Prefix the bufferedCharacter, don't suffix it. Also, file objects return the empty string at EOF.

gsnedders · gsnedders · commit dc827e85c1a8 · 2010-08-01T13:21:03.000+02:00
diff --git a/html5lib/inputstream.py b/html5lib/inputstream.py
@@ -347,10 +347,7 @@ def readChunk(self, chunkSize=None):
         
         #Deal with CR LF and surrogates broken across chunks
         if self._bufferedCharacter:
-            if data:
-                data = data + self._bufferedCharacter
-            else:
-                data = self._bufferedCharacter
+            data = self._bufferedCharacter + data
             self._bufferedCharacter = None
         elif not data:
             # We have no more data, bye-bye stream
