aio-libs / multidict / 37735945154

86%

Build Type

push

github

Committed by web-flow

Commit Message

Ignore a str subclass's own lower() in a CIMultiDict identity (#1746)

<!-- Thank you for your contribution! -->

## What do these changes do?

Computes the case-insensitive identity of a `str` subclass key as
`str.__str__(key).lower()` instead of calling the subclass's own
`lower()`, in both backends. `IStr_FromUnicode()` in the C API builds an
istr's canonical form with `str.lower()` called unbound for the same
reason. Computing an identity therefore never runs user code, which
removes a source of reentrancy from every lookup and bulk update.

In the C extension the exact copy then takes the existing ASCII fast
path instead of a method call, so a lookup with a subclass key got
faster; exact `str` and `istr` keys are unaffected.

## Are there changes in behavior for the user?

Yes. A `str` subclass that overrides `lower()` no longer changes how
`CIMultiDict` matches it: `CIMultiDict` lowercases the key's string
value with `str.lower()`, the same as for a plain `str`.

## Is it a substantial burden for the maintainers to support this?

No. The tests that relied on an overridden `lower()` to run Python code
mid-operation were dropped or moved to another hook (`__getitem__()` on
a sequence item).

## Related issue number

N/A; follows #1744. Since a key's `lower()` can no longer mutate the
source mid-update, this drops the tests #1744 added for that case and
its pure-Python source version check, which nothing could trigger any
more.

## Checklist

- [x] I think the code is well written
- [x] Unit tests for the changes exist
- [x] Documentation reflects the changes (N/A, the override was never
documented)
- [ ] If you provide code modification, please add yourself to
`CONTRIBUTORS.txt` (N/A)
- [x] Add a new news fragment into the `CHANGES/` folder
- [x] `make doc-spelling` passes and any new technical words are added
to `docs/spelling_wordlist.txt`

<details>
<summary>Agent run details (optional, for reviewers)</summary>

Tests: `pytest -q` on 3.14.7 (40... (continued)

Coverage Stats

793 of 1586 branches covered (50.0%)

Branch coverage included in aggregate %.

30 of 32 new or added lines in 5 files covered. (93.75%)

1 existing line in 1 file now uncovered.

9314 of 10179 relevant lines covered (91.5%)

1.83 hits per line