A collection of non-elegant regexps for doing exactly this can be found in
http://tclwebtest.sf.net .
When you are using aolserver 4 it is fairly easy to use tclwebtest from within with 'package require'. Otherwise it becomes tricky - one way I made it work is by calling tcl's 'source' before calling tclwebtest commands, but this is propably not very efficient.
With aolserver 4 you might want to use http://tcllib.sourceforge.net/doc/htmlparse.html instead.
Or use tdom - it has a html parsing mode, although it is not very forgiving on broken html, e.g. it crashes on things like non-closed tables etc. And I'm not sure if the html mode is available in the aolserver module.
