util: remove md5

This hash function is broken and should not be used by new code. It is
currently only used by keepalive.

mercurial/keepalive.py

@@ -1,661 +1,671 b''
 #   This library is free software; you can redistribute it and/or
 #   modify it under the terms of the GNU Lesser General Public
 #   License as published by the Free Software Foundation; either
 #   version 2.1 of the License, or (at your option) any later version.
 #
 #   This library is distributed in the hope that it will be useful,
 #   but WITHOUT ANY WARRANTY; without even the implied warranty of
 #   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 #   Lesser General Public License for more details.
 #
 #   You should have received a copy of the GNU Lesser General Public
 #   License along with this library; if not, write to the
 #      Free Software Foundation, Inc.,
 #      59 Temple Place, Suite 330,
 #      Boston, MA  02111-1307  USA
 
 # This file is part of urlgrabber, a high-level cross-protocol url-grabber
 # Copyright 2002-2004 Michael D. Stenner, Ryan Tomayko
 
 # Modified by Benoit Boissinot:
 #  - fix for digest auth (inspired from urllib2.py @ Python v2.4)
 # Modified by Dirkjan Ochtman:
 #  - import md5 function from a local util module
+# Modified by Martin Geisler:
+#  - moved md5 function from local util module to this module
 
 """An HTTP handler for urllib2 that supports HTTP 1.1 and keepalive.
 
 >>> import urllib2
 >>> from keepalive import HTTPHandler
 >>> keepalive_handler = HTTPHandler()
 >>> opener = urllib2.build_opener(keepalive_handler)
 >>> urllib2.install_opener(opener)
 >>>
 >>> fo = urllib2.urlopen('http://www.python.org')
 
 If a connection to a given host is requested, and all of the existing
 connections are still in use, another connection will be opened.  If
 the handler tries to use an existing connection but it fails in some
 way, it will be closed and removed from the pool.
 
 To remove the handler, simply re-run build_opener with no arguments, and
 install that opener.
 
 You can explicitly close connections by using the close_connection()
 method of the returned file-like object (described below) or you can
 use the handler methods:
 
   close_connection(host)
   close_all()
   open_connections()
 
 NOTE: using the close_connection and close_all methods of the handler
 should be done with care when using multiple threads.
   * there is nothing that prevents another thread from creating new
     connections immediately after connections are closed
   * no checks are done to prevent in-use connections from being closed
 
 >>> keepalive_handler.close_all()
 
 EXTRA ATTRIBUTES AND METHODS
 
   Upon a status of 200, the object returned has a few additional
   attributes and methods, which should not be used if you want to
   remain consistent with the normal urllib2-returned objects:
 
     close_connection()  -  close the connection to the host
     readlines()         -  you know, readlines()
     status              -  the return status (ie 404)
     reason              -  english translation of status (ie 'File not found')
 
   If you want the best of both worlds, use this inside an
   AttributeError-catching try:
 
   >>> try: status = fo.status
   >>> except AttributeError: status = None
 
   Unfortunately, these are ONLY there if status == 200, so it's not
   easy to distinguish between non-200 responses.  The reason is that
   urllib2 tries to do clever things with error codes 301, 302, 401,
   and 407, and it wraps the object upon return.
 
   For python versions earlier than 2.4, you can avoid this fancy error
   handling by setting the module-level global HANDLE_ERRORS to zero.
   You see, prior to 2.4, it's the HTTP Handler's job to determine what
   to handle specially, and what to just pass up.  HANDLE_ERRORS == 0
   means "pass everything up".  In python 2.4, however, this job no
   longer belongs to the HTTP Handler and is now done by a NEW handler,
   HTTPErrorProcessor.  Here's the bottom line:
 
     python version < 2.4
         HANDLE_ERRORS == 1  (default) pass up 200, treat the rest as
                             errors
         HANDLE_ERRORS == 0  pass everything up, error processing is
                             left to the calling code
     python version >= 2.4
         HANDLE_ERRORS == 1  pass up 200, treat the rest as errors
         HANDLE_ERRORS == 0  (default) pass everything up, let the
                             other handlers (specifically,
                             HTTPErrorProcessor) decide what to do
 
   In practice, setting the variable either way makes little difference
   in python 2.4, so for the most consistent behavior across versions,
   you probably just want to use the defaults, which will give you
   exceptions on errors.
 
 """
 
 # $Id: keepalive.py,v 1.14 2006/04/04 21:00:32 mstenner Exp $
 
 import urllib2
 import httplib
 import socket
 import thread
 
 DEBUG = None
 
 import sys
 if sys.version_info < (2, 4): HANDLE_ERRORS = 1
 else: HANDLE_ERRORS = 0
 
 class ConnectionManager:
     """
     The connection manager must be able to:
       * keep track of all existing
       """
     def __init__(self):
         self._lock = thread.allocate_lock()
         self._hostmap = {} # map hosts to a list of connections
         self._connmap = {} # map connections to host
         self._readymap = {} # map connection to ready state
 
     def add(self, host, connection, ready):
         self._lock.acquire()
         try:
             if not host in self._hostmap: self._hostmap[host] = []
             self._hostmap[host].append(connection)
             self._connmap[connection] = host
             self._readymap[connection] = ready
         finally:
             self._lock.release()
 
     def remove(self, connection):
         self._lock.acquire()
         try:
             try:
                 host = self._connmap[connection]
             except KeyError:
                 pass
             else:
                 del self._connmap[connection]
                 del self._readymap[connection]
                 self._hostmap[host].remove(connection)
                 if not self._hostmap[host]: del self._hostmap[host]
         finally:
             self._lock.release()
 
     def set_ready(self, connection, ready):
         try: self._readymap[connection] = ready
         except KeyError: pass
 
     def get_ready_conn(self, host):
         conn = None
         self._lock.acquire()
         try:
             if host in self._hostmap:
                 for c in self._hostmap[host]:
                     if self._readymap[c]:
                         self._readymap[c] = 0
                         conn = c
                         break
         finally:
             self._lock.release()
         return conn
 
     def get_all(self, host=None):
         if host:
             return list(self._hostmap.get(host, []))
         else:
             return dict(self._hostmap)
 
 class KeepAliveHandler:
     def __init__(self):
         self._cm = ConnectionManager()
 
     #### Connection Management
     def open_connections(self):
         """return a list of connected hosts and the number of connections
         to each.  [('foo.com:80', 2), ('bar.org', 1)]"""
         return [(host, len(li)) for (host, li) in self._cm.get_all().items()]
 
     def close_connection(self, host):
         """close connection(s) to <host>
         host is the host:port spec, as in 'www.cnn.com:8080' as passed in.
         no error occurs if there is no connection to that host."""
         for h in self._cm.get_all(host):
             self._cm.remove(h)
             h.close()
 
     def close_all(self):
         """close all open connections"""
         for host, conns in self._cm.get_all().iteritems():
             for h in conns:
                 self._cm.remove(h)
                 h.close()
 
     def _request_closed(self, request, host, connection):
         """tells us that this request is now closed and the the
         connection is ready for another request"""
         self._cm.set_ready(connection, 1)
 
     def _remove_connection(self, host, connection, close=0):
         if close: connection.close()
         self._cm.remove(connection)
 
     #### Transaction Execution
     def http_open(self, req):
         return self.do_open(HTTPConnection, req)
 
     def do_open(self, http_class, req):
         host = req.get_host()
         if not host:
             raise urllib2.URLError('no host given')
 
         try:
             h = self._cm.get_ready_conn(host)
             while h:
                 r = self._reuse_connection(h, req, host)
 
                 # if this response is non-None, then it worked and we're
                 # done.  Break out, skipping the else block.
                 if r: break
 
                 # connection is bad - possibly closed by server
                 # discard it and ask for the next free connection
                 h.close()
                 self._cm.remove(h)
                 h = self._cm.get_ready_conn(host)
             else:
                 # no (working) free connections were found.  Create a new one.
                 h = http_class(host)
                 if DEBUG: DEBUG.info("creating new connection to %s (%d)",
                                      host, id(h))
                 self._cm.add(host, h, 0)
                 self._start_transaction(h, req)
                 r = h.getresponse()
         except (socket.error, httplib.HTTPException), err:
             raise urllib2.URLError(err)
 
         # if not a persistent connection, don't try to reuse it
         if r.will_close: self._cm.remove(h)
 
         if DEBUG: DEBUG.info("STATUS: %s, %s", r.status, r.reason)
         r._handler = self
         r._host = host
         r._url = req.get_full_url()
         r._connection = h
         r.code = r.status
         r.headers = r.msg
         r.msg = r.reason
 
         if r.status == 200 or not HANDLE_ERRORS:
             return r
         else:
             return self.parent.error('http', req, r,
                                      r.status, r.msg, r.headers)
 
     def _reuse_connection(self, h, req, host):
         """start the transaction with a re-used connection
         return a response object (r) upon success or None on failure.
         This DOES not close or remove bad connections in cases where
         it returns.  However, if an unexpected exception occurs, it
         will close and remove the connection before re-raising.
         """
         try:
             self._start_transaction(h, req)
             r = h.getresponse()
             # note: just because we got something back doesn't mean it
             # worked.  We'll check the version below, too.
         except (socket.error, httplib.HTTPException):
             r = None
         except:
             # adding this block just in case we've missed
             # something we will still raise the exception, but
             # lets try and close the connection and remove it
             # first.  We previously got into a nasty loop
             # where an exception was uncaught, and so the
             # connection stayed open.  On the next try, the
             # same exception was raised, etc.  The tradeoff is
             # that it's now possible this call will raise
             # a DIFFERENT exception
             if DEBUG: DEBUG.error("unexpected exception - closing " + \
                                   "connection to %s (%d)", host, id(h))
             self._cm.remove(h)
             h.close()
             raise
 
         if r is None or r.version == 9:
             # httplib falls back to assuming HTTP 0.9 if it gets a
             # bad header back.  This is most likely to happen if
             # the socket has been closed by the server since we
             # last used the connection.
             if DEBUG: DEBUG.info("failed to re-use connection to %s (%d)",
                                  host, id(h))
             r = None
         else:
             if DEBUG: DEBUG.info("re-using connection to %s (%d)", host, id(h))
 
         return r
 
     def _start_transaction(self, h, req):
         # What follows mostly reimplements HTTPConnection.request()
         # except it adds self.parent.addheaders in the mix.
         headers = req.headers.copy()
         if sys.version_info >= (2, 4):
             headers.update(req.unredirected_hdrs)
         headers.update(self.parent.addheaders)
         headers = dict((n.lower(), v) for n,v in headers.items())
         skipheaders = {}
         for n in ('host', 'accept-encoding'):
             if n in headers:
                 skipheaders['skip_' + n.replace('-', '_')] = 1
         try:
             if req.has_data():
                 data = req.get_data()
                 h.putrequest('POST', req.get_selector(), **skipheaders)
                 if 'content-type' not in headers:
                     h.putheader('Content-type',
                                 'application/x-www-form-urlencoded')
                 if 'content-length' not in headers:
                     h.putheader('Content-length', '%d' % len(data))
             else:
                 h.putrequest('GET', req.get_selector(), **skipheaders)
         except (socket.error), err:
             raise urllib2.URLError(err)
         for k, v in headers.items():
             h.putheader(k, v)
         h.endheaders()
         if req.has_data():
             h.send(data)
 
 class HTTPHandler(KeepAliveHandler, urllib2.HTTPHandler):
     pass
 
 class HTTPResponse(httplib.HTTPResponse):
     # we need to subclass HTTPResponse in order to
     # 1) add readline() and readlines() methods
     # 2) add close_connection() methods
     # 3) add info() and geturl() methods
 
     # in order to add readline(), read must be modified to deal with a
     # buffer.  example: readline must read a buffer and then spit back
     # one line at a time.  The only real alternative is to read one
     # BYTE at a time (ick).  Once something has been read, it can't be
     # put back (ok, maybe it can, but that's even uglier than this),
     # so if you THEN do a normal read, you must first take stuff from
     # the buffer.
 
     # the read method wraps the original to accomodate buffering,
     # although read() never adds to the buffer.
     # Both readline and readlines have been stolen with almost no
     # modification from socket.py
 
 
     def __init__(self, sock, debuglevel=0, strict=0, method=None):
         if method: # the httplib in python 2.3 uses the method arg
             httplib.HTTPResponse.__init__(self, sock, debuglevel, method)
         else: # 2.2 doesn't
             httplib.HTTPResponse.__init__(self, sock, debuglevel)
         self.fileno = sock.fileno
         self.code = None
         self._rbuf = ''
         self._rbufsize = 8096
         self._handler = None # inserted by the handler later
         self._host = None    # (same)
         self._url = None     # (same)
         self._connection = None # (same)
 
     _raw_read = httplib.HTTPResponse.read
 
     def close(self):
         if self.fp:
             self.fp.close()
             self.fp = None
             if self._handler:
                 self._handler._request_closed(self, self._host,
                                               self._connection)
 
     def close_connection(self):
         self._handler._remove_connection(self._host, self._connection, close=1)
         self.close()
 
     def info(self):
         return self.headers
 
     def geturl(self):
         return self._url
 
     def read(self, amt=None):
         # the _rbuf test is only in this first if for speed.  It's not
         # logically necessary
         if self._rbuf and not amt is None:
             L = len(self._rbuf)
             if amt > L:
                 amt -= L
             else:
                 s = self._rbuf[:amt]
                 self._rbuf = self._rbuf[amt:]
                 return s
 
         s = self._rbuf + self._raw_read(amt)
         self._rbuf = ''
         return s
 
     # stolen from Python SVN #68532 to fix issue1088
     def _read_chunked(self, amt):
         chunk_left = self.chunk_left
         value = ''
 
         # XXX This accumulates chunks by repeated string concatenation,
         # which is not efficient as the number or size of chunks gets big.
         while True:
             if chunk_left is None:
                 line = self.fp.readline()
                 i = line.find(';')
                 if i >= 0:
                     line = line[:i] # strip chunk-extensions
                 try:
                     chunk_left = int(line, 16)
                 except ValueError:
                     # close the connection as protocol synchronisation is
                     # probably lost
                     self.close()
                     raise httplib.IncompleteRead(value)
                 if chunk_left == 0:
                     break
             if amt is None:
                 value += self._safe_read(chunk_left)
             elif amt < chunk_left:
                 value += self._safe_read(amt)
                 self.chunk_left = chunk_left - amt
                 return value
             elif amt == chunk_left:
                 value += self._safe_read(amt)
                 self._safe_read(2)  # toss the CRLF at the end of the chunk
                 self.chunk_left = None
                 return value
             else:
                 value += self._safe_read(chunk_left)
                 amt -= chunk_left
 
             # we read the whole chunk, get another
             self._safe_read(2)      # toss the CRLF at the end of the chunk
             chunk_left = None
 
         # read and discard trailer up to the CRLF terminator
         ### note: we shouldn't have any trailers!
         while True:
             line = self.fp.readline()
             if not line:
                 # a vanishingly small number of sites EOF without
                 # sending the trailer
                 break
             if line == '\r\n':
                 break
 
         # we read everything; close the "file"
         self.close()
 
         return value
 
     def readline(self, limit=-1):
         i = self._rbuf.find('\n')
         while i < 0 and not (0 < limit <= len(self._rbuf)):
             new = self._raw_read(self._rbufsize)
             if not new: break
             i = new.find('\n')
             if i >= 0: i = i + len(self._rbuf)
             self._rbuf = self._rbuf + new
         if i < 0: i = len(self._rbuf)
         else: i = i+1
         if 0 <= limit < len(self._rbuf): i = limit
         data, self._rbuf = self._rbuf[:i], self._rbuf[i:]
         return data
 
     def readlines(self, sizehint = 0):
         total = 0
         list = []
         while 1:
             line = self.readline()
             if not line: break
             list.append(line)
             total += len(line)
             if sizehint and total >= sizehint:
                 break
         return list
 
 
 class HTTPConnection(httplib.HTTPConnection):
     # use the modified response class
     response_class = HTTPResponse
 
 #########################################################################
 #####   TEST FUNCTIONS
 #########################################################################
 
 def error_handler(url):
     global HANDLE_ERRORS
     orig = HANDLE_ERRORS
     keepalive_handler = HTTPHandler()
     opener = urllib2.build_opener(keepalive_handler)
     urllib2.install_opener(opener)
     pos = {0: 'off', 1: 'on'}
     for i in (0, 1):
         print "  fancy error handling %s (HANDLE_ERRORS = %i)" % (pos[i], i)
         HANDLE_ERRORS = i
         try:
             fo = urllib2.urlopen(url)
             fo.read()
             fo.close()
             try: status, reason = fo.status, fo.reason
             except AttributeError: status, reason = None, None
         except IOError, e:
             print "  EXCEPTION: %s" % e
             raise
         else:
             print "  status = %s, reason = %s" % (status, reason)
     HANDLE_ERRORS = orig
     hosts = keepalive_handler.open_connections()
     print "open connections:", hosts
     keepalive_handler.close_all()
 
+def md5(s):
+    try:
+        from hashlib import md5 as _md5
+    except ImportError:
+        from md5 import md5 as _md5
+    global md5
+    md5 = _md5
+    return _md5(s)
+
 def continuity(url):
-    from util import md5
     format = '%25s: %s'
 
     # first fetch the file with the normal http handler
     opener = urllib2.build_opener()
     urllib2.install_opener(opener)
     fo = urllib2.urlopen(url)
     foo = fo.read()
     fo.close()
     m = md5.new(foo)
     print format % ('normal urllib', m.hexdigest())
 
     # now install the keepalive handler and try again
     opener = urllib2.build_opener(HTTPHandler())
     urllib2.install_opener(opener)
 
     fo = urllib2.urlopen(url)
     foo = fo.read()
     fo.close()
     m = md5.new(foo)
     print format % ('keepalive read', m.hexdigest())
 
     fo = urllib2.urlopen(url)
     foo = ''
     while 1:
         f = fo.readline()
         if f: foo = foo + f
         else: break
     fo.close()
     m = md5.new(foo)
     print format % ('keepalive readline', m.hexdigest())
 
 def comp(N, url):
     print '  making %i connections to:\n  %s' % (N, url)
 
     sys.stdout.write('  first using the normal urllib handlers')
     # first use normal opener
     opener = urllib2.build_opener()
     urllib2.install_opener(opener)
     t1 = fetch(N, url)
     print '  TIME: %.3f s' % t1
 
     sys.stdout.write('  now using the keepalive handler       ')
     # now install the keepalive handler and try again
     opener = urllib2.build_opener(HTTPHandler())
     urllib2.install_opener(opener)
     t2 = fetch(N, url)
     print '  TIME: %.3f s' % t2
     print '  improvement factor: %.2f' % (t1/t2, )
 
 def fetch(N, url, delay=0):
     import time
     lens = []
     starttime = time.time()
     for i in range(N):
         if delay and i > 0: time.sleep(delay)
         fo = urllib2.urlopen(url)
         foo = fo.read()
         fo.close()
         lens.append(len(foo))
     diff = time.time() - starttime
 
     j = 0
     for i in lens[1:]:
         j = j + 1
         if not i == lens[0]:
             print "WARNING: inconsistent length on read %i: %i" % (j, i)
 
     return diff
 
 def test_timeout(url):
     global DEBUG
     dbbackup = DEBUG
     class FakeLogger:
         def debug(self, msg, *args): print msg % args
         info = warning = error = debug
     DEBUG = FakeLogger()
     print "  fetching the file to establish a connection"
     fo = urllib2.urlopen(url)
     data1 = fo.read()
     fo.close()
 
     i = 20
     print "  waiting %i seconds for the server to close the connection" % i
     while i > 0:
         sys.stdout.write('\r  %2i' % i)
         sys.stdout.flush()
         time.sleep(1)
         i -= 1
     sys.stderr.write('\r')
 
     print "  fetching the file a second time"
     fo = urllib2.urlopen(url)
     data2 = fo.read()
     fo.close()
 
     if data1 == data2:
         print '  data are identical'
     else:
         print '  ERROR: DATA DIFFER'
 
     DEBUG = dbbackup
 
 
 def test(url, N=10):
     print "checking error hander (do this on a non-200)"
     try: error_handler(url)
     except IOError:
         print "exiting - exception will prevent further tests"
         sys.exit()
     print
     print "performing continuity test (making sure stuff isn't corrupted)"
     continuity(url)
     print
     print "performing speed comparison"
     comp(N, url)
     print
     print "performing dropped-connection check"
     test_timeout(url)
 
 if __name__ == '__main__':
     import time
     import sys
     try:
         N = int(sys.argv[1])
         url = sys.argv[2]
     except:
         print "%s <integer> <url>" % sys.argv[0]
     else:
         test(url, N)

mercurial/util.py

@@ -1,1481 +1,1471 b''
 # util.py - Mercurial utility functions and platform specfic implementations
 #
 #  Copyright 2005 K. Thananchayan <[e-mail]>
 #  Copyright 2005-2007 Matt Mackall <[e-mail]>
 #  Copyright 2006 Vadim Gelfer <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2, incorporated herein by reference.
 
 """Mercurial utility functions and platform specfic implementations.
 
 This contains helper routines that are independent of the SCM core and
 hide platform-specific details from the core.
 """
 
 from i18n import _
 import cStringIO, errno, re, shutil, sys, tempfile, traceback, error
 import os, stat, threading, time, calendar, glob, osutil, random
 import imp
 
 # Python compatibility
 
-def md5(s):
-    try:
-        import hashlib
-        _md5 = hashlib.md5
-    except ImportError:
-        from md5 import md5 as _md5
-    global md5
-    md5 = _md5
-    return _md5(s)
-
 def sha1(s):
     try:
         import hashlib
         _sha1 = hashlib.sha1
     except ImportError:
         from sha import sha as _sha1
     global sha1
     sha1 = _sha1
     return _sha1(s)
 
 import subprocess
 closefds = os.name == 'posix'
 def popen2(cmd, mode='t', bufsize=-1):
     p = subprocess.Popen(cmd, shell=True, bufsize=bufsize,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE)
     return p.stdin, p.stdout
 def popen3(cmd, mode='t', bufsize=-1):
     p = subprocess.Popen(cmd, shell=True, bufsize=bufsize,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          stderr=subprocess.PIPE)
     return p.stdin, p.stdout, p.stderr
 def Popen3(cmd, capturestderr=False, bufsize=-1):
     stderr = capturestderr and subprocess.PIPE or None
     p = subprocess.Popen(cmd, shell=True, bufsize=bufsize,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          stderr=stderr)
     p.fromchild = p.stdout
     p.tochild = p.stdin
     p.childerr = p.stderr
     return p
 
 def version():
     """Return version information if available."""
     try:
         import __version__
         return __version__.version
     except ImportError:
         return 'unknown'
 
 # used by parsedate
 defaultdateformats = (
     '%Y-%m-%d %H:%M:%S',
     '%Y-%m-%d %I:%M:%S%p',
     '%Y-%m-%d %H:%M',
     '%Y-%m-%d %I:%M%p',
     '%Y-%m-%d',
     '%m-%d',
     '%m/%d',
     '%m/%d/%y',
     '%m/%d/%Y',
     '%a %b %d %H:%M:%S %Y',
     '%a %b %d %I:%M:%S%p %Y',
     '%a, %d %b %Y %H:%M:%S',        #  GNU coreutils "/bin/date --rfc-2822"
     '%b %d %H:%M:%S %Y',
     '%b %d %I:%M:%S%p %Y',
     '%b %d %H:%M:%S',
     '%b %d %I:%M:%S%p',
     '%b %d %H:%M',
     '%b %d %I:%M%p',
     '%b %d %Y',
     '%b %d',
     '%H:%M:%S',
     '%I:%M:%SP',
     '%H:%M',
     '%I:%M%p',
 )
 
 extendeddateformats = defaultdateformats + (
     "%Y",
     "%Y-%m",
     "%b",
     "%b %Y",
     )
 
 def cachefunc(func):
     '''cache the result of function calls'''
     # XXX doesn't handle keywords args
     cache = {}
     if func.func_code.co_argcount == 1:
         # we gain a small amount of time because
         # we don't need to pack/unpack the list
         def f(arg):
             if arg not in cache:
                 cache[arg] = func(arg)
             return cache[arg]
     else:
         def f(*args):
             if args not in cache:
                 cache[args] = func(*args)
             return cache[args]
 
     return f
 
 class propertycache(object):
     def __init__(self, func):
         self.func = func
         self.name = func.__name__
     def __get__(self, obj, type=None):
         result = self.func(obj)
         setattr(obj, self.name, result)
         return result
 
 def pipefilter(s, cmd):
     '''filter string S through command CMD, returning its output'''
     (pin, pout) = popen2(cmd, 'b')
     def writer():
         try:
             pin.write(s)
             pin.close()
         except IOError, inst:
             if inst.errno != errno.EPIPE:
                 raise
 
     # we should use select instead on UNIX, but this will work on most
     # systems, including Windows
     w = threading.Thread(target=writer)
     w.start()
     f = pout.read()
     pout.close()
     w.join()
     return f
 
 def tempfilter(s, cmd):
     '''filter string S through a pair of temporary files with CMD.
     CMD is used as a template to create the real command to be run,
     with the strings INFILE and OUTFILE replaced by the real names of
     the temporary files generated.'''
     inname, outname = None, None
     try:
         infd, inname = tempfile.mkstemp(prefix='hg-filter-in-')
         fp = os.fdopen(infd, 'wb')
         fp.write(s)
         fp.close()
         outfd, outname = tempfile.mkstemp(prefix='hg-filter-out-')
         os.close(outfd)
         cmd = cmd.replace('INFILE', inname)
         cmd = cmd.replace('OUTFILE', outname)
         code = os.system(cmd)
         if sys.platform == 'OpenVMS' and code & 1:
             code = 0
         if code: raise Abort(_("command '%s' failed: %s") %
                              (cmd, explain_exit(code)))
         return open(outname, 'rb').read()
     finally:
         try:
             if inname: os.unlink(inname)
         except: pass
         try:
             if outname: os.unlink(outname)
         except: pass
 
 filtertable = {
     'tempfile:': tempfilter,
     'pipe:': pipefilter,
     }
 
 def filter(s, cmd):
     "filter a string through a command that transforms its input to its output"
     for name, fn in filtertable.iteritems():
         if cmd.startswith(name):
             return fn(s, cmd[len(name):].lstrip())
     return pipefilter(s, cmd)
 
 def binary(s):
     """return true if a string is binary data"""
     return bool(s and '\0' in s)
 
 def increasingchunks(source, min=1024, max=65536):
     '''return no less than min bytes per chunk while data remains,
     doubling min after each chunk until it reaches max'''
     def log2(x):
         if not x:
             return 0
         i = 0
         while x:
             x >>= 1
             i += 1
         return i - 1
 
     buf = []
     blen = 0
     for chunk in source:
         buf.append(chunk)
         blen += len(chunk)
         if blen >= min:
             if min < max:
                 min = min << 1
                 nmin = 1 << log2(blen)
                 if nmin > min:
                     min = nmin
                 if min > max:
                     min = max
             yield ''.join(buf)
             blen = 0
             buf = []
     if buf:
         yield ''.join(buf)
 
 Abort = error.Abort
 
 def always(fn): return True
 def never(fn): return False
 
 def patkind(name, default):
     """Split a string into an optional pattern kind prefix and the
     actual pattern."""
     for prefix in 're', 'glob', 'path', 'relglob', 'relpath', 'relre':
         if name.startswith(prefix + ':'): return name.split(':', 1)
     return default, name
 
 def globre(pat, head='^', tail='$'):
     "convert a glob pattern into a regexp"
     i, n = 0, len(pat)
     res = ''
     group = 0
     def peek(): return i < n and pat[i]
     while i < n:
         c = pat[i]
         i = i+1
         if c == '*':
             if peek() == '*':
                 i += 1
                 res += '.*'
             else:
                 res += '[^/]*'
         elif c == '?':
             res += '.'
         elif c == '[':
             j = i
             if j < n and pat[j] in '!]':
                 j += 1
             while j < n and pat[j] != ']':
                 j += 1
             if j >= n:
                 res += '\\['
             else:
                 stuff = pat[i:j].replace('\\','\\\\')
                 i = j + 1
                 if stuff[0] == '!':
                     stuff = '^' + stuff[1:]
                 elif stuff[0] == '^':
                     stuff = '\\' + stuff
                 res = '%s[%s]' % (res, stuff)
         elif c == '{':
             group += 1
             res += '(?:'
         elif c == '}' and group:
             res += ')'
             group -= 1
         elif c == ',' and group:
             res += '|'
         elif c == '\\':
             p = peek()
             if p:
                 i += 1
                 res += re.escape(p)
             else:
                 res += re.escape(c)
         else:
             res += re.escape(c)
     return head + res + tail
 
 _globchars = {'[': 1, '{': 1, '*': 1, '?': 1}
 
 def pathto(root, n1, n2):
     '''return the relative path from one place to another.
     root should use os.sep to separate directories
     n1 should use os.sep to separate directories
     n2 should use "/" to separate directories
     returns an os.sep-separated path.
 
     If n1 is a relative path, it's assumed it's
     relative to root.
     n2 should always be relative to root.
     '''
     if not n1: return localpath(n2)
     if os.path.isabs(n1):
         if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]:
             return os.path.join(root, localpath(n2))
         n2 = '/'.join((pconvert(root), n2))
     a, b = splitpath(n1), n2.split('/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return os.sep.join((['..'] * len(a)) + b) or '.'
 
 def canonpath(root, cwd, myname):
     """return the canonical path of myname, given cwd and root"""
     if root == os.sep:
         rootsep = os.sep
     elif endswithsep(root):
         rootsep = root
     else:
         rootsep = root + os.sep
     name = myname
     if not os.path.isabs(name):
         name = os.path.join(root, cwd, name)
     name = os.path.normpath(name)
     audit_path = path_auditor(root)
     if name != rootsep and name.startswith(rootsep):
         name = name[len(rootsep):]
         audit_path(name)
         return pconvert(name)
     elif name == root:
         return ''
     else:
         # Determine whether `name' is in the hierarchy at or beneath `root',
         # by iterating name=dirname(name) until that causes no change (can't
         # check name == '/', because that doesn't work on windows).  For each
         # `name', compare dev/inode numbers.  If they match, the list `rel'
         # holds the reversed list of components making up the relative file
         # name we want.
         root_st = os.stat(root)
         rel = []
         while True:
             try:
                 name_st = os.stat(name)
             except OSError:
                 break
             if samestat(name_st, root_st):
                 if not rel:
                     # name was actually the same as root (maybe a symlink)
                     return ''
                 rel.reverse()
                 name = os.path.join(*rel)
                 audit_path(name)
                 return pconvert(name)
             dirname, basename = os.path.split(name)
             rel.append(basename)
             if dirname == name:
                 break
             name = dirname
 
         raise Abort('%s not under root' % myname)
 
 def matcher(canonroot, cwd='', names=[], inc=[], exc=[], src=None, dflt_pat='glob'):
     """build a function to match a set of file patterns
 
     arguments:
     canonroot - the canonical root of the tree you're matching against
     cwd - the current working directory, if relevant
     names - patterns to find
     inc - patterns to include
     exc - patterns to exclude
     dflt_pat - if a pattern in names has no explicit type, assume this one
     src - where these patterns came from (e.g. .hgignore)
 
     a pattern is one of:
     'glob:<glob>' - a glob relative to cwd
     're:<regexp>' - a regular expression
     'path:<path>' - a path relative to canonroot
     'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs)
     'relpath:<path>' - a path relative to cwd
     'relre:<regexp>' - a regexp that doesn't have to match the start of a name
     '<something>' - one of the cases above, selected by the dflt_pat argument
 
     returns:
     a 3-tuple containing
     - list of roots (places where one should start a recursive walk of the fs);
       this often matches the explicit non-pattern names passed in, but also
       includes the initial part of glob: patterns that has no glob characters
     - a bool match(filename) function
     - a bool indicating if any patterns were passed in
     """
 
     # a common case: no patterns at all
     if not names and not inc and not exc:
         return [], always, False
 
     def contains_glob(name):
         for c in name:
             if c in _globchars: return True
         return False
 
     def regex(kind, name, tail):
         '''convert a pattern into a regular expression'''
         if not name:
             return ''
         if kind == 're':
             return name
         elif kind == 'path':
             return '^' + re.escape(name) + '(?:/|$)'
         elif kind == 'relglob':
             return globre(name, '(?:|.*/)', tail)
         elif kind == 'relpath':
             return re.escape(name) + '(?:/|$)'
         elif kind == 'relre':
             if name.startswith('^'):
                 return name
             return '.*' + name
         return globre(name, '', tail)
 
     def matchfn(pats, tail):
         """build a matching function from a set of patterns"""
         if not pats:
             return
         try:
             pat = '(?:%s)' % '|'.join([regex(k, p, tail) for (k, p) in pats])
             if len(pat) > 20000:
                 raise OverflowError()
             return re.compile(pat).match
         except OverflowError:
             # We're using a Python with a tiny regex engine and we
             # made it explode, so we'll divide the pattern list in two
             # until it works
             l = len(pats)
             if l < 2:
                 raise
             a, b = matchfn(pats[:l//2], tail), matchfn(pats[l//2:], tail)
             return lambda s: a(s) or b(s)
         except re.error:
             for k, p in pats:
                 try:
                     re.compile('(?:%s)' % regex(k, p, tail))
                 except re.error:
                     if src:
                         raise Abort("%s: invalid pattern (%s): %s" %
                                     (src, k, p))
                     else:
                         raise Abort("invalid pattern (%s): %s" % (k, p))
             raise Abort("invalid pattern")
 
     def globprefix(pat):
         '''return the non-glob prefix of a path, e.g. foo/* -> foo'''
         root = []
         for p in pat.split('/'):
             if contains_glob(p): break
             root.append(p)
         return '/'.join(root) or '.'
 
     def normalizepats(names, default):
         pats = []
         roots = []
         anypats = False
         for kind, name in [patkind(p, default) for p in names]:
             if kind in ('glob', 'relpath'):
                 name = canonpath(canonroot, cwd, name)
             elif kind in ('relglob', 'path'):
                 name = normpath(name)
 
             pats.append((kind, name))
 
             if kind in ('glob', 're', 'relglob', 'relre'):
                 anypats = True
 
             if kind == 'glob':
                 root = globprefix(name)
                 roots.append(root)
             elif kind in ('relpath', 'path'):
                 roots.append(name or '.')
             elif kind == 'relglob':
                 roots.append('.')
         return roots, pats, anypats
 
     roots, pats, anypats = normalizepats(names, dflt_pat)
 
     patmatch = matchfn(pats, '$') or always
     incmatch = always
     if inc:
         dummy, inckinds, dummy = normalizepats(inc, 'glob')
         incmatch = matchfn(inckinds, '(?:/|$)')
     excmatch = never
     if exc:
         dummy, exckinds, dummy = normalizepats(exc, 'glob')
         excmatch = matchfn(exckinds, '(?:/|$)')
 
     if not names and inc and not exc:
         # common case: hgignore patterns
         match = incmatch
     else:
         match = lambda fn: incmatch(fn) and not excmatch(fn) and patmatch(fn)
 
     return (roots, match, (inc or exc or anypats) and True)
 
 _hgexecutable = None
 
 def main_is_frozen():
     """return True if we are a frozen executable.
 
     The code supports py2exe (most common, Windows only) and tools/freeze
     (portable, not much used).
     """
     return (hasattr(sys, "frozen") or # new py2exe
             hasattr(sys, "importers") or # old py2exe
             imp.is_frozen("__main__")) # tools/freeze
 
 def hgexecutable():
     """return location of the 'hg' executable.
 
     Defaults to $HG or 'hg' in the search path.
     """
     if _hgexecutable is None:
         hg = os.environ.get('HG')
         if hg:
             set_hgexecutable(hg)
         elif main_is_frozen():
             set_hgexecutable(sys.executable)
         else:
             set_hgexecutable(find_exe('hg') or 'hg')
     return _hgexecutable
 
 def set_hgexecutable(path):
     """set location of the 'hg' executable"""
     global _hgexecutable
     _hgexecutable = path
 
 def system(cmd, environ={}, cwd=None, onerr=None, errprefix=None):
     '''enhanced shell command execution.
     run with environment maybe modified, maybe in different dir.
 
     if command fails and onerr is None, return status.  if ui object,
     print error message and return status, else raise onerr object as
     exception.'''
     def py2shell(val):
         'convert python object into string that is useful to shell'
         if val in (None, False):
             return '0'
         if val == True:
             return '1'
         return str(val)
     oldenv = {}
     for k in environ:
         oldenv[k] = os.environ.get(k)
     if cwd is not None:
         oldcwd = os.getcwd()
     origcmd = cmd
     if os.name == 'nt':
         cmd = '"%s"' % cmd
     try:
         for k, v in environ.iteritems():
             os.environ[k] = py2shell(v)
         os.environ['HG'] = hgexecutable()
         if cwd is not None and oldcwd != cwd:
             os.chdir(cwd)
         rc = os.system(cmd)
         if sys.platform == 'OpenVMS' and rc & 1:
             rc = 0
         if rc and onerr:
             errmsg = '%s %s' % (os.path.basename(origcmd.split(None, 1)[0]),
                                 explain_exit(rc)[0])
             if errprefix:
                 errmsg = '%s: %s' % (errprefix, errmsg)
             try:
                 onerr.warn(errmsg + '\n')
             except AttributeError:
                 raise onerr(errmsg)
         return rc
     finally:
         for k, v in oldenv.iteritems():
             if v is None:
                 del os.environ[k]
             else:
                 os.environ[k] = v
         if cwd is not None and oldcwd != cwd:
             os.chdir(oldcwd)
 
 def checksignature(func):
     '''wrap a function with code to check for calling errors'''
     def check(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except TypeError:
             if len(traceback.extract_tb(sys.exc_info()[2])) == 1:
                 raise error.SignatureError
             raise
 
     return check
 
 # os.path.lexists is not available on python2.3
 def lexists(filename):
     "test whether a file with this name exists. does not follow symlinks"
     try:
         os.lstat(filename)
     except:
         return False
     return True
 
 def rename(src, dst):
     """forcibly rename a file"""
     try:
         os.rename(src, dst)
     except OSError, err: # FIXME: check err (EEXIST ?)
 
         # On windows, rename to existing file is not allowed, so we
         # must delete destination first. But if a file is open, unlink
         # schedules it for delete but does not delete it. Rename
         # happens immediately even for open files, so we rename
         # destination to a temporary name, then delete that. Then
         # rename is safe to do.
         # The temporary name is chosen at random to avoid the situation
         # where a file is left lying around from a previous aborted run.
         # The usual race condition this introduces can't be avoided as
         # we need the name to rename into, and not the file itself. Due
         # to the nature of the operation however, any races will at worst
         # lead to the rename failing and the current operation aborting.
 
         def tempname(prefix):
             for tries in xrange(10):
                 temp = '%s-%08x' % (prefix, random.randint(0, 0xffffffff))
                 if not os.path.exists(temp):
                     return temp
             raise IOError, (errno.EEXIST, "No usable temporary filename found")
 
         temp = tempname(dst)
         os.rename(dst, temp)
         os.unlink(temp)
         os.rename(src, dst)
 
 def unlink(f):
     """unlink and remove the directory if it is empty"""
     os.unlink(f)
     # try removing directories that might now be empty
     try:
         os.removedirs(os.path.dirname(f))
     except OSError:
         pass
 
 def copyfile(src, dest):
     "copy a file, preserving mode and atime/mtime"
     if os.path.islink(src):
         try:
             os.unlink(dest)
         except:
             pass
         os.symlink(os.readlink(src), dest)
     else:
         try:
             shutil.copyfile(src, dest)
             shutil.copystat(src, dest)
         except shutil.Error, inst:
             raise Abort(str(inst))
 
 def copyfiles(src, dst, hardlink=None):
     """Copy a directory tree using hardlinks if possible"""
 
     if hardlink is None:
         hardlink = (os.stat(src).st_dev ==
                     os.stat(os.path.dirname(dst)).st_dev)
 
     if os.path.isdir(src):
         os.mkdir(dst)
         for name, kind in osutil.listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             copyfiles(srcname, dstname, hardlink)
     else:
         if hardlink:
             try:
                 os_link(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
 
 class path_auditor(object):
     '''ensure that a filesystem path contains no banned components.
     the following properties of a path are checked:
 
     - under top-level .hg
     - starts at the root of a windows drive
     - contains ".."
     - traverses a symlink (e.g. a/symlink_here/b)
     - inside a nested repository'''
 
     def __init__(self, root):
         self.audited = set()
         self.auditeddir = set()
         self.root = root
 
     def __call__(self, path):
         if path in self.audited:
             return
         normpath = os.path.normcase(path)
         parts = splitpath(normpath)
         if (os.path.splitdrive(path)[0]
             or parts[0].lower() in ('.hg', '.hg.', '')
             or os.pardir in parts):
             raise Abort(_("path contains illegal component: %s") % path)
         if '.hg' in path.lower():
             lparts = [p.lower() for p in parts]
             for p in '.hg', '.hg.':
                 if p in lparts[1:]:
                     pos = lparts.index(p)
                     base = os.path.join(*parts[:pos])
                     raise Abort(_('path %r is inside repo %r') % (path, base))
         def check(prefix):
             curpath = os.path.join(self.root, prefix)
             try:
                 st = os.lstat(curpath)
             except OSError, err:
                 # EINVAL can be raised as invalid path syntax under win32.
                 # They must be ignored for patterns can be checked too.
                 if err.errno not in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
                     raise
             else:
                 if stat.S_ISLNK(st.st_mode):
                     raise Abort(_('path %r traverses symbolic link %r') %
                                 (path, prefix))
                 elif (stat.S_ISDIR(st.st_mode) and
                       os.path.isdir(os.path.join(curpath, '.hg'))):
                     raise Abort(_('path %r is inside repo %r') %
                                 (path, prefix))
         parts.pop()
         prefixes = []
         for n in range(len(parts)):
             prefix = os.sep.join(parts)
             if prefix in self.auditeddir:
                 break
             check(prefix)
             prefixes.append(prefix)
             parts.pop()
 
         self.audited.add(path)
         # only add prefixes to the cache after checking everything: we don't
         # want to add "foo/bar/baz" before checking if there's a "foo/.hg"
         self.auditeddir.update(prefixes)
 
 def nlinks(pathname):
     """Return number of hardlinks for the given file."""
     return os.lstat(pathname).st_nlink
 
 if hasattr(os, 'link'):
     os_link = os.link
 else:
     def os_link(src, dst):
         raise OSError(0, _("Hardlinks not supported"))
 
 def lookup_reg(key, name=None, scope=None):
     return None
 
 if os.name == 'nt':
     from windows import *
     def expand_glob(pats):
         '''On Windows, expand the implicit globs in a list of patterns'''
         ret = []
         for p in pats:
             kind, name = patkind(p, None)
             if kind is None:
                 globbed = glob.glob(name)
                 if globbed:
                     ret.extend(globbed)
                     continue
                 # if we couldn't expand the glob, just keep it around
             ret.append(p)
         return ret
 else:
     from posix import *
 
 def makelock(info, pathname):
     try:
         return os.symlink(info, pathname)
     except OSError, why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError: # no symlink in os
         pass
 
     ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL)
     os.write(ld, info)
     os.close(ld)
 
 def readlock(pathname):
     try:
         return os.readlink(pathname)
     except OSError, why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError: # no symlink in os
         pass
     return posixfile(pathname).read()
 
 def fstat(fp):
     '''stat file object that may not have fileno method.'''
     try:
         return os.fstat(fp.fileno())
     except AttributeError:
         return os.stat(fp.name)
 
 # File system features
 
 def checkcase(path):
     """
     Check whether the given path is on a case-sensitive filesystem
 
     Requires a path (like /foo/.hg) ending with a foldable final
     directory component.
     """
     s1 = os.stat(path)
     d, b = os.path.split(path)
     p2 = os.path.join(d, b.upper())
     if path == p2:
         p2 = os.path.join(d, b.lower())
     try:
         s2 = os.stat(p2)
         if s2 == s1:
             return False
         return True
     except:
         return True
 
 _fspathcache = {}
 def fspath(name, root):
     '''Get name in the case stored in the filesystem
 
     The name is either relative to root, or it is an absolute path starting
     with root. Note that this function is unnecessary, and should not be
     called, for case-sensitive filesystems (simply because it's expensive).
     '''
     # If name is absolute, make it relative
     if name.lower().startswith(root.lower()):
         l = len(root)
         if name[l] == os.sep or name[l] == os.altsep:
             l = l + 1
         name = name[l:]
 
     if not os.path.exists(os.path.join(root, name)):
         return None
 
     seps = os.sep
     if os.altsep:
         seps = seps + os.altsep
     # Protect backslashes. This gets silly very quickly.
     seps.replace('\\','\\\\')
     pattern = re.compile(r'([^%s]+)|([%s]+)' % (seps, seps))
     dir = os.path.normcase(os.path.normpath(root))
     result = []
     for part, sep in pattern.findall(name):
         if sep:
             result.append(sep)
             continue
 
         if dir not in _fspathcache:
             _fspathcache[dir] = os.listdir(dir)
         contents = _fspathcache[dir]
 
         lpart = part.lower()
         for n in contents:
             if n.lower() == lpart:
                 result.append(n)
                 break
         else:
             # Cannot happen, as the file exists!
             result.append(part)
         dir = os.path.join(dir, lpart)
 
     return ''.join(result)
 
 def checkexec(path):
     """
     Check whether the given path is on a filesystem with UNIX-like exec flags
 
     Requires a directory (like /foo/.hg)
     """
 
     # VFAT on some Linux versions can flip mode but it doesn't persist
     # a FS remount. Frequently we can detect it if files are created
     # with exec bit on.
 
     try:
         EXECFLAGS = stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH
         fh, fn = tempfile.mkstemp("", "", path)
         try:
             os.close(fh)
             m = os.stat(fn).st_mode & 0777
             new_file_has_exec = m & EXECFLAGS
             os.chmod(fn, m ^ EXECFLAGS)
             exec_flags_cannot_flip = ((os.stat(fn).st_mode & 0777) == m)
         finally:
             os.unlink(fn)
     except (IOError, OSError):
         # we don't care, the user probably won't be able to commit anyway
         return False
     return not (new_file_has_exec or exec_flags_cannot_flip)
 
 def checklink(path):
     """check whether the given path is on a symlink-capable filesystem"""
     # mktemp is not racy because symlink creation will fail if the
     # file already exists
     name = tempfile.mktemp(dir=path)
     try:
         os.symlink(".", name)
         os.unlink(name)
         return True
     except (OSError, AttributeError):
         return False
 
 def needbinarypatch():
     """return True if patches should be applied in binary mode by default."""
     return os.name == 'nt'
 
 def endswithsep(path):
     '''Check path ends with os.sep or os.altsep.'''
     return path.endswith(os.sep) or os.altsep and path.endswith(os.altsep)
 
 def splitpath(path):
     '''Split path by os.sep.
     Note that this function does not use os.altsep because this is
     an alternative of simple "xxx.split(os.sep)".
     It is recommended to use os.path.normpath() before using this
     function if need.'''
     return path.split(os.sep)
 
 def gui():
     '''Are we running in a GUI?'''
     return os.name == "nt" or os.name == "mac" or os.environ.get("DISPLAY")
 
 def mktempcopy(name, emptyok=False, createmode=None):
     """Create a temporary file with the same contents from name
 
     The permission bits are copied from the original file.
 
     If the temporary file is going to be truncated immediately, you
     can use emptyok=True as an optimization.
 
     Returns the name of the temporary file.
     """
     d, fn = os.path.split(name)
     fd, temp = tempfile.mkstemp(prefix='.%s-' % fn, dir=d)
     os.close(fd)
     # Temporary files are created with mode 0600, which is usually not
     # what we want.  If the original file already exists, just copy
     # its mode.  Otherwise, manually obey umask.
     try:
         st_mode = os.lstat(name).st_mode & 0777
     except OSError, inst:
         if inst.errno != errno.ENOENT:
             raise
         st_mode = createmode
         if st_mode is None:
             st_mode = ~umask
         st_mode &= 0666
     os.chmod(temp, st_mode)
     if emptyok:
         return temp
     try:
         try:
             ifp = posixfile(name, "rb")
         except IOError, inst:
             if inst.errno == errno.ENOENT:
                 return temp
             if not getattr(inst, 'filename', None):
                 inst.filename = name
             raise
         ofp = posixfile(temp, "wb")
         for chunk in filechunkiter(ifp):
             ofp.write(chunk)
         ifp.close()
         ofp.close()
     except:
         try: os.unlink(temp)
         except: pass
         raise
     return temp
 
 class atomictempfile(posixfile):
     """file-like object that atomically updates a file
 
     All writes will be redirected to a temporary copy of the original
     file.  When rename is called, the copy is renamed to the original
     name, making the changes visible.
     """
     def __init__(self, name, mode, createmode):
         self.__name = name
         self.temp = mktempcopy(name, emptyok=('w' in mode),
                                createmode=createmode)
         posixfile.__init__(self, self.temp, mode)
 
     def rename(self):
         if not self.closed:
             posixfile.close(self)
             rename(self.temp, localpath(self.__name))
 
     def __del__(self):
         if not self.closed:
             try:
                 os.unlink(self.temp)
             except: pass
             posixfile.close(self)
 
 def makedirs(name, mode=None):
     """recursive directory creation with parent mode inheritance"""
     try:
         os.mkdir(name)
         if mode is not None:
             os.chmod(name, mode)
         return
     except OSError, err:
         if err.errno == errno.EEXIST:
             return
         if err.errno != errno.ENOENT:
             raise
     parent = os.path.abspath(os.path.dirname(name))
     makedirs(parent, mode)
     makedirs(name, mode)
 
 class opener(object):
     """Open files relative to a base directory
 
     This class is used to hide the details of COW semantics and
     remote file access from higher level code.
     """
     def __init__(self, base, audit=True):
         self.base = base
         if audit:
             self.audit_path = path_auditor(base)
         else:
             self.audit_path = always
         self.createmode = None
 
     def __getattr__(self, name):
         if name == '_can_symlink':
             self._can_symlink = checklink(self.base)
             return self._can_symlink
         raise AttributeError(name)
 
     def _fixfilemode(self, name):
         if self.createmode is None:
             return
         os.chmod(name, self.createmode & 0666)
 
     def __call__(self, path, mode="r", text=False, atomictemp=False):
         self.audit_path(path)
         f = os.path.join(self.base, path)
 
         if not text and "b" not in mode:
             mode += "b" # for that other OS
 
         nlink = -1
         if mode not in ("r", "rb"):
             try:
                 nlink = nlinks(f)
             except OSError:
                 nlink = 0
                 d = os.path.dirname(f)
                 if not os.path.isdir(d):
                     makedirs(d, self.createmode)
             if atomictemp:
                 return atomictempfile(f, mode, self.createmode)
             if nlink > 1:
                 rename(mktempcopy(f), f)
         fp = posixfile(f, mode)
         if nlink == 0:
             self._fixfilemode(f)
         return fp
 
     def symlink(self, src, dst):
         self.audit_path(dst)
         linkname = os.path.join(self.base, dst)
         try:
             os.unlink(linkname)
         except OSError:
             pass
 
         dirname = os.path.dirname(linkname)
         if not os.path.exists(dirname):
             makedirs(dirname, self.createmode)
 
         if self._can_symlink:
             try:
                 os.symlink(src, linkname)
             except OSError, err:
                 raise OSError(err.errno, _('could not symlink to %r: %s') %
                               (src, err.strerror), linkname)
         else:
             f = self(dst, "w")
             f.write(src)
             f.close()
             self._fixfilemode(dst)
 
 class chunkbuffer(object):
     """Allow arbitrary sized chunks of data to be efficiently read from an
     iterator over chunks of arbitrary size."""
 
     def __init__(self, in_iter):
         """in_iter is the iterator that's iterating over the input chunks.
         targetsize is how big a buffer to try to maintain."""
         self.iter = iter(in_iter)
         self.buf = ''
         self.targetsize = 2**16
 
     def read(self, l):
         """Read L bytes of data from the iterator of chunks of data.
         Returns less than L bytes if the iterator runs dry."""
         if l > len(self.buf) and self.iter:
             # Clamp to a multiple of self.targetsize
             targetsize = max(l, self.targetsize)
             collector = cStringIO.StringIO()
             collector.write(self.buf)
             collected = len(self.buf)
             for chunk in self.iter:
                 collector.write(chunk)
                 collected += len(chunk)
                 if collected >= targetsize:
                     break
             if collected < targetsize:
                 self.iter = False
             self.buf = collector.getvalue()
         if len(self.buf) == l:
             s, self.buf = str(self.buf), ''
         else:
             s, self.buf = self.buf[:l], buffer(self.buf, l)
         return s
 
 def filechunkiter(f, size=65536, limit=None):
     """Create a generator that produces the data in the file size
     (default 65536) bytes at a time, up to optional limit (default is
     to read all data).  Chunks may be less than size bytes if the
     chunk is the last chunk in the file, or the file is a socket or
     some other type of file that sometimes reads less data than is
     requested."""
     assert size >= 0
     assert limit is None or limit >= 0
     while True:
         if limit is None: nbytes = size
         else: nbytes = min(limit, size)
         s = nbytes and f.read(nbytes)
         if not s: break
         if limit: limit -= len(s)
         yield s
 
 def makedate():
     lt = time.localtime()
     if lt[8] == 1 and time.daylight:
         tz = time.altzone
     else:
         tz = time.timezone
     return time.mktime(lt), tz
 
 def datestr(date=None, format='%a %b %d %H:%M:%S %Y %1%2'):
     """represent a (unixtime, offset) tuple as a localized time.
     unixtime is seconds since the epoch, and offset is the time zone's
     number of seconds away from UTC. if timezone is false, do not
     append time zone to string."""
     t, tz = date or makedate()
     if "%1" in format or "%2" in format:
         sign = (tz > 0) and "-" or "+"
         minutes = abs(tz) / 60
         format = format.replace("%1", "%c%02d" % (sign, minutes / 60))
         format = format.replace("%2", "%02d" % (minutes % 60))
     s = time.strftime(format, time.gmtime(float(t) - tz))
     return s
 
 def shortdate(date=None):
     """turn (timestamp, tzoff) tuple into iso 8631 date."""
     return datestr(date, format='%Y-%m-%d')
 
 def strdate(string, format, defaults=[]):
     """parse a localized time string and return a (unixtime, offset) tuple.
     if the string cannot be parsed, ValueError is raised."""
     def timezone(string):
         tz = string.split()[-1]
         if tz[0] in "+-" and len(tz) == 5 and tz[1:].isdigit():
             sign = (tz[0] == "+") and 1 or -1
             hours = int(tz[1:3])
             minutes = int(tz[3:5])
             return -sign * (hours * 60 + minutes) * 60
         if tz == "GMT" or tz == "UTC":
             return 0
         return None
 
     # NOTE: unixtime = localunixtime + offset
     offset, date = timezone(string), string
     if offset != None:
         date = " ".join(string.split()[:-1])
 
     # add missing elements from defaults
     for part in defaults:
         found = [True for p in part if ("%"+p) in format]
         if not found:
             date += "@" + defaults[part]
             format += "@%" + part[0]
 
     timetuple = time.strptime(date, format)
     localunixtime = int(calendar.timegm(timetuple))
     if offset is None:
         # local timezone
         unixtime = int(time.mktime(timetuple))
         offset = unixtime - localunixtime
     else:
         unixtime = localunixtime + offset
     return unixtime, offset
 
 def parsedate(date, formats=None, defaults=None):
     """parse a localized date/time string and return a (unixtime, offset) tuple.
 
     The date may be a "unixtime offset" string or in one of the specified
     formats. If the date already is a (unixtime, offset) tuple, it is returned.
     """
     if not date:
         return 0, 0
     if isinstance(date, tuple) and len(date) == 2:
         return date
     if not formats:
         formats = defaultdateformats
     date = date.strip()
     try:
         when, offset = map(int, date.split(' '))
     except ValueError:
         # fill out defaults
         if not defaults:
             defaults = {}
         now = makedate()
         for part in "d mb yY HI M S".split():
             if part not in defaults:
                 if part[0] in "HMS":
                     defaults[part] = "00"
                 else:
                     defaults[part] = datestr(now, "%" + part[0])
 
         for format in formats:
             try:
                 when, offset = strdate(date, format, defaults)
             except (ValueError, OverflowError):
                 pass
             else:
                 break
         else:
             raise Abort(_('invalid date: %r ') % date)
     # validate explicit (probably user-specified) date and
     # time zone offset. values must fit in signed 32 bits for
     # current 32-bit linux runtimes. timezones go from UTC-12
     # to UTC+14
     if abs(when) > 0x7fffffff:
         raise Abort(_('date exceeds 32 bits: %d') % when)
     if offset < -50400 or offset > 43200:
         raise Abort(_('impossible time zone offset: %d') % offset)
     return when, offset
 
 def matchdate(date):
     """Return a function that matches a given date match specifier
 
     Formats include:
 
     '{date}' match a given date to the accuracy provided
 
     '<{date}' on or before a given date
 
     '>{date}' on or after a given date
 
     """
 
     def lower(date):
         d = dict(mb="1", d="1")
         return parsedate(date, extendeddateformats, d)[0]
 
     def upper(date):
         d = dict(mb="12", HI="23", M="59", S="59")
         for days in "31 30 29".split():
             try:
                 d["d"] = days
                 return parsedate(date, extendeddateformats, d)[0]
             except:
                 pass
         d["d"] = "28"
         return parsedate(date, extendeddateformats, d)[0]
 
     date = date.strip()
     if date[0] == "<":
         when = upper(date[1:])
         return lambda x: x <= when
     elif date[0] == ">":
         when = lower(date[1:])
         return lambda x: x >= when
     elif date[0] == "-":
         try:
             days = int(date[1:])
         except ValueError:
             raise Abort(_("invalid day spec: %s") % date[1:])
         when = makedate()[0] - days * 3600 * 24
         return lambda x: x >= when
     elif " to " in date:
         a, b = date.split(" to ")
         start, stop = lower(a), upper(b)
         return lambda x: x >= start and x <= stop
     else:
         start, stop = lower(date), upper(date)
         return lambda x: x >= start and x <= stop
 
 def shortuser(user):
     """Return a short representation of a user name or email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f+1:]
     f = user.find(' ')
     if f >= 0:
         user = user[:f]
     f = user.find('.')
     if f >= 0:
         user = user[:f]
     return user
 
 def email(author):
     '''get email of author.'''
     r = author.find('>')
     if r == -1: r = None
     return author[author.find('<')+1:r]
 
 def ellipsis(text, maxlength=400):
     """Trim string to at most maxlength (default: 400) characters."""
     if len(text) <= maxlength:
         return text
     else:
         return "%s..." % (text[:maxlength-3])
 
 def walkrepos(path, followsym=False, seen_dirs=None, recurse=False):
     '''yield every hg repository under path, recursively.'''
     def errhandler(err):
         if err.filename == path:
             raise err
     if followsym and hasattr(os.path, 'samestat'):
         def _add_dir_if_not_there(dirlst, dirname):
             match = False
             samestat = os.path.samestat
             dirstat = os.stat(dirname)
             for lstdirstat in dirlst:
                 if samestat(dirstat, lstdirstat):
                     match = True
                     break
             if not match:
                 dirlst.append(dirstat)
             return not match
     else:
         followsym = False
 
     if (seen_dirs is None) and followsym:
         seen_dirs = []
         _add_dir_if_not_there(seen_dirs, path)
     for root, dirs, files in os.walk(path, topdown=True, onerror=errhandler):
         if '.hg' in dirs:
             yield root # found a repository
             qroot = os.path.join(root, '.hg', 'patches')
             if os.path.isdir(os.path.join(qroot, '.hg')):
                 yield qroot # we have a patch queue repo here
             if recurse:
                 # avoid recursing inside the .hg directory
                 dirs.remove('.hg')
             else:
                 dirs[:] = [] # don't descend further
         elif followsym:
             newdirs = []
             for d in dirs:
                 fname = os.path.join(root, d)
                 if _add_dir_if_not_there(seen_dirs, fname):
                     if os.path.islink(fname):
                         for hgname in walkrepos(fname, True, seen_dirs):
                             yield hgname
                     else:
                         newdirs.append(d)
             dirs[:] = newdirs
 
 _rcpath = None
 
 def os_rcpath():
     '''return default os-specific hgrc search path'''
     path = system_rcpath()
     path.extend(user_rcpath())
     path = [os.path.normpath(f) for f in path]
     return path
 
 def rcpath():
     '''return hgrc search path. if env var HGRCPATH is set, use it.
     for each item in path, if directory, use files ending in .rc,
     else use item.
     make HGRCPATH empty to only look in .hg/hgrc of current repo.
     if no HGRCPATH, use default os-specific path.'''
     global _rcpath
     if _rcpath is None:
         if 'HGRCPATH' in os.environ:
             _rcpath = []
             for p in os.environ['HGRCPATH'].split(os.pathsep):
                 if not p: continue
                 if os.path.isdir(p):
                     for f, kind in osutil.listdir(p):
                         if f.endswith('.rc'):
                             _rcpath.append(os.path.join(p, f))
                 else:
                     _rcpath.append(p)
         else:
             _rcpath = os_rcpath()
     return _rcpath
 
 def bytecount(nbytes):
     '''return byte count formatted as readable string, with units'''
 
     units = (
         (100, 1<<30, _('%.0f GB')),
         (10, 1<<30, _('%.1f GB')),
         (1, 1<<30, _('%.2f GB')),
         (100, 1<<20, _('%.0f MB')),
         (10, 1<<20, _('%.1f MB')),
         (1, 1<<20, _('%.2f MB')),
         (100, 1<<10, _('%.0f KB')),
         (10, 1<<10, _('%.1f KB')),
         (1, 1<<10, _('%.2f KB')),
         (1, 1, _('%.0f bytes')),
         )
 
     for multiplier, divisor, format in units:
         if nbytes >= divisor * multiplier:
             return format % (nbytes / float(divisor))
     return units[-1][2] % nbytes
 
 def drop_scheme(scheme, path):
     sc = scheme + ':'
     if path.startswith(sc):
         path = path[len(sc):]
         if path.startswith('//'):
             path = path[2:]
     return path
 
 def uirepr(s):
     # Avoid double backslash in Windows path repr()
     return repr(s).replace('\\\\', '\\')
 
 def termwidth():
     if 'COLUMNS' in os.environ:
         try:
             return int(os.environ['COLUMNS'])
         except ValueError:
             pass
     try:
         import termios, array, fcntl
         for dev in (sys.stdout, sys.stdin):
             try:
                 fd = dev.fileno()
                 if not os.isatty(fd):
                     continue
                 arri = fcntl.ioctl(fd, termios.TIOCGWINSZ, '\0' * 8)
                 return array.array('h', arri)[1]
             except ValueError:
                 pass
     except ImportError:
         pass
     return 80
 
 def iterlines(iterator):
     for chunk in iterator:
         for line in chunk.splitlines():
             yield line

tests/test-archive

@@ -1,118 +1,121 b''
 #!/bin/sh
 
 mkdir test
 cd test
 hg init
 echo foo>foo
 hg commit -Am 1 -d '1 0'
 echo bar>bar
 hg commit -Am 2 -d '2 0'
 mkdir baz
 echo bletch>baz/bletch
 hg commit -Am 3 -d '1000000000 0'
 echo "[web]" >> .hg/hgrc
 echo "name = test-archive" >> .hg/hgrc
 cp .hg/hgrc .hg/hgrc-base
 
 # check http return codes
 test_archtype() {
     echo "allow_archive = $1" >> .hg/hgrc
     hg serve -p $HGPORT -d --pid-file=hg.pid -E errors.log
     cat hg.pid >> $DAEMON_PIDS
     echo % $1 allowed should give 200
     "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.$2" | head -n 1
     echo % $3 and $4 disallowed should both give 403
     "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.$3" | head -n 1
     "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.$4" | head -n 1
     "$TESTDIR/killdaemons.py"
     cat errors.log
     cp .hg/hgrc-base .hg/hgrc
 }
 
 echo
 test_archtype gz tar.gz tar.bz2 zip
 test_archtype bz2 tar.bz2 zip tar.gz
 test_archtype zip zip tar.gz tar.bz2
 
 echo "allow_archive = gz bz2 zip" >> .hg/hgrc
 hg serve -p $HGPORT -d --pid-file=hg.pid -E errors.log
 cat hg.pid >> $DAEMON_PIDS
 
 echo % invalid arch type should give 404
 "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.invalid" | head -n 1
 echo
 
 TIP=`hg id -v | cut -f1 -d' '`
 QTIP=`hg id -q`
 cat > getarchive.py <<EOF
 import os, sys, urllib2
 try:
     # Set stdout to binary mode for win32 platforms
     import msvcrt
     msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
 except ImportError:
     pass
     
 node, archive = sys.argv[1:]
 f = urllib2.urlopen('http://127.0.0.1:%s/?cmd=archive;node=%s;type=%s'
                     % (os.environ['HGPORT'], node, archive))
 sys.stdout.write(f.read())
 EOF
 http_proxy= python getarchive.py "$TIP" gz | gunzip | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/"
 http_proxy= python getarchive.py "$TIP" bz2 | bunzip2 | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/"
 http_proxy= python getarchive.py "$TIP" zip > archive.zip
 unzip -t archive.zip | sed "s/$QTIP/TIP/"
 
 "$TESTDIR/killdaemons.py"
 
 hg archive -t tar test.tar
 tar tf test.tar
 
 hg archive -t tbz2 -X baz test.tar.bz2
 bunzip2 -dc test.tar.bz2 | tar tf - 2>/dev/null
 
 hg archive -t tgz -p %b-%h test-%h.tar.gz
 gzip -dc test-$QTIP.tar.gz | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/"
 
 cat > md5comp.py <<EOF
-from mercurial.util import md5
+try:
+    from hashlib import md5
+except ImportError:
+    from md5 import md5
 import sys
 f1, f2 = sys.argv[1:3]
 h1 = md5(file(f1, 'rb').read()).hexdigest()
 h2 = md5(file(f2, 'rb').read()).hexdigest()
 print h1 == h2 or "md5 differ: " + repr((h1, h2))
 EOF
 
 # archive name is stored in the archive, so create similar
 # archives and rename them afterwards.
 hg archive -t tgz tip.tar.gz
 mv tip.tar.gz tip1.tar.gz
 sleep 1
 hg archive -t tgz tip.tar.gz
 mv tip.tar.gz tip2.tar.gz
 python md5comp.py tip1.tar.gz tip2.tar.gz
 
 hg archive -t zip -p /illegal test.zip
 hg archive -t zip -p very/../bad test.zip
 
 hg archive --config ui.archivemeta=false -t zip -r 2 test.zip
 unzip -t test.zip
 
 hg archive -t tar - | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/"
 
 hg archive -r 0 -t tar rev-%r.tar
 if [ -f rev-0.tar ]; then
     echo 'rev-0.tar created'
 fi
 
 hg archive -t bogus test.bogus
 
 echo % server errors
 cat errors.log
 
 echo '% empty repo'
 hg init ../empty
 cd ../empty
 hg archive ../test-empty
 
 exit 0