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
assert isinstance(version_id, basestring)
316
self._check_lines_not_unicode(lines)
317
self._check_lines_are_lines(lines)
319
sha1 = sha_strings(lines)
320
if sha1 == nostore_sha:
321
raise errors.ExistingContent
322
if version_id in self._name_map:
323
return self._check_repeated_add(version_id, parents, lines, sha1)
196
325
self._check_versions(parents)
197
## self._check_lines(text)
326
## self._check_lines(lines)
198
327
new_version = len(self._parents)
206
# if we abort after here the weave will be corrupt
207
self._parents.append(frozenset(parents))
329
# if we abort after here the (in-memory) weave will be corrupt because only
330
# some fields are updated
331
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
332
# - Robert Collins 20060226
333
self._parents.append(parents[:])
208
334
self._sha1s.append(sha1)
335
self._names.append(version_id)
336
self._name_map[version_id] = new_version
446
assert isinstance(l, basestring)
654
assert l.__class__ in (str, unicode)
447
655
if isactive is None:
448
656
isactive = (not dset) and istack and (istack[-1] in included)
450
658
result.append((istack[-1], lineno, l))
454
raise WFE("unclosed insertion blocks at end of weave",
661
raise WeaveFormatError("unclosed insertion blocks "
662
"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):
664
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
668
def _maybe_lookup(self, name_or_index):
669
"""Convert possible symbolic name to index, or pass through indexes.
673
if isinstance(name_or_index, (int, long)):
676
return self._lookup(name_or_index)
678
def get_lines(self, version_id):
679
"""See VersionedFile.get_lines()."""
680
int_index = self._maybe_lookup(version_id)
681
result = [line for (origin, lineno, line) in self._extract([int_index])]
682
expected_sha1 = self._sha1s[int_index]
683
measured_sha1 = sha_strings(result)
684
if measured_sha1 != expected_sha1:
685
raise errors.WeaveInvalidChecksum(
686
'file %s, revision %s, expected: %s, measured %s'
687
% (self._weave_name, version_id,
688
expected_sha1, measured_sha1))
691
def get_sha1(self, version_id):
692
"""See VersionedFile.get_sha1()."""
693
return self._sha1s[self._lookup(version_id)]
695
def get_sha1s(self, version_ids):
696
"""See VersionedFile.get_sha1s()."""
697
return [self._sha1s[self._lookup(v)] for v in version_ids]
699
def num_versions(self):
700
"""How many versions are in this weave?"""
490
701
l = len(self._parents)
491
702
assert l == len(self._sha1s)
496
return self.numversions()
705
__len__ = num_versions
499
707
def check(self, progress_bar=None):
500
# check no circular inclusions
501
for version in range(self.numversions()):
708
# TODO evaluate performance hit of using string sets in this routine.
709
# TODO: check no circular inclusions
710
# TODO: create a nested progress bar
711
for version in range(self.num_versions()):
502
712
inclusions = list(self._parents[version])
504
714
inclusions.sort()
506
716
raise WeaveFormatError("invalid included version %d for index %d"
507
717
% (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):
719
# try extracting all versions; parallel extraction is used
720
nv = self.num_versions()
725
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
726
# The problem is that set membership is much more expensive
727
name = self._idx_to_name(i)
728
sha1s[name] = sha.new()
730
new_inc = set([name])
731
for p in self._parents[i]:
732
new_inc.update(inclusions[self._idx_to_name(p)])
734
assert set(new_inc) == set(self.get_ancestry(name)), \
735
'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
736
inclusions[name] = new_inc
738
nlines = len(self._weave)
740
update_text = 'checking weave'
742
short_name = os.path.basename(self._weave_name)
743
update_text = 'checking %s' % (short_name,)
744
update_text = update_text[:25]
746
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]
748
progress_bar.update(update_text, lineno, nlines)
750
for name, name_inclusions in inclusions.items():
751
# The active inclusion must be an ancestor,
752
# and no ancestors must have deleted this line,
753
# because we don't support resurrection.
754
if (insert in name_inclusions) and not (deleteset & name_inclusions):
755
sha1s[name].update(line)
758
version = self._idx_to_name(i)
759
hd = sha1s[version].hexdigest()
760
expected = self._sha1s[i]
521
761
if hd != expected:
522
raise WeaveError("mismatched sha1 for version %d; "
523
"got %s, expected %s"
524
% (version, hd, expected))
762
raise errors.WeaveInvalidChecksum(
763
"mismatched sha1 for version %s: "
764
"got %s, expected %s"
765
% (version, hd, expected))
526
767
# TODO: check insertions are properly nested, that there are
527
768
# no lines outside of insertion blocks, that deletions are
528
769
# 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):
771
def _join(self, other, pb, msg, version_ids, ignore_missing):
772
"""Worker routine for join()."""
773
if not other.versions():
774
return # nothing to update, easy
777
# versions is never none, InterWeave checks this.
780
# two loops so that we do not change ourselves before verifying it
782
# work through in index order to make sure we get all dependencies
785
# get the selected versions only that are in other.versions.
786
version_ids = set(other.versions()).intersection(set(version_ids))
787
# pull in the referenced graph.
788
version_ids = other.get_ancestry(version_ids)
789
pending_parents = other.get_parent_map(version_ids)
790
pending_graph = pending_parents.items()
791
if len(pending_graph) != len(version_ids):
792
raise RevisionNotPresent(
793
set(version_ids) - set(pending_parents.keys()), self)
794
for name in topo_sort(pending_graph):
795
other_idx = other._name_map[name]
796
# returns True if we have it, False if we need it.
797
if not self._check_version_consistent(other, other_idx, name):
798
names_to_join.append((other_idx, name))
806
for other_idx, name in names_to_join:
807
# TODO: If all the parents of the other version are already
808
# present then we can avoid some work by just taking the delta
809
# and adjusting the offsets.
810
new_parents = self._imported_parents(other, other_idx)
811
sha1 = other._sha1s[other_idx]
816
pb.update(msg, merged, len(names_to_join))
818
lines = other.get_lines(other_idx)
819
self._add(name, lines, new_parents, sha1)
821
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
822
merged, processed, self._weave_name, time.time()-time0))
824
def _imported_parents(self, other, other_idx):
825
"""Return list of parents in self corresponding to indexes in other."""
827
for parent_idx in other._parents[other_idx]:
828
parent_name = other._names[parent_idx]
829
if parent_name not in self._name_map:
830
# should not be possible
831
raise WeaveError("missing parent {%s} of {%s} in %r"
832
% (parent_name, other._name_map[other_idx], self))
833
new_parents.append(self._name_map[parent_name])
836
def _check_version_consistent(self, other, other_idx, name):
837
"""Check if a version in consistent in this and other.
839
To be consistent it must have:
842
* the same direct parents (by name, not index, and disregarding
845
If present & correct return True;
846
if not present in self return False;
847
if inconsistent raise error."""
848
this_idx = self._name_map.get(name, -1)
850
if self._sha1s[this_idx] != other._sha1s[other_idx]:
851
raise errors.WeaveTextDiffers(name, self, other)
852
self_parents = self._parents[this_idx]
853
other_parents = other._parents[other_idx]
854
n1 = set([self._names[i] for i in self_parents])
855
n2 = set([other._names[i] for i in other_parents])
856
if not self._compatible_parents(n1, n2):
857
raise WeaveParentMismatch("inconsistent parents "
858
"for version {%s}: %s vs %s" % (name, n1, n2))
864
def _reweave(self, other, pb, msg):
865
"""Reweave self with other - internal helper for join().
867
:param other: The other weave to merge
868
:param pb: An optional progress bar, indicating how far done we are
869
:param msg: An optional message for the progress
871
new_weave = _reweave(self, other, pb=pb, msg=msg)
872
self._copy_weave_content(new_weave)
874
def _copy_weave_content(self, otherweave):
875
"""adsorb the content from otherweave."""
876
for attr in self.__slots__:
877
if attr != '_weave_name':
878
setattr(self, attr, copy(getattr(otherweave, attr)))
881
class WeaveFile(Weave):
882
"""A WeaveFile represents a Weave on disk and writes on change."""
884
WEAVE_SUFFIX = '.weave'
886
def __init__(self, name, transport, filemode=None, create=False, access_mode='w', get_scope=None):
887
"""Create a WeaveFile.
889
:param create: If not True, only open an existing knit.
891
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope)
892
self._transport = transport
893
self._filemode = filemode
895
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
896
except errors.NoSuchFile:
902
def _add_lines(self, version_id, parents, lines, parent_texts,
903
left_matching_blocks, nostore_sha, random_id, check_content):
904
"""Add a version and save the weave."""
905
self.check_not_reserved_id(version_id)
906
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
907
parent_texts, left_matching_blocks, nostore_sha, random_id,
912
def _clone_text(self, new_version_id, old_version_id, parents):
913
"""See VersionedFile.clone_text."""
914
super(WeaveFile, self)._clone_text(new_version_id, old_version_id, parents)
917
def copy_to(self, name, transport):
918
"""See VersionedFile.copy_to()."""
919
# as we are all in memory always, just serialise to the new place.
921
write_weave_v5(self, sio)
923
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
926
"""Save the weave."""
927
self._check_write_ok()
929
write_weave_v5(self, sio)
931
self._transport.put_file(self._weave_name + WeaveFile.WEAVE_SUFFIX,
937
"""See VersionedFile.get_suffixes()."""
938
return [WeaveFile.WEAVE_SUFFIX]
940
def join(self, other, pb=None, msg=None, version_ids=None,
941
ignore_missing=False):
942
"""Join other into self and save."""
943
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
947
def _reweave(wa, wb, pb=None, msg=None):
948
"""Combine two weaves and return the result.
950
This works even if a revision R has different parents in
951
wa and wb. In the resulting weave all the parents are given.
953
This is done by just building up a new weave, maintaining ordering
954
of the versions in the two inputs. More efficient approaches
955
might be possible but it should only be necessary to do
956
this operation rarely, when a new previously ghost version is
959
:param pb: An optional progress bar, indicating how far done we are
960
:param msg: An optional message for the progress
964
queue_a = range(wa.num_versions())
965
queue_b = range(wb.num_versions())
966
# first determine combined parents of all versions
967
# map from version name -> all parent names
968
combined_parents = _reweave_parent_graphs(wa, wb)
969
mutter("combined parents: %r", combined_parents)
970
order = topo_sort(combined_parents.iteritems())
971
mutter("order to reweave: %r", order)
976
for idx, name in enumerate(order):
978
pb.update(msg, idx, len(order))
979
if name in wa._name_map:
980
lines = wa.get_lines(name)
981
if name in wb._name_map:
982
lines_b = wb.get_lines(name)
984
mutter('Weaves differ on content. rev_id {%s}', name)
985
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
987
lines = list(difflib.unified_diff(lines, lines_b,
988
wa._weave_name, wb._weave_name))
989
mutter('lines:\n%s', ''.join(lines))
990
raise errors.WeaveTextDiffers(name, wa, wb)
992
lines = wb.get_lines(name)
993
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
996
def _reweave_parent_graphs(wa, wb):
997
"""Return combined parent ancestry for two weaves.
999
Returned as a list of (version_name, set(parent_names))"""
1001
for weave in [wa, wb]:
1002
for idx, name in enumerate(weave._names):
1003
p = combined.setdefault(name, set())
1004
p.update(map(weave._idx_to_name, weave._parents[idx]))
1009
"""Show the weave's table-of-contents"""
1010
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
1011
for i in (6, 50, 10, 10):
679
for i in range(w.numversions()):
1014
for i in range(w.num_versions()):
680
1015
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
1017
parent_str = ' '.join(map(str, w._parents[i]))
1018
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
1022
def weave_stats(weave_file, pb):
687
1023
from bzrlib.weavefile import read_weave
691
1025
wf = file(weave_file, 'rb')
692
1026
w = read_weave(wf)
693
1027
# FIXME: doesn't work on pipes
added
removed
bzrlib/weave.py
