[codenamenam]

(Explain how this PR changes mypy.)

Fixes #19660

Allow instantiation of NoneType in type checker

This change fixes the error "Cannot instantiate type 'Type[None]'"
when calling NoneType() or type(None)().

By treating NoneType as a callable that returns None, mypy can now correctly
handle such calls without raising spurious errors.

Also, I added test case testTypeUsingTypeCNoneType covering:

• direct calls to type(None)() and NoneType()
• functions accepting type[None] and type[NoneType] parameters and invoking them

This ensures proper handling of NoneType instantiation and prevents spurious errors.

[sterliakov]

Note that this does not fix #19308 (but fortunately the magic comment only affects the first issue, so that's just a plain reference):

def get_class(x: int) -> type[str | None]:
    if x > 0:
        return str
    else:
        return type(None)

                                                                                                     
x = get_class(1)
if x is type(None):
    print("hi")
else:
    reveal_type(x)  # N: Revealed type is "type[builtins.str] | type[None]"

[codenamenam]

Thanks for the feedback, @sterliakov! It seems this PR doesn't fix #19308.
I've updated the PR description to only reference #19660.

[wyattscarpenter]

Very nice!

Along with what's currently there, which ensures NoneType is callable, you should probably assert in your test case that the value produced by NoneType() and friends is None — um, or that its type is type[None], rather (or NoneType). Maybe you can use assert_type for this. Or possibly reveal_type plus expected notices, or some trick about narrowing, if assert_type doesn't work out for some reason.

[codenamenam]

@wyattscarpenter Thanks for the review! I've modified the test case to use assert_type to verify the return type, as you suggested.

[A5rocks]

Thanks. Just one nit on the test you added.

[A5rocks]

Please use standard test syntax: remove the [out] at the bottom and replace this with a # E: whatever the error was. You can see some more examples around this.

[github-actions]

Diff from mypy_primer, showing the effect of this PR on open source code:

sphinx (https://github.com/sphinx-doc/sphinx)
+ sphinx/builders/_epub_base.py:34:1: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports
- sphinx/transforms/post_transforms/code.py:9:1: note: Hint: "python3 -m pip install types-Pygments"
- sphinx/transforms/post_transforms/code.py:9:1: note: (or run "mypy --install-types" to install all missing stub packages)
- sphinx/transforms/post_transforms/code.py:9:1: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports
+ sphinx/cmd/build.py:460:1: note: Hint: "python3 -m pip install types-Pygments"
- sphinx/_cli/util/errors.py:102:1: note: Hint: "python3 -m pip install types-Pygments"
- sphinx/_cli/util/errors.py:102:1: note: (or run "mypy --install-types" to install all missing stub packages)
- sphinx/_cli/util/errors.py:102:1: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports
+ sphinx/application.py:48:1: note: Hint: "python3 -m pip install types-Pygments"
+ sphinx/application.py:48:1: note: (or run "mypy --install-types" to install all missing stub packages)

scipy-stubs (https://github.com/scipy/scipy-stubs)
+ tests/misc/test_submodules_empty.pyi:3: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports