Allow DOCTYPEs to round-trip, keeping public/system identifiers.

gsnedders · gsnedders · commit 9041a5dbc1f3 · 2009-10-17T12:23:54.000+02:00
diff --git a/src/html5lib/serializer/htmlserializer.py b/src/html5lib/serializer/htmlserializer.py
@@ -103,7 +103,23 @@ def serialize(self, treewalker, encoding=None):
         for token in treewalker:
             type = token["type"]
             if type == "Doctype":
-                doctype = u"<!DOCTYPE %s>" % token["name"]
+                doctype = u"<!DOCTYPE %s" % token["name"]
+                
+                if token["publicId"]:
+                    doctype += u' PUBLIC "%s"' % token["publicId"]
+                elif token["systemId"]:
+                    doctype += u" SYSTEM"
+                if token["systemId"]:                
+                    if token["systemId"].find(u'"') >= 0:
+                        if token["systemId"].find(u"'") >= 0:
+                            self.serializeError(_("System identifer contains both single and double quote characters"))
+                        quote_char = u"'"
+                    else:
+                        quote_char = u'"'
+                    doctype += u" %s%s%s" % (quote_char, token["systemId"], quote_char)
+                
+                doctype += u">"
+                
                 if encoding:
                     yield doctype.encode(encoding)
                 else:
diff --git a/tests/test_serializer.py b/tests/test_serializer.py
@@ -39,7 +39,12 @@ def __iter__(self):
                 for token in self.text(token[1]):
                     yield token
             elif type == "Doctype":
-                yield self.doctype(token[1])
+                if len(token) == 4:
+                    yield self.doctype(token[1], token[2], token[3])
+                elif len(token) == 3:
+                    yield self.doctype(token[1], token[2])
+                else:
+                    yield self.doctype(token[1])
             else:
                 raise ValueError("Unknown token type: " + type)
 
