1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
38
revision as _mod_revision,
43
from bzrlib.bundle import serializer
44
from bzrlib.revisiontree import RevisionTree
45
from bzrlib.store.versioned import VersionedFileStore
46
from bzrlib.store.text import TextStore
47
from bzrlib.testament import Testament
51
from bzrlib.decorators import needs_read_lock, needs_write_lock
52
from bzrlib.inter import InterObject
53
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
54
from bzrlib.symbol_versioning import (
58
from bzrlib.trace import mutter, note, warning
61
# Old formats display a warning, but only once
62
_deprecation_warning_done = False
65
######################################################################
68
class Repository(object):
69
"""Repository holding history for one or more branches.
71
The repository holds and retrieves historical information including
72
revisions and file history. It's normally accessed only by the Branch,
73
which views a particular line of development through that history.
75
The Repository builds on top of Stores and a Transport, which respectively
76
describe the disk data format and the way of accessing the (possibly
80
_file_ids_altered_regex = lazy_regex.lazy_compile(
81
r'file_id="(?P<file_id>[^"]+)"'
82
r'.*revision="(?P<revision_id>[^"]+)"'
86
def add_inventory(self, revision_id, inv, parents):
87
"""Add the inventory inv to the repository as revision_id.
89
:param parents: The revision ids of the parents that revision_id
90
is known to have and are in the repository already.
92
returns the sha1 of the serialized inventory.
94
revision_id = osutils.safe_revision_id(revision_id)
95
_mod_revision.check_not_reserved_id(revision_id)
96
assert inv.revision_id is None or inv.revision_id == revision_id, \
97
"Mismatch between inventory revision" \
98
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
99
assert inv.root is not None
100
inv_text = self.serialise_inventory(inv)
101
inv_sha1 = osutils.sha_string(inv_text)
102
inv_vf = self.control_weaves.get_weave('inventory',
103
self.get_transaction())
104
self._inventory_add_lines(inv_vf, revision_id, parents,
105
osutils.split_lines(inv_text))
108
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
110
for parent in parents:
112
final_parents.append(parent)
114
inv_vf.add_lines(revision_id, final_parents, lines)
117
def add_revision(self, revision_id, rev, inv=None, config=None):
118
"""Add rev to the revision store as revision_id.
120
:param revision_id: the revision id to use.
121
:param rev: The revision object.
122
:param inv: The inventory for the revision. if None, it will be looked
123
up in the inventory storer
124
:param config: If None no digital signature will be created.
125
If supplied its signature_needed method will be used
126
to determine if a signature should be made.
128
revision_id = osutils.safe_revision_id(revision_id)
129
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
131
_mod_revision.check_not_reserved_id(revision_id)
132
if config is not None and config.signature_needed():
134
inv = self.get_inventory(revision_id)
135
plaintext = Testament(rev, inv).as_short_text()
136
self.store_revision_signature(
137
gpg.GPGStrategy(config), plaintext, revision_id)
138
if not revision_id in self.get_inventory_weave():
140
raise errors.WeaveRevisionNotPresent(revision_id,
141
self.get_inventory_weave())
143
# yes, this is not suitable for adding with ghosts.
144
self.add_inventory(revision_id, inv, rev.parent_ids)
145
self._revision_store.add_revision(rev, self.get_transaction())
147
def _add_revision_text(self, revision_id, text):
148
revision = self._revision_store._serializer.read_revision_from_string(
150
self._revision_store._add_revision(revision, StringIO(text),
151
self.get_transaction())
154
def _all_possible_ids(self):
155
"""Return all the possible revisions that we could find."""
156
return self.get_inventory_weave().versions()
158
def all_revision_ids(self):
159
"""Returns a list of all the revision ids in the repository.
161
This is deprecated because code should generally work on the graph
162
reachable from a particular revision, and ignore any other revisions
163
that might be present. There is no direct replacement method.
165
return self._all_revision_ids()
168
def _all_revision_ids(self):
169
"""Returns a list of all the revision ids in the repository.
171
These are in as much topological order as the underlying store can
172
present: for weaves ghosts may lead to a lack of correctness until
173
the reweave updates the parents list.
175
if self._revision_store.text_store.listable():
176
return self._revision_store.all_revision_ids(self.get_transaction())
177
result = self._all_possible_ids()
178
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
179
# ids. (It should, since _revision_store's API should change to
180
# return utf8 revision_ids)
181
return self._eliminate_revisions_not_present(result)
183
def break_lock(self):
184
"""Break a lock if one is present from another instance.
186
Uses the ui factory to ask for confirmation if the lock may be from
189
self.control_files.break_lock()
192
def _eliminate_revisions_not_present(self, revision_ids):
193
"""Check every revision id in revision_ids to see if we have it.
195
Returns a set of the present revisions.
198
for id in revision_ids:
199
if self.has_revision(id):
204
def create(a_bzrdir):
205
"""Construct the current default format repository in a_bzrdir."""
206
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
208
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
209
"""instantiate a Repository.
211
:param _format: The format of the repository on disk.
212
:param a_bzrdir: The BzrDir of the repository.
214
In the future we will have a single api for all stores for
215
getting file texts, inventories and revisions, then
216
this construct will accept instances of those things.
218
super(Repository, self).__init__()
219
self._format = _format
220
# the following are part of the public API for Repository:
221
self.bzrdir = a_bzrdir
222
self.control_files = control_files
223
self._revision_store = _revision_store
224
self.text_store = text_store
225
# backwards compatibility
226
self.weave_store = text_store
227
# not right yet - should be more semantically clear ?
229
self.control_store = control_store
230
self.control_weaves = control_store
231
# TODO: make sure to construct the right store classes, etc, depending
232
# on whether escaping is required.
233
self._warn_if_deprecated()
236
return '%s(%r)' % (self.__class__.__name__,
237
self.bzrdir.transport.base)
240
return self.control_files.is_locked()
242
def lock_write(self, token=None):
243
"""Lock this repository for writing.
245
:param token: if this is already locked, then lock_write will fail
246
unless the token matches the existing lock.
247
:returns: a token if this instance supports tokens, otherwise None.
248
:raises TokenLockingNotSupported: when a token is given but this
249
instance doesn't support using token locks.
250
:raises MismatchedToken: if the specified token doesn't match the token
251
of the existing lock.
253
A token should be passed in if you know that you have locked the object
254
some other way, and need to synchronise this object's state with that
257
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
259
return self.control_files.lock_write(token=token)
262
self.control_files.lock_read()
264
def get_physical_lock_status(self):
265
return self.control_files.get_physical_lock_status()
267
def leave_lock_in_place(self):
268
"""Tell this repository not to release the physical lock when this
271
If lock_write doesn't return a token, then this method is not supported.
273
self.control_files.leave_in_place()
275
def dont_leave_lock_in_place(self):
276
"""Tell this repository to release the physical lock when this
277
object is unlocked, even if it didn't originally acquire it.
279
If lock_write doesn't return a token, then this method is not supported.
281
self.control_files.dont_leave_in_place()
284
def gather_stats(self, revid=None, committers=None):
285
"""Gather statistics from a revision id.
287
:param revid: The revision id to gather statistics from, if None, then
288
no revision specific statistics are gathered.
289
:param committers: Optional parameter controlling whether to grab
290
a count of committers from the revision specific statistics.
291
:return: A dictionary of statistics. Currently this contains:
292
committers: The number of committers if requested.
293
firstrev: A tuple with timestamp, timezone for the penultimate left
294
most ancestor of revid, if revid is not the NULL_REVISION.
295
latestrev: A tuple with timestamp, timezone for revid, if revid is
296
not the NULL_REVISION.
297
revisions: The total revision count in the repository.
298
size: An estimate disk size of the repository in bytes.
301
if revid and committers:
302
result['committers'] = 0
303
if revid and revid != _mod_revision.NULL_REVISION:
305
all_committers = set()
306
revisions = self.get_ancestry(revid)
307
# pop the leading None
309
first_revision = None
311
# ignore the revisions in the middle - just grab first and last
312
revisions = revisions[0], revisions[-1]
313
for revision in self.get_revisions(revisions):
314
if not first_revision:
315
first_revision = revision
317
all_committers.add(revision.committer)
318
last_revision = revision
320
result['committers'] = len(all_committers)
321
result['firstrev'] = (first_revision.timestamp,
322
first_revision.timezone)
323
result['latestrev'] = (last_revision.timestamp,
324
last_revision.timezone)
326
# now gather global repository information
327
if self.bzrdir.root_transport.listable():
328
c, t = self._revision_store.total_size(self.get_transaction())
329
result['revisions'] = c
334
def missing_revision_ids(self, other, revision_id=None):
335
"""Return the revision ids that other has that this does not.
337
These are returned in topological order.
339
revision_id: only return revision ids included by revision_id.
341
revision_id = osutils.safe_revision_id(revision_id)
342
return InterRepository.get(other, self).missing_revision_ids(revision_id)
346
"""Open the repository rooted at base.
348
For instance, if the repository is at URL/.bzr/repository,
349
Repository.open(URL) -> a Repository instance.
351
control = bzrdir.BzrDir.open(base)
352
return control.open_repository()
354
def copy_content_into(self, destination, revision_id=None):
355
"""Make a complete copy of the content in self into destination.
357
This is a destructive operation! Do not use it on existing
360
revision_id = osutils.safe_revision_id(revision_id)
361
return InterRepository.get(self, destination).copy_content(revision_id)
363
def fetch(self, source, revision_id=None, pb=None):
364
"""Fetch the content required to construct revision_id from source.
366
If revision_id is None all content is copied.
368
revision_id = osutils.safe_revision_id(revision_id)
369
inter = InterRepository.get(source, self)
371
return inter.fetch(revision_id=revision_id, pb=pb)
372
except NotImplementedError:
373
raise errors.IncompatibleRepositories(source, self)
375
def create_bundle(self, target, base, fileobj, format=None):
376
return serializer.write_bundle(self, target, base, fileobj, format)
378
def get_commit_builder(self, branch, parents, config, timestamp=None,
379
timezone=None, committer=None, revprops=None,
381
"""Obtain a CommitBuilder for this repository.
383
:param branch: Branch to commit to.
384
:param parents: Revision ids of the parents of the new revision.
385
:param config: Configuration to use.
386
:param timestamp: Optional timestamp recorded for commit.
387
:param timezone: Optional timezone for timestamp.
388
:param committer: Optional committer to set for commit.
389
:param revprops: Optional dictionary of revision properties.
390
:param revision_id: Optional revision id.
392
revision_id = osutils.safe_revision_id(revision_id)
393
return _CommitBuilder(self, parents, config, timestamp, timezone,
394
committer, revprops, revision_id)
397
self.control_files.unlock()
400
def clone(self, a_bzrdir, revision_id=None):
401
"""Clone this repository into a_bzrdir using the current format.
403
Currently no check is made that the format of this repository and
404
the bzrdir format are compatible. FIXME RBC 20060201.
406
:return: The newly created destination repository.
408
# TODO: deprecate after 0.16; cloning this with all its settings is
409
# probably not very useful -- mbp 20070423
410
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
411
self.copy_content_into(dest_repo, revision_id)
415
def sprout(self, to_bzrdir, revision_id=None):
416
"""Create a descendent repository for new development.
418
Unlike clone, this does not copy the settings of the repository.
420
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
421
dest_repo.fetch(self, revision_id=revision_id)
424
def _create_sprouting_repo(self, a_bzrdir, shared):
425
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
426
# use target default format.
427
dest_repo = a_bzrdir.create_repository()
429
# Most control formats need the repository to be specifically
430
# created, but on some old all-in-one formats it's not needed
432
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
433
except errors.UninitializableFormat:
434
dest_repo = a_bzrdir.open_repository()
438
def has_revision(self, revision_id):
439
"""True if this repository has a copy of the revision."""
440
revision_id = osutils.safe_revision_id(revision_id)
441
return self._revision_store.has_revision_id(revision_id,
442
self.get_transaction())
445
def get_revision_reconcile(self, revision_id):
446
"""'reconcile' helper routine that allows access to a revision always.
448
This variant of get_revision does not cross check the weave graph
449
against the revision one as get_revision does: but it should only
450
be used by reconcile, or reconcile-alike commands that are correcting
451
or testing the revision graph.
453
if not revision_id or not isinstance(revision_id, basestring):
454
raise errors.InvalidRevisionId(revision_id=revision_id,
456
return self.get_revisions([revision_id])[0]
459
def get_revisions(self, revision_ids):
460
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
461
revs = self._revision_store.get_revisions(revision_ids,
462
self.get_transaction())
464
assert not isinstance(rev.revision_id, unicode)
465
for parent_id in rev.parent_ids:
466
assert not isinstance(parent_id, unicode)
470
def get_revision_xml(self, revision_id):
471
# TODO: jam 20070210 This shouldn't be necessary since get_revision
472
# would have already do it.
473
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
474
revision_id = osutils.safe_revision_id(revision_id)
475
rev = self.get_revision(revision_id)
477
# the current serializer..
478
self._revision_store._serializer.write_revision(rev, rev_tmp)
480
return rev_tmp.getvalue()
483
def get_revision(self, revision_id):
484
"""Return the Revision object for a named revision"""
485
# TODO: jam 20070210 get_revision_reconcile should do this for us
486
revision_id = osutils.safe_revision_id(revision_id)
487
r = self.get_revision_reconcile(revision_id)
488
# weave corruption can lead to absent revision markers that should be
490
# the following test is reasonably cheap (it needs a single weave read)
491
# and the weave is cached in read transactions. In write transactions
492
# it is not cached but typically we only read a small number of
493
# revisions. For knits when they are introduced we will probably want
494
# to ensure that caching write transactions are in use.
495
inv = self.get_inventory_weave()
496
self._check_revision_parents(r, inv)
500
def get_deltas_for_revisions(self, revisions):
501
"""Produce a generator of revision deltas.
503
Note that the input is a sequence of REVISIONS, not revision_ids.
504
Trees will be held in memory until the generator exits.
505
Each delta is relative to the revision's lefthand predecessor.
507
required_trees = set()
508
for revision in revisions:
509
required_trees.add(revision.revision_id)
510
required_trees.update(revision.parent_ids[:1])
511
trees = dict((t.get_revision_id(), t) for
512
t in self.revision_trees(required_trees))
513
for revision in revisions:
514
if not revision.parent_ids:
515
old_tree = self.revision_tree(None)
517
old_tree = trees[revision.parent_ids[0]]
518
yield trees[revision.revision_id].changes_from(old_tree)
521
def get_revision_delta(self, revision_id):
522
"""Return the delta for one revision.
524
The delta is relative to the left-hand predecessor of the
527
r = self.get_revision(revision_id)
528
return list(self.get_deltas_for_revisions([r]))[0]
530
def _check_revision_parents(self, revision, inventory):
531
"""Private to Repository and Fetch.
533
This checks the parentage of revision in an inventory weave for
534
consistency and is only applicable to inventory-weave-for-ancestry
535
using repository formats & fetchers.
537
weave_parents = inventory.get_parents(revision.revision_id)
538
weave_names = inventory.versions()
539
for parent_id in revision.parent_ids:
540
if parent_id in weave_names:
541
# this parent must not be a ghost.
542
if not parent_id in weave_parents:
544
raise errors.CorruptRepository(self)
547
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
548
revision_id = osutils.safe_revision_id(revision_id)
549
signature = gpg_strategy.sign(plaintext)
550
self._revision_store.add_revision_signature_text(revision_id,
552
self.get_transaction())
554
def fileids_altered_by_revision_ids(self, revision_ids):
555
"""Find the file ids and versions affected by revisions.
557
:param revisions: an iterable containing revision ids.
558
:return: a dictionary mapping altered file-ids to an iterable of
559
revision_ids. Each altered file-ids has the exact revision_ids that
560
altered it listed explicitly.
562
assert self._serializer.support_altered_by_hack, \
563
("fileids_altered_by_revision_ids only supported for branches "
564
"which store inventory as unnested xml, not on %r" % self)
565
selected_revision_ids = set(osutils.safe_revision_id(r)
566
for r in revision_ids)
567
w = self.get_inventory_weave()
570
# this code needs to read every new line in every inventory for the
571
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
572
# not present in one of those inventories is unnecessary but not
573
# harmful because we are filtering by the revision id marker in the
574
# inventory lines : we only select file ids altered in one of those
575
# revisions. We don't need to see all lines in the inventory because
576
# only those added in an inventory in rev X can contain a revision=X
578
unescape_revid_cache = {}
579
unescape_fileid_cache = {}
581
# jam 20061218 In a big fetch, this handles hundreds of thousands
582
# of lines, so it has had a lot of inlining and optimizing done.
583
# Sorry that it is a little bit messy.
584
# Move several functions to be local variables, since this is a long
586
search = self._file_ids_altered_regex.search
587
unescape = _unescape_xml
588
setdefault = result.setdefault
589
pb = ui.ui_factory.nested_progress_bar()
591
for line in w.iter_lines_added_or_present_in_versions(
592
selected_revision_ids, pb=pb):
596
# One call to match.group() returning multiple items is quite a
597
# bit faster than 2 calls to match.group() each returning 1
598
file_id, revision_id = match.group('file_id', 'revision_id')
600
# Inlining the cache lookups helps a lot when you make 170,000
601
# lines and 350k ids, versus 8.4 unique ids.
602
# Using a cache helps in 2 ways:
603
# 1) Avoids unnecessary decoding calls
604
# 2) Re-uses cached strings, which helps in future set and
606
# (2) is enough that removing encoding entirely along with
607
# the cache (so we are using plain strings) results in no
608
# performance improvement.
610
revision_id = unescape_revid_cache[revision_id]
612
unescaped = unescape(revision_id)
613
unescape_revid_cache[revision_id] = unescaped
614
revision_id = unescaped
616
if revision_id in selected_revision_ids:
618
file_id = unescape_fileid_cache[file_id]
620
unescaped = unescape(file_id)
621
unescape_fileid_cache[file_id] = unescaped
623
setdefault(file_id, set()).add(revision_id)
629
def get_inventory_weave(self):
630
return self.control_weaves.get_weave('inventory',
631
self.get_transaction())
634
def get_inventory(self, revision_id):
635
"""Get Inventory object by hash."""
636
# TODO: jam 20070210 Technically we don't need to sanitize, since all
637
# called functions must sanitize.
638
revision_id = osutils.safe_revision_id(revision_id)
639
return self.deserialise_inventory(
640
revision_id, self.get_inventory_xml(revision_id))
642
def deserialise_inventory(self, revision_id, xml):
643
"""Transform the xml into an inventory object.
645
:param revision_id: The expected revision id of the inventory.
646
:param xml: A serialised inventory.
648
revision_id = osutils.safe_revision_id(revision_id)
649
result = self._serializer.read_inventory_from_string(xml)
650
result.root.revision = revision_id
653
def serialise_inventory(self, inv):
654
return self._serializer.write_inventory_to_string(inv)
656
def get_serializer_format(self):
657
return self._serializer.format_num
660
def get_inventory_xml(self, revision_id):
661
"""Get inventory XML as a file object."""
662
revision_id = osutils.safe_revision_id(revision_id)
664
assert isinstance(revision_id, str), type(revision_id)
665
iw = self.get_inventory_weave()
666
return iw.get_text(revision_id)
668
raise errors.HistoryMissing(self, 'inventory', revision_id)
671
def get_inventory_sha1(self, revision_id):
672
"""Return the sha1 hash of the inventory entry
674
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
675
revision_id = osutils.safe_revision_id(revision_id)
676
return self.get_revision(revision_id).inventory_sha1
679
def get_revision_graph(self, revision_id=None):
680
"""Return a dictionary containing the revision graph.
682
:param revision_id: The revision_id to get a graph from. If None, then
683
the entire revision graph is returned. This is a deprecated mode of
684
operation and will be removed in the future.
685
:return: a dictionary of revision_id->revision_parents_list.
687
# special case NULL_REVISION
688
if revision_id == _mod_revision.NULL_REVISION:
690
revision_id = osutils.safe_revision_id(revision_id)
691
a_weave = self.get_inventory_weave()
692
all_revisions = self._eliminate_revisions_not_present(
694
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
695
node in all_revisions])
696
if revision_id is None:
698
elif revision_id not in entire_graph:
699
raise errors.NoSuchRevision(self, revision_id)
701
# add what can be reached from revision_id
703
pending = set([revision_id])
704
while len(pending) > 0:
706
result[node] = entire_graph[node]
707
for revision_id in result[node]:
708
if revision_id not in result:
709
pending.add(revision_id)
713
def get_revision_graph_with_ghosts(self, revision_ids=None):
714
"""Return a graph of the revisions with ghosts marked as applicable.
716
:param revision_ids: an iterable of revisions to graph or None for all.
717
:return: a Graph object with the graph reachable from revision_ids.
719
result = deprecated_graph.Graph()
721
pending = set(self.all_revision_ids())
724
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
725
# special case NULL_REVISION
726
if _mod_revision.NULL_REVISION in pending:
727
pending.remove(_mod_revision.NULL_REVISION)
728
required = set(pending)
731
revision_id = pending.pop()
733
rev = self.get_revision(revision_id)
734
except errors.NoSuchRevision:
735
if revision_id in required:
738
result.add_ghost(revision_id)
740
for parent_id in rev.parent_ids:
741
# is this queued or done ?
742
if (parent_id not in pending and
743
parent_id not in done):
745
pending.add(parent_id)
746
result.add_node(revision_id, rev.parent_ids)
747
done.add(revision_id)
750
def _get_history_vf(self):
751
"""Get a versionedfile whose history graph reflects all revisions.
753
For weave repositories, this is the inventory weave.
755
return self.get_inventory_weave()
757
def iter_reverse_revision_history(self, revision_id):
758
"""Iterate backwards through revision ids in the lefthand history
760
:param revision_id: The revision id to start with. All its lefthand
761
ancestors will be traversed.
763
revision_id = osutils.safe_revision_id(revision_id)
764
if revision_id in (None, _mod_revision.NULL_REVISION):
766
next_id = revision_id
767
versionedfile = self._get_history_vf()
770
parents = versionedfile.get_parents(next_id)
771
if len(parents) == 0:
777
def get_revision_inventory(self, revision_id):
778
"""Return inventory of a past revision."""
779
# TODO: Unify this with get_inventory()
780
# bzr 0.0.6 and later imposes the constraint that the inventory_id
781
# must be the same as its revision, so this is trivial.
782
if revision_id is None:
783
# This does not make sense: if there is no revision,
784
# then it is the current tree inventory surely ?!
785
# and thus get_root_id() is something that looks at the last
786
# commit on the branch, and the get_root_id is an inventory check.
787
raise NotImplementedError
788
# return Inventory(self.get_root_id())
790
return self.get_inventory(revision_id)
794
"""Return True if this repository is flagged as a shared repository."""
795
raise NotImplementedError(self.is_shared)
798
def reconcile(self, other=None, thorough=False):
799
"""Reconcile this repository."""
800
from bzrlib.reconcile import RepoReconciler
801
reconciler = RepoReconciler(self, thorough=thorough)
802
reconciler.reconcile()
806
def revision_tree(self, revision_id):
807
"""Return Tree for a revision on this branch.
809
`revision_id` may be None for the empty tree revision.
811
# TODO: refactor this to use an existing revision object
812
# so we don't need to read it in twice.
813
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
814
return RevisionTree(self, Inventory(root_id=None),
815
_mod_revision.NULL_REVISION)
817
revision_id = osutils.safe_revision_id(revision_id)
818
inv = self.get_revision_inventory(revision_id)
819
return RevisionTree(self, inv, revision_id)
822
def revision_trees(self, revision_ids):
823
"""Return Tree for a revision on this branch.
825
`revision_id` may not be None or 'null:'"""
826
assert None not in revision_ids
827
assert _mod_revision.NULL_REVISION not in revision_ids
828
texts = self.get_inventory_weave().get_texts(revision_ids)
829
for text, revision_id in zip(texts, revision_ids):
830
inv = self.deserialise_inventory(revision_id, text)
831
yield RevisionTree(self, inv, revision_id)
834
def get_ancestry(self, revision_id, topo_sorted=True):
835
"""Return a list of revision-ids integrated by a revision.
837
The first element of the list is always None, indicating the origin
838
revision. This might change when we have history horizons, or
839
perhaps we should have a new API.
841
This is topologically sorted.
843
if _mod_revision.is_null(revision_id):
845
revision_id = osutils.safe_revision_id(revision_id)
846
if not self.has_revision(revision_id):
847
raise errors.NoSuchRevision(self, revision_id)
848
w = self.get_inventory_weave()
849
candidates = w.get_ancestry(revision_id, topo_sorted)
850
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
853
"""Compress the data within the repository.
855
This operation only makes sense for some repository types. For other
856
types it should be a no-op that just returns.
858
This stub method does not require a lock, but subclasses should use
859
@needs_write_lock as this is a long running call its reasonable to
860
implicitly lock for the user.
864
def print_file(self, file, revision_id):
865
"""Print `file` to stdout.
867
FIXME RBC 20060125 as John Meinel points out this is a bad api
868
- it writes to stdout, it assumes that that is valid etc. Fix
869
by creating a new more flexible convenience function.
871
revision_id = osutils.safe_revision_id(revision_id)
872
tree = self.revision_tree(revision_id)
873
# use inventory as it was in that revision
874
file_id = tree.inventory.path2id(file)
876
# TODO: jam 20060427 Write a test for this code path
877
# it had a bug in it, and was raising the wrong
879
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
880
tree.print_file(file_id)
882
def get_transaction(self):
883
return self.control_files.get_transaction()
885
def revision_parents(self, revision_id):
886
revision_id = osutils.safe_revision_id(revision_id)
887
return self.get_inventory_weave().parent_names(revision_id)
889
def get_parents(self, revision_ids):
890
"""See StackedParentsProvider.get_parents"""
892
for revision_id in revision_ids:
893
if revision_id == _mod_revision.NULL_REVISION:
897
parents = self.get_revision(revision_id).parent_ids
898
except errors.NoSuchRevision:
901
if len(parents) == 0:
902
parents = [_mod_revision.NULL_REVISION]
903
parents_list.append(parents)
906
def _make_parents_provider(self):
909
def get_graph(self, other_repository=None):
910
"""Return the graph walker for this repository format"""
911
parents_provider = self._make_parents_provider()
912
if (other_repository is not None and
913
other_repository.bzrdir.transport.base !=
914
self.bzrdir.transport.base):
915
parents_provider = graph._StackedParentsProvider(
916
[parents_provider, other_repository._make_parents_provider()])
917
return graph.Graph(parents_provider)
920
def set_make_working_trees(self, new_value):
921
"""Set the policy flag for making working trees when creating branches.
923
This only applies to branches that use this repository.
925
The default is 'True'.
926
:param new_value: True to restore the default, False to disable making
929
raise NotImplementedError(self.set_make_working_trees)
931
def make_working_trees(self):
932
"""Returns the policy for making working trees on new branches."""
933
raise NotImplementedError(self.make_working_trees)
936
def sign_revision(self, revision_id, gpg_strategy):
937
revision_id = osutils.safe_revision_id(revision_id)
938
plaintext = Testament.from_revision(self, revision_id).as_short_text()
939
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
942
def has_signature_for_revision_id(self, revision_id):
943
"""Query for a revision signature for revision_id in the repository."""
944
revision_id = osutils.safe_revision_id(revision_id)
945
return self._revision_store.has_signature(revision_id,
946
self.get_transaction())
949
def get_signature_text(self, revision_id):
950
"""Return the text for a signature."""
951
revision_id = osutils.safe_revision_id(revision_id)
952
return self._revision_store.get_signature_text(revision_id,
953
self.get_transaction())
956
def check(self, revision_ids):
957
"""Check consistency of all history of given revision_ids.
959
Different repository implementations should override _check().
961
:param revision_ids: A non-empty list of revision_ids whose ancestry
962
will be checked. Typically the last revision_id of a branch.
965
raise ValueError("revision_ids must be non-empty in %s.check"
967
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
968
return self._check(revision_ids)
970
def _check(self, revision_ids):
971
result = check.Check(self)
975
def _warn_if_deprecated(self):
976
global _deprecation_warning_done
977
if _deprecation_warning_done:
979
_deprecation_warning_done = True
980
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
981
% (self._format, self.bzrdir.transport.base))
983
def supports_rich_root(self):
984
return self._format.rich_root_data
986
def _check_ascii_revisionid(self, revision_id, method):
987
"""Private helper for ascii-only repositories."""
988
# weave repositories refuse to store revisionids that are non-ascii.
989
if revision_id is not None:
990
# weaves require ascii revision ids.
991
if isinstance(revision_id, unicode):
993
revision_id.encode('ascii')
994
except UnicodeEncodeError:
995
raise errors.NonAsciiRevisionId(method, self)
998
revision_id.decode('ascii')
999
except UnicodeDecodeError:
1000
raise errors.NonAsciiRevisionId(method, self)
1004
# remove these delegates a while after bzr 0.15
1005
def __make_delegated(name, from_module):
1006
def _deprecated_repository_forwarder():
1007
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1008
% (name, from_module),
1011
m = __import__(from_module, globals(), locals(), [name])
1013
return getattr(m, name)
1014
except AttributeError:
1015
raise AttributeError('module %s has no name %s'
1017
globals()[name] = _deprecated_repository_forwarder
1020
'AllInOneRepository',
1021
'WeaveMetaDirRepository',
1022
'PreSplitOutRepositoryFormat',
1023
'RepositoryFormat4',
1024
'RepositoryFormat5',
1025
'RepositoryFormat6',
1026
'RepositoryFormat7',
1028
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1032
'RepositoryFormatKnit',
1033
'RepositoryFormatKnit1',
1035
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1038
def install_revision(repository, rev, revision_tree):
1039
"""Install all revision data into a repository."""
1040
present_parents = []
1042
for p_id in rev.parent_ids:
1043
if repository.has_revision(p_id):
1044
present_parents.append(p_id)
1045
parent_trees[p_id] = repository.revision_tree(p_id)
1047
parent_trees[p_id] = repository.revision_tree(None)
1049
inv = revision_tree.inventory
1050
entries = inv.iter_entries()
1051
# backwards compatability hack: skip the root id.
1052
if not repository.supports_rich_root():
1053
path, root = entries.next()
1054
if root.revision != rev.revision_id:
1055
raise errors.IncompatibleRevision(repr(repository))
1056
# Add the texts that are not already present
1057
for path, ie in entries:
1058
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1059
repository.get_transaction())
1060
if ie.revision not in w:
1062
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1063
# with InventoryEntry.find_previous_heads(). if it is, then there
1064
# is a latent bug here where the parents may have ancestors of each
1066
for revision, tree in parent_trees.iteritems():
1067
if ie.file_id not in tree:
1069
parent_id = tree.inventory[ie.file_id].revision
1070
if parent_id in text_parents:
1072
text_parents.append(parent_id)
1074
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1075
repository.get_transaction())
1076
lines = revision_tree.get_file(ie.file_id).readlines()
1077
vfile.add_lines(rev.revision_id, text_parents, lines)
1079
# install the inventory
1080
repository.add_inventory(rev.revision_id, inv, present_parents)
1081
except errors.RevisionAlreadyPresent:
1083
repository.add_revision(rev.revision_id, rev, inv)
1086
class MetaDirRepository(Repository):
1087
"""Repositories in the new meta-dir layout."""
1089
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1090
super(MetaDirRepository, self).__init__(_format,
1096
dir_mode = self.control_files._dir_mode
1097
file_mode = self.control_files._file_mode
1100
def is_shared(self):
1101
"""Return True if this repository is flagged as a shared repository."""
1102
return self.control_files._transport.has('shared-storage')
1105
def set_make_working_trees(self, new_value):
1106
"""Set the policy flag for making working trees when creating branches.
1108
This only applies to branches that use this repository.
1110
The default is 'True'.
1111
:param new_value: True to restore the default, False to disable making
1116
self.control_files._transport.delete('no-working-trees')
1117
except errors.NoSuchFile:
1120
self.control_files.put_utf8('no-working-trees', '')
1122
def make_working_trees(self):
1123
"""Returns the policy for making working trees on new branches."""
1124
return not self.control_files._transport.has('no-working-trees')
1127
class RepositoryFormatRegistry(registry.Registry):
1128
"""Registry of RepositoryFormats.
1131
def get(self, format_string):
1132
r = registry.Registry.get(self, format_string)
1138
format_registry = RepositoryFormatRegistry()
1139
"""Registry of formats, indexed by their identifying format string.
1141
This can contain either format instances themselves, or classes/factories that
1142
can be called to obtain one.
1146
#####################################################################
1147
# Repository Formats
1149
class RepositoryFormat(object):
1150
"""A repository format.
1152
Formats provide three things:
1153
* An initialization routine to construct repository data on disk.
1154
* a format string which is used when the BzrDir supports versioned
1156
* an open routine which returns a Repository instance.
1158
Formats are placed in an dict by their format string for reference
1159
during opening. These should be subclasses of RepositoryFormat
1162
Once a format is deprecated, just deprecate the initialize and open
1163
methods on the format class. Do not deprecate the object, as the
1164
object will be created every system load.
1166
Common instance attributes:
1167
_matchingbzrdir - the bzrdir format that the repository format was
1168
originally written to work with. This can be used if manually
1169
constructing a bzrdir and repository, or more commonly for test suite
1174
return "<%s>" % self.__class__.__name__
1176
def __eq__(self, other):
1177
# format objects are generally stateless
1178
return isinstance(other, self.__class__)
1180
def __ne__(self, other):
1181
return not self == other
1184
def find_format(klass, a_bzrdir):
1185
"""Return the format for the repository object in a_bzrdir.
1187
This is used by bzr native formats that have a "format" file in
1188
the repository. Other methods may be used by different types of
1192
transport = a_bzrdir.get_repository_transport(None)
1193
format_string = transport.get("format").read()
1194
return format_registry.get(format_string)
1195
except errors.NoSuchFile:
1196
raise errors.NoRepositoryPresent(a_bzrdir)
1198
raise errors.UnknownFormatError(format=format_string)
1201
def register_format(klass, format):
1202
format_registry.register(format.get_format_string(), format)
1205
def unregister_format(klass, format):
1206
format_registry.remove(format.get_format_string())
1209
def get_default_format(klass):
1210
"""Return the current default format."""
1211
from bzrlib import bzrdir
1212
return bzrdir.format_registry.make_bzrdir('default').repository_format
1214
def _get_control_store(self, repo_transport, control_files):
1215
"""Return the control store for this repository."""
1216
raise NotImplementedError(self._get_control_store)
1218
def get_format_string(self):
1219
"""Return the ASCII format string that identifies this format.
1221
Note that in pre format ?? repositories the format string is
1222
not permitted nor written to disk.
1224
raise NotImplementedError(self.get_format_string)
1226
def get_format_description(self):
1227
"""Return the short description for this format."""
1228
raise NotImplementedError(self.get_format_description)
1230
def _get_revision_store(self, repo_transport, control_files):
1231
"""Return the revision store object for this a_bzrdir."""
1232
raise NotImplementedError(self._get_revision_store)
1234
def _get_text_rev_store(self,
1241
"""Common logic for getting a revision store for a repository.
1243
see self._get_revision_store for the subclass-overridable method to
1244
get the store for a repository.
1246
from bzrlib.store.revision.text import TextRevisionStore
1247
dir_mode = control_files._dir_mode
1248
file_mode = control_files._file_mode
1249
text_store = TextStore(transport.clone(name),
1251
compressed=compressed,
1253
file_mode=file_mode)
1254
_revision_store = TextRevisionStore(text_store, serializer)
1255
return _revision_store
1257
# TODO: this shouldn't be in the base class, it's specific to things that
1258
# use weaves or knits -- mbp 20070207
1259
def _get_versioned_file_store(self,
1264
versionedfile_class=None,
1265
versionedfile_kwargs={},
1267
if versionedfile_class is None:
1268
versionedfile_class = self._versionedfile_class
1269
weave_transport = control_files._transport.clone(name)
1270
dir_mode = control_files._dir_mode
1271
file_mode = control_files._file_mode
1272
return VersionedFileStore(weave_transport, prefixed=prefixed,
1274
file_mode=file_mode,
1275
versionedfile_class=versionedfile_class,
1276
versionedfile_kwargs=versionedfile_kwargs,
1279
def initialize(self, a_bzrdir, shared=False):
1280
"""Initialize a repository of this format in a_bzrdir.
1282
:param a_bzrdir: The bzrdir to put the new repository in it.
1283
:param shared: The repository should be initialized as a sharable one.
1284
:returns: The new repository object.
1286
This may raise UninitializableFormat if shared repository are not
1287
compatible the a_bzrdir.
1289
raise NotImplementedError(self.initialize)
1291
def is_supported(self):
1292
"""Is this format supported?
1294
Supported formats must be initializable and openable.
1295
Unsupported formats may not support initialization or committing or
1296
some other features depending on the reason for not being supported.
1300
def check_conversion_target(self, target_format):
1301
raise NotImplementedError(self.check_conversion_target)
1303
def open(self, a_bzrdir, _found=False):
1304
"""Return an instance of this format for the bzrdir a_bzrdir.
1306
_found is a private parameter, do not use it.
1308
raise NotImplementedError(self.open)
1311
class MetaDirRepositoryFormat(RepositoryFormat):
1312
"""Common base class for the new repositories using the metadir layout."""
1314
rich_root_data = False
1315
supports_tree_reference = False
1316
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1319
super(MetaDirRepositoryFormat, self).__init__()
1321
def _create_control_files(self, a_bzrdir):
1322
"""Create the required files and the initial control_files object."""
1323
# FIXME: RBC 20060125 don't peek under the covers
1324
# NB: no need to escape relative paths that are url safe.
1325
repository_transport = a_bzrdir.get_repository_transport(self)
1326
control_files = lockable_files.LockableFiles(repository_transport,
1327
'lock', lockdir.LockDir)
1328
control_files.create_lock()
1329
return control_files
1331
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1332
"""Upload the initial blank content."""
1333
control_files = self._create_control_files(a_bzrdir)
1334
control_files.lock_write()
1336
control_files._transport.mkdir_multi(dirs,
1337
mode=control_files._dir_mode)
1338
for file, content in files:
1339
control_files.put(file, content)
1340
for file, content in utf8_files:
1341
control_files.put_utf8(file, content)
1343
control_files.put_utf8('shared-storage', '')
1345
control_files.unlock()
1348
# formats which have no format string are not discoverable
1349
# and not independently creatable, so are not registered. They're
1350
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1351
# needed, it's constructed directly by the BzrDir. Non-native formats where
1352
# the repository is not separately opened are similar.
1354
format_registry.register_lazy(
1355
'Bazaar-NG Repository format 7',
1356
'bzrlib.repofmt.weaverepo',
1359
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1360
# default control directory format
1362
format_registry.register_lazy(
1363
'Bazaar-NG Knit Repository Format 1',
1364
'bzrlib.repofmt.knitrepo',
1365
'RepositoryFormatKnit1',
1367
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1369
format_registry.register_lazy(
1370
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1371
'bzrlib.repofmt.knitrepo',
1372
'RepositoryFormatKnit3',
1376
class InterRepository(InterObject):
1377
"""This class represents operations taking place between two repositories.
1379
Its instances have methods like copy_content and fetch, and contain
1380
references to the source and target repositories these operations can be
1383
Often we will provide convenience methods on 'repository' which carry out
1384
operations with another repository - they will always forward to
1385
InterRepository.get(other).method_name(parameters).
1389
"""The available optimised InterRepository types."""
1391
def copy_content(self, revision_id=None):
1392
raise NotImplementedError(self.copy_content)
1394
def fetch(self, revision_id=None, pb=None):
1395
"""Fetch the content required to construct revision_id.
1397
The content is copied from self.source to self.target.
1399
:param revision_id: if None all content is copied, if NULL_REVISION no
1401
:param pb: optional progress bar to use for progress reports. If not
1402
provided a default one will be created.
1404
Returns the copied revision count and the failed revisions in a tuple:
1407
raise NotImplementedError(self.fetch)
1410
def missing_revision_ids(self, revision_id=None):
1411
"""Return the revision ids that source has that target does not.
1413
These are returned in topological order.
1415
:param revision_id: only return revision ids included by this
1418
# generic, possibly worst case, slow code path.
1419
target_ids = set(self.target.all_revision_ids())
1420
if revision_id is not None:
1421
# TODO: jam 20070210 InterRepository is internal enough that it
1422
# should assume revision_ids are already utf-8
1423
revision_id = osutils.safe_revision_id(revision_id)
1424
source_ids = self.source.get_ancestry(revision_id)
1425
assert source_ids[0] is None
1428
source_ids = self.source.all_revision_ids()
1429
result_set = set(source_ids).difference(target_ids)
1430
# this may look like a no-op: its not. It preserves the ordering
1431
# other_ids had while only returning the members from other_ids
1432
# that we've decided we need.
1433
return [rev_id for rev_id in source_ids if rev_id in result_set]
1436
class InterSameDataRepository(InterRepository):
1437
"""Code for converting between repositories that represent the same data.
1439
Data format and model must match for this to work.
1443
def _get_repo_format_to_test(self):
1444
"""Repository format for testing with."""
1445
return RepositoryFormat.get_default_format()
1448
def is_compatible(source, target):
1449
if source.supports_rich_root() != target.supports_rich_root():
1451
if source._serializer != target._serializer:
1456
def copy_content(self, revision_id=None):
1457
"""Make a complete copy of the content in self into destination.
1459
This copies both the repository's revision data, and configuration information
1460
such as the make_working_trees setting.
1462
This is a destructive operation! Do not use it on existing
1465
:param revision_id: Only copy the content needed to construct
1466
revision_id and its parents.
1469
self.target.set_make_working_trees(self.source.make_working_trees())
1470
except NotImplementedError:
1472
# TODO: jam 20070210 This is fairly internal, so we should probably
1473
# just assert that revision_id is not unicode.
1474
revision_id = osutils.safe_revision_id(revision_id)
1475
# but don't bother fetching if we have the needed data now.
1476
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1477
self.target.has_revision(revision_id)):
1479
self.target.fetch(self.source, revision_id=revision_id)
1482
def fetch(self, revision_id=None, pb=None):
1483
"""See InterRepository.fetch()."""
1484
from bzrlib.fetch import GenericRepoFetcher
1485
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1486
self.source, self.source._format, self.target,
1487
self.target._format)
1488
# TODO: jam 20070210 This should be an assert, not a translate
1489
revision_id = osutils.safe_revision_id(revision_id)
1490
f = GenericRepoFetcher(to_repository=self.target,
1491
from_repository=self.source,
1492
last_revision=revision_id,
1494
return f.count_copied, f.failed_revisions
1497
class InterWeaveRepo(InterSameDataRepository):
1498
"""Optimised code paths between Weave based repositories."""
1501
def _get_repo_format_to_test(self):
1502
from bzrlib.repofmt import weaverepo
1503
return weaverepo.RepositoryFormat7()
1506
def is_compatible(source, target):
1507
"""Be compatible with known Weave formats.
1509
We don't test for the stores being of specific types because that
1510
could lead to confusing results, and there is no need to be
1513
from bzrlib.repofmt.weaverepo import (
1519
return (isinstance(source._format, (RepositoryFormat5,
1521
RepositoryFormat7)) and
1522
isinstance(target._format, (RepositoryFormat5,
1524
RepositoryFormat7)))
1525
except AttributeError:
1529
def copy_content(self, revision_id=None):
1530
"""See InterRepository.copy_content()."""
1531
# weave specific optimised path:
1532
# TODO: jam 20070210 Internal, should be an assert, not translate
1533
revision_id = osutils.safe_revision_id(revision_id)
1535
self.target.set_make_working_trees(self.source.make_working_trees())
1536
except NotImplementedError:
1538
# FIXME do not peek!
1539
if self.source.control_files._transport.listable():
1540
pb = ui.ui_factory.nested_progress_bar()
1542
self.target.weave_store.copy_all_ids(
1543
self.source.weave_store,
1545
from_transaction=self.source.get_transaction(),
1546
to_transaction=self.target.get_transaction())
1547
pb.update('copying inventory', 0, 1)
1548
self.target.control_weaves.copy_multi(
1549
self.source.control_weaves, ['inventory'],
1550
from_transaction=self.source.get_transaction(),
1551
to_transaction=self.target.get_transaction())
1552
self.target._revision_store.text_store.copy_all_ids(
1553
self.source._revision_store.text_store,
1558
self.target.fetch(self.source, revision_id=revision_id)
1561
def fetch(self, revision_id=None, pb=None):
1562
"""See InterRepository.fetch()."""
1563
from bzrlib.fetch import GenericRepoFetcher
1564
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1565
self.source, self.source._format, self.target, self.target._format)
1566
# TODO: jam 20070210 This should be an assert, not a translate
1567
revision_id = osutils.safe_revision_id(revision_id)
1568
f = GenericRepoFetcher(to_repository=self.target,
1569
from_repository=self.source,
1570
last_revision=revision_id,
1572
return f.count_copied, f.failed_revisions
1575
def missing_revision_ids(self, revision_id=None):
1576
"""See InterRepository.missing_revision_ids()."""
1577
# we want all revisions to satisfy revision_id in source.
1578
# but we don't want to stat every file here and there.
1579
# we want then, all revisions other needs to satisfy revision_id
1580
# checked, but not those that we have locally.
1581
# so the first thing is to get a subset of the revisions to
1582
# satisfy revision_id in source, and then eliminate those that
1583
# we do already have.
1584
# this is slow on high latency connection to self, but as as this
1585
# disk format scales terribly for push anyway due to rewriting
1586
# inventory.weave, this is considered acceptable.
1588
if revision_id is not None:
1589
source_ids = self.source.get_ancestry(revision_id)
1590
assert source_ids[0] is None
1593
source_ids = self.source._all_possible_ids()
1594
source_ids_set = set(source_ids)
1595
# source_ids is the worst possible case we may need to pull.
1596
# now we want to filter source_ids against what we actually
1597
# have in target, but don't try to check for existence where we know
1598
# we do not have a revision as that would be pointless.
1599
target_ids = set(self.target._all_possible_ids())
1600
possibly_present_revisions = target_ids.intersection(source_ids_set)
1601
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1602
required_revisions = source_ids_set.difference(actually_present_revisions)
1603
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1604
if revision_id is not None:
1605
# we used get_ancestry to determine source_ids then we are assured all
1606
# revisions referenced are present as they are installed in topological order.
1607
# and the tip revision was validated by get_ancestry.
1608
return required_topo_revisions
1610
# if we just grabbed the possibly available ids, then
1611
# we only have an estimate of whats available and need to validate
1612
# that against the revision records.
1613
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1616
class InterKnitRepo(InterSameDataRepository):
1617
"""Optimised code paths between Knit based repositories."""
1620
def _get_repo_format_to_test(self):
1621
from bzrlib.repofmt import knitrepo
1622
return knitrepo.RepositoryFormatKnit1()
1625
def is_compatible(source, target):
1626
"""Be compatible with known Knit formats.
1628
We don't test for the stores being of specific types because that
1629
could lead to confusing results, and there is no need to be
1632
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1634
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1635
isinstance(target._format, (RepositoryFormatKnit1)))
1636
except AttributeError:
1640
def fetch(self, revision_id=None, pb=None):
1641
"""See InterRepository.fetch()."""
1642
from bzrlib.fetch import KnitRepoFetcher
1643
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1644
self.source, self.source._format, self.target, self.target._format)
1645
# TODO: jam 20070210 This should be an assert, not a translate
1646
revision_id = osutils.safe_revision_id(revision_id)
1647
f = KnitRepoFetcher(to_repository=self.target,
1648
from_repository=self.source,
1649
last_revision=revision_id,
1651
return f.count_copied, f.failed_revisions
1654
def missing_revision_ids(self, revision_id=None):
1655
"""See InterRepository.missing_revision_ids()."""
1656
if revision_id is not None:
1657
source_ids = self.source.get_ancestry(revision_id)
1658
assert source_ids[0] is None
1661
source_ids = self.source._all_possible_ids()
1662
source_ids_set = set(source_ids)
1663
# source_ids is the worst possible case we may need to pull.
1664
# now we want to filter source_ids against what we actually
1665
# have in target, but don't try to check for existence where we know
1666
# we do not have a revision as that would be pointless.
1667
target_ids = set(self.target._all_possible_ids())
1668
possibly_present_revisions = target_ids.intersection(source_ids_set)
1669
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1670
required_revisions = source_ids_set.difference(actually_present_revisions)
1671
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1672
if revision_id is not None:
1673
# we used get_ancestry to determine source_ids then we are assured all
1674
# revisions referenced are present as they are installed in topological order.
1675
# and the tip revision was validated by get_ancestry.
1676
return required_topo_revisions
1678
# if we just grabbed the possibly available ids, then
1679
# we only have an estimate of whats available and need to validate
1680
# that against the revision records.
1681
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1684
class InterModel1and2(InterRepository):
1687
def _get_repo_format_to_test(self):
1691
def is_compatible(source, target):
1692
if not source.supports_rich_root() and target.supports_rich_root():
1698
def fetch(self, revision_id=None, pb=None):
1699
"""See InterRepository.fetch()."""
1700
from bzrlib.fetch import Model1toKnit2Fetcher
1701
# TODO: jam 20070210 This should be an assert, not a translate
1702
revision_id = osutils.safe_revision_id(revision_id)
1703
f = Model1toKnit2Fetcher(to_repository=self.target,
1704
from_repository=self.source,
1705
last_revision=revision_id,
1707
return f.count_copied, f.failed_revisions
1710
def copy_content(self, revision_id=None):
1711
"""Make a complete copy of the content in self into destination.
1713
This is a destructive operation! Do not use it on existing
1716
:param revision_id: Only copy the content needed to construct
1717
revision_id and its parents.
1720
self.target.set_make_working_trees(self.source.make_working_trees())
1721
except NotImplementedError:
1723
# TODO: jam 20070210 Internal, assert, don't translate
1724
revision_id = osutils.safe_revision_id(revision_id)
1725
# but don't bother fetching if we have the needed data now.
1726
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1727
self.target.has_revision(revision_id)):
1729
self.target.fetch(self.source, revision_id=revision_id)
1732
class InterKnit1and2(InterKnitRepo):
1735
def _get_repo_format_to_test(self):
1739
def is_compatible(source, target):
1740
"""Be compatible with Knit1 source and Knit3 target"""
1741
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1743
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1744
RepositoryFormatKnit3
1745
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1746
isinstance(target._format, (RepositoryFormatKnit3)))
1747
except AttributeError:
1751
def fetch(self, revision_id=None, pb=None):
1752
"""See InterRepository.fetch()."""
1753
from bzrlib.fetch import Knit1to2Fetcher
1754
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1755
self.source, self.source._format, self.target,
1756
self.target._format)
1757
# TODO: jam 20070210 This should be an assert, not a translate
1758
revision_id = osutils.safe_revision_id(revision_id)
1759
f = Knit1to2Fetcher(to_repository=self.target,
1760
from_repository=self.source,
1761
last_revision=revision_id,
1763
return f.count_copied, f.failed_revisions
1766
class InterRemoteRepository(InterRepository):
1767
"""Code for converting between RemoteRepository objects.
1769
This just gets an non-remote repository from the RemoteRepository, and calls
1770
InterRepository.get again.
1773
def __init__(self, source, target):
1774
if isinstance(source, remote.RemoteRepository):
1775
source._ensure_real()
1776
real_source = source._real_repository
1778
real_source = source
1779
if isinstance(target, remote.RemoteRepository):
1780
target._ensure_real()
1781
real_target = target._real_repository
1783
real_target = target
1784
self.real_inter = InterRepository.get(real_source, real_target)
1787
def is_compatible(source, target):
1788
if isinstance(source, remote.RemoteRepository):
1790
if isinstance(target, remote.RemoteRepository):
1794
def copy_content(self, revision_id=None):
1795
self.real_inter.copy_content(revision_id=revision_id)
1797
def fetch(self, revision_id=None, pb=None):
1798
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1801
def _get_repo_format_to_test(self):
1805
InterRepository.register_optimiser(InterSameDataRepository)
1806
InterRepository.register_optimiser(InterWeaveRepo)
1807
InterRepository.register_optimiser(InterKnitRepo)
1808
InterRepository.register_optimiser(InterModel1and2)
1809
InterRepository.register_optimiser(InterKnit1and2)
1810
InterRepository.register_optimiser(InterRemoteRepository)
1813
class CopyConverter(object):
1814
"""A repository conversion tool which just performs a copy of the content.
1816
This is slow but quite reliable.
1819
def __init__(self, target_format):
1820
"""Create a CopyConverter.
1822
:param target_format: The format the resulting repository should be.
1824
self.target_format = target_format
1826
def convert(self, repo, pb):
1827
"""Perform the conversion of to_convert, giving feedback via pb.
1829
:param to_convert: The disk object to convert.
1830
:param pb: a progress bar to use for progress information.
1835
# this is only useful with metadir layouts - separated repo content.
1836
# trigger an assertion if not such
1837
repo._format.get_format_string()
1838
self.repo_dir = repo.bzrdir
1839
self.step('Moving repository to repository.backup')
1840
self.repo_dir.transport.move('repository', 'repository.backup')
1841
backup_transport = self.repo_dir.transport.clone('repository.backup')
1842
repo._format.check_conversion_target(self.target_format)
1843
self.source_repo = repo._format.open(self.repo_dir,
1845
_override_transport=backup_transport)
1846
self.step('Creating new repository')
1847
converted = self.target_format.initialize(self.repo_dir,
1848
self.source_repo.is_shared())
1849
converted.lock_write()
1851
self.step('Copying content into repository.')
1852
self.source_repo.copy_content_into(converted)
1855
self.step('Deleting old repository content.')
1856
self.repo_dir.transport.delete_tree('repository.backup')
1857
self.pb.note('repository converted')
1859
def step(self, message):
1860
"""Update the pb by a step."""
1862
self.pb.update(message, self.count, self.total)
1865
class CommitBuilder(object):
1866
"""Provides an interface to build up a commit.
1868
This allows describing a tree to be committed without needing to
1869
know the internals of the format of the repository.
1872
record_root_entry = False
1873
def __init__(self, repository, parents, config, timestamp=None,
1874
timezone=None, committer=None, revprops=None,
1876
"""Initiate a CommitBuilder.
1878
:param repository: Repository to commit to.
1879
:param parents: Revision ids of the parents of the new revision.
1880
:param config: Configuration to use.
1881
:param timestamp: Optional timestamp recorded for commit.
1882
:param timezone: Optional timezone for timestamp.
1883
:param committer: Optional committer to set for commit.
1884
:param revprops: Optional dictionary of revision properties.
1885
:param revision_id: Optional revision id.
1887
self._config = config
1889
if committer is None:
1890
self._committer = self._config.username()
1892
assert isinstance(committer, basestring), type(committer)
1893
self._committer = committer
1895
self.new_inventory = Inventory(None)
1896
self._new_revision_id = osutils.safe_revision_id(revision_id)
1897
self.parents = parents
1898
self.repository = repository
1901
if revprops is not None:
1902
self._revprops.update(revprops)
1904
if timestamp is None:
1905
timestamp = time.time()
1906
# Restrict resolution to 1ms
1907
self._timestamp = round(timestamp, 3)
1909
if timezone is None:
1910
self._timezone = osutils.local_time_offset()
1912
self._timezone = int(timezone)
1914
self._generate_revision_if_needed()
1916
def commit(self, message):
1917
"""Make the actual commit.
1919
:return: The revision id of the recorded revision.
1921
rev = _mod_revision.Revision(
1922
timestamp=self._timestamp,
1923
timezone=self._timezone,
1924
committer=self._committer,
1926
inventory_sha1=self.inv_sha1,
1927
revision_id=self._new_revision_id,
1928
properties=self._revprops)
1929
rev.parent_ids = self.parents
1930
self.repository.add_revision(self._new_revision_id, rev,
1931
self.new_inventory, self._config)
1932
return self._new_revision_id
1934
def revision_tree(self):
1935
"""Return the tree that was just committed.
1937
After calling commit() this can be called to get a RevisionTree
1938
representing the newly committed tree. This is preferred to
1939
calling Repository.revision_tree() because that may require
1940
deserializing the inventory, while we already have a copy in
1943
return RevisionTree(self.repository, self.new_inventory,
1944
self._new_revision_id)
1946
def finish_inventory(self):
1947
"""Tell the builder that the inventory is finished."""
1948
if self.new_inventory.root is None:
1949
symbol_versioning.warn('Root entry should be supplied to'
1950
' record_entry_contents, as of bzr 0.10.',
1951
DeprecationWarning, stacklevel=2)
1952
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1953
self.new_inventory.revision_id = self._new_revision_id
1954
self.inv_sha1 = self.repository.add_inventory(
1955
self._new_revision_id,
1960
def _gen_revision_id(self):
1961
"""Return new revision-id."""
1962
return generate_ids.gen_revision_id(self._config.username(),
1965
def _generate_revision_if_needed(self):
1966
"""Create a revision id if None was supplied.
1968
If the repository can not support user-specified revision ids
1969
they should override this function and raise CannotSetRevisionId
1970
if _new_revision_id is not None.
1972
:raises: CannotSetRevisionId
1974
if self._new_revision_id is None:
1975
self._new_revision_id = self._gen_revision_id()
1977
def record_entry_contents(self, ie, parent_invs, path, tree):
1978
"""Record the content of ie from tree into the commit if needed.
1980
Side effect: sets ie.revision when unchanged
1982
:param ie: An inventory entry present in the commit.
1983
:param parent_invs: The inventories of the parent revisions of the
1985
:param path: The path the entry is at in the tree.
1986
:param tree: The tree which contains this entry and should be used to
1989
if self.new_inventory.root is None and ie.parent_id is not None:
1990
symbol_versioning.warn('Root entry should be supplied to'
1991
' record_entry_contents, as of bzr 0.10.',
1992
DeprecationWarning, stacklevel=2)
1993
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1995
self.new_inventory.add(ie)
1997
# ie.revision is always None if the InventoryEntry is considered
1998
# for committing. ie.snapshot will record the correct revision
1999
# which may be the sole parent if it is untouched.
2000
if ie.revision is not None:
2003
# In this revision format, root entries have no knit or weave
2004
if ie is self.new_inventory.root:
2005
# When serializing out to disk and back in
2006
# root.revision is always _new_revision_id
2007
ie.revision = self._new_revision_id
2009
previous_entries = ie.find_previous_heads(
2011
self.repository.weave_store,
2012
self.repository.get_transaction())
2013
# we are creating a new revision for ie in the history store
2015
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2017
def modified_directory(self, file_id, file_parents):
2018
"""Record the presence of a symbolic link.
2020
:param file_id: The file_id of the link to record.
2021
:param file_parents: The per-file parent revision ids.
2023
self._add_text_to_weave(file_id, [], file_parents.keys())
2025
def modified_reference(self, file_id, file_parents):
2026
"""Record the modification of a reference.
2028
:param file_id: The file_id of the link to record.
2029
:param file_parents: The per-file parent revision ids.
2031
self._add_text_to_weave(file_id, [], file_parents.keys())
2033
def modified_file_text(self, file_id, file_parents,
2034
get_content_byte_lines, text_sha1=None,
2036
"""Record the text of file file_id
2038
:param file_id: The file_id of the file to record the text of.
2039
:param file_parents: The per-file parent revision ids.
2040
:param get_content_byte_lines: A callable which will return the byte
2042
:param text_sha1: Optional SHA1 of the file contents.
2043
:param text_size: Optional size of the file contents.
2045
# mutter('storing text of file {%s} in revision {%s} into %r',
2046
# file_id, self._new_revision_id, self.repository.weave_store)
2047
# special case to avoid diffing on renames or
2049
if (len(file_parents) == 1
2050
and text_sha1 == file_parents.values()[0].text_sha1
2051
and text_size == file_parents.values()[0].text_size):
2052
previous_ie = file_parents.values()[0]
2053
versionedfile = self.repository.weave_store.get_weave(file_id,
2054
self.repository.get_transaction())
2055
versionedfile.clone_text(self._new_revision_id,
2056
previous_ie.revision, file_parents.keys())
2057
return text_sha1, text_size
2059
new_lines = get_content_byte_lines()
2060
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2061
# should return the SHA1 and size
2062
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2063
return osutils.sha_strings(new_lines), \
2064
sum(map(len, new_lines))
2066
def modified_link(self, file_id, file_parents, link_target):
2067
"""Record the presence of a symbolic link.
2069
:param file_id: The file_id of the link to record.
2070
:param file_parents: The per-file parent revision ids.
2071
:param link_target: Target location of this link.
2073
self._add_text_to_weave(file_id, [], file_parents.keys())
2075
def _add_text_to_weave(self, file_id, new_lines, parents):
2076
versionedfile = self.repository.weave_store.get_weave_or_empty(
2077
file_id, self.repository.get_transaction())
2078
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2079
versionedfile.clear_cache()
2082
class _CommitBuilder(CommitBuilder):
2083
"""Temporary class so old CommitBuilders are detected properly
2085
Note: CommitBuilder works whether or not root entry is recorded.
2088
record_root_entry = True
2091
class RootCommitBuilder(CommitBuilder):
2092
"""This commitbuilder actually records the root id"""
2094
record_root_entry = True
2096
def record_entry_contents(self, ie, parent_invs, path, tree):
2097
"""Record the content of ie from tree into the commit if needed.
2099
Side effect: sets ie.revision when unchanged
2101
:param ie: An inventory entry present in the commit.
2102
:param parent_invs: The inventories of the parent revisions of the
2104
:param path: The path the entry is at in the tree.
2105
:param tree: The tree which contains this entry and should be used to
2108
assert self.new_inventory.root is not None or ie.parent_id is None
2109
self.new_inventory.add(ie)
2111
# ie.revision is always None if the InventoryEntry is considered
2112
# for committing. ie.snapshot will record the correct revision
2113
# which may be the sole parent if it is untouched.
2114
if ie.revision is not None:
2117
previous_entries = ie.find_previous_heads(
2119
self.repository.weave_store,
2120
self.repository.get_transaction())
2121
# we are creating a new revision for ie in the history store
2123
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2135
def _unescaper(match, _map=_unescape_map):
2136
code = match.group(1)
2140
if not code.startswith('#'):
2142
return unichr(int(code[1:])).encode('utf8')
2148
def _unescape_xml(data):
2149
"""Unescape predefined XML entities in a string of data."""
2151
if _unescape_re is None:
2152
_unescape_re = re.compile('\&([^;]*);')
2153
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
