##// END OF EJS Templates
revlog: introduce a `sidedata` method...
marmoute -
r43250:0d127278 default
parent child Browse files
Show More
@@ -1,2623 +1,2633 b''
1 1 # revlog.py - storage back-end 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 """Storage back-end for Mercurial.
9 9
10 10 This provides efficient delta storage with O(1) retrieve and append
11 11 and O(changes) merge between branches.
12 12 """
13 13
14 14 from __future__ import absolute_import
15 15
16 16 import collections
17 17 import contextlib
18 18 import errno
19 19 import io
20 20 import os
21 21 import struct
22 22 import zlib
23 23
24 24 # import stuff from node for others to import from revlog
25 25 from .node import (
26 26 bin,
27 27 hex,
28 28 nullhex,
29 29 nullid,
30 30 nullrev,
31 31 short,
32 32 wdirfilenodeids,
33 33 wdirhex,
34 34 wdirid,
35 35 wdirrev,
36 36 )
37 37 from .i18n import _
38 38 from .revlogutils.constants import (
39 39 FLAG_GENERALDELTA,
40 40 FLAG_INLINE_DATA,
41 41 REVLOGV0,
42 42 REVLOGV1,
43 43 REVLOGV1_FLAGS,
44 44 REVLOGV2,
45 45 REVLOGV2_FLAGS,
46 46 REVLOG_DEFAULT_FLAGS,
47 47 REVLOG_DEFAULT_FORMAT,
48 48 REVLOG_DEFAULT_VERSION,
49 49 )
50 50 from .revlogutils.flagutil import (
51 51 REVIDX_DEFAULT_FLAGS,
52 52 REVIDX_ELLIPSIS,
53 53 REVIDX_EXTSTORED,
54 54 REVIDX_FLAGS_ORDER,
55 55 REVIDX_ISCENSORED,
56 56 REVIDX_RAWTEXT_CHANGING_FLAGS,
57 57 )
58 58 from .thirdparty import (
59 59 attr,
60 60 )
61 61 from . import (
62 62 ancestor,
63 63 dagop,
64 64 error,
65 65 mdiff,
66 66 policy,
67 67 pycompat,
68 68 templatefilters,
69 69 util,
70 70 )
71 71 from .interfaces import (
72 72 repository,
73 73 util as interfaceutil,
74 74 )
75 75 from .revlogutils import (
76 76 deltas as deltautil,
77 77 flagutil,
78 78 )
79 79 from .utils import (
80 80 storageutil,
81 81 stringutil,
82 82 )
83 83
84 84 # blanked usage of all the name to prevent pyflakes constraints
85 85 # We need these name available in the module for extensions.
86 86 REVLOGV0
87 87 REVLOGV1
88 88 REVLOGV2
89 89 FLAG_INLINE_DATA
90 90 FLAG_GENERALDELTA
91 91 REVLOG_DEFAULT_FLAGS
92 92 REVLOG_DEFAULT_FORMAT
93 93 REVLOG_DEFAULT_VERSION
94 94 REVLOGV1_FLAGS
95 95 REVLOGV2_FLAGS
96 96 REVIDX_ISCENSORED
97 97 REVIDX_ELLIPSIS
98 98 REVIDX_EXTSTORED
99 99 REVIDX_DEFAULT_FLAGS
100 100 REVIDX_FLAGS_ORDER
101 101 REVIDX_RAWTEXT_CHANGING_FLAGS
102 102
103 103 parsers = policy.importmod(r'parsers')
104 104 rustancestor = policy.importrust(r'ancestor')
105 105 rustdagop = policy.importrust(r'dagop')
106 106
107 107 # Aliased for performance.
108 108 _zlibdecompress = zlib.decompress
109 109
110 110 # max size of revlog with inline data
111 111 _maxinline = 131072
112 112 _chunksize = 1048576
113 113
114 114 # Flag processors for REVIDX_ELLIPSIS.
115 115 def ellipsisreadprocessor(rl, text):
116 116 return text, False
117 117
118 118 def ellipsiswriteprocessor(rl, text):
119 119 return text, False
120 120
121 121 def ellipsisrawprocessor(rl, text):
122 122 return False
123 123
124 124 ellipsisprocessor = (
125 125 ellipsisreadprocessor,
126 126 ellipsiswriteprocessor,
127 127 ellipsisrawprocessor,
128 128 )
129 129
130 130 def getoffset(q):
131 131 return int(q >> 16)
132 132
133 133 def gettype(q):
134 134 return int(q & 0xFFFF)
135 135
136 136 def offset_type(offset, type):
137 137 if (type & ~flagutil.REVIDX_KNOWN_FLAGS) != 0:
138 138 raise ValueError('unknown revlog index flags')
139 139 return int(int(offset) << 16 | type)
140 140
141 141 @attr.s(slots=True, frozen=True)
142 142 class _revisioninfo(object):
143 143 """Information about a revision that allows building its fulltext
144 144 node: expected hash of the revision
145 145 p1, p2: parent revs of the revision
146 146 btext: built text cache consisting of a one-element list
147 147 cachedelta: (baserev, uncompressed_delta) or None
148 148 flags: flags associated to the revision storage
149 149
150 150 One of btext[0] or cachedelta must be set.
151 151 """
152 152 node = attr.ib()
153 153 p1 = attr.ib()
154 154 p2 = attr.ib()
155 155 btext = attr.ib()
156 156 textlen = attr.ib()
157 157 cachedelta = attr.ib()
158 158 flags = attr.ib()
159 159
160 160 @interfaceutil.implementer(repository.irevisiondelta)
161 161 @attr.s(slots=True)
162 162 class revlogrevisiondelta(object):
163 163 node = attr.ib()
164 164 p1node = attr.ib()
165 165 p2node = attr.ib()
166 166 basenode = attr.ib()
167 167 flags = attr.ib()
168 168 baserevisionsize = attr.ib()
169 169 revision = attr.ib()
170 170 delta = attr.ib()
171 171 linknode = attr.ib(default=None)
172 172
173 173 @interfaceutil.implementer(repository.iverifyproblem)
174 174 @attr.s(frozen=True)
175 175 class revlogproblem(object):
176 176 warning = attr.ib(default=None)
177 177 error = attr.ib(default=None)
178 178 node = attr.ib(default=None)
179 179
180 180 # index v0:
181 181 # 4 bytes: offset
182 182 # 4 bytes: compressed length
183 183 # 4 bytes: base rev
184 184 # 4 bytes: link rev
185 185 # 20 bytes: parent 1 nodeid
186 186 # 20 bytes: parent 2 nodeid
187 187 # 20 bytes: nodeid
188 188 indexformatv0 = struct.Struct(">4l20s20s20s")
189 189 indexformatv0_pack = indexformatv0.pack
190 190 indexformatv0_unpack = indexformatv0.unpack
191 191
192 192 class revlogoldindex(list):
193 193 def __getitem__(self, i):
194 194 if i == -1:
195 195 return (0, 0, 0, -1, -1, -1, -1, nullid)
196 196 return list.__getitem__(self, i)
197 197
198 198 class revlogoldio(object):
199 199 def __init__(self):
200 200 self.size = indexformatv0.size
201 201
202 202 def parseindex(self, data, inline):
203 203 s = self.size
204 204 index = []
205 205 nodemap = {nullid: nullrev}
206 206 n = off = 0
207 207 l = len(data)
208 208 while off + s <= l:
209 209 cur = data[off:off + s]
210 210 off += s
211 211 e = indexformatv0_unpack(cur)
212 212 # transform to revlogv1 format
213 213 e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3],
214 214 nodemap.get(e[4], nullrev), nodemap.get(e[5], nullrev), e[6])
215 215 index.append(e2)
216 216 nodemap[e[6]] = n
217 217 n += 1
218 218
219 219 return revlogoldindex(index), nodemap, None
220 220
221 221 def packentry(self, entry, node, version, rev):
222 222 if gettype(entry[0]):
223 223 raise error.RevlogError(_('index entry flags need revlog '
224 224 'version 1'))
225 225 e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4],
226 226 node(entry[5]), node(entry[6]), entry[7])
227 227 return indexformatv0_pack(*e2)
228 228
229 229 # index ng:
230 230 # 6 bytes: offset
231 231 # 2 bytes: flags
232 232 # 4 bytes: compressed length
233 233 # 4 bytes: uncompressed length
234 234 # 4 bytes: base rev
235 235 # 4 bytes: link rev
236 236 # 4 bytes: parent 1 rev
237 237 # 4 bytes: parent 2 rev
238 238 # 32 bytes: nodeid
239 239 indexformatng = struct.Struct(">Qiiiiii20s12x")
240 240 indexformatng_pack = indexformatng.pack
241 241 versionformat = struct.Struct(">I")
242 242 versionformat_pack = versionformat.pack
243 243 versionformat_unpack = versionformat.unpack
244 244
245 245 # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte
246 246 # signed integer)
247 247 _maxentrysize = 0x7fffffff
248 248
249 249 class revlogio(object):
250 250 def __init__(self):
251 251 self.size = indexformatng.size
252 252
253 253 def parseindex(self, data, inline):
254 254 # call the C implementation to parse the index data
255 255 index, cache = parsers.parse_index2(data, inline)
256 256 return index, getattr(index, 'nodemap', None), cache
257 257
258 258 def packentry(self, entry, node, version, rev):
259 259 p = indexformatng_pack(*entry)
260 260 if rev == 0:
261 261 p = versionformat_pack(version) + p[4:]
262 262 return p
263 263
264 264 class revlog(flagutil.flagprocessorsmixin):
265 265 """
266 266 the underlying revision storage object
267 267
268 268 A revlog consists of two parts, an index and the revision data.
269 269
270 270 The index is a file with a fixed record size containing
271 271 information on each revision, including its nodeid (hash), the
272 272 nodeids of its parents, the position and offset of its data within
273 273 the data file, and the revision it's based on. Finally, each entry
274 274 contains a linkrev entry that can serve as a pointer to external
275 275 data.
276 276
277 277 The revision data itself is a linear collection of data chunks.
278 278 Each chunk represents a revision and is usually represented as a
279 279 delta against the previous chunk. To bound lookup time, runs of
280 280 deltas are limited to about 2 times the length of the original
281 281 version data. This makes retrieval of a version proportional to
282 282 its size, or O(1) relative to the number of revisions.
283 283
284 284 Both pieces of the revlog are written to in an append-only
285 285 fashion, which means we never need to rewrite a file to insert or
286 286 remove data, and can use some simple techniques to avoid the need
287 287 for locking while reading.
288 288
289 289 If checkambig, indexfile is opened with checkambig=True at
290 290 writing, to avoid file stat ambiguity.
291 291
292 292 If mmaplargeindex is True, and an mmapindexthreshold is set, the
293 293 index will be mmapped rather than read if it is larger than the
294 294 configured threshold.
295 295
296 296 If censorable is True, the revlog can have censored revisions.
297 297
298 298 If `upperboundcomp` is not None, this is the expected maximal gain from
299 299 compression for the data content.
300 300 """
301 301 def __init__(self, opener, indexfile, datafile=None, checkambig=False,
302 302 mmaplargeindex=False, censorable=False,
303 303 upperboundcomp=None):
304 304 """
305 305 create a revlog object
306 306
307 307 opener is a function that abstracts the file opening operation
308 308 and can be used to implement COW semantics or the like.
309 309
310 310 """
311 311 self.upperboundcomp = upperboundcomp
312 312 self.indexfile = indexfile
313 313 self.datafile = datafile or (indexfile[:-2] + ".d")
314 314 self.opener = opener
315 315 # When True, indexfile is opened with checkambig=True at writing, to
316 316 # avoid file stat ambiguity.
317 317 self._checkambig = checkambig
318 318 self._mmaplargeindex = mmaplargeindex
319 319 self._censorable = censorable
320 320 # 3-tuple of (node, rev, text) for a raw revision.
321 321 self._revisioncache = None
322 322 # Maps rev to chain base rev.
323 323 self._chainbasecache = util.lrucachedict(100)
324 324 # 2-tuple of (offset, data) of raw data from the revlog at an offset.
325 325 self._chunkcache = (0, '')
326 326 # How much data to read and cache into the raw revlog data cache.
327 327 self._chunkcachesize = 65536
328 328 self._maxchainlen = None
329 329 self._deltabothparents = True
330 330 self.index = []
331 331 # Mapping of partial identifiers to full nodes.
332 332 self._pcache = {}
333 333 # Mapping of revision integer to full node.
334 334 self._nodecache = {nullid: nullrev}
335 335 self._nodepos = None
336 336 self._compengine = 'zlib'
337 337 self._compengineopts = {}
338 338 self._maxdeltachainspan = -1
339 339 self._withsparseread = False
340 340 self._sparserevlog = False
341 341 self._srdensitythreshold = 0.50
342 342 self._srmingapsize = 262144
343 343
344 344 # Make copy of flag processors so each revlog instance can support
345 345 # custom flags.
346 346 self._flagprocessors = dict(flagutil.flagprocessors)
347 347
348 348 # 2-tuple of file handles being used for active writing.
349 349 self._writinghandles = None
350 350
351 351 self._loadindex()
352 352
353 353 def _loadindex(self):
354 354 mmapindexthreshold = None
355 355 opts = getattr(self.opener, 'options', {}) or {}
356 356
357 357 if 'revlogv2' in opts:
358 358 newversionflags = REVLOGV2 | FLAG_INLINE_DATA
359 359 elif 'revlogv1' in opts:
360 360 newversionflags = REVLOGV1 | FLAG_INLINE_DATA
361 361 if 'generaldelta' in opts:
362 362 newversionflags |= FLAG_GENERALDELTA
363 363 elif getattr(self.opener, 'options', None) is not None:
364 364 # If options provided but no 'revlog*' found, the repository
365 365 # would have no 'requires' file in it, which means we have to
366 366 # stick to the old format.
367 367 newversionflags = REVLOGV0
368 368 else:
369 369 newversionflags = REVLOG_DEFAULT_VERSION
370 370
371 371 if 'chunkcachesize' in opts:
372 372 self._chunkcachesize = opts['chunkcachesize']
373 373 if 'maxchainlen' in opts:
374 374 self._maxchainlen = opts['maxchainlen']
375 375 if 'deltabothparents' in opts:
376 376 self._deltabothparents = opts['deltabothparents']
377 377 self._lazydelta = bool(opts.get('lazydelta', True))
378 378 self._lazydeltabase = False
379 379 if self._lazydelta:
380 380 self._lazydeltabase = bool(opts.get('lazydeltabase', False))
381 381 if 'compengine' in opts:
382 382 self._compengine = opts['compengine']
383 383 if 'zlib.level' in opts:
384 384 self._compengineopts['zlib.level'] = opts['zlib.level']
385 385 if 'zstd.level' in opts:
386 386 self._compengineopts['zstd.level'] = opts['zstd.level']
387 387 if 'maxdeltachainspan' in opts:
388 388 self._maxdeltachainspan = opts['maxdeltachainspan']
389 389 if self._mmaplargeindex and 'mmapindexthreshold' in opts:
390 390 mmapindexthreshold = opts['mmapindexthreshold']
391 391 self._sparserevlog = bool(opts.get('sparse-revlog', False))
392 392 withsparseread = bool(opts.get('with-sparse-read', False))
393 393 # sparse-revlog forces sparse-read
394 394 self._withsparseread = self._sparserevlog or withsparseread
395 395 if 'sparse-read-density-threshold' in opts:
396 396 self._srdensitythreshold = opts['sparse-read-density-threshold']
397 397 if 'sparse-read-min-gap-size' in opts:
398 398 self._srmingapsize = opts['sparse-read-min-gap-size']
399 399 if opts.get('enableellipsis'):
400 400 self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor
401 401
402 402 # revlog v0 doesn't have flag processors
403 403 for flag, processor in opts.get(b'flagprocessors', {}).iteritems():
404 404 flagutil.insertflagprocessor(flag, processor, self._flagprocessors)
405 405
406 406 if self._chunkcachesize <= 0:
407 407 raise error.RevlogError(_('revlog chunk cache size %r is not '
408 408 'greater than 0') % self._chunkcachesize)
409 409 elif self._chunkcachesize & (self._chunkcachesize - 1):
410 410 raise error.RevlogError(_('revlog chunk cache size %r is not a '
411 411 'power of 2') % self._chunkcachesize)
412 412
413 413 indexdata = ''
414 414 self._initempty = True
415 415 try:
416 416 with self._indexfp() as f:
417 417 if (mmapindexthreshold is not None and
418 418 self.opener.fstat(f).st_size >= mmapindexthreshold):
419 419 # TODO: should .close() to release resources without
420 420 # relying on Python GC
421 421 indexdata = util.buffer(util.mmapread(f))
422 422 else:
423 423 indexdata = f.read()
424 424 if len(indexdata) > 0:
425 425 versionflags = versionformat_unpack(indexdata[:4])[0]
426 426 self._initempty = False
427 427 else:
428 428 versionflags = newversionflags
429 429 except IOError as inst:
430 430 if inst.errno != errno.ENOENT:
431 431 raise
432 432
433 433 versionflags = newversionflags
434 434
435 435 self.version = versionflags
436 436
437 437 flags = versionflags & ~0xFFFF
438 438 fmt = versionflags & 0xFFFF
439 439
440 440 if fmt == REVLOGV0:
441 441 if flags:
442 442 raise error.RevlogError(_('unknown flags (%#04x) in version %d '
443 443 'revlog %s') %
444 444 (flags >> 16, fmt, self.indexfile))
445 445
446 446 self._inline = False
447 447 self._generaldelta = False
448 448
449 449 elif fmt == REVLOGV1:
450 450 if flags & ~REVLOGV1_FLAGS:
451 451 raise error.RevlogError(_('unknown flags (%#04x) in version %d '
452 452 'revlog %s') %
453 453 (flags >> 16, fmt, self.indexfile))
454 454
455 455 self._inline = versionflags & FLAG_INLINE_DATA
456 456 self._generaldelta = versionflags & FLAG_GENERALDELTA
457 457
458 458 elif fmt == REVLOGV2:
459 459 if flags & ~REVLOGV2_FLAGS:
460 460 raise error.RevlogError(_('unknown flags (%#04x) in version %d '
461 461 'revlog %s') %
462 462 (flags >> 16, fmt, self.indexfile))
463 463
464 464 self._inline = versionflags & FLAG_INLINE_DATA
465 465 # generaldelta implied by version 2 revlogs.
466 466 self._generaldelta = True
467 467
468 468 else:
469 469 raise error.RevlogError(_('unknown version (%d) in revlog %s') %
470 470 (fmt, self.indexfile))
471 471 # sparse-revlog can't be on without general-delta (issue6056)
472 472 if not self._generaldelta:
473 473 self._sparserevlog = False
474 474
475 475 self._storedeltachains = True
476 476
477 477 self._io = revlogio()
478 478 if self.version == REVLOGV0:
479 479 self._io = revlogoldio()
480 480 try:
481 481 d = self._io.parseindex(indexdata, self._inline)
482 482 except (ValueError, IndexError):
483 483 raise error.RevlogError(_("index %s is corrupted") %
484 484 self.indexfile)
485 485 self.index, nodemap, self._chunkcache = d
486 486 if nodemap is not None:
487 487 self.nodemap = self._nodecache = nodemap
488 488 if not self._chunkcache:
489 489 self._chunkclear()
490 490 # revnum -> (chain-length, sum-delta-length)
491 491 self._chaininfocache = {}
492 492 # revlog header -> revlog compressor
493 493 self._decompressors = {}
494 494
495 495 @util.propertycache
496 496 def _compressor(self):
497 497 engine = util.compengines[self._compengine]
498 498 return engine.revlogcompressor(self._compengineopts)
499 499
500 500 def _indexfp(self, mode='r'):
501 501 """file object for the revlog's index file"""
502 502 args = {r'mode': mode}
503 503 if mode != 'r':
504 504 args[r'checkambig'] = self._checkambig
505 505 if mode == 'w':
506 506 args[r'atomictemp'] = True
507 507 return self.opener(self.indexfile, **args)
508 508
509 509 def _datafp(self, mode='r'):
510 510 """file object for the revlog's data file"""
511 511 return self.opener(self.datafile, mode=mode)
512 512
513 513 @contextlib.contextmanager
514 514 def _datareadfp(self, existingfp=None):
515 515 """file object suitable to read data"""
516 516 # Use explicit file handle, if given.
517 517 if existingfp is not None:
518 518 yield existingfp
519 519
520 520 # Use a file handle being actively used for writes, if available.
521 521 # There is some danger to doing this because reads will seek the
522 522 # file. However, _writeentry() performs a SEEK_END before all writes,
523 523 # so we should be safe.
524 524 elif self._writinghandles:
525 525 if self._inline:
526 526 yield self._writinghandles[0]
527 527 else:
528 528 yield self._writinghandles[1]
529 529
530 530 # Otherwise open a new file handle.
531 531 else:
532 532 if self._inline:
533 533 func = self._indexfp
534 534 else:
535 535 func = self._datafp
536 536 with func() as fp:
537 537 yield fp
538 538
539 539 def tip(self):
540 540 return self.node(len(self.index) - 1)
541 541 def __contains__(self, rev):
542 542 return 0 <= rev < len(self)
543 543 def __len__(self):
544 544 return len(self.index)
545 545 def __iter__(self):
546 546 return iter(pycompat.xrange(len(self)))
547 547 def revs(self, start=0, stop=None):
548 548 """iterate over all rev in this revlog (from start to stop)"""
549 549 return storageutil.iterrevs(len(self), start=start, stop=stop)
550 550
551 551 @util.propertycache
552 552 def nodemap(self):
553 553 if self.index:
554 554 # populate mapping down to the initial node
555 555 node0 = self.index[0][7] # get around changelog filtering
556 556 self.rev(node0)
557 557 return self._nodecache
558 558
559 559 def hasnode(self, node):
560 560 try:
561 561 self.rev(node)
562 562 return True
563 563 except KeyError:
564 564 return False
565 565
566 566 def candelta(self, baserev, rev):
567 567 """whether two revisions (baserev, rev) can be delta-ed or not"""
568 568 # Disable delta if either rev requires a content-changing flag
569 569 # processor (ex. LFS). This is because such flag processor can alter
570 570 # the rawtext content that the delta will be based on, and two clients
571 571 # could have a same revlog node with different flags (i.e. different
572 572 # rawtext contents) and the delta could be incompatible.
573 573 if ((self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS)
574 574 or (self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS)):
575 575 return False
576 576 return True
577 577
578 578 def clearcaches(self):
579 579 self._revisioncache = None
580 580 self._chainbasecache.clear()
581 581 self._chunkcache = (0, '')
582 582 self._pcache = {}
583 583
584 584 try:
585 585 # If we are using the native C version, you are in a fun case
586 586 # where self.index, self.nodemap and self._nodecaches is the same
587 587 # object.
588 588 self._nodecache.clearcaches()
589 589 except AttributeError:
590 590 self._nodecache = {nullid: nullrev}
591 591 self._nodepos = None
592 592
593 593 def rev(self, node):
594 594 try:
595 595 return self._nodecache[node]
596 596 except TypeError:
597 597 raise
598 598 except error.RevlogError:
599 599 # parsers.c radix tree lookup failed
600 600 if node == wdirid or node in wdirfilenodeids:
601 601 raise error.WdirUnsupported
602 602 raise error.LookupError(node, self.indexfile, _('no node'))
603 603 except KeyError:
604 604 # pure python cache lookup failed
605 605 n = self._nodecache
606 606 i = self.index
607 607 p = self._nodepos
608 608 if p is None:
609 609 p = len(i) - 1
610 610 else:
611 611 assert p < len(i)
612 612 for r in pycompat.xrange(p, -1, -1):
613 613 v = i[r][7]
614 614 n[v] = r
615 615 if v == node:
616 616 self._nodepos = r - 1
617 617 return r
618 618 if node == wdirid or node in wdirfilenodeids:
619 619 raise error.WdirUnsupported
620 620 raise error.LookupError(node, self.indexfile, _('no node'))
621 621
622 622 # Accessors for index entries.
623 623
624 624 # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes
625 625 # are flags.
626 626 def start(self, rev):
627 627 return int(self.index[rev][0] >> 16)
628 628
629 629 def flags(self, rev):
630 630 return self.index[rev][0] & 0xFFFF
631 631
632 632 def length(self, rev):
633 633 return self.index[rev][1]
634 634
635 635 def rawsize(self, rev):
636 636 """return the length of the uncompressed text for a given revision"""
637 637 l = self.index[rev][2]
638 638 if l >= 0:
639 639 return l
640 640
641 641 t = self.rawdata(rev)
642 642 return len(t)
643 643
644 644 def size(self, rev):
645 645 """length of non-raw text (processed by a "read" flag processor)"""
646 646 # fast path: if no "read" flag processor could change the content,
647 647 # size is rawsize. note: ELLIPSIS is known to not change the content.
648 648 flags = self.flags(rev)
649 649 if flags & (flagutil.REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0:
650 650 return self.rawsize(rev)
651 651
652 652 return len(self.revision(rev, raw=False))
653 653
654 654 def chainbase(self, rev):
655 655 base = self._chainbasecache.get(rev)
656 656 if base is not None:
657 657 return base
658 658
659 659 index = self.index
660 660 iterrev = rev
661 661 base = index[iterrev][3]
662 662 while base != iterrev:
663 663 iterrev = base
664 664 base = index[iterrev][3]
665 665
666 666 self._chainbasecache[rev] = base
667 667 return base
668 668
669 669 def linkrev(self, rev):
670 670 return self.index[rev][4]
671 671
672 672 def parentrevs(self, rev):
673 673 try:
674 674 entry = self.index[rev]
675 675 except IndexError:
676 676 if rev == wdirrev:
677 677 raise error.WdirUnsupported
678 678 raise
679 679
680 680 return entry[5], entry[6]
681 681
682 682 # fast parentrevs(rev) where rev isn't filtered
683 683 _uncheckedparentrevs = parentrevs
684 684
685 685 def node(self, rev):
686 686 try:
687 687 return self.index[rev][7]
688 688 except IndexError:
689 689 if rev == wdirrev:
690 690 raise error.WdirUnsupported
691 691 raise
692 692
693 693 # Derived from index values.
694 694
695 695 def end(self, rev):
696 696 return self.start(rev) + self.length(rev)
697 697
698 698 def parents(self, node):
699 699 i = self.index
700 700 d = i[self.rev(node)]
701 701 return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline
702 702
703 703 def chainlen(self, rev):
704 704 return self._chaininfo(rev)[0]
705 705
706 706 def _chaininfo(self, rev):
707 707 chaininfocache = self._chaininfocache
708 708 if rev in chaininfocache:
709 709 return chaininfocache[rev]
710 710 index = self.index
711 711 generaldelta = self._generaldelta
712 712 iterrev = rev
713 713 e = index[iterrev]
714 714 clen = 0
715 715 compresseddeltalen = 0
716 716 while iterrev != e[3]:
717 717 clen += 1
718 718 compresseddeltalen += e[1]
719 719 if generaldelta:
720 720 iterrev = e[3]
721 721 else:
722 722 iterrev -= 1
723 723 if iterrev in chaininfocache:
724 724 t = chaininfocache[iterrev]
725 725 clen += t[0]
726 726 compresseddeltalen += t[1]
727 727 break
728 728 e = index[iterrev]
729 729 else:
730 730 # Add text length of base since decompressing that also takes
731 731 # work. For cache hits the length is already included.
732 732 compresseddeltalen += e[1]
733 733 r = (clen, compresseddeltalen)
734 734 chaininfocache[rev] = r
735 735 return r
736 736
737 737 def _deltachain(self, rev, stoprev=None):
738 738 """Obtain the delta chain for a revision.
739 739
740 740 ``stoprev`` specifies a revision to stop at. If not specified, we
741 741 stop at the base of the chain.
742 742
743 743 Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of
744 744 revs in ascending order and ``stopped`` is a bool indicating whether
745 745 ``stoprev`` was hit.
746 746 """
747 747 # Try C implementation.
748 748 try:
749 749 return self.index.deltachain(rev, stoprev, self._generaldelta)
750 750 except AttributeError:
751 751 pass
752 752
753 753 chain = []
754 754
755 755 # Alias to prevent attribute lookup in tight loop.
756 756 index = self.index
757 757 generaldelta = self._generaldelta
758 758
759 759 iterrev = rev
760 760 e = index[iterrev]
761 761 while iterrev != e[3] and iterrev != stoprev:
762 762 chain.append(iterrev)
763 763 if generaldelta:
764 764 iterrev = e[3]
765 765 else:
766 766 iterrev -= 1
767 767 e = index[iterrev]
768 768
769 769 if iterrev == stoprev:
770 770 stopped = True
771 771 else:
772 772 chain.append(iterrev)
773 773 stopped = False
774 774
775 775 chain.reverse()
776 776 return chain, stopped
777 777
778 778 def ancestors(self, revs, stoprev=0, inclusive=False):
779 779 """Generate the ancestors of 'revs' in reverse revision order.
780 780 Does not generate revs lower than stoprev.
781 781
782 782 See the documentation for ancestor.lazyancestors for more details."""
783 783
784 784 # first, make sure start revisions aren't filtered
785 785 revs = list(revs)
786 786 checkrev = self.node
787 787 for r in revs:
788 788 checkrev(r)
789 789 # and we're sure ancestors aren't filtered as well
790 790
791 791 if rustancestor is not None:
792 792 lazyancestors = rustancestor.LazyAncestors
793 793 arg = self.index
794 794 elif util.safehasattr(parsers, 'rustlazyancestors'):
795 795 lazyancestors = ancestor.rustlazyancestors
796 796 arg = self.index
797 797 else:
798 798 lazyancestors = ancestor.lazyancestors
799 799 arg = self._uncheckedparentrevs
800 800 return lazyancestors(arg, revs, stoprev=stoprev, inclusive=inclusive)
801 801
802 802 def descendants(self, revs):
803 803 return dagop.descendantrevs(revs, self.revs, self.parentrevs)
804 804
805 805 def findcommonmissing(self, common=None, heads=None):
806 806 """Return a tuple of the ancestors of common and the ancestors of heads
807 807 that are not ancestors of common. In revset terminology, we return the
808 808 tuple:
809 809
810 810 ::common, (::heads) - (::common)
811 811
812 812 The list is sorted by revision number, meaning it is
813 813 topologically sorted.
814 814
815 815 'heads' and 'common' are both lists of node IDs. If heads is
816 816 not supplied, uses all of the revlog's heads. If common is not
817 817 supplied, uses nullid."""
818 818 if common is None:
819 819 common = [nullid]
820 820 if heads is None:
821 821 heads = self.heads()
822 822
823 823 common = [self.rev(n) for n in common]
824 824 heads = [self.rev(n) for n in heads]
825 825
826 826 # we want the ancestors, but inclusive
827 827 class lazyset(object):
828 828 def __init__(self, lazyvalues):
829 829 self.addedvalues = set()
830 830 self.lazyvalues = lazyvalues
831 831
832 832 def __contains__(self, value):
833 833 return value in self.addedvalues or value in self.lazyvalues
834 834
835 835 def __iter__(self):
836 836 added = self.addedvalues
837 837 for r in added:
838 838 yield r
839 839 for r in self.lazyvalues:
840 840 if not r in added:
841 841 yield r
842 842
843 843 def add(self, value):
844 844 self.addedvalues.add(value)
845 845
846 846 def update(self, values):
847 847 self.addedvalues.update(values)
848 848
849 849 has = lazyset(self.ancestors(common))
850 850 has.add(nullrev)
851 851 has.update(common)
852 852
853 853 # take all ancestors from heads that aren't in has
854 854 missing = set()
855 855 visit = collections.deque(r for r in heads if r not in has)
856 856 while visit:
857 857 r = visit.popleft()
858 858 if r in missing:
859 859 continue
860 860 else:
861 861 missing.add(r)
862 862 for p in self.parentrevs(r):
863 863 if p not in has:
864 864 visit.append(p)
865 865 missing = list(missing)
866 866 missing.sort()
867 867 return has, [self.node(miss) for miss in missing]
868 868
869 869 def incrementalmissingrevs(self, common=None):
870 870 """Return an object that can be used to incrementally compute the
871 871 revision numbers of the ancestors of arbitrary sets that are not
872 872 ancestors of common. This is an ancestor.incrementalmissingancestors
873 873 object.
874 874
875 875 'common' is a list of revision numbers. If common is not supplied, uses
876 876 nullrev.
877 877 """
878 878 if common is None:
879 879 common = [nullrev]
880 880
881 881 if rustancestor is not None:
882 882 return rustancestor.MissingAncestors(self.index, common)
883 883 return ancestor.incrementalmissingancestors(self.parentrevs, common)
884 884
885 885 def findmissingrevs(self, common=None, heads=None):
886 886 """Return the revision numbers of the ancestors of heads that
887 887 are not ancestors of common.
888 888
889 889 More specifically, return a list of revision numbers corresponding to
890 890 nodes N such that every N satisfies the following constraints:
891 891
892 892 1. N is an ancestor of some node in 'heads'
893 893 2. N is not an ancestor of any node in 'common'
894 894
895 895 The list is sorted by revision number, meaning it is
896 896 topologically sorted.
897 897
898 898 'heads' and 'common' are both lists of revision numbers. If heads is
899 899 not supplied, uses all of the revlog's heads. If common is not
900 900 supplied, uses nullid."""
901 901 if common is None:
902 902 common = [nullrev]
903 903 if heads is None:
904 904 heads = self.headrevs()
905 905
906 906 inc = self.incrementalmissingrevs(common=common)
907 907 return inc.missingancestors(heads)
908 908
909 909 def findmissing(self, common=None, heads=None):
910 910 """Return the ancestors of heads that are not ancestors of common.
911 911
912 912 More specifically, return a list of nodes N such that every N
913 913 satisfies the following constraints:
914 914
915 915 1. N is an ancestor of some node in 'heads'
916 916 2. N is not an ancestor of any node in 'common'
917 917
918 918 The list is sorted by revision number, meaning it is
919 919 topologically sorted.
920 920
921 921 'heads' and 'common' are both lists of node IDs. If heads is
922 922 not supplied, uses all of the revlog's heads. If common is not
923 923 supplied, uses nullid."""
924 924 if common is None:
925 925 common = [nullid]
926 926 if heads is None:
927 927 heads = self.heads()
928 928
929 929 common = [self.rev(n) for n in common]
930 930 heads = [self.rev(n) for n in heads]
931 931
932 932 inc = self.incrementalmissingrevs(common=common)
933 933 return [self.node(r) for r in inc.missingancestors(heads)]
934 934
935 935 def nodesbetween(self, roots=None, heads=None):
936 936 """Return a topological path from 'roots' to 'heads'.
937 937
938 938 Return a tuple (nodes, outroots, outheads) where 'nodes' is a
939 939 topologically sorted list of all nodes N that satisfy both of
940 940 these constraints:
941 941
942 942 1. N is a descendant of some node in 'roots'
943 943 2. N is an ancestor of some node in 'heads'
944 944
945 945 Every node is considered to be both a descendant and an ancestor
946 946 of itself, so every reachable node in 'roots' and 'heads' will be
947 947 included in 'nodes'.
948 948
949 949 'outroots' is the list of reachable nodes in 'roots', i.e., the
950 950 subset of 'roots' that is returned in 'nodes'. Likewise,
951 951 'outheads' is the subset of 'heads' that is also in 'nodes'.
952 952
953 953 'roots' and 'heads' are both lists of node IDs. If 'roots' is
954 954 unspecified, uses nullid as the only root. If 'heads' is
955 955 unspecified, uses list of all of the revlog's heads."""
956 956 nonodes = ([], [], [])
957 957 if roots is not None:
958 958 roots = list(roots)
959 959 if not roots:
960 960 return nonodes
961 961 lowestrev = min([self.rev(n) for n in roots])
962 962 else:
963 963 roots = [nullid] # Everybody's a descendant of nullid
964 964 lowestrev = nullrev
965 965 if (lowestrev == nullrev) and (heads is None):
966 966 # We want _all_ the nodes!
967 967 return ([self.node(r) for r in self], [nullid], list(self.heads()))
968 968 if heads is None:
969 969 # All nodes are ancestors, so the latest ancestor is the last
970 970 # node.
971 971 highestrev = len(self) - 1
972 972 # Set ancestors to None to signal that every node is an ancestor.
973 973 ancestors = None
974 974 # Set heads to an empty dictionary for later discovery of heads
975 975 heads = {}
976 976 else:
977 977 heads = list(heads)
978 978 if not heads:
979 979 return nonodes
980 980 ancestors = set()
981 981 # Turn heads into a dictionary so we can remove 'fake' heads.
982 982 # Also, later we will be using it to filter out the heads we can't
983 983 # find from roots.
984 984 heads = dict.fromkeys(heads, False)
985 985 # Start at the top and keep marking parents until we're done.
986 986 nodestotag = set(heads)
987 987 # Remember where the top was so we can use it as a limit later.
988 988 highestrev = max([self.rev(n) for n in nodestotag])
989 989 while nodestotag:
990 990 # grab a node to tag
991 991 n = nodestotag.pop()
992 992 # Never tag nullid
993 993 if n == nullid:
994 994 continue
995 995 # A node's revision number represents its place in a
996 996 # topologically sorted list of nodes.
997 997 r = self.rev(n)
998 998 if r >= lowestrev:
999 999 if n not in ancestors:
1000 1000 # If we are possibly a descendant of one of the roots
1001 1001 # and we haven't already been marked as an ancestor
1002 1002 ancestors.add(n) # Mark as ancestor
1003 1003 # Add non-nullid parents to list of nodes to tag.
1004 1004 nodestotag.update([p for p in self.parents(n) if
1005 1005 p != nullid])
1006 1006 elif n in heads: # We've seen it before, is it a fake head?
1007 1007 # So it is, real heads should not be the ancestors of
1008 1008 # any other heads.
1009 1009 heads.pop(n)
1010 1010 if not ancestors:
1011 1011 return nonodes
1012 1012 # Now that we have our set of ancestors, we want to remove any
1013 1013 # roots that are not ancestors.
1014 1014
1015 1015 # If one of the roots was nullid, everything is included anyway.
1016 1016 if lowestrev > nullrev:
1017 1017 # But, since we weren't, let's recompute the lowest rev to not
1018 1018 # include roots that aren't ancestors.
1019 1019
1020 1020 # Filter out roots that aren't ancestors of heads
1021 1021 roots = [root for root in roots if root in ancestors]
1022 1022 # Recompute the lowest revision
1023 1023 if roots:
1024 1024 lowestrev = min([self.rev(root) for root in roots])
1025 1025 else:
1026 1026 # No more roots? Return empty list
1027 1027 return nonodes
1028 1028 else:
1029 1029 # We are descending from nullid, and don't need to care about
1030 1030 # any other roots.
1031 1031 lowestrev = nullrev
1032 1032 roots = [nullid]
1033 1033 # Transform our roots list into a set.
1034 1034 descendants = set(roots)
1035 1035 # Also, keep the original roots so we can filter out roots that aren't
1036 1036 # 'real' roots (i.e. are descended from other roots).
1037 1037 roots = descendants.copy()
1038 1038 # Our topologically sorted list of output nodes.
1039 1039 orderedout = []
1040 1040 # Don't start at nullid since we don't want nullid in our output list,
1041 1041 # and if nullid shows up in descendants, empty parents will look like
1042 1042 # they're descendants.
1043 1043 for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1):
1044 1044 n = self.node(r)
1045 1045 isdescendant = False
1046 1046 if lowestrev == nullrev: # Everybody is a descendant of nullid
1047 1047 isdescendant = True
1048 1048 elif n in descendants:
1049 1049 # n is already a descendant
1050 1050 isdescendant = True
1051 1051 # This check only needs to be done here because all the roots
1052 1052 # will start being marked is descendants before the loop.
1053 1053 if n in roots:
1054 1054 # If n was a root, check if it's a 'real' root.
1055 1055 p = tuple(self.parents(n))
1056 1056 # If any of its parents are descendants, it's not a root.
1057 1057 if (p[0] in descendants) or (p[1] in descendants):
1058 1058 roots.remove(n)
1059 1059 else:
1060 1060 p = tuple(self.parents(n))
1061 1061 # A node is a descendant if either of its parents are
1062 1062 # descendants. (We seeded the dependents list with the roots
1063 1063 # up there, remember?)
1064 1064 if (p[0] in descendants) or (p[1] in descendants):
1065 1065 descendants.add(n)
1066 1066 isdescendant = True
1067 1067 if isdescendant and ((ancestors is None) or (n in ancestors)):
1068 1068 # Only include nodes that are both descendants and ancestors.
1069 1069 orderedout.append(n)
1070 1070 if (ancestors is not None) and (n in heads):
1071 1071 # We're trying to figure out which heads are reachable
1072 1072 # from roots.
1073 1073 # Mark this head as having been reached
1074 1074 heads[n] = True
1075 1075 elif ancestors is None:
1076 1076 # Otherwise, we're trying to discover the heads.
1077 1077 # Assume this is a head because if it isn't, the next step
1078 1078 # will eventually remove it.
1079 1079 heads[n] = True
1080 1080 # But, obviously its parents aren't.
1081 1081 for p in self.parents(n):
1082 1082 heads.pop(p, None)
1083 1083 heads = [head for head, flag in heads.iteritems() if flag]
1084 1084 roots = list(roots)
1085 1085 assert orderedout
1086 1086 assert roots
1087 1087 assert heads
1088 1088 return (orderedout, roots, heads)
1089 1089
1090 1090 def headrevs(self, revs=None):
1091 1091 if revs is None:
1092 1092 try:
1093 1093 return self.index.headrevs()
1094 1094 except AttributeError:
1095 1095 return self._headrevs()
1096 1096 if rustdagop is not None:
1097 1097 return rustdagop.headrevs(self.index, revs)
1098 1098 return dagop.headrevs(revs, self._uncheckedparentrevs)
1099 1099
1100 1100 def computephases(self, roots):
1101 1101 return self.index.computephasesmapsets(roots)
1102 1102
1103 1103 def _headrevs(self):
1104 1104 count = len(self)
1105 1105 if not count:
1106 1106 return [nullrev]
1107 1107 # we won't iter over filtered rev so nobody is a head at start
1108 1108 ishead = [0] * (count + 1)
1109 1109 index = self.index
1110 1110 for r in self:
1111 1111 ishead[r] = 1 # I may be an head
1112 1112 e = index[r]
1113 1113 ishead[e[5]] = ishead[e[6]] = 0 # my parent are not
1114 1114 return [r for r, val in enumerate(ishead) if val]
1115 1115
1116 1116 def heads(self, start=None, stop=None):
1117 1117 """return the list of all nodes that have no children
1118 1118
1119 1119 if start is specified, only heads that are descendants of
1120 1120 start will be returned
1121 1121 if stop is specified, it will consider all the revs from stop
1122 1122 as if they had no children
1123 1123 """
1124 1124 if start is None and stop is None:
1125 1125 if not len(self):
1126 1126 return [nullid]
1127 1127 return [self.node(r) for r in self.headrevs()]
1128 1128
1129 1129 if start is None:
1130 1130 start = nullrev
1131 1131 else:
1132 1132 start = self.rev(start)
1133 1133
1134 1134 stoprevs = set(self.rev(n) for n in stop or [])
1135 1135
1136 1136 revs = dagop.headrevssubset(self.revs, self.parentrevs, startrev=start,
1137 1137 stoprevs=stoprevs)
1138 1138
1139 1139 return [self.node(rev) for rev in revs]
1140 1140
1141 1141 def children(self, node):
1142 1142 """find the children of a given node"""
1143 1143 c = []
1144 1144 p = self.rev(node)
1145 1145 for r in self.revs(start=p + 1):
1146 1146 prevs = [pr for pr in self.parentrevs(r) if pr != nullrev]
1147 1147 if prevs:
1148 1148 for pr in prevs:
1149 1149 if pr == p:
1150 1150 c.append(self.node(r))
1151 1151 elif p == nullrev:
1152 1152 c.append(self.node(r))
1153 1153 return c
1154 1154
1155 1155 def commonancestorsheads(self, a, b):
1156 1156 """calculate all the heads of the common ancestors of nodes a and b"""
1157 1157 a, b = self.rev(a), self.rev(b)
1158 1158 ancs = self._commonancestorsheads(a, b)
1159 1159 return pycompat.maplist(self.node, ancs)
1160 1160
1161 1161 def _commonancestorsheads(self, *revs):
1162 1162 """calculate all the heads of the common ancestors of revs"""
1163 1163 try:
1164 1164 ancs = self.index.commonancestorsheads(*revs)
1165 1165 except (AttributeError, OverflowError): # C implementation failed
1166 1166 ancs = ancestor.commonancestorsheads(self.parentrevs, *revs)
1167 1167 return ancs
1168 1168
1169 1169 def isancestor(self, a, b):
1170 1170 """return True if node a is an ancestor of node b
1171 1171
1172 1172 A revision is considered an ancestor of itself."""
1173 1173 a, b = self.rev(a), self.rev(b)
1174 1174 return self.isancestorrev(a, b)
1175 1175
1176 1176 def isancestorrev(self, a, b):
1177 1177 """return True if revision a is an ancestor of revision b
1178 1178
1179 1179 A revision is considered an ancestor of itself.
1180 1180
1181 1181 The implementation of this is trivial but the use of
1182 1182 reachableroots is not."""
1183 1183 if a == nullrev:
1184 1184 return True
1185 1185 elif a == b:
1186 1186 return True
1187 1187 elif a > b:
1188 1188 return False
1189 1189 return bool(self.reachableroots(a, [b], [a], includepath=False))
1190 1190
1191 1191 def reachableroots(self, minroot, heads, roots, includepath=False):
1192 1192 """return (heads(::<roots> and <roots>::<heads>))
1193 1193
1194 1194 If includepath is True, return (<roots>::<heads>)."""
1195 1195 try:
1196 1196 return self.index.reachableroots2(minroot, heads, roots,
1197 1197 includepath)
1198 1198 except AttributeError:
1199 1199 return dagop._reachablerootspure(self.parentrevs,
1200 1200 minroot, roots, heads, includepath)
1201 1201
1202 1202 def ancestor(self, a, b):
1203 1203 """calculate the "best" common ancestor of nodes a and b"""
1204 1204
1205 1205 a, b = self.rev(a), self.rev(b)
1206 1206 try:
1207 1207 ancs = self.index.ancestors(a, b)
1208 1208 except (AttributeError, OverflowError):
1209 1209 ancs = ancestor.ancestors(self.parentrevs, a, b)
1210 1210 if ancs:
1211 1211 # choose a consistent winner when there's a tie
1212 1212 return min(map(self.node, ancs))
1213 1213 return nullid
1214 1214
1215 1215 def _match(self, id):
1216 1216 if isinstance(id, int):
1217 1217 # rev
1218 1218 return self.node(id)
1219 1219 if len(id) == 20:
1220 1220 # possibly a binary node
1221 1221 # odds of a binary node being all hex in ASCII are 1 in 10**25
1222 1222 try:
1223 1223 node = id
1224 1224 self.rev(node) # quick search the index
1225 1225 return node
1226 1226 except error.LookupError:
1227 1227 pass # may be partial hex id
1228 1228 try:
1229 1229 # str(rev)
1230 1230 rev = int(id)
1231 1231 if "%d" % rev != id:
1232 1232 raise ValueError
1233 1233 if rev < 0:
1234 1234 rev = len(self) + rev
1235 1235 if rev < 0 or rev >= len(self):
1236 1236 raise ValueError
1237 1237 return self.node(rev)
1238 1238 except (ValueError, OverflowError):
1239 1239 pass
1240 1240 if len(id) == 40:
1241 1241 try:
1242 1242 # a full hex nodeid?
1243 1243 node = bin(id)
1244 1244 self.rev(node)
1245 1245 return node
1246 1246 except (TypeError, error.LookupError):
1247 1247 pass
1248 1248
1249 1249 def _partialmatch(self, id):
1250 1250 # we don't care wdirfilenodeids as they should be always full hash
1251 1251 maybewdir = wdirhex.startswith(id)
1252 1252 try:
1253 1253 partial = self.index.partialmatch(id)
1254 1254 if partial and self.hasnode(partial):
1255 1255 if maybewdir:
1256 1256 # single 'ff...' match in radix tree, ambiguous with wdir
1257 1257 raise error.RevlogError
1258 1258 return partial
1259 1259 if maybewdir:
1260 1260 # no 'ff...' match in radix tree, wdir identified
1261 1261 raise error.WdirUnsupported
1262 1262 return None
1263 1263 except error.RevlogError:
1264 1264 # parsers.c radix tree lookup gave multiple matches
1265 1265 # fast path: for unfiltered changelog, radix tree is accurate
1266 1266 if not getattr(self, 'filteredrevs', None):
1267 1267 raise error.AmbiguousPrefixLookupError(
1268 1268 id, self.indexfile, _('ambiguous identifier'))
1269 1269 # fall through to slow path that filters hidden revisions
1270 1270 except (AttributeError, ValueError):
1271 1271 # we are pure python, or key was too short to search radix tree
1272 1272 pass
1273 1273
1274 1274 if id in self._pcache:
1275 1275 return self._pcache[id]
1276 1276
1277 1277 if len(id) <= 40:
1278 1278 try:
1279 1279 # hex(node)[:...]
1280 1280 l = len(id) // 2 # grab an even number of digits
1281 1281 prefix = bin(id[:l * 2])
1282 1282 nl = [e[7] for e in self.index if e[7].startswith(prefix)]
1283 1283 nl = [n for n in nl if hex(n).startswith(id) and
1284 1284 self.hasnode(n)]
1285 1285 if nullhex.startswith(id):
1286 1286 nl.append(nullid)
1287 1287 if len(nl) > 0:
1288 1288 if len(nl) == 1 and not maybewdir:
1289 1289 self._pcache[id] = nl[0]
1290 1290 return nl[0]
1291 1291 raise error.AmbiguousPrefixLookupError(
1292 1292 id, self.indexfile, _('ambiguous identifier'))
1293 1293 if maybewdir:
1294 1294 raise error.WdirUnsupported
1295 1295 return None
1296 1296 except TypeError:
1297 1297 pass
1298 1298
1299 1299 def lookup(self, id):
1300 1300 """locate a node based on:
1301 1301 - revision number or str(revision number)
1302 1302 - nodeid or subset of hex nodeid
1303 1303 """
1304 1304 n = self._match(id)
1305 1305 if n is not None:
1306 1306 return n
1307 1307 n = self._partialmatch(id)
1308 1308 if n:
1309 1309 return n
1310 1310
1311 1311 raise error.LookupError(id, self.indexfile, _('no match found'))
1312 1312
1313 1313 def shortest(self, node, minlength=1):
1314 1314 """Find the shortest unambiguous prefix that matches node."""
1315 1315 def isvalid(prefix):
1316 1316 try:
1317 1317 matchednode = self._partialmatch(prefix)
1318 1318 except error.AmbiguousPrefixLookupError:
1319 1319 return False
1320 1320 except error.WdirUnsupported:
1321 1321 # single 'ff...' match
1322 1322 return True
1323 1323 if matchednode is None:
1324 1324 raise error.LookupError(node, self.indexfile, _('no node'))
1325 1325 return True
1326 1326
1327 1327 def maybewdir(prefix):
1328 1328 return all(c == 'f' for c in pycompat.iterbytestr(prefix))
1329 1329
1330 1330 hexnode = hex(node)
1331 1331
1332 1332 def disambiguate(hexnode, minlength):
1333 1333 """Disambiguate against wdirid."""
1334 1334 for length in range(minlength, 41):
1335 1335 prefix = hexnode[:length]
1336 1336 if not maybewdir(prefix):
1337 1337 return prefix
1338 1338
1339 1339 if not getattr(self, 'filteredrevs', None):
1340 1340 try:
1341 1341 length = max(self.index.shortest(node), minlength)
1342 1342 return disambiguate(hexnode, length)
1343 1343 except error.RevlogError:
1344 1344 if node != wdirid:
1345 1345 raise error.LookupError(node, self.indexfile, _('no node'))
1346 1346 except AttributeError:
1347 1347 # Fall through to pure code
1348 1348 pass
1349 1349
1350 1350 if node == wdirid:
1351 1351 for length in range(minlength, 41):
1352 1352 prefix = hexnode[:length]
1353 1353 if isvalid(prefix):
1354 1354 return prefix
1355 1355
1356 1356 for length in range(minlength, 41):
1357 1357 prefix = hexnode[:length]
1358 1358 if isvalid(prefix):
1359 1359 return disambiguate(hexnode, length)
1360 1360
1361 1361 def cmp(self, node, text):
1362 1362 """compare text with a given file revision
1363 1363
1364 1364 returns True if text is different than what is stored.
1365 1365 """
1366 1366 p1, p2 = self.parents(node)
1367 1367 return storageutil.hashrevisionsha1(text, p1, p2) != node
1368 1368
1369 1369 def _cachesegment(self, offset, data):
1370 1370 """Add a segment to the revlog cache.
1371 1371
1372 1372 Accepts an absolute offset and the data that is at that location.
1373 1373 """
1374 1374 o, d = self._chunkcache
1375 1375 # try to add to existing cache
1376 1376 if o + len(d) == offset and len(d) + len(data) < _chunksize:
1377 1377 self._chunkcache = o, d + data
1378 1378 else:
1379 1379 self._chunkcache = offset, data
1380 1380
1381 1381 def _readsegment(self, offset, length, df=None):
1382 1382 """Load a segment of raw data from the revlog.
1383 1383
1384 1384 Accepts an absolute offset, length to read, and an optional existing
1385 1385 file handle to read from.
1386 1386
1387 1387 If an existing file handle is passed, it will be seeked and the
1388 1388 original seek position will NOT be restored.
1389 1389
1390 1390 Returns a str or buffer of raw byte data.
1391 1391
1392 1392 Raises if the requested number of bytes could not be read.
1393 1393 """
1394 1394 # Cache data both forward and backward around the requested
1395 1395 # data, in a fixed size window. This helps speed up operations
1396 1396 # involving reading the revlog backwards.
1397 1397 cachesize = self._chunkcachesize
1398 1398 realoffset = offset & ~(cachesize - 1)
1399 1399 reallength = (((offset + length + cachesize) & ~(cachesize - 1))
1400 1400 - realoffset)
1401 1401 with self._datareadfp(df) as df:
1402 1402 df.seek(realoffset)
1403 1403 d = df.read(reallength)
1404 1404
1405 1405 self._cachesegment(realoffset, d)
1406 1406 if offset != realoffset or reallength != length:
1407 1407 startoffset = offset - realoffset
1408 1408 if len(d) - startoffset < length:
1409 1409 raise error.RevlogError(
1410 1410 _('partial read of revlog %s; expected %d bytes from '
1411 1411 'offset %d, got %d') %
1412 1412 (self.indexfile if self._inline else self.datafile,
1413 1413 length, realoffset, len(d) - startoffset))
1414 1414
1415 1415 return util.buffer(d, startoffset, length)
1416 1416
1417 1417 if len(d) < length:
1418 1418 raise error.RevlogError(
1419 1419 _('partial read of revlog %s; expected %d bytes from offset '
1420 1420 '%d, got %d') %
1421 1421 (self.indexfile if self._inline else self.datafile,
1422 1422 length, offset, len(d)))
1423 1423
1424 1424 return d
1425 1425
1426 1426 def _getsegment(self, offset, length, df=None):
1427 1427 """Obtain a segment of raw data from the revlog.
1428 1428
1429 1429 Accepts an absolute offset, length of bytes to obtain, and an
1430 1430 optional file handle to the already-opened revlog. If the file
1431 1431 handle is used, it's original seek position will not be preserved.
1432 1432
1433 1433 Requests for data may be returned from a cache.
1434 1434
1435 1435 Returns a str or a buffer instance of raw byte data.
1436 1436 """
1437 1437 o, d = self._chunkcache
1438 1438 l = len(d)
1439 1439
1440 1440 # is it in the cache?
1441 1441 cachestart = offset - o
1442 1442 cacheend = cachestart + length
1443 1443 if cachestart >= 0 and cacheend <= l:
1444 1444 if cachestart == 0 and cacheend == l:
1445 1445 return d # avoid a copy
1446 1446 return util.buffer(d, cachestart, cacheend - cachestart)
1447 1447
1448 1448 return self._readsegment(offset, length, df=df)
1449 1449
1450 1450 def _getsegmentforrevs(self, startrev, endrev, df=None):
1451 1451 """Obtain a segment of raw data corresponding to a range of revisions.
1452 1452
1453 1453 Accepts the start and end revisions and an optional already-open
1454 1454 file handle to be used for reading. If the file handle is read, its
1455 1455 seek position will not be preserved.
1456 1456
1457 1457 Requests for data may be satisfied by a cache.
1458 1458
1459 1459 Returns a 2-tuple of (offset, data) for the requested range of
1460 1460 revisions. Offset is the integer offset from the beginning of the
1461 1461 revlog and data is a str or buffer of the raw byte data.
1462 1462
1463 1463 Callers will need to call ``self.start(rev)`` and ``self.length(rev)``
1464 1464 to determine where each revision's data begins and ends.
1465 1465 """
1466 1466 # Inlined self.start(startrev) & self.end(endrev) for perf reasons
1467 1467 # (functions are expensive).
1468 1468 index = self.index
1469 1469 istart = index[startrev]
1470 1470 start = int(istart[0] >> 16)
1471 1471 if startrev == endrev:
1472 1472 end = start + istart[1]
1473 1473 else:
1474 1474 iend = index[endrev]
1475 1475 end = int(iend[0] >> 16) + iend[1]
1476 1476
1477 1477 if self._inline:
1478 1478 start += (startrev + 1) * self._io.size
1479 1479 end += (endrev + 1) * self._io.size
1480 1480 length = end - start
1481 1481
1482 1482 return start, self._getsegment(start, length, df=df)
1483 1483
1484 1484 def _chunk(self, rev, df=None):
1485 1485 """Obtain a single decompressed chunk for a revision.
1486 1486
1487 1487 Accepts an integer revision and an optional already-open file handle
1488 1488 to be used for reading. If used, the seek position of the file will not
1489 1489 be preserved.
1490 1490
1491 1491 Returns a str holding uncompressed data for the requested revision.
1492 1492 """
1493 1493 return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1])
1494 1494
1495 1495 def _chunks(self, revs, df=None, targetsize=None):
1496 1496 """Obtain decompressed chunks for the specified revisions.
1497 1497
1498 1498 Accepts an iterable of numeric revisions that are assumed to be in
1499 1499 ascending order. Also accepts an optional already-open file handle
1500 1500 to be used for reading. If used, the seek position of the file will
1501 1501 not be preserved.
1502 1502
1503 1503 This function is similar to calling ``self._chunk()`` multiple times,
1504 1504 but is faster.
1505 1505
1506 1506 Returns a list with decompressed data for each requested revision.
1507 1507 """
1508 1508 if not revs:
1509 1509 return []
1510 1510 start = self.start
1511 1511 length = self.length
1512 1512 inline = self._inline
1513 1513 iosize = self._io.size
1514 1514 buffer = util.buffer
1515 1515
1516 1516 l = []
1517 1517 ladd = l.append
1518 1518
1519 1519 if not self._withsparseread:
1520 1520 slicedchunks = (revs,)
1521 1521 else:
1522 1522 slicedchunks = deltautil.slicechunk(self, revs,
1523 1523 targetsize=targetsize)
1524 1524
1525 1525 for revschunk in slicedchunks:
1526 1526 firstrev = revschunk[0]
1527 1527 # Skip trailing revisions with empty diff
1528 1528 for lastrev in revschunk[::-1]:
1529 1529 if length(lastrev) != 0:
1530 1530 break
1531 1531
1532 1532 try:
1533 1533 offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df)
1534 1534 except OverflowError:
1535 1535 # issue4215 - we can't cache a run of chunks greater than
1536 1536 # 2G on Windows
1537 1537 return [self._chunk(rev, df=df) for rev in revschunk]
1538 1538
1539 1539 decomp = self.decompress
1540 1540 for rev in revschunk:
1541 1541 chunkstart = start(rev)
1542 1542 if inline:
1543 1543 chunkstart += (rev + 1) * iosize
1544 1544 chunklength = length(rev)
1545 1545 ladd(decomp(buffer(data, chunkstart - offset, chunklength)))
1546 1546
1547 1547 return l
1548 1548
1549 1549 def _chunkclear(self):
1550 1550 """Clear the raw chunk cache."""
1551 1551 self._chunkcache = (0, '')
1552 1552
1553 1553 def deltaparent(self, rev):
1554 1554 """return deltaparent of the given revision"""
1555 1555 base = self.index[rev][3]
1556 1556 if base == rev:
1557 1557 return nullrev
1558 1558 elif self._generaldelta:
1559 1559 return base
1560 1560 else:
1561 1561 return rev - 1
1562 1562
1563 1563 def issnapshot(self, rev):
1564 1564 """tells whether rev is a snapshot
1565 1565 """
1566 1566 if not self._sparserevlog:
1567 1567 return self.deltaparent(rev) == nullrev
1568 1568 elif util.safehasattr(self.index, 'issnapshot'):
1569 1569 # directly assign the method to cache the testing and access
1570 1570 self.issnapshot = self.index.issnapshot
1571 1571 return self.issnapshot(rev)
1572 1572 if rev == nullrev:
1573 1573 return True
1574 1574 entry = self.index[rev]
1575 1575 base = entry[3]
1576 1576 if base == rev:
1577 1577 return True
1578 1578 if base == nullrev:
1579 1579 return True
1580 1580 p1 = entry[5]
1581 1581 p2 = entry[6]
1582 1582 if base == p1 or base == p2:
1583 1583 return False
1584 1584 return self.issnapshot(base)
1585 1585
1586 1586 def snapshotdepth(self, rev):
1587 1587 """number of snapshot in the chain before this one"""
1588 1588 if not self.issnapshot(rev):
1589 1589 raise error.ProgrammingError('revision %d not a snapshot')
1590 1590 return len(self._deltachain(rev)[0]) - 1
1591 1591
1592 1592 def revdiff(self, rev1, rev2):
1593 1593 """return or calculate a delta between two revisions
1594 1594
1595 1595 The delta calculated is in binary form and is intended to be written to
1596 1596 revlog data directly. So this function needs raw revision data.
1597 1597 """
1598 1598 if rev1 != nullrev and self.deltaparent(rev2) == rev1:
1599 1599 return bytes(self._chunk(rev2))
1600 1600
1601 1601 return mdiff.textdiff(self.rawdata(rev1),
1602 1602 self.rawdata(rev2))
1603 1603
1604 1604 def revision(self, nodeorrev, _df=None, raw=False):
1605 1605 """return an uncompressed revision of a given node or revision
1606 1606 number.
1607 1607
1608 1608 _df - an existing file handle to read from. (internal-only)
1609 1609 raw - an optional argument specifying if the revision data is to be
1610 1610 treated as raw data when applying flag transforms. 'raw' should be set
1611 1611 to True when generating changegroups or in debug commands.
1612 1612 """
1613 1613 if raw:
1614 1614 msg = ('revlog.revision(..., raw=True) is deprecated, '
1615 1615 'use revlog.rawdata(...)')
1616 1616 util.nouideprecwarn(msg, '5.2', stacklevel=2)
1617 1617 return self._revisiondata(nodeorrev, _df, raw=raw)
1618 1618
1619 def sidedata(self, nodeorrev, _df=None):
1620 """a map of extra data related to the changeset but not part of the hash
1621
1622 This function currently return a dictionary. However, more advanced
1623 mapping object will likely be used in the future for a more
1624 efficient/lazy code.
1625 """
1626 # XXX will actualy return data once storage is implemented.
1627 return {}
1628
1619 1629 def _revisiondata(self, nodeorrev, _df=None, raw=False):
1620 1630 # deal with <nodeorrev> argument type
1621 1631 if isinstance(nodeorrev, int):
1622 1632 rev = nodeorrev
1623 1633 node = self.node(rev)
1624 1634 else:
1625 1635 node = nodeorrev
1626 1636 rev = None
1627 1637
1628 1638 # fast path the special `nullid` rev
1629 1639 if node == nullid:
1630 1640 return ""
1631 1641
1632 1642 # The text as stored inside the revlog. Might be the revision or might
1633 1643 # need to be processed to retrieve the revision.
1634 1644 rawtext = None
1635 1645
1636 1646 rev, rawtext, validated = self._rawtext(node, rev, _df=_df)
1637 1647
1638 1648 if raw and validated:
1639 1649 # if we don't want to process the raw text and that raw
1640 1650 # text is cached, we can exit early.
1641 1651 return rawtext
1642 1652 if rev is None:
1643 1653 rev = self.rev(node)
1644 1654 # the revlog's flag for this revision
1645 1655 # (usually alter its state or content)
1646 1656 flags = self.flags(rev)
1647 1657
1648 1658 if validated and flags == REVIDX_DEFAULT_FLAGS:
1649 1659 # no extra flags set, no flag processor runs, text = rawtext
1650 1660 return rawtext
1651 1661
1652 1662 if raw:
1653 1663 validatehash = self._processflagsraw(rawtext, flags)
1654 1664 text = rawtext
1655 1665 else:
1656 1666 text, validatehash = self._processflagsread(rawtext, flags)
1657 1667 if validatehash:
1658 1668 self.checkhash(text, node, rev=rev)
1659 1669 if not validated:
1660 1670 self._revisioncache = (node, rev, rawtext)
1661 1671
1662 1672 return text
1663 1673
1664 1674 def _rawtext(self, node, rev, _df=None):
1665 1675 """return the possibly unvalidated rawtext for a revision
1666 1676
1667 1677 returns (rev, rawtext, validated)
1668 1678 """
1669 1679
1670 1680 # revision in the cache (could be useful to apply delta)
1671 1681 cachedrev = None
1672 1682 # An intermediate text to apply deltas to
1673 1683 basetext = None
1674 1684
1675 1685 # Check if we have the entry in cache
1676 1686 # The cache entry looks like (node, rev, rawtext)
1677 1687 if self._revisioncache:
1678 1688 if self._revisioncache[0] == node:
1679 1689 return (rev, self._revisioncache[2], True)
1680 1690 cachedrev = self._revisioncache[1]
1681 1691
1682 1692 if rev is None:
1683 1693 rev = self.rev(node)
1684 1694
1685 1695 chain, stopped = self._deltachain(rev, stoprev=cachedrev)
1686 1696 if stopped:
1687 1697 basetext = self._revisioncache[2]
1688 1698
1689 1699 # drop cache to save memory, the caller is expected to
1690 1700 # update self._revisioncache after validating the text
1691 1701 self._revisioncache = None
1692 1702
1693 1703 targetsize = None
1694 1704 rawsize = self.index[rev][2]
1695 1705 if 0 <= rawsize:
1696 1706 targetsize = 4 * rawsize
1697 1707
1698 1708 bins = self._chunks(chain, df=_df, targetsize=targetsize)
1699 1709 if basetext is None:
1700 1710 basetext = bytes(bins[0])
1701 1711 bins = bins[1:]
1702 1712
1703 1713 rawtext = mdiff.patches(basetext, bins)
1704 1714 del basetext # let us have a chance to free memory early
1705 1715 return (rev, rawtext, False)
1706 1716
1707 1717 def rawdata(self, nodeorrev, _df=None):
1708 1718 """return an uncompressed raw data of a given node or revision number.
1709 1719
1710 1720 _df - an existing file handle to read from. (internal-only)
1711 1721 """
1712 1722 return self._revisiondata(nodeorrev, _df, raw=True)
1713 1723
1714 1724 def hash(self, text, p1, p2):
1715 1725 """Compute a node hash.
1716 1726
1717 1727 Available as a function so that subclasses can replace the hash
1718 1728 as needed.
1719 1729 """
1720 1730 return storageutil.hashrevisionsha1(text, p1, p2)
1721 1731
1722 1732 def checkhash(self, text, node, p1=None, p2=None, rev=None):
1723 1733 """Check node hash integrity.
1724 1734
1725 1735 Available as a function so that subclasses can extend hash mismatch
1726 1736 behaviors as needed.
1727 1737 """
1728 1738 try:
1729 1739 if p1 is None and p2 is None:
1730 1740 p1, p2 = self.parents(node)
1731 1741 if node != self.hash(text, p1, p2):
1732 1742 # Clear the revision cache on hash failure. The revision cache
1733 1743 # only stores the raw revision and clearing the cache does have
1734 1744 # the side-effect that we won't have a cache hit when the raw
1735 1745 # revision data is accessed. But this case should be rare and
1736 1746 # it is extra work to teach the cache about the hash
1737 1747 # verification state.
1738 1748 if self._revisioncache and self._revisioncache[0] == node:
1739 1749 self._revisioncache = None
1740 1750
1741 1751 revornode = rev
1742 1752 if revornode is None:
1743 1753 revornode = templatefilters.short(hex(node))
1744 1754 raise error.RevlogError(_("integrity check failed on %s:%s")
1745 1755 % (self.indexfile, pycompat.bytestr(revornode)))
1746 1756 except error.RevlogError:
1747 1757 if self._censorable and storageutil.iscensoredtext(text):
1748 1758 raise error.CensoredNodeError(self.indexfile, node, text)
1749 1759 raise
1750 1760
1751 1761 def _enforceinlinesize(self, tr, fp=None):
1752 1762 """Check if the revlog is too big for inline and convert if so.
1753 1763
1754 1764 This should be called after revisions are added to the revlog. If the
1755 1765 revlog has grown too large to be an inline revlog, it will convert it
1756 1766 to use multiple index and data files.
1757 1767 """
1758 1768 tiprev = len(self) - 1
1759 1769 if (not self._inline or
1760 1770 (self.start(tiprev) + self.length(tiprev)) < _maxinline):
1761 1771 return
1762 1772
1763 1773 trinfo = tr.find(self.indexfile)
1764 1774 if trinfo is None:
1765 1775 raise error.RevlogError(_("%s not found in the transaction")
1766 1776 % self.indexfile)
1767 1777
1768 1778 trindex = trinfo[2]
1769 1779 if trindex is not None:
1770 1780 dataoff = self.start(trindex)
1771 1781 else:
1772 1782 # revlog was stripped at start of transaction, use all leftover data
1773 1783 trindex = len(self) - 1
1774 1784 dataoff = self.end(tiprev)
1775 1785
1776 1786 tr.add(self.datafile, dataoff)
1777 1787
1778 1788 if fp:
1779 1789 fp.flush()
1780 1790 fp.close()
1781 1791 # We can't use the cached file handle after close(). So prevent
1782 1792 # its usage.
1783 1793 self._writinghandles = None
1784 1794
1785 1795 with self._indexfp('r') as ifh, self._datafp('w') as dfh:
1786 1796 for r in self:
1787 1797 dfh.write(self._getsegmentforrevs(r, r, df=ifh)[1])
1788 1798
1789 1799 with self._indexfp('w') as fp:
1790 1800 self.version &= ~FLAG_INLINE_DATA
1791 1801 self._inline = False
1792 1802 io = self._io
1793 1803 for i in self:
1794 1804 e = io.packentry(self.index[i], self.node, self.version, i)
1795 1805 fp.write(e)
1796 1806
1797 1807 # the temp file replace the real index when we exit the context
1798 1808 # manager
1799 1809
1800 1810 tr.replace(self.indexfile, trindex * self._io.size)
1801 1811 self._chunkclear()
1802 1812
1803 1813 def _nodeduplicatecallback(self, transaction, node):
1804 1814 """called when trying to add a node already stored.
1805 1815 """
1806 1816
1807 1817 def addrevision(self, text, transaction, link, p1, p2, cachedelta=None,
1808 1818 node=None, flags=REVIDX_DEFAULT_FLAGS, deltacomputer=None):
1809 1819 """add a revision to the log
1810 1820
1811 1821 text - the revision data to add
1812 1822 transaction - the transaction object used for rollback
1813 1823 link - the linkrev data to add
1814 1824 p1, p2 - the parent nodeids of the revision
1815 1825 cachedelta - an optional precomputed delta
1816 1826 node - nodeid of revision; typically node is not specified, and it is
1817 1827 computed by default as hash(text, p1, p2), however subclasses might
1818 1828 use different hashing method (and override checkhash() in such case)
1819 1829 flags - the known flags to set on the revision
1820 1830 deltacomputer - an optional deltacomputer instance shared between
1821 1831 multiple calls
1822 1832 """
1823 1833 if link == nullrev:
1824 1834 raise error.RevlogError(_("attempted to add linkrev -1 to %s")
1825 1835 % self.indexfile)
1826 1836
1827 1837 if flags:
1828 1838 node = node or self.hash(text, p1, p2)
1829 1839
1830 1840 rawtext, validatehash = self._processflagswrite(text, flags)
1831 1841
1832 1842 # If the flag processor modifies the revision data, ignore any provided
1833 1843 # cachedelta.
1834 1844 if rawtext != text:
1835 1845 cachedelta = None
1836 1846
1837 1847 if len(rawtext) > _maxentrysize:
1838 1848 raise error.RevlogError(
1839 1849 _("%s: size of %d bytes exceeds maximum revlog storage of 2GiB")
1840 1850 % (self.indexfile, len(rawtext)))
1841 1851
1842 1852 node = node or self.hash(rawtext, p1, p2)
1843 1853 if node in self.nodemap:
1844 1854 return node
1845 1855
1846 1856 if validatehash:
1847 1857 self.checkhash(rawtext, node, p1=p1, p2=p2)
1848 1858
1849 1859 return self.addrawrevision(rawtext, transaction, link, p1, p2, node,
1850 1860 flags, cachedelta=cachedelta,
1851 1861 deltacomputer=deltacomputer)
1852 1862
1853 1863 def addrawrevision(self, rawtext, transaction, link, p1, p2, node, flags,
1854 1864 cachedelta=None, deltacomputer=None):
1855 1865 """add a raw revision with known flags, node and parents
1856 1866 useful when reusing a revision not stored in this revlog (ex: received
1857 1867 over wire, or read from an external bundle).
1858 1868 """
1859 1869 dfh = None
1860 1870 if not self._inline:
1861 1871 dfh = self._datafp("a+")
1862 1872 ifh = self._indexfp("a+")
1863 1873 try:
1864 1874 return self._addrevision(node, rawtext, transaction, link, p1, p2,
1865 1875 flags, cachedelta, ifh, dfh,
1866 1876 deltacomputer=deltacomputer)
1867 1877 finally:
1868 1878 if dfh:
1869 1879 dfh.close()
1870 1880 ifh.close()
1871 1881
1872 1882 def compress(self, data):
1873 1883 """Generate a possibly-compressed representation of data."""
1874 1884 if not data:
1875 1885 return '', data
1876 1886
1877 1887 compressed = self._compressor.compress(data)
1878 1888
1879 1889 if compressed:
1880 1890 # The revlog compressor added the header in the returned data.
1881 1891 return '', compressed
1882 1892
1883 1893 if data[0:1] == '\0':
1884 1894 return '', data
1885 1895 return 'u', data
1886 1896
1887 1897 def decompress(self, data):
1888 1898 """Decompress a revlog chunk.
1889 1899
1890 1900 The chunk is expected to begin with a header identifying the
1891 1901 format type so it can be routed to an appropriate decompressor.
1892 1902 """
1893 1903 if not data:
1894 1904 return data
1895 1905
1896 1906 # Revlogs are read much more frequently than they are written and many
1897 1907 # chunks only take microseconds to decompress, so performance is
1898 1908 # important here.
1899 1909 #
1900 1910 # We can make a few assumptions about revlogs:
1901 1911 #
1902 1912 # 1) the majority of chunks will be compressed (as opposed to inline
1903 1913 # raw data).
1904 1914 # 2) decompressing *any* data will likely by at least 10x slower than
1905 1915 # returning raw inline data.
1906 1916 # 3) we want to prioritize common and officially supported compression
1907 1917 # engines
1908 1918 #
1909 1919 # It follows that we want to optimize for "decompress compressed data
1910 1920 # when encoded with common and officially supported compression engines"
1911 1921 # case over "raw data" and "data encoded by less common or non-official
1912 1922 # compression engines." That is why we have the inline lookup first
1913 1923 # followed by the compengines lookup.
1914 1924 #
1915 1925 # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib
1916 1926 # compressed chunks. And this matters for changelog and manifest reads.
1917 1927 t = data[0:1]
1918 1928
1919 1929 if t == 'x':
1920 1930 try:
1921 1931 return _zlibdecompress(data)
1922 1932 except zlib.error as e:
1923 1933 raise error.RevlogError(_('revlog decompress error: %s') %
1924 1934 stringutil.forcebytestr(e))
1925 1935 # '\0' is more common than 'u' so it goes first.
1926 1936 elif t == '\0':
1927 1937 return data
1928 1938 elif t == 'u':
1929 1939 return util.buffer(data, 1)
1930 1940
1931 1941 try:
1932 1942 compressor = self._decompressors[t]
1933 1943 except KeyError:
1934 1944 try:
1935 1945 engine = util.compengines.forrevlogheader(t)
1936 1946 compressor = engine.revlogcompressor(self._compengineopts)
1937 1947 self._decompressors[t] = compressor
1938 1948 except KeyError:
1939 1949 raise error.RevlogError(_('unknown compression type %r') % t)
1940 1950
1941 1951 return compressor.decompress(data)
1942 1952
1943 1953 def _addrevision(self, node, rawtext, transaction, link, p1, p2, flags,
1944 1954 cachedelta, ifh, dfh, alwayscache=False,
1945 1955 deltacomputer=None):
1946 1956 """internal function to add revisions to the log
1947 1957
1948 1958 see addrevision for argument descriptions.
1949 1959
1950 1960 note: "addrevision" takes non-raw text, "_addrevision" takes raw text.
1951 1961
1952 1962 if "deltacomputer" is not provided or None, a defaultdeltacomputer will
1953 1963 be used.
1954 1964
1955 1965 invariants:
1956 1966 - rawtext is optional (can be None); if not set, cachedelta must be set.
1957 1967 if both are set, they must correspond to each other.
1958 1968 """
1959 1969 if node == nullid:
1960 1970 raise error.RevlogError(_("%s: attempt to add null revision") %
1961 1971 self.indexfile)
1962 1972 if node == wdirid or node in wdirfilenodeids:
1963 1973 raise error.RevlogError(_("%s: attempt to add wdir revision") %
1964 1974 self.indexfile)
1965 1975
1966 1976 if self._inline:
1967 1977 fh = ifh
1968 1978 else:
1969 1979 fh = dfh
1970 1980
1971 1981 btext = [rawtext]
1972 1982
1973 1983 curr = len(self)
1974 1984 prev = curr - 1
1975 1985 offset = self.end(prev)
1976 1986 p1r, p2r = self.rev(p1), self.rev(p2)
1977 1987
1978 1988 # full versions are inserted when the needed deltas
1979 1989 # become comparable to the uncompressed text
1980 1990 if rawtext is None:
1981 1991 # need rawtext size, before changed by flag processors, which is
1982 1992 # the non-raw size. use revlog explicitly to avoid filelog's extra
1983 1993 # logic that might remove metadata size.
1984 1994 textlen = mdiff.patchedsize(revlog.size(self, cachedelta[0]),
1985 1995 cachedelta[1])
1986 1996 else:
1987 1997 textlen = len(rawtext)
1988 1998
1989 1999 if deltacomputer is None:
1990 2000 deltacomputer = deltautil.deltacomputer(self)
1991 2001
1992 2002 revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags)
1993 2003
1994 2004 deltainfo = deltacomputer.finddeltainfo(revinfo, fh)
1995 2005
1996 2006 e = (offset_type(offset, flags), deltainfo.deltalen, textlen,
1997 2007 deltainfo.base, link, p1r, p2r, node)
1998 2008 self.index.append(e)
1999 2009 self.nodemap[node] = curr
2000 2010
2001 2011 # Reset the pure node cache start lookup offset to account for new
2002 2012 # revision.
2003 2013 if self._nodepos is not None:
2004 2014 self._nodepos = curr
2005 2015
2006 2016 entry = self._io.packentry(e, self.node, self.version, curr)
2007 2017 self._writeentry(transaction, ifh, dfh, entry, deltainfo.data,
2008 2018 link, offset)
2009 2019
2010 2020 rawtext = btext[0]
2011 2021
2012 2022 if alwayscache and rawtext is None:
2013 2023 rawtext = deltacomputer.buildtext(revinfo, fh)
2014 2024
2015 2025 if type(rawtext) == bytes: # only accept immutable objects
2016 2026 self._revisioncache = (node, curr, rawtext)
2017 2027 self._chainbasecache[curr] = deltainfo.chainbase
2018 2028 return node
2019 2029
2020 2030 def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset):
2021 2031 # Files opened in a+ mode have inconsistent behavior on various
2022 2032 # platforms. Windows requires that a file positioning call be made
2023 2033 # when the file handle transitions between reads and writes. See
2024 2034 # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other
2025 2035 # platforms, Python or the platform itself can be buggy. Some versions
2026 2036 # of Solaris have been observed to not append at the end of the file
2027 2037 # if the file was seeked to before the end. See issue4943 for more.
2028 2038 #
2029 2039 # We work around this issue by inserting a seek() before writing.
2030 2040 # Note: This is likely not necessary on Python 3. However, because
2031 2041 # the file handle is reused for reads and may be seeked there, we need
2032 2042 # to be careful before changing this.
2033 2043 ifh.seek(0, os.SEEK_END)
2034 2044 if dfh:
2035 2045 dfh.seek(0, os.SEEK_END)
2036 2046
2037 2047 curr = len(self) - 1
2038 2048 if not self._inline:
2039 2049 transaction.add(self.datafile, offset)
2040 2050 transaction.add(self.indexfile, curr * len(entry))
2041 2051 if data[0]:
2042 2052 dfh.write(data[0])
2043 2053 dfh.write(data[1])
2044 2054 ifh.write(entry)
2045 2055 else:
2046 2056 offset += curr * self._io.size
2047 2057 transaction.add(self.indexfile, offset, curr)
2048 2058 ifh.write(entry)
2049 2059 ifh.write(data[0])
2050 2060 ifh.write(data[1])
2051 2061 self._enforceinlinesize(transaction, ifh)
2052 2062
2053 2063 def addgroup(self, deltas, linkmapper, transaction, addrevisioncb=None):
2054 2064 """
2055 2065 add a delta group
2056 2066
2057 2067 given a set of deltas, add them to the revision log. the
2058 2068 first delta is against its parent, which should be in our
2059 2069 log, the rest are against the previous delta.
2060 2070
2061 2071 If ``addrevisioncb`` is defined, it will be called with arguments of
2062 2072 this revlog and the node that was added.
2063 2073 """
2064 2074
2065 2075 if self._writinghandles:
2066 2076 raise error.ProgrammingError('cannot nest addgroup() calls')
2067 2077
2068 2078 nodes = []
2069 2079
2070 2080 r = len(self)
2071 2081 end = 0
2072 2082 if r:
2073 2083 end = self.end(r - 1)
2074 2084 ifh = self._indexfp("a+")
2075 2085 isize = r * self._io.size
2076 2086 if self._inline:
2077 2087 transaction.add(self.indexfile, end + isize, r)
2078 2088 dfh = None
2079 2089 else:
2080 2090 transaction.add(self.indexfile, isize, r)
2081 2091 transaction.add(self.datafile, end)
2082 2092 dfh = self._datafp("a+")
2083 2093 def flush():
2084 2094 if dfh:
2085 2095 dfh.flush()
2086 2096 ifh.flush()
2087 2097
2088 2098 self._writinghandles = (ifh, dfh)
2089 2099
2090 2100 try:
2091 2101 deltacomputer = deltautil.deltacomputer(self)
2092 2102 # loop through our set of deltas
2093 2103 for data in deltas:
2094 2104 node, p1, p2, linknode, deltabase, delta, flags = data
2095 2105 link = linkmapper(linknode)
2096 2106 flags = flags or REVIDX_DEFAULT_FLAGS
2097 2107
2098 2108 nodes.append(node)
2099 2109
2100 2110 if node in self.nodemap:
2101 2111 self._nodeduplicatecallback(transaction, node)
2102 2112 # this can happen if two branches make the same change
2103 2113 continue
2104 2114
2105 2115 for p in (p1, p2):
2106 2116 if p not in self.nodemap:
2107 2117 raise error.LookupError(p, self.indexfile,
2108 2118 _('unknown parent'))
2109 2119
2110 2120 if deltabase not in self.nodemap:
2111 2121 raise error.LookupError(deltabase, self.indexfile,
2112 2122 _('unknown delta base'))
2113 2123
2114 2124 baserev = self.rev(deltabase)
2115 2125
2116 2126 if baserev != nullrev and self.iscensored(baserev):
2117 2127 # if base is censored, delta must be full replacement in a
2118 2128 # single patch operation
2119 2129 hlen = struct.calcsize(">lll")
2120 2130 oldlen = self.rawsize(baserev)
2121 2131 newlen = len(delta) - hlen
2122 2132 if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen):
2123 2133 raise error.CensoredBaseError(self.indexfile,
2124 2134 self.node(baserev))
2125 2135
2126 2136 if not flags and self._peek_iscensored(baserev, delta, flush):
2127 2137 flags |= REVIDX_ISCENSORED
2128 2138
2129 2139 # We assume consumers of addrevisioncb will want to retrieve
2130 2140 # the added revision, which will require a call to
2131 2141 # revision(). revision() will fast path if there is a cache
2132 2142 # hit. So, we tell _addrevision() to always cache in this case.
2133 2143 # We're only using addgroup() in the context of changegroup
2134 2144 # generation so the revision data can always be handled as raw
2135 2145 # by the flagprocessor.
2136 2146 self._addrevision(node, None, transaction, link,
2137 2147 p1, p2, flags, (baserev, delta),
2138 2148 ifh, dfh,
2139 2149 alwayscache=bool(addrevisioncb),
2140 2150 deltacomputer=deltacomputer)
2141 2151
2142 2152 if addrevisioncb:
2143 2153 addrevisioncb(self, node)
2144 2154
2145 2155 if not dfh and not self._inline:
2146 2156 # addrevision switched from inline to conventional
2147 2157 # reopen the index
2148 2158 ifh.close()
2149 2159 dfh = self._datafp("a+")
2150 2160 ifh = self._indexfp("a+")
2151 2161 self._writinghandles = (ifh, dfh)
2152 2162 finally:
2153 2163 self._writinghandles = None
2154 2164
2155 2165 if dfh:
2156 2166 dfh.close()
2157 2167 ifh.close()
2158 2168
2159 2169 return nodes
2160 2170
2161 2171 def iscensored(self, rev):
2162 2172 """Check if a file revision is censored."""
2163 2173 if not self._censorable:
2164 2174 return False
2165 2175
2166 2176 return self.flags(rev) & REVIDX_ISCENSORED
2167 2177
2168 2178 def _peek_iscensored(self, baserev, delta, flush):
2169 2179 """Quickly check if a delta produces a censored revision."""
2170 2180 if not self._censorable:
2171 2181 return False
2172 2182
2173 2183 return storageutil.deltaiscensored(delta, baserev, self.rawsize)
2174 2184
2175 2185 def getstrippoint(self, minlink):
2176 2186 """find the minimum rev that must be stripped to strip the linkrev
2177 2187
2178 2188 Returns a tuple containing the minimum rev and a set of all revs that
2179 2189 have linkrevs that will be broken by this strip.
2180 2190 """
2181 2191 return storageutil.resolvestripinfo(minlink, len(self) - 1,
2182 2192 self.headrevs(),
2183 2193 self.linkrev, self.parentrevs)
2184 2194
2185 2195 def strip(self, minlink, transaction):
2186 2196 """truncate the revlog on the first revision with a linkrev >= minlink
2187 2197
2188 2198 This function is called when we're stripping revision minlink and
2189 2199 its descendants from the repository.
2190 2200
2191 2201 We have to remove all revisions with linkrev >= minlink, because
2192 2202 the equivalent changelog revisions will be renumbered after the
2193 2203 strip.
2194 2204
2195 2205 So we truncate the revlog on the first of these revisions, and
2196 2206 trust that the caller has saved the revisions that shouldn't be
2197 2207 removed and that it'll re-add them after this truncation.
2198 2208 """
2199 2209 if len(self) == 0:
2200 2210 return
2201 2211
2202 2212 rev, _ = self.getstrippoint(minlink)
2203 2213 if rev == len(self):
2204 2214 return
2205 2215
2206 2216 # first truncate the files on disk
2207 2217 end = self.start(rev)
2208 2218 if not self._inline:
2209 2219 transaction.add(self.datafile, end)
2210 2220 end = rev * self._io.size
2211 2221 else:
2212 2222 end += rev * self._io.size
2213 2223
2214 2224 transaction.add(self.indexfile, end)
2215 2225
2216 2226 # then reset internal state in memory to forget those revisions
2217 2227 self._revisioncache = None
2218 2228 self._chaininfocache = {}
2219 2229 self._chunkclear()
2220 2230 for x in pycompat.xrange(rev, len(self)):
2221 2231 del self.nodemap[self.node(x)]
2222 2232
2223 2233 del self.index[rev:-1]
2224 2234 self._nodepos = None
2225 2235
2226 2236 def checksize(self):
2227 2237 """Check size of index and data files
2228 2238
2229 2239 return a (dd, di) tuple.
2230 2240 - dd: extra bytes for the "data" file
2231 2241 - di: extra bytes for the "index" file
2232 2242
2233 2243 A healthy revlog will return (0, 0).
2234 2244 """
2235 2245 expected = 0
2236 2246 if len(self):
2237 2247 expected = max(0, self.end(len(self) - 1))
2238 2248
2239 2249 try:
2240 2250 with self._datafp() as f:
2241 2251 f.seek(0, io.SEEK_END)
2242 2252 actual = f.tell()
2243 2253 dd = actual - expected
2244 2254 except IOError as inst:
2245 2255 if inst.errno != errno.ENOENT:
2246 2256 raise
2247 2257 dd = 0
2248 2258
2249 2259 try:
2250 2260 f = self.opener(self.indexfile)
2251 2261 f.seek(0, io.SEEK_END)
2252 2262 actual = f.tell()
2253 2263 f.close()
2254 2264 s = self._io.size
2255 2265 i = max(0, actual // s)
2256 2266 di = actual - (i * s)
2257 2267 if self._inline:
2258 2268 databytes = 0
2259 2269 for r in self:
2260 2270 databytes += max(0, self.length(r))
2261 2271 dd = 0
2262 2272 di = actual - len(self) * s - databytes
2263 2273 except IOError as inst:
2264 2274 if inst.errno != errno.ENOENT:
2265 2275 raise
2266 2276 di = 0
2267 2277
2268 2278 return (dd, di)
2269 2279
2270 2280 def files(self):
2271 2281 res = [self.indexfile]
2272 2282 if not self._inline:
2273 2283 res.append(self.datafile)
2274 2284 return res
2275 2285
2276 2286 def emitrevisions(self, nodes, nodesorder=None, revisiondata=False,
2277 2287 assumehaveparentrevisions=False,
2278 2288 deltamode=repository.CG_DELTAMODE_STD):
2279 2289 if nodesorder not in ('nodes', 'storage', 'linear', None):
2280 2290 raise error.ProgrammingError('unhandled value for nodesorder: %s' %
2281 2291 nodesorder)
2282 2292
2283 2293 if nodesorder is None and not self._generaldelta:
2284 2294 nodesorder = 'storage'
2285 2295
2286 2296 if (not self._storedeltachains and
2287 2297 deltamode != repository.CG_DELTAMODE_PREV):
2288 2298 deltamode = repository.CG_DELTAMODE_FULL
2289 2299
2290 2300 return storageutil.emitrevisions(
2291 2301 self, nodes, nodesorder, revlogrevisiondelta,
2292 2302 deltaparentfn=self.deltaparent,
2293 2303 candeltafn=self.candelta,
2294 2304 rawsizefn=self.rawsize,
2295 2305 revdifffn=self.revdiff,
2296 2306 flagsfn=self.flags,
2297 2307 deltamode=deltamode,
2298 2308 revisiondata=revisiondata,
2299 2309 assumehaveparentrevisions=assumehaveparentrevisions)
2300 2310
2301 2311 DELTAREUSEALWAYS = 'always'
2302 2312 DELTAREUSESAMEREVS = 'samerevs'
2303 2313 DELTAREUSENEVER = 'never'
2304 2314
2305 2315 DELTAREUSEFULLADD = 'fulladd'
2306 2316
2307 2317 DELTAREUSEALL = {'always', 'samerevs', 'never', 'fulladd'}
2308 2318
2309 2319 def clone(self, tr, destrevlog, addrevisioncb=None,
2310 2320 deltareuse=DELTAREUSESAMEREVS, forcedeltabothparents=None):
2311 2321 """Copy this revlog to another, possibly with format changes.
2312 2322
2313 2323 The destination revlog will contain the same revisions and nodes.
2314 2324 However, it may not be bit-for-bit identical due to e.g. delta encoding
2315 2325 differences.
2316 2326
2317 2327 The ``deltareuse`` argument control how deltas from the existing revlog
2318 2328 are preserved in the destination revlog. The argument can have the
2319 2329 following values:
2320 2330
2321 2331 DELTAREUSEALWAYS
2322 2332 Deltas will always be reused (if possible), even if the destination
2323 2333 revlog would not select the same revisions for the delta. This is the
2324 2334 fastest mode of operation.
2325 2335 DELTAREUSESAMEREVS
2326 2336 Deltas will be reused if the destination revlog would pick the same
2327 2337 revisions for the delta. This mode strikes a balance between speed
2328 2338 and optimization.
2329 2339 DELTAREUSENEVER
2330 2340 Deltas will never be reused. This is the slowest mode of execution.
2331 2341 This mode can be used to recompute deltas (e.g. if the diff/delta
2332 2342 algorithm changes).
2333 2343
2334 2344 Delta computation can be slow, so the choice of delta reuse policy can
2335 2345 significantly affect run time.
2336 2346
2337 2347 The default policy (``DELTAREUSESAMEREVS``) strikes a balance between
2338 2348 two extremes. Deltas will be reused if they are appropriate. But if the
2339 2349 delta could choose a better revision, it will do so. This means if you
2340 2350 are converting a non-generaldelta revlog to a generaldelta revlog,
2341 2351 deltas will be recomputed if the delta's parent isn't a parent of the
2342 2352 revision.
2343 2353
2344 2354 In addition to the delta policy, the ``forcedeltabothparents``
2345 2355 argument controls whether to force compute deltas against both parents
2346 2356 for merges. By default, the current default is used.
2347 2357 """
2348 2358 if deltareuse not in self.DELTAREUSEALL:
2349 2359 raise ValueError(_('value for deltareuse invalid: %s') % deltareuse)
2350 2360
2351 2361 if len(destrevlog):
2352 2362 raise ValueError(_('destination revlog is not empty'))
2353 2363
2354 2364 if getattr(self, 'filteredrevs', None):
2355 2365 raise ValueError(_('source revlog has filtered revisions'))
2356 2366 if getattr(destrevlog, 'filteredrevs', None):
2357 2367 raise ValueError(_('destination revlog has filtered revisions'))
2358 2368
2359 2369 # lazydelta and lazydeltabase controls whether to reuse a cached delta,
2360 2370 # if possible.
2361 2371 oldlazydelta = destrevlog._lazydelta
2362 2372 oldlazydeltabase = destrevlog._lazydeltabase
2363 2373 oldamd = destrevlog._deltabothparents
2364 2374
2365 2375 try:
2366 2376 if deltareuse == self.DELTAREUSEALWAYS:
2367 2377 destrevlog._lazydeltabase = True
2368 2378 destrevlog._lazydelta = True
2369 2379 elif deltareuse == self.DELTAREUSESAMEREVS:
2370 2380 destrevlog._lazydeltabase = False
2371 2381 destrevlog._lazydelta = True
2372 2382 elif deltareuse == self.DELTAREUSENEVER:
2373 2383 destrevlog._lazydeltabase = False
2374 2384 destrevlog._lazydelta = False
2375 2385
2376 2386 destrevlog._deltabothparents = forcedeltabothparents or oldamd
2377 2387
2378 2388 deltacomputer = deltautil.deltacomputer(destrevlog)
2379 2389 index = self.index
2380 2390 for rev in self:
2381 2391 entry = index[rev]
2382 2392
2383 2393 # Some classes override linkrev to take filtered revs into
2384 2394 # account. Use raw entry from index.
2385 2395 flags = entry[0] & 0xffff
2386 2396 linkrev = entry[4]
2387 2397 p1 = index[entry[5]][7]
2388 2398 p2 = index[entry[6]][7]
2389 2399 node = entry[7]
2390 2400
2391 2401 # (Possibly) reuse the delta from the revlog if allowed and
2392 2402 # the revlog chunk is a delta.
2393 2403 cachedelta = None
2394 2404 rawtext = None
2395 2405 if (deltareuse != self.DELTAREUSEFULLADD
2396 2406 and destrevlog._lazydelta):
2397 2407 dp = self.deltaparent(rev)
2398 2408 if dp != nullrev:
2399 2409 cachedelta = (dp, bytes(self._chunk(rev)))
2400 2410
2401 2411 if not cachedelta:
2402 2412 rawtext = self.rawdata(rev)
2403 2413
2404 2414
2405 2415 if deltareuse == self.DELTAREUSEFULLADD:
2406 2416 destrevlog.addrevision(rawtext, tr, linkrev, p1, p2,
2407 2417 cachedelta=cachedelta,
2408 2418 node=node, flags=flags,
2409 2419 deltacomputer=deltacomputer)
2410 2420 else:
2411 2421 ifh = destrevlog.opener(destrevlog.indexfile, 'a+',
2412 2422 checkambig=False)
2413 2423 dfh = None
2414 2424 if not destrevlog._inline:
2415 2425 dfh = destrevlog.opener(destrevlog.datafile, 'a+')
2416 2426 try:
2417 2427 destrevlog._addrevision(node, rawtext, tr, linkrev, p1,
2418 2428 p2, flags, cachedelta, ifh, dfh,
2419 2429 deltacomputer=deltacomputer)
2420 2430 finally:
2421 2431 if dfh:
2422 2432 dfh.close()
2423 2433 ifh.close()
2424 2434
2425 2435 if addrevisioncb:
2426 2436 addrevisioncb(self, rev, node)
2427 2437 finally:
2428 2438 destrevlog._lazydelta = oldlazydelta
2429 2439 destrevlog._lazydeltabase = oldlazydeltabase
2430 2440 destrevlog._deltabothparents = oldamd
2431 2441
2432 2442 def censorrevision(self, tr, censornode, tombstone=b''):
2433 2443 if (self.version & 0xFFFF) == REVLOGV0:
2434 2444 raise error.RevlogError(_('cannot censor with version %d revlogs') %
2435 2445 self.version)
2436 2446
2437 2447 censorrev = self.rev(censornode)
2438 2448 tombstone = storageutil.packmeta({b'censored': tombstone}, b'')
2439 2449
2440 2450 if len(tombstone) > self.rawsize(censorrev):
2441 2451 raise error.Abort(_('censor tombstone must be no longer than '
2442 2452 'censored data'))
2443 2453
2444 2454 # Rewriting the revlog in place is hard. Our strategy for censoring is
2445 2455 # to create a new revlog, copy all revisions to it, then replace the
2446 2456 # revlogs on transaction close.
2447 2457
2448 2458 newindexfile = self.indexfile + b'.tmpcensored'
2449 2459 newdatafile = self.datafile + b'.tmpcensored'
2450 2460
2451 2461 # This is a bit dangerous. We could easily have a mismatch of state.
2452 2462 newrl = revlog(self.opener, newindexfile, newdatafile,
2453 2463 censorable=True)
2454 2464 newrl.version = self.version
2455 2465 newrl._generaldelta = self._generaldelta
2456 2466 newrl._io = self._io
2457 2467
2458 2468 for rev in self.revs():
2459 2469 node = self.node(rev)
2460 2470 p1, p2 = self.parents(node)
2461 2471
2462 2472 if rev == censorrev:
2463 2473 newrl.addrawrevision(tombstone, tr, self.linkrev(censorrev),
2464 2474 p1, p2, censornode, REVIDX_ISCENSORED)
2465 2475
2466 2476 if newrl.deltaparent(rev) != nullrev:
2467 2477 raise error.Abort(_('censored revision stored as delta; '
2468 2478 'cannot censor'),
2469 2479 hint=_('censoring of revlogs is not '
2470 2480 'fully implemented; please report '
2471 2481 'this bug'))
2472 2482 continue
2473 2483
2474 2484 if self.iscensored(rev):
2475 2485 if self.deltaparent(rev) != nullrev:
2476 2486 raise error.Abort(_('cannot censor due to censored '
2477 2487 'revision having delta stored'))
2478 2488 rawtext = self._chunk(rev)
2479 2489 else:
2480 2490 rawtext = self.rawdata(rev)
2481 2491
2482 2492 newrl.addrawrevision(rawtext, tr, self.linkrev(rev), p1, p2, node,
2483 2493 self.flags(rev))
2484 2494
2485 2495 tr.addbackup(self.indexfile, location='store')
2486 2496 if not self._inline:
2487 2497 tr.addbackup(self.datafile, location='store')
2488 2498
2489 2499 self.opener.rename(newrl.indexfile, self.indexfile)
2490 2500 if not self._inline:
2491 2501 self.opener.rename(newrl.datafile, self.datafile)
2492 2502
2493 2503 self.clearcaches()
2494 2504 self._loadindex()
2495 2505
2496 2506 def verifyintegrity(self, state):
2497 2507 """Verifies the integrity of the revlog.
2498 2508
2499 2509 Yields ``revlogproblem`` instances describing problems that are
2500 2510 found.
2501 2511 """
2502 2512 dd, di = self.checksize()
2503 2513 if dd:
2504 2514 yield revlogproblem(error=_('data length off by %d bytes') % dd)
2505 2515 if di:
2506 2516 yield revlogproblem(error=_('index contains %d extra bytes') % di)
2507 2517
2508 2518 version = self.version & 0xFFFF
2509 2519
2510 2520 # The verifier tells us what version revlog we should be.
2511 2521 if version != state['expectedversion']:
2512 2522 yield revlogproblem(
2513 2523 warning=_("warning: '%s' uses revlog format %d; expected %d") %
2514 2524 (self.indexfile, version, state['expectedversion']))
2515 2525
2516 2526 state['skipread'] = set()
2517 2527
2518 2528 for rev in self:
2519 2529 node = self.node(rev)
2520 2530
2521 2531 # Verify contents. 4 cases to care about:
2522 2532 #
2523 2533 # common: the most common case
2524 2534 # rename: with a rename
2525 2535 # meta: file content starts with b'\1\n', the metadata
2526 2536 # header defined in filelog.py, but without a rename
2527 2537 # ext: content stored externally
2528 2538 #
2529 2539 # More formally, their differences are shown below:
2530 2540 #
2531 2541 # | common | rename | meta | ext
2532 2542 # -------------------------------------------------------
2533 2543 # flags() | 0 | 0 | 0 | not 0
2534 2544 # renamed() | False | True | False | ?
2535 2545 # rawtext[0:2]=='\1\n'| False | True | True | ?
2536 2546 #
2537 2547 # "rawtext" means the raw text stored in revlog data, which
2538 2548 # could be retrieved by "rawdata(rev)". "text"
2539 2549 # mentioned below is "revision(rev)".
2540 2550 #
2541 2551 # There are 3 different lengths stored physically:
2542 2552 # 1. L1: rawsize, stored in revlog index
2543 2553 # 2. L2: len(rawtext), stored in revlog data
2544 2554 # 3. L3: len(text), stored in revlog data if flags==0, or
2545 2555 # possibly somewhere else if flags!=0
2546 2556 #
2547 2557 # L1 should be equal to L2. L3 could be different from them.
2548 2558 # "text" may or may not affect commit hash depending on flag
2549 2559 # processors (see flagutil.addflagprocessor).
2550 2560 #
2551 2561 # | common | rename | meta | ext
2552 2562 # -------------------------------------------------
2553 2563 # rawsize() | L1 | L1 | L1 | L1
2554 2564 # size() | L1 | L2-LM | L1(*) | L1 (?)
2555 2565 # len(rawtext) | L2 | L2 | L2 | L2
2556 2566 # len(text) | L2 | L2 | L2 | L3
2557 2567 # len(read()) | L2 | L2-LM | L2-LM | L3 (?)
2558 2568 #
2559 2569 # LM: length of metadata, depending on rawtext
2560 2570 # (*): not ideal, see comment in filelog.size
2561 2571 # (?): could be "- len(meta)" if the resolved content has
2562 2572 # rename metadata
2563 2573 #
2564 2574 # Checks needed to be done:
2565 2575 # 1. length check: L1 == L2, in all cases.
2566 2576 # 2. hash check: depending on flag processor, we may need to
2567 2577 # use either "text" (external), or "rawtext" (in revlog).
2568 2578
2569 2579 try:
2570 2580 skipflags = state.get('skipflags', 0)
2571 2581 if skipflags:
2572 2582 skipflags &= self.flags(rev)
2573 2583
2574 2584 if skipflags:
2575 2585 state['skipread'].add(node)
2576 2586 else:
2577 2587 # Side-effect: read content and verify hash.
2578 2588 self.revision(node)
2579 2589
2580 2590 l1 = self.rawsize(rev)
2581 2591 l2 = len(self.rawdata(node))
2582 2592
2583 2593 if l1 != l2:
2584 2594 yield revlogproblem(
2585 2595 error=_('unpacked size is %d, %d expected') % (l2, l1),
2586 2596 node=node)
2587 2597
2588 2598 except error.CensoredNodeError:
2589 2599 if state['erroroncensored']:
2590 2600 yield revlogproblem(error=_('censored file data'),
2591 2601 node=node)
2592 2602 state['skipread'].add(node)
2593 2603 except Exception as e:
2594 2604 yield revlogproblem(
2595 2605 error=_('unpacking %s: %s') % (short(node),
2596 2606 stringutil.forcebytestr(e)),
2597 2607 node=node)
2598 2608 state['skipread'].add(node)
2599 2609
2600 2610 def storageinfo(self, exclusivefiles=False, sharedfiles=False,
2601 2611 revisionscount=False, trackedsize=False,
2602 2612 storedsize=False):
2603 2613 d = {}
2604 2614
2605 2615 if exclusivefiles:
2606 2616 d['exclusivefiles'] = [(self.opener, self.indexfile)]
2607 2617 if not self._inline:
2608 2618 d['exclusivefiles'].append((self.opener, self.datafile))
2609 2619
2610 2620 if sharedfiles:
2611 2621 d['sharedfiles'] = []
2612 2622
2613 2623 if revisionscount:
2614 2624 d['revisionscount'] = len(self)
2615 2625
2616 2626 if trackedsize:
2617 2627 d['trackedsize'] = sum(map(self.rawsize, iter(self)))
2618 2628
2619 2629 if storedsize:
2620 2630 d['storedsize'] = sum(self.opener.stat(path).st_size
2621 2631 for path in self.files())
2622 2632
2623 2633 return d
General Comments 0
You need to be logged in to leave comments. Login now