regression test for tidy

I'm "fooling" around with the source code for HTML tidy for my own pleasure/education. To ensure I don't break anything I use a regression test based on an html file and diffing the output from the original version with the output from the one I am fooling with.

I just wondered if anyone had some "better" input html files that were chosen to really excercise the various features of the HTMLTidy, as these would provide a more thorough regression test. And if so, if I could have access to them.

Thanks for any assistance.