URLParser should parse URLs including authority and a backslash after the host

• Component: Web Template Framework
• Status: RESOLVED FIXED

[dave caruso]

WTF:URLParser is unable to parse the text `
`, specifically due to the backslash.

In Safari/JSC you can run `new URL` with this string to see an error. Chrome/firefox/node.js all allow parsing this url.

The specification defines one case in host state (https://url.spec.whatwg.org/#host-state) where backslash is allowed when using special urls.

WTF::URLParser::parse does not seem to handle that. I suspect this is a one line fix around line 1486, but I have not tested the patch.

- if (*c == '/' || *c == '?' || *c == '#') {
+ if (*c == '/' || (m_urlIsSpecial && *c == '\\') || *c == '?' || *c == '#') {

[Karl Dubost]

Dave, thanks for the report.
Is there a missing string in the bug report?

> the text `
> `,

[dave caruso]

Yes it seems the url may have been auto-redacted due to it containing authorization.

"http:// a : b @ c \ hello"

i've rewritten the url with spaces. it is any "special" (protocol http/https/file and some others) but there is a backslash instead of a forward slash after the host and starting the path name. the url can also have no path but end in a backslash.

[Karl Dubost]

https://searchfox.org/wubkat/rev/c20f54808bd9cd41e49a640e8c20923340e4d14b/Source/WTF/wtf/URLParser.cpp#1480-1500

[Alex Christensen]

I think you're right, Dave.  Would you like to make a PR or two, or would you like me to?  Either way it should have a test with a special scheme and a test with a non-special scheme.  Our tests for this are in LayoutTests/imported/w3c/web-platform-tests/url/resources/urltestdata.json and ideally it would have a PR to the WPT repo as well.

[dave caruso]

I can work on a PR this weekend. Thanks for pointing me to the tests.

[Alex Christensen]

Sounds good!  Let's just make a WebKit PR for now.  WPT will be updated with this in due time.

This command will help in updating the test results:
Tools/Scripts/run-webkit-tests imported/w3c/web-platform-tests/url --reset --no-build

[Radar WebKit Bug Importer]

<rdar://problem/119219832>

[Anne van Kesteren]

Pull request: https://github.com/WebKit/WebKit/pull/24657

[EWS]

Committed 274915@main (953476de5074): <https://commits.webkit.org/274915@main>

Reviewed commits have been landed. Closing PR #24657 and removing active labels.