1
# Copyright (C) 2005-2011 Canonical Ltd
1
# Copyright (C) 2005 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
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(), """
32
revision as _mod_revision,
33
testament as _mod_testament,
37
from bzrlib.bundle import serializer
38
from bzrlib.i18n import gettext
47
from bzrlib.decorators import needs_read_lock, needs_write_lock, only_raises
48
from bzrlib.inter import InterObject
49
from bzrlib.lock import _RelockDebugMixin, LogicalLockResult
50
from bzrlib.trace import (
51
log_exception_quietly, note, mutter, mutter_callsite, warning)
54
# Old formats display a warning, but only once
55
_deprecation_warning_done = False
58
class IsInWriteGroupError(errors.InternalBzrError):
60
_fmt = "May not refresh_data of repo %(repo)s while in a write group."
62
def __init__(self, repo):
63
errors.InternalBzrError.__init__(self, repo=repo)
66
class CommitBuilder(object):
67
"""Provides an interface to build up a commit.
69
This allows describing a tree to be committed without needing to
70
know the internals of the format of the repository.
73
# all clients should supply tree roots.
74
record_root_entry = True
75
# whether this commit builder supports the record_entry_contents interface
76
supports_record_entry_contents = False
78
def __init__(self, repository, parents, config, timestamp=None,
79
timezone=None, committer=None, revprops=None,
80
revision_id=None, lossy=False):
81
"""Initiate a CommitBuilder.
83
:param repository: Repository to commit to.
84
:param parents: Revision ids of the parents of the new revision.
85
:param timestamp: Optional timestamp recorded for commit.
86
:param timezone: Optional timezone for timestamp.
87
:param committer: Optional committer to set for commit.
88
:param revprops: Optional dictionary of revision properties.
89
:param revision_id: Optional revision id.
90
:param lossy: Whether to discard data that can not be natively
91
represented, when pushing to a foreign VCS
97
self._committer = self._config.username()
98
elif not isinstance(committer, unicode):
99
self._committer = committer.decode() # throw if non-ascii
101
self._committer = committer
103
self._new_revision_id = revision_id
104
self.parents = parents
105
self.repository = repository
108
if revprops is not None:
109
self._validate_revprops(revprops)
110
self._revprops.update(revprops)
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
118
self._timezone = osutils.local_time_offset()
120
self._timezone = int(timezone)
122
self._generate_revision_if_needed()
124
def any_changes(self):
125
"""Return True if any entries were changed.
127
This includes merge-only changes. It is the core for the --unchanged
130
:return: True if any changes have occured.
132
raise NotImplementedError(self.any_changes)
134
def _validate_unicode_text(self, text, context):
135
"""Verify things like commit messages don't have bogus characters."""
137
raise ValueError('Invalid value for %s: %r' % (context, text))
139
def _validate_revprops(self, revprops):
140
for key, value in revprops.iteritems():
141
# We know that the XML serializers do not round trip '\r'
142
# correctly, so refuse to accept them
143
if not isinstance(value, basestring):
144
raise ValueError('revision property (%s) is not a valid'
145
' (unicode) string: %r' % (key, value))
146
self._validate_unicode_text(value,
147
'revision property (%s)' % (key,))
149
def commit(self, message):
150
"""Make the actual commit.
152
:return: The revision id of the recorded revision.
154
raise NotImplementedError(self.commit)
157
"""Abort the commit that is being built.
159
raise NotImplementedError(self.abort)
161
def revision_tree(self):
162
"""Return the tree that was just committed.
164
After calling commit() this can be called to get a
165
RevisionTree representing the newly committed tree. This is
166
preferred to calling Repository.revision_tree() because that may
167
require deserializing the inventory, while we already have a copy in
170
raise NotImplementedError(self.revision_tree)
172
def finish_inventory(self):
173
"""Tell the builder that the inventory is finished.
175
:return: The inventory id in the repository, which can be used with
176
repository.get_inventory.
178
raise NotImplementedError(self.finish_inventory)
180
def _gen_revision_id(self):
181
"""Return new revision-id."""
182
return generate_ids.gen_revision_id(self._committer, self._timestamp)
184
def _generate_revision_if_needed(self):
185
"""Create a revision id if None was supplied.
187
If the repository can not support user-specified revision ids
188
they should override this function and raise CannotSetRevisionId
189
if _new_revision_id is not None.
191
:raises: CannotSetRevisionId
193
if self._new_revision_id is None:
194
self._new_revision_id = self._gen_revision_id()
195
self.random_revid = True
197
self.random_revid = False
199
def will_record_deletes(self):
200
"""Tell the commit builder that deletes are being notified.
202
This enables the accumulation of an inventory delta; for the resulting
203
commit to be valid, deletes against the basis MUST be recorded via
204
builder.record_delete().
206
raise NotImplementedError(self.will_record_deletes)
208
def record_iter_changes(self, tree, basis_revision_id, iter_changes):
209
"""Record a new tree via iter_changes.
211
:param tree: The tree to obtain text contents from for changed objects.
212
:param basis_revision_id: The revision id of the tree the iter_changes
213
has been generated against. Currently assumed to be the same
214
as self.parents[0] - if it is not, errors may occur.
215
:param iter_changes: An iter_changes iterator with the changes to apply
216
to basis_revision_id. The iterator must not include any items with
217
a current kind of None - missing items must be either filtered out
218
or errored-on beefore record_iter_changes sees the item.
219
:return: A generator of (file_id, relpath, fs_hash) tuples for use with
222
raise NotImplementedError(self.record_iter_changes)
225
class RepositoryWriteLockResult(LogicalLockResult):
226
"""The result of write locking a repository.
228
:ivar repository_token: The token obtained from the underlying lock, or
230
:ivar unlock: A callable which will unlock the lock.
233
def __init__(self, unlock, repository_token):
234
LogicalLockResult.__init__(self, unlock)
235
self.repository_token = repository_token
238
return "RepositoryWriteLockResult(%s, %s)" % (self.repository_token,
242
######################################################################
246
class Repository(_RelockDebugMixin, controldir.ControlComponent):
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from copy import deepcopy
18
from cStringIO import StringIO
19
from unittest import TestSuite
20
import xml.sax.saxutils
23
from bzrlib.decorators import needs_read_lock, needs_write_lock
24
import bzrlib.errors as errors
25
from bzrlib.errors import InvalidRevisionId
26
from bzrlib.lockable_files import LockableFiles
27
from bzrlib.osutils import safe_unicode
28
from bzrlib.revision import NULL_REVISION
29
from bzrlib.store import copy_all
30
from bzrlib.store.weave import WeaveStore
31
from bzrlib.store.text import TextStore
32
from bzrlib.symbol_versioning import *
33
from bzrlib.trace import mutter
34
from bzrlib.tree import RevisionTree
35
from bzrlib.testament import Testament
36
from bzrlib.tree import EmptyTree
41
class Repository(object):
247
42
"""Repository holding history for one or more branches.
249
44
The repository holds and retrieves historical information including
250
45
revisions and file history. It's normally accessed only by the Branch,
251
46
which views a particular line of development through that history.
253
See VersionedFileRepository in bzrlib.vf_repository for the
254
base class for most Bazaar repositories.
48
The Repository builds on top of Stores and a Transport, which respectively
49
describe the disk data format and the way of accessing the (possibly
257
def abort_write_group(self, suppress_errors=False):
258
"""Commit the contents accrued within the current write group.
260
:param suppress_errors: if true, abort_write_group will catch and log
261
unexpected errors that happen during the abort, rather than
262
allowing them to propagate. Defaults to False.
264
:seealso: start_write_group.
266
if self._write_group is not self.get_transaction():
267
# has an unlock or relock occured ?
270
'(suppressed) mismatched lock context and write group. %r, %r',
271
self._write_group, self.get_transaction())
273
raise errors.BzrError(
274
'mismatched lock context and write group. %r, %r' %
275
(self._write_group, self.get_transaction()))
277
self._abort_write_group()
278
except Exception, exc:
279
self._write_group = None
280
if not suppress_errors:
282
mutter('abort_write_group failed')
283
log_exception_quietly()
284
note(gettext('bzr: ERROR (ignored): %s'), exc)
285
self._write_group = None
287
def _abort_write_group(self):
288
"""Template method for per-repository write group cleanup.
290
This is called during abort before the write group is considered to be
291
finished and should cleanup any internal state accrued during the write
292
group. There is no requirement that data handed to the repository be
293
*not* made available - this is not a rollback - but neither should any
294
attempt be made to ensure that data added is fully commited. Abort is
295
invoked when an error has occured so futher disk or network operations
296
may not be possible or may error and if possible should not be
300
def add_fallback_repository(self, repository):
301
"""Add a repository to use for looking up data not held locally.
303
:param repository: A repository.
305
raise NotImplementedError(self.add_fallback_repository)
307
def _check_fallback_repository(self, repository):
308
"""Check that this repository can fallback to repository safely.
310
Raise an error if not.
312
:param repository: A repository to fallback to.
314
return InterRepository._assert_same_model(self, repository)
54
def _all_possible_ids(self):
55
"""Return all the possible revisions that we could find."""
56
return self.get_inventory_weave().names()
316
59
def all_revision_ids(self):
317
"""Returns a list of all the revision ids in the repository.
319
This is conceptually deprecated because code should generally work on
320
the graph reachable from a particular revision, and ignore any other
321
revisions that might be present. There is no direct replacement
324
if 'evil' in debug.debug_flags:
325
mutter_callsite(2, "all_revision_ids is linear with history.")
326
return self._all_revision_ids()
328
def _all_revision_ids(self):
329
"""Returns a list of all the revision ids in the repository.
331
These are in as much topological order as the underlying store can
334
raise NotImplementedError(self._all_revision_ids)
336
def break_lock(self):
337
"""Break a lock if one is present from another instance.
339
Uses the ui factory to ask for confirmation if the lock may be from
342
self.control_files.break_lock()
60
"""Returns a list of all the revision ids in the repository.
62
These are in as much topological order as the underlying store can
63
present: for weaves ghosts may lead to a lack of correctness until
64
the reweave updates the parents list.
66
result = self._all_possible_ids()
67
return self._eliminate_revisions_not_present(result)
70
def _eliminate_revisions_not_present(self, revision_ids):
71
"""Check every revision id in revision_ids to see if we have it.
73
Returns a set of the present revisions.
76
for id in revision_ids:
77
if self.has_revision(id):
345
def create(controldir):
346
"""Construct the current default format repository in controldir."""
347
return RepositoryFormat.get_default_format().initialize(controldir)
83
"""Construct the current default format repository in a_bzrdir."""
84
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
349
def __init__(self, _format, controldir, control_files):
86
def __init__(self, _format, a_bzrdir, control_files, revision_store):
350
87
"""instantiate a Repository.
352
89
:param _format: The format of the repository on disk.
353
:param controldir: The ControlDir of the repository.
354
:param control_files: Control files to use for locking, etc.
90
:param a_bzrdir: The BzrDir of the repository.
92
In the future we will have a single api for all stores for
93
getting file texts, inventories and revisions, then
94
this construct will accept instances of those things.
356
# In the future we will have a single api for all stores for
357
# getting file texts, inventories and revisions, then
358
# this construct will accept instances of those things.
359
super(Repository, self).__init__()
360
97
self._format = _format
361
98
# the following are part of the public API for Repository:
362
self.bzrdir = controldir
99
self.bzrdir = a_bzrdir
363
100
self.control_files = control_files
365
self._write_group = None
366
# Additional places to query for data.
367
self._fallback_repositories = []
370
def user_transport(self):
371
return self.bzrdir.user_transport
374
def control_transport(self):
375
return self._transport
378
if self._fallback_repositories:
379
return '%s(%r, fallback_repositories=%r)' % (
380
self.__class__.__name__,
382
self._fallback_repositories)
384
return '%s(%r)' % (self.__class__.__name__,
387
def _has_same_fallbacks(self, other_repo):
388
"""Returns true if the repositories have the same fallbacks."""
389
my_fb = self._fallback_repositories
390
other_fb = other_repo._fallback_repositories
391
if len(my_fb) != len(other_fb):
393
for f, g in zip(my_fb, other_fb):
394
if not f.has_same_location(g):
398
def has_same_location(self, other):
399
"""Returns a boolean indicating if this repository is at the same
400
location as another repository.
402
This might return False even when two repository objects are accessing
403
the same physical repository via different URLs.
405
if self.__class__ is not other.__class__:
407
return (self.control_url == other.control_url)
409
def is_in_write_group(self):
410
"""Return True if there is an open write group.
412
:seealso: start_write_group.
414
return self._write_group is not None
417
return self.control_files.is_locked()
419
def is_write_locked(self):
420
"""Return True if this object is write locked."""
421
return self.is_locked() and self.control_files._lock_mode == 'w'
423
def lock_write(self, token=None):
424
"""Lock this repository for writing.
426
This causes caching within the repository obejct to start accumlating
427
data during reads, and allows a 'write_group' to be obtained. Write
428
groups must be used for actual data insertion.
430
A token should be passed in if you know that you have locked the object
431
some other way, and need to synchronise this object's state with that
434
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
436
:param token: if this is already locked, then lock_write will fail
437
unless the token matches the existing lock.
438
:returns: a token if this instance supports tokens, otherwise None.
439
:raises TokenLockingNotSupported: when a token is given but this
440
instance doesn't support using token locks.
441
:raises MismatchedToken: if the specified token doesn't match the token
442
of the existing lock.
443
:seealso: start_write_group.
444
:return: A RepositoryWriteLockResult.
446
locked = self.is_locked()
447
token = self.control_files.lock_write(token=token)
449
self._warn_if_deprecated()
451
for repo in self._fallback_repositories:
452
# Writes don't affect fallback repos
455
return RepositoryWriteLockResult(self.unlock, token)
101
self.revision_store = revision_store
103
def lock_write(self):
104
self.control_files.lock_write()
457
106
def lock_read(self):
458
"""Lock the repository for read operations.
460
:return: An object with an unlock method which will release the lock
463
locked = self.is_locked()
464
107
self.control_files.lock_read()
466
self._warn_if_deprecated()
468
for repo in self._fallback_repositories:
471
return LogicalLockResult(self.unlock)
473
def get_physical_lock_status(self):
474
return self.control_files.get_physical_lock_status()
476
def leave_lock_in_place(self):
477
"""Tell this repository not to release the physical lock when this
480
If lock_write doesn't return a token, then this method is not supported.
482
self.control_files.leave_in_place()
484
def dont_leave_lock_in_place(self):
485
"""Tell this repository to release the physical lock when this
486
object is unlocked, even if it didn't originally acquire it.
488
If lock_write doesn't return a token, then this method is not supported.
490
self.control_files.dont_leave_in_place()
493
def gather_stats(self, revid=None, committers=None):
494
"""Gather statistics from a revision id.
496
:param revid: The revision id to gather statistics from, if None, then
497
no revision specific statistics are gathered.
498
:param committers: Optional parameter controlling whether to grab
499
a count of committers from the revision specific statistics.
500
:return: A dictionary of statistics. Currently this contains:
501
committers: The number of committers if requested.
502
firstrev: A tuple with timestamp, timezone for the penultimate left
503
most ancestor of revid, if revid is not the NULL_REVISION.
504
latestrev: A tuple with timestamp, timezone for revid, if revid is
505
not the NULL_REVISION.
506
revisions: The total revision count in the repository.
507
size: An estimate disk size of the repository in bytes.
510
if revid and committers:
511
result['committers'] = 0
512
if revid and revid != _mod_revision.NULL_REVISION:
513
graph = self.get_graph()
515
all_committers = set()
516
revisions = [r for (r, p) in graph.iter_ancestry([revid])
517
if r != _mod_revision.NULL_REVISION]
520
# ignore the revisions in the middle - just grab first and last
521
revisions = revisions[0], revisions[-1]
522
for revision in self.get_revisions(revisions):
523
if not last_revision:
524
last_revision = revision
526
all_committers.add(revision.committer)
527
first_revision = revision
529
result['committers'] = len(all_committers)
530
result['firstrev'] = (first_revision.timestamp,
531
first_revision.timezone)
532
result['latestrev'] = (last_revision.timestamp,
533
last_revision.timezone)
536
def find_branches(self, using=False):
537
"""Find branches underneath this repository.
539
This will include branches inside other branches.
541
:param using: If True, list only branches using this repository.
543
if using and not self.is_shared():
544
return self.bzrdir.list_branches()
545
class Evaluator(object):
548
self.first_call = True
550
def __call__(self, controldir):
551
# On the first call, the parameter is always the controldir
552
# containing the current repo.
553
if not self.first_call:
555
repository = controldir.open_repository()
556
except errors.NoRepositoryPresent:
559
return False, ([], repository)
560
self.first_call = False
561
value = (controldir.list_branches(), None)
565
for branches, repository in controldir.ControlDir.find_bzrdirs(
566
self.user_transport, evaluate=Evaluator()):
567
if branches is not None:
569
if not using and repository is not None:
570
ret.extend(repository.find_branches())
574
def search_missing_revision_ids(self, other,
575
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
576
find_ghosts=True, revision_ids=None, if_present_ids=None,
110
def missing_revision_ids(self, other, revision_id=None):
578
111
"""Return the revision ids that other has that this does not.
580
113
These are returned in topological order.
582
115
revision_id: only return revision ids included by revision_id.
584
if symbol_versioning.deprecated_passed(revision_id):
585
symbol_versioning.warn(
586
'search_missing_revision_ids(revision_id=...) was '
587
'deprecated in 2.4. Use revision_ids=[...] instead.',
588
DeprecationWarning, stacklevel=3)
589
if revision_ids is not None:
590
raise AssertionError(
591
'revision_ids is mutually exclusive with revision_id')
592
if revision_id is not None:
593
revision_ids = [revision_id]
594
return InterRepository.get(other, self).search_missing_revision_ids(
595
find_ghosts=find_ghosts, revision_ids=revision_ids,
596
if_present_ids=if_present_ids, limit=limit)
117
return InterRepository.get(other, self).missing_revision_ids(revision_id)
602
123
For instance, if the repository is at URL/.bzr/repository,
603
124
Repository.open(URL) -> a Repository instance.
605
control = controldir.ControlDir.open(base)
126
control = bzrlib.bzrdir.BzrDir.open(base)
606
127
return control.open_repository()
608
def copy_content_into(self, destination, revision_id=None):
129
def copy_content_into(self, destination, revision_id=None, basis=None):
609
130
"""Make a complete copy of the content in self into destination.
611
This is a destructive operation! Do not use it on existing
132
This is a destructive operation! Do not use it on existing
614
return InterRepository.get(self, destination).copy_content(revision_id)
616
def commit_write_group(self):
617
"""Commit the contents accrued within the current write group.
619
:seealso: start_write_group.
621
:return: it may return an opaque hint that can be passed to 'pack'.
623
if self._write_group is not self.get_transaction():
624
# has an unlock or relock occured ?
625
raise errors.BzrError('mismatched lock context %r and '
627
(self.get_transaction(), self._write_group))
628
result = self._commit_write_group()
629
self._write_group = None
632
def _commit_write_group(self):
633
"""Template method for per-repository write group cleanup.
635
This is called before the write group is considered to be
636
finished and should ensure that all data handed to the repository
637
for writing during the write group is safely committed (to the
638
extent possible considering file system caching etc).
641
def suspend_write_group(self):
642
raise errors.UnsuspendableWriteGroup(self)
644
def refresh_data(self):
645
"""Re-read any data needed to synchronise with disk.
647
This method is intended to be called after another repository instance
648
(such as one used by a smart server) has inserted data into the
649
repository. On all repositories this will work outside of write groups.
650
Some repository formats (pack and newer for bzrlib native formats)
651
support refresh_data inside write groups. If called inside a write
652
group on a repository that does not support refreshing in a write group
653
IsInWriteGroupError will be raised.
657
def resume_write_group(self, tokens):
658
if not self.is_write_locked():
659
raise errors.NotWriteLocked(self)
660
if self._write_group:
661
raise errors.BzrError('already in a write group')
662
self._resume_write_group(tokens)
663
# so we can detect unlock/relock - the write group is now entered.
664
self._write_group = self.get_transaction()
666
def _resume_write_group(self, tokens):
667
raise errors.UnsuspendableWriteGroup(self)
669
def fetch(self, source, revision_id=None, find_ghosts=False,
135
return InterRepository.get(self, destination).copy_content(revision_id, basis)
137
def fetch(self, source, revision_id=None, pb=None):
671
138
"""Fetch the content required to construct revision_id from source.
673
If revision_id is None and fetch_spec is None, then all content is
676
fetch() may not be used when the repository is in a write group -
677
either finish the current write group before using fetch, or use
678
fetch before starting the write group.
680
:param find_ghosts: Find and copy revisions in the source that are
681
ghosts in the target (and not reachable directly by walking out to
682
the first-present revision in target from revision_id).
683
:param revision_id: If specified, all the content needed for this
684
revision ID will be copied to the target. Fetch will determine for
685
itself which content needs to be copied.
686
:param fetch_spec: If specified, a SearchResult or
687
PendingAncestryResult that describes which revisions to copy. This
688
allows copying multiple heads at once. Mutually exclusive with
691
if fetch_spec is not None and revision_id is not None:
692
raise AssertionError(
693
"fetch_spec and revision_id are mutually exclusive.")
694
if self.is_in_write_group():
695
raise errors.InternalBzrError(
696
"May not fetch while in a write group.")
697
# fast path same-url fetch operations
698
# TODO: lift out to somewhere common with RemoteRepository
699
# <https://bugs.launchpad.net/bzr/+bug/401646>
700
if (self.has_same_location(source)
701
and fetch_spec is None
702
and self._has_same_fallbacks(source)):
703
# check that last_revision is in 'from' and then return a
705
if (revision_id is not None and
706
not _mod_revision.is_null(revision_id)):
707
self.get_revision(revision_id)
709
inter = InterRepository.get(source, self)
710
return inter.fetch(revision_id=revision_id,
711
find_ghosts=find_ghosts, fetch_spec=fetch_spec)
713
def create_bundle(self, target, base, fileobj, format=None):
714
return serializer.write_bundle(self, target, base, fileobj, format)
716
def get_commit_builder(self, branch, parents, config, timestamp=None,
717
timezone=None, committer=None, revprops=None,
718
revision_id=None, lossy=False):
719
"""Obtain a CommitBuilder for this repository.
721
:param branch: Branch to commit to.
722
:param parents: Revision ids of the parents of the new revision.
723
:param config: Configuration to use.
724
:param timestamp: Optional timestamp recorded for commit.
725
:param timezone: Optional timezone for timestamp.
726
:param committer: Optional committer to set for commit.
727
:param revprops: Optional dictionary of revision properties.
728
:param revision_id: Optional revision id.
729
:param lossy: Whether to discard data that can not be natively
730
represented, when pushing to a foreign VCS
732
raise NotImplementedError(self.get_commit_builder)
734
@only_raises(errors.LockNotHeld, errors.LockBroken)
140
If revision_id is None all content is copied.
142
return InterRepository.get(source, self).fetch(revision_id=revision_id,
735
145
def unlock(self):
736
if (self.control_files._lock_count == 1 and
737
self.control_files._lock_mode == 'w'):
738
if self._write_group is not None:
739
self.abort_write_group()
740
self.control_files.unlock()
741
raise errors.BzrError(
742
'Must end write groups before releasing write locks.')
743
146
self.control_files.unlock()
744
if self.control_files._lock_count == 0:
745
for repo in self._fallback_repositories:
749
def clone(self, controldir, revision_id=None):
750
"""Clone this repository into controldir using the current format.
149
def clone(self, a_bzrdir, revision_id=None, basis=None):
150
"""Clone this repository into a_bzrdir using the current format.
752
152
Currently no check is made that the format of this repository and
753
153
the bzrdir format are compatible. FIXME RBC 20060201.
755
:return: The newly created destination repository.
757
# TODO: deprecate after 0.16; cloning this with all its settings is
758
# probably not very useful -- mbp 20070423
759
dest_repo = self._create_sprouting_repo(
760
controldir, shared=self.is_shared())
761
self.copy_content_into(dest_repo, revision_id)
764
def start_write_group(self):
765
"""Start a write group in the repository.
767
Write groups are used by repositories which do not have a 1:1 mapping
768
between file ids and backend store to manage the insertion of data from
769
both fetch and commit operations.
771
A write lock is required around the start_write_group/commit_write_group
772
for the support of lock-requiring repository formats.
774
One can only insert data into a repository inside a write group.
778
if not self.is_write_locked():
779
raise errors.NotWriteLocked(self)
780
if self._write_group:
781
raise errors.BzrError('already in a write group')
782
self._start_write_group()
783
# so we can detect unlock/relock - the write group is now entered.
784
self._write_group = self.get_transaction()
786
def _start_write_group(self):
787
"""Template method for per-repository write group startup.
789
This is called before the write group is considered to be
794
def sprout(self, to_bzrdir, revision_id=None):
795
"""Create a descendent repository for new development.
797
Unlike clone, this does not copy the settings of the repository.
799
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
800
dest_repo.fetch(self, revision_id=revision_id)
803
def _create_sprouting_repo(self, a_bzrdir, shared):
804
155
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
805
156
# use target default format.
806
dest_repo = a_bzrdir.create_repository()
157
result = a_bzrdir.create_repository()
158
# FIXME RBC 20060209 split out the repository type to avoid this check ?
159
elif isinstance(a_bzrdir._format,
160
(bzrlib.bzrdir.BzrDirFormat4,
161
bzrlib.bzrdir.BzrDirFormat5,
162
bzrlib.bzrdir.BzrDirFormat6)):
163
result = a_bzrdir.open_repository()
808
# Most control formats need the repository to be specifically
809
# created, but on some old all-in-one formats it's not needed
811
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
812
except errors.UninitializableFormat:
813
dest_repo = a_bzrdir.open_repository()
165
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
166
self.copy_content_into(result, revision_id, basis)
817
169
def has_revision(self, revision_id):
818
"""True if this repository has a copy of the revision."""
819
return revision_id in self.has_revisions((revision_id,))
822
def has_revisions(self, revision_ids):
823
"""Probe to find out the presence of multiple revisions.
825
:param revision_ids: An iterable of revision_ids.
826
:return: A set of the revision_ids that were present.
828
raise NotImplementedError(self.has_revisions)
170
"""True if this branch has a copy of the revision.
172
This does not necessarily imply the revision is merge
173
or on the mainline."""
174
return (revision_id is None
175
or self.revision_store.has_id(revision_id))
178
def get_revision_xml_file(self, revision_id):
179
"""Return XML file object for revision object."""
180
if not revision_id or not isinstance(revision_id, basestring):
181
raise InvalidRevisionId(revision_id=revision_id, branch=self)
183
return self.revision_store.get(revision_id)
184
except (IndexError, KeyError):
185
raise bzrlib.errors.NoSuchRevision(self, revision_id)
188
def get_revision_xml(self, revision_id):
189
return self.get_revision_xml_file(revision_id).read()
831
192
def get_revision(self, revision_id):
832
"""Return the Revision object for a named revision."""
833
return self.get_revisions([revision_id])[0]
835
def get_revision_reconcile(self, revision_id):
836
"""'reconcile' helper routine that allows access to a revision always.
838
This variant of get_revision does not cross check the weave graph
839
against the revision one as get_revision does: but it should only
840
be used by reconcile, or reconcile-alike commands that are correcting
841
or testing the revision graph.
843
raise NotImplementedError(self.get_revision_reconcile)
845
def get_revisions(self, revision_ids):
846
"""Get many revisions at once.
848
Repositories that need to check data on every revision read should
849
subclass this method.
851
raise NotImplementedError(self.get_revisions)
853
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
854
"""Produce a generator of revision deltas.
856
Note that the input is a sequence of REVISIONS, not revision_ids.
857
Trees will be held in memory until the generator exits.
858
Each delta is relative to the revision's lefthand predecessor.
860
:param specific_fileids: if not None, the result is filtered
861
so that only those file-ids, their parents and their
862
children are included.
864
# Get the revision-ids of interest
865
required_trees = set()
866
for revision in revisions:
867
required_trees.add(revision.revision_id)
868
required_trees.update(revision.parent_ids[:1])
870
# Get the matching filtered trees. Note that it's more
871
# efficient to pass filtered trees to changes_from() rather
872
# than doing the filtering afterwards. changes_from() could
873
# arguably do the filtering itself but it's path-based, not
874
# file-id based, so filtering before or afterwards is
876
if specific_fileids is None:
877
trees = dict((t.get_revision_id(), t) for
878
t in self.revision_trees(required_trees))
880
trees = dict((t.get_revision_id(), t) for
881
t in self._filtered_revision_trees(required_trees,
884
# Calculate the deltas
885
for revision in revisions:
886
if not revision.parent_ids:
887
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
889
old_tree = trees[revision.parent_ids[0]]
890
yield trees[revision.revision_id].changes_from(old_tree)
193
"""Return the Revision object for a named revision"""
194
xml_file = self.get_revision_xml_file(revision_id)
197
r = bzrlib.xml5.serializer_v5.read_revision(xml_file)
198
except SyntaxError, e:
199
raise bzrlib.errors.BzrError('failed to unpack revision_xml',
203
assert r.revision_id == revision_id
893
def get_revision_delta(self, revision_id, specific_fileids=None):
894
"""Return the delta for one revision.
896
The delta is relative to the left-hand predecessor of the
899
:param specific_fileids: if not None, the result is filtered
900
so that only those file-ids, their parents and their
901
children are included.
903
r = self.get_revision(revision_id)
904
return list(self.get_deltas_for_revisions([r],
905
specific_fileids=specific_fileids))[0]
207
def get_revision_sha1(self, revision_id):
208
"""Hash the stored value of a revision, and return it."""
209
# In the future, revision entries will be signed. At that
210
# point, it is probably best *not* to include the signature
211
# in the revision hash. Because that lets you re-sign
212
# the revision, (add signatures/remove signatures) and still
213
# have all hash pointers stay consistent.
214
# But for now, just hash the contents.
215
return bzrlib.osutils.sha_file(self.get_revision_xml_file(revision_id))
907
217
@needs_write_lock
908
218
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
909
signature = gpg_strategy.sign(plaintext)
910
self.add_signature_text(revision_id, signature)
912
def add_signature_text(self, revision_id, signature):
913
"""Store a signature text for a revision.
915
:param revision_id: Revision id of the revision
916
:param signature: Signature text.
918
raise NotImplementedError(self.add_signature_text)
920
def _find_parent_ids_of_revisions(self, revision_ids):
921
"""Find all parent ids that are mentioned in the revision graph.
923
:return: set of revisions that are parents of revision_ids which are
924
not part of revision_ids themselves
926
parent_map = self.get_parent_map(revision_ids)
928
map(parent_ids.update, parent_map.itervalues())
929
parent_ids.difference_update(revision_ids)
930
parent_ids.discard(_mod_revision.NULL_REVISION)
933
def iter_files_bytes(self, desired_files):
934
"""Iterate through file versions.
936
Files will not necessarily be returned in the order they occur in
937
desired_files. No specific order is guaranteed.
939
Yields pairs of identifier, bytes_iterator. identifier is an opaque
940
value supplied by the caller as part of desired_files. It should
941
uniquely identify the file version in the caller's context. (Examples:
942
an index number or a TreeTransform trans_id.)
944
:param desired_files: a list of (file_id, revision_id, identifier)
947
raise NotImplementedError(self.iter_files_bytes)
949
def get_rev_id_for_revno(self, revno, known_pair):
950
"""Return the revision id of a revno, given a later (revno, revid)
951
pair in the same history.
953
:return: if found (True, revid). If the available history ran out
954
before reaching the revno, then this returns
955
(False, (closest_revno, closest_revid)).
957
known_revno, known_revid = known_pair
958
partial_history = [known_revid]
959
distance_from_known = known_revno - revno
960
if distance_from_known < 0:
962
'requested revno (%d) is later than given known revno (%d)'
963
% (revno, known_revno))
219
self.revision_store.add(StringIO(gpg_strategy.sign(plaintext)),
222
def fileid_involved_between_revs(self, from_revid, to_revid):
223
"""Find file_id(s) which are involved in the changes between revisions.
225
This determines the set of revisions which are involved, and then
226
finds all file ids affected by those revisions.
228
# TODO: jam 20060119 This code assumes that w.inclusions will
229
# always be correct. But because of the presence of ghosts
230
# it is possible to be wrong.
231
# One specific example from Robert Collins:
232
# Two branches, with revisions ABC, and AD
233
# C is a ghost merge of D.
234
# Inclusions doesn't recognize D as an ancestor.
235
# If D is ever merged in the future, the weave
236
# won't be fixed, because AD never saw revision C
237
# to cause a conflict which would force a reweave.
238
w = self.get_inventory_weave()
239
from_set = set(w.inclusions([w.lookup(from_revid)]))
240
to_set = set(w.inclusions([w.lookup(to_revid)]))
241
included = to_set.difference(from_set)
242
changed = map(w.idx_to_name, included)
243
return self._fileid_involved_by_set(changed)
245
def fileid_involved(self, last_revid=None):
246
"""Find all file_ids modified in the ancestry of last_revid.
248
:param last_revid: If None, last_revision() will be used.
250
w = self.get_inventory_weave()
252
changed = set(w._names)
254
included = w.inclusions([w.lookup(last_revid)])
255
changed = map(w.idx_to_name, included)
256
return self._fileid_involved_by_set(changed)
258
def fileid_involved_by_set(self, changes):
259
"""Find all file_ids modified by the set of revisions passed in.
261
:param changes: A set() of revision ids
263
# TODO: jam 20060119 This line does *nothing*, remove it.
264
# or better yet, change _fileid_involved_by_set so
265
# that it takes the inventory weave, rather than
266
# pulling it out by itself.
267
return self._fileid_involved_by_set(changes)
269
def _fileid_involved_by_set(self, changes):
270
"""Find the set of file-ids affected by the set of revisions.
272
:param changes: A set() of revision ids.
273
:return: A set() of file ids.
275
This peaks at the Weave, interpreting each line, looking to
276
see if it mentions one of the revisions. And if so, includes
277
the file id mentioned.
278
This expects both the Weave format, and the serialization
279
to have a single line per file/directory, and to have
280
fileid="" and revision="" on that line.
282
assert isinstance(self._format, (RepositoryFormat5,
285
RepositoryFormatKnit1)), \
286
"fileid_involved only supported for branches which store inventory as unnested xml"
288
w = self.get_inventory_weave()
290
for line in w._weave:
292
# it is ugly, but it is due to the weave structure
293
if not isinstance(line, basestring): continue
295
start = line.find('file_id="')+9
296
if start < 9: continue
297
end = line.find('"', start)
299
file_id = xml.sax.saxutils.unescape(line[start:end])
301
# check if file_id is already present
302
if file_id in file_ids: continue
304
start = line.find('revision="')+10
305
if start < 10: continue
306
end = line.find('"', start)
308
revision_id = xml.sax.saxutils.unescape(line[start:end])
310
if revision_id in changes:
311
file_ids.add(file_id)
315
def get_inventory_weave(self):
316
return self.control_weaves.get_weave('inventory',
317
self.get_transaction())
320
def get_inventory(self, revision_id):
321
"""Get Inventory object by hash."""
322
xml = self.get_inventory_xml(revision_id)
323
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
326
def get_inventory_xml(self, revision_id):
327
"""Get inventory XML as a file object."""
966
self, partial_history, stop_index=distance_from_known)
967
except errors.RevisionNotPresent, err:
968
if err.revision_id == known_revid:
969
# The start revision (known_revid) wasn't found.
971
# This is a stacked repository with no fallbacks, or a there's a
972
# left-hand ghost. Either way, even though the revision named in
973
# the error isn't in this repo, we know it's the next step in this
975
partial_history.append(err.revision_id)
976
if len(partial_history) <= distance_from_known:
977
# Didn't find enough history to get a revid for the revno.
978
earliest_revno = known_revno - len(partial_history) + 1
979
return (False, (earliest_revno, partial_history[-1]))
980
if len(partial_history) - 1 > distance_from_known:
981
raise AssertionError('_iter_for_revno returned too much history')
982
return (True, partial_history[-1])
984
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
985
def iter_reverse_revision_history(self, revision_id):
986
"""Iterate backwards through revision ids in the lefthand history
988
:param revision_id: The revision id to start with. All its lefthand
989
ancestors will be traversed.
329
assert isinstance(revision_id, basestring), type(revision_id)
330
iw = self.get_inventory_weave()
331
return iw.get_text(iw.lookup(revision_id))
333
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
336
def get_inventory_sha1(self, revision_id):
337
"""Return the sha1 hash of the inventory entry
991
graph = self.get_graph()
992
stop_revisions = (None, _mod_revision.NULL_REVISION)
993
return graph.iter_lefthand_ancestry(revision_id, stop_revisions)
339
return self.get_revision(revision_id).inventory_sha1
342
def get_revision_inventory(self, revision_id):
343
"""Return inventory of a past revision."""
344
# TODO: Unify this with get_inventory()
345
# bzr 0.0.6 and later imposes the constraint that the inventory_id
346
# must be the same as its revision, so this is trivial.
347
if revision_id is None:
348
# This does not make sense: if there is no revision,
349
# then it is the current tree inventory surely ?!
350
# and thus get_root_id() is something that looks at the last
351
# commit on the branch, and the get_root_id is an inventory check.
352
raise NotImplementedError
353
# return Inventory(self.get_root_id())
355
return self.get_inventory(revision_id)
995
358
def is_shared(self):
996
359
"""Return True if this repository is flagged as a shared repository."""
997
raise NotImplementedError(self.is_shared)
1000
def reconcile(self, other=None, thorough=False):
1001
"""Reconcile this repository."""
1002
from bzrlib.reconcile import RepoReconciler
1003
reconciler = RepoReconciler(self, thorough=thorough)
1004
reconciler.reconcile()
1007
def _refresh_data(self):
1008
"""Helper called from lock_* to ensure coherency with disk.
1010
The default implementation does nothing; it is however possible
1011
for repositories to maintain loaded indices across multiple locks
1012
by checking inside their implementation of this method to see
1013
whether their indices are still valid. This depends of course on
1014
the disk format being validatable in this manner. This method is
1015
also called by the refresh_data() public interface to cause a refresh
1016
to occur while in a write lock so that data inserted by a smart server
1017
push operation is visible on the client's instance of the physical
360
# FIXME format 4-6 cannot be shared, this is technically faulty.
361
return self.control_files._transport.has('shared-storage')
1021
363
@needs_read_lock
1022
364
def revision_tree(self, revision_id):
1023
365
"""Return Tree for a revision on this branch.
1025
`revision_id` may be NULL_REVISION for the empty tree revision.
1027
raise NotImplementedError(self.revision_tree)
1029
def revision_trees(self, revision_ids):
1030
"""Return Trees for revisions in this repository.
1032
:param revision_ids: a sequence of revision-ids;
1033
a revision-id may not be None or 'null:'
1035
raise NotImplementedError(self.revision_trees)
367
`revision_id` may be None for the null revision, in which case
368
an `EmptyTree` is returned."""
369
# TODO: refactor this to use an existing revision object
370
# so we don't need to read it in twice.
371
if revision_id is None or revision_id == NULL_REVISION:
374
inv = self.get_revision_inventory(revision_id)
375
return RevisionTree(self, inv, revision_id)
1037
377
@needs_read_lock
1038
@symbol_versioning.deprecated_method(
1039
symbol_versioning.deprecated_in((2, 4, 0)))
1040
def get_ancestry(self, revision_id, topo_sorted=True):
378
def get_ancestry(self, revision_id):
1041
379
"""Return a list of revision-ids integrated by a revision.
1043
The first element of the list is always None, indicating the origin
1044
revision. This might change when we have history horizons, or
1045
perhaps we should have a new API.
1047
381
This is topologically sorted.
1049
if 'evil' in debug.debug_flags:
1050
mutter_callsite(2, "get_ancestry is linear with history.")
1051
if _mod_revision.is_null(revision_id):
383
if revision_id is None:
1053
385
if not self.has_revision(revision_id):
1054
386
raise errors.NoSuchRevision(self, revision_id)
1055
graph = self.get_graph()
1057
search = graph._make_breadth_first_searcher([revision_id])
387
w = self.get_inventory_weave()
388
return [None] + map(w.idx_to_name,
389
w.inclusions([w.lookup(revision_id)]))
392
def print_file(self, file, revision_id):
393
"""Print `file` to stdout.
395
FIXME RBC 20060125 as John Meinel points out this is a bad api
396
- it writes to stdout, it assumes that that is valid etc. Fix
397
by creating a new more flexible convenience function.
399
tree = self.revision_tree(revision_id)
400
# use inventory as it was in that revision
401
file_id = tree.inventory.path2id(file)
403
raise BzrError("%r is not present in revision %s" % (file, revno))
1060
found, ghosts = search.next_with_ghosts()
1061
except StopIteration:
1064
if _mod_revision.NULL_REVISION in keys:
1065
keys.remove(_mod_revision.NULL_REVISION)
1067
parent_map = graph.get_parent_map(keys)
1068
keys = tsort.topo_sort(parent_map)
1069
return [None] + list(keys)
1071
def pack(self, hint=None, clean_obsolete_packs=False):
1072
"""Compress the data within the repository.
1074
This operation only makes sense for some repository types. For other
1075
types it should be a no-op that just returns.
1077
This stub method does not require a lock, but subclasses should use
1078
@needs_write_lock as this is a long running call it's reasonable to
1079
implicitly lock for the user.
1081
:param hint: If not supplied, the whole repository is packed.
1082
If supplied, the repository may use the hint parameter as a
1083
hint for the parts of the repository to pack. A hint can be
1084
obtained from the result of commit_write_group(). Out of
1085
date hints are simply ignored, because concurrent operations
1086
can obsolete them rapidly.
1088
:param clean_obsolete_packs: Clean obsolete packs immediately after
405
revno = self.revision_id_to_revno(revision_id)
406
except errors.NoSuchRevision:
407
# TODO: This should not be BzrError,
408
# but NoSuchFile doesn't fit either
409
raise BzrError('%r is not present in revision %s'
410
% (file, revision_id))
412
raise BzrError('%r is not present in revision %s'
414
tree.print_file(file_id)
1092
416
def get_transaction(self):
1093
417
return self.control_files.get_transaction()
1095
def get_parent_map(self, revision_ids):
1096
"""See graph.StackedParentsProvider.get_parent_map"""
1097
raise NotImplementedError(self.get_parent_map)
1099
def _get_parent_map_no_fallbacks(self, revision_ids):
1100
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1101
# revisions index works in keys; this just works in revisions
1102
# therefore wrap and unwrap
1105
for revision_id in revision_ids:
1106
if revision_id == _mod_revision.NULL_REVISION:
1107
result[revision_id] = ()
1108
elif revision_id is None:
1109
raise ValueError('get_parent_map(None) is not valid')
1111
query_keys.append((revision_id ,))
1112
vf = self.revisions.without_fallbacks()
1113
for ((revision_id,), parent_keys) in \
1114
vf.get_parent_map(query_keys).iteritems():
1116
result[revision_id] = tuple([parent_revid
1117
for (parent_revid,) in parent_keys])
1119
result[revision_id] = (_mod_revision.NULL_REVISION,)
1122
def _make_parents_provider(self):
1123
if not self._format.supports_external_lookups:
1125
return graph.StackedParentsProvider(_LazyListJoin(
1126
[self._make_parents_provider_unstacked()],
1127
self._fallback_repositories))
1129
def _make_parents_provider_unstacked(self):
1130
return graph.CallableToParentsProviderAdapter(
1131
self._get_parent_map_no_fallbacks)
1134
def get_known_graph_ancestry(self, revision_ids):
1135
"""Return the known graph for a set of revision ids and their ancestors.
1137
raise NotImplementedError(self.get_known_graph_ancestry)
1139
def get_file_graph(self):
1140
"""Return the graph walker for files."""
1141
raise NotImplementedError(self.get_file_graph)
1143
def get_graph(self, other_repository=None):
1144
"""Return the graph walker for this repository format"""
1145
parents_provider = self._make_parents_provider()
1146
if (other_repository is not None and
1147
not self.has_same_location(other_repository)):
1148
parents_provider = graph.StackedParentsProvider(
1149
[parents_provider, other_repository._make_parents_provider()])
1150
return graph.Graph(parents_provider)
1152
def revision_ids_to_search_result(self, result_set):
1153
"""Convert a set of revision ids to a graph SearchResult."""
1154
result_parents = set()
1155
for parents in self.get_graph().get_parent_map(
1156
result_set).itervalues():
1157
result_parents.update(parents)
1158
included_keys = result_set.intersection(result_parents)
1159
start_keys = result_set.difference(included_keys)
1160
exclude_keys = result_parents.difference(result_set)
1161
result = graph.SearchResult(start_keys, exclude_keys,
1162
len(result_set), result_set)
1165
419
@needs_write_lock
1166
420
def set_make_working_trees(self, new_value):
1167
421
"""Set the policy flag for making working trees when creating branches.
1172
426
:param new_value: True to restore the default, False to disable making
1175
raise NotImplementedError(self.set_make_working_trees)
429
# FIXME: split out into a new class/strategy ?
430
if isinstance(self._format, (RepositoryFormat4,
433
raise NotImplementedError(self.set_make_working_trees)
436
self.control_files._transport.delete('no-working-trees')
437
except errors.NoSuchFile:
440
self.control_files.put_utf8('no-working-trees', '')
1177
442
def make_working_trees(self):
1178
443
"""Returns the policy for making working trees on new branches."""
1179
raise NotImplementedError(self.make_working_trees)
444
# FIXME: split out into a new class/strategy ?
445
if isinstance(self._format, (RepositoryFormat4,
449
return not self.control_files._transport.has('no-working-trees')
1181
451
@needs_write_lock
1182
452
def sign_revision(self, revision_id, gpg_strategy):
1183
testament = _mod_testament.Testament.from_revision(self, revision_id)
1184
plaintext = testament.as_short_text()
453
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1185
454
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1188
def verify_revision(self, revision_id, gpg_strategy):
1189
"""Verify the signature on a revision.
1191
:param revision_id: the revision to verify
1192
:gpg_strategy: the GPGStrategy object to used
1194
:return: gpg.SIGNATURE_VALID or a failed SIGNATURE_ value
1196
if not self.has_signature_for_revision_id(revision_id):
1197
return gpg.SIGNATURE_NOT_SIGNED, None
1198
signature = self.get_signature_text(revision_id)
1200
testament = _mod_testament.Testament.from_revision(self, revision_id)
1201
plaintext = testament.as_short_text()
1203
return gpg_strategy.verify(signature, plaintext)
1205
def has_signature_for_revision_id(self, revision_id):
1206
"""Query for a revision signature for revision_id in the repository."""
1207
raise NotImplementedError(self.has_signature_for_revision_id)
1209
def get_signature_text(self, revision_id):
1210
"""Return the text for a signature."""
1211
raise NotImplementedError(self.get_signature_text)
1213
def check(self, revision_ids=None, callback_refs=None, check_repo=True):
1214
"""Check consistency of all history of given revision_ids.
1216
Different repository implementations should override _check().
1218
:param revision_ids: A non-empty list of revision_ids whose ancestry
1219
will be checked. Typically the last revision_id of a branch.
1220
:param callback_refs: A dict of check-refs to resolve and callback
1221
the check/_check method on the items listed as wanting the ref.
1223
:param check_repo: If False do not check the repository contents, just
1224
calculate the data callback_refs requires and call them back.
1226
return self._check(revision_ids=revision_ids, callback_refs=callback_refs,
1227
check_repo=check_repo)
1229
def _check(self, revision_ids=None, callback_refs=None, check_repo=True):
1230
raise NotImplementedError(self.check)
1232
def _warn_if_deprecated(self, branch=None):
1233
if not self._format.is_deprecated():
1235
global _deprecation_warning_done
1236
if _deprecation_warning_done:
1240
conf = config.GlobalConfig()
1242
conf = branch.get_config()
1243
if conf.suppress_warning('format_deprecation'):
1245
warning("Format %s for %s is deprecated -"
1246
" please use 'bzr upgrade' to get better performance"
1247
% (self._format, self.bzrdir.transport.base))
1249
_deprecation_warning_done = True
1251
def supports_rich_root(self):
1252
return self._format.rich_root_data
1254
def _check_ascii_revisionid(self, revision_id, method):
1255
"""Private helper for ascii-only repositories."""
1256
# weave repositories refuse to store revisionids that are non-ascii.
1257
if revision_id is not None:
1258
# weaves require ascii revision ids.
1259
if isinstance(revision_id, unicode):
1261
revision_id.encode('ascii')
1262
except UnicodeEncodeError:
1263
raise errors.NonAsciiRevisionId(method, self)
1266
revision_id.decode('ascii')
1267
except UnicodeDecodeError:
1268
raise errors.NonAsciiRevisionId(method, self)
457
class AllInOneRepository(Repository):
458
"""Legacy support - the repository behaviour for all-in-one branches."""
460
def __init__(self, _format, a_bzrdir, revision_store):
461
# we reuse one control files instance.
462
dir_mode = a_bzrdir._control_files._dir_mode
463
file_mode = a_bzrdir._control_files._file_mode
465
def get_weave(name, prefixed=False):
467
name = safe_unicode(name)
470
relpath = a_bzrdir._control_files._escape(name)
471
weave_transport = a_bzrdir._control_files._transport.clone(relpath)
472
ws = WeaveStore(weave_transport, prefixed=prefixed,
475
if a_bzrdir._control_files._transport.should_cache():
476
ws.enable_cache = True
479
def get_store(name, compressed=True, prefixed=False):
480
# FIXME: This approach of assuming stores are all entirely compressed
481
# or entirely uncompressed is tidy, but breaks upgrade from
482
# some existing branches where there's a mixture; we probably
483
# still want the option to look for both.
484
relpath = a_bzrdir._control_files._escape(name)
485
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
486
prefixed=prefixed, compressed=compressed,
489
#if self._transport.should_cache():
490
# cache_path = os.path.join(self.cache_root, name)
491
# os.mkdir(cache_path)
492
# store = bzrlib.store.CachedStore(store, cache_path)
495
# not broken out yet because the controlweaves|inventory_store
496
# and text_store | weave_store bits are still different.
497
if isinstance(_format, RepositoryFormat4):
498
self.inventory_store = get_store('inventory-store')
499
self.text_store = get_store('text-store')
500
elif isinstance(_format, RepositoryFormat5):
501
self.control_weaves = get_weave('')
502
self.weave_store = get_weave('weaves')
503
elif isinstance(_format, RepositoryFormat6):
504
self.control_weaves = get_weave('')
505
self.weave_store = get_weave('weaves', prefixed=True)
507
raise errors.BzrError('unreachable code: unexpected repository'
509
revision_store.register_suffix('sig')
510
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, revision_store)
1271
513
class MetaDirRepository(Repository):
1272
"""Repositories in the new meta-dir layout.
1274
:ivar _transport: Transport for access to repository control files,
1275
typically pointing to .bzr/repository.
1278
def __init__(self, _format, a_bzrdir, control_files):
1279
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
1280
self._transport = control_files._transport
1282
def is_shared(self):
1283
"""Return True if this repository is flagged as a shared repository."""
1284
return self._transport.has('shared-storage')
1287
def set_make_working_trees(self, new_value):
1288
"""Set the policy flag for making working trees when creating branches.
1290
This only applies to branches that use this repository.
1292
The default is 'True'.
1293
:param new_value: True to restore the default, False to disable making
1298
self._transport.delete('no-working-trees')
1299
except errors.NoSuchFile:
514
"""Repositories in the new meta-dir layout."""
516
def __init__(self, _format, a_bzrdir, control_files, revision_store):
517
super(MetaDirRepository, self).__init__(_format,
522
dir_mode = self.control_files._dir_mode
523
file_mode = self.control_files._file_mode
525
def get_weave(name, prefixed=False):
527
name = safe_unicode(name)
530
relpath = self.control_files._escape(name)
531
weave_transport = self.control_files._transport.clone(relpath)
532
ws = WeaveStore(weave_transport, prefixed=prefixed,
535
if self.control_files._transport.should_cache():
536
ws.enable_cache = True
539
if isinstance(self._format, RepositoryFormat7):
540
self.control_weaves = get_weave('')
541
self.weave_store = get_weave('weaves', prefixed=True)
542
elif isinstance(self._format, RepositoryFormatKnit1):
543
self.control_weaves = get_weave('')
544
self.weave_store = get_weave('knits', prefixed=True)
1302
self._transport.put_bytes('no-working-trees', '',
1303
mode=self.bzrdir._get_file_mode())
1305
def make_working_trees(self):
1306
"""Returns the policy for making working trees on new branches."""
1307
return not self._transport.has('no-working-trees')
1310
class RepositoryFormatRegistry(controldir.ControlComponentFormatRegistry):
1311
"""Repository format registry."""
1313
def get_default(self):
1314
"""Return the current default format."""
1315
return controldir.format_registry.make_bzrdir('default').repository_format
1318
network_format_registry = registry.FormatRegistry()
1319
"""Registry of formats indexed by their network name.
1321
The network name for a repository format is an identifier that can be used when
1322
referring to formats with smart server operations. See
1323
RepositoryFormat.network_name() for more detail.
1327
format_registry = RepositoryFormatRegistry(network_format_registry)
1328
"""Registry of formats, indexed by their BzrDirMetaFormat format string.
1330
This can contain either format instances themselves, or classes/factories that
1331
can be called to obtain one.
1335
#####################################################################
1336
# Repository Formats
1338
class RepositoryFormat(controldir.ControlComponentFormat):
546
raise errors.BzrError('unreachable code: unexpected repository'
550
class RepositoryFormat(object):
1339
551
"""A repository format.
1341
Formats provide four things:
553
Formats provide three things:
1342
554
* An initialization routine to construct repository data on disk.
1343
* a optional format string which is used when the BzrDir supports
555
* a format string which is used when the BzrDir supports versioned
1345
557
* an open routine which returns a Repository instance.
1346
* A network name for referring to the format in smart server RPC
1349
There is one and only one Format subclass for each on-disk format. But
1350
there can be one Repository subclass that is used for several different
1351
formats. The _format attribute on a Repository instance can be used to
1352
determine the disk format.
1354
Formats are placed in a registry by their format string for reference
1355
during opening. These should be subclasses of RepositoryFormat for
559
Formats are placed in an dict by their format string for reference
560
during opening. These should be subclasses of RepositoryFormat
1358
563
Once a format is deprecated, just deprecate the initialize and open
1359
methods on the format class. Do not deprecate the object, as the
1360
object may be created even when a repository instance hasn't been
564
methods on the format class. Do not deprecate the object, as the
565
object will be created every system load.
1363
567
Common instance attributes:
1364
_matchingbzrdir - the controldir format that the repository format was
568
_matchingbzrdir - the bzrdir format that the repository format was
1365
569
originally written to work with. This can be used if manually
1366
570
constructing a bzrdir and repository, or more commonly for test suite
1370
# Set to True or False in derived classes. True indicates that the format
1371
# supports ghosts gracefully.
1372
supports_ghosts = None
1373
# Can this repository be given external locations to lookup additional
1374
# data. Set to True or False in derived classes.
1375
supports_external_lookups = None
1376
# Does this format support CHK bytestring lookups. Set to True or False in
1378
supports_chks = None
1379
# Should fetch trigger a reconcile after the fetch? Only needed for
1380
# some repository formats that can suffer internal inconsistencies.
1381
_fetch_reconcile = False
1382
# Does this format have < O(tree_size) delta generation. Used to hint what
1383
# code path for commit, amongst other things.
1385
# Does doing a pack operation compress data? Useful for the pack UI command
1386
# (so if there is one pack, the operation can still proceed because it may
1387
# help), and for fetching when data won't have come from the same
1389
pack_compresses = False
1390
# Does the repository storage understand references to trees?
1391
supports_tree_reference = None
1392
# Is the format experimental ?
1393
experimental = False
1394
# Does this repository format escape funky characters, or does it create
1395
# files with similar names as the versioned files in its contents on disk
1397
supports_funky_characters = None
1398
# Does this repository format support leaving locks?
1399
supports_leaving_lock = None
1400
# Does this format support the full VersionedFiles interface?
1401
supports_full_versioned_files = None
1402
# Does this format support signing revision signatures?
1403
supports_revision_signatures = True
1404
# Can the revision graph have incorrect parents?
1405
revision_graph_can_have_wrong_parents = None
1406
# Does this format support rich root data?
1407
rich_root_data = None
1408
# Does this format support explicitly versioned directories?
1409
supports_versioned_directories = None
1410
# Can other repositories be nested into one of this format?
1411
supports_nesting_repositories = None
1414
return "%s()" % self.__class__.__name__
1416
def __eq__(self, other):
1417
# format objects are generally stateless
1418
return isinstance(other, self.__class__)
1420
def __ne__(self, other):
1421
return not self == other
574
_default_format = None
575
"""The default format used for new repositories."""
578
"""The known formats."""
1424
581
def find_format(klass, a_bzrdir):
1425
"""Return the format for the repository object in a_bzrdir.
1427
This is used by bzr native formats that have a "format" file in
1428
the repository. Other methods may be used by different types of
582
"""Return the format for the repository object in a_bzrdir."""
1432
584
transport = a_bzrdir.get_repository_transport(None)
1433
format_string = transport.get_bytes("format")
1434
return format_registry.get(format_string)
585
format_string = transport.get("format").read()
586
return klass._formats[format_string]
1435
587
except errors.NoSuchFile:
1436
588
raise errors.NoRepositoryPresent(a_bzrdir)
1437
589
except KeyError:
1438
raise errors.UnknownFormatError(format=format_string,
1442
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
1443
def register_format(klass, format):
1444
format_registry.register(format)
1447
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
1448
def unregister_format(klass, format):
1449
format_registry.remove(format)
1452
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
590
raise errors.UnknownFormatError(format_string)
1453
593
def get_default_format(klass):
1454
594
"""Return the current default format."""
1455
return format_registry.get_default()
595
return klass._default_format
1457
597
def get_format_string(self):
1458
598
"""Return the ASCII format string that identifies this format.
1460
Note that in pre format ?? repositories the format string is
600
Note that in pre format ?? repositories the format string is
1461
601
not permitted nor written to disk.
1463
603
raise NotImplementedError(self.get_format_string)
1465
def get_format_description(self):
1466
"""Return the short description for this format."""
1467
raise NotImplementedError(self.get_format_description)
1469
def initialize(self, controldir, shared=False):
1470
"""Initialize a repository of this format in controldir.
1472
:param controldir: The controldir to put the new repository in it.
605
def _get_revision_store(self, repo_transport, control_files):
606
"""Return the revision store object for this a_bzrdir."""
607
raise NotImplementedError(self._get_revision_store)
609
def _get_rev_store(self,
615
"""Common logic for getting a revision store for a repository.
617
see self._get_revision_store for the method to
618
get the store for a repository.
621
name = safe_unicode(name)
624
dir_mode = control_files._dir_mode
625
file_mode = control_files._file_mode
626
revision_store =TextStore(transport.clone(name),
628
compressed=compressed,
631
revision_store.register_suffix('sig')
632
return revision_store
634
def initialize(self, a_bzrdir, shared=False):
635
"""Initialize a repository of this format in a_bzrdir.
637
:param a_bzrdir: The bzrdir to put the new repository in it.
1473
638
:param shared: The repository should be initialized as a sharable one.
1474
:returns: The new repository object.
1476
640
This may raise UninitializableFormat if shared repository are not
1477
compatible the controldir.
641
compatible the a_bzrdir.
1479
raise NotImplementedError(self.initialize)
1481
644
def is_supported(self):
1482
645
"""Is this format supported?
1484
647
Supported formats must be initializable and openable.
1485
Unsupported formats may not support initialization or committing or
648
Unsupported formats may not support initialization or committing or
1486
649
some other features depending on the reason for not being supported.
1490
def is_deprecated(self):
1491
"""Is this format deprecated?
1493
Deprecated formats may trigger a user-visible warning recommending
1494
the user to upgrade. They are still fully supported.
1498
def network_name(self):
1499
"""A simple byte string uniquely identifying this format for RPC calls.
1501
MetaDir repository formats use their disk format string to identify the
1502
repository over the wire. All in one formats such as bzr < 0.8, and
1503
foreign formats like svn/git and hg should use some marker which is
1504
unique and immutable.
1506
raise NotImplementedError(self.network_name)
1508
def check_conversion_target(self, target_format):
1509
if self.rich_root_data and not target_format.rich_root_data:
1510
raise errors.BadConversionTarget(
1511
'Does not support rich root data.', target_format,
1513
if (self.supports_tree_reference and
1514
not getattr(target_format, 'supports_tree_reference', False)):
1515
raise errors.BadConversionTarget(
1516
'Does not support nested trees', target_format,
1519
def open(self, controldir, _found=False):
1520
"""Return an instance of this format for a controldir.
653
def open(self, a_bzrdir, _found=False):
654
"""Return an instance of this format for the bzrdir a_bzrdir.
1522
656
_found is a private parameter, do not use it.
1524
658
raise NotImplementedError(self.open)
1526
def _run_post_repo_init_hooks(self, repository, controldir, shared):
1527
from bzrlib.controldir import ControlDir, RepoInitHookParams
1528
hooks = ControlDir.hooks['post_repo_init']
1531
params = RepoInitHookParams(repository, self, controldir, shared)
661
def register_format(klass, format):
662
klass._formats[format.get_format_string()] = format
665
def set_default_format(klass, format):
666
klass._default_format = format
669
def unregister_format(klass, format):
670
assert klass._formats[format.get_format_string()] is format
671
del klass._formats[format.get_format_string()]
674
class PreSplitOutRepositoryFormat(RepositoryFormat):
675
"""Base class for the pre split out repository formats."""
677
def initialize(self, a_bzrdir, shared=False, _internal=False):
678
"""Create a weave repository.
680
TODO: when creating split out bzr branch formats, move this to a common
681
base for Format5, Format6. or something like that.
683
from bzrlib.weavefile import write_weave_v5
684
from bzrlib.weave import Weave
687
raise errors.IncompatibleFormat(self, a_bzrdir._format)
690
# always initialized when the bzrdir is.
691
return self.open(a_bzrdir, _found=True)
693
# Create an empty weave
695
bzrlib.weavefile.write_weave_v5(Weave(), sio)
696
empty_weave = sio.getvalue()
698
mutter('creating repository in %s.', a_bzrdir.transport.base)
699
dirs = ['revision-store', 'weaves']
700
lock_file = 'branch-lock'
701
files = [('inventory.weave', StringIO(empty_weave)),
704
# FIXME: RBC 20060125 dont peek under the covers
705
# NB: no need to escape relative paths that are url safe.
706
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock')
707
control_files.lock_write()
708
control_files._transport.mkdir_multi(dirs,
709
mode=control_files._dir_mode)
711
for file, content in files:
712
control_files.put(file, content)
714
control_files.unlock()
715
return self.open(a_bzrdir, _found=True)
717
def open(self, a_bzrdir, _found=False):
718
"""See RepositoryFormat.open()."""
720
# we are being called directly and must probe.
721
raise NotImplementedError
723
repo_transport = a_bzrdir.get_repository_transport(None)
724
control_files = a_bzrdir._control_files
725
revision_store = self._get_revision_store(repo_transport, control_files)
726
return AllInOneRepository(_format=self,
728
revision_store=revision_store)
731
class RepositoryFormat4(PreSplitOutRepositoryFormat):
732
"""Bzr repository format 4.
734
This repository format has:
736
- TextStores for texts, inventories,revisions.
738
This format is deprecated: it indexes texts using a text id which is
739
removed in format 5; initializationa and write support for this format
744
super(RepositoryFormat4, self).__init__()
745
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
747
def initialize(self, url, shared=False, _internal=False):
748
"""Format 4 branches cannot be created."""
749
raise errors.UninitializableFormat(self)
751
def is_supported(self):
752
"""Format 4 is not supported.
754
It is not supported because the model changed from 4 to 5 and the
755
conversion logic is expensive - so doing it on the fly was not
760
def _get_revision_store(self, repo_transport, control_files):
761
"""See RepositoryFormat._get_revision_store()."""
762
return self._get_rev_store(repo_transport,
767
class RepositoryFormat5(PreSplitOutRepositoryFormat):
768
"""Bzr control format 5.
770
This repository format has:
771
- weaves for file texts and inventory
773
- TextStores for revisions and signatures.
777
super(RepositoryFormat5, self).__init__()
778
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
780
def _get_revision_store(self, repo_transport, control_files):
781
"""See RepositoryFormat._get_revision_store()."""
782
"""Return the revision store object for this a_bzrdir."""
783
return self._get_rev_store(repo_transport,
789
class RepositoryFormat6(PreSplitOutRepositoryFormat):
790
"""Bzr control format 6.
792
This repository format has:
793
- weaves for file texts and inventory
794
- hash subdirectory based stores.
795
- TextStores for revisions and signatures.
799
super(RepositoryFormat6, self).__init__()
800
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
802
def _get_revision_store(self, repo_transport, control_files):
803
"""See RepositoryFormat._get_revision_store()."""
804
return self._get_rev_store(repo_transport,
1536
811
class MetaDirRepositoryFormat(RepositoryFormat):
1537
"""Common base class for the new repositories using the metadir layout."""
1539
rich_root_data = False
1540
supports_tree_reference = False
1541
supports_external_lookups = False
1542
supports_leaving_lock = True
1543
supports_nesting_repositories = True
1546
def _matchingbzrdir(self):
1547
matching = bzrdir.BzrDirMetaFormat1()
1548
matching.repository_format = self
812
"""Common base class for the new repositories using the metadir layour."""
1551
814
def __init__(self):
1552
815
super(MetaDirRepositoryFormat, self).__init__()
816
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1554
818
def _create_control_files(self, a_bzrdir):
1555
819
"""Create the required files and the initial control_files object."""
1556
# FIXME: RBC 20060125 don't peek under the covers
820
# FIXME: RBC 20060125 dont peek under the covers
1557
821
# NB: no need to escape relative paths that are url safe.
1558
823
repository_transport = a_bzrdir.get_repository_transport(self)
1559
control_files = lockable_files.LockableFiles(repository_transport,
1560
'lock', lockdir.LockDir)
1561
control_files.create_lock()
824
repository_transport.put(lock_file, StringIO()) # TODO get the file mode from the bzrdir lock files., mode=file_mode)
825
control_files = LockableFiles(repository_transport, 'lock')
1562
826
return control_files
828
def _get_revision_store(self, repo_transport, control_files):
829
"""See RepositoryFormat._get_revision_store()."""
830
return self._get_rev_store(repo_transport,
837
def open(self, a_bzrdir, _found=False, _override_transport=None):
838
"""See RepositoryFormat.open().
840
:param _override_transport: INTERNAL USE ONLY. Allows opening the
841
repository at a slightly different url
842
than normal. I.e. during 'upgrade'.
845
format = RepositoryFormat.find_format(a_bzrdir)
846
assert format.__class__ == self.__class__
847
if _override_transport is not None:
848
repo_transport = _override_transport
850
repo_transport = a_bzrdir.get_repository_transport(None)
851
control_files = LockableFiles(repo_transport, 'lock')
852
revision_store = self._get_revision_store(repo_transport, control_files)
853
return MetaDirRepository(_format=self,
855
control_files=control_files,
856
revision_store=revision_store)
1564
858
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1565
859
"""Upload the initial blank content."""
1566
860
control_files = self._create_control_files(a_bzrdir)
1567
861
control_files.lock_write()
1568
transport = control_files._transport
1570
utf8_files += [('shared-storage', '')]
862
control_files._transport.mkdir_multi(dirs,
863
mode=control_files._dir_mode)
1572
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
1573
for (filename, content_stream) in files:
1574
transport.put_file(filename, content_stream,
1575
mode=a_bzrdir._get_file_mode())
1576
for (filename, content_bytes) in utf8_files:
1577
transport.put_bytes_non_atomic(filename, content_bytes,
1578
mode=a_bzrdir._get_file_mode())
865
for file, content in files:
866
control_files.put(file, content)
867
for file, content in utf8_files:
868
control_files.put_utf8(file, content)
870
control_files.put_utf8('shared-storage', '')
1580
872
control_files.unlock()
1582
def network_name(self):
1583
"""Metadir formats have matching disk and network format strings."""
1584
return self.get_format_string()
1587
# formats which have no format string are not discoverable or independently
1588
# creatable on disk, so are not registered in format_registry. They're
1589
# all in bzrlib.repofmt.knitreponow. When an instance of one of these is
1590
# needed, it's constructed directly by the ControlDir. Non-native formats where
1591
# the repository is not separately opened are similar.
1593
format_registry.register_lazy(
1594
'Bazaar-NG Knit Repository Format 1',
1595
'bzrlib.repofmt.knitrepo',
1596
'RepositoryFormatKnit1',
1599
format_registry.register_lazy(
1600
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1601
'bzrlib.repofmt.knitrepo',
1602
'RepositoryFormatKnit3',
1605
format_registry.register_lazy(
1606
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
1607
'bzrlib.repofmt.knitrepo',
1608
'RepositoryFormatKnit4',
1611
# Pack-based formats. There is one format for pre-subtrees, and one for
1612
# post-subtrees to allow ease of testing.
1613
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
1614
format_registry.register_lazy(
1615
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
1616
'bzrlib.repofmt.knitpack_repo',
1617
'RepositoryFormatKnitPack1',
1619
format_registry.register_lazy(
1620
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
1621
'bzrlib.repofmt.knitpack_repo',
1622
'RepositoryFormatKnitPack3',
1624
format_registry.register_lazy(
1625
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
1626
'bzrlib.repofmt.knitpack_repo',
1627
'RepositoryFormatKnitPack4',
1629
format_registry.register_lazy(
1630
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
1631
'bzrlib.repofmt.knitpack_repo',
1632
'RepositoryFormatKnitPack5',
1634
format_registry.register_lazy(
1635
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
1636
'bzrlib.repofmt.knitpack_repo',
1637
'RepositoryFormatKnitPack5RichRoot',
1639
format_registry.register_lazy(
1640
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
1641
'bzrlib.repofmt.knitpack_repo',
1642
'RepositoryFormatKnitPack5RichRootBroken',
1644
format_registry.register_lazy(
1645
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
1646
'bzrlib.repofmt.knitpack_repo',
1647
'RepositoryFormatKnitPack6',
1649
format_registry.register_lazy(
1650
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
1651
'bzrlib.repofmt.knitpack_repo',
1652
'RepositoryFormatKnitPack6RichRoot',
1654
format_registry.register_lazy(
1655
'Bazaar repository format 2a (needs bzr 1.16 or later)\n',
1656
'bzrlib.repofmt.groupcompress_repo',
1657
'RepositoryFormat2a',
1660
# Development formats.
1661
# Check their docstrings to see if/when they are obsolete.
1662
format_registry.register_lazy(
1663
("Bazaar development format 2 with subtree support "
1664
"(needs bzr.dev from before 1.8)\n"),
1665
'bzrlib.repofmt.knitpack_repo',
1666
'RepositoryFormatPackDevelopment2Subtree',
1668
format_registry.register_lazy(
1669
'Bazaar development format 8\n',
1670
'bzrlib.repofmt.groupcompress_repo',
1671
'RepositoryFormat2aSubtree',
1675
class InterRepository(InterObject):
875
class RepositoryFormat7(MetaDirRepositoryFormat):
878
This repository format has:
879
- weaves for file texts and inventory
880
- hash subdirectory based stores.
881
- TextStores for revisions and signatures.
882
- a format marker of its own
883
- an optional 'shared-storage' flag
884
- an optional 'no-working-trees' flag
887
def get_format_string(self):
888
"""See RepositoryFormat.get_format_string()."""
889
return "Bazaar-NG Repository format 7"
891
def initialize(self, a_bzrdir, shared=False):
892
"""Create a weave repository.
894
:param shared: If true the repository will be initialized as a shared
897
from bzrlib.weavefile import write_weave_v5
898
from bzrlib.weave import Weave
900
# Create an empty weave
902
bzrlib.weavefile.write_weave_v5(Weave(), sio)
903
empty_weave = sio.getvalue()
905
mutter('creating repository in %s.', a_bzrdir.transport.base)
906
dirs = ['revision-store', 'weaves']
907
files = [('inventory.weave', StringIO(empty_weave)),
909
utf8_files = [('format', self.get_format_string())]
911
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
912
return self.open(a_bzrdir=a_bzrdir, _found=True)
915
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
916
"""Bzr repository knit format 1.
918
This repository format has:
919
- knits for file texts and inventory
920
- hash subdirectory based stores.
921
- knits for revisions and signatures
922
- TextStores for revisions and signatures.
923
- a format marker of its own
924
- an optional 'shared-storage' flag
925
- an optional 'no-working-trees' flag
928
def get_format_string(self):
929
"""See RepositoryFormat.get_format_string()."""
930
return "Bazaar-NG Knit Repository Format 1"
932
def initialize(self, a_bzrdir, shared=False):
933
"""Create a knit format 1 repository.
935
:param shared: If true the repository will be initialized as a shared
937
XXX NOTE that this current uses a Weave for testing and will become
938
A Knit in due course.
940
from bzrlib.weavefile import write_weave_v5
941
from bzrlib.weave import Weave
943
# Create an empty weave
945
bzrlib.weavefile.write_weave_v5(Weave(), sio)
946
empty_weave = sio.getvalue()
948
mutter('creating repository in %s.', a_bzrdir.transport.base)
949
dirs = ['revision-store', 'knits']
950
files = [('inventory.weave', StringIO(empty_weave)),
952
utf8_files = [('format', self.get_format_string())]
954
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
955
return self.open(a_bzrdir=a_bzrdir, _found=True)
958
# formats which have no format string are not discoverable
959
# and not independently creatable, so are not registered.
960
_default_format = RepositoryFormat7()
961
RepositoryFormat.register_format(_default_format)
962
RepositoryFormat.register_format(RepositoryFormatKnit1())
963
RepositoryFormat.set_default_format(_default_format)
964
_legacy_formats = [RepositoryFormat4(),
969
class InterRepository(object):
1676
970
"""This class represents operations taking place between two repositories.
1678
972
Its instances have methods like copy_content and fetch, and contain
1679
references to the source and target repositories these operations can be
973
references to the source and target repositories these operations can be
1682
976
Often we will provide convenience methods on 'repository' which carry out
1683
977
operations with another repository - they will always forward to
1684
978
InterRepository.get(other).method_name(parameters).
980
# XXX: FIXME: FUTURE: robertc
981
# testing of these probably requires a factory in optimiser type, and
982
# then a test adapter to test each type thoroughly.
1688
986
"""The available optimised InterRepository types."""
988
def __init__(self, source, target):
989
"""Construct a default InterRepository instance. Please use 'get'.
991
Only subclasses of InterRepository should call
992
InterRepository.__init__ - clients should call InterRepository.get
993
instead which will create an optimised InterRepository if possible.
1690
998
@needs_write_lock
1691
def copy_content(self, revision_id=None):
999
def copy_content(self, revision_id=None, basis=None):
1692
1000
"""Make a complete copy of the content in self into destination.
1694
This is a destructive operation! Do not use it on existing
1002
This is a destructive operation! Do not use it on existing
1697
1005
:param revision_id: Only copy the content needed to construct
1698
1006
revision_id and its parents.
1007
:param basis: Copy the needed data preferentially from basis.
1701
1010
self.target.set_make_working_trees(self.source.make_working_trees())
1702
1011
except NotImplementedError:
1013
# grab the basis available data
1014
if basis is not None:
1015
self.target.fetch(basis, revision_id=revision_id)
1016
# but dont both fetching if we have the needed data now.
1017
if (revision_id not in (None, NULL_REVISION) and
1018
self.target.has_revision(revision_id)):
1704
1020
self.target.fetch(self.source, revision_id=revision_id)
1022
def _double_lock(self, lock_source, lock_target):
1023
"""Take out too locks, rolling back the first if the second throws."""
1028
# we want to ensure that we don't leave source locked by mistake.
1029
# and any error on target should not confuse source.
1030
self.source.unlock()
1706
1033
@needs_write_lock
1707
def fetch(self, revision_id=None, find_ghosts=False,
1034
def fetch(self, revision_id=None, pb=None):
1709
1035
"""Fetch the content required to construct revision_id.
1711
The content is copied from self.source to self.target.
1037
The content is copied from source to target.
1713
1039
:param revision_id: if None all content is copied, if NULL_REVISION no
1714
1040
content is copied.
1717
raise NotImplementedError(self.fetch)
1041
:param pb: optional progress bar to use for progress reports. If not
1042
provided a default one will be created.
1044
Returns the copied revision count and the failed revisions in a tuple:
1047
from bzrlib.fetch import RepoFetcher
1048
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1049
self.source, self.source._format, self.target, self.target._format)
1050
f = RepoFetcher(to_repository=self.target,
1051
from_repository=self.source,
1052
last_revision=revision_id,
1054
return f.count_copied, f.failed_revisions
1057
def get(klass, repository_source, repository_target):
1058
"""Retrieve a InterRepository worker object for these repositories.
1060
:param repository_source: the repository to be the 'source' member of
1061
the InterRepository instance.
1062
:param repository_target: the repository to be the 'target' member of
1063
the InterRepository instance.
1064
If an optimised InterRepository worker exists it will be used otherwise
1065
a default InterRepository instance will be created.
1067
for provider in klass._optimisers:
1068
if provider.is_compatible(repository_source, repository_target):
1069
return provider(repository_source, repository_target)
1070
return InterRepository(repository_source, repository_target)
1072
def lock_read(self):
1073
"""Take out a logical read lock.
1075
This will lock the source branch and the target branch. The source gets
1076
a read lock and the target a read lock.
1078
self._double_lock(self.source.lock_read, self.target.lock_read)
1080
def lock_write(self):
1081
"""Take out a logical write lock.
1083
This will lock the source branch and the target branch. The source gets
1084
a read lock and the target a write lock.
1086
self._double_lock(self.source.lock_read, self.target.lock_write)
1719
1088
@needs_read_lock
1720
def search_missing_revision_ids(self,
1721
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
1722
find_ghosts=True, revision_ids=None, if_present_ids=None,
1089
def missing_revision_ids(self, revision_id=None):
1724
1090
"""Return the revision ids that source has that target does not.
1092
These are returned in topological order.
1726
1094
:param revision_id: only return revision ids included by this
1728
:param revision_ids: return revision ids included by these
1729
revision_ids. NoSuchRevision will be raised if any of these
1730
revisions are not present.
1731
:param if_present_ids: like revision_ids, but will not cause
1732
NoSuchRevision if any of these are absent, instead they will simply
1733
not be in the result. This is useful for e.g. finding revisions
1734
to fetch for tags, which may reference absent revisions.
1735
:param find_ghosts: If True find missing revisions in deep history
1736
rather than just finding the surface difference.
1737
:param limit: Maximum number of revisions to return, topologically
1739
:return: A bzrlib.graph.SearchResult.
1741
raise NotImplementedError(self.search_missing_revision_ids)
1097
# generic, possibly worst case, slow code path.
1098
target_ids = set(self.target.all_revision_ids())
1099
if revision_id is not None:
1100
source_ids = self.source.get_ancestry(revision_id)
1101
assert source_ids.pop(0) == None
1103
source_ids = self.source.all_revision_ids()
1104
result_set = set(source_ids).difference(target_ids)
1105
# this may look like a no-op: its not. It preserves the ordering
1106
# other_ids had while only returning the members from other_ids
1107
# that we've decided we need.
1108
return [rev_id for rev_id in source_ids if rev_id in result_set]
1111
def register_optimiser(klass, optimiser):
1112
"""Register an InterRepository optimiser."""
1113
klass._optimisers.add(optimiser)
1116
"""Release the locks on source and target."""
1118
self.target.unlock()
1120
self.source.unlock()
1123
def unregister_optimiser(klass, optimiser):
1124
"""Unregister an InterRepository optimiser."""
1125
klass._optimisers.remove(optimiser)
1128
class InterWeaveRepo(InterRepository):
1129
"""Optimised code paths between Weave based repositories."""
1131
_matching_repo_format = _default_format
1132
"""Repository format for testing with."""
1744
def _same_model(source, target):
1745
"""True if source and target have the same data representation.
1747
Note: this is always called on the base class; overriding it in a
1748
subclass will have no effect.
1135
def is_compatible(source, target):
1136
"""Be compatible with known Weave formats.
1138
We dont test for the stores being of specific types becase that
1139
could lead to confusing results, and there is no need to be
1751
InterRepository._assert_same_model(source, target)
1753
except errors.IncompatibleRepositories, e:
1143
return (isinstance(source._format, (RepositoryFormat5,
1145
RepositoryFormat7)) and
1146
isinstance(target._format, (RepositoryFormat5,
1148
RepositoryFormat7)))
1149
except AttributeError:
1153
def copy_content(self, revision_id=None, basis=None):
1154
"""See InterRepository.copy_content()."""
1155
# weave specific optimised path:
1156
if basis is not None:
1157
# copy the basis in, then fetch remaining data.
1158
basis.copy_content_into(self.target, revision_id)
1159
# the basis copy_content_into could misset this.
1161
self.target.set_make_working_trees(self.source.make_working_trees())
1162
except NotImplementedError:
1164
self.target.fetch(self.source, revision_id=revision_id)
1167
self.target.set_make_working_trees(self.source.make_working_trees())
1168
except NotImplementedError:
1170
# FIXME do not peek!
1171
if self.source.control_files._transport.listable():
1172
pb = bzrlib.ui.ui_factory.progress_bar()
1173
copy_all(self.source.weave_store,
1174
self.target.weave_store, pb=pb)
1175
pb.update('copying inventory', 0, 1)
1176
self.target.control_weaves.copy_multi(
1177
self.source.control_weaves, ['inventory'])
1178
copy_all(self.source.revision_store,
1179
self.target.revision_store, pb=pb)
1181
self.target.fetch(self.source, revision_id=revision_id)
1184
def fetch(self, revision_id=None, pb=None):
1185
"""See InterRepository.fetch()."""
1186
from bzrlib.fetch import RepoFetcher
1187
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1188
self.source, self.source._format, self.target, self.target._format)
1189
f = RepoFetcher(to_repository=self.target,
1190
from_repository=self.source,
1191
last_revision=revision_id,
1193
return f.count_copied, f.failed_revisions
1196
def missing_revision_ids(self, revision_id=None):
1197
"""See InterRepository.missing_revision_ids()."""
1198
# we want all revisions to satisfy revision_id in source.
1199
# but we dont want to stat every file here and there.
1200
# we want then, all revisions other needs to satisfy revision_id
1201
# checked, but not those that we have locally.
1202
# so the first thing is to get a subset of the revisions to
1203
# satisfy revision_id in source, and then eliminate those that
1204
# we do already have.
1205
# this is slow on high latency connection to self, but as as this
1206
# disk format scales terribly for push anyway due to rewriting
1207
# inventory.weave, this is considered acceptable.
1209
if revision_id is not None:
1210
source_ids = self.source.get_ancestry(revision_id)
1211
assert source_ids.pop(0) == None
1213
source_ids = self.source._all_possible_ids()
1214
source_ids_set = set(source_ids)
1215
# source_ids is the worst possible case we may need to pull.
1216
# now we want to filter source_ids against what we actually
1217
# have in target, but dont try to check for existence where we know
1218
# we do not have a revision as that would be pointless.
1219
target_ids = set(self.target._all_possible_ids())
1220
possibly_present_revisions = target_ids.intersection(source_ids_set)
1221
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1222
required_revisions = source_ids_set.difference(actually_present_revisions)
1223
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1224
if revision_id is not None:
1225
# we used get_ancestry to determine source_ids then we are assured all
1226
# revisions referenced are present as they are installed in topological order.
1227
# and the tip revision was validated by get_ancestry.
1228
return required_topo_revisions
1230
# if we just grabbed the possibly available ids, then
1231
# we only have an estimate of whats available and need to validate
1232
# that against the revision records.
1233
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1236
InterRepository.register_optimiser(InterWeaveRepo)
1239
class RepositoryTestProviderAdapter(object):
1240
"""A tool to generate a suite testing multiple repository formats at once.
1242
This is done by copying the test once for each transport and injecting
1243
the transport_server, transport_readonly_server, and bzrdir_format and
1244
repository_format classes into each copy. Each copy is also given a new id()
1245
to make it easy to identify.
1248
def __init__(self, transport_server, transport_readonly_server, formats):
1249
self._transport_server = transport_server
1250
self._transport_readonly_server = transport_readonly_server
1251
self._formats = formats
1253
def adapt(self, test):
1254
result = TestSuite()
1255
for repository_format, bzrdir_format in self._formats:
1256
new_test = deepcopy(test)
1257
new_test.transport_server = self._transport_server
1258
new_test.transport_readonly_server = self._transport_readonly_server
1259
new_test.bzrdir_format = bzrdir_format
1260
new_test.repository_format = repository_format
1261
def make_new_test_id():
1262
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1263
return lambda: new_id
1264
new_test.id = make_new_test_id()
1265
result.addTest(new_test)
1269
class InterRepositoryTestProviderAdapter(object):
1270
"""A tool to generate a suite testing multiple inter repository formats.
1272
This is done by copying the test once for each interrepo provider and injecting
1273
the transport_server, transport_readonly_server, repository_format and
1274
repository_to_format classes into each copy.
1275
Each copy is also given a new id() to make it easy to identify.
1278
def __init__(self, transport_server, transport_readonly_server, formats):
1279
self._transport_server = transport_server
1280
self._transport_readonly_server = transport_readonly_server
1281
self._formats = formats
1283
def adapt(self, test):
1284
result = TestSuite()
1285
for interrepo_class, repository_format, repository_format_to in self._formats:
1286
new_test = deepcopy(test)
1287
new_test.transport_server = self._transport_server
1288
new_test.transport_readonly_server = self._transport_readonly_server
1289
new_test.interrepo_class = interrepo_class
1290
new_test.repository_format = repository_format
1291
new_test.repository_format_to = repository_format_to
1292
def make_new_test_id():
1293
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1294
return lambda: new_id
1295
new_test.id = make_new_test_id()
1296
result.addTest(new_test)
1757
def _assert_same_model(source, target):
1758
"""Raise an exception if two repositories do not use the same model.
1760
if source.supports_rich_root() != target.supports_rich_root():
1761
raise errors.IncompatibleRepositories(source, target,
1762
"different rich-root support")
1763
if source._serializer != target._serializer:
1764
raise errors.IncompatibleRepositories(source, target,
1765
"different serializers")
1300
def default_test_list():
1301
"""Generate the default list of interrepo permutations to test."""
1303
# test the default InterRepository between format 6 and the current
1305
# XXX: robertc 20060220 reinstate this when there are two supported
1306
# formats which do not have an optimal code path between them.
1307
result.append((InterRepository,
1308
RepositoryFormat6(),
1309
RepositoryFormatKnit1()))
1310
for optimiser in InterRepository._optimisers:
1311
result.append((optimiser,
1312
optimiser._matching_repo_format,
1313
optimiser._matching_repo_format
1315
# if there are specific combinations we want to use, we can add them
1768
1320
class CopyConverter(object):
1769
1321
"""A repository conversion tool which just performs a copy of the content.
1771
1323
This is slow but quite reliable.
added
removed
bzrlib/repository.py
