blob: a4ec8e53f8379e287304d4bb46be34048f1984c6 [file] [log] [blame]
Greg Stein5e0fa402000-06-26 08:28:01 +00001"""HTTP/1.1 client library
Guido van Rossum41999c11997-12-09 00:12:23 +00002
Greg Stein5e0fa402000-06-26 08:28:01 +00003<intro stuff goes here>
4<other stuff, too>
Guido van Rossum41999c11997-12-09 00:12:23 +00005
Thomas Wouters0e3f5912006-08-11 14:57:12 +00006HTTPConnection goes through a number of "states", which define when a client
Greg Stein5e0fa402000-06-26 08:28:01 +00007may legally make another request or fetch the response for a particular
8request. This diagram details these state transitions:
Guido van Rossum41999c11997-12-09 00:12:23 +00009
Greg Stein5e0fa402000-06-26 08:28:01 +000010 (null)
11 |
12 | HTTPConnection()
13 v
14 Idle
15 |
16 | putrequest()
17 v
18 Request-started
19 |
20 | ( putheader() )* endheaders()
21 v
22 Request-sent
23 |
24 | response = getresponse()
25 v
26 Unread-response [Response-headers-read]
27 |\____________________
Tim Peters5ceadc82001-01-13 19:16:21 +000028 | |
29 | response.read() | putrequest()
30 v v
31 Idle Req-started-unread-response
32 ______/|
33 / |
34 response.read() | | ( putheader() )* endheaders()
35 v v
36 Request-started Req-sent-unread-response
37 |
38 | response.read()
39 v
40 Request-sent
Greg Stein5e0fa402000-06-26 08:28:01 +000041
42This diagram presents the following rules:
43 -- a second request may not be started until {response-headers-read}
44 -- a response [object] cannot be retrieved until {request-sent}
45 -- there is no differentiation between an unread response body and a
46 partially read response body
47
48Note: this enforcement is applied by the HTTPConnection class. The
49 HTTPResponse class does not enforce this state machine, which
50 implies sophisticated clients may accelerate the request/response
51 pipeline. Caution should be taken, though: accelerating the states
52 beyond the above pattern may imply knowledge of the server's
53 connection-close behavior for certain requests. For example, it
54 is impossible to tell whether the server will close the connection
55 UNTIL the response headers have been read; this means that further
56 requests cannot be placed into the pipeline until it is known that
57 the server will NOT be closing the connection.
58
59Logical State __state __response
60------------- ------- ----------
61Idle _CS_IDLE None
62Request-started _CS_REQ_STARTED None
63Request-sent _CS_REQ_SENT None
64Unread-response _CS_IDLE <response_class>
65Req-started-unread-response _CS_REQ_STARTED <response_class>
66Req-sent-unread-response _CS_REQ_SENT <response_class>
Guido van Rossum41999c11997-12-09 00:12:23 +000067"""
Guido van Rossum23acc951994-02-21 16:36:04 +000068
Jeremy Hyltonfb35f652007-08-03 20:30:33 +000069import io
Jeremy Hylton6459c8d2001-10-11 17:47:22 +000070import socket
Barry Warsaw820c1202008-06-12 04:06:45 +000071import email.parser
72import email.message
Jeremy Hylton1afc1692008-06-18 20:49:58 +000073from urllib.parse import urlsplit
Thomas Wouters89d996e2007-09-08 17:39:28 +000074import warnings
Guido van Rossum23acc951994-02-21 16:36:04 +000075
Thomas Wouters47b49bf2007-08-30 22:15:33 +000076__all__ = ["HTTPResponse", "HTTPConnection",
Skip Montanaro951a8842001-06-01 16:25:38 +000077 "HTTPException", "NotConnected", "UnknownProtocol",
Jeremy Hylton7c75c992002-06-28 23:38:14 +000078 "UnknownTransferEncoding", "UnimplementedFileMode",
79 "IncompleteRead", "InvalidURL", "ImproperConnectionState",
80 "CannotSendRequest", "CannotSendHeader", "ResponseNotReady",
Georg Brandl6aab16e2006-02-17 19:17:25 +000081 "BadStatusLine", "error", "responses"]
Skip Montanaro2dd42762001-01-23 15:35:05 +000082
Guido van Rossum23acc951994-02-21 16:36:04 +000083HTTP_PORT = 80
Guido van Rossum09c8b6c1999-12-07 21:37:17 +000084HTTPS_PORT = 443
85
Greg Stein5e0fa402000-06-26 08:28:01 +000086_UNKNOWN = 'UNKNOWN'
87
88# connection states
89_CS_IDLE = 'Idle'
90_CS_REQ_STARTED = 'Request-started'
91_CS_REQ_SENT = 'Request-sent'
92
Martin v. Löwis39a31782004-09-18 09:03:49 +000093# status codes
94# informational
95CONTINUE = 100
96SWITCHING_PROTOCOLS = 101
97PROCESSING = 102
98
99# successful
100OK = 200
101CREATED = 201
102ACCEPTED = 202
103NON_AUTHORITATIVE_INFORMATION = 203
104NO_CONTENT = 204
105RESET_CONTENT = 205
106PARTIAL_CONTENT = 206
107MULTI_STATUS = 207
108IM_USED = 226
109
110# redirection
111MULTIPLE_CHOICES = 300
112MOVED_PERMANENTLY = 301
113FOUND = 302
114SEE_OTHER = 303
115NOT_MODIFIED = 304
116USE_PROXY = 305
117TEMPORARY_REDIRECT = 307
118
119# client error
120BAD_REQUEST = 400
121UNAUTHORIZED = 401
122PAYMENT_REQUIRED = 402
123FORBIDDEN = 403
124NOT_FOUND = 404
125METHOD_NOT_ALLOWED = 405
126NOT_ACCEPTABLE = 406
127PROXY_AUTHENTICATION_REQUIRED = 407
128REQUEST_TIMEOUT = 408
129CONFLICT = 409
130GONE = 410
131LENGTH_REQUIRED = 411
132PRECONDITION_FAILED = 412
133REQUEST_ENTITY_TOO_LARGE = 413
134REQUEST_URI_TOO_LONG = 414
135UNSUPPORTED_MEDIA_TYPE = 415
136REQUESTED_RANGE_NOT_SATISFIABLE = 416
137EXPECTATION_FAILED = 417
138UNPROCESSABLE_ENTITY = 422
139LOCKED = 423
140FAILED_DEPENDENCY = 424
141UPGRADE_REQUIRED = 426
142
143# server error
144INTERNAL_SERVER_ERROR = 500
145NOT_IMPLEMENTED = 501
146BAD_GATEWAY = 502
147SERVICE_UNAVAILABLE = 503
148GATEWAY_TIMEOUT = 504
149HTTP_VERSION_NOT_SUPPORTED = 505
150INSUFFICIENT_STORAGE = 507
151NOT_EXTENDED = 510
152
Georg Brandl6aab16e2006-02-17 19:17:25 +0000153# Mapping status codes to official W3C names
154responses = {
155 100: 'Continue',
156 101: 'Switching Protocols',
157
158 200: 'OK',
159 201: 'Created',
160 202: 'Accepted',
161 203: 'Non-Authoritative Information',
162 204: 'No Content',
163 205: 'Reset Content',
164 206: 'Partial Content',
165
166 300: 'Multiple Choices',
167 301: 'Moved Permanently',
168 302: 'Found',
169 303: 'See Other',
170 304: 'Not Modified',
171 305: 'Use Proxy',
172 306: '(Unused)',
173 307: 'Temporary Redirect',
174
175 400: 'Bad Request',
176 401: 'Unauthorized',
177 402: 'Payment Required',
178 403: 'Forbidden',
179 404: 'Not Found',
180 405: 'Method Not Allowed',
181 406: 'Not Acceptable',
182 407: 'Proxy Authentication Required',
183 408: 'Request Timeout',
184 409: 'Conflict',
185 410: 'Gone',
186 411: 'Length Required',
187 412: 'Precondition Failed',
188 413: 'Request Entity Too Large',
189 414: 'Request-URI Too Long',
190 415: 'Unsupported Media Type',
191 416: 'Requested Range Not Satisfiable',
192 417: 'Expectation Failed',
193
194 500: 'Internal Server Error',
195 501: 'Not Implemented',
196 502: 'Bad Gateway',
197 503: 'Service Unavailable',
198 504: 'Gateway Timeout',
199 505: 'HTTP Version Not Supported',
200}
201
Georg Brandl80ba8e82005-09-29 20:16:07 +0000202# maximal amount of data to read at one time in _safe_read
203MAXAMOUNT = 1048576
204
Barry Warsaw820c1202008-06-12 04:06:45 +0000205class HTTPMessage(email.message.Message):
206 def getallmatchingheaders(self, name):
207 """Find all header lines matching a given header name.
Jeremy Hylton6d0a4c72002-07-07 16:51:37 +0000208
Barry Warsaw820c1202008-06-12 04:06:45 +0000209 Look through the list of headers and find all lines matching a given
210 header name (and their continuation lines). A list of the lines is
211 returned, without interpretation. If the header does not occur, an
212 empty list is returned. If the header occurs multiple times, all
213 occurrences are returned. Case is not important in the header name.
Jeremy Hylton6d0a4c72002-07-07 16:51:37 +0000214
Jeremy Hylton6d0a4c72002-07-07 16:51:37 +0000215 """
Barry Warsaw820c1202008-06-12 04:06:45 +0000216 # XXX: copied from rfc822.Message for compatibility
217 name = name.lower() + ':'
218 n = len(name)
219 lst = []
220 hit = 0
221 for line in self.keys():
222 if line[:n].lower() == name:
223 hit = 1
224 elif not line[:1].isspace():
225 hit = 0
226 if hit:
227 lst.append(line)
228 return lst
Jeremy Hylton6d0a4c72002-07-07 16:51:37 +0000229
Barry Warsaw820c1202008-06-12 04:06:45 +0000230def parse_headers(fp):
231 """Parses only RFC2822 headers from a file pointer.
232
233 email Parser wants to see strings rather than bytes.
234 But a TextIOWrapper around self.rfile would buffer too many bytes
235 from the stream, bytes which we later need to read as bytes.
236 So we read the correct bytes here, as bytes, for email Parser
237 to parse.
238
239 """
240 # XXX: Copied from http.server.BaseHTTPRequestHandler.parse_request,
241 # maybe we can just call this function from there.
242 headers = []
243 while True:
244 line = fp.readline()
245 headers.append(line)
246 if line in (b'\r\n', b'\n', b''):
247 break
248 hstring = b''.join(headers).decode('iso-8859-1')
249
250 return email.parser.Parser(_class=HTTPMessage).parsestr(hstring)
Greg Stein5e0fa402000-06-26 08:28:01 +0000251
Jeremy Hylton97043c32007-08-04 02:34:24 +0000252class HTTPResponse:
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000253
254 # strict: If true, raise BadStatusLine if the status line can't be
255 # parsed as a valid HTTP/1.0 or 1.1 status line. By default it is
Skip Montanaro186bec22002-07-25 16:10:38 +0000256 # false because it prevents clients from talking to HTTP/0.9
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000257 # servers. Note that a response with a sufficiently corrupted
258 # status line will look like an HTTP/0.9 response.
259
260 # See RFC 2616 sec 19.6 and RFC 1945 sec 6 for details.
261
Jeremy Hylton811fc142007-08-03 13:30:02 +0000262 # The bytes from the socket object are iso-8859-1 strings.
263 # See RFC 2616 sec 2.2 which notes an exception for MIME-encoded
264 # text following RFC 2047. The basic status line parsing only
265 # accepts iso-8859-1.
266
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000267 def __init__(self, sock, debuglevel=0, strict=0, method=None):
Jeremy Hylton39b198d2007-08-04 19:22:00 +0000268 # XXX If the response includes a content-length header, we
269 # need to make sure that the client doesn't read more than the
270 # specified number of bytes. If it does, it will block until
271 # the server times out and closes the connection. (The only
272 # applies to HTTP/1.1 connections.) Since some clients access
273 # self.fp directly rather than calling read(), this is a little
274 # tricky.
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000275 self.fp = sock.makefile("rb", 0)
Jeremy Hylton30f86742000-09-18 22:50:38 +0000276 self.debuglevel = debuglevel
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000277 self.strict = strict
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000278 self._method = method
Greg Stein5e0fa402000-06-26 08:28:01 +0000279
Greg Steindd6eefb2000-07-18 09:09:48 +0000280 self.msg = None
Greg Stein5e0fa402000-06-26 08:28:01 +0000281
Greg Steindd6eefb2000-07-18 09:09:48 +0000282 # from the Status-Line of the response
Tim Peters07e99cb2001-01-14 23:47:14 +0000283 self.version = _UNKNOWN # HTTP-Version
284 self.status = _UNKNOWN # Status-Code
285 self.reason = _UNKNOWN # Reason-Phrase
Greg Stein5e0fa402000-06-26 08:28:01 +0000286
Tim Peters07e99cb2001-01-14 23:47:14 +0000287 self.chunked = _UNKNOWN # is "chunked" being used?
288 self.chunk_left = _UNKNOWN # bytes left to read in current chunk
289 self.length = _UNKNOWN # number of bytes left in response
290 self.will_close = _UNKNOWN # conn will close at end of response
Greg Stein5e0fa402000-06-26 08:28:01 +0000291
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000292 def _read_status(self):
Jeremy Hylton04319c72007-08-04 03:41:19 +0000293 # Initialize with Simple-Response defaults.
Jeremy Hylton811fc142007-08-03 13:30:02 +0000294 line = str(self.fp.readline(), "iso-8859-1")
Jeremy Hylton30f86742000-09-18 22:50:38 +0000295 if self.debuglevel > 0:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000296 print("reply:", repr(line))
Jeremy Hyltonb6769522003-06-29 17:55:05 +0000297 if not line:
298 # Presumably, the server closed the connection before
299 # sending a valid response.
300 raise BadStatusLine(line)
Greg Steindd6eefb2000-07-18 09:09:48 +0000301 try:
Guido van Rossum34735a62000-12-15 15:09:42 +0000302 [version, status, reason] = line.split(None, 2)
Greg Steindd6eefb2000-07-18 09:09:48 +0000303 except ValueError:
304 try:
Guido van Rossum34735a62000-12-15 15:09:42 +0000305 [version, status] = line.split(None, 1)
Greg Steindd6eefb2000-07-18 09:09:48 +0000306 reason = ""
307 except ValueError:
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000308 # empty version will cause next test to fail and status
309 # will be treated as 0.9 response.
310 version = ""
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000311 if not version.startswith("HTTP/"):
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000312 if self.strict:
313 self.close()
314 raise BadStatusLine(line)
315 else:
Jeremy Hylton04319c72007-08-04 03:41:19 +0000316 # Assume it's a Simple-Response from an 0.9 server.
317 # We have to convert the first line back to raw bytes
318 # because self.fp.readline() needs to return bytes.
Guido van Rossum70d0dda2007-08-29 01:53:26 +0000319 self.fp = LineAndFileWrapper(bytes(line, "ascii"), self.fp)
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000320 return "HTTP/0.9", 200, ""
Greg Stein5e0fa402000-06-26 08:28:01 +0000321
Jeremy Hylton23d40472001-04-13 14:57:08 +0000322 # The status code is a three-digit number
323 try:
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000324 status = int(status)
Jeremy Hylton23d40472001-04-13 14:57:08 +0000325 if status < 100 or status > 999:
326 raise BadStatusLine(line)
327 except ValueError:
328 raise BadStatusLine(line)
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000329 return version, status, reason
Greg Stein5e0fa402000-06-26 08:28:01 +0000330
Jeremy Hylton39c03802002-07-12 14:04:09 +0000331 def begin(self):
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000332 if self.msg is not None:
333 # we've already started reading the response
334 return
335
336 # read until we get a non-100 response
Raymond Hettingerb2e0b922003-02-26 22:45:18 +0000337 while True:
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000338 version, status, reason = self._read_status()
Martin v. Löwis39a31782004-09-18 09:03:49 +0000339 if status != CONTINUE:
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000340 break
341 # skip the header from the 100 response
Raymond Hettingerb2e0b922003-02-26 22:45:18 +0000342 while True:
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000343 skip = self.fp.readline().strip()
344 if not skip:
345 break
346 if self.debuglevel > 0:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000347 print("header:", skip)
Tim Petersc411dba2002-07-16 21:35:23 +0000348
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000349 self.status = status
350 self.reason = reason.strip()
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000351 if version == "HTTP/1.0":
Greg Steindd6eefb2000-07-18 09:09:48 +0000352 self.version = 10
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000353 elif version.startswith("HTTP/1."):
Tim Peters07e99cb2001-01-14 23:47:14 +0000354 self.version = 11 # use HTTP/1.1 code for HTTP/1.x where x>=1
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000355 elif version == "HTTP/0.9":
Jeremy Hylton110941a2000-10-12 19:58:36 +0000356 self.version = 9
Greg Steindd6eefb2000-07-18 09:09:48 +0000357 else:
358 raise UnknownProtocol(version)
Greg Stein5e0fa402000-06-26 08:28:01 +0000359
Jeremy Hylton110941a2000-10-12 19:58:36 +0000360 if self.version == 9:
Georg Brandl0aade9a2005-06-26 22:06:54 +0000361 self.length = None
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000362 self.chunked = 0
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000363 self.will_close = 1
Barry Warsaw820c1202008-06-12 04:06:45 +0000364 self.msg = email.message_from_string('')
Jeremy Hylton110941a2000-10-12 19:58:36 +0000365 return
366
Barry Warsaw820c1202008-06-12 04:06:45 +0000367 self.msg = parse_headers(self.fp)
368
Jeremy Hylton30f86742000-09-18 22:50:38 +0000369 if self.debuglevel > 0:
Barry Warsaw820c1202008-06-12 04:06:45 +0000370 for hdr in self.msg:
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000371 print("header:", hdr, end=" ")
Greg Stein5e0fa402000-06-26 08:28:01 +0000372
Greg Steindd6eefb2000-07-18 09:09:48 +0000373 # are we using the chunked-style of transfer encoding?
Barry Warsaw820c1202008-06-12 04:06:45 +0000374 tr_enc = self.msg.get("transfer-encoding")
Jeremy Hyltond229b3a2002-09-03 19:24:24 +0000375 if tr_enc and tr_enc.lower() == "chunked":
Greg Steindd6eefb2000-07-18 09:09:48 +0000376 self.chunked = 1
377 self.chunk_left = None
378 else:
379 self.chunked = 0
Greg Stein5e0fa402000-06-26 08:28:01 +0000380
Greg Steindd6eefb2000-07-18 09:09:48 +0000381 # will the connection close at the end of the response?
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000382 self.will_close = self._check_close()
Greg Stein5e0fa402000-06-26 08:28:01 +0000383
Greg Steindd6eefb2000-07-18 09:09:48 +0000384 # do we have a Content-Length?
385 # NOTE: RFC 2616, S4.4, #3 says we ignore this if tr_enc is "chunked"
Jeremy Hylton0ee5eeb2007-08-04 03:25:17 +0000386 self.length = None
Barry Warsaw820c1202008-06-12 04:06:45 +0000387 length = self.msg.get("content-length")
388
389 # are we using the chunked-style of transfer encoding?
390 tr_enc = self.msg.get("transfer-encoding")
Greg Steindd6eefb2000-07-18 09:09:48 +0000391 if length and not self.chunked:
Jeremy Hylton30a81812000-09-14 20:34:27 +0000392 try:
393 self.length = int(length)
394 except ValueError:
Christian Heimesa612dc02008-02-24 13:08:18 +0000395 self.length = None
396 else:
397 if self.length < 0: # ignore nonsensical negative lengths
398 self.length = None
399 else:
400 self.length = None
Greg Stein5e0fa402000-06-26 08:28:01 +0000401
Greg Steindd6eefb2000-07-18 09:09:48 +0000402 # does the body have a fixed length? (of zero)
Martin v. Löwis39a31782004-09-18 09:03:49 +0000403 if (status == NO_CONTENT or status == NOT_MODIFIED or
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000404 100 <= status < 200 or # 1xx codes
Jeremy Hyltone5d0e842007-08-03 13:45:24 +0000405 self._method == "HEAD"):
Greg Steindd6eefb2000-07-18 09:09:48 +0000406 self.length = 0
Greg Stein5e0fa402000-06-26 08:28:01 +0000407
Greg Steindd6eefb2000-07-18 09:09:48 +0000408 # if the connection remains open, and we aren't using chunked, and
409 # a content-length was not provided, then assume that the connection
410 # WILL close.
Jeremy Hylton0ee5eeb2007-08-04 03:25:17 +0000411 if (not self.will_close and
412 not self.chunked and
413 self.length is None):
Greg Steindd6eefb2000-07-18 09:09:48 +0000414 self.will_close = 1
Greg Stein5e0fa402000-06-26 08:28:01 +0000415
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000416 def _check_close(self):
Barry Warsaw820c1202008-06-12 04:06:45 +0000417 conn = self.msg.get("connection")
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000418 if self.version == 11:
419 # An HTTP/1.1 proxy is assumed to stay open unless
420 # explicitly closed.
Barry Warsaw820c1202008-06-12 04:06:45 +0000421 conn = self.msg.get("connection")
Raymond Hettingerbac788a2004-05-04 09:21:43 +0000422 if conn and "close" in conn.lower():
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000423 return True
424 return False
425
Jeremy Hylton2c178252004-08-07 16:28:14 +0000426 # Some HTTP/1.0 implementations have support for persistent
427 # connections, using rules different than HTTP/1.1.
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000428
Christian Heimes895627f2007-12-08 17:28:33 +0000429 # For older HTTP, Keep-Alive indicates persistent connection.
Barry Warsaw820c1202008-06-12 04:06:45 +0000430 if self.msg.get("keep-alive"):
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000431 return False
Tim Peters77c06fb2002-11-24 02:35:35 +0000432
Jeremy Hylton2c178252004-08-07 16:28:14 +0000433 # At least Akamai returns a "Connection: Keep-Alive" header,
434 # which was supposed to be sent by the client.
435 if conn and "keep-alive" in conn.lower():
436 return False
437
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000438 # Proxy-Connection is a netscape hack.
Barry Warsaw820c1202008-06-12 04:06:45 +0000439 pconn = self.msg.get("proxy-connection")
Raymond Hettingerbac788a2004-05-04 09:21:43 +0000440 if pconn and "keep-alive" in pconn.lower():
Jeremy Hylton22b3a492002-11-13 17:27:43 +0000441 return False
442
443 # otherwise, assume it will close
444 return True
445
Greg Steindd6eefb2000-07-18 09:09:48 +0000446 def close(self):
447 if self.fp:
448 self.fp.close()
449 self.fp = None
Greg Stein5e0fa402000-06-26 08:28:01 +0000450
Jeremy Hyltondf5f6b52007-08-08 17:36:33 +0000451 # These implementations are for the benefit of io.BufferedReader.
452
453 # XXX This class should probably be revised to act more like
454 # the "raw stream" that BufferedReader expects.
455
456 @property
457 def closed(self):
458 return self.isclosed()
459
460 def flush(self):
461 self.fp.flush()
462
463 # End of "raw stream" methods
464
Greg Steindd6eefb2000-07-18 09:09:48 +0000465 def isclosed(self):
466 # NOTE: it is possible that we will not ever call self.close(). This
467 # case occurs when will_close is TRUE, length is None, and we
468 # read up to the last byte, but NOT past it.
469 #
470 # IMPLIES: if will_close is FALSE, then self.close() will ALWAYS be
471 # called, meaning self.isclosed() is meaningful.
472 return self.fp is None
473
Jeremy Hylton2c178252004-08-07 16:28:14 +0000474 # XXX It would be nice to have readline and __iter__ for this, too.
475
Greg Steindd6eefb2000-07-18 09:09:48 +0000476 def read(self, amt=None):
477 if self.fp is None:
Guido van Rossum8ce8a782007-11-01 19:42:39 +0000478 return b""
Greg Steindd6eefb2000-07-18 09:09:48 +0000479
480 if self.chunked:
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000481 return self._read_chunked(amt)
Tim Peters230a60c2002-11-09 05:08:07 +0000482
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000483 if amt is None:
Greg Steindd6eefb2000-07-18 09:09:48 +0000484 # unbounded read
Jeremy Hyltondef9d2a2004-11-07 16:13:49 +0000485 if self.length is None:
Greg Steindd6eefb2000-07-18 09:09:48 +0000486 s = self.fp.read()
487 else:
488 s = self._safe_read(self.length)
Jeremy Hyltondef9d2a2004-11-07 16:13:49 +0000489 self.length = 0
Tim Peters07e99cb2001-01-14 23:47:14 +0000490 self.close() # we read everything
Greg Steindd6eefb2000-07-18 09:09:48 +0000491 return s
492
493 if self.length is not None:
494 if amt > self.length:
495 # clip the read to the "end of response"
496 amt = self.length
Greg Steindd6eefb2000-07-18 09:09:48 +0000497
498 # we do not use _safe_read() here because this may be a .will_close
499 # connection, and the user is reading more bytes than will be provided
500 # (for example, reading in 1k chunks)
501 s = self.fp.read(amt)
Jeremy Hyltondef9d2a2004-11-07 16:13:49 +0000502 if self.length is not None:
503 self.length -= len(s)
Guido van Rossum8ce8a782007-11-01 19:42:39 +0000504 if not self.length:
505 self.close()
Greg Steindd6eefb2000-07-18 09:09:48 +0000506 return s
507
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000508 def _read_chunked(self, amt):
509 assert self.chunked != _UNKNOWN
510 chunk_left = self.chunk_left
Georg Brandl95ba4692008-01-26 09:45:58 +0000511 value = b""
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000512
513 # XXX This accumulates chunks by repeated string concatenation,
514 # which is not efficient as the number or size of chunks gets big.
Raymond Hettingerb2e0b922003-02-26 22:45:18 +0000515 while True:
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000516 if chunk_left is None:
517 line = self.fp.readline()
Georg Brandl95ba4692008-01-26 09:45:58 +0000518 i = line.find(b";")
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000519 if i >= 0:
520 line = line[:i] # strip chunk-extensions
Christian Heimesa612dc02008-02-24 13:08:18 +0000521 try:
522 chunk_left = int(line, 16)
523 except ValueError:
524 # close the connection as protocol synchronisation is
525 # probably lost
526 self.close()
527 raise IncompleteRead(value)
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000528 if chunk_left == 0:
529 break
530 if amt is None:
531 value += self._safe_read(chunk_left)
532 elif amt < chunk_left:
533 value += self._safe_read(amt)
534 self.chunk_left = chunk_left - amt
535 return value
536 elif amt == chunk_left:
537 value += self._safe_read(amt)
538 self._safe_read(2) # toss the CRLF at the end of the chunk
539 self.chunk_left = None
540 return value
541 else:
542 value += self._safe_read(chunk_left)
543 amt -= chunk_left
544
545 # we read the whole chunk, get another
546 self._safe_read(2) # toss the CRLF at the end of the chunk
547 chunk_left = None
548
549 # read and discard trailer up to the CRLF terminator
550 ### note: we shouldn't have any trailers!
Raymond Hettingerb2e0b922003-02-26 22:45:18 +0000551 while True:
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000552 line = self.fp.readline()
Christian Heimes0bd4e112008-02-12 22:59:25 +0000553 if not line:
554 # a vanishingly small number of sites EOF without
555 # sending the trailer
556 break
Guido van Rossum8ce8a782007-11-01 19:42:39 +0000557 if line == b"\r\n":
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000558 break
559
560 # we read everything; close the "file"
Jeremy Hyltond4c472c2002-09-03 20:49:06 +0000561 self.close()
562
563 return value
Tim Peters230a60c2002-11-09 05:08:07 +0000564
Greg Steindd6eefb2000-07-18 09:09:48 +0000565 def _safe_read(self, amt):
566 """Read the number of bytes requested, compensating for partial reads.
567
568 Normally, we have a blocking socket, but a read() can be interrupted
569 by a signal (resulting in a partial read).
570
571 Note that we cannot distinguish between EOF and an interrupt when zero
572 bytes have been read. IncompleteRead() will be raised in this
573 situation.
574
575 This function should be used when <amt> bytes "should" be present for
576 reading. If the bytes are truly not available (due to EOF), then the
577 IncompleteRead exception can be used to detect the problem.
578 """
Georg Brandl80ba8e82005-09-29 20:16:07 +0000579 s = []
Greg Steindd6eefb2000-07-18 09:09:48 +0000580 while amt > 0:
Georg Brandl80ba8e82005-09-29 20:16:07 +0000581 chunk = self.fp.read(min(amt, MAXAMOUNT))
Greg Steindd6eefb2000-07-18 09:09:48 +0000582 if not chunk:
583 raise IncompleteRead(s)
Georg Brandl80ba8e82005-09-29 20:16:07 +0000584 s.append(chunk)
Raymond Hettingerb2e0b922003-02-26 22:45:18 +0000585 amt -= len(chunk)
Guido van Rossuma00f1232007-09-12 19:43:09 +0000586 return b"".join(s)
Greg Steindd6eefb2000-07-18 09:09:48 +0000587
588 def getheader(self, name, default=None):
589 if self.msg is None:
590 raise ResponseNotReady()
Barry Warsaw820c1202008-06-12 04:06:45 +0000591 return ', '.join(self.msg.get_all(name, default))
Greg Stein5e0fa402000-06-26 08:28:01 +0000592
Martin v. Löwisdeacce22004-08-18 12:46:26 +0000593 def getheaders(self):
594 """Return list of (header, value) tuples."""
595 if self.msg is None:
596 raise ResponseNotReady()
Guido van Rossumcc2b0162007-02-11 06:12:03 +0000597 return list(self.msg.items())
Martin v. Löwisdeacce22004-08-18 12:46:26 +0000598
Greg Stein5e0fa402000-06-26 08:28:01 +0000599
600class HTTPConnection:
601
Greg Steindd6eefb2000-07-18 09:09:48 +0000602 _http_vsn = 11
603 _http_vsn_str = 'HTTP/1.1'
Greg Stein5e0fa402000-06-26 08:28:01 +0000604
Greg Steindd6eefb2000-07-18 09:09:48 +0000605 response_class = HTTPResponse
606 default_port = HTTP_PORT
607 auto_open = 1
Jeremy Hylton30f86742000-09-18 22:50:38 +0000608 debuglevel = 0
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000609 strict = 0
Greg Stein5e0fa402000-06-26 08:28:01 +0000610
Georg Brandlf78e02b2008-06-10 17:40:04 +0000611 def __init__(self, host, port=None, strict=None,
612 timeout=socket._GLOBAL_DEFAULT_TIMEOUT):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000613 self.timeout = timeout
Greg Steindd6eefb2000-07-18 09:09:48 +0000614 self.sock = None
Jeremy Hylton8531b1b2002-07-16 21:21:11 +0000615 self._buffer = []
Greg Steindd6eefb2000-07-18 09:09:48 +0000616 self.__response = None
617 self.__state = _CS_IDLE
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000618 self._method = None
Tim Petersc411dba2002-07-16 21:35:23 +0000619
Greg Steindd6eefb2000-07-18 09:09:48 +0000620 self._set_hostport(host, port)
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000621 if strict is not None:
622 self.strict = strict
Greg Stein5e0fa402000-06-26 08:28:01 +0000623
Greg Steindd6eefb2000-07-18 09:09:48 +0000624 def _set_hostport(self, host, port):
625 if port is None:
Skip Montanaro10e6e0e2004-09-14 16:32:02 +0000626 i = host.rfind(':')
Skip Montanarocae14d22004-09-14 17:55:21 +0000627 j = host.rfind(']') # ipv6 addresses have [...]
628 if i > j:
Skip Montanaro9d389972002-03-24 16:53:50 +0000629 try:
630 port = int(host[i+1:])
631 except ValueError:
Jeremy Hyltonfbd79942002-07-02 20:19:08 +0000632 raise InvalidURL("nonnumeric port: '%s'" % host[i+1:])
Greg Steindd6eefb2000-07-18 09:09:48 +0000633 host = host[:i]
634 else:
635 port = self.default_port
Raymond Hettinger4d037912004-10-14 15:23:38 +0000636 if host and host[0] == '[' and host[-1] == ']':
Brett Cannon0a1af4a2004-09-15 23:26:23 +0000637 host = host[1:-1]
Greg Steindd6eefb2000-07-18 09:09:48 +0000638 self.host = host
639 self.port = port
Greg Stein5e0fa402000-06-26 08:28:01 +0000640
Jeremy Hylton30f86742000-09-18 22:50:38 +0000641 def set_debuglevel(self, level):
642 self.debuglevel = level
643
Greg Steindd6eefb2000-07-18 09:09:48 +0000644 def connect(self):
645 """Connect to the host and port specified in __init__."""
Guido van Rossumd8faa362007-04-27 19:54:29 +0000646 self.sock = socket.create_connection((self.host,self.port),
647 self.timeout)
Greg Stein5e0fa402000-06-26 08:28:01 +0000648
Greg Steindd6eefb2000-07-18 09:09:48 +0000649 def close(self):
650 """Close the connection to the HTTP server."""
651 if self.sock:
Tim Peters07e99cb2001-01-14 23:47:14 +0000652 self.sock.close() # close it manually... there may be other refs
Greg Steindd6eefb2000-07-18 09:09:48 +0000653 self.sock = None
654 if self.__response:
655 self.__response.close()
656 self.__response = None
657 self.__state = _CS_IDLE
Greg Stein5e0fa402000-06-26 08:28:01 +0000658
Greg Steindd6eefb2000-07-18 09:09:48 +0000659 def send(self, str):
660 """Send `str' to the server."""
661 if self.sock is None:
662 if self.auto_open:
663 self.connect()
664 else:
665 raise NotConnected()
Greg Stein5e0fa402000-06-26 08:28:01 +0000666
Jeremy Hyltone3252ec2002-07-16 21:41:43 +0000667 # send the data to the server. if we get a broken pipe, then close
Greg Steindd6eefb2000-07-18 09:09:48 +0000668 # the socket. we want to reconnect when somebody tries to send again.
669 #
670 # NOTE: we DO propagate the error, though, because we cannot simply
671 # ignore the error... the caller will know if they can retry.
Jeremy Hylton30f86742000-09-18 22:50:38 +0000672 if self.debuglevel > 0:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000673 print("send:", repr(str))
Greg Steindd6eefb2000-07-18 09:09:48 +0000674 try:
Thomas Wouters89f507f2006-12-13 04:49:30 +0000675 blocksize=8192
676 if hasattr(str,'read') :
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000677 if self.debuglevel > 0: print("sendIng a read()able")
Thomas Wouters89f507f2006-12-13 04:49:30 +0000678 data=str.read(blocksize)
679 while data:
680 self.sock.sendall(data)
681 data=str.read(blocksize)
682 else:
683 self.sock.sendall(str)
Guido van Rossumb940e112007-01-10 16:19:56 +0000684 except socket.error as v:
Guido van Rossum89df2452007-03-19 22:26:27 +0000685 if v.args[0] == 32: # Broken pipe
Greg Steindd6eefb2000-07-18 09:09:48 +0000686 self.close()
687 raise
Greg Stein5e0fa402000-06-26 08:28:01 +0000688
Jeremy Hylton8531b1b2002-07-16 21:21:11 +0000689 def _output(self, s):
690 """Add a line of output to the current request buffer.
Tim Peters469cdad2002-08-08 20:19:19 +0000691
Jeremy Hyltone3252ec2002-07-16 21:41:43 +0000692 Assumes that the line does *not* end with \\r\\n.
Jeremy Hylton8531b1b2002-07-16 21:21:11 +0000693 """
694 self._buffer.append(s)
695
696 def _send_output(self):
697 """Send the currently buffered request and clear the buffer.
698
Jeremy Hyltone3252ec2002-07-16 21:41:43 +0000699 Appends an extra \\r\\n to the buffer.
Jeremy Hylton8531b1b2002-07-16 21:21:11 +0000700 """
Martin v. Löwisdd5a8602007-06-30 09:22:09 +0000701 self._buffer.extend((b"", b""))
702 msg = b"\r\n".join(self._buffer)
Jeremy Hylton8531b1b2002-07-16 21:21:11 +0000703 del self._buffer[:]
704 self.send(msg)
705
Martin v. Löwisaf7dc8d2003-11-19 19:51:55 +0000706 def putrequest(self, method, url, skip_host=0, skip_accept_encoding=0):
Greg Steindd6eefb2000-07-18 09:09:48 +0000707 """Send a request to the server.
Greg Stein5e0fa402000-06-26 08:28:01 +0000708
Greg Steindd6eefb2000-07-18 09:09:48 +0000709 `method' specifies an HTTP request method, e.g. 'GET'.
710 `url' specifies the object being requested, e.g. '/index.html'.
Martin v. Löwisaf7dc8d2003-11-19 19:51:55 +0000711 `skip_host' if True does not add automatically a 'Host:' header
712 `skip_accept_encoding' if True does not add automatically an
713 'Accept-Encoding:' header
Greg Steindd6eefb2000-07-18 09:09:48 +0000714 """
Greg Stein5e0fa402000-06-26 08:28:01 +0000715
Greg Stein616a58d2003-06-24 06:35:19 +0000716 # if a prior response has been completed, then forget about it.
Greg Steindd6eefb2000-07-18 09:09:48 +0000717 if self.__response and self.__response.isclosed():
718 self.__response = None
Greg Stein5e0fa402000-06-26 08:28:01 +0000719
Tim Peters58eb11c2004-01-18 20:29:55 +0000720
Greg Steindd6eefb2000-07-18 09:09:48 +0000721 # in certain cases, we cannot issue another request on this connection.
722 # this occurs when:
723 # 1) we are in the process of sending a request. (_CS_REQ_STARTED)
724 # 2) a response to a previous request has signalled that it is going
725 # to close the connection upon completion.
726 # 3) the headers for the previous response have not been read, thus
727 # we cannot determine whether point (2) is true. (_CS_REQ_SENT)
728 #
729 # if there is no prior response, then we can request at will.
730 #
731 # if point (2) is true, then we will have passed the socket to the
732 # response (effectively meaning, "there is no prior response"), and
733 # will open a new one when a new request is made.
734 #
735 # Note: if a prior response exists, then we *can* start a new request.
736 # We are not allowed to begin fetching the response to this new
737 # request, however, until that prior response is complete.
738 #
739 if self.__state == _CS_IDLE:
740 self.__state = _CS_REQ_STARTED
741 else:
742 raise CannotSendRequest()
Greg Stein5e0fa402000-06-26 08:28:01 +0000743
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000744 # Save the method we use, we need it later in the response phase
745 self._method = method
Greg Steindd6eefb2000-07-18 09:09:48 +0000746 if not url:
747 url = '/'
Martin v. Löwisdd5a8602007-06-30 09:22:09 +0000748 request = '%s %s %s' % (method, url, self._http_vsn_str)
Greg Stein5e0fa402000-06-26 08:28:01 +0000749
Martin v. Löwisdd5a8602007-06-30 09:22:09 +0000750 # Non-ASCII characters should have been eliminated earlier
751 self._output(request.encode('ascii'))
Greg Stein5e0fa402000-06-26 08:28:01 +0000752
Greg Steindd6eefb2000-07-18 09:09:48 +0000753 if self._http_vsn == 11:
754 # Issue some standard headers for better HTTP/1.1 compliance
Greg Stein5e0fa402000-06-26 08:28:01 +0000755
Jeremy Hylton3921ff62002-03-09 06:07:23 +0000756 if not skip_host:
757 # this header is issued *only* for HTTP/1.1
758 # connections. more specifically, this means it is
759 # only issued when the client uses the new
760 # HTTPConnection() class. backwards-compat clients
761 # will be using HTTP/1.0 and those clients may be
762 # issuing this header themselves. we should NOT issue
763 # it twice; some web servers (such as Apache) barf
764 # when they see two Host: headers
Guido van Rossumf6922aa2001-01-14 21:03:01 +0000765
Jeremy Hylton3921ff62002-03-09 06:07:23 +0000766 # If we need a non-standard port,include it in the
767 # header. If the request is going through a proxy,
768 # but the host of the actual URL, not the host of the
769 # proxy.
Jeremy Hylton8acf1e02002-03-08 19:35:51 +0000770
Jeremy Hylton3921ff62002-03-09 06:07:23 +0000771 netloc = ''
772 if url.startswith('http'):
773 nil, netloc, nil, nil, nil = urlsplit(url)
774
775 if netloc:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000776 try:
777 netloc_enc = netloc.encode("ascii")
778 except UnicodeEncodeError:
779 netloc_enc = netloc.encode("idna")
780 self.putheader('Host', netloc_enc)
Jeremy Hylton3921ff62002-03-09 06:07:23 +0000781 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000782 try:
783 host_enc = self.host.encode("ascii")
784 except UnicodeEncodeError:
785 host_enc = self.host.encode("idna")
Georg Brandl86b2fb92008-07-16 03:43:04 +0000786 if self.port == self.default_port:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000787 self.putheader('Host', host_enc)
788 else:
Guido van Rossum98297ee2007-11-06 21:34:58 +0000789 host_enc = host_enc.decode("ascii")
Thomas Wouters477c8d52006-05-27 19:21:47 +0000790 self.putheader('Host', "%s:%s" % (host_enc, self.port))
Greg Stein5e0fa402000-06-26 08:28:01 +0000791
Greg Steindd6eefb2000-07-18 09:09:48 +0000792 # note: we are assuming that clients will not attempt to set these
793 # headers since *this* library must deal with the
794 # consequences. this also means that when the supporting
795 # libraries are updated to recognize other forms, then this
796 # code should be changed (removed or updated).
Greg Stein5e0fa402000-06-26 08:28:01 +0000797
Greg Steindd6eefb2000-07-18 09:09:48 +0000798 # we only want a Content-Encoding of "identity" since we don't
799 # support encodings such as x-gzip or x-deflate.
Martin v. Löwisaf7dc8d2003-11-19 19:51:55 +0000800 if not skip_accept_encoding:
801 self.putheader('Accept-Encoding', 'identity')
Greg Stein5e0fa402000-06-26 08:28:01 +0000802
Greg Steindd6eefb2000-07-18 09:09:48 +0000803 # we can accept "chunked" Transfer-Encodings, but no others
804 # NOTE: no TE header implies *only* "chunked"
805 #self.putheader('TE', 'chunked')
Greg Stein5e0fa402000-06-26 08:28:01 +0000806
Greg Steindd6eefb2000-07-18 09:09:48 +0000807 # if TE is supplied in the header, then it must appear in a
808 # Connection header.
809 #self.putheader('Connection', 'TE')
Greg Stein5e0fa402000-06-26 08:28:01 +0000810
Greg Steindd6eefb2000-07-18 09:09:48 +0000811 else:
812 # For HTTP/1.0, the server will assume "not chunked"
813 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000814
Benjamin Petersonf608c612008-11-16 18:33:53 +0000815 def putheader(self, header, *values):
Greg Steindd6eefb2000-07-18 09:09:48 +0000816 """Send a request header line to the server.
Greg Stein5e0fa402000-06-26 08:28:01 +0000817
Greg Steindd6eefb2000-07-18 09:09:48 +0000818 For example: h.putheader('Accept', 'text/html')
819 """
820 if self.__state != _CS_REQ_STARTED:
821 raise CannotSendHeader()
Greg Stein5e0fa402000-06-26 08:28:01 +0000822
Guido van Rossum98297ee2007-11-06 21:34:58 +0000823 if hasattr(header, 'encode'):
824 header = header.encode('ascii')
Benjamin Petersonf608c612008-11-16 18:33:53 +0000825 values = list(values)
826 for i, one_value in enumerate(values):
827 if hasattr(one_value, 'encode'):
828 values[i] = one_value.encode('ascii')
829 value = b'\r\n\t'.join(values)
Guido van Rossum98297ee2007-11-06 21:34:58 +0000830 header = header + b': ' + value
831 self._output(header)
Greg Stein5e0fa402000-06-26 08:28:01 +0000832
Greg Steindd6eefb2000-07-18 09:09:48 +0000833 def endheaders(self):
834 """Indicate that the last header line has been sent to the server."""
Greg Stein5e0fa402000-06-26 08:28:01 +0000835
Greg Steindd6eefb2000-07-18 09:09:48 +0000836 if self.__state == _CS_REQ_STARTED:
837 self.__state = _CS_REQ_SENT
838 else:
839 raise CannotSendHeader()
Greg Stein5e0fa402000-06-26 08:28:01 +0000840
Jeremy Hylton8531b1b2002-07-16 21:21:11 +0000841 self._send_output()
Greg Stein5e0fa402000-06-26 08:28:01 +0000842
Greg Steindd6eefb2000-07-18 09:09:48 +0000843 def request(self, method, url, body=None, headers={}):
844 """Send a complete request to the server."""
Greg Steindd6eefb2000-07-18 09:09:48 +0000845 try:
846 self._send_request(method, url, body, headers)
Guido van Rossumb940e112007-01-10 16:19:56 +0000847 except socket.error as v:
Greg Steindd6eefb2000-07-18 09:09:48 +0000848 # trap 'Broken pipe' if we're allowed to automatically reconnect
Guido van Rossum89df2452007-03-19 22:26:27 +0000849 if v.args[0] != 32 or not self.auto_open:
Greg Steindd6eefb2000-07-18 09:09:48 +0000850 raise
851 # try one more time
852 self._send_request(method, url, body, headers)
Greg Stein5e0fa402000-06-26 08:28:01 +0000853
Greg Steindd6eefb2000-07-18 09:09:48 +0000854 def _send_request(self, method, url, body, headers):
Jeremy Hylton2c178252004-08-07 16:28:14 +0000855 # honour explicitly requested Host: and Accept-Encoding headers
856 header_names = dict.fromkeys([k.lower() for k in headers])
857 skips = {}
858 if 'host' in header_names:
859 skips['skip_host'] = 1
860 if 'accept-encoding' in header_names:
861 skips['skip_accept_encoding'] = 1
Greg Stein5e0fa402000-06-26 08:28:01 +0000862
Jeremy Hylton2c178252004-08-07 16:28:14 +0000863 self.putrequest(method, url, **skips)
864
865 if body and ('content-length' not in header_names):
Jeremy Hylton4b878bd2007-08-10 18:49:01 +0000866 thelen = None
Thomas Wouters89f507f2006-12-13 04:49:30 +0000867 try:
Jeremy Hylton4b878bd2007-08-10 18:49:01 +0000868 thelen = str(len(body))
Guido van Rossumb940e112007-01-10 16:19:56 +0000869 except TypeError as te:
Thomas Wouters89f507f2006-12-13 04:49:30 +0000870 # If this is a file-like object, try to
871 # fstat its file descriptor
872 import os
873 try:
874 thelen = str(os.fstat(body.fileno()).st_size)
875 except (AttributeError, OSError):
876 # Don't send a length if this failed
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000877 if self.debuglevel > 0: print("Cannot stat!!")
Thomas Wouters9fe394c2007-02-05 01:24:16 +0000878
Thomas Wouters89f507f2006-12-13 04:49:30 +0000879 if thelen is not None:
880 self.putheader('Content-Length',thelen)
Guido van Rossumcc2b0162007-02-11 06:12:03 +0000881 for hdr, value in headers.items():
Greg Steindd6eefb2000-07-18 09:09:48 +0000882 self.putheader(hdr, value)
883 self.endheaders()
Greg Stein5e0fa402000-06-26 08:28:01 +0000884
Greg Steindd6eefb2000-07-18 09:09:48 +0000885 if body:
Martin v. Löwisdd5a8602007-06-30 09:22:09 +0000886 if isinstance(body, str): body = body.encode('ascii')
Greg Steindd6eefb2000-07-18 09:09:48 +0000887 self.send(body)
Greg Stein5e0fa402000-06-26 08:28:01 +0000888
Greg Steindd6eefb2000-07-18 09:09:48 +0000889 def getresponse(self):
Jeremy Hyltonfb35f652007-08-03 20:30:33 +0000890 """Get the response from the server."""
Greg Stein5e0fa402000-06-26 08:28:01 +0000891
Greg Stein616a58d2003-06-24 06:35:19 +0000892 # if a prior response has been completed, then forget about it.
Greg Steindd6eefb2000-07-18 09:09:48 +0000893 if self.__response and self.__response.isclosed():
894 self.__response = None
Greg Stein5e0fa402000-06-26 08:28:01 +0000895
Greg Steindd6eefb2000-07-18 09:09:48 +0000896 #
897 # if a prior response exists, then it must be completed (otherwise, we
898 # cannot read this response's header to determine the connection-close
899 # behavior)
900 #
901 # note: if a prior response existed, but was connection-close, then the
902 # socket and response were made independent of this HTTPConnection
903 # object since a new request requires that we open a whole new
904 # connection
905 #
906 # this means the prior response had one of two states:
907 # 1) will_close: this connection was reset and the prior socket and
908 # response operate independently
909 # 2) persistent: the response was retained and we await its
910 # isclosed() status to become true.
911 #
912 if self.__state != _CS_REQ_SENT or self.__response:
913 raise ResponseNotReady()
Greg Stein5e0fa402000-06-26 08:28:01 +0000914
Jeremy Hylton30f86742000-09-18 22:50:38 +0000915 if self.debuglevel > 0:
Jeremy Hyltond46aa372002-07-06 18:48:07 +0000916 response = self.response_class(self.sock, self.debuglevel,
Tim Petersc2659cf2003-05-12 20:19:37 +0000917 strict=self.strict,
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000918 method=self._method)
Jeremy Hylton30f86742000-09-18 22:50:38 +0000919 else:
Jeremy Hyltonc1b2cb92003-05-05 16:13:58 +0000920 response = self.response_class(self.sock, strict=self.strict,
921 method=self._method)
Greg Stein5e0fa402000-06-26 08:28:01 +0000922
Jeremy Hylton39c03802002-07-12 14:04:09 +0000923 response.begin()
Jeremy Hyltonbe4fcf12002-06-28 22:38:01 +0000924 assert response.will_close != _UNKNOWN
Greg Steindd6eefb2000-07-18 09:09:48 +0000925 self.__state = _CS_IDLE
Greg Stein5e0fa402000-06-26 08:28:01 +0000926
Greg Steindd6eefb2000-07-18 09:09:48 +0000927 if response.will_close:
Guido van Rossumd8faa362007-04-27 19:54:29 +0000928 # this effectively passes the connection to the response
929 self.close()
Greg Steindd6eefb2000-07-18 09:09:48 +0000930 else:
931 # remember this, so we can tell when it is complete
932 self.__response = response
Greg Stein5e0fa402000-06-26 08:28:01 +0000933
Greg Steindd6eefb2000-07-18 09:09:48 +0000934 return response
Greg Stein5e0fa402000-06-26 08:28:01 +0000935
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000936try:
937 import ssl
938except ImportError:
939 pass
940else:
941 class HTTPSConnection(HTTPConnection):
942 "This class allows communication via SSL."
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000943
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000944 default_port = HTTPS_PORT
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000945
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000946 def __init__(self, host, port=None, key_file=None, cert_file=None,
Georg Brandlf78e02b2008-06-10 17:40:04 +0000947 strict=None, timeout=socket._GLOBAL_DEFAULT_TIMEOUT):
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000948 HTTPConnection.__init__(self, host, port, strict, timeout)
949 self.key_file = key_file
950 self.cert_file = cert_file
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000951
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000952 def connect(self):
953 "Connect to a host on a given (SSL) port."
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000954
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000955 sock = socket.create_connection((self.host, self.port), self.timeout)
Thomas Wouters1b7f8912007-09-19 03:06:30 +0000956 self.sock = ssl.wrap_socket(sock, self.key_file, self.cert_file)
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000957
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000958
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000959 def FakeSocket (sock, sslobj):
Thomas Wouters89d996e2007-09-08 17:39:28 +0000960 warnings.warn("FakeSocket is deprecated, and won't be in 3.x. " +
Thomas Wouters1b7f8912007-09-19 03:06:30 +0000961 "Use the result of ssl.wrap_socket() directly instead.",
Thomas Wouters89d996e2007-09-08 17:39:28 +0000962 DeprecationWarning, stacklevel=2)
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000963 return sslobj
Jeremy Hylton29d27ac2002-07-09 21:22:36 +0000964
Thomas Wouters47b49bf2007-08-30 22:15:33 +0000965 __all__.append("HTTPSConnection")
Greg Stein5e0fa402000-06-26 08:28:01 +0000966
Greg Stein5e0fa402000-06-26 08:28:01 +0000967class HTTPException(Exception):
Jeremy Hylton12f4f352002-07-06 18:55:01 +0000968 # Subclasses that define an __init__ must call Exception.__init__
969 # or define self.args. Otherwise, str() will fail.
Greg Steindd6eefb2000-07-18 09:09:48 +0000970 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000971
972class NotConnected(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +0000973 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000974
Skip Montanaro9d389972002-03-24 16:53:50 +0000975class InvalidURL(HTTPException):
976 pass
977
Greg Stein5e0fa402000-06-26 08:28:01 +0000978class UnknownProtocol(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +0000979 def __init__(self, version):
Jeremy Hylton12f4f352002-07-06 18:55:01 +0000980 self.args = version,
Greg Steindd6eefb2000-07-18 09:09:48 +0000981 self.version = version
Greg Stein5e0fa402000-06-26 08:28:01 +0000982
983class UnknownTransferEncoding(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +0000984 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000985
Greg Stein5e0fa402000-06-26 08:28:01 +0000986class UnimplementedFileMode(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +0000987 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000988
989class IncompleteRead(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +0000990 def __init__(self, partial):
Jeremy Hylton12f4f352002-07-06 18:55:01 +0000991 self.args = partial,
Greg Steindd6eefb2000-07-18 09:09:48 +0000992 self.partial = partial
Greg Stein5e0fa402000-06-26 08:28:01 +0000993
994class ImproperConnectionState(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +0000995 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000996
997class CannotSendRequest(ImproperConnectionState):
Greg Steindd6eefb2000-07-18 09:09:48 +0000998 pass
Greg Stein5e0fa402000-06-26 08:28:01 +0000999
1000class CannotSendHeader(ImproperConnectionState):
Greg Steindd6eefb2000-07-18 09:09:48 +00001001 pass
Greg Stein5e0fa402000-06-26 08:28:01 +00001002
1003class ResponseNotReady(ImproperConnectionState):
Greg Steindd6eefb2000-07-18 09:09:48 +00001004 pass
Greg Stein5e0fa402000-06-26 08:28:01 +00001005
1006class BadStatusLine(HTTPException):
Greg Steindd6eefb2000-07-18 09:09:48 +00001007 def __init__(self, line):
Jeremy Hylton12f4f352002-07-06 18:55:01 +00001008 self.args = line,
Greg Steindd6eefb2000-07-18 09:09:48 +00001009 self.line = line
Greg Stein5e0fa402000-06-26 08:28:01 +00001010
1011# for backwards compatibility
1012error = HTTPException
1013
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001014class LineAndFileWrapper:
1015 """A limited file-like object for HTTP/0.9 responses."""
1016
1017 # The status-line parsing code calls readline(), which normally
1018 # get the HTTP status line. For a 0.9 response, however, this is
1019 # actually the first line of the body! Clients need to get a
1020 # readable file object that contains that line.
1021
1022 def __init__(self, line, file):
1023 self._line = line
1024 self._file = file
1025 self._line_consumed = 0
1026 self._line_offset = 0
1027 self._line_left = len(line)
1028
1029 def __getattr__(self, attr):
1030 return getattr(self._file, attr)
1031
1032 def _done(self):
1033 # called when the last byte is read from the line. After the
1034 # call, all read methods are delegated to the underlying file
Skip Montanaro74b9a7a2003-02-25 17:48:15 +00001035 # object.
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001036 self._line_consumed = 1
1037 self.read = self._file.read
1038 self.readline = self._file.readline
1039 self.readlines = self._file.readlines
1040
1041 def read(self, amt=None):
Hye-Shik Chang39aef792004-06-05 13:30:56 +00001042 if self._line_consumed:
1043 return self._file.read(amt)
1044 assert self._line_left
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001045 if amt is None or amt > self._line_left:
1046 s = self._line[self._line_offset:]
1047 self._done()
1048 if amt is None:
1049 return s + self._file.read()
1050 else:
Tim Petersc411dba2002-07-16 21:35:23 +00001051 return s + self._file.read(amt - len(s))
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001052 else:
1053 assert amt <= self._line_left
1054 i = self._line_offset
1055 j = i + amt
1056 s = self._line[i:j]
1057 self._line_offset = j
1058 self._line_left -= amt
1059 if self._line_left == 0:
1060 self._done()
1061 return s
Tim Petersc411dba2002-07-16 21:35:23 +00001062
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001063 def readline(self):
Hye-Shik Chang39aef792004-06-05 13:30:56 +00001064 if self._line_consumed:
1065 return self._file.readline()
1066 assert self._line_left
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001067 s = self._line[self._line_offset:]
1068 self._done()
1069 return s
1070
1071 def readlines(self, size=None):
Hye-Shik Chang39aef792004-06-05 13:30:56 +00001072 if self._line_consumed:
1073 return self._file.readlines(size)
1074 assert self._line_left
Jeremy Hyltond46aa372002-07-06 18:48:07 +00001075 L = [self._line[self._line_offset:]]
1076 self._done()
1077 if size is None:
1078 return L + self._file.readlines()
1079 else:
1080 return L + self._file.readlines(size)