Fix regression for _is_only_type_assignment

[DanielNoord]

• Add yourself to CONTRIBUTORS if you are a new contributor.
• Write a good description on what the PR does.

Type of Changes

	Type
✓	🐛 Bug fix

Description

Not sure if parent is the right name for the variable, but couldn't come up with a better one.

This closes #5162

[coveralls]

Pull Request Test Coverage Report for Build 1375829491

• 17 of 17 (100.0%) changed or added relevant lines in 1 file are covered.
• 111 unchanged lines in 3 files lost coverage.
• Overall coverage increased (+0.03%) to 93.277%

---

Files with Coverage Reduction	New Missed Lines	%
pylint/checkers/typecheck.py	25	95.01%
pylint/checkers/variables.py	35	95.53%
pylint/checkers/classes.py	51	94.57%

Totals
Change from base Build 1352191945:	0.03%
Covered Lines:	13666
Relevant Lines:	14651

---

💛 - Coveralls

[Pierre-Sassoulas]

Maybe 'englobing_scope' ?

[cdce8p]

I like the test case you've added with global.
Would you mind adding two more, both with and without error?

• Comprehension inside a function, similar to Crash in VariablesChecker._is_only_type_assignment #5162
• Decorator with inner function, similar to Fix regression for _is_only_type_assignment #5163 (comment)

--
Lastly, I needed some time to come up this that but assignment expressions aren't covered yet. Maybe that would be better in a new PR though.

def func():
    var: int
    if (var := var ** 2):
        print(var)

[DanielNoord]

-- Lastly, I needed some time to come up this that but assignment expressions aren't covered yet. Maybe that would be better in a new PR though.

def func():
    var: int
    if (var := var ** 2):
        print(var)

I have opened #5197 to track this. We need control-flow for this, as we currently can't recognise whether a NamedExpr has been called.
I have added some basic tests for NamedExpr which do work.

Edit: Oops, forgot to separate the assignment expressions tests into a >=3.8 file.

[cdce8p]

Thanks for opening the issue. That's definitely not something we can fix at the moment.

Left some comments. One more thing: I noticed that you used nodes.Statement as type annotation for defstmt. Technically this might be correct, but currently it would be an error as it's only recognized as NodeNG. We need to work on the astroid type annotations some more to fix that.

[cdce8p]

Suggested change

	def _is_only_type_assignment(
	node: nodes.Name, defstmt: Union[nodes.Module, nodes.Statement]
	) -> bool:
	def _is_only_type_assignment(node: nodes.Name, defstmt: nodes.NodeNG) -> bool:

I would suggest we use NodeNG for now. Can always update that later. That way we can merge this sooner.
Haven't looked at pylint-dev/astroid#1217 yet, but I feel like this should never be Module in the first place.

https://docs.python.org/3.10/library/ast.html

[DanielNoord]

Should I then add a TODO that links to the astroid issue? I fear we might forget about this in the future if we don't make some comment about it now.

[cdce8p]

I guess that would work 🤷🏻‍♂️

[cdce8p]

Looks good! Thanks, @DanielNoord 🚀