lxml gives text/tail as bytes under Py2, so accomodate for this.

gsnedders · gsnedders · commit a32418c6d251 · 2013-01-09T02:09:34.000Z
diff --git a/html5lib/treebuilders/etree_lxml.py b/html5lib/treebuilders/etree_lxml.py
@@ -11,6 +11,7 @@
 
 import warnings
 import re
+import sys
 
 from . import _base
 from html5lib.constants import DataLossWarning
@@ -71,8 +72,9 @@ def serializeElement(element, indent=0):
                 while next_element is not None:
                     serializeElement(next_element, indent+2)
                     next_element = next_element.getnext()
-            elif isinstance(element, str):
+            elif isinstance(element, str) or isinstance(element, bytes):
                 #Text in a fragment
+                assert isinstance(element, str) or sys.version_info.major == 2
                 rv.append("|%s\"%s\""%(' '*indent, element))
             else:
                 #Fragment case
@@ -84,6 +86,7 @@ def serializeElement(element, indent=0):
             if hasattr(element, "tail") and element.tail:
                 rv.append("|%s\"%s\"" %(' '*indent, element.tail))
         else:
+            assert isinstance(element, etree._Element)
             nsmatch = etree_builders.tag_regexp.match(element.tag)
             if nsmatch is not None:
                 ns = nsmatch.group(1)
diff --git a/html5lib/treewalkers/lxmletree.py b/html5lib/treewalkers/lxmletree.py
@@ -2,6 +2,7 @@
 from html5lib.treebuilders.etree import tag_regexp
 
 from gettext import gettext
+import sys
 _ = gettext
 
 from . import _base
@@ -71,7 +72,8 @@ def __init__(self, fragment_root, obj):
             self.tail = self.obj.tail
         else:
             self.tail = None
-        self.isstring = isinstance(obj, str)
+        self.isstring = isinstance(obj, str) or isinstance(obj, bytes)
+        assert not self.isstring or isinstance(obj, str) or sys.version_info.major == 2
         
     def __getattr__(self, name):
         return getattr(self.obj, name)
