.
+        """
+        # late import because of mutual dependence between these classes
+        from distutils.dist import Distribution
+
+        if not isinstance(dist, Distribution):
+            raise TypeError("dist must be a Distribution instance")
+        if self.__class__ is Command:
+            raise RuntimeError("Command is an abstract class")
+
+        self.distribution = dist
+        self.initialize_options()
+
+        # Per-command versions of the global flags, so that the user can
+        # customize Distutils' behaviour command-by-command and let some
+        # commands fall back on the Distribution's behaviour.  None means
+        # "not defined, check self.distribution's copy", while 0 or 1 mean
+        # false and true (duh).  Note that this means figuring out the real
+        # value of each flag is a touch complicated -- hence "self._dry_run"
+        # will be handled by __getattr__, below.
+        # XXX This needs to be fixed.
+        self._dry_run = None
+
+        # verbose is largely ignored, but needs to be set for
+        # backwards compatibility (I think)?
+        self.verbose = dist.verbose
+
+        # Some commands define a 'self.force' option to ignore file
+        # timestamps, but methods defined *here* assume that
+        # 'self.force' exists for all commands.  So define it here
+        # just to be safe.
+        self.force = None
+
+        # The 'help' flag is just used for command-line parsing, so
+        # none of that complicated bureaucracy is needed.
+        self.help = 0
+
+        # 'finalized' records whether or not 'finalize_options()' has been
+        # called.  'finalize_options()' itself should not pay attention to
+        # this flag: it is the business of 'ensure_finalized()', which
+        # always calls 'finalize_options()', to respect/update it.
+        self.finalized = 0
+
+    # XXX A more explicit way to customize dry_run would be better.
+    def __getattr__(self, attr):
+        if attr == 'dry_run':
+            myval = getattr(self, "_" + attr)
+            if myval is None:
+                return getattr(self.distribution, attr)
+            else:
+                return myval
+        else:
+            raise AttributeError(attr)
+
+    def ensure_finalized(self):
+        if not self.finalized:
+            self.finalize_options()
+        self.finalized = 1
+
+    # Subclasses must define:
+    #   initialize_options()
+    #     provide default values for all options; may be customized by
+    #     setup script, by options from config file(s), or by command-line
+    #     options
+    #   finalize_options()
+    #     decide on the final values for all options; this is called
+    #     after all possible intervention from the outside world
+    #     (command-line, option file, etc.) has been processed
+    #   run()
+    #     run the command: do whatever it is we're here to do,
+    #     controlled by the command's various option values
+
+    def initialize_options(self):
+        """Set default values for all the options that this command
+        supports.  Note that these defaults may be overridden by other
+        commands, by the setup script, by config files, or by the
+        command-line.  Thus, this is not the place to code dependencies
+        between options; generally, 'initialize_options()' implementations
+        are just a bunch of "self.foo = None" assignments.
+
+        This method must be implemented by all command classes.
+        """
+        raise RuntimeError("abstract method -- subclass %s must override"
+                           % self.__class__)
+
+    def finalize_options(self):
+        """Set final values for all the options that this command supports.
+        This is always called as late as possible, ie.  after any option
+        assignments from the command-line or from other commands have been
+        done.  Thus, this is the place to code option dependencies: if
+        'foo' depends on 'bar', then it is safe to set 'foo' from 'bar' as
+        long as 'foo' still has the same value it was assigned in
+        'initialize_options()'.
+
+        This method must be implemented by all command classes.
+        """
+        raise RuntimeError("abstract method -- subclass %s must override"
+                           % self.__class__)
+
+
+    def dump_options(self, header=None, indent=""):
+        from distutils.fancy_getopt import longopt_xlate
+        if header is None:
+            header = "command options for '%s':" % self.get_command_name()
+        self.announce(indent + header, level=log.INFO)
+        indent = indent + "  "
+        for (option, _, _) in self.user_options:
+            option = option.translate(longopt_xlate)
+            if option[-1] == "=":
+                option = option[:-1]
+            value = getattr(self, option)
+            self.announce(indent + "%s = %s" % (option, value),
+                          level=log.INFO)
+
+    def run(self):
+        """A command's raison d'etre: carry out the action it exists to
+        perform, controlled by the options initialized in
+        'initialize_options()', customized by other commands, the setup
+        script, the command-line, and config files, and finalized in
+        'finalize_options()'.  All terminal output and filesystem
+        interaction should be done by 'run()'.
+
+        This method must be implemented by all command classes.
+        """
+        raise RuntimeError("abstract method -- subclass %s must override"
+                           % self.__class__)
+
+    def announce(self, msg, level=1):
+        """If the current verbosity level is of greater than or equal to
+        'level' print 'msg' to stdout.
+        """
+        log.log(level, msg)
+
+    def debug_print(self, msg):
+        """Print 'msg' to stdout if the global DEBUG (taken from the
+        DISTUTILS_DEBUG environment variable) flag is true.
+        """
+        from distutils.debug import DEBUG
+        if DEBUG:
+            print(msg)
+            sys.stdout.flush()
+
+
+    # -- Option validation methods -------------------------------------
+    # (these are very handy in writing the 'finalize_options()' method)
+    #
+    # NB. the general philosophy here is to ensure that a particular option
+    # value meets certain type and value constraints.  If not, we try to
+    # force it into conformance (eg. if we expect a list but have a string,
+    # split the string on comma and/or whitespace).  If we can't force the
+    # option into conformance, raise DistutilsOptionError.  Thus, command
+    # classes need do nothing more than (eg.)
+    #   self.ensure_string_list('foo')
+    # and they can be guaranteed that thereafter, self.foo will be
+    # a list of strings.
+
+    def _ensure_stringlike(self, option, what, default=None):
+        val = getattr(self, option)
+        if val is None:
+            setattr(self, option, default)
+            return default
+        elif not isinstance(val, str):
+            raise DistutilsOptionError("'%s' must be a %s (got `%s`)"
+                                       % (option, what, val))
+        return val
+
+    def ensure_string(self, option, default=None):
+        """Ensure that 'option' is a string; if not defined, set it to
+        'default'.
+        """
+        self._ensure_stringlike(option, "string", default)
+
+    def ensure_string_list(self, option):
+        r"""Ensure that 'option' is a list of strings.  If 'option' is
+        currently a string, we split it either on /,\s*/ or /\s+/, so
+        "foo bar baz", "foo,bar,baz", and "foo,   bar baz" all become
+        ["foo", "bar", "baz"].
+        """
+        val = getattr(self, option)
+        if val is None:
+            return
+        elif isinstance(val, str):
+            setattr(self, option, re.split(r',\s*|\s+', val))
+        else:
+            if isinstance(val, list):
+                ok = all(isinstance(v, str) for v in val)
+            else:
+                ok = False
+            if not ok:
+                raise DistutilsOptionError(
+                      "'%s' must be a list of strings (got %r)"
+                      % (option, val))
+
+    def _ensure_tested_string(self, option, tester, what, error_fmt,
+                              default=None):
+        val = self._ensure_stringlike(option, what, default)
+        if val is not None and not tester(val):
+            raise DistutilsOptionError(("error in '%s' option: " + error_fmt)
+                                       % (option, val))
+
+    def ensure_filename(self, option):
+        """Ensure that 'option' is the name of an existing file."""
+        self._ensure_tested_string(option, os.path.isfile,
+                                   "filename",
+                                   "'%s' does not exist or is not a file")
+
+    def ensure_dirname(self, option):
+        self._ensure_tested_string(option, os.path.isdir,
+                                   "directory name",
+                                   "'%s' does not exist or is not a directory")
+
+
+    # -- Convenience methods for commands ------------------------------
+
+    def get_command_name(self):
+        if hasattr(self, 'command_name'):
+            return self.command_name
+        else:
+            return self.__class__.__name__
+
+    def set_undefined_options(self, src_cmd, *option_pairs):
+        """Set the values of any "undefined" options from corresponding
+        option values in some other command object.  "Undefined" here means
+        "is None", which is the convention used to indicate that an option
+        has not been changed between 'initialize_options()' and
+        'finalize_options()'.  Usually called from 'finalize_options()' for
+        options that depend on some other command rather than another
+        option of the same command.  'src_cmd' is the other command from
+        which option values will be taken (a command object will be created
+        for it if necessary); the remaining arguments are
+        '(src_option,dst_option)' tuples which mean "take the value of
+        'src_option' in the 'src_cmd' command object, and copy it to
+        'dst_option' in the current command object".
+        """
+        # Option_pairs: list of (src_option, dst_option) tuples
+        src_cmd_obj = self.distribution.get_command_obj(src_cmd)
+        src_cmd_obj.ensure_finalized()
+        for (src_option, dst_option) in option_pairs:
+            if getattr(self, dst_option) is None:
+                setattr(self, dst_option, getattr(src_cmd_obj, src_option))
+
+    def get_finalized_command(self, command, create=1):
+        """Wrapper around Distribution's 'get_command_obj()' method: find
+        (create if necessary and 'create' is true) the command object for
+        'command', call its 'ensure_finalized()' method, and return the
+        finalized command object.
+        """
+        cmd_obj = self.distribution.get_command_obj(command, create)
+        cmd_obj.ensure_finalized()
+        return cmd_obj
+
+    # XXX rename to 'get_reinitialized_command()'? (should do the
+    # same in dist.py, if so)
+    def reinitialize_command(self, command, reinit_subcommands=0):
+        return self.distribution.reinitialize_command(command,
+                                                      reinit_subcommands)
+
+    def run_command(self, command):
+        """Run some other command: uses the 'run_command()' method of
+        Distribution, which creates and finalizes the command object if
+        necessary and then invokes its 'run()' method.
+        """
+        self.distribution.run_command(command)
+
+    def get_sub_commands(self):
+        """Determine the sub-commands that are relevant in the current
+        distribution (ie., that need to be run).  This is based on the
+        'sub_commands' class attribute: each tuple in that list may include
+        a method that we call to determine if the subcommand needs to be
+        run for the current distribution.  Return a list of command names.
+        """
+        commands = []
+        for (cmd_name, method) in self.sub_commands:
+            if method is None or method(self):
+                commands.append(cmd_name)
+        return commands
+
+
+    # -- External world manipulation -----------------------------------
+
+    def warn(self, msg):
+        log.warn("warning: %s: %s\n", self.get_command_name(), msg)
+
+    def execute(self, func, args, msg=None, level=1):
+        util.execute(func, args, msg, dry_run=self.dry_run)
+
+    def mkpath(self, name, mode=0o777):
+        dir_util.mkpath(name, mode, dry_run=self.dry_run)
+
+    def copy_file(self, infile, outfile, preserve_mode=1, preserve_times=1,
+                  link=None, level=1):
+        """Copy a file respecting verbose, dry-run and force flags.  (The
+        former two default to whatever is in the Distribution object, and
+        the latter defaults to false for commands that don't define it.)"""
+        return file_util.copy_file(infile, outfile, preserve_mode,
+                                   preserve_times, not self.force, link,
+                                   dry_run=self.dry_run)
+
+    def copy_tree(self, infile, outfile, preserve_mode=1, preserve_times=1,
+                   preserve_symlinks=0, level=1):
+        """Copy an entire directory tree respecting verbose, dry-run,
+        and force flags.
+        """
+        return dir_util.copy_tree(infile, outfile, preserve_mode,
+                                  preserve_times, preserve_symlinks,
+                                  not self.force, dry_run=self.dry_run)
+
+    def move_file (self, src, dst, level=1):
+        """Move a file respecting dry-run flag."""
+        return file_util.move_file(src, dst, dry_run=self.dry_run)
+
+    def spawn(self, cmd, search_path=1, level=1):
+        """Spawn an external command respecting dry-run flag."""
+        from distutils.spawn import spawn
+        spawn(cmd, search_path, dry_run=self.dry_run)
+
+    def make_archive(self, base_name, format, root_dir=None, base_dir=None,
+                     owner=None, group=None):
+        return archive_util.make_archive(base_name, format, root_dir, base_dir,
+                                         dry_run=self.dry_run,
+                                         owner=owner, group=group)
+
+    def make_file(self, infiles, outfile, func, args,
+                  exec_msg=None, skip_msg=None, level=1):
+        """Special case of 'execute()' for operations that process one or
+        more input files and generate one output file.  Works just like
+        'execute()', except the operation is skipped and a different
+        message printed if 'outfile' already exists and is newer than all
+        files listed in 'infiles'.  If the command defined 'self.force',
+        and it is true, then the command is unconditionally run -- does no
+        timestamp checks.
+        """
+        if skip_msg is None:
+            skip_msg = "skipping %s (inputs unchanged)" % outfile
+
+        # Allow 'infiles' to be a single string
+        if isinstance(infiles, str):
+            infiles = (infiles,)
+        elif not isinstance(infiles, (list, tuple)):
+            raise TypeError(
+                  "'infiles' must be a string, or a list or tuple of strings")
+
+        if exec_msg is None:
+            exec_msg = "generating %s from %s" % (outfile, ', '.join(infiles))
+
+        # If 'outfile' must be regenerated (either because it doesn't
+        # exist, is out-of-date, or the 'force' flag is true) then
+        # perform the action that presumably regenerates it
+        if self.force or dep_util.newer_group(infiles, outfile):
+            self.execute(func, args, exec_msg, level)
+        # Otherwise, print the "skip" message
+        else:
+            log.debug(skip_msg)
+
+# XXX 'install_misc' class not currently used -- it was the base class for
+# both 'install_scripts' and 'install_data', but they outgrew it.  It might
+# still be useful for 'install_headers', though, so I'm keeping it around
+# for the time being.
+
+class install_misc(Command):
+    """Common base class for installing some files in a subdirectory.
+    Currently used by install_data and install_scripts.
+    """
+
+    user_options = [('install-dir=', 'd', "directory to install the files to")]
+
+    def initialize_options (self):
+        self.install_dir = None
+        self.outfiles = []
+
+    def _install_dir_from(self, dirname):
+        self.set_undefined_options('install', (dirname, 'install_dir'))
+
+    def _copy_files(self, filelist):
+        self.outfiles = []
+        if not filelist:
+            return
+        self.mkpath(self.install_dir)
+        for f in filelist:
+            self.copy_file(f, self.install_dir)
+            self.outfiles.append(os.path.join(self.install_dir, f))
+
+    def get_outputs(self):
+        return self.outfiles
diff --git a/Lib/distutils/command/__init__.py b/Lib/distutils/command/__init__.py
new file mode 100644
index 000000000..481eea9fd
--- /dev/null
+++ b/Lib/distutils/command/__init__.py
@@ -0,0 +1,31 @@
+"""distutils.command
+
+Package containing implementation of all the standard Distutils
+commands."""
+
+__all__ = ['build',
+           'build_py',
+           'build_ext',
+           'build_clib',
+           'build_scripts',
+           'clean',
+           'install',
+           'install_lib',
+           'install_headers',
+           'install_scripts',
+           'install_data',
+           'sdist',
+           'register',
+           'bdist',
+           'bdist_dumb',
+           'bdist_rpm',
+           'bdist_wininst',
+           'check',
+           'upload',
+           # These two are reserved for future use:
+           #'bdist_sdux',
+           #'bdist_pkgtool',
+           # Note:
+           # bdist_packager is not included because it only provides
+           # an abstract base class
+          ]
diff --git a/Lib/distutils/command/bdist.py b/Lib/distutils/command/bdist.py
new file mode 100644
index 000000000..014871d28
--- /dev/null
+++ b/Lib/distutils/command/bdist.py
@@ -0,0 +1,143 @@
+"""distutils.command.bdist
+
+Implements the Distutils 'bdist' command (create a built [binary]
+distribution)."""
+
+import os
+from distutils.core import Command
+from distutils.errors import *
+from distutils.util import get_platform
+
+
+def show_formats():
+    """Print list of available formats (arguments to "--format" option).
+    """
+    from distutils.fancy_getopt import FancyGetopt
+    formats = []
+    for format in bdist.format_commands:
+        formats.append(("formats=" + format, None,
+                        bdist.format_command[format][1]))
+    pretty_printer = FancyGetopt(formats)
+    pretty_printer.print_help("List of available distribution formats:")
+
+
+class bdist(Command):
+
+    description = "create a built (binary) distribution"
+
+    user_options = [('bdist-base=', 'b',
+                     "temporary directory for creating built distributions"),
+                    ('plat-name=', 'p',
+                     "platform name to embed in generated filenames "
+                     "(default: %s)" % get_platform()),
+                    ('formats=', None,
+                     "formats for distribution (comma-separated list)"),
+                    ('dist-dir=', 'd',
+                     "directory to put final built distributions in "
+                     "[default: dist]"),
+                    ('skip-build', None,
+                     "skip rebuilding everything (for testing/debugging)"),
+                    ('owner=', 'u',
+                     "Owner name used when creating a tar file"
+                     " [default: current user]"),
+                    ('group=', 'g',
+                     "Group name used when creating a tar file"
+                     " [default: current group]"),
+                   ]
+
+    boolean_options = ['skip-build']
+
+    help_options = [
+        ('help-formats', None,
+         "lists available distribution formats", show_formats),
+        ]
+
+    # The following commands do not take a format option from bdist
+    no_format_option = ('bdist_rpm',)
+
+    # This won't do in reality: will need to distinguish RPM-ish Linux,
+    # Debian-ish Linux, Solaris, FreeBSD, ..., Windows, Mac OS.
+    default_format = {'posix': 'gztar',
+                      'nt': 'zip'}
+
+    # Establish the preferred order (for the --help-formats option).
+    format_commands = ['rpm', 'gztar', 'bztar', 'xztar', 'ztar', 'tar',
+                       'wininst', 'zip', 'msi']
+
+    # And the real information.
+    format_command = {'rpm':   ('bdist_rpm',  "RPM distribution"),
+                      'gztar': ('bdist_dumb', "gzip'ed tar file"),
+                      'bztar': ('bdist_dumb', "bzip2'ed tar file"),
+                      'xztar': ('bdist_dumb', "xz'ed tar file"),
+                      'ztar':  ('bdist_dumb', "compressed tar file"),
+                      'tar':   ('bdist_dumb', "tar file"),
+                      'wininst': ('bdist_wininst',
+                                  "Windows executable installer"),
+                      'zip':   ('bdist_dumb', "ZIP file"),
+                      'msi':   ('bdist_msi',  "Microsoft Installer")
+                      }
+
+
+    def initialize_options(self):
+        self.bdist_base = None
+        self.plat_name = None
+        self.formats = None
+        self.dist_dir = None
+        self.skip_build = 0
+        self.group = None
+        self.owner = None
+
+    def finalize_options(self):
+        # have to finalize 'plat_name' before 'bdist_base'
+        if self.plat_name is None:
+            if self.skip_build:
+                self.plat_name = get_platform()
+            else:
+                self.plat_name = self.get_finalized_command('build').plat_name
+
+        # 'bdist_base' -- parent of per-built-distribution-format
+        # temporary directories (eg. we'll probably have
+        # "build/bdist.<plat>/dumb", "build/bdist.<plat>/rpm", etc.)
+        if self.bdist_base is None:
+            build_base = self.get_finalized_command('build').build_base
+            self.bdist_base = os.path.join(build_base,
+                                           'bdist.' + self.plat_name)
+
+        self.ensure_string_list('formats')
+        if self.formats is None:
+            try:
+                self.formats = [self.default_format[os.name]]
+            except KeyError:
+                raise DistutilsPlatformError(
+                      "don't know how to create built distributions "
+                      "on platform %s" % os.name)
+
+        if self.dist_dir is None:
+            self.dist_dir = "dist"
+
+    def run(self):
+        # Figure out which sub-commands we need to run.
+        commands = []
+        for format in self.formats:
+            try:
+                commands.append(self.format_command[format][0])
+            except KeyError:
+                raise DistutilsOptionError("invalid format '%s'" % format)
+
+        # Reinitialize and run each command.
+        for i in range(len(self.formats)):
+            cmd_name = commands[i]
+            sub_cmd = self.reinitialize_command(cmd_name)
+            if cmd_name not in self.no_format_option:
+                sub_cmd.format = self.formats[i]
+
+            # passing the owner and group names for tar archiving
+            if cmd_name == 'bdist_dumb':
+                sub_cmd.owner = self.owner
+                sub_cmd.group = self.group
+
+            # If we're going to need to run this command again, tell it to
+            # keep its temporary files around so subsequent runs go faster.
+            if cmd_name in commands[i+1:]:
+                sub_cmd.keep_temp = 1
+            self.run_command(cmd_name)
diff --git a/Lib/distutils/command/bdist_dumb.py b/Lib/distutils/command/bdist_dumb.py
new file mode 100644
index 000000000..f0d6b5b8c
--- /dev/null
+++ b/Lib/distutils/command/bdist_dumb.py
@@ -0,0 +1,123 @@
+"""distutils.command.bdist_dumb
+
+Implements the Distutils 'bdist_dumb' command (create a "dumb" built
+distribution -- i.e., just an archive to be unpacked under $prefix or
+$exec_prefix)."""
+
+import os
+from distutils.core import Command
+from distutils.util import get_platform
+from distutils.dir_util import remove_tree, ensure_relative
+from distutils.errors import *
+from distutils.sysconfig import get_python_version
+from distutils import log
+
+class bdist_dumb(Command):
+
+    description = "create a \"dumb\" built distribution"
+
+    user_options = [('bdist-dir=', 'd',
+                     "temporary directory for creating the distribution"),
+                    ('plat-name=', 'p',
+                     "platform name to embed in generated filenames "
+                     "(default: %s)" % get_platform()),
+                    ('format=', 'f',
+                     "archive format to create (tar, gztar, bztar, xztar, "
+                     "ztar, zip)"),
+                    ('keep-temp', 'k',
+                     "keep the pseudo-installation tree around after " +
+                     "creating the distribution archive"),
+                    ('dist-dir=', 'd',
+                     "directory to put final built distributions in"),
+                    ('skip-build', None,
+                     "skip rebuilding everything (for testing/debugging)"),
+                    ('relative', None,
+                     "build the archive using relative paths "
+                     "(default: false)"),
+                    ('owner=', 'u',
+                     "Owner name used when creating a tar file"
+                     " [default: current user]"),
+                    ('group=', 'g',
+                     "Group name used when creating a tar file"
+                     " [default: current group]"),
+                   ]
+
+    boolean_options = ['keep-temp', 'skip-build', 'relative']
+
+    default_format = { 'posix': 'gztar',
+                       'nt': 'zip' }
+
+    def initialize_options(self):
+        self.bdist_dir = None
+        self.plat_name = None
+        self.format = None
+        self.keep_temp = 0
+        self.dist_dir = None
+        self.skip_build = None
+        self.relative = 0
+        self.owner = None
+        self.group = None
+
+    def finalize_options(self):
+        if self.bdist_dir is None:
+            bdist_base = self.get_finalized_command('bdist').bdist_base
+            self.bdist_dir = os.path.join(bdist_base, 'dumb')
+
+        if self.format is None:
+            try:
+                self.format = self.default_format[os.name]
+            except KeyError:
+                raise DistutilsPlatformError(
+                       "don't know how to create dumb built distributions "
+                       "on platform %s" % os.name)
+
+        self.set_undefined_options('bdist',
+                                   ('dist_dir', 'dist_dir'),
+                                   ('plat_name', 'plat_name'),
+                                   ('skip_build', 'skip_build'))
+
+    def run(self):
+        if not self.skip_build:
+            self.run_command('build')
+
+        install = self.reinitialize_command('install', reinit_subcommands=1)
+        install.root = self.bdist_dir
+        install.skip_build = self.skip_build
+        install.warn_dir = 0
+
+        log.info("installing to %s", self.bdist_dir)
+        self.run_command('install')
+
+        # And make an archive relative to the root of the
+        # pseudo-installation tree.
+        archive_basename = "%s.%s" % (self.distribution.get_fullname(),
+                                      self.plat_name)
+
+        pseudoinstall_root = os.path.join(self.dist_dir, archive_basename)
+        if not self.relative:
+            archive_root = self.bdist_dir
+        else:
+            if (self.distribution.has_ext_modules() and
+                (install.install_base != install.install_platbase)):
+                raise DistutilsPlatformError(
+                       "can't make a dumb built distribution where "
+                       "base and platbase are different (%s, %s)"
+                       % (repr(install.install_base),
+                          repr(install.install_platbase)))
+            else:
+                archive_root = os.path.join(self.bdist_dir,
+                                   ensure_relative(install.install_base))
+
+        # Make the archive
+        filename = self.make_archive(pseudoinstall_root,
+                                     self.format, root_dir=archive_root,
+                                     owner=self.owner, group=self.group)
+        if self.distribution.has_ext_modules():
+            pyversion = get_python_version()
+        else:
+            pyversion = 'any'
+        self.distribution.dist_files.append(('bdist_dumb', pyversion,
+                                             filename))
+
+        if not self.keep_temp:
+            remove_tree(self.bdist_dir, dry_run=self.dry_run)
diff --git a/Lib/distutils/command/bdist_msi.py b/Lib/distutils/command/bdist_msi.py
new file mode 100644
index 000000000..80104c372
--- /dev/null
+++ b/Lib/distutils/command/bdist_msi.py
@@ -0,0 +1,741 @@
+# Copyright (C) 2005, 2006 Martin von Löwis
+# Licensed to PSF under a Contributor Agreement.
+# The bdist_wininst command proper
+# based on bdist_wininst
+"""
+Implements the bdist_msi command.
+"""
+
+import sys, os
+from distutils.core import Command
+from distutils.dir_util import remove_tree
+from distutils.sysconfig import get_python_version
+from distutils.version import StrictVersion
+from distutils.errors import DistutilsOptionError
+from distutils.util import get_platform
+from distutils import log
+import msilib
+from msilib import schema, sequence, text
+from msilib import Directory, Feature, Dialog, add_data
+
+class PyDialog(Dialog):
+    """Dialog class with a fixed layout: controls at the top, then a ruler,
+    then a list of buttons: back, next, cancel. Optionally a bitmap at the
+    left."""
+    def __init__(self, *args, **kw):
+        """Dialog(database, name, x, y, w, h, attributes, title, first,
+        default, cancel, bitmap=true)"""
+        Dialog.__init__(self, *args)
+        ruler = self.h - 36
+        bmwidth = 152*ruler/328
+        #if kw.get("bitmap", True):
+        #    self.bitmap("Bitmap", 0, 0, bmwidth, ruler, "PythonWin")
+        self.line("BottomLine", 0, ruler, self.w, 0)
+
+    def title(self, title):
+        "Set the title text of the dialog at the top."
+        # name, x, y, w, h, flags=Visible|Enabled|Transparent|NoPrefix,
+        # text, in VerdanaBold10
+        self.text("Title", 15, 10, 320, 60, 0x30003,
+                  r"{\VerdanaBold10}%s" % title)
+
+    def back(self, title, next, name = "Back", active = 1):
+        """Add a back button with a given title, the tab-next button,
+        its name in the Control table, possibly initially disabled.
+
+        Return the button, so that events can be associated"""
+        if active:
+            flags = 3 # Visible|Enabled
+        else:
+            flags = 1 # Visible
+        return self.pushbutton(name, 180, self.h-27 , 56, 17, flags, title, next)
+
+    def cancel(self, title, next, name = "Cancel", active = 1):
+        """Add a cancel button with a given title, the tab-next button,
+        its name in the Control table, possibly initially disabled.
+
+        Return the button, so that events can be associated"""
+        if active:
+            flags = 3 # Visible|Enabled
+        else:
+            flags = 1 # Visible
+        return self.pushbutton(name, 304, self.h-27, 56, 17, flags, title, next)
+
+    def next(self, title, next, name = "Next", active = 1):
+        """Add a Next button with a given title, the tab-next button,
+        its name in the Control table, possibly initially disabled.
+
+        Return the button, so that events can be associated"""
+        if active:
+            flags = 3 # Visible|Enabled
+        else:
+            flags = 1 # Visible
+        return self.pushbutton(name, 236, self.h-27, 56, 17, flags, title, next)
+
+    def xbutton(self, name, title, next, xpos):
+        """Add a button with a given title, the tab-next button,
+        its name in the Control table, giving its x position; the
+        y-position is aligned with the other buttons.
+
+        Return the button, so that events can be associated"""
+        return self.pushbutton(name, int(self.w*xpos - 28), self.h-27, 56, 17, 3, title, next)
+
+class bdist_msi(Command):
+
+    description = "create a Microsoft Installer (.msi) binary distribution"
+
+    user_options = [('bdist-dir=', None,
+                     "temporary directory for creating the distribution"),
+                    ('plat-name=', 'p',
+                     "platform name to embed in generated filenames "
+                     "(default: %s)" % get_platform()),
+                    ('keep-temp', 'k',
+                     "keep the pseudo-installation tree around after " +
+                     "creating the distribution archive"),
+                    ('target-version=', None,
+                     "require a specific python version" +
+                     " on the target system"),
+                    ('no-target-compile', 'c',
+                     "do not compile .py to .pyc on the target system"),
+                    ('no-target-optimize', 'o',
+                     "do not compile .py to .pyo (optimized) "
+                     "on the target system"),
+                    ('dist-dir=', 'd',
+                     "directory to put final built distributions in"),
+                    ('skip-build', None,
+                     "skip rebuilding everything (for testing/debugging)"),
+                    ('install-script=', None,
+                     "basename of installation script to be run after "
+                     "installation or before deinstallation"),
+                    ('pre-install-script=', None,
+                     "Fully qualified filename of a script to be run before "
+                     "any files are installed.  This script need not be in the "
+                     "distribution"),
+                   ]
+
+    boolean_options = ['keep-temp', 'no-target-compile', 'no-target-optimize',
+                       'skip-build']
+
+    all_versions = ['2.0', '2.1', '2.2', '2.3', '2.4',
+                    '2.5', '2.6', '2.7', '2.8', '2.9',
+                    '3.0', '3.1', '3.2', '3.3', '3.4',
+                    '3.5', '3.6', '3.7', '3.8', '3.9']
+    other_version = 'X'
+
+    def initialize_options(self):
+        self.bdist_dir = None
+        self.plat_name = None
+        self.keep_temp = 0
+        self.no_target_compile = 0
+        self.no_target_optimize = 0
+        self.target_version = None
+        self.dist_dir = None
+        self.skip_build = None
+        self.install_script = None
+        self.pre_install_script = None
+        self.versions = None
+
+    def finalize_options(self):
+        self.set_undefined_options('bdist', ('skip_build', 'skip_build'))
+
+        if self.bdist_dir is None:
+            bdist_base = self.get_finalized_command('bdist').bdist_base
+            self.bdist_dir = os.path.join(bdist_base, 'msi')
+
+        short_version = get_python_version()
+        if (not self.target_version) and self.distribution.has_ext_modules():
+            self.target_version = short_version
+
+        if self.target_version:
+            self.versions = [self.target_version]
+            if not self.skip_build and self.distribution.has_ext_modules()\
+               and self.target_version != short_version:
+                raise DistutilsOptionError(
+                      "target version can only be %s, or the '--skip-build'"
+                      " option must be specified" % (short_version,))
+        else:
+            self.versions = list(self.all_versions)
+
+        self.set_undefined_options('bdist',
+                                   ('dist_dir', 'dist_dir'),
+                                   ('plat_name', 'plat_name'),
+                                   )
+
+        if self.pre_install_script:
+            raise DistutilsOptionError(
+                  "the pre-install-script feature is not yet implemented")
+
+        if self.install_script:
+            for script in self.distribution.scripts:
+                if self.install_script == os.path.basename(script):
+                    break
+            else:
+                raise DistutilsOptionError(
+                      "install_script '%s' not found in scripts"
+                      % self.install_script)
+        self.install_script_key = None
+
+    def run(self):
+        if not self.skip_build:
+            self.run_command('build')
+
+        install = self.reinitialize_command('install', reinit_subcommands=1)
+        install.prefix = self.bdist_dir
+        install.skip_build = self.skip_build
+        install.warn_dir = 0
+
+        install_lib = self.reinitialize_command('install_lib')
+        # we do not want to include pyc or pyo files
+        install_lib.compile = 0
+        install_lib.optimize = 0
+
+        if self.distribution.has_ext_modules():
+            # If we are building an installer for a Python version other
+            # than the one we are currently running, then we need to ensure
+            # our build_lib reflects the other Python version rather than ours.
+            # Note that for target_version!=sys.version, we must have skipped the
+            # build step, so there is no issue with enforcing the build of this
+            # version.
+            target_version = self.target_version
+            if not target_version:
+                assert self.skip_build, "Should have already checked this"
+                target_version = '%d.%d' % sys.version_info[:2]
+            plat_specifier = ".%s-%s" % (self.plat_name, target_version)
+            build = self.get_finalized_command('build')
+            build.build_lib = os.path.join(build.build_base,
+                                           'lib' + plat_specifier)
+
+        log.info("installing to %s", self.bdist_dir)
+        install.ensure_finalized()
+
+        # avoid warning of 'install_lib' about installing
+        # into a directory not in sys.path
+        sys.path.insert(0, os.path.join(self.bdist_dir, 'PURELIB'))
+
+        install.run()
+
+        del sys.path[0]
+
+        self.mkpath(self.dist_dir)
+        fullname = self.distribution.get_fullname()
+        installer_name = self.get_installer_filename(fullname)
+        installer_name = os.path.abspath(installer_name)
+        if os.path.exists(installer_name): os.unlink(installer_name)
+
+        metadata = self.distribution.metadata
+        author = metadata.author
+        if not author:
+            author = metadata.maintainer
+        if not author:
+            author = "UNKNOWN"
+        version = metadata.get_version()
+        # ProductVersion must be strictly numeric
+        # XXX need to deal with prerelease versions
+        sversion = "%d.%d.%d" % StrictVersion(version).version
+        # Prefix ProductName with Python x.y, so that
+        # it sorts together with the other Python packages
+        # in Add-Remove-Programs (APR)
+        fullname = self.distribution.get_fullname()
+        if self.target_version:
+            product_name = "Python %s %s" % (self.target_version, fullname)
+        else:
+            product_name = "Python %s" % (fullname)
+        self.db = msilib.init_database(installer_name, schema,
+                product_name, msilib.gen_uuid(),
+                sversion, author)
+        msilib.add_tables(self.db, sequence)
+        props = [('DistVersion', version)]
+        email = metadata.author_email or metadata.maintainer_email
+        if email:
+            props.append(("ARPCONTACT", email))
+        if metadata.url:
+            props.append(("ARPURLINFOABOUT", metadata.url))
+        if props:
+            add_data(self.db, 'Property', props)
+
+        self.add_find_python()
+        self.add_files()
+        self.add_scripts()
+        self.add_ui()
+        self.db.Commit()
+
+        if hasattr(self.distribution, 'dist_files'):
+            tup = 'bdist_msi', self.target_version or 'any', fullname
+            self.distribution.dist_files.append(tup)
+
+        if not self.keep_temp:
+            remove_tree(self.bdist_dir, dry_run=self.dry_run)
+
+    def add_files(self):
+        db = self.db
+        cab = msilib.CAB("distfiles")
+        rootdir = os.path.abspath(self.bdist_dir)
+
+        root = Directory(db, cab, None, rootdir, "TARGETDIR", "SourceDir")
+        f = Feature(db, "Python", "Python", "Everything",
+                    0, 1, directory="TARGETDIR")
+
+        items = [(f, root, '')]
+        for version in self.versions + [self.other_version]:
+            target = "TARGETDIR" + version
+            name = default = "Python" + version
+            desc = "Everything"
+            if version is self.other_version:
+                title = "Python from another location"
+                level = 2
+            else:
+                title = "Python %s from registry" % version
+                level = 1
+            f = Feature(db, name, title, desc, 1, level, directory=target)
+            dir = Directory(db, cab, root, rootdir, target, default)
+            items.append((f, dir, version))
+        db.Commit()
+
+        seen = {}
+        for feature, dir, version in items:
+            todo = [dir]
+            while todo:
+                dir = todo.pop()
+                for file in os.listdir(dir.absolute):
+                    afile = os.path.join(dir.absolute, file)
+                    if os.path.isdir(afile):
+                        short = "%s|%s" % (dir.make_short(file), file)
+                        default = file + version
+                        newdir = Directory(db, cab, dir, file, default, short)
+                        todo.append(newdir)
+                    else:
+                        if not dir.component:
+                            dir.start_component(dir.logical, feature, 0)
+                        if afile not in seen:
+                            key = seen[afile] = dir.add_file(file)
+                            if file==self.install_script:
+                                if self.install_script_key:
+                                    raise DistutilsOptionError(
+                                          "Multiple files with name %s" % file)
+                                self.install_script_key = '[#%s]' % key
+                        else:
+                            key = seen[afile]
+                            add_data(self.db, "DuplicateFile",
+                                [(key + version, dir.component, key, None, dir.logical)])
+            db.Commit()
+        cab.commit(db)
+
+    def add_find_python(self):
+        """Adds code to the installer to compute the location of Python.
+
+        Properties PYTHON.MACHINE.X.Y and PYTHON.USER.X.Y will be set from the
+        registry for each version of Python.
+
+        Properties TARGETDIRX.Y will be set from PYTHON.USER.X.Y if defined,
+        else from PYTHON.MACHINE.X.Y.
+
+        Properties PYTHONX.Y will be set to TARGETDIRX.Y\\python.exe"""
+
+        start = 402
+        for ver in self.versions:
+            install_path = r"SOFTWARE\Python\PythonCore\%s\InstallPath" % ver
+            machine_reg = "python.machine." + ver
+            user_reg = "python.user." + ver
+            machine_prop = "PYTHON.MACHINE." + ver
+            user_prop = "PYTHON.USER." + ver
+            machine_action = "PythonFromMachine" + ver
+            user_action = "PythonFromUser" + ver
+            exe_action = "PythonExe" + ver
+            target_dir_prop = "TARGETDIR" + ver
+            exe_prop = "PYTHON" + ver
+            if msilib.Win64:
+                # type: msidbLocatorTypeRawValue + msidbLocatorType64bit
+                Type = 2+16
+            else:
+                Type = 2
+            add_data(self.db, "RegLocator",
+                    [(machine_reg, 2, install_path, None, Type),
+                     (user_reg, 1, install_path, None, Type)])
+            add_data(self.db, "AppSearch",
+                    [(machine_prop, machine_reg),
+                     (user_prop, user_reg)])
+            add_data(self.db, "CustomAction",
+                    [(machine_action, 51+256, target_dir_prop, "[" + machine_prop + "]"),
+                     (user_action, 51+256, target_dir_prop, "[" + user_prop + "]"),
+                     (exe_action, 51+256, exe_prop, "[" + target_dir_prop + "]\\python.exe"),
+                    ])
+            add_data(self.db, "InstallExecuteSequence",
+                    [(machine_action, machine_prop, start),
+                     (user_action, user_prop, start + 1),
+                     (exe_action, None, start + 2),
+                    ])
+            add_data(self.db, "InstallUISequence",
+                    [(machine_action, machine_prop, start),
+                     (user_action, user_prop, start + 1),
+                     (exe_action, None, start + 2),
+                    ])
+            add_data(self.db, "Condition",
+                    [("Python" + ver, 0, "NOT TARGETDIR" + ver)])
+            start += 4
+            assert start < 500
+
+    def add_scripts(self):
+        if self.install_script:
+            start = 6800
+            for ver in self.versions + [self.other_version]:
+                install_action = "install_script." + ver
+                exe_prop = "PYTHON" + ver
+                add_data(self.db, "CustomAction",
+                        [(install_action, 50, exe_prop, self.install_script_key)])
+                add_data(self.db, "InstallExecuteSequence",
+                        [(install_action, "&Python%s=3" % ver, start)])
+                start += 1
+        # XXX pre-install scripts are currently refused in finalize_options()
+        #     but if this feature is completed, it will also need to add
+        #     entries for each version as the above code does
+        if self.pre_install_script:
+            scriptfn = os.path.join(self.bdist_dir, "preinstall.bat")
+            f = open(scriptfn, "w")
+            # The batch file will be executed with [PYTHON], so that %1
+            # is the path to the Python interpreter; %0 will be the path
+            # of the batch file.
+            # rem ="""
+            # %1 %0
+            # exit
+            # """
+            # <actual script>
+            f.write('rem ="""\n%1 %0\nexit\n"""\n')
+            f.write(open(self.pre_install_script).read())
+            f.close()
+            add_data(self.db, "Binary",
+                [("PreInstall", msilib.Binary(scriptfn))
+                ])
+            add_data(self.db, "CustomAction",
+                [("PreInstall", 2, "PreInstall", None)
+                ])
+            add_data(self.db, "InstallExecuteSequence",
+                    [("PreInstall", "NOT Installed", 450)])
+
+
+    def add_ui(self):
+        db = self.db
+        x = y = 50
+        w = 370
+        h = 300
+        title = "[ProductName] Setup"
+
+        # see "Dialog Style Bits"
+        modal = 3      # visible | modal
+        modeless = 1   # visible
+        track_disk_space = 32
+
+        # UI customization properties
+        add_data(db, "Property",
+                 # See "DefaultUIFont Property"
+                 [("DefaultUIFont", "DlgFont8"),
+                  # See "ErrorDialog Style Bit"
+                  ("ErrorDialog", "ErrorDlg"),
+                  ("Progress1", "Install"),   # modified in maintenance type dlg
+                  ("Progress2", "installs"),
+                  ("MaintenanceForm_Action", "Repair"),
+                  # possible values: ALL, JUSTME
+                  ("WhichUsers", "ALL")
+                 ])
+
+        # Fonts, see "TextStyle Table"
+        add_data(db, "TextStyle",
+                 [("DlgFont8", "Tahoma", 9, None, 0),
+                  ("DlgFontBold8", "Tahoma", 8, None, 1), #bold
+                  ("VerdanaBold10", "Verdana", 10, None, 1),
+                  ("VerdanaRed9", "Verdana", 9, 255, 0),
+                 ])
+
+        # UI Sequences, see "InstallUISequence Table", "Using a Sequence Table"
+        # Numbers indicate sequence; see sequence.py for how these action integrate
+        add_data(db, "InstallUISequence",
+                 [("PrepareDlg", "Not Privileged or Windows9x or Installed", 140),
+                  ("WhichUsersDlg", "Privileged and not Windows9x and not Installed", 141),
+                  # In the user interface, assume all-users installation if privileged.
+                  ("SelectFeaturesDlg", "Not Installed", 1230),
+                  # XXX no support for resume installations yet
+                  #("ResumeDlg", "Installed AND (RESUME OR Preselected)", 1240),
+                  ("MaintenanceTypeDlg", "Installed AND NOT RESUME AND NOT Preselected", 1250),
+                  ("ProgressDlg", None, 1280)])
+
+        add_data(db, 'ActionText', text.ActionText)
+        add_data(db, 'UIText', text.UIText)
+        #####################################################################
+        # Standard dialogs: FatalError, UserExit, ExitDialog
+        fatal=PyDialog(db, "FatalError", x, y, w, h, modal, title,
+                     "Finish", "Finish", "Finish")
+        fatal.title("[ProductName] Installer ended prematurely")
+        fatal.back("< Back", "Finish", active = 0)
+        fatal.cancel("Cancel", "Back", active = 0)
+        fatal.text("Description1", 15, 70, 320, 80, 0x30003,
+                   "[ProductName] setup ended prematurely because of an error.  Your system has not been modified.  To install this program at a later time, please run the installation again.")
+        fatal.text("Description2", 15, 155, 320, 20, 0x30003,
+                   "Click the Finish button to exit the Installer.")
+        c=fatal.next("Finish", "Cancel", name="Finish")
+        c.event("EndDialog", "Exit")
+
+        user_exit=PyDialog(db, "UserExit", x, y, w, h, modal, title,
+                     "Finish", "Finish", "Finish")
+        user_exit.title("[ProductName] Installer was interrupted")
+        user_exit.back("< Back", "Finish", active = 0)
+        user_exit.cancel("Cancel", "Back", active = 0)
+        user_exit.text("Description1", 15, 70, 320, 80, 0x30003,
+                   "[ProductName] setup was interrupted.  Your system has not been modified.  "
+                   "To install this program at a later time, please run the installation again.")
+        user_exit.text("Description2", 15, 155, 320, 20, 0x30003,
+                   "Click the Finish button to exit the Installer.")
+        c = user_exit.next("Finish", "Cancel", name="Finish")
+        c.event("EndDialog", "Exit")
+
+        exit_dialog = PyDialog(db, "ExitDialog", x, y, w, h, modal, title,
+                             "Finish", "Finish", "Finish")
+        exit_dialog.title("Completing the [ProductName] Installer")
+        exit_dialog.back("< Back", "Finish", active = 0)
+        exit_dialog.cancel("Cancel", "Back", active = 0)
+        exit_dialog.text("Description", 15, 235, 320, 20, 0x30003,
+                   "Click the Finish button to exit the Installer.")
+        c = exit_dialog.next("Finish", "Cancel", name="Finish")
+        c.event("EndDialog", "Return")
+
+        #####################################################################
+        # Required dialog: FilesInUse, ErrorDlg
+        inuse = PyDialog(db, "FilesInUse",
+                         x, y, w, h,
+                         19,                # KeepModeless|Modal|Visible
+                         title,
+                         "Retry", "Retry", "Retry", bitmap=False)
+        inuse.text("Title", 15, 6, 200, 15, 0x30003,
+                   r"{\DlgFontBold8}Files in Use")
+        inuse.text("Description", 20, 23, 280, 20, 0x30003,
+               "Some files that need to be updated are currently in use.")
+        inuse.text("Text", 20, 55, 330, 50, 3,
+                   "The following applications are using files that need to be updated by this setup. Close these applications and then click Retry to continue the installation or Cancel to exit it.")
+        inuse.control("List", "ListBox", 20, 107, 330, 130, 7, "FileInUseProcess",
+                      None, None, None)
+        c=inuse.back("Exit", "Ignore", name="Exit")
+        c.event("EndDialog", "Exit")
+        c=inuse.next("Ignore", "Retry", name="Ignore")
+        c.event("EndDialog", "Ignore")
+        c=inuse.cancel("Retry", "Exit", name="Retry")
+        c.event("EndDialog","Retry")
+
+        # See "Error Dialog". See "ICE20" for the required names of the controls.
+        error = Dialog(db, "ErrorDlg",
+                       50, 10, 330, 101,
+                       65543,       # Error|Minimize|Modal|Visible
+                       title,
+                       "ErrorText", None, None)
+        error.text("ErrorText", 50,9,280,48,3, "")
+        #error.control("ErrorIcon", "Icon", 15, 9, 24, 24, 5242881, None, "py.ico", None, None)
+        error.pushbutton("N",120,72,81,21,3,"No",None).event("EndDialog","ErrorNo")
+        error.pushbutton("Y",240,72,81,21,3,"Yes",None).event("EndDialog","ErrorYes")
+        error.pushbutton("A",0,72,81,21,3,"Abort",None).event("EndDialog","ErrorAbort")
+        error.pushbutton("C",42,72,81,21,3,"Cancel",None).event("EndDialog","ErrorCancel")
+        error.pushbutton("I",81,72,81,21,3,"Ignore",None).event("EndDialog","ErrorIgnore")
+        error.pushbutton("O",159,72,81,21,3,"Ok",None).event("EndDialog","ErrorOk")
+        error.pushbutton("R",198,72,81,21,3,"Retry",None).event("EndDialog","ErrorRetry")
+
+        #####################################################################
+        # Global "Query Cancel" dialog
+        cancel = Dialog(db, "CancelDlg", 50, 10, 260, 85, 3, title,
+                        "No", "No", "No")
+        cancel.text("Text", 48, 15, 194, 30, 3,
+                    "Are you sure you want to cancel [ProductName] installation?")
+        #cancel.control("Icon", "Icon", 15, 15, 24, 24, 5242881, None,
+        #               "py.ico", None, None)
+        c=cancel.pushbutton("Yes", 72, 57, 56, 17, 3, "Yes", "No")
+        c.event("EndDialog", "Exit")
+
+        c=cancel.pushbutton("No", 132, 57, 56, 17, 3, "No", "Yes")
+        c.event("EndDialog", "Return")
+
+        #####################################################################
+        # Global "Wait for costing" dialog
+        costing = Dialog(db, "WaitForCostingDlg", 50, 10, 260, 85, modal, title,
+                         "Return", "Return", "Return")
+        costing.text("Text", 48, 15, 194, 30, 3,
+                     "Please wait while the installer finishes determining your disk space requirements.")
+        c = costing.pushbutton("Return", 102, 57, 56, 17, 3, "Return", None)
+        c.event("EndDialog", "Exit")
+
+        #####################################################################
+        # Preparation dialog: no user input except cancellation
+        prep = PyDialog(db, "PrepareDlg", x, y, w, h, modeless, title,
+                        "Cancel", "Cancel", "Cancel")
+        prep.text("Description", 15, 70, 320, 40, 0x30003,
+                  "Please wait while the Installer prepares to guide you through the installation.")
+        prep.title("Welcome to the [ProductName] Installer")
+        c=prep.text("ActionText", 15, 110, 320, 20, 0x30003, "Pondering...")
+        c.mapping("ActionText", "Text")
+        c=prep.text("ActionData", 15, 135, 320, 30, 0x30003, None)
+        c.mapping("ActionData", "Text")
+        prep.back("Back", None, active=0)
+        prep.next("Next", None, active=0)
+        c=prep.cancel("Cancel", None)
+        c.event("SpawnDialog", "CancelDlg")
+
+        #####################################################################
+        # Feature (Python directory) selection
+        seldlg = PyDialog(db, "SelectFeaturesDlg", x, y, w, h, modal, title,
+                        "Next", "Next", "Cancel")
+        seldlg.title("Select Python Installations")
+
+        seldlg.text("Hint", 15, 30, 300, 20, 3,
+                    "Select the Python locations where %s should be installed."
+                    % self.distribution.get_fullname())
+
+        seldlg.back("< Back", None, active=0)
+        c = seldlg.next("Next >", "Cancel")
+        order = 1
+        c.event("[TARGETDIR]", "[SourceDir]", ordering=order)
+        for version in self.versions + [self.other_version]:
+            order += 1
+            c.event("[TARGETDIR]", "[TARGETDIR%s]" % version,
+                    "FEATURE_SELECTED AND &Python%s=3" % version,
+                    ordering=order)
+        c.event("SpawnWaitDialog", "WaitForCostingDlg", ordering=order + 1)
+        c.event("EndDialog", "Return", ordering=order + 2)
+        c = seldlg.cancel("Cancel", "Features")
+        c.event("SpawnDialog", "CancelDlg")
+
+        c = seldlg.control("Features", "SelectionTree", 15, 60, 300, 120, 3,
+                           "FEATURE", None, "PathEdit", None)
+        c.event("[FEATURE_SELECTED]", "1")
+        ver = self.other_version
+        install_other_cond = "FEATURE_SELECTED AND &Python%s=3" % ver
+        dont_install_other_cond = "FEATURE_SELECTED AND &Python%s<>3" % ver
+
+        c = seldlg.text("Other", 15, 200, 300, 15, 3,
+                        "Provide an alternate Python location")
+        c.condition("Enable", install_other_cond)
+        c.condition("Show", install_other_cond)
+        c.condition("Disable", dont_install_other_cond)
+        c.condition("Hide", dont_install_other_cond)
+
+        c = seldlg.control("PathEdit", "PathEdit", 15, 215, 300, 16, 1,
+                           "TARGETDIR" + ver, None, "Next", None)
+        c.condition("Enable", install_other_cond)
+        c.condition("Show", install_other_cond)
+        c.condition("Disable", dont_install_other_cond)
+        c.condition("Hide", dont_install_other_cond)
+
+        #####################################################################
+        # Disk cost
+        cost = PyDialog(db, "DiskCostDlg", x, y, w, h, modal, title,
+                        "OK", "OK", "OK", bitmap=False)
+        cost.text("Title", 15, 6, 200, 15, 0x30003,
+                 r"{\DlgFontBold8}Disk Space Requirements")
+        cost.text("Description", 20, 20, 280, 20, 0x30003,
+                  "The disk space required for the installation of the selected features.")
+        cost.text("Text", 20, 53, 330, 60, 3,
+                  "The highlighted volumes (if any) do not have enough disk space "
+              "available for the currently selected features.  You can either "
+              "remove some files from the highlighted volumes, or choose to "
+              "install less features onto local drive(s), or select different "
+              "destination drive(s).")
+        cost.control("VolumeList", "VolumeCostList", 20, 100, 330, 150, 393223,
+                     None, "{120}{70}{70}{70}{70}", None, None)
+        cost.xbutton("OK", "Ok", None, 0.5).event("EndDialog", "Return")
+
+        #####################################################################
+        # WhichUsers Dialog. Only available on NT, and for privileged users.
+        # This must be run before FindRelatedProducts, because that will
+        # take into account whether the previous installation was per-user
+        # or per-machine. We currently don't support going back to this
+        # dialog after "Next" was selected; to support this, we would need to
+        # find how to reset the ALLUSERS property, and how to re-run
+        # FindRelatedProducts.
+        # On Windows9x, the ALLUSERS property is ignored on the command line
+        # and in the Property table, but installer fails according to the documentation
+        # if a dialog attempts to set ALLUSERS.
+        whichusers = PyDialog(db, "WhichUsersDlg", x, y, w, h, modal, title,
+                            "AdminInstall", "Next", "Cancel")
+        whichusers.title("Select whether to install [ProductName] for all users of this computer.")
+        # A radio group with two options: allusers, justme
+        g = whichusers.radiogroup("AdminInstall", 15, 60, 260, 50, 3,
+                                  "WhichUsers", "", "Next")
+        g.add("ALL", 0, 5, 150, 20, "Install for all users")
+        g.add("JUSTME", 0, 25, 150, 20, "Install just for me")
+
+        whichusers.back("Back", None, active=0)
+
+        c = whichusers.next("Next >", "Cancel")
+        c.event("[ALLUSERS]", "1", 'WhichUsers="ALL"', 1)
+        c.event("EndDialog", "Return", ordering = 2)
+
+        c = whichusers.cancel("Cancel", "AdminInstall")
+        c.event("SpawnDialog", "CancelDlg")
+
+        #####################################################################
+        # Installation Progress dialog (modeless)
+        progress = PyDialog(db, "ProgressDlg", x, y, w, h, modeless, title,
+                            "Cancel", "Cancel", "Cancel", bitmap=False)
+        progress.text("Title", 20, 15, 200, 15, 0x30003,
+                     r"{\DlgFontBold8}[Progress1] [ProductName]")
+        progress.text("Text", 35, 65, 300, 30, 3,
+                      "Please wait while the Installer [Progress2] [ProductName]. "
+                      "This may take several minutes.")
+        progress.text("StatusLabel", 35, 100, 35, 20, 3, "Status:")
+
+        c=progress.text("ActionText", 70, 100, w-70, 20, 3, "Pondering...")
+        c.mapping("ActionText", "Text")
+
+        #c=progress.text("ActionData", 35, 140, 300, 20, 3, None)
+        #c.mapping("ActionData", "Text")
+
+        c=progress.control("ProgressBar", "ProgressBar", 35, 120, 300, 10, 65537,
+                           None, "Progress done", None, None)
+        c.mapping("SetProgress", "Progress")
+
+        progress.back("< Back", "Next", active=False)
+        progress.next("Next >", "Cancel", active=False)
+        progress.cancel("Cancel", "Back").event("SpawnDialog", "CancelDlg")
+
+        ###################################################################
+        # Maintenance type: repair/uninstall
+        maint = PyDialog(db, "MaintenanceTypeDlg", x, y, w, h, modal, title,
+                         "Next", "Next", "Cancel")
+        maint.title("Welcome to the [ProductName] Setup Wizard")
+        maint.text("BodyText", 15, 63, 330, 42, 3,
+                   "Select whether you want to repair or remove [ProductName].")
+        g=maint.radiogroup("RepairRadioGroup", 15, 108, 330, 60, 3,
+                            "MaintenanceForm_Action", "", "Next")
+        #g.add("Change", 0, 0, 200, 17, "&Change [ProductName]")
+        g.add("Repair", 0, 18, 200, 17, "&Repair [ProductName]")
+        g.add("Remove", 0, 36, 200, 17, "Re&move [ProductName]")
+
+        maint.back("< Back", None, active=False)
+        c=maint.next("Finish", "Cancel")
+        # Change installation: Change progress dialog to "Change", then ask
+        # for feature selection
+        #c.event("[Progress1]", "Change", 'MaintenanceForm_Action="Change"', 1)
+        #c.event("[Progress2]", "changes", 'MaintenanceForm_Action="Change"', 2)
+
+        # Reinstall: Change progress dialog to "Repair", then invoke reinstall
+        # Also set list of reinstalled features to "ALL"
+        c.event("[REINSTALL]", "ALL", 'MaintenanceForm_Action="Repair"', 5)
+        c.event("[Progress1]", "Repairing", 'MaintenanceForm_Action="Repair"', 6)
+        c.event("[Progress2]", "repairs", 'MaintenanceForm_Action="Repair"', 7)
+        c.event("Reinstall", "ALL", 'MaintenanceForm_Action="Repair"', 8)
+
+        # Uninstall: Change progress to "Remove", then invoke uninstall
+        # Also set list of removed features to "ALL"
+        c.event("[REMOVE]", "ALL", 'MaintenanceForm_Action="Remove"', 11)
+        c.event("[Progress1]", "Removing", 'MaintenanceForm_Action="Remove"', 12)
+        c.event("[Progress2]", "removes", 'MaintenanceForm_Action="Remove"', 13)
+        c.event("Remove", "ALL", 'MaintenanceForm_Action="Remove"', 14)
+
+        # Close dialog when maintenance action scheduled
+        c.event("EndDialog", "Return", 'MaintenanceForm_Action<>"Change"', 20)
+        #c.event("NewDialog", "SelectFeaturesDlg", 'MaintenanceForm_Action="Change"', 21)
+
+        maint.cancel("Cancel", "RepairRadioGroup").event("SpawnDialog", "CancelDlg")
+
+    def get_installer_filename(self, fullname):
+        # Factored out to allow overriding in subclasses
+        if self.target_version:
+            base_name = "%s.%s-py%s.msi" % (fullname, self.plat_name,
+                                            self.target_version)
+        else:
+            base_name = "%s.%s.msi" % (fullname, self.plat_name)
+        installer_name = os.path.join(self.dist_dir, base_name)
+        return installer_name
diff --git a/Lib/distutils/command/bdist_rpm.py b/Lib/distutils/command/bdist_rpm.py
new file mode 100644
index 000000000..02f10dd89
--- /dev/null
+++ b/Lib/distutils/command/bdist_rpm.py
@@ -0,0 +1,582 @@
+"""distutils.command.bdist_rpm
+
+Implements the Distutils 'bdist_rpm' command (create RPM source and binary
+distributions)."""
+
+import subprocess, sys, os
+from distutils.core import Command
+from distutils.debug import DEBUG
+from distutils.util import get_platform
+from distutils.file_util import write_file
+from distutils.errors import *
+from distutils.sysconfig import get_python_version
+from distutils import log
+
+class bdist_rpm(Command):
+
+    description = "create an RPM distribution"
+
+    user_options = [
+        ('bdist-base=', None,
+         "base directory for creating built distributions"),
+        ('rpm-base=', None,
+         "base directory for creating RPMs (defaults to \"rpm\" under "
+         "--bdist-base; must be specified for RPM 2)"),
+        ('dist-dir=', 'd',
+         "directory to put final RPM files in "
+         "(and .spec files if --spec-only)"),
+        ('python=', None,
+         "path to Python interpreter to hard-code in the .spec file "
+         "(default: \"python\")"),
+        ('fix-python', None,
+         "hard-code the exact path to the current Python interpreter in "
+         "the .spec file"),
+        ('spec-only', None,
+         "only regenerate spec file"),
+        ('source-only', None,
+         "only generate source RPM"),
+        ('binary-only', None,
+         "only generate binary RPM"),
+        ('use-bzip2', None,
+         "use bzip2 instead of gzip to create source distribution"),
+
+        # More meta-data: too RPM-specific to put in the setup script,
+        # but needs to go in the .spec file -- so we make these options
+        # to "bdist_rpm".  The idea is that packagers would put this
+        # info in setup.cfg, although they are of course free to
+        # supply it on the command line.
+        ('distribution-name=', None,
+         "name of the (Linux) distribution to which this "
+         "RPM applies (*not* the name of the module distribution!)"),
+        ('group=', None,
+         "package classification [default: \"Development/Libraries\"]"),
+        ('release=', None,
+         "RPM release number"),
+        ('serial=', None,
+         "RPM serial number"),
+        ('vendor=', None,
+         "RPM \"vendor\" (eg. \"Joe Blow <joe@example.com>\") "
+         "[default: maintainer or author from setup script]"),
+        ('packager=', None,
+         "RPM packager (eg. \"Jane Doe <jane@example.net>\") "
+         "[default: vendor]"),
+        ('doc-files=', None,
+         "list of documentation files (space or comma-separated)"),
+        ('changelog=', None,
+         "RPM changelog"),
+        ('icon=', None,
+         "name of icon file"),
+        ('provides=', None,
+         "capabilities provided by this package"),
+        ('requires=', None,
+         "capabilities required by this package"),
+        ('conflicts=', None,
+         "capabilities which conflict with this package"),
+        ('build-requires=', None,
+         "capabilities required to build this package"),
+        ('obsoletes=', None,
+         "capabilities made obsolete by this package"),
+        ('no-autoreq', None,
+         "do not automatically calculate dependencies"),
+
+        # Actions to take when building RPM
+        ('keep-temp', 'k',
+         "don't clean up RPM build directory"),
+        ('no-keep-temp', None,
+         "clean up RPM build directory [default]"),
+        ('use-rpm-opt-flags', None,
+         "compile with RPM_OPT_FLAGS when building from source RPM"),
+        ('no-rpm-opt-flags', None,
+         "do not pass any RPM CFLAGS to compiler"),
+        ('rpm3-mode', None,
+         "RPM 3 compatibility mode (default)"),
+        ('rpm2-mode', None,
+         "RPM 2 compatibility mode"),
+
+        # Add the hooks necessary for specifying custom scripts
+        ('prep-script=', None,
+         "Specify a script for the PREP phase of RPM building"),
+        ('build-script=', None,
+         "Specify a script for the BUILD phase of RPM building"),
+
+        ('pre-install=', None,
+         "Specify a script for the pre-INSTALL phase of RPM building"),
+        ('install-script=', None,
+         "Specify a script for the INSTALL phase of RPM building"),
+        ('post-install=', None,
+         "Specify a script for the post-INSTALL phase of RPM building"),
+
+        ('pre-uninstall=', None,
+         "Specify a script for the pre-UNINSTALL phase of RPM building"),
+        ('post-uninstall=', None,
+         "Specify a script for the post-UNINSTALL phase of RPM building"),
+
+        ('clean-script=', None,
+         "Specify a script for the CLEAN phase of RPM building"),
+
+        ('verify-script=', None,
+         "Specify a script for the VERIFY phase of the RPM build"),
+
+        # Allow a packager to explicitly force an architecture
+        ('force-arch=', None,
+         "Force an architecture onto the RPM build process"),
+
+        ('quiet', 'q',
+         "Run the INSTALL phase of RPM building in quiet mode"),
+        ]
+
+    boolean_options = ['keep-temp', 'use-rpm-opt-flags', 'rpm3-mode',
+                       'no-autoreq', 'quiet']
+
+    negative_opt = {'no-keep-temp': 'keep-temp',
+                    'no-rpm-opt-flags': 'use-rpm-opt-flags',
+                    'rpm2-mode': 'rpm3-mode'}
+
+
+    def initialize_options(self):
+        self.bdist_base = None
+        self.rpm_base = None
+        self.dist_dir = None
+        self.python = None
+        self.fix_python = None
+        self.spec_only = None
+        self.binary_only = None
+        self.source_only = None
+        self.use_bzip2 = None
+
+        self.distribution_name = None
+        self.group = None
+        self.release = None
+        self.serial = None
+        self.vendor = None
+        self.packager = None
+        self.doc_files = None
+        self.changelog = None
+        self.icon = None
+
+        self.prep_script = None
+        self.build_script = None
+        self.install_script = None
+        self.clean_script = None
+        self.verify_script = None
+        self.pre_install = None
+        self.post_install = None
+        self.pre_uninstall = None
+        self.post_uninstall = None
+        self.prep = None
+        self.provides = None
+        self.requires = None
+        self.conflicts = None
+        self.build_requires = None
+        self.obsoletes = None
+
+        self.keep_temp = 0
+        self.use_rpm_opt_flags = 1
+        self.rpm3_mode = 1
+        self.no_autoreq = 0
+
+        self.force_arch = None
+        self.quiet = 0
+
+    def finalize_options(self):
+        self.set_undefined_options('bdist', ('bdist_base', 'bdist_base'))
+        if self.rpm_base is None:
+            if not self.rpm3_mode:
+                raise DistutilsOptionError(
+                      "you must specify --rpm-base in RPM 2 mode")
+            self.rpm_base = os.path.join(self.bdist_base, "rpm")
+
+        if self.python is None:
+            if self.fix_python:
+                self.python = sys.executable
+            else:
+                self.python = "python3"
+        elif self.fix_python:
+            raise DistutilsOptionError(
+                  "--python and --fix-python are mutually exclusive options")
+
+        if os.name != 'posix':
+            raise DistutilsPlatformError("don't know how to create RPM "
+                   "distributions on platform %s" % os.name)
+        if self.binary_only and self.source_only:
+            raise DistutilsOptionError(
+                  "cannot supply both '--source-only' and '--binary-only'")
+
+        # don't pass CFLAGS to pure python distributions
+        if not self.distribution.has_ext_modules():
+            self.use_rpm_opt_flags = 0
+
+        self.set_undefined_options('bdist', ('dist_dir', 'dist_dir'))
+        self.finalize_package_data()
+
+    def finalize_package_data(self):
+        self.ensure_string('group', "Development/Libraries")
+        self.ensure_string('vendor',
+                           "%s <%s>" % (self.distribution.get_contact(),
+                                        self.distribution.get_contact_email()))
+        self.ensure_string('packager')
+        self.ensure_string_list('doc_files')
+        if isinstance(self.doc_files, list):
+            for readme in ('README', 'README.txt'):
+                if os.path.exists(readme) and readme not in self.doc_files:
+                    self.doc_files.append(readme)
+
+        self.ensure_string('release', "1")
+        self.ensure_string('serial')   # should it be an int?
+
+        self.ensure_string('distribution_name')
+
+        self.ensure_string('changelog')
+          # Format changelog correctly
+        self.changelog = self._format_changelog(self.changelog)
+
+        self.ensure_filename('icon')
+
+        self.ensure_filename('prep_script')
+        self.ensure_filename('build_script')
+        self.ensure_filename('install_script')
+        self.ensure_filename('clean_script')
+        self.ensure_filename('verify_script')
+        self.ensure_filename('pre_install')
+        self.ensure_filename('post_install')
+        self.ensure_filename('pre_uninstall')
+        self.ensure_filename('post_uninstall')
+
+        # XXX don't forget we punted on summaries and descriptions -- they
+        # should be handled here eventually!
+
+        # Now *this* is some meta-data that belongs in the setup script...
+        self.ensure_string_list('provides')
+        self.ensure_string_list('requires')
+        self.ensure_string_list('conflicts')
+        self.ensure_string_list('build_requires')
+        self.ensure_string_list('obsoletes')
+
+        self.ensure_string('force_arch')
+
+    def run(self):
+        if DEBUG:
+            print("before _get_package_data():")
+            print("vendor =", self.vendor)
+            print("packager =", self.packager)
+            print("doc_files =", self.doc_files)
+            print("changelog =", self.changelog)
+
+        # make directories
+        if self.spec_only:
+            spec_dir = self.dist_dir
+            self.mkpath(spec_dir)
+        else:
+            rpm_dir = {}
+            for d in ('SOURCES', 'SPECS', 'BUILD', 'RPMS', 'SRPMS'):
+                rpm_dir[d] = os.path.join(self.rpm_base, d)
+                self.mkpath(rpm_dir[d])
+            spec_dir = rpm_dir['SPECS']
+
+        # Spec file goes into 'dist_dir' if '--spec-only specified',
+        # build/rpm.<plat> otherwise.
+        spec_path = os.path.join(spec_dir,
+                                 "%s.spec" % self.distribution.get_name())
+        self.execute(write_file,
+                     (spec_path,
+                      self._make_spec_file()),
+                     "writing '%s'" % spec_path)
+
+        if self.spec_only: # stop if requested
+            return
+
+        # Make a source distribution and copy to SOURCES directory with
+        # optional icon.
+        saved_dist_files = self.distribution.dist_files[:]
+        sdist = self.reinitialize_command('sdist')
+        if self.use_bzip2:
+            sdist.formats = ['bztar']
+        else:
+            sdist.formats = ['gztar']
+        self.run_command('sdist')
+        self.distribution.dist_files = saved_dist_files
+
+        source = sdist.get_archive_files()[0]
+        source_dir = rpm_dir['SOURCES']
+        self.copy_file(source, source_dir)
+
+        if self.icon:
+            if os.path.exists(self.icon):
+                self.copy_file(self.icon, source_dir)
+            else:
+                raise DistutilsFileError(
+                      "icon file '%s' does not exist" % self.icon)
+
+        # build package
+        log.info("building RPMs")
+        rpm_cmd = ['rpm']
+        if os.path.exists('/usr/bin/rpmbuild') or \
+           os.path.exists('/bin/rpmbuild'):
+            rpm_cmd = ['rpmbuild']
+
+        if self.source_only: # what kind of RPMs?
+            rpm_cmd.append('-bs')
+        elif self.binary_only:
+            rpm_cmd.append('-bb')
+        else:
+            rpm_cmd.append('-ba')
+        rpm_cmd.extend(['--define', '__python %s' % self.python])
+        if self.rpm3_mode:
+            rpm_cmd.extend(['--define',
+                             '_topdir %s' % os.path.abspath(self.rpm_base)])
+        if not self.keep_temp:
+            rpm_cmd.append('--clean')
+
+        if self.quiet:
+            rpm_cmd.append('--quiet')
+
+        rpm_cmd.append(spec_path)
+        # Determine the binary rpm names that should be built out of this spec
+        # file
+        # Note that some of these may not be really built (if the file
+        # list is empty)
+        nvr_string = "%{name}-%{version}-%{release}"
+        src_rpm = nvr_string + ".src.rpm"
+        non_src_rpm = "%{arch}/" + nvr_string + ".%{arch}.rpm"
+        q_cmd = r"rpm -q --qf '%s %s\n' --specfile '%s'" % (
+            src_rpm, non_src_rpm, spec_path)
+
+        out = os.popen(q_cmd)
+        try:
+            binary_rpms = []
+            source_rpm = None
+            while True:
+                line = out.readline()
+                if not line:
+                    break
+                l = line.strip().split()
+                assert(len(l) == 2)
+                binary_rpms.append(l[1])
+                # The source rpm is named after the first entry in the spec file
+                if source_rpm is None:
+                    source_rpm = l[0]
+
+            status = out.close()
+            if status:
+                raise DistutilsExecError("Failed to execute: %s" % repr(q_cmd))
+
+        finally:
+            out.close()
+
+        self.spawn(rpm_cmd)
+
+        if not self.dry_run:
+            if self.distribution.has_ext_modules():
+                pyversion = get_python_version()
+            else:
+                pyversion = 'any'
+
+            if not self.binary_only:
+                srpm = os.path.join(rpm_dir['SRPMS'], source_rpm)
+                assert(os.path.exists(srpm))
+                self.move_file(srpm, self.dist_dir)
+                filename = os.path.join(self.dist_dir, source_rpm)
+                self.distribution.dist_files.append(
+                    ('bdist_rpm', pyversion, filename))
+
+            if not self.source_only:
+                for rpm in binary_rpms:
+                    rpm = os.path.join(rpm_dir['RPMS'], rpm)
+                    if os.path.exists(rpm):
+                        self.move_file(rpm, self.dist_dir)
+                        filename = os.path.join(self.dist_dir,
+                                                os.path.basename(rpm))
+                        self.distribution.dist_files.append(
+                            ('bdist_rpm', pyversion, filename))
+
+    def _dist_path(self, path):
+        return os.path.join(self.dist_dir, os.path.basename(path))
+
+    def _make_spec_file(self):
+        """Generate the text of an RPM spec file and return it as a
+        list of strings (one per line).
+        """
+        # definitions and headers
+        spec_file = [
+            '%define name ' + self.distribution.get_name(),
+            '%define version ' + self.distribution.get_version().replace('-','_'),
+            '%define unmangled_version ' + self.distribution.get_version(),
+            '%define release ' + self.release.replace('-','_'),
+            '',
+            'Summary: ' + self.distribution.get_description(),
+            ]
+
+        # Workaround for #14443 which affects some RPM based systems such as
+        # RHEL6 (and probably derivatives)
+        vendor_hook = subprocess.getoutput('rpm --eval %{__os_install_post}')
+        # Generate a potential replacement value for __os_install_post (whilst
+        # normalizing the whitespace to simplify the test for whether the
+        # invocation of brp-python-bytecompile passes in __python):
+        vendor_hook = '\n'.join(['  %s \\' % line.strip()
+                                 for line in vendor_hook.splitlines()])
+        problem = "brp-python-bytecompile \\\n"
+        fixed = "brp-python-bytecompile %{__python} \\\n"
+        fixed_hook = vendor_hook.replace(problem, fixed)
+        if fixed_hook != vendor_hook:
+            spec_file.append('# Workaround for http://bugs.python.org/issue14443')
+            spec_file.append('%define __os_install_post ' + fixed_hook + '\n')
+
+        # put locale summaries into spec file
+        # XXX not supported for now (hard to put a dictionary
+        # in a config file -- arg!)
+        #for locale in self.summaries.keys():
+        #    spec_file.append('Summary(%s): %s' % (locale,
+        #                                          self.summaries[locale]))
+
+        spec_file.extend([
+            'Name: %{name}',
+            'Version: %{version}',
+            'Release: %{release}',])
+
+        # XXX yuck! this filename is available from the "sdist" command,
+        # but only after it has run: and we create the spec file before
+        # running "sdist", in case of --spec-only.
+        if self.use_bzip2:
+            spec_file.append('Source0: %{name}-%{unmangled_version}.tar.bz2')
+        else:
+            spec_file.append('Source0: %{name}-%{unmangled_version}.tar.gz')
+
+        spec_file.extend([
+            'License: ' + self.distribution.get_license(),
+            'Group: ' + self.group,
+            'BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-buildroot',
+            'Prefix: %{_prefix}', ])
+
+        if not self.force_arch:
+            # noarch if no extension modules
+            if not self.distribution.has_ext_modules():
+                spec_file.append('BuildArch: noarch')
+        else:
+            spec_file.append( 'BuildArch: %s' % self.force_arch )
+
+        for field in ('Vendor',
+                      'Packager',
+                      'Provides',
+                      'Requires',
+                      'Conflicts',
+                      'Obsoletes',
+                      ):
+            val = getattr(self, field.lower())
+            if isinstance(val, list):
+                spec_file.append('%s: %s' % (field, ' '.join(val)))
+            elif val is not None:
+                spec_file.append('%s: %s' % (field, val))
+
+
+        if self.distribution.get_url() != 'UNKNOWN':
+            spec_file.append('Url: ' + self.distribution.get_url())
+
+        if self.distribution_name:
+            spec_file.append('Distribution: ' + self.distribution_name)
+
+        if self.build_requires:
+            spec_file.append('BuildRequires: ' +
+                             ' '.join(self.build_requires))
+
+        if self.icon:
+            spec_file.append('Icon: ' + os.path.basename(self.icon))
+
+        if self.no_autoreq:
+            spec_file.append('AutoReq: 0')
+
+        spec_file.extend([
+            '',
+            '%description',
+            self.distribution.get_long_description()
+            ])
+
+        # put locale descriptions into spec file
+        # XXX again, suppressed because config file syntax doesn't
+        # easily support this ;-(
+        #for locale in self.descriptions.keys():
+        #    spec_file.extend([
+        #        '',
+        #        '%description -l ' + locale,
+        #        self.descriptions[locale],
+        #        ])
+
+        # rpm scripts
+        # figure out default build script
+        def_setup_call = "%s %s" % (self.python,os.path.basename(sys.argv[0]))
+        def_build = "%s build" % def_setup_call
+        if self.use_rpm_opt_flags:
+            def_build = 'env CFLAGS="$RPM_OPT_FLAGS" ' + def_build
+
+        # insert contents of files
+
+        # XXX this is kind of misleading: user-supplied options are files
+        # that we open and interpolate into the spec file, but the defaults
+        # are just text that we drop in as-is.  Hmmm.
+
+        install_cmd = ('%s install -O1 --root=$RPM_BUILD_ROOT '
+                       '--record=INSTALLED_FILES') % def_setup_call
+
+        script_options = [
+            ('prep', 'prep_script', "%setup -n %{name}-%{unmangled_version}"),
+            ('build', 'build_script', def_build),
+            ('install', 'install_script', install_cmd),
+            ('clean', 'clean_script', "rm -rf $RPM_BUILD_ROOT"),
+            ('verifyscript', 'verify_script', None),
+            ('pre', 'pre_install', None),
+            ('post', 'post_install', None),
+            ('preun', 'pre_uninstall', None),
+            ('postun', 'post_uninstall', None),
+        ]
+
+        for (rpm_opt, attr, default) in script_options:
+            # Insert contents of file referred to, if no file is referred to
+            # use 'default' as contents of script
+            val = getattr(self, attr)
+            if val or default:
+                spec_file.extend([
+                    '',
+                    '%' + rpm_opt,])
+                if val:
+                    spec_file.extend(open(val, 'r').read().split('\n'))
+                else:
+                    spec_file.append(default)
+
+
+        # files section
+        spec_file.extend([
+            '',
+            '%files -f INSTALLED_FILES',
+            '%defattr(-,root,root)',
+            ])
+
+        if self.doc_files:
+            spec_file.append('%doc ' + ' '.join(self.doc_files))
+
+        if self.changelog:
+            spec_file.extend([
+                '',
+                '%changelog',])
+            spec_file.extend(self.changelog)
+
+        return spec_file
+
+    def _format_changelog(self, changelog):
+        """Format the changelog correctly and convert it to a list of strings
+        """
+        if not changelog:
+            return changelog
+        new_changelog = []
+        for line in changelog.strip().split('\n'):
+            line = line.strip()
+            if line[0] == '*':
+                new_changelog.extend(['', line])
+            elif line[0] == '-':
+                new_changelog.append(line)
+            else:
+                new_changelog.append('  ' + line)
+
+        # strip trailing newline inserted by first changelog entry
+        if not new_changelog[0]:
+            del new_changelog[0]
+
+        return new_changelog
diff --git a/Lib/distutils/command/bdist_wininst.py b/Lib/distutils/command/bdist_wininst.py
new file mode 100644
index 000000000..1db47f9b9
--- /dev/null
+++ b/Lib/distutils/command/bdist_wininst.py
@@ -0,0 +1,367 @@
+"""distutils.command.bdist_wininst
+
+Implements the Distutils 'bdist_wininst' command: create a windows installer
+exe-program."""
+
+import sys, os
+from distutils.core import Command
+from distutils.util import get_platform
+from distutils.dir_util import create_tree, remove_tree
+from distutils.errors import *
+from distutils.sysconfig import get_python_version
+from distutils import log
+
+class bdist_wininst(Command):
+
+    description = "create an executable installer for MS Windows"
+
+    user_options = [('bdist-dir=', None,
+                     "temporary directory for creating the distribution"),
+                    ('plat-name=', 'p',
+                     "platform name to embed in generated filenames "
+                     "(default: %s)" % get_platform()),
+                    ('keep-temp', 'k',
+                     "keep the pseudo-installation tree around after " +
+                     "creating the distribution archive"),
+                    ('target-version=', None,
+                     "require a specific python version" +
+                     " on the target system"),
+                    ('no-target-compile', 'c',
+                     "do not compile .py to .pyc on the target system"),
+                    ('no-target-optimize', 'o',
+                     "do not compile .py to .pyo (optimized) "
+                     "on the target system"),
+                    ('dist-dir=', 'd',
+                     "directory to put final built distributions in"),
+                    ('bitmap=', 'b',
+                     "bitmap to use for the installer instead of python-powered logo"),
+                    ('title=', 't',
+                     "title to display on the installer background instead of default"),
+                    ('skip-build', None,
+                     "skip rebuilding everything (for testing/debugging)"),
+                    ('install-script=', None,
+                     "basename of installation script to be run after "
+                     "installation or before deinstallation"),
+                    ('pre-install-script=', None,
+                     "Fully qualified filename of a script to be run before "
+                     "any files are installed.  This script need not be in the "
+                     "distribution"),
+                    ('user-access-control=', None,
+                     "specify Vista's UAC handling - 'none'/default=no "
+                     "handling, 'auto'=use UAC if target Python installed for "
+                     "all users, 'force'=always use UAC"),
+                   ]
+
+    boolean_options = ['keep-temp', 'no-target-compile', 'no-target-optimize',
+                       'skip-build']
+
+    def initialize_options(self):
+        self.bdist_dir = None
+        self.plat_name = None
+        self.keep_temp = 0
+        self.no_target_compile = 0
+        self.no_target_optimize = 0
+        self.target_version = None
+        self.dist_dir = None
+        self.bitmap = None
+        self.title = None
+        self.skip_build = None
+        self.install_script = None
+        self.pre_install_script = None
+        self.user_access_control = None
+
+
+    def finalize_options(self):
+        self.set_undefined_options('bdist', ('skip_build', 'skip_build'))
+
+        if self.bdist_dir is None:
+            if self.skip_build and self.plat_name:
+                # If build is skipped and plat_name is overridden, bdist will
+                # not see the correct 'plat_name' - so set that up manually.
+                bdist = self.distribution.get_command_obj('bdist')
+                bdist.plat_name = self.plat_name
+                # next the command will be initialized using that name
+            bdist_base = self.get_finalized_command('bdist').bdist_base
+            self.bdist_dir = os.path.join(bdist_base, 'wininst')
+
+        if not self.target_version:
+            self.target_version = ""
+
+        if not self.skip_build and self.distribution.has_ext_modules():
+            short_version = get_python_version()
+            if self.target_version and self.target_version != short_version:
+                raise DistutilsOptionError(
+                      "target version can only be %s, or the '--skip-build'" \
+                      " option must be specified" % (short_version,))
+            self.target_version = short_version
+
+        self.set_undefined_options('bdist',
+                                   ('dist_dir', 'dist_dir'),
+                                   ('plat_name', 'plat_name'),
+                                  )
+
+        if self.install_script:
+            for script in self.distribution.scripts:
+                if self.install_script == os.path.basename(script):
+                    break
+            else:
+                raise DistutilsOptionError(
+                      "install_script '%s' not found in scripts"
+                      % self.install_script)
+
+    def run(self):
+        if (sys.platform != "win32" and
+            (self.distribution.has_ext_modules() or
+             self.distribution.has_c_libraries())):
+            raise DistutilsPlatformError \
+                  ("distribution contains extensions and/or C libraries; "
+                   "must be compiled on a Windows 32 platform")
+
+        if not self.skip_build:
+            self.run_command('build')
+
+        install = self.reinitialize_command('install', reinit_subcommands=1)
+        install.root = self.bdist_dir
+        install.skip_build = self.skip_build
+        install.warn_dir = 0
+        install.plat_name = self.plat_name
+
+        install_lib = self.reinitialize_command('install_lib')
+        # we do not want to include pyc or pyo files
+        install_lib.compile = 0
+        install_lib.optimize = 0
+
+        if self.distribution.has_ext_modules():
+            # If we are building an installer for a Python version other
+            # than the one we are currently running, then we need to ensure
+            # our build_lib reflects the other Python version rather than ours.
+            # Note that for target_version!=sys.version, we must have skipped the
+            # build step, so there is no issue with enforcing the build of this
+            # version.
+            target_version = self.target_version
+            if not target_version:
+                assert self.skip_build, "Should have already checked this"
+                target_version = '%d.%d' % sys.version_info[:2]
+            plat_specifier = ".%s-%s" % (self.plat_name, target_version)
+            build = self.get_finalized_command('build')
+            build.build_lib = os.path.join(build.build_base,
+                                           'lib' + plat_specifier)
+
+        # Use a custom scheme for the zip-file, because we have to decide
+        # at installation time which scheme to use.
+        for key in ('purelib', 'platlib', 'headers', 'scripts', 'data'):
+            value = key.upper()
+            if key == 'headers':
+                value = value + '/Include/$dist_name'
+            setattr(install,
+                    'install_' + key,
+                    value)
+
+        log.info("installing to %s", self.bdist_dir)
+        install.ensure_finalized()
+
+        # avoid warning of 'install_lib' about installing
+        # into a directory not in sys.path
+        sys.path.insert(0, os.path.join(self.bdist_dir, 'PURELIB'))
+
+        install.run()
+
+        del sys.path[0]
+
+        # And make an archive relative to the root of the
+        # pseudo-installation tree.
+        from tempfile import mktemp
+        archive_basename = mktemp()
+        fullname = self.distribution.get_fullname()
+        arcname = self.make_archive(archive_basename, "zip",
+                                    root_dir=self.bdist_dir)
+        # create an exe containing the zip-file
+        self.create_exe(arcname, fullname, self.bitmap)
+        if self.distribution.has_ext_modules():
+            pyversion = get_python_version()
+        else:
+            pyversion = 'any'
+        self.distribution.dist_files.append(('bdist_wininst', pyversion,
+                                             self.get_installer_filename(fullname)))
+        # remove the zip-file again
+        log.debug("removing temporary file '%s'", arcname)
+        os.remove(arcname)
+
+        if not self.keep_temp:
+            remove_tree(self.bdist_dir, dry_run=self.dry_run)
+
+    def get_inidata(self):
+        # Return data describing the installation.
+        lines = []
+        metadata = self.distribution.metadata
+
+        # Write the [metadata] section.
+        lines.append("[metadata]")
+
+        # 'info' will be displayed in the installer's dialog box,
+        # describing the items to be installed.
+        info = (metadata.long_description or '') + '\n'
+
+        # Escape newline characters
+        def escape(s):
+            return s.replace("\n", "\\n")
+
+        for name in ["author", "author_email", "description", "maintainer",
+                     "maintainer_email", "name", "url", "version"]:
+            data = getattr(metadata, name, "")
+            if data:
+                info = info + ("\n    %s: %s" % \
+                               (name.capitalize(), escape(data)))
+                lines.append("%s=%s" % (name, escape(data)))
+
+        # The [setup] section contains entries controlling
+        # the installer runtime.
+        lines.append("\n[Setup]")
+        if self.install_script:
+            lines.append("install_script=%s" % self.install_script)
+        lines.append("info=%s" % escape(info))
+        lines.append("target_compile=%d" % (not self.no_target_compile))
+        lines.append("target_optimize=%d" % (not self.no_target_optimize))
+        if self.target_version:
+            lines.append("target_version=%s" % self.target_version)
+        if self.user_access_control:
+            lines.append("user_access_control=%s" % self.user_access_control)
+
+        title = self.title or self.distribution.get_fullname()
+        lines.append("title=%s" % escape(title))
+        import time
+        import distutils
+        build_info = "Built %s with distutils-%s" % \
+                     (time.ctime(time.time()), distutils.__version__)
+        lines.append("build_info=%s" % build_info)
+        return "\n".join(lines)
+
+    def create_exe(self, arcname, fullname, bitmap=None):
+        import struct
+
+        self.mkpath(self.dist_dir)
+
+        cfgdata = self.get_inidata()
+
+        installer_name = self.get_installer_filename(fullname)
+        self.announce("creating %s" % installer_name)
+
+        if bitmap:
+            bitmapdata = open(bitmap, "rb").read()
+            bitmaplen = len(bitmapdata)
+        else:
+            bitmaplen = 0
+
+        file = open(installer_name, "wb")
+        file.write(self.get_exe_bytes())
+        if bitmap:
+            file.write(bitmapdata)
+
+        # Convert cfgdata from unicode to ascii, mbcs encoded
+        if isinstance(cfgdata, str):
+            cfgdata = cfgdata.encode("mbcs")
+
+        # Append the pre-install script
+        cfgdata = cfgdata + b"\0"
+        if self.pre_install_script:
+            # We need to normalize newlines, so we open in text mode and
+            # convert back to bytes. "latin-1" simply avoids any possible
+            # failures.
+            with open(self.pre_install_script, "r",
+                encoding="latin-1") as script:
+                script_data = script.read().encode("latin-1")
+            cfgdata = cfgdata + script_data + b"\n\0"
+        else:
+            # empty pre-install script
+            cfgdata = cfgdata + b"\0"
+        file.write(cfgdata)
+
+        # The 'magic number' 0x1234567B is used to make sure that the
+        # binary layout of 'cfgdata' is what the wininst.exe binary
+        # expects.  If the layout changes, increment that number, make
+        # the corresponding changes to the wininst.exe sources, and
+        # recompile them.
+        header = struct.pack("<iii",
+                             0x1234567B,       # tag
+                             len(cfgdata),     # length
+                             bitmaplen,        # number of bytes in bitmap
+                             )
+        file.write(header)
+        file.write(open(arcname, "rb").read())
+
+    def get_installer_filename(self, fullname):
+        # Factored out to allow overriding in subclasses
+        if self.target_version:
+            # if we create an installer for a specific python version,
+            # it's better to include this in the name
+            installer_name = os.path.join(self.dist_dir,
+                                          "%s.%s-py%s.exe" %
+                                           (fullname, self.plat_name, self.target_version))
+        else:
+            installer_name = os.path.join(self.dist_dir,
+                                          "%s.%s.exe" % (fullname, self.plat_name))
+        return installer_name
+
+    def get_exe_bytes(self):
+        # If a target-version other than the current version has been
+        # specified, then using the MSVC version from *this* build is no good.
+        # Without actually finding and executing the target version and parsing
+        # its sys.version, we just hard-code our knowledge of old versions.
+        # NOTE: Possible alternative is to allow "--target-version" to
+        # specify a Python executable rather than a simple version string.
+        # We can then execute this program to obtain any info we need, such
+        # as the real sys.version string for the build.
+        cur_version = get_python_version()
+
+        # If the target version is *later* than us, then we assume they
+        # use what we use
+        # string compares seem wrong, but are what sysconfig.py itself uses
+        if self.target_version and self.target_version < cur_version:
+            if self.target_version < "2.4":
+                bv = '6.0'
+            elif self.target_version == "2.4":
+                bv = '7.1'
+            elif self.target_version == "2.5":
+                bv = '8.0'
+            elif self.target_version <= "3.2":
+                bv = '9.0'
+            elif self.target_version <= "3.4":
+                bv = '10.0'
+            else:
+                bv = '14.0'
+        else:
+            # for current version - use authoritative check.
+            try:
+                from msvcrt import CRT_ASSEMBLY_VERSION
+            except ImportError:
+                # cross-building, so assume the latest version
+                bv = '14.0'
+            else:
+                # as far as we know, CRT is binary compatible based on
+                # the first field, so assume 'x.0' until proven otherwise
+                major = CRT_ASSEMBLY_VERSION.partition('.')[0]
+                bv = major + '.0'
+
+
+        # wininst-x.y.exe is in the same directory as this file
+        directory = os.path.dirname(__file__)
+        # we must use a wininst-x.y.exe built with the same C compiler
+        # used for python.  XXX What about mingw, borland, and so on?
+
+        # if plat_name starts with "win" but is not "win32"
+        # we want to strip "win" and leave the rest (e.g. -amd64)
+        # for all other cases, we don't want any suffix
+        if self.plat_name != 'win32' and self.plat_name[:3] == 'win':
+            sfix = self.plat_name[3:]
+        else:
+            sfix = ''
+
+        filename = os.path.join(directory, "wininst-%s%s.exe" % (bv, sfix))
+        try:
+            f = open(filename, "rb")
+        except IOError as e:
+            raise DistutilsFileError(str(e) + ', %s not included in the Debian packages.' % filename)
+        try:
+            return f.read()
+        finally:
+            f.close()
diff --git a/Lib/distutils/command/build.py b/Lib/distutils/command/build.py
new file mode 100644
index 000000000..c6f52e61e
--- /dev/null
+++ b/Lib/distutils/command/build.py
@@ -0,0 +1,157 @@
+"""distutils.command.build
+
+Implements the Distutils 'build' command."""
+
+import sys, os
+from distutils.core import Command
+from distutils.errors import DistutilsOptionError
+from distutils.util import get_platform
+
+
+def show_compilers():
+    from distutils.ccompiler import show_compilers
+    show_compilers()
+
+
+class build(Command):
+
+    description = "build everything needed to install"
+
+    user_options = [
+        ('build-base=', 'b',
+         "base directory for build library"),
+        ('build-purelib=', None,
+         "build directory for platform-neutral distributions"),
+        ('build-platlib=', None,
+         "build directory for platform-specific distributions"),
+        ('build-lib=', None,
+         "build directory for all distribution (defaults to either " +
+         "build-purelib or build-platlib"),
+        ('build-scripts=', None,
+         "build directory for scripts"),
+        ('build-temp=', 't',
+         "temporary build directory"),
+        ('plat-name=', 'p',
+         "platform name to build for, if supported "
+         "(default: %s)" % get_platform()),
+        ('compiler=', 'c',
+         "specify the compiler type"),
+        ('parallel=', 'j',
+         "number of parallel build jobs"),
+        ('debug', 'g',
+         "compile extensions and libraries with debugging information"),
+        ('force', 'f',
+         "forcibly build everything (ignore file timestamps)"),
+        ('executable=', 'e',
+         "specify final destination interpreter path (build.py)"),
+        ]
+
+    boolean_options = ['debug', 'force']
+
+    help_options = [
+        ('help-compiler', None,
+         "list available compilers", show_compilers),
+        ]
+
+    def initialize_options(self):
+        self.build_base = 'build'
+        # these are decided only after 'build_base' has its final value
+        # (unless overridden by the user or client)
+        self.build_purelib = None
+        self.build_platlib = None
+        self.build_lib = None
+        self.build_temp = None
+        self.build_scripts = None
+        self.compiler = None
+        self.plat_name = None
+        self.debug = None
+        self.force = 0
+        self.executable = None
+        self.parallel = None
+
+    def finalize_options(self):
+        if self.plat_name is None:
+            self.plat_name = get_platform()
+        else:
+            # plat-name only supported for windows (other platforms are
+            # supported via ./configure flags, if at all).  Avoid misleading
+            # other platforms.
+            if os.name != 'nt':
+                raise DistutilsOptionError(
+                            "--plat-name only supported on Windows (try "
+                            "using './configure --help' on your platform)")
+
+        plat_specifier = ".%s-%d.%d" % (self.plat_name, *sys.version_info[:2])
+
+        # Make it so Python 2.x and Python 2.x with --with-pydebug don't
+        # share the same build directories. Doing so confuses the build
+        # process for C modules
+        if hasattr(sys, 'gettotalrefcount'):
+            plat_specifier += '-pydebug'
+
+        # 'build_purelib' and 'build_platlib' just default to 'lib' and
+        # 'lib.<plat>' under the base build directory.  We only use one of
+        # them for a given distribution, though --
+        if self.build_purelib is None:
+            self.build_purelib = os.path.join(self.build_base, 'lib')
+        if self.build_platlib is None:
+            self.build_platlib = os.path.join(self.build_base,
+                                              'lib' + plat_specifier)
+
+        # 'build_lib' is the actual directory that we will use for this
+        # particular module distribution -- if user didn't supply it, pick
+        # one of 'build_purelib' or 'build_platlib'.
+        if self.build_lib is None:
+            if self.distribution.ext_modules:
+                self.build_lib = self.build_platlib
+            else:
+                self.build_lib = self.build_purelib
+
+        # 'build_temp' -- temporary directory for compiler turds,
+        # "build/temp.<plat>"
+        if self.build_temp is None:
+            self.build_temp = os.path.join(self.build_base,
+                                           'temp' + plat_specifier)
+        if self.build_scripts is None:
+            self.build_scripts = os.path.join(self.build_base,
+                                              'scripts-%d.%d' % sys.version_info[:2])
+
+        if self.executable is None:
+            self.executable = os.path.normpath(sys.executable)
+
+        if isinstance(self.parallel, str):
+            try:
+                self.parallel = int(self.parallel)
+            except ValueError:
+                raise DistutilsOptionError("parallel should be an integer")
+
+    def run(self):
+        # Run all relevant sub-commands.  This will be some subset of:
+        #  - build_py      - pure Python modules
+        #  - build_clib    - standalone C libraries
+        #  - build_ext     - Python extensions
+        #  - build_scripts - (Python) scripts
+        for cmd_name in self.get_sub_commands():
+            self.run_command(cmd_name)
+
+
+    # -- Predicates for the sub-command list ---------------------------
+
+    def has_pure_modules(self):
+        return self.distribution.has_pure_modules()
+
+    def has_c_libraries(self):
+        return self.distribution.has_c_libraries()
+
+    def has_ext_modules(self):
+        return self.distribution.has_ext_modules()
+
+    def has_scripts(self):
+        return self.distribution.has_scripts()
+
+
+    sub_commands = [('build_py',      has_pure_modules),
+                    ('build_clib',    has_c_libraries),
+                    ('build_ext',     has_ext_modules),
+                    ('build_scripts', has_scripts),
+                   ]
diff --git a/Lib/distutils/command/build_clib.py b/Lib/distutils/command/build_clib.py
new file mode 100644
index 000000000..3e20ef23c
--- /dev/null
+++ b/Lib/distutils/command/build_clib.py
@@ -0,0 +1,209 @@
+"""distutils.command.build_clib
+
+Implements the Distutils 'build_clib' command, to build a C/C++ library
+that is included in the module distribution and needed by an extension
+module."""
+
+
+# XXX this module has *lots* of code ripped-off quite transparently from
+# build_ext.py -- not surprisingly really, as the work required to build
+# a static library from a collection of C source files is not really all
+# that different from what's required to build a shared object file from
+# a collection of C source files.  Nevertheless, I haven't done the
+# necessary refactoring to account for the overlap in code between the
+# two modules, mainly because a number of subtle details changed in the
+# cut 'n paste.  Sigh.
+
+import os
+from distutils.core import Command
+from distutils.errors import *
+from distutils.sysconfig import customize_compiler
+from distutils import log
+
+def show_compilers():
+    from distutils.ccompiler import show_compilers
+    show_compilers()
+
+
+class build_clib(Command):
+
+    description = "build C/C++ libraries used by Python extensions"
+
+    user_options = [
+        ('build-clib=', 'b',
+         "directory to build C/C++ libraries to"),
+        ('build-temp=', 't',
+         "directory to put temporary build by-products"),
+        ('debug', 'g',
+         "compile with debugging information"),
+        ('force', 'f',
+         "forcibly build everything (ignore file timestamps)"),
+        ('compiler=', 'c',
+         "specify the compiler type"),
+        ]
+
+    boolean_options = ['debug', 'force']
+
+    help_options = [
+        ('help-compiler', None,
+         "list available compilers", show_compilers),
+        ]
+
+    def initialize_options(self):
+        self.build_clib = None
+        self.build_temp = None
+
+        # List of libraries to build
+        self.libraries = None
+
+        # Compilation options for all libraries
+        self.include_dirs = None
+        self.define = None
+        self.undef = None
+        self.debug = None
+        self.force = 0
+        self.compiler = None
+
+
+    def finalize_options(self):
+        # This might be confusing: both build-clib and build-temp default
+        # to build-temp as defined by the "build" command.  This is because
+        # I think that C libraries are really just temporary build
+        # by-products, at least from the point of view of building Python
+        # extensions -- but I want to keep my options open.
+        self.set_undefined_options('build',
+                                   ('build_temp', 'build_clib'),
+                                   ('build_temp', 'build_temp'),
+                                   ('compiler', 'compiler'),
+                                   ('debug', 'debug'),
+                                   ('force', 'force'))
+
+        self.libraries = self.distribution.libraries
+        if self.libraries:
+            self.check_library_list(self.libraries)
+
+        if self.include_dirs is None:
+            self.include_dirs = self.distribution.include_dirs or []
+        if isinstance(self.include_dirs, str):
+            self.include_dirs = self.include_dirs.split(os.pathsep)
+
+        # XXX same as for build_ext -- what about 'self.define' and
+        # 'self.undef' ?
+
+
+    def run(self):
+        if not self.libraries:
+            return
+
+        # Yech -- this is cut 'n pasted from build_ext.py!
+        from distutils.ccompiler import new_compiler
+        self.compiler = new_compiler(compiler=self.compiler,
+                                     dry_run=self.dry_run,
+                                     force=self.force)
+        customize_compiler(self.compiler)
+
+        if self.include_dirs is not None:
+            self.compiler.set_include_dirs(self.include_dirs)
+        if self.define is not None:
+            # 'define' option is a list of (name,value) tuples
+            for (name,value) in self.define:
+                self.compiler.define_macro(name, value)
+        if self.undef is not None:
+            for macro in self.undef:
+                self.compiler.undefine_macro(macro)
+
+        self.build_libraries(self.libraries)
+
+
+    def check_library_list(self, libraries):
+        """Ensure that the list of libraries is valid.
+
+        `library` is presumably provided as a command option 'libraries'.
+        This method checks that it is a list of 2-tuples, where the tuples
+        are (library_name, build_info_dict).
+
+        Raise DistutilsSetupError if the structure is invalid anywhere;
+        just returns otherwise.
+        """
+        if not isinstance(libraries, list):
+            raise DistutilsSetupError(
+                  "'libraries' option must be a list of tuples")
+
+        for lib in libraries:
+            if not isinstance(lib, tuple) and len(lib) != 2:
+                raise DistutilsSetupError(
+                      "each element of 'libraries' must a 2-tuple")
+
+            name, build_info = lib
+
+            if not isinstance(name, str):
+                raise DistutilsSetupError(
+                      "first element of each tuple in 'libraries' "
+                      "must be a string (the library name)")
+
+            if '/' in name or (os.sep != '/' and os.sep in name):
+                raise DistutilsSetupError("bad library name '%s': "
+                       "may not contain directory separators" % lib[0])
+
+            if not isinstance(build_info, dict):
+                raise DistutilsSetupError(
+                      "second element of each tuple in 'libraries' "
+                      "must be a dictionary (build info)")
+
+
+    def get_library_names(self):
+        # Assume the library list is valid -- 'check_library_list()' is
+        # called from 'finalize_options()', so it should be!
+        if not self.libraries:
+            return None
+
+        lib_names = []
+        for (lib_name, build_info) in self.libraries:
+            lib_names.append(lib_name)
+        return lib_names
+
+
+    def get_source_files(self):
+        self.check_library_list(self.libraries)
+        filenames = []
+        for (lib_name, build_info) in self.libraries:
+            sources = build_info.get('sources')
+            if sources is None or not isinstance(sources, (list, tuple)):
+                raise DistutilsSetupError(
+                       "in 'libraries' option (library '%s'), "
+                       "'sources' must be present and must be "
+                       "a list of source filenames" % lib_name)
+
+            filenames.extend(sources)
+        return filenames
+
+
+    def build_libraries(self, libraries):
+        for (lib_name, build_info) in libraries:
+            sources = build_info.get('sources')
+            if sources is None or not isinstance(sources, (list, tuple)):
+                raise DistutilsSetupError(
+                       "in 'libraries' option (library '%s'), "
+                       "'sources' must be present and must be "
+                       "a list of source filenames" % lib_name)
+            sources = list(sources)
+
+            log.info("building '%s' library", lib_name)
+
+            # First, compile the source code to object files in the library
+            # directory.  (This should probably change to putting object
+            # files in a temporary build directory.)
+            macros = build_info.get('macros')
+            include_dirs = build_info.get('include_dirs')
+            objects = self.compiler.compile(sources,
+                                            output_dir=self.build_temp,
+                                            macros=macros,
+                                            include_dirs=include_dirs,
+                                            debug=self.debug)
+
+            # Now "link" the object files together into a static library.
+            # (On Unix at least, this isn't really linking -- it just
+            # builds an archive.  Whatever.)
+            self.compiler.create_static_lib(objects, lib_name,
+                                            output_dir=self.build_clib,
+                                            debug=self.debug)
diff --git a/Lib/distutils/command/build_ext.py b/Lib/distutils/command/build_ext.py
new file mode 100644
index 000000000..acf2fc548
--- /dev/null
+++ b/Lib/distutils/command/build_ext.py
@@ -0,0 +1,755 @@
+"""distutils.command.build_ext
+
+Implements the Distutils 'build_ext' command, for building extension
+modules (currently limited to C extensions, should accommodate C++
+extensions ASAP)."""
+
+import contextlib
+import os
+import re
+import sys
+from distutils.core import Command
+from distutils.errors import *
+from distutils.sysconfig import customize_compiler, get_python_version
+from distutils.sysconfig import get_config_h_filename
+from distutils.dep_util import newer_group
+from distutils.extension import Extension
+from distutils.util import get_platform
+from distutils import log
+
+from site import USER_BASE
+
+# An extension name is just a dot-separated list of Python NAMEs (ie.
+# the same as a fully-qualified module name).
+extension_name_re = re.compile \
+    (r'^[a-zA-Z_][a-zA-Z_0-9]*(\.[a-zA-Z_][a-zA-Z_0-9]*)*$')
+
+
+def show_compilers ():
+    from distutils.ccompiler import show_compilers
+    show_compilers()
+
+
+class build_ext(Command):
+
+    description = "build C/C++ extensions (compile/link to build directory)"
+
+    # XXX thoughts on how to deal with complex command-line options like
+    # these, i.e. how to make it so fancy_getopt can suck them off the
+    # command line and make it look like setup.py defined the appropriate
+    # lists of tuples of what-have-you.
+    #   - each command needs a callback to process its command-line options
+    #   - Command.__init__() needs access to its share of the whole
+    #     command line (must ultimately come from
+    #     Distribution.parse_command_line())
+    #   - it then calls the current command class' option-parsing
+    #     callback to deal with weird options like -D, which have to
+    #     parse the option text and churn out some custom data
+    #     structure
+    #   - that data structure (in this case, a list of 2-tuples)
+    #     will then be present in the command object by the time
+    #     we get to finalize_options() (i.e. the constructor
+    #     takes care of both command-line and client options
+    #     in between initialize_options() and finalize_options())
+
+    sep_by = " (separated by '%s')" % os.pathsep
+    user_options = [
+        ('build-lib=', 'b',
+         "directory for compiled extension modules"),
+        ('build-temp=', 't',
+         "directory for temporary files (build by-products)"),
+        ('plat-name=', 'p',
+         "platform name to cross-compile for, if supported "
+         "(default: %s)" % get_platform()),
+        ('inplace', 'i',
+         "ignore build-lib and put compiled extensions into the source " +
+         "directory alongside your pure Python modules"),
+        ('include-dirs=', 'I',
+         "list of directories to search for header files" + sep_by),
+        ('define=', 'D',
+         "C preprocessor macros to define"),
+        ('undef=', 'U',
+         "C preprocessor macros to undefine"),
+        ('libraries=', 'l',
+         "external C libraries to link with"),
+        ('library-dirs=', 'L',
+         "directories to search for external C libraries" + sep_by),
+        ('rpath=', 'R',
+         "directories to search for shared C libraries at runtime"),
+        ('link-objects=', 'O',
+         "extra explicit link objects to include in the link"),
+        ('debug', 'g',
+         "compile/link with debugging information"),
+        ('force', 'f',
+         "forcibly build everything (ignore file timestamps)"),
+        ('compiler=', 'c',
+         "specify the compiler type"),
+        ('parallel=', 'j',
+         "number of parallel build jobs"),
+        ('swig-cpp', None,
+         "make SWIG create C++ files (default is C)"),
+        ('swig-opts=', None,
+         "list of SWIG command line options"),
+        ('swig=', None,
+         "path to the SWIG executable"),
+        ('user', None,
+         "add user include, library and rpath")
+        ]
+
+    boolean_options = ['inplace', 'debug', 'force', 'swig-cpp', 'user']
+
+    help_options = [
+        ('help-compiler', None,
+         "list available compilers", show_compilers),
+        ]
+
+    def initialize_options(self):
+        self.extensions = None
+        self.build_lib = None
+        self.plat_name = None
+        self.build_temp = None
+        self.inplace = 0
+        self.package = None
+
+        self.include_dirs = None
+        self.define = None
+        self.undef = None
+        self.libraries = None
+        self.library_dirs = None
+        self.rpath = None
+        self.link_objects = None
+        self.debug = None
+        self.force = None
+        self.compiler = None
+        self.swig = None
+        self.swig_cpp = None
+        self.swig_opts = None
+        self.user = None
+        self.parallel = None
+
+    def finalize_options(self):
+        from distutils import sysconfig
+
+        self.set_undefined_options('build',
+                                   ('build_lib', 'build_lib'),
+                                   ('build_temp', 'build_temp'),
+                                   ('compiler', 'compiler'),
+                                   ('debug', 'debug'),
+                                   ('force', 'force'),
+                                   ('parallel', 'parallel'),
+                                   ('plat_name', 'plat_name'),
+                                   )
+
+        if self.package is None:
+            self.package = self.distribution.ext_package
+
+        self.extensions = self.distribution.ext_modules
+
+        # Make sure Python's include directories (for Python.h, pyconfig.h,
+        # etc.) are in the include search path.
+        py_include = sysconfig.get_python_inc()
+        plat_py_include = sysconfig.get_python_inc(plat_specific=1)
+        if self.include_dirs is None:
+            self.include_dirs = self.distribution.include_dirs or []
+        if isinstance(self.include_dirs, str):
+            self.include_dirs = self.include_dirs.split(os.pathsep)
+
+        # If in a virtualenv, add its include directory
+        # Issue 16116
+        if sys.exec_prefix != sys.base_exec_prefix:
+            self.include_dirs.append(os.path.join(sys.exec_prefix, 'include'))
+
+        # Put the Python "system" include dir at the end, so that
+        # any local include dirs take precedence.
+        self.include_dirs.append(py_include)
+        if plat_py_include != py_include:
+            self.include_dirs.append(plat_py_include)
+
+        self.ensure_string_list('libraries')
+        self.ensure_string_list('link_objects')
+
+        # Life is easier if we're not forever checking for None, so
+        # simplify these options to empty lists if unset
+        if self.libraries is None:
+            self.libraries = []
+        if self.library_dirs is None:
+            self.library_dirs = []
+        elif isinstance(self.library_dirs, str):
+            self.library_dirs = self.library_dirs.split(os.pathsep)
+
+        if self.rpath is None:
+            self.rpath = []
+        elif isinstance(self.rpath, str):
+            self.rpath = self.rpath.split(os.pathsep)
+
+        # for extensions under windows use different directories
+        # for Release and Debug builds.
+        # also Python's library directory must be appended to library_dirs
+        if os.name == 'nt':
+            # the 'libs' directory is for binary installs - we assume that
+            # must be the *native* platform.  But we don't really support
+            # cross-compiling via a binary install anyway, so we let it go.
+            self.library_dirs.append(os.path.join(sys.exec_prefix, 'libs'))
+            if sys.base_exec_prefix != sys.prefix:  # Issue 16116
+                self.library_dirs.append(os.path.join(sys.base_exec_prefix, 'libs'))
+            if self.debug:
+                self.build_temp = os.path.join(self.build_temp, "Debug")
+            else:
+                self.build_temp = os.path.join(self.build_temp, "Release")
+
+            # Append the source distribution include and library directories,
+            # this allows distutils on windows to work in the source tree
+            self.include_dirs.append(os.path.dirname(get_config_h_filename()))
+            _sys_home = getattr(sys, '_home', None)
+            if _sys_home:
+                self.library_dirs.append(_sys_home)
+
+            # Use the .lib files for the correct architecture
+            if self.plat_name == 'win32':
+                suffix = 'win32'
+            else:
+                # win-amd64 or win-ia64
+                suffix = self.plat_name[4:]
+            new_lib = os.path.join(sys.exec_prefix, 'PCbuild')
+            if suffix:
+                new_lib = os.path.join(new_lib, suffix)
+            self.library_dirs.append(new_lib)
+
+        # for extensions under Cygwin and AtheOS Python's library directory must be
+        # appended to library_dirs
+        if sys.platform[:6] == 'cygwin' or sys.platform[:6] == 'atheos':
+            if sys.executable.startswith(os.path.join(sys.exec_prefix, "bin")):
+                # building third party extensions
+                self.library_dirs.append(os.path.join(sys.prefix, "lib",
+                                                      "python" + get_python_version(),
+                                                      "config"))
+            else:
+                # building python standard extensions
+                self.library_dirs.append('.')
+
+        # For building extensions with a shared Python library,
+        # Python's library directory must be appended to library_dirs
+        # See Issues: #1600860, #4366
+        if False and (sysconfig.get_config_var('Py_ENABLE_SHARED')):
+            if not sysconfig.python_build:
+                # building third party extensions
+                self.library_dirs.append(sysconfig.get_config_var('LIBDIR'))
+            else:
+                # building python standard extensions
+                self.library_dirs.append('.')
+
+        # The argument parsing will result in self.define being a string, but
+        # it has to be a list of 2-tuples.  All the preprocessor symbols
+        # specified by the 'define' option will be set to '1'.  Multiple
+        # symbols can be separated with commas.
+
+        if self.define:
+            defines = self.define.split(',')
+            self.define = [(symbol, '1') for symbol in defines]
+
+        # The option for macros to undefine is also a string from the
+        # option parsing, but has to be a list.  Multiple symbols can also
+        # be separated with commas here.
+        if self.undef:
+            self.undef = self.undef.split(',')
+
+        if self.swig_opts is None:
+            self.swig_opts = []
+        else:
+            self.swig_opts = self.swig_opts.split(' ')
+
+        # Finally add the user include and library directories if requested
+        if self.user:
+            user_include = os.path.join(USER_BASE, "include")
+            user_lib = os.path.join(USER_BASE, "lib")
+            if os.path.isdir(user_include):
+                self.include_dirs.append(user_include)
+            if os.path.isdir(user_lib):
+                self.library_dirs.append(user_lib)
+                self.rpath.append(user_lib)
+
+        if isinstance(self.parallel, str):
+            try:
+                self.parallel = int(self.parallel)
+            except ValueError:
+                raise DistutilsOptionError("parallel should be an integer")
+
+    def run(self):
+        from distutils.ccompiler import new_compiler
+
+        # 'self.extensions', as supplied by setup.py, is a list of
+        # Extension instances.  See the documentation for Extension (in
+        # distutils.extension) for details.
+        #
+        # For backwards compatibility with Distutils 0.8.2 and earlier, we
+        # also allow the 'extensions' list to be a list of tuples:
+        #    (ext_name, build_info)
+        # where build_info is a dictionary containing everything that
+        # Extension instances do except the name, with a few things being
+        # differently named.  We convert these 2-tuples to Extension
+        # instances as needed.
+
+        if not self.extensions:
+            return
+
+        # If we were asked to build any C/C++ libraries, make sure that the
+        # directory where we put them is in the library search path for
+        # linking extensions.
+        if self.distribution.has_c_libraries():
+            build_clib = self.get_finalized_command('build_clib')
+            self.libraries.extend(build_clib.get_library_names() or [])
+            self.library_dirs.append(build_clib.build_clib)
+
+        # Setup the CCompiler object that we'll use to do all the
+        # compiling and linking
+        self.compiler = new_compiler(compiler=self.compiler,
+                                     verbose=self.verbose,
+                                     dry_run=self.dry_run,
+                                     force=self.force)
+        customize_compiler(self.compiler)
+        # If we are cross-compiling, init the compiler now (if we are not
+        # cross-compiling, init would not hurt, but people may rely on
+        # late initialization of compiler even if they shouldn't...)
+        if os.name == 'nt' and self.plat_name != get_platform():
+            self.compiler.initialize(self.plat_name)
+
+        # And make sure that any compile/link-related options (which might
+        # come from the command-line or from the setup script) are set in
+        # that CCompiler object -- that way, they automatically apply to
+        # all compiling and linking done here.
+        if self.include_dirs is not None:
+            self.compiler.set_include_dirs(self.include_dirs)
+        if self.define is not None:
+            # 'define' option is a list of (name,value) tuples
+            for (name, value) in self.define:
+                self.compiler.define_macro(name, value)
+        if self.undef is not None:
+            for macro in self.undef:
+                self.compiler.undefine_macro(macro)
+        if self.libraries is not None:
+            self.compiler.set_libraries(self.libraries)
+        if self.library_dirs is not None:
+            self.compiler.set_library_dirs(self.library_dirs)
+        if self.rpath is not None:
+            self.compiler.set_runtime_library_dirs(self.rpath)
+        if self.link_objects is not None:
+            self.compiler.set_link_objects(self.link_objects)
+
+        # Now actually compile and link everything.
+        self.build_extensions()
+
+    def check_extensions_list(self, extensions):
+        """Ensure that the list of extensions (presumably provided as a
+        command option 'extensions') is valid, i.e. it is a list of
+        Extension objects.  We also support the old-style list of 2-tuples,
+        where the tuples are (ext_name, build_info), which are converted to
+        Extension instances here.
+
+        Raise DistutilsSetupError if the structure is invalid anywhere;
+        just returns otherwise.
+        """
+        if not isinstance(extensions, list):
+            raise DistutilsSetupError(
+                  "'ext_modules' option must be a list of Extension instances")
+
+        for i, ext in enumerate(extensions):
+            if isinstance(ext, Extension):
+                continue                # OK! (assume type-checking done
+                                        # by Extension constructor)
+
+            if not isinstance(ext, tuple) or len(ext) != 2:
+                raise DistutilsSetupError(
+                       "each element of 'ext_modules' option must be an "
+                       "Extension instance or 2-tuple")
+
+            ext_name, build_info = ext
+
+            log.warn("old-style (ext_name, build_info) tuple found in "
+                     "ext_modules for extension '%s' "
+                     "-- please convert to Extension instance", ext_name)
+
+            if not (isinstance(ext_name, str) and
+                    extension_name_re.match(ext_name)):
+                raise DistutilsSetupError(
+                       "first element of each tuple in 'ext_modules' "
+                       "must be the extension name (a string)")
+
+            if not isinstance(build_info, dict):
+                raise DistutilsSetupError(
+                       "second element of each tuple in 'ext_modules' "
+                       "must be a dictionary (build info)")
+
+            # OK, the (ext_name, build_info) dict is type-safe: convert it
+            # to an Extension instance.
+            ext = Extension(ext_name, build_info['sources'])
+
+            # Easy stuff: one-to-one mapping from dict elements to
+            # instance attributes.
+            for key in ('include_dirs', 'library_dirs', 'libraries',
+                        'extra_objects', 'extra_compile_args',
+                        'extra_link_args'):
+                val = build_info.get(key)
+                if val is not None:
+                    setattr(ext, key, val)
+
+            # Medium-easy stuff: same syntax/semantics, different names.
+            ext.runtime_library_dirs = build_info.get('rpath')
+            if 'def_file' in build_info:
+                log.warn("'def_file' element of build info dict "
+                         "no longer supported")
+
+            # Non-trivial stuff: 'macros' split into 'define_macros'
+            # and 'undef_macros'.
+            macros = build_info.get('macros')
+            if macros:
+                ext.define_macros = []
+                ext.undef_macros = []
+                for macro in macros:
+                    if not (isinstance(macro, tuple) and len(macro) in (1, 2)):
+                        raise DistutilsSetupError(
+                              "'macros' element of build info dict "
+                              "must be 1- or 2-tuple")
+                    if len(macro) == 1:
+                        ext.undef_macros.append(macro[0])
+                    elif len(macro) == 2:
+                        ext.define_macros.append(macro)
+
+            extensions[i] = ext
+
+    def get_source_files(self):
+        self.check_extensions_list(self.extensions)
+        filenames = []
+
+        # Wouldn't it be neat if we knew the names of header files too...
+        for ext in self.extensions:
+            filenames.extend(ext.sources)
+        return filenames
+
+    def get_outputs(self):
+        # Sanity check the 'extensions' list -- can't assume this is being
+        # done in the same run as a 'build_extensions()' call (in fact, we
+        # can probably assume that it *isn't*!).
+        self.check_extensions_list(self.extensions)
+
+        # And build the list of output (built) filenames.  Note that this
+        # ignores the 'inplace' flag, and assumes everything goes in the
+        # "build" tree.
+        outputs = []
+        for ext in self.extensions:
+            outputs.append(self.get_ext_fullpath(ext.name))
+        return outputs
+
+    def build_extensions(self):
+        # First, sanity-check the 'extensions' list
+        self.check_extensions_list(self.extensions)
+        if self.parallel:
+            self._build_extensions_parallel()
+        else:
+            self._build_extensions_serial()
+
+    def _build_extensions_parallel(self):
+        workers = self.parallel
+        if self.parallel is True:
+            workers = os.cpu_count()  # may return None
+        try:
+            from concurrent.futures import ThreadPoolExecutor
+        except ImportError:
+            workers = None
+
+        if workers is None:
+            self._build_extensions_serial()
+            return
+
+        with ThreadPoolExecutor(max_workers=workers) as executor:
+            futures = [executor.submit(self.build_extension, ext)
+                       for ext in self.extensions]
+            for ext, fut in zip(self.extensions, futures):
+                with self._filter_build_errors(ext):
+                    fut.result()
+
+    def _build_extensions_serial(self):
+        for ext in self.extensions:
+            with self._filter_build_errors(ext):
+                self.build_extension(ext)
+
+    @contextlib.contextmanager
+    def _filter_build_errors(self, ext):
+        try:
+            yield
+        except (CCompilerError, DistutilsError, CompileError) as e:
+            if not ext.optional:
+                raise
+            self.warn('building extension "%s" failed: %s' %
+                      (ext.name, e))
+
+    def build_extension(self, ext):
+        sources = ext.sources
+        if sources is None or not isinstance(sources, (list, tuple)):
+            raise DistutilsSetupError(
+                  "in 'ext_modules' option (extension '%s'), "
+                  "'sources' must be present and must be "
+                  "a list of source filenames" % ext.name)
+        sources = list(sources)
+
+        ext_path = self.get_ext_fullpath(ext.name)
+        depends = sources + ext.depends
+        if not (self.force or newer_group(depends, ext_path, 'newer')):
+            log.debug("skipping '%s' extension (up-to-date)", ext.name)
+            return
+        else:
+            log.info("building '%s' extension", ext.name)
+
+        # First, scan the sources for SWIG definition files (.i), run
+        # SWIG on 'em to create .c files, and modify the sources list
+        # accordingly.
+        sources = self.swig_sources(sources, ext)
+
+        # Next, compile the source code to object files.
+
+        # XXX not honouring 'define_macros' or 'undef_macros' -- the
+        # CCompiler API needs to change to accommodate this, and I
+        # want to do one thing at a time!
+
+        # Two possible sources for extra compiler arguments:
+        #   - 'extra_compile_args' in Extension object
+        #   - CFLAGS environment variable (not particularly
+        #     elegant, but people seem to expect it and I
+        #     guess it's useful)
+        # The environment variable should take precedence, and
+        # any sensible compiler will give precedence to later
+        # command line args.  Hence we combine them in order:
+        extra_args = ext.extra_compile_args or []
+
+        macros = ext.define_macros[:]
+        for undef in ext.undef_macros:
+            macros.append((undef,))
+
+        objects = self.compiler.compile(sources,
+                                         output_dir=self.build_temp,
+                                         macros=macros,
+                                         include_dirs=ext.include_dirs,
+                                         debug=self.debug,
+                                         extra_postargs=extra_args,
+                                         depends=ext.depends)
+
+        # XXX outdated variable, kept here in case third-part code
+        # needs it.
+        self._built_objects = objects[:]
+
+        # Now link the object files together into a "shared object" --
+        # of course, first we have to figure out all the other things
+        # that go into the mix.
+        if ext.extra_objects:
+            objects.extend(ext.extra_objects)
+        extra_args = ext.extra_link_args or []
+
+        # Detect target language, if not provided
+        language = ext.language or self.compiler.detect_language(sources)
+
+        self.compiler.link_shared_object(
+            objects, ext_path,
+            libraries=self.get_libraries(ext),
+            library_dirs=ext.library_dirs,
+            runtime_library_dirs=ext.runtime_library_dirs,
+            extra_postargs=extra_args,
+            export_symbols=self.get_export_symbols(ext),
+            debug=self.debug,
+            build_temp=self.build_temp,
+            target_lang=language)
+
+    def swig_sources(self, sources, extension):
+        """Walk the list of source files in 'sources', looking for SWIG
+        interface (.i) files.  Run SWIG on all that are found, and
+        return a modified 'sources' list with SWIG source files replaced
+        by the generated C (or C++) files.
+        """
+        new_sources = []
+        swig_sources = []
+        swig_targets = {}
+
+        # XXX this drops generated C/C++ files into the source tree, which
+        # is fine for developers who want to distribute the generated
+        # source -- but there should be an option to put SWIG output in
+        # the temp dir.
+
+        if self.swig_cpp:
+            log.warn("--swig-cpp is deprecated - use --swig-opts=-c++")
+
+        if self.swig_cpp or ('-c++' in self.swig_opts) or \
+           ('-c++' in extension.swig_opts):
+            target_ext = '.cpp'
+        else:
+            target_ext = '.c'
+
+        for source in sources:
+            (base, ext) = os.path.splitext(source)
+            if ext == ".i":             # SWIG interface file
+                new_sources.append(base + '_wrap' + target_ext)
+                swig_sources.append(source)
+                swig_targets[source] = new_sources[-1]
+            else:
+                new_sources.append(source)
+
+        if not swig_sources:
+            return new_sources
+
+        swig = self.swig or self.find_swig()
+        swig_cmd = [swig, "-python"]
+        swig_cmd.extend(self.swig_opts)
+        if self.swig_cpp:
+            swig_cmd.append("-c++")
+
+        # Do not override commandline arguments
+        if not self.swig_opts:
+            for o in extension.swig_opts:
+                swig_cmd.append(o)
+
+        for source in swig_sources:
+            target = swig_targets[source]
+            log.info("swigging %s to %s", source, target)
+            self.spawn(swig_cmd + ["-o", target, source])
+
+        return new_sources
+
+    def find_swig(self):
+        """Return the name of the SWIG executable.  On Unix, this is
+        just "swig" -- it should be in the PATH.  Tries a bit harder on
+        Windows.
+        """
+        if os.name == "posix":
+            return "swig"
+        elif os.name == "nt":
+            # Look for SWIG in its standard installation directory on
+            # Windows (or so I presume!).  If we find it there, great;
+            # if not, act like Unix and assume it's in the PATH.
+            for vers in ("1.3", "1.2", "1.1"):
+                fn = os.path.join("c:\\swig%s" % vers, "swig.exe")
+                if os.path.isfile(fn):
+                    return fn
+            else:
+                return "swig.exe"
+        else:
+            raise DistutilsPlatformError(
+                  "I don't know how to find (much less run) SWIG "
+                  "on platform '%s'" % os.name)
+
+    # -- Name generators -----------------------------------------------
+    # (extension names, filenames, whatever)
+    def get_ext_fullpath(self, ext_name):
+        """Returns the path of the filename for a given extension.
+
+        The file is located in `build_lib` or directly in the package
+        (inplace option).
+        """
+        fullname = self.get_ext_fullname(ext_name)
+        modpath = fullname.split('.')
+        filename = self.get_ext_filename(modpath[-1])
+
+        if not self.inplace:
+            # no further work needed
+            # returning :
+            #   build_dir/package/path/filename
+            filename = os.path.join(*modpath[:-1]+[filename])
+            return os.path.join(self.build_lib, filename)
+
+        # the inplace option requires to find the package directory
+        # using the build_py command for that
+        package = '.'.join(modpath[0:-1])
+        build_py = self.get_finalized_command('build_py')
+        package_dir = os.path.abspath(build_py.get_package_dir(package))
+
+        # returning
+        #   package_dir/filename
+        return os.path.join(package_dir, filename)
+
+    def get_ext_fullname(self, ext_name):
+        """Returns the fullname of a given extension name.
+
+        Adds the `package.` prefix"""
+        if self.package is None:
+            return ext_name
+        else:
+            return self.package + '.' + ext_name
+
+    def get_ext_filename(self, ext_name):
+        r"""Convert the name of an extension (eg. "foo.bar") into the name
+        of the file from which it will be loaded (eg. "foo/bar.so", or
+        "foo\bar.pyd").
+        """
+        from distutils.sysconfig import get_config_var
+        ext_path = ext_name.split('.')
+        ext_suffix = get_config_var('EXT_SUFFIX')
+        return os.path.join(*ext_path) + ext_suffix
+
+    def get_export_symbols(self, ext):
+        """Return the list of symbols that a shared extension has to
+        export.  This either uses 'ext.export_symbols' or, if it's not
+        provided, "PyInit_" + module_name.  Only relevant on Windows, where
+        the .pyd file (DLL) must export the module "PyInit_" function.
+        """
+        initfunc_name = "PyInit_" + ext.name.split('.')[-1]
+        if initfunc_name not in ext.export_symbols:
+            ext.export_symbols.append(initfunc_name)
+        return ext.export_symbols
+
+    def get_libraries(self, ext):
+        """Return the list of libraries to link against when building a
+        shared extension.  On most platforms, this is just 'ext.libraries';
+        on Windows, we add the Python library (eg. python20.dll).
+        """
+        # The python library is always needed on Windows.  For MSVC, this
+        # is redundant, since the library is mentioned in a pragma in
+        # pyconfig.h that MSVC groks.  The other Windows compilers all seem
+        # to need it mentioned explicitly, though, so that's what we do.
+        # Append '_d' to the python import library on debug builds.
+        if sys.platform == "win32":
+            from distutils._msvccompiler import MSVCCompiler
+            if not isinstance(self.compiler, MSVCCompiler):
+                template = "python%d%d"
+                if self.debug:
+                    template = template + '_d'
+                pythonlib = (template %
+                       (sys.hexversion >> 24, (sys.hexversion >> 16) & 0xff))
+                # don't extend ext.libraries, it may be shared with other
+                # extensions, it is a reference to the original list
+                return ext.libraries + [pythonlib]
+            else:
+                return ext.libraries
+        elif sys.platform[:6] == "cygwin":
+            template = "python%d.%d"
+            pythonlib = (template %
+                   (sys.hexversion >> 24, (sys.hexversion >> 16) & 0xff))
+            # don't extend ext.libraries, it may be shared with other
+            # extensions, it is a reference to the original list
+            return ext.libraries + [pythonlib]
+        elif sys.platform[:6] == "atheos":
+            from distutils import sysconfig
+
+            template = "python%d.%d"
+            pythonlib = (template %
+                   (sys.hexversion >> 24, (sys.hexversion >> 16) & 0xff))
+            # Get SHLIBS from Makefile
+            extra = []
+            for lib in sysconfig.get_config_var('SHLIBS').split():
+                if lib.startswith('-l'):
+                    extra.append(lib[2:])
+                else:
+                    extra.append(lib)
+            # don't extend ext.libraries, it may be shared with other
+            # extensions, it is a reference to the original list
+            return ext.libraries + [pythonlib, "m"] + extra
+        elif sys.platform == 'darwin':
+            # Don't use the default code below
+            return ext.libraries
+        elif sys.platform[:3] == 'aix':
+            # Don't use the default code below
+            return ext.libraries
+        else:
+            from distutils import sysconfig
+            if False and sysconfig.get_config_var('Py_ENABLE_SHARED'):
+                pythonlib = 'python{}.{}{}'.format(
+                    sys.hexversion >> 24, (sys.hexversion >> 16) & 0xff,
+                    sysconfig.get_config_var('ABIFLAGS'))
+                return ext.libraries + [pythonlib]
+            else:
+                return ext.libraries
diff --git a/Lib/distutils/command/build_py.py b/Lib/distutils/command/build_py.py
new file mode 100644
index 000000000..cf0ca57c3
--- /dev/null
+++ b/Lib/distutils/command/build_py.py
@@ -0,0 +1,416 @@
+"""distutils.command.build_py
+
+Implements the Distutils 'build_py' command."""
+
+import os
+import importlib.util
+import sys
+from glob import glob
+
+from distutils.core import Command
+from distutils.errors import *
+from distutils.util import convert_path, Mixin2to3
+from distutils import log
+
+class build_py (Command):
+
+    description = "\"build\" pure Python modules (copy to build directory)"
+
+    user_options = [
+        ('build-lib=', 'd', "directory to \"build\" (copy) to"),
+        ('compile', 'c', "compile .py to .pyc"),
+        ('no-compile', None, "don't compile .py files [default]"),
+        ('optimize=', 'O',
+         "also compile with optimization: -O1 for \"python -O\", "
+         "-O2 for \"python -OO\", and -O0 to disable [default: -O0]"),
+        ('force', 'f', "forcibly build everything (ignore file timestamps)"),
+        ]
+
+    boolean_options = ['compile', 'force']
+    negative_opt = {'no-compile' : 'compile'}
+
+    def initialize_options(self):
+        self.build_lib = None
+        self.py_modules = None
+        self.package = None
+        self.package_data = None
+        self.package_dir = None
+        self.compile = 0
+        self.optimize = 0
+        self.force = None
+
+    def finalize_options(self):
+        self.set_undefined_options('build',
+                                   ('build_lib', 'build_lib'),
+                                   ('force', 'force'))
+
+        # Get the distribution options that are aliases for build_py
+        # options -- list of packages and list of modules.
+        self.packages = self.distribution.packages
+        self.py_modules = self.distribution.py_modules
+        self.package_data = self.distribution.package_data
+        self.package_dir = {}
+        if self.distribution.package_dir:
+            for name, path in self.distribution.package_dir.items():
+                self.package_dir[name] = convert_path(path)
+        self.data_files = self.get_data_files()
+
+        # Ick, copied straight from install_lib.py (fancy_getopt needs a
+        # type system!  Hell, *everything* needs a type system!!!)
+        if not isinstance(self.optimize, int):
+            try:
+                self.optimize = int(self.optimize)
+                assert 0 <= self.optimize <= 2
+            except (ValueError, AssertionError):
+                raise DistutilsOptionError("optimize must be 0, 1, or 2")
+
+    def run(self):
+        # XXX copy_file by default preserves atime and mtime.  IMHO this is
+        # the right thing to do, but perhaps it should be an option -- in
+        # particular, a site administrator might want installed files to
+        # reflect the time of installation rather than the last
+        # modification time before the installed release.
+
+        # XXX copy_file by default preserves mode, which appears to be the
+        # wrong thing to do: if a file is read-only in the working
+        # directory, we want it to be installed read/write so that the next
+        # installation of the same module distribution can overwrite it
+        # without problems.  (This might be a Unix-specific issue.)  Thus
+        # we turn off 'preserve_mode' when copying to the build directory,
+        # since the build directory is supposed to be exactly what the
+        # installation will look like (ie. we preserve mode when
+        # installing).
+
+        # Two options control which modules will be installed: 'packages'
+        # and 'py_modules'.  The former lets us work with whole packages, not
+        # specifying individual modules at all; the latter is for
+        # specifying modules one-at-a-time.
+
+        if self.py_modules:
+            self.build_modules()
+        if self.packages:
+            self.build_packages()
+            self.build_package_data()
+
+        self.byte_compile(self.get_outputs(include_bytecode=0))
+
+    def get_data_files(self):
+        """Generate list of '(package,src_dir,build_dir,filenames)' tuples"""
+        data = []
+        if not self.packages:
+            return data
+        for package in self.packages:
+            # Locate package source directory
+            src_dir = self.get_package_dir(package)
+
+            # Compute package build directory
+            build_dir = os.path.join(*([self.build_lib] + package.split('.')))
+
+            # Length of path to strip from found files
+            plen = 0
+            if src_dir:
+                plen = len(src_dir)+1
+
+            # Strip directory from globbed filenames
+            filenames = [
+                file[plen:] for file in self.find_data_files(package, src_dir)
+                ]
+            data.append((package, src_dir, build_dir, filenames))
+        return data
+
+    def find_data_files(self, package, src_dir):
+        """Return filenames for package's data files in 'src_dir'"""
+        globs = (self.package_data.get('', [])
+                 + self.package_data.get(package, []))
+        files = []
+        for pattern in globs:
+            # Each pattern has to be converted to a platform-specific path
+            filelist = glob(os.path.join(src_dir, convert_path(pattern)))
+            # Files that match more than one pattern are only added once
+            files.extend([fn for fn in filelist if fn not in files
+                and os.path.isfile(fn)])
+        return files
+
+    def build_package_data(self):
+        """Copy data files into build directory"""
+        lastdir = None
+        for package, src_dir, build_dir, filenames in self.data_files:
+            for filename in filenames:
+                target = os.path.join(build_dir, filename)
+                self.mkpath(os.path.dirname(target))
+                self.copy_file(os.path.join(src_dir, filename), target,
+                               preserve_mode=False)
+
+    def get_package_dir(self, package):
+        """Return the directory, relative to the top of the source
+           distribution, where package 'package' should be found
+           (at least according to the 'package_dir' option, if any)."""
+        path = package.split('.')
+
+        if not self.package_dir:
+            if path:
+                return os.path.join(*path)
+            else:
+                return ''
+        else:
+            tail = []
+            while path:
+                try:
+                    pdir = self.package_dir['.'.join(path)]
+                except KeyError:
+                    tail.insert(0, path[-1])
+                    del path[-1]
+                else:
+                    tail.insert(0, pdir)
+                    return os.path.join(*tail)
+            else:
+                # Oops, got all the way through 'path' without finding a
+                # match in package_dir.  If package_dir defines a directory
+                # for the root (nameless) package, then fallback on it;
+                # otherwise, we might as well have not consulted
+                # package_dir at all, as we just use the directory implied
+                # by 'tail' (which should be the same as the original value
+                # of 'path' at this point).
+                pdir = self.package_dir.get('')
+                if pdir is not None:
+                    tail.insert(0, pdir)
+
+                if tail:
+                    return os.path.join(*tail)
+                else:
+                    return ''
+
+    def check_package(self, package, package_dir):
+        # Empty dir name means current directory, which we can probably
+        # assume exists.  Also, os.path.exists and isdir don't know about
+        # my "empty string means current dir" convention, so we have to
+        # circumvent them.
+        if package_dir != "":
+            if not os.path.exists(package_dir):
+                raise DistutilsFileError(
+                      "package directory '%s' does not exist" % package_dir)
+            if not os.path.isdir(package_dir):
+                raise DistutilsFileError(
+                       "supposed package directory '%s' exists, "
+                       "but is not a directory" % package_dir)
+
+        # Require __init__.py for all but the "root package"
+        if package:
+            init_py = os.path.join(package_dir, "__init__.py")
+            if os.path.isfile(init_py):
+                return init_py
+            else:
+                log.warn(("package init file '%s' not found " +
+                          "(or not a regular file)"), init_py)
+
+        # Either not in a package at all (__init__.py not expected), or
+        # __init__.py doesn't exist -- so don't return the filename.
+        return None
+
+    def check_module(self, module, module_file):
+        if not os.path.isfile(module_file):
+            log.warn("file %s (for module %s) not found", module_file, module)
+            return False
+        else:
+            return True
+
+    def find_package_modules(self, package, package_dir):
+        self.check_package(package, package_dir)
+        module_files = glob(os.path.join(package_dir, "*.py"))
+        modules = []
+        setup_script = os.path.abspath(self.distribution.script_name)
+
+        for f in module_files:
+            abs_f = os.path.abspath(f)
+            if abs_f != setup_script:
+                module = os.path.splitext(os.path.basename(f))[0]
+                modules.append((package, module, f))
+            else:
+                self.debug_print("excluding %s" % setup_script)
+        return modules
+
+    def find_modules(self):
+        """Finds individually-specified Python modules, ie. those listed by
+        module name in 'self.py_modules'.  Returns a list of tuples (package,
+        module_base, filename): 'package' is a tuple of the path through
+        package-space to the module; 'module_base' is the bare (no
+        packages, no dots) module name, and 'filename' is the path to the
+        ".py" file (relative to the distribution root) that implements the
+        module.
+        """
+        # Map package names to tuples of useful info about the package:
+        #    (package_dir, checked)
+        # package_dir - the directory where we'll find source files for
+        #   this package
+        # checked - true if we have checked that the package directory
+        #   is valid (exists, contains __init__.py, ... ?)
+        packages = {}
+
+        # List of (package, module, filename) tuples to return
+        modules = []
+
+        # We treat modules-in-packages almost the same as toplevel modules,
+        # just the "package" for a toplevel is empty (either an empty
+        # string or empty list, depending on context).  Differences:
+        #   - don't check for __init__.py in directory for empty package
+        for module in self.py_modules:
+            path = module.split('.')
+            package = '.'.join(path[0:-1])
+            module_base = path[-1]
+
+            try:
+                (package_dir, checked) = packages[package]
+            except KeyError:
+                package_dir = self.get_package_dir(package)
+                checked = 0
+
+            if not checked:
+                init_py = self.check_package(package, package_dir)
+                packages[package] = (package_dir, 1)
+                if init_py:
+                    modules.append((package, "__init__", init_py))
+
+            # XXX perhaps we should also check for just .pyc files
+            # (so greedy closed-source bastards can distribute Python
+            # modules too)
+            module_file = os.path.join(package_dir, module_base + ".py")
+            if not self.check_module(module, module_file):
+                continue
+
+            modules.append((package, module_base, module_file))
+
+        return modules
+
+    def find_all_modules(self):
+        """Compute the list of all modules that will be built, whether
+        they are specified one-module-at-a-time ('self.py_modules') or
+        by whole packages ('self.packages').  Return a list of tuples
+        (package, module, module_file), just like 'find_modules()' and
+        'find_package_modules()' do."""
+        modules = []
+        if self.py_modules:
+            modules.extend(self.find_modules())
+        if self.packages:
+            for package in self.packages:
+                package_dir = self.get_package_dir(package)
+                m = self.find_package_modules(package, package_dir)
+                modules.extend(m)
+        return modules
+
+    def get_source_files(self):
+        return [module[-1] for module in self.find_all_modules()]
+
+    def get_module_outfile(self, build_dir, package, module):
+        outfile_path = [build_dir] + list(package) + [module + ".py"]
+        return os.path.join(*outfile_path)
+
+    def get_outputs(self, include_bytecode=1):
+        modules = self.find_all_modules()
+        outputs = []
+        for (package, module, module_file) in modules:
+            package = package.split('.')
+            filename = self.get_module_outfile(self.build_lib, package, module)
+            outputs.append(filename)
+            if include_bytecode:
+                if self.compile:
+                    outputs.append(importlib.util.cache_from_source(
+                        filename, optimization=''))
+                if self.optimize > 0:
+                    outputs.append(importlib.util.cache_from_source(
+                        filename, optimization=self.optimize))
+
+        outputs += [
+            os.path.join(build_dir, filename)
+            for package, src_dir, build_dir, filenames in self.data_files
+            for filename in filenames
+            ]
+
+        return outputs
+
+    def build_module(self, module, module_file, package):
+        if isinstance(package, str):
+            package = package.split('.')
+        elif not isinstance(package, (list, tuple)):
+            raise TypeError(
+                  "'package' must be a string (dot-separated), list, or tuple")
+
+        # Now put the module source file into the "build" area -- this is
+        # easy, we just copy it somewhere under self.build_lib (the build
+        # directory for Python source).
+        outfile = self.get_module_outfile(self.build_lib, package, module)
+        dir = os.path.dirname(outfile)
+        self.mkpath(dir)
+        return self.copy_file(module_file, outfile, preserve_mode=0)
+
+    def build_modules(self):
+        modules = self.find_modules()
+        for (package, module, module_file) in modules:
+            # Now "build" the module -- ie. copy the source file to
+            # self.build_lib (the build directory for Python source).
+            # (Actually, it gets copied to the directory for this package
+            # under self.build_lib.)
+            self.build_module(module, module_file, package)
+
+    def build_packages(self):
+        for package in self.packages:
+            # Get list of (package, module, module_file) tuples based on
+            # scanning the package directory.  'package' is only included
+            # in the tuple so that 'find_modules()' and
+            # 'find_package_tuples()' have a consistent interface; it's
+            # ignored here (apart from a sanity check).  Also, 'module' is
+            # the *unqualified* module name (ie. no dots, no package -- we
+            # already know its package!), and 'module_file' is the path to
+            # the .py file, relative to the current directory
+            # (ie. including 'package_dir').
+            package_dir = self.get_package_dir(package)
+            modules = self.find_package_modules(package, package_dir)
+
+            # Now loop over the modules we found, "building" each one (just
+            # copy it to self.build_lib).
+            for (package_, module, module_file) in modules:
+                assert package == package_
+                self.build_module(module, module_file, package)
+
+    def byte_compile(self, files):
+        if sys.dont_write_bytecode:
+            self.warn('byte-compiling is disabled, skipping.')
+            return
+
+        from distutils.util import byte_compile
+        prefix = self.build_lib
+        if prefix[-1] != os.sep:
+            prefix = prefix + os.sep
+
+        # XXX this code is essentially the same as the 'byte_compile()
+        # method of the "install_lib" command, except for the determination
+        # of the 'prefix' string.  Hmmm.
+        if self.compile:
+            byte_compile(files, optimize=0,
+                         force=self.force, prefix=prefix, dry_run=self.dry_run)
+        if self.optimize > 0:
+            byte_compile(files, optimize=self.optimize,
+                         force=self.force, prefix=prefix, dry_run=self.dry_run)
+
+class build_py_2to3(build_py, Mixin2to3):
+    def run(self):
+        self.updated_files = []
+
+        # Base class code
+        if self.py_modules:
+            self.build_modules()
+        if self.packages:
+            self.build_packages()
+            self.build_package_data()
+
+        # 2to3
+        self.run_2to3(self.updated_files)
+
+        # Remaining base class code
+        self.byte_compile(self.get_outputs(include_bytecode=0))
+
+    def build_module(self, module, module_file, package):
+        res = build_py.build_module(self, module, module_file, package)
+        if res[1]:
+            # file was copied
+            self.updated_files.append(res[0])
+        return res
diff --git a/Lib/distutils/command/build_scripts.py b/Lib/distutils/command/build_scripts.py
new file mode 100644
index 000000000..ccc70e646
--- /dev/null
+++ b/Lib/distutils/command/build_scripts.py
@@ -0,0 +1,160 @@
+"""distutils.command.build_scripts
+
+Implements the Distutils 'build_scripts' command."""
+
+import os, re
+from stat import ST_MODE
+from distutils import sysconfig
+from distutils.core import Command
+from distutils.dep_util import newer
+from distutils.util import convert_path, Mixin2to3
+from distutils import log
+import tokenize
+
+# check if Python is called on the first line with this expression
+first_line_re = re.compile(b'^#!.*python[0-9.]*([ \t].*)?$')
+
+class build_scripts(Command):
+
+    description = "\"build\" scripts (copy and fixup #! line)"
+
+    user_options = [
+        ('build-dir=', 'd', "directory to \"build\" (copy) to"),
+        ('force', 'f', "forcibly build everything (ignore file timestamps"),
+        ('executable=', 'e', "specify final destination interpreter path"),
+        ]
+
+    boolean_options = ['force']
+
+
+    def initialize_options(self):
+        self.build_dir = None
+        self.scripts = None
+        self.force = None
+        self.executable = None
+        self.outfiles = None
+
+    def finalize_options(self):
+        self.set_undefined_options('build',
+                                   ('build_scripts', 'build_dir'),
+                                   ('force', 'force'),
+                                   ('executable', 'executable'))
+        self.scripts = self.distribution.scripts
+
+    def get_source_files(self):
+        return self.scripts
+
+    def run(self):
+        if not self.scripts:
+            return
+        self.copy_scripts()
+
+
+    def copy_scripts(self):
+        r"""Copy each script listed in 'self.scripts'; if it's marked as a
+        Python script in the Unix way (first line matches 'first_line_re',
+        ie. starts with "\#!" and contains "python"), then adjust the first
+        line to refer to the current Python interpreter as we copy.
+        """
+        self.mkpath(self.build_dir)
+        outfiles = []
+        updated_files = []
+        for script in self.scripts:
+            adjust = False
+            script = convert_path(script)
+            outfile = os.path.join(self.build_dir, os.path.basename(script))
+            outfiles.append(outfile)
+
+            if not self.force and not newer(script, outfile):
+                log.debug("not copying %s (up-to-date)", script)
+                continue
+
+            # Always open the file, but ignore failures in dry-run mode --
+            # that way, we'll get accurate feedback if we can read the
+            # script.
+            try:
+                f = open(script, "rb")
+            except OSError:
+                if not self.dry_run:
+                    raise
+                f = None
+            else:
+                encoding, lines = tokenize.detect_encoding(f.readline)
+                f.seek(0)
+                first_line = f.readline()
+                if not first_line:
+                    self.warn("%s is an empty file (skipping)" % script)
+                    continue
+
+                match = first_line_re.match(first_line)
+                if match:
+                    adjust = True
+                    post_interp = match.group(1) or b''
+
+            if adjust:
+                log.info("copying and adjusting %s -> %s", script,
+                         self.build_dir)
+                updated_files.append(outfile)
+                if not self.dry_run:
+                    if not sysconfig.python_build:
+                        executable = self.executable
+                    else:
+                        executable = os.path.join(
+                            sysconfig.get_config_var("BINDIR"),
+                           "python%s%s" % (sysconfig.get_config_var("VERSION"),
+                                           sysconfig.get_config_var("EXE")))
+                    executable = os.fsencode(executable)
+                    shebang = b"#!" + executable + post_interp + b"\n"
+                    # Python parser starts to read a script using UTF-8 until
+                    # it gets a #coding:xxx cookie. The shebang has to be the
+                    # first line of a file, the #coding:xxx cookie cannot be
+                    # written before. So the shebang has to be decodable from
+                    # UTF-8.
+                    try:
+                        shebang.decode('utf-8')
+                    except UnicodeDecodeError:
+                        raise ValueError(
+                            "The shebang ({!r}) is not decodable "
+                            "from utf-8".format(shebang))
+                    # If the script is encoded to a custom encoding (use a
+                    # #coding:xxx cookie), the shebang has to be decodable from
+                    # the script encoding too.
+                    try:
+                        shebang.decode(encoding)
+                    except UnicodeDecodeError:
+                        raise ValueError(
+                            "The shebang ({!r}) is not decodable "
+                            "from the script encoding ({})"
+                            .format(shebang, encoding))
+                    with open(outfile, "wb") as outf:
+                        outf.write(shebang)
+                        outf.writelines(f.readlines())
+                if f:
+                    f.close()
+            else:
+                if f:
+                    f.close()
+                updated_files.append(outfile)
+                self.copy_file(script, outfile)
+
+        if os.name == 'posix':
+            for file in outfiles:
+                if self.dry_run:
+                    log.info("changing mode of %s", file)
+                else:
+                    oldmode = os.stat(file)[ST_MODE] & 0o7777
+                    newmode = (oldmode | 0o555) & 0o7777
+                    if newmode != oldmode:
+                        log.info("changing mode of %s from %o to %o",
+                                 file, oldmode, newmode)
+                        os.chmod(file, newmode)
+        # XXX should we modify self.outfiles?
+        return outfiles, updated_files
+
+class build_scripts_2to3(build_scripts, Mixin2to3):
+
+    def copy_scripts(self):
+        outfiles, updated_files = build_scripts.copy_scripts(self)
+        if not self.dry_run:
+            self.run_2to3(updated_files)
+        return outfiles, updated_files
diff --git a/Lib/distutils/command/check.py b/Lib/distutils/command/check.py
new file mode 100644
index 000000000..7ebe707cf
--- /dev/null
+++ b/Lib/distutils/command/check.py
@@ -0,0 +1,145 @@
+"""distutils.command.check
+
+Implements the Distutils 'check' command.
+"""
+from distutils.core import Command
+from distutils.errors import DistutilsSetupError
+
+try:
+    # docutils is installed
+    from docutils.utils import Reporter
+    from docutils.parsers.rst import Parser
+    from docutils import frontend
+    from docutils import nodes
+    from io import StringIO
+
+    class SilentReporter(Reporter):
+
+        def __init__(self, source, report_level, halt_level, stream=None,
+                     debug=0, encoding='ascii', error_handler='replace'):
+            self.messages = []
+            Reporter.__init__(self, source, report_level, halt_level, stream,
+                              debug, encoding, error_handler)
+
+        def system_message(self, level, message, *children, **kwargs):
+            self.messages.append((level, message, children, kwargs))
+            return nodes.system_message(message, level=level,
+                                        type=self.levels[level],
+                                        *children, **kwargs)
+
+    HAS_DOCUTILS = True
+except Exception:
+    # Catch all exceptions because exceptions besides ImportError probably
+    # indicate that docutils is not ported to Py3k.
+    HAS_DOCUTILS = False
+
+class check(Command):
+    """This command checks the meta-data of the package.
+    """
+    description = ("perform some checks on the package")
+    user_options = [('metadata', 'm', 'Verify meta-data'),
+                    ('restructuredtext', 'r',
+                     ('Checks if long string meta-data syntax '
+                      'are reStructuredText-compliant')),
+                    ('strict', 's',
+                     'Will exit with an error if a check fails')]
+
+    boolean_options = ['metadata', 'restructuredtext', 'strict']
+
+    def initialize_options(self):
+        """Sets default values for options."""
+        self.restructuredtext = 0
+        self.metadata = 1
+        self.strict = 0
+        self._warnings = 0
+
+    def finalize_options(self):
+        pass
+
+    def warn(self, msg):
+        """Counts the number of warnings that occurs."""
+        self._warnings += 1
+        return Command.warn(self, msg)
+
+    def run(self):
+        """Runs the command."""
+        # perform the various tests
+        if self.metadata:
+            self.check_metadata()
+        if self.restructuredtext:
+            if HAS_DOCUTILS:
+                self.check_restructuredtext()
+            elif self.strict:
+                raise DistutilsSetupError('The docutils package is needed.')
+
+        # let's raise an error in strict mode, if we have at least
+        # one warning
+        if self.strict and self._warnings > 0:
+            raise DistutilsSetupError('Please correct your package.')
+
+    def check_metadata(self):
+        """Ensures that all required elements of meta-data are supplied.
+
+        name, version, URL, (author and author_email) or
+        (maintainer and maintainer_email)).
+
+        Warns if any are missing.
+        """
+        metadata = self.distribution.metadata
+
+        missing = []
+        for attr in ('name', 'version', 'url'):
+            if not (hasattr(metadata, attr) and getattr(metadata, attr)):
+                missing.append(attr)
+
+        if missing:
+            self.warn("missing required meta-data: %s"  % ', '.join(missing))
+        if metadata.author:
+            if not metadata.author_email:
+                self.warn("missing meta-data: if 'author' supplied, " +
+                          "'author_email' must be supplied too")
+        elif metadata.maintainer:
+            if not metadata.maintainer_email:
+                self.warn("missing meta-data: if 'maintainer' supplied, " +
+                          "'maintainer_email' must be supplied too")
+        else:
+            self.warn("missing meta-data: either (author and author_email) " +
+                      "or (maintainer and maintainer_email) " +
+                      "must be supplied")
+
+    def check_restructuredtext(self):
+        """Checks if the long string fields are reST-compliant."""
+        data = self.distribution.get_long_description()
+        for warning in self._check_rst_data(data):
+            line = warning[-1].get('line')
+            if line is None:
+                warning = warning[1]
+            else:
+                warning = '%s (line %s)' % (warning[1], line)
+            self.warn(warning)
+
+    def _check_rst_data(self, data):
+        """Returns warnings when the provided data doesn't compile."""
+        source_path = StringIO()
+        parser = Parser()
+        settings = frontend.OptionParser(components=(Parser,)).get_default_values()
+        settings.tab_width = 4
+        settings.pep_references = None
+        settings.rfc_references = None
+        reporter = SilentReporter(source_path,
+                          settings.report_level,
+                          settings.halt_level,
+                          stream=settings.warning_stream,
+                          debug=settings.debug,
+                          encoding=settings.error_encoding,
+                          error_handler=settings.error_encoding_error_handler)
+
+        document = nodes.document(settings, reporter, source=source_path)
+        document.note_source(source_path, -1)
+        try:
+            parser.parse(data, document)
+        except AttributeError as e:
+            reporter.messages.append(
+                (-1, 'Could not finish the parsing: %s.' % e, '', {}))
+
+        return reporter.messages
diff --git a/Lib/distutils/command/clean.py b/Lib/distutils/command/clean.py
new file mode 100644
index 000000000..0cb270166
--- /dev/null
+++ b/Lib/distutils/command/clean.py
@@ -0,0 +1,76 @@
+"""distutils.command.clean
+
+Implements the Distutils 'clean' command."""
+
+# contributed by Bastian Kleineidam <calvin@cs.uni-sb.de>, added 2000-03-18
+
+import os
+from distutils.core import Command
+from distutils.dir_util import remove_tree
+from distutils import log
+
+class clean(Command):
+
+    description = "clean up temporary files from 'build' command"
+    user_options = [
+        ('build-base=', 'b',
+         "base build directory (default: 'build.build-base')"),
+        ('build-lib=', None,
+         "build directory for all modules (default: 'build.build-lib')"),
+        ('build-temp=', 't',
+         "temporary build directory (default: 'build.build-temp')"),
+        ('build-scripts=', None,
+         "build directory for scripts (default: 'build.build-scripts')"),
+        ('bdist-base=', None,
+         "temporary directory for built distributions"),
+        ('all', 'a',
+         "remove all build output, not just temporary by-products")
+    ]
+
+    boolean_options = ['all']
+
+    def initialize_options(self):
+        self.build_base = None
+        self.build_lib = None
+        self.build_temp = None
+        self.build_scripts = None
+        self.bdist_base = None
+        self.all = None
+
+    def finalize_options(self):
+        self.set_undefined_options('build',
+                                   ('build_base', 'build_base'),
+                                   ('build_lib', 'build_lib'),
+                                   ('build_scripts', 'build_scripts'),
+                                   ('build_temp', 'build_temp'))
+        self.set_undefined_options('bdist',
+                                   ('bdist_base', 'bdist_base'))
+
+    def run(self):
+        # remove the build/temp.<plat> directory (unless it's already
+        # gone)
+        if os.path.exists(self.build_temp):
+            remove_tree(self.build_temp, dry_run=self.dry_run)
+        else:
+            log.debug("'%s' does not exist -- can't clean it",
+                      self.build_temp)
+
+        if self.all:
+            # remove build directories
+            for directory in (self.build_lib,
+                              self.bdist_base,
+                              self.build_scripts):
+                if os.path.exists(directory):
+                    remove_tree(directory, dry_run=self.dry_run)
+                else:
+                    log.warn("'%s' does not exist -- can't clean it",
+                             directory)
+
+        # just for the heck of it, try to remove the base build directory:
+        # we might have emptied it right now, but if not we don't care
+        if not self.dry_run:
+            try:
+                os.rmdir(self.build_base)
+                log.info("removing '%s'", self.build_base)
+            except OSError:
+                pass
diff --git a/Lib/distutils/command/command_template b/Lib/distutils/command/command_template
new file mode 100644
index 000000000..6106819db
--- /dev/null
+++ b/Lib/distutils/command/command_template
@@ -0,0 +1,33 @@
+"""distutils.command.x
+
+Implements the Distutils 'x' command.
+"""
+
+# created 2000/mm/dd, John Doe
+
+__revision__ = "$Id$"
+
+from distutils.core import Command
+
+
+class x(Command):
+
+    # Brief (40-50 characters) description of the command
+    description = ""
+
+    # List of option tuples: long name, short name (None if no short
+    # name), and help string.
+    user_options = [('', '',
+                     ""),
+                   ]
+
+    def initialize_options(self):
+        self. = None
+        self. = None
+        self. = None
+
+    def finalize_options(self):
+        if self.x is None:
+            self.x = 
+
+    def run(self):
diff --git a/Lib/distutils/command/config.py b/Lib/distutils/command/config.py
new file mode 100644
index 000000000..4ae153d19
--- /dev/null
+++ b/Lib/distutils/command/config.py
@@ -0,0 +1,347 @@
+"""distutils.command.config
+
+Implements the Distutils 'config' command, a (mostly) empty command class
+that exists mainly to be sub-classed by specific module distributions and
+applications.  The idea is that while every "config" command is different,
+at least they're all named the same, and users always see "config" in the
+list of standard commands.  Also, this is a good place to put common
+configure-like tasks: "try to compile this C code", or "figure out where
+this header file lives".
+"""
+
+import os, re
+
+from distutils.core import Command
+from distutils.errors import DistutilsExecError
+from distutils.sysconfig import customize_compiler
+from distutils import log
+
+LANG_EXT = {"c": ".c", "c++": ".cxx"}
+
+class config(Command):
+
+    description = "prepare to build"
+
+    user_options = [
+        ('compiler=', None,
+         "specify the compiler type"),
+        ('cc=', None,
+         "specify the compiler executable"),
+        ('include-dirs=', 'I',
+         "list of directories to search for header files"),
+        ('define=', 'D',
+         "C preprocessor macros to define"),
+        ('undef=', 'U',
+         "C preprocessor macros to undefine"),
+        ('libraries=', 'l',
+         "external C libraries to link with"),
+        ('library-dirs=', 'L',
+         "directories to search for external C libraries"),
+
+        ('noisy', None,
+         "show every action (compile, link, run, ...) taken"),
+        ('dump-source', None,
+         "dump generated source files before attempting to compile them"),
+        ]
+
+
+    # The three standard command methods: since the "config" command
+    # does nothing by default, these are empty.
+
+    def initialize_options(self):
+        self.compiler = None
+        self.cc = None
+        self.include_dirs = None
+        self.libraries = None
+        self.library_dirs = None
+
+        # maximal output for now
+        self.noisy = 1
+        self.dump_source = 1
+
+        # list of temporary files generated along-the-way that we have
+        # to clean at some point
+        self.temp_files = []
+
+    def finalize_options(self):
+        if self.include_dirs is None:
+            self.include_dirs = self.distribution.include_dirs or []
+        elif isinstance(self.include_dirs, str):
+            self.include_dirs = self.include_dirs.split(os.pathsep)
+
+        if self.libraries is None:
+            self.libraries = []
+        elif isinstance(self.libraries, str):
+            self.libraries = [self.libraries]
+
+        if self.library_dirs is None:
+            self.library_dirs = []
+        elif isinstance(self.library_dirs, str):
+            self.library_dirs = self.library_dirs.split(os.pathsep)
+
+    def run(self):
+        pass
+
+    # Utility methods for actual "config" commands.  The interfaces are
+    # loosely based on Autoconf macros of similar names.  Sub-classes
+    # may use these freely.
+
+    def _check_compiler(self):
+        """Check that 'self.compiler' really is a CCompiler object;
+        if not, make it one.
+        """
+        # We do this late, and only on-demand, because this is an expensive
+        # import.
+        from distutils.ccompiler import CCompiler, new_compiler
+        if not isinstance(self.compiler, CCompiler):
+            self.compiler = new_compiler(compiler=self.compiler,
+                                         dry_run=self.dry_run, force=1)
+            customize_compiler(self.compiler)
+            if self.include_dirs:
+                self.compiler.set_include_dirs(self.include_dirs)
+            if self.libraries:
+                self.compiler.set_libraries(self.libraries)
+            if self.library_dirs:
+                self.compiler.set_library_dirs(self.library_dirs)
+
+    def _gen_temp_sourcefile(self, body, headers, lang):
+        filename = "_configtest" + LANG_EXT[lang]
+        file = open(filename, "w")
+        if headers:
+            for header in headers:
+                file.write("#include <%s>\n" % header)
+            file.write("\n")
+        file.write(body)
+        if body[-1] != "\n":
+            file.write("\n")
+        file.close()
+        return filename
+
+    def _preprocess(self, body, headers, include_dirs, lang):
+        src = self._gen_temp_sourcefile(body, headers, lang)
+        out = "_configtest.i"
+        self.temp_files.extend([src, out])
+        self.compiler.preprocess(src, out, include_dirs=include_dirs)
+        return (src, out)
+
+    def _compile(self, body, headers, include_dirs, lang):
+        src = self._gen_temp_sourcefile(body, headers, lang)
+        if self.dump_source:
+            dump_file(src, "compiling '%s':" % src)
+        (obj,) = self.compiler.object_filenames([src])
+        self.temp_files.extend([src, obj])
+        self.compiler.compile([src], include_dirs=include_dirs)
+        return (src, obj)
+
+    def _link(self, body, headers, include_dirs, libraries, library_dirs,
+              lang):
+        (src, obj) = self._compile(body, headers, include_dirs, lang)
+        prog = os.path.splitext(os.path.basename(src))[0]
+        self.compiler.link_executable([obj], prog,
+                                      libraries=libraries,
+                                      library_dirs=library_dirs,
+                                      target_lang=lang)
+
+        if self.compiler.exe_extension is not None:
+            prog = prog + self.compiler.exe_extension
+        self.temp_files.append(prog)
+
+        return (src, obj, prog)
+
+    def _clean(self, *filenames):
+        if not filenames:
+            filenames = self.temp_files
+            self.temp_files = []
+        log.info("removing: %s", ' '.join(filenames))
+        for filename in filenames:
+            try:
+                os.remove(filename)
+            except OSError:
+                pass
+
+
+    # XXX these ignore the dry-run flag: what to do, what to do? even if
+    # you want a dry-run build, you still need some sort of configuration
+    # info.  My inclination is to make it up to the real config command to
+    # consult 'dry_run', and assume a default (minimal) configuration if
+    # true.  The problem with trying to do it here is that you'd have to
+    # return either true or false from all the 'try' methods, neither of
+    # which is correct.
+
+    # XXX need access to the header search path and maybe default macros.
+
+    def try_cpp(self, body=None, headers=None, include_dirs=None, lang="c"):
+        """Construct a source file from 'body' (a string containing lines
+        of C/C++ code) and 'headers' (a list of header files to include)
+        and run it through the preprocessor.  Return true if the
+        preprocessor succeeded, false if there were any errors.
+        ('body' probably isn't of much use, but what the heck.)
+        """
+        from distutils.ccompiler import CompileError
+        self._check_compiler()
+        ok = True
+        try:
+            self._preprocess(body, headers, include_dirs, lang)
+        except CompileError:
+            ok = False
+
+        self._clean()
+        return ok
+
+    def search_cpp(self, pattern, body=None, headers=None, include_dirs=None,
+                   lang="c"):
+        """Construct a source file (just like 'try_cpp()'), run it through
+        the preprocessor, and return true if any line of the output matches
+        'pattern'.  'pattern' should either be a compiled regex object or a
+        string containing a regex.  If both 'body' and 'headers' are None,
+        preprocesses an empty file -- which can be useful to determine the
+        symbols the preprocessor and compiler set by default.
+        """
+        self._check_compiler()
+        src, out = self._preprocess(body, headers, include_dirs, lang)
+
+        if isinstance(pattern, str):
+            pattern = re.compile(pattern)
+
+        file = open(out)
+        match = False
+        while True:
+            line = file.readline()
+            if line == '':
+                break
+            if pattern.search(line):
+                match = True
+                break
+
+        file.close()
+        self._clean()
+        return match
+
+    def try_compile(self, body, headers=None, include_dirs=None, lang="c"):
+        """Try to compile a source file built from 'body' and 'headers'.
+        Return true on success, false otherwise.
+        """
+        from distutils.ccompiler import CompileError
+        self._check_compiler()
+        try:
+            self._compile(body, headers, include_dirs, lang)
+            ok = True
+        except CompileError:
+            ok = False
+
+        log.info(ok and "success!" or "failure.")
+        self._clean()
+        return ok
+
+    def try_link(self, body, headers=None, include_dirs=None, libraries=None,
+                 library_dirs=None, lang="c"):
+        """Try to compile and link a source file, built from 'body' and
+        'headers', to executable form.  Return true on success, false
+        otherwise.
+        """
+        from distutils.ccompiler import CompileError, LinkError
+        self._check_compiler()
+        try:
+            self._link(body, headers, include_dirs,
+                       libraries, library_dirs, lang)
+            ok = True
+        except (CompileError, LinkError):
+            ok = False
+
+        log.info(ok and "success!" or "failure.")
+        self._clean()
+        return ok
+
+    def try_run(self, body, headers=None, include_dirs=None, libraries=None,
+                library_dirs=None, lang="c"):
+        """Try to compile, link to an executable, and run a program
+        built from 'body' and 'headers'.  Return true on success, false
+        otherwise.
+        """
+        from distutils.ccompiler import CompileError, LinkError
+        self._check_compiler()
+        try:
+            src, obj, exe = self._link(body, headers, include_dirs,
+                                       libraries, library_dirs, lang)
+            self.spawn([exe])
+            ok = True
+        except (CompileError, LinkError, DistutilsExecError):
+            ok = False
+
+        log.info(ok and "success!" or "failure.")
+        self._clean()
+        return ok
+
+
+    # -- High-level methods --------------------------------------------
+    # (these are the ones that are actually likely to be useful
+    # when implementing a real-world config command!)
+
+    def check_func(self, func, headers=None, include_dirs=None,
+                   libraries=None, library_dirs=None, decl=0, call=0):
+        """Determine if function 'func' is available by constructing a
+        source file that refers to 'func', and compiles and links it.
+        If everything succeeds, returns true; otherwise returns false.
+
+        The constructed source file starts out by including the header
+        files listed in 'headers'.  If 'decl' is true, it then declares
+        'func' (as "int func()"); you probably shouldn't supply 'headers'
+        and set 'decl' true in the same call, or you might get errors about
+        a conflicting declarations for 'func'.  Finally, the constructed
+        'main()' function either references 'func' or (if 'call' is true)
+        calls it.  'libraries' and 'library_dirs' are used when
+        linking.
+        """
+        self._check_compiler()
+        body = []
+        if decl:
+            body.append("int %s ();" % func)
+        body.append("int main () {")
+        if call:
+            body.append("  %s();" % func)
+        else:
+            body.append("  %s;" % func)
+        body.append("}")
+        body = "\n".join(body) + "\n"
+
+        return self.try_link(body, headers, include_dirs,
+                             libraries, library_dirs)
+
+    def check_lib(self, library, library_dirs=None, headers=None,
+                  include_dirs=None, other_libraries=[]):
+        """Determine if 'library' is available to be linked against,
+        without actually checking that any particular symbols are provided
+        by it.  'headers' will be used in constructing the source file to
+        be compiled, but the only effect of this is to check if all the
+        header files listed are available.  Any libraries listed in
+        'other_libraries' will be included in the link, in case 'library'
+        has symbols that depend on other libraries.
+        """
+        self._check_compiler()
+        return self.try_link("int main (void) { }", headers, include_dirs,
+                             [library] + other_libraries, library_dirs)
+
+    def check_header(self, header, include_dirs=None, library_dirs=None,
+                     lang="c"):
+        """Determine if the system header file named by 'header_file'
+        exists and can be found by the preprocessor; return true if so,
+        false otherwise.
+        """
+        return self.try_cpp(body="/* No body */", headers=[header],
+                            include_dirs=include_dirs)
+
+
+def dump_file(filename, head=None):
+    """Dumps a file content into log.info.
+
+    If head is not None, will be dumped before the file content.
+    """
+    if head is None:
+        log.info('%s', filename)
+    else:
+        log.info(head)
+    file = open(filename)
+    try:
+        log.info(file.read())
+    finally:
+        file.close()
diff --git a/Lib/distutils/command/install.py b/Lib/distutils/command/install.py
new file mode 100644
index 000000000..078b3d59b
--- /dev/null
+++ b/Lib/distutils/command/install.py
@@ -0,0 +1,700 @@
+"""distutils.command.install
+
+Implements the Distutils 'install' command."""
+
+import sys
+import os
+
+from distutils import log
+from distutils.core import Command
+from distutils.debug import DEBUG
+from distutils.sysconfig import get_config_vars
+from distutils.errors import DistutilsPlatformError
+from distutils.file_util import write_file
+from distutils.util import convert_path, subst_vars, change_root
+from distutils.util import get_platform
+from distutils.errors import DistutilsOptionError
+
+from site import USER_BASE
+from site import USER_SITE
+HAS_USER_SITE = True
+
+WINDOWS_SCHEME = {
+    'purelib': '$base/Lib/site-packages',
+    'platlib': '$base/Lib/site-packages',
+    'headers': '$base/Include/$dist_name',
+    'scripts': '$base/Scripts',
+    'data'   : '$base',
+}
+
+INSTALL_SCHEMES = {
+    'unix_prefix': {
+        'purelib': '$base/lib/python$py_version_short/site-packages',
+        'platlib': '$platbase/lib/python$py_version_short/site-packages',
+        'headers': '$base/include/python$py_version_short$abiflags/$dist_name',
+        'scripts': '$base/bin',
+        'data'   : '$base',
+        },
+    'unix_local': {
+        'purelib': '$base/local/lib/python$py_version_short/dist-packages',
+        'platlib': '$platbase/local/lib/python$py_version_short/dist-packages',
+        'headers': '$base/local/include/python$py_version_short/$dist_name',
+        'scripts': '$base/local/bin',
+        'data'   : '$base/local',
+        },
+    'deb_system': {
+        'purelib': '$base/lib/python3/dist-packages',
+        'platlib': '$platbase/lib/python3/dist-packages',
+        'headers': '$base/include/python$py_version_short/$dist_name',
+        'scripts': '$base/bin',
+        'data'   : '$base',
+        },
+    'unix_home': {
+        'purelib': '$base/lib/python',
+        'platlib': '$base/lib/python',
+        'headers': '$base/include/python/$dist_name',
+        'scripts': '$base/bin',
+        'data'   : '$base',
+        },
+    'nt': WINDOWS_SCHEME,
+    }
+
+# user site schemes
+if HAS_USER_SITE:
+    INSTALL_SCHEMES['nt_user'] = {
+        'purelib': '$usersite',
+        'platlib': '$usersite',
+        'headers': '$userbase/Python$py_version_nodot/Include/$dist_name',
+        'scripts': '$userbase/Python$py_version_nodot/Scripts',
+        'data'   : '$userbase',
+        }
+
+    INSTALL_SCHEMES['unix_user'] = {
+        'purelib': '$usersite',
+        'platlib': '$usersite',
+        'headers':
+            '$userbase/include/python$py_version_short$abiflags/$dist_name',
+        'scripts': '$userbase/bin',
+        'data'   : '$userbase',
+        }
+
+# The keys to an installation scheme; if any new types of files are to be
+# installed, be sure to add an entry to every installation scheme above,
+# and to SCHEME_KEYS here.
+SCHEME_KEYS = ('purelib', 'platlib', 'headers', 'scripts', 'data')
+
+
+class install(Command):
+
+    description = "install everything from build directory"
+
+    user_options = [
+        # Select installation scheme and set base director(y|ies)
+        ('prefix=', None,
+         "installation prefix"),
+        ('exec-prefix=', None,
+         "(Unix only) prefix for platform-specific files"),
+        ('home=', None,
+         "(Unix only) home directory to install under"),
+
+        # Or, just set the base director(y|ies)
+        ('install-base=', None,
+         "base installation directory (instead of --prefix or --home)"),
+        ('install-platbase=', None,
+         "base installation directory for platform-specific files " +
+         "(instead of --exec-prefix or --home)"),
+        ('root=', None,
+         "install everything relative to this alternate root directory"),
+
+        # Or, explicitly set the installation scheme
+        ('install-purelib=', None,
+         "installation directory for pure Python module distributions"),
+        ('install-platlib=', None,
+         "installation directory for non-pure module distributions"),
+        ('install-lib=', None,
+         "installation directory for all module distributions " +
+         "(overrides --install-purelib and --install-platlib)"),
+
+        ('install-headers=', None,
+         "installation directory for C/C++ headers"),
+        ('install-scripts=', None,
+         "installation directory for Python scripts"),
+        ('install-data=', None,
+         "installation directory for data files"),
+
+        # Byte-compilation options -- see install_lib.py for details, as
+        # these are duplicated from there (but only install_lib does
+        # anything with them).
+        ('compile', 'c', "compile .py to .pyc [default]"),
+        ('no-compile', None, "don't compile .py files"),
+        ('optimize=', 'O',
+         "also compile with optimization: -O1 for \"python -O\", "
+         "-O2 for \"python -OO\", and -O0 to disable [default: -O0]"),
+
+        # Miscellaneous control options
+        ('force', 'f',
+         "force installation (overwrite any existing files)"),
+        ('skip-build', None,
+         "skip rebuilding everything (for testing/debugging)"),
+
+        # Where to install documentation (eventually!)
+        #('doc-format=', None, "format of documentation to generate"),
+        #('install-man=', None, "directory for Unix man pages"),
+        #('install-html=', None, "directory for HTML documentation"),
+        #('install-info=', None, "directory for GNU info files"),
+
+        ('record=', None,
+         "filename in which to record list of installed files"),
+
+        ('install-layout=', None,
+         "installation layout to choose (known values: deb, unix)"),
+        ]
+
+    boolean_options = ['compile', 'force', 'skip-build']
+
+    if HAS_USER_SITE:
+        user_options.append(('user', None,
+                             "install in user site-package '%s'" % USER_SITE))
+        boolean_options.append('user')
+
+    negative_opt = {'no-compile' : 'compile'}
+
+
+    def initialize_options(self):
+        """Initializes options."""
+        # High-level options: these select both an installation base
+        # and scheme.
+        self.prefix = None
+        self.exec_prefix = None
+        self.home = None
+        self.user = 0
+        self.prefix_option = None
+
+        # These select only the installation base; it's up to the user to
+        # specify the installation scheme (currently, that means supplying
+        # the --install-{platlib,purelib,scripts,data} options).
+        self.install_base = None
+        self.install_platbase = None
+        self.root = None
+
+        # These options are the actual installation directories; if not
+        # supplied by the user, they are filled in using the installation
+        # scheme implied by prefix/exec-prefix/home and the contents of
+        # that installation scheme.
+        self.install_purelib = None     # for pure module distributions
+        self.install_platlib = None     # non-pure (dists w/ extensions)
+        self.install_headers = None     # for C/C++ headers
+        self.install_lib = None         # set to either purelib or platlib
+        self.install_scripts = None
+        self.install_data = None
+        self.install_userbase = USER_BASE
+        self.install_usersite = USER_SITE
+
+        # enable custom installation, known values: deb
+        self.install_layout = None
+        self.multiarch = None
+        
+        self.compile = None
+        self.optimize = None
+
+        # Deprecated
+        # These two are for putting non-packagized distributions into their
+        # own directory and creating a .pth file if it makes sense.
+        # 'extra_path' comes from the setup file; 'install_path_file' can
+        # be turned off if it makes no sense to install a .pth file.  (But
+        # better to install it uselessly than to guess wrong and not
+        # install it when it's necessary and would be used!)  Currently,
+        # 'install_path_file' is always true unless some outsider meddles
+        # with it.
+        self.extra_path = None
+        self.install_path_file = 1
+
+        # 'force' forces installation, even if target files are not
+        # out-of-date.  'skip_build' skips running the "build" command,
+        # handy if you know it's not necessary.  'warn_dir' (which is *not*
+        # a user option, it's just there so the bdist_* commands can turn
+        # it off) determines whether we warn about installing to a
+        # directory not in sys.path.
+        self.force = 0
+        self.skip_build = 0
+        self.warn_dir = 1
+
+        # These are only here as a conduit from the 'build' command to the
+        # 'install_*' commands that do the real work.  ('build_base' isn't
+        # actually used anywhere, but it might be useful in future.)  They
+        # are not user options, because if the user told the install
+        # command where the build directory is, that wouldn't affect the
+        # build command.
+        self.build_base = None
+        self.build_lib = None
+
+        # Not defined yet because we don't know anything about
+        # documentation yet.
+        #self.install_man = None
+        #self.install_html = None
+        #self.install_info = None
+
+        self.record = None
+
+
+    # -- Option finalizing methods -------------------------------------
+    # (This is rather more involved than for most commands,
+    # because this is where the policy for installing third-
+    # party Python modules on various platforms given a wide
+    # array of user input is decided.  Yes, it's quite complex!)
+
+    def finalize_options(self):
+        """Finalizes options."""
+        # This method (and its pliant slaves, like 'finalize_unix()',
+        # 'finalize_other()', and 'select_scheme()') is where the default
+        # installation directories for modules, extension modules, and
+        # anything else we care to install from a Python module
+        # distribution.  Thus, this code makes a pretty important policy
+        # statement about how third-party stuff is added to a Python
+        # installation!  Note that the actual work of installation is done
+        # by the relatively simple 'install_*' commands; they just take
+        # their orders from the installation directory options determined
+        # here.
+
+        # Check for errors/inconsistencies in the options; first, stuff
+        # that's wrong on any platform.
+
+        if ((self.prefix or self.exec_prefix or self.home) and
+            (self.install_base or self.install_platbase)):
+            raise DistutilsOptionError(
+                   "must supply either prefix/exec-prefix/home or " +
+                   "install-base/install-platbase -- not both")
+
+        if self.home and (self.prefix or self.exec_prefix):
+            raise DistutilsOptionError(
+                  "must supply either home or prefix/exec-prefix -- not both")
+
+        if self.user and (self.prefix or self.exec_prefix or self.home or
+                self.install_base or self.install_platbase):
+            raise DistutilsOptionError("can't combine user with prefix, "
+                                       "exec_prefix/home, or install_(plat)base")
+
+        # Next, stuff that's wrong (or dubious) only on certain platforms.
+        if os.name != "posix":
+            if self.exec_prefix:
+                self.warn("exec-prefix option ignored on this platform")
+                self.exec_prefix = None
+
+        # Now the interesting logic -- so interesting that we farm it out
+        # to other methods.  The goal of these methods is to set the final
+        # values for the install_{lib,scripts,data,...}  options, using as
+        # input a heady brew of prefix, exec_prefix, home, install_base,
+        # install_platbase, user-supplied versions of
+        # install_{purelib,platlib,lib,scripts,data,...}, and the
+        # INSTALL_SCHEME dictionary above.  Phew!
+
+        self.dump_dirs("pre-finalize_{unix,other}")
+
+        if os.name == 'posix':
+            self.finalize_unix()
+        else:
+            self.finalize_other()
+
+        self.dump_dirs("post-finalize_{unix,other}()")
+
+        # Expand configuration variables, tilde, etc. in self.install_base
+        # and self.install_platbase -- that way, we can use $base or
+        # $platbase in the other installation directories and not worry
+        # about needing recursive variable expansion (shudder).
+
+        py_version = sys.version.split()[0]
+        (prefix, exec_prefix) = get_config_vars('prefix', 'exec_prefix')
+        try:
+            abiflags = sys.abiflags
+        except AttributeError:
+            # sys.abiflags may not be defined on all platforms.
+            abiflags = ''
+        self.config_vars = {'dist_name': self.distribution.get_name(),
+                            'dist_version': self.distribution.get_version(),
+                            'dist_fullname': self.distribution.get_fullname(),
+                            'py_version': py_version,
+                            'py_version_short': '%d.%d' % sys.version_info[:2],
+                            'py_version_nodot': '%d%d' % sys.version_info[:2],
+                            'sys_prefix': prefix,
+                            'prefix': prefix,
+                            'sys_exec_prefix': exec_prefix,
+                            'exec_prefix': exec_prefix,
+                            'abiflags': abiflags,
+                           }
+
+        if HAS_USER_SITE:
+            self.config_vars['userbase'] = self.install_userbase
+            self.config_vars['usersite'] = self.install_usersite
+
+        self.expand_basedirs()
+
+        self.dump_dirs("post-expand_basedirs()")
+
+        # Now define config vars for the base directories so we can expand
+        # everything else.
+        self.config_vars['base'] = self.install_base
+        self.config_vars['platbase'] = self.install_platbase
+
+        if DEBUG:
+            from pprint import pprint
+            print("config vars:")
+            pprint(self.config_vars)
+
+        # Expand "~" and configuration variables in the installation
+        # directories.
+        self.expand_dirs()
+
+        self.dump_dirs("post-expand_dirs()")
+
+        # Create directories in the home dir:
+        if self.user:
+            self.create_home_path()
+
+        # Pick the actual directory to install all modules to: either
+        # install_purelib or install_platlib, depending on whether this
+        # module distribution is pure or not.  Of course, if the user
+        # already specified install_lib, use their selection.
+        if self.install_lib is None:
+            if self.distribution.ext_modules: # has extensions: non-pure
+                self.install_lib = self.install_platlib
+            else:
+                self.install_lib = self.install_purelib
+
+
+        # Convert directories from Unix /-separated syntax to the local
+        # convention.
+        self.convert_paths('lib', 'purelib', 'platlib',
+                           'scripts', 'data', 'headers',
+                           'userbase', 'usersite')
+
+        # Deprecated
+        # Well, we're not actually fully completely finalized yet: we still
+        # have to deal with 'extra_path', which is the hack for allowing
+        # non-packagized module distributions (hello, Numerical Python!) to
+        # get their own directories.
+        self.handle_extra_path()
+        self.install_libbase = self.install_lib # needed for .pth file
+        self.install_lib = os.path.join(self.install_lib, self.extra_dirs)
+
+        # If a new root directory was supplied, make all the installation
+        # dirs relative to it.
+        if self.root is not None:
+            self.change_roots('libbase', 'lib', 'purelib', 'platlib',
+                              'scripts', 'data', 'headers')
+
+        self.dump_dirs("after prepending root")
+
+        # Find out the build directories, ie. where to install from.
+        self.set_undefined_options('build',
+                                   ('build_base', 'build_base'),
+                                   ('build_lib', 'build_lib'))
+
+        # Punt on doc directories for now -- after all, we're punting on
+        # documentation completely!
+
+    def dump_dirs(self, msg):
+        """Dumps the list of user options."""
+        if not DEBUG:
+            return
+        from distutils.fancy_getopt import longopt_xlate
+        log.debug(msg + ":")
+        for opt in self.user_options:
+            opt_name = opt[0]
+            if opt_name[-1] == "=":
+                opt_name = opt_name[0:-1]
+            if opt_name in self.negative_opt:
+                opt_name = self.negative_opt[opt_name]
+                opt_name = opt_name.translate(longopt_xlate)
+                val = not getattr(self, opt_name)
+            else:
+                opt_name = opt_name.translate(longopt_xlate)
+                val = getattr(self, opt_name)
+            log.debug("  %s: %s", opt_name, val)
+
+    def finalize_unix(self):
+        """Finalizes options for posix platforms."""
+        if self.install_base is not None or self.install_platbase is not None:
+            if ((self.install_lib is None and
+                 self.install_purelib is None and
+                 self.install_platlib is None) or
+                self.install_headers is None or
+                self.install_scripts is None or
+                self.install_data is None):
+                raise DistutilsOptionError(
+                      "install-base or install-platbase supplied, but "
+                      "installation scheme is incomplete")
+            return
+
+        if self.user:
+            if self.install_userbase is None:
+                raise DistutilsPlatformError(
+                    "User base directory is not specified")
+            self.install_base = self.install_platbase = self.install_userbase
+            self.select_scheme("unix_user")
+        elif self.home is not None:
+            self.install_base = self.install_platbase = self.home
+            self.select_scheme("unix_home")
+        else:
+            self.prefix_option = self.prefix
+            if self.prefix is None:
+                if self.exec_prefix is not None:
+                    raise DistutilsOptionError(
+                          "must not supply exec-prefix without prefix")
+
+                self.prefix = os.path.normpath(sys.prefix)
+                self.exec_prefix = os.path.normpath(sys.exec_prefix)
+
+            else:
+                if self.exec_prefix is None:
+                    self.exec_prefix = self.prefix
+
+            self.install_base = self.prefix
+            self.install_platbase = self.exec_prefix
+            if self.install_layout:
+                if self.install_layout.lower() in ['deb']:
+                    import sysconfig
+                    self.multiarch = sysconfig.get_config_var('MULTIARCH')
+                    self.select_scheme("deb_system")
+                elif self.install_layout.lower() in ['unix']:
+                    self.select_scheme("unix_prefix")
+                else:
+                    raise DistutilsOptionError(
+                        "unknown value for --install-layout")
+            elif ((self.prefix_option and
+                   os.path.normpath(self.prefix) != '/usr/local')
+                  or sys.base_prefix != sys.prefix
+                  or 'PYTHONUSERBASE' in os.environ
+                  or 'VIRTUAL_ENV' in os.environ
+                  or 'real_prefix' in sys.__dict__):
+                self.select_scheme("unix_prefix")
+            else:
+                if os.path.normpath(self.prefix) == '/usr/local':
+                    self.prefix = self.exec_prefix = '/usr'
+                    self.install_base = self.install_platbase = '/usr'
+                self.select_scheme("unix_local")
+
+    def finalize_other(self):
+        """Finalizes options for non-posix platforms"""
+        if self.user:
+            if self.install_userbase is None:
+                raise DistutilsPlatformError(
+                    "User base directory is not specified")
+            self.install_base = self.install_platbase = self.install_userbase
+            self.select_scheme(os.name + "_user")
+        elif self.home is not None:
+            self.install_base = self.install_platbase = self.home
+            self.select_scheme("unix_home")
+        else:
+            if self.prefix is None:
+                self.prefix = os.path.normpath(sys.prefix)
+
+            self.install_base = self.install_platbase = self.prefix
+            try:
+                self.select_scheme(os.name)
+            except KeyError:
+                raise DistutilsPlatformError(
+                      "I don't know how to install stuff on '%s'" % os.name)
+
+    def select_scheme(self, name):
+        """Sets the install directories by applying the install schemes."""
+        # it's the caller's problem if they supply a bad name!
+        scheme = INSTALL_SCHEMES[name]
+        for key in SCHEME_KEYS:
+            attrname = 'install_' + key
+            if getattr(self, attrname) is None:
+                setattr(self, attrname, scheme[key])
+
+    def _expand_attrs(self, attrs):
+        for attr in attrs:
+            val = getattr(self, attr)
+            if val is not None:
+                if os.name == 'posix' or os.name == 'nt':
+                    val = os.path.expanduser(val)
+                val = subst_vars(val, self.config_vars)
+                setattr(self, attr, val)
+
+    def expand_basedirs(self):
+        """Calls `os.path.expanduser` on install_base, install_platbase and
+        root."""
+        self._expand_attrs(['install_base', 'install_platbase', 'root'])
+
+    def expand_dirs(self):
+        """Calls `os.path.expanduser` on install dirs."""
+        self._expand_attrs(['install_purelib', 'install_platlib',
+                            'install_lib', 'install_headers',
+                            'install_scripts', 'install_data',])
+
+    def convert_paths(self, *names):
+        """Call `convert_path` over `names`."""
+        for name in names:
+            attr = "install_" + name
+            setattr(self, attr, convert_path(getattr(self, attr)))
+
+    def handle_extra_path(self):
+        """Set `path_file` and `extra_dirs` using `extra_path`."""
+        if self.extra_path is None:
+            self.extra_path = self.distribution.extra_path
+
+        if self.extra_path is not None:
+            log.warn(
+                "Distribution option extra_path is deprecated. "
+                "See issue27919 for details."
+            )
+            if isinstance(self.extra_path, str):
+                self.extra_path = self.extra_path.split(',')
+
+            if len(self.extra_path) == 1:
+                path_file = extra_dirs = self.extra_path[0]
+            elif len(self.extra_path) == 2:
+                path_file, extra_dirs = self.extra_path
+            else:
+                raise DistutilsOptionError(
+                      "'extra_path' option must be a list, tuple, or "
+                      "comma-separated string with 1 or 2 elements")
+
+            # convert to local form in case Unix notation used (as it
+            # should be in setup scripts)
+            extra_dirs = convert_path(extra_dirs)
+        else:
+            path_file = None
+            extra_dirs = ''
+
+        # XXX should we warn if path_file and not extra_dirs? (in which
+        # case the path file would be harmless but pointless)
+        self.path_file = path_file
+        self.extra_dirs = extra_dirs
+
+    def change_roots(self, *names):
+        """Change the install directories pointed by name using root."""
+        for name in names:
+            attr = "install_" + name
+            setattr(self, attr, change_root(self.root, getattr(self, attr)))
+
+    def create_home_path(self):
+        """Create directories under ~."""
+        if not self.user:
+            return
+        home = convert_path(os.path.expanduser("~"))
+        for name, path in self.config_vars.items():
+            if path.startswith(home) and not os.path.isdir(path):
+                self.debug_print("os.makedirs('%s', 0o700)" % path)
+                os.makedirs(path, 0o700)
+
+    # -- Command execution methods -------------------------------------
+
+    def run(self):
+        """Runs the command."""
+        # Obviously have to build before we can install
+        if not self.skip_build:
+            self.run_command('build')
+            # If we built for any other platform, we can't install.
+            build_plat = self.distribution.get_command_obj('build').plat_name
+            # check warn_dir - it is a clue that the 'install' is happening
+            # internally, and not to sys.path, so we don't check the platform
+            # matches what we are running.
+            if self.warn_dir and build_plat != get_platform():
+                raise DistutilsPlatformError("Can't install when "
+                                             "cross-compiling")
+
+        # Run all sub-commands (at least those that need to be run)
+        for cmd_name in self.get_sub_commands():
+            self.run_command(cmd_name)
+
+        if self.path_file:
+            self.create_path_file()
+
+        # write list of installed files, if requested.
+        if self.record:
+            outputs = self.get_outputs()
+            if self.root:               # strip any package prefix
+                root_len = len(self.root)
+                for counter in range(len(outputs)):
+                    outputs[counter] = outputs[counter][root_len:]
+            self.execute(write_file,
+                         (self.record, outputs),
+                         "writing list of installed files to '%s'" %
+                         self.record)
+
+        sys_path = map(os.path.normpath, sys.path)
+        sys_path = map(os.path.normcase, sys_path)
+        install_lib = os.path.normcase(os.path.normpath(self.install_lib))
+        if (self.warn_dir and
+            not (self.path_file and self.install_path_file) and
+            install_lib not in sys_path):
+            log.debug(("modules installed to '%s', which is not in "
+                       "Python's module search path (sys.path) -- "
+                       "you'll have to change the search path yourself"),
+                       self.install_lib)
+
+    def create_path_file(self):
+        """Creates the .pth file"""
+        filename = os.path.join(self.install_libbase,
+                                self.path_file + ".pth")
+        if self.install_path_file:
+            self.execute(write_file,
+                         (filename, [self.extra_dirs]),
+                         "creating %s" % filename)
+        else:
+            self.warn("path file '%s' not created" % filename)
+
+
+    # -- Reporting methods ---------------------------------------------
+
+    def get_outputs(self):
+        """Assembles the outputs of all the sub-commands."""
+        outputs = []
+        for cmd_name in self.get_sub_commands():
+            cmd = self.get_finalized_command(cmd_name)
+            # Add the contents of cmd.get_outputs(), ensuring
+            # that outputs doesn't contain duplicate entries
+            for filename in cmd.get_outputs():
+                if filename not in outputs:
+                    outputs.append(filename)
+
+        if self.path_file and self.install_path_file:
+            outputs.append(os.path.join(self.install_libbase,
+                                        self.path_file + ".pth"))
+
+        return outputs
+
+    def get_inputs(self):
+        """Returns the inputs of all the sub-commands"""
+        # XXX gee, this looks familiar ;-(
+        inputs = []
+        for cmd_name in self.get_sub_commands():
+            cmd = self.get_finalized_command(cmd_name)
+            inputs.extend(cmd.get_inputs())
+
+        return inputs
+
+    # -- Predicates for sub-command list -------------------------------
+
+    def has_lib(self):
+        """Returns true if the current distribution has any Python
+        modules to install."""
+        return (self.distribution.has_pure_modules() or
+                self.distribution.has_ext_modules())
+
+    def has_headers(self):
+        """Returns true if the current distribution has any headers to
+        install."""
+        return self.distribution.has_headers()
+
+    def has_scripts(self):
+        """Returns true if the current distribution has any scripts to.
+        install."""
+        return self.distribution.has_scripts()
+
+    def has_data(self):
+        """Returns true if the current distribution has any data to.
+        install."""
+        return self.distribution.has_data_files()
+
+    # 'sub_commands': a list of commands this command might have to run to
+    # get its work done.  See cmd.py for more info.
+    sub_commands = [('install_lib',     has_lib),
+                    ('install_headers', has_headers),
+                    ('install_scripts', has_scripts),
+                    ('install_data',    has_data),
+                    ('install_egg_info', lambda self:True),
+                   ]
diff --git a/Lib/distutils/command/install_data.py b/Lib/distutils/command/install_data.py
new file mode 100644
index 000000000..947cd76a9
--- /dev/null
+++ b/Lib/distutils/command/install_data.py
@@ -0,0 +1,79 @@
+"""distutils.command.install_data
+
+Implements the Distutils 'install_data' command, for installing
+platform-independent data files."""
+
+# contributed by Bastian Kleineidam
+
+import os
+from distutils.core import Command
+from distutils.util import change_root, convert_path
+
+class install_data(Command):
+
+    description = "install data files"
+
+    user_options = [
+        ('install-dir=', 'd',
+         "base directory for installing data files "
+         "(default: installation base dir)"),
+        ('root=', None,
+         "install everything relative to this alternate root directory"),
+        ('force', 'f', "force installation (overwrite existing files)"),
+        ]
+
+    boolean_options = ['force']
+
+    def initialize_options(self):
+        self.install_dir = None
+        self.outfiles = []
+        self.root = None
+        self.force = 0
+        self.data_files = self.distribution.data_files
+        self.warn_dir = 1
+
+    def finalize_options(self):
+        self.set_undefined_options('install',
+                                   ('install_data', 'install_dir'),
+                                   ('root', 'root'),
+                                   ('force', 'force'),
+                                  )
+
+    def run(self):
+        self.mkpath(self.install_dir)
+        for f in self.data_files:
+            if isinstance(f, str):
+                # it's a simple file, so copy it
+                f = convert_path(f)
+                if self.warn_dir:
+                    self.warn("setup script did not provide a directory for "
+                              "'%s' -- installing right in '%s'" %
+                              (f, self.install_dir))
+                (out, _) = self.copy_file(f, self.install_dir)
+                self.outfiles.append(out)
+            else:
+                # it's a tuple with path to install to and a list of files
+                dir = convert_path(f[0])
+                if not os.path.isabs(dir):
+                    dir = os.path.join(self.install_dir, dir)
+                elif self.root:
+                    dir = change_root(self.root, dir)
+                self.mkpath(dir)
+
+                if f[1] == []:
+                    # If there are no files listed, the user must be
+                    # trying to create an empty directory, so add the
+                    # directory to the list of output files.
+                    self.outfiles.append(dir)
+                else:
+                    # Copy files, adding them to the list of output files.
+                    for data in f[1]:
+                        data = convert_path(data)
+                        (out, _) = self.copy_file(data, dir)
+                        self.outfiles.append(out)
+
+    def get_inputs(self):
+        return self.data_files or []
+
+    def get_outputs(self):
+        return self.outfiles
diff --git a/Lib/distutils/command/install_egg_info.py b/Lib/distutils/command/install_egg_info.py
new file mode 100644
index 000000000..0a71b6100
--- /dev/null
+++ b/Lib/distutils/command/install_egg_info.py
@@ -0,0 +1,97 @@
+"""distutils.command.install_egg_info
+
+Implements the Distutils 'install_egg_info' command, for installing
+a package's PKG-INFO metadata."""
+
+
+from distutils.cmd import Command
+from distutils import log, dir_util
+import os, sys, re
+
+class install_egg_info(Command):
+    """Install an .egg-info file for the package"""
+
+    description = "Install package's PKG-INFO metadata as an .egg-info file"
+    user_options = [
+        ('install-dir=', 'd', "directory to install to"),
+        ('install-layout', None, "custom installation layout"),
+    ]
+
+    def initialize_options(self):
+        self.install_dir = None
+        self.install_layout = None
+        self.prefix_option = None
+
+    def finalize_options(self):
+        self.set_undefined_options('install_lib',('install_dir','install_dir'))
+        self.set_undefined_options('install',('install_layout','install_layout'))
+        self.set_undefined_options('install',('prefix_option','prefix_option'))
+        if self.install_layout:
+            if not self.install_layout.lower() in ['deb', 'unix']:
+                raise DistutilsOptionError(
+                    "unknown value for --install-layout")
+            no_pyver = (self.install_layout.lower() == 'deb')
+        elif self.prefix_option:
+            no_pyver = False
+        else:
+            no_pyver = True
+        if no_pyver:
+            basename = "%s-%s.egg-info" % (
+                to_filename(safe_name(self.distribution.get_name())),
+                to_filename(safe_version(self.distribution.get_version()))
+                )
+        else:
+            basename = "%s-%s-py%d.%d.egg-info" % (
+                to_filename(safe_name(self.distribution.get_name())),
+                to_filename(safe_version(self.distribution.get_version())),
+                *sys.version_info[:2]
+            )
+        self.target = os.path.join(self.install_dir, basename)
+        self.outputs = [self.target]
+
+    def run(self):
+        target = self.target
+        if os.path.isdir(target) and not os.path.islink(target):
+            dir_util.remove_tree(target, dry_run=self.dry_run)
+        elif os.path.exists(target):
+            self.execute(os.unlink,(self.target,),"Removing "+target)
+        elif not os.path.isdir(self.install_dir):
+            self.execute(os.makedirs, (self.install_dir,),
+                         "Creating "+self.install_dir)
+        log.info("Writing %s", target)
+        if not self.dry_run:
+            with open(target, 'w', encoding='UTF-8') as f:
+                self.distribution.metadata.write_pkg_file(f)
+
+    def get_outputs(self):
+        return self.outputs
+
+
+# The following routines are taken from setuptools' pkg_resources module and
+# can be replaced by importing them from pkg_resources once it is included
+# in the stdlib.
+
+def safe_name(name):
+    """Convert an arbitrary string to a standard distribution name
+
+    Any runs of non-alphanumeric/. characters are replaced with a single '-'.
+    """
+    return re.sub('[^A-Za-z0-9.]+', '-', name)
+
+
+def safe_version(version):
+    """Convert an arbitrary string to a standard version string
+
+    Spaces become dots, and all other non-alphanumeric characters become
+    dashes, with runs of multiple dashes condensed to a single dash.
+    """
+    version = version.replace(' ','.')
+    return re.sub('[^A-Za-z0-9.]+', '-', version)
+
+
+def to_filename(name):
+    """Convert a project or version name to its filename-escaped form
+
+    Any '-' characters are currently replaced with '_'.
+    """
+    return name.replace('-','_')
diff --git a/Lib/distutils/command/install_headers.py b/Lib/distutils/command/install_headers.py
new file mode 100644
index 000000000..9bb0b18dc
--- /dev/null
+++ b/Lib/distutils/command/install_headers.py
@@ -0,0 +1,47 @@
+"""distutils.command.install_headers
+
+Implements the Distutils 'install_headers' command, to install C/C++ header
+files to the Python include directory."""
+
+from distutils.core import Command
+
+
+# XXX force is never used
+class install_headers(Command):
+
+    description = "install C/C++ header files"
+
+    user_options = [('install-dir=', 'd',
+                     "directory to install header files to"),
+                    ('force', 'f',
+                     "force installation (overwrite existing files)"),
+                   ]
+
+    boolean_options = ['force']
+
+    def initialize_options(self):
+        self.install_dir = None
+        self.force = 0
+        self.outfiles = []
+
+    def finalize_options(self):
+        self.set_undefined_options('install',
+                                   ('install_headers', 'install_dir'),
+                                   ('force', 'force'))
+
+
+    def run(self):
+        headers = self.distribution.headers
+        if not headers:
+            return
+
+        self.mkpath(self.install_dir)
+        for header in headers:
+            (out, _) = self.copy_file(header, self.install_dir)
+            self.outfiles.append(out)
+
+    def get_inputs(self):
+        return self.distribution.headers or []
+
+    def get_outputs(self):
+        return self.outfiles
diff --git a/Lib/distutils/command/install_lib.py b/Lib/distutils/command/install_lib.py
new file mode 100644
index 000000000..eef63626f
--- /dev/null
+++ b/Lib/distutils/command/install_lib.py
@@ -0,0 +1,221 @@
+"""distutils.command.install_lib
+
+Implements the Distutils 'install_lib' command
+(install all Python modules)."""
+
+import os
+import importlib.util
+import sys
+
+from distutils.core import Command
+from distutils.errors import DistutilsOptionError
+
+
+# Extension for Python source files.
+PYTHON_SOURCE_EXTENSION = ".py"
+
+class install_lib(Command):
+
+    description = "install all Python modules (extensions and pure Python)"
+
+    # The byte-compilation options are a tad confusing.  Here are the
+    # possible scenarios:
+    #   1) no compilation at all (--no-compile --no-optimize)
+    #   2) compile .pyc only (--compile --no-optimize; default)
+    #   3) compile .pyc and "opt-1" .pyc (--compile --optimize)
+    #   4) compile "opt-1" .pyc only (--no-compile --optimize)
+    #   5) compile .pyc and "opt-2" .pyc (--compile --optimize-more)
+    #   6) compile "opt-2" .pyc only (--no-compile --optimize-more)
+    #
+    # The UI for this is two options, 'compile' and 'optimize'.
+    # 'compile' is strictly boolean, and only decides whether to
+    # generate .pyc files.  'optimize' is three-way (0, 1, or 2), and
+    # decides both whether to generate .pyc files and what level of
+    # optimization to use.
+
+    user_options = [
+        ('install-dir=', 'd', "directory to install to"),
+        ('build-dir=','b', "build directory (where to install from)"),
+        ('force', 'f', "force installation (overwrite existing files)"),
+        ('compile', 'c', "compile .py to .pyc [default]"),
+        ('no-compile', None, "don't compile .py files"),
+        ('optimize=', 'O',
+         "also compile with optimization: -O1 for \"python -O\", "
+         "-O2 for \"python -OO\", and -O0 to disable [default: -O0]"),
+        ('skip-build', None, "skip the build steps"),
+        ]
+
+    boolean_options = ['force', 'compile', 'skip-build']
+    negative_opt = {'no-compile' : 'compile'}
+
+    def initialize_options(self):
+        # let the 'install' command dictate our installation directory
+        self.install_dir = None
+        self.build_dir = None
+        self.force = 0
+        self.compile = None
+        self.optimize = None
+        self.skip_build = None
+        self.multiarch = None # if we should rename the extensions
+
+    def finalize_options(self):
+        # Get all the information we need to install pure Python modules
+        # from the umbrella 'install' command -- build (source) directory,
+        # install (target) directory, and whether to compile .py files.
+        self.set_undefined_options('install',
+                                   ('build_lib', 'build_dir'),
+                                   ('install_lib', 'install_dir'),
+                                   ('force', 'force'),
+                                   ('compile', 'compile'),
+                                   ('optimize', 'optimize'),
+                                   ('skip_build', 'skip_build'),
+                                   ('multiarch', 'multiarch'),
+                                  )
+
+        if self.compile is None:
+            self.compile = True
+        if self.optimize is None:
+            self.optimize = False
+
+        if not isinstance(self.optimize, int):
+            try:
+                self.optimize = int(self.optimize)
+                if self.optimize not in (0, 1, 2):
+                    raise AssertionError
+            except (ValueError, AssertionError):
+                raise DistutilsOptionError("optimize must be 0, 1, or 2")
+
+    def run(self):
+        # Make sure we have built everything we need first
+        self.build()
+
+        # Install everything: simply dump the entire contents of the build
+        # directory to the installation directory (that's the beauty of
+        # having a build directory!)
+        outfiles = self.install()
+
+        # (Optionally) compile .py to .pyc
+        if outfiles is not None and self.distribution.has_pure_modules():
+            self.byte_compile(outfiles)
+
+    # -- Top-level worker functions ------------------------------------
+    # (called from 'run()')
+
+    def build(self):
+        if not self.skip_build:
+            if self.distribution.has_pure_modules():
+                self.run_command('build_py')
+            if self.distribution.has_ext_modules():
+                self.run_command('build_ext')
+
+    def install(self):
+        if os.path.isdir(self.build_dir):
+            import distutils.dir_util
+            distutils.dir_util._multiarch = self.multiarch
+            outfiles = self.copy_tree(self.build_dir, self.install_dir)
+        else:
+            self.warn("'%s' does not exist -- no Python modules to install" %
+                      self.build_dir)
+            return
+        return outfiles
+
+    def byte_compile(self, files):
+        if sys.dont_write_bytecode:
+            self.warn('byte-compiling is disabled, skipping.')
+            return
+
+        from distutils.util import byte_compile
+
+        # Get the "--root" directory supplied to the "install" command,
+        # and use it as a prefix to strip off the purported filename
+        # encoded in bytecode files.  This is far from complete, but it
+        # should at least generate usable bytecode in RPM distributions.
+        install_root = self.get_finalized_command('install').root
+
+        if self.compile:
+            byte_compile(files, optimize=0,
+                         force=self.force, prefix=install_root,
+                         dry_run=self.dry_run)
+        if self.optimize > 0:
+            byte_compile(files, optimize=self.optimize,
+                         force=self.force, prefix=install_root,
+                         verbose=self.verbose, dry_run=self.dry_run)
+
+
+    # -- Utility methods -----------------------------------------------
+
+    def _mutate_outputs(self, has_any, build_cmd, cmd_option, output_dir):
+        if not has_any:
+            return []
+
+        build_cmd = self.get_finalized_command(build_cmd)
+        build_files = build_cmd.get_outputs()
+        build_dir = getattr(build_cmd, cmd_option)
+
+        prefix_len = len(build_dir) + len(os.sep)
+        outputs = []
+        for file in build_files:
+            outputs.append(os.path.join(output_dir, file[prefix_len:]))
+
+        return outputs
+
+    def _bytecode_filenames(self, py_filenames):
+        bytecode_files = []
+        for py_file in py_filenames:
+            # Since build_py handles package data installation, the
+            # list of outputs can contain more than just .py files.
+            # Make sure we only report bytecode for the .py files.
+            ext = os.path.splitext(os.path.normcase(py_file))[1]
+            if ext != PYTHON_SOURCE_EXTENSION:
+                continue
+            if self.compile:
+                bytecode_files.append(importlib.util.cache_from_source(
+                    py_file, optimization=''))
+            if self.optimize > 0:
+                bytecode_files.append(importlib.util.cache_from_source(
+                    py_file, optimization=self.optimize))
+
+        return bytecode_files
+
+
+    # -- External interface --------------------------------------------
+    # (called by outsiders)
+
+    def get_outputs(self):
+        """Return the list of files that would be installed if this command
+        were actually run.  Not affected by the "dry-run" flag or whether
+        modules have actually been built yet.
+        """
+        pure_outputs = \
+            self._mutate_outputs(self.distribution.has_pure_modules(),
+                                 'build_py', 'build_lib',
+                                 self.install_dir)
+        if self.compile:
+            bytecode_outputs = self._bytecode_filenames(pure_outputs)
+        else:
+            bytecode_outputs = []
+
+        ext_outputs = \
+            self._mutate_outputs(self.distribution.has_ext_modules(),
+                                 'build_ext', 'build_lib',
+                                 self.install_dir)
+
+        return pure_outputs + bytecode_outputs + ext_outputs
+
+    def get_inputs(self):
+        """Get the list of files that are input to this command, ie. the
+        files that get installed as they are named in the build tree.
+        The files in this list correspond one-to-one to the output
+        filenames returned by 'get_outputs()'.
+        """
+        inputs = []
+
+        if self.distribution.has_pure_modules():
+            build_py = self.get_finalized_command('build_py')
+            inputs.extend(build_py.get_outputs())
+
+        if self.distribution.has_ext_modules():
+            build_ext = self.get_finalized_command('build_ext')
+            inputs.extend(build_ext.get_outputs())
+
+        return inputs
diff --git a/Lib/distutils/command/install_scripts.py b/Lib/distutils/command/install_scripts.py
new file mode 100644
index 000000000..31a1130ee
--- /dev/null
+++ b/Lib/distutils/command/install_scripts.py
@@ -0,0 +1,60 @@
+"""distutils.command.install_scripts
+
+Implements the Distutils 'install_scripts' command, for installing
+Python scripts."""
+
+# contributed by Bastian Kleineidam
+
+import os
+from distutils.core import Command
+from distutils import log
+from stat import ST_MODE
+
+
+class install_scripts(Command):
+
+    description = "install scripts (Python or otherwise)"
+
+    user_options = [
+        ('install-dir=', 'd', "directory to install scripts to"),
+        ('build-dir=','b', "build directory (where to install from)"),
+        ('force', 'f', "force installation (overwrite existing files)"),
+        ('skip-build', None, "skip the build steps"),
+    ]
+
+    boolean_options = ['force', 'skip-build']
+
+    def initialize_options(self):
+        self.install_dir = None
+        self.force = 0
+        self.build_dir = None
+        self.skip_build = None
+
+    def finalize_options(self):
+        self.set_undefined_options('build', ('build_scripts', 'build_dir'))
+        self.set_undefined_options('install',
+                                   ('install_scripts', 'install_dir'),
+                                   ('force', 'force'),
+                                   ('skip_build', 'skip_build'),
+                                  )
+
+    def run(self):
+        if not self.skip_build:
+            self.run_command('build_scripts')
+        self.outfiles = self.copy_tree(self.build_dir, self.install_dir)
+        if os.name == 'posix':
+            # Set the executable bits (owner, group, and world) on
+            # all the scripts we just installed.
+            for file in self.get_outputs():
+                if self.dry_run:
+                    log.info("changing mode of %s", file)
+                else:
+                    mode = ((os.stat(file)[ST_MODE]) | 0o555) & 0o7777
+                    log.info("changing mode of %s to %o", file, mode)
+                    os.chmod(file, mode)
+
+    def get_inputs(self):
+        return self.distribution.scripts or []
+
+    def get_outputs(self):
+        return self.outfiles or []
diff --git a/Lib/distutils/command/register.py b/Lib/distutils/command/register.py
new file mode 100644
index 000000000..0fac94e9e
--- /dev/null
+++ b/Lib/distutils/command/register.py
@@ -0,0 +1,304 @@
+"""distutils.command.register
+
+Implements the Distutils 'register' command (register with the repository).
+"""
+
+# created 2002/10/21, Richard Jones
+
+import getpass
+import io
+import urllib.parse, urllib.request
+from warnings import warn
+
+from distutils.core import PyPIRCCommand
+from distutils.errors import *
+from distutils import log
+
+class register(PyPIRCCommand):
+
+    description = ("register the distribution with the Python package index")
+    user_options = PyPIRCCommand.user_options + [
+        ('list-classifiers', None,
+         'list the valid Trove classifiers'),
+        ('strict', None ,
+         'Will stop the registering if the meta-data are not fully compliant')
+        ]
+    boolean_options = PyPIRCCommand.boolean_options + [
+        'verify', 'list-classifiers', 'strict']
+
+    sub_commands = [('check', lambda self: True)]
+
+    def initialize_options(self):
+        PyPIRCCommand.initialize_options(self)
+        self.list_classifiers = 0
+        self.strict = 0
+
+    def finalize_options(self):
+        PyPIRCCommand.finalize_options(self)
+        # setting options for the `check` subcommand
+        check_options = {'strict': ('register', self.strict),
+                         'restructuredtext': ('register', 1)}
+        self.distribution.command_options['check'] = check_options
+
+    def run(self):
+        self.finalize_options()
+        self._set_config()
+
+        # Run sub commands
+        for cmd_name in self.get_sub_commands():
+            self.run_command(cmd_name)
+
+        if self.dry_run:
+            self.verify_metadata()
+        elif self.list_classifiers:
+            self.classifiers()
+        else:
+            self.send_metadata()
+
+    def check_metadata(self):
+        """Deprecated API."""
+        warn("distutils.command.register.check_metadata is deprecated, \
+              use the check command instead", PendingDeprecationWarning)
+        check = self.distribution.get_command_obj('check')
+        check.ensure_finalized()
+        check.strict = self.strict
+        check.restructuredtext = 1
+        check.run()
+
+    def _set_config(self):
+        ''' Reads the configuration file and set attributes.
+        '''
+        config = self._read_pypirc()
+        if config != {}:
+            self.username = config['username']
+            self.password = config['password']
+            self.repository = config['repository']
+            self.realm = config['realm']
+            self.has_config = True
+        else:
+            if self.repository not in ('pypi', self.DEFAULT_REPOSITORY):
+                raise ValueError('%s not found in .pypirc' % self.repository)
+            if self.repository == 'pypi':
+                self.repository = self.DEFAULT_REPOSITORY
+            self.has_config = False
+
+    def classifiers(self):
+        ''' Fetch the list of classifiers from the server.
+        '''
+        url = self.repository+'?:action=list_classifiers'
+        response = urllib.request.urlopen(url)
+        log.info(self._read_pypi_response(response))
+
+    def verify_metadata(self):
+        ''' Send the metadata to the package index server to be checked.
+        '''
+        # send the info to the server and report the result
+        (code, result) = self.post_to_server(self.build_post_data('verify'))
+        log.info('Server response (%s): %s', code, result)
+
+    def send_metadata(self):
+        ''' Send the metadata to the package index server.
+
+            Well, do the following:
+            1. figure who the user is, and then
+            2. send the data as a Basic auth'ed POST.
+
+            First we try to read the username/password from $HOME/.pypirc,
+            which is a ConfigParser-formatted file with a section
+            [distutils] containing username and password entries (both
+            in clear text). Eg:
+
+                [distutils]
+                index-servers =
+                    pypi
+
+                [pypi]
+                username: fred
+                password: sekrit
+
+            Otherwise, to figure who the user is, we offer the user three
+            choices:
+
+             1. use existing login,
+             2. register as a new user, or
+             3. set the password to a random string and email the user.
+
+        '''
+        # see if we can short-cut and get the username/password from the
+        # config
+        if self.has_config:
+            choice = '1'
+            username = self.username
+            password = self.password
+        else:
+            choice = 'x'
+            username = password = ''
+
+        # get the user's login info
+        choices = '1 2 3 4'.split()
+        while choice not in choices:
+            self.announce('''\
+We need to know who you are, so please choose either:
+ 1. use your existing login,
+ 2. register as a new user,
+ 3. have the server generate a new password for you (and email it to you), or
+ 4. quit
+Your selection [default 1]: ''', log.INFO)
+            choice = input()
+            if not choice:
+                choice = '1'
+            elif choice not in choices:
+                print('Please choose one of the four options!')
+
+        if choice == '1':
+            # get the username and password
+            while not username:
+                username = input('Username: ')
+            while not password:
+                password = getpass.getpass('Password: ')
+
+            # set up the authentication
+            auth = urllib.request.HTTPPasswordMgr()
+            host = urllib.parse.urlparse(self.repository)[1]
+            auth.add_password(self.realm, host, username, password)
+            # send the info to the server and report the result
+            code, result = self.post_to_server(self.build_post_data('submit'),
+                auth)
+            self.announce('Server response (%s): %s' % (code, result),
+                          log.INFO)
+
+            # possibly save the login
+            if code == 200:
+                if self.has_config:
+                    # sharing the password in the distribution instance
+                    # so the upload command can reuse it
+                    self.distribution.password = password
+                else:
+                    self.announce(('I can store your PyPI login so future '
+                                   'submissions will be faster.'), log.INFO)
+                    self.announce('(the login will be stored in %s)' % \
+                                  self._get_rc_file(), log.INFO)
+                    choice = 'X'
+                    while choice.lower() not in 'yn':
+                        choice = input('Save your login (y/N)?')
+                        if not choice:
+                            choice = 'n'
+                    if choice.lower() == 'y':
+                        self._store_pypirc(username, password)
+
+        elif choice == '2':
+            data = {':action': 'user'}
+            data['name'] = data['password'] = data['email'] = ''
+            data['confirm'] = None
+            while not data['name']:
+                data['name'] = input('Username: ')
+            while data['password'] != data['confirm']:
+                while not data['password']:
+                    data['password'] = getpass.getpass('Password: ')
+                while not data['confirm']:
+                    data['confirm'] = getpass.getpass(' Confirm: ')
+                if data['password'] != data['confirm']:
+                    data['password'] = ''
+                    data['confirm'] = None
+                    print("Password and confirm don't match!")
+            while not data['email']:
+                data['email'] = input('   EMail: ')
+            code, result = self.post_to_server(data)
+            if code != 200:
+                log.info('Server response (%s): %s', code, result)
+            else:
+                log.info('You will receive an email shortly.')
+                log.info(('Follow the instructions in it to '
+                          'complete registration.'))
+        elif choice == '3':
+            data = {':action': 'password_reset'}
+            data['email'] = ''
+            while not data['email']:
+                data['email'] = input('Your email address: ')
+            code, result = self.post_to_server(data)
+            log.info('Server response (%s): %s', code, result)
+
+    def build_post_data(self, action):
+        # figure the data to send - the metadata plus some additional
+        # information used by the package server
+        meta = self.distribution.metadata
+        data = {
+            ':action': action,
+            'metadata_version' : '1.0',
+            'name': meta.get_name(),
+            'version': meta.get_version(),
+            'summary': meta.get_description(),
+            'home_page': meta.get_url(),
+            'author': meta.get_contact(),
+            'author_email': meta.get_contact_email(),
+            'license': meta.get_licence(),
+            'description': meta.get_long_description(),
+            'keywords': meta.get_keywords(),
+            'platform': meta.get_platforms(),
+            'classifiers': meta.get_classifiers(),
+            'download_url': meta.get_download_url(),
+            # PEP 314
+            'provides': meta.get_provides(),
+            'requires': meta.get_requires(),
+            'obsoletes': meta.get_obsoletes(),
+        }
+        if data['provides'] or data['requires'] or data['obsoletes']:
+            data['metadata_version'] = '1.1'
+        return data
+
+    def post_to_server(self, data, auth=None):
+        ''' Post a query to the server, and return a string response.
+        '''
+        if 'name' in data:
+            self.announce('Registering %s to %s' % (data['name'],
+                                                    self.repository),
+                                                    log.INFO)
+        # Build up the MIME payload for the urllib2 POST data
+        boundary = '--------------GHSKFJDLGDS7543FJKLFHRE75642756743254'
+        sep_boundary = '\n--' + boundary
+        end_boundary = sep_boundary + '--'
+        body = io.StringIO()
+        for key, value in data.items():
+            # handle multiple entries for the same name
+            if type(value) not in (type([]), type( () )):
+                value = [value]
+            for value in value:
+                value = str(value)
+                body.write(sep_boundary)
+                body.write('\nContent-Disposition: form-data; name="%s"'%key)
+                body.write("\n\n")
+                body.write(value)
+                if value and value[-1] == '\r':
+                    body.write('\n')  # write an extra newline (lurve Macs)
+        body.write(end_boundary)
+        body.write("\n")
+        body = body.getvalue().encode("utf-8")
+
+        # build the Request
+        headers = {
+            'Content-type': 'multipart/form-data; boundary=%s; charset=utf-8'%boundary,
+            'Content-length': str(len(body))
+        }
+        req = urllib.request.Request(self.repository, body, headers)
+
+        # handle HTTP and include the Basic Auth handler
+        opener = urllib.request.build_opener(
+            urllib.request.HTTPBasicAuthHandler(password_mgr=auth)
+        )
+        data = ''
+        try:
+            result = opener.open(req)
+        except urllib.error.HTTPError as e:
+            if self.show_response:
+                data = e.fp.read()
+            result = e.code, e.msg
+        except urllib.error.URLError as e:
+            result = 500, str(e)
+        else:
+            if self.show_response:
+                data = self._read_pypi_response(result)
+            result = 200, 'OK'
+        if self.show_response:
+            msg = '\n'.join(('-' * 75, data, '-' * 75))
+            self.announce(msg, log.INFO)
+        return result
diff --git a/Lib/distutils/command/sdist.py b/Lib/distutils/command/sdist.py
new file mode 100644
index 000000000..4fd1d4715
--- /dev/null
+++ b/Lib/distutils/command/sdist.py
@@ -0,0 +1,456 @@
+"""distutils.command.sdist
+
+Implements the Distutils 'sdist' command (create a source distribution)."""
+
+import os
+import sys
+from types import *
+from glob import glob
+from warnings import warn
+
+from distutils.core import Command
+from distutils import dir_util, dep_util, file_util, archive_util
+from distutils.text_file import TextFile
+from distutils.errors import *
+from distutils.filelist import FileList
+from distutils import log
+from distutils.util import convert_path
+
+def show_formats():
+    """Print all possible values for the 'formats' option (used by
+    the "--help-formats" command-line option).
+    """
+    from distutils.fancy_getopt import FancyGetopt
+    from distutils.archive_util import ARCHIVE_FORMATS
+    formats = []
+    for format in ARCHIVE_FORMATS.keys():
+        formats.append(("formats=" + format, None,
+                        ARCHIVE_FORMATS[format][2]))
+    formats.sort()
+    FancyGetopt(formats).print_help(
+        "List of available source distribution formats:")
+
+class sdist(Command):
+
+    description = "create a source distribution (tarball, zip file, etc.)"
+
+    def checking_metadata(self):
+        """Callable used for the check sub-command.
+
+        Placed here so user_options can view it"""
+        return self.metadata_check
+
+    user_options = [
+        ('template=', 't',
+         "name of manifest template file [default: MANIFEST.in]"),
+        ('manifest=', 'm',
+         "name of manifest file [default: MANIFEST]"),
+        ('use-defaults', None,
+         "include the default file set in the manifest "
+         "[default; disable with --no-defaults]"),
+        ('no-defaults', None,
+         "don't include the default file set"),
+        ('prune', None,
+         "specifically exclude files/directories that should not be "
+         "distributed (build tree, RCS/CVS dirs, etc.) "
+         "[default; disable with --no-prune]"),
+        ('no-prune', None,
+         "don't automatically exclude anything"),
+        ('manifest-only', 'o',
+         "just regenerate the manifest and then stop "
+         "(implies --force-manifest)"),
+        ('force-manifest', 'f',
+         "forcibly regenerate the manifest and carry on as usual. "
+         "Deprecated: now the manifest is always regenerated."),
+        ('formats=', None,
+         "formats for source distribution (comma-separated list)"),
+        ('keep-temp', 'k',
+         "keep the distribution tree around after creating " +
+         "archive file(s)"),
+        ('dist-dir=', 'd',
+         "directory to put the source distribution archive(s) in "
+         "[default: dist]"),
+        ('metadata-check', None,
+         "Ensure that all required elements of meta-data "
+         "are supplied. Warn if any missing. [default]"),
+        ('owner=', 'u',
+         "Owner name used when creating a tar file [default: current user]"),
+        ('group=', 'g',
+         "Group name used when creating a tar file [default: current group]"),
+        ]
+
+    boolean_options = ['use-defaults', 'prune',
+                       'manifest-only', 'force-manifest',
+                       'keep-temp', 'metadata-check']
+
+    help_options = [
+        ('help-formats', None,
+         "list available distribution formats", show_formats),
+        ]
+
+    negative_opt = {'no-defaults': 'use-defaults',
+                    'no-prune': 'prune' }
+
+    sub_commands = [('check', checking_metadata)]
+
+    def initialize_options(self):
+        # 'template' and 'manifest' are, respectively, the names of
+        # the manifest template and manifest file.
+        self.template = None
+        self.manifest = None
+
+        # 'use_defaults': if true, we will include the default file set
+        # in the manifest
+        self.use_defaults = 1
+        self.prune = 1
+
+        self.manifest_only = 0
+        self.force_manifest = 0
+
+        self.formats = ['gztar']
+        self.keep_temp = 0
+        self.dist_dir = None
+
+        self.archive_files = None
+        self.metadata_check = 1
+        self.owner = None
+        self.group = None
+
+    def finalize_options(self):
+        if self.manifest is None:
+            self.manifest = "MANIFEST"
+        if self.template is None:
+            self.template = "MANIFEST.in"
+
+        self.ensure_string_list('formats')
+
+        bad_format = archive_util.check_archive_formats(self.formats)
+        if bad_format:
+            raise DistutilsOptionError(
+                  "unknown archive format '%s'" % bad_format)
+
+        if self.dist_dir is None:
+            self.dist_dir = "dist"
+
+    def run(self):
+        # 'filelist' contains the list of files that will make up the
+        # manifest
+        self.filelist = FileList()
+
+        # Run sub commands
+        for cmd_name in self.get_sub_commands():
+            self.run_command(cmd_name)
+
+        # Do whatever it takes to get the list of files to process
+        # (process the manifest template, read an existing manifest,
+        # whatever).  File list is accumulated in 'self.filelist'.
+        self.get_file_list()
+
+        # If user just wanted us to regenerate the manifest, stop now.
+        if self.manifest_only:
+            return
+
+        # Otherwise, go ahead and create the source distribution tarball,
+        # or zipfile, or whatever.
+        self.make_distribution()
+
+    def check_metadata(self):
+        """Deprecated API."""
+        warn("distutils.command.sdist.check_metadata is deprecated, \
+              use the check command instead", PendingDeprecationWarning)
+        check = self.distribution.get_command_obj('check')
+        check.ensure_finalized()
+        check.run()
+
+    def get_file_list(self):
+        """Figure out the list of files to include in the source
+        distribution, and put it in 'self.filelist'.  This might involve
+        reading the manifest template (and writing the manifest), or just
+        reading the manifest, or just using the default file set -- it all
+        depends on the user's options.
+        """
+        # new behavior when using a template:
+        # the file list is recalculated every time because
+        # even if MANIFEST.in or setup.py are not changed
+        # the user might have added some files in the tree that
+        # need to be included.
+        #
+        #  This makes --force the default and only behavior with templates.
+        template_exists = os.path.isfile(self.template)
+        if not template_exists and self._manifest_is_not_generated():
+            self.read_manifest()
+            self.filelist.sort()
+            self.filelist.remove_duplicates()
+            return
+
+        if not template_exists:
+            self.warn(("manifest template '%s' does not exist " +
+                        "(using default file list)") %
+                        self.template)
+        self.filelist.findall()
+
+        if self.use_defaults:
+            self.add_defaults()
+
+        if template_exists:
+            self.read_template()
+
+        if self.prune:
+            self.prune_file_list()
+
+        self.filelist.sort()
+        self.filelist.remove_duplicates()
+        self.write_manifest()
+
+    def add_defaults(self):
+        """Add all the default files to self.filelist:
+          - README or README.txt
+          - setup.py
+          - test/test*.py
+          - all pure Python modules mentioned in setup script
+          - all files pointed by package_data (build_py)
+          - all files defined in data_files.
+          - all files defined as scripts.
+          - all C sources listed as part of extensions or C libraries
+            in the setup script (doesn't catch C headers!)
+        Warns if (README or README.txt) or setup.py are missing; everything
+        else is optional.
+        """
+        standards = [('README', 'README.txt'), self.distribution.script_name]
+        for fn in standards:
+            if isinstance(fn, tuple):
+                alts = fn
+                got_it = False
+                for fn in alts:
+                    if os.path.exists(fn):
+                        got_it = True
+                        self.filelist.append(fn)
+                        break
+
+                if not got_it:
+                    self.warn("standard file not found: should have one of " +
+                              ', '.join(alts))
+            else:
+                if os.path.exists(fn):
+                    self.filelist.append(fn)
+                else:
+                    self.warn("standard file '%s' not found" % fn)
+
+        optional = ['test/test*.py', 'setup.cfg']
+        for pattern in optional:
+            files = filter(os.path.isfile, glob(pattern))
+            self.filelist.extend(files)
+
+        # build_py is used to get:
+        #  - python modules
+        #  - files defined in package_data
+        build_py = self.get_finalized_command('build_py')
+
+        # getting python files
+        if self.distribution.has_pure_modules():
+            self.filelist.extend(build_py.get_source_files())
+
+        # getting package_data files
+        # (computed in build_py.data_files by build_py.finalize_options)
+        for pkg, src_dir, build_dir, filenames in build_py.data_files:
+            for filename in filenames:
+                self.filelist.append(os.path.join(src_dir, filename))
+
+        # getting distribution.data_files
+        if self.distribution.has_data_files():
+            for item in self.distribution.data_files:
+                if isinstance(item, str): # plain file
+                    item = convert_path(item)
+                    if os.path.isfile(item):
+                        self.filelist.append(item)
+                else:    # a (dirname, filenames) tuple
+                    dirname, filenames = item
+                    for f in filenames:
+                        f = convert_path(f)
+                        if os.path.isfile(f):
+                            self.filelist.append(f)
+
+        if self.distribution.has_ext_modules():
+            build_ext = self.get_finalized_command('build_ext')
+            self.filelist.extend(build_ext.get_source_files())
+
+        if self.distribution.has_c_libraries():
+            build_clib = self.get_finalized_command('build_clib')
+            self.filelist.extend(build_clib.get_source_files())
+
+        if self.distribution.has_scripts():
+            build_scripts = self.get_finalized_command('build_scripts')
+            self.filelist.extend(build_scripts.get_source_files())
+
+    def read_template(self):
+        """Read and parse manifest template file named by self.template.
+
+        (usually "MANIFEST.in") The parsing and processing is done by
+        'self.filelist', which updates itself accordingly.
+        """
+        log.info("reading manifest template '%s'", self.template)
+        template = TextFile(self.template, strip_comments=1, skip_blanks=1,
+                            join_lines=1, lstrip_ws=1, rstrip_ws=1,
+                            collapse_join=1)
+
+        try:
+            while True:
+                line = template.readline()
+                if line is None:            # end of file
+                    break
+
+                try:
+                    self.filelist.process_template_line(line)
+                # the call above can raise a DistutilsTemplateError for
+                # malformed lines, or a ValueError from the lower-level
+                # convert_path function
+                except (DistutilsTemplateError, ValueError) as msg:
+                    self.warn("%s, line %d: %s" % (template.filename,
+                                                   template.current_line,
+                                                   msg))
+        finally:
+            template.close()
+
+    def prune_file_list(self):
+        """Prune off branches that might slip into the file list as created
+        by 'read_template()', but really don't belong there:
+          * the build tree (typically "build")
+          * the release tree itself (only an issue if we ran "sdist"
+            previously with --keep-temp, or it aborted)
+          * any RCS, CVS, .svn, .hg, .git, .bzr, _darcs directories
+        """
+        build = self.get_finalized_command('build')
+        base_dir = self.distribution.get_fullname()
+
+        self.filelist.exclude_pattern(None, prefix=build.build_base)
+        self.filelist.exclude_pattern(None, prefix=base_dir)
+
+        if sys.platform == 'win32':
+            seps = r'/|\\'
+        else:
+            seps = '/'
+
+        vcs_dirs = ['RCS', 'CVS', r'\.svn', r'\.hg', r'\.git', r'\.bzr',
+                    '_darcs']
+        vcs_ptrn = r'(^|%s)(%s)(%s).*' % (seps, '|'.join(vcs_dirs), seps)
+        self.filelist.exclude_pattern(vcs_ptrn, is_regex=1)
+
+    def write_manifest(self):
+        """Write the file list in 'self.filelist' (presumably as filled in
+        by 'add_defaults()' and 'read_template()') to the manifest file
+        named by 'self.manifest'.
+        """
+        if self._manifest_is_not_generated():
+            log.info("not writing to manually maintained "
+                     "manifest file '%s'" % self.manifest)
+            return
+
+        content = self.filelist.files[:]
+        content.insert(0, '# file GENERATED by distutils, do NOT edit')
+        self.execute(file_util.write_file, (self.manifest, content),
+                     "writing manifest file '%s'" % self.manifest)
+
+    def _manifest_is_not_generated(self):
+        # check for special comment used in 3.1.3 and higher
+        if not os.path.isfile(self.manifest):
+            return False
+
+        fp = open(self.manifest)
+        try:
+            first_line = fp.readline()
+        finally:
+            fp.close()
+        return first_line != '# file GENERATED by distutils, do NOT edit\n'
+
+    def read_manifest(self):
+        """Read the manifest file (named by 'self.manifest') and use it to
+        fill in 'self.filelist', the list of files to include in the source
+        distribution.
+        """
+        log.info("reading manifest file '%s'", self.manifest)
+        manifest = open(self.manifest)
+        for line in manifest:
+            # ignore comments and blank lines
+            line = line.strip()
+            if line.startswith('#') or not line:
+                continue
+            self.filelist.append(line)
+        manifest.close()
+
+    def make_release_tree(self, base_dir, files):
+        """Create the directory tree that will become the source
+        distribution archive.  All directories implied by the filenames in
+        'files' are created under 'base_dir', and then we hard link or copy
+        (if hard linking is unavailable) those files into place.
+        Essentially, this duplicates the developer's source tree, but in a
+        directory named after the distribution, containing only the files
+        to be distributed.
+        """
+        # Create all the directories under 'base_dir' necessary to
+        # put 'files' there; the 'mkpath()' is just so we don't die
+        # if the manifest happens to be empty.
+        self.mkpath(base_dir)
+        dir_util.create_tree(base_dir, files, dry_run=self.dry_run)
+
+        # And walk over the list of files, either making a hard link (if
+        # os.link exists) to each one that doesn't already exist in its
+        # corresponding location under 'base_dir', or copying each file
+        # that's out-of-date in 'base_dir'.  (Usually, all files will be
+        # out-of-date, because by default we blow away 'base_dir' when
+        # we're done making the distribution archives.)
+
+        if hasattr(os, 'link'):        # can make hard links on this system
+            link = 'hard'
+            msg = "making hard links in %s..." % base_dir
+        else:                           # nope, have to copy
+            link = None
+            msg = "copying files to %s..." % base_dir
+
+        if not files:
+            log.warn("no files to distribute -- empty manifest?")
+        else:
+            log.info(msg)
+        for file in files:
+            if not os.path.isfile(file):
+                log.warn("'%s' not a regular file -- skipping", file)
+            else:
+                dest = os.path.join(base_dir, file)
+                self.copy_file(file, dest, link=link)
+
+        self.distribution.metadata.write_pkg_info(base_dir)
+
+    def make_distribution(self):
+        """Create the source distribution(s).  First, we create the release
+        tree with 'make_release_tree()'; then, we create all required
+        archive files (according to 'self.formats') from the release tree.
+        Finally, we clean up by blowing away the release tree (unless
+        'self.keep_temp' is true).  The list of archive files created is
+        stored so it can be retrieved later by 'get_archive_files()'.
+        """
+        # Don't warn about missing meta-data here -- should be (and is!)
+        # done elsewhere.
+        base_dir = self.distribution.get_fullname()
+        base_name = os.path.join(self.dist_dir, base_dir)
+
+        self.make_release_tree(base_dir, self.filelist.files)
+        archive_files = []              # remember names of files we create
+        # tar archive must be created last to avoid overwrite and remove
+        if 'tar' in self.formats:
+            self.formats.append(self.formats.pop(self.formats.index('tar')))
+
+        for fmt in self.formats:
+            file = self.make_archive(base_name, fmt, base_dir=base_dir,
+                                     owner=self.owner, group=self.group)
+            archive_files.append(file)
+            self.distribution.dist_files.append(('sdist', '', file))
+
+        self.archive_files = archive_files
+
+        if not self.keep_temp:
+            dir_util.remove_tree(base_dir, dry_run=self.dry_run)
+
+    def get_archive_files(self):
+        """Return the list of archive files created when the command
+        was run, or None if the command hasn't run yet.
+        """
+        return self.archive_files
diff --git a/Lib/distutils/command/upload.py b/Lib/distutils/command/upload.py
new file mode 100644
index 000000000..32dda359b
--- /dev/null
+++ b/Lib/distutils/command/upload.py
@@ -0,0 +1,200 @@
+"""
+distutils.command.upload
+
+Implements the Distutils 'upload' subcommand (upload package to a package
+index).
+"""
+
+import os
+import io
+import platform
+import hashlib
+from base64 import standard_b64encode
+from urllib.request import urlopen, Request, HTTPError
+from urllib.parse import urlparse
+from distutils.errors import DistutilsError, DistutilsOptionError
+from distutils.core import PyPIRCCommand
+from distutils.spawn import spawn
+from distutils import log
+
+class upload(PyPIRCCommand):
+
+    description = "upload binary package to PyPI"
+
+    user_options = PyPIRCCommand.user_options + [
+        ('sign', 's',
+         'sign files to upload using gpg'),
+        ('identity=', 'i', 'GPG identity used to sign files'),
+        ]
+
+    boolean_options = PyPIRCCommand.boolean_options + ['sign']
+
+    def initialize_options(self):
+        PyPIRCCommand.initialize_options(self)
+        self.username = ''
+        self.password = ''
+        self.show_response = 0
+        self.sign = False
+        self.identity = None
+
+    def finalize_options(self):
+        PyPIRCCommand.finalize_options(self)
+        if self.identity and not self.sign:
+            raise DistutilsOptionError(
+                "Must use --sign for --identity to have meaning"
+            )
+        config = self._read_pypirc()
+        if config != {}:
+            self.username = config['username']
+            self.password = config['password']
+            self.repository = config['repository']
+            self.realm = config['realm']
+
+        # getting the password from the distribution
+        # if previously set by the register command
+        if not self.password and self.distribution.password:
+            self.password = self.distribution.password
+
+    def run(self):
+        if not self.distribution.dist_files:
+            msg = ("Must create and upload files in one command "
+                   "(e.g. setup.py sdist upload)")
+            raise DistutilsOptionError(msg)
+        for command, pyversion, filename in self.distribution.dist_files:
+            self.upload_file(command, pyversion, filename)
+
+    def upload_file(self, command, pyversion, filename):
+        # Makes sure the repository URL is compliant
+        schema, netloc, url, params, query, fragments = \
+            urlparse(self.repository)
+        if params or query or fragments:
+            raise AssertionError("Incompatible url %s" % self.repository)
+
+        if schema not in ('http', 'https'):
+            raise AssertionError("unsupported schema " + schema)
+
+        # Sign if requested
+        if self.sign:
+            gpg_args = ["gpg", "--detach-sign", "-a", filename]
+            if self.identity:
+                gpg_args[2:2] = ["--local-user", self.identity]
+            spawn(gpg_args,
+                  dry_run=self.dry_run)
+
+        # Fill in the data - send all the meta-data in case we need to
+        # register a new release
+        f = open(filename,'rb')
+        try:
+            content = f.read()
+        finally:
+            f.close()
+        meta = self.distribution.metadata
+        data = {
+            # action
+            ':action': 'file_upload',
+            'protocol_version': '1',
+
+            # identify release
+            'name': meta.get_name(),
+            'version': meta.get_version(),
+
+            # file content
+            'content': (os.path.basename(filename),content),
+            'filetype': command,
+            'pyversion': pyversion,
+            'md5_digest': hashlib.md5(content).hexdigest(),
+
+            # additional meta-data
+            'metadata_version': '1.0',
+            'summary': meta.get_description(),
+            'home_page': meta.get_url(),
+            'author': meta.get_contact(),
+            'author_email': meta.get_contact_email(),
+            'license': meta.get_licence(),
+            'description': meta.get_long_description(),
+            'keywords': meta.get_keywords(),
+            'platform': meta.get_platforms(),
+            'classifiers': meta.get_classifiers(),
+            'download_url': meta.get_download_url(),
+            # PEP 314
+            'provides': meta.get_provides(),
+            'requires': meta.get_requires(),
+            'obsoletes': meta.get_obsoletes(),
+            }
+        comment = ''
+        if command == 'bdist_rpm':
+            dist, version, id = platform.dist()
+            if dist:
+                comment = 'built for %s %s' % (dist, version)
+        elif command == 'bdist_dumb':
+            comment = 'built for %s' % platform.platform(terse=1)
+        data['comment'] = comment
+
+        if self.sign:
+            data['gpg_signature'] = (os.path.basename(filename) + ".asc",
+                                     open(filename+".asc", "rb").read())
+
+        # set up the authentication
+        user_pass = (self.username + ":" + self.password).encode('ascii')
+        # The exact encoding of the authentication string is debated.
+        # Anyway PyPI only accepts ascii for both username or password.
+        auth = "Basic " + standard_b64encode(user_pass).decode('ascii')
+
+        # Build up the MIME payload for the POST data
+        boundary = '--------------GHSKFJDLGDS7543FJKLFHRE75642756743254'
+        sep_boundary = b'\r\n--' + boundary.encode('ascii')
+        end_boundary = sep_boundary + b'--\r\n'
+        body = io.BytesIO()
+        for key, value in data.items():
+            title = '\r\nContent-Disposition: form-data; name="%s"' % key
+            # handle multiple entries for the same name
+            if not isinstance(value, list):
+                value = [value]
+            for value in value:
+                if type(value) is tuple:
+                    title += '; filename="%s"' % value[0]
+                    value = value[1]
+                else:
+                    value = str(value).encode('utf-8')
+                body.write(sep_boundary)
+                body.write(title.encode('utf-8'))
+                body.write(b"\r\n\r\n")
+                body.write(value)
+        body.write(end_boundary)
+        body = body.getvalue()
+
+        msg = "Submitting %s to %s" % (filename, self.repository)
+        self.announce(msg, log.INFO)
+
+        # build the Request
+        headers = {
+            'Content-type': 'multipart/form-data; boundary=%s' % boundary,
+            'Content-length': str(len(body)),
+            'Authorization': auth,
+        }
+
+        request = Request(self.repository, data=body,
+                          headers=headers)
+        # send the data
+        try:
+            result = urlopen(request)
+            status = result.getcode()
+            reason = result.msg
+        except HTTPError as e:
+            status = e.code
+            reason = e.msg
+        except OSError as e:
+            self.announce(str(e), log.ERROR)
+            raise
+
+        if status == 200:
+            self.announce('Server response (%s): %s' % (status, reason),
+                          log.INFO)
+            if self.show_response:
+                text = self._read_pypi_response(result)
+                msg = '\n'.join(('-' * 75, text, '-' * 75))
+                self.announce(msg, log.INFO)
+        else:
+            msg = 'Upload failed (%s): %s' % (status, reason)
+            self.announce(msg, log.ERROR)
+            raise DistutilsError(msg)
diff --git a/Lib/distutils/config.py b/Lib/distutils/config.py
new file mode 100644
index 000000000..bf8d8dd2f
--- /dev/null
+++ b/Lib/distutils/config.py
@@ -0,0 +1,131 @@
+"""distutils.pypirc
+
+Provides the PyPIRCCommand class, the base class for the command classes
+that uses .pypirc in the distutils.command package.
+"""
+import os
+from configparser import RawConfigParser
+
+from distutils.cmd import Command
+
+DEFAULT_PYPIRC = """\
+[distutils]
+index-servers =
+    pypi
+
+[pypi]
+username:%s
+password:%s
+"""
+
+class PyPIRCCommand(Command):
+    """Base command that knows how to handle the .pypirc file
+    """
+    DEFAULT_REPOSITORY = 'https://upload.pypi.org/legacy/'
+    DEFAULT_REALM = 'pypi'
+    repository = None
+    realm = None
+
+    user_options = [
+        ('repository=', 'r',
+         "url of repository [default: %s]" % \
+            DEFAULT_REPOSITORY),
+        ('show-response', None,
+         'display full response text from server')]
+
+    boolean_options = ['show-response']
+
+    def _get_rc_file(self):
+        """Returns rc file path."""
+        return os.path.join(os.path.expanduser('~'), '.pypirc')
+
+    def _store_pypirc(self, username, password):
+        """Creates a default .pypirc file."""
+        rc = self._get_rc_file()
+        with os.fdopen(os.open(rc, os.O_CREAT | os.O_WRONLY, 0o600), 'w') as f:
+            f.write(DEFAULT_PYPIRC % (username, password))
+
+    def _read_pypirc(self):
+        """Reads the .pypirc file."""
+        rc = self._get_rc_file()
+        if os.path.exists(rc):
+            self.announce('Using PyPI login from %s' % rc)
+            repository = self.repository or self.DEFAULT_REPOSITORY
+            realm = self.realm or self.DEFAULT_REALM
+
+            config = RawConfigParser()
+            config.read(rc)
+            sections = config.sections()
+            if 'distutils' in sections:
+                # let's get the list of servers
+                index_servers = config.get('distutils', 'index-servers')
+                _servers = [server.strip() for server in
+                            index_servers.split('\n')
+                            if server.strip() != '']
+                if _servers == []:
+                    # nothing set, let's try to get the default pypi
+                    if 'pypi' in sections:
+                        _servers = ['pypi']
+                    else:
+                        # the file is not properly defined, returning
+                        # an empty dict
+                        return {}
+                for server in _servers:
+                    current = {'server': server}
+                    current['username'] = config.get(server, 'username')
+
+                    # optional params
+                    for key, default in (('repository',
+                                          self.DEFAULT_REPOSITORY),
+                                         ('realm', self.DEFAULT_REALM),
+                                         ('password', None)):
+                        if config.has_option(server, key):
+                            current[key] = config.get(server, key)
+                        else:
+                            current[key] = default
+
+                    # work around people having "repository" for the "pypi"
+                    # section of their config set to the HTTP (rather than
+                    # HTTPS) URL
+                    if (server == 'pypi' and
+                        repository in (self.DEFAULT_REPOSITORY, 'pypi')):
+                        current['repository'] = self.DEFAULT_REPOSITORY
+                        return current
+
+                    if (current['server'] == repository or
+                        current['repository'] == repository):
+                        return current
+            elif 'server-login' in sections:
+                # old format
+                server = 'server-login'
+                if config.has_option(server, 'repository'):
+                    repository = config.get(server, 'repository')
+                else:
+                    repository = self.DEFAULT_REPOSITORY
+                return {'username': config.get(server, 'username'),
+                        'password': config.get(server, 'password'),
+                        'repository': repository,
+                        'server': server,
+                        'realm': self.DEFAULT_REALM}
+
+        return {}
+
+    def _read_pypi_response(self, response):
+        """Read and decode a PyPI HTTP response."""
+        import cgi
+        content_type = response.getheader('content-type', 'text/plain')
+        encoding = cgi.parse_header(content_type)[1].get('charset', 'ascii')
+        return response.read().decode(encoding)
+
+    def initialize_options(self):
+        """Initialize options."""
+        self.repository = None
+        self.realm = None
+        self.show_response = 0
+
+    def finalize_options(self):
+        """Finalizes options."""
+        if self.repository is None:
+            self.repository = self.DEFAULT_REPOSITORY
+        if self.realm is None:
+            self.realm = self.DEFAULT_REALM
diff --git a/Lib/distutils/core.py b/Lib/distutils/core.py
new file mode 100644
index 000000000..d603d4a45
--- /dev/null
+++ b/Lib/distutils/core.py
@@ -0,0 +1,234 @@
+"""distutils.core
+
+The only module that needs to be imported to use the Distutils; provides
+the 'setup' function (which is to be called from the setup script).  Also
+indirectly provides the Distribution and Command classes, although they are
+really defined in distutils.dist and distutils.cmd.
+"""
+
+import os
+import sys
+
+from distutils.debug import DEBUG
+from distutils.errors import *
+
+# Mainly import these so setup scripts can "from distutils.core import" them.
+from distutils.dist import Distribution
+from distutils.cmd import Command
+from distutils.config import PyPIRCCommand
+from distutils.extension import Extension
+
+# This is a barebones help message generated displayed when the user
+# runs the setup script with no arguments at all.  More useful help
+# is generated with various --help options: global help, list commands,
+# and per-command help.
+USAGE = """\
+usage: %(script)s [global_opts] cmd1 [cmd1_opts] [cmd2 [cmd2_opts] ...]
+   or: %(script)s --help [cmd1 cmd2 ...]
+   or: %(script)s --help-commands
+   or: %(script)s cmd --help
+"""
+
+def gen_usage (script_name):
+    script = os.path.basename(script_name)
+    return USAGE % vars()
+
+
+# Some mild magic to control the behaviour of 'setup()' from 'run_setup()'.
+_setup_stop_after = None
+_setup_distribution = None
+
+# Legal keyword arguments for the setup() function
+setup_keywords = ('distclass', 'script_name', 'script_args', 'options',
+                  'name', 'version', 'author', 'author_email',
+                  'maintainer', 'maintainer_email', 'url', 'license',
+                  'description', 'long_description', 'keywords',
+                  'platforms', 'classifiers', 'download_url',
+                  'requires', 'provides', 'obsoletes',
+                  )
+
+# Legal keyword arguments for the Extension constructor
+extension_keywords = ('name', 'sources', 'include_dirs',
+                      'define_macros', 'undef_macros',
+                      'library_dirs', 'libraries', 'runtime_library_dirs',
+                      'extra_objects', 'extra_compile_args', 'extra_link_args',
+                      'swig_opts', 'export_symbols', 'depends', 'language')
+
+def setup (**attrs):
+    """The gateway to the Distutils: do everything your setup script needs
+    to do, in a highly flexible and user-driven way.  Briefly: create a
+    Distribution instance; find and parse config files; parse the command
+    line; run each Distutils command found there, customized by the options
+    supplied to 'setup()' (as keyword arguments), in config files, and on
+    the command line.
+
+    The Distribution instance might be an instance of a class supplied via
+    the 'distclass' keyword argument to 'setup'; if no such class is
+    supplied, then the Distribution class (in dist.py) is instantiated.
+    All other arguments to 'setup' (except for 'cmdclass') are used to set
+    attributes of the Distribution instance.
+
+    The 'cmdclass' argument, if supplied, is a dictionary mapping command
+    names to command classes.  Each command encountered on the command line
+    will be turned into a command class, which is in turn instantiated; any
+    class found in 'cmdclass' is used in place of the default, which is
+    (for command 'foo_bar') class 'foo_bar' in module
+    'distutils.command.foo_bar'.  The command class must provide a
+    'user_options' attribute which is a list of option specifiers for
+    'distutils.fancy_getopt'.  Any command-line options between the current
+    and the next command are used to set attributes of the current command
+    object.
+
+    When the entire command-line has been successfully parsed, calls the
+    'run()' method on each command object in turn.  This method will be
+    driven entirely by the Distribution object (which each command object
+    has a reference to, thanks to its constructor), and the
+    command-specific options that became attributes of each command
+    object.
+    """
+
+    global _setup_stop_after, _setup_distribution
+
+    # Determine the distribution class -- either caller-supplied or
+    # our Distribution (see below).
+    klass = attrs.get('distclass')
+    if klass:
+        del attrs['distclass']
+    else:
+        klass = Distribution
+
+    if 'script_name' not in attrs:
+        attrs['script_name'] = os.path.basename(sys.argv[0])
+    if 'script_args'  not in attrs:
+        attrs['script_args'] = sys.argv[1:]
+
+    # Create the Distribution instance, using the remaining arguments
+    # (ie. everything except distclass) to initialize it
+    try:
+        _setup_distribution = dist = klass(attrs)
+    except DistutilsSetupError as msg:
+        if 'name' not in attrs:
+            raise SystemExit("error in setup command: %s" % msg)
+        else:
+            raise SystemExit("error in %s setup command: %s" % \
+                  (attrs['name'], msg))
+
+    if _setup_stop_after == "init":
+        return dist
+
+    # Find and parse the config file(s): they will override options from
+    # the setup script, but be overridden by the command line.
+    dist.parse_config_files()
+
+    if DEBUG:
+        print("options (after parsing config files):")
+        dist.dump_option_dicts()
+
+    if _setup_stop_after == "config":
+        return dist
+
+    # Parse the command line and override config files; any
+    # command-line errors are the end user's fault, so turn them into
+    # SystemExit to suppress tracebacks.
+    try:
+        ok = dist.parse_command_line()
+    except DistutilsArgError as msg:
+        raise SystemExit(gen_usage(dist.script_name) + "\nerror: %s" % msg)
+
+    if DEBUG:
+        print("options (after parsing command line):")
+        dist.dump_option_dicts()
+
+    if _setup_stop_after == "commandline":
+        return dist
+
+    # And finally, run all the commands found on the command line.
+    if ok:
+        try:
+            dist.run_commands()
+        except KeyboardInterrupt:
+            raise SystemExit("interrupted")
+        except OSError as exc:
+            if DEBUG:
+                sys.stderr.write("error: %s\n" % (exc,))
+                raise
+            else:
+                raise SystemExit("error: %s" % (exc,))
+
+        except (DistutilsError,
+                CCompilerError) as msg:
+            if DEBUG:
+                raise
+            else:
+                raise SystemExit("error: " + str(msg))
+
+    return dist
+
+# setup ()
+
+
+def run_setup (script_name, script_args=None, stop_after="run"):
+    """Run a setup script in a somewhat controlled environment, and
+    return the Distribution instance that drives things.  This is useful
+    if you need to find out the distribution meta-data (passed as
+    keyword args from 'script' to 'setup()', or the contents of the
+    config files or command-line.
+
+    'script_name' is a file that will be read and run with 'exec()';
+    'sys.argv[0]' will be replaced with 'script' for the duration of the
+    call.  'script_args' is a list of strings; if supplied,
+    'sys.argv[1:]' will be replaced by 'script_args' for the duration of
+    the call.
+
+    'stop_after' tells 'setup()' when to stop processing; possible
+    values:
+      init
+        stop after the Distribution instance has been created and
+        populated with the keyword arguments to 'setup()'
+      config
+        stop after config files have been parsed (and their data
+        stored in the Distribution instance)
+      commandline
+        stop after the command-line ('sys.argv[1:]' or 'script_args')
+        have been parsed (and the data stored in the Distribution)
+      run [default]
+        stop after all commands have been run (the same as if 'setup()'
+        had been called in the usual way
+
+    Returns the Distribution instance, which provides all information
+    used to drive the Distutils.
+    """
+    if stop_after not in ('init', 'config', 'commandline', 'run'):
+        raise ValueError("invalid value for 'stop_after': %r" % (stop_after,))
+
+    global _setup_stop_after, _setup_distribution
+    _setup_stop_after = stop_after
+
+    save_argv = sys.argv.copy()
+    g = {'__file__': script_name}
+    try:
+        try:
+            sys.argv[0] = script_name
+            if script_args is not None:
+                sys.argv[1:] = script_args
+            with open(script_name, 'rb') as f:
+                exec(f.read(), g)
+        finally:
+            sys.argv = save_argv
+            _setup_stop_after = None
+    except SystemExit:
+        # Hmm, should we do something if exiting with a non-zero code
+        # (ie. error)?
+        pass
+
+    if _setup_distribution is None:
+        raise RuntimeError(("'distutils.core.setup()' was never called -- "
+               "perhaps '%s' is not a Distutils setup script?") % \
+              script_name)
+
+    # I wonder if the setup script's namespace -- g and l -- would be of
+    # any interest to callers?
+    #print "_setup_distribution:", _setup_distribution
+    return _setup_distribution
+
+# run_setup ()
diff --git a/Lib/distutils/cygwinccompiler.py b/Lib/distutils/cygwinccompiler.py
new file mode 100644
index 000000000..1c3699034
--- /dev/null
+++ b/Lib/distutils/cygwinccompiler.py
@@ -0,0 +1,405 @@
+"""distutils.cygwinccompiler
+
+Provides the CygwinCCompiler class, a subclass of UnixCCompiler that
+handles the Cygwin port of the GNU C compiler to Windows.  It also contains
+the Mingw32CCompiler class which handles the mingw32 port of GCC (same as
+cygwin in no-cygwin mode).
+"""
+
+# problems:
+#
+# * if you use a msvc compiled python version (1.5.2)
+#   1. you have to insert a __GNUC__ section in its config.h
+#   2. you have to generate an import library for its dll
+#      - create a def-file for python??.dll
+#      - create an import library using
+#             dlltool --dllname python15.dll --def python15.def \
+#                       --output-lib libpython15.a
+#
+#   see also http://starship.python.net/crew/kernr/mingw32/Notes.html
+#
+# * We put export_symbols in a def-file, and don't use
+#   --export-all-symbols because it doesn't worked reliable in some
+#   tested configurations. And because other windows compilers also
+#   need their symbols specified this no serious problem.
+#
+# tested configurations:
+#
+# * cygwin gcc 2.91.57/ld 2.9.4/dllwrap 0.2.4 works
+#   (after patching python's config.h and for C++ some other include files)
+#   see also http://starship.python.net/crew/kernr/mingw32/Notes.html
+# * mingw32 gcc 2.95.2/ld 2.9.4/dllwrap 0.2.4 works
+#   (ld doesn't support -shared, so we use dllwrap)
+# * cygwin gcc 2.95.2/ld 2.10.90/dllwrap 2.10.90 works now
+#   - its dllwrap doesn't work, there is a bug in binutils 2.10.90
+#     see also http://sources.redhat.com/ml/cygwin/2000-06/msg01274.html
+#   - using gcc -mdll instead dllwrap doesn't work without -static because
+#     it tries to link against dlls instead their import libraries. (If
+#     it finds the dll first.)
+#     By specifying -static we force ld to link against the import libraries,
+#     this is windows standard and there are normally not the necessary symbols
+#     in the dlls.
+#   *** only the version of June 2000 shows these problems
+# * cygwin gcc 3.2/ld 2.13.90 works
+#   (ld supports -shared)
+# * mingw gcc 3.2/ld 2.13 works
+#   (ld supports -shared)
+
+import os
+import sys
+import copy
+from subprocess import Popen, PIPE, check_output
+import re
+
+from distutils.ccompiler import gen_preprocess_options, gen_lib_options
+from distutils.unixccompiler import UnixCCompiler
+from distutils.file_util import write_file
+from distutils.errors import (DistutilsExecError, CCompilerError,
+        CompileError, UnknownFileError)
+from distutils import log
+from distutils.version import LooseVersion
+from distutils.spawn import find_executable
+
+def get_msvcr():
+    """Include the appropriate MSVC runtime library if Python was built
+    with MSVC 7.0 or later.
+    """
+    msc_pos = sys.version.find('MSC v.')
+    if msc_pos != -1:
+        msc_ver = sys.version[msc_pos+6:msc_pos+10]
+        if msc_ver == '1300':
+            # MSVC 7.0
+            return ['msvcr70']
+        elif msc_ver == '1310':
+            # MSVC 7.1
+            return ['msvcr71']
+        elif msc_ver == '1400':
+            # VS2005 / MSVC 8.0
+            return ['msvcr80']
+        elif msc_ver == '1500':
+            # VS2008 / MSVC 9.0
+            return ['msvcr90']
+        elif msc_ver == '1600':
+            # VS2010 / MSVC 10.0
+            return ['msvcr100']
+        else:
+            raise ValueError("Unknown MS Compiler version %s " % msc_ver)
+
+
+class CygwinCCompiler(UnixCCompiler):
+    """ Handles the Cygwin port of the GNU C compiler to Windows.
+    """
+    compiler_type = 'cygwin'
+    obj_extension = ".o"
+    static_lib_extension = ".a"
+    shared_lib_extension = ".dll"
+    static_lib_format = "lib%s%s"
+    shared_lib_format = "%s%s"
+    exe_extension = ".exe"
+
+    def __init__(self, verbose=0, dry_run=0, force=0):
+
+        UnixCCompiler.__init__(self, verbose, dry_run, force)
+
+        status, details = check_config_h()
+        self.debug_print("Python's GCC status: %s (details: %s)" %
+                         (status, details))
+        if status is not CONFIG_H_OK:
+            self.warn(
+                "Python's pyconfig.h doesn't seem to support your compiler. "
+                "Reason: %s. "
+                "Compiling may fail because of undefined preprocessor macros."
+                % details)
+
+        self.gcc_version, self.ld_version, self.dllwrap_version = \
+            get_versions()
+        self.debug_print(self.compiler_type + ": gcc %s, ld %s, dllwrap %s\n" %
+                         (self.gcc_version,
+                          self.ld_version,
+                          self.dllwrap_version) )
+
+        # ld_version >= "2.10.90" and < "2.13" should also be able to use
+        # gcc -mdll instead of dllwrap
+        # Older dllwraps had own version numbers, newer ones use the
+        # same as the rest of binutils ( also ld )
+        # dllwrap 2.10.90 is buggy
+        if self.ld_version >= "2.10.90":
+            self.linker_dll = "gcc"
+        else:
+            self.linker_dll = "dllwrap"
+
+        # ld_version >= "2.13" support -shared so use it instead of
+        # -mdll -static
+        if self.ld_version >= "2.13":
+            shared_option = "-shared"
+        else:
+            shared_option = "-mdll -static"
+
+        # Hard-code GCC because that's what this is all about.
+        # XXX optimization, warnings etc. should be customizable.
+        self.set_executables(compiler='gcc -mcygwin -O -Wall',
+                             compiler_so='gcc -mcygwin -mdll -O -Wall',
+                             compiler_cxx='g++ -mcygwin -O -Wall',
+                             linker_exe='gcc -mcygwin',
+                             linker_so=('%s -mcygwin %s' %
+                                        (self.linker_dll, shared_option)))
+
+        # cygwin and mingw32 need different sets of libraries
+        if self.gcc_version == "2.91.57":
+            # cygwin shouldn't need msvcrt, but without the dlls will crash
+            # (gcc version 2.91.57) -- perhaps something about initialization
+            self.dll_libraries=["msvcrt"]
+            self.warn(
+                "Consider upgrading to a newer version of gcc")
+        else:
+            # Include the appropriate MSVC runtime library if Python was built
+            # with MSVC 7.0 or later.
+            self.dll_libraries = get_msvcr()
+
+    def _compile(self, obj, src, ext, cc_args, extra_postargs, pp_opts):
+        """Compiles the source by spawning GCC and windres if needed."""
+        if ext == '.rc' or ext == '.res':
+            # gcc needs '.res' and '.rc' compiled to object files !!!
+            try:
+                self.spawn(["windres", "-i", src, "-o", obj])
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+        else: # for other files use the C-compiler
+            try:
+                self.spawn(self.compiler_so + cc_args + [src, '-o', obj] +
+                           extra_postargs)
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+
+    def link(self, target_desc, objects, output_filename, output_dir=None,
+             libraries=None, library_dirs=None, runtime_library_dirs=None,
+             export_symbols=None, debug=0, extra_preargs=None,
+             extra_postargs=None, build_temp=None, target_lang=None):
+        """Link the objects."""
+        # use separate copies, so we can modify the lists
+        extra_preargs = copy.copy(extra_preargs or [])
+        libraries = copy.copy(libraries or [])
+        objects = copy.copy(objects or [])
+
+        # Additional libraries
+        libraries.extend(self.dll_libraries)
+
+        # handle export symbols by creating a def-file
+        # with executables this only works with gcc/ld as linker
+        if ((export_symbols is not None) and
+            (target_desc != self.EXECUTABLE or self.linker_dll == "gcc")):
+            # (The linker doesn't do anything if output is up-to-date.
+            # So it would probably better to check if we really need this,
+            # but for this we had to insert some unchanged parts of
+            # UnixCCompiler, and this is not what we want.)
+
+            # we want to put some files in the same directory as the
+            # object files are, build_temp doesn't help much
+            # where are the object files
+            temp_dir = os.path.dirname(objects[0])
+            # name of dll to give the helper files the same base name
+            (dll_name, dll_extension) = os.path.splitext(
+                os.path.basename(output_filename))
+
+            # generate the filenames for these files
+            def_file = os.path.join(temp_dir, dll_name + ".def")
+            lib_file = os.path.join(temp_dir, 'lib' + dll_name + ".a")
+
+            # Generate .def file
+            contents = [
+                "LIBRARY %s" % os.path.basename(output_filename),
+                "EXPORTS"]
+            for sym in export_symbols:
+                contents.append(sym)
+            self.execute(write_file, (def_file, contents),
+                         "writing %s" % def_file)
+
+            # next add options for def-file and to creating import libraries
+
+            # dllwrap uses different options than gcc/ld
+            if self.linker_dll == "dllwrap":
+                extra_preargs.extend(["--output-lib", lib_file])
+                # for dllwrap we have to use a special option
+                extra_preargs.extend(["--def", def_file])
+            # we use gcc/ld here and can be sure ld is >= 2.9.10
+            else:
+                # doesn't work: bfd_close build\...\libfoo.a: Invalid operation
+                #extra_preargs.extend(["-Wl,--out-implib,%s" % lib_file])
+                # for gcc/ld the def-file is specified as any object files
+                objects.append(def_file)
+
+        #end: if ((export_symbols is not None) and
+        #        (target_desc != self.EXECUTABLE or self.linker_dll == "gcc")):
+
+        # who wants symbols and a many times larger output file
+        # should explicitly switch the debug mode on
+        # otherwise we let dllwrap/ld strip the output file
+        # (On my machine: 10KB < stripped_file < ??100KB
+        #   unstripped_file = stripped_file + XXX KB
+        #  ( XXX=254 for a typical python extension))
+        if not debug:
+            extra_preargs.append("-s")
+
+        UnixCCompiler.link(self, target_desc, objects, output_filename,
+                           output_dir, libraries, library_dirs,
+                           runtime_library_dirs,
+                           None, # export_symbols, we do this in our def-file
+                           debug, extra_preargs, extra_postargs, build_temp,
+                           target_lang)
+
+    # -- Miscellaneous methods -----------------------------------------
+
+    def object_filenames(self, source_filenames, strip_dir=0, output_dir=''):
+        """Adds supports for rc and res files."""
+        if output_dir is None:
+            output_dir = ''
+        obj_names = []
+        for src_name in source_filenames:
+            # use normcase to make sure '.rc' is really '.rc' and not '.RC'
+            base, ext = os.path.splitext(os.path.normcase(src_name))
+            if ext not in (self.src_extensions + ['.rc','.res']):
+                raise UnknownFileError("unknown file type '%s' (from '%s')" % \
+                      (ext, src_name))
+            if strip_dir:
+                base = os.path.basename (base)
+            if ext in ('.res', '.rc'):
+                # these need to be compiled to object files
+                obj_names.append (os.path.join(output_dir,
+                                              base + ext + self.obj_extension))
+            else:
+                obj_names.append (os.path.join(output_dir,
+                                               base + self.obj_extension))
+        return obj_names
+
+# the same as cygwin plus some additional parameters
+class Mingw32CCompiler(CygwinCCompiler):
+    """ Handles the Mingw32 port of the GNU C compiler to Windows.
+    """
+    compiler_type = 'mingw32'
+
+    def __init__(self, verbose=0, dry_run=0, force=0):
+
+        CygwinCCompiler.__init__ (self, verbose, dry_run, force)
+
+        # ld_version >= "2.13" support -shared so use it instead of
+        # -mdll -static
+        if self.ld_version >= "2.13":
+            shared_option = "-shared"
+        else:
+            shared_option = "-mdll -static"
+
+        # A real mingw32 doesn't need to specify a different entry point,
+        # but cygwin 2.91.57 in no-cygwin-mode needs it.
+        if self.gcc_version <= "2.91.57":
+            entry_point = '--entry _DllMain@12'
+        else:
+            entry_point = ''
+
+        if is_cygwingcc():
+            raise CCompilerError(
+                'Cygwin gcc cannot be used with --compiler=mingw32')
+
+        self.set_executables(compiler='gcc -O -Wall',
+                             compiler_so='gcc -mdll -O -Wall',
+                             compiler_cxx='g++ -O -Wall',
+                             linker_exe='gcc',
+                             linker_so='%s %s %s'
+                                        % (self.linker_dll, shared_option,
+                                           entry_point))
+        # Maybe we should also append -mthreads, but then the finished
+        # dlls need another dll (mingwm10.dll see Mingw32 docs)
+        # (-mthreads: Support thread-safe exception handling on `Mingw32')
+
+        # no additional libraries needed
+        self.dll_libraries=[]
+
+        # Include the appropriate MSVC runtime library if Python was built
+        # with MSVC 7.0 or later.
+        self.dll_libraries = get_msvcr()
+
+# Because these compilers aren't configured in Python's pyconfig.h file by
+# default, we should at least warn the user if he is using an unmodified
+# version.
+
+CONFIG_H_OK = "ok"
+CONFIG_H_NOTOK = "not ok"
+CONFIG_H_UNCERTAIN = "uncertain"
+
+def check_config_h():
+    """Check if the current Python installation appears amenable to building
+    extensions with GCC.
+
+    Returns a tuple (status, details), where 'status' is one of the following
+    constants:
+
+    - CONFIG_H_OK: all is well, go ahead and compile
+    - CONFIG_H_NOTOK: doesn't look good
+    - CONFIG_H_UNCERTAIN: not sure -- unable to read pyconfig.h
+
+    'details' is a human-readable string explaining the situation.
+
+    Note there are two ways to conclude "OK": either 'sys.version' contains
+    the string "GCC" (implying that this Python was built with GCC), or the
+    installed "pyconfig.h" contains the string "__GNUC__".
+    """
+
+    # XXX since this function also checks sys.version, it's not strictly a
+    # "pyconfig.h" check -- should probably be renamed...
+
+    from distutils import sysconfig
+
+    # if sys.version contains GCC then python was compiled with GCC, and the
+    # pyconfig.h file should be OK
+    if "GCC" in sys.version:
+        return CONFIG_H_OK, "sys.version mentions 'GCC'"
+
+    # let's see if __GNUC__ is mentioned in python.h
+    fn = sysconfig.get_config_h_filename()
+    try:
+        config_h = open(fn)
+        try:
+            if "__GNUC__" in config_h.read():
+                return CONFIG_H_OK, "'%s' mentions '__GNUC__'" % fn
+            else:
+                return CONFIG_H_NOTOK, "'%s' does not mention '__GNUC__'" % fn
+        finally:
+            config_h.close()
+    except OSError as exc:
+        return (CONFIG_H_UNCERTAIN,
+                "couldn't read '%s': %s" % (fn, exc.strerror))
+
+RE_VERSION = re.compile(br'(\d+\.\d+(\.\d+)*)')
+
+def _find_exe_version(cmd):
+    """Find the version of an executable by running `cmd` in the shell.
+
+    If the command is not found, or the output does not match
+    `RE_VERSION`, returns None.
+    """
+    executable = cmd.split()[0]
+    if find_executable(executable) is None:
+        return None
+    out = Popen(cmd, shell=True, stdout=PIPE).stdout
+    try:
+        out_string = out.read()
+    finally:
+        out.close()
+    result = RE_VERSION.search(out_string)
+    if result is None:
+        return None
+    # LooseVersion works with strings
+    # so we need to decode our bytes
+    return LooseVersion(result.group(1).decode())
+
+def get_versions():
+    """ Try to find out the versions of gcc, ld and dllwrap.
+
+    If not possible it returns None for it.
+    """
+    commands = ['gcc -dumpversion', 'ld -v', 'dllwrap --version']
+    return tuple([_find_exe_version(cmd) for cmd in commands])
+
+def is_cygwingcc():
+    '''Try to determine if the gcc that would be used is from cygwin.'''
+    out_string = check_output(['gcc', '-dumpmachine'])
+    return out_string.strip().endswith(b'cygwin')
diff --git a/Lib/distutils/debug.py b/Lib/distutils/debug.py
new file mode 100644
index 000000000..daf1660f0
--- /dev/null
+++ b/Lib/distutils/debug.py
@@ -0,0 +1,5 @@
+import os
+
+# If DISTUTILS_DEBUG is anything other than the empty string, we run in
+# debug mode.
+DEBUG = os.environ.get('DISTUTILS_DEBUG')
diff --git a/Lib/distutils/dep_util.py b/Lib/distutils/dep_util.py
new file mode 100644
index 000000000..d74f5e4e9
--- /dev/null
+++ b/Lib/distutils/dep_util.py
@@ -0,0 +1,92 @@
+"""distutils.dep_util
+
+Utility functions for simple, timestamp-based dependency of files
+and groups of files; also, function based entirely on such
+timestamp dependency analysis."""
+
+import os
+from distutils.errors import DistutilsFileError
+
+
+def newer (source, target):
+    """Return true if 'source' exists and is more recently modified than
+    'target', or if 'source' exists and 'target' doesn't.  Return false if
+    both exist and 'target' is the same age or younger than 'source'.
+    Raise DistutilsFileError if 'source' does not exist.
+    """
+    if not os.path.exists(source):
+        raise DistutilsFileError("file '%s' does not exist" %
+                                 os.path.abspath(source))
+    if not os.path.exists(target):
+        return 1
+
+    from stat import ST_MTIME
+    mtime1 = os.stat(source)[ST_MTIME]
+    mtime2 = os.stat(target)[ST_MTIME]
+
+    return mtime1 > mtime2
+
+# newer ()
+
+
+def newer_pairwise (sources, targets):
+    """Walk two filename lists in parallel, testing if each source is newer
+    than its corresponding target.  Return a pair of lists (sources,
+    targets) where source is newer than target, according to the semantics
+    of 'newer()'.
+    """
+    if len(sources) != len(targets):
+        raise ValueError("'sources' and 'targets' must be same length")
+
+    # build a pair of lists (sources, targets) where  source is newer
+    n_sources = []
+    n_targets = []
+    for i in range(len(sources)):
+        if newer(sources[i], targets[i]):
+            n_sources.append(sources[i])
+            n_targets.append(targets[i])
+
+    return (n_sources, n_targets)
+
+# newer_pairwise ()
+
+
+def newer_group (sources, target, missing='error'):
+    """Return true if 'target' is out-of-date with respect to any file
+    listed in 'sources'.  In other words, if 'target' exists and is newer
+    than every file in 'sources', return false; otherwise return true.
+    'missing' controls what we do when a source file is missing; the
+    default ("error") is to blow up with an OSError from inside 'stat()';
+    if it is "ignore", we silently drop any missing source files; if it is
+    "newer", any missing source files make us assume that 'target' is
+    out-of-date (this is handy in "dry-run" mode: it'll make you pretend to
+    carry out commands that wouldn't work because inputs are missing, but
+    that doesn't matter because you're not actually going to run the
+    commands).
+    """
+    # If the target doesn't even exist, then it's definitely out-of-date.
+    if not os.path.exists(target):
+        return 1
+
+    # Otherwise we have to find out the hard way: if *any* source file
+    # is more recent than 'target', then 'target' is out-of-date and
+    # we can immediately return true.  If we fall through to the end
+    # of the loop, then 'target' is up-to-date and we return false.
+    from stat import ST_MTIME
+    target_mtime = os.stat(target)[ST_MTIME]
+    for source in sources:
+        if not os.path.exists(source):
+            if missing == 'error':      # blow up when we stat() the file
+                pass
+            elif missing == 'ignore':   # missing source dropped from
+                continue                #  target's dependency list
+            elif missing == 'newer':    # missing source means target is
+                return 1                #  out-of-date
+
+        source_mtime = os.stat(source)[ST_MTIME]
+        if source_mtime > target_mtime:
+            return 1
+    else:
+        return 0
+
+# newer_group ()
diff --git a/Lib/distutils/dir_util.py b/Lib/distutils/dir_util.py
new file mode 100644
index 000000000..df4d751c9
--- /dev/null
+++ b/Lib/distutils/dir_util.py
@@ -0,0 +1,223 @@
+"""distutils.dir_util
+
+Utility functions for manipulating directories and directory trees."""
+
+import os
+import errno
+from distutils.errors import DistutilsFileError, DistutilsInternalError
+from distutils import log
+
+# cache for by mkpath() -- in addition to cheapening redundant calls,
+# eliminates redundant "creating /foo/bar/baz" messages in dry-run mode
+_path_created = {}
+
+# I don't use os.makedirs because a) it's new to Python 1.5.2, and
+# b) it blows up if the directory already exists (I want to silently
+# succeed in that case).
+def mkpath(name, mode=0o777, verbose=1, dry_run=0):
+    """Create a directory and any missing ancestor directories.
+
+    If the directory already exists (or if 'name' is the empty string, which
+    means the current directory, which of course exists), then do nothing.
+    Raise DistutilsFileError if unable to create some directory along the way
+    (eg. some sub-path exists, but is a file rather than a directory).
+    If 'verbose' is true, print a one-line summary of each mkdir to stdout.
+    Return the list of directories actually created.
+    """
+
+    global _path_created
+
+    # Detect a common bug -- name is None
+    if not isinstance(name, str):
+        raise DistutilsInternalError(
+              "mkpath: 'name' must be a string (got %r)" % (name,))
+
+    # XXX what's the better way to handle verbosity? print as we create
+    # each directory in the path (the current behaviour), or only announce
+    # the creation of the whole path? (quite easy to do the latter since
+    # we're not using a recursive algorithm)
+
+    name = os.path.normpath(name)
+    created_dirs = []
+    if os.path.isdir(name) or name == '':
+        return created_dirs
+    if _path_created.get(os.path.abspath(name)):
+        return created_dirs
+
+    (head, tail) = os.path.split(name)
+    tails = [tail]                      # stack of lone dirs to create
+
+    while head and tail and not os.path.isdir(head):
+        (head, tail) = os.path.split(head)
+        tails.insert(0, tail)          # push next higher dir onto stack
+
+    # now 'head' contains the deepest directory that already exists
+    # (that is, the child of 'head' in 'name' is the highest directory
+    # that does *not* exist)
+    for d in tails:
+        #print "head = %s, d = %s: " % (head, d),
+        head = os.path.join(head, d)
+        abs_head = os.path.abspath(head)
+
+        if _path_created.get(abs_head):
+            continue
+
+        if verbose >= 1:
+            log.info("creating %s", head)
+
+        if not dry_run:
+            try:
+                os.mkdir(head, mode)
+            except OSError as exc:
+                if not (exc.errno == errno.EEXIST and os.path.isdir(head)):
+                    raise DistutilsFileError(
+                          "could not create '%s': %s" % (head, exc.args[-1]))
+            created_dirs.append(head)
+
+        _path_created[abs_head] = 1
+    return created_dirs
+
+def create_tree(base_dir, files, mode=0o777, verbose=1, dry_run=0):
+    """Create all the empty directories under 'base_dir' needed to put 'files'
+    there.
+
+    'base_dir' is just the name of a directory which doesn't necessarily
+    exist yet; 'files' is a list of filenames to be interpreted relative to
+    'base_dir'.  'base_dir' + the directory portion of every file in 'files'
+    will be created if it doesn't already exist.  'mode', 'verbose' and
+    'dry_run' flags are as for 'mkpath()'.
+    """
+    # First get the list of directories to create
+    need_dir = set()
+    for file in files:
+        need_dir.add(os.path.join(base_dir, os.path.dirname(file)))
+
+    # Now create them
+    for dir in sorted(need_dir):
+        mkpath(dir, mode, verbose=verbose, dry_run=dry_run)
+
+import sysconfig
+_multiarch = None
+
+def copy_tree(src, dst, preserve_mode=1, preserve_times=1,
+              preserve_symlinks=0, update=0, verbose=1, dry_run=0):
+    """Copy an entire directory tree 'src' to a new location 'dst'.
+
+    Both 'src' and 'dst' must be directory names.  If 'src' is not a
+    directory, raise DistutilsFileError.  If 'dst' does not exist, it is
+    created with 'mkpath()'.  The end result of the copy is that every
+    file in 'src' is copied to 'dst', and directories under 'src' are
+    recursively copied to 'dst'.  Return the list of files that were
+    copied or might have been copied, using their output name.  The
+    return value is unaffected by 'update' or 'dry_run': it is simply
+    the list of all files under 'src', with the names changed to be
+    under 'dst'.
+
+    'preserve_mode' and 'preserve_times' are the same as for
+    'copy_file'; note that they only apply to regular files, not to
+    directories.  If 'preserve_symlinks' is true, symlinks will be
+    copied as symlinks (on platforms that support them!); otherwise
+    (the default), the destination of the symlink will be copied.
+    'update' and 'verbose' are the same as for 'copy_file'.
+    """
+    from distutils.file_util import copy_file
+
+    if not dry_run and not os.path.isdir(src):
+        raise DistutilsFileError(
+              "cannot copy tree '%s': not a directory" % src)
+    try:
+        names = os.listdir(src)
+    except OSError as e:
+        if dry_run:
+            names = []
+        else:
+            raise DistutilsFileError(
+                  "error listing files in '%s': %s" % (src, e.strerror))
+
+    ext_suffix = sysconfig.get_config_var ('EXT_SUFFIX')
+    _multiarch = sysconfig.get_config_var ('MULTIARCH')
+    if ext_suffix.endswith(_multiarch + ext_suffix[-3:]):
+        new_suffix = None
+    else:
+        new_suffix = "%s-%s%s" % (ext_suffix[:-3], _multiarch, ext_suffix[-3:])
+
+    if not dry_run:
+        mkpath(dst, verbose=verbose)
+
+    outputs = []
+
+    for n in names:
+        src_name = os.path.join(src, n)
+        dst_name = os.path.join(dst, n)
+        if new_suffix and _multiarch and n.endswith(ext_suffix) and not n.endswith(new_suffix):
+            dst_name = os.path.join(dst, n.replace(ext_suffix, new_suffix))
+            log.info("renaming extension %s -> %s", n, n.replace(ext_suffix, new_suffix))
+
+        if n.startswith('.nfs'):
+            # skip NFS rename files
+            continue
+
+        if preserve_symlinks and os.path.islink(src_name):
+            link_dest = os.readlink(src_name)
+            if verbose >= 1:
+                log.info("linking %s -> %s", dst_name, link_dest)
+            if not dry_run:
+                os.symlink(link_dest, dst_name)
+            outputs.append(dst_name)
+
+        elif os.path.isdir(src_name):
+            outputs.extend(
+                copy_tree(src_name, dst_name, preserve_mode,
+                          preserve_times, preserve_symlinks, update,
+                          verbose=verbose, dry_run=dry_run))
+        else:
+            copy_file(src_name, dst_name, preserve_mode,
+                      preserve_times, update, verbose=verbose,
+                      dry_run=dry_run)
+            outputs.append(dst_name)
+
+    return outputs
+
+def _build_cmdtuple(path, cmdtuples):
+    """Helper for remove_tree()."""
+    for f in os.listdir(path):
+        real_f = os.path.join(path,f)
+        if os.path.isdir(real_f) and not os.path.islink(real_f):
+            _build_cmdtuple(real_f, cmdtuples)
+        else:
+            cmdtuples.append((os.remove, real_f))
+    cmdtuples.append((os.rmdir, path))
+
+def remove_tree(directory, verbose=1, dry_run=0):
+    """Recursively remove an entire directory tree.
+
+    Any errors are ignored (apart from being reported to stdout if 'verbose'
+    is true).
+    """
+    global _path_created
+
+    if verbose >= 1:
+        log.info("removing '%s' (and everything under it)", directory)
+    if dry_run:
+        return
+    cmdtuples = []
+    _build_cmdtuple(directory, cmdtuples)
+    for cmd in cmdtuples:
+        try:
+            cmd[0](cmd[1])
+            # remove dir from cache if it's already there
+            abspath = os.path.abspath(cmd[1])
+            if abspath in _path_created:
+                del _path_created[abspath]
+        except OSError as exc:
+            log.warn("error removing %s: %s", directory, exc)
+
+def ensure_relative(path):
+    """Take the full path 'path', and make it a relative path.
+
+    This is useful to make 'path' the second argument to os.path.join().
+    """
+    drive, path = os.path.splitdrive(path)
+    if path[0:1] == os.sep:
+        path = drive + path[1:]
+    return path
diff --git a/Lib/distutils/dist.py b/Lib/distutils/dist.py
new file mode 100644
index 000000000..62a24516c
--- /dev/null
+++ b/Lib/distutils/dist.py
@@ -0,0 +1,1236 @@
+"""distutils.dist
+
+Provides the Distribution class, which represents the module distribution
+being built/installed/distributed.
+"""
+
+import sys
+import os
+import re
+from email import message_from_file
+
+try:
+    import warnings
+except ImportError:
+    warnings = None
+
+from distutils.errors import *
+from distutils.fancy_getopt import FancyGetopt, translate_longopt
+from distutils.util import check_environ, strtobool, rfc822_escape
+from distutils import log
+from distutils.debug import DEBUG
+
+# Regex to define acceptable Distutils command names.  This is not *quite*
+# the same as a Python NAME -- I don't allow leading underscores.  The fact
+# that they're very similar is no coincidence; the default naming scheme is
+# to look for a Python module named after the command.
+command_re = re.compile(r'^[a-zA-Z]([a-zA-Z0-9_]*)$')
+
+
+class Distribution:
+    """The core of the Distutils.  Most of the work hiding behind 'setup'
+    is really done within a Distribution instance, which farms the work out
+    to the Distutils commands specified on the command line.
+
+    Setup scripts will almost never instantiate Distribution directly,
+    unless the 'setup()' function is totally inadequate to their needs.
+    However, it is conceivable that a setup script might wish to subclass
+    Distribution for some specialized purpose, and then pass the subclass
+    to 'setup()' as the 'distclass' keyword argument.  If so, it is
+    necessary to respect the expectations that 'setup' has of Distribution.
+    See the code for 'setup()', in core.py, for details.
+    """
+
+    # 'global_options' describes the command-line options that may be
+    # supplied to the setup script prior to any actual commands.
+    # Eg. "./setup.py -n" or "./setup.py --quiet" both take advantage of
+    # these global options.  This list should be kept to a bare minimum,
+    # since every global option is also valid as a command option -- and we
+    # don't want to pollute the commands with too many options that they
+    # have minimal control over.
+    # The fourth entry for verbose means that it can be repeated.
+    global_options = [
+        ('verbose', 'v', "run verbosely (default)", 1),
+        ('quiet', 'q', "run quietly (turns verbosity off)"),
+        ('dry-run', 'n', "don't actually do anything"),
+        ('help', 'h', "show detailed help message"),
+        ('no-user-cfg', None,
+            'ignore pydistutils.cfg in your home directory'),
+    ]
+
+    # 'common_usage' is a short (2-3 line) string describing the common
+    # usage of the setup script.
+    common_usage = """\
+Common commands: (see '--help-commands' for more)
+
+  setup.py build      will build the package underneath 'build/'
+  setup.py install    will install the package
+"""
+
+    # options that are not propagated to the commands
+    display_options = [
+        ('help-commands', None,
+         "list all available commands"),
+        ('name', None,
+         "print package name"),
+        ('version', 'V',
+         "print package version"),
+        ('fullname', None,
+         "print <package name>-<version>"),
+        ('author', None,
+         "print the author's name"),
+        ('author-email', None,
+         "print the author's email address"),
+        ('maintainer', None,
+         "print the maintainer's name"),
+        ('maintainer-email', None,
+         "print the maintainer's email address"),
+        ('contact', None,
+         "print the maintainer's name if known, else the author's"),
+        ('contact-email', None,
+         "print the maintainer's email address if known, else the author's"),
+        ('url', None,
+         "print the URL for this package"),
+        ('license', None,
+         "print the license of the package"),
+        ('licence', None,
+         "alias for --license"),
+        ('description', None,
+         "print the package description"),
+        ('long-description', None,
+         "print the long package description"),
+        ('platforms', None,
+         "print the list of platforms"),
+        ('classifiers', None,
+         "print the list of classifiers"),
+        ('keywords', None,
+         "print the list of keywords"),
+        ('provides', None,
+         "print the list of packages/modules provided"),
+        ('requires', None,
+         "print the list of packages/modules required"),
+        ('obsoletes', None,
+         "print the list of packages/modules made obsolete")
+        ]
+    display_option_names = [translate_longopt(x[0]) for x in display_options]
+
+    # negative options are options that exclude other options
+    negative_opt = {'quiet': 'verbose'}
+
+    # -- Creation/initialization methods -------------------------------
+
+    def __init__(self, attrs=None):
+        """Construct a new Distribution instance: initialize all the
+        attributes of a Distribution, and then use 'attrs' (a dictionary
+        mapping attribute names to values) to assign some of those
+        attributes their "real" values.  (Any attributes not mentioned in
+        'attrs' will be assigned to some null value: 0, None, an empty list
+        or dictionary, etc.)  Most importantly, initialize the
+        'command_obj' attribute to the empty dictionary; this will be
+        filled in with real command objects by 'parse_command_line()'.
+        """
+
+        # Default values for our command-line options
+        self.verbose = 1
+        self.dry_run = 0
+        self.help = 0
+        for attr in self.display_option_names:
+            setattr(self, attr, 0)
+
+        # Store the distribution meta-data (name, version, author, and so
+        # forth) in a separate object -- we're getting to have enough
+        # information here (and enough command-line options) that it's
+        # worth it.  Also delegate 'get_XXX()' methods to the 'metadata'
+        # object in a sneaky and underhanded (but efficient!) way.
+        self.metadata = DistributionMetadata()
+        for basename in self.metadata._METHOD_BASENAMES:
+            method_name = "get_" + basename
+            setattr(self, method_name, getattr(self.metadata, method_name))
+
+        # 'cmdclass' maps command names to class objects, so we
+        # can 1) quickly figure out which class to instantiate when
+        # we need to create a new command object, and 2) have a way
+        # for the setup script to override command classes
+        self.cmdclass = {}
+
+        # 'command_packages' is a list of packages in which commands
+        # are searched for.  The factory for command 'foo' is expected
+        # to be named 'foo' in the module 'foo' in one of the packages
+        # named here.  This list is searched from the left; an error
+        # is raised if no named package provides the command being
+        # searched for.  (Always access using get_command_packages().)
+        self.command_packages = None
+
+        # 'script_name' and 'script_args' are usually set to sys.argv[0]
+        # and sys.argv[1:], but they can be overridden when the caller is
+        # not necessarily a setup script run from the command-line.
+        self.script_name = None
+        self.script_args = None
+
+        # 'command_options' is where we store command options between
+        # parsing them (from config files, the command-line, etc.) and when
+        # they are actually needed -- ie. when the command in question is
+        # instantiated.  It is a dictionary of dictionaries of 2-tuples:
+        #   command_options = { command_name : { option : (source, value) } }
+        self.command_options = {}
+
+        # 'dist_files' is the list of (command, pyversion, file) that
+        # have been created by any dist commands run so far. This is
+        # filled regardless of whether the run is dry or not. pyversion
+        # gives sysconfig.get_python_version() if the dist file is
+        # specific to a Python version, 'any' if it is good for all
+        # Python versions on the target platform, and '' for a source
+        # file. pyversion should not be used to specify minimum or
+        # maximum required Python versions; use the metainfo for that
+        # instead.
+        self.dist_files = []
+
+        # These options are really the business of various commands, rather
+        # than of the Distribution itself.  We provide aliases for them in
+        # Distribution as a convenience to the developer.
+        self.packages = None
+        self.package_data = {}
+        self.package_dir = None
+        self.py_modules = None
+        self.libraries = None
+        self.headers = None
+        self.ext_modules = None
+        self.ext_package = None
+        self.include_dirs = None
+        self.extra_path = None
+        self.scripts = None
+        self.data_files = None
+        self.password = ''
+
+        # And now initialize bookkeeping stuff that can't be supplied by
+        # the caller at all.  'command_obj' maps command names to
+        # Command instances -- that's how we enforce that every command
+        # class is a singleton.
+        self.command_obj = {}
+
+        # 'have_run' maps command names to boolean values; it keeps track
+        # of whether we have actually run a particular command, to make it
+        # cheap to "run" a command whenever we think we might need to -- if
+        # it's already been done, no need for expensive filesystem
+        # operations, we just check the 'have_run' dictionary and carry on.
+        # It's only safe to query 'have_run' for a command class that has
+        # been instantiated -- a false value will be inserted when the
+        # command object is created, and replaced with a true value when
+        # the command is successfully run.  Thus it's probably best to use
+        # '.get()' rather than a straight lookup.
+        self.have_run = {}
+
+        # Now we'll use the attrs dictionary (ultimately, keyword args from
+        # the setup script) to possibly override any or all of these
+        # distribution options.
+
+        if attrs:
+            # Pull out the set of command options and work on them
+            # specifically.  Note that this order guarantees that aliased
+            # command options will override any supplied redundantly
+            # through the general options dictionary.
+            options = attrs.get('options')
+            if options is not None:
+                del attrs['options']
+                for (command, cmd_options) in options.items():
+                    opt_dict = self.get_option_dict(command)
+                    for (opt, val) in cmd_options.items():
+                        opt_dict[opt] = ("setup script", val)
+
+            if 'licence' in attrs:
+                attrs['license'] = attrs['licence']
+                del attrs['licence']
+                msg = "'licence' distribution option is deprecated; use 'license'"
+                if warnings is not None:
+                    warnings.warn(msg)
+                else:
+                    sys.stderr.write(msg + "\n")
+
+            # Now work on the rest of the attributes.  Any attribute that's
+            # not already defined is invalid!
+            for (key, val) in attrs.items():
+                if hasattr(self.metadata, "set_" + key):
+                    getattr(self.metadata, "set_" + key)(val)
+                elif hasattr(self.metadata, key):
+                    setattr(self.metadata, key, val)
+                elif hasattr(self, key):
+                    setattr(self, key, val)
+                else:
+                    msg = "Unknown distribution option: %s" % repr(key)
+                    if warnings is not None:
+                        warnings.warn(msg)
+                    else:
+                        sys.stderr.write(msg + "\n")
+
+        # no-user-cfg is handled before other command line args
+        # because other args override the config files, and this
+        # one is needed before we can load the config files.
+        # If attrs['script_args'] wasn't passed, assume false.
+        #
+        # This also make sure we just look at the global options
+        self.want_user_cfg = True
+
+        if self.script_args is not None:
+            for arg in self.script_args:
+                if not arg.startswith('-'):
+                    break
+                if arg == '--no-user-cfg':
+                    self.want_user_cfg = False
+                    break
+
+        self.finalize_options()
+
+    def get_option_dict(self, command):
+        """Get the option dictionary for a given command.  If that
+        command's option dictionary hasn't been created yet, then create it
+        and return the new dictionary; otherwise, return the existing
+        option dictionary.
+        """
+        dict = self.command_options.get(command)
+        if dict is None:
+            dict = self.command_options[command] = {}
+        return dict
+
+    def dump_option_dicts(self, header=None, commands=None, indent=""):
+        from pprint import pformat
+
+        if commands is None:             # dump all command option dicts
+            commands = sorted(self.command_options.keys())
+
+        if header is not None:
+            self.announce(indent + header)
+            indent = indent + "  "
+
+        if not commands:
+            self.announce(indent + "no commands known yet")
+            return
+
+        for cmd_name in commands:
+            opt_dict = self.command_options.get(cmd_name)
+            if opt_dict is None:
+                self.announce(indent +
+                              "no option dict for '%s' command" % cmd_name)
+            else:
+                self.announce(indent +
+                              "option dict for '%s' command:" % cmd_name)
+                out = pformat(opt_dict)
+                for line in out.split('\n'):
+                    self.announce(indent + "  " + line)
+
+    # -- Config file finding/parsing methods ---------------------------
+
+    def find_config_files(self):
+        """Find as many configuration files as should be processed for this
+        platform, and return a list of filenames in the order in which they
+        should be parsed.  The filenames returned are guaranteed to exist
+        (modulo nasty race conditions).
+
+        There are three possible config files: distutils.cfg in the
+        Distutils installation directory (ie. where the top-level
+        Distutils __inst__.py file lives), a file in the user's home
+        directory named .pydistutils.cfg on Unix and pydistutils.cfg
+        on Windows/Mac; and setup.cfg in the current directory.
+
+        The file in the user's home directory can be disabled with the
+        --no-user-cfg option.
+        """
+        files = []
+        check_environ()
+
+        # Where to look for the system-wide Distutils config file
+        sys_dir = os.path.dirname(sys.modules['distutils'].__file__)
+
+        # Look for the system config file
+        sys_file = os.path.join(sys_dir, "distutils.cfg")
+        if os.path.isfile(sys_file):
+            files.append(sys_file)
+
+        # What to call the per-user config file
+        if os.name == 'posix':
+            user_filename = ".pydistutils.cfg"
+        else:
+            user_filename = "pydistutils.cfg"
+
+        # And look for the user config file
+        if self.want_user_cfg:
+            user_file = os.path.join(os.path.expanduser('~'), user_filename)
+            if os.path.isfile(user_file):
+                files.append(user_file)
+
+        # All platforms support local setup.cfg
+        local_file = "setup.cfg"
+        if os.path.isfile(local_file):
+            files.append(local_file)
+
+        if DEBUG:
+            self.announce("using config files: %s" % ', '.join(files))
+
+        return files
+
+    def parse_config_files(self, filenames=None):
+        from configparser import ConfigParser
+
+        # Ignore install directory options if we have a venv
+        if sys.prefix != sys.base_prefix:
+            ignore_options = [
+                'install-base', 'install-platbase', 'install-lib',
+                'install-platlib', 'install-purelib', 'install-headers',
+                'install-scripts', 'install-data', 'prefix', 'exec-prefix',
+                'home', 'user', 'root']
+        else:
+            ignore_options = []
+
+        ignore_options = frozenset(ignore_options)
+
+        if filenames is None:
+            filenames = self.find_config_files()
+
+        if DEBUG:
+            self.announce("Distribution.parse_config_files():")
+
+        parser = ConfigParser()
+        for filename in filenames:
+            if DEBUG:
+                self.announce("  reading %s" % filename)
+            parser.read(filename)
+            for section in parser.sections():
+                options = parser.options(section)
+                opt_dict = self.get_option_dict(section)
+
+                for opt in options:
+                    if opt != '__name__' and opt not in ignore_options:
+                        val = parser.get(section,opt)
+                        opt = opt.replace('-', '_')
+                        opt_dict[opt] = (filename, val)
+
+            # Make the ConfigParser forget everything (so we retain
+            # the original filenames that options come from)
+            parser.__init__()
+
+        # If there was a "global" section in the config file, use it
+        # to set Distribution options.
+
+        if 'global' in self.command_options:
+            for (opt, (src, val)) in self.command_options['global'].items():
+                alias = self.negative_opt.get(opt)
+                try:
+                    if alias:
+                        setattr(self, alias, not strtobool(val))
+                    elif opt in ('verbose', 'dry_run'): # ugh!
+                        setattr(self, opt, strtobool(val))
+                    else:
+                        setattr(self, opt, val)
+                except ValueError as msg:
+                    raise DistutilsOptionError(msg)
+
+    # -- Command-line parsing methods ----------------------------------
+
+    def parse_command_line(self):
+        """Parse the setup script's command line, taken from the
+        'script_args' instance attribute (which defaults to 'sys.argv[1:]'
+        -- see 'setup()' in core.py).  This list is first processed for
+        "global options" -- options that set attributes of the Distribution
+        instance.  Then, it is alternately scanned for Distutils commands
+        and options for that command.  Each new command terminates the
+        options for the previous command.  The allowed options for a
+        command are determined by the 'user_options' attribute of the
+        command class -- thus, we have to be able to load command classes
+        in order to parse the command line.  Any error in that 'options'
+        attribute raises DistutilsGetoptError; any error on the
+        command-line raises DistutilsArgError.  If no Distutils commands
+        were found on the command line, raises DistutilsArgError.  Return
+        true if command-line was successfully parsed and we should carry
+        on with executing commands; false if no errors but we shouldn't
+        execute commands (currently, this only happens if user asks for
+        help).
+        """
+        #
+        # We now have enough information to show the Macintosh dialog
+        # that allows the user to interactively specify the "command line".
+        #
+        toplevel_options = self._get_toplevel_options()
+
+        # We have to parse the command line a bit at a time -- global
+        # options, then the first command, then its options, and so on --
+        # because each command will be handled by a different class, and
+        # the options that are valid for a particular class aren't known
+        # until we have loaded the command class, which doesn't happen
+        # until we know what the command is.
+
+        self.commands = []
+        parser = FancyGetopt(toplevel_options + self.display_options)
+        parser.set_negative_aliases(self.negative_opt)
+        parser.set_aliases({'licence': 'license'})
+        args = parser.getopt(args=self.script_args, object=self)
+        option_order = parser.get_option_order()
+        log.set_verbosity(self.verbose)
+
+        # for display options we return immediately
+        if self.handle_display_options(option_order):
+            return
+        while args:
+            args = self._parse_command_opts(parser, args)
+            if args is None:            # user asked for help (and got it)
+                return
+
+        # Handle the cases of --help as a "global" option, ie.
+        # "setup.py --help" and "setup.py --help command ...".  For the
+        # former, we show global options (--verbose, --dry-run, etc.)
+        # and display-only options (--name, --version, etc.); for the
+        # latter, we omit the display-only options and show help for
+        # each command listed on the command line.
+        if self.help:
+            self._show_help(parser,
+                            display_options=len(self.commands) == 0,
+                            commands=self.commands)
+            return
+
+        # Oops, no commands found -- an end-user error
+        if not self.commands:
+            raise DistutilsArgError("no commands supplied")
+
+        # All is well: return true
+        return True
+
+    def _get_toplevel_options(self):
+        """Return the non-display options recognized at the top level.
+
+        This includes options that are recognized *only* at the top
+        level as well as options recognized for commands.
+        """
+        return self.global_options + [
+            ("command-packages=", None,
+             "list of packages that provide distutils commands"),
+            ]
+
+    def _parse_command_opts(self, parser, args):
+        """Parse the command-line options for a single command.
+        'parser' must be a FancyGetopt instance; 'args' must be the list
+        of arguments, starting with the current command (whose options
+        we are about to parse).  Returns a new version of 'args' with
+        the next command at the front of the list; will be the empty
+        list if there are no more commands on the command line.  Returns
+        None if the user asked for help on this command.
+        """
+        # late import because of mutual dependence between these modules
+        from distutils.cmd import Command
+
+        # Pull the current command from the head of the command line
+        command = args[0]
+        if not command_re.match(command):
+            raise SystemExit("invalid command name '%s'" % command)
+        self.commands.append(command)
+
+        # Dig up the command class that implements this command, so we
+        # 1) know that it's a valid command, and 2) know which options
+        # it takes.
+        try:
+            cmd_class = self.get_command_class(command)
+        except DistutilsModuleError as msg:
+            raise DistutilsArgError(msg)
+
+        # Require that the command class be derived from Command -- want
+        # to be sure that the basic "command" interface is implemented.
+        if not issubclass(cmd_class, Command):
+            raise DistutilsClassError(
+                "command class %s must subclass Command" % cmd_class)
+
+        # Also make sure that the command object provides a list of its
+        # known options.
+        if not (hasattr(cmd_class, 'user_options') and
+                isinstance(cmd_class.user_options, list)):
+            msg = ("command class %s must provide "
+                "'user_options' attribute (a list of tuples)")
+            raise DistutilsClassError(msg % cmd_class)
+
+        # If the command class has a list of negative alias options,
+        # merge it in with the global negative aliases.
+        negative_opt = self.negative_opt
+        if hasattr(cmd_class, 'negative_opt'):
+            negative_opt = negative_opt.copy()
+            negative_opt.update(cmd_class.negative_opt)
+
+        # Check for help_options in command class.  They have a different
+        # format (tuple of four) so we need to preprocess them here.
+        if (hasattr(cmd_class, 'help_options') and
+                isinstance(cmd_class.help_options, list)):
+            help_options = fix_help_options(cmd_class.help_options)
+        else:
+            help_options = []
+
+        # All commands support the global options too, just by adding
+        # in 'global_options'.
+        parser.set_option_table(self.global_options +
+                                cmd_class.user_options +
+                                help_options)
+        parser.set_negative_aliases(negative_opt)
+        (args, opts) = parser.getopt(args[1:])
+        if hasattr(opts, 'help') and opts.help:
+            self._show_help(parser, display_options=0, commands=[cmd_class])
+            return
+
+        if (hasattr(cmd_class, 'help_options') and
+                isinstance(cmd_class.help_options, list)):
+            help_option_found=0
+            for (help_option, short, desc, func) in cmd_class.help_options:
+                if hasattr(opts, parser.get_attr_name(help_option)):
+                    help_option_found=1
+                    if callable(func):
+                        func()
+                    else:
+                        raise DistutilsClassError(
+                            "invalid help function %r for help option '%s': "
+                            "must be a callable object (function, etc.)"
+                            % (func, help_option))
+
+            if help_option_found:
+                return
+
+        # Put the options from the command-line into their official
+        # holding pen, the 'command_options' dictionary.
+        opt_dict = self.get_option_dict(command)
+        for (name, value) in vars(opts).items():
+            opt_dict[name] = ("command line", value)
+
+        return args
+
+    def finalize_options(self):
+        """Set final values for all the options on the Distribution
+        instance, analogous to the .finalize_options() method of Command
+        objects.
+        """
+        for attr in ('keywords', 'platforms'):
+            value = getattr(self.metadata, attr)
+            if value is None:
+                continue
+            if isinstance(value, str):
+                value = [elm.strip() for elm in value.split(',')]
+                setattr(self.metadata, attr, value)
+
+    def _show_help(self, parser, global_options=1, display_options=1,
+                   commands=[]):
+        """Show help for the setup script command-line in the form of
+        several lists of command-line options.  'parser' should be a
+        FancyGetopt instance; do not expect it to be returned in the
+        same state, as its option table will be reset to make it
+        generate the correct help text.
+
+        If 'global_options' is true, lists the global options:
+        --verbose, --dry-run, etc.  If 'display_options' is true, lists
+        the "display-only" options: --name, --version, etc.  Finally,
+        lists per-command help for every command name or command class
+        in 'commands'.
+        """
+        # late import because of mutual dependence between these modules
+        from distutils.core import gen_usage
+        from distutils.cmd import Command
+
+        if global_options:
+            if display_options:
+                options = self._get_toplevel_options()
+            else:
+                options = self.global_options
+            parser.set_option_table(options)
+            parser.print_help(self.common_usage + "\nGlobal options:")
+            print('')
+
+        if display_options:
+            parser.set_option_table(self.display_options)
+            parser.print_help(
+                "Information display options (just display " +
+                "information, ignore any commands)")
+            print('')
+
+        for command in self.commands:
+            if isinstance(command, type) and issubclass(command, Command):
+                klass = command
+            else:
+                klass = self.get_command_class(command)
+            if (hasattr(klass, 'help_options') and
+                    isinstance(klass.help_options, list)):
+                parser.set_option_table(klass.user_options +
+                                        fix_help_options(klass.help_options))
+            else:
+                parser.set_option_table(klass.user_options)
+            parser.print_help("Options for '%s' command:" % klass.__name__)
+            print('')
+
+        print(gen_usage(self.script_name))
+
+    def handle_display_options(self, option_order):
+        """If there were any non-global "display-only" options
+        (--help-commands or the metadata display options) on the command
+        line, display the requested info and return true; else return
+        false.
+        """
+        from distutils.core import gen_usage
+
+        # User just wants a list of commands -- we'll print it out and stop
+        # processing now (ie. if they ran "setup --help-commands foo bar",
+        # we ignore "foo bar").
+        if self.help_commands:
+            self.print_commands()
+            print('')
+            print(gen_usage(self.script_name))
+            return 1
+
+        # If user supplied any of the "display metadata" options, then
+        # display that metadata in the order in which the user supplied the
+        # metadata options.
+        any_display_options = 0
+        is_display_option = {}
+        for option in self.display_options:
+            is_display_option[option[0]] = 1
+
+        for (opt, val) in option_order:
+            if val and is_display_option.get(opt):
+                opt = translate_longopt(opt)
+                value = getattr(self.metadata, "get_"+opt)()
+                if opt in ['keywords', 'platforms']:
+                    print(','.join(value))
+                elif opt in ('classifiers', 'provides', 'requires',
+                             'obsoletes'):
+                    print('\n'.join(value))
+                else:
+                    print(value)
+                any_display_options = 1
+
+        return any_display_options
+
+    def print_command_list(self, commands, header, max_length):
+        """Print a subset of the list of all commands -- used by
+        'print_commands()'.
+        """
+        print(header + ":")
+
+        for cmd in commands:
+            klass = self.cmdclass.get(cmd)
+            if not klass:
+                klass = self.get_command_class(cmd)
+            try:
+                description = klass.description
+            except AttributeError:
+                description = "(no description available)"
+
+            print("  %-*s  %s" % (max_length, cmd, description))
+
+    def print_commands(self):
+        """Print out a help message listing all available commands with a
+        description of each.  The list is divided into "standard commands"
+        (listed in distutils.command.__all__) and "extra commands"
+        (mentioned in self.cmdclass, but not a standard command).  The
+        descriptions come from the command class attribute
+        'description'.
+        """
+        import distutils.command
+        std_commands = distutils.command.__all__
+        is_std = {}
+        for cmd in std_commands:
+            is_std[cmd] = 1
+
+        extra_commands = []
+        for cmd in self.cmdclass.keys():
+            if not is_std.get(cmd):
+                extra_commands.append(cmd)
+
+        max_length = 0
+        for cmd in (std_commands + extra_commands):
+            if len(cmd) > max_length:
+                max_length = len(cmd)
+
+        self.print_command_list(std_commands,
+                                "Standard commands",
+                                max_length)
+        if extra_commands:
+            print()
+            self.print_command_list(extra_commands,
+                                    "Extra commands",
+                                    max_length)
+
+    def get_command_list(self):
+        """Get a list of (command, description) tuples.
+        The list is divided into "standard commands" (listed in
+        distutils.command.__all__) and "extra commands" (mentioned in
+        self.cmdclass, but not a standard command).  The descriptions come
+        from the command class attribute 'description'.
+        """
+        # Currently this is only used on Mac OS, for the Mac-only GUI
+        # Distutils interface (by Jack Jansen)
+        import distutils.command
+        std_commands = distutils.command.__all__
+        is_std = {}
+        for cmd in std_commands:
+            is_std[cmd] = 1
+
+        extra_commands = []
+        for cmd in self.cmdclass.keys():
+            if not is_std.get(cmd):
+                extra_commands.append(cmd)
+
+        rv = []
+        for cmd in (std_commands + extra_commands):
+            klass = self.cmdclass.get(cmd)
+            if not klass:
+                klass = self.get_command_class(cmd)
+            try:
+                description = klass.description
+            except AttributeError:
+                description = "(no description available)"
+            rv.append((cmd, description))
+        return rv
+
+    # -- Command class/object methods ----------------------------------
+
+    def get_command_packages(self):
+        """Return a list of packages from which commands are loaded."""
+        pkgs = self.command_packages
+        if not isinstance(pkgs, list):
+            if pkgs is None:
+                pkgs = ''
+            pkgs = [pkg.strip() for pkg in pkgs.split(',') if pkg != '']
+            if "distutils.command" not in pkgs:
+                pkgs.insert(0, "distutils.command")
+            self.command_packages = pkgs
+        return pkgs
+
+    def get_command_class(self, command):
+        """Return the class that implements the Distutils command named by
+        'command'.  First we check the 'cmdclass' dictionary; if the
+        command is mentioned there, we fetch the class object from the
+        dictionary and return it.  Otherwise we load the command module
+        ("distutils.command." + command) and fetch the command class from
+        the module.  The loaded class is also stored in 'cmdclass'
+        to speed future calls to 'get_command_class()'.
+
+        Raises DistutilsModuleError if the expected module could not be
+        found, or if that module does not define the expected class.
+        """
+        klass = self.cmdclass.get(command)
+        if klass:
+            return klass
+
+        for pkgname in self.get_command_packages():
+            module_name = "%s.%s" % (pkgname, command)
+            klass_name = command
+
+            try:
+                __import__(module_name)
+                module = sys.modules[module_name]
+            except ImportError:
+                continue
+
+            try:
+                klass = getattr(module, klass_name)
+            except AttributeError:
+                raise DistutilsModuleError(
+                    "invalid command '%s' (no class '%s' in module '%s')"
+                    % (command, klass_name, module_name))
+
+            self.cmdclass[command] = klass
+            return klass
+
+        raise DistutilsModuleError("invalid command '%s'" % command)
+
+    def get_command_obj(self, command, create=1):
+        """Return the command object for 'command'.  Normally this object
+        is cached on a previous call to 'get_command_obj()'; if no command
+        object for 'command' is in the cache, then we either create and
+        return it (if 'create' is true) or return None.
+        """
+        cmd_obj = self.command_obj.get(command)
+        if not cmd_obj and create:
+            if DEBUG:
+                self.announce("Distribution.get_command_obj(): "
+                              "creating '%s' command object" % command)
+
+            klass = self.get_command_class(command)
+            cmd_obj = self.command_obj[command] = klass(self)
+            self.have_run[command] = 0
+
+            # Set any options that were supplied in config files
+            # or on the command line.  (NB. support for error
+            # reporting is lame here: any errors aren't reported
+            # until 'finalize_options()' is called, which means
+            # we won't report the source of the error.)
+            options = self.command_options.get(command)
+            if options:
+                self._set_command_options(cmd_obj, options)
+
+        return cmd_obj
+
+    def _set_command_options(self, command_obj, option_dict=None):
+        """Set the options for 'command_obj' from 'option_dict'.  Basically
+        this means copying elements of a dictionary ('option_dict') to
+        attributes of an instance ('command').
+
+        'command_obj' must be a Command instance.  If 'option_dict' is not
+        supplied, uses the standard option dictionary for this command
+        (from 'self.command_options').
+        """
+        command_name = command_obj.get_command_name()
+        if option_dict is None:
+            option_dict = self.get_option_dict(command_name)
+
+        if DEBUG:
+            self.announce("  setting options for '%s' command:" % command_name)
+        for (option, (source, value)) in option_dict.items():
+            if DEBUG:
+                self.announce("    %s = %s (from %s)" % (option, value,
+                                                         source))
+            try:
+                bool_opts = [translate_longopt(o)
+                             for o in command_obj.boolean_options]
+            except AttributeError:
+                bool_opts = []
+            try:
+                neg_opt = command_obj.negative_opt
+            except AttributeError:
+                neg_opt = {}
+
+            try:
+                is_string = isinstance(value, str)
+                if option in neg_opt and is_string:
+                    setattr(command_obj, neg_opt[option], not strtobool(value))
+                elif option in bool_opts and is_string:
+                    setattr(command_obj, option, strtobool(value))
+                elif hasattr(command_obj, option):
+                    setattr(command_obj, option, value)
+                else:
+                    raise DistutilsOptionError(
+                        "error in %s: command '%s' has no such option '%s'"
+                        % (source, command_name, option))
+            except ValueError as msg:
+                raise DistutilsOptionError(msg)
+
+    def reinitialize_command(self, command, reinit_subcommands=0):
+        """Reinitializes a command to the state it was in when first
+        returned by 'get_command_obj()': ie., initialized but not yet
+        finalized.  This provides the opportunity to sneak option
+        values in programmatically, overriding or supplementing
+        user-supplied values from the config files and command line.
+        You'll have to re-finalize the command object (by calling
+        'finalize_options()' or 'ensure_finalized()') before using it for
+        real.
+
+        'command' should be a command name (string) or command object.  If
+        'reinit_subcommands' is true, also reinitializes the command's
+        sub-commands, as declared by the 'sub_commands' class attribute (if
+        it has one).  See the "install" command for an example.  Only
+        reinitializes the sub-commands that actually matter, ie. those
+        whose test predicates return true.
+
+        Returns the reinitialized command object.
+        """
+        from distutils.cmd import Command
+        if not isinstance(command, Command):
+            command_name = command
+            command = self.get_command_obj(command_name)
+        else:
+            command_name = command.get_command_name()
+
+        if not command.finalized:
+            return command
+        command.initialize_options()
+        command.finalized = 0
+        self.have_run[command_name] = 0
+        self._set_command_options(command)
+
+        if reinit_subcommands:
+            for sub in command.get_sub_commands():
+                self.reinitialize_command(sub, reinit_subcommands)
+
+        return command
+
+    # -- Methods that operate on the Distribution ----------------------
+
+    def announce(self, msg, level=log.INFO):
+        log.log(level, msg)
+
+    def run_commands(self):
+        """Run each command that was seen on the setup script command line.
+        Uses the list of commands found and cache of command objects
+        created by 'get_command_obj()'.
+        """
+        for cmd in self.commands:
+            self.run_command(cmd)
+
+    # -- Methods that operate on its Commands --------------------------
+
+    def run_command(self, command):
+        """Do whatever it takes to run a command (including nothing at all,
+        if the command has already been run).  Specifically: if we have
+        already created and run the command named by 'command', return
+        silently without doing anything.  If the command named by 'command'
+        doesn't even have a command object yet, create one.  Then invoke
+        'run()' on that command object (or an existing one).
+        """
+        # Already been here, done that? then return silently.
+        if self.have_run.get(command):
+            return
+
+        log.info("running %s", command)
+        cmd_obj = self.get_command_obj(command)
+        cmd_obj.ensure_finalized()
+        cmd_obj.run()
+        self.have_run[command] = 1
+
+    # -- Distribution query methods ------------------------------------
+
+    def has_pure_modules(self):
+        return len(self.packages or self.py_modules or []) > 0
+
+    def has_ext_modules(self):
+        return self.ext_modules and len(self.ext_modules) > 0
+
+    def has_c_libraries(self):
+        return self.libraries and len(self.libraries) > 0
+
+    def has_modules(self):
+        return self.has_pure_modules() or self.has_ext_modules()
+
+    def has_headers(self):
+        return self.headers and len(self.headers) > 0
+
+    def has_scripts(self):
+        return self.scripts and len(self.scripts) > 0
+
+    def has_data_files(self):
+        return self.data_files and len(self.data_files) > 0
+
+    def is_pure(self):
+        return (self.has_pure_modules() and
+                not self.has_ext_modules() and
+                not self.has_c_libraries())
+
+    # -- Metadata query methods ----------------------------------------
+
+    # If you're looking for 'get_name()', 'get_version()', and so forth,
+    # they are defined in a sneaky way: the constructor binds self.get_XXX
+    # to self.metadata.get_XXX.  The actual code is in the
+    # DistributionMetadata class, below.
+
+class DistributionMetadata:
+    """Dummy class to hold the distribution meta-data: name, version,
+    author, and so forth.
+    """
+
+    _METHOD_BASENAMES = ("name", "version", "author", "author_email",
+                         "maintainer", "maintainer_email", "url",
+                         "license", "description", "long_description",
+                         "keywords", "platforms", "fullname", "contact",
+                         "contact_email", "classifiers", "download_url",
+                         # PEP 314
+                         "provides", "requires", "obsoletes",
+                         )
+
+    def __init__(self, path=None):
+        if path is not None:
+            self.read_pkg_file(open(path))
+        else:
+            self.name = None
+            self.version = None
+            self.author = None
+            self.author_email = None
+            self.maintainer = None
+            self.maintainer_email = None
+            self.url = None
+            self.license = None
+            self.description = None
+            self.long_description = None
+            self.keywords = None
+            self.platforms = None
+            self.classifiers = None
+            self.download_url = None
+            # PEP 314
+            self.provides = None
+            self.requires = None
+            self.obsoletes = None
+
+    def read_pkg_file(self, file):
+        """Reads the metadata values from a file object."""
+        msg = message_from_file(file)
+
+        def _read_field(name):
+            value = msg[name]
+            if value == 'UNKNOWN':
+                return None
+            return value
+
+        def _read_list(name):
+            values = msg.get_all(name, None)
+            if values == []:
+                return None
+            return values
+
+        metadata_version = msg['metadata-version']
+        self.name = _read_field('name')
+        self.version = _read_field('version')
+        self.description = _read_field('summary')
+        # we are filling author only.
+        self.author = _read_field('author')
+        self.maintainer = None
+        self.author_email = _read_field('author-email')
+        self.maintainer_email = None
+        self.url = _read_field('home-page')
+        self.license = _read_field('license')
+
+        if 'download-url' in msg:
+            self.download_url = _read_field('download-url')
+        else:
+            self.download_url = None
+
+        self.long_description = _read_field('description')
+        self.description = _read_field('summary')
+
+        if 'keywords' in msg:
+            self.keywords = _read_field('keywords').split(',')
+
+        self.platforms = _read_list('platform')
+        self.classifiers = _read_list('classifier')
+
+        # PEP 314 - these fields only exist in 1.1
+        if metadata_version == '1.1':
+            self.requires = _read_list('requires')
+            self.provides = _read_list('provides')
+            self.obsoletes = _read_list('obsoletes')
+        else:
+            self.requires = None
+            self.provides = None
+            self.obsoletes = None
+
+    def write_pkg_info(self, base_dir):
+        """Write the PKG-INFO file into the release tree.
+        """
+        with open(os.path.join(base_dir, 'PKG-INFO'), 'w',
+                  encoding='UTF-8') as pkg_info:
+            self.write_pkg_file(pkg_info)
+
+    def write_pkg_file(self, file):
+        """Write the PKG-INFO format data to a file object.
+        """
+        version = '1.0'
+        if (self.provides or self.requires or self.obsoletes or
+                self.classifiers or self.download_url):
+            version = '1.1'
+
+        file.write('Metadata-Version: %s\n' % version)
+        file.write('Name: %s\n' % self.get_name())
+        file.write('Version: %s\n' % self.get_version())
+        file.write('Summary: %s\n' % self.get_description())
+        file.write('Home-page: %s\n' % self.get_url())
+        file.write('Author: %s\n' % self.get_contact())
+        file.write('Author-email: %s\n' % self.get_contact_email())
+        file.write('License: %s\n' % self.get_license())
+        if self.download_url:
+            file.write('Download-URL: %s\n' % self.download_url)
+
+        long_desc = rfc822_escape(self.get_long_description())
+        file.write('Description: %s\n' % long_desc)
+
+        keywords = ','.join(self.get_keywords())
+        if keywords:
+            file.write('Keywords: %s\n' % keywords)
+
+        self._write_list(file, 'Platform', self.get_platforms())
+        self._write_list(file, 'Classifier', self.get_classifiers())
+
+        # PEP 314
+        self._write_list(file, 'Requires', self.get_requires())
+        self._write_list(file, 'Provides', self.get_provides())
+        self._write_list(file, 'Obsoletes', self.get_obsoletes())
+
+    def _write_list(self, file, name, values):
+        for value in values:
+            file.write('%s: %s\n' % (name, value))
+
+    # -- Metadata query methods ----------------------------------------
+
+    def get_name(self):
+        return self.name or "UNKNOWN"
+
+    def get_version(self):
+        return self.version or "0.0.0"
+
+    def get_fullname(self):
+        return "%s-%s" % (self.get_name(), self.get_version())
+
+    def get_author(self):
+        return self.author or "UNKNOWN"
+
+    def get_author_email(self):
+        return self.author_email or "UNKNOWN"
+
+    def get_maintainer(self):
+        return self.maintainer or "UNKNOWN"
+
+    def get_maintainer_email(self):
+        return self.maintainer_email or "UNKNOWN"
+
+    def get_contact(self):
+        return self.maintainer or self.author or "UNKNOWN"
+
+    def get_contact_email(self):
+        return self.maintainer_email or self.author_email or "UNKNOWN"
+
+    def get_url(self):
+        return self.url or "UNKNOWN"
+
+    def get_license(self):
+        return self.license or "UNKNOWN"
+    get_licence = get_license
+
+    def get_description(self):
+        return self.description or "UNKNOWN"
+
+    def get_long_description(self):
+        return self.long_description or "UNKNOWN"
+
+    def get_keywords(self):
+        return self.keywords or []
+
+    def get_platforms(self):
+        return self.platforms or ["UNKNOWN"]
+
+    def get_classifiers(self):
+        return self.classifiers or []
+
+    def get_download_url(self):
+        return self.download_url or "UNKNOWN"
+
+    # PEP 314
+    def get_requires(self):
+        return self.requires or []
+
+    def set_requires(self, value):
+        import distutils.versionpredicate
+        for v in value:
+            distutils.versionpredicate.VersionPredicate(v)
+        self.requires = value
+
+    def get_provides(self):
+        return self.provides or []
+
+    def set_provides(self, value):
+        value = [v.strip() for v in value]
+        for v in value:
+            import distutils.versionpredicate
+            distutils.versionpredicate.split_provision(v)
+        self.provides = value
+
+    def get_obsoletes(self):
+        return self.obsoletes or []
+
+    def set_obsoletes(self, value):
+        import distutils.versionpredicate
+        for v in value:
+            distutils.versionpredicate.VersionPredicate(v)
+        self.obsoletes = value
+
+def fix_help_options(options):
+    """Convert a 4-tuple 'help_options' list as found in various command
+    classes to the 3-tuple form required by FancyGetopt.
+    """
+    new_options = []
+    for help_tuple in options:
+        new_options.append(help_tuple[0:3])
+    return new_options
diff --git a/Lib/distutils/errors.py b/Lib/distutils/errors.py
new file mode 100644
index 000000000..8b93059e1
--- /dev/null
+++ b/Lib/distutils/errors.py
@@ -0,0 +1,97 @@
+"""distutils.errors
+
+Provides exceptions used by the Distutils modules.  Note that Distutils
+modules may raise standard exceptions; in particular, SystemExit is
+usually raised for errors that are obviously the end-user's fault
+(eg. bad command-line arguments).
+
+This module is safe to use in "from ... import *" mode; it only exports
+symbols whose names start with "Distutils" and end with "Error"."""
+
+class DistutilsError (Exception):
+    """The root of all Distutils evil."""
+    pass
+
+class DistutilsModuleError (DistutilsError):
+    """Unable to load an expected module, or to find an expected class
+    within some module (in particular, command modules and classes)."""
+    pass
+
+class DistutilsClassError (DistutilsError):
+    """Some command class (or possibly distribution class, if anyone
+    feels a need to subclass Distribution) is found not to be holding
+    up its end of the bargain, ie. implementing some part of the
+    "command "interface."""
+    pass
+
+class DistutilsGetoptError (DistutilsError):
+    """The option table provided to 'fancy_getopt()' is bogus."""
+    pass
+
+class DistutilsArgError (DistutilsError):
+    """Raised by fancy_getopt in response to getopt.error -- ie. an
+    error in the command line usage."""
+    pass
+
+class DistutilsFileError (DistutilsError):
+    """Any problems in the filesystem: expected file not found, etc.
+    Typically this is for problems that we detect before OSError
+    could be raised."""
+    pass
+
+class DistutilsOptionError (DistutilsError):
+    """Syntactic/semantic errors in command options, such as use of
+    mutually conflicting options, or inconsistent options,
+    badly-spelled values, etc.  No distinction is made between option
+    values originating in the setup script, the command line, config
+    files, or what-have-you -- but if we *know* something originated in
+    the setup script, we'll raise DistutilsSetupError instead."""
+    pass
+
+class DistutilsSetupError (DistutilsError):
+    """For errors that can be definitely blamed on the setup script,
+    such as invalid keyword arguments to 'setup()'."""
+    pass
+
+class DistutilsPlatformError (DistutilsError):
+    """We don't know how to do something on the current platform (but
+    we do know how to do it on some platform) -- eg. trying to compile
+    C files on a platform not supported by a CCompiler subclass."""
+    pass
+
+class DistutilsExecError (DistutilsError):
+    """Any problems executing an external program (such as the C
+    compiler, when compiling C files)."""
+    pass
+
+class DistutilsInternalError (DistutilsError):
+    """Internal inconsistencies or impossibilities (obviously, this
+    should never be seen if the code is working!)."""
+    pass
+
+class DistutilsTemplateError (DistutilsError):
+    """Syntax error in a file list template."""
+
+class DistutilsByteCompileError(DistutilsError):
+    """Byte compile error."""
+
+# Exception classes used by the CCompiler implementation classes
+class CCompilerError (Exception):
+    """Some compile/link operation failed."""
+
+class PreprocessError (CCompilerError):
+    """Failure to preprocess one or more C/C++ files."""
+
+class CompileError (CCompilerError):
+    """Failure to compile one or more C/C++ source files."""
+
+class LibError (CCompilerError):
+    """Failure to create a static library from one or more C/C++ object
+    files."""
+
+class LinkError (CCompilerError):
+    """Failure to link one or more C/C++ object files into an executable
+    or shared library file."""
+
+class UnknownFileError (CCompilerError):
+    """Attempt to process an unknown file type."""
diff --git a/Lib/distutils/extension.py b/Lib/distutils/extension.py
new file mode 100644
index 000000000..c507da360
--- /dev/null
+++ b/Lib/distutils/extension.py
@@ -0,0 +1,240 @@
+"""distutils.extension
+
+Provides the Extension class, used to describe C/C++ extension
+modules in setup scripts."""
+
+import os
+import warnings
+
+# This class is really only used by the "build_ext" command, so it might
+# make sense to put it in distutils.command.build_ext.  However, that
+# module is already big enough, and I want to make this class a bit more
+# complex to simplify some common cases ("foo" module in "foo.c") and do
+# better error-checking ("foo.c" actually exists).
+#
+# Also, putting this in build_ext.py means every setup script would have to
+# import that large-ish module (indirectly, through distutils.core) in
+# order to do anything.
+
+class Extension:
+    """Just a collection of attributes that describes an extension
+    module and everything needed to build it (hopefully in a portable
+    way, but there are hooks that let you be as unportable as you need).
+
+    Instance attributes:
+      name : string
+        the full name of the extension, including any packages -- ie.
+        *not* a filename or pathname, but Python dotted name
+      sources : [string]
+        list of source filenames, relative to the distribution root
+        (where the setup script lives), in Unix form (slash-separated)
+        for portability.  Source files may be C, C++, SWIG (.i),
+        platform-specific resource files, or whatever else is recognized
+        by the "build_ext" command as source for a Python extension.
+      include_dirs : [string]
+        list of directories to search for C/C++ header files (in Unix
+        form for portability)
+      define_macros : [(name : string, value : string|None)]
+        list of macros to define; each macro is defined using a 2-tuple,
+        where 'value' is either the string to define it to or None to
+        define it without a particular value (equivalent of "#define
+        FOO" in source or -DFOO on Unix C compiler command line)
+      undef_macros : [string]
+        list of macros to undefine explicitly
+      library_dirs : [string]
+        list of directories to search for C/C++ libraries at link time
+      libraries : [string]
+        list of library names (not filenames or paths) to link against
+      runtime_library_dirs : [string]
+        list of directories to search for C/C++ libraries at run time
+        (for shared extensions, this is when the extension is loaded)
+      extra_objects : [string]
+        list of extra files to link with (eg. object files not implied
+        by 'sources', static library that must be explicitly specified,
+        binary resource files, etc.)
+      extra_compile_args : [string]
+        any extra platform- and compiler-specific information to use
+        when compiling the source files in 'sources'.  For platforms and
+        compilers where "command line" makes sense, this is typically a
+        list of command-line arguments, but for other platforms it could
+        be anything.
+      extra_link_args : [string]
+        any extra platform- and compiler-specific information to use
+        when linking object files together to create the extension (or
+        to create a new static Python interpreter).  Similar
+        interpretation as for 'extra_compile_args'.
+      export_symbols : [string]
+        list of symbols to be exported from a shared extension.  Not
+        used on all platforms, and not generally necessary for Python
+        extensions, which typically export exactly one symbol: "init" +
+        extension_name.
+      swig_opts : [string]
+        any extra options to pass to SWIG if a source file has the .i
+        extension.
+      depends : [string]
+        list of files that the extension depends on
+      language : string
+        extension language (i.e. "c", "c++", "objc"). Will be detected
+        from the source extensions if not provided.
+      optional : boolean
+        specifies that a build failure in the extension should not abort the
+        build process, but simply not install the failing extension.
+    """
+
+    # When adding arguments to this constructor, be sure to update
+    # setup_keywords in core.py.
+    def __init__(self, name, sources,
+                  include_dirs=None,
+                  define_macros=None,
+                  undef_macros=None,
+                  library_dirs=None,
+                  libraries=None,
+                  runtime_library_dirs=None,
+                  extra_objects=None,
+                  extra_compile_args=None,
+                  extra_link_args=None,
+                  export_symbols=None,
+                  swig_opts = None,
+                  depends=None,
+                  language=None,
+                  optional=None,
+                  **kw                      # To catch unknown keywords
+                 ):
+        if not isinstance(name, str):
+            raise AssertionError("'name' must be a string")
+        if not (isinstance(sources, list) and
+                all(isinstance(v, str) for v in sources)):
+            raise AssertionError("'sources' must be a list of strings")
+
+        self.name = name
+        self.sources = sources
+        self.include_dirs = include_dirs or []
+        self.define_macros = define_macros or []
+        self.undef_macros = undef_macros or []
+        self.library_dirs = library_dirs or []
+        self.libraries = libraries or []
+        self.runtime_library_dirs = runtime_library_dirs or []
+        self.extra_objects = extra_objects or []
+        self.extra_compile_args = extra_compile_args or []
+        self.extra_link_args = extra_link_args or []
+        self.export_symbols = export_symbols or []
+        self.swig_opts = swig_opts or []
+        self.depends = depends or []
+        self.language = language
+        self.optional = optional
+
+        # If there are unknown keyword options, warn about them
+        if len(kw) > 0:
+            options = [repr(option) for option in kw]
+            options = ', '.join(sorted(options))
+            msg = "Unknown Extension options: %s" % options
+            warnings.warn(msg)
+
+    def __repr__(self):
+        return '<%s.%s(%r) at %#x>' % (
+            self.__class__.__module__,
+            self.__class__.__qualname__,
+            self.name,
+            id(self))
+
+
+def read_setup_file(filename):
+    """Reads a Setup file and returns Extension instances."""
+    from distutils.sysconfig import (parse_makefile, expand_makefile_vars,
+                                     _variable_rx)
+
+    from distutils.text_file import TextFile
+    from distutils.util import split_quoted
+
+    # First pass over the file to gather "VAR = VALUE" assignments.
+    vars = parse_makefile(filename)
+
+    # Second pass to gobble up the real content: lines of the form
+    #   <module> ... [<sourcefile> ...] [<cpparg> ...] [<library> ...]
+    file = TextFile(filename,
+                    strip_comments=1, skip_blanks=1, join_lines=1,
+                    lstrip_ws=1, rstrip_ws=1)
+    try:
+        extensions = []
+
+        while True:
+            line = file.readline()
+            if line is None:                # eof
+                break
+            if _variable_rx.match(line):    # VAR=VALUE, handled in first pass
+                continue
+
+            if line[0] == line[-1] == "*":
+                file.warn("'%s' lines not handled yet" % line)
+                continue
+
+            line = expand_makefile_vars(line, vars)
+            words = split_quoted(line)
+
+            # NB. this parses a slightly different syntax than the old
+            # makesetup script: here, there must be exactly one extension per
+            # line, and it must be the first word of the line.  I have no idea
+            # why the old syntax supported multiple extensions per line, as
+            # they all wind up being the same.
+
+            module = words[0]
+            ext = Extension(module, [])
+            append_next_word = None
+
+            for word in words[1:]:
+                if append_next_word is not None:
+                    append_next_word.append(word)
+                    append_next_word = None
+                    continue
+
+                suffix = os.path.splitext(word)[1]
+                switch = word[0:2] ; value = word[2:]
+
+                if suffix in (".c", ".cc", ".cpp", ".cxx", ".c++", ".m", ".mm"):
+                    # hmm, should we do something about C vs. C++ sources?
+                    # or leave it up to the CCompiler implementation to
+                    # worry about?
+                    ext.sources.append(word)
+                elif switch == "-I":
+                    ext.include_dirs.append(value)
+                elif switch == "-D":
+                    equals = value.find("=")
+                    if equals == -1:        # bare "-DFOO" -- no value
+                        ext.define_macros.append((value, None))
+                    else:                   # "-DFOO=blah"
+                        ext.define_macros.append((value[0:equals],
+                                                  value[equals+2:]))
+                elif switch == "-U":
+                    ext.undef_macros.append(value)
+                elif switch == "-C":        # only here 'cause makesetup has it!
+                    ext.extra_compile_args.append(word)
+                elif switch == "-l":
+                    ext.libraries.append(value)
+                elif switch == "-L":
+                    ext.library_dirs.append(value)
+                elif switch == "-R":
+                    ext.runtime_library_dirs.append(value)
+                elif word == "-rpath":
+                    append_next_word = ext.runtime_library_dirs
+                elif word == "-Xlinker":
+                    append_next_word = ext.extra_link_args
+                elif word == "-Xcompiler":
+                    append_next_word = ext.extra_compile_args
+                elif switch == "-u":
+                    ext.extra_link_args.append(word)
+                    if not value:
+                        append_next_word = ext.extra_link_args
+                elif suffix in (".a", ".so", ".sl", ".o", ".dylib"):
+                    # NB. a really faithful emulation of makesetup would
+                    # append a .o file to extra_objects only if it
+                    # had a slash in it; otherwise, it would s/.o/.c/
+                    # and append it to sources.  Hmmmm.
+                    ext.extra_objects.append(word)
+                else:
+                    file.warn("unrecognized argument '%s'" % word)
+
+            extensions.append(ext)
+    finally:
+        file.close()
+
+    return extensions
diff --git a/Lib/distutils/fancy_getopt.py b/Lib/distutils/fancy_getopt.py
new file mode 100644
index 000000000..7d170dd27
--- /dev/null
+++ b/Lib/distutils/fancy_getopt.py
@@ -0,0 +1,457 @@
+"""distutils.fancy_getopt
+
+Wrapper around the standard getopt module that provides the following
+additional features:
+  * short and long options are tied together
+  * options have help strings, so fancy_getopt could potentially
+    create a complete usage summary
+  * options set attributes of a passed-in object
+"""
+
+import sys, string, re
+import getopt
+from distutils.errors import *
+
+# Much like command_re in distutils.core, this is close to but not quite
+# the same as a Python NAME -- except, in the spirit of most GNU
+# utilities, we use '-' in place of '_'.  (The spirit of LISP lives on!)
+# The similarities to NAME are again not a coincidence...
+longopt_pat = r'[a-zA-Z](?:[a-zA-Z0-9-]*)'
+longopt_re = re.compile(r'^%s$' % longopt_pat)
+
+# For recognizing "negative alias" options, eg. "quiet=!verbose"
+neg_alias_re = re.compile("^(%s)=!(%s)$" % (longopt_pat, longopt_pat))
+
+# This is used to translate long options to legitimate Python identifiers
+# (for use as attributes of some object).
+longopt_xlate = str.maketrans('-', '_')
+
+class FancyGetopt:
+    """Wrapper around the standard 'getopt()' module that provides some
+    handy extra functionality:
+      * short and long options are tied together
+      * options have help strings, and help text can be assembled
+        from them
+      * options set attributes of a passed-in object
+      * boolean options can have "negative aliases" -- eg. if
+        --quiet is the "negative alias" of --verbose, then "--quiet"
+        on the command line sets 'verbose' to false
+    """
+
+    def __init__(self, option_table=None):
+        # The option table is (currently) a list of tuples.  The
+        # tuples may have 3 or four values:
+        #   (long_option, short_option, help_string [, repeatable])
+        # if an option takes an argument, its long_option should have '='
+        # appended; short_option should just be a single character, no ':'
+        # in any case.  If a long_option doesn't have a corresponding
+        # short_option, short_option should be None.  All option tuples
+        # must have long options.
+        self.option_table = option_table
+
+        # 'option_index' maps long option names to entries in the option
+        # table (ie. those 3-tuples).
+        self.option_index = {}
+        if self.option_table:
+            self._build_index()
+
+        # 'alias' records (duh) alias options; {'foo': 'bar'} means
+        # --foo is an alias for --bar
+        self.alias = {}
+
+        # 'negative_alias' keeps track of options that are the boolean
+        # opposite of some other option
+        self.negative_alias = {}
+
+        # These keep track of the information in the option table.  We
+        # don't actually populate these structures until we're ready to
+        # parse the command-line, since the 'option_table' passed in here
+        # isn't necessarily the final word.
+        self.short_opts = []
+        self.long_opts = []
+        self.short2long = {}
+        self.attr_name = {}
+        self.takes_arg = {}
+
+        # And 'option_order' is filled up in 'getopt()'; it records the
+        # original order of options (and their values) on the command-line,
+        # but expands short options, converts aliases, etc.
+        self.option_order = []
+
+    def _build_index(self):
+        self.option_index.clear()
+        for option in self.option_table:
+            self.option_index[option[0]] = option
+
+    def set_option_table(self, option_table):
+        self.option_table = option_table
+        self._build_index()
+
+    def add_option(self, long_option, short_option=None, help_string=None):
+        if long_option in self.option_index:
+            raise DistutilsGetoptError(
+                  "option conflict: already an option '%s'" % long_option)
+        else:
+            option = (long_option, short_option, help_string)
+            self.option_table.append(option)
+            self.option_index[long_option] = option
+
+    def has_option(self, long_option):
+        """Return true if the option table for this parser has an
+        option with long name 'long_option'."""
+        return long_option in self.option_index
+
+    def get_attr_name(self, long_option):
+        """Translate long option name 'long_option' to the form it
+        has as an attribute of some object: ie., translate hyphens
+        to underscores."""
+        return long_option.translate(longopt_xlate)
+
+    def _check_alias_dict(self, aliases, what):
+        assert isinstance(aliases, dict)
+        for (alias, opt) in aliases.items():
+            if alias not in self.option_index:
+                raise DistutilsGetoptError(("invalid %s '%s': "
+                       "option '%s' not defined") % (what, alias, alias))
+            if opt not in self.option_index:
+                raise DistutilsGetoptError(("invalid %s '%s': "
+                       "aliased option '%s' not defined") % (what, alias, opt))
+
+    def set_aliases(self, alias):
+        """Set the aliases for this option parser."""
+        self._check_alias_dict(alias, "alias")
+        self.alias = alias
+
+    def set_negative_aliases(self, negative_alias):
+        """Set the negative aliases for this option parser.
+        'negative_alias' should be a dictionary mapping option names to
+        option names, both the key and value must already be defined
+        in the option table."""
+        self._check_alias_dict(negative_alias, "negative alias")
+        self.negative_alias = negative_alias
+
+    def _grok_option_table(self):
+        """Populate the various data structures that keep tabs on the
+        option table.  Called by 'getopt()' before it can do anything
+        worthwhile.
+        """
+        self.long_opts = []
+        self.short_opts = []
+        self.short2long.clear()
+        self.repeat = {}
+
+        for option in self.option_table:
+            if len(option) == 3:
+                long, short, help = option
+                repeat = 0
+            elif len(option) == 4:
+                long, short, help, repeat = option
+            else:
+                # the option table is part of the code, so simply
+                # assert that it is correct
+                raise ValueError("invalid option tuple: %r" % (option,))
+
+            # Type- and value-check the option names
+            if not isinstance(long, str) or len(long) < 2:
+                raise DistutilsGetoptError(("invalid long option '%s': "
+                       "must be a string of length >= 2") % long)
+
+            if (not ((short is None) or
+                     (isinstance(short, str) and len(short) == 1))):
+                raise DistutilsGetoptError("invalid short option '%s': "
+                       "must a single character or None" % short)
+
+            self.repeat[long] = repeat
+            self.long_opts.append(long)
+
+            if long[-1] == '=':             # option takes an argument?
+                if short: short = short + ':'
+                long = long[0:-1]
+                self.takes_arg[long] = 1
+            else:
+                # Is option is a "negative alias" for some other option (eg.
+                # "quiet" == "!verbose")?
+                alias_to = self.negative_alias.get(long)
+                if alias_to is not None:
+                    if self.takes_arg[alias_to]:
+                        raise DistutilsGetoptError(
+                              "invalid negative alias '%s': "
+                              "aliased option '%s' takes a value"
+                              % (long, alias_to))
+
+                    self.long_opts[-1] = long # XXX redundant?!
+                self.takes_arg[long] = 0
+
+            # If this is an alias option, make sure its "takes arg" flag is
+            # the same as the option it's aliased to.
+            alias_to = self.alias.get(long)
+            if alias_to is not None:
+                if self.takes_arg[long] != self.takes_arg[alias_to]:
+                    raise DistutilsGetoptError(
+                          "invalid alias '%s': inconsistent with "
+                          "aliased option '%s' (one of them takes a value, "
+                          "the other doesn't"
+                          % (long, alias_to))
+
+            # Now enforce some bondage on the long option name, so we can
+            # later translate it to an attribute name on some object.  Have
+            # to do this a bit late to make sure we've removed any trailing
+            # '='.
+            if not longopt_re.match(long):
+                raise DistutilsGetoptError(
+                       "invalid long option name '%s' "
+                       "(must be letters, numbers, hyphens only" % long)
+
+            self.attr_name[long] = self.get_attr_name(long)
+            if short:
+                self.short_opts.append(short)
+                self.short2long[short[0]] = long
+
+    def getopt(self, args=None, object=None):
+        """Parse command-line options in args. Store as attributes on object.
+
+        If 'args' is None or not supplied, uses 'sys.argv[1:]'.  If
+        'object' is None or not supplied, creates a new OptionDummy
+        object, stores option values there, and returns a tuple (args,
+        object).  If 'object' is supplied, it is modified in place and
+        'getopt()' just returns 'args'; in both cases, the returned
+        'args' is a modified copy of the passed-in 'args' list, which
+        is left untouched.
+        """
+        if args is None:
+            args = sys.argv[1:]
+        if object is None:
+            object = OptionDummy()
+            created_object = True
+        else:
+            created_object = False
+
+        self._grok_option_table()
+
+        short_opts = ' '.join(self.short_opts)
+        try:
+            opts, args = getopt.getopt(args, short_opts, self.long_opts)
+        except getopt.error as msg:
+            raise DistutilsArgError(msg)
+
+        for opt, val in opts:
+            if len(opt) == 2 and opt[0] == '-': # it's a short option
+                opt = self.short2long[opt[1]]
+            else:
+                assert len(opt) > 2 and opt[:2] == '--'
+                opt = opt[2:]
+
+            alias = self.alias.get(opt)
+            if alias:
+                opt = alias
+
+            if not self.takes_arg[opt]:     # boolean option?
+                assert val == '', "boolean option can't have value"
+                alias = self.negative_alias.get(opt)
+                if alias:
+                    opt = alias
+                    val = 0
+                else:
+                    val = 1
+
+            attr = self.attr_name[opt]
+            # The only repeating option at the moment is 'verbose'.
+            # It has a negative option -q quiet, which should set verbose = 0.
+            if val and self.repeat.get(attr) is not None:
+                val = getattr(object, attr, 0) + 1
+            setattr(object, attr, val)
+            self.option_order.append((opt, val))
+
+        # for opts
+        if created_object:
+            return args, object
+        else:
+            return args
+
+    def get_option_order(self):
+        """Returns the list of (option, value) tuples processed by the
+        previous run of 'getopt()'.  Raises RuntimeError if
+        'getopt()' hasn't been called yet.
+        """
+        if self.option_order is None:
+            raise RuntimeError("'getopt()' hasn't been called yet")
+        else:
+            return self.option_order
+
+    def generate_help(self, header=None):
+        """Generate help text (a list of strings, one per suggested line of
+        output) from the option table for this FancyGetopt object.
+        """
+        # Blithely assume the option table is good: probably wouldn't call
+        # 'generate_help()' unless you've already called 'getopt()'.
+
+        # First pass: determine maximum length of long option names
+        max_opt = 0
+        for option in self.option_table:
+            long = option[0]
+            short = option[1]
+            l = len(long)
+            if long[-1] == '=':
+                l = l - 1
+            if short is not None:
+                l = l + 5                   # " (-x)" where short == 'x'
+            if l > max_opt:
+                max_opt = l
+
+        opt_width = max_opt + 2 + 2 + 2     # room for indent + dashes + gutter
+
+        # Typical help block looks like this:
+        #   --foo       controls foonabulation
+        # Help block for longest option looks like this:
+        #   --flimflam  set the flim-flam level
+        # and with wrapped text:
+        #   --flimflam  set the flim-flam level (must be between
+        #               0 and 100, except on Tuesdays)
+        # Options with short names will have the short name shown (but
+        # it doesn't contribute to max_opt):
+        #   --foo (-f)  controls foonabulation
+        # If adding the short option would make the left column too wide,
+        # we push the explanation off to the next line
+        #   --flimflam (-l)
+        #               set the flim-flam level
+        # Important parameters:
+        #   - 2 spaces before option block start lines
+        #   - 2 dashes for each long option name
+        #   - min. 2 spaces between option and explanation (gutter)
+        #   - 5 characters (incl. space) for short option name
+
+        # Now generate lines of help text.  (If 80 columns were good enough
+        # for Jesus, then 78 columns are good enough for me!)
+        line_width = 78
+        text_width = line_width - opt_width
+        big_indent = ' ' * opt_width
+        if header:
+            lines = [header]
+        else:
+            lines = ['Option summary:']
+
+        for option in self.option_table:
+            long, short, help = option[:3]
+            text = wrap_text(help, text_width)
+            if long[-1] == '=':
+                long = long[0:-1]
+
+            # Case 1: no short option at all (makes life easy)
+            if short is None:
+                if text:
+                    lines.append("  --%-*s  %s" % (max_opt, long, text[0]))
+                else:
+                    lines.append("  --%-*s  " % (max_opt, long))
+
+            # Case 2: we have a short option, so we have to include it
+            # just after the long option
+            else:
+                opt_names = "%s (-%s)" % (long, short)
+                if text:
+                    lines.append("  --%-*s  %s" %
+                                 (max_opt, opt_names, text[0]))
+                else:
+                    lines.append("  --%-*s" % opt_names)
+
+            for l in text[1:]:
+                lines.append(big_indent + l)
+        return lines
+
+    def print_help(self, header=None, file=None):
+        if file is None:
+            file = sys.stdout
+        for line in self.generate_help(header):
+            file.write(line + "\n")
+
+
+def fancy_getopt(options, negative_opt, object, args):
+    parser = FancyGetopt(options)
+    parser.set_negative_aliases(negative_opt)
+    return parser.getopt(args, object)
+
+
+WS_TRANS = {ord(_wschar) : ' ' for _wschar in string.whitespace}
+
+def wrap_text(text, width):
+    """wrap_text(text : string, width : int) -> [string]
+
+    Split 'text' into multiple lines of no more than 'width' characters
+    each, and return the list of strings that results.
+    """
+    if text is None:
+        return []
+    if len(text) <= width:
+        return [text]
+
+    text = text.expandtabs()
+    text = text.translate(WS_TRANS)
+    chunks = re.split(r'( +|-+)', text)
+    chunks = [ch for ch in chunks if ch] # ' - ' results in empty strings
+    lines = []
+
+    while chunks:
+        cur_line = []                   # list of chunks (to-be-joined)
+        cur_len = 0                     # length of current line
+
+        while chunks:
+            l = len(chunks[0])
+            if cur_len + l <= width:    # can squeeze (at least) this chunk in
+                cur_line.append(chunks[0])
+                del chunks[0]
+                cur_len = cur_len + l
+            else:                       # this line is full
+                # drop last chunk if all space
+                if cur_line and cur_line[-1][0] == ' ':
+                    del cur_line[-1]
+                break
+
+        if chunks:                      # any chunks left to process?
+            # if the current line is still empty, then we had a single
+            # chunk that's too big too fit on a line -- so we break
+            # down and break it up at the line width
+            if cur_len == 0:
+                cur_line.append(chunks[0][0:width])
+                chunks[0] = chunks[0][width:]
+
+            # all-whitespace chunks at the end of a line can be discarded
+            # (and we know from the re.split above that if a chunk has
+            # *any* whitespace, it is *all* whitespace)
+            if chunks[0][0] == ' ':
+                del chunks[0]
+
+        # and store this line in the list-of-all-lines -- as a single
+        # string, of course!
+        lines.append(''.join(cur_line))
+
+    return lines
+
+
+def translate_longopt(opt):
+    """Convert a long option name to a valid Python identifier by
+    changing "-" to "_".
+    """
+    return opt.translate(longopt_xlate)
+
+
+class OptionDummy:
+    """Dummy class just used as a place to hold command-line option
+    values as instance attributes."""
+
+    def __init__(self, options=[]):
+        """Create a new OptionDummy instance.  The attributes listed in
+        'options' will be initialized to None."""
+        for opt in options:
+            setattr(self, opt, None)
+
+
+if __name__ == "__main__":
+    text = """\
+Tra-la-la, supercalifragilisticexpialidocious.
+How *do* you spell that odd word, anyways?
+(Someone ask Mary -- she'll know [or she'll
+say, "How should I know?"].)"""
+
+    for w in (10, 20, 30, 40):
+        print("width: %d" % w)
+        print("\n".join(wrap_text(text, w)))
+        print()
diff --git a/Lib/distutils/file_util.py b/Lib/distutils/file_util.py
new file mode 100644
index 000000000..b3fee35a6
--- /dev/null
+++ b/Lib/distutils/file_util.py
@@ -0,0 +1,238 @@
+"""distutils.file_util
+
+Utility functions for operating on single files.
+"""
+
+import os
+from distutils.errors import DistutilsFileError
+from distutils import log
+
+# for generating verbose output in 'copy_file()'
+_copy_action = { None:   'copying',
+                 'hard': 'hard linking',
+                 'sym':  'symbolically linking' }
+
+
+def _copy_file_contents(src, dst, buffer_size=16*1024):
+    """Copy the file 'src' to 'dst'; both must be filenames.  Any error
+    opening either file, reading from 'src', or writing to 'dst', raises
+    DistutilsFileError.  Data is read/written in chunks of 'buffer_size'
+    bytes (default 16k).  No attempt is made to handle anything apart from
+    regular files.
+    """
+    # Stolen from shutil module in the standard library, but with
+    # custom error-handling added.
+    fsrc = None
+    fdst = None
+    try:
+        try:
+            fsrc = open(src, 'rb')
+        except OSError as e:
+            raise DistutilsFileError("could not open '%s': %s" % (src, e.strerror))
+
+        if os.path.exists(dst):
+            try:
+                os.unlink(dst)
+            except OSError as e:
+                raise DistutilsFileError(
+                      "could not delete '%s': %s" % (dst, e.strerror))
+
+        try:
+            fdst = open(dst, 'wb')
+        except OSError as e:
+            raise DistutilsFileError(
+                  "could not create '%s': %s" % (dst, e.strerror))
+
+        while True:
+            try:
+                buf = fsrc.read(buffer_size)
+            except OSError as e:
+                raise DistutilsFileError(
+                      "could not read from '%s': %s" % (src, e.strerror))
+
+            if not buf:
+                break
+
+            try:
+                fdst.write(buf)
+            except OSError as e:
+                raise DistutilsFileError(
+                      "could not write to '%s': %s" % (dst, e.strerror))
+    finally:
+        if fdst:
+            fdst.close()
+        if fsrc:
+            fsrc.close()
+
+def copy_file(src, dst, preserve_mode=1, preserve_times=1, update=0,
+              link=None, verbose=1, dry_run=0):
+    """Copy a file 'src' to 'dst'.  If 'dst' is a directory, then 'src' is
+    copied there with the same name; otherwise, it must be a filename.  (If
+    the file exists, it will be ruthlessly clobbered.)  If 'preserve_mode'
+    is true (the default), the file's mode (type and permission bits, or
+    whatever is analogous on the current platform) is copied.  If
+    'preserve_times' is true (the default), the last-modified and
+    last-access times are copied as well.  If 'update' is true, 'src' will
+    only be copied if 'dst' does not exist, or if 'dst' does exist but is
+    older than 'src'.
+
+    'link' allows you to make hard links (os.link) or symbolic links
+    (os.symlink) instead of copying: set it to "hard" or "sym"; if it is
+    None (the default), files are copied.  Don't set 'link' on systems that
+    don't support it: 'copy_file()' doesn't check if hard or symbolic
+    linking is available. If hardlink fails, falls back to
+    _copy_file_contents().
+
+    Under Mac OS, uses the native file copy function in macostools; on
+    other systems, uses '_copy_file_contents()' to copy file contents.
+
+    Return a tuple (dest_name, copied): 'dest_name' is the actual name of
+    the output file, and 'copied' is true if the file was copied (or would
+    have been copied, if 'dry_run' true).
+    """
+    # XXX if the destination file already exists, we clobber it if
+    # copying, but blow up if linking.  Hmmm.  And I don't know what
+    # macostools.copyfile() does.  Should definitely be consistent, and
+    # should probably blow up if destination exists and we would be
+    # changing it (ie. it's not already a hard/soft link to src OR
+    # (not update) and (src newer than dst).
+
+    from distutils.dep_util import newer
+    from stat import ST_ATIME, ST_MTIME, ST_MODE, S_IMODE
+
+    if not os.path.isfile(src):
+        raise DistutilsFileError(
+              "can't copy '%s': doesn't exist or not a regular file" % src)
+
+    if os.path.isdir(dst):
+        dir = dst
+        dst = os.path.join(dst, os.path.basename(src))
+    else:
+        dir = os.path.dirname(dst)
+
+    if update and not newer(src, dst):
+        if verbose >= 1:
+            log.debug("not copying %s (output up-to-date)", src)
+        return (dst, 0)
+
+    try:
+        action = _copy_action[link]
+    except KeyError:
+        raise ValueError("invalid value '%s' for 'link' argument" % link)
+
+    if verbose >= 1:
+        if os.path.basename(dst) == os.path.basename(src):
+            log.info("%s %s -> %s", action, src, dir)
+        else:
+            log.info("%s %s -> %s", action, src, dst)
+
+    if dry_run:
+        return (dst, 1)
+
+    # If linking (hard or symbolic), use the appropriate system call
+    # (Unix only, of course, but that's the caller's responsibility)
+    elif link == 'hard':
+        if not (os.path.exists(dst) and os.path.samefile(src, dst)):
+            try:
+                os.link(src, dst)
+                return (dst, 1)
+            except OSError:
+                # If hard linking fails, fall back on copying file
+                # (some special filesystems don't support hard linking
+                #  even under Unix, see issue #8876).
+                pass
+    elif link == 'sym':
+        if not (os.path.exists(dst) and os.path.samefile(src, dst)):
+            os.symlink(src, dst)
+            return (dst, 1)
+
+    # Otherwise (non-Mac, not linking), copy the file contents and
+    # (optionally) copy the times and mode.
+    _copy_file_contents(src, dst)
+    if preserve_mode or preserve_times:
+        st = os.stat(src)
+
+        # According to David Ascher <da@ski.org>, utime() should be done
+        # before chmod() (at least under NT).
+        if preserve_times:
+            os.utime(dst, (st[ST_ATIME], st[ST_MTIME]))
+        if preserve_mode:
+            os.chmod(dst, S_IMODE(st[ST_MODE]))
+
+    return (dst, 1)
+
+
+# XXX I suspect this is Unix-specific -- need porting help!
+def move_file (src, dst,
+               verbose=1,
+               dry_run=0):
+
+    """Move a file 'src' to 'dst'.  If 'dst' is a directory, the file will
+    be moved into it with the same name; otherwise, 'src' is just renamed
+    to 'dst'.  Return the new full name of the file.
+
+    Handles cross-device moves on Unix using 'copy_file()'.  What about
+    other systems???
+    """
+    from os.path import exists, isfile, isdir, basename, dirname
+    import errno
+
+    if verbose >= 1:
+        log.info("moving %s -> %s", src, dst)
+
+    if dry_run:
+        return dst
+
+    if not isfile(src):
+        raise DistutilsFileError("can't move '%s': not a regular file" % src)
+
+    if isdir(dst):
+        dst = os.path.join(dst, basename(src))
+    elif exists(dst):
+        raise DistutilsFileError(
+              "can't move '%s': destination '%s' already exists" %
+              (src, dst))
+
+    if not isdir(dirname(dst)):
+        raise DistutilsFileError(
+              "can't move '%s': destination '%s' not a valid path" %
+              (src, dst))
+
+    copy_it = False
+    try:
+        os.rename(src, dst)
+    except OSError as e:
+        (num, msg) = e.args
+        if num == errno.EXDEV:
+            copy_it = True
+        else:
+            raise DistutilsFileError(
+                  "couldn't move '%s' to '%s': %s" % (src, dst, msg))
+
+    if copy_it:
+        copy_file(src, dst, verbose=verbose)
+        try:
+            os.unlink(src)
+        except OSError as e:
+            (num, msg) = e.args
+            try:
+                os.unlink(dst)
+            except OSError:
+                pass
+            raise DistutilsFileError(
+                  "couldn't move '%s' to '%s' by copy/delete: "
+                  "delete '%s' failed: %s"
+                  % (src, dst, src, msg))
+    return dst
+
+
+def write_file (filename, contents):
+    """Create a file with the specified name and write 'contents' (a
+    sequence of strings without line terminators) to it.
+    """
+    f = open(filename, "w")
+    try:
+        for line in contents:
+            f.write(line + "\n")
+    finally:
+        f.close()
diff --git a/Lib/distutils/filelist.py b/Lib/distutils/filelist.py
new file mode 100644
index 000000000..c92d5fdba
--- /dev/null
+++ b/Lib/distutils/filelist.py
@@ -0,0 +1,327 @@
+"""distutils.filelist
+
+Provides the FileList class, used for poking about the filesystem
+and building lists of files.
+"""
+
+import os, re
+import fnmatch
+import functools
+from distutils.util import convert_path
+from distutils.errors import DistutilsTemplateError, DistutilsInternalError
+from distutils import log
+
+class FileList:
+    """A list of files built by on exploring the filesystem and filtered by
+    applying various patterns to what we find there.
+
+    Instance attributes:
+      dir
+        directory from which files will be taken -- only used if
+        'allfiles' not supplied to constructor
+      files
+        list of filenames currently being built/filtered/manipulated
+      allfiles
+        complete list of files under consideration (ie. without any
+        filtering applied)
+    """
+
+    def __init__(self, warn=None, debug_print=None):
+        # ignore argument to FileList, but keep them for backwards
+        # compatibility
+        self.allfiles = None
+        self.files = []
+
+    def set_allfiles(self, allfiles):
+        self.allfiles = allfiles
+
+    def findall(self, dir=os.curdir):
+        self.allfiles = findall(dir)
+
+    def debug_print(self, msg):
+        """Print 'msg' to stdout if the global DEBUG (taken from the
+        DISTUTILS_DEBUG environment variable) flag is true.
+        """
+        from distutils.debug import DEBUG
+        if DEBUG:
+            print(msg)
+
+    # -- List-like methods ---------------------------------------------
+
+    def append(self, item):
+        self.files.append(item)
+
+    def extend(self, items):
+        self.files.extend(items)
+
+    def sort(self):
+        # Not a strict lexical sort!
+        sortable_files = sorted(map(os.path.split, self.files))
+        self.files = []
+        for sort_tuple in sortable_files:
+            self.files.append(os.path.join(*sort_tuple))
+
+
+    # -- Other miscellaneous utility methods ---------------------------
+
+    def remove_duplicates(self):
+        # Assumes list has been sorted!
+        for i in range(len(self.files) - 1, 0, -1):
+            if self.files[i] == self.files[i - 1]:
+                del self.files[i]
+
+
+    # -- "File template" methods ---------------------------------------
+
+    def _parse_template_line(self, line):
+        words = line.split()
+        action = words[0]
+
+        patterns = dir = dir_pattern = None
+
+        if action in ('include', 'exclude',
+                      'global-include', 'global-exclude'):
+            if len(words) < 2:
+                raise DistutilsTemplateError(
+                      "'%s' expects <pattern1> <pattern2> ..." % action)
+            patterns = [convert_path(w) for w in words[1:]]
+        elif action in ('recursive-include', 'recursive-exclude'):
+            if len(words) < 3:
+                raise DistutilsTemplateError(
+                      "'%s' expects <dir> <pattern1> <pattern2> ..." % action)
+            dir = convert_path(words[1])
+            patterns = [convert_path(w) for w in words[2:]]
+        elif action in ('graft', 'prune'):
+            if len(words) != 2:
+                raise DistutilsTemplateError(
+                      "'%s' expects a single <dir_pattern>" % action)
+            dir_pattern = convert_path(words[1])
+        else:
+            raise DistutilsTemplateError("unknown action '%s'" % action)
+
+        return (action, patterns, dir, dir_pattern)
+
+    def process_template_line(self, line):
+        # Parse the line: split it up, make sure the right number of words
+        # is there, and return the relevant words.  'action' is always
+        # defined: it's the first word of the line.  Which of the other
+        # three are defined depends on the action; it'll be either
+        # patterns, (dir and patterns), or (dir_pattern).
+        (action, patterns, dir, dir_pattern) = self._parse_template_line(line)
+
+        # OK, now we know that the action is valid and we have the
+        # right number of words on the line for that action -- so we
+        # can proceed with minimal error-checking.
+        if action == 'include':
+            self.debug_print("include " + ' '.join(patterns))
+            for pattern in patterns:
+                if not self.include_pattern(pattern, anchor=1):
+                    log.warn("warning: no files found matching '%s'",
+                             pattern)
+
+        elif action == 'exclude':
+            self.debug_print("exclude " + ' '.join(patterns))
+            for pattern in patterns:
+                if not self.exclude_pattern(pattern, anchor=1):
+                    log.warn(("warning: no previously-included files "
+                              "found matching '%s'"), pattern)
+
+        elif action == 'global-include':
+            self.debug_print("global-include " + ' '.join(patterns))
+            for pattern in patterns:
+                if not self.include_pattern(pattern, anchor=0):
+                    log.warn(("warning: no files found matching '%s' "
+                              "anywhere in distribution"), pattern)
+
+        elif action == 'global-exclude':
+            self.debug_print("global-exclude " + ' '.join(patterns))
+            for pattern in patterns:
+                if not self.exclude_pattern(pattern, anchor=0):
+                    log.warn(("warning: no previously-included files matching "
+                              "'%s' found anywhere in distribution"),
+                             pattern)
+
+        elif action == 'recursive-include':
+            self.debug_print("recursive-include %s %s" %
+                             (dir, ' '.join(patterns)))
+            for pattern in patterns:
+                if not self.include_pattern(pattern, prefix=dir):
+                    log.warn(("warning: no files found matching '%s' "
+                                "under directory '%s'"),
+                             pattern, dir)
+
+        elif action == 'recursive-exclude':
+            self.debug_print("recursive-exclude %s %s" %
+                             (dir, ' '.join(patterns)))
+            for pattern in patterns:
+                if not self.exclude_pattern(pattern, prefix=dir):
+                    log.warn(("warning: no previously-included files matching "
+                              "'%s' found under directory '%s'"),
+                             pattern, dir)
+
+        elif action == 'graft':
+            self.debug_print("graft " + dir_pattern)
+            if not self.include_pattern(None, prefix=dir_pattern):
+                log.warn("warning: no directories found matching '%s'",
+                         dir_pattern)
+
+        elif action == 'prune':
+            self.debug_print("prune " + dir_pattern)
+            if not self.exclude_pattern(None, prefix=dir_pattern):
+                log.warn(("no previously-included directories found "
+                          "matching '%s'"), dir_pattern)
+        else:
+            raise DistutilsInternalError(
+                  "this cannot happen: invalid action '%s'" % action)
+
+
+    # -- Filtering/selection methods -----------------------------------
+
+    def include_pattern(self, pattern, anchor=1, prefix=None, is_regex=0):
+        """Select strings (presumably filenames) from 'self.files' that
+        match 'pattern', a Unix-style wildcard (glob) pattern.  Patterns
+        are not quite the same as implemented by the 'fnmatch' module: '*'
+        and '?'  match non-special characters, where "special" is platform-
+        dependent: slash on Unix; colon, slash, and backslash on
+        DOS/Windows; and colon on Mac OS.
+
+        If 'anchor' is true (the default), then the pattern match is more
+        stringent: "*.py" will match "foo.py" but not "foo/bar.py".  If
+        'anchor' is false, both of these will match.
+
+        If 'prefix' is supplied, then only filenames starting with 'prefix'
+        (itself a pattern) and ending with 'pattern', with anything in between
+        them, will match.  'anchor' is ignored in this case.
+
+        If 'is_regex' is true, 'anchor' and 'prefix' are ignored, and
+        'pattern' is assumed to be either a string containing a regex or a
+        regex object -- no translation is done, the regex is just compiled
+        and used as-is.
+
+        Selected strings will be added to self.files.
+
+        Return True if files are found, False otherwise.
+        """
+        # XXX docstring lying about what the special chars are?
+        files_found = False
+        pattern_re = translate_pattern(pattern, anchor, prefix, is_regex)
+        self.debug_print("include_pattern: applying regex r'%s'" %
+                         pattern_re.pattern)
+
+        # delayed loading of allfiles list
+        if self.allfiles is None:
+            self.findall()
+
+        for name in self.allfiles:
+            if pattern_re.search(name):
+                self.debug_print(" adding " + name)
+                self.files.append(name)
+                files_found = True
+        return files_found
+
+
+    def exclude_pattern (self, pattern,
+                         anchor=1, prefix=None, is_regex=0):
+        """Remove strings (presumably filenames) from 'files' that match
+        'pattern'.  Other parameters are the same as for
+        'include_pattern()', above.
+        The list 'self.files' is modified in place.
+        Return True if files are found, False otherwise.
+        """
+        files_found = False
+        pattern_re = translate_pattern(pattern, anchor, prefix, is_regex)
+        self.debug_print("exclude_pattern: applying regex r'%s'" %
+                         pattern_re.pattern)
+        for i in range(len(self.files)-1, -1, -1):
+            if pattern_re.search(self.files[i]):
+                self.debug_print(" removing " + self.files[i])
+                del self.files[i]
+                files_found = True
+        return files_found
+
+
+# ----------------------------------------------------------------------
+# Utility functions
+
+def _find_all_simple(path):
+    """
+    Find all files under 'path'
+    """
+    results = (
+        os.path.join(base, file)
+        for base, dirs, files in os.walk(path, followlinks=True)
+        for file in files
+    )
+    return filter(os.path.isfile, results)
+
+
+def findall(dir=os.curdir):
+    """
+    Find all files under 'dir' and return the list of full filenames.
+    Unless dir is '.', return full filenames with dir prepended.
+    """
+    files = _find_all_simple(dir)
+    if dir == os.curdir:
+        make_rel = functools.partial(os.path.relpath, start=dir)
+        files = map(make_rel, files)
+    return list(files)
+
+
+def glob_to_re(pattern):
+    """Translate a shell-like glob pattern to a regular expression; return
+    a string containing the regex.  Differs from 'fnmatch.translate()' in
+    that '*' does not match "special characters" (which are
+    platform-specific).
+    """
+    pattern_re = fnmatch.translate(pattern)
+
+    # '?' and '*' in the glob pattern become '.' and '.*' in the RE, which
+    # IMHO is wrong -- '?' and '*' aren't supposed to match slash in Unix,
+    # and by extension they shouldn't match such "special characters" under
+    # any OS.  So change all non-escaped dots in the RE to match any
+    # character except the special characters (currently: just os.sep).
+    sep = os.sep
+    if os.sep == '\\':
+        # we're using a regex to manipulate a regex, so we need
+        # to escape the backslash twice
+        sep = r'\\\\'
+    escaped = r'\1[^%s]' % sep
+    pattern_re = re.sub(r'((?<!\\)(\\\\)*)\.', escaped, pattern_re)
+    return pattern_re
+
+
+def translate_pattern(pattern, anchor=1, prefix=None, is_regex=0):
+    """Translate a shell-like wildcard pattern to a compiled regular
+    expression.  Return the compiled regex.  If 'is_regex' true,
+    then 'pattern' is directly compiled to a regex (if it's a string)
+    or just returned as-is (assumes it's a regex object).
+    """
+    if is_regex:
+        if isinstance(pattern, str):
+            return re.compile(pattern)
+        else:
+            return pattern
+
+    # ditch start and end characters
+    start, _, end = glob_to_re('_').partition('_')
+
+    if pattern:
+        pattern_re = glob_to_re(pattern)
+        assert pattern_re.startswith(start) and pattern_re.endswith(end)
+    else:
+        pattern_re = ''
+
+    if prefix is not None:
+        prefix_re = glob_to_re(prefix)
+        assert prefix_re.startswith(start) and prefix_re.endswith(end)
+        prefix_re = prefix_re[len(start): len(prefix_re) - len(end)]
+        sep = os.sep
+        if os.sep == '\\':
+            sep = r'\\'
+        pattern_re = pattern_re[len(start): len(pattern_re) - len(end)]
+        pattern_re = r'%s\A%s%s.*%s%s' % (start, prefix_re, sep, pattern_re, end)
+    else:                               # no prefix -- respect anchor flag
+        if anchor:
+            pattern_re = r'%s\A%s' % (start, pattern_re[len(start):])
+
+    return re.compile(pattern_re)
diff --git a/Lib/distutils/log.py b/Lib/distutils/log.py
new file mode 100644
index 000000000..8ef6b28ea
--- /dev/null
+++ b/Lib/distutils/log.py
@@ -0,0 +1,77 @@
+"""A simple log mechanism styled after PEP 282."""
+
+# The class here is styled after PEP 282 so that it could later be
+# replaced with a standard Python logging implementation.
+
+DEBUG = 1
+INFO = 2
+WARN = 3
+ERROR = 4
+FATAL = 5
+
+import sys
+
+class Log:
+
+    def __init__(self, threshold=WARN):
+        self.threshold = threshold
+
+    def _log(self, level, msg, args):
+        if level not in (DEBUG, INFO, WARN, ERROR, FATAL):
+            raise ValueError('%s wrong log level' % str(level))
+
+        if level >= self.threshold:
+            if args:
+                msg = msg % args
+            if level in (WARN, ERROR, FATAL):
+                stream = sys.stderr
+            else:
+                stream = sys.stdout
+            try:
+                stream.write('%s\n' % msg)
+            except UnicodeEncodeError:
+                # emulate backslashreplace error handler
+                encoding = stream.encoding
+                msg = msg.encode(encoding, "backslashreplace").decode(encoding)
+                stream.write('%s\n' % msg)
+            stream.flush()
+
+    def log(self, level, msg, *args):
+        self._log(level, msg, args)
+
+    def debug(self, msg, *args):
+        self._log(DEBUG, msg, args)
+
+    def info(self, msg, *args):
+        self._log(INFO, msg, args)
+
+    def warn(self, msg, *args):
+        self._log(WARN, msg, args)
+
+    def error(self, msg, *args):
+        self._log(ERROR, msg, args)
+
+    def fatal(self, msg, *args):
+        self._log(FATAL, msg, args)
+
+_global_log = Log()
+log = _global_log.log
+debug = _global_log.debug
+info = _global_log.info
+warn = _global_log.warn
+error = _global_log.error
+fatal = _global_log.fatal
+
+def set_threshold(level):
+    # return the old threshold for use from tests
+    old = _global_log.threshold
+    _global_log.threshold = level
+    return old
+
+def set_verbosity(v):
+    if v <= 0:
+        set_threshold(WARN)
+    elif v == 1:
+        set_threshold(INFO)
+    elif v >= 2:
+        set_threshold(DEBUG)
diff --git a/Lib/distutils/msvc9compiler.py b/Lib/distutils/msvc9compiler.py
new file mode 100644
index 000000000..211912762
--- /dev/null
+++ b/Lib/distutils/msvc9compiler.py
@@ -0,0 +1,791 @@
+"""distutils.msvc9compiler
+
+Contains MSVCCompiler, an implementation of the abstract CCompiler class
+for the Microsoft Visual Studio 2008.
+
+The module is compatible with VS 2005 and VS 2008. You can find legacy support
+for older versions of VS in distutils.msvccompiler.
+"""
+
+# Written by Perry Stoll
+# hacked by Robin Becker and Thomas Heller to do a better job of
+#   finding DevStudio (through the registry)
+# ported to VS2005 and VS 2008 by Christian Heimes
+
+import os
+import subprocess
+import sys
+import re
+
+from distutils.errors import DistutilsExecError, DistutilsPlatformError, \
+                             CompileError, LibError, LinkError
+from distutils.ccompiler import CCompiler, gen_preprocess_options, \
+                                gen_lib_options
+from distutils import log
+from distutils.util import get_platform
+
+import winreg
+
+RegOpenKeyEx = winreg.OpenKeyEx
+RegEnumKey = winreg.EnumKey
+RegEnumValue = winreg.EnumValue
+RegError = winreg.error
+
+HKEYS = (winreg.HKEY_USERS,
+         winreg.HKEY_CURRENT_USER,
+         winreg.HKEY_LOCAL_MACHINE,
+         winreg.HKEY_CLASSES_ROOT)
+
+NATIVE_WIN64 = (sys.platform == 'win32' and sys.maxsize > 2**32)
+if NATIVE_WIN64:
+    # Visual C++ is a 32-bit application, so we need to look in
+    # the corresponding registry branch, if we're running a
+    # 64-bit Python on Win64
+    VS_BASE = r"Software\Wow6432Node\Microsoft\VisualStudio\%0.1f"
+    WINSDK_BASE = r"Software\Wow6432Node\Microsoft\Microsoft SDKs\Windows"
+    NET_BASE = r"Software\Wow6432Node\Microsoft\.NETFramework"
+else:
+    VS_BASE = r"Software\Microsoft\VisualStudio\%0.1f"
+    WINSDK_BASE = r"Software\Microsoft\Microsoft SDKs\Windows"
+    NET_BASE = r"Software\Microsoft\.NETFramework"
+
+# A map keyed by get_platform() return values to values accepted by
+# 'vcvarsall.bat'.  Note a cross-compile may combine these (eg, 'x86_amd64' is
+# the param to cross-compile on x86 targeting amd64.)
+PLAT_TO_VCVARS = {
+    'win32' : 'x86',
+    'win-amd64' : 'amd64',
+    'win-ia64' : 'ia64',
+}
+
+class Reg:
+    """Helper class to read values from the registry
+    """
+
+    def get_value(cls, path, key):
+        for base in HKEYS:
+            d = cls.read_values(base, path)
+            if d and key in d:
+                return d[key]
+        raise KeyError(key)
+    get_value = classmethod(get_value)
+
+    def read_keys(cls, base, key):
+        """Return list of registry keys."""
+        try:
+            handle = RegOpenKeyEx(base, key)
+        except RegError:
+            return None
+        L = []
+        i = 0
+        while True:
+            try:
+                k = RegEnumKey(handle, i)
+            except RegError:
+                break
+            L.append(k)
+            i += 1
+        return L
+    read_keys = classmethod(read_keys)
+
+    def read_values(cls, base, key):
+        """Return dict of registry keys and values.
+
+        All names are converted to lowercase.
+        """
+        try:
+            handle = RegOpenKeyEx(base, key)
+        except RegError:
+            return None
+        d = {}
+        i = 0
+        while True:
+            try:
+                name, value, type = RegEnumValue(handle, i)
+            except RegError:
+                break
+            name = name.lower()
+            d[cls.convert_mbcs(name)] = cls.convert_mbcs(value)
+            i += 1
+        return d
+    read_values = classmethod(read_values)
+
+    def convert_mbcs(s):
+        dec = getattr(s, "decode", None)
+        if dec is not None:
+            try:
+                s = dec("mbcs")
+            except UnicodeError:
+                pass
+        return s
+    convert_mbcs = staticmethod(convert_mbcs)
+
+class MacroExpander:
+
+    def __init__(self, version):
+        self.macros = {}
+        self.vsbase = VS_BASE % version
+        self.load_macros(version)
+
+    def set_macro(self, macro, path, key):
+        self.macros["$(%s)" % macro] = Reg.get_value(path, key)
+
+    def load_macros(self, version):
+        self.set_macro("VCInstallDir", self.vsbase + r"\Setup\VC", "productdir")
+        self.set_macro("VSInstallDir", self.vsbase + r"\Setup\VS", "productdir")
+        self.set_macro("FrameworkDir", NET_BASE, "installroot")
+        try:
+            if version >= 8.0:
+                self.set_macro("FrameworkSDKDir", NET_BASE,
+                               "sdkinstallrootv2.0")
+            else:
+                raise KeyError("sdkinstallrootv2.0")
+        except KeyError:
+            raise DistutilsPlatformError(
+            """Python was built with Visual Studio 2008;
+extensions must be built with a compiler than can generate compatible binaries.
+Visual Studio 2008 was not found on this system. If you have Cygwin installed,
+you can try compiling with MingW32, by passing "-c mingw32" to setup.py.""")
+
+        if version >= 9.0:
+            self.set_macro("FrameworkVersion", self.vsbase, "clr version")
+            self.set_macro("WindowsSdkDir", WINSDK_BASE, "currentinstallfolder")
+        else:
+            p = r"Software\Microsoft\NET Framework Setup\Product"
+            for base in HKEYS:
+                try:
+                    h = RegOpenKeyEx(base, p)
+                except RegError:
+                    continue
+                key = RegEnumKey(h, 0)
+                d = Reg.get_value(base, r"%s\%s" % (p, key))
+                self.macros["$(FrameworkVersion)"] = d["version"]
+
+    def sub(self, s):
+        for k, v in self.macros.items():
+            s = s.replace(k, v)
+        return s
+
+def get_build_version():
+    """Return the version of MSVC that was used to build Python.
+
+    For Python 2.3 and up, the version number is included in
+    sys.version.  For earlier versions, assume the compiler is MSVC 6.
+    """
+    prefix = "MSC v."
+    i = sys.version.find(prefix)
+    if i == -1:
+        return 6
+    i = i + len(prefix)
+    s, rest = sys.version[i:].split(" ", 1)
+    majorVersion = int(s[:-2]) - 6
+    if majorVersion >= 13:
+        # v13 was skipped and should be v14
+        majorVersion += 1
+    minorVersion = int(s[2:3]) / 10.0
+    # I don't think paths are affected by minor version in version 6
+    if majorVersion == 6:
+        minorVersion = 0
+    if majorVersion >= 6:
+        return majorVersion + minorVersion
+    # else we don't know what version of the compiler this is
+    return None
+
+def normalize_and_reduce_paths(paths):
+    """Return a list of normalized paths with duplicates removed.
+
+    The current order of paths is maintained.
+    """
+    # Paths are normalized so things like:  /a and /a/ aren't both preserved.
+    reduced_paths = []
+    for p in paths:
+        np = os.path.normpath(p)
+        # XXX(nnorwitz): O(n**2), if reduced_paths gets long perhaps use a set.
+        if np not in reduced_paths:
+            reduced_paths.append(np)
+    return reduced_paths
+
+def removeDuplicates(variable):
+    """Remove duplicate values of an environment variable.
+    """
+    oldList = variable.split(os.pathsep)
+    newList = []
+    for i in oldList:
+        if i not in newList:
+            newList.append(i)
+    newVariable = os.pathsep.join(newList)
+    return newVariable
+
+def find_vcvarsall(version):
+    """Find the vcvarsall.bat file
+
+    At first it tries to find the productdir of VS 2008 in the registry. If
+    that fails it falls back to the VS90COMNTOOLS env var.
+    """
+    vsbase = VS_BASE % version
+    try:
+        productdir = Reg.get_value(r"%s\Setup\VC" % vsbase,
+                                   "productdir")
+    except KeyError:
+        log.debug("Unable to find productdir in registry")
+        productdir = None
+
+    if not productdir or not os.path.isdir(productdir):
+        toolskey = "VS%0.f0COMNTOOLS" % version
+        toolsdir = os.environ.get(toolskey, None)
+
+        if toolsdir and os.path.isdir(toolsdir):
+            productdir = os.path.join(toolsdir, os.pardir, os.pardir, "VC")
+            productdir = os.path.abspath(productdir)
+            if not os.path.isdir(productdir):
+                log.debug("%s is not a valid directory" % productdir)
+                return None
+        else:
+            log.debug("Env var %s is not set or invalid" % toolskey)
+    if not productdir:
+        log.debug("No productdir found")
+        return None
+    vcvarsall = os.path.join(productdir, "vcvarsall.bat")
+    if os.path.isfile(vcvarsall):
+        return vcvarsall
+    log.debug("Unable to find vcvarsall.bat")
+    return None
+
+def query_vcvarsall(version, arch="x86"):
+    """Launch vcvarsall.bat and read the settings from its environment
+    """
+    vcvarsall = find_vcvarsall(version)
+    interesting = set(("include", "lib", "libpath", "path"))
+    result = {}
+
+    if vcvarsall is None:
+        raise DistutilsPlatformError("Unable to find vcvarsall.bat")
+    log.debug("Calling 'vcvarsall.bat %s' (version=%s)", arch, version)
+    popen = subprocess.Popen('"%s" %s & set' % (vcvarsall, arch),
+                             stdout=subprocess.PIPE,
+                             stderr=subprocess.PIPE)
+    try:
+        stdout, stderr = popen.communicate()
+        if popen.wait() != 0:
+            raise DistutilsPlatformError(stderr.decode("mbcs"))
+
+        stdout = stdout.decode("mbcs")
+        for line in stdout.split("\n"):
+            line = Reg.convert_mbcs(line)
+            if '=' not in line:
+                continue
+            line = line.strip()
+            key, value = line.split('=', 1)
+            key = key.lower()
+            if key in interesting:
+                if value.endswith(os.pathsep):
+                    value = value[:-1]
+                result[key] = removeDuplicates(value)
+
+    finally:
+        popen.stdout.close()
+        popen.stderr.close()
+
+    if len(result) != len(interesting):
+        raise ValueError(str(list(result.keys())))
+
+    return result
+
+# More globals
+VERSION = get_build_version()
+if VERSION < 8.0:
+    raise DistutilsPlatformError("VC %0.1f is not supported by this module" % VERSION)
+# MACROS = MacroExpander(VERSION)
+
+class MSVCCompiler(CCompiler) :
+    """Concrete class that implements an interface to Microsoft Visual C++,
+       as defined by the CCompiler abstract class."""
+
+    compiler_type = 'msvc'
+
+    # Just set this so CCompiler's constructor doesn't barf.  We currently
+    # don't use the 'set_executables()' bureaucracy provided by CCompiler,
+    # as it really isn't necessary for this sort of single-compiler class.
+    # Would be nice to have a consistent interface with UnixCCompiler,
+    # though, so it's worth thinking about.
+    executables = {}
+
+    # Private class data (need to distinguish C from C++ source for compiler)
+    _c_extensions = ['.c']
+    _cpp_extensions = ['.cc', '.cpp', '.cxx']
+    _rc_extensions = ['.rc']
+    _mc_extensions = ['.mc']
+
+    # Needed for the filename generation methods provided by the
+    # base class, CCompiler.
+    src_extensions = (_c_extensions + _cpp_extensions +
+                      _rc_extensions + _mc_extensions)
+    res_extension = '.res'
+    obj_extension = '.obj'
+    static_lib_extension = '.lib'
+    shared_lib_extension = '.dll'
+    static_lib_format = shared_lib_format = '%s%s'
+    exe_extension = '.exe'
+
+    def __init__(self, verbose=0, dry_run=0, force=0):
+        CCompiler.__init__ (self, verbose, dry_run, force)
+        self.__version = VERSION
+        self.__root = r"Software\Microsoft\VisualStudio"
+        # self.__macros = MACROS
+        self.__paths = []
+        # target platform (.plat_name is consistent with 'bdist')
+        self.plat_name = None
+        self.__arch = None # deprecated name
+        self.initialized = False
+
+    def initialize(self, plat_name=None):
+        # multi-init means we would need to check platform same each time...
+        assert not self.initialized, "don't init multiple times"
+        if plat_name is None:
+            plat_name = get_platform()
+        # sanity check for platforms to prevent obscure errors later.
+        ok_plats = 'win32', 'win-amd64', 'win-ia64'
+        if plat_name not in ok_plats:
+            raise DistutilsPlatformError("--plat-name must be one of %s" %
+                                         (ok_plats,))
+
+        if "DISTUTILS_USE_SDK" in os.environ and "MSSdk" in os.environ and self.find_exe("cl.exe"):
+            # Assume that the SDK set up everything alright; don't try to be
+            # smarter
+            self.cc = "cl.exe"
+            self.linker = "link.exe"
+            self.lib = "lib.exe"
+            self.rc = "rc.exe"
+            self.mc = "mc.exe"
+        else:
+            # On x86, 'vcvars32.bat amd64' creates an env that doesn't work;
+            # to cross compile, you use 'x86_amd64'.
+            # On AMD64, 'vcvars32.bat amd64' is a native build env; to cross
+            # compile use 'x86' (ie, it runs the x86 compiler directly)
+            # No idea how itanium handles this, if at all.
+            if plat_name == get_platform() or plat_name == 'win32':
+                # native build or cross-compile to win32
+                plat_spec = PLAT_TO_VCVARS[plat_name]
+            else:
+                # cross compile from win32 -> some 64bit
+                plat_spec = PLAT_TO_VCVARS[get_platform()] + '_' + \
+                            PLAT_TO_VCVARS[plat_name]
+
+            vc_env = query_vcvarsall(VERSION, plat_spec)
+
+            self.__paths = vc_env['path'].split(os.pathsep)
+            os.environ['lib'] = vc_env['lib']
+            os.environ['include'] = vc_env['include']
+
+            if len(self.__paths) == 0:
+                raise DistutilsPlatformError("Python was built with %s, "
+                       "and extensions need to be built with the same "
+                       "version of the compiler, but it isn't installed."
+                       % self.__product)
+
+            self.cc = self.find_exe("cl.exe")
+            self.linker = self.find_exe("link.exe")
+            self.lib = self.find_exe("lib.exe")
+            self.rc = self.find_exe("rc.exe")   # resource compiler
+            self.mc = self.find_exe("mc.exe")   # message compiler
+            #self.set_path_env_var('lib')
+            #self.set_path_env_var('include')
+
+        # extend the MSVC path with the current path
+        try:
+            for p in os.environ['path'].split(';'):
+                self.__paths.append(p)
+        except KeyError:
+            pass
+        self.__paths = normalize_and_reduce_paths(self.__paths)
+        os.environ['path'] = ";".join(self.__paths)
+
+        self.preprocess_options = None
+        if self.__arch == "x86":
+            self.compile_options = [ '/nologo', '/Ox', '/MD', '/W3',
+                                     '/DNDEBUG']
+            self.compile_options_debug = ['/nologo', '/Od', '/MDd', '/W3',
+                                          '/Z7', '/D_DEBUG']
+        else:
+            # Win64
+            self.compile_options = [ '/nologo', '/Ox', '/MD', '/W3', '/GS-' ,
+                                     '/DNDEBUG']
+            self.compile_options_debug = ['/nologo', '/Od', '/MDd', '/W3', '/GS-',
+                                          '/Z7', '/D_DEBUG']
+
+        self.ldflags_shared = ['/DLL', '/nologo', '/INCREMENTAL:NO']
+        if self.__version >= 7:
+            self.ldflags_shared_debug = [
+                '/DLL', '/nologo', '/INCREMENTAL:no', '/DEBUG'
+                ]
+        self.ldflags_static = [ '/nologo']
+
+        self.initialized = True
+
+    # -- Worker methods ------------------------------------------------
+
+    def object_filenames(self,
+                         source_filenames,
+                         strip_dir=0,
+                         output_dir=''):
+        # Copied from ccompiler.py, extended to return .res as 'object'-file
+        # for .rc input file
+        if output_dir is None: output_dir = ''
+        obj_names = []
+        for src_name in source_filenames:
+            (base, ext) = os.path.splitext (src_name)
+            base = os.path.splitdrive(base)[1] # Chop off the drive
+            base = base[os.path.isabs(base):]  # If abs, chop off leading /
+            if ext not in self.src_extensions:
+                # Better to raise an exception instead of silently continuing
+                # and later complain about sources and targets having
+                # different lengths
+                raise CompileError ("Don't know how to compile %s" % src_name)
+            if strip_dir:
+                base = os.path.basename (base)
+            if ext in self._rc_extensions:
+                obj_names.append (os.path.join (output_dir,
+                                                base + self.res_extension))
+            elif ext in self._mc_extensions:
+                obj_names.append (os.path.join (output_dir,
+                                                base + self.res_extension))
+            else:
+                obj_names.append (os.path.join (output_dir,
+                                                base + self.obj_extension))
+        return obj_names
+
+
+    def compile(self, sources,
+                output_dir=None, macros=None, include_dirs=None, debug=0,
+                extra_preargs=None, extra_postargs=None, depends=None):
+
+        if not self.initialized:
+            self.initialize()
+        compile_info = self._setup_compile(output_dir, macros, include_dirs,
+                                           sources, depends, extra_postargs)
+        macros, objects, extra_postargs, pp_opts, build = compile_info
+
+        compile_opts = extra_preargs or []
+        compile_opts.append ('/c')
+        if debug:
+            compile_opts.extend(self.compile_options_debug)
+        else:
+            compile_opts.extend(self.compile_options)
+
+        for obj in objects:
+            try:
+                src, ext = build[obj]
+            except KeyError:
+                continue
+            if debug:
+                # pass the full pathname to MSVC in debug mode,
+                # this allows the debugger to find the source file
+                # without asking the user to browse for it
+                src = os.path.abspath(src)
+
+            if ext in self._c_extensions:
+                input_opt = "/Tc" + src
+            elif ext in self._cpp_extensions:
+                input_opt = "/Tp" + src
+            elif ext in self._rc_extensions:
+                # compile .RC to .RES file
+                input_opt = src
+                output_opt = "/fo" + obj
+                try:
+                    self.spawn([self.rc] + pp_opts +
+                               [output_opt] + [input_opt])
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue
+            elif ext in self._mc_extensions:
+                # Compile .MC to .RC file to .RES file.
+                #   * '-h dir' specifies the directory for the
+                #     generated include file
+                #   * '-r dir' specifies the target directory of the
+                #     generated RC file and the binary message resource
+                #     it includes
+                #
+                # For now (since there are no options to change this),
+                # we use the source-directory for the include file and
+                # the build directory for the RC file and message
+                # resources. This works at least for win32all.
+                h_dir = os.path.dirname(src)
+                rc_dir = os.path.dirname(obj)
+                try:
+                    # first compile .MC to .RC and .H file
+                    self.spawn([self.mc] +
+                               ['-h', h_dir, '-r', rc_dir] + [src])
+                    base, _ = os.path.splitext (os.path.basename (src))
+                    rc_file = os.path.join (rc_dir, base + '.rc')
+                    # then compile .RC to .RES file
+                    self.spawn([self.rc] +
+                               ["/fo" + obj] + [rc_file])
+
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue
+            else:
+                # how to handle this file?
+                raise CompileError("Don't know how to compile %s to %s"
+                                   % (src, obj))
+
+            output_opt = "/Fo" + obj
+            try:
+                self.spawn([self.cc] + compile_opts + pp_opts +
+                           [input_opt, output_opt] +
+                           extra_postargs)
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+
+        return objects
+
+
+    def create_static_lib(self,
+                          objects,
+                          output_libname,
+                          output_dir=None,
+                          debug=0,
+                          target_lang=None):
+
+        if not self.initialized:
+            self.initialize()
+        (objects, output_dir) = self._fix_object_args(objects, output_dir)
+        output_filename = self.library_filename(output_libname,
+                                                output_dir=output_dir)
+
+        if self._need_link(objects, output_filename):
+            lib_args = objects + ['/OUT:' + output_filename]
+            if debug:
+                pass # XXX what goes here?
+            try:
+                self.spawn([self.lib] + lib_args)
+            except DistutilsExecError as msg:
+                raise LibError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+
+    def link(self,
+             target_desc,
+             objects,
+             output_filename,
+             output_dir=None,
+             libraries=None,
+             library_dirs=None,
+             runtime_library_dirs=None,
+             export_symbols=None,
+             debug=0,
+             extra_preargs=None,
+             extra_postargs=None,
+             build_temp=None,
+             target_lang=None):
+
+        if not self.initialized:
+            self.initialize()
+        (objects, output_dir) = self._fix_object_args(objects, output_dir)
+        fixed_args = self._fix_lib_args(libraries, library_dirs,
+                                        runtime_library_dirs)
+        (libraries, library_dirs, runtime_library_dirs) = fixed_args
+
+        if runtime_library_dirs:
+            self.warn ("I don't know what to do with 'runtime_library_dirs': "
+                       + str (runtime_library_dirs))
+
+        lib_opts = gen_lib_options(self,
+                                   library_dirs, runtime_library_dirs,
+                                   libraries)
+        if output_dir is not None:
+            output_filename = os.path.join(output_dir, output_filename)
+
+        if self._need_link(objects, output_filename):
+            if target_desc == CCompiler.EXECUTABLE:
+                if debug:
+                    ldflags = self.ldflags_shared_debug[1:]
+                else:
+                    ldflags = self.ldflags_shared[1:]
+            else:
+                if debug:
+                    ldflags = self.ldflags_shared_debug
+                else:
+                    ldflags = self.ldflags_shared
+
+            export_opts = []
+            for sym in (export_symbols or []):
+                export_opts.append("/EXPORT:" + sym)
+
+            ld_args = (ldflags + lib_opts + export_opts +
+                       objects + ['/OUT:' + output_filename])
+
+            # The MSVC linker generates .lib and .exp files, which cannot be
+            # suppressed by any linker switches. The .lib files may even be
+            # needed! Make sure they are generated in the temporary build
+            # directory. Since they have different names for debug and release
+            # builds, they can go into the same directory.
+            build_temp = os.path.dirname(objects[0])
+            if export_symbols is not None:
+                (dll_name, dll_ext) = os.path.splitext(
+                    os.path.basename(output_filename))
+                implib_file = os.path.join(
+                    build_temp,
+                    self.library_filename(dll_name))
+                ld_args.append ('/IMPLIB:' + implib_file)
+
+            self.manifest_setup_ldargs(output_filename, build_temp, ld_args)
+
+            if extra_preargs:
+                ld_args[:0] = extra_preargs
+            if extra_postargs:
+                ld_args.extend(extra_postargs)
+
+            self.mkpath(os.path.dirname(output_filename))
+            try:
+                self.spawn([self.linker] + ld_args)
+            except DistutilsExecError as msg:
+                raise LinkError(msg)
+
+            # embed the manifest
+            # XXX - this is somewhat fragile - if mt.exe fails, distutils
+            # will still consider the DLL up-to-date, but it will not have a
+            # manifest.  Maybe we should link to a temp file?  OTOH, that
+            # implies a build environment error that shouldn't go undetected.
+            mfinfo = self.manifest_get_embed_info(target_desc, ld_args)
+            if mfinfo is not None:
+                mffilename, mfid = mfinfo
+                out_arg = '-outputresource:%s;%s' % (output_filename, mfid)
+                try:
+                    self.spawn(['mt.exe', '-nologo', '-manifest',
+                                mffilename, out_arg])
+                except DistutilsExecError as msg:
+                    raise LinkError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+    def manifest_setup_ldargs(self, output_filename, build_temp, ld_args):
+        # If we need a manifest at all, an embedded manifest is recommended.
+        # See MSDN article titled
+        # "How to: Embed a Manifest Inside a C/C++ Application"
+        # (currently at http://msdn2.microsoft.com/en-us/library/ms235591(VS.80).aspx)
+        # Ask the linker to generate the manifest in the temp dir, so
+        # we can check it, and possibly embed it, later.
+        temp_manifest = os.path.join(
+                build_temp,
+                os.path.basename(output_filename) + ".manifest")
+        ld_args.append('/MANIFESTFILE:' + temp_manifest)
+
+    def manifest_get_embed_info(self, target_desc, ld_args):
+        # If a manifest should be embedded, return a tuple of
+        # (manifest_filename, resource_id).  Returns None if no manifest
+        # should be embedded.  See http://bugs.python.org/issue7833 for why
+        # we want to avoid any manifest for extension modules if we can)
+        for arg in ld_args:
+            if arg.startswith("/MANIFESTFILE:"):
+                temp_manifest = arg.split(":", 1)[1]
+                break
+        else:
+            # no /MANIFESTFILE so nothing to do.
+            return None
+        if target_desc == CCompiler.EXECUTABLE:
+            # by default, executables always get the manifest with the
+            # CRT referenced.
+            mfid = 1
+        else:
+            # Extension modules try and avoid any manifest if possible.
+            mfid = 2
+            temp_manifest = self._remove_visual_c_ref(temp_manifest)
+        if temp_manifest is None:
+            return None
+        return temp_manifest, mfid
+
+    def _remove_visual_c_ref(self, manifest_file):
+        try:
+            # Remove references to the Visual C runtime, so they will
+            # fall through to the Visual C dependency of Python.exe.
+            # This way, when installed for a restricted user (e.g.
+            # runtimes are not in WinSxS folder, but in Python's own
+            # folder), the runtimes do not need to be in every folder
+            # with .pyd's.
+            # Returns either the filename of the modified manifest or
+            # None if no manifest should be embedded.
+            manifest_f = open(manifest_file)
+            try:
+                manifest_buf = manifest_f.read()
+            finally:
+                manifest_f.close()
+            pattern = re.compile(
+                r"""<assemblyIdentity.*?name=("|')Microsoft\."""\
+                r"""VC\d{2}\.CRT("|').*?(/>|</assemblyIdentity>)""",
+                re.DOTALL)
+            manifest_buf = re.sub(pattern, "", manifest_buf)
+            pattern = r"<dependentAssembly>\s*</dependentAssembly>"
+            manifest_buf = re.sub(pattern, "", manifest_buf)
+            # Now see if any other assemblies are referenced - if not, we
+            # don't want a manifest embedded.
+            pattern = re.compile(
+                r"""<assemblyIdentity.*?name=(?:"|')(.+?)(?:"|')"""
+                r""".*?(?:/>|</assemblyIdentity>)""", re.DOTALL)
+            if re.search(pattern, manifest_buf) is None:
+                return None
+
+            manifest_f = open(manifest_file, 'w')
+            try:
+                manifest_f.write(manifest_buf)
+                return manifest_file
+            finally:
+                manifest_f.close()
+        except OSError:
+            pass
+
+    # -- Miscellaneous methods -----------------------------------------
+    # These are all used by the 'gen_lib_options() function, in
+    # ccompiler.py.
+
+    def library_dir_option(self, dir):
+        return "/LIBPATH:" + dir
+
+    def runtime_library_dir_option(self, dir):
+        raise DistutilsPlatformError(
+              "don't know how to set runtime library search path for MSVC++")
+
+    def library_option(self, lib):
+        return self.library_filename(lib)
+
+
+    def find_library_file(self, dirs, lib, debug=0):
+        # Prefer a debugging library if found (and requested), but deal
+        # with it if we don't have one.
+        if debug:
+            try_names = [lib + "_d", lib]
+        else:
+            try_names = [lib]
+        for dir in dirs:
+            for name in try_names:
+                libfile = os.path.join(dir, self.library_filename (name))
+                if os.path.exists(libfile):
+                    return libfile
+        else:
+            # Oops, didn't find it in *any* of 'dirs'
+            return None
+
+    # Helper methods for using the MSVC registry settings
+
+    def find_exe(self, exe):
+        """Return path to an MSVC executable program.
+
+        Tries to find the program in several places: first, one of the
+        MSVC program search paths from the registry; next, the directories
+        in the PATH environment variable.  If any of those work, return an
+        absolute path that is known to exist.  If none of them work, just
+        return the original program name, 'exe'.
+        """
+        for p in self.__paths:
+            fn = os.path.join(os.path.abspath(p), exe)
+            if os.path.isfile(fn):
+                return fn
+
+        # didn't find it; try existing path
+        for p in os.environ['Path'].split(';'):
+            fn = os.path.join(os.path.abspath(p),exe)
+            if os.path.isfile(fn):
+                return fn
+
+        return exe
diff --git a/Lib/distutils/msvccompiler.py b/Lib/distutils/msvccompiler.py
new file mode 100644
index 000000000..1048cd415
--- /dev/null
+++ b/Lib/distutils/msvccompiler.py
@@ -0,0 +1,643 @@
+"""distutils.msvccompiler
+
+Contains MSVCCompiler, an implementation of the abstract CCompiler class
+for the Microsoft Visual Studio.
+"""
+
+# Written by Perry Stoll
+# hacked by Robin Becker and Thomas Heller to do a better job of
+#   finding DevStudio (through the registry)
+
+import sys, os
+from distutils.errors import \
+     DistutilsExecError, DistutilsPlatformError, \
+     CompileError, LibError, LinkError
+from distutils.ccompiler import \
+     CCompiler, gen_preprocess_options, gen_lib_options
+from distutils import log
+
+_can_read_reg = False
+try:
+    import winreg
+
+    _can_read_reg = True
+    hkey_mod = winreg
+
+    RegOpenKeyEx = winreg.OpenKeyEx
+    RegEnumKey = winreg.EnumKey
+    RegEnumValue = winreg.EnumValue
+    RegError = winreg.error
+
+except ImportError:
+    try:
+        import win32api
+        import win32con
+        _can_read_reg = True
+        hkey_mod = win32con
+
+        RegOpenKeyEx = win32api.RegOpenKeyEx
+        RegEnumKey = win32api.RegEnumKey
+        RegEnumValue = win32api.RegEnumValue
+        RegError = win32api.error
+    except ImportError:
+        log.info("Warning: Can't read registry to find the "
+                 "necessary compiler setting\n"
+                 "Make sure that Python modules winreg, "
+                 "win32api or win32con are installed.")
+        pass
+
+if _can_read_reg:
+    HKEYS = (hkey_mod.HKEY_USERS,
+             hkey_mod.HKEY_CURRENT_USER,
+             hkey_mod.HKEY_LOCAL_MACHINE,
+             hkey_mod.HKEY_CLASSES_ROOT)
+
+def read_keys(base, key):
+    """Return list of registry keys."""
+    try:
+        handle = RegOpenKeyEx(base, key)
+    except RegError:
+        return None
+    L = []
+    i = 0
+    while True:
+        try:
+            k = RegEnumKey(handle, i)
+        except RegError:
+            break
+        L.append(k)
+        i += 1
+    return L
+
+def read_values(base, key):
+    """Return dict of registry keys and values.
+
+    All names are converted to lowercase.
+    """
+    try:
+        handle = RegOpenKeyEx(base, key)
+    except RegError:
+        return None
+    d = {}
+    i = 0
+    while True:
+        try:
+            name, value, type = RegEnumValue(handle, i)
+        except RegError:
+            break
+        name = name.lower()
+        d[convert_mbcs(name)] = convert_mbcs(value)
+        i += 1
+    return d
+
+def convert_mbcs(s):
+    dec = getattr(s, "decode", None)
+    if dec is not None:
+        try:
+            s = dec("mbcs")
+        except UnicodeError:
+            pass
+    return s
+
+class MacroExpander:
+    def __init__(self, version):
+        self.macros = {}
+        self.load_macros(version)
+
+    def set_macro(self, macro, path, key):
+        for base in HKEYS:
+            d = read_values(base, path)
+            if d:
+                self.macros["$(%s)" % macro] = d[key]
+                break
+
+    def load_macros(self, version):
+        vsbase = r"Software\Microsoft\VisualStudio\%0.1f" % version
+        self.set_macro("VCInstallDir", vsbase + r"\Setup\VC", "productdir")
+        self.set_macro("VSInstallDir", vsbase + r"\Setup\VS", "productdir")
+        net = r"Software\Microsoft\.NETFramework"
+        self.set_macro("FrameworkDir", net, "installroot")
+        try:
+            if version > 7.0:
+                self.set_macro("FrameworkSDKDir", net, "sdkinstallrootv1.1")
+            else:
+                self.set_macro("FrameworkSDKDir", net, "sdkinstallroot")
+        except KeyError as exc: #
+            raise DistutilsPlatformError(
+            """Python was built with Visual Studio 2003;
+extensions must be built with a compiler than can generate compatible binaries.
+Visual Studio 2003 was not found on this system. If you have Cygwin installed,
+you can try compiling with MingW32, by passing "-c mingw32" to setup.py.""")
+
+        p = r"Software\Microsoft\NET Framework Setup\Product"
+        for base in HKEYS:
+            try:
+                h = RegOpenKeyEx(base, p)
+            except RegError:
+                continue
+            key = RegEnumKey(h, 0)
+            d = read_values(base, r"%s\%s" % (p, key))
+            self.macros["$(FrameworkVersion)"] = d["version"]
+
+    def sub(self, s):
+        for k, v in self.macros.items():
+            s = s.replace(k, v)
+        return s
+
+def get_build_version():
+    """Return the version of MSVC that was used to build Python.
+
+    For Python 2.3 and up, the version number is included in
+    sys.version.  For earlier versions, assume the compiler is MSVC 6.
+    """
+    prefix = "MSC v."
+    i = sys.version.find(prefix)
+    if i == -1:
+        return 6
+    i = i + len(prefix)
+    s, rest = sys.version[i:].split(" ", 1)
+    majorVersion = int(s[:-2]) - 6
+    if majorVersion >= 13:
+        # v13 was skipped and should be v14
+        majorVersion += 1
+    minorVersion = int(s[2:3]) / 10.0
+    # I don't think paths are affected by minor version in version 6
+    if majorVersion == 6:
+        minorVersion = 0
+    if majorVersion >= 6:
+        return majorVersion + minorVersion
+    # else we don't know what version of the compiler this is
+    return None
+
+def get_build_architecture():
+    """Return the processor architecture.
+
+    Possible results are "Intel", "Itanium", or "AMD64".
+    """
+
+    prefix = " bit ("
+    i = sys.version.find(prefix)
+    if i == -1:
+        return "Intel"
+    j = sys.version.find(")", i)
+    return sys.version[i+len(prefix):j]
+
+def normalize_and_reduce_paths(paths):
+    """Return a list of normalized paths with duplicates removed.
+
+    The current order of paths is maintained.
+    """
+    # Paths are normalized so things like:  /a and /a/ aren't both preserved.
+    reduced_paths = []
+    for p in paths:
+        np = os.path.normpath(p)
+        # XXX(nnorwitz): O(n**2), if reduced_paths gets long perhaps use a set.
+        if np not in reduced_paths:
+            reduced_paths.append(np)
+    return reduced_paths
+
+
+class MSVCCompiler(CCompiler) :
+    """Concrete class that implements an interface to Microsoft Visual C++,
+       as defined by the CCompiler abstract class."""
+
+    compiler_type = 'msvc'
+
+    # Just set this so CCompiler's constructor doesn't barf.  We currently
+    # don't use the 'set_executables()' bureaucracy provided by CCompiler,
+    # as it really isn't necessary for this sort of single-compiler class.
+    # Would be nice to have a consistent interface with UnixCCompiler,
+    # though, so it's worth thinking about.
+    executables = {}
+
+    # Private class data (need to distinguish C from C++ source for compiler)
+    _c_extensions = ['.c']
+    _cpp_extensions = ['.cc', '.cpp', '.cxx']
+    _rc_extensions = ['.rc']
+    _mc_extensions = ['.mc']
+
+    # Needed for the filename generation methods provided by the
+    # base class, CCompiler.
+    src_extensions = (_c_extensions + _cpp_extensions +
+                      _rc_extensions + _mc_extensions)
+    res_extension = '.res'
+    obj_extension = '.obj'
+    static_lib_extension = '.lib'
+    shared_lib_extension = '.dll'
+    static_lib_format = shared_lib_format = '%s%s'
+    exe_extension = '.exe'
+
+    def __init__(self, verbose=0, dry_run=0, force=0):
+        CCompiler.__init__ (self, verbose, dry_run, force)
+        self.__version = get_build_version()
+        self.__arch = get_build_architecture()
+        if self.__arch == "Intel":
+            # x86
+            if self.__version >= 7:
+                self.__root = r"Software\Microsoft\VisualStudio"
+                self.__macros = MacroExpander(self.__version)
+            else:
+                self.__root = r"Software\Microsoft\Devstudio"
+            self.__product = "Visual Studio version %s" % self.__version
+        else:
+            # Win64. Assume this was built with the platform SDK
+            self.__product = "Microsoft SDK compiler %s" % (self.__version + 6)
+
+        self.initialized = False
+
+    def initialize(self):
+        self.__paths = []
+        if "DISTUTILS_USE_SDK" in os.environ and "MSSdk" in os.environ and self.find_exe("cl.exe"):
+            # Assume that the SDK set up everything alright; don't try to be
+            # smarter
+            self.cc = "cl.exe"
+            self.linker = "link.exe"
+            self.lib = "lib.exe"
+            self.rc = "rc.exe"
+            self.mc = "mc.exe"
+        else:
+            self.__paths = self.get_msvc_paths("path")
+
+            if len(self.__paths) == 0:
+                raise DistutilsPlatformError("Python was built with %s, "
+                       "and extensions need to be built with the same "
+                       "version of the compiler, but it isn't installed."
+                       % self.__product)
+
+            self.cc = self.find_exe("cl.exe")
+            self.linker = self.find_exe("link.exe")
+            self.lib = self.find_exe("lib.exe")
+            self.rc = self.find_exe("rc.exe")   # resource compiler
+            self.mc = self.find_exe("mc.exe")   # message compiler
+            self.set_path_env_var('lib')
+            self.set_path_env_var('include')
+
+        # extend the MSVC path with the current path
+        try:
+            for p in os.environ['path'].split(';'):
+                self.__paths.append(p)
+        except KeyError:
+            pass
+        self.__paths = normalize_and_reduce_paths(self.__paths)
+        os.environ['path'] = ";".join(self.__paths)
+
+        self.preprocess_options = None
+        if self.__arch == "Intel":
+            self.compile_options = [ '/nologo', '/Ox', '/MD', '/W3', '/GX' ,
+                                     '/DNDEBUG']
+            self.compile_options_debug = ['/nologo', '/Od', '/MDd', '/W3', '/GX',
+                                          '/Z7', '/D_DEBUG']
+        else:
+            # Win64
+            self.compile_options = [ '/nologo', '/Ox', '/MD', '/W3', '/GS-' ,
+                                     '/DNDEBUG']
+            self.compile_options_debug = ['/nologo', '/Od', '/MDd', '/W3', '/GS-',
+                                          '/Z7', '/D_DEBUG']
+
+        self.ldflags_shared = ['/DLL', '/nologo', '/INCREMENTAL:NO']
+        if self.__version >= 7:
+            self.ldflags_shared_debug = [
+                '/DLL', '/nologo', '/INCREMENTAL:no', '/DEBUG'
+                ]
+        else:
+            self.ldflags_shared_debug = [
+                '/DLL', '/nologo', '/INCREMENTAL:no', '/pdb:None', '/DEBUG'
+                ]
+        self.ldflags_static = [ '/nologo']
+
+        self.initialized = True
+
+    # -- Worker methods ------------------------------------------------
+
+    def object_filenames(self,
+                         source_filenames,
+                         strip_dir=0,
+                         output_dir=''):
+        # Copied from ccompiler.py, extended to return .res as 'object'-file
+        # for .rc input file
+        if output_dir is None: output_dir = ''
+        obj_names = []
+        for src_name in source_filenames:
+            (base, ext) = os.path.splitext (src_name)
+            base = os.path.splitdrive(base)[1] # Chop off the drive
+            base = base[os.path.isabs(base):]  # If abs, chop off leading /
+            if ext not in self.src_extensions:
+                # Better to raise an exception instead of silently continuing
+                # and later complain about sources and targets having
+                # different lengths
+                raise CompileError ("Don't know how to compile %s" % src_name)
+            if strip_dir:
+                base = os.path.basename (base)
+            if ext in self._rc_extensions:
+                obj_names.append (os.path.join (output_dir,
+                                                base + self.res_extension))
+            elif ext in self._mc_extensions:
+                obj_names.append (os.path.join (output_dir,
+                                                base + self.res_extension))
+            else:
+                obj_names.append (os.path.join (output_dir,
+                                                base + self.obj_extension))
+        return obj_names
+
+
+    def compile(self, sources,
+                output_dir=None, macros=None, include_dirs=None, debug=0,
+                extra_preargs=None, extra_postargs=None, depends=None):
+
+        if not self.initialized:
+            self.initialize()
+        compile_info = self._setup_compile(output_dir, macros, include_dirs,
+                                           sources, depends, extra_postargs)
+        macros, objects, extra_postargs, pp_opts, build = compile_info
+
+        compile_opts = extra_preargs or []
+        compile_opts.append ('/c')
+        if debug:
+            compile_opts.extend(self.compile_options_debug)
+        else:
+            compile_opts.extend(self.compile_options)
+
+        for obj in objects:
+            try:
+                src, ext = build[obj]
+            except KeyError:
+                continue
+            if debug:
+                # pass the full pathname to MSVC in debug mode,
+                # this allows the debugger to find the source file
+                # without asking the user to browse for it
+                src = os.path.abspath(src)
+
+            if ext in self._c_extensions:
+                input_opt = "/Tc" + src
+            elif ext in self._cpp_extensions:
+                input_opt = "/Tp" + src
+            elif ext in self._rc_extensions:
+                # compile .RC to .RES file
+                input_opt = src
+                output_opt = "/fo" + obj
+                try:
+                    self.spawn([self.rc] + pp_opts +
+                               [output_opt] + [input_opt])
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue
+            elif ext in self._mc_extensions:
+                # Compile .MC to .RC file to .RES file.
+                #   * '-h dir' specifies the directory for the
+                #     generated include file
+                #   * '-r dir' specifies the target directory of the
+                #     generated RC file and the binary message resource
+                #     it includes
+                #
+                # For now (since there are no options to change this),
+                # we use the source-directory for the include file and
+                # the build directory for the RC file and message
+                # resources. This works at least for win32all.
+                h_dir = os.path.dirname(src)
+                rc_dir = os.path.dirname(obj)
+                try:
+                    # first compile .MC to .RC and .H file
+                    self.spawn([self.mc] +
+                               ['-h', h_dir, '-r', rc_dir] + [src])
+                    base, _ = os.path.splitext (os.path.basename (src))
+                    rc_file = os.path.join (rc_dir, base + '.rc')
+                    # then compile .RC to .RES file
+                    self.spawn([self.rc] +
+                               ["/fo" + obj] + [rc_file])
+
+                except DistutilsExecError as msg:
+                    raise CompileError(msg)
+                continue
+            else:
+                # how to handle this file?
+                raise CompileError("Don't know how to compile %s to %s"
+                                   % (src, obj))
+
+            output_opt = "/Fo" + obj
+            try:
+                self.spawn([self.cc] + compile_opts + pp_opts +
+                           [input_opt, output_opt] +
+                           extra_postargs)
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+
+        return objects
+
+
+    def create_static_lib(self,
+                          objects,
+                          output_libname,
+                          output_dir=None,
+                          debug=0,
+                          target_lang=None):
+
+        if not self.initialized:
+            self.initialize()
+        (objects, output_dir) = self._fix_object_args(objects, output_dir)
+        output_filename = self.library_filename(output_libname,
+                                                output_dir=output_dir)
+
+        if self._need_link(objects, output_filename):
+            lib_args = objects + ['/OUT:' + output_filename]
+            if debug:
+                pass # XXX what goes here?
+            try:
+                self.spawn([self.lib] + lib_args)
+            except DistutilsExecError as msg:
+                raise LibError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+
+    def link(self,
+             target_desc,
+             objects,
+             output_filename,
+             output_dir=None,
+             libraries=None,
+             library_dirs=None,
+             runtime_library_dirs=None,
+             export_symbols=None,
+             debug=0,
+             extra_preargs=None,
+             extra_postargs=None,
+             build_temp=None,
+             target_lang=None):
+
+        if not self.initialized:
+            self.initialize()
+        (objects, output_dir) = self._fix_object_args(objects, output_dir)
+        fixed_args = self._fix_lib_args(libraries, library_dirs,
+                                        runtime_library_dirs)
+        (libraries, library_dirs, runtime_library_dirs) = fixed_args
+
+        if runtime_library_dirs:
+            self.warn ("I don't know what to do with 'runtime_library_dirs': "
+                       + str (runtime_library_dirs))
+
+        lib_opts = gen_lib_options(self,
+                                   library_dirs, runtime_library_dirs,
+                                   libraries)
+        if output_dir is not None:
+            output_filename = os.path.join(output_dir, output_filename)
+
+        if self._need_link(objects, output_filename):
+            if target_desc == CCompiler.EXECUTABLE:
+                if debug:
+                    ldflags = self.ldflags_shared_debug[1:]
+                else:
+                    ldflags = self.ldflags_shared[1:]
+            else:
+                if debug:
+                    ldflags = self.ldflags_shared_debug
+                else:
+                    ldflags = self.ldflags_shared
+
+            export_opts = []
+            for sym in (export_symbols or []):
+                export_opts.append("/EXPORT:" + sym)
+
+            ld_args = (ldflags + lib_opts + export_opts +
+                       objects + ['/OUT:' + output_filename])
+
+            # The MSVC linker generates .lib and .exp files, which cannot be
+            # suppressed by any linker switches. The .lib files may even be
+            # needed! Make sure they are generated in the temporary build
+            # directory. Since they have different names for debug and release
+            # builds, they can go into the same directory.
+            if export_symbols is not None:
+                (dll_name, dll_ext) = os.path.splitext(
+                    os.path.basename(output_filename))
+                implib_file = os.path.join(
+                    os.path.dirname(objects[0]),
+                    self.library_filename(dll_name))
+                ld_args.append ('/IMPLIB:' + implib_file)
+
+            if extra_preargs:
+                ld_args[:0] = extra_preargs
+            if extra_postargs:
+                ld_args.extend(extra_postargs)
+
+            self.mkpath(os.path.dirname(output_filename))
+            try:
+                self.spawn([self.linker] + ld_args)
+            except DistutilsExecError as msg:
+                raise LinkError(msg)
+
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+
+    # -- Miscellaneous methods -----------------------------------------
+    # These are all used by the 'gen_lib_options() function, in
+    # ccompiler.py.
+
+    def library_dir_option(self, dir):
+        return "/LIBPATH:" + dir
+
+    def runtime_library_dir_option(self, dir):
+        raise DistutilsPlatformError(
+              "don't know how to set runtime library search path for MSVC++")
+
+    def library_option(self, lib):
+        return self.library_filename(lib)
+
+
+    def find_library_file(self, dirs, lib, debug=0):
+        # Prefer a debugging library if found (and requested), but deal
+        # with it if we don't have one.
+        if debug:
+            try_names = [lib + "_d", lib]
+        else:
+            try_names = [lib]
+        for dir in dirs:
+            for name in try_names:
+                libfile = os.path.join(dir, self.library_filename (name))
+                if os.path.exists(libfile):
+                    return libfile
+        else:
+            # Oops, didn't find it in *any* of 'dirs'
+            return None
+
+    # Helper methods for using the MSVC registry settings
+
+    def find_exe(self, exe):
+        """Return path to an MSVC executable program.
+
+        Tries to find the program in several places: first, one of the
+        MSVC program search paths from the registry; next, the directories
+        in the PATH environment variable.  If any of those work, return an
+        absolute path that is known to exist.  If none of them work, just
+        return the original program name, 'exe'.
+        """
+        for p in self.__paths:
+            fn = os.path.join(os.path.abspath(p), exe)
+            if os.path.isfile(fn):
+                return fn
+
+        # didn't find it; try existing path
+        for p in os.environ['Path'].split(';'):
+            fn = os.path.join(os.path.abspath(p),exe)
+            if os.path.isfile(fn):
+                return fn
+
+        return exe
+
+    def get_msvc_paths(self, path, platform='x86'):
+        """Get a list of devstudio directories (include, lib or path).
+
+        Return a list of strings.  The list will be empty if unable to
+        access the registry or appropriate registry keys not found.
+        """
+        if not _can_read_reg:
+            return []
+
+        path = path + " dirs"
+        if self.__version >= 7:
+            key = (r"%s\%0.1f\VC\VC_OBJECTS_PLATFORM_INFO\Win32\Directories"
+                   % (self.__root, self.__version))
+        else:
+            key = (r"%s\6.0\Build System\Components\Platforms"
+                   r"\Win32 (%s)\Directories" % (self.__root, platform))
+
+        for base in HKEYS:
+            d = read_values(base, key)
+            if d:
+                if self.__version >= 7:
+                    return self.__macros.sub(d[path]).split(";")
+                else:
+                    return d[path].split(";")
+        # MSVC 6 seems to create the registry entries we need only when
+        # the GUI is run.
+        if self.__version == 6:
+            for base in HKEYS:
+                if read_values(base, r"%s\6.0" % self.__root) is not None:
+                    self.warn("It seems you have Visual Studio 6 installed, "
+                        "but the expected registry settings are not present.\n"
+                        "You must at least run the Visual Studio GUI once "
+                        "so that these entries are created.")
+                    break
+        return []
+
+    def set_path_env_var(self, name):
+        """Set environment variable 'name' to an MSVC path type value.
+
+        This is equivalent to a SET command prior to execution of spawned
+        commands.
+        """
+
+        if name == "lib":
+            p = self.get_msvc_paths("library")
+        else:
+            p = self.get_msvc_paths(name)
+        if p:
+            os.environ[name] = ';'.join(p)
+
+
+if get_build_version() >= 8.0:
+    log.debug("Importing new compiler from distutils.msvc9compiler")
+    OldMSVCCompiler = MSVCCompiler
+    from distutils.msvc9compiler import MSVCCompiler
+    # get_build_architecture not really relevant now we support cross-compile
+    from distutils.msvc9compiler import MacroExpander
diff --git a/Lib/distutils/spawn.py b/Lib/distutils/spawn.py
new file mode 100644
index 000000000..538768809
--- /dev/null
+++ b/Lib/distutils/spawn.py
@@ -0,0 +1,192 @@
+"""distutils.spawn
+
+Provides the 'spawn()' function, a front-end to various platform-
+specific functions for launching another program in a sub-process.
+Also provides the 'find_executable()' to search the path for a given
+executable name.
+"""
+
+import sys
+import os
+
+from distutils.errors import DistutilsPlatformError, DistutilsExecError
+from distutils.debug import DEBUG
+from distutils import log
+
+def spawn(cmd, search_path=1, verbose=0, dry_run=0):
+    """Run another program, specified as a command list 'cmd', in a new process.
+
+    'cmd' is just the argument list for the new process, ie.
+    cmd[0] is the program to run and cmd[1:] are the rest of its arguments.
+    There is no way to run a program with a name different from that of its
+    executable.
+
+    If 'search_path' is true (the default), the system's executable
+    search path will be used to find the program; otherwise, cmd[0]
+    must be the exact path to the executable.  If 'dry_run' is true,
+    the command will not actually be run.
+
+    Raise DistutilsExecError if running the program fails in any way; just
+    return on success.
+    """
+    # cmd is documented as a list, but just in case some code passes a tuple
+    # in, protect our %-formatting code against horrible death
+    cmd = list(cmd)
+    if os.name == 'posix':
+        _spawn_posix(cmd, search_path, dry_run=dry_run)
+    elif os.name == 'nt':
+        _spawn_nt(cmd, search_path, dry_run=dry_run)
+    else:
+        raise DistutilsPlatformError(
+              "don't know how to spawn programs on platform '%s'" % os.name)
+
+def _nt_quote_args(args):
+    """Quote command-line arguments for DOS/Windows conventions.
+
+    Just wraps every argument which contains blanks in double quotes, and
+    returns a new argument list.
+    """
+    # XXX this doesn't seem very robust to me -- but if the Windows guys
+    # say it'll work, I guess I'll have to accept it.  (What if an arg
+    # contains quotes?  What other magic characters, other than spaces,
+    # have to be escaped?  Is there an escaping mechanism other than
+    # quoting?)
+    for i, arg in enumerate(args):
+        if ' ' in arg:
+            args[i] = '"%s"' % arg
+    return args
+
+def _spawn_nt(cmd, search_path=1, verbose=0, dry_run=0):
+    executable = cmd[0]
+    cmd = _nt_quote_args(cmd)
+    if search_path:
+        # either we find one or it stays the same
+        executable = find_executable(executable) or executable
+    log.info(' '.join([executable] + cmd[1:]))
+    if not dry_run:
+        # spawn for NT requires a full path to the .exe
+        try:
+            rc = os.spawnv(os.P_WAIT, executable, cmd)
+        except OSError as exc:
+            # this seems to happen when the command isn't found
+            if not DEBUG:
+                cmd = executable
+            raise DistutilsExecError(
+                  "command %r failed: %s" % (cmd, exc.args[-1]))
+        if rc != 0:
+            # and this reflects the command running but failing
+            if not DEBUG:
+                cmd = executable
+            raise DistutilsExecError(
+                  "command %r failed with exit status %d" % (cmd, rc))
+
+if sys.platform == 'darwin':
+    from distutils import sysconfig
+    _cfg_target = None
+    _cfg_target_split = None
+
+def _spawn_posix(cmd, search_path=1, verbose=0, dry_run=0):
+    log.info(' '.join(cmd))
+    if dry_run:
+        return
+    executable = cmd[0]
+    exec_fn = search_path and os.execvp or os.execv
+    env = None
+    if sys.platform == 'darwin':
+        global _cfg_target, _cfg_target_split
+        if _cfg_target is None:
+            _cfg_target = sysconfig.get_config_var(
+                                  'MACOSX_DEPLOYMENT_TARGET') or ''
+            if _cfg_target:
+                _cfg_target_split = [int(x) for x in _cfg_target.split('.')]
+        if _cfg_target:
+            # ensure that the deployment target of build process is not less
+            # than that used when the interpreter was built. This ensures
+            # extension modules are built with correct compatibility values
+            cur_target = os.environ.get('MACOSX_DEPLOYMENT_TARGET', _cfg_target)
+            if _cfg_target_split > [int(x) for x in cur_target.split('.')]:
+                my_msg = ('$MACOSX_DEPLOYMENT_TARGET mismatch: '
+                          'now "%s" but "%s" during configure'
+                                % (cur_target, _cfg_target))
+                raise DistutilsPlatformError(my_msg)
+            env = dict(os.environ,
+                       MACOSX_DEPLOYMENT_TARGET=cur_target)
+            exec_fn = search_path and os.execvpe or os.execve
+    pid = os.fork()
+    if pid == 0: # in the child
+        try:
+            if env is None:
+                exec_fn(executable, cmd)
+            else:
+                exec_fn(executable, cmd, env)
+        except OSError as e:
+            if not DEBUG:
+                cmd = executable
+            sys.stderr.write("unable to execute %r: %s\n"
+                             % (cmd, e.strerror))
+            os._exit(1)
+
+        if not DEBUG:
+            cmd = executable
+        sys.stderr.write("unable to execute %r for unknown reasons" % cmd)
+        os._exit(1)
+    else: # in the parent
+        # Loop until the child either exits or is terminated by a signal
+        # (ie. keep waiting if it's merely stopped)
+        while True:
+            try:
+                pid, status = os.waitpid(pid, 0)
+            except OSError as exc:
+                if not DEBUG:
+                    cmd = executable
+                raise DistutilsExecError(
+                      "command %r failed: %s" % (cmd, exc.args[-1]))
+            if os.WIFSIGNALED(status):
+                if not DEBUG:
+                    cmd = executable
+                raise DistutilsExecError(
+                      "command %r terminated by signal %d"
+                      % (cmd, os.WTERMSIG(status)))
+            elif os.WIFEXITED(status):
+                exit_status = os.WEXITSTATUS(status)
+                if exit_status == 0:
+                    return   # hey, it succeeded!
+                else:
+                    if not DEBUG:
+                        cmd = executable
+                    raise DistutilsExecError(
+                          "command %r failed with exit status %d"
+                          % (cmd, exit_status))
+            elif os.WIFSTOPPED(status):
+                continue
+            else:
+                if not DEBUG:
+                    cmd = executable
+                raise DistutilsExecError(
+                      "unknown error executing %r: termination status %d"
+                      % (cmd, status))
+
+def find_executable(executable, path=None):
+    """Tries to find 'executable' in the directories listed in 'path'.
+
+    A string listing directories separated by 'os.pathsep'; defaults to
+    os.environ['PATH'].  Returns the complete filename or None if not found.
+    """
+    if path is None:
+        path = os.environ.get('PATH', os.defpath)
+
+    paths = path.split(os.pathsep)
+    base, ext = os.path.splitext(executable)
+
+    if (sys.platform == 'win32') and (ext != '.exe'):
+        executable = executable + '.exe'
+
+    if not os.path.isfile(executable):
+        for p in paths:
+            f = os.path.join(p, executable)
+            if os.path.isfile(f):
+                # the file exists, we have a shot at spawn working
+                return f
+        return None
+    else:
+        return executable
diff --git a/Lib/distutils/sysconfig.py b/Lib/distutils/sysconfig.py
new file mode 100644
index 000000000..b1a98539c
--- /dev/null
+++ b/Lib/distutils/sysconfig.py
@@ -0,0 +1,558 @@
+"""Provide access to Python's configuration information.  The specific
+configuration variables available depend heavily on the platform and
+configuration.  The values may be retrieved using
+get_config_var(name), and the list of variables is available via
+get_config_vars().keys().  Additional convenience functions are also
+available.
+
+Written by:   Fred L. Drake, Jr.
+Email:        <fdrake@acm.org>
+"""
+
+import _imp
+import os
+import re
+import sys
+import fnmatch
+
+from .errors import DistutilsPlatformError
+
+# These are needed in a couple of spots, so just compute them once.
+PREFIX = os.path.normpath(sys.prefix)
+EXEC_PREFIX = os.path.normpath(sys.exec_prefix)
+BASE_PREFIX = os.path.normpath(sys.base_prefix)
+BASE_EXEC_PREFIX = os.path.normpath(sys.base_exec_prefix)
+
+# Path to the base directory of the project. On Windows the binary may
+# live in project/PCBuild/win32 or project/PCBuild/amd64.
+# set for cross builds
+if "_PYTHON_PROJECT_BASE" in os.environ:
+    project_base = os.path.abspath(os.environ["_PYTHON_PROJECT_BASE"])
+else:
+    project_base = os.path.dirname(os.path.abspath(sys.executable))
+if (os.name == 'nt' and
+    project_base.lower().endswith(('\\pcbuild\\win32', '\\pcbuild\\amd64'))):
+    project_base = os.path.dirname(os.path.dirname(project_base))
+
+# python_build: (Boolean) if true, we're either building Python or
+# building an extension with an un-installed Python, so we use
+# different (hard-wired) directories.
+# Setup.local is available for Makefile builds including VPATH builds,
+# Setup.dist is available on Windows
+def _is_python_source_dir(d):
+    for fn in ("Setup.dist", "Setup.local"):
+        if os.path.isfile(os.path.join(d, "Modules", fn)):
+            return True
+    return False
+_sys_home = getattr(sys, '_home', None)
+if (_sys_home and os.name == 'nt' and
+    _sys_home.lower().endswith(('\\pcbuild\\win32', '\\pcbuild\\amd64'))):
+    _sys_home = os.path.dirname(os.path.dirname(_sys_home))
+def _python_build():
+    if _sys_home:
+        return _is_python_source_dir(_sys_home)
+    return _is_python_source_dir(project_base)
+python_build = _python_build()
+
+# Calculate the build qualifier flags if they are defined.  Adding the flags
+# to the include and lib directories only makes sense for an installation, not
+# an in-source build.
+build_flags = ''
+try:
+    if not python_build:
+        build_flags = sys.abiflags
+except AttributeError:
+    # It's not a configure-based build, so the sys module doesn't have
+    # this attribute, which is fine.
+    pass
+
+def get_python_version():
+    """Return a string containing the major and minor Python version,
+    leaving off the patchlevel.  Sample return values could be '1.5'
+    or '2.2'.
+    """
+    return '%d.%d' % sys.version_info[:2]
+
+
+def get_python_inc(plat_specific=0, prefix=None):
+    """Return the directory containing installed Python header files.
+
+    If 'plat_specific' is false (the default), this is the path to the
+    non-platform-specific header files, i.e. Python.h and so on;
+    otherwise, this is the path to platform-specific header files
+    (namely pyconfig.h).
+
+    If 'prefix' is supplied, use it instead of sys.base_prefix or
+    sys.base_exec_prefix -- i.e., ignore 'plat_specific'.
+    """
+    if prefix is None:
+        prefix = plat_specific and BASE_EXEC_PREFIX or BASE_PREFIX
+    if os.name == "posix":
+        if python_build:
+            # Assume the executable is in the build directory.  The
+            # pyconfig.h file should be in the same directory.  Since
+            # the build directory may not be the source directory, we
+            # must use "srcdir" from the makefile to find the "Include"
+            # directory.
+            if plat_specific:
+                return _sys_home or project_base
+            else:
+                incdir = os.path.join(get_config_var('srcdir'), 'Include')
+                return os.path.normpath(incdir)
+        python_dir = 'python' + get_python_version() + build_flags
+        if not python_build and plat_specific:
+            import sysconfig
+            return sysconfig.get_path('platinclude')
+        return os.path.join(prefix, "include", python_dir)
+    elif os.name == "nt":
+        return os.path.join(prefix, "include")
+    else:
+        raise DistutilsPlatformError(
+            "I don't know where Python installs its C header files "
+            "on platform '%s'" % os.name)
+
+
+def get_python_lib(plat_specific=0, standard_lib=0, prefix=None):
+    """Return the directory containing the Python library (standard or
+    site additions).
+
+    If 'plat_specific' is true, return the directory containing
+    platform-specific modules, i.e. any module from a non-pure-Python
+    module distribution; otherwise, return the platform-shared library
+    directory.  If 'standard_lib' is true, return the directory
+    containing standard Python library modules; otherwise, return the
+    directory for site-specific modules.
+
+    If 'prefix' is supplied, use it instead of sys.base_prefix or
+    sys.base_exec_prefix -- i.e., ignore 'plat_specific'.
+    """
+    is_default_prefix = not prefix or os.path.normpath(prefix) in ('/usr', '/usr/local')
+    if prefix is None:
+        if standard_lib:
+            prefix = plat_specific and BASE_EXEC_PREFIX or BASE_PREFIX
+        else:
+            prefix = plat_specific and EXEC_PREFIX or PREFIX
+
+    if os.name == "posix":
+        libpython = os.path.join(prefix,
+                                 "lib", "python" + get_python_version())
+        if standard_lib:
+            return libpython
+        elif (is_default_prefix and
+              'PYTHONUSERBASE' not in os.environ and
+              'VIRTUAL_ENV' not in os.environ and
+              'real_prefix' not in sys.__dict__ and
+              sys.prefix == sys.base_prefix):
+            return os.path.join(prefix, "lib", "python3", "dist-packages")
+        else:
+            return os.path.join(libpython, "site-packages")
+    elif os.name == "nt":
+        if standard_lib:
+            return os.path.join(prefix, "Lib")
+        else:
+            return os.path.join(prefix, "Lib", "site-packages")
+    else:
+        raise DistutilsPlatformError(
+            "I don't know where Python installs its library "
+            "on platform '%s'" % os.name)
+
+
+
+def customize_compiler(compiler):
+    """Do any platform-specific customization of a CCompiler instance.
+
+    Mainly needed on Unix, so we can plug in the information that
+    varies across Unices and is stored in Python's Makefile.
+    """
+    if compiler.compiler_type == "unix":
+        if sys.platform == "darwin":
+            # Perform first-time customization of compiler-related
+            # config vars on OS X now that we know we need a compiler.
+            # This is primarily to support Pythons from binary
+            # installers.  The kind and paths to build tools on
+            # the user system may vary significantly from the system
+            # that Python itself was built on.  Also the user OS
+            # version and build tools may not support the same set
+            # of CPU architectures for universal builds.
+            global _config_vars
+            # Use get_config_var() to ensure _config_vars is initialized.
+            if not get_config_var('CUSTOMIZED_OSX_COMPILER'):
+                import _osx_support
+                _osx_support.customize_compiler(_config_vars)
+                _config_vars['CUSTOMIZED_OSX_COMPILER'] = 'True'
+
+        (cc, cxx, opt, cflags, ccshared, ldshared, shlib_suffix, ar, ar_flags,
+         configure_cppflags, configure_cflags, configure_ldflags) = \
+            get_config_vars('CC', 'CXX', 'OPT', 'CFLAGS',
+                            'CCSHARED', 'LDSHARED', 'SHLIB_SUFFIX', 'AR', 'ARFLAGS',
+                            'CONFIGURE_CPPFLAGS', 'CONFIGURE_CFLAGS', 'CONFIGURE_LDFLAGS')
+
+        if 'CC' in os.environ:
+            newcc = os.environ['CC']
+            if (sys.platform == 'darwin'
+                    and 'LDSHARED' not in os.environ
+                    and ldshared.startswith(cc)):
+                # On OS X, if CC is overridden, use that as the default
+                #       command for LDSHARED as well
+                ldshared = newcc + ldshared[len(cc):]
+            cc = newcc
+        if 'CXX' in os.environ:
+            cxx = os.environ['CXX']
+        if fnmatch.filter([cc, cxx], '*-4.[0-8]'):
+            configure_cflags = configure_cflags.replace('-fstack-protector-strong', '-fstack-protector')
+            ldshared = ldshared.replace('-fstack-protector-strong', '-fstack-protector')
+            cflags = cflags.replace('-fstack-protector-strong', '-fstack-protector')
+        if 'LDSHARED' in os.environ:
+            ldshared = os.environ['LDSHARED']
+        if 'CPP' in os.environ:
+            cpp = os.environ['CPP']
+        else:
+            cpp = cc + " -E"           # not always
+        if 'LDFLAGS' in os.environ:
+            ldshared = ldshared + ' ' + os.environ['LDFLAGS']
+        elif configure_ldflags:
+            ldshared = ldshared + ' ' + configure_ldflags
+        if 'CFLAGS' in os.environ:
+            cflags = opt + ' ' + os.environ['CFLAGS']
+            ldshared = ldshared + ' ' + os.environ['CFLAGS']
+        elif configure_cflags:
+            cflags = opt + ' ' + configure_cflags
+            ldshared = ldshared + ' ' + configure_cflags
+        if 'CPPFLAGS' in os.environ:
+            cpp = cpp + ' ' + os.environ['CPPFLAGS']
+            cflags = cflags + ' ' + os.environ['CPPFLAGS']
+            ldshared = ldshared + ' ' + os.environ['CPPFLAGS']
+        elif configure_cppflags:
+            cpp = cpp + ' ' + configure_cppflags
+            cflags = cflags + ' ' + configure_cppflags
+            ldshared = ldshared + ' ' + configure_cppflags
+        if 'AR' in os.environ:
+            ar = os.environ['AR']
+        if 'ARFLAGS' in os.environ:
+            archiver = ar + ' ' + os.environ['ARFLAGS']
+        else:
+            archiver = ar + ' ' + ar_flags
+
+        cc_cmd = cc + ' ' + cflags
+        compiler.set_executables(
+            preprocessor=cpp,
+            compiler=cc_cmd,
+            compiler_so=cc_cmd + ' ' + ccshared,
+            compiler_cxx=cxx,
+            linker_so=ldshared,
+            linker_exe=cc,
+            archiver=archiver)
+
+        compiler.shared_lib_extension = shlib_suffix
+
+
+def get_config_h_filename():
+    """Return full pathname of installed pyconfig.h file."""
+    if python_build:
+        if os.name == "nt":
+            inc_dir = os.path.join(_sys_home or project_base, "PC")
+        else:
+            inc_dir = _sys_home or project_base
+    else:
+        inc_dir = get_python_inc(plat_specific=1)
+
+    return os.path.join(inc_dir, 'pyconfig.h')
+
+
+def get_makefile_filename():
+    """Return full pathname of installed Makefile from the Python build."""
+    if python_build:
+        return os.path.join(_sys_home or project_base, "Makefile")
+    lib_dir = get_python_lib(plat_specific=0, standard_lib=1)
+    config_file = 'config-{}{}'.format(get_python_version(), build_flags)
+    if hasattr(sys.implementation, '_multiarch'):
+        config_file += '-%s' % sys.implementation._multiarch
+    return os.path.join(lib_dir, config_file, 'Makefile')
+
+
+def parse_config_h(fp, g=None):
+    """Parse a config.h-style file.
+
+    A dictionary containing name/value pairs is returned.  If an
+    optional dictionary is passed in as the second argument, it is
+    used instead of a new dictionary.
+    """
+    if g is None:
+        g = {}
+    define_rx = re.compile("#define ([A-Z][A-Za-z0-9_]+) (.*)\n")
+    undef_rx = re.compile("/[*] #undef ([A-Z][A-Za-z0-9_]+) [*]/\n")
+    #
+    while True:
+        line = fp.readline()
+        if not line:
+            break
+        m = define_rx.match(line)
+        if m:
+            n, v = m.group(1, 2)
+            try: v = int(v)
+            except ValueError: pass
+            g[n] = v
+        else:
+            m = undef_rx.match(line)
+            if m:
+                g[m.group(1)] = 0
+    return g
+
+
+# Regexes needed for parsing Makefile (and similar syntaxes,
+# like old-style Setup files).
+_variable_rx = re.compile(r"([a-zA-Z][a-zA-Z0-9_]+)\s*=\s*(.*)")
+_findvar1_rx = re.compile(r"\$\(([A-Za-z][A-Za-z0-9_]*)\)")
+_findvar2_rx = re.compile(r"\${([A-Za-z][A-Za-z0-9_]*)}")
+
+def parse_makefile(fn, g=None):
+    """Parse a Makefile-style file.
+
+    A dictionary containing name/value pairs is returned.  If an
+    optional dictionary is passed in as the second argument, it is
+    used instead of a new dictionary.
+    """
+    from distutils.text_file import TextFile
+    fp = TextFile(fn, strip_comments=1, skip_blanks=1, join_lines=1, errors="surrogateescape")
+
+    if g is None:
+        g = {}
+    done = {}
+    notdone = {}
+
+    while True:
+        line = fp.readline()
+        if line is None: # eof
+            break
+        m = _variable_rx.match(line)
+        if m:
+            n, v = m.group(1, 2)
+            v = v.strip()
+            # `$$' is a literal `$' in make
+            tmpv = v.replace('$$', '')
+
+            if "$" in tmpv:
+                notdone[n] = v
+            else:
+                try:
+                    v = int(v)
+                except ValueError:
+                    # insert literal `$'
+                    done[n] = v.replace('$$', '$')
+                else:
+                    done[n] = v
+
+    # Variables with a 'PY_' prefix in the makefile. These need to
+    # be made available without that prefix through sysconfig.
+    # Special care is needed to ensure that variable expansion works, even
+    # if the expansion uses the name without a prefix.
+    renamed_variables = ('CFLAGS', 'LDFLAGS', 'CPPFLAGS')
+
+    # do variable interpolation here
+    while notdone:
+        for name in list(notdone):
+            value = notdone[name]
+            m = _findvar1_rx.search(value) or _findvar2_rx.search(value)
+            if m:
+                n = m.group(1)
+                found = True
+                if n in done:
+                    item = str(done[n])
+                elif n in notdone:
+                    # get it on a subsequent round
+                    found = False
+                elif n in os.environ:
+                    # do it like make: fall back to environment
+                    item = os.environ[n]
+
+                elif n in renamed_variables:
+                    if name.startswith('PY_') and name[3:] in renamed_variables:
+                        item = ""
+
+                    elif 'PY_' + n in notdone:
+                        found = False
+
+                    else:
+                        item = str(done['PY_' + n])
+                else:
+                    done[n] = item = ""
+                if found:
+                    after = value[m.end():]
+                    value = value[:m.start()] + item + after
+                    if "$" in after:
+                        notdone[name] = value
+                    else:
+                        try: value = int(value)
+                        except ValueError:
+                            done[name] = value.strip()
+                        else:
+                            done[name] = value
+                        del notdone[name]
+
+                        if name.startswith('PY_') \
+                            and name[3:] in renamed_variables:
+
+                            name = name[3:]
+                            if name not in done:
+                                done[name] = value
+            else:
+                # bogus variable reference; just drop it since we can't deal
+                del notdone[name]
+
+    fp.close()
+
+    # strip spurious spaces
+    for k, v in done.items():
+        if isinstance(v, str):
+            done[k] = v.strip()
+
+    # save the results in the global dictionary
+    g.update(done)
+    return g
+
+
+def expand_makefile_vars(s, vars):
+    """Expand Makefile-style variables -- "${foo}" or "$(foo)" -- in
+    'string' according to 'vars' (a dictionary mapping variable names to
+    values).  Variables not present in 'vars' are silently expanded to the
+    empty string.  The variable values in 'vars' should not contain further
+    variable expansions; if 'vars' is the output of 'parse_makefile()',
+    you're fine.  Returns a variable-expanded version of 's'.
+    """
+
+    # This algorithm does multiple expansion, so if vars['foo'] contains
+    # "${bar}", it will expand ${foo} to ${bar}, and then expand
+    # ${bar}... and so forth.  This is fine as long as 'vars' comes from
+    # 'parse_makefile()', which takes care of such expansions eagerly,
+    # according to make's variable expansion semantics.
+
+    while True:
+        m = _findvar1_rx.search(s) or _findvar2_rx.search(s)
+        if m:
+            (beg, end) = m.span()
+            s = s[0:beg] + vars.get(m.group(1)) + s[end:]
+        else:
+            break
+    return s
+
+
+_config_vars = None
+
+def _init_posix():
+    """Initialize the module as appropriate for POSIX systems."""
+    # _sysconfigdata is generated at build time, see the sysconfig module
+    name = os.environ.get('_PYTHON_SYSCONFIGDATA_NAME',
+        '_sysconfigdata_{abi}_{platform}_{multiarch}'.format(
+        abi=sys.abiflags,
+        platform=sys.platform,
+        multiarch=getattr(sys.implementation, '_multiarch', ''),
+    ))
+    _temp = __import__(name, globals(), locals(), ['build_time_vars'], 0)
+    build_time_vars = _temp.build_time_vars
+    global _config_vars
+    _config_vars = {}
+    _config_vars.update(build_time_vars)
+
+
+def _init_nt():
+    """Initialize the module as appropriate for NT"""
+    g = {}
+    # set basic install directories
+    g['LIBDEST'] = get_python_lib(plat_specific=0, standard_lib=1)
+    g['BINLIBDEST'] = get_python_lib(plat_specific=1, standard_lib=1)
+
+    # XXX hmmm.. a normal install puts include files here
+    g['INCLUDEPY'] = get_python_inc(plat_specific=0)
+
+    g['EXT_SUFFIX'] = _imp.extension_suffixes()[0]
+    g['EXE'] = ".exe"
+    g['VERSION'] = get_python_version().replace(".", "")
+    g['BINDIR'] = os.path.dirname(os.path.abspath(sys.executable))
+
+    global _config_vars
+    _config_vars = g
+
+
+def get_config_vars(*args):
+    """With no arguments, return a dictionary of all configuration
+    variables relevant for the current platform.  Generally this includes
+    everything needed to build extensions and install both pure modules and
+    extensions.  On Unix, this means every variable defined in Python's
+    installed Makefile; on Windows it's a much smaller set.
+
+    With arguments, return a list of values that result from looking up
+    each argument in the configuration variable dictionary.
+    """
+    global _config_vars
+    if _config_vars is None:
+        func = globals().get("_init_" + os.name)
+        if func:
+            func()
+        else:
+            _config_vars = {}
+
+        # Normalized versions of prefix and exec_prefix are handy to have;
+        # in fact, these are the standard versions used most places in the
+        # Distutils.
+        _config_vars['prefix'] = PREFIX
+        _config_vars['exec_prefix'] = EXEC_PREFIX
+
+        # For backward compatibility, see issue19555
+        SO = _config_vars.get('EXT_SUFFIX')
+        if SO is not None:
+            _config_vars['SO'] = SO
+
+        # Always convert srcdir to an absolute path
+        srcdir = _config_vars.get('srcdir', project_base)
+        if os.name == 'posix':
+            if python_build:
+                # If srcdir is a relative path (typically '.' or '..')
+                # then it should be interpreted relative to the directory
+                # containing Makefile.
+                base = os.path.dirname(get_makefile_filename())
+                srcdir = os.path.join(base, srcdir)
+            else:
+                # srcdir is not meaningful since the installation is
+                # spread about the filesystem.  We choose the
+                # directory containing the Makefile since we know it
+                # exists.
+                srcdir = os.path.dirname(get_makefile_filename())
+        _config_vars['srcdir'] = os.path.abspath(os.path.normpath(srcdir))
+
+        # Convert srcdir into an absolute path if it appears necessary.
+        # Normally it is relative to the build directory.  However, during
+        # testing, for example, we might be running a non-installed python
+        # from a different directory.
+        if python_build and os.name == "posix":
+            base = project_base
+            if (not os.path.isabs(_config_vars['srcdir']) and
+                base != os.getcwd()):
+                # srcdir is relative and we are not in the same directory
+                # as the executable. Assume executable is in the build
+                # directory and make srcdir absolute.
+                srcdir = os.path.join(base, _config_vars['srcdir'])
+                _config_vars['srcdir'] = os.path.normpath(srcdir)
+
+        # OS X platforms require special customization to handle
+        # multi-architecture, multi-os-version installers
+        if sys.platform == 'darwin':
+            import _osx_support
+            _osx_support.customize_config_vars(_config_vars)
+
+    if args:
+        vals = []
+        for name in args:
+            vals.append(_config_vars.get(name))
+        return vals
+    else:
+        return _config_vars
+
+def get_config_var(name):
+    """Return the value of a single variable using the dictionary
+    returned by 'get_config_vars()'.  Equivalent to
+    get_config_vars().get(name)
+    """
+    if name == 'SO':
+        import warnings
+        warnings.warn('SO is deprecated, use EXT_SUFFIX', DeprecationWarning, 2)
+    return get_config_vars().get(name)
diff --git a/Lib/distutils/text_file.py b/Lib/distutils/text_file.py
new file mode 100644
index 000000000..93abad38f
--- /dev/null
+++ b/Lib/distutils/text_file.py
@@ -0,0 +1,286 @@
+"""text_file
+
+provides the TextFile class, which gives an interface to text files
+that (optionally) takes care of stripping comments, ignoring blank
+lines, and joining lines with backslashes."""
+
+import sys, io
+
+
+class TextFile:
+    """Provides a file-like object that takes care of all the things you
+       commonly want to do when processing a text file that has some
+       line-by-line syntax: strip comments (as long as "#" is your
+       comment character), skip blank lines, join adjacent lines by
+       escaping the newline (ie. backslash at end of line), strip
+       leading and/or trailing whitespace.  All of these are optional
+       and independently controllable.
+
+       Provides a 'warn()' method so you can generate warning messages that
+       report physical line number, even if the logical line in question
+       spans multiple physical lines.  Also provides 'unreadline()' for
+       implementing line-at-a-time lookahead.
+
+       Constructor is called as:
+
+           TextFile (filename=None, file=None, **options)
+
+       It bombs (RuntimeError) if both 'filename' and 'file' are None;
+       'filename' should be a string, and 'file' a file object (or
+       something that provides 'readline()' and 'close()' methods).  It is
+       recommended that you supply at least 'filename', so that TextFile
+       can include it in warning messages.  If 'file' is not supplied,
+       TextFile creates its own using 'io.open()'.
+
+       The options are all boolean, and affect the value returned by
+       'readline()':
+         strip_comments [default: true]
+           strip from "#" to end-of-line, as well as any whitespace
+           leading up to the "#" -- unless it is escaped by a backslash
+         lstrip_ws [default: false]
+           strip leading whitespace from each line before returning it
+         rstrip_ws [default: true]
+           strip trailing whitespace (including line terminator!) from
+           each line before returning it
+         skip_blanks [default: true}
+           skip lines that are empty *after* stripping comments and
+           whitespace.  (If both lstrip_ws and rstrip_ws are false,
+           then some lines may consist of solely whitespace: these will
+           *not* be skipped, even if 'skip_blanks' is true.)
+         join_lines [default: false]
+           if a backslash is the last non-newline character on a line
+           after stripping comments and whitespace, join the following line
+           to it to form one "logical line"; if N consecutive lines end
+           with a backslash, then N+1 physical lines will be joined to
+           form one logical line.
+         collapse_join [default: false]
+           strip leading whitespace from lines that are joined to their
+           predecessor; only matters if (join_lines and not lstrip_ws)
+         errors [default: 'strict']
+           error handler used to decode the file content
+
+       Note that since 'rstrip_ws' can strip the trailing newline, the
+       semantics of 'readline()' must differ from those of the builtin file
+       object's 'readline()' method!  In particular, 'readline()' returns
+       None for end-of-file: an empty string might just be a blank line (or
+       an all-whitespace line), if 'rstrip_ws' is true but 'skip_blanks' is
+       not."""
+
+    default_options = { 'strip_comments': 1,
+                        'skip_blanks':    1,
+                        'lstrip_ws':      0,
+                        'rstrip_ws':      1,
+                        'join_lines':     0,
+                        'collapse_join':  0,
+                        'errors':         'strict',
+                      }
+
+    def __init__(self, filename=None, file=None, **options):
+        """Construct a new TextFile object.  At least one of 'filename'
+           (a string) and 'file' (a file-like object) must be supplied.
+           They keyword argument options are described above and affect
+           the values returned by 'readline()'."""
+        if filename is None and file is None:
+            raise RuntimeError("you must supply either or both of 'filename' and 'file'")
+
+        # set values for all options -- either from client option hash
+        # or fallback to default_options
+        for opt in self.default_options.keys():
+            if opt in options:
+                setattr(self, opt, options[opt])
+            else:
+                setattr(self, opt, self.default_options[opt])
+
+        # sanity check client option hash
+        for opt in options.keys():
+            if opt not in self.default_options:
+                raise KeyError("invalid TextFile option '%s'" % opt)
+
+        if file is None:
+            self.open(filename)
+        else:
+            self.filename = filename
+            self.file = file
+            self.current_line = 0       # assuming that file is at BOF!
+
+        # 'linebuf' is a stack of lines that will be emptied before we
+        # actually read from the file; it's only populated by an
+        # 'unreadline()' operation
+        self.linebuf = []
+
+    def open(self, filename):
+        """Open a new file named 'filename'.  This overrides both the
+           'filename' and 'file' arguments to the constructor."""
+        self.filename = filename
+        self.file = io.open(self.filename, 'r', errors=self.errors)
+        self.current_line = 0
+
+    def close(self):
+        """Close the current file and forget everything we know about it
+           (filename, current line number)."""
+        file = self.file
+        self.file = None
+        self.filename = None
+        self.current_line = None
+        file.close()
+
+    def gen_error(self, msg, line=None):
+        outmsg = []
+        if line is None:
+            line = self.current_line
+        outmsg.append(self.filename + ", ")
+        if isinstance(line, (list, tuple)):
+            outmsg.append("lines %d-%d: " % tuple(line))
+        else:
+            outmsg.append("line %d: " % line)
+        outmsg.append(str(msg))
+        return "".join(outmsg)
+
+    def error(self, msg, line=None):
+        raise ValueError("error: " + self.gen_error(msg, line))
+
+    def warn(self, msg, line=None):
+        """Print (to stderr) a warning message tied to the current logical
+           line in the current file.  If the current logical line in the
+           file spans multiple physical lines, the warning refers to the
+           whole range, eg. "lines 3-5".  If 'line' supplied, it overrides
+           the current line number; it may be a list or tuple to indicate a
+           range of physical lines, or an integer for a single physical
+           line."""
+        sys.stderr.write("warning: " + self.gen_error(msg, line) + "\n")
+
+    def readline(self):
+        """Read and return a single logical line from the current file (or
+           from an internal buffer if lines have previously been "unread"
+           with 'unreadline()').  If the 'join_lines' option is true, this
+           may involve reading multiple physical lines concatenated into a
+           single string.  Updates the current line number, so calling
+           'warn()' after 'readline()' emits a warning about the physical
+           line(s) just read.  Returns None on end-of-file, since the empty
+           string can occur if 'rstrip_ws' is true but 'strip_blanks' is
+           not."""
+        # If any "unread" lines waiting in 'linebuf', return the top
+        # one.  (We don't actually buffer read-ahead data -- lines only
+        # get put in 'linebuf' if the client explicitly does an
+        # 'unreadline()'.
+        if self.linebuf:
+            line = self.linebuf[-1]
+            del self.linebuf[-1]
+            return line
+
+        buildup_line = ''
+
+        while True:
+            # read the line, make it None if EOF
+            line = self.file.readline()
+            if line == '':
+                line = None
+
+            if self.strip_comments and line:
+
+                # Look for the first "#" in the line.  If none, never
+                # mind.  If we find one and it's the first character, or
+                # is not preceded by "\", then it starts a comment --
+                # strip the comment, strip whitespace before it, and
+                # carry on.  Otherwise, it's just an escaped "#", so
+                # unescape it (and any other escaped "#"'s that might be
+                # lurking in there) and otherwise leave the line alone.
+
+                pos = line.find("#")
+                if pos == -1: # no "#" -- no comments
+                    pass
+
+                # It's definitely a comment -- either "#" is the first
+                # character, or it's elsewhere and unescaped.
+                elif pos == 0 or line[pos-1] != "\\":
+                    # Have to preserve the trailing newline, because it's
+                    # the job of a later step (rstrip_ws) to remove it --
+                    # and if rstrip_ws is false, we'd better preserve it!
+                    # (NB. this means that if the final line is all comment
+                    # and has no trailing newline, we will think that it's
+                    # EOF; I think that's OK.)
+                    eol = (line[-1] == '\n') and '\n' or ''
+                    line = line[0:pos] + eol
+
+                    # If all that's left is whitespace, then skip line
+                    # *now*, before we try to join it to 'buildup_line' --
+                    # that way constructs like
+                    #   hello \\
+                    #   # comment that should be ignored
+                    #   there
+                    # result in "hello there".
+                    if line.strip() == "":
+                        continue
+                else: # it's an escaped "#"
+                    line = line.replace("\\#", "#")
+
+            # did previous line end with a backslash? then accumulate
+            if self.join_lines and buildup_line:
+                # oops: end of file
+                if line is None:
+                    self.warn("continuation line immediately precedes "
+                              "end-of-file")
+                    return buildup_line
+
+                if self.collapse_join:
+                    line = line.lstrip()
+                line = buildup_line + line
+
+                # careful: pay attention to line number when incrementing it
+                if isinstance(self.current_line, list):
+                    self.current_line[1] = self.current_line[1] + 1
+                else:
+                    self.current_line = [self.current_line,
+                                         self.current_line + 1]
+            # just an ordinary line, read it as usual
+            else:
+                if line is None: # eof
+                    return None
+
+                # still have to be careful about incrementing the line number!
+                if isinstance(self.current_line, list):
+                    self.current_line = self.current_line[1] + 1
+                else:
+                    self.current_line = self.current_line + 1
+
+            # strip whitespace however the client wants (leading and
+            # trailing, or one or the other, or neither)
+            if self.lstrip_ws and self.rstrip_ws:
+                line = line.strip()
+            elif self.lstrip_ws:
+                line = line.lstrip()
+            elif self.rstrip_ws:
+                line = line.rstrip()
+
+            # blank line (whether we rstrip'ed or not)? skip to next line
+            # if appropriate
+            if (line == '' or line == '\n') and self.skip_blanks:
+                continue
+
+            if self.join_lines:
+                if line[-1] == '\\':
+                    buildup_line = line[:-1]
+                    continue
+
+                if line[-2:] == '\\\n':
+                    buildup_line = line[0:-2] + '\n'
+                    continue
+
+            # well, I guess there's some actual content there: return it
+            return line
+
+    def readlines(self):
+        """Read and return the list of all logical lines remaining in the
+           current file."""
+        lines = []
+        while True:
+            line = self.readline()
+            if line is None:
+                return lines
+            lines.append(line)
+
+    def unreadline(self, line):
+        """Push 'line' (a string) onto an internal buffer that will be
+           checked by future 'readline()' calls.  Handy for implementing
+           a parser with line-at-a-time lookahead."""
+        self.linebuf.append(line)
diff --git a/Lib/distutils/unixccompiler.py b/Lib/distutils/unixccompiler.py
new file mode 100644
index 000000000..4524417e6
--- /dev/null
+++ b/Lib/distutils/unixccompiler.py
@@ -0,0 +1,333 @@
+"""distutils.unixccompiler
+
+Contains the UnixCCompiler class, a subclass of CCompiler that handles
+the "typical" Unix-style command-line C compiler:
+  * macros defined with -Dname[=value]
+  * macros undefined with -Uname
+  * include search directories specified with -Idir
+  * libraries specified with -lllib
+  * library search directories specified with -Ldir
+  * compile handled by 'cc' (or similar) executable with -c option:
+    compiles .c to .o
+  * link static library handled by 'ar' command (possibly with 'ranlib')
+  * link shared library handled by 'cc -shared'
+"""
+
+import os, sys, re
+
+from distutils import sysconfig
+from distutils.dep_util import newer
+from distutils.ccompiler import \
+     CCompiler, gen_preprocess_options, gen_lib_options
+from distutils.errors import \
+     DistutilsExecError, CompileError, LibError, LinkError
+from distutils import log
+
+if sys.platform == 'darwin':
+    import _osx_support
+
+# XXX Things not currently handled:
+#   * optimization/debug/warning flags; we just use whatever's in Python's
+#     Makefile and live with it.  Is this adequate?  If not, we might
+#     have to have a bunch of subclasses GNUCCompiler, SGICCompiler,
+#     SunCCompiler, and I suspect down that road lies madness.
+#   * even if we don't know a warning flag from an optimization flag,
+#     we need some way for outsiders to feed preprocessor/compiler/linker
+#     flags in to us -- eg. a sysadmin might want to mandate certain flags
+#     via a site config file, or a user might want to set something for
+#     compiling this module distribution only via the setup.py command
+#     line, whatever.  As long as these options come from something on the
+#     current system, they can be as system-dependent as they like, and we
+#     should just happily stuff them into the preprocessor/compiler/linker
+#     options and carry on.
+
+
+class UnixCCompiler(CCompiler):
+
+    compiler_type = 'unix'
+
+    # These are used by CCompiler in two places: the constructor sets
+    # instance attributes 'preprocessor', 'compiler', etc. from them, and
+    # 'set_executable()' allows any of these to be set.  The defaults here
+    # are pretty generic; they will probably have to be set by an outsider
+    # (eg. using information discovered by the sysconfig about building
+    # Python extensions).
+    executables = {'preprocessor' : None,
+                   'compiler'     : ["cc"],
+                   'compiler_so'  : ["cc"],
+                   'compiler_cxx' : ["cc"],
+                   'linker_so'    : ["cc", "-shared"],
+                   'linker_exe'   : ["cc"],
+                   'archiver'     : ["ar", "-cr"],
+                   'ranlib'       : None,
+                  }
+
+    if sys.platform[:6] == "darwin":
+        executables['ranlib'] = ["ranlib"]
+
+    # Needed for the filename generation methods provided by the base
+    # class, CCompiler.  NB. whoever instantiates/uses a particular
+    # UnixCCompiler instance should set 'shared_lib_ext' -- we set a
+    # reasonable common default here, but it's not necessarily used on all
+    # Unices!
+
+    src_extensions = [".c",".C",".cc",".cxx",".cpp",".m"]
+    obj_extension = ".o"
+    static_lib_extension = ".a"
+    shared_lib_extension = ".so"
+    dylib_lib_extension = ".dylib"
+    xcode_stub_lib_extension = ".tbd"
+    static_lib_format = shared_lib_format = dylib_lib_format = "lib%s%s"
+    xcode_stub_lib_format = dylib_lib_format
+    if sys.platform == "cygwin":
+        exe_extension = ".exe"
+
+    def preprocess(self, source, output_file=None, macros=None,
+                   include_dirs=None, extra_preargs=None, extra_postargs=None):
+        fixed_args = self._fix_compile_args(None, macros, include_dirs)
+        ignore, macros, include_dirs = fixed_args
+        pp_opts = gen_preprocess_options(macros, include_dirs)
+        pp_args = self.preprocessor + pp_opts
+        if output_file:
+            pp_args.extend(['-o', output_file])
+        if extra_preargs:
+            pp_args[:0] = extra_preargs
+        if extra_postargs:
+            pp_args.extend(extra_postargs)
+        pp_args.append(source)
+
+        # We need to preprocess: either we're being forced to, or we're
+        # generating output to stdout, or there's a target output file and
+        # the source file is newer than the target (or the target doesn't
+        # exist).
+        if self.force or output_file is None or newer(source, output_file):
+            if output_file:
+                self.mkpath(os.path.dirname(output_file))
+            try:
+                self.spawn(pp_args)
+            except DistutilsExecError as msg:
+                raise CompileError(msg)
+
+    def _compile(self, obj, src, ext, cc_args, extra_postargs, pp_opts):
+        compiler_so = self.compiler_so
+        if sys.platform == 'darwin':
+            compiler_so = _osx_support.compiler_fixup(compiler_so,
+                                                    cc_args + extra_postargs)
+        try:
+            self.spawn(compiler_so + cc_args + [src, '-o', obj] +
+                       extra_postargs)
+        except DistutilsExecError as msg:
+            raise CompileError(msg)
+
+    def create_static_lib(self, objects, output_libname,
+                          output_dir=None, debug=0, target_lang=None):
+        objects, output_dir = self._fix_object_args(objects, output_dir)
+
+        output_filename = \
+            self.library_filename(output_libname, output_dir=output_dir)
+
+        if self._need_link(objects, output_filename):
+            self.mkpath(os.path.dirname(output_filename))
+            self.spawn(self.archiver +
+                       [output_filename] +
+                       objects + self.objects)
+
+            # Not many Unices required ranlib anymore -- SunOS 4.x is, I
+            # think the only major Unix that does.  Maybe we need some
+            # platform intelligence here to skip ranlib if it's not
+            # needed -- or maybe Python's configure script took care of
+            # it for us, hence the check for leading colon.
+            if self.ranlib:
+                try:
+                    self.spawn(self.ranlib + [output_filename])
+                except DistutilsExecError as msg:
+                    raise LibError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+    def link(self, target_desc, objects,
+             output_filename, output_dir=None, libraries=None,
+             library_dirs=None, runtime_library_dirs=None,
+             export_symbols=None, debug=0, extra_preargs=None,
+             extra_postargs=None, build_temp=None, target_lang=None):
+        objects, output_dir = self._fix_object_args(objects, output_dir)
+        fixed_args = self._fix_lib_args(libraries, library_dirs,
+                                        runtime_library_dirs)
+        libraries, library_dirs, runtime_library_dirs = fixed_args
+
+        # filter out standard library paths, which are not explicitely needed
+        # for linking
+        system_libdirs = ['/lib', '/lib64', '/usr/lib', '/usr/lib64']
+        multiarch = sysconfig.get_config_var("MULTIARCH")
+        if multiarch:
+            system_libdirs.extend(['/lib/%s' % multiarch, '/usr/lib/%s' % multiarch])
+        library_dirs = [dir for dir in library_dirs
+                        if not dir in system_libdirs]
+        runtime_library_dirs = [dir for dir in runtime_library_dirs
+                                if not dir in system_libdirs]
+
+        lib_opts = gen_lib_options(self, library_dirs, runtime_library_dirs,
+                                   libraries)
+        if not isinstance(output_dir, (str, type(None))):
+            raise TypeError("'output_dir' must be a string or None")
+        if output_dir is not None:
+            output_filename = os.path.join(output_dir, output_filename)
+
+        if self._need_link(objects, output_filename):
+            ld_args = (objects + self.objects +
+                       lib_opts + ['-o', output_filename])
+            if debug:
+                ld_args[:0] = ['-g']
+            if extra_preargs:
+                ld_args[:0] = extra_preargs
+            if extra_postargs:
+                ld_args.extend(extra_postargs)
+            self.mkpath(os.path.dirname(output_filename))
+            try:
+                if target_desc == CCompiler.EXECUTABLE:
+                    linker = self.linker_exe[:]
+                else:
+                    linker = self.linker_so[:]
+                if target_lang == "c++" and self.compiler_cxx:
+                    # skip over environment variable settings if /usr/bin/env
+                    # is used to set up the linker's environment.
+                    # This is needed on OSX. Note: this assumes that the
+                    # normal and C++ compiler have the same environment
+                    # settings.
+                    i = 0
+                    if os.path.basename(linker[0]) == "env":
+                        i = 1
+                        while '=' in linker[i]:
+                            i += 1
+                    linker[i] = self.compiler_cxx[i]
+
+                if sys.platform == 'darwin':
+                    linker = _osx_support.compiler_fixup(linker, ld_args)
+
+                self.spawn(linker + ld_args)
+            except DistutilsExecError as msg:
+                raise LinkError(msg)
+        else:
+            log.debug("skipping %s (up-to-date)", output_filename)
+
+    # -- Miscellaneous methods -----------------------------------------
+    # These are all used by the 'gen_lib_options() function, in
+    # ccompiler.py.
+
+    def library_dir_option(self, dir):
+        return "-L" + dir
+
+    def _is_gcc(self, compiler_name):
+        return "gcc" in compiler_name or "g++" in compiler_name
+
+    def runtime_library_dir_option(self, dir):
+        # XXX Hackish, at the very least.  See Python bug #445902:
+        # http://sourceforge.net/tracker/index.php
+        #   ?func=detail&aid=445902&group_id=5470&atid=105470
+        # Linkers on different platforms need different options to
+        # specify that directories need to be added to the list of
+        # directories searched for dependencies when a dynamic library
+        # is sought.  GCC on GNU systems (Linux, FreeBSD, ...) has to
+        # be told to pass the -R option through to the linker, whereas
+        # other compilers and gcc on other systems just know this.
+        # Other compilers may need something slightly different.  At
+        # this time, there's no way to determine this information from
+        # the configuration data stored in the Python installation, so
+        # we use this hack.
+        compiler = os.path.basename(sysconfig.get_config_var("CC"))
+        if sys.platform[:6] == "darwin":
+            # MacOSX's linker doesn't understand the -R flag at all
+            return "-L" + dir
+        elif sys.platform[:7] == "freebsd":
+            return "-Wl,-rpath=" + dir
+        elif sys.platform[:5] == "hp-ux":
+            if self._is_gcc(compiler):
+                return ["-Wl,+s", "-L" + dir]
+            return ["+s", "-L" + dir]
+        elif sys.platform[:7] == "irix646" or sys.platform[:6] == "osf1V5":
+            return ["-rpath", dir]
+        else:
+            if self._is_gcc(compiler):
+                # gcc on non-GNU systems does not need -Wl, but can
+                # use it anyway.  Since distutils has always passed in
+                # -Wl whenever gcc was used in the past it is probably
+                # safest to keep doing so.
+                if sysconfig.get_config_var("GNULD") == "yes":
+                    # GNU ld needs an extra option to get a RUNPATH
+                    # instead of just an RPATH.
+                    return "-Wl,--enable-new-dtags,-R" + dir
+                else:
+                    return "-Wl,-R" + dir
+            else:
+                # No idea how --enable-new-dtags would be passed on to
+                # ld if this system was using GNU ld.  Don't know if a
+                # system like this even exists.
+                return "-R" + dir
+
+    def library_option(self, lib):
+        return "-l" + lib
+
+    def find_library_file(self, dirs, lib, debug=0):
+        shared_f = self.library_filename(lib, lib_type='shared')
+        dylib_f = self.library_filename(lib, lib_type='dylib')
+        xcode_stub_f = self.library_filename(lib, lib_type='xcode_stub')
+        static_f = self.library_filename(lib, lib_type='static')
+
+        if sys.platform == 'darwin':
+            # On OSX users can specify an alternate SDK using
+            # '-isysroot', calculate the SDK root if it is specified
+            # (and use it further on)
+            #
+            # Note that, as of Xcode 7, Apple SDKs may contain textual stub
+            # libraries with .tbd extensions rather than the normal .dylib
+            # shared libraries installed in /.  The Apple compiler tool
+            # chain handles this transparently but it can cause problems
+            # for programs that are being built with an SDK and searching
+            # for specific libraries.  Callers of find_library_file need to
+            # keep in mind that the base filename of the returned SDK library
+            # file might have a different extension from that of the library
+            # file installed on the running system, for example:
+            #   /Applications/Xcode.app/Contents/Developer/Platforms/
+            #       MacOSX.platform/Developer/SDKs/MacOSX10.11.sdk/
+            #       usr/lib/libedit.tbd
+            # vs
+            #   /usr/lib/libedit.dylib
+            cflags = sysconfig.get_config_var('CFLAGS')
+            m = re.search(r'-isysroot\s+(\S+)', cflags)
+            if m is None:
+                sysroot = '/'
+            else:
+                sysroot = m.group(1)
+
+
+
+        for dir in dirs:
+            shared = os.path.join(dir, shared_f)
+            dylib = os.path.join(dir, dylib_f)
+            static = os.path.join(dir, static_f)
+            xcode_stub = os.path.join(dir, xcode_stub_f)
+
+            if sys.platform == 'darwin' and (
+                dir.startswith('/System/') or (
+                dir.startswith('/usr/') and not dir.startswith('/usr/local/'))):
+
+                shared = os.path.join(sysroot, dir[1:], shared_f)
+                dylib = os.path.join(sysroot, dir[1:], dylib_f)
+                static = os.path.join(sysroot, dir[1:], static_f)
+                xcode_stub = os.path.join(sysroot, dir[1:], xcode_stub_f)
+
+            # We're second-guessing the linker here, with not much hard
+            # data to go on: GCC seems to prefer the shared library, so I'm
+            # assuming that *all* Unix C compilers do.  And of course I'm
+            # ignoring even GCC's "-static" option.  So sue me.
+            if os.path.exists(dylib):
+                return dylib
+            elif os.path.exists(xcode_stub):
+                return xcode_stub
+            elif os.path.exists(shared):
+                return shared
+            elif os.path.exists(static):
+                return static
+
+        # Oops, didn't find it in *any* of 'dirs'
+        return None
diff --git a/Lib/distutils/util.py b/Lib/distutils/util.py
new file mode 100644
index 000000000..fdcf6faba
--- /dev/null
+++ b/Lib/distutils/util.py
@@ -0,0 +1,557 @@
+"""distutils.util
+
+Miscellaneous utility functions -- anything that doesn't fit into
+one of the other *util.py modules.
+"""
+
+import os
+import re
+import importlib.util
+import string
+import sys
+from distutils.errors import DistutilsPlatformError
+from distutils.dep_util import newer
+from distutils.spawn import spawn
+from distutils import log
+from distutils.errors import DistutilsByteCompileError
+
+def get_platform ():
+    """Return a string that identifies the current platform.  This is used
+    mainly to distinguish platform-specific build directories and
+    platform-specific built distributions.  Typically includes the OS name
+    and version and the architecture (as supplied by 'os.uname()'),
+    although the exact information included depends on the OS; eg. for IRIX
+    the architecture isn't particularly important (IRIX only runs on SGI
+    hardware), but for Linux the kernel version isn't particularly
+    important.
+
+    Examples of returned values:
+       linux-i586
+       linux-alpha (?)
+       solaris-2.6-sun4u
+       irix-5.3
+       irix64-6.2
+
+    Windows will return one of:
+       win-amd64 (64bit Windows on AMD64 (aka x86_64, Intel64, EM64T, etc)
+       win-ia64 (64bit Windows on Itanium)
+       win32 (all others - specifically, sys.platform is returned)
+
+    For other non-POSIX platforms, currently just returns 'sys.platform'.
+    """
+    if os.name == 'nt':
+        # sniff sys.version for architecture.
+        prefix = " bit ("
+        i = sys.version.find(prefix)
+        if i == -1:
+            return sys.platform
+        j = sys.version.find(")", i)
+        look = sys.version[i+len(prefix):j].lower()
+        if look == 'amd64':
+            return 'win-amd64'
+        if look == 'itanium':
+            return 'win-ia64'
+        return sys.platform
+
+    # Set for cross builds explicitly
+    if "_PYTHON_HOST_PLATFORM" in os.environ:
+        return os.environ["_PYTHON_HOST_PLATFORM"]
+
+    if os.name != "posix" or not hasattr(os, 'uname'):
+        # XXX what about the architecture? NT is Intel or Alpha,
+        # Mac OS is M68k or PPC, etc.
+        return sys.platform
+
+    # Try to distinguish various flavours of Unix
+
+    (osname, host, release, version, machine) = os.uname()
+
+    # Convert the OS name to lowercase, remove '/' characters
+    # (to accommodate BSD/OS), and translate spaces (for "Power Macintosh")
+    osname = osname.lower().replace('/', '')
+    machine = machine.replace(' ', '_')
+    machine = machine.replace('/', '-')
+
+    if osname[:5] == "linux":
+        # At least on Linux/Intel, 'machine' is the processor --
+        # i386, etc.
+        # XXX what about Alpha, SPARC, etc?
+        return  "%s-%s" % (osname, machine)
+    elif osname[:5] == "sunos":
+        if release[0] >= "5":           # SunOS 5 == Solaris 2
+            osname = "solaris"
+            release = "%d.%s" % (int(release[0]) - 3, release[2:])
+            # We can't use "platform.architecture()[0]" because a
+            # bootstrap problem. We use a dict to get an error
+            # if some suspicious happens.
+            bitness = {2147483647:"32bit", 9223372036854775807:"64bit"}
+            machine += ".%s" % bitness[sys.maxsize]
+        # fall through to standard osname-release-machine representation
+    elif osname[:4] == "irix":              # could be "irix64"!
+        return "%s-%s" % (osname, release)
+    elif osname[:3] == "aix":
+        return "%s-%s.%s" % (osname, version, release)
+    elif osname[:6] == "cygwin":
+        osname = "cygwin"
+        rel_re = re.compile (r'[\d.]+', re.ASCII)
+        m = rel_re.match(release)
+        if m:
+            release = m.group()
+    elif osname[:6] == "darwin":
+        import _osx_support, distutils.sysconfig
+        osname, release, machine = _osx_support.get_platform_osx(
+                                        distutils.sysconfig.get_config_vars(),
+                                        osname, release, machine)
+
+    return "%s-%s-%s" % (osname, release, machine)
+
+# get_platform ()
+
+
+def convert_path (pathname):
+    """Return 'pathname' as a name that will work on the native filesystem,
+    i.e. split it on '/' and put it back together again using the current
+    directory separator.  Needed because filenames in the setup script are
+    always supplied in Unix style, and have to be converted to the local
+    convention before we can actually use them in the filesystem.  Raises
+    ValueError on non-Unix-ish systems if 'pathname' either starts or
+    ends with a slash.
+    """
+    if os.sep == '/':
+        return pathname
+    if not pathname:
+        return pathname
+    if pathname[0] == '/':
+        raise ValueError("path '%s' cannot be absolute" % pathname)
+    if pathname[-1] == '/':
+        raise ValueError("path '%s' cannot end with '/'" % pathname)
+
+    paths = pathname.split('/')
+    while '.' in paths:
+        paths.remove('.')
+    if not paths:
+        return os.curdir
+    return os.path.join(*paths)
+
+# convert_path ()
+
+
+def change_root (new_root, pathname):
+    """Return 'pathname' with 'new_root' prepended.  If 'pathname' is
+    relative, this is equivalent to "os.path.join(new_root,pathname)".
+    Otherwise, it requires making 'pathname' relative and then joining the
+    two, which is tricky on DOS/Windows and Mac OS.
+    """
+    if os.name == 'posix':
+        if not os.path.isabs(pathname):
+            return os.path.join(new_root, pathname)
+        else:
+            return os.path.join(new_root, pathname[1:])
+
+    elif os.name == 'nt':
+        (drive, path) = os.path.splitdrive(pathname)
+        if path[0] == '\\':
+            path = path[1:]
+        return os.path.join(new_root, path)
+
+    else:
+        raise DistutilsPlatformError("nothing known about platform '%s'" % os.name)
+
+
+_environ_checked = 0
+def check_environ ():
+    """Ensure that 'os.environ' has all the environment variables we
+    guarantee that users can use in config files, command-line options,
+    etc.  Currently this includes:
+      HOME - user's home directory (Unix only)
+      PLAT - description of the current platform, including hardware
+             and OS (see 'get_platform()')
+    """
+    global _environ_checked
+    if _environ_checked:
+        return
+
+    if os.name == 'posix' and 'HOME' not in os.environ:
+        import pwd
+        os.environ['HOME'] = pwd.getpwuid(os.getuid())[5]
+
+    if 'PLAT' not in os.environ:
+        os.environ['PLAT'] = get_platform()
+
+    _environ_checked = 1
+
+
+def subst_vars (s, local_vars):
+    """Perform shell/Perl-style variable substitution on 'string'.  Every
+    occurrence of '$' followed by a name is considered a variable, and
+    variable is substituted by the value found in the 'local_vars'
+    dictionary, or in 'os.environ' if it's not in 'local_vars'.
+    'os.environ' is first checked/augmented to guarantee that it contains
+    certain values: see 'check_environ()'.  Raise ValueError for any
+    variables not found in either 'local_vars' or 'os.environ'.
+    """
+    check_environ()
+    def _subst (match, local_vars=local_vars):
+        var_name = match.group(1)
+        if var_name in local_vars:
+            return str(local_vars[var_name])
+        else:
+            return os.environ[var_name]
+
+    try:
+        return re.sub(r'\$([a-zA-Z_][a-zA-Z_0-9]*)', _subst, s)
+    except KeyError as var:
+        raise ValueError("invalid variable '$%s'" % var)
+
+# subst_vars ()
+
+
+def grok_environment_error (exc, prefix="error: "):
+    # Function kept for backward compatibility.
+    # Used to try clever things with EnvironmentErrors,
+    # but nowadays str(exception) produces good messages.
+    return prefix + str(exc)
+
+
+# Needed by 'split_quoted()'
+_wordchars_re = _squote_re = _dquote_re = None
+def _init_regex():
+    global _wordchars_re, _squote_re, _dquote_re
+    _wordchars_re = re.compile(r'[^\\\'\"%s ]*' % string.whitespace)
+    _squote_re = re.compile(r"'(?:[^'\\]|\\.)*'")
+    _dquote_re = re.compile(r'"(?:[^"\\]|\\.)*"')
+
+def split_quoted (s):
+    """Split a string up according to Unix shell-like rules for quotes and
+    backslashes.  In short: words are delimited by spaces, as long as those
+    spaces are not escaped by a backslash, or inside a quoted string.
+    Single and double quotes are equivalent, and the quote characters can
+    be backslash-escaped.  The backslash is stripped from any two-character
+    escape sequence, leaving only the escaped character.  The quote
+    characters are stripped from any quoted string.  Returns a list of
+    words.
+    """
+
+    # This is a nice algorithm for splitting up a single string, since it
+    # doesn't require character-by-character examination.  It was a little
+    # bit of a brain-bender to get it working right, though...
+    if _wordchars_re is None: _init_regex()
+
+    s = s.strip()
+    words = []
+    pos = 0
+
+    while s:
+        m = _wordchars_re.match(s, pos)
+        end = m.end()
+        if end == len(s):
+            words.append(s[:end])
+            break
+
+        if s[end] in string.whitespace: # unescaped, unquoted whitespace: now
+            words.append(s[:end])       # we definitely have a word delimiter
+            s = s[end:].lstrip()
+            pos = 0
+
+        elif s[end] == '\\':            # preserve whatever is being escaped;
+                                        # will become part of the current word
+            s = s[:end] + s[end+1:]
+            pos = end+1
+
+        else:
+            if s[end] == "'":           # slurp singly-quoted string
+                m = _squote_re.match(s, end)
+            elif s[end] == '"':         # slurp doubly-quoted string
+                m = _dquote_re.match(s, end)
+            else:
+                raise RuntimeError("this can't happen (bad char '%c')" % s[end])
+
+            if m is None:
+                raise ValueError("bad string (mismatched %s quotes?)" % s[end])
+
+            (beg, end) = m.span()
+            s = s[:beg] + s[beg+1:end-1] + s[end:]
+            pos = m.end() - 2
+
+        if pos >= len(s):
+            words.append(s)
+            break
+
+    return words
+
+# split_quoted ()
+
+
+def execute (func, args, msg=None, verbose=0, dry_run=0):
+    """Perform some action that affects the outside world (eg.  by
+    writing to the filesystem).  Such actions are special because they
+    are disabled by the 'dry_run' flag.  This method takes care of all
+    that bureaucracy for you; all you have to do is supply the
+    function to call and an argument tuple for it (to embody the
+    "external action" being performed), and an optional message to
+    print.
+    """
+    if msg is None:
+        msg = "%s%r" % (func.__name__, args)
+        if msg[-2:] == ',)':        # correct for singleton tuple
+            msg = msg[0:-2] + ')'
+
+    log.info(msg)
+    if not dry_run:
+        func(*args)
+
+
+def strtobool (val):
+    """Convert a string representation of truth to true (1) or false (0).
+
+    True values are 'y', 'yes', 't', 'true', 'on', and '1'; false values
+    are 'n', 'no', 'f', 'false', 'off', and '0'.  Raises ValueError if
+    'val' is anything else.
+    """
+    val = val.lower()
+    if val in ('y', 'yes', 't', 'true', 'on', '1'):
+        return 1
+    elif val in ('n', 'no', 'f', 'false', 'off', '0'):
+        return 0
+    else:
+        raise ValueError("invalid truth value %r" % (val,))
+
+
+def byte_compile (py_files,
+                  optimize=0, force=0,
+                  prefix=None, base_dir=None,
+                  verbose=1, dry_run=0,
+                  direct=None):
+    """Byte-compile a collection of Python source files to .pyc
+    files in a __pycache__ subdirectory.  'py_files' is a list
+    of files to compile; any files that don't end in ".py" are silently
+    skipped.  'optimize' must be one of the following:
+      0 - don't optimize
+      1 - normal optimization (like "python -O")
+      2 - extra optimization (like "python -OO")
+    If 'force' is true, all files are recompiled regardless of
+    timestamps.
+
+    The source filename encoded in each bytecode file defaults to the
+    filenames listed in 'py_files'; you can modify these with 'prefix' and
+    'basedir'.  'prefix' is a string that will be stripped off of each
+    source filename, and 'base_dir' is a directory name that will be
+    prepended (after 'prefix' is stripped).  You can supply either or both
+    (or neither) of 'prefix' and 'base_dir', as you wish.
+
+    If 'dry_run' is true, doesn't actually do anything that would
+    affect the filesystem.
+
+    Byte-compilation is either done directly in this interpreter process
+    with the standard py_compile module, or indirectly by writing a
+    temporary script and executing it.  Normally, you should let
+    'byte_compile()' figure out to use direct compilation or not (see
+    the source for details).  The 'direct' flag is used by the script
+    generated in indirect mode; unless you know what you're doing, leave
+    it set to None.
+    """
+
+    # Late import to fix a bootstrap issue: _posixsubprocess is built by
+    # setup.py, but setup.py uses distutils.
+    import subprocess
+
+    # nothing is done if sys.dont_write_bytecode is True
+    if sys.dont_write_bytecode:
+        raise DistutilsByteCompileError('byte-compiling is disabled.')
+
+    # First, if the caller didn't force us into direct or indirect mode,
+    # figure out which mode we should be in.  We take a conservative
+    # approach: choose direct mode *only* if the current interpreter is
+    # in debug mode and optimize is 0.  If we're not in debug mode (-O
+    # or -OO), we don't know which level of optimization this
+    # interpreter is running with, so we can't do direct
+    # byte-compilation and be certain that it's the right thing.  Thus,
+    # always compile indirectly if the current interpreter is in either
+    # optimize mode, or if either optimization level was requested by
+    # the caller.
+    if direct is None:
+        direct = (__debug__ and optimize == 0)
+
+    # "Indirect" byte-compilation: write a temporary script and then
+    # run it with the appropriate flags.
+    if not direct:
+        try:
+            from tempfile import mkstemp
+            (script_fd, script_name) = mkstemp(".py")
+        except ImportError:
+            from tempfile import mktemp
+            (script_fd, script_name) = None, mktemp(".py")
+        log.info("writing byte-compilation script '%s'", script_name)
+        if not dry_run:
+            if script_fd is not None:
+                script = os.fdopen(script_fd, "w")
+            else:
+                script = open(script_name, "w")
+
+            script.write("""\
+from distutils.util import byte_compile
+files = [
+""")
+
+            # XXX would be nice to write absolute filenames, just for
+            # safety's sake (script should be more robust in the face of
+            # chdir'ing before running it).  But this requires abspath'ing
+            # 'prefix' as well, and that breaks the hack in build_lib's
+            # 'byte_compile()' method that carefully tacks on a trailing
+            # slash (os.sep really) to make sure the prefix here is "just
+            # right".  This whole prefix business is rather delicate -- the
+            # problem is that it's really a directory, but I'm treating it
+            # as a dumb string, so trailing slashes and so forth matter.
+
+            #py_files = map(os.path.abspath, py_files)
+            #if prefix:
+            #    prefix = os.path.abspath(prefix)
+
+            script.write(",\n".join(map(repr, py_files)) + "]\n")
+            script.write("""
+byte_compile(files, optimize=%r, force=%r,
+             prefix=%r, base_dir=%r,
+             verbose=%r, dry_run=0,
+             direct=1)
+""" % (optimize, force, prefix, base_dir, verbose))
+
+            script.close()
+
+        cmd = [sys.executable]
+        cmd.extend(subprocess._optim_args_from_interpreter_flags())
+        cmd.append(script_name)
+        spawn(cmd, dry_run=dry_run)
+        execute(os.remove, (script_name,), "removing %s" % script_name,
+                dry_run=dry_run)
+
+    # "Direct" byte-compilation: use the py_compile module to compile
+    # right here, right now.  Note that the script generated in indirect
+    # mode simply calls 'byte_compile()' in direct mode, a weird sort of
+    # cross-process recursion.  Hey, it works!
+    else:
+        from py_compile import compile
+
+        for file in py_files:
+            if file[-3:] != ".py":
+                # This lets us be lazy and not filter filenames in
+                # the "install_lib" command.
+                continue
+
+            # Terminology from the py_compile module:
+            #   cfile - byte-compiled file
+            #   dfile - purported source filename (same as 'file' by default)
+            if optimize >= 0:
+                opt = '' if optimize == 0 else optimize
+                cfile = importlib.util.cache_from_source(
+                    file, optimization=opt)
+            else:
+                cfile = importlib.util.cache_from_source(file)
+            dfile = file
+            if prefix:
+                if file[:len(prefix)] != prefix:
+                    raise ValueError("invalid prefix: filename %r doesn't start with %r"
+                           % (file, prefix))
+                dfile = dfile[len(prefix):]
+            if base_dir:
+                dfile = os.path.join(base_dir, dfile)
+
+            cfile_base = os.path.basename(cfile)
+            if direct:
+                if force or newer(file, cfile):
+                    log.info("byte-compiling %s to %s", file, cfile_base)
+                    if not dry_run:
+                        compile(file, cfile, dfile)
+                else:
+                    log.debug("skipping byte-compilation of %s to %s",
+                              file, cfile_base)
+
+# byte_compile ()
+
+def rfc822_escape (header):
+    """Return a version of the string escaped for inclusion in an
+    RFC-822 header, by ensuring there are 8 spaces space after each newline.
+    """
+    lines = header.split('\n')
+    sep = '\n' + 8 * ' '
+    return sep.join(lines)
+
+# 2to3 support
+
+def run_2to3(files, fixer_names=None, options=None, explicit=None):
+    """Invoke 2to3 on a list of Python files.
+    The files should all come from the build area, as the
+    modification is done in-place. To reduce the build time,
+    only files modified since the last invocation of this
+    function should be passed in the files argument."""
+
+    if not files:
+        return
+
+    # Make this class local, to delay import of 2to3
+    from lib2to3.refactor import RefactoringTool, get_fixers_from_package
+    class DistutilsRefactoringTool(RefactoringTool):
+        def log_error(self, msg, *args, **kw):
+            log.error(msg, *args)
+
+        def log_message(self, msg, *args):
+            log.info(msg, *args)
+
+        def log_debug(self, msg, *args):
+            log.debug(msg, *args)
+
+    if fixer_names is None:
+        fixer_names = get_fixers_from_package('lib2to3.fixes')
+    r = DistutilsRefactoringTool(fixer_names, options=options)
+    r.refactor(files, write=True)
+
+def copydir_run_2to3(src, dest, template=None, fixer_names=None,
+                     options=None, explicit=None):
+    """Recursively copy a directory, only copying new and changed files,
+    running run_2to3 over all newly copied Python modules afterward.
+
+    If you give a template string, it's parsed like a MANIFEST.in.
+    """
+    from distutils.dir_util import mkpath
+    from distutils.file_util import copy_file
+    from distutils.filelist import FileList
+    filelist = FileList()
+    curdir = os.getcwd()
+    os.chdir(src)
+    try:
+        filelist.findall()
+    finally:
+        os.chdir(curdir)
+    filelist.files[:] = filelist.allfiles
+    if template:
+        for line in template.splitlines():
+            line = line.strip()
+            if not line: continue
+            filelist.process_template_line(line)
+    copied = []
+    for filename in filelist.files:
+        outname = os.path.join(dest, filename)
+        mkpath(os.path.dirname(outname))
+        res = copy_file(os.path.join(src, filename), outname, update=1)
+        if res[1]: copied.append(outname)
+    run_2to3([fn for fn in copied if fn.lower().endswith('.py')],
+             fixer_names=fixer_names, options=options, explicit=explicit)
+    return copied
+
+class Mixin2to3:
+    '''Mixin class for commands that run 2to3.
+    To configure 2to3, setup scripts may either change
+    the class variables, or inherit from individual commands
+    to override how 2to3 is invoked.'''
+
+    # provide list of fixers to run;
+    # defaults to all from lib2to3.fixers
+    fixer_names = None
+
+    # options dictionary
+    options = None
+
+    # list of fixers to invoke even though they are marked as explicit
+    explicit = None
+
+    def run_2to3(self, files):
+        return run_2to3(files, self.fixer_names, self.options, self.explicit)
diff --git a/Lib/distutils/version.py b/Lib/distutils/version.py
new file mode 100644
index 000000000..af14cc134
--- /dev/null
+++ b/Lib/distutils/version.py
@@ -0,0 +1,343 @@
+#
+# distutils/version.py
+#
+# Implements multiple version numbering conventions for the
+# Python Module Distribution Utilities.
+#
+# $Id$
+#
+
+"""Provides classes to represent module version numbers (one class for
+each style of version numbering).  There are currently two such classes
+implemented: StrictVersion and LooseVersion.
+
+Every version number class implements the following interface:
+  * the 'parse' method takes a string and parses it to some internal
+    representation; if the string is an invalid version number,
+    'parse' raises a ValueError exception
+  * the class constructor takes an optional string argument which,
+    if supplied, is passed to 'parse'
+  * __str__ reconstructs the string that was passed to 'parse' (or
+    an equivalent string -- ie. one that will generate an equivalent
+    version number instance)
+  * __repr__ generates Python code to recreate the version number instance
+  * _cmp compares the current instance with either another instance
+    of the same class or a string (which will be parsed to an instance
+    of the same class, thus must follow the same rules)
+"""
+
+import re
+
+class Version:
+    """Abstract base class for version numbering classes.  Just provides
+    constructor (__init__) and reproducer (__repr__), because those
+    seem to be the same for all version numbering classes; and route
+    rich comparisons to _cmp.
+    """
+
+    def __init__ (self, vstring=None):
+        if vstring:
+            self.parse(vstring)
+
+    def __repr__ (self):
+        return "%s ('%s')" % (self.__class__.__name__, str(self))
+
+    def __eq__(self, other):
+        c = self._cmp(other)
+        if c is NotImplemented:
+            return c
+        return c == 0
+
+    def __lt__(self, other):
+        c = self._cmp(other)
+        if c is NotImplemented:
+            return c
+        return c < 0
+
+    def __le__(self, other):
+        c = self._cmp(other)
+        if c is NotImplemented:
+            return c
+        return c <= 0
+
+    def __gt__(self, other):
+        c = self._cmp(other)
+        if c is NotImplemented:
+            return c
+        return c > 0
+
+    def __ge__(self, other):
+        c = self._cmp(other)
+        if c is NotImplemented:
+            return c
+        return c >= 0
+
+
+# Interface for version-number classes -- must be implemented
+# by the following classes (the concrete ones -- Version should
+# be treated as an abstract class).
+#    __init__ (string) - create and take same action as 'parse'
+#                        (string parameter is optional)
+#    parse (string)    - convert a string representation to whatever
+#                        internal representation is appropriate for
+#                        this style of version numbering
+#    __str__ (self)    - convert back to a string; should be very similar
+#                        (if not identical to) the string supplied to parse
+#    __repr__ (self)   - generate Python code to recreate
+#                        the instance
+#    _cmp (self, other) - compare two version numbers ('other' may
+#                        be an unparsed version string, or another
+#                        instance of your version class)
+
+
+class StrictVersion (Version):
+
+    """Version numbering for anal retentives and software idealists.
+    Implements the standard interface for version number classes as
+    described above.  A version number consists of two or three
+    dot-separated numeric components, with an optional "pre-release" tag
+    on the end.  The pre-release tag consists of the letter 'a' or 'b'
+    followed by a number.  If the numeric components of two version
+    numbers are equal, then one with a pre-release tag will always
+    be deemed earlier (lesser) than one without.
+
+    The following are valid version numbers (shown in the order that
+    would be obtained by sorting according to the supplied cmp function):
+
+        0.4       0.4.0  (these two are equivalent)
+        0.4.1
+        0.5a1
+        0.5b3
+        0.5
+        0.9.6
+        1.0
+        1.0.4a3
+        1.0.4b1
+        1.0.4
+
+    The following are examples of invalid version numbers:
+
+        1
+        2.7.2.2
+        1.3.a4
+        1.3pl1
+        1.3c4
+
+    The rationale for this version numbering system will be explained
+    in the distutils documentation.
+    """
+
+    version_re = re.compile(r'^(\d+) \. (\d+) (\. (\d+))? ([ab](\d+))?$',
+                            re.VERBOSE | re.ASCII)
+
+
+    def parse (self, vstring):
+        match = self.version_re.match(vstring)
+        if not match:
+            raise ValueError("invalid version number '%s'" % vstring)
+
+        (major, minor, patch, prerelease, prerelease_num) = \
+            match.group(1, 2, 4, 5, 6)
+
+        if patch:
+            self.version = tuple(map(int, [major, minor, patch]))
+        else:
+            self.version = tuple(map(int, [major, minor])) + (0,)
+
+        if prerelease:
+            self.prerelease = (prerelease[0], int(prerelease_num))
+        else:
+            self.prerelease = None
+
+
+    def __str__ (self):
+
+        if self.version[2] == 0:
+            vstring = '.'.join(map(str, self.version[0:2]))
+        else:
+            vstring = '.'.join(map(str, self.version))
+
+        if self.prerelease:
+            vstring = vstring + self.prerelease[0] + str(self.prerelease[1])
+
+        return vstring
+
+
+    def _cmp (self, other):
+        if isinstance(other, str):
+            other = StrictVersion(other)
+
+        if self.version != other.version:
+            # numeric versions don't match
+            # prerelease stuff doesn't matter
+            if self.version < other.version:
+                return -1
+            else:
+                return 1
+
+        # have to compare prerelease
+        # case 1: neither has prerelease; they're equal
+        # case 2: self has prerelease, other doesn't; other is greater
+        # case 3: self doesn't have prerelease, other does: self is greater
+        # case 4: both have prerelease: must compare them!
+
+        if (not self.prerelease and not other.prerelease):
+            return 0
+        elif (self.prerelease and not other.prerelease):
+            return -1
+        elif (not self.prerelease and other.prerelease):
+            return 1
+        elif (self.prerelease and other.prerelease):
+            if self.prerelease == other.prerelease:
+                return 0
+            elif self.prerelease < other.prerelease:
+                return -1
+            else:
+                return 1
+        else:
+            assert False, "never get here"
+
+# end class StrictVersion
+
+
+# The rules according to Greg Stein:
+# 1) a version number has 1 or more numbers separated by a period or by
+#    sequences of letters. If only periods, then these are compared
+#    left-to-right to determine an ordering.
+# 2) sequences of letters are part of the tuple for comparison and are
+#    compared lexicographically
+# 3) recognize the numeric components may have leading zeroes
+#
+# The LooseVersion class below implements these rules: a version number
+# string is split up into a tuple of integer and string components, and
+# comparison is a simple tuple comparison.  This means that version
+# numbers behave in a predictable and obvious way, but a way that might
+# not necessarily be how people *want* version numbers to behave.  There
+# wouldn't be a problem if people could stick to purely numeric version
+# numbers: just split on period and compare the numbers as tuples.
+# However, people insist on putting letters into their version numbers;
+# the most common purpose seems to be:
+#   - indicating a "pre-release" version
+#     ('alpha', 'beta', 'a', 'b', 'pre', 'p')
+#   - indicating a post-release patch ('p', 'pl', 'patch')
+# but of course this can't cover all version number schemes, and there's
+# no way to know what a programmer means without asking him.
+#
+# The problem is what to do with letters (and other non-numeric
+# characters) in a version number.  The current implementation does the
+# obvious and predictable thing: keep them as strings and compare
+# lexically within a tuple comparison.  This has the desired effect if
+# an appended letter sequence implies something "post-release":
+# eg. "0.99" < "0.99pl14" < "1.0", and "5.001" < "5.001m" < "5.002".
+#
+# However, if letters in a version number imply a pre-release version,
+# the "obvious" thing isn't correct.  Eg. you would expect that
+# "1.5.1" < "1.5.2a2" < "1.5.2", but under the tuple/lexical comparison
+# implemented here, this just isn't so.
+#
+# Two possible solutions come to mind.  The first is to tie the
+# comparison algorithm to a particular set of semantic rules, as has
+# been done in the StrictVersion class above.  This works great as long
+# as everyone can go along with bondage and discipline.  Hopefully a
+# (large) subset of Python module programmers will agree that the
+# particular flavour of bondage and discipline provided by StrictVersion
+# provides enough benefit to be worth using, and will submit their
+# version numbering scheme to its domination.  The free-thinking
+# anarchists in the lot will never give in, though, and something needs
+# to be done to accommodate them.
+#
+# Perhaps a "moderately strict" version class could be implemented that
+# lets almost anything slide (syntactically), and makes some heuristic
+# assumptions about non-digits in version number strings.  This could
+# sink into special-case-hell, though; if I was as talented and
+# idiosyncratic as Larry Wall, I'd go ahead and implement a class that
+# somehow knows that "1.2.1" < "1.2.2a2" < "1.2.2" < "1.2.2pl3", and is
+# just as happy dealing with things like "2g6" and "1.13++".  I don't
+# think I'm smart enough to do it right though.
+#
+# In any case, I've coded the test suite for this module (see
+# ../test/test_version.py) specifically to fail on things like comparing
+# "1.2a2" and "1.2".  That's not because the *code* is doing anything
+# wrong, it's because the simple, obvious design doesn't match my
+# complicated, hairy expectations for real-world version numbers.  It
+# would be a snap to fix the test suite to say, "Yep, LooseVersion does
+# the Right Thing" (ie. the code matches the conception).  But I'd rather
+# have a conception that matches common notions about version numbers.
+
+class LooseVersion (Version):
+
+    """Version numbering for anarchists and software realists.
+    Implements the standard interface for version number classes as
+    described above.  A version number consists of a series of numbers,
+    separated by either periods or strings of letters.  When comparing
+    version numbers, the numeric components will be compared
+    numerically, and the alphabetic components lexically.  The following
+    are all valid version numbers, in no particular order:
+
+        1.5.1
+        1.5.2b2
+        161
+        3.10a
+        8.02
+        3.4j
+        1996.07.12
+        3.2.pl0
+        3.1.1.6
+        2g6
+        11g
+        0.960923
+        2.2beta29
+        1.13++
+        5.5.kw
+        2.0b1pl0
+
+    In fact, there is no such thing as an invalid version number under
+    this scheme; the rules for comparison are simple and predictable,
+    but may not always give the results you want (for some definition
+    of "want").
+    """
+
+    component_re = re.compile(r'(\d+ | [a-z]+ | \.)', re.VERBOSE)
+
+    def __init__ (self, vstring=None):
+        if vstring:
+            self.parse(vstring)
+
+
+    def parse (self, vstring):
+        # I've given up on thinking I can reconstruct the version string
+        # from the parsed tuple -- so I just store the string here for
+        # use by __str__
+        self.vstring = vstring
+        components = [x for x in self.component_re.split(vstring)
+                              if x and x != '.']
+        for i, obj in enumerate(components):
+            try:
+                components[i] = int(obj)
+            except ValueError:
+                pass
+
+        self.version = components
+
+
+    def __str__ (self):
+        return self.vstring
+
+
+    def __repr__ (self):
+        return "LooseVersion ('%s')" % str(self)
+
+
+    def _cmp (self, other):
+        if isinstance(other, str):
+            other = LooseVersion(other)
+
+        if self.version == other.version:
+            return 0
+        if self.version < other.version:
+            return -1
+        if self.version > other.version:
+            return 1
+
+
+# end class LooseVersion
diff --git a/Lib/distutils/versionpredicate.py b/Lib/distutils/versionpredicate.py
new file mode 100644
index 000000000..062c98f24
--- /dev/null
+++ b/Lib/distutils/versionpredicate.py
@@ -0,0 +1,166 @@
+"""Module for parsing and testing package version predicate strings.
+"""
+import re
+import distutils.version
+import operator
+
+
+re_validPackage = re.compile(r"(?i)^\s*([a-z_]\w*(?:\.[a-z_]\w*)*)(.*)",
+    re.ASCII)
+# (package) (rest)
+
+re_paren = re.compile(r"^\s*\((.*)\)\s*$") # (list) inside of parentheses
+re_splitComparison = re.compile(r"^\s*(<=|>=|<|>|!=|==)\s*([^\s,]+)\s*$")
+# (comp) (version)
+
+
+def splitUp(pred):
+    """Parse a single version comparison.
+
+    Return (comparison string, StrictVersion)
+    """
+    res = re_splitComparison.match(pred)
+    if not res:
+        raise ValueError("bad package restriction syntax: %r" % pred)
+    comp, verStr = res.groups()
+    return (comp, distutils.version.StrictVersion(verStr))
+
+compmap = {"<": operator.lt, "<=": operator.le, "==": operator.eq,
+           ">": operator.gt, ">=": operator.ge, "!=": operator.ne}
+
+class VersionPredicate:
+    """Parse and test package version predicates.
+
+    >>> v = VersionPredicate('pyepat.abc (>1.0, <3333.3a1, !=1555.1b3)')
+
+    The `name` attribute provides the full dotted name that is given::
+
+    >>> v.name
+    'pyepat.abc'
+
+    The str() of a `VersionPredicate` provides a normalized
+    human-readable version of the expression::
+
+    >>> print(v)
+    pyepat.abc (> 1.0, < 3333.3a1, != 1555.1b3)
+
+    The `satisfied_by()` method can be used to determine with a given
+    version number is included in the set described by the version
+    restrictions::
+
+    >>> v.satisfied_by('1.1')
+    True
+    >>> v.satisfied_by('1.4')
+    True
+    >>> v.satisfied_by('1.0')
+    False
+    >>> v.satisfied_by('4444.4')
+    False
+    >>> v.satisfied_by('1555.1b3')
+    False
+
+    `VersionPredicate` is flexible in accepting extra whitespace::
+
+    >>> v = VersionPredicate(' pat( ==  0.1  )  ')
+    >>> v.name
+    'pat'
+    >>> v.satisfied_by('0.1')
+    True
+    >>> v.satisfied_by('0.2')
+    False
+
+    If any version numbers passed in do not conform to the
+    restrictions of `StrictVersion`, a `ValueError` is raised::
+
+    >>> v = VersionPredicate('p1.p2.p3.p4(>=1.0, <=1.3a1, !=1.2zb3)')
+    Traceback (most recent call last):
+      ...
+    ValueError: invalid version number '1.2zb3'
+
+    It the module or package name given does not conform to what's
+    allowed as a legal module or package name, `ValueError` is
+    raised::
+
+    >>> v = VersionPredicate('foo-bar')
+    Traceback (most recent call last):
+      ...
+    ValueError: expected parenthesized list: '-bar'
+
+    >>> v = VersionPredicate('foo bar (12.21)')
+    Traceback (most recent call last):
+      ...
+    ValueError: expected parenthesized list: 'bar (12.21)'
+
+    """
+
+    def __init__(self, versionPredicateStr):
+        """Parse a version predicate string.
+        """
+        # Fields:
+        #    name:  package name
+        #    pred:  list of (comparison string, StrictVersion)
+
+        versionPredicateStr = versionPredicateStr.strip()
+        if not versionPredicateStr:
+            raise ValueError("empty package restriction")
+        match = re_validPackage.match(versionPredicateStr)
+        if not match:
+            raise ValueError("bad package name in %r" % versionPredicateStr)
+        self.name, paren = match.groups()
+        paren = paren.strip()
+        if paren:
+            match = re_paren.match(paren)
+            if not match:
+                raise ValueError("expected parenthesized list: %r" % paren)
+            str = match.groups()[0]
+            self.pred = [splitUp(aPred) for aPred in str.split(",")]
+            if not self.pred:
+                raise ValueError("empty parenthesized list in %r"
+                                 % versionPredicateStr)
+        else:
+            self.pred = []
+
+    def __str__(self):
+        if self.pred:
+            seq = [cond + " " + str(ver) for cond, ver in self.pred]
+            return self.name + " (" + ", ".join(seq) + ")"
+        else:
+            return self.name
+
+    def satisfied_by(self, version):
+        """True if version is compatible with all the predicates in self.
+        The parameter version must be acceptable to the StrictVersion
+        constructor.  It may be either a string or StrictVersion.
+        """
+        for cond, ver in self.pred:
+            if not compmap[cond](version, ver):
+                return False
+        return True
+
+
+_provision_rx = None
+
+def split_provision(value):
+    """Return the name and optional version number of a provision.
+
+    The version number, if given, will be returned as a `StrictVersion`
+    instance, otherwise it will be `None`.
+
+    >>> split_provision('mypkg')
+    ('mypkg', None)
+    >>> split_provision(' mypkg( 1.2 ) ')
+    ('mypkg', StrictVersion ('1.2'))
+    """
+    global _provision_rx
+    if _provision_rx is None:
+        _provision_rx = re.compile(
+            r"([a-zA-Z_]\w*(?:\.[a-zA-Z_]\w*)*)(?:\s*\(\s*([^)\s]+)\s*\))?$",
+            re.ASCII)
+    value = value.strip()
+    m = _provision_rx.match(value)
+    if not m:
+        raise ValueError("illegal provides specification: %r" % value)
+    ver = m.group(2) or None
+    if ver:
+        ver = distutils.version.StrictVersion(ver)
+    return m.group(1), ver
diff --git a/Lib/email/__init__.py b/Lib/email/__init__.py
new file mode 100644
index 000000000..fae872439
--- /dev/null
+++ b/Lib/email/__init__.py
@@ -0,0 +1,62 @@
+# Copyright (C) 2001-2007 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""A package for parsing, handling, and generating email messages."""
+
+__all__ = [
+    'base64mime',
+    'charset',
+    'encoders',
+    'errors',
+    'feedparser',
+    'generator',
+    'header',
+    'iterators',
+    'message',
+    'message_from_file',
+    'message_from_binary_file',
+    'message_from_string',
+    'message_from_bytes',
+    'mime',
+    'parser',
+    'quoprimime',
+    'utils',
+    ]
+
+
+
+# Some convenience routines.  Don't import Parser and Message as side-effects
+# of importing email since those cascadingly import most of the rest of the
+# email package.
+def message_from_string(s, *args, **kws):
+    """Parse a string into a Message object model.
+
+    Optional _class and strict are passed to the Parser constructor.
+    """
+    from email.parser import Parser
+    return Parser(*args, **kws).parsestr(s)
+
+def message_from_bytes(s, *args, **kws):
+    """Parse a bytes string into a Message object model.
+
+    Optional _class and strict are passed to the Parser constructor.
+    """
+    from email.parser import BytesParser
+    return BytesParser(*args, **kws).parsebytes(s)
+
+def message_from_file(fp, *args, **kws):
+    """Read a file and parse its contents into a Message object model.
+
+    Optional _class and strict are passed to the Parser constructor.
+    """
+    from email.parser import Parser
+    return Parser(*args, **kws).parse(fp)
+
+def message_from_binary_file(fp, *args, **kws):
+    """Read a binary file and parse its contents into a Message object model.
+
+    Optional _class and strict are passed to the Parser constructor.
+    """
+    from email.parser import BytesParser
+    return BytesParser(*args, **kws).parse(fp)
diff --git a/Lib/email/_encoded_words.py b/Lib/email/_encoded_words.py
new file mode 100644
index 000000000..5eaab36ed
--- /dev/null
+++ b/Lib/email/_encoded_words.py
@@ -0,0 +1,221 @@
+""" Routines for manipulating RFC2047 encoded words.
+
+This is currently a package-private API, but will be considered for promotion
+to a public API if there is demand.
+
+"""
+
+# An ecoded word looks like this:
+#
+#        =?charset[*lang]?cte?encoded_string?=
+#
+# for more information about charset see the charset module.  Here it is one
+# of the preferred MIME charset names (hopefully; you never know when parsing).
+# cte (Content Transfer Encoding) is either 'q' or 'b' (ignoring case).  In
+# theory other letters could be used for other encodings, but in practice this
+# (almost?) never happens.  There could be a public API for adding entries
+# to the CTE tables, but YAGNI for now.  'q' is Quoted Printable, 'b' is
+# Base64.  The meaning of encoded_string should be obvious.  'lang' is optional
+# as indicated by the brackets (they are not part of the syntax) but is almost
+# never encountered in practice.
+#
+# The general interface for a CTE decoder is that it takes the encoded_string
+# as its argument, and returns a tuple (cte_decoded_string, defects).  The
+# cte_decoded_string is the original binary that was encoded using the
+# specified cte.  'defects' is a list of MessageDefect instances indicating any
+# problems encountered during conversion.  'charset' and 'lang' are the
+# corresponding strings extracted from the EW, case preserved.
+#
+# The general interface for a CTE encoder is that it takes a binary sequence
+# as input and returns the cte_encoded_string, which is an ascii-only string.
+#
+# Each decoder must also supply a length function that takes the binary
+# sequence as its argument and returns the length of the resulting encoded
+# string.
+#
+# The main API functions for the module are decode, which calls the decoder
+# referenced by the cte specifier, and encode, which adds the appropriate
+# RFC 2047 "chrome" to the encoded string, and can optionally automatically
+# select the shortest possible encoding.  See their docstrings below for
+# details.
+
+import re
+import base64
+import binascii
+import functools
+from string import ascii_letters, digits
+from email import errors
+
+__all__ = ['decode_q',
+           'encode_q',
+           'decode_b',
+           'encode_b',
+           'len_q',
+           'len_b',
+           'decode',
+           'encode',
+           ]
+
+#
+# Quoted Printable
+#
+
+# regex based decoder.
+_q_byte_subber = functools.partial(re.compile(br'=([a-fA-F0-9]{2})').sub,
+        lambda m: bytes([int(m.group(1), 16)]))
+
+def decode_q(encoded):
+    encoded = encoded.replace(b'_', b' ')
+    return _q_byte_subber(encoded), []
+
+
+# dict mapping bytes to their encoded form
+class _QByteMap(dict):
+
+    safe = b'-!*+/' + ascii_letters.encode('ascii') + digits.encode('ascii')
+
+    def __missing__(self, key):
+        if key in self.safe:
+            self[key] = chr(key)
+        else:
+            self[key] = "={:02X}".format(key)
+        return self[key]
+
+_q_byte_map = _QByteMap()
+
+# In headers spaces are mapped to '_'.
+_q_byte_map[ord(' ')] = '_'
+
+def encode_q(bstring):
+    return ''.join(_q_byte_map[x] for x in bstring)
+
+def len_q(bstring):
+    return sum(len(_q_byte_map[x]) for x in bstring)
+
+
+#
+# Base64
+#
+
+def decode_b(encoded):
+    defects = []
+    pad_err = len(encoded) % 4
+    if pad_err:
+        defects.append(errors.InvalidBase64PaddingDefect())
+        padded_encoded = encoded + b'==='[:4-pad_err]
+    else:
+        padded_encoded = encoded
+    try:
+        return base64.b64decode(padded_encoded, validate=True), defects
+    except binascii.Error:
+        # Since we had correct padding, this must an invalid char error.
+        defects = [errors.InvalidBase64CharactersDefect()]
+        # The non-alphabet characters are ignored as far as padding
+        # goes, but we don't know how many there are.  So we'll just
+        # try various padding lengths until something works.
+        for i in 0, 1, 2, 3:
+            try:
+                return base64.b64decode(encoded+b'='*i, validate=False), defects
+            except binascii.Error:
+                if i==0:
+                    defects.append(errors.InvalidBase64PaddingDefect())
+        else:
+            # This should never happen.
+            raise AssertionError("unexpected binascii.Error")
+
+def encode_b(bstring):
+    return base64.b64encode(bstring).decode('ascii')
+
+def len_b(bstring):
+    groups_of_3, leftover = divmod(len(bstring), 3)
+    # 4 bytes out for each 3 bytes (or nonzero fraction thereof) in.
+    return groups_of_3 * 4 + (4 if leftover else 0)
+
+
+_cte_decoders = {
+    'q': decode_q,
+    'b': decode_b,
+    }
+
+def decode(ew):
+    """Decode encoded word and return (string, charset, lang, defects) tuple.
+
+    An RFC 2047/2243 encoded word has the form:
+
+        =?charset*lang?cte?encoded_string?=
+
+    where '*lang' may be omitted but the other parts may not be.
+
+    This function expects exactly such a string (that is, it does not check the
+    syntax and may raise errors if the string is not well formed), and returns
+    the encoded_string decoded first from its Content Transfer Encoding and
+    then from the resulting bytes into unicode using the specified charset.  If
+    the cte-decoded string does not successfully decode using the specified
+    character set, a defect is added to the defects list and the unknown octets
+    are replaced by the unicode 'unknown' character \\uFDFF.
+
+    The specified charset and language are returned.  The default for language,
+    which is rarely if ever encountered, is the empty string.
+
+    """
+    _, charset, cte, cte_string, _ = ew.split('?')
+    charset, _, lang = charset.partition('*')
+    cte = cte.lower()
+    # Recover the original bytes and do CTE decoding.
+    bstring = cte_string.encode('ascii', 'surrogateescape')
+    bstring, defects = _cte_decoders[cte](bstring)
+    # Turn the CTE decoded bytes into unicode.
+    try:
+        string = bstring.decode(charset)
+    except UnicodeError:
+        defects.append(errors.UndecodableBytesDefect("Encoded word "
+            "contains bytes not decodable using {} charset".format(charset)))
+        string = bstring.decode(charset, 'surrogateescape')
+    except LookupError:
+        string = bstring.decode('ascii', 'surrogateescape')
+        if charset.lower() != 'unknown-8bit':
+            defects.append(errors.CharsetError("Unknown charset {} "
+                "in encoded word; decoded as unknown bytes".format(charset)))
+    return string, charset, lang, defects
+
+
+_cte_encoders = {
+    'q': encode_q,
+    'b': encode_b,
+    }
+
+_cte_encode_length = {
+    'q': len_q,
+    'b': len_b,
+    }
+
+def encode(string, charset='utf-8', encoding=None, lang=''):
+    """Encode string using the CTE encoding that produces the shorter result.
+
+    Produces an RFC 2047/2243 encoded word of the form:
+
+        =?charset*lang?cte?encoded_string?=
+
+    where '*lang' is omitted unless the 'lang' parameter is given a value.
+    Optional argument charset (defaults to utf-8) specifies the charset to use
+    to encode the string to binary before CTE encoding it.  Optional argument
+    'encoding' is the cte specifier for the encoding that should be used ('q'
+    or 'b'); if it is None (the default) the encoding which produces the
+    shortest encoded sequence is used, except that 'q' is preferred if it is up
+    to five characters longer.  Optional argument 'lang' (default '') gives the
+    RFC 2243 language string to specify in the encoded word.
+
+    """
+    if charset == 'unknown-8bit':
+        bstring = string.encode('ascii', 'surrogateescape')
+    else:
+        bstring = string.encode(charset)
+    if encoding is None:
+        qlen = _cte_encode_length['q'](bstring)
+        blen = _cte_encode_length['b'](bstring)
+        # Bias toward q.  5 is arbitrary.
+        encoding = 'q' if qlen - blen < 5 else 'b'
+    encoded = _cte_encoders[encoding](bstring)
+    if lang:
+        lang = '*' + lang
+    return "=?{}{}?{}?{}?=".format(charset, lang, encoding, encoded)
diff --git a/Lib/email/_header_value_parser.py b/Lib/email/_header_value_parser.py
new file mode 100644
index 000000000..57d01fbcb
--- /dev/null
+++ b/Lib/email/_header_value_parser.py
@@ -0,0 +1,2970 @@
+"""Header value parser implementing various email-related RFC parsing rules.
+
+The parsing methods defined in this module implement various email related
+parsing rules.  Principal among them is RFC 5322, which is the followon
+to RFC 2822 and primarily a clarification of the former.  It also implements
+RFC 2047 encoded word decoding.
+
+RFC 5322 goes to considerable trouble to maintain backward compatibility with
+RFC 822 in the parse phase, while cleaning up the structure on the generation
+phase.  This parser supports correct RFC 5322 generation by tagging white space
+as folding white space only when folding is allowed in the non-obsolete rule
+sets.  Actually, the parser is even more generous when accepting input than RFC
+5322 mandates, following the spirit of Postel's Law, which RFC 5322 encourages.
+Where possible deviations from the standard are annotated on the 'defects'
+attribute of tokens that deviate.
+
+The general structure of the parser follows RFC 5322, and uses its terminology
+where there is a direct correspondence.  Where the implementation requires a
+somewhat different structure than that used by the formal grammar, new terms
+that mimic the closest existing terms are used.  Thus, it really helps to have
+a copy of RFC 5322 handy when studying this code.
+
+Input to the parser is a string that has already been unfolded according to
+RFC 5322 rules.  According to the RFC this unfolding is the very first step, and
+this parser leaves the unfolding step to a higher level message parser, which
+will have already detected the line breaks that need unfolding while
+determining the beginning and end of each header.
+
+The output of the parser is a TokenList object, which is a list subclass.  A
+TokenList is a recursive data structure.  The terminal nodes of the structure
+are Terminal objects, which are subclasses of str.  These do not correspond
+directly to terminal objects in the formal grammar, but are instead more
+practical higher level combinations of true terminals.
+
+All TokenList and Terminal objects have a 'value' attribute, which produces the
+semantically meaningful value of that part of the parse subtree.  The value of
+all whitespace tokens (no matter how many sub-tokens they may contain) is a
+single space, as per the RFC rules.  This includes 'CFWS', which is herein
+included in the general class of whitespace tokens.  There is one exception to
+the rule that whitespace tokens are collapsed into single spaces in values: in
+the value of a 'bare-quoted-string' (a quoted-string with no leading or
+trailing whitespace), any whitespace that appeared between the quotation marks
+is preserved in the returned value.  Note that in all Terminal strings quoted
+pairs are turned into their unquoted values.
+
+All TokenList and Terminal objects also have a string value, which attempts to
+be a "canonical" representation of the RFC-compliant form of the substring that
+produced the parsed subtree, including minimal use of quoted pair quoting.
+Whitespace runs are not collapsed.
+
+Comment tokens also have a 'content' attribute providing the string found
+between the parens (including any nested comments) with whitespace preserved.
+
+All TokenList and Terminal objects have a 'defects' attribute which is a
+possibly empty list all of the defects found while creating the token.  Defects
+may appear on any token in the tree, and a composite list of all defects in the
+subtree is available through the 'all_defects' attribute of any node.  (For
+Terminal notes x.defects == x.all_defects.)
+
+Each object in a parse tree is called a 'token', and each has a 'token_type'
+attribute that gives the name from the RFC 5322 grammar that it represents.
+Not all RFC 5322 nodes are produced, and there is one non-RFC 5322 node that
+may be produced: 'ptext'.  A 'ptext' is a string of printable ascii characters.
+It is returned in place of lists of (ctext/quoted-pair) and
+(qtext/quoted-pair).
+
+XXX: provide complete list of token types.
+"""
+
+import re
+import urllib   # For urllib.parse.unquote
+from string import hexdigits
+from collections import OrderedDict
+from operator import itemgetter
+from email import _encoded_words as _ew
+from email import errors
+from email import utils
+
+#
+# Useful constants and functions
+#
+
+WSP = set(' \t')
+CFWS_LEADER = WSP | set('(')
+SPECIALS = set(r'()<>@,:;.\"[]')
+ATOM_ENDS = SPECIALS | WSP
+DOT_ATOM_ENDS = ATOM_ENDS - set('.')
+# '.', '"', and '(' do not end phrases in order to support obs-phrase
+PHRASE_ENDS = SPECIALS - set('."(')
+TSPECIALS = (SPECIALS | set('/?=')) - set('.')
+TOKEN_ENDS = TSPECIALS | WSP
+ASPECIALS = TSPECIALS | set("*'%")
+ATTRIBUTE_ENDS = ASPECIALS | WSP
+EXTENDED_ATTRIBUTE_ENDS = ATTRIBUTE_ENDS - set('%')
+
+def quote_string(value):
+    return '"'+str(value).replace('\\', '\\\\').replace('"', r'\"')+'"'
+
+#
+# Accumulator for header folding
+#
+
+class _Folded:
+
+    def __init__(self, maxlen, policy):
+        self.maxlen = maxlen
+        self.policy = policy
+        self.lastlen = 0
+        self.stickyspace = None
+        self.firstline = True
+        self.done = []
+        self.current = []
+
+    def newline(self):
+        self.done.extend(self.current)
+        self.done.append(self.policy.linesep)
+        self.current.clear()
+        self.lastlen = 0
+
+    def finalize(self):
+        if self.current:
+            self.newline()
+
+    def __str__(self):
+        return ''.join(self.done)
+
+    def append(self, stoken):
+        self.current.append(stoken)
+
+    def append_if_fits(self, token, stoken=None):
+        if stoken is None:
+            stoken = str(token)
+        l = len(stoken)
+        if self.stickyspace is not None:
+            stickyspace_len = len(self.stickyspace)
+            if self.lastlen + stickyspace_len + l <= self.maxlen:
+                self.current.append(self.stickyspace)
+                self.lastlen += stickyspace_len
+                self.current.append(stoken)
+                self.lastlen += l
+                self.stickyspace = None
+                self.firstline = False
+                return True
+            if token.has_fws:
+                ws = token.pop_leading_fws()
+                if ws is not None:
+                    self.stickyspace += str(ws)
+                    stickyspace_len += len(ws)
+                token._fold(self)
+                return True
+            if stickyspace_len and l + 1 <= self.maxlen:
+                margin = self.maxlen - l
+                if 0 < margin < stickyspace_len:
+                    trim = stickyspace_len - margin
+                    self.current.append(self.stickyspace[:trim])
+                    self.stickyspace = self.stickyspace[trim:]
+                    stickyspace_len = trim
+                self.newline()
+                self.current.append(self.stickyspace)
+                self.current.append(stoken)
+                self.lastlen = l + stickyspace_len
+                self.stickyspace = None
+                self.firstline = False
+                return True
+            if not self.firstline:
+                self.newline()
+            self.current.append(self.stickyspace)
+            self.current.append(stoken)
+            self.stickyspace = None
+            self.firstline = False
+            return True
+        if self.lastlen + l <= self.maxlen:
+            self.current.append(stoken)
+            self.lastlen += l
+            return True
+        if l < self.maxlen:
+            self.newline()
+            self.current.append(stoken)
+            self.lastlen = l
+            return True
+        return False
+
+#
+# TokenList and its subclasses
+#
+
+class TokenList(list):
+
+    token_type = None
+
+    def __init__(self, *args, **kw):
+        super().__init__(*args, **kw)
+        self.defects = []
+
+    def __str__(self):
+        return ''.join(str(x) for x in self)
+
+    def __repr__(self):
+        return '{}({})'.format(self.__class__.__name__,
+                             super().__repr__())
+
+    @property
+    def value(self):
+        return ''.join(x.value for x in self if x.value)
+
+    @property
+    def all_defects(self):
+        return sum((x.all_defects for x in self), self.defects)
+
+    #
+    # Folding API
+    #
+    # parts():
+    #
+    # return a list of objects that constitute the "higher level syntactic
+    # objects" specified by the RFC as the best places to fold a header line.
+    # The returned objects must include leading folding white space, even if
+    # this means mutating the underlying parse tree of the object.  Each object
+    # is only responsible for returning *its* parts, and should not drill down
+    # to any lower level except as required to meet the leading folding white
+    # space constraint.
+    #
+    # _fold(folded):
+    #
+    #   folded: the result accumulator.  This is an instance of _Folded.
+    #       (XXX: I haven't finished factoring this out yet, the folding code
+    #       pretty much uses this as a state object.) When the folded.current
+    #       contains as much text as will fit, the _fold method should call
+    #       folded.newline.
+    #  folded.lastlen: the current length of the test stored in folded.current.
+    #  folded.maxlen: The maximum number of characters that may appear on a
+    #       folded line.  Differs from the policy setting in that "no limit" is
+    #       represented by +inf, which means it can be used in the trivially
+    #       logical fashion in comparisons.
+    #
+    # Currently no subclasses implement parts, and I think this will remain
+    # true.  A subclass only needs to implement _fold when the generic version
+    # isn't sufficient.  _fold will need to be implemented primarily when it is
+    # possible for encoded words to appear in the specialized token-list, since
+    # there is no generic algorithm that can know where exactly the encoded
+    # words are allowed.  A _fold implementation is responsible for filling
+    # lines in the same general way that the top level _fold does. It may, and
+    # should, call the _fold method of sub-objects in a similar fashion to that
+    # of the top level _fold.
+    #
+    # XXX: I'm hoping it will be possible to factor the existing code further
+    # to reduce redundancy and make the logic clearer.
+
+    @property
+    def parts(self):
+        klass = self.__class__
+        this = []
+        for token in self:
+            if token.startswith_fws():
+                if this:
+                    yield this[0] if len(this)==1 else klass(this)
+                    this.clear()
+            end_ws = token.pop_trailing_ws()
+            this.append(token)
+            if end_ws:
+                yield klass(this)
+                this = [end_ws]
+        if this:
+            yield this[0] if len(this)==1 else klass(this)
+
+    def startswith_fws(self):
+        return self[0].startswith_fws()
+
+    def pop_leading_fws(self):
+        if self[0].token_type == 'fws':
+            return self.pop(0)
+        return self[0].pop_leading_fws()
+
+    def pop_trailing_ws(self):
+        if self[-1].token_type == 'cfws':
+            return self.pop(-1)
+        return self[-1].pop_trailing_ws()
+
+    @property
+    def has_fws(self):
+        for part in self:
+            if part.has_fws:
+                return True
+        return False
+
+    def has_leading_comment(self):
+        return self[0].has_leading_comment()
+
+    @property
+    def comments(self):
+        comments = []
+        for token in self:
+            comments.extend(token.comments)
+        return comments
+
+    def fold(self, *, policy):
+        # max_line_length 0/None means no limit, ie: infinitely long.
+        maxlen = policy.max_line_length or float("+inf")
+        folded = _Folded(maxlen, policy)
+        self._fold(folded)
+        folded.finalize()
+        return str(folded)
+
+    def as_encoded_word(self, charset):
+        # This works only for things returned by 'parts', which include
+        # the leading fws, if any, that should be used.
+        res = []
+        ws = self.pop_leading_fws()
+        if ws:
+            res.append(ws)
+        trailer = self.pop(-1) if self[-1].token_type=='fws' else ''
+        res.append(_ew.encode(str(self), charset))
+        res.append(trailer)
+        return ''.join(res)
+
+    def cte_encode(self, charset, policy):
+        res = []
+        for part in self:
+            res.append(part.cte_encode(charset, policy))
+        return ''.join(res)
+
+    def _fold(self, folded):
+        encoding = 'utf-8' if folded.policy.utf8 else 'ascii'
+        for part in self.parts:
+            tstr = str(part)
+            tlen = len(tstr)
+            try:
+                str(part).encode(encoding)
+            except UnicodeEncodeError:
+                if any(isinstance(x, errors.UndecodableBytesDefect)
+                        for x in part.all_defects):
+                    charset = 'unknown-8bit'
+                else:
+                    # XXX: this should be a policy setting when utf8 is False.
+                    charset = 'utf-8'
+                tstr = part.cte_encode(charset, folded.policy)
+                tlen = len(tstr)
+            if folded.append_if_fits(part, tstr):
+                continue
+            # Peel off the leading whitespace if any and make it sticky, to
+            # avoid infinite recursion.
+            ws = part.pop_leading_fws()
+            if ws is not None:
+                # Peel off the leading whitespace and make it sticky, to
+                # avoid infinite recursion.
+                folded.stickyspace = str(part.pop(0))
+                if folded.append_if_fits(part):
+                    continue
+            if part.has_fws:
+                part._fold(folded)
+                continue
+            # There are no fold points in this one; it is too long for a single
+            # line and can't be split...we just have to put it on its own line.
+            folded.append(tstr)
+            folded.newline()
+
+    def pprint(self, indent=''):
+        print('\n'.join(self._pp(indent='')))
+
+    def ppstr(self, indent=''):
+        return '\n'.join(self._pp(indent=''))
+
+    def _pp(self, indent=''):
+        yield '{}{}/{}('.format(
+            indent,
+            self.__class__.__name__,
+            self.token_type)
+        for token in self:
+            if not hasattr(token, '_pp'):
+                yield (indent + '    !! invalid element in token '
+                                        'list: {!r}'.format(token))
+            else:
+                yield from token._pp(indent+'    ')
+        if self.defects:
+            extra = ' Defects: {}'.format(self.defects)
+        else:
+            extra = ''
+        yield '{}){}'.format(indent, extra)
+
+
+class WhiteSpaceTokenList(TokenList):
+
+    @property
+    def value(self):
+        return ' '
+
+    @property
+    def comments(self):
+        return [x.content for x in self if x.token_type=='comment']
+
+
+class UnstructuredTokenList(TokenList):
+
+    token_type = 'unstructured'
+
+    def _fold(self, folded):
+        last_ew = None
+        encoding = 'utf-8' if folded.policy.utf8 else 'ascii'
+        for part in self.parts:
+            tstr = str(part)
+            is_ew = False
+            try:
+                str(part).encode(encoding)
+            except UnicodeEncodeError:
+                if any(isinstance(x, errors.UndecodableBytesDefect)
+                       for x in part.all_defects):
+                    charset = 'unknown-8bit'
+                else:
+                    charset = 'utf-8'
+                if last_ew is not None:
+                    # We've already done an EW, combine this one with it
+                    # if there's room.
+                    chunk = get_unstructured(
+                        ''.join(folded.current[last_ew:]+[tstr])).as_encoded_word(charset)
+                    oldlastlen = sum(len(x) for x in folded.current[:last_ew])
+                    schunk = str(chunk)
+                    lchunk = len(schunk)
+                    if oldlastlen + lchunk <= folded.maxlen:
+                        del folded.current[last_ew:]
+                        folded.append(schunk)
+                        folded.lastlen = oldlastlen + lchunk
+                        continue
+                tstr = part.as_encoded_word(charset)
+                is_ew = True
+            if folded.append_if_fits(part, tstr):
+                if is_ew:
+                    last_ew = len(folded.current) - 1
+                continue
+            if is_ew or last_ew:
+                # It's too big to fit on the line, but since we've
+                # got encoded words we can use encoded word folding.
+                part._fold_as_ew(folded)
+                continue
+            # Peel off the leading whitespace if any and make it sticky, to
+            # avoid infinite recursion.
+            ws = part.pop_leading_fws()
+            if ws is not None:
+                folded.stickyspace = str(ws)
+                if folded.append_if_fits(part):
+                    continue
+            if part.has_fws:
+                part._fold(folded)
+                continue
+            # It can't be split...we just have to put it on its own line.
+            folded.append(tstr)
+            folded.newline()
+            last_ew = None
+
+    def cte_encode(self, charset, policy):
+        res = []
+        last_ew = None
+        for part in self:
+            spart = str(part)
+            try:
+                spart.encode('us-ascii')
+                res.append(spart)
+            except UnicodeEncodeError:
+                if last_ew is None:
+                    res.append(part.cte_encode(charset, policy))
+                    last_ew = len(res)
+                else:
+                    tl = get_unstructured(''.join(res[last_ew:] + [spart]))
+                    res.append(tl.as_encoded_word(charset))
+        return ''.join(res)
+
+
+class Phrase(TokenList):
+
+    token_type = 'phrase'
+
+    def _fold(self, folded):
+        # As with Unstructured, we can have pure ASCII with or without
+        # surrogateescape encoded bytes, or we could have unicode.  But this
+        # case is more complicated, since we have to deal with the various
+        # sub-token types and how they can be composed in the face of
+        # unicode-that-needs-CTE-encoding, and the fact that if a token a
+        # comment that becomes a barrier across which we can't compose encoded
+        # words.
+        last_ew = None
+        encoding = 'utf-8' if folded.policy.utf8 else 'ascii'
+        for part in self.parts:
+            tstr = str(part)
+            tlen = len(tstr)
+            has_ew = False
+            try:
+                str(part).encode(encoding)
+            except UnicodeEncodeError:
+                if any(isinstance(x, errors.UndecodableBytesDefect)
+                        for x in part.all_defects):
+                    charset = 'unknown-8bit'
+                else:
+                    charset = 'utf-8'
+                if last_ew is not None and not part.has_leading_comment():
+                    # We've already done an EW, let's see if we can combine
+                    # this one with it.  The last_ew logic ensures that all we
+                    # have at this point is atoms, no comments or quoted
+                    # strings.  So we can treat the text between the last
+                    # encoded word and the content of this token as
+                    # unstructured text, and things will work correctly.  But
+                    # we have to strip off any trailing comment on this token
+                    # first, and if it is a quoted string we have to pull out
+                    # the content (we're encoding it, so it no longer needs to
+                    # be quoted).
+                    if part[-1].token_type == 'cfws' and part.comments:
+                        remainder = part.pop(-1)
+                    else:
+                        remainder = ''
+                    for i, token in enumerate(part):
+                        if token.token_type == 'bare-quoted-string':
+                            part[i] = UnstructuredTokenList(token[:])
+                    chunk = get_unstructured(
+                        ''.join(folded.current[last_ew:]+[tstr])).as_encoded_word(charset)
+                    schunk = str(chunk)
+                    lchunk = len(schunk)
+                    if last_ew + lchunk <= folded.maxlen:
+                        del folded.current[last_ew:]
+                        folded.append(schunk)
+                        folded.lastlen = sum(len(x) for x in folded.current)
+                        continue
+                tstr = part.as_encoded_word(charset)
+                tlen = len(tstr)
+                has_ew = True
+            if folded.append_if_fits(part, tstr):
+                if has_ew and not part.comments:
+                    last_ew = len(folded.current) - 1
+                elif part.comments or part.token_type == 'quoted-string':
+                    # If a comment is involved we can't combine EWs.  And if a
+                    # quoted string is involved, it's not worth the effort to
+                    # try to combine them.
+                    last_ew = None
+                continue
+            part._fold(folded)
+
+    def cte_encode(self, charset, policy):
+        res = []
+        last_ew = None
+        is_ew = False
+        for part in self:
+            spart = str(part)
+            try:
+                spart.encode('us-ascii')
+                res.append(spart)
+            except UnicodeEncodeError:
+                is_ew = True
+                if last_ew is None:
+                    if not part.comments:
+                        last_ew = len(res)
+                    res.append(part.cte_encode(charset, policy))
+                elif not part.has_leading_comment():
+                    if part[-1].token_type == 'cfws' and part.comments:
+                        remainder = part.pop(-1)
+                    else:
+                        remainder = ''
+                    for i, token in enumerate(part):
+                        if token.token_type == 'bare-quoted-string':
+                            part[i] = UnstructuredTokenList(token[:])
+                    tl = get_unstructured(''.join(res[last_ew:] + [spart]))
+                    res[last_ew:] = [tl.as_encoded_word(charset)]
+            if part.comments or (not is_ew and part.token_type == 'quoted-string'):
+                last_ew = None
+        return ''.join(res)
+
+class Word(TokenList):
+
+    token_type = 'word'
+
+
+class CFWSList(WhiteSpaceTokenList):
+
+    token_type = 'cfws'
+
+    def has_leading_comment(self):
+        return bool(self.comments)
+
+
+class Atom(TokenList):
+
+    token_type = 'atom'
+
+
+class Token(TokenList):
+
+    token_type = 'token'
+
+
+class EncodedWord(TokenList):
+
+    token_type = 'encoded-word'
+    cte = None
+    charset = None
+    lang = None
+
+    @property
+    def encoded(self):
+        if self.cte is not None:
+            return self.cte
+        _ew.encode(str(self), self.charset)
+
+
+
+class QuotedString(TokenList):
+
+    token_type = 'quoted-string'
+
+    @property
+    def content(self):
+        for x in self:
+            if x.token_type == 'bare-quoted-string':
+                return x.value
+
+    @property
+    def quoted_value(self):
+        res = []
+        for x in self:
+            if x.token_type == 'bare-quoted-string':
+                res.append(str(x))
+            else:
+                res.append(x.value)
+        return ''.join(res)
+
+    @property
+    def stripped_value(self):
+        for token in self:
+            if token.token_type == 'bare-quoted-string':
+                return token.value
+
+
+class BareQuotedString(QuotedString):
+
+    token_type = 'bare-quoted-string'
+
+    def __str__(self):
+        return quote_string(''.join(str(x) for x in self))
+
+    @property
+    def value(self):
+        return ''.join(str(x) for x in self)
+
+
+class Comment(WhiteSpaceTokenList):
+
+    token_type = 'comment'
+
+    def __str__(self):
+        return ''.join(sum([
+                            ["("],
+                            [self.quote(x) for x in self],
+                            [")"],
+                            ], []))
+
+    def quote(self, value):
+        if value.token_type == 'comment':
+            return str(value)
+        return str(value).replace('\\', '\\\\').replace(
+                                  '(', r'\(').replace(
+                                  ')', r'\)')
+
+    @property
+    def content(self):
+        return ''.join(str(x) for x in self)
+
+    @property
+    def comments(self):
+        return [self.content]
+
+class AddressList(TokenList):
+
+    token_type = 'address-list'
+
+    @property
+    def addresses(self):
+        return [x for x in self if x.token_type=='address']
+
+    @property
+    def mailboxes(self):
+        return sum((x.mailboxes
+                    for x in self if x.token_type=='address'), [])
+
+    @property
+    def all_mailboxes(self):
+        return sum((x.all_mailboxes
+                    for x in self if x.token_type=='address'), [])
+
+
+class Address(TokenList):
+
+    token_type = 'address'
+
+    @property
+    def display_name(self):
+        if self[0].token_type == 'group':
+            return self[0].display_name
+
+    @property
+    def mailboxes(self):
+        if self[0].token_type == 'mailbox':
+            return [self[0]]
+        elif self[0].token_type == 'invalid-mailbox':
+            return []
+        return self[0].mailboxes
+
+    @property
+    def all_mailboxes(self):
+        if self[0].token_type == 'mailbox':
+            return [self[0]]
+        elif self[0].token_type == 'invalid-mailbox':
+            return [self[0]]
+        return self[0].all_mailboxes
+
+class MailboxList(TokenList):
+
+    token_type = 'mailbox-list'
+
+    @property
+    def mailboxes(self):
+        return [x for x in self if x.token_type=='mailbox']
+
+    @property
+    def all_mailboxes(self):
+        return [x for x in self
+            if x.token_type in ('mailbox', 'invalid-mailbox')]
+
+
+class GroupList(TokenList):
+
+    token_type = 'group-list'
+
+    @property
+    def mailboxes(self):
+        if not self or self[0].token_type != 'mailbox-list':
+            return []
+        return self[0].mailboxes
+
+    @property
+    def all_mailboxes(self):
+        if not self or self[0].token_type != 'mailbox-list':
+            return []
+        return self[0].all_mailboxes
+
+
+class Group(TokenList):
+
+    token_type = "group"
+
+    @property
+    def mailboxes(self):
+        if self[2].token_type != 'group-list':
+            return []
+        return self[2].mailboxes
+
+    @property
+    def all_mailboxes(self):
+        if self[2].token_type != 'group-list':
+            return []
+        return self[2].all_mailboxes
+
+    @property
+    def display_name(self):
+        return self[0].display_name
+
+
+class NameAddr(TokenList):
+
+    token_type = 'name-addr'
+
+    @property
+    def display_name(self):
+        if len(self) == 1:
+            return None
+        return self[0].display_name
+
+    @property
+    def local_part(self):
+        return self[-1].local_part
+
+    @property
+    def domain(self):
+        return self[-1].domain
+
+    @property
+    def route(self):
+        return self[-1].route
+
+    @property
+    def addr_spec(self):
+        return self[-1].addr_spec
+
+
+class AngleAddr(TokenList):
+
+    token_type = 'angle-addr'
+
+    @property
+    def local_part(self):
+        for x in self:
+            if x.token_type == 'addr-spec':
+                return x.local_part
+
+    @property
+    def domain(self):
+        for x in self:
+            if x.token_type == 'addr-spec':
+                return x.domain
+
+    @property
+    def route(self):
+        for x in self:
+            if x.token_type == 'obs-route':
+                return x.domains
+
+    @property
+    def addr_spec(self):
+        for x in self:
+            if x.token_type == 'addr-spec':
+                return x.addr_spec
+        else:
+            return '<>'
+
+
+class ObsRoute(TokenList):
+
+    token_type = 'obs-route'
+
+    @property
+    def domains(self):
+        return [x.domain for x in self if x.token_type == 'domain']
+
+
+class Mailbox(TokenList):
+
+    token_type = 'mailbox'
+
+    @property
+    def display_name(self):
+        if self[0].token_type == 'name-addr':
+            return self[0].display_name
+
+    @property
+    def local_part(self):
+        return self[0].local_part
+
+    @property
+    def domain(self):
+        return self[0].domain
+
+    @property
+    def route(self):
+        if self[0].token_type == 'name-addr':
+            return self[0].route
+
+    @property
+    def addr_spec(self):
+        return self[0].addr_spec
+
+
+class InvalidMailbox(TokenList):
+
+    token_type = 'invalid-mailbox'
+
+    @property
+    def display_name(self):
+        return None
+
+    local_part = domain = route = addr_spec = display_name
+
+
+class Domain(TokenList):
+
+    token_type = 'domain'
+
+    @property
+    def domain(self):
+        return ''.join(super().value.split())
+
+
+class DotAtom(TokenList):
+
+    token_type = 'dot-atom'
+
+
+class DotAtomText(TokenList):
+
+    token_type = 'dot-atom-text'
+
+
+class AddrSpec(TokenList):
+
+    token_type = 'addr-spec'
+
+    @property
+    def local_part(self):
+        return self[0].local_part
+
+    @property
+    def domain(self):
+        if len(self) < 3:
+            return None
+        return self[-1].domain
+
+    @property
+    def value(self):
+        if len(self) < 3:
+            return self[0].value
+        return self[0].value.rstrip()+self[1].value+self[2].value.lstrip()
+
+    @property
+    def addr_spec(self):
+        nameset = set(self.local_part)
+        if len(nameset) > len(nameset-DOT_ATOM_ENDS):
+            lp = quote_string(self.local_part)
+        else:
+            lp = self.local_part
+        if self.domain is not None:
+            return lp + '@' + self.domain
+        return lp
+
+
+class ObsLocalPart(TokenList):
+
+    token_type = 'obs-local-part'
+
+
+class DisplayName(Phrase):
+
+    token_type = 'display-name'
+
+    @property
+    def display_name(self):
+        res = TokenList(self)
+        if res[0].token_type == 'cfws':
+            res.pop(0)
+        else:
+            if res[0][0].token_type == 'cfws':
+                res[0] = TokenList(res[0][1:])
+        if res[-1].token_type == 'cfws':
+            res.pop()
+        else:
+            if res[-1][-1].token_type == 'cfws':
+                res[-1] = TokenList(res[-1][:-1])
+        return res.value
+
+    @property
+    def value(self):
+        quote = False
+        if self.defects:
+            quote = True
+        else:
+            for x in self:
+                if x.token_type == 'quoted-string':
+                    quote = True
+        if quote:
+            pre = post = ''
+            if self[0].token_type=='cfws' or self[0][0].token_type=='cfws':
+                pre = ' '
+            if self[-1].token_type=='cfws' or self[-1][-1].token_type=='cfws':
+                post = ' '
+            return pre+quote_string(self.display_name)+post
+        else:
+            return super().value
+
+
+class LocalPart(TokenList):
+
+    token_type = 'local-part'
+
+    @property
+    def value(self):
+        if self[0].token_type == "quoted-string":
+            return self[0].quoted_value
+        else:
+            return self[0].value
+
+    @property
+    def local_part(self):
+        # Strip whitespace from front, back, and around dots.
+        res = [DOT]
+        last = DOT
+        last_is_tl = False
+        for tok in self[0] + [DOT]:
+            if tok.token_type == 'cfws':
+                continue
+            if (last_is_tl and tok.token_type == 'dot' and
+                    last[-1].token_type == 'cfws'):
+                res[-1] = TokenList(last[:-1])
+            is_tl = isinstance(tok, TokenList)
+            if (is_tl and last.token_type == 'dot' and
+                    tok[0].token_type == 'cfws'):
+                res.append(TokenList(tok[1:]))
+            else:
+                res.append(tok)
+            last = res[-1]
+            last_is_tl = is_tl
+        res = TokenList(res[1:-1])
+        return res.value
+
+
+class DomainLiteral(TokenList):
+
+    token_type = 'domain-literal'
+
+    @property
+    def domain(self):
+        return ''.join(super().value.split())
+
+    @property
+    def ip(self):
+        for x in self:
+            if x.token_type == 'ptext':
+                return x.value
+
+
+class MIMEVersion(TokenList):
+
+    token_type = 'mime-version'
+    major = None
+    minor = None
+
+
+class Parameter(TokenList):
+
+    token_type = 'parameter'
+    sectioned = False
+    extended = False
+    charset = 'us-ascii'
+
+    @property
+    def section_number(self):
+        # Because the first token, the attribute (name) eats CFWS, the second
+        # token is always the section if there is one.
+        return self[1].number if self.sectioned else 0
+
+    @property
+    def param_value(self):
+        # This is part of the "handle quoted extended parameters" hack.
+        for token in self:
+            if token.token_type == 'value':
+                return token.stripped_value
+            if token.token_type == 'quoted-string':
+                for token in token:
+                    if token.token_type == 'bare-quoted-string':
+                        for token in token:
+                            if token.token_type == 'value':
+                                return token.stripped_value
+        return ''
+
+
+class InvalidParameter(Parameter):
+
+    token_type = 'invalid-parameter'
+
+
+class Attribute(TokenList):
+
+    token_type = 'attribute'
+
+    @property
+    def stripped_value(self):
+        for token in self:
+            if token.token_type.endswith('attrtext'):
+                return token.value
+
+class Section(TokenList):
+
+    token_type = 'section'
+    number = None
+
+
+class Value(TokenList):
+
+    token_type = 'value'
+
+    @property
+    def stripped_value(self):
+        token = self[0]
+        if token.token_type == 'cfws':
+            token = self[1]
+        if token.token_type.endswith(
+                ('quoted-string', 'attribute', 'extended-attribute')):
+            return token.stripped_value
+        return self.value
+
+
+class MimeParameters(TokenList):
+
+    token_type = 'mime-parameters'
+
+    @property
+    def params(self):
+        # The RFC specifically states that the ordering of parameters is not
+        # guaranteed and may be reordered by the transport layer.  So we have
+        # to assume the RFC 2231 pieces can come in any order.  However, we
+        # output them in the order that we first see a given name, which gives
+        # us a stable __str__.
+        params = OrderedDict()
+        for token in self:
+            if not token.token_type.endswith('parameter'):
+                continue
+            if token[0].token_type != 'attribute':
+                continue
+            name = token[0].value.strip()
+            if name not in params:
+                params[name] = []
+            params[name].append((token.section_number, token))
+        for name, parts in params.items():
+            parts = sorted(parts, key=itemgetter(0))
+            first_param = parts[0][1]
+            charset = first_param.charset
+            # Our arbitrary error recovery is to ignore duplicate parameters,
+            # to use appearance order if there are duplicate rfc 2231 parts,
+            # and to ignore gaps.  This mimics the error recovery of get_param.
+            if not first_param.extended and len(parts) > 1:
+                if parts[1][0] == 0:
+                    parts[1][1].defects.append(errors.InvalidHeaderDefect(
+                        'duplicate parameter name; duplicate(s) ignored'))
+                    parts = parts[:1]
+                # Else assume the *0* was missing...note that this is different
+                # from get_param, but we registered a defect for this earlier.
+            value_parts = []
+            i = 0
+            for section_number, param in parts:
+                if section_number != i:
+                    # We could get fancier here and look for a complete
+                    # duplicate extended parameter and ignore the second one
+                    # seen.  But we're not doing that.  The old code didn't.
+                    if not param.extended:
+                        param.defects.append(errors.InvalidHeaderDefect(
+                            'duplicate parameter name; duplicate ignored'))
+                        continue
+                    else:
+                        param.defects.append(errors.InvalidHeaderDefect(
+                            "inconsistent RFC2231 parameter numbering"))
+                i += 1
+                value = param.param_value
+                if param.extended:
+                    try:
+                        value = urllib.parse.unquote_to_bytes(value)
+                    except UnicodeEncodeError:
+                        # source had surrogate escaped bytes.  What we do now
+                        # is a bit of an open question.  I'm not sure this is
+                        # the best choice, but it is what the old algorithm did
+                        value = urllib.parse.unquote(value, encoding='latin-1')
+                    else:
+                        try:
+                            value = value.decode(charset, 'surrogateescape')
+                        except LookupError:
+                            # XXX: there should really be a custom defect for
+                            # unknown character set to make it easy to find,
+                            # because otherwise unknown charset is a silent
+                            # failure.
+                            value = value.decode('us-ascii', 'surrogateescape')
+                        if utils._has_surrogates(value):
+                            param.defects.append(errors.UndecodableBytesDefect())
+                value_parts.append(value)
+            value = ''.join(value_parts)
+            yield name, value
+
+    def __str__(self):
+        params = []
+        for name, value in self.params:
+            if value:
+                params.append('{}={}'.format(name, quote_string(value)))
+            else:
+                params.append(name)
+        params = '; '.join(params)
+        return ' ' + params if params else ''
+
+
+class ParameterizedHeaderValue(TokenList):
+
+    @property
+    def params(self):
+        for token in reversed(self):
+            if token.token_type == 'mime-parameters':
+                return token.params
+        return {}
+
+    @property
+    def parts(self):
+        if self and self[-1].token_type == 'mime-parameters':
+            # We don't want to start a new line if all of the params don't fit
+            # after the value, so unwrap the parameter list.
+            return TokenList(self[:-1] + self[-1])
+        return TokenList(self).parts
+
+
+class ContentType(ParameterizedHeaderValue):
+
+    token_type = 'content-type'
+    maintype = 'text'
+    subtype = 'plain'
+
+
+class ContentDisposition(ParameterizedHeaderValue):
+
+    token_type = 'content-disposition'
+    content_disposition = None
+
+
+class ContentTransferEncoding(TokenList):
+
+    token_type = 'content-transfer-encoding'
+    cte = '7bit'
+
+
+class HeaderLabel(TokenList):
+
+    token_type = 'header-label'
+
+
+class Header(TokenList):
+
+    token_type = 'header'
+
+    def _fold(self, folded):
+        folded.append(str(self.pop(0)))
+        folded.lastlen = len(folded.current[0])
+        # The first line of the header is different from all others: we don't
+        # want to start a new object on a new line if it has any fold points in
+        # it that would allow part of it to be on the first header line.
+        # Further, if the first fold point would fit on the new line, we want
+        # to do that, but if it doesn't we want to put it on the first line.
+        # Folded supports this via the stickyspace attribute.  If this
+        # attribute is not None, it does the special handling.
+        folded.stickyspace = str(self.pop(0)) if self[0].token_type == 'cfws' else ''
+        rest = self.pop(0)
+        if self:
+            raise ValueError("Malformed Header token list")
+        rest._fold(folded)
+
+
+#
+# Terminal classes and instances
+#
+
+class Terminal(str):
+
+    def __new__(cls, value, token_type):
+        self = super().__new__(cls, value)
+        self.token_type = token_type
+        self.defects = []
+        return self
+
+    def __repr__(self):
+        return "{}({})".format(self.__class__.__name__, super().__repr__())
+
+    @property
+    def all_defects(self):
+        return list(self.defects)
+
+    def _pp(self, indent=''):
+        return ["{}{}/{}({}){}".format(
+            indent,
+            self.__class__.__name__,
+            self.token_type,
+            super().__repr__(),
+            '' if not self.defects else ' {}'.format(self.defects),
+            )]
+
+    def cte_encode(self, charset, policy):
+        value = str(self)
+        try:
+            value.encode('us-ascii')
+            return value
+        except UnicodeEncodeError:
+            return _ew.encode(value, charset)
+
+    def pop_trailing_ws(self):
+        # This terminates the recursion.
+        return None
+
+    def pop_leading_fws(self):
+        # This terminates the recursion.
+        return None
+
+    @property
+    def comments(self):
+        return []
+
+    def has_leading_comment(self):
+        return False
+
+    def __getnewargs__(self):
+        return(str(self), self.token_type)
+
+
+class WhiteSpaceTerminal(Terminal):
+
+    @property
+    def value(self):
+        return ' '
+
+    def startswith_fws(self):
+        return True
+
+    has_fws = True
+
+
+class ValueTerminal(Terminal):
+
+    @property
+    def value(self):
+        return self
+
+    def startswith_fws(self):
+        return False
+
+    has_fws = False
+
+    def as_encoded_word(self, charset):
+        return _ew.encode(str(self), charset)
+
+
+class EWWhiteSpaceTerminal(WhiteSpaceTerminal):
+
+    @property
+    def value(self):
+        return ''
+
+    @property
+    def encoded(self):
+        return self[:]
+
+    def __str__(self):
+        return ''
+
+    has_fws = True
+
+
+# XXX these need to become classes and used as instances so
+# that a program can't change them in a parse tree and screw
+# up other parse trees.  Maybe should have  tests for that, too.
+DOT = ValueTerminal('.', 'dot')
+ListSeparator = ValueTerminal(',', 'list-separator')
+RouteComponentMarker = ValueTerminal('@', 'route-component-marker')
+
+#
+# Parser
+#
+
+# Parse strings according to RFC822/2047/2822/5322 rules.
+#
+# This is a stateless parser.  Each get_XXX function accepts a string and
+# returns either a Terminal or a TokenList representing the RFC object named
+# by the method and a string containing the remaining unparsed characters
+# from the input.  Thus a parser method consumes the next syntactic construct
+# of a given type and returns a token representing the construct plus the
+# unparsed remainder of the input string.
+#
+# For example, if the first element of a structured header is a 'phrase',
+# then:
+#
+#     phrase, value = get_phrase(value)
+#
+# returns the complete phrase from the start of the string value, plus any
+# characters left in the string after the phrase is removed.
+
+_wsp_splitter = re.compile(r'([{}]+)'.format(''.join(WSP))).split
+_non_atom_end_matcher = re.compile(r"[^{}]+".format(
+    ''.join(ATOM_ENDS).replace('\\','\\\\').replace(']',r'\]'))).match
+_non_printable_finder = re.compile(r"[\x00-\x20\x7F]").findall
+_non_token_end_matcher = re.compile(r"[^{}]+".format(
+    ''.join(TOKEN_ENDS).replace('\\','\\\\').replace(']',r'\]'))).match
+_non_attribute_end_matcher = re.compile(r"[^{}]+".format(
+    ''.join(ATTRIBUTE_ENDS).replace('\\','\\\\').replace(']',r'\]'))).match
+_non_extended_attribute_end_matcher = re.compile(r"[^{}]+".format(
+    ''.join(EXTENDED_ATTRIBUTE_ENDS).replace(
+                                    '\\','\\\\').replace(']',r'\]'))).match
+
+def _validate_xtext(xtext):
+    """If input token contains ASCII non-printables, register a defect."""
+
+    non_printables = _non_printable_finder(xtext)
+    if non_printables:
+        xtext.defects.append(errors.NonPrintableDefect(non_printables))
+    if utils._has_surrogates(xtext):
+        xtext.defects.append(errors.UndecodableBytesDefect(
+            "Non-ASCII characters found in header token"))
+
+def _get_ptext_to_endchars(value, endchars):
+    """Scan printables/quoted-pairs until endchars and return unquoted ptext.
+
+    This function turns a run of qcontent, ccontent-without-comments, or
+    dtext-with-quoted-printables into a single string by unquoting any
+    quoted printables.  It returns the string, the remaining value, and
+    a flag that is True iff there were any quoted printables decoded.
+
+    """
+    fragment, *remainder = _wsp_splitter(value, 1)
+    vchars = []
+    escape = False
+    had_qp = False
+    for pos in range(len(fragment)):
+        if fragment[pos] == '\\':
+            if escape:
+                escape = False
+                had_qp = True
+            else:
+                escape = True
+                continue
+        if escape:
+            escape = False
+        elif fragment[pos] in endchars:
+            break
+        vchars.append(fragment[pos])
+    else:
+        pos = pos + 1
+    return ''.join(vchars), ''.join([fragment[pos:]] + remainder), had_qp
+
+def get_fws(value):
+    """FWS = 1*WSP
+
+    This isn't the RFC definition.  We're using fws to represent tokens where
+    folding can be done, but when we are parsing the *un*folding has already
+    been done so we don't need to watch out for CRLF.
+
+    """
+    newvalue = value.lstrip()
+    fws = WhiteSpaceTerminal(value[:len(value)-len(newvalue)], 'fws')
+    return fws, newvalue
+
+def get_encoded_word(value):
+    """ encoded-word = "=?" charset "?" encoding "?" encoded-text "?="
+
+    """
+    ew = EncodedWord()
+    if not value.startswith('=?'):
+        raise errors.HeaderParseError(
+            "expected encoded word but found {}".format(value))
+    tok, *remainder = value[2:].split('?=', 1)
+    if tok == value[2:]:
+        raise errors.HeaderParseError(
+            "expected encoded word but found {}".format(value))
+    remstr = ''.join(remainder)
+    if len(remstr) > 1 and remstr[0] in hexdigits and remstr[1] in hexdigits:
+        # The ? after the CTE was followed by an encoded word escape (=XX).
+        rest, *remainder = remstr.split('?=', 1)
+        tok = tok + '?=' + rest
+    if len(tok.split()) > 1:
+        ew.defects.append(errors.InvalidHeaderDefect(
+            "whitespace inside encoded word"))
+    ew.cte = value
+    value = ''.join(remainder)
+    try:
+        text, charset, lang, defects = _ew.decode('=?' + tok + '?=')
+    except ValueError:
+        raise errors.HeaderParseError(
+            "encoded word format invalid: '{}'".format(ew.cte))
+    ew.charset = charset
+    ew.lang = lang
+    ew.defects.extend(defects)
+    while text:
+        if text[0] in WSP:
+            token, text = get_fws(text)
+            ew.append(token)
+            continue
+        chars, *remainder = _wsp_splitter(text, 1)
+        vtext = ValueTerminal(chars, 'vtext')
+        _validate_xtext(vtext)
+        ew.append(vtext)
+        text = ''.join(remainder)
+    return ew, value
+
+def get_unstructured(value):
+    """unstructured = (*([FWS] vchar) *WSP) / obs-unstruct
+       obs-unstruct = *((*LF *CR *(obs-utext) *LF *CR)) / FWS)
+       obs-utext = %d0 / obs-NO-WS-CTL / LF / CR
+
+       obs-NO-WS-CTL is control characters except WSP/CR/LF.
+
+    So, basically, we have printable runs, plus control characters or nulls in
+    the obsolete syntax, separated by whitespace.  Since RFC 2047 uses the
+    obsolete syntax in its specification, but requires whitespace on either
+    side of the encoded words, I can see no reason to need to separate the
+    non-printable-non-whitespace from the printable runs if they occur, so we
+    parse this into xtext tokens separated by WSP tokens.
+
+    Because an 'unstructured' value must by definition constitute the entire
+    value, this 'get' routine does not return a remaining value, only the
+    parsed TokenList.
+
+    """
+    # XXX: but what about bare CR and LF?  They might signal the start or
+    # end of an encoded word.  YAGNI for now, since our current parsers
+    # will never send us strings with bare CR or LF.
+
+    unstructured = UnstructuredTokenList()
+    while value:
+        if value[0] in WSP:
+            token, value = get_fws(value)
+            unstructured.append(token)
+            continue
+        if value.startswith('=?'):
+            try:
+                token, value = get_encoded_word(value)
+            except errors.HeaderParseError:
+                # XXX: Need to figure out how to register defects when
+                # appropriate here.
+                pass
+            else:
+                have_ws = True
+                if len(unstructured) > 0:
+                    if unstructured[-1].token_type != 'fws':
+                        unstructured.defects.append(errors.InvalidHeaderDefect(
+                            "missing whitespace before encoded word"))
+                        have_ws = False
+                if have_ws and len(unstructured) > 1:
+                    if unstructured[-2].token_type == 'encoded-word':
+                        unstructured[-1] = EWWhiteSpaceTerminal(
+                            unstructured[-1], 'fws')
+                unstructured.append(token)
+                continue
+        tok, *remainder = _wsp_splitter(value, 1)
+        vtext = ValueTerminal(tok, 'vtext')
+        _validate_xtext(vtext)
+        unstructured.append(vtext)
+        value = ''.join(remainder)
+    return unstructured
+
+def get_qp_ctext(value):
+    r"""ctext = <printable ascii except \ ( )>
+
+    This is not the RFC ctext, since we are handling nested comments in comment
+    and unquoting quoted-pairs here.  We allow anything except the '()'
+    characters, but if we find any ASCII other than the RFC defined printable
+    ASCII, a NonPrintableDefect is added to the token's defects list.  Since
+    quoted pairs are converted to their unquoted values, what is returned is
+    a 'ptext' token.  In this case it is a WhiteSpaceTerminal, so it's value
+    is ' '.
+
+    """
+    ptext, value, _ = _get_ptext_to_endchars(value, '()')
+    ptext = WhiteSpaceTerminal(ptext, 'ptext')
+    _validate_xtext(ptext)
+    return ptext, value
+
+def get_qcontent(value):
+    """qcontent = qtext / quoted-pair
+
+    We allow anything except the DQUOTE character, but if we find any ASCII
+    other than the RFC defined printable ASCII, a NonPrintableDefect is
+    added to the token's defects list.  Any quoted pairs are converted to their
+    unquoted values, so what is returned is a 'ptext' token.  In this case it
+    is a ValueTerminal.
+
+    """
+    ptext, value, _ = _get_ptext_to_endchars(value, '"')
+    ptext = ValueTerminal(ptext, 'ptext')
+    _validate_xtext(ptext)
+    return ptext, value
+
+def get_atext(value):
+    """atext = <matches _atext_matcher>
+
+    We allow any non-ATOM_ENDS in atext, but add an InvalidATextDefect to
+    the token's defects list if we find non-atext characters.
+    """
+    m = _non_atom_end_matcher(value)
+    if not m:
+        raise errors.HeaderParseError(
+            "expected atext but found '{}'".format(value))
+    atext = m.group()
+    value = value[len(atext):]
+    atext = ValueTerminal(atext, 'atext')
+    _validate_xtext(atext)
+    return atext, value
+
+def get_bare_quoted_string(value):
+    """bare-quoted-string = DQUOTE *([FWS] qcontent) [FWS] DQUOTE
+
+    A quoted-string without the leading or trailing white space.  Its
+    value is the text between the quote marks, with whitespace
+    preserved and quoted pairs decoded.
+    """
+    if value[0] != '"':
+        raise errors.HeaderParseError(
+            "expected '\"' but found '{}'".format(value))
+    bare_quoted_string = BareQuotedString()
+    value = value[1:]
+    while value and value[0] != '"':
+        if value[0] in WSP:
+            token, value = get_fws(value)
+        elif value[:2] == '=?':
+            try:
+                token, value = get_encoded_word(value)
+                bare_quoted_string.defects.append(errors.InvalidHeaderDefect(
+                    "encoded word inside quoted string"))
+            except errors.HeaderParseError:
+                token, value = get_qcontent(value)
+        else:
+            token, value = get_qcontent(value)
+        bare_quoted_string.append(token)
+    if not value:
+        bare_quoted_string.defects.append(errors.InvalidHeaderDefect(
+            "end of header inside quoted string"))
+        return bare_quoted_string, value
+    return bare_quoted_string, value[1:]
+
+def get_comment(value):
+    """comment = "(" *([FWS] ccontent) [FWS] ")"
+       ccontent = ctext / quoted-pair / comment
+
+    We handle nested comments here, and quoted-pair in our qp-ctext routine.
+    """
+    if value and value[0] != '(':
+        raise errors.HeaderParseError(
+            "expected '(' but found '{}'".format(value))
+    comment = Comment()
+    value = value[1:]
+    while value and value[0] != ")":
+        if value[0] in WSP:
+            token, value = get_fws(value)
+        elif value[0] == '(':
+            token, value = get_comment(value)
+        else:
+            token, value = get_qp_ctext(value)
+        comment.append(token)
+    if not value:
+        comment.defects.append(errors.InvalidHeaderDefect(
+            "end of header inside comment"))
+        return comment, value
+    return comment, value[1:]
+
+def get_cfws(value):
+    """CFWS = (1*([FWS] comment) [FWS]) / FWS
+
+    """
+    cfws = CFWSList()
+    while value and value[0] in CFWS_LEADER:
+        if value[0] in WSP:
+            token, value = get_fws(value)
+        else:
+            token, value = get_comment(value)
+        cfws.append(token)
+    return cfws, value
+
+def get_quoted_string(value):
+    """quoted-string = [CFWS] <bare-quoted-string> [CFWS]
+
+    'bare-quoted-string' is an intermediate class defined by this
+    parser and not by the RFC grammar.  It is the quoted string
+    without any attached CFWS.
+    """
+    quoted_string = QuotedString()
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        quoted_string.append(token)
+    token, value = get_bare_quoted_string(value)
+    quoted_string.append(token)
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        quoted_string.append(token)
+    return quoted_string, value
+
+def get_atom(value):
+    """atom = [CFWS] 1*atext [CFWS]
+
+    An atom could be an rfc2047 encoded word.
+    """
+    atom = Atom()
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        atom.append(token)
+    if value and value[0] in ATOM_ENDS:
+        raise errors.HeaderParseError(
+            "expected atom but found '{}'".format(value))
+    if value.startswith('=?'):
+        try:
+            token, value = get_encoded_word(value)
+        except errors.HeaderParseError:
+            # XXX: need to figure out how to register defects when
+            # appropriate here.
+            token, value = get_atext(value)
+    else:
+        token, value = get_atext(value)
+    atom.append(token)
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        atom.append(token)
+    return atom, value
+
+def get_dot_atom_text(value):
+    """ dot-text = 1*atext *("." 1*atext)
+
+    """
+    dot_atom_text = DotAtomText()
+    if not value or value[0] in ATOM_ENDS:
+        raise errors.HeaderParseError("expected atom at a start of "
+            "dot-atom-text but found '{}'".format(value))
+    while value and value[0] not in ATOM_ENDS:
+        token, value = get_atext(value)
+        dot_atom_text.append(token)
+        if value and value[0] == '.':
+            dot_atom_text.append(DOT)
+            value = value[1:]
+    if dot_atom_text[-1] is DOT:
+        raise errors.HeaderParseError("expected atom at end of dot-atom-text "
+            "but found '{}'".format('.'+value))
+    return dot_atom_text, value
+
+def get_dot_atom(value):
+    """ dot-atom = [CFWS] dot-atom-text [CFWS]
+
+    Any place we can have a dot atom, we could instead have an rfc2047 encoded
+    word.
+    """
+    dot_atom = DotAtom()
+    if value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        dot_atom.append(token)
+    if value.startswith('=?'):
+        try:
+            token, value = get_encoded_word(value)
+        except errors.HeaderParseError:
+            # XXX: need to figure out how to register defects when
+            # appropriate here.
+            token, value = get_dot_atom_text(value)
+    else:
+        token, value = get_dot_atom_text(value)
+    dot_atom.append(token)
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        dot_atom.append(token)
+    return dot_atom, value
+
+def get_word(value):
+    """word = atom / quoted-string
+
+    Either atom or quoted-string may start with CFWS.  We have to peel off this
+    CFWS first to determine which type of word to parse.  Afterward we splice
+    the leading CFWS, if any, into the parsed sub-token.
+
+    If neither an atom or a quoted-string is found before the next special, a
+    HeaderParseError is raised.
+
+    The token returned is either an Atom or a QuotedString, as appropriate.
+    This means the 'word' level of the formal grammar is not represented in the
+    parse tree; this is because having that extra layer when manipulating the
+    parse tree is more confusing than it is helpful.
+
+    """
+    if value[0] in CFWS_LEADER:
+        leader, value = get_cfws(value)
+    else:
+        leader = None
+    if value[0]=='"':
+        token, value = get_quoted_string(value)
+    elif value[0] in SPECIALS:
+        raise errors.HeaderParseError("Expected 'atom' or 'quoted-string' "
+                                      "but found '{}'".format(value))
+    else:
+        token, value = get_atom(value)
+    if leader is not None:
+        token[:0] = [leader]
+    return token, value
+
+def get_phrase(value):
+    """ phrase = 1*word / obs-phrase
+        obs-phrase = word *(word / "." / CFWS)
+
+    This means a phrase can be a sequence of words, periods, and CFWS in any
+    order as long as it starts with at least one word.  If anything other than
+    words is detected, an ObsoleteHeaderDefect is added to the token's defect
+    list.  We also accept a phrase that starts with CFWS followed by a dot;
+    this is registered as an InvalidHeaderDefect, since it is not supported by
+    even the obsolete grammar.
+
+    """
+    phrase = Phrase()
+    try:
+        token, value = get_word(value)
+        phrase.append(token)
+    except errors.HeaderParseError:
+        phrase.defects.append(errors.InvalidHeaderDefect(
+            "phrase does not start with word"))
+    while value and value[0] not in PHRASE_ENDS:
+        if value[0]=='.':
+            phrase.append(DOT)
+            phrase.defects.append(errors.ObsoleteHeaderDefect(
+                "period in 'phrase'"))
+            value = value[1:]
+        else:
+            try:
+                token, value = get_word(value)
+            except errors.HeaderParseError:
+                if value[0] in CFWS_LEADER:
+                    token, value = get_cfws(value)
+                    phrase.defects.append(errors.ObsoleteHeaderDefect(
+                        "comment found without atom"))
+                else:
+                    raise
+            phrase.append(token)
+    return phrase, value
+
+def get_local_part(value):
+    """ local-part = dot-atom / quoted-string / obs-local-part
+
+    """
+    local_part = LocalPart()
+    leader = None
+    if value[0] in CFWS_LEADER:
+        leader, value = get_cfws(value)
+    if not value:
+        raise errors.HeaderParseError(
+            "expected local-part but found '{}'".format(value))
+    try:
+        token, value = get_dot_atom(value)
+    except errors.HeaderParseError:
+        try:
+            token, value = get_word(value)
+        except errors.HeaderParseError:
+            if value[0] != '\\' and value[0] in PHRASE_ENDS:
+                raise
+            token = TokenList()
+    if leader is not None:
+        token[:0] = [leader]
+    local_part.append(token)
+    if value and (value[0]=='\\' or value[0] not in PHRASE_ENDS):
+        obs_local_part, value = get_obs_local_part(str(local_part) + value)
+        if obs_local_part.token_type == 'invalid-obs-local-part':
+            local_part.defects.append(errors.InvalidHeaderDefect(
+                "local-part is not dot-atom, quoted-string, or obs-local-part"))
+        else:
+            local_part.defects.append(errors.ObsoleteHeaderDefect(
+                "local-part is not a dot-atom (contains CFWS)"))
+        local_part[0] = obs_local_part
+    try:
+        local_part.value.encode('ascii')
+    except UnicodeEncodeError:
+        local_part.defects.append(errors.NonASCIILocalPartDefect(
+                "local-part contains non-ASCII characters)"))
+    return local_part, value
+
+def get_obs_local_part(value):
+    """ obs-local-part = word *("." word)
+    """
+    obs_local_part = ObsLocalPart()
+    last_non_ws_was_dot = False
+    while value and (value[0]=='\\' or value[0] not in PHRASE_ENDS):
+        if value[0] == '.':
+            if last_non_ws_was_dot:
+                obs_local_part.defects.append(errors.InvalidHeaderDefect(
+                    "invalid repeated '.'"))
+            obs_local_part.append(DOT)
+            last_non_ws_was_dot = True
+            value = value[1:]
+            continue
+        elif value[0]=='\\':
+            obs_local_part.append(ValueTerminal(value[0],
+                                                'misplaced-special'))
+            value = value[1:]
+            obs_local_part.defects.append(errors.InvalidHeaderDefect(
+                "'\\' character outside of quoted-string/ccontent"))
+            last_non_ws_was_dot = False
+            continue
+        if obs_local_part and obs_local_part[-1].token_type != 'dot':
+            obs_local_part.defects.append(errors.InvalidHeaderDefect(
+                "missing '.' between words"))
+        try:
+            token, value = get_word(value)
+            last_non_ws_was_dot = False
+        except errors.HeaderParseError:
+            if value[0] not in CFWS_LEADER:
+                raise
+            token, value = get_cfws(value)
+        obs_local_part.append(token)
+    if (obs_local_part[0].token_type == 'dot' or
+            obs_local_part[0].token_type=='cfws' and
+            obs_local_part[1].token_type=='dot'):
+        obs_local_part.defects.append(errors.InvalidHeaderDefect(
+            "Invalid leading '.' in local part"))
+    if (obs_local_part[-1].token_type == 'dot' or
+            obs_local_part[-1].token_type=='cfws' and
+            obs_local_part[-2].token_type=='dot'):
+        obs_local_part.defects.append(errors.InvalidHeaderDefect(
+            "Invalid trailing '.' in local part"))
+    if obs_local_part.defects:
+        obs_local_part.token_type = 'invalid-obs-local-part'
+    return obs_local_part, value
+
+def get_dtext(value):
+    r""" dtext = <printable ascii except \ [ ]> / obs-dtext
+        obs-dtext = obs-NO-WS-CTL / quoted-pair
+
+    We allow anything except the excluded characters, but if we find any
+    ASCII other than the RFC defined printable ASCII, a NonPrintableDefect is
+    added to the token's defects list.  Quoted pairs are converted to their
+    unquoted values, so what is returned is a ptext token, in this case a
+    ValueTerminal.  If there were quoted-printables, an ObsoleteHeaderDefect is
+    added to the returned token's defect list.
+
+    """
+    ptext, value, had_qp = _get_ptext_to_endchars(value, '[]')
+    ptext = ValueTerminal(ptext, 'ptext')
+    if had_qp:
+        ptext.defects.append(errors.ObsoleteHeaderDefect(
+            "quoted printable found in domain-literal"))
+    _validate_xtext(ptext)
+    return ptext, value
+
+def _check_for_early_dl_end(value, domain_literal):
+    if value:
+        return False
+    domain_literal.append(errors.InvalidHeaderDefect(
+        "end of input inside domain-literal"))
+    domain_literal.append(ValueTerminal(']', 'domain-literal-end'))
+    return True
+
+def get_domain_literal(value):
+    """ domain-literal = [CFWS] "[" *([FWS] dtext) [FWS] "]" [CFWS]
+
+    """
+    domain_literal = DomainLiteral()
+    if value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        domain_literal.append(token)
+    if not value:
+        raise errors.HeaderParseError("expected domain-literal")
+    if value[0] != '[':
+        raise errors.HeaderParseError("expected '[' at start of domain-literal "
+                "but found '{}'".format(value))
+    value = value[1:]
+    if _check_for_early_dl_end(value, domain_literal):
+        return domain_literal, value
+    domain_literal.append(ValueTerminal('[', 'domain-literal-start'))
+    if value[0] in WSP:
+        token, value = get_fws(value)
+        domain_literal.append(token)
+    token, value = get_dtext(value)
+    domain_literal.append(token)
+    if _check_for_early_dl_end(value, domain_literal):
+        return domain_literal, value
+    if value[0] in WSP:
+        token, value = get_fws(value)
+        domain_literal.append(token)
+    if _check_for_early_dl_end(value, domain_literal):
+        return domain_literal, value
+    if value[0] != ']':
+        raise errors.HeaderParseError("expected ']' at end of domain-literal "
+                "but found '{}'".format(value))
+    domain_literal.append(ValueTerminal(']', 'domain-literal-end'))
+    value = value[1:]
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        domain_literal.append(token)
+    return domain_literal, value
+
+def get_domain(value):
+    """ domain = dot-atom / domain-literal / obs-domain
+        obs-domain = atom *("." atom))
+
+    """
+    domain = Domain()
+    leader = None
+    if value[0] in CFWS_LEADER:
+        leader, value = get_cfws(value)
+    if not value:
+        raise errors.HeaderParseError(
+            "expected domain but found '{}'".format(value))
+    if value[0] == '[':
+        token, value = get_domain_literal(value)
+        if leader is not None:
+            token[:0] = [leader]
+        domain.append(token)
+        return domain, value
+    try:
+        token, value = get_dot_atom(value)
+    except errors.HeaderParseError:
+        token, value = get_atom(value)
+    if leader is not None:
+        token[:0] = [leader]
+    domain.append(token)
+    if value and value[0] == '.':
+        domain.defects.append(errors.ObsoleteHeaderDefect(
+            "domain is not a dot-atom (contains CFWS)"))
+        if domain[0].token_type == 'dot-atom':
+            domain[:] = domain[0]
+        while value and value[0] == '.':
+            domain.append(DOT)
+            token, value = get_atom(value[1:])
+            domain.append(token)
+    return domain, value
+
+def get_addr_spec(value):
+    """ addr-spec = local-part "@" domain
+
+    """
+    addr_spec = AddrSpec()
+    token, value = get_local_part(value)
+    addr_spec.append(token)
+    if not value or value[0] != '@':
+        addr_spec.defects.append(errors.InvalidHeaderDefect(
+            "add-spec local part with no domain"))
+        return addr_spec, value
+    addr_spec.append(ValueTerminal('@', 'address-at-symbol'))
+    token, value = get_domain(value[1:])
+    addr_spec.append(token)
+    return addr_spec, value
+
+def get_obs_route(value):
+    """ obs-route = obs-domain-list ":"
+        obs-domain-list = *(CFWS / ",") "@" domain *("," [CFWS] ["@" domain])
+
+        Returns an obs-route token with the appropriate sub-tokens (that is,
+        there is no obs-domain-list in the parse tree).
+    """
+    obs_route = ObsRoute()
+    while value and (value[0]==',' or value[0] in CFWS_LEADER):
+        if value[0] in CFWS_LEADER:
+            token, value = get_cfws(value)
+            obs_route.append(token)
+        elif value[0] == ',':
+            obs_route.append(ListSeparator)
+            value = value[1:]
+    if not value or value[0] != '@':
+        raise errors.HeaderParseError(
+            "expected obs-route domain but found '{}'".format(value))
+    obs_route.append(RouteComponentMarker)
+    token, value = get_domain(value[1:])
+    obs_route.append(token)
+    while value and value[0]==',':
+        obs_route.append(ListSeparator)
+        value = value[1:]
+        if not value:
+            break
+        if value[0] in CFWS_LEADER:
+            token, value = get_cfws(value)
+            obs_route.append(token)
+        if value[0] == '@':
+            obs_route.append(RouteComponentMarker)
+            token, value = get_domain(value[1:])
+            obs_route.append(token)
+    if not value:
+        raise errors.HeaderParseError("end of header while parsing obs-route")
+    if value[0] != ':':
+        raise errors.HeaderParseError( "expected ':' marking end of "
+            "obs-route but found '{}'".format(value))
+    obs_route.append(ValueTerminal(':', 'end-of-obs-route-marker'))
+    return obs_route, value[1:]
+
+def get_angle_addr(value):
+    """ angle-addr = [CFWS] "<" addr-spec ">" [CFWS] / obs-angle-addr
+        obs-angle-addr = [CFWS] "<" obs-route addr-spec ">" [CFWS]
+
+    """
+    angle_addr = AngleAddr()
+    if value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        angle_addr.append(token)
+    if not value or value[0] != '<':
+        raise errors.HeaderParseError(
+            "expected angle-addr but found '{}'".format(value))
+    angle_addr.append(ValueTerminal('<', 'angle-addr-start'))
+    value = value[1:]
+    # Although it is not legal per RFC5322, SMTP uses '<>' in certain
+    # circumstances.
+    if value[0] == '>':
+        angle_addr.append(ValueTerminal('>', 'angle-addr-end'))
+        angle_addr.defects.append(errors.InvalidHeaderDefect(
+            "null addr-spec in angle-addr"))
+        value = value[1:]
+        return angle_addr, value
+    try:
+        token, value = get_addr_spec(value)
+    except errors.HeaderParseError:
+        try:
+            token, value = get_obs_route(value)
+            angle_addr.defects.append(errors.ObsoleteHeaderDefect(
+                "obsolete route specification in angle-addr"))
+        except errors.HeaderParseError:
+            raise errors.HeaderParseError(
+                "expected addr-spec or obs-route but found '{}'".format(value))
+        angle_addr.append(token)
+        token, value = get_addr_spec(value)
+    angle_addr.append(token)
+    if value and value[0] == '>':
+        value = value[1:]
+    else:
+        angle_addr.defects.append(errors.InvalidHeaderDefect(
+            "missing trailing '>' on angle-addr"))
+    angle_addr.append(ValueTerminal('>', 'angle-addr-end'))
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        angle_addr.append(token)
+    return angle_addr, value
+
+def get_display_name(value):
+    """ display-name = phrase
+
+    Because this is simply a name-rule, we don't return a display-name
+    token containing a phrase, but rather a display-name token with
+    the content of the phrase.
+
+    """
+    display_name = DisplayName()
+    token, value = get_phrase(value)
+    display_name.extend(token[:])
+    display_name.defects = token.defects[:]
+    return display_name, value
+
+
+def get_name_addr(value):
+    """ name-addr = [display-name] angle-addr
+
+    """
+    name_addr = NameAddr()
+    # Both the optional display name and the angle-addr can start with cfws.
+    leader = None
+    if value[0] in CFWS_LEADER:
+        leader, value = get_cfws(value)
+        if not value:
+            raise errors.HeaderParseError(
+                "expected name-addr but found '{}'".format(leader))
+    if value[0] != '<':
+        if value[0] in PHRASE_ENDS:
+            raise errors.HeaderParseError(
+                "expected name-addr but found '{}'".format(value))
+        token, value = get_display_name(value)
+        if not value:
+            raise errors.HeaderParseError(
+                "expected name-addr but found '{}'".format(token))
+        if leader is not None:
+            token[0][:0] = [leader]
+            leader = None
+        name_addr.append(token)
+    token, value = get_angle_addr(value)
+    if leader is not None:
+        token[:0] = [leader]
+    name_addr.append(token)
+    return name_addr, value
+
+def get_mailbox(value):
+    """ mailbox = name-addr / addr-spec
+
+    """
+    # The only way to figure out if we are dealing with a name-addr or an
+    # addr-spec is to try parsing each one.
+    mailbox = Mailbox()
+    try:
+        token, value = get_name_addr(value)
+    except errors.HeaderParseError:
+        try:
+            token, value = get_addr_spec(value)
+        except errors.HeaderParseError:
+            raise errors.HeaderParseError(
+                "expected mailbox but found '{}'".format(value))
+    if any(isinstance(x, errors.InvalidHeaderDefect)
+                       for x in token.all_defects):
+        mailbox.token_type = 'invalid-mailbox'
+    mailbox.append(token)
+    return mailbox, value
+
+def get_invalid_mailbox(value, endchars):
+    """ Read everything up to one of the chars in endchars.
+
+    This is outside the formal grammar.  The InvalidMailbox TokenList that is
+    returned acts like a Mailbox, but the data attributes are None.
+
+    """
+    invalid_mailbox = InvalidMailbox()
+    while value and value[0] not in endchars:
+        if value[0] in PHRASE_ENDS:
+            invalid_mailbox.append(ValueTerminal(value[0],
+                                                 'misplaced-special'))
+            value = value[1:]
+        else:
+            token, value = get_phrase(value)
+            invalid_mailbox.append(token)
+    return invalid_mailbox, value
+
+def get_mailbox_list(value):
+    """ mailbox-list = (mailbox *("," mailbox)) / obs-mbox-list
+        obs-mbox-list = *([CFWS] ",") mailbox *("," [mailbox / CFWS])
+
+    For this routine we go outside the formal grammar in order to improve error
+    handling.  We recognize the end of the mailbox list only at the end of the
+    value or at a ';' (the group terminator).  This is so that we can turn
+    invalid mailboxes into InvalidMailbox tokens and continue parsing any
+    remaining valid mailboxes.  We also allow all mailbox entries to be null,
+    and this condition is handled appropriately at a higher level.
+
+    """
+    mailbox_list = MailboxList()
+    while value and value[0] != ';':
+        try:
+            token, value = get_mailbox(value)
+            mailbox_list.append(token)
+        except errors.HeaderParseError:
+            leader = None
+            if value[0] in CFWS_LEADER:
+                leader, value = get_cfws(value)
+                if not value or value[0] in ',;':
+                    mailbox_list.append(leader)
+                    mailbox_list.defects.append(errors.ObsoleteHeaderDefect(
+                        "empty element in mailbox-list"))
+                else:
+                    token, value = get_invalid_mailbox(value, ',;')
+                    if leader is not None:
+                        token[:0] = [leader]
+                    mailbox_list.append(token)
+                    mailbox_list.defects.append(errors.InvalidHeaderDefect(
+                        "invalid mailbox in mailbox-list"))
+            elif value[0] == ',':
+                mailbox_list.defects.append(errors.ObsoleteHeaderDefect(
+                    "empty element in mailbox-list"))
+            else:
+                token, value = get_invalid_mailbox(value, ',;')
+                if leader is not None:
+                    token[:0] = [leader]
+                mailbox_list.append(token)
+                mailbox_list.defects.append(errors.InvalidHeaderDefect(
+                    "invalid mailbox in mailbox-list"))
+        if value and value[0] not in ',;':
+            # Crap after mailbox; treat it as an invalid mailbox.
+            # The mailbox info will still be available.
+            mailbox = mailbox_list[-1]
+            mailbox.token_type = 'invalid-mailbox'
+            token, value = get_invalid_mailbox(value, ',;')
+            mailbox.extend(token)
+            mailbox_list.defects.append(errors.InvalidHeaderDefect(
+                "invalid mailbox in mailbox-list"))
+        if value and value[0] == ',':
+            mailbox_list.append(ListSeparator)
+            value = value[1:]
+    return mailbox_list, value
+
+
+def get_group_list(value):
+    """ group-list = mailbox-list / CFWS / obs-group-list
+        obs-group-list = 1*([CFWS] ",") [CFWS]
+
+    """
+    group_list = GroupList()
+    if not value:
+        group_list.defects.append(errors.InvalidHeaderDefect(
+            "end of header before group-list"))
+        return group_list, value
+    leader = None
+    if value and value[0] in CFWS_LEADER:
+        leader, value = get_cfws(value)
+        if not value:
+            # This should never happen in email parsing, since CFWS-only is a
+            # legal alternative to group-list in a group, which is the only
+            # place group-list appears.
+            group_list.defects.append(errors.InvalidHeaderDefect(
+                "end of header in group-list"))
+            group_list.append(leader)
+            return group_list, value
+        if value[0] == ';':
+            group_list.append(leader)
+            return group_list, value
+    token, value = get_mailbox_list(value)
+    if len(token.all_mailboxes)==0:
+        if leader is not None:
+            group_list.append(leader)
+        group_list.extend(token)
+        group_list.defects.append(errors.ObsoleteHeaderDefect(
+            "group-list with empty entries"))
+        return group_list, value
+    if leader is not None:
+        token[:0] = [leader]
+    group_list.append(token)
+    return group_list, value
+
+def get_group(value):
+    """ group = display-name ":" [group-list] ";" [CFWS]
+
+    """
+    group = Group()
+    token, value = get_display_name(value)
+    if not value or value[0] != ':':
+        raise errors.HeaderParseError("expected ':' at end of group "
+            "display name but found '{}'".format(value))
+    group.append(token)
+    group.append(ValueTerminal(':', 'group-display-name-terminator'))
+    value = value[1:]
+    if value and value[0] == ';':
+        group.append(ValueTerminal(';', 'group-terminator'))
+        return group, value[1:]
+    token, value = get_group_list(value)
+    group.append(token)
+    if not value:
+        group.defects.append(errors.InvalidHeaderDefect(
+            "end of header in group"))
+    if value[0] != ';':
+        raise errors.HeaderParseError(
+            "expected ';' at end of group but found {}".format(value))
+    group.append(ValueTerminal(';', 'group-terminator'))
+    value = value[1:]
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        group.append(token)
+    return group, value
+
+def get_address(value):
+    """ address = mailbox / group
+
+    Note that counter-intuitively, an address can be either a single address or
+    a list of addresses (a group).  This is why the returned Address object has
+    a 'mailboxes' attribute which treats a single address as a list of length
+    one.  When you need to differentiate between to two cases, extract the single
+    element, which is either a mailbox or a group token.
+
+    """
+    # The formal grammar isn't very helpful when parsing an address.  mailbox
+    # and group, especially when allowing for obsolete forms, start off very
+    # similarly.  It is only when you reach one of @, <, or : that you know
+    # what you've got.  So, we try each one in turn, starting with the more
+    # likely of the two.  We could perhaps make this more efficient by looking
+    # for a phrase and then branching based on the next character, but that
+    # would be a premature optimization.
+    address = Address()
+    try:
+        token, value = get_group(value)
+    except errors.HeaderParseError:
+        try:
+            token, value = get_mailbox(value)
+        except errors.HeaderParseError:
+            raise errors.HeaderParseError(
+                "expected address but found '{}'".format(value))
+    address.append(token)
+    return address, value
+
+def get_address_list(value):
+    """ address_list = (address *("," address)) / obs-addr-list
+        obs-addr-list = *([CFWS] ",") address *("," [address / CFWS])
+
+    We depart from the formal grammar here by continuing to parse until the end
+    of the input, assuming the input to be entirely composed of an
+    address-list.  This is always true in email parsing, and allows us
+    to skip invalid addresses to parse additional valid ones.
+
+    """
+    address_list = AddressList()
+    while value:
+        try:
+            token, value = get_address(value)
+            address_list.append(token)
+        except errors.HeaderParseError as err:
+            leader = None
+            if value[0] in CFWS_LEADER:
+                leader, value = get_cfws(value)
+                if not value or value[0] == ',':
+                    address_list.append(leader)
+                    address_list.defects.append(errors.ObsoleteHeaderDefect(
+                        "address-list entry with no content"))
+                else:
+                    token, value = get_invalid_mailbox(value, ',')
+                    if leader is not None:
+                        token[:0] = [leader]
+                    address_list.append(Address([token]))
+                    address_list.defects.append(errors.InvalidHeaderDefect(
+                        "invalid address in address-list"))
+            elif value[0] == ',':
+                address_list.defects.append(errors.ObsoleteHeaderDefect(
+                    "empty element in address-list"))
+            else:
+                token, value = get_invalid_mailbox(value, ',')
+                if leader is not None:
+                    token[:0] = [leader]
+                address_list.append(Address([token]))
+                address_list.defects.append(errors.InvalidHeaderDefect(
+                    "invalid address in address-list"))
+        if value and value[0] != ',':
+            # Crap after address; treat it as an invalid mailbox.
+            # The mailbox info will still be available.
+            mailbox = address_list[-1][0]
+            mailbox.token_type = 'invalid-mailbox'
+            token, value = get_invalid_mailbox(value, ',')
+            mailbox.extend(token)
+            address_list.defects.append(errors.InvalidHeaderDefect(
+                "invalid address in address-list"))
+        if value:  # Must be a , at this point.
+            address_list.append(ValueTerminal(',', 'list-separator'))
+            value = value[1:]
+    return address_list, value
+
+#
+# XXX: As I begin to add additional header parsers, I'm realizing we probably
+# have two level of parser routines: the get_XXX methods that get a token in
+# the grammar, and parse_XXX methods that parse an entire field value.  So
+# get_address_list above should really be a parse_ method, as probably should
+# be get_unstructured.
+#
+
+def parse_mime_version(value):
+    """ mime-version = [CFWS] 1*digit [CFWS] "." [CFWS] 1*digit [CFWS]
+
+    """
+    # The [CFWS] is implicit in the RFC 2045 BNF.
+    # XXX: This routine is a bit verbose, should factor out a get_int method.
+    mime_version = MIMEVersion()
+    if not value:
+        mime_version.defects.append(errors.HeaderMissingRequiredValue(
+            "Missing MIME version number (eg: 1.0)"))
+        return mime_version
+    if value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        mime_version.append(token)
+        if not value:
+            mime_version.defects.append(errors.HeaderMissingRequiredValue(
+                "Expected MIME version number but found only CFWS"))
+    digits = ''
+    while value and value[0] != '.' and value[0] not in CFWS_LEADER:
+        digits += value[0]
+        value = value[1:]
+    if not digits.isdigit():
+        mime_version.defects.append(errors.InvalidHeaderDefect(
+            "Expected MIME major version number but found {!r}".format(digits)))
+        mime_version.append(ValueTerminal(digits, 'xtext'))
+    else:
+        mime_version.major = int(digits)
+        mime_version.append(ValueTerminal(digits, 'digits'))
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        mime_version.append(token)
+    if not value or value[0] != '.':
+        if mime_version.major is not None:
+            mime_version.defects.append(errors.InvalidHeaderDefect(
+                "Incomplete MIME version; found only major number"))
+        if value:
+            mime_version.append(ValueTerminal(value, 'xtext'))
+        return mime_version
+    mime_version.append(ValueTerminal('.', 'version-separator'))
+    value = value[1:]
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        mime_version.append(token)
+    if not value:
+        if mime_version.major is not None:
+            mime_version.defects.append(errors.InvalidHeaderDefect(
+                "Incomplete MIME version; found only major number"))
+        return mime_version
+    digits = ''
+    while value and value[0] not in CFWS_LEADER:
+        digits += value[0]
+        value = value[1:]
+    if not digits.isdigit():
+        mime_version.defects.append(errors.InvalidHeaderDefect(
+            "Expected MIME minor version number but found {!r}".format(digits)))
+        mime_version.append(ValueTerminal(digits, 'xtext'))
+    else:
+        mime_version.minor = int(digits)
+        mime_version.append(ValueTerminal(digits, 'digits'))
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        mime_version.append(token)
+    if value:
+        mime_version.defects.append(errors.InvalidHeaderDefect(
+            "Excess non-CFWS text after MIME version"))
+        mime_version.append(ValueTerminal(value, 'xtext'))
+    return mime_version
+
+def get_invalid_parameter(value):
+    """ Read everything up to the next ';'.
+
+    This is outside the formal grammar.  The InvalidParameter TokenList that is
+    returned acts like a Parameter, but the data attributes are None.
+
+    """
+    invalid_parameter = InvalidParameter()
+    while value and value[0] != ';':
+        if value[0] in PHRASE_ENDS:
+            invalid_parameter.append(ValueTerminal(value[0],
+                                                   'misplaced-special'))
+            value = value[1:]
+        else:
+            token, value = get_phrase(value)
+            invalid_parameter.append(token)
+    return invalid_parameter, value
+
+def get_ttext(value):
+    """ttext = <matches _ttext_matcher>
+
+    We allow any non-TOKEN_ENDS in ttext, but add defects to the token's
+    defects list if we find non-ttext characters.  We also register defects for
+    *any* non-printables even though the RFC doesn't exclude all of them,
+    because we follow the spirit of RFC 5322.
+
+    """
+    m = _non_token_end_matcher(value)
+    if not m:
+        raise errors.HeaderParseError(
+            "expected ttext but found '{}'".format(value))
+    ttext = m.group()
+    value = value[len(ttext):]
+    ttext = ValueTerminal(ttext, 'ttext')
+    _validate_xtext(ttext)
+    return ttext, value
+
+def get_token(value):
+    """token = [CFWS] 1*ttext [CFWS]
+
+    The RFC equivalent of ttext is any US-ASCII chars except space, ctls, or
+    tspecials.  We also exclude tabs even though the RFC doesn't.
+
+    The RFC implies the CFWS but is not explicit about it in the BNF.
+
+    """
+    mtoken = Token()
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        mtoken.append(token)
+    if value and value[0] in TOKEN_ENDS:
+        raise errors.HeaderParseError(
+            "expected token but found '{}'".format(value))
+    token, value = get_ttext(value)
+    mtoken.append(token)
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        mtoken.append(token)
+    return mtoken, value
+
+def get_attrtext(value):
+    """attrtext = 1*(any non-ATTRIBUTE_ENDS character)
+
+    We allow any non-ATTRIBUTE_ENDS in attrtext, but add defects to the
+    token's defects list if we find non-attrtext characters.  We also register
+    defects for *any* non-printables even though the RFC doesn't exclude all of
+    them, because we follow the spirit of RFC 5322.
+
+    """
+    m = _non_attribute_end_matcher(value)
+    if not m:
+        raise errors.HeaderParseError(
+            "expected attrtext but found {!r}".format(value))
+    attrtext = m.group()
+    value = value[len(attrtext):]
+    attrtext = ValueTerminal(attrtext, 'attrtext')
+    _validate_xtext(attrtext)
+    return attrtext, value
+
+def get_attribute(value):
+    """ [CFWS] 1*attrtext [CFWS]
+
+    This version of the BNF makes the CFWS explicit, and as usual we use a
+    value terminal for the actual run of characters.  The RFC equivalent of
+    attrtext is the token characters, with the subtraction of '*', "'", and '%'.
+    We include tab in the excluded set just as we do for token.
+
+    """
+    attribute = Attribute()
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        attribute.append(token)
+    if value and value[0] in ATTRIBUTE_ENDS:
+        raise errors.HeaderParseError(
+            "expected token but found '{}'".format(value))
+    token, value = get_attrtext(value)
+    attribute.append(token)
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        attribute.append(token)
+    return attribute, value
+
+def get_extended_attrtext(value):
+    """attrtext = 1*(any non-ATTRIBUTE_ENDS character plus '%')
+
+    This is a special parsing routine so that we get a value that
+    includes % escapes as a single string (which we decode as a single
+    string later).
+
+    """
+    m = _non_extended_attribute_end_matcher(value)
+    if not m:
+        raise errors.HeaderParseError(
+            "expected extended attrtext but found {!r}".format(value))
+    attrtext = m.group()
+    value = value[len(attrtext):]
+    attrtext = ValueTerminal(attrtext, 'extended-attrtext')
+    _validate_xtext(attrtext)
+    return attrtext, value
+
+def get_extended_attribute(value):
+    """ [CFWS] 1*extended_attrtext [CFWS]
+
+    This is like the non-extended version except we allow % characters, so that
+    we can pick up an encoded value as a single string.
+
+    """
+    # XXX: should we have an ExtendedAttribute TokenList?
+    attribute = Attribute()
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        attribute.append(token)
+    if value and value[0] in EXTENDED_ATTRIBUTE_ENDS:
+        raise errors.HeaderParseError(
+            "expected token but found '{}'".format(value))
+    token, value = get_extended_attrtext(value)
+    attribute.append(token)
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        attribute.append(token)
+    return attribute, value
+
+def get_section(value):
+    """ '*' digits
+
+    The formal BNF is more complicated because leading 0s are not allowed.  We
+    check for that and add a defect.  We also assume no CFWS is allowed between
+    the '*' and the digits, though the RFC is not crystal clear on that.
+    The caller should already have dealt with leading CFWS.
+
+    """
+    section = Section()
+    if not value or value[0] != '*':
+        raise errors.HeaderParseError("Expected section but found {}".format(
+                                        value))
+    section.append(ValueTerminal('*', 'section-marker'))
+    value = value[1:]
+    if not value or not value[0].isdigit():
+        raise errors.HeaderParseError("Expected section number but "
+                                      "found {}".format(value))
+    digits = ''
+    while value and value[0].isdigit():
+        digits += value[0]
+        value = value[1:]
+    if digits[0] == '0' and digits != '0':
+        section.defects.append(errors.InvalidHeaderError("section number"
+            "has an invalid leading 0"))
+    section.number = int(digits)
+    section.append(ValueTerminal(digits, 'digits'))
+    return section, value
+
+
+def get_value(value):
+    """ quoted-string / attribute
+
+    """
+    v = Value()
+    if not value:
+        raise errors.HeaderParseError("Expected value but found end of string")
+    leader = None
+    if value[0] in CFWS_LEADER:
+        leader, value = get_cfws(value)
+    if not value:
+        raise errors.HeaderParseError("Expected value but found "
+                                      "only {}".format(leader))
+    if value[0] == '"':
+        token, value = get_quoted_string(value)
+    else:
+        token, value = get_extended_attribute(value)
+    if leader is not None:
+        token[:0] = [leader]
+    v.append(token)
+    return v, value
+
+def get_parameter(value):
+    """ attribute [section] ["*"] [CFWS] "=" value
+
+    The CFWS is implied by the RFC but not made explicit in the BNF.  This
+    simplified form of the BNF from the RFC is made to conform with the RFC BNF
+    through some extra checks.  We do it this way because it makes both error
+    recovery and working with the resulting parse tree easier.
+    """
+    # It is possible CFWS would also be implicitly allowed between the section
+    # and the 'extended-attribute' marker (the '*') , but we've never seen that
+    # in the wild and we will therefore ignore the possibility.
+    param = Parameter()
+    token, value = get_attribute(value)
+    param.append(token)
+    if not value or value[0] == ';':
+        param.defects.append(errors.InvalidHeaderDefect("Parameter contains "
+            "name ({}) but no value".format(token)))
+        return param, value
+    if value[0] == '*':
+        try:
+            token, value = get_section(value)
+            param.sectioned = True
+            param.append(token)
+        except errors.HeaderParseError:
+            pass
+        if not value:
+            raise errors.HeaderParseError("Incomplete parameter")
+        if value[0] == '*':
+            param.append(ValueTerminal('*', 'extended-parameter-marker'))
+            value = value[1:]
+            param.extended = True
+    if value[0] != '=':
+        raise errors.HeaderParseError("Parameter not followed by '='")
+    param.append(ValueTerminal('=', 'parameter-separator'))
+    value = value[1:]
+    leader = None
+    if value and value[0] in CFWS_LEADER:
+        token, value = get_cfws(value)
+        param.append(token)
+    remainder = None
+    appendto = param
+    if param.extended and value and value[0] == '"':
+        # Now for some serious hackery to handle the common invalid case of
+        # double quotes around an extended value.  We also accept (with defect)
+        # a value marked as encoded that isn't really.
+        qstring, remainder = get_quoted_string(value)
+        inner_value = qstring.stripped_value
+        semi_valid = False
+        if param.section_number == 0:
+            if inner_value and inner_value[0] == "'":
+                semi_valid = True
+            else:
+                token, rest = get_attrtext(inner_value)
+                if rest and rest[0] == "'":
+                    semi_valid = True
+        else:
+            try:
+                token, rest = get_extended_attrtext(inner_value)
+            except:
+                pass
+            else:
+                if not rest:
+                    semi_valid = True
+        if semi_valid:
+            param.defects.append(errors.InvalidHeaderDefect(
+                "Quoted string value for extended parameter is invalid"))
+            param.append(qstring)
+            for t in qstring:
+                if t.token_type == 'bare-quoted-string':
+                    t[:] = []
+                    appendto = t
+                    break
+            value = inner_value
+        else:
+            remainder = None
+            param.defects.append(errors.InvalidHeaderDefect(
+                "Parameter marked as extended but appears to have a "
+                "quoted string value that is non-encoded"))
+    if value and value[0] == "'":
+        token = None
+    else:
+        token, value = get_value(value)
+    if not param.extended or param.section_number > 0:
+        if not value or value[0] != "'":
+            appendto.append(token)
+            if remainder is not None:
+                assert not value, value
+                value = remainder
+            return param, value
+        param.defects.append(errors.InvalidHeaderDefect(
+            "Apparent initial-extended-value but attribute "
+            "was not marked as extended or was not initial section"))
+    if not value:
+        # Assume the charset/lang is missing and the token is the value.
+        param.defects.append(errors.InvalidHeaderDefect(
+            "Missing required charset/lang delimiters"))
+        appendto.append(token)
+        if remainder is None:
+            return param, value
+    else:
+        if token is not None:
+            for t in token:
+                if t.token_type == 'extended-attrtext':
+                    break
+            t.token_type == 'attrtext'
+            appendto.append(t)
+            param.charset = t.value
+        if value[0] != "'":
+            raise errors.HeaderParseError("Expected RFC2231 char/lang encoding "
+                                          "delimiter, but found {!r}".format(value))
+        appendto.append(ValueTerminal("'", 'RFC2231 delimiter'))
+        value = value[1:]
+        if value and value[0] != "'":
+            token, value = get_attrtext(value)
+            appendto.append(token)
+            param.lang = token.value
+            if not value or value[0] != "'":
+                raise errors.HeaderParseError("Expected RFC2231 char/lang encoding "
+                                  "delimiter, but found {}".format(value))
+        appendto.append(ValueTerminal("'", 'RFC2231 delimiter'))
+        value = value[1:]
+    if remainder is not None:
+        # Treat the rest of value as bare quoted string content.
+        v = Value()
+        while value:
+            if value[0] in WSP:
+                token, value = get_fws(value)
+            else:
+                token, value = get_qcontent(value)
+            v.append(token)
+        token = v
+    else:
+        token, value = get_value(value)
+    appendto.append(token)
+    if remainder is not None:
+        assert not value, value
+        value = remainder
+    return param, value
+
+def parse_mime_parameters(value):
+    """ parameter *( ";" parameter )
+
+    That BNF is meant to indicate this routine should only be called after
+    finding and handling the leading ';'.  There is no corresponding rule in
+    the formal RFC grammar, but it is more convenient for us for the set of
+    parameters to be treated as its own TokenList.
+
+    This is 'parse' routine because it consumes the reminaing value, but it
+    would never be called to parse a full header.  Instead it is called to
+    parse everything after the non-parameter value of a specific MIME header.
+
+    """
+    mime_parameters = MimeParameters()
+    while value:
+        try:
+            token, value = get_parameter(value)
+            mime_parameters.append(token)
+        except errors.HeaderParseError as err:
+            leader = None
+            if value[0] in CFWS_LEADER:
+                leader, value = get_cfws(value)
+            if not value:
+                mime_parameters.append(leader)
+                return mime_parameters
+            if value[0] == ';':
+                if leader is not None:
+                    mime_parameters.append(leader)
+                mime_parameters.defects.append(errors.InvalidHeaderDefect(
+                    "parameter entry with no content"))
+            else:
+                token, value = get_invalid_parameter(value)
+                if leader:
+                    token[:0] = [leader]
+                mime_parameters.append(token)
+                mime_parameters.defects.append(errors.InvalidHeaderDefect(
+                    "invalid parameter {!r}".format(token)))
+        if value and value[0] != ';':
+            # Junk after the otherwise valid parameter.  Mark it as
+            # invalid, but it will have a value.
+            param = mime_parameters[-1]
+            param.token_type = 'invalid-parameter'
+            token, value = get_invalid_parameter(value)
+            param.extend(token)
+            mime_parameters.defects.append(errors.InvalidHeaderDefect(
+                "parameter with invalid trailing text {!r}".format(token)))
+        if value:
+            # Must be a ';' at this point.
+            mime_parameters.append(ValueTerminal(';', 'parameter-separator'))
+            value = value[1:]
+    return mime_parameters
+
+def _find_mime_parameters(tokenlist, value):
+    """Do our best to find the parameters in an invalid MIME header
+
+    """
+    while value and value[0] != ';':
+        if value[0] in PHRASE_ENDS:
+            tokenlist.append(ValueTerminal(value[0], 'misplaced-special'))
+            value = value[1:]
+        else:
+            token, value = get_phrase(value)
+            tokenlist.append(token)
+    if not value:
+        return
+    tokenlist.append(ValueTerminal(';', 'parameter-separator'))
+    tokenlist.append(parse_mime_parameters(value[1:]))
+
+def parse_content_type_header(value):
+    """ maintype "/" subtype *( ";" parameter )
+
+    The maintype and substype are tokens.  Theoretically they could
+    be checked against the official IANA list + x-token, but we
+    don't do that.
+    """
+    ctype = ContentType()
+    recover = False
+    if not value:
+        ctype.defects.append(errors.HeaderMissingRequiredValue(
+            "Missing content type specification"))
+        return ctype
+    try:
+        token, value = get_token(value)
+    except errors.HeaderParseError:
+        ctype.defects.append(errors.InvalidHeaderDefect(
+            "Expected content maintype but found {!r}".format(value)))
+        _find_mime_parameters(ctype, value)
+        return ctype
+    ctype.append(token)
+    # XXX: If we really want to follow the formal grammar we should make
+    # mantype and subtype specialized TokenLists here.  Probably not worth it.
+    if not value or value[0] != '/':
+        ctype.defects.append(errors.InvalidHeaderDefect(
+            "Invalid content type"))
+        if value:
+            _find_mime_parameters(ctype, value)
+        return ctype
+    ctype.maintype = token.value.strip().lower()
+    ctype.append(ValueTerminal('/', 'content-type-separator'))
+    value = value[1:]
+    try:
+        token, value = get_token(value)
+    except errors.HeaderParseError:
+        ctype.defects.append(errors.InvalidHeaderDefect(
+            "Expected content subtype but found {!r}".format(value)))
+        _find_mime_parameters(ctype, value)
+        return ctype
+    ctype.append(token)
+    ctype.subtype = token.value.strip().lower()
+    if not value:
+        return ctype
+    if value[0] != ';':
+        ctype.defects.append(errors.InvalidHeaderDefect(
+            "Only parameters are valid after content type, but "
+            "found {!r}".format(value)))
+        # The RFC requires that a syntactically invalid content-type be treated
+        # as text/plain.  Perhaps we should postel this, but we should probably
+        # only do that if we were checking the subtype value against IANA.
+        del ctype.maintype, ctype.subtype
+        _find_mime_parameters(ctype, value)
+        return ctype
+    ctype.append(ValueTerminal(';', 'parameter-separator'))
+    ctype.append(parse_mime_parameters(value[1:]))
+    return ctype
+
+def parse_content_disposition_header(value):
+    """ disposition-type *( ";" parameter )
+
+    """
+    disp_header = ContentDisposition()
+    if not value:
+        disp_header.defects.append(errors.HeaderMissingRequiredValue(
+            "Missing content disposition"))
+        return disp_header
+    try:
+        token, value = get_token(value)
+    except errors.HeaderParseError:
+        disp_header.defects.append(errors.InvalidHeaderDefect(
+            "Expected content disposition but found {!r}".format(value)))
+        _find_mime_parameters(disp_header, value)
+        return disp_header
+    disp_header.append(token)
+    disp_header.content_disposition = token.value.strip().lower()
+    if not value:
+        return disp_header
+    if value[0] != ';':
+        disp_header.defects.append(errors.InvalidHeaderDefect(
+            "Only parameters are valid after content disposition, but "
+            "found {!r}".format(value)))
+        _find_mime_parameters(disp_header, value)
+        return disp_header
+    disp_header.append(ValueTerminal(';', 'parameter-separator'))
+    disp_header.append(parse_mime_parameters(value[1:]))
+    return disp_header
+
+def parse_content_transfer_encoding_header(value):
+    """ mechanism
+
+    """
+    # We should probably validate the values, since the list is fixed.
+    cte_header = ContentTransferEncoding()
+    if not value:
+        cte_header.defects.append(errors.HeaderMissingRequiredValue(
+            "Missing content transfer encoding"))
+        return cte_header
+    try:
+        token, value = get_token(value)
+    except errors.HeaderParseError:
+        cte_header.defects.append(errors.InvalidHeaderDefect(
+            "Expected content transfer encoding but found {!r}".format(value)))
+    else:
+        cte_header.append(token)
+        cte_header.cte = token.value.strip().lower()
+    if not value:
+        return cte_header
+    while value:
+        cte_header.defects.append(errors.InvalidHeaderDefect(
+            "Extra text after content transfer encoding"))
+        if value[0] in PHRASE_ENDS:
+            cte_header.append(ValueTerminal(value[0], 'misplaced-special'))
+            value = value[1:]
+        else:
+            token, value = get_phrase(value)
+            cte_header.append(token)
+    return cte_header
diff --git a/Lib/email/_parseaddr.py b/Lib/email/_parseaddr.py
new file mode 100644
index 000000000..cdfa3729a
--- /dev/null
+++ b/Lib/email/_parseaddr.py
@@ -0,0 +1,540 @@
+# Copyright (C) 2002-2007 Python Software Foundation
+# Contact: email-sig@python.org
+
+"""Email address parsing code.
+
+Lifted directly from rfc822.py.  This should eventually be rewritten.
+"""
+
+__all__ = [
+    'mktime_tz',
+    'parsedate',
+    'parsedate_tz',
+    'quote',
+    ]
+
+import time, calendar
+
+SPACE = ' '
+EMPTYSTRING = ''
+COMMASPACE = ', '
+
+# Parse a date field
+_monthnames = ['jan', 'feb', 'mar', 'apr', 'may', 'jun', 'jul',
+               'aug', 'sep', 'oct', 'nov', 'dec',
+               'january', 'february', 'march', 'april', 'may', 'june', 'july',
+               'august', 'september', 'october', 'november', 'december']
+
+_daynames = ['mon', 'tue', 'wed', 'thu', 'fri', 'sat', 'sun']
+
+# The timezone table does not include the military time zones defined
+# in RFC822, other than Z.  According to RFC1123, the description in
+# RFC822 gets the signs wrong, so we can't rely on any such time
+# zones.  RFC1123 recommends that numeric timezone indicators be used
+# instead of timezone names.
+
+_timezones = {'UT':0, 'UTC':0, 'GMT':0, 'Z':0,
+              'AST': -400, 'ADT': -300,  # Atlantic (used in Canada)
+              'EST': -500, 'EDT': -400,  # Eastern
+              'CST': -600, 'CDT': -500,  # Central
+              'MST': -700, 'MDT': -600,  # Mountain
+              'PST': -800, 'PDT': -700   # Pacific
+              }
+
+
+def parsedate_tz(data):
+    """Convert a date string to a time tuple.
+
+    Accounts for military timezones.
+    """
+    res = _parsedate_tz(data)
+    if not res:
+        return
+    if res[9] is None:
+        res[9] = 0
+    return tuple(res)
+
+def _parsedate_tz(data):
+    """Convert date to extended time tuple.
+
+    The last (additional) element is the time zone offset in seconds, except if
+    the timezone was specified as -0000.  In that case the last element is
+    None.  This indicates a UTC timestamp that explicitly declaims knowledge of
+    the source timezone, as opposed to a +0000 timestamp that indicates the
+    source timezone really was UTC.
+
+    """
+    if not data:
+        return
+    data = data.split()
+    # The FWS after the comma after the day-of-week is optional, so search and
+    # adjust for this.
+    if data[0].endswith(',') or data[0].lower() in _daynames:
+        # There's a dayname here. Skip it
+        del data[0]
+    else:
+        i = data[0].rfind(',')
+        if i >= 0:
+            data[0] = data[0][i+1:]
+    if len(data) == 3: # RFC 850 date, deprecated
+        stuff = data[0].split('-')
+        if len(stuff) == 3:
+            data = stuff + data[1:]
+    if len(data) == 4:
+        s = data[3]
+        i = s.find('+')
+        if i == -1:
+            i = s.find('-')
+        if i > 0:
+            data[3:] = [s[:i], s[i:]]
+        else:
+            data.append('') # Dummy tz
+    if len(data) < 5:
+        return None
+    data = data[:5]
+    [dd, mm, yy, tm, tz] = data
+    mm = mm.lower()
+    if mm not in _monthnames:
+        dd, mm = mm, dd.lower()
+        if mm not in _monthnames:
+            return None
+    mm = _monthnames.index(mm) + 1
+    if mm > 12:
+        mm -= 12
+    if dd[-1] == ',':
+        dd = dd[:-1]
+    i = yy.find(':')
+    if i > 0:
+        yy, tm = tm, yy
+    if yy[-1] == ',':
+        yy = yy[:-1]
+    if not yy[0].isdigit():
+        yy, tz = tz, yy
+    if tm[-1] == ',':
+        tm = tm[:-1]
+    tm = tm.split(':')
+    if len(tm) == 2:
+        [thh, tmm] = tm
+        tss = '0'
+    elif len(tm) == 3:
+        [thh, tmm, tss] = tm
+    elif len(tm) == 1 and '.' in tm[0]:
+        # Some non-compliant MUAs use '.' to separate time elements.
+        tm = tm[0].split('.')
+        if len(tm) == 2:
+            [thh, tmm] = tm
+            tss = 0
+        elif len(tm) == 3:
+            [thh, tmm, tss] = tm
+    else:
+        return None
+    try:
+        yy = int(yy)
+        dd = int(dd)
+        thh = int(thh)
+        tmm = int(tmm)
+        tss = int(tss)
+    except ValueError:
+        return None
+    # Check for a yy specified in two-digit format, then convert it to the
+    # appropriate four-digit format, according to the POSIX standard. RFC 822
+    # calls for a two-digit yy, but RFC 2822 (which obsoletes RFC 822)
+    # mandates a 4-digit yy. For more information, see the documentation for
+    # the time module.
+    if yy < 100:
+        # The year is between 1969 and 1999 (inclusive).
+        if yy > 68:
+            yy += 1900
+        # The year is between 2000 and 2068 (inclusive).
+        else:
+            yy += 2000
+    tzoffset = None
+    tz = tz.upper()
+    if tz in _timezones:
+        tzoffset = _timezones[tz]
+    else:
+        try:
+            tzoffset = int(tz)
+        except ValueError:
+            pass
+        if tzoffset==0 and tz.startswith('-'):
+            tzoffset = None
+    # Convert a timezone offset into seconds ; -0500 -> -18000
+    if tzoffset:
+        if tzoffset < 0:
+            tzsign = -1
+            tzoffset = -tzoffset
+        else:
+            tzsign = 1
+        tzoffset = tzsign * ( (tzoffset//100)*3600 + (tzoffset % 100)*60)
+    # Daylight Saving Time flag is set to -1, since DST is unknown.
+    return [yy, mm, dd, thh, tmm, tss, 0, 1, -1, tzoffset]
+
+
+def parsedate(data):
+    """Convert a time string to a time tuple."""
+    t = parsedate_tz(data)
+    if isinstance(t, tuple):
+        return t[:9]
+    else:
+        return t
+
+
+def mktime_tz(data):
+    """Turn a 10-tuple as returned by parsedate_tz() into a POSIX timestamp."""
+    if data[9] is None:
+        # No zone info, so localtime is better assumption than GMT
+        return time.mktime(data[:8] + (-1,))
+    else:
+        t = calendar.timegm(data)
+        return t - data[9]
+
+
+def quote(str):
+    """Prepare string to be used in a quoted string.
+
+    Turns backslash and double quote characters into quoted pairs.  These
+    are the only characters that need to be quoted inside a quoted string.
+    Does not add the surrounding double quotes.
+    """
+    return str.replace('\\', '\\\\').replace('"', '\\"')
+
+
+class AddrlistClass:
+    """Address parser class by Ben Escoto.
+
+    To understand what this class does, it helps to have a copy of RFC 2822 in
+    front of you.
+
+    Note: this class interface is deprecated and may be removed in the future.
+    Use email.utils.AddressList instead.
+    """
+
+    def __init__(self, field):
+        """Initialize a new instance.
+
+        `field' is an unparsed address header field, containing
+        one or more addresses.
+        """
+        self.specials = '()<>@,:;.\"[]'
+        self.pos = 0
+        self.LWS = ' \t'
+        self.CR = '\r\n'
+        self.FWS = self.LWS + self.CR
+        self.atomends = self.specials + self.LWS + self.CR
+        # Note that RFC 2822 now specifies `.' as obs-phrase, meaning that it
+        # is obsolete syntax.  RFC 2822 requires that we recognize obsolete
+        # syntax, so allow dots in phrases.
+        self.phraseends = self.atomends.replace('.', '')
+        self.field = field
+        self.commentlist = []
+
+    def gotonext(self):
+        """Skip white space and extract comments."""
+        wslist = []
+        while self.pos < len(self.field):
+            if self.field[self.pos] in self.LWS + '\n\r':
+                if self.field[self.pos] not in '\n\r':
+                    wslist.append(self.field[self.pos])
+                self.pos += 1
+            elif self.field[self.pos] == '(':
+                self.commentlist.append(self.getcomment())
+            else:
+                break
+        return EMPTYSTRING.join(wslist)
+
+    def getaddrlist(self):
+        """Parse all addresses.
+
+        Returns a list containing all of the addresses.
+        """
+        result = []
+        while self.pos < len(self.field):
+            ad = self.getaddress()
+            if ad:
+                result += ad
+            else:
+                result.append(('', ''))
+        return result
+
+    def getaddress(self):
+        """Parse the next address."""
+        self.commentlist = []
+        self.gotonext()
+
+        oldpos = self.pos
+        oldcl = self.commentlist
+        plist = self.getphraselist()
+
+        self.gotonext()
+        returnlist = []
+
+        if self.pos >= len(self.field):
+            # Bad email address technically, no domain.
+            if plist:
+                returnlist = [(SPACE.join(self.commentlist), plist[0])]
+
+        elif self.field[self.pos] in '.@':
+            # email address is just an addrspec
+            # this isn't very efficient since we start over
+            self.pos = oldpos
+            self.commentlist = oldcl
+            addrspec = self.getaddrspec()
+            returnlist = [(SPACE.join(self.commentlist), addrspec)]
+
+        elif self.field[self.pos] == ':':
+            # address is a group
+            returnlist = []
+
+            fieldlen = len(self.field)
+            self.pos += 1
+            while self.pos < len(self.field):
+                self.gotonext()
+                if self.pos < fieldlen and self.field[self.pos] == ';':
+                    self.pos += 1
+                    break
+                returnlist = returnlist + self.getaddress()
+
+        elif self.field[self.pos] == '<':
+            # Address is a phrase then a route addr
+            routeaddr = self.getrouteaddr()
+
+            if self.commentlist:
+                returnlist = [(SPACE.join(plist) + ' (' +
+                               ' '.join(self.commentlist) + ')', routeaddr)]
+            else:
+                returnlist = [(SPACE.join(plist), routeaddr)]
+
+        else:
+            if plist:
+                returnlist = [(SPACE.join(self.commentlist), plist[0])]
+            elif self.field[self.pos] in self.specials:
+                self.pos += 1
+
+        self.gotonext()
+        if self.pos < len(self.field) and self.field[self.pos] == ',':
+            self.pos += 1
+        return returnlist
+
+    def getrouteaddr(self):
+        """Parse a route address (Return-path value).
+
+        This method just skips all the route stuff and returns the addrspec.
+        """
+        if self.field[self.pos] != '<':
+            return
+
+        expectroute = False
+        self.pos += 1
+        self.gotonext()
+        adlist = ''
+        while self.pos < len(self.field):
+            if expectroute:
+                self.getdomain()
+                expectroute = False
+            elif self.field[self.pos] == '>':
+                self.pos += 1
+                break
+            elif self.field[self.pos] == '@':
+                self.pos += 1
+                expectroute = True
+            elif self.field[self.pos] == ':':
+                self.pos += 1
+            else:
+                adlist = self.getaddrspec()
+                self.pos += 1
+                break
+            self.gotonext()
+
+        return adlist
+
+    def getaddrspec(self):
+        """Parse an RFC 2822 addr-spec."""
+        aslist = []
+
+        self.gotonext()
+        while self.pos < len(self.field):
+            preserve_ws = True
+            if self.field[self.pos] == '.':
+                if aslist and not aslist[-1].strip():
+                    aslist.pop()
+                aslist.append('.')
+                self.pos += 1
+                preserve_ws = False
+            elif self.field[self.pos] == '"':
+                aslist.append('"%s"' % quote(self.getquote()))
+            elif self.field[self.pos] in self.atomends:
+                if aslist and not aslist[-1].strip():
+                    aslist.pop()
+                break
+            else:
+                aslist.append(self.getatom())
+            ws = self.gotonext()
+            if preserve_ws and ws:
+                aslist.append(ws)
+
+        if self.pos >= len(self.field) or self.field[self.pos] != '@':
+            return EMPTYSTRING.join(aslist)
+
+        aslist.append('@')
+        self.pos += 1
+        self.gotonext()
+        return EMPTYSTRING.join(aslist) + self.getdomain()
+
+    def getdomain(self):
+        """Get the complete domain name from an address."""
+        sdlist = []
+        while self.pos < len(self.field):
+            if self.field[self.pos] in self.LWS:
+                self.pos += 1
+            elif self.field[self.pos] == '(':
+                self.commentlist.append(self.getcomment())
+            elif self.field[self.pos] == '[':
+                sdlist.append(self.getdomainliteral())
+            elif self.field[self.pos] == '.':
+                self.pos += 1
+                sdlist.append('.')
+            elif self.field[self.pos] in self.atomends:
+                break
+            else:
+                sdlist.append(self.getatom())
+        return EMPTYSTRING.join(sdlist)
+
+    def getdelimited(self, beginchar, endchars, allowcomments=True):
+        """Parse a header fragment delimited by special characters.
+
+        `beginchar' is the start character for the fragment.
+        If self is not looking at an instance of `beginchar' then
+        getdelimited returns the empty string.
+
+        `endchars' is a sequence of allowable end-delimiting characters.
+        Parsing stops when one of these is encountered.
+
+        If `allowcomments' is non-zero, embedded RFC 2822 comments are allowed
+        within the parsed fragment.
+        """
+        if self.field[self.pos] != beginchar:
+            return ''
+
+        slist = ['']
+        quote = False
+        self.pos += 1
+        while self.pos < len(self.field):
+            if quote:
+                slist.append(self.field[self.pos])
+                quote = False
+            elif self.field[self.pos] in endchars:
+                self.pos += 1
+                break
+            elif allowcomments and self.field[self.pos] == '(':
+                slist.append(self.getcomment())
+                continue        # have already advanced pos from getcomment
+            elif self.field[self.pos] == '\\':
+                quote = True
+            else:
+                slist.append(self.field[self.pos])
+            self.pos += 1
+
+        return EMPTYSTRING.join(slist)
+
+    def getquote(self):
+        """Get a quote-delimited fragment from self's field."""
+        return self.getdelimited('"', '"\r', False)
+
+    def getcomment(self):
+        """Get a parenthesis-delimited fragment from self's field."""
+        return self.getdelimited('(', ')\r', True)
+
+    def getdomainliteral(self):
+        """Parse an RFC 2822 domain-literal."""
+        return '[%s]' % self.getdelimited('[', ']\r', False)
+
+    def getatom(self, atomends=None):
+        """Parse an RFC 2822 atom.
+
+        Optional atomends specifies a different set of end token delimiters
+        (the default is to use self.atomends).  This is used e.g. in
+        getphraselist() since phrase endings must not include the `.' (which
+        is legal in phrases)."""
+        atomlist = ['']
+        if atomends is None:
+            atomends = self.atomends
+
+        while self.pos < len(self.field):
+            if self.field[self.pos] in atomends:
+                break
+            else:
+                atomlist.append(self.field[self.pos])
+            self.pos += 1
+
+        return EMPTYSTRING.join(atomlist)
+
+    def getphraselist(self):
+        """Parse a sequence of RFC 2822 phrases.
+
+        A phrase is a sequence of words, which are in turn either RFC 2822
+        atoms or quoted-strings.  Phrases are canonicalized by squeezing all
+        runs of continuous whitespace into one space.
+        """
+        plist = []
+
+        while self.pos < len(self.field):
+            if self.field[self.pos] in self.FWS:
+                self.pos += 1
+            elif self.field[self.pos] == '"':
+                plist.append(self.getquote())
+            elif self.field[self.pos] == '(':
+                self.commentlist.append(self.getcomment())
+            elif self.field[self.pos] in self.phraseends:
+                break
+            else:
+                plist.append(self.getatom(self.phraseends))
+
+        return plist
+
+class AddressList(AddrlistClass):
+    """An AddressList encapsulates a list of parsed RFC 2822 addresses."""
+    def __init__(self, field):
+        AddrlistClass.__init__(self, field)
+        if field:
+            self.addresslist = self.getaddrlist()
+        else:
+            self.addresslist = []
+
+    def __len__(self):
+        return len(self.addresslist)
+
+    def __add__(self, other):
+        # Set union
+        newaddr = AddressList(None)
+        newaddr.addresslist = self.addresslist[:]
+        for x in other.addresslist:
+            if not x in self.addresslist:
+                newaddr.addresslist.append(x)
+        return newaddr
+
+    def __iadd__(self, other):
+        # Set union, in-place
+        for x in other.addresslist:
+            if not x in self.addresslist:
+                self.addresslist.append(x)
+        return self
+
+    def __sub__(self, other):
+        # Set difference
+        newaddr = AddressList(None)
+        for x in self.addresslist:
+            if not x in other.addresslist:
+                newaddr.addresslist.append(x)
+        return newaddr
+
+    def __isub__(self, other):
+        # Set difference, in-place
+        for x in other.addresslist:
+            if x in self.addresslist:
+                self.addresslist.remove(x)
+        return self
+
+    def __getitem__(self, index):
+        # Make indexing, slices, and 'in' work
+        return self.addresslist[index]
diff --git a/Lib/email/_policybase.py b/Lib/email/_policybase.py
new file mode 100644
index 000000000..df4649676
--- /dev/null
+++ b/Lib/email/_policybase.py
@@ -0,0 +1,370 @@
+"""Policy framework for the email package.
+
+Allows fine grained feature control of how the package parses and emits data.
+"""
+
+import abc
+from email import header
+from email import charset as _charset
+from email.utils import _has_surrogates
+
+__all__ = [
+    'Policy',
+    'Compat32',
+    'compat32',
+    ]
+
+
+class _PolicyBase:
+
+    """Policy Object basic framework.
+
+    This class is useless unless subclassed.  A subclass should define
+    class attributes with defaults for any values that are to be
+    managed by the Policy object.  The constructor will then allow
+    non-default values to be set for these attributes at instance
+    creation time.  The instance will be callable, taking these same
+    attributes keyword arguments, and returning a new instance
+    identical to the called instance except for those values changed
+    by the keyword arguments.  Instances may be added, yielding new
+    instances with any non-default values from the right hand
+    operand overriding those in the left hand operand.  That is,
+
+        A + B == A(<non-default values of B>)
+
+    The repr of an instance can be used to reconstruct the object
+    if and only if the repr of the values can be used to reconstruct
+    those values.
+
+    """
+
+    def __init__(self, **kw):
+        """Create new Policy, possibly overriding some defaults.
+
+        See class docstring for a list of overridable attributes.
+
+        """
+        for name, value in kw.items():
+            if hasattr(self, name):
+                super(_PolicyBase,self).__setattr__(name, value)
+            else:
+                raise TypeError(
+                    "{!r} is an invalid keyword argument for {}".format(
+                        name, self.__class__.__name__))
+
+    def __repr__(self):
+        args = [ "{}={!r}".format(name, value)
+                 for name, value in self.__dict__.items() ]
+        return "{}({})".format(self.__class__.__name__, ', '.join(args))
+
+    def clone(self, **kw):
+        """Return a new instance with specified attributes changed.
+
+        The new instance has the same attribute values as the current object,
+        except for the changes passed in as keyword arguments.
+
+        """
+        newpolicy = self.__class__.__new__(self.__class__)
+        for attr, value in self.__dict__.items():
+            object.__setattr__(newpolicy, attr, value)
+        for attr, value in kw.items():
+            if not hasattr(self, attr):
+                raise TypeError(
+                    "{!r} is an invalid keyword argument for {}".format(
+                        attr, self.__class__.__name__))
+            object.__setattr__(newpolicy, attr, value)
+        return newpolicy
+
+    def __setattr__(self, name, value):
+        if hasattr(self, name):
+            msg = "{!r} object attribute {!r} is read-only"
+        else:
+            msg = "{!r} object has no attribute {!r}"
+        raise AttributeError(msg.format(self.__class__.__name__, name))
+
+    def __add__(self, other):
+        """Non-default values from right operand override those from left.
+
+        The object returned is a new instance of the subclass.
+
+        """
+        return self.clone(**other.__dict__)
+
+
+def _append_doc(doc, added_doc):
+    doc = doc.rsplit('\n', 1)[0]
+    added_doc = added_doc.split('\n', 1)[1]
+    return doc + '\n' + added_doc
+
+def _extend_docstrings(cls):
+    if cls.__doc__ and cls.__doc__.startswith('+'):
+        cls.__doc__ = _append_doc(cls.__bases__[0].__doc__, cls.__doc__)
+    for name, attr in cls.__dict__.items():
+        if attr.__doc__ and attr.__doc__.startswith('+'):
+            for c in (c for base in cls.__bases__ for c in base.mro()):
+                doc = getattr(getattr(c, name), '__doc__')
+                if doc:
+                    attr.__doc__ = _append_doc(doc, attr.__doc__)
+                    break
+    return cls
+
+
+class Policy(_PolicyBase, metaclass=abc.ABCMeta):
+
+    r"""Controls for how messages are interpreted and formatted.
+
+    Most of the classes and many of the methods in the email package accept
+    Policy objects as parameters.  A Policy object contains a set of values and
+    functions that control how input is interpreted and how output is rendered.
+    For example, the parameter 'raise_on_defect' controls whether or not an RFC
+    violation results in an error being raised or not, while 'max_line_length'
+    controls the maximum length of output lines when a Message is serialized.
+
+    Any valid attribute may be overridden when a Policy is created by passing
+    it as a keyword argument to the constructor.  Policy objects are immutable,
+    but a new Policy object can be created with only certain values changed by
+    calling the Policy instance with keyword arguments.  Policy objects can
+    also be added, producing a new Policy object in which the non-default
+    attributes set in the right hand operand overwrite those specified in the
+    left operand.
+
+    Settable attributes:
+
+    raise_on_defect     -- If true, then defects should be raised as errors.
+                           Default: False.
+
+    linesep             -- string containing the value to use as separation
+                           between output lines.  Default '\n'.
+
+    cte_type            -- Type of allowed content transfer encodings
+
+                           7bit  -- ASCII only
+                           8bit  -- Content-Transfer-Encoding: 8bit is allowed
+
+                           Default: 8bit.  Also controls the disposition of
+                           (RFC invalid) binary data in headers; see the
+                           documentation of the binary_fold method.
+
+    max_line_length     -- maximum length of lines, excluding 'linesep',
+                           during serialization.  None or 0 means no line
+                           wrapping is done.  Default is 78.
+
+    mangle_from_        -- a flag that, when True escapes From_ lines in the
+                           body of the message by putting a `>' in front of
+                           them. This is used when the message is being
+                           serialized by a generator. Default: True.
+
+    message_factory     -- the class to use to create new message objects.
+                           If the value is None, the default is Message.
+
+    """
+
+    raise_on_defect = False
+    linesep = '\n'
+    cte_type = '8bit'
+    max_line_length = 78
+    mangle_from_ = False
+    message_factory = None
+
+    def handle_defect(self, obj, defect):
+        """Based on policy, either raise defect or call register_defect.
+
+            handle_defect(obj, defect)
+
+        defect should be a Defect subclass, but in any case must be an
+        Exception subclass.  obj is the object on which the defect should be
+        registered if it is not raised.  If the raise_on_defect is True, the
+        defect is raised as an error, otherwise the object and the defect are
+        passed to register_defect.
+
+        This method is intended to be called by parsers that discover defects.
+        The email package parsers always call it with Defect instances.
+
+        """
+        if self.raise_on_defect:
+            raise defect
+        self.register_defect(obj, defect)
+
+    def register_defect(self, obj, defect):
+        """Record 'defect' on 'obj'.
+
+        Called by handle_defect if raise_on_defect is False.  This method is
+        part of the Policy API so that Policy subclasses can implement custom
+        defect handling.  The default implementation calls the append method of
+        the defects attribute of obj.  The objects used by the email package by
+        default that get passed to this method will always have a defects
+        attribute with an append method.
+
+        """
+        obj.defects.append(defect)
+
+    def header_max_count(self, name):
+        """Return the maximum allowed number of headers named 'name'.
+
+        Called when a header is added to a Message object.  If the returned
+        value is not 0 or None, and there are already a number of headers with
+        the name 'name' equal to the value returned, a ValueError is raised.
+
+        Because the default behavior of Message's __setitem__ is to append the
+        value to the list of headers, it is easy to create duplicate headers
+        without realizing it.  This method allows certain headers to be limited
+        in the number of instances of that header that may be added to a
+        Message programmatically.  (The limit is not observed by the parser,
+        which will faithfully produce as many headers as exist in the message
+        being parsed.)
+
+        The default implementation returns None for all header names.
+        """
+        return None
+
+    @abc.abstractmethod
+    def header_source_parse(self, sourcelines):
+        """Given a list of linesep terminated strings constituting the lines of
+        a single header, return the (name, value) tuple that should be stored
+        in the model.  The input lines should retain their terminating linesep
+        characters.  The lines passed in by the email package may contain
+        surrogateescaped binary data.
+        """
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def header_store_parse(self, name, value):
+        """Given the header name and the value provided by the application
+        program, return the (name, value) that should be stored in the model.
+        """
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def header_fetch_parse(self, name, value):
+        """Given the header name and the value from the model, return the value
+        to be returned to the application program that is requesting that
+        header.  The value passed in by the email package may contain
+        surrogateescaped binary data if the lines were parsed by a BytesParser.
+        The returned value should not contain any surrogateescaped data.
+
+        """
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def fold(self, name, value):
+        """Given the header name and the value from the model, return a string
+        containing linesep characters that implement the folding of the header
+        according to the policy controls.  The value passed in by the email
+        package may contain surrogateescaped binary data if the lines were
+        parsed by a BytesParser.  The returned value should not contain any
+        surrogateescaped data.
+
+        """
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def fold_binary(self, name, value):
+        """Given the header name and the value from the model, return binary
+        data containing linesep characters that implement the folding of the
+        header according to the policy controls.  The value passed in by the
+        email package may contain surrogateescaped binary data.
+
+        """
+        raise NotImplementedError
+
+
+@_extend_docstrings
+class Compat32(Policy):
+
+    """+
+    This particular policy is the backward compatibility Policy.  It
+    replicates the behavior of the email package version 5.1.
+    """
+
+    mangle_from_ = True
+
+    def _sanitize_header(self, name, value):
+        # If the header value contains surrogates, return a Header using
+        # the unknown-8bit charset to encode the bytes as encoded words.
+        if not isinstance(value, str):
+            # Assume it is already a header object
+            return value
+        if _has_surrogates(value):
+            return header.Header(value, charset=_charset.UNKNOWN8BIT,
+                                 header_name=name)
+        else:
+            return value
+
+    def header_source_parse(self, sourcelines):
+        """+
+        The name is parsed as everything up to the ':' and returned unmodified.
+        The value is determined by stripping leading whitespace off the
+        remainder of the first line, joining all subsequent lines together, and
+        stripping any trailing carriage return or linefeed characters.
+
+        """
+        name, value = sourcelines[0].split(':', 1)
+        value = value.lstrip(' \t') + ''.join(sourcelines[1:])
+        return (name, value.rstrip('\r\n'))
+
+    def header_store_parse(self, name, value):
+        """+
+        The name and value are returned unmodified.
+        """
+        return (name, value)
+
+    def header_fetch_parse(self, name, value):
+        """+
+        If the value contains binary data, it is converted into a Header object
+        using the unknown-8bit charset.  Otherwise it is returned unmodified.
+        """
+        return self._sanitize_header(name, value)
+
+    def fold(self, name, value):
+        """+
+        Headers are folded using the Header folding algorithm, which preserves
+        existing line breaks in the value, and wraps each resulting line to the
+        max_line_length.  Non-ASCII binary data are CTE encoded using the
+        unknown-8bit charset.
+
+        """
+        return self._fold(name, value, sanitize=True)
+
+    def fold_binary(self, name, value):
+        """+
+        Headers are folded using the Header folding algorithm, which preserves
+        existing line breaks in the value, and wraps each resulting line to the
+        max_line_length.  If cte_type is 7bit, non-ascii binary data is CTE
+        encoded using the unknown-8bit charset.  Otherwise the original source
+        header is used, with its existing line breaks and/or binary data.
+
+        """
+        folded = self._fold(name, value, sanitize=self.cte_type=='7bit')
+        return folded.encode('ascii', 'surrogateescape')
+
+    def _fold(self, name, value, sanitize):
+        parts = []
+        parts.append('%s: ' % name)
+        if isinstance(value, str):
+            if _has_surrogates(value):
+                if sanitize:
+                    h = header.Header(value,
+                                      charset=_charset.UNKNOWN8BIT,
+                                      header_name=name)
+                else:
+                    # If we have raw 8bit data in a byte string, we have no idea
+                    # what the encoding is.  There is no safe way to split this
+                    # string.  If it's ascii-subset, then we could do a normal
+                    # ascii split, but if it's multibyte then we could break the
+                    # string.  There's no way to know so the least harm seems to
+                    # be to not split the string and risk it being too long.
+                    parts.append(value)
+                    h = None
+            else:
+                h = header.Header(value, header_name=name)
+        else:
+            # Assume it is a Header-like object.
+            h = value
+        if h is not None:
+            parts.append(h.encode(linesep=self.linesep,
+                                  maxlinelen=self.max_line_length))
+        parts.append(self.linesep)
+        return ''.join(parts)
+
+
+compat32 = Compat32()
diff --git a/Lib/email/architecture.rst b/Lib/email/architecture.rst
new file mode 100644
index 000000000..78572ae63
--- /dev/null
+++ b/Lib/email/architecture.rst
@@ -0,0 +1,216 @@
+:mod:`email` Package Architecture
+=================================
+
+Overview
+--------
+
+The email package consists of three major components:
+
+    Model
+        An object structure that represents an email message, and provides an
+        API for creating, querying, and modifying a message.
+
+    Parser
+        Takes a sequence of characters or bytes and produces a model of the
+        email message represented by those characters or bytes.
+
+    Generator
+        Takes a model and turns it into a sequence of characters or bytes.  The
+        sequence can either be intended for human consumption (a printable
+        unicode string) or bytes suitable for transmission over the wire.  In
+        the latter case all data is properly encoded using the content transfer
+        encodings specified by the relevant RFCs.
+
+Conceptually the package is organized around the model.  The model provides both
+"external" APIs intended for use by application programs using the library,
+and "internal" APIs intended for use by the Parser and Generator components.
+This division is intentionally a bit fuzzy; the API described by this
+documentation is all a public, stable API.  This allows for an application
+with special needs to implement its own parser and/or generator.
+
+In addition to the three major functional components, there is a third key
+component to the architecture:
+
+    Policy
+        An object that specifies various behavioral settings and carries
+        implementations of various behavior-controlling methods.
+
+The Policy framework provides a simple and convenient way to control the
+behavior of the library, making it possible for the library to be used in a
+very flexible fashion while leveraging the common code required to parse,
+represent, and generate message-like objects.  For example, in addition to the
+default :rfc:`5322` email message policy, we also have a policy that manages
+HTTP headers in a fashion compliant with :rfc:`2616`.  Individual policy
+controls, such as the maximum line length produced by the generator, can also
+be controlled individually to meet specialized application requirements.
+
+
+The Model
+---------
+
+The message model is implemented by the :class:`~email.message.Message` class.
+The model divides a message into the two fundamental parts discussed by the
+RFC: the header section and the body.  The `Message` object acts as a
+pseudo-dictionary of named headers.  Its dictionary interface provides
+convenient access to individual headers by name.  However, all headers are kept
+internally in an ordered list, so that the information about the order of the
+headers in the original message is preserved.
+
+The `Message` object also has a `payload` that holds the body.  A `payload` can
+be one of two things: data, or a list of `Message` objects.  The latter is used
+to represent a multipart MIME message.  Lists can be nested arbitrarily deeply
+in order to represent the message, with all terminal leaves having non-list
+data payloads.
+
+
+Message Lifecycle
+-----------------
+
+The general lifecyle of a message is:
+
+    Creation
+        A `Message` object can be created by a Parser, or it can be
+        instantiated as an empty message by an application.
+
+    Manipulation
+        The application may examine one or more headers, and/or the
+        payload, and it may modify one or more headers and/or
+        the payload.  This may be done on the top level `Message`
+        object, or on any sub-object.
+
+    Finalization
+        The Model is converted into a unicode or binary stream,
+        or the model is discarded.
+
+
+
+Header Policy Control During Lifecycle
+--------------------------------------
+
+One of the major controls exerted by the Policy is the management of headers
+during the `Message` lifecycle.  Most applications don't need to be aware of
+this.
+
+A header enters the model in one of two ways: via a Parser, or by being set to
+a specific value by an application program after the Model already exists.
+Similarly, a header exits the model in one of two ways: by being serialized by
+a Generator, or by being retrieved from a Model by an application program.  The
+Policy object provides hooks for all four of these pathways.
+
+The model storage for headers is a list of (name, value) tuples.
+
+The Parser identifies headers during parsing, and passes them to the
+:meth:`~email.policy.Policy.header_source_parse` method of the Policy.  The
+result of that method is the (name, value) tuple to be stored in the model.
+
+When an application program supplies a header value (for example, through the
+`Message` object `__setitem__` interface), the name and the value are passed to
+the :meth:`~email.policy.Policy.header_store_parse` method of the Policy, which
+returns the (name, value) tuple to be stored in the model.
+
+When an application program retrieves a header (through any of the dict or list
+interfaces of `Message`), the name and value are passed to the
+:meth:`~email.policy.Policy.header_fetch_parse` method of the Policy to
+obtain the value returned to the application.
+
+When a Generator requests a header during serialization, the name and value are
+passed to the :meth:`~email.policy.Policy.fold` method of the Policy, which
+returns a string containing line breaks in the appropriate places.  The
+:meth:`~email.policy.Policy.cte_type` Policy control determines whether or
+not Content Transfer Encoding is performed on the data in the header.  There is
+also a :meth:`~email.policy.Policy.binary_fold` method for use by generators
+that produce binary output, which returns the folded header as binary data,
+possibly folded at different places than the corresponding string would be.
+
+
+Handling Binary Data
+--------------------
+
+In an ideal world all message data would conform to the RFCs, meaning that the
+parser could decode the message into the idealized unicode message that the
+sender originally wrote.  In the real world, the email package must also be
+able to deal with badly formatted messages, including messages containing
+non-ASCII characters that either have no indicated character set or are not
+valid characters in the indicated character set.
+
+Since email messages are *primarily* text data, and operations on message data
+are primarily text operations (except for binary payloads of course), the model
+stores all text data as unicode strings.  Un-decodable binary inside text
+data is handled by using the `surrogateescape` error handler of the ASCII
+codec.  As with the binary filenames the error handler was introduced to
+handle, this allows the email package to "carry" the binary data received
+during parsing along until the output stage, at which time it is regenerated
+in its original form.
+
+This carried binary data is almost entirely an implementation detail.  The one
+place where it is visible in the API is in the "internal" API.  A Parser must
+do the `surrogateescape` encoding of binary input data, and pass that data to
+the appropriate Policy method.  The "internal" interface used by the Generator
+to access header values preserves the `surrogateescaped` bytes.  All other
+interfaces convert the binary data either back into bytes or into a safe form
+(losing information in some cases).
+
+
+Backward Compatibility
+----------------------
+
+The :class:`~email.policy.Policy.Compat32` Policy provides backward
+compatibility with version 5.1 of the email package.  It does this via the
+following implementation of the four+1 Policy methods described above:
+
+header_source_parse
+    Splits the first line on the colon to obtain the name, discards any spaces
+    after the colon, and joins the remainder of the line with all of the
+    remaining lines, preserving the linesep characters to obtain the value.
+    Trailing carriage return and/or linefeed characters are stripped from the
+    resulting value string.
+
+header_store_parse
+    Returns the name and value exactly as received from the application.
+
+header_fetch_parse
+    If the value contains any `surrogateescaped` binary data, return the value
+    as a :class:`~email.header.Header` object, using the character set
+    `unknown-8bit`.  Otherwise just returns the value.
+
+fold
+    Uses :class:`~email.header.Header`'s folding to fold headers in the
+    same way the email5.1 generator did.
+
+binary_fold
+    Same as fold, but encodes to 'ascii'.
+
+
+New Algorithm
+-------------
+
+header_source_parse
+    Same as legacy behavior.
+
+header_store_parse
+    Same as legacy behavior.
+
+header_fetch_parse
+    If the value is already a header object, returns it.  Otherwise, parses the
+    value using the new parser, and returns the resulting object as the value.
+    `surrogateescaped` bytes get turned into unicode unknown character code
+    points.
+
+fold
+    Uses the new header folding algorithm, respecting the policy settings.
+    surrogateescaped bytes are encoded using the ``unknown-8bit`` charset for
+    ``cte_type=7bit`` or ``8bit``.  Returns a string.
+
+    At some point there will also be a ``cte_type=unicode``, and for that
+    policy fold will serialize the idealized unicode message with RFC-like
+    folding, converting any surrogateescaped bytes into the unicode
+    unknown character glyph.
+
+binary_fold
+    Uses the new header folding algorithm, respecting the policy settings.
+    surrogateescaped bytes are encoded using the `unknown-8bit` charset for
+    ``cte_type=7bit``, and get turned back into bytes for ``cte_type=8bit``.
+    Returns bytes.
+
+    At some point there will also be a ``cte_type=unicode``, and for that
+    policy binary_fold will serialize the message according to :rfc:``5335``.
diff --git a/Lib/email/base64mime.py b/Lib/email/base64mime.py
new file mode 100644
index 000000000..17f0818f6
--- /dev/null
+++ b/Lib/email/base64mime.py
@@ -0,0 +1,119 @@
+# Copyright (C) 2002-2007 Python Software Foundation
+# Author: Ben Gertzfield
+# Contact: email-sig@python.org
+
+"""Base64 content transfer encoding per RFCs 2045-2047.
+
+This module handles the content transfer encoding method defined in RFC 2045
+to encode arbitrary 8-bit data using the three 8-bit bytes in four 7-bit
+characters encoding known as Base64.
+
+It is used in the MIME standards for email to attach images, audio, and text
+using some 8-bit character sets to messages.
+
+This module provides an interface to encode and decode both headers and bodies
+with Base64 encoding.
+
+RFC 2045 defines a method for including character set information in an
+`encoded-word' in a header.  This method is commonly used for 8-bit real names
+in To:, From:, Cc:, etc. fields, as well as Subject: lines.
+
+This module does not do the line wrapping or end-of-line character conversion
+necessary for proper internationalized headers; it only does dumb encoding and
+decoding.  To deal with the various line wrapping issues, use the email.header
+module.
+"""
+
+__all__ = [
+    'body_decode',
+    'body_encode',
+    'decode',
+    'decodestring',
+    'header_encode',
+    'header_length',
+    ]
+
+
+from base64 import b64encode
+from binascii import b2a_base64, a2b_base64
+
+CRLF = '\r\n'
+NL = '\n'
+EMPTYSTRING = ''
+
+# See also Charset.py
+MISC_LEN = 7
+
+
+
+# Helpers
+def header_length(bytearray):
+    """Return the length of s when it is encoded with base64."""
+    groups_of_3, leftover = divmod(len(bytearray), 3)
+    # 4 bytes out for each 3 bytes (or nonzero fraction thereof) in.
+    n = groups_of_3 * 4
+    if leftover:
+        n += 4
+    return n
+
+
+
+def header_encode(header_bytes, charset='iso-8859-1'):
+    """Encode a single header line with Base64 encoding in a given charset.
+
+    charset names the character set to use to encode the header.  It defaults
+    to iso-8859-1.  Base64 encoding is defined in RFC 2045.
+    """
+    if not header_bytes:
+        return ""
+    if isinstance(header_bytes, str):
+        header_bytes = header_bytes.encode(charset)
+    encoded = b64encode(header_bytes).decode("ascii")
+    return '=?%s?b?%s?=' % (charset, encoded)
+
+
+
+def body_encode(s, maxlinelen=76, eol=NL):
+    r"""Encode a string with base64.
+
+    Each line will be wrapped at, at most, maxlinelen characters (defaults to
+    76 characters).
+
+    Each line of encoded text will end with eol, which defaults to "\n".  Set
+    this to "\r\n" if you will be using the result of this function directly
+    in an email.
+    """
+    if not s:
+        return s
+
+    encvec = []
+    max_unencoded = maxlinelen * 3 // 4
+    for i in range(0, len(s), max_unencoded):
+        # BAW: should encode() inherit b2a_base64()'s dubious behavior in
+        # adding a newline to the encoded string?
+        enc = b2a_base64(s[i:i + max_unencoded]).decode("ascii")
+        if enc.endswith(NL) and eol != NL:
+            enc = enc[:-1] + eol
+        encvec.append(enc)
+    return EMPTYSTRING.join(encvec)
+
+
+
+def decode(string):
+    """Decode a raw base64 string, returning a bytes object.
+
+    This function does not parse a full MIME header value encoded with
+    base64 (like =?iso-8859-1?b?bmloISBuaWgh?=) -- please use the high
+    level email.header class for that functionality.
+    """
+    if not string:
+        return bytes()
+    elif isinstance(string, str):
+        return a2b_base64(string.encode('raw-unicode-escape'))
+    else:
+        return a2b_base64(string)
+
+
+# For convenience and backwards compatibility w/ standard base64 module
+body_decode = decode
+decodestring = decode
diff --git a/Lib/email/charset.py b/Lib/email/charset.py
new file mode 100644
index 000000000..ee564040c
--- /dev/null
+++ b/Lib/email/charset.py
@@ -0,0 +1,406 @@
+# Copyright (C) 2001-2007 Python Software Foundation
+# Author: Ben Gertzfield, Barry Warsaw
+# Contact: email-sig@python.org
+
+__all__ = [
+    'Charset',
+    'add_alias',
+    'add_charset',
+    'add_codec',
+    ]
+
+from functools import partial
+
+import email.base64mime
+import email.quoprimime
+
+from email import errors
+from email.encoders import encode_7or8bit
+
+
+
+# Flags for types of header encodings
+QP          = 1 # Quoted-Printable
+BASE64      = 2 # Base64
+SHORTEST    = 3 # the shorter of QP and base64, but only for headers
+
+# In "=?charset?q?hello_world?=", the =?, ?q?, and ?= add up to 7
+RFC2047_CHROME_LEN = 7
+
+DEFAULT_CHARSET = 'us-ascii'
+UNKNOWN8BIT = 'unknown-8bit'
+EMPTYSTRING = ''
+
+
+
+# Defaults
+CHARSETS = {
+    # input        header enc  body enc output conv
+    'iso-8859-1':  (QP,        QP,      None),
+    'iso-8859-2':  (QP,        QP,      None),
+    'iso-8859-3':  (QP,        QP,      None),
+    'iso-8859-4':  (QP,        QP,      None),
+    # iso-8859-5 is Cyrillic, and not especially used
+    # iso-8859-6 is Arabic, also not particularly used
+    # iso-8859-7 is Greek, QP will not make it readable
+    # iso-8859-8 is Hebrew, QP will not make it readable
+    'iso-8859-9':  (QP,        QP,      None),
+    'iso-8859-10': (QP,        QP,      None),
+    # iso-8859-11 is Thai, QP will not make it readable
+    'iso-8859-13': (QP,        QP,      None),
+    'iso-8859-14': (QP,        QP,      None),
+    'iso-8859-15': (QP,        QP,      None),
+    'iso-8859-16': (QP,        QP,      None),
+    'windows-1252':(QP,        QP,      None),
+    'viscii':      (QP,        QP,      None),
+    'us-ascii':    (None,      None,    None),
+    'big5':        (BASE64,    BASE64,  None),
+    'gb2312':      (BASE64,    BASE64,  None),
+    'euc-jp':      (BASE64,    None,    'iso-2022-jp'),
+    'shift_jis':   (BASE64,    None,    'iso-2022-jp'),
+    'iso-2022-jp': (BASE64,    None,    None),
+    'koi8-r':      (BASE64,    BASE64,  None),
+    'utf-8':       (SHORTEST,  BASE64, 'utf-8'),
+    }
+
+# Aliases for other commonly-used names for character sets.  Map
+# them to the real ones used in email.
+ALIASES = {
+    'latin_1': 'iso-8859-1',
+    'latin-1': 'iso-8859-1',
+    'latin_2': 'iso-8859-2',
+    'latin-2': 'iso-8859-2',
+    'latin_3': 'iso-8859-3',
+    'latin-3': 'iso-8859-3',
+    'latin_4': 'iso-8859-4',
+    'latin-4': 'iso-8859-4',
+    'latin_5': 'iso-8859-9',
+    'latin-5': 'iso-8859-9',
+    'latin_6': 'iso-8859-10',
+    'latin-6': 'iso-8859-10',
+    'latin_7': 'iso-8859-13',
+    'latin-7': 'iso-8859-13',
+    'latin_8': 'iso-8859-14',
+    'latin-8': 'iso-8859-14',
+    'latin_9': 'iso-8859-15',
+    'latin-9': 'iso-8859-15',
+    'latin_10':'iso-8859-16',
+    'latin-10':'iso-8859-16',
+    'cp949':   'ks_c_5601-1987',
+    'euc_jp':  'euc-jp',
+    'euc_kr':  'euc-kr',
+    'ascii':   'us-ascii',
+    }
+
+
+# Map charsets to their Unicode codec strings.
+CODEC_MAP = {
+    'gb2312':      'eucgb2312_cn',
+    'big5':        'big5_tw',
+    # Hack: We don't want *any* conversion for stuff marked us-ascii, as all
+    # sorts of garbage might be sent to us in the guise of 7-bit us-ascii.
+    # Let that stuff pass through without conversion to/from Unicode.
+    'us-ascii':    None,
+    }
+
+
+
+# Convenience functions for extending the above mappings
+def add_charset(charset, header_enc=None, body_enc=None, output_charset=None):
+    """Add character set properties to the global registry.
+
+    charset is the input character set, and must be the canonical name of a
+    character set.
+
+    Optional header_enc and body_enc is either Charset.QP for
+    quoted-printable, Charset.BASE64 for base64 encoding, Charset.SHORTEST for
+    the shortest of qp or base64 encoding, or None for no encoding.  SHORTEST
+    is only valid for header_enc.  It describes how message headers and
+    message bodies in the input charset are to be encoded.  Default is no
+    encoding.
+
+    Optional output_charset is the character set that the output should be
+    in.  Conversions will proceed from input charset, to Unicode, to the
+    output charset when the method Charset.convert() is called.  The default
+    is to output in the same character set as the input.
+
+    Both input_charset and output_charset must have Unicode codec entries in
+    the module's charset-to-codec mapping; use add_codec(charset, codecname)
+    to add codecs the module does not know about.  See the codecs module's
+    documentation for more information.
+    """
+    if body_enc == SHORTEST:
+        raise ValueError('SHORTEST not allowed for body_enc')
+    CHARSETS[charset] = (header_enc, body_enc, output_charset)
+
+
+def add_alias(alias, canonical):
+    """Add a character set alias.
+
+    alias is the alias name, e.g. latin-1
+    canonical is the character set's canonical name, e.g. iso-8859-1
+    """
+    ALIASES[alias] = canonical
+
+
+def add_codec(charset, codecname):
+    """Add a codec that map characters in the given charset to/from Unicode.
+
+    charset is the canonical name of a character set.  codecname is the name
+    of a Python codec, as appropriate for the second argument to the unicode()
+    built-in, or to the encode() method of a Unicode string.
+    """
+    CODEC_MAP[charset] = codecname
+
+
+
+# Convenience function for encoding strings, taking into account
+# that they might be unknown-8bit (ie: have surrogate-escaped bytes)
+def _encode(string, codec):
+    if codec == UNKNOWN8BIT:
+        return string.encode('ascii', 'surrogateescape')
+    else:
+        return string.encode(codec)
+
+
+
+class Charset:
+    """Map character sets to their email properties.
+
+    This class provides information about the requirements imposed on email
+    for a specific character set.  It also provides convenience routines for
+    converting between character sets, given the availability of the
+    applicable codecs.  Given a character set, it will do its best to provide
+    information on how to use that character set in an email in an
+    RFC-compliant way.
+
+    Certain character sets must be encoded with quoted-printable or base64
+    when used in email headers or bodies.  Certain character sets must be
+    converted outright, and are not allowed in email.  Instances of this
+    module expose the following information about a character set:
+
+    input_charset: The initial character set specified.  Common aliases
+                   are converted to their `official' email names (e.g. latin_1
+                   is converted to iso-8859-1).  Defaults to 7-bit us-ascii.
+
+    header_encoding: If the character set must be encoded before it can be
+                     used in an email header, this attribute will be set to
+                     Charset.QP (for quoted-printable), Charset.BASE64 (for
+                     base64 encoding), or Charset.SHORTEST for the shortest of
+                     QP or BASE64 encoding.  Otherwise, it will be None.
+
+    body_encoding: Same as header_encoding, but describes the encoding for the
+                   mail message's body, which indeed may be different than the
+                   header encoding.  Charset.SHORTEST is not allowed for
+                   body_encoding.
+
+    output_charset: Some character sets must be converted before they can be
+                    used in email headers or bodies.  If the input_charset is
+                    one of them, this attribute will contain the name of the
+                    charset output will be converted to.  Otherwise, it will
+                    be None.
+
+    input_codec: The name of the Python codec used to convert the
+                 input_charset to Unicode.  If no conversion codec is
+                 necessary, this attribute will be None.
+
+    output_codec: The name of the Python codec used to convert Unicode
+                  to the output_charset.  If no conversion codec is necessary,
+                  this attribute will have the same value as the input_codec.
+    """
+    def __init__(self, input_charset=DEFAULT_CHARSET):
+        # RFC 2046, $4.1.2 says charsets are not case sensitive.  We coerce to
+        # unicode because its .lower() is locale insensitive.  If the argument
+        # is already a unicode, we leave it at that, but ensure that the
+        # charset is ASCII, as the standard (RFC XXX) requires.
+        try:
+            if isinstance(input_charset, str):
+                input_charset.encode('ascii')
+            else:
+                input_charset = str(input_charset, 'ascii')
+        except UnicodeError:
+            raise errors.CharsetError(input_charset)
+        input_charset = input_charset.lower()
+        # Set the input charset after filtering through the aliases
+        self.input_charset = ALIASES.get(input_charset, input_charset)
+        # We can try to guess which encoding and conversion to use by the
+        # charset_map dictionary.  Try that first, but let the user override
+        # it.
+        henc, benc, conv = CHARSETS.get(self.input_charset,
+                                        (SHORTEST, BASE64, None))
+        if not conv:
+            conv = self.input_charset
+        # Set the attributes, allowing the arguments to override the default.
+        self.header_encoding = henc
+        self.body_encoding = benc
+        self.output_charset = ALIASES.get(conv, conv)
+        # Now set the codecs.  If one isn't defined for input_charset,
+        # guess and try a Unicode codec with the same name as input_codec.
+        self.input_codec = CODEC_MAP.get(self.input_charset,
+                                         self.input_charset)
+        self.output_codec = CODEC_MAP.get(self.output_charset,
+                                          self.output_charset)
+
+    def __str__(self):
+        return self.input_charset.lower()
+
+    __repr__ = __str__
+
+    def __eq__(self, other):
+        return str(self) == str(other).lower()
+
+    def get_body_encoding(self):
+        """Return the content-transfer-encoding used for body encoding.
+
+        This is either the string `quoted-printable' or `base64' depending on
+        the encoding used, or it is a function in which case you should call
+        the function with a single argument, the Message object being
+        encoded.  The function should then set the Content-Transfer-Encoding
+        header itself to whatever is appropriate.
+
+        Returns "quoted-printable" if self.body_encoding is QP.
+        Returns "base64" if self.body_encoding is BASE64.
+        Returns conversion function otherwise.
+        """
+        assert self.body_encoding != SHORTEST
+        if self.body_encoding == QP:
+            return 'quoted-printable'
+        elif self.body_encoding == BASE64:
+            return 'base64'
+        else:
+            return encode_7or8bit
+
+    def get_output_charset(self):
+        """Return the output character set.
+
+        This is self.output_charset if that is not None, otherwise it is
+        self.input_charset.
+        """
+        return self.output_charset or self.input_charset
+
+    def header_encode(self, string):
+        """Header-encode a string by converting it first to bytes.
+
+        The type of encoding (base64 or quoted-printable) will be based on
+        this charset's `header_encoding`.
+
+        :param string: A unicode string for the header.  It must be possible
+            to encode this string to bytes using the character set's
+            output codec.
+        :return: The encoded string, with RFC 2047 chrome.
+        """
+        codec = self.output_codec or 'us-ascii'
+        header_bytes = _encode(string, codec)
+        # 7bit/8bit encodings return the string unchanged (modulo conversions)
+        encoder_module = self._get_encoder(header_bytes)
+        if encoder_module is None:
+            return string
+        return encoder_module.header_encode(header_bytes, codec)
+
+    def header_encode_lines(self, string, maxlengths):
+        """Header-encode a string by converting it first to bytes.
+
+        This is similar to `header_encode()` except that the string is fit
+        into maximum line lengths as given by the argument.
+
+        :param string: A unicode string for the header.  It must be possible
+            to encode this string to bytes using the character set's
+            output codec.
+        :param maxlengths: Maximum line length iterator.  Each element
+            returned from this iterator will provide the next maximum line
+            length.  This parameter is used as an argument to built-in next()
+            and should never be exhausted.  The maximum line lengths should
+            not count the RFC 2047 chrome.  These line lengths are only a
+            hint; the splitter does the best it can.
+        :return: Lines of encoded strings, each with RFC 2047 chrome.
+        """
+        # See which encoding we should use.
+        codec = self.output_codec or 'us-ascii'
+        header_bytes = _encode(string, codec)
+        encoder_module = self._get_encoder(header_bytes)
+        encoder = partial(encoder_module.header_encode, charset=codec)
+        # Calculate the number of characters that the RFC 2047 chrome will
+        # contribute to each line.
+        charset = self.get_output_charset()
+        extra = len(charset) + RFC2047_CHROME_LEN
+        # Now comes the hard part.  We must encode bytes but we can't split on
+        # bytes because some character sets are variable length and each
+        # encoded word must stand on its own.  So the problem is you have to
+        # encode to bytes to figure out this word's length, but you must split
+        # on characters.  This causes two problems: first, we don't know how
+        # many octets a specific substring of unicode characters will get
+        # encoded to, and second, we don't know how many ASCII characters
+        # those octets will get encoded to.  Unless we try it.  Which seems
+        # inefficient.  In the interest of being correct rather than fast (and
+        # in the hope that there will be few encoded headers in any such
+        # message), brute force it. :(
+        lines = []
+        current_line = []
+        maxlen = next(maxlengths) - extra
+        for character in string:
+            current_line.append(character)
+            this_line = EMPTYSTRING.join(current_line)
+            length = encoder_module.header_length(_encode(this_line, charset))
+            if length > maxlen:
+                # This last character doesn't fit so pop it off.
+                current_line.pop()
+                # Does nothing fit on the first line?
+                if not lines and not current_line:
+                    lines.append(None)
+                else:
+                    separator = (' ' if lines else '')
+                    joined_line = EMPTYSTRING.join(current_line)
+                    header_bytes = _encode(joined_line, codec)
+                    lines.append(encoder(header_bytes))
+                current_line = [character]
+                maxlen = next(maxlengths) - extra
+        joined_line = EMPTYSTRING.join(current_line)
+        header_bytes = _encode(joined_line, codec)
+        lines.append(encoder(header_bytes))
+        return lines
+
+    def _get_encoder(self, header_bytes):
+        if self.header_encoding == BASE64:
+            return email.base64mime
+        elif self.header_encoding == QP:
+            return email.quoprimime
+        elif self.header_encoding == SHORTEST:
+            len64 = email.base64mime.header_length(header_bytes)
+            lenqp = email.quoprimime.header_length(header_bytes)
+            if len64 < lenqp:
+                return email.base64mime
+            else:
+                return email.quoprimime
+        else:
+            return None
+
+    def body_encode(self, string):
+        """Body-encode a string by converting it first to bytes.
+
+        The type of encoding (base64 or quoted-printable) will be based on
+        self.body_encoding.  If body_encoding is None, we assume the
+        output charset is a 7bit encoding, so re-encoding the decoded
+        string using the ascii codec produces the correct string version
+        of the content.
+        """
+        if not string:
+            return string
+        if self.body_encoding is BASE64:
+            if isinstance(string, str):
+                string = string.encode(self.output_charset)
+            return email.base64mime.body_encode(string)
+        elif self.body_encoding is QP:
+            # quopromime.body_encode takes a string, but operates on it as if
+            # it were a list of byte codes.  For a (minimal) history on why
+            # this is so, see changeset 0cf700464177.  To correctly encode a
+            # character set, then, we must turn it into pseudo bytes via the
+            # latin1 charset, which will encode any byte as a single code point
+            # between 0 and 255, which is what body_encode is expecting.
+            if isinstance(string, str):
+                string = string.encode(self.output_charset)
+            string = string.decode('latin1')
+            return email.quoprimime.body_encode(string)
+        else:
+            if isinstance(string, str):
+                string = string.encode(self.output_charset).decode('ascii')
+            return string
diff --git a/Lib/email/contentmanager.py b/Lib/email/contentmanager.py
new file mode 100644
index 000000000..b904ded94
--- /dev/null
+++ b/Lib/email/contentmanager.py
@@ -0,0 +1,250 @@
+import binascii
+import email.charset
+import email.message
+import email.errors
+from email import quoprimime
+
+class ContentManager:
+
+    def __init__(self):
+        self.get_handlers = {}
+        self.set_handlers = {}
+
+    def add_get_handler(self, key, handler):
+        self.get_handlers[key] = handler
+
+    def get_content(self, msg, *args, **kw):
+        content_type = msg.get_content_type()
+        if content_type in self.get_handlers:
+            return self.get_handlers[content_type](msg, *args, **kw)
+        maintype = msg.get_content_maintype()
+        if maintype in self.get_handlers:
+            return self.get_handlers[maintype](msg, *args, **kw)
+        if '' in self.get_handlers:
+            return self.get_handlers[''](msg, *args, **kw)
+        raise KeyError(content_type)
+
+    def add_set_handler(self, typekey, handler):
+        self.set_handlers[typekey] = handler
+
+    def set_content(self, msg, obj, *args, **kw):
+        if msg.get_content_maintype() == 'multipart':
+            # XXX: is this error a good idea or not?  We can remove it later,
+            # but we can't add it later, so do it for now.
+            raise TypeError("set_content not valid on multipart")
+        handler = self._find_set_handler(msg, obj)
+        msg.clear_content()
+        handler(msg, obj, *args, **kw)
+
+    def _find_set_handler(self, msg, obj):
+        full_path_for_error = None
+        for typ in type(obj).__mro__:
+            if typ in self.set_handlers:
+                return self.set_handlers[typ]
+            qname = typ.__qualname__
+            modname = getattr(typ, '__module__', '')
+            full_path = '.'.join((modname, qname)) if modname else qname
+            if full_path_for_error is None:
+                full_path_for_error = full_path
+            if full_path in self.set_handlers:
+                return self.set_handlers[full_path]
+            if qname in self.set_handlers:
+                return self.set_handlers[qname]
+            name = typ.__name__
+            if name in self.set_handlers:
+                return self.set_handlers[name]
+        if None in self.set_handlers:
+            return self.set_handlers[None]
+        raise KeyError(full_path_for_error)
+
+
+raw_data_manager = ContentManager()
+
+
+def get_text_content(msg, errors='replace'):
+    content = msg.get_payload(decode=True)
+    charset = msg.get_param('charset', 'ASCII')
+    return content.decode(charset, errors=errors)
+raw_data_manager.add_get_handler('text', get_text_content)
+
+
+def get_non_text_content(msg):
+    return msg.get_payload(decode=True)
+for maintype in 'audio image video application'.split():
+    raw_data_manager.add_get_handler(maintype, get_non_text_content)
+
+
+def get_message_content(msg):
+    return msg.get_payload(0)
+for subtype in 'rfc822 external-body'.split():
+    raw_data_manager.add_get_handler('message/'+subtype, get_message_content)
+
+
+def get_and_fixup_unknown_message_content(msg):
+    # If we don't understand a message subtype, we are supposed to treat it as
+    # if it were application/octet-stream, per
+    # tools.ietf.org/html/rfc2046#section-5.2.4.  Feedparser doesn't do that,
+    # so do our best to fix things up.  Note that it is *not* appropriate to
+    # model message/partial content as Message objects, so they are handled
+    # here as well.  (How to reassemble them is out of scope for this comment :)
+    return bytes(msg.get_payload(0))
+raw_data_manager.add_get_handler('message',
+                                 get_and_fixup_unknown_message_content)
+
+
+def _prepare_set(msg, maintype, subtype, headers):
+    msg['Content-Type'] = '/'.join((maintype, subtype))
+    if headers:
+        if not hasattr(headers[0], 'name'):
+            mp = msg.policy
+            headers = [mp.header_factory(*mp.header_source_parse([header]))
+                       for header in headers]
+        try:
+            for header in headers:
+                if header.defects:
+                    raise header.defects[0]
+                msg[header.name] = header
+        except email.errors.HeaderDefect as exc:
+            raise ValueError("Invalid header: {}".format(
+                                header.fold(policy=msg.policy))) from exc
+
+
+def _finalize_set(msg, disposition, filename, cid, params):
+    if disposition is None and filename is not None:
+        disposition = 'attachment'
+    if disposition is not None:
+        msg['Content-Disposition'] = disposition
+    if filename is not None:
+        msg.set_param('filename',
+                      filename,
+                      header='Content-Disposition',
+                      replace=True)
+    if cid is not None:
+        msg['Content-ID'] = cid
+    if params is not None:
+        for key, value in params.items():
+            msg.set_param(key, value)
+
+
+# XXX: This is a cleaned-up version of base64mime.body_encode (including a bug
+# fix in the calculation of unencoded_bytes_per_line).  It would be nice to
+# drop both this and quoprimime.body_encode in favor of enhanced binascii
+# routines that accepted a max_line_length parameter.
+def _encode_base64(data, max_line_length):
+    encoded_lines = []
+    unencoded_bytes_per_line = max_line_length // 4 * 3
+    for i in range(0, len(data), unencoded_bytes_per_line):
+        thisline = data[i:i+unencoded_bytes_per_line]
+        encoded_lines.append(binascii.b2a_base64(thisline).decode('ascii'))
+    return ''.join(encoded_lines)
+
+
+def _encode_text(string, charset, cte, policy):
+    lines = string.encode(charset).splitlines()
+    linesep = policy.linesep.encode('ascii')
+    def embedded_body(lines): return linesep.join(lines) + linesep
+    def normal_body(lines): return b'\n'.join(lines) + b'\n'
+    if cte==None:
+        # Use heuristics to decide on the "best" encoding.
+        try:
+            return '7bit', normal_body(lines).decode('ascii')
+        except UnicodeDecodeError:
+            pass
+        if (policy.cte_type == '8bit' and
+                max(len(x) for x in lines) <= policy.max_line_length):
+            return '8bit', normal_body(lines).decode('ascii', 'surrogateescape')
+        sniff = embedded_body(lines[:10])
+        sniff_qp = quoprimime.body_encode(sniff.decode('latin-1'),
+                                          policy.max_line_length)
+        sniff_base64 = binascii.b2a_base64(sniff)
+        # This is a little unfair to qp; it includes lineseps, base64 doesn't.
+        if len(sniff_qp) > len(sniff_base64):
+            cte = 'base64'
+        else:
+            cte = 'quoted-printable'
+            if len(lines) <= 10:
+                return cte, sniff_qp
+    if cte == '7bit':
+        data = normal_body(lines).decode('ascii')
+    elif cte == '8bit':
+        data = normal_body(lines).decode('ascii', 'surrogateescape')
+    elif cte == 'quoted-printable':
+        data = quoprimime.body_encode(normal_body(lines).decode('latin-1'),
+                                      policy.max_line_length)
+    elif cte == 'base64':
+        data = _encode_base64(embedded_body(lines), policy.max_line_length)
+    else:
+        raise ValueError("Unknown content transfer encoding {}".format(cte))
+    return cte, data
+
+
+def set_text_content(msg, string, subtype="plain", charset='utf-8', cte=None,
+                     disposition=None, filename=None, cid=None,
+                     params=None, headers=None):
+    _prepare_set(msg, 'text', subtype, headers)
+    cte, payload = _encode_text(string, charset, cte, msg.policy)
+    msg.set_payload(payload)
+    msg.set_param('charset',
+                  email.charset.ALIASES.get(charset, charset),
+                  replace=True)
+    msg['Content-Transfer-Encoding'] = cte
+    _finalize_set(msg, disposition, filename, cid, params)
+raw_data_manager.add_set_handler(str, set_text_content)
+
+
+def set_message_content(msg, message, subtype="rfc822", cte=None,
+                       disposition=None, filename=None, cid=None,
+                       params=None, headers=None):
+    if subtype == 'partial':
+        raise ValueError("message/partial is not supported for Message objects")
+    if subtype == 'rfc822':
+        if cte not in (None, '7bit', '8bit', 'binary'):
+            # http://tools.ietf.org/html/rfc2046#section-5.2.1 mandate.
+            raise ValueError(
+                "message/rfc822 parts do not support cte={}".format(cte))
+        # 8bit will get coerced on serialization if policy.cte_type='7bit'.  We
+        # may end up claiming 8bit when it isn't needed, but the only negative
+        # result of that should be a gateway that needs to coerce to 7bit
+        # having to look through the whole embedded message to discover whether
+        # or not it actually has to do anything.
+        cte = '8bit' if cte is None else cte
+    elif subtype == 'external-body':
+        if cte not in (None, '7bit'):
+            # http://tools.ietf.org/html/rfc2046#section-5.2.3 mandate.
+            raise ValueError(
+                "message/external-body parts do not support cte={}".format(cte))
+        cte = '7bit'
+    elif cte is None:
+        # http://tools.ietf.org/html/rfc2046#section-5.2.4 says all future
+        # subtypes should be restricted to 7bit, so assume that.
+        cte = '7bit'
+    _prepare_set(msg, 'message', subtype, headers)
+    msg.set_payload([message])
+    msg['Content-Transfer-Encoding'] = cte
+    _finalize_set(msg, disposition, filename, cid, params)
+raw_data_manager.add_set_handler(email.message.Message, set_message_content)
+
+
+def set_bytes_content(msg, data, maintype, subtype, cte='base64',
+                     disposition=None, filename=None, cid=None,
+                     params=None, headers=None):
+    _prepare_set(msg, maintype, subtype, headers)
+    if cte == 'base64':
+        data = _encode_base64(data, max_line_length=msg.policy.max_line_length)
+    elif cte == 'quoted-printable':
+        # XXX: quoprimime.body_encode won't encode newline characters in data,
+        # so we can't use it.  This means max_line_length is ignored.  Another
+        # bug to fix later.  (Note: encoders.quopri is broken on line ends.)
+        data = binascii.b2a_qp(data, istext=False, header=False, quotetabs=True)
+        data = data.decode('ascii')
+    elif cte == '7bit':
+        # Make sure it really is only ASCII.  The early warning here seems
+        # worth the overhead...if you care write your own content manager :).
+        data.encode('ascii')
+    elif cte in ('8bit', 'binary'):
+        data = data.decode('ascii', 'surrogateescape')
+    msg.set_payload(data)
+    msg['Content-Transfer-Encoding'] = cte
+    _finalize_set(msg, disposition, filename, cid, params)
+for typ in (bytes, bytearray, memoryview):
+    raw_data_manager.add_set_handler(typ, set_bytes_content)
diff --git a/Lib/email/encoders.py b/Lib/email/encoders.py
new file mode 100644
index 000000000..0a66acb62
--- /dev/null
+++ b/Lib/email/encoders.py
@@ -0,0 +1,69 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Encodings and related functions."""
+
+__all__ = [
+    'encode_7or8bit',
+    'encode_base64',
+    'encode_noop',
+    'encode_quopri',
+    ]
+
+
+from base64 import encodebytes as _bencode
+from quopri import encodestring as _encodestring
+
+
+
+def _qencode(s):
+    enc = _encodestring(s, quotetabs=True)
+    # Must encode spaces, which quopri.encodestring() doesn't do
+    return enc.replace(b' ', b'=20')
+
+
+def encode_base64(msg):
+    """Encode the message's payload in Base64.
+
+    Also, add an appropriate Content-Transfer-Encoding header.
+    """
+    orig = msg.get_payload(decode=True)
+    encdata = str(_bencode(orig), 'ascii')
+    msg.set_payload(encdata)
+    msg['Content-Transfer-Encoding'] = 'base64'
+
+
+
+def encode_quopri(msg):
+    """Encode the message's payload in quoted-printable.
+
+    Also, add an appropriate Content-Transfer-Encoding header.
+    """
+    orig = msg.get_payload(decode=True)
+    encdata = _qencode(orig)
+    msg.set_payload(encdata)
+    msg['Content-Transfer-Encoding'] = 'quoted-printable'
+
+
+
+def encode_7or8bit(msg):
+    """Set the Content-Transfer-Encoding header to 7bit or 8bit."""
+    orig = msg.get_payload(decode=True)
+    if orig is None:
+        # There's no payload.  For backwards compatibility we use 7bit
+        msg['Content-Transfer-Encoding'] = '7bit'
+        return
+    # We play a trick to make this go fast.  If decoding from ASCII succeeds,
+    # we know the data must be 7bit, otherwise treat it as 8bit.
+    try:
+        orig.decode('ascii')
+    except UnicodeError:
+        msg['Content-Transfer-Encoding'] = '8bit'
+    else:
+        msg['Content-Transfer-Encoding'] = '7bit'
+
+
+
+def encode_noop(msg):
+    """Do nothing."""
diff --git a/Lib/email/errors.py b/Lib/email/errors.py
new file mode 100644
index 000000000..791239fa6
--- /dev/null
+++ b/Lib/email/errors.py
@@ -0,0 +1,107 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""email package exception classes."""
+
+
+class MessageError(Exception):
+    """Base class for errors in the email package."""
+
+
+class MessageParseError(MessageError):
+    """Base class for message parsing errors."""
+
+
+class HeaderParseError(MessageParseError):
+    """Error while parsing headers."""
+
+
+class BoundaryError(MessageParseError):
+    """Couldn't find terminating boundary."""
+
+
+class MultipartConversionError(MessageError, TypeError):
+    """Conversion to a multipart is prohibited."""
+
+
+class CharsetError(MessageError):
+    """An illegal charset was given."""
+
+
+# These are parsing defects which the parser was able to work around.
+class MessageDefect(ValueError):
+    """Base class for a message defect."""
+
+    def __init__(self, line=None):
+        if line is not None:
+            super().__init__(line)
+        self.line = line
+
+class NoBoundaryInMultipartDefect(MessageDefect):
+    """A message claimed to be a multipart but had no boundary parameter."""
+
+class StartBoundaryNotFoundDefect(MessageDefect):
+    """The claimed start boundary was never found."""
+
+class CloseBoundaryNotFoundDefect(MessageDefect):
+    """A start boundary was found, but not the corresponding close boundary."""
+
+class FirstHeaderLineIsContinuationDefect(MessageDefect):
+    """A message had a continuation line as its first header line."""
+
+class MisplacedEnvelopeHeaderDefect(MessageDefect):
+    """A 'Unix-from' header was found in the middle of a header block."""
+
+class MissingHeaderBodySeparatorDefect(MessageDefect):
+    """Found line with no leading whitespace and no colon before blank line."""
+# XXX: backward compatibility, just in case (it was never emitted).
+MalformedHeaderDefect = MissingHeaderBodySeparatorDefect
+
+class MultipartInvariantViolationDefect(MessageDefect):
+    """A message claimed to be a multipart but no subparts were found."""
+
+class InvalidMultipartContentTransferEncodingDefect(MessageDefect):
+    """An invalid content transfer encoding was set on the multipart itself."""
+
+class UndecodableBytesDefect(MessageDefect):
+    """Header contained bytes that could not be decoded"""
+
+class InvalidBase64PaddingDefect(MessageDefect):
+    """base64 encoded sequence had an incorrect length"""
+
+class InvalidBase64CharactersDefect(MessageDefect):
+    """base64 encoded sequence had characters not in base64 alphabet"""
+
+# These errors are specific to header parsing.
+
+class HeaderDefect(MessageDefect):
+    """Base class for a header defect."""
+
+    def __init__(self, *args, **kw):
+        super().__init__(*args, **kw)
+
+class InvalidHeaderDefect(HeaderDefect):
+    """Header is not valid, message gives details."""
+
+class HeaderMissingRequiredValue(HeaderDefect):
+    """A header that must have a value had none"""
+
+class NonPrintableDefect(HeaderDefect):
+    """ASCII characters outside the ascii-printable range found"""
+
+    def __init__(self, non_printables):
+        super().__init__(non_printables)
+        self.non_printables = non_printables
+
+    def __str__(self):
+        return ("the following ASCII non-printables found in header: "
+            "{}".format(self.non_printables))
+
+class ObsoleteHeaderDefect(HeaderDefect):
+    """Header uses syntax declared obsolete by RFC 5322"""
+
+class NonASCIILocalPartDefect(HeaderDefect):
+    """local_part contains non-ASCII characters"""
+    # This defect only occurs during unicode parsing, not when
+    # parsing messages decoded from binary.
diff --git a/Lib/email/feedparser.py b/Lib/email/feedparser.py
new file mode 100644
index 000000000..7c07ca864
--- /dev/null
+++ b/Lib/email/feedparser.py
@@ -0,0 +1,536 @@
+# Copyright (C) 2004-2006 Python Software Foundation
+# Authors: Baxter, Wouters and Warsaw
+# Contact: email-sig@python.org
+
+"""FeedParser - An email feed parser.
+
+The feed parser implements an interface for incrementally parsing an email
+message, line by line.  This has advantages for certain applications, such as
+those reading email messages off a socket.
+
+FeedParser.feed() is the primary interface for pushing new data into the
+parser.  It returns when there's nothing more it can do with the available
+data.  When you have no more data to push into the parser, call .close().
+This completes the parsing and returns the root message object.
+
+The other advantage of this parser is that it will never raise a parsing
+exception.  Instead, when it finds something unexpected, it adds a 'defect' to
+the current message.  Defects are just instances that live on the message
+object's .defects attribute.
+"""
+
+__all__ = ['FeedParser', 'BytesFeedParser']
+
+import re
+
+from email import errors
+from email._policybase import compat32
+from collections import deque
+from io import StringIO
+
+NLCRE = re.compile(r'\r\n|\r|\n')
+NLCRE_bol = re.compile(r'(\r\n|\r|\n)')
+NLCRE_eol = re.compile(r'(\r\n|\r|\n)\Z')
+NLCRE_crack = re.compile(r'(\r\n|\r|\n)')
+# RFC 2822 $3.6.8 Optional fields.  ftext is %d33-57 / %d59-126, Any character
+# except controls, SP, and ":".
+headerRE = re.compile(r'^(From |[\041-\071\073-\176]*:|[\t ])')
+EMPTYSTRING = ''
+NL = '\n'
+
+NeedMoreData = object()
+
+
+
+class BufferedSubFile(object):
+    """A file-ish object that can have new data loaded into it.
+
+    You can also push and pop line-matching predicates onto a stack.  When the
+    current predicate matches the current line, a false EOF response
+    (i.e. empty string) is returned instead.  This lets the parser adhere to a
+    simple abstraction -- it parses until EOF closes the current message.
+    """
+    def __init__(self):
+        # Text stream of the last partial line pushed into this object.
+        # See issue 22233 for why this is a text stream and not a list.
+        self._partial = StringIO(newline='')
+        # A deque of full, pushed lines
+        self._lines = deque()
+        # The stack of false-EOF checking predicates.
+        self._eofstack = []
+        # A flag indicating whether the file has been closed or not.
+        self._closed = False
+
+    def push_eof_matcher(self, pred):
+        self._eofstack.append(pred)
+
+    def pop_eof_matcher(self):
+        return self._eofstack.pop()
+
+    def close(self):
+        # Don't forget any trailing partial line.
+        self._partial.seek(0)
+        self.pushlines(self._partial.readlines())
+        self._partial.seek(0)
+        self._partial.truncate()
+        self._closed = True
+
+    def readline(self):
+        if not self._lines:
+            if self._closed:
+                return ''
+            return NeedMoreData
+        # Pop the line off the stack and see if it matches the current
+        # false-EOF predicate.
+        line = self._lines.popleft()
+        # RFC 2046, section 5.1.2 requires us to recognize outer level
+        # boundaries at any level of inner nesting.  Do this, but be sure it's
+        # in the order of most to least nested.
+        for ateof in reversed(self._eofstack):
+            if ateof(line):
+                # We're at the false EOF.  But push the last line back first.
+                self._lines.appendleft(line)
+                return ''
+        return line
+
+    def unreadline(self, line):
+        # Let the consumer push a line back into the buffer.
+        assert line is not NeedMoreData
+        self._lines.appendleft(line)
+
+    def push(self, data):
+        """Push some new data into this object."""
+        self._partial.write(data)
+        if '\n' not in data and '\r' not in data:
+            # No new complete lines, wait for more.
+            return
+
+        # Crack into lines, preserving the linesep characters.
+        self._partial.seek(0)
+        parts = self._partial.readlines()
+        self._partial.seek(0)
+        self._partial.truncate()
+
+        # If the last element of the list does not end in a newline, then treat
+        # it as a partial line.  We only check for '\n' here because a line
+        # ending with '\r' might be a line that was split in the middle of a
+        # '\r\n' sequence (see bugs 1555570 and 1721862).
+        if not parts[-1].endswith('\n'):
+            self._partial.write(parts.pop())
+        self.pushlines(parts)
+
+    def pushlines(self, lines):
+        self._lines.extend(lines)
+
+    def __iter__(self):
+        return self
+
+    def __next__(self):
+        line = self.readline()
+        if line == '':
+            raise StopIteration
+        return line
+
+
+
+class FeedParser:
+    """A feed-style parser of email."""
+
+    def __init__(self, _factory=None, *, policy=compat32):
+        """_factory is called with no arguments to create a new message obj
+
+        The policy keyword specifies a policy object that controls a number of
+        aspects of the parser's operation.  The default policy maintains
+        backward compatibility.
+
+        """
+        self.policy = policy
+        self._old_style_factory = False
+        if _factory is None:
+            if policy.message_factory is None:
+                from email.message import Message
+                self._factory = Message
+            else:
+                self._factory = policy.message_factory
+        else:
+            self._factory = _factory
+            try:
+                _factory(policy=self.policy)
+            except TypeError:
+                # Assume this is an old-style factory
+                self._old_style_factory = True
+        self._input = BufferedSubFile()
+        self._msgstack = []
+        self._parse = self._parsegen().__next__
+        self._cur = None
+        self._last = None
+        self._headersonly = False
+
+    # Non-public interface for supporting Parser's headersonly flag
+    def _set_headersonly(self):
+        self._headersonly = True
+
+    def feed(self, data):
+        """Push more data into the parser."""
+        self._input.push(data)
+        self._call_parse()
+
+    def _call_parse(self):
+        try:
+            self._parse()
+        except StopIteration:
+            pass
+
+    def close(self):
+        """Parse all remaining data and return the root message object."""
+        self._input.close()
+        self._call_parse()
+        root = self._pop_message()
+        assert not self._msgstack
+        # Look for final set of defects
+        if root.get_content_maintype() == 'multipart' \
+               and not root.is_multipart():
+            defect = errors.MultipartInvariantViolationDefect()
+            self.policy.handle_defect(root, defect)
+        return root
+
+    def _new_message(self):
+        if self._old_style_factory:
+            msg = self._factory()
+        else:
+            msg = self._factory(policy=self.policy)
+        if self._cur and self._cur.get_content_type() == 'multipart/digest':
+            msg.set_default_type('message/rfc822')
+        if self._msgstack:
+            self._msgstack[-1].attach(msg)
+        self._msgstack.append(msg)
+        self._cur = msg
+        self._last = msg
+
+    def _pop_message(self):
+        retval = self._msgstack.pop()
+        if self._msgstack:
+            self._cur = self._msgstack[-1]
+        else:
+            self._cur = None
+        return retval
+
+    def _parsegen(self):
+        # Create a new message and start by parsing headers.
+        self._new_message()
+        headers = []
+        # Collect the headers, searching for a line that doesn't match the RFC
+        # 2822 header or continuation pattern (including an empty line).
+        for line in self._input:
+            if line is NeedMoreData:
+                yield NeedMoreData
+                continue
+            if not headerRE.match(line):
+                # If we saw the RFC defined header/body separator
+                # (i.e. newline), just throw it away. Otherwise the line is
+                # part of the body so push it back.
+                if not NLCRE.match(line):
+                    defect = errors.MissingHeaderBodySeparatorDefect()
+                    self.policy.handle_defect(self._cur, defect)
+                    self._input.unreadline(line)
+                break
+            headers.append(line)
+        # Done with the headers, so parse them and figure out what we're
+        # supposed to see in the body of the message.
+        self._parse_headers(headers)
+        # Headers-only parsing is a backwards compatibility hack, which was
+        # necessary in the older parser, which could raise errors.  All
+        # remaining lines in the input are thrown into the message body.
+        if self._headersonly:
+            lines = []
+            while True:
+                line = self._input.readline()
+                if line is NeedMoreData:
+                    yield NeedMoreData
+                    continue
+                if line == '':
+                    break
+                lines.append(line)
+            self._cur.set_payload(EMPTYSTRING.join(lines))
+            return
+        if self._cur.get_content_type() == 'message/delivery-status':
+            # message/delivery-status contains blocks of headers separated by
+            # a blank line.  We'll represent each header block as a separate
+            # nested message object, but the processing is a bit different
+            # than standard message/* types because there is no body for the
+            # nested messages.  A blank line separates the subparts.
+            while True:
+                self._input.push_eof_matcher(NLCRE.match)
+                for retval in self._parsegen():
+                    if retval is NeedMoreData:
+                        yield NeedMoreData
+                        continue
+                    break
+                msg = self._pop_message()
+                # We need to pop the EOF matcher in order to tell if we're at
+                # the end of the current file, not the end of the last block
+                # of message headers.
+                self._input.pop_eof_matcher()
+                # The input stream must be sitting at the newline or at the
+                # EOF.  We want to see if we're at the end of this subpart, so
+                # first consume the blank line, then test the next line to see
+                # if we're at this subpart's EOF.
+                while True:
+                    line = self._input.readline()
+                    if line is NeedMoreData:
+                        yield NeedMoreData
+                        continue
+                    break
+                while True:
+                    line = self._input.readline()
+                    if line is NeedMoreData:
+                        yield NeedMoreData
+                        continue
+                    break
+                if line == '':
+                    break
+                # Not at EOF so this is a line we're going to need.
+                self._input.unreadline(line)
+            return
+        if self._cur.get_content_maintype() == 'message':
+            # The message claims to be a message/* type, then what follows is
+            # another RFC 2822 message.
+            for retval in self._parsegen():
+                if retval is NeedMoreData:
+                    yield NeedMoreData
+                    continue
+                break
+            self._pop_message()
+            return
+        if self._cur.get_content_maintype() == 'multipart':
+            boundary = self._cur.get_boundary()
+            if boundary is None:
+                # The message /claims/ to be a multipart but it has not
+                # defined a boundary.  That's a problem which we'll handle by
+                # reading everything until the EOF and marking the message as
+                # defective.
+                defect = errors.NoBoundaryInMultipartDefect()
+                self.policy.handle_defect(self._cur, defect)
+                lines = []
+                for line in self._input:
+                    if line is NeedMoreData:
+                        yield NeedMoreData
+                        continue
+                    lines.append(line)
+                self._cur.set_payload(EMPTYSTRING.join(lines))
+                return
+            # Make sure a valid content type was specified per RFC 2045:6.4.
+            if (self._cur.get('content-transfer-encoding', '8bit').lower()
+                    not in ('7bit', '8bit', 'binary')):
+                defect = errors.InvalidMultipartContentTransferEncodingDefect()
+                self.policy.handle_defect(self._cur, defect)
+            # Create a line match predicate which matches the inter-part
+            # boundary as well as the end-of-multipart boundary.  Don't push
+            # this onto the input stream until we've scanned past the
+            # preamble.
+            separator = '--' + boundary
+            boundaryre = re.compile(
+                '(?P<sep>' + re.escape(separator) +
+                r')(?P<end>--)?(?P<ws>[ \t]*)(?P<linesep>\r\n|\r|\n)?$')
+            capturing_preamble = True
+            preamble = []
+            linesep = False
+            close_boundary_seen = False
+            while True:
+                line = self._input.readline()
+                if line is NeedMoreData:
+                    yield NeedMoreData
+                    continue
+                if line == '':
+                    break
+                mo = boundaryre.match(line)
+                if mo:
+                    # If we're looking at the end boundary, we're done with
+                    # this multipart.  If there was a newline at the end of
+                    # the closing boundary, then we need to initialize the
+                    # epilogue with the empty string (see below).
+                    if mo.group('end'):
+                        close_boundary_seen = True
+                        linesep = mo.group('linesep')
+                        break
+                    # We saw an inter-part boundary.  Were we in the preamble?
+                    if capturing_preamble:
+                        if preamble:
+                            # According to RFC 2046, the last newline belongs
+                            # to the boundary.
+                            lastline = preamble[-1]
+                            eolmo = NLCRE_eol.search(lastline)
+                            if eolmo:
+                                preamble[-1] = lastline[:-len(eolmo.group(0))]
+                            self._cur.preamble = EMPTYSTRING.join(preamble)
+                        capturing_preamble = False
+                        self._input.unreadline(line)
+                        continue
+                    # We saw a boundary separating two parts.  Consume any
+                    # multiple boundary lines that may be following.  Our
+                    # interpretation of RFC 2046 BNF grammar does not produce
+                    # body parts within such double boundaries.
+                    while True:
+                        line = self._input.readline()
+                        if line is NeedMoreData:
+                            yield NeedMoreData
+                            continue
+                        mo = boundaryre.match(line)
+                        if not mo:
+                            self._input.unreadline(line)
+                            break
+                    # Recurse to parse this subpart; the input stream points
+                    # at the subpart's first line.
+                    self._input.push_eof_matcher(boundaryre.match)
+                    for retval in self._parsegen():
+                        if retval is NeedMoreData:
+                            yield NeedMoreData
+                            continue
+                        break
+                    # Because of RFC 2046, the newline preceding the boundary
+                    # separator actually belongs to the boundary, not the
+                    # previous subpart's payload (or epilogue if the previous
+                    # part is a multipart).
+                    if self._last.get_content_maintype() == 'multipart':
+                        epilogue = self._last.epilogue
+                        if epilogue == '':
+                            self._last.epilogue = None
+                        elif epilogue is not None:
+                            mo = NLCRE_eol.search(epilogue)
+                            if mo:
+                                end = len(mo.group(0))
+                                self._last.epilogue = epilogue[:-end]
+                    else:
+                        payload = self._last._payload
+                        if isinstance(payload, str):
+                            mo = NLCRE_eol.search(payload)
+                            if mo:
+                                payload = payload[:-len(mo.group(0))]
+                                self._last._payload = payload
+                    self._input.pop_eof_matcher()
+                    self._pop_message()
+                    # Set the multipart up for newline cleansing, which will
+                    # happen if we're in a nested multipart.
+                    self._last = self._cur
+                else:
+                    # I think we must be in the preamble
+                    assert capturing_preamble
+                    preamble.append(line)
+            # We've seen either the EOF or the end boundary.  If we're still
+            # capturing the preamble, we never saw the start boundary.  Note
+            # that as a defect and store the captured text as the payload.
+            if capturing_preamble:
+                defect = errors.StartBoundaryNotFoundDefect()
+                self.policy.handle_defect(self._cur, defect)
+                self._cur.set_payload(EMPTYSTRING.join(preamble))
+                epilogue = []
+                for line in self._input:
+                    if line is NeedMoreData:
+                        yield NeedMoreData
+                        continue
+                self._cur.epilogue = EMPTYSTRING.join(epilogue)
+                return
+            # If we're not processing the preamble, then we might have seen
+            # EOF without seeing that end boundary...that is also a defect.
+            if not close_boundary_seen:
+                defect = errors.CloseBoundaryNotFoundDefect()
+                self.policy.handle_defect(self._cur, defect)
+                return
+            # Everything from here to the EOF is epilogue.  If the end boundary
+            # ended in a newline, we'll need to make sure the epilogue isn't
+            # None
+            if linesep:
+                epilogue = ['']
+            else:
+                epilogue = []
+            for line in self._input:
+                if line is NeedMoreData:
+                    yield NeedMoreData
+                    continue
+                epilogue.append(line)
+            # Any CRLF at the front of the epilogue is not technically part of
+            # the epilogue.  Also, watch out for an empty string epilogue,
+            # which means a single newline.
+            if epilogue:
+                firstline = epilogue[0]
+                bolmo = NLCRE_bol.match(firstline)
+                if bolmo:
+                    epilogue[0] = firstline[len(bolmo.group(0)):]
+            self._cur.epilogue = EMPTYSTRING.join(epilogue)
+            return
+        # Otherwise, it's some non-multipart type, so the entire rest of the
+        # file contents becomes the payload.
+        lines = []
+        for line in self._input:
+            if line is NeedMoreData:
+                yield NeedMoreData
+                continue
+            lines.append(line)
+        self._cur.set_payload(EMPTYSTRING.join(lines))
+
+    def _parse_headers(self, lines):
+        # Passed a list of lines that make up the headers for the current msg
+        lastheader = ''
+        lastvalue = []
+        for lineno, line in enumerate(lines):
+            # Check for continuation
+            if line[0] in ' \t':
+                if not lastheader:
+                    # The first line of the headers was a continuation.  This
+                    # is illegal, so let's note the defect, store the illegal
+                    # line, and ignore it for purposes of headers.
+                    defect = errors.FirstHeaderLineIsContinuationDefect(line)
+                    self.policy.handle_defect(self._cur, defect)
+                    continue
+                lastvalue.append(line)
+                continue
+            if lastheader:
+                self._cur.set_raw(*self.policy.header_source_parse(lastvalue))
+                lastheader, lastvalue = '', []
+            # Check for envelope header, i.e. unix-from
+            if line.startswith('From '):
+                if lineno == 0:
+                    # Strip off the trailing newline
+                    mo = NLCRE_eol.search(line)
+                    if mo:
+                        line = line[:-len(mo.group(0))]
+                    self._cur.set_unixfrom(line)
+                    continue
+                elif lineno == len(lines) - 1:
+                    # Something looking like a unix-from at the end - it's
+                    # probably the first line of the body, so push back the
+                    # line and stop.
+                    self._input.unreadline(line)
+                    return
+                else:
+                    # Weirdly placed unix-from line.  Note this as a defect
+                    # and ignore it.
+                    defect = errors.MisplacedEnvelopeHeaderDefect(line)
+                    self._cur.defects.append(defect)
+                    continue
+            # Split the line on the colon separating field name from value.
+            # There will always be a colon, because if there wasn't the part of
+            # the parser that calls us would have started parsing the body.
+            i = line.find(':')
+
+            # If the colon is on the start of the line the header is clearly
+            # malformed, but we might be able to salvage the rest of the
+            # message. Track the error but keep going.
+            if i == 0:
+                defect = errors.InvalidHeaderDefect("Missing header name.")
+                self._cur.defects.append(defect)
+                continue
+
+            assert i>0, "_parse_headers fed line with no : and no leading WS"
+            lastheader = line[:i]
+            lastvalue = [line]
+        # Done with all the lines, so handle the last header.
+        if lastheader:
+            self._cur.set_raw(*self.policy.header_source_parse(lastvalue))
+
+
+class BytesFeedParser(FeedParser):
+    """Like FeedParser, but feed accepts bytes."""
+
+    def feed(self, data):
+        super().feed(data.decode('ascii', 'surrogateescape'))
diff --git a/Lib/email/generator.py b/Lib/email/generator.py
new file mode 100644
index 000000000..ae670c235
--- /dev/null
+++ b/Lib/email/generator.py
@@ -0,0 +1,508 @@
+# Copyright (C) 2001-2010 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Classes to generate plain text from a message object tree."""
+
+__all__ = ['Generator', 'DecodedGenerator', 'BytesGenerator']
+
+import re
+import sys
+import time
+import random
+
+from copy import deepcopy
+from io import StringIO, BytesIO
+from email.utils import _has_surrogates
+
+UNDERSCORE = '_'
+NL = '\n'  # XXX: no longer used by the code below.
+
+NLCRE = re.compile(r'\r\n|\r|\n')
+fcre = re.compile(r'^From ', re.MULTILINE)
+
+
+
+class Generator:
+    """Generates output from a Message object tree.
+
+    This basic generator writes the message to the given file object as plain
+    text.
+    """
+    #
+    # Public interface
+    #
+
+    def __init__(self, outfp, mangle_from_=None, maxheaderlen=None, *,
+                 policy=None):
+        """Create the generator for message flattening.
+
+        outfp is the output file-like object for writing the message to.  It
+        must have a write() method.
+
+        Optional mangle_from_ is a flag that, when True (the default if policy
+        is not set), escapes From_ lines in the body of the message by putting
+        a `>' in front of them.
+
+        Optional maxheaderlen specifies the longest length for a non-continued
+        header.  When a header line is longer (in characters, with tabs
+        expanded to 8 spaces) than maxheaderlen, the header will split as
+        defined in the Header class.  Set maxheaderlen to zero to disable
+        header wrapping.  The default is 78, as recommended (but not required)
+        by RFC 2822.
+
+        The policy keyword specifies a policy object that controls a number of
+        aspects of the generator's operation.  If no policy is specified,
+        the policy associated with the Message object passed to the
+        flatten method is used.
+
+        """
+
+        if mangle_from_ is None:
+            mangle_from_ = True if policy is None else policy.mangle_from_
+        self._fp = outfp
+        self._mangle_from_ = mangle_from_
+        self.maxheaderlen = maxheaderlen
+        self.policy = policy
+
+    def write(self, s):
+        # Just delegate to the file object
+        self._fp.write(s)
+
+    def flatten(self, msg, unixfrom=False, linesep=None):
+        r"""Print the message object tree rooted at msg to the output file
+        specified when the Generator instance was created.
+
+        unixfrom is a flag that forces the printing of a Unix From_ delimiter
+        before the first object in the message tree.  If the original message
+        has no From_ delimiter, a `standard' one is crafted.  By default, this
+        is False to inhibit the printing of any From_ delimiter.
+
+        Note that for subobjects, no From_ line is printed.
+
+        linesep specifies the characters used to indicate a new line in
+        the output.  The default value is determined by the policy specified
+        when the Generator instance was created or, if none was specified,
+        from the policy associated with the msg.
+
+        """
+        # We use the _XXX constants for operating on data that comes directly
+        # from the msg, and _encoded_XXX constants for operating on data that
+        # has already been converted (to bytes in the BytesGenerator) and
+        # inserted into a temporary buffer.
+        policy = msg.policy if self.policy is None else self.policy
+        if linesep is not None:
+            policy = policy.clone(linesep=linesep)
+        if self.maxheaderlen is not None:
+            policy = policy.clone(max_line_length=self.maxheaderlen)
+        self._NL = policy.linesep
+        self._encoded_NL = self._encode(self._NL)
+        self._EMPTY = ''
+        self._encoded_EMPTY = self._encode(self._EMPTY)
+        # Because we use clone (below) when we recursively process message
+        # subparts, and because clone uses the computed policy (not None),
+        # submessages will automatically get set to the computed policy when
+        # they are processed by this code.
+        old_gen_policy = self.policy
+        old_msg_policy = msg.policy
+        try:
+            self.policy = policy
+            msg.policy = policy
+            if unixfrom:
+                ufrom = msg.get_unixfrom()
+                if not ufrom:
+                    ufrom = 'From nobody ' + time.ctime(time.time())
+                self.write(ufrom + self._NL)
+            self._write(msg)
+        finally:
+            self.policy = old_gen_policy
+            msg.policy = old_msg_policy
+
+    def clone(self, fp):
+        """Clone this generator with the exact same options."""
+        return self.__class__(fp,
+                              self._mangle_from_,
+                              None, # Use policy setting, which we've adjusted
+                              policy=self.policy)
+
+    #
+    # Protected interface - undocumented ;/
+    #
+
+    # Note that we use 'self.write' when what we are writing is coming from
+    # the source, and self._fp.write when what we are writing is coming from a
+    # buffer (because the Bytes subclass has already had a chance to transform
+    # the data in its write method in that case).  This is an entirely
+    # pragmatic split determined by experiment; we could be more general by
+    # always using write and having the Bytes subclass write method detect when
+    # it has already transformed the input; but, since this whole thing is a
+    # hack anyway this seems good enough.
+
+    def _new_buffer(self):
+        # BytesGenerator overrides this to return BytesIO.
+        return StringIO()
+
+    def _encode(self, s):
+        # BytesGenerator overrides this to encode strings to bytes.
+        return s
+
+    def _write_lines(self, lines):
+        # We have to transform the line endings.
+        if not lines:
+            return
+        lines = NLCRE.split(lines)
+        for line in lines[:-1]:
+            self.write(line)
+            self.write(self._NL)
+        if lines[-1]:
+            self.write(lines[-1])
+        # XXX logic tells me this else should be needed, but the tests fail
+        # with it and pass without it.  (NLCRE.split ends with a blank element
+        # if and only if there was a trailing newline.)
+        #else:
+        #    self.write(self._NL)
+
+    def _write(self, msg):
+        # We can't write the headers yet because of the following scenario:
+        # say a multipart message includes the boundary string somewhere in
+        # its body.  We'd have to calculate the new boundary /before/ we write
+        # the headers so that we can write the correct Content-Type:
+        # parameter.
+        #
+        # The way we do this, so as to make the _handle_*() methods simpler,
+        # is to cache any subpart writes into a buffer.  The we write the
+        # headers and the buffer contents.  That way, subpart handlers can
+        # Do The Right Thing, and can still modify the Content-Type: header if
+        # necessary.
+        oldfp = self._fp
+        try:
+            self._munge_cte = None
+            self._fp = sfp = self._new_buffer()
+            self._dispatch(msg)
+        finally:
+            self._fp = oldfp
+            munge_cte = self._munge_cte
+            del self._munge_cte
+        # If we munged the cte, copy the message again and re-fix the CTE.
+        if munge_cte:
+            msg = deepcopy(msg)
+            msg.replace_header('content-transfer-encoding', munge_cte[0])
+            msg.replace_header('content-type', munge_cte[1])
+        # Write the headers.  First we see if the message object wants to
+        # handle that itself.  If not, we'll do it generically.
+        meth = getattr(msg, '_write_headers', None)
+        if meth is None:
+            self._write_headers(msg)
+        else:
+            meth(self)
+        self._fp.write(sfp.getvalue())
+
+    def _dispatch(self, msg):
+        # Get the Content-Type: for the message, then try to dispatch to
+        # self._handle_<maintype>_<subtype>().  If there's no handler for the
+        # full MIME type, then dispatch to self._handle_<maintype>().  If
+        # that's missing too, then dispatch to self._writeBody().
+        main = msg.get_content_maintype()
+        sub = msg.get_content_subtype()
+        specific = UNDERSCORE.join((main, sub)).replace('-', '_')
+        meth = getattr(self, '_handle_' + specific, None)
+        if meth is None:
+            generic = main.replace('-', '_')
+            meth = getattr(self, '_handle_' + generic, None)
+            if meth is None:
+                meth = self._writeBody
+        meth(msg)
+
+    #
+    # Default handlers
+    #
+
+    def _write_headers(self, msg):
+        for h, v in msg.raw_items():
+            self.write(self.policy.fold(h, v))
+        # A blank line always separates headers from body
+        self.write(self._NL)
+
+    #
+    # Handlers for writing types and subtypes
+    #
+
+    def _handle_text(self, msg):
+        payload = msg.get_payload()
+        if payload is None:
+            return
+        if not isinstance(payload, str):
+            raise TypeError('string payload expected: %s' % type(payload))
+        if _has_surrogates(msg._payload):
+            charset = msg.get_param('charset')
+            if charset is not None:
+                # XXX: This copy stuff is an ugly hack to avoid modifying the
+                # existing message.
+                msg = deepcopy(msg)
+                del msg['content-transfer-encoding']
+                msg.set_payload(payload, charset)
+                payload = msg.get_payload()
+                self._munge_cte = (msg['content-transfer-encoding'],
+                                   msg['content-type'])
+        if self._mangle_from_:
+            payload = fcre.sub('>From ', payload)
+        self._write_lines(payload)
+
+    # Default body handler
+    _writeBody = _handle_text
+
+    def _handle_multipart(self, msg):
+        # The trick here is to write out each part separately, merge them all
+        # together, and then make sure that the boundary we've chosen isn't
+        # present in the payload.
+        msgtexts = []
+        subparts = msg.get_payload()
+        if subparts is None:
+            subparts = []
+        elif isinstance(subparts, str):
+            # e.g. a non-strict parse of a message with no starting boundary.
+            self.write(subparts)
+            return
+        elif not isinstance(subparts, list):
+            # Scalar payload
+            subparts = [subparts]
+        for part in subparts:
+            s = self._new_buffer()
+            g = self.clone(s)
+            g.flatten(part, unixfrom=False, linesep=self._NL)
+            msgtexts.append(s.getvalue())
+        # BAW: What about boundaries that are wrapped in double-quotes?
+        boundary = msg.get_boundary()
+        if not boundary:
+            # Create a boundary that doesn't appear in any of the
+            # message texts.
+            alltext = self._encoded_NL.join(msgtexts)
+            boundary = self._make_boundary(alltext)
+            msg.set_boundary(boundary)
+        # If there's a preamble, write it out, with a trailing CRLF
+        if msg.preamble is not None:
+            if self._mangle_from_:
+                preamble = fcre.sub('>From ', msg.preamble)
+            else:
+                preamble = msg.preamble
+            self._write_lines(preamble)
+            self.write(self._NL)
+        # dash-boundary transport-padding CRLF
+        self.write('--' + boundary + self._NL)
+        # body-part
+        if msgtexts:
+            self._fp.write(msgtexts.pop(0))
+        # *encapsulation
+        # --> delimiter transport-padding
+        # --> CRLF body-part
+        for body_part in msgtexts:
+            # delimiter transport-padding CRLF
+            self.write(self._NL + '--' + boundary + self._NL)
+            # body-part
+            self._fp.write(body_part)
+        # close-delimiter transport-padding
+        self.write(self._NL + '--' + boundary + '--' + self._NL)
+        if msg.epilogue is not None:
+            if self._mangle_from_:
+                epilogue = fcre.sub('>From ', msg.epilogue)
+            else:
+                epilogue = msg.epilogue
+            self._write_lines(epilogue)
+
+    def _handle_multipart_signed(self, msg):
+        # The contents of signed parts has to stay unmodified in order to keep
+        # the signature intact per RFC1847 2.1, so we disable header wrapping.
+        # RDM: This isn't enough to completely preserve the part, but it helps.
+        p = self.policy
+        self.policy = p.clone(max_line_length=0)
+        try:
+            self._handle_multipart(msg)
+        finally:
+            self.policy = p
+
+    def _handle_message_delivery_status(self, msg):
+        # We can't just write the headers directly to self's file object
+        # because this will leave an extra newline between the last header
+        # block and the boundary.  Sigh.
+        blocks = []
+        for part in msg.get_payload():
+            s = self._new_buffer()
+            g = self.clone(s)
+            g.flatten(part, unixfrom=False, linesep=self._NL)
+            text = s.getvalue()
+            lines = text.split(self._encoded_NL)
+            # Strip off the unnecessary trailing empty line
+            if lines and lines[-1] == self._encoded_EMPTY:
+                blocks.append(self._encoded_NL.join(lines[:-1]))
+            else:
+                blocks.append(text)
+        # Now join all the blocks with an empty line.  This has the lovely
+        # effect of separating each block with an empty line, but not adding
+        # an extra one after the last one.
+        self._fp.write(self._encoded_NL.join(blocks))
+
+    def _handle_message(self, msg):
+        s = self._new_buffer()
+        g = self.clone(s)
+        # The payload of a message/rfc822 part should be a multipart sequence
+        # of length 1.  The zeroth element of the list should be the Message
+        # object for the subpart.  Extract that object, stringify it, and
+        # write it out.
+        # Except, it turns out, when it's a string instead, which happens when
+        # and only when HeaderParser is used on a message of mime type
+        # message/rfc822.  Such messages are generated by, for example,
+        # Groupwise when forwarding unadorned messages.  (Issue 7970.)  So
+        # in that case we just emit the string body.
+        payload = msg._payload
+        if isinstance(payload, list):
+            g.flatten(msg.get_payload(0), unixfrom=False, linesep=self._NL)
+            payload = s.getvalue()
+        else:
+            payload = self._encode(payload)
+        self._fp.write(payload)
+
+    # This used to be a module level function; we use a classmethod for this
+    # and _compile_re so we can continue to provide the module level function
+    # for backward compatibility by doing
+    #   _make_boundary = Generator._make_boundary
+    # at the end of the module.  It *is* internal, so we could drop that...
+    @classmethod
+    def _make_boundary(cls, text=None):
+        # Craft a random boundary.  If text is given, ensure that the chosen
+        # boundary doesn't appear in the text.
+        token = random.randrange(sys.maxsize)
+        boundary = ('=' * 15) + (_fmt % token) + '=='
+        if text is None:
+            return boundary
+        b = boundary
+        counter = 0
+        while True:
+            cre = cls._compile_re('^--' + re.escape(b) + '(--)?$', re.MULTILINE)
+            if not cre.search(text):
+                break
+            b = boundary + '.' + str(counter)
+            counter += 1
+        return b
+
+    @classmethod
+    def _compile_re(cls, s, flags):
+        return re.compile(s, flags)
+
+
+class BytesGenerator(Generator):
+    """Generates a bytes version of a Message object tree.
+
+    Functionally identical to the base Generator except that the output is
+    bytes and not string.  When surrogates were used in the input to encode
+    bytes, these are decoded back to bytes for output.  If the policy has
+    cte_type set to 7bit, then the message is transformed such that the
+    non-ASCII bytes are properly content transfer encoded, using the charset
+    unknown-8bit.
+
+    The outfp object must accept bytes in its write method.
+    """
+
+    def write(self, s):
+        self._fp.write(s.encode('ascii', 'surrogateescape'))
+
+    def _new_buffer(self):
+        return BytesIO()
+
+    def _encode(self, s):
+        return s.encode('ascii')
+
+    def _write_headers(self, msg):
+        # This is almost the same as the string version, except for handling
+        # strings with 8bit bytes.
+        for h, v in msg.raw_items():
+            self._fp.write(self.policy.fold_binary(h, v))
+        # A blank line always separates headers from body
+        self.write(self._NL)
+
+    def _handle_text(self, msg):
+        # If the string has surrogates the original source was bytes, so
+        # just write it back out.
+        if msg._payload is None:
+            return
+        if _has_surrogates(msg._payload) and not self.policy.cte_type=='7bit':
+            if self._mangle_from_:
+                msg._payload = fcre.sub(">From ", msg._payload)
+            self._write_lines(msg._payload)
+        else:
+            super(BytesGenerator,self)._handle_text(msg)
+
+    # Default body handler
+    _writeBody = _handle_text
+
+    @classmethod
+    def _compile_re(cls, s, flags):
+        return re.compile(s.encode('ascii'), flags)
+
+
+
+_FMT = '[Non-text (%(type)s) part of message omitted, filename %(filename)s]'
+
+class DecodedGenerator(Generator):
+    """Generates a text representation of a message.
+
+    Like the Generator base class, except that non-text parts are substituted
+    with a format string representing the part.
+    """
+    def __init__(self, outfp, mangle_from_=None, maxheaderlen=None, fmt=None, *,
+                 policy=None):
+        """Like Generator.__init__() except that an additional optional
+        argument is allowed.
+
+        Walks through all subparts of a message.  If the subpart is of main
+        type `text', then it prints the decoded payload of the subpart.
+
+        Otherwise, fmt is a format string that is used instead of the message
+        payload.  fmt is expanded with the following keywords (in
+        %(keyword)s format):
+
+        type       : Full MIME type of the non-text part
+        maintype   : Main MIME type of the non-text part
+        subtype    : Sub-MIME type of the non-text part
+        filename   : Filename of the non-text part
+        description: Description associated with the non-text part
+        encoding   : Content transfer encoding of the non-text part
+
+        The default value for fmt is None, meaning
+
+        [Non-text (%(type)s) part of message omitted, filename %(filename)s]
+        """
+        Generator.__init__(self, outfp, mangle_from_, maxheaderlen,
+                           policy=policy)
+        if fmt is None:
+            self._fmt = _FMT
+        else:
+            self._fmt = fmt
+
+    def _dispatch(self, msg):
+        for part in msg.walk():
+            maintype = part.get_content_maintype()
+            if maintype == 'text':
+                print(part.get_payload(decode=False), file=self)
+            elif maintype == 'multipart':
+                # Just skip this
+                pass
+            else:
+                print(self._fmt % {
+                    'type'       : part.get_content_type(),
+                    'maintype'   : part.get_content_maintype(),
+                    'subtype'    : part.get_content_subtype(),
+                    'filename'   : part.get_filename('[no filename]'),
+                    'description': part.get('Content-Description',
+                                            '[no description]'),
+                    'encoding'   : part.get('Content-Transfer-Encoding',
+                                            '[no encoding]'),
+                    }, file=self)
+
+
+
+# Helper used by Generator._make_boundary
+_width = len(repr(sys.maxsize-1))
+_fmt = '%%0%dd' % _width
+
+# Backward compatibility
+_make_boundary = Generator._make_boundary
diff --git a/Lib/email/header.py b/Lib/email/header.py
new file mode 100644
index 000000000..c7b2dd9f3
--- /dev/null
+++ b/Lib/email/header.py
@@ -0,0 +1,578 @@
+# Copyright (C) 2002-2007 Python Software Foundation
+# Author: Ben Gertzfield, Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Header encoding and decoding functionality."""
+
+__all__ = [
+    'Header',
+    'decode_header',
+    'make_header',
+    ]
+
+import re
+import binascii
+
+import email.quoprimime
+import email.base64mime
+
+from email.errors import HeaderParseError
+from email import charset as _charset
+Charset = _charset.Charset
+
+NL = '\n'
+SPACE = ' '
+BSPACE = b' '
+SPACE8 = ' ' * 8
+EMPTYSTRING = ''
+MAXLINELEN = 78
+FWS = ' \t'
+
+USASCII = Charset('us-ascii')
+UTF8 = Charset('utf-8')
+
+# Match encoded-word strings in the form =?charset?q?Hello_World?=
+ecre = re.compile(r'''
+  =\?                   # literal =?
+  (?P<charset>[^?]*?)   # non-greedy up to the next ? is the charset
+  \?                    # literal ?
+  (?P<encoding>[qb])    # either a "q" or a "b", case insensitive
+  \?                    # literal ?
+  (?P<encoded>.*?)      # non-greedy up to the next ?= is the encoded string
+  \?=                   # literal ?=
+  ''', re.VERBOSE | re.IGNORECASE | re.MULTILINE)
+
+# Field name regexp, including trailing colon, but not separating whitespace,
+# according to RFC 2822.  Character range is from tilde to exclamation mark.
+# For use with .match()
+fcre = re.compile(r'[\041-\176]+:$')
+
+# Find a header embedded in a putative header value.  Used to check for
+# header injection attack.
+_embedded_header = re.compile(r'\n[^ \t]+:')
+
+
+
+# Helpers
+_max_append = email.quoprimime._max_append
+
+
+
+def decode_header(header):
+    """Decode a message header value without converting charset.
+
+    Returns a list of (string, charset) pairs containing each of the decoded
+    parts of the header.  Charset is None for non-encoded parts of the header,
+    otherwise a lower-case string containing the name of the character set
+    specified in the encoded string.
+
+    header may be a string that may or may not contain RFC2047 encoded words,
+    or it may be a Header object.
+
+    An email.errors.HeaderParseError may be raised when certain decoding error
+    occurs (e.g. a base64 decoding exception).
+    """
+    # If it is a Header object, we can just return the encoded chunks.
+    if hasattr(header, '_chunks'):
+        return [(_charset._encode(string, str(charset)), str(charset))
+                    for string, charset in header._chunks]
+    # If no encoding, just return the header with no charset.
+    if not ecre.search(header):
+        return [(header, None)]
+    # First step is to parse all the encoded parts into triplets of the form
+    # (encoded_string, encoding, charset).  For unencoded strings, the last
+    # two parts will be None.
+    words = []
+    for line in header.splitlines():
+        parts = ecre.split(line)
+        first = True
+        while parts:
+            unencoded = parts.pop(0)
+            if first:
+                unencoded = unencoded.lstrip()
+                first = False
+            if unencoded:
+                words.append((unencoded, None, None))
+            if parts:
+                charset = parts.pop(0).lower()
+                encoding = parts.pop(0).lower()
+                encoded = parts.pop(0)
+                words.append((encoded, encoding, charset))
+    # Now loop over words and remove words that consist of whitespace
+    # between two encoded strings.
+    droplist = []
+    for n, w in enumerate(words):
+        if n>1 and w[1] and words[n-2][1] and words[n-1][0].isspace():
+            droplist.append(n-1)
+    for d in reversed(droplist):
+        del words[d]
+
+    # The next step is to decode each encoded word by applying the reverse
+    # base64 or quopri transformation.  decoded_words is now a list of the
+    # form (decoded_word, charset).
+    decoded_words = []
+    for encoded_string, encoding, charset in words:
+        if encoding is None:
+            # This is an unencoded word.
+            decoded_words.append((encoded_string, charset))
+        elif encoding == 'q':
+            word = email.quoprimime.header_decode(encoded_string)
+            decoded_words.append((word, charset))
+        elif encoding == 'b':
+            paderr = len(encoded_string) % 4   # Postel's law: add missing padding
+            if paderr:
+                encoded_string += '==='[:4 - paderr]
+            try:
+                word = email.base64mime.decode(encoded_string)
+            except binascii.Error:
+                raise HeaderParseError('Base64 decoding error')
+            else:
+                decoded_words.append((word, charset))
+        else:
+            raise AssertionError('Unexpected encoding: ' + encoding)
+    # Now convert all words to bytes and collapse consecutive runs of
+    # similarly encoded words.
+    collapsed = []
+    last_word = last_charset = None
+    for word, charset in decoded_words:
+        if isinstance(word, str):
+            word = bytes(word, 'raw-unicode-escape')
+        if last_word is None:
+            last_word = word
+            last_charset = charset
+        elif charset != last_charset:
+            collapsed.append((last_word, last_charset))
+            last_word = word
+            last_charset = charset
+        elif last_charset is None:
+            last_word += BSPACE + word
+        else:
+            last_word += word
+    collapsed.append((last_word, last_charset))
+    return collapsed
+
+
+
+def make_header(decoded_seq, maxlinelen=None, header_name=None,
+                continuation_ws=' '):
+    """Create a Header from a sequence of pairs as returned by decode_header()
+
+    decode_header() takes a header value string and returns a sequence of
+    pairs of the format (decoded_string, charset) where charset is the string
+    name of the character set.
+
+    This function takes one of those sequence of pairs and returns a Header
+    instance.  Optional maxlinelen, header_name, and continuation_ws are as in
+    the Header constructor.
+    """
+    h = Header(maxlinelen=maxlinelen, header_name=header_name,
+               continuation_ws=continuation_ws)
+    for s, charset in decoded_seq:
+        # None means us-ascii but we can simply pass it on to h.append()
+        if charset is not None and not isinstance(charset, Charset):
+            charset = Charset(charset)
+        h.append(s, charset)
+    return h
+
+
+
+class Header:
+    def __init__(self, s=None, charset=None,
+                 maxlinelen=None, header_name=None,
+                 continuation_ws=' ', errors='strict'):
+        """Create a MIME-compliant header that can contain many character sets.
+
+        Optional s is the initial header value.  If None, the initial header
+        value is not set.  You can later append to the header with .append()
+        method calls.  s may be a byte string or a Unicode string, but see the
+        .append() documentation for semantics.
+
+        Optional charset serves two purposes: it has the same meaning as the
+        charset argument to the .append() method.  It also sets the default
+        character set for all subsequent .append() calls that omit the charset
+        argument.  If charset is not provided in the constructor, the us-ascii
+        charset is used both as s's initial charset and as the default for
+        subsequent .append() calls.
+
+        The maximum line length can be specified explicitly via maxlinelen. For
+        splitting the first line to a shorter value (to account for the field
+        header which isn't included in s, e.g. `Subject') pass in the name of
+        the field in header_name.  The default maxlinelen is 78 as recommended
+        by RFC 2822.
+
+        continuation_ws must be RFC 2822 compliant folding whitespace (usually
+        either a space or a hard tab) which will be prepended to continuation
+        lines.
+
+        errors is passed through to the .append() call.
+        """
+        if charset is None:
+            charset = USASCII
+        elif not isinstance(charset, Charset):
+            charset = Charset(charset)
+        self._charset = charset
+        self._continuation_ws = continuation_ws
+        self._chunks = []
+        if s is not None:
+            self.append(s, charset, errors)
+        if maxlinelen is None:
+            maxlinelen = MAXLINELEN
+        self._maxlinelen = maxlinelen
+        if header_name is None:
+            self._headerlen = 0
+        else:
+            # Take the separating colon and space into account.
+            self._headerlen = len(header_name) + 2
+
+    def __str__(self):
+        """Return the string value of the header."""
+        self._normalize()
+        uchunks = []
+        lastcs = None
+        lastspace = None
+        for string, charset in self._chunks:
+            # We must preserve spaces between encoded and non-encoded word
+            # boundaries, which means for us we need to add a space when we go
+            # from a charset to None/us-ascii, or from None/us-ascii to a
+            # charset.  Only do this for the second and subsequent chunks.
+            # Don't add a space if the None/us-ascii string already has
+            # a space (trailing or leading depending on transition)
+            nextcs = charset
+            if nextcs == _charset.UNKNOWN8BIT:
+                original_bytes = string.encode('ascii', 'surrogateescape')
+                string = original_bytes.decode('ascii', 'replace')
+            if uchunks:
+                hasspace = string and self._nonctext(string[0])
+                if lastcs not in (None, 'us-ascii'):
+                    if nextcs in (None, 'us-ascii') and not hasspace:
+                        uchunks.append(SPACE)
+                        nextcs = None
+                elif nextcs not in (None, 'us-ascii') and not lastspace:
+                    uchunks.append(SPACE)
+            lastspace = string and self._nonctext(string[-1])
+            lastcs = nextcs
+            uchunks.append(string)
+        return EMPTYSTRING.join(uchunks)
+
+    # Rich comparison operators for equality only.  BAW: does it make sense to
+    # have or explicitly disable <, <=, >, >= operators?
+    def __eq__(self, other):
+        # other may be a Header or a string.  Both are fine so coerce
+        # ourselves to a unicode (of the unencoded header value), swap the
+        # args and do another comparison.
+        return other == str(self)
+
+    def append(self, s, charset=None, errors='strict'):
+        """Append a string to the MIME header.
+
+        Optional charset, if given, should be a Charset instance or the name
+        of a character set (which will be converted to a Charset instance).  A
+        value of None (the default) means that the charset given in the
+        constructor is used.
+
+        s may be a byte string or a Unicode string.  If it is a byte string
+        (i.e. isinstance(s, str) is false), then charset is the encoding of
+        that byte string, and a UnicodeError will be raised if the string
+        cannot be decoded with that charset.  If s is a Unicode string, then
+        charset is a hint specifying the character set of the characters in
+        the string.  In either case, when producing an RFC 2822 compliant
+        header using RFC 2047 rules, the string will be encoded using the
+        output codec of the charset.  If the string cannot be encoded to the
+        output codec, a UnicodeError will be raised.
+
+        Optional `errors' is passed as the errors argument to the decode
+        call if s is a byte string.
+        """
+        if charset is None:
+            charset = self._charset
+        elif not isinstance(charset, Charset):
+            charset = Charset(charset)
+        if not isinstance(s, str):
+            input_charset = charset.input_codec or 'us-ascii'
+            if input_charset == _charset.UNKNOWN8BIT:
+                s = s.decode('us-ascii', 'surrogateescape')
+            else:
+                s = s.decode(input_charset, errors)
+        # Ensure that the bytes we're storing can be decoded to the output
+        # character set, otherwise an early error is raised.
+        output_charset = charset.output_codec or 'us-ascii'
+        if output_charset != _charset.UNKNOWN8BIT:
+            try:
+                s.encode(output_charset, errors)
+            except UnicodeEncodeError:
+                if output_charset!='us-ascii':
+                    raise
+                charset = UTF8
+        self._chunks.append((s, charset))
+
+    def _nonctext(self, s):
+        """True if string s is not a ctext character of RFC822.
+        """
+        return s.isspace() or s in ('(', ')', '\\')
+
+    def encode(self, splitchars=';, \t', maxlinelen=None, linesep='\n'):
+        r"""Encode a message header into an RFC-compliant format.
+
+        There are many issues involved in converting a given string for use in
+        an email header.  Only certain character sets are readable in most
+        email clients, and as header strings can only contain a subset of
+        7-bit ASCII, care must be taken to properly convert and encode (with
+        Base64 or quoted-printable) header strings.  In addition, there is a
+        75-character length limit on any given encoded header field, so
+        line-wrapping must be performed, even with double-byte character sets.
+
+        Optional maxlinelen specifies the maximum length of each generated
+        line, exclusive of the linesep string.  Individual lines may be longer
+        than maxlinelen if a folding point cannot be found.  The first line
+        will be shorter by the length of the header name plus ": " if a header
+        name was specified at Header construction time.  The default value for
+        maxlinelen is determined at header construction time.
+
+        Optional splitchars is a string containing characters which should be
+        given extra weight by the splitting algorithm during normal header
+        wrapping.  This is in very rough support of RFC 2822's `higher level
+        syntactic breaks':  split points preceded by a splitchar are preferred
+        during line splitting, with the characters preferred in the order in
+        which they appear in the string.  Space and tab may be included in the
+        string to indicate whether preference should be given to one over the
+        other as a split point when other split chars do not appear in the line
+        being split.  Splitchars does not affect RFC 2047 encoded lines.
+
+        Optional linesep is a string to be used to separate the lines of
+        the value.  The default value is the most useful for typical
+        Python applications, but it can be set to \r\n to produce RFC-compliant
+        line separators when needed.
+        """
+        self._normalize()
+        if maxlinelen is None:
+            maxlinelen = self._maxlinelen
+        # A maxlinelen of 0 means don't wrap.  For all practical purposes,
+        # choosing a huge number here accomplishes that and makes the
+        # _ValueFormatter algorithm much simpler.
+        if maxlinelen == 0:
+            maxlinelen = 1000000
+        formatter = _ValueFormatter(self._headerlen, maxlinelen,
+                                    self._continuation_ws, splitchars)
+        lastcs = None
+        hasspace = lastspace = None
+        for string, charset in self._chunks:
+            if hasspace is not None:
+                hasspace = string and self._nonctext(string[0])
+                if lastcs not in (None, 'us-ascii'):
+                    if not hasspace or charset not in (None, 'us-ascii'):
+                        formatter.add_transition()
+                elif charset not in (None, 'us-ascii') and not lastspace:
+                    formatter.add_transition()
+            lastspace = string and self._nonctext(string[-1])
+            lastcs = charset
+            hasspace = False
+            lines = string.splitlines()
+            if lines:
+                formatter.feed('', lines[0], charset)
+            else:
+                formatter.feed('', '', charset)
+            for line in lines[1:]:
+                formatter.newline()
+                if charset.header_encoding is not None:
+                    formatter.feed(self._continuation_ws, ' ' + line.lstrip(),
+                                   charset)
+                else:
+                    sline = line.lstrip()
+                    fws = line[:len(line)-len(sline)]
+                    formatter.feed(fws, sline, charset)
+            if len(lines) > 1:
+                formatter.newline()
+        if self._chunks:
+            formatter.add_transition()
+        value = formatter._str(linesep)
+        if _embedded_header.search(value):
+            raise HeaderParseError("header value appears to contain "
+                "an embedded header: {!r}".format(value))
+        return value
+
+    def _normalize(self):
+        # Step 1: Normalize the chunks so that all runs of identical charsets
+        # get collapsed into a single unicode string.
+        chunks = []
+        last_charset = None
+        last_chunk = []
+        for string, charset in self._chunks:
+            if charset == last_charset:
+                last_chunk.append(string)
+            else:
+                if last_charset is not None:
+                    chunks.append((SPACE.join(last_chunk), last_charset))
+                last_chunk = [string]
+                last_charset = charset
+        if last_chunk:
+            chunks.append((SPACE.join(last_chunk), last_charset))
+        self._chunks = chunks
+
+
+
+class _ValueFormatter:
+    def __init__(self, headerlen, maxlen, continuation_ws, splitchars):
+        self._maxlen = maxlen
+        self._continuation_ws = continuation_ws
+        self._continuation_ws_len = len(continuation_ws)
+        self._splitchars = splitchars
+        self._lines = []
+        self._current_line = _Accumulator(headerlen)
+
+    def _str(self, linesep):
+        self.newline()
+        return linesep.join(self._lines)
+
+    def __str__(self):
+        return self._str(NL)
+
+    def newline(self):
+        end_of_line = self._current_line.pop()
+        if end_of_line != (' ', ''):
+            self._current_line.push(*end_of_line)
+        if len(self._current_line) > 0:
+            if self._current_line.is_onlyws():
+                self._lines[-1] += str(self._current_line)
+            else:
+                self._lines.append(str(self._current_line))
+        self._current_line.reset()
+
+    def add_transition(self):
+        self._current_line.push(' ', '')
+
+    def feed(self, fws, string, charset):
+        # If the charset has no header encoding (i.e. it is an ASCII encoding)
+        # then we must split the header at the "highest level syntactic break"
+        # possible. Note that we don't have a lot of smarts about field
+        # syntax; we just try to break on semi-colons, then commas, then
+        # whitespace.  Eventually, this should be pluggable.
+        if charset.header_encoding is None:
+            self._ascii_split(fws, string, self._splitchars)
+            return
+        # Otherwise, we're doing either a Base64 or a quoted-printable
+        # encoding which means we don't need to split the line on syntactic
+        # breaks.  We can basically just find enough characters to fit on the
+        # current line, minus the RFC 2047 chrome.  What makes this trickier
+        # though is that we have to split at octet boundaries, not character
+        # boundaries but it's only safe to split at character boundaries so at
+        # best we can only get close.
+        encoded_lines = charset.header_encode_lines(string, self._maxlengths())
+        # The first element extends the current line, but if it's None then
+        # nothing more fit on the current line so start a new line.
+        try:
+            first_line = encoded_lines.pop(0)
+        except IndexError:
+            # There are no encoded lines, so we're done.
+            return
+        if first_line is not None:
+            self._append_chunk(fws, first_line)
+        try:
+            last_line = encoded_lines.pop()
+        except IndexError:
+            # There was only one line.
+            return
+        self.newline()
+        self._current_line.push(self._continuation_ws, last_line)
+        # Everything else are full lines in themselves.
+        for line in encoded_lines:
+            self._lines.append(self._continuation_ws + line)
+
+    def _maxlengths(self):
+        # The first line's length.
+        yield self._maxlen - len(self._current_line)
+        while True:
+            yield self._maxlen - self._continuation_ws_len
+
+    def _ascii_split(self, fws, string, splitchars):
+        # The RFC 2822 header folding algorithm is simple in principle but
+        # complex in practice.  Lines may be folded any place where "folding
+        # white space" appears by inserting a linesep character in front of the
+        # FWS.  The complication is that not all spaces or tabs qualify as FWS,
+        # and we are also supposed to prefer to break at "higher level
+        # syntactic breaks".  We can't do either of these without intimate
+        # knowledge of the structure of structured headers, which we don't have
+        # here.  So the best we can do here is prefer to break at the specified
+        # splitchars, and hope that we don't choose any spaces or tabs that
+        # aren't legal FWS.  (This is at least better than the old algorithm,
+        # where we would sometimes *introduce* FWS after a splitchar, or the
+        # algorithm before that, where we would turn all white space runs into
+        # single spaces or tabs.)
+        parts = re.split("(["+FWS+"]+)", fws+string)
+        if parts[0]:
+            parts[:0] = ['']
+        else:
+            parts.pop(0)
+        for fws, part in zip(*[iter(parts)]*2):
+            self._append_chunk(fws, part)
+
+    def _append_chunk(self, fws, string):
+        self._current_line.push(fws, string)
+        if len(self._current_line) > self._maxlen:
+            # Find the best split point, working backward from the end.
+            # There might be none, on a long first line.
+            for ch in self._splitchars:
+                for i in range(self._current_line.part_count()-1, 0, -1):
+                    if ch.isspace():
+                        fws = self._current_line[i][0]
+                        if fws and fws[0]==ch:
+                            break
+                    prevpart = self._current_line[i-1][1]
+                    if prevpart and prevpart[-1]==ch:
+                        break
+                else:
+                    continue
+                break
+            else:
+                fws, part = self._current_line.pop()
+                if self._current_line._initial_size > 0:
+                    # There will be a header, so leave it on a line by itself.
+                    self.newline()
+                    if not fws:
+                        # We don't use continuation_ws here because the whitespace
+                        # after a header should always be a space.
+                        fws = ' '
+                self._current_line.push(fws, part)
+                return
+            remainder = self._current_line.pop_from(i)
+            self._lines.append(str(self._current_line))
+            self._current_line.reset(remainder)
+
+
+class _Accumulator(list):
+
+    def __init__(self, initial_size=0):
+        self._initial_size = initial_size
+        super().__init__()
+
+    def push(self, fws, string):
+        self.append((fws, string))
+
+    def pop_from(self, i=0):
+        popped = self[i:]
+        self[i:] = []
+        return popped
+
+    def pop(self):
+        if self.part_count()==0:
+            return ('', '')
+        return super().pop()
+
+    def __len__(self):
+        return sum((len(fws)+len(part) for fws, part in self),
+                   self._initial_size)
+
+    def __str__(self):
+        return EMPTYSTRING.join((EMPTYSTRING.join((fws, part))
+                                for fws, part in self))
+
+    def reset(self, startval=None):
+        if startval is None:
+            startval = []
+        self[:] = startval
+        self._initial_size = 0
+
+    def is_onlyws(self):
+        return self._initial_size==0 and (not self or str(self).isspace())
+
+    def part_count(self):
+        return super().__len__()
diff --git a/Lib/email/headerregistry.py b/Lib/email/headerregistry.py
new file mode 100644
index 000000000..0fc2231e5
--- /dev/null
+++ b/Lib/email/headerregistry.py
@@ -0,0 +1,586 @@
+"""Representing and manipulating email headers via custom objects.
+
+This module provides an implementation of the HeaderRegistry API.
+The implementation is designed to flexibly follow RFC5322 rules.
+
+Eventually HeaderRegistry will be a public API, but it isn't yet,
+and will probably change some before that happens.
+
+"""
+from types import MappingProxyType
+
+from email import utils
+from email import errors
+from email import _header_value_parser as parser
+
+class Address:
+
+    def __init__(self, display_name='', username='', domain='', addr_spec=None):
+        """Create an object representing a full email address.
+
+        An address can have a 'display_name', a 'username', and a 'domain'.  In
+        addition to specifying the username and domain separately, they may be
+        specified together by using the addr_spec keyword *instead of* the
+        username and domain keywords.  If an addr_spec string is specified it
+        must be properly quoted according to RFC 5322 rules; an error will be
+        raised if it is not.
+
+        An Address object has display_name, username, domain, and addr_spec
+        attributes, all of which are read-only.  The addr_spec and the string
+        value of the object are both quoted according to RFC5322 rules, but
+        without any Content Transfer Encoding.
+
+        """
+        # This clause with its potential 'raise' may only happen when an
+        # application program creates an Address object using an addr_spec
+        # keyword.  The email library code itself must always supply username
+        # and domain.
+        if addr_spec is not None:
+            if username or domain:
+                raise TypeError("addrspec specified when username and/or "
+                                "domain also specified")
+            a_s, rest = parser.get_addr_spec(addr_spec)
+            if rest:
+                raise ValueError("Invalid addr_spec; only '{}' "
+                                 "could be parsed from '{}'".format(
+                                    a_s, addr_spec))
+            if a_s.all_defects:
+                raise a_s.all_defects[0]
+            username = a_s.local_part
+            domain = a_s.domain
+        self._display_name = display_name
+        self._username = username
+        self._domain = domain
+
+    @property
+    def display_name(self):
+        return self._display_name
+
+    @property
+    def username(self):
+        return self._username
+
+    @property
+    def domain(self):
+        return self._domain
+
+    @property
+    def addr_spec(self):
+        """The addr_spec (username@domain) portion of the address, quoted
+        according to RFC 5322 rules, but with no Content Transfer Encoding.
+        """
+        nameset = set(self.username)
+        if len(nameset) > len(nameset-parser.DOT_ATOM_ENDS):
+            lp = parser.quote_string(self.username)
+        else:
+            lp = self.username
+        if self.domain:
+            return lp + '@' + self.domain
+        if not lp:
+            return '<>'
+        return lp
+
+    def __repr__(self):
+        return "{}(display_name={!r}, username={!r}, domain={!r})".format(
+                        self.__class__.__name__,
+                        self.display_name, self.username, self.domain)
+
+    def __str__(self):
+        nameset = set(self.display_name)
+        if len(nameset) > len(nameset-parser.SPECIALS):
+            disp = parser.quote_string(self.display_name)
+        else:
+            disp = self.display_name
+        if disp:
+            addr_spec = '' if self.addr_spec=='<>' else self.addr_spec
+            return "{} <{}>".format(disp, addr_spec)
+        return self.addr_spec
+
+    def __eq__(self, other):
+        if type(other) != type(self):
+            return False
+        return (self.display_name == other.display_name and
+                self.username == other.username and
+                self.domain == other.domain)
+
+
+class Group:
+
+    def __init__(self, display_name=None, addresses=None):
+        """Create an object representing an address group.
+
+        An address group consists of a display_name followed by colon and a
+        list of addresses (see Address) terminated by a semi-colon.  The Group
+        is created by specifying a display_name and a possibly empty list of
+        Address objects.  A Group can also be used to represent a single
+        address that is not in a group, which is convenient when manipulating
+        lists that are a combination of Groups and individual Addresses.  In
+        this case the display_name should be set to None.  In particular, the
+        string representation of a Group whose display_name is None is the same
+        as the Address object, if there is one and only one Address object in
+        the addresses list.
+
+        """
+        self._display_name = display_name
+        self._addresses = tuple(addresses) if addresses else tuple()
+
+    @property
+    def display_name(self):
+        return self._display_name
+
+    @property
+    def addresses(self):
+        return self._addresses
+
+    def __repr__(self):
+        return "{}(display_name={!r}, addresses={!r}".format(
+                 self.__class__.__name__,
+                 self.display_name, self.addresses)
+
+    def __str__(self):
+        if self.display_name is None and len(self.addresses)==1:
+            return str(self.addresses[0])
+        disp = self.display_name
+        if disp is not None:
+            nameset = set(disp)
+            if len(nameset) > len(nameset-parser.SPECIALS):
+                disp = parser.quote_string(disp)
+        adrstr = ", ".join(str(x) for x in self.addresses)
+        adrstr = ' ' + adrstr if adrstr else adrstr
+        return "{}:{};".format(disp, adrstr)
+
+    def __eq__(self, other):
+        if type(other) != type(self):
+            return False
+        return (self.display_name == other.display_name and
+                self.addresses == other.addresses)
+
+
+# Header Classes #
+
+class BaseHeader(str):
+
+    """Base class for message headers.
+
+    Implements generic behavior and provides tools for subclasses.
+
+    A subclass must define a classmethod named 'parse' that takes an unfolded
+    value string and a dictionary as its arguments.  The dictionary will
+    contain one key, 'defects', initialized to an empty list.  After the call
+    the dictionary must contain two additional keys: parse_tree, set to the
+    parse tree obtained from parsing the header, and 'decoded', set to the
+    string value of the idealized representation of the data from the value.
+    (That is, encoded words are decoded, and values that have canonical
+    representations are so represented.)
+
+    The defects key is intended to collect parsing defects, which the message
+    parser will subsequently dispose of as appropriate.  The parser should not,
+    insofar as practical, raise any errors.  Defects should be added to the
+    list instead.  The standard header parsers register defects for RFC
+    compliance issues, for obsolete RFC syntax, and for unrecoverable parsing
+    errors.
+
+    The parse method may add additional keys to the dictionary.  In this case
+    the subclass must define an 'init' method, which will be passed the
+    dictionary as its keyword arguments.  The method should use (usually by
+    setting them as the value of similarly named attributes) and remove all the
+    extra keys added by its parse method, and then use super to call its parent
+    class with the remaining arguments and keywords.
+
+    The subclass should also make sure that a 'max_count' attribute is defined
+    that is either None or 1. XXX: need to better define this API.
+
+    """
+
+    def __new__(cls, name, value):
+        kwds = {'defects': []}
+        cls.parse(value, kwds)
+        if utils._has_surrogates(kwds['decoded']):
+            kwds['decoded'] = utils._sanitize(kwds['decoded'])
+        self = str.__new__(cls, kwds['decoded'])
+        del kwds['decoded']
+        self.init(name, **kwds)
+        return self
+
+    def init(self, name, *, parse_tree, defects):
+        self._name = name
+        self._parse_tree = parse_tree
+        self._defects = defects
+
+    @property
+    def name(self):
+        return self._name
+
+    @property
+    def defects(self):
+        return tuple(self._defects)
+
+    def __reduce__(self):
+        return (
+            _reconstruct_header,
+            (
+                self.__class__.__name__,
+                self.__class__.__bases__,
+                str(self),
+            ),
+            self.__dict__)
+
+    @classmethod
+    def _reconstruct(cls, value):
+        return str.__new__(cls, value)
+
+    def fold(self, *, policy):
+        """Fold header according to policy.
+
+        The parsed representation of the header is folded according to
+        RFC5322 rules, as modified by the policy.  If the parse tree
+        contains surrogateescaped bytes, the bytes are CTE encoded using
+        the charset 'unknown-8bit".
+
+        Any non-ASCII characters in the parse tree are CTE encoded using
+        charset utf-8. XXX: make this a policy setting.
+
+        The returned value is an ASCII-only string possibly containing linesep
+        characters, and ending with a linesep character.  The string includes
+        the header name and the ': ' separator.
+
+        """
+        # At some point we need to only put fws here if it was in the source.
+        header = parser.Header([
+            parser.HeaderLabel([
+                parser.ValueTerminal(self.name, 'header-name'),
+                parser.ValueTerminal(':', 'header-sep')]),
+            parser.CFWSList([parser.WhiteSpaceTerminal(' ', 'fws')]),
+                             self._parse_tree])
+        return header.fold(policy=policy)
+
+
+def _reconstruct_header(cls_name, bases, value):
+    return type(cls_name, bases, {})._reconstruct(value)
+
+
+class UnstructuredHeader:
+
+    max_count = None
+    value_parser = staticmethod(parser.get_unstructured)
+
+    @classmethod
+    def parse(cls, value, kwds):
+        kwds['parse_tree'] = cls.value_parser(value)
+        kwds['decoded'] = str(kwds['parse_tree'])
+
+
+class UniqueUnstructuredHeader(UnstructuredHeader):
+
+    max_count = 1
+
+
+class DateHeader:
+
+    """Header whose value consists of a single timestamp.
+
+    Provides an additional attribute, datetime, which is either an aware
+    datetime using a timezone, or a naive datetime if the timezone
+    in the input string is -0000.  Also accepts a datetime as input.
+    The 'value' attribute is the normalized form of the timestamp,
+    which means it is the output of format_datetime on the datetime.
+    """
+
+    max_count = None
+
+    # This is used only for folding, not for creating 'decoded'.
+    value_parser = staticmethod(parser.get_unstructured)
+
+    @classmethod
+    def parse(cls, value, kwds):
+        if not value:
+            kwds['defects'].append(errors.HeaderMissingRequiredValue())
+            kwds['datetime'] = None
+            kwds['decoded'] = ''
+            kwds['parse_tree'] = parser.TokenList()
+            return
+        if isinstance(value, str):
+            value = utils.parsedate_to_datetime(value)
+        kwds['datetime'] = value
+        kwds['decoded'] = utils.format_datetime(kwds['datetime'])
+        kwds['parse_tree'] = cls.value_parser(kwds['decoded'])
+
+    def init(self, *args, **kw):
+        self._datetime = kw.pop('datetime')
+        super().init(*args, **kw)
+
+    @property
+    def datetime(self):
+        return self._datetime
+
+
+class UniqueDateHeader(DateHeader):
+
+    max_count = 1
+
+
+class AddressHeader:
+
+    max_count = None
+
+    @staticmethod
+    def value_parser(value):
+        address_list, value = parser.get_address_list(value)
+        assert not value, 'this should not happen'
+        return address_list
+
+    @classmethod
+    def parse(cls, value, kwds):
+        if isinstance(value, str):
+            # We are translating here from the RFC language (address/mailbox)
+            # to our API language (group/address).
+            kwds['parse_tree'] = address_list = cls.value_parser(value)
+            groups = []
+            for addr in address_list.addresses:
+                groups.append(Group(addr.display_name,
+                                    [Address(mb.display_name or '',
+                                             mb.local_part or '',
+                                             mb.domain or '')
+                                     for mb in addr.all_mailboxes]))
+            defects = list(address_list.all_defects)
+        else:
+            # Assume it is Address/Group stuff
+            if not hasattr(value, '__iter__'):
+                value = [value]
+            groups = [Group(None, [item]) if not hasattr(item, 'addresses')
+                                          else item
+                                    for item in value]
+            defects = []
+        kwds['groups'] = groups
+        kwds['defects'] = defects
+        kwds['decoded'] = ', '.join([str(item) for item in groups])
+        if 'parse_tree' not in kwds:
+            kwds['parse_tree'] = cls.value_parser(kwds['decoded'])
+
+    def init(self, *args, **kw):
+        self._groups = tuple(kw.pop('groups'))
+        self._addresses = None
+        super().init(*args, **kw)
+
+    @property
+    def groups(self):
+        return self._groups
+
+    @property
+    def addresses(self):
+        if self._addresses is None:
+            self._addresses = tuple([address for group in self._groups
+                                             for address in group.addresses])
+        return self._addresses
+
+
+class UniqueAddressHeader(AddressHeader):
+
+    max_count = 1
+
+
+class SingleAddressHeader(AddressHeader):
+
+    @property
+    def address(self):
+        if len(self.addresses)!=1:
+            raise ValueError(("value of single address header {} is not "
+                "a single address").format(self.name))
+        return self.addresses[0]
+
+
+class UniqueSingleAddressHeader(SingleAddressHeader):
+
+    max_count = 1
+
+
+class MIMEVersionHeader:
+
+    max_count = 1
+
+    value_parser = staticmethod(parser.parse_mime_version)
+
+    @classmethod
+    def parse(cls, value, kwds):
+        kwds['parse_tree'] = parse_tree = cls.value_parser(value)
+        kwds['decoded'] = str(parse_tree)
+        kwds['defects'].extend(parse_tree.all_defects)
+        kwds['major'] = None if parse_tree.minor is None else parse_tree.major
+        kwds['minor'] = parse_tree.minor
+        if parse_tree.minor is not None:
+            kwds['version'] = '{}.{}'.format(kwds['major'], kwds['minor'])
+        else:
+            kwds['version'] = None
+
+    def init(self, *args, **kw):
+        self._version = kw.pop('version')
+        self._major = kw.pop('major')
+        self._minor = kw.pop('minor')
+        super().init(*args, **kw)
+
+    @property
+    def major(self):
+        return self._major
+
+    @property
+    def minor(self):
+        return self._minor
+
+    @property
+    def version(self):
+        return self._version
+
+
+class ParameterizedMIMEHeader:
+
+    # Mixin that handles the params dict.  Must be subclassed and
+    # a property value_parser for the specific header provided.
+
+    max_count = 1
+
+    @classmethod
+    def parse(cls, value, kwds):
+        kwds['parse_tree'] = parse_tree = cls.value_parser(value)
+        kwds['decoded'] = str(parse_tree)
+        kwds['defects'].extend(parse_tree.all_defects)
+        if parse_tree.params is None:
+            kwds['params'] = {}
+        else:
+            # The MIME RFCs specify that parameter ordering is arbitrary.
+            kwds['params'] = {utils._sanitize(name).lower():
+                                    utils._sanitize(value)
+                               for name, value in parse_tree.params}
+
+    def init(self, *args, **kw):
+        self._params = kw.pop('params')
+        super().init(*args, **kw)
+
+    @property
+    def params(self):
+        return MappingProxyType(self._params)
+
+
+class ContentTypeHeader(ParameterizedMIMEHeader):
+
+    value_parser = staticmethod(parser.parse_content_type_header)
+
+    def init(self, *args, **kw):
+        super().init(*args, **kw)
+        self._maintype = utils._sanitize(self._parse_tree.maintype)
+        self._subtype = utils._sanitize(self._parse_tree.subtype)
+
+    @property
+    def maintype(self):
+        return self._maintype
+
+    @property
+    def subtype(self):
+        return self._subtype
+
+    @property
+    def content_type(self):
+        return self.maintype + '/' + self.subtype
+
+
+class ContentDispositionHeader(ParameterizedMIMEHeader):
+
+    value_parser = staticmethod(parser.parse_content_disposition_header)
+
+    def init(self, *args, **kw):
+        super().init(*args, **kw)
+        cd = self._parse_tree.content_disposition
+        self._content_disposition = cd if cd is None else utils._sanitize(cd)
+
+    @property
+    def content_disposition(self):
+        return self._content_disposition
+
+
+class ContentTransferEncodingHeader:
+
+    max_count = 1
+
+    value_parser = staticmethod(parser.parse_content_transfer_encoding_header)
+
+    @classmethod
+    def parse(cls, value, kwds):
+        kwds['parse_tree'] = parse_tree = cls.value_parser(value)
+        kwds['decoded'] = str(parse_tree)
+        kwds['defects'].extend(parse_tree.all_defects)
+
+    def init(self, *args, **kw):
+        super().init(*args, **kw)
+        self._cte = utils._sanitize(self._parse_tree.cte)
+
+    @property
+    def cte(self):
+        return self._cte
+
+
+# The header factory #
+
+_default_header_map = {
+    'subject':                      UniqueUnstructuredHeader,
+    'date':                         UniqueDateHeader,
+    'resent-date':                  DateHeader,
+    'orig-date':                    UniqueDateHeader,
+    'sender':                       UniqueSingleAddressHeader,
+    'resent-sender':                SingleAddressHeader,
+    'to':                           UniqueAddressHeader,
+    'resent-to':                    AddressHeader,
+    'cc':                           UniqueAddressHeader,
+    'resent-cc':                    AddressHeader,
+    'bcc':                          UniqueAddressHeader,
+    'resent-bcc':                   AddressHeader,
+    'from':                         UniqueAddressHeader,
+    'resent-from':                  AddressHeader,
+    'reply-to':                     UniqueAddressHeader,
+    'mime-version':                 MIMEVersionHeader,
+    'content-type':                 ContentTypeHeader,
+    'content-disposition':          ContentDispositionHeader,
+    'content-transfer-encoding':    ContentTransferEncodingHeader,
+    }
+
+class HeaderRegistry:
+
+    """A header_factory and header registry."""
+
+    def __init__(self, base_class=BaseHeader, default_class=UnstructuredHeader,
+                       use_default_map=True):
+        """Create a header_factory that works with the Policy API.
+
+        base_class is the class that will be the last class in the created
+        header class's __bases__ list.  default_class is the class that will be
+        used if "name" (see __call__) does not appear in the registry.
+        use_default_map controls whether or not the default mapping of names to
+        specialized classes is copied in to the registry when the factory is
+        created.  The default is True.
+
+        """
+        self.registry = {}
+        self.base_class = base_class
+        self.default_class = default_class
+        if use_default_map:
+            self.registry.update(_default_header_map)
+
+    def map_to_type(self, name, cls):
+        """Register cls as the specialized class for handling "name" headers.
+
+        """
+        self.registry[name.lower()] = cls
+
+    def __getitem__(self, name):
+        cls = self.registry.get(name.lower(), self.default_class)
+        return type('_'+cls.__name__, (cls, self.base_class), {})
+
+    def __call__(self, name, value):
+        """Create a header instance for header 'name' from 'value'.
+
+        Creates a header instance by creating a specialized class for parsing
+        and representing the specified header by combining the factory
+        base_class with a specialized class from the registry or the
+        default_class, and passing the name and value to the constructed
+        class's constructor.
+
+        """
+        return self[name](name, value)
diff --git a/Lib/email/iterators.py b/Lib/email/iterators.py
new file mode 100644
index 000000000..b5502ee97
--- /dev/null
+++ b/Lib/email/iterators.py
@@ -0,0 +1,71 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Various types of useful iterators and generators."""
+
+__all__ = [
+    'body_line_iterator',
+    'typed_subpart_iterator',
+    'walk',
+    # Do not include _structure() since it's part of the debugging API.
+    ]
+
+import sys
+from io import StringIO
+
+
+
+# This function will become a method of the Message class
+def walk(self):
+    """Walk over the message tree, yielding each subpart.
+
+    The walk is performed in depth-first order.  This method is a
+    generator.
+    """
+    yield self
+    if self.is_multipart():
+        for subpart in self.get_payload():
+            yield from subpart.walk()
+
+
+
+# These two functions are imported into the Iterators.py interface module.
+def body_line_iterator(msg, decode=False):
+    """Iterate over the parts, returning string payloads line-by-line.
+
+    Optional decode (default False) is passed through to .get_payload().
+    """
+    for subpart in msg.walk():
+        payload = subpart.get_payload(decode=decode)
+        if isinstance(payload, str):
+            yield from StringIO(payload)
+
+
+def typed_subpart_iterator(msg, maintype='text', subtype=None):
+    """Iterate over the subparts with a given MIME type.
+
+    Use `maintype' as the main MIME type to match against; this defaults to
+    "text".  Optional `subtype' is the MIME subtype to match against; if
+    omitted, only the main type is matched.
+    """
+    for subpart in msg.walk():
+        if subpart.get_content_maintype() == maintype:
+            if subtype is None or subpart.get_content_subtype() == subtype:
+                yield subpart
+
+
+
+def _structure(msg, fp=None, level=0, include_default=False):
+    """A handy debugging aid"""
+    if fp is None:
+        fp = sys.stdout
+    tab = ' ' * (level * 4)
+    print(tab + msg.get_content_type(), end='', file=fp)
+    if include_default:
+        print(' [%s]' % msg.get_default_type(), file=fp)
+    else:
+        print(file=fp)
+    if msg.is_multipart():
+        for subpart in msg.get_payload():
+            _structure(subpart, fp, level+1, include_default)
diff --git a/Lib/email/message.py b/Lib/email/message.py
new file mode 100644
index 000000000..b6512f219
--- /dev/null
+++ b/Lib/email/message.py
@@ -0,0 +1,1164 @@
+# Copyright (C) 2001-2007 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Basic message object for the email package object model."""
+
+__all__ = ['Message', 'EmailMessage']
+
+import re
+import uu
+import quopri
+from io import BytesIO, StringIO
+
+# Intrapackage imports
+from email import utils
+from email import errors
+from email._policybase import Policy, compat32
+from email import charset as _charset
+from email._encoded_words import decode_b
+Charset = _charset.Charset
+
+SEMISPACE = '; '
+
+# Regular expression that matches `special' characters in parameters, the
+# existence of which force quoting of the parameter value.
+tspecials = re.compile(r'[ \(\)<>@,;:\\"/\[\]\?=]')
+
+
+def _splitparam(param):
+    # Split header parameters.  BAW: this may be too simple.  It isn't
+    # strictly RFC 2045 (section 5.1) compliant, but it catches most headers
+    # found in the wild.  We may eventually need a full fledged parser.
+    # RDM: we might have a Header here; for now just stringify it.
+    a, sep, b = str(param).partition(';')
+    if not sep:
+        return a.strip(), None
+    return a.strip(), b.strip()
+
+def _formatparam(param, value=None, quote=True):
+    """Convenience function to format and return a key=value pair.
+
+    This will quote the value if needed or if quote is true.  If value is a
+    three tuple (charset, language, value), it will be encoded according
+    to RFC2231 rules.  If it contains non-ascii characters it will likewise
+    be encoded according to RFC2231 rules, using the utf-8 charset and
+    a null language.
+    """
+    if value is not None and len(value) > 0:
+        # A tuple is used for RFC 2231 encoded parameter values where items
+        # are (charset, language, value).  charset is a string, not a Charset
+        # instance.  RFC 2231 encoded values are never quoted, per RFC.
+        if isinstance(value, tuple):
+            # Encode as per RFC 2231
+            param += '*'
+            value = utils.encode_rfc2231(value[2], value[0], value[1])
+            return '%s=%s' % (param, value)
+        else:
+            try:
+                value.encode('ascii')
+            except UnicodeEncodeError:
+                param += '*'
+                value = utils.encode_rfc2231(value, 'utf-8', '')
+                return '%s=%s' % (param, value)
+        # BAW: Please check this.  I think that if quote is set it should
+        # force quoting even if not necessary.
+        if quote or tspecials.search(value):
+            return '%s="%s"' % (param, utils.quote(value))
+        else:
+            return '%s=%s' % (param, value)
+    else:
+        return param
+
+def _parseparam(s):
+    # RDM This might be a Header, so for now stringify it.
+    s = ';' + str(s)
+    plist = []
+    while s[:1] == ';':
+        s = s[1:]
+        end = s.find(';')
+        while end > 0 and (s.count('"', 0, end) - s.count('\\"', 0, end)) % 2:
+            end = s.find(';', end + 1)
+        if end < 0:
+            end = len(s)
+        f = s[:end]
+        if '=' in f:
+            i = f.index('=')
+            f = f[:i].strip().lower() + '=' + f[i+1:].strip()
+        plist.append(f.strip())
+        s = s[end:]
+    return plist
+
+
+def _unquotevalue(value):
+    # This is different than utils.collapse_rfc2231_value() because it doesn't
+    # try to convert the value to a unicode.  Message.get_param() and
+    # Message.get_params() are both currently defined to return the tuple in
+    # the face of RFC 2231 parameters.
+    if isinstance(value, tuple):
+        return value[0], value[1], utils.unquote(value[2])
+    else:
+        return utils.unquote(value)
+
+
+
+class Message:
+    """Basic message object.
+
+    A message object is defined as something that has a bunch of RFC 2822
+    headers and a payload.  It may optionally have an envelope header
+    (a.k.a. Unix-From or From_ header).  If the message is a container (i.e. a
+    multipart or a message/rfc822), then the payload is a list of Message
+    objects, otherwise it is a string.
+
+    Message objects implement part of the `mapping' interface, which assumes
+    there is exactly one occurrence of the header per message.  Some headers
+    do in fact appear multiple times (e.g. Received) and for those headers,
+    you must use the explicit API to set or get all the headers.  Not all of
+    the mapping methods are implemented.
+    """
+    def __init__(self, policy=compat32):
+        self.policy = policy
+        self._headers = []
+        self._unixfrom = None
+        self._payload = None
+        self._charset = None
+        # Defaults for multipart messages
+        self.preamble = self.epilogue = None
+        self.defects = []
+        # Default content type
+        self._default_type = 'text/plain'
+
+    def __str__(self):
+        """Return the entire formatted message as a string.
+        """
+        return self.as_string()
+
+    def as_string(self, unixfrom=False, maxheaderlen=0, policy=None):
+        """Return the entire formatted message as a string.
+
+        Optional 'unixfrom', when true, means include the Unix From_ envelope
+        header.  For backward compatibility reasons, if maxheaderlen is
+        not specified it defaults to 0, so you must override it explicitly
+        if you want a different maxheaderlen.  'policy' is passed to the
+        Generator instance used to serialize the mesasge; if it is not
+        specified the policy associated with the message instance is used.
+
+        If the message object contains binary data that is not encoded
+        according to RFC standards, the non-compliant data will be replaced by
+        unicode "unknown character" code points.
+        """
+        from email.generator import Generator
+        policy = self.policy if policy is None else policy
+        fp = StringIO()
+        g = Generator(fp,
+                      mangle_from_=False,
+                      maxheaderlen=maxheaderlen,
+                      policy=policy)
+        g.flatten(self, unixfrom=unixfrom)
+        return fp.getvalue()
+
+    def __bytes__(self):
+        """Return the entire formatted message as a bytes object.
+        """
+        return self.as_bytes()
+
+    def as_bytes(self, unixfrom=False, policy=None):
+        """Return the entire formatted message as a bytes object.
+
+        Optional 'unixfrom', when true, means include the Unix From_ envelope
+        header.  'policy' is passed to the BytesGenerator instance used to
+        serialize the message; if not specified the policy associated with
+        the message instance is used.
+        """
+        from email.generator import BytesGenerator
+        policy = self.policy if policy is None else policy
+        fp = BytesIO()
+        g = BytesGenerator(fp, mangle_from_=False, policy=policy)
+        g.flatten(self, unixfrom=unixfrom)
+        return fp.getvalue()
+
+    def is_multipart(self):
+        """Return True if the message consists of multiple parts."""
+        return isinstance(self._payload, list)
+
+    #
+    # Unix From_ line
+    #
+    def set_unixfrom(self, unixfrom):
+        self._unixfrom = unixfrom
+
+    def get_unixfrom(self):
+        return self._unixfrom
+
+    #
+    # Payload manipulation.
+    #
+    def attach(self, payload):
+        """Add the given payload to the current payload.
+
+        The current payload will always be a list of objects after this method
+        is called.  If you want to set the payload to a scalar object, use
+        set_payload() instead.
+        """
+        if self._payload is None:
+            self._payload = [payload]
+        else:
+            try:
+                self._payload.append(payload)
+            except AttributeError:
+                raise TypeError("Attach is not valid on a message with a"
+                                " non-multipart payload")
+
+    def get_payload(self, i=None, decode=False):
+        """Return a reference to the payload.
+
+        The payload will either be a list object or a string.  If you mutate
+        the list object, you modify the message's payload in place.  Optional
+        i returns that index into the payload.
+
+        Optional decode is a flag indicating whether the payload should be
+        decoded or not, according to the Content-Transfer-Encoding header
+        (default is False).
+
+        When True and the message is not a multipart, the payload will be
+        decoded if this header's value is `quoted-printable' or `base64'.  If
+        some other encoding is used, or the header is missing, or if the
+        payload has bogus data (i.e. bogus base64 or uuencoded data), the
+        payload is returned as-is.
+
+        If the message is a multipart and the decode flag is True, then None
+        is returned.
+        """
+        # Here is the logic table for this code, based on the email5.0.0 code:
+        #   i     decode  is_multipart  result
+        # ------  ------  ------------  ------------------------------
+        #  None   True    True          None
+        #   i     True    True          None
+        #  None   False   True          _payload (a list)
+        #   i     False   True          _payload element i (a Message)
+        #   i     False   False         error (not a list)
+        #   i     True    False         error (not a list)
+        #  None   False   False         _payload
+        #  None   True    False         _payload decoded (bytes)
+        # Note that Barry planned to factor out the 'decode' case, but that
+        # isn't so easy now that we handle the 8 bit data, which needs to be
+        # converted in both the decode and non-decode path.
+        if self.is_multipart():
+            if decode:
+                return None
+            if i is None:
+                return self._payload
+            else:
+                return self._payload[i]
+        # For backward compatibility, Use isinstance and this error message
+        # instead of the more logical is_multipart test.
+        if i is not None and not isinstance(self._payload, list):
+            raise TypeError('Expected list, got %s' % type(self._payload))
+        payload = self._payload
+        # cte might be a Header, so for now stringify it.
+        cte = str(self.get('content-transfer-encoding', '')).lower()
+        # payload may be bytes here.
+        if isinstance(payload, str):
+            if utils._has_surrogates(payload):
+                bpayload = payload.encode('ascii', 'surrogateescape')
+                if not decode:
+                    try:
+                        payload = bpayload.decode(self.get_param('charset', 'ascii'), 'replace')
+                    except LookupError:
+                        payload = bpayload.decode('ascii', 'replace')
+            elif decode:
+                try:
+                    bpayload = payload.encode('ascii')
+                except UnicodeError:
+                    # This won't happen for RFC compliant messages (messages
+                    # containing only ASCII code points in the unicode input).
+                    # If it does happen, turn the string into bytes in a way
+                    # guaranteed not to fail.
+                    bpayload = payload.encode('raw-unicode-escape')
+        if not decode:
+            return payload
+        if cte == 'quoted-printable':
+            return quopri.decodestring(bpayload)
+        elif cte == 'base64':
+            # XXX: this is a bit of a hack; decode_b should probably be factored
+            # out somewhere, but I haven't figured out where yet.
+            value, defects = decode_b(b''.join(bpayload.splitlines()))
+            for defect in defects:
+                self.policy.handle_defect(self, defect)
+            return value
+        elif cte in ('x-uuencode', 'uuencode', 'uue', 'x-uue'):
+            in_file = BytesIO(bpayload)
+            out_file = BytesIO()
+            try:
+                uu.decode(in_file, out_file, quiet=True)
+                return out_file.getvalue()
+            except uu.Error:
+                # Some decoding problem
+                return bpayload
+        if isinstance(payload, str):
+            return bpayload
+        return payload
+
+    def set_payload(self, payload, charset=None):
+        """Set the payload to the given value.
+
+        Optional charset sets the message's default character set.  See
+        set_charset() for details.
+        """
+        if hasattr(payload, 'encode'):
+            if charset is None:
+                self._payload = payload
+                return
+            if not isinstance(charset, Charset):
+                charset = Charset(charset)
+            payload = payload.encode(charset.output_charset)
+        if hasattr(payload, 'decode'):
+            self._payload = payload.decode('ascii', 'surrogateescape')
+        else:
+            self._payload = payload
+        if charset is not None:
+            self.set_charset(charset)
+
+    def set_charset(self, charset):
+        """Set the charset of the payload to a given character set.
+
+        charset can be a Charset instance, a string naming a character set, or
+        None.  If it is a string it will be converted to a Charset instance.
+        If charset is None, the charset parameter will be removed from the
+        Content-Type field.  Anything else will generate a TypeError.
+
+        The message will be assumed to be of type text/* encoded with
+        charset.input_charset.  It will be converted to charset.output_charset
+        and encoded properly, if needed, when generating the plain text
+        representation of the message.  MIME headers (MIME-Version,
+        Content-Type, Content-Transfer-Encoding) will be added as needed.
+        """
+        if charset is None:
+            self.del_param('charset')
+            self._charset = None
+            return
+        if not isinstance(charset, Charset):
+            charset = Charset(charset)
+        self._charset = charset
+        if 'MIME-Version' not in self:
+            self.add_header('MIME-Version', '1.0')
+        if 'Content-Type' not in self:
+            self.add_header('Content-Type', 'text/plain',
+                            charset=charset.get_output_charset())
+        else:
+            self.set_param('charset', charset.get_output_charset())
+        if charset != charset.get_output_charset():
+            self._payload = charset.body_encode(self._payload)
+        if 'Content-Transfer-Encoding' not in self:
+            cte = charset.get_body_encoding()
+            try:
+                cte(self)
+            except TypeError:
+                # This 'if' is for backward compatibility, it allows unicode
+                # through even though that won't work correctly if the
+                # message is serialized.
+                payload = self._payload
+                if payload:
+                    try:
+                        payload = payload.encode('ascii', 'surrogateescape')
+                    except UnicodeError:
+                        payload = payload.encode(charset.output_charset)
+                self._payload = charset.body_encode(payload)
+                self.add_header('Content-Transfer-Encoding', cte)
+
+    def get_charset(self):
+        """Return the Charset instance associated with the message's payload.
+        """
+        return self._charset
+
+    #
+    # MAPPING INTERFACE (partial)
+    #
+    def __len__(self):
+        """Return the total number of headers, including duplicates."""
+        return len(self._headers)
+
+    def __getitem__(self, name):
+        """Get a header value.
+
+        Return None if the header is missing instead of raising an exception.
+
+        Note that if the header appeared multiple times, exactly which
+        occurrence gets returned is undefined.  Use get_all() to get all
+        the values matching a header field name.
+        """
+        return self.get(name)
+
+    def __setitem__(self, name, val):
+        """Set the value of a header.
+
+        Note: this does not overwrite an existing header with the same field
+        name.  Use __delitem__() first to delete any existing headers.
+        """
+        max_count = self.policy.header_max_count(name)
+        if max_count:
+            lname = name.lower()
+            found = 0
+            for k, v in self._headers:
+                if k.lower() == lname:
+                    found += 1
+                    if found >= max_count:
+                        raise ValueError("There may be at most {} {} headers "
+                                         "in a message".format(max_count, name))
+        self._headers.append(self.policy.header_store_parse(name, val))
+
+    def __delitem__(self, name):
+        """Delete all occurrences of a header, if present.
+
+        Does not raise an exception if the header is missing.
+        """
+        name = name.lower()
+        newheaders = []
+        for k, v in self._headers:
+            if k.lower() != name:
+                newheaders.append((k, v))
+        self._headers = newheaders
+
+    def __contains__(self, name):
+        return name.lower() in [k.lower() for k, v in self._headers]
+
+    def __iter__(self):
+        for field, value in self._headers:
+            yield field
+
+    def keys(self):
+        """Return a list of all the message's header field names.
+
+        These will be sorted in the order they appeared in the original
+        message, or were added to the message, and may contain duplicates.
+        Any fields deleted and re-inserted are always appended to the header
+        list.
+        """
+        return [k for k, v in self._headers]
+
+    def values(self):
+        """Return a list of all the message's header values.
+
+        These will be sorted in the order they appeared in the original
+        message, or were added to the message, and may contain duplicates.
+        Any fields deleted and re-inserted are always appended to the header
+        list.
+        """
+        return [self.policy.header_fetch_parse(k, v)
+                for k, v in self._headers]
+
+    def items(self):
+        """Get all the message's header fields and values.
+
+        These will be sorted in the order they appeared in the original
+        message, or were added to the message, and may contain duplicates.
+        Any fields deleted and re-inserted are always appended to the header
+        list.
+        """
+        return [(k, self.policy.header_fetch_parse(k, v))
+                for k, v in self._headers]
+
+    def get(self, name, failobj=None):
+        """Get a header value.
+
+        Like __getitem__() but return failobj instead of None when the field
+        is missing.
+        """
+        name = name.lower()
+        for k, v in self._headers:
+            if k.lower() == name:
+                return self.policy.header_fetch_parse(k, v)
+        return failobj
+
+    #
+    # "Internal" methods (public API, but only intended for use by a parser
+    # or generator, not normal application code.
+    #
+
+    def set_raw(self, name, value):
+        """Store name and value in the model without modification.
+
+        This is an "internal" API, intended only for use by a parser.
+        """
+        self._headers.append((name, value))
+
+    def raw_items(self):
+        """Return the (name, value) header pairs without modification.
+
+        This is an "internal" API, intended only for use by a generator.
+        """
+        return iter(self._headers.copy())
+
+    #
+    # Additional useful stuff
+    #
+
+    def get_all(self, name, failobj=None):
+        """Return a list of all the values for the named field.
+
+        These will be sorted in the order they appeared in the original
+        message, and may contain duplicates.  Any fields deleted and
+        re-inserted are always appended to the header list.
+
+        If no such fields exist, failobj is returned (defaults to None).
+        """
+        values = []
+        name = name.lower()
+        for k, v in self._headers:
+            if k.lower() == name:
+                values.append(self.policy.header_fetch_parse(k, v))
+        if not values:
+            return failobj
+        return values
+
+    def add_header(self, _name, _value, **_params):
+        """Extended header setting.
+
+        name is the header field to add.  keyword arguments can be used to set
+        additional parameters for the header field, with underscores converted
+        to dashes.  Normally the parameter will be added as key="value" unless
+        value is None, in which case only the key will be added.  If a
+        parameter value contains non-ASCII characters it can be specified as a
+        three-tuple of (charset, language, value), in which case it will be
+        encoded according to RFC2231 rules.  Otherwise it will be encoded using
+        the utf-8 charset and a language of ''.
+
+        Examples:
+
+        msg.add_header('content-disposition', 'attachment', filename='bud.gif')
+        msg.add_header('content-disposition', 'attachment',
+                       filename=('utf-8', '', Fußballer.ppt'))
+        msg.add_header('content-disposition', 'attachment',
+                       filename='Fußballer.ppt'))
+        """
+        parts = []
+        for k, v in _params.items():
+            if v is None:
+                parts.append(k.replace('_', '-'))
+            else:
+                parts.append(_formatparam(k.replace('_', '-'), v))
+        if _value is not None:
+            parts.insert(0, _value)
+        self[_name] = SEMISPACE.join(parts)
+
+    def replace_header(self, _name, _value):
+        """Replace a header.
+
+        Replace the first matching header found in the message, retaining
+        header order and case.  If no matching header was found, a KeyError is
+        raised.
+        """
+        _name = _name.lower()
+        for i, (k, v) in zip(range(len(self._headers)), self._headers):
+            if k.lower() == _name:
+                self._headers[i] = self.policy.header_store_parse(k, _value)
+                break
+        else:
+            raise KeyError(_name)
+
+    #
+    # Use these three methods instead of the three above.
+    #
+
+    def get_content_type(self):
+        """Return the message's content type.
+
+        The returned string is coerced to lower case of the form
+        `maintype/subtype'.  If there was no Content-Type header in the
+        message, the default type as given by get_default_type() will be
+        returned.  Since according to RFC 2045, messages always have a default
+        type this will always return a value.
+
+        RFC 2045 defines a message's default type to be text/plain unless it
+        appears inside a multipart/digest container, in which case it would be
+        message/rfc822.
+        """
+        missing = object()
+        value = self.get('content-type', missing)
+        if value is missing:
+            # This should have no parameters
+            return self.get_default_type()
+        ctype = _splitparam(value)[0].lower()
+        # RFC 2045, section 5.2 says if its invalid, use text/plain
+        if ctype.count('/') != 1:
+            return 'text/plain'
+        return ctype
+
+    def get_content_maintype(self):
+        """Return the message's main content type.
+
+        This is the `maintype' part of the string returned by
+        get_content_type().
+        """
+        ctype = self.get_content_type()
+        return ctype.split('/')[0]
+
+    def get_content_subtype(self):
+        """Returns the message's sub-content type.
+
+        This is the `subtype' part of the string returned by
+        get_content_type().
+        """
+        ctype = self.get_content_type()
+        return ctype.split('/')[1]
+
+    def get_default_type(self):
+        """Return the `default' content type.
+
+        Most messages have a default content type of text/plain, except for
+        messages that are subparts of multipart/digest containers.  Such
+        subparts have a default content type of message/rfc822.
+        """
+        return self._default_type
+
+    def set_default_type(self, ctype):
+        """Set the `default' content type.
+
+        ctype should be either "text/plain" or "message/rfc822", although this
+        is not enforced.  The default content type is not stored in the
+        Content-Type header.
+        """
+        self._default_type = ctype
+
+    def _get_params_preserve(self, failobj, header):
+        # Like get_params() but preserves the quoting of values.  BAW:
+        # should this be part of the public interface?
+        missing = object()
+        value = self.get(header, missing)
+        if value is missing:
+            return failobj
+        params = []
+        for p in _parseparam(value):
+            try:
+                name, val = p.split('=', 1)
+                name = name.strip()
+                val = val.strip()
+            except ValueError:
+                # Must have been a bare attribute
+                name = p.strip()
+                val = ''
+            params.append((name, val))
+        params = utils.decode_params(params)
+        return params
+
+    def get_params(self, failobj=None, header='content-type', unquote=True):
+        """Return the message's Content-Type parameters, as a list.
+
+        The elements of the returned list are 2-tuples of key/value pairs, as
+        split on the `=' sign.  The left hand side of the `=' is the key,
+        while the right hand side is the value.  If there is no `=' sign in
+        the parameter the value is the empty string.  The value is as
+        described in the get_param() method.
+
+        Optional failobj is the object to return if there is no Content-Type
+        header.  Optional header is the header to search instead of
+        Content-Type.  If unquote is True, the value is unquoted.
+        """
+        missing = object()
+        params = self._get_params_preserve(missing, header)
+        if params is missing:
+            return failobj
+        if unquote:
+            return [(k, _unquotevalue(v)) for k, v in params]
+        else:
+            return params
+
+    def get_param(self, param, failobj=None, header='content-type',
+                  unquote=True):
+        """Return the parameter value if found in the Content-Type header.
+
+        Optional failobj is the object to return if there is no Content-Type
+        header, or the Content-Type header has no such parameter.  Optional
+        header is the header to search instead of Content-Type.
+
+        Parameter keys are always compared case insensitively.  The return
+        value can either be a string, or a 3-tuple if the parameter was RFC
+        2231 encoded.  When it's a 3-tuple, the elements of the value are of
+        the form (CHARSET, LANGUAGE, VALUE).  Note that both CHARSET and
+        LANGUAGE can be None, in which case you should consider VALUE to be
+        encoded in the us-ascii charset.  You can usually ignore LANGUAGE.
+        The parameter value (either the returned string, or the VALUE item in
+        the 3-tuple) is always unquoted, unless unquote is set to False.
+
+        If your application doesn't care whether the parameter was RFC 2231
+        encoded, it can turn the return value into a string as follows:
+
+            rawparam = msg.get_param('foo')
+            param = email.utils.collapse_rfc2231_value(rawparam)
+
+        """
+        if header not in self:
+            return failobj
+        for k, v in self._get_params_preserve(failobj, header):
+            if k.lower() == param.lower():
+                if unquote:
+                    return _unquotevalue(v)
+                else:
+                    return v
+        return failobj
+
+    def set_param(self, param, value, header='Content-Type', requote=True,
+                  charset=None, language='', replace=False):
+        """Set a parameter in the Content-Type header.
+
+        If the parameter already exists in the header, its value will be
+        replaced with the new value.
+
+        If header is Content-Type and has not yet been defined for this
+        message, it will be set to "text/plain" and the new parameter and
+        value will be appended as per RFC 2045.
+
+        An alternate header can be specified in the header argument, and all
+        parameters will be quoted as necessary unless requote is False.
+
+        If charset is specified, the parameter will be encoded according to RFC
+        2231.  Optional language specifies the RFC 2231 language, defaulting
+        to the empty string.  Both charset and language should be strings.
+        """
+        if not isinstance(value, tuple) and charset:
+            value = (charset, language, value)
+
+        if header not in self and header.lower() == 'content-type':
+            ctype = 'text/plain'
+        else:
+            ctype = self.get(header)
+        if not self.get_param(param, header=header):
+            if not ctype:
+                ctype = _formatparam(param, value, requote)
+            else:
+                ctype = SEMISPACE.join(
+                    [ctype, _formatparam(param, value, requote)])
+        else:
+            ctype = ''
+            for old_param, old_value in self.get_params(header=header,
+                                                        unquote=requote):
+                append_param = ''
+                if old_param.lower() == param.lower():
+                    append_param = _formatparam(param, value, requote)
+                else:
+                    append_param = _formatparam(old_param, old_value, requote)
+                if not ctype:
+                    ctype = append_param
+                else:
+                    ctype = SEMISPACE.join([ctype, append_param])
+        if ctype != self.get(header):
+            if replace:
+                self.replace_header(header, ctype)
+            else:
+                del self[header]
+                self[header] = ctype
+
+    def del_param(self, param, header='content-type', requote=True):
+        """Remove the given parameter completely from the Content-Type header.
+
+        The header will be re-written in place without the parameter or its
+        value. All values will be quoted as necessary unless requote is
+        False.  Optional header specifies an alternative to the Content-Type
+        header.
+        """
+        if header not in self:
+            return
+        new_ctype = ''
+        for p, v in self.get_params(header=header, unquote=requote):
+            if p.lower() != param.lower():
+                if not new_ctype:
+                    new_ctype = _formatparam(p, v, requote)
+                else:
+                    new_ctype = SEMISPACE.join([new_ctype,
+                                                _formatparam(p, v, requote)])
+        if new_ctype != self.get(header):
+            del self[header]
+            self[header] = new_ctype
+
+    def set_type(self, type, header='Content-Type', requote=True):
+        """Set the main type and subtype for the Content-Type header.
+
+        type must be a string in the form "maintype/subtype", otherwise a
+        ValueError is raised.
+
+        This method replaces the Content-Type header, keeping all the
+        parameters in place.  If requote is False, this leaves the existing
+        header's quoting as is.  Otherwise, the parameters will be quoted (the
+        default).
+
+        An alternative header can be specified in the header argument.  When
+        the Content-Type header is set, we'll always also add a MIME-Version
+        header.
+        """
+        # BAW: should we be strict?
+        if not type.count('/') == 1:
+            raise ValueError
+        # Set the Content-Type, you get a MIME-Version
+        if header.lower() == 'content-type':
+            del self['mime-version']
+            self['MIME-Version'] = '1.0'
+        if header not in self:
+            self[header] = type
+            return
+        params = self.get_params(header=header, unquote=requote)
+        del self[header]
+        self[header] = type
+        # Skip the first param; it's the old type.
+        for p, v in params[1:]:
+            self.set_param(p, v, header, requote)
+
+    def get_filename(self, failobj=None):
+        """Return the filename associated with the payload if present.
+
+        The filename is extracted from the Content-Disposition header's
+        `filename' parameter, and it is unquoted.  If that header is missing
+        the `filename' parameter, this method falls back to looking for the
+        `name' parameter.
+        """
+        missing = object()
+        filename = self.get_param('filename', missing, 'content-disposition')
+        if filename is missing:
+            filename = self.get_param('name', missing, 'content-type')
+        if filename is missing:
+            return failobj
+        return utils.collapse_rfc2231_value(filename).strip()
+
+    def get_boundary(self, failobj=None):
+        """Return the boundary associated with the payload if present.
+
+        The boundary is extracted from the Content-Type header's `boundary'
+        parameter, and it is unquoted.
+        """
+        missing = object()
+        boundary = self.get_param('boundary', missing)
+        if boundary is missing:
+            return failobj
+        # RFC 2046 says that boundaries may begin but not end in w/s
+        return utils.collapse_rfc2231_value(boundary).rstrip()
+
+    def set_boundary(self, boundary):
+        """Set the boundary parameter in Content-Type to 'boundary'.
+
+        This is subtly different than deleting the Content-Type header and
+        adding a new one with a new boundary parameter via add_header().  The
+        main difference is that using the set_boundary() method preserves the
+        order of the Content-Type header in the original message.
+
+        HeaderParseError is raised if the message has no Content-Type header.
+        """
+        missing = object()
+        params = self._get_params_preserve(missing, 'content-type')
+        if params is missing:
+            # There was no Content-Type header, and we don't know what type
+            # to set it to, so raise an exception.
+            raise errors.HeaderParseError('No Content-Type header found')
+        newparams = []
+        foundp = False
+        for pk, pv in params:
+            if pk.lower() == 'boundary':
+                newparams.append(('boundary', '"%s"' % boundary))
+                foundp = True
+            else:
+                newparams.append((pk, pv))
+        if not foundp:
+            # The original Content-Type header had no boundary attribute.
+            # Tack one on the end.  BAW: should we raise an exception
+            # instead???
+            newparams.append(('boundary', '"%s"' % boundary))
+        # Replace the existing Content-Type header with the new value
+        newheaders = []
+        for h, v in self._headers:
+            if h.lower() == 'content-type':
+                parts = []
+                for k, v in newparams:
+                    if v == '':
+                        parts.append(k)
+                    else:
+                        parts.append('%s=%s' % (k, v))
+                val = SEMISPACE.join(parts)
+                newheaders.append(self.policy.header_store_parse(h, val))
+
+            else:
+                newheaders.append((h, v))
+        self._headers = newheaders
+
+    def get_content_charset(self, failobj=None):
+        """Return the charset parameter of the Content-Type header.
+
+        The returned string is always coerced to lower case.  If there is no
+        Content-Type header, or if that header has no charset parameter,
+        failobj is returned.
+        """
+        missing = object()
+        charset = self.get_param('charset', missing)
+        if charset is missing:
+            return failobj
+        if isinstance(charset, tuple):
+            # RFC 2231 encoded, so decode it, and it better end up as ascii.
+            pcharset = charset[0] or 'us-ascii'
+            try:
+                # LookupError will be raised if the charset isn't known to
+                # Python.  UnicodeError will be raised if the encoded text
+                # contains a character not in the charset.
+                as_bytes = charset[2].encode('raw-unicode-escape')
+                charset = str(as_bytes, pcharset)
+            except (LookupError, UnicodeError):
+                charset = charset[2]
+        # charset characters must be in us-ascii range
+        try:
+            charset.encode('us-ascii')
+        except UnicodeError:
+            return failobj
+        # RFC 2046, $4.1.2 says charsets are not case sensitive
+        return charset.lower()
+
+    def get_charsets(self, failobj=None):
+        """Return a list containing the charset(s) used in this message.
+
+        The returned list of items describes the Content-Type headers'
+        charset parameter for this message and all the subparts in its
+        payload.
+
+        Each item will either be a string (the value of the charset parameter
+        in the Content-Type header of that part) or the value of the
+        'failobj' parameter (defaults to None), if the part does not have a
+        main MIME type of "text", or the charset is not defined.
+
+        The list will contain one string for each part of the message, plus
+        one for the container message (i.e. self), so that a non-multipart
+        message will still return a list of length 1.
+        """
+        return [part.get_content_charset(failobj) for part in self.walk()]
+
+    def get_content_disposition(self):
+        """Return the message's content-disposition if it exists, or None.
+
+        The return values can be either 'inline', 'attachment' or None
+        according to the rfc2183.
+        """
+        value = self.get('content-disposition')
+        if value is None:
+            return None
+        c_d = _splitparam(value)[0].lower()
+        return c_d
+
+    # I.e. def walk(self): ...
+    from email.iterators import walk
+
+
+class MIMEPart(Message):
+
+    def __init__(self, policy=None):
+        if policy is None:
+            from email.policy import default
+            policy = default
+        Message.__init__(self, policy)
+
+
+    def as_string(self, unixfrom=False, maxheaderlen=None, policy=None):
+        """Return the entire formatted message as a string.
+
+        Optional 'unixfrom', when true, means include the Unix From_ envelope
+        header.  maxheaderlen is retained for backward compatibility with the
+        base Message class, but defaults to None, meaning that the policy value
+        for max_line_length controls the header maximum length.  'policy' is
+        passed to the Generator instance used to serialize the mesasge; if it
+        is not specified the policy associated with the message instance is
+        used.
+        """
+        policy = self.policy if policy is None else policy
+        if maxheaderlen is None:
+            maxheaderlen = policy.max_line_length
+        return super().as_string(maxheaderlen=maxheaderlen, policy=policy)
+
+    def __str__(self):
+        return self.as_string(policy=self.policy.clone(utf8=True))
+
+    def is_attachment(self):
+        c_d = self.get('content-disposition')
+        return False if c_d is None else c_d.content_disposition == 'attachment'
+
+    def _find_body(self, part, preferencelist):
+        if part.is_attachment():
+            return
+        maintype, subtype = part.get_content_type().split('/')
+        if maintype == 'text':
+            if subtype in preferencelist:
+                yield (preferencelist.index(subtype), part)
+            return
+        if maintype != 'multipart':
+            return
+        if subtype != 'related':
+            for subpart in part.iter_parts():
+                yield from self._find_body(subpart, preferencelist)
+            return
+        if 'related' in preferencelist:
+            yield (preferencelist.index('related'), part)
+        candidate = None
+        start = part.get_param('start')
+        if start:
+            for subpart in part.iter_parts():
+                if subpart['content-id'] == start:
+                    candidate = subpart
+                    break
+        if candidate is None:
+            subparts = part.get_payload()
+            candidate = subparts[0] if subparts else None
+        if candidate is not None:
+            yield from self._find_body(candidate, preferencelist)
+
+    def get_body(self, preferencelist=('related', 'html', 'plain')):
+        """Return best candidate mime part for display as 'body' of message.
+
+        Do a depth first search, starting with self, looking for the first part
+        matching each of the items in preferencelist, and return the part
+        corresponding to the first item that has a match, or None if no items
+        have a match.  If 'related' is not included in preferencelist, consider
+        the root part of any multipart/related encountered as a candidate
+        match.  Ignore parts with 'Content-Disposition: attachment'.
+        """
+        best_prio = len(preferencelist)
+        body = None
+        for prio, part in self._find_body(self, preferencelist):
+            if prio < best_prio:
+                best_prio = prio
+                body = part
+                if prio == 0:
+                    break
+        return body
+
+    _body_types = {('text', 'plain'),
+                   ('text', 'html'),
+                   ('multipart', 'related'),
+                   ('multipart', 'alternative')}
+    def iter_attachments(self):
+        """Return an iterator over the non-main parts of a multipart.
+
+        Skip the first of each occurrence of text/plain, text/html,
+        multipart/related, or multipart/alternative in the multipart (unless
+        they have a 'Content-Disposition: attachment' header) and include all
+        remaining subparts in the returned iterator.  When applied to a
+        multipart/related, return all parts except the root part.  Return an
+        empty iterator when applied to a multipart/alternative or a
+        non-multipart.
+        """
+        maintype, subtype = self.get_content_type().split('/')
+        if maintype != 'multipart' or subtype == 'alternative':
+            return
+        parts = self.get_payload().copy()
+        if maintype == 'multipart' and subtype == 'related':
+            # For related, we treat everything but the root as an attachment.
+            # The root may be indicated by 'start'; if there's no start or we
+            # can't find the named start, treat the first subpart as the root.
+            start = self.get_param('start')
+            if start:
+                found = False
+                attachments = []
+                for part in parts:
+                    if part.get('content-id') == start:
+                        found = True
+                    else:
+                        attachments.append(part)
+                if found:
+                    yield from attachments
+                    return
+            parts.pop(0)
+            yield from parts
+            return
+        # Otherwise we more or less invert the remaining logic in get_body.
+        # This only really works in edge cases (ex: non-text related or
+        # alternatives) if the sending agent sets content-disposition.
+        seen = []   # Only skip the first example of each candidate type.
+        for part in parts:
+            maintype, subtype = part.get_content_type().split('/')
+            if ((maintype, subtype) in self._body_types and
+                    not part.is_attachment() and subtype not in seen):
+                seen.append(subtype)
+                continue
+            yield part
+
+    def iter_parts(self):
+        """Return an iterator over all immediate subparts of a multipart.
+
+        Return an empty iterator for a non-multipart.
+        """
+        if self.get_content_maintype() == 'multipart':
+            yield from self.get_payload()
+
+    def get_content(self, *args, content_manager=None, **kw):
+        if content_manager is None:
+            content_manager = self.policy.content_manager
+        return content_manager.get_content(self, *args, **kw)
+
+    def set_content(self, *args, content_manager=None, **kw):
+        if content_manager is None:
+            content_manager = self.policy.content_manager
+        content_manager.set_content(self, *args, **kw)
+
+    def _make_multipart(self, subtype, disallowed_subtypes, boundary):
+        if self.get_content_maintype() == 'multipart':
+            existing_subtype = self.get_content_subtype()
+            disallowed_subtypes = disallowed_subtypes + (subtype,)
+            if existing_subtype in disallowed_subtypes:
+                raise ValueError("Cannot convert {} to {}".format(
+                    existing_subtype, subtype))
+        keep_headers = []
+        part_headers = []
+        for name, value in self._headers:
+            if name.lower().startswith('content-'):
+                part_headers.append((name, value))
+            else:
+                keep_headers.append((name, value))
+        if part_headers:
+            # There is existing content, move it to the first subpart.
+            part = type(self)(policy=self.policy)
+            part._headers = part_headers
+            part._payload = self._payload
+            self._payload = [part]
+        else:
+            self._payload = []
+        self._headers = keep_headers
+        self['Content-Type'] = 'multipart/' + subtype
+        if boundary is not None:
+            self.set_param('boundary', boundary)
+
+    def make_related(self, boundary=None):
+        self._make_multipart('related', ('alternative', 'mixed'), boundary)
+
+    def make_alternative(self, boundary=None):
+        self._make_multipart('alternative', ('mixed',), boundary)
+
+    def make_mixed(self, boundary=None):
+        self._make_multipart('mixed', (), boundary)
+
+    def _add_multipart(self, _subtype, *args, _disp=None, **kw):
+        if (self.get_content_maintype() != 'multipart' or
+                self.get_content_subtype() != _subtype):
+            getattr(self, 'make_' + _subtype)()
+        part = type(self)(policy=self.policy)
+        part.set_content(*args, **kw)
+        if _disp and 'content-disposition' not in part:
+            part['Content-Disposition'] = _disp
+        self.attach(part)
+
+    def add_related(self, *args, **kw):
+        self._add_multipart('related', *args, _disp='inline', **kw)
+
+    def add_alternative(self, *args, **kw):
+        self._add_multipart('alternative', *args, **kw)
+
+    def add_attachment(self, *args, **kw):
+        self._add_multipart('mixed', *args, _disp='attachment', **kw)
+
+    def clear(self):
+        self._headers = []
+        self._payload = None
+
+    def clear_content(self):
+        self._headers = [(n, v) for n, v in self._headers
+                         if not n.lower().startswith('content-')]
+        self._payload = None
+
+
+class EmailMessage(MIMEPart):
+
+    def set_content(self, *args, **kw):
+        super().set_content(*args, **kw)
+        if 'MIME-Version' not in self:
+            self['MIME-Version'] = '1.0'
diff --git a/Lib/email/mime/__init__.py b/Lib/email/mime/__init__.py
new file mode 100644
index 000000000..e69de29bb
diff --git a/Lib/email/mime/application.py b/Lib/email/mime/application.py
new file mode 100644
index 000000000..6877e554e
--- /dev/null
+++ b/Lib/email/mime/application.py
@@ -0,0 +1,37 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Keith Dart
+# Contact: email-sig@python.org
+
+"""Class representing application/* type MIME documents."""
+
+__all__ = ["MIMEApplication"]
+
+from email import encoders
+from email.mime.nonmultipart import MIMENonMultipart
+
+
+class MIMEApplication(MIMENonMultipart):
+    """Class for generating application/* MIME documents."""
+
+    def __init__(self, _data, _subtype='octet-stream',
+                 _encoder=encoders.encode_base64, *, policy=None, **_params):
+        """Create an application/* type MIME document.
+
+        _data is a string containing the raw application data.
+
+        _subtype is the MIME content type subtype, defaulting to
+        'octet-stream'.
+
+        _encoder is a function which will perform the actual encoding for
+        transport of the application data, defaulting to base64 encoding.
+
+        Any additional keyword arguments are passed to the base class
+        constructor, which turns them into parameters on the Content-Type
+        header.
+        """
+        if _subtype is None:
+            raise TypeError('Invalid application MIME subtype')
+        MIMENonMultipart.__init__(self, 'application', _subtype, policy=policy,
+                                  **_params)
+        self.set_payload(_data)
+        _encoder(self)
diff --git a/Lib/email/mime/audio.py b/Lib/email/mime/audio.py
new file mode 100644
index 000000000..4bcd7b224
--- /dev/null
+++ b/Lib/email/mime/audio.py
@@ -0,0 +1,74 @@
+# Copyright (C) 2001-2007 Python Software Foundation
+# Author: Anthony Baxter
+# Contact: email-sig@python.org
+
+"""Class representing audio/* type MIME documents."""
+
+__all__ = ['MIMEAudio']
+
+import sndhdr
+
+from io import BytesIO
+from email import encoders
+from email.mime.nonmultipart import MIMENonMultipart
+
+
+
+_sndhdr_MIMEmap = {'au'  : 'basic',
+                   'wav' :'x-wav',
+                   'aiff':'x-aiff',
+                   'aifc':'x-aiff',
+                   }
+
+# There are others in sndhdr that don't have MIME types. :(
+# Additional ones to be added to sndhdr? midi, mp3, realaudio, wma??
+def _whatsnd(data):
+    """Try to identify a sound file type.
+
+    sndhdr.what() has a pretty cruddy interface, unfortunately.  This is why
+    we re-do it here.  It would be easier to reverse engineer the Unix 'file'
+    command and use the standard 'magic' file, as shipped with a modern Unix.
+    """
+    hdr = data[:512]
+    fakefile = BytesIO(hdr)
+    for testfn in sndhdr.tests:
+        res = testfn(hdr, fakefile)
+        if res is not None:
+            return _sndhdr_MIMEmap.get(res[0])
+    return None
+
+
+
+class MIMEAudio(MIMENonMultipart):
+    """Class for generating audio/* MIME documents."""
+
+    def __init__(self, _audiodata, _subtype=None,
+                 _encoder=encoders.encode_base64, *, policy=None, **_params):
+        """Create an audio/* type MIME document.
+
+        _audiodata is a string containing the raw audio data.  If this data
+        can be decoded by the standard Python `sndhdr' module, then the
+        subtype will be automatically included in the Content-Type header.
+        Otherwise, you can specify  the specific audio subtype via the
+        _subtype parameter.  If _subtype is not given, and no subtype can be
+        guessed, a TypeError is raised.
+
+        _encoder is a function which will perform the actual encoding for
+        transport of the image data.  It takes one argument, which is this
+        Image instance.  It should use get_payload() and set_payload() to
+        change the payload to the encoded form.  It should also add any
+        Content-Transfer-Encoding or other headers to the message as
+        necessary.  The default encoding is Base64.
+
+        Any additional keyword arguments are passed to the base class
+        constructor, which turns them into parameters on the Content-Type
+        header.
+        """
+        if _subtype is None:
+            _subtype = _whatsnd(_audiodata)
+        if _subtype is None:
+            raise TypeError('Could not find audio MIME subtype')
+        MIMENonMultipart.__init__(self, 'audio', _subtype, policy=policy,
+                                  **_params)
+        self.set_payload(_audiodata)
+        _encoder(self)
diff --git a/Lib/email/mime/base.py b/Lib/email/mime/base.py
new file mode 100644
index 000000000..1a3f9b51f
--- /dev/null
+++ b/Lib/email/mime/base.py
@@ -0,0 +1,30 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Base class for MIME specializations."""
+
+__all__ = ['MIMEBase']
+
+import email.policy
+
+from email import message
+
+
+
+class MIMEBase(message.Message):
+    """Base class for MIME specializations."""
+
+    def __init__(self, _maintype, _subtype, *, policy=None, **_params):
+        """This constructor adds a Content-Type: and a MIME-Version: header.
+
+        The Content-Type: header is taken from the _maintype and _subtype
+        arguments.  Additional parameters for this header are taken from the
+        keyword arguments.
+        """
+        if policy is None:
+            policy = email.policy.compat32
+        message.Message.__init__(self, policy=policy)
+        ctype = '%s/%s' % (_maintype, _subtype)
+        self.add_header('Content-Type', ctype, **_params)
+        self['MIME-Version'] = '1.0'
diff --git a/Lib/email/mime/image.py b/Lib/email/mime/image.py
new file mode 100644
index 000000000..92724643c
--- /dev/null
+++ b/Lib/email/mime/image.py
@@ -0,0 +1,47 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Class representing image/* type MIME documents."""
+
+__all__ = ['MIMEImage']
+
+import imghdr
+
+from email import encoders
+from email.mime.nonmultipart import MIMENonMultipart
+
+
+
+class MIMEImage(MIMENonMultipart):
+    """Class for generating image/* type MIME documents."""
+
+    def __init__(self, _imagedata, _subtype=None,
+                 _encoder=encoders.encode_base64, *, policy=None, **_params):
+        """Create an image/* type MIME document.
+
+        _imagedata is a string containing the raw image data.  If this data
+        can be decoded by the standard Python `imghdr' module, then the
+        subtype will be automatically included in the Content-Type header.
+        Otherwise, you can specify the specific image subtype via the _subtype
+        parameter.
+
+        _encoder is a function which will perform the actual encoding for
+        transport of the image data.  It takes one argument, which is this
+        Image instance.  It should use get_payload() and set_payload() to
+        change the payload to the encoded form.  It should also add any
+        Content-Transfer-Encoding or other headers to the message as
+        necessary.  The default encoding is Base64.
+
+        Any additional keyword arguments are passed to the base class
+        constructor, which turns them into parameters on the Content-Type
+        header.
+        """
+        if _subtype is None:
+            _subtype = imghdr.what(None, _imagedata)
+        if _subtype is None:
+            raise TypeError('Could not guess image MIME subtype')
+        MIMENonMultipart.__init__(self, 'image', _subtype, policy=policy,
+                                  **_params)
+        self.set_payload(_imagedata)
+        _encoder(self)
diff --git a/Lib/email/mime/message.py b/Lib/email/mime/message.py
new file mode 100644
index 000000000..07e4f2d11
--- /dev/null
+++ b/Lib/email/mime/message.py
@@ -0,0 +1,34 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Class representing message/* MIME documents."""
+
+__all__ = ['MIMEMessage']
+
+from email import message
+from email.mime.nonmultipart import MIMENonMultipart
+
+
+
+class MIMEMessage(MIMENonMultipart):
+    """Class representing message/* MIME documents."""
+
+    def __init__(self, _msg, _subtype='rfc822', *, policy=None):
+        """Create a message/* type MIME document.
+
+        _msg is a message object and must be an instance of Message, or a
+        derived class of Message, otherwise a TypeError is raised.
+
+        Optional _subtype defines the subtype of the contained message.  The
+        default is "rfc822" (this is defined by the MIME standard, even though
+        the term "rfc822" is technically outdated by RFC 2822).
+        """
+        MIMENonMultipart.__init__(self, 'message', _subtype, policy=policy)
+        if not isinstance(_msg, message.Message):
+            raise TypeError('Argument is not an instance of Message')
+        # It's convenient to use this base class method.  We need to do it
+        # this way or we'll get an exception
+        message.Message.attach(self, _msg)
+        # And be sure our default type is set correctly
+        self.set_default_type('message/rfc822')
diff --git a/Lib/email/mime/multipart.py b/Lib/email/mime/multipart.py
new file mode 100644
index 000000000..2d3f28881
--- /dev/null
+++ b/Lib/email/mime/multipart.py
@@ -0,0 +1,48 @@
+# Copyright (C) 2002-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Base class for MIME multipart/* type messages."""
+
+__all__ = ['MIMEMultipart']
+
+from email.mime.base import MIMEBase
+
+
+
+class MIMEMultipart(MIMEBase):
+    """Base class for MIME multipart/* type messages."""
+
+    def __init__(self, _subtype='mixed', boundary=None, _subparts=None,
+                 *, policy=None,
+                 **_params):
+        """Creates a multipart/* type message.
+
+        By default, creates a multipart/mixed message, with proper
+        Content-Type and MIME-Version headers.
+
+        _subtype is the subtype of the multipart content type, defaulting to
+        `mixed'.
+
+        boundary is the multipart boundary string.  By default it is
+        calculated as needed.
+
+        _subparts is a sequence of initial subparts for the payload.  It
+        must be an iterable object, such as a list.  You can always
+        attach new subparts to the message by using the attach() method.
+
+        Additional parameters for the Content-Type header are taken from the
+        keyword arguments (or passed into the _params argument).
+        """
+        MIMEBase.__init__(self, 'multipart', _subtype, policy=policy, **_params)
+
+        # Initialise _payload to an empty list as the Message superclass's
+        # implementation of is_multipart assumes that _payload is a list for
+        # multipart messages.
+        self._payload = []
+
+        if _subparts:
+            for p in _subparts:
+                self.attach(p)
+        if boundary:
+            self.set_boundary(boundary)
diff --git a/Lib/email/mime/nonmultipart.py b/Lib/email/mime/nonmultipart.py
new file mode 100644
index 000000000..e1f51968b
--- /dev/null
+++ b/Lib/email/mime/nonmultipart.py
@@ -0,0 +1,22 @@
+# Copyright (C) 2002-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Base class for MIME type messages that are not multipart."""
+
+__all__ = ['MIMENonMultipart']
+
+from email import errors
+from email.mime.base import MIMEBase
+
+
+
+class MIMENonMultipart(MIMEBase):
+    """Base class for MIME non-multipart type messages."""
+
+    def attach(self, payload):
+        # The public API prohibits attaching multiple subparts to MIMEBase
+        # derived subtypes since none of them are, by definition, of content
+        # type multipart/*
+        raise errors.MultipartConversionError(
+            'Cannot attach additional subparts to non-multipart/*')
diff --git a/Lib/email/mime/text.py b/Lib/email/mime/text.py
new file mode 100644
index 000000000..35b442383
--- /dev/null
+++ b/Lib/email/mime/text.py
@@ -0,0 +1,42 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Class representing text/* type MIME documents."""
+
+__all__ = ['MIMEText']
+
+from email.charset import Charset
+from email.mime.nonmultipart import MIMENonMultipart
+
+
+
+class MIMEText(MIMENonMultipart):
+    """Class for generating text/* type MIME documents."""
+
+    def __init__(self, _text, _subtype='plain', _charset=None, *, policy=None):
+        """Create a text/* type MIME document.
+
+        _text is the string for this message object.
+
+        _subtype is the MIME sub content type, defaulting to "plain".
+
+        _charset is the character set parameter added to the Content-Type
+        header.  This defaults to "us-ascii".  Note that as a side-effect, the
+        Content-Transfer-Encoding header will also be set.
+        """
+
+        # If no _charset was specified, check to see if there are non-ascii
+        # characters present. If not, use 'us-ascii', otherwise use utf-8.
+        # XXX: This can be removed once #7304 is fixed.
+        if _charset is None:
+            try:
+                _text.encode('us-ascii')
+                _charset = 'us-ascii'
+            except UnicodeEncodeError:
+                _charset = 'utf-8'
+
+        MIMENonMultipart.__init__(self, 'text', _subtype, policy=policy,
+                                  **{'charset': str(_charset)})
+
+        self.set_payload(_text, _charset)
diff --git a/Lib/email/parser.py b/Lib/email/parser.py
new file mode 100644
index 000000000..555b17256
--- /dev/null
+++ b/Lib/email/parser.py
@@ -0,0 +1,132 @@
+# Copyright (C) 2001-2007 Python Software Foundation
+# Author: Barry Warsaw, Thomas Wouters, Anthony Baxter
+# Contact: email-sig@python.org
+
+"""A parser of RFC 2822 and MIME email messages."""
+
+__all__ = ['Parser', 'HeaderParser', 'BytesParser', 'BytesHeaderParser',
+           'FeedParser', 'BytesFeedParser']
+
+from io import StringIO, TextIOWrapper
+
+from email.feedparser import FeedParser, BytesFeedParser
+from email._policybase import compat32
+
+
+
+class Parser:
+    def __init__(self, _class=None, *, policy=compat32):
+        """Parser of RFC 2822 and MIME email messages.
+
+        Creates an in-memory object tree representing the email message, which
+        can then be manipulated and turned over to a Generator to return the
+        textual representation of the message.
+
+        The string must be formatted as a block of RFC 2822 headers and header
+        continuation lines, optionally preceded by a `Unix-from' header.  The
+        header block is terminated either by the end of the string or by a
+        blank line.
+
+        _class is the class to instantiate for new message objects when they
+        must be created.  This class must have a constructor that can take
+        zero arguments.  Default is Message.Message.
+
+        The policy keyword specifies a policy object that controls a number of
+        aspects of the parser's operation.  The default policy maintains
+        backward compatibility.
+
+        """
+        self._class = _class
+        self.policy = policy
+
+    def parse(self, fp, headersonly=False):
+        """Create a message structure from the data in a file.
+
+        Reads all the data from the file and returns the root of the message
+        structure.  Optional headersonly is a flag specifying whether to stop
+        parsing after reading the headers or not.  The default is False,
+        meaning it parses the entire contents of the file.
+        """
+        feedparser = FeedParser(self._class, policy=self.policy)
+        if headersonly:
+            feedparser._set_headersonly()
+        while True:
+            data = fp.read(8192)
+            if not data:
+                break
+            feedparser.feed(data)
+        return feedparser.close()
+
+    def parsestr(self, text, headersonly=False):
+        """Create a message structure from a string.
+
+        Returns the root of the message structure.  Optional headersonly is a
+        flag specifying whether to stop parsing after reading the headers or
+        not.  The default is False, meaning it parses the entire contents of
+        the file.
+        """
+        return self.parse(StringIO(text), headersonly=headersonly)
+
+
+
+class HeaderParser(Parser):
+    def parse(self, fp, headersonly=True):
+        return Parser.parse(self, fp, True)
+
+    def parsestr(self, text, headersonly=True):
+        return Parser.parsestr(self, text, True)
+
+
+class BytesParser:
+
+    def __init__(self, *args, **kw):
+        """Parser of binary RFC 2822 and MIME email messages.
+
+        Creates an in-memory object tree representing the email message, which
+        can then be manipulated and turned over to a Generator to return the
+        textual representation of the message.
+
+        The input must be formatted as a block of RFC 2822 headers and header
+        continuation lines, optionally preceded by a `Unix-from' header.  The
+        header block is terminated either by the end of the input or by a
+        blank line.
+
+        _class is the class to instantiate for new message objects when they
+        must be created.  This class must have a constructor that can take
+        zero arguments.  Default is Message.Message.
+        """
+        self.parser = Parser(*args, **kw)
+
+    def parse(self, fp, headersonly=False):
+        """Create a message structure from the data in a binary file.
+
+        Reads all the data from the file and returns the root of the message
+        structure.  Optional headersonly is a flag specifying whether to stop
+        parsing after reading the headers or not.  The default is False,
+        meaning it parses the entire contents of the file.
+        """
+        fp = TextIOWrapper(fp, encoding='ascii', errors='surrogateescape')
+        try:
+            return self.parser.parse(fp, headersonly)
+        finally:
+            fp.detach()
+
+
+    def parsebytes(self, text, headersonly=False):
+        """Create a message structure from a byte string.
+
+        Returns the root of the message structure.  Optional headersonly is a
+        flag specifying whether to stop parsing after reading the headers or
+        not.  The default is False, meaning it parses the entire contents of
+        the file.
+        """
+        text = text.decode('ASCII', errors='surrogateescape')
+        return self.parser.parsestr(text, headersonly)
+
+
+class BytesHeaderParser(BytesParser):
+    def parse(self, fp, headersonly=True):
+        return BytesParser.parse(self, fp, headersonly=True)
+
+    def parsebytes(self, text, headersonly=True):
+        return BytesParser.parsebytes(self, text, headersonly=True)
diff --git a/Lib/email/policy.py b/Lib/email/policy.py
new file mode 100644
index 000000000..5131311ac
--- /dev/null
+++ b/Lib/email/policy.py
@@ -0,0 +1,223 @@
+"""This will be the home for the policy that hooks in the new
+code that adds all the email6 features.
+"""
+
+import re
+from email._policybase import Policy, Compat32, compat32, _extend_docstrings
+from email.utils import _has_surrogates
+from email.headerregistry import HeaderRegistry as HeaderRegistry
+from email.contentmanager import raw_data_manager
+from email.message import EmailMessage
+
+__all__ = [
+    'Compat32',
+    'compat32',
+    'Policy',
+    'EmailPolicy',
+    'default',
+    'strict',
+    'SMTP',
+    'HTTP',
+    ]
+
+linesep_splitter = re.compile(r'\n|\r')
+
+@_extend_docstrings
+class EmailPolicy(Policy):
+
+    """+
+    PROVISIONAL
+
+    The API extensions enabled by this policy are currently provisional.
+    Refer to the documentation for details.
+
+    This policy adds new header parsing and folding algorithms.  Instead of
+    simple strings, headers are custom objects with custom attributes
+    depending on the type of the field.  The folding algorithm fully
+    implements RFCs 2047 and 5322.
+
+    In addition to the settable attributes listed above that apply to
+    all Policies, this policy adds the following additional attributes:
+
+    utf8                -- if False (the default) message headers will be
+                           serialized as ASCII, using encoded words to encode
+                           any non-ASCII characters in the source strings.  If
+                           True, the message headers will be serialized using
+                           utf8 and will not contain encoded words (see RFC
+                           6532 for more on this serialization format).
+
+    refold_source       -- if the value for a header in the Message object
+                           came from the parsing of some source, this attribute
+                           indicates whether or not a generator should refold
+                           that value when transforming the message back into
+                           stream form.  The possible values are:
+
+                           none  -- all source values use original folding
+                           long  -- source values that have any line that is
+                                    longer than max_line_length will be
+                                    refolded
+                           all  -- all values are refolded.
+
+                           The default is 'long'.
+
+    header_factory      -- a callable that takes two arguments, 'name' and
+                           'value', where 'name' is a header field name and
+                           'value' is an unfolded header field value, and
+                           returns a string-like object that represents that
+                           header.  A default header_factory is provided that
+                           understands some of the RFC5322 header field types.
+                           (Currently address fields and date fields have
+                           special treatment, while all other fields are
+                           treated as unstructured.  This list will be
+                           completed before the extension is marked stable.)
+
+    content_manager     -- an object with at least two methods: get_content
+                           and set_content.  When the get_content or
+                           set_content method of a Message object is called,
+                           it calls the corresponding method of this object,
+                           passing it the message object as its first argument,
+                           and any arguments or keywords that were passed to
+                           it as additional arguments.  The default
+                           content_manager is
+                           :data:`~email.contentmanager.raw_data_manager`.
+
+    """
+
+    message_factory = EmailMessage
+    utf8 = False
+    refold_source = 'long'
+    header_factory = HeaderRegistry()
+    content_manager = raw_data_manager
+
+    def __init__(self, **kw):
+        # Ensure that each new instance gets a unique header factory
+        # (as opposed to clones, which share the factory).
+        if 'header_factory' not in kw:
+            object.__setattr__(self, 'header_factory', HeaderRegistry())
+        super().__init__(**kw)
+
+    def header_max_count(self, name):
+        """+
+        The implementation for this class returns the max_count attribute from
+        the specialized header class that would be used to construct a header
+        of type 'name'.
+        """
+        return self.header_factory[name].max_count
+
+    # The logic of the next three methods is chosen such that it is possible to
+    # switch a Message object between a Compat32 policy and a policy derived
+    # from this class and have the results stay consistent.  This allows a
+    # Message object constructed with this policy to be passed to a library
+    # that only handles Compat32 objects, or to receive such an object and
+    # convert it to use the newer style by just changing its policy.  It is
+    # also chosen because it postpones the relatively expensive full rfc5322
+    # parse until as late as possible when parsing from source, since in many
+    # applications only a few headers will actually be inspected.
+
+    def header_source_parse(self, sourcelines):
+        """+
+        The name is parsed as everything up to the ':' and returned unmodified.
+        The value is determined by stripping leading whitespace off the
+        remainder of the first line, joining all subsequent lines together, and
+        stripping any trailing carriage return or linefeed characters.  (This
+        is the same as Compat32).
+
+        """
+        name, value = sourcelines[0].split(':', 1)
+        value = value.lstrip(' \t') + ''.join(sourcelines[1:])
+        return (name, value.rstrip('\r\n'))
+
+    def header_store_parse(self, name, value):
+        """+
+        The name is returned unchanged.  If the input value has a 'name'
+        attribute and it matches the name ignoring case, the value is returned
+        unchanged.  Otherwise the name and value are passed to header_factory
+        method, and the resulting custom header object is returned as the
+        value.  In this case a ValueError is raised if the input value contains
+        CR or LF characters.
+
+        """
+        if hasattr(value, 'name') and value.name.lower() == name.lower():
+            return (name, value)
+        if isinstance(value, str) and len(value.splitlines())>1:
+            # XXX this error message isn't quite right when we use splitlines
+            # (see issue 22233), but I'm not sure what should happen here.
+            raise ValueError("Header values may not contain linefeed "
+                             "or carriage return characters")
+        return (name, self.header_factory(name, value))
+
+    def header_fetch_parse(self, name, value):
+        """+
+        If the value has a 'name' attribute, it is returned to unmodified.
+        Otherwise the name and the value with any linesep characters removed
+        are passed to the header_factory method, and the resulting custom
+        header object is returned.  Any surrogateescaped bytes get turned
+        into the unicode unknown-character glyph.
+
+        """
+        if hasattr(value, 'name'):
+            return value
+        # We can't use splitlines here because it splits on more than \r and \n.
+        value = ''.join(linesep_splitter.split(value))
+        return self.header_factory(name, value)
+
+    def fold(self, name, value):
+        """+
+        Header folding is controlled by the refold_source policy setting.  A
+        value is considered to be a 'source value' if and only if it does not
+        have a 'name' attribute (having a 'name' attribute means it is a header
+        object of some sort).  If a source value needs to be refolded according
+        to the policy, it is converted into a custom header object by passing
+        the name and the value with any linesep characters removed to the
+        header_factory method.  Folding of a custom header object is done by
+        calling its fold method with the current policy.
+
+        Source values are split into lines using splitlines.  If the value is
+        not to be refolded, the lines are rejoined using the linesep from the
+        policy and returned.  The exception is lines containing non-ascii
+        binary data.  In that case the value is refolded regardless of the
+        refold_source setting, which causes the binary data to be CTE encoded
+        using the unknown-8bit charset.
+
+        """
+        return self._fold(name, value, refold_binary=True)
+
+    def fold_binary(self, name, value):
+        """+
+        The same as fold if cte_type is 7bit, except that the returned value is
+        bytes.
+
+        If cte_type is 8bit, non-ASCII binary data is converted back into
+        bytes.  Headers with binary data are not refolded, regardless of the
+        refold_header setting, since there is no way to know whether the binary
+        data consists of single byte characters or multibyte characters.
+
+        If utf8 is true, headers are encoded to utf8, otherwise to ascii with
+        non-ASCII unicode rendered as encoded words.
+
+        """
+        folded = self._fold(name, value, refold_binary=self.cte_type=='7bit')
+        charset = 'utf8' if self.utf8 else 'ascii'
+        return folded.encode(charset, 'surrogateescape')
+
+    def _fold(self, name, value, refold_binary=False):
+        if hasattr(value, 'name'):
+            return value.fold(policy=self)
+        maxlen = self.max_line_length if self.max_line_length else float('inf')
+        lines = value.splitlines()
+        refold = (self.refold_source == 'all' or
+                  self.refold_source == 'long' and
+                    (lines and len(lines[0])+len(name)+2 > maxlen or
+                     any(len(x) > maxlen for x in lines[1:])))
+        if refold or refold_binary and _has_surrogates(value):
+            return self.header_factory(name, ''.join(lines)).fold(policy=self)
+        return name + ': ' + self.linesep.join(lines) + self.linesep
+
+
+default = EmailPolicy()
+# Make the default policy use the class default header_factory
+del default.header_factory
+strict = default.clone(raise_on_defect=True)
+SMTP = default.clone(linesep='\r\n')
+HTTP = default.clone(linesep='\r\n', max_line_length=None)
+SMTPUTF8 = SMTP.clone(utf8=True)
diff --git a/Lib/email/quoprimime.py b/Lib/email/quoprimime.py
new file mode 100644
index 000000000..c543eb59a
--- /dev/null
+++ b/Lib/email/quoprimime.py
@@ -0,0 +1,299 @@
+# Copyright (C) 2001-2006 Python Software Foundation
+# Author: Ben Gertzfield
+# Contact: email-sig@python.org
+
+"""Quoted-printable content transfer encoding per RFCs 2045-2047.
+
+This module handles the content transfer encoding method defined in RFC 2045
+to encode US ASCII-like 8-bit data called `quoted-printable'.  It is used to
+safely encode text that is in a character set similar to the 7-bit US ASCII
+character set, but that includes some 8-bit characters that are normally not
+allowed in email bodies or headers.
+
+Quoted-printable is very space-inefficient for encoding binary files; use the
+email.base64mime module for that instead.
+
+This module provides an interface to encode and decode both headers and bodies
+with quoted-printable encoding.
+
+RFC 2045 defines a method for including character set information in an
+`encoded-word' in a header.  This method is commonly used for 8-bit real names
+in To:/From:/Cc: etc. fields, as well as Subject: lines.
+
+This module does not do the line wrapping or end-of-line character
+conversion necessary for proper internationalized headers; it only
+does dumb encoding and decoding.  To deal with the various line
+wrapping issues, use the email.header module.
+"""
+
+__all__ = [
+    'body_decode',
+    'body_encode',
+    'body_length',
+    'decode',
+    'decodestring',
+    'header_decode',
+    'header_encode',
+    'header_length',
+    'quote',
+    'unquote',
+    ]
+
+import re
+
+from string import ascii_letters, digits, hexdigits
+
+CRLF = '\r\n'
+NL = '\n'
+EMPTYSTRING = ''
+
+# Build a mapping of octets to the expansion of that octet.  Since we're only
+# going to have 256 of these things, this isn't terribly inefficient
+# space-wise.  Remember that headers and bodies have different sets of safe
+# characters.  Initialize both maps with the full expansion, and then override
+# the safe bytes with the more compact form.
+_QUOPRI_MAP = ['=%02X' % c for c in range(256)]
+_QUOPRI_HEADER_MAP = _QUOPRI_MAP[:]
+_QUOPRI_BODY_MAP = _QUOPRI_MAP[:]
+
+# Safe header bytes which need no encoding.
+for c in b'-!*+/' + ascii_letters.encode('ascii') + digits.encode('ascii'):
+    _QUOPRI_HEADER_MAP[c] = chr(c)
+# Headers have one other special encoding; spaces become underscores.
+_QUOPRI_HEADER_MAP[ord(' ')] = '_'
+
+# Safe body bytes which need no encoding.
+for c in (b' !"#$%&\'()*+,-./0123456789:;<>'
+          b'?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`'
+          b'abcdefghijklmnopqrstuvwxyz{|}~\t'):
+    _QUOPRI_BODY_MAP[c] = chr(c)
+
+
+
+# Helpers
+def header_check(octet):
+    """Return True if the octet should be escaped with header quopri."""
+    return chr(octet) != _QUOPRI_HEADER_MAP[octet]
+
+
+def body_check(octet):
+    """Return True if the octet should be escaped with body quopri."""
+    return chr(octet) != _QUOPRI_BODY_MAP[octet]
+
+
+def header_length(bytearray):
+    """Return a header quoted-printable encoding length.
+
+    Note that this does not include any RFC 2047 chrome added by
+    `header_encode()`.
+
+    :param bytearray: An array of bytes (a.k.a. octets).
+    :return: The length in bytes of the byte array when it is encoded with
+        quoted-printable for headers.
+    """
+    return sum(len(_QUOPRI_HEADER_MAP[octet]) for octet in bytearray)
+
+
+def body_length(bytearray):
+    """Return a body quoted-printable encoding length.
+
+    :param bytearray: An array of bytes (a.k.a. octets).
+    :return: The length in bytes of the byte array when it is encoded with
+        quoted-printable for bodies.
+    """
+    return sum(len(_QUOPRI_BODY_MAP[octet]) for octet in bytearray)
+
+
+def _max_append(L, s, maxlen, extra=''):
+    if not isinstance(s, str):
+        s = chr(s)
+    if not L:
+        L.append(s.lstrip())
+    elif len(L[-1]) + len(s) <= maxlen:
+        L[-1] += extra + s
+    else:
+        L.append(s.lstrip())
+
+
+def unquote(s):
+    """Turn a string in the form =AB to the ASCII character with value 0xab"""
+    return chr(int(s[1:3], 16))
+
+
+def quote(c):
+    return _QUOPRI_MAP[ord(c)]
+
+
+def header_encode(header_bytes, charset='iso-8859-1'):
+    """Encode a single header line with quoted-printable (like) encoding.
+
+    Defined in RFC 2045, this `Q' encoding is similar to quoted-printable, but
+    used specifically for email header fields to allow charsets with mostly 7
+    bit characters (and some 8 bit) to remain more or less readable in non-RFC
+    2045 aware mail clients.
+
+    charset names the character set to use in the RFC 2046 header.  It
+    defaults to iso-8859-1.
+    """
+    # Return empty headers as an empty string.
+    if not header_bytes:
+        return ''
+    # Iterate over every byte, encoding if necessary.
+    encoded = header_bytes.decode('latin1').translate(_QUOPRI_HEADER_MAP)
+    # Now add the RFC chrome to each encoded chunk and glue the chunks
+    # together.
+    return '=?%s?q?%s?=' % (charset, encoded)
+
+
+_QUOPRI_BODY_ENCODE_MAP = _QUOPRI_BODY_MAP[:]
+for c in b'\r\n':
+    _QUOPRI_BODY_ENCODE_MAP[c] = chr(c)
+
+def body_encode(body, maxlinelen=76, eol=NL):
+    """Encode with quoted-printable, wrapping at maxlinelen characters.
+
+    Each line of encoded text will end with eol, which defaults to "\\n".  Set
+    this to "\\r\\n" if you will be using the result of this function directly
+    in an email.
+
+    Each line will be wrapped at, at most, maxlinelen characters before the
+    eol string (maxlinelen defaults to 76 characters, the maximum value
+    permitted by RFC 2045).  Long lines will have the 'soft line break'
+    quoted-printable character "=" appended to them, so the decoded text will
+    be identical to the original text.
+
+    The minimum maxlinelen is 4 to have room for a quoted character ("=XX")
+    followed by a soft line break.  Smaller values will generate a
+    ValueError.
+
+    """
+
+    if maxlinelen < 4:
+        raise ValueError("maxlinelen must be at least 4")
+    if not body:
+        return body
+
+    # quote speacial characters
+    body = body.translate(_QUOPRI_BODY_ENCODE_MAP)
+
+    soft_break = '=' + eol
+    # leave space for the '=' at the end of a line
+    maxlinelen1 = maxlinelen - 1
+
+    encoded_body = []
+    append = encoded_body.append
+
+    for line in body.splitlines():
+        # break up the line into pieces no longer than maxlinelen - 1
+        start = 0
+        laststart = len(line) - 1 - maxlinelen
+        while start <= laststart:
+            stop = start + maxlinelen1
+            # make sure we don't break up an escape sequence
+            if line[stop - 2] == '=':
+                append(line[start:stop - 1])
+                start = stop - 2
+            elif line[stop - 1] == '=':
+                append(line[start:stop])
+                start = stop - 1
+            else:
+                append(line[start:stop] + '=')
+                start = stop
+
+        # handle rest of line, special case if line ends in whitespace
+        if line and line[-1] in ' \t':
+            room = start - laststart
+            if room >= 3:
+                # It's a whitespace character at end-of-line, and we have room
+                # for the three-character quoted encoding.
+                q = quote(line[-1])
+            elif room == 2:
+                # There's room for the whitespace character and a soft break.
+                q = line[-1] + soft_break
+            else:
+                # There's room only for a soft break.  The quoted whitespace
+                # will be the only content on the subsequent line.
+                q = soft_break + quote(line[-1])
+            append(line[start:-1] + q)
+        else:
+            append(line[start:])
+
+    # add back final newline if present
+    if body[-1] in CRLF:
+        append('')
+
+    return eol.join(encoded_body)
+
+
+
+# BAW: I'm not sure if the intent was for the signature of this function to be
+# the same as base64MIME.decode() or not...
+def decode(encoded, eol=NL):
+    """Decode a quoted-printable string.
+
+    Lines are separated with eol, which defaults to \\n.
+    """
+    if not encoded:
+        return encoded
+    # BAW: see comment in encode() above.  Again, we're building up the
+    # decoded string with string concatenation, which could be done much more
+    # efficiently.
+    decoded = ''
+
+    for line in encoded.splitlines():
+        line = line.rstrip()
+        if not line:
+            decoded += eol
+            continue
+
+        i = 0
+        n = len(line)
+        while i < n:
+            c = line[i]
+            if c != '=':
+                decoded += c
+                i += 1
+            # Otherwise, c == "=".  Are we at the end of the line?  If so, add
+            # a soft line break.
+            elif i+1 == n:
+                i += 1
+                continue
+            # Decode if in form =AB
+            elif i+2 < n and line[i+1] in hexdigits and line[i+2] in hexdigits:
+                decoded += unquote(line[i:i+3])
+                i += 3
+            # Otherwise, not in form =AB, pass literally
+            else:
+                decoded += c
+                i += 1
+
+            if i == n:
+                decoded += eol
+    # Special case if original string did not end with eol
+    if encoded[-1] not in '\r\n' and decoded.endswith(eol):
+        decoded = decoded[:-1]
+    return decoded
+
+
+# For convenience and backwards compatibility w/ standard base64 module
+body_decode = decode
+decodestring = decode
+
+
+
+def _unquote_match(match):
+    """Turn a match in the form =AB to the ASCII character with value 0xab"""
+    s = match.group(0)
+    return unquote(s)
+
+
+# Header decoding is done a bit differently
+def header_decode(s):
+    """Decode a string encoded with RFC 2045 MIME header `Q' encoding.
+
+    This function does not parse a full MIME header value encoded with
+    quoted-printable (like =?iso-8859-1?q?Hello_World?=) -- please use
+    the high level email.header class for that functionality.
+    """
+    s = s.replace('_', ' ')
+    return re.sub(r'=[a-fA-F0-9]{2}', _unquote_match, s, flags=re.ASCII)
diff --git a/Lib/email/utils.py b/Lib/email/utils.py
new file mode 100644
index 000000000..a759d2330
--- /dev/null
+++ b/Lib/email/utils.py
@@ -0,0 +1,382 @@
+# Copyright (C) 2001-2010 Python Software Foundation
+# Author: Barry Warsaw
+# Contact: email-sig@python.org
+
+"""Miscellaneous utilities."""
+
+__all__ = [
+    'collapse_rfc2231_value',
+    'decode_params',
+    'decode_rfc2231',
+    'encode_rfc2231',
+    'formataddr',
+    'formatdate',
+    'format_datetime',
+    'getaddresses',
+    'make_msgid',
+    'mktime_tz',
+    'parseaddr',
+    'parsedate',
+    'parsedate_tz',
+    'parsedate_to_datetime',
+    'unquote',
+    ]
+
+import os
+import re
+import time
+import random
+import socket
+import datetime
+import urllib.parse
+
+from email._parseaddr import quote
+from email._parseaddr import AddressList as _AddressList
+from email._parseaddr import mktime_tz
+
+from email._parseaddr import parsedate, parsedate_tz, _parsedate_tz
+
+# Intrapackage imports
+from email.charset import Charset
+
+COMMASPACE = ', '
+EMPTYSTRING = ''
+UEMPTYSTRING = ''
+CRLF = '\r\n'
+TICK = "'"
+
+specialsre = re.compile(r'[][\\()<>@,:;".]')
+escapesre = re.compile(r'[\\"]')
+
+def _has_surrogates(s):
+    """Return True if s contains surrogate-escaped binary data."""
+    # This check is based on the fact that unless there are surrogates, utf8
+    # (Python's default encoding) can encode any string.  This is the fastest
+    # way to check for surrogates, see issue 11454 for timings.
+    try:
+        s.encode()
+        return False
+    except UnicodeEncodeError:
+        return True
+
+# How to deal with a string containing bytes before handing it to the
+# application through the 'normal' interface.
+def _sanitize(string):
+    # Turn any escaped bytes into unicode 'unknown' char.  If the escaped
+    # bytes happen to be utf-8 they will instead get decoded, even if they
+    # were invalid in the charset the source was supposed to be in.  This
+    # seems like it is not a bad thing; a defect was still registered.
+    original_bytes = string.encode('utf-8', 'surrogateescape')
+    return original_bytes.decode('utf-8', 'replace')
+
+
+
+# Helpers
+
+def formataddr(pair, charset='utf-8'):
+    """The inverse of parseaddr(), this takes a 2-tuple of the form
+    (realname, email_address) and returns the string value suitable
+    for an RFC 2822 From, To or Cc header.
+
+    If the first element of pair is false, then the second element is
+    returned unmodified.
+
+    Optional charset if given is the character set that is used to encode
+    realname in case realname is not ASCII safe.  Can be an instance of str or
+    a Charset-like object which has a header_encode method.  Default is
+    'utf-8'.
+    """
+    name, address = pair
+    # The address MUST (per RFC) be ascii, so raise a UnicodeError if it isn't.
+    address.encode('ascii')
+    if name:
+        try:
+            name.encode('ascii')
+        except UnicodeEncodeError:
+            if isinstance(charset, str):
+                charset = Charset(charset)
+            encoded_name = charset.header_encode(name)
+            return "%s <%s>" % (encoded_name, address)
+        else:
+            quotes = ''
+            if specialsre.search(name):
+                quotes = '"'
+            name = escapesre.sub(r'\\\g<0>', name)
+            return '%s%s%s <%s>' % (quotes, name, quotes, address)
+    return address
+
+
+
+def getaddresses(fieldvalues):
+    """Return a list of (REALNAME, EMAIL) for each fieldvalue."""
+    all = COMMASPACE.join(fieldvalues)
+    a = _AddressList(all)
+    return a.addresslist
+
+
+
+ecre = re.compile(r'''
+  =\?                   # literal =?
+  (?P<charset>[^?]*?)   # non-greedy up to the next ? is the charset
+  \?                    # literal ?
+  (?P<encoding>[qb])    # either a "q" or a "b", case insensitive
+  \?                    # literal ?
+  (?P<atom>.*?)         # non-greedy up to the next ?= is the atom
+  \?=                   # literal ?=
+  ''', re.VERBOSE | re.IGNORECASE)
+
+
+def _format_timetuple_and_zone(timetuple, zone):
+    return '%s, %02d %s %04d %02d:%02d:%02d %s' % (
+        ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun'][timetuple[6]],
+        timetuple[2],
+        ['Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
+         'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'][timetuple[1] - 1],
+        timetuple[0], timetuple[3], timetuple[4], timetuple[5],
+        zone)
+
+def formatdate(timeval=None, localtime=False, usegmt=False):
+    """Returns a date string as specified by RFC 2822, e.g.:
+
+    Fri, 09 Nov 2001 01:08:47 -0000
+
+    Optional timeval if given is a floating point time value as accepted by
+    gmtime() and localtime(), otherwise the current time is used.
+
+    Optional localtime is a flag that when True, interprets timeval, and
+    returns a date relative to the local timezone instead of UTC, properly
+    taking daylight savings time into account.
+
+    Optional argument usegmt means that the timezone is written out as
+    an ascii string, not numeric one (so "GMT" instead of "+0000"). This
+    is needed for HTTP, and is only used when localtime==False.
+    """
+    # Note: we cannot use strftime() because that honors the locale and RFC
+    # 2822 requires that day and month names be the English abbreviations.
+    if timeval is None:
+        timeval = time.time()
+    if localtime or usegmt:
+        dt = datetime.datetime.fromtimestamp(timeval, datetime.timezone.utc)
+    else:
+        dt = datetime.datetime.utcfromtimestamp(timeval)
+    if localtime:
+        dt = dt.astimezone()
+        usegmt = False
+    return format_datetime(dt, usegmt)
+
+def format_datetime(dt, usegmt=False):
+    """Turn a datetime into a date string as specified in RFC 2822.
+
+    If usegmt is True, dt must be an aware datetime with an offset of zero.  In
+    this case 'GMT' will be rendered instead of the normal +0000 required by
+    RFC2822.  This is to support HTTP headers involving date stamps.
+    """
+    now = dt.timetuple()
+    if usegmt:
+        if dt.tzinfo is None or dt.tzinfo != datetime.timezone.utc:
+            raise ValueError("usegmt option requires a UTC datetime")
+        zone = 'GMT'
+    elif dt.tzinfo is None:
+        zone = '-0000'
+    else:
+        zone = dt.strftime("%z")
+    return _format_timetuple_and_zone(now, zone)
+
+
+def make_msgid(idstring=None, domain=None):
+    """Returns a string suitable for RFC 2822 compliant Message-ID, e.g:
+
+    <142480216486.20800.16526388040877946887@nightshade.la.mastaler.com>
+
+    Optional idstring if given is a string used to strengthen the
+    uniqueness of the message id.  Optional domain if given provides the
+    portion of the message id after the '@'.  It defaults to the locally
+    defined hostname.
+    """
+    timeval = int(time.time()*100)
+    pid = os.getpid()
+    randint = random.getrandbits(64)
+    if idstring is None:
+        idstring = ''
+    else:
+        idstring = '.' + idstring
+    if domain is None:
+        domain = socket.getfqdn()
+    msgid = '<%d.%d.%d%s@%s>' % (timeval, pid, randint, idstring, domain)
+    return msgid
+
+
+def parsedate_to_datetime(data):
+    *dtuple, tz = _parsedate_tz(data)
+    if tz is None:
+        return datetime.datetime(*dtuple[:6])
+    return datetime.datetime(*dtuple[:6],
+            tzinfo=datetime.timezone(datetime.timedelta(seconds=tz)))
+
+
+def parseaddr(addr):
+    addrs = _AddressList(addr).addresslist
+    if not addrs:
+        return '', ''
+    return addrs[0]
+
+
+# rfc822.unquote() doesn't properly de-backslash-ify in Python pre-2.3.
+def unquote(str):
+    """Remove quotes from a string."""
+    if len(str) > 1:
+        if str.startswith('"') and str.endswith('"'):
+            return str[1:-1].replace('\\\\', '\\').replace('\\"', '"')
+        if str.startswith('<') and str.endswith('>'):
+            return str[1:-1]
+    return str
+
+
+
+# RFC2231-related functions - parameter encoding and decoding
+def decode_rfc2231(s):
+    """Decode string according to RFC 2231"""
+    parts = s.split(TICK, 2)
+    if len(parts) <= 2:
+        return None, None, s
+    return parts
+
+
+def encode_rfc2231(s, charset=None, language=None):
+    """Encode string according to RFC 2231.
+
+    If neither charset nor language is given, then s is returned as-is.  If
+    charset is given but not language, the string is encoded using the empty
+    string for language.
+    """
+    s = urllib.parse.quote(s, safe='', encoding=charset or 'ascii')
+    if charset is None and language is None:
+        return s
+    if language is None:
+        language = ''
+    return "%s'%s'%s" % (charset, language, s)
+
+
+rfc2231_continuation = re.compile(r'^(?P<name>\w+)\*((?P<num>[0-9]+)\*?)?$',
+    re.ASCII)
+
+def decode_params(params):
+    """Decode parameters list according to RFC 2231.
+
+    params is a sequence of 2-tuples containing (param name, string value).
+    """
+    # Copy params so we don't mess with the original
+    params = params[:]
+    new_params = []
+    # Map parameter's name to a list of continuations.  The values are a
+    # 3-tuple of the continuation number, the string value, and a flag
+    # specifying whether a particular segment is %-encoded.
+    rfc2231_params = {}
+    name, value = params.pop(0)
+    new_params.append((name, value))
+    while params:
+        name, value = params.pop(0)
+        if name.endswith('*'):
+            encoded = True
+        else:
+            encoded = False
+        value = unquote(value)
+        mo = rfc2231_continuation.match(name)
+        if mo:
+            name, num = mo.group('name', 'num')
+            if num is not None:
+                num = int(num)
+            rfc2231_params.setdefault(name, []).append((num, value, encoded))
+        else:
+            new_params.append((name, '"%s"' % quote(value)))
+    if rfc2231_params:
+        for name, continuations in rfc2231_params.items():
+            value = []
+            extended = False
+            # Sort by number
+            continuations.sort()
+            # And now append all values in numerical order, converting
+            # %-encodings for the encoded segments.  If any of the
+            # continuation names ends in a *, then the entire string, after
+            # decoding segments and concatenating, must have the charset and
+            # language specifiers at the beginning of the string.
+            for num, s, encoded in continuations:
+                if encoded:
+                    # Decode as "latin-1", so the characters in s directly
+                    # represent the percent-encoded octet values.
+                    # collapse_rfc2231_value treats this as an octet sequence.
+                    s = urllib.parse.unquote(s, encoding="latin-1")
+                    extended = True
+                value.append(s)
+            value = quote(EMPTYSTRING.join(value))
+            if extended:
+                charset, language, value = decode_rfc2231(value)
+                new_params.append((name, (charset, language, '"%s"' % value)))
+            else:
+                new_params.append((name, '"%s"' % value))
+    return new_params
+
+def collapse_rfc2231_value(value, errors='replace',
+                           fallback_charset='us-ascii'):
+    if not isinstance(value, tuple) or len(value) != 3:
+        return unquote(value)
+    # While value comes to us as a unicode string, we need it to be a bytes
+    # object.  We do not want bytes() normal utf-8 decoder, we want a straight
+    # interpretation of the string as character bytes.
+    charset, language, text = value
+    if charset is None:
+        # Issue 17369: if charset/lang is None, decode_rfc2231 couldn't parse
+        # the value, so use the fallback_charset.
+        charset = fallback_charset
+    rawbytes = bytes(text, 'raw-unicode-escape')
+    try:
+        return str(rawbytes, charset, errors)
+    except LookupError:
+        # charset is not a known codec.
+        return unquote(text)
+
+
+#
+# datetime doesn't provide a localtime function yet, so provide one.  Code
+# adapted from the patch in issue 9527.  This may not be perfect, but it is
+# better than not having it.
+#
+
+def localtime(dt=None, isdst=-1):
+    """Return local time as an aware datetime object.
+
+    If called without arguments, return current time.  Otherwise *dt*
+    argument should be a datetime instance, and it is converted to the
+    local time zone according to the system time zone database.  If *dt* is
+    naive (that is, dt.tzinfo is None), it is assumed to be in local time.
+    In this case, a positive or zero value for *isdst* causes localtime to
+    presume initially that summer time (for example, Daylight Saving Time)
+    is or is not (respectively) in effect for the specified time.  A
+    negative value for *isdst* causes the localtime() function to attempt
+    to divine whether summer time is in effect for the specified time.
+
+    """
+    if dt is None:
+        return datetime.datetime.now(datetime.timezone.utc).astimezone()
+    if dt.tzinfo is not None:
+        return dt.astimezone()
+    # We have a naive datetime.  Convert to a (localtime) timetuple and pass to
+    # system mktime together with the isdst hint.  System mktime will return
+    # seconds since epoch.
+    tm = dt.timetuple()[:-1] + (isdst,)
+    seconds = time.mktime(tm)
+    localtm = time.localtime(seconds)
+    try:
+        delta = datetime.timedelta(seconds=localtm.tm_gmtoff)
+        tz = datetime.timezone(delta, localtm.tm_zone)
+    except AttributeError:
+        # Compute UTC offset and compare with the value implied by tm_isdst.
+        # If the values match, use the zone name implied by tm_isdst.
+        delta = dt - datetime.datetime(*time.gmtime(seconds)[:6])
+        dst = time.daylight and localtm.tm_isdst > 0
+        gmtoff = -(time.altzone if dst else time.timezone)
+        if delta == datetime.timedelta(seconds=gmtoff):
+            tz = datetime.timezone(delta, time.tzname[dst])
+        else:
+            tz = datetime.timezone(delta)
+    return dt.replace(tzinfo=tz)
diff --git a/Lib/getopt.py b/Lib/getopt.py
new file mode 100644
index 000000000..9d4cab1ba
--- /dev/null
+++ b/Lib/getopt.py
@@ -0,0 +1,215 @@
+"""Parser for command line options.
+
+This module helps scripts to parse the command line arguments in
+sys.argv.  It supports the same conventions as the Unix getopt()
+function (including the special meanings of arguments of the form `-'
+and `--').  Long options similar to those supported by GNU software
+may be used as well via an optional third argument.  This module
+provides two functions and an exception:
+
+getopt() -- Parse command line options
+gnu_getopt() -- Like getopt(), but allow option and non-option arguments
+to be intermixed.
+GetoptError -- exception (class) raised with 'opt' attribute, which is the
+option involved with the exception.
+"""
+
+# Long option support added by Lars Wirzenius <liw@iki.fi>.
+#
+# Gerrit Holl <gerrit@nl.linux.org> moved the string-based exceptions
+# to class-based exceptions.
+#
+# Peter Åstrand <astrand@lysator.liu.se> added gnu_getopt().
+#
+# TODO for gnu_getopt():
+#
+# - GNU getopt_long_only mechanism
+# - allow the caller to specify ordering
+# - RETURN_IN_ORDER option
+# - GNU extension with '-' as first character of option string
+# - optional arguments, specified by double colons
+# - an option string with a W followed by semicolon should
+#   treat "-W foo" as "--foo"
+
+__all__ = ["GetoptError","error","getopt","gnu_getopt"]
+
+import os
+try:
+    from gettext import gettext as _
+except ImportError:
+    # Bootstrapping Python: gettext's dependencies not built yet
+    def _(s): return s
+
+class GetoptError(Exception):
+    opt = ''
+    msg = ''
+    def __init__(self, msg, opt=''):
+        self.msg = msg
+        self.opt = opt
+        Exception.__init__(self, msg, opt)
+
+    def __str__(self):
+        return self.msg
+
+error = GetoptError # backward compatibility
+
+def getopt(args, shortopts, longopts = []):
+    """getopt(args, options[, long_options]) -> opts, args
+
+    Parses command line options and parameter list.  args is the
+    argument list to be parsed, without the leading reference to the
+    running program.  Typically, this means "sys.argv[1:]".  shortopts
+    is the string of option letters that the script wants to
+    recognize, with options that require an argument followed by a
+    colon (i.e., the same format that Unix getopt() uses).  If
+    specified, longopts is a list of strings with the names of the
+    long options which should be supported.  The leading '--'
+    characters should not be included in the option name.  Options
+    which require an argument should be followed by an equal sign
+    ('=').
+
+    The return value consists of two elements: the first is a list of
+    (option, value) pairs; the second is the list of program arguments
+    left after the option list was stripped (this is a trailing slice
+    of the first argument).  Each option-and-value pair returned has
+    the option as its first element, prefixed with a hyphen (e.g.,
+    '-x'), and the option argument as its second element, or an empty
+    string if the option has no argument.  The options occur in the
+    list in the same order in which they were found, thus allowing
+    multiple occurrences.  Long and short options may be mixed.
+
+    """
+
+    opts = []
+    if type(longopts) == type(""):
+        longopts = [longopts]
+    else:
+        longopts = list(longopts)
+    while args and args[0].startswith('-') and args[0] != '-':
+        if args[0] == '--':
+            args = args[1:]
+            break
+        if args[0].startswith('--'):
+            opts, args = do_longs(opts, args[0][2:], longopts, args[1:])
+        else:
+            opts, args = do_shorts(opts, args[0][1:], shortopts, args[1:])
+
+    return opts, args
+
+def gnu_getopt(args, shortopts, longopts = []):
+    """getopt(args, options[, long_options]) -> opts, args
+
+    This function works like getopt(), except that GNU style scanning
+    mode is used by default. This means that option and non-option
+    arguments may be intermixed. The getopt() function stops
+    processing options as soon as a non-option argument is
+    encountered.
+
+    If the first character of the option string is `+', or if the
+    environment variable POSIXLY_CORRECT is set, then option
+    processing stops as soon as a non-option argument is encountered.
+
+    """
+
+    opts = []
+    prog_args = []
+    if isinstance(longopts, str):
+        longopts = [longopts]
+    else:
+        longopts = list(longopts)
+
+    # Allow options after non-option arguments?
+    if shortopts.startswith('+'):
+        shortopts = shortopts[1:]
+        all_options_first = True
+    elif os.environ.get("POSIXLY_CORRECT"):
+        all_options_first = True
+    else:
+        all_options_first = False
+
+    while args:
+        if args[0] == '--':
+            prog_args += args[1:]
+            break
+
+        if args[0][:2] == '--':
+            opts, args = do_longs(opts, args[0][2:], longopts, args[1:])
+        elif args[0][:1] == '-' and args[0] != '-':
+            opts, args = do_shorts(opts, args[0][1:], shortopts, args[1:])
+        else:
+            if all_options_first:
+                prog_args += args
+                break
+            else:
+                prog_args.append(args[0])
+                args = args[1:]
+
+    return opts, prog_args
+
+def do_longs(opts, opt, longopts, args):
+    try:
+        i = opt.index('=')
+    except ValueError:
+        optarg = None
+    else:
+        opt, optarg = opt[:i], opt[i+1:]
+
+    has_arg, opt = long_has_args(opt, longopts)
+    if has_arg:
+        if optarg is None:
+            if not args:
+                raise GetoptError(_('option --%s requires argument') % opt, opt)
+            optarg, args = args[0], args[1:]
+    elif optarg is not None:
+        raise GetoptError(_('option --%s must not have an argument') % opt, opt)
+    opts.append(('--' + opt, optarg or ''))
+    return opts, args
+
+# Return:
+#   has_arg?
+#   full option name
+def long_has_args(opt, longopts):
+    possibilities = [o for o in longopts if o.startswith(opt)]
+    if not possibilities:
+        raise GetoptError(_('option --%s not recognized') % opt, opt)
+    # Is there an exact match?
+    if opt in possibilities:
+        return False, opt
+    elif opt + '=' in possibilities:
+        return True, opt
+    # No exact match, so better be unique.
+    if len(possibilities) > 1:
+        # XXX since possibilities contains all valid continuations, might be
+        # nice to work them into the error msg
+        raise GetoptError(_('option --%s not a unique prefix') % opt, opt)
+    assert len(possibilities) == 1
+    unique_match = possibilities[0]
+    has_arg = unique_match.endswith('=')
+    if has_arg:
+        unique_match = unique_match[:-1]
+    return has_arg, unique_match
+
+def do_shorts(opts, optstring, shortopts, args):
+    while optstring != '':
+        opt, optstring = optstring[0], optstring[1:]
+        if short_has_arg(opt, shortopts):
+            if optstring == '':
+                if not args:
+                    raise GetoptError(_('option -%s requires argument') % opt,
+                                      opt)
+                optstring, args = args[0], args[1:]
+            optarg, optstring = optstring, ''
+        else:
+            optarg = ''
+        opts.append(('-' + opt, optarg))
+    return opts, args
+
+def short_has_arg(opt, shortopts):
+    for i in range(len(shortopts)):
+        if opt == shortopts[i] != ':':
+            return shortopts.startswith(':', i+1)
+    raise GetoptError(_('option -%s not recognized') % opt, opt)
+
+if __name__ == '__main__':
+    import sys
+    print(getopt(sys.argv[1:], "a:b", ["alpha=", "beta"]))
diff --git a/Lib/os.py b/Lib/os.py
index cb49481df..59a55c99b 100644
--- a/Lib/os.py
+++ b/Lib/os.py
@@ -1,8 +1,55 @@
-import sys
+r"""OS routines for NT or Posix depending on what system we're on.
 
+This exports:
+  - all functions from posix or nt, e.g. unlink, stat, etc.
+  - os.path is either posixpath or ntpath
+  - os.name is either 'posix' or 'nt'
+  - os.curdir is a string representing the current directory (always '.')
+  - os.pardir is a string representing the parent directory (always '..')
+  - os.sep is the (or a most common) pathname separator ('/' or '\\')
+  - os.extsep is the extension separator (always '.')
+  - os.altsep is the alternate pathname separator (None or '/')
+  - os.pathsep is the component separator used in $PATH etc
+  - os.linesep is the line separator in text files ('\r' or '\n' or '\r\n')
+  - os.defpath is the default search path for executables
+  - os.devnull is the file path of the null device ('/dev/null', etc.)
+
+Programs that import and use 'os' stand a better chance of being
+portable between different platforms.  Of course, they must then
+only use functions that are defined by all platforms (e.g., unlink
+and opendir), and leave all pathname manipulation to os.path
+(e.g., split and join).
+"""
+
+#'
+import abc
+import sys, errno
+import stat as st
+
+_names = sys.builtin_module_names
+
+# Note:  more names are added to __all__ later.
+__all__ = ["altsep", "curdir", "pardir", "sep", "pathsep", "linesep",
+           "defpath", "name", "path", "devnull", "SEEK_SET", "SEEK_CUR",
+           "SEEK_END", "fsencode", "fsdecode", "get_exec_path", "fdopen",
+           "popen", "extsep"]
+
+def _exists(name):
+    return name in globals()
+
+def _get_exports_list(module):
+    try:
+        return list(module.__all__)
+    except AttributeError:
+        return [n for n in dir(module) if n[0] != '_']
+
+import _os
 from _os import *
+__all__.extend(_get_exports_list(_os))
+del _os
 
-
+# Any new dependencies of the os module and/or changes in path separator
+# requires updating importlib as well.
 if name == 'nt':
     linesep = '\r\n'
     import ntpath as path
@@ -15,6 +62,472 @@ sys.modules['os.path'] = path
 from os.path import (curdir, pardir, sep, pathsep, defpath, extsep, altsep,
     devnull)
 
+# Python uses fixed values for the SEEK_ constants; they are mapped
+# to native constants if necessary in posixmodule.c
+# Other possible SEEK values are directly imported from posixmodule.c
+SEEK_SET = 0
+SEEK_CUR = 1
+SEEK_END = 2
+
+# Super directory utilities.
+# (Inspired by Eric Raymond; the doc strings are mostly his)
+
+def makedirs(name, mode=0o777, exist_ok=False):
+    """makedirs(name [, mode=0o777][, exist_ok=False])
+
+    Super-mkdir; create a leaf directory and all intermediate ones.  Works like
+    mkdir, except that any intermediate path segment (not just the rightmost)
+    will be created if it does not exist. If the target directory already
+    exists, raise an OSError if exist_ok is False. Otherwise no exception is
+    raised.  This is recursive.
+
+    """
+    head, tail = path.split(name)
+    if not tail:
+        head, tail = path.split(head)
+    if head and tail and not path.exists(head):
+        try:
+            makedirs(head, mode, exist_ok)
+        except FileExistsError:
+            # Defeats race condition when another thread created the path
+            pass
+        cdir = curdir
+        if isinstance(tail, bytes):
+            cdir = bytes(curdir, 'ASCII')
+        if tail == cdir:           # xxx/newdir/. exists if xxx/newdir exists
+            return
+    try:
+        mkdir(name, mode)
+    except OSError:
+        # Cannot rely on checking for EEXIST, since the operating system
+        # could give priority to other errors like EACCES or EROFS
+        if not exist_ok or not path.isdir(name):
+            raise
+
+def removedirs(name):
+    """removedirs(name)
+
+    Super-rmdir; remove a leaf directory and all empty intermediate
+    ones.  Works like rmdir except that, if the leaf directory is
+    successfully removed, directories corresponding to rightmost path
+    segments will be pruned away until either the whole path is
+    consumed or an error occurs.  Errors during this latter phase are
+    ignored -- they generally mean that a directory was not empty.
+
+    """
+    rmdir(name)
+    head, tail = path.split(name)
+    if not tail:
+        head, tail = path.split(head)
+    while head and tail:
+        try:
+            rmdir(head)
+        except OSError:
+            break
+        head, tail = path.split(head)
+
+def renames(old, new):
+    """renames(old, new)
+
+    Super-rename; create directories as necessary and delete any left
+    empty.  Works like rename, except creation of any intermediate
+    directories needed to make the new pathname good is attempted
+    first.  After the rename, directories corresponding to rightmost
+    path segments of the old name will be pruned until either the
+    whole path is consumed or a nonempty directory is found.
+
+    Note: this function can fail with the new directory structure made
+    if you lack permissions needed to unlink the leaf directory or
+    file.
+
+    """
+    head, tail = path.split(new)
+    if head and tail and not path.exists(head):
+        makedirs(head)
+    rename(old, new)
+    head, tail = path.split(old)
+    if head and tail:
+        try:
+            removedirs(head)
+        except OSError:
+            pass
+
+__all__.extend(["makedirs", "removedirs", "renames"])
+
+def walk(top, topdown=True, onerror=None, followlinks=False):
+    """Directory tree generator.
+
+    For each directory in the directory tree rooted at top (including top
+    itself, but excluding '.' and '..'), yields a 3-tuple
+
+        dirpath, dirnames, filenames
+
+    dirpath is a string, the path to the directory.  dirnames is a list of
+    the names of the subdirectories in dirpath (excluding '.' and '..').
+    filenames is a list of the names of the non-directory files in dirpath.
+    Note that the names in the lists are just names, with no path components.
+    To get a full path (which begins with top) to a file or directory in
+    dirpath, do os.path.join(dirpath, name).
+
+    If optional arg 'topdown' is true or not specified, the triple for a
+    directory is generated before the triples for any of its subdirectories
+    (directories are generated top down).  If topdown is false, the triple
+    for a directory is generated after the triples for all of its
+    subdirectories (directories are generated bottom up).
+
+    When topdown is true, the caller can modify the dirnames list in-place
+    (e.g., via del or slice assignment), and walk will only recurse into the
+    subdirectories whose names remain in dirnames; this can be used to prune the
+    search, or to impose a specific order of visiting.  Modifying dirnames when
+    topdown is false is ineffective, since the directories in dirnames have
+    already been generated by the time dirnames itself is generated. No matter
+    the value of topdown, the list of subdirectories is retrieved before the
+    tuples for the directory and its subdirectories are generated.
+
+    By default errors from the os.scandir() call are ignored.  If
+    optional arg 'onerror' is specified, it should be a function; it
+    will be called with one argument, an OSError instance.  It can
+    report the error to continue with the walk, or raise the exception
+    to abort the walk.  Note that the filename is available as the
+    filename attribute of the exception object.
+
+    By default, os.walk does not follow symbolic links to subdirectories on
+    systems that support them.  In order to get this functionality, set the
+    optional argument 'followlinks' to true.
+
+    Caution:  if you pass a relative pathname for top, don't change the
+    current working directory between resumptions of walk.  walk never
+    changes the current directory, and assumes that the client doesn't
+    either.
+
+    Example:
+
+    import os
+    from os.path import join, getsize
+    for root, dirs, files in os.walk('python/Lib/email'):
+        print(root, "consumes", end="")
+        print(sum([getsize(join(root, name)) for name in files]), end="")
+        print("bytes in", len(files), "non-directory files")
+        if 'CVS' in dirs:
+            dirs.remove('CVS')  # don't visit CVS directories
+
+    """
+    top = fspath(top)
+    dirs = []
+    nondirs = []
+    walk_dirs = []
+
+    # We may not have read permission for top, in which case we can't
+    # get a list of the files the directory contains.  os.walk
+    # always suppressed the exception then, rather than blow up for a
+    # minor reason when (say) a thousand readable directories are still
+    # left to visit.  That logic is copied here.
+    try:
+        # Note that scandir is global in this module due
+        # to earlier import-*.
+        scandir_it = scandir(top)
+    except OSError as error:
+        if onerror is not None:
+            onerror(error)
+        return
+
+    with scandir_it:
+        while True:
+            try:
+                try:
+                    entry = next(scandir_it)
+                except StopIteration:
+                    break
+            except OSError as error:
+                if onerror is not None:
+                    onerror(error)
+                return
+
+            try:
+                is_dir = entry.is_dir()
+            except OSError:
+                # If is_dir() raises an OSError, consider that the entry is not
+                # a directory, same behaviour than os.path.isdir().
+                is_dir = False
+
+            if is_dir:
+                dirs.append(entry.name)
+            else:
+                nondirs.append(entry.name)
+
+            if not topdown and is_dir:
+                # Bottom-up: recurse into sub-directory, but exclude symlinks to
+                # directories if followlinks is False
+                if followlinks:
+                    walk_into = True
+                else:
+                    try:
+                        is_symlink = entry.is_symlink()
+                    except OSError:
+                        # If is_symlink() raises an OSError, consider that the
+                        # entry is not a symbolic link, same behaviour than
+                        # os.path.islink().
+                        is_symlink = False
+                    walk_into = not is_symlink
+
+                if walk_into:
+                    walk_dirs.append(entry.path)
+
+    # Yield before recursion if going top down
+    if topdown:
+        yield top, dirs, nondirs
+
+        # Recurse into sub-directories
+        islink, join = path.islink, path.join
+        for dirname in dirs:
+            new_path = join(top, dirname)
+            # Issue #23605: os.path.islink() is used instead of caching
+            # entry.is_symlink() result during the loop on os.scandir() because
+            # the caller can replace the directory entry during the "yield"
+            # above.
+            if followlinks or not islink(new_path):
+                yield from walk(new_path, topdown, onerror, followlinks)
+    else:
+        # Recurse into sub-directories
+        for new_path in walk_dirs:
+            yield from walk(new_path, topdown, onerror, followlinks)
+        # Yield after recursion if going bottom up
+        yield top, dirs, nondirs
+
+__all__.append("walk")
+
+if {open, stat} <= supports_dir_fd and {listdir, stat} <= supports_fd:
+
+    def fwalk(top=".", topdown=True, onerror=None, *, follow_symlinks=False, dir_fd=None):
+        """Directory tree generator.
+
+        This behaves exactly like walk(), except that it yields a 4-tuple
+
+            dirpath, dirnames, filenames, dirfd
+
+        `dirpath`, `dirnames` and `filenames` are identical to walk() output,
+        and `dirfd` is a file descriptor referring to the directory `dirpath`.
+
+        The advantage of fwalk() over walk() is that it's safe against symlink
+        races (when follow_symlinks is False).
+
+        If dir_fd is not None, it should be a file descriptor open to a directory,
+          and top should be relative; top will then be relative to that directory.
+          (dir_fd is always supported for fwalk.)
+
+        Caution:
+        Since fwalk() yields file descriptors, those are only valid until the
+        next iteration step, so you should dup() them if you want to keep them
+        for a longer period.
+
+        Example:
+
+        import os
+        for root, dirs, files, rootfd in os.fwalk('python/Lib/email'):
+            print(root, "consumes", end="")
+            print(sum([os.stat(name, dir_fd=rootfd).st_size for name in files]),
+                  end="")
+            print("bytes in", len(files), "non-directory files")
+            if 'CVS' in dirs:
+                dirs.remove('CVS')  # don't visit CVS directories
+        """
+        if not isinstance(top, int) or not hasattr(top, '__index__'):
+            top = fspath(top)
+        # Note: To guard against symlink races, we use the standard
+        # lstat()/open()/fstat() trick.
+        orig_st = stat(top, follow_symlinks=False, dir_fd=dir_fd)
+        topfd = open(top, O_RDONLY, dir_fd=dir_fd)
+        try:
+            if (follow_symlinks or (st.S_ISDIR(orig_st.st_mode) and
+                                    path.samestat(orig_st, stat(topfd)))):
+                yield from _fwalk(topfd, top, topdown, onerror, follow_symlinks)
+        finally:
+            close(topfd)
+
+    def _fwalk(topfd, toppath, topdown, onerror, follow_symlinks):
+        # Note: This uses O(depth of the directory tree) file descriptors: if
+        # necessary, it can be adapted to only require O(1) FDs, see issue
+        # #13734.
+
+        names = listdir(topfd)
+        dirs, nondirs = [], []
+        for name in names:
+            try:
+                # Here, we don't use AT_SYMLINK_NOFOLLOW to be consistent with
+                # walk() which reports symlinks to directories as directories.
+                # We do however check for symlinks before recursing into
+                # a subdirectory.
+                if st.S_ISDIR(stat(name, dir_fd=topfd).st_mode):
+                    dirs.append(name)
+                else:
+                    nondirs.append(name)
+            except OSError:
+                try:
+                    # Add dangling symlinks, ignore disappeared files
+                    if st.S_ISLNK(stat(name, dir_fd=topfd, follow_symlinks=False)
+                                .st_mode):
+                        nondirs.append(name)
+                except OSError:
+                    continue
+
+        if topdown:
+            yield toppath, dirs, nondirs, topfd
+
+        for name in dirs:
+            try:
+                orig_st = stat(name, dir_fd=topfd, follow_symlinks=follow_symlinks)
+                dirfd = open(name, O_RDONLY, dir_fd=topfd)
+            except OSError as err:
+                if onerror is not None:
+                    onerror(err)
+                continue
+            try:
+                if follow_symlinks or path.samestat(orig_st, stat(dirfd)):
+                    dirpath = path.join(toppath, name)
+                    yield from _fwalk(dirfd, dirpath, topdown, onerror, follow_symlinks)
+            finally:
+                close(dirfd)
+
+        if not topdown:
+            yield toppath, dirs, nondirs, topfd
+
+    __all__.append("fwalk")
+
+# Make sure os.environ exists, at least
+try:
+    environ
+except NameError:
+    environ = {}
+
+def execl(file, *args):
+    """execl(file, *args)
+
+    Execute the executable file with argument list args, replacing the
+    current process. """
+    execv(file, args)
+
+def execle(file, *args):
+    """execle(file, *args, env)
+
+    Execute the executable file with argument list args and
+    environment env, replacing the current process. """
+    env = args[-1]
+    execve(file, args[:-1], env)
+
+def execlp(file, *args):
+    """execlp(file, *args)
+
+    Execute the executable file (which is searched for along $PATH)
+    with argument list args, replacing the current process. """
+    execvp(file, args)
+
+def execlpe(file, *args):
+    """execlpe(file, *args, env)
+
+    Execute the executable file (which is searched for along $PATH)
+    with argument list args and environment env, replacing the current
+    process. """
+    env = args[-1]
+    execvpe(file, args[:-1], env)
+
+def execvp(file, args):
+    """execvp(file, args)
+
+    Execute the executable file (which is searched for along $PATH)
+    with argument list args, replacing the current process.
+    args may be a list or tuple of strings. """
+    _execvpe(file, args)
+
+def execvpe(file, args, env):
+    """execvpe(file, args, env)
+
+    Execute the executable file (which is searched for along $PATH)
+    with argument list args and environment env , replacing the
+    current process.
+    args may be a list or tuple of strings. """
+    _execvpe(file, args, env)
+
+__all__.extend(["execl","execle","execlp","execlpe","execvp","execvpe"])
+
+def _execvpe(file, args, env=None):
+    if env is not None:
+        exec_func = execve
+        argrest = (args, env)
+    else:
+        exec_func = execv
+        argrest = (args,)
+        env = environ
+
+    head, tail = path.split(file)
+    if head:
+        exec_func(file, *argrest)
+        return
+    last_exc = saved_exc = None
+    saved_tb = None
+    path_list = get_exec_path(env)
+    if name != 'nt':
+        file = fsencode(file)
+        path_list = map(fsencode, path_list)
+    for dir in path_list:
+        fullname = path.join(dir, file)
+        try:
+            exec_func(fullname, *argrest)
+        except OSError as e:
+            last_exc = e
+            tb = sys.exc_info()[2]
+            if (e.errno != errno.ENOENT and e.errno != errno.ENOTDIR
+                and saved_exc is None):
+                saved_exc = e
+                saved_tb = tb
+    if saved_exc:
+        raise saved_exc.with_traceback(saved_tb)
+    raise last_exc.with_traceback(tb)
+
+
+def get_exec_path(env=None):
+    """Returns the sequence of directories that will be searched for the
+    named executable (similar to a shell) when launching a process.
+
+    *env* must be an environment variable dict or None.  If *env* is None,
+    os.environ will be used.
+    """
+    # Use a local import instead of a global import to limit the number of
+    # modules loaded at startup: the os module is always loaded at startup by
+    # Python. It may also avoid a bootstrap issue.
+    import warnings
+
+    if env is None:
+        env = environ
+
+    # {b'PATH': ...}.get('PATH') and {'PATH': ...}.get(b'PATH') emit a
+    # BytesWarning when using python -b or python -bb: ignore the warning
+    with warnings.catch_warnings():
+        warnings.simplefilter("ignore", BytesWarning)
+
+        try:
+            path_list = env.get('PATH')
+        except TypeError:
+            path_list = None
+
+        if supports_bytes_environ:
+            try:
+                path_listb = env[b'PATH']
+            except (KeyError, TypeError):
+                pass
+            else:
+                if path_list is not None:
+                    raise ValueError(
+                        "env cannot contain 'PATH' and b'PATH' keys")
+                path_list = path_listb
+
+            if path_list is not None and isinstance(path_list, bytes):
+                path_list = fsdecode(path_list)
+
+    if path_list is None:
+        path_list = defpath
+    return path_list.split(pathsep)
+
+
 # Change environ to automatically call putenv(), unsetenv if they exist.
 from _collections_abc import MutableMapping
 
@@ -34,7 +547,6 @@ class _Environ(MutableMapping):
         except KeyError:
             # raise KeyError with the original key value
             raise KeyError(key) from None
-
         return self.decodevalue(value)
 
     def __setitem__(self, key, value):
@@ -53,9 +565,7 @@ class _Environ(MutableMapping):
             raise KeyError(key) from None
 
     def __iter__(self):
-        # list() from dict object is an atomic operation
-        keys = list(self._data)
-        for key in keys:
+        for key in self._data:
             yield self.decodekey(key)
 
     def __len__(self):
@@ -78,46 +588,43 @@ try:
     _putenv = putenv
 except NameError:
     _putenv = lambda key, value: None
-# else:
-#     if "putenv" not in __all__:
-#         __all__.append("putenv")
+else:
+    if "putenv" not in __all__:
+        __all__.append("putenv")
 
 try:
     _unsetenv = unsetenv
 except NameError:
     _unsetenv = lambda key: _putenv(key, "")
-# else:
-#     if "unsetenv" not in __all__:
-#         __all__.append("unsetenv")
+else:
+    if "unsetenv" not in __all__:
+        __all__.append("unsetenv")
 
 def _createenviron():
-    # if name == 'nt':
-    #     # Where Env Var Names Must Be UPPERCASE
-    #     def check_str(value):
-    #         if not isinstance(value, str):
-    #             raise TypeError("str expected, not %s" % type(value).__name__)
-    #         return value
-    #     encode = check_str
-    #     decode = str
-    #     def encodekey(key):
-    #         return encode(key).upper()
-    #     data = {}
-    #     for key, value in environ.items():
-    #         data[encodekey(key)] = value
-    # else:
-    #     # Where Env Var Names Can Be Mixed Case
-    #     encoding = sys.getfilesystemencoding()
-    #     def encode(value):
-    #         if not isinstance(value, str):
-    #             raise TypeError("str expected, not %s" % type(value).__name__)
-    #         return value.encode(encoding, 'surrogateescape')
-    #     def decode(value):
-    #         return value.decode(encoding, 'surrogateescape')
-    #     encodekey = encode
-    decode = str
-    encode = str
-    encodekey = encode
-    data = environ
+    if name == 'nt':
+        # Where Env Var Names Must Be UPPERCASE
+        def check_str(value):
+            if not isinstance(value, str):
+                raise TypeError("str expected, not %s" % type(value).__name__)
+            return value
+        encode = check_str
+        decode = str
+        def encodekey(key):
+            return encode(key).upper()
+        data = {}
+        for key, value in environ.items():
+            data[encodekey(key)] = value
+    else:
+        # Where Env Var Names Can Be Mixed Case
+        encoding = sys.getfilesystemencoding()
+        def encode(value):
+            if not isinstance(value, str):
+                raise TypeError("str expected, not %s" % type(value).__name__)
+            return value.encode(encoding, 'surrogateescape')
+        def decode(value):
+            return value.decode(encoding, 'surrogateescape')
+        encodekey = encode
+        data = environ
     return _Environ(data,
         encodekey, decode,
         encode, decode,
@@ -133,3 +640,308 @@ def getenv(key, default=None):
     The optional second argument can specify an alternate default.
     key, default and the result are str."""
     return environ.get(key, default)
+
+supports_bytes_environ = (name != 'nt')
+__all__.extend(("getenv", "supports_bytes_environ"))
+
+if supports_bytes_environ:
+    def _check_bytes(value):
+        if not isinstance(value, bytes):
+            raise TypeError("bytes expected, not %s" % type(value).__name__)
+        return value
+
+    # bytes environ
+    environb = _Environ(environ._data,
+        _check_bytes, bytes,
+        _check_bytes, bytes,
+        _putenv, _unsetenv)
+    del _check_bytes
+
+    def getenvb(key, default=None):
+        """Get an environment variable, return None if it doesn't exist.
+        The optional second argument can specify an alternate default.
+        key, default and the result are bytes."""
+        return environb.get(key, default)
+
+    __all__.extend(("environb", "getenvb"))
+
+def _fscodec():
+    encoding = sys.getfilesystemencoding()
+    errors = sys.getfilesystemencodeerrors()
+
+    def fsencode(filename):
+        """Encode filename (an os.PathLike, bytes, or str) to the filesystem
+        encoding with 'surrogateescape' error handler, return bytes unchanged.
+        On Windows, use 'strict' error handler if the file system encoding is
+        'mbcs' (which is the default encoding).
+        """
+        filename = fspath(filename)  # Does type-checking of `filename`.
+        if isinstance(filename, str):
+            return filename.encode(encoding, errors)
+        else:
+            return filename
+
+    def fsdecode(filename):
+        """Decode filename (an os.PathLike, bytes, or str) from the filesystem
+        encoding with 'surrogateescape' error handler, return str unchanged. On
+        Windows, use 'strict' error handler if the file system encoding is
+        'mbcs' (which is the default encoding).
+        """
+        filename = fspath(filename)  # Does type-checking of `filename`.
+        if isinstance(filename, bytes):
+            return filename.decode(encoding, errors)
+        else:
+            return filename
+
+    return fsencode, fsdecode
+
+fsencode, fsdecode = _fscodec()
+del _fscodec
+
+# Supply spawn*() (probably only for Unix)
+if _exists("fork") and not _exists("spawnv") and _exists("execv"):
+
+    P_WAIT = 0
+    P_NOWAIT = P_NOWAITO = 1
+
+    __all__.extend(["P_WAIT", "P_NOWAIT", "P_NOWAITO"])
+
+    # XXX Should we support P_DETACH?  I suppose it could fork()**2
+    # and close the std I/O streams.  Also, P_OVERLAY is the same
+    # as execv*()?
+
+    def _spawnvef(mode, file, args, env, func):
+        # Internal helper; func is the exec*() function to use
+        if not isinstance(args, (tuple, list)):
+            raise TypeError('argv must be a tuple or a list')
+        if not args or not args[0]:
+            raise ValueError('argv first element cannot be empty')
+        pid = fork()
+        if not pid:
+            # Child
+            try:
+                if env is None:
+                    func(file, args)
+                else:
+                    func(file, args, env)
+            except:
+                _exit(127)
+        else:
+            # Parent
+            if mode == P_NOWAIT:
+                return pid # Caller is responsible for waiting!
+            while 1:
+                wpid, sts = waitpid(pid, 0)
+                if WIFSTOPPED(sts):
+                    continue
+                elif WIFSIGNALED(sts):
+                    return -WTERMSIG(sts)
+                elif WIFEXITED(sts):
+                    return WEXITSTATUS(sts)
+                else:
+                    raise OSError("Not stopped, signaled or exited???")
+
+    def spawnv(mode, file, args):
+        """spawnv(mode, file, args) -> integer
+
+Execute file with arguments from args in a subprocess.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        return _spawnvef(mode, file, args, None, execv)
+
+    def spawnve(mode, file, args, env):
+        """spawnve(mode, file, args, env) -> integer
+
+Execute file with arguments from args in a subprocess with the
+specified environment.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        return _spawnvef(mode, file, args, env, execve)
+
+    # Note: spawnvp[e] is't currently supported on Windows
+
+    def spawnvp(mode, file, args):
+        """spawnvp(mode, file, args) -> integer
+
+Execute file (which is looked for along $PATH) with arguments from
+args in a subprocess.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        return _spawnvef(mode, file, args, None, execvp)
+
+    def spawnvpe(mode, file, args, env):
+        """spawnvpe(mode, file, args, env) -> integer
+
+Execute file (which is looked for along $PATH) with arguments from
+args in a subprocess with the supplied environment.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        return _spawnvef(mode, file, args, env, execvpe)
+
+
+    __all__.extend(["spawnv", "spawnve", "spawnvp", "spawnvpe"])
+
+
+if _exists("spawnv"):
+    # These aren't supplied by the basic Windows code
+    # but can be easily implemented in Python
+
+    def spawnl(mode, file, *args):
+        """spawnl(mode, file, *args) -> integer
+
+Execute file with arguments from args in a subprocess.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        return spawnv(mode, file, args)
+
+    def spawnle(mode, file, *args):
+        """spawnle(mode, file, *args, env) -> integer
+
+Execute file with arguments from args in a subprocess with the
+supplied environment.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        env = args[-1]
+        return spawnve(mode, file, args[:-1], env)
+
+
+    __all__.extend(["spawnl", "spawnle"])
+
+
+if _exists("spawnvp"):
+    # At the moment, Windows doesn't implement spawnvp[e],
+    # so it won't have spawnlp[e] either.
+    def spawnlp(mode, file, *args):
+        """spawnlp(mode, file, *args) -> integer
+
+Execute file (which is looked for along $PATH) with arguments from
+args in a subprocess with the supplied environment.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        return spawnvp(mode, file, args)
+
+    def spawnlpe(mode, file, *args):
+        """spawnlpe(mode, file, *args, env) -> integer
+
+Execute file (which is looked for along $PATH) with arguments from
+args in a subprocess with the supplied environment.
+If mode == P_NOWAIT return the pid of the process.
+If mode == P_WAIT return the process's exit code if it exits normally;
+otherwise return -SIG, where SIG is the signal that killed it. """
+        env = args[-1]
+        return spawnvpe(mode, file, args[:-1], env)
+
+
+    __all__.extend(["spawnlp", "spawnlpe"])
+
+
+# Supply os.popen()
+def popen(cmd, mode="r", buffering=-1):
+    if not isinstance(cmd, str):
+        raise TypeError("invalid cmd type (%s, expected string)" % type(cmd))
+    if mode not in ("r", "w"):
+        raise ValueError("invalid mode %r" % mode)
+    if buffering == 0 or buffering is None:
+        raise ValueError("popen() does not support unbuffered streams")
+    import subprocess, io
+    if mode == "r":
+        proc = subprocess.Popen(cmd,
+                                shell=True,
+                                stdout=subprocess.PIPE,
+                                bufsize=buffering)
+        return _wrap_close(io.TextIOWrapper(proc.stdout), proc)
+    else:
+        proc = subprocess.Popen(cmd,
+                                shell=True,
+                                stdin=subprocess.PIPE,
+                                bufsize=buffering)
+        return _wrap_close(io.TextIOWrapper(proc.stdin), proc)
+
+# Helper for popen() -- a proxy for a file whose close waits for the process
+class _wrap_close:
+    def __init__(self, stream, proc):
+        self._stream = stream
+        self._proc = proc
+    def close(self):
+        self._stream.close()
+        returncode = self._proc.wait()
+        if returncode == 0:
+            return None
+        if name == 'nt':
+            return returncode
+        else:
+            return returncode << 8  # Shift left to match old behavior
+    def __enter__(self):
+        return self
+    def __exit__(self, *args):
+        self.close()
+    def __getattr__(self, name):
+        return getattr(self._stream, name)
+    def __iter__(self):
+        return iter(self._stream)
+
+# Supply os.fdopen()
+def fdopen(fd, *args, **kwargs):
+    if not isinstance(fd, int):
+        raise TypeError("invalid fd type (%s, expected integer)" % type(fd))
+    import io
+    return io.open(fd, *args, **kwargs)
+
+
+# For testing purposes, make sure the function is available when the C
+# implementation exists.
+def _fspath(path):
+    """Return the path representation of a path-like object.
+
+    If str or bytes is passed in, it is returned unchanged. Otherwise the
+    os.PathLike interface is used to get the path representation. If the
+    path representation is not str or bytes, TypeError is raised. If the
+    provided path is not str, bytes, or os.PathLike, TypeError is raised.
+    """
+    if isinstance(path, (str, bytes)):
+        return path
+
+    # Work from the object's type to match method resolution of other magic
+    # methods.
+    path_type = type(path)
+    try:
+        path_repr = path_type.__fspath__(path)
+    except AttributeError:
+        if hasattr(path_type, '__fspath__'):
+            raise
+        else:
+            raise TypeError("expected str, bytes or os.PathLike object, "
+                            "not " + path_type.__name__)
+    if isinstance(path_repr, (str, bytes)):
+        return path_repr
+    else:
+        raise TypeError("expected {}.__fspath__() to return str or bytes, "
+                        "not {}".format(path_type.__name__,
+                                        type(path_repr).__name__))
+
+# If there is no C implementation, make the pure Python version the
+# implementation as transparently as possible.
+if not _exists('fspath'):
+    fspath = _fspath
+    fspath.__name__ = "fspath"
+
+
+class PathLike(abc.ABC):
+
+    """Abstract base class for implementing the file system path protocol."""
+
+    @abc.abstractmethod
+    def __fspath__(self):
+        """Return the file system path representation of the object."""
+        raise NotImplementedError
+
+    @classmethod
+    def __subclasshook__(cls, subclass):
+        return hasattr(subclass, '__fspath__')
diff --git a/Lib/py_compile.py b/Lib/py_compile.py
new file mode 100644
index 000000000..11c5b505c
--- /dev/null
+++ b/Lib/py_compile.py
@@ -0,0 +1,186 @@
+"""Routine to "compile" a .py file to a .pyc file.
+
+This module has intimate knowledge of the format of .pyc files.
+"""
+
+import importlib._bootstrap_external
+import importlib.machinery
+import importlib.util
+import os
+import os.path
+import sys
+import traceback
+
+__all__ = ["compile", "main", "PyCompileError"]
+
+
+class PyCompileError(Exception):
+    """Exception raised when an error occurs while attempting to
+    compile the file.
+
+    To raise this exception, use
+
+        raise PyCompileError(exc_type,exc_value,file[,msg])
+
+    where
+
+        exc_type:   exception type to be used in error message
+                    type name can be accesses as class variable
+                    'exc_type_name'
+
+        exc_value:  exception value to be used in error message
+                    can be accesses as class variable 'exc_value'
+
+        file:       name of file being compiled to be used in error message
+                    can be accesses as class variable 'file'
+
+        msg:        string message to be written as error message
+                    If no value is given, a default exception message will be
+                    given, consistent with 'standard' py_compile output.
+                    message (or default) can be accesses as class variable
+                    'msg'
+
+    """
+
+    def __init__(self, exc_type, exc_value, file, msg=''):
+        exc_type_name = exc_type.__name__
+        if exc_type is SyntaxError:
+            tbtext = ''.join(traceback.format_exception_only(
+                exc_type, exc_value))
+            errmsg = tbtext.replace('File "<string>"', 'File "%s"' % file)
+        else:
+            errmsg = "Sorry: %s: %s" % (exc_type_name,exc_value)
+
+        Exception.__init__(self,msg or errmsg,exc_type_name,exc_value,file)
+
+        self.exc_type_name = exc_type_name
+        self.exc_value = exc_value
+        self.file = file
+        self.msg = msg or errmsg
+
+    def __str__(self):
+        return self.msg
+
+
+def compile(file, cfile=None, dfile=None, doraise=False, optimize=-1):
+    """Byte-compile one Python source file to Python bytecode.
+
+    :param file: The source file name.
+    :param cfile: The target byte compiled file name.  When not given, this
+        defaults to the PEP 3147/PEP 488 location.
+    :param dfile: Purported file name, i.e. the file name that shows up in
+        error messages.  Defaults to the source file name.
+    :param doraise: Flag indicating whether or not an exception should be
+        raised when a compile error is found.  If an exception occurs and this
+        flag is set to False, a string indicating the nature of the exception
+        will be printed, and the function will return to the caller. If an
+        exception occurs and this flag is set to True, a PyCompileError
+        exception will be raised.
+    :param optimize: The optimization level for the compiler.  Valid values
+        are -1, 0, 1 and 2.  A value of -1 means to use the optimization
+        level of the current interpreter, as given by -O command line options.
+
+    :return: Path to the resulting byte compiled file.
+
+    Note that it isn't necessary to byte-compile Python modules for
+    execution efficiency -- Python itself byte-compiles a module when
+    it is loaded, and if it can, writes out the bytecode to the
+    corresponding .pyc file.
+
+    However, if a Python installation is shared between users, it is a
+    good idea to byte-compile all modules upon installation, since
+    other users may not be able to write in the source directories,
+    and thus they won't be able to write the .pyc file, and then
+    they would be byte-compiling every module each time it is loaded.
+    This can slow down program start-up considerably.
+
+    See compileall.py for a script/module that uses this module to
+    byte-compile all installed files (or all files in selected
+    directories).
+
+    Do note that FileExistsError is raised if cfile ends up pointing at a
+    non-regular file or symlink. Because the compilation uses a file renaming,
+    the resulting file would be regular and thus not the same type of file as
+    it was previously.
+    """
+    if cfile is None:
+        if optimize >= 0:
+            optimization = optimize if optimize >= 1 else ''
+            cfile = importlib.util.cache_from_source(file,
+                                                     optimization=optimization)
+        else:
+            cfile = importlib.util.cache_from_source(file)
+    if os.path.islink(cfile):
+        msg = ('{} is a symlink and will be changed into a regular file if '
+               'import writes a byte-compiled file to it')
+        raise FileExistsError(msg.format(cfile))
+    elif os.path.exists(cfile) and not os.path.isfile(cfile):
+        msg = ('{} is a non-regular file and will be changed into a regular '
+               'one if import writes a byte-compiled file to it')
+        raise FileExistsError(msg.format(cfile))
+    loader = importlib.machinery.SourceFileLoader('<py_compile>', file)
+    source_bytes = loader.get_data(file)
+    try:
+        code = loader.source_to_code(source_bytes, dfile or file,
+                                     _optimize=optimize)
+    except Exception as err:
+        py_exc = PyCompileError(err.__class__, err, dfile or file)
+        if doraise:
+            raise py_exc
+        else:
+            sys.stderr.write(py_exc.msg + '\n')
+            return
+    try:
+        dirname = os.path.dirname(cfile)
+        if dirname:
+            os.makedirs(dirname)
+    except FileExistsError:
+        pass
+    source_stats = loader.path_stats(file)
+    bytecode = importlib._bootstrap_external._code_to_bytecode(
+            code, source_stats['mtime'], source_stats['size'])
+    mode = importlib._bootstrap_external._calc_mode(file)
+    importlib._bootstrap_external._write_atomic(cfile, bytecode, mode)
+    return cfile
+
+
+def main(args=None):
+    """Compile several source files.
+
+    The files named in 'args' (or on the command line, if 'args' is
+    not specified) are compiled and the resulting bytecode is cached
+    in the normal manner.  This function does not search a directory
+    structure to locate source files; it only compiles files named
+    explicitly.  If '-' is the only parameter in args, the list of
+    files is taken from standard input.
+
+    """
+    if args is None:
+        args = sys.argv[1:]
+    rv = 0
+    if args == ['-']:
+        while True:
+            filename = sys.stdin.readline()
+            if not filename:
+                break
+            filename = filename.rstrip('\n')
+            try:
+                compile(filename, doraise=True)
+            except PyCompileError as error:
+                rv = 1
+                sys.stderr.write("%s\n" % error.msg)
+            except OSError as error:
+                rv = 1
+                sys.stderr.write("%s\n" % error)
+    else:
+        for filename in args:
+            try:
+                compile(filename, doraise=True)
+            except PyCompileError as error:
+                # return value to indicate at least one failure
+                rv = 1
+                sys.stderr.write("%s\n" % error.msg)
+    return rv
+
+if __name__ == "__main__":
+    sys.exit(main())
diff --git a/Lib/sysconfig.py b/Lib/sysconfig.py
new file mode 100644
index 000000000..9314e71a2
--- /dev/null
+++ b/Lib/sysconfig.py
@@ -0,0 +1,721 @@
+"""Access to Python's configuration information."""
+
+import os
+import sys
+from os.path import pardir, realpath
+
+__all__ = [
+    'get_config_h_filename',
+    'get_config_var',
+    'get_config_vars',
+    'get_makefile_filename',
+    'get_path',
+    'get_path_names',
+    'get_paths',
+    'get_platform',
+    'get_python_version',
+    'get_scheme_names',
+    'parse_config_h',
+]
+
+_INSTALL_SCHEMES = {
+    'posix_prefix': {
+        'stdlib': '{installed_base}/lib/python{py_version_short}',
+        'platstdlib': '{platbase}/lib/python{py_version_short}',
+        'purelib': '{base}/lib/python{py_version_short}/site-packages',
+        'platlib': '{platbase}/lib/python{py_version_short}/site-packages',
+        'include':
+            '{installed_base}/include/python{py_version_short}{abiflags}',
+        'platinclude':
+            '{installed_platbase}/include/python{py_version_short}{abiflags}',
+        'scripts': '{base}/bin',
+        'data': '{base}',
+        },
+    'posix_home': {
+        'stdlib': '{installed_base}/lib/python',
+        'platstdlib': '{base}/lib/python',
+        'purelib': '{base}/lib/python',
+        'platlib': '{base}/lib/python',
+        'include': '{installed_base}/include/python',
+        'platinclude': '{installed_base}/include/python',
+        'scripts': '{base}/bin',
+        'data': '{base}',
+        },
+    'nt': {
+        'stdlib': '{installed_base}/Lib',
+        'platstdlib': '{base}/Lib',
+        'purelib': '{base}/Lib/site-packages',
+        'platlib': '{base}/Lib/site-packages',
+        'include': '{installed_base}/Include',
+        'platinclude': '{installed_base}/Include',
+        'scripts': '{base}/Scripts',
+        'data': '{base}',
+        },
+    'nt_user': {
+        'stdlib': '{userbase}/Python{py_version_nodot}',
+        'platstdlib': '{userbase}/Python{py_version_nodot}',
+        'purelib': '{userbase}/Python{py_version_nodot}/site-packages',
+        'platlib': '{userbase}/Python{py_version_nodot}/site-packages',
+        'include': '{userbase}/Python{py_version_nodot}/Include',
+        'scripts': '{userbase}/Python{py_version_nodot}/Scripts',
+        'data': '{userbase}',
+        },
+    'posix_user': {
+        'stdlib': '{userbase}/lib/python{py_version_short}',
+        'platstdlib': '{userbase}/lib/python{py_version_short}',
+        'purelib': '{userbase}/lib/python{py_version_short}/site-packages',
+        'platlib': '{userbase}/lib/python{py_version_short}/site-packages',
+        'include': '{userbase}/include/python{py_version_short}',
+        'scripts': '{userbase}/bin',
+        'data': '{userbase}',
+        },
+    'osx_framework_user': {
+        'stdlib': '{userbase}/lib/python',
+        'platstdlib': '{userbase}/lib/python',
+        'purelib': '{userbase}/lib/python/site-packages',
+        'platlib': '{userbase}/lib/python/site-packages',
+        'include': '{userbase}/include',
+        'scripts': '{userbase}/bin',
+        'data': '{userbase}',
+        },
+    }
+
+_SCHEME_KEYS = ('stdlib', 'platstdlib', 'purelib', 'platlib', 'include',
+                'scripts', 'data')
+
+ # FIXME don't rely on sys.version here, its format is an implementation detail
+ # of CPython, use sys.version_info or sys.hexversion
+_PY_VERSION = sys.version.split()[0]
+_PY_VERSION_SHORT = '%d.%d' % sys.version_info[:2]
+_PY_VERSION_SHORT_NO_DOT = '%d%d' % sys.version_info[:2]
+_PREFIX = os.path.normpath(sys.prefix)
+_BASE_PREFIX = os.path.normpath(sys.base_prefix)
+_EXEC_PREFIX = os.path.normpath(sys.exec_prefix)
+_BASE_EXEC_PREFIX = os.path.normpath(sys.base_exec_prefix)
+_CONFIG_VARS = None
+_USER_BASE = None
+
+
+def _safe_realpath(path):
+    try:
+        return realpath(path)
+    except OSError:
+        return path
+
+if sys.executable:
+    _PROJECT_BASE = os.path.dirname(_safe_realpath(sys.executable))
+else:
+    # sys.executable can be empty if argv[0] has been changed and Python is
+    # unable to retrieve the real program name
+    _PROJECT_BASE = _safe_realpath(os.getcwd())
+
+if (os.name == 'nt' and
+    _PROJECT_BASE.lower().endswith(('\\pcbuild\\win32', '\\pcbuild\\amd64'))):
+    _PROJECT_BASE = _safe_realpath(os.path.join(_PROJECT_BASE, pardir, pardir))
+
+# set for cross builds
+if "_PYTHON_PROJECT_BASE" in os.environ:
+    _PROJECT_BASE = _safe_realpath(os.environ["_PYTHON_PROJECT_BASE"])
+
+def _is_python_source_dir(d):
+    for fn in ("Setup.dist", "Setup.local"):
+        if os.path.isfile(os.path.join(d, "Modules", fn)):
+            return True
+    return False
+
+_sys_home = getattr(sys, '_home', None)
+if (_sys_home and os.name == 'nt' and
+    _sys_home.lower().endswith(('\\pcbuild\\win32', '\\pcbuild\\amd64'))):
+    _sys_home = os.path.dirname(os.path.dirname(_sys_home))
+def is_python_build(check_home=False):
+    if check_home and _sys_home:
+        return _is_python_source_dir(_sys_home)
+    return _is_python_source_dir(_PROJECT_BASE)
+
+_PYTHON_BUILD = is_python_build(True)
+
+if _PYTHON_BUILD:
+    for scheme in ('posix_prefix', 'posix_home'):
+        _INSTALL_SCHEMES[scheme]['include'] = '{srcdir}/Include'
+        _INSTALL_SCHEMES[scheme]['platinclude'] = '{projectbase}/.'
+
+
+def _subst_vars(s, local_vars):
+    try:
+        return s.format(**local_vars)
+    except KeyError:
+        try:
+            return s.format(**os.environ)
+        except KeyError as var:
+            raise AttributeError('{%s}' % var)
+
+def _extend_dict(target_dict, other_dict):
+    target_keys = target_dict.keys()
+    for key, value in other_dict.items():
+        if key in target_keys:
+            continue
+        target_dict[key] = value
+
+
+def _expand_vars(scheme, vars):
+    res = {}
+    if vars is None:
+        vars = {}
+    _extend_dict(vars, get_config_vars())
+
+    for key, value in _INSTALL_SCHEMES[scheme].items():
+        if os.name in ('posix', 'nt'):
+            value = os.path.expanduser(value)
+        res[key] = os.path.normpath(_subst_vars(value, vars))
+    return res
+
+
+def _get_default_scheme():
+    if os.name == 'posix':
+        # the default scheme for posix is posix_prefix
+        return 'posix_prefix'
+    return os.name
+
+
+def _getuserbase():
+    env_base = os.environ.get("PYTHONUSERBASE", None)
+
+    def joinuser(*args):
+        return os.path.expanduser(os.path.join(*args))
+
+    if os.name == "nt":
+        base = os.environ.get("APPDATA") or "~"
+        if env_base:
+            return env_base
+        else:
+            return joinuser(base, "Python")
+
+    if sys.platform == "darwin":
+        framework = get_config_var("PYTHONFRAMEWORK")
+        if framework:
+            if env_base:
+                return env_base
+            else:
+                return joinuser("~", "Library", framework, "%d.%d" %
+                                sys.version_info[:2])
+
+    if env_base:
+        return env_base
+    else:
+        return joinuser("~", ".local")
+
+
+def _parse_makefile(filename, vars=None):
+    """Parse a Makefile-style file.
+
+    A dictionary containing name/value pairs is returned.  If an
+    optional dictionary is passed in as the second argument, it is
+    used instead of a new dictionary.
+    """
+    # Regexes needed for parsing Makefile (and similar syntaxes,
+    # like old-style Setup files).
+    import re
+    _variable_rx = re.compile(r"([a-zA-Z][a-zA-Z0-9_]+)\s*=\s*(.*)")
+    _findvar1_rx = re.compile(r"\$\(([A-Za-z][A-Za-z0-9_]*)\)")
+    _findvar2_rx = re.compile(r"\${([A-Za-z][A-Za-z0-9_]*)}")
+
+    if vars is None:
+        vars = {}
+    done = {}
+    notdone = {}
+
+    with open(filename, errors="surrogateescape") as f:
+        lines = f.readlines()
+
+    for line in lines:
+        if line.startswith('#') or line.strip() == '':
+            continue
+        m = _variable_rx.match(line)
+        if m:
+            n, v = m.group(1, 2)
+            v = v.strip()
+            # `$$' is a literal `$' in make
+            tmpv = v.replace('$$', '')
+
+            if "$" in tmpv:
+                notdone[n] = v
+            else:
+                try:
+                    v = int(v)
+                except ValueError:
+                    # insert literal `$'
+                    done[n] = v.replace('$$', '$')
+                else:
+                    done[n] = v
+
+    # do variable interpolation here
+    variables = list(notdone.keys())
+
+    # Variables with a 'PY_' prefix in the makefile. These need to
+    # be made available without that prefix through sysconfig.
+    # Special care is needed to ensure that variable expansion works, even
+    # if the expansion uses the name without a prefix.
+    renamed_variables = ('CFLAGS', 'LDFLAGS', 'CPPFLAGS')
+
+    while len(variables) > 0:
+        for name in tuple(variables):
+            value = notdone[name]
+            m1 = _findvar1_rx.search(value)
+            m2 = _findvar2_rx.search(value)
+            if m1 and m2:
+                m = m1 if m1.start() < m2.start() else m2
+            else:
+                m = m1 if m1 else m2
+            if m is not None:
+                n = m.group(1)
+                found = True
+                if n in done:
+                    item = str(done[n])
+                elif n in notdone:
+                    # get it on a subsequent round
+                    found = False
+                elif n in os.environ:
+                    # do it like make: fall back to environment
+                    item = os.environ[n]
+
+                elif n in renamed_variables:
+                    if (name.startswith('PY_') and
+                        name[3:] in renamed_variables):
+                        item = ""
+
+                    elif 'PY_' + n in notdone:
+                        found = False
+
+                    else:
+                        item = str(done['PY_' + n])
+
+                else:
+                    done[n] = item = ""
+
+                if found:
+                    after = value[m.end():]
+                    value = value[:m.start()] + item + after
+                    if "$" in after:
+                        notdone[name] = value
+                    else:
+                        try:
+                            value = int(value)
+                        except ValueError:
+                            done[name] = value.strip()
+                        else:
+                            done[name] = value
+                        variables.remove(name)
+
+                        if name.startswith('PY_') \
+                        and name[3:] in renamed_variables:
+
+                            name = name[3:]
+                            if name not in done:
+                                done[name] = value
+
+            else:
+                # bogus variable reference (e.g. "prefix=$/opt/python");
+                # just drop it since we can't deal
+                done[name] = value
+                variables.remove(name)
+
+    # strip spurious spaces
+    for k, v in done.items():
+        if isinstance(v, str):
+            done[k] = v.strip()
+
+    # save the results in the global dictionary
+    vars.update(done)
+    return vars
+
+
+def get_makefile_filename():
+    """Return the path of the Makefile."""
+    if _PYTHON_BUILD:
+        return os.path.join(_sys_home or _PROJECT_BASE, "Makefile")
+    if hasattr(sys, 'abiflags'):
+        config_dir_name = 'config-%s%s' % (_PY_VERSION_SHORT, sys.abiflags)
+    else:
+        config_dir_name = 'config'
+    if hasattr(sys.implementation, '_multiarch'):
+        config_dir_name += '-%s' % sys.implementation._multiarch
+    return os.path.join(get_path('stdlib'), config_dir_name, 'Makefile')
+
+
+def _get_sysconfigdata_name():
+    return os.environ.get('_PYTHON_SYSCONFIGDATA_NAME',
+        '_sysconfigdata_{abi}_{platform}_{multiarch}'.format(
+        abi=sys.abiflags,
+        platform=sys.platform,
+        multiarch=getattr(sys.implementation, '_multiarch', ''),
+    ))
+
+
+def _generate_posix_vars():
+    """Generate the Python module containing build-time variables."""
+    import pprint
+    vars = {}
+    # load the installed Makefile:
+    makefile = get_makefile_filename()
+    try:
+        _parse_makefile(makefile, vars)
+    except OSError as e:
+        msg = "invalid Python installation: unable to open %s" % makefile
+        if hasattr(e, "strerror"):
+            msg = msg + " (%s)" % e.strerror
+        raise OSError(msg)
+    # load the installed pyconfig.h:
+    config_h = get_config_h_filename()
+    try:
+        with open(config_h) as f:
+            parse_config_h(f, vars)
+    except OSError as e:
+        msg = "invalid Python installation: unable to open %s" % config_h
+        if hasattr(e, "strerror"):
+            msg = msg + " (%s)" % e.strerror
+        raise OSError(msg)
+    # On AIX, there are wrong paths to the linker scripts in the Makefile
+    # -- these paths are relative to the Python source, but when installed
+    # the scripts are in another directory.
+    if _PYTHON_BUILD:
+        vars['BLDSHARED'] = vars['LDSHARED']
+
+    # There's a chicken-and-egg situation on OS X with regards to the
+    # _sysconfigdata module after the changes introduced by #15298:
+    # get_config_vars() is called by get_platform() as part of the
+    # `make pybuilddir.txt` target -- which is a precursor to the
+    # _sysconfigdata.py module being constructed.  Unfortunately,
+    # get_config_vars() eventually calls _init_posix(), which attempts
+    # to import _sysconfigdata, which we won't have built yet.  In order
+    # for _init_posix() to work, if we're on Darwin, just mock up the
+    # _sysconfigdata module manually and populate it with the build vars.
+    # This is more than sufficient for ensuring the subsequent call to
+    # get_platform() succeeds.
+    name = _get_sysconfigdata_name()
+    if 'darwin' in sys.platform:
+        import types
+        module = types.ModuleType(name)
+        module.build_time_vars = vars
+        sys.modules[name] = module
+
+    pybuilddir = 'build/lib.%s-%s' % (get_platform(), _PY_VERSION_SHORT)
+    if hasattr(sys, "gettotalrefcount"):
+        pybuilddir += '-pydebug'
+    os.makedirs(pybuilddir, exist_ok=True)
+    destfile = os.path.join(pybuilddir, name + '.py')
+
+    with open(destfile, 'w', encoding='utf8') as f:
+        f.write('# system configuration generated and used by'
+                ' the sysconfig module\n')
+        f.write('build_time_vars = ')
+        pprint.pprint(vars, stream=f)
+
+    # Create file used for sys.path fixup -- see Modules/getpath.c
+    with open('pybuilddir.txt', 'w', encoding='ascii') as f:
+        f.write(pybuilddir)
+
+def _init_posix(vars):
+    """Initialize the module as appropriate for POSIX systems."""
+    # _sysconfigdata is generated at build time, see _generate_posix_vars()
+    name = _get_sysconfigdata_name()
+    _temp = __import__(name, globals(), locals(), ['build_time_vars'], 0)
+    build_time_vars = _temp.build_time_vars
+    vars.update(build_time_vars)
+
+def _init_non_posix(vars):
+    """Initialize the module as appropriate for NT"""
+    # set basic install directories
+    vars['LIBDEST'] = get_path('stdlib')
+    vars['BINLIBDEST'] = get_path('platstdlib')
+    vars['INCLUDEPY'] = get_path('include')
+    vars['EXT_SUFFIX'] = '.pyd'
+    vars['EXE'] = '.exe'
+    vars['VERSION'] = _PY_VERSION_SHORT_NO_DOT
+    vars['BINDIR'] = os.path.dirname(_safe_realpath(sys.executable))
+
+#
+# public APIs
+#
+
+
+def parse_config_h(fp, vars=None):
+    """Parse a config.h-style file.
+
+    A dictionary containing name/value pairs is returned.  If an
+    optional dictionary is passed in as the second argument, it is
+    used instead of a new dictionary.
+    """
+    if vars is None:
+        vars = {}
+    import re
+    define_rx = re.compile("#define ([A-Z][A-Za-z0-9_]+) (.*)\n")
+    undef_rx = re.compile("/[*] #undef ([A-Z][A-Za-z0-9_]+) [*]/\n")
+
+    while True:
+        line = fp.readline()
+        if not line:
+            break
+        m = define_rx.match(line)
+        if m:
+            n, v = m.group(1, 2)
+            try:
+                v = int(v)
+            except ValueError:
+                pass
+            vars[n] = v
+        else:
+            m = undef_rx.match(line)
+            if m:
+                vars[m.group(1)] = 0
+    return vars
+
+
+def get_config_h_filename():
+    """Return the path of pyconfig.h."""
+    if _PYTHON_BUILD:
+        if os.name == "nt":
+            inc_dir = os.path.join(_sys_home or _PROJECT_BASE, "PC")
+        else:
+            inc_dir = _sys_home or _PROJECT_BASE
+    else:
+        inc_dir = get_path('platinclude')
+    return os.path.join(inc_dir, 'pyconfig.h')
+
+
+def get_scheme_names():
+    """Return a tuple containing the schemes names."""
+    return tuple(sorted(_INSTALL_SCHEMES))
+
+
+def get_path_names():
+    """Return a tuple containing the paths names."""
+    return _SCHEME_KEYS
+
+
+def get_paths(scheme=_get_default_scheme(), vars=None, expand=True):
+    """Return a mapping containing an install scheme.
+
+    ``scheme`` is the install scheme name. If not provided, it will
+    return the default scheme for the current platform.
+    """
+    if expand:
+        return _expand_vars(scheme, vars)
+    else:
+        return _INSTALL_SCHEMES[scheme]
+
+
+def get_path(name, scheme=_get_default_scheme(), vars=None, expand=True):
+    """Return a path corresponding to the scheme.
+
+    ``scheme`` is the install scheme name.
+    """
+    return get_paths(scheme, vars, expand)[name]
+
+
+def get_config_vars(*args):
+    """With no arguments, return a dictionary of all configuration
+    variables relevant for the current platform.
+
+    On Unix, this means every variable defined in Python's installed Makefile;
+    On Windows it's a much smaller set.
+
+    With arguments, return a list of values that result from looking up
+    each argument in the configuration variable dictionary.
+    """
+    global _CONFIG_VARS
+    if _CONFIG_VARS is None:
+        _CONFIG_VARS = {}
+        # Normalized versions of prefix and exec_prefix are handy to have;
+        # in fact, these are the standard versions used most places in the
+        # Distutils.
+        _CONFIG_VARS['prefix'] = _PREFIX
+        _CONFIG_VARS['exec_prefix'] = _EXEC_PREFIX
+        _CONFIG_VARS['py_version'] = _PY_VERSION
+        _CONFIG_VARS['py_version_short'] = _PY_VERSION_SHORT
+        _CONFIG_VARS['py_version_nodot'] = _PY_VERSION_SHORT_NO_DOT
+        _CONFIG_VARS['installed_base'] = _BASE_PREFIX
+        _CONFIG_VARS['base'] = _PREFIX
+        _CONFIG_VARS['installed_platbase'] = _BASE_EXEC_PREFIX
+        _CONFIG_VARS['platbase'] = _EXEC_PREFIX
+        _CONFIG_VARS['projectbase'] = _PROJECT_BASE
+        try:
+            _CONFIG_VARS['abiflags'] = sys.abiflags
+        except AttributeError:
+            # sys.abiflags may not be defined on all platforms.
+            _CONFIG_VARS['abiflags'] = ''
+
+        if os.name == 'nt':
+            _init_non_posix(_CONFIG_VARS)
+        if os.name == 'posix':
+            _init_posix(_CONFIG_VARS)
+        # For backward compatibility, see issue19555
+        SO = _CONFIG_VARS.get('EXT_SUFFIX')
+        if SO is not None:
+            _CONFIG_VARS['SO'] = SO
+        # Setting 'userbase' is done below the call to the
+        # init function to enable using 'get_config_var' in
+        # the init-function.
+        _CONFIG_VARS['userbase'] = _getuserbase()
+
+        # Always convert srcdir to an absolute path
+        srcdir = _CONFIG_VARS.get('srcdir', _PROJECT_BASE)
+        if os.name == 'posix':
+            if _PYTHON_BUILD:
+                # If srcdir is a relative path (typically '.' or '..')
+                # then it should be interpreted relative to the directory
+                # containing Makefile.
+                base = os.path.dirname(get_makefile_filename())
+                srcdir = os.path.join(base, srcdir)
+            else:
+                # srcdir is not meaningful since the installation is
+                # spread about the filesystem.  We choose the
+                # directory containing the Makefile since we know it
+                # exists.
+                srcdir = os.path.dirname(get_makefile_filename())
+        _CONFIG_VARS['srcdir'] = _safe_realpath(srcdir)
+
+        # OS X platforms require special customization to handle
+        # multi-architecture, multi-os-version installers
+        if sys.platform == 'darwin':
+            import _osx_support
+            _osx_support.customize_config_vars(_CONFIG_VARS)
+
+    if args:
+        vals = []
+        for name in args:
+            vals.append(_CONFIG_VARS.get(name))
+        return vals
+    else:
+        return _CONFIG_VARS
+
+
+def get_config_var(name):
+    """Return the value of a single variable using the dictionary returned by
+    'get_config_vars()'.
+
+    Equivalent to get_config_vars().get(name)
+    """
+    if name == 'SO':
+        import warnings
+        warnings.warn('SO is deprecated, use EXT_SUFFIX', DeprecationWarning, 2)
+    return get_config_vars().get(name)
+
+
+def get_platform():
+    """Return a string that identifies the current platform.
+
+    This is used mainly to distinguish platform-specific build directories and
+    platform-specific built distributions.  Typically includes the OS name
+    and version and the architecture (as supplied by 'os.uname()'),
+    although the exact information included depends on the OS; eg. for IRIX
+    the architecture isn't particularly important (IRIX only runs on SGI
+    hardware), but for Linux the kernel version isn't particularly
+    important.
+
+    Examples of returned values:
+       linux-i586
+       linux-alpha (?)
+       solaris-2.6-sun4u
+       irix-5.3
+       irix64-6.2
+
+    Windows will return one of:
+       win-amd64 (64bit Windows on AMD64 (aka x86_64, Intel64, EM64T, etc)
+       win-ia64 (64bit Windows on Itanium)
+       win32 (all others - specifically, sys.platform is returned)
+
+    For other non-POSIX platforms, currently just returns 'sys.platform'.
+    """
+    if os.name == 'nt':
+        # sniff sys.version for architecture.
+        prefix = " bit ("
+        i = sys.version.find(prefix)
+        if i == -1:
+            return sys.platform
+        j = sys.version.find(")", i)
+        look = sys.version[i+len(prefix):j].lower()
+        if look == 'amd64':
+            return 'win-amd64'
+        if look == 'itanium':
+            return 'win-ia64'
+        return sys.platform
+
+    if os.name != "posix" or not hasattr(os, 'uname'):
+        # XXX what about the architecture? NT is Intel or Alpha
+        return sys.platform
+
+    # Set for cross builds explicitly
+    if "_PYTHON_HOST_PLATFORM" in os.environ:
+        return os.environ["_PYTHON_HOST_PLATFORM"]
+
+    # Try to distinguish various flavours of Unix
+    osname, host, release, version, machine = os.uname()
+
+    # Convert the OS name to lowercase, remove '/' characters
+    # (to accommodate BSD/OS), and translate spaces (for "Power Macintosh")
+    osname = osname.lower().replace('/', '')
+    machine = machine.replace(' ', '_')
+    machine = machine.replace('/', '-')
+
+    if osname[:5] == "linux":
+        # At least on Linux/Intel, 'machine' is the processor --
+        # i386, etc.
+        # XXX what about Alpha, SPARC, etc?
+        return  "%s-%s" % (osname, machine)
+    elif osname[:5] == "sunos":
+        if release[0] >= "5":           # SunOS 5 == Solaris 2
+            osname = "solaris"
+            release = "%d.%s" % (int(release[0]) - 3, release[2:])
+            # We can't use "platform.architecture()[0]" because a
+            # bootstrap problem. We use a dict to get an error
+            # if some suspicious happens.
+            bitness = {2147483647:"32bit", 9223372036854775807:"64bit"}
+            machine += ".%s" % bitness[sys.maxsize]
+        # fall through to standard osname-release-machine representation
+    elif osname[:4] == "irix":              # could be "irix64"!
+        return "%s-%s" % (osname, release)
+    elif osname[:3] == "aix":
+        return "%s-%s.%s" % (osname, version, release)
+    elif osname[:6] == "cygwin":
+        osname = "cygwin"
+        import re
+        rel_re = re.compile(r'[\d.]+')
+        m = rel_re.match(release)
+        if m:
+            release = m.group()
+    elif osname[:6] == "darwin":
+        import _osx_support
+        osname, release, machine = _osx_support.get_platform_osx(
+                                            get_config_vars(),
+                                            osname, release, machine)
+
+    return "%s-%s-%s" % (osname, release, machine)
+
+
+def get_python_version():
+    return _PY_VERSION_SHORT
+
+
+def _print_dict(title, data):
+    for index, (key, value) in enumerate(sorted(data.items())):
+        if index == 0:
+            print('%s: ' % (title))
+        print('\t%s = "%s"' % (key, value))
+
+
+def _main():
+    """Display all information sysconfig detains."""
+    if '--generate-posix-vars' in sys.argv:
+        _generate_posix_vars()
+        return
+    print('Platform: "%s"' % get_platform())
+    print('Python version: "%s"' % get_python_version())
+    print('Current installation scheme: "%s"' % _get_default_scheme())
+    print()
+    _print_dict('Paths', get_paths())
+    print()
+    _print_dict('Variables', get_config_vars())
+
+
+if __name__ == '__main__':
+    _main()
diff --git a/vm/src/stdlib/os.rs b/vm/src/stdlib/os.rs
index 0c1f240d7..e89092965 100644
--- a/vm/src/stdlib/os.rs
+++ b/vm/src/stdlib/os.rs
@@ -26,7 +26,8 @@ use crate::obj::objset::PySet;
 use crate::obj::objstr::{self, PyString, PyStringRef};
 use crate::obj::objtype::{self, PyClassRef};
 use crate::pyobject::{
-    ItemProtocol, PyClassImpl, PyObjectRef, PyRef, PyResult, PyValue, TryIntoRef, TypeProtocol,
+    Either, ItemProtocol, PyClassImpl, PyObjectRef, PyRef, PyResult, PyValue, TryIntoRef,
+    TypeProtocol,
 };
 use crate::vm::VirtualMachine;
 
@@ -416,12 +417,46 @@ fn os_listdir(path: PyStringRef, vm: &VirtualMachine) -> PyResult {
     }
 }
 
-fn os_putenv(key: PyStringRef, value: PyStringRef, _vm: &VirtualMachine) {
-    env::set_var(&key.value, &value.value)
+fn bytes_as_osstr<'a>(b: &'a [u8], vm: &VirtualMachine) -> PyResult<&'a ffi::OsStr> {
+    let os_str = {
+        #[cfg(unix)]
+        {
+            use std::os::unix::ffi::OsStrExt;
+            Some(ffi::OsStr::from_bytes(b))
+        }
+        #[cfg(windows)]
+        {
+            std::str::from_utf8(b).ok().map(|s| s.as_ref())
+        }
+    };
+    os_str
+        .ok_or_else(|| vm.new_value_error("Can't convert bytes to str for env function".to_owned()))
 }
 
-fn os_unsetenv(key: PyStringRef, _vm: &VirtualMachine) {
-    env::remove_var(&key.value)
+fn os_putenv(
+    key: Either<PyStringRef, PyBytesRef>,
+    value: Either<PyStringRef, PyBytesRef>,
+    vm: &VirtualMachine,
+) -> PyResult<()> {
+    let key: &ffi::OsStr = match key {
+        Either::A(ref s) => s.as_str().as_ref(),
+        Either::B(ref b) => bytes_as_osstr(b.get_value(), vm)?,
+    };
+    let value: &ffi::OsStr = match value {
+        Either::A(ref s) => s.as_str().as_ref(),
+        Either::B(ref b) => bytes_as_osstr(b.get_value(), vm)?,
+    };
+    env::set_var(key, value);
+    Ok(())
+}
+
+fn os_unsetenv(key: Either<PyStringRef, PyBytesRef>, vm: &VirtualMachine) -> PyResult<()> {
+    let key: &ffi::OsStr = match key {
+        Either::A(ref s) => s.as_str().as_ref(),
+        Either::B(ref b) => bytes_as_osstr(b.get_value(), vm)?,
+    };
+    env::remove_var(key);
+    Ok(())
 }
 
 fn _os_environ(vm: &VirtualMachine) -> PyDictRef {
diff --git a/vm/src/sysmodule.rs b/vm/src/sysmodule.rs
index 14528a639..a1e1fa415 100644
--- a/vm/src/sysmodule.rs
+++ b/vm/src/sysmodule.rs
@@ -322,6 +322,7 @@ settrace() -- set the global debug tracing function
     let prefix = option_env!("RUSTPYTHON_PREFIX").unwrap_or("/usr/local");
     let base_prefix = option_env!("RUSTPYTHON_BASEPREFIX").unwrap_or(prefix);
     let exec_prefix = option_env!("RUSTPYTHON_EXECPREFIX").unwrap_or(prefix);
+    let base_exec_prefix = option_env!("RUSTPYTHON_BASEEXECPREFIX").unwrap_or(exec_prefix);
 
     extend_module!(vm, module, {
       "__name__" => ctx.new_str(String::from("sys")),
@@ -363,6 +364,7 @@ settrace() -- set the global debug tracing function
       "prefix" => ctx.new_str(prefix.to_string()),
       "base_prefix" => ctx.new_str(base_prefix.to_string()),
       "exec_prefix" => ctx.new_str(exec_prefix.to_string()),
+      "base_exec_prefix" => ctx.new_str(base_exec_prefix.to_string()),
       "exit" => ctx.new_rustfunc(sys_exit),
     });