dataclass

Dataclass is a Python construct that provides a decorator and helper functions for creating classes intended primarily for storing data. It automates common boilerplate tasks by generating methods such as __init__, __repr__, and __eq__, and, if requested, ordering methods. Introduced in Python 3.7 as part of the dataclasses module, it can also be backported to earlier versions via a separate package.

Usage: declare a class with type-annotated attributes and apply @dataclass. Fields with defaults must follow non-default fields. Default values are shared across instances, so use default_factory for mutable types like list or dict. Per-field behavior can be customized with field(), which can set default, default_factory, init, repr, compare, metadata, or exclude a field from certain generated methods. Class-level options include init, repr, eq, order, frozen, and post-init validation via __post_init__.

Dataclasses provide helpful utilities such as asdict() and astuple() to convert instances to dictionaries or tuples, and replace() to create a copy with some fields altered. They pair well with type hints and support inheritance.

Limitations: dataclasses rely on type annotations and simple storage patterns; they are not a full object system replacement. Complex invariants should be enforced in __post_init__, and mutable fields require careful handling to avoid shared state. Although frozen dataclasses prevent attribute reassignment, they can still contain mutable objects that may be modified in place.