Changeset 178:37f36ea93838

Timestamp:

Jan 12, 2013, 4:40:38 PM (9 years ago)

Author:

Ralph Meijer <ralphm@…>

Branch:

default

Message:

Properly encode IDN domain names for SRV lookups.

Before Twisted 12.3.0, the SRV lookup done for outgoing
connections succeeded if passed a unicode string with all-ASCII
code points. A recent change made the DNS code more strict, and
only byte string are accepted as domain name. See
​http://twistedmatrix.com/trac/ticket/6245 for details.

This change makes sure domain names are encoded to their ASCII
Compatible Equivalent (ACE) version before passing the resulting
byte string to twisted.names.srvconnect.SRVConnector, as per
RFC 3490.

Note that while connecting to servers with an IDN domain name
now works properly, authentication using MD5-Digest SASL mechanism
will fail until ​http://twistedmatrix.com/trac/ticket/5066 has been
resolved.

Fixes: #77.

Location:

wokkel

Files:

• client.py (modified) (2 diffs)
• generic.py (modified) (2 diffs)
• server.py (modified) (2 diffs)
• test/test_client.py (modified) (2 diffs)
• test/test_generic.py (modified) (1 diff)

wokkel/client.py

@@ -88 +88 @@
     def __init__(self, jid, password, host=None, port=5222):
         self.jid = jid
-        self.domain = jid.host
+        self.domain = generic.prepareIDNName(jid.host)
         self.host = host
         self.port = port
@@ -183 +183 @@
 
 def clientCreator(factory):
-    domain = factory.authenticator.jid.host
+    domain = generic.prepareIDNName(factory.authenticator.jid.host)
     c = XMPPClientConnector(reactor, domain, factory)
     c.connect()

wokkel/generic.py

@@ -7 +7 @@
 Generic XMPP protocol helpers.
 """
+
+from encodings import idna
 
 from zope.interface import implements
@@ -328 +330 @@
     def clientConnectionFailed(self, connector, reason):
         self.deferred.errback(reason)
+
+
+
+def prepareIDNName(name):
+    """
+    Encode a unicode IDN Domain Name into its ACE equivalent.
+
+    This will encode the domain labels, separated by allowed dot code points,
+    to their ASCII Compatible Encoding (ACE) equivalent, using punycode. The
+    result is an ASCII byte string of the encoded labels, separated by the
+    standard full stop.
+    """
+    result = []
+    labels = idna.dots.split(name)
+
+    if labels and len(labels[-1]) == 0:
+        trailing_dot = b'.'
+        del labels[-1]
+    else:
+        trailing_dot = b''
+
+    for label in labels:
+        result.append(idna.ToASCII(label))
+
+    return b'.'.join(result) + trailing_dot

wokkel/server.py

@@ -30 +30 @@
 from twisted.words.xish import domish
 
-from wokkel.generic import DeferredXmlStreamFactory, XmlPipe
+from wokkel.generic import DeferredXmlStreamFactory, XmlPipe, prepareIDNName
 
 NS_DIALBACK = 'jabber:server:dialback'
@@ -469 +469 @@
 
 def initiateS2S(factory):
-    domain = factory.authenticator.otherHost
+    domain = prepareIDNName(factory.authenticator.otherHost)
     c = XMPPServerConnector(reactor, domain, factory)
     c.connect()

wokkel/test/test_client.py

@@ -41 +41 @@
         xs.dispatch(xs, xmlstream.STREAM_AUTHD_EVENT)
         self.assertEquals(JID('user@example.org/test'), self.client.jid)
+
+
+    def test_domain(self):
+        """
+        The domain to connect to is a byte string derived from the JID host.
+        """
+        self.assertIsInstance(self.client.domain, bytes)
+        self.assertEqual(b'example.org', self.client.domain)
 
 
@@ -143 +151 @@
         def cb(connector):
             self.assertEqual('xmpp-client', connector.service)
-            self.assertEqual('example.org', connector.domain)
+            self.assertIsInstance(connector.domain, bytes)
+            self.assertEqual(b'example.org', connector.domain)
             self.assertEqual(factory, connector.factory)
 

wokkel/test/test_generic.py

@@ -269 +269 @@
         """
         self.assertIdentical(None, self.request.timeout)
+
+
+
+class PrepareIDNNameTests(unittest.TestCase):
+    """
+    Tests for L{wokkel.generic.prepareIDNName}.
+    """
+
+    def test_bytestring(self):
+        """
+        An ASCII-encoded byte string is left as-is.
+        """
+        name = b"example.com"
+        result = generic.prepareIDNName(name)
+        self.assertEqual(b"example.com", result)
+
+
+    def test_unicode(self):
+        """
+        A unicode all-ASCII name is converted to an ASCII byte string.
+        """
+        name = u"example.com"
+        result = generic.prepareIDNName(name)
+        self.assertEqual(b"example.com", result)
+
+
+    def test_unicodeNonASCII(self):
+        """
+        A unicode with non-ASCII is converted to its ACE equivalent.
+        """
+        name = u"\u00e9chec.example.com"
+        result = generic.prepareIDNName(name)
+        self.assertEqual(b"xn--chec-9oa.example.com", result)
+
+
+    def test_unicodeHalfwidthIdeographicFullStop(self):
+        """
+        Exotic dots in unicode names are converted to Full Stop.
+        """
+        name = u"\u00e9chec.example\uff61com"
+        result = generic.prepareIDNName(name)
+        self.assertEqual(b"xn--chec-9oa.example.com", result)
+
+
+    def test_unicodeTrailingDot(self):
+        """
+        Unicode names with trailing dots retain the trailing dot.
+
+        L{encodings.idna.ToASCII} doesn't allow the empty string as the input,
+        hence the implementation needs to strip a trailing dot, and re-add it
+        after encoding the labels.
+        """
+        name = u"example.com."
+        result = generic.prepareIDNName(name)
+        self.assertEqual(b"example.com.", result)