Fix parse errors for < in attributes

jgraham · jgraham · commit 1ad89ed72c6f · 2009-07-12T17:43:52.000+02:00
diff --git a/src/html5lib/tokenizer.py b/src/html5lib/tokenizer.py
@@ -455,7 +455,7 @@ def beforeAttributeNameState(self):
             self.emitCurrentToken()
         elif data == u"/":
             self.state = self.selfClosingStartTagState
-        elif data == u"'" or data == u'"' or data == u"=":
+        elif data in (u"'", u'"', u"=", u"<"):
             self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
               "invalid-character-in-attribute-name"})
             self.currentToken["data"].append([data, ""])
@@ -488,7 +488,7 @@ def attributeNameState(self):
             self.state = self.afterAttributeNameState
         elif data == u"/":
             self.state = self.selfClosingStartTagState
-        elif data == u"'" or data == u'"':
+        elif data in (u"'", u'"', u"<"):
             self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
               "invalid-character-in-attribute-name"})
             self.currentToken["data"][-1][0] += data
@@ -532,7 +532,7 @@ def afterAttributeNameState(self):
             self.state = self.attributeNameState
         elif data == u"/":
             self.state = self.selfClosingStartTagState
-        elif data == u"'" or data == u'"':
+        elif data in (u"'", u'"', u"<"):
             self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
               "invalid-character-after-attribute-name"})
             self.currentToken["data"].append([data, ""])
@@ -561,7 +561,7 @@ def beforeAttributeValueState(self):
             self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
               "expected-attribute-value-but-got-right-bracket"})
             self.emitCurrentToken()
-        elif data == u"=":
+        elif data in (u"=", u"<"):
             self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
               "equals-in-unquoted-attribute-value"})
             self.currentToken["data"][-1][1] += data
