B023 False positive when there is a list comprehension with the same variable name in a loop.

[vivodi]

Description

minimal code snippet:

for _ in range(3):
    [x for x in []]
    def func():
        lambda x: x

command:

ruff check --isolated --select B023 demo.py

log:

demo.py:5:19: B023 Function definition does not bind loop variable `x`
  |
4 |     def func():
5 |         lambda x: x
  |                   ^ B023
  |

Found 1 error.

ruff version: 0.9.2

This is NOT a duplicate of #7847, because x is only used in a list comprehension in the same loop. Ruff should not raise a false positive in this situation.

[orgua]

I got a similar false positive when variable name is not reused, but pandas.apply() is used:

import pandas as pd

data = pd.DataFrame()
data.loc[0, "hex"] = "72756666"
data.loc[1, "hex"] = "75767576"

def modifier(value):
    return chr(int(value, 16))


for _i in range(0, 4):
    data[f"v{_i}"] = data["hex"].apply(
        lambda x: modifier(x[2 * _i : 2 * _i + 2]),
    )

print(data)

I am pretty sure that the code is correct and it returns what I want:

        hex v0 v1 v2 v3
0  72756666  r  u  f  f
1  75767576  u  v  u  v

but the linting command

ruff check --isolated --select B023 demo.py

triggers

demo.py:13:34: B023 Function definition does not bind loop variable `_i`
   |
11 | for _i in range(0, 4):
12 |     data[f"v{_i}"] = data["hex"].apply(
13 |         lambda x: modifier(x[2 * _i : 2 * _i + 2]),
   |                                  ^^ B023
14 |     )
   |

demo.py:13:43: B023 Function definition does not bind loop variable `_i`
   |
11 | for _i in range(0, 4):
12 |     data[f"v{_i}"] = data["hex"].apply(
13 |         lambda x: modifier(x[2 * _i : 2 * _i + 2]),
   |                                           ^^ B023
14 |     )
   |

Found 2 errors.