59
61
# where the basis and destination are unchanged.
61
63
# FIXME: Sometimes we will be given a parents list for a revision
62
# that includes some redundant parents (i.e. already a parent of
63
# something in the list.) We should eliminate them. This can
64
# that includes some redundant parents (i.e. already a parent of
65
# something in the list.) We should eliminate them. This can
64
66
# be done fairly efficiently because the sequence numbers constrain
65
67
# the possible relationships.
67
# FIXME: the conflict markers should be *7* characters
73
from ..lazy_import import lazy_import
74
lazy_import(globals(), """
75
from breezy import tsort
81
from ..errors import (
82
RevisionAlreadyPresent,
84
UnavailableRepresentation,
86
from ..osutils import dirname, sha, sha_strings, split_lines
87
from ..revision import NULL_REVISION
88
from ..sixish import (
91
from ..trace import mutter
92
from .versionedfile import (
99
from .weavefile import _read_weave_v5, write_weave_v5
102
class WeaveError(errors.BzrError):
104
_fmt = "Error in processing weave: %(msg)s"
106
def __init__(self, msg=None):
107
errors.BzrError.__init__(self)
111
class WeaveRevisionAlreadyPresent(WeaveError):
113
_fmt = "Revision {%(revision_id)s} already present in %(weave)s"
115
def __init__(self, revision_id, weave):
117
WeaveError.__init__(self)
118
self.revision_id = revision_id
122
class WeaveRevisionNotPresent(WeaveError):
124
_fmt = "Revision {%(revision_id)s} not present in %(weave)s"
126
def __init__(self, revision_id, weave):
127
WeaveError.__init__(self)
128
self.revision_id = revision_id
132
class WeaveFormatError(WeaveError):
134
_fmt = "Weave invariant violated: %(what)s"
136
def __init__(self, what):
137
WeaveError.__init__(self)
141
class WeaveParentMismatch(WeaveError):
143
_fmt = "Parents are mismatched between two revisions. %(msg)s"
146
class WeaveInvalidChecksum(WeaveError):
148
_fmt = "Text did not match its checksum: %(msg)s"
151
class WeaveTextDiffers(WeaveError):
153
_fmt = ("Weaves differ on text content. Revision:"
154
" {%(revision_id)s}, %(weave_a)s, %(weave_b)s")
156
def __init__(self, revision_id, weave_a, weave_b):
157
WeaveError.__init__(self)
158
self.revision_id = revision_id
159
self.weave_a = weave_a
160
self.weave_b = weave_b
163
class WeaveContentFactory(ContentFactory):
164
"""Content factory for streaming from weaves.
166
:seealso ContentFactory:
169
def __init__(self, version, weave):
170
"""Create a WeaveContentFactory for version from weave."""
171
ContentFactory.__init__(self)
172
self.sha1 = weave.get_sha1s([version])[version]
173
self.key = (version,)
174
parents = weave.get_parent_map([version])[version]
175
self.parents = tuple((parent,) for parent in parents)
176
self.storage_kind = 'fulltext'
179
def get_bytes_as(self, storage_kind):
180
if storage_kind == 'fulltext':
181
return self._weave.get_text(self.key[-1])
182
elif storage_kind in ('chunked', 'lines'):
183
return self._weave.get_lines(self.key[-1])
185
raise UnavailableRepresentation(self.key, storage_kind, 'fulltext')
187
def iter_bytes_as(self, storage_kind):
188
if storage_kind in ('chunked', 'lines'):
189
return iter(self._weave.get_lines(self.key[-1]))
191
raise UnavailableRepresentation(self.key, storage_kind, 'fulltext')
194
class Weave(VersionedFile):
71
from difflib import SequenceMatcher
73
from bzrlib.trace import mutter
74
from bzrlib.errors import WeaveError, WeaveFormatError, WeaveParentMismatch, \
75
WeaveRevisionNotPresent, WeaveRevisionAlreadyPresent
76
from bzrlib.tsort import topo_sort
195
80
"""weave - versioned text file storage.
197
82
A Weave manages versions of line-based text files, keeping track
198
83
of the originating version for each line.
337
194
if not isinstance(other, Weave):
339
196
return self._parents == other._parents \
340
and self._weave == other._weave \
341
and self._sha1s == other._sha1s
197
and self._weave == other._weave \
198
and self._sha1s == other._sha1s
343
201
def __ne__(self, other):
344
202
return not self.__eq__(other)
346
def _idx_to_name(self, version):
347
return self._names[version]
349
def _lookup(self, name):
204
def __contains__(self, name):
205
return self._name_map.has_key(name)
207
def maybe_lookup(self, name_or_index):
208
"""Convert possible symbolic name to index, or pass through indexes."""
209
if isinstance(name_or_index, (int, long)):
212
return self.lookup(name_or_index)
215
def lookup(self, name):
350
216
"""Convert symbolic version name to index."""
351
if not self._allow_reserved:
352
self.check_not_reserved_id(name)
354
218
return self._name_map[name]
356
raise RevisionNotPresent(name, self._weave_name)
220
raise WeaveRevisionNotPresent(name, self)
359
"""See VersionedFile.versions."""
360
223
return self._names[:]
362
def has_version(self, version_id):
363
"""See VersionedFile.has_version."""
364
return (version_id in self._name_map)
366
__contains__ = has_version
368
def get_record_stream(self, versions, ordering, include_delta_closure):
369
"""Get a stream of records for versions.
371
:param versions: The versions to include. Each version is a tuple
373
:param ordering: Either 'unordered' or 'topological'. A topologically
374
sorted stream has compression parents strictly before their
376
:param include_delta_closure: If True then the closure across any
377
compression parents will be included (in the opaque data).
378
:return: An iterator of ContentFactory objects, each of which is only
379
valid until the iterator is advanced.
381
versions = [version[-1] for version in versions]
382
if ordering == 'topological':
383
parents = self.get_parent_map(versions)
384
new_versions = tsort.topo_sort(parents)
385
new_versions.extend(set(versions).difference(set(parents)))
386
versions = new_versions
387
elif ordering == 'groupcompress':
388
parents = self.get_parent_map(versions)
389
new_versions = sort_groupcompress(parents)
390
new_versions.extend(set(versions).difference(set(parents)))
391
versions = new_versions
392
for version in versions:
394
yield WeaveContentFactory(version, self)
396
yield AbsentContentFactory((version,))
398
def get_parent_map(self, version_ids):
399
"""See VersionedFile.get_parent_map."""
401
for version_id in version_ids:
402
if version_id == NULL_REVISION:
407
map(self._idx_to_name,
408
self._parents[self._lookup(version_id)]))
409
except RevisionNotPresent:
411
result[version_id] = parents
414
def get_parents_with_ghosts(self, version_id):
415
raise NotImplementedError(self.get_parents_with_ghosts)
417
def insert_record_stream(self, stream):
418
"""Insert a record stream into this versioned file.
420
:param stream: A stream of records to insert.
422
:seealso VersionedFile.get_record_stream:
425
for record in stream:
426
# Raise an error when a record is missing.
427
if record.storage_kind == 'absent':
428
raise RevisionNotPresent([record.key[0]], self)
429
# adapt to non-tuple interface
430
parents = [parent[0] for parent in record.parents]
431
if record.storage_kind in ('fulltext', 'chunked', 'lines'):
433
record.key[0], parents,
434
record.get_bytes_as('lines'))
436
adapter_key = record.storage_kind, 'lines'
438
adapter = adapters[adapter_key]
440
adapter_factory = adapter_registry.get(adapter_key)
441
adapter = adapter_factory(self)
442
adapters[adapter_key] = adapter
443
lines = adapter.get_bytes(record, 'lines')
445
self.add_lines(record.key[0], parents, lines)
446
except RevisionAlreadyPresent:
225
def iter_names(self):
226
"""Yield a list of all names in this weave."""
227
return iter(self._names)
229
def idx_to_name(self, version):
230
return self._names[version]
449
232
def _check_repeated_add(self, name, parents, text, sha1):
450
233
"""Check that a duplicated add is OK.
452
235
If it is, return the (old) index; otherwise raise an exception.
454
idx = self._lookup(name)
237
idx = self.lookup(name)
455
238
if sorted(self._parents[idx]) != sorted(parents) \
456
or sha1 != self._sha1s[idx]:
457
raise RevisionAlreadyPresent(name, self._weave_name)
239
or sha1 != self._sha1s[idx]:
240
raise WeaveRevisionAlreadyPresent(name, self)
460
def _add_lines(self, version_id, parents, lines, parent_texts,
461
left_matching_blocks, nostore_sha, random_id,
463
"""See VersionedFile.add_lines."""
464
idx = self._add(version_id, lines, list(map(self._lookup, parents)),
465
nostore_sha=nostore_sha)
466
return sha_strings(lines), sum(map(len, lines)), idx
468
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
243
def add(self, name, parents, text, sha1=None):
469
244
"""Add a single text on top of the weave.
471
246
Returns the index number of the newly added version.
474
249
Symbolic name for this version.
475
250
(Typically the revision-id of the revision that added it.)
476
If None, a name will be allocated based on the hash. (sha1:SHAHASH)
479
253
List or set of direct parent version numbers.
482
256
Sequence of lines to be added in the new version.
484
:param nostore_sha: See VersionedFile.add_lines.
258
sha -- SHA-1 of the file, if known. This is trusted to be
486
self._check_lines_not_unicode(lines)
487
self._check_lines_are_lines(lines)
489
sha1 = sha_strings(lines)
490
if sha1 == nostore_sha:
491
raise errors.ExistingContent
492
if version_id is None:
493
version_id = b"sha1:" + sha1
494
if version_id in self._name_map:
495
return self._check_repeated_add(version_id, parents, lines, sha1)
261
from bzrlib.osutils import sha_strings
263
assert isinstance(name, basestring)
265
sha1 = sha_strings(text)
266
if name in self._name_map:
267
return self._check_repeated_add(name, parents, text, sha1)
269
parents = map(self.maybe_lookup, parents)
497
270
self._check_versions(parents)
271
## self._check_lines(text)
498
272
new_version = len(self._parents)
500
# if we abort after here the (in-memory) weave will be corrupt because
501
# only some fields are updated
502
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
503
# - Robert Collins 20060226
275
# if we abort after here the (in-memory) weave will be corrupt because only
276
# some fields are updated
504
277
self._parents.append(parents[:])
505
278
self._sha1s.append(sha1)
506
self._names.append(version_id)
507
self._name_map[version_id] = new_version
279
self._names.append(name)
280
self._name_map[name] = new_version
510
284
# special case; adding with no parents revision; can do
511
285
# this more quickly by just appending unconditionally.
512
286
# even more specially, if we're adding an empty text we
513
287
# need do nothing at all.
515
self._weave.append((b'{', new_version))
516
self._weave.extend(lines)
517
self._weave.append((b'}', None))
289
self._weave.append(('{', new_version))
290
self._weave.extend(text)
291
self._weave.append(('}', None))
518
293
return new_version
520
295
if len(parents) == 1:
608
433
except IndexError:
609
434
raise IndexError("invalid version number %r" % i)
611
def _compatible_parents(self, my_parents, other_parents):
612
"""During join check that other_parents are joinable with my_parents.
614
Joinable is defined as 'is a subset of' - supersets may require
615
regeneration of diffs, but subsets do not.
617
return len(other_parents.difference(my_parents)) == 0
619
def annotate(self, version_id):
620
"""Return a list of (version-id, line) tuples for version_id.
437
def annotate(self, name_or_index):
438
return list(self.annotate_iter(name_or_index))
441
def annotate_iter(self, name_or_index):
442
"""Yield list of (index-id, line) pairs for the specified version.
622
444
The index indicates when the line originated in the weave."""
623
incls = [self._lookup(version_id)]
624
return [(self._idx_to_name(origin), text) for origin, lineno, text in
625
self._extract(incls)]
627
def iter_lines_added_or_present_in_versions(self, version_ids=None,
629
"""See VersionedFile.iter_lines_added_or_present_in_versions()."""
630
if version_ids is None:
631
version_ids = self.versions()
632
version_ids = set(version_ids)
633
for lineno, inserted, deletes, line in self._walk_internal(
635
if inserted not in version_ids:
637
if not line.endswith(b'\n'):
638
yield line + b'\n', inserted
642
def _walk_internal(self, version_ids=None):
643
"""Helper method for weave actions."""
445
incls = [self.maybe_lookup(name_or_index)]
446
for origin, lineno, text in self._extract(incls):
453
(lineno, insert, deletes, text)
454
for each literal line.
648
460
lineno = 0 # line of weave, 0-based
650
462
for l in self._weave:
651
if l.__class__ == tuple:
463
if isinstance(l, tuple):
654
istack.append(self._names[v])
658
dset.add(self._names[v])
660
dset.remove(self._names[v])
662
476
raise WeaveFormatError('unexpected instruction %r' % v)
664
yield lineno, istack[-1], frozenset(dset), l
478
assert isinstance(l, basestring)
480
yield lineno, istack[-1], dset, l
485
def _extract(self, versions):
486
"""Yield annotation of lines in included set.
488
Yields a sequence of tuples (origin, lineno, text), where
489
origin is the origin version, lineno the index in the weave,
490
and text the text of the line.
492
The set typically but not necessarily corresponds to a version.
495
if not isinstance(i, int):
498
included = self.inclusions(versions)
503
lineno = 0 # line of weave, 0-based
509
WFE = WeaveFormatError
511
for l in self._weave:
512
if isinstance(l, tuple):
516
assert v not in istack
530
assert isinstance(l, basestring)
532
isactive = (not dset) and istack and (istack[-1] in included)
534
result.append((istack[-1], lineno, l))
668
537
raise WeaveFormatError("unclosed insertion blocks "
669
"at end of weave: %s" % istack)
538
"at end of weave: %s" % istack)
671
raise WeaveFormatError(
672
"unclosed deletion blocks at end of weave: %s" % dset)
540
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
545
def get_iter(self, name_or_index):
546
"""Yield lines for the specified version."""
547
incls = [self.maybe_lookup(name_or_index)]
548
for origin, lineno, line in self._extract(incls):
552
def get_text(self, name_or_index):
553
return ''.join(self.get_iter(name_or_index))
554
assert isinstance(version, int)
557
def get_lines(self, name_or_index):
558
return list(self.get_iter(name_or_index))
564
def mash_iter(self, included):
565
"""Return composed version of multiple included versions."""
566
included = map(self.maybe_lookup, included)
567
for origin, lineno, text in self._extract(included):
571
def dump(self, to_file):
572
from pprint import pprint
573
print >>to_file, "Weave._weave = ",
574
pprint(self._weave, to_file)
575
print >>to_file, "Weave._parents = ",
576
pprint(self._parents, to_file)
580
def numversions(self):
581
l = len(self._parents)
582
assert l == len(self._sha1s)
587
return self.numversions()
590
def check(self, progress_bar=None):
591
# check no circular inclusions
592
for version in range(self.numversions()):
593
inclusions = list(self._parents[version])
596
if inclusions[-1] >= version:
597
raise WeaveFormatError("invalid included version %d for index %d"
598
% (inclusions[-1], version))
600
# try extracting all versions; this is a bit slow and parallel
601
# extraction could be used
602
nv = self.numversions()
603
for version in range(nv):
605
progress_bar.update('checking text', version, nv)
607
for l in self.get_iter(version):
610
expected = self._sha1s[version]
612
raise WeaveError("mismatched sha1 for version %d; "
613
"got %s, expected %s"
614
% (version, hd, expected))
616
# TODO: check insertions are properly nested, that there are
617
# no lines outside of insertion blocks, that deletions are
618
# properly paired, etc.
622
def merge(self, merge_versions):
623
"""Automerge and mark conflicts between versions.
625
This returns a sequence, each entry describing alternatives
626
for a chunk of the file. Each of the alternatives is given as
629
If there is a chunk of the file where there's no diagreement,
630
only one alternative is given.
632
# approach: find the included versions common to all the
634
raise NotImplementedError()
638
def _delta(self, included, lines):
639
"""Return changes from basis to new revision.
641
The old text for comparison is the union of included revisions.
643
This is used in inserting a new text.
645
Delta is returned as a sequence of
646
(weave1, weave2, newlines).
648
This indicates that weave1:weave2 of the old weave should be
649
replaced by the sequence of lines in newlines. Note that
650
these line numbers are positions in the total weave and don't
651
correspond to the lines in any extracted version, or even the
652
extracted union of included versions.
654
If line1=line2, this is a pure insert; if newlines=[] this is a
655
pure delete. (Similar to difflib.)
657
raise NotImplementedError()
674
660
def plan_merge(self, ver_a, ver_b):
675
661
"""Return pseudo-annotation indicating how the two versions merge.
716
701
# not in either revision
717
702
yield 'irrelevant', line
719
def _extract(self, versions):
720
"""Yield annotation of lines in included set.
722
Yields a sequence of tuples (origin, lineno, text), where
723
origin is the origin version, lineno the index in the weave,
724
and text the text of the line.
726
The set typically but not necessarily corresponds to a version.
729
if not isinstance(i, int):
732
included = self._inclusions(versions)
738
lineno = 0 # line of weave, 0-based
745
# 449 0 4474.6820 2356.5590 breezy.weave:556(_extract)
746
# +285282 0 1676.8040 1676.8040 +<isinstance>
747
# 1.6 seconds in 'isinstance'.
748
# changing the first isinstance:
749
# 449 0 2814.2660 1577.1760 breezy.weave:556(_extract)
750
# +140414 0 762.8050 762.8050 +<isinstance>
751
# note that the inline time actually dropped (less function calls)
752
# and total processing time was halved.
753
# we're still spending ~1/4 of the method in isinstance though.
754
# so lets hard code the acceptable string classes we expect:
755
# 449 0 1202.9420 786.2930 breezy.weave:556(_extract)
756
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
758
# yay, down to ~1/4 the initial extract time, and our inline time
759
# has shrunk again, with isinstance no longer dominating.
760
# tweaking the stack inclusion test to use a set gives:
761
# 449 0 1122.8030 713.0080 breezy.weave:556(_extract)
762
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
764
# - a 5% win, or possibly just noise. However with large istacks that
765
# 'in' test could dominate, so I'm leaving this change in place - when
766
# its fast enough to consider profiling big datasets we can review.
768
for l in self._weave:
769
if l.__class__ == tuple:
776
iset.remove(istack.pop())
704
yield 'unchanged', '' # terminator
708
def weave_merge(self, plan):
713
for state, line in plan:
714
if state == 'unchanged' or state == 'killed-both':
715
# resync and flush queued conflicts changes if any
716
if not lines_a and not lines_b:
718
elif ch_a and not ch_b:
720
for l in lines_a: yield l
721
elif ch_b and not ch_a:
722
for l in lines_b: yield l
723
elif lines_a == lines_b:
724
for l in lines_a: yield l
784
raise AssertionError()
727
for l in lines_a: yield l
729
for l in lines_b: yield l
736
if state == 'unchanged':
739
elif state == 'killed-a':
742
elif state == 'killed-b':
745
elif state == 'new-a':
748
elif state == 'new-b':
787
isactive = (not dset) and istack and (
788
istack[-1] in included)
790
result.append((istack[-1], lineno, l))
793
raise WeaveFormatError("unclosed insertion blocks "
794
"at end of weave: %s" % istack)
796
raise WeaveFormatError(
797
"unclosed deletion blocks at end of weave: %s" % dset)
800
def _maybe_lookup(self, name_or_index):
801
"""Convert possible symbolic name to index, or pass through indexes.
752
assert state in ('irrelevant', 'ghost-a', 'ghost-b', 'killed-base',
757
def join(self, other):
758
"""Integrate versions from other into this weave.
760
The resulting weave contains all the history of both weaves;
761
any version you could retrieve from either self or other can be
762
retrieved from self after this call.
764
It is illegal for the two weaves to contain different values
765
or different parents for any version. See also reweave().
805
# GZ 2017-04-01: This used to check for long as well, but I don't think
806
# there are python implementations with sys.maxsize > sys.maxint
807
if isinstance(name_or_index, int):
810
return self._lookup(name_or_index)
812
def get_lines(self, version_id):
813
"""See VersionedFile.get_lines()."""
814
int_index = self._maybe_lookup(version_id)
815
result = [line for (origin, lineno, line)
816
in self._extract([int_index])]
817
expected_sha1 = self._sha1s[int_index]
818
measured_sha1 = sha_strings(result)
819
if measured_sha1 != expected_sha1:
820
raise WeaveInvalidChecksum(
821
'file %s, revision %s, expected: %s, measured %s'
822
% (self._weave_name, version_id,
823
expected_sha1, measured_sha1))
826
def get_sha1s(self, version_ids):
827
"""See VersionedFile.get_sha1s()."""
829
for v in version_ids:
830
result[v] = self._sha1s[self._lookup(v)]
833
def num_versions(self):
834
"""How many versions are in this weave?"""
835
return len(self._parents)
837
__len__ = num_versions
839
def check(self, progress_bar=None):
840
# TODO evaluate performance hit of using string sets in this routine.
841
# TODO: check no circular inclusions
842
# TODO: create a nested progress bar
843
for version in range(self.num_versions()):
844
inclusions = list(self._parents[version])
847
if inclusions[-1] >= version:
848
raise WeaveFormatError(
849
"invalid included version %d for index %d"
850
% (inclusions[-1], version))
852
# try extracting all versions; parallel extraction is used
853
nv = self.num_versions()
858
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
859
# The problem is that set membership is much more expensive
860
name = self._idx_to_name(i)
864
for p in self._parents[i]:
865
new_inc.update(inclusions[self._idx_to_name(p)])
867
if set(new_inc) != set(self.get_ancestry(name)):
868
raise AssertionError(
870
% (set(new_inc), set(self.get_ancestry(name))))
871
inclusions[name] = new_inc
873
nlines = len(self._weave)
875
update_text = 'checking weave'
877
short_name = os.path.basename(self._weave_name)
878
update_text = 'checking %s' % (short_name,)
879
update_text = update_text[:25]
881
for lineno, insert, deleteset, line in self._walk_internal():
883
progress_bar.update(update_text, lineno, nlines)
885
for name, name_inclusions in inclusions.items():
886
# The active inclusion must be an ancestor,
887
# and no ancestors must have deleted this line,
888
# because we don't support resurrection.
889
if ((insert in name_inclusions) and
890
not (deleteset & name_inclusions)):
891
sha1s[name].update(line)
894
version = self._idx_to_name(i)
895
hd = sha1s[version].hexdigest().encode()
896
expected = self._sha1s[i]
898
raise WeaveInvalidChecksum(
899
"mismatched sha1 for version %s: "
900
"got %s, expected %s"
901
% (version, hd, expected))
903
# TODO: check insertions are properly nested, that there are
904
# no lines outside of insertion blocks, that deletions are
905
# properly paired, etc.
767
if other.numversions() == 0:
768
return # nothing to update, easy
769
# two loops so that we do not change ourselves before verifying it
771
# work through in index order to make sure we get all dependencies
772
for other_idx, name in enumerate(other._names):
773
if self._check_version_consistent(other, other_idx, name):
775
for other_idx, name in enumerate(other._names):
776
# TODO: If all the parents of the other version are already
777
# present then we can avoid some work by just taking the delta
778
# and adjusting the offsets.
779
new_parents = self._imported_parents(other, other_idx)
780
lines = other.get_lines(other_idx)
781
sha1 = other._sha1s[other_idx]
782
self.add(name, new_parents, lines, sha1)
907
785
def _imported_parents(self, other, other_idx):
908
786
"""Return list of parents in self corresponding to indexes in other."""
910
788
for parent_idx in other._parents[other_idx]:
911
789
parent_name = other._names[parent_idx]
912
if parent_name not in self._name_map:
790
if parent_name not in self._names:
913
791
# should not be possible
914
raise WeaveError("missing parent {%s} of {%s} in %r"
792
raise WeaveError("missing parent {%s} of {%s} in %r"
915
793
% (parent_name, other._name_map[other_idx], self))
916
794
new_parents.append(self._name_map[parent_name])
917
795
return new_parents
925
803
* the same direct parents (by name, not index, and disregarding
928
806
If present & correct return True;
929
if not present in self return False;
807
if not present in self return False;
930
808
if inconsistent raise error."""
931
809
this_idx = self._name_map.get(name, -1)
932
810
if this_idx != -1:
933
811
if self._sha1s[this_idx] != other._sha1s[other_idx]:
934
raise WeaveTextDiffers(name, self, other)
812
raise WeaveError("inconsistent texts for version {%s} "
813
"when joining weaves"
935
815
self_parents = self._parents[this_idx]
936
816
other_parents = other._parents[other_idx]
937
n1 = {self._names[i] for i in self_parents}
938
n2 = {other._names[i] for i in other_parents}
939
if not self._compatible_parents(n1, n2):
940
raise WeaveParentMismatch(
941
"inconsistent parents "
817
n1 = [self._names[i] for i in self_parents]
818
n2 = [other._names[i] for i in other_parents]
822
raise WeaveParentMismatch("inconsistent parents "
942
823
"for version {%s}: %s vs %s" % (name, n1, n2))
944
825
return True # ok!
948
def _reweave(self, other, pb, msg):
949
"""Reweave self with other - internal helper for join().
951
:param other: The other weave to merge
952
:param pb: An optional progress bar, indicating how far done we are
953
:param msg: An optional message for the progress
955
new_weave = _reweave(self, other, pb=pb, msg=msg)
956
self._copy_weave_content(new_weave)
958
def _copy_weave_content(self, otherweave):
959
"""adsorb the content from otherweave."""
829
def reweave(self, other):
830
"""Reweave self with other."""
831
new_weave = reweave(self, other)
960
832
for attr in self.__slots__:
961
if attr != '_weave_name':
962
setattr(self, attr, copy(getattr(otherweave, attr)))
965
class WeaveFile(Weave):
966
"""A WeaveFile represents a Weave on disk and writes on change."""
968
WEAVE_SUFFIX = '.weave'
970
def __init__(self, name, transport, filemode=None, create=False,
971
access_mode='w', get_scope=None):
972
"""Create a WeaveFile.
974
:param create: If not True, only open an existing knit.
976
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope,
977
allow_reserved=False)
978
self._transport = transport
979
self._filemode = filemode
981
f = self._transport.get(name + WeaveFile.WEAVE_SUFFIX)
982
_read_weave_v5(BytesIO(f.read()), self)
983
except errors.NoSuchFile:
989
def _add_lines(self, version_id, parents, lines, parent_texts,
990
left_matching_blocks, nostore_sha, random_id,
992
"""Add a version and save the weave."""
993
self.check_not_reserved_id(version_id)
994
result = super(WeaveFile, self)._add_lines(
995
version_id, parents, lines, parent_texts, left_matching_blocks,
996
nostore_sha, random_id, check_content)
1000
def copy_to(self, name, transport):
1001
"""See VersionedFile.copy_to()."""
1002
# as we are all in memory always, just serialise to the new place.
1004
write_weave_v5(self, sio)
1006
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
1009
"""Save the weave."""
1010
self._check_write_ok()
1012
write_weave_v5(self, sio)
1014
bytes = sio.getvalue()
1015
path = self._weave_name + WeaveFile.WEAVE_SUFFIX
1017
self._transport.put_bytes(path, bytes, self._filemode)
1018
except errors.NoSuchFile:
1019
self._transport.mkdir(dirname(path))
1020
self._transport.put_bytes(path, bytes, self._filemode)
1024
"""See VersionedFile.get_suffixes()."""
1025
return [WeaveFile.WEAVE_SUFFIX]
1027
def insert_record_stream(self, stream):
1028
super(WeaveFile, self).insert_record_stream(stream)
1032
def _reweave(wa, wb, pb=None, msg=None):
833
setattr(self, attr, getattr(new_weave, attr))
1033
837
"""Combine two weaves and return the result.
1035
This works even if a revision R has different parents in
839
This works even if a revision R has different parents in
1036
840
wa and wb. In the resulting weave all the parents are given.
1038
This is done by just building up a new weave, maintaining ordering
842
This is done by just building up a new weave, maintaining ordering
1039
843
of the versions in the two inputs. More efficient approaches
1040
might be possible but it should only be necessary to do
1041
this operation rarely, when a new previously ghost version is
844
might be possible but it should only be necessary to do
845
this operation rarely, when a new previously ghost version is
1044
:param pb: An optional progress bar, indicating how far done we are
1045
:param msg: An optional message for the progress
850
queue_a = range(wa.numversions())
851
queue_b = range(wb.numversions())
1048
852
# first determine combined parents of all versions
1049
853
# map from version name -> all parent names
1050
854
combined_parents = _reweave_parent_graphs(wa, wb)
1051
855
mutter("combined parents: %r", combined_parents)
1052
order = tsort.topo_sort(combined_parents.items())
856
order = topo_sort(combined_parents.iteritems())
1053
857
mutter("order to reweave: %r", order)
1058
for idx, name in enumerate(order):
1060
pb.update(msg, idx, len(order))
1061
859
if name in wa._name_map:
1062
860
lines = wa.get_lines(name)
1063
861
if name in wb._name_map:
1064
lines_b = wb.get_lines(name)
1065
if lines != lines_b:
1066
mutter('Weaves differ on content. rev_id {%s}', name)
1067
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
1069
lines = list(difflib.unified_diff(lines, lines_b,
1070
wa._weave_name, wb._weave_name))
1071
mutter('lines:\n%s', ''.join(lines))
1072
raise WeaveTextDiffers(name, wa, wb)
862
assert lines == wb.get_lines(name)
1074
864
lines = wb.get_lines(name)
1075
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
865
wr.add(name, combined_parents[name], lines)
1079
869
def _reweave_parent_graphs(wa, wb):
1080
870
"""Return combined parent ancestry for two weaves.
1082
872
Returned as a list of (version_name, set(parent_names))"""
1084
874
for weave in [wa, wb]:
1085
875
for idx, name in enumerate(weave._names):
1086
876
p = combined.setdefault(name, set())
1087
p.update(map(weave._idx_to_name, weave._parents[idx]))
877
p.update(map(weave.idx_to_name, weave._parents[idx]))
882
"""Show the weave's table-of-contents"""
883
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
884
for i in (6, 50, 10, 10):
887
for i in range(w.numversions()):
890
parent_str = ' '.join(map(str, w._parents[i]))
891
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
895
def weave_stats(weave_file, pb):
896
from bzrlib.weavefile import read_weave
898
wf = file(weave_file, 'rb')
900
# FIXME: doesn't work on pipes
901
weave_size = wf.tell()
905
for i in range(vers):
906
pb.update('checking sizes', i, vers)
907
for origin, lineno, line in w._extract([i]):
912
print 'versions %9d' % vers
913
print 'weave file %9d bytes' % weave_size
914
print 'total contents %9d bytes' % total
915
print 'compression ratio %9.2fx' % (float(total) / float(weave_size))
918
print 'average size %9d bytes' % avg
919
print 'relative size %9.2fx' % (float(weave_size) / float(avg))
923
print """bzr weave tool
925
Experimental tool for weave algorithm.
929
Create an empty weave file
930
weave get WEAVEFILE VERSION
931
Write out specified version.
932
weave check WEAVEFILE
933
Check consistency of all versions.
935
Display table of contents.
936
weave add WEAVEFILE NAME [BASE...] < NEWTEXT
937
Add NEWTEXT, with specified parent versions.
938
weave annotate WEAVEFILE VERSION
939
Display origin of each line.
940
weave mash WEAVEFILE VERSION...
941
Display composite of all selected versions.
942
weave merge WEAVEFILE VERSION1 VERSION2 > OUT
943
Auto-merge two versions and display conflicts.
944
weave diff WEAVEFILE VERSION1 VERSION2
945
Show differences between two versions.
949
% weave init foo.weave
951
% weave add foo.weave ver0 < foo.txt
954
(create updated version)
956
% weave get foo.weave 0 | diff -u - foo.txt
957
% weave add foo.weave ver1 0 < foo.txt
960
% weave get foo.weave 0 > foo.txt (create forked version)
962
% weave add foo.weave ver2 0 < foo.txt
965
% weave merge foo.weave 1 2 > foo.txt (merge them)
966
% vi foo.txt (resolve conflicts)
967
% weave add foo.weave merged 1 2 < foo.txt (commit merged version)
979
# in case we're run directly from the subdirectory
980
sys.path.append('..')
982
from bzrlib.weavefile import write_weave, read_weave
983
from bzrlib.progress import ProgressBar
998
return read_weave(file(argv[2], 'rb'))
1004
# at the moment, based on everything in the file
1006
parents = map(int, argv[4:])
1007
lines = sys.stdin.readlines()
1008
ver = w.add(name, parents, lines)
1009
write_weave(w, file(argv[2], 'wb'))
1010
print 'added version %r %d' % (name, ver)
1013
if os.path.exists(fn):
1014
raise IOError("file exists")
1016
write_weave(w, file(fn, 'wb'))
1017
elif cmd == 'get': # get one version
1019
sys.stdout.writelines(w.get_iter(int(argv[3])))
1021
elif cmd == 'mash': # get composite
1023
sys.stdout.writelines(w.mash_iter(map(int, argv[3:])))
1026
from difflib import unified_diff
1029
v1, v2 = map(int, argv[3:5])
1032
diff_gen = unified_diff(lines1, lines2,
1033
'%s version %d' % (fn, v1),
1034
'%s version %d' % (fn, v2))
1035
sys.stdout.writelines(diff_gen)
1037
elif cmd == 'annotate':
1039
# newline is added to all lines regardless; too hard to get
1040
# reasonable formatting otherwise
1042
for origin, text in w.annotate(int(argv[3])):
1043
text = text.rstrip('\r\n')
1045
print ' | %s' % (text)
1047
print '%5d | %s' % (origin, text)
1053
elif cmd == 'stats':
1054
weave_stats(argv[2], ProgressBar())
1056
elif cmd == 'check':
1061
print '%d versions ok' % w.numversions()
1063
elif cmd == 'inclusions':
1065
print ' '.join(map(str, w.inclusions([int(argv[3])])))
1067
elif cmd == 'parents':
1069
print ' '.join(map(str, w._parents[int(argv[3])]))
1071
elif cmd == 'plan-merge':
1073
for state, line in w.plan_merge(int(argv[3]), int(argv[4])):
1075
print '%14s | %s' % (state, line),
1077
elif cmd == 'merge':
1079
p = w.plan_merge(int(argv[3]), int(argv[4]))
1080
sys.stdout.writelines(w.weave_merge(p))
1082
elif cmd == 'mash-merge':
1088
v1, v2 = map(int, argv[3:5])
1090
basis = w.inclusions([v1]).intersection(w.inclusions([v2]))
1092
base_lines = list(w.mash_iter(basis))
1093
a_lines = list(w.get(v1))
1094
b_lines = list(w.get(v2))
1096
from bzrlib.merge3 import Merge3
1097
m3 = Merge3(base_lines, a_lines, b_lines)
1099
name_a = 'version %d' % v1
1100
name_b = 'version %d' % v2
1101
sys.stdout.writelines(m3.merge_lines(name_a=name_a, name_b=name_b))
1103
raise ValueError('unknown command %r' % cmd)
1107
def profile_main(argv):
1108
import tempfile, hotshot, hotshot.stats
1110
prof_f = tempfile.NamedTemporaryFile()
1112
prof = hotshot.Profile(prof_f.name)
1114
ret = prof.runcall(main, argv)
1117
stats = hotshot.stats.load(prof_f.name)
1119
stats.sort_stats('cumulative')
1120
## XXX: Might like to write to stderr or the trace file instead but
1121
## print_stats seems hardcoded to stdout
1122
stats.print_stats(20)
1127
if __name__ == '__main__':
1129
if '--profile' in sys.argv:
1131
args.remove('--profile')
1132
sys.exit(profile_main(args))
1134
sys.exit(main(sys.argv))