It's been a while since I used that module but if I recall correctly, it parses everything in to a token and the tokens not defined as an HTML tag should be defined as a text token.

Take a look at HTML::TokeParser help - parsing headlines and you'll see a quick program I wrote to dump an HTML page to tokenized output. Run that on your page and I think you'll see you don't need to do the regex per se, but rather need to check text tokens to find what you're after.

Good luck!

Update: Changed link from scratchpad to node as per suggestion by ww