Opened 10 years ago
Last modified 5 days ago
#39535 new defect (bug)
Canonical redirects disallow tag named comments
| Reported by: | dwc | Owned by: | |
|---|---|---|---|
| Priority: | normal | Milestone: | Future Release |
| Component: | Canonical | Version: | |
| Severity: | normal | Keywords: | has-patch has-unit-tests changes-requested |
| Cc: | Focuses: |
Description
The redirect_canonical function includes a regular expression replacement which effectively disallows tags named "comments" to have an RSS feed:
https://github.com/WordPress/WordPress/blob/master/wp-includes/canonical.php#L285
For example:
/blog1/tag/comments/feed/
This is redirected to:
/blog1/tag/feed/
Could / should the regular expression on line 285 could pay attention to the first placeholder it matched (comments/?) or check whether the page is a tag request?
Attachments (2)
Change History (22)
#2
@
10 years ago
@dd32 @markjaquith I looked a bit at the source code and this seems like it has been around for a while. Sorry to pull you in on this one, not sure who would be assigned :)
Here is the commit on Git:
https://github.com/WordPress/WordPress/commit/ecffc4649f4cb90b43fe0b38ab0f4b18967cb3fd
I was able to reproduce the bug myself, real easy to do so. Just create a tag named "comments" and then try to load the rss feed for the tag: http://URL/tag/comments/feed/
Edit: tagged the canonical gods
This ticket was mentioned in Slack in #core by asalce. View the logs.
10 years ago
#6
@
10 years ago
@swissspidy I submitted the unit-test, The best way I could test this out was to test if the query vars were set correctly after a go_to("/tag/comments/feed/") the query vars should have a "feed" and "tag" should equal "comments" , I also added some other tests to make sure tests were running correctly.
@dwc perfect! :)
This ticket was mentioned in Slack in #core by asalce. View the logs.
10 years ago
This ticket was mentioned in Slack in #core by asalce. View the logs.
10 years ago
#10
in reply to: ↑ 9
@
10 years ago
Replying to SergeyBiryukov:
I ran into this bug with another site, Any chance this could be moved into 4.7.x ?
This ticket was mentioned in Slack in #core by obenland. View the logs.
10 years ago
This ticket was mentioned in Slack in #core by asalce. View the logs.
9 years ago
#15
@
8 years ago
Can this one also be put up for review?
https://make.wordpress.org/core/2019/02/20/reverting-the-bulk-ticket-closing/ @obenland @JeffPaul
#16
@
8 years ago
@asalce I don't think this ticket relates to the bulk ticket closing, but if you're asking about getting this milestoned then your best bet is to check with the Canonical component maintainers (see https://make.wordpress.org/core/components/permalinks/canonical/).
#17
@
8 years ago
@JeffPaul Thanks!
@markjaquith this ticket has been in my account for a bit. I supplied a patch and unit testing awhile ago. Let me know what else is needed to get this into a release. Thanks!
- Alberto S.
This ticket was mentioned in PR #8981 on WordPress/wordpress-develop by @SirLouen.
17 months ago
#18
- Keywords has-unit-tests added; needs-unit-tests removed
This patch is not working. Unit Tests not passing. Needs more work.
Trac ticket: https://core.trac.wordpress.org/ticket/39535
#19
@
17 months ago
- Keywords changes-requested added; needs-testing removed
Patch has been refreshed, but still unit tests are not passing. Needs more work.
cc @asalce
This ticket was mentioned in PR #15982 on WordPress/wordpress-develop by @irozum.
5 days ago
#20
redirect_canonical() strips an optional comments/ segment before any feed ending, so /tag/comments/feed/ is redirected to /tag/feed/ (and /category/comments/feed/ to /category/feed/). A term with the slug comments could therefore never serve its feed.
Only a comments feed has a comments/ segment to strip, and those always set withcomments. The add-back step already only re-adds comments/ when withcomments is set, so this makes the strip step match it: the comments/ segment is only stripped when withcomments is set.
The site-wide comments feed (/comments/feed/, ?feed=comments-rss2) and per-post comment feeds are unaffected and covered by the existing canonical tests. New tests cover tag and category feeds for a term named comments.
Earlier work on this ticket, including the 2017 patch and the refresh in #8981, was a starting point for the investigation. #8981 notes its tests did not pass, so this takes a different approach that does not depend on the host or permalink structure.
A related issue is not addressed here: a term named feed, rss or atom (e.g. /tag/feed/feed/) is still over-stripped by the repeated stripping in the while loop. That could be a follow-up.
Use of AI Tools
AI assistance: Yes
Tool(s): Claude Code
Model(s): Claude Sonnet 5.5
Used for: Implementation, tests, and PR description.
---
This Pull Request is for code review only. Please keep all other discussion in the Trac ticket. Do not merge this Pull Request. See GitHub Pull Requests for Code Review in the Core Handbook for more details.
🤖 Generated with Claude Code
@dwc I attached a patch which should work. The safest way was to check against the generated site-wide comments feed. Could you take look and let me know if that solves your issue?