instead of printing it for every new connection

twisted.internet.error.UnsupportedAddressFamily was added in twisted-12.1.0.

This commit adds a stub that emulates it for platforms that don't have a
recent Twisted (e.g. Debian squeeze/wheezy).

Server-side obfsproxy does not need txsocksx.

From now on, obfsproxy when configured to connect to a proxy (either
using --proxy or TOR_PT_PROXY) will refuse to run if it doesn't have
txsocksx or its Twisted is too old (< 13.2.0).

Conflicts:
	ChangeLog

Clean some long lines, and some instances of pylint's
logging-not-lazy (although the rest of the codebase has many more
instances of this).

Some HTTP proxys opt to close the connection instead of returning a
HTTP status code on certain failures (Eg: polipo auth faliure, privoxy
ACL denial).  This change logs a better error message to the obfsproxy
log when that happens.

As an added bonus the SOCKS5 server now knows about EHOSTUNREACH, so
a more informative response is sent on that failure type.

This adds a HTTPS CONNECT client.  It's been lightly tested with both
privoxy and apache2 and appears to work, both without authentication
and with Basic.

On all the systems I've checked so far running a U*IX variant,
`python2` is a valid symlink to the Python 2.x.x interpreter.  Since
certain systems are also shipping Python 3 and some make it the default
system python, the shebang should be explicit.

This fixes bug #11190.

TransportConfig already carries global obfsproxy information (like the
state location, etc.). The only difference with the settings.py idea
that settings.py:Config was a singleton.

Conflicts:
	ChangeLog

This works around <http://bugs.python.org/issue10212> which prevented
the SOCKS5 server from working for clients running Python < 2.7.4,
fixing <https://bugs.torproject.org/11329>.

Found by Yawning Angel.

Patch originally by Arturo Filasto, with changes by Yawning Angel.  This
patch depends on pyptlib modifications as it also supports using the
managed TOR_PT_PROXY enviornment variable.

WARNING: Attempting to use a http proxy will break mysteriously as the
connect routines are just stubbed out.

The attribute "creationTime" does not exist.  Instead, update the correct
attribute "keyCreation".  This bug broke the server's key rotation.  The patch,
written by Yawning Angel, fixes <https://bugs.torproject.org/11100>.

 * ACCEPTABLE_CMDS now only contains CONNECT, since that is the only command
   used and actually implemented.
 * Removed trailing whitespace from the socks5 tests.

Not going to touch #10240 for now since the SOCKS args handler will eventually
be totally rewritten (changed to return a dict containing the separated [k=v]
pairs).

 * Since the compat inet_ntop is used, no longer need to try/catch
 * Make _ByteBuffer.get(self, length) return a string and remove a ton of
   str() calls

Apparently, these routines are missing on Windows.

Changed based on feedback from hellias:
 * handleCmdConnectFailure now traps errors
 * sendReply no longer deals with parsing out the outgoing local address
 * logging uses the obfsproxy logger again
 * Fixed a bug where send_reply was used instead of sendReply
 * The obfsproxy specific SOCKSv5 classes were renamed to OBFSSOCKSv5Outgoing
   and OBFSSOCKSv5Protocol

Bugs I found:
 * SOCKSv5Protocol.dataReceived() should actually work now (never called by
   obfsproxy since it's overridden)
 * _ByteBuffer.get_uint32() was always doing byte order conversion (method never
   used)

 * Instead of using methodcaller, just store lambdas in the auth vtable
 * Add doc strings to _ByteBuffer
 * Do not use "len" as a variable name
 * Fix a bug in processEstablishedData() that pylint caught (obfsproxy
   would never have triggered it since the routine is overriden)

 * Use a wrapper class to isolate the scary struct.pack/unpack calls
 * Refactor the code to look like twisted.protocols.socks
 * Support DOMAINNAME, though it's untested

 * SOCKSv5Reply changed to a newstyle class
 * Move the pluggable transport specific auth code into it's own routine, with
   lots of comments.

Thanks to isis for the help.

 * It works over the loopback address at least, still haven't gotten a tunnel.