PLC1901 false positive in __bool__ methods

[trag1c]

class X:
    val: str

    def __bool__(self) -> bool:
        return self.val != ""

$ ruff check test.py --select=PLC1901
test.py:5:28: PLC1901 `self.val != ""` can be simplified to `self.val` as an empty string is falsey
Found 1 error.

Trying to correct this to self.val will lead to an error:

TypeError: __bool__ should return bool, returned str

[MichaReiser]

Yeah, The "safe" fix for this rule should probably be bool(self.val). But even that is not entirely safe, because it now no longer reads self.val, which could have side effects.

[trag1c]

In one case I specifically need the comparison :)

[charliermarsh]

I dislike this rule, too many false positives.

[karolinepauls]

This rule is fundamentally incorrect for assertions or assignments, since None != "" - unless it can infer that the variable is a string.

a = 0

def test_sth() -> None:
    assert a == ""

Command: ruff --isolated --select PLC1901 a.py

Output:

a.py:4:17: PLC1901 `a == ""` can be simplified to `not a` as an empty string is falsey
Found 1 error.

$ ruff --version
ruff 0.0.274

[charliermarsh]

Yes, Pylint's implementation has the same problem:

❯ pylint --load-plugins=pylint.extensions.emptystring foo.py
************* Module foo
foo.py:1:0: C0114: Missing module docstring (missing-module-docstring)
foo.py:1:0: C0104: Disallowed name "foo" (disallowed-name)
foo.py:1:0: C0103: Constant name "a" doesn't conform to UPPER_CASE naming style (invalid-name)
foo.py:3:0: C0116: Missing function or method docstring (missing-function-docstring)
foo.py:4:11: C1901: "a == ''" can be simplified to "not a" as an empty string is falsey (compare-to-empty-string)

I'm going to move this rule to the nursery so that it's explicitly opt-in.

[adifelice-godaddy]

Looks like the rule was renamed/revisited:
https://pylint.pycqa.org/en/latest/user_guide/messages/convention/use-implicit-booleaness-not-comparison-to-string.html

[bersbersbers]

Here's another example where this rule creates false positives with numpy:

import numpy as np

array = np.array(("X", "", "X"), dtype=str)
indices = array == ""
print(indices)

print(not array)

At least in this case, not array raises a ValueError, so incorrect fixes become obvious (at least at runtime).