1
2
3
4
5
6 import cgi
7 import errno
8 import logging
9 import mimetypes
10 import os
11 import socket
12 import time
13 try:
14 from cStringIO import StringIO
15 except ImportError:
16 from StringIO import StringIO
17 import types
18 import urlparse
19 import uuid
20
21 from restkit import __version__
22 from restkit.errors import RequestError, InvalidUrl, RedirectLimit, \
23 AlreadyRead
24 from restkit.filters import Filters
25 from restkit.forms import multipart_form_encode, form_encode
26 from restkit.util import sock
27 from restkit import tee
28 from restkit import util
29 from restkit.util.misc import deprecated_property
30 from restkit import http
31
32 MAX_FOLLOW_REDIRECTS = 5
33
34 USER_AGENT = "restkit/%s" % __version__
35
36 log = logging.getLogger(__name__)
39 """ Http Response object returned by HttpConnction"""
40
41 charset = "utf8"
42 unicode_errors = 'strict'
43
44 - def __init__(self, response, final_url):
45 self.response = response
46 self.status = response.status
47 self.status_int = response.status_int
48 self.version = response.version
49 self.headerslist = response.headers
50 self.final_url = final_url
51
52 headers = {}
53 for key, value in response.headers:
54 headers[key.lower()] = value
55 self.headers = headers
56 self.closed = False
57
59 try:
60 return getattr(self, key)
61 except AttributeError:
62 pass
63 return self.headers[key.lower()]
64
66 return (key.lower() in self.headers)
67
69 for item in list(self.headers.items()):
70 yield item
71
72 - def body_string(self, charset=None, unicode_errors="strict"):
73 """ return body string, by default in bytestring """
74 if self.closed:
75 raise AlreadyRead("The response have already been read")
76 body = self.response.body.read()
77 if charset is not None:
78 try:
79 body = body.decode(charset, unicode_errors)
80 except UnicodeDecodeError:
81 pass
82 self.close()
83 return body
84
85 - def body_stream(self):
86 """ return full body stream """
87 if self.closed:
88 raise AlreadyRead("The response have already been read")
89 return self.response.body
90
92 """ release the socket """
93 self.response.body.close()
94
95 @property
97 """ body in bytestring """
98 return self.body_string()
99
100 body = deprecated_property(
101 body, 'body', 'use body_string() instead',
102 warning=True)
103
104 @property
105 - def body_file(self):
106 """ return body as a file like object"""
107 return self.body_stream()
108
109 body_file = deprecated_property(
110 body_file, 'body_file', 'use body_stream() instead',
111 warning=True)
112
113 @property
114 - def unicode_body(self):
115 """ like body but converted to unicode"""
116 if not self.charset:
117 raise AttributeError(
118 "You cannot access HttpResponse.unicode_body unless charset is set")
119 body = self.body_string()
120 return body.decode(self.charset, self.unicode_errors)
121
122 unicode_body = deprecated_property(
123 unicode_body, 'unicode_body', 'replaced by body_string()',
124 warning=True)
125
127 """ Http Connection object. """
128
129 version = (1, 1)
130 response_class = HttpResponse
131
132
138
139 """ HttpConnection constructor
140
141 :param timeout: socket timeout
142 :param filters: list, list of http filters. see the doc of http filters
143 for more info
144 :param follow_redirect: boolean, by default is false. If true,
145 if the HTTP status is 301, 302 or 303 the client will follow
146 the location.
147 :param max_follow_redirect: max number of redirection. If max is reached
148 the RedirectLimit exception is raised.
149 :param pool_instance: a pool instance inherited from
150 `restkit.pool.PoolInterface`
151 :param ssl_args: ssl arguments. See http://docs.python.org/library/ssl.html
152 for more information.
153 """
154 self._sock = None
155 self.timeout = timeout
156 self.headers = []
157 self.req_headers = []
158 self.ua = USER_AGENT
159 self.url = None
160
161 self.follow_redirect = follow_redirect
162 self.nb_redirections = max_follow_redirect
163 self.force_follow_redirect = force_follow_redirect
164 self.method = 'GET'
165 self.body = None
166 self.response_body = StringIO()
167 self.final_url = None
168
169
170 self.filters = Filters(filters)
171 self.ssl_args = ssl_args
172
173 if not pool_instance:
174 self.should_close = True
175 self.pool = None
176 else:
177 self.pool = pool_instance
178 self.should_close = False
179
180 if response_class is not None:
181 self.response_class = response_class
182
184 """ initate a connection if needed or reuse a socket"""
185
186
187 self.filters.apply("on_connect", self)
188 if self._sock is not None:
189 return self._sock
190
191 addr = (self.host, self.port)
192 s = None
193
194 if self.pool is not None:
195 s = self.pool.get(addr)
196
197 if not s:
198
199 if self.uri.scheme == "https":
200 s = sock.connect(addr, self.timeout, **self.ssl_args)
201 else:
202 s = sock.connect(addr, self.timeout)
203 return s
204
209
211 if not self.pool:
212 return
213 self.pool.put(address, self._sock)
214
216 """ parse url and get host/port"""
217 self.uri = urlparse.urlparse(url)
218 if self.uri.scheme not in ('http', 'https'):
219 raise InvalidUrl("None valid url")
220
221 host, port = util.parse_netloc(self.uri)
222 self.host = host
223 self.port = port
224
225 - def set_body(self, body, content_type=None, content_length=None,
226 chunked=False):
227 """ set HTTP body and manage form if needed """
228 if not body:
229 self.headers.append(('Content-Type', content_type))
230 if self.method in ('POST', 'PUT'):
231 self.headers.append(("Content-Length", "0"))
232 return
233
234
235 if isinstance(body, dict):
236 if content_type is not None and \
237 content_type.startswith("multipart/form-data"):
238 type_, opts = cgi.parse_header(content_type)
239 boundary = opts.get('boundary', uuid.uuid4().hex)
240 body, self.headers = multipart_form_encode(body,
241 self.headers, boundary)
242 else:
243 content_type = "application/x-www-form-urlencoded; charset=utf-8"
244 body = form_encode(body)
245 elif hasattr(body, "boundary"):
246 content_type = "multipart/form-data; boundary=%s" % body.boundary
247 content_length = body.get_size()
248
249 if not content_type:
250 content_type = 'application/octet-stream'
251 if hasattr(body, 'name'):
252 content_type = mimetypes.guess_type(body.name)[0]
253
254 if not content_length:
255 if hasattr(body, 'fileno'):
256 try:
257 body.flush()
258 except IOError:
259 pass
260 content_length = str(os.fstat(body.fileno())[6])
261 elif hasattr(body, 'getvalue'):
262 try:
263 content_length = str(len(body.getvalue()))
264 except AttributeError:
265 pass
266 elif isinstance(body, types.StringTypes):
267 body = util.to_bytestring(body)
268 content_length = len(body)
269
270 if content_length:
271 self.headers.append(("Content-Length", content_length))
272 self.headers.append(('Content-Type', content_type))
273
274 elif not chunked:
275 raise RequestError("Can't determine content length and" +
276 "Transfer-Encoding header is not chunked")
277
278 self.body = body
279
280
281 - def request(self, url, method='GET', body=None, headers=None):
282 """ make effective request
283
284 :param url: str, url string
285 :param method: str, by default GET. http verbs
286 :param body: the body, could be a string, an iterator or a file-like object
287 :param headers: dict or list of tupple, http headers
288 """
289 self._sock = None
290 self.url = url
291 self.final_url = url
292 self.parse_url(url)
293 self.method = method.upper()
294 self.headers = []
295
296
297 headers = headers or []
298 if isinstance(headers, dict):
299 headers = list(headers.items())
300
301 ua = USER_AGENT
302 content_length = None
303 accept_encoding = 'identity'
304 chunked = False
305 content_type = None
306
307
308 try:
309 host = self.uri.netloc.encode('ascii')
310 except UnicodeEncodeError:
311 host = self.uri.netloc.encode('idna')
312
313
314 for name, value in headers:
315 name = name.title()
316 if name == "User-Agent":
317 ua = value
318 elif name == "Content-Type":
319 content_type = value
320 elif name == "Content-Length":
321 content_length = str(value)
322 elif name == "Accept-Encoding":
323 accept_encoding = 'identity'
324 elif name == "Host":
325 host = value
326 elif name == "Transfer-Encoding":
327 if value.lower() == "chunked":
328 chunked = True
329 self.headers.append((name, value))
330 else:
331 if not isinstance(value, types.StringTypes):
332 value = str(value)
333 self.headers.append((name, value))
334
335 self.set_body(body, content_type=content_type,
336 content_length=content_length, chunked=chunked)
337
338 self.ua = ua
339 self.chunked = chunked
340 self.host_hdr = host
341 self.accept_encoding = accept_encoding
342
343
344 return self.do_send()
345
347
348 if self.version == (1,1):
349 httpver = "HTTP/1.1"
350 else:
351 httpver = "HTTP/1.0"
352
353
354 path = self.uri.path or "/"
355 req_path = urlparse.urlunparse(('','', path, '',
356 self.uri.query, self.uri.fragment))
357
358
359 req_headers = [
360 "%s %s %s\r\n" % (self.method, req_path, httpver),
361 "Host: %s\r\n" % self.host_hdr,
362 "User-Agent: %s\r\n" % self.ua,
363 "Accept-Encoding: %s\r\n" % self.accept_encoding
364 ]
365 req_headers.extend(["%s: %s\r\n" % (k, v) for k, v in self.headers])
366 req_headers.append('\r\n')
367 return req_headers
368
370 tries = 2
371 while True:
372 try:
373
374 self._sock = self.make_connection()
375
376
377 self.filters.apply("on_request", self)
378
379
380 self.req_headers = req_headers = self._req_headers()
381
382
383 log.info('Start request: %s %s', self.method, self.url)
384 log.debug("Request headers: [%s]", req_headers)
385
386 self._sock.sendall("".join(req_headers))
387
388 if self.body is not None:
389 if hasattr(self.body, 'read'):
390 if hasattr(self.body, 'seek'): self.body.seek(0)
391 sock.sendfile(self._sock, self.body, self.chunked)
392 elif isinstance(self.body, types.StringTypes):
393 sock.send(self._sock, self.body, self.chunked)
394 else:
395 sock.sendlines(self._sock, self.body, self.chunked)
396
397 if self.chunked:
398 sock.send_chunk(self._sock, "")
399
400 return self.start_response()
401 except socket.gaierror, e:
402 self.clean_connections()
403 raise
404 except socket.error, e:
405 if e[0] not in (errno.EAGAIN, errno.ECONNABORTED, errno.EPIPE,
406 errno.ECONNREFUSED) or tries <= 0:
407 self.clean_connections()
408 raise
409 if e[0] == errno.EPIPE:
410 log.debug("Got EPIPE")
411 self.clean_connections()
412 except:
413 if tries <= 0:
414 raise
415
416 self.clean_connections()
417 time.sleep(0.2)
418 tries -= 1
419
420
422 """ follow redirections if needed"""
423 if self.nb_redirections <= 0:
424 raise RedirectLimit("Redirection limit is reached")
425
426 if not location:
427 raise RequestError('no Location header')
428
429 new_uri = urlparse.urlparse(location)
430 if not new_uri.netloc:
431 absolute_uri = "%s://%s" % (self.uri.scheme, self.uri.netloc)
432 location = urlparse.urljoin(absolute_uri, location)
433
434 log.debug("Redirect to %s" % location)
435
436 self.final_url = location
437 response.body.read()
438 self.nb_redirections -= 1
439 sock.close(self._sock)
440 return self.request(location, self.method, self.body, self.headers)
441
443 """
444 Get headers, set Body object and return HttpResponse
445 """
446
447 parser = http.ResponseParser(self._sock,
448 release_source = lambda:self.release_connection(
449 self.uri.netloc, self._sock))
450 resp = parser.next()
451
452 log.debug("Start response: %s", resp.status)
453 log.debug("Response headers: [%s]", resp.headers)
454
455 location = None
456 for hdr_name, hdr_value in resp.headers:
457 if hdr_name.lower() == "location":
458 location = hdr_value
459 break
460
461 if self.follow_redirect:
462 if resp.status_int in (301, 302, 307):
463 if self.method in ('GET', 'HEAD') or \
464 self.force_follow_redirect:
465 if self.method not in ('GET', 'HEAD') and \
466 hasattr(self.body, 'seek'):
467 self.body.seek(0)
468 return self.do_redirect(resp, location)
469 elif resp.status_int == 303 and self.method in ('GET',
470 'HEAD'):
471
472
473 return self.do_redirect(resp, location)
474
475
476
477 self.filters.apply("on_response", self)
478
479 self.final_url = location or self.final_url
480 log.debug("Return response: %s" % self.final_url)
481 if self.method == "HEAD":
482 resp.body = StringIO()
483
484 return self.response_class(resp, self.final_url)
485