Flag PEP 585 and PEP 604 violations in quoted annotations

[charliermarsh]

Summary

At present, we don't flag PEP 585 and PEP 604 errors in quoted annotations. This mirrors pyupgrade, which doesn't fix quoted annotations. However, in our case, we should at least be flagging them as errors.

As a reminder, PEP 585 introduced standard library generics, so you can do list[int] instead of from typing import List; List[int]. PEP 604 added new syntax for unions, so you can do str | int instead of from typing import Union; Union[str, int].

Note that we can fix simple quoted annotations, but... we need to take care in certain cases.

For example: a quoted annotation can contain an implicit string concatenation, like x: "Li" "st[int] = []. This is also valid:

x: """
List[str]
""" = []

Fixing those cases will be challenging, so for now, it's safer just to avoid fixing them at all (though we should add autofix for "simple" cases soon).

Closes #3555.

[github-actions]

PR Check Results

Ecosystem

✅ ecosystem check detected no changes.

Benchmark

Linux

group                                      main                                   pr
-----                                      ----                                   --
linter/all-rules/large/dataset.py          1.00     13.4±0.08ms     3.0 MB/sec    1.00     13.4±0.08ms     3.0 MB/sec
linter/all-rules/numpy/ctypeslib.py        1.00      3.5±0.01ms     4.7 MB/sec    1.00      3.5±0.01ms     4.7 MB/sec
linter/all-rules/numpy/globals.py          1.00    479.4±0.94µs     6.2 MB/sec    1.01    482.8±1.02µs     6.1 MB/sec
linter/all-rules/pydantic/types.py         1.00      5.9±0.02ms     4.3 MB/sec    1.00      5.9±0.02ms     4.3 MB/sec
linter/default-rules/large/dataset.py      1.00      7.1±0.02ms     5.7 MB/sec    1.00      7.1±0.02ms     5.7 MB/sec
linter/default-rules/numpy/ctypeslib.py    1.00   1604.0±1.56µs    10.4 MB/sec    1.00   1604.8±1.38µs    10.4 MB/sec
linter/default-rules/numpy/globals.py      1.00    180.0±0.35µs    16.4 MB/sec    1.00    180.2±0.42µs    16.4 MB/sec
linter/default-rules/pydantic/types.py     1.00      3.4±0.01ms     7.6 MB/sec    1.00      3.4±0.01ms     7.6 MB/sec

Windows

group                                      main                                   pr
-----                                      ----                                   --
linter/all-rules/large/dataset.py          1.00     14.8±0.19ms     2.7 MB/sec    1.01     15.0±0.21ms     2.7 MB/sec
linter/all-rules/numpy/ctypeslib.py        1.00      4.1±0.06ms     4.1 MB/sec    1.01      4.1±0.07ms     4.0 MB/sec
linter/all-rules/numpy/globals.py          1.00   453.0±24.56µs     6.5 MB/sec    1.01   458.5±33.09µs     6.4 MB/sec
linter/all-rules/pydantic/types.py         1.00      6.7±0.09ms     3.8 MB/sec    1.01      6.7±0.10ms     3.8 MB/sec
linter/default-rules/large/dataset.py      1.00      8.1±0.09ms     5.0 MB/sec    1.01      8.2±0.07ms     5.0 MB/sec
linter/default-rules/numpy/ctypeslib.py    1.00  1746.5±21.26µs     9.5 MB/sec    1.01  1768.3±32.04µs     9.4 MB/sec
linter/default-rules/numpy/globals.py      1.00    180.6±4.35µs    16.3 MB/sec    1.02    183.6±7.80µs    16.1 MB/sec
linter/default-rules/pydantic/types.py     1.00      3.8±0.06ms     6.7 MB/sec    1.02      3.9±0.06ms     6.6 MB/sec

[MichaReiser]

My understanding is that our guideline is to use rule names that work well with allow. Is this an internal-only name and what's the motivation of using the PEP names?

[konstin]

python features often lack a name beyond the PEP number. e.g. https://docs.python.org/3/library/typing.html says:

PEP 604: Allow writing union types as X | Y
Introducing types.UnionType and the ability to use the binary-or operator | to signify a union of types

we could invent our own name such as NonPipeUnion.

[MichaReiser]

NIt: Unrelated to this change: A Availability::None would remove the need for Option<AutofixKind>. It would then simply be AUTOFX: AutofixKind = AutofixKind::NONE (we could even inline Availability into AutofixKind

[charliermarsh]

Yeah this would be a good change, I think.