13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
from .lazy_import import lazy_import
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
18
20
lazy_import(globals(), """
27
37
revision as _mod_revision,
30
from breezy.i18n import gettext
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
38
from .decorators import only_raises
39
from .inter import InterObject
40
from .lock import _RelockDebugMixin, LogicalLockResult
42
log_exception_quietly, note, mutter, mutter_callsite, warning)
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
56
from bzrlib.trace import mutter, note, warning
45
59
# Old formats display a warning, but only once
46
60
_deprecation_warning_done = False
49
class IsInWriteGroupError(errors.InternalBzrError):
51
_fmt = "May not refresh_data of repo %(repo)s while in a write group."
53
def __init__(self, repo):
54
errors.InternalBzrError.__init__(self, repo=repo)
57
class CannotSetRevisionId(errors.BzrError):
59
_fmt = "Repository format does not support setting revision ids."
62
class FetchResult(object):
63
"""Result of a fetch operation.
65
:ivar revidmap: For lossy fetches, map from source revid to target revid.
66
:ivar total_fetched: Number of revisions fetched
69
def __init__(self, total_fetched=None, revidmap=None):
70
self.total_fetched = total_fetched
71
self.revidmap = revidmap
74
class CommitBuilder(object):
75
"""Provides an interface to build up a commit.
77
This allows describing a tree to be committed without needing to
78
know the internals of the format of the repository.
81
# all clients should supply tree roots.
82
record_root_entry = True
83
# whether this commit builder will automatically update the branch that is
85
updates_branch = False
87
def __init__(self, repository, parents, config_stack, timestamp=None,
88
timezone=None, committer=None, revprops=None,
89
revision_id=None, lossy=False):
90
"""Initiate a CommitBuilder.
92
:param repository: Repository to commit to.
93
:param parents: Revision ids of the parents of the new revision.
94
:param timestamp: Optional timestamp recorded for commit.
95
:param timezone: Optional timezone for timestamp.
96
:param committer: Optional committer to set for commit.
97
:param revprops: Optional dictionary of revision properties.
98
:param revision_id: Optional revision id.
99
:param lossy: Whether to discard data that can not be natively
100
represented, when pushing to a foreign VCS
102
self._config_stack = config_stack
105
if committer is None:
106
self._committer = self._config_stack.get('email')
107
elif not isinstance(committer, str):
108
self._committer = committer.decode() # throw if non-ascii
110
self._committer = committer
112
self.parents = parents
113
self.repository = repository
116
if revprops is not None:
117
self._validate_revprops(revprops)
118
self._revprops.update(revprops)
120
if timestamp is None:
121
timestamp = time.time()
122
# Restrict resolution to 1ms
123
self._timestamp = round(timestamp, 3)
126
self._timezone = osutils.local_time_offset()
128
self._timezone = int(timezone)
130
self._generate_revision_if_needed(revision_id)
132
def any_changes(self):
133
"""Return True if any entries were changed.
135
This includes merge-only changes. It is the core for the --unchanged
138
:return: True if any changes have occured.
140
raise NotImplementedError(self.any_changes)
142
def _validate_unicode_text(self, text, context):
143
"""Verify things like commit messages don't have bogus characters."""
144
# TODO(jelmer): Make this repository-format specific
146
raise ValueError('Invalid value for %s: %r' % (context, text))
148
def _validate_revprops(self, revprops):
149
for key, value in revprops.items():
150
# We know that the XML serializers do not round trip '\r'
151
# correctly, so refuse to accept them
152
if not isinstance(value, str):
153
raise ValueError('revision property (%s) is not a valid'
154
' (unicode) string: %r' % (key, value))
155
# TODO(jelmer): Make this repository-format specific
156
self._validate_unicode_text(value,
157
'revision property (%s)' % (key,))
159
def commit(self, message):
160
"""Make the actual commit.
162
:return: The revision id of the recorded revision.
164
raise NotImplementedError(self.commit)
167
"""Abort the commit that is being built.
169
raise NotImplementedError(self.abort)
171
def revision_tree(self):
172
"""Return the tree that was just committed.
174
After calling commit() this can be called to get a
175
RevisionTree representing the newly committed tree. This is
176
preferred to calling Repository.revision_tree() because that may
177
require deserializing the inventory, while we already have a copy in
180
raise NotImplementedError(self.revision_tree)
182
def finish_inventory(self):
183
"""Tell the builder that the inventory is finished.
185
:return: The inventory id in the repository, which can be used with
186
repository.get_inventory.
188
raise NotImplementedError(self.finish_inventory)
190
def _generate_revision_if_needed(self, revision_id):
191
"""Create a revision id if None was supplied.
193
If the repository can not support user-specified revision ids
194
they should override this function and raise CannotSetRevisionId
195
if _new_revision_id is not None.
197
:raises: CannotSetRevisionId
199
if not self.repository._format.supports_setting_revision_ids:
200
if revision_id is not None:
201
raise CannotSetRevisionId()
203
if revision_id is None:
204
self._new_revision_id = self._gen_revision_id()
205
self.random_revid = True
207
self._new_revision_id = revision_id
208
self.random_revid = False
210
def record_iter_changes(self, tree, basis_revision_id, iter_changes):
211
"""Record a new tree via iter_changes.
213
:param tree: The tree to obtain text contents from for changed objects.
214
:param basis_revision_id: The revision id of the tree the iter_changes
215
has been generated against. Currently assumed to be the same
216
as self.parents[0] - if it is not, errors may occur.
217
:param iter_changes: An iter_changes iterator with the changes to apply
218
to basis_revision_id. The iterator must not include any items with
219
a current kind of None - missing items must be either filtered out
220
or errored-on beefore record_iter_changes sees the item.
221
:return: A generator of (relpath, fs_hash) tuples for use with
224
raise NotImplementedError(self.record_iter_changes)
227
class RepositoryWriteLockResult(LogicalLockResult):
228
"""The result of write locking a repository.
230
:ivar repository_token: The token obtained from the underlying lock, or
232
:ivar unlock: A callable which will unlock the lock.
235
def __init__(self, unlock, repository_token):
236
LogicalLockResult.__init__(self, unlock)
237
self.repository_token = repository_token
240
return "RepositoryWriteLockResult(%s, %s)" % (self.repository_token,
244
class WriteGroup(object):
245
"""Context manager that manages a write group.
247
Raising an exception will result in the write group being aborted.
250
def __init__(self, repository, suppress_errors=False):
251
self.repository = repository
252
self._suppress_errors = suppress_errors
255
self.repository.start_write_group()
258
def __exit__(self, exc_type, exc_val, exc_tb):
260
self.repository.abort_write_group(self._suppress_errors)
263
self.repository.commit_write_group()
266
63
######################################################################
270
class Repository(controldir.ControlComponent, _RelockDebugMixin):
66
class Repository(object):
271
67
"""Repository holding history for one or more branches.
273
69
The repository holds and retrieves historical information including
274
70
revisions and file history. It's normally accessed only by the Branch,
275
71
which views a particular line of development through that history.
277
See VersionedFileRepository in breezy.vf_repository for the
278
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
281
# Does this repository implementation support random access to
282
# items in the tree, or just bulk fetching/pushing of data?
283
supports_random_access = True
285
def abort_write_group(self, suppress_errors=False):
286
"""Commit the contents accrued within the current write group.
288
:param suppress_errors: if true, abort_write_group will catch and log
289
unexpected errors that happen during the abort, rather than
290
allowing them to propagate. Defaults to False.
292
:seealso: start_write_group.
294
if self._write_group is not self.get_transaction():
295
# has an unlock or relock occured ?
298
'(suppressed) mismatched lock context and write group. %r, %r',
299
self._write_group, self.get_transaction())
301
raise errors.BzrError(
302
'mismatched lock context and write group. %r, %r' %
303
(self._write_group, self.get_transaction()))
305
self._abort_write_group()
306
except Exception as exc:
307
self._write_group = None
308
if not suppress_errors:
310
mutter('abort_write_group failed')
311
log_exception_quietly()
312
note(gettext('brz: ERROR (ignored): %s'), exc)
313
self._write_group = None
315
def _abort_write_group(self):
316
"""Template method for per-repository write group cleanup.
318
This is called during abort before the write group is considered to be
319
finished and should cleanup any internal state accrued during the write
320
group. There is no requirement that data handed to the repository be
321
*not* made available - this is not a rollback - but neither should any
322
attempt be made to ensure that data added is fully commited. Abort is
323
invoked when an error has occured so futher disk or network operations
324
may not be possible or may error and if possible should not be
328
def add_fallback_repository(self, repository):
329
"""Add a repository to use for looking up data not held locally.
331
:param repository: A repository.
333
raise NotImplementedError(self.add_fallback_repository)
335
def _check_fallback_repository(self, repository):
336
"""Check that this repository can fallback to repository safely.
338
Raise an error if not.
340
:param repository: A repository to fallback to.
342
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()
344
150
def all_revision_ids(self):
345
"""Returns a list of all the revision ids in the repository.
151
"""Returns a list of all the revision ids in the repository.
347
This is conceptually deprecated because code should generally work on
348
the graph reachable from a particular revision, and ignore any other
349
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.
352
if 'evil' in debug.debug_flags:
353
mutter_callsite(2, "all_revision_ids is linear with history.")
354
157
return self._all_revision_ids()
356
160
def _all_revision_ids(self):
357
"""Returns a list of all the revision ids in the repository.
161
"""Returns a list of all the revision ids in the repository.
359
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.
362
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)
364
175
def break_lock(self):
365
176
"""Break a lock if one is present from another instance.
622
340
For instance, if the repository is at URL/.bzr/repository,
623
341
Repository.open(URL) -> a Repository instance.
625
control = controldir.ControlDir.open(base)
343
control = bzrdir.BzrDir.open(base)
626
344
return control.open_repository()
628
346
def copy_content_into(self, destination, revision_id=None):
629
347
"""Make a complete copy of the content in self into destination.
631
This is a destructive operation! Do not use it on existing
349
This is a destructive operation! Do not use it on existing
352
revision_id = osutils.safe_revision_id(revision_id)
634
353
return InterRepository.get(self, destination).copy_content(revision_id)
636
def commit_write_group(self):
637
"""Commit the contents accrued within the current write group.
639
:seealso: start_write_group.
641
:return: it may return an opaque hint that can be passed to 'pack'.
643
if self._write_group is not self.get_transaction():
644
# has an unlock or relock occured ?
645
raise errors.BzrError('mismatched lock context %r and '
647
(self.get_transaction(), self._write_group))
648
result = self._commit_write_group()
649
self._write_group = None
652
def _commit_write_group(self):
653
"""Template method for per-repository write group cleanup.
655
This is called before the write group is considered to be
656
finished and should ensure that all data handed to the repository
657
for writing during the write group is safely committed (to the
658
extent possible considering file system caching etc).
661
def suspend_write_group(self):
662
"""Suspend a write group.
664
:raise UnsuspendableWriteGroup: If the write group can not be
666
:return: List of tokens
668
raise errors.UnsuspendableWriteGroup(self)
670
def refresh_data(self):
671
"""Re-read any data needed to synchronise with disk.
673
This method is intended to be called after another repository instance
674
(such as one used by a smart server) has inserted data into the
675
repository. On all repositories this will work outside of write groups.
676
Some repository formats (pack and newer for breezy native formats)
677
support refresh_data inside write groups. If called inside a write
678
group on a repository that does not support refreshing in a write group
679
IsInWriteGroupError will be raised.
683
def resume_write_group(self, tokens):
684
if not self.is_write_locked():
685
raise errors.NotWriteLocked(self)
686
if self._write_group:
687
raise errors.BzrError('already in a write group')
688
self._resume_write_group(tokens)
689
# so we can detect unlock/relock - the write group is now entered.
690
self._write_group = self.get_transaction()
692
def _resume_write_group(self, tokens):
693
raise errors.UnsuspendableWriteGroup(self)
695
def fetch(self, source, revision_id=None, find_ghosts=False, lossy=False):
355
def fetch(self, source, revision_id=None, pb=None):
696
356
"""Fetch the content required to construct revision_id from source.
698
If revision_id is None, then all content is copied.
700
fetch() may not be used when the repository is in a write group -
701
either finish the current write group before using fetch, or use
702
fetch before starting the write group.
704
:param find_ghosts: Find and copy revisions in the source that are
705
ghosts in the target (and not reachable directly by walking out to
706
the first-present revision in target from revision_id).
707
:param revision_id: If specified, all the content needed for this
708
revision ID will be copied to the target. Fetch will determine for
709
itself which content needs to be copied.
710
:return: A FetchResult object
358
If revision_id is None all content is copied.
712
if self.is_in_write_group():
713
raise errors.InternalBzrError(
714
"May not fetch while in a write group.")
715
# fast path same-url fetch operations
716
# TODO: lift out to somewhere common with RemoteRepository
717
# <https://bugs.launchpad.net/bzr/+bug/401646>
718
if (self.has_same_location(source)
719
and self._has_same_fallbacks(source)):
720
# check that last_revision is in 'from' and then return a
722
if (revision_id is not None and
723
not _mod_revision.is_null(revision_id)):
724
self.get_revision(revision_id)
360
revision_id = osutils.safe_revision_id(revision_id)
726
361
inter = InterRepository.get(source, self)
728
revision_id=revision_id, find_ghosts=find_ghosts, lossy=lossy)
363
return inter.fetch(revision_id=revision_id, pb=pb)
364
except NotImplementedError:
365
raise errors.IncompatibleRepositories(source, self)
730
def get_commit_builder(self, branch, parents, config_stack, timestamp=None,
731
timezone=None, committer=None, revprops=None,
732
revision_id=None, lossy=False):
367
def get_commit_builder(self, branch, parents, config, timestamp=None,
368
timezone=None, committer=None, revprops=None,
733
370
"""Obtain a CommitBuilder for this repository.
735
372
:param branch: Branch to commit to.
736
373
:param parents: Revision ids of the parents of the new revision.
737
:param config_stack: Configuration stack to use.
374
:param config: Configuration to use.
738
375
:param timestamp: Optional timestamp recorded for commit.
739
376
:param timezone: Optional timezone for timestamp.
740
377
:param committer: Optional committer to set for commit.
741
378
:param revprops: Optional dictionary of revision properties.
742
379
:param revision_id: Optional revision id.
743
:param lossy: Whether to discard data that can not be natively
744
represented, when pushing to a foreign VCS
746
raise NotImplementedError(self.get_commit_builder)
381
revision_id = osutils.safe_revision_id(revision_id)
382
return _CommitBuilder(self, parents, config, timestamp, timezone,
383
committer, revprops, revision_id)
748
@only_raises(errors.LockNotHeld, errors.LockBroken)
749
385
def unlock(self):
750
if (self.control_files._lock_count == 1 and
751
self.control_files._lock_mode == 'w'):
752
if self._write_group is not None:
753
self.abort_write_group()
754
self.control_files.unlock()
755
raise errors.BzrError(
756
'Must end write groups before releasing write locks.')
757
386
self.control_files.unlock()
758
if self.control_files._lock_count == 0:
759
for repo in self._fallback_repositories:
762
def clone(self, controldir, revision_id=None):
763
"""Clone this repository into controldir using the current format.
389
def clone(self, a_bzrdir, revision_id=None):
390
"""Clone this repository into a_bzrdir using the current format.
765
392
Currently no check is made that the format of this repository and
766
393
the bzrdir format are compatible. FIXME RBC 20060201.
768
395
:return: The newly created destination repository.
770
with self.lock_read():
771
# TODO: deprecate after 0.16; cloning this with all its settings is
772
# probably not very useful -- mbp 20070423
773
dest_repo = self._create_sprouting_repo(
774
controldir, shared=self.is_shared())
775
self.copy_content_into(dest_repo, revision_id)
778
def start_write_group(self):
779
"""Start a write group in the repository.
781
Write groups are used by repositories which do not have a 1:1 mapping
782
between file ids and backend store to manage the insertion of data from
783
both fetch and commit operations.
785
A write lock is required around the
786
start_write_group/commit_write_group for the support of lock-requiring
789
One can only insert data into a repository inside a write group.
793
if not self.is_write_locked():
794
raise errors.NotWriteLocked(self)
795
if self._write_group:
796
raise errors.BzrError('already in a write group')
797
self._start_write_group()
798
# so we can detect unlock/relock - the write group is now entered.
799
self._write_group = self.get_transaction()
801
def _start_write_group(self):
802
"""Template method for per-repository write group startup.
804
This is called before the write group is considered to be
808
def sprout(self, to_bzrdir, revision_id=None):
809
"""Create a descendent repository for new development.
811
Unlike clone, this does not copy the settings of the repository.
813
with self.lock_read():
814
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
815
dest_repo.fetch(self, revision_id=revision_id)
818
def _create_sprouting_repo(self, a_controldir, shared):
820
a_controldir._format, self.controldir._format.__class__):
397
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
821
398
# use target default format.
822
dest_repo = a_controldir.create_repository()
399
dest_repo = a_bzrdir.create_repository()
824
401
# Most control formats need the repository to be specifically
825
402
# created, but on some old all-in-one formats it's not needed
827
dest_repo = self._format.initialize(
828
a_controldir, shared=shared)
404
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
829
405
except errors.UninitializableFormat:
830
dest_repo = a_controldir.open_repository()
406
dest_repo = a_bzrdir.open_repository()
407
self.copy_content_into(dest_repo, revision_id)
833
411
def has_revision(self, revision_id):
834
412
"""True if this repository has a copy of the revision."""
835
with self.lock_read():
836
return revision_id in self.has_revisions((revision_id,))
838
def has_revisions(self, revision_ids):
839
"""Probe to find out the presence of multiple revisions.
841
:param revision_ids: An iterable of revision_ids.
842
:return: A set of the revision_ids that were present.
844
raise NotImplementedError(self.has_revisions)
846
def get_revision(self, revision_id):
847
"""Return the Revision object for a named revision."""
848
with self.lock_read():
849
return self.get_revisions([revision_id])[0]
413
revision_id = osutils.safe_revision_id(revision_id)
414
return self._revision_store.has_revision_id(revision_id,
415
self.get_transaction())
851
418
def get_revision_reconcile(self, revision_id):
852
419
"""'reconcile' helper routine that allows access to a revision always.
854
421
This variant of get_revision does not cross check the weave graph
855
422
against the revision one as get_revision does: but it should only
856
423
be used by reconcile, or reconcile-alike commands that are correcting
857
424
or testing the revision graph.
859
raise NotImplementedError(self.get_revision_reconcile)
426
if not revision_id or not isinstance(revision_id, basestring):
427
raise errors.InvalidRevisionId(revision_id=revision_id,
429
return self.get_revisions([revision_id])[0]
861
432
def get_revisions(self, revision_ids):
862
"""Get many revisions at once.
864
Repositories that need to check data on every revision read should
865
subclass this method.
868
for revid, rev in self.iter_revisions(revision_ids):
870
raise errors.NoSuchRevision(self, revid)
872
return [revs[revid] for revid in revision_ids]
874
def iter_revisions(self, revision_ids):
875
"""Iterate over revision objects.
877
:param revision_ids: An iterable of revisions to examine. None may be
878
passed to request all revisions known to the repository. Note that
879
not all repositories can find unreferenced revisions; for those
880
repositories only referenced ones will be returned.
881
:return: An iterator of (revid, revision) tuples. Absent revisions (
882
those asked for but not available) are returned as (revid, None).
883
N.B.: Revisions are not necessarily yielded in order.
885
raise NotImplementedError(self.iter_revisions)
887
def get_revision_delta(self, revision_id):
888
"""Return the delta for one revision.
890
The delta is relative to the left-hand predecessor of the
893
with self.lock_read():
894
r = self.get_revision(revision_id)
895
return list(self.get_revision_deltas([r]))[0]
897
def get_revision_deltas(self, revisions, specific_files=None):
433
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
434
revs = self._revision_store.get_revisions(revision_ids,
435
self.get_transaction())
437
assert not isinstance(rev.revision_id, unicode)
438
for parent_id in rev.parent_ids:
439
assert not isinstance(parent_id, unicode)
443
def get_revision_xml(self, revision_id):
444
# TODO: jam 20070210 This shouldn't be necessary since get_revision
445
# would have already do it.
446
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
447
revision_id = osutils.safe_revision_id(revision_id)
448
rev = self.get_revision(revision_id)
450
# the current serializer..
451
self._revision_store._serializer.write_revision(rev, rev_tmp)
453
return rev_tmp.getvalue()
456
def get_revision(self, revision_id):
457
"""Return the Revision object for a named revision"""
458
# TODO: jam 20070210 get_revision_reconcile should do this for us
459
revision_id = osutils.safe_revision_id(revision_id)
460
r = self.get_revision_reconcile(revision_id)
461
# weave corruption can lead to absent revision markers that should be
463
# the following test is reasonably cheap (it needs a single weave read)
464
# and the weave is cached in read transactions. In write transactions
465
# it is not cached but typically we only read a small number of
466
# revisions. For knits when they are introduced we will probably want
467
# to ensure that caching write transactions are in use.
468
inv = self.get_inventory_weave()
469
self._check_revision_parents(r, inv)
473
def get_deltas_for_revisions(self, revisions):
898
474
"""Produce a generator of revision deltas.
900
Note that the input is a sequence of REVISIONS, not revision ids.
476
Note that the input is a sequence of REVISIONS, not revision_ids.
901
477
Trees will be held in memory until the generator exits.
902
478
Each delta is relative to the revision's lefthand predecessor.
904
specific_files should exist in the first revision.
906
:param specific_files: if not None, the result is filtered
907
so that only those files, their parents and their
908
children are included.
910
from .tree import InterTree
911
# Get the revision-ids of interest
912
480
required_trees = set()
913
481
for revision in revisions:
914
482
required_trees.add(revision.revision_id)
915
483
required_trees.update(revision.parent_ids[:1])
918
t.get_revision_id(): t
919
for t in self.revision_trees(required_trees)}
921
# Calculate the deltas
484
trees = dict((t.get_revision_id(), t) for
485
t in self.revision_trees(required_trees))
922
486
for revision in revisions:
923
487
if not revision.parent_ids:
924
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
488
old_tree = self.revision_tree(None)
926
490
old_tree = trees[revision.parent_ids[0]]
927
intertree = InterTree.get(old_tree, trees[revision.revision_id])
928
yield intertree.compare(specific_files=specific_files)
929
if specific_files is not None:
931
p for p in intertree.find_source_paths(
932
specific_files).values()
491
yield trees[revision.revision_id].changes_from(old_tree)
494
def get_revision_delta(self, revision_id):
495
"""Return the delta for one revision.
497
The delta is relative to the left-hand predecessor of the
500
r = self.get_revision(revision_id)
501
return list(self.get_deltas_for_revisions([r]))[0]
503
def _check_revision_parents(self, revision, inventory):
504
"""Private to Repository and Fetch.
506
This checks the parentage of revision in an inventory weave for
507
consistency and is only applicable to inventory-weave-for-ancestry
508
using repository formats & fetchers.
510
weave_parents = inventory.get_parents(revision.revision_id)
511
weave_names = inventory.versions()
512
for parent_id in revision.parent_ids:
513
if parent_id in weave_names:
514
# this parent must not be a ghost.
515
if not parent_id in weave_parents:
517
raise errors.CorruptRepository(self)
935
520
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
936
raise NotImplementedError(self.store_revision_signature)
938
def add_signature_text(self, revision_id, signature):
939
"""Store a signature text for a revision.
941
:param revision_id: Revision id of the revision
942
:param signature: Signature text.
944
raise NotImplementedError(self.add_signature_text)
946
def iter_files_bytes(self, desired_files):
947
"""Iterate through file versions.
949
Files will not necessarily be returned in the order they occur in
950
desired_files. No specific order is guaranteed.
952
Yields pairs of identifier, bytes_iterator. identifier is an opaque
953
value supplied by the caller as part of desired_files. It should
954
uniquely identify the file version in the caller's context. (Examples:
955
an index number or a TreeTransform trans_id.)
957
:param desired_files: a list of (file_id, revision_id, identifier)
960
raise NotImplementedError(self.iter_files_bytes)
962
def get_rev_id_for_revno(self, revno, known_pair):
963
"""Return the revision id of a revno, given a later (revno, revid)
964
pair in the same history.
966
:return: if found (True, revid). If the available history ran out
967
before reaching the revno, then this returns
968
(False, (closest_revno, closest_revid)).
970
known_revno, known_revid = known_pair
971
partial_history = [known_revid]
972
distance_from_known = known_revno - revno
973
if distance_from_known < 0:
974
raise errors.RevnoOutOfBounds(revno, (0, known_revno))
977
self, partial_history, stop_index=distance_from_known)
978
except errors.RevisionNotPresent as err:
979
if err.revision_id == known_revid:
980
# The start revision (known_revid) wasn't found.
981
raise errors.NoSuchRevision(self, known_revid)
982
# This is a stacked repository with no fallbacks, or a there's a
983
# left-hand ghost. Either way, even though the revision named in
984
# the error isn't in this repo, we know it's the next step in this
986
partial_history.append(err.revision_id)
987
if len(partial_history) <= distance_from_known:
988
# Didn't find enough history to get a revid for the revno.
989
earliest_revno = known_revno - len(partial_history) + 1
990
return (False, (earliest_revno, partial_history[-1]))
991
if len(partial_history) - 1 > distance_from_known:
992
raise AssertionError('_iter_for_revno returned too much history')
993
return (True, partial_history[-1])
521
revision_id = osutils.safe_revision_id(revision_id)
522
signature = gpg_strategy.sign(plaintext)
523
self._revision_store.add_revision_signature_text(revision_id,
525
self.get_transaction())
527
def fileids_altered_by_revision_ids(self, revision_ids):
528
"""Find the file ids and versions affected by revisions.
530
:param revisions: an iterable containing revision ids.
531
:return: a dictionary mapping altered file-ids to an iterable of
532
revision_ids. Each altered file-ids has the exact revision_ids that
533
altered it listed explicitly.
535
assert self._serializer.support_altered_by_hack, \
536
("fileids_altered_by_revision_ids only supported for branches "
537
"which store inventory as unnested xml, not on %r" % self)
538
selected_revision_ids = set(osutils.safe_revision_id(r)
539
for r in revision_ids)
540
w = self.get_inventory_weave()
543
# this code needs to read every new line in every inventory for the
544
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
545
# not present in one of those inventories is unnecessary but not
546
# harmful because we are filtering by the revision id marker in the
547
# inventory lines : we only select file ids altered in one of those
548
# revisions. We don't need to see all lines in the inventory because
549
# only those added in an inventory in rev X can contain a revision=X
551
unescape_revid_cache = {}
552
unescape_fileid_cache = {}
554
# jam 20061218 In a big fetch, this handles hundreds of thousands
555
# of lines, so it has had a lot of inlining and optimizing done.
556
# Sorry that it is a little bit messy.
557
# Move several functions to be local variables, since this is a long
559
search = self._file_ids_altered_regex.search
560
unescape = _unescape_xml
561
setdefault = result.setdefault
562
pb = ui.ui_factory.nested_progress_bar()
564
for line in w.iter_lines_added_or_present_in_versions(
565
selected_revision_ids, pb=pb):
569
# One call to match.group() returning multiple items is quite a
570
# bit faster than 2 calls to match.group() each returning 1
571
file_id, revision_id = match.group('file_id', 'revision_id')
573
# Inlining the cache lookups helps a lot when you make 170,000
574
# lines and 350k ids, versus 8.4 unique ids.
575
# Using a cache helps in 2 ways:
576
# 1) Avoids unnecessary decoding calls
577
# 2) Re-uses cached strings, which helps in future set and
579
# (2) is enough that removing encoding entirely along with
580
# the cache (so we are using plain strings) results in no
581
# performance improvement.
583
revision_id = unescape_revid_cache[revision_id]
585
unescaped = unescape(revision_id)
586
unescape_revid_cache[revision_id] = unescaped
587
revision_id = unescaped
589
if revision_id in selected_revision_ids:
591
file_id = unescape_fileid_cache[file_id]
593
unescaped = unescape(file_id)
594
unescape_fileid_cache[file_id] = unescaped
596
setdefault(file_id, set()).add(revision_id)
602
def get_inventory_weave(self):
603
return self.control_weaves.get_weave('inventory',
604
self.get_transaction())
607
def get_inventory(self, revision_id):
608
"""Get Inventory object by hash."""
609
# TODO: jam 20070210 Technically we don't need to sanitize, since all
610
# called functions must sanitize.
611
revision_id = osutils.safe_revision_id(revision_id)
612
return self.deserialise_inventory(
613
revision_id, self.get_inventory_xml(revision_id))
615
def deserialise_inventory(self, revision_id, xml):
616
"""Transform the xml into an inventory object.
618
:param revision_id: The expected revision id of the inventory.
619
:param xml: A serialised inventory.
621
revision_id = osutils.safe_revision_id(revision_id)
622
result = self._serializer.read_inventory_from_string(xml)
623
result.root.revision = revision_id
626
def serialise_inventory(self, inv):
627
return self._serializer.write_inventory_to_string(inv)
630
def get_inventory_xml(self, revision_id):
631
"""Get inventory XML as a file object."""
632
revision_id = osutils.safe_revision_id(revision_id)
634
assert isinstance(revision_id, str), type(revision_id)
635
iw = self.get_inventory_weave()
636
return iw.get_text(revision_id)
638
raise errors.HistoryMissing(self, 'inventory', revision_id)
641
def get_inventory_sha1(self, revision_id):
642
"""Return the sha1 hash of the inventory entry
644
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
645
revision_id = osutils.safe_revision_id(revision_id)
646
return self.get_revision(revision_id).inventory_sha1
649
def get_revision_graph(self, revision_id=None):
650
"""Return a dictionary containing the revision graph.
652
:param revision_id: The revision_id to get a graph from. If None, then
653
the entire revision graph is returned. This is a deprecated mode of
654
operation and will be removed in the future.
655
:return: a dictionary of revision_id->revision_parents_list.
657
# special case NULL_REVISION
658
if revision_id == _mod_revision.NULL_REVISION:
660
revision_id = osutils.safe_revision_id(revision_id)
661
a_weave = self.get_inventory_weave()
662
all_revisions = self._eliminate_revisions_not_present(
664
entire_graph = dict([(node, a_weave.get_parents(node)) for
665
node in all_revisions])
666
if revision_id is None:
668
elif revision_id not in entire_graph:
669
raise errors.NoSuchRevision(self, revision_id)
671
# add what can be reached from revision_id
673
pending = set([revision_id])
674
while len(pending) > 0:
676
result[node] = entire_graph[node]
677
for revision_id in result[node]:
678
if revision_id not in result:
679
pending.add(revision_id)
683
def get_revision_graph_with_ghosts(self, revision_ids=None):
684
"""Return a graph of the revisions with ghosts marked as applicable.
686
:param revision_ids: an iterable of revisions to graph or None for all.
687
:return: a Graph object with the graph reachable from revision_ids.
689
result = graph.Graph()
691
pending = set(self.all_revision_ids())
694
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
695
# special case NULL_REVISION
696
if _mod_revision.NULL_REVISION in pending:
697
pending.remove(_mod_revision.NULL_REVISION)
698
required = set(pending)
701
revision_id = pending.pop()
703
rev = self.get_revision(revision_id)
704
except errors.NoSuchRevision:
705
if revision_id in required:
708
result.add_ghost(revision_id)
710
for parent_id in rev.parent_ids:
711
# is this queued or done ?
712
if (parent_id not in pending and
713
parent_id not in done):
715
pending.add(parent_id)
716
result.add_node(revision_id, rev.parent_ids)
717
done.add(revision_id)
720
def _get_history_vf(self):
721
"""Get a versionedfile whose history graph reflects all revisions.
723
For weave repositories, this is the inventory weave.
725
return self.get_inventory_weave()
727
def iter_reverse_revision_history(self, revision_id):
728
"""Iterate backwards through revision ids in the lefthand history
730
:param revision_id: The revision id to start with. All its lefthand
731
ancestors will be traversed.
733
revision_id = osutils.safe_revision_id(revision_id)
734
if revision_id in (None, _mod_revision.NULL_REVISION):
736
next_id = revision_id
737
versionedfile = self._get_history_vf()
740
parents = versionedfile.get_parents(next_id)
741
if len(parents) == 0:
747
def get_revision_inventory(self, revision_id):
748
"""Return inventory of a past revision."""
749
# TODO: Unify this with get_inventory()
750
# bzr 0.0.6 and later imposes the constraint that the inventory_id
751
# must be the same as its revision, so this is trivial.
752
if revision_id is None:
753
# This does not make sense: if there is no revision,
754
# then it is the current tree inventory surely ?!
755
# and thus get_root_id() is something that looks at the last
756
# commit on the branch, and the get_root_id is an inventory check.
757
raise NotImplementedError
758
# return Inventory(self.get_root_id())
760
return self.get_inventory(revision_id)
995
763
def is_shared(self):
996
764
"""Return True if this repository is flagged as a shared repository."""
997
765
raise NotImplementedError(self.is_shared)
999
768
def reconcile(self, other=None, thorough=False):
1000
769
"""Reconcile this repository."""
1001
raise NotImplementedError(self.reconcile)
1003
def _refresh_data(self):
1004
"""Helper called from lock_* to ensure coherency with disk.
1006
The default implementation does nothing; it is however possible
1007
for repositories to maintain loaded indices across multiple locks
1008
by checking inside their implementation of this method to see
1009
whether their indices are still valid. This depends of course on
1010
the disk format being validatable in this manner. This method is
1011
also called by the refresh_data() public interface to cause a refresh
1012
to occur while in a write lock so that data inserted by a smart server
1013
push operation is visible on the client's instance of the physical
770
from bzrlib.reconcile import RepoReconciler
771
reconciler = RepoReconciler(self, thorough=thorough)
772
reconciler.reconcile()
1017
776
def revision_tree(self, revision_id):
1018
777
"""Return Tree for a revision on this branch.
1020
`revision_id` may be NULL_REVISION for the empty tree revision.
779
`revision_id` may be None for the empty tree revision.
1022
raise NotImplementedError(self.revision_tree)
781
# TODO: refactor this to use an existing revision object
782
# so we don't need to read it in twice.
783
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
784
return RevisionTree(self, Inventory(root_id=None),
785
_mod_revision.NULL_REVISION)
787
revision_id = osutils.safe_revision_id(revision_id)
788
inv = self.get_revision_inventory(revision_id)
789
return RevisionTree(self, inv, revision_id)
1024
792
def revision_trees(self, revision_ids):
1025
"""Return Trees for revisions in this repository.
1027
:param revision_ids: a sequence of revision-ids;
1028
a revision-id may not be None or b'null:'
1030
raise NotImplementedError(self.revision_trees)
1032
def pack(self, hint=None, clean_obsolete_packs=False):
1033
"""Compress the data within the repository.
1035
This operation only makes sense for some repository types. For other
1036
types it should be a no-op that just returns.
1038
This stub method does not require a lock, but subclasses should use
1039
self.write_lock as this is a long running call it's reasonable to
1040
implicitly lock for the user.
1042
:param hint: If not supplied, the whole repository is packed.
1043
If supplied, the repository may use the hint parameter as a
1044
hint for the parts of the repository to pack. A hint can be
1045
obtained from the result of commit_write_group(). Out of
1046
date hints are simply ignored, because concurrent operations
1047
can obsolete them rapidly.
1049
:param clean_obsolete_packs: Clean obsolete packs immediately after
793
"""Return Tree for a revision on this branch.
795
`revision_id` may not be None or 'null:'"""
796
assert None not in revision_ids
797
assert _mod_revision.NULL_REVISION not in revision_ids
798
texts = self.get_inventory_weave().get_texts(revision_ids)
799
for text, revision_id in zip(texts, revision_ids):
800
inv = self.deserialise_inventory(revision_id, text)
801
yield RevisionTree(self, inv, revision_id)
804
def get_ancestry(self, revision_id):
805
"""Return a list of revision-ids integrated by a revision.
807
The first element of the list is always None, indicating the origin
808
revision. This might change when we have history horizons, or
809
perhaps we should have a new API.
811
This is topologically sorted.
813
if revision_id is None:
815
revision_id = osutils.safe_revision_id(revision_id)
816
if not self.has_revision(revision_id):
817
raise errors.NoSuchRevision(self, revision_id)
818
w = self.get_inventory_weave()
819
candidates = w.get_ancestry(revision_id)
820
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
823
def print_file(self, file, revision_id):
824
"""Print `file` to stdout.
826
FIXME RBC 20060125 as John Meinel points out this is a bad api
827
- it writes to stdout, it assumes that that is valid etc. Fix
828
by creating a new more flexible convenience function.
830
revision_id = osutils.safe_revision_id(revision_id)
831
tree = self.revision_tree(revision_id)
832
# use inventory as it was in that revision
833
file_id = tree.inventory.path2id(file)
835
# TODO: jam 20060427 Write a test for this code path
836
# it had a bug in it, and was raising the wrong
838
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
839
tree.print_file(file_id)
1053
841
def get_transaction(self):
1054
842
return self.control_files.get_transaction()
1056
def get_parent_map(self, revision_ids):
1057
"""See graph.StackedParentsProvider.get_parent_map"""
1058
raise NotImplementedError(self.get_parent_map)
1060
def _get_parent_map_no_fallbacks(self, revision_ids):
1061
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1062
# revisions index works in keys; this just works in revisions
1063
# therefore wrap and unwrap
1066
for revision_id in revision_ids:
1067
if revision_id == _mod_revision.NULL_REVISION:
1068
result[revision_id] = ()
1069
elif revision_id is None:
1070
raise ValueError('get_parent_map(None) is not valid')
1072
query_keys.append((revision_id,))
1073
vf = self.revisions.without_fallbacks()
1074
for (revision_id,), parent_keys in (
1075
vf.get_parent_map(query_keys).items()):
1077
result[revision_id] = tuple([parent_revid
1078
for (parent_revid,) in parent_keys])
1080
result[revision_id] = (_mod_revision.NULL_REVISION,)
1083
def _make_parents_provider(self):
1084
if not self._format.supports_external_lookups:
1086
return graph.StackedParentsProvider(_LazyListJoin(
1087
[self._make_parents_provider_unstacked()],
1088
self._fallback_repositories))
1090
def _make_parents_provider_unstacked(self):
1091
return graph.CallableToParentsProviderAdapter(
1092
self._get_parent_map_no_fallbacks)
1094
def get_known_graph_ancestry(self, revision_ids):
1095
"""Return the known graph for a set of revision ids and their ancestors.
1097
raise NotImplementedError(self.get_known_graph_ancestry)
1099
def get_file_graph(self):
1100
"""Return the graph walker for files."""
1101
raise NotImplementedError(self.get_file_graph)
1103
def get_graph(self, other_repository=None):
1104
"""Return the graph walker for this repository format"""
1105
parents_provider = self._make_parents_provider()
1106
if (other_repository is not None and
1107
not self.has_same_location(other_repository)):
1108
parents_provider = graph.StackedParentsProvider(
1109
[parents_provider, other_repository._make_parents_provider()])
1110
return graph.Graph(parents_provider)
844
def revision_parents(self, revision_id):
845
revision_id = osutils.safe_revision_id(revision_id)
846
return self.get_inventory_weave().parent_names(revision_id)
1112
849
def set_make_working_trees(self, new_value):
1113
850
"""Set the policy flag for making working trees when creating branches.
1342
1109
def __ne__(self, other):
1343
1110
return not self == other
1113
def find_format(klass, a_bzrdir):
1114
"""Return the format for the repository object in a_bzrdir.
1116
This is used by bzr native formats that have a "format" file in
1117
the repository. Other methods may be used by different types of
1121
transport = a_bzrdir.get_repository_transport(None)
1122
format_string = transport.get("format").read()
1123
return format_registry.get(format_string)
1124
except errors.NoSuchFile:
1125
raise errors.NoRepositoryPresent(a_bzrdir)
1127
raise errors.UnknownFormatError(format=format_string)
1130
def register_format(klass, format):
1131
format_registry.register(format.get_format_string(), format)
1134
def unregister_format(klass, format):
1135
format_registry.remove(format.get_format_string())
1138
def get_default_format(klass):
1139
"""Return the current default format."""
1140
from bzrlib import bzrdir
1141
return bzrdir.format_registry.make_bzrdir('default').repository_format
1143
def _get_control_store(self, repo_transport, control_files):
1144
"""Return the control store for this repository."""
1145
raise NotImplementedError(self._get_control_store)
1147
def get_format_string(self):
1148
"""Return the ASCII format string that identifies this format.
1150
Note that in pre format ?? repositories the format string is
1151
not permitted nor written to disk.
1153
raise NotImplementedError(self.get_format_string)
1345
1155
def get_format_description(self):
1346
1156
"""Return the short description for this format."""
1347
1157
raise NotImplementedError(self.get_format_description)
1349
def initialize(self, controldir, shared=False):
1350
"""Initialize a repository of this format in controldir.
1352
:param controldir: The controldir to put the new repository in it.
1159
def _get_revision_store(self, repo_transport, control_files):
1160
"""Return the revision store object for this a_bzrdir."""
1161
raise NotImplementedError(self._get_revision_store)
1163
def _get_text_rev_store(self,
1170
"""Common logic for getting a revision store for a repository.
1172
see self._get_revision_store for the subclass-overridable method to
1173
get the store for a repository.
1175
from bzrlib.store.revision.text import TextRevisionStore
1176
dir_mode = control_files._dir_mode
1177
file_mode = control_files._file_mode
1178
text_store = TextStore(transport.clone(name),
1180
compressed=compressed,
1182
file_mode=file_mode)
1183
_revision_store = TextRevisionStore(text_store, serializer)
1184
return _revision_store
1186
# TODO: this shouldn't be in the base class, it's specific to things that
1187
# use weaves or knits -- mbp 20070207
1188
def _get_versioned_file_store(self,
1193
versionedfile_class=None,
1194
versionedfile_kwargs={},
1196
if versionedfile_class is None:
1197
versionedfile_class = self._versionedfile_class
1198
weave_transport = control_files._transport.clone(name)
1199
dir_mode = control_files._dir_mode
1200
file_mode = control_files._file_mode
1201
return VersionedFileStore(weave_transport, prefixed=prefixed,
1203
file_mode=file_mode,
1204
versionedfile_class=versionedfile_class,
1205
versionedfile_kwargs=versionedfile_kwargs,
1208
def initialize(self, a_bzrdir, shared=False):
1209
"""Initialize a repository of this format in a_bzrdir.
1211
:param a_bzrdir: The bzrdir to put the new repository in it.
1353
1212
:param shared: The repository should be initialized as a sharable one.
1354
:returns: The new repository object.
1356
1214
This may raise UninitializableFormat if shared repository are not
1357
compatible the controldir.
1215
compatible the a_bzrdir.
1359
raise NotImplementedError(self.initialize)
1361
1218
def is_supported(self):
1362
1219
"""Is this format supported?
1364
1221
Supported formats must be initializable and openable.
1365
Unsupported formats may not support initialization or committing or
1222
Unsupported formats may not support initialization or committing or
1366
1223
some other features depending on the reason for not being supported.
1370
def is_deprecated(self):
1371
"""Is this format deprecated?
1373
Deprecated formats may trigger a user-visible warning recommending
1374
the user to upgrade. They are still fully supported.
1378
def network_name(self):
1379
"""A simple byte string uniquely identifying this format for RPC calls.
1381
MetaDir repository formats use their disk format string to identify the
1382
repository over the wire. All in one formats such as bzr < 0.8, and
1383
foreign formats like svn/git and hg should use some marker which is
1384
unique and immutable.
1386
raise NotImplementedError(self.network_name)
1388
1227
def check_conversion_target(self, target_format):
1389
if self.rich_root_data and not target_format.rich_root_data:
1390
raise errors.BadConversionTarget(
1391
'Does not support rich root data.', target_format,
1393
if (self.supports_tree_reference
1394
and not getattr(target_format, 'supports_tree_reference', False)):
1395
raise errors.BadConversionTarget(
1396
'Does not support nested trees', target_format,
1399
def open(self, controldir, _found=False):
1400
"""Return an instance of this format for a controldir.
1228
raise NotImplementedError(self.check_conversion_target)
1230
def open(self, a_bzrdir, _found=False):
1231
"""Return an instance of this format for the bzrdir a_bzrdir.
1402
1233
_found is a private parameter, do not use it.
1404
1235
raise NotImplementedError(self.open)
1406
def _run_post_repo_init_hooks(self, repository, controldir, shared):
1407
from .controldir import ControlDir, RepoInitHookParams
1408
hooks = ControlDir.hooks['post_repo_init']
1411
params = RepoInitHookParams(repository, self, controldir, shared)
1416
# formats which have no format string are not discoverable or independently
1417
# creatable on disk, so are not registered in format_registry. They're
1418
# all in breezy.bzr.knitreponow. When an instance of one of these is
1419
# needed, it's constructed directly by the ControlDir. Non-native formats where
1238
class MetaDirRepositoryFormat(RepositoryFormat):
1239
"""Common base class for the new repositories using the metadir layout."""
1241
rich_root_data = False
1242
supports_tree_reference = False
1243
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1246
super(MetaDirRepositoryFormat, self).__init__()
1248
def _create_control_files(self, a_bzrdir):
1249
"""Create the required files and the initial control_files object."""
1250
# FIXME: RBC 20060125 don't peek under the covers
1251
# NB: no need to escape relative paths that are url safe.
1252
repository_transport = a_bzrdir.get_repository_transport(self)
1253
control_files = lockable_files.LockableFiles(repository_transport,
1254
'lock', lockdir.LockDir)
1255
control_files.create_lock()
1256
return control_files
1258
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1259
"""Upload the initial blank content."""
1260
control_files = self._create_control_files(a_bzrdir)
1261
control_files.lock_write()
1263
control_files._transport.mkdir_multi(dirs,
1264
mode=control_files._dir_mode)
1265
for file, content in files:
1266
control_files.put(file, content)
1267
for file, content in utf8_files:
1268
control_files.put_utf8(file, content)
1270
control_files.put_utf8('shared-storage', '')
1272
control_files.unlock()
1275
# formats which have no format string are not discoverable
1276
# and not independently creatable, so are not registered. They're
1277
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1278
# needed, it's constructed directly by the BzrDir. Non-native formats where
1420
1279
# the repository is not separately opened are similar.
1422
1281
format_registry.register_lazy(
1423
b'Bazaar-NG Knit Repository Format 1',
1424
'breezy.bzr.knitrepo',
1282
'Bazaar-NG Repository format 7',
1283
'bzrlib.repofmt.weaverepo',
1286
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1287
# default control directory format
1289
format_registry.register_lazy(
1290
'Bazaar-NG Knit Repository Format 1',
1291
'bzrlib.repofmt.knitrepo',
1425
1292
'RepositoryFormatKnit1',
1294
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1428
1296
format_registry.register_lazy(
1429
b'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1430
'breezy.bzr.knitrepo',
1297
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1298
'bzrlib.repofmt.knitrepo',
1431
1299
'RepositoryFormatKnit3',
1434
format_registry.register_lazy(
1435
b'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
1436
'breezy.bzr.knitrepo',
1437
'RepositoryFormatKnit4',
1440
# Pack-based formats. There is one format for pre-subtrees, and one for
1441
# post-subtrees to allow ease of testing.
1442
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
1443
format_registry.register_lazy(
1444
b'Bazaar pack repository format 1 (needs bzr 0.92)\n',
1445
'breezy.bzr.knitpack_repo',
1446
'RepositoryFormatKnitPack1',
1448
format_registry.register_lazy(
1449
b'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
1450
'breezy.bzr.knitpack_repo',
1451
'RepositoryFormatKnitPack3',
1453
format_registry.register_lazy(
1454
b'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
1455
'breezy.bzr.knitpack_repo',
1456
'RepositoryFormatKnitPack4',
1458
format_registry.register_lazy(
1459
b'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
1460
'breezy.bzr.knitpack_repo',
1461
'RepositoryFormatKnitPack5',
1463
format_registry.register_lazy(
1464
b'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
1465
'breezy.bzr.knitpack_repo',
1466
'RepositoryFormatKnitPack5RichRoot',
1468
format_registry.register_lazy(
1469
b'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
1470
'breezy.bzr.knitpack_repo',
1471
'RepositoryFormatKnitPack5RichRootBroken',
1473
format_registry.register_lazy(
1474
b'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
1475
'breezy.bzr.knitpack_repo',
1476
'RepositoryFormatKnitPack6',
1478
format_registry.register_lazy(
1479
b'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
1480
'breezy.bzr.knitpack_repo',
1481
'RepositoryFormatKnitPack6RichRoot',
1483
format_registry.register_lazy(
1484
b'Bazaar repository format 2a (needs bzr 1.16 or later)\n',
1485
'breezy.bzr.groupcompress_repo',
1486
'RepositoryFormat2a',
1489
# Development formats.
1490
# Check their docstrings to see if/when they are obsolete.
1491
format_registry.register_lazy(
1492
(b"Bazaar development format 2 with subtree support "
1493
b"(needs bzr.dev from before 1.8)\n"),
1494
'breezy.bzr.knitpack_repo',
1495
'RepositoryFormatPackDevelopment2Subtree',
1497
format_registry.register_lazy(
1498
b'Bazaar development format 8\n',
1499
'breezy.bzr.groupcompress_repo',
1500
'RepositoryFormat2aSubtree',
1504
1303
class InterRepository(InterObject):
1505
1304
"""This class represents operations taking place between two repositories.
1507
1306
Its instances have methods like copy_content and fetch, and contain
1508
references to the source and target repositories these operations can be
1307
references to the source and target repositories these operations can be
1509
1308
carried out on.
1511
1310
Often we will provide convenience methods on 'repository' which carry out
1517
1316
"""The available optimised InterRepository types."""
1519
1318
def copy_content(self, revision_id=None):
1520
"""Make a complete copy of the content in self into destination.
1522
This is a destructive operation! Do not use it on existing
1525
:param revision_id: Only copy the content needed to construct
1526
revision_id and its parents.
1528
with self.lock_write():
1530
self.target.set_make_working_trees(
1531
self.source.make_working_trees())
1532
except (NotImplementedError, errors.RepositoryUpgradeRequired):
1534
self.target.fetch(self.source, revision_id=revision_id)
1536
def fetch(self, revision_id=None, find_ghosts=False, lossy=False):
1319
raise NotImplementedError(self.copy_content)
1321
def fetch(self, revision_id=None, pb=None):
1537
1322
"""Fetch the content required to construct revision_id.
1539
1324
The content is copied from self.source to self.target.
1541
1326
:param revision_id: if None all content is copied, if NULL_REVISION no
1542
1327
content is copied.
1543
:return: FetchResult
1328
:param pb: optional progress bar to use for progress reports. If not
1329
provided a default one will be created.
1331
Returns the copied revision count and the failed revisions in a tuple:
1545
1334
raise NotImplementedError(self.fetch)
1547
def search_missing_revision_ids(
1548
self, find_ghosts=True, revision_ids=None, if_present_ids=None,
1337
def missing_revision_ids(self, revision_id=None):
1550
1338
"""Return the revision ids that source has that target does not.
1552
:param revision_ids: return revision ids included by these
1553
revision_ids. NoSuchRevision will be raised if any of these
1554
revisions are not present.
1555
:param if_present_ids: like revision_ids, but will not cause
1556
NoSuchRevision if any of these are absent, instead they will simply
1557
not be in the result. This is useful for e.g. finding revisions
1558
to fetch for tags, which may reference absent revisions.
1559
:param find_ghosts: If True find missing revisions in deep history
1560
rather than just finding the surface difference.
1561
:param limit: Maximum number of revisions to return, topologically
1563
:return: A breezy.graph.SearchResult.
1565
raise NotImplementedError(self.search_missing_revision_ids)
1568
def _same_model(source, target):
1569
"""True if source and target have the same data representation.
1571
Note: this is always called on the base class; overriding it in a
1572
subclass will have no effect.
1575
InterRepository._assert_same_model(source, target)
1577
except errors.IncompatibleRepositories as e:
1581
def _assert_same_model(source, target):
1582
"""Raise an exception if two repositories do not use the same model.
1340
These are returned in topological order.
1342
:param revision_id: only return revision ids included by this
1345
# generic, possibly worst case, slow code path.
1346
target_ids = set(self.target.all_revision_ids())
1347
if revision_id is not None:
1348
# TODO: jam 20070210 InterRepository is internal enough that it
1349
# should assume revision_ids are already utf-8
1350
revision_id = osutils.safe_revision_id(revision_id)
1351
source_ids = self.source.get_ancestry(revision_id)
1352
assert source_ids[0] is None
1355
source_ids = self.source.all_revision_ids()
1356
result_set = set(source_ids).difference(target_ids)
1357
# this may look like a no-op: its not. It preserves the ordering
1358
# other_ids had while only returning the members from other_ids
1359
# that we've decided we need.
1360
return [rev_id for rev_id in source_ids if rev_id in result_set]
1363
class InterSameDataRepository(InterRepository):
1364
"""Code for converting between repositories that represent the same data.
1366
Data format and model must match for this to work.
1370
def _get_repo_format_to_test(self):
1371
"""Repository format for testing with."""
1372
return RepositoryFormat.get_default_format()
1375
def is_compatible(source, target):
1584
1376
if source.supports_rich_root() != target.supports_rich_root():
1585
raise errors.IncompatibleRepositories(source, target,
1586
"different rich-root support")
1587
1378
if source._serializer != target._serializer:
1588
raise errors.IncompatibleRepositories(source, target,
1589
"different serializers")
1383
def copy_content(self, revision_id=None):
1384
"""Make a complete copy of the content in self into destination.
1386
This is a destructive operation! Do not use it on existing
1389
:param revision_id: Only copy the content needed to construct
1390
revision_id and its parents.
1393
self.target.set_make_working_trees(self.source.make_working_trees())
1394
except NotImplementedError:
1396
# TODO: jam 20070210 This is fairly internal, so we should probably
1397
# just assert that revision_id is not unicode.
1398
revision_id = osutils.safe_revision_id(revision_id)
1399
# but don't bother fetching if we have the needed data now.
1400
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1401
self.target.has_revision(revision_id)):
1403
self.target.fetch(self.source, revision_id=revision_id)
1406
def fetch(self, revision_id=None, pb=None):
1407
"""See InterRepository.fetch()."""
1408
from bzrlib.fetch import GenericRepoFetcher
1409
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1410
self.source, self.source._format, self.target,
1411
self.target._format)
1412
# TODO: jam 20070210 This should be an assert, not a translate
1413
revision_id = osutils.safe_revision_id(revision_id)
1414
f = GenericRepoFetcher(to_repository=self.target,
1415
from_repository=self.source,
1416
last_revision=revision_id,
1418
return f.count_copied, f.failed_revisions
1421
class InterWeaveRepo(InterSameDataRepository):
1422
"""Optimised code paths between Weave based repositories."""
1425
def _get_repo_format_to_test(self):
1426
from bzrlib.repofmt import weaverepo
1427
return weaverepo.RepositoryFormat7()
1430
def is_compatible(source, target):
1431
"""Be compatible with known Weave formats.
1433
We don't test for the stores being of specific types because that
1434
could lead to confusing results, and there is no need to be
1437
from bzrlib.repofmt.weaverepo import (
1443
return (isinstance(source._format, (RepositoryFormat5,
1445
RepositoryFormat7)) and
1446
isinstance(target._format, (RepositoryFormat5,
1448
RepositoryFormat7)))
1449
except AttributeError:
1453
def copy_content(self, revision_id=None):
1454
"""See InterRepository.copy_content()."""
1455
# weave specific optimised path:
1456
# TODO: jam 20070210 Internal, should be an assert, not translate
1457
revision_id = osutils.safe_revision_id(revision_id)
1459
self.target.set_make_working_trees(self.source.make_working_trees())
1460
except NotImplementedError:
1462
# FIXME do not peek!
1463
if self.source.control_files._transport.listable():
1464
pb = ui.ui_factory.nested_progress_bar()
1466
self.target.weave_store.copy_all_ids(
1467
self.source.weave_store,
1469
from_transaction=self.source.get_transaction(),
1470
to_transaction=self.target.get_transaction())
1471
pb.update('copying inventory', 0, 1)
1472
self.target.control_weaves.copy_multi(
1473
self.source.control_weaves, ['inventory'],
1474
from_transaction=self.source.get_transaction(),
1475
to_transaction=self.target.get_transaction())
1476
self.target._revision_store.text_store.copy_all_ids(
1477
self.source._revision_store.text_store,
1482
self.target.fetch(self.source, revision_id=revision_id)
1485
def fetch(self, revision_id=None, pb=None):
1486
"""See InterRepository.fetch()."""
1487
from bzrlib.fetch import GenericRepoFetcher
1488
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1489
self.source, self.source._format, self.target, self.target._format)
1490
# TODO: jam 20070210 This should be an assert, not a translate
1491
revision_id = osutils.safe_revision_id(revision_id)
1492
f = GenericRepoFetcher(to_repository=self.target,
1493
from_repository=self.source,
1494
last_revision=revision_id,
1496
return f.count_copied, f.failed_revisions
1499
def missing_revision_ids(self, revision_id=None):
1500
"""See InterRepository.missing_revision_ids()."""
1501
# we want all revisions to satisfy revision_id in source.
1502
# but we don't want to stat every file here and there.
1503
# we want then, all revisions other needs to satisfy revision_id
1504
# checked, but not those that we have locally.
1505
# so the first thing is to get a subset of the revisions to
1506
# satisfy revision_id in source, and then eliminate those that
1507
# we do already have.
1508
# this is slow on high latency connection to self, but as as this
1509
# disk format scales terribly for push anyway due to rewriting
1510
# inventory.weave, this is considered acceptable.
1512
if revision_id is not None:
1513
source_ids = self.source.get_ancestry(revision_id)
1514
assert source_ids[0] is None
1517
source_ids = self.source._all_possible_ids()
1518
source_ids_set = set(source_ids)
1519
# source_ids is the worst possible case we may need to pull.
1520
# now we want to filter source_ids against what we actually
1521
# have in target, but don't try to check for existence where we know
1522
# we do not have a revision as that would be pointless.
1523
target_ids = set(self.target._all_possible_ids())
1524
possibly_present_revisions = target_ids.intersection(source_ids_set)
1525
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1526
required_revisions = source_ids_set.difference(actually_present_revisions)
1527
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1528
if revision_id is not None:
1529
# we used get_ancestry to determine source_ids then we are assured all
1530
# revisions referenced are present as they are installed in topological order.
1531
# and the tip revision was validated by get_ancestry.
1532
return required_topo_revisions
1534
# if we just grabbed the possibly available ids, then
1535
# we only have an estimate of whats available and need to validate
1536
# that against the revision records.
1537
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1540
class InterKnitRepo(InterSameDataRepository):
1541
"""Optimised code paths between Knit based repositories."""
1544
def _get_repo_format_to_test(self):
1545
from bzrlib.repofmt import knitrepo
1546
return knitrepo.RepositoryFormatKnit1()
1549
def is_compatible(source, target):
1550
"""Be compatible with known Knit formats.
1552
We don't test for the stores being of specific types because that
1553
could lead to confusing results, and there is no need to be
1556
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1558
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1559
isinstance(target._format, (RepositoryFormatKnit1)))
1560
except AttributeError:
1564
def fetch(self, revision_id=None, pb=None):
1565
"""See InterRepository.fetch()."""
1566
from bzrlib.fetch import KnitRepoFetcher
1567
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1568
self.source, self.source._format, self.target, self.target._format)
1569
# TODO: jam 20070210 This should be an assert, not a translate
1570
revision_id = osutils.safe_revision_id(revision_id)
1571
f = KnitRepoFetcher(to_repository=self.target,
1572
from_repository=self.source,
1573
last_revision=revision_id,
1575
return f.count_copied, f.failed_revisions
1578
def missing_revision_ids(self, revision_id=None):
1579
"""See InterRepository.missing_revision_ids()."""
1580
if revision_id is not None:
1581
source_ids = self.source.get_ancestry(revision_id)
1582
assert source_ids[0] is None
1585
source_ids = self.source._all_possible_ids()
1586
source_ids_set = set(source_ids)
1587
# source_ids is the worst possible case we may need to pull.
1588
# now we want to filter source_ids against what we actually
1589
# have in target, but don't try to check for existence where we know
1590
# we do not have a revision as that would be pointless.
1591
target_ids = set(self.target._all_possible_ids())
1592
possibly_present_revisions = target_ids.intersection(source_ids_set)
1593
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1594
required_revisions = source_ids_set.difference(actually_present_revisions)
1595
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1596
if revision_id is not None:
1597
# we used get_ancestry to determine source_ids then we are assured all
1598
# revisions referenced are present as they are installed in topological order.
1599
# and the tip revision was validated by get_ancestry.
1600
return required_topo_revisions
1602
# if we just grabbed the possibly available ids, then
1603
# we only have an estimate of whats available and need to validate
1604
# that against the revision records.
1605
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1608
class InterModel1and2(InterRepository):
1611
def _get_repo_format_to_test(self):
1615
def is_compatible(source, target):
1616
if not source.supports_rich_root() and target.supports_rich_root():
1622
def fetch(self, revision_id=None, pb=None):
1623
"""See InterRepository.fetch()."""
1624
from bzrlib.fetch import Model1toKnit2Fetcher
1625
# TODO: jam 20070210 This should be an assert, not a translate
1626
revision_id = osutils.safe_revision_id(revision_id)
1627
f = Model1toKnit2Fetcher(to_repository=self.target,
1628
from_repository=self.source,
1629
last_revision=revision_id,
1631
return f.count_copied, f.failed_revisions
1634
def copy_content(self, revision_id=None):
1635
"""Make a complete copy of the content in self into destination.
1637
This is a destructive operation! Do not use it on existing
1640
:param revision_id: Only copy the content needed to construct
1641
revision_id and its parents.
1644
self.target.set_make_working_trees(self.source.make_working_trees())
1645
except NotImplementedError:
1647
# TODO: jam 20070210 Internal, assert, don't translate
1648
revision_id = osutils.safe_revision_id(revision_id)
1649
# but don't bother fetching if we have the needed data now.
1650
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1651
self.target.has_revision(revision_id)):
1653
self.target.fetch(self.source, revision_id=revision_id)
1656
class InterKnit1and2(InterKnitRepo):
1659
def _get_repo_format_to_test(self):
1663
def is_compatible(source, target):
1664
"""Be compatible with Knit1 source and Knit3 target"""
1665
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1667
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1668
RepositoryFormatKnit3
1669
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1670
isinstance(target._format, (RepositoryFormatKnit3)))
1671
except AttributeError:
1675
def fetch(self, revision_id=None, pb=None):
1676
"""See InterRepository.fetch()."""
1677
from bzrlib.fetch import Knit1to2Fetcher
1678
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1679
self.source, self.source._format, self.target,
1680
self.target._format)
1681
# TODO: jam 20070210 This should be an assert, not a translate
1682
revision_id = osutils.safe_revision_id(revision_id)
1683
f = Knit1to2Fetcher(to_repository=self.target,
1684
from_repository=self.source,
1685
last_revision=revision_id,
1687
return f.count_copied, f.failed_revisions
1690
InterRepository.register_optimiser(InterSameDataRepository)
1691
InterRepository.register_optimiser(InterWeaveRepo)
1692
InterRepository.register_optimiser(InterKnitRepo)
1693
InterRepository.register_optimiser(InterModel1and2)
1694
InterRepository.register_optimiser(InterKnit1and2)
1697
class RepositoryTestProviderAdapter(object):
1698
"""A tool to generate a suite testing multiple repository formats at once.
1700
This is done by copying the test once for each transport and injecting
1701
the transport_server, transport_readonly_server, and bzrdir_format and
1702
repository_format classes into each copy. Each copy is also given a new id()
1703
to make it easy to identify.
1706
def __init__(self, transport_server, transport_readonly_server, formats):
1707
self._transport_server = transport_server
1708
self._transport_readonly_server = transport_readonly_server
1709
self._formats = formats
1711
def adapt(self, test):
1712
result = unittest.TestSuite()
1713
for repository_format, bzrdir_format in self._formats:
1714
from copy import deepcopy
1715
new_test = deepcopy(test)
1716
new_test.transport_server = self._transport_server
1717
new_test.transport_readonly_server = self._transport_readonly_server
1718
new_test.bzrdir_format = bzrdir_format
1719
new_test.repository_format = repository_format
1720
def make_new_test_id():
1721
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1722
return lambda: new_id
1723
new_test.id = make_new_test_id()
1724
result.addTest(new_test)
1728
class InterRepositoryTestProviderAdapter(object):
1729
"""A tool to generate a suite testing multiple inter repository formats.
1731
This is done by copying the test once for each interrepo provider and injecting
1732
the transport_server, transport_readonly_server, repository_format and
1733
repository_to_format classes into each copy.
1734
Each copy is also given a new id() to make it easy to identify.
1737
def __init__(self, transport_server, transport_readonly_server, formats):
1738
self._transport_server = transport_server
1739
self._transport_readonly_server = transport_readonly_server
1740
self._formats = formats
1742
def adapt(self, test):
1743
result = unittest.TestSuite()
1744
for interrepo_class, repository_format, repository_format_to in self._formats:
1745
from copy import deepcopy
1746
new_test = deepcopy(test)
1747
new_test.transport_server = self._transport_server
1748
new_test.transport_readonly_server = self._transport_readonly_server
1749
new_test.interrepo_class = interrepo_class
1750
new_test.repository_format = repository_format
1751
new_test.repository_format_to = repository_format_to
1752
def make_new_test_id():
1753
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1754
return lambda: new_id
1755
new_test.id = make_new_test_id()
1756
result.addTest(new_test)
1760
def default_test_list():
1761
"""Generate the default list of interrepo permutations to test."""
1762
from bzrlib.repofmt import knitrepo, weaverepo
1764
# test the default InterRepository between format 6 and the current
1766
# XXX: robertc 20060220 reinstate this when there are two supported
1767
# formats which do not have an optimal code path between them.
1768
#result.append((InterRepository,
1769
# RepositoryFormat6(),
1770
# RepositoryFormatKnit1()))
1771
for optimiser_class in InterRepository._optimisers:
1772
format_to_test = optimiser_class._get_repo_format_to_test()
1773
if format_to_test is not None:
1774
result.append((optimiser_class,
1775
format_to_test, format_to_test))
1776
# if there are specific combinations we want to use, we can add them
1778
result.append((InterModel1and2,
1779
weaverepo.RepositoryFormat5(),
1780
knitrepo.RepositoryFormatKnit3()))
1781
result.append((InterKnit1and2,
1782
knitrepo.RepositoryFormatKnit1(),
1783
knitrepo.RepositoryFormatKnit3()))
1592
1787
class CopyConverter(object):
1593
1788
"""A repository conversion tool which just performs a copy of the content.
1595
1790
This is slow but quite reliable.
1601
1796
:param target_format: The format the resulting repository should be.
1603
1798
self.target_format = target_format
1605
1800
def convert(self, repo, pb):
1606
1801
"""Perform the conversion of to_convert, giving feedback via pb.
1608
1803
:param to_convert: The disk object to convert.
1609
1804
:param pb: a progress bar to use for progress information.
1611
with ui.ui_factory.nested_progress_bar() as pb:
1614
# this is only useful with metadir layouts - separated repo content.
1615
# trigger an assertion if not such
1616
repo._format.get_format_string()
1617
self.repo_dir = repo.controldir
1618
pb.update(gettext('Moving repository to repository.backup'))
1619
self.repo_dir.transport.move('repository', 'repository.backup')
1620
backup_transport = self.repo_dir.transport.clone(
1621
'repository.backup')
1622
repo._format.check_conversion_target(self.target_format)
1623
self.source_repo = repo._format.open(self.repo_dir,
1625
_override_transport=backup_transport)
1626
pb.update(gettext('Creating new repository'))
1627
converted = self.target_format.initialize(self.repo_dir,
1628
self.source_repo.is_shared())
1629
with converted.lock_write():
1630
pb.update(gettext('Copying content'))
1631
self.source_repo.copy_content_into(converted)
1632
pb.update(gettext('Deleting old repository content'))
1633
self.repo_dir.transport.delete_tree('repository.backup')
1634
ui.ui_factory.note(gettext('repository converted'))
1637
def _strip_NULL_ghosts(revision_graph):
1638
"""Also don't use this. more compatibility code for unmigrated clients."""
1639
# Filter ghosts, and null:
1640
if _mod_revision.NULL_REVISION in revision_graph:
1641
del revision_graph[_mod_revision.NULL_REVISION]
1642
for key, parents in revision_graph.items():
1643
revision_graph[key] = tuple(parent for parent in parents if parent
1645
return revision_graph
1648
def _iter_for_revno(repo, partial_history_cache, stop_index=None,
1649
stop_revision=None):
1650
"""Extend the partial history to include a given index
1652
If a stop_index is supplied, stop when that index has been reached.
1653
If a stop_revision is supplied, stop when that revision is
1654
encountered. Otherwise, stop when the beginning of history is
1657
:param stop_index: The index which should be present. When it is
1658
present, history extension will stop.
1659
:param stop_revision: The revision id which should be present. When
1660
it is encountered, history extension will stop.
1662
start_revision = partial_history_cache[-1]
1663
graph = repo.get_graph()
1664
iterator = graph.iter_lefthand_ancestry(start_revision,
1665
(_mod_revision.NULL_REVISION,))
1809
# this is only useful with metadir layouts - separated repo content.
1810
# trigger an assertion if not such
1811
repo._format.get_format_string()
1812
self.repo_dir = repo.bzrdir
1813
self.step('Moving repository to repository.backup')
1814
self.repo_dir.transport.move('repository', 'repository.backup')
1815
backup_transport = self.repo_dir.transport.clone('repository.backup')
1816
repo._format.check_conversion_target(self.target_format)
1817
self.source_repo = repo._format.open(self.repo_dir,
1819
_override_transport=backup_transport)
1820
self.step('Creating new repository')
1821
converted = self.target_format.initialize(self.repo_dir,
1822
self.source_repo.is_shared())
1823
converted.lock_write()
1825
self.step('Copying content into repository.')
1826
self.source_repo.copy_content_into(converted)
1829
self.step('Deleting old repository content.')
1830
self.repo_dir.transport.delete_tree('repository.backup')
1831
self.pb.note('repository converted')
1833
def step(self, message):
1834
"""Update the pb by a step."""
1836
self.pb.update(message, self.count, self.total)
1839
class CommitBuilder(object):
1840
"""Provides an interface to build up a commit.
1842
This allows describing a tree to be committed without needing to
1843
know the internals of the format of the repository.
1846
record_root_entry = False
1847
def __init__(self, repository, parents, config, timestamp=None,
1848
timezone=None, committer=None, revprops=None,
1850
"""Initiate a CommitBuilder.
1852
:param repository: Repository to commit to.
1853
:param parents: Revision ids of the parents of the new revision.
1854
:param config: Configuration to use.
1855
:param timestamp: Optional timestamp recorded for commit.
1856
:param timezone: Optional timezone for timestamp.
1857
:param committer: Optional committer to set for commit.
1858
:param revprops: Optional dictionary of revision properties.
1859
:param revision_id: Optional revision id.
1861
self._config = config
1863
if committer is None:
1864
self._committer = self._config.username()
1866
assert isinstance(committer, basestring), type(committer)
1867
self._committer = committer
1869
self.new_inventory = Inventory(None)
1870
self._new_revision_id = osutils.safe_revision_id(revision_id)
1871
self.parents = parents
1872
self.repository = repository
1875
if revprops is not None:
1876
self._revprops.update(revprops)
1878
if timestamp is None:
1879
timestamp = time.time()
1880
# Restrict resolution to 1ms
1881
self._timestamp = round(timestamp, 3)
1883
if timezone is None:
1884
self._timezone = osutils.local_time_offset()
1886
self._timezone = int(timezone)
1888
self._generate_revision_if_needed()
1890
def commit(self, message):
1891
"""Make the actual commit.
1893
:return: The revision id of the recorded revision.
1895
rev = _mod_revision.Revision(
1896
timestamp=self._timestamp,
1897
timezone=self._timezone,
1898
committer=self._committer,
1900
inventory_sha1=self.inv_sha1,
1901
revision_id=self._new_revision_id,
1902
properties=self._revprops)
1903
rev.parent_ids = self.parents
1904
self.repository.add_revision(self._new_revision_id, rev,
1905
self.new_inventory, self._config)
1906
return self._new_revision_id
1908
def revision_tree(self):
1909
"""Return the tree that was just committed.
1911
After calling commit() this can be called to get a RevisionTree
1912
representing the newly committed tree. This is preferred to
1913
calling Repository.revision_tree() because that may require
1914
deserializing the inventory, while we already have a copy in
1917
return RevisionTree(self.repository, self.new_inventory,
1918
self._new_revision_id)
1920
def finish_inventory(self):
1921
"""Tell the builder that the inventory is finished."""
1922
if self.new_inventory.root is None:
1923
symbol_versioning.warn('Root entry should be supplied to'
1924
' record_entry_contents, as of bzr 0.10.',
1925
DeprecationWarning, stacklevel=2)
1926
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1927
self.new_inventory.revision_id = self._new_revision_id
1928
self.inv_sha1 = self.repository.add_inventory(
1929
self._new_revision_id,
1934
def _gen_revision_id(self):
1935
"""Return new revision-id."""
1936
return generate_ids.gen_revision_id(self._config.username(),
1939
def _generate_revision_if_needed(self):
1940
"""Create a revision id if None was supplied.
1942
If the repository can not support user-specified revision ids
1943
they should override this function and raise CannotSetRevisionId
1944
if _new_revision_id is not None.
1946
:raises: CannotSetRevisionId
1948
if self._new_revision_id is None:
1949
self._new_revision_id = self._gen_revision_id()
1951
def record_entry_contents(self, ie, parent_invs, path, tree):
1952
"""Record the content of ie from tree into the commit if needed.
1954
Side effect: sets ie.revision when unchanged
1956
:param ie: An inventory entry present in the commit.
1957
:param parent_invs: The inventories of the parent revisions of the
1959
:param path: The path the entry is at in the tree.
1960
:param tree: The tree which contains this entry and should be used to
1963
if self.new_inventory.root is None and ie.parent_id is not None:
1964
symbol_versioning.warn('Root entry should be supplied to'
1965
' record_entry_contents, as of bzr 0.10.',
1966
DeprecationWarning, stacklevel=2)
1967
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1969
self.new_inventory.add(ie)
1971
# ie.revision is always None if the InventoryEntry is considered
1972
# for committing. ie.snapshot will record the correct revision
1973
# which may be the sole parent if it is untouched.
1974
if ie.revision is not None:
1977
# In this revision format, root entries have no knit or weave
1978
if ie is self.new_inventory.root:
1979
# When serializing out to disk and back in
1980
# root.revision is always _new_revision_id
1981
ie.revision = self._new_revision_id
1983
previous_entries = ie.find_previous_heads(
1985
self.repository.weave_store,
1986
self.repository.get_transaction())
1987
# we are creating a new revision for ie in the history store
1989
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1991
def modified_directory(self, file_id, file_parents):
1992
"""Record the presence of a symbolic link.
1994
:param file_id: The file_id of the link to record.
1995
:param file_parents: The per-file parent revision ids.
1997
self._add_text_to_weave(file_id, [], file_parents.keys())
1999
def modified_reference(self, file_id, file_parents):
2000
"""Record the modification of a reference.
2002
:param file_id: The file_id of the link to record.
2003
:param file_parents: The per-file parent revision ids.
2005
self._add_text_to_weave(file_id, [], file_parents.keys())
2007
def modified_file_text(self, file_id, file_parents,
2008
get_content_byte_lines, text_sha1=None,
2010
"""Record the text of file file_id
2012
:param file_id: The file_id of the file to record the text of.
2013
:param file_parents: The per-file parent revision ids.
2014
:param get_content_byte_lines: A callable which will return the byte
2016
:param text_sha1: Optional SHA1 of the file contents.
2017
:param text_size: Optional size of the file contents.
2019
# mutter('storing text of file {%s} in revision {%s} into %r',
2020
# file_id, self._new_revision_id, self.repository.weave_store)
2021
# special case to avoid diffing on renames or
2023
if (len(file_parents) == 1
2024
and text_sha1 == file_parents.values()[0].text_sha1
2025
and text_size == file_parents.values()[0].text_size):
2026
previous_ie = file_parents.values()[0]
2027
versionedfile = self.repository.weave_store.get_weave(file_id,
2028
self.repository.get_transaction())
2029
versionedfile.clone_text(self._new_revision_id,
2030
previous_ie.revision, file_parents.keys())
2031
return text_sha1, text_size
2033
new_lines = get_content_byte_lines()
2034
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2035
# should return the SHA1 and size
2036
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2037
return osutils.sha_strings(new_lines), \
2038
sum(map(len, new_lines))
2040
def modified_link(self, file_id, file_parents, link_target):
2041
"""Record the presence of a symbolic link.
2043
:param file_id: The file_id of the link to record.
2044
:param file_parents: The per-file parent revision ids.
2045
:param link_target: Target location of this link.
2047
self._add_text_to_weave(file_id, [], file_parents.keys())
2049
def _add_text_to_weave(self, file_id, new_lines, parents):
2050
versionedfile = self.repository.weave_store.get_weave_or_empty(
2051
file_id, self.repository.get_transaction())
2052
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2053
versionedfile.clear_cache()
2056
class _CommitBuilder(CommitBuilder):
2057
"""Temporary class so old CommitBuilders are detected properly
2059
Note: CommitBuilder works whether or not root entry is recorded.
2062
record_root_entry = True
2065
class RootCommitBuilder(CommitBuilder):
2066
"""This commitbuilder actually records the root id"""
2068
record_root_entry = True
2070
def record_entry_contents(self, ie, parent_invs, path, tree):
2071
"""Record the content of ie from tree into the commit if needed.
2073
Side effect: sets ie.revision when unchanged
2075
:param ie: An inventory entry present in the commit.
2076
:param parent_invs: The inventories of the parent revisions of the
2078
:param path: The path the entry is at in the tree.
2079
:param tree: The tree which contains this entry and should be used to
2082
assert self.new_inventory.root is not None or ie.parent_id is None
2083
self.new_inventory.add(ie)
2085
# ie.revision is always None if the InventoryEntry is considered
2086
# for committing. ie.snapshot will record the correct revision
2087
# which may be the sole parent if it is untouched.
2088
if ie.revision is not None:
2091
previous_entries = ie.find_previous_heads(
2093
self.repository.weave_store,
2094
self.repository.get_transaction())
2095
# we are creating a new revision for ie in the history store
2097
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2109
def _unescaper(match, _map=_unescape_map):
2110
code = match.group(1)
1667
# skip the last revision in the list
1670
if (stop_index is not None and
1671
len(partial_history_cache) > stop_index):
1673
if partial_history_cache[-1] == stop_revision:
1675
revision_id = next(iterator)
1676
partial_history_cache.append(revision_id)
1677
except StopIteration:
1682
class _LazyListJoin(object):
1683
"""An iterable yielding the contents of many lists as one list.
1685
Each iterator made from this will reflect the current contents of the lists
1686
at the time the iterator is made.
1688
This is used by Repository's _make_parents_provider implementation so that
1691
pp = repo._make_parents_provider() # uses a list of fallback repos
1692
pp.add_fallback_repository(other_repo) # appends to that list
1693
result = pp.get_parent_map(...)
1694
# The result will include revs from other_repo
1697
def __init__(self, *list_parts):
1698
self.list_parts = list_parts
1702
for list_part in self.list_parts:
1703
full_list.extend(list_part)
1704
return iter(full_list)
1707
return "%s.%s(%s)" % (self.__module__, self.__class__.__name__,
2114
if not code.startswith('#'):
2116
return unichr(int(code[1:])).encode('utf8')
2122
def _unescape_xml(data):
2123
"""Unescape predefined XML entities in a string of data."""
2125
if _unescape_re is None:
2126
_unescape_re = re.compile('\&([^;]*);')
2127
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
