Edge case bug in table tokenizing
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	• Elitre
	Dec 31 2014, 5:37 PM

Description

The template is https://nl.wikipedia.org/wiki/Sjabloon:Infobox_luchthaven , and you can see examples at https://nl.wikipedia.org/w/index.php?title=Vliegbasis_Eindhoven&veaction=edit or https://nl.wikipedia.org/wiki/Luchthaven_Schiphol?veaction=edit . Tagging @Arlolra who's a wizard with templates (and can tell if Parsoiders need to be involved).

Related Objects

Mentioned In: rGPAR95ff6d60b615: Merge "T85627: Fix edge case tokenizing table lines"
rGPAR50056f6e5dea: T85627: Fix edge case tokenizing table lines

Event Timeline

• Elitre created this task.Dec 31 2014, 5:37 PM

• Elitre raised the priority of this task from to Needs Triage.

• Elitre updated the task description. (Show Details)

• Elitre added a project: VisualEditor.

• Elitre subscribed.

Clearly a Parsoid-related issue. The template (which generates a table) has several tables inside, and their closing |} are "indented" with some leading spaces (this is visible in template expansion). Parsoid misparses one of these as preformatted text, thus one of the tables is unclosed, this everything following the template gets included in the table it generates. Perhaps related to the use of {{!}} to build the tables?

http://parsoid.wmflabs.org/nlwiki/Vliegbasis_Eindhoven

Reproducible with this test snippet. There are space chars on the 3 lines after "|foo". It looks at least 2 empty lines with an empty leading space is required for the indented " |}" to be tokenized as a string instead of a table end tag.