Dataclass ABC

Library that lets you define abstract properties for dataclasses.

Installation

pip install dataclass-abc

Usage

The dataclass_abc class decorator resolves the abstract properties overwritten by a field.

from abc import ABC, abstractmethod

from dataclass_abc import dataclass_abc

class A(ABC):
    @property
    @abstractmethod
    def name(self) -> str:
        ...

@dataclass_abc(frozen=True)
class B(A):
    name: str        # overwrites the abstract property 'name' in 'A'

Define mutable variables

Define a mutable variable name in the abstract class A by using the name.setter decorator.

from abc import ABC, abstractmethod

from dataclass_abc import dataclass_abc

class A(ABC):
    @property
    @abstractmethod
    def name(self) -> str:
        ...

    @name.setter
    @abstractmethod
    def name(self, val: str):
        ...

    def set_name(self, val: str):
        self.name = val

@dataclass_abc
class B(A):
    name: str

b = B(name='A')
b.set_name('B')

Example

The example implements the code snippets taken from RealPython with abstract properties.

Design pattern

This library suggests the following design pattern:

• mixins - a mixin is an abstract class that implements data as abstract properties and methods based on the abstract properties.
• classes - an abstract class inherits from one or more mixins (see City or CapitalCity in the example). This class is used for pattern matching, e.g. using isinstance method.
• impl - an implementation class implements the abstract properties. (see CityImpl or CapitalCityImpl in the example). This class is decorated with dataclass_abc and resolve_abc_prop and should always be called through an initialize function.
• init - an initialize function (or constructor function) initializes an implementation class.