71
62
# some existing branches where there's a mixture; we probably
72
63
# still want the option to look for both.
73
64
relpath = a_bzrdir._control_files._escape(name)
74
store = TextStore(a_bzrdir.transport.clone(relpath),
65
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
75
66
prefixed=prefixed, compressed=compressed,
77
68
file_mode=file_mode)
80
71
# not broken out yet because the controlweaves|inventory_store
81
# and texts bits are still different.
72
# and text_store | weave_store bits are still different.
82
73
if isinstance(_format, RepositoryFormat4):
83
74
# cannot remove these - there is still no consistent api
84
75
# which allows access to this old info.
85
76
self.inventory_store = get_store('inventory-store')
86
self._text_store = get_store('text-store')
87
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files)
88
self._fetch_order = 'topological'
89
self._fetch_reconcile = True
77
text_store = get_store('text-store')
78
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
92
81
def _all_possible_ids(self):
93
82
"""Return all the possible revisions that we could find."""
94
83
if 'evil' in debug.debug_flags:
95
84
mutter_callsite(3, "_all_possible_ids scales with size of history.")
96
return [key[-1] for key in self.inventories.keys()]
85
return self.get_inventory_weave().versions()
99
88
def _all_revision_ids(self):
103
92
present: for weaves ghosts may lead to a lack of correctness until
104
93
the reweave updates the parents list.
106
return [key[-1] for key in self.revisions.keys()]
108
def _activate_new_inventory(self):
109
"""Put a replacement inventory.new into use as inventories."""
110
# Copy the content across
111
t = self.bzrdir._control_files._transport
112
t.copy('inventory.new.weave', 'inventory.weave')
113
# delete the temp inventory
114
t.delete('inventory.new.weave')
115
# Check we can parse the new weave properly as a sanity check
116
self.inventories.keys()
118
def _backup_inventory(self):
119
t = self.bzrdir._control_files._transport
120
t.copy('inventory.weave', 'inventory.backup.weave')
122
def _temp_inventories(self):
123
t = self.bzrdir._control_files._transport
124
return self._format._get_inventories(t, self, 'inventory.new')
95
if self._revision_store.text_store.listable():
96
return self._revision_store.all_revision_ids(self.get_transaction())
97
result = self._all_possible_ids()
98
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
99
# ids. (It should, since _revision_store's API should change to
100
# return utf8 revision_ids)
101
return self._eliminate_revisions_not_present(result)
103
def _check_revision_parents(self, revision, inventory):
104
"""Private to Repository and Fetch.
106
This checks the parentage of revision in an inventory weave for
107
consistency and is only applicable to inventory-weave-for-ancestry
108
using repository formats & fetchers.
110
weave_parents = inventory.get_parents(revision.revision_id)
111
weave_names = inventory.versions()
112
for parent_id in revision.parent_ids:
113
if parent_id in weave_names:
114
# this parent must not be a ghost.
115
if not parent_id in weave_parents:
117
raise errors.CorruptRepository(self)
126
119
def get_commit_builder(self, branch, parents, config, timestamp=None,
127
120
timezone=None, committer=None, revprops=None,
128
121
revision_id=None):
129
122
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
130
result = CommitBuilder(self, parents, config, timestamp, timezone,
123
result = WeaveCommitBuilder(self, parents, config, timestamp, timezone,
131
124
committer, revprops, revision_id)
132
125
self.start_write_group()
136
129
def get_revisions(self, revision_ids):
137
130
revs = self._get_revisions(revision_ids)
131
# weave corruption can lead to absent revision markers that should be
133
# the following test is reasonably cheap (it needs a single weave read)
134
# and the weave is cached in read transactions. In write transactions
135
# it is not cached but typically we only read a small number of
136
# revisions. For knits when they are introduced we will probably want
137
# to ensure that caching write transactions are in use.
138
inv = self.get_inventory_weave()
140
self._check_revision_parents(rev, inv)
140
def _inventory_add_lines(self, revision_id, parents, lines,
142
"""Store lines in inv_vf and return the sha1 of the inventory."""
143
present_parents = self.get_graph().get_parent_map(parents)
145
for parent in parents:
146
if parent in present_parents:
147
final_parents.append((parent,))
148
return self.inventories.add_lines((revision_id,), final_parents, lines,
149
check_content=check_content)[0]
144
def get_revision_graph(self, revision_id=None):
145
"""Return a dictionary containing the revision graph.
147
:param revision_id: The revision_id to get a graph from. If None, then
148
the entire revision graph is returned. This is a deprecated mode of
149
operation and will be removed in the future.
150
:return: a dictionary of revision_id->revision_parents_list.
152
if 'evil' in debug.debug_flags:
154
"get_revision_graph scales with size of history.")
155
# special case NULL_REVISION
156
if revision_id == _mod_revision.NULL_REVISION:
158
a_weave = self.get_inventory_weave()
159
all_revisions = self._eliminate_revisions_not_present(
161
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
162
node in all_revisions])
163
if revision_id is None:
165
elif revision_id not in entire_graph:
166
raise errors.NoSuchRevision(self, revision_id)
168
# add what can be reached from revision_id
170
pending = set([revision_id])
171
while len(pending) > 0:
173
result[node] = entire_graph[node]
174
for revision_id in result[node]:
175
if revision_id not in result:
176
pending.add(revision_id)
151
180
def is_shared(self):
152
181
"""AllInOne repositories cannot be shared."""
199
223
present: for weaves ghosts may lead to a lack of correctness until
200
224
the reweave updates the parents list.
202
return [key[-1] for key in self.revisions.keys()]
204
def _activate_new_inventory(self):
205
"""Put a replacement inventory.new into use as inventories."""
206
# Copy the content across
208
t.copy('inventory.new.weave', 'inventory.weave')
209
# delete the temp inventory
210
t.delete('inventory.new.weave')
211
# Check we can parse the new weave properly as a sanity check
212
self.inventories.keys()
214
def _backup_inventory(self):
216
t.copy('inventory.weave', 'inventory.backup.weave')
218
def _temp_inventories(self):
220
return self._format._get_inventories(t, self, 'inventory.new')
226
if self._revision_store.text_store.listable():
227
return self._revision_store.all_revision_ids(self.get_transaction())
228
result = self._all_possible_ids()
229
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
230
# ids. (It should, since _revision_store's API should change to
231
# return utf8 revision_ids)
232
return self._eliminate_revisions_not_present(result)
234
def _check_revision_parents(self, revision, inventory):
235
"""Private to Repository and Fetch.
237
This checks the parentage of revision in an inventory weave for
238
consistency and is only applicable to inventory-weave-for-ancestry
239
using repository formats & fetchers.
241
weave_parents = inventory.get_parents(revision.revision_id)
242
weave_names = inventory.versions()
243
for parent_id in revision.parent_ids:
244
if parent_id in weave_names:
245
# this parent must not be a ghost.
246
if not parent_id in weave_parents:
248
raise errors.CorruptRepository(self)
222
250
def get_commit_builder(self, branch, parents, config, timestamp=None,
223
251
timezone=None, committer=None, revprops=None,
224
252
revision_id=None):
225
253
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
226
result = CommitBuilder(self, parents, config, timestamp, timezone,
254
result = WeaveCommitBuilder(self, parents, config, timestamp, timezone,
227
255
committer, revprops, revision_id)
228
256
self.start_write_group()
232
260
def get_revision(self, revision_id):
233
261
"""Return the Revision object for a named revision"""
262
# TODO: jam 20070210 get_revision_reconcile should do this for us
234
263
r = self.get_revision_reconcile(revision_id)
264
# weave corruption can lead to absent revision markers that should be
266
# the following test is reasonably cheap (it needs a single weave read)
267
# and the weave is cached in read transactions. In write transactions
268
# it is not cached but typically we only read a small number of
269
# revisions. For knits when they are introduced we will probably want
270
# to ensure that caching write transactions are in use.
271
inv = self.get_inventory_weave()
272
self._check_revision_parents(r, inv)
237
def _inventory_add_lines(self, revision_id, parents, lines,
239
"""Store lines in inv_vf and return the sha1 of the inventory."""
240
present_parents = self.get_graph().get_parent_map(parents)
242
for parent in parents:
243
if parent in present_parents:
244
final_parents.append((parent,))
245
return self.inventories.add_lines((revision_id,), final_parents, lines,
246
check_content=check_content)[0]
276
def get_revision_graph(self, revision_id=None):
277
"""Return a dictionary containing the revision graph.
279
:param revision_id: The revision_id to get a graph from. If None, then
280
the entire revision graph is returned. This is a deprecated mode of
281
operation and will be removed in the future.
282
:return: a dictionary of revision_id->revision_parents_list.
284
if 'evil' in debug.debug_flags:
286
"get_revision_graph scales with size of history.")
287
# special case NULL_REVISION
288
if revision_id == _mod_revision.NULL_REVISION:
290
a_weave = self.get_inventory_weave()
291
all_revisions = self._eliminate_revisions_not_present(
293
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
294
node in all_revisions])
295
if revision_id is None:
297
elif revision_id not in entire_graph:
298
raise errors.NoSuchRevision(self, revision_id)
300
# add what can be reached from revision_id
302
pending = set([revision_id])
303
while len(pending) > 0:
305
result[node] = entire_graph[node]
306
for revision_id in result[node]:
307
if revision_id not in result:
308
pending.add(revision_id)
248
311
def revision_graph_can_have_wrong_parents(self):
312
# XXX: This is an old format that we don't support full checking on, so
313
# just claim that checking for this inconsistency is not required.
272
336
empty_weave = sio.getvalue()
274
338
mutter('creating repository in %s.', a_bzrdir.transport.base)
339
dirs = ['revision-store', 'weaves']
340
files = [('inventory.weave', StringIO(empty_weave)),
276
343
# FIXME: RBC 20060125 don't peek under the covers
277
344
# NB: no need to escape relative paths that are url safe.
278
345
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
279
'branch-lock', lockable_files.TransportLock)
346
'branch-lock', lockable_files.TransportLock)
280
347
control_files.create_lock()
281
348
control_files.lock_write()
282
transport = a_bzrdir.transport
349
control_files._transport.mkdir_multi(dirs,
350
mode=control_files._dir_mode)
284
transport.mkdir_multi(['revision-store', 'weaves'],
285
mode=a_bzrdir._get_dir_mode())
286
transport.put_bytes_non_atomic('inventory.weave', empty_weave)
352
for file, content in files:
353
control_files.put(file, content)
288
355
control_files.unlock()
289
356
return self.open(a_bzrdir, _found=True)
358
def _get_control_store(self, repo_transport, control_files):
359
"""Return the control store for this repository."""
360
return self._get_versioned_file_store('',
365
def _get_text_store(self, transport, control_files):
366
"""Get a store for file texts for this format."""
367
raise NotImplementedError(self._get_text_store)
291
369
def open(self, a_bzrdir, _found=False):
292
370
"""See RepositoryFormat.open()."""
346
def _get_inventories(self, repo_transport, repo, name='inventory'):
347
# No inventories store written so far.
424
def _get_control_store(self, repo_transport, control_files):
425
"""Format 4 repositories have no formal control store at this point.
427
This will cause any control-file-needing apis to fail - this is desired.
350
def _get_revisions(self, repo_transport, repo):
431
def _get_revision_store(self, repo_transport, control_files):
432
"""See RepositoryFormat._get_revision_store()."""
351
433
from bzrlib.xml4 import serializer_v4
352
return RevisionTextStore(repo_transport.clone('revision-store'),
353
serializer_v4, True, versionedfile.PrefixMapper(),
354
repo.is_locked, repo.is_write_locked)
356
def _get_signatures(self, repo_transport, repo):
357
return SignatureTextStore(repo_transport.clone('revision-store'),
358
False, versionedfile.PrefixMapper(),
359
repo.is_locked, repo.is_write_locked)
361
def _get_texts(self, repo_transport, repo):
434
return self._get_text_rev_store(repo_transport,
437
serializer=serializer_v4)
439
def _get_text_store(self, transport, control_files):
440
"""See RepositoryFormat._get_text_store()."""
365
443
class RepositoryFormat5(PreSplitOutRepositoryFormat):
377
455
def __init__(self):
378
456
super(RepositoryFormat5, self).__init__()
379
self._fetch_order = 'topological'
380
self._fetch_reconcile = True
382
458
def get_format_description(self):
383
459
"""See RepositoryFormat.get_format_description()."""
384
460
return "Weave repository format 5"
386
def _get_inventories(self, repo_transport, repo, name='inventory'):
387
mapper = versionedfile.ConstantMapper(name)
388
return versionedfile.ThunkedVersionedFiles(repo_transport,
389
weave.WeaveFile, mapper, repo.is_locked)
391
def _get_revisions(self, repo_transport, repo):
392
from bzrlib.xml5 import serializer_v5
393
return RevisionTextStore(repo_transport.clone('revision-store'),
394
serializer_v5, False, versionedfile.PrefixMapper(),
395
repo.is_locked, repo.is_write_locked)
397
def _get_signatures(self, repo_transport, repo):
398
return SignatureTextStore(repo_transport.clone('revision-store'),
399
False, versionedfile.PrefixMapper(),
400
repo.is_locked, repo.is_write_locked)
402
def _get_texts(self, repo_transport, repo):
403
mapper = versionedfile.PrefixMapper()
404
base_transport = repo_transport.clone('weaves')
405
return versionedfile.ThunkedVersionedFiles(base_transport,
406
weave.WeaveFile, mapper, repo.is_locked)
462
def _get_revision_store(self, repo_transport, control_files):
463
"""See RepositoryFormat._get_revision_store()."""
464
"""Return the revision store object for this a_bzrdir."""
465
return self._get_text_rev_store(repo_transport,
470
def _get_text_store(self, transport, control_files):
471
"""See RepositoryFormat._get_text_store()."""
472
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
409
475
class RepositoryFormat6(PreSplitOutRepositoryFormat):
421
487
def __init__(self):
422
488
super(RepositoryFormat6, self).__init__()
423
self._fetch_order = 'topological'
424
self._fetch_reconcile = True
426
490
def get_format_description(self):
427
491
"""See RepositoryFormat.get_format_description()."""
428
492
return "Weave repository format 6"
430
def _get_inventories(self, repo_transport, repo, name='inventory'):
431
mapper = versionedfile.ConstantMapper(name)
432
return versionedfile.ThunkedVersionedFiles(repo_transport,
433
weave.WeaveFile, mapper, repo.is_locked)
435
def _get_revisions(self, repo_transport, repo):
436
from bzrlib.xml5 import serializer_v5
437
return RevisionTextStore(repo_transport.clone('revision-store'),
438
serializer_v5, False, versionedfile.HashPrefixMapper(),
439
repo.is_locked, repo.is_write_locked)
441
def _get_signatures(self, repo_transport, repo):
442
return SignatureTextStore(repo_transport.clone('revision-store'),
443
False, versionedfile.HashPrefixMapper(),
444
repo.is_locked, repo.is_write_locked)
446
def _get_texts(self, repo_transport, repo):
447
mapper = versionedfile.HashPrefixMapper()
448
base_transport = repo_transport.clone('weaves')
449
return versionedfile.ThunkedVersionedFiles(base_transport,
450
weave.WeaveFile, mapper, repo.is_locked)
494
def _get_revision_store(self, repo_transport, control_files):
495
"""See RepositoryFormat._get_revision_store()."""
496
return self._get_text_rev_store(repo_transport,
502
def _get_text_store(self, transport, control_files):
503
"""See RepositoryFormat._get_text_store()."""
504
return self._get_versioned_file_store('weaves', transport, control_files)
453
506
class RepositoryFormat7(MetaDirRepositoryFormat):
454
507
"""Bzr repository 7.
476
536
def check_conversion_target(self, target_format):
479
def _get_inventories(self, repo_transport, repo, name='inventory'):
480
mapper = versionedfile.ConstantMapper(name)
481
return versionedfile.ThunkedVersionedFiles(repo_transport,
482
weave.WeaveFile, mapper, repo.is_locked)
484
def _get_revisions(self, repo_transport, repo):
485
from bzrlib.xml5 import serializer_v5
486
return RevisionTextStore(repo_transport.clone('revision-store'),
487
serializer_v5, True, versionedfile.HashPrefixMapper(),
488
repo.is_locked, repo.is_write_locked)
490
def _get_signatures(self, repo_transport, repo):
491
return SignatureTextStore(repo_transport.clone('revision-store'),
492
True, versionedfile.HashPrefixMapper(),
493
repo.is_locked, repo.is_write_locked)
495
def _get_texts(self, repo_transport, repo):
496
mapper = versionedfile.HashPrefixMapper()
497
base_transport = repo_transport.clone('weaves')
498
return versionedfile.ThunkedVersionedFiles(base_transport,
499
weave.WeaveFile, mapper, repo.is_locked)
539
def _get_revision_store(self, repo_transport, control_files):
540
"""See RepositoryFormat._get_revision_store()."""
541
return self._get_text_rev_store(repo_transport,
548
def _get_text_store(self, transport, control_files):
549
"""See RepositoryFormat._get_text_store()."""
550
return self._get_versioned_file_store('weaves',
501
554
def initialize(self, a_bzrdir, shared=False):
502
555
"""Create a weave repository.
529
582
format = RepositoryFormat.find_format(a_bzrdir)
583
assert format.__class__ == self.__class__
530
584
if _override_transport is not None:
531
585
repo_transport = _override_transport
533
587
repo_transport = a_bzrdir.get_repository_transport(None)
534
588
control_files = lockable_files.LockableFiles(repo_transport,
535
589
'lock', lockdir.LockDir)
536
result = WeaveMetaDirRepository(_format=self, a_bzrdir=a_bzrdir,
537
control_files=control_files)
538
result.revisions = self._get_revisions(repo_transport, result)
539
result.signatures = self._get_signatures(repo_transport, result)
540
result.inventories = self._get_inventories(repo_transport, result)
541
result.texts = self._get_texts(repo_transport, result)
542
result._transport = repo_transport
546
class TextVersionedFiles(VersionedFiles):
547
"""Just-a-bunch-of-files based VersionedFile stores."""
549
def __init__(self, transport, compressed, mapper, is_locked, can_write):
550
self._compressed = compressed
551
self._transport = transport
552
self._mapper = mapper
557
self._is_locked = is_locked
558
self._can_write = can_write
560
def add_lines(self, key, parents, lines):
561
"""Add a revision to the store."""
562
if not self._is_locked():
563
raise errors.ObjectNotLocked(self)
564
if not self._can_write():
565
raise errors.ReadOnlyError(self)
567
raise ValueError('bad idea to put / in %r' % (key,))
568
text = ''.join(lines)
570
text = bytes_to_gzip(text)
571
path = self._map(key)
572
self._transport.put_bytes_non_atomic(path, text, create_parent_dir=True)
574
def insert_record_stream(self, stream):
576
for record in stream:
577
# Raise an error when a record is missing.
578
if record.storage_kind == 'absent':
579
raise errors.RevisionNotPresent([record.key[0]], self)
580
# adapt to non-tuple interface
581
if record.storage_kind == 'fulltext':
582
self.add_lines(record.key, None,
583
osutils.split_lines(record.get_bytes_as('fulltext')))
585
adapter_key = record.storage_kind, 'fulltext'
587
adapter = adapters[adapter_key]
589
adapter_factory = adapter_registry.get(adapter_key)
590
adapter = adapter_factory(self)
591
adapters[adapter_key] = adapter
592
lines = osutils.split_lines(adapter.get_bytes(
593
record, record.get_bytes_as(record.storage_kind)))
595
self.add_lines(record.key, None, lines)
596
except RevisionAlreadyPresent:
599
def _load_text(self, key):
600
if not self._is_locked():
601
raise errors.ObjectNotLocked(self)
602
path = self._map(key)
604
text = self._transport.get_bytes(path)
605
compressed = self._compressed
606
except errors.NoSuchFile:
608
# try without the .gz
611
text = self._transport.get_bytes(path)
613
except errors.NoSuchFile:
618
text = GzipFile(mode='rb', fileobj=StringIO(text)).read()
622
return self._mapper.map(key) + self._ext
625
class RevisionTextStore(TextVersionedFiles):
626
"""Legacy thunk for format 4 repositories."""
628
def __init__(self, transport, serializer, compressed, mapper, is_locked,
630
"""Create a RevisionTextStore at transport with serializer."""
631
TextVersionedFiles.__init__(self, transport, compressed, mapper,
632
is_locked, can_write)
633
self._serializer = serializer
635
def _load_text_parents(self, key):
636
text = self._load_text(key)
639
parents = self._serializer.read_revision_from_string(text).parent_ids
640
return text, tuple((parent,) for parent in parents)
642
def get_parent_map(self, keys):
645
parents = self._load_text_parents(key)[1]
648
result[key] = parents
651
def get_record_stream(self, keys, sort_order, include_delta_closure):
653
text, parents = self._load_text_parents(key)
655
yield AbsentContentFactory(key)
657
yield FulltextContentFactory(key, parents, None, text)
660
if not self._is_locked():
661
raise errors.ObjectNotLocked(self)
663
for quoted_relpath in self._transport.iter_files_recursive():
664
relpath = urllib.unquote(quoted_relpath)
665
path, ext = os.path.splitext(relpath)
668
if '.sig' not in relpath:
669
relpaths.add(relpath)
670
paths = list(relpaths)
671
return set([self._mapper.unmap(path) for path in paths])
674
class SignatureTextStore(TextVersionedFiles):
675
"""Legacy thunk for format 4-7 repositories."""
677
def __init__(self, transport, compressed, mapper, is_locked, can_write):
678
TextVersionedFiles.__init__(self, transport, compressed, mapper,
679
is_locked, can_write)
680
self._ext = '.sig' + self._ext
682
def get_parent_map(self, keys):
685
text = self._load_text(key)
691
def get_record_stream(self, keys, sort_order, include_delta_closure):
693
text = self._load_text(key)
695
yield AbsentContentFactory(key)
697
yield FulltextContentFactory(key, None, None, text)
700
if not self._is_locked():
701
raise errors.ObjectNotLocked(self)
703
for quoted_relpath in self._transport.iter_files_recursive():
704
relpath = urllib.unquote(quoted_relpath)
705
path, ext = os.path.splitext(relpath)
708
if not relpath.endswith('.sig'):
710
relpaths.add(relpath[:-4])
711
paths = list(relpaths)
712
return set([self._mapper.unmap(path) for path in paths])
590
text_store = self._get_text_store(repo_transport, control_files)
591
control_store = self._get_control_store(repo_transport, control_files)
592
_revision_store = self._get_revision_store(repo_transport, control_files)
593
return WeaveMetaDirRepository(_format=self,
595
control_files=control_files,
596
_revision_store=_revision_store,
597
control_store=control_store,
598
text_store=text_store)
601
class WeaveCommitBuilder(CommitBuilder):
602
"""A builder for weave based repos that don't support ghosts."""
604
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
605
versionedfile = self.repository.weave_store.get_weave_or_empty(
606
file_id, self.repository.get_transaction())
607
result = versionedfile.add_lines(
608
self._new_revision_id, parents, new_lines,
609
nostore_sha=nostore_sha)[0:2]
610
versionedfile.clear_cache()
714
614
_legacy_formats = [RepositoryFormat4(),
715
615
RepositoryFormat5(),
added
removed
bzrlib/repofmt/weaverepo.py
