The thread "Hotlinking images from wikipedia ?" was specially interesting for me.

I'm trying to keep a local mirror of wikipedia (it is the second most accessed website from my faculty, second only to google). My idea was/is to download and insert the databases (done), and to download and keep a local cache of the images on demand. (and then find a way to update both the images and the databases).

Is there an easy way to download the images when they are not found locally? I'm willing to start poking at the code if it is not already done, but I'd prefer to diverge as little as possible from the original mediawiki code.

Cheers,