libxml2's parser (which lxml uses) was written before HTML 5. It generally works quite well, but I've found a bug or two in it, and it doesn't get updated much

libxml2's parser (which lxml uses) was written before HTML 5. It generally works quite well, but I've found a bug or two in it, and it doesn't get updated much. For instance, weird things can happen with tags after the </body> tag (I think maybe you can get a second body or head element). lxml can't parse Google's homepage as another example, but it's an extreme example.