1 """HTTP/1.1 client library
3 <intro stuff goes here>
6 HTTPConnection goes through a number of "states", which define when a client
7 may legally make another request or fetch the response for a particular
8 request. This diagram details these state transitions:
20 | ( putheader() )* endheaders()
24 | response = getresponse()
26 Unread-response [Response-headers-read]
27 |\____________________
29 | response.read() | putrequest()
31 Idle Req-started-unread-response
34 response.read() | | ( putheader() )* endheaders()
36 Request-started Req-sent-unread-response
42 This 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
48 Note: 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.
59 Logical State __state __response
60 ------------- ------- ----------
62 Request-started _CS_REQ_STARTED None
63 Request-sent _CS_REQ_SENT None
64 Unread-response _CS_IDLE <response_class>
65 Req-started-unread-response _CS_REQ_STARTED <response_class>
66 Req-sent-unread-response _CS_REQ_SENT <response_class>
71 from urlparse
import urlsplit
75 from cStringIO
import StringIO
77 from StringIO
import StringIO
79 __all__
= ["HTTP", "HTTPResponse", "HTTPConnection",
80 "HTTPException", "NotConnected", "UnknownProtocol",
81 "UnknownTransferEncoding", "UnimplementedFileMode",
82 "IncompleteRead", "InvalidURL", "ImproperConnectionState",
83 "CannotSendRequest", "CannotSendHeader", "ResponseNotReady",
84 "BadStatusLine", "error", "responses"]
93 _CS_REQ_STARTED
= 'Request-started'
94 _CS_REQ_SENT
= 'Request-sent'
99 SWITCHING_PROTOCOLS
= 101
106 NON_AUTHORITATIVE_INFORMATION
= 203
109 PARTIAL_CONTENT
= 206
114 MULTIPLE_CHOICES
= 300
115 MOVED_PERMANENTLY
= 301
120 TEMPORARY_REDIRECT
= 307
125 PAYMENT_REQUIRED
= 402
128 METHOD_NOT_ALLOWED
= 405
130 PROXY_AUTHENTICATION_REQUIRED
= 407
131 REQUEST_TIMEOUT
= 408
134 LENGTH_REQUIRED
= 411
135 PRECONDITION_FAILED
= 412
136 REQUEST_ENTITY_TOO_LARGE
= 413
137 REQUEST_URI_TOO_LONG
= 414
138 UNSUPPORTED_MEDIA_TYPE
= 415
139 REQUESTED_RANGE_NOT_SATISFIABLE
= 416
140 EXPECTATION_FAILED
= 417
141 UNPROCESSABLE_ENTITY
= 422
143 FAILED_DEPENDENCY
= 424
144 UPGRADE_REQUIRED
= 426
147 INTERNAL_SERVER_ERROR
= 500
148 NOT_IMPLEMENTED
= 501
150 SERVICE_UNAVAILABLE
= 503
151 GATEWAY_TIMEOUT
= 504
152 HTTP_VERSION_NOT_SUPPORTED
= 505
153 INSUFFICIENT_STORAGE
= 507
156 # Mapping status codes to official W3C names
159 101: 'Switching Protocols',
164 203: 'Non-Authoritative Information',
166 205: 'Reset Content',
167 206: 'Partial Content',
169 300: 'Multiple Choices',
170 301: 'Moved Permanently',
176 307: 'Temporary Redirect',
180 402: 'Payment Required',
183 405: 'Method Not Allowed',
184 406: 'Not Acceptable',
185 407: 'Proxy Authentication Required',
186 408: 'Request Timeout',
189 411: 'Length Required',
190 412: 'Precondition Failed',
191 413: 'Request Entity Too Large',
192 414: 'Request-URI Too Long',
193 415: 'Unsupported Media Type',
194 416: 'Requested Range Not Satisfiable',
195 417: 'Expectation Failed',
197 500: 'Internal Server Error',
198 501: 'Not Implemented',
200 503: 'Service Unavailable',
201 504: 'Gateway Timeout',
202 505: 'HTTP Version Not Supported',
205 # maximal amount of data to read at one time in _safe_read
208 class HTTPMessage(mimetools
.Message
):
210 def addheader(self
, key
, value
):
211 """Add header for field key handling repeats."""
212 prev
= self
.dict.get(key
)
214 self
.dict[key
] = value
216 combined
= ", ".join((prev
, value
))
217 self
.dict[key
] = combined
219 def addcontinue(self
, key
, more
):
220 """Add more field data from a continuation line."""
221 prev
= self
.dict[key
]
222 self
.dict[key
] = prev
+ "\n " + more
224 def readheaders(self
):
225 """Read header lines.
227 Read header lines up to the entirely blank line that terminates them.
228 The (normally blank) line that ends the headers is skipped, but not
229 included in the returned list. If a non-header line ends the headers,
230 (which is an error), an attempt is made to backspace over it; it is
231 never included in the returned list.
233 The variable self.status is set to the empty string if all went well,
234 otherwise it is an error message. The variable self.headers is a
235 completely uninterpreted list of lines contained in the header (so
236 printing them will reproduce the header exactly as it appears in the
239 If multiple header fields with the same name occur, they are combined
240 according to the rules in RFC 2616 sec 4.2:
242 Appending each subsequent field-value to the first, each separated
243 by a comma. The order in which header fields with the same field-name
244 are received is significant to the interpretation of the combined
247 # XXX The implementation overrides the readheaders() method of
248 # rfc822.Message. The base class design isn't amenable to
249 # customized behavior here so the method here is a copy of the
250 # base class code with a few small changes.
254 self
.headers
= hlist
= []
258 startofline
= unread
= tell
= None
259 if hasattr(self
.fp
, 'unread'):
260 unread
= self
.fp
.unread
268 startofline
= tell
= None
270 line
= self
.fp
.readline()
272 self
.status
= 'EOF in headers'
274 # Skip unix From name time lines
275 if firstline
and line
.startswith('From '):
276 self
.unixfrom
= self
.unixfrom
+ line
279 if headerseen
and line
[0] in ' \t':
280 # XXX Not sure if continuation lines are handled properly
281 # for http and/or for repeating headers
282 # It's a continuation line.
284 self
.addcontinue(headerseen
, line
.strip())
286 elif self
.iscomment(line
):
287 # It's a comment. Ignore it.
289 elif self
.islast(line
):
290 # Note! No pushback here! The delimiter line gets eaten.
292 headerseen
= self
.isheader(line
)
294 # It's a legal header line, save it.
296 self
.addheader(headerseen
, line
[len(headerseen
)+1:].strip())
299 # It's not a header line; throw it back and stop here.
301 self
.status
= 'No headers'
303 self
.status
= 'Non-header line where header expected'
304 # Try to undo the read.
308 self
.fp
.seek(startofline
)
310 self
.status
= self
.status
+ '; bad seek'
315 # strict: If true, raise BadStatusLine if the status line can't be
316 # parsed as a valid HTTP/1.0 or 1.1 status line. By default it is
317 # false because it prevents clients from talking to HTTP/0.9
318 # servers. Note that a response with a sufficiently corrupted
319 # status line will look like an HTTP/0.9 response.
321 # See RFC 2616 sec 19.6 and RFC 1945 sec 6 for details.
323 def __init__(self
, sock
, debuglevel
=0, strict
=0, method
=None):
324 self
.fp
= sock
.makefile('rb', 0)
325 self
.debuglevel
= debuglevel
327 self
._method
= method
331 # from the Status-Line of the response
332 self
.version
= _UNKNOWN
# HTTP-Version
333 self
.status
= _UNKNOWN
# Status-Code
334 self
.reason
= _UNKNOWN
# Reason-Phrase
336 self
.chunked
= _UNKNOWN
# is "chunked" being used?
337 self
.chunk_left
= _UNKNOWN
# bytes left to read in current chunk
338 self
.length
= _UNKNOWN
# number of bytes left in response
339 self
.will_close
= _UNKNOWN
# conn will close at end of response
341 def _read_status(self
):
342 # Initialize with Simple-Response defaults
343 line
= self
.fp
.readline()
344 if self
.debuglevel
> 0:
345 print "reply:", repr(line
)
347 # Presumably, the server closed the connection before
348 # sending a valid response.
349 raise BadStatusLine(line
)
351 [version
, status
, reason
] = line
.split(None, 2)
354 [version
, status
] = line
.split(None, 1)
357 # empty version will cause next test to fail and status
358 # will be treated as 0.9 response.
360 if not version
.startswith('HTTP/'):
363 raise BadStatusLine(line
)
365 # assume it's a Simple-Response from an 0.9 server
366 self
.fp
= LineAndFileWrapper(line
, self
.fp
)
367 return "HTTP/0.9", 200, ""
369 # The status code is a three-digit number
372 if status
< 100 or status
> 999:
373 raise BadStatusLine(line
)
375 raise BadStatusLine(line
)
376 return version
, status
, reason
379 if self
.msg
is not None:
380 # we've already started reading the response
383 # read until we get a non-100 response
385 version
, status
, reason
= self
._read
_status
()
386 if status
!= CONTINUE
:
388 # skip the header from the 100 response
390 skip
= self
.fp
.readline().strip()
393 if self
.debuglevel
> 0:
394 print "header:", skip
397 self
.reason
= reason
.strip()
398 if version
== 'HTTP/1.0':
400 elif version
.startswith('HTTP/1.'):
401 self
.version
= 11 # use HTTP/1.1 code for HTTP/1.x where x>=1
402 elif version
== 'HTTP/0.9':
405 raise UnknownProtocol(version
)
407 if self
.version
== 9:
411 self
.msg
= HTTPMessage(StringIO())
414 self
.msg
= HTTPMessage(self
.fp
, 0)
415 if self
.debuglevel
> 0:
416 for hdr
in self
.msg
.headers
:
417 print "header:", hdr
,
419 # don't let the msg keep an fp
422 # are we using the chunked-style of transfer encoding?
423 tr_enc
= self
.msg
.getheader('transfer-encoding')
424 if tr_enc
and tr_enc
.lower() == "chunked":
426 self
.chunk_left
= None
430 # will the connection close at the end of the response?
431 self
.will_close
= self
._check
_close
()
433 # do we have a Content-Length?
434 # NOTE: RFC 2616, S4.4, #3 says we ignore this if tr_enc is "chunked"
435 length
= self
.msg
.getheader('content-length')
436 if length
and not self
.chunked
:
438 self
.length
= int(length
)
442 if self
.length
< 0: # ignore nonsensical negative lengths
447 # does the body have a fixed length? (of zero)
448 if (status
== NO_CONTENT
or status
== NOT_MODIFIED
or
449 100 <= status
< 200 or # 1xx codes
450 self
._method
== 'HEAD'):
453 # if the connection remains open, and we aren't using chunked, and
454 # a content-length was not provided, then assume that the connection
456 if not self
.will_close
and \
457 not self
.chunked
and \
461 def _check_close(self
):
462 conn
= self
.msg
.getheader('connection')
463 if self
.version
== 11:
464 # An HTTP/1.1 proxy is assumed to stay open unless
466 conn
= self
.msg
.getheader('connection')
467 if conn
and "close" in conn
.lower():
471 # Some HTTP/1.0 implementations have support for persistent
472 # connections, using rules different than HTTP/1.1.
474 # For older HTTP, Keep-Alive indicates persistent connection.
475 if self
.msg
.getheader('keep-alive'):
478 # At least Akamai returns a "Connection: Keep-Alive" header,
479 # which was supposed to be sent by the client.
480 if conn
and "keep-alive" in conn
.lower():
483 # Proxy-Connection is a netscape hack.
484 pconn
= self
.msg
.getheader('proxy-connection')
485 if pconn
and "keep-alive" in pconn
.lower():
488 # otherwise, assume it will close
497 # NOTE: it is possible that we will not ever call self.close(). This
498 # case occurs when will_close is TRUE, length is None, and we
499 # read up to the last byte, but NOT past it.
501 # IMPLIES: if will_close is FALSE, then self.close() will ALWAYS be
502 # called, meaning self.isclosed() is meaningful.
503 return self
.fp
is None
505 # XXX It would be nice to have readline and __iter__ for this, too.
507 def read(self
, amt
=None):
512 return self
._read
_chunked
(amt
)
516 if self
.length
is None:
519 s
= self
._safe
_read
(self
.length
)
521 self
.close() # we read everything
524 if self
.length
is not None:
525 if amt
> self
.length
:
526 # clip the read to the "end of response"
529 # we do not use _safe_read() here because this may be a .will_close
530 # connection, and the user is reading more bytes than will be provided
531 # (for example, reading in 1k chunks)
532 s
= self
.fp
.read(amt
)
533 if self
.length
is not None:
534 self
.length
-= len(s
)
539 def _read_chunked(self
, amt
):
540 assert self
.chunked
!= _UNKNOWN
541 chunk_left
= self
.chunk_left
544 # XXX This accumulates chunks by repeated string concatenation,
545 # which is not efficient as the number or size of chunks gets big.
547 if chunk_left
is None:
548 line
= self
.fp
.readline()
551 line
= line
[:i
] # strip chunk-extensions
553 chunk_left
= int(line
, 16)
555 # close the connection as protocol synchronisation is
558 raise IncompleteRead(value
)
562 value
+= self
._safe
_read
(chunk_left
)
563 elif amt
< chunk_left
:
564 value
+= self
._safe
_read
(amt
)
565 self
.chunk_left
= chunk_left
- amt
567 elif amt
== chunk_left
:
568 value
+= self
._safe
_read
(amt
)
569 self
._safe
_read
(2) # toss the CRLF at the end of the chunk
570 self
.chunk_left
= None
573 value
+= self
._safe
_read
(chunk_left
)
576 # we read the whole chunk, get another
577 self
._safe
_read
(2) # toss the CRLF at the end of the chunk
580 # read and discard trailer up to the CRLF terminator
581 ### note: we shouldn't have any trailers!
583 line
= self
.fp
.readline()
585 # a vanishingly small number of sites EOF without
586 # sending the trailer
591 # we read everything; close the "file"
596 def _safe_read(self
, amt
):
597 """Read the number of bytes requested, compensating for partial reads.
599 Normally, we have a blocking socket, but a read() can be interrupted
600 by a signal (resulting in a partial read).
602 Note that we cannot distinguish between EOF and an interrupt when zero
603 bytes have been read. IncompleteRead() will be raised in this
606 This function should be used when <amt> bytes "should" be present for
607 reading. If the bytes are truly not available (due to EOF), then the
608 IncompleteRead exception can be used to detect the problem.
612 chunk
= self
.fp
.read(min(amt
, MAXAMOUNT
))
614 raise IncompleteRead(s
)
619 def getheader(self
, name
, default
=None):
621 raise ResponseNotReady()
622 return self
.msg
.getheader(name
, default
)
624 def getheaders(self
):
625 """Return list of (header, value) tuples."""
627 raise ResponseNotReady()
628 return self
.msg
.items()
631 class HTTPConnection
:
634 _http_vsn_str
= 'HTTP/1.1'
636 response_class
= HTTPResponse
637 default_port
= HTTP_PORT
642 def __init__(self
, host
, port
=None, strict
=None, timeout
=None):
643 self
.timeout
= timeout
646 self
.__response
= None
647 self
.__state
= _CS_IDLE
650 self
._set
_hostport
(host
, port
)
651 if strict
is not None:
654 def _set_hostport(self
, host
, port
):
657 j
= host
.rfind(']') # ipv6 addresses have [...]
660 port
= int(host
[i
+1:])
662 raise InvalidURL("nonnumeric port: '%s'" % host
[i
+1:])
665 port
= self
.default_port
666 if host
and host
[0] == '[' and host
[-1] == ']':
671 def set_debuglevel(self
, level
):
672 self
.debuglevel
= level
675 """Connect to the host and port specified in __init__."""
676 self
.sock
= socket
.create_connection((self
.host
,self
.port
),
680 """Close the connection to the HTTP server."""
682 self
.sock
.close() # close it manually... there may be other refs
685 self
.__response
.close()
686 self
.__response
= None
687 self
.__state
= _CS_IDLE
690 """Send `str' to the server."""
691 if self
.sock
is None:
697 # send the data to the server. if we get a broken pipe, then close
698 # the socket. we want to reconnect when somebody tries to send again.
700 # NOTE: we DO propagate the error, though, because we cannot simply
701 # ignore the error... the caller will know if they can retry.
702 if self
.debuglevel
> 0:
703 print "send:", repr(str)
706 if hasattr(str,'read') :
707 if self
.debuglevel
> 0: print "sendIng a read()able"
708 data
=str.read(blocksize
)
710 self
.sock
.sendall(data
)
711 data
=str.read(blocksize
)
713 self
.sock
.sendall(str)
714 except socket
.error
, v
:
715 if v
[0] == 32: # Broken pipe
719 def _output(self
, s
):
720 """Add a line of output to the current request buffer.
722 Assumes that the line does *not* end with \\r\\n.
724 self
._buffer
.append(s
)
726 def _send_output(self
):
727 """Send the currently buffered request and clear the buffer.
729 Appends an extra \\r\\n to the buffer.
731 self
._buffer
.extend(("", ""))
732 msg
= "\r\n".join(self
._buffer
)
736 def putrequest(self
, method
, url
, skip_host
=0, skip_accept_encoding
=0):
737 """Send a request to the server.
739 `method' specifies an HTTP request method, e.g. 'GET'.
740 `url' specifies the object being requested, e.g. '/index.html'.
741 `skip_host' if True does not add automatically a 'Host:' header
742 `skip_accept_encoding' if True does not add automatically an
743 'Accept-Encoding:' header
746 # if a prior response has been completed, then forget about it.
747 if self
.__response
and self
.__response
.isclosed():
748 self
.__response
= None
751 # in certain cases, we cannot issue another request on this connection.
753 # 1) we are in the process of sending a request. (_CS_REQ_STARTED)
754 # 2) a response to a previous request has signalled that it is going
755 # to close the connection upon completion.
756 # 3) the headers for the previous response have not been read, thus
757 # we cannot determine whether point (2) is true. (_CS_REQ_SENT)
759 # if there is no prior response, then we can request at will.
761 # if point (2) is true, then we will have passed the socket to the
762 # response (effectively meaning, "there is no prior response"), and
763 # will open a new one when a new request is made.
765 # Note: if a prior response exists, then we *can* start a new request.
766 # We are not allowed to begin fetching the response to this new
767 # request, however, until that prior response is complete.
769 if self
.__state
== _CS_IDLE
:
770 self
.__state
= _CS_REQ_STARTED
772 raise CannotSendRequest()
774 # Save the method we use, we need it later in the response phase
775 self
._method
= method
778 str = '%s %s %s' % (method
, url
, self
._http
_vsn
_str
)
782 if self
._http
_vsn
== 11:
783 # Issue some standard headers for better HTTP/1.1 compliance
786 # this header is issued *only* for HTTP/1.1
787 # connections. more specifically, this means it is
788 # only issued when the client uses the new
789 # HTTPConnection() class. backwards-compat clients
790 # will be using HTTP/1.0 and those clients may be
791 # issuing this header themselves. we should NOT issue
792 # it twice; some web servers (such as Apache) barf
793 # when they see two Host: headers
795 # If we need a non-standard port,include it in the
796 # header. If the request is going through a proxy,
797 # but the host of the actual URL, not the host of the
801 if url
.startswith('http'):
802 nil
, netloc
, nil
, nil
, nil
= urlsplit(url
)
806 netloc_enc
= netloc
.encode("ascii")
807 except UnicodeEncodeError:
808 netloc_enc
= netloc
.encode("idna")
809 self
.putheader('Host', netloc_enc
)
812 host_enc
= self
.host
.encode("ascii")
813 except UnicodeEncodeError:
814 host_enc
= self
.host
.encode("idna")
815 if self
.port
== HTTP_PORT
:
816 self
.putheader('Host', host_enc
)
818 self
.putheader('Host', "%s:%s" % (host_enc
, self
.port
))
820 # note: we are assuming that clients will not attempt to set these
821 # headers since *this* library must deal with the
822 # consequences. this also means that when the supporting
823 # libraries are updated to recognize other forms, then this
824 # code should be changed (removed or updated).
826 # we only want a Content-Encoding of "identity" since we don't
827 # support encodings such as x-gzip or x-deflate.
828 if not skip_accept_encoding
:
829 self
.putheader('Accept-Encoding', 'identity')
831 # we can accept "chunked" Transfer-Encodings, but no others
832 # NOTE: no TE header implies *only* "chunked"
833 #self.putheader('TE', 'chunked')
835 # if TE is supplied in the header, then it must appear in a
837 #self.putheader('Connection', 'TE')
840 # For HTTP/1.0, the server will assume "not chunked"
843 def putheader(self
, header
, value
):
844 """Send a request header line to the server.
846 For example: h.putheader('Accept', 'text/html')
848 if self
.__state
!= _CS_REQ_STARTED
:
849 raise CannotSendHeader()
851 str = '%s: %s' % (header
, value
)
854 def endheaders(self
):
855 """Indicate that the last header line has been sent to the server."""
857 if self
.__state
== _CS_REQ_STARTED
:
858 self
.__state
= _CS_REQ_SENT
860 raise CannotSendHeader()
864 def request(self
, method
, url
, body
=None, headers
={}):
865 """Send a complete request to the server."""
868 self
._send
_request
(method
, url
, body
, headers
)
869 except socket
.error
, v
:
870 # trap 'Broken pipe' if we're allowed to automatically reconnect
871 if v
[0] != 32 or not self
.auto_open
:
874 self
._send
_request
(method
, url
, body
, headers
)
876 def _send_request(self
, method
, url
, body
, headers
):
877 # honour explicitly requested Host: and Accept-Encoding headers
878 header_names
= dict.fromkeys([k
.lower() for k
in headers
])
880 if 'host' in header_names
:
881 skips
['skip_host'] = 1
882 if 'accept-encoding' in header_names
:
883 skips
['skip_accept_encoding'] = 1
885 self
.putrequest(method
, url
, **skips
)
887 if body
and ('content-length' not in header_names
):
890 thelen
=str(len(body
))
891 except TypeError, te
:
892 # If this is a file-like object, try to
893 # fstat its file descriptor
896 thelen
= str(os
.fstat(body
.fileno()).st_size
)
897 except (AttributeError, OSError):
898 # Don't send a length if this failed
899 if self
.debuglevel
> 0: print "Cannot stat!!"
901 if thelen
is not None:
902 self
.putheader('Content-Length',thelen
)
903 for hdr
, value
in headers
.iteritems():
904 self
.putheader(hdr
, value
)
910 def getresponse(self
):
911 "Get the response from the server."
913 # if a prior response has been completed, then forget about it.
914 if self
.__response
and self
.__response
.isclosed():
915 self
.__response
= None
918 # if a prior response exists, then it must be completed (otherwise, we
919 # cannot read this response's header to determine the connection-close
922 # note: if a prior response existed, but was connection-close, then the
923 # socket and response were made independent of this HTTPConnection
924 # object since a new request requires that we open a whole new
927 # this means the prior response had one of two states:
928 # 1) will_close: this connection was reset and the prior socket and
929 # response operate independently
930 # 2) persistent: the response was retained and we await its
931 # isclosed() status to become true.
933 if self
.__state
!= _CS_REQ_SENT
or self
.__response
:
934 raise ResponseNotReady()
936 if self
.debuglevel
> 0:
937 response
= self
.response_class(self
.sock
, self
.debuglevel
,
941 response
= self
.response_class(self
.sock
, strict
=self
.strict
,
945 assert response
.will_close
!= _UNKNOWN
946 self
.__state
= _CS_IDLE
948 if response
.will_close
:
949 # this effectively passes the connection to the response
952 # remember this, so we can tell when it is complete
953 self
.__response
= response
959 "Compatibility class with httplib.py from 1.5."
962 _http_vsn_str
= 'HTTP/1.0'
966 _connection_class
= HTTPConnection
968 def __init__(self
, host
='', port
=None, strict
=None):
969 "Provide a default host, since the superclass requires one."
971 # some joker passed 0 explicitly, meaning default port
975 # Note that we may pass an empty string as the host; this will throw
976 # an error when we attempt to connect. Presumably, the client code
977 # will call connect before then, with a proper host.
978 self
._setup
(self
._connection
_class
(host
, port
, strict
))
980 def _setup(self
, conn
):
983 # set up delegation to flesh out interface
984 self
.send
= conn
.send
985 self
.putrequest
= conn
.putrequest
986 self
.endheaders
= conn
.endheaders
987 self
.set_debuglevel
= conn
.set_debuglevel
989 conn
._http
_vsn
= self
._http
_vsn
990 conn
._http
_vsn
_str
= self
._http
_vsn
_str
994 def connect(self
, host
=None, port
=None):
995 "Accept arguments to set the host/port, since the superclass doesn't."
998 self
._conn
._set
_hostport
(host
, port
)
1002 "Provide a getfile, since the superclass' does not use this concept."
1005 def putheader(self
, header
, *values
):
1006 "The superclass allows only one value argument."
1007 self
._conn
.putheader(header
, '\r\n\t'.join(values
))
1010 """Compat definition since superclass does not define it.
1012 Returns a tuple consisting of:
1013 - server status code (e.g. '200' if all goes well)
1014 - server "reason" corresponding to status code
1015 - any RFC822 headers in the response from the server
1018 response
= self
._conn
.getresponse()
1019 except BadStatusLine
, e
:
1020 ### hmm. if getresponse() ever closes the socket on a bad request,
1021 ### then we are going to have problems with self.sock
1023 ### should we keep this behavior? do people use it?
1024 # keep the socket open (as a file), and return it
1025 self
.file = self
._conn
.sock
.makefile('rb', 0)
1027 # close our socket -- we want to restart after any protocol error
1031 return -1, e
.line
, None
1033 self
.headers
= response
.msg
1034 self
.file = response
.fp
1035 return response
.status
, response
.reason
, response
.msg
1040 # note that self.file == response.fp, which gets closed by the
1041 # superclass. just clear the object ref here.
1042 ### hmm. messy. if status==-1, then self.file is owned by us.
1043 ### well... we aren't explicitly closing, but losing this ref will
1052 class HTTPSConnection(HTTPConnection
):
1053 "This class allows communication via SSL."
1055 default_port
= HTTPS_PORT
1057 def __init__(self
, host
, port
=None, key_file
=None, cert_file
=None,
1058 strict
=None, timeout
=None):
1059 HTTPConnection
.__init
__(self
, host
, port
, strict
, timeout
)
1060 self
.key_file
= key_file
1061 self
.cert_file
= cert_file
1064 "Connect to a host on a given (SSL) port."
1066 sock
= socket
.create_connection((self
.host
, self
.port
), self
.timeout
)
1067 self
.sock
= ssl
.wrap_socket(sock
, self
.key_file
, self
.cert_file
)
1069 __all__
.append("HTTPSConnection")
1072 """Compatibility with 1.5 httplib interface
1074 Python 1.5.2 did not have an HTTPS class, but it defined an
1075 interface for sending http requests that is also useful for
1079 _connection_class
= HTTPSConnection
1081 def __init__(self
, host
='', port
=None, key_file
=None, cert_file
=None,
1083 # provide a default host, pass the X509 cert info
1085 # urf. compensate for bad input.
1088 self
._setup
(self
._connection
_class
(host
, port
, key_file
,
1091 # we never actually use these for anything, but we keep them
1092 # here for compatibility with post-1.5.2 CVS.
1093 self
.key_file
= key_file
1094 self
.cert_file
= cert_file
1097 def FakeSocket (sock
, sslobj
):
1098 warnings
.warn("FakeSocket is deprecated, and won't be in 3.x. " +
1099 "Use the result of ssl.wrap_socket() directly instead.",
1100 DeprecationWarning, stacklevel
=2)
1104 class HTTPException(Exception):
1105 # Subclasses that define an __init__ must call Exception.__init__
1106 # or define self.args. Otherwise, str() will fail.
1109 class NotConnected(HTTPException
):
1112 class InvalidURL(HTTPException
):
1115 class UnknownProtocol(HTTPException
):
1116 def __init__(self
, version
):
1117 self
.args
= version
,
1118 self
.version
= version
1120 class UnknownTransferEncoding(HTTPException
):
1123 class UnimplementedFileMode(HTTPException
):
1126 class IncompleteRead(HTTPException
):
1127 def __init__(self
, partial
):
1128 self
.args
= partial
,
1129 self
.partial
= partial
1131 class ImproperConnectionState(HTTPException
):
1134 class CannotSendRequest(ImproperConnectionState
):
1137 class CannotSendHeader(ImproperConnectionState
):
1140 class ResponseNotReady(ImproperConnectionState
):
1143 class BadStatusLine(HTTPException
):
1144 def __init__(self
, line
):
1148 # for backwards compatibility
1149 error
= HTTPException
1151 class LineAndFileWrapper
:
1152 """A limited file-like object for HTTP/0.9 responses."""
1154 # The status-line parsing code calls readline(), which normally
1155 # get the HTTP status line. For a 0.9 response, however, this is
1156 # actually the first line of the body! Clients need to get a
1157 # readable file object that contains that line.
1159 def __init__(self
, line
, file):
1162 self
._line
_consumed
= 0
1163 self
._line
_offset
= 0
1164 self
._line
_left
= len(line
)
1166 def __getattr__(self
, attr
):
1167 return getattr(self
._file
, attr
)
1170 # called when the last byte is read from the line. After the
1171 # call, all read methods are delegated to the underlying file
1173 self
._line
_consumed
= 1
1174 self
.read
= self
._file
.read
1175 self
.readline
= self
._file
.readline
1176 self
.readlines
= self
._file
.readlines
1178 def read(self
, amt
=None):
1179 if self
._line
_consumed
:
1180 return self
._file
.read(amt
)
1181 assert self
._line
_left
1182 if amt
is None or amt
> self
._line
_left
:
1183 s
= self
._line
[self
._line
_offset
:]
1186 return s
+ self
._file
.read()
1188 return s
+ self
._file
.read(amt
- len(s
))
1190 assert amt
<= self
._line
_left
1191 i
= self
._line
_offset
1194 self
._line
_offset
= j
1195 self
._line
_left
-= amt
1196 if self
._line
_left
== 0:
1201 if self
._line
_consumed
:
1202 return self
._file
.readline()
1203 assert self
._line
_left
1204 s
= self
._line
[self
._line
_offset
:]
1208 def readlines(self
, size
=None):
1209 if self
._line
_consumed
:
1210 return self
._file
.readlines(size
)
1211 assert self
._line
_left
1212 L
= [self
._line
[self
._line
_offset
:]]
1215 return L
+ self
._file
.readlines()
1217 return L
+ self
._file
.readlines(size
)
1220 """Test this module.
1222 A hodge podge of tests collected here, because they have too many
1223 external dependencies for the regular test suite.
1228 opts
, args
= getopt
.getopt(sys
.argv
[1:], 'd')
1231 if o
== '-d': dl
= dl
+ 1
1232 host
= 'www.python.org'
1234 if args
[0:]: host
= args
[0]
1235 if args
[1:]: selector
= args
[1]
1237 h
.set_debuglevel(dl
)
1239 h
.putrequest('GET', selector
)
1241 status
, reason
, headers
= h
.getreply()
1242 print 'status =', status
1243 print 'reason =', reason
1244 print "read", len(h
.getfile().read())
1247 for header
in headers
.headers
: print header
.strip()
1250 # minimal test that code to extract host from url works
1253 _http_vsn_str
= 'HTTP/1.1'
1255 h
= HTTP11('www.python.org')
1256 h
.putrequest('GET', 'http://www.python.org/~jeremy/')
1267 for host
, selector
in (('sourceforge.net', '/projects/python'),
1269 print "https://%s%s" % (host
, selector
)
1271 hs
.set_debuglevel(dl
)
1273 hs
.putrequest('GET', selector
)
1275 status
, reason
, headers
= hs
.getreply()
1276 print 'status =', status
1277 print 'reason =', reason
1278 print "read", len(hs
.getfile().read())
1281 for header
in headers
.headers
: print header
.strip()
1284 if __name__
== '__main__':