I think you've painted yourself into a corner. Monster regexes can be constructed to parse HTML and tables, but they are quite difficult to perfect. As you have noted, HTML::TableExtract is a far superior way to what you have tried to create.

The HTML::TableExtract module is surprisingly simple to use. You really should give it a shot, as it will likely take less time to figure out how to use it than it would to diagnose your problem.

The reason why it may not match and return false is because it's spinning trying to find a match. Maybe there's a billion different ways to try and get a match, and it will investigate them all just to be sure. Doing a series of mini-matches is much better, and using HTML::Parser is better yet.