OLD | NEW |
1 # urllib3/poolmanager.py | 1 # urllib3/poolmanager.py |
2 # Copyright 2008-2013 Andrey Petrov and contributors (see CONTRIBUTORS.txt) | 2 # Copyright 2008-2013 Andrey Petrov and contributors (see CONTRIBUTORS.txt) |
3 # | 3 # |
4 # This module is part of urllib3 and is released under | 4 # This module is part of urllib3 and is released under |
5 # the MIT License: http://www.opensource.org/licenses/mit-license.php | 5 # the MIT License: http://www.opensource.org/licenses/mit-license.php |
6 | 6 |
7 import logging | 7 import logging |
8 | 8 |
| 9 try: # Python 3 |
| 10 from urllib.parse import urljoin |
| 11 except ImportError: |
| 12 from urlparse import urljoin |
| 13 |
9 from ._collections import RecentlyUsedContainer | 14 from ._collections import RecentlyUsedContainer |
10 from .connectionpool import HTTPConnectionPool, HTTPSConnectionPool | 15 from .connectionpool import HTTPConnectionPool, HTTPSConnectionPool |
11 from .connectionpool import connection_from_url, port_by_scheme | 16 from .connectionpool import port_by_scheme |
12 from .request import RequestMethods | 17 from .request import RequestMethods |
13 from .util import parse_url | 18 from .util import parse_url |
14 | 19 |
15 | 20 |
16 __all__ = ['PoolManager', 'ProxyManager', 'proxy_from_url'] | 21 __all__ = ['PoolManager', 'ProxyManager', 'proxy_from_url'] |
17 | 22 |
18 | 23 |
19 pool_classes_by_scheme = { | 24 pool_classes_by_scheme = { |
20 'http': HTTPConnectionPool, | 25 'http': HTTPConnectionPool, |
21 'https': HTTPSConnectionPool, | 26 'https': HTTPSConnectionPool, |
(...skipping 26 matching lines...) Expand all Loading... |
48 | 53 |
49 >>> manager = PoolManager(num_pools=2) | 54 >>> manager = PoolManager(num_pools=2) |
50 >>> r = manager.request('GET', 'http://google.com/') | 55 >>> r = manager.request('GET', 'http://google.com/') |
51 >>> r = manager.request('GET', 'http://google.com/mail') | 56 >>> r = manager.request('GET', 'http://google.com/mail') |
52 >>> r = manager.request('GET', 'http://yahoo.com/') | 57 >>> r = manager.request('GET', 'http://yahoo.com/') |
53 >>> len(manager.pools) | 58 >>> len(manager.pools) |
54 2 | 59 2 |
55 | 60 |
56 """ | 61 """ |
57 | 62 |
| 63 proxy = None |
| 64 |
58 def __init__(self, num_pools=10, headers=None, **connection_pool_kw): | 65 def __init__(self, num_pools=10, headers=None, **connection_pool_kw): |
59 RequestMethods.__init__(self, headers) | 66 RequestMethods.__init__(self, headers) |
60 self.connection_pool_kw = connection_pool_kw | 67 self.connection_pool_kw = connection_pool_kw |
61 self.pools = RecentlyUsedContainer(num_pools, | 68 self.pools = RecentlyUsedContainer(num_pools, |
62 dispose_func=lambda p: p.close()) | 69 dispose_func=lambda p: p.close()) |
63 | 70 |
64 def _new_pool(self, scheme, host, port): | 71 def _new_pool(self, scheme, host, port): |
65 """ | 72 """ |
66 Create a new :class:`ConnectionPool` based on host, port and scheme. | 73 Create a new :class:`ConnectionPool` based on host, port and scheme. |
67 | 74 |
(...skipping 19 matching lines...) Expand all Loading... |
87 """ | 94 """ |
88 self.pools.clear() | 95 self.pools.clear() |
89 | 96 |
90 def connection_from_host(self, host, port=None, scheme='http'): | 97 def connection_from_host(self, host, port=None, scheme='http'): |
91 """ | 98 """ |
92 Get a :class:`ConnectionPool` based on the host, port, and scheme. | 99 Get a :class:`ConnectionPool` based on the host, port, and scheme. |
93 | 100 |
94 If ``port`` isn't given, it will be derived from the ``scheme`` using | 101 If ``port`` isn't given, it will be derived from the ``scheme`` using |
95 ``urllib3.connectionpool.port_by_scheme``. | 102 ``urllib3.connectionpool.port_by_scheme``. |
96 """ | 103 """ |
| 104 |
97 scheme = scheme or 'http' | 105 scheme = scheme or 'http' |
| 106 |
98 port = port or port_by_scheme.get(scheme, 80) | 107 port = port or port_by_scheme.get(scheme, 80) |
99 | 108 |
100 pool_key = (scheme, host, port) | 109 pool_key = (scheme, host, port) |
101 | 110 |
102 # If the scheme, host, or port doesn't match existing open connections, | 111 with self.pools.lock: |
103 # open a new ConnectionPool. | 112 # If the scheme, host, or port doesn't match existing open |
104 pool = self.pools.get(pool_key) | 113 # connections, open a new ConnectionPool. |
105 if pool: | 114 pool = self.pools.get(pool_key) |
106 return pool | 115 if pool: |
| 116 return pool |
107 | 117 |
108 # Make a fresh ConnectionPool of the desired type | 118 # Make a fresh ConnectionPool of the desired type |
109 pool = self._new_pool(scheme, host, port) | 119 pool = self._new_pool(scheme, host, port) |
110 self.pools[pool_key] = pool | 120 self.pools[pool_key] = pool |
111 return pool | 121 return pool |
112 | 122 |
113 def connection_from_url(self, url): | 123 def connection_from_url(self, url): |
114 """ | 124 """ |
115 Similar to :func:`urllib3.connectionpool.connection_from_url` but | 125 Similar to :func:`urllib3.connectionpool.connection_from_url` but |
116 doesn't pass any additional parameters to the | 126 doesn't pass any additional parameters to the |
117 :class:`urllib3.connectionpool.ConnectionPool` constructor. | 127 :class:`urllib3.connectionpool.ConnectionPool` constructor. |
118 | 128 |
119 Additional parameters are taken from the :class:`.PoolManager` | 129 Additional parameters are taken from the :class:`.PoolManager` |
120 constructor. | 130 constructor. |
(...skipping 11 matching lines...) Expand all Loading... |
132 :class:`urllib3.connectionpool.ConnectionPool` can be chosen for it. | 142 :class:`urllib3.connectionpool.ConnectionPool` can be chosen for it. |
133 """ | 143 """ |
134 u = parse_url(url) | 144 u = parse_url(url) |
135 conn = self.connection_from_host(u.host, port=u.port, scheme=u.scheme) | 145 conn = self.connection_from_host(u.host, port=u.port, scheme=u.scheme) |
136 | 146 |
137 kw['assert_same_host'] = False | 147 kw['assert_same_host'] = False |
138 kw['redirect'] = False | 148 kw['redirect'] = False |
139 if 'headers' not in kw: | 149 if 'headers' not in kw: |
140 kw['headers'] = self.headers | 150 kw['headers'] = self.headers |
141 | 151 |
142 response = conn.urlopen(method, u.request_uri, **kw) | 152 if self.proxy is not None and u.scheme == "http": |
| 153 response = conn.urlopen(method, url, **kw) |
| 154 else: |
| 155 response = conn.urlopen(method, u.request_uri, **kw) |
143 | 156 |
144 redirect_location = redirect and response.get_redirect_location() | 157 redirect_location = redirect and response.get_redirect_location() |
145 if not redirect_location: | 158 if not redirect_location: |
146 return response | 159 return response |
147 | 160 |
| 161 # Support relative URLs for redirecting. |
| 162 redirect_location = urljoin(url, redirect_location) |
| 163 |
| 164 # RFC 2616, Section 10.3.4 |
148 if response.status == 303: | 165 if response.status == 303: |
149 method = 'GET' | 166 method = 'GET' |
150 | 167 |
151 log.info("Redirecting %s -> %s" % (url, redirect_location)) | 168 log.info("Redirecting %s -> %s" % (url, redirect_location)) |
152 kw['retries'] = kw.get('retries', 3) - 1 # Persist retries countdown | 169 kw['retries'] = kw.get('retries', 3) - 1 # Persist retries countdown |
153 kw['redirect'] = redirect | 170 kw['redirect'] = redirect |
154 return self.urlopen(method, redirect_location, **kw) | 171 return self.urlopen(method, redirect_location, **kw) |
155 | 172 |
156 | 173 |
157 class ProxyManager(RequestMethods): | 174 class ProxyManager(PoolManager): |
158 """ | 175 """ |
159 Given a ConnectionPool to a proxy, the ProxyManager's ``urlopen`` method | 176 Behaves just like :class:`PoolManager`, but sends all requests through |
160 will make requests to any url through the defined proxy. The ProxyManager | 177 the defined proxy, using the CONNECT method for HTTPS URLs. |
161 class will automatically set the 'Host' header if it is not provided. | 178 |
| 179 :param poxy_url: |
| 180 The URL of the proxy to be used. |
| 181 |
| 182 :param proxy_headers: |
| 183 A dictionary contaning headers that will be sent to the proxy. In case |
| 184 of HTTP they are being sent with each request, while in the |
| 185 HTTPS/CONNECT case they are sent only once. Could be used for proxy |
| 186 authentication. |
| 187 |
| 188 Example: |
| 189 >>> proxy = urllib3.ProxyManager('http://localhost:3128/') |
| 190 >>> r1 = proxy.request('GET', 'http://google.com/') |
| 191 >>> r2 = proxy.request('GET', 'http://httpbin.org/') |
| 192 >>> len(proxy.pools) |
| 193 1 |
| 194 >>> r3 = proxy.request('GET', 'https://httpbin.org/') |
| 195 >>> r4 = proxy.request('GET', 'https://twitter.com/') |
| 196 >>> len(proxy.pools) |
| 197 3 |
| 198 |
162 """ | 199 """ |
163 | 200 |
164 def __init__(self, proxy_pool): | 201 def __init__(self, proxy_url, num_pools=10, headers=None, |
165 self.proxy_pool = proxy_pool | 202 proxy_headers=None, **connection_pool_kw): |
| 203 |
| 204 if isinstance(proxy_url, HTTPConnectionPool): |
| 205 proxy_url = '%s://%s:%i' % (proxy_url.scheme, proxy_url.host, |
| 206 proxy_url.port) |
| 207 proxy = parse_url(proxy_url) |
| 208 if not proxy.port: |
| 209 port = port_by_scheme.get(proxy.scheme, 80) |
| 210 proxy = proxy._replace(port=port) |
| 211 self.proxy = proxy |
| 212 self.proxy_headers = proxy_headers or {} |
| 213 assert self.proxy.scheme in ("http", "https"), \ |
| 214 'Not supported proxy scheme %s' % self.proxy.scheme |
| 215 connection_pool_kw['_proxy'] = self.proxy |
| 216 connection_pool_kw['_proxy_headers'] = self.proxy_headers |
| 217 super(ProxyManager, self).__init__( |
| 218 num_pools, headers, **connection_pool_kw) |
| 219 |
| 220 def connection_from_host(self, host, port=None, scheme='http'): |
| 221 if scheme == "https": |
| 222 return super(ProxyManager, self).connection_from_host( |
| 223 host, port, scheme) |
| 224 |
| 225 return super(ProxyManager, self).connection_from_host( |
| 226 self.proxy.host, self.proxy.port, self.proxy.scheme) |
166 | 227 |
167 def _set_proxy_headers(self, url, headers=None): | 228 def _set_proxy_headers(self, url, headers=None): |
168 """ | 229 """ |
169 Sets headers needed by proxies: specifically, the Accept and Host | 230 Sets headers needed by proxies: specifically, the Accept and Host |
170 headers. Only sets headers not provided by the user. | 231 headers. Only sets headers not provided by the user. |
171 """ | 232 """ |
172 headers_ = {'Accept': '*/*'} | 233 headers_ = {'Accept': '*/*'} |
173 | 234 |
174 host = parse_url(url).host | 235 netloc = parse_url(url).netloc |
175 if host: | 236 if netloc: |
176 headers_['Host'] = host | 237 headers_['Host'] = netloc |
177 | 238 |
178 if headers: | 239 if headers: |
179 headers_.update(headers) | 240 headers_.update(headers) |
180 | |
181 return headers_ | 241 return headers_ |
182 | 242 |
183 def urlopen(self, method, url, **kw): | 243 def urlopen(self, method, url, redirect=True, **kw): |
184 "Same as HTTP(S)ConnectionPool.urlopen, ``url`` must be absolute." | 244 "Same as HTTP(S)ConnectionPool.urlopen, ``url`` must be absolute." |
185 kw['assert_same_host'] = False | 245 u = parse_url(url) |
186 kw['headers'] = self._set_proxy_headers(url, headers=kw.get('headers')) | 246 |
187 return self.proxy_pool.urlopen(method, url, **kw) | 247 if u.scheme == "http": |
| 248 # It's too late to set proxy headers on per-request basis for |
| 249 # tunnelled HTTPS connections, should use |
| 250 # constructor's proxy_headers instead. |
| 251 kw['headers'] = self._set_proxy_headers(url, kw.get('headers', |
| 252 self.headers)) |
| 253 kw['headers'].update(self.proxy_headers) |
| 254 |
| 255 return super(ProxyManager, self).urlopen(method, url, redirect, **kw) |
188 | 256 |
189 | 257 |
190 def proxy_from_url(url, **pool_kw): | 258 def proxy_from_url(url, **kw): |
191 proxy_pool = connection_from_url(url, **pool_kw) | 259 return ProxyManager(proxy_url=url, **kw) |
192 return ProxyManager(proxy_pool) | |
OLD | NEW |