cpython/Tools/unicode/mkstringprep.py

import re
import os
import unicodedata as unicodedata_current
from unicodedata import ucd_3_2_0 as unicodedata_320

FILENAME = "Tools/unicode/data/rfc3454.txt"
URL = "https://www.rfc-editor.org/rfc/rfc3454.txt"

def gen_category(cats):
    for i in range(0, 0x110000):
        if unicodedata_320.category(chr(i)) in cats:
            yield i

def gen_bidirectional(cats):
    for i in range(0, 0x110000):
        if unicodedata_320.bidirectional(chr(i)) in cats:
            yield i

def compact_set(l):
    single = []
    tuple = []
    prev = None
    span = 0
    for e in l:
        if prev is None:
            prev = e
            span = 0
            continue
        if prev+span+1 != e:
            if span > 2:
                tuple.append((prev,prev+span+1))
            else:
                for i in range(prev, prev+span+1):
                    single.append(i)
            prev = e
            span = 0
        else:
            span += 1
    if span:
        tuple.append((prev,prev+span+1))
    else:
        single.append(prev)
    if not single and len(tuple) == 1:
        tuple = "range(%d,%d)" % tuple[0]
    else:
        tuple = " + ".join("list(range(%d,%d))" % t for t in tuple)
    if not single:
        return "set(%s)" % tuple
    if not tuple:
        return "set(%r)" % (single,)
    return "set(%r + %s)" % (single, tuple)

############## Read the tables in the RFC #######################

try:
    data_file = open(FILENAME, encoding='utf-8')
except FileNotFoundError:
    import urllib.request
    os.makedirs(os.path.dirname(FILENAME), exist_ok=True)
    urllib.request.urlretrieve(URL, filename=FILENAME)
    data_file = open(FILENAME, encoding='utf-8')

with data_file:
    data = data_file.readlines()

tables = []
curname = None
for l in data:
    l = l.strip()
    if not l:
        continue
    # Skip RFC page breaks
    if l.startswith(("Hoffman & Blanchet", "RFC 3454")):
        continue
    # Find start/end lines
    m = re.match("----- (Start|End) Table ([A-Z](.[0-9])+) -----", l)
    if m:
        if m.group(1) == "Start":
            if curname:
                raise RuntimeError("Double Start", (curname, l))
            curname = m.group(2)
            table = {}
            tables.append((curname, table))
            continue
        else:
            if not curname:
                raise RuntimeError("End without start", l)
            if curname != m.group(2):
                raise RuntimeError("Unexpected end", l)
            curname = None
            continue
    if not curname:
        continue
    # Now we are in a table
    fields = l.split(";")
    if len(fields) > 1:
        # Drop comment field
        fields = fields[:-1]
    if len(fields) == 1:
        fields = fields[0].split("-")
        if len(fields) > 1:
            # range
            try:
                start, end = fields
            except ValueError:
                raise RuntimeError("Unpacking problem", l)
        else:
            start = end = fields[0]
        start = int(start, 16)
        end = int(end, 16)
        for i in range(start, end+1):
            table[i] = i
    else:
        code, value = fields
        value = value.strip()
        if value:
            value = [int(v, 16) for v in value.split(" ")]
        else:
            # table B.1
            value = None
        table[int(code, 16)] = value

########### Generate compact Python versions of the tables #############

print("""# This file is generated by mkstringprep.py. DO NOT EDIT.
\"\"\"Library that exposes various tables found in the StringPrep RFC 3454.

There are two kinds of tables: sets, for which a member test is provided,
and mappings, for which a mapping function is provided.
\"\"\"

# This check asserts that mkstringprep.py has been run
# when unicodedata is modified to ensure conformant behavior.
import unicodedata
""")

print("assert unicodedata.unidata_version == %r" % (unicodedata_current.unidata_version,))

print("""
from unicodedata import ucd_3_2_0 as unicodedata_320
""")

print("assert unicodedata_320.unidata_version == %r" % (unicodedata_320.unidata_version,))

# A.1 is the table of unassigned characters
# XXX Plane 15 PUA is listed as unassigned in Python.
name, table = tables[0]
del tables[0]
assert name == "A.1"
table = set(table.keys())
Cn = set(gen_category(["Cn"]))

# FDD0..FDEF are process internal codes
Cn -= set(range(0xFDD0, 0xFDF0))
# not a character
Cn -= set(range(0xFFFE, 0x110000, 0x10000))
Cn -= set(range(0xFFFF, 0x110000, 0x10000))

# assert table == Cn

print("""
def in_table_a1(code):
    if unicodedata_320.category(code) != 'Cn': return False
    c = ord(code)
    if 0xFDD0 <= c < 0xFDF0: return False
    return (c & 0xFFFF) not in (0xFFFE, 0xFFFF)
""")

# B.1 cannot easily be derived
name, table = tables[0]
del tables[0]
assert name == "B.1"
table = sorted(table.keys())
print("""
b1_set = """ + compact_set(table) + """
def in_table_b1(code):
    return ord(code) in b1_set
""")

# B.2 and B.3 is case folding.
# It takes CaseFolding.txt into account, which is
# not available in the Python database. Since
# B.2 is derived from B.3, we process B.3 first.
# B.3 supposedly *is* CaseFolding-3.2.0.txt.

name, table_b2 = tables[0]
del tables[0]
assert name == "B.2"

name, table_b3 = tables[0]
del tables[0]
assert name == "B.3"

# B.3 is mostly Python's .lower, except for a number
# of special cases, e.g. considering canonical forms.
# To enforce Unicode 3.2.0 behavior of .lower instead of
# whatever Unicode version is included with Python we
# add unassigned or newly case-folding codepoints to
# the exception map, too.

b3_exceptions = {}

for k,v in table_b2.items():
    if list(map(ord, chr(k).lower())) != v:
        b3_exceptions[k] = "".join(map(chr,v))
for cp in range(0x110000):
    ch = chr(cp)
    # Assigned in current Unicode version
    # and supports case folding, but not
    # explicitly in B.2 or B.3 tables.
    if (unicodedata_current.category(ch) != "Cn"
            and ch.lower() != ch
            and cp not in table_b2
            and cp not in table_b3):
        b3_exceptions[cp] = ch  # Identity.

b3 = sorted(b3_exceptions.items())

print("""
b3_exceptions = {""")
for i, kv in enumerate(b3):
    print("0x%x:%a," % kv, end='\n' if i % 4 == 3 else ' ')
print("}")

print("""
def map_table_b3(code):
    r = b3_exceptions.get(ord(code))
    if r is not None: return r
    return code.lower()
""")

def map_table_b3(code):
    r = b3_exceptions.get(ord(code))
    if r is not None: return r
    return code.lower()

# B.2 is case folding for NFKC. This is the same as B.3,
# except where NormalizeWithKC(Fold(a)) !=
# NormalizeWithKC(Fold(NormalizeWithKC(Fold(a))))

def map_table_b2(a):
    al = map_table_b3(a)
    b = unicodedata_320.normalize("NFKC", al)
    bl = "".join([map_table_b3(ch) for ch in b])
    c = unicodedata_320.normalize("NFKC", bl)
    if b != c:
        return c
    else:
        return al

specials = {}
for k,v in table_b2.items():
    if list(map(ord, map_table_b2(chr(k)))) != v:
        specials[k] = v

# B.3 should not add any additional special cases
assert specials == {}

print("""
def map_table_b2(a):
    al = map_table_b3(a)
    b = unicodedata_320.normalize("NFKC", al)
    bl = "".join([map_table_b3(ch) for ch in b])
    c = unicodedata_320.normalize("NFKC", bl)
    if b != c:
        return c
    else:
        return al
""")

# C.1.1 is a table with a single character
name, table = tables[0]
del tables[0]
assert name == "C.1.1"
assert table == {0x20:0x20}

print("""
def in_table_c11(code):
    return code == " "
""")

# C.1.2 is the rest of all space characters
name, table = tables[0]
del tables[0]
assert name == "C.1.2"

table = set(table.keys())
Zs = set(gen_category(["Zs"])) - {0x20}
assert Zs == table

print("""
def in_table_c12(code):
    return unicodedata_320.category(code) == "Zs" and code != " "

def in_table_c11_c12(code):
    return unicodedata_320.category(code) == "Zs"
""")

# C.2.1 ASCII control characters
name, table_c21 = tables[0]
del tables[0]
assert name == "C.2.1"

Cc = set(gen_category(["Cc"]))
Cc_ascii = Cc & set(range(128))
table_c21 = set(table_c21.keys())
assert Cc_ascii == table_c21

print("""
def in_table_c21(code):
    return ord(code) < 128 and unicodedata_320.category(code) == "Cc"
""")

# C.2.2 Non-ASCII control characters. It also includes
# a number of characters in category Cf.
name, table_c22 = tables[0]
del tables[0]
assert name == "C.2.2"

Cc_nonascii = Cc - Cc_ascii
table_c22 = set(table_c22.keys())
assert len(Cc_nonascii - table_c22) == 0

specials = list(table_c22 - Cc_nonascii)
specials.sort()

print("""c22_specials = """ + compact_set(specials) + """
def in_table_c22(code):
    c = ord(code)
    if c < 128: return False
    if unicodedata_320.category(code) == "Cc": return True
    return c in c22_specials

def in_table_c21_c22(code):
    return unicodedata_320.category(code) == "Cc" or \\
           ord(code) in c22_specials
""")

# C.3 Private use
name, table = tables[0]
del tables[0]
assert name == "C.3"

Co = set(gen_category(["Co"]))
assert set(table.keys()) == Co

print("""
def in_table_c3(code):
    return unicodedata_320.category(code) == "Co"
""")

# C.4 Non-character code points, xFFFE, xFFFF
# plus process internal codes
name, table = tables[0]
del tables[0]
assert name == "C.4"

nonchar = set(range(0xFDD0,0xFDF0))
nonchar.update(range(0xFFFE,0x110000,0x10000))
nonchar.update(range(0xFFFF,0x110000,0x10000))
table = set(table.keys())
assert table == nonchar

print("""
def in_table_c4(code):
    c = ord(code)
    if c < 0xFDD0: return False
    if c < 0xFDF0: return True
    return (ord(code) & 0xFFFF) in (0xFFFE, 0xFFFF)
""")

# C.5 Surrogate codes
name, table = tables[0]
del tables[0]
assert name == "C.5"

Cs = set(gen_category(["Cs"]))
assert set(table.keys()) == Cs

print("""
def in_table_c5(code):
    return unicodedata_320.category(code) == "Cs"
""")

# C.6 Inappropriate for plain text
name, table = tables[0]
del tables[0]
assert name == "C.6"

table = sorted(table.keys())

print("""
c6_set = """ + compact_set(table) + """
def in_table_c6(code):
    return ord(code) in c6_set
""")

# C.7 Inappropriate for canonical representation
name, table = tables[0]
del tables[0]
assert name == "C.7"

table = sorted(table.keys())

print("""
c7_set = """ + compact_set(table) + """
def in_table_c7(code):
    return ord(code) in c7_set
""")

# C.8 Change display properties or are deprecated
name, table = tables[0]
del tables[0]
assert name == "C.8"

table = sorted(table.keys())

print("""
c8_set = """ + compact_set(table) + """
def in_table_c8(code):
    return ord(code) in c8_set
""")

# C.9 Tagging characters
name, table = tables[0]
del tables[0]
assert name == "C.9"

table = sorted(table.keys())

print("""
c9_set = """ + compact_set(table) + """
def in_table_c9(code):
    return ord(code) in c9_set
""")

# D.1 Characters with bidirectional property "R" or "AL"
name, table = tables[0]
del tables[0]
assert name == "D.1"

RandAL = set(gen_bidirectional(["R","AL"]))
assert set(table.keys()) == RandAL

print("""
def in_table_d1(code):
    return unicodedata_320.bidirectional(code) in ("R","AL")
""")

# D.2 Characters with bidirectional property "L"
name, table = tables[0]
del tables[0]
assert name == "D.2"

L = set(gen_bidirectional(["L"]))
assert set(table.keys()) == L

print("""
def in_table_d2(code):
    return unicodedata_320.bidirectional(code) == "L"
""", end="")