Statistics
| Branch: | Tag: | Revision:

root / lib / http / __init__.py @ 77e1d753

History | View | Annotate | Download (27.5 kB)

1 a43f68dc Michael Hanselmann
#
2 a43f68dc Michael Hanselmann
#
3 4ce6007f Michael Hanselmann
4 4ce6007f Michael Hanselmann
# Copyright (C) 2007, 2008 Google Inc.
5 4ce6007f Michael Hanselmann
#
6 a43f68dc Michael Hanselmann
# This program is free software; you can redistribute it and/or modify
7 a43f68dc Michael Hanselmann
# it under the terms of the GNU General Public License as published by
8 a43f68dc Michael Hanselmann
# the Free Software Foundation; either version 2 of the License, or
9 a43f68dc Michael Hanselmann
# (at your option) any later version.
10 a43f68dc Michael Hanselmann
#
11 a43f68dc Michael Hanselmann
# This program is distributed in the hope that it will be useful, but
12 a43f68dc Michael Hanselmann
# WITHOUT ANY WARRANTY; without even the implied warranty of
13 a43f68dc Michael Hanselmann
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14 a43f68dc Michael Hanselmann
# General Public License for more details.
15 a43f68dc Michael Hanselmann
#
16 a43f68dc Michael Hanselmann
# You should have received a copy of the GNU General Public License
17 a43f68dc Michael Hanselmann
# along with this program; if not, write to the Free Software
18 a43f68dc Michael Hanselmann
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
19 a43f68dc Michael Hanselmann
# 02110-1301, USA.
20 a43f68dc Michael Hanselmann
21 02cab3e7 Michael Hanselmann
"""HTTP module.
22 a43f68dc Michael Hanselmann

23 a43f68dc Michael Hanselmann
"""
24 a43f68dc Michael Hanselmann
25 42242313 Michael Hanselmann
import logging
26 42242313 Michael Hanselmann
import mimetools
27 a43f68dc Michael Hanselmann
import OpenSSL
28 42242313 Michael Hanselmann
import select
29 42242313 Michael Hanselmann
import socket
30 8a0b06d2 Michael Hanselmann
import errno
31 8a0b06d2 Michael Hanselmann
32 8a0b06d2 Michael Hanselmann
from cStringIO import StringIO
33 a43f68dc Michael Hanselmann
34 42242313 Michael Hanselmann
from ganeti import constants
35 a43f68dc Michael Hanselmann
from ganeti import serializer
36 f2a6fc9e Michael Hanselmann
from ganeti import utils
37 a43f68dc Michael Hanselmann
38 a43f68dc Michael Hanselmann
39 8a9f9060 Michael Hanselmann
HTTP_GANETI_VERSION = "Ganeti %s" % constants.RELEASE_VERSION
40 8a9f9060 Michael Hanselmann
41 42242313 Michael Hanselmann
HTTP_OK = 200
42 42242313 Michael Hanselmann
HTTP_NO_CONTENT = 204
43 42242313 Michael Hanselmann
HTTP_NOT_MODIFIED = 304
44 42242313 Michael Hanselmann
45 42242313 Michael Hanselmann
HTTP_0_9 = "HTTP/0.9"
46 42242313 Michael Hanselmann
HTTP_1_0 = "HTTP/1.0"
47 42242313 Michael Hanselmann
HTTP_1_1 = "HTTP/1.1"
48 42242313 Michael Hanselmann
49 42242313 Michael Hanselmann
HTTP_GET = "GET"
50 42242313 Michael Hanselmann
HTTP_HEAD = "HEAD"
51 8a9f9060 Michael Hanselmann
HTTP_POST = "POST"
52 8a9f9060 Michael Hanselmann
HTTP_PUT = "PUT"
53 99b5ef90 Michael Hanselmann
HTTP_DELETE = "DELETE"
54 8a9f9060 Michael Hanselmann
55 713faea6 Oleksiy Mishchenko
HTTP_ETAG = "ETag"
56 8a9f9060 Michael Hanselmann
HTTP_HOST = "Host"
57 8a9f9060 Michael Hanselmann
HTTP_SERVER = "Server"
58 8a9f9060 Michael Hanselmann
HTTP_DATE = "Date"
59 8a9f9060 Michael Hanselmann
HTTP_USER_AGENT = "User-Agent"
60 8a9f9060 Michael Hanselmann
HTTP_CONTENT_TYPE = "Content-Type"
61 8a9f9060 Michael Hanselmann
HTTP_CONTENT_LENGTH = "Content-Length"
62 8a9f9060 Michael Hanselmann
HTTP_CONNECTION = "Connection"
63 8a9f9060 Michael Hanselmann
HTTP_KEEP_ALIVE = "Keep-Alive"
64 b3660886 Michael Hanselmann
HTTP_WWW_AUTHENTICATE = "WWW-Authenticate"
65 b3660886 Michael Hanselmann
HTTP_AUTHORIZATION = "Authorization"
66 b3660886 Michael Hanselmann
HTTP_AUTHENTICATION_INFO = "Authentication-Info"
67 b3660886 Michael Hanselmann
HTTP_ALLOW = "Allow"
68 42242313 Michael Hanselmann
69 d7bace1b Michael Hanselmann
_SSL_UNEXPECTED_EOF = "Unexpected EOF"
70 d7bace1b Michael Hanselmann
71 73a59d9e Michael Hanselmann
# Socket operations
72 73a59d9e Michael Hanselmann
(SOCKOP_SEND,
73 73a59d9e Michael Hanselmann
 SOCKOP_RECV,
74 d7fa9982 Michael Hanselmann
 SOCKOP_SHUTDOWN,
75 d7fa9982 Michael Hanselmann
 SOCKOP_HANDSHAKE) = range(4)
76 73a59d9e Michael Hanselmann
77 b18dd019 Iustin Pop
# send/receive quantum
78 b18dd019 Iustin Pop
SOCK_BUF_SIZE = 32768
79 42242313 Michael Hanselmann
80 13b63666 Michael Hanselmann
81 02cab3e7 Michael Hanselmann
class HttpError(Exception):
82 02cab3e7 Michael Hanselmann
  """Internal exception for HTTP errors.
83 02cab3e7 Michael Hanselmann

84 02cab3e7 Michael Hanselmann
  This should only be used for internal error reporting.
85 02cab3e7 Michael Hanselmann

86 02cab3e7 Michael Hanselmann
  """
87 02cab3e7 Michael Hanselmann
88 02cab3e7 Michael Hanselmann
89 39cfc25b Michael Hanselmann
class HttpConnectionClosed(Exception):
90 39cfc25b Michael Hanselmann
  """Internal exception for a closed connection.
91 39cfc25b Michael Hanselmann

92 39cfc25b Michael Hanselmann
  This should only be used for internal error reporting. Only use
93 39cfc25b Michael Hanselmann
  it if there's no other way to report this condition.
94 39cfc25b Michael Hanselmann

95 39cfc25b Michael Hanselmann
  """
96 39cfc25b Michael Hanselmann
97 39cfc25b Michael Hanselmann
98 d7fa9982 Michael Hanselmann
class HttpSessionHandshakeUnexpectedEOF(HttpError):
99 d7fa9982 Michael Hanselmann
  """Internal exception for errors during SSL handshake.
100 d7fa9982 Michael Hanselmann

101 d7fa9982 Michael Hanselmann
  This should only be used for internal error reporting.
102 d7fa9982 Michael Hanselmann

103 d7fa9982 Michael Hanselmann
  """
104 d7fa9982 Michael Hanselmann
105 d7fa9982 Michael Hanselmann
106 c9d0fa8a Michael Hanselmann
class HttpSocketTimeout(Exception):
107 73a59d9e Michael Hanselmann
  """Internal exception for socket timeouts.
108 73a59d9e Michael Hanselmann

109 73a59d9e Michael Hanselmann
  This should only be used for internal error reporting.
110 73a59d9e Michael Hanselmann

111 73a59d9e Michael Hanselmann
  """
112 8a0b06d2 Michael Hanselmann
113 8a0b06d2 Michael Hanselmann
114 84f2756e Michael Hanselmann
class HttpException(Exception):
115 a43f68dc Michael Hanselmann
  code = None
116 a43f68dc Michael Hanselmann
  message = None
117 a43f68dc Michael Hanselmann
118 a8e01e9f Michael Hanselmann
  def __init__(self, message=None, headers=None):
119 42242313 Michael Hanselmann
    Exception.__init__(self)
120 a8e01e9f Michael Hanselmann
    self.message = message
121 a8e01e9f Michael Hanselmann
    self.headers = headers
122 a43f68dc Michael Hanselmann
123 a43f68dc Michael Hanselmann
124 84f2756e Michael Hanselmann
class HttpBadRequest(HttpException):
125 b3660886 Michael Hanselmann
  """400 Bad Request
126 b3660886 Michael Hanselmann

127 25e7b43f Iustin Pop
  RFC2616, 10.4.1: The request could not be understood by the server
128 25e7b43f Iustin Pop
  due to malformed syntax. The client SHOULD NOT repeat the request
129 25e7b43f Iustin Pop
  without modifications.
130 b3660886 Michael Hanselmann

131 b3660886 Michael Hanselmann
  """
132 a43f68dc Michael Hanselmann
  code = 400
133 a43f68dc Michael Hanselmann
134 a43f68dc Michael Hanselmann
135 b3660886 Michael Hanselmann
class HttpUnauthorized(HttpException):
136 b3660886 Michael Hanselmann
  """401 Unauthorized
137 b3660886 Michael Hanselmann

138 25e7b43f Iustin Pop
  RFC2616, section 10.4.2: The request requires user
139 25e7b43f Iustin Pop
  authentication. The response MUST include a WWW-Authenticate header
140 25e7b43f Iustin Pop
  field (section 14.47) containing a challenge applicable to the
141 25e7b43f Iustin Pop
  requested resource.
142 b3660886 Michael Hanselmann

143 b3660886 Michael Hanselmann
  """
144 b3660886 Michael Hanselmann
  code = 401
145 b3660886 Michael Hanselmann
146 b3660886 Michael Hanselmann
147 84f2756e Michael Hanselmann
class HttpForbidden(HttpException):
148 b3660886 Michael Hanselmann
  """403 Forbidden
149 b3660886 Michael Hanselmann

150 25e7b43f Iustin Pop
  RFC2616, 10.4.4: The server understood the request, but is refusing
151 25e7b43f Iustin Pop
  to fulfill it.  Authorization will not help and the request SHOULD
152 25e7b43f Iustin Pop
  NOT be repeated.
153 b3660886 Michael Hanselmann

154 b3660886 Michael Hanselmann
  """
155 a43f68dc Michael Hanselmann
  code = 403
156 a43f68dc Michael Hanselmann
157 a43f68dc Michael Hanselmann
158 84f2756e Michael Hanselmann
class HttpNotFound(HttpException):
159 b3660886 Michael Hanselmann
  """404 Not Found
160 b3660886 Michael Hanselmann

161 25e7b43f Iustin Pop
  RFC2616, 10.4.5: The server has not found anything matching the
162 25e7b43f Iustin Pop
  Request-URI.  No indication is given of whether the condition is
163 25e7b43f Iustin Pop
  temporary or permanent.
164 b3660886 Michael Hanselmann

165 b3660886 Michael Hanselmann
  """
166 a43f68dc Michael Hanselmann
  code = 404
167 a43f68dc Michael Hanselmann
168 a43f68dc Michael Hanselmann
169 b3660886 Michael Hanselmann
class HttpMethodNotAllowed(HttpException):
170 b3660886 Michael Hanselmann
  """405 Method Not Allowed
171 b3660886 Michael Hanselmann

172 25e7b43f Iustin Pop
  RFC2616, 10.4.6: The method specified in the Request-Line is not
173 25e7b43f Iustin Pop
  allowed for the resource identified by the Request-URI. The response
174 25e7b43f Iustin Pop
  MUST include an Allow header containing a list of valid methods for
175 25e7b43f Iustin Pop
  the requested resource.
176 b3660886 Michael Hanselmann

177 b3660886 Michael Hanselmann
  """
178 b3660886 Michael Hanselmann
  code = 405
179 b3660886 Michael Hanselmann
180 b3660886 Michael Hanselmann
181 b3660886 Michael Hanselmann
class HttpRequestTimeout(HttpException):
182 b3660886 Michael Hanselmann
  """408 Request Timeout
183 b3660886 Michael Hanselmann

184 25e7b43f Iustin Pop
  RFC2616, 10.4.9: The client did not produce a request within the
185 25e7b43f Iustin Pop
  time that the server was prepared to wait. The client MAY repeat the
186 25e7b43f Iustin Pop
  request without modifications at any later time.
187 b3660886 Michael Hanselmann

188 b3660886 Michael Hanselmann
  """
189 b3660886 Michael Hanselmann
  code = 408
190 b3660886 Michael Hanselmann
191 b3660886 Michael Hanselmann
192 b3660886 Michael Hanselmann
class HttpConflict(HttpException):
193 b3660886 Michael Hanselmann
  """409 Conflict
194 b3660886 Michael Hanselmann

195 25e7b43f Iustin Pop
  RFC2616, 10.4.10: The request could not be completed due to a
196 25e7b43f Iustin Pop
  conflict with the current state of the resource. This code is only
197 25e7b43f Iustin Pop
  allowed in situations where it is expected that the user might be
198 25e7b43f Iustin Pop
  able to resolve the conflict and resubmit the request.
199 b3660886 Michael Hanselmann

200 b3660886 Michael Hanselmann
  """
201 b3660886 Michael Hanselmann
  code = 409
202 b3660886 Michael Hanselmann
203 b3660886 Michael Hanselmann
204 84f2756e Michael Hanselmann
class HttpGone(HttpException):
205 b3660886 Michael Hanselmann
  """410 Gone
206 b3660886 Michael Hanselmann

207 25e7b43f Iustin Pop
  RFC2616, 10.4.11: The requested resource is no longer available at
208 25e7b43f Iustin Pop
  the server and no forwarding address is known. This condition is
209 25e7b43f Iustin Pop
  expected to be considered permanent.
210 b3660886 Michael Hanselmann

211 b3660886 Michael Hanselmann
  """
212 a43f68dc Michael Hanselmann
  code = 410
213 a43f68dc Michael Hanselmann
214 a43f68dc Michael Hanselmann
215 84f2756e Michael Hanselmann
class HttpLengthRequired(HttpException):
216 b3660886 Michael Hanselmann
  """411 Length Required
217 b3660886 Michael Hanselmann

218 25e7b43f Iustin Pop
  RFC2616, 10.4.12: The server refuses to accept the request without a
219 25e7b43f Iustin Pop
  defined Content-Length. The client MAY repeat the request if it adds
220 25e7b43f Iustin Pop
  a valid Content-Length header field containing the length of the
221 25e7b43f Iustin Pop
  message-body in the request message.
222 b3660886 Michael Hanselmann

223 b3660886 Michael Hanselmann
  """
224 a43f68dc Michael Hanselmann
  code = 411
225 a43f68dc Michael Hanselmann
226 a43f68dc Michael Hanselmann
227 b3660886 Michael Hanselmann
class HttpPreconditionFailed(HttpException):
228 b3660886 Michael Hanselmann
  """412 Precondition Failed
229 b3660886 Michael Hanselmann

230 25e7b43f Iustin Pop
  RFC2616, 10.4.13: The precondition given in one or more of the
231 25e7b43f Iustin Pop
  request-header fields evaluated to false when it was tested on the
232 25e7b43f Iustin Pop
  server.
233 b3660886 Michael Hanselmann

234 b3660886 Michael Hanselmann
  """
235 b3660886 Michael Hanselmann
  code = 412
236 b3660886 Michael Hanselmann
237 b3660886 Michael Hanselmann
238 79589f25 Michael Hanselmann
class HttpInternalServerError(HttpException):
239 b3660886 Michael Hanselmann
  """500 Internal Server Error
240 b3660886 Michael Hanselmann

241 25e7b43f Iustin Pop
  RFC2616, 10.5.1: The server encountered an unexpected condition
242 25e7b43f Iustin Pop
  which prevented it from fulfilling the request.
243 b3660886 Michael Hanselmann

244 b3660886 Michael Hanselmann
  """
245 a43f68dc Michael Hanselmann
  code = 500
246 a43f68dc Michael Hanselmann
247 a43f68dc Michael Hanselmann
248 84f2756e Michael Hanselmann
class HttpNotImplemented(HttpException):
249 b3660886 Michael Hanselmann
  """501 Not Implemented
250 b3660886 Michael Hanselmann

251 25e7b43f Iustin Pop
  RFC2616, 10.5.2: The server does not support the functionality
252 25e7b43f Iustin Pop
  required to fulfill the request.
253 b3660886 Michael Hanselmann

254 b3660886 Michael Hanselmann
  """
255 a43f68dc Michael Hanselmann
  code = 501
256 a43f68dc Michael Hanselmann
257 a43f68dc Michael Hanselmann
258 77e1d753 Iustin Pop
class HttpBadGateway(HttpException):
259 77e1d753 Iustin Pop
  """502 Bad Gateway
260 77e1d753 Iustin Pop

261 77e1d753 Iustin Pop
  RFC2616, 10.5.3: The server, while acting as a gateway or proxy,
262 77e1d753 Iustin Pop
  received an invalid response from the upstream server it accessed in
263 77e1d753 Iustin Pop
  attempting to fulfill the request.
264 77e1d753 Iustin Pop

265 77e1d753 Iustin Pop
  """
266 77e1d753 Iustin Pop
  code = 502
267 77e1d753 Iustin Pop
268 77e1d753 Iustin Pop
269 84f2756e Michael Hanselmann
class HttpServiceUnavailable(HttpException):
270 b3660886 Michael Hanselmann
  """503 Service Unavailable
271 b3660886 Michael Hanselmann

272 25e7b43f Iustin Pop
  RFC2616, 10.5.4: The server is currently unable to handle the
273 25e7b43f Iustin Pop
  request due to a temporary overloading or maintenance of the server.
274 b3660886 Michael Hanselmann

275 b3660886 Michael Hanselmann
  """
276 a43f68dc Michael Hanselmann
  code = 503
277 a43f68dc Michael Hanselmann
278 a43f68dc Michael Hanselmann
279 77e1d753 Iustin Pop
class HttpGatewayTimeout(HttpException):
280 77e1d753 Iustin Pop
  """504 Gateway Timeout
281 77e1d753 Iustin Pop

282 77e1d753 Iustin Pop
  RFC2616, 10.5.5: The server, while acting as a gateway or proxy, did
283 77e1d753 Iustin Pop
  not receive a timely response from the upstream server specified by
284 77e1d753 Iustin Pop
  the URI (e.g.  HTTP, FTP, LDAP) or some other auxiliary server
285 77e1d753 Iustin Pop
  (e.g. DNS) it needed to access in attempting to complete the
286 77e1d753 Iustin Pop
  request.
287 77e1d753 Iustin Pop

288 77e1d753 Iustin Pop
  """
289 77e1d753 Iustin Pop
  code = 504
290 77e1d753 Iustin Pop
291 77e1d753 Iustin Pop
292 84f2756e Michael Hanselmann
class HttpVersionNotSupported(HttpException):
293 b3660886 Michael Hanselmann
  """505 HTTP Version Not Supported
294 b3660886 Michael Hanselmann

295 25e7b43f Iustin Pop
  RFC2616, 10.5.6: The server does not support, or refuses to support,
296 25e7b43f Iustin Pop
  the HTTP protocol version that was used in the request message.
297 b3660886 Michael Hanselmann

298 b3660886 Michael Hanselmann
  """
299 42242313 Michael Hanselmann
  code = 505
300 42242313 Michael Hanselmann
301 42242313 Michael Hanselmann
302 84f2756e Michael Hanselmann
class HttpJsonConverter:
303 a43f68dc Michael Hanselmann
  CONTENT_TYPE = "application/json"
304 a43f68dc Michael Hanselmann
305 a43f68dc Michael Hanselmann
  def Encode(self, data):
306 a43f68dc Michael Hanselmann
    return serializer.DumpJson(data)
307 a43f68dc Michael Hanselmann
308 a43f68dc Michael Hanselmann
  def Decode(self, data):
309 a43f68dc Michael Hanselmann
    return serializer.LoadJson(data)
310 a43f68dc Michael Hanselmann
311 a43f68dc Michael Hanselmann
312 aea0ed67 Michael Hanselmann
def WaitForSocketCondition(sock, event, timeout):
313 f22c1cea Michael Hanselmann
  """Waits for a condition to occur on the socket.
314 f22c1cea Michael Hanselmann

315 73a59d9e Michael Hanselmann
  @type sock: socket
316 358a8811 Michael Hanselmann
  @param sock: Wait for events on this socket
317 f22c1cea Michael Hanselmann
  @type event: int
318 f22c1cea Michael Hanselmann
  @param event: ORed condition (see select module)
319 f22c1cea Michael Hanselmann
  @type timeout: float or None
320 f22c1cea Michael Hanselmann
  @param timeout: Timeout in seconds
321 f22c1cea Michael Hanselmann
  @rtype: int or None
322 f22c1cea Michael Hanselmann
  @return: None for timeout, otherwise occured conditions
323 f22c1cea Michael Hanselmann

324 f22c1cea Michael Hanselmann
  """
325 f22c1cea Michael Hanselmann
  check = (event | select.POLLPRI |
326 f22c1cea Michael Hanselmann
           select.POLLNVAL | select.POLLHUP | select.POLLERR)
327 f22c1cea Michael Hanselmann
328 f22c1cea Michael Hanselmann
  if timeout is not None:
329 f22c1cea Michael Hanselmann
    # Poller object expects milliseconds
330 f22c1cea Michael Hanselmann
    timeout *= 1000
331 f22c1cea Michael Hanselmann
332 aea0ed67 Michael Hanselmann
  poller = select.poll()
333 f22c1cea Michael Hanselmann
  poller.register(sock, event)
334 f22c1cea Michael Hanselmann
  try:
335 f22c1cea Michael Hanselmann
    while True:
336 f22c1cea Michael Hanselmann
      # TODO: If the main thread receives a signal and we have no timeout, we
337 f22c1cea Michael Hanselmann
      # could wait forever. This should check a global "quit" flag or
338 f22c1cea Michael Hanselmann
      # something every so often.
339 f22c1cea Michael Hanselmann
      io_events = poller.poll(timeout)
340 f22c1cea Michael Hanselmann
      if not io_events:
341 f22c1cea Michael Hanselmann
        # Timeout
342 f22c1cea Michael Hanselmann
        return None
343 f22c1cea Michael Hanselmann
      for (evfd, evcond) in io_events:
344 f22c1cea Michael Hanselmann
        if evcond & check:
345 f22c1cea Michael Hanselmann
          return evcond
346 f22c1cea Michael Hanselmann
  finally:
347 f22c1cea Michael Hanselmann
    poller.unregister(sock)
348 f22c1cea Michael Hanselmann
349 f22c1cea Michael Hanselmann
350 aea0ed67 Michael Hanselmann
def SocketOperation(sock, op, arg1, timeout):
351 73a59d9e Michael Hanselmann
  """Wrapper around socket functions.
352 73a59d9e Michael Hanselmann

353 73a59d9e Michael Hanselmann
  This function abstracts error handling for socket operations, especially
354 73a59d9e Michael Hanselmann
  for the complicated interaction with OpenSSL.
355 73a59d9e Michael Hanselmann

356 73a59d9e Michael Hanselmann
  @type sock: socket
357 358a8811 Michael Hanselmann
  @param sock: Socket for the operation
358 73a59d9e Michael Hanselmann
  @type op: int
359 73a59d9e Michael Hanselmann
  @param op: Operation to execute (SOCKOP_* constants)
360 73a59d9e Michael Hanselmann
  @type arg1: any
361 73a59d9e Michael Hanselmann
  @param arg1: Parameter for function (if needed)
362 73a59d9e Michael Hanselmann
  @type timeout: None or float
363 73a59d9e Michael Hanselmann
  @param timeout: Timeout in seconds or None
364 358a8811 Michael Hanselmann
  @return: Return value of socket function
365 73a59d9e Michael Hanselmann

366 73a59d9e Michael Hanselmann
  """
367 73a59d9e Michael Hanselmann
  # TODO: event_poll/event_check/override
368 d7fa9982 Michael Hanselmann
  if op in (SOCKOP_SEND, SOCKOP_HANDSHAKE):
369 73a59d9e Michael Hanselmann
    event_poll = select.POLLOUT
370 73a59d9e Michael Hanselmann
    event_check = select.POLLOUT
371 73a59d9e Michael Hanselmann
372 73a59d9e Michael Hanselmann
  elif op == SOCKOP_RECV:
373 73a59d9e Michael Hanselmann
    event_poll = select.POLLIN
374 73a59d9e Michael Hanselmann
    event_check = select.POLLIN | select.POLLPRI
375 73a59d9e Michael Hanselmann
376 73a59d9e Michael Hanselmann
  elif op == SOCKOP_SHUTDOWN:
377 73a59d9e Michael Hanselmann
    event_poll = None
378 73a59d9e Michael Hanselmann
    event_check = None
379 73a59d9e Michael Hanselmann
380 73a59d9e Michael Hanselmann
    # The timeout is only used when OpenSSL requests polling for a condition.
381 73a59d9e Michael Hanselmann
    # It is not advisable to have no timeout for shutdown.
382 73a59d9e Michael Hanselmann
    assert timeout
383 73a59d9e Michael Hanselmann
384 73a59d9e Michael Hanselmann
  else:
385 73a59d9e Michael Hanselmann
    raise AssertionError("Invalid socket operation")
386 73a59d9e Michael Hanselmann
387 d7fa9982 Michael Hanselmann
  # Handshake is only supported by SSL sockets
388 d7fa9982 Michael Hanselmann
  if (op == SOCKOP_HANDSHAKE and
389 d7fa9982 Michael Hanselmann
      not isinstance(sock, OpenSSL.SSL.ConnectionType)):
390 d7fa9982 Michael Hanselmann
    return
391 d7fa9982 Michael Hanselmann
392 73a59d9e Michael Hanselmann
  # No override by default
393 73a59d9e Michael Hanselmann
  event_override = 0
394 73a59d9e Michael Hanselmann
395 73a59d9e Michael Hanselmann
  while True:
396 73a59d9e Michael Hanselmann
    # Poll only for certain operations and when asked for by an override
397 d7fa9982 Michael Hanselmann
    if event_override or op in (SOCKOP_SEND, SOCKOP_RECV, SOCKOP_HANDSHAKE):
398 73a59d9e Michael Hanselmann
      if event_override:
399 73a59d9e Michael Hanselmann
        wait_for_event = event_override
400 73a59d9e Michael Hanselmann
      else:
401 73a59d9e Michael Hanselmann
        wait_for_event = event_poll
402 73a59d9e Michael Hanselmann
403 aea0ed67 Michael Hanselmann
      event = WaitForSocketCondition(sock, wait_for_event, timeout)
404 73a59d9e Michael Hanselmann
      if event is None:
405 c9d0fa8a Michael Hanselmann
        raise HttpSocketTimeout()
406 73a59d9e Michael Hanselmann
407 73a59d9e Michael Hanselmann
      if (op == SOCKOP_RECV and
408 73a59d9e Michael Hanselmann
          event & (select.POLLNVAL | select.POLLHUP | select.POLLERR)):
409 73a59d9e Michael Hanselmann
        return ""
410 73a59d9e Michael Hanselmann
411 73a59d9e Michael Hanselmann
      if not event & wait_for_event:
412 73a59d9e Michael Hanselmann
        continue
413 73a59d9e Michael Hanselmann
414 73a59d9e Michael Hanselmann
    # Reset override
415 73a59d9e Michael Hanselmann
    event_override = 0
416 73a59d9e Michael Hanselmann
417 73a59d9e Michael Hanselmann
    try:
418 73a59d9e Michael Hanselmann
      try:
419 73a59d9e Michael Hanselmann
        if op == SOCKOP_SEND:
420 73a59d9e Michael Hanselmann
          return sock.send(arg1)
421 73a59d9e Michael Hanselmann
422 73a59d9e Michael Hanselmann
        elif op == SOCKOP_RECV:
423 73a59d9e Michael Hanselmann
          return sock.recv(arg1)
424 73a59d9e Michael Hanselmann
425 73a59d9e Michael Hanselmann
        elif op == SOCKOP_SHUTDOWN:
426 73a59d9e Michael Hanselmann
          if isinstance(sock, OpenSSL.SSL.ConnectionType):
427 73a59d9e Michael Hanselmann
            # PyOpenSSL's shutdown() doesn't take arguments
428 73a59d9e Michael Hanselmann
            return sock.shutdown()
429 73a59d9e Michael Hanselmann
          else:
430 73a59d9e Michael Hanselmann
            return sock.shutdown(arg1)
431 73a59d9e Michael Hanselmann
432 d7fa9982 Michael Hanselmann
        elif op == SOCKOP_HANDSHAKE:
433 d7fa9982 Michael Hanselmann
          return sock.do_handshake()
434 d7fa9982 Michael Hanselmann
435 73a59d9e Michael Hanselmann
      except OpenSSL.SSL.WantWriteError:
436 73a59d9e Michael Hanselmann
        # OpenSSL wants to write, poll for POLLOUT
437 73a59d9e Michael Hanselmann
        event_override = select.POLLOUT
438 73a59d9e Michael Hanselmann
        continue
439 73a59d9e Michael Hanselmann
440 73a59d9e Michael Hanselmann
      except OpenSSL.SSL.WantReadError:
441 73a59d9e Michael Hanselmann
        # OpenSSL wants to read, poll for POLLIN
442 73a59d9e Michael Hanselmann
        event_override = select.POLLIN | select.POLLPRI
443 73a59d9e Michael Hanselmann
        continue
444 73a59d9e Michael Hanselmann
445 73a59d9e Michael Hanselmann
      except OpenSSL.SSL.WantX509LookupError:
446 73a59d9e Michael Hanselmann
        continue
447 73a59d9e Michael Hanselmann
448 39cfc25b Michael Hanselmann
      except OpenSSL.SSL.ZeroReturnError, err:
449 39cfc25b Michael Hanselmann
        # SSL Connection has been closed. In SSL 3.0 and TLS 1.0, this only
450 39cfc25b Michael Hanselmann
        # occurs if a closure alert has occurred in the protocol, i.e. the
451 39cfc25b Michael Hanselmann
        # connection has been closed cleanly. Note that this does not
452 39cfc25b Michael Hanselmann
        # necessarily mean that the transport layer (e.g. a socket) has been
453 39cfc25b Michael Hanselmann
        # closed.
454 39cfc25b Michael Hanselmann
        if op == SOCKOP_SEND:
455 39cfc25b Michael Hanselmann
          # Can happen during a renegotiation
456 39cfc25b Michael Hanselmann
          raise HttpConnectionClosed(err.args)
457 39cfc25b Michael Hanselmann
        elif op == SOCKOP_RECV:
458 39cfc25b Michael Hanselmann
          return ""
459 39cfc25b Michael Hanselmann
460 39cfc25b Michael Hanselmann
        # SSL_shutdown shouldn't return SSL_ERROR_ZERO_RETURN
461 39cfc25b Michael Hanselmann
        raise socket.error(err.args)
462 39cfc25b Michael Hanselmann
463 73a59d9e Michael Hanselmann
      except OpenSSL.SSL.SysCallError, err:
464 73a59d9e Michael Hanselmann
        if op == SOCKOP_SEND:
465 73a59d9e Michael Hanselmann
          # arg1 is the data when writing
466 73a59d9e Michael Hanselmann
          if err.args and err.args[0] == -1 and arg1 == "":
467 73a59d9e Michael Hanselmann
            # errors when writing empty strings are expected
468 73a59d9e Michael Hanselmann
            # and can be ignored
469 73a59d9e Michael Hanselmann
            return 0
470 73a59d9e Michael Hanselmann
471 d7fa9982 Michael Hanselmann
        if err.args == (-1, _SSL_UNEXPECTED_EOF):
472 d7fa9982 Michael Hanselmann
          if op == SOCKOP_RECV:
473 73a59d9e Michael Hanselmann
            return ""
474 d7fa9982 Michael Hanselmann
          elif op == SOCKOP_HANDSHAKE:
475 d7fa9982 Michael Hanselmann
            # Can happen if peer disconnects directly after the connection is
476 d7fa9982 Michael Hanselmann
            # opened.
477 d7fa9982 Michael Hanselmann
            raise HttpSessionHandshakeUnexpectedEOF(err.args)
478 73a59d9e Michael Hanselmann
479 73a59d9e Michael Hanselmann
        raise socket.error(err.args)
480 73a59d9e Michael Hanselmann
481 73a59d9e Michael Hanselmann
      except OpenSSL.SSL.Error, err:
482 73a59d9e Michael Hanselmann
        raise socket.error(err.args)
483 73a59d9e Michael Hanselmann
484 73a59d9e Michael Hanselmann
    except socket.error, err:
485 73a59d9e Michael Hanselmann
      if err.args and err.args[0] == errno.EAGAIN:
486 73a59d9e Michael Hanselmann
        # Ignore EAGAIN
487 73a59d9e Michael Hanselmann
        continue
488 73a59d9e Michael Hanselmann
489 73a59d9e Michael Hanselmann
      raise
490 73a59d9e Michael Hanselmann
491 73a59d9e Michael Hanselmann
492 aea0ed67 Michael Hanselmann
def ShutdownConnection(sock, close_timeout, write_timeout, msgreader, force):
493 02cab3e7 Michael Hanselmann
  """Closes the connection.
494 02cab3e7 Michael Hanselmann

495 358a8811 Michael Hanselmann
  @type sock: socket
496 358a8811 Michael Hanselmann
  @param sock: Socket to be shut down
497 358a8811 Michael Hanselmann
  @type close_timeout: float
498 25e7b43f Iustin Pop
  @param close_timeout: How long to wait for the peer to close
499 25e7b43f Iustin Pop
      the connection
500 358a8811 Michael Hanselmann
  @type write_timeout: float
501 358a8811 Michael Hanselmann
  @param write_timeout: Write timeout for shutdown
502 358a8811 Michael Hanselmann
  @type msgreader: http.HttpMessageReader
503 25e7b43f Iustin Pop
  @param msgreader: Request message reader, used to determine whether
504 25e7b43f Iustin Pop
      peer should close connection
505 358a8811 Michael Hanselmann
  @type force: bool
506 25e7b43f Iustin Pop
  @param force: Whether to forcibly close the connection without
507 25e7b43f Iustin Pop
      waiting for peer
508 358a8811 Michael Hanselmann

509 02cab3e7 Michael Hanselmann
  """
510 02cab3e7 Michael Hanselmann
  #print msgreader.peer_will_close, force
511 02cab3e7 Michael Hanselmann
  if msgreader and msgreader.peer_will_close and not force:
512 02cab3e7 Michael Hanselmann
    # Wait for peer to close
513 02cab3e7 Michael Hanselmann
    try:
514 02cab3e7 Michael Hanselmann
      # Check whether it's actually closed
515 aea0ed67 Michael Hanselmann
      if not SocketOperation(sock, SOCKOP_RECV, 1, close_timeout):
516 02cab3e7 Michael Hanselmann
        return
517 02cab3e7 Michael Hanselmann
    except (socket.error, HttpError, HttpSocketTimeout):
518 02cab3e7 Michael Hanselmann
      # Ignore errors at this stage
519 02cab3e7 Michael Hanselmann
      pass
520 02cab3e7 Michael Hanselmann
521 02cab3e7 Michael Hanselmann
  # Close the connection from our side
522 02cab3e7 Michael Hanselmann
  try:
523 39cfc25b Michael Hanselmann
    # We don't care about the return value, see NOTES in SSL_shutdown(3).
524 aea0ed67 Michael Hanselmann
    SocketOperation(sock, SOCKOP_SHUTDOWN, socket.SHUT_RDWR,
525 02cab3e7 Michael Hanselmann
                    write_timeout)
526 02cab3e7 Michael Hanselmann
  except HttpSocketTimeout:
527 02cab3e7 Michael Hanselmann
    raise HttpError("Timeout while shutting down connection")
528 02cab3e7 Michael Hanselmann
  except socket.error, err:
529 45eac583 Michael Hanselmann
    # Ignore ENOTCONN
530 45eac583 Michael Hanselmann
    if not (err.args and err.args[0] == errno.ENOTCONN):
531 45eac583 Michael Hanselmann
      raise HttpError("Error while shutting down connection: %s" % err)
532 02cab3e7 Michael Hanselmann
533 02cab3e7 Michael Hanselmann
534 aea0ed67 Michael Hanselmann
def Handshake(sock, write_timeout):
535 d7fa9982 Michael Hanselmann
  """Shakes peer's hands.
536 d7fa9982 Michael Hanselmann

537 d7fa9982 Michael Hanselmann
  @type sock: socket
538 d7fa9982 Michael Hanselmann
  @param sock: Socket to be shut down
539 d7fa9982 Michael Hanselmann
  @type write_timeout: float
540 d7fa9982 Michael Hanselmann
  @param write_timeout: Write timeout for handshake
541 d7fa9982 Michael Hanselmann

542 d7fa9982 Michael Hanselmann
  """
543 d7fa9982 Michael Hanselmann
  try:
544 aea0ed67 Michael Hanselmann
    return SocketOperation(sock, SOCKOP_HANDSHAKE, None, write_timeout)
545 d7fa9982 Michael Hanselmann
  except HttpSocketTimeout:
546 d7fa9982 Michael Hanselmann
    raise HttpError("Timeout during SSL handshake")
547 d7fa9982 Michael Hanselmann
  except socket.error, err:
548 d7fa9982 Michael Hanselmann
    raise HttpError("Error in SSL handshake: %s" % err)
549 d7fa9982 Michael Hanselmann
550 d7fa9982 Michael Hanselmann
551 f20cbea2 Michael Hanselmann
class HttpSslParams(object):
552 f20cbea2 Michael Hanselmann
  """Data class for SSL key and certificate.
553 f20cbea2 Michael Hanselmann

554 f20cbea2 Michael Hanselmann
  """
555 f20cbea2 Michael Hanselmann
  def __init__(self, ssl_key_path, ssl_cert_path):
556 f20cbea2 Michael Hanselmann
    """Initializes this class.
557 f20cbea2 Michael Hanselmann

558 f20cbea2 Michael Hanselmann
    @type ssl_key_path: string
559 f20cbea2 Michael Hanselmann
    @param ssl_key_path: Path to file containing SSL key in PEM format
560 f20cbea2 Michael Hanselmann
    @type ssl_cert_path: string
561 25e7b43f Iustin Pop
    @param ssl_cert_path: Path to file containing SSL certificate
562 25e7b43f Iustin Pop
        in PEM format
563 f20cbea2 Michael Hanselmann

564 f20cbea2 Michael Hanselmann
    """
565 65c6b8e0 Michael Hanselmann
    self.ssl_key_pem = utils.ReadFile(ssl_key_path)
566 65c6b8e0 Michael Hanselmann
    self.ssl_cert_pem = utils.ReadFile(ssl_cert_path)
567 f20cbea2 Michael Hanselmann
568 65c6b8e0 Michael Hanselmann
  def GetKey(self):
569 65c6b8e0 Michael Hanselmann
    return OpenSSL.crypto.load_privatekey(OpenSSL.crypto.FILETYPE_PEM,
570 65c6b8e0 Michael Hanselmann
                                          self.ssl_key_pem)
571 65c6b8e0 Michael Hanselmann
572 65c6b8e0 Michael Hanselmann
  def GetCertificate(self):
573 65c6b8e0 Michael Hanselmann
    return OpenSSL.crypto.load_certificate(OpenSSL.crypto.FILETYPE_PEM,
574 65c6b8e0 Michael Hanselmann
                                           self.ssl_cert_pem)
575 f20cbea2 Michael Hanselmann
576 f20cbea2 Michael Hanselmann
577 f4322a1e Michael Hanselmann
class HttpBase(object):
578 b14f759e Michael Hanselmann
  """Base class for HTTP server and client.
579 b14f759e Michael Hanselmann

580 b14f759e Michael Hanselmann
  """
581 b14f759e Michael Hanselmann
  def __init__(self):
582 22692e48 Michael Hanselmann
    self.using_ssl = None
583 f20cbea2 Michael Hanselmann
    self._ssl_params = None
584 65c6b8e0 Michael Hanselmann
    self._ssl_key = None
585 65c6b8e0 Michael Hanselmann
    self._ssl_cert = None
586 b14f759e Michael Hanselmann
587 f20cbea2 Michael Hanselmann
  def _CreateSocket(self, ssl_params, ssl_verify_peer):
588 b14f759e Michael Hanselmann
    """Creates a TCP socket and initializes SSL if needed.
589 b14f759e Michael Hanselmann

590 f20cbea2 Michael Hanselmann
    @type ssl_params: HttpSslParams
591 f20cbea2 Michael Hanselmann
    @param ssl_params: SSL key and certificate
592 b14f759e Michael Hanselmann
    @type ssl_verify_peer: bool
593 25e7b43f Iustin Pop
    @param ssl_verify_peer: Whether to require client certificate
594 25e7b43f Iustin Pop
        and compare it with our certificate
595 b14f759e Michael Hanselmann

596 b14f759e Michael Hanselmann
    """
597 f20cbea2 Michael Hanselmann
    self._ssl_params = ssl_params
598 f20cbea2 Michael Hanselmann
599 b14f759e Michael Hanselmann
    sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
600 b14f759e Michael Hanselmann
601 b14f759e Michael Hanselmann
    # Should we enable SSL?
602 22692e48 Michael Hanselmann
    self.using_ssl = ssl_params is not None
603 b14f759e Michael Hanselmann
604 22692e48 Michael Hanselmann
    if not self.using_ssl:
605 b14f759e Michael Hanselmann
      return sock
606 b14f759e Michael Hanselmann
607 65c6b8e0 Michael Hanselmann
    self._ssl_key = ssl_params.GetKey()
608 65c6b8e0 Michael Hanselmann
    self._ssl_cert = ssl_params.GetCertificate()
609 65c6b8e0 Michael Hanselmann
610 b14f759e Michael Hanselmann
    ctx = OpenSSL.SSL.Context(OpenSSL.SSL.SSLv23_METHOD)
611 b14f759e Michael Hanselmann
    ctx.set_options(OpenSSL.SSL.OP_NO_SSLv2)
612 b14f759e Michael Hanselmann
613 65c6b8e0 Michael Hanselmann
    ctx.use_privatekey(self._ssl_key)
614 65c6b8e0 Michael Hanselmann
    ctx.use_certificate(self._ssl_cert)
615 b14f759e Michael Hanselmann
    ctx.check_privatekey()
616 b14f759e Michael Hanselmann
617 b14f759e Michael Hanselmann
    if ssl_verify_peer:
618 b14f759e Michael Hanselmann
      ctx.set_verify(OpenSSL.SSL.VERIFY_PEER |
619 b14f759e Michael Hanselmann
                     OpenSSL.SSL.VERIFY_FAIL_IF_NO_PEER_CERT,
620 b14f759e Michael Hanselmann
                     self._SSLVerifyCallback)
621 b14f759e Michael Hanselmann
622 b14f759e Michael Hanselmann
    return OpenSSL.SSL.Connection(ctx, sock)
623 b14f759e Michael Hanselmann
624 b14f759e Michael Hanselmann
  def _SSLVerifyCallback(self, conn, cert, errnum, errdepth, ok):
625 b14f759e Michael Hanselmann
    """Verify the certificate provided by the peer
626 b14f759e Michael Hanselmann

627 b14f759e Michael Hanselmann
    We only compare fingerprints. The client must use the same certificate as
628 b14f759e Michael Hanselmann
    we do on our side.
629 b14f759e Michael Hanselmann

630 b14f759e Michael Hanselmann
    """
631 f20cbea2 Michael Hanselmann
    assert self._ssl_params, "SSL not initialized"
632 b14f759e Michael Hanselmann
633 65c6b8e0 Michael Hanselmann
    return (self._ssl_cert.digest("sha1") == cert.digest("sha1") and
634 65c6b8e0 Michael Hanselmann
            self._ssl_cert.digest("md5") == cert.digest("md5"))
635 b14f759e Michael Hanselmann
636 b14f759e Michael Hanselmann
637 02cab3e7 Michael Hanselmann
class HttpMessage(object):
638 02cab3e7 Michael Hanselmann
  """Data structure for HTTP message.
639 02cab3e7 Michael Hanselmann

640 02cab3e7 Michael Hanselmann
  """
641 02cab3e7 Michael Hanselmann
  def __init__(self):
642 02cab3e7 Michael Hanselmann
    self.start_line = None
643 02cab3e7 Michael Hanselmann
    self.headers = None
644 02cab3e7 Michael Hanselmann
    self.body = None
645 02cab3e7 Michael Hanselmann
    self.decoded_body = None
646 02cab3e7 Michael Hanselmann
647 02cab3e7 Michael Hanselmann
648 02cab3e7 Michael Hanselmann
class HttpClientToServerStartLine(object):
649 02cab3e7 Michael Hanselmann
  """Data structure for HTTP request start line.
650 02cab3e7 Michael Hanselmann

651 02cab3e7 Michael Hanselmann
  """
652 02cab3e7 Michael Hanselmann
  def __init__(self, method, path, version):
653 02cab3e7 Michael Hanselmann
    self.method = method
654 02cab3e7 Michael Hanselmann
    self.path = path
655 02cab3e7 Michael Hanselmann
    self.version = version
656 02cab3e7 Michael Hanselmann
657 02cab3e7 Michael Hanselmann
  def __str__(self):
658 02cab3e7 Michael Hanselmann
    return "%s %s %s" % (self.method, self.path, self.version)
659 02cab3e7 Michael Hanselmann
660 02cab3e7 Michael Hanselmann
661 02cab3e7 Michael Hanselmann
class HttpServerToClientStartLine(object):
662 02cab3e7 Michael Hanselmann
  """Data structure for HTTP response start line.
663 02cab3e7 Michael Hanselmann

664 02cab3e7 Michael Hanselmann
  """
665 02cab3e7 Michael Hanselmann
  def __init__(self, version, code, reason):
666 02cab3e7 Michael Hanselmann
    self.version = version
667 02cab3e7 Michael Hanselmann
    self.code = code
668 02cab3e7 Michael Hanselmann
    self.reason = reason
669 02cab3e7 Michael Hanselmann
670 02cab3e7 Michael Hanselmann
  def __str__(self):
671 02cab3e7 Michael Hanselmann
    return "%s %s %s" % (self.version, self.code, self.reason)
672 02cab3e7 Michael Hanselmann
673 02cab3e7 Michael Hanselmann
674 02cab3e7 Michael Hanselmann
class HttpMessageWriter(object):
675 02cab3e7 Michael Hanselmann
  """Writes an HTTP message to a socket.
676 02cab3e7 Michael Hanselmann

677 02cab3e7 Michael Hanselmann
  """
678 02cab3e7 Michael Hanselmann
  def __init__(self, sock, msg, write_timeout):
679 358a8811 Michael Hanselmann
    """Initializes this class and writes an HTTP message to a socket.
680 358a8811 Michael Hanselmann

681 358a8811 Michael Hanselmann
    @type sock: socket
682 358a8811 Michael Hanselmann
    @param sock: Socket to be written to
683 358a8811 Michael Hanselmann
    @type msg: http.HttpMessage
684 358a8811 Michael Hanselmann
    @param msg: HTTP message to be written
685 358a8811 Michael Hanselmann
    @type write_timeout: float
686 358a8811 Michael Hanselmann
    @param write_timeout: Write timeout for socket
687 358a8811 Michael Hanselmann

688 358a8811 Michael Hanselmann
    """
689 02cab3e7 Michael Hanselmann
    self._msg = msg
690 02cab3e7 Michael Hanselmann
691 02cab3e7 Michael Hanselmann
    self._PrepareMessage()
692 02cab3e7 Michael Hanselmann
693 02cab3e7 Michael Hanselmann
    buf = self._FormatMessage()
694 02cab3e7 Michael Hanselmann
695 b18dd019 Iustin Pop
    pos = 0
696 b18dd019 Iustin Pop
    end = len(buf)
697 b18dd019 Iustin Pop
    while pos < end:
698 b18dd019 Iustin Pop
      # Send only SOCK_BUF_SIZE bytes at a time
699 358a8811 Michael Hanselmann
      data = buf[pos:(pos + SOCK_BUF_SIZE)]
700 02cab3e7 Michael Hanselmann
701 aea0ed67 Michael Hanselmann
      sent = SocketOperation(sock, SOCKOP_SEND, data, write_timeout)
702 02cab3e7 Michael Hanselmann
703 02cab3e7 Michael Hanselmann
      # Remove sent bytes
704 b18dd019 Iustin Pop
      pos += sent
705 02cab3e7 Michael Hanselmann
706 b18dd019 Iustin Pop
    assert pos == end, "Message wasn't sent completely"
707 02cab3e7 Michael Hanselmann
708 02cab3e7 Michael Hanselmann
  def _PrepareMessage(self):
709 02cab3e7 Michael Hanselmann
    """Prepares the HTTP message by setting mandatory headers.
710 02cab3e7 Michael Hanselmann

711 02cab3e7 Michael Hanselmann
    """
712 02cab3e7 Michael Hanselmann
    # RFC2616, section 4.3: "The presence of a message-body in a request is
713 02cab3e7 Michael Hanselmann
    # signaled by the inclusion of a Content-Length or Transfer-Encoding header
714 02cab3e7 Michael Hanselmann
    # field in the request's message-headers."
715 02cab3e7 Michael Hanselmann
    if self._msg.body:
716 02cab3e7 Michael Hanselmann
      self._msg.headers[HTTP_CONTENT_LENGTH] = len(self._msg.body)
717 02cab3e7 Michael Hanselmann
718 02cab3e7 Michael Hanselmann
  def _FormatMessage(self):
719 02cab3e7 Michael Hanselmann
    """Serializes the HTTP message into a string.
720 02cab3e7 Michael Hanselmann

721 02cab3e7 Michael Hanselmann
    """
722 02cab3e7 Michael Hanselmann
    buf = StringIO()
723 02cab3e7 Michael Hanselmann
724 02cab3e7 Michael Hanselmann
    # Add start line
725 02cab3e7 Michael Hanselmann
    buf.write(str(self._msg.start_line))
726 02cab3e7 Michael Hanselmann
    buf.write("\r\n")
727 02cab3e7 Michael Hanselmann
728 02cab3e7 Michael Hanselmann
    # Add headers
729 02cab3e7 Michael Hanselmann
    if self._msg.start_line.version != HTTP_0_9:
730 02cab3e7 Michael Hanselmann
      for name, value in self._msg.headers.iteritems():
731 02cab3e7 Michael Hanselmann
        buf.write("%s: %s\r\n" % (name, value))
732 02cab3e7 Michael Hanselmann
733 02cab3e7 Michael Hanselmann
    buf.write("\r\n")
734 02cab3e7 Michael Hanselmann
735 02cab3e7 Michael Hanselmann
    # Add message body if needed
736 02cab3e7 Michael Hanselmann
    if self.HasMessageBody():
737 02cab3e7 Michael Hanselmann
      buf.write(self._msg.body)
738 02cab3e7 Michael Hanselmann
739 02cab3e7 Michael Hanselmann
    elif self._msg.body:
740 02cab3e7 Michael Hanselmann
      logging.warning("Ignoring message body")
741 02cab3e7 Michael Hanselmann
742 02cab3e7 Michael Hanselmann
    return buf.getvalue()
743 02cab3e7 Michael Hanselmann
744 02cab3e7 Michael Hanselmann
  def HasMessageBody(self):
745 02cab3e7 Michael Hanselmann
    """Checks whether the HTTP message contains a body.
746 02cab3e7 Michael Hanselmann

747 02cab3e7 Michael Hanselmann
    Can be overriden by subclasses.
748 02cab3e7 Michael Hanselmann

749 02cab3e7 Michael Hanselmann
    """
750 02cab3e7 Michael Hanselmann
    return bool(self._msg.body)
751 02cab3e7 Michael Hanselmann
752 02cab3e7 Michael Hanselmann
753 02cab3e7 Michael Hanselmann
class HttpMessageReader(object):
754 02cab3e7 Michael Hanselmann
  """Reads HTTP message from socket.
755 02cab3e7 Michael Hanselmann

756 02cab3e7 Michael Hanselmann
  """
757 02cab3e7 Michael Hanselmann
  # Length limits
758 02cab3e7 Michael Hanselmann
  START_LINE_LENGTH_MAX = None
759 02cab3e7 Michael Hanselmann
  HEADER_LENGTH_MAX = None
760 02cab3e7 Michael Hanselmann
761 02cab3e7 Michael Hanselmann
  # Parser state machine
762 02cab3e7 Michael Hanselmann
  PS_START_LINE = "start-line"
763 02cab3e7 Michael Hanselmann
  PS_HEADERS = "headers"
764 02cab3e7 Michael Hanselmann
  PS_BODY = "entity-body"
765 02cab3e7 Michael Hanselmann
  PS_COMPLETE = "complete"
766 02cab3e7 Michael Hanselmann
767 02cab3e7 Michael Hanselmann
  def __init__(self, sock, msg, read_timeout):
768 358a8811 Michael Hanselmann
    """Reads an HTTP message from a socket.
769 358a8811 Michael Hanselmann

770 358a8811 Michael Hanselmann
    @type sock: socket
771 358a8811 Michael Hanselmann
    @param sock: Socket to be read from
772 358a8811 Michael Hanselmann
    @type msg: http.HttpMessage
773 358a8811 Michael Hanselmann
    @param msg: Object for the read message
774 358a8811 Michael Hanselmann
    @type read_timeout: float
775 358a8811 Michael Hanselmann
    @param read_timeout: Read timeout for socket
776 358a8811 Michael Hanselmann

777 358a8811 Michael Hanselmann
    """
778 02cab3e7 Michael Hanselmann
    self.sock = sock
779 02cab3e7 Michael Hanselmann
    self.msg = msg
780 02cab3e7 Michael Hanselmann
781 02cab3e7 Michael Hanselmann
    self.start_line_buffer = None
782 02cab3e7 Michael Hanselmann
    self.header_buffer = StringIO()
783 02cab3e7 Michael Hanselmann
    self.body_buffer = StringIO()
784 02cab3e7 Michael Hanselmann
    self.parser_status = self.PS_START_LINE
785 02cab3e7 Michael Hanselmann
    self.content_length = None
786 02cab3e7 Michael Hanselmann
    self.peer_will_close = None
787 02cab3e7 Michael Hanselmann
788 02cab3e7 Michael Hanselmann
    buf = ""
789 02cab3e7 Michael Hanselmann
    eof = False
790 02cab3e7 Michael Hanselmann
    while self.parser_status != self.PS_COMPLETE:
791 f088165d Michael Hanselmann
      # TODO: Don't read more than necessary (Content-Length), otherwise
792 f088165d Michael Hanselmann
      # data might be lost and/or an error could occur
793 aea0ed67 Michael Hanselmann
      data = SocketOperation(sock, SOCKOP_RECV, SOCK_BUF_SIZE, read_timeout)
794 02cab3e7 Michael Hanselmann
795 02cab3e7 Michael Hanselmann
      if data:
796 02cab3e7 Michael Hanselmann
        buf += data
797 02cab3e7 Michael Hanselmann
      else:
798 02cab3e7 Michael Hanselmann
        eof = True
799 02cab3e7 Michael Hanselmann
800 02cab3e7 Michael Hanselmann
      # Do some parsing and error checking while more data arrives
801 02cab3e7 Michael Hanselmann
      buf = self._ContinueParsing(buf, eof)
802 02cab3e7 Michael Hanselmann
803 02cab3e7 Michael Hanselmann
      # Must be done only after the buffer has been evaluated
804 02cab3e7 Michael Hanselmann
      # TODO: Connection-length < len(data read) and connection closed
805 02cab3e7 Michael Hanselmann
      if (eof and
806 02cab3e7 Michael Hanselmann
          self.parser_status in (self.PS_START_LINE,
807 02cab3e7 Michael Hanselmann
                                 self.PS_HEADERS)):
808 02cab3e7 Michael Hanselmann
        raise HttpError("Connection closed prematurely")
809 02cab3e7 Michael Hanselmann
810 02cab3e7 Michael Hanselmann
    # Parse rest
811 02cab3e7 Michael Hanselmann
    buf = self._ContinueParsing(buf, True)
812 02cab3e7 Michael Hanselmann
813 02cab3e7 Michael Hanselmann
    assert self.parser_status == self.PS_COMPLETE
814 02cab3e7 Michael Hanselmann
    assert not buf, "Parser didn't read full response"
815 02cab3e7 Michael Hanselmann
816 02cab3e7 Michael Hanselmann
    msg.body = self.body_buffer.getvalue()
817 02cab3e7 Michael Hanselmann
818 02cab3e7 Michael Hanselmann
    # TODO: Content-type, error handling
819 02cab3e7 Michael Hanselmann
    if msg.body:
820 02cab3e7 Michael Hanselmann
      msg.decoded_body = HttpJsonConverter().Decode(msg.body)
821 02cab3e7 Michael Hanselmann
    else:
822 02cab3e7 Michael Hanselmann
      msg.decoded_body = None
823 02cab3e7 Michael Hanselmann
824 02cab3e7 Michael Hanselmann
    if msg.decoded_body:
825 02cab3e7 Michael Hanselmann
      logging.debug("Message body: %s", msg.decoded_body)
826 02cab3e7 Michael Hanselmann
827 02cab3e7 Michael Hanselmann
  def _ContinueParsing(self, buf, eof):
828 02cab3e7 Michael Hanselmann
    """Main function for HTTP message state machine.
829 02cab3e7 Michael Hanselmann

830 02cab3e7 Michael Hanselmann
    @type buf: string
831 02cab3e7 Michael Hanselmann
    @param buf: Receive buffer
832 02cab3e7 Michael Hanselmann
    @type eof: bool
833 02cab3e7 Michael Hanselmann
    @param eof: Whether we've reached EOF on the socket
834 02cab3e7 Michael Hanselmann
    @rtype: string
835 02cab3e7 Michael Hanselmann
    @return: Updated receive buffer
836 02cab3e7 Michael Hanselmann

837 02cab3e7 Michael Hanselmann
    """
838 f088165d Michael Hanselmann
    # TODO: Use offset instead of slicing when possible
839 02cab3e7 Michael Hanselmann
    if self.parser_status == self.PS_START_LINE:
840 02cab3e7 Michael Hanselmann
      # Expect start line
841 02cab3e7 Michael Hanselmann
      while True:
842 02cab3e7 Michael Hanselmann
        idx = buf.find("\r\n")
843 02cab3e7 Michael Hanselmann
844 02cab3e7 Michael Hanselmann
        # RFC2616, section 4.1: "In the interest of robustness, servers SHOULD
845 02cab3e7 Michael Hanselmann
        # ignore any empty line(s) received where a Request-Line is expected.
846 02cab3e7 Michael Hanselmann
        # In other words, if the server is reading the protocol stream at the
847 02cab3e7 Michael Hanselmann
        # beginning of a message and receives a CRLF first, it should ignore
848 02cab3e7 Michael Hanselmann
        # the CRLF."
849 02cab3e7 Michael Hanselmann
        if idx == 0:
850 358a8811 Michael Hanselmann
          # TODO: Limit number of CRLFs/empty lines for safety?
851 02cab3e7 Michael Hanselmann
          buf = buf[:2]
852 02cab3e7 Michael Hanselmann
          continue
853 02cab3e7 Michael Hanselmann
854 02cab3e7 Michael Hanselmann
        if idx > 0:
855 02cab3e7 Michael Hanselmann
          self.start_line_buffer = buf[:idx]
856 02cab3e7 Michael Hanselmann
857 02cab3e7 Michael Hanselmann
          self._CheckStartLineLength(len(self.start_line_buffer))
858 02cab3e7 Michael Hanselmann
859 02cab3e7 Michael Hanselmann
          # Remove status line, including CRLF
860 02cab3e7 Michael Hanselmann
          buf = buf[idx + 2:]
861 02cab3e7 Michael Hanselmann
862 02cab3e7 Michael Hanselmann
          self.msg.start_line = self.ParseStartLine(self.start_line_buffer)
863 02cab3e7 Michael Hanselmann
864 02cab3e7 Michael Hanselmann
          self.parser_status = self.PS_HEADERS
865 02cab3e7 Michael Hanselmann
        else:
866 02cab3e7 Michael Hanselmann
          # Check whether incoming data is getting too large, otherwise we just
867 02cab3e7 Michael Hanselmann
          # fill our read buffer.
868 02cab3e7 Michael Hanselmann
          self._CheckStartLineLength(len(buf))
869 02cab3e7 Michael Hanselmann
870 02cab3e7 Michael Hanselmann
        break
871 02cab3e7 Michael Hanselmann
872 02cab3e7 Michael Hanselmann
    # TODO: Handle messages without headers
873 02cab3e7 Michael Hanselmann
    if self.parser_status == self.PS_HEADERS:
874 02cab3e7 Michael Hanselmann
      # Wait for header end
875 02cab3e7 Michael Hanselmann
      idx = buf.find("\r\n\r\n")
876 02cab3e7 Michael Hanselmann
      if idx >= 0:
877 02cab3e7 Michael Hanselmann
        self.header_buffer.write(buf[:idx + 2])
878 02cab3e7 Michael Hanselmann
879 02cab3e7 Michael Hanselmann
        self._CheckHeaderLength(self.header_buffer.tell())
880 02cab3e7 Michael Hanselmann
881 02cab3e7 Michael Hanselmann
        # Remove headers, including CRLF
882 02cab3e7 Michael Hanselmann
        buf = buf[idx + 4:]
883 02cab3e7 Michael Hanselmann
884 02cab3e7 Michael Hanselmann
        self._ParseHeaders()
885 02cab3e7 Michael Hanselmann
886 02cab3e7 Michael Hanselmann
        self.parser_status = self.PS_BODY
887 02cab3e7 Michael Hanselmann
      else:
888 02cab3e7 Michael Hanselmann
        # Check whether incoming data is getting too large, otherwise we just
889 02cab3e7 Michael Hanselmann
        # fill our read buffer.
890 02cab3e7 Michael Hanselmann
        self._CheckHeaderLength(len(buf))
891 02cab3e7 Michael Hanselmann
892 02cab3e7 Michael Hanselmann
    if self.parser_status == self.PS_BODY:
893 02cab3e7 Michael Hanselmann
      # TODO: Implement max size for body_buffer
894 02cab3e7 Michael Hanselmann
      self.body_buffer.write(buf)
895 02cab3e7 Michael Hanselmann
      buf = ""
896 02cab3e7 Michael Hanselmann
897 02cab3e7 Michael Hanselmann
      # Check whether we've read everything
898 02cab3e7 Michael Hanselmann
      #
899 02cab3e7 Michael Hanselmann
      # RFC2616, section 4.4: "When a message-body is included with a message,
900 02cab3e7 Michael Hanselmann
      # the transfer-length of that body is determined by one of the following
901 02cab3e7 Michael Hanselmann
      # [...] 5. By the server closing the connection. (Closing the connection
902 02cab3e7 Michael Hanselmann
      # cannot be used to indicate the end of a request body, since that would
903 02cab3e7 Michael Hanselmann
      # leave no possibility for the server to send back a response.)"
904 f088165d Michael Hanselmann
      #
905 f088165d Michael Hanselmann
      # TODO: Error when buffer length > Content-Length header
906 02cab3e7 Michael Hanselmann
      if (eof or
907 02cab3e7 Michael Hanselmann
          self.content_length is None or
908 02cab3e7 Michael Hanselmann
          (self.content_length is not None and
909 02cab3e7 Michael Hanselmann
           self.body_buffer.tell() >= self.content_length)):
910 02cab3e7 Michael Hanselmann
        self.parser_status = self.PS_COMPLETE
911 02cab3e7 Michael Hanselmann
912 02cab3e7 Michael Hanselmann
    return buf
913 02cab3e7 Michael Hanselmann
914 02cab3e7 Michael Hanselmann
  def _CheckStartLineLength(self, length):
915 02cab3e7 Michael Hanselmann
    """Limits the start line buffer size.
916 02cab3e7 Michael Hanselmann

917 02cab3e7 Michael Hanselmann
    @type length: int
918 02cab3e7 Michael Hanselmann
    @param length: Buffer size
919 02cab3e7 Michael Hanselmann

920 02cab3e7 Michael Hanselmann
    """
921 02cab3e7 Michael Hanselmann
    if (self.START_LINE_LENGTH_MAX is not None and
922 02cab3e7 Michael Hanselmann
        length > self.START_LINE_LENGTH_MAX):
923 02cab3e7 Michael Hanselmann
      raise HttpError("Start line longer than %d chars" %
924 02cab3e7 Michael Hanselmann
                       self.START_LINE_LENGTH_MAX)
925 02cab3e7 Michael Hanselmann
926 02cab3e7 Michael Hanselmann
  def _CheckHeaderLength(self, length):
927 02cab3e7 Michael Hanselmann
    """Limits the header buffer size.
928 02cab3e7 Michael Hanselmann

929 02cab3e7 Michael Hanselmann
    @type length: int
930 02cab3e7 Michael Hanselmann
    @param length: Buffer size
931 02cab3e7 Michael Hanselmann

932 02cab3e7 Michael Hanselmann
    """
933 02cab3e7 Michael Hanselmann
    if (self.HEADER_LENGTH_MAX is not None and
934 02cab3e7 Michael Hanselmann
        length > self.HEADER_LENGTH_MAX):
935 02cab3e7 Michael Hanselmann
      raise HttpError("Headers longer than %d chars" % self.HEADER_LENGTH_MAX)
936 02cab3e7 Michael Hanselmann
937 02cab3e7 Michael Hanselmann
  def ParseStartLine(self, start_line):
938 02cab3e7 Michael Hanselmann
    """Parses the start line of a message.
939 02cab3e7 Michael Hanselmann

940 02cab3e7 Michael Hanselmann
    Must be overriden by subclass.
941 02cab3e7 Michael Hanselmann

942 02cab3e7 Michael Hanselmann
    @type start_line: string
943 02cab3e7 Michael Hanselmann
    @param start_line: Start line string
944 02cab3e7 Michael Hanselmann

945 02cab3e7 Michael Hanselmann
    """
946 02cab3e7 Michael Hanselmann
    raise NotImplementedError()
947 02cab3e7 Michael Hanselmann
948 02cab3e7 Michael Hanselmann
  def _WillPeerCloseConnection(self):
949 02cab3e7 Michael Hanselmann
    """Evaluate whether peer will close the connection.
950 02cab3e7 Michael Hanselmann

951 02cab3e7 Michael Hanselmann
    @rtype: bool
952 02cab3e7 Michael Hanselmann
    @return: Whether peer will close the connection
953 02cab3e7 Michael Hanselmann

954 02cab3e7 Michael Hanselmann
    """
955 02cab3e7 Michael Hanselmann
    # RFC2616, section 14.10: "HTTP/1.1 defines the "close" connection option
956 02cab3e7 Michael Hanselmann
    # for the sender to signal that the connection will be closed after
957 02cab3e7 Michael Hanselmann
    # completion of the response. For example,
958 02cab3e7 Michael Hanselmann
    #
959 02cab3e7 Michael Hanselmann
    #        Connection: close
960 02cab3e7 Michael Hanselmann
    #
961 02cab3e7 Michael Hanselmann
    # in either the request or the response header fields indicates that the
962 02cab3e7 Michael Hanselmann
    # connection SHOULD NOT be considered `persistent' (section 8.1) after the
963 02cab3e7 Michael Hanselmann
    # current request/response is complete."
964 02cab3e7 Michael Hanselmann
965 02cab3e7 Michael Hanselmann
    hdr_connection = self.msg.headers.get(HTTP_CONNECTION, None)
966 02cab3e7 Michael Hanselmann
    if hdr_connection:
967 02cab3e7 Michael Hanselmann
      hdr_connection = hdr_connection.lower()
968 02cab3e7 Michael Hanselmann
969 02cab3e7 Michael Hanselmann
    # An HTTP/1.1 server is assumed to stay open unless explicitly closed.
970 02cab3e7 Michael Hanselmann
    if self.msg.start_line.version == HTTP_1_1:
971 02cab3e7 Michael Hanselmann
      return (hdr_connection and "close" in hdr_connection)
972 02cab3e7 Michael Hanselmann
973 02cab3e7 Michael Hanselmann
    # Some HTTP/1.0 implementations have support for persistent connections,
974 02cab3e7 Michael Hanselmann
    # using rules different than HTTP/1.1.
975 02cab3e7 Michael Hanselmann
976 02cab3e7 Michael Hanselmann
    # For older HTTP, Keep-Alive indicates persistent connection.
977 02cab3e7 Michael Hanselmann
    if self.msg.headers.get(HTTP_KEEP_ALIVE):
978 02cab3e7 Michael Hanselmann
      return False
979 02cab3e7 Michael Hanselmann
980 02cab3e7 Michael Hanselmann
    # At least Akamai returns a "Connection: Keep-Alive" header, which was
981 02cab3e7 Michael Hanselmann
    # supposed to be sent by the client.
982 02cab3e7 Michael Hanselmann
    if hdr_connection and "keep-alive" in hdr_connection:
983 02cab3e7 Michael Hanselmann
      return False
984 02cab3e7 Michael Hanselmann
985 02cab3e7 Michael Hanselmann
    return True
986 02cab3e7 Michael Hanselmann
987 02cab3e7 Michael Hanselmann
  def _ParseHeaders(self):
988 02cab3e7 Michael Hanselmann
    """Parses the headers.
989 02cab3e7 Michael Hanselmann

990 02cab3e7 Michael Hanselmann
    This function also adjusts internal variables based on header values.
991 02cab3e7 Michael Hanselmann

992 25e7b43f Iustin Pop
    RFC2616, section 4.3: The presence of a message-body in a request is
993 02cab3e7 Michael Hanselmann
    signaled by the inclusion of a Content-Length or Transfer-Encoding header
994 25e7b43f Iustin Pop
    field in the request's message-headers.
995 02cab3e7 Michael Hanselmann

996 02cab3e7 Michael Hanselmann
    """
997 02cab3e7 Michael Hanselmann
    # Parse headers
998 02cab3e7 Michael Hanselmann
    self.header_buffer.seek(0, 0)
999 02cab3e7 Michael Hanselmann
    self.msg.headers = mimetools.Message(self.header_buffer, 0)
1000 02cab3e7 Michael Hanselmann
1001 02cab3e7 Michael Hanselmann
    self.peer_will_close = self._WillPeerCloseConnection()
1002 02cab3e7 Michael Hanselmann
1003 02cab3e7 Michael Hanselmann
    # Do we have a Content-Length header?
1004 02cab3e7 Michael Hanselmann
    hdr_content_length = self.msg.headers.get(HTTP_CONTENT_LENGTH, None)
1005 02cab3e7 Michael Hanselmann
    if hdr_content_length:
1006 02cab3e7 Michael Hanselmann
      try:
1007 02cab3e7 Michael Hanselmann
        self.content_length = int(hdr_content_length)
1008 02cab3e7 Michael Hanselmann
      except ValueError:
1009 02cab3e7 Michael Hanselmann
        self.content_length = None
1010 02cab3e7 Michael Hanselmann
      if self.content_length is not None and self.content_length < 0:
1011 02cab3e7 Michael Hanselmann
        self.content_length = None
1012 02cab3e7 Michael Hanselmann
1013 02cab3e7 Michael Hanselmann
    # if the connection remains open and a content-length was not provided,
1014 02cab3e7 Michael Hanselmann
    # then assume that the connection WILL close.
1015 02cab3e7 Michael Hanselmann
    if self.content_length is None:
1016 02cab3e7 Michael Hanselmann
      self.peer_will_close = True