python - How do I correctly add type-hints to Mixin classes?

Question

Consider the following example. The example is contrived but illustrates the point in a runnable example:

class MultiplicatorMixin:

    def multiply(self, m: int) -> int:
        return self.value * m


class AdditionMixin:

    def add(self, b: int) -> int:
        return self.value + b


class MyClass(MultiplicatorMixin, AdditionMixin):

    def __init__(self, value: int) -> None:
        self.value = value


instance = MyClass(10)
print(instance.add(2))
print(instance.multiply(2))

When executed this will give the following output:

12
20

The code works.

But running mypy on it, yields the following errors:

example.py:4: error: "MultiplicatorMixin" has no attribute "value"
example.py:10: error: "AdditionMixin" has no attribute "value"

I understand why mypy gives this result. But the mixin classes are never used by themselves. They are always used as additional superclasses.

For context, this is a pattern which has been used in an existing application and I am in the process of adding type-hints. And in this case, the errors are false-positives. I am thinking about rewriting the part using the mixins as I don't particularly like it and the same could probably be done with reorganising the class hierarchy.

But I still would like to know how something like this could be properly hinted.

Answer 1

For reference, mypy recommends to implement mixins through a protocol (documentation here).

It works with mypy >= 750.

from typing import Protocol


class HasValueProtocol(Protocol):
    @property
    def value(self) -> int: ...


class MultiplicationMixin:

    def multiply(self: HasValueProtocol, m: int) -> int:
        return self.value * m


class AdditionMixin:

    def add(self: HasValueProtocol, b: int) -> int:
        return self.value + b


class MyClass(MultiplicationMixin, AdditionMixin):

    def __init__(self, value: int) -> None:
        self.value = value

The Protocol base class is provided in the typing_extensions package for Python 2.7 and 3.4-3.7.