mypy 不理解为 None 值引发异常

Question

我在另一个函数中创建一个函数，如果其中一个变量是 None 以引发 ValueError。但是mypy看不懂，我的代码是：

from typing import Mapping, Optional, Union


def make_setting(name: str, age: Optional[int], score: Optional[int]) -> Mapping[str, Union[str, int]]:
    def check(the_first_name,*arg):
        if None in arg:
            raise ValueError('The args of "{}" should not be None.'.format(the_first_name))
    check(name, age, score)
    return make_setting_of_policy(name, age, score)

def make_setting_of_policy(name: str, age:int, score: int) -> Mapping[str, Union[str, int]]:
    return {'name':name, 'age':age, 'score':score}

mypy 不理解 make_setting_of_policy 的输入不是 None，并显示以下错误：

Argument 2 to "make_setting_of_policy" has incompatible type "Optional[int]"; expected "int"
Argument 3 to "make_setting_of_policy" has incompatible type "Optional[int]"; expected "int"

Answer 1

mypy doesn't understand the input of make_setting_of_policy is not None

是的，问题是关于 type narrowing Optional 参数以删除潜在的 None。 Mypy 不考虑引发 run-time 异常。最简单的解决方案是直接检查参数（这应该适用于所有支持 PEP 484 的 Python 3.5+ 版本。）

from typing import Mapping, Optional, Union


def make_setting(name: str, age: Optional[int], score: Optional[int]) -> Mapping[str, Union[str, int]]:

    if age is not None and score is not None:
        return make_setting_of_policy(name, age, score)  # runtime type check
    else:
        raise ValueError('The args of "{}" should not be None.'.format(name))  # runtime exception


def make_setting_of_policy(name: str, age: int, score: int) -> Mapping[str, Union[str, int]]:
    return {'name': name, 'age': age, 'score': score}

mypy 结果：

Success: no issues found in 1 source file

如果使用 Python 3.10，您可以使用 TypeGuards，注意其特定规则。

PEP 647 - TypeGuard Type

When a conditional statement includes a call to a user-defined type guard function, and that function returns true, the expression passed as the first positional argument to the type guard function should be assumed by a static type checker to take on the type specified in the TypeGuard return type, unless and until it is further narrowed within the conditional code block.

使用 TypeGuard 的解决方案（使用单个参数，因此我们不重复前面示例的缩小）如下所示：

from typing import Mapping, Optional, Union, TypeGuard


def check(args: tuple[str, Optional[int], Optional[int]]) -> TypeGuard[tuple[str, int, int]]:
    if None in args:
        raise ValueError('The args of "{}" should not be None.'.format(args[0]))
    return all(x is not None for x in args)


def make_setting(name: str, age: Optional[int], score: Optional[int]) -> Mapping[str, Union[str, int]]:

    the_args = (name, age, score)
    # reveal_type(the_args)
    if check(the_args):
        # reveal_type(the_args)
        return make_setting_of_policy(*the_args)  # runtime type check
    else:
        raise ValueError('The args of "{}" should not be None.'.format(name))  # runtime exception


def make_setting_of_policy(name: str, age: int, score: int) -> Mapping[str, Union[str, int]]:
    return {'name': name, 'age': age, 'score': score}

mypy 结果：

Success: no issues found in 1 source file

取消注释 reveal_type() 调用使 mypy 显示预期的类型缩小：

Q71223933.py:13: note: Revealed type is "Tuple[builtins.str, Union[builtins.int, None], Union[builtins.int, None]]"
Q71223933.py:15: note: Revealed type is "Tuple[builtins.str, builtins.int, builtins.int]"

但是您的问题中有一个微妙的暗示，即在 def check(the_first_name,*arg) 函数中对 *args 进行类型缩小。使用 Python 3.11 可以使用 PEP 646 -- Variadic Generics, but since your Optionals where written with Python 3.9 syntax I'm assuming that's beyond the scope of this question, see PEP 604 -- Allow writing union types as X | Y.

mypy 不理解为 None 值引发异常

mypy doesn't undestand raise Exception for None value

python

narrowing

python-3.x

mypy

python-typing