In order to understand if the behavior being used by the getTargetLanguage, i. e. the target language being the page language versus the user language, we should collect some information about this usage to decide how to create parity in the Parsoid interface.
Add a logging message in the target language function when page and user language are different.
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Open | None | T309166 Review language handling in Kartographer & Kartotherian | |||
| Resolved | MSantos | T311592 Collect information about old Parser `getTargetLanguage` usage |
Change 809587 had a related patch set uploaded (by MSantos; author: MSantos):
[mediawiki/core@master] Log target language behavior
Change 809937 had a related patch set uploaded (by MSantos; author: MSantos):
[mediawiki/extensions/Kartographer@master] Log target language behavior
Change 809587 abandoned by MSantos:
[mediawiki/core@master] Log target language behavior
Reason:
As per discussion in slack, this change is spammy and we should do it in Kartographer only hence abandoning it in favor of I3f944b81ff8b4e998741dc53a3cd6d92b1b2f254
Change 809937 merged by jenkins-bot:
[mediawiki/extensions/Kartographer@master] Log target language behavior
After this log being available for a while, we can see some insights from the language usage: https://logstash.wikimedia.org/goto/ed727275e91709dcc1187f4a31a9ed58.
We can see a few occurrences of user languages not being the same as the page language, see the top 3 results below:
normalized_message Count Target language (hu) is different than page language (en) (T311592) 7,463 Target language (fr) is different than page language (en) (T311592) 4,676 Target language (de) is different than page language (en) (T311592) 3,170
The wikis affected are listed below:
server # uk.wikipedia.org 115,580 ar.wikipedia.org 49,937 vec.wikipedia.org 28,870 commons.wikimedia.org 8,888 ru.wikipedia.org 6,291
Assuming the Page Language is always the same for TargetLanguage is not secure, although the impact is low. cc/ @ssastry, @Arlolra, and @cscott
I'm closing this ticket since the investigation is done, but some actions will be needed in the Kartographer compatibility with parsoid.
Change 883845 had a related patch set uploaded (by Thiemo Kreuz (WMDE); author: Thiemo Kreuz (WMDE)):
[mediawiki/extensions/Kartographer@master] [DNM] Remove obsolete page language logging
As far as I can tell the log entries are all from Wikimedia Commons. On Commons we split the parser cache by language. The basic wiki language and therefor the page language for all pages is English. Still we render the page in the user language and store it as such in the parser cache. The user language becomes the target language.
What this means is – as far as I understand it – that we need to continue using the parser's target language. We can not simplify this to use the page language instead.
Change #883845 merged by jenkins-bot:
[mediawiki/extensions/Kartographer@master] Remove obsolete page language logging