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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
from cStringIO import StringIO
17
from __future__ import absolute_import
19
19
from bzrlib.lazy_import import lazy_import
20
20
lazy_import(globals(), """
25
24
from bzrlib import (
33
revision as _mod_revision,
34
testament as _mod_testament,
38
from bzrlib.bundle import serializer
39
from bzrlib.i18n import gettext
37
revision as _mod_revision,
42
from bzrlib.revisiontree import RevisionTree
43
from bzrlib.store.versioned import VersionedFileStore
44
from bzrlib.store.text import TextStore
45
from bzrlib.testament import Testament
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
49
from bzrlib.decorators import needs_read_lock, needs_write_lock, only_raises
50
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
56
from bzrlib.trace import mutter, note, warning
51
from bzrlib.lock import _RelockDebugMixin, LogicalLockResult
52
from bzrlib.trace import (
53
log_exception_quietly, note, mutter, mutter_callsite, warning)
59
56
# Old formats display a warning, but only once
60
57
_deprecation_warning_done = False
60
class IsInWriteGroupError(errors.InternalBzrError):
62
_fmt = "May not refresh_data of repo %(repo)s while in a write group."
64
def __init__(self, repo):
65
errors.InternalBzrError.__init__(self, repo=repo)
68
class CommitBuilder(object):
69
"""Provides an interface to build up a commit.
71
This allows describing a tree to be committed without needing to
72
know the internals of the format of the repository.
75
# all clients should supply tree roots.
76
record_root_entry = True
77
# whether this commit builder supports the record_entry_contents interface
78
supports_record_entry_contents = False
79
# whether this commit builder will automatically update the branch that is
81
updates_branch = False
83
def __init__(self, repository, parents, config_stack, timestamp=None,
84
timezone=None, committer=None, revprops=None,
85
revision_id=None, lossy=False):
86
"""Initiate a CommitBuilder.
88
:param repository: Repository to commit to.
89
:param parents: Revision ids of the parents of the new revision.
90
:param timestamp: Optional timestamp recorded for commit.
91
:param timezone: Optional timezone for timestamp.
92
:param committer: Optional committer to set for commit.
93
:param revprops: Optional dictionary of revision properties.
94
:param revision_id: Optional revision id.
95
:param lossy: Whether to discard data that can not be natively
96
represented, when pushing to a foreign VCS
98
self._config_stack = config_stack
101
if committer is None:
102
self._committer = self._config_stack.get('email')
103
elif not isinstance(committer, unicode):
104
self._committer = committer.decode() # throw if non-ascii
106
self._committer = committer
108
self._new_revision_id = revision_id
109
self.parents = parents
110
self.repository = repository
113
if revprops is not None:
114
self._validate_revprops(revprops)
115
self._revprops.update(revprops)
117
if timestamp is None:
118
timestamp = time.time()
119
# Restrict resolution to 1ms
120
self._timestamp = round(timestamp, 3)
123
self._timezone = osutils.local_time_offset()
125
self._timezone = int(timezone)
127
self._generate_revision_if_needed()
129
def any_changes(self):
130
"""Return True if any entries were changed.
132
This includes merge-only changes. It is the core for the --unchanged
135
:return: True if any changes have occured.
137
raise NotImplementedError(self.any_changes)
139
def _validate_unicode_text(self, text, context):
140
"""Verify things like commit messages don't have bogus characters."""
142
raise ValueError('Invalid value for %s: %r' % (context, text))
144
def _validate_revprops(self, revprops):
145
for key, value in revprops.iteritems():
146
# We know that the XML serializers do not round trip '\r'
147
# correctly, so refuse to accept them
148
if not isinstance(value, basestring):
149
raise ValueError('revision property (%s) is not a valid'
150
' (unicode) string: %r' % (key, value))
151
self._validate_unicode_text(value,
152
'revision property (%s)' % (key,))
154
def commit(self, message):
155
"""Make the actual commit.
157
:return: The revision id of the recorded revision.
159
raise NotImplementedError(self.commit)
162
"""Abort the commit that is being built.
164
raise NotImplementedError(self.abort)
166
def revision_tree(self):
167
"""Return the tree that was just committed.
169
After calling commit() this can be called to get a
170
RevisionTree representing the newly committed tree. This is
171
preferred to calling Repository.revision_tree() because that may
172
require deserializing the inventory, while we already have a copy in
175
raise NotImplementedError(self.revision_tree)
177
def finish_inventory(self):
178
"""Tell the builder that the inventory is finished.
180
:return: The inventory id in the repository, which can be used with
181
repository.get_inventory.
183
raise NotImplementedError(self.finish_inventory)
185
def _gen_revision_id(self):
186
"""Return new revision-id."""
187
return generate_ids.gen_revision_id(self._committer, self._timestamp)
189
def _generate_revision_if_needed(self):
190
"""Create a revision id if None was supplied.
192
If the repository can not support user-specified revision ids
193
they should override this function and raise CannotSetRevisionId
194
if _new_revision_id is not None.
196
:raises: CannotSetRevisionId
198
if self._new_revision_id is None:
199
self._new_revision_id = self._gen_revision_id()
200
self.random_revid = True
202
self.random_revid = False
204
def will_record_deletes(self):
205
"""Tell the commit builder that deletes are being notified.
207
This enables the accumulation of an inventory delta; for the resulting
208
commit to be valid, deletes against the basis MUST be recorded via
209
builder.record_delete().
211
raise NotImplementedError(self.will_record_deletes)
213
def record_iter_changes(self, tree, basis_revision_id, iter_changes):
214
"""Record a new tree via iter_changes.
216
:param tree: The tree to obtain text contents from for changed objects.
217
:param basis_revision_id: The revision id of the tree the iter_changes
218
has been generated against. Currently assumed to be the same
219
as self.parents[0] - if it is not, errors may occur.
220
:param iter_changes: An iter_changes iterator with the changes to apply
221
to basis_revision_id. The iterator must not include any items with
222
a current kind of None - missing items must be either filtered out
223
or errored-on beefore record_iter_changes sees the item.
224
:return: A generator of (file_id, relpath, fs_hash) tuples for use with
227
raise NotImplementedError(self.record_iter_changes)
230
class RepositoryWriteLockResult(LogicalLockResult):
231
"""The result of write locking a repository.
233
:ivar repository_token: The token obtained from the underlying lock, or
235
:ivar unlock: A callable which will unlock the lock.
238
def __init__(self, unlock, repository_token):
239
LogicalLockResult.__init__(self, unlock)
240
self.repository_token = repository_token
243
return "RepositoryWriteLockResult(%s, %s)" % (self.repository_token,
63
247
######################################################################
66
class Repository(object):
251
class Repository(_RelockDebugMixin, controldir.ControlComponent):
67
252
"""Repository holding history for one or more branches.
69
254
The repository holds and retrieves historical information including
70
255
revisions and file history. It's normally accessed only by the Branch,
71
256
which views a particular line of development through that history.
73
The Repository builds on top of Stores and a Transport, which respectively
74
describe the disk data format and the way of accessing the (possibly
258
See VersionedFileRepository in bzrlib.vf_repository for the
259
base class for most Bazaar repositories.
78
_file_ids_altered_regex = lazy_regex.lazy_compile(
79
r'file_id="(?P<file_id>[^"]+)"'
80
r'.*revision="(?P<revision_id>[^"]+)"'
84
def add_inventory(self, revision_id, inv, parents):
85
"""Add the inventory inv to the repository as revision_id.
87
:param parents: The revision ids of the parents that revision_id
88
is known to have and are in the repository already.
90
returns the sha1 of the serialized inventory.
92
revision_id = osutils.safe_revision_id(revision_id)
93
_mod_revision.check_not_reserved_id(revision_id)
94
assert inv.revision_id is None or inv.revision_id == revision_id, \
95
"Mismatch between inventory revision" \
96
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
97
assert inv.root is not None
98
inv_text = self.serialise_inventory(inv)
99
inv_sha1 = osutils.sha_string(inv_text)
100
inv_vf = self.control_weaves.get_weave('inventory',
101
self.get_transaction())
102
self._inventory_add_lines(inv_vf, revision_id, parents,
103
osutils.split_lines(inv_text))
106
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
108
for parent in parents:
110
final_parents.append(parent)
112
inv_vf.add_lines(revision_id, final_parents, lines)
115
def add_revision(self, revision_id, rev, inv=None, config=None):
116
"""Add rev to the revision store as revision_id.
118
:param revision_id: the revision id to use.
119
:param rev: The revision object.
120
:param inv: The inventory for the revision. if None, it will be looked
121
up in the inventory storer
122
:param config: If None no digital signature will be created.
123
If supplied its signature_needed method will be used
124
to determine if a signature should be made.
126
revision_id = osutils.safe_revision_id(revision_id)
127
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
129
_mod_revision.check_not_reserved_id(revision_id)
130
if config is not None and config.signature_needed():
132
inv = self.get_inventory(revision_id)
133
plaintext = Testament(rev, inv).as_short_text()
134
self.store_revision_signature(
135
gpg.GPGStrategy(config), plaintext, revision_id)
136
if not revision_id in self.get_inventory_weave():
138
raise errors.WeaveRevisionNotPresent(revision_id,
139
self.get_inventory_weave())
141
# yes, this is not suitable for adding with ghosts.
142
self.add_inventory(revision_id, inv, rev.parent_ids)
143
self._revision_store.add_revision(rev, self.get_transaction())
146
def _all_possible_ids(self):
147
"""Return all the possible revisions that we could find."""
148
return self.get_inventory_weave().versions()
262
def abort_write_group(self, suppress_errors=False):
263
"""Commit the contents accrued within the current write group.
265
:param suppress_errors: if true, abort_write_group will catch and log
266
unexpected errors that happen during the abort, rather than
267
allowing them to propagate. Defaults to False.
269
:seealso: start_write_group.
271
if self._write_group is not self.get_transaction():
272
# has an unlock or relock occured ?
275
'(suppressed) mismatched lock context and write group. %r, %r',
276
self._write_group, self.get_transaction())
278
raise errors.BzrError(
279
'mismatched lock context and write group. %r, %r' %
280
(self._write_group, self.get_transaction()))
282
self._abort_write_group()
283
except Exception, exc:
284
self._write_group = None
285
if not suppress_errors:
287
mutter('abort_write_group failed')
288
log_exception_quietly()
289
note(gettext('bzr: ERROR (ignored): %s'), exc)
290
self._write_group = None
292
def _abort_write_group(self):
293
"""Template method for per-repository write group cleanup.
295
This is called during abort before the write group is considered to be
296
finished and should cleanup any internal state accrued during the write
297
group. There is no requirement that data handed to the repository be
298
*not* made available - this is not a rollback - but neither should any
299
attempt be made to ensure that data added is fully commited. Abort is
300
invoked when an error has occured so futher disk or network operations
301
may not be possible or may error and if possible should not be
305
def add_fallback_repository(self, repository):
306
"""Add a repository to use for looking up data not held locally.
308
:param repository: A repository.
310
raise NotImplementedError(self.add_fallback_repository)
312
def _check_fallback_repository(self, repository):
313
"""Check that this repository can fallback to repository safely.
315
Raise an error if not.
317
:param repository: A repository to fallback to.
319
return InterRepository._assert_same_model(self, repository)
150
321
def all_revision_ids(self):
151
"""Returns a list of all the revision ids in the repository.
322
"""Returns a list of all the revision ids in the repository.
153
This is deprecated because code should generally work on the graph
154
reachable from a particular revision, and ignore any other revisions
155
that might be present. There is no direct replacement method.
324
This is conceptually deprecated because code should generally work on
325
the graph reachable from a particular revision, and ignore any other
326
revisions that might be present. There is no direct replacement
329
if 'evil' in debug.debug_flags:
330
mutter_callsite(2, "all_revision_ids is linear with history.")
157
331
return self._all_revision_ids()
160
333
def _all_revision_ids(self):
161
"""Returns a list of all the revision ids in the repository.
334
"""Returns a list of all the revision ids in the repository.
163
These are in as much topological order as the underlying store can
164
present: for weaves ghosts may lead to a lack of correctness until
165
the reweave updates the parents list.
336
These are in as much topological order as the underlying store can
167
if self._revision_store.text_store.listable():
168
return self._revision_store.all_revision_ids(self.get_transaction())
169
result = self._all_possible_ids()
170
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
171
# ids. (It should, since _revision_store's API should change to
172
# return utf8 revision_ids)
173
return self._eliminate_revisions_not_present(result)
339
raise NotImplementedError(self._all_revision_ids)
175
341
def break_lock(self):
176
342
"""Break a lock if one is present from another instance.
181
347
self.control_files.break_lock()
184
def _eliminate_revisions_not_present(self, revision_ids):
185
"""Check every revision id in revision_ids to see if we have it.
187
Returns a set of the present revisions.
190
for id in revision_ids:
191
if self.has_revision(id):
196
def create(a_bzrdir):
197
"""Construct the current default format repository in a_bzrdir."""
198
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
350
def create(controldir):
351
"""Construct the current default format repository in controldir."""
352
return RepositoryFormat.get_default_format().initialize(controldir)
200
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
354
def __init__(self, _format, controldir, control_files):
201
355
"""instantiate a Repository.
203
357
:param _format: The format of the repository on disk.
204
:param a_bzrdir: The BzrDir of the repository.
206
In the future we will have a single api for all stores for
207
getting file texts, inventories and revisions, then
208
this construct will accept instances of those things.
358
:param controldir: The ControlDir of the repository.
359
:param control_files: Control files to use for locking, etc.
361
# In the future we will have a single api for all stores for
362
# getting file texts, inventories and revisions, then
363
# this construct will accept instances of those things.
210
364
super(Repository, self).__init__()
211
365
self._format = _format
212
366
# the following are part of the public API for Repository:
213
self.bzrdir = a_bzrdir
367
self.bzrdir = controldir
214
368
self.control_files = control_files
215
self._revision_store = _revision_store
216
self.text_store = text_store
217
# backwards compatibility
218
self.weave_store = text_store
219
# not right yet - should be more semantically clear ?
221
self.control_store = control_store
222
self.control_weaves = control_store
223
# TODO: make sure to construct the right store classes, etc, depending
224
# on whether escaping is required.
225
self._warn_if_deprecated()
370
self._write_group = None
371
# Additional places to query for data.
372
self._fallback_repositories = []
375
def user_transport(self):
376
return self.bzrdir.user_transport
379
def control_transport(self):
380
return self._transport
227
382
def __repr__(self):
228
return '%s(%r)' % (self.__class__.__name__,
229
self.bzrdir.transport.base)
383
if self._fallback_repositories:
384
return '%s(%r, fallback_repositories=%r)' % (
385
self.__class__.__name__,
387
self._fallback_repositories)
389
return '%s(%r)' % (self.__class__.__name__,
392
def _has_same_fallbacks(self, other_repo):
393
"""Returns true if the repositories have the same fallbacks."""
394
my_fb = self._fallback_repositories
395
other_fb = other_repo._fallback_repositories
396
if len(my_fb) != len(other_fb):
398
for f, g in zip(my_fb, other_fb):
399
if not f.has_same_location(g):
403
def has_same_location(self, other):
404
"""Returns a boolean indicating if this repository is at the same
405
location as another repository.
407
This might return False even when two repository objects are accessing
408
the same physical repository via different URLs.
410
if self.__class__ is not other.__class__:
412
return (self.control_url == other.control_url)
414
def is_in_write_group(self):
415
"""Return True if there is an open write group.
417
:seealso: start_write_group.
419
return self._write_group is not None
231
421
def is_locked(self):
232
422
return self.control_files.is_locked()
234
def lock_write(self):
235
self.control_files.lock_write()
424
def is_write_locked(self):
425
"""Return True if this object is write locked."""
426
return self.is_locked() and self.control_files._lock_mode == 'w'
428
def lock_write(self, token=None):
429
"""Lock this repository for writing.
431
This causes caching within the repository obejct to start accumlating
432
data during reads, and allows a 'write_group' to be obtained. Write
433
groups must be used for actual data insertion.
435
A token should be passed in if you know that you have locked the object
436
some other way, and need to synchronise this object's state with that
439
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
441
:param token: if this is already locked, then lock_write will fail
442
unless the token matches the existing lock.
443
:returns: a token if this instance supports tokens, otherwise None.
444
:raises TokenLockingNotSupported: when a token is given but this
445
instance doesn't support using token locks.
446
:raises MismatchedToken: if the specified token doesn't match the token
447
of the existing lock.
448
:seealso: start_write_group.
449
:return: A RepositoryWriteLockResult.
451
locked = self.is_locked()
452
token = self.control_files.lock_write(token=token)
454
self._warn_if_deprecated()
456
for repo in self._fallback_repositories:
457
# Writes don't affect fallback repos
460
return RepositoryWriteLockResult(self.unlock, token)
237
462
def lock_read(self):
463
"""Lock the repository for read operations.
465
:return: An object with an unlock method which will release the lock
468
locked = self.is_locked()
238
469
self.control_files.lock_read()
471
self._warn_if_deprecated()
473
for repo in self._fallback_repositories:
476
return LogicalLockResult(self.unlock)
240
478
def get_physical_lock_status(self):
241
479
return self.control_files.get_physical_lock_status()
481
def leave_lock_in_place(self):
482
"""Tell this repository not to release the physical lock when this
485
If lock_write doesn't return a token, then this method is not supported.
487
self.control_files.leave_in_place()
489
def dont_leave_lock_in_place(self):
490
"""Tell this repository to release the physical lock when this
491
object is unlocked, even if it didn't originally acquire it.
493
If lock_write doesn't return a token, then this method is not supported.
495
self.control_files.dont_leave_in_place()
244
498
def gather_stats(self, revid=None, committers=None):
245
499
"""Gather statistics from a revision id.
308
607
For instance, if the repository is at URL/.bzr/repository,
309
608
Repository.open(URL) -> a Repository instance.
311
control = bzrdir.BzrDir.open(base)
610
control = controldir.ControlDir.open(base)
312
611
return control.open_repository()
314
def copy_content_into(self, destination, revision_id=None, basis=None):
613
def copy_content_into(self, destination, revision_id=None):
315
614
"""Make a complete copy of the content in self into destination.
616
This is a destructive operation! Do not use it on existing
619
return InterRepository.get(self, destination).copy_content(revision_id)
621
def commit_write_group(self):
622
"""Commit the contents accrued within the current write group.
624
:seealso: start_write_group.
317
This is a destructive operation! Do not use it on existing
320
revision_id = osutils.safe_revision_id(revision_id)
321
return InterRepository.get(self, destination).copy_content(revision_id, basis)
323
def fetch(self, source, revision_id=None, pb=None):
626
:return: it may return an opaque hint that can be passed to 'pack'.
628
if self._write_group is not self.get_transaction():
629
# has an unlock or relock occured ?
630
raise errors.BzrError('mismatched lock context %r and '
632
(self.get_transaction(), self._write_group))
633
result = self._commit_write_group()
634
self._write_group = None
637
def _commit_write_group(self):
638
"""Template method for per-repository write group cleanup.
640
This is called before the write group is considered to be
641
finished and should ensure that all data handed to the repository
642
for writing during the write group is safely committed (to the
643
extent possible considering file system caching etc).
646
def suspend_write_group(self):
647
"""Suspend a write group.
649
:raise UnsuspendableWriteGroup: If the write group can not be
651
:return: List of tokens
653
raise errors.UnsuspendableWriteGroup(self)
655
def refresh_data(self):
656
"""Re-read any data needed to synchronise with disk.
658
This method is intended to be called after another repository instance
659
(such as one used by a smart server) has inserted data into the
660
repository. On all repositories this will work outside of write groups.
661
Some repository formats (pack and newer for bzrlib native formats)
662
support refresh_data inside write groups. If called inside a write
663
group on a repository that does not support refreshing in a write group
664
IsInWriteGroupError will be raised.
668
def resume_write_group(self, tokens):
669
if not self.is_write_locked():
670
raise errors.NotWriteLocked(self)
671
if self._write_group:
672
raise errors.BzrError('already in a write group')
673
self._resume_write_group(tokens)
674
# so we can detect unlock/relock - the write group is now entered.
675
self._write_group = self.get_transaction()
677
def _resume_write_group(self, tokens):
678
raise errors.UnsuspendableWriteGroup(self)
680
def fetch(self, source, revision_id=None, find_ghosts=False):
324
681
"""Fetch the content required to construct revision_id from source.
326
If revision_id is None all content is copied.
683
If revision_id is None, then all content is copied.
685
fetch() may not be used when the repository is in a write group -
686
either finish the current write group before using fetch, or use
687
fetch before starting the write group.
689
:param find_ghosts: Find and copy revisions in the source that are
690
ghosts in the target (and not reachable directly by walking out to
691
the first-present revision in target from revision_id).
692
:param revision_id: If specified, all the content needed for this
693
revision ID will be copied to the target. Fetch will determine for
694
itself which content needs to be copied.
328
revision_id = osutils.safe_revision_id(revision_id)
329
return InterRepository.get(source, self).fetch(revision_id=revision_id,
332
def get_commit_builder(self, branch, parents, config, timestamp=None,
333
timezone=None, committer=None, revprops=None,
696
if self.is_in_write_group():
697
raise errors.InternalBzrError(
698
"May not fetch while in a write group.")
699
# fast path same-url fetch operations
700
# TODO: lift out to somewhere common with RemoteRepository
701
# <https://bugs.launchpad.net/bzr/+bug/401646>
702
if (self.has_same_location(source)
703
and self._has_same_fallbacks(source)):
704
# check that last_revision is in 'from' and then return a
706
if (revision_id is not None and
707
not _mod_revision.is_null(revision_id)):
708
self.get_revision(revision_id)
710
inter = InterRepository.get(source, self)
711
return inter.fetch(revision_id=revision_id, find_ghosts=find_ghosts)
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_stack, timestamp=None,
717
timezone=None, committer=None, revprops=None,
718
revision_id=None, lossy=False):
335
719
"""Obtain a CommitBuilder for this repository.
337
721
:param branch: Branch to commit to.
338
722
:param parents: Revision ids of the parents of the new revision.
339
:param config: Configuration to use.
723
:param config_stack: Configuration stack to use.
340
724
:param timestamp: Optional timestamp recorded for commit.
341
725
:param timezone: Optional timezone for timestamp.
342
726
:param committer: Optional committer to set for commit.
343
727
:param revprops: Optional dictionary of revision properties.
344
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
346
revision_id = osutils.safe_revision_id(revision_id)
347
return _CommitBuilder(self, parents, config, timestamp, timezone,
348
committer, revprops, revision_id)
732
raise NotImplementedError(self.get_commit_builder)
734
@only_raises(errors.LockNotHeld, errors.LockBroken)
350
735
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.')
351
743
self.control_files.unlock()
744
if self.control_files._lock_count == 0:
745
for repo in self._fallback_repositories:
354
def clone(self, a_bzrdir, revision_id=None, basis=None):
355
"""Clone this repository into a_bzrdir using the current format.
749
def clone(self, controldir, revision_id=None):
750
"""Clone this repository into controldir using the current format.
357
752
Currently no check is made that the format of this repository and
358
753
the bzrdir format are compatible. FIXME RBC 20060201.
360
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):
362
804
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
363
805
# use target default format.
364
806
dest_repo = a_bzrdir.create_repository()
366
808
# Most control formats need the repository to be specifically
367
809
# created, but on some old all-in-one formats it's not needed
369
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
811
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
370
812
except errors.UninitializableFormat:
371
813
dest_repo = a_bzrdir.open_repository()
372
self.copy_content_into(dest_repo, revision_id, basis)
376
817
def has_revision(self, revision_id):
377
818
"""True if this repository has a copy of the revision."""
378
revision_id = osutils.safe_revision_id(revision_id)
379
return self._revision_store.has_revision_id(revision_id,
380
self.get_transaction())
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)
831
def get_revision(self, revision_id):
832
"""Return the Revision object for a named revision."""
833
return self.get_revisions([revision_id])[0]
383
835
def get_revision_reconcile(self, revision_id):
384
836
"""'reconcile' helper routine that allows access to a revision always.
386
838
This variant of get_revision does not cross check the weave graph
387
839
against the revision one as get_revision does: but it should only
388
840
be used by reconcile, or reconcile-alike commands that are correcting
389
841
or testing the revision graph.
391
if not revision_id or not isinstance(revision_id, basestring):
392
raise errors.InvalidRevisionId(revision_id=revision_id,
394
return self.get_revisions([revision_id])[0]
843
raise NotImplementedError(self.get_revision_reconcile)
397
845
def get_revisions(self, revision_ids):
398
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
399
revs = self._revision_store.get_revisions(revision_ids,
400
self.get_transaction())
402
assert not isinstance(rev.revision_id, unicode)
403
for parent_id in rev.parent_ids:
404
assert not isinstance(parent_id, unicode)
408
def get_revision_xml(self, revision_id):
409
# TODO: jam 20070210 This shouldn't be necessary since get_revision
410
# would have already do it.
411
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
412
revision_id = osutils.safe_revision_id(revision_id)
413
rev = self.get_revision(revision_id)
415
# the current serializer..
416
self._revision_store._serializer.write_revision(rev, rev_tmp)
418
return rev_tmp.getvalue()
421
def get_revision(self, revision_id):
422
"""Return the Revision object for a named revision"""
423
# TODO: jam 20070210 get_revision_reconcile should do this for us
424
revision_id = osutils.safe_revision_id(revision_id)
425
r = self.get_revision_reconcile(revision_id)
426
# weave corruption can lead to absent revision markers that should be
428
# the following test is reasonably cheap (it needs a single weave read)
429
# and the weave is cached in read transactions. In write transactions
430
# it is not cached but typically we only read a small number of
431
# revisions. For knits when they are introduced we will probably want
432
# to ensure that caching write transactions are in use.
433
inv = self.get_inventory_weave()
434
self._check_revision_parents(r, inv)
438
def get_deltas_for_revisions(self, revisions):
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):
439
854
"""Produce a generator of revision deltas.
441
856
Note that the input is a sequence of REVISIONS, not revision_ids.
442
857
Trees will be held in memory until the generator exits.
443
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
445
865
required_trees = set()
446
866
for revision in revisions:
447
867
required_trees.add(revision.revision_id)
448
868
required_trees.update(revision.parent_ids[:1])
449
trees = dict((t.get_revision_id(), t) for
450
t in self.revision_trees(required_trees))
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
451
885
for revision in revisions:
452
886
if not revision.parent_ids:
453
old_tree = self.revision_tree(None)
887
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
455
889
old_tree = trees[revision.parent_ids[0]]
456
890
yield trees[revision.revision_id].changes_from(old_tree)
459
def get_revision_delta(self, revision_id):
893
def get_revision_delta(self, revision_id, specific_fileids=None):
460
894
"""Return the delta for one revision.
462
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.
465
903
r = self.get_revision(revision_id)
466
return list(self.get_deltas_for_revisions([r]))[0]
468
def _check_revision_parents(self, revision, inventory):
469
"""Private to Repository and Fetch.
471
This checks the parentage of revision in an inventory weave for
472
consistency and is only applicable to inventory-weave-for-ancestry
473
using repository formats & fetchers.
475
weave_parents = inventory.get_parents(revision.revision_id)
476
weave_names = inventory.versions()
477
for parent_id in revision.parent_ids:
478
if parent_id in weave_names:
479
# this parent must not be a ghost.
480
if not parent_id in weave_parents:
482
raise errors.CorruptRepository(self)
904
return list(self.get_deltas_for_revisions([r],
905
specific_fileids=specific_fileids))[0]
484
907
@needs_write_lock
485
908
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
486
revision_id = osutils.safe_revision_id(revision_id)
487
909
signature = gpg_strategy.sign(plaintext)
488
self._revision_store.add_revision_signature_text(revision_id,
490
self.get_transaction())
492
def fileids_altered_by_revision_ids(self, revision_ids):
493
"""Find the file ids and versions affected by revisions.
495
:param revisions: an iterable containing revision ids.
496
:return: a dictionary mapping altered file-ids to an iterable of
497
revision_ids. Each altered file-ids has the exact revision_ids that
498
altered it listed explicitly.
500
assert self._serializer.support_altered_by_hack, \
501
("fileids_altered_by_revision_ids only supported for branches "
502
"which store inventory as unnested xml, not on %r" % self)
503
selected_revision_ids = set(osutils.safe_revision_id(r)
504
for r in revision_ids)
505
w = self.get_inventory_weave()
508
# this code needs to read every new line in every inventory for the
509
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
510
# not present in one of those inventories is unnecessary but not
511
# harmful because we are filtering by the revision id marker in the
512
# inventory lines : we only select file ids altered in one of those
513
# revisions. We don't need to see all lines in the inventory because
514
# only those added in an inventory in rev X can contain a revision=X
516
unescape_revid_cache = {}
517
unescape_fileid_cache = {}
519
# jam 20061218 In a big fetch, this handles hundreds of thousands
520
# of lines, so it has had a lot of inlining and optimizing done.
521
# Sorry that it is a little bit messy.
522
# Move several functions to be local variables, since this is a long
524
search = self._file_ids_altered_regex.search
525
unescape = _unescape_xml
526
setdefault = result.setdefault
527
pb = ui.ui_factory.nested_progress_bar()
529
for line in w.iter_lines_added_or_present_in_versions(
530
selected_revision_ids, pb=pb):
534
# One call to match.group() returning multiple items is quite a
535
# bit faster than 2 calls to match.group() each returning 1
536
file_id, revision_id = match.group('file_id', 'revision_id')
538
# Inlining the cache lookups helps a lot when you make 170,000
539
# lines and 350k ids, versus 8.4 unique ids.
540
# Using a cache helps in 2 ways:
541
# 1) Avoids unnecessary decoding calls
542
# 2) Re-uses cached strings, which helps in future set and
544
# (2) is enough that removing encoding entirely along with
545
# the cache (so we are using plain strings) results in no
546
# performance improvement.
548
revision_id = unescape_revid_cache[revision_id]
550
unescaped = unescape(revision_id)
551
unescape_revid_cache[revision_id] = unescaped
552
revision_id = unescaped
554
if revision_id in selected_revision_ids:
556
file_id = unescape_fileid_cache[file_id]
558
unescaped = unescape(file_id)
559
unescape_fileid_cache[file_id] = unescaped
561
setdefault(file_id, set()).add(revision_id)
567
def get_inventory_weave(self):
568
return self.control_weaves.get_weave('inventory',
569
self.get_transaction())
572
def get_inventory(self, revision_id):
573
"""Get Inventory object by hash."""
574
# TODO: jam 20070210 Technically we don't need to sanitize, since all
575
# called functions must sanitize.
576
revision_id = osutils.safe_revision_id(revision_id)
577
return self.deserialise_inventory(
578
revision_id, self.get_inventory_xml(revision_id))
580
def deserialise_inventory(self, revision_id, xml):
581
"""Transform the xml into an inventory object.
583
:param revision_id: The expected revision id of the inventory.
584
:param xml: A serialised inventory.
586
revision_id = osutils.safe_revision_id(revision_id)
587
result = self._serializer.read_inventory_from_string(xml)
588
result.root.revision = revision_id
591
def serialise_inventory(self, inv):
592
return self._serializer.write_inventory_to_string(inv)
595
def get_inventory_xml(self, revision_id):
596
"""Get inventory XML as a file object."""
597
revision_id = osutils.safe_revision_id(revision_id)
599
assert isinstance(revision_id, str), type(revision_id)
600
iw = self.get_inventory_weave()
601
return iw.get_text(revision_id)
603
raise errors.HistoryMissing(self, 'inventory', revision_id)
606
def get_inventory_sha1(self, revision_id):
607
"""Return the sha1 hash of the inventory entry
609
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
610
revision_id = osutils.safe_revision_id(revision_id)
611
return self.get_revision(revision_id).inventory_sha1
614
def get_revision_graph(self, revision_id=None):
615
"""Return a dictionary containing the revision graph.
617
:param revision_id: The revision_id to get a graph from. If None, then
618
the entire revision graph is returned. This is a deprecated mode of
619
operation and will be removed in the future.
620
:return: a dictionary of revision_id->revision_parents_list.
622
# special case NULL_REVISION
623
if revision_id == _mod_revision.NULL_REVISION:
625
revision_id = osutils.safe_revision_id(revision_id)
626
a_weave = self.get_inventory_weave()
627
all_revisions = self._eliminate_revisions_not_present(
629
entire_graph = dict([(node, a_weave.get_parents(node)) for
630
node in all_revisions])
631
if revision_id is None:
633
elif revision_id not in entire_graph:
634
raise errors.NoSuchRevision(self, revision_id)
636
# add what can be reached from revision_id
638
pending = set([revision_id])
639
while len(pending) > 0:
641
result[node] = entire_graph[node]
642
for revision_id in result[node]:
643
if revision_id not in result:
644
pending.add(revision_id)
648
def get_revision_graph_with_ghosts(self, revision_ids=None):
649
"""Return a graph of the revisions with ghosts marked as applicable.
651
:param revision_ids: an iterable of revisions to graph or None for all.
652
:return: a Graph object with the graph reachable from revision_ids.
654
result = graph.Graph()
656
pending = set(self.all_revision_ids())
659
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
660
# special case NULL_REVISION
661
if _mod_revision.NULL_REVISION in pending:
662
pending.remove(_mod_revision.NULL_REVISION)
663
required = set(pending)
666
revision_id = pending.pop()
668
rev = self.get_revision(revision_id)
669
except errors.NoSuchRevision:
670
if revision_id in required:
673
result.add_ghost(revision_id)
675
for parent_id in rev.parent_ids:
676
# is this queued or done ?
677
if (parent_id not in pending and
678
parent_id not in done):
680
pending.add(parent_id)
681
result.add_node(revision_id, rev.parent_ids)
682
done.add(revision_id)
685
def _get_history_vf(self):
686
"""Get a versionedfile whose history graph reflects all revisions.
688
For weave repositories, this is the inventory weave.
690
return self.get_inventory_weave()
692
def iter_reverse_revision_history(self, revision_id):
693
"""Iterate backwards through revision ids in the lefthand history
695
:param revision_id: The revision id to start with. All its lefthand
696
ancestors will be traversed.
698
revision_id = osutils.safe_revision_id(revision_id)
699
if revision_id in (None, _mod_revision.NULL_REVISION):
701
next_id = revision_id
702
versionedfile = self._get_history_vf()
705
parents = versionedfile.get_parents(next_id)
706
if len(parents) == 0:
712
def get_revision_inventory(self, revision_id):
713
"""Return inventory of a past revision."""
714
# TODO: Unify this with get_inventory()
715
# bzr 0.0.6 and later imposes the constraint that the inventory_id
716
# must be the same as its revision, so this is trivial.
717
if revision_id is None:
718
# This does not make sense: if there is no revision,
719
# then it is the current tree inventory surely ?!
720
# and thus get_root_id() is something that looks at the last
721
# commit on the branch, and the get_root_id is an inventory check.
722
raise NotImplementedError
723
# return Inventory(self.get_root_id())
725
return self.get_inventory(revision_id)
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))
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])
728
984
def is_shared(self):
729
985
"""Return True if this repository is flagged as a shared repository."""
730
986
raise NotImplementedError(self.is_shared)
736
992
reconciler = RepoReconciler(self, thorough=thorough)
737
993
reconciler.reconcile()
738
994
return reconciler
996
def _refresh_data(self):
997
"""Helper called from lock_* to ensure coherency with disk.
999
The default implementation does nothing; it is however possible
1000
for repositories to maintain loaded indices across multiple locks
1001
by checking inside their implementation of this method to see
1002
whether their indices are still valid. This depends of course on
1003
the disk format being validatable in this manner. This method is
1004
also called by the refresh_data() public interface to cause a refresh
1005
to occur while in a write lock so that data inserted by a smart server
1006
push operation is visible on the client's instance of the physical
740
1010
@needs_read_lock
741
1011
def revision_tree(self, revision_id):
742
1012
"""Return Tree for a revision on this branch.
744
`revision_id` may be None for the empty tree revision.
1014
`revision_id` may be NULL_REVISION for the empty tree revision.
746
# TODO: refactor this to use an existing revision object
747
# so we don't need to read it in twice.
748
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
749
return RevisionTree(self, Inventory(root_id=None),
750
_mod_revision.NULL_REVISION)
752
revision_id = osutils.safe_revision_id(revision_id)
753
inv = self.get_revision_inventory(revision_id)
754
return RevisionTree(self, inv, revision_id)
1016
raise NotImplementedError(self.revision_tree)
757
1018
def revision_trees(self, revision_ids):
758
"""Return Tree for a revision on this branch.
760
`revision_id` may not be None or 'null:'"""
761
assert None not in revision_ids
762
assert _mod_revision.NULL_REVISION not in revision_ids
763
texts = self.get_inventory_weave().get_texts(revision_ids)
764
for text, revision_id in zip(texts, revision_ids):
765
inv = self.deserialise_inventory(revision_id, text)
766
yield RevisionTree(self, inv, revision_id)
769
def get_ancestry(self, revision_id):
770
"""Return a list of revision-ids integrated by a revision.
772
The first element of the list is always None, indicating the origin
773
revision. This might change when we have history horizons, or
774
perhaps we should have a new API.
776
This is topologically sorted.
778
if revision_id is None:
780
revision_id = osutils.safe_revision_id(revision_id)
781
if not self.has_revision(revision_id):
782
raise errors.NoSuchRevision(self, revision_id)
783
w = self.get_inventory_weave()
784
candidates = w.get_ancestry(revision_id)
785
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
788
def print_file(self, file, revision_id):
789
"""Print `file` to stdout.
791
FIXME RBC 20060125 as John Meinel points out this is a bad api
792
- it writes to stdout, it assumes that that is valid etc. Fix
793
by creating a new more flexible convenience function.
795
revision_id = osutils.safe_revision_id(revision_id)
796
tree = self.revision_tree(revision_id)
797
# use inventory as it was in that revision
798
file_id = tree.inventory.path2id(file)
800
# TODO: jam 20060427 Write a test for this code path
801
# it had a bug in it, and was raising the wrong
803
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
804
tree.print_file(file_id)
1019
"""Return Trees for revisions in this repository.
1021
:param revision_ids: a sequence of revision-ids;
1022
a revision-id may not be None or 'null:'
1024
raise NotImplementedError(self.revision_trees)
1026
def pack(self, hint=None, clean_obsolete_packs=False):
1027
"""Compress the data within the repository.
1029
This operation only makes sense for some repository types. For other
1030
types it should be a no-op that just returns.
1032
This stub method does not require a lock, but subclasses should use
1033
@needs_write_lock as this is a long running call it's reasonable to
1034
implicitly lock for the user.
1036
:param hint: If not supplied, the whole repository is packed.
1037
If supplied, the repository may use the hint parameter as a
1038
hint for the parts of the repository to pack. A hint can be
1039
obtained from the result of commit_write_group(). Out of
1040
date hints are simply ignored, because concurrent operations
1041
can obsolete them rapidly.
1043
:param clean_obsolete_packs: Clean obsolete packs immediately after
806
1047
def get_transaction(self):
807
1048
return self.control_files.get_transaction()
809
def revision_parents(self, revision_id):
810
revision_id = osutils.safe_revision_id(revision_id)
811
return self.get_inventory_weave().parent_names(revision_id)
1050
def get_parent_map(self, revision_ids):
1051
"""See graph.StackedParentsProvider.get_parent_map"""
1052
raise NotImplementedError(self.get_parent_map)
1054
def _get_parent_map_no_fallbacks(self, revision_ids):
1055
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1056
# revisions index works in keys; this just works in revisions
1057
# therefore wrap and unwrap
1060
for revision_id in revision_ids:
1061
if revision_id == _mod_revision.NULL_REVISION:
1062
result[revision_id] = ()
1063
elif revision_id is None:
1064
raise ValueError('get_parent_map(None) is not valid')
1066
query_keys.append((revision_id ,))
1067
vf = self.revisions.without_fallbacks()
1068
for ((revision_id,), parent_keys) in \
1069
vf.get_parent_map(query_keys).iteritems():
1071
result[revision_id] = tuple([parent_revid
1072
for (parent_revid,) in parent_keys])
1074
result[revision_id] = (_mod_revision.NULL_REVISION,)
1077
def _make_parents_provider(self):
1078
if not self._format.supports_external_lookups:
1080
return graph.StackedParentsProvider(_LazyListJoin(
1081
[self._make_parents_provider_unstacked()],
1082
self._fallback_repositories))
1084
def _make_parents_provider_unstacked(self):
1085
return graph.CallableToParentsProviderAdapter(
1086
self._get_parent_map_no_fallbacks)
1089
def get_known_graph_ancestry(self, revision_ids):
1090
"""Return the known graph for a set of revision ids and their ancestors.
1092
raise NotImplementedError(self.get_known_graph_ancestry)
1094
def get_file_graph(self):
1095
"""Return the graph walker for files."""
1096
raise NotImplementedError(self.get_file_graph)
1098
def get_graph(self, other_repository=None):
1099
"""Return the graph walker for this repository format"""
1100
parents_provider = self._make_parents_provider()
1101
if (other_repository is not None and
1102
not self.has_same_location(other_repository)):
1103
parents_provider = graph.StackedParentsProvider(
1104
[parents_provider, other_repository._make_parents_provider()])
1105
return graph.Graph(parents_provider)
813
1107
@needs_write_lock
814
1108
def set_make_working_trees(self, new_value):
823
1117
raise NotImplementedError(self.set_make_working_trees)
825
1119
def make_working_trees(self):
826
1120
"""Returns the policy for making working trees on new branches."""
827
1121
raise NotImplementedError(self.make_working_trees)
829
1123
@needs_write_lock
830
1124
def sign_revision(self, revision_id, gpg_strategy):
831
revision_id = osutils.safe_revision_id(revision_id)
832
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1125
testament = _mod_testament.Testament.from_revision(self, revision_id)
1126
plaintext = testament.as_short_text()
833
1127
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
835
1129
@needs_read_lock
1130
def verify_revision_signature(self, revision_id, gpg_strategy):
1131
"""Verify the signature on a revision.
1133
:param revision_id: the revision to verify
1134
:gpg_strategy: the GPGStrategy object to used
1136
:return: gpg.SIGNATURE_VALID or a failed SIGNATURE_ value
1138
if not self.has_signature_for_revision_id(revision_id):
1139
return gpg.SIGNATURE_NOT_SIGNED, None
1140
signature = self.get_signature_text(revision_id)
1142
testament = _mod_testament.Testament.from_revision(self, revision_id)
1143
plaintext = testament.as_short_text()
1145
return gpg_strategy.verify(signature, plaintext)
1148
def verify_revision_signatures(self, revision_ids, gpg_strategy):
1149
"""Verify revision signatures for a number of revisions.
1151
:param revision_id: the revision to verify
1152
:gpg_strategy: the GPGStrategy object to used
1153
:return: Iterator over tuples with revision id, result and keys
1155
for revid in revision_ids:
1156
(result, key) = self.verify_revision_signature(revid, gpg_strategy)
1157
yield revid, result, key
836
1159
def has_signature_for_revision_id(self, revision_id):
837
1160
"""Query for a revision signature for revision_id in the repository."""
838
revision_id = osutils.safe_revision_id(revision_id)
839
return self._revision_store.has_signature(revision_id,
840
self.get_transaction())
1161
raise NotImplementedError(self.has_signature_for_revision_id)
843
1163
def get_signature_text(self, revision_id):
844
1164
"""Return the text for a signature."""
845
revision_id = osutils.safe_revision_id(revision_id)
846
return self._revision_store.get_signature_text(revision_id,
847
self.get_transaction())
1165
raise NotImplementedError(self.get_signature_text)
850
def check(self, revision_ids):
1167
def check(self, revision_ids=None, callback_refs=None, check_repo=True):
851
1168
"""Check consistency of all history of given revision_ids.
853
1170
Different repository implementations should override _check().
855
1172
:param revision_ids: A non-empty list of revision_ids whose ancestry
856
1173
will be checked. Typically the last revision_id of a branch.
1174
:param callback_refs: A dict of check-refs to resolve and callback
1175
the check/_check method on the items listed as wanting the ref.
1177
:param check_repo: If False do not check the repository contents, just
1178
calculate the data callback_refs requires and call them back.
859
raise ValueError("revision_ids must be non-empty in %s.check"
861
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
862
return self._check(revision_ids)
864
def _check(self, revision_ids):
865
result = check.Check(self)
869
def _warn_if_deprecated(self):
1180
return self._check(revision_ids=revision_ids, callback_refs=callback_refs,
1181
check_repo=check_repo)
1183
def _check(self, revision_ids=None, callback_refs=None, check_repo=True):
1184
raise NotImplementedError(self.check)
1186
def _warn_if_deprecated(self, branch=None):
1187
if not self._format.is_deprecated():
870
1189
global _deprecation_warning_done
871
1190
if _deprecation_warning_done:
873
_deprecation_warning_done = True
874
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
875
% (self._format, self.bzrdir.transport.base))
1194
conf = config.GlobalStack()
1196
conf = branch.get_config_stack()
1197
if 'format_deprecation' in conf.get('suppress_warnings'):
1199
warning("Format %s for %s is deprecated -"
1200
" please use 'bzr upgrade' to get better performance"
1201
% (self._format, self.bzrdir.transport.base))
1203
_deprecation_warning_done = True
877
1205
def supports_rich_root(self):
878
1206
return self._format.rich_root_data
1074
1387
def __ne__(self, other):
1075
1388
return not self == other
1078
def find_format(klass, a_bzrdir):
1079
"""Return the format for the repository object in a_bzrdir.
1081
This is used by bzr native formats that have a "format" file in
1082
the repository. Other methods may be used by different types of
1086
transport = a_bzrdir.get_repository_transport(None)
1087
format_string = transport.get("format").read()
1088
return format_registry.get(format_string)
1089
except errors.NoSuchFile:
1090
raise errors.NoRepositoryPresent(a_bzrdir)
1092
raise errors.UnknownFormatError(format=format_string)
1095
def register_format(klass, format):
1096
format_registry.register(format.get_format_string(), format)
1099
def unregister_format(klass, format):
1100
format_registry.remove(format.get_format_string())
1103
def get_default_format(klass):
1104
"""Return the current default format."""
1105
from bzrlib import bzrdir
1106
return bzrdir.format_registry.make_bzrdir('default').repository_format
1108
def _get_control_store(self, repo_transport, control_files):
1109
"""Return the control store for this repository."""
1110
raise NotImplementedError(self._get_control_store)
1112
def get_format_string(self):
1113
"""Return the ASCII format string that identifies this format.
1115
Note that in pre format ?? repositories the format string is
1116
not permitted nor written to disk.
1118
raise NotImplementedError(self.get_format_string)
1120
1390
def get_format_description(self):
1121
1391
"""Return the short description for this format."""
1122
1392
raise NotImplementedError(self.get_format_description)
1124
def _get_revision_store(self, repo_transport, control_files):
1125
"""Return the revision store object for this a_bzrdir."""
1126
raise NotImplementedError(self._get_revision_store)
1128
def _get_text_rev_store(self,
1135
"""Common logic for getting a revision store for a repository.
1137
see self._get_revision_store for the subclass-overridable method to
1138
get the store for a repository.
1140
from bzrlib.store.revision.text import TextRevisionStore
1141
dir_mode = control_files._dir_mode
1142
file_mode = control_files._file_mode
1143
text_store = TextStore(transport.clone(name),
1145
compressed=compressed,
1147
file_mode=file_mode)
1148
_revision_store = TextRevisionStore(text_store, serializer)
1149
return _revision_store
1151
# TODO: this shouldn't be in the base class, it's specific to things that
1152
# use weaves or knits -- mbp 20070207
1153
def _get_versioned_file_store(self,
1158
versionedfile_class=None,
1159
versionedfile_kwargs={},
1161
if versionedfile_class is None:
1162
versionedfile_class = self._versionedfile_class
1163
weave_transport = control_files._transport.clone(name)
1164
dir_mode = control_files._dir_mode
1165
file_mode = control_files._file_mode
1166
return VersionedFileStore(weave_transport, prefixed=prefixed,
1168
file_mode=file_mode,
1169
versionedfile_class=versionedfile_class,
1170
versionedfile_kwargs=versionedfile_kwargs,
1173
def initialize(self, a_bzrdir, shared=False):
1174
"""Initialize a repository of this format in a_bzrdir.
1176
:param a_bzrdir: The bzrdir to put the new repository in it.
1394
def initialize(self, controldir, shared=False):
1395
"""Initialize a repository of this format in controldir.
1397
:param controldir: The controldir to put the new repository in it.
1177
1398
:param shared: The repository should be initialized as a sharable one.
1399
:returns: The new repository object.
1179
1401
This may raise UninitializableFormat if shared repository are not
1180
compatible the a_bzrdir.
1402
compatible the controldir.
1404
raise NotImplementedError(self.initialize)
1183
1406
def is_supported(self):
1184
1407
"""Is this format supported?
1186
1409
Supported formats must be initializable and openable.
1187
Unsupported formats may not support initialization or committing or
1410
Unsupported formats may not support initialization or committing or
1188
1411
some other features depending on the reason for not being supported.
1415
def is_deprecated(self):
1416
"""Is this format deprecated?
1418
Deprecated formats may trigger a user-visible warning recommending
1419
the user to upgrade. They are still fully supported.
1423
def network_name(self):
1424
"""A simple byte string uniquely identifying this format for RPC calls.
1426
MetaDir repository formats use their disk format string to identify the
1427
repository over the wire. All in one formats such as bzr < 0.8, and
1428
foreign formats like svn/git and hg should use some marker which is
1429
unique and immutable.
1431
raise NotImplementedError(self.network_name)
1192
1433
def check_conversion_target(self, target_format):
1193
raise NotImplementedError(self.check_conversion_target)
1195
def open(self, a_bzrdir, _found=False):
1196
"""Return an instance of this format for the bzrdir a_bzrdir.
1434
if self.rich_root_data and not target_format.rich_root_data:
1435
raise errors.BadConversionTarget(
1436
'Does not support rich root data.', target_format,
1438
if (self.supports_tree_reference and
1439
not getattr(target_format, 'supports_tree_reference', False)):
1440
raise errors.BadConversionTarget(
1441
'Does not support nested trees', target_format,
1444
def open(self, controldir, _found=False):
1445
"""Return an instance of this format for a controldir.
1198
1447
_found is a private parameter, do not use it.
1200
1449
raise NotImplementedError(self.open)
1203
class MetaDirRepositoryFormat(RepositoryFormat):
1451
def _run_post_repo_init_hooks(self, repository, controldir, shared):
1452
from bzrlib.controldir import ControlDir, RepoInitHookParams
1453
hooks = ControlDir.hooks['post_repo_init']
1456
params = RepoInitHookParams(repository, self, controldir, shared)
1461
class RepositoryFormatMetaDir(bzrdir.BzrFormat, RepositoryFormat):
1204
1462
"""Common base class for the new repositories using the metadir layout."""
1206
1464
rich_root_data = False
1207
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1465
supports_tree_reference = False
1466
supports_external_lookups = False
1467
supports_leaving_lock = True
1468
supports_nesting_repositories = True
1471
def _matchingbzrdir(self):
1472
matching = bzrdir.BzrDirMetaFormat1()
1473
matching.repository_format = self
1209
1476
def __init__(self):
1210
super(MetaDirRepositoryFormat, self).__init__()
1477
RepositoryFormat.__init__(self)
1478
bzrdir.BzrFormat.__init__(self)
1212
1480
def _create_control_files(self, a_bzrdir):
1213
1481
"""Create the required files and the initial control_files object."""
1279
1632
_optimisers = []
1280
1633
"""The available optimised InterRepository types."""
1282
def copy_content(self, revision_id=None, basis=None):
1283
raise NotImplementedError(self.copy_content)
1285
def fetch(self, revision_id=None, pb=None):
1636
def copy_content(self, revision_id=None):
1637
"""Make a complete copy of the content in self into destination.
1639
This is a destructive operation! Do not use it on existing
1642
:param revision_id: Only copy the content needed to construct
1643
revision_id and its parents.
1646
self.target.set_make_working_trees(self.source.make_working_trees())
1647
except NotImplementedError:
1649
self.target.fetch(self.source, revision_id=revision_id)
1652
def fetch(self, revision_id=None, find_ghosts=False):
1286
1653
"""Fetch the content required to construct revision_id.
1288
1655
The content is copied from self.source to self.target.
1290
1657
:param revision_id: if None all content is copied, if NULL_REVISION no
1291
1658
content is copied.
1292
:param pb: optional progress bar to use for progress reports. If not
1293
provided a default one will be created.
1295
Returns the copied revision count and the failed revisions in a tuple:
1298
1661
raise NotImplementedError(self.fetch)
1300
1663
@needs_read_lock
1301
def missing_revision_ids(self, revision_id=None):
1664
def search_missing_revision_ids(self,
1665
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
1666
find_ghosts=True, revision_ids=None, if_present_ids=None,
1302
1668
"""Return the revision ids that source has that target does not.
1304
These are returned in topological order.
1306
1670
:param revision_id: only return revision ids included by this
1309
# generic, possibly worst case, slow code path.
1310
target_ids = set(self.target.all_revision_ids())
1311
if revision_id is not None:
1312
# TODO: jam 20070210 InterRepository is internal enough that it
1313
# should assume revision_ids are already utf-8
1314
revision_id = osutils.safe_revision_id(revision_id)
1315
source_ids = self.source.get_ancestry(revision_id)
1316
assert source_ids[0] is None
1319
source_ids = self.source.all_revision_ids()
1320
result_set = set(source_ids).difference(target_ids)
1321
# this may look like a no-op: its not. It preserves the ordering
1322
# other_ids had while only returning the members from other_ids
1323
# that we've decided we need.
1324
return [rev_id for rev_id in source_ids if rev_id in result_set]
1327
class InterSameDataRepository(InterRepository):
1328
"""Code for converting between repositories that represent the same data.
1330
Data format and model must match for this to work.
1334
def _get_repo_format_to_test(self):
1335
"""Repository format for testing with."""
1336
return RepositoryFormat.get_default_format()
1339
def is_compatible(source, target):
1340
if not isinstance(source, Repository):
1342
if not isinstance(target, Repository):
1344
if source._format.rich_root_data != target._format.rich_root_data:
1672
:param revision_ids: return revision ids included by these
1673
revision_ids. NoSuchRevision will be raised if any of these
1674
revisions are not present.
1675
:param if_present_ids: like revision_ids, but will not cause
1676
NoSuchRevision if any of these are absent, instead they will simply
1677
not be in the result. This is useful for e.g. finding revisions
1678
to fetch for tags, which may reference absent revisions.
1679
:param find_ghosts: If True find missing revisions in deep history
1680
rather than just finding the surface difference.
1681
:param limit: Maximum number of revisions to return, topologically
1683
:return: A bzrlib.graph.SearchResult.
1685
raise NotImplementedError(self.search_missing_revision_ids)
1688
def _same_model(source, target):
1689
"""True if source and target have the same data representation.
1691
Note: this is always called on the base class; overriding it in a
1692
subclass will have no effect.
1695
InterRepository._assert_same_model(source, target)
1697
except errors.IncompatibleRepositories, e:
1701
def _assert_same_model(source, target):
1702
"""Raise an exception if two repositories do not use the same model.
1704
if source.supports_rich_root() != target.supports_rich_root():
1705
raise errors.IncompatibleRepositories(source, target,
1706
"different rich-root support")
1346
1707
if source._serializer != target._serializer:
1352
def copy_content(self, revision_id=None, basis=None):
1353
"""Make a complete copy of the content in self into destination.
1355
This is a destructive operation! Do not use it on existing
1358
:param revision_id: Only copy the content needed to construct
1359
revision_id and its parents.
1360
:param basis: Copy the needed data preferentially from basis.
1363
self.target.set_make_working_trees(self.source.make_working_trees())
1364
except NotImplementedError:
1366
# TODO: jam 20070210 This is fairly internal, so we should probably
1367
# just assert that revision_id is not unicode.
1368
revision_id = osutils.safe_revision_id(revision_id)
1369
# grab the basis available data
1370
if basis is not None:
1371
self.target.fetch(basis, revision_id=revision_id)
1372
# but don't bother fetching if we have the needed data now.
1373
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1374
self.target.has_revision(revision_id)):
1376
self.target.fetch(self.source, revision_id=revision_id)
1379
def fetch(self, revision_id=None, pb=None):
1380
"""See InterRepository.fetch()."""
1381
from bzrlib.fetch import GenericRepoFetcher
1382
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1383
self.source, self.source._format, self.target,
1384
self.target._format)
1385
# TODO: jam 20070210 This should be an assert, not a translate
1386
revision_id = osutils.safe_revision_id(revision_id)
1387
f = GenericRepoFetcher(to_repository=self.target,
1388
from_repository=self.source,
1389
last_revision=revision_id,
1391
return f.count_copied, f.failed_revisions
1394
class InterWeaveRepo(InterSameDataRepository):
1395
"""Optimised code paths between Weave based repositories."""
1398
def _get_repo_format_to_test(self):
1399
from bzrlib.repofmt import weaverepo
1400
return weaverepo.RepositoryFormat7()
1403
def is_compatible(source, target):
1404
"""Be compatible with known Weave formats.
1406
We don't test for the stores being of specific types because that
1407
could lead to confusing results, and there is no need to be
1410
from bzrlib.repofmt.weaverepo import (
1416
return (isinstance(source._format, (RepositoryFormat5,
1418
RepositoryFormat7)) and
1419
isinstance(target._format, (RepositoryFormat5,
1421
RepositoryFormat7)))
1422
except AttributeError:
1426
def copy_content(self, revision_id=None, basis=None):
1427
"""See InterRepository.copy_content()."""
1428
# weave specific optimised path:
1429
# TODO: jam 20070210 Internal, should be an assert, not translate
1430
revision_id = osutils.safe_revision_id(revision_id)
1431
if basis is not None:
1432
# copy the basis in, then fetch remaining data.
1433
basis.copy_content_into(self.target, revision_id)
1434
# the basis copy_content_into could miss-set this.
1436
self.target.set_make_working_trees(self.source.make_working_trees())
1437
except NotImplementedError:
1439
self.target.fetch(self.source, revision_id=revision_id)
1442
self.target.set_make_working_trees(self.source.make_working_trees())
1443
except NotImplementedError:
1445
# FIXME do not peek!
1446
if self.source.control_files._transport.listable():
1447
pb = ui.ui_factory.nested_progress_bar()
1449
self.target.weave_store.copy_all_ids(
1450
self.source.weave_store,
1452
from_transaction=self.source.get_transaction(),
1453
to_transaction=self.target.get_transaction())
1454
pb.update('copying inventory', 0, 1)
1455
self.target.control_weaves.copy_multi(
1456
self.source.control_weaves, ['inventory'],
1457
from_transaction=self.source.get_transaction(),
1458
to_transaction=self.target.get_transaction())
1459
self.target._revision_store.text_store.copy_all_ids(
1460
self.source._revision_store.text_store,
1465
self.target.fetch(self.source, revision_id=revision_id)
1468
def fetch(self, revision_id=None, pb=None):
1469
"""See InterRepository.fetch()."""
1470
from bzrlib.fetch import GenericRepoFetcher
1471
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1472
self.source, self.source._format, self.target, self.target._format)
1473
# TODO: jam 20070210 This should be an assert, not a translate
1474
revision_id = osutils.safe_revision_id(revision_id)
1475
f = GenericRepoFetcher(to_repository=self.target,
1476
from_repository=self.source,
1477
last_revision=revision_id,
1479
return f.count_copied, f.failed_revisions
1482
def missing_revision_ids(self, revision_id=None):
1483
"""See InterRepository.missing_revision_ids()."""
1484
# we want all revisions to satisfy revision_id in source.
1485
# but we don't want to stat every file here and there.
1486
# we want then, all revisions other needs to satisfy revision_id
1487
# checked, but not those that we have locally.
1488
# so the first thing is to get a subset of the revisions to
1489
# satisfy revision_id in source, and then eliminate those that
1490
# we do already have.
1491
# this is slow on high latency connection to self, but as as this
1492
# disk format scales terribly for push anyway due to rewriting
1493
# inventory.weave, this is considered acceptable.
1495
if revision_id is not None:
1496
source_ids = self.source.get_ancestry(revision_id)
1497
assert source_ids[0] is None
1500
source_ids = self.source._all_possible_ids()
1501
source_ids_set = set(source_ids)
1502
# source_ids is the worst possible case we may need to pull.
1503
# now we want to filter source_ids against what we actually
1504
# have in target, but don't try to check for existence where we know
1505
# we do not have a revision as that would be pointless.
1506
target_ids = set(self.target._all_possible_ids())
1507
possibly_present_revisions = target_ids.intersection(source_ids_set)
1508
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1509
required_revisions = source_ids_set.difference(actually_present_revisions)
1510
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1511
if revision_id is not None:
1512
# we used get_ancestry to determine source_ids then we are assured all
1513
# revisions referenced are present as they are installed in topological order.
1514
# and the tip revision was validated by get_ancestry.
1515
return required_topo_revisions
1517
# if we just grabbed the possibly available ids, then
1518
# we only have an estimate of whats available and need to validate
1519
# that against the revision records.
1520
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1523
class InterKnitRepo(InterSameDataRepository):
1524
"""Optimised code paths between Knit based repositories."""
1527
def _get_repo_format_to_test(self):
1528
from bzrlib.repofmt import knitrepo
1529
return knitrepo.RepositoryFormatKnit1()
1532
def is_compatible(source, target):
1533
"""Be compatible with known Knit formats.
1535
We don't test for the stores being of specific types because that
1536
could lead to confusing results, and there is no need to be
1539
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1541
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1542
isinstance(target._format, (RepositoryFormatKnit1)))
1543
except AttributeError:
1547
def fetch(self, revision_id=None, pb=None):
1548
"""See InterRepository.fetch()."""
1549
from bzrlib.fetch import KnitRepoFetcher
1550
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1551
self.source, self.source._format, self.target, self.target._format)
1552
# TODO: jam 20070210 This should be an assert, not a translate
1553
revision_id = osutils.safe_revision_id(revision_id)
1554
f = KnitRepoFetcher(to_repository=self.target,
1555
from_repository=self.source,
1556
last_revision=revision_id,
1558
return f.count_copied, f.failed_revisions
1561
def missing_revision_ids(self, revision_id=None):
1562
"""See InterRepository.missing_revision_ids()."""
1563
if revision_id is not None:
1564
source_ids = self.source.get_ancestry(revision_id)
1565
assert source_ids[0] is None
1568
source_ids = self.source._all_possible_ids()
1569
source_ids_set = set(source_ids)
1570
# source_ids is the worst possible case we may need to pull.
1571
# now we want to filter source_ids against what we actually
1572
# have in target, but don't try to check for existence where we know
1573
# we do not have a revision as that would be pointless.
1574
target_ids = set(self.target._all_possible_ids())
1575
possibly_present_revisions = target_ids.intersection(source_ids_set)
1576
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1577
required_revisions = source_ids_set.difference(actually_present_revisions)
1578
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1579
if revision_id is not None:
1580
# we used get_ancestry to determine source_ids then we are assured all
1581
# revisions referenced are present as they are installed in topological order.
1582
# and the tip revision was validated by get_ancestry.
1583
return required_topo_revisions
1585
# if we just grabbed the possibly available ids, then
1586
# we only have an estimate of whats available and need to validate
1587
# that against the revision records.
1588
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1591
class InterModel1and2(InterRepository):
1594
def _get_repo_format_to_test(self):
1598
def is_compatible(source, target):
1599
if not isinstance(source, Repository):
1601
if not isinstance(target, Repository):
1603
if not source._format.rich_root_data and target._format.rich_root_data:
1609
def fetch(self, revision_id=None, pb=None):
1610
"""See InterRepository.fetch()."""
1611
from bzrlib.fetch import Model1toKnit2Fetcher
1612
# TODO: jam 20070210 This should be an assert, not a translate
1613
revision_id = osutils.safe_revision_id(revision_id)
1614
f = Model1toKnit2Fetcher(to_repository=self.target,
1615
from_repository=self.source,
1616
last_revision=revision_id,
1618
return f.count_copied, f.failed_revisions
1621
def copy_content(self, revision_id=None, basis=None):
1622
"""Make a complete copy of the content in self into destination.
1624
This is a destructive operation! Do not use it on existing
1627
:param revision_id: Only copy the content needed to construct
1628
revision_id and its parents.
1629
:param basis: Copy the needed data preferentially from basis.
1632
self.target.set_make_working_trees(self.source.make_working_trees())
1633
except NotImplementedError:
1635
# TODO: jam 20070210 Internal, assert, don't translate
1636
revision_id = osutils.safe_revision_id(revision_id)
1637
# grab the basis available data
1638
if basis is not None:
1639
self.target.fetch(basis, revision_id=revision_id)
1640
# but don't bother fetching if we have the needed data now.
1641
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1642
self.target.has_revision(revision_id)):
1644
self.target.fetch(self.source, revision_id=revision_id)
1647
class InterKnit1and2(InterKnitRepo):
1650
def _get_repo_format_to_test(self):
1654
def is_compatible(source, target):
1655
"""Be compatible with Knit1 source and Knit3 target"""
1656
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1658
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1659
RepositoryFormatKnit3
1660
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1661
isinstance(target._format, (RepositoryFormatKnit3)))
1662
except AttributeError:
1666
def fetch(self, revision_id=None, pb=None):
1667
"""See InterRepository.fetch()."""
1668
from bzrlib.fetch import Knit1to2Fetcher
1669
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1670
self.source, self.source._format, self.target,
1671
self.target._format)
1672
# TODO: jam 20070210 This should be an assert, not a translate
1673
revision_id = osutils.safe_revision_id(revision_id)
1674
f = Knit1to2Fetcher(to_repository=self.target,
1675
from_repository=self.source,
1676
last_revision=revision_id,
1678
return f.count_copied, f.failed_revisions
1681
InterRepository.register_optimiser(InterSameDataRepository)
1682
InterRepository.register_optimiser(InterWeaveRepo)
1683
InterRepository.register_optimiser(InterKnitRepo)
1684
InterRepository.register_optimiser(InterModel1and2)
1685
InterRepository.register_optimiser(InterKnit1and2)
1688
class RepositoryTestProviderAdapter(object):
1689
"""A tool to generate a suite testing multiple repository formats at once.
1691
This is done by copying the test once for each transport and injecting
1692
the transport_server, transport_readonly_server, and bzrdir_format and
1693
repository_format classes into each copy. Each copy is also given a new id()
1694
to make it easy to identify.
1697
def __init__(self, transport_server, transport_readonly_server, formats):
1698
self._transport_server = transport_server
1699
self._transport_readonly_server = transport_readonly_server
1700
self._formats = formats
1702
def adapt(self, test):
1703
result = unittest.TestSuite()
1704
for repository_format, bzrdir_format in self._formats:
1705
from copy import deepcopy
1706
new_test = deepcopy(test)
1707
new_test.transport_server = self._transport_server
1708
new_test.transport_readonly_server = self._transport_readonly_server
1709
new_test.bzrdir_format = bzrdir_format
1710
new_test.repository_format = repository_format
1711
def make_new_test_id():
1712
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1713
return lambda: new_id
1714
new_test.id = make_new_test_id()
1715
result.addTest(new_test)
1719
class InterRepositoryTestProviderAdapter(object):
1720
"""A tool to generate a suite testing multiple inter repository formats.
1722
This is done by copying the test once for each interrepo provider and injecting
1723
the transport_server, transport_readonly_server, repository_format and
1724
repository_to_format classes into each copy.
1725
Each copy is also given a new id() to make it easy to identify.
1728
def __init__(self, transport_server, transport_readonly_server, formats):
1729
self._transport_server = transport_server
1730
self._transport_readonly_server = transport_readonly_server
1731
self._formats = formats
1733
def adapt(self, test):
1734
result = unittest.TestSuite()
1735
for interrepo_class, repository_format, repository_format_to in self._formats:
1736
from copy import deepcopy
1737
new_test = deepcopy(test)
1738
new_test.transport_server = self._transport_server
1739
new_test.transport_readonly_server = self._transport_readonly_server
1740
new_test.interrepo_class = interrepo_class
1741
new_test.repository_format = repository_format
1742
new_test.repository_format_to = repository_format_to
1743
def make_new_test_id():
1744
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1745
return lambda: new_id
1746
new_test.id = make_new_test_id()
1747
result.addTest(new_test)
1751
def default_test_list():
1752
"""Generate the default list of interrepo permutations to test."""
1753
from bzrlib.repofmt import knitrepo, weaverepo
1755
# test the default InterRepository between format 6 and the current
1757
# XXX: robertc 20060220 reinstate this when there are two supported
1758
# formats which do not have an optimal code path between them.
1759
#result.append((InterRepository,
1760
# RepositoryFormat6(),
1761
# RepositoryFormatKnit1()))
1762
for optimiser_class in InterRepository._optimisers:
1763
format_to_test = optimiser_class._get_repo_format_to_test()
1764
if format_to_test is not None:
1765
result.append((optimiser_class,
1766
format_to_test, format_to_test))
1767
# if there are specific combinations we want to use, we can add them
1769
result.append((InterModel1and2,
1770
weaverepo.RepositoryFormat5(),
1771
knitrepo.RepositoryFormatKnit3()))
1772
result.append((InterKnit1and2,
1773
knitrepo.RepositoryFormatKnit1(),
1774
knitrepo.RepositoryFormatKnit3()))
1708
raise errors.IncompatibleRepositories(source, target,
1709
"different serializers")
1778
1712
class CopyConverter(object):
1779
1713
"""A repository conversion tool which just performs a copy of the content.
1781
1715
This is slow but quite reliable.
1787
1721
:param target_format: The format the resulting repository should be.
1789
1723
self.target_format = target_format
1791
1725
def convert(self, repo, pb):
1792
1726
"""Perform the conversion of to_convert, giving feedback via pb.
1794
1728
:param to_convert: The disk object to convert.
1795
1729
:param pb: a progress bar to use for progress information.
1731
pb = ui.ui_factory.nested_progress_bar()
1800
1734
# this is only useful with metadir layouts - separated repo content.
1801
1735
# trigger an assertion if not such
1802
1736
repo._format.get_format_string()
1803
1737
self.repo_dir = repo.bzrdir
1804
self.step('Moving repository to repository.backup')
1738
pb.update(gettext('Moving repository to repository.backup'))
1805
1739
self.repo_dir.transport.move('repository', 'repository.backup')
1806
1740
backup_transport = self.repo_dir.transport.clone('repository.backup')
1807
1741
repo._format.check_conversion_target(self.target_format)
1808
1742
self.source_repo = repo._format.open(self.repo_dir,
1810
1744
_override_transport=backup_transport)
1811
self.step('Creating new repository')
1745
pb.update(gettext('Creating new repository'))
1812
1746
converted = self.target_format.initialize(self.repo_dir,
1813
1747
self.source_repo.is_shared())
1814
1748
converted.lock_write()
1816
self.step('Copying content into repository.')
1750
pb.update(gettext('Copying content'))
1817
1751
self.source_repo.copy_content_into(converted)
1819
1753
converted.unlock()
1820
self.step('Deleting old repository content.')
1754
pb.update(gettext('Deleting old repository content'))
1821
1755
self.repo_dir.transport.delete_tree('repository.backup')
1822
self.pb.note('repository converted')
1824
def step(self, message):
1825
"""Update the pb by a step."""
1827
self.pb.update(message, self.count, self.total)
1830
class CommitBuilder(object):
1831
"""Provides an interface to build up a commit.
1833
This allows describing a tree to be committed without needing to
1834
know the internals of the format of the repository.
1837
record_root_entry = False
1838
def __init__(self, repository, parents, config, timestamp=None,
1839
timezone=None, committer=None, revprops=None,
1841
"""Initiate a CommitBuilder.
1843
:param repository: Repository to commit to.
1844
:param parents: Revision ids of the parents of the new revision.
1845
:param config: Configuration to use.
1846
:param timestamp: Optional timestamp recorded for commit.
1847
:param timezone: Optional timezone for timestamp.
1848
:param committer: Optional committer to set for commit.
1849
:param revprops: Optional dictionary of revision properties.
1850
:param revision_id: Optional revision id.
1852
self._config = config
1854
if committer is None:
1855
self._committer = self._config.username()
1857
assert isinstance(committer, basestring), type(committer)
1858
self._committer = committer
1860
self.new_inventory = Inventory(None)
1861
self._new_revision_id = osutils.safe_revision_id(revision_id)
1862
self.parents = parents
1863
self.repository = repository
1866
if revprops is not None:
1867
self._revprops.update(revprops)
1869
if timestamp is None:
1870
timestamp = time.time()
1871
# Restrict resolution to 1ms
1872
self._timestamp = round(timestamp, 3)
1874
if timezone is None:
1875
self._timezone = osutils.local_time_offset()
1877
self._timezone = int(timezone)
1879
self._generate_revision_if_needed()
1881
def commit(self, message):
1882
"""Make the actual commit.
1884
:return: The revision id of the recorded revision.
1886
rev = _mod_revision.Revision(
1887
timestamp=self._timestamp,
1888
timezone=self._timezone,
1889
committer=self._committer,
1891
inventory_sha1=self.inv_sha1,
1892
revision_id=self._new_revision_id,
1893
properties=self._revprops)
1894
rev.parent_ids = self.parents
1895
self.repository.add_revision(self._new_revision_id, rev,
1896
self.new_inventory, self._config)
1897
return self._new_revision_id
1899
def revision_tree(self):
1900
"""Return the tree that was just committed.
1902
After calling commit() this can be called to get a RevisionTree
1903
representing the newly committed tree. This is preferred to
1904
calling Repository.revision_tree() because that may require
1905
deserializing the inventory, while we already have a copy in
1908
return RevisionTree(self.repository, self.new_inventory,
1909
self._new_revision_id)
1911
def finish_inventory(self):
1912
"""Tell the builder that the inventory is finished."""
1913
if self.new_inventory.root is None:
1914
symbol_versioning.warn('Root entry should be supplied to'
1915
' record_entry_contents, as of bzr 0.10.',
1916
DeprecationWarning, stacklevel=2)
1917
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1918
self.new_inventory.revision_id = self._new_revision_id
1919
self.inv_sha1 = self.repository.add_inventory(
1920
self._new_revision_id,
1925
def _gen_revision_id(self):
1926
"""Return new revision-id."""
1927
return generate_ids.gen_revision_id(self._config.username(),
1930
def _generate_revision_if_needed(self):
1931
"""Create a revision id if None was supplied.
1933
If the repository can not support user-specified revision ids
1934
they should override this function and raise CannotSetRevisionId
1935
if _new_revision_id is not None.
1937
:raises: CannotSetRevisionId
1939
if self._new_revision_id is None:
1940
self._new_revision_id = self._gen_revision_id()
1942
def record_entry_contents(self, ie, parent_invs, path, tree):
1943
"""Record the content of ie from tree into the commit if needed.
1945
Side effect: sets ie.revision when unchanged
1947
:param ie: An inventory entry present in the commit.
1948
:param parent_invs: The inventories of the parent revisions of the
1950
:param path: The path the entry is at in the tree.
1951
:param tree: The tree which contains this entry and should be used to
1954
if self.new_inventory.root is None and ie.parent_id is not None:
1955
symbol_versioning.warn('Root entry should be supplied to'
1956
' record_entry_contents, as of bzr 0.10.',
1957
DeprecationWarning, stacklevel=2)
1958
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1960
self.new_inventory.add(ie)
1962
# ie.revision is always None if the InventoryEntry is considered
1963
# for committing. ie.snapshot will record the correct revision
1964
# which may be the sole parent if it is untouched.
1965
if ie.revision is not None:
1968
# In this revision format, root entries have no knit or weave
1969
if ie is self.new_inventory.root:
1970
# When serializing out to disk and back in
1971
# root.revision is always _new_revision_id
1972
ie.revision = self._new_revision_id
1974
previous_entries = ie.find_previous_heads(
1976
self.repository.weave_store,
1977
self.repository.get_transaction())
1978
# we are creating a new revision for ie in the history store
1980
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1982
def modified_directory(self, file_id, file_parents):
1983
"""Record the presence of a symbolic link.
1985
:param file_id: The file_id of the link to record.
1986
:param file_parents: The per-file parent revision ids.
1988
self._add_text_to_weave(file_id, [], file_parents.keys())
1990
def modified_reference(self, file_id, file_parents):
1991
"""Record the modification of a reference.
1993
:param file_id: The file_id of the link to record.
1994
:param file_parents: The per-file parent revision ids.
1996
self._add_text_to_weave(file_id, [], file_parents.keys())
1998
def modified_file_text(self, file_id, file_parents,
1999
get_content_byte_lines, text_sha1=None,
2001
"""Record the text of file file_id
2003
:param file_id: The file_id of the file to record the text of.
2004
:param file_parents: The per-file parent revision ids.
2005
:param get_content_byte_lines: A callable which will return the byte
2007
:param text_sha1: Optional SHA1 of the file contents.
2008
:param text_size: Optional size of the file contents.
2010
# mutter('storing text of file {%s} in revision {%s} into %r',
2011
# file_id, self._new_revision_id, self.repository.weave_store)
2012
# special case to avoid diffing on renames or
2014
if (len(file_parents) == 1
2015
and text_sha1 == file_parents.values()[0].text_sha1
2016
and text_size == file_parents.values()[0].text_size):
2017
previous_ie = file_parents.values()[0]
2018
versionedfile = self.repository.weave_store.get_weave(file_id,
2019
self.repository.get_transaction())
2020
versionedfile.clone_text(self._new_revision_id,
2021
previous_ie.revision, file_parents.keys())
2022
return text_sha1, text_size
2024
new_lines = get_content_byte_lines()
2025
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2026
# should return the SHA1 and size
2027
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2028
return osutils.sha_strings(new_lines), \
2029
sum(map(len, new_lines))
2031
def modified_link(self, file_id, file_parents, link_target):
2032
"""Record the presence of a symbolic link.
2034
:param file_id: The file_id of the link to record.
2035
:param file_parents: The per-file parent revision ids.
2036
:param link_target: Target location of this link.
2038
self._add_text_to_weave(file_id, [], file_parents.keys())
2040
def _add_text_to_weave(self, file_id, new_lines, parents):
2041
versionedfile = self.repository.weave_store.get_weave_or_empty(
2042
file_id, self.repository.get_transaction())
2043
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2044
versionedfile.clear_cache()
2047
class _CommitBuilder(CommitBuilder):
2048
"""Temporary class so old CommitBuilders are detected properly
2050
Note: CommitBuilder works whether or not root entry is recorded.
2053
record_root_entry = True
2056
class RootCommitBuilder(CommitBuilder):
2057
"""This commitbuilder actually records the root id"""
2059
record_root_entry = True
2061
def record_entry_contents(self, ie, parent_invs, path, tree):
2062
"""Record the content of ie from tree into the commit if needed.
2064
Side effect: sets ie.revision when unchanged
2066
:param ie: An inventory entry present in the commit.
2067
:param parent_invs: The inventories of the parent revisions of the
2069
:param path: The path the entry is at in the tree.
2070
:param tree: The tree which contains this entry and should be used to
2073
assert self.new_inventory.root is not None or ie.parent_id is None
2074
self.new_inventory.add(ie)
2076
# ie.revision is always None if the InventoryEntry is considered
2077
# for committing. ie.snapshot will record the correct revision
2078
# which may be the sole parent if it is untouched.
2079
if ie.revision is not None:
2082
previous_entries = ie.find_previous_heads(
2084
self.repository.weave_store,
2085
self.repository.get_transaction())
2086
# we are creating a new revision for ie in the history store
2088
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2100
def _unescaper(match, _map=_unescape_map):
2101
code = match.group(1)
1756
ui.ui_factory.note(gettext('repository converted'))
1760
def _strip_NULL_ghosts(revision_graph):
1761
"""Also don't use this. more compatibility code for unmigrated clients."""
1762
# Filter ghosts, and null:
1763
if _mod_revision.NULL_REVISION in revision_graph:
1764
del revision_graph[_mod_revision.NULL_REVISION]
1765
for key, parents in revision_graph.items():
1766
revision_graph[key] = tuple(parent for parent in parents if parent
1768
return revision_graph
1771
def _iter_for_revno(repo, partial_history_cache, stop_index=None,
1772
stop_revision=None):
1773
"""Extend the partial history to include a given index
1775
If a stop_index is supplied, stop when that index has been reached.
1776
If a stop_revision is supplied, stop when that revision is
1777
encountered. Otherwise, stop when the beginning of history is
1780
:param stop_index: The index which should be present. When it is
1781
present, history extension will stop.
1782
:param stop_revision: The revision id which should be present. When
1783
it is encountered, history extension will stop.
1785
start_revision = partial_history_cache[-1]
1786
graph = repo.get_graph()
1787
iterator = graph.iter_lefthand_ancestry(start_revision,
1788
(_mod_revision.NULL_REVISION,))
2105
if not code.startswith('#'):
2107
return unichr(int(code[1:])).encode('utf8')
2113
def _unescape_xml(data):
2114
"""Unescape predefined XML entities in a string of data."""
2116
if _unescape_re is None:
2117
_unescape_re = re.compile('\&([^;]*);')
2118
return _unescape_re.sub(_unescaper, data)
1790
# skip the last revision in the list
1793
if (stop_index is not None and
1794
len(partial_history_cache) > stop_index):
1796
if partial_history_cache[-1] == stop_revision:
1798
revision_id = iterator.next()
1799
partial_history_cache.append(revision_id)
1800
except StopIteration:
1805
class _LazyListJoin(object):
1806
"""An iterable yielding the contents of many lists as one list.
1808
Each iterator made from this will reflect the current contents of the lists
1809
at the time the iterator is made.
1811
This is used by Repository's _make_parents_provider implementation so that
1814
pp = repo._make_parents_provider() # uses a list of fallback repos
1815
pp.add_fallback_repository(other_repo) # appends to that list
1816
result = pp.get_parent_map(...)
1817
# The result will include revs from other_repo
1820
def __init__(self, *list_parts):
1821
self.list_parts = list_parts
1825
for list_part in self.list_parts:
1826
full_list.extend(list_part)
1827
return iter(full_list)
1830
return "%s.%s(%s)" % (self.__module__, self.__class__.__name__,
added
removed
bzrlib/repository.py
