use HTML::TokeParser. It really does a good job of parsing out links.

an example would be here

-p