Fix false positive no-member in except * handler

[matusvalo]

Type of Changes

	Type
✓	🐛 Bug fix
	✨ New feature
	🔨 Refactoring
	📜 Docs

Description

This PR allows astroid to understand that except * produces ExceptionGroup instead of specified exception.

Note

This PR is mainly interested in ExceptionGroup.exceptions attribute. Other attributes/methods are left for future improvements.

Refs pylint-dev/pylint#9056

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.26%. Comparing base (9256b31) to head (0f362f5).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2786   +/-   ##
=======================================
  Coverage   93.26%   93.26%           
=======================================
  Files          91       91           
  Lines       11010    11017    +7     
=======================================
+ Hits        10268    10275    +7     
  Misses        742      742           

Flag	Coverage Δ
linux	93.12% <100.00%> (+<0.01%)	⬆️
pypy	93.26% <100.00%> (+<0.01%)	⬆️
windows	93.24% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
astroid/protocols.py	90.04% <100.00%> (+0.18%)	⬆️

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

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

[DanielNoord]

I'm wondering if we can just get this from builtins and infer it from there? Why do we need to create the ClassDef ourselves?

[matusvalo]

I will try that.

[matusvalo]

OK ExceptionGroup constructed and infered from builtins module...

[Pierre-Sassoulas]

Thank you for the merge request. Are pylint and Ruff disagreeing about next ? It seems making pylint happy makes Ruff sad. Maybe there's another way to make pylint happy, I'm on mobile so I was not able to check the pipeline for the commit where pylint is sad.

[matusvalo]

Thank you for the merge request. Are pylint and Ruff disagreeing about next ? It seems making pylint happy makes Ruff sad. Maybe there's another way to make pylint happy, I'm on mobile so I was not able to check the pipeline for the commit where pylint is sad.

The issue is following code:

        eg = list(
            node_classes.unpack_infer(
                extract_node(
                    """
from builtins import ExceptionGroup
ExceptionGroup
"""
                )
            )
        )[0]

ruff does not like list with zero index, it prefers next() (RUF015). On the other hand pylint does not like next() version:

        eg = next(
            node_classes.unpack_infer(
                extract_node(
                    """
from builtins import ExceptionGroup
ExceptionGroup
"""
                )
            )
        )

without handled StopIteration exception (stop-iteration-return).

If I understand code correctly, it is guaranteed that at least one item is returned by unpack_infer() hence I propose to use next() version of the code and disable stop-iteration-return. Let me know if you agree.

EDIT: Or we can add assertion:

try:
    ...
except StopIteration:
   assert False, "Should not happen"

[Pierre-Sassoulas]

Let's noqa pylint