Implement Windows file path handling in URL parsing (WHATWG URL #874)

[mertcanaltin]

The WHATWG URL specification is adding Windows file path
handling support. Windows drive letter patterns should
automatically convert to file:/// URLs when parsing.

Expected behavior

new URL("C:\\folder\\file.txt")
// Should output: file:///C:/folder/file.txt

Related links

• Spec PR: feat: handle Windows file paths with backslashes in URL parsing whatwg/url#874
• Original issue: Windows file paths in URL parsing whatwg/url#873
• Tests:
  accepting windows backslash paths web-platform-tests/wpt#53459
• Chromium: https://issues.chromium.org/issues/437449112
• Gecko: https://webcompat.com/issues/170937
• WebKit: https://bugs.webkit.org/show_bug.cgi?id=297169

[raloy1102]

I’m working on this feature and noticed that parse_url currently takes base_href: Option<&Url>.
However, because Rust’s Url parser canonicalizes file URLs and removes the host, the host info is not available on the Url object, which makes WHATWG‑compliant handling difficult (e.g., UNC hosts).
Would it be acceptable to change base_href to take a &str so the original host can be extracted before parse and restored later? Or is there a preferred alternative?

[crowlKats]

@raloy1102 sounds reasonable