Show More
@@ -1,661 +1,671 b'' | |||
|
1 | 1 | # This library is free software; you can redistribute it and/or |
|
2 | 2 | # modify it under the terms of the GNU Lesser General Public |
|
3 | 3 | # License as published by the Free Software Foundation; either |
|
4 | 4 | # version 2.1 of the License, or (at your option) any later version. |
|
5 | 5 | # |
|
6 | 6 | # This library is distributed in the hope that it will be useful, |
|
7 | 7 | # but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
8 | 8 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
9 | 9 | # Lesser General Public License for more details. |
|
10 | 10 | # |
|
11 | 11 | # You should have received a copy of the GNU Lesser General Public |
|
12 | 12 | # License along with this library; if not, write to the |
|
13 | 13 | # Free Software Foundation, Inc., |
|
14 | 14 | # 59 Temple Place, Suite 330, |
|
15 | 15 | # Boston, MA 02111-1307 USA |
|
16 | 16 | |
|
17 | 17 | # This file is part of urlgrabber, a high-level cross-protocol url-grabber |
|
18 | 18 | # Copyright 2002-2004 Michael D. Stenner, Ryan Tomayko |
|
19 | 19 | |
|
20 | 20 | # Modified by Benoit Boissinot: |
|
21 | 21 | # - fix for digest auth (inspired from urllib2.py @ Python v2.4) |
|
22 | 22 | # Modified by Dirkjan Ochtman: |
|
23 | 23 | # - import md5 function from a local util module |
|
24 | # Modified by Martin Geisler: | |
|
25 | # - moved md5 function from local util module to this module | |
|
24 | 26 | |
|
25 | 27 | """An HTTP handler for urllib2 that supports HTTP 1.1 and keepalive. |
|
26 | 28 | |
|
27 | 29 | >>> import urllib2 |
|
28 | 30 | >>> from keepalive import HTTPHandler |
|
29 | 31 | >>> keepalive_handler = HTTPHandler() |
|
30 | 32 | >>> opener = urllib2.build_opener(keepalive_handler) |
|
31 | 33 | >>> urllib2.install_opener(opener) |
|
32 | 34 | >>> |
|
33 | 35 | >>> fo = urllib2.urlopen('http://www.python.org') |
|
34 | 36 | |
|
35 | 37 | If a connection to a given host is requested, and all of the existing |
|
36 | 38 | connections are still in use, another connection will be opened. If |
|
37 | 39 | the handler tries to use an existing connection but it fails in some |
|
38 | 40 | way, it will be closed and removed from the pool. |
|
39 | 41 | |
|
40 | 42 | To remove the handler, simply re-run build_opener with no arguments, and |
|
41 | 43 | install that opener. |
|
42 | 44 | |
|
43 | 45 | You can explicitly close connections by using the close_connection() |
|
44 | 46 | method of the returned file-like object (described below) or you can |
|
45 | 47 | use the handler methods: |
|
46 | 48 | |
|
47 | 49 | close_connection(host) |
|
48 | 50 | close_all() |
|
49 | 51 | open_connections() |
|
50 | 52 | |
|
51 | 53 | NOTE: using the close_connection and close_all methods of the handler |
|
52 | 54 | should be done with care when using multiple threads. |
|
53 | 55 | * there is nothing that prevents another thread from creating new |
|
54 | 56 | connections immediately after connections are closed |
|
55 | 57 | * no checks are done to prevent in-use connections from being closed |
|
56 | 58 | |
|
57 | 59 | >>> keepalive_handler.close_all() |
|
58 | 60 | |
|
59 | 61 | EXTRA ATTRIBUTES AND METHODS |
|
60 | 62 | |
|
61 | 63 | Upon a status of 200, the object returned has a few additional |
|
62 | 64 | attributes and methods, which should not be used if you want to |
|
63 | 65 | remain consistent with the normal urllib2-returned objects: |
|
64 | 66 | |
|
65 | 67 | close_connection() - close the connection to the host |
|
66 | 68 | readlines() - you know, readlines() |
|
67 | 69 | status - the return status (ie 404) |
|
68 | 70 | reason - english translation of status (ie 'File not found') |
|
69 | 71 | |
|
70 | 72 | If you want the best of both worlds, use this inside an |
|
71 | 73 | AttributeError-catching try: |
|
72 | 74 | |
|
73 | 75 | >>> try: status = fo.status |
|
74 | 76 | >>> except AttributeError: status = None |
|
75 | 77 | |
|
76 | 78 | Unfortunately, these are ONLY there if status == 200, so it's not |
|
77 | 79 | easy to distinguish between non-200 responses. The reason is that |
|
78 | 80 | urllib2 tries to do clever things with error codes 301, 302, 401, |
|
79 | 81 | and 407, and it wraps the object upon return. |
|
80 | 82 | |
|
81 | 83 | For python versions earlier than 2.4, you can avoid this fancy error |
|
82 | 84 | handling by setting the module-level global HANDLE_ERRORS to zero. |
|
83 | 85 | You see, prior to 2.4, it's the HTTP Handler's job to determine what |
|
84 | 86 | to handle specially, and what to just pass up. HANDLE_ERRORS == 0 |
|
85 | 87 | means "pass everything up". In python 2.4, however, this job no |
|
86 | 88 | longer belongs to the HTTP Handler and is now done by a NEW handler, |
|
87 | 89 | HTTPErrorProcessor. Here's the bottom line: |
|
88 | 90 | |
|
89 | 91 | python version < 2.4 |
|
90 | 92 | HANDLE_ERRORS == 1 (default) pass up 200, treat the rest as |
|
91 | 93 | errors |
|
92 | 94 | HANDLE_ERRORS == 0 pass everything up, error processing is |
|
93 | 95 | left to the calling code |
|
94 | 96 | python version >= 2.4 |
|
95 | 97 | HANDLE_ERRORS == 1 pass up 200, treat the rest as errors |
|
96 | 98 | HANDLE_ERRORS == 0 (default) pass everything up, let the |
|
97 | 99 | other handlers (specifically, |
|
98 | 100 | HTTPErrorProcessor) decide what to do |
|
99 | 101 | |
|
100 | 102 | In practice, setting the variable either way makes little difference |
|
101 | 103 | in python 2.4, so for the most consistent behavior across versions, |
|
102 | 104 | you probably just want to use the defaults, which will give you |
|
103 | 105 | exceptions on errors. |
|
104 | 106 | |
|
105 | 107 | """ |
|
106 | 108 | |
|
107 | 109 | # $Id: keepalive.py,v 1.14 2006/04/04 21:00:32 mstenner Exp $ |
|
108 | 110 | |
|
109 | 111 | import urllib2 |
|
110 | 112 | import httplib |
|
111 | 113 | import socket |
|
112 | 114 | import thread |
|
113 | 115 | |
|
114 | 116 | DEBUG = None |
|
115 | 117 | |
|
116 | 118 | import sys |
|
117 | 119 | if sys.version_info < (2, 4): HANDLE_ERRORS = 1 |
|
118 | 120 | else: HANDLE_ERRORS = 0 |
|
119 | 121 | |
|
120 | 122 | class ConnectionManager: |
|
121 | 123 | """ |
|
122 | 124 | The connection manager must be able to: |
|
123 | 125 | * keep track of all existing |
|
124 | 126 | """ |
|
125 | 127 | def __init__(self): |
|
126 | 128 | self._lock = thread.allocate_lock() |
|
127 | 129 | self._hostmap = {} # map hosts to a list of connections |
|
128 | 130 | self._connmap = {} # map connections to host |
|
129 | 131 | self._readymap = {} # map connection to ready state |
|
130 | 132 | |
|
131 | 133 | def add(self, host, connection, ready): |
|
132 | 134 | self._lock.acquire() |
|
133 | 135 | try: |
|
134 | 136 | if not host in self._hostmap: self._hostmap[host] = [] |
|
135 | 137 | self._hostmap[host].append(connection) |
|
136 | 138 | self._connmap[connection] = host |
|
137 | 139 | self._readymap[connection] = ready |
|
138 | 140 | finally: |
|
139 | 141 | self._lock.release() |
|
140 | 142 | |
|
141 | 143 | def remove(self, connection): |
|
142 | 144 | self._lock.acquire() |
|
143 | 145 | try: |
|
144 | 146 | try: |
|
145 | 147 | host = self._connmap[connection] |
|
146 | 148 | except KeyError: |
|
147 | 149 | pass |
|
148 | 150 | else: |
|
149 | 151 | del self._connmap[connection] |
|
150 | 152 | del self._readymap[connection] |
|
151 | 153 | self._hostmap[host].remove(connection) |
|
152 | 154 | if not self._hostmap[host]: del self._hostmap[host] |
|
153 | 155 | finally: |
|
154 | 156 | self._lock.release() |
|
155 | 157 | |
|
156 | 158 | def set_ready(self, connection, ready): |
|
157 | 159 | try: self._readymap[connection] = ready |
|
158 | 160 | except KeyError: pass |
|
159 | 161 | |
|
160 | 162 | def get_ready_conn(self, host): |
|
161 | 163 | conn = None |
|
162 | 164 | self._lock.acquire() |
|
163 | 165 | try: |
|
164 | 166 | if host in self._hostmap: |
|
165 | 167 | for c in self._hostmap[host]: |
|
166 | 168 | if self._readymap[c]: |
|
167 | 169 | self._readymap[c] = 0 |
|
168 | 170 | conn = c |
|
169 | 171 | break |
|
170 | 172 | finally: |
|
171 | 173 | self._lock.release() |
|
172 | 174 | return conn |
|
173 | 175 | |
|
174 | 176 | def get_all(self, host=None): |
|
175 | 177 | if host: |
|
176 | 178 | return list(self._hostmap.get(host, [])) |
|
177 | 179 | else: |
|
178 | 180 | return dict(self._hostmap) |
|
179 | 181 | |
|
180 | 182 | class KeepAliveHandler: |
|
181 | 183 | def __init__(self): |
|
182 | 184 | self._cm = ConnectionManager() |
|
183 | 185 | |
|
184 | 186 | #### Connection Management |
|
185 | 187 | def open_connections(self): |
|
186 | 188 | """return a list of connected hosts and the number of connections |
|
187 | 189 | to each. [('foo.com:80', 2), ('bar.org', 1)]""" |
|
188 | 190 | return [(host, len(li)) for (host, li) in self._cm.get_all().items()] |
|
189 | 191 | |
|
190 | 192 | def close_connection(self, host): |
|
191 | 193 | """close connection(s) to <host> |
|
192 | 194 | host is the host:port spec, as in 'www.cnn.com:8080' as passed in. |
|
193 | 195 | no error occurs if there is no connection to that host.""" |
|
194 | 196 | for h in self._cm.get_all(host): |
|
195 | 197 | self._cm.remove(h) |
|
196 | 198 | h.close() |
|
197 | 199 | |
|
198 | 200 | def close_all(self): |
|
199 | 201 | """close all open connections""" |
|
200 | 202 | for host, conns in self._cm.get_all().iteritems(): |
|
201 | 203 | for h in conns: |
|
202 | 204 | self._cm.remove(h) |
|
203 | 205 | h.close() |
|
204 | 206 | |
|
205 | 207 | def _request_closed(self, request, host, connection): |
|
206 | 208 | """tells us that this request is now closed and the the |
|
207 | 209 | connection is ready for another request""" |
|
208 | 210 | self._cm.set_ready(connection, 1) |
|
209 | 211 | |
|
210 | 212 | def _remove_connection(self, host, connection, close=0): |
|
211 | 213 | if close: connection.close() |
|
212 | 214 | self._cm.remove(connection) |
|
213 | 215 | |
|
214 | 216 | #### Transaction Execution |
|
215 | 217 | def http_open(self, req): |
|
216 | 218 | return self.do_open(HTTPConnection, req) |
|
217 | 219 | |
|
218 | 220 | def do_open(self, http_class, req): |
|
219 | 221 | host = req.get_host() |
|
220 | 222 | if not host: |
|
221 | 223 | raise urllib2.URLError('no host given') |
|
222 | 224 | |
|
223 | 225 | try: |
|
224 | 226 | h = self._cm.get_ready_conn(host) |
|
225 | 227 | while h: |
|
226 | 228 | r = self._reuse_connection(h, req, host) |
|
227 | 229 | |
|
228 | 230 | # if this response is non-None, then it worked and we're |
|
229 | 231 | # done. Break out, skipping the else block. |
|
230 | 232 | if r: break |
|
231 | 233 | |
|
232 | 234 | # connection is bad - possibly closed by server |
|
233 | 235 | # discard it and ask for the next free connection |
|
234 | 236 | h.close() |
|
235 | 237 | self._cm.remove(h) |
|
236 | 238 | h = self._cm.get_ready_conn(host) |
|
237 | 239 | else: |
|
238 | 240 | # no (working) free connections were found. Create a new one. |
|
239 | 241 | h = http_class(host) |
|
240 | 242 | if DEBUG: DEBUG.info("creating new connection to %s (%d)", |
|
241 | 243 | host, id(h)) |
|
242 | 244 | self._cm.add(host, h, 0) |
|
243 | 245 | self._start_transaction(h, req) |
|
244 | 246 | r = h.getresponse() |
|
245 | 247 | except (socket.error, httplib.HTTPException), err: |
|
246 | 248 | raise urllib2.URLError(err) |
|
247 | 249 | |
|
248 | 250 | # if not a persistent connection, don't try to reuse it |
|
249 | 251 | if r.will_close: self._cm.remove(h) |
|
250 | 252 | |
|
251 | 253 | if DEBUG: DEBUG.info("STATUS: %s, %s", r.status, r.reason) |
|
252 | 254 | r._handler = self |
|
253 | 255 | r._host = host |
|
254 | 256 | r._url = req.get_full_url() |
|
255 | 257 | r._connection = h |
|
256 | 258 | r.code = r.status |
|
257 | 259 | r.headers = r.msg |
|
258 | 260 | r.msg = r.reason |
|
259 | 261 | |
|
260 | 262 | if r.status == 200 or not HANDLE_ERRORS: |
|
261 | 263 | return r |
|
262 | 264 | else: |
|
263 | 265 | return self.parent.error('http', req, r, |
|
264 | 266 | r.status, r.msg, r.headers) |
|
265 | 267 | |
|
266 | 268 | def _reuse_connection(self, h, req, host): |
|
267 | 269 | """start the transaction with a re-used connection |
|
268 | 270 | return a response object (r) upon success or None on failure. |
|
269 | 271 | This DOES not close or remove bad connections in cases where |
|
270 | 272 | it returns. However, if an unexpected exception occurs, it |
|
271 | 273 | will close and remove the connection before re-raising. |
|
272 | 274 | """ |
|
273 | 275 | try: |
|
274 | 276 | self._start_transaction(h, req) |
|
275 | 277 | r = h.getresponse() |
|
276 | 278 | # note: just because we got something back doesn't mean it |
|
277 | 279 | # worked. We'll check the version below, too. |
|
278 | 280 | except (socket.error, httplib.HTTPException): |
|
279 | 281 | r = None |
|
280 | 282 | except: |
|
281 | 283 | # adding this block just in case we've missed |
|
282 | 284 | # something we will still raise the exception, but |
|
283 | 285 | # lets try and close the connection and remove it |
|
284 | 286 | # first. We previously got into a nasty loop |
|
285 | 287 | # where an exception was uncaught, and so the |
|
286 | 288 | # connection stayed open. On the next try, the |
|
287 | 289 | # same exception was raised, etc. The tradeoff is |
|
288 | 290 | # that it's now possible this call will raise |
|
289 | 291 | # a DIFFERENT exception |
|
290 | 292 | if DEBUG: DEBUG.error("unexpected exception - closing " + \ |
|
291 | 293 | "connection to %s (%d)", host, id(h)) |
|
292 | 294 | self._cm.remove(h) |
|
293 | 295 | h.close() |
|
294 | 296 | raise |
|
295 | 297 | |
|
296 | 298 | if r is None or r.version == 9: |
|
297 | 299 | # httplib falls back to assuming HTTP 0.9 if it gets a |
|
298 | 300 | # bad header back. This is most likely to happen if |
|
299 | 301 | # the socket has been closed by the server since we |
|
300 | 302 | # last used the connection. |
|
301 | 303 | if DEBUG: DEBUG.info("failed to re-use connection to %s (%d)", |
|
302 | 304 | host, id(h)) |
|
303 | 305 | r = None |
|
304 | 306 | else: |
|
305 | 307 | if DEBUG: DEBUG.info("re-using connection to %s (%d)", host, id(h)) |
|
306 | 308 | |
|
307 | 309 | return r |
|
308 | 310 | |
|
309 | 311 | def _start_transaction(self, h, req): |
|
310 | 312 | # What follows mostly reimplements HTTPConnection.request() |
|
311 | 313 | # except it adds self.parent.addheaders in the mix. |
|
312 | 314 | headers = req.headers.copy() |
|
313 | 315 | if sys.version_info >= (2, 4): |
|
314 | 316 | headers.update(req.unredirected_hdrs) |
|
315 | 317 | headers.update(self.parent.addheaders) |
|
316 | 318 | headers = dict((n.lower(), v) for n,v in headers.items()) |
|
317 | 319 | skipheaders = {} |
|
318 | 320 | for n in ('host', 'accept-encoding'): |
|
319 | 321 | if n in headers: |
|
320 | 322 | skipheaders['skip_' + n.replace('-', '_')] = 1 |
|
321 | 323 | try: |
|
322 | 324 | if req.has_data(): |
|
323 | 325 | data = req.get_data() |
|
324 | 326 | h.putrequest('POST', req.get_selector(), **skipheaders) |
|
325 | 327 | if 'content-type' not in headers: |
|
326 | 328 | h.putheader('Content-type', |
|
327 | 329 | 'application/x-www-form-urlencoded') |
|
328 | 330 | if 'content-length' not in headers: |
|
329 | 331 | h.putheader('Content-length', '%d' % len(data)) |
|
330 | 332 | else: |
|
331 | 333 | h.putrequest('GET', req.get_selector(), **skipheaders) |
|
332 | 334 | except (socket.error), err: |
|
333 | 335 | raise urllib2.URLError(err) |
|
334 | 336 | for k, v in headers.items(): |
|
335 | 337 | h.putheader(k, v) |
|
336 | 338 | h.endheaders() |
|
337 | 339 | if req.has_data(): |
|
338 | 340 | h.send(data) |
|
339 | 341 | |
|
340 | 342 | class HTTPHandler(KeepAliveHandler, urllib2.HTTPHandler): |
|
341 | 343 | pass |
|
342 | 344 | |
|
343 | 345 | class HTTPResponse(httplib.HTTPResponse): |
|
344 | 346 | # we need to subclass HTTPResponse in order to |
|
345 | 347 | # 1) add readline() and readlines() methods |
|
346 | 348 | # 2) add close_connection() methods |
|
347 | 349 | # 3) add info() and geturl() methods |
|
348 | 350 | |
|
349 | 351 | # in order to add readline(), read must be modified to deal with a |
|
350 | 352 | # buffer. example: readline must read a buffer and then spit back |
|
351 | 353 | # one line at a time. The only real alternative is to read one |
|
352 | 354 | # BYTE at a time (ick). Once something has been read, it can't be |
|
353 | 355 | # put back (ok, maybe it can, but that's even uglier than this), |
|
354 | 356 | # so if you THEN do a normal read, you must first take stuff from |
|
355 | 357 | # the buffer. |
|
356 | 358 | |
|
357 | 359 | # the read method wraps the original to accomodate buffering, |
|
358 | 360 | # although read() never adds to the buffer. |
|
359 | 361 | # Both readline and readlines have been stolen with almost no |
|
360 | 362 | # modification from socket.py |
|
361 | 363 | |
|
362 | 364 | |
|
363 | 365 | def __init__(self, sock, debuglevel=0, strict=0, method=None): |
|
364 | 366 | if method: # the httplib in python 2.3 uses the method arg |
|
365 | 367 | httplib.HTTPResponse.__init__(self, sock, debuglevel, method) |
|
366 | 368 | else: # 2.2 doesn't |
|
367 | 369 | httplib.HTTPResponse.__init__(self, sock, debuglevel) |
|
368 | 370 | self.fileno = sock.fileno |
|
369 | 371 | self.code = None |
|
370 | 372 | self._rbuf = '' |
|
371 | 373 | self._rbufsize = 8096 |
|
372 | 374 | self._handler = None # inserted by the handler later |
|
373 | 375 | self._host = None # (same) |
|
374 | 376 | self._url = None # (same) |
|
375 | 377 | self._connection = None # (same) |
|
376 | 378 | |
|
377 | 379 | _raw_read = httplib.HTTPResponse.read |
|
378 | 380 | |
|
379 | 381 | def close(self): |
|
380 | 382 | if self.fp: |
|
381 | 383 | self.fp.close() |
|
382 | 384 | self.fp = None |
|
383 | 385 | if self._handler: |
|
384 | 386 | self._handler._request_closed(self, self._host, |
|
385 | 387 | self._connection) |
|
386 | 388 | |
|
387 | 389 | def close_connection(self): |
|
388 | 390 | self._handler._remove_connection(self._host, self._connection, close=1) |
|
389 | 391 | self.close() |
|
390 | 392 | |
|
391 | 393 | def info(self): |
|
392 | 394 | return self.headers |
|
393 | 395 | |
|
394 | 396 | def geturl(self): |
|
395 | 397 | return self._url |
|
396 | 398 | |
|
397 | 399 | def read(self, amt=None): |
|
398 | 400 | # the _rbuf test is only in this first if for speed. It's not |
|
399 | 401 | # logically necessary |
|
400 | 402 | if self._rbuf and not amt is None: |
|
401 | 403 | L = len(self._rbuf) |
|
402 | 404 | if amt > L: |
|
403 | 405 | amt -= L |
|
404 | 406 | else: |
|
405 | 407 | s = self._rbuf[:amt] |
|
406 | 408 | self._rbuf = self._rbuf[amt:] |
|
407 | 409 | return s |
|
408 | 410 | |
|
409 | 411 | s = self._rbuf + self._raw_read(amt) |
|
410 | 412 | self._rbuf = '' |
|
411 | 413 | return s |
|
412 | 414 | |
|
413 | 415 | # stolen from Python SVN #68532 to fix issue1088 |
|
414 | 416 | def _read_chunked(self, amt): |
|
415 | 417 | chunk_left = self.chunk_left |
|
416 | 418 | value = '' |
|
417 | 419 | |
|
418 | 420 | # XXX This accumulates chunks by repeated string concatenation, |
|
419 | 421 | # which is not efficient as the number or size of chunks gets big. |
|
420 | 422 | while True: |
|
421 | 423 | if chunk_left is None: |
|
422 | 424 | line = self.fp.readline() |
|
423 | 425 | i = line.find(';') |
|
424 | 426 | if i >= 0: |
|
425 | 427 | line = line[:i] # strip chunk-extensions |
|
426 | 428 | try: |
|
427 | 429 | chunk_left = int(line, 16) |
|
428 | 430 | except ValueError: |
|
429 | 431 | # close the connection as protocol synchronisation is |
|
430 | 432 | # probably lost |
|
431 | 433 | self.close() |
|
432 | 434 | raise httplib.IncompleteRead(value) |
|
433 | 435 | if chunk_left == 0: |
|
434 | 436 | break |
|
435 | 437 | if amt is None: |
|
436 | 438 | value += self._safe_read(chunk_left) |
|
437 | 439 | elif amt < chunk_left: |
|
438 | 440 | value += self._safe_read(amt) |
|
439 | 441 | self.chunk_left = chunk_left - amt |
|
440 | 442 | return value |
|
441 | 443 | elif amt == chunk_left: |
|
442 | 444 | value += self._safe_read(amt) |
|
443 | 445 | self._safe_read(2) # toss the CRLF at the end of the chunk |
|
444 | 446 | self.chunk_left = None |
|
445 | 447 | return value |
|
446 | 448 | else: |
|
447 | 449 | value += self._safe_read(chunk_left) |
|
448 | 450 | amt -= chunk_left |
|
449 | 451 | |
|
450 | 452 | # we read the whole chunk, get another |
|
451 | 453 | self._safe_read(2) # toss the CRLF at the end of the chunk |
|
452 | 454 | chunk_left = None |
|
453 | 455 | |
|
454 | 456 | # read and discard trailer up to the CRLF terminator |
|
455 | 457 | ### note: we shouldn't have any trailers! |
|
456 | 458 | while True: |
|
457 | 459 | line = self.fp.readline() |
|
458 | 460 | if not line: |
|
459 | 461 | # a vanishingly small number of sites EOF without |
|
460 | 462 | # sending the trailer |
|
461 | 463 | break |
|
462 | 464 | if line == '\r\n': |
|
463 | 465 | break |
|
464 | 466 | |
|
465 | 467 | # we read everything; close the "file" |
|
466 | 468 | self.close() |
|
467 | 469 | |
|
468 | 470 | return value |
|
469 | 471 | |
|
470 | 472 | def readline(self, limit=-1): |
|
471 | 473 | i = self._rbuf.find('\n') |
|
472 | 474 | while i < 0 and not (0 < limit <= len(self._rbuf)): |
|
473 | 475 | new = self._raw_read(self._rbufsize) |
|
474 | 476 | if not new: break |
|
475 | 477 | i = new.find('\n') |
|
476 | 478 | if i >= 0: i = i + len(self._rbuf) |
|
477 | 479 | self._rbuf = self._rbuf + new |
|
478 | 480 | if i < 0: i = len(self._rbuf) |
|
479 | 481 | else: i = i+1 |
|
480 | 482 | if 0 <= limit < len(self._rbuf): i = limit |
|
481 | 483 | data, self._rbuf = self._rbuf[:i], self._rbuf[i:] |
|
482 | 484 | return data |
|
483 | 485 | |
|
484 | 486 | def readlines(self, sizehint = 0): |
|
485 | 487 | total = 0 |
|
486 | 488 | list = [] |
|
487 | 489 | while 1: |
|
488 | 490 | line = self.readline() |
|
489 | 491 | if not line: break |
|
490 | 492 | list.append(line) |
|
491 | 493 | total += len(line) |
|
492 | 494 | if sizehint and total >= sizehint: |
|
493 | 495 | break |
|
494 | 496 | return list |
|
495 | 497 | |
|
496 | 498 | |
|
497 | 499 | class HTTPConnection(httplib.HTTPConnection): |
|
498 | 500 | # use the modified response class |
|
499 | 501 | response_class = HTTPResponse |
|
500 | 502 | |
|
501 | 503 | ######################################################################### |
|
502 | 504 | ##### TEST FUNCTIONS |
|
503 | 505 | ######################################################################### |
|
504 | 506 | |
|
505 | 507 | def error_handler(url): |
|
506 | 508 | global HANDLE_ERRORS |
|
507 | 509 | orig = HANDLE_ERRORS |
|
508 | 510 | keepalive_handler = HTTPHandler() |
|
509 | 511 | opener = urllib2.build_opener(keepalive_handler) |
|
510 | 512 | urllib2.install_opener(opener) |
|
511 | 513 | pos = {0: 'off', 1: 'on'} |
|
512 | 514 | for i in (0, 1): |
|
513 | 515 | print " fancy error handling %s (HANDLE_ERRORS = %i)" % (pos[i], i) |
|
514 | 516 | HANDLE_ERRORS = i |
|
515 | 517 | try: |
|
516 | 518 | fo = urllib2.urlopen(url) |
|
517 | 519 | fo.read() |
|
518 | 520 | fo.close() |
|
519 | 521 | try: status, reason = fo.status, fo.reason |
|
520 | 522 | except AttributeError: status, reason = None, None |
|
521 | 523 | except IOError, e: |
|
522 | 524 | print " EXCEPTION: %s" % e |
|
523 | 525 | raise |
|
524 | 526 | else: |
|
525 | 527 | print " status = %s, reason = %s" % (status, reason) |
|
526 | 528 | HANDLE_ERRORS = orig |
|
527 | 529 | hosts = keepalive_handler.open_connections() |
|
528 | 530 | print "open connections:", hosts |
|
529 | 531 | keepalive_handler.close_all() |
|
530 | 532 | |
|
533 | def md5(s): | |
|
534 | try: | |
|
535 | from hashlib import md5 as _md5 | |
|
536 | except ImportError: | |
|
537 | from md5 import md5 as _md5 | |
|
538 | global md5 | |
|
539 | md5 = _md5 | |
|
540 | return _md5(s) | |
|
541 | ||
|
531 | 542 | def continuity(url): |
|
532 | from util import md5 | |
|
533 | 543 | format = '%25s: %s' |
|
534 | 544 | |
|
535 | 545 | # first fetch the file with the normal http handler |
|
536 | 546 | opener = urllib2.build_opener() |
|
537 | 547 | urllib2.install_opener(opener) |
|
538 | 548 | fo = urllib2.urlopen(url) |
|
539 | 549 | foo = fo.read() |
|
540 | 550 | fo.close() |
|
541 | 551 | m = md5.new(foo) |
|
542 | 552 | print format % ('normal urllib', m.hexdigest()) |
|
543 | 553 | |
|
544 | 554 | # now install the keepalive handler and try again |
|
545 | 555 | opener = urllib2.build_opener(HTTPHandler()) |
|
546 | 556 | urllib2.install_opener(opener) |
|
547 | 557 | |
|
548 | 558 | fo = urllib2.urlopen(url) |
|
549 | 559 | foo = fo.read() |
|
550 | 560 | fo.close() |
|
551 | 561 | m = md5.new(foo) |
|
552 | 562 | print format % ('keepalive read', m.hexdigest()) |
|
553 | 563 | |
|
554 | 564 | fo = urllib2.urlopen(url) |
|
555 | 565 | foo = '' |
|
556 | 566 | while 1: |
|
557 | 567 | f = fo.readline() |
|
558 | 568 | if f: foo = foo + f |
|
559 | 569 | else: break |
|
560 | 570 | fo.close() |
|
561 | 571 | m = md5.new(foo) |
|
562 | 572 | print format % ('keepalive readline', m.hexdigest()) |
|
563 | 573 | |
|
564 | 574 | def comp(N, url): |
|
565 | 575 | print ' making %i connections to:\n %s' % (N, url) |
|
566 | 576 | |
|
567 | 577 | sys.stdout.write(' first using the normal urllib handlers') |
|
568 | 578 | # first use normal opener |
|
569 | 579 | opener = urllib2.build_opener() |
|
570 | 580 | urllib2.install_opener(opener) |
|
571 | 581 | t1 = fetch(N, url) |
|
572 | 582 | print ' TIME: %.3f s' % t1 |
|
573 | 583 | |
|
574 | 584 | sys.stdout.write(' now using the keepalive handler ') |
|
575 | 585 | # now install the keepalive handler and try again |
|
576 | 586 | opener = urllib2.build_opener(HTTPHandler()) |
|
577 | 587 | urllib2.install_opener(opener) |
|
578 | 588 | t2 = fetch(N, url) |
|
579 | 589 | print ' TIME: %.3f s' % t2 |
|
580 | 590 | print ' improvement factor: %.2f' % (t1/t2, ) |
|
581 | 591 | |
|
582 | 592 | def fetch(N, url, delay=0): |
|
583 | 593 | import time |
|
584 | 594 | lens = [] |
|
585 | 595 | starttime = time.time() |
|
586 | 596 | for i in range(N): |
|
587 | 597 | if delay and i > 0: time.sleep(delay) |
|
588 | 598 | fo = urllib2.urlopen(url) |
|
589 | 599 | foo = fo.read() |
|
590 | 600 | fo.close() |
|
591 | 601 | lens.append(len(foo)) |
|
592 | 602 | diff = time.time() - starttime |
|
593 | 603 | |
|
594 | 604 | j = 0 |
|
595 | 605 | for i in lens[1:]: |
|
596 | 606 | j = j + 1 |
|
597 | 607 | if not i == lens[0]: |
|
598 | 608 | print "WARNING: inconsistent length on read %i: %i" % (j, i) |
|
599 | 609 | |
|
600 | 610 | return diff |
|
601 | 611 | |
|
602 | 612 | def test_timeout(url): |
|
603 | 613 | global DEBUG |
|
604 | 614 | dbbackup = DEBUG |
|
605 | 615 | class FakeLogger: |
|
606 | 616 | def debug(self, msg, *args): print msg % args |
|
607 | 617 | info = warning = error = debug |
|
608 | 618 | DEBUG = FakeLogger() |
|
609 | 619 | print " fetching the file to establish a connection" |
|
610 | 620 | fo = urllib2.urlopen(url) |
|
611 | 621 | data1 = fo.read() |
|
612 | 622 | fo.close() |
|
613 | 623 | |
|
614 | 624 | i = 20 |
|
615 | 625 | print " waiting %i seconds for the server to close the connection" % i |
|
616 | 626 | while i > 0: |
|
617 | 627 | sys.stdout.write('\r %2i' % i) |
|
618 | 628 | sys.stdout.flush() |
|
619 | 629 | time.sleep(1) |
|
620 | 630 | i -= 1 |
|
621 | 631 | sys.stderr.write('\r') |
|
622 | 632 | |
|
623 | 633 | print " fetching the file a second time" |
|
624 | 634 | fo = urllib2.urlopen(url) |
|
625 | 635 | data2 = fo.read() |
|
626 | 636 | fo.close() |
|
627 | 637 | |
|
628 | 638 | if data1 == data2: |
|
629 | 639 | print ' data are identical' |
|
630 | 640 | else: |
|
631 | 641 | print ' ERROR: DATA DIFFER' |
|
632 | 642 | |
|
633 | 643 | DEBUG = dbbackup |
|
634 | 644 | |
|
635 | 645 | |
|
636 | 646 | def test(url, N=10): |
|
637 | 647 | print "checking error hander (do this on a non-200)" |
|
638 | 648 | try: error_handler(url) |
|
639 | 649 | except IOError: |
|
640 | 650 | print "exiting - exception will prevent further tests" |
|
641 | 651 | sys.exit() |
|
642 | 652 | |
|
643 | 653 | print "performing continuity test (making sure stuff isn't corrupted)" |
|
644 | 654 | continuity(url) |
|
645 | 655 | |
|
646 | 656 | print "performing speed comparison" |
|
647 | 657 | comp(N, url) |
|
648 | 658 | |
|
649 | 659 | print "performing dropped-connection check" |
|
650 | 660 | test_timeout(url) |
|
651 | 661 | |
|
652 | 662 | if __name__ == '__main__': |
|
653 | 663 | import time |
|
654 | 664 | import sys |
|
655 | 665 | try: |
|
656 | 666 | N = int(sys.argv[1]) |
|
657 | 667 | url = sys.argv[2] |
|
658 | 668 | except: |
|
659 | 669 | print "%s <integer> <url>" % sys.argv[0] |
|
660 | 670 | else: |
|
661 | 671 | test(url, N) |
@@ -1,1481 +1,1471 b'' | |||
|
1 | 1 | # util.py - Mercurial utility functions and platform specfic implementations |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005 K. Thananchayan <thananck@yahoo.com> |
|
4 | 4 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
5 | 5 | # Copyright 2006 Vadim Gelfer <vadim.gelfer@gmail.com> |
|
6 | 6 | # |
|
7 | 7 | # This software may be used and distributed according to the terms of the |
|
8 | 8 | # GNU General Public License version 2, incorporated herein by reference. |
|
9 | 9 | |
|
10 | 10 | """Mercurial utility functions and platform specfic implementations. |
|
11 | 11 | |
|
12 | 12 | This contains helper routines that are independent of the SCM core and |
|
13 | 13 | hide platform-specific details from the core. |
|
14 | 14 | """ |
|
15 | 15 | |
|
16 | 16 | from i18n import _ |
|
17 | 17 | import cStringIO, errno, re, shutil, sys, tempfile, traceback, error |
|
18 | 18 | import os, stat, threading, time, calendar, glob, osutil, random |
|
19 | 19 | import imp |
|
20 | 20 | |
|
21 | 21 | # Python compatibility |
|
22 | 22 | |
|
23 | def md5(s): | |
|
24 | try: | |
|
25 | import hashlib | |
|
26 | _md5 = hashlib.md5 | |
|
27 | except ImportError: | |
|
28 | from md5 import md5 as _md5 | |
|
29 | global md5 | |
|
30 | md5 = _md5 | |
|
31 | return _md5(s) | |
|
32 | ||
|
33 | 23 | def sha1(s): |
|
34 | 24 | try: |
|
35 | 25 | import hashlib |
|
36 | 26 | _sha1 = hashlib.sha1 |
|
37 | 27 | except ImportError: |
|
38 | 28 | from sha import sha as _sha1 |
|
39 | 29 | global sha1 |
|
40 | 30 | sha1 = _sha1 |
|
41 | 31 | return _sha1(s) |
|
42 | 32 | |
|
43 | 33 | import subprocess |
|
44 | 34 | closefds = os.name == 'posix' |
|
45 | 35 | def popen2(cmd, mode='t', bufsize=-1): |
|
46 | 36 | p = subprocess.Popen(cmd, shell=True, bufsize=bufsize, |
|
47 | 37 | close_fds=closefds, |
|
48 | 38 | stdin=subprocess.PIPE, stdout=subprocess.PIPE) |
|
49 | 39 | return p.stdin, p.stdout |
|
50 | 40 | def popen3(cmd, mode='t', bufsize=-1): |
|
51 | 41 | p = subprocess.Popen(cmd, shell=True, bufsize=bufsize, |
|
52 | 42 | close_fds=closefds, |
|
53 | 43 | stdin=subprocess.PIPE, stdout=subprocess.PIPE, |
|
54 | 44 | stderr=subprocess.PIPE) |
|
55 | 45 | return p.stdin, p.stdout, p.stderr |
|
56 | 46 | def Popen3(cmd, capturestderr=False, bufsize=-1): |
|
57 | 47 | stderr = capturestderr and subprocess.PIPE or None |
|
58 | 48 | p = subprocess.Popen(cmd, shell=True, bufsize=bufsize, |
|
59 | 49 | close_fds=closefds, |
|
60 | 50 | stdin=subprocess.PIPE, stdout=subprocess.PIPE, |
|
61 | 51 | stderr=stderr) |
|
62 | 52 | p.fromchild = p.stdout |
|
63 | 53 | p.tochild = p.stdin |
|
64 | 54 | p.childerr = p.stderr |
|
65 | 55 | return p |
|
66 | 56 | |
|
67 | 57 | def version(): |
|
68 | 58 | """Return version information if available.""" |
|
69 | 59 | try: |
|
70 | 60 | import __version__ |
|
71 | 61 | return __version__.version |
|
72 | 62 | except ImportError: |
|
73 | 63 | return 'unknown' |
|
74 | 64 | |
|
75 | 65 | # used by parsedate |
|
76 | 66 | defaultdateformats = ( |
|
77 | 67 | '%Y-%m-%d %H:%M:%S', |
|
78 | 68 | '%Y-%m-%d %I:%M:%S%p', |
|
79 | 69 | '%Y-%m-%d %H:%M', |
|
80 | 70 | '%Y-%m-%d %I:%M%p', |
|
81 | 71 | '%Y-%m-%d', |
|
82 | 72 | '%m-%d', |
|
83 | 73 | '%m/%d', |
|
84 | 74 | '%m/%d/%y', |
|
85 | 75 | '%m/%d/%Y', |
|
86 | 76 | '%a %b %d %H:%M:%S %Y', |
|
87 | 77 | '%a %b %d %I:%M:%S%p %Y', |
|
88 | 78 | '%a, %d %b %Y %H:%M:%S', # GNU coreutils "/bin/date --rfc-2822" |
|
89 | 79 | '%b %d %H:%M:%S %Y', |
|
90 | 80 | '%b %d %I:%M:%S%p %Y', |
|
91 | 81 | '%b %d %H:%M:%S', |
|
92 | 82 | '%b %d %I:%M:%S%p', |
|
93 | 83 | '%b %d %H:%M', |
|
94 | 84 | '%b %d %I:%M%p', |
|
95 | 85 | '%b %d %Y', |
|
96 | 86 | '%b %d', |
|
97 | 87 | '%H:%M:%S', |
|
98 | 88 | '%I:%M:%SP', |
|
99 | 89 | '%H:%M', |
|
100 | 90 | '%I:%M%p', |
|
101 | 91 | ) |
|
102 | 92 | |
|
103 | 93 | extendeddateformats = defaultdateformats + ( |
|
104 | 94 | "%Y", |
|
105 | 95 | "%Y-%m", |
|
106 | 96 | "%b", |
|
107 | 97 | "%b %Y", |
|
108 | 98 | ) |
|
109 | 99 | |
|
110 | 100 | def cachefunc(func): |
|
111 | 101 | '''cache the result of function calls''' |
|
112 | 102 | # XXX doesn't handle keywords args |
|
113 | 103 | cache = {} |
|
114 | 104 | if func.func_code.co_argcount == 1: |
|
115 | 105 | # we gain a small amount of time because |
|
116 | 106 | # we don't need to pack/unpack the list |
|
117 | 107 | def f(arg): |
|
118 | 108 | if arg not in cache: |
|
119 | 109 | cache[arg] = func(arg) |
|
120 | 110 | return cache[arg] |
|
121 | 111 | else: |
|
122 | 112 | def f(*args): |
|
123 | 113 | if args not in cache: |
|
124 | 114 | cache[args] = func(*args) |
|
125 | 115 | return cache[args] |
|
126 | 116 | |
|
127 | 117 | return f |
|
128 | 118 | |
|
129 | 119 | class propertycache(object): |
|
130 | 120 | def __init__(self, func): |
|
131 | 121 | self.func = func |
|
132 | 122 | self.name = func.__name__ |
|
133 | 123 | def __get__(self, obj, type=None): |
|
134 | 124 | result = self.func(obj) |
|
135 | 125 | setattr(obj, self.name, result) |
|
136 | 126 | return result |
|
137 | 127 | |
|
138 | 128 | def pipefilter(s, cmd): |
|
139 | 129 | '''filter string S through command CMD, returning its output''' |
|
140 | 130 | (pin, pout) = popen2(cmd, 'b') |
|
141 | 131 | def writer(): |
|
142 | 132 | try: |
|
143 | 133 | pin.write(s) |
|
144 | 134 | pin.close() |
|
145 | 135 | except IOError, inst: |
|
146 | 136 | if inst.errno != errno.EPIPE: |
|
147 | 137 | raise |
|
148 | 138 | |
|
149 | 139 | # we should use select instead on UNIX, but this will work on most |
|
150 | 140 | # systems, including Windows |
|
151 | 141 | w = threading.Thread(target=writer) |
|
152 | 142 | w.start() |
|
153 | 143 | f = pout.read() |
|
154 | 144 | pout.close() |
|
155 | 145 | w.join() |
|
156 | 146 | return f |
|
157 | 147 | |
|
158 | 148 | def tempfilter(s, cmd): |
|
159 | 149 | '''filter string S through a pair of temporary files with CMD. |
|
160 | 150 | CMD is used as a template to create the real command to be run, |
|
161 | 151 | with the strings INFILE and OUTFILE replaced by the real names of |
|
162 | 152 | the temporary files generated.''' |
|
163 | 153 | inname, outname = None, None |
|
164 | 154 | try: |
|
165 | 155 | infd, inname = tempfile.mkstemp(prefix='hg-filter-in-') |
|
166 | 156 | fp = os.fdopen(infd, 'wb') |
|
167 | 157 | fp.write(s) |
|
168 | 158 | fp.close() |
|
169 | 159 | outfd, outname = tempfile.mkstemp(prefix='hg-filter-out-') |
|
170 | 160 | os.close(outfd) |
|
171 | 161 | cmd = cmd.replace('INFILE', inname) |
|
172 | 162 | cmd = cmd.replace('OUTFILE', outname) |
|
173 | 163 | code = os.system(cmd) |
|
174 | 164 | if sys.platform == 'OpenVMS' and code & 1: |
|
175 | 165 | code = 0 |
|
176 | 166 | if code: raise Abort(_("command '%s' failed: %s") % |
|
177 | 167 | (cmd, explain_exit(code))) |
|
178 | 168 | return open(outname, 'rb').read() |
|
179 | 169 | finally: |
|
180 | 170 | try: |
|
181 | 171 | if inname: os.unlink(inname) |
|
182 | 172 | except: pass |
|
183 | 173 | try: |
|
184 | 174 | if outname: os.unlink(outname) |
|
185 | 175 | except: pass |
|
186 | 176 | |
|
187 | 177 | filtertable = { |
|
188 | 178 | 'tempfile:': tempfilter, |
|
189 | 179 | 'pipe:': pipefilter, |
|
190 | 180 | } |
|
191 | 181 | |
|
192 | 182 | def filter(s, cmd): |
|
193 | 183 | "filter a string through a command that transforms its input to its output" |
|
194 | 184 | for name, fn in filtertable.iteritems(): |
|
195 | 185 | if cmd.startswith(name): |
|
196 | 186 | return fn(s, cmd[len(name):].lstrip()) |
|
197 | 187 | return pipefilter(s, cmd) |
|
198 | 188 | |
|
199 | 189 | def binary(s): |
|
200 | 190 | """return true if a string is binary data""" |
|
201 | 191 | return bool(s and '\0' in s) |
|
202 | 192 | |
|
203 | 193 | def increasingchunks(source, min=1024, max=65536): |
|
204 | 194 | '''return no less than min bytes per chunk while data remains, |
|
205 | 195 | doubling min after each chunk until it reaches max''' |
|
206 | 196 | def log2(x): |
|
207 | 197 | if not x: |
|
208 | 198 | return 0 |
|
209 | 199 | i = 0 |
|
210 | 200 | while x: |
|
211 | 201 | x >>= 1 |
|
212 | 202 | i += 1 |
|
213 | 203 | return i - 1 |
|
214 | 204 | |
|
215 | 205 | buf = [] |
|
216 | 206 | blen = 0 |
|
217 | 207 | for chunk in source: |
|
218 | 208 | buf.append(chunk) |
|
219 | 209 | blen += len(chunk) |
|
220 | 210 | if blen >= min: |
|
221 | 211 | if min < max: |
|
222 | 212 | min = min << 1 |
|
223 | 213 | nmin = 1 << log2(blen) |
|
224 | 214 | if nmin > min: |
|
225 | 215 | min = nmin |
|
226 | 216 | if min > max: |
|
227 | 217 | min = max |
|
228 | 218 | yield ''.join(buf) |
|
229 | 219 | blen = 0 |
|
230 | 220 | buf = [] |
|
231 | 221 | if buf: |
|
232 | 222 | yield ''.join(buf) |
|
233 | 223 | |
|
234 | 224 | Abort = error.Abort |
|
235 | 225 | |
|
236 | 226 | def always(fn): return True |
|
237 | 227 | def never(fn): return False |
|
238 | 228 | |
|
239 | 229 | def patkind(name, default): |
|
240 | 230 | """Split a string into an optional pattern kind prefix and the |
|
241 | 231 | actual pattern.""" |
|
242 | 232 | for prefix in 're', 'glob', 'path', 'relglob', 'relpath', 'relre': |
|
243 | 233 | if name.startswith(prefix + ':'): return name.split(':', 1) |
|
244 | 234 | return default, name |
|
245 | 235 | |
|
246 | 236 | def globre(pat, head='^', tail='$'): |
|
247 | 237 | "convert a glob pattern into a regexp" |
|
248 | 238 | i, n = 0, len(pat) |
|
249 | 239 | res = '' |
|
250 | 240 | group = 0 |
|
251 | 241 | def peek(): return i < n and pat[i] |
|
252 | 242 | while i < n: |
|
253 | 243 | c = pat[i] |
|
254 | 244 | i = i+1 |
|
255 | 245 | if c == '*': |
|
256 | 246 | if peek() == '*': |
|
257 | 247 | i += 1 |
|
258 | 248 | res += '.*' |
|
259 | 249 | else: |
|
260 | 250 | res += '[^/]*' |
|
261 | 251 | elif c == '?': |
|
262 | 252 | res += '.' |
|
263 | 253 | elif c == '[': |
|
264 | 254 | j = i |
|
265 | 255 | if j < n and pat[j] in '!]': |
|
266 | 256 | j += 1 |
|
267 | 257 | while j < n and pat[j] != ']': |
|
268 | 258 | j += 1 |
|
269 | 259 | if j >= n: |
|
270 | 260 | res += '\\[' |
|
271 | 261 | else: |
|
272 | 262 | stuff = pat[i:j].replace('\\','\\\\') |
|
273 | 263 | i = j + 1 |
|
274 | 264 | if stuff[0] == '!': |
|
275 | 265 | stuff = '^' + stuff[1:] |
|
276 | 266 | elif stuff[0] == '^': |
|
277 | 267 | stuff = '\\' + stuff |
|
278 | 268 | res = '%s[%s]' % (res, stuff) |
|
279 | 269 | elif c == '{': |
|
280 | 270 | group += 1 |
|
281 | 271 | res += '(?:' |
|
282 | 272 | elif c == '}' and group: |
|
283 | 273 | res += ')' |
|
284 | 274 | group -= 1 |
|
285 | 275 | elif c == ',' and group: |
|
286 | 276 | res += '|' |
|
287 | 277 | elif c == '\\': |
|
288 | 278 | p = peek() |
|
289 | 279 | if p: |
|
290 | 280 | i += 1 |
|
291 | 281 | res += re.escape(p) |
|
292 | 282 | else: |
|
293 | 283 | res += re.escape(c) |
|
294 | 284 | else: |
|
295 | 285 | res += re.escape(c) |
|
296 | 286 | return head + res + tail |
|
297 | 287 | |
|
298 | 288 | _globchars = {'[': 1, '{': 1, '*': 1, '?': 1} |
|
299 | 289 | |
|
300 | 290 | def pathto(root, n1, n2): |
|
301 | 291 | '''return the relative path from one place to another. |
|
302 | 292 | root should use os.sep to separate directories |
|
303 | 293 | n1 should use os.sep to separate directories |
|
304 | 294 | n2 should use "/" to separate directories |
|
305 | 295 | returns an os.sep-separated path. |
|
306 | 296 | |
|
307 | 297 | If n1 is a relative path, it's assumed it's |
|
308 | 298 | relative to root. |
|
309 | 299 | n2 should always be relative to root. |
|
310 | 300 | ''' |
|
311 | 301 | if not n1: return localpath(n2) |
|
312 | 302 | if os.path.isabs(n1): |
|
313 | 303 | if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]: |
|
314 | 304 | return os.path.join(root, localpath(n2)) |
|
315 | 305 | n2 = '/'.join((pconvert(root), n2)) |
|
316 | 306 | a, b = splitpath(n1), n2.split('/') |
|
317 | 307 | a.reverse() |
|
318 | 308 | b.reverse() |
|
319 | 309 | while a and b and a[-1] == b[-1]: |
|
320 | 310 | a.pop() |
|
321 | 311 | b.pop() |
|
322 | 312 | b.reverse() |
|
323 | 313 | return os.sep.join((['..'] * len(a)) + b) or '.' |
|
324 | 314 | |
|
325 | 315 | def canonpath(root, cwd, myname): |
|
326 | 316 | """return the canonical path of myname, given cwd and root""" |
|
327 | 317 | if root == os.sep: |
|
328 | 318 | rootsep = os.sep |
|
329 | 319 | elif endswithsep(root): |
|
330 | 320 | rootsep = root |
|
331 | 321 | else: |
|
332 | 322 | rootsep = root + os.sep |
|
333 | 323 | name = myname |
|
334 | 324 | if not os.path.isabs(name): |
|
335 | 325 | name = os.path.join(root, cwd, name) |
|
336 | 326 | name = os.path.normpath(name) |
|
337 | 327 | audit_path = path_auditor(root) |
|
338 | 328 | if name != rootsep and name.startswith(rootsep): |
|
339 | 329 | name = name[len(rootsep):] |
|
340 | 330 | audit_path(name) |
|
341 | 331 | return pconvert(name) |
|
342 | 332 | elif name == root: |
|
343 | 333 | return '' |
|
344 | 334 | else: |
|
345 | 335 | # Determine whether `name' is in the hierarchy at or beneath `root', |
|
346 | 336 | # by iterating name=dirname(name) until that causes no change (can't |
|
347 | 337 | # check name == '/', because that doesn't work on windows). For each |
|
348 | 338 | # `name', compare dev/inode numbers. If they match, the list `rel' |
|
349 | 339 | # holds the reversed list of components making up the relative file |
|
350 | 340 | # name we want. |
|
351 | 341 | root_st = os.stat(root) |
|
352 | 342 | rel = [] |
|
353 | 343 | while True: |
|
354 | 344 | try: |
|
355 | 345 | name_st = os.stat(name) |
|
356 | 346 | except OSError: |
|
357 | 347 | break |
|
358 | 348 | if samestat(name_st, root_st): |
|
359 | 349 | if not rel: |
|
360 | 350 | # name was actually the same as root (maybe a symlink) |
|
361 | 351 | return '' |
|
362 | 352 | rel.reverse() |
|
363 | 353 | name = os.path.join(*rel) |
|
364 | 354 | audit_path(name) |
|
365 | 355 | return pconvert(name) |
|
366 | 356 | dirname, basename = os.path.split(name) |
|
367 | 357 | rel.append(basename) |
|
368 | 358 | if dirname == name: |
|
369 | 359 | break |
|
370 | 360 | name = dirname |
|
371 | 361 | |
|
372 | 362 | raise Abort('%s not under root' % myname) |
|
373 | 363 | |
|
374 | 364 | def matcher(canonroot, cwd='', names=[], inc=[], exc=[], src=None, dflt_pat='glob'): |
|
375 | 365 | """build a function to match a set of file patterns |
|
376 | 366 | |
|
377 | 367 | arguments: |
|
378 | 368 | canonroot - the canonical root of the tree you're matching against |
|
379 | 369 | cwd - the current working directory, if relevant |
|
380 | 370 | names - patterns to find |
|
381 | 371 | inc - patterns to include |
|
382 | 372 | exc - patterns to exclude |
|
383 | 373 | dflt_pat - if a pattern in names has no explicit type, assume this one |
|
384 | 374 | src - where these patterns came from (e.g. .hgignore) |
|
385 | 375 | |
|
386 | 376 | a pattern is one of: |
|
387 | 377 | 'glob:<glob>' - a glob relative to cwd |
|
388 | 378 | 're:<regexp>' - a regular expression |
|
389 | 379 | 'path:<path>' - a path relative to canonroot |
|
390 | 380 | 'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs) |
|
391 | 381 | 'relpath:<path>' - a path relative to cwd |
|
392 | 382 | 'relre:<regexp>' - a regexp that doesn't have to match the start of a name |
|
393 | 383 | '<something>' - one of the cases above, selected by the dflt_pat argument |
|
394 | 384 | |
|
395 | 385 | returns: |
|
396 | 386 | a 3-tuple containing |
|
397 | 387 | - list of roots (places where one should start a recursive walk of the fs); |
|
398 | 388 | this often matches the explicit non-pattern names passed in, but also |
|
399 | 389 | includes the initial part of glob: patterns that has no glob characters |
|
400 | 390 | - a bool match(filename) function |
|
401 | 391 | - a bool indicating if any patterns were passed in |
|
402 | 392 | """ |
|
403 | 393 | |
|
404 | 394 | # a common case: no patterns at all |
|
405 | 395 | if not names and not inc and not exc: |
|
406 | 396 | return [], always, False |
|
407 | 397 | |
|
408 | 398 | def contains_glob(name): |
|
409 | 399 | for c in name: |
|
410 | 400 | if c in _globchars: return True |
|
411 | 401 | return False |
|
412 | 402 | |
|
413 | 403 | def regex(kind, name, tail): |
|
414 | 404 | '''convert a pattern into a regular expression''' |
|
415 | 405 | if not name: |
|
416 | 406 | return '' |
|
417 | 407 | if kind == 're': |
|
418 | 408 | return name |
|
419 | 409 | elif kind == 'path': |
|
420 | 410 | return '^' + re.escape(name) + '(?:/|$)' |
|
421 | 411 | elif kind == 'relglob': |
|
422 | 412 | return globre(name, '(?:|.*/)', tail) |
|
423 | 413 | elif kind == 'relpath': |
|
424 | 414 | return re.escape(name) + '(?:/|$)' |
|
425 | 415 | elif kind == 'relre': |
|
426 | 416 | if name.startswith('^'): |
|
427 | 417 | return name |
|
428 | 418 | return '.*' + name |
|
429 | 419 | return globre(name, '', tail) |
|
430 | 420 | |
|
431 | 421 | def matchfn(pats, tail): |
|
432 | 422 | """build a matching function from a set of patterns""" |
|
433 | 423 | if not pats: |
|
434 | 424 | return |
|
435 | 425 | try: |
|
436 | 426 | pat = '(?:%s)' % '|'.join([regex(k, p, tail) for (k, p) in pats]) |
|
437 | 427 | if len(pat) > 20000: |
|
438 | 428 | raise OverflowError() |
|
439 | 429 | return re.compile(pat).match |
|
440 | 430 | except OverflowError: |
|
441 | 431 | # We're using a Python with a tiny regex engine and we |
|
442 | 432 | # made it explode, so we'll divide the pattern list in two |
|
443 | 433 | # until it works |
|
444 | 434 | l = len(pats) |
|
445 | 435 | if l < 2: |
|
446 | 436 | raise |
|
447 | 437 | a, b = matchfn(pats[:l//2], tail), matchfn(pats[l//2:], tail) |
|
448 | 438 | return lambda s: a(s) or b(s) |
|
449 | 439 | except re.error: |
|
450 | 440 | for k, p in pats: |
|
451 | 441 | try: |
|
452 | 442 | re.compile('(?:%s)' % regex(k, p, tail)) |
|
453 | 443 | except re.error: |
|
454 | 444 | if src: |
|
455 | 445 | raise Abort("%s: invalid pattern (%s): %s" % |
|
456 | 446 | (src, k, p)) |
|
457 | 447 | else: |
|
458 | 448 | raise Abort("invalid pattern (%s): %s" % (k, p)) |
|
459 | 449 | raise Abort("invalid pattern") |
|
460 | 450 | |
|
461 | 451 | def globprefix(pat): |
|
462 | 452 | '''return the non-glob prefix of a path, e.g. foo/* -> foo''' |
|
463 | 453 | root = [] |
|
464 | 454 | for p in pat.split('/'): |
|
465 | 455 | if contains_glob(p): break |
|
466 | 456 | root.append(p) |
|
467 | 457 | return '/'.join(root) or '.' |
|
468 | 458 | |
|
469 | 459 | def normalizepats(names, default): |
|
470 | 460 | pats = [] |
|
471 | 461 | roots = [] |
|
472 | 462 | anypats = False |
|
473 | 463 | for kind, name in [patkind(p, default) for p in names]: |
|
474 | 464 | if kind in ('glob', 'relpath'): |
|
475 | 465 | name = canonpath(canonroot, cwd, name) |
|
476 | 466 | elif kind in ('relglob', 'path'): |
|
477 | 467 | name = normpath(name) |
|
478 | 468 | |
|
479 | 469 | pats.append((kind, name)) |
|
480 | 470 | |
|
481 | 471 | if kind in ('glob', 're', 'relglob', 'relre'): |
|
482 | 472 | anypats = True |
|
483 | 473 | |
|
484 | 474 | if kind == 'glob': |
|
485 | 475 | root = globprefix(name) |
|
486 | 476 | roots.append(root) |
|
487 | 477 | elif kind in ('relpath', 'path'): |
|
488 | 478 | roots.append(name or '.') |
|
489 | 479 | elif kind == 'relglob': |
|
490 | 480 | roots.append('.') |
|
491 | 481 | return roots, pats, anypats |
|
492 | 482 | |
|
493 | 483 | roots, pats, anypats = normalizepats(names, dflt_pat) |
|
494 | 484 | |
|
495 | 485 | patmatch = matchfn(pats, '$') or always |
|
496 | 486 | incmatch = always |
|
497 | 487 | if inc: |
|
498 | 488 | dummy, inckinds, dummy = normalizepats(inc, 'glob') |
|
499 | 489 | incmatch = matchfn(inckinds, '(?:/|$)') |
|
500 | 490 | excmatch = never |
|
501 | 491 | if exc: |
|
502 | 492 | dummy, exckinds, dummy = normalizepats(exc, 'glob') |
|
503 | 493 | excmatch = matchfn(exckinds, '(?:/|$)') |
|
504 | 494 | |
|
505 | 495 | if not names and inc and not exc: |
|
506 | 496 | # common case: hgignore patterns |
|
507 | 497 | match = incmatch |
|
508 | 498 | else: |
|
509 | 499 | match = lambda fn: incmatch(fn) and not excmatch(fn) and patmatch(fn) |
|
510 | 500 | |
|
511 | 501 | return (roots, match, (inc or exc or anypats) and True) |
|
512 | 502 | |
|
513 | 503 | _hgexecutable = None |
|
514 | 504 | |
|
515 | 505 | def main_is_frozen(): |
|
516 | 506 | """return True if we are a frozen executable. |
|
517 | 507 | |
|
518 | 508 | The code supports py2exe (most common, Windows only) and tools/freeze |
|
519 | 509 | (portable, not much used). |
|
520 | 510 | """ |
|
521 | 511 | return (hasattr(sys, "frozen") or # new py2exe |
|
522 | 512 | hasattr(sys, "importers") or # old py2exe |
|
523 | 513 | imp.is_frozen("__main__")) # tools/freeze |
|
524 | 514 | |
|
525 | 515 | def hgexecutable(): |
|
526 | 516 | """return location of the 'hg' executable. |
|
527 | 517 | |
|
528 | 518 | Defaults to $HG or 'hg' in the search path. |
|
529 | 519 | """ |
|
530 | 520 | if _hgexecutable is None: |
|
531 | 521 | hg = os.environ.get('HG') |
|
532 | 522 | if hg: |
|
533 | 523 | set_hgexecutable(hg) |
|
534 | 524 | elif main_is_frozen(): |
|
535 | 525 | set_hgexecutable(sys.executable) |
|
536 | 526 | else: |
|
537 | 527 | set_hgexecutable(find_exe('hg') or 'hg') |
|
538 | 528 | return _hgexecutable |
|
539 | 529 | |
|
540 | 530 | def set_hgexecutable(path): |
|
541 | 531 | """set location of the 'hg' executable""" |
|
542 | 532 | global _hgexecutable |
|
543 | 533 | _hgexecutable = path |
|
544 | 534 | |
|
545 | 535 | def system(cmd, environ={}, cwd=None, onerr=None, errprefix=None): |
|
546 | 536 | '''enhanced shell command execution. |
|
547 | 537 | run with environment maybe modified, maybe in different dir. |
|
548 | 538 | |
|
549 | 539 | if command fails and onerr is None, return status. if ui object, |
|
550 | 540 | print error message and return status, else raise onerr object as |
|
551 | 541 | exception.''' |
|
552 | 542 | def py2shell(val): |
|
553 | 543 | 'convert python object into string that is useful to shell' |
|
554 | 544 | if val in (None, False): |
|
555 | 545 | return '0' |
|
556 | 546 | if val == True: |
|
557 | 547 | return '1' |
|
558 | 548 | return str(val) |
|
559 | 549 | oldenv = {} |
|
560 | 550 | for k in environ: |
|
561 | 551 | oldenv[k] = os.environ.get(k) |
|
562 | 552 | if cwd is not None: |
|
563 | 553 | oldcwd = os.getcwd() |
|
564 | 554 | origcmd = cmd |
|
565 | 555 | if os.name == 'nt': |
|
566 | 556 | cmd = '"%s"' % cmd |
|
567 | 557 | try: |
|
568 | 558 | for k, v in environ.iteritems(): |
|
569 | 559 | os.environ[k] = py2shell(v) |
|
570 | 560 | os.environ['HG'] = hgexecutable() |
|
571 | 561 | if cwd is not None and oldcwd != cwd: |
|
572 | 562 | os.chdir(cwd) |
|
573 | 563 | rc = os.system(cmd) |
|
574 | 564 | if sys.platform == 'OpenVMS' and rc & 1: |
|
575 | 565 | rc = 0 |
|
576 | 566 | if rc and onerr: |
|
577 | 567 | errmsg = '%s %s' % (os.path.basename(origcmd.split(None, 1)[0]), |
|
578 | 568 | explain_exit(rc)[0]) |
|
579 | 569 | if errprefix: |
|
580 | 570 | errmsg = '%s: %s' % (errprefix, errmsg) |
|
581 | 571 | try: |
|
582 | 572 | onerr.warn(errmsg + '\n') |
|
583 | 573 | except AttributeError: |
|
584 | 574 | raise onerr(errmsg) |
|
585 | 575 | return rc |
|
586 | 576 | finally: |
|
587 | 577 | for k, v in oldenv.iteritems(): |
|
588 | 578 | if v is None: |
|
589 | 579 | del os.environ[k] |
|
590 | 580 | else: |
|
591 | 581 | os.environ[k] = v |
|
592 | 582 | if cwd is not None and oldcwd != cwd: |
|
593 | 583 | os.chdir(oldcwd) |
|
594 | 584 | |
|
595 | 585 | def checksignature(func): |
|
596 | 586 | '''wrap a function with code to check for calling errors''' |
|
597 | 587 | def check(*args, **kwargs): |
|
598 | 588 | try: |
|
599 | 589 | return func(*args, **kwargs) |
|
600 | 590 | except TypeError: |
|
601 | 591 | if len(traceback.extract_tb(sys.exc_info()[2])) == 1: |
|
602 | 592 | raise error.SignatureError |
|
603 | 593 | raise |
|
604 | 594 | |
|
605 | 595 | return check |
|
606 | 596 | |
|
607 | 597 | # os.path.lexists is not available on python2.3 |
|
608 | 598 | def lexists(filename): |
|
609 | 599 | "test whether a file with this name exists. does not follow symlinks" |
|
610 | 600 | try: |
|
611 | 601 | os.lstat(filename) |
|
612 | 602 | except: |
|
613 | 603 | return False |
|
614 | 604 | return True |
|
615 | 605 | |
|
616 | 606 | def rename(src, dst): |
|
617 | 607 | """forcibly rename a file""" |
|
618 | 608 | try: |
|
619 | 609 | os.rename(src, dst) |
|
620 | 610 | except OSError, err: # FIXME: check err (EEXIST ?) |
|
621 | 611 | |
|
622 | 612 | # On windows, rename to existing file is not allowed, so we |
|
623 | 613 | # must delete destination first. But if a file is open, unlink |
|
624 | 614 | # schedules it for delete but does not delete it. Rename |
|
625 | 615 | # happens immediately even for open files, so we rename |
|
626 | 616 | # destination to a temporary name, then delete that. Then |
|
627 | 617 | # rename is safe to do. |
|
628 | 618 | # The temporary name is chosen at random to avoid the situation |
|
629 | 619 | # where a file is left lying around from a previous aborted run. |
|
630 | 620 | # The usual race condition this introduces can't be avoided as |
|
631 | 621 | # we need the name to rename into, and not the file itself. Due |
|
632 | 622 | # to the nature of the operation however, any races will at worst |
|
633 | 623 | # lead to the rename failing and the current operation aborting. |
|
634 | 624 | |
|
635 | 625 | def tempname(prefix): |
|
636 | 626 | for tries in xrange(10): |
|
637 | 627 | temp = '%s-%08x' % (prefix, random.randint(0, 0xffffffff)) |
|
638 | 628 | if not os.path.exists(temp): |
|
639 | 629 | return temp |
|
640 | 630 | raise IOError, (errno.EEXIST, "No usable temporary filename found") |
|
641 | 631 | |
|
642 | 632 | temp = tempname(dst) |
|
643 | 633 | os.rename(dst, temp) |
|
644 | 634 | os.unlink(temp) |
|
645 | 635 | os.rename(src, dst) |
|
646 | 636 | |
|
647 | 637 | def unlink(f): |
|
648 | 638 | """unlink and remove the directory if it is empty""" |
|
649 | 639 | os.unlink(f) |
|
650 | 640 | # try removing directories that might now be empty |
|
651 | 641 | try: |
|
652 | 642 | os.removedirs(os.path.dirname(f)) |
|
653 | 643 | except OSError: |
|
654 | 644 | pass |
|
655 | 645 | |
|
656 | 646 | def copyfile(src, dest): |
|
657 | 647 | "copy a file, preserving mode and atime/mtime" |
|
658 | 648 | if os.path.islink(src): |
|
659 | 649 | try: |
|
660 | 650 | os.unlink(dest) |
|
661 | 651 | except: |
|
662 | 652 | pass |
|
663 | 653 | os.symlink(os.readlink(src), dest) |
|
664 | 654 | else: |
|
665 | 655 | try: |
|
666 | 656 | shutil.copyfile(src, dest) |
|
667 | 657 | shutil.copystat(src, dest) |
|
668 | 658 | except shutil.Error, inst: |
|
669 | 659 | raise Abort(str(inst)) |
|
670 | 660 | |
|
671 | 661 | def copyfiles(src, dst, hardlink=None): |
|
672 | 662 | """Copy a directory tree using hardlinks if possible""" |
|
673 | 663 | |
|
674 | 664 | if hardlink is None: |
|
675 | 665 | hardlink = (os.stat(src).st_dev == |
|
676 | 666 | os.stat(os.path.dirname(dst)).st_dev) |
|
677 | 667 | |
|
678 | 668 | if os.path.isdir(src): |
|
679 | 669 | os.mkdir(dst) |
|
680 | 670 | for name, kind in osutil.listdir(src): |
|
681 | 671 | srcname = os.path.join(src, name) |
|
682 | 672 | dstname = os.path.join(dst, name) |
|
683 | 673 | copyfiles(srcname, dstname, hardlink) |
|
684 | 674 | else: |
|
685 | 675 | if hardlink: |
|
686 | 676 | try: |
|
687 | 677 | os_link(src, dst) |
|
688 | 678 | except (IOError, OSError): |
|
689 | 679 | hardlink = False |
|
690 | 680 | shutil.copy(src, dst) |
|
691 | 681 | else: |
|
692 | 682 | shutil.copy(src, dst) |
|
693 | 683 | |
|
694 | 684 | class path_auditor(object): |
|
695 | 685 | '''ensure that a filesystem path contains no banned components. |
|
696 | 686 | the following properties of a path are checked: |
|
697 | 687 | |
|
698 | 688 | - under top-level .hg |
|
699 | 689 | - starts at the root of a windows drive |
|
700 | 690 | - contains ".." |
|
701 | 691 | - traverses a symlink (e.g. a/symlink_here/b) |
|
702 | 692 | - inside a nested repository''' |
|
703 | 693 | |
|
704 | 694 | def __init__(self, root): |
|
705 | 695 | self.audited = set() |
|
706 | 696 | self.auditeddir = set() |
|
707 | 697 | self.root = root |
|
708 | 698 | |
|
709 | 699 | def __call__(self, path): |
|
710 | 700 | if path in self.audited: |
|
711 | 701 | return |
|
712 | 702 | normpath = os.path.normcase(path) |
|
713 | 703 | parts = splitpath(normpath) |
|
714 | 704 | if (os.path.splitdrive(path)[0] |
|
715 | 705 | or parts[0].lower() in ('.hg', '.hg.', '') |
|
716 | 706 | or os.pardir in parts): |
|
717 | 707 | raise Abort(_("path contains illegal component: %s") % path) |
|
718 | 708 | if '.hg' in path.lower(): |
|
719 | 709 | lparts = [p.lower() for p in parts] |
|
720 | 710 | for p in '.hg', '.hg.': |
|
721 | 711 | if p in lparts[1:]: |
|
722 | 712 | pos = lparts.index(p) |
|
723 | 713 | base = os.path.join(*parts[:pos]) |
|
724 | 714 | raise Abort(_('path %r is inside repo %r') % (path, base)) |
|
725 | 715 | def check(prefix): |
|
726 | 716 | curpath = os.path.join(self.root, prefix) |
|
727 | 717 | try: |
|
728 | 718 | st = os.lstat(curpath) |
|
729 | 719 | except OSError, err: |
|
730 | 720 | # EINVAL can be raised as invalid path syntax under win32. |
|
731 | 721 | # They must be ignored for patterns can be checked too. |
|
732 | 722 | if err.errno not in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL): |
|
733 | 723 | raise |
|
734 | 724 | else: |
|
735 | 725 | if stat.S_ISLNK(st.st_mode): |
|
736 | 726 | raise Abort(_('path %r traverses symbolic link %r') % |
|
737 | 727 | (path, prefix)) |
|
738 | 728 | elif (stat.S_ISDIR(st.st_mode) and |
|
739 | 729 | os.path.isdir(os.path.join(curpath, '.hg'))): |
|
740 | 730 | raise Abort(_('path %r is inside repo %r') % |
|
741 | 731 | (path, prefix)) |
|
742 | 732 | parts.pop() |
|
743 | 733 | prefixes = [] |
|
744 | 734 | for n in range(len(parts)): |
|
745 | 735 | prefix = os.sep.join(parts) |
|
746 | 736 | if prefix in self.auditeddir: |
|
747 | 737 | break |
|
748 | 738 | check(prefix) |
|
749 | 739 | prefixes.append(prefix) |
|
750 | 740 | parts.pop() |
|
751 | 741 | |
|
752 | 742 | self.audited.add(path) |
|
753 | 743 | # only add prefixes to the cache after checking everything: we don't |
|
754 | 744 | # want to add "foo/bar/baz" before checking if there's a "foo/.hg" |
|
755 | 745 | self.auditeddir.update(prefixes) |
|
756 | 746 | |
|
757 | 747 | def nlinks(pathname): |
|
758 | 748 | """Return number of hardlinks for the given file.""" |
|
759 | 749 | return os.lstat(pathname).st_nlink |
|
760 | 750 | |
|
761 | 751 | if hasattr(os, 'link'): |
|
762 | 752 | os_link = os.link |
|
763 | 753 | else: |
|
764 | 754 | def os_link(src, dst): |
|
765 | 755 | raise OSError(0, _("Hardlinks not supported")) |
|
766 | 756 | |
|
767 | 757 | def lookup_reg(key, name=None, scope=None): |
|
768 | 758 | return None |
|
769 | 759 | |
|
770 | 760 | if os.name == 'nt': |
|
771 | 761 | from windows import * |
|
772 | 762 | def expand_glob(pats): |
|
773 | 763 | '''On Windows, expand the implicit globs in a list of patterns''' |
|
774 | 764 | ret = [] |
|
775 | 765 | for p in pats: |
|
776 | 766 | kind, name = patkind(p, None) |
|
777 | 767 | if kind is None: |
|
778 | 768 | globbed = glob.glob(name) |
|
779 | 769 | if globbed: |
|
780 | 770 | ret.extend(globbed) |
|
781 | 771 | continue |
|
782 | 772 | # if we couldn't expand the glob, just keep it around |
|
783 | 773 | ret.append(p) |
|
784 | 774 | return ret |
|
785 | 775 | else: |
|
786 | 776 | from posix import * |
|
787 | 777 | |
|
788 | 778 | def makelock(info, pathname): |
|
789 | 779 | try: |
|
790 | 780 | return os.symlink(info, pathname) |
|
791 | 781 | except OSError, why: |
|
792 | 782 | if why.errno == errno.EEXIST: |
|
793 | 783 | raise |
|
794 | 784 | except AttributeError: # no symlink in os |
|
795 | 785 | pass |
|
796 | 786 | |
|
797 | 787 | ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL) |
|
798 | 788 | os.write(ld, info) |
|
799 | 789 | os.close(ld) |
|
800 | 790 | |
|
801 | 791 | def readlock(pathname): |
|
802 | 792 | try: |
|
803 | 793 | return os.readlink(pathname) |
|
804 | 794 | except OSError, why: |
|
805 | 795 | if why.errno not in (errno.EINVAL, errno.ENOSYS): |
|
806 | 796 | raise |
|
807 | 797 | except AttributeError: # no symlink in os |
|
808 | 798 | pass |
|
809 | 799 | return posixfile(pathname).read() |
|
810 | 800 | |
|
811 | 801 | def fstat(fp): |
|
812 | 802 | '''stat file object that may not have fileno method.''' |
|
813 | 803 | try: |
|
814 | 804 | return os.fstat(fp.fileno()) |
|
815 | 805 | except AttributeError: |
|
816 | 806 | return os.stat(fp.name) |
|
817 | 807 | |
|
818 | 808 | # File system features |
|
819 | 809 | |
|
820 | 810 | def checkcase(path): |
|
821 | 811 | """ |
|
822 | 812 | Check whether the given path is on a case-sensitive filesystem |
|
823 | 813 | |
|
824 | 814 | Requires a path (like /foo/.hg) ending with a foldable final |
|
825 | 815 | directory component. |
|
826 | 816 | """ |
|
827 | 817 | s1 = os.stat(path) |
|
828 | 818 | d, b = os.path.split(path) |
|
829 | 819 | p2 = os.path.join(d, b.upper()) |
|
830 | 820 | if path == p2: |
|
831 | 821 | p2 = os.path.join(d, b.lower()) |
|
832 | 822 | try: |
|
833 | 823 | s2 = os.stat(p2) |
|
834 | 824 | if s2 == s1: |
|
835 | 825 | return False |
|
836 | 826 | return True |
|
837 | 827 | except: |
|
838 | 828 | return True |
|
839 | 829 | |
|
840 | 830 | _fspathcache = {} |
|
841 | 831 | def fspath(name, root): |
|
842 | 832 | '''Get name in the case stored in the filesystem |
|
843 | 833 | |
|
844 | 834 | The name is either relative to root, or it is an absolute path starting |
|
845 | 835 | with root. Note that this function is unnecessary, and should not be |
|
846 | 836 | called, for case-sensitive filesystems (simply because it's expensive). |
|
847 | 837 | ''' |
|
848 | 838 | # If name is absolute, make it relative |
|
849 | 839 | if name.lower().startswith(root.lower()): |
|
850 | 840 | l = len(root) |
|
851 | 841 | if name[l] == os.sep or name[l] == os.altsep: |
|
852 | 842 | l = l + 1 |
|
853 | 843 | name = name[l:] |
|
854 | 844 | |
|
855 | 845 | if not os.path.exists(os.path.join(root, name)): |
|
856 | 846 | return None |
|
857 | 847 | |
|
858 | 848 | seps = os.sep |
|
859 | 849 | if os.altsep: |
|
860 | 850 | seps = seps + os.altsep |
|
861 | 851 | # Protect backslashes. This gets silly very quickly. |
|
862 | 852 | seps.replace('\\','\\\\') |
|
863 | 853 | pattern = re.compile(r'([^%s]+)|([%s]+)' % (seps, seps)) |
|
864 | 854 | dir = os.path.normcase(os.path.normpath(root)) |
|
865 | 855 | result = [] |
|
866 | 856 | for part, sep in pattern.findall(name): |
|
867 | 857 | if sep: |
|
868 | 858 | result.append(sep) |
|
869 | 859 | continue |
|
870 | 860 | |
|
871 | 861 | if dir not in _fspathcache: |
|
872 | 862 | _fspathcache[dir] = os.listdir(dir) |
|
873 | 863 | contents = _fspathcache[dir] |
|
874 | 864 | |
|
875 | 865 | lpart = part.lower() |
|
876 | 866 | for n in contents: |
|
877 | 867 | if n.lower() == lpart: |
|
878 | 868 | result.append(n) |
|
879 | 869 | break |
|
880 | 870 | else: |
|
881 | 871 | # Cannot happen, as the file exists! |
|
882 | 872 | result.append(part) |
|
883 | 873 | dir = os.path.join(dir, lpart) |
|
884 | 874 | |
|
885 | 875 | return ''.join(result) |
|
886 | 876 | |
|
887 | 877 | def checkexec(path): |
|
888 | 878 | """ |
|
889 | 879 | Check whether the given path is on a filesystem with UNIX-like exec flags |
|
890 | 880 | |
|
891 | 881 | Requires a directory (like /foo/.hg) |
|
892 | 882 | """ |
|
893 | 883 | |
|
894 | 884 | # VFAT on some Linux versions can flip mode but it doesn't persist |
|
895 | 885 | # a FS remount. Frequently we can detect it if files are created |
|
896 | 886 | # with exec bit on. |
|
897 | 887 | |
|
898 | 888 | try: |
|
899 | 889 | EXECFLAGS = stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH |
|
900 | 890 | fh, fn = tempfile.mkstemp("", "", path) |
|
901 | 891 | try: |
|
902 | 892 | os.close(fh) |
|
903 | 893 | m = os.stat(fn).st_mode & 0777 |
|
904 | 894 | new_file_has_exec = m & EXECFLAGS |
|
905 | 895 | os.chmod(fn, m ^ EXECFLAGS) |
|
906 | 896 | exec_flags_cannot_flip = ((os.stat(fn).st_mode & 0777) == m) |
|
907 | 897 | finally: |
|
908 | 898 | os.unlink(fn) |
|
909 | 899 | except (IOError, OSError): |
|
910 | 900 | # we don't care, the user probably won't be able to commit anyway |
|
911 | 901 | return False |
|
912 | 902 | return not (new_file_has_exec or exec_flags_cannot_flip) |
|
913 | 903 | |
|
914 | 904 | def checklink(path): |
|
915 | 905 | """check whether the given path is on a symlink-capable filesystem""" |
|
916 | 906 | # mktemp is not racy because symlink creation will fail if the |
|
917 | 907 | # file already exists |
|
918 | 908 | name = tempfile.mktemp(dir=path) |
|
919 | 909 | try: |
|
920 | 910 | os.symlink(".", name) |
|
921 | 911 | os.unlink(name) |
|
922 | 912 | return True |
|
923 | 913 | except (OSError, AttributeError): |
|
924 | 914 | return False |
|
925 | 915 | |
|
926 | 916 | def needbinarypatch(): |
|
927 | 917 | """return True if patches should be applied in binary mode by default.""" |
|
928 | 918 | return os.name == 'nt' |
|
929 | 919 | |
|
930 | 920 | def endswithsep(path): |
|
931 | 921 | '''Check path ends with os.sep or os.altsep.''' |
|
932 | 922 | return path.endswith(os.sep) or os.altsep and path.endswith(os.altsep) |
|
933 | 923 | |
|
934 | 924 | def splitpath(path): |
|
935 | 925 | '''Split path by os.sep. |
|
936 | 926 | Note that this function does not use os.altsep because this is |
|
937 | 927 | an alternative of simple "xxx.split(os.sep)". |
|
938 | 928 | It is recommended to use os.path.normpath() before using this |
|
939 | 929 | function if need.''' |
|
940 | 930 | return path.split(os.sep) |
|
941 | 931 | |
|
942 | 932 | def gui(): |
|
943 | 933 | '''Are we running in a GUI?''' |
|
944 | 934 | return os.name == "nt" or os.name == "mac" or os.environ.get("DISPLAY") |
|
945 | 935 | |
|
946 | 936 | def mktempcopy(name, emptyok=False, createmode=None): |
|
947 | 937 | """Create a temporary file with the same contents from name |
|
948 | 938 | |
|
949 | 939 | The permission bits are copied from the original file. |
|
950 | 940 | |
|
951 | 941 | If the temporary file is going to be truncated immediately, you |
|
952 | 942 | can use emptyok=True as an optimization. |
|
953 | 943 | |
|
954 | 944 | Returns the name of the temporary file. |
|
955 | 945 | """ |
|
956 | 946 | d, fn = os.path.split(name) |
|
957 | 947 | fd, temp = tempfile.mkstemp(prefix='.%s-' % fn, dir=d) |
|
958 | 948 | os.close(fd) |
|
959 | 949 | # Temporary files are created with mode 0600, which is usually not |
|
960 | 950 | # what we want. If the original file already exists, just copy |
|
961 | 951 | # its mode. Otherwise, manually obey umask. |
|
962 | 952 | try: |
|
963 | 953 | st_mode = os.lstat(name).st_mode & 0777 |
|
964 | 954 | except OSError, inst: |
|
965 | 955 | if inst.errno != errno.ENOENT: |
|
966 | 956 | raise |
|
967 | 957 | st_mode = createmode |
|
968 | 958 | if st_mode is None: |
|
969 | 959 | st_mode = ~umask |
|
970 | 960 | st_mode &= 0666 |
|
971 | 961 | os.chmod(temp, st_mode) |
|
972 | 962 | if emptyok: |
|
973 | 963 | return temp |
|
974 | 964 | try: |
|
975 | 965 | try: |
|
976 | 966 | ifp = posixfile(name, "rb") |
|
977 | 967 | except IOError, inst: |
|
978 | 968 | if inst.errno == errno.ENOENT: |
|
979 | 969 | return temp |
|
980 | 970 | if not getattr(inst, 'filename', None): |
|
981 | 971 | inst.filename = name |
|
982 | 972 | raise |
|
983 | 973 | ofp = posixfile(temp, "wb") |
|
984 | 974 | for chunk in filechunkiter(ifp): |
|
985 | 975 | ofp.write(chunk) |
|
986 | 976 | ifp.close() |
|
987 | 977 | ofp.close() |
|
988 | 978 | except: |
|
989 | 979 | try: os.unlink(temp) |
|
990 | 980 | except: pass |
|
991 | 981 | raise |
|
992 | 982 | return temp |
|
993 | 983 | |
|
994 | 984 | class atomictempfile(posixfile): |
|
995 | 985 | """file-like object that atomically updates a file |
|
996 | 986 | |
|
997 | 987 | All writes will be redirected to a temporary copy of the original |
|
998 | 988 | file. When rename is called, the copy is renamed to the original |
|
999 | 989 | name, making the changes visible. |
|
1000 | 990 | """ |
|
1001 | 991 | def __init__(self, name, mode, createmode): |
|
1002 | 992 | self.__name = name |
|
1003 | 993 | self.temp = mktempcopy(name, emptyok=('w' in mode), |
|
1004 | 994 | createmode=createmode) |
|
1005 | 995 | posixfile.__init__(self, self.temp, mode) |
|
1006 | 996 | |
|
1007 | 997 | def rename(self): |
|
1008 | 998 | if not self.closed: |
|
1009 | 999 | posixfile.close(self) |
|
1010 | 1000 | rename(self.temp, localpath(self.__name)) |
|
1011 | 1001 | |
|
1012 | 1002 | def __del__(self): |
|
1013 | 1003 | if not self.closed: |
|
1014 | 1004 | try: |
|
1015 | 1005 | os.unlink(self.temp) |
|
1016 | 1006 | except: pass |
|
1017 | 1007 | posixfile.close(self) |
|
1018 | 1008 | |
|
1019 | 1009 | def makedirs(name, mode=None): |
|
1020 | 1010 | """recursive directory creation with parent mode inheritance""" |
|
1021 | 1011 | try: |
|
1022 | 1012 | os.mkdir(name) |
|
1023 | 1013 | if mode is not None: |
|
1024 | 1014 | os.chmod(name, mode) |
|
1025 | 1015 | return |
|
1026 | 1016 | except OSError, err: |
|
1027 | 1017 | if err.errno == errno.EEXIST: |
|
1028 | 1018 | return |
|
1029 | 1019 | if err.errno != errno.ENOENT: |
|
1030 | 1020 | raise |
|
1031 | 1021 | parent = os.path.abspath(os.path.dirname(name)) |
|
1032 | 1022 | makedirs(parent, mode) |
|
1033 | 1023 | makedirs(name, mode) |
|
1034 | 1024 | |
|
1035 | 1025 | class opener(object): |
|
1036 | 1026 | """Open files relative to a base directory |
|
1037 | 1027 | |
|
1038 | 1028 | This class is used to hide the details of COW semantics and |
|
1039 | 1029 | remote file access from higher level code. |
|
1040 | 1030 | """ |
|
1041 | 1031 | def __init__(self, base, audit=True): |
|
1042 | 1032 | self.base = base |
|
1043 | 1033 | if audit: |
|
1044 | 1034 | self.audit_path = path_auditor(base) |
|
1045 | 1035 | else: |
|
1046 | 1036 | self.audit_path = always |
|
1047 | 1037 | self.createmode = None |
|
1048 | 1038 | |
|
1049 | 1039 | def __getattr__(self, name): |
|
1050 | 1040 | if name == '_can_symlink': |
|
1051 | 1041 | self._can_symlink = checklink(self.base) |
|
1052 | 1042 | return self._can_symlink |
|
1053 | 1043 | raise AttributeError(name) |
|
1054 | 1044 | |
|
1055 | 1045 | def _fixfilemode(self, name): |
|
1056 | 1046 | if self.createmode is None: |
|
1057 | 1047 | return |
|
1058 | 1048 | os.chmod(name, self.createmode & 0666) |
|
1059 | 1049 | |
|
1060 | 1050 | def __call__(self, path, mode="r", text=False, atomictemp=False): |
|
1061 | 1051 | self.audit_path(path) |
|
1062 | 1052 | f = os.path.join(self.base, path) |
|
1063 | 1053 | |
|
1064 | 1054 | if not text and "b" not in mode: |
|
1065 | 1055 | mode += "b" # for that other OS |
|
1066 | 1056 | |
|
1067 | 1057 | nlink = -1 |
|
1068 | 1058 | if mode not in ("r", "rb"): |
|
1069 | 1059 | try: |
|
1070 | 1060 | nlink = nlinks(f) |
|
1071 | 1061 | except OSError: |
|
1072 | 1062 | nlink = 0 |
|
1073 | 1063 | d = os.path.dirname(f) |
|
1074 | 1064 | if not os.path.isdir(d): |
|
1075 | 1065 | makedirs(d, self.createmode) |
|
1076 | 1066 | if atomictemp: |
|
1077 | 1067 | return atomictempfile(f, mode, self.createmode) |
|
1078 | 1068 | if nlink > 1: |
|
1079 | 1069 | rename(mktempcopy(f), f) |
|
1080 | 1070 | fp = posixfile(f, mode) |
|
1081 | 1071 | if nlink == 0: |
|
1082 | 1072 | self._fixfilemode(f) |
|
1083 | 1073 | return fp |
|
1084 | 1074 | |
|
1085 | 1075 | def symlink(self, src, dst): |
|
1086 | 1076 | self.audit_path(dst) |
|
1087 | 1077 | linkname = os.path.join(self.base, dst) |
|
1088 | 1078 | try: |
|
1089 | 1079 | os.unlink(linkname) |
|
1090 | 1080 | except OSError: |
|
1091 | 1081 | pass |
|
1092 | 1082 | |
|
1093 | 1083 | dirname = os.path.dirname(linkname) |
|
1094 | 1084 | if not os.path.exists(dirname): |
|
1095 | 1085 | makedirs(dirname, self.createmode) |
|
1096 | 1086 | |
|
1097 | 1087 | if self._can_symlink: |
|
1098 | 1088 | try: |
|
1099 | 1089 | os.symlink(src, linkname) |
|
1100 | 1090 | except OSError, err: |
|
1101 | 1091 | raise OSError(err.errno, _('could not symlink to %r: %s') % |
|
1102 | 1092 | (src, err.strerror), linkname) |
|
1103 | 1093 | else: |
|
1104 | 1094 | f = self(dst, "w") |
|
1105 | 1095 | f.write(src) |
|
1106 | 1096 | f.close() |
|
1107 | 1097 | self._fixfilemode(dst) |
|
1108 | 1098 | |
|
1109 | 1099 | class chunkbuffer(object): |
|
1110 | 1100 | """Allow arbitrary sized chunks of data to be efficiently read from an |
|
1111 | 1101 | iterator over chunks of arbitrary size.""" |
|
1112 | 1102 | |
|
1113 | 1103 | def __init__(self, in_iter): |
|
1114 | 1104 | """in_iter is the iterator that's iterating over the input chunks. |
|
1115 | 1105 | targetsize is how big a buffer to try to maintain.""" |
|
1116 | 1106 | self.iter = iter(in_iter) |
|
1117 | 1107 | self.buf = '' |
|
1118 | 1108 | self.targetsize = 2**16 |
|
1119 | 1109 | |
|
1120 | 1110 | def read(self, l): |
|
1121 | 1111 | """Read L bytes of data from the iterator of chunks of data. |
|
1122 | 1112 | Returns less than L bytes if the iterator runs dry.""" |
|
1123 | 1113 | if l > len(self.buf) and self.iter: |
|
1124 | 1114 | # Clamp to a multiple of self.targetsize |
|
1125 | 1115 | targetsize = max(l, self.targetsize) |
|
1126 | 1116 | collector = cStringIO.StringIO() |
|
1127 | 1117 | collector.write(self.buf) |
|
1128 | 1118 | collected = len(self.buf) |
|
1129 | 1119 | for chunk in self.iter: |
|
1130 | 1120 | collector.write(chunk) |
|
1131 | 1121 | collected += len(chunk) |
|
1132 | 1122 | if collected >= targetsize: |
|
1133 | 1123 | break |
|
1134 | 1124 | if collected < targetsize: |
|
1135 | 1125 | self.iter = False |
|
1136 | 1126 | self.buf = collector.getvalue() |
|
1137 | 1127 | if len(self.buf) == l: |
|
1138 | 1128 | s, self.buf = str(self.buf), '' |
|
1139 | 1129 | else: |
|
1140 | 1130 | s, self.buf = self.buf[:l], buffer(self.buf, l) |
|
1141 | 1131 | return s |
|
1142 | 1132 | |
|
1143 | 1133 | def filechunkiter(f, size=65536, limit=None): |
|
1144 | 1134 | """Create a generator that produces the data in the file size |
|
1145 | 1135 | (default 65536) bytes at a time, up to optional limit (default is |
|
1146 | 1136 | to read all data). Chunks may be less than size bytes if the |
|
1147 | 1137 | chunk is the last chunk in the file, or the file is a socket or |
|
1148 | 1138 | some other type of file that sometimes reads less data than is |
|
1149 | 1139 | requested.""" |
|
1150 | 1140 | assert size >= 0 |
|
1151 | 1141 | assert limit is None or limit >= 0 |
|
1152 | 1142 | while True: |
|
1153 | 1143 | if limit is None: nbytes = size |
|
1154 | 1144 | else: nbytes = min(limit, size) |
|
1155 | 1145 | s = nbytes and f.read(nbytes) |
|
1156 | 1146 | if not s: break |
|
1157 | 1147 | if limit: limit -= len(s) |
|
1158 | 1148 | yield s |
|
1159 | 1149 | |
|
1160 | 1150 | def makedate(): |
|
1161 | 1151 | lt = time.localtime() |
|
1162 | 1152 | if lt[8] == 1 and time.daylight: |
|
1163 | 1153 | tz = time.altzone |
|
1164 | 1154 | else: |
|
1165 | 1155 | tz = time.timezone |
|
1166 | 1156 | return time.mktime(lt), tz |
|
1167 | 1157 | |
|
1168 | 1158 | def datestr(date=None, format='%a %b %d %H:%M:%S %Y %1%2'): |
|
1169 | 1159 | """represent a (unixtime, offset) tuple as a localized time. |
|
1170 | 1160 | unixtime is seconds since the epoch, and offset is the time zone's |
|
1171 | 1161 | number of seconds away from UTC. if timezone is false, do not |
|
1172 | 1162 | append time zone to string.""" |
|
1173 | 1163 | t, tz = date or makedate() |
|
1174 | 1164 | if "%1" in format or "%2" in format: |
|
1175 | 1165 | sign = (tz > 0) and "-" or "+" |
|
1176 | 1166 | minutes = abs(tz) / 60 |
|
1177 | 1167 | format = format.replace("%1", "%c%02d" % (sign, minutes / 60)) |
|
1178 | 1168 | format = format.replace("%2", "%02d" % (minutes % 60)) |
|
1179 | 1169 | s = time.strftime(format, time.gmtime(float(t) - tz)) |
|
1180 | 1170 | return s |
|
1181 | 1171 | |
|
1182 | 1172 | def shortdate(date=None): |
|
1183 | 1173 | """turn (timestamp, tzoff) tuple into iso 8631 date.""" |
|
1184 | 1174 | return datestr(date, format='%Y-%m-%d') |
|
1185 | 1175 | |
|
1186 | 1176 | def strdate(string, format, defaults=[]): |
|
1187 | 1177 | """parse a localized time string and return a (unixtime, offset) tuple. |
|
1188 | 1178 | if the string cannot be parsed, ValueError is raised.""" |
|
1189 | 1179 | def timezone(string): |
|
1190 | 1180 | tz = string.split()[-1] |
|
1191 | 1181 | if tz[0] in "+-" and len(tz) == 5 and tz[1:].isdigit(): |
|
1192 | 1182 | sign = (tz[0] == "+") and 1 or -1 |
|
1193 | 1183 | hours = int(tz[1:3]) |
|
1194 | 1184 | minutes = int(tz[3:5]) |
|
1195 | 1185 | return -sign * (hours * 60 + minutes) * 60 |
|
1196 | 1186 | if tz == "GMT" or tz == "UTC": |
|
1197 | 1187 | return 0 |
|
1198 | 1188 | return None |
|
1199 | 1189 | |
|
1200 | 1190 | # NOTE: unixtime = localunixtime + offset |
|
1201 | 1191 | offset, date = timezone(string), string |
|
1202 | 1192 | if offset != None: |
|
1203 | 1193 | date = " ".join(string.split()[:-1]) |
|
1204 | 1194 | |
|
1205 | 1195 | # add missing elements from defaults |
|
1206 | 1196 | for part in defaults: |
|
1207 | 1197 | found = [True for p in part if ("%"+p) in format] |
|
1208 | 1198 | if not found: |
|
1209 | 1199 | date += "@" + defaults[part] |
|
1210 | 1200 | format += "@%" + part[0] |
|
1211 | 1201 | |
|
1212 | 1202 | timetuple = time.strptime(date, format) |
|
1213 | 1203 | localunixtime = int(calendar.timegm(timetuple)) |
|
1214 | 1204 | if offset is None: |
|
1215 | 1205 | # local timezone |
|
1216 | 1206 | unixtime = int(time.mktime(timetuple)) |
|
1217 | 1207 | offset = unixtime - localunixtime |
|
1218 | 1208 | else: |
|
1219 | 1209 | unixtime = localunixtime + offset |
|
1220 | 1210 | return unixtime, offset |
|
1221 | 1211 | |
|
1222 | 1212 | def parsedate(date, formats=None, defaults=None): |
|
1223 | 1213 | """parse a localized date/time string and return a (unixtime, offset) tuple. |
|
1224 | 1214 | |
|
1225 | 1215 | The date may be a "unixtime offset" string or in one of the specified |
|
1226 | 1216 | formats. If the date already is a (unixtime, offset) tuple, it is returned. |
|
1227 | 1217 | """ |
|
1228 | 1218 | if not date: |
|
1229 | 1219 | return 0, 0 |
|
1230 | 1220 | if isinstance(date, tuple) and len(date) == 2: |
|
1231 | 1221 | return date |
|
1232 | 1222 | if not formats: |
|
1233 | 1223 | formats = defaultdateformats |
|
1234 | 1224 | date = date.strip() |
|
1235 | 1225 | try: |
|
1236 | 1226 | when, offset = map(int, date.split(' ')) |
|
1237 | 1227 | except ValueError: |
|
1238 | 1228 | # fill out defaults |
|
1239 | 1229 | if not defaults: |
|
1240 | 1230 | defaults = {} |
|
1241 | 1231 | now = makedate() |
|
1242 | 1232 | for part in "d mb yY HI M S".split(): |
|
1243 | 1233 | if part not in defaults: |
|
1244 | 1234 | if part[0] in "HMS": |
|
1245 | 1235 | defaults[part] = "00" |
|
1246 | 1236 | else: |
|
1247 | 1237 | defaults[part] = datestr(now, "%" + part[0]) |
|
1248 | 1238 | |
|
1249 | 1239 | for format in formats: |
|
1250 | 1240 | try: |
|
1251 | 1241 | when, offset = strdate(date, format, defaults) |
|
1252 | 1242 | except (ValueError, OverflowError): |
|
1253 | 1243 | pass |
|
1254 | 1244 | else: |
|
1255 | 1245 | break |
|
1256 | 1246 | else: |
|
1257 | 1247 | raise Abort(_('invalid date: %r ') % date) |
|
1258 | 1248 | # validate explicit (probably user-specified) date and |
|
1259 | 1249 | # time zone offset. values must fit in signed 32 bits for |
|
1260 | 1250 | # current 32-bit linux runtimes. timezones go from UTC-12 |
|
1261 | 1251 | # to UTC+14 |
|
1262 | 1252 | if abs(when) > 0x7fffffff: |
|
1263 | 1253 | raise Abort(_('date exceeds 32 bits: %d') % when) |
|
1264 | 1254 | if offset < -50400 or offset > 43200: |
|
1265 | 1255 | raise Abort(_('impossible time zone offset: %d') % offset) |
|
1266 | 1256 | return when, offset |
|
1267 | 1257 | |
|
1268 | 1258 | def matchdate(date): |
|
1269 | 1259 | """Return a function that matches a given date match specifier |
|
1270 | 1260 | |
|
1271 | 1261 | Formats include: |
|
1272 | 1262 | |
|
1273 | 1263 | '{date}' match a given date to the accuracy provided |
|
1274 | 1264 | |
|
1275 | 1265 | '<{date}' on or before a given date |
|
1276 | 1266 | |
|
1277 | 1267 | '>{date}' on or after a given date |
|
1278 | 1268 | |
|
1279 | 1269 | """ |
|
1280 | 1270 | |
|
1281 | 1271 | def lower(date): |
|
1282 | 1272 | d = dict(mb="1", d="1") |
|
1283 | 1273 | return parsedate(date, extendeddateformats, d)[0] |
|
1284 | 1274 | |
|
1285 | 1275 | def upper(date): |
|
1286 | 1276 | d = dict(mb="12", HI="23", M="59", S="59") |
|
1287 | 1277 | for days in "31 30 29".split(): |
|
1288 | 1278 | try: |
|
1289 | 1279 | d["d"] = days |
|
1290 | 1280 | return parsedate(date, extendeddateformats, d)[0] |
|
1291 | 1281 | except: |
|
1292 | 1282 | pass |
|
1293 | 1283 | d["d"] = "28" |
|
1294 | 1284 | return parsedate(date, extendeddateformats, d)[0] |
|
1295 | 1285 | |
|
1296 | 1286 | date = date.strip() |
|
1297 | 1287 | if date[0] == "<": |
|
1298 | 1288 | when = upper(date[1:]) |
|
1299 | 1289 | return lambda x: x <= when |
|
1300 | 1290 | elif date[0] == ">": |
|
1301 | 1291 | when = lower(date[1:]) |
|
1302 | 1292 | return lambda x: x >= when |
|
1303 | 1293 | elif date[0] == "-": |
|
1304 | 1294 | try: |
|
1305 | 1295 | days = int(date[1:]) |
|
1306 | 1296 | except ValueError: |
|
1307 | 1297 | raise Abort(_("invalid day spec: %s") % date[1:]) |
|
1308 | 1298 | when = makedate()[0] - days * 3600 * 24 |
|
1309 | 1299 | return lambda x: x >= when |
|
1310 | 1300 | elif " to " in date: |
|
1311 | 1301 | a, b = date.split(" to ") |
|
1312 | 1302 | start, stop = lower(a), upper(b) |
|
1313 | 1303 | return lambda x: x >= start and x <= stop |
|
1314 | 1304 | else: |
|
1315 | 1305 | start, stop = lower(date), upper(date) |
|
1316 | 1306 | return lambda x: x >= start and x <= stop |
|
1317 | 1307 | |
|
1318 | 1308 | def shortuser(user): |
|
1319 | 1309 | """Return a short representation of a user name or email address.""" |
|
1320 | 1310 | f = user.find('@') |
|
1321 | 1311 | if f >= 0: |
|
1322 | 1312 | user = user[:f] |
|
1323 | 1313 | f = user.find('<') |
|
1324 | 1314 | if f >= 0: |
|
1325 | 1315 | user = user[f+1:] |
|
1326 | 1316 | f = user.find(' ') |
|
1327 | 1317 | if f >= 0: |
|
1328 | 1318 | user = user[:f] |
|
1329 | 1319 | f = user.find('.') |
|
1330 | 1320 | if f >= 0: |
|
1331 | 1321 | user = user[:f] |
|
1332 | 1322 | return user |
|
1333 | 1323 | |
|
1334 | 1324 | def email(author): |
|
1335 | 1325 | '''get email of author.''' |
|
1336 | 1326 | r = author.find('>') |
|
1337 | 1327 | if r == -1: r = None |
|
1338 | 1328 | return author[author.find('<')+1:r] |
|
1339 | 1329 | |
|
1340 | 1330 | def ellipsis(text, maxlength=400): |
|
1341 | 1331 | """Trim string to at most maxlength (default: 400) characters.""" |
|
1342 | 1332 | if len(text) <= maxlength: |
|
1343 | 1333 | return text |
|
1344 | 1334 | else: |
|
1345 | 1335 | return "%s..." % (text[:maxlength-3]) |
|
1346 | 1336 | |
|
1347 | 1337 | def walkrepos(path, followsym=False, seen_dirs=None, recurse=False): |
|
1348 | 1338 | '''yield every hg repository under path, recursively.''' |
|
1349 | 1339 | def errhandler(err): |
|
1350 | 1340 | if err.filename == path: |
|
1351 | 1341 | raise err |
|
1352 | 1342 | if followsym and hasattr(os.path, 'samestat'): |
|
1353 | 1343 | def _add_dir_if_not_there(dirlst, dirname): |
|
1354 | 1344 | match = False |
|
1355 | 1345 | samestat = os.path.samestat |
|
1356 | 1346 | dirstat = os.stat(dirname) |
|
1357 | 1347 | for lstdirstat in dirlst: |
|
1358 | 1348 | if samestat(dirstat, lstdirstat): |
|
1359 | 1349 | match = True |
|
1360 | 1350 | break |
|
1361 | 1351 | if not match: |
|
1362 | 1352 | dirlst.append(dirstat) |
|
1363 | 1353 | return not match |
|
1364 | 1354 | else: |
|
1365 | 1355 | followsym = False |
|
1366 | 1356 | |
|
1367 | 1357 | if (seen_dirs is None) and followsym: |
|
1368 | 1358 | seen_dirs = [] |
|
1369 | 1359 | _add_dir_if_not_there(seen_dirs, path) |
|
1370 | 1360 | for root, dirs, files in os.walk(path, topdown=True, onerror=errhandler): |
|
1371 | 1361 | if '.hg' in dirs: |
|
1372 | 1362 | yield root # found a repository |
|
1373 | 1363 | qroot = os.path.join(root, '.hg', 'patches') |
|
1374 | 1364 | if os.path.isdir(os.path.join(qroot, '.hg')): |
|
1375 | 1365 | yield qroot # we have a patch queue repo here |
|
1376 | 1366 | if recurse: |
|
1377 | 1367 | # avoid recursing inside the .hg directory |
|
1378 | 1368 | dirs.remove('.hg') |
|
1379 | 1369 | else: |
|
1380 | 1370 | dirs[:] = [] # don't descend further |
|
1381 | 1371 | elif followsym: |
|
1382 | 1372 | newdirs = [] |
|
1383 | 1373 | for d in dirs: |
|
1384 | 1374 | fname = os.path.join(root, d) |
|
1385 | 1375 | if _add_dir_if_not_there(seen_dirs, fname): |
|
1386 | 1376 | if os.path.islink(fname): |
|
1387 | 1377 | for hgname in walkrepos(fname, True, seen_dirs): |
|
1388 | 1378 | yield hgname |
|
1389 | 1379 | else: |
|
1390 | 1380 | newdirs.append(d) |
|
1391 | 1381 | dirs[:] = newdirs |
|
1392 | 1382 | |
|
1393 | 1383 | _rcpath = None |
|
1394 | 1384 | |
|
1395 | 1385 | def os_rcpath(): |
|
1396 | 1386 | '''return default os-specific hgrc search path''' |
|
1397 | 1387 | path = system_rcpath() |
|
1398 | 1388 | path.extend(user_rcpath()) |
|
1399 | 1389 | path = [os.path.normpath(f) for f in path] |
|
1400 | 1390 | return path |
|
1401 | 1391 | |
|
1402 | 1392 | def rcpath(): |
|
1403 | 1393 | '''return hgrc search path. if env var HGRCPATH is set, use it. |
|
1404 | 1394 | for each item in path, if directory, use files ending in .rc, |
|
1405 | 1395 | else use item. |
|
1406 | 1396 | make HGRCPATH empty to only look in .hg/hgrc of current repo. |
|
1407 | 1397 | if no HGRCPATH, use default os-specific path.''' |
|
1408 | 1398 | global _rcpath |
|
1409 | 1399 | if _rcpath is None: |
|
1410 | 1400 | if 'HGRCPATH' in os.environ: |
|
1411 | 1401 | _rcpath = [] |
|
1412 | 1402 | for p in os.environ['HGRCPATH'].split(os.pathsep): |
|
1413 | 1403 | if not p: continue |
|
1414 | 1404 | if os.path.isdir(p): |
|
1415 | 1405 | for f, kind in osutil.listdir(p): |
|
1416 | 1406 | if f.endswith('.rc'): |
|
1417 | 1407 | _rcpath.append(os.path.join(p, f)) |
|
1418 | 1408 | else: |
|
1419 | 1409 | _rcpath.append(p) |
|
1420 | 1410 | else: |
|
1421 | 1411 | _rcpath = os_rcpath() |
|
1422 | 1412 | return _rcpath |
|
1423 | 1413 | |
|
1424 | 1414 | def bytecount(nbytes): |
|
1425 | 1415 | '''return byte count formatted as readable string, with units''' |
|
1426 | 1416 | |
|
1427 | 1417 | units = ( |
|
1428 | 1418 | (100, 1<<30, _('%.0f GB')), |
|
1429 | 1419 | (10, 1<<30, _('%.1f GB')), |
|
1430 | 1420 | (1, 1<<30, _('%.2f GB')), |
|
1431 | 1421 | (100, 1<<20, _('%.0f MB')), |
|
1432 | 1422 | (10, 1<<20, _('%.1f MB')), |
|
1433 | 1423 | (1, 1<<20, _('%.2f MB')), |
|
1434 | 1424 | (100, 1<<10, _('%.0f KB')), |
|
1435 | 1425 | (10, 1<<10, _('%.1f KB')), |
|
1436 | 1426 | (1, 1<<10, _('%.2f KB')), |
|
1437 | 1427 | (1, 1, _('%.0f bytes')), |
|
1438 | 1428 | ) |
|
1439 | 1429 | |
|
1440 | 1430 | for multiplier, divisor, format in units: |
|
1441 | 1431 | if nbytes >= divisor * multiplier: |
|
1442 | 1432 | return format % (nbytes / float(divisor)) |
|
1443 | 1433 | return units[-1][2] % nbytes |
|
1444 | 1434 | |
|
1445 | 1435 | def drop_scheme(scheme, path): |
|
1446 | 1436 | sc = scheme + ':' |
|
1447 | 1437 | if path.startswith(sc): |
|
1448 | 1438 | path = path[len(sc):] |
|
1449 | 1439 | if path.startswith('//'): |
|
1450 | 1440 | path = path[2:] |
|
1451 | 1441 | return path |
|
1452 | 1442 | |
|
1453 | 1443 | def uirepr(s): |
|
1454 | 1444 | # Avoid double backslash in Windows path repr() |
|
1455 | 1445 | return repr(s).replace('\\\\', '\\') |
|
1456 | 1446 | |
|
1457 | 1447 | def termwidth(): |
|
1458 | 1448 | if 'COLUMNS' in os.environ: |
|
1459 | 1449 | try: |
|
1460 | 1450 | return int(os.environ['COLUMNS']) |
|
1461 | 1451 | except ValueError: |
|
1462 | 1452 | pass |
|
1463 | 1453 | try: |
|
1464 | 1454 | import termios, array, fcntl |
|
1465 | 1455 | for dev in (sys.stdout, sys.stdin): |
|
1466 | 1456 | try: |
|
1467 | 1457 | fd = dev.fileno() |
|
1468 | 1458 | if not os.isatty(fd): |
|
1469 | 1459 | continue |
|
1470 | 1460 | arri = fcntl.ioctl(fd, termios.TIOCGWINSZ, '\0' * 8) |
|
1471 | 1461 | return array.array('h', arri)[1] |
|
1472 | 1462 | except ValueError: |
|
1473 | 1463 | pass |
|
1474 | 1464 | except ImportError: |
|
1475 | 1465 | pass |
|
1476 | 1466 | return 80 |
|
1477 | 1467 | |
|
1478 | 1468 | def iterlines(iterator): |
|
1479 | 1469 | for chunk in iterator: |
|
1480 | 1470 | for line in chunk.splitlines(): |
|
1481 | 1471 | yield line |
@@ -1,118 +1,121 b'' | |||
|
1 | 1 | #!/bin/sh |
|
2 | 2 | |
|
3 | 3 | mkdir test |
|
4 | 4 | cd test |
|
5 | 5 | hg init |
|
6 | 6 | echo foo>foo |
|
7 | 7 | hg commit -Am 1 -d '1 0' |
|
8 | 8 | echo bar>bar |
|
9 | 9 | hg commit -Am 2 -d '2 0' |
|
10 | 10 | mkdir baz |
|
11 | 11 | echo bletch>baz/bletch |
|
12 | 12 | hg commit -Am 3 -d '1000000000 0' |
|
13 | 13 | echo "[web]" >> .hg/hgrc |
|
14 | 14 | echo "name = test-archive" >> .hg/hgrc |
|
15 | 15 | cp .hg/hgrc .hg/hgrc-base |
|
16 | 16 | |
|
17 | 17 | # check http return codes |
|
18 | 18 | test_archtype() { |
|
19 | 19 | echo "allow_archive = $1" >> .hg/hgrc |
|
20 | 20 | hg serve -p $HGPORT -d --pid-file=hg.pid -E errors.log |
|
21 | 21 | cat hg.pid >> $DAEMON_PIDS |
|
22 | 22 | echo % $1 allowed should give 200 |
|
23 | 23 | "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.$2" | head -n 1 |
|
24 | 24 | echo % $3 and $4 disallowed should both give 403 |
|
25 | 25 | "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.$3" | head -n 1 |
|
26 | 26 | "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.$4" | head -n 1 |
|
27 | 27 | "$TESTDIR/killdaemons.py" |
|
28 | 28 | cat errors.log |
|
29 | 29 | cp .hg/hgrc-base .hg/hgrc |
|
30 | 30 | } |
|
31 | 31 | |
|
32 | 32 | echo |
|
33 | 33 | test_archtype gz tar.gz tar.bz2 zip |
|
34 | 34 | test_archtype bz2 tar.bz2 zip tar.gz |
|
35 | 35 | test_archtype zip zip tar.gz tar.bz2 |
|
36 | 36 | |
|
37 | 37 | echo "allow_archive = gz bz2 zip" >> .hg/hgrc |
|
38 | 38 | hg serve -p $HGPORT -d --pid-file=hg.pid -E errors.log |
|
39 | 39 | cat hg.pid >> $DAEMON_PIDS |
|
40 | 40 | |
|
41 | 41 | echo % invalid arch type should give 404 |
|
42 | 42 | "$TESTDIR/get-with-headers.py" localhost:$HGPORT "/archive/tip.invalid" | head -n 1 |
|
43 | 43 | echo |
|
44 | 44 | |
|
45 | 45 | TIP=`hg id -v | cut -f1 -d' '` |
|
46 | 46 | QTIP=`hg id -q` |
|
47 | 47 | cat > getarchive.py <<EOF |
|
48 | 48 | import os, sys, urllib2 |
|
49 | 49 | try: |
|
50 | 50 | # Set stdout to binary mode for win32 platforms |
|
51 | 51 | import msvcrt |
|
52 | 52 | msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY) |
|
53 | 53 | except ImportError: |
|
54 | 54 | pass |
|
55 | 55 | |
|
56 | 56 | node, archive = sys.argv[1:] |
|
57 | 57 | f = urllib2.urlopen('http://127.0.0.1:%s/?cmd=archive;node=%s;type=%s' |
|
58 | 58 | % (os.environ['HGPORT'], node, archive)) |
|
59 | 59 | sys.stdout.write(f.read()) |
|
60 | 60 | EOF |
|
61 | 61 | http_proxy= python getarchive.py "$TIP" gz | gunzip | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/" |
|
62 | 62 | http_proxy= python getarchive.py "$TIP" bz2 | bunzip2 | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/" |
|
63 | 63 | http_proxy= python getarchive.py "$TIP" zip > archive.zip |
|
64 | 64 | unzip -t archive.zip | sed "s/$QTIP/TIP/" |
|
65 | 65 | |
|
66 | 66 | "$TESTDIR/killdaemons.py" |
|
67 | 67 | |
|
68 | 68 | hg archive -t tar test.tar |
|
69 | 69 | tar tf test.tar |
|
70 | 70 | |
|
71 | 71 | hg archive -t tbz2 -X baz test.tar.bz2 |
|
72 | 72 | bunzip2 -dc test.tar.bz2 | tar tf - 2>/dev/null |
|
73 | 73 | |
|
74 | 74 | hg archive -t tgz -p %b-%h test-%h.tar.gz |
|
75 | 75 | gzip -dc test-$QTIP.tar.gz | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/" |
|
76 | 76 | |
|
77 | 77 | cat > md5comp.py <<EOF |
|
78 | from mercurial.util import md5 | |
|
78 | try: | |
|
79 | from hashlib import md5 | |
|
80 | except ImportError: | |
|
81 | from md5 import md5 | |
|
79 | 82 | import sys |
|
80 | 83 | f1, f2 = sys.argv[1:3] |
|
81 | 84 | h1 = md5(file(f1, 'rb').read()).hexdigest() |
|
82 | 85 | h2 = md5(file(f2, 'rb').read()).hexdigest() |
|
83 | 86 | print h1 == h2 or "md5 differ: " + repr((h1, h2)) |
|
84 | 87 | EOF |
|
85 | 88 | |
|
86 | 89 | # archive name is stored in the archive, so create similar |
|
87 | 90 | # archives and rename them afterwards. |
|
88 | 91 | hg archive -t tgz tip.tar.gz |
|
89 | 92 | mv tip.tar.gz tip1.tar.gz |
|
90 | 93 | sleep 1 |
|
91 | 94 | hg archive -t tgz tip.tar.gz |
|
92 | 95 | mv tip.tar.gz tip2.tar.gz |
|
93 | 96 | python md5comp.py tip1.tar.gz tip2.tar.gz |
|
94 | 97 | |
|
95 | 98 | hg archive -t zip -p /illegal test.zip |
|
96 | 99 | hg archive -t zip -p very/../bad test.zip |
|
97 | 100 | |
|
98 | 101 | hg archive --config ui.archivemeta=false -t zip -r 2 test.zip |
|
99 | 102 | unzip -t test.zip |
|
100 | 103 | |
|
101 | 104 | hg archive -t tar - | tar tf - 2>/dev/null | sed "s/$QTIP/TIP/" |
|
102 | 105 | |
|
103 | 106 | hg archive -r 0 -t tar rev-%r.tar |
|
104 | 107 | if [ -f rev-0.tar ]; then |
|
105 | 108 | echo 'rev-0.tar created' |
|
106 | 109 | fi |
|
107 | 110 | |
|
108 | 111 | hg archive -t bogus test.bogus |
|
109 | 112 | |
|
110 | 113 | echo % server errors |
|
111 | 114 | cat errors.log |
|
112 | 115 | |
|
113 | 116 | echo '% empty repo' |
|
114 | 117 | hg init ../empty |
|
115 | 118 | cd ../empty |
|
116 | 119 | hg archive ../test-empty |
|
117 | 120 | |
|
118 | 121 | exit 0 |
General Comments 0
You need to be logged in to leave comments.
Login now