Fix CDATA state when it ends with more than two square brackets.

gsnedders · gsnedders · commit 4213ee24bee2 · 2013-01-06T22:07:09.000Z
diff --git a/html5lib/tokenizer.py b/html5lib/tokenizer.py
@@ -1716,21 +1716,18 @@ def cdataSectionState(self):
         data = []
         while True:
             data.append(self.stream.charsUntil("]"))
-            charStack = []
-
-            for expected in ["]", "]", ">"]:
-                charStack.append(self.stream.char())
-                matched = True
-                if charStack[-1] == EOF:
-                    data.extend(charStack[:-1])
-                    break
-                elif charStack[-1] != expected:
-                    matched = False
-                    data.extend(charStack)
+            data.append(self.stream.charsUntil(">"))
+            char = self.stream.char()
+            if char == EOF:
+                break
+            else:
+                assert char == ">"
+                if data[-1][-2:] == "]]":
+                    data[-1] = data[-1][:-2]
                     break
+                else:
+                    data.append(char)
 
-            if matched:
-                break
         data = "".join(data)
         #Deal with null here rather than in the parser
         nullCount = data.count("\u0000")
