Merge pull request #35 from alexander255/master

Allow parsing of DNS names and non-ASCII unix sockets

Author: mhchia

multiaddr/codec.py

@@ -1,10 +1,15 @@
 import base58
 import base64
 import binascii
+import os
 
+import idna
 from netaddr import IPAddress
 
 from .protocols import code_to_varint
+from .protocols import P_DNS
+from .protocols import P_DNS4
+from .protocols import P_DNS6
 from .protocols import P_DCCP
 from .protocols import P_IP4
 from .protocols import P_IP6
@@ -156,7 +161,11 @@ def address_string_to_bytes(proto, addr_string):
             raise ValueError("invalid P2P multihash: %s" % mm)
         return b''.join([size, mm])
     elif proto.code == P_UNIX:
-        addr_string_bytes = addr_string.encode("ascii")
+        addr_string_bytes = os.fsencode(addr_string)
+        size = code_to_varint(len(addr_string_bytes))
+        return b''.join([size, binascii.hexlify(addr_string_bytes)])
+    elif proto.code in (P_DNS, P_DNS4, P_DNS6):
+        addr_string_bytes = idna.encode(addr_string, uts46=True)
         size = code_to_varint(len(addr_string_bytes))
         return b''.join([size, binascii.hexlify(addr_string_bytes)])
     else:
@@ -192,7 +201,11 @@ def address_bytes_to_string(proto, buf):
     elif proto.code == P_UNIX:
         buf = binascii.unhexlify(buf)
         size, num_bytes_read = read_varint_code(buf)
-        return buf[num_bytes_read:].decode('ascii')
+        return os.fsdecode(buf[num_bytes_read:])
+    elif proto.code in (P_DNS, P_DNS4, P_DNS6):
+        buf = binascii.unhexlify(buf)
+        size, num_bytes_read = read_varint_code(buf)
+        return idna.decode(buf[num_bytes_read:])
     raise ValueError("unknown protocol")
 
 

requirements.txt

@@ -2,3 +2,4 @@ varint
 six
 base58
 netaddr
+idna

tests/test_codec.py

@@ -23,6 +23,14 @@
     (_names_to_protocols['p2p'],
      b'221220d52ebb89d85b02a284948203a62ff28389c57c9f42beec4ec20db76a68911c0b',
      'QmcgpsyWgH8Y8ajJz1Cu72KnS5uo2Aa2LpzU7kinSupNKC'),
+
+# Additional test data
+    (_names_to_protocols['dns4'],
+     b'30786e2d2d34676272696d2e786e2d2d2d2d796d63626161616a6c6336646a3762786e6532632e786e2d2d776762683163',
+     u'موقع.وزارة-الاتصالات.مصر'),  # Explicietly mark as unicode to force it LTR
+    (_names_to_protocols['dns4'],
+     b'16786e2d2d667562616c6c2d6374612e6578616d706c65',
+     'fußball.example'),  # This will fail if IDNA-2003/NamePrep is used
 ]
 
 BYTES_MAP_STR_TEST_DATA = [

tests/test_multiaddr.py

@@ -45,7 +45,10 @@
      "/ip4/127.0.0.1/p2p/tcp",
      "/unix",
      "/ip4/1.2.3.4/tcp/80/unix",
-     "/ip4/127.0.0.1/tcp/9090/http/p2p-webcrt-direct"])
+     "/ip4/127.0.0.1/tcp/9090/http/p2p-webcrt-direct",
+     "/dns",
+     "/dns4",
+     "/dns6"])
 def test_invalid(addr_str):
     with pytest.raises(ValueError):
         Multiaddr(addr_str)
@@ -77,10 +80,13 @@ def test_invalid(addr_str):
      "/ip4/127.0.0.1/udp/1234",
      "/ip4/127.0.0.1/p2p/QmcgpsyWgH8Y8ajJz1Cu72KnS5uo2Aa2LpzU7kinSupNKC/tcp/1234",
      "/unix/a/b/c/d/e",
+     "/unix/Überrschung!/大柱",
      "/unix/stdio",
      "/ip4/1.2.3.4/tcp/80/unix/a/b/c/d/e/f",
      "/ip4/127.0.0.1/p2p/QmcgpsyWgH8Y8ajJz1Cu72KnS5uo2Aa2LpzU7kinSupNKC/tcp/1234/unix/stdio",
-     "/ip4/127.0.0.1/tcp/9090/http/p2p-webrtc-direct"])  # nopep8
+     "/ip4/127.0.0.1/tcp/9090/http/p2p-webrtc-direct",
+     "/dns/example.com",
+     "/dns4/موقع.وزارة-الاتصالات.مصر"])  # nopep8
 def test_valid(addr_str):
     ma = Multiaddr(addr_str)
     assert str(ma) == addr_str.rstrip("/")