Page MenuHomePhabricator

Inconsistent data in api.wikimedia.org/feed/v1/wikipedia/en/featured
Closed, InvalidPublic

Description

Apologies for not having a consistent reproduce case, but I noticed while querying https://api.wikimedia.org/feed/v1/wikipedia/en/featured/2022/11/14 (no tokens or authentication of any kind), I got some mismatch in the tfa (today's featured article) portion of the feed.

The tfa section should, as far as I can tell, look like this:

{
  "type": "standard",
  "title": "Roberta_Williams",
  "displaytitle": "<span class=\"mw-page-title-main\">Roberta Williams</span>",
  "namespace": {
    "id": 0,
    "text": ""
  },
  "wikibase_item": "Q55176",
  "titles": {
    "canonical": "Roberta_Williams",
    "normalized": "Roberta Williams",
    "display": "<span class=\"mw-page-title-main\">Roberta Williams</span>"
  },
  "pageid": 268015,
  "thumbnail": {
    "source": "https://upload.wikimedia.org/wikipedia/commons/thumb/b/b3/Roberta_Williams_headshot.jpg/587px-Roberta_Williams_headshot.jpg",
    "width": 587,
    "height": 640
  },
  "originalimage": {
    "source": "https://upload.wikimedia.org/wikipedia/commons/b/b3/Roberta_Williams_headshot.jpg",
    "width": 705,
    "height": 768
  },
  "lang": "en",
  "dir": "ltr",
  "revision": "1121874264",
  "tid": "207056a0-6439-11ed-909b-691000e41133",
  "timestamp": "2022-11-14T16:26:39Z",
  "description": "American video game designer (born 1953)",
  "description_source": "local",
  "content_urls": {
    "desktop": {
      "page": "https://en.wikipedia.org/wiki/Roberta_Williams",
      "revisions": "https://en.wikipedia.org/wiki/Roberta_Williams?action=history",
      "edit": "https://en.wikipedia.org/wiki/Roberta_Williams?action=edit",
      "talk": "https://en.wikipedia.org/wiki/Talk:Roberta_Williams"
    },
    "mobile": {
      "page": "https://en.m.wikipedia.org/wiki/Roberta_Williams",
      "revisions": "https://en.m.wikipedia.org/wiki/Special:History/Roberta_Williams",
      "edit": "https://en.m.wikipedia.org/wiki/Roberta_Williams?action=edit",
      "talk": "https://en.m.wikipedia.org/wiki/Talk:Roberta_Williams"
    }
  },
  "extract": "Roberta Lynn Williams is an American video game designer and writer, who co-founded Sierra On-Line with her husband, game developer Ken Williams. In 1980 her first game, Mystery House, became a modest commercial success; it is credited as the first graphic adventure game. She is also known for creating and maintaining the King's Quest series, as well as designing the full motion video game Phantasmagoria in 1995.",
  "extract_html": "<p><b>Roberta Lynn Williams</b> is an American video game designer and writer, who co-founded Sierra On-Line with her husband, game developer Ken Williams. In 1980 her first game, <i>Mystery House</i>, became a modest commercial success; it is credited as the first graphic adventure game. She is also known for creating and maintaining the <i>King's Quest</i> series, as well as designing the full motion video game <span><i>Phantasmagoria</i></span> in 1995.</p>",
  "normalizedtitle": "Roberta Williams"
}

And that is indeed what I get consistently. But just a few minutes ago, I got this for the extract:

Khoja is a village in Multan Punjab Pakistan khoja enrich full culture different sects of societies are living here i.e #Mayo # Dogar #sharkies, #Baloch khoja UC144 MNA 158 AND PP203.

I am not sure what to make of that.

Event Timeline

This seems like potentially a serious bug (T324801 reminded me that I filed this), so hopefully someone working on RESTBase sunsetting work could take a look?

I am not away of any RESTbase Sunsetting work that would impact the operation of the feed endpoints in production. @Jgiannelos may know more.

daniel triaged this task as High priority.Dec 14 2022, 12:20 PM
daniel moved this task from Unsorted to PCS Service Pile on the RESTBase Sunsetting board.

Nothing changed on wikifeeds lately that would affect production. The usual culprit is caching that is pointing to an older revision.
That said can you clarify how to reproduce this? The API request returns the expected response according to the ticket description. I don't find any references to the snippet you posted.

Nothing changed on wikifeeds lately that would affect production. The usual culprit is caching that is pointing to an older revision.
That said can you clarify how to reproduce this? The API request returns the expected response according to the ticket description. I don't find any references to the snippet you posted.

It's something I observed while visiting https://api.wikimedia.org/feed/v1/wikipedia/en/featured/2022/11/14 in my browser, back on November 14. Some code in the app was reading data from a different revision, it seems, and swapping the wrong revision's extract into the data for the today's featured article section.

This doesn't seem actionable, since we can't reproduce it. Not sure what to do with the ticket. Let it sit on low priority? Or close it?

This doesn't seem actionable, since we can't reproduce it. Not sure what to do with the ticket. Let it sit on low priority? Or close it?

Yeah, I suppose we can close it.

Can't reproduce. Please re-open if new information becomes available.