dataclasses

A decorator that synthesises __init__, __repr__, __eq__, ordering, hash, slots, and replacement on a regular Python class. Decorator-generated code is appended to the class at decoration time; introspection sees real methods.

Source-of-record: Lib/dataclasses.py, dataclasses docs.

@dataclass

@dataclass(*, init=True, repr=True, eq=True, order=False,
           unsafe_hash=False, frozen=False, match_args=True,
           kw_only=False, slots=False, weakref_slot=False)

Parameter	Default	Effect
init	True	Generate __init__.
repr	True	Generate __repr__.
eq	True	Generate __eq__.
order	False	Generate <, <=, >, >=.
unsafe_hash	False	Force __hash__ even when mutable.
frozen	False	Immutable: assignments raise FrozenInstanceError.
match_args	True	Generate __match_args__ (PEP 634).
kw_only	False	All fields keyword-only by default.
slots	False	Generate __slots__ and return a new class.
weakref_slot	False	Add __weakref__ to slots.

Fields

field(*, default=MISSING, default_factory=MISSING, init=True,
      repr=True, hash=None, compare=True, metadata=None,
      kw_only=MISSING)

Parameter	Effect
default	Static default value.
default_factory	Zero-arg callable producing a default per instance.
init	Include in __init__.
repr	Include in __repr__.
compare	Use for __eq__ / ordering.
hash	Override; default tracks compare.
metadata	Arbitrary mapping for downstream tools.
kw_only	Override the class-level setting per field.

The sentinel dataclasses.MISSING indicates "no value supplied".

Inspection

Function	Returns
fields(class_or_instance)	Tuple of Field objects in declaration order.
asdict(obj, *, dict_factory=dict)	Recursively convert to dict.
astuple(obj, *, tuple_factory=tuple)	Recursively convert to tuple.
is_dataclass(obj)	Predicate.
replace(obj, /, **changes)	New instance with fields replaced.
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)	Build a dataclass at runtime.

Field attributes: name, type, default, default_factory, repr, hash, init, compare, metadata, kw_only.

Frozen

Frozen instances raise FrozenInstanceError on assignment. __hash__ is generated from the compare fields. Cycle: frozen=True and eq=True together produce a hashable value class.

Slots

slots=True rewrites the class with __slots__ set to the field names. Because __slots__ cannot be added after class creation, the decorator returns a fresh class object. Subclasses must also set __slots__ to keep instances slot-only.

Post-init hook

__post_init__(self) runs at the end of the generated __init__. Use to derive fields from inputs. With init=False, the user provides the __init__ and __post_init__ is not called.

InitVar

from dataclasses import InitVar
@dataclass
class C:
    x: int
    y: InitVar[int] = 0
    def __post_init__(self, y): ...

InitVar[T] fields are passed to __init__ and __post_init__ but are not stored on the instance.

ClassVar

typing.ClassVar[T] annotations are excluded from the field set; they remain ordinary class attributes.

Gopy status

Area	State
@dataclass and all options	Complete.
field, MISSING	Complete.
fields, asdict, astuple, replace, make_dataclass	Complete.
__post_init__, InitVar, ClassVar	Complete.
slots=True, weakref_slot=True	Complete.

Reference

• CPython 3.14: dataclasses.
• Lib/dataclasses.py.
• module/dataclasses/. gopy port.
• PEP 557.