Why the Lucky Stiff has a great parser, hpricot http://code.whytheluckystiff.net/hpricot/
If you need to follow links or fill out forms as well, the trunk of mechanize can use hpricot as it’s parser. Deadly combo!
joshua
Google for Rubyful Soup - it's a port (by the original author) of the excellent Python parser "Beautiful Soup", which is explicitly designed to deal with messy, badly-formed, awkward HTML - ie, the real-world examples of it.