The ElementTidy parser can read (almost) arbitrary HTML files, and turn them into well-formed element trees. This parser uses a library version of Dave Raggett's HTML Tidy utility to fix any problems with the HTML before converting it to XHTML (the XML version of HTML).