Fix persistent markup leakage from Content Translation tool (Community Wishlist/W550)
The Content Translation tool (CX2) frequently injects malformed wikitext into published articles, leading to errors in reference lists (example).
Namely: in repeated references, the tool is leaking <span class="cx-segment"> tags (example), and corrupted citation structures, such as <ref name="...">...<cite class="citation web cs1" data-ve-ignore="true">...</cite></ref>; (example), resulting in error messages like Cite error: Invalid <ref> tag; name "Author2026" defined multiple times with different content.
This is a longstanding issue that persists despite years of reporting (T144167 and T287936). These errors require significant manual intervention from experienced editors to fix after publication:
- An older global filter has caught more than 6000 cases of the former issue: Global Filter 192 (~1 page/day)
- A new filter has been set up to track the latter issue: Global Filter 401 (~170 pages/day; disabled after acquiring enough samples, over around 4½ days)
Content Translation is intended to lower the barrier to entry for editors, but it is currently confusing inexperienced editors and creating a cleanup debt that falls on the shoulders of the community.
Unassigned
editors, patrollers
This wish currently has 25 supporters. Voting for this wish is open until it is completed.