Mypy 子类中更具体的参数

Question

我想用抽象方法声明一个基础 class，该方法具有类型化参数，以便实现 classes 可以为该参数指定更具体的类型，例如：

from abc import ABC, abstractmethod

class Job(ABC):
    pass

class EasyJob(Job):
    pass

class HardJob(Job):
    pass

class Worker(ABC):
    @abstractmethod
    def run(self, job: Job) -> None:
        raise NotImplementedError()

class EasyWorker(Worker):
    def run(self, job: EasyJob) -> None:
        pass

class HardWorker(Worker):
    def run(self, job: HardJob) -> None:
        pass

然而，mypy 对此抱怨是可以理解的：

line 14: error: Argument 1 of "run" is incompatible with supertype "Worker"; supertype defines the argument type as "Job"
line 18: error: Argument 1 of "run" is incompatible with supertype "Worker"; supertype defines the argument type as "Job"

Python有什么方法可以促进这样的结构吗？

Answer 1

您可能想要 Bounded Parametric Polymorphism. See also this section about bounded type variables。

在 Worker 中，您想说您的方法运行是通用的，因为它需要一个值某些未指定类型 T 而 T 是 Job 的子类型。类派生自 Worker 然后用 T 代替具体类型：

from abc import ABC, abstractmethod
from typing import Generic, TypeVar


class Job(ABC):
    pass


class EasyJob(Job):
    pass


class HardJob(Job):
    pass

T = TypeVar('T', bound=Job)


class Worker(ABC, Generic[T]):
    @abstractmethod
    def run(self, job: T) -> None:
        raise NotImplementedError()


class EasyWorker(Worker[EasyJob]):
    def run(self, job: EasyJob) -> None:
        pass


class HardWorker(Worker[HardJob]):
    def run(self, job: HardJob) -> None:
        pass

Mypy 子类中更具体的参数

Mypy more specific parameter in subclass

python

type-hinting

mypy

python-typing