Since I'm not anxious to register on some website just so I can see what on earth the tables you're talking about look like, how about trying to give a semi-technical description for us? You said they're not regular HTML tables, so what are they? Are they graphic images? Are they javascript entities? Are they PDF files?

And since you've already tried everything you can think of, can you tell us what you thought of thus far, so that we know where you've already invested time? While you're at it, you might also let us know in what way your attempts fell short of meeting the need.

Since I don't know better, I'll suggest that most websites worth their weight in salt will also be lynx-friendly. That being the case, perhaps the easiest way to get at the data from the tables in question is to parse the all-text output from lynx. It's easy to grab the output from lynx. ..of course this assumes you're on a linux/unix type system. In this way, you can use the robustness of lynx -- a full-fledged text-based browser capable of handling cookies, and all sorts of curve-balls -- to intelligently dump the site to text.