[perflint] Simplify finding the loop target in PERF401

[w0nder1ng]

Fixes #15012.

def f():
    # panics when the code can't find the loop variable
    values = [1, 2, 3]
    result = []
    for i in values:
        result.append(i + 1)
    del i

I'm not sure exactly why this test case panics, but I suspect the del i removes the binding from the semantic model's symbols.

I changed the code to search for the correct binding by directly iterating through the bindings. Since we know exactly which binding we want, this should find the loop variable without any complications.

[github-actions]

ruff-ecosystem results

Linter (stable)

ℹ️ ecosystem check detected linter changes. (+0 -1 violations, +0 -0 fixes in 2 projects; 53 projects unchanged)

pandas-dev/pandas (+0 -0 violations, +0 -0 fixes)

python-poetry/poetry (+0 -1 violations, +0 -0 fixes)

- tests/console/commands/test_show.py:127:5: B018 Found useless expression. Either assign it to a variable or remove it.

Changes by rule (1 rules affected)

code	total	+ violation	- violation	+ fix	- fix
B018	1	0	1	0	0

Linter (preview)

✅ ecosystem check detected no linter changes.

[dylwil3]

I'm not sure exactly why this test case panics, but I suspect the del i removes the binding from the semantic model's symbols.

Yep. This is the cause of a few cases of strange behavior... When the semantic model is called "as it's built" then this behavior is correct (because at the program point of interest, the delete has not yet ocurred), but it can be counter-intuitive when called (like in the current situation) on deferred scopes/for-loops/etc.

[dhruvmanila]

Is this unwrap safe? If so, we should use expect but I'd possibly just do an early return if this is None.

[MichaReiser]

It should be fine. The rule only applies to for-loop targets that are names; a binding must exist.