Page MenuHomePhabricator

Issue with converting a specific post from HTML to wikitext with Parsoid
Closed, ResolvedPublic

Description

See parent task.

The HTML:

<meta typeof="mw:Includes/IncludeOnly" id="mwAg"/><meta typeof="mw:Includes/IncludeOnly/End" id="mwAw"/>

Real HTML in context:

<p about="#mwt1" typeof="mw:Transclusion" id="mwAQ" data-mw='{"parts":[{"template":{"target":{"wt":"从Wiki文本讨论页转换为Flow页面","href":"./Template:从Wiki文本讨论页转换为Flow页面"},"params":{"archive":{"wt":"User talk:Gakmo/存档 1"},"date":{"wt":"2015-10-19"}},"i":0}}]}'>既往讨论已于2015-10-19存档在<span class="flow-link-to-archive"><a rel="mw:WikiLink" href="./User_talk:Gakmo/存档_1" title="User talk:Gakmo/存档 1">User talk:Gakmo/存档 1</a></span>。</p>
<meta typeof="mw:Includes/IncludeOnly" id="mwAg"/><meta typeof="mw:Includes/IncludeOnly/End" id="mwAw"/>

will not decode at http://parsoid-lb.eqiad.wikimedia.org/_html/

Event Timeline

Mattflaschen-WMF claimed this task.
Mattflaschen-WMF raised the priority of this task from to Unbreak Now!.
Mattflaschen-WMF lowered the priority of this task from Unbreak Now! to High.
Mattflaschen-WMF updated the task description. (Show Details)
Mattflaschen-WMF set Security to None.
Mattflaschen-WMF renamed this task from Issue with converting a specific post from HTML to wikitext with RESTBase to Issue with converting a specific post from HTML to wikitext with Parsoid.Oct 27 2015, 12:24 AM
Mattflaschen-WMF updated the task description. (Show Details)
Mattflaschen-WMF edited projects, added Parsoid; removed RESTBase.

Sorry, I copied it wrong before. I didn't realize there was a second line.

I removed the paragraph, since it's not necessary to reproduce it.

I guess it might have originally been:

<includeonly></includeonly>

but don't really know.

Quite possibly this is related to T115236: Flow posts being serialized from HTML -> WT without providing Parsoid data-parsoid attributes?. We just changed Flow in production to stop using RESTBase. This HTML comes from when it was still on, though.

So, looks like we have to support conversion from html -> wt even when data-parsoid is absent. Should not be a bad thing to fix ... will make our code more robust. But, yes, wikitext will be normalized, but that is a known thing.