69
71
from copy import copy
70
72
from cStringIO import StringIO
73
from bzrlib.lazy_import import lazy_import
74
lazy_import(globals(), """
75
from bzrlib import tsort
77
78
from bzrlib import (
81
from bzrlib.trace import mutter
81
82
from bzrlib.errors import (WeaveError, WeaveFormatError, WeaveParentMismatch,
82
83
RevisionAlreadyPresent,
83
84
RevisionNotPresent,
84
UnavailableRepresentation,
85
WeaveRevisionAlreadyPresent,
86
WeaveRevisionNotPresent,
86
from bzrlib.osutils import dirname, sha, sha_strings, split_lines
88
import bzrlib.errors as errors
89
from bzrlib.osutils import sha_strings
87
90
import bzrlib.patiencediff
88
from bzrlib.revision import NULL_REVISION
89
from bzrlib.symbol_versioning import *
90
from bzrlib.trace import mutter
91
from bzrlib.versionedfile import (
91
from bzrlib.tsort import topo_sort
92
from bzrlib.versionedfile import VersionedFile, InterVersionedFile
98
93
from bzrlib.weavefile import _read_weave_v5, write_weave_v5
101
class WeaveContentFactory(ContentFactory):
102
"""Content factory for streaming from weaves.
104
:seealso ContentFactory:
107
def __init__(self, version, weave):
108
"""Create a WeaveContentFactory for version from weave."""
109
ContentFactory.__init__(self)
110
self.sha1 = weave.get_sha1s([version])[version]
111
self.key = (version,)
112
parents = weave.get_parent_map([version])[version]
113
self.parents = tuple((parent,) for parent in parents)
114
self.storage_kind = 'fulltext'
117
def get_bytes_as(self, storage_kind):
118
if storage_kind == 'fulltext':
119
return self._weave.get_text(self.key[-1])
120
elif storage_kind == 'chunked':
121
return self._weave.get_lines(self.key[-1])
123
raise UnavailableRepresentation(self.key, storage_kind, 'fulltext')
126
96
class Weave(VersionedFile):
127
97
"""weave - versioned text file storage.
129
99
A Weave manages versions of line-based text files, keeping track
130
100
of the originating version for each line.
297
246
__contains__ = has_version
299
def get_record_stream(self, versions, ordering, include_delta_closure):
300
"""Get a stream of records for versions.
302
:param versions: The versions to include. Each version is a tuple
304
:param ordering: Either 'unordered' or 'topological'. A topologically
305
sorted stream has compression parents strictly before their
307
:param include_delta_closure: If True then the closure across any
308
compression parents will be included (in the opaque data).
309
:return: An iterator of ContentFactory objects, each of which is only
310
valid until the iterator is advanced.
312
versions = [version[-1] for version in versions]
313
if ordering == 'topological':
314
parents = self.get_parent_map(versions)
315
new_versions = tsort.topo_sort(parents)
316
new_versions.extend(set(versions).difference(set(parents)))
317
versions = new_versions
318
elif ordering == 'groupcompress':
319
parents = self.get_parent_map(versions)
320
new_versions = sort_groupcompress(parents)
321
new_versions.extend(set(versions).difference(set(parents)))
322
versions = new_versions
323
for version in versions:
325
yield WeaveContentFactory(version, self)
327
yield AbsentContentFactory((version,))
329
def get_parent_map(self, version_ids):
330
"""See VersionedFile.get_parent_map."""
332
for version_id in version_ids:
333
if version_id == NULL_REVISION:
338
map(self._idx_to_name,
339
self._parents[self._lookup(version_id)]))
340
except RevisionNotPresent:
342
result[version_id] = parents
345
def get_parents_with_ghosts(self, version_id):
346
raise NotImplementedError(self.get_parents_with_ghosts)
348
def insert_record_stream(self, stream):
349
"""Insert a record stream into this versioned file.
351
:param stream: A stream of records to insert.
353
:seealso VersionedFile.get_record_stream:
356
for record in stream:
357
# Raise an error when a record is missing.
358
if record.storage_kind == 'absent':
359
raise RevisionNotPresent([record.key[0]], self)
360
# adapt to non-tuple interface
361
parents = [parent[0] for parent in record.parents]
362
if (record.storage_kind == 'fulltext'
363
or record.storage_kind == 'chunked'):
364
self.add_lines(record.key[0], parents,
365
osutils.chunks_to_lines(record.get_bytes_as('chunked')))
367
adapter_key = record.storage_kind, 'fulltext'
369
adapter = adapters[adapter_key]
371
adapter_factory = adapter_registry.get(adapter_key)
372
adapter = adapter_factory(self)
373
adapters[adapter_key] = adapter
374
lines = split_lines(adapter.get_bytes(record))
376
self.add_lines(record.key[0], parents, lines)
377
except RevisionAlreadyPresent:
248
def get_parents(self, version_id):
249
"""See VersionedFile.get_parent."""
250
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
380
252
def _check_repeated_add(self, name, parents, text, sha1):
381
253
"""Check that a duplicated add is OK.
398
270
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
399
271
"""Add a single text on top of the weave.
401
273
Returns the index number of the newly added version.
404
276
Symbolic name for this version.
405
277
(Typically the revision-id of the revision that added it.)
406
If None, a name will be allocated based on the hash. (sha1:SHAHASH)
409
280
List or set of direct parent version numbers.
412
283
Sequence of lines to be added in the new version.
414
285
:param nostore_sha: See VersionedFile.add_lines.
287
assert isinstance(version_id, basestring)
416
288
self._check_lines_not_unicode(lines)
417
289
self._check_lines_are_lines(lines)
419
291
sha1 = sha_strings(lines)
420
292
if sha1 == nostore_sha:
421
293
raise errors.ExistingContent
422
if version_id is None:
423
version_id = "sha1:" + sha1
424
294
if version_id in self._name_map:
425
295
return self._check_repeated_add(version_id, parents, lines, sha1)
557
438
def _compatible_parents(self, my_parents, other_parents):
558
439
"""During join check that other_parents are joinable with my_parents.
560
Joinable is defined as 'is a subset of' - supersets may require
441
Joinable is defined as 'is a subset of' - supersets may require
561
442
regeneration of diffs, but subsets do not.
563
444
return len(other_parents.difference(my_parents)) == 0
565
def annotate(self, version_id):
566
"""Return a list of (version-id, line) tuples for version_id.
446
def annotate_iter(self, version_id):
447
"""Yield list of (version-id, line) pairs for the specified version.
568
449
The index indicates when the line originated in the weave."""
569
450
incls = [self._lookup(version_id)]
570
return [(self._idx_to_name(origin), text) for origin, lineno, text in
571
self._extract(incls)]
451
for origin, lineno, text in self._extract(incls):
452
yield self._idx_to_name(origin), text
573
454
def iter_lines_added_or_present_in_versions(self, version_ids=None,
699
588
# we're still spending ~1/4 of the method in isinstance though.
700
589
# so lets hard code the acceptable string classes we expect:
701
590
# 449 0 1202.9420 786.2930 bzrlib.weave:556(_extract)
702
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
591
# +71352 0 377.5560 377.5560 +<method 'append' of 'list'
704
593
# yay, down to ~1/4 the initial extract time, and our inline time
705
594
# has shrunk again, with isinstance no longer dominating.
706
595
# tweaking the stack inclusion test to use a set gives:
707
596
# 449 0 1122.8030 713.0080 bzrlib.weave:556(_extract)
708
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
597
# +71352 0 354.9980 354.9980 +<method 'append' of 'list'
710
599
# - a 5% win, or possibly just noise. However with large istacks that
711
600
# 'in' test could dominate, so I'm leaving this change in place -
712
601
# when its fast enough to consider profiling big datasets we can review.
717
606
for l in self._weave:
718
607
if l.__class__ == tuple:
725
615
iset.remove(istack.pop())
727
617
if v in included:
730
622
if v in included:
733
raise AssertionError()
626
assert l.__class__ in (str, unicode)
735
627
if isactive is None:
736
628
isactive = (not dset) and istack and (istack[-1] in included)
848
740
# no lines outside of insertion blocks, that deletions are
849
741
# properly paired, etc.
743
def _join(self, other, pb, msg, version_ids, ignore_missing):
744
"""Worker routine for join()."""
745
if not other.versions():
746
return # nothing to update, easy
749
# versions is never none, InterWeave checks this.
752
# two loops so that we do not change ourselves before verifying it
754
# work through in index order to make sure we get all dependencies
757
# get the selected versions only that are in other.versions.
758
version_ids = set(other.versions()).intersection(set(version_ids))
759
# pull in the referenced graph.
760
version_ids = other.get_ancestry(version_ids)
761
pending_graph = [(version, other.get_parents(version)) for
762
version in version_ids]
763
for name in topo_sort(pending_graph):
764
other_idx = other._name_map[name]
765
# returns True if we have it, False if we need it.
766
if not self._check_version_consistent(other, other_idx, name):
767
names_to_join.append((other_idx, name))
776
for other_idx, name in names_to_join:
777
# TODO: If all the parents of the other version are already
778
# present then we can avoid some work by just taking the delta
779
# and adjusting the offsets.
780
new_parents = self._imported_parents(other, other_idx)
781
sha1 = other._sha1s[other_idx]
786
pb.update(msg, merged, len(names_to_join))
788
lines = other.get_lines(other_idx)
789
self._add(name, lines, new_parents, sha1)
791
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
792
merged, processed, self._weave_name, time.time()-time0))
851
794
def _imported_parents(self, other, other_idx):
852
795
"""Return list of parents in self corresponding to indexes in other."""
909
852
"""A WeaveFile represents a Weave on disk and writes on change."""
911
854
WEAVE_SUFFIX = '.weave'
913
def __init__(self, name, transport, filemode=None, create=False, access_mode='w', get_scope=None):
856
def __init__(self, name, transport, filemode=None, create=False, access_mode='w'):
914
857
"""Create a WeaveFile.
916
859
:param create: If not True, only open an existing knit.
918
super(WeaveFile, self).__init__(name, access_mode, get_scope=get_scope,
919
allow_reserved=False)
861
super(WeaveFile, self).__init__(name, access_mode)
920
862
self._transport = transport
921
863
self._filemode = filemode
923
f = self._transport.get(name + WeaveFile.WEAVE_SUFFIX)
924
_read_weave_v5(StringIO(f.read()), self)
865
_read_weave_v5(self._transport.get(name + WeaveFile.WEAVE_SUFFIX), self)
925
866
except errors.NoSuchFile:
947
893
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
895
def create_empty(self, name, transport, filemode=None):
896
return WeaveFile(name, transport, filemode, create=True)
950
899
"""Save the weave."""
951
900
self._check_write_ok()
953
902
write_weave_v5(self, sio)
955
bytes = sio.getvalue()
956
path = self._weave_name + WeaveFile.WEAVE_SUFFIX
958
self._transport.put_bytes(path, bytes, self._filemode)
959
except errors.NoSuchFile:
960
self._transport.mkdir(dirname(path))
961
self._transport.put_bytes(path, bytes, self._filemode)
904
self._transport.put_file(self._weave_name + WeaveFile.WEAVE_SUFFIX,
964
909
def get_suffixes():
965
910
"""See VersionedFile.get_suffixes()."""
966
911
return [WeaveFile.WEAVE_SUFFIX]
968
def insert_record_stream(self, stream):
969
super(WeaveFile, self).insert_record_stream(stream)
913
def join(self, other, pb=None, msg=None, version_ids=None,
914
ignore_missing=False):
915
"""Join other into self and save."""
916
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
973
920
def _reweave(wa, wb, pb=None, msg=None):
974
921
"""Combine two weaves and return the result.
976
This works even if a revision R has different parents in
923
This works even if a revision R has different parents in
977
924
wa and wb. In the resulting weave all the parents are given.
979
This is done by just building up a new weave, maintaining ordering
926
This is done by just building up a new weave, maintaining ordering
980
927
of the versions in the two inputs. More efficient approaches
981
might be possible but it should only be necessary to do
982
this operation rarely, when a new previously ghost version is
928
might be possible but it should only be necessary to do
929
this operation rarely, when a new previously ghost version is
985
932
:param pb: An optional progress bar, indicating how far done we are
1030
976
p = combined.setdefault(name, set())
1031
977
p.update(map(weave._idx_to_name, weave._parents[idx]))
982
"""Show the weave's table-of-contents"""
983
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
984
for i in (6, 50, 10, 10):
987
for i in range(w.num_versions()):
990
parent_str = ' '.join(map(str, w._parents[i]))
991
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
995
def weave_stats(weave_file, pb):
996
from bzrlib.weavefile import read_weave
998
wf = file(weave_file, 'rb')
1000
# FIXME: doesn't work on pipes
1001
weave_size = wf.tell()
1005
for i in range(vers):
1006
pb.update('checking sizes', i, vers)
1007
for origin, lineno, line in w._extract([i]):
1012
print 'versions %9d' % vers
1013
print 'weave file %9d bytes' % weave_size
1014
print 'total contents %9d bytes' % total
1015
print 'compression ratio %9.2fx' % (float(total) / float(weave_size))
1018
print 'average size %9d bytes' % avg
1019
print 'relative size %9.2fx' % (float(weave_size) / float(avg))
1023
print """bzr weave tool
1025
Experimental tool for weave algorithm.
1028
weave init WEAVEFILE
1029
Create an empty weave file
1030
weave get WEAVEFILE VERSION
1031
Write out specified version.
1032
weave check WEAVEFILE
1033
Check consistency of all versions.
1035
Display table of contents.
1036
weave add WEAVEFILE NAME [BASE...] < NEWTEXT
1037
Add NEWTEXT, with specified parent versions.
1038
weave annotate WEAVEFILE VERSION
1039
Display origin of each line.
1040
weave merge WEAVEFILE VERSION1 VERSION2 > OUT
1041
Auto-merge two versions and display conflicts.
1042
weave diff WEAVEFILE VERSION1 VERSION2
1043
Show differences between two versions.
1047
% weave init foo.weave
1049
% weave add foo.weave ver0 < foo.txt
1052
(create updated version)
1054
% weave get foo.weave 0 | diff -u - foo.txt
1055
% weave add foo.weave ver1 0 < foo.txt
1058
% weave get foo.weave 0 > foo.txt (create forked version)
1060
% weave add foo.weave ver2 0 < foo.txt
1063
% weave merge foo.weave 1 2 > foo.txt (merge them)
1064
% vi foo.txt (resolve conflicts)
1065
% weave add foo.weave merged 1 2 < foo.txt (commit merged version)
1077
# in case we're run directly from the subdirectory
1078
sys.path.append('..')
1080
from bzrlib.weavefile import write_weave, read_weave
1081
from bzrlib.progress import ProgressBar
1096
return read_weave(file(argv[2], 'rb'))
1102
# at the moment, based on everything in the file
1104
parents = map(int, argv[4:])
1105
lines = sys.stdin.readlines()
1106
ver = w.add(name, parents, lines)
1107
write_weave(w, file(argv[2], 'wb'))
1108
print 'added version %r %d' % (name, ver)
1111
if os.path.exists(fn):
1112
raise IOError("file exists")
1114
write_weave(w, file(fn, 'wb'))
1115
elif cmd == 'get': # get one version
1117
sys.stdout.writelines(w.get_iter(int(argv[3])))
1122
v1, v2 = map(int, argv[3:5])
1125
diff_gen = bzrlib.patiencediff.unified_diff(lines1, lines2,
1126
'%s version %d' % (fn, v1),
1127
'%s version %d' % (fn, v2))
1128
sys.stdout.writelines(diff_gen)
1130
elif cmd == 'annotate':
1132
# newline is added to all lines regardless; too hard to get
1133
# reasonable formatting otherwise
1135
for origin, text in w.annotate(int(argv[3])):
1136
text = text.rstrip('\r\n')
1138
print ' | %s' % (text)
1140
print '%5d | %s' % (origin, text)
1146
elif cmd == 'stats':
1147
weave_stats(argv[2], ProgressBar())
1149
elif cmd == 'check':
1154
print '%d versions ok' % w.num_versions()
1156
elif cmd == 'inclusions':
1158
print ' '.join(map(str, w.inclusions([int(argv[3])])))
1160
elif cmd == 'parents':
1162
print ' '.join(map(str, w._parents[int(argv[3])]))
1164
elif cmd == 'plan-merge':
1165
# replaced by 'bzr weave-plan-merge'
1167
for state, line in w.plan_merge(int(argv[3]), int(argv[4])):
1169
print '%14s | %s' % (state, line),
1170
elif cmd == 'merge':
1171
# replaced by 'bzr weave-merge-text'
1173
p = w.plan_merge(int(argv[3]), int(argv[4]))
1174
sys.stdout.writelines(w.weave_merge(p))
1176
raise ValueError('unknown command %r' % cmd)
1179
if __name__ == '__main__':
1181
sys.exit(main(sys.argv))
1184
class InterWeave(InterVersionedFile):
1185
"""Optimised code paths for weave to weave operations."""
1187
_matching_file_from_factory = staticmethod(WeaveFile)
1188
_matching_file_to_factory = staticmethod(WeaveFile)
1191
def is_compatible(source, target):
1192
"""Be compatible with weaves."""
1194
return (isinstance(source, Weave) and
1195
isinstance(target, Weave))
1196
except AttributeError:
1199
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
1200
"""See InterVersionedFile.join."""
1201
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
1202
if self.target.versions() == [] and version_ids is None:
1203
self.target._copy_weave_content(self.source)
1205
self.target._join(self.source, pb, msg, version_ids, ignore_missing)
1208
InterVersionedFile.register_optimiser(InterWeave)
added
removed
bzrlib/weave.py
