190
137
should be no way to get an earlier version deleting a later
194
Text of the weave; list of control instruction tuples and strings.
197
144
List of parents, indexed by version number.
198
145
It is only necessary to store the minimal set of parents for
199
146
each version; the parent's parents are implied.
202
List of hex SHA-1 of each version.
205
List of symbolic names for each version. Each should be unique.
208
For each name, the version number.
211
Descriptive name of this weave; typically the filename if known.
149
List of hex SHA-1 of each version, or None if not recorded.
215
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
216
'_weave_name', '_matcher', '_allow_reserved']
218
def __init__(self, weave_name=None, access_mode='w', matcher=None,
219
get_scope=None, allow_reserved=False):
222
:param get_scope: A callable that returns an opaque object to be used
223
for detecting when this weave goes out of scope (should stop
224
answering requests or allowing mutation).
226
super(Weave, self).__init__()
232
self._weave_name = weave_name
234
self._matcher = bzrlib.patiencediff.PatienceSequenceMatcher
236
self._matcher = matcher
237
if get_scope is None:
238
get_scope = lambda:None
239
self._get_scope = get_scope
240
self._scope = get_scope()
241
self._access_mode = access_mode
242
self._allow_reserved = allow_reserved
245
return "Weave(%r)" % self._weave_name
247
def _check_write_ok(self):
248
"""Is the versioned file marked as 'finished' ? Raise if it is."""
249
if self._get_scope() != self._scope:
250
raise errors.OutSideTransaction()
251
if self._access_mode != 'w':
252
raise errors.ReadOnlyObjectDirtiedError(self)
255
"""Return a deep copy of self.
257
The copy can be modified without affecting the original weave."""
259
other._weave = self._weave[:]
260
other._parents = self._parents[:]
261
other._sha1s = self._sha1s[:]
262
other._names = self._names[:]
263
other._name_map = self._name_map.copy()
264
other._weave_name = self._weave_name
267
157
def __eq__(self, other):
268
158
if not isinstance(other, Weave):
270
return self._parents == other._parents \
271
and self._weave == other._weave \
272
and self._sha1s == other._sha1s
160
return self._v == other._v \
161
and self._l == other._l
274
164
def __ne__(self, other):
275
165
return not self.__eq__(other)
277
def _idx_to_name(self, version):
278
return self._names[version]
280
def _lookup(self, name):
281
"""Convert symbolic version name to index."""
282
if not self._allow_reserved:
283
self.check_not_reserved_id(name)
285
return self._name_map[name]
287
raise RevisionNotPresent(name, self._weave_name)
290
"""See VersionedFile.versions."""
291
return self._names[:]
293
def has_version(self, version_id):
294
"""See VersionedFile.has_version."""
295
return (version_id in self._name_map)
297
__contains__ = has_version
299
def get_record_stream(self, versions, ordering, include_delta_closure):
300
"""Get a stream of records for versions.
302
:param versions: The versions to include. Each version is a tuple
304
:param ordering: Either 'unordered' or 'topological'. A topologically
305
sorted stream has compression parents strictly before their
307
:param include_delta_closure: If True then the closure across any
308
compression parents will be included (in the opaque data).
309
:return: An iterator of ContentFactory objects, each of which is only
310
valid until the iterator is advanced.
312
versions = [version[-1] for version in versions]
313
if ordering == 'topological':
314
parents = self.get_parent_map(versions)
315
new_versions = tsort.topo_sort(parents)
316
new_versions.extend(set(versions).difference(set(parents)))
317
versions = new_versions
318
elif ordering == 'groupcompress':
319
parents = self.get_parent_map(versions)
320
new_versions = sort_groupcompress(parents)
321
new_versions.extend(set(versions).difference(set(parents)))
322
versions = new_versions
323
for version in versions:
325
yield WeaveContentFactory(version, self)
327
yield AbsentContentFactory((version,))
329
def get_parent_map(self, version_ids):
330
"""See VersionedFile.get_parent_map."""
332
for version_id in version_ids:
333
if version_id == NULL_REVISION:
338
map(self._idx_to_name,
339
self._parents[self._lookup(version_id)]))
340
except RevisionNotPresent:
342
result[version_id] = parents
345
def get_parents_with_ghosts(self, version_id):
346
raise NotImplementedError(self.get_parents_with_ghosts)
348
def insert_record_stream(self, stream):
349
"""Insert a record stream into this versioned file.
351
:param stream: A stream of records to insert.
353
:seealso VersionedFile.get_record_stream:
356
for record in stream:
357
# Raise an error when a record is missing.
358
if record.storage_kind == 'absent':
359
raise RevisionNotPresent([record.key[0]], self)
360
# adapt to non-tuple interface
361
parents = [parent[0] for parent in record.parents]
362
if (record.storage_kind == 'fulltext'
363
or record.storage_kind == 'chunked'):
364
self.add_lines(record.key[0], parents,
365
osutils.chunks_to_lines(record.get_bytes_as('chunked')))
367
adapter_key = record.storage_kind, 'fulltext'
369
adapter = adapters[adapter_key]
371
adapter_factory = adapter_registry.get(adapter_key)
372
adapter = adapter_factory(self)
373
adapters[adapter_key] = adapter
374
lines = split_lines(adapter.get_bytes(record))
376
self.add_lines(record.key[0], parents, lines)
377
except RevisionAlreadyPresent:
380
def _check_repeated_add(self, name, parents, text, sha1):
381
"""Check that a duplicated add is OK.
383
If it is, return the (old) index; otherwise raise an exception.
385
idx = self._lookup(name)
386
if sorted(self._parents[idx]) != sorted(parents) \
387
or sha1 != self._sha1s[idx]:
388
raise RevisionAlreadyPresent(name, self._weave_name)
391
def _add_lines(self, version_id, parents, lines, parent_texts,
392
left_matching_blocks, nostore_sha, random_id, check_content):
393
"""See VersionedFile.add_lines."""
394
idx = self._add(version_id, lines, map(self._lookup, parents),
395
nostore_sha=nostore_sha)
396
return sha_strings(lines), sum(map(len, lines)), idx
398
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
168
def add(self, parents, text):
399
169
"""Add a single text on top of the weave.
401
171
Returns the index number of the newly added version.
404
Symbolic name for this version.
405
(Typically the revision-id of the revision that added it.)
406
If None, a name will be allocated based on the hash. (sha1:SHAHASH)
409
174
List or set of direct parent version numbers.
412
Sequence of lines to be added in the new version.
414
:param nostore_sha: See VersionedFile.add_lines.
416
self._check_lines_not_unicode(lines)
417
self._check_lines_are_lines(lines)
419
sha1 = sha_strings(lines)
420
if sha1 == nostore_sha:
421
raise errors.ExistingContent
422
if version_id is None:
423
version_id = "sha1:" + sha1
424
if version_id in self._name_map:
425
return self._check_repeated_add(version_id, parents, lines, sha1)
427
self._check_versions(parents)
428
## self._check_lines(lines)
429
new_version = len(self._parents)
431
# if we abort after here the (in-memory) weave will be corrupt because only
432
# some fields are updated
433
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
434
# - Robert Collins 20060226
435
self._parents.append(parents[:])
177
Sequence of lines to be added in the new version."""
178
## self._check_versions(parents)
179
## self._check_lines(text)
189
# TODO: It'd probably be faster to append things on to a new
190
# list rather than modifying the existing one, which is likely
191
# to cause a lot of copying.
194
ancestors = self.inclusions(parents)
195
delta = self._delta(ancestors, text)
197
# offset gives the number of lines that have been inserted
198
# into the weave up to the current point; if the original edit instruction
199
# says to change line A then we actually change (A+offset)
202
for i1, i2, newlines in delta:
205
assert i2 <= len(self._l)
207
# the deletion and insertion are handled separately.
208
# first delete the region.
210
self._l.insert(i1+offset, ('[', idx))
211
self._l.insert(i2+offset+1, (']', idx))
216
# there may have been a deletion spanning up to
217
# i2; we want to insert after this region to make sure
218
# we don't destroy ourselves
220
self._l[i:i] = [('{', idx)] \
223
offset += 2 + len(newlines)
225
self._addversion(parents)
227
# special case; adding with no parents revision; can do this
228
# more quickly by just appending unconditionally
229
self._l.append(('{', idx))
231
self._l.append(('}', idx))
233
self._addversion(None)
436
235
self._sha1s.append(sha1)
437
self._names.append(version_id)
438
self._name_map[version_id] = new_version
442
# special case; adding with no parents revision; can do
443
# this more quickly by just appending unconditionally.
444
# even more specially, if we're adding an empty text we
445
# need do nothing at all.
447
self._weave.append(('{', new_version))
448
self._weave.extend(lines)
449
self._weave.append(('}', None))
452
if len(parents) == 1:
453
pv = list(parents)[0]
454
if sha1 == self._sha1s[pv]:
455
# special case: same as the single parent
459
ancestors = self._inclusions(parents)
463
# basis a list of (origin, lineno, line)
466
for origin, lineno, line in self._extract(ancestors):
467
basis_lineno.append(lineno)
468
basis_lines.append(line)
470
# another small special case: a merge, producing the same text
472
if lines == basis_lines:
475
# add a sentinel, because we can also match against the final line
476
basis_lineno.append(len(self._weave))
478
# XXX: which line of the weave should we really consider
479
# matches the end of the file? the current code says it's the
480
# last line of the weave?
482
#print 'basis_lines:', basis_lines
483
#print 'new_lines: ', lines
485
s = self._matcher(None, basis_lines, lines)
487
# offset gives the number of lines that have been inserted
488
# into the weave up to the current point; if the original edit instruction
489
# says to change line A then we actually change (A+offset)
492
for tag, i1, i2, j1, j2 in s.get_opcodes():
493
# i1,i2 are given in offsets within basis_lines; we need to map them
494
# back to offsets within the entire weave
495
#print 'raw match', tag, i1, i2, j1, j2
498
i1 = basis_lineno[i1]
499
i2 = basis_lineno[i2]
500
# the deletion and insertion are handled separately.
501
# first delete the region.
503
self._weave.insert(i1+offset, ('[', new_version))
504
self._weave.insert(i2+offset+1, (']', new_version))
508
# there may have been a deletion spanning up to
509
# i2; we want to insert after this region to make sure
510
# we don't destroy ourselves
512
self._weave[i:i] = ([('{', new_version)]
515
offset += 2 + (j2 - j1)
518
def _inclusions(self, versions):
240
def inclusions(self, versions):
519
241
"""Return set of all ancestors of given version(s)."""
520
if not len(versions):
247
# include all its parents
252
raise ValueError("version %d not present in weave" % v)
255
def minimal_parents(self, version):
256
"""Find the minimal set of parents for the version."""
257
included = self._v[version]
523
for v in xrange(max(versions), 0, -1):
525
# include all its parents
526
i.update(self._parents[v])
528
## except IndexError:
529
## raise ValueError("version %d not present in weave" % v)
531
def get_ancestry(self, version_ids, topo_sorted=True):
532
"""See VersionedFile.get_ancestry."""
533
if isinstance(version_ids, basestring):
534
version_ids = [version_ids]
535
i = self._inclusions([self._lookup(v) for v in version_ids])
536
return [self._idx_to_name(v) for v in i]
262
li.sort(reverse=True)
270
gotit.update(self.inclusions(pv))
272
assert mininc[0] >= 0
273
assert mininc[-1] < version
277
def _addversion(self, parents):
279
self._v.append(parents)
281
self._v.append(set())
538
284
def _check_lines(self, text):
539
285
if not isinstance(text, list):
543
289
if not isinstance(l, basestring):
544
290
raise ValueError("text line should be a string or unicode, not %s"
549
295
def _check_versions(self, indexes):
550
296
"""Check everything in the sequence of indexes is valid"""
551
297
for i in indexes:
554
300
except IndexError:
555
301
raise IndexError("invalid version number %r" % i)
557
def _compatible_parents(self, my_parents, other_parents):
558
"""During join check that other_parents are joinable with my_parents.
560
Joinable is defined as 'is a subset of' - supersets may require
561
regeneration of diffs, but subsets do not.
563
return len(other_parents.difference(my_parents)) == 0
565
def annotate(self, version_id):
566
"""Return a list of (version-id, line) tuples for version_id.
304
def annotate(self, index):
305
return list(self.annotate_iter(index))
308
def annotate_iter(self, version):
309
"""Yield list of (index-id, line) pairs for the specified version.
568
311
The index indicates when the line originated in the weave."""
569
incls = [self._lookup(version_id)]
570
return [(self._idx_to_name(origin), text) for origin, lineno, text in
571
self._extract(incls)]
573
def iter_lines_added_or_present_in_versions(self, version_ids=None,
575
"""See VersionedFile.iter_lines_added_or_present_in_versions()."""
576
if version_ids is None:
577
version_ids = self.versions()
578
version_ids = set(version_ids)
579
for lineno, inserted, deletes, line in self._walk_internal(version_ids):
580
if inserted not in version_ids: continue
582
yield line + '\n', inserted
586
def _walk_internal(self, version_ids=None):
587
"""Helper method for weave actions."""
312
for origin, lineno, text in self._extract([version]):
320
(lineno, insert, deletes, text)
321
for each literal line.
592
327
lineno = 0 # line of weave, 0-based
594
for l in self._weave:
595
if l.__class__ == tuple:
330
if isinstance(l, tuple):
599
istack.append(self._names[v])
603
dset.add(self._names[v])
605
dset.remove(self._names[v])
607
raise WeaveFormatError('unexpected instruction %r' % v)
609
yield lineno, istack[-1], frozenset(dset), l
343
raise WeaveFormatError('unexpected instruction %r'
346
assert isinstance(l, basestring)
348
yield lineno, istack[-1], dset, l
353
def _extract(self, versions):
354
"""Yield annotation of lines in included set.
356
Yields a sequence of tuples (origin, lineno, text), where
357
origin is the origin version, lineno the index in the weave,
358
and text the text of the line.
360
The set typically but not necessarily corresponds to a version.
362
included = self.inclusions(versions)
367
lineno = 0 # line of weave, 0-based
373
WFE = WeaveFormatError
376
if isinstance(l, tuple):
380
assert v not in istack
395
assert isinstance(l, basestring)
397
isactive = (not dset) and istack and (istack[-1] in included)
399
result.append((istack[-1], lineno, l))
613
raise WeaveFormatError("unclosed insertion blocks "
614
"at end of weave: %s" % istack)
403
raise WFE("unclosed insertion blocks at end of weave",
616
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
406
raise WFE("unclosed deletion blocks at end of weave",
413
def get_iter(self, version):
414
"""Yield lines for the specified version."""
415
for origin, lineno, line in self._extract([version]):
419
def get(self, index):
420
return list(self.get_iter(index))
423
def mash_iter(self, included):
424
"""Return composed version of multiple included versions."""
425
for origin, lineno, text in self._extract(included):
429
def dump(self, to_file):
430
from pprint import pprint
431
print >>to_file, "Weave._l = ",
432
pprint(self._l, to_file)
433
print >>to_file, "Weave._v = ",
434
pprint(self._v, to_file)
438
def numversions(self):
440
assert l == len(self._sha1s)
444
def check(self, progress_bar=None):
445
# check no circular inclusions
446
for version in range(self.numversions()):
447
inclusions = list(self._v[version])
450
if inclusions[-1] >= version:
451
raise WeaveFormatError("invalid included version %d for index %d"
452
% (inclusions[-1], version))
454
# try extracting all versions; this is a bit slow and parallel
455
# extraction could be used
457
nv = self.numversions()
458
for version in range(nv):
460
progress_bar.update('checking text', version, nv)
462
for l in self.get_iter(version):
465
expected = self._sha1s[version]
467
raise WeaveError("mismatched sha1 for version %d; "
468
"got %s, expected %s"
469
% (version, hd, expected))
471
# TODO: check insertions are properly nested, that there are
472
# no lines outside of insertion blocks, that deletions are
473
# properly paired, etc.
477
def merge(self, merge_versions):
478
"""Automerge and mark conflicts between versions.
480
This returns a sequence, each entry describing alternatives
481
for a chunk of the file. Each of the alternatives is given as
484
If there is a chunk of the file where there's no diagreement,
485
only one alternative is given.
488
# approach: find the included versions common to all the
490
raise NotImplementedError()
494
def _delta(self, included, lines):
495
"""Return changes from basis to new revision.
497
The old text for comparison is the union of included revisions.
499
This is used in inserting a new text.
501
Delta is returned as a sequence of
502
(weave1, weave2, newlines).
504
This indicates that weave1:weave2 of the old weave should be
505
replaced by the sequence of lines in newlines. Note that
506
these line numbers are positions in the total weave and don't
507
correspond to the lines in any extracted version, or even the
508
extracted union of included versions.
510
If line1=line2, this is a pure insert; if newlines=[] this is a
511
pure delete. (Similar to difflib.)
513
# basis a list of (origin, lineno, line)
516
for origin, lineno, line in self._extract(included):
517
basis_lineno.append(lineno)
518
basis_lines.append(line)
520
# add a sentinal, because we can also match against the final line
521
basis_lineno.append(len(self._l))
523
# XXX: which line of the weave should we really consider
524
# matches the end of the file? the current code says it's the
525
# last line of the weave?
527
from difflib import SequenceMatcher
528
s = SequenceMatcher(None, basis_lines, lines)
530
# TODO: Perhaps return line numbers from composed weave as well?
532
for tag, i1, i2, j1, j2 in s.get_opcodes():
533
##print tag, i1, i2, j1, j2
538
# i1,i2 are given in offsets within basis_lines; we need to map them
539
# back to offsets within the entire weave
540
real_i1 = basis_lineno[i1]
541
real_i2 = basis_lineno[i2]
545
assert j2 <= len(lines)
547
yield real_i1, real_i2, lines[j1:j2]
619
551
def plan_merge(self, ver_a, ver_b):
620
552
"""Return pseudo-annotation indicating how the two versions merge.
660
592
# not in either revision
661
593
yield 'irrelevant', line
663
def _extract(self, versions):
664
"""Yield annotation of lines in included set.
666
Yields a sequence of tuples (origin, lineno, text), where
667
origin is the origin version, lineno the index in the weave,
668
and text the text of the line.
670
The set typically but not necessarily corresponds to a version.
673
if not isinstance(i, int):
676
included = self._inclusions(versions)
682
lineno = 0 # line of weave, 0-based
688
WFE = WeaveFormatError
691
# 449 0 4474.6820 2356.5590 bzrlib.weave:556(_extract)
692
# +285282 0 1676.8040 1676.8040 +<isinstance>
693
# 1.6 seconds in 'isinstance'.
694
# changing the first isinstance:
695
# 449 0 2814.2660 1577.1760 bzrlib.weave:556(_extract)
696
# +140414 0 762.8050 762.8050 +<isinstance>
697
# note that the inline time actually dropped (less function calls)
698
# and total processing time was halved.
699
# we're still spending ~1/4 of the method in isinstance though.
700
# so lets hard code the acceptable string classes we expect:
701
# 449 0 1202.9420 786.2930 bzrlib.weave:556(_extract)
702
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
704
# yay, down to ~1/4 the initial extract time, and our inline time
705
# has shrunk again, with isinstance no longer dominating.
706
# tweaking the stack inclusion test to use a set gives:
707
# 449 0 1122.8030 713.0080 bzrlib.weave:556(_extract)
708
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
710
# - a 5% win, or possibly just noise. However with large istacks that
711
# 'in' test could dominate, so I'm leaving this change in place -
712
# when its fast enough to consider profiling big datasets we can review.
717
for l in self._weave:
718
if l.__class__ == tuple:
725
iset.remove(istack.pop())
595
yield 'unchanged', '' # terminator
599
def weave_merge(self, plan):
604
for state, line in plan:
605
if state == 'unchanged' or state == 'killed-both':
606
# resync and flush queued conflicts changes if any
607
if not lines_a and not lines_b:
609
elif ch_a and not ch_b:
611
for l in lines_a: yield l
612
elif ch_b and not ch_a:
613
for l in lines_b: yield l
614
elif lines_a == lines_b:
615
for l in lines_a: yield l
733
raise AssertionError()
736
isactive = (not dset) and istack and (istack[-1] in included)
738
result.append((istack[-1], lineno, l))
741
raise WeaveFormatError("unclosed insertion blocks "
742
"at end of weave: %s" % istack)
744
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
748
def _maybe_lookup(self, name_or_index):
749
"""Convert possible symbolic name to index, or pass through indexes.
753
if isinstance(name_or_index, (int, long)):
756
return self._lookup(name_or_index)
758
def get_lines(self, version_id):
759
"""See VersionedFile.get_lines()."""
760
int_index = self._maybe_lookup(version_id)
761
result = [line for (origin, lineno, line) in self._extract([int_index])]
762
expected_sha1 = self._sha1s[int_index]
763
measured_sha1 = sha_strings(result)
764
if measured_sha1 != expected_sha1:
765
raise errors.WeaveInvalidChecksum(
766
'file %s, revision %s, expected: %s, measured %s'
767
% (self._weave_name, version_id,
768
expected_sha1, measured_sha1))
771
def get_sha1s(self, version_ids):
772
"""See VersionedFile.get_sha1s()."""
774
for v in version_ids:
775
result[v] = self._sha1s[self._lookup(v)]
778
def num_versions(self):
779
"""How many versions are in this weave?"""
780
l = len(self._parents)
783
__len__ = num_versions
785
def check(self, progress_bar=None):
786
# TODO evaluate performance hit of using string sets in this routine.
787
# TODO: check no circular inclusions
788
# TODO: create a nested progress bar
789
for version in range(self.num_versions()):
790
inclusions = list(self._parents[version])
793
if inclusions[-1] >= version:
794
raise WeaveFormatError("invalid included version %d for index %d"
795
% (inclusions[-1], version))
797
# try extracting all versions; parallel extraction is used
798
nv = self.num_versions()
803
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
804
# The problem is that set membership is much more expensive
805
name = self._idx_to_name(i)
808
new_inc = set([name])
809
for p in self._parents[i]:
810
new_inc.update(inclusions[self._idx_to_name(p)])
812
if set(new_inc) != set(self.get_ancestry(name)):
813
raise AssertionError(
815
% (set(new_inc), set(self.get_ancestry(name))))
816
inclusions[name] = new_inc
818
nlines = len(self._weave)
820
update_text = 'checking weave'
822
short_name = os.path.basename(self._weave_name)
823
update_text = 'checking %s' % (short_name,)
824
update_text = update_text[:25]
826
for lineno, insert, deleteset, line in self._walk_internal():
828
progress_bar.update(update_text, lineno, nlines)
830
for name, name_inclusions in inclusions.items():
831
# The active inclusion must be an ancestor,
832
# and no ancestors must have deleted this line,
833
# because we don't support resurrection.
834
if (insert in name_inclusions) and not (deleteset & name_inclusions):
835
sha1s[name].update(line)
838
version = self._idx_to_name(i)
839
hd = sha1s[version].hexdigest()
840
expected = self._sha1s[i]
842
raise errors.WeaveInvalidChecksum(
843
"mismatched sha1 for version %s: "
844
"got %s, expected %s"
845
% (version, hd, expected))
847
# TODO: check insertions are properly nested, that there are
848
# no lines outside of insertion blocks, that deletions are
849
# properly paired, etc.
851
def _imported_parents(self, other, other_idx):
852
"""Return list of parents in self corresponding to indexes in other."""
854
for parent_idx in other._parents[other_idx]:
855
parent_name = other._names[parent_idx]
856
if parent_name not in self._name_map:
857
# should not be possible
858
raise WeaveError("missing parent {%s} of {%s} in %r"
859
% (parent_name, other._name_map[other_idx], self))
860
new_parents.append(self._name_map[parent_name])
863
def _check_version_consistent(self, other, other_idx, name):
864
"""Check if a version in consistent in this and other.
866
To be consistent it must have:
869
* the same direct parents (by name, not index, and disregarding
872
If present & correct return True;
873
if not present in self return False;
874
if inconsistent raise error."""
875
this_idx = self._name_map.get(name, -1)
877
if self._sha1s[this_idx] != other._sha1s[other_idx]:
878
raise errors.WeaveTextDiffers(name, self, other)
879
self_parents = self._parents[this_idx]
880
other_parents = other._parents[other_idx]
881
n1 = set([self._names[i] for i in self_parents])
882
n2 = set([other._names[i] for i in other_parents])
883
if not self._compatible_parents(n1, n2):
884
raise WeaveParentMismatch("inconsistent parents "
885
"for version {%s}: %s vs %s" % (name, n1, n2))
891
def _reweave(self, other, pb, msg):
892
"""Reweave self with other - internal helper for join().
894
:param other: The other weave to merge
895
:param pb: An optional progress bar, indicating how far done we are
896
:param msg: An optional message for the progress
898
new_weave = _reweave(self, other, pb=pb, msg=msg)
899
self._copy_weave_content(new_weave)
901
def _copy_weave_content(self, otherweave):
902
"""adsorb the content from otherweave."""
903
for attr in self.__slots__:
904
if attr != '_weave_name':
905
setattr(self, attr, copy(getattr(otherweave, attr)))
908
class WeaveFile(Weave):
909
"""A WeaveFile represents a Weave on disk and writes on change."""
911
WEAVE_SUFFIX = '.weave'
913
def __init__(self, name, transport, filemode=None, create=False, access_mode='w', get_scope=None):
914
"""Create a WeaveFile.
916
:param create: If not True, only open an existing knit.
918
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope,
919
allow_reserved=False)
920
self._transport = transport
921
self._filemode = filemode
923
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
924
except errors.NoSuchFile:
930
def _add_lines(self, version_id, parents, lines, parent_texts,
931
left_matching_blocks, nostore_sha, random_id, check_content):
932
"""Add a version and save the weave."""
933
self.check_not_reserved_id(version_id)
934
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
935
parent_texts, left_matching_blocks, nostore_sha, random_id,
940
def copy_to(self, name, transport):
941
"""See VersionedFile.copy_to()."""
942
# as we are all in memory always, just serialise to the new place.
944
write_weave_v5(self, sio)
946
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
949
"""Save the weave."""
950
self._check_write_ok()
952
write_weave_v5(self, sio)
954
bytes = sio.getvalue()
955
path = self._weave_name + WeaveFile.WEAVE_SUFFIX
957
self._transport.put_bytes(path, bytes, self._filemode)
958
except errors.NoSuchFile:
959
self._transport.mkdir(dirname(path))
960
self._transport.put_bytes(path, bytes, self._filemode)
964
"""See VersionedFile.get_suffixes()."""
965
return [WeaveFile.WEAVE_SUFFIX]
967
def insert_record_stream(self, stream):
968
super(WeaveFile, self).insert_record_stream(stream)
972
def _reweave(wa, wb, pb=None, msg=None):
973
"""Combine two weaves and return the result.
975
This works even if a revision R has different parents in
976
wa and wb. In the resulting weave all the parents are given.
978
This is done by just building up a new weave, maintaining ordering
979
of the versions in the two inputs. More efficient approaches
980
might be possible but it should only be necessary to do
981
this operation rarely, when a new previously ghost version is
984
:param pb: An optional progress bar, indicating how far done we are
985
:param msg: An optional message for the progress
989
queue_a = range(wa.num_versions())
990
queue_b = range(wb.num_versions())
991
# first determine combined parents of all versions
992
# map from version name -> all parent names
993
combined_parents = _reweave_parent_graphs(wa, wb)
994
mutter("combined parents: %r", combined_parents)
995
order = tsort.topo_sort(combined_parents.iteritems())
996
mutter("order to reweave: %r", order)
1001
for idx, name in enumerate(order):
1003
pb.update(msg, idx, len(order))
1004
if name in wa._name_map:
1005
lines = wa.get_lines(name)
1006
if name in wb._name_map:
1007
lines_b = wb.get_lines(name)
1008
if lines != lines_b:
1009
mutter('Weaves differ on content. rev_id {%s}', name)
1010
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
1012
lines = list(difflib.unified_diff(lines, lines_b,
1013
wa._weave_name, wb._weave_name))
1014
mutter('lines:\n%s', ''.join(lines))
1015
raise errors.WeaveTextDiffers(name, wa, wb)
1017
lines = wb.get_lines(name)
1018
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
1022
def _reweave_parent_graphs(wa, wb):
1023
"""Return combined parent ancestry for two weaves.
1025
Returned as a list of (version_name, set(parent_names))"""
1027
for weave in [wa, wb]:
1028
for idx, name in enumerate(weave._names):
1029
p = combined.setdefault(name, set())
1030
p.update(map(weave._idx_to_name, weave._parents[idx]))
618
for l in lines_a: yield l
620
for l in lines_b: yield l
627
if state == 'unchanged':
630
elif state == 'killed-a':
633
elif state == 'killed-b':
636
elif state == 'new-a':
639
elif state == 'new-b':
643
assert state in ('irrelevant', 'ghost-a', 'ghost-b', 'killed-base',
653
def weave_info(filename, out):
654
"""Show some text information about the weave."""
655
from weavefile import read_weave
656
wf = file(filename, 'rb')
658
# FIXME: doesn't work on pipes
659
weave_size = wf.tell()
660
print >>out, "weave file size %d bytes" % weave_size
661
print >>out, "weave contains %d versions" % len(w._v)
664
print '%6s %6s %8s %40s %20s' % ('ver', 'lines', 'bytes', 'sha1', 'parents')
665
for i in (6, 6, 8, 40, 20):
668
for i in range(len(w._v)):
671
bytes = sum((len(a) for a in text))
673
print '%6d %6d %8d %40s' % (i, lines, bytes, sha1),
679
print >>out, "versions total %d bytes" % total
680
print >>out, "compression ratio %.3f" % (float(total)/float(weave_size))
684
print """bzr weave tool
686
Experimental tool for weave algorithm.
690
Create an empty weave file
691
weave get WEAVEFILE VERSION
692
Write out specified version.
693
weave check WEAVEFILE
694
Check consistency of all versions.
696
Display table of contents.
697
weave add WEAVEFILE [BASE...] < NEWTEXT
698
Add NEWTEXT, with specified parent versions.
699
weave annotate WEAVEFILE VERSION
700
Display origin of each line.
701
weave mash WEAVEFILE VERSION...
702
Display composite of all selected versions.
703
weave merge WEAVEFILE VERSION1 VERSION2 > OUT
704
Auto-merge two versions and display conflicts.
708
% weave init foo.weave
710
% weave add foo.weave < foo.txt
713
(create updated version)
715
% weave get foo.weave 0 | diff -u - foo.txt
716
% weave add foo.weave 0 < foo.txt
719
% weave get foo.weave 0 > foo.txt (create forked version)
721
% weave add foo.weave 0 < foo.txt
724
% weave merge foo.weave 1 2 > foo.txt (merge them)
725
% vi foo.txt (resolve conflicts)
726
% weave add foo.weave 1 2 < foo.txt (commit merged version)
735
from weavefile import write_weave, read_weave
736
from bzrlib.progress import ProgressBar
744
return read_weave(file(argv[2], 'rb'))
750
# at the moment, based on everything in the file
751
parents = map(int, argv[3:])
752
lines = sys.stdin.readlines()
753
ver = w.add(parents, lines)
754
write_weave(w, file(argv[2], 'wb'))
755
print 'added version %d' % ver
758
if os.path.exists(fn):
759
raise IOError("file exists")
761
write_weave(w, file(fn, 'wb'))
762
elif cmd == 'get': # get one version
764
sys.stdout.writelines(w.get_iter(int(argv[3])))
766
elif cmd == 'mash': # get composite
768
sys.stdout.writelines(w.mash_iter(map(int, argv[3:])))
770
elif cmd == 'annotate':
772
# newline is added to all lines regardless; too hard to get
773
# reasonable formatting otherwise
775
for origin, text in w.annotate(int(argv[3])):
776
text = text.rstrip('\r\n')
778
print ' | %s' % (text)
780
print '%5d | %s' % (origin, text)
784
weave_info(argv[2], sys.stdout)
792
elif cmd == 'inclusions':
794
print ' '.join(map(str, w.inclusions([int(argv[3])])))
796
elif cmd == 'parents':
798
print ' '.join(map(str, w._v[int(argv[3])]))
800
elif cmd == 'plan-merge':
802
for state, line in w.plan_merge(int(argv[3]), int(argv[4])):
804
print '%14s | %s' % (state, line),
808
p = w.plan_merge(int(argv[3]), int(argv[4]))
809
sys.stdout.writelines(w.weave_merge(p))
811
elif cmd == 'mash-merge':
817
v1, v2 = map(int, argv[3:5])
819
basis = w.inclusions([v1]).intersection(w.inclusions([v2]))
821
base_lines = list(w.mash_iter(basis))
822
a_lines = list(w.get(v1))
823
b_lines = list(w.get(v2))
825
from bzrlib.merge3 import Merge3
826
m3 = Merge3(base_lines, a_lines, b_lines)
828
name_a = 'version %d' % v1
829
name_b = 'version %d' % v2
830
sys.stdout.writelines(m3.merge_lines(name_a=name_a, name_b=name_b))
832
raise ValueError('unknown command %r' % cmd)
835
if __name__ == '__main__':
837
sys.exit(main(sys.argv))
added
removed
bzrlib/weave.py
