Guido van Rossum | e7b146f | 2000-02-04 15:28:42 +0000 | [diff] [blame] | 1 | """An extensible library for opening URLs using a variety of protocols |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 2 | |
| 3 | The simplest way to use this module is to call the urlopen function, |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 4 | which accepts a string containing a URL or a Request object (described |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 5 | below). It opens the URL and returns the results as file-like |
| 6 | object; the returned object has some extra methods described below. |
| 7 | |
Jeremy Hylton | e190663 | 2002-10-11 17:27:55 +0000 | [diff] [blame] | 8 | The OpenerDirector manages a collection of Handler objects that do |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 9 | all the actual work. Each Handler implements a particular protocol or |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 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 |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 14 | HTTP 301, 302, 303 and 307 redirect errors, and the HTTPDigestAuthHandler |
| 15 | deals with digest authentication. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 16 | |
Georg Brandl | c5ffd91 | 2006-04-02 20:48:11 +0000 | [diff] [blame] | 17 | urlopen(url, data=None) -- basic usage is the same as original |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 18 | urllib. pass the url and optionally data to post to an HTTP URL, and |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 19 | get a file-like object back. One difference is that you can also pass |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 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. |
| 23 | |
| 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. |
| 29 | |
| 30 | install_opener -- installs a new opener as the default opener. |
| 31 | |
| 32 | objects of interest: |
| 33 | OpenerDirector -- |
| 34 | |
| 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. |
| 38 | |
| 39 | BaseHandler -- |
| 40 | |
| 41 | exceptions: |
| 42 | URLError-- a subclass of IOError, individual protocols have their own |
| 43 | specific subclass |
| 44 | |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 45 | HTTPError-- also a valid HTTP response, so you can treat an HTTP error |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 46 | as an exceptional event or valid response |
| 47 | |
| 48 | internals: |
| 49 | BaseHandler and parent |
| 50 | _call_chain conventions |
| 51 | |
| 52 | Example usage: |
| 53 | |
| 54 | import urllib2 |
| 55 | |
| 56 | # set up authentication info |
| 57 | authinfo = urllib2.HTTPBasicAuthHandler() |
| 58 | authinfo.add_password('realm', 'host', 'username', 'password') |
| 59 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 60 | proxy_support = urllib2.ProxyHandler({"http" : "http://ahad-haam:3128"}) |
| 61 | |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 62 | # build a new opener that adds authentication and caching FTP handlers |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 63 | opener = urllib2.build_opener(proxy_support, authinfo, urllib2.CacheFTPHandler) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 64 | |
| 65 | # install it |
| 66 | urllib2.install_opener(opener) |
| 67 | |
| 68 | f = urllib2.urlopen('http://www.python.org/') |
| 69 | |
| 70 | |
| 71 | """ |
| 72 | |
| 73 | # XXX issues: |
| 74 | # If an authentication error handler that tries to perform |
Fred Drake | d5214b0 | 2001-11-08 17:19:29 +0000 | [diff] [blame] | 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. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 80 | # ftp errors aren't handled cleanly |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 81 | # check digest against correct (i.e. non-apache) implementation |
| 82 | |
Georg Brandl | c5ffd91 | 2006-04-02 20:48:11 +0000 | [diff] [blame] | 83 | # Possible extensions: |
| 84 | # complex proxies XXX not sure what exactly was meant by this |
| 85 | # abstract factory for opener |
| 86 | |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 87 | import base64 |
| 88 | import ftplib |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 89 | import httplib |
Jeremy Hylton | 8b78b99 | 2001-10-09 16:18:45 +0000 | [diff] [blame] | 90 | import inspect |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 91 | import md5 |
| 92 | import mimetypes |
| 93 | import mimetools |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 94 | import os |
| 95 | import posixpath |
| 96 | import random |
| 97 | import re |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 98 | import sha |
| 99 | import socket |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 100 | import sys |
| 101 | import time |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 102 | import urlparse |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 103 | import bisect |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 104 | import cookielib |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 105 | |
| 106 | try: |
| 107 | from cStringIO import StringIO |
| 108 | except ImportError: |
| 109 | from StringIO import StringIO |
| 110 | |
Georg Brandl | 7fff58c | 2006-04-02 21:13:13 +0000 | [diff] [blame] | 111 | from urllib import (unwrap, unquote, splittype, splithost, quote, |
Andrew M. Kuchling | 33ad28b | 2004-08-31 11:38:12 +0000 | [diff] [blame] | 112 | addinfourl, splitport, splitgophertype, splitquery, |
| 113 | splitattr, ftpwrapper, noheaders, splituser, splitpasswd, splitvalue) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 114 | |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 115 | # support for FileHandler, proxies via environment variables |
| 116 | from urllib import localhost, url2pathname, getproxies |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 117 | |
Georg Brandl | 720096a | 2006-04-02 20:45:34 +0000 | [diff] [blame] | 118 | # used in User-Agent header sent |
| 119 | __version__ = sys.version[:3] |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 120 | |
| 121 | _opener = None |
| 122 | def urlopen(url, data=None): |
| 123 | global _opener |
| 124 | if _opener is None: |
| 125 | _opener = build_opener() |
| 126 | return _opener.open(url, data) |
| 127 | |
| 128 | def install_opener(opener): |
| 129 | global _opener |
| 130 | _opener = opener |
| 131 | |
| 132 | # do these error classes make sense? |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 133 | # make sure all of the IOError stuff is overridden. we just want to be |
Fred Drake | a87a521 | 2002-08-13 13:59:55 +0000 | [diff] [blame] | 134 | # subtypes. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 135 | |
| 136 | class URLError(IOError): |
| 137 | # URLError is a sub-type of IOError, but it doesn't share any of |
Jeremy Hylton | 0a4a50d | 2003-10-06 05:15:13 +0000 | [diff] [blame] | 138 | # the implementation. need to override __init__ and __str__. |
| 139 | # It sets self.args for compatibility with other EnvironmentError |
| 140 | # subclasses, but args doesn't have the typical format with errno in |
| 141 | # slot 0 and strerror in slot 1. This may be better than nothing. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 142 | def __init__(self, reason): |
Jeremy Hylton | 0a4a50d | 2003-10-06 05:15:13 +0000 | [diff] [blame] | 143 | self.args = reason, |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 144 | self.reason = reason |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 145 | |
| 146 | def __str__(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 147 | return '<urlopen error %s>' % self.reason |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 148 | |
| 149 | class HTTPError(URLError, addinfourl): |
| 150 | """Raised when HTTP error occurs, but also acts like non-error return""" |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 151 | __super_init = addinfourl.__init__ |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 152 | |
| 153 | def __init__(self, url, code, msg, hdrs, fp): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 154 | self.code = code |
| 155 | self.msg = msg |
| 156 | self.hdrs = hdrs |
| 157 | self.fp = fp |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 158 | self.filename = url |
Jeremy Hylton | 40bbae3 | 2002-06-03 16:53:00 +0000 | [diff] [blame] | 159 | # The addinfourl classes depend on fp being a valid file |
| 160 | # object. In some cases, the HTTPError may not have a valid |
| 161 | # file object. If this happens, the simplest workaround is to |
Tim Peters | c411dba | 2002-07-16 21:35:23 +0000 | [diff] [blame] | 162 | # not initialize the base classes. |
Jeremy Hylton | 40bbae3 | 2002-06-03 16:53:00 +0000 | [diff] [blame] | 163 | if fp is not None: |
| 164 | self.__super_init(fp, hdrs, url) |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 165 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 166 | def __str__(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 167 | return 'HTTP Error %s: %s' % (self.code, self.msg) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 168 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 169 | class GopherError(URLError): |
| 170 | pass |
| 171 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 172 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 173 | class Request: |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 174 | |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 175 | def __init__(self, url, data=None, headers={}, |
| 176 | origin_req_host=None, unverifiable=False): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 177 | # unwrap('<URL:type://host/path>') --> 'type://host/path' |
| 178 | self.__original = unwrap(url) |
| 179 | self.type = None |
| 180 | # self.__r_type is what's left after doing the splittype |
| 181 | self.host = None |
| 182 | self.port = None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 183 | self.data = data |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 184 | self.headers = {} |
Brett Cannon | c8b188a | 2003-05-17 19:51:26 +0000 | [diff] [blame] | 185 | for key, value in headers.items(): |
Brett Cannon | 86503b1 | 2003-05-12 07:29:42 +0000 | [diff] [blame] | 186 | self.add_header(key, value) |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 187 | self.unredirected_hdrs = {} |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 188 | if origin_req_host is None: |
| 189 | origin_req_host = cookielib.request_host(self) |
| 190 | self.origin_req_host = origin_req_host |
| 191 | self.unverifiable = unverifiable |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 192 | |
| 193 | def __getattr__(self, attr): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 194 | # XXX this is a fallback mechanism to guard against these |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 195 | # methods getting called in a non-standard order. this may be |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 196 | # too complicated and/or unnecessary. |
| 197 | # XXX should the __r_XXX attributes be public? |
| 198 | if attr[:12] == '_Request__r_': |
| 199 | name = attr[12:] |
| 200 | if hasattr(Request, 'get_' + name): |
| 201 | getattr(self, 'get_' + name)() |
| 202 | return getattr(self, attr) |
| 203 | raise AttributeError, attr |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 204 | |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 205 | def get_method(self): |
| 206 | if self.has_data(): |
| 207 | return "POST" |
| 208 | else: |
| 209 | return "GET" |
| 210 | |
Jeremy Hylton | 023518a | 2003-12-17 18:52:16 +0000 | [diff] [blame] | 211 | # XXX these helper methods are lame |
| 212 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 213 | def add_data(self, data): |
| 214 | self.data = data |
| 215 | |
| 216 | def has_data(self): |
| 217 | return self.data is not None |
| 218 | |
| 219 | def get_data(self): |
| 220 | return self.data |
| 221 | |
| 222 | def get_full_url(self): |
| 223 | return self.__original |
| 224 | |
| 225 | def get_type(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 226 | if self.type is None: |
| 227 | self.type, self.__r_type = splittype(self.__original) |
Jeremy Hylton | 78cae61 | 2001-05-09 15:49:24 +0000 | [diff] [blame] | 228 | if self.type is None: |
| 229 | raise ValueError, "unknown url type: %s" % self.__original |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 230 | return self.type |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 231 | |
| 232 | def get_host(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 233 | if self.host is None: |
| 234 | self.host, self.__r_host = splithost(self.__r_type) |
| 235 | if self.host: |
| 236 | self.host = unquote(self.host) |
| 237 | return self.host |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 238 | |
| 239 | def get_selector(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 240 | return self.__r_host |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 241 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 242 | def set_proxy(self, host, type): |
| 243 | self.host, self.type = host, type |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 244 | self.__r_host = self.__original |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 245 | |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 246 | def get_origin_req_host(self): |
| 247 | return self.origin_req_host |
| 248 | |
| 249 | def is_unverifiable(self): |
| 250 | return self.unverifiable |
| 251 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 252 | def add_header(self, key, val): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 253 | # useful for something like authentication |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 254 | self.headers[key.capitalize()] = val |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 255 | |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 256 | def add_unredirected_header(self, key, val): |
| 257 | # will not be added to a redirected request |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 258 | self.unredirected_hdrs[key.capitalize()] = val |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 259 | |
| 260 | def has_header(self, header_name): |
Neal Norwitz | 1cdd363 | 2004-06-07 03:49:50 +0000 | [diff] [blame] | 261 | return (header_name in self.headers or |
| 262 | header_name in self.unredirected_hdrs) |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 263 | |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 264 | def get_header(self, header_name, default=None): |
| 265 | return self.headers.get( |
| 266 | header_name, |
| 267 | self.unredirected_hdrs.get(header_name, default)) |
| 268 | |
| 269 | def header_items(self): |
| 270 | hdrs = self.unredirected_hdrs.copy() |
| 271 | hdrs.update(self.headers) |
| 272 | return hdrs.items() |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 273 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 274 | class OpenerDirector: |
| 275 | def __init__(self): |
Georg Brandl | 8d457c7 | 2005-06-26 22:01:35 +0000 | [diff] [blame] | 276 | client_version = "Python-urllib/%s" % __version__ |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 277 | self.addheaders = [('User-agent', client_version)] |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 278 | # manage the individual handlers |
| 279 | self.handlers = [] |
| 280 | self.handle_open = {} |
| 281 | self.handle_error = {} |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 282 | self.process_response = {} |
| 283 | self.process_request = {} |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 284 | |
| 285 | def add_handler(self, handler): |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 286 | added = False |
Jeremy Hylton | 8b78b99 | 2001-10-09 16:18:45 +0000 | [diff] [blame] | 287 | for meth in dir(handler): |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 288 | i = meth.find("_") |
| 289 | protocol = meth[:i] |
| 290 | condition = meth[i+1:] |
| 291 | |
| 292 | if condition.startswith("error"): |
Neal Norwitz | 1cdd363 | 2004-06-07 03:49:50 +0000 | [diff] [blame] | 293 | j = condition.find("_") + i + 1 |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 294 | kind = meth[j+1:] |
| 295 | try: |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 296 | kind = int(kind) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 297 | except ValueError: |
| 298 | pass |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 299 | lookup = self.handle_error.get(protocol, {}) |
| 300 | self.handle_error[protocol] = lookup |
| 301 | elif condition == "open": |
| 302 | kind = protocol |
Raymond Hettinger | f7bf02d | 2005-02-05 14:37:06 +0000 | [diff] [blame] | 303 | lookup = self.handle_open |
| 304 | elif condition == "response": |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 305 | kind = protocol |
Raymond Hettinger | f7bf02d | 2005-02-05 14:37:06 +0000 | [diff] [blame] | 306 | lookup = self.process_response |
| 307 | elif condition == "request": |
| 308 | kind = protocol |
| 309 | lookup = self.process_request |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 310 | else: |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 311 | continue |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 312 | |
| 313 | handlers = lookup.setdefault(kind, []) |
| 314 | if handlers: |
| 315 | bisect.insort(handlers, handler) |
| 316 | else: |
| 317 | handlers.append(handler) |
| 318 | added = True |
| 319 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 320 | if added: |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 321 | # XXX why does self.handlers need to be sorted? |
| 322 | bisect.insort(self.handlers, handler) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 323 | handler.add_parent(self) |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 324 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 325 | def close(self): |
Jeremy Hylton | dce391c | 2003-12-15 16:08:48 +0000 | [diff] [blame] | 326 | # Only exists for backwards compatibility. |
| 327 | pass |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 328 | |
| 329 | def _call_chain(self, chain, kind, meth_name, *args): |
Georg Brandl | c5ffd91 | 2006-04-02 20:48:11 +0000 | [diff] [blame] | 330 | # Handlers raise an exception if no one else should try to handle |
| 331 | # the request, or return None if they can't but another handler |
| 332 | # could. Otherwise, they return the response. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 333 | handlers = chain.get(kind, ()) |
| 334 | for handler in handlers: |
| 335 | func = getattr(handler, meth_name) |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 336 | |
| 337 | result = func(*args) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 338 | if result is not None: |
| 339 | return result |
| 340 | |
| 341 | def open(self, fullurl, data=None): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 342 | # accept a URL or a Request object |
Walter Dörwald | 65230a2 | 2002-06-03 15:58:32 +0000 | [diff] [blame] | 343 | if isinstance(fullurl, basestring): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 344 | req = Request(fullurl, data) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 345 | else: |
| 346 | req = fullurl |
| 347 | if data is not None: |
| 348 | req.add_data(data) |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 349 | |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 350 | protocol = req.get_type() |
| 351 | |
| 352 | # pre-process request |
| 353 | meth_name = protocol+"_request" |
| 354 | for processor in self.process_request.get(protocol, []): |
| 355 | meth = getattr(processor, meth_name) |
| 356 | req = meth(req) |
| 357 | |
| 358 | response = self._open(req, data) |
| 359 | |
| 360 | # post-process response |
| 361 | meth_name = protocol+"_response" |
| 362 | for processor in self.process_response.get(protocol, []): |
| 363 | meth = getattr(processor, meth_name) |
| 364 | response = meth(req, response) |
| 365 | |
| 366 | return response |
| 367 | |
| 368 | def _open(self, req, data=None): |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 369 | result = self._call_chain(self.handle_open, 'default', |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 370 | 'default_open', req) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 371 | if result: |
| 372 | return result |
| 373 | |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 374 | protocol = req.get_type() |
| 375 | result = self._call_chain(self.handle_open, protocol, protocol + |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 376 | '_open', req) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 377 | if result: |
| 378 | return result |
| 379 | |
| 380 | return self._call_chain(self.handle_open, 'unknown', |
| 381 | 'unknown_open', req) |
| 382 | |
| 383 | def error(self, proto, *args): |
Raymond Hettinger | dbecd93 | 2005-02-06 06:57:08 +0000 | [diff] [blame] | 384 | if proto in ('http', 'https'): |
Fred Drake | d5214b0 | 2001-11-08 17:19:29 +0000 | [diff] [blame] | 385 | # XXX http[s] protocols are special-cased |
| 386 | dict = self.handle_error['http'] # https is not different than http |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 387 | proto = args[2] # YUCK! |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 388 | meth_name = 'http_error_%s' % proto |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 389 | http_err = 1 |
| 390 | orig_args = args |
| 391 | else: |
| 392 | dict = self.handle_error |
| 393 | meth_name = proto + '_error' |
| 394 | http_err = 0 |
| 395 | args = (dict, proto, meth_name) + args |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 396 | result = self._call_chain(*args) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 397 | if result: |
| 398 | return result |
| 399 | |
| 400 | if http_err: |
| 401 | args = (dict, 'default', 'http_error_default') + orig_args |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 402 | return self._call_chain(*args) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 403 | |
Gustavo Niemeyer | 9556fba | 2003-06-07 17:53:08 +0000 | [diff] [blame] | 404 | # XXX probably also want an abstract factory that knows when it makes |
| 405 | # sense to skip a superclass in favor of a subclass and when it might |
| 406 | # make sense to include both |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 407 | |
| 408 | def build_opener(*handlers): |
| 409 | """Create an opener object from a list of handlers. |
| 410 | |
| 411 | The opener will use several default handlers, including support |
Gustavo Niemeyer | 9556fba | 2003-06-07 17:53:08 +0000 | [diff] [blame] | 412 | for HTTP and FTP. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 413 | |
| 414 | If any of the handlers passed as arguments are subclasses of the |
| 415 | default handlers, the default handlers will not be used. |
| 416 | """ |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 417 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 418 | opener = OpenerDirector() |
| 419 | default_classes = [ProxyHandler, UnknownHandler, HTTPHandler, |
| 420 | HTTPDefaultErrorHandler, HTTPRedirectHandler, |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 421 | FTPHandler, FileHandler, HTTPErrorProcessor] |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 422 | if hasattr(httplib, 'HTTPS'): |
| 423 | default_classes.append(HTTPSHandler) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 424 | skip = [] |
| 425 | for klass in default_classes: |
| 426 | for check in handlers: |
Jeremy Hylton | 8b78b99 | 2001-10-09 16:18:45 +0000 | [diff] [blame] | 427 | if inspect.isclass(check): |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 428 | if issubclass(check, klass): |
| 429 | skip.append(klass) |
Jeremy Hylton | 8b78b99 | 2001-10-09 16:18:45 +0000 | [diff] [blame] | 430 | elif isinstance(check, klass): |
| 431 | skip.append(klass) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 432 | for klass in skip: |
| 433 | default_classes.remove(klass) |
| 434 | |
| 435 | for klass in default_classes: |
| 436 | opener.add_handler(klass()) |
| 437 | |
| 438 | for h in handlers: |
Jeremy Hylton | 8b78b99 | 2001-10-09 16:18:45 +0000 | [diff] [blame] | 439 | if inspect.isclass(h): |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 440 | h = h() |
| 441 | opener.add_handler(h) |
| 442 | return opener |
| 443 | |
| 444 | class BaseHandler: |
Gustavo Niemeyer | 9556fba | 2003-06-07 17:53:08 +0000 | [diff] [blame] | 445 | handler_order = 500 |
| 446 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 447 | def add_parent(self, parent): |
| 448 | self.parent = parent |
Tim Peters | 58eb11c | 2004-01-18 20:29:55 +0000 | [diff] [blame] | 449 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 450 | def close(self): |
Jeremy Hylton | dce391c | 2003-12-15 16:08:48 +0000 | [diff] [blame] | 451 | # Only exists for backwards compatibility |
| 452 | pass |
Tim Peters | 58eb11c | 2004-01-18 20:29:55 +0000 | [diff] [blame] | 453 | |
Gustavo Niemeyer | 9556fba | 2003-06-07 17:53:08 +0000 | [diff] [blame] | 454 | def __lt__(self, other): |
| 455 | if not hasattr(other, "handler_order"): |
| 456 | # Try to preserve the old behavior of having custom classes |
| 457 | # inserted after default ones (works only for custom user |
| 458 | # classes which are not aware of handler_order). |
| 459 | return True |
| 460 | return self.handler_order < other.handler_order |
Tim Peters | f545baa | 2003-06-15 23:26:30 +0000 | [diff] [blame] | 461 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 462 | |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 463 | class HTTPErrorProcessor(BaseHandler): |
| 464 | """Process HTTP error responses.""" |
| 465 | handler_order = 1000 # after all other processing |
| 466 | |
| 467 | def http_response(self, request, response): |
| 468 | code, msg, hdrs = response.code, response.msg, response.info() |
| 469 | |
Andrew M. Kuchling | 08c08bb | 2004-06-29 13:19:19 +0000 | [diff] [blame] | 470 | if code not in (200, 206): |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 471 | response = self.parent.error( |
| 472 | 'http', request, response, code, msg, hdrs) |
| 473 | |
| 474 | return response |
| 475 | |
| 476 | https_response = http_response |
| 477 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 478 | class HTTPDefaultErrorHandler(BaseHandler): |
| 479 | def http_error_default(self, req, fp, code, msg, hdrs): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 480 | raise HTTPError(req.get_full_url(), code, msg, hdrs, fp) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 481 | |
| 482 | class HTTPRedirectHandler(BaseHandler): |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 483 | # maximum number of redirections to any single URL |
| 484 | # this is needed because of the state that cookies introduce |
| 485 | max_repeats = 4 |
| 486 | # maximum total number of redirections (regardless of URL) before |
| 487 | # assuming we're in a loop |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 488 | max_redirections = 10 |
| 489 | |
Jeremy Hylton | 0389295 | 2003-05-05 04:09:13 +0000 | [diff] [blame] | 490 | def redirect_request(self, req, fp, code, msg, headers, newurl): |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 491 | """Return a Request or None in response to a redirect. |
| 492 | |
Jeremy Hylton | aefae55 | 2003-07-10 13:30:12 +0000 | [diff] [blame] | 493 | This is called by the http_error_30x methods when a |
| 494 | redirection response is received. If a redirection should |
| 495 | take place, return a new Request to allow http_error_30x to |
| 496 | perform the redirect. Otherwise, raise HTTPError if no-one |
| 497 | else should try to handle this url. Return None if you can't |
| 498 | but another Handler might. |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 499 | """ |
Jeremy Hylton | 828023b | 2003-05-04 23:44:49 +0000 | [diff] [blame] | 500 | m = req.get_method() |
| 501 | if (code in (301, 302, 303, 307) and m in ("GET", "HEAD") |
Martin v. Löwis | 162f081 | 2003-07-12 07:33:32 +0000 | [diff] [blame] | 502 | or code in (301, 302, 303) and m == "POST"): |
| 503 | # Strictly (according to RFC 2616), 301 or 302 in response |
| 504 | # to a POST MUST NOT cause a redirection without confirmation |
Jeremy Hylton | 828023b | 2003-05-04 23:44:49 +0000 | [diff] [blame] | 505 | # from the user (of urllib2, in this case). In practice, |
| 506 | # essentially all clients do redirect in this case, so we |
| 507 | # do the same. |
Georg Brandl | ddb84d7 | 2006-03-18 11:35:18 +0000 | [diff] [blame] | 508 | # be conciliant with URIs containing a space |
| 509 | newurl = newurl.replace(' ', '%20') |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 510 | return Request(newurl, |
| 511 | headers=req.headers, |
| 512 | origin_req_host=req.get_origin_req_host(), |
| 513 | unverifiable=True) |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 514 | else: |
Martin v. Löwis | e3b67bc | 2003-06-14 05:51:25 +0000 | [diff] [blame] | 515 | raise HTTPError(req.get_full_url(), code, msg, headers, fp) |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 516 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 517 | # Implementation note: To avoid the server sending us into an |
| 518 | # infinite loop, the request object needs to track what URLs we |
| 519 | # have already seen. Do this by adding a handler-specific |
| 520 | # attribute to the Request object. |
| 521 | def http_error_302(self, req, fp, code, msg, headers): |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 522 | # Some servers (incorrectly) return multiple Location headers |
| 523 | # (so probably same goes for URI). Use first header. |
Raymond Hettinger | 54f0222 | 2002-06-01 14:18:47 +0000 | [diff] [blame] | 524 | if 'location' in headers: |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 525 | newurl = headers.getheaders('location')[0] |
Raymond Hettinger | 54f0222 | 2002-06-01 14:18:47 +0000 | [diff] [blame] | 526 | elif 'uri' in headers: |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 527 | newurl = headers.getheaders('uri')[0] |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 528 | else: |
| 529 | return |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 530 | newurl = urlparse.urljoin(req.get_full_url(), newurl) |
| 531 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 532 | # XXX Probably want to forget about the state of the current |
| 533 | # request, although that might interact poorly with other |
| 534 | # handlers that also use handler-specific request attributes |
Jeremy Hylton | 0389295 | 2003-05-05 04:09:13 +0000 | [diff] [blame] | 535 | new = self.redirect_request(req, fp, code, msg, headers, newurl) |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 536 | if new is None: |
| 537 | return |
| 538 | |
| 539 | # loop detection |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 540 | # .redirect_dict has a key url if url was previously visited. |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 541 | if hasattr(req, 'redirect_dict'): |
| 542 | visited = new.redirect_dict = req.redirect_dict |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 543 | if (visited.get(newurl, 0) >= self.max_repeats or |
| 544 | len(visited) >= self.max_redirections): |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 545 | raise HTTPError(req.get_full_url(), code, |
Jeremy Hylton | 54e99e8 | 2001-08-07 21:12:25 +0000 | [diff] [blame] | 546 | self.inf_msg + msg, headers, fp) |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 547 | else: |
| 548 | visited = new.redirect_dict = req.redirect_dict = {} |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 549 | visited[newurl] = visited.get(newurl, 0) + 1 |
Jeremy Hylton | 54e99e8 | 2001-08-07 21:12:25 +0000 | [diff] [blame] | 550 | |
| 551 | # Don't close the fp until we are sure that we won't use it |
Tim Peters | ab9ba27 | 2001-08-09 21:40:30 +0000 | [diff] [blame] | 552 | # with HTTPError. |
Jeremy Hylton | 54e99e8 | 2001-08-07 21:12:25 +0000 | [diff] [blame] | 553 | fp.read() |
| 554 | fp.close() |
| 555 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 556 | return self.parent.open(new) |
| 557 | |
Raymond Hettinger | 024aaa1 | 2003-04-24 15:32:12 +0000 | [diff] [blame] | 558 | http_error_301 = http_error_303 = http_error_307 = http_error_302 |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 559 | |
Martin v. Löwis | 162f081 | 2003-07-12 07:33:32 +0000 | [diff] [blame] | 560 | inf_msg = "The HTTP server returned a redirect error that would " \ |
Thomas Wouters | 7e47402 | 2000-07-16 12:04:32 +0000 | [diff] [blame] | 561 | "lead to an infinite loop.\n" \ |
Martin v. Löwis | 162f081 | 2003-07-12 07:33:32 +0000 | [diff] [blame] | 562 | "The last 30x error message was:\n" |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 563 | |
Georg Brandl | 720096a | 2006-04-02 20:45:34 +0000 | [diff] [blame] | 564 | |
| 565 | def _parse_proxy(proxy): |
| 566 | """Return (scheme, user, password, host/port) given a URL or an authority. |
| 567 | |
| 568 | If a URL is supplied, it must have an authority (host:port) component. |
| 569 | According to RFC 3986, having an authority component means the URL must |
| 570 | have two slashes after the scheme: |
| 571 | |
| 572 | >>> _parse_proxy('file:/ftp.example.com/') |
| 573 | Traceback (most recent call last): |
| 574 | ValueError: proxy URL with no authority: 'file:/ftp.example.com/' |
| 575 | |
| 576 | The first three items of the returned tuple may be None. |
| 577 | |
| 578 | Examples of authority parsing: |
| 579 | |
| 580 | >>> _parse_proxy('proxy.example.com') |
| 581 | (None, None, None, 'proxy.example.com') |
| 582 | >>> _parse_proxy('proxy.example.com:3128') |
| 583 | (None, None, None, 'proxy.example.com:3128') |
| 584 | |
| 585 | The authority component may optionally include userinfo (assumed to be |
| 586 | username:password): |
| 587 | |
| 588 | >>> _parse_proxy('joe:password@proxy.example.com') |
| 589 | (None, 'joe', 'password', 'proxy.example.com') |
| 590 | >>> _parse_proxy('joe:password@proxy.example.com:3128') |
| 591 | (None, 'joe', 'password', 'proxy.example.com:3128') |
| 592 | |
| 593 | Same examples, but with URLs instead: |
| 594 | |
| 595 | >>> _parse_proxy('http://proxy.example.com/') |
| 596 | ('http', None, None, 'proxy.example.com') |
| 597 | >>> _parse_proxy('http://proxy.example.com:3128/') |
| 598 | ('http', None, None, 'proxy.example.com:3128') |
| 599 | >>> _parse_proxy('http://joe:password@proxy.example.com/') |
| 600 | ('http', 'joe', 'password', 'proxy.example.com') |
| 601 | >>> _parse_proxy('http://joe:password@proxy.example.com:3128') |
| 602 | ('http', 'joe', 'password', 'proxy.example.com:3128') |
| 603 | |
| 604 | Everything after the authority is ignored: |
| 605 | |
| 606 | >>> _parse_proxy('ftp://joe:password@proxy.example.com/rubbish:3128') |
| 607 | ('ftp', 'joe', 'password', 'proxy.example.com') |
| 608 | |
| 609 | Test for no trailing '/' case: |
| 610 | |
| 611 | >>> _parse_proxy('http://joe:password@proxy.example.com') |
| 612 | ('http', 'joe', 'password', 'proxy.example.com') |
| 613 | |
| 614 | """ |
| 615 | from urlparse import _splitnetloc |
| 616 | scheme, r_scheme = splittype(proxy) |
| 617 | if not r_scheme.startswith("/"): |
| 618 | # authority |
| 619 | scheme = None |
| 620 | authority = proxy |
| 621 | else: |
| 622 | # URL |
| 623 | if not r_scheme.startswith("//"): |
| 624 | raise ValueError("proxy URL with no authority: %r" % proxy) |
| 625 | # We have an authority, so for RFC 3986-compliant URLs (by ss 3. |
| 626 | # and 3.3.), path is empty or starts with '/' |
| 627 | end = r_scheme.find("/", 2) |
| 628 | if end == -1: |
| 629 | end = None |
| 630 | authority = r_scheme[2:end] |
| 631 | userinfo, hostport = splituser(authority) |
| 632 | if userinfo is not None: |
| 633 | user, password = splitpasswd(userinfo) |
| 634 | else: |
| 635 | user = password = None |
| 636 | return scheme, user, password, hostport |
| 637 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 638 | class ProxyHandler(BaseHandler): |
Gustavo Niemeyer | 9556fba | 2003-06-07 17:53:08 +0000 | [diff] [blame] | 639 | # Proxies must be in front |
| 640 | handler_order = 100 |
| 641 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 642 | def __init__(self, proxies=None): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 643 | if proxies is None: |
| 644 | proxies = getproxies() |
| 645 | assert hasattr(proxies, 'has_key'), "proxies must be a mapping" |
| 646 | self.proxies = proxies |
Brett Cannon | df0d87a | 2003-05-18 02:25:07 +0000 | [diff] [blame] | 647 | for type, url in proxies.items(): |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 648 | setattr(self, '%s_open' % type, |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 649 | lambda r, proxy=url, type=type, meth=self.proxy_open: \ |
| 650 | meth(r, proxy, type)) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 651 | |
| 652 | def proxy_open(self, req, proxy, type): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 653 | orig_type = req.get_type() |
Georg Brandl | 720096a | 2006-04-02 20:45:34 +0000 | [diff] [blame] | 654 | proxy_type, user, password, hostport = _parse_proxy(proxy) |
| 655 | if proxy_type is None: |
| 656 | proxy_type = orig_type |
Georg Brandl | 531ceba | 2006-01-21 07:20:56 +0000 | [diff] [blame] | 657 | if user and password: |
Georg Brandl | 720096a | 2006-04-02 20:45:34 +0000 | [diff] [blame] | 658 | user_pass = '%s:%s' % (unquote(user), unquote(password)) |
| 659 | creds = base64.encodestring(user_pass).strip() |
| 660 | req.add_header('Proxy-authorization', 'Basic ' + creds) |
| 661 | hostport = unquote(hostport) |
| 662 | req.set_proxy(hostport, proxy_type) |
| 663 | if orig_type == proxy_type: |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 664 | # let other handlers take care of it |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 665 | return None |
| 666 | else: |
| 667 | # need to start over, because the other handlers don't |
| 668 | # grok the proxy's URL type |
Georg Brandl | 720096a | 2006-04-02 20:45:34 +0000 | [diff] [blame] | 669 | # e.g. if we have a constructor arg proxies like so: |
| 670 | # {'http': 'ftp://proxy.example.com'}, we may end up turning |
| 671 | # a request for http://acme.example.com/a into one for |
| 672 | # ftp://proxy.example.com/a |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 673 | return self.parent.open(req) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 674 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 675 | class HTTPPasswordMgr: |
| 676 | def __init__(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 677 | self.passwd = {} |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 678 | |
| 679 | def add_password(self, realm, uri, user, passwd): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 680 | # uri could be a single URI or a sequence |
Walter Dörwald | 65230a2 | 2002-06-03 15:58:32 +0000 | [diff] [blame] | 681 | if isinstance(uri, basestring): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 682 | uri = [uri] |
| 683 | uri = tuple(map(self.reduce_uri, uri)) |
Raymond Hettinger | 54f0222 | 2002-06-01 14:18:47 +0000 | [diff] [blame] | 684 | if not realm in self.passwd: |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 685 | self.passwd[realm] = {} |
| 686 | self.passwd[realm][uri] = (user, passwd) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 687 | |
| 688 | def find_user_password(self, realm, authuri): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 689 | domains = self.passwd.get(realm, {}) |
| 690 | authuri = self.reduce_uri(authuri) |
Brett Cannon | 86503b1 | 2003-05-12 07:29:42 +0000 | [diff] [blame] | 691 | for uris, authinfo in domains.iteritems(): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 692 | for uri in uris: |
| 693 | if self.is_suburi(uri, authuri): |
| 694 | return authinfo |
| 695 | return None, None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 696 | |
| 697 | def reduce_uri(self, uri): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 698 | """Accept netloc or URI and extract only the netloc and path""" |
| 699 | parts = urlparse.urlparse(uri) |
| 700 | if parts[1]: |
| 701 | return parts[1], parts[2] or '/' |
| 702 | else: |
| 703 | return parts[2], '/' |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 704 | |
| 705 | def is_suburi(self, base, test): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 706 | """Check if test is below base in a URI tree |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 707 | |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 708 | Both args must be URIs in reduced form. |
| 709 | """ |
| 710 | if base == test: |
Guido van Rossum | 8ca162f | 2002-04-07 06:36:23 +0000 | [diff] [blame] | 711 | return True |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 712 | if base[0] != test[0]: |
Guido van Rossum | 8ca162f | 2002-04-07 06:36:23 +0000 | [diff] [blame] | 713 | return False |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 714 | common = posixpath.commonprefix((base[1], test[1])) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 715 | if len(common) == len(base[1]): |
Guido van Rossum | 8ca162f | 2002-04-07 06:36:23 +0000 | [diff] [blame] | 716 | return True |
| 717 | return False |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 718 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 719 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 720 | class HTTPPasswordMgrWithDefaultRealm(HTTPPasswordMgr): |
| 721 | |
| 722 | def find_user_password(self, realm, authuri): |
Jeremy Hylton | aefae55 | 2003-07-10 13:30:12 +0000 | [diff] [blame] | 723 | user, password = HTTPPasswordMgr.find_user_password(self, realm, |
| 724 | authuri) |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 725 | if user is not None: |
| 726 | return user, password |
| 727 | return HTTPPasswordMgr.find_user_password(self, None, authuri) |
| 728 | |
| 729 | |
| 730 | class AbstractBasicAuthHandler: |
| 731 | |
Neal Norwitz | 853ddd5 | 2002-10-09 23:17:04 +0000 | [diff] [blame] | 732 | rx = re.compile('[ \t]*([^ \t]+)[ \t]+realm="([^"]*)"', re.I) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 733 | |
| 734 | # XXX there can actually be multiple auth-schemes in a |
| 735 | # www-authenticate header. should probably be a lot more careful |
| 736 | # in parsing them to extract multiple alternatives |
| 737 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 738 | def __init__(self, password_mgr=None): |
| 739 | if password_mgr is None: |
| 740 | password_mgr = HTTPPasswordMgr() |
| 741 | self.passwd = password_mgr |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 742 | self.add_password = self.passwd.add_password |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 743 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 744 | def http_error_auth_reqed(self, authreq, host, req, headers): |
| 745 | # XXX could be multiple headers |
| 746 | authreq = headers.get(authreq, None) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 747 | if authreq: |
Martin v. Löwis | 65a7975 | 2004-08-03 12:59:55 +0000 | [diff] [blame] | 748 | mo = AbstractBasicAuthHandler.rx.search(authreq) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 749 | if mo: |
| 750 | scheme, realm = mo.groups() |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 751 | if scheme.lower() == 'basic': |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 752 | return self.retry_http_basic_auth(host, req, realm) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 753 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 754 | def retry_http_basic_auth(self, host, req, realm): |
Jeremy Hylton | b300ae3 | 2004-12-22 14:27:19 +0000 | [diff] [blame] | 755 | # TODO(jhylton): Remove the host argument? It depends on whether |
| 756 | # retry_http_basic_auth() is consider part of the public API. |
| 757 | # It probably is. |
| 758 | user, pw = self.passwd.find_user_password(realm, req.get_full_url()) |
Martin v. Löwis | 8b3e871 | 2004-05-06 01:41:26 +0000 | [diff] [blame] | 759 | if pw is not None: |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 760 | raw = "%s:%s" % (user, pw) |
Jeremy Hylton | 52a17be | 2001-11-09 16:46:51 +0000 | [diff] [blame] | 761 | auth = 'Basic %s' % base64.encodestring(raw).strip() |
| 762 | if req.headers.get(self.auth_header, None) == auth: |
| 763 | return None |
| 764 | req.add_header(self.auth_header, auth) |
| 765 | return self.parent.open(req) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 766 | else: |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 767 | return None |
| 768 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 769 | class HTTPBasicAuthHandler(AbstractBasicAuthHandler, BaseHandler): |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 770 | |
Jeremy Hylton | 52a17be | 2001-11-09 16:46:51 +0000 | [diff] [blame] | 771 | auth_header = 'Authorization' |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 772 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 773 | def http_error_401(self, req, fp, code, msg, headers): |
| 774 | host = urlparse.urlparse(req.get_full_url())[1] |
Tim Peters | 30edd23 | 2001-03-16 08:29:48 +0000 | [diff] [blame] | 775 | return self.http_error_auth_reqed('www-authenticate', |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 776 | host, req, headers) |
| 777 | |
| 778 | |
| 779 | class ProxyBasicAuthHandler(AbstractBasicAuthHandler, BaseHandler): |
| 780 | |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 781 | auth_header = 'Proxy-authorization' |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 782 | |
| 783 | def http_error_407(self, req, fp, code, msg, headers): |
| 784 | host = req.get_host() |
Tim Peters | 30edd23 | 2001-03-16 08:29:48 +0000 | [diff] [blame] | 785 | return self.http_error_auth_reqed('proxy-authenticate', |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 786 | host, req, headers) |
| 787 | |
| 788 | |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 789 | def randombytes(n): |
| 790 | """Return n random bytes.""" |
| 791 | # Use /dev/urandom if it is available. Fall back to random module |
| 792 | # if not. It might be worthwhile to extend this function to use |
| 793 | # other platform-specific mechanisms for getting random bytes. |
| 794 | if os.path.exists("/dev/urandom"): |
| 795 | f = open("/dev/urandom") |
| 796 | s = f.read(n) |
| 797 | f.close() |
| 798 | return s |
| 799 | else: |
| 800 | L = [chr(random.randrange(0, 256)) for i in range(n)] |
| 801 | return "".join(L) |
| 802 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 803 | class AbstractDigestAuthHandler: |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 804 | # Digest authentication is specified in RFC 2617. |
| 805 | |
| 806 | # XXX The client does not inspect the Authentication-Info header |
| 807 | # in a successful response. |
| 808 | |
| 809 | # XXX It should be possible to test this implementation against |
| 810 | # a mock server that just generates a static set of challenges. |
| 811 | |
| 812 | # XXX qop="auth-int" supports is shaky |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 813 | |
| 814 | def __init__(self, passwd=None): |
| 815 | if passwd is None: |
Jeremy Hylton | 54e99e8 | 2001-08-07 21:12:25 +0000 | [diff] [blame] | 816 | passwd = HTTPPasswordMgr() |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 817 | self.passwd = passwd |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 818 | self.add_password = self.passwd.add_password |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 819 | self.retried = 0 |
| 820 | self.nonce_count = 0 |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 821 | |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 822 | def reset_retry_count(self): |
| 823 | self.retried = 0 |
| 824 | |
| 825 | def http_error_auth_reqed(self, auth_header, host, req, headers): |
| 826 | authreq = headers.get(auth_header, None) |
| 827 | if self.retried > 5: |
| 828 | # Don't fail endlessly - if we failed once, we'll probably |
| 829 | # fail a second time. Hm. Unless the Password Manager is |
| 830 | # prompting for the information. Crap. This isn't great |
| 831 | # but it's better than the current 'repeat until recursion |
| 832 | # depth exceeded' approach <wink> |
Tim Peters | 58eb11c | 2004-01-18 20:29:55 +0000 | [diff] [blame] | 833 | raise HTTPError(req.get_full_url(), 401, "digest auth failed", |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 834 | headers, None) |
| 835 | else: |
| 836 | self.retried += 1 |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 837 | if authreq: |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 838 | scheme = authreq.split()[0] |
| 839 | if scheme.lower() == 'digest': |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 840 | return self.retry_http_digest_auth(req, authreq) |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 841 | else: |
| 842 | raise ValueError("AbstractDigestAuthHandler doesn't know " |
| 843 | "about %s"%(scheme)) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 844 | |
| 845 | def retry_http_digest_auth(self, req, auth): |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 846 | token, challenge = auth.split(' ', 1) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 847 | chal = parse_keqv_list(parse_http_list(challenge)) |
| 848 | auth = self.get_authorization(req, chal) |
| 849 | if auth: |
Jeremy Hylton | 52a17be | 2001-11-09 16:46:51 +0000 | [diff] [blame] | 850 | auth_val = 'Digest %s' % auth |
| 851 | if req.headers.get(self.auth_header, None) == auth_val: |
| 852 | return None |
| 853 | req.add_header(self.auth_header, auth_val) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 854 | resp = self.parent.open(req) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 855 | return resp |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 856 | |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 857 | def get_cnonce(self, nonce): |
| 858 | # The cnonce-value is an opaque |
| 859 | # quoted string value provided by the client and used by both client |
| 860 | # and server to avoid chosen plaintext attacks, to provide mutual |
| 861 | # authentication, and to provide some message integrity protection. |
| 862 | # This isn't a fabulous effort, but it's probably Good Enough. |
| 863 | dig = sha.new("%s:%s:%s:%s" % (self.nonce_count, nonce, time.ctime(), |
| 864 | randombytes(8))).hexdigest() |
| 865 | return dig[:16] |
| 866 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 867 | def get_authorization(self, req, chal): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 868 | try: |
| 869 | realm = chal['realm'] |
| 870 | nonce = chal['nonce'] |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 871 | qop = chal.get('qop') |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 872 | algorithm = chal.get('algorithm', 'MD5') |
| 873 | # mod_digest doesn't send an opaque, even though it isn't |
| 874 | # supposed to be optional |
| 875 | opaque = chal.get('opaque', None) |
| 876 | except KeyError: |
| 877 | return None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 878 | |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 879 | H, KD = self.get_algorithm_impls(algorithm) |
| 880 | if H is None: |
| 881 | return None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 882 | |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 883 | user, pw = self.passwd.find_user_password(realm, req.get_full_url()) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 884 | if user is None: |
| 885 | return None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 886 | |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 887 | # XXX not implemented yet |
| 888 | if req.has_data(): |
| 889 | entdig = self.get_entity_digest(req.get_data(), chal) |
| 890 | else: |
| 891 | entdig = None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 892 | |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 893 | A1 = "%s:%s:%s" % (user, realm, pw) |
Johannes Gijsbers | cdd625a | 2005-01-09 05:51:49 +0000 | [diff] [blame] | 894 | A2 = "%s:%s" % (req.get_method(), |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 895 | # XXX selector: what about proxies and full urls |
| 896 | req.get_selector()) |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 897 | if qop == 'auth': |
| 898 | self.nonce_count += 1 |
| 899 | ncvalue = '%08x' % self.nonce_count |
| 900 | cnonce = self.get_cnonce(nonce) |
| 901 | noncebit = "%s:%s:%s:%s:%s" % (nonce, ncvalue, cnonce, qop, H(A2)) |
| 902 | respdig = KD(H(A1), noncebit) |
| 903 | elif qop is None: |
| 904 | respdig = KD(H(A1), "%s:%s" % (nonce, H(A2))) |
| 905 | else: |
| 906 | # XXX handle auth-int. |
| 907 | pass |
Tim Peters | 58eb11c | 2004-01-18 20:29:55 +0000 | [diff] [blame] | 908 | |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 909 | # XXX should the partial digests be encoded too? |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 910 | |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 911 | base = 'username="%s", realm="%s", nonce="%s", uri="%s", ' \ |
| 912 | 'response="%s"' % (user, realm, nonce, req.get_selector(), |
| 913 | respdig) |
| 914 | if opaque: |
Jeremy Hylton | b300ae3 | 2004-12-22 14:27:19 +0000 | [diff] [blame] | 915 | base += ', opaque="%s"' % opaque |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 916 | if entdig: |
Jeremy Hylton | b300ae3 | 2004-12-22 14:27:19 +0000 | [diff] [blame] | 917 | base += ', digest="%s"' % entdig |
| 918 | base += ', algorithm="%s"' % algorithm |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 919 | if qop: |
Jeremy Hylton | b300ae3 | 2004-12-22 14:27:19 +0000 | [diff] [blame] | 920 | base += ', qop=auth, nc=%s, cnonce="%s"' % (ncvalue, cnonce) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 921 | return base |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 922 | |
| 923 | def get_algorithm_impls(self, algorithm): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 924 | # lambdas assume digest modules are imported at the top level |
| 925 | if algorithm == 'MD5': |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 926 | H = lambda x: md5.new(x).hexdigest() |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 927 | elif algorithm == 'SHA': |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 928 | H = lambda x: sha.new(x).hexdigest() |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 929 | # XXX MD5-sess |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 930 | KD = lambda s, d: H("%s:%s" % (s, d)) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 931 | return H, KD |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 932 | |
| 933 | def get_entity_digest(self, data, chal): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 934 | # XXX not implemented yet |
| 935 | return None |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 936 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 937 | |
| 938 | class HTTPDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler): |
| 939 | """An authentication protocol defined by RFC 2069 |
| 940 | |
| 941 | Digest authentication improves on basic authentication because it |
| 942 | does not transmit passwords in the clear. |
| 943 | """ |
| 944 | |
Jeremy Hylton | aefae55 | 2003-07-10 13:30:12 +0000 | [diff] [blame] | 945 | auth_header = 'Authorization' |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 946 | |
| 947 | def http_error_401(self, req, fp, code, msg, headers): |
| 948 | host = urlparse.urlparse(req.get_full_url())[1] |
Tim Peters | 58eb11c | 2004-01-18 20:29:55 +0000 | [diff] [blame] | 949 | retry = self.http_error_auth_reqed('www-authenticate', |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 950 | host, req, headers) |
| 951 | self.reset_retry_count() |
| 952 | return retry |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 953 | |
| 954 | |
| 955 | class ProxyDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler): |
| 956 | |
Jeremy Hylton | aefae55 | 2003-07-10 13:30:12 +0000 | [diff] [blame] | 957 | auth_header = 'Proxy-Authorization' |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 958 | |
| 959 | def http_error_407(self, req, fp, code, msg, headers): |
| 960 | host = req.get_host() |
Tim Peters | 58eb11c | 2004-01-18 20:29:55 +0000 | [diff] [blame] | 961 | retry = self.http_error_auth_reqed('proxy-authenticate', |
Jeremy Hylton | fcefd0d | 2003-10-21 18:07:07 +0000 | [diff] [blame] | 962 | host, req, headers) |
| 963 | self.reset_retry_count() |
| 964 | return retry |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 965 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 966 | class AbstractHTTPHandler(BaseHandler): |
| 967 | |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 968 | def __init__(self, debuglevel=0): |
| 969 | self._debuglevel = debuglevel |
| 970 | |
| 971 | def set_http_debuglevel(self, level): |
| 972 | self._debuglevel = level |
| 973 | |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 974 | def do_request_(self, request): |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 975 | host = request.get_host() |
| 976 | if not host: |
| 977 | raise URLError('no host given') |
| 978 | |
| 979 | if request.has_data(): # POST |
| 980 | data = request.get_data() |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 981 | if not request.has_header('Content-type'): |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 982 | request.add_unredirected_header( |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 983 | 'Content-type', |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 984 | 'application/x-www-form-urlencoded') |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 985 | if not request.has_header('Content-length'): |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 986 | request.add_unredirected_header( |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 987 | 'Content-length', '%d' % len(data)) |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 988 | |
| 989 | scheme, sel = splittype(request.get_selector()) |
| 990 | sel_host, sel_path = splithost(sel) |
| 991 | if not request.has_header('Host'): |
| 992 | request.add_unredirected_header('Host', sel_host or host) |
| 993 | for name, value in self.parent.addheaders: |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 994 | name = name.capitalize() |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 995 | if not request.has_header(name): |
| 996 | request.add_unredirected_header(name, value) |
| 997 | |
| 998 | return request |
| 999 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 1000 | def do_open(self, http_class, req): |
Jeremy Hylton | 023518a | 2003-12-17 18:52:16 +0000 | [diff] [blame] | 1001 | """Return an addinfourl object for the request, using http_class. |
| 1002 | |
| 1003 | http_class must implement the HTTPConnection API from httplib. |
| 1004 | The addinfourl return value is a file-like object. It also |
| 1005 | has methods and attributes including: |
| 1006 | - info(): return a mimetools.Message object for the headers |
| 1007 | - geturl(): return the original request URL |
| 1008 | - code: HTTP status code |
| 1009 | """ |
Moshe Zadka | 7667680 | 2001-04-11 07:44:53 +0000 | [diff] [blame] | 1010 | host = req.get_host() |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1011 | if not host: |
| 1012 | raise URLError('no host given') |
| 1013 | |
Jeremy Hylton | 828023b | 2003-05-04 23:44:49 +0000 | [diff] [blame] | 1014 | h = http_class(host) # will parse host:port |
Jeremy Hylton | c1be59f | 2003-12-14 05:27:34 +0000 | [diff] [blame] | 1015 | h.set_debuglevel(self._debuglevel) |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1016 | |
Jeremy Hylton | 023518a | 2003-12-17 18:52:16 +0000 | [diff] [blame] | 1017 | headers = dict(req.headers) |
| 1018 | headers.update(req.unredirected_hdrs) |
Jeremy Hylton | b3ee6f9 | 2004-02-24 19:40:35 +0000 | [diff] [blame] | 1019 | # We want to make an HTTP/1.1 request, but the addinfourl |
| 1020 | # class isn't prepared to deal with a persistent connection. |
| 1021 | # It will try to read all remaining data from the socket, |
| 1022 | # which will block while the server waits for the next request. |
| 1023 | # So make sure the connection gets closed after the (only) |
| 1024 | # request. |
| 1025 | headers["Connection"] = "close" |
Jeremy Hylton | 828023b | 2003-05-04 23:44:49 +0000 | [diff] [blame] | 1026 | try: |
Jeremy Hylton | 023518a | 2003-12-17 18:52:16 +0000 | [diff] [blame] | 1027 | h.request(req.get_method(), req.get_selector(), req.data, headers) |
| 1028 | r = h.getresponse() |
| 1029 | except socket.error, err: # XXX what error? |
Jeremy Hylton | 828023b | 2003-05-04 23:44:49 +0000 | [diff] [blame] | 1030 | raise URLError(err) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1031 | |
Andrew M. Kuchling | f9ea7c0 | 2004-07-10 15:34:34 +0000 | [diff] [blame] | 1032 | # Pick apart the HTTPResponse object to get the addinfourl |
Jeremy Hylton | 5d9c303 | 2004-08-07 17:40:50 +0000 | [diff] [blame] | 1033 | # object initialized properly. |
| 1034 | |
| 1035 | # Wrap the HTTPResponse object in socket's file object adapter |
| 1036 | # for Windows. That adapter calls recv(), so delegate recv() |
| 1037 | # to read(). This weird wrapping allows the returned object to |
| 1038 | # have readline() and readlines() methods. |
Tim Peters | 9ca3f85 | 2004-08-08 01:05:14 +0000 | [diff] [blame] | 1039 | |
Jeremy Hylton | 5d9c303 | 2004-08-07 17:40:50 +0000 | [diff] [blame] | 1040 | # XXX It might be better to extract the read buffering code |
| 1041 | # out of socket._fileobject() and into a base class. |
Tim Peters | 9ca3f85 | 2004-08-08 01:05:14 +0000 | [diff] [blame] | 1042 | |
Jeremy Hylton | 5d9c303 | 2004-08-07 17:40:50 +0000 | [diff] [blame] | 1043 | r.recv = r.read |
| 1044 | fp = socket._fileobject(r) |
Tim Peters | 9ca3f85 | 2004-08-08 01:05:14 +0000 | [diff] [blame] | 1045 | |
Jeremy Hylton | 5d9c303 | 2004-08-07 17:40:50 +0000 | [diff] [blame] | 1046 | resp = addinfourl(fp, r.msg, req.get_full_url()) |
Andrew M. Kuchling | f9ea7c0 | 2004-07-10 15:34:34 +0000 | [diff] [blame] | 1047 | resp.code = r.status |
| 1048 | resp.msg = r.reason |
| 1049 | return resp |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1050 | |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 1051 | |
| 1052 | class HTTPHandler(AbstractHTTPHandler): |
| 1053 | |
| 1054 | def http_open(self, req): |
Jeremy Hylton | 023518a | 2003-12-17 18:52:16 +0000 | [diff] [blame] | 1055 | return self.do_open(httplib.HTTPConnection, req) |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 1056 | |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 1057 | http_request = AbstractHTTPHandler.do_request_ |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 1058 | |
| 1059 | if hasattr(httplib, 'HTTPS'): |
| 1060 | class HTTPSHandler(AbstractHTTPHandler): |
| 1061 | |
| 1062 | def https_open(self, req): |
Jeremy Hylton | 023518a | 2003-12-17 18:52:16 +0000 | [diff] [blame] | 1063 | return self.do_open(httplib.HTTPSConnection, req) |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 1064 | |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 1065 | https_request = AbstractHTTPHandler.do_request_ |
| 1066 | |
| 1067 | class HTTPCookieProcessor(BaseHandler): |
| 1068 | def __init__(self, cookiejar=None): |
| 1069 | if cookiejar is None: |
Neal Norwitz | 1cdd363 | 2004-06-07 03:49:50 +0000 | [diff] [blame] | 1070 | cookiejar = cookielib.CookieJar() |
Martin v. Löwis | 2a6ba90 | 2004-05-31 18:22:40 +0000 | [diff] [blame] | 1071 | self.cookiejar = cookiejar |
| 1072 | |
| 1073 | def http_request(self, request): |
| 1074 | self.cookiejar.add_cookie_header(request) |
| 1075 | return request |
| 1076 | |
| 1077 | def http_response(self, request, response): |
| 1078 | self.cookiejar.extract_cookies(response, request) |
| 1079 | return response |
| 1080 | |
| 1081 | https_request = http_request |
| 1082 | https_response = http_response |
Moshe Zadka | 8a18e99 | 2001-03-01 08:40:42 +0000 | [diff] [blame] | 1083 | |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1084 | class UnknownHandler(BaseHandler): |
| 1085 | def unknown_open(self, req): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1086 | type = req.get_type() |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1087 | raise URLError('unknown url type: %s' % type) |
| 1088 | |
| 1089 | def parse_keqv_list(l): |
| 1090 | """Parse list of key=value strings where keys are not duplicated.""" |
| 1091 | parsed = {} |
| 1092 | for elt in l: |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 1093 | k, v = elt.split('=', 1) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1094 | if v[0] == '"' and v[-1] == '"': |
| 1095 | v = v[1:-1] |
| 1096 | parsed[k] = v |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1097 | return parsed |
| 1098 | |
| 1099 | def parse_http_list(s): |
| 1100 | """Parse lists as described by RFC 2068 Section 2. |
Tim Peters | 9e34c04 | 2005-08-26 15:20:46 +0000 | [diff] [blame] | 1101 | |
Andrew M. Kuchling | 22ab06e | 2004-04-06 19:43:03 +0000 | [diff] [blame] | 1102 | In particular, parse comma-separated lists where the elements of |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1103 | the list may include quoted-strings. A quoted-string could |
Georg Brandl | e1b13d2 | 2005-08-24 22:20:32 +0000 | [diff] [blame] | 1104 | contain a comma. A non-quoted string could have quotes in the |
| 1105 | middle. Neither commas nor quotes count if they are escaped. |
| 1106 | Only double-quotes count, not single-quotes. |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1107 | """ |
Georg Brandl | e1b13d2 | 2005-08-24 22:20:32 +0000 | [diff] [blame] | 1108 | res = [] |
| 1109 | part = '' |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1110 | |
Georg Brandl | e1b13d2 | 2005-08-24 22:20:32 +0000 | [diff] [blame] | 1111 | escape = quote = False |
| 1112 | for cur in s: |
| 1113 | if escape: |
| 1114 | part += cur |
| 1115 | escape = False |
| 1116 | continue |
| 1117 | if quote: |
| 1118 | if cur == '\\': |
| 1119 | escape = True |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1120 | continue |
Georg Brandl | e1b13d2 | 2005-08-24 22:20:32 +0000 | [diff] [blame] | 1121 | elif cur == '"': |
| 1122 | quote = False |
| 1123 | part += cur |
| 1124 | continue |
| 1125 | |
| 1126 | if cur == ',': |
| 1127 | res.append(part) |
| 1128 | part = '' |
| 1129 | continue |
| 1130 | |
| 1131 | if cur == '"': |
| 1132 | quote = True |
Tim Peters | 9e34c04 | 2005-08-26 15:20:46 +0000 | [diff] [blame] | 1133 | |
Georg Brandl | e1b13d2 | 2005-08-24 22:20:32 +0000 | [diff] [blame] | 1134 | part += cur |
| 1135 | |
| 1136 | # append last part |
| 1137 | if part: |
| 1138 | res.append(part) |
| 1139 | |
| 1140 | return [part.strip() for part in res] |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1141 | |
| 1142 | class FileHandler(BaseHandler): |
| 1143 | # Use local file or FTP depending on form of URL |
| 1144 | def file_open(self, req): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1145 | url = req.get_selector() |
| 1146 | if url[:2] == '//' and url[2:3] != '/': |
| 1147 | req.type = 'ftp' |
| 1148 | return self.parent.open(req) |
| 1149 | else: |
| 1150 | return self.open_local_file(req) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1151 | |
| 1152 | # names for the localhost |
| 1153 | names = None |
| 1154 | def get_names(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1155 | if FileHandler.names is None: |
Georg Brandl | 4eb521e | 2006-04-02 20:37:17 +0000 | [diff] [blame] | 1156 | try: |
| 1157 | FileHandler.names = (socket.gethostbyname('localhost'), |
| 1158 | socket.gethostbyname(socket.gethostname())) |
| 1159 | except socket.gaierror: |
| 1160 | FileHandler.names = (socket.gethostbyname('localhost'),) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1161 | return FileHandler.names |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1162 | |
| 1163 | # not entirely sure what the rules are here |
| 1164 | def open_local_file(self, req): |
Anthony Baxter | 3dd9e46 | 2004-10-11 13:53:08 +0000 | [diff] [blame] | 1165 | import email.Utils |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1166 | host = req.get_host() |
| 1167 | file = req.get_selector() |
Jeremy Hylton | 6d8c1aa | 2001-08-27 20:16:53 +0000 | [diff] [blame] | 1168 | localfile = url2pathname(file) |
| 1169 | stats = os.stat(localfile) |
Martin v. Löwis | 9d3eba8 | 2002-03-18 08:37:19 +0000 | [diff] [blame] | 1170 | size = stats.st_size |
Anthony Baxter | 3dd9e46 | 2004-10-11 13:53:08 +0000 | [diff] [blame] | 1171 | modified = email.Utils.formatdate(stats.st_mtime, usegmt=True) |
Jeremy Hylton | 6d8c1aa | 2001-08-27 20:16:53 +0000 | [diff] [blame] | 1172 | mtype = mimetypes.guess_type(file)[0] |
Jeremy Hylton | 6d8c1aa | 2001-08-27 20:16:53 +0000 | [diff] [blame] | 1173 | headers = mimetools.Message(StringIO( |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 1174 | 'Content-type: %s\nContent-length: %d\nLast-modified: %s\n' % |
Jeremy Hylton | 6d8c1aa | 2001-08-27 20:16:53 +0000 | [diff] [blame] | 1175 | (mtype or 'text/plain', size, modified))) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1176 | if host: |
| 1177 | host, port = splitport(host) |
| 1178 | if not host or \ |
| 1179 | (not port and socket.gethostbyname(host) in self.get_names()): |
Jeremy Hylton | 6d8c1aa | 2001-08-27 20:16:53 +0000 | [diff] [blame] | 1180 | return addinfourl(open(localfile, 'rb'), |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1181 | headers, 'file:'+file) |
| 1182 | raise URLError('file not on local host') |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1183 | |
| 1184 | class FTPHandler(BaseHandler): |
| 1185 | def ftp_open(self, req): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1186 | host = req.get_host() |
| 1187 | if not host: |
| 1188 | raise IOError, ('ftp error', 'no host given') |
Martin v. Löwis | a79449e | 2004-02-15 21:19:18 +0000 | [diff] [blame] | 1189 | host, port = splitport(host) |
| 1190 | if port is None: |
| 1191 | port = ftplib.FTP_PORT |
Kurt B. Kaiser | 3f7cb5d | 2004-07-11 17:14:13 +0000 | [diff] [blame] | 1192 | else: |
| 1193 | port = int(port) |
Martin v. Löwis | a79449e | 2004-02-15 21:19:18 +0000 | [diff] [blame] | 1194 | |
| 1195 | # username/password handling |
| 1196 | user, host = splituser(host) |
| 1197 | if user: |
| 1198 | user, passwd = splitpasswd(user) |
| 1199 | else: |
| 1200 | passwd = None |
| 1201 | host = unquote(host) |
| 1202 | user = unquote(user or '') |
| 1203 | passwd = unquote(passwd or '') |
| 1204 | |
Jeremy Hylton | 73574ee | 2000-10-12 18:54:18 +0000 | [diff] [blame] | 1205 | try: |
| 1206 | host = socket.gethostbyname(host) |
| 1207 | except socket.error, msg: |
| 1208 | raise URLError(msg) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1209 | path, attrs = splitattr(req.get_selector()) |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 1210 | dirs = path.split('/') |
Martin v. Löwis | 7db04e7 | 2004-02-15 20:51:39 +0000 | [diff] [blame] | 1211 | dirs = map(unquote, dirs) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1212 | dirs, file = dirs[:-1], dirs[-1] |
| 1213 | if dirs and not dirs[0]: |
| 1214 | dirs = dirs[1:] |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1215 | try: |
| 1216 | fw = self.connect_ftp(user, passwd, host, port, dirs) |
| 1217 | type = file and 'I' or 'D' |
| 1218 | for attr in attrs: |
Kurt B. Kaiser | 3f7cb5d | 2004-07-11 17:14:13 +0000 | [diff] [blame] | 1219 | attr, value = splitvalue(attr) |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 1220 | if attr.lower() == 'type' and \ |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1221 | value in ('a', 'A', 'i', 'I', 'd', 'D'): |
Eric S. Raymond | b08b2d3 | 2001-02-09 11:10:16 +0000 | [diff] [blame] | 1222 | type = value.upper() |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1223 | fp, retrlen = fw.retrfile(file, type) |
Guido van Rossum | 833a8d8 | 2001-08-24 13:10:13 +0000 | [diff] [blame] | 1224 | headers = "" |
| 1225 | mtype = mimetypes.guess_type(req.get_full_url())[0] |
| 1226 | if mtype: |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 1227 | headers += "Content-type: %s\n" % mtype |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1228 | if retrlen is not None and retrlen >= 0: |
Georg Brandl | 80bb2bb | 2006-03-28 19:19:56 +0000 | [diff] [blame] | 1229 | headers += "Content-length: %d\n" % retrlen |
Guido van Rossum | 833a8d8 | 2001-08-24 13:10:13 +0000 | [diff] [blame] | 1230 | sf = StringIO(headers) |
| 1231 | headers = mimetools.Message(sf) |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1232 | return addinfourl(fp, headers, req.get_full_url()) |
| 1233 | except ftplib.all_errors, msg: |
| 1234 | raise IOError, ('ftp error', msg), sys.exc_info()[2] |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1235 | |
| 1236 | def connect_ftp(self, user, passwd, host, port, dirs): |
| 1237 | fw = ftpwrapper(user, passwd, host, port, dirs) |
| 1238 | ## fw.ftp.set_debuglevel(1) |
| 1239 | return fw |
| 1240 | |
| 1241 | class CacheFTPHandler(FTPHandler): |
| 1242 | # XXX would be nice to have pluggable cache strategies |
| 1243 | # XXX this stuff is definitely not thread safe |
| 1244 | def __init__(self): |
| 1245 | self.cache = {} |
| 1246 | self.timeout = {} |
| 1247 | self.soonest = 0 |
| 1248 | self.delay = 60 |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1249 | self.max_conns = 16 |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1250 | |
| 1251 | def setTimeout(self, t): |
| 1252 | self.delay = t |
| 1253 | |
| 1254 | def setMaxConns(self, m): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1255 | self.max_conns = m |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1256 | |
| 1257 | def connect_ftp(self, user, passwd, host, port, dirs): |
Mark Hammond | c533c98 | 2004-05-10 07:35:33 +0000 | [diff] [blame] | 1258 | key = user, host, port, '/'.join(dirs) |
Raymond Hettinger | 54f0222 | 2002-06-01 14:18:47 +0000 | [diff] [blame] | 1259 | if key in self.cache: |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1260 | self.timeout[key] = time.time() + self.delay |
| 1261 | else: |
| 1262 | self.cache[key] = ftpwrapper(user, passwd, host, port, dirs) |
| 1263 | self.timeout[key] = time.time() + self.delay |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1264 | self.check_cache() |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1265 | return self.cache[key] |
| 1266 | |
| 1267 | def check_cache(self): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1268 | # first check for old ones |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1269 | t = time.time() |
| 1270 | if self.soonest <= t: |
Raymond Hettinger | 4ec4fa2 | 2003-05-23 08:51:51 +0000 | [diff] [blame] | 1271 | for k, v in self.timeout.items(): |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1272 | if v < t: |
| 1273 | self.cache[k].close() |
| 1274 | del self.cache[k] |
| 1275 | del self.timeout[k] |
| 1276 | self.soonest = min(self.timeout.values()) |
| 1277 | |
| 1278 | # then check the size |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1279 | if len(self.cache) == self.max_conns: |
Brett Cannon | c8b188a | 2003-05-17 19:51:26 +0000 | [diff] [blame] | 1280 | for k, v in self.timeout.items(): |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1281 | if v == self.soonest: |
| 1282 | del self.cache[k] |
| 1283 | del self.timeout[k] |
| 1284 | break |
| 1285 | self.soonest = min(self.timeout.values()) |
Jeremy Hylton | 6d7e47b | 2000-01-20 18:19:08 +0000 | [diff] [blame] | 1286 | |
| 1287 | class GopherHandler(BaseHandler): |
| 1288 | def gopher_open(self, req): |
Georg Brandl | c5ffd91 | 2006-04-02 20:48:11 +0000 | [diff] [blame] | 1289 | # XXX can raise socket.error |
Tim Peters | e2c9a6c | 2006-02-18 04:14:16 +0000 | [diff] [blame] | 1290 | import gopherlib # this raises DeprecationWarning in 2.5 |
Fred Drake | 13a2c27 | 2000-02-10 17:17:14 +0000 | [diff] [blame] | 1291 | host = req.get_host() |
| 1292 | if not host: |
| 1293 | raise GopherError('no host given') |
| 1294 | host = unquote(host) |
| 1295 | selector = req.get_selector() |
| 1296 | type, selector = splitgophertype(selector) |
| 1297 | selector, query = splitquery(selector) |
| 1298 | selector = unquote(selector) |
| 1299 | if query: |
| 1300 | query = unquote(query) |
| 1301 | fp = gopherlib.send_query(selector, query, host) |
| 1302 | else: |
| 1303 | fp = gopherlib.send_selector(selector, host) |
| 1304 | return addinfourl(fp, noheaders(), req.get_full_url()) |