As stated in T172642, there are a bunch of lowercase q-IDs (like q2120834) in page_props tables for various wikis. Correct ID is uppercase (Q2120834) and there may be situations - e.g. when integrating with other data sets - where it can be a problem. It would be nice to clean up such cases and put proper uppercase IDs there.
Description
Description
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Resolved | Smalyshev | T172642 Lowercase QIDs returned by wikibase:mwapi | |||
Resolved | Addshore | T119738 Problems with page_props on clients | |||
Resolved | Addshore | T172748 Clean up lowercase q-IDs from page_props table |
Event Timeline
Comment Actions
Before moving forward with this we should see if it is still an issue.
select count(*) from page_props where pp_propname = 'wikibase_item' and left(pp_value,1) = 'q' limit 1;
I should run this on the analytics cluster for all wikis, but it seems the sql server there is currently down...
Comment Actions
I did:
addshore@mwmaint1002:~$ foreachwikiindblist wikidataclient sql.php --query "/*T172748-addshore-wikidataclient*/select * from page_props where pp_propname = 'wikibase_item' and left(pp_value,1) = 'q' lim it 1;" > wikibase_item_page_props_lowercase_q_output
Which told me the follow wikis still have some lower case entity ids in page props
bat_smgwiki be_x_oldwiki bewiki eowiki hsbwiki kywiki map_bmswiki nds_nlwiki nlwiki scnwiki skwiki test2wiki zh_classicalwiki zh_min_nanwiki zh_yuewiki
Comment Actions
- bat_smgwiki
- be_x_oldwiki
- bewiki
- eowiki
- hsbwiki
- kywiki
- map_bmswiki
- nds_nlwiki
- nlwiki
- scnwiki
- skwiki
- test2wiki
- zh_classicalwiki
- zh_min_nanwiki
- zh_yuewiki
Comment Actions
I wrote a bot to touch the pages and reparse them, it fixed around half of them. The rest are either protected or has something that abusefilter or spamfilter blocks. I don't know what to do with those...
Comment Actions
Here are the number of rows that we are talking about.
1 @ bewiki 1 @ eowiki 6 @ hsbwiki 1 @ kywiki 2 @ scnwiki 1 @ skwiki 1 @ test2wiki
We could just go through and manually fix these and resolve this.