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(), """
39
revision as _mod_revision,
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 get_commit_builder(self, branch, parents, config, timestamp=None,
376
timezone=None, committer=None, revprops=None,
378
"""Obtain a CommitBuilder for this repository.
380
:param branch: Branch to commit to.
381
:param parents: Revision ids of the parents of the new revision.
382
:param config: Configuration to use.
383
:param timestamp: Optional timestamp recorded for commit.
384
:param timezone: Optional timezone for timestamp.
385
:param committer: Optional committer to set for commit.
386
:param revprops: Optional dictionary of revision properties.
387
:param revision_id: Optional revision id.
389
revision_id = osutils.safe_revision_id(revision_id)
390
return _CommitBuilder(self, parents, config, timestamp, timezone,
391
committer, revprops, revision_id)
394
self.control_files.unlock()
397
def clone(self, a_bzrdir, revision_id=None):
398
"""Clone this repository into a_bzrdir using the current format.
400
Currently no check is made that the format of this repository and
401
the bzrdir format are compatible. FIXME RBC 20060201.
403
:return: The newly created destination repository.
405
# TODO: deprecate after 0.16; cloning this with all its settings is
406
# probably not very useful -- mbp 20070423
407
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
408
self.copy_content_into(dest_repo, revision_id)
412
def sprout(self, to_bzrdir, revision_id=None):
413
"""Create a descendent repository for new development.
415
Unlike clone, this does not copy the settings of the repository.
417
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
418
dest_repo.fetch(self, revision_id=revision_id)
421
def _create_sprouting_repo(self, a_bzrdir, shared):
422
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
423
# use target default format.
424
dest_repo = a_bzrdir.create_repository()
426
# Most control formats need the repository to be specifically
427
# created, but on some old all-in-one formats it's not needed
429
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
430
except errors.UninitializableFormat:
431
dest_repo = a_bzrdir.open_repository()
435
def has_revision(self, revision_id):
436
"""True if this repository has a copy of the revision."""
437
revision_id = osutils.safe_revision_id(revision_id)
438
return self._revision_store.has_revision_id(revision_id,
439
self.get_transaction())
442
def get_revision_reconcile(self, revision_id):
443
"""'reconcile' helper routine that allows access to a revision always.
445
This variant of get_revision does not cross check the weave graph
446
against the revision one as get_revision does: but it should only
447
be used by reconcile, or reconcile-alike commands that are correcting
448
or testing the revision graph.
450
if not revision_id or not isinstance(revision_id, basestring):
451
raise errors.InvalidRevisionId(revision_id=revision_id,
453
return self.get_revisions([revision_id])[0]
456
def get_revisions(self, revision_ids):
457
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
458
revs = self._revision_store.get_revisions(revision_ids,
459
self.get_transaction())
461
assert not isinstance(rev.revision_id, unicode)
462
for parent_id in rev.parent_ids:
463
assert not isinstance(parent_id, unicode)
467
def get_revision_xml(self, revision_id):
468
# TODO: jam 20070210 This shouldn't be necessary since get_revision
469
# would have already do it.
470
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
471
revision_id = osutils.safe_revision_id(revision_id)
472
rev = self.get_revision(revision_id)
474
# the current serializer..
475
self._revision_store._serializer.write_revision(rev, rev_tmp)
477
return rev_tmp.getvalue()
480
def get_revision(self, revision_id):
481
"""Return the Revision object for a named revision"""
482
# TODO: jam 20070210 get_revision_reconcile should do this for us
483
revision_id = osutils.safe_revision_id(revision_id)
484
r = self.get_revision_reconcile(revision_id)
485
# weave corruption can lead to absent revision markers that should be
487
# the following test is reasonably cheap (it needs a single weave read)
488
# and the weave is cached in read transactions. In write transactions
489
# it is not cached but typically we only read a small number of
490
# revisions. For knits when they are introduced we will probably want
491
# to ensure that caching write transactions are in use.
492
inv = self.get_inventory_weave()
493
self._check_revision_parents(r, inv)
497
def get_deltas_for_revisions(self, revisions):
498
"""Produce a generator of revision deltas.
500
Note that the input is a sequence of REVISIONS, not revision_ids.
501
Trees will be held in memory until the generator exits.
502
Each delta is relative to the revision's lefthand predecessor.
504
required_trees = set()
505
for revision in revisions:
506
required_trees.add(revision.revision_id)
507
required_trees.update(revision.parent_ids[:1])
508
trees = dict((t.get_revision_id(), t) for
509
t in self.revision_trees(required_trees))
510
for revision in revisions:
511
if not revision.parent_ids:
512
old_tree = self.revision_tree(None)
514
old_tree = trees[revision.parent_ids[0]]
515
yield trees[revision.revision_id].changes_from(old_tree)
518
def get_revision_delta(self, revision_id):
519
"""Return the delta for one revision.
521
The delta is relative to the left-hand predecessor of the
524
r = self.get_revision(revision_id)
525
return list(self.get_deltas_for_revisions([r]))[0]
527
def _check_revision_parents(self, revision, inventory):
528
"""Private to Repository and Fetch.
530
This checks the parentage of revision in an inventory weave for
531
consistency and is only applicable to inventory-weave-for-ancestry
532
using repository formats & fetchers.
534
weave_parents = inventory.get_parents(revision.revision_id)
535
weave_names = inventory.versions()
536
for parent_id in revision.parent_ids:
537
if parent_id in weave_names:
538
# this parent must not be a ghost.
539
if not parent_id in weave_parents:
541
raise errors.CorruptRepository(self)
544
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
545
revision_id = osutils.safe_revision_id(revision_id)
546
signature = gpg_strategy.sign(plaintext)
547
self._revision_store.add_revision_signature_text(revision_id,
549
self.get_transaction())
551
def fileids_altered_by_revision_ids(self, revision_ids):
552
"""Find the file ids and versions affected by revisions.
554
:param revisions: an iterable containing revision ids.
555
:return: a dictionary mapping altered file-ids to an iterable of
556
revision_ids. Each altered file-ids has the exact revision_ids that
557
altered it listed explicitly.
559
assert self._serializer.support_altered_by_hack, \
560
("fileids_altered_by_revision_ids only supported for branches "
561
"which store inventory as unnested xml, not on %r" % self)
562
selected_revision_ids = set(osutils.safe_revision_id(r)
563
for r in revision_ids)
564
w = self.get_inventory_weave()
567
# this code needs to read every new line in every inventory for the
568
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
569
# not present in one of those inventories is unnecessary but not
570
# harmful because we are filtering by the revision id marker in the
571
# inventory lines : we only select file ids altered in one of those
572
# revisions. We don't need to see all lines in the inventory because
573
# only those added in an inventory in rev X can contain a revision=X
575
unescape_revid_cache = {}
576
unescape_fileid_cache = {}
578
# jam 20061218 In a big fetch, this handles hundreds of thousands
579
# of lines, so it has had a lot of inlining and optimizing done.
580
# Sorry that it is a little bit messy.
581
# Move several functions to be local variables, since this is a long
583
search = self._file_ids_altered_regex.search
584
unescape = _unescape_xml
585
setdefault = result.setdefault
586
pb = ui.ui_factory.nested_progress_bar()
588
for line in w.iter_lines_added_or_present_in_versions(
589
selected_revision_ids, pb=pb):
593
# One call to match.group() returning multiple items is quite a
594
# bit faster than 2 calls to match.group() each returning 1
595
file_id, revision_id = match.group('file_id', 'revision_id')
597
# Inlining the cache lookups helps a lot when you make 170,000
598
# lines and 350k ids, versus 8.4 unique ids.
599
# Using a cache helps in 2 ways:
600
# 1) Avoids unnecessary decoding calls
601
# 2) Re-uses cached strings, which helps in future set and
603
# (2) is enough that removing encoding entirely along with
604
# the cache (so we are using plain strings) results in no
605
# performance improvement.
607
revision_id = unescape_revid_cache[revision_id]
609
unescaped = unescape(revision_id)
610
unescape_revid_cache[revision_id] = unescaped
611
revision_id = unescaped
613
if revision_id in selected_revision_ids:
615
file_id = unescape_fileid_cache[file_id]
617
unescaped = unescape(file_id)
618
unescape_fileid_cache[file_id] = unescaped
620
setdefault(file_id, set()).add(revision_id)
626
def get_inventory_weave(self):
627
return self.control_weaves.get_weave('inventory',
628
self.get_transaction())
631
def get_inventory(self, revision_id):
632
"""Get Inventory object by hash."""
633
# TODO: jam 20070210 Technically we don't need to sanitize, since all
634
# called functions must sanitize.
635
revision_id = osutils.safe_revision_id(revision_id)
636
return self.deserialise_inventory(
637
revision_id, self.get_inventory_xml(revision_id))
639
def deserialise_inventory(self, revision_id, xml):
640
"""Transform the xml into an inventory object.
642
:param revision_id: The expected revision id of the inventory.
643
:param xml: A serialised inventory.
645
revision_id = osutils.safe_revision_id(revision_id)
646
result = self._serializer.read_inventory_from_string(xml)
647
result.root.revision = revision_id
650
def serialise_inventory(self, inv):
651
return self._serializer.write_inventory_to_string(inv)
654
def get_inventory_xml(self, revision_id):
655
"""Get inventory XML as a file object."""
656
revision_id = osutils.safe_revision_id(revision_id)
658
assert isinstance(revision_id, str), type(revision_id)
659
iw = self.get_inventory_weave()
660
return iw.get_text(revision_id)
662
raise errors.HistoryMissing(self, 'inventory', revision_id)
665
def get_inventory_sha1(self, revision_id):
666
"""Return the sha1 hash of the inventory entry
668
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
669
revision_id = osutils.safe_revision_id(revision_id)
670
return self.get_revision(revision_id).inventory_sha1
673
def get_revision_graph(self, revision_id=None):
674
"""Return a dictionary containing the revision graph.
676
:param revision_id: The revision_id to get a graph from. If None, then
677
the entire revision graph is returned. This is a deprecated mode of
678
operation and will be removed in the future.
679
:return: a dictionary of revision_id->revision_parents_list.
681
# special case NULL_REVISION
682
if revision_id == _mod_revision.NULL_REVISION:
684
revision_id = osutils.safe_revision_id(revision_id)
685
a_weave = self.get_inventory_weave()
686
all_revisions = self._eliminate_revisions_not_present(
688
entire_graph = dict([(node, a_weave.get_parents(node)) for
689
node in all_revisions])
690
if revision_id is None:
692
elif revision_id not in entire_graph:
693
raise errors.NoSuchRevision(self, revision_id)
695
# add what can be reached from revision_id
697
pending = set([revision_id])
698
while len(pending) > 0:
700
result[node] = entire_graph[node]
701
for revision_id in result[node]:
702
if revision_id not in result:
703
pending.add(revision_id)
707
def get_revision_graph_with_ghosts(self, revision_ids=None):
708
"""Return a graph of the revisions with ghosts marked as applicable.
710
:param revision_ids: an iterable of revisions to graph or None for all.
711
:return: a Graph object with the graph reachable from revision_ids.
713
result = deprecated_graph.Graph()
715
pending = set(self.all_revision_ids())
718
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
719
# special case NULL_REVISION
720
if _mod_revision.NULL_REVISION in pending:
721
pending.remove(_mod_revision.NULL_REVISION)
722
required = set(pending)
725
revision_id = pending.pop()
727
rev = self.get_revision(revision_id)
728
except errors.NoSuchRevision:
729
if revision_id in required:
732
result.add_ghost(revision_id)
734
for parent_id in rev.parent_ids:
735
# is this queued or done ?
736
if (parent_id not in pending and
737
parent_id not in done):
739
pending.add(parent_id)
740
result.add_node(revision_id, rev.parent_ids)
741
done.add(revision_id)
744
def _get_history_vf(self):
745
"""Get a versionedfile whose history graph reflects all revisions.
747
For weave repositories, this is the inventory weave.
749
return self.get_inventory_weave()
751
def iter_reverse_revision_history(self, revision_id):
752
"""Iterate backwards through revision ids in the lefthand history
754
:param revision_id: The revision id to start with. All its lefthand
755
ancestors will be traversed.
757
revision_id = osutils.safe_revision_id(revision_id)
758
if revision_id in (None, _mod_revision.NULL_REVISION):
760
next_id = revision_id
761
versionedfile = self._get_history_vf()
764
parents = versionedfile.get_parents(next_id)
765
if len(parents) == 0:
771
def get_revision_inventory(self, revision_id):
772
"""Return inventory of a past revision."""
773
# TODO: Unify this with get_inventory()
774
# bzr 0.0.6 and later imposes the constraint that the inventory_id
775
# must be the same as its revision, so this is trivial.
776
if revision_id is None:
777
# This does not make sense: if there is no revision,
778
# then it is the current tree inventory surely ?!
779
# and thus get_root_id() is something that looks at the last
780
# commit on the branch, and the get_root_id is an inventory check.
781
raise NotImplementedError
782
# return Inventory(self.get_root_id())
784
return self.get_inventory(revision_id)
788
"""Return True if this repository is flagged as a shared repository."""
789
raise NotImplementedError(self.is_shared)
792
def reconcile(self, other=None, thorough=False):
793
"""Reconcile this repository."""
794
from bzrlib.reconcile import RepoReconciler
795
reconciler = RepoReconciler(self, thorough=thorough)
796
reconciler.reconcile()
800
def revision_tree(self, revision_id):
801
"""Return Tree for a revision on this branch.
803
`revision_id` may be None for the empty tree revision.
805
# TODO: refactor this to use an existing revision object
806
# so we don't need to read it in twice.
807
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
808
return RevisionTree(self, Inventory(root_id=None),
809
_mod_revision.NULL_REVISION)
811
revision_id = osutils.safe_revision_id(revision_id)
812
inv = self.get_revision_inventory(revision_id)
813
return RevisionTree(self, inv, revision_id)
816
def revision_trees(self, revision_ids):
817
"""Return Tree for a revision on this branch.
819
`revision_id` may not be None or 'null:'"""
820
assert None not in revision_ids
821
assert _mod_revision.NULL_REVISION not in revision_ids
822
texts = self.get_inventory_weave().get_texts(revision_ids)
823
for text, revision_id in zip(texts, revision_ids):
824
inv = self.deserialise_inventory(revision_id, text)
825
yield RevisionTree(self, inv, revision_id)
828
def get_ancestry(self, revision_id):
829
"""Return a list of revision-ids integrated by a revision.
831
The first element of the list is always None, indicating the origin
832
revision. This might change when we have history horizons, or
833
perhaps we should have a new API.
835
This is topologically sorted.
837
if revision_id is None:
839
revision_id = osutils.safe_revision_id(revision_id)
840
if not self.has_revision(revision_id):
841
raise errors.NoSuchRevision(self, revision_id)
842
w = self.get_inventory_weave()
843
candidates = w.get_ancestry(revision_id)
844
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
847
def print_file(self, file, revision_id):
848
"""Print `file` to stdout.
850
FIXME RBC 20060125 as John Meinel points out this is a bad api
851
- it writes to stdout, it assumes that that is valid etc. Fix
852
by creating a new more flexible convenience function.
854
revision_id = osutils.safe_revision_id(revision_id)
855
tree = self.revision_tree(revision_id)
856
# use inventory as it was in that revision
857
file_id = tree.inventory.path2id(file)
859
# TODO: jam 20060427 Write a test for this code path
860
# it had a bug in it, and was raising the wrong
862
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
863
tree.print_file(file_id)
865
def get_transaction(self):
866
return self.control_files.get_transaction()
868
def revision_parents(self, revision_id):
869
revision_id = osutils.safe_revision_id(revision_id)
870
return self.get_inventory_weave().parent_names(revision_id)
872
def get_parents(self, revision_ids):
873
"""See StackedParentsProvider.get_parents"""
875
for revision_id in revision_ids:
876
if revision_id == _mod_revision.NULL_REVISION:
880
parents = self.get_revision(revision_id).parent_ids
881
except errors.NoSuchRevision:
884
if len(parents) == 0:
885
parents = [_mod_revision.NULL_REVISION]
886
parents_list.append(parents)
889
def _make_parents_provider(self):
892
def get_graph(self, other_repository=None):
893
"""Return the graph walker for this repository format"""
894
parents_provider = self._make_parents_provider()
895
if (other_repository is not None and
896
other_repository.bzrdir.transport.base !=
897
self.bzrdir.transport.base):
898
parents_provider = graph._StackedParentsProvider(
899
[parents_provider, other_repository._make_parents_provider()])
900
return graph.Graph(parents_provider)
903
def set_make_working_trees(self, new_value):
904
"""Set the policy flag for making working trees when creating branches.
906
This only applies to branches that use this repository.
908
The default is 'True'.
909
:param new_value: True to restore the default, False to disable making
912
raise NotImplementedError(self.set_make_working_trees)
914
def make_working_trees(self):
915
"""Returns the policy for making working trees on new branches."""
916
raise NotImplementedError(self.make_working_trees)
919
def sign_revision(self, revision_id, gpg_strategy):
920
revision_id = osutils.safe_revision_id(revision_id)
921
plaintext = Testament.from_revision(self, revision_id).as_short_text()
922
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
925
def has_signature_for_revision_id(self, revision_id):
926
"""Query for a revision signature for revision_id in the repository."""
927
revision_id = osutils.safe_revision_id(revision_id)
928
return self._revision_store.has_signature(revision_id,
929
self.get_transaction())
932
def get_signature_text(self, revision_id):
933
"""Return the text for a signature."""
934
revision_id = osutils.safe_revision_id(revision_id)
935
return self._revision_store.get_signature_text(revision_id,
936
self.get_transaction())
939
def check(self, revision_ids):
940
"""Check consistency of all history of given revision_ids.
942
Different repository implementations should override _check().
944
:param revision_ids: A non-empty list of revision_ids whose ancestry
945
will be checked. Typically the last revision_id of a branch.
948
raise ValueError("revision_ids must be non-empty in %s.check"
950
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
951
return self._check(revision_ids)
953
def _check(self, revision_ids):
954
result = check.Check(self)
958
def _warn_if_deprecated(self):
959
global _deprecation_warning_done
960
if _deprecation_warning_done:
962
_deprecation_warning_done = True
963
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
964
% (self._format, self.bzrdir.transport.base))
966
def supports_rich_root(self):
967
return self._format.rich_root_data
969
def _check_ascii_revisionid(self, revision_id, method):
970
"""Private helper for ascii-only repositories."""
971
# weave repositories refuse to store revisionids that are non-ascii.
972
if revision_id is not None:
973
# weaves require ascii revision ids.
974
if isinstance(revision_id, unicode):
976
revision_id.encode('ascii')
977
except UnicodeEncodeError:
978
raise errors.NonAsciiRevisionId(method, self)
981
revision_id.decode('ascii')
982
except UnicodeDecodeError:
983
raise errors.NonAsciiRevisionId(method, self)
987
# remove these delegates a while after bzr 0.15
988
def __make_delegated(name, from_module):
989
def _deprecated_repository_forwarder():
990
symbol_versioning.warn('%s moved to %s in bzr 0.15'
991
% (name, from_module),
994
m = __import__(from_module, globals(), locals(), [name])
996
return getattr(m, name)
997
except AttributeError:
998
raise AttributeError('module %s has no name %s'
1000
globals()[name] = _deprecated_repository_forwarder
1003
'AllInOneRepository',
1004
'WeaveMetaDirRepository',
1005
'PreSplitOutRepositoryFormat',
1006
'RepositoryFormat4',
1007
'RepositoryFormat5',
1008
'RepositoryFormat6',
1009
'RepositoryFormat7',
1011
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1015
'RepositoryFormatKnit',
1016
'RepositoryFormatKnit1',
1018
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1021
def install_revision(repository, rev, revision_tree):
1022
"""Install all revision data into a repository."""
1023
present_parents = []
1025
for p_id in rev.parent_ids:
1026
if repository.has_revision(p_id):
1027
present_parents.append(p_id)
1028
parent_trees[p_id] = repository.revision_tree(p_id)
1030
parent_trees[p_id] = repository.revision_tree(None)
1032
inv = revision_tree.inventory
1033
entries = inv.iter_entries()
1034
# backwards compatability hack: skip the root id.
1035
if not repository.supports_rich_root():
1036
path, root = entries.next()
1037
if root.revision != rev.revision_id:
1038
raise errors.IncompatibleRevision(repr(repository))
1039
# Add the texts that are not already present
1040
for path, ie in entries:
1041
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1042
repository.get_transaction())
1043
if ie.revision not in w:
1045
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1046
# with InventoryEntry.find_previous_heads(). if it is, then there
1047
# is a latent bug here where the parents may have ancestors of each
1049
for revision, tree in parent_trees.iteritems():
1050
if ie.file_id not in tree:
1052
parent_id = tree.inventory[ie.file_id].revision
1053
if parent_id in text_parents:
1055
text_parents.append(parent_id)
1057
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1058
repository.get_transaction())
1059
lines = revision_tree.get_file(ie.file_id).readlines()
1060
vfile.add_lines(rev.revision_id, text_parents, lines)
1062
# install the inventory
1063
repository.add_inventory(rev.revision_id, inv, present_parents)
1064
except errors.RevisionAlreadyPresent:
1066
repository.add_revision(rev.revision_id, rev, inv)
1069
class MetaDirRepository(Repository):
1070
"""Repositories in the new meta-dir layout."""
1072
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1073
super(MetaDirRepository, self).__init__(_format,
1079
dir_mode = self.control_files._dir_mode
1080
file_mode = self.control_files._file_mode
1083
def is_shared(self):
1084
"""Return True if this repository is flagged as a shared repository."""
1085
return self.control_files._transport.has('shared-storage')
1088
def set_make_working_trees(self, new_value):
1089
"""Set the policy flag for making working trees when creating branches.
1091
This only applies to branches that use this repository.
1093
The default is 'True'.
1094
:param new_value: True to restore the default, False to disable making
1099
self.control_files._transport.delete('no-working-trees')
1100
except errors.NoSuchFile:
1103
self.control_files.put_utf8('no-working-trees', '')
1105
def make_working_trees(self):
1106
"""Returns the policy for making working trees on new branches."""
1107
return not self.control_files._transport.has('no-working-trees')
1110
class RepositoryFormatRegistry(registry.Registry):
1111
"""Registry of RepositoryFormats.
1114
def get(self, format_string):
1115
r = registry.Registry.get(self, format_string)
1121
format_registry = RepositoryFormatRegistry()
1122
"""Registry of formats, indexed by their identifying format string.
1124
This can contain either format instances themselves, or classes/factories that
1125
can be called to obtain one.
1129
#####################################################################
1130
# Repository Formats
1132
class RepositoryFormat(object):
1133
"""A repository format.
1135
Formats provide three things:
1136
* An initialization routine to construct repository data on disk.
1137
* a format string which is used when the BzrDir supports versioned
1139
* an open routine which returns a Repository instance.
1141
Formats are placed in an dict by their format string for reference
1142
during opening. These should be subclasses of RepositoryFormat
1145
Once a format is deprecated, just deprecate the initialize and open
1146
methods on the format class. Do not deprecate the object, as the
1147
object will be created every system load.
1149
Common instance attributes:
1150
_matchingbzrdir - the bzrdir format that the repository format was
1151
originally written to work with. This can be used if manually
1152
constructing a bzrdir and repository, or more commonly for test suite
1157
return "<%s>" % self.__class__.__name__
1159
def __eq__(self, other):
1160
# format objects are generally stateless
1161
return isinstance(other, self.__class__)
1163
def __ne__(self, other):
1164
return not self == other
1167
def find_format(klass, a_bzrdir):
1168
"""Return the format for the repository object in a_bzrdir.
1170
This is used by bzr native formats that have a "format" file in
1171
the repository. Other methods may be used by different types of
1175
transport = a_bzrdir.get_repository_transport(None)
1176
format_string = transport.get("format").read()
1177
return format_registry.get(format_string)
1178
except errors.NoSuchFile:
1179
raise errors.NoRepositoryPresent(a_bzrdir)
1181
raise errors.UnknownFormatError(format=format_string)
1184
def register_format(klass, format):
1185
format_registry.register(format.get_format_string(), format)
1188
def unregister_format(klass, format):
1189
format_registry.remove(format.get_format_string())
1192
def get_default_format(klass):
1193
"""Return the current default format."""
1194
from bzrlib import bzrdir
1195
return bzrdir.format_registry.make_bzrdir('default').repository_format
1197
def _get_control_store(self, repo_transport, control_files):
1198
"""Return the control store for this repository."""
1199
raise NotImplementedError(self._get_control_store)
1201
def get_format_string(self):
1202
"""Return the ASCII format string that identifies this format.
1204
Note that in pre format ?? repositories the format string is
1205
not permitted nor written to disk.
1207
raise NotImplementedError(self.get_format_string)
1209
def get_format_description(self):
1210
"""Return the short description for this format."""
1211
raise NotImplementedError(self.get_format_description)
1213
def _get_revision_store(self, repo_transport, control_files):
1214
"""Return the revision store object for this a_bzrdir."""
1215
raise NotImplementedError(self._get_revision_store)
1217
def _get_text_rev_store(self,
1224
"""Common logic for getting a revision store for a repository.
1226
see self._get_revision_store for the subclass-overridable method to
1227
get the store for a repository.
1229
from bzrlib.store.revision.text import TextRevisionStore
1230
dir_mode = control_files._dir_mode
1231
file_mode = control_files._file_mode
1232
text_store = TextStore(transport.clone(name),
1234
compressed=compressed,
1236
file_mode=file_mode)
1237
_revision_store = TextRevisionStore(text_store, serializer)
1238
return _revision_store
1240
# TODO: this shouldn't be in the base class, it's specific to things that
1241
# use weaves or knits -- mbp 20070207
1242
def _get_versioned_file_store(self,
1247
versionedfile_class=None,
1248
versionedfile_kwargs={},
1250
if versionedfile_class is None:
1251
versionedfile_class = self._versionedfile_class
1252
weave_transport = control_files._transport.clone(name)
1253
dir_mode = control_files._dir_mode
1254
file_mode = control_files._file_mode
1255
return VersionedFileStore(weave_transport, prefixed=prefixed,
1257
file_mode=file_mode,
1258
versionedfile_class=versionedfile_class,
1259
versionedfile_kwargs=versionedfile_kwargs,
1262
def initialize(self, a_bzrdir, shared=False):
1263
"""Initialize a repository of this format in a_bzrdir.
1265
:param a_bzrdir: The bzrdir to put the new repository in it.
1266
:param shared: The repository should be initialized as a sharable one.
1267
:returns: The new repository object.
1269
This may raise UninitializableFormat if shared repository are not
1270
compatible the a_bzrdir.
1272
raise NotImplementedError(self.initialize)
1274
def is_supported(self):
1275
"""Is this format supported?
1277
Supported formats must be initializable and openable.
1278
Unsupported formats may not support initialization or committing or
1279
some other features depending on the reason for not being supported.
1283
def check_conversion_target(self, target_format):
1284
raise NotImplementedError(self.check_conversion_target)
1286
def open(self, a_bzrdir, _found=False):
1287
"""Return an instance of this format for the bzrdir a_bzrdir.
1289
_found is a private parameter, do not use it.
1291
raise NotImplementedError(self.open)
1294
class MetaDirRepositoryFormat(RepositoryFormat):
1295
"""Common base class for the new repositories using the metadir layout."""
1297
rich_root_data = False
1298
supports_tree_reference = False
1299
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1302
super(MetaDirRepositoryFormat, self).__init__()
1304
def _create_control_files(self, a_bzrdir):
1305
"""Create the required files and the initial control_files object."""
1306
# FIXME: RBC 20060125 don't peek under the covers
1307
# NB: no need to escape relative paths that are url safe.
1308
repository_transport = a_bzrdir.get_repository_transport(self)
1309
control_files = lockable_files.LockableFiles(repository_transport,
1310
'lock', lockdir.LockDir)
1311
control_files.create_lock()
1312
return control_files
1314
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1315
"""Upload the initial blank content."""
1316
control_files = self._create_control_files(a_bzrdir)
1317
control_files.lock_write()
1319
control_files._transport.mkdir_multi(dirs,
1320
mode=control_files._dir_mode)
1321
for file, content in files:
1322
control_files.put(file, content)
1323
for file, content in utf8_files:
1324
control_files.put_utf8(file, content)
1326
control_files.put_utf8('shared-storage', '')
1328
control_files.unlock()
1331
# formats which have no format string are not discoverable
1332
# and not independently creatable, so are not registered. They're
1333
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1334
# needed, it's constructed directly by the BzrDir. Non-native formats where
1335
# the repository is not separately opened are similar.
1337
format_registry.register_lazy(
1338
'Bazaar-NG Repository format 7',
1339
'bzrlib.repofmt.weaverepo',
1342
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1343
# default control directory format
1345
format_registry.register_lazy(
1346
'Bazaar-NG Knit Repository Format 1',
1347
'bzrlib.repofmt.knitrepo',
1348
'RepositoryFormatKnit1',
1350
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1352
format_registry.register_lazy(
1353
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1354
'bzrlib.repofmt.knitrepo',
1355
'RepositoryFormatKnit3',
1359
class InterRepository(InterObject):
1360
"""This class represents operations taking place between two repositories.
1362
Its instances have methods like copy_content and fetch, and contain
1363
references to the source and target repositories these operations can be
1366
Often we will provide convenience methods on 'repository' which carry out
1367
operations with another repository - they will always forward to
1368
InterRepository.get(other).method_name(parameters).
1372
"""The available optimised InterRepository types."""
1374
def copy_content(self, revision_id=None):
1375
raise NotImplementedError(self.copy_content)
1377
def fetch(self, revision_id=None, pb=None):
1378
"""Fetch the content required to construct revision_id.
1380
The content is copied from self.source to self.target.
1382
:param revision_id: if None all content is copied, if NULL_REVISION no
1384
:param pb: optional progress bar to use for progress reports. If not
1385
provided a default one will be created.
1387
Returns the copied revision count and the failed revisions in a tuple:
1390
raise NotImplementedError(self.fetch)
1393
def missing_revision_ids(self, revision_id=None):
1394
"""Return the revision ids that source has that target does not.
1396
These are returned in topological order.
1398
:param revision_id: only return revision ids included by this
1401
# generic, possibly worst case, slow code path.
1402
target_ids = set(self.target.all_revision_ids())
1403
if revision_id is not None:
1404
# TODO: jam 20070210 InterRepository is internal enough that it
1405
# should assume revision_ids are already utf-8
1406
revision_id = osutils.safe_revision_id(revision_id)
1407
source_ids = self.source.get_ancestry(revision_id)
1408
assert source_ids[0] is None
1411
source_ids = self.source.all_revision_ids()
1412
result_set = set(source_ids).difference(target_ids)
1413
# this may look like a no-op: its not. It preserves the ordering
1414
# other_ids had while only returning the members from other_ids
1415
# that we've decided we need.
1416
return [rev_id for rev_id in source_ids if rev_id in result_set]
1419
class InterSameDataRepository(InterRepository):
1420
"""Code for converting between repositories that represent the same data.
1422
Data format and model must match for this to work.
1426
def _get_repo_format_to_test(self):
1427
"""Repository format for testing with."""
1428
return RepositoryFormat.get_default_format()
1431
def is_compatible(source, target):
1432
if source.supports_rich_root() != target.supports_rich_root():
1434
if source._serializer != target._serializer:
1439
def copy_content(self, revision_id=None):
1440
"""Make a complete copy of the content in self into destination.
1442
This copies both the repository's revision data, and configuration information
1443
such as the make_working_trees setting.
1445
This is a destructive operation! Do not use it on existing
1448
:param revision_id: Only copy the content needed to construct
1449
revision_id and its parents.
1452
self.target.set_make_working_trees(self.source.make_working_trees())
1453
except NotImplementedError:
1455
# TODO: jam 20070210 This is fairly internal, so we should probably
1456
# just assert that revision_id is not unicode.
1457
revision_id = osutils.safe_revision_id(revision_id)
1458
# but don't bother fetching if we have the needed data now.
1459
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1460
self.target.has_revision(revision_id)):
1462
self.target.fetch(self.source, revision_id=revision_id)
1465
def fetch(self, revision_id=None, pb=None):
1466
"""See InterRepository.fetch()."""
1467
from bzrlib.fetch import GenericRepoFetcher
1468
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1469
self.source, self.source._format, self.target,
1470
self.target._format)
1471
# TODO: jam 20070210 This should be an assert, not a translate
1472
revision_id = osutils.safe_revision_id(revision_id)
1473
f = GenericRepoFetcher(to_repository=self.target,
1474
from_repository=self.source,
1475
last_revision=revision_id,
1477
return f.count_copied, f.failed_revisions
1480
class InterWeaveRepo(InterSameDataRepository):
1481
"""Optimised code paths between Weave based repositories."""
1484
def _get_repo_format_to_test(self):
1485
from bzrlib.repofmt import weaverepo
1486
return weaverepo.RepositoryFormat7()
1489
def is_compatible(source, target):
1490
"""Be compatible with known Weave formats.
1492
We don't test for the stores being of specific types because that
1493
could lead to confusing results, and there is no need to be
1496
from bzrlib.repofmt.weaverepo import (
1502
return (isinstance(source._format, (RepositoryFormat5,
1504
RepositoryFormat7)) and
1505
isinstance(target._format, (RepositoryFormat5,
1507
RepositoryFormat7)))
1508
except AttributeError:
1512
def copy_content(self, revision_id=None):
1513
"""See InterRepository.copy_content()."""
1514
# weave specific optimised path:
1515
# TODO: jam 20070210 Internal, should be an assert, not translate
1516
revision_id = osutils.safe_revision_id(revision_id)
1518
self.target.set_make_working_trees(self.source.make_working_trees())
1519
except NotImplementedError:
1521
# FIXME do not peek!
1522
if self.source.control_files._transport.listable():
1523
pb = ui.ui_factory.nested_progress_bar()
1525
self.target.weave_store.copy_all_ids(
1526
self.source.weave_store,
1528
from_transaction=self.source.get_transaction(),
1529
to_transaction=self.target.get_transaction())
1530
pb.update('copying inventory', 0, 1)
1531
self.target.control_weaves.copy_multi(
1532
self.source.control_weaves, ['inventory'],
1533
from_transaction=self.source.get_transaction(),
1534
to_transaction=self.target.get_transaction())
1535
self.target._revision_store.text_store.copy_all_ids(
1536
self.source._revision_store.text_store,
1541
self.target.fetch(self.source, revision_id=revision_id)
1544
def fetch(self, revision_id=None, pb=None):
1545
"""See InterRepository.fetch()."""
1546
from bzrlib.fetch import GenericRepoFetcher
1547
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1548
self.source, self.source._format, self.target, self.target._format)
1549
# TODO: jam 20070210 This should be an assert, not a translate
1550
revision_id = osutils.safe_revision_id(revision_id)
1551
f = GenericRepoFetcher(to_repository=self.target,
1552
from_repository=self.source,
1553
last_revision=revision_id,
1555
return f.count_copied, f.failed_revisions
1558
def missing_revision_ids(self, revision_id=None):
1559
"""See InterRepository.missing_revision_ids()."""
1560
# we want all revisions to satisfy revision_id in source.
1561
# but we don't want to stat every file here and there.
1562
# we want then, all revisions other needs to satisfy revision_id
1563
# checked, but not those that we have locally.
1564
# so the first thing is to get a subset of the revisions to
1565
# satisfy revision_id in source, and then eliminate those that
1566
# we do already have.
1567
# this is slow on high latency connection to self, but as as this
1568
# disk format scales terribly for push anyway due to rewriting
1569
# inventory.weave, this is considered acceptable.
1571
if revision_id is not None:
1572
source_ids = self.source.get_ancestry(revision_id)
1573
assert source_ids[0] is None
1576
source_ids = self.source._all_possible_ids()
1577
source_ids_set = set(source_ids)
1578
# source_ids is the worst possible case we may need to pull.
1579
# now we want to filter source_ids against what we actually
1580
# have in target, but don't try to check for existence where we know
1581
# we do not have a revision as that would be pointless.
1582
target_ids = set(self.target._all_possible_ids())
1583
possibly_present_revisions = target_ids.intersection(source_ids_set)
1584
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1585
required_revisions = source_ids_set.difference(actually_present_revisions)
1586
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1587
if revision_id is not None:
1588
# we used get_ancestry to determine source_ids then we are assured all
1589
# revisions referenced are present as they are installed in topological order.
1590
# and the tip revision was validated by get_ancestry.
1591
return required_topo_revisions
1593
# if we just grabbed the possibly available ids, then
1594
# we only have an estimate of whats available and need to validate
1595
# that against the revision records.
1596
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1599
class InterKnitRepo(InterSameDataRepository):
1600
"""Optimised code paths between Knit based repositories."""
1603
def _get_repo_format_to_test(self):
1604
from bzrlib.repofmt import knitrepo
1605
return knitrepo.RepositoryFormatKnit1()
1608
def is_compatible(source, target):
1609
"""Be compatible with known Knit formats.
1611
We don't test for the stores being of specific types because that
1612
could lead to confusing results, and there is no need to be
1615
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1617
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1618
isinstance(target._format, (RepositoryFormatKnit1)))
1619
except AttributeError:
1623
def fetch(self, revision_id=None, pb=None):
1624
"""See InterRepository.fetch()."""
1625
from bzrlib.fetch import KnitRepoFetcher
1626
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1627
self.source, self.source._format, self.target, self.target._format)
1628
# TODO: jam 20070210 This should be an assert, not a translate
1629
revision_id = osutils.safe_revision_id(revision_id)
1630
f = KnitRepoFetcher(to_repository=self.target,
1631
from_repository=self.source,
1632
last_revision=revision_id,
1634
return f.count_copied, f.failed_revisions
1637
def missing_revision_ids(self, revision_id=None):
1638
"""See InterRepository.missing_revision_ids()."""
1639
if revision_id is not None:
1640
source_ids = self.source.get_ancestry(revision_id)
1641
assert source_ids[0] is None
1644
source_ids = self.source._all_possible_ids()
1645
source_ids_set = set(source_ids)
1646
# source_ids is the worst possible case we may need to pull.
1647
# now we want to filter source_ids against what we actually
1648
# have in target, but don't try to check for existence where we know
1649
# we do not have a revision as that would be pointless.
1650
target_ids = set(self.target._all_possible_ids())
1651
possibly_present_revisions = target_ids.intersection(source_ids_set)
1652
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1653
required_revisions = source_ids_set.difference(actually_present_revisions)
1654
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1655
if revision_id is not None:
1656
# we used get_ancestry to determine source_ids then we are assured all
1657
# revisions referenced are present as they are installed in topological order.
1658
# and the tip revision was validated by get_ancestry.
1659
return required_topo_revisions
1661
# if we just grabbed the possibly available ids, then
1662
# we only have an estimate of whats available and need to validate
1663
# that against the revision records.
1664
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1667
class InterModel1and2(InterRepository):
1670
def _get_repo_format_to_test(self):
1674
def is_compatible(source, target):
1675
if not source.supports_rich_root() and target.supports_rich_root():
1681
def fetch(self, revision_id=None, pb=None):
1682
"""See InterRepository.fetch()."""
1683
from bzrlib.fetch import Model1toKnit2Fetcher
1684
# TODO: jam 20070210 This should be an assert, not a translate
1685
revision_id = osutils.safe_revision_id(revision_id)
1686
f = Model1toKnit2Fetcher(to_repository=self.target,
1687
from_repository=self.source,
1688
last_revision=revision_id,
1690
return f.count_copied, f.failed_revisions
1693
def copy_content(self, revision_id=None):
1694
"""Make a complete copy of the content in self into destination.
1696
This is a destructive operation! Do not use it on existing
1699
:param revision_id: Only copy the content needed to construct
1700
revision_id and its parents.
1703
self.target.set_make_working_trees(self.source.make_working_trees())
1704
except NotImplementedError:
1706
# TODO: jam 20070210 Internal, assert, don't translate
1707
revision_id = osutils.safe_revision_id(revision_id)
1708
# but don't bother fetching if we have the needed data now.
1709
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1710
self.target.has_revision(revision_id)):
1712
self.target.fetch(self.source, revision_id=revision_id)
1715
class InterKnit1and2(InterKnitRepo):
1718
def _get_repo_format_to_test(self):
1722
def is_compatible(source, target):
1723
"""Be compatible with Knit1 source and Knit3 target"""
1724
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1726
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1727
RepositoryFormatKnit3
1728
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1729
isinstance(target._format, (RepositoryFormatKnit3)))
1730
except AttributeError:
1734
def fetch(self, revision_id=None, pb=None):
1735
"""See InterRepository.fetch()."""
1736
from bzrlib.fetch import Knit1to2Fetcher
1737
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1738
self.source, self.source._format, self.target,
1739
self.target._format)
1740
# TODO: jam 20070210 This should be an assert, not a translate
1741
revision_id = osutils.safe_revision_id(revision_id)
1742
f = Knit1to2Fetcher(to_repository=self.target,
1743
from_repository=self.source,
1744
last_revision=revision_id,
1746
return f.count_copied, f.failed_revisions
1749
class InterRemoteRepository(InterRepository):
1750
"""Code for converting between RemoteRepository objects.
1752
This just gets an non-remote repository from the RemoteRepository, and calls
1753
InterRepository.get again.
1756
def __init__(self, source, target):
1757
if isinstance(source, remote.RemoteRepository):
1758
source._ensure_real()
1759
real_source = source._real_repository
1761
real_source = source
1762
if isinstance(target, remote.RemoteRepository):
1763
target._ensure_real()
1764
real_target = target._real_repository
1766
real_target = target
1767
self.real_inter = InterRepository.get(real_source, real_target)
1770
def is_compatible(source, target):
1771
if isinstance(source, remote.RemoteRepository):
1773
if isinstance(target, remote.RemoteRepository):
1777
def copy_content(self, revision_id=None):
1778
self.real_inter.copy_content(revision_id=revision_id)
1780
def fetch(self, revision_id=None, pb=None):
1781
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1784
def _get_repo_format_to_test(self):
1788
InterRepository.register_optimiser(InterSameDataRepository)
1789
InterRepository.register_optimiser(InterWeaveRepo)
1790
InterRepository.register_optimiser(InterKnitRepo)
1791
InterRepository.register_optimiser(InterModel1and2)
1792
InterRepository.register_optimiser(InterKnit1and2)
1793
InterRepository.register_optimiser(InterRemoteRepository)
1796
class RepositoryTestProviderAdapter(object):
1797
"""A tool to generate a suite testing multiple repository formats at once.
1799
This is done by copying the test once for each transport and injecting
1800
the transport_server, transport_readonly_server, and bzrdir_format and
1801
repository_format classes into each copy. Each copy is also given a new id()
1802
to make it easy to identify.
1805
def __init__(self, transport_server, transport_readonly_server, formats,
1806
vfs_transport_factory=None):
1807
self._transport_server = transport_server
1808
self._transport_readonly_server = transport_readonly_server
1809
self._vfs_transport_factory = vfs_transport_factory
1810
self._formats = formats
1812
def adapt(self, test):
1813
result = unittest.TestSuite()
1814
for repository_format, bzrdir_format in self._formats:
1815
from copy import deepcopy
1816
new_test = deepcopy(test)
1817
new_test.transport_server = self._transport_server
1818
new_test.transport_readonly_server = self._transport_readonly_server
1819
# Only override the test's vfs_transport_factory if one was
1820
# specified, otherwise just leave the default in place.
1821
if self._vfs_transport_factory:
1822
new_test.vfs_transport_factory = self._vfs_transport_factory
1823
new_test.bzrdir_format = bzrdir_format
1824
new_test.repository_format = repository_format
1825
def make_new_test_id():
1826
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1827
return lambda: new_id
1828
new_test.id = make_new_test_id()
1829
result.addTest(new_test)
1833
class InterRepositoryTestProviderAdapter(object):
1834
"""A tool to generate a suite testing multiple inter repository formats.
1836
This is done by copying the test once for each interrepo provider and injecting
1837
the transport_server, transport_readonly_server, repository_format and
1838
repository_to_format classes into each copy.
1839
Each copy is also given a new id() to make it easy to identify.
1842
def __init__(self, transport_server, transport_readonly_server, formats):
1843
self._transport_server = transport_server
1844
self._transport_readonly_server = transport_readonly_server
1845
self._formats = formats
1847
def adapt(self, test):
1848
result = unittest.TestSuite()
1849
for interrepo_class, repository_format, repository_format_to in self._formats:
1850
from copy import deepcopy
1851
new_test = deepcopy(test)
1852
new_test.transport_server = self._transport_server
1853
new_test.transport_readonly_server = self._transport_readonly_server
1854
new_test.interrepo_class = interrepo_class
1855
new_test.repository_format = repository_format
1856
new_test.repository_format_to = repository_format_to
1857
def make_new_test_id():
1858
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1859
return lambda: new_id
1860
new_test.id = make_new_test_id()
1861
result.addTest(new_test)
1865
def default_test_list():
1866
"""Generate the default list of interrepo permutations to test."""
1867
from bzrlib.repofmt import knitrepo, weaverepo
1869
# test the default InterRepository between format 6 and the current
1871
# XXX: robertc 20060220 reinstate this when there are two supported
1872
# formats which do not have an optimal code path between them.
1873
#result.append((InterRepository,
1874
# RepositoryFormat6(),
1875
# RepositoryFormatKnit1()))
1876
for optimiser_class in InterRepository._optimisers:
1877
format_to_test = optimiser_class._get_repo_format_to_test()
1878
if format_to_test is not None:
1879
result.append((optimiser_class,
1880
format_to_test, format_to_test))
1881
# if there are specific combinations we want to use, we can add them
1883
result.append((InterModel1and2,
1884
weaverepo.RepositoryFormat5(),
1885
knitrepo.RepositoryFormatKnit3()))
1886
result.append((InterKnit1and2,
1887
knitrepo.RepositoryFormatKnit1(),
1888
knitrepo.RepositoryFormatKnit3()))
1892
class CopyConverter(object):
1893
"""A repository conversion tool which just performs a copy of the content.
1895
This is slow but quite reliable.
1898
def __init__(self, target_format):
1899
"""Create a CopyConverter.
1901
:param target_format: The format the resulting repository should be.
1903
self.target_format = target_format
1905
def convert(self, repo, pb):
1906
"""Perform the conversion of to_convert, giving feedback via pb.
1908
:param to_convert: The disk object to convert.
1909
:param pb: a progress bar to use for progress information.
1914
# this is only useful with metadir layouts - separated repo content.
1915
# trigger an assertion if not such
1916
repo._format.get_format_string()
1917
self.repo_dir = repo.bzrdir
1918
self.step('Moving repository to repository.backup')
1919
self.repo_dir.transport.move('repository', 'repository.backup')
1920
backup_transport = self.repo_dir.transport.clone('repository.backup')
1921
repo._format.check_conversion_target(self.target_format)
1922
self.source_repo = repo._format.open(self.repo_dir,
1924
_override_transport=backup_transport)
1925
self.step('Creating new repository')
1926
converted = self.target_format.initialize(self.repo_dir,
1927
self.source_repo.is_shared())
1928
converted.lock_write()
1930
self.step('Copying content into repository.')
1931
self.source_repo.copy_content_into(converted)
1934
self.step('Deleting old repository content.')
1935
self.repo_dir.transport.delete_tree('repository.backup')
1936
self.pb.note('repository converted')
1938
def step(self, message):
1939
"""Update the pb by a step."""
1941
self.pb.update(message, self.count, self.total)
1944
class CommitBuilder(object):
1945
"""Provides an interface to build up a commit.
1947
This allows describing a tree to be committed without needing to
1948
know the internals of the format of the repository.
1951
record_root_entry = False
1952
def __init__(self, repository, parents, config, timestamp=None,
1953
timezone=None, committer=None, revprops=None,
1955
"""Initiate a CommitBuilder.
1957
:param repository: Repository to commit to.
1958
:param parents: Revision ids of the parents of the new revision.
1959
:param config: Configuration to use.
1960
:param timestamp: Optional timestamp recorded for commit.
1961
:param timezone: Optional timezone for timestamp.
1962
:param committer: Optional committer to set for commit.
1963
:param revprops: Optional dictionary of revision properties.
1964
:param revision_id: Optional revision id.
1966
self._config = config
1968
if committer is None:
1969
self._committer = self._config.username()
1971
assert isinstance(committer, basestring), type(committer)
1972
self._committer = committer
1974
self.new_inventory = Inventory(None)
1975
self._new_revision_id = osutils.safe_revision_id(revision_id)
1976
self.parents = parents
1977
self.repository = repository
1980
if revprops is not None:
1981
self._revprops.update(revprops)
1983
if timestamp is None:
1984
timestamp = time.time()
1985
# Restrict resolution to 1ms
1986
self._timestamp = round(timestamp, 3)
1988
if timezone is None:
1989
self._timezone = osutils.local_time_offset()
1991
self._timezone = int(timezone)
1993
self._generate_revision_if_needed()
1995
def commit(self, message):
1996
"""Make the actual commit.
1998
:return: The revision id of the recorded revision.
2000
rev = _mod_revision.Revision(
2001
timestamp=self._timestamp,
2002
timezone=self._timezone,
2003
committer=self._committer,
2005
inventory_sha1=self.inv_sha1,
2006
revision_id=self._new_revision_id,
2007
properties=self._revprops)
2008
rev.parent_ids = self.parents
2009
self.repository.add_revision(self._new_revision_id, rev,
2010
self.new_inventory, self._config)
2011
return self._new_revision_id
2013
def revision_tree(self):
2014
"""Return the tree that was just committed.
2016
After calling commit() this can be called to get a RevisionTree
2017
representing the newly committed tree. This is preferred to
2018
calling Repository.revision_tree() because that may require
2019
deserializing the inventory, while we already have a copy in
2022
return RevisionTree(self.repository, self.new_inventory,
2023
self._new_revision_id)
2025
def finish_inventory(self):
2026
"""Tell the builder that the inventory is finished."""
2027
if self.new_inventory.root is None:
2028
symbol_versioning.warn('Root entry should be supplied to'
2029
' record_entry_contents, as of bzr 0.10.',
2030
DeprecationWarning, stacklevel=2)
2031
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2032
self.new_inventory.revision_id = self._new_revision_id
2033
self.inv_sha1 = self.repository.add_inventory(
2034
self._new_revision_id,
2039
def _gen_revision_id(self):
2040
"""Return new revision-id."""
2041
return generate_ids.gen_revision_id(self._config.username(),
2044
def _generate_revision_if_needed(self):
2045
"""Create a revision id if None was supplied.
2047
If the repository can not support user-specified revision ids
2048
they should override this function and raise CannotSetRevisionId
2049
if _new_revision_id is not None.
2051
:raises: CannotSetRevisionId
2053
if self._new_revision_id is None:
2054
self._new_revision_id = self._gen_revision_id()
2056
def record_entry_contents(self, ie, parent_invs, path, tree):
2057
"""Record the content of ie from tree into the commit if needed.
2059
Side effect: sets ie.revision when unchanged
2061
:param ie: An inventory entry present in the commit.
2062
:param parent_invs: The inventories of the parent revisions of the
2064
:param path: The path the entry is at in the tree.
2065
:param tree: The tree which contains this entry and should be used to
2068
if self.new_inventory.root is None and ie.parent_id is not None:
2069
symbol_versioning.warn('Root entry should be supplied to'
2070
' record_entry_contents, as of bzr 0.10.',
2071
DeprecationWarning, stacklevel=2)
2072
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2074
self.new_inventory.add(ie)
2076
# ie.revision is always None if the InventoryEntry is considered
2077
# for committing. ie.snapshot will record the correct revision
2078
# which may be the sole parent if it is untouched.
2079
if ie.revision is not None:
2082
# In this revision format, root entries have no knit or weave
2083
if ie is self.new_inventory.root:
2084
# When serializing out to disk and back in
2085
# root.revision is always _new_revision_id
2086
ie.revision = self._new_revision_id
2088
previous_entries = ie.find_previous_heads(
2090
self.repository.weave_store,
2091
self.repository.get_transaction())
2092
# we are creating a new revision for ie in the history store
2094
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2096
def modified_directory(self, file_id, file_parents):
2097
"""Record the presence of a symbolic link.
2099
:param file_id: The file_id of the link to record.
2100
:param file_parents: The per-file parent revision ids.
2102
self._add_text_to_weave(file_id, [], file_parents.keys())
2104
def modified_reference(self, file_id, file_parents):
2105
"""Record the modification of a reference.
2107
:param file_id: The file_id of the link to record.
2108
:param file_parents: The per-file parent revision ids.
2110
self._add_text_to_weave(file_id, [], file_parents.keys())
2112
def modified_file_text(self, file_id, file_parents,
2113
get_content_byte_lines, text_sha1=None,
2115
"""Record the text of file file_id
2117
:param file_id: The file_id of the file to record the text of.
2118
:param file_parents: The per-file parent revision ids.
2119
:param get_content_byte_lines: A callable which will return the byte
2121
:param text_sha1: Optional SHA1 of the file contents.
2122
:param text_size: Optional size of the file contents.
2124
# mutter('storing text of file {%s} in revision {%s} into %r',
2125
# file_id, self._new_revision_id, self.repository.weave_store)
2126
# special case to avoid diffing on renames or
2128
if (len(file_parents) == 1
2129
and text_sha1 == file_parents.values()[0].text_sha1
2130
and text_size == file_parents.values()[0].text_size):
2131
previous_ie = file_parents.values()[0]
2132
versionedfile = self.repository.weave_store.get_weave(file_id,
2133
self.repository.get_transaction())
2134
versionedfile.clone_text(self._new_revision_id,
2135
previous_ie.revision, file_parents.keys())
2136
return text_sha1, text_size
2138
new_lines = get_content_byte_lines()
2139
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2140
# should return the SHA1 and size
2141
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2142
return osutils.sha_strings(new_lines), \
2143
sum(map(len, new_lines))
2145
def modified_link(self, file_id, file_parents, link_target):
2146
"""Record the presence of a symbolic link.
2148
:param file_id: The file_id of the link to record.
2149
:param file_parents: The per-file parent revision ids.
2150
:param link_target: Target location of this link.
2152
self._add_text_to_weave(file_id, [], file_parents.keys())
2154
def _add_text_to_weave(self, file_id, new_lines, parents):
2155
versionedfile = self.repository.weave_store.get_weave_or_empty(
2156
file_id, self.repository.get_transaction())
2157
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2158
versionedfile.clear_cache()
2161
class _CommitBuilder(CommitBuilder):
2162
"""Temporary class so old CommitBuilders are detected properly
2164
Note: CommitBuilder works whether or not root entry is recorded.
2167
record_root_entry = True
2170
class RootCommitBuilder(CommitBuilder):
2171
"""This commitbuilder actually records the root id"""
2173
record_root_entry = True
2175
def record_entry_contents(self, ie, parent_invs, path, tree):
2176
"""Record the content of ie from tree into the commit if needed.
2178
Side effect: sets ie.revision when unchanged
2180
:param ie: An inventory entry present in the commit.
2181
:param parent_invs: The inventories of the parent revisions of the
2183
:param path: The path the entry is at in the tree.
2184
:param tree: The tree which contains this entry and should be used to
2187
assert self.new_inventory.root is not None or ie.parent_id is None
2188
self.new_inventory.add(ie)
2190
# ie.revision is always None if the InventoryEntry is considered
2191
# for committing. ie.snapshot will record the correct revision
2192
# which may be the sole parent if it is untouched.
2193
if ie.revision is not None:
2196
previous_entries = ie.find_previous_heads(
2198
self.repository.weave_store,
2199
self.repository.get_transaction())
2200
# we are creating a new revision for ie in the history store
2202
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2214
def _unescaper(match, _map=_unescape_map):
2215
code = match.group(1)
2219
if not code.startswith('#'):
2221
return unichr(int(code[1:])).encode('utf8')
2227
def _unescape_xml(data):
2228
"""Unescape predefined XML entities in a string of data."""
2230
if _unescape_re is None:
2231
_unescape_re = re.compile('\&([^;]*);')
2232
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
