https_forwarder.py revision 5f1c94371a64b3196d4be9466099bb892df9b88e
15821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)# Copyright 2014 The Chromium Authors. All rights reserved.
25821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)# Use of this source code is governed by a BSD-style license that can be
35821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)# found in the LICENSE file.
45821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
55821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)"""An https server that forwards requests to another server. This allows a
65821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)server that supports http only to be accessed over https.
75821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)"""
85821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
95821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import BaseHTTPServer
105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import os
115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import SocketServer
125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import sys
135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import urllib2
145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import urlparse
155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import testserver_base
165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)import tlslite.api
175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)class RedirectSuppressor(urllib2.HTTPErrorProcessor):
205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  """Prevents urllib2 from following http redirects.
215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  If this class is placed in an urllib2.OpenerDirector's handler chain before
235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  the default urllib2.HTTPRedirectHandler, it will terminate the processing of
245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  responses containing redirect codes (301, 302, 303, 307) before they reach the
255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  default redirect handler.
265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  """
275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  def http_response(self, req, response):
295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    return response
305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  def https_response(self, req, response):
325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    return response
335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)class RequestForwarder(BaseHTTPServer.BaseHTTPRequestHandler):
365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  """Handles requests received by forwarding them to the another server."""
375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  def do_GET(self):
395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    """Forwards GET requests."""
405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self._forward(None)
415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  def do_POST(self):
435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    """Forwards POST requests."""
445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self._forward(self.rfile.read(int(self.headers['Content-Length'])))
455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  def _forward(self, body):
475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    """Forwards a GET or POST request to another server.
485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    Args:
505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      body: The request body. This should be |None| for GET requests.
515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    """
525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    request_url = urlparse.urlparse(self.path)
535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    url = urlparse.urlunparse((self.server.forward_scheme,
545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               self.server.forward_netloc,
555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               self.server.forward_path + request_url[2],
565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               request_url[3],
575821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               request_url[4],
585821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               request_url[5]))
595821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
605821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    headers = dict((key, value) for key, value in dict(self.headers).iteritems()
615821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                   if key.lower() != 'host')
625821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    opener = urllib2.build_opener(RedirectSuppressor)
635821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    forward = opener.open(urllib2.Request(url, body, headers))
645821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
655821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self.send_response(forward.getcode())
665821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    for key, value in dict(forward.info()).iteritems():
675821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # RFC 6265 states in section 3:
685821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      #
695821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # Origin servers SHOULD NOT fold multiple Set-Cookie header fields into
705821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # a single header field.
715821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      #
725821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # Python 2 does not obey this requirement and folds multiple Set-Cookie
735821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # header fields into one. The following code undoes this folding by
745821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # splitting the Set-Cookie header field at each comma. Note that this is a
755821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # hack because the code does not (and cannot reliably) distinguish between
765821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # commas inserted by Python while folding multiple headers and commas that
775821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      # were part of the original Set-Cookie headers.
785821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      if key == 'set-cookie':
795821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)        for cookie in value.split(','):
805821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)          self.send_header(key, cookie)
815821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      else:
825821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)        self.send_header(key, value)
835821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self.end_headers()
845821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self.wfile.write(forward.read())
855821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
865821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
875821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)class MultiThreadedHTTPSServer(SocketServer.ThreadingMixIn,
885821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               tlslite.api.TLSSocketServerMixIn,
895821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               testserver_base.ClientRestrictingServerMixIn,
905821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               testserver_base.BrokenPipeHandlerMixIn,
915821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                               testserver_base.StoppableHTTPServer):
925821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  """A multi-threaded version of testserver.HTTPSServer."""
935821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
945821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  def __init__(self, server_address, request_hander_class, pem_cert_and_key):
955821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    """Initializes the server.
96a3f6a49ab37290eeeb8db0f41ec0f1cb74a68be7Torne (Richard Coles)
97a3f6a49ab37290eeeb8db0f41ec0f1cb74a68be7Torne (Richard Coles)    Args:
98a3f6a49ab37290eeeb8db0f41ec0f1cb74a68be7Torne (Richard Coles)      server_address: Server host and port.
995821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)      request_hander_class: The class that will handle requests to the server.
100a3f6a49ab37290eeeb8db0f41ec0f1cb74a68be7Torne (Richard Coles)      pem_cert_and_key: Path to file containing the https cert and private key.
101a3f6a49ab37290eeeb8db0f41ec0f1cb74a68be7Torne (Richard Coles)    """
1025821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self.cert_chain = tlslite.api.X509CertChain()
1035821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)    self.cert_chain.parsePemList(pem_cert_and_key)
104    # Force using only python implementation - otherwise behavior is different
105    # depending on whether m2crypto Python module is present (error is thrown
106    # when it is). m2crypto uses a C (based on OpenSSL) implementation under
107    # the hood.
108    self.private_key = tlslite.api.parsePEMKey(pem_cert_and_key,
109                                               private=True,
110                                               implementations=['python'])
111
112    testserver_base.StoppableHTTPServer.__init__(self,
113                                                 server_address,
114                                                 request_hander_class)
115
116  def handshake(self, tlsConnection):
117    """Performs the SSL handshake for an https connection.
118
119    Args:
120      tlsConnection: The https connection.
121    Returns:
122      Whether the SSL handshake succeeded.
123    """
124    try:
125      self.tlsConnection = tlsConnection
126      tlsConnection.handshakeServer(certChain=self.cert_chain,
127                                    privateKey=self.private_key)
128      tlsConnection.ignoreAbruptClose = True
129      return True
130    except:
131      return False
132
133
134class ServerRunner(testserver_base.TestServerRunner):
135  """Runner that starts an https server which forwards requests to another
136  server.
137  """
138
139  def create_server(self, server_data):
140    """Performs the SSL handshake for an https connection.
141
142    Args:
143      server_data: Dictionary that holds information about the server.
144    Returns:
145      The started server.
146    """
147    port = self.options.port
148    host = self.options.host
149
150    if not os.path.isfile(self.options.cert_and_key_file):
151      raise testserver_base.OptionError(
152          'Specified server cert file not found: ' +
153          self.options.cert_and_key_file)
154    pem_cert_and_key = open(self.options.cert_and_key_file).read()
155
156    server = MultiThreadedHTTPSServer((host, port),
157                                      RequestForwarder,
158                                      pem_cert_and_key)
159    print 'HTTPS server started on %s:%d...' % (host, server.server_port)
160
161    forward_target = urlparse.urlparse(self.options.forward_target)
162    server.forward_scheme = forward_target[0]
163    server.forward_netloc = forward_target[1]
164    server.forward_path = forward_target[2].rstrip('/')
165    server.forward_host = forward_target.hostname
166    if forward_target.port:
167      server.forward_host += ':' + str(forward_target.port)
168    server_data['port'] = server.server_port
169    return server
170
171  def add_options(self):
172    """Specifies the command-line options understood by the server."""
173    testserver_base.TestServerRunner.add_options(self)
174    self.option_parser.add_option('--https', action='store_true',
175                                  help='Ignored (provided for compatibility '
176                                  'only).')
177    self.option_parser.add_option('--cert-and-key-file', help='The path to the '
178                                  'file containing the certificate and private '
179                                  'key for the server in PEM format.')
180    self.option_parser.add_option('--forward-target', help='The URL prefix to '
181                                  'which requests will be forwarded.')
182
183
184if __name__ == '__main__':
185  sys.exit(ServerRunner().main())
186