69
71
from copy import copy
70
72
from cStringIO import StringIO
75
from bzrlib.lazy_import import lazy_import
76
lazy_import(globals(), """
77
from bzrlib import tsort
79
78
from bzrlib import (
81
from bzrlib.trace import mutter
84
82
from bzrlib.errors import (WeaveError, WeaveFormatError, WeaveParentMismatch,
85
83
RevisionAlreadyPresent,
86
84
RevisionNotPresent,
87
UnavailableRepresentation,
88
85
WeaveRevisionAlreadyPresent,
89
86
WeaveRevisionNotPresent,
91
from bzrlib.osutils import dirname, sha, sha_strings, split_lines
88
import bzrlib.errors as errors
89
from bzrlib.osutils import sha_strings
92
90
import bzrlib.patiencediff
93
from bzrlib.revision import NULL_REVISION
94
from bzrlib.symbol_versioning import *
95
from bzrlib.trace import mutter
96
from bzrlib.versionedfile import (
91
from bzrlib.symbol_versioning import (deprecated_method,
95
from bzrlib.tsort import topo_sort
96
from bzrlib.versionedfile import VersionedFile, InterVersionedFile
103
97
from bzrlib.weavefile import _read_weave_v5, write_weave_v5
106
class WeaveContentFactory(ContentFactory):
107
"""Content factory for streaming from weaves.
109
:seealso ContentFactory:
112
def __init__(self, version, weave):
113
"""Create a WeaveContentFactory for version from weave."""
114
ContentFactory.__init__(self)
115
self.sha1 = weave.get_sha1s([version])[version]
116
self.key = (version,)
117
parents = weave.get_parent_map([version])[version]
118
self.parents = tuple((parent,) for parent in parents)
119
self.storage_kind = 'fulltext'
122
def get_bytes_as(self, storage_kind):
123
if storage_kind == 'fulltext':
124
return self._weave.get_text(self.key[-1])
125
elif storage_kind == 'chunked':
126
return self._weave.get_lines(self.key[-1])
128
raise UnavailableRepresentation(self.key, storage_kind, 'fulltext')
131
100
class Weave(VersionedFile):
132
101
"""weave - versioned text file storage.
134
103
A Weave manages versions of line-based text files, keeping track
135
104
of the originating version for each line.
302
274
__contains__ = has_version
304
def get_record_stream(self, versions, ordering, include_delta_closure):
305
"""Get a stream of records for versions.
307
:param versions: The versions to include. Each version is a tuple
309
:param ordering: Either 'unordered' or 'topological'. A topologically
310
sorted stream has compression parents strictly before their
312
:param include_delta_closure: If True then the closure across any
313
compression parents will be included (in the opaque data).
314
:return: An iterator of ContentFactory objects, each of which is only
315
valid until the iterator is advanced.
317
versions = [version[-1] for version in versions]
318
if ordering == 'topological':
319
parents = self.get_parent_map(versions)
320
new_versions = tsort.topo_sort(parents)
321
new_versions.extend(set(versions).difference(set(parents)))
322
versions = new_versions
323
elif ordering == 'groupcompress':
324
parents = self.get_parent_map(versions)
325
new_versions = sort_groupcompress(parents)
326
new_versions.extend(set(versions).difference(set(parents)))
327
versions = new_versions
328
for version in versions:
330
yield WeaveContentFactory(version, self)
332
yield AbsentContentFactory((version,))
334
def get_parent_map(self, version_ids):
335
"""See VersionedFile.get_parent_map."""
276
def get_delta(self, version_id):
277
"""See VersionedFile.get_delta."""
278
return self.get_deltas([version_id])[version_id]
280
def get_deltas(self, version_ids):
281
"""See VersionedFile.get_deltas."""
282
version_ids = self.get_ancestry(version_ids)
337
283
for version_id in version_ids:
338
if version_id == NULL_REVISION:
343
map(self._idx_to_name,
344
self._parents[self._lookup(version_id)]))
345
except RevisionNotPresent:
284
if not self.has_version(version_id):
285
raise RevisionNotPresent(version_id, self)
286
# try extracting all versions; parallel extraction is used
287
nv = self.num_versions()
293
last_parent_lines = {}
295
parent_inclusions = {}
300
# its simplest to generate a full set of prepared variables.
302
name = self._names[i]
303
sha1s[name] = self.get_sha1(name)
304
parents_list = self.get_parents(name)
306
parent = parents_list[0]
307
parents[name] = parent
308
parent_inclusions[name] = inclusions[parent]
311
parent_inclusions[name] = set()
312
# we want to emit start, finish, replacement_length, replacement_lines tuples.
313
diff_hunks[name] = []
314
current_hunks[name] = [0, 0, 0, []] # #start, finish, repl_length, repl_tuples
315
parent_linenums[name] = 0
317
parent_noeols[name] = False
318
last_parent_lines[name] = None
319
new_inc = set([name])
320
for p in self._parents[i]:
321
new_inc.update(inclusions[self._idx_to_name(p)])
322
# debug only, known good so far.
323
#assert set(new_inc) == set(self.get_ancestry(name)), \
324
# 'failed %s != %s' % (set(new_inc), set(self.get_ancestry(name)))
325
inclusions[name] = new_inc
327
nlines = len(self._weave)
329
for lineno, inserted, deletes, line in self._walk_internal():
330
# a line is active in a version if:
331
# insert is in the versions inclusions
333
# deleteset & the versions inclusions is an empty set.
334
# so - if we have a included by mapping - version is included by
335
# children, we get a list of children to examine for deletes affect
336
# ing them, which is less than the entire set of children.
337
for version_id in version_ids:
338
# The active inclusion must be an ancestor,
339
# and no ancestors must have deleted this line,
340
# because we don't support resurrection.
341
parent_inclusion = parent_inclusions[version_id]
342
inclusion = inclusions[version_id]
343
parent_active = inserted in parent_inclusion and not (deletes & parent_inclusion)
344
version_active = inserted in inclusion and not (deletes & inclusion)
345
if not parent_active and not version_active:
346
# unrelated line of ancestry
347
result[version_id] = parents
348
elif parent_active and version_active:
350
parent_linenum = parent_linenums[version_id]
351
if current_hunks[version_id] != [parent_linenum, parent_linenum, 0, []]:
352
diff_hunks[version_id].append(tuple(current_hunks[version_id]))
354
current_hunks[version_id] = [parent_linenum, parent_linenum, 0, []]
355
parent_linenums[version_id] = parent_linenum
358
noeols[version_id] = True
361
elif parent_active and not version_active:
363
current_hunks[version_id][1] += 1
364
parent_linenums[version_id] += 1
365
last_parent_lines[version_id] = line
366
elif not parent_active and version_active:
368
# noeol only occurs at the end of a file because we
369
# diff linewise. We want to show noeol changes as a
370
# empty diff unless the actual eol-less content changed.
373
if last_parent_lines[version_id][-1] != '\n':
374
parent_noeols[version_id] = True
375
except (TypeError, IndexError):
378
if theline[-1] != '\n':
379
noeols[version_id] = True
383
parent_should_go = False
385
if parent_noeols[version_id] == noeols[version_id]:
386
# no noeol toggle, so trust the weaves statement
387
# that this line is changed.
389
if parent_noeols[version_id]:
390
theline = theline + '\n'
391
elif parent_noeols[version_id]:
392
# parent has no eol, we do:
393
# our line is new, report as such..
395
elif noeols[version_id]:
396
# append a eol so that it looks like
398
theline = theline + '\n'
399
if parents[version_id] is not None:
400
#if last_parent_lines[version_id] is not None:
401
parent_should_go = True
402
if last_parent_lines[version_id] != theline:
405
#parent_should_go = False
407
current_hunks[version_id][2] += 1
408
current_hunks[version_id][3].append((inserted, theline))
410
# last hunk last parent line is not eaten
411
current_hunks[version_id][1] -= 1
412
if current_hunks[version_id][1] < 0:
413
current_hunks[version_id][1] = 0
414
# import pdb;pdb.set_trace()
415
# assert current_hunks[version_id][1] >= 0
419
version = self._idx_to_name(i)
420
if current_hunks[version] != [0, 0, 0, []]:
421
diff_hunks[version].append(tuple(current_hunks[version]))
423
for version_id in version_ids:
424
result[version_id] = (
428
diff_hunks[version_id],
350
def get_parents_with_ghosts(self, version_id):
351
raise NotImplementedError(self.get_parents_with_ghosts)
353
def insert_record_stream(self, stream):
354
"""Insert a record stream into this versioned file.
356
:param stream: A stream of records to insert.
358
:seealso VersionedFile.get_record_stream:
361
for record in stream:
362
# Raise an error when a record is missing.
363
if record.storage_kind == 'absent':
364
raise RevisionNotPresent([record.key[0]], self)
365
# adapt to non-tuple interface
366
parents = [parent[0] for parent in record.parents]
367
if (record.storage_kind == 'fulltext'
368
or record.storage_kind == 'chunked'):
369
self.add_lines(record.key[0], parents,
370
osutils.chunks_to_lines(record.get_bytes_as('chunked')))
372
adapter_key = record.storage_kind, 'fulltext'
374
adapter = adapters[adapter_key]
376
adapter_factory = adapter_registry.get(adapter_key)
377
adapter = adapter_factory(self)
378
adapters[adapter_key] = adapter
379
lines = split_lines(adapter.get_bytes(record))
381
self.add_lines(record.key[0], parents, lines)
382
except RevisionAlreadyPresent:
432
def get_parents(self, version_id):
433
"""See VersionedFile.get_parent."""
434
return map(self._idx_to_name, self._parents[self._lookup(version_id)])
385
436
def _check_repeated_add(self, name, parents, text, sha1):
386
437
"""Check that a duplicated add is OK.
393
444
raise RevisionAlreadyPresent(name, self._weave_name)
396
def _add_lines(self, version_id, parents, lines, parent_texts,
397
left_matching_blocks, nostore_sha, random_id, check_content):
447
@deprecated_method(zero_eight)
448
def add_identical(self, old_rev_id, new_rev_id, parents):
449
"""Please use Weave.clone_text now."""
450
return self.clone_text(new_rev_id, old_rev_id, parents)
452
def _add_lines(self, version_id, parents, lines, parent_texts):
398
453
"""See VersionedFile.add_lines."""
399
idx = self._add(version_id, lines, map(self._lookup, parents),
400
nostore_sha=nostore_sha)
401
return sha_strings(lines), sum(map(len, lines)), idx
403
def _add(self, version_id, lines, parents, sha1=None, nostore_sha=None):
454
return self._add(version_id, lines, map(self._lookup, parents))
456
@deprecated_method(zero_eight)
457
def add(self, name, parents, text, sha1=None):
458
"""See VersionedFile.add_lines for the non deprecated api."""
459
return self._add(name, text, map(self._maybe_lookup, parents), sha1)
461
def _add(self, version_id, lines, parents, sha1=None):
404
462
"""Add a single text on top of the weave.
406
464
Returns the index number of the newly added version.
409
467
Symbolic name for this version.
410
468
(Typically the revision-id of the revision that added it.)
411
If None, a name will be allocated based on the hash. (sha1:SHAHASH)
414
471
List or set of direct parent version numbers.
417
474
Sequence of lines to be added in the new version.
419
:param nostore_sha: See VersionedFile.add_lines.
477
assert isinstance(version_id, basestring)
421
478
self._check_lines_not_unicode(lines)
422
479
self._check_lines_are_lines(lines)
424
481
sha1 = sha_strings(lines)
425
if sha1 == nostore_sha:
426
raise errors.ExistingContent
427
if version_id is None:
428
version_id = "sha1:" + sha1
429
482
if version_id in self._name_map:
430
483
return self._check_repeated_add(version_id, parents, lines, sha1)
853
975
# no lines outside of insertion blocks, that deletions are
854
976
# properly paired, etc.
978
def _join(self, other, pb, msg, version_ids, ignore_missing):
979
"""Worker routine for join()."""
980
if not other.versions():
981
return # nothing to update, easy
984
# versions is never none, InterWeave checks this.
987
# two loops so that we do not change ourselves before verifying it
989
# work through in index order to make sure we get all dependencies
992
# get the selected versions only that are in other.versions.
993
version_ids = set(other.versions()).intersection(set(version_ids))
994
# pull in the referenced graph.
995
version_ids = other.get_ancestry(version_ids)
996
pending_graph = [(version, other.get_parents(version)) for
997
version in version_ids]
998
for name in topo_sort(pending_graph):
999
other_idx = other._name_map[name]
1000
# returns True if we have it, False if we need it.
1001
if not self._check_version_consistent(other, other_idx, name):
1002
names_to_join.append((other_idx, name))
1011
for other_idx, name in names_to_join:
1012
# TODO: If all the parents of the other version are already
1013
# present then we can avoid some work by just taking the delta
1014
# and adjusting the offsets.
1015
new_parents = self._imported_parents(other, other_idx)
1016
sha1 = other._sha1s[other_idx]
1021
pb.update(msg, merged, len(names_to_join))
1023
lines = other.get_lines(other_idx)
1024
self._add(name, lines, new_parents, sha1)
1026
mutter("merged = %d, processed = %d, file_id=%s; deltat=%d"%(
1027
merged, processed, self._weave_name, time.time()-time0))
856
1029
def _imported_parents(self, other, other_idx):
857
1030
"""Return list of parents in self corresponding to indexes in other."""
858
1031
new_parents = []
951
1131
transport.put_file(name + WeaveFile.WEAVE_SUFFIX, sio, self._filemode)
1133
def create_empty(self, name, transport, filemode=None):
1134
return WeaveFile(name, transport, filemode, create=True)
953
1136
def _save(self):
954
1137
"""Save the weave."""
955
1138
self._check_write_ok()
956
1139
sio = StringIO()
957
1140
write_weave_v5(self, sio)
959
bytes = sio.getvalue()
960
path = self._weave_name + WeaveFile.WEAVE_SUFFIX
962
self._transport.put_bytes(path, bytes, self._filemode)
963
except errors.NoSuchFile:
964
self._transport.mkdir(dirname(path))
965
self._transport.put_bytes(path, bytes, self._filemode)
1142
self._transport.put_file(self._weave_name + WeaveFile.WEAVE_SUFFIX,
968
1147
def get_suffixes():
969
1148
"""See VersionedFile.get_suffixes()."""
970
1149
return [WeaveFile.WEAVE_SUFFIX]
972
def insert_record_stream(self, stream):
973
super(WeaveFile, self).insert_record_stream(stream)
1151
def join(self, other, pb=None, msg=None, version_ids=None,
1152
ignore_missing=False):
1153
"""Join other into self and save."""
1154
super(WeaveFile, self).join(other, pb, msg, version_ids, ignore_missing)
1158
@deprecated_function(zero_eight)
1159
def reweave(wa, wb, pb=None, msg=None):
1160
"""reweaving is deprecation, please just use weave.join()."""
1161
_reweave(wa, wb, pb, msg)
977
1163
def _reweave(wa, wb, pb=None, msg=None):
978
1164
"""Combine two weaves and return the result.
980
This works even if a revision R has different parents in
1166
This works even if a revision R has different parents in
981
1167
wa and wb. In the resulting weave all the parents are given.
983
This is done by just building up a new weave, maintaining ordering
1169
This is done by just building up a new weave, maintaining ordering
984
1170
of the versions in the two inputs. More efficient approaches
985
might be possible but it should only be necessary to do
986
this operation rarely, when a new previously ghost version is
1171
might be possible but it should only be necessary to do
1172
this operation rarely, when a new previously ghost version is
989
1175
:param pb: An optional progress bar, indicating how far done we are
1034
1219
p = combined.setdefault(name, set())
1035
1220
p.update(map(weave._idx_to_name, weave._parents[idx]))
1036
1221
return combined
1225
"""Show the weave's table-of-contents"""
1226
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
1227
for i in (6, 50, 10, 10):
1230
for i in range(w.num_versions()):
1233
parent_str = ' '.join(map(str, w._parents[i]))
1234
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
1238
def weave_stats(weave_file, pb):
1239
from bzrlib.weavefile import read_weave
1241
wf = file(weave_file, 'rb')
1243
# FIXME: doesn't work on pipes
1244
weave_size = wf.tell()
1248
for i in range(vers):
1249
pb.update('checking sizes', i, vers)
1250
for origin, lineno, line in w._extract([i]):
1255
print 'versions %9d' % vers
1256
print 'weave file %9d bytes' % weave_size
1257
print 'total contents %9d bytes' % total
1258
print 'compression ratio %9.2fx' % (float(total) / float(weave_size))
1261
print 'average size %9d bytes' % avg
1262
print 'relative size %9.2fx' % (float(weave_size) / float(avg))
1266
print """bzr weave tool
1268
Experimental tool for weave algorithm.
1271
weave init WEAVEFILE
1272
Create an empty weave file
1273
weave get WEAVEFILE VERSION
1274
Write out specified version.
1275
weave check WEAVEFILE
1276
Check consistency of all versions.
1278
Display table of contents.
1279
weave add WEAVEFILE NAME [BASE...] < NEWTEXT
1280
Add NEWTEXT, with specified parent versions.
1281
weave annotate WEAVEFILE VERSION
1282
Display origin of each line.
1283
weave merge WEAVEFILE VERSION1 VERSION2 > OUT
1284
Auto-merge two versions and display conflicts.
1285
weave diff WEAVEFILE VERSION1 VERSION2
1286
Show differences between two versions.
1290
% weave init foo.weave
1292
% weave add foo.weave ver0 < foo.txt
1295
(create updated version)
1297
% weave get foo.weave 0 | diff -u - foo.txt
1298
% weave add foo.weave ver1 0 < foo.txt
1301
% weave get foo.weave 0 > foo.txt (create forked version)
1303
% weave add foo.weave ver2 0 < foo.txt
1306
% weave merge foo.weave 1 2 > foo.txt (merge them)
1307
% vi foo.txt (resolve conflicts)
1308
% weave add foo.weave merged 1 2 < foo.txt (commit merged version)
1320
# in case we're run directly from the subdirectory
1321
sys.path.append('..')
1323
from bzrlib.weavefile import write_weave, read_weave
1324
from bzrlib.progress import ProgressBar
1339
return read_weave(file(argv[2], 'rb'))
1345
# at the moment, based on everything in the file
1347
parents = map(int, argv[4:])
1348
lines = sys.stdin.readlines()
1349
ver = w.add(name, parents, lines)
1350
write_weave(w, file(argv[2], 'wb'))
1351
print 'added version %r %d' % (name, ver)
1354
if os.path.exists(fn):
1355
raise IOError("file exists")
1357
write_weave(w, file(fn, 'wb'))
1358
elif cmd == 'get': # get one version
1360
sys.stdout.writelines(w.get_iter(int(argv[3])))
1365
v1, v2 = map(int, argv[3:5])
1368
diff_gen = bzrlib.patiencediff.unified_diff(lines1, lines2,
1369
'%s version %d' % (fn, v1),
1370
'%s version %d' % (fn, v2))
1371
sys.stdout.writelines(diff_gen)
1373
elif cmd == 'annotate':
1375
# newline is added to all lines regardless; too hard to get
1376
# reasonable formatting otherwise
1378
for origin, text in w.annotate(int(argv[3])):
1379
text = text.rstrip('\r\n')
1381
print ' | %s' % (text)
1383
print '%5d | %s' % (origin, text)
1389
elif cmd == 'stats':
1390
weave_stats(argv[2], ProgressBar())
1392
elif cmd == 'check':
1397
print '%d versions ok' % w.num_versions()
1399
elif cmd == 'inclusions':
1401
print ' '.join(map(str, w.inclusions([int(argv[3])])))
1403
elif cmd == 'parents':
1405
print ' '.join(map(str, w._parents[int(argv[3])]))
1407
elif cmd == 'plan-merge':
1408
# replaced by 'bzr weave-plan-merge'
1410
for state, line in w.plan_merge(int(argv[3]), int(argv[4])):
1412
print '%14s | %s' % (state, line),
1413
elif cmd == 'merge':
1414
# replaced by 'bzr weave-merge-text'
1416
p = w.plan_merge(int(argv[3]), int(argv[4]))
1417
sys.stdout.writelines(w.weave_merge(p))
1419
raise ValueError('unknown command %r' % cmd)
1422
if __name__ == '__main__':
1424
sys.exit(main(sys.argv))
1427
class InterWeave(InterVersionedFile):
1428
"""Optimised code paths for weave to weave operations."""
1430
_matching_file_from_factory = staticmethod(WeaveFile)
1431
_matching_file_to_factory = staticmethod(WeaveFile)
1434
def is_compatible(source, target):
1435
"""Be compatible with weaves."""
1437
return (isinstance(source, Weave) and
1438
isinstance(target, Weave))
1439
except AttributeError:
1442
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
1443
"""See InterVersionedFile.join."""
1444
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
1445
if self.target.versions() == [] and version_ids is None:
1446
self.target._copy_weave_content(self.source)
1449
self.target._join(self.source, pb, msg, version_ids, ignore_missing)
1450
except errors.WeaveParentMismatch:
1451
self.target._reweave(self.source, pb, msg)
1454
InterVersionedFile.register_optimiser(InterWeave)
added
removed
bzrlib/weave.py
