Fixed recent regression of recent minor charsUntil optimisation

philiptaylor · philiptaylor · commit 22886b16a256 · 2008-12-18T17:07:14.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%401245
diff --git a/src/html5lib/inputstream.py b/src/html5lib/inputstream.py
@@ -343,15 +343,20 @@ def charsUntil(self, characters, opposite = False):
             # Find the longest matching prefix
             m = chars.match(self.chunk, self.chunkOffset)
             if m is None:
-                end = self.chunkOffset
+                # If nothing matched, and it wasn't because we ran out of chunk,
+                # then stop
+                if self.chunkOffset != self.chunkSize:
+                    break
             else:
                 end = m.end()
-            # If not everything matched, return everything up to the part that didn't match
-            if end != self.chunkSize:
-                rv.append(self.chunk[self.chunkOffset:end])
-                self.chunkOffset = end
-                break
-            # If the whole chunk matched, use it all and read the next chunk
+                # If not the whole chunk matched, return everything
+                # up to the part that didn't match
+                if end != self.chunkSize:
+                    rv.append(self.chunk[self.chunkOffset:end])
+                    self.chunkOffset = end
+                    break
+            # If the whole remainder of the chunk matched,
+            # use it all and read the next chunk
             rv.append(self.chunk[self.chunkOffset:])
             if not self.readChunk():
                 # Reached EOF
