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):
189
super(Weave, self).__init__(access_mode)
170
191
self._parents = []
195
self._weave_name = weave_name
197
self._matcher = bzrlib.patiencediff.PatienceSequenceMatcher
199
self._matcher = matcher
202
return "Weave(%r)" % self._weave_name
205
"""Return a deep copy of self.
207
The copy can be modified without affecting the original weave."""
209
other._weave = self._weave[:]
210
other._parents = self._parents[:]
211
other._sha1s = self._sha1s[:]
212
other._names = self._names[:]
213
other._name_map = self._name_map.copy()
214
other._weave_name = self._weave_name
174
217
def __eq__(self, other):
175
218
if not isinstance(other, Weave):
177
220
return self._parents == other._parents \
178
and self._weave == other._weave
221
and self._weave == other._weave \
222
and self._sha1s == other._sha1s
181
224
def __ne__(self, other):
182
225
return not self.__eq__(other)
185
def add(self, parents, text):
227
def _idx_to_name(self, version):
228
return self._names[version]
230
def _lookup(self, name):
231
"""Convert symbolic version name to index."""
232
self.check_not_reserved_id(name)
234
return self._name_map[name]
236
raise RevisionNotPresent(name, self._weave_name)
239
"""See VersionedFile.versions."""
240
return self._names[:]
242
def has_version(self, version_id):
243
"""See VersionedFile.has_version."""
244
return (version_id in self._name_map)
246
__contains__ = has_version
248
def get_parents(self, version_id):
249
"""See VersionedFile.get_parent."""
250
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
252
def _check_repeated_add(self, name, parents, text, sha1):
253
"""Check that a duplicated add is OK.
255
If it is, return the (old) index; otherwise raise an exception.
257
idx = self._lookup(name)
258
if sorted(self._parents[idx]) != sorted(parents) \
259
or sha1 != self._sha1s[idx]:
260
raise RevisionAlreadyPresent(name, self._weave_name)
263
def _add_lines(self, version_id, parents, lines, parent_texts,
264
left_matching_blocks, nostore_sha, random_id, check_content):
265
"""See VersionedFile.add_lines."""
266
idx = self._add(version_id, lines, map(self._lookup, parents),
267
nostore_sha=nostore_sha)
268
return sha_strings(lines), sum(map(len, lines)), idx
270
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
186
271
"""Add a single text on top of the weave.
188
273
Returns the index number of the newly added version.
276
Symbolic name for this version.
277
(Typically the revision-id of the revision that added it.)
191
280
List or set of direct parent version numbers.
194
Sequence of lines to be added in the new version."""
283
Sequence of lines to be added in the new version.
285
:param nostore_sha: See VersionedFile.add_lines.
287
assert isinstance(version_id, basestring)
288
self._check_lines_not_unicode(lines)
289
self._check_lines_are_lines(lines)
291
sha1 = sha_strings(lines)
292
if sha1 == nostore_sha:
293
raise errors.ExistingContent
294
if version_id in self._name_map:
295
return self._check_repeated_add(version_id, parents, lines, sha1)
196
297
self._check_versions(parents)
197
## self._check_lines(text)
298
## self._check_lines(lines)
198
299
new_version = len(self._parents)
206
# if we abort after here the weave will be corrupt
207
self._parents.append(frozenset(parents))
301
# if we abort after here the (in-memory) weave will be corrupt because only
302
# some fields are updated
303
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
304
# - Robert Collins 20060226
305
self._parents.append(parents[:])
208
306
self._sha1s.append(sha1)
307
self._names.append(version_id)
308
self._name_map[version_id] = new_version
446
assert isinstance(l, basestring)
626
assert l.__class__ in (str, unicode)
447
627
if isactive is None:
448
628
isactive = (not dset) and istack and (istack[-1] in included)
450
630
result.append((istack[-1], lineno, l))
454
raise WFE("unclosed insertion blocks at end of weave",
633
raise WeaveFormatError("unclosed insertion blocks "
634
"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):
636
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
640
def _maybe_lookup(self, name_or_index):
641
"""Convert possible symbolic name to index, or pass through indexes.
645
if isinstance(name_or_index, (int, long)):
648
return self._lookup(name_or_index)
650
def get_lines(self, version_id):
651
"""See VersionedFile.get_lines()."""
652
int_index = self._maybe_lookup(version_id)
653
result = [line for (origin, lineno, line) in self._extract([int_index])]
654
expected_sha1 = self._sha1s[int_index]
655
measured_sha1 = sha_strings(result)
656
if measured_sha1 != expected_sha1:
657
raise errors.WeaveInvalidChecksum(
658
'file %s, revision %s, expected: %s, measured %s'
659
% (self._weave_name, version_id,
660
expected_sha1, measured_sha1))
663
def get_sha1(self, version_id):
664
"""See VersionedFile.get_sha1()."""
665
return self._sha1s[self._lookup(version_id)]
667
def get_sha1s(self, version_ids):
668
"""See VersionedFile.get_sha1s()."""
669
return [self._sha1s[self._lookup(v)] for v in version_ids]
671
def num_versions(self):
672
"""How many versions are in this weave?"""
490
673
l = len(self._parents)
491
674
assert l == len(self._sha1s)
496
return self.numversions()
677
__len__ = num_versions
499
679
def check(self, progress_bar=None):
500
# check no circular inclusions
501
for version in range(self.numversions()):
680
# TODO evaluate performance hit of using string sets in this routine.
681
# TODO: check no circular inclusions
682
# TODO: create a nested progress bar
683
for version in range(self.num_versions()):
502
684
inclusions = list(self._parents[version])
504
686
inclusions.sort()
506
688
raise WeaveFormatError("invalid included version %d for index %d"
507
689
% (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):
691
# try extracting all versions; parallel extraction is used
692
nv = self.num_versions()
697
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
698
# The problem is that set membership is much more expensive
699
name = self._idx_to_name(i)
700
sha1s[name] = sha.new()
702
new_inc = set([name])
703
for p in self._parents[i]:
704
new_inc.update(inclusions[self._idx_to_name(p)])
706
assert set(new_inc) == set(self.get_ancestry(name)), \
707
'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
708
inclusions[name] = new_inc
710
nlines = len(self._weave)
712
update_text = 'checking weave'
714
short_name = os.path.basename(self._weave_name)
715
update_text = 'checking %s' % (short_name,)
716
update_text = update_text[:25]
718
for lineno, insert, deleteset, line in self._walk_internal():
515
progress_bar.update('checking text', version, nv)
517
for l in self.get_iter(version):
520
expected = self._sha1s[version]
720
progress_bar.update(update_text, lineno, nlines)
722
for name, name_inclusions in inclusions.items():
723
# The active inclusion must be an ancestor,
724
# and no ancestors must have deleted this line,
725
# because we don't support resurrection.
726
if (insert in name_inclusions) and not (deleteset & name_inclusions):
727
sha1s[name].update(line)
730
version = self._idx_to_name(i)
731
hd = sha1s[version].hexdigest()
732
expected = self._sha1s[i]
521
733
if hd != expected:
522
raise WeaveError("mismatched sha1 for version %d; "
523
"got %s, expected %s"
524
% (version, hd, expected))
734
raise errors.WeaveInvalidChecksum(
735
"mismatched sha1 for version %s: "
736
"got %s, expected %s"
737
% (version, hd, expected))
526
739
# TODO: check insertions are properly nested, that there are
527
740
# no lines outside of insertion blocks, that deletions are
528
741
# 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.)
571
def plan_merge(self, ver_a, ver_b):
572
"""Return pseudo-annotation indicating how the two versions merge.
574
This is computed between versions a and b and their common
577
Weave lines present in none of them are skipped entirely.
579
inc_a = self.inclusions([ver_a])
580
inc_b = self.inclusions([ver_b])
581
inc_c = inc_a & inc_b
583
for lineno, insert, deleteset, line in self._walk():
584
if deleteset & inc_c:
585
# killed in parent; can't be in either a or b
586
# not relevant to our work
587
yield 'killed-base', line
588
elif insert in inc_c:
589
# was inserted in base
590
killed_a = bool(deleteset & inc_a)
591
killed_b = bool(deleteset & inc_b)
592
if killed_a and killed_b:
593
yield 'killed-both', line
595
yield 'killed-a', line
597
yield 'killed-b', line
599
yield 'unchanged', line
600
elif insert in inc_a:
601
if deleteset & inc_a:
602
yield 'ghost-a', line
606
elif insert in inc_b:
607
if deleteset & inc_b:
608
yield 'ghost-b', line
612
# not in either revision
613
yield 'irrelevant', line
615
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
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):
743
def _join(self, other, pb, msg, version_ids, ignore_missing):
744
"""Worker routine for join()."""
745
if not other.versions():
746
return # nothing to update, easy
749
# versions is never none, InterWeave checks this.
752
# two loops so that we do not change ourselves before verifying it
754
# work through in index order to make sure we get all dependencies
757
# get the selected versions only that are in other.versions.
758
version_ids = set(other.versions()).intersection(set(version_ids))
759
# pull in the referenced graph.
760
version_ids = other.get_ancestry(version_ids)
761
pending_graph = [(version, other.get_parents(version)) for
762
version in version_ids]
763
for name in topo_sort(pending_graph):
764
other_idx = other._name_map[name]
765
# returns True if we have it, False if we need it.
766
if not self._check_version_consistent(other, other_idx, name):
767
names_to_join.append((other_idx, name))
776
for other_idx, name in names_to_join:
777
# TODO: If all the parents of the other version are already
778
# present then we can avoid some work by just taking the delta
779
# and adjusting the offsets.
780
new_parents = self._imported_parents(other, other_idx)
781
sha1 = other._sha1s[other_idx]
786
pb.update(msg, merged, len(names_to_join))
788
lines = other.get_lines(other_idx)
789
self._add(name, lines, new_parents, sha1)
791
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
792
merged, processed, self._weave_name, time.time()-time0))
794
def _imported_parents(self, other, other_idx):
795
"""Return list of parents in self corresponding to indexes in other."""
797
for parent_idx in other._parents[other_idx]:
798
parent_name = other._names[parent_idx]
799
if parent_name not in self._name_map:
800
# should not be possible
801
raise WeaveError("missing parent {%s} of {%s} in %r"
802
% (parent_name, other._name_map[other_idx], self))
803
new_parents.append(self._name_map[parent_name])
806
def _check_version_consistent(self, other, other_idx, name):
807
"""Check if a version in consistent in this and other.
809
To be consistent it must have:
812
* the same direct parents (by name, not index, and disregarding
815
If present & correct return True;
816
if not present in self return False;
817
if inconsistent raise error."""
818
this_idx = self._name_map.get(name, -1)
820
if self._sha1s[this_idx] != other._sha1s[other_idx]:
821
raise errors.WeaveTextDiffers(name, self, other)
822
self_parents = self._parents[this_idx]
823
other_parents = other._parents[other_idx]
824
n1 = set([self._names[i] for i in self_parents])
825
n2 = set([other._names[i] for i in other_parents])
826
if not self._compatible_parents(n1, n2):
827
raise WeaveParentMismatch("inconsistent parents "
828
"for version {%s}: %s vs %s" % (name, n1, n2))
834
def _reweave(self, other, pb, msg):
835
"""Reweave self with other - internal helper for join().
837
:param other: The other weave to merge
838
:param pb: An optional progress bar, indicating how far done we are
839
:param msg: An optional message for the progress
841
new_weave = _reweave(self, other, pb=pb, msg=msg)
842
self._copy_weave_content(new_weave)
844
def _copy_weave_content(self, otherweave):
845
"""adsorb the content from otherweave."""
846
for attr in self.__slots__:
847
if attr != '_weave_name':
848
setattr(self, attr, copy(getattr(otherweave, attr)))
851
class WeaveFile(Weave):
852
"""A WeaveFile represents a Weave on disk and writes on change."""
854
WEAVE_SUFFIX = '.weave'
856
def __init__(self, name, transport, filemode=None, create=False, access_mode='w'):
857
"""Create a WeaveFile.
859
:param create: If not True, only open an existing knit.
861
super(WeaveFile, self).__init__(name, access_mode)
862
self._transport = transport
863
self._filemode = filemode
865
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
866
except errors.NoSuchFile:
872
def _add_lines(self, version_id, parents, lines, parent_texts,
873
left_matching_blocks, nostore_sha, random_id, check_content):
874
"""Add a version and save the weave."""
875
self.check_not_reserved_id(version_id)
876
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
877
parent_texts, left_matching_blocks, nostore_sha, random_id,
882
def _clone_text(self, new_version_id, old_version_id, parents):
883
"""See VersionedFile.clone_text."""
884
super(WeaveFile, self)._clone_text(new_version_id, old_version_id, parents)
887
def copy_to(self, name, transport):
888
"""See VersionedFile.copy_to()."""
889
# as we are all in memory always, just serialise to the new place.
891
write_weave_v5(self, sio)
893
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
895
def create_empty(self, name, transport, filemode=None):
896
return WeaveFile(name, transport, filemode, create=True)
899
"""Save the weave."""
900
self._check_write_ok()
902
write_weave_v5(self, sio)
904
self._transport.put_file(self._weave_name + WeaveFile.WEAVE_SUFFIX,
910
"""See VersionedFile.get_suffixes()."""
911
return [WeaveFile.WEAVE_SUFFIX]
913
def join(self, other, pb=None, msg=None, version_ids=None,
914
ignore_missing=False):
915
"""Join other into self and save."""
916
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
920
def _reweave(wa, wb, pb=None, msg=None):
921
"""Combine two weaves and return the result.
923
This works even if a revision R has different parents in
924
wa and wb. In the resulting weave all the parents are given.
926
This is done by just building up a new weave, maintaining ordering
927
of the versions in the two inputs. More efficient approaches
928
might be possible but it should only be necessary to do
929
this operation rarely, when a new previously ghost version is
932
:param pb: An optional progress bar, indicating how far done we are
933
:param msg: An optional message for the progress
937
queue_a = range(wa.num_versions())
938
queue_b = range(wb.num_versions())
939
# first determine combined parents of all versions
940
# map from version name -> all parent names
941
combined_parents = _reweave_parent_graphs(wa, wb)
942
mutter("combined parents: %r", combined_parents)
943
order = topo_sort(combined_parents.iteritems())
944
mutter("order to reweave: %r", order)
949
for idx, name in enumerate(order):
951
pb.update(msg, idx, len(order))
952
if name in wa._name_map:
953
lines = wa.get_lines(name)
954
if name in wb._name_map:
955
lines_b = wb.get_lines(name)
957
mutter('Weaves differ on content. rev_id {%s}', name)
958
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
960
lines = list(difflib.unified_diff(lines, lines_b,
961
wa._weave_name, wb._weave_name))
962
mutter('lines:\n%s', ''.join(lines))
963
raise errors.WeaveTextDiffers(name, wa, wb)
965
lines = wb.get_lines(name)
966
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
969
def _reweave_parent_graphs(wa, wb):
970
"""Return combined parent ancestry for two weaves.
972
Returned as a list of (version_name, set(parent_names))"""
974
for weave in [wa, wb]:
975
for idx, name in enumerate(weave._names):
976
p = combined.setdefault(name, set())
977
p.update(map(weave._idx_to_name, weave._parents[idx]))
982
"""Show the weave's table-of-contents"""
983
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
984
for i in (6, 50, 10, 10):
679
for i in range(w.numversions()):
987
for i in range(w.num_versions()):
680
988
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
990
parent_str = ' '.join(map(str, w._parents[i]))
991
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
995
def weave_stats(weave_file, pb):
687
996
from bzrlib.weavefile import read_weave
691
998
wf = file(weave_file, 'rb')
692
999
w = read_weave(wf)
693
1000
# FIXME: doesn't work on pipes
added
removed
bzrlib/weave.py
