Fix file:// URL roundtrip bugs (#1101, #1102)

[jrey8343]

Summary

This PR fixes two bugs found through OSS-Fuzz that caused file:// URLs to fail roundtrip tests (parse → serialize → parse).

Fixes #1101 - File URLs with hosts and paths containing multiple slashes were losing their host component during roundtrip
Fixes #1102 - set_host("localhost") on file:// URLs didn't normalize localhost to empty host like the parser does

Changes

Bug #1101: Path structure preservation

Problem: When parsing file URLs like file://host//path, the path normalization was stripping all leading slashes, causing the host to be lost on re-parse.

Fix: Modified parse_path() in parser.rs to preserve path structure when a host component is present. Leading slash normalization now only applies to hostless file:// URLs.

Example:

let url1 = Url::parse("file://.host//path").unwrap();
let serialized = url1.to_string();
let url2 = Url::parse(&serialized).unwrap();
assert_eq!(url1, url2); // Now passes

Bug #1102: localhost normalization in set_host()

Problem: The URL parser normalizes "localhost" to empty host for file:// URLs per WHATWG spec, but set_host() wasn't applying the same normalization, causing asymmetric behavior.

Fix: Modified set_host() in lib.rs to normalize "localhost" to None for file:// URLs, matching the parser's behavior.

Example:

let mut url = Url::parse("file:///path").unwrap();
url.set_host(Some("localhost")).unwrap();
let reparsed = Url::parse(&url.to_string()).unwrap();
assert_eq!(url, reparsed); // Now passes

Impact

• ✅ All existing tests pass (14,095 tests)
• ✅ Fuzzer verified with 1M+ iterations, no crashes
• ✅ Improved WHATWG spec compliance
• ✅ Resolved 4 Web Platform Tests that were previously expected failures:
  • file://spider///
  • file://monkey/ with pathname set to \\\\
  • file:///unicorn with pathname set to //\\/
  • file:///unicorn with pathname set to //monkey/..//

Testing

Added comprehensive test suite in url/tests/roundtrip_bugs.rs that reproduces both bugs and verifies the fixes.

WHATWG Spec Compliance

Both fixes align with the WHATWG URL Standard:

• File URL host normalization: https://url.spec.whatwg.org/#file-host-state
• Path normalization: https://url.spec.whatwg.org/#file-slash-state

---

Found while integrating rust-url with OSS-Fuzz for continuous fuzzing.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (main@a66f422). Learn more about missing BASE report.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1103   +/-   ##
=======================================
  Coverage        ?   86.41%           
=======================================
  Files           ?       27           
  Lines           ?     5337           
  Branches        ?        0           
=======================================
  Hits            ?     4612           
  Misses          ?      725           
  Partials        ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Manishearth]

Could you rebase this fix off of the fuzz PR?