bytearray instead of bytes

Since Python 2 has no distinct bytes type, bytearray is used instead for the “packed” (binary) address representation:

>>> ipaddress.ip_address(bytearray('\xc0\xa8\x54\x17'))
IPv4Address('192.168.84.23')
>>> ipaddress.ip_address('127.0.0.17').packed
bytearray(b'\x7f\x00\x00\x11')

This means py2-ipaddress can accept both Python 2 string types (str and unicode) for the textual address representation.

If you prefer semantics closer to Python 3, you may be interested in Philipp Hagemeister’s ipaddress backport, which uses str for the “packed” address representation, but then requires all textual IP addresses to be given as unicode strings. That backport also supports Python 3.0–3.2.

Caching of is_private and is_global

Since Python 2.7’s functools module does not have Python 3.2’s lru_cache, no caching is performed for the is_private and is_global properties; this should be a minor problem as Python 3.3’s ipaddress did not use lru_cache either.

py2-ipaddress 3.4.2

Test suite and documentation improvements.

py2-ipaddress 3.4.1

Python 2.6 support and a bugfix.

py2-ipaddress 3.4

Since Python 2 does not distinguish between bytes and str like Python 3 does, version 2.0.1 and earlier of py2-ipaddress attempted to interpret str arguments as both and do the “right” thing.

This unfortunately led to surprising behavior in py2-ipaddress:

>>> ipaddress.ip_address('test.example.org')
IPv6Address('7465:7374:2e65:7861:6d70:6c65:2e6f:7267')

The ipaddress module does not, of course, perform DNS resolution. Rather, the argument is interpreted as a byte string (of length 16) and converted bit-for-bit into an IPv6 address. In Python 3, ipaddress correctly rejects such a constructor argument (unless the b prefix is used to explicitly mark the literal as a byte string).

Even worse, there is not always a single right interpretation. Python 3 example:

>>> ipaddress.ip_address('::1234:5678:9abc')
IPv6Address('::1234:5678:9abc')
>>> ipaddress.ip_address(b'::1234:5678:9abc')
IPv6Address('3a3a:3132:3334:3a35:3637:383a:3961:6263')

There is no way to distinguish the two invocations in Python 2. As a result, py2-ipaddress 3.4 uses bytearray for all byte strings, and str for text strings only:

>>> ipaddress.ip_address('::1234:5678:9abc')
IPv6Address('::1234:5678:9abc')
>>> ipaddress.ip_address(b'::1234:5678:9abc')
IPv6Address('::1234:5678:9abc')
>>> ipaddress.ip_address(bytearray('::1234:5678:9abc'))
IPv6Address('3a3a:3132:3334:3a35:3637:383a:3961:6263')