Page MenuHomePhabricator

LanguageConverter should work in category listings
Open, Needs TriagePublic

Description

Take a look at this category. While most articles are in Cyrillic, there are four at the end that are in the Latin script. When sr-ec is selected from the LC tab at the top, the Latin page names don't change into Cyrillic ones, and vice versa.

Event Timeline

Note that this request is not going to work correctly all the time. In Serbian Wikipedia, articles can be named either in Cyrillic (90% of the time, this is the case) or in Latin.

Sometimes, page name is in Latin, but plain Serbian name, but lot of times it is in English or other native Latin language (e.g. names of music bands, brands, songs/albums/etc, pseudonyms which cannot be transliterated... Latin names of species...) Sometimes, it is even mixed, like "The Beatles (албум)"...

In category https://sr.wikipedia.org/sr-ec/Категорија:Поврће, there are, for example, three articles with titles in Latin:

  • Pasulji
  • Phaseolus polyanthus
  • Povrtarstvo

First and third are native Serbian words, and cam be transliterated. Middle one is a Latin nomenklature name, and should be always in Latin, regardless of selected variant. So, if user would select "sr-ec", it should be like:

  • Пасуљи
  • Phaseolus polyanthus
  • Повртарство

However, as far as I know, it would near impossible to implement this, as category is constructed. For correct transliteration of title of the article, you would have to render the article in question using language converter, and then deduct what correct title should be.

OK, that may be useful. I assume that stored canonical representation would be used at places where page is linked (e.g. in other article)?

In any case, that work may help with

T257309: LanguageConverter should support preventing the conversion of a category name

...I guess.