HTML tables are either well-formed or not. If they are well-formed
then you can load them in an XML Document. If they are not well
formed, then you'll have to parse them by hand.
Darren from RegexLib.com is working on a resilient HTML parser, so
you might head on over there and check out some of his code. Don't
feel bashful to ask him
for what he is doing, since he might not have it posted readily on
the site.