On 12/29/05, Rob Church robchur@gmail.com wrote:
Undelete the pages; yes, undelete. Then nip into the page table and find out the page id...delete that page's row from the page table. Hop over to the revisions table and delete all rows where rev_page is the same as that page id. That's a permanent deletion; the page never existed.
For the small touches, check that corresponding revisions are wiped from the recent changes table, too.
Is there wiki-side documentation on spelunking around the database like this? I don't know where to begin looking. (the mediawiki hacking topics?)
This looks like it's only good for totally nuking a complete page, can it be varied to delete the history of a particular page and leave the the present revision intact? I understand that it might not work that way now (with revisions being tracked, and not complete page copies).
This stuff wouldn't work for bulk stuff as-is, but if it's simple enough I could pull off some Ruby scripting it if there were some nice mysql tools for Ruby (I'm not sure). That, and I know some people I can pay to do it and test it if need be. =)