Often "talking to the teddy bear" is all that's needed to sort out your problem, as happened in this case. Thanks for updating with the solution you discovered!

Note that as a general thing there are stacks of modules in CPAN for parsing and manipulating HTML. If you are working on this to gain a tool rather than just as an educational exercise it may be worth the time to investigate a few of the most useful looking modules. HTML is tricky stuff to parse so using a module that gets a reasonable amount of love from its author and is widely used by others is likely to save you a lot of grief!