diff --git a/Lib/dataclasses.py b/Lib/dataclasses.py index e1687a117..7883ce78e 100644 --- a/Lib/dataclasses.py +++ b/Lib/dataclasses.py @@ -4,11 +4,9 @@ import copy import types import inspect import keyword -import builtins -import functools +import itertools import abc -import _thread -from types import FunctionType, GenericAlias +from reprlib import recursive_repr __all__ = ['dataclass', @@ -222,6 +220,30 @@ _POST_INIT_NAME = '__post_init__' # https://bugs.python.org/issue33453 for details. _MODULE_IDENTIFIER_RE = re.compile(r'^(?:\s*(\w+)\s*\.)?\s*(\w+)') +# Atomic immutable types which don't require any recursive handling and for which deepcopy +# returns the same object. We can provide a fast-path for these types in asdict and astuple. +_ATOMIC_TYPES = frozenset({ + # Common JSON Serializable types + types.NoneType, + bool, + int, + float, + str, + # Other common types + complex, + bytes, + # Other types that are also unaffected by deepcopy + types.EllipsisType, + types.NotImplementedType, + types.CodeType, + types.BuiltinFunctionType, + types.FunctionType, + type, + range, + property, +}) + + class InitVar: __slots__ = ('type', ) @@ -229,7 +251,7 @@ class InitVar: self.type = type def __repr__(self): - if isinstance(self.type, type) and not isinstance(self.type, GenericAlias): + if isinstance(self.type, type): type_name = self.type.__name__ else: # typing objects, e.g. List[int] @@ -279,6 +301,7 @@ class Field: self.kw_only = kw_only self._field_type = None + @recursive_repr() def __repr__(self): return ('Field(' f'name={self.name!r},' @@ -297,7 +320,7 @@ class Field: # This is used to support the PEP 487 __set_name__ protocol in the # case where we're using a field that contains a descriptor as a # default value. For details on __set_name__, see - # https://www.python.org/dev/peps/pep-0487/#implementation-details. + # https://peps.python.org/pep-0487/#implementation-details. # # Note that in _process_class, this Field object is overwritten # with the default value, so the end result is a descriptor that @@ -309,7 +332,7 @@ class Field: # it. func(self.default, owner, name) - __class_getitem__ = classmethod(GenericAlias) + __class_getitem__ = classmethod(types.GenericAlias) class _DataclassParams: @@ -319,15 +342,25 @@ class _DataclassParams: 'order', 'unsafe_hash', 'frozen', + 'match_args', + 'kw_only', + 'slots', + 'weakref_slot', ) - def __init__(self, init, repr, eq, order, unsafe_hash, frozen): + def __init__(self, + init, repr, eq, order, unsafe_hash, frozen, + match_args, kw_only, slots, weakref_slot): self.init = init self.repr = repr self.eq = eq self.order = order self.unsafe_hash = unsafe_hash self.frozen = frozen + self.match_args = match_args + self.kw_only = kw_only + self.slots = slots + self.weakref_slot = weakref_slot def __repr__(self): return ('_DataclassParams(' @@ -336,7 +369,11 @@ class _DataclassParams: f'eq={self.eq!r},' f'order={self.order!r},' f'unsafe_hash={self.unsafe_hash!r},' - f'frozen={self.frozen!r}' + f'frozen={self.frozen!r},' + f'match_args={self.match_args!r},' + f'kw_only={self.kw_only!r},' + f'slots={self.slots!r},' + f'weakref_slot={self.weakref_slot!r}' ')') @@ -388,49 +425,95 @@ def _tuple_str(obj_name, fields): return f'({",".join([f"{obj_name}.{f.name}" for f in fields])},)' -# This function's logic is copied from "recursive_repr" function in -# reprlib module to avoid dependency. -def _recursive_repr(user_function): - # Decorator to make a repr function return "..." for a recursive - # call. - repr_running = set() +class _FuncBuilder: + def __init__(self, globals): + self.names = [] + self.src = [] + self.globals = globals + self.locals = {} + self.overwrite_errors = {} + self.unconditional_adds = {} - @functools.wraps(user_function) - def wrapper(self): - key = id(self), _thread.get_ident() - if key in repr_running: - return '...' - repr_running.add(key) - try: - result = user_function(self) - finally: - repr_running.discard(key) - return result - return wrapper + def add_fn(self, name, args, body, *, locals=None, return_type=MISSING, + overwrite_error=False, unconditional_add=False, decorator=None): + if locals is not None: + self.locals.update(locals) + # Keep track if this method is allowed to be overwritten if it already + # exists in the class. The error is method-specific, so keep it with + # the name. We'll use this when we generate all of the functions in + # the add_fns_to_class call. overwrite_error is either True, in which + # case we'll raise an error, or it's a string, in which case we'll + # raise an error and append this string. + if overwrite_error: + self.overwrite_errors[name] = overwrite_error -def _create_fn(name, args, body, *, globals=None, locals=None, - return_type=MISSING): - # Note that we may mutate locals. Callers beware! - # The only callers are internal to this module, so no - # worries about external callers. - if locals is None: - locals = {} - return_annotation = '' - if return_type is not MISSING: - locals['_return_type'] = return_type - return_annotation = '->_return_type' - args = ','.join(args) - body = '\n'.join(f' {b}' for b in body) + # Should this function always overwrite anything that's already in the + # class? The default is to not overwrite a function that already + # exists. + if unconditional_add: + self.unconditional_adds[name] = True - # Compute the text of the entire function. - txt = f' def {name}({args}){return_annotation}:\n{body}' + self.names.append(name) - local_vars = ', '.join(locals.keys()) - txt = f"def __create_fn__({local_vars}):\n{txt}\n return {name}" - ns = {} - exec(txt, globals, ns) - return ns['__create_fn__'](**locals) + if return_type is not MISSING: + self.locals[f'__dataclass_{name}_return_type__'] = return_type + return_annotation = f'->__dataclass_{name}_return_type__' + else: + return_annotation = '' + args = ','.join(args) + body = '\n'.join(body) + + # Compute the text of the entire function, add it to the text we're generating. + self.src.append(f'{f' {decorator}\n' if decorator else ''} def {name}({args}){return_annotation}:\n{body}') + + def add_fns_to_class(self, cls): + # The source to all of the functions we're generating. + fns_src = '\n'.join(self.src) + + # The locals they use. + local_vars = ','.join(self.locals.keys()) + + # The names of all of the functions, used for the return value of the + # outer function. Need to handle the 0-tuple specially. + if len(self.names) == 0: + return_names = '()' + else: + return_names =f'({",".join(self.names)},)' + + # txt is the entire function we're going to execute, including the + # bodies of the functions we're defining. Here's a greatly simplified + # version: + # def __create_fn__(): + # def __init__(self, x, y): + # self.x = x + # self.y = y + # @recursive_repr + # def __repr__(self): + # return f"cls(x={self.x!r},y={self.y!r})" + # return __init__,__repr__ + + txt = f"def __create_fn__({local_vars}):\n{fns_src}\n return {return_names}" + ns = {} + exec(txt, self.globals, ns) + fns = ns['__create_fn__'](**self.locals) + + # Now that we've generated the functions, assign them into cls. + for name, fn in zip(self.names, fns): + fn.__qualname__ = f"{cls.__qualname__}.{fn.__name__}" + if self.unconditional_adds.get(name, False): + setattr(cls, name, fn) + else: + already_exists = _set_new_attribute(cls, name, fn) + + # See if it's an error to overwrite this particular function. + if already_exists and (msg_extra := self.overwrite_errors.get(name)): + error_msg = (f'Cannot overwrite attribute {fn.__name__} ' + f'in class {cls.__name__}') + if not msg_extra is True: + error_msg = f'{error_msg} {msg_extra}' + + raise TypeError(error_msg) def _field_assign(frozen, name, value, self_name): @@ -441,22 +524,22 @@ def _field_assign(frozen, name, value, self_name): # self_name is what "self" is called in this function: don't # hard-code "self", since that might be a field name. if frozen: - return f'__dataclass_builtins_object__.__setattr__({self_name},{name!r},{value})' - return f'{self_name}.{name}={value}' + return f' __dataclass_builtins_object__.__setattr__({self_name},{name!r},{value})' + return f' {self_name}.{name}={value}' def _field_init(f, frozen, globals, self_name, slots): # Return the text of the line in the body of __init__ that will # initialize this field. - default_name = f'_dflt_{f.name}' + default_name = f'__dataclass_dflt_{f.name}__' if f.default_factory is not MISSING: if f.init: # This field has a default factory. If a parameter is # given, use it. If not, call the factory. globals[default_name] = f.default_factory value = (f'{default_name}() ' - f'if {f.name} is _HAS_DEFAULT_FACTORY ' + f'if {f.name} is __dataclass_HAS_DEFAULT_FACTORY__ ' f'else {f.name}') else: # This is a field that's not in the __init__ params, but @@ -517,15 +600,15 @@ def _init_param(f): elif f.default is not MISSING: # There's a default, this will be the name that's used to look # it up. - default = f'=_dflt_{f.name}' + default = f'=__dataclass_dflt_{f.name}__' elif f.default_factory is not MISSING: # There's a factory function. Set a marker. - default = '=_HAS_DEFAULT_FACTORY' - return f'{f.name}:_type_{f.name}{default}' + default = '=__dataclass_HAS_DEFAULT_FACTORY__' + return f'{f.name}:__dataclass_type_{f.name}__{default}' def _init_fn(fields, std_fields, kw_only_fields, frozen, has_post_init, - self_name, globals, slots): + self_name, func_builder, slots): # fields contains both real fields and InitVar pseudo-fields. # Make sure we don't have fields without defaults following fields @@ -534,22 +617,21 @@ def _init_fn(fields, std_fields, kw_only_fields, frozen, has_post_init, # message, and future-proofs us in case we build up the function # using ast. - seen_default = False + seen_default = None for f in std_fields: # Only consider the non-kw-only fields in the __init__ call. if f.init: if not (f.default is MISSING and f.default_factory is MISSING): - seen_default = True + seen_default = f elif seen_default: raise TypeError(f'non-default argument {f.name!r} ' - 'follows default argument') + f'follows default argument {seen_default.name!r}') - locals = {f'_type_{f.name}': f.type for f in fields} - locals.update({ - 'MISSING': MISSING, - '_HAS_DEFAULT_FACTORY': _HAS_DEFAULT_FACTORY, - '__dataclass_builtins_object__': object, - }) + locals = {**{f'__dataclass_type_{f.name}__': f.type for f in fields}, + **{'__dataclass_HAS_DEFAULT_FACTORY__': _HAS_DEFAULT_FACTORY, + '__dataclass_builtins_object__': object, + } + } body_lines = [] for f in fields: @@ -563,11 +645,11 @@ def _init_fn(fields, std_fields, kw_only_fields, frozen, has_post_init, if has_post_init: params_str = ','.join(f.name for f in fields if f._field_type is _FIELD_INITVAR) - body_lines.append(f'{self_name}.{_POST_INIT_NAME}({params_str})') + body_lines.append(f' {self_name}.{_POST_INIT_NAME}({params_str})') # If no body lines, use 'pass'. if not body_lines: - body_lines = ['pass'] + body_lines = [' pass'] _init_params = [_init_param(f) for f in std_fields] if kw_only_fields: @@ -576,70 +658,34 @@ def _init_fn(fields, std_fields, kw_only_fields, frozen, has_post_init, # (instead of just concatenting the lists together). _init_params += ['*'] _init_params += [_init_param(f) for f in kw_only_fields] - return _create_fn('__init__', - [self_name] + _init_params, - body_lines, - locals=locals, - globals=globals, - return_type=None) + func_builder.add_fn('__init__', + [self_name] + _init_params, + body_lines, + locals=locals, + return_type=None) -def _repr_fn(fields, globals): - fn = _create_fn('__repr__', - ('self',), - ['return self.__class__.__qualname__ + f"(' + - ', '.join([f"{f.name}={{self.{f.name}!r}}" - for f in fields]) + - ')"'], - globals=globals) - return _recursive_repr(fn) - - -def _frozen_get_del_attr(cls, fields, globals): +def _frozen_get_del_attr(cls, fields, func_builder): locals = {'cls': cls, 'FrozenInstanceError': FrozenInstanceError} + condition = 'type(self) is cls' if fields: - fields_str = '(' + ','.join(repr(f.name) for f in fields) + ',)' - else: - # Special case for the zero-length tuple. - fields_str = '()' - return (_create_fn('__setattr__', - ('self', 'name', 'value'), - (f'if type(self) is cls or name in {fields_str}:', - ' raise FrozenInstanceError(f"cannot assign to field {name!r}")', - f'super(cls, self).__setattr__(name, value)'), - locals=locals, - globals=globals), - _create_fn('__delattr__', - ('self', 'name'), - (f'if type(self) is cls or name in {fields_str}:', - ' raise FrozenInstanceError(f"cannot delete field {name!r}")', - f'super(cls, self).__delattr__(name)'), - locals=locals, - globals=globals), - ) + condition += ' or name in {' + ', '.join(repr(f.name) for f in fields) + '}' - -def _cmp_fn(name, op, self_tuple, other_tuple, globals): - # Create a comparison function. If the fields in the object are - # named 'x' and 'y', then self_tuple is the string - # '(self.x,self.y)' and other_tuple is the string - # '(other.x,other.y)'. - - return _create_fn(name, - ('self', 'other'), - [ 'if other.__class__ is self.__class__:', - f' return {self_tuple}{op}{other_tuple}', - 'return NotImplemented'], - globals=globals) - - -def _hash_fn(fields, globals): - self_tuple = _tuple_str('self', fields) - return _create_fn('__hash__', - ('self',), - [f'return hash({self_tuple})'], - globals=globals) + func_builder.add_fn('__setattr__', + ('self', 'name', 'value'), + (f' if {condition}:', + ' raise FrozenInstanceError(f"cannot assign to field {name!r}")', + f' super(cls, self).__setattr__(name, value)'), + locals=locals, + overwrite_error=True) + func_builder.add_fn('__delattr__', + ('self', 'name'), + (f' if {condition}:', + ' raise FrozenInstanceError(f"cannot delete field {name!r}")', + f' super(cls, self).__delattr__(name)'), + locals=locals, + overwrite_error=True) def _is_classvar(a_type, typing): @@ -807,26 +853,20 @@ def _get_field(cls, a_name, a_type, default_kw_only): raise TypeError(f'field {f.name} is a ClassVar but specifies ' 'kw_only') - # For real fields, disallow mutable defaults for known types. - if f._field_type is _FIELD and isinstance(f.default, (list, dict, set)): + # For real fields, disallow mutable defaults. Use unhashable as a proxy + # indicator for mutability. Read the __hash__ attribute from the class, + # not the instance. + if f._field_type is _FIELD and f.default.__class__.__hash__ is None: raise ValueError(f'mutable default {type(f.default)} for field ' f'{f.name} is not allowed: use default_factory') return f -def _set_qualname(cls, value): - # Ensure that the functions returned from _create_fn uses the proper - # __qualname__ (the class they belong to). - if isinstance(value, FunctionType): - value.__qualname__ = f"{cls.__qualname__}.{value.__name__}" - return value - def _set_new_attribute(cls, name, value): # Never overwrites an existing attribute. Returns True if the # attribute already exists. if name in cls.__dict__: return True - _set_qualname(cls, value) setattr(cls, name, value) return False @@ -836,14 +876,22 @@ def _set_new_attribute(cls, name, value): # take. The common case is to do nothing, so instead of providing a # function that is a no-op, use None to signify that. -def _hash_set_none(cls, fields, globals): - return None +def _hash_set_none(cls, fields, func_builder): + # It's sort of a hack that I'm setting this here, instead of at + # func_builder.add_fns_to_class time, but since this is an exceptional case + # (it's not setting an attribute to a function, but to a scalar value), + # just do it directly here. I might come to regret this. + cls.__hash__ = None -def _hash_add(cls, fields, globals): +def _hash_add(cls, fields, func_builder): flds = [f for f in fields if (f.compare if f.hash is None else f.hash)] - return _set_qualname(cls, _hash_fn(flds, globals)) + self_tuple = _tuple_str('self', flds) + func_builder.add_fn('__hash__', + ('self',), + [f' return hash({self_tuple})'], + unconditional_add=True) -def _hash_exception(cls, fields, globals): +def _hash_exception(cls, fields, func_builder): # Raise an exception. raise TypeError(f'Cannot overwrite attribute __hash__ ' f'in class {cls.__name__}') @@ -879,7 +927,7 @@ _hash_action = {(False, False, False, False): None, def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, - match_args, kw_only, slots): + match_args, kw_only, slots, weakref_slot): # Now that dicts retain insertion order, there's no reason to use # an ordered dict. I am leveraging that ordering here, because # derived class fields overwrite base class fields, but the order @@ -897,13 +945,18 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, globals = {} setattr(cls, _PARAMS, _DataclassParams(init, repr, eq, order, - unsafe_hash, frozen)) + unsafe_hash, frozen, + match_args, kw_only, + slots, weakref_slot)) # Find our base classes in reverse MRO order, and exclude # ourselves. In reversed order so that more derived classes # override earlier field definitions in base classes. As long as - # we're iterating over them, see if any are frozen. + # we're iterating over them, see if all or any of them are frozen. any_frozen_base = False + # By default `all_frozen_bases` is `None` to represent a case, + # where some dataclasses does not have any bases with `_FIELDS` + all_frozen_bases = None has_dataclass_bases = False for b in cls.__mro__[-1:0:-1]: # Only process classes that have been processed by our @@ -913,13 +966,13 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, has_dataclass_bases = True for f in base_fields.values(): fields[f.name] = f - if getattr(b, _PARAMS).frozen: - any_frozen_base = True + if all_frozen_bases is None: + all_frozen_bases = True + current_frozen = getattr(b, _PARAMS).frozen + all_frozen_bases = all_frozen_bases and current_frozen + any_frozen_base = any_frozen_base or current_frozen - # Annotations that are defined in this class (not in base - # classes). If __annotations__ isn't present, then this class - # adds no new annotations. We use this to compute fields that are - # added by this class. + # Annotations defined specifically in this class (not in base classes). # # Fields are found from cls_annotations, which is guaranteed to be # ordered. Default values are from class attributes, if a field @@ -928,7 +981,7 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, # actual default value. Pseudo-fields ClassVars and InitVars are # included, despite the fact that they're not real fields. That's # dealt with later. - cls_annotations = cls.__dict__.get('__annotations__', {}) + cls_annotations = inspect.get_annotations(cls) # Now find fields in our class. While doing so, validate some # things, and set the default values (as class attributes) where @@ -986,7 +1039,7 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, 'frozen one') # Raise an exception if we're frozen, but none of our bases are. - if not any_frozen_base and frozen: + if all_frozen_bases is False and frozen: raise TypeError('cannot inherit frozen dataclass from a ' 'non-frozen one') @@ -997,7 +1050,7 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, # Was this class defined with an explicit __hash__? Note that if # __eq__ is defined in this class, then python will automatically # set __hash__ to None. This is a heuristic, as it's possible - # that such a __hash__ == None was not auto-generated, but it + # that such a __hash__ == None was not auto-generated, but it's # close enough. class_hash = cls.__dict__.get('__hash__', MISSING) has_explicit_hash = not (class_hash is MISSING or @@ -1016,24 +1069,27 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, (std_init_fields, kw_only_init_fields) = _fields_in_init_order(all_init_fields) + func_builder = _FuncBuilder(globals) + if init: # Does this class have a post-init function? has_post_init = hasattr(cls, _POST_INIT_NAME) - _set_new_attribute(cls, '__init__', - _init_fn(all_init_fields, - std_init_fields, - kw_only_init_fields, - frozen, - has_post_init, - # The name to use for the "self" - # param in __init__. Use "self" - # if possible. - '__dataclass_self__' if 'self' in fields - else 'self', - globals, - slots, - )) + _init_fn(all_init_fields, + std_init_fields, + kw_only_init_fields, + frozen, + has_post_init, + # The name to use for the "self" + # param in __init__. Use "self" + # if possible. + '__dataclass_self__' if 'self' in fields + else 'self', + func_builder, + slots, + ) + + _set_new_attribute(cls, '__replace__', _replace) # Get the fields as a list, and include only real fields. This is # used in all of the following methods. @@ -1041,18 +1097,27 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, if repr: flds = [f for f in field_list if f.repr] - _set_new_attribute(cls, '__repr__', _repr_fn(flds, globals)) + func_builder.add_fn('__repr__', + ('self',), + [' return f"{self.__class__.__qualname__}(' + + ', '.join([f"{f.name}={{self.{f.name}!r}}" + for f in flds]) + ')"'], + locals={'__dataclasses_recursive_repr': recursive_repr}, + decorator="@__dataclasses_recursive_repr()") if eq: # Create __eq__ method. There's no need for a __ne__ method, # since python will call __eq__ and negate it. - flds = [f for f in field_list if f.compare] - self_tuple = _tuple_str('self', flds) - other_tuple = _tuple_str('other', flds) - _set_new_attribute(cls, '__eq__', - _cmp_fn('__eq__', '==', - self_tuple, other_tuple, - globals=globals)) + cmp_fields = (field for field in field_list if field.compare) + terms = [f'self.{field.name}==other.{field.name}' for field in cmp_fields] + field_comparisons = ' and '.join(terms) or 'True' + func_builder.add_fn('__eq__', + ('self', 'other'), + [ ' if self is other:', + ' return True', + ' if other.__class__ is self.__class__:', + f' return {field_comparisons}', + ' return NotImplemented']) if order: # Create and set the ordering methods. @@ -1064,18 +1129,19 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, ('__gt__', '>'), ('__ge__', '>='), ]: - if _set_new_attribute(cls, name, - _cmp_fn(name, op, self_tuple, other_tuple, - globals=globals)): - raise TypeError(f'Cannot overwrite attribute {name} ' - f'in class {cls.__name__}. Consider using ' - 'functools.total_ordering') + # Create a comparison function. If the fields in the object are + # named 'x' and 'y', then self_tuple is the string + # '(self.x,self.y)' and other_tuple is the string + # '(other.x,other.y)'. + func_builder.add_fn(name, + ('self', 'other'), + [ ' if other.__class__ is self.__class__:', + f' return {self_tuple}{op}{other_tuple}', + ' return NotImplemented'], + overwrite_error='Consider using functools.total_ordering') if frozen: - for fn in _frozen_get_del_attr(cls, field_list, globals): - if _set_new_attribute(cls, fn.__name__, fn): - raise TypeError(f'Cannot overwrite attribute {fn.__name__} ' - f'in class {cls.__name__}') + _frozen_get_del_attr(cls, field_list, func_builder) # Decide if/how we're going to create a hash function. hash_action = _hash_action[bool(unsafe_hash), @@ -1083,22 +1149,33 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, bool(frozen), has_explicit_hash] if hash_action: - # No need to call _set_new_attribute here, since by the time - # we're here the overwriting is unconditional. - cls.__hash__ = hash_action(cls, field_list, globals) + cls.__hash__ = hash_action(cls, field_list, func_builder) + + # Generate the methods and add them to the class. This needs to be done + # before the __doc__ logic below, since inspect will look at the __init__ + # signature. + func_builder.add_fns_to_class(cls) if not getattr(cls, '__doc__'): # Create a class doc-string. - cls.__doc__ = (cls.__name__ + - str(inspect.signature(cls)).replace(' -> None', '')) + try: + # In some cases fetching a signature is not possible. + # But, we surely should not fail in this case. + text_sig = str(inspect.signature(cls)).replace(' -> None', '') + except (TypeError, ValueError): + text_sig = '' + cls.__doc__ = (cls.__name__ + text_sig) if match_args: - # I could probably compute this once + # I could probably compute this once. _set_new_attribute(cls, '__match_args__', tuple(f.name for f in std_init_fields)) + # It's an error to specify weakref_slot if slots is False. + if weakref_slot and not slots: + raise TypeError('weakref_slot is True but slots is False') if slots: - cls = _add_slots(cls, frozen) + cls = _add_slots(cls, frozen, weakref_slot) abc.update_abstractmethods(cls) @@ -1106,7 +1183,7 @@ def _process_class(cls, init, repr, eq, order, unsafe_hash, frozen, # _dataclass_getstate and _dataclass_setstate are needed for pickling frozen -# classes with slots. These could be slighly more performant if we generated +# classes with slots. These could be slightly more performant if we generated # the code instead of iterating over fields. But that can be a project for # another day, if performance becomes an issue. def _dataclass_getstate(self): @@ -1119,7 +1196,30 @@ def _dataclass_setstate(self, state): object.__setattr__(self, field.name, value) -def _add_slots(cls, is_frozen): +def _get_slots(cls): + match cls.__dict__.get('__slots__'): + # `__dictoffset__` and `__weakrefoffset__` can tell us whether + # the base type has dict/weakref slots, in a way that works correctly + # for both Python classes and C extension types. Extension types + # don't use `__slots__` for slot creation + case None: + slots = [] + if getattr(cls, '__weakrefoffset__', -1) != 0: + slots.append('__weakref__') + if getattr(cls, '__dictoffset__', -1) != 0: + slots.append('__dict__') + yield from slots + case str(slot): + yield slot + # Slots may be any iterable, but we cannot handle an iterator + # because it will already be (partially) consumed. + case iterable if not hasattr(iterable, '__next__'): + yield from iterable + case _: + raise TypeError(f"Slots of '{cls.__name__}' cannot be determined") + + +def _add_slots(cls, is_frozen, weakref_slot): # Need to create a new class, since we can't set __slots__ # after a class has been created. @@ -1130,7 +1230,23 @@ def _add_slots(cls, is_frozen): # Create a new dict for our new class. cls_dict = dict(cls.__dict__) field_names = tuple(f.name for f in fields(cls)) - cls_dict['__slots__'] = field_names + # Make sure slots don't overlap with those in base classes. + inherited_slots = set( + itertools.chain.from_iterable(map(_get_slots, cls.__mro__[1:-1])) + ) + # The slots for our class. Remove slots from our base classes. Add + # '__weakref__' if weakref_slot was given, unless it is already present. + cls_dict["__slots__"] = tuple( + itertools.filterfalse( + inherited_slots.__contains__, + itertools.chain( + # gh-93521: '__weakref__' also needs to be filtered out if + # already present in inherited_slots + field_names, ('__weakref__',) if weakref_slot else () + ) + ), + ) + for field_name in field_names: # Remove our attributes, if present. They'll still be # available in _MARKER. @@ -1139,6 +1255,9 @@ def _add_slots(cls, is_frozen): # Remove __dict__ itself. cls_dict.pop('__dict__', None) + # Clear existing `__weakref__` descriptor, it belongs to a previous type: + cls_dict.pop('__weakref__', None) # gh-102069 + # And finally create the class. qualname = getattr(cls, '__qualname__', None) cls = type(cls)(cls.__name__, cls.__bases__, cls_dict) @@ -1147,33 +1266,35 @@ def _add_slots(cls, is_frozen): if is_frozen: # Need this for pickling frozen classes with slots. - cls.__getstate__ = _dataclass_getstate - cls.__setstate__ = _dataclass_setstate + if '__getstate__' not in cls_dict: + cls.__getstate__ = _dataclass_getstate + if '__setstate__' not in cls_dict: + cls.__setstate__ = _dataclass_setstate return cls def dataclass(cls=None, /, *, init=True, repr=True, eq=True, order=False, unsafe_hash=False, frozen=False, match_args=True, - kw_only=False, slots=False): - """Returns the same class as was passed in, with dunder methods - added based on the fields defined in the class. + kw_only=False, slots=False, weakref_slot=False): + """Add dunder methods based on the fields defined in the class. Examines PEP 526 __annotations__ to determine fields. - If init is true, an __init__() method is added to the class. If - repr is true, a __repr__() method is added. If order is true, rich + If init is true, an __init__() method is added to the class. If repr + is true, a __repr__() method is added. If order is true, rich comparison dunder methods are added. If unsafe_hash is true, a - __hash__() method function is added. If frozen is true, fields may - not be assigned to after instance creation. If match_args is true, - the __match_args__ tuple is added. If kw_only is true, then by - default all fields are keyword-only. If slots is true, an - __slots__ attribute is added. + __hash__() method is added. If frozen is true, fields may not be + assigned to after instance creation. If match_args is true, the + __match_args__ tuple is added. If kw_only is true, then by default + all fields are keyword-only. If slots is true, a new class with a + __slots__ attribute is returned. """ def wrap(cls): return _process_class(cls, init, repr, eq, order, unsafe_hash, - frozen, match_args, kw_only, slots) + frozen, match_args, kw_only, slots, + weakref_slot) # See if we're being called as @dataclass or @dataclass(). if cls is None: @@ -1195,7 +1316,7 @@ def fields(class_or_instance): try: fields = getattr(class_or_instance, _FIELDS) except AttributeError: - raise TypeError('must be called with a dataclass type or instance') + raise TypeError('must be called with a dataclass type or instance') from None # Exclude pseudo-fields. Note that fields is sorted by insertion # order, so the order of the tuple is as the fields were defined. @@ -1210,7 +1331,7 @@ def _is_dataclass_instance(obj): def is_dataclass(obj): """Returns True if obj is a dataclass or an instance of a dataclass.""" - cls = obj if isinstance(obj, type) and not isinstance(obj, GenericAlias) else type(obj) + cls = obj if isinstance(obj, type) else type(obj) return hasattr(cls, _FIELDS) @@ -1218,7 +1339,7 @@ def asdict(obj, *, dict_factory=dict): """Return the fields of a dataclass instance as a new dictionary mapping field names to field values. - Example usage: + Example usage:: @dataclass class C: @@ -1231,7 +1352,7 @@ def asdict(obj, *, dict_factory=dict): If given, 'dict_factory' will be used instead of built-in dict. The function applies recursively to field values that are dataclass instances. This will also look into built-in containers: - tuples, lists, and dicts. + tuples, lists, and dicts. Other objects are copied with 'copy.deepcopy()'. """ if not _is_dataclass_instance(obj): raise TypeError("asdict() should be called on dataclass instances") @@ -1239,42 +1360,69 @@ def asdict(obj, *, dict_factory=dict): def _asdict_inner(obj, dict_factory): - if _is_dataclass_instance(obj): - result = [] - for f in fields(obj): - value = _asdict_inner(getattr(obj, f.name), dict_factory) - result.append((f.name, value)) - return dict_factory(result) - elif isinstance(obj, tuple) and hasattr(obj, '_fields'): - # obj is a namedtuple. Recurse into it, but the returned - # object is another namedtuple of the same type. This is - # similar to how other list- or tuple-derived classes are - # treated (see below), but we just need to create them - # differently because a namedtuple's __init__ needs to be - # called differently (see bpo-34363). + obj_type = type(obj) + if obj_type in _ATOMIC_TYPES: + return obj + elif hasattr(obj_type, _FIELDS): + # dataclass instance: fast path for the common case + if dict_factory is dict: + return { + f.name: _asdict_inner(getattr(obj, f.name), dict) + for f in fields(obj) + } + else: + return dict_factory([ + (f.name, _asdict_inner(getattr(obj, f.name), dict_factory)) + for f in fields(obj) + ]) + # handle the builtin types first for speed; subclasses handled below + elif obj_type is list: + return [_asdict_inner(v, dict_factory) for v in obj] + elif obj_type is dict: + return { + _asdict_inner(k, dict_factory): _asdict_inner(v, dict_factory) + for k, v in obj.items() + } + elif obj_type is tuple: + return tuple([_asdict_inner(v, dict_factory) for v in obj]) + elif issubclass(obj_type, tuple): + if hasattr(obj, '_fields'): + # obj is a namedtuple. Recurse into it, but the returned + # object is another namedtuple of the same type. This is + # similar to how other list- or tuple-derived classes are + # treated (see below), but we just need to create them + # differently because a namedtuple's __init__ needs to be + # called differently (see bpo-34363). - # I'm not using namedtuple's _asdict() - # method, because: - # - it does not recurse in to the namedtuple fields and - # convert them to dicts (using dict_factory). - # - I don't actually want to return a dict here. The main - # use case here is json.dumps, and it handles converting - # namedtuples to lists. Admittedly we're losing some - # information here when we produce a json list instead of a - # dict. Note that if we returned dicts here instead of - # namedtuples, we could no longer call asdict() on a data - # structure where a namedtuple was used as a dict key. - - return type(obj)(*[_asdict_inner(v, dict_factory) for v in obj]) - elif isinstance(obj, (list, tuple)): + # I'm not using namedtuple's _asdict() + # method, because: + # - it does not recurse in to the namedtuple fields and + # convert them to dicts (using dict_factory). + # - I don't actually want to return a dict here. The main + # use case here is json.dumps, and it handles converting + # namedtuples to lists. Admittedly we're losing some + # information here when we produce a json list instead of a + # dict. Note that if we returned dicts here instead of + # namedtuples, we could no longer call asdict() on a data + # structure where a namedtuple was used as a dict key. + return obj_type(*[_asdict_inner(v, dict_factory) for v in obj]) + else: + return obj_type(_asdict_inner(v, dict_factory) for v in obj) + elif issubclass(obj_type, dict): + if hasattr(obj_type, 'default_factory'): + # obj is a defaultdict, which has a different constructor from + # dict as it requires the default_factory as its first arg. + result = obj_type(obj.default_factory) + for k, v in obj.items(): + result[_asdict_inner(k, dict_factory)] = _asdict_inner(v, dict_factory) + return result + return obj_type((_asdict_inner(k, dict_factory), + _asdict_inner(v, dict_factory)) + for k, v in obj.items()) + elif issubclass(obj_type, list): # Assume we can create an object of this type by passing in a - # generator (which is not true for namedtuples, handled - # above). - return type(obj)(_asdict_inner(v, dict_factory) for v in obj) - elif isinstance(obj, dict): - return type(obj)((_asdict_inner(k, dict_factory), - _asdict_inner(v, dict_factory)) - for k, v in obj.items()) + # generator + return obj_type(_asdict_inner(v, dict_factory) for v in obj) else: return copy.deepcopy(obj) @@ -1289,13 +1437,13 @@ def astuple(obj, *, tuple_factory=tuple): x: int y: int - c = C(1, 2) - assert astuple(c) == (1, 2) + c = C(1, 2) + assert astuple(c) == (1, 2) If given, 'tuple_factory' will be used instead of built-in tuple. The function applies recursively to field values that are dataclass instances. This will also look into built-in containers: - tuples, lists, and dicts. + tuples, lists, and dicts. Other objects are copied with 'copy.deepcopy()'. """ if not _is_dataclass_instance(obj): @@ -1304,12 +1452,13 @@ def astuple(obj, *, tuple_factory=tuple): def _astuple_inner(obj, tuple_factory): - if _is_dataclass_instance(obj): - result = [] - for f in fields(obj): - value = _astuple_inner(getattr(obj, f.name), tuple_factory) - result.append(value) - return tuple_factory(result) + if type(obj) in _ATOMIC_TYPES: + return obj + elif _is_dataclass_instance(obj): + return tuple_factory([ + _astuple_inner(getattr(obj, f.name), tuple_factory) + for f in fields(obj) + ]) elif isinstance(obj, tuple) and hasattr(obj, '_fields'): # obj is a namedtuple. Recurse into it, but the returned # object is another namedtuple of the same type. This is @@ -1324,7 +1473,15 @@ def _astuple_inner(obj, tuple_factory): # above). return type(obj)(_astuple_inner(v, tuple_factory) for v in obj) elif isinstance(obj, dict): - return type(obj)((_astuple_inner(k, tuple_factory), _astuple_inner(v, tuple_factory)) + obj_type = type(obj) + if hasattr(obj_type, 'default_factory'): + # obj is a defaultdict, which has a different constructor from + # dict as it requires the default_factory as its first arg. + result = obj_type(getattr(obj, 'default_factory')) + for k, v in obj.items(): + result[_astuple_inner(k, tuple_factory)] = _astuple_inner(v, tuple_factory) + return result + return obj_type((_astuple_inner(k, tuple_factory), _astuple_inner(v, tuple_factory)) for k, v in obj.items()) else: return copy.deepcopy(obj) @@ -1332,17 +1489,18 @@ def _astuple_inner(obj, tuple_factory): def make_dataclass(cls_name, fields, *, bases=(), namespace=None, init=True, repr=True, eq=True, order=False, unsafe_hash=False, - frozen=False, match_args=True, kw_only=False, slots=False): + frozen=False, match_args=True, kw_only=False, slots=False, + weakref_slot=False, module=None): """Return a new dynamically created dataclass. The dataclass name will be 'cls_name'. 'fields' is an iterable of either (name), (name, type) or (name, type, Field) objects. If type is omitted, use the string 'typing.Any'. Field objects are created by - the equivalent of calling 'field(name, type [, Field-info])'. + the equivalent of calling 'field(name, type [, Field-info])'.:: C = make_dataclass('C', ['x', ('y', int), ('z', int, field(init=False))], bases=(Base,)) - is equivalent to: + is equivalent to:: @dataclass class C(Base): @@ -1352,8 +1510,11 @@ def make_dataclass(cls_name, fields, *, bases=(), namespace=None, init=True, For the bases and namespace parameters, see the builtin type() function. - The parameters init, repr, eq, order, unsafe_hash, and frozen are passed to - dataclass(). + The parameters init, repr, eq, order, unsafe_hash, frozen, match_args, kw_only, + slots, and weakref_slot are passed to dataclass(). + + If module parameter is defined, the '__module__' attribute of the dataclass is + set to that value. """ if namespace is None: @@ -1396,16 +1557,30 @@ def make_dataclass(cls_name, fields, *, bases=(), namespace=None, init=True, # of generic dataclasses. cls = types.new_class(cls_name, bases, {}, exec_body_callback) + # For pickling to work, the __module__ variable needs to be set to the frame + # where the dataclass is created. + if module is None: + try: + module = sys._getframemodulename(1) or '__main__' + except AttributeError: + try: + module = sys._getframe(1).f_globals.get('__name__', '__main__') + except (AttributeError, ValueError): + pass + if module is not None: + cls.__module__ = module + # Apply the normal decorator. return dataclass(cls, init=init, repr=repr, eq=eq, order=order, unsafe_hash=unsafe_hash, frozen=frozen, - match_args=match_args, kw_only=kw_only, slots=slots) + match_args=match_args, kw_only=kw_only, slots=slots, + weakref_slot=weakref_slot) def replace(obj, /, **changes): """Return a new object replacing specified fields with new values. - This is especially useful for frozen classes. Example usage: + This is especially useful for frozen classes. Example usage:: @dataclass(frozen=True) class C: @@ -1415,18 +1590,20 @@ def replace(obj, /, **changes): c = C(1, 2) c1 = replace(c, x=3) assert c1.x == 3 and c1.y == 2 - """ - - # We're going to mutate 'changes', but that's okay because it's a - # new dict, even if called with 'replace(obj, **my_changes)'. - + """ if not _is_dataclass_instance(obj): raise TypeError("replace() should be called on dataclass instances") + return _replace(obj, **changes) + + +def _replace(self, /, **changes): + # We're going to mutate 'changes', but that's okay because it's a + # new dict, even if called with 'replace(self, **my_changes)'. # It's an error to have init=False fields in 'changes'. - # If a field is not in 'changes', read its value from the provided obj. + # If a field is not in 'changes', read its value from the provided 'self'. - for f in getattr(obj, _FIELDS).values(): + for f in getattr(self, _FIELDS).values(): # Only consider normal fields or InitVars. if f._field_type is _FIELD_CLASSVAR: continue @@ -1434,20 +1611,20 @@ def replace(obj, /, **changes): if not f.init: # Error if this field is specified in changes. if f.name in changes: - raise ValueError(f'field {f.name} is declared with ' - 'init=False, it cannot be specified with ' - 'replace()') + raise TypeError(f'field {f.name} is declared with ' + f'init=False, it cannot be specified with ' + f'replace()') continue if f.name not in changes: if f._field_type is _FIELD_INITVAR and f.default is MISSING: - raise ValueError(f"InitVar {f.name!r} " - 'must be specified with replace()') - changes[f.name] = getattr(obj, f.name) + raise TypeError(f"InitVar {f.name!r} " + f'must be specified with replace()') + changes[f.name] = getattr(self, f.name) # Create the new object, which calls __init__() and # __post_init__() (if defined), using all of the init fields we've # added and/or left in 'changes'. If there are values supplied in # changes that aren't fields, this will correctly raise a # TypeError. - return obj.__class__(**changes) + return self.__class__(**changes) diff --git a/Lib/test/test_copy.py b/Lib/test/test_copy.py index 2f9d8ed9b..456767bbe 100644 --- a/Lib/test/test_copy.py +++ b/Lib/test/test_copy.py @@ -961,8 +961,6 @@ class TestReplace(unittest.TestCase): with self.assertRaisesRegex(TypeError, 'unexpected field name'): copy.replace(p, x=1, error=2) - # TODO: RUSTPYTHON - @unittest.expectedFailure def test_dataclass(self): from dataclasses import dataclass @dataclass diff --git a/Lib/test/test_dataclasses.py b/Lib/test/test_dataclasses/__init__.py similarity index 80% rename from Lib/test/test_dataclasses.py rename to Lib/test/test_dataclasses/__init__.py index f11edd957..1b16da426 100644 --- a/Lib/test/test_dataclasses.py +++ b/Lib/test/test_dataclasses/__init__.py @@ -5,20 +5,26 @@ from dataclasses import * import abc +import io import pickle import inspect import builtins import types +import weakref +import traceback import unittest from unittest.mock import Mock -from typing import ClassVar, Any, List, Union, Tuple, Dict, Generic, TypeVar, Optional, Protocol +from typing import ClassVar, Any, List, Union, Tuple, Dict, Generic, TypeVar, Optional, Protocol, DefaultDict from typing import get_type_hints -from collections import deque, OrderedDict, namedtuple +from collections import deque, OrderedDict, namedtuple, defaultdict +from copy import deepcopy from functools import total_ordering import typing # Needed for the string "typing.ClassVar[int]" to work as an annotation. import dataclasses # Needed for the string "dataclasses.InitVar[int]" to work as an annotation. +from test import support + # Just any custom exception we can catch. class CustomError(Exception): pass @@ -67,6 +73,50 @@ class TestCase(unittest.TestCase): self.assertEqual(repr_output, expected_output) + def test_field_recursive_repr(self): + rec_field = field() + rec_field.type = rec_field + rec_field.name = "id" + repr_output = repr(rec_field) + + self.assertIn(",type=...,", repr_output) + + def test_recursive_annotation(self): + class C: + pass + + @dataclass + class D: + C: C = field() + + self.assertIn(",type=...,", repr(D.__dataclass_fields__["C"])) + + def test_dataclass_params_repr(self): + # Even though this is testing an internal implementation detail, + # it's testing a feature we want to make sure is correctly implemented + # for the sake of dataclasses itself + @dataclass(slots=True, frozen=True) + class Some: pass + + repr_output = repr(Some.__dataclass_params__) + expected_output = "_DataclassParams(init=True,repr=True," \ + "eq=True,order=False,unsafe_hash=False,frozen=True," \ + "match_args=True,kw_only=False," \ + "slots=True,weakref_slot=False)" + self.assertEqual(repr_output, expected_output) + + def test_dataclass_params_signature(self): + # Even though this is testing an internal implementation detail, + # it's testing a feature we want to make sure is correctly implemented + # for the sake of dataclasses itself + @dataclass + class Some: pass + + for param in inspect.signature(dataclass).parameters: + if param == 'cls': + continue + self.assertTrue(hasattr(Some.__dataclass_params__, param), msg=param) + def test_named_init_params(self): @dataclass class C: @@ -87,7 +137,7 @@ class TestCase(unittest.TestCase): # Non-defaults following defaults. with self.assertRaisesRegex(TypeError, "non-default argument 'y' follows " - "default argument"): + "default argument 'x'"): @dataclass class C: x: int = 0 @@ -96,7 +146,7 @@ class TestCase(unittest.TestCase): # A derived class adds a non-default field after a default one. with self.assertRaisesRegex(TypeError, "non-default argument 'y' follows " - "default argument"): + "default argument 'x'"): @dataclass class B: x: int = 0 @@ -109,7 +159,7 @@ class TestCase(unittest.TestCase): # a field which didn't use to have a default. with self.assertRaisesRegex(TypeError, "non-default argument 'y' follows " - "default argument"): + "default argument 'x'"): @dataclass class B: x: int @@ -230,6 +280,31 @@ class TestCase(unittest.TestCase): c = C('foo') self.assertEqual(c.object, 'foo') + def test_field_named_BUILTINS_frozen(self): + # gh-96151 + @dataclass(frozen=True) + class C: + BUILTINS: int + c = C(5) + self.assertEqual(c.BUILTINS, 5) + + def test_field_with_special_single_underscore_names(self): + # gh-98886 + + @dataclass + class X: + x: int = field(default_factory=lambda: 111) + _dflt_x: int = field(default_factory=lambda: 222) + + X() + + @dataclass + class Y: + y: int = field(default_factory=lambda: 111) + _HAS_DEFAULT_FACTORY: int = 222 + + assert Y(y=222).y == 222 + def test_field_named_like_builtin(self): # Attribute names can shadow built-in names # since code generation is used. @@ -501,6 +576,32 @@ class TestCase(unittest.TestCase): self.assertNotEqual(C(3), C(4, 10)) self.assertNotEqual(C(3, 10), C(4, 10)) + def test_no_unhashable_default(self): + # See bpo-44674. + class Unhashable: + __hash__ = None + + unhashable_re = 'mutable default .* for field a is not allowed' + with self.assertRaisesRegex(ValueError, unhashable_re): + @dataclass + class A: + a: dict = {} + + with self.assertRaisesRegex(ValueError, unhashable_re): + @dataclass + class A: + a: Any = Unhashable() + + # Make sure that the machinery looking for hashability is using the + # class's __hash__, not the instance's __hash__. + with self.assertRaisesRegex(ValueError, unhashable_re): + unhashable = Unhashable() + # This shouldn't make the variable hashable. + unhashable.__hash__ = lambda: 0 + @dataclass + class A: + a: Any = unhashable + def test_hash_field_rules(self): # Test all 6 cases of: # hash=True/False/None @@ -659,8 +760,8 @@ class TestCase(unittest.TestCase): class Subclass(typ): pass with self.assertRaisesRegex(ValueError, - f"mutable default .*Subclass'>" - ' for field z is not allowed' + "mutable default .*Subclass'>" + " for field z is not allowed" ): @dataclass class Point: @@ -668,12 +769,12 @@ class TestCase(unittest.TestCase): # Because this is a ClassVar, it can be mutable. @dataclass - class C: + class UsesMutableClassVar: z: ClassVar[typ] = typ() # Because this is a ClassVar, it can be mutable. @dataclass - class C: + class UsesMutableClassVarWithSubType: x: ClassVar[typ] = Subclass() def test_deliberately_mutable_defaults(self): @@ -990,6 +1091,65 @@ class TestCase(unittest.TestCase): self.assertEqual((c.x, c.y), (3, 4)) self.assertTrue(C.flag) + def test_post_init_not_auto_added(self): + # See bpo-46757, which had proposed always adding __post_init__. As + # Raymond Hettinger pointed out, that would be a breaking change. So, + # add a test to make sure that the current behavior doesn't change. + + @dataclass + class A0: + pass + + @dataclass + class B0: + b_called: bool = False + def __post_init__(self): + self.b_called = True + + @dataclass + class C0(A0, B0): + c_called: bool = False + def __post_init__(self): + super().__post_init__() + self.c_called = True + + # Since A0 has no __post_init__, and one wasn't automatically added + # (because that's the rule: it's never added by @dataclass, it's only + # the class author that can add it), then B0.__post_init__ is called. + # Verify that. + c = C0() + self.assertTrue(c.b_called) + self.assertTrue(c.c_called) + + ###################################### + # Now, the same thing, except A1 defines __post_init__. + @dataclass + class A1: + def __post_init__(self): + pass + + @dataclass + class B1: + b_called: bool = False + def __post_init__(self): + self.b_called = True + + @dataclass + class C1(A1, B1): + c_called: bool = False + def __post_init__(self): + super().__post_init__() + self.c_called = True + + # This time, B1.__post_init__ isn't being called. This mimics what + # would happen if A1.__post_init__ had been automatically added, + # instead of manually added as we see here. This test isn't really + # needed, but I'm including it just to demonstrate the changed + # behavior when A1 does define __post_init__. + c = C1() + self.assertFalse(c.b_called) + self.assertTrue(c.c_called) + def test_class_var(self): # Make sure ClassVars are ignored in __init__, __repr__, etc. @dataclass @@ -1158,6 +1318,29 @@ class TestCase(unittest.TestCase): c = C(10, 11, 50, 51) self.assertEqual(vars(c), {'x': 21, 'y': 101}) + def test_init_var_name_shadowing(self): + # Because dataclasses rely exclusively on `__annotations__` for + # handling InitVar and `__annotations__` preserves shadowed definitions, + # you can actually shadow an InitVar with a method or property. + # + # This only works when there is no default value; `dataclasses` uses the + # actual name (which will be bound to the shadowing method) for default + # values. + @dataclass + class C: + shadowed: InitVar[int] + _shadowed: int = field(init=False) + + def __post_init__(self, shadowed): + self._shadowed = shadowed * 2 + + @property + def shadowed(self): + return self._shadowed * 3 + + c = C(5) + self.assertEqual(c.shadowed, 30) + def test_default_factory(self): # Test a factory that returns a new list. @dataclass @@ -1365,6 +1548,24 @@ class TestCase(unittest.TestCase): self.assertTrue(is_dataclass(type(a))) self.assertTrue(is_dataclass(a)) + def test_is_dataclass_inheritance(self): + @dataclass + class X: + y: int + + class Z(X): + pass + + self.assertTrue(is_dataclass(X), "X should be a dataclass") + self.assertTrue( + is_dataclass(Z), + "Z should be a dataclass because it inherits from X", + ) + z_instance = Z(y=5) + self.assertTrue( + is_dataclass(z_instance), + "z_instance should be a dataclass because it is an instance of Z", + ) def test_helper_fields_with_class_instance(self): # Check that we can call fields() on either a class or instance, @@ -1388,6 +1589,16 @@ class TestCase(unittest.TestCase): with self.assertRaisesRegex(TypeError, 'dataclass type or instance'): fields(C()) + def test_clean_traceback_from_fields_exception(self): + stdout = io.StringIO() + try: + fields(object) + except TypeError as exc: + traceback.print_exception(exc, file=stdout) + printed_traceback = stdout.getvalue() + self.assertNotIn("AttributeError", printed_traceback) + self.assertNotIn("__dataclass_fields__", printed_traceback) + def test_helper_asdict(self): # Basic tests for asdict(), it should return a new dictionary. @dataclass @@ -1565,6 +1776,23 @@ class TestCase(unittest.TestCase): self.assertIsNot(d['f'], t) self.assertEqual(d['f'].my_a(), 6) + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_helper_asdict_defaultdict(self): + # Ensure asdict() does not throw exceptions when a + # defaultdict is a member of a dataclass + @dataclass + class C: + mp: DefaultDict[str, List] + + dd = defaultdict(list) + dd["x"].append(12) + c = C(mp=dd) + d = asdict(c) + + self.assertEqual(d, {"mp": {"x": [12]}}) + self.assertTrue(d["mp"] is not c.mp) # make sure defaultdict is copied + def test_helper_astuple(self): # Basic tests for astuple(), it should return a new tuple. @dataclass @@ -1692,6 +1920,23 @@ class TestCase(unittest.TestCase): t = astuple(c, tuple_factory=list) self.assertEqual(t, ['outer', T(1, ['inner', T(11, 12, 13)], 2)]) + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_helper_astuple_defaultdict(self): + # Ensure astuple() does not throw exceptions when a + # defaultdict is a member of a dataclass + @dataclass + class C: + mp: DefaultDict[str, List] + + dd = defaultdict(list) + dd["x"].append(12) + c = C(mp=dd) + t = astuple(c) + + self.assertEqual(t, ({"x": [12]},)) + self.assertTrue(t[0] is not dd) # make sure defaultdict is copied + def test_dynamic_class_creation(self): cls_dict = {'__annotations__': {'x': int, 'y': int}, } @@ -2019,6 +2264,7 @@ class TestDocString(unittest.TestCase): # whitespace stripped. self.assertEqual(a.replace(' ', ''), b.replace(' ', '')) + @support.requires_docstrings def test_existing_docstring_not_overridden(self): @dataclass class C: @@ -2100,13 +2346,25 @@ class TestDocString(unittest.TestCase): self.assertDocStrEqual(C.__doc__, "C(x:collections.deque=)") + def test_docstring_with_no_signature(self): + # See https://github.com/python/cpython/issues/103449 + class Meta(type): + __call__ = dict + class Base(metaclass=Meta): + pass + + @dataclass + class C(Base): + pass + + self.assertDocStrEqual(C.__doc__, "C") + class TestInit(unittest.TestCase): def test_base_has_init(self): class B: def __init__(self): self.z = 100 - pass # Make sure that declaring this class doesn't raise an error. # The issue is that we can't override __init__ in our class, @@ -2129,12 +2387,12 @@ class TestInit(unittest.TestCase): self.assertEqual(c.z, 100) def test_no_init(self): - dataclass(init=False) + @dataclass(init=False) class C: i: int = 0 self.assertEqual(C().i, 0) - dataclass(init=False) + @dataclass(init=False) class C: i: int = 2 def __init__(self): @@ -2259,6 +2517,15 @@ class TestRepr(unittest.TestCase): class TestEq(unittest.TestCase): + def test_recursive_eq(self): + # Test a class with recursive child + @dataclass + class C: + recursive: object = ... + c = C() + c.recursive = c + self.assertEqual(c, c) + def test_no_eq(self): # Test a class with no __eq__ and eq=False. @dataclass(eq=False) @@ -2613,6 +2880,19 @@ class TestFrozen(unittest.TestCase): c.i = 5 self.assertEqual(c.i, 10) + def test_frozen_empty(self): + @dataclass(frozen=True) + class C: + pass + + c = C() + self.assertFalse(hasattr(c, 'i')) + with self.assertRaises(FrozenInstanceError): + c.i = 5 + self.assertFalse(hasattr(c, 'i')) + with self.assertRaises(FrozenInstanceError): + del c.i + def test_inherit(self): @dataclass(frozen=True) class C: @@ -2641,6 +2921,101 @@ class TestFrozen(unittest.TestCase): class D(C): j: int + def test_inherit_frozen_mutliple_inheritance(self): + @dataclass + class NotFrozen: + pass + + @dataclass(frozen=True) + class Frozen: + pass + + class NotDataclass: + pass + + for bases in ( + (NotFrozen, Frozen), + (Frozen, NotFrozen), + (Frozen, NotDataclass), + (NotDataclass, Frozen), + ): + with self.subTest(bases=bases): + with self.assertRaisesRegex( + TypeError, + 'cannot inherit non-frozen dataclass from a frozen one', + ): + @dataclass + class NotFrozenChild(*bases): + pass + + for bases in ( + (NotFrozen, Frozen), + (Frozen, NotFrozen), + (NotFrozen, NotDataclass), + (NotDataclass, NotFrozen), + ): + with self.subTest(bases=bases): + with self.assertRaisesRegex( + TypeError, + 'cannot inherit frozen dataclass from a non-frozen one', + ): + @dataclass(frozen=True) + class FrozenChild(*bases): + pass + + def test_inherit_frozen_mutliple_inheritance_regular_mixins(self): + @dataclass(frozen=True) + class Frozen: + pass + + class NotDataclass: + pass + + class C1(Frozen, NotDataclass): + pass + self.assertEqual(C1.__mro__, (C1, Frozen, NotDataclass, object)) + + class C2(NotDataclass, Frozen): + pass + self.assertEqual(C2.__mro__, (C2, NotDataclass, Frozen, object)) + + @dataclass(frozen=True) + class C3(Frozen, NotDataclass): + pass + self.assertEqual(C3.__mro__, (C3, Frozen, NotDataclass, object)) + + @dataclass(frozen=True) + class C4(NotDataclass, Frozen): + pass + self.assertEqual(C4.__mro__, (C4, NotDataclass, Frozen, object)) + + def test_multiple_frozen_dataclasses_inheritance(self): + @dataclass(frozen=True) + class FrozenA: + pass + + @dataclass(frozen=True) + class FrozenB: + pass + + class C1(FrozenA, FrozenB): + pass + self.assertEqual(C1.__mro__, (C1, FrozenA, FrozenB, object)) + + class C2(FrozenB, FrozenA): + pass + self.assertEqual(C2.__mro__, (C2, FrozenB, FrozenA, object)) + + @dataclass(frozen=True) + class C3(FrozenA, FrozenB): + pass + self.assertEqual(C3.__mro__, (C3, FrozenA, FrozenB, object)) + + @dataclass(frozen=True) + class C4(FrozenB, FrozenA): + pass + self.assertEqual(C4.__mro__, (C4, FrozenB, FrozenA, object)) + def test_inherit_nonfrozen_from_empty(self): @dataclass class C: @@ -2736,6 +3111,37 @@ class TestFrozen(unittest.TestCase): self.assertEqual(s.y, 10) self.assertEqual(s.cached, True) + with self.assertRaises(FrozenInstanceError): + del s.x + self.assertEqual(s.x, 3) + with self.assertRaises(FrozenInstanceError): + del s.y + self.assertEqual(s.y, 10) + del s.cached + self.assertFalse(hasattr(s, 'cached')) + with self.assertRaises(AttributeError) as cm: + del s.cached + self.assertNotIsInstance(cm.exception, FrozenInstanceError) + + def test_non_frozen_normal_derived_from_empty_frozen(self): + @dataclass(frozen=True) + class D: + pass + + class S(D): + pass + + s = S() + self.assertFalse(hasattr(s, 'x')) + s.x = 5 + self.assertEqual(s.x, 5) + + del s.x + self.assertFalse(hasattr(s, 'x')) + with self.assertRaises(AttributeError) as cm: + del s.x + self.assertNotIsInstance(cm.exception, FrozenInstanceError) + def test_overwriting_frozen(self): # frozen uses __setattr__ and __delattr__. with self.assertRaisesRegex(TypeError, @@ -2774,6 +3180,48 @@ class TestFrozen(unittest.TestCase): with self.assertRaisesRegex(TypeError, 'unhashable type'): hash(C({})) + def test_frozen_deepcopy_without_slots(self): + # see: https://github.com/python/cpython/issues/89683 + @dataclass(frozen=True, slots=False) + class C: + s: str + + c = C('hello') + self.assertEqual(deepcopy(c), c) + + def test_frozen_deepcopy_with_slots(self): + # see: https://github.com/python/cpython/issues/89683 + with self.subTest('generated __slots__'): + @dataclass(frozen=True, slots=True) + class C: + s: str + + c = C('hello') + self.assertEqual(deepcopy(c), c) + + with self.subTest('user-defined __slots__ and no __{get,set}state__'): + @dataclass(frozen=True, slots=False) + class C: + __slots__ = ('s',) + s: str + + # with user-defined slots, __getstate__ and __setstate__ are not + # automatically added, hence the error + err = r"^cannot\ assign\ to\ field\ 's'$" + self.assertRaisesRegex(FrozenInstanceError, err, deepcopy, C('')) + + with self.subTest('user-defined __slots__ and __{get,set}state__'): + @dataclass(frozen=True, slots=False) + class C: + __slots__ = ('s',) + __getstate__ = dataclasses._dataclass_getstate + __setstate__ = dataclasses._dataclass_setstate + + s: str + + c = C('hello') + self.assertEqual(deepcopy(c), c) + class TestSlots(unittest.TestCase): def test_simple(self): @@ -2841,23 +3289,58 @@ class TestSlots(unittest.TestCase): x: int def test_generated_slots_value(self): - @dataclass(slots=True) - class Base: - x: int - self.assertEqual(Base.__slots__, ('x',)) + class Root: + __slots__ = {'x'} + + class Root2(Root): + __slots__ = {'k': '...', 'j': ''} + + class Root3(Root2): + __slots__ = ['h'] + + class Root4(Root3): + __slots__ = 'aa' @dataclass(slots=True) - class Delivered(Base): + class Base(Root4): y: int + j: str + h: str - self.assertEqual(Delivered.__slots__, ('x', 'y')) + self.assertEqual(Base.__slots__, ('y', )) + + @dataclass(slots=True) + class Derived(Base): + aa: float + x: str + z: int + k: str + h: str + + self.assertEqual(Derived.__slots__, ('z', )) @dataclass - class AnotherDelivered(Base): + class AnotherDerived(Base): z: int - self.assertTrue('__slots__' not in AnotherDelivered.__dict__) + self.assertNotIn('__slots__', AnotherDerived.__dict__) + + def test_cant_inherit_from_iterator_slots(self): + + class Root: + __slots__ = iter(['a']) + + class Root2(Root): + __slots__ = ('b', ) + + with self.assertRaisesRegex( + TypeError, + "^Slots of 'Root' cannot be determined" + ): + @dataclass(slots=True) + class C(Root2): + x: int def test_returns_new_class(self): class A: @@ -2896,6 +3379,74 @@ class TestSlots(unittest.TestCase): self.assertIsNot(obj, p) self.assertEqual(obj, p) + @dataclass(frozen=True, slots=True) + class FrozenSlotsGetStateClass: + foo: str + bar: int + + getstate_called: bool = field(default=False, compare=False) + + def __getstate__(self): + object.__setattr__(self, 'getstate_called', True) + return [self.foo, self.bar] + + @dataclass(frozen=True, slots=True) + class FrozenSlotsSetStateClass: + foo: str + bar: int + + setstate_called: bool = field(default=False, compare=False) + + def __setstate__(self, state): + object.__setattr__(self, 'setstate_called', True) + object.__setattr__(self, 'foo', state[0]) + object.__setattr__(self, 'bar', state[1]) + + @dataclass(frozen=True, slots=True) + class FrozenSlotsAllStateClass: + foo: str + bar: int + + getstate_called: bool = field(default=False, compare=False) + setstate_called: bool = field(default=False, compare=False) + + def __getstate__(self): + object.__setattr__(self, 'getstate_called', True) + return [self.foo, self.bar] + + def __setstate__(self, state): + object.__setattr__(self, 'setstate_called', True) + object.__setattr__(self, 'foo', state[0]) + object.__setattr__(self, 'bar', state[1]) + + def test_frozen_slots_pickle_custom_state(self): + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + obj = self.FrozenSlotsGetStateClass('a', 1) + dumped = pickle.dumps(obj, protocol=proto) + + self.assertTrue(obj.getstate_called) + self.assertEqual(obj, pickle.loads(dumped)) + + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + obj = self.FrozenSlotsSetStateClass('a', 1) + obj2 = pickle.loads(pickle.dumps(obj, protocol=proto)) + + self.assertTrue(obj2.setstate_called) + self.assertEqual(obj, obj2) + + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + obj = self.FrozenSlotsAllStateClass('a', 1) + dumped = pickle.dumps(obj, protocol=proto) + + self.assertTrue(obj.getstate_called) + + obj2 = pickle.loads(dumped) + self.assertTrue(obj2.setstate_called) + self.assertEqual(obj, obj2) + def test_slots_with_default_no_init(self): # Originally reported in bpo-44649. @dataclass(slots=True) @@ -2918,6 +3469,311 @@ class TestSlots(unittest.TestCase): self.assertEqual(obj.a, 'a') self.assertEqual(obj.b, 'b') + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_slots_no_weakref(self): + @dataclass(slots=True) + class A: + # No weakref. + pass + + self.assertNotIn("__weakref__", A.__slots__) + a = A() + with self.assertRaisesRegex(TypeError, + "cannot create weak reference"): + weakref.ref(a) + with self.assertRaises(AttributeError): + a.__weakref__ + + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_slots_weakref(self): + @dataclass(slots=True, weakref_slot=True) + class A: + a: int + + self.assertIn("__weakref__", A.__slots__) + a = A(1) + a_ref = weakref.ref(a) + + self.assertIs(a.__weakref__, a_ref) + + def test_slots_weakref_base_str(self): + class Base: + __slots__ = '__weakref__' + + @dataclass(slots=True) + class A(Base): + a: int + + # __weakref__ is in the base class, not A. But an A is still weakref-able. + self.assertIn("__weakref__", Base.__slots__) + self.assertNotIn("__weakref__", A.__slots__) + a = A(1) + weakref.ref(a) + + def test_slots_weakref_base_tuple(self): + # Same as test_slots_weakref_base, but use a tuple instead of a string + # in the base class. + class Base: + __slots__ = ('__weakref__',) + + @dataclass(slots=True) + class A(Base): + a: int + + # __weakref__ is in the base class, not A. But an A is still + # weakref-able. + self.assertIn("__weakref__", Base.__slots__) + self.assertNotIn("__weakref__", A.__slots__) + a = A(1) + weakref.ref(a) + + def test_weakref_slot_without_slot(self): + with self.assertRaisesRegex(TypeError, + "weakref_slot is True but slots is False"): + @dataclass(weakref_slot=True) + class A: + a: int + + def test_weakref_slot_make_dataclass(self): + A = make_dataclass('A', [('a', int),], slots=True, weakref_slot=True) + self.assertIn("__weakref__", A.__slots__) + a = A(1) + weakref.ref(a) + + # And make sure if raises if slots=True is not given. + with self.assertRaisesRegex(TypeError, + "weakref_slot is True but slots is False"): + B = make_dataclass('B', [('a', int),], weakref_slot=True) + + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_weakref_slot_subclass_weakref_slot(self): + @dataclass(slots=True, weakref_slot=True) + class Base: + field: int + + # A *can* also specify weakref_slot=True if it wants to (gh-93521) + @dataclass(slots=True, weakref_slot=True) + class A(Base): + ... + + # __weakref__ is in the base class, not A. But an instance of A + # is still weakref-able. + self.assertIn("__weakref__", Base.__slots__) + self.assertNotIn("__weakref__", A.__slots__) + a = A(1) + a_ref = weakref.ref(a) + self.assertIs(a.__weakref__, a_ref) + + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_weakref_slot_subclass_no_weakref_slot(self): + @dataclass(slots=True, weakref_slot=True) + class Base: + field: int + + @dataclass(slots=True) + class A(Base): + ... + + # __weakref__ is in the base class, not A. Even though A doesn't + # specify weakref_slot, it should still be weakref-able. + self.assertIn("__weakref__", Base.__slots__) + self.assertNotIn("__weakref__", A.__slots__) + a = A(1) + a_ref = weakref.ref(a) + self.assertIs(a.__weakref__, a_ref) + + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_weakref_slot_normal_base_weakref_slot(self): + class Base: + __slots__ = ('__weakref__',) + + @dataclass(slots=True, weakref_slot=True) + class A(Base): + field: int + + # __weakref__ is in the base class, not A. But an instance of + # A is still weakref-able. + self.assertIn("__weakref__", Base.__slots__) + self.assertNotIn("__weakref__", A.__slots__) + a = A(1) + a_ref = weakref.ref(a) + self.assertIs(a.__weakref__, a_ref) + + def test_dataclass_derived_weakref_slot(self): + class A: + pass + + @dataclass(slots=True, weakref_slot=True) + class B(A): + pass + + self.assertEqual(B.__slots__, ()) + B() + + def test_dataclass_derived_generic(self): + T = typing.TypeVar('T') + + @dataclass(slots=True, weakref_slot=True) + class A(typing.Generic[T]): + pass + self.assertEqual(A.__slots__, ('__weakref__',)) + self.assertTrue(A.__weakref__) + A() + + @dataclass(slots=True, weakref_slot=True) + class B[T2]: + pass + self.assertEqual(B.__slots__, ('__weakref__',)) + self.assertTrue(B.__weakref__) + B() + + # TODO: RUSTPYTHON + @unittest.expectedFailure + def test_dataclass_derived_generic_from_base(self): + T = typing.TypeVar('T') + + class RawBase: ... + + @dataclass(slots=True, weakref_slot=True) + class C1(typing.Generic[T], RawBase): + pass + self.assertEqual(C1.__slots__, ()) + self.assertTrue(C1.__weakref__) + C1() + @dataclass(slots=True, weakref_slot=True) + class C2(RawBase, typing.Generic[T]): + pass + self.assertEqual(C2.__slots__, ()) + self.assertTrue(C2.__weakref__) + C2() + + @dataclass(slots=True, weakref_slot=True) + class D[T2](RawBase): + pass + self.assertEqual(D.__slots__, ()) + self.assertTrue(D.__weakref__) + D() + + def test_dataclass_derived_generic_from_slotted_base(self): + T = typing.TypeVar('T') + + class WithSlots: + __slots__ = ('a', 'b') + + @dataclass(slots=True, weakref_slot=True) + class E1(WithSlots, Generic[T]): + pass + self.assertEqual(E1.__slots__, ('__weakref__',)) + self.assertTrue(E1.__weakref__) + E1() + @dataclass(slots=True, weakref_slot=True) + class E2(Generic[T], WithSlots): + pass + self.assertEqual(E2.__slots__, ('__weakref__',)) + self.assertTrue(E2.__weakref__) + E2() + + @dataclass(slots=True, weakref_slot=True) + class F[T2](WithSlots): + pass + self.assertEqual(F.__slots__, ('__weakref__',)) + self.assertTrue(F.__weakref__) + F() + + def test_dataclass_derived_generic_from_slotted_base_with_weakref(self): + T = typing.TypeVar('T') + + class WithWeakrefSlot: + __slots__ = ('__weakref__',) + + @dataclass(slots=True, weakref_slot=True) + class G1(WithWeakrefSlot, Generic[T]): + pass + self.assertEqual(G1.__slots__, ()) + self.assertTrue(G1.__weakref__) + G1() + @dataclass(slots=True, weakref_slot=True) + class G2(Generic[T], WithWeakrefSlot): + pass + self.assertEqual(G2.__slots__, ()) + self.assertTrue(G2.__weakref__) + G2() + + @dataclass(slots=True, weakref_slot=True) + class H[T2](WithWeakrefSlot): + pass + self.assertEqual(H.__slots__, ()) + self.assertTrue(H.__weakref__) + H() + + def test_dataclass_slot_dict(self): + class WithDictSlot: + __slots__ = ('__dict__',) + + @dataclass(slots=True) + class A(WithDictSlot): ... + + self.assertEqual(A.__slots__, ()) + self.assertEqual(A().__dict__, {}) + A() + + @support.cpython_only + def test_dataclass_slot_dict_ctype(self): + # https://github.com/python/cpython/issues/123935 + from test.support import import_helper + # Skips test if `_testcapi` is not present: + _testcapi = import_helper.import_module('_testcapi') + + @dataclass(slots=True) + class HasDictOffset(_testcapi.HeapCTypeWithDict): + __dict__: dict = {} + self.assertNotEqual(_testcapi.HeapCTypeWithDict.__dictoffset__, 0) + self.assertEqual(HasDictOffset.__slots__, ()) + + @dataclass(slots=True) + class DoesNotHaveDictOffset(_testcapi.HeapCTypeWithWeakref): + __dict__: dict = {} + self.assertEqual(_testcapi.HeapCTypeWithWeakref.__dictoffset__, 0) + self.assertEqual(DoesNotHaveDictOffset.__slots__, ('__dict__',)) + + @support.cpython_only + def test_slots_with_wrong_init_subclass(self): + # TODO: This test is for a kinda-buggy behavior. + # Ideally, it should be fixed and `__init_subclass__` + # should be fully supported in the future versions. + # See https://github.com/python/cpython/issues/91126 + class WrongSuper: + def __init_subclass__(cls, arg): + pass + + with self.assertRaisesRegex( + TypeError, + "missing 1 required positional argument: 'arg'", + ): + @dataclass(slots=True) + class WithWrongSuper(WrongSuper, arg=1): + pass + + class CorrectSuper: + args = [] + def __init_subclass__(cls, arg="default"): + cls.args.append(arg) + + @dataclass(slots=True) + class WithCorrectSuper(CorrectSuper): + pass + + # __init_subclass__ is called twice: once for `WithCorrectSuper` + # and once for `WithCorrectSuper__slots__` new class + # that we create internally. + self.assertEqual(CorrectSuper.args, ["default", "default"]) + + class TestDescriptors(unittest.TestCase): def test_set_name(self): # See bpo-33141. @@ -3202,10 +4058,10 @@ class TestStringAnnotations(unittest.TestCase): self.assertEqual(C(10).x, 10) def test_classvar_module_level_import(self): - from test import dataclass_module_1 - from test import dataclass_module_1_str - from test import dataclass_module_2 - from test import dataclass_module_2_str + from test.test_dataclasses import dataclass_module_1 + from test.test_dataclasses import dataclass_module_1_str + from test.test_dataclasses import dataclass_module_2 + from test.test_dataclasses import dataclass_module_2_str for m in (dataclass_module_1, dataclass_module_1_str, dataclass_module_2, dataclass_module_2_str, @@ -3243,7 +4099,7 @@ class TestStringAnnotations(unittest.TestCase): self.assertNotIn('not_iv4', c.__dict__) def test_text_annotations(self): - from test import dataclass_textanno + from test.test_dataclasses import dataclass_textanno self.assertEqual( get_type_hints(dataclass_textanno.Bar), @@ -3254,6 +4110,15 @@ class TestStringAnnotations(unittest.TestCase): 'return': type(None)}) +ByMakeDataClass = make_dataclass('ByMakeDataClass', [('x', int)]) +ManualModuleMakeDataClass = make_dataclass('ManualModuleMakeDataClass', + [('x', int)], + module=__name__) +WrongNameMakeDataclass = make_dataclass('Wrong', [('x', int)]) +WrongModuleMakeDataclass = make_dataclass('WrongModuleMakeDataclass', + [('x', int)], + module='custom') + class TestMakeDataclass(unittest.TestCase): def test_simple(self): C = make_dataclass('C', @@ -3363,6 +4228,36 @@ class TestMakeDataclass(unittest.TestCase): 'y': int, 'z': 'typing.Any'}) + def test_module_attr(self): + self.assertEqual(ByMakeDataClass.__module__, __name__) + self.assertEqual(ByMakeDataClass(1).__module__, __name__) + self.assertEqual(WrongModuleMakeDataclass.__module__, "custom") + Nested = make_dataclass('Nested', []) + self.assertEqual(Nested.__module__, __name__) + self.assertEqual(Nested().__module__, __name__) + + def test_pickle_support(self): + for klass in [ByMakeDataClass, ManualModuleMakeDataClass]: + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + self.assertEqual( + pickle.loads(pickle.dumps(klass, proto)), + klass, + ) + self.assertEqual( + pickle.loads(pickle.dumps(klass(1), proto)), + klass(1), + ) + + def test_cannot_be_pickled(self): + for klass in [WrongNameMakeDataclass, WrongModuleMakeDataclass]: + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + with self.assertRaises(pickle.PickleError): + pickle.dumps(klass, proto) + with self.assertRaises(pickle.PickleError): + pickle.dumps(klass(1), proto) + def test_invalid_type_specification(self): for bad_field in [(), (1, 2, 3, 4), @@ -3444,9 +4339,9 @@ class TestReplace(unittest.TestCase): self.assertEqual((c1.x, c1.y, c1.z, c1.t), (3, 2, 10, 100)) - with self.assertRaisesRegex(ValueError, 'init=False'): + with self.assertRaisesRegex(TypeError, 'init=False'): replace(c, x=3, z=20, t=50) - with self.assertRaisesRegex(ValueError, 'init=False'): + with self.assertRaisesRegex(TypeError, 'init=False'): replace(c, z=20) replace(c, x=3, z=20, t=50) @@ -3499,10 +4394,10 @@ class TestReplace(unittest.TestCase): self.assertEqual((c1.x, c1.y), (5, 10)) # Trying to replace y is an error. - with self.assertRaisesRegex(ValueError, 'init=False'): + with self.assertRaisesRegex(TypeError, 'init=False'): replace(c, x=2, y=30) - with self.assertRaisesRegex(ValueError, 'init=False'): + with self.assertRaisesRegex(TypeError, 'init=False'): replace(c, y=30) def test_classvar(self): @@ -3535,8 +4430,8 @@ class TestReplace(unittest.TestCase): c = C(1, 10) self.assertEqual(c.x, 10) - with self.assertRaisesRegex(ValueError, r"InitVar 'y' must be " - "specified with replace()"): + with self.assertRaisesRegex(TypeError, r"InitVar 'y' must be " + r"specified with replace\(\)"): replace(c, x=3) c = replace(c, x=3, y=5) self.assertEqual(c.x, 15) @@ -3666,6 +4561,7 @@ class TestAbstract(unittest.TestCase): self.assertFalse(inspect.isabstract(Date)) self.assertGreater(Date(2020,12,25), Date(2020,8,31)) + # TODO: RUSTPYTHON @unittest.expectedFailure def test_maintain_abc(self): class A(abc.ABC): @@ -3680,7 +4576,7 @@ class TestAbstract(unittest.TestCase): day: 'int' self.assertTrue(inspect.isabstract(Date)) - msg = 'class Date with abstract method foo' + msg = "class Date without an implementation for abstract method 'foo'" self.assertRaisesRegex(TypeError, msg, Date) @@ -3926,7 +4822,7 @@ class TestKeywordArgs(unittest.TestCase): # But this usage is okay, since it's not using KW_ONLY. @dataclass - class A: + class NoDuplicateKwOnlyAnnotation: a: int _: KW_ONLY b: int @@ -3934,13 +4830,13 @@ class TestKeywordArgs(unittest.TestCase): # And if inheriting, it's okay. @dataclass - class A: + class BaseUsesKwOnly: a: int _: KW_ONLY b: int c: int @dataclass - class B(A): + class SubclassUsesKwOnly(BaseUsesKwOnly): _: KW_ONLY d: int @@ -4003,7 +4899,7 @@ class TestKeywordArgs(unittest.TestCase): # Make sure we still check for non-kwarg non-defaults not following # defaults. - err_regex = "non-default argument 'z' follows default argument" + err_regex = "non-default argument 'z' follows default argument 'a'" with self.assertRaisesRegex(TypeError, err_regex): @dataclass class A: @@ -4026,4 +4922,4 @@ class TestKeywordArgs(unittest.TestCase): if __name__ == '__main__': - unittest.main() + unittest.main() \ No newline at end of file diff --git a/Lib/test/test_dataclasses/dataclass_module_1.py b/Lib/test/test_dataclasses/dataclass_module_1.py new file mode 100644 index 000000000..87a33f819 --- /dev/null +++ b/Lib/test/test_dataclasses/dataclass_module_1.py @@ -0,0 +1,32 @@ +#from __future__ import annotations +USING_STRINGS = False + +# dataclass_module_1.py and dataclass_module_1_str.py are identical +# except only the latter uses string annotations. + +import dataclasses +import typing + +T_CV2 = typing.ClassVar[int] +T_CV3 = typing.ClassVar + +T_IV2 = dataclasses.InitVar[int] +T_IV3 = dataclasses.InitVar + +@dataclasses.dataclass +class CV: + T_CV4 = typing.ClassVar + cv0: typing.ClassVar[int] = 20 + cv1: typing.ClassVar = 30 + cv2: T_CV2 + cv3: T_CV3 + not_cv4: T_CV4 # When using string annotations, this field is not recognized as a ClassVar. + +@dataclasses.dataclass +class IV: + T_IV4 = dataclasses.InitVar + iv0: dataclasses.InitVar[int] + iv1: dataclasses.InitVar + iv2: T_IV2 + iv3: T_IV3 + not_iv4: T_IV4 # When using string annotations, this field is not recognized as an InitVar. diff --git a/Lib/test/test_dataclasses/dataclass_module_1_str.py b/Lib/test/test_dataclasses/dataclass_module_1_str.py new file mode 100644 index 000000000..6de490b7a --- /dev/null +++ b/Lib/test/test_dataclasses/dataclass_module_1_str.py @@ -0,0 +1,32 @@ +from __future__ import annotations +USING_STRINGS = True + +# dataclass_module_1.py and dataclass_module_1_str.py are identical +# except only the latter uses string annotations. + +import dataclasses +import typing + +T_CV2 = typing.ClassVar[int] +T_CV3 = typing.ClassVar + +T_IV2 = dataclasses.InitVar[int] +T_IV3 = dataclasses.InitVar + +@dataclasses.dataclass +class CV: + T_CV4 = typing.ClassVar + cv0: typing.ClassVar[int] = 20 + cv1: typing.ClassVar = 30 + cv2: T_CV2 + cv3: T_CV3 + not_cv4: T_CV4 # When using string annotations, this field is not recognized as a ClassVar. + +@dataclasses.dataclass +class IV: + T_IV4 = dataclasses.InitVar + iv0: dataclasses.InitVar[int] + iv1: dataclasses.InitVar + iv2: T_IV2 + iv3: T_IV3 + not_iv4: T_IV4 # When using string annotations, this field is not recognized as an InitVar. diff --git a/Lib/test/test_dataclasses/dataclass_module_2.py b/Lib/test/test_dataclasses/dataclass_module_2.py new file mode 100644 index 000000000..68fb733e2 --- /dev/null +++ b/Lib/test/test_dataclasses/dataclass_module_2.py @@ -0,0 +1,32 @@ +#from __future__ import annotations +USING_STRINGS = False + +# dataclass_module_2.py and dataclass_module_2_str.py are identical +# except only the latter uses string annotations. + +from dataclasses import dataclass, InitVar +from typing import ClassVar + +T_CV2 = ClassVar[int] +T_CV3 = ClassVar + +T_IV2 = InitVar[int] +T_IV3 = InitVar + +@dataclass +class CV: + T_CV4 = ClassVar + cv0: ClassVar[int] = 20 + cv1: ClassVar = 30 + cv2: T_CV2 + cv3: T_CV3 + not_cv4: T_CV4 # When using string annotations, this field is not recognized as a ClassVar. + +@dataclass +class IV: + T_IV4 = InitVar + iv0: InitVar[int] + iv1: InitVar + iv2: T_IV2 + iv3: T_IV3 + not_iv4: T_IV4 # When using string annotations, this field is not recognized as an InitVar. diff --git a/Lib/test/test_dataclasses/dataclass_module_2_str.py b/Lib/test/test_dataclasses/dataclass_module_2_str.py new file mode 100644 index 000000000..b363d17c1 --- /dev/null +++ b/Lib/test/test_dataclasses/dataclass_module_2_str.py @@ -0,0 +1,32 @@ +from __future__ import annotations +USING_STRINGS = True + +# dataclass_module_2.py and dataclass_module_2_str.py are identical +# except only the latter uses string annotations. + +from dataclasses import dataclass, InitVar +from typing import ClassVar + +T_CV2 = ClassVar[int] +T_CV3 = ClassVar + +T_IV2 = InitVar[int] +T_IV3 = InitVar + +@dataclass +class CV: + T_CV4 = ClassVar + cv0: ClassVar[int] = 20 + cv1: ClassVar = 30 + cv2: T_CV2 + cv3: T_CV3 + not_cv4: T_CV4 # When using string annotations, this field is not recognized as a ClassVar. + +@dataclass +class IV: + T_IV4 = InitVar + iv0: InitVar[int] + iv1: InitVar + iv2: T_IV2 + iv3: T_IV3 + not_iv4: T_IV4 # When using string annotations, this field is not recognized as an InitVar. diff --git a/Lib/test/test_dataclasses/dataclass_textanno.py b/Lib/test/test_dataclasses/dataclass_textanno.py new file mode 100644 index 000000000..3eb6c943d --- /dev/null +++ b/Lib/test/test_dataclasses/dataclass_textanno.py @@ -0,0 +1,12 @@ +from __future__ import annotations + +import dataclasses + + +class Foo: + pass + + +@dataclasses.dataclass +class Bar: + foo: Foo