13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
from bzrlib.lazy_import import lazy_import
18
lazy_import(globals(), """
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21
17
from bzrlib import (
28
revision as _mod_revision,
37
30
from bzrlib.decorators import needs_read_lock, needs_write_lock
38
31
from bzrlib.repository import (
33
MetaDirRepositoryFormat,
43
from bzrlib.vf_repository import (
44
InterSameDataRepository,
45
MetaDirVersionedFileRepository,
46
MetaDirVersionedFileRepositoryFormat,
47
VersionedFileCommitBuilder,
48
VersionedFileRootCommitBuilder,
50
from bzrlib import symbol_versioning
37
import bzrlib.revision as _mod_revision
38
from bzrlib.store.versioned import VersionedFileStore
39
from bzrlib.trace import mutter, note, warning
53
42
class _KnitParentsProvider(object):
58
47
def __repr__(self):
59
48
return 'KnitParentsProvider(%r)' % self._knit
61
def get_parent_map(self, keys):
62
"""See graph.StackedParentsProvider.get_parent_map"""
64
for revision_id in keys:
65
if revision_id is None:
66
raise ValueError('get_parent_map(None) is not valid')
50
def get_parents(self, revision_ids):
52
for revision_id in revision_ids:
67
53
if revision_id == _mod_revision.NULL_REVISION:
68
parent_map[revision_id] = ()
72
self._knit.get_parents_with_ghosts(revision_id))
57
parents = self._knit.get_parents_with_ghosts(revision_id)
73
58
except errors.RevisionNotPresent:
76
61
if len(parents) == 0:
77
parents = (_mod_revision.NULL_REVISION,)
78
parent_map[revision_id] = parents
82
class _KnitsParentsProvider(object):
84
def __init__(self, knit, prefix=()):
85
"""Create a parent provider for string keys mapped to tuple keys."""
90
return 'KnitsParentsProvider(%r)' % self._knit
92
def get_parent_map(self, keys):
93
"""See graph.StackedParentsProvider.get_parent_map"""
94
parent_map = self._knit.get_parent_map(
95
[self._prefix + (key,) for key in keys])
97
for key, parents in parent_map.items():
100
parents = (_mod_revision.NULL_REVISION,)
102
parents = tuple(parent[-1] for parent in parents)
103
result[revid] = parents
104
for revision_id in keys:
105
if revision_id == _mod_revision.NULL_REVISION:
106
result[revision_id] = ()
110
class KnitRepository(MetaDirVersionedFileRepository):
62
parents = [_mod_revision.NULL_REVISION]
63
parents_list.append(parents)
67
class KnitRepository(MetaDirRepository):
111
68
"""Knit format repository."""
113
# These attributes are inherited from the Repository base class. Setting
114
# them to None ensures that if the constructor is changed to not initialize
115
# them, or a subclass fails to call the constructor, that an error will
116
# occur rather than the system working but generating incorrect data.
117
_commit_builder_class = None
120
def __init__(self, _format, a_bzrdir, control_files, _commit_builder_class,
122
super(KnitRepository, self).__init__(_format, a_bzrdir, control_files)
123
self._commit_builder_class = _commit_builder_class
124
self._serializer = _serializer
125
self._reconcile_fixes_text_parents = True
70
_serializer = xml5.serializer_v5
72
def _warn_if_deprecated(self):
73
# This class isn't deprecated
76
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
77
inv_vf.add_lines_with_ghosts(revid, parents, lines)
128
80
def _all_revision_ids(self):
129
81
"""See Repository.all_revision_ids()."""
130
return [key[0] for key in self.revisions.keys()]
132
def _activate_new_inventory(self):
133
"""Put a replacement inventory.new into use as inventories."""
134
# Copy the content across
136
t.copy('inventory.new.kndx', 'inventory.kndx')
138
t.copy('inventory.new.knit', 'inventory.knit')
139
except errors.NoSuchFile:
140
# empty inventories knit
141
t.delete('inventory.knit')
142
# delete the temp inventory
143
t.delete('inventory.new.kndx')
145
t.delete('inventory.new.knit')
146
except errors.NoSuchFile:
147
# empty inventories knit
149
# Force index reload (sanity check)
150
self.inventories._index._reset_cache()
151
self.inventories.keys()
153
def _backup_inventory(self):
155
t.copy('inventory.kndx', 'inventory.backup.kndx')
156
t.copy('inventory.knit', 'inventory.backup.knit')
158
def _move_file_id(self, from_id, to_id):
159
t = self._transport.clone('knits')
160
from_rel_url = self.texts._index._mapper.map((from_id, None))
161
to_rel_url = self.texts._index._mapper.map((to_id, None))
162
# We expect both files to always exist in this case.
163
for suffix in ('.knit', '.kndx'):
164
t.rename(from_rel_url + suffix, to_rel_url + suffix)
166
def _remove_file_id(self, file_id):
167
t = self._transport.clone('knits')
168
rel_url = self.texts._index._mapper.map((file_id, None))
169
for suffix in ('.kndx', '.knit'):
171
t.delete(rel_url + suffix)
172
except errors.NoSuchFile:
175
def _temp_inventories(self):
176
result = self._format._get_inventories(self._transport, self,
178
# Reconciling when the output has no revisions would result in no
179
# writes - but we want to ensure there is an inventory for
180
# compatibility with older clients that don't lazy-load.
181
result.get_parent_map([('A',)])
82
# Knits get the revision graph from the index of the revision knit, so
83
# it's always possible even if they're on an unlistable transport.
84
return self._revision_store.all_revision_ids(self.get_transaction())
86
def fileid_involved_between_revs(self, from_revid, to_revid):
87
"""Find file_id(s) which are involved in the changes between revisions.
89
This determines the set of revisions which are involved, and then
90
finds all file ids affected by those revisions.
92
from_revid = osutils.safe_revision_id(from_revid)
93
to_revid = osutils.safe_revision_id(to_revid)
94
vf = self._get_revision_vf()
95
from_set = set(vf.get_ancestry(from_revid))
96
to_set = set(vf.get_ancestry(to_revid))
97
changed = to_set.difference(from_set)
98
return self._fileid_involved_by_set(changed)
100
def fileid_involved(self, last_revid=None):
101
"""Find all file_ids modified in the ancestry of last_revid.
103
:param last_revid: If None, last_revision() will be used.
106
changed = set(self.all_revision_ids())
108
changed = set(self.get_ancestry(last_revid))
111
return self._fileid_involved_by_set(changed)
114
def get_ancestry(self, revision_id, topo_sorted=True):
115
"""Return a list of revision-ids integrated by a revision.
117
This is topologically sorted, unless 'topo_sorted' is specified as
120
if _mod_revision.is_null(revision_id):
122
revision_id = osutils.safe_revision_id(revision_id)
123
vf = self._get_revision_vf()
125
return [None] + vf.get_ancestry(revision_id, topo_sorted)
126
except errors.RevisionNotPresent:
127
raise errors.NoSuchRevision(self, revision_id)
185
130
def get_revision(self, revision_id):
187
132
revision_id = osutils.safe_revision_id(revision_id)
188
133
return self.get_revision_reconcile(revision_id)
190
def _refresh_data(self):
191
if not self.is_locked():
193
if self.is_in_write_group():
194
raise IsInWriteGroupError(self)
195
# Create a new transaction to force all knits to see the scope change.
196
# This is safe because we're outside a write group.
197
self.control_files._finish_transaction()
198
if self.is_write_locked():
199
self.control_files._set_write_transaction()
201
self.control_files._set_read_transaction()
136
def get_revision_graph(self, revision_id=None):
137
"""Return a dictionary containing the revision graph.
139
:param revision_id: The revision_id to get a graph from. If None, then
140
the entire revision graph is returned. This is a deprecated mode of
141
operation and will be removed in the future.
142
:return: a dictionary of revision_id->revision_parents_list.
144
# special case NULL_REVISION
145
if revision_id == _mod_revision.NULL_REVISION:
147
revision_id = osutils.safe_revision_id(revision_id)
148
a_weave = self._get_revision_vf()
149
if revision_id is None:
150
return a_weave.get_graph()
151
elif revision_id not in a_weave:
152
raise errors.NoSuchRevision(self, revision_id)
154
# add what can be reached from revision_id
155
return a_weave.get_graph([revision_id])
158
def get_revision_graph_with_ghosts(self, revision_ids=None):
159
"""Return a graph of the revisions with ghosts marked as applicable.
161
:param revision_ids: an iterable of revisions to graph or None for all.
162
:return: a Graph object with the graph reachable from revision_ids.
164
result = deprecated_graph.Graph()
165
vf = self._get_revision_vf()
166
versions = set(vf.versions())
168
pending = set(self.all_revision_ids())
171
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
172
# special case NULL_REVISION
173
if _mod_revision.NULL_REVISION in pending:
174
pending.remove(_mod_revision.NULL_REVISION)
175
required = set(pending)
178
revision_id = pending.pop()
179
if not revision_id in versions:
180
if revision_id in required:
181
raise errors.NoSuchRevision(self, revision_id)
183
result.add_ghost(revision_id)
184
# mark it as done so we don't try for it again.
185
done.add(revision_id)
187
parent_ids = vf.get_parents_with_ghosts(revision_id)
188
for parent_id in parent_ids:
189
# is this queued or done ?
190
if (parent_id not in pending and
191
parent_id not in done):
193
pending.add(parent_id)
194
result.add_node(revision_id, parent_ids)
195
done.add(revision_id)
198
def _get_revision_vf(self):
199
""":return: a versioned file containing the revisions."""
200
vf = self._revision_store.get_revision_file(self.get_transaction())
203
def _get_history_vf(self):
204
"""Get a versionedfile whose history graph reflects all revisions.
206
For knit repositories, this is the revision knit.
208
return self._get_revision_vf()
203
210
@needs_write_lock
204
211
def reconcile(self, other=None, thorough=False):
207
214
reconciler = KnitReconciler(self, thorough=thorough)
208
215
reconciler.reconcile()
209
216
return reconciler
218
def revision_parents(self, revision_id):
219
revision_id = osutils.safe_revision_id(revision_id)
220
return self._get_revision_vf().get_parents(revision_id)
211
222
def _make_parents_provider(self):
212
return _KnitsParentsProvider(self.revisions)
215
class RepositoryFormatKnit(MetaDirVersionedFileRepositoryFormat):
216
"""Bzr repository knit format (generalized).
223
return _KnitParentsProvider(self._get_revision_vf())
226
class KnitRepository3(KnitRepository):
228
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
229
control_store, text_store):
230
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
231
_revision_store, control_store, text_store)
232
self._serializer = xml7.serializer_v7
234
def deserialise_inventory(self, revision_id, xml):
235
"""Transform the xml into an inventory object.
237
:param revision_id: The expected revision id of the inventory.
238
:param xml: A serialised inventory.
240
result = self._serializer.read_inventory_from_string(xml)
241
assert result.root.revision is not None
244
def serialise_inventory(self, inv):
245
"""Transform the inventory object into XML text.
247
:param revision_id: The expected revision id of the inventory.
248
:param xml: A serialised inventory.
250
assert inv.revision_id is not None
251
assert inv.root.revision is not None
252
return KnitRepository.serialise_inventory(self, inv)
254
def get_commit_builder(self, branch, parents, config, timestamp=None,
255
timezone=None, committer=None, revprops=None,
257
"""Obtain a CommitBuilder for this repository.
259
:param branch: Branch to commit to.
260
:param parents: Revision ids of the parents of the new revision.
261
:param config: Configuration to use.
262
:param timestamp: Optional timestamp recorded for commit.
263
:param timezone: Optional timezone for timestamp.
264
:param committer: Optional committer to set for commit.
265
:param revprops: Optional dictionary of revision properties.
266
:param revision_id: Optional revision id.
268
revision_id = osutils.safe_revision_id(revision_id)
269
result = RootCommitBuilder(self, parents, config, timestamp, timezone,
270
committer, revprops, revision_id)
271
self.start_write_group()
275
class RepositoryFormatKnit(MetaDirRepositoryFormat):
276
"""Bzr repository knit format (generalized).
218
278
This repository format has:
219
279
- knits for file texts and inventory
229
# Set this attribute in derived classes to control the repository class
230
# created by open and initialize.
231
repository_class = None
232
# Set this attribute in derived classes to control the
233
# _commit_builder_class that the repository objects will have passed to
235
_commit_builder_class = None
236
# Set this attribute in derived clases to control the _serializer that the
237
# repository objects will have passed to their constructor.
239
def _serializer(self):
240
return xml5.serializer_v5
241
# Knit based repositories handle ghosts reasonably well.
242
supports_ghosts = True
243
# External lookups are not supported in this format.
244
supports_external_lookups = False
246
supports_chks = False
247
_fetch_order = 'topological'
248
_fetch_uses_deltas = True
250
supports_funky_characters = True
251
# The revision.kndx could potentially claim a revision has a different
252
# parent to the revision text.
253
revision_graph_can_have_wrong_parents = True
255
def _get_inventories(self, repo_transport, repo, name='inventory'):
256
mapper = versionedfile.ConstantMapper(name)
257
index = _mod_knit._KndxIndex(repo_transport, mapper,
258
repo.get_transaction, repo.is_write_locked, repo.is_locked)
259
access = _mod_knit._KnitKeyAccess(repo_transport, mapper)
260
return _mod_knit.KnitVersionedFiles(index, access, annotated=False)
262
def _get_revisions(self, repo_transport, repo):
263
mapper = versionedfile.ConstantMapper('revisions')
264
index = _mod_knit._KndxIndex(repo_transport, mapper,
265
repo.get_transaction, repo.is_write_locked, repo.is_locked)
266
access = _mod_knit._KnitKeyAccess(repo_transport, mapper)
267
return _mod_knit.KnitVersionedFiles(index, access, max_delta_chain=0,
270
def _get_signatures(self, repo_transport, repo):
271
mapper = versionedfile.ConstantMapper('signatures')
272
index = _mod_knit._KndxIndex(repo_transport, mapper,
273
repo.get_transaction, repo.is_write_locked, repo.is_locked)
274
access = _mod_knit._KnitKeyAccess(repo_transport, mapper)
275
return _mod_knit.KnitVersionedFiles(index, access, max_delta_chain=0,
278
def _get_texts(self, repo_transport, repo):
279
mapper = versionedfile.HashEscapedPrefixMapper()
280
base_transport = repo_transport.clone('knits')
281
index = _mod_knit._KndxIndex(base_transport, mapper,
282
repo.get_transaction, repo.is_write_locked, repo.is_locked)
283
access = _mod_knit._KnitKeyAccess(base_transport, mapper)
284
return _mod_knit.KnitVersionedFiles(index, access, max_delta_chain=200,
289
def _get_control_store(self, repo_transport, control_files):
290
"""Return the control store for this repository."""
291
return VersionedFileStore(
294
file_mode=control_files._file_mode,
295
versionedfile_class=knit.KnitVersionedFile,
296
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
299
def _get_revision_store(self, repo_transport, control_files):
300
"""See RepositoryFormat._get_revision_store()."""
301
from bzrlib.store.revision.knit import KnitRevisionStore
302
versioned_file_store = VersionedFileStore(
304
file_mode=control_files._file_mode,
307
versionedfile_class=knit.KnitVersionedFile,
308
versionedfile_kwargs={'delta':False,
309
'factory':knit.KnitPlainFactory(),
313
return KnitRevisionStore(versioned_file_store)
315
def _get_text_store(self, transport, control_files):
316
"""See RepositoryFormat._get_text_store()."""
317
return self._get_versioned_file_store('knits',
320
versionedfile_class=knit.KnitVersionedFile,
321
versionedfile_kwargs={
322
'create_parent_dir':True,
324
'dir_mode':control_files._dir_mode,
287
328
def initialize(self, a_bzrdir, shared=False):
288
329
"""Create a knit format 1 repository.
292
333
:param shared: If true the repository will be initialized as a shared
295
trace.mutter('creating repository in %s.', a_bzrdir.transport.base)
336
mutter('creating repository in %s.', a_bzrdir.transport.base)
337
dirs = ['revision-store', 'knits']
298
339
utf8_files = [('format', self.get_format_string())]
300
341
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
301
342
repo_transport = a_bzrdir.get_repository_transport(None)
302
343
control_files = lockable_files.LockableFiles(repo_transport,
303
344
'lock', lockdir.LockDir)
345
control_store = self._get_control_store(repo_transport, control_files)
304
346
transaction = transactions.WriteTransaction()
305
result = self.open(a_bzrdir=a_bzrdir, _found=True)
347
# trigger a write of the inventory store.
348
control_store.get_weave_or_empty('inventory', transaction)
349
_revision_store = self._get_revision_store(repo_transport, control_files)
307
350
# the revision id here is irrelevant: it will not be stored, and cannot
308
# already exist, we do this to create files on disk for older clients.
309
result.inventories.get_parent_map([('A',)])
310
result.revisions.get_parent_map([('A',)])
311
result.signatures.get_parent_map([('A',)])
313
self._run_post_repo_init_hooks(result, a_bzrdir, shared)
352
_revision_store.has_revision_id('A', transaction)
353
_revision_store.get_signature_file(transaction)
354
return self.open(a_bzrdir=a_bzrdir, _found=True)
316
356
def open(self, a_bzrdir, _found=False, _override_transport=None):
317
357
"""See RepositoryFormat.open().
319
359
:param _override_transport: INTERNAL USE ONLY. Allows opening the
320
360
repository at a slightly different url
321
361
than normal. I.e. during 'upgrade'.
324
364
format = RepositoryFormat.find_format(a_bzrdir)
365
assert format.__class__ == self.__class__
325
366
if _override_transport is not None:
326
367
repo_transport = _override_transport
328
369
repo_transport = a_bzrdir.get_repository_transport(None)
329
370
control_files = lockable_files.LockableFiles(repo_transport,
330
371
'lock', lockdir.LockDir)
331
repo = self.repository_class(_format=self,
372
text_store = self._get_text_store(repo_transport, control_files)
373
control_store = self._get_control_store(repo_transport, control_files)
374
_revision_store = self._get_revision_store(repo_transport, control_files)
375
return KnitRepository(_format=self,
332
376
a_bzrdir=a_bzrdir,
333
377
control_files=control_files,
334
_commit_builder_class=self._commit_builder_class,
335
_serializer=self._serializer)
336
repo.revisions = self._get_revisions(repo_transport, repo)
337
repo.signatures = self._get_signatures(repo_transport, repo)
338
repo.inventories = self._get_inventories(repo_transport, repo)
339
repo.texts = self._get_texts(repo_transport, repo)
340
repo.chk_bytes = None
341
repo._transport = repo_transport
378
_revision_store=_revision_store,
379
control_store=control_store,
380
text_store=text_store)
345
383
class RepositoryFormatKnit1(RepositoryFormatKnit):
417
455
"""See RepositoryFormat.get_format_description()."""
418
456
return "Knit repository format 3"
421
class RepositoryFormatKnit4(RepositoryFormatKnit):
422
"""Bzr repository knit format 4.
424
This repository format has everything in format 3, except for
426
- knits for file texts and inventory
427
- hash subdirectory based stores.
428
- knits for revisions and signatures
429
- TextStores for revisions and signatures.
430
- a format marker of its own
431
- an optional 'shared-storage' flag
432
- an optional 'no-working-trees' flag
434
- support for recording full info about the tree root
437
repository_class = KnitRepository
438
_commit_builder_class = VersionedFileRootCommitBuilder
439
rich_root_data = True
440
supports_tree_reference = False
442
def _serializer(self):
443
return xml6.serializer_v6
445
def _get_matching_bzrdir(self):
446
return bzrdir.format_registry.make_bzrdir('rich-root')
448
def _ignore_setting_bzrdir(self, format):
451
_matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)
453
def get_format_string(self):
454
"""See RepositoryFormat.get_format_string()."""
455
return 'Bazaar Knit Repository Format 4 (bzr 1.0)\n'
457
def get_format_description(self):
458
"""See RepositoryFormat.get_format_description()."""
459
return "Knit repository format 4"
462
class InterKnitRepo(InterSameDataRepository):
463
"""Optimised code paths between Knit based repositories."""
466
def _get_repo_format_to_test(self):
467
return RepositoryFormatKnit1()
470
def is_compatible(source, target):
471
"""Be compatible with known Knit formats.
473
We don't test for the stores being of specific types because that
474
could lead to confusing results, and there is no need to be
458
def open(self, a_bzrdir, _found=False, _override_transport=None):
459
"""See RepositoryFormat.open().
461
:param _override_transport: INTERNAL USE ONLY. Allows opening the
462
repository at a slightly different url
463
than normal. I.e. during 'upgrade'.
478
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
479
isinstance(target._format, RepositoryFormatKnit))
480
except AttributeError:
482
return are_knits and InterRepository._same_model(source, target)
485
def search_missing_revision_ids(self,
486
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
487
find_ghosts=True, revision_ids=None, if_present_ids=None,
489
"""See InterRepository.search_missing_revision_ids()."""
490
if symbol_versioning.deprecated_passed(revision_id):
491
symbol_versioning.warn(
492
'search_missing_revision_ids(revision_id=...) was '
493
'deprecated in 2.4. Use revision_ids=[...] instead.',
494
DeprecationWarning, stacklevel=2)
495
if revision_ids is not None:
496
raise AssertionError(
497
'revision_ids is mutually exclusive with revision_id')
498
if revision_id is not None:
499
revision_ids = [revision_id]
501
source_ids_set = self._present_source_revisions_for(
502
revision_ids, if_present_ids)
503
# source_ids is the worst possible case we may need to pull.
504
# now we want to filter source_ids against what we actually
505
# have in target, but don't try to check for existence where we know
506
# we do not have a revision as that would be pointless.
507
target_ids = set(self.target.all_revision_ids())
508
possibly_present_revisions = target_ids.intersection(source_ids_set)
509
actually_present_revisions = set(
510
self.target._eliminate_revisions_not_present(possibly_present_revisions))
511
required_revisions = source_ids_set.difference(actually_present_revisions)
512
if revision_ids is not None:
513
# we used get_ancestry to determine source_ids then we are assured all
514
# revisions referenced are present as they are installed in topological order.
515
# and the tip revision was validated by get_ancestry.
516
result_set = required_revisions
466
format = RepositoryFormat.find_format(a_bzrdir)
467
assert format.__class__ == self.__class__
468
if _override_transport is not None:
469
repo_transport = _override_transport
518
# if we just grabbed the possibly available ids, then
519
# we only have an estimate of whats available and need to validate
520
# that against the revision records.
522
self.source._eliminate_revisions_not_present(required_revisions))
523
if limit is not None:
524
topo_ordered = self.source.get_graph().iter_topo_order(result_set)
525
result_set = set(itertools.islice(topo_ordered, limit))
526
return self.source.revision_ids_to_search_result(result_set)
529
InterRepository.register_optimiser(InterKnitRepo)
471
repo_transport = a_bzrdir.get_repository_transport(None)
472
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
474
text_store = self._get_text_store(repo_transport, control_files)
475
control_store = self._get_control_store(repo_transport, control_files)
476
_revision_store = self._get_revision_store(repo_transport, control_files)
477
return self.repository_class(_format=self,
479
control_files=control_files,
480
_revision_store=_revision_store,
481
control_store=control_store,
482
text_store=text_store)
added
removed
bzrlib/repofmt/knitrepo.py
