importhtml5libparser=html5lib.HTMLParser(tree=html5lib.treebuilders.getTreeBuilder("lxml"),namespaceHTMLElements=False)serializer=html5lib.serializer.HTMLSerializer(omit_optional_tags=False)walker=html5lib.treewalkers.getTreeWalker("lxml")# workssrc=u"experiences"tree=parser.parseFragment(src,container="div")stream=walker(tree)output=serializer.serialize(stream)print("\n".join(output))# Doesn't worksrc=u"exp\xe9riences"tree=parser.parseFragment(src,container="div")stream=walker(tree)output=serializer.serialize(stream)print("\n".join(output))

defensure_str(s):ifsisNone:returnNoneelifisinstance(s,text_type):returnselse:returns.decode("utf-8","strict")classFragmentWrapper(object):def__init__(self,fragment_root,obj):        ...self.isstring=isinstance(obj,str)orisinstance(obj,bytes)# Support for bytes here is Py2ifself.isstring:self.obj=ensure_str(self.obj)

defensure_str(s):ifsisNone:returnNoneelifisinstance(s,text_type):returnselse:returns.decode("utf-8","strict")classFragmentWrapper(object):def__init__(self,fragment_root,obj):        ...self.isstring=isinstance(obj,str)orisinstance(obj,bytes)orisinstance(obj,text_type)# Support for bytes here is Py2ifself.isstring:self.obj=ensure_str(self.obj)