56
45
from bzrlib.store.text import TextStore
57
from bzrlib.tuned_gzip import GzipFile, bytes_to_gzip
58
from bzrlib.versionedfile import (
60
FulltextContentFactory,
46
from bzrlib.trace import mutter
65
49
class AllInOneRepository(Repository):
66
50
"""Legacy support - the repository behaviour for all-in-one branches."""
69
def _serializer(self):
70
return xml5.serializer_v5
72
def _escape(self, file_or_path):
73
if not isinstance(file_or_path, basestring):
74
file_or_path = '/'.join(file_or_path)
75
if file_or_path == '':
77
return urlutils.escape(osutils.safe_unicode(file_or_path))
79
def __init__(self, _format, a_bzrdir):
80
# we reuse one control files instance.
52
_serializer = xml5.serializer_v5
54
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
81
55
dir_mode = a_bzrdir._get_dir_mode()
82
56
file_mode = a_bzrdir._get_file_mode()
84
58
def get_store(name, compressed=True, prefixed=False):
85
59
# FIXME: This approach of assuming stores are all entirely compressed
86
# or entirely uncompressed is tidy, but breaks upgrade from
87
# some existing branches where there's a mixture; we probably
60
# or entirely uncompressed is tidy, but breaks upgrade from
61
# some existing branches where there's a mixture; we probably
88
62
# still want the option to look for both.
89
relpath = self._escape(name)
63
relpath = a_bzrdir._control_files._escape(name)
90
64
store = TextStore(a_bzrdir.transport.clone(relpath),
91
65
prefixed=prefixed, compressed=compressed,
96
70
# not broken out yet because the controlweaves|inventory_store
97
# and texts bits are still different.
71
# and text_store | weave_store bits are still different.
98
72
if isinstance(_format, RepositoryFormat4):
99
# cannot remove these - there is still no consistent api
73
# cannot remove these - there is still no consistent api
100
74
# which allows access to this old info.
101
75
self.inventory_store = get_store('inventory-store')
102
self._text_store = get_store('text-store')
103
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files)
76
text_store = get_store('text-store')
77
super(AllInOneRepository, self).__init__(_format,
78
a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
79
if control_store is not None:
80
control_store.get_scope = self.get_transaction
81
text_store.get_scope = self.get_transaction
106
84
def _all_possible_ids(self):
107
85
"""Return all the possible revisions that we could find."""
108
86
if 'evil' in debug.debug_flags:
109
trace.mutter_callsite(
110
3, "_all_possible_ids scales with size of history.")
111
return [key[-1] for key in self.inventories.keys()]
87
mutter_callsite(3, "_all_possible_ids scales with size of history.")
88
return self.get_inventory_weave().versions()
114
91
def _all_revision_ids(self):
115
"""Returns a list of all the revision ids in the repository.
92
"""Returns a list of all the revision ids in the repository.
117
These are in as much topological order as the underlying store can
94
These are in as much topological order as the underlying store can
118
95
present: for weaves ghosts may lead to a lack of correctness until
119
96
the reweave updates the parents list.
121
return [key[-1] for key in self.revisions.keys()]
123
def _activate_new_inventory(self):
124
"""Put a replacement inventory.new into use as inventories."""
125
# Copy the content across
126
t = self.bzrdir._control_files._transport
127
t.copy('inventory.new.weave', 'inventory.weave')
128
# delete the temp inventory
129
t.delete('inventory.new.weave')
130
# Check we can parse the new weave properly as a sanity check
131
self.inventories.keys()
133
def _backup_inventory(self):
134
t = self.bzrdir._control_files._transport
135
t.copy('inventory.weave', 'inventory.backup.weave')
137
def _temp_inventories(self):
138
t = self.bzrdir._control_files._transport
139
return self._format._get_inventories(t, self, 'inventory.new')
98
if self._revision_store.text_store.listable():
99
return self._revision_store.all_revision_ids(self.get_transaction())
100
result = self._all_possible_ids()
101
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
102
# ids. (It should, since _revision_store's API should change to
103
# return utf8 revision_ids)
104
return self._eliminate_revisions_not_present(result)
106
def _check_revision_parents(self, revision, inventory):
107
"""Private to Repository and Fetch.
109
This checks the parentage of revision in an inventory weave for
110
consistency and is only applicable to inventory-weave-for-ancestry
111
using repository formats & fetchers.
113
weave_parents = inventory.get_parent_map(
114
[revision.revision_id])[revision.revision_id]
115
parent_map = inventory.get_parent_map(revision.parent_ids)
116
for parent_id in revision.parent_ids:
117
if parent_id in parent_map:
118
# this parent must not be a ghost.
119
if not parent_id in weave_parents:
121
raise errors.CorruptRepository(self)
141
123
def get_commit_builder(self, branch, parents, config, timestamp=None,
142
124
timezone=None, committer=None, revprops=None,
143
125
revision_id=None):
144
126
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
145
result = CommitBuilder(self, parents, config, timestamp, timezone,
127
result = WeaveCommitBuilder(self, parents, config, timestamp, timezone,
146
128
committer, revprops, revision_id)
147
129
self.start_write_group()
151
133
def get_revisions(self, revision_ids):
152
134
revs = self._get_revisions(revision_ids)
135
# weave corruption can lead to absent revision markers that should be
137
# the following test is reasonably cheap (it needs a single weave read)
138
# and the weave is cached in read transactions. In write transactions
139
# it is not cached but typically we only read a small number of
140
# revisions. For knits when they are introduced we will probably want
141
# to ensure that caching write transactions are in use.
142
inv = self.get_inventory_weave()
144
self._check_revision_parents(rev, inv)
155
def _inventory_add_lines(self, revision_id, parents, lines,
157
"""Store lines in inv_vf and return the sha1 of the inventory."""
158
present_parents = self.get_graph().get_parent_map(parents)
160
for parent in parents:
161
if parent in present_parents:
162
final_parents.append((parent,))
163
return self.inventories.add_lines((revision_id,), final_parents, lines,
164
check_content=check_content)[0]
147
def has_revisions(self, revision_ids):
148
"""See Repository.has_revisions()."""
150
transaction = self.get_transaction()
151
for revision_id in revision_ids:
152
if self._revision_store.has_revision_id(revision_id, transaction):
153
result.add(revision_id)
166
157
def is_shared(self):
167
158
"""AllInOne repositories cannot be shared."""
192
183
class WeaveMetaDirRepository(MetaDirVersionedFileRepository):
193
184
"""A subclass of MetaDirRepository to set weave specific policy."""
195
def __init__(self, _format, a_bzrdir, control_files):
196
super(WeaveMetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
197
self._serializer = _format._serializer
186
_serializer = xml5.serializer_v5
200
189
def _all_possible_ids(self):
201
190
"""Return all the possible revisions that we could find."""
202
191
if 'evil' in debug.debug_flags:
203
trace.mutter_callsite(
204
3, "_all_possible_ids scales with size of history.")
205
return [key[-1] for key in self.inventories.keys()]
192
mutter_callsite(3, "_all_possible_ids scales with size of history.")
193
return self.get_inventory_weave().versions()
208
196
def _all_revision_ids(self):
209
"""Returns a list of all the revision ids in the repository.
197
"""Returns a list of all the revision ids in the repository.
211
These are in as much topological order as the underlying store can
199
These are in as much topological order as the underlying store can
212
200
present: for weaves ghosts may lead to a lack of correctness until
213
201
the reweave updates the parents list.
215
return [key[-1] for key in self.revisions.keys()]
217
def _activate_new_inventory(self):
218
"""Put a replacement inventory.new into use as inventories."""
219
# Copy the content across
221
t.copy('inventory.new.weave', 'inventory.weave')
222
# delete the temp inventory
223
t.delete('inventory.new.weave')
224
# Check we can parse the new weave properly as a sanity check
225
self.inventories.keys()
227
def _backup_inventory(self):
229
t.copy('inventory.weave', 'inventory.backup.weave')
231
def _temp_inventories(self):
233
return self._format._get_inventories(t, self, 'inventory.new')
203
if self._revision_store.text_store.listable():
204
return self._revision_store.all_revision_ids(self.get_transaction())
205
result = self._all_possible_ids()
206
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
207
# ids. (It should, since _revision_store's API should change to
208
# return utf8 revision_ids)
209
return self._eliminate_revisions_not_present(result)
211
def _check_revision_parents(self, revision, inventory):
212
"""Private to Repository and Fetch.
214
This checks the parentage of revision in an inventory weave for
215
consistency and is only applicable to inventory-weave-for-ancestry
216
using repository formats & fetchers.
218
weave_parents = inventory.get_parent_map(
219
[revision.revision_id])[revision.revision_id]
220
parent_map = inventory.get_parent_map(revision.parent_ids)
221
for parent_id in revision.parent_ids:
222
if parent_id in parent_map:
223
# this parent must not be a ghost.
224
if not parent_id in weave_parents:
226
raise errors.CorruptRepository(self)
235
228
def get_commit_builder(self, branch, parents, config, timestamp=None,
236
229
timezone=None, committer=None, revprops=None,
237
230
revision_id=None):
238
231
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
239
result = CommitBuilder(self, parents, config, timestamp, timezone,
232
result = WeaveCommitBuilder(self, parents, config, timestamp, timezone,
240
233
committer, revprops, revision_id)
241
234
self.start_write_group()
245
238
def get_revision(self, revision_id):
246
239
"""Return the Revision object for a named revision"""
240
# TODO: jam 20070210 get_revision_reconcile should do this for us
247
241
r = self.get_revision_reconcile(revision_id)
242
# weave corruption can lead to absent revision markers that should be
244
# the following test is reasonably cheap (it needs a single weave read)
245
# and the weave is cached in read transactions. In write transactions
246
# it is not cached but typically we only read a small number of
247
# revisions. For knits when they are introduced we will probably want
248
# to ensure that caching write transactions are in use.
249
inv = self.get_inventory_weave()
250
self._check_revision_parents(r, inv)
250
def _inventory_add_lines(self, revision_id, parents, lines,
252
"""Store lines in inv_vf and return the sha1 of the inventory."""
253
present_parents = self.get_graph().get_parent_map(parents)
255
for parent in parents:
256
if parent in present_parents:
257
final_parents.append((parent,))
258
return self.inventories.add_lines((revision_id,), final_parents, lines,
259
check_content=check_content)[0]
253
def has_revisions(self, revision_ids):
254
"""See Repository.has_revisions()."""
256
transaction = self.get_transaction()
257
for revision_id in revision_ids:
258
if self._revision_store.has_revision_id(revision_id, transaction):
259
result.add(revision_id)
261
262
def revision_graph_can_have_wrong_parents(self):
263
# XXX: This is an old format that we don't support full checking on, so
264
# just claim that checking for this inconsistency is not required.
351
365
"""Format 4 is not supported.
353
367
It is not supported because the model changed from 4 to 5 and the
354
conversion logic is expensive - so doing it on the fly was not
368
conversion logic is expensive - so doing it on the fly was not
359
def _get_inventories(self, repo_transport, repo, name='inventory'):
360
# No inventories store written so far.
373
def _get_control_store(self, repo_transport, control_files):
374
"""Format 4 repositories have no formal control store at this point.
376
This will cause any control-file-needing apis to fail - this is desired.
363
def _get_revisions(self, repo_transport, repo):
380
def _get_revision_store(self, repo_transport, control_files):
381
"""See RepositoryFormat._get_revision_store()."""
364
382
from bzrlib.xml4 import serializer_v4
365
return RevisionTextStore(repo_transport.clone('revision-store'),
366
serializer_v4, True, versionedfile.PrefixMapper(),
367
repo.is_locked, repo.is_write_locked)
369
def _get_signatures(self, repo_transport, repo):
370
return SignatureTextStore(repo_transport.clone('revision-store'),
371
False, versionedfile.PrefixMapper(),
372
repo.is_locked, repo.is_write_locked)
374
def _get_texts(self, repo_transport, repo):
383
return self._get_text_rev_store(repo_transport,
386
serializer=serializer_v4)
388
def _get_text_store(self, transport, control_files):
389
"""See RepositoryFormat._get_text_store()."""
378
392
class RepositoryFormat5(PreSplitOutRepositoryFormat):
387
401
_versionedfile_class = weave.WeaveFile
388
402
_matchingbzrdir = bzrdir.BzrDirFormat5()
390
def _serializer(self):
391
return xml5.serializer_v5
405
super(RepositoryFormat5, self).__init__()
393
407
def get_format_description(self):
394
408
"""See RepositoryFormat.get_format_description()."""
395
409
return "Weave repository format 5"
397
def network_name(self):
398
"""The network name for this format is the control dirs disk label."""
399
return self._matchingbzrdir.get_format_string()
401
def _get_inventories(self, repo_transport, repo, name='inventory'):
402
mapper = versionedfile.ConstantMapper(name)
403
return versionedfile.ThunkedVersionedFiles(repo_transport,
404
weave.WeaveFile, mapper, repo.is_locked)
406
def _get_revisions(self, repo_transport, repo):
407
return RevisionTextStore(repo_transport.clone('revision-store'),
408
xml5.serializer_v5, False, versionedfile.PrefixMapper(),
409
repo.is_locked, repo.is_write_locked)
411
def _get_signatures(self, repo_transport, repo):
412
return SignatureTextStore(repo_transport.clone('revision-store'),
413
False, versionedfile.PrefixMapper(),
414
repo.is_locked, repo.is_write_locked)
416
def _get_texts(self, repo_transport, repo):
417
mapper = versionedfile.PrefixMapper()
418
base_transport = repo_transport.clone('weaves')
419
return versionedfile.ThunkedVersionedFiles(base_transport,
420
weave.WeaveFile, mapper, repo.is_locked)
411
def _get_revision_store(self, repo_transport, control_files):
412
"""See RepositoryFormat._get_revision_store()."""
413
"""Return the revision store object for this a_bzrdir."""
414
return self._get_text_rev_store(repo_transport,
419
def _get_text_store(self, transport, control_files):
420
"""See RepositoryFormat._get_text_store()."""
421
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
423
424
class RepositoryFormat6(PreSplitOutRepositoryFormat):
432
433
_versionedfile_class = weave.WeaveFile
433
434
_matchingbzrdir = bzrdir.BzrDirFormat6()
435
def _serializer(self):
436
return xml5.serializer_v5
437
super(RepositoryFormat6, self).__init__()
438
439
def get_format_description(self):
439
440
"""See RepositoryFormat.get_format_description()."""
440
441
return "Weave repository format 6"
442
def network_name(self):
443
"""The network name for this format is the control dirs disk label."""
444
return self._matchingbzrdir.get_format_string()
446
def _get_inventories(self, repo_transport, repo, name='inventory'):
447
mapper = versionedfile.ConstantMapper(name)
448
return versionedfile.ThunkedVersionedFiles(repo_transport,
449
weave.WeaveFile, mapper, repo.is_locked)
451
def _get_revisions(self, repo_transport, repo):
452
return RevisionTextStore(repo_transport.clone('revision-store'),
453
xml5.serializer_v5, False, versionedfile.HashPrefixMapper(),
454
repo.is_locked, repo.is_write_locked)
456
def _get_signatures(self, repo_transport, repo):
457
return SignatureTextStore(repo_transport.clone('revision-store'),
458
False, versionedfile.HashPrefixMapper(),
459
repo.is_locked, repo.is_write_locked)
461
def _get_texts(self, repo_transport, repo):
462
mapper = versionedfile.HashPrefixMapper()
463
base_transport = repo_transport.clone('weaves')
464
return versionedfile.ThunkedVersionedFiles(base_transport,
465
weave.WeaveFile, mapper, repo.is_locked)
443
def _get_revision_store(self, repo_transport, control_files):
444
"""See RepositoryFormat._get_revision_store()."""
445
return self._get_text_rev_store(repo_transport,
451
def _get_text_store(self, transport, control_files):
452
"""See RepositoryFormat._get_text_store()."""
453
return self._get_versioned_file_store('weaves', transport, control_files)
468
455
class RepositoryFormat7(MetaDirRepositoryFormat):
469
456
"""Bzr repository 7.
496
482
"""See RepositoryFormat.get_format_description()."""
497
483
return "Weave repository format 7"
499
def _get_inventories(self, repo_transport, repo, name='inventory'):
500
mapper = versionedfile.ConstantMapper(name)
501
return versionedfile.ThunkedVersionedFiles(repo_transport,
502
weave.WeaveFile, mapper, repo.is_locked)
504
def _get_revisions(self, repo_transport, repo):
505
return RevisionTextStore(repo_transport.clone('revision-store'),
506
xml5.serializer_v5, True, versionedfile.HashPrefixMapper(),
507
repo.is_locked, repo.is_write_locked)
509
def _get_signatures(self, repo_transport, repo):
510
return SignatureTextStore(repo_transport.clone('revision-store'),
511
True, versionedfile.HashPrefixMapper(),
512
repo.is_locked, repo.is_write_locked)
514
def _get_texts(self, repo_transport, repo):
515
mapper = versionedfile.HashPrefixMapper()
516
base_transport = repo_transport.clone('weaves')
517
return versionedfile.ThunkedVersionedFiles(base_transport,
518
weave.WeaveFile, mapper, repo.is_locked)
485
def check_conversion_target(self, target_format):
488
def _get_revision_store(self, repo_transport, control_files):
489
"""See RepositoryFormat._get_revision_store()."""
490
return self._get_text_rev_store(repo_transport,
497
def _get_text_store(self, transport, control_files):
498
"""See RepositoryFormat._get_text_store()."""
499
return self._get_versioned_file_store('weaves',
520
503
def initialize(self, a_bzrdir, shared=False):
521
504
"""Create a weave repository.
552
535
repo_transport = a_bzrdir.get_repository_transport(None)
553
536
control_files = lockable_files.LockableFiles(repo_transport,
554
537
'lock', lockdir.LockDir)
555
result = WeaveMetaDirRepository(_format=self, a_bzrdir=a_bzrdir,
556
control_files=control_files)
557
result.revisions = self._get_revisions(repo_transport, result)
558
result.signatures = self._get_signatures(repo_transport, result)
559
result.inventories = self._get_inventories(repo_transport, result)
560
result.texts = self._get_texts(repo_transport, result)
561
result.chk_bytes = None
562
result._transport = repo_transport
566
class TextVersionedFiles(VersionedFiles):
567
"""Just-a-bunch-of-files based VersionedFile stores."""
569
def __init__(self, transport, compressed, mapper, is_locked, can_write):
570
self._compressed = compressed
571
self._transport = transport
572
self._mapper = mapper
577
self._is_locked = is_locked
578
self._can_write = can_write
580
def add_lines(self, key, parents, lines):
581
"""Add a revision to the store."""
582
if not self._is_locked():
583
raise errors.ObjectNotLocked(self)
584
if not self._can_write():
585
raise errors.ReadOnlyError(self)
587
raise ValueError('bad idea to put / in %r' % (key,))
588
text = ''.join(lines)
590
text = bytes_to_gzip(text)
591
path = self._map(key)
592
self._transport.put_bytes_non_atomic(path, text, create_parent_dir=True)
594
def insert_record_stream(self, stream):
596
for record in stream:
597
# Raise an error when a record is missing.
598
if record.storage_kind == 'absent':
599
raise errors.RevisionNotPresent([record.key[0]], self)
600
# adapt to non-tuple interface
601
if record.storage_kind == 'fulltext':
602
self.add_lines(record.key, None,
603
osutils.split_lines(record.get_bytes_as('fulltext')))
605
adapter_key = record.storage_kind, 'fulltext'
607
adapter = adapters[adapter_key]
609
adapter_factory = adapter_registry.get(adapter_key)
610
adapter = adapter_factory(self)
611
adapters[adapter_key] = adapter
612
lines = osutils.split_lines(adapter.get_bytes(
613
record, record.get_bytes_as(record.storage_kind)))
615
self.add_lines(record.key, None, lines)
616
except RevisionAlreadyPresent:
619
def _load_text(self, key):
620
if not self._is_locked():
621
raise errors.ObjectNotLocked(self)
622
path = self._map(key)
624
text = self._transport.get_bytes(path)
625
compressed = self._compressed
626
except errors.NoSuchFile:
628
# try without the .gz
631
text = self._transport.get_bytes(path)
633
except errors.NoSuchFile:
638
text = GzipFile(mode='rb', fileobj=StringIO(text)).read()
642
return self._mapper.map(key) + self._ext
645
class RevisionTextStore(TextVersionedFiles):
646
"""Legacy thunk for format 4 repositories."""
648
def __init__(self, transport, serializer, compressed, mapper, is_locked,
650
"""Create a RevisionTextStore at transport with serializer."""
651
TextVersionedFiles.__init__(self, transport, compressed, mapper,
652
is_locked, can_write)
653
self._serializer = serializer
655
def _load_text_parents(self, key):
656
text = self._load_text(key)
659
parents = self._serializer.read_revision_from_string(text).parent_ids
660
return text, tuple((parent,) for parent in parents)
662
def get_parent_map(self, keys):
665
parents = self._load_text_parents(key)[1]
668
result[key] = parents
671
def get_known_graph_ancestry(self, keys):
672
"""Get a KnownGraph instance with the ancestry of keys."""
674
parent_map = self.get_parent_map(keys)
675
kg = _mod_graph.KnownGraph(parent_map)
678
def get_record_stream(self, keys, sort_order, include_delta_closure):
680
text, parents = self._load_text_parents(key)
682
yield AbsentContentFactory(key)
684
yield FulltextContentFactory(key, parents, None, text)
687
if not self._is_locked():
688
raise errors.ObjectNotLocked(self)
690
for quoted_relpath in self._transport.iter_files_recursive():
691
relpath = urllib.unquote(quoted_relpath)
692
path, ext = os.path.splitext(relpath)
695
if not relpath.endswith('.sig'):
696
relpaths.add(relpath)
697
paths = list(relpaths)
698
return set([self._mapper.unmap(path) for path in paths])
701
class SignatureTextStore(TextVersionedFiles):
702
"""Legacy thunk for format 4-7 repositories."""
704
def __init__(self, transport, compressed, mapper, is_locked, can_write):
705
TextVersionedFiles.__init__(self, transport, compressed, mapper,
706
is_locked, can_write)
707
self._ext = '.sig' + self._ext
709
def get_parent_map(self, keys):
712
text = self._load_text(key)
718
def get_record_stream(self, keys, sort_order, include_delta_closure):
720
text = self._load_text(key)
722
yield AbsentContentFactory(key)
724
yield FulltextContentFactory(key, None, None, text)
727
if not self._is_locked():
728
raise errors.ObjectNotLocked(self)
730
for quoted_relpath in self._transport.iter_files_recursive():
731
relpath = urllib.unquote(quoted_relpath)
732
path, ext = os.path.splitext(relpath)
735
if not relpath.endswith('.sig'):
737
relpaths.add(relpath[:-4])
738
paths = list(relpaths)
739
return set([self._mapper.unmap(path) for path in paths])
538
text_store = self._get_text_store(repo_transport, control_files)
539
control_store = self._get_control_store(repo_transport, control_files)
540
_revision_store = self._get_revision_store(repo_transport, control_files)
541
return WeaveMetaDirRepository(_format=self,
543
control_files=control_files,
544
_revision_store=_revision_store,
545
control_store=control_store,
546
text_store=text_store)
549
class WeaveCommitBuilder(CommitBuilder):
550
"""A builder for weave based repos that don't support ghosts."""
552
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
553
versionedfile = self.repository.weave_store.get_weave_or_empty(
554
file_id, self.repository.get_transaction())
555
result = versionedfile.add_lines(
556
self._new_revision_id, parents, new_lines,
557
nostore_sha=nostore_sha)[0:2]
741
561
_legacy_formats = [RepositoryFormat4(),
742
562
RepositoryFormat5(),
added
removed
bzrlib/repofmt/weaverepo.py
