Merge branch 'braaen-eui' into rel-0.7.x

5 files changed, 170 insertions, 46 deletions

diff --git a/netaddr/__init__.py b/netaddr/__init__.py
index 504535f..bd43eec 100644
--- a/netaddr/__init__.py
+++ b/netaddr/__init__.py
@@ -42,6 +42,9 @@ from netaddr.strategy.ipv6 import (valid_str as valid_ipv6, ipv6_compact,
 from netaddr.strategy.eui48 import (mac_eui48, mac_unix, mac_unix_expanded,
         mac_cisco, mac_bare, mac_pgsql, valid_str as valid_mac)
 
+from netaddr.strategy.eui64 import (eui64_base, eui64_unix, eui64_unix_expanded,
+        eui64_cisco, eui64_bare, valid_str as valid_mac)
+
 __all__ = [
     #   Constants.
     'ZEROFILL', 'Z', 'INET_PTON', 'P', 'NOHOST', 'N',
diff --git a/netaddr/eui/__init__.py b/netaddr/eui/__init__.py
index 3affc21..11a050b 100644
--- a/netaddr/eui/__init__.py
+++ b/netaddr/eui/__init__.py
@@ -11,6 +11,7 @@ identifiers.
 from netaddr.core import NotRegisteredError, AddrFormatError, DictDotLookup
 from netaddr.strategy import eui48 as _eui48, eui64 as _eui64
 from netaddr.strategy.eui48 import mac_eui48
+from netaddr.strategy.eui64 import eui64_base
 from netaddr.ip import IPAddress
 from netaddr.compat import _is_int, _is_str
 
@@ -456,7 +457,10 @@ class EUI(BaseIdentifier):
 
     def _set_dialect(self, value):
         if value is None:
-            self._dialect = mac_eui48
+            if self._module is _eui64:
+                self._dialect = eui64_base
+            else:
+                self._dialect = mac_eui48
         else:
             if hasattr(value, 'word_size') and hasattr(value, 'word_fmt'):
                 self._dialect = value
diff --git a/netaddr/strategy/eui48.py b/netaddr/strategy/eui48.py
index c960dd6..68c8232 100644
--- a/netaddr/strategy/eui48.py
+++ b/netaddr/strategy/eui48.py
@@ -19,6 +19,7 @@ except ImportError:
     AF_LINK = 48
 
 from netaddr.core import AddrFormatError
+from netaddr.compat import _is_str
 from netaddr.strategy import (
     valid_words as _valid_words, int_to_words as _int_to_words,
     words_to_int as _words_to_int, valid_bits as _valid_bits,
@@ -26,8 +27,6 @@ from netaddr.strategy import (
     valid_bin as _valid_bin, int_to_bin as _int_to_bin,
     bin_to_int as _bin_to_int)
 
-from netaddr.compat import _is_str
-
 #: The width (in bits) of this address type.
 width = 48
 
diff --git a/netaddr/strategy/eui64.py b/netaddr/strategy/eui64.py
index 0100a18..01c98ea 100644
--- a/netaddr/strategy/eui64.py
+++ b/netaddr/strategy/eui64.py
@@ -9,11 +9,7 @@ IEEE 64-bit EUI (Extended Unique Indentifier) logic.
 import struct as _struct
 import re as _re
 
-#   This is a fake constant that doesn't really exist. Here for completeness.
-AF_EUI64 = 64
-
 from netaddr.core import AddrFormatError
-from netaddr.compat import _is_str
 from netaddr.strategy import (
     valid_words as _valid_words, int_to_words as _int_to_words,
     words_to_int as _words_to_int, valid_bits as _valid_bits,
@@ -21,17 +17,12 @@ from netaddr.strategy import (
     valid_bin as _valid_bin, int_to_bin as _int_to_bin,
     bin_to_int as _bin_to_int)
 
-#: The width (in bits) of this address type.
-width = 64
 
-#: The individual word size (in bits) of this address type.
-word_size = 8
-
-#: The format string to be used when converting words to string values.
-word_fmt = '%.2X'
+#   This is a fake constant that doesn't really exist. Here for completeness.
+AF_EUI64 = 64
 
-#: The separator character used between each word.
-word_sep = '-'
+#: The width (in bits) of this address type.
+width = 64
 
 #: The AF_* constant value of this address type.
 family = AF_EUI64
@@ -42,24 +33,88 @@ family_name = 'EUI-64'
 #: The version of this address type.
 version = 64
 
-#: The number base to be used when interpreting word values as integers.
-word_base = 16
-
 #: The maximum integer value that can be represented by this address type.
 max_int = 2 ** width - 1
 
-#: The number of words in this address type.
-num_words = width // word_size
+#-----------------------------------------------------------------------------
+#   Dialect classes.
+#-----------------------------------------------------------------------------
+
+class eui64_base(object):
+    """A standard IEEE EUI-64 dialect class."""
+    #: The individual word size (in bits) of this address type.
+    word_size = 8
+
+    #: The number of words in this address type.
+    num_words = width // word_size
+
+    #: The maximum integer value for an individual word in this address type.
+    max_word = 2 ** word_size - 1
+
+    #: The separator character used between each word.
+    word_sep = '-'
+
+    #: The format string to be used when converting words to string values.
+    word_fmt = '%.2X'
 
-#: The maximum integer value for an individual word in this address type.
-max_word = 2 ** word_size - 1
+    #: The number base to be used when interpreting word values as integers.
+    word_base = 16
 
-#: Compiled regular expression for detecting value EUI-64 identifiers.
-RE_EUI64_FORMATS = [
-    _re.compile('^' + ':'.join(['([0-9A-F]{1,2})'] * 8) + '$', _re.IGNORECASE),
-    _re.compile('^' + '-'.join(['([0-9A-F]{1,2})'] * 8) + '$', _re.IGNORECASE),
-    _re.compile('^(' + '[0-9A-F]' * 16 + ')$', _re.IGNORECASE),
-]
+
+class eui64_unix(eui64_base):
+    """A UNIX-style MAC address dialect class."""
+    word_size = 8
+    num_words = width // word_size
+    word_sep = ':'
+    word_fmt = '%x'
+    word_base = 16
+
+
+class eui64_unix_expanded(eui64_unix):
+    """A UNIX-style MAC address dialect class with leading zeroes."""
+    word_fmt = '%.2x'
+
+
+class eui64_cisco(eui64_base):
+    """A Cisco 'triple hextet' MAC address dialect class."""
+    word_size = 16
+    num_words = width // word_size
+    word_sep = '.'
+    word_fmt = '%.4x'
+    word_base = 16
+
+
+class eui64_bare(eui64_base):
+    """A bare (no delimiters) MAC address dialect class."""
+    word_size = 64
+    num_words = width // word_size
+    word_sep = ''
+    word_fmt = '%.16X'
+    word_base = 16
+
+
+#: The default dialect to be used when not specified by the user.
+
+DEFAULT_EUI64_DIALECT = eui64_base
+
+#-----------------------------------------------------------------------------
+#: Regular expressions to match all supported MAC address formats.
+RE_EUI64_FORMATS = (
+    #   2 bytes x 8 (UNIX, Windows, EUI-64)
+    '^' + ':'.join(['([0-9A-F]{1,2})'] * 8) + '$',
+    '^' + '-'.join(['([0-9A-F]{1,2})'] * 8) + '$',
+
+    #   4 bytes x 4 (Cisco like)
+    '^' + ':'.join(['([0-9A-F]{1,4})'] * 4) + '$',
+    '^' + '-'.join(['([0-9A-F]{1,4})'] * 4) + '$',
+    '^' + '\.'.join(['([0-9A-F]{1,4})'] * 4) + '$',
+
+    #   16 bytes (bare, no delimiters)
+    '^(' + ''.join(['[0-9A-F]'] * 16) + ')$',
+)
+#   For efficiency, each string regexp converted in place to its compiled
+#   counterpart.
+RE_EUI64_FORMATS = [_re.compile(_, _re.IGNORECASE) for _ in RE_EUI64_FORMATS]
 
 
 def _get_match_result(address, formats):
@@ -89,7 +144,7 @@ def str_to_int(addr):
     :param addr: An IEEE EUI-64 indentifier in string form.
 
     :return: An unsigned integer that is equivalent to value represented
-        by EUI-64 string identifier.
+        by EUI-64 string address formatted according to the dialect
     """
     words = []
 
@@ -100,13 +155,25 @@ def str_to_int(addr):
     except TypeError:
         raise AddrFormatError('invalid IEEE EUI-64 identifier: %r!' % addr)
 
-    if _is_str(words):
-        return int(words, 16)
-    if len(words) != num_words:
+    if isinstance(words, tuple):
+        pass
+    else:
+        words = (words,)
+
+    if len(words) == 8:
+        #   2 bytes x 8 (UNIX, Windows, EUI-48)
+        int_val = int(''.join(['%.2x' % int(w, 16) for w in words]), 16)
+    elif len(words) == 4:
+        #   4 bytes x 4 (Cisco like)
+        int_val = int(''.join(['%.4x' % int(w, 16) for w in words]), 16)
+    elif len(words) == 1:
+        #   16 bytes (bare, no delimiters)
+        int_val = int('%016x' % int(words[0], 16), 16)
+    else:
         raise AddrFormatError(
             'bad word count for EUI-64 identifier: %r!' % addr)
 
-    return int(''.join(['%.2x' % int(w, 16) for w in words]), 16)
+    return int_val
 
 
 def int_to_str(int_val, dialect=None):
@@ -114,13 +181,14 @@ def int_to_str(int_val, dialect=None):
     :param int_val: An unsigned integer.
 
     :param dialect: (optional) a Python class defining formatting options
-        (Please Note - not currently in use).
 
     :return: An IEEE EUI-64 identifier that is equivalent to unsigned integer.
     """
-    words = int_to_words(int_val)
-    tokens = [word_fmt % i for i in words]
-    addr = word_sep.join(tokens)
+    if dialect is None:
+        dialect = eui64_base
+    words = int_to_words(int_val, dialect)
+    tokens = [dialect.word_fmt % i for i in words]
+    addr = dialect.word_sep.join(tokens)
     return addr
 
 
@@ -155,30 +223,45 @@ def packed_to_int(packed_int):
 
 
 def valid_words(words, dialect=None):
-    return _valid_words(words, word_size, num_words)
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
+    return _valid_words(words, dialect.word_size, dialect.num_words)
 
 
 def int_to_words(int_val, dialect=None):
-    return _int_to_words(int_val, word_size, num_words)
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
+    return _int_to_words(int_val, dialect.word_size, dialect.num_words)
 
 
 def words_to_int(words, dialect=None):
-    return _words_to_int(words, word_size, num_words)
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
+    return _words_to_int(words, dialect.word_size, dialect.num_words)
 
 
 def valid_bits(bits, dialect=None):
-    return _valid_bits(bits, width, word_sep)
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
+    return _valid_bits(bits, width, dialect.word_sep)
 
 
 def bits_to_int(bits, dialect=None):
-    return _bits_to_int(bits, width, word_sep)
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
+    return _bits_to_int(bits, width, dialect.word_sep)
 
 
 def int_to_bits(int_val, dialect=None):
-    return _int_to_bits(int_val, word_size, num_words, word_sep)
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
+    return _int_to_bits(
+        int_val, dialect.word_size, dialect.num_words, dialect.word_sep)
 
 
-def valid_bin(bin_val):
+def valid_bin(bin_val, dialect=None):
+    if dialect is None:
+        dialect = DEFAULT_EUI64_DIALECT
     return _valid_bin(bin_val, width)
 
 
diff --git a/netaddr/tests/eui/test_eui.py b/netaddr/tests/eui/test_eui.py
index 48ef7d2..c806028 100644
--- a/netaddr/tests/eui/test_eui.py
+++ b/netaddr/tests/eui/test_eui.py
@@ -4,7 +4,9 @@ import random
 
 import pytest
 
-from netaddr import EUI, mac_unix, mac_unix_expanded, mac_cisco, mac_bare, mac_pgsql, OUI, IAB, IPAddress
+from netaddr import (EUI, mac_unix, mac_unix_expanded, mac_cisco,
+    mac_bare, mac_pgsql, eui64_unix, eui64_unix_expanded,
+    eui64_cisco, eui64_bare, OUI, IAB, IPAddress)
 
 
 def test_mac_address_properties():
@@ -99,6 +101,39 @@ def test_eui_custom_dialect():
     assert str(mac) == '00:1B:77:49:54:FD'
 
 
+def test_eui64_dialects():
+    mac = EUI('00-1B-77-49-54-FD-12-34')
+    assert str(mac) == '00-1B-77-49-54-FD-12-34'
+
+    mac = EUI('00-1B-77-49-54-FD-12-34', dialect=eui64_unix)
+    assert str(mac) == '0:1b:77:49:54:fd:12:34'
+
+    mac = EUI('00-1B-77-49-54-FD-12-34', dialect=eui64_unix_expanded)
+    assert str(mac) == '00:1b:77:49:54:fd:12:34'
+
+    mac = EUI('00-1B-77-49-54-FD-12-34', dialect=eui64_cisco)
+    assert str(mac) == '001b.7749.54fd.1234'
+
+    mac = EUI('00-1B-77-49-54-FD-12-34', dialect=eui64_bare)
+    assert str(mac) == '001B774954FD1234'
+
+
+def test_eui64_dialect_property_assignment():
+    mac = EUI('00-1B-77-49-54-FD-12-34')
+    assert str(mac) == '00-1B-77-49-54-FD-12-34'
+
+    mac.dialect = eui64_cisco
+    assert str(mac) == '001b.7749.54fd.1234'
+
+
+def test_eui64_custom_dialect():
+    class eui64_custom(eui64_unix):
+        word_fmt = '%.2X'
+
+    mac = EUI('00-1B-77-49-54-FD-12-34', dialect=eui64_custom)
+    assert str(mac) == '00:1B:77:49:54:FD:12:34'
+
+
 def test_eui_oui_information():
     mac = EUI('00-1B-77-49-54-FD')
     oui = mac.oui