awesome-python
diff --git a/‎examples/sanitizer/sanitizer.py
Lines changed: 3 additions & 1 deletion b/‎examples/sanitizer/sanitizer.py
Lines changed: 3 additions & 1 deletion
diff --git a/‎src/treebuilders/dom.py
Lines changed: 34 additions & 6 deletions b/‎src/treebuilders/dom.py
Lines changed: 34 additions & 6 deletions
diff --git a/‎src/treebuilders/simpletree.py
Lines changed: 1 addition & 1 deletion b/‎src/treebuilders/simpletree.py
Lines changed: 1 addition & 1 deletion
@@ -118,7 +118,9 @@ def _sanitizeTree(self, tree, escapeRemovedMarkup):
         return tree
 
     def acceptableURI(self, uri):
-        return urlparse.urlparse(uri)[0] in self.acceptable_schemes
+        #This is wrong
+        parsedURI = urlparse.urlparse(uri) 
+        return parsedURI[0] in self.acceptable_schemes or not parsedURI[0]
 
     def nodeToText(self, node, endTag=False):
         """Create an unescaped text node containing a serialization of node's
 
@@ -1,6 +1,8 @@
 import _base
 from xml.dom import minidom, Node, XML_NAMESPACE, XMLNS_NAMESPACE
 import new
+from xml.sax.saxutils import escape
+from constants import voidElements
 
 import re
 illegal_xml_chars = re.compile("[\x01-\x08\x0B\x0C\x0E-\x1F]")
@@ -86,9 +88,9 @@ def elementClass(self, name):
         return NodeBuilder(self.dom.createElement(name))
 
     def commentClass(self, data):
-        return NodeBuilder(self.dom.createComment(data))
-    
-    def fragmentClass(self):
+        return NodeBuilder(self.dom.createComment(data))
+    
+    def fragmentClass(self):
         return NodeBuilder(self.dom.createDocumentFragment())
 
     def appendChild(self, node):
@@ -98,9 +100,9 @@ def testSerializer(self, element):
         return testSerializer(element)
 
     def getDocument(self):
-        return self.dom
-    
-    def getFragment(self):
+        return self.dom
+    
+    def getFragment(self):
         return _base.TreeBuilder.getFragment(self).element
 
     def insertText(self, data, parent=None):
@@ -143,6 +145,32 @@ def serializeElement(element, indent=0):
 
     return "\n".join(rv)
 
+class HTMLSerializer(object):
+    def serialize(self, node):
+        rv = self.serializeNode(node)
+        for child in node.childNodes:
+            rv += self.serialize(child)
+        if node.nodeType == Node.ELEMENT_NODE and node.nodeName not in voidElements:
+            rv += "</%s>\n"%node.nodeName
+        return rv
+    
+    def serializeNode(self, node):
+        if node.nodeType == Node.TEXT_NODE:
+            rv = node.nodeValue
+        elif node.nodeType == Node.ELEMENT_NODE:
+            rv = "<%s"%node.nodeName
+            if node.hasAttributes():
+                rv = rv+"".join([" %s='%s'"%(key, escape(value)) for key,value in
+                                 node.attributes.items()])
+            rv += ">"
+        elif node.nodeType == Node.COMMENT_NODE:
+            rv = "<!-- %s -->" % escape(node.nodeValue)        
+        elif node.nodeType == Node.DOCUMENT_TYPE_NODE:
+            rv = "<!DOCTYPE %s>" % node.name
+        else:
+            rv = ""
+        return rv
+
 def dom2sax(node, handler, nsmap={'xml':XML_NAMESPACE}):
   if node.nodeType == Node.ELEMENT_NODE:
     if not nsmap:
 
@@ -210,7 +210,7 @@ def serializeNode(self, node):
                                  node.attributes.iteritems()])
             rv += ">"
         elif node.type == CommentNode.type:
-            rv = "<!-- %s -->" % escape(self.data)        
+            rv = "<!-- %s -->" % escape(node.data)        
         elif node.type == DocumentType.type:
             rv = "<!DOCTYPE %s>" % node.name
         else: