Dataclasses

When working with classes, a lot of boiler-plate code is often required: always __init__, __repr__, often __eq__, __lt__ for comparisons, often __hash__ when instances are used in sets or as dict keys, sometimes ‘neutral representation’ astuple, asdict as an export interface. Dataclasses does the job for you and are a huge time saver.

Note

It seems at first glance that class attributes are defined. However, this syntax in the context of a dataclass is called a field. It will actually create class instances with independent attributes. To explicitly define a class attribute, the typing.ClassVar type is required.

dataclasses_usage

from dataclasses import dataclass, field
from typing import ClassVar


@dataclass(order=True)  # allow ordering between instances
class Student:
    last_name: str
    first_name: str
    middle_names: list = field(default_factory=list)  # for a list instance attribute

    id: int = field(init=False)  # not set by user, but automatically after init
    nb_students: ClassVar[int] = 0  # a class attribute shared between instances

    def __post_init__(self):  # optional, here used to fill automatically the "id"
        self.id = self._generate_unique_id()

    @classmethod
    def _generate_unique_id(cls):
        cls.nb_students += 1
        return cls.nb_students


def main() -> None:
    student = Student("D", "a", ["b", "c"])  # using positional
    other_student = Student(
        first_name="a",
        last_name="E",
    )  # using keywords
    print(student)
    print(other_student)
    print(student < other_student)
    # the field order is the comparison order (lastname, then firstname, ...)


if __name__ == "__main__":
    main()