Re-assign filteredEntityList as charStack grows.

gsnedders · gsnedders · commit 587483d90877 · 2012-04-18T23:47:33.000+01:00
There's no way entities that failed to match the prefix will match in
future, so remove them. 6% perf gain parsing the spec.
diff --git a/html5lib/tokenizer.py b/html5lib/tokenizer.py
@@ -184,8 +184,10 @@ def consumeEntity(self, allowedChar=None, fromAttribute=False):
             def entitiesStartingWith(name):
                 return [e for e in filteredEntityList if e.startswith(name)]
 
-            while (charStack[-1] is not EOF and
-                   entitiesStartingWith("".join(charStack))):
+            while (charStack[-1] is not EOF):
+                filteredEntityList = entitiesStartingWith(u"".join(charStack))
+                if not filteredEntityList:
+                    break
                 charStack.append(self.stream.char())
 
             # At this point we have a string that starts with some characters
