cpython/Lib/urllib2.py

1468 lines
50 KiB
Python
Raw Normal View History

"""An extensible library for opening URLs using a variety of protocols
The simplest way to use this module is to call the urlopen function,
which accepts a string containing a URL or a Request object (described
below). It opens the URL and returns the results as file-like
object; the returned object has some extra methods described below.
2002-10-11 14:27:55 -03:00
The OpenerDirector manages a collection of Handler objects that do
all the actual work. Each Handler implements a particular protocol or
option. The OpenerDirector is a composite object that invokes the
Handlers needed to open the requested URL. For example, the
HTTPHandler performs HTTP GET and POST requests and deals with
non-error returns. The HTTPRedirectHandler automatically deals with
HTTP 301, 302, 303 and 307 redirect errors, and the HTTPDigestAuthHandler
deals with digest authentication.
urlopen(url, data=None) -- Basic usage is the same as original
urllib. pass the url and optionally data to post to an HTTP URL, and
get a file-like object back. One difference is that you can also pass
a Request instance instead of URL. Raises a URLError (subclass of
IOError); for HTTP errors, raises an HTTPError, which can also be
treated as a valid response.
build_opener -- Function that creates a new OpenerDirector instance.
Will install the default handlers. Accepts one or more Handlers as
arguments, either instances or Handler classes that it will
instantiate. If one of the argument is a subclass of the default
handler, the argument will be installed instead of the default.
install_opener -- Installs a new opener as the default opener.
objects of interest:
OpenerDirector -- Sets up the User Agent as the Python-urllib client and manages
the Handler classes, while dealing with requests and responses.
Request -- An object that encapsulates the state of a request. The
state can be as simple as the URL. It can also include extra HTTP
headers, e.g. a User-Agent.
BaseHandler --
exceptions:
URLError -- A subclass of IOError, individual protocols have their own
specific subclass.
HTTPError -- Also a valid HTTP response, so you can treat an HTTP error
as an exceptional event or valid response.
internals:
BaseHandler and parent
_call_chain conventions
Example usage:
import urllib2
# set up authentication info
authinfo = urllib2.HTTPBasicAuthHandler()
authinfo.add_password(realm='PDQ Application',
uri='https://mahler:8092/site-updates.py',
user='klem',
passwd='geheim$parole')
proxy_support = urllib2.ProxyHandler({"http" : "http://ahad-haam:3128"})
# build a new opener that adds authentication and caching FTP handlers
opener = urllib2.build_opener(proxy_support, authinfo, urllib2.CacheFTPHandler)
# install it
urllib2.install_opener(opener)
f = urllib2.urlopen('http://www.python.org/')
"""
# XXX issues:
# If an authentication error handler that tries to perform
# authentication for some reason but fails, how should the error be
# signalled? The client needs to know the HTTP error code. But if
# the handler knows that the problem was, e.g., that it didn't know
# that hash algo that requested in the challenge, it would be good to
# pass that information along to the client, too.
# ftp errors aren't handled cleanly
# check digest against correct (i.e. non-apache) implementation
# Possible extensions:
# complex proxies XXX not sure what exactly was meant by this
# abstract factory for opener
import base64
import hashlib
import httplib
import mimetools
import os
import posixpath
import random
import re
import socket
import sys
import time
import urlparse
import bisect
import warnings
try:
from cStringIO import StringIO
except ImportError:
from StringIO import StringIO
from urllib import (unwrap, unquote, splittype, splithost, quote,
addinfourl, splitport, splittag, toBytes,
splitattr, ftpwrapper, splituser, splitpasswd, splitvalue)
# support for FileHandler, proxies via environment variables
from urllib import localhost, url2pathname, getproxies, proxy_bypass
# used in User-Agent header sent
__version__ = sys.version[:3]
_opener = None
def urlopen(url, data=None, timeout=socket._GLOBAL_DEFAULT_TIMEOUT):
global _opener
if _opener is None:
_opener = build_opener()
return _opener.open(url, data, timeout)
def install_opener(opener):
global _opener
_opener = opener
# do these error classes make sense?
# make sure all of the IOError stuff is overridden. we just want to be
# subtypes.
class URLError(IOError):
# URLError is a sub-type of IOError, but it doesn't share any of
# the implementation. need to override __init__ and __str__.
# It sets self.args for compatibility with other EnvironmentError
# subclasses, but args doesn't have the typical format with errno in
# slot 0 and strerror in slot 1. This may be better than nothing.
def __init__(self, reason):
self.args = reason,
2000-02-10 13:17:14 -04:00
self.reason = reason
def __str__(self):
2000-02-10 13:17:14 -04:00
return '<urlopen error %s>' % self.reason
class HTTPError(URLError, addinfourl):
"""Raised when HTTP error occurs, but also acts like non-error return"""
__super_init = addinfourl.__init__
def __init__(self, url, code, msg, hdrs, fp):
2000-02-10 13:17:14 -04:00
self.code = code
self.msg = msg
self.hdrs = hdrs
self.fp = fp
self.filename = url
# The addinfourl classes depend on fp being a valid file
# object. In some cases, the HTTPError may not have a valid
# file object. If this happens, the simplest workaround is to
2002-07-16 18:35:23 -03:00
# not initialize the base classes.
if fp is not None:
self.__super_init(fp, hdrs, url, code)
def __str__(self):
2000-02-10 13:17:14 -04:00
return 'HTTP Error %s: %s' % (self.code, self.msg)
# since URLError specifies a .reason attribute, HTTPError should also
# provide this attribute. See issue13211 fo discussion.
@property
def reason(self):
return self.msg
# copied from cookielib.py
2006-05-18 03:51:46 -03:00
_cut_port_re = re.compile(r":\d+$")
def request_host(request):
"""Return request-host, as defined by RFC 2965.
Variation from RFC: returned value is lowercased, for convenient
comparison.
"""
url = request.get_full_url()
host = urlparse.urlparse(url)[1]
if host == "":
host = request.get_header("Host", "")
# remove port, if present
2006-05-18 03:51:46 -03:00
host = _cut_port_re.sub("", host, 1)
return host.lower()
class Request:
def __init__(self, url, data=None, headers={},
origin_req_host=None, unverifiable=False):
2000-02-10 13:17:14 -04:00
# unwrap('<URL:type://host/path>') --> 'type://host/path'
self.__original = unwrap(url)
self.__original, self.__fragment = splittag(self.__original)
2000-02-10 13:17:14 -04:00
self.type = None
# self.__r_type is what's left after doing the splittype
self.host = None
self.port = None
self._tunnel_host = None
self.data = data
2000-02-10 13:17:14 -04:00
self.headers = {}
for key, value in headers.items():
self.add_header(key, value)
self.unredirected_hdrs = {}
if origin_req_host is None:
origin_req_host = request_host(self)
self.origin_req_host = origin_req_host
self.unverifiable = unverifiable
def __getattr__(self, attr):
2000-02-10 13:17:14 -04:00
# XXX this is a fallback mechanism to guard against these
# methods getting called in a non-standard order. this may be
2000-02-10 13:17:14 -04:00
# too complicated and/or unnecessary.
# XXX should the __r_XXX attributes be public?
if attr[:12] == '_Request__r_':
name = attr[12:]
if hasattr(Request, 'get_' + name):
getattr(self, 'get_' + name)()
return getattr(self, attr)
raise AttributeError, attr
def get_method(self):
if self.has_data():
return "POST"
else:
return "GET"
# XXX these helper methods are lame
def add_data(self, data):
self.data = data
def has_data(self):
return self.data is not None
def get_data(self):
return self.data
def get_full_url(self):
if self.__fragment:
return '%s#%s' % (self.__original, self.__fragment)
else:
return self.__original
def get_type(self):
2000-02-10 13:17:14 -04:00
if self.type is None:
self.type, self.__r_type = splittype(self.__original)
if self.type is None:
raise ValueError, "unknown url type: %s" % self.__original
2000-02-10 13:17:14 -04:00
return self.type
def get_host(self):
2000-02-10 13:17:14 -04:00
if self.host is None:
self.host, self.__r_host = splithost(self.__r_type)
if self.host:
self.host = unquote(self.host)
return self.host
def get_selector(self):
2000-02-10 13:17:14 -04:00
return self.__r_host
def set_proxy(self, host, type):
if self.type == 'https' and not self._tunnel_host:
self._tunnel_host = self.host
else:
self.type = type
self.__r_host = self.__original
self.host = host
def has_proxy(self):
return self.__r_host == self.__original
def get_origin_req_host(self):
return self.origin_req_host
def is_unverifiable(self):
return self.unverifiable
def add_header(self, key, val):
2000-02-10 13:17:14 -04:00
# useful for something like authentication
self.headers[key.capitalize()] = val
def add_unredirected_header(self, key, val):
# will not be added to a redirected request
self.unredirected_hdrs[key.capitalize()] = val
def has_header(self, header_name):
return (header_name in self.headers or
header_name in self.unredirected_hdrs)
def get_header(self, header_name, default=None):
return self.headers.get(
header_name,
self.unredirected_hdrs.get(header_name, default))
def header_items(self):
hdrs = self.unredirected_hdrs.copy()
hdrs.update(self.headers)
return hdrs.items()
class OpenerDirector:
def __init__(self):
client_version = "Python-urllib/%s" % __version__
self.addheaders = [('User-agent', client_version)]
# self.handlers is retained only for backward compatibility
self.handlers = []
# manage the individual handlers
self.handle_open = {}
self.handle_error = {}
self.process_response = {}
self.process_request = {}
def add_handler(self, handler):
if not hasattr(handler, "add_parent"):
raise TypeError("expected BaseHandler instance, got %r" %
type(handler))
added = False
for meth in dir(handler):
if meth in ["redirect_request", "do_open", "proxy_open"]:
# oops, coincidental match
continue
i = meth.find("_")
protocol = meth[:i]
condition = meth[i+1:]
if condition.startswith("error"):
j = condition.find("_") + i + 1
kind = meth[j+1:]
try:
2001-02-09 07:10:16 -04:00
kind = int(kind)
except ValueError:
pass
lookup = self.handle_error.get(protocol, {})
self.handle_error[protocol] = lookup
elif condition == "open":
kind = protocol
lookup = self.handle_open
elif condition == "response":
kind = protocol
lookup = self.process_response
elif condition == "request":
kind = protocol
lookup = self.process_request
else:
continue
handlers = lookup.setdefault(kind, [])
if handlers:
bisect.insort(handlers, handler)
else:
handlers.append(handler)
added = True
if added:
bisect.insort(self.handlers, handler)
handler.add_parent(self)
def close(self):
# Only exists for backwards compatibility.
pass
def _call_chain(self, chain, kind, meth_name, *args):
# Handlers raise an exception if no one else should try to handle
# the request, or return None if they can't but another handler
# could. Otherwise, they return the response.
handlers = chain.get(kind, ())
for handler in handlers:
func = getattr(handler, meth_name)
result = func(*args)
if result is not None:
return result
def open(self, fullurl, data=None, timeout=socket._GLOBAL_DEFAULT_TIMEOUT):
2000-02-10 13:17:14 -04:00
# accept a URL or a Request object
if isinstance(fullurl, basestring):
2000-02-10 13:17:14 -04:00
req = Request(fullurl, data)
else:
req = fullurl
if data is not None:
req.add_data(data)
req.timeout = timeout
protocol = req.get_type()
# pre-process request
meth_name = protocol+"_request"
for processor in self.process_request.get(protocol, []):
meth = getattr(processor, meth_name)
req = meth(req)
response = self._open(req, data)
# post-process response
meth_name = protocol+"_response"
for processor in self.process_response.get(protocol, []):
meth = getattr(processor, meth_name)
response = meth(req, response)
return response
def _open(self, req, data=None):
result = self._call_chain(self.handle_open, 'default',
'default_open', req)
if result:
return result
protocol = req.get_type()
result = self._call_chain(self.handle_open, protocol, protocol +
'_open', req)
if result:
return result
return self._call_chain(self.handle_open, 'unknown',
'unknown_open', req)
def error(self, proto, *args):
if proto in ('http', 'https'):
# XXX http[s] protocols are special-cased
dict = self.handle_error['http'] # https is not different than http
proto = args[2] # YUCK!
meth_name = 'http_error_%s' % proto
http_err = 1
orig_args = args
else:
dict = self.handle_error
meth_name = proto + '_error'
http_err = 0
args = (dict, proto, meth_name) + args
result = self._call_chain(*args)
if result:
return result
if http_err:
args = (dict, 'default', 'http_error_default') + orig_args
return self._call_chain(*args)
# XXX probably also want an abstract factory that knows when it makes
# sense to skip a superclass in favor of a subclass and when it might
# make sense to include both
def build_opener(*handlers):
"""Create an opener object from a list of handlers.
The opener will use several default handlers, including support
for HTTP, FTP and when applicable, HTTPS.
If any of the handlers passed as arguments are subclasses of the
default handlers, the default handlers will not be used.
"""
import types
def isclass(obj):
return isinstance(obj, (types.ClassType, type))
opener = OpenerDirector()
default_classes = [ProxyHandler, UnknownHandler, HTTPHandler,
HTTPDefaultErrorHandler, HTTPRedirectHandler,
FTPHandler, FileHandler, HTTPErrorProcessor]
if hasattr(httplib, 'HTTPS'):
default_classes.append(HTTPSHandler)
skip = set()
for klass in default_classes:
for check in handlers:
if isclass(check):
if issubclass(check, klass):
skip.add(klass)
elif isinstance(check, klass):
skip.add(klass)
for klass in skip:
default_classes.remove(klass)
for klass in default_classes:
opener.add_handler(klass())
for h in handlers:
if isclass(h):
h = h()
opener.add_handler(h)
return opener
class BaseHandler:
handler_order = 500
def add_parent(self, parent):
self.parent = parent
2004-01-18 16:29:55 -04:00
def close(self):
# Only exists for backwards compatibility
pass
2004-01-18 16:29:55 -04:00
def __lt__(self, other):
if not hasattr(other, "handler_order"):
# Try to preserve the old behavior of having custom classes
# inserted after default ones (works only for custom user
# classes which are not aware of handler_order).
return True
return self.handler_order < other.handler_order
2003-06-15 20:26:30 -03:00
class HTTPErrorProcessor(BaseHandler):
"""Process HTTP error responses."""
handler_order = 1000 # after all other processing
def http_response(self, request, response):
code, msg, hdrs = response.code, response.msg, response.info()
# According to RFC 2616, "2xx" code indicates that the client's
# request was successfully received, understood, and accepted.
if not (200 <= code < 300):
response = self.parent.error(
'http', request, response, code, msg, hdrs)
return response
https_response = http_response
class HTTPDefaultErrorHandler(BaseHandler):
def http_error_default(self, req, fp, code, msg, hdrs):
2000-02-10 13:17:14 -04:00
raise HTTPError(req.get_full_url(), code, msg, hdrs, fp)
class HTTPRedirectHandler(BaseHandler):
# maximum number of redirections to any single URL
# this is needed because of the state that cookies introduce
max_repeats = 4
# maximum total number of redirections (regardless of URL) before
# assuming we're in a loop
max_redirections = 10
def redirect_request(self, req, fp, code, msg, headers, newurl):
"""Return a Request or None in response to a redirect.
This is called by the http_error_30x methods when a
redirection response is received. If a redirection should
take place, return a new Request to allow http_error_30x to
perform the redirect. Otherwise, raise HTTPError if no-one
else should try to handle this url. Return None if you can't
but another Handler might.
"""
m = req.get_method()
if (code in (301, 302, 303, 307) and m in ("GET", "HEAD")
or code in (301, 302, 303) and m == "POST"):
# Strictly (according to RFC 2616), 301 or 302 in response
# to a POST MUST NOT cause a redirection without confirmation
# from the user (of urllib2, in this case). In practice,
# essentially all clients do redirect in this case, so we
# do the same.
# be conciliant with URIs containing a space
newurl = newurl.replace(' ', '%20')
newheaders = dict((k,v) for k,v in req.headers.items()
if k.lower() not in ("content-length", "content-type")
)
return Request(newurl,
headers=newheaders,
origin_req_host=req.get_origin_req_host(),
unverifiable=True)
else:
raise HTTPError(req.get_full_url(), code, msg, headers, fp)
# Implementation note: To avoid the server sending us into an
# infinite loop, the request object needs to track what URLs we
# have already seen. Do this by adding a handler-specific
# attribute to the Request object.
def http_error_302(self, req, fp, code, msg, headers):
# Some servers (incorrectly) return multiple Location headers
# (so probably same goes for URI). Use first header.
if 'location' in headers:
newurl = headers.getheaders('location')[0]
elif 'uri' in headers:
newurl = headers.getheaders('uri')[0]
else:
return
# fix a possible malformed URL
urlparts = urlparse.urlparse(newurl)
if not urlparts.path:
urlparts = list(urlparts)
urlparts[2] = "/"
newurl = urlparse.urlunparse(urlparts)
newurl = urlparse.urljoin(req.get_full_url(), newurl)
# For security reasons we do not allow redirects to protocols
# other than HTTP, HTTPS or FTP.
newurl_lower = newurl.lower()
if not (newurl_lower.startswith('http://') or
newurl_lower.startswith('https://') or
newurl_lower.startswith('ftp://')):
raise HTTPError(newurl, code,
msg + " - Redirection to url '%s' is not allowed" %
newurl,
headers, fp)
# XXX Probably want to forget about the state of the current
# request, although that might interact poorly with other
# handlers that also use handler-specific request attributes
new = self.redirect_request(req, fp, code, msg, headers, newurl)
if new is None:
return
# loop detection
# .redirect_dict has a key url if url was previously visited.
if hasattr(req, 'redirect_dict'):
visited = new.redirect_dict = req.redirect_dict
if (visited.get(newurl, 0) >= self.max_repeats or
len(visited) >= self.max_redirections):
raise HTTPError(req.get_full_url(), code,
self.inf_msg + msg, headers, fp)
else:
visited = new.redirect_dict = req.redirect_dict = {}
visited[newurl] = visited.get(newurl, 0) + 1
# Don't close the fp until we are sure that we won't use it
2001-08-09 18:40:30 -03:00
# with HTTPError.
fp.read()
fp.close()
return self.parent.open(new, timeout=req.timeout)
http_error_301 = http_error_303 = http_error_307 = http_error_302
inf_msg = "The HTTP server returned a redirect error that would " \
"lead to an infinite loop.\n" \
"The last 30x error message was:\n"
def _parse_proxy(proxy):
"""Return (scheme, user, password, host/port) given a URL or an authority.
If a URL is supplied, it must have an authority (host:port) component.
According to RFC 3986, having an authority component means the URL must
have two slashes after the scheme:
>>> _parse_proxy('file:/ftp.example.com/')
Traceback (most recent call last):
ValueError: proxy URL with no authority: 'file:/ftp.example.com/'
The first three items of the returned tuple may be None.
Examples of authority parsing:
>>> _parse_proxy('proxy.example.com')
(None, None, None, 'proxy.example.com')
>>> _parse_proxy('proxy.example.com:3128')
(None, None, None, 'proxy.example.com:3128')
The authority component may optionally include userinfo (assumed to be
username:password):
>>> _parse_proxy('joe:password@proxy.example.com')
(None, 'joe', 'password', 'proxy.example.com')
>>> _parse_proxy('joe:password@proxy.example.com:3128')
(None, 'joe', 'password', 'proxy.example.com:3128')
Same examples, but with URLs instead:
>>> _parse_proxy('http://proxy.example.com/')
('http', None, None, 'proxy.example.com')
>>> _parse_proxy('http://proxy.example.com:3128/')
('http', None, None, 'proxy.example.com:3128')
>>> _parse_proxy('http://joe:password@proxy.example.com/')
('http', 'joe', 'password', 'proxy.example.com')
>>> _parse_proxy('http://joe:password@proxy.example.com:3128')
('http', 'joe', 'password', 'proxy.example.com:3128')
Everything after the authority is ignored:
>>> _parse_proxy('ftp://joe:password@proxy.example.com/rubbish:3128')
('ftp', 'joe', 'password', 'proxy.example.com')
Test for no trailing '/' case:
>>> _parse_proxy('http://joe:password@proxy.example.com')
('http', 'joe', 'password', 'proxy.example.com')
"""
scheme, r_scheme = splittype(proxy)
if not r_scheme.startswith("/"):
# authority
scheme = None
authority = proxy
else:
# URL
if not r_scheme.startswith("//"):
raise ValueError("proxy URL with no authority: %r" % proxy)
# We have an authority, so for RFC 3986-compliant URLs (by ss 3.
# and 3.3.), path is empty or starts with '/'
end = r_scheme.find("/", 2)
if end == -1:
end = None
authority = r_scheme[2:end]
userinfo, hostport = splituser(authority)
if userinfo is not None:
user, password = splitpasswd(userinfo)
else:
user = password = None
return scheme, user, password, hostport
class ProxyHandler(BaseHandler):
# Proxies must be in front
handler_order = 100
def __init__(self, proxies=None):
2000-02-10 13:17:14 -04:00
if proxies is None:
proxies = getproxies()
assert hasattr(proxies, 'has_key'), "proxies must be a mapping"
self.proxies = proxies
for type, url in proxies.items():
setattr(self, '%s_open' % type,
2000-02-10 13:17:14 -04:00
lambda r, proxy=url, type=type, meth=self.proxy_open: \
meth(r, proxy, type))
def proxy_open(self, req, proxy, type):
2000-02-10 13:17:14 -04:00
orig_type = req.get_type()
proxy_type, user, password, hostport = _parse_proxy(proxy)
if proxy_type is None:
proxy_type = orig_type
if req.host and proxy_bypass(req.host):
return None
if user and password:
user_pass = '%s:%s' % (unquote(user), unquote(password))
creds = base64.b64encode(user_pass).strip()
req.add_header('Proxy-authorization', 'Basic ' + creds)
hostport = unquote(hostport)
req.set_proxy(hostport, proxy_type)
if orig_type == proxy_type or orig_type == 'https':
2000-02-10 13:17:14 -04:00
# let other handlers take care of it
return None
else:
# need to start over, because the other handlers don't
# grok the proxy's URL type
# e.g. if we have a constructor arg proxies like so:
# {'http': 'ftp://proxy.example.com'}, we may end up turning
# a request for http://acme.example.com/a into one for
# ftp://proxy.example.com/a
return self.parent.open(req, timeout=req.timeout)
class HTTPPasswordMgr:
def __init__(self):
2000-02-10 13:17:14 -04:00
self.passwd = {}
def add_password(self, realm, uri, user, passwd):
2000-02-10 13:17:14 -04:00
# uri could be a single URI or a sequence
if isinstance(uri, basestring):
2000-02-10 13:17:14 -04:00
uri = [uri]
if not realm in self.passwd:
2000-02-10 13:17:14 -04:00
self.passwd[realm] = {}
for default_port in True, False:
reduced_uri = tuple(
[self.reduce_uri(u, default_port) for u in uri])
self.passwd[realm][reduced_uri] = (user, passwd)
def find_user_password(self, realm, authuri):
2000-02-10 13:17:14 -04:00
domains = self.passwd.get(realm, {})
for default_port in True, False:
reduced_authuri = self.reduce_uri(authuri, default_port)
for uris, authinfo in domains.iteritems():
for uri in uris:
if self.is_suburi(uri, reduced_authuri):
return authinfo
2000-02-10 13:17:14 -04:00
return None, None
def reduce_uri(self, uri, default_port=True):
"""Accept authority or URI and extract only the authority and path."""
# note HTTP URLs do not have a userinfo component
parts = urlparse.urlsplit(uri)
2000-02-10 13:17:14 -04:00
if parts[1]:
# URI
scheme = parts[0]
authority = parts[1]
path = parts[2] or '/'
2000-02-10 13:17:14 -04:00
else:
# host or host:port
scheme = None
authority = uri
path = '/'
host, port = splitport(authority)
if default_port and port is None and scheme is not None:
dport = {"http": 80,
"https": 443,
}.get(scheme)
if dport is not None:
authority = "%s:%d" % (host, dport)
return authority, path
def is_suburi(self, base, test):
2000-02-10 13:17:14 -04:00
"""Check if test is below base in a URI tree
Both args must be URIs in reduced form.
"""
if base == test:
return True
2000-02-10 13:17:14 -04:00
if base[0] != test[0]:
return False
common = posixpath.commonprefix((base[1], test[1]))
2000-02-10 13:17:14 -04:00
if len(common) == len(base[1]):
return True
return False
class HTTPPasswordMgrWithDefaultRealm(HTTPPasswordMgr):
def find_user_password(self, realm, authuri):
user, password = HTTPPasswordMgr.find_user_password(self, realm,
authuri)
if user is not None:
return user, password
return HTTPPasswordMgr.find_user_password(self, None, authuri)
class AbstractBasicAuthHandler:
# XXX this allows for multiple auth-schemes, but will stupidly pick
# the last one with a realm specified.
# allow for double- and single-quoted realm values
# (single quotes are a violation of the RFC, but appear in the wild)
rx = re.compile('(?:.*,)*[ \t]*([^ \t]+)[ \t]+'
'realm=(["\']?)([^"\']*)\\2', re.I)
# XXX could pre-emptively send auth info already accepted (RFC 2617,
# end of section 2, and section 1.2 immediately after "credentials"
# production).
def __init__(self, password_mgr=None):
if password_mgr is None:
password_mgr = HTTPPasswordMgr()
self.passwd = password_mgr
2000-02-10 13:17:14 -04:00
self.add_password = self.passwd.add_password
self.retried = 0
def reset_retry_count(self):
self.retried = 0
def http_error_auth_reqed(self, authreq, host, req, headers):
# host may be an authority (without userinfo) or a URL with an
# authority
# XXX could be multiple headers
authreq = headers.get(authreq, None)
if self.retried > 5:
# retry sending the username:password 5 times before failing.
raise HTTPError(req.get_full_url(), 401, "basic auth failed",
headers, None)
else:
self.retried += 1
if authreq:
mo = AbstractBasicAuthHandler.rx.search(authreq)
if mo:
scheme, quote, realm = mo.groups()
if quote not in ['"', "'"]:
warnings.warn("Basic Auth Realm was unquoted",
UserWarning, 2)
2001-02-09 07:10:16 -04:00
if scheme.lower() == 'basic':
response = self.retry_http_basic_auth(host, req, realm)
if response and response.code != 401:
self.retried = 0
return response
def retry_http_basic_auth(self, host, req, realm):
user, pw = self.passwd.find_user_password(realm, host)
if pw is not None:
2000-02-10 13:17:14 -04:00
raw = "%s:%s" % (user, pw)
auth = 'Basic %s' % base64.b64encode(raw).strip()
if req.headers.get(self.auth_header, None) == auth:
return None
req.add_unredirected_header(self.auth_header, auth)
return self.parent.open(req, timeout=req.timeout)
else:
return None
class HTTPBasicAuthHandler(AbstractBasicAuthHandler, BaseHandler):
auth_header = 'Authorization'
def http_error_401(self, req, fp, code, msg, headers):
url = req.get_full_url()
response = self.http_error_auth_reqed('www-authenticate',
url, req, headers)
self.reset_retry_count()
return response
class ProxyBasicAuthHandler(AbstractBasicAuthHandler, BaseHandler):
auth_header = 'Proxy-authorization'
def http_error_407(self, req, fp, code, msg, headers):
# http_error_auth_reqed requires that there is no userinfo component in
# authority. Assume there isn't one, since urllib2 does not (and
# should not, RFC 3986 s. 3.2.1) support requests for URLs containing
# userinfo.
authority = req.get_host()
response = self.http_error_auth_reqed('proxy-authenticate',
authority, req, headers)
self.reset_retry_count()
return response
def randombytes(n):
"""Return n random bytes."""
# Use /dev/urandom if it is available. Fall back to random module
# if not. It might be worthwhile to extend this function to use
# other platform-specific mechanisms for getting random bytes.
if os.path.exists("/dev/urandom"):
f = open("/dev/urandom")
s = f.read(n)
f.close()
return s
else:
L = [chr(random.randrange(0, 256)) for i in range(n)]
return "".join(L)
class AbstractDigestAuthHandler:
# Digest authentication is specified in RFC 2617.
# XXX The client does not inspect the Authentication-Info header
# in a successful response.
# XXX It should be possible to test this implementation against
# a mock server that just generates a static set of challenges.
# XXX qop="auth-int" supports is shaky
def __init__(self, passwd=None):
if passwd is None:
passwd = HTTPPasswordMgr()
self.passwd = passwd
2000-02-10 13:17:14 -04:00
self.add_password = self.passwd.add_password
self.retried = 0
self.nonce_count = 0
self.last_nonce = None
def reset_retry_count(self):
self.retried = 0
def http_error_auth_reqed(self, auth_header, host, req, headers):
authreq = headers.get(auth_header, None)
if self.retried > 5:
# Don't fail endlessly - if we failed once, we'll probably
# fail a second time. Hm. Unless the Password Manager is
# prompting for the information. Crap. This isn't great
# but it's better than the current 'repeat until recursion
# depth exceeded' approach <wink>
2004-01-18 16:29:55 -04:00
raise HTTPError(req.get_full_url(), 401, "digest auth failed",
headers, None)
else:
self.retried += 1
2000-02-10 13:17:14 -04:00
if authreq:
scheme = authreq.split()[0]
if scheme.lower() == 'digest':
2000-02-10 13:17:14 -04:00
return self.retry_http_digest_auth(req, authreq)
def retry_http_digest_auth(self, req, auth):
2001-02-09 07:10:16 -04:00
token, challenge = auth.split(' ', 1)
2000-02-10 13:17:14 -04:00
chal = parse_keqv_list(parse_http_list(challenge))
auth = self.get_authorization(req, chal)
if auth:
auth_val = 'Digest %s' % auth
if req.headers.get(self.auth_header, None) == auth_val:
return None
req.add_unredirected_header(self.auth_header, auth_val)
resp = self.parent.open(req, timeout=req.timeout)
2000-02-10 13:17:14 -04:00
return resp
def get_cnonce(self, nonce):
# The cnonce-value is an opaque
# quoted string value provided by the client and used by both client
# and server to avoid chosen plaintext attacks, to provide mutual
# authentication, and to provide some message integrity protection.
# This isn't a fabulous effort, but it's probably Good Enough.
dig = hashlib.sha1("%s:%s:%s:%s" % (self.nonce_count, nonce, time.ctime(),
randombytes(8))).hexdigest()
return dig[:16]
def get_authorization(self, req, chal):
2000-02-10 13:17:14 -04:00
try:
realm = chal['realm']
nonce = chal['nonce']
qop = chal.get('qop')
2000-02-10 13:17:14 -04:00
algorithm = chal.get('algorithm', 'MD5')
# mod_digest doesn't send an opaque, even though it isn't
# supposed to be optional
opaque = chal.get('opaque', None)
except KeyError:
return None
H, KD = self.get_algorithm_impls(algorithm)
if H is None:
return None
user, pw = self.passwd.find_user_password(realm, req.get_full_url())
2000-02-10 13:17:14 -04:00
if user is None:
return None
# XXX not implemented yet
if req.has_data():
entdig = self.get_entity_digest(req.get_data(), chal)
else:
entdig = None
A1 = "%s:%s:%s" % (user, realm, pw)
A2 = "%s:%s" % (req.get_method(),
2000-02-10 13:17:14 -04:00
# XXX selector: what about proxies and full urls
req.get_selector())
if qop == 'auth':
if nonce == self.last_nonce:
self.nonce_count += 1
else:
self.nonce_count = 1
self.last_nonce = nonce
ncvalue = '%08x' % self.nonce_count
cnonce = self.get_cnonce(nonce)
noncebit = "%s:%s:%s:%s:%s" % (nonce, ncvalue, cnonce, qop, H(A2))
respdig = KD(H(A1), noncebit)
elif qop is None:
respdig = KD(H(A1), "%s:%s" % (nonce, H(A2)))
else:
# XXX handle auth-int.
raise URLError("qop '%s' is not supported." % qop)
2004-01-18 16:29:55 -04:00
2000-02-10 13:17:14 -04:00
# XXX should the partial digests be encoded too?
base = 'username="%s", realm="%s", nonce="%s", uri="%s", ' \
'response="%s"' % (user, realm, nonce, req.get_selector(),
respdig)
if opaque:
base += ', opaque="%s"' % opaque
2000-02-10 13:17:14 -04:00
if entdig:
base += ', digest="%s"' % entdig
base += ', algorithm="%s"' % algorithm
if qop:
base += ', qop=auth, nc=%s, cnonce="%s"' % (ncvalue, cnonce)
2000-02-10 13:17:14 -04:00
return base
def get_algorithm_impls(self, algorithm):
# algorithm should be case-insensitive according to RFC2617
algorithm = algorithm.upper()
2000-02-10 13:17:14 -04:00
# lambdas assume digest modules are imported at the top level
if algorithm == 'MD5':
H = lambda x: hashlib.md5(x).hexdigest()
2000-02-10 13:17:14 -04:00
elif algorithm == 'SHA':
H = lambda x: hashlib.sha1(x).hexdigest()
2000-02-10 13:17:14 -04:00
# XXX MD5-sess
KD = lambda s, d: H("%s:%s" % (s, d))
2000-02-10 13:17:14 -04:00
return H, KD
def get_entity_digest(self, data, chal):
2000-02-10 13:17:14 -04:00
# XXX not implemented yet
return None
class HTTPDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler):
"""An authentication protocol defined by RFC 2069
Digest authentication improves on basic authentication because it
does not transmit passwords in the clear.
"""
auth_header = 'Authorization'
handler_order = 490 # before Basic auth
def http_error_401(self, req, fp, code, msg, headers):
host = urlparse.urlparse(req.get_full_url())[1]
2004-01-18 16:29:55 -04:00
retry = self.http_error_auth_reqed('www-authenticate',
host, req, headers)
self.reset_retry_count()
return retry
class ProxyDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler):
auth_header = 'Proxy-Authorization'
handler_order = 490 # before Basic auth
def http_error_407(self, req, fp, code, msg, headers):
host = req.get_host()
2004-01-18 16:29:55 -04:00
retry = self.http_error_auth_reqed('proxy-authenticate',
host, req, headers)
self.reset_retry_count()
return retry
class AbstractHTTPHandler(BaseHandler):
def __init__(self, debuglevel=0):
self._debuglevel = debuglevel
def set_http_debuglevel(self, level):
self._debuglevel = level
def do_request_(self, request):
host = request.get_host()
if not host:
raise URLError('no host given')
if request.has_data(): # POST
data = request.get_data()
if not request.has_header('Content-type'):
request.add_unredirected_header(
'Content-type',
'application/x-www-form-urlencoded')
if not request.has_header('Content-length'):
request.add_unredirected_header(
'Content-length', '%d' % len(data))
sel_host = host
if request.has_proxy():
scheme, sel = splittype(request.get_selector())
sel_host, sel_path = splithost(sel)
if not request.has_header('Host'):
request.add_unredirected_header('Host', sel_host)
for name, value in self.parent.addheaders:
name = name.capitalize()
if not request.has_header(name):
request.add_unredirected_header(name, value)
return request
def do_open(self, http_class, req):
"""Return an addinfourl object for the request, using http_class.
http_class must implement the HTTPConnection API from httplib.
The addinfourl return value is a file-like object. It also
has methods and attributes including:
- info(): return a mimetools.Message object for the headers
- geturl(): return the original request URL
- code: HTTP status code
"""
host = req.get_host()
if not host:
raise URLError('no host given')
h = http_class(host, timeout=req.timeout) # will parse host:port
h.set_debuglevel(self._debuglevel)
headers = dict(req.unredirected_hdrs)
headers.update(dict((k, v) for k, v in req.headers.items()
if k not in headers))
# We want to make an HTTP/1.1 request, but the addinfourl
# class isn't prepared to deal with a persistent connection.
# It will try to read all remaining data from the socket,
# which will block while the server waits for the next request.
# So make sure the connection gets closed after the (only)
# request.
headers["Connection"] = "close"
headers = dict(
(name.title(), val) for name, val in headers.items())
if req._tunnel_host:
tunnel_headers = {}
proxy_auth_hdr = "Proxy-Authorization"
if proxy_auth_hdr in headers:
tunnel_headers[proxy_auth_hdr] = headers[proxy_auth_hdr]
# Proxy-Authorization should not be sent to origin
# server.
del headers[proxy_auth_hdr]
h.set_tunnel(req._tunnel_host, headers=tunnel_headers)
try:
h.request(req.get_method(), req.get_selector(), req.data, headers)
except socket.error, err: # XXX what error?
h.close()
raise URLError(err)
else:
try:
r = h.getresponse(buffering=True)
except TypeError: # buffering kw not supported
r = h.getresponse()
# Pick apart the HTTPResponse object to get the addinfourl
# object initialized properly.
# Wrap the HTTPResponse object in socket's file object adapter
# for Windows. That adapter calls recv(), so delegate recv()
# to read(). This weird wrapping allows the returned object to
# have readline() and readlines() methods.
2004-08-07 22:05:14 -03:00
# XXX It might be better to extract the read buffering code
# out of socket._fileobject() and into a base class.
2004-08-07 22:05:14 -03:00
r.recv = r.read
fp = socket._fileobject(r, close=True)
2004-08-07 22:05:14 -03:00
resp = addinfourl(fp, r.msg, req.get_full_url())
resp.code = r.status
resp.msg = r.reason
return resp
class HTTPHandler(AbstractHTTPHandler):
def http_open(self, req):
return self.do_open(httplib.HTTPConnection, req)
http_request = AbstractHTTPHandler.do_request_
if hasattr(httplib, 'HTTPS'):
class HTTPSHandler(AbstractHTTPHandler):
def https_open(self, req):
return self.do_open(httplib.HTTPSConnection, req)
https_request = AbstractHTTPHandler.do_request_
class HTTPCookieProcessor(BaseHandler):
def __init__(self, cookiejar=None):
import cookielib
if cookiejar is None:
cookiejar = cookielib.CookieJar()
self.cookiejar = cookiejar
def http_request(self, request):
self.cookiejar.add_cookie_header(request)
return request
def http_response(self, request, response):
self.cookiejar.extract_cookies(response, request)
return response
https_request = http_request
https_response = http_response
class UnknownHandler(BaseHandler):
def unknown_open(self, req):
2000-02-10 13:17:14 -04:00
type = req.get_type()
raise URLError('unknown url type: %s' % type)
def parse_keqv_list(l):
"""Parse list of key=value strings where keys are not duplicated."""
parsed = {}
for elt in l:
2001-02-09 07:10:16 -04:00
k, v = elt.split('=', 1)
2000-02-10 13:17:14 -04:00
if v[0] == '"' and v[-1] == '"':
v = v[1:-1]
parsed[k] = v
return parsed
def parse_http_list(s):
"""Parse lists as described by RFC 2068 Section 2.
2004-04-06 16:43:03 -03:00
In particular, parse comma-separated lists where the elements of
the list may include quoted-strings. A quoted-string could
contain a comma. A non-quoted string could have quotes in the
middle. Neither commas nor quotes count if they are escaped.
Only double-quotes count, not single-quotes.
"""
res = []
part = ''
escape = quote = False
for cur in s:
if escape:
part += cur
escape = False
continue
if quote:
if cur == '\\':
escape = True
2000-02-10 13:17:14 -04:00
continue
elif cur == '"':
quote = False
part += cur
continue
if cur == ',':
res.append(part)
part = ''
continue
if cur == '"':
quote = True
part += cur
# append last part
if part:
res.append(part)
return [part.strip() for part in res]
def _safe_gethostbyname(host):
try:
return socket.gethostbyname(host)
except socket.gaierror:
return None
class FileHandler(BaseHandler):
# Use local file or FTP depending on form of URL
def file_open(self, req):
2000-02-10 13:17:14 -04:00
url = req.get_selector()
if url[:2] == '//' and url[2:3] != '/' and (req.host and
req.host != 'localhost'):
2000-02-10 13:17:14 -04:00
req.type = 'ftp'
return self.parent.open(req)
else:
return self.open_local_file(req)
# names for the localhost
names = None
def get_names(self):
2000-02-10 13:17:14 -04:00
if FileHandler.names is None:
try:
FileHandler.names = tuple(
socket.gethostbyname_ex('localhost')[2] +
socket.gethostbyname_ex(socket.gethostname())[2])
except socket.gaierror:
FileHandler.names = (socket.gethostbyname('localhost'),)
2000-02-10 13:17:14 -04:00
return FileHandler.names
# not entirely sure what the rules are here
def open_local_file(self, req):
import email.utils
import mimetypes
2000-02-10 13:17:14 -04:00
host = req.get_host()
filename = req.get_selector()
localfile = url2pathname(filename)
try:
stats = os.stat(localfile)
size = stats.st_size
modified = email.utils.formatdate(stats.st_mtime, usegmt=True)
mtype = mimetypes.guess_type(filename)[0]
headers = mimetools.Message(StringIO(
'Content-type: %s\nContent-length: %d\nLast-modified: %s\n' %
(mtype or 'text/plain', size, modified)))
if host:
host, port = splitport(host)
if not host or \
(not port and _safe_gethostbyname(host) in self.get_names()):
if host:
origurl = 'file://' + host + filename
else:
origurl = 'file://' + filename
return addinfourl(open(localfile, 'rb'), headers, origurl)
except OSError, msg:
# urllib2 users shouldn't expect OSErrors coming from urlopen()
raise URLError(msg)
2000-02-10 13:17:14 -04:00
raise URLError('file not on local host')
class FTPHandler(BaseHandler):
def ftp_open(self, req):
import ftplib
import mimetypes
2000-02-10 13:17:14 -04:00
host = req.get_host()
if not host:
raise URLError('ftp error: no host given')
host, port = splitport(host)
if port is None:
port = ftplib.FTP_PORT
else:
port = int(port)
# username/password handling
user, host = splituser(host)
if user:
user, passwd = splitpasswd(user)
else:
passwd = None
host = unquote(host)
user = user or ''
passwd = passwd or ''
try:
host = socket.gethostbyname(host)
except socket.error, msg:
raise URLError(msg)
2000-02-10 13:17:14 -04:00
path, attrs = splitattr(req.get_selector())
2001-02-09 07:10:16 -04:00
dirs = path.split('/')
dirs = map(unquote, dirs)
2000-02-10 13:17:14 -04:00
dirs, file = dirs[:-1], dirs[-1]
if dirs and not dirs[0]:
dirs = dirs[1:]
try:
fw = self.connect_ftp(user, passwd, host, port, dirs, req.timeout)
2000-02-10 13:17:14 -04:00
type = file and 'I' or 'D'
for attr in attrs:
attr, value = splitvalue(attr)
2001-02-09 07:10:16 -04:00
if attr.lower() == 'type' and \
2000-02-10 13:17:14 -04:00
value in ('a', 'A', 'i', 'I', 'd', 'D'):
2001-02-09 07:10:16 -04:00
type = value.upper()
2000-02-10 13:17:14 -04:00
fp, retrlen = fw.retrfile(file, type)
headers = ""
mtype = mimetypes.guess_type(req.get_full_url())[0]
if mtype:
headers += "Content-type: %s\n" % mtype
2000-02-10 13:17:14 -04:00
if retrlen is not None and retrlen >= 0:
headers += "Content-length: %d\n" % retrlen
sf = StringIO(headers)
headers = mimetools.Message(sf)
2000-02-10 13:17:14 -04:00
return addinfourl(fp, headers, req.get_full_url())
except ftplib.all_errors, msg:
raise URLError, ('ftp error: %s' % msg), sys.exc_info()[2]
def connect_ftp(self, user, passwd, host, port, dirs, timeout):
fw = ftpwrapper(user, passwd, host, port, dirs, timeout,
persistent=False)
## fw.ftp.set_debuglevel(1)
return fw
class CacheFTPHandler(FTPHandler):
# XXX would be nice to have pluggable cache strategies
# XXX this stuff is definitely not thread safe
def __init__(self):
self.cache = {}
self.timeout = {}
self.soonest = 0
self.delay = 60
2000-02-10 13:17:14 -04:00
self.max_conns = 16
def setTimeout(self, t):
self.delay = t
def setMaxConns(self, m):
2000-02-10 13:17:14 -04:00
self.max_conns = m
def connect_ftp(self, user, passwd, host, port, dirs, timeout):
key = user, host, port, '/'.join(dirs), timeout
if key in self.cache:
self.timeout[key] = time.time() + self.delay
else:
self.cache[key] = ftpwrapper(user, passwd, host, port, dirs, timeout)
self.timeout[key] = time.time() + self.delay
2000-02-10 13:17:14 -04:00
self.check_cache()
return self.cache[key]
def check_cache(self):
2000-02-10 13:17:14 -04:00
# first check for old ones
t = time.time()
if self.soonest <= t:
for k, v in self.timeout.items():
if v < t:
self.cache[k].close()
del self.cache[k]
del self.timeout[k]
self.soonest = min(self.timeout.values())
# then check the size
2000-02-10 13:17:14 -04:00
if len(self.cache) == self.max_conns:
for k, v in self.timeout.items():
2000-02-10 13:17:14 -04:00
if v == self.soonest:
del self.cache[k]
del self.timeout[k]
break
self.soonest = min(self.timeout.values())
def clear_cache(self):
for conn in self.cache.values():
conn.close()
self.cache.clear()
self.timeout.clear()