Page MenuHomePhabricator

PageImages does not consider Wikisource page covers when selecting image
Closed, ResolvedPublicFeature

Description

Feature summary (what you would like to be able to do and where):
When DesktopImprovements (Vector 2022) search is used on Wikisource, instead of showing a blank image for every page, we could show the user the cover page for the book.

Given that some Wikisources (en, bn, fr definitely do) already mark the cover page of the book in their index pages, it would make sense to use these as thumbnails for books.

Screenshot from 2022-11-22 18-18-03.png (1,397×716 px, 208 KB)

Use case(s) (list the steps that you performed to discover that problem, and describe the actual underlying problem which you want to solve. Do not describe only a solution):

  • Making it easier to distinguish different editions of the same book with different cover art.

Benefits (why should this be implemented?):

  • From a reader's POV it makes the experience of finding a book a lot simpler.
  • A editor might get confused between two editions of the same book, this would make it harder to make that mistake (especially if the cover art is different)

Event Timeline

Jdlrobson moved this task from Incoming to Tracking (Waiting on other teams) on the Vector 2022 board.
Jdlrobson subscribed.

This is actually an issue with the PageImages extension. Presumably that needs to be updated to be aware of Wikisource covers https://en.wikisource.org/wiki/War_and_Peace_(Tolstoy)?action=info

Jdlrobson renamed this task from Use cover page/image as thumbnail for search to PageImages does not consider Wikisource page covers.Nov 28 2022, 5:48 PM
Jdlrobson renamed this task from PageImages does not consider Wikisource page covers to PageImages does not consider Wikisource page covers when selecting image.

Question. Is there ANYTHING other than convention, which indicates that something is the cover of a book ? I currently see.. a cover is defined by the community as:

  • The first page transclusion on a book page,
  • should be the front cover,
  • which is a transcribed version of the book cover
  • generally this page has a an associated image, which is the n’th page of a upload

I’m seeing several problems that have to be solved.

  • page images has no concept of or way to address, a page within a multi page files right now
  • covers of books are not actually images transcluded on the book pages (they are transcribed wiki pages, which might look like a cover image)
  • page images of books are hard to guess (conventions fail)

There is no data that says: “book x: cover is page y of z” ? Hmm. The Index page seems to have a field for cover, but then the book page, doesn’t seem to have a back reference to the index in a reliable way ?

I think it were best if there is a hook for page images, that the Wikisource extension can hook into, to provide the image to page images extension directly, but even then it might not be simple.

On enWS the {{header}} template has a cover parameter used by the ws-export tool for just this purpose, which, provided PageImages can pick up signals or microdata in rendered markup, could be used for this. The alternative would almost have to be making PageImages ask Proofread Page to provide the image. I don't think there is any sane algorithm that PageImages could use to try to determine it on its own with any kind of accuracy.

Mainspace pages are connected to the Index: page through the "Source" tab at the top of the page, and by the non-prefixed page name in the <pages tag, both of which are Proofread Page-specific (it adds the former based on the latter). The downside is that not all mainspace pages are scan-backed (do not use the Proofread Page extension facilities). On enWS this is mainly legacy texts, but some Wikisourcen do not use it at all. I think ignoring those and only supporting cover images when Proofread Page is used is reasonable.

I was just making some tests here and even by adding class=pageimage on the thumbnail syntax, PageImages doesn't pick the cover. But there it is, a JPG thumbnail of the desired cover, just waiting to be picked up...

But it turns out the PageImages extension isn't even installed in Wikisource??? I will need to open a new ticket

So, this apparently runs deeper...

12 years ago, somebody opened T68455. Then, the next day the disabled PageImages for every Wikisource and Wikibooks without any consult! And just a decade later he.wikisource (T362851) and this year bn.wikisource (T416800) opted in, but this didn't attract awareness in the wide Wikisource community.

Please, simply revert this hasty and unconsulted decision.

Ironically, they missed multilingual Wikisource.

Note PageImages is now enable on all Wikisource and Wikibooks project (T417538).

If I head to https://en.wikisource.org/wiki/Main_Page and at the top search for War and P, the entry for the Tolstoi does have an icon (unlike the screenshot in this task description). Thus presumably this is fixed?

No reply; assuming this is resolved also in my understanding