Requests for comment/Image information

This proposal was obsoleted by Extension:CommonsMetadata and is now abandoned.

Request for comment (RFC)
Image information
Component General
Creation date
Author(s) MaxSem, Brion VIBBER, Yuvi Panda
Document status declined

We need a facility to access information from file description pages in a machine-readable format. Bits of information needed are:

  • License information
  • Image description, in user's preferred language is possible
  • Author information
  • Date image was produced
  • Source of the image
  • etc

Commons already has a rich metadata system for Images, Artwork and Books - we should use it.

Storage: page_props vs. dedicated table edit

  • Some fields (e.g. license information HTML) can be quite large - is it desireable to store them in page_props? It currently uses BLOB for pp_value which can store up to 64k.
  • Some information (e.g. descriptions) might come in several languages. Storage options include: page_props with all languages packed into one blob or having a dedicated column for language in a separtate table? MaxSem leans towards the former option.

Query requirements edit

In the long term it would be nice to be able to do general queries, like "give me all the images with CC-BY-compatible license that are photographs from 1970 to present within this geo box". In the short term we do not need this. :)

In the short term, we need to be able to request the data for a specific file by name. We should also be able to ask for specific set of properties only - 'get me property X, Y, Z for item File:ABC.jpeg', rather than only 'get me all properties of item File:ABC.jpeg'.

Exposing data from a shared repo edit

  • How?
  • How important it is, do we need to implement it from the very beginning?

Proposed API draft edit

api.php?action=query&prop=filemetadata&fmdprop=description|licenseurl&fmdlanguage=ru

  • Allow selecting information in several languages or provide several fallback languages in case information in requested language is not found? Both?

Extension name edit

Please suggest something, or the most stupid name will be chosen:P

See also edit