add type annotations

[chdsbd]

I think it may be helpful to add type annotations to this package.

[nicoddemus]

Thanks for the suggestion @chdsbd. I think this would be best when pytest-mock supports Python 3 only so we can use native type annotations.

[JakeSummers]

There are some reasonable workarounds to this problem.

Right now I have a test like this:

def test_car(mocker: MockFixture):
    mocker.patch.object(Engine, "run")
    car = Car()
    car.start()
    Engine.run.assert_any_call(... )

When I run it through mypy, I get the following error:

test_car.py:30: error: "Callable[[Engine, ...], ...]" has no attribute "assert_any_call"

Workaround #1 - Ignore the error

def test_car(mocker: MockFixture):
    mocker.patch.object(Engine, "run")
    car = Car()
    car.start()
    Engine.run.assert_any_call(... ) # ignore: type

Workaround #2 - Use the result of the patch command.

def test_car(mocker: MockFixture):
    engine_mock:MagicMock = mocker.patch.object(Engine, "run")
    car = Car()
    car.start()
    engine_mock.assert_any_call(... ) # ignore: type

[srittau]

FWIW, I have just opened python/typeshed#4275. This will add stubs to typeshed for now. Of course, I'd be glad to see the stubs in pytest-mock directly some day, either via separate stub files or inline.

[bluetech]

@srittau What gets priority when a package has its own stubs, and is also defined in typeshed?

[JelleZijlstra]

@bluetech stub packages get priority, see https://www.python.org/dev/peps/pep-0561/#type-checker-module-resolution-order.

[bluetech]

Thanks @JelleZijlstra. So it should be fine if we ever decide to ship type annotations inline.

[srittau]

As soon as that happens, we should remove the stubs from typeshed anyway, to avoid confusion and reduce typeshed's maintenance burden.

[jtpavlock]

Pytest 6.0 now supports type annotations and pytest-mock is now python3.5+ 👀

[nicoddemus]

In #194 we now run mypy without errors on pytest-mock itself (thanks to @staticdev), so at least that step is done.

The next step would probably be to type annotate the public API and export a py.typed file along the package.