Fixes duplicate-code check with ignore-imports

[theirix]

Type of Changes

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

Description

Improves detection of imports under conditionals. Limited checks by constant conditions such as True/typing.TYPE_CHECKING.

Closes #8914

[codecov]

Codecov Report

Merging #9147 (c5cbe23) into main (0796dfa) will increase coverage by 0.00%.
The diff coverage is 100.00%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #9147   +/-   ##
=======================================
  Coverage   95.76%   95.76%           
=======================================
  Files         173      173           
  Lines       18680    18681    +1     
=======================================
+ Hits        17889    17890    +1     
  Misses        791      791           

Files	Coverage Δ
pylint/checkers/similar.py	96.29% <100.00%> (+<0.01%)	⬆️

[jacobtylerwalls]

Thanks for the patch!

Inference seems like overkill here, since we just need to traverse the ast. Something like this passes the unit test you've contributed. It causes a failure in test_no_hide_code_with_imports, but that puzzles me as that expected result looks incorrect (we should remove the --ignore-imports flag from that case, no?

[jacobtylerwalls]

Thanks for the updates; very helpful. Requesting one last test case to pass and then this should be ready to merge!

[github-actions]

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

This comment was generated for commit c5cbe23

[jacobtylerwalls]

Thanks for the contribution!

[theirix]

Thanks for the contribution!

Sure, thank you for the extensive review and suggestions.

[Pierre-Sassoulas]

We're not checking duplicate-code in the primers, might be beneficial to check locally or exceptionally/temporarily enable it in the CI in this case ? (wdyt @jacobtylerwalls ?)

[jacobtylerwalls]

Yeah, that's worth a try.

[jacobtylerwalls]

I ran the primer locally, minus pandas and home-assistant. Generated a lot of noisy, trivial output. Too much to paste here. Next time I'll do this on CI on a fork so we get a more readable output.

I did see a handful of wanted changes like this:

No longer emitted: (astroid)
{'type': 'refactor', 'module': 'astroid.brain.brain_hashlib', 'obj': '', 'line': 1, 'column': 0, 'endLine': None, 'endColumn': None, 'path': 'tests/.pylint_primer_tests/pylint-dev/astroid/astroid/brain/brain_hashlib.py', 'symbol': 'duplicate-code', 'message': 'Similar lines in 2 files\n==astroid.nodes.node_classes:[56:63]\n==astroid.nodes.node_ng:[40:51]\nif sys.version_info >= (3, 11):\n from typing import Self\nelse:\n from typing_extensions import Self\n\nif TYPE_CHECKING:\n from astroid import nodes', 'message-id': 'R0801'}

[jacobtylerwalls]

(Just clarifying for folks who might not be aware of the flakiness in the primer checks: the "noisy, trivial output" I referred to is probably from some unrelated mutability of cached objects in astroid, not from anything wrong in this PR.)

[Pierre-Sassoulas]

Thank you for checking Jacob !