Port source code back to Python 2

Author: ntninja

multiaddr/codec.py

@@ -5,6 +5,7 @@
 
 import idna
 from netaddr import IPAddress
+import six
 
 from .protocols import code_to_varint
 from .protocols import P_DNS
@@ -24,6 +25,23 @@
 from .protocols import read_varint_code
 
 
+if hasattr(os, "fsencode") and hasattr(os, "fsdecode"):
+    fsencode = os.fsencode
+    fsdecode = os.fsdecode
+else:  # PY2
+    import sys
+
+    def fsencode(path):
+        if not isinstance(path, six.binary_type):
+            path = path.encode(sys.getfilesystemencoding())
+        return path
+
+    def fsdecode(path):
+        if not isinstance(path, six.text_type):
+            path = path.decode(sys.getfilesystemencoding())
+        return path
+
+
 def string_to_bytes(string):
     if not string:
         return b''
@@ -150,6 +168,8 @@ def address_string_to_bytes(proto, addr_string):
     elif proto.code == P_P2P:  # ipfs
         # the address is a varint prefixed multihash string representation
         try:
+            if six.PY2 and isinstance(addr_string, unicode):
+                addr_string = addr_string.encode("ascii")
             mm = base58.b58decode(addr_string)
         except Exception as ex:
             raise ValueError("failed to parse p2p addr: %s %s"
@@ -161,7 +181,7 @@ 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 = os.fsencode(addr_string)
+        addr_string_bytes = 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):
@@ -201,7 +221,7 @@ 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 os.fsdecode(buf[num_bytes_read:])
+        return 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)

multiaddr/multiaddr.py

@@ -2,6 +2,8 @@
 import binascii
 from copy import copy
 
+import six
+
 from .codec import size_for_addr
 from .codec import string_to_bytes
 from .codec import bytes_to_string
@@ -37,9 +39,16 @@ def __init__(self, addr):
             addr : A string-encoded or a byte-encoded Multiaddr
 
         """
-        if isinstance(addr, str):
+        # On Python 2 text string will often be binary anyways so detect the
+        # obvious case of a “binary-encoded” multiaddr starting with a slash
+        # and decode it into text
+        if six.PY2 and isinstance(addr, str) and addr.startswith("/"):
+            import locale
+            addr = addr.decode(locale.getpreferredencoding())
+
+        if isinstance(addr, six.text_type):
             self._bytes = string_to_bytes(addr)
-        elif isinstance(addr, bytes):
+        elif isinstance(addr, six.binary_type):
             self._bytes = addr
         else:
             raise ValueError("Invalid address type, must be bytes or str")
@@ -62,6 +71,16 @@ def __str__(self):
             raise ValueError(
                 "multiaddr failed to convert back to string. corrupted?")
 
+    # On Python 2 __str__ needs to return binary text, so expose the original
+    # function as __unicode__ and transparently encode its returned text based
+    # on the current locale
+    if six.PY2:
+        __unicode__ = __str__
+
+        def __str__(self):
+            import locale
+            return self.__unicode__().encode(locale.getpreferredencoding())
+
     def __repr__(self):
         return "<Multiaddr %s>" % str(self)
 

tests/test_codec.py

@@ -1,3 +1,4 @@
+# -*- encoding: utf-8 -*-
 import pytest
 
 from multiaddr.codec import address_bytes_to_string
@@ -27,10 +28,10 @@
 # Additional test data
     (_names_to_protocols['dns4'],
      b'30786e2d2d34676272696d2e786e2d2d2d2d796d63626161616a6c6336646a3762786e6532632e786e2d2d776762683163',
-     u'موقع.وزارة-الاتصالات.مصر'),  # Explicietly mark as unicode to force it LTR
+     u'موقع.وزارة-الاتصالات.مصر'),  # Explicietly mark this as unicode to force the text to be LTR in editors
     (_names_to_protocols['dns4'],
      b'16786e2d2d667562616c6c2d6374612e6578616d706c65',
-     'fußball.example'),  # This will fail if IDNA-2003/NamePrep is used
+     u'fußball.example'),  # This will fail if IDNA-2003/NamePrep is used
 ]
 
 BYTES_MAP_STR_TEST_DATA = [

tests/test_multiaddr.py

@@ -1,6 +1,9 @@
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
+from __future__ import unicode_literals
+
 import pytest
+import six
 
 from multiaddr.multiaddr import Multiaddr
 from multiaddr.multiaddr import ProtocolNotFoundException
@@ -89,7 +92,7 @@ def test_invalid(addr_str):
      "/dns4/موقع.وزارة-الاتصالات.مصر"])  # nopep8
 def test_valid(addr_str):
     ma = Multiaddr(addr_str)
-    assert str(ma) == addr_str.rstrip("/")
+    assert six.text_type(ma) == addr_str.rstrip("/")
 
 
 def test_eq():
@@ -250,7 +253,7 @@ def test_get_value_too_many_fields_protocol(monkeypatch):
     that the constructor specifies is ignored by the test.
     """
     monkeypatch.setattr("multiaddr.multiaddr.Multiaddr.__str__",
-                        lambda ignore: '/udp/1234/5678')
+                        lambda ignore: str('/udp/1234/5678'))
     a = Multiaddr("/ip4/127.0.0.1/udp/1234")
     with pytest.raises(ValueError):
         a.value_for_protocol(P_UDP)