160
164
each version; the parent's parents are implied.
163
List of hex SHA-1 of each version, or None if not recorded.
167
List of hex SHA-1 of each version.
170
List of symbolic names for each version. Each should be unique.
173
For each name, the version number.
176
Descriptive name of this weave; typically the filename if known.
166
__slots__ = ['_weave', '_parents', '_sha1s']
180
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
183
def __init__(self, weave_name=None):
170
185
self._parents = []
189
self._weave_name = weave_name
192
return "Weave(%r)" % self._weave_name
195
"""Return a deep copy of self.
197
The copy can be modified without affecting the original weave."""
199
other._weave = self._weave[:]
200
other._parents = self._parents[:]
201
other._sha1s = self._sha1s[:]
202
other._names = self._names[:]
203
other._name_map = self._name_map.copy()
204
other._weave_name = self._weave_name
174
207
def __eq__(self, other):
175
208
if not isinstance(other, Weave):
177
210
return self._parents == other._parents \
178
and self._weave == other._weave
211
and self._weave == other._weave \
212
and self._sha1s == other._sha1s
181
214
def __ne__(self, other):
182
215
return not self.__eq__(other)
185
def add(self, parents, text):
217
@deprecated_method(zero_eight)
218
def idx_to_name(self, index):
219
"""Old public interface, the public interface is all names now."""
222
def _idx_to_name(self, version):
223
return self._names[version]
225
@deprecated_method(zero_eight)
226
def lookup(self, name):
227
"""Backwards compatability thunk:
229
Return name, as name is valid in the api now, and spew deprecation
234
def _lookup(self, name):
235
"""Convert symbolic version name to index."""
237
return self._name_map[name]
239
raise RevisionNotPresent(name, self._weave_name)
241
@deprecated_method(zero_eight)
242
def iter_names(self):
243
"""Deprecated convenience function, please see VersionedFile.names()."""
244
return iter(self.names())
246
@deprecated_method(zero_eight)
248
"""See Weave.versions for the current api."""
249
return self.versions()
252
"""See VersionedFile.versions."""
253
return self._names[:]
255
def has_version(self, version_id):
256
"""See VersionedFile.has_version."""
257
return self._name_map.has_key(version_id)
259
__contains__ = has_version
261
def get_parents(self, version_id):
262
"""See VersionedFile.get_parent."""
263
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
265
def _check_repeated_add(self, name, parents, text, sha1):
266
"""Check that a duplicated add is OK.
268
If it is, return the (old) index; otherwise raise an exception.
270
idx = self._lookup(name)
271
if sorted(self._parents[idx]) != sorted(parents) \
272
or sha1 != self._sha1s[idx]:
273
raise RevisionAlreadyPresent(name, self._weave_name)
276
@deprecated_method(zero_eight)
277
def add_identical(self, old_rev_id, new_rev_id, parents):
278
"""Please use Weave.clone_text now."""
279
return self.clone_text(new_rev_id, old_rev_id, parents)
281
def add_lines(self, version_id, parents, lines):
282
"""See VersionedFile.add_lines."""
283
return self._add(version_id, lines, map(self._lookup, parents))
285
@deprecated_method(zero_eight)
286
def add(self, name, parents, text, sha1=None):
287
"""See VersionedFile.add_lines for the non deprecated api."""
288
return self._add(name, text, map(self._maybe_lookup, parents), sha1)
290
def _add(self, version_id, lines, parents, sha1=None):
186
291
"""Add a single text on top of the weave.
188
293
Returns the index number of the newly added version.
296
Symbolic name for this version.
297
(Typically the revision-id of the revision that added it.)
191
300
List or set of direct parent version numbers.
194
Sequence of lines to be added in the new version."""
303
Sequence of lines to be added in the new version.
306
assert isinstance(version_id, basestring)
308
sha1 = sha_strings(lines)
309
if version_id in self._name_map:
310
return self._check_repeated_add(version_id, parents, lines, sha1)
196
312
self._check_versions(parents)
197
## self._check_lines(text)
313
## self._check_lines(lines)
198
314
new_version = len(self._parents)
206
# if we abort after here the weave will be corrupt
207
self._parents.append(frozenset(parents))
316
# if we abort after here the (in-memory) weave will be corrupt because only
317
# some fields are updated
318
self._parents.append(parents[:])
208
319
self._sha1s.append(sha1)
320
self._names.append(version_id)
321
self._name_map[version_id] = new_version
450
571
result.append((istack[-1], lineno, l))
454
raise WFE("unclosed insertion blocks at end of weave",
574
raise WeaveFormatError("unclosed insertion blocks "
575
"at end of weave: %s" % istack)
457
raise WFE("unclosed deletion blocks at end of weave",
577
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
464
def get_iter(self, version):
581
@deprecated_method(zero_eight)
582
def get_iter(self, name_or_index):
583
"""Deprecated, please do not use. Lookups are not not needed.
585
Please use get_lines now.
587
return self._get_iter(self._maybe_lookup(name_or_index))
589
@deprecated_method(zero_eight)
590
def maybe_lookup(self, name_or_index):
591
"""Deprecated, please do not use. Lookups are not not needed."""
592
return self._maybe_lookup(name_or_index)
594
def _maybe_lookup(self, name_or_index):
595
"""Convert possible symbolic name to index, or pass through indexes.
599
if isinstance(name_or_index, (int, long)):
602
return self._lookup(name_or_index)
604
def _get_iter(self, version_id):
465
605
"""Yield lines for the specified version."""
466
for origin, lineno, line in self._extract([version]):
606
incls = [self._maybe_lookup(version_id)]
611
# We don't have sha1 sums for multiple entries
613
for origin, lineno, line in self._extract(incls):
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)
618
expected_sha1 = self._sha1s[index]
619
measured_sha1 = cur_sha.hexdigest()
620
if measured_sha1 != expected_sha1:
621
raise errors.WeaveInvalidChecksum(
622
'file %s, revision %s, expected: %s, measured %s'
623
% (self._weave_name, self._names[index],
624
expected_sha1, measured_sha1))
626
@deprecated_method(zero_eight)
627
def get(self, version_id):
628
"""Please use either Weave.get_text or Weave.get_lines as desired."""
629
return self.get_lines(version_id)
631
def get_lines(self, version_id):
632
"""See VersionedFile.get_lines()."""
633
return list(self._get_iter(version_id))
635
def get_sha1(self, name):
636
"""Get the stored sha1 sum for the given revision.
638
:param name: The name of the version to lookup
640
return self._sha1s[self._lookup(name)]
642
@deprecated_method(zero_eight)
489
643
def numversions(self):
644
"""How many versions are in this weave?
646
Deprecated in favour of num_versions.
648
return self.num_versions()
650
def num_versions(self):
651
"""How many versions are in this weave?"""
490
652
l = len(self._parents)
491
653
assert l == len(self._sha1s)
496
return self.numversions()
656
__len__ = num_versions
499
658
def check(self, progress_bar=None):
659
# TODO evaluate performance hit of using string sets in this routine.
500
660
# check no circular inclusions
501
for version in range(self.numversions()):
661
for version in range(self.num_versions()):
502
662
inclusions = list(self._parents[version])
504
664
inclusions.sort()
506
666
raise WeaveFormatError("invalid included version %d for index %d"
507
667
% (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):
669
# try extracting all versions; parallel extraction is used
670
nv = self.num_versions()
675
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
676
# The problem is that set membership is much more expensive
677
name = self._idx_to_name(i)
678
sha1s[name] = sha.new()
680
new_inc = set([name])
681
for p in self._parents[i]:
682
new_inc.update(inclusions[self._idx_to_name(p)])
684
assert set(new_inc) == set(self.get_ancestry(name)), \
685
'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
686
inclusions[name] = new_inc
688
nlines = len(self._weave)
690
update_text = 'checking weave'
692
short_name = os.path.basename(self._weave_name)
693
update_text = 'checking %s' % (short_name,)
694
update_text = update_text[:25]
696
for lineno, insert, deleteset, line in self.walk():
515
progress_bar.update('checking text', version, nv)
517
for l in self.get_iter(version):
520
expected = self._sha1s[version]
698
progress_bar.update(update_text, lineno, nlines)
700
for name, name_inclusions in inclusions.items():
701
# The active inclusion must be an ancestor,
702
# and no ancestors must have deleted this line,
703
# because we don't support resurrection.
704
if (insert in name_inclusions) and not (deleteset & name_inclusions):
705
sha1s[name].update(line)
708
version = self._idx_to_name(i)
709
hd = sha1s[version].hexdigest()
710
expected = self._sha1s[i]
521
711
if hd != expected:
522
raise WeaveError("mismatched sha1 for version %d; "
523
"got %s, expected %s"
524
% (version, hd, expected))
712
raise errors.WeaveInvalidChecksum(
713
"mismatched sha1 for version %s: "
714
"got %s, expected %s"
715
% (version, hd, expected))
526
717
# TODO: check insertions are properly nested, that there are
527
718
# no lines outside of insertion blocks, that deletions are
528
719
# 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):
721
def _join(self, other, pb, msg, version_ids):
722
"""Worker routine for join()."""
723
if not other.versions():
724
return # nothing to update, easy
727
for version_id in version_ids:
728
if not self.has_version(version_id):
729
raise RevisionNotPresent(version_id, self._weave_name)
730
assert version_ids == None
732
# two loops so that we do not change ourselves before verifying it
734
# work through in index order to make sure we get all dependencies
737
for other_idx, name in enumerate(other._names):
738
self._check_version_consistent(other, other_idx, name)
739
sha1 = other._sha1s[other_idx]
743
if name in self._name_map:
744
idx = self._lookup(name)
745
n1 = set(map(other._idx_to_name, other._parents[other_idx]))
746
n2 = set(map(self._idx_to_name, self._parents[idx]))
747
if sha1 == self._sha1s[idx] and n1 == n2:
750
names_to_join.append((other_idx, name))
757
for other_idx, name in names_to_join:
758
# TODO: If all the parents of the other version are already
759
# present then we can avoid some work by just taking the delta
760
# and adjusting the offsets.
761
new_parents = self._imported_parents(other, other_idx)
762
sha1 = other._sha1s[other_idx]
767
pb.update(msg, merged, len(names_to_join))
769
lines = other.get_lines(other_idx)
770
self._add(name, lines, new_parents, sha1)
772
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
773
merged, processed, self._weave_name, time.time()-time0))
775
def _imported_parents(self, other, other_idx):
776
"""Return list of parents in self corresponding to indexes in other."""
778
for parent_idx in other._parents[other_idx]:
779
parent_name = other._names[parent_idx]
780
if parent_name not in self._names:
781
# should not be possible
782
raise WeaveError("missing parent {%s} of {%s} in %r"
783
% (parent_name, other._name_map[other_idx], self))
784
new_parents.append(self._name_map[parent_name])
787
def _check_version_consistent(self, other, other_idx, name):
788
"""Check if a version in consistent in this and other.
790
To be consistent it must have:
793
* the same direct parents (by name, not index, and disregarding
796
If present & correct return True;
797
if not present in self return False;
798
if inconsistent raise error."""
799
this_idx = self._name_map.get(name, -1)
801
if self._sha1s[this_idx] != other._sha1s[other_idx]:
802
raise errors.WeaveTextDiffers(name, self, other)
803
self_parents = self._parents[this_idx]
804
other_parents = other._parents[other_idx]
805
n1 = set([self._names[i] for i in self_parents])
806
n2 = set([other._names[i] for i in other_parents])
808
raise WeaveParentMismatch("inconsistent parents "
809
"for version {%s}: %s vs %s" % (name, n1, n2))
815
@deprecated_method(zero_eight)
816
def reweave(self, other, pb=None, msg=None):
817
"""reweave has been superceded by plain use of join."""
818
return self.join(other, pb, msg)
820
def _reweave(self, other, pb, msg):
821
"""Reweave self with other - internal helper for join().
823
:param other: The other weave to merge
824
:param pb: An optional progress bar, indicating how far done we are
825
:param msg: An optional message for the progress
827
new_weave = _reweave(self, other, pb=pb, msg=msg)
828
for attr in self.__slots__:
829
if attr != '_weave_name':
830
setattr(self, attr, getattr(new_weave, attr))
833
class WeaveFile(Weave):
834
"""A WeaveFile represents a Weave on disk and writes on change."""
836
WEAVE_SUFFIX = '.weave'
838
def __init__(self, name, transport, mode=None):
839
super(WeaveFile, self).__init__(name)
840
self._transport = transport
843
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
844
except errors.NoSuchFile:
848
def add_lines(self, version_id, parents, lines):
849
"""Add a version and save the weave."""
850
super(WeaveFile, self).add_lines(version_id, parents, lines)
853
def copy_to(self, name, transport):
854
"""See VersionedFile.copy_to()."""
855
# as we are all in memory always, just serialise to the new place.
857
write_weave_v5(self, sio)
859
transport.put(name + WeaveFile.WEAVE_SUFFIX, sio, self._mode)
861
def create_empty(self, name, transport, mode=None):
862
return WeaveFile(name, transport, mode)
865
"""Save the weave."""
867
write_weave_v5(self, sio)
869
self._transport.put(self._weave_name + WeaveFile.WEAVE_SUFFIX,
875
"""See VersionedFile.get_suffixes()."""
876
return [WeaveFile.WEAVE_SUFFIX]
878
def join(self, other, pb=None, msg=None, version_ids=None):
879
"""Join other into self and save."""
880
super(WeaveFile, self).join(other, pb, msg, version_ids)
884
@deprecated_function(zero_eight)
885
def reweave(wa, wb, pb=None, msg=None):
886
"""reweaving is deprecation, please just use weave.join()."""
887
_reweave(wa, wb, pb, msg)
889
def _reweave(wa, wb, pb=None, msg=None):
890
"""Combine two weaves and return the result.
892
This works even if a revision R has different parents in
893
wa and wb. In the resulting weave all the parents are given.
895
This is done by just building up a new weave, maintaining ordering
896
of the versions in the two inputs. More efficient approaches
897
might be possible but it should only be necessary to do
898
this operation rarely, when a new previously ghost version is
901
:param pb: An optional progress bar, indicating how far done we are
902
:param msg: An optional message for the progress
906
queue_a = range(wa.num_versions())
907
queue_b = range(wb.num_versions())
908
# first determine combined parents of all versions
909
# map from version name -> all parent names
910
combined_parents = _reweave_parent_graphs(wa, wb)
911
mutter("combined parents: %r", combined_parents)
912
order = topo_sort(combined_parents.iteritems())
913
mutter("order to reweave: %r", order)
918
for idx, name in enumerate(order):
920
pb.update(msg, idx, len(order))
921
if name in wa._name_map:
922
lines = wa.get_lines(name)
923
if name in wb._name_map:
924
lines_b = wb.get_lines(name)
926
mutter('Weaves differ on content. rev_id {%s}', name)
927
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
929
lines = list(difflib.unified_diff(lines, lines_b,
930
wa._weave_name, wb._weave_name))
931
mutter('lines:\n%s', ''.join(lines))
932
raise errors.WeaveTextDiffers(name, wa, wb)
934
lines = wb.get_lines(name)
935
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
938
def _reweave_parent_graphs(wa, wb):
939
"""Return combined parent ancestry for two weaves.
941
Returned as a list of (version_name, set(parent_names))"""
943
for weave in [wa, wb]:
944
for idx, name in enumerate(weave._names):
945
p = combined.setdefault(name, set())
946
p.update(map(weave._idx_to_name, weave._parents[idx]))
951
"""Show the weave's table-of-contents"""
952
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
953
for i in (6, 50, 10, 10):
679
for i in range(w.numversions()):
956
for i in range(w.num_versions()):
680
957
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
959
parent_str = ' '.join(map(str, w._parents[i]))
960
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
964
def weave_stats(weave_file, pb):
687
965
from bzrlib.weavefile import read_weave
691
967
wf = file(weave_file, 'rb')
968
w = read_weave(wf, WeaveVersionedFile)
693
969
# FIXME: doesn't work on pipes
694
970
weave_size = wf.tell()
added
removed
bzrlib/weave.py
