It allows for editing meta tags, title tag, title attribute in anchor tags, and alt attributes in image tags. That part works fine and HTML::TreeBuidler makes it fairly simple. The problem lies in the reconstruction of pages with non compliant HTML.

I have looked at HTML::Parser and HTML::TokeParser, but they don't do the heavy lifting that HTML::TreeBuilder does for me. I am trying to avoid reinventing the wheel by using HTML::TreeBuilder. According to the change log HTML::TreeBuilder there was a known bug with handling out of place information inside of table, but doubt if they invisioned anything as bad as non table tags outside of th, tr, caption, td, etc. tags within a table.