I think it returns an XML as default. I want to make the query as smaller as possible. I'm only interested in new additions, as far as I could see, new additions are inside <td class="diff-addedline"> and there in <span class="diffchange">. At least this is true in the non-API query. I manage to get the diff output with the API and those tags are represented in this way: http://pastebin.com/Myv1976Y this is for XMLFM (default output). API DIFF: http://es.wikipedia.org/w/api.php?action=query&prop=revisions&rvsta… NON-API DIFF: http://es.wikipedia.org/w/index.php?diff=37372500&oldid=32780367&di… I think this is not yet implemented in the framework. Matias.

Look into site.py:loadrevisions for directions on how to do it. When is the query actually been executed? Since I'm kind of lost in the code. I want to see how loadrevisions executes the query and manages the results. I've tried to "magically" make it work with this UAH patch ( http://pastebin.com/rhg9n7NM) for the site.py but doesn't seem to work as easy as I wished. Any hint would be great ! --Matias.

On Tue, May 25, 2010 at 05:14:18PM -0300, Matias wrote: The 'problem' of loadrevisions() is that it doesn't return anything, it just loads revisions (sic) and store them in the _revisions attribute of the Page() object, which is a dictionnary of Revision() objects (see update_page() in api.py). And these Revision() objects are not meant to store diffs. My suggestion would be to directly use a PropertyGenerator and handle the result yourself instead of parsing it via update_page. Which will be almost the same thing you've shown on your pastebin link, with the last part (api.update_page(...)) removed and replaced by something useful for you. stan.