Try HTML::Parser. using regexes to parse HTML is a nightmare as many will attest to, and HTML::Parser is usually the defacto reccommendation around here when you've got HTML to munge.

HTH