Use loop variables or variables defined in a loop outside the loop

[ShaiAvr]

Summary

I recently got bitten by the accidental use of a loop variable outside the loop. For example, this code works since the loop's variable name still exists after the loop is over and holds the value from the last iteration:

for x in container:
    print(x)

print(x)  # I can still use the value from the last iteration, which is most likely a mistake.

Most of the time, code like that indicates an error or a mistake. If container is empty, for example, the loop won't run, and we'd get a NameError when trying to use the variable x. Even if we can guarantee that the loop runs and x will be valid after the loop, it's likely a logical error to use it. The same applies to other variables created inside a loop:

for x in container:
    y = manipulate(x)
    print(x, y)

print(y)  # Likely a mistake

I tried to enable the ALL rule of ruff, but it didn't catch these mistakes. I think a rule should be added to warn about such code. It could be legitimate to use this technique to capture the last value of a loop, but in most cases, such code indicates a mistake or logical error, and it should be avoided. For the cases where this behaviour is used intentionally, the user can always ignore the rule.

[ntBre]

I think this would be a nice rule, I've been bitten by this before too. There's previous work in #11769, but there are some unresolved questions around balancing false positives and negatives. This kind of rule is pretty tricky to write in Ruff.

[MichaReiser]

ty can catch this kind of error if you enable possibly-unbound-reference (https://play.ty.dev/671c6094-729f-4d97-941e-f6c52ee3d236).

But it doesn't cover:

Even if we can guarantee that the loop runs and x will be valid after the loop, it's likely a logical error to use it.

which may justify a dedicated rule