Using regexps to parse "delimited text" (such as HTML) is a bit of a fool's errand. HTML::TokeParser would be good for this. Or better yet Text::Balanced::Knockers.