Mostly done, code uploaded

Aside from bugs, I'm done with the parsing and database part of the code. I have completely switched the database to Berkeley DB from SQL, making it around 300 times faster. There are still some issues with Unicode and link formats that I need to work out. Debugging is slow because each time I make a correction I have to wait for it to re-parse. Soon I will be starting work on a algorithm that will find related articles. The latest code has been uploaded to svn.