aio-libs / yarl / 36471421231

98%

push

github

Compare URLs with empty components the way they print (#1934)

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

## What do these changes do?

Two review findings on #1929 were marked resolved but not fixed:

- `==` and `hash()` ignored an empty authority for every special scheme,
so in RFC mode `URL("http:/x", mode="rfc")` and `URL("http:///x",
mode="rfc")` printed differently but compared equal and collapsed into
one dictionary key. The comparison now follows what `str()` writes: the
empty authority is ignored only where it is always printed, i.e. a
special scheme in WHATWG mode.
- `URL.build(path="//a", encoded=True)` prints `/.//a`, but
`URL("/.//a", encoded=True)` kept the serialization-only `/.` in its
path. The encoded parser now drops it the same way the unencoded one
does. So that a literal `/.` segment in that place survives too, `str()`
adds the `/.` to any authority-less path that is `//...` behind zero or
more `/.` segments and the parsers drop exactly one: `/.//a` is written
`/././/a`.

The CodeQL "redundant comparison" alerts on `test_ordering` are gone
too: each ordering operator is now checked on its own pair of URLs.

## Are there changes in behavior for the user?

Yes, for URLs without an authority in RFC mode only: `file:/p` and
`file:///p` (and `http:/x` vs `http:///x`) no longer compare equal
there, because they print differently. In the default WHATWG mode both
forms still print `file:///p` and stay equal. RFC mode is not released
yet, so this is a bugfix rather than a breaking change.

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

No, it narrows one property and adds three lines to the encoded parser.

## Related issue number

Follow-up to #1929.

## Checklist

- [x] I think the code is well written
- [x] Unit tests for the changes exist
- [x] Documentation reflects the changes
- [ ] If you provide code modification, please add yourself to
`CONTRIBUTORS.txt`
- [x] Add a new news fragment into the `CHANGES/` folde... (continued)

96 of 192 branches covered (50.0%)

Branch coverage included in aggregate %.

76 of 76 new or added lines in 3 files covered. (100.0%)

8749 of 8866 relevant lines covered (98.68%)

0.99 hits per line