Use None to represent missing state for doctype name in tokenizer

gsnedders · gsnedders · commit e170e3a1a9dc · 2013-07-20T19:06:04.000+01:00
Previously, the empty string was used, which should never occur per
spec. This moves us to being more closely in line with the spec.
diff --git a/html5lib/tokenizer.py b/html5lib/tokenizer.py
@@ -1134,7 +1134,7 @@ def markupDeclarationOpenState(self):
                     break
             if matched:
                 self.currentToken = {"type": tokenTypes["Doctype"],
-                                     "name": "",
+                                     "name": None,
                                      "publicId": None, "systemId": None,
                                      "correct": True}
                 self.state = self.doctypeState
diff --git a/html5lib/treebuilders/etree.py b/html5lib/treebuilders/etree.py
@@ -161,7 +161,8 @@ def _setData(self, value):
     class DocumentType(Element):
         def __init__(self, name, publicId, systemId):
             Element.__init__(self, "<!DOCTYPE>")
-            self._element.text = name
+            assert name is None or len(name) > 0
+            self._element.text = name if name else ""
             self.publicId = publicId
             self.systemId = systemId
 
