# url.py - HTTP handling for mercurial
#
# Copyright 2005, 2006, 2007, 2008 Matt Mackall <[email protected]>
# Copyright 2006, 2007 Alexis S. L. Carvalho <[email protected]>
# Copyright 2006 Vadim Gelfer <[email protected]>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2, incorporated herein by reference.

import urllib, urllib2, urlparse, httplib, os, re, socket, cStringIO
from i18n import _
import keepalive, util

def hidepassword(url):
   '''hide user credential in a url string'''
   scheme, netloc, path, params, query, fragment = urlparse.urlparse(url)
   netloc = re.sub('([^:]*):([^@]*)@(.*)', r'\1:***@\3', netloc)
   return urlparse.urlunparse((scheme, netloc, path, params, query, fragment))

def removeauth(url):
   '''remove all authentication information from a url string'''
   scheme, netloc, path, params, query, fragment = urlparse.urlparse(url)
   netloc = netloc[netloc.find('@')+1:]
   return urlparse.urlunparse((scheme, netloc, path, params, query, fragment))

def netlocsplit(netloc):
   '''split [user[:passwd]@]host[:port] into 4-tuple.'''

   a = netloc.find('@')
   if a == -1:
       user, passwd = None, None
   else:
       userpass, netloc = netloc[:a], netloc[a+1:]
       c = userpass.find(':')
       if c == -1:
           user, passwd = urllib.unquote(userpass), None
       else:
           user = urllib.unquote(userpass[:c])
           passwd = urllib.unquote(userpass[c+1:])
   c = netloc.find(':')
   if c == -1:
       host, port = netloc, None
   else:
       host, port = netloc[:c], netloc[c+1:]
   return host, port, user, passwd

def netlocunsplit(host, port, user=None, passwd=None):
   '''turn host, port, user, passwd into [user[:passwd]@]host[:port].'''
   if port:
       hostport = host + ':' + port
   else:
       hostport = host
   if user:
       if passwd:
           userpass = urllib.quote(user) + ':' + urllib.quote(passwd)
       else:
           userpass = urllib.quote(user)
       return userpass + '@' + hostport
   return hostport

_safe = ('abcdefghijklmnopqrstuvwxyz'
        'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
        '0123456789' '_.-/')
_safeset = None
_hex = None
def quotepath(path):
   '''quote the path part of a URL

   This is similar to urllib.quote, but it also tries to avoid
   quoting things twice (inspired by wget):

   >>> quotepath('abc def')
   'abc%20def'
   >>> quotepath('abc%20def')
   'abc%20def'
   >>> quotepath('abc%20 def')
   'abc%20%20def'
   >>> quotepath('abc def%20')
   'abc%20def%20'
   >>> quotepath('abc def%2')
   'abc%20def%252'
   >>> quotepath('abc def%')
   'abc%20def%25'
   '''
   global _safeset, _hex
   if _safeset is None:
       _safeset = set(_safe)
       _hex = set('abcdefABCDEF0123456789')
   l = list(path)
   for i in xrange(len(l)):
       c = l[i]
       if c == '%' and i + 2 < len(l) and (l[i+1] in _hex and l[i+2] in _hex):
           pass
       elif c not in _safeset:
           l[i] = '%%%02X' % ord(c)
   return ''.join(l)

class passwordmgr(urllib2.HTTPPasswordMgrWithDefaultRealm):
   def __init__(self, ui):
       urllib2.HTTPPasswordMgrWithDefaultRealm.__init__(self)
       self.ui = ui

   def find_user_password(self, realm, authuri):
       authinfo = urllib2.HTTPPasswordMgrWithDefaultRealm.find_user_password(
           self, realm, authuri)
       user, passwd = authinfo
       if user and passwd:
           self._writedebug(user, passwd)
           return (user, passwd)

       if not user:
           auth = self.readauthtoken(authuri)
           if auth:
               user, passwd = auth.get('username'), auth.get('password')
       if not user or not passwd:
           if not self.ui.interactive():
               raise util.Abort(_('http authorization required'))

           self.ui.write(_("http authorization required\n"))
           self.ui.status(_("realm: %s\n") % realm)
           if user:
               self.ui.status(_("user: %s\n") % user)
           else:
               user = self.ui.prompt(_("user:"), default=None)

           if not passwd:
               passwd = self.ui.getpass()

       self.add_password(realm, authuri, user, passwd)
       self._writedebug(user, passwd)
       return (user, passwd)

   def _writedebug(self, user, passwd):
       msg = _('http auth: user %s, password %s\n')
       self.ui.debug(msg % (user, passwd and '*' * len(passwd) or 'not set'))

   def readauthtoken(self, uri):
       # Read configuration
       config = dict()
       for key, val in self.ui.configitems('auth'):
           group, setting = key.split('.', 1)
           gdict = config.setdefault(group, dict())
           gdict[setting] = val

       # Find the best match
       scheme, hostpath = uri.split('://', 1)
       bestlen = 0
       bestauth = None
       for auth in config.itervalues():
           prefix = auth.get('prefix')
           if not prefix: continue
           p = prefix.split('://', 1)
           if len(p) > 1:
               schemes, prefix = [p[0]], p[1]
           else:
               schemes = (auth.get('schemes') or 'https').split()
           if (prefix == '*' or hostpath.startswith(prefix)) and \
               len(prefix) > bestlen and scheme in schemes:
               bestlen = len(prefix)
               bestauth = auth
       return bestauth

class proxyhandler(urllib2.ProxyHandler):
   def __init__(self, ui):
       proxyurl = ui.config("http_proxy", "host") or os.getenv('http_proxy')
       # XXX proxyauthinfo = None

       if proxyurl:
           # proxy can be proper url or host[:port]
           if not (proxyurl.startswith('http:') or
                   proxyurl.startswith('https:')):
               proxyurl = 'http://' + proxyurl + '/'
           snpqf = urlparse.urlsplit(proxyurl)
           proxyscheme, proxynetloc, proxypath, proxyquery, proxyfrag = snpqf
           hpup = netlocsplit(proxynetloc)

           proxyhost, proxyport, proxyuser, proxypasswd = hpup
           if not proxyuser:
               proxyuser = ui.config("http_proxy", "user")
               proxypasswd = ui.config("http_proxy", "passwd")

           # see if we should use a proxy for this url
           no_list = [ "localhost", "127.0.0.1" ]
           no_list.extend([p.lower() for
                           p in ui.configlist("http_proxy", "no")])
           no_list.extend([p.strip().lower() for
                           p in os.getenv("no_proxy", '').split(',')
                           if p.strip()])
           # "http_proxy.always" config is for running tests on localhost
           if ui.configbool("http_proxy", "always"):
               self.no_list = []
           else:
               self.no_list = no_list

           proxyurl = urlparse.urlunsplit((
               proxyscheme, netlocunsplit(proxyhost, proxyport,
                                               proxyuser, proxypasswd or ''),
               proxypath, proxyquery, proxyfrag))
           proxies = {'http': proxyurl, 'https': proxyurl}
           ui.debug(_('proxying through http://%s:%s\n') %
                     (proxyhost, proxyport))
       else:
           proxies = {}

       # urllib2 takes proxy values from the environment and those
       # will take precedence if found, so drop them
       for env in ["HTTP_PROXY", "http_proxy", "no_proxy"]:
           try:
               if env in os.environ:
                   del os.environ[env]
           except OSError:
               pass

       urllib2.ProxyHandler.__init__(self, proxies)
       self.ui = ui

   def proxy_open(self, req, proxy, type_):
       host = req.get_host().split(':')[0]
       if host in self.no_list:
           return None

       # work around a bug in Python < 2.4.2
       # (it leaves a "\n" at the end of Proxy-authorization headers)
       baseclass = req.__class__
       class _request(baseclass):
           def add_header(self, key, val):
               if key.lower() == 'proxy-authorization':
                   val = val.strip()
               return baseclass.add_header(self, key, val)
       req.__class__ = _request

       return urllib2.ProxyHandler.proxy_open(self, req, proxy, type_)

class httpsendfile(file):
   def __len__(self):
       return os.fstat(self.fileno()).st_size

def _gen_sendfile(connection):
   def _sendfile(self, data):
       # send a file
       if isinstance(data, httpsendfile):
           # if auth required, some data sent twice, so rewind here
           data.seek(0)
           for chunk in util.filechunkiter(data):
               connection.send(self, chunk)
       else:
           connection.send(self, data)
   return _sendfile

has_https = hasattr(urllib2, 'HTTPSHandler')
if has_https:
   try:
       # avoid using deprecated/broken FakeSocket in python 2.6
       import ssl
       _ssl_wrap_socket = ssl.wrap_socket
   except ImportError:
       def _ssl_wrap_socket(sock, key_file, cert_file):
           ssl = socket.ssl(sock, key_file, cert_file)
           return httplib.FakeSocket(sock, ssl)

class httpconnection(keepalive.HTTPConnection):
   # must be able to send big bundle as stream.
   send = _gen_sendfile(keepalive.HTTPConnection)

   def _proxytunnel(self):
       proxyheaders = dict(
               [(x, self.headers[x]) for x in self.headers
                if x.lower().startswith('proxy-')])
       self._set_hostport(self.host, self.port)
       self.send('CONNECT %s:%d HTTP/1.0\r\n' % (self.realhost, self.realport))
       for header in proxyheaders.iteritems():
           self.send('%s: %s\r\n' % header)
       self.send('\r\n')

       # majority of the following code is duplicated from
       # httplib.HTTPConnection as there are no adequate places to
       # override functions to provide the needed functionality
       res = self.response_class(self.sock,
                                 strict=self.strict,
                                 method=self._method)

       while True:
           version, status, reason = res._read_status()
           if status != httplib.CONTINUE:
               break
           while True:
               skip = res.fp.readline().strip()
               if not skip:
                   break
       res.status = status
       res.reason = reason.strip()

       if res.status == 200:
           while True:
               line = res.fp.readline()
               if line == '\r\n':
                   break
           return True

       if version == 'HTTP/1.0':
           res.version = 10
       elif version.startswith('HTTP/1.'):
           res.version = 11
       elif version == 'HTTP/0.9':
           res.version = 9
       else:
           raise httplib.UnknownProtocol(version)

       if res.version == 9:
           res.length = None
           res.chunked = 0
           res.will_close = 1
           res.msg = httplib.HTTPMessage(cStringIO.StringIO())
           return False

       res.msg = httplib.HTTPMessage(res.fp)
       res.msg.fp = None

       # are we using the chunked-style of transfer encoding?
       trenc = res.msg.getheader('transfer-encoding')
       if trenc and trenc.lower() == "chunked":
           res.chunked = 1
           res.chunk_left = None
       else:
           res.chunked = 0

       # will the connection close at the end of the response?
       res.will_close = res._check_close()

       # do we have a Content-Length?
       # NOTE: RFC 2616, S4.4, #3 says we ignore this if tr_enc is "chunked"
       length = res.msg.getheader('content-length')
       if length and not res.chunked:
           try:
               res.length = int(length)
           except ValueError:
               res.length = None
           else:
               if res.length < 0:  # ignore nonsensical negative lengths
                   res.length = None
       else:
           res.length = None

       # does the body have a fixed length? (of zero)
       if (status == httplib.NO_CONTENT or status == httplib.NOT_MODIFIED or
           100 <= status < 200 or # 1xx codes
           res._method == 'HEAD'):
           res.length = 0

       # if the connection remains open, and we aren't using chunked, and
       # a content-length was not provided, then assume that the connection
       # WILL close.
       if (not res.will_close and
          not res.chunked and
          res.length is None):
           res.will_close = 1

       self.proxyres = res

       return False

   def connect(self):
       if has_https and self.realhost: # use CONNECT proxy
           self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
           self.sock.connect((self.host, self.port))
           if self._proxytunnel():
               # we do not support client x509 certificates
               self.sock = _ssl_wrap_socket(self.sock, None, None)
       else:
           keepalive.HTTPConnection.connect(self)

   def getresponse(self):
       proxyres = getattr(self, 'proxyres', None)
       if proxyres:
           if proxyres.will_close:
               self.close()
           self.proxyres = None
           return proxyres
       return keepalive.HTTPConnection.getresponse(self)

class httphandler(keepalive.HTTPHandler):
   def http_open(self, req):
       return self.do_open(httpconnection, req)

   def _start_transaction(self, h, req):
       if req.get_selector() == req.get_full_url(): # has proxy
           urlparts = urlparse.urlparse(req.get_selector())
           if urlparts[0] == 'https': # only use CONNECT for HTTPS
               if ':' in urlparts[1]:
                   realhost, realport = urlparts[1].split(':')
                   realport = int(realport)
               else:
                   realhost = urlparts[1]
                   realport = 443

               h.realhost = realhost
               h.realport = realport
               h.headers = req.headers.copy()
               h.headers.update(self.parent.addheaders)
               return keepalive.HTTPHandler._start_transaction(self, h, req)

       h.realhost = None
       h.realport = None
       h.headers = None
       return keepalive.HTTPHandler._start_transaction(self, h, req)

   def __del__(self):
       self.close_all()

if has_https:
   class httpsconnection(httplib.HTTPSConnection):
       response_class = keepalive.HTTPResponse
       # must be able to send big bundle as stream.
       send = _gen_sendfile(httplib.HTTPSConnection)

   class httpshandler(keepalive.KeepAliveHandler, urllib2.HTTPSHandler):
       def __init__(self, ui):
           keepalive.KeepAliveHandler.__init__(self)
           urllib2.HTTPSHandler.__init__(self)
           self.ui = ui
           self.pwmgr = passwordmgr(self.ui)

       def https_open(self, req):
           self.auth = self.pwmgr.readauthtoken(req.get_full_url())
           return self.do_open(self._makeconnection, req)

       def _makeconnection(self, host, port=443, *args, **kwargs):
           keyfile = None
           certfile = None

           if args: # key_file
               keyfile = args.pop(0)
           if args: # cert_file
               certfile = args.pop(0)

           # if the user has specified different key/cert files in
           # hgrc, we prefer these
           if self.auth and 'key' in self.auth and 'cert' in self.auth:
               keyfile = self.auth['key']
               certfile = self.auth['cert']

           # let host port take precedence
           if ':' in host and '[' not in host or ']:' in host:
               host, port = host.rsplit(':', 1)
               port = int(port)
               if '[' in host:
                   host = host[1:-1]

           return httpsconnection(host, port, keyfile, certfile, *args, **kwargs)

# In python < 2.5 AbstractDigestAuthHandler raises a ValueError if
# it doesn't know about the auth type requested.  This can happen if
# somebody is using BasicAuth and types a bad password.
class httpdigestauthhandler(urllib2.HTTPDigestAuthHandler):
   def http_error_auth_reqed(self, auth_header, host, req, headers):
       try:
           return urllib2.HTTPDigestAuthHandler.http_error_auth_reqed(
                       self, auth_header, host, req, headers)
       except ValueError, inst:
           arg = inst.args[0]
           if arg.startswith("AbstractDigestAuthHandler doesn't know "):
               return
           raise

def getauthinfo(path):
   scheme, netloc, urlpath, query, frag = urlparse.urlsplit(path)
   if not urlpath:
       urlpath = '/'
   if scheme != 'file':
       # XXX: why are we quoting the path again with some smart
       # heuristic here? Anyway, it cannot be done with file://
       # urls since path encoding is os/fs dependent (see
       # urllib.pathname2url() for details).
       urlpath = quotepath(urlpath)
   host, port, user, passwd = netlocsplit(netloc)

   # urllib cannot handle URLs with embedded user or passwd
   url = urlparse.urlunsplit((scheme, netlocunsplit(host, port),
                             urlpath, query, frag))
   if user:
       netloc = host
       if port:
           netloc += ':' + port
       # Python < 2.4.3 uses only the netloc to search for a password
       authinfo = (None, (url, netloc), user, passwd or '')
   else:
       authinfo = None
   return url, authinfo

handlerfuncs = []

def opener(ui, authinfo=None):
   '''
   construct an opener suitable for urllib2
   authinfo will be added to the password manager
   '''
   handlers = [httphandler()]
   if has_https:
       handlers.append(httpshandler(ui))

   handlers.append(proxyhandler(ui))

   passmgr = passwordmgr(ui)
   if authinfo is not None:
       passmgr.add_password(*authinfo)
       user, passwd = authinfo[2:4]
       ui.debug(_('http auth: user %s, password %s\n') %
                (user, passwd and '*' * len(passwd) or 'not set'))

   handlers.extend((urllib2.HTTPBasicAuthHandler(passmgr),
                    httpdigestauthhandler(passmgr)))
   handlers.extend([h(ui, passmgr) for h in handlerfuncs])
   opener = urllib2.build_opener(*handlers)

   # 1.0 here is the _protocol_ version
   opener.addheaders = [('User-agent', 'mercurial/proto-1.0')]
   opener.addheaders.append(('Accept', 'application/mercurial-0.1'))
   return opener

scheme_re = re.compile(r'^([a-zA-Z0-9+-.]+)://')

def open(ui, url, data=None):
   scheme = None
   m = scheme_re.search(url)
   if m:
       scheme = m.group(1).lower()
   if not scheme:
       path = util.normpath(os.path.abspath(url))
       url = 'file://' + urllib.pathname2url(path)
       authinfo = None
   else:
       url, authinfo = getauthinfo(url)
   return opener(ui, authinfo).open(url, data)

You are viewing proxied material from sdf.org. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.