Page MenuHomePhabricator

Content-Transform-TeamGroup
ActivePublic

Milestones

Watchers (3)

Details

Description

The WMF Content Transform team is responsible for maintaining the wikitext parsing products and Maps, among other things. Please see our team page for more details.

Content-Transform-Team-WIP lists current work in progress.

Recent Activity

Today

Maryann-Onyinye moved T398724: Fragments in redirects returned by action=parse are wrong when they contain an underscore from Backlog to Hacking Projects on the Wikimania-Hackathon-2025 board.
Thu, Jul 17, 8:52 AM · MW-Interfaces-Team, MediaWiki-Redirects, MediaWiki-Action-API, Wikimania-Hackathon-2025, affects-Kiwix-and-openZIM, Content-Transform-Team
Dcljr updated the task description for T399785: Post-creation work for zghwiktionary.
Thu, Jul 17, 5:03 AM · Patch-For-Review, Countervandalism-Network, Content-Transform-Team, Wiki-Setup
Pppery moved T399785: Post-creation work for zghwiktionary from Untriaged to Post-creation tasks (wiki not yet created) on the Wiki-Setup board.
Thu, Jul 17, 12:54 AM · Patch-For-Review, Countervandalism-Network, Content-Transform-Team, Wiki-Setup
gerritbot added a project to T399785: Post-creation work for zghwiktionary: Patch-For-Review.
Thu, Jul 17, 12:45 AM · Patch-For-Review, Countervandalism-Network, Content-Transform-Team, Wiki-Setup
gerritbot added a comment to T399785: Post-creation work for zghwiktionary.

Change #1170221 had a related patch set uploaded (by Gerrit maintenance bot; author: Gerrit maintenance bot):

[analytics/refinery@master] Add zgh.wiktionary to pageview allowlist

https://gerrit.wikimedia.org/r/1170221

Thu, Jul 17, 12:45 AM · Patch-For-Review, Countervandalism-Network, Content-Transform-Team, Wiki-Setup
Maintenance_bot created T399785: Post-creation work for zghwiktionary.
Thu, Jul 17, 12:44 AM · Patch-For-Review, Countervandalism-Network, Content-Transform-Team, Wiki-Setup
Djackson-ctr moved T398952: Inconsistent/confusing styles for temporary account links from Needs QA to Done on the Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)) board.

I have verified that the code changes have been reverted (User-related link colors have been removed).
QA was completed for this ticket using Test wiki 1.45.0-wmf.9 (a9e4ca5) and 1.45.0-wmf.10 (06dce30), and localhost 1.45.0-alpha (1ba8f04).
Thank you @Tchanders.

Thu, Jul 17, 12:35 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team

Yesterday

ssastry created T399776: Edge case wikilink parsing bug.
Wed, Jul 16, 9:55 PM · Parsoid-Read-Views (Small Size Wikipedias), Content-Transform-Team, Parsoid

Tue, Jul 15

ssastry updated the task description for T399638: Edge case rendering difference for <poem> content in Parsoid.
Tue, Jul 15, 9:36 PM · Parsoid-Read-Views (Small Size Wikipedias), Content-Transform-Team, Parsoid
ssastry created T399638: Edge case rendering difference for <poem> content in Parsoid.
Tue, Jul 15, 9:35 PM · Parsoid-Read-Views (Small Size Wikipedias), Content-Transform-Team, Parsoid
HCoplin-WMF added a comment to T399197: Transform endpoints not working as expected.

@ABreault-WMF -- I took a look at the diff. It's incredibly hard to parse though, given changes to formatting and wikitext nuance. I don't think it would actually be useful to a customer trying to figure out what changed so they can decide whether or not they are safe to upgrade. Do y'all have actual spec files that we could compare or something? Or are the wikis the only place specs are described?

Tue, Jul 15, 9:01 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
ssastry created T399635: {{#ifexpr:{{{1}}}>1|{{{2}}}}} errors in Parsoid.
Tue, Jul 15, 8:49 PM · Content-Transform-Team, Parsoid-Read-Views (Small Size Wikipedias), Parsoid
ssastry updated the task description for T399634: Parsoid incorrectly tokenizes links with tplargs in them.
Tue, Jul 15, 8:46 PM · Parsoid-Read-Views (Small Size Wikipedias), Content-Transform-Team, Parsoid
ssastry created T399634: Parsoid incorrectly tokenizes links with tplargs in them.
Tue, Jul 15, 8:44 PM · Parsoid-Read-Views (Small Size Wikipedias), Content-Transform-Team, Parsoid
HNordeenWMF moved T398838: transform/wikitext endpoint returns error if content too large. from Needs Triage to Tracking on the Wikipedia-iOS-App-Backlog board.
Tue, Jul 15, 5:57 PM · Wikipedia-iOS-App-Backlog, Content-Transform-Team, Wikipedia-Android-App-Backlog
Tchanders moved T398952: Inconsistent/confusing styles for temporary account links from To triage to In progress on the Temporary accounts (Global wiki rollout) board.
Tue, Jul 15, 5:23 PM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
Tchanders edited projects for T398952: Inconsistent/confusing styles for temporary account links, added: Temporary accounts (Global wiki rollout), Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)); removed Temporary accounts.
Tue, Jul 15, 5:22 PM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
Tchanders moved T398952: Inconsistent/confusing styles for temporary account links from Needs Other Teams to Someday / Maybe on the Temporary accounts board.
Tue, Jul 15, 5:22 PM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
Tchanders edited projects for T398952: Inconsistent/confusing styles for temporary account links, added: Temporary accounts; removed Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout).
Tue, Jul 15, 5:21 PM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
OWresch-WMF edited projects for T389587: [SPIKE] Optimize PageHTMLHandler::getETag to avoid ParserOutput, added: MW-Interfaces-Team (MWI-Sprint-14 (2025-07-15 to 2025-07-29)); removed MW-Interfaces-Team (MWI-Sprint-13 (2025-07-01 to 2025-07-15)).
Tue, Jul 15, 2:56 PM · MW-Interfaces-Team (MWI-Sprint-14 (2025-07-15 to 2025-07-29)), Content-Transform-Team, MediaWiki-REST-API, Wikimedia-Performance-recommendation
Niharika added a comment to T398952: Inconsistent/confusing styles for temporary account links.

4 Grey-background link styles are sometimes applied to the new-messages-from-users bubble - on wikis without Notifications (Echo) installed, if you leave a message on a temporary account's talk page, the person behind that temporary account will see the link styling applied to the "a new message" and "last change" links within this bubble when they're on a page history/user contributions page, but (seemingly) not if they're on most other pages. -- "a new message" and "last change" links should have the grey background.

@Niharika Should this say instead that they should not have the grey background? Just asking because this seems to be the one example of a link having a grey background that isn't a user link.

Tue, Jul 15, 10:26 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team

Mon, Jul 14

ReleaseTaggerBot added a project to T399081: Lua `mw.title.new("Z42"):getContent()` stopped working on wikifunctions: MW-1.45-notes (1.45.0-wmf.10; 2025-07-15).
Mon, Jul 14, 2:00 PM · MW-1.45-notes (1.45.0-wmf.10; 2025-07-15), Parsoid (Tracking), Content-Transform-Team, Abstract Wikipedia team, Parsoid-Read-Views
Maintenance_bot removed a project from T399081: Lua `mw.title.new("Z42"):getContent()` stopped working on wikifunctions: Patch-For-Review.
Mon, Jul 14, 1:31 PM · MW-1.45-notes (1.45.0-wmf.10; 2025-07-15), Parsoid (Tracking), Content-Transform-Team, Abstract Wikipedia team, Parsoid-Read-Views
TheDJ added a comment to T383127: Wikimedia maps show up as gray boxes.

FYI: For me the (static) map now renders in the iOS app.

Mon, Jul 14, 1:19 PM · Content-Transform-Team, iOS-Bug-Backlog, Maps (Kartographer), Wikipedia-iOS-App-Backlog
gerritbot added a comment to T399081: Lua `mw.title.new("Z42"):getContent()` stopped working on wikifunctions.

Change #1167859 merged by jenkins-bot:

[mediawiki/extensions/WikiLambda@master] ZObjectContentHandler::fillParserOutput: Add labels as parser page properties

https://gerrit.wikimedia.org/r/1167859

Mon, Jul 14, 1:01 PM · MW-1.45-notes (1.45.0-wmf.10; 2025-07-15), Parsoid (Tracking), Content-Transform-Team, Abstract Wikipedia team, Parsoid-Read-Views
Seddon added a project to T383127: Wikimedia maps show up as gray boxes: Content-Transform-Team.
Mon, Jul 14, 10:40 AM · Content-Transform-Team, iOS-Bug-Backlog, Maps (Kartographer), Wikipedia-iOS-App-Backlog
Tchanders added a comment to T398952: Inconsistent/confusing styles for temporary account links.

4 Grey-background link styles are sometimes applied to the new-messages-from-users bubble - on wikis without Notifications (Echo) installed, if you leave a message on a temporary account's talk page, the person behind that temporary account will see the link styling applied to the "a new message" and "last change" links within this bubble when they're on a page history/user contributions page, but (seemingly) not if they're on most other pages. -- "a new message" and "last change" links should have the grey background.

Mon, Jul 14, 8:53 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team

Sat, Jul 12

Niharika added a comment to T398952: Inconsistent/confusing styles for temporary account links.

@Esanders That makes sense to me.

we should style all contrib links (links to [[Special:Contributions/<tempuser>]] with any label) and user page links in the content area, as this will give us a reasonably low false-positive and false-negative rate.

Noting for @Niharika that this would mean styling links that have the temp user name as text, but link to the user page rather than the contribs page (as happens with @ mentions). Does this sound OK?

Sat, Jul 12, 2:48 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team

Fri, Jul 11

OSleger-WMF claimed T371617: Wikimedia\Assert\InvariantException: Invariant failed: Bad UTF-8 at end of string (2 byte sequence).
Fri, Jul 11, 8:29 PM · Essential-Work, User-brennen, Content-Transform-Team, Parsoid, Wikimedia-production-error
ABreault-WMF added a comment to T399197: Transform endpoints not working as expected.

@ABreault-WMF -- was v1.x formally deprecated? If so, when was it, and what does the announcement process look like? I see we're on 2.7 now, but I'd like to have something we can point to with the comms about the versioning changes.

Fri, Jul 11, 3:32 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
Tchanders added a comment to T398952: Inconsistent/confusing styles for temporary account links.

@Esanders That makes sense to me.

Fri, Jul 11, 1:16 PM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
Esanders added a comment to T398952: Inconsistent/confusing styles for temporary account links.

For non article content links, we should maintain the status quo before this patch which is to apply the mw-tempuserlink class to anything that currently has mw-userlink and is for a temp user.

Fri, Jul 11, 1:01 PM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
HCoplin-WMF added a comment to T399197: Transform endpoints not working as expected.

@ABreault-WMF -- was v1.x formally deprecated? If so, when was it, and what does the announcement process look like? I see we're on 2.7 now, but I'd like to have something we can point to with the comms about the versioning changes.

Fri, Jul 11, 12:49 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
OSleger-WMF added a comment to T371617: Wikimedia\Assert\InvariantException: Invariant failed: Bad UTF-8 at end of string (2 byte sequence).

Possible replication:

Template:Ta

<includeonly>{|
<div>
</includeonly>

Template:Tb

<includeonly>
</div>|}
</includeonly>

Page

{{Ta}}
řaa
<table>
<tr><td>
a
</tr></td>
</table>
{{Tb}}

Letter ř in this case can be replace with any 2 byte utf-8 character.
Error stack trace:

Fri, Jul 11, 11:25 AM · Essential-Work, User-brennen, Content-Transform-Team, Parsoid, Wikimedia-production-error
Winston_Sung added a comment to T399081: Lua `mw.title.new("Z42"):getContent()` stopped working on wikifunctions.

Emm... I think the correct way should be getWikitextForTransclusion instead of $wgNonincludableNamespaces.

Fri, Jul 11, 11:18 AM · MW-1.45-notes (1.45.0-wmf.10; 2025-07-15), Parsoid (Tracking), Content-Transform-Team, Abstract Wikipedia team, Parsoid-Read-Views
Niharika added a comment to T398952: Inconsistent/confusing styles for temporary account links.

Question that came to mind: What if a temporary account is pinged by another user in a discussion (e.g. by using @[[User:~2025-3105|~2025-3105]])? Should the temporary account username in the ping have a grey background?
& what if an editor pings them in a way such as Thank [[User:~2025-3105|you]] for the question! [...]? Would/should the word "you" have a grey background in this context?

Fri, Jul 11, 11:08 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
Bugreporter added a comment to T399081: Lua `mw.title.new("Z42"):getContent()` stopped working on wikifunctions.

I would say WikiLambda ZObject pages are actually JSON pages (not something like Wikibase entity pages) and therefore not something really should be added to $wgNonincludableNamespaces.

https://www.wikifunctions.org/wiki/Z4?action=raw

vs.

https://www.wikidata.org/wiki/Property:P31?action=raw

Fri, Jul 11, 11:05 AM · MW-1.45-notes (1.45.0-wmf.10; 2025-07-15), Parsoid (Tracking), Content-Transform-Team, Abstract Wikipedia team, Parsoid-Read-Views
A_smart_kitten added a comment to T398952: Inconsistent/confusing styles for temporary account links.

Question that came to mind: What if a temporary account is pinged by another user in a discussion (e.g. by using @[[User:~2025-3105|~2025-3105]])? Should the temporary account username in the ping have a grey background?

Fri, Jul 11, 9:53 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team
Niharika added a comment to T398952: Inconsistent/confusing styles for temporary account links.

Generally speaking we only want the grey background for temporary account usernames and not on other links, even if they play a supporting role.

Fri, Jul 11, 9:01 AM · Trust and Safety Product Sprint (Sprint Cannoli (July 7 - July 25)), Temporary accounts (Global wiki rollout), MW-1.45-notes (1.45.0-wmf.9; 2025-07-08), Content-Transform-Team, Trust and Safety Product Team

Thu, Jul 10

daniel added a comment to T399197: Transform endpoints not working as expected.

Parsoid generally doesn't protect against this because we assume the input is valid UTF-8 and sanitized before it is given to the parser. It appears that perhaps the REST API needs to call UtfNormal\Validator::cleanUp on its input at some point before it is given to Parsoid.

Thu, Jul 10, 8:47 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
ABreault-WMF added a comment to T399197: Transform endpoints not working as expected.

Removing the profile would mean they are willing to accept any breaking changes we may make. The point of content negotiation was
that we were going to give a grace period between major version bumps for clients to upgrade. I'm not sure how well we've adhered to semver though and if the client is still on 1.2.1 then they haven't been following any announcements about version upgrades anyways.

Thu, Jul 10, 8:10 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
daniel added a comment to T399197: Transform endpoints not working as expected.

@ABreault-WMF is right. They confirmed they're getting a 406.

Thu, Jul 10, 8:02 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
HCoplin-WMF added a comment to T399197: Transform endpoints not working as expected.

@ABreault-WMF is right. They confirmed they're getting a 406. The error code was not reported in the initial response, so I assumed 500 (especially after messing around and being able to make 500s trigger). Does this mean the fix would be to either ask them to remove the spec version, or add support for specifying a spec version?

Thu, Jul 10, 6:36 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
Jdforrester-WMF added a comment to T399081: Lua `mw.title.new("Z42"):getContent()` stopped working on wikifunctions.

I would say WikiLambda ZObject pages are actually JSON pages (not something like Wikibase entity pages) and therefore not something really should be added to $wgNonincludableNamespaces.

https://www.wikifunctions.org/wiki/Z4?action=raw

vs.

https://www.wikidata.org/wiki/Property:P31?action=raw

Thu, Jul 10, 4:29 PM · MW-1.45-notes (1.45.0-wmf.10; 2025-07-15), Parsoid (Tracking), Content-Transform-Team, Abstract Wikipedia team, Parsoid-Read-Views
cscott added a comment to T399197: Transform endpoints not working as expected.

I am seeing another kind of failure on this endpoint, starting on the 18th, e.g. reqId c69c2570-2213-4e2e-901d-357481cd5ef9:

Error message: PHP Warning: Uninitialized string offset 39

Stack trace:

from /srv/mediawiki/php-1.45.0-wmf.6/vendor/wikimedia/wikipeg/src/PEGParserBase.php(144)
Thu, Jul 10, 3:53 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
daniel added a comment to T399197: Transform endpoints not working as expected.

I found an instance on logstash, requestId 6d0f6eda-1e06-4446-ab0a-ebb59b65783a:

The error message is TypeError: normalizer_normalize(): Argument #1 ($string) must be of type string, array given.

...

However, they are all from today - perhaps from someone trying to reproduce this issue, and triggering another issue...

Thu, Jul 10, 3:18 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
ABreault-WMF added a comment to T399197: Transform endpoints not working as expected.

Not sure how relevant this is to the current issue but Parsoid can't generate a semantically equivalent version for that. This should at least be returning a 406

Thu, Jul 10, 3:06 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
HCoplin-WMF added a comment to T399197: Transform endpoints not working as expected.

Steps to reproduce the 500 error in the REST framework:

Thu, Jul 10, 2:50 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
daniel added a comment to T399197: Transform endpoints not working as expected.

Analysis of the TypeError:

Thu, Jul 10, 2:46 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team
daniel updated subscribers of T399197: Transform endpoints not working as expected.

Tagging Content-Transform-Team for the parsoid issue. @cscott any thoughts?

Thu, Jul 10, 2:28 PM · MediaWiki-REST-API, RESTBase Sunsetting, Content-Transform-Team, MW-Interfaces-Team