1)
A shareware, open-source C program named "Tidy" lives at the venerable W3C, at http://www.w3.org/People/Raggett/tidy/

It is written by Dave Ragget (the editor of the HTML 4 spec). Is very current, sophisticated, to some extent handles XML, cascading style sheets, ASP and PHP as well. Configurable, is ported/tested on just about any current platform, plus a Java port. I haven't tried it yet, but i'm gonna!

2)
MicroSoft has seen the light ..
yeah right!
Microsoft decided to get in on the action, after several other folks started incorporating "clean up MS2000 mess" tools into their own apps, and has a little program for it at the MS website. http://officeupdate.microsoft.com/2000/downloadDetails/Msohtmf2.htm

It works, if you check *all* the options in the configuration menu. It still leaves cell size, font size, and cell color info though, which you may want, or you can get rid of with a regex. A co-worker tells me an option in DreamWeaver will remove cell size info.