Incorrect result for else/elif about cognitive_complexity HOT 4 CLOSED

Thank you for your kind words and for the report.
It really seems like a bug.
Problem is I don't have time to fix it right now, so it would be great if you submit the pull request.

from cognitive_complexity.

Apologies for the delay; I've been pulled to other projects the last few weeks (both personal and professional). Finally got a chance to work on cognitive complexity again this weekend, and found the cause.

First, ast.iter_child_nodes() just yields the members of orelse without indicating the else itself. Second, an elif was incrementing the nesting level because it looked no different than a nested if. This means that a simple if..else is undercounted by 1, and a simple if..elif..else is correct but for the wrong reasons (+2 to the elif when it should be +1 to the elif and +1 to the else).

Since for and while have optional else clauses, I've attempted to account for those as well.

I'll create the PR momentarily.

@Melevir, I also noticed two other cases where complexity seems to be overcounted. Since the fix involves changes in the same areas, I can add these to the PR I'm working on, or write separate issues and PRs if you prefer.

• I think ast.BoolOp should not receive a nesting increment (test_real_function() should have a cognitive complexity of 9)
• I think break and continue should not receive any increment (unlike C and Java there is no goto and no break/continue to a label, and no increment is given in the addVersion example in the Cognitive Complexity white paper).

from cognitive_complexity.

Merged. Thanks a lot!
Please, create sepatate issues for other problems. It seems like we should fix them and make new major release since thouse changes are not backward compatible.

from cognitive_complexity.

Happy to help! Thanks for the quick review and merge. Will create additional issues now.

from cognitive_complexity.