Is there a reason why you don't want to use TokeParse, or another CPAN module that does HTML parsing?

I mean, it can't be because you can do better (otherwise, you wouldn't have failed).

Abigail