from html5lib import HTMLParser, sanitizer, serializer, treebuilders, treewalkersopts = {}opts['tokenizer'] = sanitizer.HTMLSanitizerparser = html5lib.HTMLParser(tree=treebuilders.getTreeBuilder("dom"), **opts)data = u'Hello World<img src="/static/cms/img/icons/plugins/link.png" alt="Link - A Link" title="Link - A Link" />'dom_tree = parser.parseFragment(data)walker = treewalkers.getTreeWalker("dom")stream = walker(dom_tree)s = serializer.htmlserializer.HTMLSerializer()text = ''.join(s.serialize(stream))assert text == data