160
169
each version; the parent's parents are implied.
163
List of hex SHA-1 of each version, or None if not recorded.
172
List of hex SHA-1 of each version.
175
List of symbolic names for each version. Each should be unique.
178
For each name, the version number.
181
Descriptive name of this weave; typically the filename if known.
166
__slots__ = ['_weave', '_parents', '_sha1s']
185
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
186
'_weave_name', '_matcher']
188
def __init__(self, weave_name=None, access_mode='w', matcher=None, get_scope=None):
191
:param get_scope: A callable that returns an opaque object to be used
192
for detecting when this weave goes out of scope (should stop
193
answering requests or allowing mutation).
195
super(Weave, self).__init__(access_mode)
170
197
self._parents = []
201
self._weave_name = weave_name
203
self._matcher = bzrlib.patiencediff.PatienceSequenceMatcher
205
self._matcher = matcher
206
if get_scope is None:
207
get_scope = lambda:None
208
self._get_scope = get_scope
209
self._scope = get_scope()
210
self._access_mode = access_mode
213
return "Weave(%r)" % self._weave_name
215
def _check_write_ok(self):
216
"""Is the versioned file marked as 'finished' ? Raise if it is."""
217
if self._get_scope() != self._scope:
218
raise errors.OutSideTransaction()
219
if self._access_mode != 'w':
220
raise errors.ReadOnlyObjectDirtiedError(self)
223
"""Return a deep copy of self.
225
The copy can be modified without affecting the original weave."""
227
other._weave = self._weave[:]
228
other._parents = self._parents[:]
229
other._sha1s = self._sha1s[:]
230
other._names = self._names[:]
231
other._name_map = self._name_map.copy()
232
other._weave_name = self._weave_name
174
235
def __eq__(self, other):
175
236
if not isinstance(other, Weave):
177
238
return self._parents == other._parents \
178
and self._weave == other._weave
239
and self._weave == other._weave \
240
and self._sha1s == other._sha1s
181
242
def __ne__(self, other):
182
243
return not self.__eq__(other)
185
def add(self, parents, text):
245
def _idx_to_name(self, version):
246
return self._names[version]
248
def _lookup(self, name):
249
"""Convert symbolic version name to index."""
250
self.check_not_reserved_id(name)
252
return self._name_map[name]
254
raise RevisionNotPresent(name, self._weave_name)
257
"""See VersionedFile.versions."""
258
return self._names[:]
260
def has_version(self, version_id):
261
"""See VersionedFile.has_version."""
262
return (version_id in self._name_map)
264
__contains__ = has_version
266
def get_parent_map(self, version_ids):
267
"""See VersionedFile.get_parent_map."""
269
for version_id in version_ids:
271
result[version_id] = tuple(
272
map(self._idx_to_name, self._parents[self._lookup(version_id)]))
273
except RevisionNotPresent:
277
def get_parents_with_ghosts(self, version_id):
278
raise NotImplementedError(self.get_parents_with_ghosts)
280
def _check_repeated_add(self, name, parents, text, sha1):
281
"""Check that a duplicated add is OK.
283
If it is, return the (old) index; otherwise raise an exception.
285
idx = self._lookup(name)
286
if sorted(self._parents[idx]) != sorted(parents) \
287
or sha1 != self._sha1s[idx]:
288
raise RevisionAlreadyPresent(name, self._weave_name)
291
def _add_lines(self, version_id, parents, lines, parent_texts,
292
left_matching_blocks, nostore_sha, random_id, check_content):
293
"""See VersionedFile.add_lines."""
294
idx = self._add(version_id, lines, map(self._lookup, parents),
295
nostore_sha=nostore_sha)
296
return sha_strings(lines), sum(map(len, lines)), idx
298
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
186
299
"""Add a single text on top of the weave.
188
301
Returns the index number of the newly added version.
304
Symbolic name for this version.
305
(Typically the revision-id of the revision that added it.)
191
308
List or set of direct parent version numbers.
194
Sequence of lines to be added in the new version."""
311
Sequence of lines to be added in the new version.
313
:param nostore_sha: See VersionedFile.add_lines.
315
self._check_lines_not_unicode(lines)
316
self._check_lines_are_lines(lines)
318
sha1 = sha_strings(lines)
319
if sha1 == nostore_sha:
320
raise errors.ExistingContent
321
if version_id in self._name_map:
322
return self._check_repeated_add(version_id, parents, lines, sha1)
196
324
self._check_versions(parents)
197
## self._check_lines(text)
325
## self._check_lines(lines)
198
326
new_version = len(self._parents)
206
# if we abort after here the weave will be corrupt
207
self._parents.append(frozenset(parents))
328
# if we abort after here the (in-memory) weave will be corrupt because only
329
# some fields are updated
330
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
331
# - Robert Collins 20060226
332
self._parents.append(parents[:])
208
333
self._sha1s.append(sha1)
334
self._names.append(version_id)
335
self._name_map[version_id] = new_version
378
492
lineno = 0 # line of weave, 0-based
380
494
for l in self._weave:
381
if isinstance(l, tuple):
495
if l.__class__ == tuple:
499
istack.append(self._names[v])
503
dset.add(self._names[v])
394
raise WeaveFormatError('unexpected instruction %r'
397
assert isinstance(l, basestring)
399
yield lineno, istack[-1], dset, l
404
def _extract(self, versions):
405
"""Yield annotation of lines in included set.
407
Yields a sequence of tuples (origin, lineno, text), where
408
origin is the origin version, lineno the index in the weave,
409
and text the text of the line.
411
The set typically but not necessarily corresponds to a version.
413
included = self.inclusions(versions)
418
lineno = 0 # line of weave, 0-based
424
WFE = WeaveFormatError
426
for l in self._weave:
427
if isinstance(l, tuple):
431
assert v not in istack
446
assert isinstance(l, basestring)
448
isactive = (not dset) and istack and (istack[-1] in included)
450
result.append((istack[-1], lineno, l))
505
dset.remove(self._names[v])
507
raise WeaveFormatError('unexpected instruction %r' % v)
509
yield lineno, istack[-1], frozenset(dset), l
454
raise WFE("unclosed insertion blocks at end of weave",
513
raise WeaveFormatError("unclosed insertion blocks "
514
"at end of weave: %s" % istack)
457
raise WFE("unclosed deletion blocks at end of weave",
464
def get_iter(self, version):
465
"""Yield lines for the specified version."""
466
for origin, lineno, line in self._extract([version]):
470
def get(self, index):
471
return list(self.get_iter(index))
474
def mash_iter(self, included):
475
"""Return composed version of multiple included versions."""
476
for origin, lineno, text in self._extract(included):
480
def dump(self, to_file):
481
from pprint import pprint
482
print >>to_file, "Weave._weave = ",
483
pprint(self._weave, to_file)
484
print >>to_file, "Weave._parents = ",
485
pprint(self._parents, to_file)
489
def numversions(self):
490
l = len(self._parents)
491
assert l == len(self._sha1s)
496
return self.numversions()
499
def check(self, progress_bar=None):
500
# check no circular inclusions
501
for version in range(self.numversions()):
502
inclusions = list(self._parents[version])
505
if inclusions[-1] >= version:
506
raise WeaveFormatError("invalid included version %d for index %d"
507
% (inclusions[-1], version))
509
# try extracting all versions; this is a bit slow and parallel
510
# extraction could be used
512
nv = self.numversions()
513
for version in range(nv):
515
progress_bar.update('checking text', version, nv)
517
for l in self.get_iter(version):
520
expected = self._sha1s[version]
522
raise WeaveError("mismatched sha1 for version %d; "
523
"got %s, expected %s"
524
% (version, hd, expected))
526
# TODO: check insertions are properly nested, that there are
527
# no lines outside of insertion blocks, that deletions are
528
# properly paired, etc.
532
def merge(self, merge_versions):
533
"""Automerge and mark conflicts between versions.
535
This returns a sequence, each entry describing alternatives
536
for a chunk of the file. Each of the alternatives is given as
539
If there is a chunk of the file where there's no diagreement,
540
only one alternative is given.
543
# approach: find the included versions common to all the
545
raise NotImplementedError()
549
def _delta(self, included, lines):
550
"""Return changes from basis to new revision.
552
The old text for comparison is the union of included revisions.
554
This is used in inserting a new text.
556
Delta is returned as a sequence of
557
(weave1, weave2, newlines).
559
This indicates that weave1:weave2 of the old weave should be
560
replaced by the sequence of lines in newlines. Note that
561
these line numbers are positions in the total weave and don't
562
correspond to the lines in any extracted version, or even the
563
extracted union of included versions.
565
If line1=line2, this is a pure insert; if newlines=[] this is a
566
pure delete. (Similar to difflib.)
516
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
571
519
def plan_merge(self, ver_a, ver_b):
572
520
"""Return pseudo-annotation indicating how the two versions merge.
615
563
yield 'unchanged', '' # terminator
619
def weave_merge(self, plan):
624
for state, line in plan:
625
if state == 'unchanged' or state == 'killed-both':
626
# resync and flush queued conflicts changes if any
627
if not lines_a and not lines_b:
629
elif ch_a and not ch_b:
631
for l in lines_a: yield l
632
elif ch_b and not ch_a:
633
for l in lines_b: yield l
634
elif lines_a == lines_b:
635
for l in lines_a: yield l
565
def _extract(self, versions):
566
"""Yield annotation of lines in included set.
568
Yields a sequence of tuples (origin, lineno, text), where
569
origin is the origin version, lineno the index in the weave,
570
and text the text of the line.
572
The set typically but not necessarily corresponds to a version.
575
if not isinstance(i, int):
578
included = self._inclusions(versions)
584
lineno = 0 # line of weave, 0-based
590
WFE = WeaveFormatError
593
# 449 0 4474.6820 2356.5590 bzrlib.weave:556(_extract)
594
# +285282 0 1676.8040 1676.8040 +<isinstance>
595
# 1.6 seconds in 'isinstance'.
596
# changing the first isinstance:
597
# 449 0 2814.2660 1577.1760 bzrlib.weave:556(_extract)
598
# +140414 0 762.8050 762.8050 +<isinstance>
599
# note that the inline time actually dropped (less function calls)
600
# and total processing time was halved.
601
# we're still spending ~1/4 of the method in isinstance though.
602
# so lets hard code the acceptable string classes we expect:
603
# 449 0 1202.9420 786.2930 bzrlib.weave:556(_extract)
604
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
606
# yay, down to ~1/4 the initial extract time, and our inline time
607
# has shrunk again, with isinstance no longer dominating.
608
# tweaking the stack inclusion test to use a set gives:
609
# 449 0 1122.8030 713.0080 bzrlib.weave:556(_extract)
610
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
612
# - a 5% win, or possibly just noise. However with large istacks that
613
# 'in' test could dominate, so I'm leaving this change in place -
614
# when its fast enough to consider profiling big datasets we can review.
619
for l in self._weave:
620
if l.__class__ == tuple:
627
iset.remove(istack.pop())
638
for l in lines_a: yield l
640
for l in lines_b: yield l
647
if state == 'unchanged':
650
elif state == 'killed-a':
653
elif state == 'killed-b':
656
elif state == 'new-a':
659
elif state == 'new-b':
663
assert state in ('irrelevant', 'ghost-a', 'ghost-b', 'killed-base',
674
"""Show some text information about the weave."""
675
print '%6s %40s %20s' % ('ver', 'sha1', 'parents')
676
for i in (6, 40, 20):
635
raise AssertionError()
638
isactive = (not dset) and istack and (istack[-1] in included)
640
result.append((istack[-1], lineno, l))
643
raise WeaveFormatError("unclosed insertion blocks "
644
"at end of weave: %s" % istack)
646
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
650
def _maybe_lookup(self, name_or_index):
651
"""Convert possible symbolic name to index, or pass through indexes.
655
if isinstance(name_or_index, (int, long)):
658
return self._lookup(name_or_index)
660
def get_lines(self, version_id):
661
"""See VersionedFile.get_lines()."""
662
int_index = self._maybe_lookup(version_id)
663
result = [line for (origin, lineno, line) in self._extract([int_index])]
664
expected_sha1 = self._sha1s[int_index]
665
measured_sha1 = sha_strings(result)
666
if measured_sha1 != expected_sha1:
667
raise errors.WeaveInvalidChecksum(
668
'file %s, revision %s, expected: %s, measured %s'
669
% (self._weave_name, version_id,
670
expected_sha1, measured_sha1))
673
def get_sha1s(self, version_ids):
674
"""See VersionedFile.get_sha1s()."""
675
return [self._sha1s[self._lookup(v)] for v in version_ids]
677
def num_versions(self):
678
"""How many versions are in this weave?"""
679
l = len(self._parents)
682
__len__ = num_versions
684
def check(self, progress_bar=None):
685
# TODO evaluate performance hit of using string sets in this routine.
686
# TODO: check no circular inclusions
687
# TODO: create a nested progress bar
688
for version in range(self.num_versions()):
689
inclusions = list(self._parents[version])
692
if inclusions[-1] >= version:
693
raise WeaveFormatError("invalid included version %d for index %d"
694
% (inclusions[-1], version))
696
# try extracting all versions; parallel extraction is used
697
nv = self.num_versions()
702
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
703
# The problem is that set membership is much more expensive
704
name = self._idx_to_name(i)
705
sha1s[name] = sha.new()
707
new_inc = set([name])
708
for p in self._parents[i]:
709
new_inc.update(inclusions[self._idx_to_name(p)])
711
if set(new_inc) != set(self.get_ancestry(name)):
712
raise AssertionError(
714
% (set(new_inc), set(self.get_ancestry(name))))
715
inclusions[name] = new_inc
717
nlines = len(self._weave)
719
update_text = 'checking weave'
721
short_name = os.path.basename(self._weave_name)
722
update_text = 'checking %s' % (short_name,)
723
update_text = update_text[:25]
725
for lineno, insert, deleteset, line in self._walk_internal():
727
progress_bar.update(update_text, lineno, nlines)
729
for name, name_inclusions in inclusions.items():
730
# The active inclusion must be an ancestor,
731
# and no ancestors must have deleted this line,
732
# because we don't support resurrection.
733
if (insert in name_inclusions) and not (deleteset & name_inclusions):
734
sha1s[name].update(line)
737
version = self._idx_to_name(i)
738
hd = sha1s[version].hexdigest()
739
expected = self._sha1s[i]
741
raise errors.WeaveInvalidChecksum(
742
"mismatched sha1 for version %s: "
743
"got %s, expected %s"
744
% (version, hd, expected))
746
# TODO: check insertions are properly nested, that there are
747
# no lines outside of insertion blocks, that deletions are
748
# properly paired, etc.
750
def _join(self, other, pb, msg, version_ids, ignore_missing):
751
"""Worker routine for join()."""
752
if not other.versions():
753
return # nothing to update, easy
756
# versions is never none, InterWeave checks this.
759
# two loops so that we do not change ourselves before verifying it
761
# work through in index order to make sure we get all dependencies
764
# get the selected versions only that are in other.versions.
765
version_ids = set(other.versions()).intersection(set(version_ids))
766
# pull in the referenced graph.
767
version_ids = other.get_ancestry(version_ids)
768
pending_parents = other.get_parent_map(version_ids)
769
pending_graph = pending_parents.items()
770
if len(pending_graph) != len(version_ids):
771
raise RevisionNotPresent(
772
set(version_ids) - set(pending_parents.keys()), self)
773
for name in topo_sort(pending_graph):
774
other_idx = other._name_map[name]
775
# returns True if we have it, False if we need it.
776
if not self._check_version_consistent(other, other_idx, name):
777
names_to_join.append((other_idx, name))
785
for other_idx, name in names_to_join:
786
# TODO: If all the parents of the other version are already
787
# present then we can avoid some work by just taking the delta
788
# and adjusting the offsets.
789
new_parents = self._imported_parents(other, other_idx)
790
sha1 = other._sha1s[other_idx]
795
pb.update(msg, merged, len(names_to_join))
797
lines = other.get_lines(other_idx)
798
self._add(name, lines, new_parents, sha1)
800
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
801
merged, processed, self._weave_name, time.time()-time0))
803
def _imported_parents(self, other, other_idx):
804
"""Return list of parents in self corresponding to indexes in other."""
806
for parent_idx in other._parents[other_idx]:
807
parent_name = other._names[parent_idx]
808
if parent_name not in self._name_map:
809
# should not be possible
810
raise WeaveError("missing parent {%s} of {%s} in %r"
811
% (parent_name, other._name_map[other_idx], self))
812
new_parents.append(self._name_map[parent_name])
815
def _check_version_consistent(self, other, other_idx, name):
816
"""Check if a version in consistent in this and other.
818
To be consistent it must have:
821
* the same direct parents (by name, not index, and disregarding
824
If present & correct return True;
825
if not present in self return False;
826
if inconsistent raise error."""
827
this_idx = self._name_map.get(name, -1)
829
if self._sha1s[this_idx] != other._sha1s[other_idx]:
830
raise errors.WeaveTextDiffers(name, self, other)
831
self_parents = self._parents[this_idx]
832
other_parents = other._parents[other_idx]
833
n1 = set([self._names[i] for i in self_parents])
834
n2 = set([other._names[i] for i in other_parents])
835
if not self._compatible_parents(n1, n2):
836
raise WeaveParentMismatch("inconsistent parents "
837
"for version {%s}: %s vs %s" % (name, n1, n2))
843
def _reweave(self, other, pb, msg):
844
"""Reweave self with other - internal helper for join().
846
:param other: The other weave to merge
847
:param pb: An optional progress bar, indicating how far done we are
848
:param msg: An optional message for the progress
850
new_weave = _reweave(self, other, pb=pb, msg=msg)
851
self._copy_weave_content(new_weave)
853
def _copy_weave_content(self, otherweave):
854
"""adsorb the content from otherweave."""
855
for attr in self.__slots__:
856
if attr != '_weave_name':
857
setattr(self, attr, copy(getattr(otherweave, attr)))
860
class WeaveFile(Weave):
861
"""A WeaveFile represents a Weave on disk and writes on change."""
863
WEAVE_SUFFIX = '.weave'
865
def __init__(self, name, transport, filemode=None, create=False, access_mode='w', get_scope=None):
866
"""Create a WeaveFile.
868
:param create: If not True, only open an existing knit.
870
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope)
871
self._transport = transport
872
self._filemode = filemode
874
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
875
except errors.NoSuchFile:
881
def _add_lines(self, version_id, parents, lines, parent_texts,
882
left_matching_blocks, nostore_sha, random_id, check_content):
883
"""Add a version and save the weave."""
884
self.check_not_reserved_id(version_id)
885
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
886
parent_texts, left_matching_blocks, nostore_sha, random_id,
891
def copy_to(self, name, transport):
892
"""See VersionedFile.copy_to()."""
893
# as we are all in memory always, just serialise to the new place.
895
write_weave_v5(self, sio)
897
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
900
"""Save the weave."""
901
self._check_write_ok()
903
write_weave_v5(self, sio)
905
self._transport.put_file(self._weave_name + WeaveFile.WEAVE_SUFFIX,
911
"""See VersionedFile.get_suffixes()."""
912
return [WeaveFile.WEAVE_SUFFIX]
914
def join(self, other, pb=None, msg=None, version_ids=None,
915
ignore_missing=False):
916
"""Join other into self and save."""
917
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
921
def _reweave(wa, wb, pb=None, msg=None):
922
"""Combine two weaves and return the result.
924
This works even if a revision R has different parents in
925
wa and wb. In the resulting weave all the parents are given.
927
This is done by just building up a new weave, maintaining ordering
928
of the versions in the two inputs. More efficient approaches
929
might be possible but it should only be necessary to do
930
this operation rarely, when a new previously ghost version is
933
:param pb: An optional progress bar, indicating how far done we are
934
:param msg: An optional message for the progress
938
queue_a = range(wa.num_versions())
939
queue_b = range(wb.num_versions())
940
# first determine combined parents of all versions
941
# map from version name -> all parent names
942
combined_parents = _reweave_parent_graphs(wa, wb)
943
mutter("combined parents: %r", combined_parents)
944
order = topo_sort(combined_parents.iteritems())
945
mutter("order to reweave: %r", order)
950
for idx, name in enumerate(order):
952
pb.update(msg, idx, len(order))
953
if name in wa._name_map:
954
lines = wa.get_lines(name)
955
if name in wb._name_map:
956
lines_b = wb.get_lines(name)
958
mutter('Weaves differ on content. rev_id {%s}', name)
959
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
961
lines = list(difflib.unified_diff(lines, lines_b,
962
wa._weave_name, wb._weave_name))
963
mutter('lines:\n%s', ''.join(lines))
964
raise errors.WeaveTextDiffers(name, wa, wb)
966
lines = wb.get_lines(name)
967
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
970
def _reweave_parent_graphs(wa, wb):
971
"""Return combined parent ancestry for two weaves.
973
Returned as a list of (version_name, set(parent_names))"""
975
for weave in [wa, wb]:
976
for idx, name in enumerate(weave._names):
977
p = combined.setdefault(name, set())
978
p.update(map(weave._idx_to_name, weave._parents[idx]))
983
"""Show the weave's table-of-contents"""
984
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
985
for i in (6, 50, 10, 10):
679
for i in range(w.numversions()):
988
for i in range(w.num_versions()):
680
989
sha1 = w._sha1s[i]
681
print '%6d %40s %s' % (i, sha1, ' '.join(map(str, w._parents[i])))
685
def weave_stats(weave_file):
686
from bzrlib.progress import ProgressBar
991
parent_str = ' '.join(map(str, w._parents[i]))
992
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
996
def weave_stats(weave_file, pb):
687
997
from bzrlib.weavefile import read_weave
691
999
wf = file(weave_file, 'rb')
692
1000
w = read_weave(wf)
693
1001
# FIXME: doesn't work on pipes
added
removed
bzrlib/weave.py
