dataclasses

Dataclasses are a Python feature provided by the dataclasses module that simplify the creation of classes whose primary purpose is to store data. Introduced in Python 3.7 (PEP 557), they reduce boilerplate by automatically generating common methods. When a class is decorated with @dataclass and its fields are annotated with types, Python generates an init, repr, and eq methods by default, and can also generate ordering methods if requested.

Fields in a dataclass are defined with type annotations. You can assign default values or use default_factory for fields that require a mutable default, such as lists or dicts. The dataclass decorator accepts several options, including init,repr, eq, order, and frozen, which control whether the corresponding methods are generated and whether instances are immutable. The field() function provides per-field customization, including default values, default_factory, whether a field participates in init, and metadata attached to the field.

Special markers include InitVar, used for constructor parameters that are not stored as attributes, and ClassVar, which marks class-level data that should be ignored by dataclass-generated methods.

Post-initialization and utilities: post_init is invoked after the generated init to perform extra initialization. In frozen dataclasses, attributes cannot be assigned after creation, though object.setattr can be used within post_init if necessary. The module also provides asdict and astuple to convert dataclass instances to dicts or tuples, and replace to create a new instance with some fields updated. Dataclasses are well suited for simple data containers and can be less verbose than manually writing boilerplate code.