Page MenuHomePhabricator

[Bug] page_image entry (with empty value) added to page_props for entities with no relevant page image
Closed, ResolvedPublic

Description

page_image is currently always set and added to the page_props table, even if there is no relevant image. Instead of not adding the property, the property is added with a blank value.

e.g. for https://www.wikidata.org/wiki/Q5160

select * from page_props where pp_page = 6102;
+---------+--------------+------------------------------------------------------------------------------------------------------------------------------------------+------------+
| pp_page | pp_propname  | pp_value                                                                                                                                 | pp_sortkey |
+---------+--------------+------------------------------------------------------------------------------------------------------------------------------------------+------------+
|    6102 | displaytitle | <span class="wikibase-title ">
<span class="wikibase-title-label">'Are'are</span>
<span class="wikibase-title-id">(Q5160)</span>
</span> |       NULL |
|    6102 | page_image   |                                                                                                                                          |       NULL |
|    6102 | wb-claims    | 7                                                                                                                                        |          7 |
|    6102 | wb-sitelinks | 6                                                                                                                                        |          6 |
+---------+--------------+------------------------------------------------------------------------------------------------------------------------------------------+------------+
4 rows in set (0.00 sec)

we should instead should just not set the page_image property in the ParserOutput in this case. This is how it is done for wikitext pages.

Event Timeline

aude raised the priority of this task from to Needs Triage.
aude updated the task description. (Show Details)
aude subscribed.
aude set Security to None.

Change 249355 had a related patch set uploaded (by Aude):
Don't set page_image property when there is no 'best' image

https://gerrit.wikimedia.org/r/249355

Change 249355 merged by jenkins-bot:
Don't set page_image property when there is no 'best' image

https://gerrit.wikimedia.org/r/249355

Change 249445 had a related patch set uploaded (by Aude):
Don't set page_image property when there is no 'best' image

https://gerrit.wikimedia.org/r/249445

thiemowmde triaged this task as Medium priority.
thiemowmde edited projects, added Wikidata-Sprint-2015-10-13; removed Patch-For-Review.

Oh. I wrote the code how it was because I had, for some reason, the impression a page property is deleted by setting it to null.

Change 249445 abandoned by Aude:
Don't set page_image property when there is no 'best' image

Reason:
at this point, i think it's ok to wait until next scheduled deployment on wednesday

https://gerrit.wikimedia.org/r/249445