Page MenuHomePhabricator

Vandalism on Structured Data tasks
Closed, ResolvedPublic

Description

Originally reported by @Ramsey-WMF , there were about 125 tasks affected on Structured-Data-Backlog by @Sarahmarie1981 . Approximately 25 have been reverted (thanks @Aklapper !), so about 100 remain. I'm not aware of a way to search or filter by user-has-edited, so that may have to happen on the backend.

Happy to help with this if Phab provides a tool to help me identify more automatically than searching each task. We know the tasks exist because of email notifications, but it's not easy to quickly make a query from that. :-/

Related Objects

Mentioned Here
T174031: MCR: Include all slots in XML dumps
T214201: Implement NSFW image classifier using Open NSFW
T215305: "Depicts and other statements on a bicycle": Qualifiers, and search by depicts statements, and other statements
T220459: [stub] Add "click here to see more" functionality to statement panels
T220955: MediawikiEditEntity::checkEditPermissions fatals with unknown message 'no-permission'
T221015: Enforce statement uniqueness on Filepage
T221233: Unable to submit structured data on IE11
T221681: Clarify the MediaInfo data model in relation to Wikibase [Epic]
T222159: WikibaseMediaInfo serializes empty statements as empty array [] instead of empty object {} in JSON
T222319: Pages are not added to watchlist when structured data edits are undone (but the user's preferences are set to do so)
T222322: Edit summary box is empty (no default value) when undoing structured data edits
T223823: Add structured data to api response
T224139: Allow campaign admin to suggest statements for display in UploadWizard
T224142: Suggested statements (without values) in UploadWizard configurable via query string
T224147: Mechanism to allow a campaign admin to add help text for pre-selected statements
T224214: Allow structured data to be added via API:Upload
T225172: Explore new informational UI elements to explain SDC functionalities
T227264: When js is turned off the 'captions' header appears below the actual captions on the File page
T228770: Design a system for serving images with depicts suggestions needing confirmation
T229007: Show structured data from remote image repositories on file pages
T229008: Allow remote editing of structured data from other wikis
T229437: How should we expose historical information on suggested labels (and vote(s) on them)?
T230314: Show constraint violations on SDC statements
T230563: Suggested statements WITH values in UploadWizard configurable via url query string
T230607: stop using page_content_model
T231917: Enable users to actually edit structured data for relevant files through this tool
T232087: Statements not editable on a File page whose id has changed
T232101: Incorrect suggested statements appearing for a campaign
T232200: Write selenium tests for QA checklist "Add Structured data during file upload with Upload Wizard" section
T232201: Write selenium tests for QA checklist "Check page/file history AND recent changes" section
T232203: Write selenium tests for QA checklist "Check that watchlists work properly" section
T232204: Write selenium tests for QA checklist "Check that diffs between past revisions/edits works" section
T232205: Write selenium tests for QA checklist "Check that Search works properly" section
T232207: Write selenium tests for QA checklist "Check UI design integrity" section
T232208: Write selenium tests for QA checklist "Check Language support beyond English" section
T233036: Enable support for all WB data types in top-level Statements and Qualifiers
T235658: Unnecessary calls to wbformatvalue on File page
T235942: "somevalue" SDOC statements not visible in search index
T236431: Data dumps for the MachineVision extension
T237089: Create CQS puppet configs by applying query_service module
T240265: Order depicts autocomplete suggestions in search box by number of times they used
T240281: Clarify confusion between Multimedia and Structured Data team and update docs accordingly
T240516: Do not serve label suggestions that have already been added as depicts statements
T240897: Mobile version of haswbstatement:P<depicts>=Q<your term> autocomplete suggestions
T241005: Qualifier switches statements on cancel
T242054: Edit links should not be printable
T242144: Set custom summary while editing structured data
T242225: Document input format for all data types for wbcreateclaim, wbsetclaim, wbsetclaimvalue, etc. somewhere
T242414: Consider adding suggestion widget to OOUI
T242662: "machinevision-machineaidedtagging-intro" message should not refer to production interwiki codes
T242958: InvalidArgumentException: "property is not provided by source" (via wbeditentity API)
T243410: Geo-coord inputs marked as invalid when they're not

Event Timeline

Oh, hang on. Herald re-added it

Screenshot 2020-03-17 at 18.32.37.png (308×1 px, 86 KB)

It would look like it's all cleaned up TBH

Using the user's history page text, and a spreadsheet, I pulled out some IDs to investigate:

T220459, T220955, T221233, T221681, T222159, T222319, T222322, T224139, T224142, T224147, T224214, T225172, T227264, T228770, T229007, T229008, T215305, T230563, T230607, T231917, T232101, T232200, T232201, T232203, T232204, T232205, T232207, T232208, T235658, T235942, T237089, T230314, T174031, T233036, T221015, T223823, T236431, T240265, T240281, T240516, T232087, T240897, T241005, T242054, T242144, T242225, T242414, T242662, T229437, T242958, T243410, T214201

It used to be possible to plug this into advanced search, but I am struggling to figure that out now.

@Ramsey-WMF , can you confirm @Reedy 's assessment? Anything stand out to you?

One thing to check is what herald rule is adding that. For example T231917 did get the extra project removed, but herald didn't re-add the one removed

One thing to check is what herald rule is adding that. For example T231917 did get the extra project removed, but herald didn't re-add the one removed

Which is https://phabricator.wikimedia.org/herald/transcript/3585845/

Screenshot 2020-03-17 at 18.37.05.png (414×2 px, 170 KB)

https://phabricator.wikimedia.org/H331

@Reedy I believe that is because the Herald is set to only add the project once. That is to stop it from constantly re-adding tasks that were intended to have the tag removed. That, however, doesn't account for vandalism, so what would be really handy is a simple list of anything @Sarahmarie1981 edited so we can review which of those tasks don't have the Structured-Data-Backlog tag.

Yeah Herald missed re-adding tags for a good number of the tickets but @JJMC89 appears to be going through and fixing that manually right now 😺 (thanks!)

JJMC89 claimed this task.

I've restored Structured-Data-Backlog where Herald didn't.

See https://wikitech.wikimedia.org/wiki/Phabricator#Revert_all_activity_of_a_given_user for a script to mass-rollback, but can only be run by system administrator (@Aklapper for notice).

Too late for that, if manual action has been taken, I believe

Phabricators tooling for this stuff is awful, basically. People complain about MediaWiki, but it's amazing in comparison

Also Thanks @JJMC89 as I missed some tasks (sorry for that)!

It used to be possible to plug this into advanced search, but I am struggling to figure that out now.

https://phabricator.wikimedia.org/maniphest/?ids=1,2,3 basically

@Aklapper That's it! I could not find it because that field does not show unless the URL has been altered:
https://phabricator.wikimedia.org/maniphest/query/advanced/