13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
17
19
from bzrlib.lazy_import import lazy_import
18
20
lazy_import(globals(), """
22
25
from bzrlib import (
32
37
revision as _mod_revision,
33
testament as _mod_testament,
37
from bzrlib.bundle import serializer
38
from bzrlib.i18n import gettext
47
from bzrlib.decorators import needs_read_lock, needs_write_lock, only_raises
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
48
50
from bzrlib.inter import InterObject
49
from bzrlib.lock import _RelockDebugMixin, LogicalLockResult
50
from bzrlib.trace import (
51
log_exception_quietly, note, mutter, mutter_callsite, warning)
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
56
from bzrlib.trace import mutter, note, warning
54
59
# Old formats display a warning, but only once
55
60
_deprecation_warning_done = False
58
class IsInWriteGroupError(errors.InternalBzrError):
60
_fmt = "May not refresh_data of repo %(repo)s while in a write group."
62
def __init__(self, repo):
63
errors.InternalBzrError.__init__(self, repo=repo)
66
class CommitBuilder(object):
67
"""Provides an interface to build up a commit.
69
This allows describing a tree to be committed without needing to
70
know the internals of the format of the repository.
73
# all clients should supply tree roots.
74
record_root_entry = True
75
# whether this commit builder supports the record_entry_contents interface
76
supports_record_entry_contents = False
78
def __init__(self, repository, parents, config, timestamp=None,
79
timezone=None, committer=None, revprops=None,
80
revision_id=None, lossy=False):
81
"""Initiate a CommitBuilder.
83
:param repository: Repository to commit to.
84
:param parents: Revision ids of the parents of the new revision.
85
:param timestamp: Optional timestamp recorded for commit.
86
:param timezone: Optional timezone for timestamp.
87
:param committer: Optional committer to set for commit.
88
:param revprops: Optional dictionary of revision properties.
89
:param revision_id: Optional revision id.
90
:param lossy: Whether to discard data that can not be natively
91
represented, when pushing to a foreign VCS
97
self._committer = self._config.username()
98
elif not isinstance(committer, unicode):
99
self._committer = committer.decode() # throw if non-ascii
101
self._committer = committer
103
self._new_revision_id = revision_id
104
self.parents = parents
105
self.repository = repository
108
if revprops is not None:
109
self._validate_revprops(revprops)
110
self._revprops.update(revprops)
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
118
self._timezone = osutils.local_time_offset()
120
self._timezone = int(timezone)
122
self._generate_revision_if_needed()
124
def any_changes(self):
125
"""Return True if any entries were changed.
127
This includes merge-only changes. It is the core for the --unchanged
130
:return: True if any changes have occured.
132
raise NotImplementedError(self.any_changes)
134
def _validate_unicode_text(self, text, context):
135
"""Verify things like commit messages don't have bogus characters."""
137
raise ValueError('Invalid value for %s: %r' % (context, text))
139
def _validate_revprops(self, revprops):
140
for key, value in revprops.iteritems():
141
# We know that the XML serializers do not round trip '\r'
142
# correctly, so refuse to accept them
143
if not isinstance(value, basestring):
144
raise ValueError('revision property (%s) is not a valid'
145
' (unicode) string: %r' % (key, value))
146
self._validate_unicode_text(value,
147
'revision property (%s)' % (key,))
149
def commit(self, message):
150
"""Make the actual commit.
152
:return: The revision id of the recorded revision.
154
raise NotImplementedError(self.commit)
157
"""Abort the commit that is being built.
159
raise NotImplementedError(self.abort)
161
def revision_tree(self):
162
"""Return the tree that was just committed.
164
After calling commit() this can be called to get a
165
RevisionTree representing the newly committed tree. This is
166
preferred to calling Repository.revision_tree() because that may
167
require deserializing the inventory, while we already have a copy in
170
raise NotImplementedError(self.revision_tree)
172
def finish_inventory(self):
173
"""Tell the builder that the inventory is finished.
175
:return: The inventory id in the repository, which can be used with
176
repository.get_inventory.
178
raise NotImplementedError(self.finish_inventory)
180
def _gen_revision_id(self):
181
"""Return new revision-id."""
182
return generate_ids.gen_revision_id(self._committer, self._timestamp)
184
def _generate_revision_if_needed(self):
185
"""Create a revision id if None was supplied.
187
If the repository can not support user-specified revision ids
188
they should override this function and raise CannotSetRevisionId
189
if _new_revision_id is not None.
191
:raises: CannotSetRevisionId
193
if self._new_revision_id is None:
194
self._new_revision_id = self._gen_revision_id()
195
self.random_revid = True
197
self.random_revid = False
199
def will_record_deletes(self):
200
"""Tell the commit builder that deletes are being notified.
202
This enables the accumulation of an inventory delta; for the resulting
203
commit to be valid, deletes against the basis MUST be recorded via
204
builder.record_delete().
206
raise NotImplementedError(self.will_record_deletes)
208
def record_iter_changes(self, tree, basis_revision_id, iter_changes):
209
"""Record a new tree via iter_changes.
211
:param tree: The tree to obtain text contents from for changed objects.
212
:param basis_revision_id: The revision id of the tree the iter_changes
213
has been generated against. Currently assumed to be the same
214
as self.parents[0] - if it is not, errors may occur.
215
:param iter_changes: An iter_changes iterator with the changes to apply
216
to basis_revision_id. The iterator must not include any items with
217
a current kind of None - missing items must be either filtered out
218
or errored-on beefore record_iter_changes sees the item.
219
:return: A generator of (file_id, relpath, fs_hash) tuples for use with
222
raise NotImplementedError(self.record_iter_changes)
225
class RepositoryWriteLockResult(LogicalLockResult):
226
"""The result of write locking a repository.
228
:ivar repository_token: The token obtained from the underlying lock, or
230
:ivar unlock: A callable which will unlock the lock.
233
def __init__(self, unlock, repository_token):
234
LogicalLockResult.__init__(self, unlock)
235
self.repository_token = repository_token
238
return "RepositoryWriteLockResult(%s, %s)" % (self.repository_token,
242
63
######################################################################
246
class Repository(_RelockDebugMixin, controldir.ControlComponent):
66
class Repository(object):
247
67
"""Repository holding history for one or more branches.
249
69
The repository holds and retrieves historical information including
250
70
revisions and file history. It's normally accessed only by the Branch,
251
71
which views a particular line of development through that history.
253
See VersionedFileRepository in bzrlib.vf_repository for the
254
base class for most Bazaar repositories.
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
257
def abort_write_group(self, suppress_errors=False):
258
"""Commit the contents accrued within the current write group.
260
:param suppress_errors: if true, abort_write_group will catch and log
261
unexpected errors that happen during the abort, rather than
262
allowing them to propagate. Defaults to False.
264
:seealso: start_write_group.
266
if self._write_group is not self.get_transaction():
267
# has an unlock or relock occured ?
270
'(suppressed) mismatched lock context and write group. %r, %r',
271
self._write_group, self.get_transaction())
273
raise errors.BzrError(
274
'mismatched lock context and write group. %r, %r' %
275
(self._write_group, self.get_transaction()))
277
self._abort_write_group()
278
except Exception, exc:
279
self._write_group = None
280
if not suppress_errors:
282
mutter('abort_write_group failed')
283
log_exception_quietly()
284
note(gettext('bzr: ERROR (ignored): %s'), exc)
285
self._write_group = None
287
def _abort_write_group(self):
288
"""Template method for per-repository write group cleanup.
290
This is called during abort before the write group is considered to be
291
finished and should cleanup any internal state accrued during the write
292
group. There is no requirement that data handed to the repository be
293
*not* made available - this is not a rollback - but neither should any
294
attempt be made to ensure that data added is fully commited. Abort is
295
invoked when an error has occured so futher disk or network operations
296
may not be possible or may error and if possible should not be
300
def add_fallback_repository(self, repository):
301
"""Add a repository to use for looking up data not held locally.
303
:param repository: A repository.
305
raise NotImplementedError(self.add_fallback_repository)
307
def _check_fallback_repository(self, repository):
308
"""Check that this repository can fallback to repository safely.
310
Raise an error if not.
312
:param repository: A repository to fallback to.
314
return InterRepository._assert_same_model(self, repository)
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()
316
150
def all_revision_ids(self):
317
"""Returns a list of all the revision ids in the repository.
151
"""Returns a list of all the revision ids in the repository.
319
This is conceptually deprecated because code should generally work on
320
the graph reachable from a particular revision, and ignore any other
321
revisions that might be present. There is no direct replacement
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
if 'evil' in debug.debug_flags:
325
mutter_callsite(2, "all_revision_ids is linear with history.")
326
157
return self._all_revision_ids()
328
160
def _all_revision_ids(self):
329
"""Returns a list of all the revision ids in the repository.
161
"""Returns a list of all the revision ids in the repository.
331
These are in as much topological order as the underlying store can
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.
334
raise NotImplementedError(self._all_revision_ids)
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)
336
175
def break_lock(self):
337
176
"""Break a lock if one is present from another instance.
342
181
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):
345
196
def create(a_bzrdir):
346
197
"""Construct the current default format repository in a_bzrdir."""
347
198
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
349
def __init__(self, _format, a_bzrdir, control_files):
200
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
350
201
"""instantiate a Repository.
352
203
:param _format: The format of the repository on disk.
353
204
:param a_bzrdir: The BzrDir of the repository.
354
:param control_files: Control files to use for locking, etc.
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.
356
# In the future we will have a single api for all stores for
357
# getting file texts, inventories and revisions, then
358
# this construct will accept instances of those things.
359
210
super(Repository, self).__init__()
360
211
self._format = _format
361
212
# the following are part of the public API for Repository:
362
213
self.bzrdir = a_bzrdir
363
214
self.control_files = control_files
365
self._write_group = None
366
# Additional places to query for data.
367
self._fallback_repositories = []
370
def user_transport(self):
371
return self.bzrdir.user_transport
374
def control_transport(self):
375
return self._transport
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()
377
227
def __repr__(self):
378
if self._fallback_repositories:
379
return '%s(%r, fallback_repositories=%r)' % (
380
self.__class__.__name__,
382
self._fallback_repositories)
384
return '%s(%r)' % (self.__class__.__name__,
387
def _has_same_fallbacks(self, other_repo):
388
"""Returns true if the repositories have the same fallbacks."""
389
my_fb = self._fallback_repositories
390
other_fb = other_repo._fallback_repositories
391
if len(my_fb) != len(other_fb):
393
for f, g in zip(my_fb, other_fb):
394
if not f.has_same_location(g):
398
def has_same_location(self, other):
399
"""Returns a boolean indicating if this repository is at the same
400
location as another repository.
402
This might return False even when two repository objects are accessing
403
the same physical repository via different URLs.
405
if self.__class__ is not other.__class__:
407
return (self.control_url == other.control_url)
409
def is_in_write_group(self):
410
"""Return True if there is an open write group.
412
:seealso: start_write_group.
414
return self._write_group is not None
228
return '%s(%r)' % (self.__class__.__name__,
229
self.bzrdir.transport.base)
416
231
def is_locked(self):
417
232
return self.control_files.is_locked()
419
def is_write_locked(self):
420
"""Return True if this object is write locked."""
421
return self.is_locked() and self.control_files._lock_mode == 'w'
423
def lock_write(self, token=None):
424
"""Lock this repository for writing.
426
This causes caching within the repository obejct to start accumlating
427
data during reads, and allows a 'write_group' to be obtained. Write
428
groups must be used for actual data insertion.
430
A token should be passed in if you know that you have locked the object
431
some other way, and need to synchronise this object's state with that
434
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
436
:param token: if this is already locked, then lock_write will fail
437
unless the token matches the existing lock.
438
:returns: a token if this instance supports tokens, otherwise None.
439
:raises TokenLockingNotSupported: when a token is given but this
440
instance doesn't support using token locks.
441
:raises MismatchedToken: if the specified token doesn't match the token
442
of the existing lock.
443
:seealso: start_write_group.
444
:return: A RepositoryWriteLockResult.
446
locked = self.is_locked()
447
token = self.control_files.lock_write(token=token)
449
self._warn_if_deprecated()
451
for repo in self._fallback_repositories:
452
# Writes don't affect fallback repos
455
return RepositoryWriteLockResult(self.unlock, token)
234
def lock_write(self):
235
self.control_files.lock_write()
457
237
def lock_read(self):
458
"""Lock the repository for read operations.
460
:return: An object with an unlock method which will release the lock
463
locked = self.is_locked()
464
238
self.control_files.lock_read()
466
self._warn_if_deprecated()
468
for repo in self._fallback_repositories:
471
return LogicalLockResult(self.unlock)
473
240
def get_physical_lock_status(self):
474
241
return self.control_files.get_physical_lock_status()
476
def leave_lock_in_place(self):
477
"""Tell this repository not to release the physical lock when this
480
If lock_write doesn't return a token, then this method is not supported.
482
self.control_files.leave_in_place()
484
def dont_leave_lock_in_place(self):
485
"""Tell this repository to release the physical lock when this
486
object is unlocked, even if it didn't originally acquire it.
488
If lock_write doesn't return a token, then this method is not supported.
490
self.control_files.dont_leave_in_place()
493
244
def gather_stats(self, revid=None, committers=None):
494
245
"""Gather statistics from a revision id.
605
311
control = bzrdir.BzrDir.open(base)
606
312
return control.open_repository()
608
def copy_content_into(self, destination, revision_id=None):
314
def copy_content_into(self, destination, revision_id=None, basis=None):
609
315
"""Make a complete copy of the content in self into destination.
611
This is a destructive operation! Do not use it on existing
317
This is a destructive operation! Do not use it on existing
614
return InterRepository.get(self, destination).copy_content(revision_id)
616
def commit_write_group(self):
617
"""Commit the contents accrued within the current write group.
619
:seealso: start_write_group.
621
:return: it may return an opaque hint that can be passed to 'pack'.
623
if self._write_group is not self.get_transaction():
624
# has an unlock or relock occured ?
625
raise errors.BzrError('mismatched lock context %r and '
627
(self.get_transaction(), self._write_group))
628
result = self._commit_write_group()
629
self._write_group = None
632
def _commit_write_group(self):
633
"""Template method for per-repository write group cleanup.
635
This is called before the write group is considered to be
636
finished and should ensure that all data handed to the repository
637
for writing during the write group is safely committed (to the
638
extent possible considering file system caching etc).
641
def suspend_write_group(self):
642
raise errors.UnsuspendableWriteGroup(self)
644
def refresh_data(self):
645
"""Re-read any data needed to synchronise with disk.
647
This method is intended to be called after another repository instance
648
(such as one used by a smart server) has inserted data into the
649
repository. On all repositories this will work outside of write groups.
650
Some repository formats (pack and newer for bzrlib native formats)
651
support refresh_data inside write groups. If called inside a write
652
group on a repository that does not support refreshing in a write group
653
IsInWriteGroupError will be raised.
657
def resume_write_group(self, tokens):
658
if not self.is_write_locked():
659
raise errors.NotWriteLocked(self)
660
if self._write_group:
661
raise errors.BzrError('already in a write group')
662
self._resume_write_group(tokens)
663
# so we can detect unlock/relock - the write group is now entered.
664
self._write_group = self.get_transaction()
666
def _resume_write_group(self, tokens):
667
raise errors.UnsuspendableWriteGroup(self)
669
def fetch(self, source, revision_id=None, find_ghosts=False,
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):
671
324
"""Fetch the content required to construct revision_id from source.
673
If revision_id is None and fetch_spec is None, then all content is
676
fetch() may not be used when the repository is in a write group -
677
either finish the current write group before using fetch, or use
678
fetch before starting the write group.
680
:param find_ghosts: Find and copy revisions in the source that are
681
ghosts in the target (and not reachable directly by walking out to
682
the first-present revision in target from revision_id).
683
:param revision_id: If specified, all the content needed for this
684
revision ID will be copied to the target. Fetch will determine for
685
itself which content needs to be copied.
686
:param fetch_spec: If specified, a SearchResult or
687
PendingAncestryResult that describes which revisions to copy. This
688
allows copying multiple heads at once. Mutually exclusive with
326
If revision_id is None all content is copied.
691
if fetch_spec is not None and revision_id is not None:
692
raise AssertionError(
693
"fetch_spec and revision_id are mutually exclusive.")
694
if self.is_in_write_group():
695
raise errors.InternalBzrError(
696
"May not fetch while in a write group.")
697
# fast path same-url fetch operations
698
# TODO: lift out to somewhere common with RemoteRepository
699
# <https://bugs.launchpad.net/bzr/+bug/401646>
700
if (self.has_same_location(source)
701
and fetch_spec is None
702
and self._has_same_fallbacks(source)):
703
# check that last_revision is in 'from' and then return a
705
if (revision_id is not None and
706
not _mod_revision.is_null(revision_id)):
707
self.get_revision(revision_id)
709
inter = InterRepository.get(source, self)
710
return inter.fetch(revision_id=revision_id,
711
find_ghosts=find_ghosts, fetch_spec=fetch_spec)
713
def create_bundle(self, target, base, fileobj, format=None):
714
return serializer.write_bundle(self, target, base, fileobj, format)
716
def get_commit_builder(self, branch, parents, config, timestamp=None,
717
timezone=None, committer=None, revprops=None,
718
revision_id=None, lossy=False):
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,
719
335
"""Obtain a CommitBuilder for this repository.
721
337
:param branch: Branch to commit to.
722
338
:param parents: Revision ids of the parents of the new revision.
723
339
:param config: Configuration to use.
807
366
# Most control formats need the repository to be specifically
808
367
# created, but on some old all-in-one formats it's not needed
810
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
369
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
811
370
except errors.UninitializableFormat:
812
371
dest_repo = a_bzrdir.open_repository()
372
self.copy_content_into(dest_repo, revision_id, basis)
816
376
def has_revision(self, revision_id):
817
377
"""True if this repository has a copy of the revision."""
818
return revision_id in self.has_revisions((revision_id,))
821
def has_revisions(self, revision_ids):
822
"""Probe to find out the presence of multiple revisions.
824
:param revision_ids: An iterable of revision_ids.
825
:return: A set of the revision_ids that were present.
827
raise NotImplementedError(self.has_revisions)
830
def get_revision(self, revision_id):
831
"""Return the Revision object for a named revision."""
832
return self.get_revisions([revision_id])[0]
378
revision_id = osutils.safe_revision_id(revision_id)
379
return self._revision_store.has_revision_id(revision_id,
380
self.get_transaction())
834
383
def get_revision_reconcile(self, revision_id):
835
384
"""'reconcile' helper routine that allows access to a revision always.
837
386
This variant of get_revision does not cross check the weave graph
838
387
against the revision one as get_revision does: but it should only
839
388
be used by reconcile, or reconcile-alike commands that are correcting
840
389
or testing the revision graph.
842
raise NotImplementedError(self.get_revision_reconcile)
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]
844
397
def get_revisions(self, revision_ids):
845
"""Get many revisions at once.
847
Repositories that need to check data on every revision read should
848
subclass this method.
850
raise NotImplementedError(self.get_revisions)
852
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
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):
853
439
"""Produce a generator of revision deltas.
855
441
Note that the input is a sequence of REVISIONS, not revision_ids.
856
442
Trees will be held in memory until the generator exits.
857
443
Each delta is relative to the revision's lefthand predecessor.
859
:param specific_fileids: if not None, the result is filtered
860
so that only those file-ids, their parents and their
861
children are included.
863
# Get the revision-ids of interest
864
445
required_trees = set()
865
446
for revision in revisions:
866
447
required_trees.add(revision.revision_id)
867
448
required_trees.update(revision.parent_ids[:1])
869
# Get the matching filtered trees. Note that it's more
870
# efficient to pass filtered trees to changes_from() rather
871
# than doing the filtering afterwards. changes_from() could
872
# arguably do the filtering itself but it's path-based, not
873
# file-id based, so filtering before or afterwards is
875
if specific_fileids is None:
876
trees = dict((t.get_revision_id(), t) for
877
t in self.revision_trees(required_trees))
879
trees = dict((t.get_revision_id(), t) for
880
t in self._filtered_revision_trees(required_trees,
883
# Calculate the deltas
449
trees = dict((t.get_revision_id(), t) for
450
t in self.revision_trees(required_trees))
884
451
for revision in revisions:
885
452
if not revision.parent_ids:
886
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
453
old_tree = self.revision_tree(None)
888
455
old_tree = trees[revision.parent_ids[0]]
889
456
yield trees[revision.revision_id].changes_from(old_tree)
892
def get_revision_delta(self, revision_id, specific_fileids=None):
459
def get_revision_delta(self, revision_id):
893
460
"""Return the delta for one revision.
895
462
The delta is relative to the left-hand predecessor of the
898
:param specific_fileids: if not None, the result is filtered
899
so that only those file-ids, their parents and their
900
children are included.
902
465
r = self.get_revision(revision_id)
903
return list(self.get_deltas_for_revisions([r],
904
specific_fileids=specific_fileids))[0]
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)
906
484
@needs_write_lock
907
485
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
486
revision_id = osutils.safe_revision_id(revision_id)
908
487
signature = gpg_strategy.sign(plaintext)
909
self.add_signature_text(revision_id, signature)
911
def add_signature_text(self, revision_id, signature):
912
"""Store a signature text for a revision.
914
:param revision_id: Revision id of the revision
915
:param signature: Signature text.
917
raise NotImplementedError(self.add_signature_text)
919
def _find_parent_ids_of_revisions(self, revision_ids):
920
"""Find all parent ids that are mentioned in the revision graph.
922
:return: set of revisions that are parents of revision_ids which are
923
not part of revision_ids themselves
925
parent_map = self.get_parent_map(revision_ids)
927
map(parent_ids.update, parent_map.itervalues())
928
parent_ids.difference_update(revision_ids)
929
parent_ids.discard(_mod_revision.NULL_REVISION)
488
self._revision_store.add_revision_signature_text(revision_id,
490
self.get_transaction())
932
492
def fileids_altered_by_revision_ids(self, revision_ids):
933
493
"""Find the file ids and versions affected by revisions.
935
495
:param revisions: an iterable containing revision ids.
936
496
:return: a dictionary mapping altered file-ids to an iterable of
937
revision_ids. Each altered file-ids has the exact revision_ids
938
that altered it listed explicitly.
940
raise NotImplementedError(self.fileids_altered_by_revision_ids)
942
def iter_files_bytes(self, desired_files):
943
"""Iterate through file versions.
945
Files will not necessarily be returned in the order they occur in
946
desired_files. No specific order is guaranteed.
948
Yields pairs of identifier, bytes_iterator. identifier is an opaque
949
value supplied by the caller as part of desired_files. It should
950
uniquely identify the file version in the caller's context. (Examples:
951
an index number or a TreeTransform trans_id.)
953
:param desired_files: a list of (file_id, revision_id, identifier)
956
raise NotImplementedError(self.iter_files_bytes)
958
def get_rev_id_for_revno(self, revno, known_pair):
959
"""Return the revision id of a revno, given a later (revno, revid)
960
pair in the same history.
962
:return: if found (True, revid). If the available history ran out
963
before reaching the revno, then this returns
964
(False, (closest_revno, closest_revid)).
966
known_revno, known_revid = known_pair
967
partial_history = [known_revid]
968
distance_from_known = known_revno - revno
969
if distance_from_known < 0:
971
'requested revno (%d) is later than given known revno (%d)'
972
% (revno, known_revno))
975
self, partial_history, stop_index=distance_from_known)
976
except errors.RevisionNotPresent, err:
977
if err.revision_id == known_revid:
978
# The start revision (known_revid) wasn't found.
980
# This is a stacked repository with no fallbacks, or a there's a
981
# left-hand ghost. Either way, even though the revision named in
982
# the error isn't in this repo, we know it's the next step in this
984
partial_history.append(err.revision_id)
985
if len(partial_history) <= distance_from_known:
986
# Didn't find enough history to get a revid for the revno.
987
earliest_revno = known_revno - len(partial_history) + 1
988
return (False, (earliest_revno, partial_history[-1]))
989
if len(partial_history) - 1 > distance_from_known:
990
raise AssertionError('_iter_for_revno returned too much history')
991
return (True, partial_history[-1])
993
@symbol_versioning.deprecated_method(symbol_versioning.deprecated_in((2, 4, 0)))
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()
994
692
def iter_reverse_revision_history(self, revision_id):
995
693
"""Iterate backwards through revision ids in the lefthand history
997
695
:param revision_id: The revision id to start with. All its lefthand
998
696
ancestors will be traversed.
1000
graph = self.get_graph()
1001
stop_revisions = (None, _mod_revision.NULL_REVISION)
1002
return graph.iter_lefthand_ancestry(revision_id, stop_revisions)
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)
1004
728
def is_shared(self):
1005
729
"""Return True if this repository is flagged as a shared repository."""
1006
730
raise NotImplementedError(self.is_shared)
1012
736
reconciler = RepoReconciler(self, thorough=thorough)
1013
737
reconciler.reconcile()
1014
738
return reconciler
1016
def _refresh_data(self):
1017
"""Helper called from lock_* to ensure coherency with disk.
1019
The default implementation does nothing; it is however possible
1020
for repositories to maintain loaded indices across multiple locks
1021
by checking inside their implementation of this method to see
1022
whether their indices are still valid. This depends of course on
1023
the disk format being validatable in this manner. This method is
1024
also called by the refresh_data() public interface to cause a refresh
1025
to occur while in a write lock so that data inserted by a smart server
1026
push operation is visible on the client's instance of the physical
1030
740
@needs_read_lock
1031
741
def revision_tree(self, revision_id):
1032
742
"""Return Tree for a revision on this branch.
1034
`revision_id` may be NULL_REVISION for the empty tree revision.
744
`revision_id` may be None for the empty tree revision.
1036
raise NotImplementedError(self.revision_tree)
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)
1038
757
def revision_trees(self, revision_ids):
1039
"""Return Trees for revisions in this repository.
758
"""Return Tree for a revision on this branch.
1041
:param revision_ids: a sequence of revision-ids;
1042
a revision-id may not be None or 'null:'
1044
raise NotImplementedError(self.revision_trees)
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)
1046
768
@needs_read_lock
1047
@symbol_versioning.deprecated_method(
1048
symbol_versioning.deprecated_in((2, 4, 0)))
1049
def get_ancestry(self, revision_id, topo_sorted=True):
769
def get_ancestry(self, revision_id):
1050
770
"""Return a list of revision-ids integrated by a revision.
1052
The first element of the list is always None, indicating the origin
1053
revision. This might change when we have history horizons, or
772
The first element of the list is always None, indicating the origin
773
revision. This might change when we have history horizons, or
1054
774
perhaps we should have a new API.
1056
776
This is topologically sorted.
1058
if 'evil' in debug.debug_flags:
1059
mutter_callsite(2, "get_ancestry is linear with history.")
1060
if _mod_revision.is_null(revision_id):
778
if revision_id is None:
780
revision_id = osutils.safe_revision_id(revision_id)
1062
781
if not self.has_revision(revision_id):
1063
782
raise errors.NoSuchRevision(self, revision_id)
1064
graph = self.get_graph()
1066
search = graph._make_breadth_first_searcher([revision_id])
1069
found, ghosts = search.next_with_ghosts()
1070
except StopIteration:
1073
if _mod_revision.NULL_REVISION in keys:
1074
keys.remove(_mod_revision.NULL_REVISION)
1076
parent_map = graph.get_parent_map(keys)
1077
keys = tsort.topo_sort(parent_map)
1078
return [None] + list(keys)
1080
def pack(self, hint=None, clean_obsolete_packs=False):
1081
"""Compress the data within the repository.
1083
This operation only makes sense for some repository types. For other
1084
types it should be a no-op that just returns.
1086
This stub method does not require a lock, but subclasses should use
1087
@needs_write_lock as this is a long running call it's reasonable to
1088
implicitly lock for the user.
1090
:param hint: If not supplied, the whole repository is packed.
1091
If supplied, the repository may use the hint parameter as a
1092
hint for the parts of the repository to pack. A hint can be
1093
obtained from the result of commit_write_group(). Out of
1094
date hints are simply ignored, because concurrent operations
1095
can obsolete them rapidly.
1097
:param clean_obsolete_packs: Clean obsolete packs immediately after
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)
1101
806
def get_transaction(self):
1102
807
return self.control_files.get_transaction()
1104
def get_parent_map(self, revision_ids):
1105
"""See graph.StackedParentsProvider.get_parent_map"""
1106
raise NotImplementedError(self.get_parent_map)
1108
def _get_parent_map_no_fallbacks(self, revision_ids):
1109
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1110
# revisions index works in keys; this just works in revisions
1111
# therefore wrap and unwrap
1114
for revision_id in revision_ids:
1115
if revision_id == _mod_revision.NULL_REVISION:
1116
result[revision_id] = ()
1117
elif revision_id is None:
1118
raise ValueError('get_parent_map(None) is not valid')
1120
query_keys.append((revision_id ,))
1121
vf = self.revisions.without_fallbacks()
1122
for ((revision_id,), parent_keys) in \
1123
vf.get_parent_map(query_keys).iteritems():
1125
result[revision_id] = tuple([parent_revid
1126
for (parent_revid,) in parent_keys])
1128
result[revision_id] = (_mod_revision.NULL_REVISION,)
1131
def _make_parents_provider(self):
1132
if not self._format.supports_external_lookups:
1134
return graph.StackedParentsProvider(_LazyListJoin(
1135
[self._make_parents_provider_unstacked()],
1136
self._fallback_repositories))
1138
def _make_parents_provider_unstacked(self):
1139
return graph.CallableToParentsProviderAdapter(
1140
self._get_parent_map_no_fallbacks)
1143
def get_known_graph_ancestry(self, revision_ids):
1144
"""Return the known graph for a set of revision ids and their ancestors.
1146
raise NotImplementedError(self.get_known_graph_ancestry)
1148
def get_file_graph(self):
1149
"""Return the graph walker for files."""
1150
raise NotImplementedError(self.get_file_graph)
1152
def get_graph(self, other_repository=None):
1153
"""Return the graph walker for this repository format"""
1154
parents_provider = self._make_parents_provider()
1155
if (other_repository is not None and
1156
not self.has_same_location(other_repository)):
1157
parents_provider = graph.StackedParentsProvider(
1158
[parents_provider, other_repository._make_parents_provider()])
1159
return graph.Graph(parents_provider)
1161
def revision_ids_to_search_result(self, result_set):
1162
"""Convert a set of revision ids to a graph SearchResult."""
1163
result_parents = set()
1164
for parents in self.get_graph().get_parent_map(
1165
result_set).itervalues():
1166
result_parents.update(parents)
1167
included_keys = result_set.intersection(result_parents)
1168
start_keys = result_set.difference(included_keys)
1169
exclude_keys = result_parents.difference(result_set)
1170
result = graph.SearchResult(start_keys, exclude_keys,
1171
len(result_set), result_set)
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)
1174
813
@needs_write_lock
1175
814
def set_make_working_trees(self, new_value):
1476
1123
"""Return the short description for this format."""
1477
1124
raise NotImplementedError(self.get_format_description)
1126
def _get_revision_store(self, repo_transport, control_files):
1127
"""Return the revision store object for this a_bzrdir."""
1128
raise NotImplementedError(self._get_revision_store)
1130
def _get_text_rev_store(self,
1137
"""Common logic for getting a revision store for a repository.
1139
see self._get_revision_store for the subclass-overridable method to
1140
get the store for a repository.
1142
from bzrlib.store.revision.text import TextRevisionStore
1143
dir_mode = control_files._dir_mode
1144
file_mode = control_files._file_mode
1145
text_store = TextStore(transport.clone(name),
1147
compressed=compressed,
1149
file_mode=file_mode)
1150
_revision_store = TextRevisionStore(text_store, serializer)
1151
return _revision_store
1153
# TODO: this shouldn't be in the base class, it's specific to things that
1154
# use weaves or knits -- mbp 20070207
1155
def _get_versioned_file_store(self,
1160
versionedfile_class=None,
1161
versionedfile_kwargs={},
1163
if versionedfile_class is None:
1164
versionedfile_class = self._versionedfile_class
1165
weave_transport = control_files._transport.clone(name)
1166
dir_mode = control_files._dir_mode
1167
file_mode = control_files._file_mode
1168
return VersionedFileStore(weave_transport, prefixed=prefixed,
1170
file_mode=file_mode,
1171
versionedfile_class=versionedfile_class,
1172
versionedfile_kwargs=versionedfile_kwargs,
1479
1175
def initialize(self, a_bzrdir, shared=False):
1480
1176
"""Initialize a repository of this format in a_bzrdir.
1482
1178
:param a_bzrdir: The bzrdir to put the new repository in it.
1483
1179
:param shared: The repository should be initialized as a sharable one.
1484
:returns: The new repository object.
1486
1181
This may raise UninitializableFormat if shared repository are not
1487
1182
compatible the a_bzrdir.
1489
raise NotImplementedError(self.initialize)
1491
1185
def is_supported(self):
1492
1186
"""Is this format supported?
1494
1188
Supported formats must be initializable and openable.
1495
Unsupported formats may not support initialization or committing or
1189
Unsupported formats may not support initialization or committing or
1496
1190
some other features depending on the reason for not being supported.
1500
def is_deprecated(self):
1501
"""Is this format deprecated?
1503
Deprecated formats may trigger a user-visible warning recommending
1504
the user to upgrade. They are still fully supported.
1508
def network_name(self):
1509
"""A simple byte string uniquely identifying this format for RPC calls.
1511
MetaDir repository formats use their disk format string to identify the
1512
repository over the wire. All in one formats such as bzr < 0.8, and
1513
foreign formats like svn/git and hg should use some marker which is
1514
unique and immutable.
1516
raise NotImplementedError(self.network_name)
1518
1194
def check_conversion_target(self, target_format):
1519
if self.rich_root_data and not target_format.rich_root_data:
1520
raise errors.BadConversionTarget(
1521
'Does not support rich root data.', target_format,
1523
if (self.supports_tree_reference and
1524
not getattr(target_format, 'supports_tree_reference', False)):
1525
raise errors.BadConversionTarget(
1526
'Does not support nested trees', target_format,
1195
raise NotImplementedError(self.check_conversion_target)
1529
1197
def open(self, a_bzrdir, _found=False):
1530
1198
"""Return an instance of this format for the bzrdir a_bzrdir.
1532
1200
_found is a private parameter, do not use it.
1534
1202
raise NotImplementedError(self.open)
1536
def _run_post_repo_init_hooks(self, repository, a_bzrdir, shared):
1537
from bzrlib.bzrdir import BzrDir, RepoInitHookParams
1538
hooks = BzrDir.hooks['post_repo_init']
1541
params = RepoInitHookParams(repository, self, a_bzrdir, shared)
1546
1205
class MetaDirRepositoryFormat(RepositoryFormat):
1547
1206
"""Common base class for the new repositories using the metadir layout."""
1549
1208
rich_root_data = False
1550
supports_tree_reference = False
1551
supports_external_lookups = False
1552
supports_leaving_lock = True
1553
supports_nesting_repositories = True
1556
def _matchingbzrdir(self):
1557
matching = bzrdir.BzrDirMetaFormat1()
1558
matching.repository_format = self
1209
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1561
1211
def __init__(self):
1562
1212
super(MetaDirRepositoryFormat, self).__init__()
1575
1225
"""Upload the initial blank content."""
1576
1226
control_files = self._create_control_files(a_bzrdir)
1577
1227
control_files.lock_write()
1578
transport = control_files._transport
1580
utf8_files += [('shared-storage', '')]
1582
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
1583
for (filename, content_stream) in files:
1584
transport.put_file(filename, content_stream,
1585
mode=a_bzrdir._get_file_mode())
1586
for (filename, content_bytes) in utf8_files:
1587
transport.put_bytes_non_atomic(filename, content_bytes,
1588
mode=a_bzrdir._get_file_mode())
1229
control_files._transport.mkdir_multi(dirs,
1230
mode=control_files._dir_mode)
1231
for file, content in files:
1232
control_files.put(file, content)
1233
for file, content in utf8_files:
1234
control_files.put_utf8(file, content)
1236
control_files.put_utf8('shared-storage', '')
1590
1238
control_files.unlock()
1592
def network_name(self):
1593
"""Metadir formats have matching disk and network format strings."""
1594
return self.get_format_string()
1597
# formats which have no format string are not discoverable or independently
1598
# creatable on disk, so are not registered in format_registry. They're
1599
# all in bzrlib.repofmt.knitreponow. When an instance of one of these is
1241
# formats which have no format string are not discoverable
1242
# and not independently creatable, so are not registered. They're
1243
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1600
1244
# needed, it's constructed directly by the BzrDir. Non-native formats where
1601
1245
# the repository is not separately opened are similar.
1603
1247
format_registry.register_lazy(
1248
'Bazaar-NG Repository format 7',
1249
'bzrlib.repofmt.weaverepo',
1252
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1253
# default control directory format
1255
format_registry.register_lazy(
1604
1256
'Bazaar-NG Knit Repository Format 1',
1605
1257
'bzrlib.repofmt.knitrepo',
1606
1258
'RepositoryFormatKnit1',
1609
format_registry.register_lazy(
1610
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1260
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1262
format_registry.register_lazy(
1263
'Bazaar Knit Repository Format 2\n',
1264
'bzrlib.repofmt.knitrepo',
1265
'RepositoryFormatKnit2',
1268
format_registry.register_lazy(
1269
'Bazaar Knit Repository Format 3\n',
1611
1270
'bzrlib.repofmt.knitrepo',
1612
1271
'RepositoryFormatKnit3',
1615
format_registry.register_lazy(
1616
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
1617
'bzrlib.repofmt.knitrepo',
1618
'RepositoryFormatKnit4',
1621
# Pack-based formats. There is one format for pre-subtrees, and one for
1622
# post-subtrees to allow ease of testing.
1623
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
1624
format_registry.register_lazy(
1625
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
1626
'bzrlib.repofmt.knitpack_repo',
1627
'RepositoryFormatKnitPack1',
1629
format_registry.register_lazy(
1630
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
1631
'bzrlib.repofmt.knitpack_repo',
1632
'RepositoryFormatKnitPack3',
1634
format_registry.register_lazy(
1635
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
1636
'bzrlib.repofmt.knitpack_repo',
1637
'RepositoryFormatKnitPack4',
1639
format_registry.register_lazy(
1640
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
1641
'bzrlib.repofmt.knitpack_repo',
1642
'RepositoryFormatKnitPack5',
1644
format_registry.register_lazy(
1645
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
1646
'bzrlib.repofmt.knitpack_repo',
1647
'RepositoryFormatKnitPack5RichRoot',
1649
format_registry.register_lazy(
1650
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
1651
'bzrlib.repofmt.knitpack_repo',
1652
'RepositoryFormatKnitPack5RichRootBroken',
1654
format_registry.register_lazy(
1655
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
1656
'bzrlib.repofmt.knitpack_repo',
1657
'RepositoryFormatKnitPack6',
1659
format_registry.register_lazy(
1660
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
1661
'bzrlib.repofmt.knitpack_repo',
1662
'RepositoryFormatKnitPack6RichRoot',
1664
format_registry.register_lazy(
1665
'Bazaar repository format 2a (needs bzr 1.16 or later)\n',
1666
'bzrlib.repofmt.groupcompress_repo',
1667
'RepositoryFormat2a',
1670
# Development formats.
1671
# Check their docstrings to see if/when they are obsolete.
1672
format_registry.register_lazy(
1673
("Bazaar development format 2 with subtree support "
1674
"(needs bzr.dev from before 1.8)\n"),
1675
'bzrlib.repofmt.knitpack_repo',
1676
'RepositoryFormatPackDevelopment2Subtree',
1678
format_registry.register_lazy(
1679
'Bazaar development format 8\n',
1680
'bzrlib.repofmt.groupcompress_repo',
1681
'RepositoryFormat2aSubtree',
1685
1275
class InterRepository(InterObject):
1686
1276
"""This class represents operations taking place between two repositories.
1688
1278
Its instances have methods like copy_content and fetch, and contain
1689
references to the source and target repositories these operations can be
1279
references to the source and target repositories these operations can be
1690
1280
carried out on.
1692
1282
Often we will provide convenience methods on 'repository' which carry out
1697
1287
_optimisers = []
1698
1288
"""The available optimised InterRepository types."""
1701
def copy_content(self, revision_id=None):
1702
"""Make a complete copy of the content in self into destination.
1704
This is a destructive operation! Do not use it on existing
1707
:param revision_id: Only copy the content needed to construct
1708
revision_id and its parents.
1711
self.target.set_make_working_trees(self.source.make_working_trees())
1712
except NotImplementedError:
1714
self.target.fetch(self.source, revision_id=revision_id)
1717
def fetch(self, revision_id=None, find_ghosts=False,
1290
def copy_content(self, revision_id=None, basis=None):
1291
raise NotImplementedError(self.copy_content)
1293
def fetch(self, revision_id=None, pb=None):
1719
1294
"""Fetch the content required to construct revision_id.
1721
1296
The content is copied from self.source to self.target.
1723
1298
:param revision_id: if None all content is copied, if NULL_REVISION no
1724
1299
content is copied.
1300
:param pb: optional progress bar to use for progress reports. If not
1301
provided a default one will be created.
1303
Returns the copied revision count and the failed revisions in a tuple:
1727
1306
raise NotImplementedError(self.fetch)
1729
1308
@needs_read_lock
1730
def search_missing_revision_ids(self,
1731
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
1732
find_ghosts=True, revision_ids=None, if_present_ids=None,
1309
def missing_revision_ids(self, revision_id=None):
1734
1310
"""Return the revision ids that source has that target does not.
1312
These are returned in topological order.
1736
1314
:param revision_id: only return revision ids included by this
1738
:param revision_ids: return revision ids included by these
1739
revision_ids. NoSuchRevision will be raised if any of these
1740
revisions are not present.
1741
:param if_present_ids: like revision_ids, but will not cause
1742
NoSuchRevision if any of these are absent, instead they will simply
1743
not be in the result. This is useful for e.g. finding revisions
1744
to fetch for tags, which may reference absent revisions.
1745
:param find_ghosts: If True find missing revisions in deep history
1746
rather than just finding the surface difference.
1747
:param limit: Maximum number of revisions to return, topologically
1749
:return: A bzrlib.graph.SearchResult.
1751
raise NotImplementedError(self.search_missing_revision_ids)
1754
def _same_model(source, target):
1755
"""True if source and target have the same data representation.
1757
Note: this is always called on the base class; overriding it in a
1758
subclass will have no effect.
1761
InterRepository._assert_same_model(source, target)
1317
# generic, possibly worst case, slow code path.
1318
target_ids = set(self.target.all_revision_ids())
1319
if revision_id is not None:
1320
# TODO: jam 20070210 InterRepository is internal enough that it
1321
# should assume revision_ids are already utf-8
1322
revision_id = osutils.safe_revision_id(revision_id)
1323
source_ids = self.source.get_ancestry(revision_id)
1324
assert source_ids[0] is None
1327
source_ids = self.source.all_revision_ids()
1328
result_set = set(source_ids).difference(target_ids)
1329
# this may look like a no-op: its not. It preserves the ordering
1330
# other_ids had while only returning the members from other_ids
1331
# that we've decided we need.
1332
return [rev_id for rev_id in source_ids if rev_id in result_set]
1335
class InterSameDataRepository(InterRepository):
1336
"""Code for converting between repositories that represent the same data.
1338
Data format and model must match for this to work.
1342
def _get_repo_format_to_test(self):
1343
"""Repository format for testing with."""
1344
return RepositoryFormat.get_default_format()
1347
def is_compatible(source, target):
1348
if not isinstance(source, Repository):
1350
if not isinstance(target, Repository):
1352
if source._format.rich_root_data != target._format.rich_root_data:
1354
if source._serializer != target._serializer:
1360
def copy_content(self, revision_id=None, basis=None):
1361
"""Make a complete copy of the content in self into destination.
1363
This is a destructive operation! Do not use it on existing
1366
:param revision_id: Only copy the content needed to construct
1367
revision_id and its parents.
1368
:param basis: Copy the needed data preferentially from basis.
1371
self.target.set_make_working_trees(self.source.make_working_trees())
1372
except NotImplementedError:
1374
# TODO: jam 20070210 This is fairly internal, so we should probably
1375
# just assert that revision_id is not unicode.
1376
revision_id = osutils.safe_revision_id(revision_id)
1377
# grab the basis available data
1378
if basis is not None:
1379
self.target.fetch(basis, revision_id=revision_id)
1380
# but don't bother fetching if we have the needed data now.
1381
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1382
self.target.has_revision(revision_id)):
1384
self.target.fetch(self.source, revision_id=revision_id)
1387
def fetch(self, revision_id=None, pb=None):
1388
"""See InterRepository.fetch()."""
1389
from bzrlib.fetch import GenericRepoFetcher
1390
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1391
self.source, self.source._format, self.target,
1392
self.target._format)
1393
# TODO: jam 20070210 This should be an assert, not a translate
1394
revision_id = osutils.safe_revision_id(revision_id)
1395
f = GenericRepoFetcher(to_repository=self.target,
1396
from_repository=self.source,
1397
last_revision=revision_id,
1399
return f.count_copied, f.failed_revisions
1402
class InterWeaveRepo(InterSameDataRepository):
1403
"""Optimised code paths between Weave based repositories."""
1406
def _get_repo_format_to_test(self):
1407
from bzrlib.repofmt import weaverepo
1408
return weaverepo.RepositoryFormat7()
1411
def is_compatible(source, target):
1412
"""Be compatible with known Weave formats.
1414
We don't test for the stores being of specific types because that
1415
could lead to confusing results, and there is no need to be
1418
from bzrlib.repofmt.weaverepo import (
1424
return (isinstance(source._format, (RepositoryFormat5,
1426
RepositoryFormat7)) and
1427
isinstance(target._format, (RepositoryFormat5,
1429
RepositoryFormat7)))
1430
except AttributeError:
1434
def copy_content(self, revision_id=None, basis=None):
1435
"""See InterRepository.copy_content()."""
1436
# weave specific optimised path:
1437
# TODO: jam 20070210 Internal, should be an assert, not translate
1438
revision_id = osutils.safe_revision_id(revision_id)
1439
if basis is not None:
1440
# copy the basis in, then fetch remaining data.
1441
basis.copy_content_into(self.target, revision_id)
1442
# the basis copy_content_into could miss-set this.
1444
self.target.set_make_working_trees(self.source.make_working_trees())
1445
except NotImplementedError:
1447
self.target.fetch(self.source, revision_id=revision_id)
1450
self.target.set_make_working_trees(self.source.make_working_trees())
1451
except NotImplementedError:
1453
# FIXME do not peek!
1454
if self.source.control_files._transport.listable():
1455
pb = ui.ui_factory.nested_progress_bar()
1457
self.target.weave_store.copy_all_ids(
1458
self.source.weave_store,
1460
from_transaction=self.source.get_transaction(),
1461
to_transaction=self.target.get_transaction())
1462
pb.update('copying inventory', 0, 1)
1463
self.target.control_weaves.copy_multi(
1464
self.source.control_weaves, ['inventory'],
1465
from_transaction=self.source.get_transaction(),
1466
to_transaction=self.target.get_transaction())
1467
self.target._revision_store.text_store.copy_all_ids(
1468
self.source._revision_store.text_store,
1473
self.target.fetch(self.source, revision_id=revision_id)
1476
def fetch(self, revision_id=None, pb=None):
1477
"""See InterRepository.fetch()."""
1478
from bzrlib.fetch import GenericRepoFetcher
1479
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1480
self.source, self.source._format, self.target, self.target._format)
1481
# TODO: jam 20070210 This should be an assert, not a translate
1482
revision_id = osutils.safe_revision_id(revision_id)
1483
f = GenericRepoFetcher(to_repository=self.target,
1484
from_repository=self.source,
1485
last_revision=revision_id,
1487
return f.count_copied, f.failed_revisions
1490
def missing_revision_ids(self, revision_id=None):
1491
"""See InterRepository.missing_revision_ids()."""
1492
# we want all revisions to satisfy revision_id in source.
1493
# but we don't want to stat every file here and there.
1494
# we want then, all revisions other needs to satisfy revision_id
1495
# checked, but not those that we have locally.
1496
# so the first thing is to get a subset of the revisions to
1497
# satisfy revision_id in source, and then eliminate those that
1498
# we do already have.
1499
# this is slow on high latency connection to self, but as as this
1500
# disk format scales terribly for push anyway due to rewriting
1501
# inventory.weave, this is considered acceptable.
1503
if revision_id is not None:
1504
source_ids = self.source.get_ancestry(revision_id)
1505
assert source_ids[0] is None
1508
source_ids = self.source._all_possible_ids()
1509
source_ids_set = set(source_ids)
1510
# source_ids is the worst possible case we may need to pull.
1511
# now we want to filter source_ids against what we actually
1512
# have in target, but don't try to check for existence where we know
1513
# we do not have a revision as that would be pointless.
1514
target_ids = set(self.target._all_possible_ids())
1515
possibly_present_revisions = target_ids.intersection(source_ids_set)
1516
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1517
required_revisions = source_ids_set.difference(actually_present_revisions)
1518
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1519
if revision_id is not None:
1520
# we used get_ancestry to determine source_ids then we are assured all
1521
# revisions referenced are present as they are installed in topological order.
1522
# and the tip revision was validated by get_ancestry.
1523
return required_topo_revisions
1525
# if we just grabbed the possibly available ids, then
1526
# we only have an estimate of whats available and need to validate
1527
# that against the revision records.
1528
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1531
class InterKnitRepo(InterSameDataRepository):
1532
"""Optimised code paths between Knit based repositories."""
1535
def _get_repo_format_to_test(self):
1536
from bzrlib.repofmt import knitrepo
1537
return knitrepo.RepositoryFormatKnit1()
1540
def is_compatible(source, target):
1541
"""Be compatible with known Knit formats.
1543
We don't test for the stores being of specific types because that
1544
could lead to confusing results, and there is no need to be
1547
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1549
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1550
isinstance(target._format, (RepositoryFormatKnit1)))
1551
except AttributeError:
1555
def fetch(self, revision_id=None, pb=None):
1556
"""See InterRepository.fetch()."""
1557
from bzrlib.fetch import KnitRepoFetcher
1558
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1559
self.source, self.source._format, self.target, self.target._format)
1560
# TODO: jam 20070210 This should be an assert, not a translate
1561
revision_id = osutils.safe_revision_id(revision_id)
1562
f = KnitRepoFetcher(to_repository=self.target,
1563
from_repository=self.source,
1564
last_revision=revision_id,
1566
return f.count_copied, f.failed_revisions
1569
def missing_revision_ids(self, revision_id=None):
1570
"""See InterRepository.missing_revision_ids()."""
1571
if revision_id is not None:
1572
source_ids = self.source.get_ancestry(revision_id)
1573
assert source_ids[0] is None
1576
source_ids = self.source._all_possible_ids()
1577
source_ids_set = set(source_ids)
1578
# source_ids is the worst possible case we may need to pull.
1579
# now we want to filter source_ids against what we actually
1580
# have in target, but don't try to check for existence where we know
1581
# we do not have a revision as that would be pointless.
1582
target_ids = set(self.target._all_possible_ids())
1583
possibly_present_revisions = target_ids.intersection(source_ids_set)
1584
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1585
required_revisions = source_ids_set.difference(actually_present_revisions)
1586
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1587
if revision_id is not None:
1588
# we used get_ancestry to determine source_ids then we are assured all
1589
# revisions referenced are present as they are installed in topological order.
1590
# and the tip revision was validated by get_ancestry.
1591
return required_topo_revisions
1593
# if we just grabbed the possibly available ids, then
1594
# we only have an estimate of whats available and need to validate
1595
# that against the revision records.
1596
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1599
class InterModel1and2(InterRepository):
1602
def _get_repo_format_to_test(self):
1606
def is_compatible(source, target):
1607
if not isinstance(source, Repository):
1609
if not isinstance(target, Repository):
1611
if not source._format.rich_root_data and target._format.rich_root_data:
1763
except errors.IncompatibleRepositories, e:
1767
def _assert_same_model(source, target):
1768
"""Raise an exception if two repositories do not use the same model.
1617
def fetch(self, revision_id=None, pb=None):
1618
"""See InterRepository.fetch()."""
1619
from bzrlib.fetch import Model1toKnit2Fetcher
1620
# TODO: jam 20070210 This should be an assert, not a translate
1621
revision_id = osutils.safe_revision_id(revision_id)
1622
f = Model1toKnit2Fetcher(to_repository=self.target,
1623
from_repository=self.source,
1624
last_revision=revision_id,
1626
return f.count_copied, f.failed_revisions
1629
def copy_content(self, revision_id=None, basis=None):
1630
"""Make a complete copy of the content in self into destination.
1632
This is a destructive operation! Do not use it on existing
1635
:param revision_id: Only copy the content needed to construct
1636
revision_id and its parents.
1637
:param basis: Copy the needed data preferentially from basis.
1770
if source.supports_rich_root() != target.supports_rich_root():
1771
raise errors.IncompatibleRepositories(source, target,
1772
"different rich-root support")
1773
if source._serializer != target._serializer:
1774
raise errors.IncompatibleRepositories(source, target,
1775
"different serializers")
1640
self.target.set_make_working_trees(self.source.make_working_trees())
1641
except NotImplementedError:
1643
# TODO: jam 20070210 Internal, assert, don't translate
1644
revision_id = osutils.safe_revision_id(revision_id)
1645
# grab the basis available data
1646
if basis is not None:
1647
self.target.fetch(basis, revision_id=revision_id)
1648
# but don't bother fetching if we have the needed data now.
1649
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1650
self.target.has_revision(revision_id)):
1652
self.target.fetch(self.source, revision_id=revision_id)
1655
class InterKnit1and2(InterKnitRepo):
1658
def _get_repo_format_to_test(self):
1662
def is_compatible(source, target):
1663
"""Be compatible with Knit1 source and Knit2 target"""
1664
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit2
1666
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1667
RepositoryFormatKnit2
1668
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1669
isinstance(target._format, (RepositoryFormatKnit2)))
1670
except AttributeError:
1674
def fetch(self, revision_id=None, pb=None):
1675
"""See InterRepository.fetch()."""
1676
from bzrlib.fetch import Knit1to2Fetcher
1677
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1678
self.source, self.source._format, self.target,
1679
self.target._format)
1680
# TODO: jam 20070210 This should be an assert, not a translate
1681
revision_id = osutils.safe_revision_id(revision_id)
1682
f = Knit1to2Fetcher(to_repository=self.target,
1683
from_repository=self.source,
1684
last_revision=revision_id,
1686
return f.count_copied, f.failed_revisions
1689
InterRepository.register_optimiser(InterSameDataRepository)
1690
InterRepository.register_optimiser(InterWeaveRepo)
1691
InterRepository.register_optimiser(InterKnitRepo)
1692
InterRepository.register_optimiser(InterModel1and2)
1693
InterRepository.register_optimiser(InterKnit1and2)
1696
class RepositoryTestProviderAdapter(object):
1697
"""A tool to generate a suite testing multiple repository formats at once.
1699
This is done by copying the test once for each transport and injecting
1700
the transport_server, transport_readonly_server, and bzrdir_format and
1701
repository_format classes into each copy. Each copy is also given a new id()
1702
to make it easy to identify.
1705
def __init__(self, transport_server, transport_readonly_server, formats):
1706
self._transport_server = transport_server
1707
self._transport_readonly_server = transport_readonly_server
1708
self._formats = formats
1710
def adapt(self, test):
1711
result = unittest.TestSuite()
1712
for repository_format, bzrdir_format in self._formats:
1713
from copy import deepcopy
1714
new_test = deepcopy(test)
1715
new_test.transport_server = self._transport_server
1716
new_test.transport_readonly_server = self._transport_readonly_server
1717
new_test.bzrdir_format = bzrdir_format
1718
new_test.repository_format = repository_format
1719
def make_new_test_id():
1720
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1721
return lambda: new_id
1722
new_test.id = make_new_test_id()
1723
result.addTest(new_test)
1727
class InterRepositoryTestProviderAdapter(object):
1728
"""A tool to generate a suite testing multiple inter repository formats.
1730
This is done by copying the test once for each interrepo provider and injecting
1731
the transport_server, transport_readonly_server, repository_format and
1732
repository_to_format classes into each copy.
1733
Each copy is also given a new id() to make it easy to identify.
1736
def __init__(self, transport_server, transport_readonly_server, formats):
1737
self._transport_server = transport_server
1738
self._transport_readonly_server = transport_readonly_server
1739
self._formats = formats
1741
def adapt(self, test):
1742
result = unittest.TestSuite()
1743
for interrepo_class, repository_format, repository_format_to in self._formats:
1744
from copy import deepcopy
1745
new_test = deepcopy(test)
1746
new_test.transport_server = self._transport_server
1747
new_test.transport_readonly_server = self._transport_readonly_server
1748
new_test.interrepo_class = interrepo_class
1749
new_test.repository_format = repository_format
1750
new_test.repository_format_to = repository_format_to
1751
def make_new_test_id():
1752
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1753
return lambda: new_id
1754
new_test.id = make_new_test_id()
1755
result.addTest(new_test)
1759
def default_test_list():
1760
"""Generate the default list of interrepo permutations to test."""
1761
from bzrlib.repofmt import knitrepo, weaverepo
1763
# test the default InterRepository between format 6 and the current
1765
# XXX: robertc 20060220 reinstate this when there are two supported
1766
# formats which do not have an optimal code path between them.
1767
#result.append((InterRepository,
1768
# RepositoryFormat6(),
1769
# RepositoryFormatKnit1()))
1770
for optimiser_class in InterRepository._optimisers:
1771
format_to_test = optimiser_class._get_repo_format_to_test()
1772
if format_to_test is not None:
1773
result.append((optimiser_class,
1774
format_to_test, format_to_test))
1775
# if there are specific combinations we want to use, we can add them
1777
result.append((InterModel1and2,
1778
weaverepo.RepositoryFormat5(),
1779
knitrepo.RepositoryFormatKnit2()))
1780
result.append((InterKnit1and2,
1781
knitrepo.RepositoryFormatKnit1(),
1782
knitrepo.RepositoryFormatKnit2()))
1778
1786
class CopyConverter(object):
1779
1787
"""A repository conversion tool which just performs a copy of the content.
1781
1789
This is slow but quite reliable.
1787
1795
:param target_format: The format the resulting repository should be.
1789
1797
self.target_format = target_format
1791
1799
def convert(self, repo, pb):
1792
1800
"""Perform the conversion of to_convert, giving feedback via pb.
1794
1802
:param to_convert: The disk object to convert.
1795
1803
:param pb: a progress bar to use for progress information.
1797
pb = ui.ui_factory.nested_progress_bar()
1800
1808
# this is only useful with metadir layouts - separated repo content.
1801
1809
# trigger an assertion if not such
1802
1810
repo._format.get_format_string()
1803
1811
self.repo_dir = repo.bzrdir
1804
pb.update(gettext('Moving repository to repository.backup'))
1812
self.step('Moving repository to repository.backup')
1805
1813
self.repo_dir.transport.move('repository', 'repository.backup')
1806
1814
backup_transport = self.repo_dir.transport.clone('repository.backup')
1807
1815
repo._format.check_conversion_target(self.target_format)
1808
1816
self.source_repo = repo._format.open(self.repo_dir,
1810
1818
_override_transport=backup_transport)
1811
pb.update(gettext('Creating new repository'))
1819
self.step('Creating new repository')
1812
1820
converted = self.target_format.initialize(self.repo_dir,
1813
1821
self.source_repo.is_shared())
1814
1822
converted.lock_write()
1816
pb.update(gettext('Copying content'))
1824
self.step('Copying content into repository.')
1817
1825
self.source_repo.copy_content_into(converted)
1819
1827
converted.unlock()
1820
pb.update(gettext('Deleting old repository content'))
1828
self.step('Deleting old repository content.')
1821
1829
self.repo_dir.transport.delete_tree('repository.backup')
1822
ui.ui_factory.note(gettext('repository converted'))
1826
def _strip_NULL_ghosts(revision_graph):
1827
"""Also don't use this. more compatibility code for unmigrated clients."""
1828
# Filter ghosts, and null:
1829
if _mod_revision.NULL_REVISION in revision_graph:
1830
del revision_graph[_mod_revision.NULL_REVISION]
1831
for key, parents in revision_graph.items():
1832
revision_graph[key] = tuple(parent for parent in parents if parent
1834
return revision_graph
1837
def _iter_for_revno(repo, partial_history_cache, stop_index=None,
1838
stop_revision=None):
1839
"""Extend the partial history to include a given index
1841
If a stop_index is supplied, stop when that index has been reached.
1842
If a stop_revision is supplied, stop when that revision is
1843
encountered. Otherwise, stop when the beginning of history is
1846
:param stop_index: The index which should be present. When it is
1847
present, history extension will stop.
1848
:param stop_revision: The revision id which should be present. When
1849
it is encountered, history extension will stop.
1851
start_revision = partial_history_cache[-1]
1852
graph = repo.get_graph()
1853
iterator = graph.iter_lefthand_ancestry(start_revision,
1854
(_mod_revision.NULL_REVISION,))
1830
self.pb.note('repository converted')
1832
def step(self, message):
1833
"""Update the pb by a step."""
1835
self.pb.update(message, self.count, self.total)
1838
class CommitBuilder(object):
1839
"""Provides an interface to build up a commit.
1841
This allows describing a tree to be committed without needing to
1842
know the internals of the format of the repository.
1845
record_root_entry = False
1846
def __init__(self, repository, parents, config, timestamp=None,
1847
timezone=None, committer=None, revprops=None,
1849
"""Initiate a CommitBuilder.
1851
:param repository: Repository to commit to.
1852
:param parents: Revision ids of the parents of the new revision.
1853
:param config: Configuration to use.
1854
:param timestamp: Optional timestamp recorded for commit.
1855
:param timezone: Optional timezone for timestamp.
1856
:param committer: Optional committer to set for commit.
1857
:param revprops: Optional dictionary of revision properties.
1858
:param revision_id: Optional revision id.
1860
self._config = config
1862
if committer is None:
1863
self._committer = self._config.username()
1865
assert isinstance(committer, basestring), type(committer)
1866
self._committer = committer
1868
self.new_inventory = Inventory(None)
1869
self._new_revision_id = osutils.safe_revision_id(revision_id)
1870
self.parents = parents
1871
self.repository = repository
1874
if revprops is not None:
1875
self._revprops.update(revprops)
1877
if timestamp is None:
1878
timestamp = time.time()
1879
# Restrict resolution to 1ms
1880
self._timestamp = round(timestamp, 3)
1882
if timezone is None:
1883
self._timezone = osutils.local_time_offset()
1885
self._timezone = int(timezone)
1887
self._generate_revision_if_needed()
1889
def commit(self, message):
1890
"""Make the actual commit.
1892
:return: The revision id of the recorded revision.
1894
rev = _mod_revision.Revision(
1895
timestamp=self._timestamp,
1896
timezone=self._timezone,
1897
committer=self._committer,
1899
inventory_sha1=self.inv_sha1,
1900
revision_id=self._new_revision_id,
1901
properties=self._revprops)
1902
rev.parent_ids = self.parents
1903
self.repository.add_revision(self._new_revision_id, rev,
1904
self.new_inventory, self._config)
1905
return self._new_revision_id
1907
def revision_tree(self):
1908
"""Return the tree that was just committed.
1910
After calling commit() this can be called to get a RevisionTree
1911
representing the newly committed tree. This is preferred to
1912
calling Repository.revision_tree() because that may require
1913
deserializing the inventory, while we already have a copy in
1916
return RevisionTree(self.repository, self.new_inventory,
1917
self._new_revision_id)
1919
def finish_inventory(self):
1920
"""Tell the builder that the inventory is finished."""
1921
if self.new_inventory.root is None:
1922
symbol_versioning.warn('Root entry should be supplied to'
1923
' record_entry_contents, as of bzr 0.10.',
1924
DeprecationWarning, stacklevel=2)
1925
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1926
self.new_inventory.revision_id = self._new_revision_id
1927
self.inv_sha1 = self.repository.add_inventory(
1928
self._new_revision_id,
1933
def _gen_revision_id(self):
1934
"""Return new revision-id."""
1935
return generate_ids.gen_revision_id(self._config.username(),
1938
def _generate_revision_if_needed(self):
1939
"""Create a revision id if None was supplied.
1941
If the repository can not support user-specified revision ids
1942
they should override this function and raise CannotSetRevisionId
1943
if _new_revision_id is not None.
1945
:raises: CannotSetRevisionId
1947
if self._new_revision_id is None:
1948
self._new_revision_id = self._gen_revision_id()
1950
def record_entry_contents(self, ie, parent_invs, path, tree):
1951
"""Record the content of ie from tree into the commit if needed.
1953
Side effect: sets ie.revision when unchanged
1955
:param ie: An inventory entry present in the commit.
1956
:param parent_invs: The inventories of the parent revisions of the
1958
:param path: The path the entry is at in the tree.
1959
:param tree: The tree which contains this entry and should be used to
1962
if self.new_inventory.root is None and ie.parent_id is not None:
1963
symbol_versioning.warn('Root entry should be supplied to'
1964
' record_entry_contents, as of bzr 0.10.',
1965
DeprecationWarning, stacklevel=2)
1966
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1968
self.new_inventory.add(ie)
1970
# ie.revision is always None if the InventoryEntry is considered
1971
# for committing. ie.snapshot will record the correct revision
1972
# which may be the sole parent if it is untouched.
1973
if ie.revision is not None:
1976
# In this revision format, root entries have no knit or weave
1977
if ie is self.new_inventory.root:
1978
# When serializing out to disk and back in
1979
# root.revision is always _new_revision_id
1980
ie.revision = self._new_revision_id
1982
previous_entries = ie.find_previous_heads(
1984
self.repository.weave_store,
1985
self.repository.get_transaction())
1986
# we are creating a new revision for ie in the history store
1988
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1990
def modified_directory(self, file_id, file_parents):
1991
"""Record the presence of a symbolic link.
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_reference(self, file_id, file_parents):
1999
"""Record the modification of a reference.
2001
:param file_id: The file_id of the link to record.
2002
:param file_parents: The per-file parent revision ids.
2004
self._add_text_to_weave(file_id, [], file_parents.keys())
2006
def modified_file_text(self, file_id, file_parents,
2007
get_content_byte_lines, text_sha1=None,
2009
"""Record the text of file file_id
2011
:param file_id: The file_id of the file to record the text of.
2012
:param file_parents: The per-file parent revision ids.
2013
:param get_content_byte_lines: A callable which will return the byte
2015
:param text_sha1: Optional SHA1 of the file contents.
2016
:param text_size: Optional size of the file contents.
2018
# mutter('storing text of file {%s} in revision {%s} into %r',
2019
# file_id, self._new_revision_id, self.repository.weave_store)
2020
# special case to avoid diffing on renames or
2022
if (len(file_parents) == 1
2023
and text_sha1 == file_parents.values()[0].text_sha1
2024
and text_size == file_parents.values()[0].text_size):
2025
previous_ie = file_parents.values()[0]
2026
versionedfile = self.repository.weave_store.get_weave(file_id,
2027
self.repository.get_transaction())
2028
versionedfile.clone_text(self._new_revision_id,
2029
previous_ie.revision, file_parents.keys())
2030
return text_sha1, text_size
2032
new_lines = get_content_byte_lines()
2033
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2034
# should return the SHA1 and size
2035
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2036
return osutils.sha_strings(new_lines), \
2037
sum(map(len, new_lines))
2039
def modified_link(self, file_id, file_parents, link_target):
2040
"""Record the presence of a symbolic link.
2042
:param file_id: The file_id of the link to record.
2043
:param file_parents: The per-file parent revision ids.
2044
:param link_target: Target location of this link.
2046
self._add_text_to_weave(file_id, [], file_parents.keys())
2048
def _add_text_to_weave(self, file_id, new_lines, parents):
2049
versionedfile = self.repository.weave_store.get_weave_or_empty(
2050
file_id, self.repository.get_transaction())
2051
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2052
versionedfile.clear_cache()
2055
class _CommitBuilder(CommitBuilder):
2056
"""Temporary class so old CommitBuilders are detected properly
2058
Note: CommitBuilder works whether or not root entry is recorded.
2061
record_root_entry = True
2064
class RootCommitBuilder(CommitBuilder):
2065
"""This commitbuilder actually records the root id"""
2067
record_root_entry = True
2069
def record_entry_contents(self, ie, parent_invs, path, tree):
2070
"""Record the content of ie from tree into the commit if needed.
2072
Side effect: sets ie.revision when unchanged
2074
:param ie: An inventory entry present in the commit.
2075
:param parent_invs: The inventories of the parent revisions of the
2077
:param path: The path the entry is at in the tree.
2078
:param tree: The tree which contains this entry and should be used to
2081
assert self.new_inventory.root is not None or ie.parent_id is None
2082
self.new_inventory.add(ie)
2084
# ie.revision is always None if the InventoryEntry is considered
2085
# for committing. ie.snapshot will record the correct revision
2086
# which may be the sole parent if it is untouched.
2087
if ie.revision is not None:
2090
previous_entries = ie.find_previous_heads(
2092
self.repository.weave_store,
2093
self.repository.get_transaction())
2094
# we are creating a new revision for ie in the history store
2096
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2108
def _unescaper(match, _map=_unescape_map):
2109
code = match.group(1)
1856
# skip the last revision in the list
1859
if (stop_index is not None and
1860
len(partial_history_cache) > stop_index):
1862
if partial_history_cache[-1] == stop_revision:
1864
revision_id = iterator.next()
1865
partial_history_cache.append(revision_id)
1866
except StopIteration:
1871
class _LazyListJoin(object):
1872
"""An iterable yielding the contents of many lists as one list.
1874
Each iterator made from this will reflect the current contents of the lists
1875
at the time the iterator is made.
1877
This is used by Repository's _make_parents_provider implementation so that
1880
pp = repo._make_parents_provider() # uses a list of fallback repos
1881
pp.add_fallback_repository(other_repo) # appends to that list
1882
result = pp.get_parent_map(...)
1883
# The result will include revs from other_repo
1886
def __init__(self, *list_parts):
1887
self.list_parts = list_parts
1891
for list_part in self.list_parts:
1892
full_list.extend(list_part)
1893
return iter(full_list)
1896
return "%s.%s(%s)" % (self.__module__, self.__class__.__name__,
2113
if not code.startswith('#'):
2115
return unichr(int(code[1:])).encode('utf8')
2121
def _unescape_xml(data):
2122
"""Unescape predefined XML entities in a string of data."""
2124
if _unescape_re is None:
2125
_unescape_re = re.compile('\&([^;]*);')
2126
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
