Once I'd found where the databases are kept the DB bit didn't take too long: the main challenge was the fact that the structure of the database is really wacky. The annotation table goes something like | ID | Page Number | Binary field containing a bunch of annotation data for that page | - so you can't easily extract just the information you're interested in using SQL.
Actually extracting the information from the DB probably took less than an hour, but working out a reliable way to extract just the annotation text from the binary field probably a smidge under another half day. Then probably about half a day to write the script. So about a day total, probably.
edit: plus an embarrassingly long time to figure out where the database was actually stored. Because the file path is so similar to where the books are stored, I kinda skipped over it when looking, and ended up scratching my head for a while.
Last edited by Zaltys; 05-12-2012 at 03:07 AM.