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_delta(self, version_id):
249
"""See VersionedFile.get_delta."""
250
return self.get_deltas([version_id])[version_id]
252
def get_deltas(self, version_ids):
253
"""See VersionedFile.get_deltas."""
254
version_ids = self.get_ancestry(version_ids)
255
for version_id in version_ids:
256
if not self.has_version(version_id):
257
raise RevisionNotPresent(version_id, self)
258
# try extracting all versions; parallel extraction is used
259
nv = self.num_versions()
265
last_parent_lines = {}
267
parent_inclusions = {}
272
# its simplest to generate a full set of prepared variables.
274
name = self._names[i]
275
sha1s[name] = self.get_sha1(name)
276
parents_list = self.get_parents(name)
278
parent = parents_list[0]
279
parents[name] = parent
280
parent_inclusions[name] = inclusions[parent]
283
parent_inclusions[name] = set()
284
# we want to emit start, finish, replacement_length, replacement_lines tuples.
285
diff_hunks[name] = []
286
current_hunks[name] = [0, 0, 0, []] # #start, finish, repl_length, repl_tuples
287
parent_linenums[name] = 0
289
parent_noeols[name] = False
290
last_parent_lines[name] = None
291
new_inc = set([name])
292
for p in self._parents[i]:
293
new_inc.update(inclusions[self._idx_to_name(p)])
294
# debug only, known good so far.
295
#assert set(new_inc) == set(self.get_ancestry(name)), \
296
# 'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
297
inclusions[name] = new_inc
299
nlines = len(self._weave)
301
for lineno, inserted, deletes, line in self._walk_internal():
302
# a line is active in a version if:
303
# insert is in the versions inclusions
305
# deleteset & the versions inclusions is an empty set.
306
# so - if we have a included by mapping - version is included by
307
# children, we get a list of children to examine for deletes affect
308
# ing them, which is less than the entire set of children.
309
for version_id in version_ids:
310
# The active inclusion must be an ancestor,
311
# and no ancestors must have deleted this line,
312
# because we don't support resurrection.
313
parent_inclusion = parent_inclusions[version_id]
314
inclusion = inclusions[version_id]
315
parent_active = inserted in parent_inclusion and not (deletes & parent_inclusion)
316
version_active = inserted in inclusion and not (deletes & inclusion)
317
if not parent_active and not version_active:
318
# unrelated line of ancestry
320
elif parent_active and version_active:
322
parent_linenum = parent_linenums[version_id]
323
if current_hunks[version_id] != [parent_linenum, parent_linenum, 0, []]:
324
diff_hunks[version_id].append(tuple(current_hunks[version_id]))
326
current_hunks[version_id] = [parent_linenum, parent_linenum, 0, []]
327
parent_linenums[version_id] = parent_linenum
330
noeols[version_id] = True
333
elif parent_active and not version_active:
335
current_hunks[version_id][1] += 1
336
parent_linenums[version_id] += 1
337
last_parent_lines[version_id] = line
338
elif not parent_active and version_active:
340
# noeol only occurs at the end of a file because we
341
# diff linewise. We want to show noeol changes as a
342
# empty diff unless the actual eol-less content changed.
345
if last_parent_lines[version_id][-1] != '\n':
346
parent_noeols[version_id] = True
347
except (TypeError, IndexError):
350
if theline[-1] != '\n':
351
noeols[version_id] = True
355
parent_should_go = False
357
if parent_noeols[version_id] == noeols[version_id]:
358
# no noeol toggle, so trust the weaves statement
359
# that this line is changed.
361
if parent_noeols[version_id]:
362
theline = theline + '\n'
363
elif parent_noeols[version_id]:
364
# parent has no eol, we do:
365
# our line is new, report as such..
367
elif noeols[version_id]:
368
# append a eol so that it looks like
370
theline = theline + '\n'
371
if parents[version_id] is not None:
372
#if last_parent_lines[version_id] is not None:
373
parent_should_go = True
374
if last_parent_lines[version_id] != theline:
377
#parent_should_go = False
379
current_hunks[version_id][2] += 1
380
current_hunks[version_id][3].append((inserted, theline))
382
# last hunk last parent line is not eaten
383
current_hunks[version_id][1] -= 1
384
if current_hunks[version_id][1] < 0:
385
current_hunks[version_id][1] = 0
386
# import pdb;pdb.set_trace()
387
# assert current_hunks[version_id][1] >= 0
391
version = self._idx_to_name(i)
392
if current_hunks[version] != [0, 0, 0, []]:
393
diff_hunks[version].append(tuple(current_hunks[version]))
395
for version_id in version_ids:
396
result[version_id] = (
400
diff_hunks[version_id],
404
def get_parents(self, version_id):
405
"""See VersionedFile.get_parent."""
406
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
408
def _check_repeated_add(self, name, parents, text, sha1):
409
"""Check that a duplicated add is OK.
411
If it is, return the (old) index; otherwise raise an exception.
413
idx = self._lookup(name)
414
if sorted(self._parents[idx]) != sorted(parents) \
415
or sha1 != self._sha1s[idx]:
416
raise RevisionAlreadyPresent(name, self._weave_name)
419
def _add_lines(self, version_id, parents, lines, parent_texts,
420
left_matching_blocks=None):
421
"""See VersionedFile.add_lines."""
422
idx = self._add(version_id, lines, map(self._lookup, parents))
423
return sha_strings(lines), sum(map(len, lines)), idx
425
def _add(self, version_id, lines, parents, sha1=None):
186
426
"""Add a single text on top of the weave.
188
428
Returns the index number of the newly added version.
431
Symbolic name for this version.
432
(Typically the revision-id of the revision that added it.)
191
435
List or set of direct parent version numbers.
194
Sequence of lines to be added in the new version."""
438
Sequence of lines to be added in the new version.
441
assert isinstance(version_id, basestring)
442
self._check_lines_not_unicode(lines)
443
self._check_lines_are_lines(lines)
445
sha1 = sha_strings(lines)
446
if version_id in self._name_map:
447
return self._check_repeated_add(version_id, parents, lines, sha1)
196
449
self._check_versions(parents)
197
## self._check_lines(text)
450
## self._check_lines(lines)
198
451
new_version = len(self._parents)
206
# if we abort after here the weave will be corrupt
207
self._parents.append(frozenset(parents))
453
# if we abort after here the (in-memory) weave will be corrupt because only
454
# some fields are updated
455
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
456
# - Robert Collins 20060226
457
self._parents.append(parents[:])
208
458
self._sha1s.append(sha1)
459
self._names.append(version_id)
460
self._name_map[version_id] = new_version
446
assert isinstance(l, basestring)
778
assert l.__class__ in (str, unicode)
447
779
if isactive is None:
448
780
isactive = (not dset) and istack and (istack[-1] in included)
450
782
result.append((istack[-1], lineno, l))
454
raise WFE("unclosed insertion blocks at end of weave",
785
raise WeaveFormatError("unclosed insertion blocks "
786
"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):
788
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
792
def _maybe_lookup(self, name_or_index):
793
"""Convert possible symbolic name to index, or pass through indexes.
797
if isinstance(name_or_index, (int, long)):
800
return self._lookup(name_or_index)
802
def get_lines(self, version_id):
803
"""See VersionedFile.get_lines()."""
804
int_index = self._maybe_lookup(version_id)
805
result = [line for (origin, lineno, line) in self._extract([int_index])]
806
expected_sha1 = self._sha1s[int_index]
807
measured_sha1 = sha_strings(result)
808
if measured_sha1 != expected_sha1:
809
raise errors.WeaveInvalidChecksum(
810
'file %s, revision %s, expected: %s, measured %s'
811
% (self._weave_name, version_id,
812
expected_sha1, measured_sha1))
815
def get_sha1(self, version_id):
816
"""See VersionedFile.get_sha1()."""
817
return self._sha1s[self._lookup(version_id)]
819
def get_sha1s(self, version_ids):
820
"""See VersionedFile.get_sha1s()."""
821
return [self._sha1s[self._lookup(v)] for v in version_ids]
823
def num_versions(self):
824
"""How many versions are in this weave?"""
490
825
l = len(self._parents)
491
826
assert l == len(self._sha1s)
496
return self.numversions()
829
__len__ = num_versions
499
831
def check(self, progress_bar=None):
500
# check no circular inclusions
501
for version in range(self.numversions()):
832
# TODO evaluate performance hit of using string sets in this routine.
833
# TODO: check no circular inclusions
834
# TODO: create a nested progress bar
835
for version in range(self.num_versions()):
502
836
inclusions = list(self._parents[version])
504
838
inclusions.sort()
506
840
raise WeaveFormatError("invalid included version %d for index %d"
507
841
% (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):
843
# try extracting all versions; parallel extraction is used
844
nv = self.num_versions()
849
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
850
# The problem is that set membership is much more expensive
851
name = self._idx_to_name(i)
852
sha1s[name] = sha.new()
854
new_inc = set([name])
855
for p in self._parents[i]:
856
new_inc.update(inclusions[self._idx_to_name(p)])
858
assert set(new_inc) == set(self.get_ancestry(name)), \
859
'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
860
inclusions[name] = new_inc
862
nlines = len(self._weave)
864
update_text = 'checking weave'
866
short_name = os.path.basename(self._weave_name)
867
update_text = 'checking %s' % (short_name,)
868
update_text = update_text[:25]
870
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]
872
progress_bar.update(update_text, lineno, nlines)
874
for name, name_inclusions in inclusions.items():
875
# The active inclusion must be an ancestor,
876
# and no ancestors must have deleted this line,
877
# because we don't support resurrection.
878
if (insert in name_inclusions) and not (deleteset & name_inclusions):
879
sha1s[name].update(line)
882
version = self._idx_to_name(i)
883
hd = sha1s[version].hexdigest()
884
expected = self._sha1s[i]
521
885
if hd != expected:
522
raise WeaveError("mismatched sha1 for version %d; "
523
"got %s, expected %s"
524
% (version, hd, expected))
886
raise errors.WeaveInvalidChecksum(
887
"mismatched sha1 for version %s: "
888
"got %s, expected %s"
889
% (version, hd, expected))
526
891
# TODO: check insertions are properly nested, that there are
527
892
# no lines outside of insertion blocks, that deletions are
528
893
# 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):
895
def _join(self, other, pb, msg, version_ids, ignore_missing):
896
"""Worker routine for join()."""
897
if not other.versions():
898
return # nothing to update, easy
901
# versions is never none, InterWeave checks this.
904
# two loops so that we do not change ourselves before verifying it
906
# work through in index order to make sure we get all dependencies
909
# get the selected versions only that are in other.versions.
910
version_ids = set(other.versions()).intersection(set(version_ids))
911
# pull in the referenced graph.
912
version_ids = other.get_ancestry(version_ids)
913
pending_graph = [(version, other.get_parents(version)) for
914
version in version_ids]
915
for name in topo_sort(pending_graph):
916
other_idx = other._name_map[name]
917
# returns True if we have it, False if we need it.
918
if not self._check_version_consistent(other, other_idx, name):
919
names_to_join.append((other_idx, name))
928
for other_idx, name in names_to_join:
929
# TODO: If all the parents of the other version are already
930
# present then we can avoid some work by just taking the delta
931
# and adjusting the offsets.
932
new_parents = self._imported_parents(other, other_idx)
933
sha1 = other._sha1s[other_idx]
938
pb.update(msg, merged, len(names_to_join))
940
lines = other.get_lines(other_idx)
941
self._add(name, lines, new_parents, sha1)
943
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
944
merged, processed, self._weave_name, time.time()-time0))
946
def _imported_parents(self, other, other_idx):
947
"""Return list of parents in self corresponding to indexes in other."""
949
for parent_idx in other._parents[other_idx]:
950
parent_name = other._names[parent_idx]
951
if parent_name not in self._name_map:
952
# should not be possible
953
raise WeaveError("missing parent {%s} of {%s} in %r"
954
% (parent_name, other._name_map[other_idx], self))
955
new_parents.append(self._name_map[parent_name])
958
def _check_version_consistent(self, other, other_idx, name):
959
"""Check if a version in consistent in this and other.
961
To be consistent it must have:
964
* the same direct parents (by name, not index, and disregarding
967
If present & correct return True;
968
if not present in self return False;
969
if inconsistent raise error."""
970
this_idx = self._name_map.get(name, -1)
972
if self._sha1s[this_idx] != other._sha1s[other_idx]:
973
raise errors.WeaveTextDiffers(name, self, other)
974
self_parents = self._parents[this_idx]
975
other_parents = other._parents[other_idx]
976
n1 = set([self._names[i] for i in self_parents])
977
n2 = set([other._names[i] for i in other_parents])
978
if not self._compatible_parents(n1, n2):
979
raise WeaveParentMismatch("inconsistent parents "
980
"for version {%s}: %s vs %s" % (name, n1, n2))
986
def _reweave(self, other, pb, msg):
987
"""Reweave self with other - internal helper for join().
989
:param other: The other weave to merge
990
:param pb: An optional progress bar, indicating how far done we are
991
:param msg: An optional message for the progress
993
new_weave = _reweave(self, other, pb=pb, msg=msg)
994
self._copy_weave_content(new_weave)
996
def _copy_weave_content(self, otherweave):
997
"""adsorb the content from otherweave."""
998
for attr in self.__slots__:
999
if attr != '_weave_name':
1000
setattr(self, attr, copy(getattr(otherweave, attr)))
1003
class WeaveFile(Weave):
1004
"""A WeaveFile represents a Weave on disk and writes on change."""
1006
WEAVE_SUFFIX = '.weave'
1008
def __init__(self, name, transport, filemode=None, create=False, access_mode='w'):
1009
"""Create a WeaveFile.
1011
:param create: If not True, only open an existing knit.
1013
super(WeaveFile, self).__init__(name, access_mode)
1014
self._transport = transport
1015
self._filemode = filemode
1017
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
1018
except errors.NoSuchFile:
1024
def _add_lines(self, version_id, parents, lines, parent_texts,
1025
left_matching_blocks=None):
1026
"""Add a version and save the weave."""
1027
self.check_not_reserved_id(version_id)
1028
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
1033
def _clone_text(self, new_version_id, old_version_id, parents):
1034
"""See VersionedFile.clone_text."""
1035
super(WeaveFile, self)._clone_text(new_version_id, old_version_id, parents)
1038
def copy_to(self, name, transport):
1039
"""See VersionedFile.copy_to()."""
1040
# as we are all in memory always, just serialise to the new place.
1042
write_weave_v5(self, sio)
1044
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
1046
def create_empty(self, name, transport, filemode=None):
1047
return WeaveFile(name, transport, filemode, create=True)
1050
"""Save the weave."""
1051
self._check_write_ok()
1053
write_weave_v5(self, sio)
1055
self._transport.put_file(self._weave_name + WeaveFile.WEAVE_SUFFIX,
1061
"""See VersionedFile.get_suffixes()."""
1062
return [WeaveFile.WEAVE_SUFFIX]
1064
def join(self, other, pb=None, msg=None, version_ids=None,
1065
ignore_missing=False):
1066
"""Join other into self and save."""
1067
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
1071
def _reweave(wa, wb, pb=None, msg=None):
1072
"""Combine two weaves and return the result.
1074
This works even if a revision R has different parents in
1075
wa and wb. In the resulting weave all the parents are given.
1077
This is done by just building up a new weave, maintaining ordering
1078
of the versions in the two inputs. More efficient approaches
1079
might be possible but it should only be necessary to do
1080
this operation rarely, when a new previously ghost version is
1083
:param pb: An optional progress bar, indicating how far done we are
1084
:param msg: An optional message for the progress
1088
queue_a = range(wa.num_versions())
1089
queue_b = range(wb.num_versions())
1090
# first determine combined parents of all versions
1091
# map from version name -> all parent names
1092
combined_parents = _reweave_parent_graphs(wa, wb)
1093
mutter("combined parents: %r", combined_parents)
1094
order = topo_sort(combined_parents.iteritems())
1095
mutter("order to reweave: %r", order)
1100
for idx, name in enumerate(order):
1102
pb.update(msg, idx, len(order))
1103
if name in wa._name_map:
1104
lines = wa.get_lines(name)
1105
if name in wb._name_map:
1106
lines_b = wb.get_lines(name)
1107
if lines != lines_b:
1108
mutter('Weaves differ on content. rev_id {%s}', name)
1109
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
1111
lines = list(difflib.unified_diff(lines, lines_b,
1112
wa._weave_name, wb._weave_name))
1113
mutter('lines:\n%s', ''.join(lines))
1114
raise errors.WeaveTextDiffers(name, wa, wb)
1116
lines = wb.get_lines(name)
1117
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
1120
def _reweave_parent_graphs(wa, wb):
1121
"""Return combined parent ancestry for two weaves.
1123
Returned as a list of (version_name, set(parent_names))"""
1125
for weave in [wa, wb]:
1126
for idx, name in enumerate(weave._names):
1127
p = combined.setdefault(name, set())
1128
p.update(map(weave._idx_to_name, weave._parents[idx]))
1133
"""Show the weave's table-of-contents"""
1134
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
1135
for i in (6, 50, 10, 10):
679
for i in range(w.numversions()):
1138
for i in range(w.num_versions()):
680
1139
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
1141
parent_str = ' '.join(map(str, w._parents[i]))
1142
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
1146
def weave_stats(weave_file, pb):
687
1147
from bzrlib.weavefile import read_weave
691
1149
wf = file(weave_file, 'rb')
692
1150
w = read_weave(wf)
693
1151
# FIXME: doesn't work on pipes
added
removed
bzrlib/weave.py
