Show More
@@ -1,761 +1,761 b'' | |||||
1 | # This library is free software; you can redistribute it and/or |
|
1 | # This library is free software; you can redistribute it and/or | |
2 | # modify it under the terms of the GNU Lesser General Public |
|
2 | # modify it under the terms of the GNU Lesser General Public | |
3 | # License as published by the Free Software Foundation; either |
|
3 | # License as published by the Free Software Foundation; either | |
4 | # version 2.1 of the License, or (at your option) any later version. |
|
4 | # version 2.1 of the License, or (at your option) any later version. | |
5 | # |
|
5 | # | |
6 | # This library is distributed in the hope that it will be useful, |
|
6 | # This library is distributed in the hope that it will be useful, | |
7 | # but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
7 | # but WITHOUT ANY WARRANTY; without even the implied warranty of | |
8 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
8 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
9 | # Lesser General Public License for more details. |
|
9 | # Lesser General Public License for more details. | |
10 | # |
|
10 | # | |
11 | # You should have received a copy of the GNU Lesser General Public |
|
11 | # You should have received a copy of the GNU Lesser General Public | |
12 | # License along with this library; if not, see |
|
12 | # License along with this library; if not, see | |
13 | # <http://www.gnu.org/licenses/>. |
|
13 | # <http://www.gnu.org/licenses/>. | |
14 |
|
14 | |||
15 | # This file is part of urlgrabber, a high-level cross-protocol url-grabber |
|
15 | # This file is part of urlgrabber, a high-level cross-protocol url-grabber | |
16 | # Copyright 2002-2004 Michael D. Stenner, Ryan Tomayko |
|
16 | # Copyright 2002-2004 Michael D. Stenner, Ryan Tomayko | |
17 |
|
17 | |||
18 | # Modified by Benoit Boissinot: |
|
18 | # Modified by Benoit Boissinot: | |
19 | # - fix for digest auth (inspired from urllib2.py @ Python v2.4) |
|
19 | # - fix for digest auth (inspired from urllib2.py @ Python v2.4) | |
20 | # Modified by Dirkjan Ochtman: |
|
20 | # Modified by Dirkjan Ochtman: | |
21 | # - import md5 function from a local util module |
|
21 | # - import md5 function from a local util module | |
22 | # Modified by Martin Geisler: |
|
22 | # Modified by Martin Geisler: | |
23 | # - moved md5 function from local util module to this module |
|
23 | # - moved md5 function from local util module to this module | |
24 | # Modified by Augie Fackler: |
|
24 | # Modified by Augie Fackler: | |
25 | # - add safesend method and use it to prevent broken pipe errors |
|
25 | # - add safesend method and use it to prevent broken pipe errors | |
26 | # on large POST requests |
|
26 | # on large POST requests | |
27 |
|
27 | |||
28 | """An HTTP handler for urllib2 that supports HTTP 1.1 and keepalive. |
|
28 | """An HTTP handler for urllib2 that supports HTTP 1.1 and keepalive. | |
29 |
|
29 | |||
30 | >>> import urllib2 |
|
30 | >>> import urllib2 | |
31 | >>> from keepalive import HTTPHandler |
|
31 | >>> from keepalive import HTTPHandler | |
32 | >>> keepalive_handler = HTTPHandler() |
|
32 | >>> keepalive_handler = HTTPHandler() | |
33 | >>> opener = urllib2.build_opener(keepalive_handler) |
|
33 | >>> opener = urllib2.build_opener(keepalive_handler) | |
34 | >>> urllib2.install_opener(opener) |
|
34 | >>> urllib2.install_opener(opener) | |
35 | >>> |
|
35 | >>> | |
36 | >>> fo = urllib2.urlopen('http://www.python.org') |
|
36 | >>> fo = urllib2.urlopen('http://www.python.org') | |
37 |
|
37 | |||
38 | If a connection to a given host is requested, and all of the existing |
|
38 | If a connection to a given host is requested, and all of the existing | |
39 | connections are still in use, another connection will be opened. If |
|
39 | connections are still in use, another connection will be opened. If | |
40 | the handler tries to use an existing connection but it fails in some |
|
40 | the handler tries to use an existing connection but it fails in some | |
41 | way, it will be closed and removed from the pool. |
|
41 | way, it will be closed and removed from the pool. | |
42 |
|
42 | |||
43 | To remove the handler, simply re-run build_opener with no arguments, and |
|
43 | To remove the handler, simply re-run build_opener with no arguments, and | |
44 | install that opener. |
|
44 | install that opener. | |
45 |
|
45 | |||
46 | You can explicitly close connections by using the close_connection() |
|
46 | You can explicitly close connections by using the close_connection() | |
47 | method of the returned file-like object (described below) or you can |
|
47 | method of the returned file-like object (described below) or you can | |
48 | use the handler methods: |
|
48 | use the handler methods: | |
49 |
|
49 | |||
50 | close_connection(host) |
|
50 | close_connection(host) | |
51 | close_all() |
|
51 | close_all() | |
52 | open_connections() |
|
52 | open_connections() | |
53 |
|
53 | |||
54 | NOTE: using the close_connection and close_all methods of the handler |
|
54 | NOTE: using the close_connection and close_all methods of the handler | |
55 | should be done with care when using multiple threads. |
|
55 | should be done with care when using multiple threads. | |
56 | * there is nothing that prevents another thread from creating new |
|
56 | * there is nothing that prevents another thread from creating new | |
57 | connections immediately after connections are closed |
|
57 | connections immediately after connections are closed | |
58 | * no checks are done to prevent in-use connections from being closed |
|
58 | * no checks are done to prevent in-use connections from being closed | |
59 |
|
59 | |||
60 | >>> keepalive_handler.close_all() |
|
60 | >>> keepalive_handler.close_all() | |
61 |
|
61 | |||
62 | EXTRA ATTRIBUTES AND METHODS |
|
62 | EXTRA ATTRIBUTES AND METHODS | |
63 |
|
63 | |||
64 | Upon a status of 200, the object returned has a few additional |
|
64 | Upon a status of 200, the object returned has a few additional | |
65 | attributes and methods, which should not be used if you want to |
|
65 | attributes and methods, which should not be used if you want to | |
66 | remain consistent with the normal urllib2-returned objects: |
|
66 | remain consistent with the normal urllib2-returned objects: | |
67 |
|
67 | |||
68 | close_connection() - close the connection to the host |
|
68 | close_connection() - close the connection to the host | |
69 | readlines() - you know, readlines() |
|
69 | readlines() - you know, readlines() | |
70 | status - the return status (i.e. 404) |
|
70 | status - the return status (i.e. 404) | |
71 | reason - english translation of status (i.e. 'File not found') |
|
71 | reason - english translation of status (i.e. 'File not found') | |
72 |
|
72 | |||
73 | If you want the best of both worlds, use this inside an |
|
73 | If you want the best of both worlds, use this inside an | |
74 | AttributeError-catching try: |
|
74 | AttributeError-catching try: | |
75 |
|
75 | |||
76 | >>> try: status = fo.status |
|
76 | >>> try: status = fo.status | |
77 | >>> except AttributeError: status = None |
|
77 | >>> except AttributeError: status = None | |
78 |
|
78 | |||
79 | Unfortunately, these are ONLY there if status == 200, so it's not |
|
79 | Unfortunately, these are ONLY there if status == 200, so it's not | |
80 | easy to distinguish between non-200 responses. The reason is that |
|
80 | easy to distinguish between non-200 responses. The reason is that | |
81 | urllib2 tries to do clever things with error codes 301, 302, 401, |
|
81 | urllib2 tries to do clever things with error codes 301, 302, 401, | |
82 | and 407, and it wraps the object upon return. |
|
82 | and 407, and it wraps the object upon return. | |
83 |
|
83 | |||
84 | For python versions earlier than 2.4, you can avoid this fancy error |
|
84 | For python versions earlier than 2.4, you can avoid this fancy error | |
85 | handling by setting the module-level global HANDLE_ERRORS to zero. |
|
85 | handling by setting the module-level global HANDLE_ERRORS to zero. | |
86 | You see, prior to 2.4, it's the HTTP Handler's job to determine what |
|
86 | You see, prior to 2.4, it's the HTTP Handler's job to determine what | |
87 | to handle specially, and what to just pass up. HANDLE_ERRORS == 0 |
|
87 | to handle specially, and what to just pass up. HANDLE_ERRORS == 0 | |
88 | means "pass everything up". In python 2.4, however, this job no |
|
88 | means "pass everything up". In python 2.4, however, this job no | |
89 | longer belongs to the HTTP Handler and is now done by a NEW handler, |
|
89 | longer belongs to the HTTP Handler and is now done by a NEW handler, | |
90 | HTTPErrorProcessor. Here's the bottom line: |
|
90 | HTTPErrorProcessor. Here's the bottom line: | |
91 |
|
91 | |||
92 | python version < 2.4 |
|
92 | python version < 2.4 | |
93 | HANDLE_ERRORS == 1 (default) pass up 200, treat the rest as |
|
93 | HANDLE_ERRORS == 1 (default) pass up 200, treat the rest as | |
94 | errors |
|
94 | errors | |
95 | HANDLE_ERRORS == 0 pass everything up, error processing is |
|
95 | HANDLE_ERRORS == 0 pass everything up, error processing is | |
96 | left to the calling code |
|
96 | left to the calling code | |
97 | python version >= 2.4 |
|
97 | python version >= 2.4 | |
98 | HANDLE_ERRORS == 1 pass up 200, treat the rest as errors |
|
98 | HANDLE_ERRORS == 1 pass up 200, treat the rest as errors | |
99 | HANDLE_ERRORS == 0 (default) pass everything up, let the |
|
99 | HANDLE_ERRORS == 0 (default) pass everything up, let the | |
100 | other handlers (specifically, |
|
100 | other handlers (specifically, | |
101 | HTTPErrorProcessor) decide what to do |
|
101 | HTTPErrorProcessor) decide what to do | |
102 |
|
102 | |||
103 | In practice, setting the variable either way makes little difference |
|
103 | In practice, setting the variable either way makes little difference | |
104 | in python 2.4, so for the most consistent behavior across versions, |
|
104 | in python 2.4, so for the most consistent behavior across versions, | |
105 | you probably just want to use the defaults, which will give you |
|
105 | you probably just want to use the defaults, which will give you | |
106 | exceptions on errors. |
|
106 | exceptions on errors. | |
107 |
|
107 | |||
108 | """ |
|
108 | """ | |
109 |
|
109 | |||
110 | # $Id: keepalive.py,v 1.14 2006/04/04 21:00:32 mstenner Exp $ |
|
110 | # $Id: keepalive.py,v 1.14 2006/04/04 21:00:32 mstenner Exp $ | |
111 |
|
111 | |||
112 | import errno |
|
112 | import errno | |
113 | import httplib |
|
113 | import httplib | |
114 | import socket |
|
114 | import socket | |
115 | import thread |
|
115 | import thread | |
116 | import urllib2 |
|
116 | import urllib2 | |
117 |
|
117 | |||
118 | DEBUG = None |
|
118 | DEBUG = None | |
119 |
|
119 | |||
120 | import sys |
|
120 | import sys | |
121 | if sys.version_info < (2, 4): |
|
121 | if sys.version_info < (2, 4): | |
122 | HANDLE_ERRORS = 1 |
|
122 | HANDLE_ERRORS = 1 | |
123 | else: HANDLE_ERRORS = 0 |
|
123 | else: HANDLE_ERRORS = 0 | |
124 |
|
124 | |||
125 | class ConnectionManager(object): |
|
125 | class ConnectionManager(object): | |
126 | """ |
|
126 | """ | |
127 | The connection manager must be able to: |
|
127 | The connection manager must be able to: | |
128 | * keep track of all existing |
|
128 | * keep track of all existing | |
129 | """ |
|
129 | """ | |
130 | def __init__(self): |
|
130 | def __init__(self): | |
131 | self._lock = thread.allocate_lock() |
|
131 | self._lock = thread.allocate_lock() | |
132 | self._hostmap = {} # map hosts to a list of connections |
|
132 | self._hostmap = {} # map hosts to a list of connections | |
133 | self._connmap = {} # map connections to host |
|
133 | self._connmap = {} # map connections to host | |
134 | self._readymap = {} # map connection to ready state |
|
134 | self._readymap = {} # map connection to ready state | |
135 |
|
135 | |||
136 | def add(self, host, connection, ready): |
|
136 | def add(self, host, connection, ready): | |
137 | self._lock.acquire() |
|
137 | self._lock.acquire() | |
138 | try: |
|
138 | try: | |
139 | if host not in self._hostmap: |
|
139 | if host not in self._hostmap: | |
140 | self._hostmap[host] = [] |
|
140 | self._hostmap[host] = [] | |
141 | self._hostmap[host].append(connection) |
|
141 | self._hostmap[host].append(connection) | |
142 | self._connmap[connection] = host |
|
142 | self._connmap[connection] = host | |
143 | self._readymap[connection] = ready |
|
143 | self._readymap[connection] = ready | |
144 | finally: |
|
144 | finally: | |
145 | self._lock.release() |
|
145 | self._lock.release() | |
146 |
|
146 | |||
147 | def remove(self, connection): |
|
147 | def remove(self, connection): | |
148 | self._lock.acquire() |
|
148 | self._lock.acquire() | |
149 | try: |
|
149 | try: | |
150 | try: |
|
150 | try: | |
151 | host = self._connmap[connection] |
|
151 | host = self._connmap[connection] | |
152 | except KeyError: |
|
152 | except KeyError: | |
153 | pass |
|
153 | pass | |
154 | else: |
|
154 | else: | |
155 | del self._connmap[connection] |
|
155 | del self._connmap[connection] | |
156 | del self._readymap[connection] |
|
156 | del self._readymap[connection] | |
157 | self._hostmap[host].remove(connection) |
|
157 | self._hostmap[host].remove(connection) | |
158 | if not self._hostmap[host]: del self._hostmap[host] |
|
158 | if not self._hostmap[host]: del self._hostmap[host] | |
159 | finally: |
|
159 | finally: | |
160 | self._lock.release() |
|
160 | self._lock.release() | |
161 |
|
161 | |||
162 | def set_ready(self, connection, ready): |
|
162 | def set_ready(self, connection, ready): | |
163 | try: |
|
163 | try: | |
164 | self._readymap[connection] = ready |
|
164 | self._readymap[connection] = ready | |
165 | except KeyError: |
|
165 | except KeyError: | |
166 | pass |
|
166 | pass | |
167 |
|
167 | |||
168 | def get_ready_conn(self, host): |
|
168 | def get_ready_conn(self, host): | |
169 | conn = None |
|
169 | conn = None | |
170 | self._lock.acquire() |
|
170 | self._lock.acquire() | |
171 | try: |
|
171 | try: | |
172 | if host in self._hostmap: |
|
172 | if host in self._hostmap: | |
173 | for c in self._hostmap[host]: |
|
173 | for c in self._hostmap[host]: | |
174 | if self._readymap[c]: |
|
174 | if self._readymap[c]: | |
175 | self._readymap[c] = 0 |
|
175 | self._readymap[c] = 0 | |
176 | conn = c |
|
176 | conn = c | |
177 | break |
|
177 | break | |
178 | finally: |
|
178 | finally: | |
179 | self._lock.release() |
|
179 | self._lock.release() | |
180 | return conn |
|
180 | return conn | |
181 |
|
181 | |||
182 | def get_all(self, host=None): |
|
182 | def get_all(self, host=None): | |
183 | if host: |
|
183 | if host: | |
184 | return list(self._hostmap.get(host, [])) |
|
184 | return list(self._hostmap.get(host, [])) | |
185 | else: |
|
185 | else: | |
186 | return dict(self._hostmap) |
|
186 | return dict(self._hostmap) | |
187 |
|
187 | |||
188 | class KeepAliveHandler(object): |
|
188 | class KeepAliveHandler(object): | |
189 | def __init__(self): |
|
189 | def __init__(self): | |
190 | self._cm = ConnectionManager() |
|
190 | self._cm = ConnectionManager() | |
191 |
|
191 | |||
192 | #### Connection Management |
|
192 | #### Connection Management | |
193 | def open_connections(self): |
|
193 | def open_connections(self): | |
194 | """return a list of connected hosts and the number of connections |
|
194 | """return a list of connected hosts and the number of connections | |
195 | to each. [('foo.com:80', 2), ('bar.org', 1)]""" |
|
195 | to each. [('foo.com:80', 2), ('bar.org', 1)]""" | |
196 | return [(host, len(li)) for (host, li) in self._cm.get_all().items()] |
|
196 | return [(host, len(li)) for (host, li) in self._cm.get_all().items()] | |
197 |
|
197 | |||
198 | def close_connection(self, host): |
|
198 | def close_connection(self, host): | |
199 | """close connection(s) to <host> |
|
199 | """close connection(s) to <host> | |
200 | host is the host:port spec, as in 'www.cnn.com:8080' as passed in. |
|
200 | host is the host:port spec, as in 'www.cnn.com:8080' as passed in. | |
201 | no error occurs if there is no connection to that host.""" |
|
201 | no error occurs if there is no connection to that host.""" | |
202 | for h in self._cm.get_all(host): |
|
202 | for h in self._cm.get_all(host): | |
203 | self._cm.remove(h) |
|
203 | self._cm.remove(h) | |
204 | h.close() |
|
204 | h.close() | |
205 |
|
205 | |||
206 | def close_all(self): |
|
206 | def close_all(self): | |
207 | """close all open connections""" |
|
207 | """close all open connections""" | |
208 | for host, conns in self._cm.get_all().iteritems(): |
|
208 | for host, conns in self._cm.get_all().iteritems(): | |
209 | for h in conns: |
|
209 | for h in conns: | |
210 | self._cm.remove(h) |
|
210 | self._cm.remove(h) | |
211 | h.close() |
|
211 | h.close() | |
212 |
|
212 | |||
213 | def _request_closed(self, request, host, connection): |
|
213 | def _request_closed(self, request, host, connection): | |
214 | """tells us that this request is now closed and that the |
|
214 | """tells us that this request is now closed and that the | |
215 | connection is ready for another request""" |
|
215 | connection is ready for another request""" | |
216 | self._cm.set_ready(connection, 1) |
|
216 | self._cm.set_ready(connection, 1) | |
217 |
|
217 | |||
218 | def _remove_connection(self, host, connection, close=0): |
|
218 | def _remove_connection(self, host, connection, close=0): | |
219 | if close: |
|
219 | if close: | |
220 | connection.close() |
|
220 | connection.close() | |
221 | self._cm.remove(connection) |
|
221 | self._cm.remove(connection) | |
222 |
|
222 | |||
223 | #### Transaction Execution |
|
223 | #### Transaction Execution | |
224 | def http_open(self, req): |
|
224 | def http_open(self, req): | |
225 | return self.do_open(HTTPConnection, req) |
|
225 | return self.do_open(HTTPConnection, req) | |
226 |
|
226 | |||
227 | def do_open(self, http_class, req): |
|
227 | def do_open(self, http_class, req): | |
228 | host = req.get_host() |
|
228 | host = req.get_host() | |
229 | if not host: |
|
229 | if not host: | |
230 | raise urllib2.URLError('no host given') |
|
230 | raise urllib2.URLError('no host given') | |
231 |
|
231 | |||
232 | try: |
|
232 | try: | |
233 | h = self._cm.get_ready_conn(host) |
|
233 | h = self._cm.get_ready_conn(host) | |
234 | while h: |
|
234 | while h: | |
235 | r = self._reuse_connection(h, req, host) |
|
235 | r = self._reuse_connection(h, req, host) | |
236 |
|
236 | |||
237 | # if this response is non-None, then it worked and we're |
|
237 | # if this response is non-None, then it worked and we're | |
238 | # done. Break out, skipping the else block. |
|
238 | # done. Break out, skipping the else block. | |
239 | if r: |
|
239 | if r: | |
240 | break |
|
240 | break | |
241 |
|
241 | |||
242 | # connection is bad - possibly closed by server |
|
242 | # connection is bad - possibly closed by server | |
243 | # discard it and ask for the next free connection |
|
243 | # discard it and ask for the next free connection | |
244 | h.close() |
|
244 | h.close() | |
245 | self._cm.remove(h) |
|
245 | self._cm.remove(h) | |
246 | h = self._cm.get_ready_conn(host) |
|
246 | h = self._cm.get_ready_conn(host) | |
247 | else: |
|
247 | else: | |
248 | # no (working) free connections were found. Create a new one. |
|
248 | # no (working) free connections were found. Create a new one. | |
249 | h = http_class(host) |
|
249 | h = http_class(host) | |
250 | if DEBUG: |
|
250 | if DEBUG: | |
251 | DEBUG.info("creating new connection to %s (%d)", |
|
251 | DEBUG.info("creating new connection to %s (%d)", | |
252 | host, id(h)) |
|
252 | host, id(h)) | |
253 | self._cm.add(host, h, 0) |
|
253 | self._cm.add(host, h, 0) | |
254 | self._start_transaction(h, req) |
|
254 | self._start_transaction(h, req) | |
255 | r = h.getresponse() |
|
255 | r = h.getresponse() | |
256 | except (socket.error, httplib.HTTPException), err: |
|
256 | except (socket.error, httplib.HTTPException), err: | |
257 | raise urllib2.URLError(err) |
|
257 | raise urllib2.URLError(err) | |
258 |
|
258 | |||
259 | # if not a persistent connection, don't try to reuse it |
|
259 | # if not a persistent connection, don't try to reuse it | |
260 | if r.will_close: |
|
260 | if r.will_close: | |
261 | self._cm.remove(h) |
|
261 | self._cm.remove(h) | |
262 |
|
262 | |||
263 | if DEBUG: |
|
263 | if DEBUG: | |
264 | DEBUG.info("STATUS: %s, %s", r.status, r.reason) |
|
264 | DEBUG.info("STATUS: %s, %s", r.status, r.reason) | |
265 | r._handler = self |
|
265 | r._handler = self | |
266 | r._host = host |
|
266 | r._host = host | |
267 | r._url = req.get_full_url() |
|
267 | r._url = req.get_full_url() | |
268 | r._connection = h |
|
268 | r._connection = h | |
269 | r.code = r.status |
|
269 | r.code = r.status | |
270 | r.headers = r.msg |
|
270 | r.headers = r.msg | |
271 | r.msg = r.reason |
|
271 | r.msg = r.reason | |
272 |
|
272 | |||
273 | if r.status == 200 or not HANDLE_ERRORS: |
|
273 | if r.status == 200 or not HANDLE_ERRORS: | |
274 | return r |
|
274 | return r | |
275 | else: |
|
275 | else: | |
276 | return self.parent.error('http', req, r, |
|
276 | return self.parent.error('http', req, r, | |
277 | r.status, r.msg, r.headers) |
|
277 | r.status, r.msg, r.headers) | |
278 |
|
278 | |||
279 | def _reuse_connection(self, h, req, host): |
|
279 | def _reuse_connection(self, h, req, host): | |
280 | """start the transaction with a re-used connection |
|
280 | """start the transaction with a re-used connection | |
281 | return a response object (r) upon success or None on failure. |
|
281 | return a response object (r) upon success or None on failure. | |
282 | This DOES not close or remove bad connections in cases where |
|
282 | This DOES not close or remove bad connections in cases where | |
283 | it returns. However, if an unexpected exception occurs, it |
|
283 | it returns. However, if an unexpected exception occurs, it | |
284 | will close and remove the connection before re-raising. |
|
284 | will close and remove the connection before re-raising. | |
285 | """ |
|
285 | """ | |
286 | try: |
|
286 | try: | |
287 | self._start_transaction(h, req) |
|
287 | self._start_transaction(h, req) | |
288 | r = h.getresponse() |
|
288 | r = h.getresponse() | |
289 | # note: just because we got something back doesn't mean it |
|
289 | # note: just because we got something back doesn't mean it | |
290 | # worked. We'll check the version below, too. |
|
290 | # worked. We'll check the version below, too. | |
291 | except (socket.error, httplib.HTTPException): |
|
291 | except (socket.error, httplib.HTTPException): | |
292 | r = None |
|
292 | r = None | |
293 | except: # re-raises |
|
293 | except: # re-raises | |
294 | # adding this block just in case we've missed |
|
294 | # adding this block just in case we've missed | |
295 | # something we will still raise the exception, but |
|
295 | # something we will still raise the exception, but | |
296 | # lets try and close the connection and remove it |
|
296 | # lets try and close the connection and remove it | |
297 | # first. We previously got into a nasty loop |
|
297 | # first. We previously got into a nasty loop | |
298 | # where an exception was uncaught, and so the |
|
298 | # where an exception was uncaught, and so the | |
299 | # connection stayed open. On the next try, the |
|
299 | # connection stayed open. On the next try, the | |
300 | # same exception was raised, etc. The trade-off is |
|
300 | # same exception was raised, etc. The trade-off is | |
301 | # that it's now possible this call will raise |
|
301 | # that it's now possible this call will raise | |
302 | # a DIFFERENT exception |
|
302 | # a DIFFERENT exception | |
303 | if DEBUG: |
|
303 | if DEBUG: | |
304 | DEBUG.error("unexpected exception - closing " |
|
304 | DEBUG.error("unexpected exception - closing " | |
305 | "connection to %s (%d)", host, id(h)) |
|
305 | "connection to %s (%d)", host, id(h)) | |
306 | self._cm.remove(h) |
|
306 | self._cm.remove(h) | |
307 | h.close() |
|
307 | h.close() | |
308 | raise |
|
308 | raise | |
309 |
|
309 | |||
310 | if r is None or r.version == 9: |
|
310 | if r is None or r.version == 9: | |
311 | # httplib falls back to assuming HTTP 0.9 if it gets a |
|
311 | # httplib falls back to assuming HTTP 0.9 if it gets a | |
312 | # bad header back. This is most likely to happen if |
|
312 | # bad header back. This is most likely to happen if | |
313 | # the socket has been closed by the server since we |
|
313 | # the socket has been closed by the server since we | |
314 | # last used the connection. |
|
314 | # last used the connection. | |
315 | if DEBUG: |
|
315 | if DEBUG: | |
316 | DEBUG.info("failed to re-use connection to %s (%d)", |
|
316 | DEBUG.info("failed to re-use connection to %s (%d)", | |
317 | host, id(h)) |
|
317 | host, id(h)) | |
318 | r = None |
|
318 | r = None | |
319 | else: |
|
319 | else: | |
320 | if DEBUG: |
|
320 | if DEBUG: | |
321 | DEBUG.info("re-using connection to %s (%d)", host, id(h)) |
|
321 | DEBUG.info("re-using connection to %s (%d)", host, id(h)) | |
322 |
|
322 | |||
323 | return r |
|
323 | return r | |
324 |
|
324 | |||
325 | def _start_transaction(self, h, req): |
|
325 | def _start_transaction(self, h, req): | |
326 | # What follows mostly reimplements HTTPConnection.request() |
|
326 | # What follows mostly reimplements HTTPConnection.request() | |
327 | # except it adds self.parent.addheaders in the mix. |
|
327 | # except it adds self.parent.addheaders in the mix. | |
328 | headers = req.headers.copy() |
|
328 | headers = req.headers.copy() | |
329 | if sys.version_info >= (2, 4): |
|
329 | if sys.version_info >= (2, 4): | |
330 | headers.update(req.unredirected_hdrs) |
|
330 | headers.update(req.unredirected_hdrs) | |
331 | headers.update(self.parent.addheaders) |
|
331 | headers.update(self.parent.addheaders) | |
332 | headers = dict((n.lower(), v) for n, v in headers.items()) |
|
332 | headers = dict((n.lower(), v) for n, v in headers.items()) | |
333 | skipheaders = {} |
|
333 | skipheaders = {} | |
334 | for n in ('host', 'accept-encoding'): |
|
334 | for n in ('host', 'accept-encoding'): | |
335 | if n in headers: |
|
335 | if n in headers: | |
336 | skipheaders['skip_' + n.replace('-', '_')] = 1 |
|
336 | skipheaders['skip_' + n.replace('-', '_')] = 1 | |
337 | try: |
|
337 | try: | |
338 | if req.has_data(): |
|
338 | if req.has_data(): | |
339 | data = req.get_data() |
|
339 | data = req.get_data() | |
340 | h.putrequest('POST', req.get_selector(), **skipheaders) |
|
340 | h.putrequest('POST', req.get_selector(), **skipheaders) | |
341 | if 'content-type' not in headers: |
|
341 | if 'content-type' not in headers: | |
342 | h.putheader('Content-type', |
|
342 | h.putheader('Content-type', | |
343 | 'application/x-www-form-urlencoded') |
|
343 | 'application/x-www-form-urlencoded') | |
344 | if 'content-length' not in headers: |
|
344 | if 'content-length' not in headers: | |
345 | h.putheader('Content-length', '%d' % len(data)) |
|
345 | h.putheader('Content-length', '%d' % len(data)) | |
346 | else: |
|
346 | else: | |
347 | h.putrequest('GET', req.get_selector(), **skipheaders) |
|
347 | h.putrequest('GET', req.get_selector(), **skipheaders) | |
348 | except (socket.error), err: |
|
348 | except (socket.error), err: | |
349 | raise urllib2.URLError(err) |
|
349 | raise urllib2.URLError(err) | |
350 | for k, v in headers.items(): |
|
350 | for k, v in headers.items(): | |
351 | h.putheader(k, v) |
|
351 | h.putheader(k, v) | |
352 | h.endheaders() |
|
352 | h.endheaders() | |
353 | if req.has_data(): |
|
353 | if req.has_data(): | |
354 | h.send(data) |
|
354 | h.send(data) | |
355 |
|
355 | |||
356 | class HTTPHandler(KeepAliveHandler, urllib2.HTTPHandler): |
|
356 | class HTTPHandler(KeepAliveHandler, urllib2.HTTPHandler): | |
357 | pass |
|
357 | pass | |
358 |
|
358 | |||
359 | class HTTPResponse(httplib.HTTPResponse): |
|
359 | class HTTPResponse(httplib.HTTPResponse): | |
360 | # we need to subclass HTTPResponse in order to |
|
360 | # we need to subclass HTTPResponse in order to | |
361 | # 1) add readline() and readlines() methods |
|
361 | # 1) add readline() and readlines() methods | |
362 | # 2) add close_connection() methods |
|
362 | # 2) add close_connection() methods | |
363 | # 3) add info() and geturl() methods |
|
363 | # 3) add info() and geturl() methods | |
364 |
|
364 | |||
365 | # in order to add readline(), read must be modified to deal with a |
|
365 | # in order to add readline(), read must be modified to deal with a | |
366 | # buffer. example: readline must read a buffer and then spit back |
|
366 | # buffer. example: readline must read a buffer and then spit back | |
367 | # one line at a time. The only real alternative is to read one |
|
367 | # one line at a time. The only real alternative is to read one | |
368 | # BYTE at a time (ick). Once something has been read, it can't be |
|
368 | # BYTE at a time (ick). Once something has been read, it can't be | |
369 | # put back (ok, maybe it can, but that's even uglier than this), |
|
369 | # put back (ok, maybe it can, but that's even uglier than this), | |
370 | # so if you THEN do a normal read, you must first take stuff from |
|
370 | # so if you THEN do a normal read, you must first take stuff from | |
371 | # the buffer. |
|
371 | # the buffer. | |
372 |
|
372 | |||
373 | # the read method wraps the original to accommodate buffering, |
|
373 | # the read method wraps the original to accommodate buffering, | |
374 | # although read() never adds to the buffer. |
|
374 | # although read() never adds to the buffer. | |
375 | # Both readline and readlines have been stolen with almost no |
|
375 | # Both readline and readlines have been stolen with almost no | |
376 | # modification from socket.py |
|
376 | # modification from socket.py | |
377 |
|
377 | |||
378 |
|
378 | |||
379 | def __init__(self, sock, debuglevel=0, strict=0, method=None): |
|
379 | def __init__(self, sock, debuglevel=0, strict=0, method=None): | |
380 | httplib.HTTPResponse.__init__(self, sock, debuglevel, method) |
|
380 | httplib.HTTPResponse.__init__(self, sock, debuglevel, method) | |
381 | self.fileno = sock.fileno |
|
381 | self.fileno = sock.fileno | |
382 | self.code = None |
|
382 | self.code = None | |
383 | self._rbuf = '' |
|
383 | self._rbuf = '' | |
384 | self._rbufsize = 8096 |
|
384 | self._rbufsize = 8096 | |
385 | self._handler = None # inserted by the handler later |
|
385 | self._handler = None # inserted by the handler later | |
386 | self._host = None # (same) |
|
386 | self._host = None # (same) | |
387 | self._url = None # (same) |
|
387 | self._url = None # (same) | |
388 | self._connection = None # (same) |
|
388 | self._connection = None # (same) | |
389 |
|
389 | |||
390 | _raw_read = httplib.HTTPResponse.read |
|
390 | _raw_read = httplib.HTTPResponse.read | |
391 |
|
391 | |||
392 | def close(self): |
|
392 | def close(self): | |
393 | if self.fp: |
|
393 | if self.fp: | |
394 | self.fp.close() |
|
394 | self.fp.close() | |
395 | self.fp = None |
|
395 | self.fp = None | |
396 | if self._handler: |
|
396 | if self._handler: | |
397 | self._handler._request_closed(self, self._host, |
|
397 | self._handler._request_closed(self, self._host, | |
398 | self._connection) |
|
398 | self._connection) | |
399 |
|
399 | |||
400 | def close_connection(self): |
|
400 | def close_connection(self): | |
401 | self._handler._remove_connection(self._host, self._connection, close=1) |
|
401 | self._handler._remove_connection(self._host, self._connection, close=1) | |
402 | self.close() |
|
402 | self.close() | |
403 |
|
403 | |||
404 | def info(self): |
|
404 | def info(self): | |
405 | return self.headers |
|
405 | return self.headers | |
406 |
|
406 | |||
407 | def geturl(self): |
|
407 | def geturl(self): | |
408 | return self._url |
|
408 | return self._url | |
409 |
|
409 | |||
410 | def read(self, amt=None): |
|
410 | def read(self, amt=None): | |
411 | # the _rbuf test is only in this first if for speed. It's not |
|
411 | # the _rbuf test is only in this first if for speed. It's not | |
412 | # logically necessary |
|
412 | # logically necessary | |
413 | if self._rbuf and not amt is None: |
|
413 | if self._rbuf and not amt is None: | |
414 | L = len(self._rbuf) |
|
414 | L = len(self._rbuf) | |
415 | if amt > L: |
|
415 | if amt > L: | |
416 | amt -= L |
|
416 | amt -= L | |
417 | else: |
|
417 | else: | |
418 | s = self._rbuf[:amt] |
|
418 | s = self._rbuf[:amt] | |
419 | self._rbuf = self._rbuf[amt:] |
|
419 | self._rbuf = self._rbuf[amt:] | |
420 | return s |
|
420 | return s | |
421 |
|
421 | |||
422 | s = self._rbuf + self._raw_read(amt) |
|
422 | s = self._rbuf + self._raw_read(amt) | |
423 | self._rbuf = '' |
|
423 | self._rbuf = '' | |
424 | return s |
|
424 | return s | |
425 |
|
425 | |||
426 | # stolen from Python SVN #68532 to fix issue1088 |
|
426 | # stolen from Python SVN #68532 to fix issue1088 | |
427 | def _read_chunked(self, amt): |
|
427 | def _read_chunked(self, amt): | |
428 | chunk_left = self.chunk_left |
|
428 | chunk_left = self.chunk_left | |
429 | value = '' |
|
429 | value = '' | |
430 |
|
430 | |||
431 | # XXX This accumulates chunks by repeated string concatenation, |
|
431 | # XXX This accumulates chunks by repeated string concatenation, | |
432 | # which is not efficient as the number or size of chunks gets big. |
|
432 | # which is not efficient as the number or size of chunks gets big. | |
433 | while True: |
|
433 | while True: | |
434 | if chunk_left is None: |
|
434 | if chunk_left is None: | |
435 | line = self.fp.readline() |
|
435 | line = self.fp.readline() | |
436 | i = line.find(';') |
|
436 | i = line.find(';') | |
437 | if i >= 0: |
|
437 | if i >= 0: | |
438 | line = line[:i] # strip chunk-extensions |
|
438 | line = line[:i] # strip chunk-extensions | |
439 | try: |
|
439 | try: | |
440 | chunk_left = int(line, 16) |
|
440 | chunk_left = int(line, 16) | |
441 | except ValueError: |
|
441 | except ValueError: | |
442 | # close the connection as protocol synchronization is |
|
442 | # close the connection as protocol synchronization is | |
443 | # probably lost |
|
443 | # probably lost | |
444 | self.close() |
|
444 | self.close() | |
445 | raise httplib.IncompleteRead(value) |
|
445 | raise httplib.IncompleteRead(value) | |
446 | if chunk_left == 0: |
|
446 | if chunk_left == 0: | |
447 | break |
|
447 | break | |
448 | if amt is None: |
|
448 | if amt is None: | |
449 | value += self._safe_read(chunk_left) |
|
449 | value += self._safe_read(chunk_left) | |
450 | elif amt < chunk_left: |
|
450 | elif amt < chunk_left: | |
451 | value += self._safe_read(amt) |
|
451 | value += self._safe_read(amt) | |
452 | self.chunk_left = chunk_left - amt |
|
452 | self.chunk_left = chunk_left - amt | |
453 | return value |
|
453 | return value | |
454 | elif amt == chunk_left: |
|
454 | elif amt == chunk_left: | |
455 | value += self._safe_read(amt) |
|
455 | value += self._safe_read(amt) | |
456 | self._safe_read(2) # toss the CRLF at the end of the chunk |
|
456 | self._safe_read(2) # toss the CRLF at the end of the chunk | |
457 | self.chunk_left = None |
|
457 | self.chunk_left = None | |
458 | return value |
|
458 | return value | |
459 | else: |
|
459 | else: | |
460 | value += self._safe_read(chunk_left) |
|
460 | value += self._safe_read(chunk_left) | |
461 | amt -= chunk_left |
|
461 | amt -= chunk_left | |
462 |
|
462 | |||
463 | # we read the whole chunk, get another |
|
463 | # we read the whole chunk, get another | |
464 | self._safe_read(2) # toss the CRLF at the end of the chunk |
|
464 | self._safe_read(2) # toss the CRLF at the end of the chunk | |
465 | chunk_left = None |
|
465 | chunk_left = None | |
466 |
|
466 | |||
467 | # read and discard trailer up to the CRLF terminator |
|
467 | # read and discard trailer up to the CRLF terminator | |
468 | ### note: we shouldn't have any trailers! |
|
468 | ### note: we shouldn't have any trailers! | |
469 | while True: |
|
469 | while True: | |
470 | line = self.fp.readline() |
|
470 | line = self.fp.readline() | |
471 | if not line: |
|
471 | if not line: | |
472 | # a vanishingly small number of sites EOF without |
|
472 | # a vanishingly small number of sites EOF without | |
473 | # sending the trailer |
|
473 | # sending the trailer | |
474 | break |
|
474 | break | |
475 | if line == '\r\n': |
|
475 | if line == '\r\n': | |
476 | break |
|
476 | break | |
477 |
|
477 | |||
478 | # we read everything; close the "file" |
|
478 | # we read everything; close the "file" | |
479 | self.close() |
|
479 | self.close() | |
480 |
|
480 | |||
481 | return value |
|
481 | return value | |
482 |
|
482 | |||
483 | def readline(self, limit=-1): |
|
483 | def readline(self, limit=-1): | |
484 | i = self._rbuf.find('\n') |
|
484 | i = self._rbuf.find('\n') | |
485 | while i < 0 and not (0 < limit <= len(self._rbuf)): |
|
485 | while i < 0 and not (0 < limit <= len(self._rbuf)): | |
486 | new = self._raw_read(self._rbufsize) |
|
486 | new = self._raw_read(self._rbufsize) | |
487 | if not new: |
|
487 | if not new: | |
488 | break |
|
488 | break | |
489 | i = new.find('\n') |
|
489 | i = new.find('\n') | |
490 | if i >= 0: |
|
490 | if i >= 0: | |
491 | i = i + len(self._rbuf) |
|
491 | i = i + len(self._rbuf) | |
492 | self._rbuf = self._rbuf + new |
|
492 | self._rbuf = self._rbuf + new | |
493 | if i < 0: |
|
493 | if i < 0: | |
494 | i = len(self._rbuf) |
|
494 | i = len(self._rbuf) | |
495 | else: |
|
495 | else: | |
496 | i = i + 1 |
|
496 | i = i + 1 | |
497 | if 0 <= limit < len(self._rbuf): |
|
497 | if 0 <= limit < len(self._rbuf): | |
498 | i = limit |
|
498 | i = limit | |
499 | data, self._rbuf = self._rbuf[:i], self._rbuf[i:] |
|
499 | data, self._rbuf = self._rbuf[:i], self._rbuf[i:] | |
500 | return data |
|
500 | return data | |
501 |
|
501 | |||
502 | def readlines(self, sizehint=0): |
|
502 | def readlines(self, sizehint=0): | |
503 | total = 0 |
|
503 | total = 0 | |
504 | list = [] |
|
504 | list = [] | |
505 | while True: |
|
505 | while True: | |
506 | line = self.readline() |
|
506 | line = self.readline() | |
507 | if not line: |
|
507 | if not line: | |
508 | break |
|
508 | break | |
509 | list.append(line) |
|
509 | list.append(line) | |
510 | total += len(line) |
|
510 | total += len(line) | |
511 | if sizehint and total >= sizehint: |
|
511 | if sizehint and total >= sizehint: | |
512 | break |
|
512 | break | |
513 | return list |
|
513 | return list | |
514 |
|
514 | |||
515 | def safesend(self, str): |
|
515 | def safesend(self, str): | |
516 | """Send `str' to the server. |
|
516 | """Send `str' to the server. | |
517 |
|
517 | |||
518 | Shamelessly ripped off from httplib to patch a bad behavior. |
|
518 | Shamelessly ripped off from httplib to patch a bad behavior. | |
519 | """ |
|
519 | """ | |
520 | # _broken_pipe_resp is an attribute we set in this function |
|
520 | # _broken_pipe_resp is an attribute we set in this function | |
521 | # if the socket is closed while we're sending data but |
|
521 | # if the socket is closed while we're sending data but | |
522 | # the server sent us a response before hanging up. |
|
522 | # the server sent us a response before hanging up. | |
523 | # In that case, we want to pretend to send the rest of the |
|
523 | # In that case, we want to pretend to send the rest of the | |
524 | # outgoing data, and then let the user use getresponse() |
|
524 | # outgoing data, and then let the user use getresponse() | |
525 | # (which we wrap) to get this last response before |
|
525 | # (which we wrap) to get this last response before | |
526 | # opening a new socket. |
|
526 | # opening a new socket. | |
527 | if getattr(self, '_broken_pipe_resp', None) is not None: |
|
527 | if getattr(self, '_broken_pipe_resp', None) is not None: | |
528 | return |
|
528 | return | |
529 |
|
529 | |||
530 | if self.sock is None: |
|
530 | if self.sock is None: | |
531 | if self.auto_open: |
|
531 | if self.auto_open: | |
532 | self.connect() |
|
532 | self.connect() | |
533 | else: |
|
533 | else: | |
534 | raise httplib.NotConnected |
|
534 | raise httplib.NotConnected | |
535 |
|
535 | |||
536 | # send the data to the server. if we get a broken pipe, then close |
|
536 | # send the data to the server. if we get a broken pipe, then close | |
537 | # the socket. we want to reconnect when somebody tries to send again. |
|
537 | # the socket. we want to reconnect when somebody tries to send again. | |
538 | # |
|
538 | # | |
539 | # NOTE: we DO propagate the error, though, because we cannot simply |
|
539 | # NOTE: we DO propagate the error, though, because we cannot simply | |
540 | # ignore the error... the caller will know if they can retry. |
|
540 | # ignore the error... the caller will know if they can retry. | |
541 | if self.debuglevel > 0: |
|
541 | if self.debuglevel > 0: | |
542 | print "send:", repr(str) |
|
542 | print "send:", repr(str) | |
543 | try: |
|
543 | try: | |
544 | blocksize = 8192 |
|
544 | blocksize = 8192 | |
545 | read = getattr(str, 'read', None) |
|
545 | read = getattr(str, 'read', None) | |
546 | if read is not None: |
|
546 | if read is not None: | |
547 | if self.debuglevel > 0: |
|
547 | if self.debuglevel > 0: | |
548 | print "sending a read()able" |
|
548 | print "sending a read()able" | |
549 | data = read(blocksize) |
|
549 | data = read(blocksize) | |
550 | while data: |
|
550 | while data: | |
551 | self.sock.sendall(data) |
|
551 | self.sock.sendall(data) | |
552 | data = read(blocksize) |
|
552 | data = read(blocksize) | |
553 | else: |
|
553 | else: | |
554 | self.sock.sendall(str) |
|
554 | self.sock.sendall(str) | |
555 | except socket.error, v: |
|
555 | except socket.error, v: | |
556 | reraise = True |
|
556 | reraise = True | |
557 | if v[0] == errno.EPIPE: # Broken pipe |
|
557 | if v[0] == errno.EPIPE: # Broken pipe | |
558 | if self._HTTPConnection__state == httplib._CS_REQ_SENT: |
|
558 | if self._HTTPConnection__state == httplib._CS_REQ_SENT: | |
559 | self._broken_pipe_resp = None |
|
559 | self._broken_pipe_resp = None | |
560 | self._broken_pipe_resp = self.getresponse() |
|
560 | self._broken_pipe_resp = self.getresponse() | |
561 | reraise = False |
|
561 | reraise = False | |
562 | self.close() |
|
562 | self.close() | |
563 | if reraise: |
|
563 | if reraise: | |
564 | raise |
|
564 | raise | |
565 |
|
565 | |||
566 | def wrapgetresponse(cls): |
|
566 | def wrapgetresponse(cls): | |
567 | """Wraps getresponse in cls with a broken-pipe sane version. |
|
567 | """Wraps getresponse in cls with a broken-pipe sane version. | |
568 | """ |
|
568 | """ | |
569 | def safegetresponse(self): |
|
569 | def safegetresponse(self): | |
570 | # In safesend() we might set the _broken_pipe_resp |
|
570 | # In safesend() we might set the _broken_pipe_resp | |
571 | # attribute, in which case the socket has already |
|
571 | # attribute, in which case the socket has already | |
572 | # been closed and we just need to give them the response |
|
572 | # been closed and we just need to give them the response | |
573 | # back. Otherwise, we use the normal response path. |
|
573 | # back. Otherwise, we use the normal response path. | |
574 | r = getattr(self, '_broken_pipe_resp', None) |
|
574 | r = getattr(self, '_broken_pipe_resp', None) | |
575 | if r is not None: |
|
575 | if r is not None: | |
576 | return r |
|
576 | return r | |
577 | return cls.getresponse(self) |
|
577 | return cls.getresponse(self) | |
578 | safegetresponse.__doc__ = cls.getresponse.__doc__ |
|
578 | safegetresponse.__doc__ = cls.getresponse.__doc__ | |
579 | return safegetresponse |
|
579 | return safegetresponse | |
580 |
|
580 | |||
581 | class HTTPConnection(httplib.HTTPConnection): |
|
581 | class HTTPConnection(httplib.HTTPConnection): | |
582 | # use the modified response class |
|
582 | # use the modified response class | |
583 | response_class = HTTPResponse |
|
583 | response_class = HTTPResponse | |
584 | send = safesend |
|
584 | send = safesend | |
585 | getresponse = wrapgetresponse(httplib.HTTPConnection) |
|
585 | getresponse = wrapgetresponse(httplib.HTTPConnection) | |
586 |
|
586 | |||
587 |
|
587 | |||
588 | ######################################################################### |
|
588 | ######################################################################### | |
589 | ##### TEST FUNCTIONS |
|
589 | ##### TEST FUNCTIONS | |
590 | ######################################################################### |
|
590 | ######################################################################### | |
591 |
|
591 | |||
592 | def error_handler(url): |
|
592 | def error_handler(url): | |
593 | global HANDLE_ERRORS |
|
593 | global HANDLE_ERRORS | |
594 | orig = HANDLE_ERRORS |
|
594 | orig = HANDLE_ERRORS | |
595 | keepalive_handler = HTTPHandler() |
|
595 | keepalive_handler = HTTPHandler() | |
596 | opener = urllib2.build_opener(keepalive_handler) |
|
596 | opener = urllib2.build_opener(keepalive_handler) | |
597 | urllib2.install_opener(opener) |
|
597 | urllib2.install_opener(opener) | |
598 | pos = {0: 'off', 1: 'on'} |
|
598 | pos = {0: 'off', 1: 'on'} | |
599 | for i in (0, 1): |
|
599 | for i in (0, 1): | |
600 | print " fancy error handling %s (HANDLE_ERRORS = %i)" % (pos[i], i) |
|
600 | print " fancy error handling %s (HANDLE_ERRORS = %i)" % (pos[i], i) | |
601 | HANDLE_ERRORS = i |
|
601 | HANDLE_ERRORS = i | |
602 | try: |
|
602 | try: | |
603 | fo = urllib2.urlopen(url) |
|
603 | fo = urllib2.urlopen(url) | |
604 | fo.read() |
|
604 | fo.read() | |
605 | fo.close() |
|
605 | fo.close() | |
606 | try: |
|
606 | try: | |
607 | status, reason = fo.status, fo.reason |
|
607 | status, reason = fo.status, fo.reason | |
608 | except AttributeError: |
|
608 | except AttributeError: | |
609 | status, reason = None, None |
|
609 | status, reason = None, None | |
610 | except IOError, e: |
|
610 | except IOError, e: | |
611 | print " EXCEPTION: %s" % e |
|
611 | print " EXCEPTION: %s" % e | |
612 | raise |
|
612 | raise | |
613 | else: |
|
613 | else: | |
614 | print " status = %s, reason = %s" % (status, reason) |
|
614 | print " status = %s, reason = %s" % (status, reason) | |
615 | HANDLE_ERRORS = orig |
|
615 | HANDLE_ERRORS = orig | |
616 | hosts = keepalive_handler.open_connections() |
|
616 | hosts = keepalive_handler.open_connections() | |
617 | print "open connections:", hosts |
|
617 | print "open connections:", hosts | |
618 | keepalive_handler.close_all() |
|
618 | keepalive_handler.close_all() | |
619 |
|
619 | |||
620 | def md5(s): |
|
620 | def md5(s): | |
621 | try: |
|
621 | try: | |
622 | from hashlib import md5 as _md5 |
|
622 | from hashlib import md5 as _md5 | |
623 | except ImportError: |
|
623 | except ImportError: | |
624 | from md5 import md5 as _md5 |
|
624 | from md5 import md5 as _md5 | |
625 | global md5 |
|
625 | global md5 | |
626 | md5 = _md5 |
|
626 | md5 = _md5 | |
627 | return _md5(s) |
|
627 | return _md5(s) | |
628 |
|
628 | |||
629 | def continuity(url): |
|
629 | def continuity(url): | |
630 | format = '%25s: %s' |
|
630 | format = '%25s: %s' | |
631 |
|
631 | |||
632 | # first fetch the file with the normal http handler |
|
632 | # first fetch the file with the normal http handler | |
633 | opener = urllib2.build_opener() |
|
633 | opener = urllib2.build_opener() | |
634 | urllib2.install_opener(opener) |
|
634 | urllib2.install_opener(opener) | |
635 | fo = urllib2.urlopen(url) |
|
635 | fo = urllib2.urlopen(url) | |
636 | foo = fo.read() |
|
636 | foo = fo.read() | |
637 | fo.close() |
|
637 | fo.close() | |
638 |
m = md5 |
|
638 | m = md5(foo) | |
639 | print format % ('normal urllib', m.hexdigest()) |
|
639 | print format % ('normal urllib', m.hexdigest()) | |
640 |
|
640 | |||
641 | # now install the keepalive handler and try again |
|
641 | # now install the keepalive handler and try again | |
642 | opener = urllib2.build_opener(HTTPHandler()) |
|
642 | opener = urllib2.build_opener(HTTPHandler()) | |
643 | urllib2.install_opener(opener) |
|
643 | urllib2.install_opener(opener) | |
644 |
|
644 | |||
645 | fo = urllib2.urlopen(url) |
|
645 | fo = urllib2.urlopen(url) | |
646 | foo = fo.read() |
|
646 | foo = fo.read() | |
647 | fo.close() |
|
647 | fo.close() | |
648 |
m = md5 |
|
648 | m = md5(foo) | |
649 | print format % ('keepalive read', m.hexdigest()) |
|
649 | print format % ('keepalive read', m.hexdigest()) | |
650 |
|
650 | |||
651 | fo = urllib2.urlopen(url) |
|
651 | fo = urllib2.urlopen(url) | |
652 | foo = '' |
|
652 | foo = '' | |
653 | while True: |
|
653 | while True: | |
654 | f = fo.readline() |
|
654 | f = fo.readline() | |
655 | if f: |
|
655 | if f: | |
656 | foo = foo + f |
|
656 | foo = foo + f | |
657 | else: break |
|
657 | else: break | |
658 | fo.close() |
|
658 | fo.close() | |
659 |
m = md5 |
|
659 | m = md5(foo) | |
660 | print format % ('keepalive readline', m.hexdigest()) |
|
660 | print format % ('keepalive readline', m.hexdigest()) | |
661 |
|
661 | |||
662 | def comp(N, url): |
|
662 | def comp(N, url): | |
663 | print ' making %i connections to:\n %s' % (N, url) |
|
663 | print ' making %i connections to:\n %s' % (N, url) | |
664 |
|
664 | |||
665 | sys.stdout.write(' first using the normal urllib handlers') |
|
665 | sys.stdout.write(' first using the normal urllib handlers') | |
666 | # first use normal opener |
|
666 | # first use normal opener | |
667 | opener = urllib2.build_opener() |
|
667 | opener = urllib2.build_opener() | |
668 | urllib2.install_opener(opener) |
|
668 | urllib2.install_opener(opener) | |
669 | t1 = fetch(N, url) |
|
669 | t1 = fetch(N, url) | |
670 | print ' TIME: %.3f s' % t1 |
|
670 | print ' TIME: %.3f s' % t1 | |
671 |
|
671 | |||
672 | sys.stdout.write(' now using the keepalive handler ') |
|
672 | sys.stdout.write(' now using the keepalive handler ') | |
673 | # now install the keepalive handler and try again |
|
673 | # now install the keepalive handler and try again | |
674 | opener = urllib2.build_opener(HTTPHandler()) |
|
674 | opener = urllib2.build_opener(HTTPHandler()) | |
675 | urllib2.install_opener(opener) |
|
675 | urllib2.install_opener(opener) | |
676 | t2 = fetch(N, url) |
|
676 | t2 = fetch(N, url) | |
677 | print ' TIME: %.3f s' % t2 |
|
677 | print ' TIME: %.3f s' % t2 | |
678 | print ' improvement factor: %.2f' % (t1 / t2) |
|
678 | print ' improvement factor: %.2f' % (t1 / t2) | |
679 |
|
679 | |||
680 | def fetch(N, url, delay=0): |
|
680 | def fetch(N, url, delay=0): | |
681 | import time |
|
681 | import time | |
682 | lens = [] |
|
682 | lens = [] | |
683 | starttime = time.time() |
|
683 | starttime = time.time() | |
684 | for i in range(N): |
|
684 | for i in range(N): | |
685 | if delay and i > 0: |
|
685 | if delay and i > 0: | |
686 | time.sleep(delay) |
|
686 | time.sleep(delay) | |
687 | fo = urllib2.urlopen(url) |
|
687 | fo = urllib2.urlopen(url) | |
688 | foo = fo.read() |
|
688 | foo = fo.read() | |
689 | fo.close() |
|
689 | fo.close() | |
690 | lens.append(len(foo)) |
|
690 | lens.append(len(foo)) | |
691 | diff = time.time() - starttime |
|
691 | diff = time.time() - starttime | |
692 |
|
692 | |||
693 | j = 0 |
|
693 | j = 0 | |
694 | for i in lens[1:]: |
|
694 | for i in lens[1:]: | |
695 | j = j + 1 |
|
695 | j = j + 1 | |
696 | if not i == lens[0]: |
|
696 | if not i == lens[0]: | |
697 | print "WARNING: inconsistent length on read %i: %i" % (j, i) |
|
697 | print "WARNING: inconsistent length on read %i: %i" % (j, i) | |
698 |
|
698 | |||
699 | return diff |
|
699 | return diff | |
700 |
|
700 | |||
701 | def test_timeout(url): |
|
701 | def test_timeout(url): | |
702 | global DEBUG |
|
702 | global DEBUG | |
703 | dbbackup = DEBUG |
|
703 | dbbackup = DEBUG | |
704 | class FakeLogger(object): |
|
704 | class FakeLogger(object): | |
705 | def debug(self, msg, *args): |
|
705 | def debug(self, msg, *args): | |
706 | print msg % args |
|
706 | print msg % args | |
707 | info = warning = error = debug |
|
707 | info = warning = error = debug | |
708 | DEBUG = FakeLogger() |
|
708 | DEBUG = FakeLogger() | |
709 | print " fetching the file to establish a connection" |
|
709 | print " fetching the file to establish a connection" | |
710 | fo = urllib2.urlopen(url) |
|
710 | fo = urllib2.urlopen(url) | |
711 | data1 = fo.read() |
|
711 | data1 = fo.read() | |
712 | fo.close() |
|
712 | fo.close() | |
713 |
|
713 | |||
714 | i = 20 |
|
714 | i = 20 | |
715 | print " waiting %i seconds for the server to close the connection" % i |
|
715 | print " waiting %i seconds for the server to close the connection" % i | |
716 | while i > 0: |
|
716 | while i > 0: | |
717 | sys.stdout.write('\r %2i' % i) |
|
717 | sys.stdout.write('\r %2i' % i) | |
718 | sys.stdout.flush() |
|
718 | sys.stdout.flush() | |
719 | time.sleep(1) |
|
719 | time.sleep(1) | |
720 | i -= 1 |
|
720 | i -= 1 | |
721 | sys.stderr.write('\r') |
|
721 | sys.stderr.write('\r') | |
722 |
|
722 | |||
723 | print " fetching the file a second time" |
|
723 | print " fetching the file a second time" | |
724 | fo = urllib2.urlopen(url) |
|
724 | fo = urllib2.urlopen(url) | |
725 | data2 = fo.read() |
|
725 | data2 = fo.read() | |
726 | fo.close() |
|
726 | fo.close() | |
727 |
|
727 | |||
728 | if data1 == data2: |
|
728 | if data1 == data2: | |
729 | print ' data are identical' |
|
729 | print ' data are identical' | |
730 | else: |
|
730 | else: | |
731 | print ' ERROR: DATA DIFFER' |
|
731 | print ' ERROR: DATA DIFFER' | |
732 |
|
732 | |||
733 | DEBUG = dbbackup |
|
733 | DEBUG = dbbackup | |
734 |
|
734 | |||
735 |
|
735 | |||
736 | def test(url, N=10): |
|
736 | def test(url, N=10): | |
737 | print "checking error handler (do this on a non-200)" |
|
737 | print "checking error handler (do this on a non-200)" | |
738 | try: error_handler(url) |
|
738 | try: error_handler(url) | |
739 | except IOError: |
|
739 | except IOError: | |
740 | print "exiting - exception will prevent further tests" |
|
740 | print "exiting - exception will prevent further tests" | |
741 | sys.exit() |
|
741 | sys.exit() | |
742 |
|
742 | |||
743 | print "performing continuity test (making sure stuff isn't corrupted)" |
|
743 | print "performing continuity test (making sure stuff isn't corrupted)" | |
744 | continuity(url) |
|
744 | continuity(url) | |
745 |
|
745 | |||
746 | print "performing speed comparison" |
|
746 | print "performing speed comparison" | |
747 | comp(N, url) |
|
747 | comp(N, url) | |
748 |
|
748 | |||
749 | print "performing dropped-connection check" |
|
749 | print "performing dropped-connection check" | |
750 | test_timeout(url) |
|
750 | test_timeout(url) | |
751 |
|
751 | |||
752 | if __name__ == '__main__': |
|
752 | if __name__ == '__main__': | |
753 | import time |
|
753 | import time | |
754 | import sys |
|
754 | import sys | |
755 | try: |
|
755 | try: | |
756 | N = int(sys.argv[1]) |
|
756 | N = int(sys.argv[1]) | |
757 | url = sys.argv[2] |
|
757 | url = sys.argv[2] | |
758 | except (IndexError, ValueError): |
|
758 | except (IndexError, ValueError): | |
759 | print "%s <integer> <url>" % sys.argv[0] |
|
759 | print "%s <integer> <url>" % sys.argv[0] | |
760 | else: |
|
760 | else: | |
761 | test(url, N) |
|
761 | test(url, N) |
General Comments 0
You need to be logged in to leave comments.
Login now