Page MenuHomePhabricator

CX2: Strange display of a source paragraph in French with the {{citation}} template
Open, MediumPublic

Description

To reproduce, translate the article https://fr.wikipedia.org/wiki/Pierre_Gripari using CX2. I can reproduce it if I'm translating to both Hebrew and Russian.

Go to the paragraph titled "Personnalité". I see at least two issues with how it's displayed in the the source column:

  • There is a "↵" character in the beginning of the source paragraph. There is no reason for it to be displayed there.
  • More importantly, the content of the {{citation}} template is displayed twice: "« un Martien observant le monde des hommes avec une curiosité amusée, étranger au monde terrestre. « un Martien observant le monde des hommes avec une curiosité amusée, étranger au monde terrestre.". It's supposed to appear just once, and to end with a "»" character: "« un Martien observant le monde des hommes avec une curiosité amusée, étranger au monde terrestre. »"

Screenshot:

Screenshot_2018-11-25 תרגום דף – ויקיפדיה.png (932×1 px, 186 KB)

This may be a Parsoid or a Visual Editor issue, although I cannot reproduce it with the usual VisualEditor.

I just saw a similar phenomenon when translating Tove Skutnabb-Kangas from the German Wikipedia into the English Wikipedia. Here's how it looks in CX2:

Screenshot_2018-12-17 Translate page - Note This utility is currently restricted to extended confirmed editors on the Engli[...].png (601×2 px, 134 KB)

The original citation in the German Wikipedia is rendered like this:

Screenshot_2018-12-17 Tove Skutnabb-Kangas – Wikipedia.png (163×2 px, 42 KB)

So the adaptation into English by CX2 works quite correctly, but the rendering of the original is wrong, and this is quite confusing to me as a translator.

Event Timeline

I doubt this is a Parsoid issue -- whenever you run into this, check the Parsoid HTML for the page from the REST API. For example, https://fr.wikipedia.org/api/rest_v1/page/html/Pierre_Gripari ... My suspicion is that there is something going on in the CX code itself. IIRC, it splits content into atomic chunks for adaptation, so maybe the duplication is happening in rendering these chunks if there is something goofy with the keys assigned to these chunks?

Arrbee triaged this task as Medium priority.Jul 8 2019, 2:57 PM
Arrbee moved this task from Check & Move to Bugs on the ContentTranslation board.
Arrbee removed a project: Parsoid.