Don't let the lxml treewalker walk above the fragment root

gsnedders · gsnedders · commit 270a2ca14faf · 2016-01-12T23:24:18.000+01:00
diff --git a/html5lib/treewalkers/lxmletree.py b/html5lib/treewalkers/lxmletree.py
@@ -118,8 +118,10 @@ def __len__(self):
 class TreeWalker(_base.NonRecursiveTreeWalker):
     def __init__(self, tree):
         if hasattr(tree, "getroot"):
+            self.fragmentChildren = set()
             tree = Root(tree)
         elif isinstance(tree, list):
+            self.fragmentChildren = set(tree)
             tree = FragmentRoot(tree)
         _base.NonRecursiveTreeWalker.__init__(self, tree)
         self.filter = ihatexml.InfosetFilter()
@@ -197,5 +199,7 @@ def getParentNode(self, node):
             if key == "text":
                 return node
             # else: fallback to "normal" processing
+        elif node in self.fragmentChildren:
+            return None
 
         return node.getparent()
