61
61
# where the basis and destination are unchanged.
63
63
# FIXME: Sometimes we will be given a parents list for a revision
64
# that includes some redundant parents (i.e. already a parent of
65
# 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
66
66
# be done fairly efficiently because the sequence numbers constrain
67
67
# the possible relationships.
69
# FIXME: the conflict markers should be *7* characters
72
from cStringIO import StringIO
77
from bzrlib.lazy_import import lazy_import
78
lazy_import(globals(), """
79
from bzrlib import tsort
86
from bzrlib.errors import (WeaveError, WeaveFormatError, WeaveParentMismatch,
87
RevisionAlreadyPresent,
89
UnavailableRepresentation,
90
WeaveRevisionAlreadyPresent,
91
WeaveRevisionNotPresent,
93
from bzrlib.osutils import dirname, sha, sha_strings, split_lines
94
import bzrlib.patiencediff
95
from bzrlib.revision import NULL_REVISION
96
from bzrlib.symbol_versioning import *
71
from difflib import SequenceMatcher
97
73
from bzrlib.trace import mutter
98
from bzrlib.versionedfile import (
105
from bzrlib.weavefile import _read_weave_v5, write_weave_v5
108
class WeaveContentFactory(ContentFactory):
109
"""Content factory for streaming from weaves.
111
:seealso ContentFactory:
114
def __init__(self, version, weave):
115
"""Create a WeaveContentFactory for version from weave."""
116
ContentFactory.__init__(self)
117
self.sha1 = weave.get_sha1s([version])[version]
118
self.key = (version,)
119
parents = weave.get_parent_map([version])[version]
120
self.parents = tuple((parent,) for parent in parents)
121
self.storage_kind = 'fulltext'
124
def get_bytes_as(self, storage_kind):
125
if storage_kind == 'fulltext':
126
return self._weave.get_text(self.key[-1])
127
elif storage_kind == 'chunked':
128
return self._weave.get_lines(self.key[-1])
130
raise UnavailableRepresentation(self.key, storage_kind, 'fulltext')
133
class Weave(VersionedFile):
74
from bzrlib.errors import WeaveError, WeaveFormatError, WeaveParentMismatch, \
75
WeaveRevisionNotPresent, WeaveRevisionAlreadyPresent
76
from bzrlib.tsort import topo_sort
134
80
"""weave - versioned text file storage.
136
82
A Weave manages versions of line-based text files, keeping track
137
83
of the originating version for each line.
222
168
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
223
'_weave_name', '_matcher', '_allow_reserved']
225
def __init__(self, weave_name=None, access_mode='w', matcher=None,
226
get_scope=None, allow_reserved=False):
229
:param get_scope: A callable that returns an opaque object to be used
230
for detecting when this weave goes out of scope (should stop
231
answering requests or allowing mutation).
233
super(Weave, self).__init__()
171
def __init__(self, weave_name=None):
235
173
self._parents = []
238
176
self._name_map = {}
239
177
self._weave_name = weave_name
241
self._matcher = bzrlib.patiencediff.PatienceSequenceMatcher
243
self._matcher = matcher
244
if get_scope is None:
245
get_scope = lambda:None
246
self._get_scope = get_scope
247
self._scope = get_scope()
248
self._access_mode = access_mode
249
self._allow_reserved = allow_reserved
251
179
def __repr__(self):
252
180
return "Weave(%r)" % self._weave_name
254
def _check_write_ok(self):
255
"""Is the versioned file marked as 'finished' ? Raise if it is."""
256
if self._get_scope() != self._scope:
257
raise errors.OutSideTransaction()
258
if self._access_mode != 'w':
259
raise errors.ReadOnlyObjectDirtiedError(self)
262
184
"""Return a deep copy of self.
264
186
The copy can be modified without affecting the original weave."""
266
188
other._weave = self._weave[:]
277
199
return self._parents == other._parents \
278
200
and self._weave == other._weave \
279
and self._sha1s == other._sha1s
201
and self._sha1s == other._sha1s
281
204
def __ne__(self, other):
282
205
return not self.__eq__(other)
284
def _idx_to_name(self, version):
285
return self._names[version]
287
def _lookup(self, name):
207
def __contains__(self, name):
208
return self._name_map.has_key(name)
210
def maybe_lookup(self, name_or_index):
211
"""Convert possible symbolic name to index, or pass through indexes."""
212
if isinstance(name_or_index, (int, long)):
215
return self.lookup(name_or_index)
218
def lookup(self, name):
288
219
"""Convert symbolic version name to index."""
289
if not self._allow_reserved:
290
self.check_not_reserved_id(name)
292
221
return self._name_map[name]
294
raise RevisionNotPresent(name, self._weave_name)
223
raise WeaveRevisionNotPresent(name, self)
297
"""See VersionedFile.versions."""
298
226
return self._names[:]
300
def has_version(self, version_id):
301
"""See VersionedFile.has_version."""
302
return (version_id in self._name_map)
304
__contains__ = has_version
306
def get_record_stream(self, versions, ordering, include_delta_closure):
307
"""Get a stream of records for versions.
309
:param versions: The versions to include. Each version is a tuple
311
:param ordering: Either 'unordered' or 'topological'. A topologically
312
sorted stream has compression parents strictly before their
314
:param include_delta_closure: If True then the closure across any
315
compression parents will be included (in the opaque data).
316
:return: An iterator of ContentFactory objects, each of which is only
317
valid until the iterator is advanced.
319
versions = [version[-1] for version in versions]
320
if ordering == 'topological':
321
parents = self.get_parent_map(versions)
322
new_versions = tsort.topo_sort(parents)
323
new_versions.extend(set(versions).difference(set(parents)))
324
versions = new_versions
325
elif ordering == 'groupcompress':
326
parents = self.get_parent_map(versions)
327
new_versions = sort_groupcompress(parents)
328
new_versions.extend(set(versions).difference(set(parents)))
329
versions = new_versions
330
for version in versions:
332
yield WeaveContentFactory(version, self)
334
yield AbsentContentFactory((version,))
336
def get_parent_map(self, version_ids):
337
"""See VersionedFile.get_parent_map."""
339
for version_id in version_ids:
340
if version_id == NULL_REVISION:
345
map(self._idx_to_name,
346
self._parents[self._lookup(version_id)]))
347
except RevisionNotPresent:
349
result[version_id] = parents
352
def get_parents_with_ghosts(self, version_id):
353
raise NotImplementedError(self.get_parents_with_ghosts)
355
def insert_record_stream(self, stream):
356
"""Insert a record stream into this versioned file.
358
:param stream: A stream of records to insert.
360
:seealso VersionedFile.get_record_stream:
363
for record in stream:
364
# Raise an error when a record is missing.
365
if record.storage_kind == 'absent':
366
raise RevisionNotPresent([record.key[0]], self)
367
# adapt to non-tuple interface
368
parents = [parent[0] for parent in record.parents]
369
if (record.storage_kind == 'fulltext'
370
or record.storage_kind == 'chunked'):
371
self.add_lines(record.key[0], parents,
372
osutils.chunks_to_lines(record.get_bytes_as('chunked')))
374
adapter_key = record.storage_kind, 'fulltext'
376
adapter = adapters[adapter_key]
378
adapter_factory = adapter_registry.get(adapter_key)
379
adapter = adapter_factory(self)
380
adapters[adapter_key] = adapter
381
lines = split_lines(adapter.get_bytes(record))
383
self.add_lines(record.key[0], parents, lines)
384
except RevisionAlreadyPresent:
228
def iter_names(self):
229
"""Yield a list of all names in this weave."""
230
return iter(self._names)
232
def idx_to_name(self, version):
233
return self._names[version]
387
235
def _check_repeated_add(self, name, parents, text, sha1):
388
236
"""Check that a duplicated add is OK.
390
238
If it is, return the (old) index; otherwise raise an exception.
392
idx = self._lookup(name)
240
idx = self.lookup(name)
393
241
if sorted(self._parents[idx]) != sorted(parents) \
394
242
or sha1 != self._sha1s[idx]:
395
raise RevisionAlreadyPresent(name, self._weave_name)
243
raise WeaveRevisionAlreadyPresent(name, self)
398
def _add_lines(self, version_id, parents, lines, parent_texts,
399
left_matching_blocks, nostore_sha, random_id, check_content):
400
"""See VersionedFile.add_lines."""
401
idx = self._add(version_id, lines, map(self._lookup, parents),
402
nostore_sha=nostore_sha)
403
return sha_strings(lines), sum(map(len, lines)), idx
405
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
246
def add(self, name, parents, text, sha1=None):
406
247
"""Add a single text on top of the weave.
408
249
Returns the index number of the newly added version.
411
252
Symbolic name for this version.
412
253
(Typically the revision-id of the revision that added it.)
413
If None, a name will be allocated based on the hash. (sha1:SHAHASH)
416
256
List or set of direct parent version numbers.
419
259
Sequence of lines to be added in the new version.
421
:param nostore_sha: See VersionedFile.add_lines.
261
sha -- SHA-1 of the file, if known. This is trusted to be
423
self._check_lines_not_unicode(lines)
424
self._check_lines_are_lines(lines)
426
sha1 = sha_strings(lines)
427
if sha1 == nostore_sha:
428
raise errors.ExistingContent
429
if version_id is None:
430
version_id = "sha1:" + sha1
431
if version_id in self._name_map:
432
return self._check_repeated_add(version_id, parents, lines, sha1)
264
from bzrlib.osutils import sha_strings
266
assert isinstance(name, basestring)
268
sha1 = sha_strings(text)
269
if name in self._name_map:
270
return self._check_repeated_add(name, parents, text, sha1)
272
parents = map(self.maybe_lookup, parents)
434
273
self._check_versions(parents)
435
## self._check_lines(lines)
274
## self._check_lines(text)
436
275
new_version = len(self._parents)
438
278
# if we abort after here the (in-memory) weave will be corrupt because only
439
279
# some fields are updated
440
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
441
# - Robert Collins 20060226
442
280
self._parents.append(parents[:])
443
281
self._sha1s.append(sha1)
444
self._names.append(version_id)
445
self._name_map[version_id] = new_version
282
self._names.append(name)
283
self._name_map[name] = new_version
449
287
# special case; adding with no parents revision; can do
450
288
# this more quickly by just appending unconditionally.
451
289
# even more specially, if we're adding an empty text we
452
290
# need do nothing at all.
454
292
self._weave.append(('{', new_version))
455
self._weave.extend(lines)
293
self._weave.extend(text)
456
294
self._weave.append(('}', None))
457
296
return new_version
459
298
if len(parents) == 1:
561
436
except IndexError:
562
437
raise IndexError("invalid version number %r" % i)
564
def _compatible_parents(self, my_parents, other_parents):
565
"""During join check that other_parents are joinable with my_parents.
567
Joinable is defined as 'is a subset of' - supersets may require
568
regeneration of diffs, but subsets do not.
570
return len(other_parents.difference(my_parents)) == 0
572
def annotate(self, version_id):
573
"""Return a list of (version-id, line) tuples for version_id.
440
def annotate(self, name_or_index):
441
return list(self.annotate_iter(name_or_index))
444
def annotate_iter(self, name_or_index):
445
"""Yield list of (index-id, line) pairs for the specified version.
575
447
The index indicates when the line originated in the weave."""
576
incls = [self._lookup(version_id)]
577
return [(self._idx_to_name(origin), text) for origin, lineno, text in
578
self._extract(incls)]
580
def iter_lines_added_or_present_in_versions(self, version_ids=None,
582
"""See VersionedFile.iter_lines_added_or_present_in_versions()."""
583
if version_ids is None:
584
version_ids = self.versions()
585
version_ids = set(version_ids)
586
for lineno, inserted, deletes, line in self._walk_internal(version_ids):
587
if inserted not in version_ids: continue
589
yield line + '\n', inserted
593
def _walk_internal(self, version_ids=None):
594
"""Helper method for weave actions."""
448
incls = [self.maybe_lookup(name_or_index)]
449
for origin, lineno, text in self._extract(incls):
456
(lineno, insert, deletes, text)
457
for each literal line.
599
463
lineno = 0 # line of weave, 0-based
601
465
for l in self._weave:
602
if l.__class__ == tuple:
466
if isinstance(l, tuple):
606
istack.append(self._names[v])
610
dset.add(self._names[v])
612
dset.remove(self._names[v])
614
479
raise WeaveFormatError('unexpected instruction %r' % v)
616
yield lineno, istack[-1], frozenset(dset), l
481
assert isinstance(l, basestring)
483
yield lineno, istack[-1], dset, l
488
def _extract(self, versions):
489
"""Yield annotation of lines in included set.
491
Yields a sequence of tuples (origin, lineno, text), where
492
origin is the origin version, lineno the index in the weave,
493
and text the text of the line.
495
The set typically but not necessarily corresponds to a version.
498
if not isinstance(i, int):
501
included = self.inclusions(versions)
506
lineno = 0 # line of weave, 0-based
512
WFE = WeaveFormatError
514
for l in self._weave:
515
if isinstance(l, tuple):
519
assert v not in istack
533
assert isinstance(l, basestring)
535
isactive = (not dset) and istack and (istack[-1] in included)
537
result.append((istack[-1], lineno, l))
620
540
raise WeaveFormatError("unclosed insertion blocks "
621
541
"at end of weave: %s" % istack)
623
543
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
548
def get_iter(self, name_or_index):
549
"""Yield lines for the specified version."""
550
incls = [self.maybe_lookup(name_or_index)]
551
for origin, lineno, line in self._extract(incls):
555
def get_text(self, name_or_index):
556
return ''.join(self.get_iter(name_or_index))
557
assert isinstance(version, int)
560
def get_lines(self, name_or_index):
561
return list(self.get_iter(name_or_index))
567
def mash_iter(self, included):
568
"""Return composed version of multiple included versions."""
569
included = map(self.maybe_lookup, included)
570
for origin, lineno, text in self._extract(included):
574
def dump(self, to_file):
575
from pprint import pprint
576
print >>to_file, "Weave._weave = ",
577
pprint(self._weave, to_file)
578
print >>to_file, "Weave._parents = ",
579
pprint(self._parents, to_file)
583
def numversions(self):
584
l = len(self._parents)
585
assert l == len(self._sha1s)
590
return self.numversions()
593
def check(self, progress_bar=None):
594
# check no circular inclusions
595
for version in range(self.numversions()):
596
inclusions = list(self._parents[version])
599
if inclusions[-1] >= version:
600
raise WeaveFormatError("invalid included version %d for index %d"
601
% (inclusions[-1], version))
603
# try extracting all versions; this is a bit slow and parallel
604
# extraction could be used
605
nv = self.numversions()
606
for version in range(nv):
608
progress_bar.update('checking text', version, nv)
610
for l in self.get_iter(version):
613
expected = self._sha1s[version]
615
raise WeaveError("mismatched sha1 for version %d; "
616
"got %s, expected %s"
617
% (version, hd, expected))
619
# TODO: check insertions are properly nested, that there are
620
# no lines outside of insertion blocks, that deletions are
621
# properly paired, etc.
625
def merge(self, merge_versions):
626
"""Automerge and mark conflicts between versions.
628
This returns a sequence, each entry describing alternatives
629
for a chunk of the file. Each of the alternatives is given as
632
If there is a chunk of the file where there's no diagreement,
633
only one alternative is given.
635
# approach: find the included versions common to all the
637
raise NotImplementedError()
641
def _delta(self, included, lines):
642
"""Return changes from basis to new revision.
644
The old text for comparison is the union of included revisions.
646
This is used in inserting a new text.
648
Delta is returned as a sequence of
649
(weave1, weave2, newlines).
651
This indicates that weave1:weave2 of the old weave should be
652
replaced by the sequence of lines in newlines. Note that
653
these line numbers are positions in the total weave and don't
654
correspond to the lines in any extracted version, or even the
655
extracted union of included versions.
657
If line1=line2, this is a pure insert; if newlines=[] this is a
658
pure delete. (Similar to difflib.)
660
raise NotImplementedError()
626
663
def plan_merge(self, ver_a, ver_b):
627
664
"""Return pseudo-annotation indicating how the two versions merge.
667
704
# not in either revision
668
705
yield 'irrelevant', line
670
def _extract(self, versions):
671
"""Yield annotation of lines in included set.
673
Yields a sequence of tuples (origin, lineno, text), where
674
origin is the origin version, lineno the index in the weave,
675
and text the text of the line.
677
The set typically but not necessarily corresponds to a version.
680
if not isinstance(i, int):
683
included = self._inclusions(versions)
689
lineno = 0 # line of weave, 0-based
695
WFE = WeaveFormatError
698
# 449 0 4474.6820 2356.5590 bzrlib.weave:556(_extract)
699
# +285282 0 1676.8040 1676.8040 +<isinstance>
700
# 1.6 seconds in 'isinstance'.
701
# changing the first isinstance:
702
# 449 0 2814.2660 1577.1760 bzrlib.weave:556(_extract)
703
# +140414 0 762.8050 762.8050 +<isinstance>
704
# note that the inline time actually dropped (less function calls)
705
# and total processing time was halved.
706
# we're still spending ~1/4 of the method in isinstance though.
707
# so lets hard code the acceptable string classes we expect:
708
# 449 0 1202.9420 786.2930 bzrlib.weave:556(_extract)
709
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
711
# yay, down to ~1/4 the initial extract time, and our inline time
712
# has shrunk again, with isinstance no longer dominating.
713
# tweaking the stack inclusion test to use a set gives:
714
# 449 0 1122.8030 713.0080 bzrlib.weave:556(_extract)
715
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
717
# - a 5% win, or possibly just noise. However with large istacks that
718
# 'in' test could dominate, so I'm leaving this change in place -
719
# when its fast enough to consider profiling big datasets we can review.
724
for l in self._weave:
725
if l.__class__ == tuple:
732
iset.remove(istack.pop())
707
yield 'unchanged', '' # terminator
711
def weave_merge(self, plan):
716
for state, line in plan:
717
if state == 'unchanged' or state == 'killed-both':
718
# resync and flush queued conflicts changes if any
719
if not lines_a and not lines_b:
721
elif ch_a and not ch_b:
723
for l in lines_a: yield l
724
elif ch_b and not ch_a:
725
for l in lines_b: yield l
726
elif lines_a == lines_b:
727
for l in lines_a: yield l
740
raise AssertionError()
730
for l in lines_a: yield l
732
for l in lines_b: yield l
739
if state == 'unchanged':
742
elif state == 'killed-a':
745
elif state == 'killed-b':
748
elif state == 'new-a':
751
elif state == 'new-b':
743
isactive = (not dset) and istack and (istack[-1] in included)
745
result.append((istack[-1], lineno, l))
748
raise WeaveFormatError("unclosed insertion blocks "
749
"at end of weave: %s" % istack)
751
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
755
def _maybe_lookup(self, name_or_index):
756
"""Convert possible symbolic name to index, or pass through indexes.
755
assert state in ('irrelevant', 'ghost-a', 'ghost-b', 'killed-base',
760
def join(self, other):
761
"""Integrate versions from other into this weave.
763
The resulting weave contains all the history of both weaves;
764
any version you could retrieve from either self or other can be
765
retrieved from self after this call.
767
It is illegal for the two weaves to contain different values
768
or different parents for any version. See also reweave().
760
if isinstance(name_or_index, (int, long)):
763
return self._lookup(name_or_index)
765
def get_lines(self, version_id):
766
"""See VersionedFile.get_lines()."""
767
int_index = self._maybe_lookup(version_id)
768
result = [line for (origin, lineno, line) in self._extract([int_index])]
769
expected_sha1 = self._sha1s[int_index]
770
measured_sha1 = sha_strings(result)
771
if measured_sha1 != expected_sha1:
772
raise errors.WeaveInvalidChecksum(
773
'file %s, revision %s, expected: %s, measured %s'
774
% (self._weave_name, version_id,
775
expected_sha1, measured_sha1))
778
def get_sha1s(self, version_ids):
779
"""See VersionedFile.get_sha1s()."""
781
for v in version_ids:
782
result[v] = self._sha1s[self._lookup(v)]
785
def num_versions(self):
786
"""How many versions are in this weave?"""
787
l = len(self._parents)
790
__len__ = num_versions
792
def check(self, progress_bar=None):
793
# TODO evaluate performance hit of using string sets in this routine.
794
# TODO: check no circular inclusions
795
# TODO: create a nested progress bar
796
for version in range(self.num_versions()):
797
inclusions = list(self._parents[version])
800
if inclusions[-1] >= version:
801
raise WeaveFormatError("invalid included version %d for index %d"
802
% (inclusions[-1], version))
804
# try extracting all versions; parallel extraction is used
805
nv = self.num_versions()
810
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
811
# The problem is that set membership is much more expensive
812
name = self._idx_to_name(i)
815
new_inc = set([name])
816
for p in self._parents[i]:
817
new_inc.update(inclusions[self._idx_to_name(p)])
819
if set(new_inc) != set(self.get_ancestry(name)):
820
raise AssertionError(
822
% (set(new_inc), set(self.get_ancestry(name))))
823
inclusions[name] = new_inc
825
nlines = len(self._weave)
827
update_text = 'checking weave'
829
short_name = os.path.basename(self._weave_name)
830
update_text = 'checking %s' % (short_name,)
831
update_text = update_text[:25]
833
for lineno, insert, deleteset, line in self._walk_internal():
835
progress_bar.update(update_text, lineno, nlines)
837
for name, name_inclusions in inclusions.items():
838
# The active inclusion must be an ancestor,
839
# and no ancestors must have deleted this line,
840
# because we don't support resurrection.
841
if (insert in name_inclusions) and not (deleteset & name_inclusions):
842
sha1s[name].update(line)
845
version = self._idx_to_name(i)
846
hd = sha1s[version].hexdigest()
847
expected = self._sha1s[i]
849
raise errors.WeaveInvalidChecksum(
850
"mismatched sha1 for version %s: "
851
"got %s, expected %s"
852
% (version, hd, expected))
854
# TODO: check insertions are properly nested, that there are
855
# no lines outside of insertion blocks, that deletions are
856
# properly paired, etc.
770
if other.numversions() == 0:
771
return # nothing to update, easy
772
# two loops so that we do not change ourselves before verifying it
774
# work through in index order to make sure we get all dependencies
775
for other_idx, name in enumerate(other._names):
776
if self._check_version_consistent(other, other_idx, name):
778
for other_idx, name in enumerate(other._names):
779
# TODO: If all the parents of the other version are already
780
# present then we can avoid some work by just taking the delta
781
# and adjusting the offsets.
782
new_parents = self._imported_parents(other, other_idx)
783
lines = other.get_lines(other_idx)
784
sha1 = other._sha1s[other_idx]
785
self.add(name, new_parents, lines, sha1)
858
788
def _imported_parents(self, other, other_idx):
859
789
"""Return list of parents in self corresponding to indexes in other."""
861
791
for parent_idx in other._parents[other_idx]:
862
792
parent_name = other._names[parent_idx]
863
if parent_name not in self._name_map:
793
if parent_name not in self._names:
864
794
# should not be possible
865
raise WeaveError("missing parent {%s} of {%s} in %r"
795
raise WeaveError("missing parent {%s} of {%s} in %r"
866
796
% (parent_name, other._name_map[other_idx], self))
867
797
new_parents.append(self._name_map[parent_name])
868
798
return new_parents
898
def _reweave(self, other, pb, msg):
899
"""Reweave self with other - internal helper for join().
901
:param other: The other weave to merge
902
:param pb: An optional progress bar, indicating how far done we are
903
:param msg: An optional message for the progress
905
new_weave = _reweave(self, other, pb=pb, msg=msg)
906
self._copy_weave_content(new_weave)
908
def _copy_weave_content(self, otherweave):
909
"""adsorb the content from otherweave."""
832
def reweave(self, other):
833
"""Reweave self with other."""
834
new_weave = reweave(self, other)
910
835
for attr in self.__slots__:
911
if attr != '_weave_name':
912
setattr(self, attr, copy(getattr(otherweave, attr)))
915
class WeaveFile(Weave):
916
"""A WeaveFile represents a Weave on disk and writes on change."""
918
WEAVE_SUFFIX = '.weave'
920
def __init__(self, name, transport, filemode=None, create=False, access_mode='w', get_scope=None):
921
"""Create a WeaveFile.
923
:param create: If not True, only open an existing knit.
925
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope,
926
allow_reserved=False)
927
self._transport = transport
928
self._filemode = filemode
930
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
931
except errors.NoSuchFile:
937
def _add_lines(self, version_id, parents, lines, parent_texts,
938
left_matching_blocks, nostore_sha, random_id, check_content):
939
"""Add a version and save the weave."""
940
self.check_not_reserved_id(version_id)
941
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
942
parent_texts, left_matching_blocks, nostore_sha, random_id,
947
def copy_to(self, name, transport):
948
"""See VersionedFile.copy_to()."""
949
# as we are all in memory always, just serialise to the new place.
951
write_weave_v5(self, sio)
953
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
956
"""Save the weave."""
957
self._check_write_ok()
959
write_weave_v5(self, sio)
961
bytes = sio.getvalue()
962
path = self._weave_name + WeaveFile.WEAVE_SUFFIX
964
self._transport.put_bytes(path, bytes, self._filemode)
965
except errors.NoSuchFile:
966
self._transport.mkdir(dirname(path))
967
self._transport.put_bytes(path, bytes, self._filemode)
971
"""See VersionedFile.get_suffixes()."""
972
return [WeaveFile.WEAVE_SUFFIX]
974
def insert_record_stream(self, stream):
975
super(WeaveFile, self).insert_record_stream(stream)
979
def _reweave(wa, wb, pb=None, msg=None):
836
setattr(self, attr, getattr(new_weave, attr))
980
840
"""Combine two weaves and return the result.
982
This works even if a revision R has different parents in
842
This works even if a revision R has different parents in
983
843
wa and wb. In the resulting weave all the parents are given.
985
This is done by just building up a new weave, maintaining ordering
845
This is done by just building up a new weave, maintaining ordering
986
846
of the versions in the two inputs. More efficient approaches
987
might be possible but it should only be necessary to do
988
this operation rarely, when a new previously ghost version is
847
might be possible but it should only be necessary to do
848
this operation rarely, when a new previously ghost version is
991
:param pb: An optional progress bar, indicating how far done we are
992
:param msg: An optional message for the progress
996
queue_a = range(wa.num_versions())
997
queue_b = range(wb.num_versions())
853
queue_a = range(wa.numversions())
854
queue_b = range(wb.numversions())
998
855
# first determine combined parents of all versions
999
856
# map from version name -> all parent names
1000
857
combined_parents = _reweave_parent_graphs(wa, wb)
1001
858
mutter("combined parents: %r", combined_parents)
1002
order = tsort.topo_sort(combined_parents.iteritems())
859
order = topo_sort(combined_parents.iteritems())
1003
860
mutter("order to reweave: %r", order)
1008
for idx, name in enumerate(order):
1010
pb.update(msg, idx, len(order))
1011
862
if name in wa._name_map:
1012
863
lines = wa.get_lines(name)
1013
864
if name in wb._name_map:
1014
lines_b = wb.get_lines(name)
1015
if lines != lines_b:
1016
mutter('Weaves differ on content. rev_id {%s}', name)
1017
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
1019
lines = list(difflib.unified_diff(lines, lines_b,
1020
wa._weave_name, wb._weave_name))
1021
mutter('lines:\n%s', ''.join(lines))
1022
raise errors.WeaveTextDiffers(name, wa, wb)
865
assert lines == wb.get_lines(name)
1024
867
lines = wb.get_lines(name)
1025
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
868
wr.add(name, combined_parents[name], lines)
1028
872
def _reweave_parent_graphs(wa, wb):
1029
873
"""Return combined parent ancestry for two weaves.
1031
875
Returned as a list of (version_name, set(parent_names))"""
1033
877
for weave in [wa, wb]:
1034
878
for idx, name in enumerate(weave._names):
1035
879
p = combined.setdefault(name, set())
1036
p.update(map(weave._idx_to_name, weave._parents[idx]))
880
p.update(map(weave.idx_to_name, weave._parents[idx]))
1221
1072
print ' '.join(map(str, w._parents[int(argv[3])]))
1223
1074
elif cmd == 'plan-merge':
1224
# replaced by 'bzr weave-plan-merge'
1226
1076
for state, line in w.plan_merge(int(argv[3]), int(argv[4])):
1228
1078
print '%14s | %s' % (state, line),
1229
1080
elif cmd == 'merge':
1230
# replaced by 'bzr weave-merge-text'
1232
1082
p = w.plan_merge(int(argv[3]), int(argv[4]))
1233
1083
sys.stdout.writelines(w.weave_merge(p))
1085
elif cmd == 'mash-merge':
1091
v1, v2 = map(int, argv[3:5])
1093
basis = w.inclusions([v1]).intersection(w.inclusions([v2]))
1095
base_lines = list(w.mash_iter(basis))
1096
a_lines = list(w.get(v1))
1097
b_lines = list(w.get(v2))
1099
from bzrlib.merge3 import Merge3
1100
m3 = Merge3(base_lines, a_lines, b_lines)
1102
name_a = 'version %d' % v1
1103
name_b = 'version %d' % v2
1104
sys.stdout.writelines(m3.merge_lines(name_a=name_a, name_b=name_b))
1235
1106
raise ValueError('unknown command %r' % cmd)
1110
def profile_main(argv):
1111
import tempfile, hotshot, hotshot.stats
1113
prof_f = tempfile.NamedTemporaryFile()
1115
prof = hotshot.Profile(prof_f.name)
1117
ret = prof.runcall(main, argv)
1120
stats = hotshot.stats.load(prof_f.name)
1122
stats.sort_stats('cumulative')
1123
## XXX: Might like to write to stderr or the trace file instead but
1124
## print_stats seems hardcoded to stdout
1125
stats.print_stats(20)
1238
1130
if __name__ == '__main__':
1240
sys.exit(main(sys.argv))
1132
if '--profile' in sys.argv:
1134
args.remove('--profile')
1135
sys.exit(profile_main(args))
1137
sys.exit(main(sys.argv))
added
removed
bzrlib/weave.py
