160
165
each version; the parent's parents are implied.
163
List of hex SHA-1 of each version, or None if not recorded.
168
List of hex SHA-1 of each version.
171
List of symbolic names for each version. Each should be unique.
174
For each name, the version number.
177
Descriptive name of this weave; typically the filename if known.
166
__slots__ = ['_weave', '_parents', '_sha1s']
181
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
184
def __init__(self, weave_name=None, access_mode='w'):
185
super(Weave, self).__init__(access_mode)
170
187
self._parents = []
191
self._weave_name = weave_name
194
return "Weave(%r)" % self._weave_name
197
"""Return a deep copy of self.
199
The copy can be modified without affecting the original weave."""
201
other._weave = self._weave[:]
202
other._parents = self._parents[:]
203
other._sha1s = self._sha1s[:]
204
other._names = self._names[:]
205
other._name_map = self._name_map.copy()
206
other._weave_name = self._weave_name
174
209
def __eq__(self, other):
175
210
if not isinstance(other, Weave):
177
212
return self._parents == other._parents \
178
and self._weave == other._weave
213
and self._weave == other._weave \
214
and self._sha1s == other._sha1s
181
216
def __ne__(self, other):
182
217
return not self.__eq__(other)
185
def add(self, parents, text):
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)
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."""
239
return self._name_map[name]
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):
186
293
"""Add a single text on top of the weave.
188
295
Returns the index number of the newly added version.
298
Symbolic name for this version.
299
(Typically the revision-id of the revision that added it.)
191
302
List or set of direct parent version numbers.
194
Sequence of lines to be added in the new version."""
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)
196
314
self._check_versions(parents)
197
## self._check_lines(text)
315
## self._check_lines(lines)
198
316
new_version = len(self._parents)
206
# if we abort after here the weave will be corrupt
207
self._parents.append(frozenset(parents))
318
# if we abort after here the (in-memory) weave will be corrupt because only
319
# some fields are updated
320
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
321
# - Robert Collins 20060226
322
self._parents.append(parents[:])
208
323
self._sha1s.append(sha1)
324
self._names.append(version_id)
325
self._name_map[version_id] = new_version
450
605
result.append((istack[-1], lineno, l))
454
raise WFE("unclosed insertion blocks at end of weave",
608
raise WeaveFormatError("unclosed insertion blocks "
609
"at end of weave: %s" % istack)
457
raise WFE("unclosed deletion blocks at end of weave",
611
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
464
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):
465
639
"""Yield lines for the specified version."""
466
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):
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)
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)
489
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?"""
490
686
l = len(self._parents)
491
687
assert l == len(self._sha1s)
496
return self.numversions()
690
__len__ = num_versions
499
692
def check(self, progress_bar=None):
693
# TODO evaluate performance hit of using string sets in this routine.
500
694
# check no circular inclusions
501
for version in range(self.numversions()):
695
for version in range(self.num_versions()):
502
696
inclusions = list(self._parents[version])
504
698
inclusions.sort()
506
700
raise WeaveFormatError("invalid included version %d for index %d"
507
701
% (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):
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():
515
progress_bar.update('checking text', version, nv)
517
for l in self.get_iter(version):
520
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]
521
745
if hd != expected:
522
raise WeaveError("mismatched sha1 for version %d; "
523
"got %s, expected %s"
524
% (version, hd, expected))
746
raise errors.WeaveInvalidChecksum(
747
"mismatched sha1 for version %s: "
748
"got %s, expected %s"
749
% (version, hd, expected))
526
751
# TODO: check insertions are properly nested, that there are
527
752
# no lines outside of insertion blocks, that deletions are
528
753
# 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):
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
# returns True if we have it, False if we need it.
781
if not self._check_version_consistent(other, other_idx, name):
782
names_to_join.append((other_idx, name))
791
for other_idx, name in names_to_join:
792
# TODO: If all the parents of the other version are already
793
# present then we can avoid some work by just taking the delta
794
# and adjusting the offsets.
795
new_parents = self._imported_parents(other, other_idx)
796
sha1 = other._sha1s[other_idx]
801
pb.update(msg, merged, len(names_to_join))
803
lines = other.get_lines(other_idx)
804
self._add(name, lines, new_parents, sha1)
806
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
807
merged, processed, self._weave_name, time.time()-time0))
809
def _imported_parents(self, other, other_idx):
810
"""Return list of parents in self corresponding to indexes in other."""
812
for parent_idx in other._parents[other_idx]:
813
parent_name = other._names[parent_idx]
814
if parent_name not in self._names:
815
# should not be possible
816
raise WeaveError("missing parent {%s} of {%s} in %r"
817
% (parent_name, other._name_map[other_idx], self))
818
new_parents.append(self._name_map[parent_name])
821
def _check_version_consistent(self, other, other_idx, name):
822
"""Check if a version in consistent in this and other.
824
To be consistent it must have:
827
* the same direct parents (by name, not index, and disregarding
830
If present & correct return True;
831
if not present in self return False;
832
if inconsistent raise error."""
833
this_idx = self._name_map.get(name, -1)
835
if self._sha1s[this_idx] != other._sha1s[other_idx]:
836
raise errors.WeaveTextDiffers(name, self, other)
837
self_parents = self._parents[this_idx]
838
other_parents = other._parents[other_idx]
839
n1 = set([self._names[i] for i in self_parents])
840
n2 = set([other._names[i] for i in other_parents])
841
if not self._compatible_parents(n1, n2):
842
raise WeaveParentMismatch("inconsistent parents "
843
"for version {%s}: %s vs %s" % (name, n1, n2))
849
@deprecated_method(zero_eight)
850
def reweave(self, other, pb=None, msg=None):
851
"""reweave has been superceded by plain use of join."""
852
return self.join(other, pb, msg)
854
def _reweave(self, other, pb, msg):
855
"""Reweave self with other - internal helper for join().
857
:param other: The other weave to merge
858
:param pb: An optional progress bar, indicating how far done we are
859
:param msg: An optional message for the progress
861
new_weave = _reweave(self, other, pb=pb, msg=msg)
862
self._copy_weave_content(new_weave)
864
def _copy_weave_content(self, otherweave):
865
"""adsorb the content from otherweave."""
866
for attr in self.__slots__:
867
if attr != '_weave_name':
868
setattr(self, attr, copy(getattr(otherweave, attr)))
871
class WeaveFile(Weave):
872
"""A WeaveFile represents a Weave on disk and writes on change."""
874
WEAVE_SUFFIX = '.weave'
876
def __init__(self, name, transport, filemode=None, create=False, access_mode='w'):
877
"""Create a WeaveFile.
879
:param create: If not True, only open an existing knit.
881
super(WeaveFile, self).__init__(name, access_mode)
882
self._transport = transport
883
self._filemode = filemode
885
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
886
except errors.NoSuchFile:
892
def _add_lines(self, version_id, parents, lines):
893
"""Add a version and save the weave."""
894
super(WeaveFile, self)._add_lines(version_id, parents, lines)
897
def _clone_text(self, new_version_id, old_version_id, parents):
898
"""See VersionedFile.clone_text."""
899
super(WeaveFile, self)._clone_text(new_version_id, old_version_id, parents)
902
def copy_to(self, name, transport):
903
"""See VersionedFile.copy_to()."""
904
# as we are all in memory always, just serialise to the new place.
906
write_weave_v5(self, sio)
908
transport.put(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
910
def create_empty(self, name, transport, filemode=None):
911
return WeaveFile(name, transport, filemode, create=True)
914
"""Save the weave."""
915
self._check_write_ok()
917
write_weave_v5(self, sio)
919
self._transport.put(self._weave_name + WeaveFile.WEAVE_SUFFIX,
925
"""See VersionedFile.get_suffixes()."""
926
return [WeaveFile.WEAVE_SUFFIX]
928
def join(self, other, pb=None, msg=None, version_ids=None,
929
ignore_missing=False):
930
"""Join other into self and save."""
931
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
935
@deprecated_function(zero_eight)
936
def reweave(wa, wb, pb=None, msg=None):
937
"""reweaving is deprecation, please just use weave.join()."""
938
_reweave(wa, wb, pb, msg)
940
def _reweave(wa, wb, pb=None, msg=None):
941
"""Combine two weaves and return the result.
943
This works even if a revision R has different parents in
944
wa and wb. In the resulting weave all the parents are given.
946
This is done by just building up a new weave, maintaining ordering
947
of the versions in the two inputs. More efficient approaches
948
might be possible but it should only be necessary to do
949
this operation rarely, when a new previously ghost version is
952
:param pb: An optional progress bar, indicating how far done we are
953
:param msg: An optional message for the progress
957
queue_a = range(wa.num_versions())
958
queue_b = range(wb.num_versions())
959
# first determine combined parents of all versions
960
# map from version name -> all parent names
961
combined_parents = _reweave_parent_graphs(wa, wb)
962
mutter("combined parents: %r", combined_parents)
963
order = topo_sort(combined_parents.iteritems())
964
mutter("order to reweave: %r", order)
969
for idx, name in enumerate(order):
971
pb.update(msg, idx, len(order))
972
if name in wa._name_map:
973
lines = wa.get_lines(name)
974
if name in wb._name_map:
975
lines_b = wb.get_lines(name)
977
mutter('Weaves differ on content. rev_id {%s}', name)
978
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
980
lines = list(difflib.unified_diff(lines, lines_b,
981
wa._weave_name, wb._weave_name))
982
mutter('lines:\n%s', ''.join(lines))
983
raise errors.WeaveTextDiffers(name, wa, wb)
985
lines = wb.get_lines(name)
986
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
989
def _reweave_parent_graphs(wa, wb):
990
"""Return combined parent ancestry for two weaves.
992
Returned as a list of (version_name, set(parent_names))"""
994
for weave in [wa, wb]:
995
for idx, name in enumerate(weave._names):
996
p = combined.setdefault(name, set())
997
p.update(map(weave._idx_to_name, weave._parents[idx]))
1002
"""Show the weave's table-of-contents"""
1003
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
1004
for i in (6, 50, 10, 10):
679
for i in range(w.numversions()):
1007
for i in range(w.num_versions()):
680
1008
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
1010
parent_str = ' '.join(map(str, w._parents[i]))
1011
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
1015
def weave_stats(weave_file, pb):
687
1016
from bzrlib.weavefile import read_weave
691
1018
wf = file(weave_file, 'rb')
1019
w = read_weave(wf, WeaveVersionedFile)
693
1020
# FIXME: doesn't work on pipes
694
1021
weave_size = wf.tell()
added
removed
bzrlib/weave.py
