161
138
should be no way to get an earlier version deleting a later
165
Text of the weave; list of control instruction tuples and strings.
168
145
List of parents, indexed by version number.
169
146
It is only necessary to store the minimal set of parents for
170
147
each version; the parent's parents are implied.
173
List of hex SHA-1 of each version.
176
List of symbolic names for each version. Each should be unique.
179
For each name, the version number.
182
Descriptive name of this weave; typically the filename if known.
150
List of hex SHA-1 of each version, or None if not recorded.
186
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
187
'_weave_name', '_matcher']
189
def __init__(self, weave_name=None, access_mode='w', matcher=None):
190
super(Weave, self).__init__(access_mode)
196
self._weave_name = weave_name
198
self._matcher = bzrlib.patiencediff.PatienceSequenceMatcher
200
self._matcher = matcher
203
return "Weave(%r)" % self._weave_name
206
"""Return a deep copy of self.
208
The copy can be modified without affecting the original weave."""
210
other._weave = self._weave[:]
211
other._parents = self._parents[:]
212
other._sha1s = self._sha1s[:]
213
other._names = self._names[:]
214
other._name_map = self._name_map.copy()
215
other._weave_name = self._weave_name
218
158
def __eq__(self, other):
219
159
if not isinstance(other, Weave):
221
return self._parents == other._parents \
222
and self._weave == other._weave \
223
and self._sha1s == other._sha1s
161
return self._v == other._v \
162
and self._l == other._l
225
165
def __ne__(self, other):
226
166
return not self.__eq__(other)
228
@deprecated_method(zero_eight)
229
def idx_to_name(self, index):
230
"""Old public interface, the public interface is all names now."""
233
def _idx_to_name(self, version):
234
return self._names[version]
236
@deprecated_method(zero_eight)
237
def lookup(self, name):
238
"""Backwards compatibility thunk:
240
Return name, as name is valid in the api now, and spew deprecation
245
def _lookup(self, name):
246
"""Convert symbolic version name to index."""
248
return self._name_map[name]
250
raise RevisionNotPresent(name, self._weave_name)
252
@deprecated_method(zero_eight)
253
def iter_names(self):
254
"""Deprecated convenience function, please see VersionedFile.names()."""
255
return iter(self.names())
257
@deprecated_method(zero_eight)
259
"""See Weave.versions for the current api."""
260
return self.versions()
263
"""See VersionedFile.versions."""
264
return self._names[:]
266
def has_version(self, version_id):
267
"""See VersionedFile.has_version."""
268
return self._name_map.has_key(version_id)
270
__contains__ = has_version
272
def get_delta(self, version_id):
273
"""See VersionedFile.get_delta."""
274
return self.get_deltas([version_id])[version_id]
276
def get_deltas(self, version_ids):
277
"""See VersionedFile.get_deltas."""
278
version_ids = self.get_ancestry(version_ids)
279
for version_id in version_ids:
280
if not self.has_version(version_id):
281
raise RevisionNotPresent(version_id, self)
282
# try extracting all versions; parallel extraction is used
283
nv = self.num_versions()
289
last_parent_lines = {}
291
parent_inclusions = {}
296
# its simplest to generate a full set of prepared variables.
298
name = self._names[i]
299
sha1s[name] = self.get_sha1(name)
300
parents_list = self.get_parents(name)
302
parent = parents_list[0]
303
parents[name] = parent
304
parent_inclusions[name] = inclusions[parent]
307
parent_inclusions[name] = set()
308
# we want to emit start, finish, replacement_length, replacement_lines tuples.
309
diff_hunks[name] = []
310
current_hunks[name] = [0, 0, 0, []] # #start, finish, repl_length, repl_tuples
311
parent_linenums[name] = 0
313
parent_noeols[name] = False
314
last_parent_lines[name] = None
315
new_inc = set([name])
316
for p in self._parents[i]:
317
new_inc.update(inclusions[self._idx_to_name(p)])
318
# debug only, known good so far.
319
#assert set(new_inc) == set(self.get_ancestry(name)), \
320
# 'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
321
inclusions[name] = new_inc
323
nlines = len(self._weave)
325
for lineno, inserted, deletes, line in self._walk_internal():
326
# a line is active in a version if:
327
# insert is in the versions inclusions
329
# deleteset & the versions inclusions is an empty set.
330
# so - if we have a included by mapping - version is included by
331
# children, we get a list of children to examine for deletes affect
332
# ing them, which is less than the entire set of children.
333
for version_id in version_ids:
334
# The active inclusion must be an ancestor,
335
# and no ancestors must have deleted this line,
336
# because we don't support resurrection.
337
parent_inclusion = parent_inclusions[version_id]
338
inclusion = inclusions[version_id]
339
parent_active = inserted in parent_inclusion and not (deletes & parent_inclusion)
340
version_active = inserted in inclusion and not (deletes & inclusion)
341
if not parent_active and not version_active:
342
# unrelated line of ancestry
344
elif parent_active and version_active:
346
parent_linenum = parent_linenums[version_id]
347
if current_hunks[version_id] != [parent_linenum, parent_linenum, 0, []]:
348
diff_hunks[version_id].append(tuple(current_hunks[version_id]))
350
current_hunks[version_id] = [parent_linenum, parent_linenum, 0, []]
351
parent_linenums[version_id] = parent_linenum
354
noeols[version_id] = True
357
elif parent_active and not version_active:
359
current_hunks[version_id][1] += 1
360
parent_linenums[version_id] += 1
361
last_parent_lines[version_id] = line
362
elif not parent_active and version_active:
364
# noeol only occurs at the end of a file because we
365
# diff linewise. We want to show noeol changes as a
366
# empty diff unless the actual eol-less content changed.
369
if last_parent_lines[version_id][-1] != '\n':
370
parent_noeols[version_id] = True
371
except (TypeError, IndexError):
374
if theline[-1] != '\n':
375
noeols[version_id] = True
379
parent_should_go = False
381
if parent_noeols[version_id] == noeols[version_id]:
382
# no noeol toggle, so trust the weaves statement
383
# that this line is changed.
385
if parent_noeols[version_id]:
386
theline = theline + '\n'
387
elif parent_noeols[version_id]:
388
# parent has no eol, we do:
389
# our line is new, report as such..
391
elif noeols[version_id]:
392
# append a eol so that it looks like
394
theline = theline + '\n'
395
if parents[version_id] is not None:
396
#if last_parent_lines[version_id] is not None:
397
parent_should_go = True
398
if last_parent_lines[version_id] != theline:
401
#parent_should_go = False
403
current_hunks[version_id][2] += 1
404
current_hunks[version_id][3].append((inserted, theline))
406
# last hunk last parent line is not eaten
407
current_hunks[version_id][1] -= 1
408
if current_hunks[version_id][1] < 0:
409
current_hunks[version_id][1] = 0
410
# import pdb;pdb.set_trace()
411
# assert current_hunks[version_id][1] >= 0
415
version = self._idx_to_name(i)
416
if current_hunks[version] != [0, 0, 0, []]:
417
diff_hunks[version].append(tuple(current_hunks[version]))
419
for version_id in version_ids:
420
result[version_id] = (
424
diff_hunks[version_id],
428
def get_parents(self, version_id):
429
"""See VersionedFile.get_parent."""
430
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
432
def _check_repeated_add(self, name, parents, text, sha1):
433
"""Check that a duplicated add is OK.
435
If it is, return the (old) index; otherwise raise an exception.
437
idx = self._lookup(name)
438
if sorted(self._parents[idx]) != sorted(parents) \
439
or sha1 != self._sha1s[idx]:
440
raise RevisionAlreadyPresent(name, self._weave_name)
443
@deprecated_method(zero_eight)
444
def add_identical(self, old_rev_id, new_rev_id, parents):
445
"""Please use Weave.clone_text now."""
446
return self.clone_text(new_rev_id, old_rev_id, parents)
448
def _add_lines(self, version_id, parents, lines, parent_texts):
449
"""See VersionedFile.add_lines."""
450
return self._add(version_id, lines, map(self._lookup, parents))
452
@deprecated_method(zero_eight)
453
def add(self, name, parents, text, sha1=None):
454
"""See VersionedFile.add_lines for the non deprecated api."""
455
return self._add(name, text, map(self._maybe_lookup, parents), sha1)
457
def _add(self, version_id, lines, parents, sha1=None):
169
def add(self, parents, text):
458
170
"""Add a single text on top of the weave.
460
172
Returns the index number of the newly added version.
463
Symbolic name for this version.
464
(Typically the revision-id of the revision that added it.)
467
175
List or set of direct parent version numbers.
470
Sequence of lines to be added in the new version.
473
assert isinstance(version_id, basestring)
474
self._check_lines_not_unicode(lines)
475
self._check_lines_are_lines(lines)
477
sha1 = sha_strings(lines)
478
if version_id in self._name_map:
479
return self._check_repeated_add(version_id, parents, lines, sha1)
481
self._check_versions(parents)
482
## self._check_lines(lines)
483
new_version = len(self._parents)
485
# if we abort after here the (in-memory) weave will be corrupt because only
486
# some fields are updated
487
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
488
# - Robert Collins 20060226
489
self._parents.append(parents[:])
178
Sequence of lines to be added in the new version."""
179
## self._check_versions(parents)
180
## self._check_lines(text)
191
ancestors = self.inclusions(parents)
192
delta = self._delta(ancestors, text)
194
# offset gives the number of lines that have been inserted
195
# into the weave up to the current point; if the original edit instruction
196
# says to change line A then we actually change (A+offset)
199
for i1, i2, newlines in delta:
202
assert i2 <= len(self._l)
204
# the deletion and insertion are handled separately.
205
# first delete the region.
207
self._l.insert(i1+offset, ('[', idx))
208
self._l.insert(i2+offset+1, (']', idx))
213
# there may have been a deletion spanning up to
214
# i2; we want to insert after this region to make sure
215
# we don't destroy ourselves
217
self._l[i:i] = [('{', idx)] \
220
offset += 2 + len(newlines)
222
self._addversion(parents)
224
# special case; adding with no parents revision; can do this
225
# more quickly by just appending unconditionally
226
self._l.append(('{', idx))
228
self._l.append(('}', idx))
230
self._addversion(None)
490
232
self._sha1s.append(sha1)
491
self._names.append(version_id)
492
self._name_map[version_id] = new_version
496
# special case; adding with no parents revision; can do
497
# this more quickly by just appending unconditionally.
498
# even more specially, if we're adding an empty text we
499
# need do nothing at all.
501
self._weave.append(('{', new_version))
502
self._weave.extend(lines)
503
self._weave.append(('}', None))
506
if len(parents) == 1:
507
pv = list(parents)[0]
508
if sha1 == self._sha1s[pv]:
509
# special case: same as the single parent
513
ancestors = self._inclusions(parents)
517
# basis a list of (origin, lineno, line)
520
for origin, lineno, line in self._extract(ancestors):
521
basis_lineno.append(lineno)
522
basis_lines.append(line)
524
# another small special case: a merge, producing the same text
526
if lines == basis_lines:
529
# add a sentinel, because we can also match against the final line
530
basis_lineno.append(len(self._weave))
532
# XXX: which line of the weave should we really consider
533
# matches the end of the file? the current code says it's the
534
# last line of the weave?
536
#print 'basis_lines:', basis_lines
537
#print 'new_lines: ', lines
539
s = self._matcher(None, basis_lines, lines)
541
# offset gives the number of lines that have been inserted
542
# into the weave up to the current point; if the original edit instruction
543
# says to change line A then we actually change (A+offset)
546
for tag, i1, i2, j1, j2 in s.get_opcodes():
547
# i1,i2 are given in offsets within basis_lines; we need to map them
548
# back to offsets within the entire weave
549
#print 'raw match', tag, i1, i2, j1, j2
553
i1 = basis_lineno[i1]
554
i2 = basis_lineno[i2]
556
assert 0 <= j1 <= j2 <= len(lines)
558
#print tag, i1, i2, j1, j2
560
# the deletion and insertion are handled separately.
561
# first delete the region.
563
self._weave.insert(i1+offset, ('[', new_version))
564
self._weave.insert(i2+offset+1, (']', new_version))
568
# there may have been a deletion spanning up to
569
# i2; we want to insert after this region to make sure
570
# we don't destroy ourselves
572
self._weave[i:i] = ([('{', new_version)]
575
offset += 2 + (j2 - j1)
578
def _clone_text(self, new_version_id, old_version_id, parents):
579
"""See VersionedFile.clone_text."""
580
old_lines = self.get_text(old_version_id)
581
self.add_lines(new_version_id, parents, old_lines)
583
def _inclusions(self, versions):
237
def inclusions(self, versions):
584
238
"""Return set of all ancestors of given version(s)."""
585
if not len(versions):
587
239
i = set(versions)
588
for v in xrange(max(versions), 0, -1):
590
# include all its parents
591
i.update(self._parents[v])
593
## except IndexError:
594
## raise ValueError("version %d not present in weave" % v)
596
@deprecated_method(zero_eight)
597
def inclusions(self, version_ids):
598
"""Deprecated - see VersionedFile.get_ancestry for the replacement."""
244
# include all its parents
249
raise ValueError("version %d not present in weave" % v)
252
def minimal_parents(self, version):
253
"""Find the minimal set of parents for the version."""
254
included = self._v[version]
601
if isinstance(version_ids[0], int):
602
return [self._idx_to_name(v) for v in self._inclusions(version_ids)]
259
li.sort(reverse=True)
267
gotit.update(self.inclusions(pv))
269
assert mininc[0] >= 0
270
assert mininc[-1] < version
274
def _addversion(self, parents):
276
self._v.append(parents)
604
return self.get_ancestry(version_ids)
278
self._v.append(frozenset())
606
def get_ancestry(self, version_ids):
607
"""See VersionedFile.get_ancestry."""
608
if isinstance(version_ids, basestring):
609
version_ids = [version_ids]
610
i = self._inclusions([self._lookup(v) for v in version_ids])
611
return [self._idx_to_name(v) for v in i]
613
281
def _check_lines(self, text):
614
282
if not isinstance(text, list):
625
293
"""Check everything in the sequence of indexes is valid"""
626
294
for i in indexes:
629
297
except IndexError:
630
298
raise IndexError("invalid version number %r" % i)
632
def _compatible_parents(self, my_parents, other_parents):
633
"""During join check that other_parents are joinable with my_parents.
635
Joinable is defined as 'is a subset of' - supersets may require
636
regeneration of diffs, but subsets do not.
638
return len(other_parents.difference(my_parents)) == 0
640
def annotate(self, version_id):
641
if isinstance(version_id, int):
642
warnings.warn('Weave.annotate(int) is deprecated. Please use version names'
643
' in all circumstances as of 0.8',
648
for origin, lineno, text in self._extract([version_id]):
649
result.append((origin, text))
652
return super(Weave, self).annotate(version_id)
654
def annotate_iter(self, version_id):
655
"""Yield list of (version-id, line) pairs for the specified version.
301
def annotate(self, index):
302
return list(self.annotate_iter(index))
305
def annotate_iter(self, version):
306
"""Yield list of (index-id, line) pairs for the specified version.
657
308
The index indicates when the line originated in the weave."""
658
incls = [self._lookup(version_id)]
659
for origin, lineno, text in self._extract(incls):
660
yield self._idx_to_name(origin), text
662
@deprecated_method(zero_eight)
664
"""_walk has become visit, a supported api."""
665
return self._walk_internal()
667
def iter_lines_added_or_present_in_versions(self, version_ids=None):
668
"""See VersionedFile.iter_lines_added_or_present_in_versions()."""
669
if version_ids is None:
670
version_ids = self.versions()
671
version_ids = set(version_ids)
672
for lineno, inserted, deletes, line in self._walk_internal(version_ids):
673
# if inserted not in version_ids then it was inserted before the
674
# versions we care about, but because weaves cannot represent ghosts
675
# properly, we do not filter down to that
676
# if inserted not in version_ids: continue
682
#@deprecated_method(zero_eight)
683
def walk(self, version_ids=None):
684
"""See VersionedFile.walk."""
685
return self._walk_internal(version_ids)
687
def _walk_internal(self, version_ids=None):
688
"""Helper method for weave actions."""
693
lineno = 0 # line of weave, 0-based
695
for l in self._weave:
696
if l.__class__ == tuple:
700
istack.append(self._names[v])
704
assert self._names[v] not in dset
705
dset.add(self._names[v])
707
dset.remove(self._names[v])
709
raise WeaveFormatError('unexpected instruction %r' % v)
711
assert l.__class__ in (str, unicode)
713
yield lineno, istack[-1], frozenset(dset), l
717
raise WeaveFormatError("unclosed insertion blocks "
718
"at end of weave: %s" % istack)
720
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
723
def plan_merge(self, ver_a, ver_b):
724
"""Return pseudo-annotation indicating how the two versions merge.
726
This is computed between versions a and b and their common
729
Weave lines present in none of them are skipped entirely.
731
inc_a = set(self.get_ancestry([ver_a]))
732
inc_b = set(self.get_ancestry([ver_b]))
733
inc_c = inc_a & inc_b
735
for lineno, insert, deleteset, line in\
736
self.walk([ver_a, ver_b]):
737
if deleteset & inc_c:
738
# killed in parent; can't be in either a or b
739
# not relevant to our work
740
yield 'killed-base', line
741
elif insert in inc_c:
742
# was inserted in base
743
killed_a = bool(deleteset & inc_a)
744
killed_b = bool(deleteset & inc_b)
745
if killed_a and killed_b:
746
yield 'killed-both', line
748
yield 'killed-a', line
750
yield 'killed-b', line
752
yield 'unchanged', line
753
elif insert in inc_a:
754
if deleteset & inc_a:
755
yield 'ghost-a', line
759
elif insert in inc_b:
760
if deleteset & inc_b:
761
yield 'ghost-b', line
765
# not in either revision
766
yield 'irrelevant', line
768
yield 'unchanged', '' # terminator
309
for origin, lineno, text in self._extract([version]):
770
313
def _extract(self, versions):
771
314
"""Yield annotation of lines in included set.
844
assert l.__class__ in (str, unicode)
353
assert isinstance(l, basestring)
845
354
if isactive is None:
846
355
isactive = (not dset) and istack and (istack[-1] in included)
848
result.append((istack[-1], lineno, l))
357
yield istack[-1], lineno, l
851
raise WeaveFormatError("unclosed insertion blocks "
852
"at end of weave: %s" % istack)
361
raise WFE("unclosed insertion blocks at end of weave",
854
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
858
@deprecated_method(zero_eight)
859
def get_iter(self, name_or_index):
860
"""Deprecated, please do not use. Lookups are not not needed.
862
Please use get_lines now.
864
return iter(self.get_lines(self._maybe_lookup(name_or_index)))
866
@deprecated_method(zero_eight)
867
def maybe_lookup(self, name_or_index):
868
"""Deprecated, please do not use. Lookups are not not needed."""
869
return self._maybe_lookup(name_or_index)
871
def _maybe_lookup(self, name_or_index):
872
"""Convert possible symbolic name to index, or pass through indexes.
876
if isinstance(name_or_index, (int, long)):
879
return self._lookup(name_or_index)
881
@deprecated_method(zero_eight)
882
def get(self, version_id):
883
"""Please use either Weave.get_text or Weave.get_lines as desired."""
884
return self.get_lines(version_id)
886
def get_lines(self, version_id):
887
"""See VersionedFile.get_lines()."""
888
int_index = self._maybe_lookup(version_id)
889
result = [line for (origin, lineno, line) in self._extract([int_index])]
890
expected_sha1 = self._sha1s[int_index]
891
measured_sha1 = sha_strings(result)
892
if measured_sha1 != expected_sha1:
893
raise errors.WeaveInvalidChecksum(
894
'file %s, revision %s, expected: %s, measured %s'
895
% (self._weave_name, version_id,
896
expected_sha1, measured_sha1))
899
def get_sha1(self, version_id):
900
"""See VersionedFile.get_sha1()."""
901
return self._sha1s[self._lookup(version_id)]
903
@deprecated_method(zero_eight)
364
raise WFE("unclosed deletion blocks at end of weave",
368
def get_iter(self, version):
369
"""Yield lines for the specified version."""
370
for origin, lineno, line in self._extract([version]):
374
def get(self, index):
375
return list(self.get_iter(index))
378
def mash_iter(self, included):
379
"""Return composed version of multiple included versions."""
380
included = frozenset(included)
381
for origin, lineno, text in self._extract(included):
385
def dump(self, to_file):
386
from pprint import pprint
387
print >>to_file, "Weave._l = ",
388
pprint(self._l, to_file)
389
print >>to_file, "Weave._v = ",
390
pprint(self._v, to_file)
904
394
def numversions(self):
905
"""How many versions are in this weave?
907
Deprecated in favour of num_versions.
909
return self.num_versions()
911
def num_versions(self):
912
"""How many versions are in this weave?"""
913
l = len(self._parents)
914
396
assert l == len(self._sha1s)
917
__len__ = num_versions
919
400
def check(self, progress_bar=None):
920
# TODO evaluate performance hit of using string sets in this routine.
921
# TODO: check no circular inclusions
922
# TODO: create a nested progress bar
923
for version in range(self.num_versions()):
924
inclusions = list(self._parents[version])
401
# check no circular inclusions
402
for version in range(self.numversions()):
403
inclusions = list(self._v[version])
926
405
inclusions.sort()
927
406
if inclusions[-1] >= version:
928
407
raise WeaveFormatError("invalid included version %d for index %d"
929
408
% (inclusions[-1], version))
931
# try extracting all versions; parallel extraction is used
932
nv = self.num_versions()
937
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
938
# The problem is that set membership is much more expensive
939
name = self._idx_to_name(i)
940
sha1s[name] = sha.new()
942
new_inc = set([name])
943
for p in self._parents[i]:
944
new_inc.update(inclusions[self._idx_to_name(p)])
946
assert set(new_inc) == set(self.get_ancestry(name)), \
947
'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
948
inclusions[name] = new_inc
950
nlines = len(self._weave)
952
update_text = 'checking weave'
954
short_name = os.path.basename(self._weave_name)
955
update_text = 'checking %s' % (short_name,)
956
update_text = update_text[:25]
958
for lineno, insert, deleteset, line in self._walk_internal():
410
# try extracting all versions; this is a bit slow and parallel
411
# extraction could be used
413
nv = self.numversions()
414
for version in range(nv):
960
progress_bar.update(update_text, lineno, nlines)
962
for name, name_inclusions in inclusions.items():
963
# The active inclusion must be an ancestor,
964
# and no ancestors must have deleted this line,
965
# because we don't support resurrection.
966
if (insert in name_inclusions) and not (deleteset & name_inclusions):
967
sha1s[name].update(line)
970
version = self._idx_to_name(i)
971
hd = sha1s[version].hexdigest()
972
expected = self._sha1s[i]
416
progress_bar.update('checking text', version, nv)
418
for l in self.get_iter(version):
421
expected = self._sha1s[version]
973
422
if hd != expected:
974
raise errors.WeaveInvalidChecksum(
975
"mismatched sha1 for version %s: "
976
"got %s, expected %s"
977
% (version, hd, expected))
423
raise WeaveError("mismatched sha1 for version %d; "
424
"got %s, expected %s"
425
% (version, hd, expected))
979
427
# TODO: check insertions are properly nested, that there are
980
428
# no lines outside of insertion blocks, that deletions are
981
429
# properly paired, etc.
983
def _join(self, other, pb, msg, version_ids, ignore_missing):
984
"""Worker routine for join()."""
985
if not other.versions():
986
return # nothing to update, easy
989
# versions is never none, InterWeave checks this.
992
# two loops so that we do not change ourselves before verifying it
994
# work through in index order to make sure we get all dependencies
997
# get the selected versions only that are in other.versions.
998
version_ids = set(other.versions()).intersection(set(version_ids))
999
# pull in the referenced graph.
1000
version_ids = other.get_ancestry(version_ids)
1001
pending_graph = [(version, other.get_parents(version)) for
1002
version in version_ids]
1003
for name in topo_sort(pending_graph):
1004
other_idx = other._name_map[name]
1005
# returns True if we have it, False if we need it.
1006
if not self._check_version_consistent(other, other_idx, name):
1007
names_to_join.append((other_idx, name))
1016
for other_idx, name in names_to_join:
1017
# TODO: If all the parents of the other version are already
1018
# present then we can avoid some work by just taking the delta
1019
# and adjusting the offsets.
1020
new_parents = self._imported_parents(other, other_idx)
1021
sha1 = other._sha1s[other_idx]
1026
pb.update(msg, merged, len(names_to_join))
1028
lines = other.get_lines(other_idx)
1029
self._add(name, lines, new_parents, sha1)
1031
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
1032
merged, processed, self._weave_name, time.time()-time0))
1034
def _imported_parents(self, other, other_idx):
1035
"""Return list of parents in self corresponding to indexes in other."""
1037
for parent_idx in other._parents[other_idx]:
1038
parent_name = other._names[parent_idx]
1039
if parent_name not in self._name_map:
1040
# should not be possible
1041
raise WeaveError("missing parent {%s} of {%s} in %r"
1042
% (parent_name, other._name_map[other_idx], self))
1043
new_parents.append(self._name_map[parent_name])
1046
def _check_version_consistent(self, other, other_idx, name):
1047
"""Check if a version in consistent in this and other.
1049
To be consistent it must have:
1052
* the same direct parents (by name, not index, and disregarding
1055
If present & correct return True;
1056
if not present in self return False;
1057
if inconsistent raise error."""
1058
this_idx = self._name_map.get(name, -1)
1060
if self._sha1s[this_idx] != other._sha1s[other_idx]:
1061
raise errors.WeaveTextDiffers(name, self, other)
1062
self_parents = self._parents[this_idx]
1063
other_parents = other._parents[other_idx]
1064
n1 = set([self._names[i] for i in self_parents])
1065
n2 = set([other._names[i] for i in other_parents])
1066
if not self._compatible_parents(n1, n2):
1067
raise WeaveParentMismatch("inconsistent parents "
1068
"for version {%s}: %s vs %s" % (name, n1, n2))
1074
@deprecated_method(zero_eight)
1075
def reweave(self, other, pb=None, msg=None):
1076
"""reweave has been superseded by plain use of join."""
1077
return self.join(other, pb, msg)
1079
def _reweave(self, other, pb, msg):
1080
"""Reweave self with other - internal helper for join().
1082
:param other: The other weave to merge
1083
:param pb: An optional progress bar, indicating how far done we are
1084
:param msg: An optional message for the progress
1086
new_weave = _reweave(self, other, pb=pb, msg=msg)
1087
self._copy_weave_content(new_weave)
1089
def _copy_weave_content(self, otherweave):
1090
"""adsorb the content from otherweave."""
1091
for attr in self.__slots__:
1092
if attr != '_weave_name':
1093
setattr(self, attr, copy(getattr(otherweave, attr)))
1096
class WeaveFile(Weave):
1097
"""A WeaveFile represents a Weave on disk and writes on change."""
1099
WEAVE_SUFFIX = '.weave'
1101
def __init__(self, name, transport, filemode=None, create=False, access_mode='w'):
1102
"""Create a WeaveFile.
1104
:param create: If not True, only open an existing knit.
1106
super(WeaveFile, self).__init__(name, access_mode)
1107
self._transport = transport
1108
self._filemode = filemode
1110
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
1111
except errors.NoSuchFile:
1117
def _add_lines(self, version_id, parents, lines, parent_texts):
1118
"""Add a version and save the weave."""
1119
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
1124
def _clone_text(self, new_version_id, old_version_id, parents):
1125
"""See VersionedFile.clone_text."""
1126
super(WeaveFile, self)._clone_text(new_version_id, old_version_id, parents)
1129
def copy_to(self, name, transport):
1130
"""See VersionedFile.copy_to()."""
1131
# as we are all in memory always, just serialise to the new place.
1133
write_weave_v5(self, sio)
1135
transport.put(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
1137
def create_empty(self, name, transport, filemode=None):
1138
return WeaveFile(name, transport, filemode, create=True)
1141
"""Save the weave."""
1142
self._check_write_ok()
1144
write_weave_v5(self, sio)
1146
self._transport.put(self._weave_name + WeaveFile.WEAVE_SUFFIX,
1152
"""See VersionedFile.get_suffixes()."""
1153
return [WeaveFile.WEAVE_SUFFIX]
1155
def join(self, other, pb=None, msg=None, version_ids=None,
1156
ignore_missing=False):
1157
"""Join other into self and save."""
1158
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
1162
@deprecated_function(zero_eight)
1163
def reweave(wa, wb, pb=None, msg=None):
1164
"""reweaving is deprecation, please just use weave.join()."""
1165
_reweave(wa, wb, pb, msg)
1167
def _reweave(wa, wb, pb=None, msg=None):
1168
"""Combine two weaves and return the result.
1170
This works even if a revision R has different parents in
1171
wa and wb. In the resulting weave all the parents are given.
1173
This is done by just building up a new weave, maintaining ordering
1174
of the versions in the two inputs. More efficient approaches
1175
might be possible but it should only be necessary to do
1176
this operation rarely, when a new previously ghost version is
1179
:param pb: An optional progress bar, indicating how far done we are
1180
:param msg: An optional message for the progress
1184
queue_a = range(wa.num_versions())
1185
queue_b = range(wb.num_versions())
1186
# first determine combined parents of all versions
1187
# map from version name -> all parent names
1188
combined_parents = _reweave_parent_graphs(wa, wb)
1189
mutter("combined parents: %r", combined_parents)
1190
order = topo_sort(combined_parents.iteritems())
1191
mutter("order to reweave: %r", order)
1196
for idx, name in enumerate(order):
1198
pb.update(msg, idx, len(order))
1199
if name in wa._name_map:
1200
lines = wa.get_lines(name)
1201
if name in wb._name_map:
1202
lines_b = wb.get_lines(name)
1203
if lines != lines_b:
1204
mutter('Weaves differ on content. rev_id {%s}', name)
1205
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
1207
lines = list(difflib.unified_diff(lines, lines_b,
1208
wa._weave_name, wb._weave_name))
1209
mutter('lines:\n%s', ''.join(lines))
1210
raise errors.WeaveTextDiffers(name, wa, wb)
1212
lines = wb.get_lines(name)
1213
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
1216
def _reweave_parent_graphs(wa, wb):
1217
"""Return combined parent ancestry for two weaves.
1219
Returned as a list of (version_name, set(parent_names))"""
1221
for weave in [wa, wb]:
1222
for idx, name in enumerate(weave._names):
1223
p = combined.setdefault(name, set())
1224
p.update(map(weave._idx_to_name, weave._parents[idx]))
1229
"""Show the weave's table-of-contents"""
1230
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
1231
for i in (6, 50, 10, 10):
1234
for i in range(w.num_versions()):
1237
parent_str = ' '.join(map(str, w._parents[i]))
1238
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
1242
def weave_stats(weave_file, pb):
1243
from bzrlib.weavefile import read_weave
1245
wf = file(weave_file, 'rb')
433
def merge(self, merge_versions):
434
"""Automerge and mark conflicts between versions.
436
This returns a sequence, each entry describing alternatives
437
for a chunk of the file. Each of the alternatives is given as
440
If there is a chunk of the file where there's no diagreement,
441
only one alternative is given.
444
# approach: find the included versions common to all the
446
raise NotImplementedError()
450
def _delta(self, included, lines):
451
"""Return changes from basis to new revision.
453
The old text for comparison is the union of included revisions.
455
This is used in inserting a new text.
457
Delta is returned as a sequence of
458
(weave1, weave2, newlines).
460
This indicates that weave1:weave2 of the old weave should be
461
replaced by the sequence of lines in newlines. Note that
462
these line numbers are positions in the total weave and don't
463
correspond to the lines in any extracted version, or even the
464
extracted union of included versions.
466
If line1=line2, this is a pure insert; if newlines=[] this is a
467
pure delete. (Similar to difflib.)
469
# basis a list of (origin, lineno, line)
472
for origin, lineno, line in self._extract(included):
473
basis_lineno.append(lineno)
474
basis_lines.append(line)
476
# add a sentinal, because we can also match against the final line
477
basis_lineno.append(len(self._l))
479
# XXX: which line of the weave should we really consider
480
# matches the end of the file? the current code says it's the
481
# last line of the weave?
483
from difflib import SequenceMatcher
484
s = SequenceMatcher(None, basis_lines, lines)
486
# TODO: Perhaps return line numbers from composed weave as well?
488
for tag, i1, i2, j1, j2 in s.get_opcodes():
489
##print tag, i1, i2, j1, j2
494
# i1,i2 are given in offsets within basis_lines; we need to map them
495
# back to offsets within the entire weave
496
real_i1 = basis_lineno[i1]
497
real_i2 = basis_lineno[i2]
501
assert j2 <= len(lines)
503
yield real_i1, real_i2, lines[j1:j2]
507
def weave_info(filename, out):
508
"""Show some text information about the weave."""
509
from weavefile import read_weave
510
wf = file(filename, 'rb')
1246
511
w = read_weave(wf)
1247
512
# FIXME: doesn't work on pipes
1248
513
weave_size = wf.tell()
514
print >>out, "weave file size %d bytes" % weave_size
515
print >>out, "weave contains %d versions" % len(w._v)
1252
for i in range(vers):
1253
pb.update('checking sizes', i, vers)
1254
for origin, lineno, line in w._extract([i]):
1259
print 'versions %9d' % vers
1260
print 'weave file %9d bytes' % weave_size
1261
print 'total contents %9d bytes' % total
1262
print 'compression ratio %9.2fx' % (float(total) / float(weave_size))
1265
print 'average size %9d bytes' % avg
1266
print 'relative size %9.2fx' % (float(weave_size) / float(avg))
518
print '%6s %6s %8s %40s %20s' % ('ver', 'lines', 'bytes', 'sha1', 'parents')
519
for i in (6, 6, 8, 40, 20):
522
for i in range(len(w._v)):
525
bytes = sum((len(a) for a in text))
527
print '%6d %6d %8d %40s' % (i, lines, bytes, sha1),
533
print >>out, "versions total %d bytes" % total
534
print >>out, "compression ratio %.3f" % (float(total)/float(weave_size))
added
removed
bzrlib/weave.py
