1 """An extensible library for opening URLs using a variety of protocols
3 The simplest way to use this module is to call the urlopen function,
4 which accepts a string containing a URL or a Request object (described
5 below). It opens the URL and returns the results as file-like
6 object; the returned object has some extra methods described below.
8 The OpenerDirector manages a collection of Handler objects that do
9 all the actual work. Each Handler implements a particular protocol or
10 option. The OpenerDirector is a composite object that invokes the
11 Handlers needed to open the requested URL. For example, the
12 HTTPHandler performs HTTP GET and POST requests and deals with
13 non-error returns. The HTTPRedirectHandler automatically deals with
14 HTTP 301, 302, 303 and 307 redirect errors, and the HTTPDigestAuthHandler
15 deals with digest authentication.
17 urlopen(url, data=None) -- basic usage is the same as original
18 urllib. pass the url and optionally data to post to an HTTP URL, and
19 get a file-like object back. One difference is that you can also pass
20 a Request instance instead of URL. Raises a URLError (subclass of
21 IOError); for HTTP errors, raises an HTTPError, which can also be
22 treated as a valid response.
24 build_opener -- function that creates a new OpenerDirector instance.
25 will install the default handlers. accepts one or more Handlers as
26 arguments, either instances or Handler classes that it will
27 instantiate. if one of the argument is a subclass of the default
28 handler, the argument will be installed instead of the default.
30 install_opener -- installs a new opener as the default opener.
35 Request -- an object that encapsulates the state of a request. the
36 state can be a simple as the URL. it can also include extra HTTP
37 headers, e.g. a User-Agent.
42 URLError-- a subclass of IOError, individual protocols have their own
45 HTTPError-- also a valid HTTP response, so you can treat an HTTP error
46 as an exceptional event or valid response
49 BaseHandler and parent
50 _call_chain conventions
56 # set up authentication info
57 authinfo = urllib2.HTTPBasicAuthHandler()
58 authinfo.add_password('realm', 'host', 'username', 'password')
60 proxy_support = urllib2.ProxyHandler({"http" : "http://ahad-haam:3128"})
62 # build a new opener that adds authentication and caching FTP handlers
63 opener = urllib2.build_opener(proxy_support, authinfo, urllib2.CacheFTPHandler)
66 urllib2.install_opener(opener)
68 f = urllib2.urlopen('http://www.python.org/')
74 # If an authentication error handler that tries to perform
75 # authentication for some reason but fails, how should the error be
76 # signalled? The client needs to know the HTTP error code. But if
77 # the handler knows that the problem was, e.g., that it didn't know
78 # that hash algo that requested in the challenge, it would be good to
79 # pass that information along to the client, too.
80 # ftp errors aren't handled cleanly
81 # check digest against correct (i.e. non-apache) implementation
83 # Possible extensions:
84 # complex proxies XXX not sure what exactly was meant by this
85 # abstract factory for opener
102 from cStringIO import StringIO
104 from StringIO import StringIO
106 from urllib import (unwrap, unquote, splittype, splithost, quote,
107 addinfourl, splitport, splitgophertype, splitquery,
108 splitattr, ftpwrapper, noheaders, splituser, splitpasswd, splitvalue)
110 # support for FileHandler, proxies via environment variables
111 from urllib import localhost, url2pathname, getproxies
113 # used in User-Agent header sent
114 __version__ = sys.version[:3]
117 def urlopen(url, data=None):
120 _opener = build_opener()
121 return _opener.open(url, data)
123 def install_opener(opener):
127 # do these error classes make sense?
128 # make sure all of the IOError stuff is overridden. we just want to be
131 class URLError(IOError):
132 # URLError is a sub-type of IOError, but it doesn't share any of
133 # the implementation. need to override __init__ and __str__.
134 # It sets self.args for compatibility with other EnvironmentError
135 # subclasses, but args doesn't have the typical format with errno in
136 # slot 0 and strerror in slot 1. This may be better than nothing.
137 def __init__(self, reason):
142 return '<urlopen error %s>' % self.reason
144 class HTTPError(URLError, addinfourl):
145 """Raised when HTTP error occurs, but also acts like non-error return"""
146 __super_init = addinfourl.__init__
148 def __init__(self, url, code, msg, hdrs, fp):
154 # The addinfourl classes depend on fp being a valid file
155 # object. In some cases, the HTTPError may not have a valid
156 # file object. If this happens, the simplest workaround is to
157 # not initialize the base classes.
159 self.__super_init(fp, hdrs, url)
162 return 'HTTP Error %s: %s' % (self.code, self.msg)
164 class GopherError(URLError):
167 # copied from cookielib.py
168 _cut_port_re = re.compile(r":\d+$")
169 def request_host(request):
170 """Return request-host, as defined by RFC 2965.
172 Variation from RFC: returned value is lowercased, for convenient
176 url = request.get_full_url()
177 host = urlparse.urlparse(url)[1]
179 host = request.get_header("Host", "")
181 # remove port, if present
182 host = _cut_port_re.sub("", host, 1)
187 def __init__(self, url, data=None, headers={},
188 origin_req_host=None, unverifiable=False):
189 # unwrap('<URL:type://host/path>') --> 'type://host/path'
190 self.__original = unwrap(url)
192 # self.__r_type is what's left after doing the splittype
197 for key, value in headers.items():
198 self.add_header(key, value)
199 self.unredirected_hdrs = {}
200 if origin_req_host is None:
201 origin_req_host = request_host(self)
202 self.origin_req_host = origin_req_host
203 self.unverifiable = unverifiable
205 def __getattr__(self, attr):
206 # XXX this is a fallback mechanism to guard against these
207 # methods getting called in a non-standard order. this may be
208 # too complicated and/or unnecessary.
209 # XXX should the __r_XXX attributes be public?
210 if attr[:12] == '_Request__r_':
212 if hasattr(Request, 'get_' + name):
213 getattr(self, 'get_' + name)()
214 return getattr(self, attr)
215 raise AttributeError, attr
217 def get_method(self):
223 # XXX these helper methods are lame
225 def add_data(self, data):
229 return self.data is not None
234 def get_full_url(self):
235 return self.__original
238 if self.type is None:
239 self.type, self.__r_type = splittype(self.__original)
240 if self.type is None:
241 raise ValueError, "unknown url type: %s" % self.__original
245 if self.host is None:
246 self.host, self.__r_host = splithost(self.__r_type)
248 self.host = unquote(self.host)
251 def get_selector(self):
254 def set_proxy(self, host, type):
255 self.host, self.type = host, type
256 self.__r_host = self.__original
258 def get_origin_req_host(self):
259 return self.origin_req_host
261 def is_unverifiable(self):
262 return self.unverifiable
264 def add_header(self, key, val):
265 # useful for something like authentication
266 self.headers[key.capitalize()] = val
268 def add_unredirected_header(self, key, val):
269 # will not be added to a redirected request
270 self.unredirected_hdrs[key.capitalize()] = val
272 def has_header(self, header_name):
273 return (header_name in self.headers or
274 header_name in self.unredirected_hdrs)
276 def get_header(self, header_name, default=None):
277 return self.headers.get(
279 self.unredirected_hdrs.get(header_name, default))
281 def header_items(self):
282 hdrs = self.unredirected_hdrs.copy()
283 hdrs.update(self.headers)
286 class OpenerDirector:
288 client_version = "Python-urllib/%s" % __version__
289 self.addheaders = [('User-agent', client_version)]
290 # manage the individual handlers
292 self.handle_open = {}
293 self.handle_error = {}
294 self.process_response = {}
295 self.process_request = {}
297 def add_handler(self, handler):
299 for meth in dir(handler):
300 if meth in ["redirect_request", "do_open", "proxy_open"]:
301 # oops, coincidental match
306 condition = meth[i+1:]
308 if condition.startswith("error"):
309 j = condition.find("_") + i + 1
315 lookup = self.handle_error.get(protocol, {})
316 self.handle_error[protocol] = lookup
317 elif condition == "open":
319 lookup = self.handle_open
320 elif condition == "response":
322 lookup = self.process_response
323 elif condition == "request":
325 lookup = self.process_request
329 handlers = lookup.setdefault(kind, [])
331 bisect.insort(handlers, handler)
333 handlers.append(handler)
337 # XXX why does self.handlers need to be sorted?
338 bisect.insort(self.handlers, handler)
339 handler.add_parent(self)
342 # Only exists for backwards compatibility.
345 def _call_chain(self, chain, kind, meth_name, *args):
346 # Handlers raise an exception if no one else should try to handle
347 # the request, or return None if they can't but another handler
348 # could. Otherwise, they return the response.
349 handlers = chain.get(kind, ())
350 for handler in handlers:
351 func = getattr(handler, meth_name)
354 if result is not None:
357 def open(self, fullurl, data=None):
358 # accept a URL or a Request object
359 if isinstance(fullurl, basestring):
360 req = Request(fullurl, data)
366 protocol = req.get_type()
368 # pre-process request
369 meth_name = protocol+"_request"
370 for processor in self.process_request.get(protocol, []):
371 meth = getattr(processor, meth_name)
374 response = self._open(req, data)
376 # post-process response
377 meth_name = protocol+"_response"
378 for processor in self.process_response.get(protocol, []):
379 meth = getattr(processor, meth_name)
380 response = meth(req, response)
384 def _open(self, req, data=None):
385 result = self._call_chain(self.handle_open, 'default',
390 protocol = req.get_type()
391 result = self._call_chain(self.handle_open, protocol, protocol +
396 return self._call_chain(self.handle_open, 'unknown',
399 def error(self, proto, *args):
400 if proto in ('http', 'https'):
401 # XXX http[s] protocols are special-cased
402 dict = self.handle_error['http'] # https is not different than http
403 proto = args[2] # YUCK!
404 meth_name = 'http_error_%s' % proto
408 dict = self.handle_error
409 meth_name = proto + '_error'
411 args = (dict, proto, meth_name) + args
412 result = self._call_chain(*args)
417 args = (dict, 'default', 'http_error_default') + orig_args
418 return self._call_chain(*args)
420 # XXX probably also want an abstract factory that knows when it makes
421 # sense to skip a superclass in favor of a subclass and when it might
422 # make sense to include both
424 def build_opener(*handlers):
425 """Create an opener object from a list of handlers.
427 The opener will use several default handlers, including support
430 If any of the handlers passed as arguments are subclasses of the
431 default handlers, the default handlers will not be used.
435 return isinstance(obj, types.ClassType) or hasattr(obj, "__bases__")
437 opener = OpenerDirector()
438 default_classes = [ProxyHandler, UnknownHandler, HTTPHandler,
439 HTTPDefaultErrorHandler, HTTPRedirectHandler,
440 FTPHandler, FileHandler, HTTPErrorProcessor]
441 if hasattr(httplib, 'HTTPS'):
442 default_classes.append(HTTPSHandler)
444 for klass in default_classes:
445 for check in handlers:
447 if issubclass(check, klass):
449 elif isinstance(check, klass):
452 default_classes.remove(klass)
454 for klass in default_classes:
455 opener.add_handler(klass())
460 opener.add_handler(h)
466 def add_parent(self, parent):
470 # Only exists for backwards compatibility
473 def __lt__(self, other):
474 if not hasattr(other, "handler_order"):
475 # Try to preserve the old behavior of having custom classes
476 # inserted after default ones (works only for custom user
477 # classes which are not aware of handler_order).
479 return self.handler_order < other.handler_order
482 class HTTPErrorProcessor(BaseHandler):
483 """Process HTTP error responses."""
484 handler_order = 1000 # after all other processing
486 def http_response(self, request, response):
487 code, msg, hdrs = response.code, response.msg, response.info()
489 if code not in (200, 206):
490 response = self.parent.error(
491 'http', request, response, code, msg, hdrs)
495 https_response = http_response
497 class HTTPDefaultErrorHandler(BaseHandler):
498 def http_error_default(self, req, fp, code, msg, hdrs):
499 raise HTTPError(req.get_full_url(), code, msg, hdrs, fp)
501 class HTTPRedirectHandler(BaseHandler):
502 # maximum number of redirections to any single URL
503 # this is needed because of the state that cookies introduce
505 # maximum total number of redirections (regardless of URL) before
506 # assuming we're in a loop
507 max_redirections = 10
509 def redirect_request(self, req, fp, code, msg, headers, newurl):
510 """Return a Request or None in response to a redirect.
512 This is called by the http_error_30x methods when a
513 redirection response is received. If a redirection should
514 take place, return a new Request to allow http_error_30x to
515 perform the redirect. Otherwise, raise HTTPError if no-one
516 else should try to handle this url. Return None if you can't
517 but another Handler might.
520 if (code in (301, 302, 303, 307) and m in ("GET", "HEAD")
521 or code in (301, 302, 303) and m == "POST"):
522 # Strictly (according to RFC 2616), 301 or 302 in response
523 # to a POST MUST NOT cause a redirection without confirmation
524 # from the user (of urllib2, in this case). In practice,
525 # essentially all clients do redirect in this case, so we
527 # be conciliant with URIs containing a space
528 newurl = newurl.replace(' ', '%20')
529 return Request(newurl,
531 origin_req_host=req.get_origin_req_host(),
534 raise HTTPError(req.get_full_url(), code, msg, headers, fp)
536 # Implementation note: To avoid the server sending us into an
537 # infinite loop, the request object needs to track what URLs we
538 # have already seen. Do this by adding a handler-specific
539 # attribute to the Request object.
540 def http_error_302(self, req, fp, code, msg, headers):
541 # Some servers (incorrectly) return multiple Location headers
542 # (so probably same goes for URI). Use first header.
543 if 'location' in headers:
544 newurl = headers.getheaders('location')[0]
545 elif 'uri' in headers:
546 newurl = headers.getheaders('uri')[0]
549 newurl = urlparse.urljoin(req.get_full_url(), newurl)
551 # XXX Probably want to forget about the state of the current
552 # request, although that might interact poorly with other
553 # handlers that also use handler-specific request attributes
554 new = self.redirect_request(req, fp, code, msg, headers, newurl)
559 # .redirect_dict has a key url if url was previously visited.
560 if hasattr(req, 'redirect_dict'):
561 visited = new.redirect_dict = req.redirect_dict
562 if (visited.get(newurl, 0) >= self.max_repeats or
563 len(visited) >= self.max_redirections):
564 raise HTTPError(req.get_full_url(), code,
565 self.inf_msg + msg, headers, fp)
567 visited = new.redirect_dict = req.redirect_dict = {}
568 visited[newurl] = visited.get(newurl, 0) + 1
570 # Don't close the fp until we are sure that we won't use it
575 return self.parent.open(new)
577 http_error_301 = http_error_303 = http_error_307 = http_error_302
579 inf_msg = "The HTTP server returned a redirect error that would " \
580 "lead to an infinite loop.\n" \
581 "The last 30x error message was:\n"
584 def _parse_proxy(proxy):
585 """Return (scheme, user, password, host/port) given a URL or an authority.
587 If a URL is supplied, it must have an authority (host:port) component.
588 According to RFC 3986, having an authority component means the URL must
589 have two slashes after the scheme:
591 >>> _parse_proxy('file:/ftp.example.com/')
592 Traceback (most recent call last):
593 ValueError: proxy URL with no authority: 'file:/ftp.example.com/'
595 The first three items of the returned tuple may be None.
597 Examples of authority parsing:
599 >>> _parse_proxy('proxy.example.com')
600 (None, None, None, 'proxy.example.com')
601 >>> _parse_proxy('proxy.example.com:3128')
602 (None, None, None, 'proxy.example.com:3128')
604 The authority component may optionally include userinfo (assumed to be
607 >>> _parse_proxy('joe:password@proxy.example.com')
608 (None, 'joe', 'password', 'proxy.example.com')
609 >>> _parse_proxy('joe:password@proxy.example.com:3128')
610 (None, 'joe', 'password', 'proxy.example.com:3128')
612 Same examples, but with URLs instead:
614 >>> _parse_proxy('http://proxy.example.com/')
615 ('http', None, None, 'proxy.example.com')
616 >>> _parse_proxy('http://proxy.example.com:3128/')
617 ('http', None, None, 'proxy.example.com:3128')
618 >>> _parse_proxy('http://joe:password@proxy.example.com/')
619 ('http', 'joe', 'password', 'proxy.example.com')
620 >>> _parse_proxy('http://joe:password@proxy.example.com:3128')
621 ('http', 'joe', 'password', 'proxy.example.com:3128')
623 Everything after the authority is ignored:
625 >>> _parse_proxy('ftp://joe:password@proxy.example.com/rubbish:3128')
626 ('ftp', 'joe', 'password', 'proxy.example.com')
628 Test for no trailing '/' case:
630 >>> _parse_proxy('http://joe:password@proxy.example.com')
631 ('http', 'joe', 'password', 'proxy.example.com')
634 scheme, r_scheme = splittype(proxy)
635 if not r_scheme.startswith("/"):
641 if not r_scheme.startswith("//"):
642 raise ValueError("proxy URL with no authority: %r" % proxy)
643 # We have an authority, so for RFC 3986-compliant URLs (by ss 3.
644 # and 3.3.), path is empty or starts with '/'
645 end = r_scheme.find("/", 2)
648 authority = r_scheme[2:end]
649 userinfo, hostport = splituser(authority)
650 if userinfo is not None:
651 user, password = splitpasswd(userinfo)
653 user = password = None
654 return scheme, user, password, hostport
656 class ProxyHandler(BaseHandler):
657 # Proxies must be in front
660 def __init__(self, proxies=None):
662 proxies = getproxies()
663 assert hasattr(proxies, 'has_key'), "proxies must be a mapping"
664 self.proxies = proxies
665 for type, url in proxies.items():
666 setattr(self, '%s_open' % type,
667 lambda r, proxy=url, type=type, meth=self.proxy_open: \
668 meth(r, proxy, type))
670 def proxy_open(self, req, proxy, type):
671 orig_type = req.get_type()
672 proxy_type, user, password, hostport = _parse_proxy(proxy)
673 if proxy_type is None:
674 proxy_type = orig_type
675 if user and password:
676 user_pass = '%s:%s' % (unquote(user), unquote(password))
677 creds = base64.b64encode(user_pass).strip()
678 req.add_header('Proxy-authorization', 'Basic ' + creds)
679 hostport = unquote(hostport)
680 req.set_proxy(hostport, proxy_type)
681 if orig_type == proxy_type:
682 # let other handlers take care of it
685 # need to start over, because the other handlers don't
686 # grok the proxy's URL type
687 # e.g. if we have a constructor arg proxies like so:
688 # {'http': 'ftp://proxy.example.com'}, we may end up turning
689 # a request for http://acme.example.com/a into one for
690 # ftp://proxy.example.com/a
691 return self.parent.open(req)
693 class HTTPPasswordMgr:
698 def add_password(self, realm, uri, user, passwd):
699 # uri could be a single URI or a sequence
700 if isinstance(uri, basestring):
702 if not realm in self.passwd:
703 self.passwd[realm] = {}
704 for default_port in True, False:
706 [self.reduce_uri(u, default_port) for u in uri])
707 self.passwd[realm][reduced_uri] = (user, passwd)
709 def find_user_password(self, realm, authuri):
710 domains = self.passwd.get(realm, {})
711 for default_port in True, False:
712 reduced_authuri = self.reduce_uri(authuri, default_port)
713 for uris, authinfo in domains.iteritems():
715 if self.is_suburi(uri, reduced_authuri):
719 def reduce_uri(self, uri, default_port=True):
720 """Accept authority or URI and extract only the authority and path."""
721 # note HTTP URLs do not have a userinfo component
722 parts = urlparse.urlsplit(uri)
727 path = parts[2] or '/'
733 host, port = splitport(authority)
734 if default_port and port is None and scheme is not None:
738 if dport is not None:
739 authority = "%s:%d" % (host, dport)
740 return authority, path
742 def is_suburi(self, base, test):
743 """Check if test is below base in a URI tree
745 Both args must be URIs in reduced form.
749 if base[0] != test[0]:
751 common = posixpath.commonprefix((base[1], test[1]))
752 if len(common) == len(base[1]):
757 class HTTPPasswordMgrWithDefaultRealm(HTTPPasswordMgr):
759 def find_user_password(self, realm, authuri):
760 user, password = HTTPPasswordMgr.find_user_password(self, realm,
763 return user, password
764 return HTTPPasswordMgr.find_user_password(self, None, authuri)
767 class AbstractBasicAuthHandler:
769 # XXX this allows for multiple auth-schemes, but will stupidly pick
770 # the last one with a realm specified.
772 rx = re.compile('(?:.*,)*[ \t]*([^ \t]+)[ \t]+realm="([^"]*)"', re.I)
774 # XXX could pre-emptively send auth info already accepted (RFC 2617,
775 # end of section 2, and section 1.2 immediately after "credentials"
778 def __init__(self, password_mgr=None):
779 if password_mgr is None:
780 password_mgr = HTTPPasswordMgr()
781 self.passwd = password_mgr
782 self.add_password = self.passwd.add_password
784 def http_error_auth_reqed(self, authreq, host, req, headers):
785 # host may be an authority (without userinfo) or a URL with an
787 # XXX could be multiple headers
788 authreq = headers.get(authreq, None)
790 mo = AbstractBasicAuthHandler.rx.search(authreq)
792 scheme, realm = mo.groups()
793 if scheme.lower() == 'basic':
794 return self.retry_http_basic_auth(host, req, realm)
796 def retry_http_basic_auth(self, host, req, realm):
797 user, pw = self.passwd.find_user_password(realm, host)
799 raw = "%s:%s" % (user, pw)
800 auth = 'Basic %s' % base64.b64encode(raw).strip()
801 if req.headers.get(self.auth_header, None) == auth:
803 req.add_header(self.auth_header, auth)
804 return self.parent.open(req)
809 class HTTPBasicAuthHandler(AbstractBasicAuthHandler, BaseHandler):
811 auth_header = 'Authorization'
813 def http_error_401(self, req, fp, code, msg, headers):
814 url = req.get_full_url()
815 return self.http_error_auth_reqed('www-authenticate',
819 class ProxyBasicAuthHandler(AbstractBasicAuthHandler, BaseHandler):
821 auth_header = 'Proxy-authorization'
823 def http_error_407(self, req, fp, code, msg, headers):
824 # http_error_auth_reqed requires that there is no userinfo component in
825 # authority. Assume there isn't one, since urllib2 does not (and
826 # should not, RFC 3986 s. 3.2.1) support requests for URLs containing
828 authority = req.get_host()
829 return self.http_error_auth_reqed('proxy-authenticate',
830 authority, req, headers)
834 """Return n random bytes."""
835 # Use /dev/urandom if it is available. Fall back to random module
836 # if not. It might be worthwhile to extend this function to use
837 # other platform-specific mechanisms for getting random bytes.
838 if os.path.exists("/dev/urandom"):
839 f = open("/dev/urandom")
844 L = [chr(random.randrange(0, 256)) for i in range(n)]
847 class AbstractDigestAuthHandler:
848 # Digest authentication is specified in RFC 2617.
850 # XXX The client does not inspect the Authentication-Info header
851 # in a successful response.
853 # XXX It should be possible to test this implementation against
854 # a mock server that just generates a static set of challenges.
856 # XXX qop="auth-int" supports is shaky
858 def __init__(self, passwd=None):
860 passwd = HTTPPasswordMgr()
862 self.add_password = self.passwd.add_password
866 def reset_retry_count(self):
869 def http_error_auth_reqed(self, auth_header, host, req, headers):
870 authreq = headers.get(auth_header, None)
872 # Don't fail endlessly - if we failed once, we'll probably
873 # fail a second time. Hm. Unless the Password Manager is
874 # prompting for the information. Crap. This isn't great
875 # but it's better than the current 'repeat until recursion
876 # depth exceeded' approach <wink>
877 raise HTTPError(req.get_full_url(), 401, "digest auth failed",
882 scheme = authreq.split()[0]
883 if scheme.lower() == 'digest':
884 return self.retry_http_digest_auth(req, authreq)
886 def retry_http_digest_auth(self, req, auth):
887 token, challenge = auth.split(' ', 1)
888 chal = parse_keqv_list(parse_http_list(challenge))
889 auth = self.get_authorization(req, chal)
891 auth_val = 'Digest %s' % auth
892 if req.headers.get(self.auth_header, None) == auth_val:
894 req.add_unredirected_header(self.auth_header, auth_val)
895 resp = self.parent.open(req)
898 def get_cnonce(self, nonce):
899 # The cnonce-value is an opaque
900 # quoted string value provided by the client and used by both client
901 # and server to avoid chosen plaintext attacks, to provide mutual
902 # authentication, and to provide some message integrity protection.
903 # This isn't a fabulous effort, but it's probably Good Enough.
904 dig = hashlib.sha1("%s:%s:%s:%s" % (self.nonce_count, nonce, time.ctime(),
905 randombytes(8))).hexdigest()
908 def get_authorization(self, req, chal):
910 realm = chal['realm']
911 nonce = chal['nonce']
912 qop = chal.get('qop')
913 algorithm = chal.get('algorithm', 'MD5')
914 # mod_digest doesn't send an opaque, even though it isn't
915 # supposed to be optional
916 opaque = chal.get('opaque', None)
920 H, KD = self.get_algorithm_impls(algorithm)
924 user, pw = self.passwd.find_user_password(realm, req.get_full_url())
928 # XXX not implemented yet
930 entdig = self.get_entity_digest(req.get_data(), chal)
934 A1 = "%s:%s:%s" % (user, realm, pw)
935 A2 = "%s:%s" % (req.get_method(),
936 # XXX selector: what about proxies and full urls
939 self.nonce_count += 1
940 ncvalue = '%08x' % self.nonce_count
941 cnonce = self.get_cnonce(nonce)
942 noncebit = "%s:%s:%s:%s:%s" % (nonce, ncvalue, cnonce, qop, H(A2))
943 respdig = KD(H(A1), noncebit)
945 respdig = KD(H(A1), "%s:%s" % (nonce, H(A2)))
947 # XXX handle auth-int.
950 # XXX should the partial digests be encoded too?
952 base = 'username="%s", realm="%s", nonce="%s", uri="%s", ' \
953 'response="%s"' % (user, realm, nonce, req.get_selector(),
956 base += ', opaque="%s"' % opaque
958 base += ', digest="%s"' % entdig
959 base += ', algorithm="%s"' % algorithm
961 base += ', qop=auth, nc=%s, cnonce="%s"' % (ncvalue, cnonce)
964 def get_algorithm_impls(self, algorithm):
965 # lambdas assume digest modules are imported at the top level
966 if algorithm == 'MD5':
967 H = lambda x: hashlib.md5(x).hexdigest()
968 elif algorithm == 'SHA':
969 H = lambda x: hashlib.sha1(x).hexdigest()
971 KD = lambda s, d: H("%s:%s" % (s, d))
974 def get_entity_digest(self, data, chal):
975 # XXX not implemented yet
979 class HTTPDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler):
980 """An authentication protocol defined by RFC 2069
982 Digest authentication improves on basic authentication because it
983 does not transmit passwords in the clear.
986 auth_header = 'Authorization'
987 handler_order = 490 # before Basic auth
989 def http_error_401(self, req, fp, code, msg, headers):
990 host = urlparse.urlparse(req.get_full_url())[1]
991 retry = self.http_error_auth_reqed('www-authenticate',
993 self.reset_retry_count()
997 class ProxyDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler):
999 auth_header = 'Proxy-Authorization'
1000 handler_order = 490 # before Basic auth
1002 def http_error_407(self, req, fp, code, msg, headers):
1003 host = req.get_host()
1004 retry = self.http_error_auth_reqed('proxy-authenticate',
1006 self.reset_retry_count()
1009 class AbstractHTTPHandler(BaseHandler):
1011 def __init__(self, debuglevel=0):
1012 self._debuglevel = debuglevel
1014 def set_http_debuglevel(self, level):
1015 self._debuglevel = level
1017 def do_request_(self, request):
1018 host = request.get_host()
1020 raise URLError('no host given')
1022 if request.has_data(): # POST
1023 data = request.get_data()
1024 if not request.has_header('Content-type'):
1025 request.add_unredirected_header(
1027 'application/x-www-form-urlencoded')
1028 if not request.has_header('Content-length'):
1029 request.add_unredirected_header(
1030 'Content-length', '%d' % len(data))
1032 scheme, sel = splittype(request.get_selector())
1033 sel_host, sel_path = splithost(sel)
1034 if not request.has_header('Host'):
1035 request.add_unredirected_header('Host', sel_host or host)
1036 for name, value in self.parent.addheaders:
1037 name = name.capitalize()
1038 if not request.has_header(name):
1039 request.add_unredirected_header(name, value)
1043 def do_open(self, http_class, req):
1044 """Return an addinfourl object for the request, using http_class.
1046 http_class must implement the HTTPConnection API from httplib.
1047 The addinfourl return value is a file-like object. It also
1048 has methods and attributes including:
1049 - info(): return a mimetools.Message object for the headers
1050 - geturl(): return the original request URL
1051 - code: HTTP status code
1053 host = req.get_host()
1055 raise URLError('no host given')
1057 h = http_class(host) # will parse host:port
1058 h.set_debuglevel(self._debuglevel)
1060 headers = dict(req.headers)
1061 headers.update(req.unredirected_hdrs)
1062 # We want to make an HTTP/1.1 request, but the addinfourl
1063 # class isn't prepared to deal with a persistent connection.
1064 # It will try to read all remaining data from the socket,
1065 # which will block while the server waits for the next request.
1066 # So make sure the connection gets closed after the (only)
1068 headers["Connection"] = "close"
1070 (name.title(), val) for name, val in headers.items())
1072 h.request(req.get_method(), req.get_selector(), req.data, headers)
1074 except socket.error, err: # XXX what error?
1077 # Pick apart the HTTPResponse object to get the addinfourl
1078 # object initialized properly.
1080 # Wrap the HTTPResponse object in socket's file object adapter
1081 # for Windows. That adapter calls recv(), so delegate recv()
1082 # to read(). This weird wrapping allows the returned object to
1083 # have readline() and readlines() methods.
1085 # XXX It might be better to extract the read buffering code
1086 # out of socket._fileobject() and into a base class.
1089 fp = socket._fileobject(r, close=True)
1091 resp = addinfourl(fp, r.msg, req.get_full_url())
1092 resp.code = r.status
1097 class HTTPHandler(AbstractHTTPHandler):
1099 def http_open(self, req):
1100 return self.do_open(httplib.HTTPConnection, req)
1102 http_request = AbstractHTTPHandler.do_request_
1104 if hasattr(httplib, 'HTTPS'):
1105 class HTTPSHandler(AbstractHTTPHandler):
1107 def https_open(self, req):
1108 return self.do_open(httplib.HTTPSConnection, req)
1110 https_request = AbstractHTTPHandler.do_request_
1112 class HTTPCookieProcessor(BaseHandler):
1113 def __init__(self, cookiejar=None):
1115 if cookiejar is None:
1116 cookiejar = cookielib.CookieJar()
1117 self.cookiejar = cookiejar
1119 def http_request(self, request):
1120 self.cookiejar.add_cookie_header(request)
1123 def http_response(self, request, response):
1124 self.cookiejar.extract_cookies(response, request)
1127 https_request = http_request
1128 https_response = http_response
1130 class UnknownHandler(BaseHandler):
1131 def unknown_open(self, req):
1132 type = req.get_type()
1133 raise URLError('unknown url type: %s' % type)
1135 def parse_keqv_list(l):
1136 """Parse list of key=value strings where keys are not duplicated."""
1139 k, v = elt.split('=', 1)
1140 if v[0] == '"' and v[-1] == '"':
1145 def parse_http_list(s):
1146 """Parse lists as described by RFC 2068 Section 2.
1148 In particular, parse comma-separated lists where the elements of
1149 the list may include quoted-strings. A quoted-string could
1150 contain a comma. A non-quoted string could have quotes in the
1151 middle. Neither commas nor quotes count if they are escaped.
1152 Only double-quotes count, not single-quotes.
1157 escape = quote = False
1186 return [part.strip() for part in res]
1188 class FileHandler(BaseHandler):
1189 # Use local file or FTP depending on form of URL
1190 def file_open(self, req):
1191 url = req.get_selector()
1192 if url[:2] == '//' and url[2:3] != '/':
1194 return self.parent.open(req)
1196 return self.open_local_file(req)
1198 # names for the localhost
1200 def get_names(self):
1201 if FileHandler.names is None:
1203 FileHandler.names = (socket.gethostbyname('localhost'),
1204 socket.gethostbyname(socket.gethostname()))
1205 except socket.gaierror:
1206 FileHandler.names = (socket.gethostbyname('localhost'),)
1207 return FileHandler.names
1209 # not entirely sure what the rules are here
1210 def open_local_file(self, req):
1213 host = req.get_host()
1214 file = req.get_selector()
1215 localfile = url2pathname(file)
1216 stats = os.stat(localfile)
1217 size = stats.st_size
1218 modified = email.Utils.formatdate(stats.st_mtime, usegmt=True)
1219 mtype = mimetypes.guess_type(file)[0]
1220 headers = mimetools.Message(StringIO(
1221 'Content-type: %s\nContent-length: %d\nLast-modified: %s\n' %
1222 (mtype or 'text/plain', size, modified)))
1224 host, port = splitport(host)
1226 (not port and socket.gethostbyname(host) in self.get_names()):
1227 return addinfourl(open(localfile, 'rb'),
1228 headers, 'file:'+file)
1229 raise URLError('file not on local host')
1231 class FTPHandler(BaseHandler):
1232 def ftp_open(self, req):
1235 host = req.get_host()
1237 raise IOError, ('ftp error', 'no host given')
1238 host, port = splitport(host)
1240 port = ftplib.FTP_PORT
1244 # username/password handling
1245 user, host = splituser(host)
1247 user, passwd = splitpasswd(user)
1250 host = unquote(host)
1251 user = unquote(user or '')
1252 passwd = unquote(passwd or '')
1255 host = socket.gethostbyname(host)
1256 except socket.error, msg:
1258 path, attrs = splitattr(req.get_selector())
1259 dirs = path.split('/')
1260 dirs = map(unquote, dirs)
1261 dirs, file = dirs[:-1], dirs[-1]
1262 if dirs and not dirs[0]:
1265 fw = self.connect_ftp(user, passwd, host, port, dirs)
1266 type = file and 'I' or 'D'
1268 attr, value = splitvalue(attr)
1269 if attr.lower() == 'type' and \
1270 value in ('a', 'A', 'i', 'I', 'd', 'D'):
1271 type = value.upper()
1272 fp, retrlen = fw.retrfile(file, type)
1274 mtype = mimetypes.guess_type(req.get_full_url())[0]
1276 headers += "Content-type: %s\n" % mtype
1277 if retrlen is not None and retrlen >= 0:
1278 headers += "Content-length: %d\n" % retrlen
1279 sf = StringIO(headers)
1280 headers = mimetools.Message(sf)
1281 return addinfourl(fp, headers, req.get_full_url())
1282 except ftplib.all_errors, msg:
1283 raise IOError, ('ftp error', msg), sys.exc_info()[2]
1285 def connect_ftp(self, user, passwd, host, port, dirs):
1286 fw = ftpwrapper(user, passwd, host, port, dirs)
1287 ## fw.ftp.set_debuglevel(1)
1290 class CacheFTPHandler(FTPHandler):
1291 # XXX would be nice to have pluggable cache strategies
1292 # XXX this stuff is definitely not thread safe
1300 def setTimeout(self, t):
1303 def setMaxConns(self, m):
1306 def connect_ftp(self, user, passwd, host, port, dirs):
1307 key = user, host, port, '/'.join(dirs)
1308 if key in self.cache:
1309 self.timeout[key] = time.time() + self.delay
1311 self.cache[key] = ftpwrapper(user, passwd, host, port, dirs)
1312 self.timeout[key] = time.time() + self.delay
1314 return self.cache[key]
1316 def check_cache(self):
1317 # first check for old ones
1319 if self.soonest <= t:
1320 for k, v in self.timeout.items():
1322 self.cache[k].close()
1325 self.soonest = min(self.timeout.values())
1327 # then check the size
1328 if len(self.cache) == self.max_conns:
1329 for k, v in self.timeout.items():
1330 if v == self.soonest:
1334 self.soonest = min(self.timeout.values())
1336 class GopherHandler(BaseHandler):
1337 def gopher_open(self, req):
1338 # XXX can raise socket.error
1339 import gopherlib # this raises DeprecationWarning in 2.5
1340 host = req.get_host()
1342 raise GopherError('no host given')
1343 host = unquote(host)
1344 selector = req.get_selector()
1345 type, selector = splitgophertype(selector)
1346 selector, query = splitquery(selector)
1347 selector = unquote(selector)
1349 query = unquote(query)
1350 fp = gopherlib.send_query(selector, query, host)
1352 fp = gopherlib.send_selector(selector, host)
1353 return addinfourl(fp, noheaders(), req.get_full_url())