Fix class checker crashes on x.__class__ targets outside plain assignments

[codeAnqiang-ma]

Type of Changes

	Type
✓	🐛 Bug fix

Description

Closes #11267

visit_assignattr fires for every AssignAttr, but the two __class__ code paths assumed the target's parent is a plain assignment carrying a literal value, so legal constructs like for foo.__class__ in [Foo]: aborted the whole module with a fatal astroid-error (F0002):

• _check_invalid_class_object read node.parent.value, which does not exist for a for-loop, with, or comprehension target and is None for a bare annotation; its tuple branch indexed .value.elts, which crashes when the right-hand side is a call, a nested tuple, or a too-short literal (AttributeError / IndexError).
• _check_in_slots read node.parent.value the same way for slotted classes, and _has_same_layout_slots called next(assigned_value.infer()) without handling astroid.InferenceError, so an unresolvable value such as an undefined name also crashed.

The fix skips the checks when there is no single assigned value to inspect (the same approach as #11173's non-name loop target fix), and gives an unresolvable value in _has_same_layout_slots the same answer as any other non-class value, matching the existing behavior for Uninferable (assigning-non-slot is still emitted — covered by the existing ClassReassingingInvalidLayoutClass expectation and a new test).

Existing behavior is preserved: the #7467 tuple-unpacking cases keep their messages, including starred unpacking like self.__class__, *rest = ... (the bound check only skips genuinely unbalanced unpacking).

Test evidence

Both extended functional tests fail before the fix and pass after it:

$ python -m pytest tests/test_functional.py -k "invalid_class_object or assigning_non_slot" -q
# before the fix: 2 failed (AttributeError: 'For' object has no attribute 'value', ...), 1 passed
# after the fix:  3 passed

Full functional suite (python -m pytest tests/test_functional.py -q -n auto): 874 passed, 29 skipped, 14 failed — the failing set is byte-for-byte identical on current main in the same environment (macOS, Python 3.13.3; e.g. wrong_import_order, used_before_assignment_py310/311), i.e. pre-existing and unrelated. mypy and a self-run of pylint on the changed file are clean.

This PR was prepared with AI assistance as part of a code correctness review; every crash variant was reproduced locally and the fix and tests were reviewed and run by the author.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.40%. Comparing base (4be9585) to head (ffba5c1).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #11268      +/-   ##
==========================================
+ Coverage   96.36%   96.40%   +0.04%     
==========================================
  Files         178      178              
  Lines       19953    20042      +89     
==========================================
+ Hits        19228    19322      +94     
+ Misses        725      720       -5     

Files with missing lines	Coverage Δ
pylint/checkers/classes/class_checker.py	94.80% <100.00%> (+0.72%)	⬆️

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Pierre-Sassoulas]

Suggested change

	Fix a crash in the class checker when ``x.__class__`` is an assignment target
	outside a plain assignment, such as a for-loop, ``with``, or comprehension
	target, a bare annotation, a list target, or a tuple target unpacked from a
	non-literal or too-short right-hand side. This affected both
	``invalid-class-object`` and ``assigning-non-slot``, which also crashed when the
	assigned value could not be resolved by astroid, or when ``__class__`` was
	assigned a class that does not define ``__slots__``.
	Closes #11267
	Fix a crash in :ref:`invalid-class-object` and :ref:`assigning-non-slot` when ``__class__`` is assigned outside a simple assignment (e.g. ``for obj.__class__ in classes:``).
	Closes #11267

[github-actions]

🤖 According to the primer, this change has no effect on the checked open source code. 🤖🎉

This comment was generated for commit ffba5c1