See how htmltreexpather.pl does it, it uses this to parse "xml"

            $tree->implicit_tags(0);
            $tree->no_expand_entities(1);
            $tree->ignore_unknown(0);
            $tree->ignore_ignorable_whitespace(0);
            $tree->no_space_compacting(1);
            $tree->store_comments(1);
            $tree->store_pis(1);
[download]