1225255 - nsStandardURL should use nsNCRFallbackEncoderWrapper instead of raw nsIUnicodeEncoder

Status: RESOLVED FIXED

(Core :: Networking, defect)

Description

[Henri Sivonen (:hsivonen)]

The URL spec requires the "HTML" error handling mode when encoding the query string for parsing:
https://encoding.spec.whatwg.org/#encode

Therefore, nsStandardURL should use nsNCRFallbackEncoderWrapper instead of raw nsIUnicodeEncoder. (nsNCRFallbackEncoderWrapper can't fail for non-OOM reasons, so the UTF-8 fallback should be removed, too.)

Comment 2

[Henri Sivonen (:hsivonen)]

Attachment: Gecko changes

This is a minimal patch to the Gecko code to fix this. (Previously, I tried doing other cleanup around how the encoding info for the URL is kept around, but that caused the scope of the fix to grow too large very quickly. Let's not go there at this time and wait for rust-url to fix things like that.)

The bigger part of the fix here is going to be adjusting all the relevant test case expectations. I'll be away from Bugzilla for a couple of weeks. If someone else wants to pursue the test case adjustments before I get to it, please go ahead.

Comment 3

[Henri Sivonen (:hsivonen)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=781f64a6af5a

Comment 4

[Henri Sivonen (:hsivonen)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=86c4a29bf86e

Comment 5

[Henri Sivonen (:hsivonen)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=147c0e75306f

Comment 6

[Henri Sivonen (:hsivonen)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=3caacd3ae0242df429e61c7d936674fc1d5e3686

Comment 7

[Henri Sivonen (:hsivonen)]

Attachment: Bug 1225255 - Encode URL query string segments to bytes with HTML numeric character references for unmappable characters.

Review commit: https://reviewboard.mozilla.org/r/95756/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/95756/

Comment 8

[Patrick McManus [:mcmanus]]

I really apologize for not rerouting this review earlier - I thought I had and just found it in a bz report assigned to me. I think I forgot to hit the publish review button on mozreview; probly cause I didn't actually review it.

again - apologies.

Comment 9

[Valentin Gosu [:valentin] (he/him)]

Comment on attachment 8814564 [details]
Bug 1225255 - Encode URL query string segments to bytes with HTML numeric character references for unmappable characters.

https://reviewboard.mozilla.org/r/95756/#review100696

Looks good. Thanks!

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/501d5da362c3
Encode URL query string segments to bytes with HTML numeric character references for unmappable characters. r=valentin

Comment 11

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/501d5da362c3