22
22
"""Weave - storage of related text file versions"""
24
# TODO: Perhaps have copy method for Weave instances?
25
26
# XXX: If we do weaves this way, will a merge still behave the same
26
27
# way if it's done in a different order? That's a pretty desirable
30
# TODO: How to write these to disk? One option is cPickle, which
31
# would be fast but less friendly to C, and perhaps not portable. Another is
29
33
# TODO: Nothing here so far assumes the lines are really \n newlines,
30
# rather than being split up in some other way. We could accommodate
34
# rather than being split up in some other way. We could accomodate
31
35
# binaries, perhaps by naively splitting on \n or perhaps using
32
36
# something like a rolling checksum.
38
# TODO: Perhaps track SHA-1 in the header for protection? This would
39
# be redundant with it being stored in the inventory, but perhaps
42
# TODO: Track version names as well as indexes.
44
# TODO: Probably do transitive expansion when specifying parents?
46
# TODO: Separate out some code to read and write weaves.
34
48
# TODO: End marker for each version so we can stop reading?
36
50
# TODO: Check that no insertion occurs inside a deletion that was
37
51
# active in the version of the insertion.
39
# TODO: In addition to the SHA-1 check, perhaps have some code that
40
# checks structural constraints of the weave: ie that insertions are
41
# properly nested, that there is no text outside of an insertion, that
42
# insertions or deletions are not repeated, etc.
44
# TODO: Parallel-extract that passes back each line along with a
45
# description of which revisions include it. Nice for checking all
46
# shas or calculating stats in parallel.
48
# TODO: Using a single _extract routine and then processing the output
49
# is probably inefficient. It's simple enough that we can afford to
50
# have slight specializations for different ways its used: annotate,
51
# basis for add, get, etc.
53
# TODO: Probably the API should work only in names to hide the integer
54
# indexes from the user.
56
# TODO: Is there any potential performance win by having an add()
57
# variant that is passed a pre-cooked version of the single basis
60
# TODO: Reweave can possibly be made faster by remembering diffs
61
# where the basis and destination are unchanged.
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
66
# be done fairly efficiently because the sequence numbers constrain
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 *
97
from bzrlib.trace import mutter
98
from bzrlib.versionedfile import (
104
from bzrlib.weavefile import _read_weave_v5, write_weave_v5
107
class WeaveContentFactory(ContentFactory):
108
"""Content factory for streaming from weaves.
110
:seealso ContentFactory:
113
def __init__(self, version, weave):
114
"""Create a WeaveContentFactory for version from weave."""
115
ContentFactory.__init__(self)
116
self.sha1 = weave.get_sha1s([version])[version]
117
self.key = (version,)
118
parents = weave.get_parent_map([version])[version]
119
self.parents = tuple((parent,) for parent in parents)
120
self.storage_kind = 'fulltext'
123
def get_bytes_as(self, storage_kind):
124
if storage_kind == 'fulltext':
125
return self._weave.get_text(self.key[-1])
126
elif storage_kind == 'chunked':
127
return self._weave.get_lines(self.key[-1])
129
raise UnavailableRepresentation(self.key, storage_kind, 'fulltext')
132
class Weave(VersionedFile):
53
# TODO: Perhaps a special slower check() method that verifies more
54
# nesting constraints and the MD5 of each version?
62
from sets import Set, ImmutableSet
64
frozenset = ImmutableSet
68
class WeaveError(Exception):
69
"""Exception in processing weave"""
72
class WeaveFormatError(WeaveError):
73
"""Weave invariant violated"""
133
77
"""weave - versioned text file storage.
135
79
A Weave manages versions of line-based text files, keeping track
196
138
should be no way to get an earlier version deleting a later
200
Text of the weave; list of control instruction tuples and strings.
203
145
List of parents, indexed by version number.
204
146
It is only necessary to store the minimal set of parents for
205
147
each version; the parent's parents are implied.
208
List of hex SHA-1 of each version.
211
List of symbolic names for each version. Each should be unique.
214
For each name, the version number.
217
Descriptive name of this weave; typically the filename if known.
150
List of hex SHA-1 of each version, or None if not recorded.
221
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
222
'_weave_name', '_matcher', '_allow_reserved']
224
def __init__(self, weave_name=None, access_mode='w', matcher=None,
225
get_scope=None, allow_reserved=False):
228
:param get_scope: A callable that returns an opaque object to be used
229
for detecting when this weave goes out of scope (should stop
230
answering requests or allowing mutation).
232
super(Weave, self).__init__()
238
self._weave_name = weave_name
240
self._matcher = bzrlib.patiencediff.PatienceSequenceMatcher
242
self._matcher = matcher
243
if get_scope is None:
244
get_scope = lambda:None
245
self._get_scope = get_scope
246
self._scope = get_scope()
247
self._access_mode = access_mode
248
self._allow_reserved = allow_reserved
251
return "Weave(%r)" % self._weave_name
253
def _check_write_ok(self):
254
"""Is the versioned file marked as 'finished' ? Raise if it is."""
255
if self._get_scope() != self._scope:
256
raise errors.OutSideTransaction()
257
if self._access_mode != 'w':
258
raise errors.ReadOnlyObjectDirtiedError(self)
261
"""Return a deep copy of self.
263
The copy can be modified without affecting the original weave."""
265
other._weave = self._weave[:]
266
other._parents = self._parents[:]
267
other._sha1s = self._sha1s[:]
268
other._names = self._names[:]
269
other._name_map = self._name_map.copy()
270
other._weave_name = self._weave_name
273
158
def __eq__(self, other):
274
159
if not isinstance(other, Weave):
276
return self._parents == other._parents \
277
and self._weave == other._weave \
278
and self._sha1s == other._sha1s
161
return self._v == other._v \
162
and self._l == other._l
280
165
def __ne__(self, other):
281
166
return not self.__eq__(other)
283
def _idx_to_name(self, version):
284
return self._names[version]
286
def _lookup(self, name):
287
"""Convert symbolic version name to index."""
288
if not self._allow_reserved:
289
self.check_not_reserved_id(name)
291
return self._name_map[name]
293
raise RevisionNotPresent(name, self._weave_name)
296
"""See VersionedFile.versions."""
297
return self._names[:]
299
def has_version(self, version_id):
300
"""See VersionedFile.has_version."""
301
return (version_id in self._name_map)
303
__contains__ = has_version
305
def get_record_stream(self, versions, ordering, include_delta_closure):
306
"""Get a stream of records for versions.
308
:param versions: The versions to include. Each version is a tuple
310
:param ordering: Either 'unordered' or 'topological'. A topologically
311
sorted stream has compression parents strictly before their
313
:param include_delta_closure: If True then the closure across any
314
compression parents will be included (in the opaque data).
315
:return: An iterator of ContentFactory objects, each of which is only
316
valid until the iterator is advanced.
318
versions = [version[-1] for version in versions]
319
if ordering == 'topological':
320
parents = self.get_parent_map(versions)
321
new_versions = tsort.topo_sort(parents)
322
new_versions.extend(set(versions).difference(set(parents)))
323
versions = new_versions
324
for version in versions:
326
yield WeaveContentFactory(version, self)
328
yield AbsentContentFactory((version,))
330
def get_parent_map(self, version_ids):
331
"""See VersionedFile.get_parent_map."""
333
for version_id in version_ids:
334
if version_id == NULL_REVISION:
339
map(self._idx_to_name,
340
self._parents[self._lookup(version_id)]))
341
except RevisionNotPresent:
343
result[version_id] = parents
346
def get_parents_with_ghosts(self, version_id):
347
raise NotImplementedError(self.get_parents_with_ghosts)
349
def insert_record_stream(self, stream):
350
"""Insert a record stream into this versioned file.
352
:param stream: A stream of records to insert.
354
:seealso VersionedFile.get_record_stream:
357
for record in stream:
358
# Raise an error when a record is missing.
359
if record.storage_kind == 'absent':
360
raise RevisionNotPresent([record.key[0]], self)
361
# adapt to non-tuple interface
362
parents = [parent[0] for parent in record.parents]
363
if (record.storage_kind == 'fulltext'
364
or record.storage_kind == 'chunked'):
365
self.add_lines(record.key[0], parents,
366
osutils.chunks_to_lines(record.get_bytes_as('chunked')))
368
adapter_key = record.storage_kind, 'fulltext'
370
adapter = adapters[adapter_key]
372
adapter_factory = adapter_registry.get(adapter_key)
373
adapter = adapter_factory(self)
374
adapters[adapter_key] = adapter
375
lines = split_lines(adapter.get_bytes(
376
record, record.get_bytes_as(record.storage_kind)))
378
self.add_lines(record.key[0], parents, lines)
379
except RevisionAlreadyPresent:
382
def _check_repeated_add(self, name, parents, text, sha1):
383
"""Check that a duplicated add is OK.
385
If it is, return the (old) index; otherwise raise an exception.
387
idx = self._lookup(name)
388
if sorted(self._parents[idx]) != sorted(parents) \
389
or sha1 != self._sha1s[idx]:
390
raise RevisionAlreadyPresent(name, self._weave_name)
393
def _add_lines(self, version_id, parents, lines, parent_texts,
394
left_matching_blocks, nostore_sha, random_id, check_content):
395
"""See VersionedFile.add_lines."""
396
idx = self._add(version_id, lines, map(self._lookup, parents),
397
nostore_sha=nostore_sha)
398
return sha_strings(lines), sum(map(len, lines)), idx
400
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
169
def add(self, parents, text):
401
170
"""Add a single text on top of the weave.
403
172
Returns the index number of the newly added version.
406
Symbolic name for this version.
407
(Typically the revision-id of the revision that added it.)
410
175
List or set of direct parent version numbers.
413
Sequence of lines to be added in the new version.
415
:param nostore_sha: See VersionedFile.add_lines.
417
self._check_lines_not_unicode(lines)
418
self._check_lines_are_lines(lines)
420
sha1 = sha_strings(lines)
421
if sha1 == nostore_sha:
422
raise errors.ExistingContent
423
if version_id in self._name_map:
424
return self._check_repeated_add(version_id, parents, lines, sha1)
426
self._check_versions(parents)
427
## self._check_lines(lines)
428
new_version = len(self._parents)
430
# if we abort after here the (in-memory) weave will be corrupt because only
431
# some fields are updated
432
# XXX: FIXME implement a succeed-or-fail of the rest of this routine.
433
# - Robert Collins 20060226
434
self._parents.append(parents[:])
178
Sequence of lines to be added in the new version."""
179
## self._check_versions(parents)
180
## self._check_lines(text)
191
ancestors = self.inclusions(parents)
192
delta = self._delta(ancestors, text)
194
# offset gives the number of lines that have been inserted
195
# into the weave up to the current point; if the original edit instruction
196
# says to change line A then we actually change (A+offset)
199
for i1, i2, newlines in delta:
202
assert i2 <= len(self._l)
204
# the deletion and insertion are handled separately.
205
# first delete the region.
207
self._l.insert(i1+offset, ('[', idx))
208
self._l.insert(i2+offset+1, (']', idx))
213
# there may have been a deletion spanning up to
214
# i2; we want to insert after this region to make sure
215
# we don't destroy ourselves
217
self._l[i:i] = [('{', idx)] \
220
offset += 2 + len(newlines)
222
self._addversion(parents)
224
# special case; adding with no parents revision; can do this
225
# more quickly by just appending unconditionally
226
self._l.append(('{', idx))
228
self._l.append(('}', idx))
230
self._addversion(None)
435
232
self._sha1s.append(sha1)
436
self._names.append(version_id)
437
self._name_map[version_id] = new_version
441
# special case; adding with no parents revision; can do
442
# this more quickly by just appending unconditionally.
443
# even more specially, if we're adding an empty text we
444
# need do nothing at all.
446
self._weave.append(('{', new_version))
447
self._weave.extend(lines)
448
self._weave.append(('}', None))
451
if len(parents) == 1:
452
pv = list(parents)[0]
453
if sha1 == self._sha1s[pv]:
454
# special case: same as the single parent
458
ancestors = self._inclusions(parents)
462
# basis a list of (origin, lineno, line)
465
for origin, lineno, line in self._extract(ancestors):
466
basis_lineno.append(lineno)
467
basis_lines.append(line)
469
# another small special case: a merge, producing the same text
471
if lines == basis_lines:
474
# add a sentinel, because we can also match against the final line
475
basis_lineno.append(len(self._weave))
477
# XXX: which line of the weave should we really consider
478
# matches the end of the file? the current code says it's the
479
# last line of the weave?
481
#print 'basis_lines:', basis_lines
482
#print 'new_lines: ', lines
484
s = self._matcher(None, basis_lines, lines)
486
# offset gives the number of lines that have been inserted
487
# into the weave up to the current point; if the original edit instruction
488
# says to change line A then we actually change (A+offset)
491
for tag, i1, i2, j1, j2 in s.get_opcodes():
492
# i1,i2 are given in offsets within basis_lines; we need to map them
493
# back to offsets within the entire weave
494
#print 'raw match', tag, i1, i2, j1, j2
497
i1 = basis_lineno[i1]
498
i2 = basis_lineno[i2]
499
# the deletion and insertion are handled separately.
500
# first delete the region.
502
self._weave.insert(i1+offset, ('[', new_version))
503
self._weave.insert(i2+offset+1, (']', new_version))
507
# there may have been a deletion spanning up to
508
# i2; we want to insert after this region to make sure
509
# we don't destroy ourselves
511
self._weave[i:i] = ([('{', new_version)]
514
offset += 2 + (j2 - j1)
517
def _inclusions(self, versions):
237
def inclusions(self, versions):
518
238
"""Return set of all ancestors of given version(s)."""
519
if not len(versions):
244
# include all its parents
249
raise ValueError("version %d not present in weave" % v)
252
def minimal_parents(self, version):
253
"""Find the minimal set of parents for the version."""
254
included = self._v[version]
522
for v in xrange(max(versions), 0, -1):
524
# include all its parents
525
i.update(self._parents[v])
527
## except IndexError:
528
## raise ValueError("version %d not present in weave" % v)
530
def get_ancestry(self, version_ids, topo_sorted=True):
531
"""See VersionedFile.get_ancestry."""
532
if isinstance(version_ids, basestring):
533
version_ids = [version_ids]
534
i = self._inclusions([self._lookup(v) for v in version_ids])
535
return [self._idx_to_name(v) for v in i]
259
li.sort(reverse=True)
267
gotit.update(self.inclusions(pv))
269
assert mininc[0] >= 0
270
assert mininc[-1] < version
274
def _addversion(self, parents):
276
self._v.append(parents)
278
self._v.append(frozenset())
537
281
def _check_lines(self, text):
538
282
if not isinstance(text, list):
672
320
The set typically but not necessarily corresponds to a version.
675
if not isinstance(i, int):
678
included = self._inclusions(versions)
322
included = self.inclusions(versions)
684
327
lineno = 0 # line of weave, 0-based
690
331
WFE = WeaveFormatError
693
# 449 0 4474.6820 2356.5590 bzrlib.weave:556(_extract)
694
# +285282 0 1676.8040 1676.8040 +<isinstance>
695
# 1.6 seconds in 'isinstance'.
696
# changing the first isinstance:
697
# 449 0 2814.2660 1577.1760 bzrlib.weave:556(_extract)
698
# +140414 0 762.8050 762.8050 +<isinstance>
699
# note that the inline time actually dropped (less function calls)
700
# and total processing time was halved.
701
# we're still spending ~1/4 of the method in isinstance though.
702
# so lets hard code the acceptable string classes we expect:
703
# 449 0 1202.9420 786.2930 bzrlib.weave:556(_extract)
704
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
706
# yay, down to ~1/4 the initial extract time, and our inline time
707
# has shrunk again, with isinstance no longer dominating.
708
# tweaking the stack inclusion test to use a set gives:
709
# 449 0 1122.8030 713.0080 bzrlib.weave:556(_extract)
710
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
712
# - a 5% win, or possibly just noise. However with large istacks that
713
# 'in' test could dominate, so I'm leaving this change in place -
714
# when its fast enough to consider profiling big datasets we can review.
719
for l in self._weave:
720
if l.__class__ == tuple:
334
if isinstance(l, tuple):
338
assert v not in istack
727
iset.remove(istack.pop())
729
344
if v in included:
732
349
if v in included:
735
raise AssertionError()
353
assert isinstance(l, basestring)
737
354
if isactive is None:
738
355
isactive = (not dset) and istack and (istack[-1] in included)
740
result.append((istack[-1], lineno, l))
357
yield istack[-1], lineno, l
743
raise WeaveFormatError("unclosed insertion blocks "
744
"at end of weave: %s" % istack)
361
raise WFE("unclosed insertion blocks at end of weave",
746
raise WeaveFormatError("unclosed deletion blocks at end of weave: %s"
750
def _maybe_lookup(self, name_or_index):
751
"""Convert possible symbolic name to index, or pass through indexes.
755
if isinstance(name_or_index, (int, long)):
758
return self._lookup(name_or_index)
760
def get_lines(self, version_id):
761
"""See VersionedFile.get_lines()."""
762
int_index = self._maybe_lookup(version_id)
763
result = [line for (origin, lineno, line) in self._extract([int_index])]
764
expected_sha1 = self._sha1s[int_index]
765
measured_sha1 = sha_strings(result)
766
if measured_sha1 != expected_sha1:
767
raise errors.WeaveInvalidChecksum(
768
'file %s, revision %s, expected: %s, measured %s'
769
% (self._weave_name, version_id,
770
expected_sha1, measured_sha1))
773
def get_sha1s(self, version_ids):
774
"""See VersionedFile.get_sha1s()."""
776
for v in version_ids:
777
result[v] = self._sha1s[self._lookup(v)]
780
def num_versions(self):
781
"""How many versions are in this weave?"""
782
l = len(self._parents)
364
raise WFE("unclosed deletion blocks at end of weave",
368
def get_iter(self, version):
369
"""Yield lines for the specified version."""
370
for origin, lineno, line in self._extract([version]):
374
def get(self, index):
375
return list(self.get_iter(index))
378
def mash_iter(self, included):
379
"""Return composed version of multiple included versions."""
380
included = frozenset(included)
381
for origin, lineno, text in self._extract(included):
385
def dump(self, to_file):
386
from pprint import pprint
387
print >>to_file, "Weave._l = ",
388
pprint(self._l, to_file)
389
print >>to_file, "Weave._v = ",
390
pprint(self._v, to_file)
394
def numversions(self):
396
assert l == len(self._sha1s)
785
__len__ = num_versions
787
400
def check(self, progress_bar=None):
788
# TODO evaluate performance hit of using string sets in this routine.
789
# TODO: check no circular inclusions
790
# TODO: create a nested progress bar
791
for version in range(self.num_versions()):
792
inclusions = list(self._parents[version])
401
# check no circular inclusions
402
for version in range(self.numversions()):
403
inclusions = list(self._v[version])
794
405
inclusions.sort()
795
406
if inclusions[-1] >= version:
796
407
raise WeaveFormatError("invalid included version %d for index %d"
797
408
% (inclusions[-1], version))
799
# try extracting all versions; parallel extraction is used
800
nv = self.num_versions()
805
# For creating the ancestry, IntSet is much faster (3.7s vs 0.17s)
806
# The problem is that set membership is much more expensive
807
name = self._idx_to_name(i)
810
new_inc = set([name])
811
for p in self._parents[i]:
812
new_inc.update(inclusions[self._idx_to_name(p)])
814
if set(new_inc) != set(self.get_ancestry(name)):
815
raise AssertionError(
817
% (set(new_inc), set(self.get_ancestry(name))))
818
inclusions[name] = new_inc
820
nlines = len(self._weave)
822
update_text = 'checking weave'
824
short_name = os.path.basename(self._weave_name)
825
update_text = 'checking %s' % (short_name,)
826
update_text = update_text[:25]
828
for lineno, insert, deleteset, line in self._walk_internal():
410
# try extracting all versions; this is a bit slow and parallel
411
# extraction could be used
413
nv = self.numversions()
414
for version in range(nv):
830
progress_bar.update(update_text, lineno, nlines)
832
for name, name_inclusions in inclusions.items():
833
# The active inclusion must be an ancestor,
834
# and no ancestors must have deleted this line,
835
# because we don't support resurrection.
836
if (insert in name_inclusions) and not (deleteset & name_inclusions):
837
sha1s[name].update(line)
840
version = self._idx_to_name(i)
841
hd = sha1s[version].hexdigest()
842
expected = self._sha1s[i]
416
progress_bar.update('checking text', version, nv)
418
for l in self.get_iter(version):
421
expected = self._sha1s[version]
843
422
if hd != expected:
844
raise errors.WeaveInvalidChecksum(
845
"mismatched sha1 for version %s: "
846
"got %s, expected %s"
847
% (version, hd, expected))
423
raise WeaveError("mismatched sha1 for version %d; "
424
"got %s, expected %s"
425
% (version, hd, expected))
849
427
# TODO: check insertions are properly nested, that there are
850
428
# no lines outside of insertion blocks, that deletions are
851
429
# properly paired, etc.
853
def _imported_parents(self, other, other_idx):
854
"""Return list of parents in self corresponding to indexes in other."""
856
for parent_idx in other._parents[other_idx]:
857
parent_name = other._names[parent_idx]
858
if parent_name not in self._name_map:
859
# should not be possible
860
raise WeaveError("missing parent {%s} of {%s} in %r"
861
% (parent_name, other._name_map[other_idx], self))
862
new_parents.append(self._name_map[parent_name])
865
def _check_version_consistent(self, other, other_idx, name):
866
"""Check if a version in consistent in this and other.
868
To be consistent it must have:
871
* the same direct parents (by name, not index, and disregarding
874
If present & correct return True;
875
if not present in self return False;
876
if inconsistent raise error."""
877
this_idx = self._name_map.get(name, -1)
879
if self._sha1s[this_idx] != other._sha1s[other_idx]:
880
raise errors.WeaveTextDiffers(name, self, other)
881
self_parents = self._parents[this_idx]
882
other_parents = other._parents[other_idx]
883
n1 = set([self._names[i] for i in self_parents])
884
n2 = set([other._names[i] for i in other_parents])
885
if not self._compatible_parents(n1, n2):
886
raise WeaveParentMismatch("inconsistent parents "
887
"for version {%s}: %s vs %s" % (name, n1, n2))
893
def _reweave(self, other, pb, msg):
894
"""Reweave self with other - internal helper for join().
896
:param other: The other weave to merge
897
:param pb: An optional progress bar, indicating how far done we are
898
:param msg: An optional message for the progress
900
new_weave = _reweave(self, other, pb=pb, msg=msg)
901
self._copy_weave_content(new_weave)
903
def _copy_weave_content(self, otherweave):
904
"""adsorb the content from otherweave."""
905
for attr in self.__slots__:
906
if attr != '_weave_name':
907
setattr(self, attr, copy(getattr(otherweave, attr)))
910
class WeaveFile(Weave):
911
"""A WeaveFile represents a Weave on disk and writes on change."""
913
WEAVE_SUFFIX = '.weave'
915
def __init__(self, name, transport, filemode=None, create=False, access_mode='w', get_scope=None):
916
"""Create a WeaveFile.
918
:param create: If not True, only open an existing knit.
920
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope,
921
allow_reserved=False)
922
self._transport = transport
923
self._filemode = filemode
925
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
926
except errors.NoSuchFile:
932
def _add_lines(self, version_id, parents, lines, parent_texts,
933
left_matching_blocks, nostore_sha, random_id, check_content):
934
"""Add a version and save the weave."""
935
self.check_not_reserved_id(version_id)
936
result = super(WeaveFile, self)._add_lines(version_id, parents, lines,
937
parent_texts, left_matching_blocks, nostore_sha, random_id,
942
def copy_to(self, name, transport):
943
"""See VersionedFile.copy_to()."""
944
# as we are all in memory always, just serialise to the new place.
946
write_weave_v5(self, sio)
948
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
951
"""Save the weave."""
952
self._check_write_ok()
954
write_weave_v5(self, sio)
956
bytes = sio.getvalue()
957
path = self._weave_name + WeaveFile.WEAVE_SUFFIX
959
self._transport.put_bytes(path, bytes, self._filemode)
960
except errors.NoSuchFile:
961
self._transport.mkdir(dirname(path))
962
self._transport.put_bytes(path, bytes, self._filemode)
966
"""See VersionedFile.get_suffixes()."""
967
return [WeaveFile.WEAVE_SUFFIX]
969
def insert_record_stream(self, stream):
970
super(WeaveFile, self).insert_record_stream(stream)
973
@deprecated_method(one_five)
974
def join(self, other, pb=None, msg=None, version_ids=None,
975
ignore_missing=False):
976
"""Join other into self and save."""
977
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
981
def _reweave(wa, wb, pb=None, msg=None):
982
"""Combine two weaves and return the result.
984
This works even if a revision R has different parents in
985
wa and wb. In the resulting weave all the parents are given.
987
This is done by just building up a new weave, maintaining ordering
988
of the versions in the two inputs. More efficient approaches
989
might be possible but it should only be necessary to do
990
this operation rarely, when a new previously ghost version is
993
:param pb: An optional progress bar, indicating how far done we are
994
:param msg: An optional message for the progress
998
queue_a = range(wa.num_versions())
999
queue_b = range(wb.num_versions())
1000
# first determine combined parents of all versions
1001
# map from version name -> all parent names
1002
combined_parents = _reweave_parent_graphs(wa, wb)
1003
mutter("combined parents: %r", combined_parents)
1004
order = tsort.topo_sort(combined_parents.iteritems())
1005
mutter("order to reweave: %r", order)
1010
for idx, name in enumerate(order):
1012
pb.update(msg, idx, len(order))
1013
if name in wa._name_map:
1014
lines = wa.get_lines(name)
1015
if name in wb._name_map:
1016
lines_b = wb.get_lines(name)
1017
if lines != lines_b:
1018
mutter('Weaves differ on content. rev_id {%s}', name)
1019
mutter('weaves: %s, %s', wa._weave_name, wb._weave_name)
1021
lines = list(difflib.unified_diff(lines, lines_b,
1022
wa._weave_name, wb._weave_name))
1023
mutter('lines:\n%s', ''.join(lines))
1024
raise errors.WeaveTextDiffers(name, wa, wb)
1026
lines = wb.get_lines(name)
1027
wr._add(name, lines, [wr._lookup(i) for i in combined_parents[name]])
1030
def _reweave_parent_graphs(wa, wb):
1031
"""Return combined parent ancestry for two weaves.
1033
Returned as a list of (version_name, set(parent_names))"""
1035
for weave in [wa, wb]:
1036
for idx, name in enumerate(weave._names):
1037
p = combined.setdefault(name, set())
1038
p.update(map(weave._idx_to_name, weave._parents[idx]))
1043
"""Show the weave's table-of-contents"""
1044
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
1045
for i in (6, 50, 10, 10):
1048
for i in range(w.num_versions()):
1051
parent_str = ' '.join(map(str, w._parents[i]))
1052
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
1056
def weave_stats(weave_file, pb):
1057
from bzrlib.weavefile import read_weave
1059
wf = file(weave_file, 'rb')
433
def merge(self, merge_versions):
434
"""Automerge and mark conflicts between versions.
436
This returns a sequence, each entry describing alternatives
437
for a chunk of the file. Each of the alternatives is given as
440
If there is a chunk of the file where there's no diagreement,
441
only one alternative is given.
444
# approach: find the included versions common to all the
446
raise NotImplementedError()
450
def _delta(self, included, lines):
451
"""Return changes from basis to new revision.
453
The old text for comparison is the union of included revisions.
455
This is used in inserting a new text.
457
Delta is returned as a sequence of
458
(weave1, weave2, newlines).
460
This indicates that weave1:weave2 of the old weave should be
461
replaced by the sequence of lines in newlines. Note that
462
these line numbers are positions in the total weave and don't
463
correspond to the lines in any extracted version, or even the
464
extracted union of included versions.
466
If line1=line2, this is a pure insert; if newlines=[] this is a
467
pure delete. (Similar to difflib.)
469
# basis a list of (origin, lineno, line)
472
for origin, lineno, line in self._extract(included):
473
basis_lineno.append(lineno)
474
basis_lines.append(line)
476
# add a sentinal, because we can also match against the final line
477
basis_lineno.append(len(self._l))
479
# XXX: which line of the weave should we really consider
480
# matches the end of the file? the current code says it's the
481
# last line of the weave?
483
from difflib import SequenceMatcher
484
s = SequenceMatcher(None, basis_lines, lines)
486
# TODO: Perhaps return line numbers from composed weave as well?
488
for tag, i1, i2, j1, j2 in s.get_opcodes():
489
##print tag, i1, i2, j1, j2
494
# i1,i2 are given in offsets within basis_lines; we need to map them
495
# back to offsets within the entire weave
496
real_i1 = basis_lineno[i1]
497
real_i2 = basis_lineno[i2]
501
assert j2 <= len(lines)
503
yield real_i1, real_i2, lines[j1:j2]
507
def weave_info(filename, out):
508
"""Show some text information about the weave."""
509
from weavefile import read_weave
510
wf = file(filename, 'rb')
1060
511
w = read_weave(wf)
1061
512
# FIXME: doesn't work on pipes
1062
513
weave_size = wf.tell()
514
print >>out, "weave file size %d bytes" % weave_size
515
print >>out, "weave contains %d versions" % len(w._v)
1066
for i in range(vers):
1067
pb.update('checking sizes', i, vers)
1068
for origin, lineno, line in w._extract([i]):
1073
print 'versions %9d' % vers
1074
print 'weave file %9d bytes' % weave_size
1075
print 'total contents %9d bytes' % total
1076
print 'compression ratio %9.2fx' % (float(total) / float(weave_size))
1079
print 'average size %9d bytes' % avg
1080
print 'relative size %9.2fx' % (float(weave_size) / float(avg))
518
print '%6s %6s %8s %40s %20s' % ('ver', 'lines', 'bytes', 'sha1', 'parents')
519
for i in (6, 6, 8, 40, 20):
522
for i in range(len(w._v)):
525
bytes = sum((len(a) for a in text))
527
print '%6d %6d %8d %40s' % (i, lines, bytes, sha1),
533
print >>out, "versions total %d bytes" % total
534
print >>out, "compression ratio %.3f" % (float(total)/float(weave_size))
added
removed
bzrlib/weave.py
