Let me preface my reply by noting that I barely understand your question (but that never stopped me from blurting out something :-) ). I think that what you want is to parse the HTML contents with a tool like HTML::TokeParser, which has facilities for easily listing tag attributes.