User talk:Yurik/Query API/Postponed issues

From Wikipedia, the free encyclopedia

Contents 1 Existing thumbnails 2 Get Page Summary 3 Revision sizes 4 Move to Meta

[edit] Existing thumbnails

Could the imageinfo query be augmented (or a new query added) to return a list of existing thumbnails for a given image? It'd also be nice if there was a mechanism for requesting that the server generate a new thumbnail of a particular size and only returning output when that process was complete. (This could be expensive, of course, but it's nothing that cannot be done now by submitting a dummy preview). Lupin|talk|popups 03:30, 14 June 2006 (UTC)

I'm not sure how i would delay the response. I guess it should be possible... Admins will kill me for that though :) On IRC, brion said its not possible to get the list of already rendered images. So i guess the whole thing is postponned. --Yurik 06:35, 14 June 2006 (UTC)

[edit] Get Page Summary

It would be very useful to get the page summary -- first several paragraphs of the text instead of the entire content. Currently, Lupin's popup use this algorithm to show just that. Note that this is language dependent. (Written by Lupin at his talk page):

1. locate the target heading (for a link like User:Lupin#Other things) and delete everything before it
2. remove HTML comments
3. remove HTML divs
4. remove galleries
5. remove box templates like infobox, elementbox, frame*, floats, *box_start and so on
6. remove all templates, or just templates which span several lines, depending on user preferences
7. remove tables, wikitext and HTML
8. remove all images
9. remove chunks of italic text, as are used to alert the reader to dab pages or other tangential remarks. This is not an exact science!
10. mopup: remove horizontal rules, all indented lines, __TOC__ and friends
11. The remaining text is what we work with. Figure out how much to preview based on user prefs (only return whole sentences - again, English language dependency here). Note that some users prefer longer previews than others, sometimes spanning several paragraphs.
12. remove unsightly whitespace from what remains (this shouldn't really be needed - it's an artifact of sloppy whitespace removal in previous steps)
13. convert to HTML, passing any templates straight through
14. remove any remaining templates which are still "too long".

--Yurik 23:50, 9 June 2006 (UTC)

[edit] Revision sizes

From IRC:
pgk> vandalism detection:
in the recent changes, given older revision X and new revision Y, search X for words a,b,c (given as a query parameter); do the same for Y, and compare if the list of words found in X is smaller than in Y.
Return revision size. --Yurik 17:18, 8 June 2006 (UTC)

This requires database support for efficient implementation. --Yurik 14:16, 26 July 2006 (UTC)

[edit] Move to Meta

Since this is installed globally on all Wikimedia websites, wouldn't it be more appropriate if this was moved to Meta? — Edward Z. Yang^(Talk) 02:06, 13 July 2006 (UTC)

I would rather keep it here until the query API gets stabilized. Once most of the internals gets sorted out (month or two), it should be moved to the meta, thus declaring it as an endorsed feature. --Yurik 22:20, 13 July 2006 (UTC)

Perfectly reasonable. Great work on the query API, just judging from its popularity, a lot of people have been waiting for this feature. — Edward Z. Yang^(Talk) 00:24, 14 July 2006 (UTC)