##// END OF EJS Templates
changegroup*(): use enumerate when possible
Benoit Boissinot -
r11665:1f7f8bc4 default
parent child Browse files
Show More
@@ -1,1806 +1,1799 b''
1 1 # localrepo.py - read/write repository class for mercurial
2 2 #
3 3 # Copyright 2005-2007 Matt Mackall <mpm@selenic.com>
4 4 #
5 5 # This software may be used and distributed according to the terms of the
6 6 # GNU General Public License version 2 or any later version.
7 7
8 8 from node import bin, hex, nullid, nullrev, short
9 9 from i18n import _
10 10 import repo, changegroup, subrepo, discovery, pushkey
11 11 import changelog, dirstate, filelog, manifest, context
12 12 import lock, transaction, store, encoding
13 13 import util, extensions, hook, error
14 14 import match as matchmod
15 15 import merge as mergemod
16 16 import tags as tagsmod
17 17 import url as urlmod
18 18 from lock import release
19 19 import weakref, errno, os, time, inspect
20 20 propertycache = util.propertycache
21 21
22 22 class localrepository(repo.repository):
23 23 capabilities = set(('lookup', 'changegroupsubset', 'branchmap', 'pushkey'))
24 24 supported = set('revlogv1 store fncache shared'.split())
25 25
26 26 def __init__(self, baseui, path=None, create=0):
27 27 repo.repository.__init__(self)
28 28 self.root = os.path.realpath(util.expandpath(path))
29 29 self.path = os.path.join(self.root, ".hg")
30 30 self.origroot = path
31 31 self.opener = util.opener(self.path)
32 32 self.wopener = util.opener(self.root)
33 33 self.baseui = baseui
34 34 self.ui = baseui.copy()
35 35
36 36 try:
37 37 self.ui.readconfig(self.join("hgrc"), self.root)
38 38 extensions.loadall(self.ui)
39 39 except IOError:
40 40 pass
41 41
42 42 if not os.path.isdir(self.path):
43 43 if create:
44 44 if not os.path.exists(path):
45 45 util.makedirs(path)
46 46 os.mkdir(self.path)
47 47 requirements = ["revlogv1"]
48 48 if self.ui.configbool('format', 'usestore', True):
49 49 os.mkdir(os.path.join(self.path, "store"))
50 50 requirements.append("store")
51 51 if self.ui.configbool('format', 'usefncache', True):
52 52 requirements.append("fncache")
53 53 # create an invalid changelog
54 54 self.opener("00changelog.i", "a").write(
55 55 '\0\0\0\2' # represents revlogv2
56 56 ' dummy changelog to prevent using the old repo layout'
57 57 )
58 58 reqfile = self.opener("requires", "w")
59 59 for r in requirements:
60 60 reqfile.write("%s\n" % r)
61 61 reqfile.close()
62 62 else:
63 63 raise error.RepoError(_("repository %s not found") % path)
64 64 elif create:
65 65 raise error.RepoError(_("repository %s already exists") % path)
66 66 else:
67 67 # find requirements
68 68 requirements = set()
69 69 try:
70 70 requirements = set(self.opener("requires").read().splitlines())
71 71 except IOError, inst:
72 72 if inst.errno != errno.ENOENT:
73 73 raise
74 74 for r in requirements - self.supported:
75 75 raise error.RepoError(_("requirement '%s' not supported") % r)
76 76
77 77 self.sharedpath = self.path
78 78 try:
79 79 s = os.path.realpath(self.opener("sharedpath").read())
80 80 if not os.path.exists(s):
81 81 raise error.RepoError(
82 82 _('.hg/sharedpath points to nonexistent directory %s') % s)
83 83 self.sharedpath = s
84 84 except IOError, inst:
85 85 if inst.errno != errno.ENOENT:
86 86 raise
87 87
88 88 self.store = store.store(requirements, self.sharedpath, util.opener)
89 89 self.spath = self.store.path
90 90 self.sopener = self.store.opener
91 91 self.sjoin = self.store.join
92 92 self.opener.createmode = self.store.createmode
93 93 self.sopener.options = {}
94 94
95 95 # These two define the set of tags for this repository. _tags
96 96 # maps tag name to node; _tagtypes maps tag name to 'global' or
97 97 # 'local'. (Global tags are defined by .hgtags across all
98 98 # heads, and local tags are defined in .hg/localtags.) They
99 99 # constitute the in-memory cache of tags.
100 100 self._tags = None
101 101 self._tagtypes = None
102 102
103 103 self._branchcache = None # in UTF-8
104 104 self._branchcachetip = None
105 105 self.nodetagscache = None
106 106 self.filterpats = {}
107 107 self._datafilters = {}
108 108 self._transref = self._lockref = self._wlockref = None
109 109
110 110 @propertycache
111 111 def changelog(self):
112 112 c = changelog.changelog(self.sopener)
113 113 if 'HG_PENDING' in os.environ:
114 114 p = os.environ['HG_PENDING']
115 115 if p.startswith(self.root):
116 116 c.readpending('00changelog.i.a')
117 117 self.sopener.options['defversion'] = c.version
118 118 return c
119 119
120 120 @propertycache
121 121 def manifest(self):
122 122 return manifest.manifest(self.sopener)
123 123
124 124 @propertycache
125 125 def dirstate(self):
126 126 return dirstate.dirstate(self.opener, self.ui, self.root)
127 127
128 128 def __getitem__(self, changeid):
129 129 if changeid is None:
130 130 return context.workingctx(self)
131 131 return context.changectx(self, changeid)
132 132
133 133 def __contains__(self, changeid):
134 134 try:
135 135 return bool(self.lookup(changeid))
136 136 except error.RepoLookupError:
137 137 return False
138 138
139 139 def __nonzero__(self):
140 140 return True
141 141
142 142 def __len__(self):
143 143 return len(self.changelog)
144 144
145 145 def __iter__(self):
146 146 for i in xrange(len(self)):
147 147 yield i
148 148
149 149 def url(self):
150 150 return 'file:' + self.root
151 151
152 152 def hook(self, name, throw=False, **args):
153 153 return hook.hook(self.ui, self, name, throw, **args)
154 154
155 155 tag_disallowed = ':\r\n'
156 156
157 157 def _tag(self, names, node, message, local, user, date, extra={}):
158 158 if isinstance(names, str):
159 159 allchars = names
160 160 names = (names,)
161 161 else:
162 162 allchars = ''.join(names)
163 163 for c in self.tag_disallowed:
164 164 if c in allchars:
165 165 raise util.Abort(_('%r cannot be used in a tag name') % c)
166 166
167 167 branches = self.branchmap()
168 168 for name in names:
169 169 self.hook('pretag', throw=True, node=hex(node), tag=name,
170 170 local=local)
171 171 if name in branches:
172 172 self.ui.warn(_("warning: tag %s conflicts with existing"
173 173 " branch name\n") % name)
174 174
175 175 def writetags(fp, names, munge, prevtags):
176 176 fp.seek(0, 2)
177 177 if prevtags and prevtags[-1] != '\n':
178 178 fp.write('\n')
179 179 for name in names:
180 180 m = munge and munge(name) or name
181 181 if self._tagtypes and name in self._tagtypes:
182 182 old = self._tags.get(name, nullid)
183 183 fp.write('%s %s\n' % (hex(old), m))
184 184 fp.write('%s %s\n' % (hex(node), m))
185 185 fp.close()
186 186
187 187 prevtags = ''
188 188 if local:
189 189 try:
190 190 fp = self.opener('localtags', 'r+')
191 191 except IOError:
192 192 fp = self.opener('localtags', 'a')
193 193 else:
194 194 prevtags = fp.read()
195 195
196 196 # local tags are stored in the current charset
197 197 writetags(fp, names, None, prevtags)
198 198 for name in names:
199 199 self.hook('tag', node=hex(node), tag=name, local=local)
200 200 return
201 201
202 202 try:
203 203 fp = self.wfile('.hgtags', 'rb+')
204 204 except IOError:
205 205 fp = self.wfile('.hgtags', 'ab')
206 206 else:
207 207 prevtags = fp.read()
208 208
209 209 # committed tags are stored in UTF-8
210 210 writetags(fp, names, encoding.fromlocal, prevtags)
211 211
212 212 if '.hgtags' not in self.dirstate:
213 213 self[None].add(['.hgtags'])
214 214
215 215 m = matchmod.exact(self.root, '', ['.hgtags'])
216 216 tagnode = self.commit(message, user, date, extra=extra, match=m)
217 217
218 218 for name in names:
219 219 self.hook('tag', node=hex(node), tag=name, local=local)
220 220
221 221 return tagnode
222 222
223 223 def tag(self, names, node, message, local, user, date):
224 224 '''tag a revision with one or more symbolic names.
225 225
226 226 names is a list of strings or, when adding a single tag, names may be a
227 227 string.
228 228
229 229 if local is True, the tags are stored in a per-repository file.
230 230 otherwise, they are stored in the .hgtags file, and a new
231 231 changeset is committed with the change.
232 232
233 233 keyword arguments:
234 234
235 235 local: whether to store tags in non-version-controlled file
236 236 (default False)
237 237
238 238 message: commit message to use if committing
239 239
240 240 user: name of user to use if committing
241 241
242 242 date: date tuple to use if committing'''
243 243
244 244 for x in self.status()[:5]:
245 245 if '.hgtags' in x:
246 246 raise util.Abort(_('working copy of .hgtags is changed '
247 247 '(please commit .hgtags manually)'))
248 248
249 249 self.tags() # instantiate the cache
250 250 self._tag(names, node, message, local, user, date)
251 251
252 252 def tags(self):
253 253 '''return a mapping of tag to node'''
254 254 if self._tags is None:
255 255 (self._tags, self._tagtypes) = self._findtags()
256 256
257 257 return self._tags
258 258
259 259 def _findtags(self):
260 260 '''Do the hard work of finding tags. Return a pair of dicts
261 261 (tags, tagtypes) where tags maps tag name to node, and tagtypes
262 262 maps tag name to a string like \'global\' or \'local\'.
263 263 Subclasses or extensions are free to add their own tags, but
264 264 should be aware that the returned dicts will be retained for the
265 265 duration of the localrepo object.'''
266 266
267 267 # XXX what tagtype should subclasses/extensions use? Currently
268 268 # mq and bookmarks add tags, but do not set the tagtype at all.
269 269 # Should each extension invent its own tag type? Should there
270 270 # be one tagtype for all such "virtual" tags? Or is the status
271 271 # quo fine?
272 272
273 273 alltags = {} # map tag name to (node, hist)
274 274 tagtypes = {}
275 275
276 276 tagsmod.findglobaltags(self.ui, self, alltags, tagtypes)
277 277 tagsmod.readlocaltags(self.ui, self, alltags, tagtypes)
278 278
279 279 # Build the return dicts. Have to re-encode tag names because
280 280 # the tags module always uses UTF-8 (in order not to lose info
281 281 # writing to the cache), but the rest of Mercurial wants them in
282 282 # local encoding.
283 283 tags = {}
284 284 for (name, (node, hist)) in alltags.iteritems():
285 285 if node != nullid:
286 286 tags[encoding.tolocal(name)] = node
287 287 tags['tip'] = self.changelog.tip()
288 288 tagtypes = dict([(encoding.tolocal(name), value)
289 289 for (name, value) in tagtypes.iteritems()])
290 290 return (tags, tagtypes)
291 291
292 292 def tagtype(self, tagname):
293 293 '''
294 294 return the type of the given tag. result can be:
295 295
296 296 'local' : a local tag
297 297 'global' : a global tag
298 298 None : tag does not exist
299 299 '''
300 300
301 301 self.tags()
302 302
303 303 return self._tagtypes.get(tagname)
304 304
305 305 def tagslist(self):
306 306 '''return a list of tags ordered by revision'''
307 307 l = []
308 308 for t, n in self.tags().iteritems():
309 309 try:
310 310 r = self.changelog.rev(n)
311 311 except:
312 312 r = -2 # sort to the beginning of the list if unknown
313 313 l.append((r, t, n))
314 314 return [(t, n) for r, t, n in sorted(l)]
315 315
316 316 def nodetags(self, node):
317 317 '''return the tags associated with a node'''
318 318 if not self.nodetagscache:
319 319 self.nodetagscache = {}
320 320 for t, n in self.tags().iteritems():
321 321 self.nodetagscache.setdefault(n, []).append(t)
322 322 for tags in self.nodetagscache.itervalues():
323 323 tags.sort()
324 324 return self.nodetagscache.get(node, [])
325 325
326 326 def _branchtags(self, partial, lrev):
327 327 # TODO: rename this function?
328 328 tiprev = len(self) - 1
329 329 if lrev != tiprev:
330 330 ctxgen = (self[r] for r in xrange(lrev + 1, tiprev + 1))
331 331 self._updatebranchcache(partial, ctxgen)
332 332 self._writebranchcache(partial, self.changelog.tip(), tiprev)
333 333
334 334 return partial
335 335
336 336 def branchmap(self):
337 337 '''returns a dictionary {branch: [branchheads]}'''
338 338 tip = self.changelog.tip()
339 339 if self._branchcache is not None and self._branchcachetip == tip:
340 340 return self._branchcache
341 341
342 342 oldtip = self._branchcachetip
343 343 self._branchcachetip = tip
344 344 if oldtip is None or oldtip not in self.changelog.nodemap:
345 345 partial, last, lrev = self._readbranchcache()
346 346 else:
347 347 lrev = self.changelog.rev(oldtip)
348 348 partial = self._branchcache
349 349
350 350 self._branchtags(partial, lrev)
351 351 # this private cache holds all heads (not just tips)
352 352 self._branchcache = partial
353 353
354 354 return self._branchcache
355 355
356 356 def branchtags(self):
357 357 '''return a dict where branch names map to the tipmost head of
358 358 the branch, open heads come before closed'''
359 359 bt = {}
360 360 for bn, heads in self.branchmap().iteritems():
361 361 tip = heads[-1]
362 362 for h in reversed(heads):
363 363 if 'close' not in self.changelog.read(h)[5]:
364 364 tip = h
365 365 break
366 366 bt[bn] = tip
367 367 return bt
368 368
369 369
370 370 def _readbranchcache(self):
371 371 partial = {}
372 372 try:
373 373 f = self.opener("branchheads.cache")
374 374 lines = f.read().split('\n')
375 375 f.close()
376 376 except (IOError, OSError):
377 377 return {}, nullid, nullrev
378 378
379 379 try:
380 380 last, lrev = lines.pop(0).split(" ", 1)
381 381 last, lrev = bin(last), int(lrev)
382 382 if lrev >= len(self) or self[lrev].node() != last:
383 383 # invalidate the cache
384 384 raise ValueError('invalidating branch cache (tip differs)')
385 385 for l in lines:
386 386 if not l:
387 387 continue
388 388 node, label = l.split(" ", 1)
389 389 partial.setdefault(label.strip(), []).append(bin(node))
390 390 except KeyboardInterrupt:
391 391 raise
392 392 except Exception, inst:
393 393 if self.ui.debugflag:
394 394 self.ui.warn(str(inst), '\n')
395 395 partial, last, lrev = {}, nullid, nullrev
396 396 return partial, last, lrev
397 397
398 398 def _writebranchcache(self, branches, tip, tiprev):
399 399 try:
400 400 f = self.opener("branchheads.cache", "w", atomictemp=True)
401 401 f.write("%s %s\n" % (hex(tip), tiprev))
402 402 for label, nodes in branches.iteritems():
403 403 for node in nodes:
404 404 f.write("%s %s\n" % (hex(node), label))
405 405 f.rename()
406 406 except (IOError, OSError):
407 407 pass
408 408
409 409 def _updatebranchcache(self, partial, ctxgen):
410 410 # collect new branch entries
411 411 newbranches = {}
412 412 for c in ctxgen:
413 413 newbranches.setdefault(c.branch(), []).append(c.node())
414 414 # if older branchheads are reachable from new ones, they aren't
415 415 # really branchheads. Note checking parents is insufficient:
416 416 # 1 (branch a) -> 2 (branch b) -> 3 (branch a)
417 417 for branch, newnodes in newbranches.iteritems():
418 418 bheads = partial.setdefault(branch, [])
419 419 bheads.extend(newnodes)
420 420 if len(bheads) <= 1:
421 421 continue
422 422 # starting from tip means fewer passes over reachable
423 423 while newnodes:
424 424 latest = newnodes.pop()
425 425 if latest not in bheads:
426 426 continue
427 427 minbhrev = self[min([self[bh].rev() for bh in bheads])].node()
428 428 reachable = self.changelog.reachable(latest, minbhrev)
429 429 reachable.remove(latest)
430 430 bheads = [b for b in bheads if b not in reachable]
431 431 partial[branch] = bheads
432 432
433 433 def lookup(self, key):
434 434 if isinstance(key, int):
435 435 return self.changelog.node(key)
436 436 elif key == '.':
437 437 return self.dirstate.parents()[0]
438 438 elif key == 'null':
439 439 return nullid
440 440 elif key == 'tip':
441 441 return self.changelog.tip()
442 442 n = self.changelog._match(key)
443 443 if n:
444 444 return n
445 445 if key in self.tags():
446 446 return self.tags()[key]
447 447 if key in self.branchtags():
448 448 return self.branchtags()[key]
449 449 n = self.changelog._partialmatch(key)
450 450 if n:
451 451 return n
452 452
453 453 # can't find key, check if it might have come from damaged dirstate
454 454 if key in self.dirstate.parents():
455 455 raise error.Abort(_("working directory has unknown parent '%s'!")
456 456 % short(key))
457 457 try:
458 458 if len(key) == 20:
459 459 key = hex(key)
460 460 except:
461 461 pass
462 462 raise error.RepoLookupError(_("unknown revision '%s'") % key)
463 463
464 464 def lookupbranch(self, key, remote=None):
465 465 repo = remote or self
466 466 if key in repo.branchmap():
467 467 return key
468 468
469 469 repo = (remote and remote.local()) and remote or self
470 470 return repo[key].branch()
471 471
472 472 def local(self):
473 473 return True
474 474
475 475 def join(self, f):
476 476 return os.path.join(self.path, f)
477 477
478 478 def wjoin(self, f):
479 479 return os.path.join(self.root, f)
480 480
481 481 def rjoin(self, f):
482 482 return os.path.join(self.root, util.pconvert(f))
483 483
484 484 def file(self, f):
485 485 if f[0] == '/':
486 486 f = f[1:]
487 487 return filelog.filelog(self.sopener, f)
488 488
489 489 def changectx(self, changeid):
490 490 return self[changeid]
491 491
492 492 def parents(self, changeid=None):
493 493 '''get list of changectxs for parents of changeid'''
494 494 return self[changeid].parents()
495 495
496 496 def filectx(self, path, changeid=None, fileid=None):
497 497 """changeid can be a changeset revision, node, or tag.
498 498 fileid can be a file revision or node."""
499 499 return context.filectx(self, path, changeid, fileid)
500 500
501 501 def getcwd(self):
502 502 return self.dirstate.getcwd()
503 503
504 504 def pathto(self, f, cwd=None):
505 505 return self.dirstate.pathto(f, cwd)
506 506
507 507 def wfile(self, f, mode='r'):
508 508 return self.wopener(f, mode)
509 509
510 510 def _link(self, f):
511 511 return os.path.islink(self.wjoin(f))
512 512
513 513 def _filter(self, filter, filename, data):
514 514 if filter not in self.filterpats:
515 515 l = []
516 516 for pat, cmd in self.ui.configitems(filter):
517 517 if cmd == '!':
518 518 continue
519 519 mf = matchmod.match(self.root, '', [pat])
520 520 fn = None
521 521 params = cmd
522 522 for name, filterfn in self._datafilters.iteritems():
523 523 if cmd.startswith(name):
524 524 fn = filterfn
525 525 params = cmd[len(name):].lstrip()
526 526 break
527 527 if not fn:
528 528 fn = lambda s, c, **kwargs: util.filter(s, c)
529 529 # Wrap old filters not supporting keyword arguments
530 530 if not inspect.getargspec(fn)[2]:
531 531 oldfn = fn
532 532 fn = lambda s, c, **kwargs: oldfn(s, c)
533 533 l.append((mf, fn, params))
534 534 self.filterpats[filter] = l
535 535
536 536 for mf, fn, cmd in self.filterpats[filter]:
537 537 if mf(filename):
538 538 self.ui.debug("filtering %s through %s\n" % (filename, cmd))
539 539 data = fn(data, cmd, ui=self.ui, repo=self, filename=filename)
540 540 break
541 541
542 542 return data
543 543
544 544 def adddatafilter(self, name, filter):
545 545 self._datafilters[name] = filter
546 546
547 547 def wread(self, filename):
548 548 if self._link(filename):
549 549 data = os.readlink(self.wjoin(filename))
550 550 else:
551 551 data = self.wopener(filename, 'r').read()
552 552 return self._filter("encode", filename, data)
553 553
554 554 def wwrite(self, filename, data, flags):
555 555 data = self._filter("decode", filename, data)
556 556 try:
557 557 os.unlink(self.wjoin(filename))
558 558 except OSError:
559 559 pass
560 560 if 'l' in flags:
561 561 self.wopener.symlink(data, filename)
562 562 else:
563 563 self.wopener(filename, 'w').write(data)
564 564 if 'x' in flags:
565 565 util.set_flags(self.wjoin(filename), False, True)
566 566
567 567 def wwritedata(self, filename, data):
568 568 return self._filter("decode", filename, data)
569 569
570 570 def transaction(self, desc):
571 571 tr = self._transref and self._transref() or None
572 572 if tr and tr.running():
573 573 return tr.nest()
574 574
575 575 # abort here if the journal already exists
576 576 if os.path.exists(self.sjoin("journal")):
577 577 raise error.RepoError(
578 578 _("abandoned transaction found - run hg recover"))
579 579
580 580 # save dirstate for rollback
581 581 try:
582 582 ds = self.opener("dirstate").read()
583 583 except IOError:
584 584 ds = ""
585 585 self.opener("journal.dirstate", "w").write(ds)
586 586 self.opener("journal.branch", "w").write(self.dirstate.branch())
587 587 self.opener("journal.desc", "w").write("%d\n%s\n" % (len(self), desc))
588 588
589 589 renames = [(self.sjoin("journal"), self.sjoin("undo")),
590 590 (self.join("journal.dirstate"), self.join("undo.dirstate")),
591 591 (self.join("journal.branch"), self.join("undo.branch")),
592 592 (self.join("journal.desc"), self.join("undo.desc"))]
593 593 tr = transaction.transaction(self.ui.warn, self.sopener,
594 594 self.sjoin("journal"),
595 595 aftertrans(renames),
596 596 self.store.createmode)
597 597 self._transref = weakref.ref(tr)
598 598 return tr
599 599
600 600 def recover(self):
601 601 lock = self.lock()
602 602 try:
603 603 if os.path.exists(self.sjoin("journal")):
604 604 self.ui.status(_("rolling back interrupted transaction\n"))
605 605 transaction.rollback(self.sopener, self.sjoin("journal"),
606 606 self.ui.warn)
607 607 self.invalidate()
608 608 return True
609 609 else:
610 610 self.ui.warn(_("no interrupted transaction available\n"))
611 611 return False
612 612 finally:
613 613 lock.release()
614 614
615 615 def rollback(self, dryrun=False):
616 616 wlock = lock = None
617 617 try:
618 618 wlock = self.wlock()
619 619 lock = self.lock()
620 620 if os.path.exists(self.sjoin("undo")):
621 621 try:
622 622 args = self.opener("undo.desc", "r").read().splitlines()
623 623 if len(args) >= 3 and self.ui.verbose:
624 624 desc = _("rolling back to revision %s"
625 625 " (undo %s: %s)\n") % (
626 626 int(args[0]) - 1, args[1], args[2])
627 627 elif len(args) >= 2:
628 628 desc = _("rolling back to revision %s (undo %s)\n") % (
629 629 int(args[0]) - 1, args[1])
630 630 except IOError:
631 631 desc = _("rolling back unknown transaction\n")
632 632 self.ui.status(desc)
633 633 if dryrun:
634 634 return
635 635 transaction.rollback(self.sopener, self.sjoin("undo"),
636 636 self.ui.warn)
637 637 util.rename(self.join("undo.dirstate"), self.join("dirstate"))
638 638 try:
639 639 branch = self.opener("undo.branch").read()
640 640 self.dirstate.setbranch(branch)
641 641 except IOError:
642 642 self.ui.warn(_("Named branch could not be reset, "
643 643 "current branch still is: %s\n")
644 644 % encoding.tolocal(self.dirstate.branch()))
645 645 self.invalidate()
646 646 self.dirstate.invalidate()
647 647 self.destroyed()
648 648 else:
649 649 self.ui.warn(_("no rollback information available\n"))
650 650 return 1
651 651 finally:
652 652 release(lock, wlock)
653 653
654 654 def invalidatecaches(self):
655 655 self._tags = None
656 656 self._tagtypes = None
657 657 self.nodetagscache = None
658 658 self._branchcache = None # in UTF-8
659 659 self._branchcachetip = None
660 660
661 661 def invalidate(self):
662 662 for a in "changelog manifest".split():
663 663 if a in self.__dict__:
664 664 delattr(self, a)
665 665 self.invalidatecaches()
666 666
667 667 def _lock(self, lockname, wait, releasefn, acquirefn, desc):
668 668 try:
669 669 l = lock.lock(lockname, 0, releasefn, desc=desc)
670 670 except error.LockHeld, inst:
671 671 if not wait:
672 672 raise
673 673 self.ui.warn(_("waiting for lock on %s held by %r\n") %
674 674 (desc, inst.locker))
675 675 # default to 600 seconds timeout
676 676 l = lock.lock(lockname, int(self.ui.config("ui", "timeout", "600")),
677 677 releasefn, desc=desc)
678 678 if acquirefn:
679 679 acquirefn()
680 680 return l
681 681
682 682 def lock(self, wait=True):
683 683 '''Lock the repository store (.hg/store) and return a weak reference
684 684 to the lock. Use this before modifying the store (e.g. committing or
685 685 stripping). If you are opening a transaction, get a lock as well.)'''
686 686 l = self._lockref and self._lockref()
687 687 if l is not None and l.held:
688 688 l.lock()
689 689 return l
690 690
691 691 l = self._lock(self.sjoin("lock"), wait, None, self.invalidate,
692 692 _('repository %s') % self.origroot)
693 693 self._lockref = weakref.ref(l)
694 694 return l
695 695
696 696 def wlock(self, wait=True):
697 697 '''Lock the non-store parts of the repository (everything under
698 698 .hg except .hg/store) and return a weak reference to the lock.
699 699 Use this before modifying files in .hg.'''
700 700 l = self._wlockref and self._wlockref()
701 701 if l is not None and l.held:
702 702 l.lock()
703 703 return l
704 704
705 705 l = self._lock(self.join("wlock"), wait, self.dirstate.write,
706 706 self.dirstate.invalidate, _('working directory of %s') %
707 707 self.origroot)
708 708 self._wlockref = weakref.ref(l)
709 709 return l
710 710
711 711 def _filecommit(self, fctx, manifest1, manifest2, linkrev, tr, changelist):
712 712 """
713 713 commit an individual file as part of a larger transaction
714 714 """
715 715
716 716 fname = fctx.path()
717 717 text = fctx.data()
718 718 flog = self.file(fname)
719 719 fparent1 = manifest1.get(fname, nullid)
720 720 fparent2 = fparent2o = manifest2.get(fname, nullid)
721 721
722 722 meta = {}
723 723 copy = fctx.renamed()
724 724 if copy and copy[0] != fname:
725 725 # Mark the new revision of this file as a copy of another
726 726 # file. This copy data will effectively act as a parent
727 727 # of this new revision. If this is a merge, the first
728 728 # parent will be the nullid (meaning "look up the copy data")
729 729 # and the second one will be the other parent. For example:
730 730 #
731 731 # 0 --- 1 --- 3 rev1 changes file foo
732 732 # \ / rev2 renames foo to bar and changes it
733 733 # \- 2 -/ rev3 should have bar with all changes and
734 734 # should record that bar descends from
735 735 # bar in rev2 and foo in rev1
736 736 #
737 737 # this allows this merge to succeed:
738 738 #
739 739 # 0 --- 1 --- 3 rev4 reverts the content change from rev2
740 740 # \ / merging rev3 and rev4 should use bar@rev2
741 741 # \- 2 --- 4 as the merge base
742 742 #
743 743
744 744 cfname = copy[0]
745 745 crev = manifest1.get(cfname)
746 746 newfparent = fparent2
747 747
748 748 if manifest2: # branch merge
749 749 if fparent2 == nullid or crev is None: # copied on remote side
750 750 if cfname in manifest2:
751 751 crev = manifest2[cfname]
752 752 newfparent = fparent1
753 753
754 754 # find source in nearest ancestor if we've lost track
755 755 if not crev:
756 756 self.ui.debug(" %s: searching for copy revision for %s\n" %
757 757 (fname, cfname))
758 758 for ancestor in self['.'].ancestors():
759 759 if cfname in ancestor:
760 760 crev = ancestor[cfname].filenode()
761 761 break
762 762
763 763 self.ui.debug(" %s: copy %s:%s\n" % (fname, cfname, hex(crev)))
764 764 meta["copy"] = cfname
765 765 meta["copyrev"] = hex(crev)
766 766 fparent1, fparent2 = nullid, newfparent
767 767 elif fparent2 != nullid:
768 768 # is one parent an ancestor of the other?
769 769 fparentancestor = flog.ancestor(fparent1, fparent2)
770 770 if fparentancestor == fparent1:
771 771 fparent1, fparent2 = fparent2, nullid
772 772 elif fparentancestor == fparent2:
773 773 fparent2 = nullid
774 774
775 775 # is the file changed?
776 776 if fparent2 != nullid or flog.cmp(fparent1, text) or meta:
777 777 changelist.append(fname)
778 778 return flog.add(text, meta, tr, linkrev, fparent1, fparent2)
779 779
780 780 # are just the flags changed during merge?
781 781 if fparent1 != fparent2o and manifest1.flags(fname) != fctx.flags():
782 782 changelist.append(fname)
783 783
784 784 return fparent1
785 785
786 786 def commit(self, text="", user=None, date=None, match=None, force=False,
787 787 editor=False, extra={}):
788 788 """Add a new revision to current repository.
789 789
790 790 Revision information is gathered from the working directory,
791 791 match can be used to filter the committed files. If editor is
792 792 supplied, it is called to get a commit message.
793 793 """
794 794
795 795 def fail(f, msg):
796 796 raise util.Abort('%s: %s' % (f, msg))
797 797
798 798 if not match:
799 799 match = matchmod.always(self.root, '')
800 800
801 801 if not force:
802 802 vdirs = []
803 803 match.dir = vdirs.append
804 804 match.bad = fail
805 805
806 806 wlock = self.wlock()
807 807 try:
808 808 wctx = self[None]
809 809 merge = len(wctx.parents()) > 1
810 810
811 811 if (not force and merge and match and
812 812 (match.files() or match.anypats())):
813 813 raise util.Abort(_('cannot partially commit a merge '
814 814 '(do not specify files or patterns)'))
815 815
816 816 changes = self.status(match=match, clean=force)
817 817 if force:
818 818 changes[0].extend(changes[6]) # mq may commit unchanged files
819 819
820 820 # check subrepos
821 821 subs = []
822 822 removedsubs = set()
823 823 for p in wctx.parents():
824 824 removedsubs.update(s for s in p.substate if match(s))
825 825 for s in wctx.substate:
826 826 removedsubs.discard(s)
827 827 if match(s) and wctx.sub(s).dirty():
828 828 subs.append(s)
829 829 if (subs or removedsubs):
830 830 if (not match('.hgsub') and
831 831 '.hgsub' in (wctx.modified() + wctx.added())):
832 832 raise util.Abort(_("can't commit subrepos without .hgsub"))
833 833 if '.hgsubstate' not in changes[0]:
834 834 changes[0].insert(0, '.hgsubstate')
835 835
836 836 # make sure all explicit patterns are matched
837 837 if not force and match.files():
838 838 matched = set(changes[0] + changes[1] + changes[2])
839 839
840 840 for f in match.files():
841 841 if f == '.' or f in matched or f in wctx.substate:
842 842 continue
843 843 if f in changes[3]: # missing
844 844 fail(f, _('file not found!'))
845 845 if f in vdirs: # visited directory
846 846 d = f + '/'
847 847 for mf in matched:
848 848 if mf.startswith(d):
849 849 break
850 850 else:
851 851 fail(f, _("no match under directory!"))
852 852 elif f not in self.dirstate:
853 853 fail(f, _("file not tracked!"))
854 854
855 855 if (not force and not extra.get("close") and not merge
856 856 and not (changes[0] or changes[1] or changes[2])
857 857 and wctx.branch() == wctx.p1().branch()):
858 858 return None
859 859
860 860 ms = mergemod.mergestate(self)
861 861 for f in changes[0]:
862 862 if f in ms and ms[f] == 'u':
863 863 raise util.Abort(_("unresolved merge conflicts "
864 864 "(see hg resolve)"))
865 865
866 866 cctx = context.workingctx(self, text, user, date, extra, changes)
867 867 if editor:
868 868 cctx._text = editor(self, cctx, subs)
869 869 edited = (text != cctx._text)
870 870
871 871 # commit subs
872 872 if subs or removedsubs:
873 873 state = wctx.substate.copy()
874 874 for s in subs:
875 875 sub = wctx.sub(s)
876 876 self.ui.status(_('committing subrepository %s\n') %
877 877 subrepo.relpath(sub))
878 878 sr = sub.commit(cctx._text, user, date)
879 879 state[s] = (state[s][0], sr)
880 880 subrepo.writestate(self, state)
881 881
882 882 # Save commit message in case this transaction gets rolled back
883 883 # (e.g. by a pretxncommit hook). Leave the content alone on
884 884 # the assumption that the user will use the same editor again.
885 885 msgfile = self.opener('last-message.txt', 'wb')
886 886 msgfile.write(cctx._text)
887 887 msgfile.close()
888 888
889 889 p1, p2 = self.dirstate.parents()
890 890 hookp1, hookp2 = hex(p1), (p2 != nullid and hex(p2) or '')
891 891 try:
892 892 self.hook("precommit", throw=True, parent1=hookp1, parent2=hookp2)
893 893 ret = self.commitctx(cctx, True)
894 894 except:
895 895 if edited:
896 896 msgfn = self.pathto(msgfile.name[len(self.root)+1:])
897 897 self.ui.write(
898 898 _('note: commit message saved in %s\n') % msgfn)
899 899 raise
900 900
901 901 # update dirstate and mergestate
902 902 for f in changes[0] + changes[1]:
903 903 self.dirstate.normal(f)
904 904 for f in changes[2]:
905 905 self.dirstate.forget(f)
906 906 self.dirstate.setparents(ret)
907 907 ms.reset()
908 908 finally:
909 909 wlock.release()
910 910
911 911 self.hook("commit", node=hex(ret), parent1=hookp1, parent2=hookp2)
912 912 return ret
913 913
914 914 def commitctx(self, ctx, error=False):
915 915 """Add a new revision to current repository.
916 916 Revision information is passed via the context argument.
917 917 """
918 918
919 919 tr = lock = None
920 920 removed = ctx.removed()
921 921 p1, p2 = ctx.p1(), ctx.p2()
922 922 m1 = p1.manifest().copy()
923 923 m2 = p2.manifest()
924 924 user = ctx.user()
925 925
926 926 lock = self.lock()
927 927 try:
928 928 tr = self.transaction("commit")
929 929 trp = weakref.proxy(tr)
930 930
931 931 # check in files
932 932 new = {}
933 933 changed = []
934 934 linkrev = len(self)
935 935 for f in sorted(ctx.modified() + ctx.added()):
936 936 self.ui.note(f + "\n")
937 937 try:
938 938 fctx = ctx[f]
939 939 new[f] = self._filecommit(fctx, m1, m2, linkrev, trp,
940 940 changed)
941 941 m1.set(f, fctx.flags())
942 942 except OSError, inst:
943 943 self.ui.warn(_("trouble committing %s!\n") % f)
944 944 raise
945 945 except IOError, inst:
946 946 errcode = getattr(inst, 'errno', errno.ENOENT)
947 947 if error or errcode and errcode != errno.ENOENT:
948 948 self.ui.warn(_("trouble committing %s!\n") % f)
949 949 raise
950 950 else:
951 951 removed.append(f)
952 952
953 953 # update manifest
954 954 m1.update(new)
955 955 removed = [f for f in sorted(removed) if f in m1 or f in m2]
956 956 drop = [f for f in removed if f in m1]
957 957 for f in drop:
958 958 del m1[f]
959 959 mn = self.manifest.add(m1, trp, linkrev, p1.manifestnode(),
960 960 p2.manifestnode(), (new, drop))
961 961
962 962 # update changelog
963 963 self.changelog.delayupdate()
964 964 n = self.changelog.add(mn, changed + removed, ctx.description(),
965 965 trp, p1.node(), p2.node(),
966 966 user, ctx.date(), ctx.extra().copy())
967 967 p = lambda: self.changelog.writepending() and self.root or ""
968 968 xp1, xp2 = p1.hex(), p2 and p2.hex() or ''
969 969 self.hook('pretxncommit', throw=True, node=hex(n), parent1=xp1,
970 970 parent2=xp2, pending=p)
971 971 self.changelog.finalize(trp)
972 972 tr.close()
973 973
974 974 if self._branchcache:
975 975 self.branchtags()
976 976 return n
977 977 finally:
978 978 if tr:
979 979 tr.release()
980 980 lock.release()
981 981
982 982 def destroyed(self):
983 983 '''Inform the repository that nodes have been destroyed.
984 984 Intended for use by strip and rollback, so there's a common
985 985 place for anything that has to be done after destroying history.'''
986 986 # XXX it might be nice if we could take the list of destroyed
987 987 # nodes, but I don't see an easy way for rollback() to do that
988 988
989 989 # Ensure the persistent tag cache is updated. Doing it now
990 990 # means that the tag cache only has to worry about destroyed
991 991 # heads immediately after a strip/rollback. That in turn
992 992 # guarantees that "cachetip == currenttip" (comparing both rev
993 993 # and node) always means no nodes have been added or destroyed.
994 994
995 995 # XXX this is suboptimal when qrefresh'ing: we strip the current
996 996 # head, refresh the tag cache, then immediately add a new head.
997 997 # But I think doing it this way is necessary for the "instant
998 998 # tag cache retrieval" case to work.
999 999 self.invalidatecaches()
1000 1000
1001 1001 def walk(self, match, node=None):
1002 1002 '''
1003 1003 walk recursively through the directory tree or a given
1004 1004 changeset, finding all files matched by the match
1005 1005 function
1006 1006 '''
1007 1007 return self[node].walk(match)
1008 1008
1009 1009 def status(self, node1='.', node2=None, match=None,
1010 1010 ignored=False, clean=False, unknown=False):
1011 1011 """return status of files between two nodes or node and working directory
1012 1012
1013 1013 If node1 is None, use the first dirstate parent instead.
1014 1014 If node2 is None, compare node1 with working directory.
1015 1015 """
1016 1016
1017 1017 def mfmatches(ctx):
1018 1018 mf = ctx.manifest().copy()
1019 1019 for fn in mf.keys():
1020 1020 if not match(fn):
1021 1021 del mf[fn]
1022 1022 return mf
1023 1023
1024 1024 if isinstance(node1, context.changectx):
1025 1025 ctx1 = node1
1026 1026 else:
1027 1027 ctx1 = self[node1]
1028 1028 if isinstance(node2, context.changectx):
1029 1029 ctx2 = node2
1030 1030 else:
1031 1031 ctx2 = self[node2]
1032 1032
1033 1033 working = ctx2.rev() is None
1034 1034 parentworking = working and ctx1 == self['.']
1035 1035 match = match or matchmod.always(self.root, self.getcwd())
1036 1036 listignored, listclean, listunknown = ignored, clean, unknown
1037 1037
1038 1038 # load earliest manifest first for caching reasons
1039 1039 if not working and ctx2.rev() < ctx1.rev():
1040 1040 ctx2.manifest()
1041 1041
1042 1042 if not parentworking:
1043 1043 def bad(f, msg):
1044 1044 if f not in ctx1:
1045 1045 self.ui.warn('%s: %s\n' % (self.dirstate.pathto(f), msg))
1046 1046 match.bad = bad
1047 1047
1048 1048 if working: # we need to scan the working dir
1049 1049 subrepos = []
1050 1050 if '.hgsub' in self.dirstate:
1051 1051 subrepos = ctx1.substate.keys()
1052 1052 s = self.dirstate.status(match, subrepos, listignored,
1053 1053 listclean, listunknown)
1054 1054 cmp, modified, added, removed, deleted, unknown, ignored, clean = s
1055 1055
1056 1056 # check for any possibly clean files
1057 1057 if parentworking and cmp:
1058 1058 fixup = []
1059 1059 # do a full compare of any files that might have changed
1060 1060 for f in sorted(cmp):
1061 1061 if (f not in ctx1 or ctx2.flags(f) != ctx1.flags(f)
1062 1062 or ctx1[f].cmp(ctx2[f].data())):
1063 1063 modified.append(f)
1064 1064 else:
1065 1065 fixup.append(f)
1066 1066
1067 1067 if listclean:
1068 1068 clean += fixup
1069 1069
1070 1070 # update dirstate for files that are actually clean
1071 1071 if fixup:
1072 1072 try:
1073 1073 # updating the dirstate is optional
1074 1074 # so we don't wait on the lock
1075 1075 wlock = self.wlock(False)
1076 1076 try:
1077 1077 for f in fixup:
1078 1078 self.dirstate.normal(f)
1079 1079 finally:
1080 1080 wlock.release()
1081 1081 except error.LockError:
1082 1082 pass
1083 1083
1084 1084 if not parentworking:
1085 1085 mf1 = mfmatches(ctx1)
1086 1086 if working:
1087 1087 # we are comparing working dir against non-parent
1088 1088 # generate a pseudo-manifest for the working dir
1089 1089 mf2 = mfmatches(self['.'])
1090 1090 for f in cmp + modified + added:
1091 1091 mf2[f] = None
1092 1092 mf2.set(f, ctx2.flags(f))
1093 1093 for f in removed:
1094 1094 if f in mf2:
1095 1095 del mf2[f]
1096 1096 else:
1097 1097 # we are comparing two revisions
1098 1098 deleted, unknown, ignored = [], [], []
1099 1099 mf2 = mfmatches(ctx2)
1100 1100
1101 1101 modified, added, clean = [], [], []
1102 1102 for fn in mf2:
1103 1103 if fn in mf1:
1104 1104 if (mf1.flags(fn) != mf2.flags(fn) or
1105 1105 (mf1[fn] != mf2[fn] and
1106 1106 (mf2[fn] or ctx1[fn].cmp(ctx2[fn].data())))):
1107 1107 modified.append(fn)
1108 1108 elif listclean:
1109 1109 clean.append(fn)
1110 1110 del mf1[fn]
1111 1111 else:
1112 1112 added.append(fn)
1113 1113 removed = mf1.keys()
1114 1114
1115 1115 r = modified, added, removed, deleted, unknown, ignored, clean
1116 1116 [l.sort() for l in r]
1117 1117 return r
1118 1118
1119 1119 def heads(self, start=None):
1120 1120 heads = self.changelog.heads(start)
1121 1121 # sort the output in rev descending order
1122 1122 heads = [(-self.changelog.rev(h), h) for h in heads]
1123 1123 return [n for (r, n) in sorted(heads)]
1124 1124
1125 1125 def branchheads(self, branch=None, start=None, closed=False):
1126 1126 '''return a (possibly filtered) list of heads for the given branch
1127 1127
1128 1128 Heads are returned in topological order, from newest to oldest.
1129 1129 If branch is None, use the dirstate branch.
1130 1130 If start is not None, return only heads reachable from start.
1131 1131 If closed is True, return heads that are marked as closed as well.
1132 1132 '''
1133 1133 if branch is None:
1134 1134 branch = self[None].branch()
1135 1135 branches = self.branchmap()
1136 1136 if branch not in branches:
1137 1137 return []
1138 1138 # the cache returns heads ordered lowest to highest
1139 1139 bheads = list(reversed(branches[branch]))
1140 1140 if start is not None:
1141 1141 # filter out the heads that cannot be reached from startrev
1142 1142 fbheads = set(self.changelog.nodesbetween([start], bheads)[2])
1143 1143 bheads = [h for h in bheads if h in fbheads]
1144 1144 if not closed:
1145 1145 bheads = [h for h in bheads if
1146 1146 ('close' not in self.changelog.read(h)[5])]
1147 1147 return bheads
1148 1148
1149 1149 def branches(self, nodes):
1150 1150 if not nodes:
1151 1151 nodes = [self.changelog.tip()]
1152 1152 b = []
1153 1153 for n in nodes:
1154 1154 t = n
1155 1155 while 1:
1156 1156 p = self.changelog.parents(n)
1157 1157 if p[1] != nullid or p[0] == nullid:
1158 1158 b.append((t, n, p[0], p[1]))
1159 1159 break
1160 1160 n = p[0]
1161 1161 return b
1162 1162
1163 1163 def between(self, pairs):
1164 1164 r = []
1165 1165
1166 1166 for top, bottom in pairs:
1167 1167 n, l, i = top, [], 0
1168 1168 f = 1
1169 1169
1170 1170 while n != bottom and n != nullid:
1171 1171 p = self.changelog.parents(n)[0]
1172 1172 if i == f:
1173 1173 l.append(n)
1174 1174 f = f * 2
1175 1175 n = p
1176 1176 i += 1
1177 1177
1178 1178 r.append(l)
1179 1179
1180 1180 return r
1181 1181
1182 1182 def pull(self, remote, heads=None, force=False):
1183 1183 lock = self.lock()
1184 1184 try:
1185 1185 tmp = discovery.findcommonincoming(self, remote, heads=heads,
1186 1186 force=force)
1187 1187 common, fetch, rheads = tmp
1188 1188 if not fetch:
1189 1189 self.ui.status(_("no changes found\n"))
1190 1190 return 0
1191 1191
1192 1192 if fetch == [nullid]:
1193 1193 self.ui.status(_("requesting all changes\n"))
1194 1194 elif heads is None and remote.capable('changegroupsubset'):
1195 1195 # issue1320, avoid a race if remote changed after discovery
1196 1196 heads = rheads
1197 1197
1198 1198 if heads is None:
1199 1199 cg = remote.changegroup(fetch, 'pull')
1200 1200 else:
1201 1201 if not remote.capable('changegroupsubset'):
1202 1202 raise util.Abort(_("Partial pull cannot be done because "
1203 1203 "other repository doesn't support "
1204 1204 "changegroupsubset."))
1205 1205 cg = remote.changegroupsubset(fetch, heads, 'pull')
1206 1206 return self.addchangegroup(cg, 'pull', remote.url(), lock=lock)
1207 1207 finally:
1208 1208 lock.release()
1209 1209
1210 1210 def push(self, remote, force=False, revs=None, newbranch=False):
1211 1211 '''Push outgoing changesets (limited by revs) from the current
1212 1212 repository to remote. Return an integer:
1213 1213 - 0 means HTTP error *or* nothing to push
1214 1214 - 1 means we pushed and remote head count is unchanged *or*
1215 1215 we have outgoing changesets but refused to push
1216 1216 - other values as described by addchangegroup()
1217 1217 '''
1218 1218 # there are two ways to push to remote repo:
1219 1219 #
1220 1220 # addchangegroup assumes local user can lock remote
1221 1221 # repo (local filesystem, old ssh servers).
1222 1222 #
1223 1223 # unbundle assumes local user cannot lock remote repo (new ssh
1224 1224 # servers, http servers).
1225 1225
1226 1226 lock = None
1227 1227 unbundle = remote.capable('unbundle')
1228 1228 if not unbundle:
1229 1229 lock = remote.lock()
1230 1230 try:
1231 1231 ret = discovery.prepush(self, remote, force, revs, newbranch)
1232 1232 if ret[0] is None:
1233 1233 # and here we return 0 for "nothing to push" or 1 for
1234 1234 # "something to push but I refuse"
1235 1235 return ret[1]
1236 1236
1237 1237 cg, remote_heads = ret
1238 1238 if unbundle:
1239 1239 # local repo finds heads on server, finds out what revs it must
1240 1240 # push. once revs transferred, if server finds it has
1241 1241 # different heads (someone else won commit/push race), server
1242 1242 # aborts.
1243 1243 if force:
1244 1244 remote_heads = ['force']
1245 1245 # ssh: return remote's addchangegroup()
1246 1246 # http: return remote's addchangegroup() or 0 for error
1247 1247 return remote.unbundle(cg, remote_heads, 'push')
1248 1248 else:
1249 1249 # we return an integer indicating remote head count change
1250 1250 return remote.addchangegroup(cg, 'push', self.url(), lock=lock)
1251 1251 finally:
1252 1252 if lock is not None:
1253 1253 lock.release()
1254 1254
1255 1255 def changegroupinfo(self, nodes, source):
1256 1256 if self.ui.verbose or source == 'bundle':
1257 1257 self.ui.status(_("%d changesets found\n") % len(nodes))
1258 1258 if self.ui.debugflag:
1259 1259 self.ui.debug("list of changesets:\n")
1260 1260 for node in nodes:
1261 1261 self.ui.debug("%s\n" % hex(node))
1262 1262
1263 1263 def changegroupsubset(self, bases, heads, source, extranodes=None):
1264 1264 """Compute a changegroup consisting of all the nodes that are
1265 1265 descendents of any of the bases and ancestors of any of the heads.
1266 1266 Return a chunkbuffer object whose read() method will return
1267 1267 successive changegroup chunks.
1268 1268
1269 1269 It is fairly complex as determining which filenodes and which
1270 1270 manifest nodes need to be included for the changeset to be complete
1271 1271 is non-trivial.
1272 1272
1273 1273 Another wrinkle is doing the reverse, figuring out which changeset in
1274 1274 the changegroup a particular filenode or manifestnode belongs to.
1275 1275
1276 1276 The caller can specify some nodes that must be included in the
1277 1277 changegroup using the extranodes argument. It should be a dict
1278 1278 where the keys are the filenames (or 1 for the manifest), and the
1279 1279 values are lists of (node, linknode) tuples, where node is a wanted
1280 1280 node and linknode is the changelog node that should be transmitted as
1281 1281 the linkrev.
1282 1282 """
1283 1283
1284 1284 # Set up some initial variables
1285 1285 # Make it easy to refer to self.changelog
1286 1286 cl = self.changelog
1287 1287 # Compute the list of changesets in this changegroup.
1288 1288 # Some bases may turn out to be superfluous, and some heads may be
1289 1289 # too. nodesbetween will return the minimal set of bases and heads
1290 1290 # necessary to re-create the changegroup.
1291 1291 if not bases:
1292 1292 bases = [nullid]
1293 1293 msng_cl_lst, bases, heads = cl.nodesbetween(bases, heads)
1294 1294
1295 1295 if extranodes is None:
1296 1296 # can we go through the fast path ?
1297 1297 heads.sort()
1298 1298 allheads = self.heads()
1299 1299 allheads.sort()
1300 1300 if heads == allheads:
1301 1301 return self._changegroup(msng_cl_lst, source)
1302 1302
1303 1303 # slow path
1304 1304 self.hook('preoutgoing', throw=True, source=source)
1305 1305
1306 1306 self.changegroupinfo(msng_cl_lst, source)
1307 1307
1308 1308 # We assume that all ancestors of bases are known
1309 1309 commonrevs = set(cl.ancestors(*[cl.rev(n) for n in bases]))
1310 1310
1311 1311 # Make it easy to refer to self.manifest
1312 1312 mnfst = self.manifest
1313 1313 # We don't know which manifests are missing yet
1314 1314 msng_mnfst_set = {}
1315 1315 # Nor do we know which filenodes are missing.
1316 1316 msng_filenode_set = {}
1317 1317
1318 1318 junk = mnfst.index[len(mnfst) - 1] # Get around a bug in lazyindex
1319 1319 junk = None
1320 1320
1321 1321 # A changeset always belongs to itself, so the changenode lookup
1322 1322 # function for a changenode is identity.
1323 1323 def identity(x):
1324 1324 return x
1325 1325
1326 1326 # A function generating function that sets up the initial environment
1327 1327 # the inner function.
1328 1328 def filenode_collector(changedfiles):
1329 1329 # This gathers information from each manifestnode included in the
1330 1330 # changegroup about which filenodes the manifest node references
1331 1331 # so we can include those in the changegroup too.
1332 1332 #
1333 1333 # It also remembers which changenode each filenode belongs to. It
1334 1334 # does this by assuming the a filenode belongs to the changenode
1335 1335 # the first manifest that references it belongs to.
1336 1336 def collect_msng_filenodes(mnfstnode):
1337 1337 r = mnfst.rev(mnfstnode)
1338 1338 if r - 1 in mnfst.parentrevs(r):
1339 1339 # If the previous rev is one of the parents,
1340 1340 # we only need to see a diff.
1341 1341 deltamf = mnfst.readdelta(mnfstnode)
1342 1342 # For each line in the delta
1343 1343 for f, fnode in deltamf.iteritems():
1344 1344 # And if the file is in the list of files we care
1345 1345 # about.
1346 1346 if f in changedfiles:
1347 1347 # Get the changenode this manifest belongs to
1348 1348 clnode = msng_mnfst_set[mnfstnode]
1349 1349 # Create the set of filenodes for the file if
1350 1350 # there isn't one already.
1351 1351 ndset = msng_filenode_set.setdefault(f, {})
1352 1352 # And set the filenode's changelog node to the
1353 1353 # manifest's if it hasn't been set already.
1354 1354 ndset.setdefault(fnode, clnode)
1355 1355 else:
1356 1356 # Otherwise we need a full manifest.
1357 1357 m = mnfst.read(mnfstnode)
1358 1358 # For every file in we care about.
1359 1359 for f in changedfiles:
1360 1360 fnode = m.get(f, None)
1361 1361 # If it's in the manifest
1362 1362 if fnode is not None:
1363 1363 # See comments above.
1364 1364 clnode = msng_mnfst_set[mnfstnode]
1365 1365 ndset = msng_filenode_set.setdefault(f, {})
1366 1366 ndset.setdefault(fnode, clnode)
1367 1367 return collect_msng_filenodes
1368 1368
1369 1369 # If we determine that a particular file or manifest node must be a
1370 1370 # node that the recipient of the changegroup will already have, we can
1371 1371 # also assume the recipient will have all the parents. This function
1372 1372 # prunes them from the set of missing nodes.
1373 1373 def prune(revlog, missingnodes):
1374 1374 hasset = set()
1375 1375 # If a 'missing' filenode thinks it belongs to a changenode we
1376 1376 # assume the recipient must have, then the recipient must have
1377 1377 # that filenode.
1378 1378 for n in missingnodes:
1379 1379 clrev = revlog.linkrev(revlog.rev(n))
1380 1380 if clrev in commonrevs:
1381 1381 hasset.add(n)
1382 1382 for n in hasset:
1383 1383 missingnodes.pop(n, None)
1384 1384 for r in revlog.ancestors(*[revlog.rev(n) for n in hasset]):
1385 1385 missingnodes.pop(revlog.node(r), None)
1386 1386
1387 1387 # Add the nodes that were explicitly requested.
1388 1388 def add_extra_nodes(name, nodes):
1389 1389 if not extranodes or name not in extranodes:
1390 1390 return
1391 1391
1392 1392 for node, linknode in extranodes[name]:
1393 1393 if node not in nodes:
1394 1394 nodes[node] = linknode
1395 1395
1396 1396 # Now that we have all theses utility functions to help out and
1397 1397 # logically divide up the task, generate the group.
1398 1398 def gengroup():
1399 1399 # The set of changed files starts empty.
1400 1400 changedfiles = set()
1401 1401 collect = changegroup.collector(cl, msng_mnfst_set, changedfiles)
1402 1402
1403 1403 # Create a changenode group generator that will call our functions
1404 1404 # back to lookup the owning changenode and collect information.
1405 1405 group = cl.group(msng_cl_lst, identity, collect)
1406 cnt = 0
1407 for chnk in group:
1406 for cnt, chnk in enumerate(group):
1408 1407 yield chnk
1409 1408 self.ui.progress(_('bundling changes'), cnt, unit=_('chunks'))
1410 cnt += 1
1411 1409 self.ui.progress(_('bundling changes'), None)
1412 1410
1413 1411 prune(mnfst, msng_mnfst_set)
1414 1412 add_extra_nodes(1, msng_mnfst_set)
1415 1413 msng_mnfst_lst = msng_mnfst_set.keys()
1416 1414 # Sort the manifestnodes by revision number.
1417 1415 msng_mnfst_lst.sort(key=mnfst.rev)
1418 1416 # Create a generator for the manifestnodes that calls our lookup
1419 1417 # and data collection functions back.
1420 1418 group = mnfst.group(msng_mnfst_lst,
1421 1419 lambda mnode: msng_mnfst_set[mnode],
1422 1420 filenode_collector(changedfiles))
1423 cnt = 0
1424 for chnk in group:
1421 for cnt, chnk in enumerate(group):
1425 1422 yield chnk
1426 1423 self.ui.progress(_('bundling manifests'), cnt, unit=_('chunks'))
1427 cnt += 1
1428 1424 self.ui.progress(_('bundling manifests'), None)
1429 1425
1430 1426 # These are no longer needed, dereference and toss the memory for
1431 1427 # them.
1432 1428 msng_mnfst_lst = None
1433 1429 msng_mnfst_set.clear()
1434 1430
1435 1431 if extranodes:
1436 1432 for fname in extranodes:
1437 1433 if isinstance(fname, int):
1438 1434 continue
1439 1435 msng_filenode_set.setdefault(fname, {})
1440 1436 changedfiles.add(fname)
1441 1437 # Go through all our files in order sorted by name.
1442 1438 cnt = 0
1443 1439 for fname in sorted(changedfiles):
1444 1440 filerevlog = self.file(fname)
1445 1441 if not len(filerevlog):
1446 1442 raise util.Abort(_("empty or missing revlog for %s") % fname)
1447 1443 # Toss out the filenodes that the recipient isn't really
1448 1444 # missing.
1449 1445 missingfnodes = msng_filenode_set.pop(fname, {})
1450 1446 prune(filerevlog, missingfnodes)
1451 1447 add_extra_nodes(fname, missingfnodes)
1452 1448 # If any filenodes are left, generate the group for them,
1453 1449 # otherwise don't bother.
1454 1450 if missingfnodes:
1455 1451 yield changegroup.chunkheader(len(fname))
1456 1452 yield fname
1457 1453 # Sort the filenodes by their revision # (topological order)
1458 1454 nodeiter = list(missingfnodes)
1459 1455 nodeiter.sort(key=filerevlog.rev)
1460 1456 # Create a group generator and only pass in a changenode
1461 1457 # lookup function as we need to collect no information
1462 1458 # from filenodes.
1463 1459 group = filerevlog.group(nodeiter,
1464 1460 lambda fnode: missingfnodes[fnode])
1465 1461 for chnk in group:
1466 1462 self.ui.progress(
1467 1463 _('bundling files'), cnt, item=fname, unit=_('chunks'))
1468 1464 cnt += 1
1469 1465 yield chnk
1470 1466 # Signal that no more groups are left.
1471 1467 yield changegroup.closechunk()
1472 1468 self.ui.progress(_('bundling files'), None)
1473 1469
1474 1470 if msng_cl_lst:
1475 1471 self.hook('outgoing', node=hex(msng_cl_lst[0]), source=source)
1476 1472
1477 1473 return util.chunkbuffer(gengroup())
1478 1474
1479 1475 def changegroup(self, basenodes, source):
1480 1476 # to avoid a race we use changegroupsubset() (issue1320)
1481 1477 return self.changegroupsubset(basenodes, self.heads(), source)
1482 1478
1483 1479 def _changegroup(self, nodes, source):
1484 1480 """Compute the changegroup of all nodes that we have that a recipient
1485 1481 doesn't. Return a chunkbuffer object whose read() method will return
1486 1482 successive changegroup chunks.
1487 1483
1488 1484 This is much easier than the previous function as we can assume that
1489 1485 the recipient has any changenode we aren't sending them.
1490 1486
1491 1487 nodes is the set of nodes to send"""
1492 1488
1493 1489 self.hook('preoutgoing', throw=True, source=source)
1494 1490
1495 1491 cl = self.changelog
1496 1492 revset = set([cl.rev(n) for n in nodes])
1497 1493 self.changegroupinfo(nodes, source)
1498 1494
1499 1495 def identity(x):
1500 1496 return x
1501 1497
1502 1498 def gennodelst(log):
1503 1499 for r in log:
1504 1500 if log.linkrev(r) in revset:
1505 1501 yield log.node(r)
1506 1502
1507 1503 def lookuplinkrev_func(revlog):
1508 1504 def lookuplinkrev(n):
1509 1505 return cl.node(revlog.linkrev(revlog.rev(n)))
1510 1506 return lookuplinkrev
1511 1507
1512 1508 def gengroup():
1513 1509 '''yield a sequence of changegroup chunks (strings)'''
1514 1510 # construct a list of all changed files
1515 1511 changedfiles = set()
1516 1512 mmfs = {}
1517 1513 collect = changegroup.collector(cl, mmfs, changedfiles)
1518 1514
1519 cnt = 0
1520 for chnk in cl.group(nodes, identity, collect):
1515 for cnt, chnk in enumerate(cl.group(nodes, identity, collect)):
1521 1516 self.ui.progress(_('bundling changes'), cnt, unit=_('chunks'))
1522 cnt += 1
1523 1517 yield chnk
1524 1518 self.ui.progress(_('bundling changes'), None)
1525 1519
1526 1520 mnfst = self.manifest
1527 1521 nodeiter = gennodelst(mnfst)
1528 cnt = 0
1529 for chnk in mnfst.group(nodeiter, lookuplinkrev_func(mnfst)):
1522 for cnt, chnk in enumerate(mnfst.group(nodeiter,
1523 lookuplinkrev_func(mnfst))):
1530 1524 self.ui.progress(_('bundling manifests'), cnt, unit=_('chunks'))
1531 cnt += 1
1532 1525 yield chnk
1533 1526 self.ui.progress(_('bundling manifests'), None)
1534 1527
1535 1528 cnt = 0
1536 1529 for fname in sorted(changedfiles):
1537 1530 filerevlog = self.file(fname)
1538 1531 if not len(filerevlog):
1539 1532 raise util.Abort(_("empty or missing revlog for %s") % fname)
1540 1533 nodeiter = gennodelst(filerevlog)
1541 1534 nodeiter = list(nodeiter)
1542 1535 if nodeiter:
1543 1536 yield changegroup.chunkheader(len(fname))
1544 1537 yield fname
1545 1538 lookup = lookuplinkrev_func(filerevlog)
1546 1539 for chnk in filerevlog.group(nodeiter, lookup):
1547 1540 self.ui.progress(
1548 1541 _('bundling files'), cnt, item=fname, unit=_('chunks'))
1549 1542 cnt += 1
1550 1543 yield chnk
1551 1544 self.ui.progress(_('bundling files'), None)
1552 1545
1553 1546 yield changegroup.closechunk()
1554 1547
1555 1548 if nodes:
1556 1549 self.hook('outgoing', node=hex(nodes[0]), source=source)
1557 1550
1558 1551 return util.chunkbuffer(gengroup())
1559 1552
1560 1553 def addchangegroup(self, source, srctype, url, emptyok=False, lock=None):
1561 1554 """Add the changegroup returned by source.read() to this repo.
1562 1555 srctype is a string like 'push', 'pull', or 'unbundle'. url is
1563 1556 the URL of the repo where this changegroup is coming from.
1564 1557
1565 1558 Return an integer summarizing the change to this repo:
1566 1559 - nothing changed or no source: 0
1567 1560 - more heads than before: 1+added heads (2..n)
1568 1561 - fewer heads than before: -1-removed heads (-2..-n)
1569 1562 - number of heads stays the same: 1
1570 1563 """
1571 1564 def csmap(x):
1572 1565 self.ui.debug("add changeset %s\n" % short(x))
1573 1566 return len(cl)
1574 1567
1575 1568 def revmap(x):
1576 1569 return cl.rev(x)
1577 1570
1578 1571 if not source:
1579 1572 return 0
1580 1573
1581 1574 self.hook('prechangegroup', throw=True, source=srctype, url=url)
1582 1575
1583 1576 changesets = files = revisions = 0
1584 1577 efiles = set()
1585 1578
1586 1579 # write changelog data to temp files so concurrent readers will not see
1587 1580 # inconsistent view
1588 1581 cl = self.changelog
1589 1582 cl.delayupdate()
1590 1583 oldheads = len(cl.heads())
1591 1584
1592 1585 tr = self.transaction("\n".join([srctype, urlmod.hidepassword(url)]))
1593 1586 try:
1594 1587 trp = weakref.proxy(tr)
1595 1588 # pull off the changeset group
1596 1589 self.ui.status(_("adding changesets\n"))
1597 1590 clstart = len(cl)
1598 1591 class prog(object):
1599 1592 step = _('changesets')
1600 1593 count = 1
1601 1594 ui = self.ui
1602 1595 total = None
1603 1596 def __call__(self):
1604 1597 self.ui.progress(self.step, self.count, unit=_('chunks'),
1605 1598 total=self.total)
1606 1599 self.count += 1
1607 1600 pr = prog()
1608 1601 chunkiter = changegroup.chunkiter(source, progress=pr)
1609 1602 if cl.addgroup(chunkiter, csmap, trp) is None and not emptyok:
1610 1603 raise util.Abort(_("received changelog group is empty"))
1611 1604 clend = len(cl)
1612 1605 changesets = clend - clstart
1613 1606 for c in xrange(clstart, clend):
1614 1607 efiles.update(self[c].files())
1615 1608 efiles = len(efiles)
1616 1609 self.ui.progress(_('changesets'), None)
1617 1610
1618 1611 # pull off the manifest group
1619 1612 self.ui.status(_("adding manifests\n"))
1620 1613 pr.step = _('manifests')
1621 1614 pr.count = 1
1622 1615 pr.total = changesets # manifests <= changesets
1623 1616 chunkiter = changegroup.chunkiter(source, progress=pr)
1624 1617 # no need to check for empty manifest group here:
1625 1618 # if the result of the merge of 1 and 2 is the same in 3 and 4,
1626 1619 # no new manifest will be created and the manifest group will
1627 1620 # be empty during the pull
1628 1621 self.manifest.addgroup(chunkiter, revmap, trp)
1629 1622 self.ui.progress(_('manifests'), None)
1630 1623
1631 1624 needfiles = {}
1632 1625 if self.ui.configbool('server', 'validate', default=False):
1633 1626 # validate incoming csets have their manifests
1634 1627 for cset in xrange(clstart, clend):
1635 1628 mfest = self.changelog.read(self.changelog.node(cset))[0]
1636 1629 mfest = self.manifest.readdelta(mfest)
1637 1630 # store file nodes we must see
1638 1631 for f, n in mfest.iteritems():
1639 1632 needfiles.setdefault(f, set()).add(n)
1640 1633
1641 1634 # process the files
1642 1635 self.ui.status(_("adding file changes\n"))
1643 1636 pr.step = 'files'
1644 1637 pr.count = 1
1645 1638 pr.total = efiles
1646 1639 while 1:
1647 1640 f = changegroup.getchunk(source)
1648 1641 if not f:
1649 1642 break
1650 1643 self.ui.debug("adding %s revisions\n" % f)
1651 1644 pr()
1652 1645 fl = self.file(f)
1653 1646 o = len(fl)
1654 1647 chunkiter = changegroup.chunkiter(source)
1655 1648 if fl.addgroup(chunkiter, revmap, trp) is None:
1656 1649 raise util.Abort(_("received file revlog group is empty"))
1657 1650 revisions += len(fl) - o
1658 1651 files += 1
1659 1652 if f in needfiles:
1660 1653 needs = needfiles[f]
1661 1654 for new in xrange(o, len(fl)):
1662 1655 n = fl.node(new)
1663 1656 if n in needs:
1664 1657 needs.remove(n)
1665 1658 if not needs:
1666 1659 del needfiles[f]
1667 1660 self.ui.progress(_('files'), None)
1668 1661
1669 1662 for f, needs in needfiles.iteritems():
1670 1663 fl = self.file(f)
1671 1664 for n in needs:
1672 1665 try:
1673 1666 fl.rev(n)
1674 1667 except error.LookupError:
1675 1668 raise util.Abort(
1676 1669 _('missing file data for %s:%s - run hg verify') %
1677 1670 (f, hex(n)))
1678 1671
1679 1672 newheads = len(cl.heads())
1680 1673 heads = ""
1681 1674 if oldheads and newheads != oldheads:
1682 1675 heads = _(" (%+d heads)") % (newheads - oldheads)
1683 1676
1684 1677 self.ui.status(_("added %d changesets"
1685 1678 " with %d changes to %d files%s\n")
1686 1679 % (changesets, revisions, files, heads))
1687 1680
1688 1681 if changesets > 0:
1689 1682 p = lambda: cl.writepending() and self.root or ""
1690 1683 self.hook('pretxnchangegroup', throw=True,
1691 1684 node=hex(cl.node(clstart)), source=srctype,
1692 1685 url=url, pending=p)
1693 1686
1694 1687 # make changelog see real files again
1695 1688 cl.finalize(trp)
1696 1689
1697 1690 tr.close()
1698 1691 finally:
1699 1692 tr.release()
1700 1693 if lock:
1701 1694 lock.release()
1702 1695
1703 1696 if changesets > 0:
1704 1697 # forcefully update the on-disk branch cache
1705 1698 self.ui.debug("updating the branch cache\n")
1706 1699 self.branchtags()
1707 1700 self.hook("changegroup", node=hex(cl.node(clstart)),
1708 1701 source=srctype, url=url)
1709 1702
1710 1703 for i in xrange(clstart, clend):
1711 1704 self.hook("incoming", node=hex(cl.node(i)),
1712 1705 source=srctype, url=url)
1713 1706
1714 1707 # never return 0 here:
1715 1708 if newheads < oldheads:
1716 1709 return newheads - oldheads - 1
1717 1710 else:
1718 1711 return newheads - oldheads + 1
1719 1712
1720 1713
1721 1714 def stream_in(self, remote):
1722 1715 fp = remote.stream_out()
1723 1716 l = fp.readline()
1724 1717 try:
1725 1718 resp = int(l)
1726 1719 except ValueError:
1727 1720 raise error.ResponseError(
1728 1721 _('Unexpected response from remote server:'), l)
1729 1722 if resp == 1:
1730 1723 raise util.Abort(_('operation forbidden by server'))
1731 1724 elif resp == 2:
1732 1725 raise util.Abort(_('locking the remote repository failed'))
1733 1726 elif resp != 0:
1734 1727 raise util.Abort(_('the server sent an unknown error code'))
1735 1728 self.ui.status(_('streaming all changes\n'))
1736 1729 l = fp.readline()
1737 1730 try:
1738 1731 total_files, total_bytes = map(int, l.split(' ', 1))
1739 1732 except (ValueError, TypeError):
1740 1733 raise error.ResponseError(
1741 1734 _('Unexpected response from remote server:'), l)
1742 1735 self.ui.status(_('%d files to transfer, %s of data\n') %
1743 1736 (total_files, util.bytecount(total_bytes)))
1744 1737 start = time.time()
1745 1738 for i in xrange(total_files):
1746 1739 # XXX doesn't support '\n' or '\r' in filenames
1747 1740 l = fp.readline()
1748 1741 try:
1749 1742 name, size = l.split('\0', 1)
1750 1743 size = int(size)
1751 1744 except (ValueError, TypeError):
1752 1745 raise error.ResponseError(
1753 1746 _('Unexpected response from remote server:'), l)
1754 1747 self.ui.debug('adding %s (%s)\n' % (name, util.bytecount(size)))
1755 1748 # for backwards compat, name was partially encoded
1756 1749 ofp = self.sopener(store.decodedir(name), 'w')
1757 1750 for chunk in util.filechunkiter(fp, limit=size):
1758 1751 ofp.write(chunk)
1759 1752 ofp.close()
1760 1753 elapsed = time.time() - start
1761 1754 if elapsed <= 0:
1762 1755 elapsed = 0.001
1763 1756 self.ui.status(_('transferred %s in %.1f seconds (%s/sec)\n') %
1764 1757 (util.bytecount(total_bytes), elapsed,
1765 1758 util.bytecount(total_bytes / elapsed)))
1766 1759 self.invalidate()
1767 1760 return len(self.heads()) + 1
1768 1761
1769 1762 def clone(self, remote, heads=[], stream=False):
1770 1763 '''clone remote repository.
1771 1764
1772 1765 keyword arguments:
1773 1766 heads: list of revs to clone (forces use of pull)
1774 1767 stream: use streaming clone if possible'''
1775 1768
1776 1769 # now, all clients that can request uncompressed clones can
1777 1770 # read repo formats supported by all servers that can serve
1778 1771 # them.
1779 1772
1780 1773 # if revlog format changes, client will have to check version
1781 1774 # and format flags on "stream" capability, and use
1782 1775 # uncompressed only if compatible.
1783 1776
1784 1777 if stream and not heads and remote.capable('stream'):
1785 1778 return self.stream_in(remote)
1786 1779 return self.pull(remote, heads)
1787 1780
1788 1781 def pushkey(self, namespace, key, old, new):
1789 1782 return pushkey.push(self, namespace, key, old, new)
1790 1783
1791 1784 def listkeys(self, namespace):
1792 1785 return pushkey.list(self, namespace)
1793 1786
1794 1787 # used to avoid circular references so destructors work
1795 1788 def aftertrans(files):
1796 1789 renamefiles = [tuple(t) for t in files]
1797 1790 def a():
1798 1791 for src, dest in renamefiles:
1799 1792 util.rename(src, dest)
1800 1793 return a
1801 1794
1802 1795 def instance(ui, path, create):
1803 1796 return localrepository(ui, util.drop_scheme('file', path), create)
1804 1797
1805 1798 def islocal(path):
1806 1799 return True
General Comments 0
You need to be logged in to leave comments. Login now