Encode domain names using UTF-8 with UTS-46 canonicalization

See multiformats/multiaddr#94

multiaddr/codecs/idna.py → multiaddr/codecs/domain.py

@@ -10,8 +10,11 @@
 
 
 def to_bytes(proto, string):
-    return idna.encode(string, uts46=True)
+    return idna.uts46_remap(string).encode("utf-8")
 
 
 def to_string(proto, buf):
-    return idna.decode(buf)
+    string = buf.decode("utf-8")
+    for label in string.split("."):
+        idna.check_label(label)
+    return string

multiaddr/protocols.py

@@ -127,10 +127,10 @@ def __repr__(self):
     Protocol(P_DCCP, 'dccp', 'uint16be'),
     Protocol(P_IP6, 'ip6', 'ip6'),
     Protocol(P_IP6ZONE, 'ip6zone', 'utf8'),
-    Protocol(P_DNS, 'dns', 'idna'),
-    Protocol(P_DNS4, 'dns4', 'idna'),
-    Protocol(P_DNS6, 'dns6', 'idna'),
-    Protocol(P_DNSADDR, 'dnsaddr', 'idna'),
+    Protocol(P_DNS, 'dns', 'domain'),
+    Protocol(P_DNS4, 'dns4', 'domain'),
+    Protocol(P_DNS6, 'dns6', 'domain'),
+    Protocol(P_DNSADDR, 'dnsaddr', 'domain'),
     Protocol(P_SCTP, 'sctp', 'uint16be'),
     Protocol(P_UDT, 'udt', None),
     Protocol(P_UTP, 'utp', None),

tests/test_transforms.py

@@ -37,11 +37,13 @@
 
     # Additional test data
     (_names_to_protocols['dns4'],
-     b'xn--4gbrim.xn----ymcbaaajlc6dj7bxne2c.xn--wgbh1c',
+     b'\xd9\x85\xd9\x88\xd9\x82\xd8\xb9.\xd9\x88\xd8\xb2\xd8\xa7\xd8\xb1\xd8\xa9'
+     b'-\xd8\xa7\xd9\x84\xd8\xa7\xd8\xaa\xd8\xb5\xd8\xa7\xd9\x84\xd8\xa7\xd8\xaa'
+     b'.\xd9\x85\xd8\xb5\xd8\xb1',
      # Explicitly mark this as unicode to force the text to be LTR in editors
      u'موقع.وزارة-الاتصالات.مصر'),
     (_names_to_protocols['dns4'],
-     b'xn--fuball-cta.example',
+     b'fu\xc3\x9fball.example',
      u'fußball.example'),  # This will fail if IDNA-2003/NamePrep is used
 ]