1
# Copyright (C) 2005-2011 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
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(), """
32
revision as _mod_revision,
33
testament as _mod_testament,
36
from bzrlib.bundle import serializer
45
from bzrlib.decorators import needs_read_lock, needs_write_lock, only_raises
46
from bzrlib.inter import InterObject
47
from bzrlib.lock import _RelockDebugMixin, LogicalLockResult
48
from bzrlib.trace import (
49
log_exception_quietly, note, mutter, mutter_callsite, warning)
52
# Old formats display a warning, but only once
53
_deprecation_warning_done = False
56
class IsInWriteGroupError(errors.InternalBzrError):
58
_fmt = "May not refresh_data of repo %(repo)s while in a write group."
60
def __init__(self, repo):
61
errors.InternalBzrError.__init__(self, repo=repo)
64
class CommitBuilder(object):
65
"""Provides an interface to build up a commit.
67
This allows describing a tree to be committed without needing to
68
know the internals of the format of the repository.
71
# all clients should supply tree roots.
72
record_root_entry = True
73
# whether this commit builder supports the record_entry_contents interface
74
supports_record_entry_contents = False
76
def __init__(self, repository, parents, config, timestamp=None,
77
timezone=None, committer=None, revprops=None,
78
revision_id=None, lossy=False):
79
"""Initiate a CommitBuilder.
81
:param repository: Repository to commit to.
82
:param parents: Revision ids of the parents of the new revision.
83
:param timestamp: Optional timestamp recorded for commit.
84
:param timezone: Optional timezone for timestamp.
85
:param committer: Optional committer to set for commit.
86
:param revprops: Optional dictionary of revision properties.
87
:param revision_id: Optional revision id.
88
:param lossy: Whether to discard data that can not be natively
89
represented, when pushing to a foreign VCS
95
self._committer = self._config.username()
96
elif not isinstance(committer, unicode):
97
self._committer = committer.decode() # throw if non-ascii
99
self._committer = committer
101
self._new_revision_id = revision_id
102
self.parents = parents
103
self.repository = repository
106
if revprops is not None:
107
self._validate_revprops(revprops)
108
self._revprops.update(revprops)
110
if timestamp is None:
111
timestamp = time.time()
112
# Restrict resolution to 1ms
113
self._timestamp = round(timestamp, 3)
116
self._timezone = osutils.local_time_offset()
118
self._timezone = int(timezone)
120
self._generate_revision_if_needed()
122
def any_changes(self):
123
"""Return True if any entries were changed.
125
This includes merge-only changes. It is the core for the --unchanged
128
:return: True if any changes have occured.
130
raise NotImplementedError(self.any_changes)
132
def _validate_unicode_text(self, text, context):
133
"""Verify things like commit messages don't have bogus characters."""
135
raise ValueError('Invalid value for %s: %r' % (context, text))
137
def _validate_revprops(self, revprops):
138
for key, value in revprops.iteritems():
139
# We know that the XML serializers do not round trip '\r'
140
# correctly, so refuse to accept them
141
if not isinstance(value, basestring):
142
raise ValueError('revision property (%s) is not a valid'
143
' (unicode) string: %r' % (key, value))
144
self._validate_unicode_text(value,
145
'revision property (%s)' % (key,))
147
def commit(self, message):
148
"""Make the actual commit.
150
:return: The revision id of the recorded revision.
152
raise NotImplementedError(self.commit)
155
"""Abort the commit that is being built.
157
raise NotImplementedError(self.abort)
159
def revision_tree(self):
160
"""Return the tree that was just committed.
162
After calling commit() this can be called to get a
163
RevisionTree representing the newly committed tree. This is
164
preferred to calling Repository.revision_tree() because that may
165
require deserializing the inventory, while we already have a copy in
168
raise NotImplementedError(self.revision_tree)
170
def finish_inventory(self):
171
"""Tell the builder that the inventory is finished.
173
:return: The inventory id in the repository, which can be used with
174
repository.get_inventory.
176
raise NotImplementedError(self.finish_inventory)
178
def _gen_revision_id(self):
179
"""Return new revision-id."""
180
return generate_ids.gen_revision_id(self._committer, self._timestamp)
182
def _generate_revision_if_needed(self):
183
"""Create a revision id if None was supplied.
185
If the repository can not support user-specified revision ids
186
they should override this function and raise CannotSetRevisionId
187
if _new_revision_id is not None.
189
:raises: CannotSetRevisionId
191
if self._new_revision_id is None:
192
self._new_revision_id = self._gen_revision_id()
193
self.random_revid = True
195
self.random_revid = False
197
def will_record_deletes(self):
198
"""Tell the commit builder that deletes are being notified.
200
This enables the accumulation of an inventory delta; for the resulting
201
commit to be valid, deletes against the basis MUST be recorded via
202
builder.record_delete().
204
raise NotImplementedError(self.will_record_deletes)
206
def record_iter_changes(self, tree, basis_revision_id, iter_changes):
207
"""Record a new tree via iter_changes.
209
:param tree: The tree to obtain text contents from for changed objects.
210
:param basis_revision_id: The revision id of the tree the iter_changes
211
has been generated against. Currently assumed to be the same
212
as self.parents[0] - if it is not, errors may occur.
213
:param iter_changes: An iter_changes iterator with the changes to apply
214
to basis_revision_id. The iterator must not include any items with
215
a current kind of None - missing items must be either filtered out
216
or errored-on beefore record_iter_changes sees the item.
217
:return: A generator of (file_id, relpath, fs_hash) tuples for use with
220
raise NotImplementedError(self.record_iter_changes)
223
class RepositoryWriteLockResult(LogicalLockResult):
224
"""The result of write locking a repository.
226
:ivar repository_token: The token obtained from the underlying lock, or
228
:ivar unlock: A callable which will unlock the lock.
231
def __init__(self, unlock, repository_token):
232
LogicalLockResult.__init__(self, unlock)
233
self.repository_token = repository_token
236
return "RepositoryWriteLockResult(%s, %s)" % (self.repository_token,
240
######################################################################
244
class Repository(_RelockDebugMixin, controldir.ControlComponent):
245
"""Repository holding history for one or more branches.
247
The repository holds and retrieves historical information including
248
revisions and file history. It's normally accessed only by the Branch,
249
which views a particular line of development through that history.
251
See VersionedFileRepository in bzrlib.vf_repository for the
252
base class for most Bazaar repositories.
255
def abort_write_group(self, suppress_errors=False):
256
"""Commit the contents accrued within the current write group.
258
:param suppress_errors: if true, abort_write_group will catch and log
259
unexpected errors that happen during the abort, rather than
260
allowing them to propagate. Defaults to False.
262
:seealso: start_write_group.
264
if self._write_group is not self.get_transaction():
265
# has an unlock or relock occured ?
268
'(suppressed) mismatched lock context and write group. %r, %r',
269
self._write_group, self.get_transaction())
271
raise errors.BzrError(
272
'mismatched lock context and write group. %r, %r' %
273
(self._write_group, self.get_transaction()))
275
self._abort_write_group()
276
except Exception, exc:
277
self._write_group = None
278
if not suppress_errors:
280
mutter('abort_write_group failed')
281
log_exception_quietly()
282
note('bzr: ERROR (ignored): %s', exc)
283
self._write_group = None
285
def _abort_write_group(self):
286
"""Template method for per-repository write group cleanup.
288
This is called during abort before the write group is considered to be
289
finished and should cleanup any internal state accrued during the write
290
group. There is no requirement that data handed to the repository be
291
*not* made available - this is not a rollback - but neither should any
292
attempt be made to ensure that data added is fully commited. Abort is
293
invoked when an error has occured so futher disk or network operations
294
may not be possible or may error and if possible should not be
298
def add_fallback_repository(self, repository):
299
"""Add a repository to use for looking up data not held locally.
301
:param repository: A repository.
303
raise NotImplementedError(self.add_fallback_repository)
305
def _check_fallback_repository(self, repository):
306
"""Check that this repository can fallback to repository safely.
308
Raise an error if not.
310
:param repository: A repository to fallback to.
312
return InterRepository._assert_same_model(self, repository)
314
def all_revision_ids(self):
315
"""Returns a list of all the revision ids in the repository.
317
This is conceptually deprecated because code should generally work on
318
the graph reachable from a particular revision, and ignore any other
319
revisions that might be present. There is no direct replacement
322
if 'evil' in debug.debug_flags:
323
mutter_callsite(2, "all_revision_ids is linear with history.")
324
return self._all_revision_ids()
326
def _all_revision_ids(self):
327
"""Returns a list of all the revision ids in the repository.
329
These are in as much topological order as the underlying store can
332
raise NotImplementedError(self._all_revision_ids)
334
def break_lock(self):
335
"""Break a lock if one is present from another instance.
337
Uses the ui factory to ask for confirmation if the lock may be from
340
self.control_files.break_lock()
343
def _eliminate_revisions_not_present(self, revision_ids):
344
"""Check every revision id in revision_ids to see if we have it.
346
Returns a set of the present revisions.
349
graph = self.get_graph()
350
parent_map = graph.get_parent_map(revision_ids)
351
# The old API returned a list, should this actually be a set?
352
return parent_map.keys()
355
def create(a_bzrdir):
356
"""Construct the current default format repository in a_bzrdir."""
357
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
359
def __init__(self, _format, a_bzrdir, control_files):
360
"""instantiate a Repository.
362
:param _format: The format of the repository on disk.
363
:param a_bzrdir: The BzrDir of the repository.
364
:param control_files: Control files to use for locking, etc.
366
# In the future we will have a single api for all stores for
367
# getting file texts, inventories and revisions, then
368
# this construct will accept instances of those things.
369
super(Repository, self).__init__()
370
self._format = _format
371
# the following are part of the public API for Repository:
372
self.bzrdir = a_bzrdir
373
self.control_files = control_files
374
self._transport = control_files._transport
375
self.base = self._transport.base
377
self._write_group = None
378
# Additional places to query for data.
379
self._fallback_repositories = []
382
def user_transport(self):
383
return self.bzrdir.user_transport
386
def control_transport(self):
387
return self._transport
390
if self._fallback_repositories:
391
return '%s(%r, fallback_repositories=%r)' % (
392
self.__class__.__name__,
394
self._fallback_repositories)
396
return '%s(%r)' % (self.__class__.__name__,
399
def _has_same_fallbacks(self, other_repo):
400
"""Returns true if the repositories have the same fallbacks."""
401
my_fb = self._fallback_repositories
402
other_fb = other_repo._fallback_repositories
403
if len(my_fb) != len(other_fb):
405
for f, g in zip(my_fb, other_fb):
406
if not f.has_same_location(g):
410
def has_same_location(self, other):
411
"""Returns a boolean indicating if this repository is at the same
412
location as another repository.
414
This might return False even when two repository objects are accessing
415
the same physical repository via different URLs.
417
if self.__class__ is not other.__class__:
419
return (self._transport.base == other._transport.base)
421
def is_in_write_group(self):
422
"""Return True if there is an open write group.
424
:seealso: start_write_group.
426
return self._write_group is not None
429
return self.control_files.is_locked()
431
def is_write_locked(self):
432
"""Return True if this object is write locked."""
433
return self.is_locked() and self.control_files._lock_mode == 'w'
435
def lock_write(self, token=None):
436
"""Lock this repository for writing.
438
This causes caching within the repository obejct to start accumlating
439
data during reads, and allows a 'write_group' to be obtained. Write
440
groups must be used for actual data insertion.
442
A token should be passed in if you know that you have locked the object
443
some other way, and need to synchronise this object's state with that
446
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
448
:param token: if this is already locked, then lock_write will fail
449
unless the token matches the existing lock.
450
:returns: a token if this instance supports tokens, otherwise None.
451
:raises TokenLockingNotSupported: when a token is given but this
452
instance doesn't support using token locks.
453
:raises MismatchedToken: if the specified token doesn't match the token
454
of the existing lock.
455
:seealso: start_write_group.
456
:return: A RepositoryWriteLockResult.
458
locked = self.is_locked()
459
token = self.control_files.lock_write(token=token)
461
self._warn_if_deprecated()
463
for repo in self._fallback_repositories:
464
# Writes don't affect fallback repos
467
return RepositoryWriteLockResult(self.unlock, token)
470
"""Lock the repository for read operations.
472
:return: An object with an unlock method which will release the lock
475
locked = self.is_locked()
476
self.control_files.lock_read()
478
self._warn_if_deprecated()
480
for repo in self._fallback_repositories:
483
return LogicalLockResult(self.unlock)
485
def get_physical_lock_status(self):
486
return self.control_files.get_physical_lock_status()
488
def leave_lock_in_place(self):
489
"""Tell this repository not to release the physical lock when this
492
If lock_write doesn't return a token, then this method is not supported.
494
self.control_files.leave_in_place()
496
def dont_leave_lock_in_place(self):
497
"""Tell this repository to release the physical lock when this
498
object is unlocked, even if it didn't originally acquire it.
500
If lock_write doesn't return a token, then this method is not supported.
502
self.control_files.dont_leave_in_place()
505
def gather_stats(self, revid=None, committers=None):
506
"""Gather statistics from a revision id.
508
:param revid: The revision id to gather statistics from, if None, then
509
no revision specific statistics are gathered.
510
:param committers: Optional parameter controlling whether to grab
511
a count of committers from the revision specific statistics.
512
:return: A dictionary of statistics. Currently this contains:
513
committers: The number of committers if requested.
514
firstrev: A tuple with timestamp, timezone for the penultimate left
515
most ancestor of revid, if revid is not the NULL_REVISION.
516
latestrev: A tuple with timestamp, timezone for revid, if revid is
517
not the NULL_REVISION.
518
revisions: The total revision count in the repository.
519
size: An estimate disk size of the repository in bytes.
522
if revid and committers:
523
result['committers'] = 0
524
if revid and revid != _mod_revision.NULL_REVISION:
526
all_committers = set()
527
revisions = self.get_ancestry(revid)
528
# pop the leading None
530
first_revision = None
532
# ignore the revisions in the middle - just grab first and last
533
revisions = revisions[0], revisions[-1]
534
for revision in self.get_revisions(revisions):
535
if not first_revision:
536
first_revision = revision
538
all_committers.add(revision.committer)
539
last_revision = revision
541
result['committers'] = len(all_committers)
542
result['firstrev'] = (first_revision.timestamp,
543
first_revision.timezone)
544
result['latestrev'] = (last_revision.timestamp,
545
last_revision.timezone)
548
def find_branches(self, using=False):
549
"""Find branches underneath this repository.
551
This will include branches inside other branches.
553
:param using: If True, list only branches using this repository.
555
if using and not self.is_shared():
556
return self.bzrdir.list_branches()
557
class Evaluator(object):
560
self.first_call = True
562
def __call__(self, bzrdir):
563
# On the first call, the parameter is always the bzrdir
564
# containing the current repo.
565
if not self.first_call:
567
repository = bzrdir.open_repository()
568
except errors.NoRepositoryPresent:
571
return False, ([], repository)
572
self.first_call = False
573
value = (bzrdir.list_branches(), None)
577
for branches, repository in bzrdir.BzrDir.find_bzrdirs(
578
self.user_transport, evaluate=Evaluator()):
579
if branches is not None:
581
if not using and repository is not None:
582
ret.extend(repository.find_branches())
586
def search_missing_revision_ids(self, other,
587
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
588
find_ghosts=True, revision_ids=None, if_present_ids=None):
589
"""Return the revision ids that other has that this does not.
591
These are returned in topological order.
593
revision_id: only return revision ids included by revision_id.
595
if symbol_versioning.deprecated_passed(revision_id):
596
symbol_versioning.warn(
597
'search_missing_revision_ids(revision_id=...) was '
598
'deprecated in 2.4. Use revision_ids=[...] instead.',
599
DeprecationWarning, stacklevel=3)
600
if revision_ids is not None:
601
raise AssertionError(
602
'revision_ids is mutually exclusive with revision_id')
603
if revision_id is not None:
604
revision_ids = [revision_id]
605
return InterRepository.get(other, self).search_missing_revision_ids(
606
find_ghosts=find_ghosts, revision_ids=revision_ids,
607
if_present_ids=if_present_ids)
611
"""Open the repository rooted at base.
613
For instance, if the repository is at URL/.bzr/repository,
614
Repository.open(URL) -> a Repository instance.
616
control = bzrdir.BzrDir.open(base)
617
return control.open_repository()
619
def copy_content_into(self, destination, revision_id=None):
620
"""Make a complete copy of the content in self into destination.
622
This is a destructive operation! Do not use it on existing
625
return InterRepository.get(self, destination).copy_content(revision_id)
627
def commit_write_group(self):
628
"""Commit the contents accrued within the current write group.
630
:seealso: start_write_group.
632
:return: it may return an opaque hint that can be passed to 'pack'.
634
if self._write_group is not self.get_transaction():
635
# has an unlock or relock occured ?
636
raise errors.BzrError('mismatched lock context %r and '
638
(self.get_transaction(), self._write_group))
639
result = self._commit_write_group()
640
self._write_group = None
643
def _commit_write_group(self):
644
"""Template method for per-repository write group cleanup.
646
This is called before the write group is considered to be
647
finished and should ensure that all data handed to the repository
648
for writing during the write group is safely committed (to the
649
extent possible considering file system caching etc).
652
def suspend_write_group(self):
653
raise errors.UnsuspendableWriteGroup(self)
655
def refresh_data(self):
656
"""Re-read any data needed to synchronise with disk.
658
This method is intended to be called after another repository instance
659
(such as one used by a smart server) has inserted data into the
660
repository. On all repositories this will work outside of write groups.
661
Some repository formats (pack and newer for bzrlib native formats)
662
support refresh_data inside write groups. If called inside a write
663
group on a repository that does not support refreshing in a write group
664
IsInWriteGroupError will be raised.
668
def resume_write_group(self, tokens):
669
if not self.is_write_locked():
670
raise errors.NotWriteLocked(self)
671
if self._write_group:
672
raise errors.BzrError('already in a write group')
673
self._resume_write_group(tokens)
674
# so we can detect unlock/relock - the write group is now entered.
675
self._write_group = self.get_transaction()
677
def _resume_write_group(self, tokens):
678
raise errors.UnsuspendableWriteGroup(self)
680
def fetch(self, source, revision_id=None, find_ghosts=False,
682
"""Fetch the content required to construct revision_id from source.
684
If revision_id is None and fetch_spec is None, then all content is
687
fetch() may not be used when the repository is in a write group -
688
either finish the current write group before using fetch, or use
689
fetch before starting the write group.
691
:param find_ghosts: Find and copy revisions in the source that are
692
ghosts in the target (and not reachable directly by walking out to
693
the first-present revision in target from revision_id).
694
:param revision_id: If specified, all the content needed for this
695
revision ID will be copied to the target. Fetch will determine for
696
itself which content needs to be copied.
697
:param fetch_spec: If specified, a SearchResult or
698
PendingAncestryResult that describes which revisions to copy. This
699
allows copying multiple heads at once. Mutually exclusive with
702
if fetch_spec is not None and revision_id is not None:
703
raise AssertionError(
704
"fetch_spec and revision_id are mutually exclusive.")
705
if self.is_in_write_group():
706
raise errors.InternalBzrError(
707
"May not fetch while in a write group.")
708
# fast path same-url fetch operations
709
# TODO: lift out to somewhere common with RemoteRepository
710
# <https://bugs.launchpad.net/bzr/+bug/401646>
711
if (self.has_same_location(source)
712
and fetch_spec is None
713
and self._has_same_fallbacks(source)):
714
# check that last_revision is in 'from' and then return a
716
if (revision_id is not None and
717
not _mod_revision.is_null(revision_id)):
718
self.get_revision(revision_id)
720
inter = InterRepository.get(source, self)
721
return inter.fetch(revision_id=revision_id,
722
find_ghosts=find_ghosts, fetch_spec=fetch_spec)
724
def create_bundle(self, target, base, fileobj, format=None):
725
return serializer.write_bundle(self, target, base, fileobj, format)
727
def get_commit_builder(self, branch, parents, config, timestamp=None,
728
timezone=None, committer=None, revprops=None,
729
revision_id=None, lossy=False):
730
"""Obtain a CommitBuilder for this repository.
732
:param branch: Branch to commit to.
733
:param parents: Revision ids of the parents of the new revision.
734
:param config: Configuration to use.
735
:param timestamp: Optional timestamp recorded for commit.
736
:param timezone: Optional timezone for timestamp.
737
:param committer: Optional committer to set for commit.
738
:param revprops: Optional dictionary of revision properties.
739
:param revision_id: Optional revision id.
740
:param lossy: Whether to discard data that can not be natively
741
represented, when pushing to a foreign VCS
743
raise NotImplementedError(self.get_commit_builder)
745
@only_raises(errors.LockNotHeld, errors.LockBroken)
747
if (self.control_files._lock_count == 1 and
748
self.control_files._lock_mode == 'w'):
749
if self._write_group is not None:
750
self.abort_write_group()
751
self.control_files.unlock()
752
raise errors.BzrError(
753
'Must end write groups before releasing write locks.')
754
self.control_files.unlock()
755
if self.control_files._lock_count == 0:
756
for repo in self._fallback_repositories:
760
def clone(self, a_bzrdir, revision_id=None):
761
"""Clone this repository into a_bzrdir using the current format.
763
Currently no check is made that the format of this repository and
764
the bzrdir format are compatible. FIXME RBC 20060201.
766
:return: The newly created destination repository.
768
# TODO: deprecate after 0.16; cloning this with all its settings is
769
# probably not very useful -- mbp 20070423
770
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
771
self.copy_content_into(dest_repo, revision_id)
774
def start_write_group(self):
775
"""Start a write group in the repository.
777
Write groups are used by repositories which do not have a 1:1 mapping
778
between file ids and backend store to manage the insertion of data from
779
both fetch and commit operations.
781
A write lock is required around the start_write_group/commit_write_group
782
for the support of lock-requiring repository formats.
784
One can only insert data into a repository inside a write group.
788
if not self.is_write_locked():
789
raise errors.NotWriteLocked(self)
790
if self._write_group:
791
raise errors.BzrError('already in a write group')
792
self._start_write_group()
793
# so we can detect unlock/relock - the write group is now entered.
794
self._write_group = self.get_transaction()
796
def _start_write_group(self):
797
"""Template method for per-repository write group startup.
799
This is called before the write group is considered to be
804
def sprout(self, to_bzrdir, revision_id=None):
805
"""Create a descendent repository for new development.
807
Unlike clone, this does not copy the settings of the repository.
809
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
810
dest_repo.fetch(self, revision_id=revision_id)
813
def _create_sprouting_repo(self, a_bzrdir, shared):
814
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
815
# use target default format.
816
dest_repo = a_bzrdir.create_repository()
818
# Most control formats need the repository to be specifically
819
# created, but on some old all-in-one formats it's not needed
821
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
822
except errors.UninitializableFormat:
823
dest_repo = a_bzrdir.open_repository()
827
def has_revision(self, revision_id):
828
"""True if this repository has a copy of the revision."""
829
return revision_id in self.has_revisions((revision_id,))
832
def has_revisions(self, revision_ids):
833
"""Probe to find out the presence of multiple revisions.
835
:param revision_ids: An iterable of revision_ids.
836
:return: A set of the revision_ids that were present.
838
raise NotImplementedError(self.has_revisions)
841
def get_revision(self, revision_id):
842
"""Return the Revision object for a named revision."""
843
return self.get_revisions([revision_id])[0]
845
def get_revision_reconcile(self, revision_id):
846
"""'reconcile' helper routine that allows access to a revision always.
848
This variant of get_revision does not cross check the weave graph
849
against the revision one as get_revision does: but it should only
850
be used by reconcile, or reconcile-alike commands that are correcting
851
or testing the revision graph.
853
raise NotImplementedError(self.get_revision_reconcile)
855
def get_revisions(self, revision_ids):
856
"""Get many revisions at once.
858
Repositories that need to check data on every revision read should
859
subclass this method.
861
raise NotImplementedError(self.get_revisions)
863
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
864
"""Produce a generator of revision deltas.
866
Note that the input is a sequence of REVISIONS, not revision_ids.
867
Trees will be held in memory until the generator exits.
868
Each delta is relative to the revision's lefthand predecessor.
870
:param specific_fileids: if not None, the result is filtered
871
so that only those file-ids, their parents and their
872
children are included.
874
# Get the revision-ids of interest
875
required_trees = set()
876
for revision in revisions:
877
required_trees.add(revision.revision_id)
878
required_trees.update(revision.parent_ids[:1])
880
# Get the matching filtered trees. Note that it's more
881
# efficient to pass filtered trees to changes_from() rather
882
# than doing the filtering afterwards. changes_from() could
883
# arguably do the filtering itself but it's path-based, not
884
# file-id based, so filtering before or afterwards is
886
if specific_fileids is None:
887
trees = dict((t.get_revision_id(), t) for
888
t in self.revision_trees(required_trees))
890
trees = dict((t.get_revision_id(), t) for
891
t in self._filtered_revision_trees(required_trees,
894
# Calculate the deltas
895
for revision in revisions:
896
if not revision.parent_ids:
897
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
899
old_tree = trees[revision.parent_ids[0]]
900
yield trees[revision.revision_id].changes_from(old_tree)
903
def get_revision_delta(self, revision_id, specific_fileids=None):
904
"""Return the delta for one revision.
906
The delta is relative to the left-hand predecessor of the
909
:param specific_fileids: if not None, the result is filtered
910
so that only those file-ids, their parents and their
911
children are included.
913
r = self.get_revision(revision_id)
914
return list(self.get_deltas_for_revisions([r],
915
specific_fileids=specific_fileids))[0]
918
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
919
signature = gpg_strategy.sign(plaintext)
920
self.add_signature_text(revision_id, signature)
922
def add_signature_text(self, revision_id, signature):
923
"""Store a signature text for a revision.
925
:param revision_id: Revision id of the revision
926
:param signature: Signature text.
928
raise NotImplementedError(self.add_signature_text)
930
def find_text_key_references(self):
931
"""Find the text key references within the repository.
933
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
934
to whether they were referred to by the tree of the
935
revision_id that they contain.
937
raise NotImplementedError(self.find_text_key_references)
939
def _find_parent_ids_of_revisions(self, revision_ids):
940
"""Find all parent ids that are mentioned in the revision graph.
942
:return: set of revisions that are parents of revision_ids which are
943
not part of revision_ids themselves
945
parent_map = self.get_parent_map(revision_ids)
947
map(parent_ids.update, parent_map.itervalues())
948
parent_ids.difference_update(revision_ids)
949
parent_ids.discard(_mod_revision.NULL_REVISION)
952
def fileids_altered_by_revision_ids(self, revision_ids):
953
"""Find the file ids and versions affected by revisions.
955
:param revisions: an iterable containing revision ids.
956
:return: a dictionary mapping altered file-ids to an iterable of
957
revision_ids. Each altered file-ids has the exact revision_ids
958
that altered it listed explicitly.
960
raise NotImplementedError(self.fileids_altered_by_revision_ids)
962
def iter_files_bytes(self, desired_files):
963
"""Iterate through file versions.
965
Files will not necessarily be returned in the order they occur in
966
desired_files. No specific order is guaranteed.
968
Yields pairs of identifier, bytes_iterator. identifier is an opaque
969
value supplied by the caller as part of desired_files. It should
970
uniquely identify the file version in the caller's context. (Examples:
971
an index number or a TreeTransform trans_id.)
973
:param desired_files: a list of (file_id, revision_id, identifier)
976
raise NotImplementedError(self.iter_files_bytes)
978
def get_rev_id_for_revno(self, revno, known_pair):
979
"""Return the revision id of a revno, given a later (revno, revid)
980
pair in the same history.
982
:return: if found (True, revid). If the available history ran out
983
before reaching the revno, then this returns
984
(False, (closest_revno, closest_revid)).
986
known_revno, known_revid = known_pair
987
partial_history = [known_revid]
988
distance_from_known = known_revno - revno
989
if distance_from_known < 0:
991
'requested revno (%d) is later than given known revno (%d)'
992
% (revno, known_revno))
995
self, partial_history, stop_index=distance_from_known)
996
except errors.RevisionNotPresent, err:
997
if err.revision_id == known_revid:
998
# The start revision (known_revid) wasn't found.
1000
# This is a stacked repository with no fallbacks, or a there's a
1001
# left-hand ghost. Either way, even though the revision named in
1002
# the error isn't in this repo, we know it's the next step in this
1003
# left-hand history.
1004
partial_history.append(err.revision_id)
1005
if len(partial_history) <= distance_from_known:
1006
# Didn't find enough history to get a revid for the revno.
1007
earliest_revno = known_revno - len(partial_history) + 1
1008
return (False, (earliest_revno, partial_history[-1]))
1009
if len(partial_history) - 1 > distance_from_known:
1010
raise AssertionError('_iter_for_revno returned too much history')
1011
return (True, partial_history[-1])
1013
def iter_reverse_revision_history(self, revision_id):
1014
"""Iterate backwards through revision ids in the lefthand history
1016
:param revision_id: The revision id to start with. All its lefthand
1017
ancestors will be traversed.
1019
graph = self.get_graph()
1020
stop_revisions = (None, _mod_revision.NULL_REVISION)
1021
return graph.iter_lefthand_ancestry(revision_id, stop_revisions)
1023
def is_shared(self):
1024
"""Return True if this repository is flagged as a shared repository."""
1025
raise NotImplementedError(self.is_shared)
1028
def reconcile(self, other=None, thorough=False):
1029
"""Reconcile this repository."""
1030
from bzrlib.reconcile import RepoReconciler
1031
reconciler = RepoReconciler(self, thorough=thorough)
1032
reconciler.reconcile()
1035
def _refresh_data(self):
1036
"""Helper called from lock_* to ensure coherency with disk.
1038
The default implementation does nothing; it is however possible
1039
for repositories to maintain loaded indices across multiple locks
1040
by checking inside their implementation of this method to see
1041
whether their indices are still valid. This depends of course on
1042
the disk format being validatable in this manner. This method is
1043
also called by the refresh_data() public interface to cause a refresh
1044
to occur while in a write lock so that data inserted by a smart server
1045
push operation is visible on the client's instance of the physical
1050
def revision_tree(self, revision_id):
1051
"""Return Tree for a revision on this branch.
1053
`revision_id` may be NULL_REVISION for the empty tree revision.
1055
raise NotImplementedError(self.revision_tree)
1057
def revision_trees(self, revision_ids):
1058
"""Return Trees for revisions in this repository.
1060
:param revision_ids: a sequence of revision-ids;
1061
a revision-id may not be None or 'null:'
1063
raise NotImplementedError(self.revision_trees)
1066
def get_ancestry(self, revision_id, topo_sorted=True):
1067
"""Return a list of revision-ids integrated by a revision.
1069
The first element of the list is always None, indicating the origin
1070
revision. This might change when we have history horizons, or
1071
perhaps we should have a new API.
1073
This is topologically sorted.
1075
if _mod_revision.is_null(revision_id):
1077
if not self.has_revision(revision_id):
1078
raise errors.NoSuchRevision(self, revision_id)
1079
graph = self.get_graph()
1081
search = graph._make_breadth_first_searcher([revision_id])
1084
found, ghosts = search.next_with_ghosts()
1085
except StopIteration:
1088
if _mod_revision.NULL_REVISION in keys:
1089
keys.remove(_mod_revision.NULL_REVISION)
1091
parent_map = graph.get_parent_map(keys)
1092
keys = tsort.topo_sort(parent_map)
1093
return [None] + list(keys)
1095
def pack(self, hint=None, clean_obsolete_packs=False):
1096
"""Compress the data within the repository.
1098
This operation only makes sense for some repository types. For other
1099
types it should be a no-op that just returns.
1101
This stub method does not require a lock, but subclasses should use
1102
@needs_write_lock as this is a long running call it's reasonable to
1103
implicitly lock for the user.
1105
:param hint: If not supplied, the whole repository is packed.
1106
If supplied, the repository may use the hint parameter as a
1107
hint for the parts of the repository to pack. A hint can be
1108
obtained from the result of commit_write_group(). Out of
1109
date hints are simply ignored, because concurrent operations
1110
can obsolete them rapidly.
1112
:param clean_obsolete_packs: Clean obsolete packs immediately after
1116
def get_transaction(self):
1117
return self.control_files.get_transaction()
1119
def get_parent_map(self, revision_ids):
1120
"""See graph.StackedParentsProvider.get_parent_map"""
1121
raise NotImplementedError(self.get_parent_map)
1123
def _get_parent_map_no_fallbacks(self, revision_ids):
1124
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1125
# revisions index works in keys; this just works in revisions
1126
# therefore wrap and unwrap
1129
for revision_id in revision_ids:
1130
if revision_id == _mod_revision.NULL_REVISION:
1131
result[revision_id] = ()
1132
elif revision_id is None:
1133
raise ValueError('get_parent_map(None) is not valid')
1135
query_keys.append((revision_id ,))
1136
vf = self.revisions.without_fallbacks()
1137
for ((revision_id,), parent_keys) in \
1138
vf.get_parent_map(query_keys).iteritems():
1140
result[revision_id] = tuple([parent_revid
1141
for (parent_revid,) in parent_keys])
1143
result[revision_id] = (_mod_revision.NULL_REVISION,)
1146
def _make_parents_provider(self):
1147
if not self._format.supports_external_lookups:
1149
return graph.StackedParentsProvider(_LazyListJoin(
1150
[self._make_parents_provider_unstacked()],
1151
self._fallback_repositories))
1153
def _make_parents_provider_unstacked(self):
1154
return graph.CallableToParentsProviderAdapter(
1155
self._get_parent_map_no_fallbacks)
1158
def get_known_graph_ancestry(self, revision_ids):
1159
"""Return the known graph for a set of revision ids and their ancestors.
1161
raise NotImplementedError(self.get_known_graph_ancestry)
1163
def get_graph(self, other_repository=None):
1164
"""Return the graph walker for this repository format"""
1165
parents_provider = self._make_parents_provider()
1166
if (other_repository is not None and
1167
not self.has_same_location(other_repository)):
1168
parents_provider = graph.StackedParentsProvider(
1169
[parents_provider, other_repository._make_parents_provider()])
1170
return graph.Graph(parents_provider)
1172
def revision_ids_to_search_result(self, result_set):
1173
"""Convert a set of revision ids to a graph SearchResult."""
1174
result_parents = set()
1175
for parents in self.get_graph().get_parent_map(
1176
result_set).itervalues():
1177
result_parents.update(parents)
1178
included_keys = result_set.intersection(result_parents)
1179
start_keys = result_set.difference(included_keys)
1180
exclude_keys = result_parents.difference(result_set)
1181
result = graph.SearchResult(start_keys, exclude_keys,
1182
len(result_set), result_set)
1186
def set_make_working_trees(self, new_value):
1187
"""Set the policy flag for making working trees when creating branches.
1189
This only applies to branches that use this repository.
1191
The default is 'True'.
1192
:param new_value: True to restore the default, False to disable making
1195
raise NotImplementedError(self.set_make_working_trees)
1197
def make_working_trees(self):
1198
"""Returns the policy for making working trees on new branches."""
1199
raise NotImplementedError(self.make_working_trees)
1202
def sign_revision(self, revision_id, gpg_strategy):
1203
testament = _mod_testament.Testament.from_revision(self, revision_id)
1204
plaintext = testament.as_short_text()
1205
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1207
def has_signature_for_revision_id(self, revision_id):
1208
"""Query for a revision signature for revision_id in the repository."""
1209
raise NotImplementedError(self.has_signature_for_revision_id)
1211
def get_signature_text(self, revision_id):
1212
"""Return the text for a signature."""
1213
raise NotImplementedError(self.get_signature_text)
1216
def check(self, revision_ids=None, callback_refs=None, check_repo=True):
1217
"""Check consistency of all history of given revision_ids.
1219
Different repository implementations should override _check().
1221
:param revision_ids: A non-empty list of revision_ids whose ancestry
1222
will be checked. Typically the last revision_id of a branch.
1223
:param callback_refs: A dict of check-refs to resolve and callback
1224
the check/_check method on the items listed as wanting the ref.
1226
:param check_repo: If False do not check the repository contents, just
1227
calculate the data callback_refs requires and call them back.
1229
return self._check(revision_ids, callback_refs=callback_refs,
1230
check_repo=check_repo)
1232
def _check(self, revision_ids, callback_refs, check_repo):
1233
result = check.Check(self, check_repo=check_repo)
1234
result.check(callback_refs)
1237
def _warn_if_deprecated(self, branch=None):
1238
if not self._format.is_deprecated():
1240
global _deprecation_warning_done
1241
if _deprecation_warning_done:
1245
conf = config.GlobalConfig()
1247
conf = branch.get_config()
1248
if conf.suppress_warning('format_deprecation'):
1250
warning("Format %s for %s is deprecated -"
1251
" please use 'bzr upgrade' to get better performance"
1252
% (self._format, self.bzrdir.transport.base))
1254
_deprecation_warning_done = True
1256
def supports_rich_root(self):
1257
return self._format.rich_root_data
1259
def _check_ascii_revisionid(self, revision_id, method):
1260
"""Private helper for ascii-only repositories."""
1261
# weave repositories refuse to store revisionids that are non-ascii.
1262
if revision_id is not None:
1263
# weaves require ascii revision ids.
1264
if isinstance(revision_id, unicode):
1266
revision_id.encode('ascii')
1267
except UnicodeEncodeError:
1268
raise errors.NonAsciiRevisionId(method, self)
1271
revision_id.decode('ascii')
1272
except UnicodeDecodeError:
1273
raise errors.NonAsciiRevisionId(method, self)
1276
class MetaDirRepository(Repository):
1277
"""Repositories in the new meta-dir layout.
1279
:ivar _transport: Transport for access to repository control files,
1280
typically pointing to .bzr/repository.
1283
def __init__(self, _format, a_bzrdir, control_files):
1284
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
1285
self._transport = control_files._transport
1287
def is_shared(self):
1288
"""Return True if this repository is flagged as a shared repository."""
1289
return self._transport.has('shared-storage')
1292
def set_make_working_trees(self, new_value):
1293
"""Set the policy flag for making working trees when creating branches.
1295
This only applies to branches that use this repository.
1297
The default is 'True'.
1298
:param new_value: True to restore the default, False to disable making
1303
self._transport.delete('no-working-trees')
1304
except errors.NoSuchFile:
1307
self._transport.put_bytes('no-working-trees', '',
1308
mode=self.bzrdir._get_file_mode())
1310
def make_working_trees(self):
1311
"""Returns the policy for making working trees on new branches."""
1312
return not self._transport.has('no-working-trees')
1315
class RepositoryFormatRegistry(controldir.ControlComponentFormatRegistry):
1316
"""Repository format registry."""
1318
def get_default(self):
1319
"""Return the current default format."""
1320
from bzrlib import bzrdir
1321
return bzrdir.format_registry.make_bzrdir('default').repository_format
1324
network_format_registry = registry.FormatRegistry()
1325
"""Registry of formats indexed by their network name.
1327
The network name for a repository format is an identifier that can be used when
1328
referring to formats with smart server operations. See
1329
RepositoryFormat.network_name() for more detail.
1333
format_registry = RepositoryFormatRegistry(network_format_registry)
1334
"""Registry of formats, indexed by their BzrDirMetaFormat format string.
1336
This can contain either format instances themselves, or classes/factories that
1337
can be called to obtain one.
1341
#####################################################################
1342
# Repository Formats
1344
class RepositoryFormat(controldir.ControlComponentFormat):
1345
"""A repository format.
1347
Formats provide four things:
1348
* An initialization routine to construct repository data on disk.
1349
* a optional format string which is used when the BzrDir supports
1351
* an open routine which returns a Repository instance.
1352
* A network name for referring to the format in smart server RPC
1355
There is one and only one Format subclass for each on-disk format. But
1356
there can be one Repository subclass that is used for several different
1357
formats. The _format attribute on a Repository instance can be used to
1358
determine the disk format.
1360
Formats are placed in a registry by their format string for reference
1361
during opening. These should be subclasses of RepositoryFormat for
1364
Once a format is deprecated, just deprecate the initialize and open
1365
methods on the format class. Do not deprecate the object, as the
1366
object may be created even when a repository instance hasn't been
1369
Common instance attributes:
1370
_matchingbzrdir - the bzrdir format that the repository format was
1371
originally written to work with. This can be used if manually
1372
constructing a bzrdir and repository, or more commonly for test suite
1376
# Set to True or False in derived classes. True indicates that the format
1377
# supports ghosts gracefully.
1378
supports_ghosts = None
1379
# Can this repository be given external locations to lookup additional
1380
# data. Set to True or False in derived classes.
1381
supports_external_lookups = None
1382
# Does this format support CHK bytestring lookups. Set to True or False in
1384
supports_chks = None
1385
# Should fetch trigger a reconcile after the fetch? Only needed for
1386
# some repository formats that can suffer internal inconsistencies.
1387
_fetch_reconcile = False
1388
# Does this format have < O(tree_size) delta generation. Used to hint what
1389
# code path for commit, amongst other things.
1391
# Does doing a pack operation compress data? Useful for the pack UI command
1392
# (so if there is one pack, the operation can still proceed because it may
1393
# help), and for fetching when data won't have come from the same
1395
pack_compresses = False
1396
# Does the repository storage understand references to trees?
1397
supports_tree_reference = None
1398
# Is the format experimental ?
1399
experimental = False
1400
# Does this repository format escape funky characters, or does it create
1401
# files with similar names as the versioned files in its contents on disk
1403
supports_funky_characters = None
1404
# Does this repository format support leaving locks?
1405
supports_leaving_lock = None
1406
# Does this format support the full VersionedFiles interface?
1407
supports_full_versioned_files = None
1408
# Does this format support signing revision signatures?
1409
supports_revision_signatures = True
1410
# Can the revision graph have incorrect parents?
1411
revision_graph_can_have_wrong_parents = None
1412
# Does this format support rich root data?
1413
rich_root_data = None
1416
return "%s()" % self.__class__.__name__
1418
def __eq__(self, other):
1419
# format objects are generally stateless
1420
return isinstance(other, self.__class__)
1422
def __ne__(self, other):
1423
return not self == other
1426
def find_format(klass, a_bzrdir):
1427
"""Return the format for the repository object in a_bzrdir.
1429
This is used by bzr native formats that have a "format" file in
1430
the repository. Other methods may be used by different types of
1434
transport = a_bzrdir.get_repository_transport(None)
1435
format_string = transport.get_bytes("format")
1436
return format_registry.get(format_string)
1437
except errors.NoSuchFile:
1438
raise errors.NoRepositoryPresent(a_bzrdir)
1440
raise errors.UnknownFormatError(format=format_string,
1444
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
1445
def register_format(klass, format):
1446
format_registry.register(format)
1449
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
1450
def unregister_format(klass, format):
1451
format_registry.remove(format)
1454
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
1455
def get_default_format(klass):
1456
"""Return the current default format."""
1457
return format_registry.get_default()
1459
def get_format_string(self):
1460
"""Return the ASCII format string that identifies this format.
1462
Note that in pre format ?? repositories the format string is
1463
not permitted nor written to disk.
1465
raise NotImplementedError(self.get_format_string)
1467
def get_format_description(self):
1468
"""Return the short description for this format."""
1469
raise NotImplementedError(self.get_format_description)
1471
def initialize(self, a_bzrdir, shared=False):
1472
"""Initialize a repository of this format in a_bzrdir.
1474
:param a_bzrdir: The bzrdir to put the new repository in it.
1475
:param shared: The repository should be initialized as a sharable one.
1476
:returns: The new repository object.
1478
This may raise UninitializableFormat if shared repository are not
1479
compatible the a_bzrdir.
1481
raise NotImplementedError(self.initialize)
1483
def is_supported(self):
1484
"""Is this format supported?
1486
Supported formats must be initializable and openable.
1487
Unsupported formats may not support initialization or committing or
1488
some other features depending on the reason for not being supported.
1492
def is_deprecated(self):
1493
"""Is this format deprecated?
1495
Deprecated formats may trigger a user-visible warning recommending
1496
the user to upgrade. They are still fully supported.
1500
def network_name(self):
1501
"""A simple byte string uniquely identifying this format for RPC calls.
1503
MetaDir repository formats use their disk format string to identify the
1504
repository over the wire. All in one formats such as bzr < 0.8, and
1505
foreign formats like svn/git and hg should use some marker which is
1506
unique and immutable.
1508
raise NotImplementedError(self.network_name)
1510
def check_conversion_target(self, target_format):
1511
if self.rich_root_data and not target_format.rich_root_data:
1512
raise errors.BadConversionTarget(
1513
'Does not support rich root data.', target_format,
1515
if (self.supports_tree_reference and
1516
not getattr(target_format, 'supports_tree_reference', False)):
1517
raise errors.BadConversionTarget(
1518
'Does not support nested trees', target_format,
1521
def open(self, a_bzrdir, _found=False):
1522
"""Return an instance of this format for the bzrdir a_bzrdir.
1524
_found is a private parameter, do not use it.
1526
raise NotImplementedError(self.open)
1528
def _run_post_repo_init_hooks(self, repository, a_bzrdir, shared):
1529
from bzrlib.bzrdir import BzrDir, RepoInitHookParams
1530
hooks = BzrDir.hooks['post_repo_init']
1533
params = RepoInitHookParams(repository, self, a_bzrdir, shared)
1538
class MetaDirRepositoryFormat(RepositoryFormat):
1539
"""Common base class for the new repositories using the metadir layout."""
1541
rich_root_data = False
1542
supports_tree_reference = False
1543
supports_external_lookups = False
1544
supports_leaving_lock = True
1547
def _matchingbzrdir(self):
1548
matching = bzrdir.BzrDirMetaFormat1()
1549
matching.repository_format = self
1553
super(MetaDirRepositoryFormat, self).__init__()
1555
def _create_control_files(self, a_bzrdir):
1556
"""Create the required files and the initial control_files object."""
1557
# FIXME: RBC 20060125 don't peek under the covers
1558
# NB: no need to escape relative paths that are url safe.
1559
repository_transport = a_bzrdir.get_repository_transport(self)
1560
control_files = lockable_files.LockableFiles(repository_transport,
1561
'lock', lockdir.LockDir)
1562
control_files.create_lock()
1563
return control_files
1565
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1566
"""Upload the initial blank content."""
1567
control_files = self._create_control_files(a_bzrdir)
1568
control_files.lock_write()
1569
transport = control_files._transport
1571
utf8_files += [('shared-storage', '')]
1573
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
1574
for (filename, content_stream) in files:
1575
transport.put_file(filename, content_stream,
1576
mode=a_bzrdir._get_file_mode())
1577
for (filename, content_bytes) in utf8_files:
1578
transport.put_bytes_non_atomic(filename, content_bytes,
1579
mode=a_bzrdir._get_file_mode())
1581
control_files.unlock()
1583
def network_name(self):
1584
"""Metadir formats have matching disk and network format strings."""
1585
return self.get_format_string()
1588
# formats which have no format string are not discoverable or independently
1589
# creatable on disk, so are not registered in format_registry. They're
1590
# all in bzrlib.repofmt.knitreponow. When an instance of one of these is
1591
# needed, it's constructed directly by the BzrDir. Non-native formats where
1592
# the repository is not separately opened are similar.
1594
format_registry.register_lazy(
1595
'Bazaar-NG Knit Repository Format 1',
1596
'bzrlib.repofmt.knitrepo',
1597
'RepositoryFormatKnit1',
1600
format_registry.register_lazy(
1601
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1602
'bzrlib.repofmt.knitrepo',
1603
'RepositoryFormatKnit3',
1606
format_registry.register_lazy(
1607
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
1608
'bzrlib.repofmt.knitrepo',
1609
'RepositoryFormatKnit4',
1612
# Pack-based formats. There is one format for pre-subtrees, and one for
1613
# post-subtrees to allow ease of testing.
1614
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
1615
format_registry.register_lazy(
1616
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
1617
'bzrlib.repofmt.knitpack_repo',
1618
'RepositoryFormatKnitPack1',
1620
format_registry.register_lazy(
1621
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
1622
'bzrlib.repofmt.knitpack_repo',
1623
'RepositoryFormatKnitPack3',
1625
format_registry.register_lazy(
1626
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
1627
'bzrlib.repofmt.knitpack_repo',
1628
'RepositoryFormatKnitPack4',
1630
format_registry.register_lazy(
1631
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
1632
'bzrlib.repofmt.knitpack_repo',
1633
'RepositoryFormatKnitPack5',
1635
format_registry.register_lazy(
1636
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
1637
'bzrlib.repofmt.knitpack_repo',
1638
'RepositoryFormatKnitPack5RichRoot',
1640
format_registry.register_lazy(
1641
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
1642
'bzrlib.repofmt.knitpack_repo',
1643
'RepositoryFormatKnitPack5RichRootBroken',
1645
format_registry.register_lazy(
1646
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
1647
'bzrlib.repofmt.knitpack_repo',
1648
'RepositoryFormatKnitPack6',
1650
format_registry.register_lazy(
1651
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
1652
'bzrlib.repofmt.knitpack_repo',
1653
'RepositoryFormatKnitPack6RichRoot',
1655
format_registry.register_lazy(
1656
'Bazaar repository format 2a (needs bzr 1.16 or later)\n',
1657
'bzrlib.repofmt.groupcompress_repo',
1658
'RepositoryFormat2a',
1661
# Development formats.
1662
# Check their docstrings to see if/when they are obsolete.
1663
format_registry.register_lazy(
1664
("Bazaar development format 2 with subtree support "
1665
"(needs bzr.dev from before 1.8)\n"),
1666
'bzrlib.repofmt.knitpack_repo',
1667
'RepositoryFormatPackDevelopment2Subtree',
1669
format_registry.register_lazy(
1670
'Bazaar development format 8\n',
1671
'bzrlib.repofmt.groupcompress_repo',
1672
'RepositoryFormat2aSubtree',
1676
class InterRepository(InterObject):
1677
"""This class represents operations taking place between two repositories.
1679
Its instances have methods like copy_content and fetch, and contain
1680
references to the source and target repositories these operations can be
1683
Often we will provide convenience methods on 'repository' which carry out
1684
operations with another repository - they will always forward to
1685
InterRepository.get(other).method_name(parameters).
1688
_walk_to_common_revisions_batch_size = 50
1690
"""The available optimised InterRepository types."""
1693
def copy_content(self, revision_id=None):
1694
"""Make a complete copy of the content in self into destination.
1696
This is a destructive operation! Do not use it on existing
1699
:param revision_id: Only copy the content needed to construct
1700
revision_id and its parents.
1703
self.target.set_make_working_trees(self.source.make_working_trees())
1704
except NotImplementedError:
1706
self.target.fetch(self.source, revision_id=revision_id)
1709
def fetch(self, revision_id=None, find_ghosts=False,
1711
"""Fetch the content required to construct revision_id.
1713
The content is copied from self.source to self.target.
1715
:param revision_id: if None all content is copied, if NULL_REVISION no
1719
ui.ui_factory.warn_experimental_format_fetch(self)
1720
from bzrlib.fetch import RepoFetcher
1721
# See <https://launchpad.net/bugs/456077> asking for a warning here
1722
if self.source._format.network_name() != self.target._format.network_name():
1723
ui.ui_factory.show_user_warning('cross_format_fetch',
1724
from_format=self.source._format,
1725
to_format=self.target._format)
1726
f = RepoFetcher(to_repository=self.target,
1727
from_repository=self.source,
1728
last_revision=revision_id,
1729
fetch_spec=fetch_spec,
1730
find_ghosts=find_ghosts)
1732
def _walk_to_common_revisions(self, revision_ids, if_present_ids=None):
1733
"""Walk out from revision_ids in source to revisions target has.
1735
:param revision_ids: The start point for the search.
1736
:return: A set of revision ids.
1738
target_graph = self.target.get_graph()
1739
revision_ids = frozenset(revision_ids)
1741
all_wanted_revs = revision_ids.union(if_present_ids)
1743
all_wanted_revs = revision_ids
1744
missing_revs = set()
1745
source_graph = self.source.get_graph()
1746
# ensure we don't pay silly lookup costs.
1747
searcher = source_graph._make_breadth_first_searcher(all_wanted_revs)
1748
null_set = frozenset([_mod_revision.NULL_REVISION])
1749
searcher_exhausted = False
1753
# Iterate the searcher until we have enough next_revs
1754
while len(next_revs) < self._walk_to_common_revisions_batch_size:
1756
next_revs_part, ghosts_part = searcher.next_with_ghosts()
1757
next_revs.update(next_revs_part)
1758
ghosts.update(ghosts_part)
1759
except StopIteration:
1760
searcher_exhausted = True
1762
# If there are ghosts in the source graph, and the caller asked for
1763
# them, make sure that they are present in the target.
1764
# We don't care about other ghosts as we can't fetch them and
1765
# haven't been asked to.
1766
ghosts_to_check = set(revision_ids.intersection(ghosts))
1767
revs_to_get = set(next_revs).union(ghosts_to_check)
1769
have_revs = set(target_graph.get_parent_map(revs_to_get))
1770
# we always have NULL_REVISION present.
1771
have_revs = have_revs.union(null_set)
1772
# Check if the target is missing any ghosts we need.
1773
ghosts_to_check.difference_update(have_revs)
1775
# One of the caller's revision_ids is a ghost in both the
1776
# source and the target.
1777
raise errors.NoSuchRevision(
1778
self.source, ghosts_to_check.pop())
1779
missing_revs.update(next_revs - have_revs)
1780
# Because we may have walked past the original stop point, make
1781
# sure everything is stopped
1782
stop_revs = searcher.find_seen_ancestors(have_revs)
1783
searcher.stop_searching_any(stop_revs)
1784
if searcher_exhausted:
1786
return searcher.get_result()
1789
def search_missing_revision_ids(self,
1790
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
1791
find_ghosts=True, revision_ids=None, if_present_ids=None):
1792
"""Return the revision ids that source has that target does not.
1794
:param revision_id: only return revision ids included by this
1796
:param revision_ids: return revision ids included by these
1797
revision_ids. NoSuchRevision will be raised if any of these
1798
revisions are not present.
1799
:param if_present_ids: like revision_ids, but will not cause
1800
NoSuchRevision if any of these are absent, instead they will simply
1801
not be in the result. This is useful for e.g. finding revisions
1802
to fetch for tags, which may reference absent revisions.
1803
:param find_ghosts: If True find missing revisions in deep history
1804
rather than just finding the surface difference.
1805
:return: A bzrlib.graph.SearchResult.
1807
if symbol_versioning.deprecated_passed(revision_id):
1808
symbol_versioning.warn(
1809
'search_missing_revision_ids(revision_id=...) was '
1810
'deprecated in 2.4. Use revision_ids=[...] instead.',
1811
DeprecationWarning, stacklevel=2)
1812
if revision_ids is not None:
1813
raise AssertionError(
1814
'revision_ids is mutually exclusive with revision_id')
1815
if revision_id is not None:
1816
revision_ids = [revision_id]
1818
# stop searching at found target revisions.
1819
if not find_ghosts and (revision_ids is not None or if_present_ids is
1821
return self._walk_to_common_revisions(revision_ids,
1822
if_present_ids=if_present_ids)
1823
# generic, possibly worst case, slow code path.
1824
target_ids = set(self.target.all_revision_ids())
1825
source_ids = self._present_source_revisions_for(
1826
revision_ids, if_present_ids)
1827
result_set = set(source_ids).difference(target_ids)
1828
return self.source.revision_ids_to_search_result(result_set)
1830
def _present_source_revisions_for(self, revision_ids, if_present_ids=None):
1831
"""Returns set of all revisions in ancestry of revision_ids present in
1834
:param revision_ids: if None, all revisions in source are returned.
1835
:param if_present_ids: like revision_ids, but if any/all of these are
1836
absent no error is raised.
1838
if revision_ids is not None or if_present_ids is not None:
1839
# First, ensure all specified revisions exist. Callers expect
1840
# NoSuchRevision when they pass absent revision_ids here.
1841
if revision_ids is None:
1842
revision_ids = set()
1843
if if_present_ids is None:
1844
if_present_ids = set()
1845
revision_ids = set(revision_ids)
1846
if_present_ids = set(if_present_ids)
1847
all_wanted_ids = revision_ids.union(if_present_ids)
1848
graph = self.source.get_graph()
1849
present_revs = set(graph.get_parent_map(all_wanted_ids))
1850
missing = revision_ids.difference(present_revs)
1852
raise errors.NoSuchRevision(self.source, missing.pop())
1853
found_ids = all_wanted_ids.intersection(present_revs)
1854
source_ids = [rev_id for (rev_id, parents) in
1855
graph.iter_ancestry(found_ids)
1856
if rev_id != _mod_revision.NULL_REVISION
1857
and parents is not None]
1859
source_ids = self.source.all_revision_ids()
1860
return set(source_ids)
1863
def _same_model(source, target):
1864
"""True if source and target have the same data representation.
1866
Note: this is always called on the base class; overriding it in a
1867
subclass will have no effect.
1870
InterRepository._assert_same_model(source, target)
1872
except errors.IncompatibleRepositories, e:
1876
def _assert_same_model(source, target):
1877
"""Raise an exception if two repositories do not use the same model.
1879
if source.supports_rich_root() != target.supports_rich_root():
1880
raise errors.IncompatibleRepositories(source, target,
1881
"different rich-root support")
1882
if source._serializer != target._serializer:
1883
raise errors.IncompatibleRepositories(source, target,
1884
"different serializers")
1887
def _get_repo_format_to_test(self):
1891
def is_compatible(cls, source, target):
1892
# The default implementation is compatible with everything
1896
InterRepository.register_optimiser(InterRepository)
1899
class CopyConverter(object):
1900
"""A repository conversion tool which just performs a copy of the content.
1902
This is slow but quite reliable.
1905
def __init__(self, target_format):
1906
"""Create a CopyConverter.
1908
:param target_format: The format the resulting repository should be.
1910
self.target_format = target_format
1912
def convert(self, repo, pb):
1913
"""Perform the conversion of to_convert, giving feedback via pb.
1915
:param to_convert: The disk object to convert.
1916
:param pb: a progress bar to use for progress information.
1918
pb = ui.ui_factory.nested_progress_bar()
1921
# this is only useful with metadir layouts - separated repo content.
1922
# trigger an assertion if not such
1923
repo._format.get_format_string()
1924
self.repo_dir = repo.bzrdir
1925
pb.update('Moving repository to repository.backup')
1926
self.repo_dir.transport.move('repository', 'repository.backup')
1927
backup_transport = self.repo_dir.transport.clone('repository.backup')
1928
repo._format.check_conversion_target(self.target_format)
1929
self.source_repo = repo._format.open(self.repo_dir,
1931
_override_transport=backup_transport)
1932
pb.update('Creating new repository')
1933
converted = self.target_format.initialize(self.repo_dir,
1934
self.source_repo.is_shared())
1935
converted.lock_write()
1937
pb.update('Copying content')
1938
self.source_repo.copy_content_into(converted)
1941
pb.update('Deleting old repository content')
1942
self.repo_dir.transport.delete_tree('repository.backup')
1943
ui.ui_factory.note('repository converted')
1947
def _strip_NULL_ghosts(revision_graph):
1948
"""Also don't use this. more compatibility code for unmigrated clients."""
1949
# Filter ghosts, and null:
1950
if _mod_revision.NULL_REVISION in revision_graph:
1951
del revision_graph[_mod_revision.NULL_REVISION]
1952
for key, parents in revision_graph.items():
1953
revision_graph[key] = tuple(parent for parent in parents if parent
1955
return revision_graph
1958
def _iter_for_revno(repo, partial_history_cache, stop_index=None,
1959
stop_revision=None):
1960
"""Extend the partial history to include a given index
1962
If a stop_index is supplied, stop when that index has been reached.
1963
If a stop_revision is supplied, stop when that revision is
1964
encountered. Otherwise, stop when the beginning of history is
1967
:param stop_index: The index which should be present. When it is
1968
present, history extension will stop.
1969
:param stop_revision: The revision id which should be present. When
1970
it is encountered, history extension will stop.
1972
start_revision = partial_history_cache[-1]
1973
iterator = repo.iter_reverse_revision_history(start_revision)
1975
#skip the last revision in the list
1978
if (stop_index is not None and
1979
len(partial_history_cache) > stop_index):
1981
if partial_history_cache[-1] == stop_revision:
1983
revision_id = iterator.next()
1984
partial_history_cache.append(revision_id)
1985
except StopIteration:
1990
class _LazyListJoin(object):
1991
"""An iterable yielding the contents of many lists as one list.
1993
Each iterator made from this will reflect the current contents of the lists
1994
at the time the iterator is made.
1996
This is used by Repository's _make_parents_provider implementation so that
1999
pp = repo._make_parents_provider() # uses a list of fallback repos
2000
pp.add_fallback_repository(other_repo) # appends to that list
2001
result = pp.get_parent_map(...)
2002
# The result will include revs from other_repo
2005
def __init__(self, *list_parts):
2006
self.list_parts = list_parts
2010
for list_part in self.list_parts:
2011
full_list.extend(list_part)
2012
return iter(full_list)
added
removed
bzrlib/repository.py
