205
212
return self._parents == other._parents \
206
213
and self._weave == other._weave \
207
214
and self._sha1s == other._sha1s
210
216
def __ne__(self, other):
211
217
return not self.__eq__(other)
219
@deprecated_method(zero_eight)
220
def idx_to_name(self, index):
221
"""Old public interface, the public interface is all names now."""
224
def _idx_to_name(self, version):
225
return self._names[version]
227
@deprecated_method(zero_eight)
214
228
def lookup(self, name):
229
"""Backwards compatability thunk:
231
Return name, as name is valid in the api now, and spew deprecation
236
def _lookup(self, name):
237
"""Convert symbolic version name to index."""
216
239
return self._name_map[name]
218
raise WeaveError("name %s not present in weave %s" %
219
(name, self._weave_name))
222
def add(self, name, parents, text):
241
raise RevisionNotPresent(name, self._weave_name)
243
@deprecated_method(zero_eight)
244
def iter_names(self):
245
"""Deprecated convenience function, please see VersionedFile.names()."""
246
return iter(self.names())
248
@deprecated_method(zero_eight)
250
"""See Weave.versions for the current api."""
251
return self.versions()
254
"""See VersionedFile.versions."""
255
return self._names[:]
257
def has_version(self, version_id):
258
"""See VersionedFile.has_version."""
259
return self._name_map.has_key(version_id)
261
__contains__ = has_version
263
def get_parents(self, version_id):
264
"""See VersionedFile.get_parent."""
265
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
267
def _check_repeated_add(self, name, parents, text, sha1):
268
"""Check that a duplicated add is OK.
270
If it is, return the (old) index; otherwise raise an exception.
272
idx = self._lookup(name)
273
if sorted(self._parents[idx]) != sorted(parents) \
274
or sha1 != self._sha1s[idx]:
275
raise RevisionAlreadyPresent(name, self._weave_name)
278
@deprecated_method(zero_eight)
279
def add_identical(self, old_rev_id, new_rev_id, parents):
280
"""Please use Weave.clone_text now."""
281
return self.clone_text(new_rev_id, old_rev_id, parents)
283
def _add_lines(self, version_id, parents, lines):
284
"""See VersionedFile.add_lines."""
285
return self._add(version_id, lines, map(self._lookup, parents))
287
@deprecated_method(zero_eight)
288
def add(self, name, parents, text, sha1=None):
289
"""See VersionedFile.add_lines for the non deprecated api."""
290
return self._add(name, text, map(self._maybe_lookup, parents), sha1)
292
def _add(self, version_id, lines, parents, sha1=None):
223
293
"""Add a single text on top of the weave.
225
295
Returns the index number of the newly added version.
228
298
Symbolic name for this version.
229
299
(Typically the revision-id of the revision that added it.)
232
302
List or set of direct parent version numbers.
235
Sequence of lines to be added in the new version."""
237
assert isinstance(name, basestring)
238
if name in self._name_map:
239
raise WeaveError("name %r already present in weave" % name)
305
Sequence of lines to be added in the new version.
308
assert isinstance(version_id, basestring)
310
sha1 = sha_strings(lines)
311
if version_id in self._name_map:
312
return self._check_repeated_add(version_id, parents, lines, sha1)
241
314
self._check_versions(parents)
242
## self._check_lines(text)
315
## self._check_lines(lines)
243
316
new_version = len(self._parents)
250
318
# if we abort after here the (in-memory) weave will be corrupt because only
251
319
# some fields are updated
320
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
321
# - Robert Collins 20060226
252
322
self._parents.append(parents[:])
253
323
self._sha1s.append(sha1)
254
self._names.append(name)
255
self._name_map[name] = new_version
324
self._names.append(version_id)
325
self._name_map[version_id] = new_version
501
605
result.append((istack[-1], lineno, l))
505
raise WFE("unclosed insertion blocks at end of weave",
608
raise WeaveFormatError("unclosed insertion blocks "
609
"at end of weave: %s" % istack)
508
raise WFE("unclosed deletion blocks at end of weave",
611
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
515
def get_iter(self, version):
615
@deprecated_method(zero_eight)
616
def get_iter(self, name_or_index):
617
"""Deprecated, please do not use. Lookups are not not needed.
619
Please use get_lines now.
621
return self._get_iter(self._maybe_lookup(name_or_index))
623
@deprecated_method(zero_eight)
624
def maybe_lookup(self, name_or_index):
625
"""Deprecated, please do not use. Lookups are not not needed."""
626
return self._maybe_lookup(name_or_index)
628
def _maybe_lookup(self, name_or_index):
629
"""Convert possible symbolic name to index, or pass through indexes.
633
if isinstance(name_or_index, (int, long)):
636
return self._lookup(name_or_index)
638
def _get_iter(self, version_id):
516
639
"""Yield lines for the specified version."""
517
for origin, lineno, line in self._extract([version]):
640
incls = [self._maybe_lookup(version_id)]
645
# We don't have sha1 sums for multiple entries
647
for origin, lineno, line in self._extract(incls):
521
def get_text(self, version):
522
assert isinstance(version, int)
524
s.writelines(self.get_iter(version))
528
def get(self, index):
529
return list(self.get_iter(index))
532
def mash_iter(self, included):
533
"""Return composed version of multiple included versions."""
534
for origin, lineno, text in self._extract(included):
538
def dump(self, to_file):
539
from pprint import pprint
540
print >>to_file, "Weave._weave = ",
541
pprint(self._weave, to_file)
542
print >>to_file, "Weave._parents = ",
543
pprint(self._parents, to_file)
652
expected_sha1 = self._sha1s[index]
653
measured_sha1 = cur_sha.hexdigest()
654
if measured_sha1 != expected_sha1:
655
raise errors.WeaveInvalidChecksum(
656
'file %s, revision %s, expected: %s, measured %s'
657
% (self._weave_name, self._names[index],
658
expected_sha1, measured_sha1))
660
@deprecated_method(zero_eight)
661
def get(self, version_id):
662
"""Please use either Weave.get_text or Weave.get_lines as desired."""
663
return self.get_lines(version_id)
665
def get_lines(self, version_id):
666
"""See VersionedFile.get_lines()."""
667
return list(self._get_iter(version_id))
669
def get_sha1(self, name):
670
"""Get the stored sha1 sum for the given revision.
672
:param name: The name of the version to lookup
674
return self._sha1s[self._lookup(name)]
676
@deprecated_method(zero_eight)
547
677
def numversions(self):
678
"""How many versions are in this weave?
680
Deprecated in favour of num_versions.
682
return self.num_versions()
684
def num_versions(self):
685
"""How many versions are in this weave?"""
548
686
l = len(self._parents)
549
687
assert l == len(self._sha1s)
554
return self.numversions()
690
__len__ = num_versions
557
692
def check(self, progress_bar=None):
693
# TODO evaluate performance hit of using string sets in this routine.
558
694
# check no circular inclusions
559
for version in range(self.numversions()):
695
for version in range(self.num_versions()):
560
696
inclusions = list(self._parents[version])
562
698
inclusions.sort()
564
700
raise WeaveFormatError("invalid included version %d for index %d"
565
701
% (inclusions[-1], version))
567
# try extracting all versions; this is a bit slow and parallel
568
# extraction could be used
569
nv = self.numversions()
570
for version in range(nv):
703
# try extracting all versions; parallel extraction is used
704
nv = self.num_versions()
709
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
710
# The problem is that set membership is much more expensive
711
name = self._idx_to_name(i)
712
sha1s[name] = sha.new()
714
new_inc = set([name])
715
for p in self._parents[i]:
716
new_inc.update(inclusions[self._idx_to_name(p)])
718
assert set(new_inc) == set(self.get_ancestry(name)), \
719
'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
720
inclusions[name] = new_inc
722
nlines = len(self._weave)
724
update_text = 'checking weave'
726
short_name = os.path.basename(self._weave_name)
727
update_text = 'checking %s' % (short_name,)
728
update_text = update_text[:25]
730
for lineno, insert, deleteset, line in self._walk_internal():
572
progress_bar.update('checking text', version, nv)
574
for l in self.get_iter(version):
577
expected = self._sha1s[version]
732
progress_bar.update(update_text, lineno, nlines)
734
for name, name_inclusions in inclusions.items():
735
# The active inclusion must be an ancestor,
736
# and no ancestors must have deleted this line,
737
# because we don't support resurrection.
738
if (insert in name_inclusions) and not (deleteset & name_inclusions):
739
sha1s[name].update(line)
742
version = self._idx_to_name(i)
743
hd = sha1s[version].hexdigest()
744
expected = self._sha1s[i]
578
745
if hd != expected:
579
raise WeaveError("mismatched sha1 for version %d; "
580
"got %s, expected %s"
581
% (version, hd, expected))
746
raise errors.WeaveInvalidChecksum(
747
"mismatched sha1 for version %s: "
748
"got %s, expected %s"
749
% (version, hd, expected))
583
751
# TODO: check insertions are properly nested, that there are
584
752
# no lines outside of insertion blocks, that deletions are
585
753
# properly paired, etc.
589
def merge(self, merge_versions):
590
"""Automerge and mark conflicts between versions.
592
This returns a sequence, each entry describing alternatives
593
for a chunk of the file. Each of the alternatives is given as
596
If there is a chunk of the file where there's no diagreement,
597
only one alternative is given.
600
# approach: find the included versions common to all the
602
raise NotImplementedError()
606
def _delta(self, included, lines):
607
"""Return changes from basis to new revision.
609
The old text for comparison is the union of included revisions.
611
This is used in inserting a new text.
613
Delta is returned as a sequence of
614
(weave1, weave2, newlines).
616
This indicates that weave1:weave2 of the old weave should be
617
replaced by the sequence of lines in newlines. Note that
618
these line numbers are positions in the total weave and don't
619
correspond to the lines in any extracted version, or even the
620
extracted union of included versions.
622
If line1=line2, this is a pure insert; if newlines=[] this is a
623
pure delete. (Similar to difflib.)
628
def plan_merge(self, ver_a, ver_b):
629
"""Return pseudo-annotation indicating how the two versions merge.
631
This is computed between versions a and b and their common
634
Weave lines present in none of them are skipped entirely.
636
inc_a = self.inclusions([ver_a])
637
inc_b = self.inclusions([ver_b])
638
inc_c = inc_a & inc_b
640
for lineno, insert, deleteset, line in self._walk():
641
if deleteset & inc_c:
642
# killed in parent; can't be in either a or b
643
# not relevant to our work
644
yield 'killed-base', line
645
elif insert in inc_c:
646
# was inserted in base
647
killed_a = bool(deleteset & inc_a)
648
killed_b = bool(deleteset & inc_b)
649
if killed_a and killed_b:
650
yield 'killed-both', line
652
yield 'killed-a', line
654
yield 'killed-b', line
656
yield 'unchanged', line
657
elif insert in inc_a:
658
if deleteset & inc_a:
659
yield 'ghost-a', line
663
elif insert in inc_b:
664
if deleteset & inc_b:
665
yield 'ghost-b', line
669
# not in either revision
670
yield 'irrelevant', line
672
yield 'unchanged', '' # terminator
676
def weave_merge(self, plan):
681
for state, line in plan:
682
if state == 'unchanged' or state == 'killed-both':
683
# resync and flush queued conflicts changes if any
684
if not lines_a and not lines_b:
686
elif ch_a and not ch_b:
688
for l in lines_a: yield l
689
elif ch_b and not ch_a:
690
for l in lines_b: yield l
691
elif lines_a == lines_b:
692
for l in lines_a: yield l
695
for l in lines_a: yield l
697
for l in lines_b: yield l
704
if state == 'unchanged':
707
elif state == 'killed-a':
710
elif state == 'killed-b':
713
elif state == 'new-a':
716
elif state == 'new-b':
720
assert state in ('irrelevant', 'ghost-a', 'ghost-b', 'killed-base',
755
def _join(self, other, pb, msg, version_ids, ignore_missing):
756
"""Worker routine for join()."""
757
if not other.versions():
758
return # nothing to update, easy
761
for version_id in version_ids:
762
if not other.has_version(version_id) and not ignore_missing:
763
raise RevisionNotPresent(version_id, self._weave_name)
765
version_ids = other.versions()
767
# two loops so that we do not change ourselves before verifying it
769
# work through in index order to make sure we get all dependencies
772
# get the selected versions only that are in other.versions.
773
version_ids = set(other.versions()).intersection(set(version_ids))
774
# pull in the referenced graph.
775
version_ids = other.get_ancestry(version_ids)
776
pending_graph = [(version, other.get_parents(version)) for
777
version in version_ids]
778
for name in topo_sort(pending_graph):
779
other_idx = other._name_map[name]
780
self._check_version_consistent(other, other_idx, name)
781
sha1 = other._sha1s[other_idx]
785
if name in self._name_map:
786
idx = self._lookup(name)
787
n1 = set(map(other._idx_to_name, other._parents[other_idx]))
788
n2 = set(map(self._idx_to_name, self._parents[idx]))
789
if sha1 == self._sha1s[idx] and n1 == n2:
792
names_to_join.append((other_idx, name))
799
for other_idx, name in names_to_join:
800
# TODO: If all the parents of the other version are already
801
# present then we can avoid some work by just taking the delta
802
# and adjusting the offsets.
803
new_parents = self._imported_parents(other, other_idx)
804
sha1 = other._sha1s[other_idx]
809
pb.update(msg, merged, len(names_to_join))
811
lines = other.get_lines(other_idx)
812
self._add(name, lines, new_parents, sha1)
814
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
815
merged, processed, self._weave_name, time.time()-time0))
817
def _imported_parents(self, other, other_idx):
818
"""Return list of parents in self corresponding to indexes in other."""
820
for parent_idx in other._parents[other_idx]:
821
parent_name = other._names[parent_idx]
822
if parent_name not in self._names:
823
# should not be possible
824
raise WeaveError("missing parent {%s} of {%s} in %r"
825
% (parent_name, other._name_map[other_idx], self))
826
new_parents.append(self._name_map[parent_name])
829
def _check_version_consistent(self, other, other_idx, name):
830
"""Check if a version in consistent in this and other.
832
To be consistent it must have:
835
* the same direct parents (by name, not index, and disregarding
838
If present & correct return True;
839
if not present in self return False;
840
if inconsistent raise error."""
841
this_idx = self._name_map.get(name, -1)
843
if self._sha1s[this_idx] != other._sha1s[other_idx]:
844
raise errors.WeaveTextDiffers(name, self, other)
845
self_parents = self._parents[this_idx]
846
other_parents = other._parents[other_idx]
847
n1 = set([self._names[i] for i in self_parents])
848
n2 = set([other._names[i] for i in other_parents])
849
if not self._compatible_parents(n1, n2):
850
raise WeaveParentMismatch("inconsistent parents "
851
"for version {%s}: %s vs %s" % (name, n1, n2))
857
@deprecated_method(zero_eight)
858
def reweave(self, other, pb=None, msg=None):
859
"""reweave has been superceded by plain use of join."""
860
return self.join(other, pb, msg)
862
def _reweave(self, other, pb, msg):
863
"""Reweave self with other - internal helper for join().
865
:param other: The other weave to merge
866
:param pb: An optional progress bar, indicating how far done we are
867
:param msg: An optional message for the progress
869
new_weave = _reweave(self, other, pb=pb, msg=msg)
870
self._copy_weave_content(new_weave)
872
def _copy_weave_content(self, otherweave):
873
"""adsorb the content from otherweave."""
874
for attr in self.__slots__:
875
if attr != '_weave_name':
876
setattr(self, attr, copy(getattr(otherweave, attr)))
879
class WeaveFile(Weave):
880
"""A WeaveFile represents a Weave on disk and writes on change."""
882
WEAVE_SUFFIX = '.weave'
884
def __init__(self, name, transport, filemode=None, create=False, access_mode='w'):
885
"""Create a WeaveFile.
887
:param create: If not True, only open an existing knit.
889
super(WeaveFile, self).__init__(name, access_mode)
890
self._transport = transport
891
self._filemode = filemode
893
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
894
except errors.NoSuchFile:
900
def _add_lines(self, version_id, parents, lines):
901
"""Add a version and save the weave."""
902
super(WeaveFile, self)._add_lines(version_id, parents, lines)
905
def _clone_text(self, new_version_id, old_version_id, parents):
906
"""See VersionedFile.clone_text."""
907
super(WeaveFile, self)._clone_text(new_version_id, old_version_id, parents)
910
def copy_to(self, name, transport):
911
"""See VersionedFile.copy_to()."""
912
# as we are all in memory always, just serialise to the new place.
914
write_weave_v5(self, sio)
916
transport.put(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
918
def create_empty(self, name, transport, filemode=None):
919
return WeaveFile(name, transport, filemode, create=True)
922
"""Save the weave."""
923
self._check_write_ok()
925
write_weave_v5(self, sio)
927
self._transport.put(self._weave_name + WeaveFile.WEAVE_SUFFIX,
933
"""See VersionedFile.get_suffixes()."""
934
return [WeaveFile.WEAVE_SUFFIX]
936
def join(self, other, pb=None, msg=None, version_ids=None,
937
ignore_missing=False):
938
"""Join other into self and save."""
939
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
943
@deprecated_function(zero_eight)
944
def reweave(wa, wb, pb=None, msg=None):
945
"""reweaving is deprecation, please just use weave.join()."""
946
_reweave(wa, wb, pb, msg)
948
def _reweave(wa, wb, pb=None, msg=None):
949
"""Combine two weaves and return the result.
951
This works even if a revision R has different parents in
952
wa and wb. In the resulting weave all the parents are given.
954
This is done by just building up a new weave, maintaining ordering
955
of the versions in the two inputs. More efficient approaches
956
might be possible but it should only be necessary to do
957
this operation rarely, when a new previously ghost version is
960
:param pb: An optional progress bar, indicating how far done we are
961
:param msg: An optional message for the progress
965
queue_a = range(wa.num_versions())
966
queue_b = range(wb.num_versions())
967
# first determine combined parents of all versions
968
# map from version name -> all parent names
969
combined_parents = _reweave_parent_graphs(wa, wb)
970
mutter("combined parents: %r", combined_parents)
971
order = topo_sort(combined_parents.iteritems())
972
mutter("order to reweave: %r", order)
977
for idx, name in enumerate(order):
979
pb.update(msg, idx, len(order))
980
if name in wa._name_map:
981
lines = wa.get_lines(name)
982
if name in wb._name_map:
983
lines_b = wb.get_lines(name)
985
mutter('Weaves differ on content. rev_id {%s}', name)
986
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
988
lines = list(difflib.unified_diff(lines, lines_b,
989
wa._weave_name, wb._weave_name))
990
mutter('lines:\n%s', ''.join(lines))
991
raise errors.WeaveTextDiffers(name, wa, wb)
993
lines = wb.get_lines(name)
994
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
997
def _reweave_parent_graphs(wa, wb):
998
"""Return combined parent ancestry for two weaves.
1000
Returned as a list of (version_name, set(parent_names))"""
1002
for weave in [wa, wb]:
1003
for idx, name in enumerate(weave._names):
1004
p = combined.setdefault(name, set())
1005
p.update(map(weave._idx_to_name, weave._parents[idx]))
730
1009
def weave_toc(w):
added
removed
bzrlib/weave.py
