Page MenuHomePhabricator
Create Task
Maniphest T78090

Fix base href and red link rewriter to handle new Parsoid relative links
Closed, ResolvedPublic
Actions

Description

We need to update our code for T72743: Parsoid base URL should be independent of page. Symptom is that page points to the wrong URL. E.g. first version of header at https://ca.wikipedia.org/wiki/Viquip%C3%A8dia:La_taverna/Tecnicismes points to https://ca.wikipedia.org/wiki/Viquip%C3%A8dia:La_taverna/Viquip%C3%A8dia:La_taverna/Tecnicismes/Arxius_2 .

C. Scott provided some help in IRC:

[11/13/14 16:53] <cscott> superm401: if you actually properly resolve the given URL with the given base HREF, your pages will continue to work both before and after

I don't know if we preserve old base hrefs given by parsoid though.

He also said:

[11/13/14 16:54] <cscott> superm401: and if you ignore base href and just replace(/([.][.]?\/)*/, '') on your hrefs to get article titles, your pages will also continue to work both before and after

Since I think we're doing some server-side processing, I think the second one should be doable.

Details

SubjectRepoBranchLines +/-
Gracefully handle non-ascii hrefs in redlinkermediawiki/extensions/Flowwmf/1.25wmf12+13 -3
Gracefully handle non-ascii hrefs in redlinkermediawiki/extensions/Flowmaster+13 -3
Dont output base hrefsmediawiki/extensions/Flowmaster+0 -7
Add ConfirmEdit dependency to Flow phpunitintegration/configmaster+1 -1
Customize query in gerrit

Revisions and Commits

rEFLW extension-Flow
rEFLW234c58d7a9c7 Gracefully handle non-ascii hrefs in redlinker
rEFLW168ee8c77ed3 Gracefully handle non-ascii hrefs in redlinker
rEFLW0bac08ec24a1 Gracefully handle non-ascii hrefs in redlinker
rMEXT MediaWiki Extensions
rMEXTc990db92ce61 Updated mediawiki/extensions Project: mediawiki/extensions/Flow…

Related Objects
Search...

Event Timeline

• Mattflaschen-WMF created this task.Dec 9 2014, 11:13 PM
• Mattflaschen-WMF raised the priority of this task from to High.
• Mattflaschen-WMF updated the task description. (Show Details)
• Mattflaschen-WMF added projects: Patch-For-Review, StructuredDiscussions.
• Mattflaschen-WMF changed Security from none to None.
• Mattflaschen-WMF added subscribers: Unknown Object (MLST), Quiddity, • DannyH and 7 others.
• marcoil subscribed.Dec 10 2014, 3:35 PM
• Mattflaschen-WMF merged a task: T78128: Internal links point to wrong targets in description and comments of subpages with Flow.Dec 10 2014, 6:46 PM
• Mattflaschen-WMF added subscribers: Qgil, Vriullop.
Qgil added a comment.EditedDec 10 2014, 7:58 PM

Now internal links work! Except this case:

gerritbot added a comment.Dec 10 2014, 11:39 PM

Change 179007 had a related patch set uploaded (by EBernhardson):
Gracefully handle non-ascii hrefs in redlinker

https://gerrit.wikimedia.org/r/179007

Patch-For-Review

EBernhardson claimed this task.Dec 10 2014, 11:40 PM
EBernhardson added a project: Collaboration-Team-Triage.
EBernhardson moved this task from Untriaged to Sprint O (ends Jan 27) on the Collaboration-Team-Triage board.
EBernhardson moved this task from Sprint O (ends Jan 27) to Code Review on the Collaboration-Team-Triage board.
gerritbot added a comment.Dec 11 2014, 12:31 AM

Change 179024 had a related patch set uploaded (by EBernhardson):
Add ConfirmEdit dependency to Flow phpunit

https://gerrit.wikimedia.org/r/179024

Patch-For-Review

• Spage added a project: WMF-Server-Backports.Dec 11 2014, 1:22 AM

Collaboration-Team-Triage want to backport this to 1.25wmf11 to fix Catalan wiki.

• marcoil added a comment.Dec 11 2014, 12:14 PM

The problem appears only when viewing a thread in the main discussion page, not when viewing it in the thread's page. For example, wikitext [[Viquipèdia_Discussió:Flow]] gives:

gerritbot added a comment.Dec 11 2014, 3:32 PM

Change 179024 merged by jenkins-bot:
Add ConfirmEdit dependency to Flow phpunit

https://gerrit.wikimedia.org/r/179024

EBernhardson added a comment.Dec 11 2014, 4:11 PM

The problem looks to be related to non-ascii in the title, flow was not appropriately calling urldecode on the value taken from an href before passing it into the Title class. This resulted in the original parsoid anchors being output instead of the rewritten anchors from Linker::link.

We can also look into converting red linker back into only processing red links. It was switched to handle all links instead of only non-existant due to issues we had with subpage links, but the changes to parsoid anchor output since then should have fixed that.

• Mattflaschen-WMF renamed this task from Fix base href to handle new Parsoid relative links to Fix base href and red link rewriter to handle new Parsoid relative links.Dec 12 2014, 7:02 PM
gerritbot added a comment.Dec 12 2014, 7:03 PM

Change 179512 had a related patch set uploaded (by Mattflaschen):
Dont output base hrefs

https://gerrit.wikimedia.org/r/179512

Patch-For-Review

cscott added a comment.Dec 12 2014, 8:51 PM

Just for the record, there's one more patch possibly coming down the pipe here:

Currently Parsoid emits a leading ./ for every link, because otherwise browsers would try to parse File:Foo.jpg as the url *protocol* file: (ie, instead of http). Initial colons are special in urls. So currently we emit <a href="./File:Foo.jpg"> -- the slash ensures that the colon isn't misinterpreted as a protocol marker.

We are considering eventually URL-encoding these initial colons (which don't otherwise need to be URL-encoded) so that links to File:Foo.jpg would become <a href="File%3AFoo.jpg">. This would make it even easier for folks to consume our HTML -- you just need to URL-decode the href to get the article title -- at the cost of making it a little less pretty for humans to read.

So if you are feeling forward thinking you might want to double-check that your href-parsing will handle File%3A:Foo.jpg just as well as ./File:Foo.jpg.

• Mattflaschen-WMF added a comment.Dec 12 2014, 9:50 PM
In T78090#844918, @cscott wrote:

So if you are feeling forward thinking you might want to double-check that your href-parsing will handle File%3A:Foo.jpg just as well as ./File:Foo.jpg.

Thanks for the heads-up. Erik's fix (https://gerrit.wikimedia.org/r/#/c/179007/) adds a urldecode which is required anyway for non-ASCII characters, but I think will address this too.

gerritbot added a comment.Dec 15 2014, 11:51 PM

Change 179007 merged by jenkins-bot:
Gracefully handle non-ascii hrefs in redlinker

https://gerrit.wikimedia.org/r/179007

EBernhardson closed this task as Resolved.Dec 15 2014, 11:51 PM
EBernhardson added a commit: rEFLW0bac08ec24a1: Gracefully handle non-ascii hrefs in redlinker.

Closed by commit rEFLW0bac08ec24a1.

Jdlrobson unsubscribed.Dec 16 2014, 12:08 AM
Qgil added a comment.Dec 16 2014, 8:51 AM

Thanks! When do you expect this change to be deployed in ca.wiki?

gerritbot added a comment.Dec 16 2014, 4:08 PM

Change 179512 merged by jenkins-bot:
Dont output base hrefs

https://gerrit.wikimedia.org/r/179512

• Mattflaschen-WMF mentioned this in rEFLW54ef9ea19e6d: Dont output base hrefs.Dec 16 2014, 4:09 PM
Quiddity added a comment.Dec 16 2014, 9:18 PM
In T78090#849953, @Qgil wrote:

Thanks! When do you expect this change to be deployed in ca.wiki?

It should be deployed in tomorrow's release train, as it's in 1.25wmf12. (I asked in IRC)

gerritbot added a comment.Dec 16 2014, 9:18 PM

Change 180303 had a related patch set uploaded (by EBernhardson):
Gracefully handle non-ascii hrefs in redlinker

https://gerrit.wikimedia.org/r/180303

Patch-For-Review

gerritbot added a comment.Dec 17 2014, 12:31 AM

Change 180303 merged by jenkins-bot:
Gracefully handle non-ascii hrefs in redlinker

https://gerrit.wikimedia.org/r/180303

EBernhardson added a commit: rEFLW168ee8c77ed3: Gracefully handle non-ascii hrefs in redlinker.Dec 17 2014, 12:32 AM
Quiddity removed a subscriber: Maryana.Dec 19 2014, 1:25 AM
EBernhardson mentioned this in rCICFeb6f80f0490b: Add ConfirmEdit dependency to Flow phpunit.Dec 31 2014, 6:27 PM
Diffusion added a commit: rMEXTc990db92ce61: Updated mediawiki/extensions Project: mediawiki/extensions/Flow….Mar 11 2015, 4:38 AM
Diffusion mentioned this in rMEXT713c0c91b6f3: Updated mediawiki/extensions Project: mediawiki/extensions/Flow….Mar 11 2015, 6:22 AM
Quiddity mentioned this in T97660: Links in preview don't link correctly (Flow boards on subpage).Apr 30 2015, 5:24 PM
EBernhardson added a commit: rEFLW234c58d7a9c7: Gracefully handle non-ascii hrefs in redlinker.Apr 28 2016, 2:41 AM
• Phabricator_maintenance removed subscribers: • wikibugs-l-list, Collaboration-Team-Triage.Aug 12 2016, 11:51 AM
Content licensed under Creative Commons Attribution-ShareAlike (CC BY-SA) 4.0 unless otherwise noted; code licensed under GNU General Public License (GPL) 2.0 or later and other open source licenses. By using this site, you agree to the Terms of Use, Privacy Policy, and Code of Conduct. · Wikimedia Foundation · Privacy Policy · Code of Conduct · Terms of Use · Disclaimer · CC-BY-SA · GPL · Credits