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.revisiontree import RevisionTree
44
from bzrlib.store.versioned import VersionedFileStore
45
from bzrlib.store.text import TextStore
46
from bzrlib.testament import Testament
50
from bzrlib.decorators import needs_read_lock, needs_write_lock
51
from bzrlib.inter import InterObject
52
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
53
from bzrlib.symbol_versioning import (
57
from bzrlib.trace import mutter, note, warning
60
# Old formats display a warning, but only once
61
_deprecation_warning_done = False
64
######################################################################
67
class Repository(object):
68
"""Repository holding history for one or more branches.
70
The repository holds and retrieves historical information including
71
revisions and file history. It's normally accessed only by the Branch,
72
which views a particular line of development through that history.
74
The Repository builds on top of Stores and a Transport, which respectively
75
describe the disk data format and the way of accessing the (possibly
79
_file_ids_altered_regex = lazy_regex.lazy_compile(
80
r'file_id="(?P<file_id>[^"]+)"'
81
r'.*revision="(?P<revision_id>[^"]+)"'
85
def add_inventory(self, revision_id, inv, parents):
86
"""Add the inventory inv to the repository as revision_id.
88
:param parents: The revision ids of the parents that revision_id
89
is known to have and are in the repository already.
91
returns the sha1 of the serialized inventory.
93
revision_id = osutils.safe_revision_id(revision_id)
94
_mod_revision.check_not_reserved_id(revision_id)
95
assert inv.revision_id is None or inv.revision_id == revision_id, \
96
"Mismatch between inventory revision" \
97
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
98
assert inv.root is not None
99
inv_text = self.serialise_inventory(inv)
100
inv_sha1 = osutils.sha_string(inv_text)
101
inv_vf = self.control_weaves.get_weave('inventory',
102
self.get_transaction())
103
self._inventory_add_lines(inv_vf, revision_id, parents,
104
osutils.split_lines(inv_text))
107
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
109
for parent in parents:
111
final_parents.append(parent)
113
inv_vf.add_lines(revision_id, final_parents, lines)
116
def add_revision(self, revision_id, rev, inv=None, config=None):
117
"""Add rev to the revision store as revision_id.
119
:param revision_id: the revision id to use.
120
:param rev: The revision object.
121
:param inv: The inventory for the revision. if None, it will be looked
122
up in the inventory storer
123
:param config: If None no digital signature will be created.
124
If supplied its signature_needed method will be used
125
to determine if a signature should be made.
127
revision_id = osutils.safe_revision_id(revision_id)
128
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
130
_mod_revision.check_not_reserved_id(revision_id)
131
if config is not None and config.signature_needed():
133
inv = self.get_inventory(revision_id)
134
plaintext = Testament(rev, inv).as_short_text()
135
self.store_revision_signature(
136
gpg.GPGStrategy(config), plaintext, revision_id)
137
if not revision_id in self.get_inventory_weave():
139
raise errors.WeaveRevisionNotPresent(revision_id,
140
self.get_inventory_weave())
142
# yes, this is not suitable for adding with ghosts.
143
self.add_inventory(revision_id, inv, rev.parent_ids)
144
self._revision_store.add_revision(rev, self.get_transaction())
146
def _add_revision_text(self, revision_id, text):
147
revision = self._revision_store._serializer.read_revision_from_string(
149
self._revision_store._add_revision(revision, StringIO(text),
150
self.get_transaction())
153
def _all_possible_ids(self):
154
"""Return all the possible revisions that we could find."""
155
return self.get_inventory_weave().versions()
157
def all_revision_ids(self):
158
"""Returns a list of all the revision ids in the repository.
160
This is deprecated because code should generally work on the graph
161
reachable from a particular revision, and ignore any other revisions
162
that might be present. There is no direct replacement method.
164
return self._all_revision_ids()
167
def _all_revision_ids(self):
168
"""Returns a list of all the revision ids in the repository.
170
These are in as much topological order as the underlying store can
171
present: for weaves ghosts may lead to a lack of correctness until
172
the reweave updates the parents list.
174
if self._revision_store.text_store.listable():
175
return self._revision_store.all_revision_ids(self.get_transaction())
176
result = self._all_possible_ids()
177
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
178
# ids. (It should, since _revision_store's API should change to
179
# return utf8 revision_ids)
180
return self._eliminate_revisions_not_present(result)
182
def break_lock(self):
183
"""Break a lock if one is present from another instance.
185
Uses the ui factory to ask for confirmation if the lock may be from
188
self.control_files.break_lock()
191
def _eliminate_revisions_not_present(self, revision_ids):
192
"""Check every revision id in revision_ids to see if we have it.
194
Returns a set of the present revisions.
197
for id in revision_ids:
198
if self.has_revision(id):
203
def create(a_bzrdir):
204
"""Construct the current default format repository in a_bzrdir."""
205
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
207
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
208
"""instantiate a Repository.
210
:param _format: The format of the repository on disk.
211
:param a_bzrdir: The BzrDir of the repository.
213
In the future we will have a single api for all stores for
214
getting file texts, inventories and revisions, then
215
this construct will accept instances of those things.
217
super(Repository, self).__init__()
218
self._format = _format
219
# the following are part of the public API for Repository:
220
self.bzrdir = a_bzrdir
221
self.control_files = control_files
222
self._revision_store = _revision_store
223
self.text_store = text_store
224
# backwards compatibility
225
self.weave_store = text_store
226
# not right yet - should be more semantically clear ?
228
self.control_store = control_store
229
self.control_weaves = control_store
230
# TODO: make sure to construct the right store classes, etc, depending
231
# on whether escaping is required.
232
self._warn_if_deprecated()
235
return '%s(%r)' % (self.__class__.__name__,
236
self.bzrdir.transport.base)
239
return self.control_files.is_locked()
241
def lock_write(self, token=None):
242
"""Lock this repository for writing.
244
:param token: if this is already locked, then lock_write will fail
245
unless the token matches the existing lock.
246
:returns: a token if this instance supports tokens, otherwise None.
247
:raises TokenLockingNotSupported: when a token is given but this
248
instance doesn't support using token locks.
249
:raises MismatchedToken: if the specified token doesn't match the token
250
of the existing lock.
252
A token should be passed in if you know that you have locked the object
253
some other way, and need to synchronise this object's state with that
256
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
258
return self.control_files.lock_write(token=token)
261
self.control_files.lock_read()
263
def get_physical_lock_status(self):
264
return self.control_files.get_physical_lock_status()
266
def leave_lock_in_place(self):
267
"""Tell this repository not to release the physical lock when this
270
If lock_write doesn't return a token, then this method is not supported.
272
self.control_files.leave_in_place()
274
def dont_leave_lock_in_place(self):
275
"""Tell this repository to release the physical lock when this
276
object is unlocked, even if it didn't originally acquire it.
278
If lock_write doesn't return a token, then this method is not supported.
280
self.control_files.dont_leave_in_place()
283
def gather_stats(self, revid=None, committers=None):
284
"""Gather statistics from a revision id.
286
:param revid: The revision id to gather statistics from, if None, then
287
no revision specific statistics are gathered.
288
:param committers: Optional parameter controlling whether to grab
289
a count of committers from the revision specific statistics.
290
:return: A dictionary of statistics. Currently this contains:
291
committers: The number of committers if requested.
292
firstrev: A tuple with timestamp, timezone for the penultimate left
293
most ancestor of revid, if revid is not the NULL_REVISION.
294
latestrev: A tuple with timestamp, timezone for revid, if revid is
295
not the NULL_REVISION.
296
revisions: The total revision count in the repository.
297
size: An estimate disk size of the repository in bytes.
300
if revid and committers:
301
result['committers'] = 0
302
if revid and revid != _mod_revision.NULL_REVISION:
304
all_committers = set()
305
revisions = self.get_ancestry(revid)
306
# pop the leading None
308
first_revision = None
310
# ignore the revisions in the middle - just grab first and last
311
revisions = revisions[0], revisions[-1]
312
for revision in self.get_revisions(revisions):
313
if not first_revision:
314
first_revision = revision
316
all_committers.add(revision.committer)
317
last_revision = revision
319
result['committers'] = len(all_committers)
320
result['firstrev'] = (first_revision.timestamp,
321
first_revision.timezone)
322
result['latestrev'] = (last_revision.timestamp,
323
last_revision.timezone)
325
# now gather global repository information
326
if self.bzrdir.root_transport.listable():
327
c, t = self._revision_store.total_size(self.get_transaction())
328
result['revisions'] = c
333
def missing_revision_ids(self, other, revision_id=None):
334
"""Return the revision ids that other has that this does not.
336
These are returned in topological order.
338
revision_id: only return revision ids included by revision_id.
340
revision_id = osutils.safe_revision_id(revision_id)
341
return InterRepository.get(other, self).missing_revision_ids(revision_id)
345
"""Open the repository rooted at base.
347
For instance, if the repository is at URL/.bzr/repository,
348
Repository.open(URL) -> a Repository instance.
350
control = bzrdir.BzrDir.open(base)
351
return control.open_repository()
353
def copy_content_into(self, destination, revision_id=None):
354
"""Make a complete copy of the content in self into destination.
356
This is a destructive operation! Do not use it on existing
359
revision_id = osutils.safe_revision_id(revision_id)
360
return InterRepository.get(self, destination).copy_content(revision_id)
362
def fetch(self, source, revision_id=None, pb=None):
363
"""Fetch the content required to construct revision_id from source.
365
If revision_id is None all content is copied.
367
revision_id = osutils.safe_revision_id(revision_id)
368
inter = InterRepository.get(source, self)
370
return inter.fetch(revision_id=revision_id, pb=pb)
371
except NotImplementedError:
372
raise errors.IncompatibleRepositories(source, self)
374
def get_commit_builder(self, branch, parents, config, timestamp=None,
375
timezone=None, committer=None, revprops=None,
377
"""Obtain a CommitBuilder for this repository.
379
:param branch: Branch to commit to.
380
:param parents: Revision ids of the parents of the new revision.
381
:param config: Configuration to use.
382
:param timestamp: Optional timestamp recorded for commit.
383
:param timezone: Optional timezone for timestamp.
384
:param committer: Optional committer to set for commit.
385
:param revprops: Optional dictionary of revision properties.
386
:param revision_id: Optional revision id.
388
revision_id = osutils.safe_revision_id(revision_id)
389
return _CommitBuilder(self, parents, config, timestamp, timezone,
390
committer, revprops, revision_id)
393
self.control_files.unlock()
396
def clone(self, a_bzrdir, revision_id=None):
397
"""Clone this repository into a_bzrdir using the current format.
399
Currently no check is made that the format of this repository and
400
the bzrdir format are compatible. FIXME RBC 20060201.
402
:return: The newly created destination repository.
404
# TODO: deprecate after 0.16; cloning this with all its settings is
405
# probably not very useful -- mbp 20070423
406
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
407
self.copy_content_into(dest_repo, revision_id)
411
def sprout(self, to_bzrdir, revision_id=None):
412
"""Create a descendent repository for new development.
414
Unlike clone, this does not copy the settings of the repository.
416
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
417
dest_repo.fetch(self, revision_id=revision_id)
420
def _create_sprouting_repo(self, a_bzrdir, shared):
421
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
422
# use target default format.
423
dest_repo = a_bzrdir.create_repository()
425
# Most control formats need the repository to be specifically
426
# created, but on some old all-in-one formats it's not needed
428
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
429
except errors.UninitializableFormat:
430
dest_repo = a_bzrdir.open_repository()
434
def has_revision(self, revision_id):
435
"""True if this repository has a copy of the revision."""
436
revision_id = osutils.safe_revision_id(revision_id)
437
return self._revision_store.has_revision_id(revision_id,
438
self.get_transaction())
441
def get_revision_reconcile(self, revision_id):
442
"""'reconcile' helper routine that allows access to a revision always.
444
This variant of get_revision does not cross check the weave graph
445
against the revision one as get_revision does: but it should only
446
be used by reconcile, or reconcile-alike commands that are correcting
447
or testing the revision graph.
449
if not revision_id or not isinstance(revision_id, basestring):
450
raise errors.InvalidRevisionId(revision_id=revision_id,
452
return self.get_revisions([revision_id])[0]
455
def get_revisions(self, revision_ids):
456
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
457
revs = self._revision_store.get_revisions(revision_ids,
458
self.get_transaction())
460
assert not isinstance(rev.revision_id, unicode)
461
for parent_id in rev.parent_ids:
462
assert not isinstance(parent_id, unicode)
466
def get_revision_xml(self, revision_id):
467
# TODO: jam 20070210 This shouldn't be necessary since get_revision
468
# would have already do it.
469
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
470
revision_id = osutils.safe_revision_id(revision_id)
471
rev = self.get_revision(revision_id)
473
# the current serializer..
474
self._revision_store._serializer.write_revision(rev, rev_tmp)
476
return rev_tmp.getvalue()
479
def get_revision(self, revision_id):
480
"""Return the Revision object for a named revision"""
481
# TODO: jam 20070210 get_revision_reconcile should do this for us
482
revision_id = osutils.safe_revision_id(revision_id)
483
r = self.get_revision_reconcile(revision_id)
484
# weave corruption can lead to absent revision markers that should be
486
# the following test is reasonably cheap (it needs a single weave read)
487
# and the weave is cached in read transactions. In write transactions
488
# it is not cached but typically we only read a small number of
489
# revisions. For knits when they are introduced we will probably want
490
# to ensure that caching write transactions are in use.
491
inv = self.get_inventory_weave()
492
self._check_revision_parents(r, inv)
496
def get_deltas_for_revisions(self, revisions):
497
"""Produce a generator of revision deltas.
499
Note that the input is a sequence of REVISIONS, not revision_ids.
500
Trees will be held in memory until the generator exits.
501
Each delta is relative to the revision's lefthand predecessor.
503
required_trees = set()
504
for revision in revisions:
505
required_trees.add(revision.revision_id)
506
required_trees.update(revision.parent_ids[:1])
507
trees = dict((t.get_revision_id(), t) for
508
t in self.revision_trees(required_trees))
509
for revision in revisions:
510
if not revision.parent_ids:
511
old_tree = self.revision_tree(None)
513
old_tree = trees[revision.parent_ids[0]]
514
yield trees[revision.revision_id].changes_from(old_tree)
517
def get_revision_delta(self, revision_id):
518
"""Return the delta for one revision.
520
The delta is relative to the left-hand predecessor of the
523
r = self.get_revision(revision_id)
524
return list(self.get_deltas_for_revisions([r]))[0]
526
def _check_revision_parents(self, revision, inventory):
527
"""Private to Repository and Fetch.
529
This checks the parentage of revision in an inventory weave for
530
consistency and is only applicable to inventory-weave-for-ancestry
531
using repository formats & fetchers.
533
weave_parents = inventory.get_parents(revision.revision_id)
534
weave_names = inventory.versions()
535
for parent_id in revision.parent_ids:
536
if parent_id in weave_names:
537
# this parent must not be a ghost.
538
if not parent_id in weave_parents:
540
raise errors.CorruptRepository(self)
543
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
544
revision_id = osutils.safe_revision_id(revision_id)
545
signature = gpg_strategy.sign(plaintext)
546
self._revision_store.add_revision_signature_text(revision_id,
548
self.get_transaction())
550
def fileids_altered_by_revision_ids(self, revision_ids):
551
"""Find the file ids and versions affected by revisions.
553
:param revisions: an iterable containing revision ids.
554
:return: a dictionary mapping altered file-ids to an iterable of
555
revision_ids. Each altered file-ids has the exact revision_ids that
556
altered it listed explicitly.
558
assert self._serializer.support_altered_by_hack, \
559
("fileids_altered_by_revision_ids only supported for branches "
560
"which store inventory as unnested xml, not on %r" % self)
561
selected_revision_ids = set(osutils.safe_revision_id(r)
562
for r in revision_ids)
563
w = self.get_inventory_weave()
566
# this code needs to read every new line in every inventory for the
567
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
568
# not present in one of those inventories is unnecessary but not
569
# harmful because we are filtering by the revision id marker in the
570
# inventory lines : we only select file ids altered in one of those
571
# revisions. We don't need to see all lines in the inventory because
572
# only those added in an inventory in rev X can contain a revision=X
574
unescape_revid_cache = {}
575
unescape_fileid_cache = {}
577
# jam 20061218 In a big fetch, this handles hundreds of thousands
578
# of lines, so it has had a lot of inlining and optimizing done.
579
# Sorry that it is a little bit messy.
580
# Move several functions to be local variables, since this is a long
582
search = self._file_ids_altered_regex.search
583
unescape = _unescape_xml
584
setdefault = result.setdefault
585
pb = ui.ui_factory.nested_progress_bar()
587
for line in w.iter_lines_added_or_present_in_versions(
588
selected_revision_ids, pb=pb):
592
# One call to match.group() returning multiple items is quite a
593
# bit faster than 2 calls to match.group() each returning 1
594
file_id, revision_id = match.group('file_id', 'revision_id')
596
# Inlining the cache lookups helps a lot when you make 170,000
597
# lines and 350k ids, versus 8.4 unique ids.
598
# Using a cache helps in 2 ways:
599
# 1) Avoids unnecessary decoding calls
600
# 2) Re-uses cached strings, which helps in future set and
602
# (2) is enough that removing encoding entirely along with
603
# the cache (so we are using plain strings) results in no
604
# performance improvement.
606
revision_id = unescape_revid_cache[revision_id]
608
unescaped = unescape(revision_id)
609
unescape_revid_cache[revision_id] = unescaped
610
revision_id = unescaped
612
if revision_id in selected_revision_ids:
614
file_id = unescape_fileid_cache[file_id]
616
unescaped = unescape(file_id)
617
unescape_fileid_cache[file_id] = unescaped
619
setdefault(file_id, set()).add(revision_id)
625
def get_inventory_weave(self):
626
return self.control_weaves.get_weave('inventory',
627
self.get_transaction())
630
def get_inventory(self, revision_id):
631
"""Get Inventory object by hash."""
632
# TODO: jam 20070210 Technically we don't need to sanitize, since all
633
# called functions must sanitize.
634
revision_id = osutils.safe_revision_id(revision_id)
635
return self.deserialise_inventory(
636
revision_id, self.get_inventory_xml(revision_id))
638
def deserialise_inventory(self, revision_id, xml):
639
"""Transform the xml into an inventory object.
641
:param revision_id: The expected revision id of the inventory.
642
:param xml: A serialised inventory.
644
revision_id = osutils.safe_revision_id(revision_id)
645
result = self._serializer.read_inventory_from_string(xml)
646
result.root.revision = revision_id
649
def serialise_inventory(self, inv):
650
return self._serializer.write_inventory_to_string(inv)
653
def get_inventory_xml(self, revision_id):
654
"""Get inventory XML as a file object."""
655
revision_id = osutils.safe_revision_id(revision_id)
657
assert isinstance(revision_id, str), type(revision_id)
658
iw = self.get_inventory_weave()
659
return iw.get_text(revision_id)
661
raise errors.HistoryMissing(self, 'inventory', revision_id)
664
def get_inventory_sha1(self, revision_id):
665
"""Return the sha1 hash of the inventory entry
667
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
668
revision_id = osutils.safe_revision_id(revision_id)
669
return self.get_revision(revision_id).inventory_sha1
672
def get_revision_graph(self, revision_id=None):
673
"""Return a dictionary containing the revision graph.
675
:param revision_id: The revision_id to get a graph from. If None, then
676
the entire revision graph is returned. This is a deprecated mode of
677
operation and will be removed in the future.
678
:return: a dictionary of revision_id->revision_parents_list.
680
# special case NULL_REVISION
681
if revision_id == _mod_revision.NULL_REVISION:
683
revision_id = osutils.safe_revision_id(revision_id)
684
a_weave = self.get_inventory_weave()
685
all_revisions = self._eliminate_revisions_not_present(
687
entire_graph = dict([(node, a_weave.get_parents(node)) for
688
node in all_revisions])
689
if revision_id is None:
691
elif revision_id not in entire_graph:
692
raise errors.NoSuchRevision(self, revision_id)
694
# add what can be reached from revision_id
696
pending = set([revision_id])
697
while len(pending) > 0:
699
result[node] = entire_graph[node]
700
for revision_id in result[node]:
701
if revision_id not in result:
702
pending.add(revision_id)
706
def get_revision_graph_with_ghosts(self, revision_ids=None):
707
"""Return a graph of the revisions with ghosts marked as applicable.
709
:param revision_ids: an iterable of revisions to graph or None for all.
710
:return: a Graph object with the graph reachable from revision_ids.
712
result = graph.Graph()
714
pending = set(self.all_revision_ids())
717
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
718
# special case NULL_REVISION
719
if _mod_revision.NULL_REVISION in pending:
720
pending.remove(_mod_revision.NULL_REVISION)
721
required = set(pending)
724
revision_id = pending.pop()
726
rev = self.get_revision(revision_id)
727
except errors.NoSuchRevision:
728
if revision_id in required:
731
result.add_ghost(revision_id)
733
for parent_id in rev.parent_ids:
734
# is this queued or done ?
735
if (parent_id not in pending and
736
parent_id not in done):
738
pending.add(parent_id)
739
result.add_node(revision_id, rev.parent_ids)
740
done.add(revision_id)
743
def _get_history_vf(self):
744
"""Get a versionedfile whose history graph reflects all revisions.
746
For weave repositories, this is the inventory weave.
748
return self.get_inventory_weave()
750
def iter_reverse_revision_history(self, revision_id):
751
"""Iterate backwards through revision ids in the lefthand history
753
:param revision_id: The revision id to start with. All its lefthand
754
ancestors will be traversed.
756
revision_id = osutils.safe_revision_id(revision_id)
757
if revision_id in (None, _mod_revision.NULL_REVISION):
759
next_id = revision_id
760
versionedfile = self._get_history_vf()
763
parents = versionedfile.get_parents(next_id)
764
if len(parents) == 0:
770
def get_revision_inventory(self, revision_id):
771
"""Return inventory of a past revision."""
772
# TODO: Unify this with get_inventory()
773
# bzr 0.0.6 and later imposes the constraint that the inventory_id
774
# must be the same as its revision, so this is trivial.
775
if revision_id is None:
776
# This does not make sense: if there is no revision,
777
# then it is the current tree inventory surely ?!
778
# and thus get_root_id() is something that looks at the last
779
# commit on the branch, and the get_root_id is an inventory check.
780
raise NotImplementedError
781
# return Inventory(self.get_root_id())
783
return self.get_inventory(revision_id)
787
"""Return True if this repository is flagged as a shared repository."""
788
raise NotImplementedError(self.is_shared)
791
def reconcile(self, other=None, thorough=False):
792
"""Reconcile this repository."""
793
from bzrlib.reconcile import RepoReconciler
794
reconciler = RepoReconciler(self, thorough=thorough)
795
reconciler.reconcile()
799
def revision_tree(self, revision_id):
800
"""Return Tree for a revision on this branch.
802
`revision_id` may be None for the empty tree revision.
804
# TODO: refactor this to use an existing revision object
805
# so we don't need to read it in twice.
806
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
807
return RevisionTree(self, Inventory(root_id=None),
808
_mod_revision.NULL_REVISION)
810
revision_id = osutils.safe_revision_id(revision_id)
811
inv = self.get_revision_inventory(revision_id)
812
return RevisionTree(self, inv, revision_id)
815
def revision_trees(self, revision_ids):
816
"""Return Tree for a revision on this branch.
818
`revision_id` may not be None or 'null:'"""
819
assert None not in revision_ids
820
assert _mod_revision.NULL_REVISION not in revision_ids
821
texts = self.get_inventory_weave().get_texts(revision_ids)
822
for text, revision_id in zip(texts, revision_ids):
823
inv = self.deserialise_inventory(revision_id, text)
824
yield RevisionTree(self, inv, revision_id)
827
def get_ancestry(self, revision_id):
828
"""Return a list of revision-ids integrated by a revision.
830
The first element of the list is always None, indicating the origin
831
revision. This might change when we have history horizons, or
832
perhaps we should have a new API.
834
This is topologically sorted.
836
if revision_id is None:
838
revision_id = osutils.safe_revision_id(revision_id)
839
if not self.has_revision(revision_id):
840
raise errors.NoSuchRevision(self, revision_id)
841
w = self.get_inventory_weave()
842
candidates = w.get_ancestry(revision_id)
843
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
846
def print_file(self, file, revision_id):
847
"""Print `file` to stdout.
849
FIXME RBC 20060125 as John Meinel points out this is a bad api
850
- it writes to stdout, it assumes that that is valid etc. Fix
851
by creating a new more flexible convenience function.
853
revision_id = osutils.safe_revision_id(revision_id)
854
tree = self.revision_tree(revision_id)
855
# use inventory as it was in that revision
856
file_id = tree.inventory.path2id(file)
858
# TODO: jam 20060427 Write a test for this code path
859
# it had a bug in it, and was raising the wrong
861
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
862
tree.print_file(file_id)
864
def get_transaction(self):
865
return self.control_files.get_transaction()
867
def revision_parents(self, revision_id):
868
revision_id = osutils.safe_revision_id(revision_id)
869
return self.get_inventory_weave().parent_names(revision_id)
872
def set_make_working_trees(self, new_value):
873
"""Set the policy flag for making working trees when creating branches.
875
This only applies to branches that use this repository.
877
The default is 'True'.
878
:param new_value: True to restore the default, False to disable making
881
raise NotImplementedError(self.set_make_working_trees)
883
def make_working_trees(self):
884
"""Returns the policy for making working trees on new branches."""
885
raise NotImplementedError(self.make_working_trees)
888
def sign_revision(self, revision_id, gpg_strategy):
889
revision_id = osutils.safe_revision_id(revision_id)
890
plaintext = Testament.from_revision(self, revision_id).as_short_text()
891
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
894
def has_signature_for_revision_id(self, revision_id):
895
"""Query for a revision signature for revision_id in the repository."""
896
revision_id = osutils.safe_revision_id(revision_id)
897
return self._revision_store.has_signature(revision_id,
898
self.get_transaction())
901
def get_signature_text(self, revision_id):
902
"""Return the text for a signature."""
903
revision_id = osutils.safe_revision_id(revision_id)
904
return self._revision_store.get_signature_text(revision_id,
905
self.get_transaction())
908
def check(self, revision_ids):
909
"""Check consistency of all history of given revision_ids.
911
Different repository implementations should override _check().
913
:param revision_ids: A non-empty list of revision_ids whose ancestry
914
will be checked. Typically the last revision_id of a branch.
917
raise ValueError("revision_ids must be non-empty in %s.check"
919
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
920
return self._check(revision_ids)
922
def _check(self, revision_ids):
923
result = check.Check(self)
927
def _warn_if_deprecated(self):
928
global _deprecation_warning_done
929
if _deprecation_warning_done:
931
_deprecation_warning_done = True
932
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
933
% (self._format, self.bzrdir.transport.base))
935
def supports_rich_root(self):
936
return self._format.rich_root_data
938
def _check_ascii_revisionid(self, revision_id, method):
939
"""Private helper for ascii-only repositories."""
940
# weave repositories refuse to store revisionids that are non-ascii.
941
if revision_id is not None:
942
# weaves require ascii revision ids.
943
if isinstance(revision_id, unicode):
945
revision_id.encode('ascii')
946
except UnicodeEncodeError:
947
raise errors.NonAsciiRevisionId(method, self)
950
revision_id.decode('ascii')
951
except UnicodeDecodeError:
952
raise errors.NonAsciiRevisionId(method, self)
956
# remove these delegates a while after bzr 0.15
957
def __make_delegated(name, from_module):
958
def _deprecated_repository_forwarder():
959
symbol_versioning.warn('%s moved to %s in bzr 0.15'
960
% (name, from_module),
963
m = __import__(from_module, globals(), locals(), [name])
965
return getattr(m, name)
966
except AttributeError:
967
raise AttributeError('module %s has no name %s'
969
globals()[name] = _deprecated_repository_forwarder
972
'AllInOneRepository',
973
'WeaveMetaDirRepository',
974
'PreSplitOutRepositoryFormat',
980
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
984
'RepositoryFormatKnit',
985
'RepositoryFormatKnit1',
987
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
990
def install_revision(repository, rev, revision_tree):
991
"""Install all revision data into a repository."""
994
for p_id in rev.parent_ids:
995
if repository.has_revision(p_id):
996
present_parents.append(p_id)
997
parent_trees[p_id] = repository.revision_tree(p_id)
999
parent_trees[p_id] = repository.revision_tree(None)
1001
inv = revision_tree.inventory
1002
entries = inv.iter_entries()
1003
# backwards compatability hack: skip the root id.
1004
if not repository.supports_rich_root():
1005
path, root = entries.next()
1006
if root.revision != rev.revision_id:
1007
raise errors.IncompatibleRevision(repr(repository))
1008
# Add the texts that are not already present
1009
for path, ie in entries:
1010
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1011
repository.get_transaction())
1012
if ie.revision not in w:
1014
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1015
# with InventoryEntry.find_previous_heads(). if it is, then there
1016
# is a latent bug here where the parents may have ancestors of each
1018
for revision, tree in parent_trees.iteritems():
1019
if ie.file_id not in tree:
1021
parent_id = tree.inventory[ie.file_id].revision
1022
if parent_id in text_parents:
1024
text_parents.append(parent_id)
1026
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1027
repository.get_transaction())
1028
lines = revision_tree.get_file(ie.file_id).readlines()
1029
vfile.add_lines(rev.revision_id, text_parents, lines)
1031
# install the inventory
1032
repository.add_inventory(rev.revision_id, inv, present_parents)
1033
except errors.RevisionAlreadyPresent:
1035
repository.add_revision(rev.revision_id, rev, inv)
1038
class MetaDirRepository(Repository):
1039
"""Repositories in the new meta-dir layout."""
1041
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1042
super(MetaDirRepository, self).__init__(_format,
1048
dir_mode = self.control_files._dir_mode
1049
file_mode = self.control_files._file_mode
1052
def is_shared(self):
1053
"""Return True if this repository is flagged as a shared repository."""
1054
return self.control_files._transport.has('shared-storage')
1057
def set_make_working_trees(self, new_value):
1058
"""Set the policy flag for making working trees when creating branches.
1060
This only applies to branches that use this repository.
1062
The default is 'True'.
1063
:param new_value: True to restore the default, False to disable making
1068
self.control_files._transport.delete('no-working-trees')
1069
except errors.NoSuchFile:
1072
self.control_files.put_utf8('no-working-trees', '')
1074
def make_working_trees(self):
1075
"""Returns the policy for making working trees on new branches."""
1076
return not self.control_files._transport.has('no-working-trees')
1079
class RepositoryFormatRegistry(registry.Registry):
1080
"""Registry of RepositoryFormats.
1083
def get(self, format_string):
1084
r = registry.Registry.get(self, format_string)
1090
format_registry = RepositoryFormatRegistry()
1091
"""Registry of formats, indexed by their identifying format string.
1093
This can contain either format instances themselves, or classes/factories that
1094
can be called to obtain one.
1098
#####################################################################
1099
# Repository Formats
1101
class RepositoryFormat(object):
1102
"""A repository format.
1104
Formats provide three things:
1105
* An initialization routine to construct repository data on disk.
1106
* a format string which is used when the BzrDir supports versioned
1108
* an open routine which returns a Repository instance.
1110
Formats are placed in an dict by their format string for reference
1111
during opening. These should be subclasses of RepositoryFormat
1114
Once a format is deprecated, just deprecate the initialize and open
1115
methods on the format class. Do not deprecate the object, as the
1116
object will be created every system load.
1118
Common instance attributes:
1119
_matchingbzrdir - the bzrdir format that the repository format was
1120
originally written to work with. This can be used if manually
1121
constructing a bzrdir and repository, or more commonly for test suite
1126
return "<%s>" % self.__class__.__name__
1128
def __eq__(self, other):
1129
# format objects are generally stateless
1130
return isinstance(other, self.__class__)
1132
def __ne__(self, other):
1133
return not self == other
1136
def find_format(klass, a_bzrdir):
1137
"""Return the format for the repository object in a_bzrdir.
1139
This is used by bzr native formats that have a "format" file in
1140
the repository. Other methods may be used by different types of
1144
transport = a_bzrdir.get_repository_transport(None)
1145
format_string = transport.get("format").read()
1146
return format_registry.get(format_string)
1147
except errors.NoSuchFile:
1148
raise errors.NoRepositoryPresent(a_bzrdir)
1150
raise errors.UnknownFormatError(format=format_string)
1153
def register_format(klass, format):
1154
format_registry.register(format.get_format_string(), format)
1157
def unregister_format(klass, format):
1158
format_registry.remove(format.get_format_string())
1161
def get_default_format(klass):
1162
"""Return the current default format."""
1163
from bzrlib import bzrdir
1164
return bzrdir.format_registry.make_bzrdir('default').repository_format
1166
def _get_control_store(self, repo_transport, control_files):
1167
"""Return the control store for this repository."""
1168
raise NotImplementedError(self._get_control_store)
1170
def get_format_string(self):
1171
"""Return the ASCII format string that identifies this format.
1173
Note that in pre format ?? repositories the format string is
1174
not permitted nor written to disk.
1176
raise NotImplementedError(self.get_format_string)
1178
def get_format_description(self):
1179
"""Return the short description for this format."""
1180
raise NotImplementedError(self.get_format_description)
1182
def _get_revision_store(self, repo_transport, control_files):
1183
"""Return the revision store object for this a_bzrdir."""
1184
raise NotImplementedError(self._get_revision_store)
1186
def _get_text_rev_store(self,
1193
"""Common logic for getting a revision store for a repository.
1195
see self._get_revision_store for the subclass-overridable method to
1196
get the store for a repository.
1198
from bzrlib.store.revision.text import TextRevisionStore
1199
dir_mode = control_files._dir_mode
1200
file_mode = control_files._file_mode
1201
text_store = TextStore(transport.clone(name),
1203
compressed=compressed,
1205
file_mode=file_mode)
1206
_revision_store = TextRevisionStore(text_store, serializer)
1207
return _revision_store
1209
# TODO: this shouldn't be in the base class, it's specific to things that
1210
# use weaves or knits -- mbp 20070207
1211
def _get_versioned_file_store(self,
1216
versionedfile_class=None,
1217
versionedfile_kwargs={},
1219
if versionedfile_class is None:
1220
versionedfile_class = self._versionedfile_class
1221
weave_transport = control_files._transport.clone(name)
1222
dir_mode = control_files._dir_mode
1223
file_mode = control_files._file_mode
1224
return VersionedFileStore(weave_transport, prefixed=prefixed,
1226
file_mode=file_mode,
1227
versionedfile_class=versionedfile_class,
1228
versionedfile_kwargs=versionedfile_kwargs,
1231
def initialize(self, a_bzrdir, shared=False):
1232
"""Initialize a repository of this format in a_bzrdir.
1234
:param a_bzrdir: The bzrdir to put the new repository in it.
1235
:param shared: The repository should be initialized as a sharable one.
1236
:returns: The new repository object.
1238
This may raise UninitializableFormat if shared repository are not
1239
compatible the a_bzrdir.
1241
raise NotImplementedError(self.initialize)
1243
def is_supported(self):
1244
"""Is this format supported?
1246
Supported formats must be initializable and openable.
1247
Unsupported formats may not support initialization or committing or
1248
some other features depending on the reason for not being supported.
1252
def check_conversion_target(self, target_format):
1253
raise NotImplementedError(self.check_conversion_target)
1255
def open(self, a_bzrdir, _found=False):
1256
"""Return an instance of this format for the bzrdir a_bzrdir.
1258
_found is a private parameter, do not use it.
1260
raise NotImplementedError(self.open)
1263
class MetaDirRepositoryFormat(RepositoryFormat):
1264
"""Common base class for the new repositories using the metadir layout."""
1266
rich_root_data = False
1267
supports_tree_reference = False
1268
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1271
super(MetaDirRepositoryFormat, self).__init__()
1273
def _create_control_files(self, a_bzrdir):
1274
"""Create the required files and the initial control_files object."""
1275
# FIXME: RBC 20060125 don't peek under the covers
1276
# NB: no need to escape relative paths that are url safe.
1277
repository_transport = a_bzrdir.get_repository_transport(self)
1278
control_files = lockable_files.LockableFiles(repository_transport,
1279
'lock', lockdir.LockDir)
1280
control_files.create_lock()
1281
return control_files
1283
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1284
"""Upload the initial blank content."""
1285
control_files = self._create_control_files(a_bzrdir)
1286
control_files.lock_write()
1288
control_files._transport.mkdir_multi(dirs,
1289
mode=control_files._dir_mode)
1290
for file, content in files:
1291
control_files.put(file, content)
1292
for file, content in utf8_files:
1293
control_files.put_utf8(file, content)
1295
control_files.put_utf8('shared-storage', '')
1297
control_files.unlock()
1300
# formats which have no format string are not discoverable
1301
# and not independently creatable, so are not registered. They're
1302
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1303
# needed, it's constructed directly by the BzrDir. Non-native formats where
1304
# the repository is not separately opened are similar.
1306
format_registry.register_lazy(
1307
'Bazaar-NG Repository format 7',
1308
'bzrlib.repofmt.weaverepo',
1311
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1312
# default control directory format
1314
format_registry.register_lazy(
1315
'Bazaar-NG Knit Repository Format 1',
1316
'bzrlib.repofmt.knitrepo',
1317
'RepositoryFormatKnit1',
1319
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1321
format_registry.register_lazy(
1322
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1323
'bzrlib.repofmt.knitrepo',
1324
'RepositoryFormatKnit3',
1328
class InterRepository(InterObject):
1329
"""This class represents operations taking place between two repositories.
1331
Its instances have methods like copy_content and fetch, and contain
1332
references to the source and target repositories these operations can be
1335
Often we will provide convenience methods on 'repository' which carry out
1336
operations with another repository - they will always forward to
1337
InterRepository.get(other).method_name(parameters).
1341
"""The available optimised InterRepository types."""
1343
def copy_content(self, revision_id=None):
1344
raise NotImplementedError(self.copy_content)
1346
def fetch(self, revision_id=None, pb=None):
1347
"""Fetch the content required to construct revision_id.
1349
The content is copied from self.source to self.target.
1351
:param revision_id: if None all content is copied, if NULL_REVISION no
1353
:param pb: optional progress bar to use for progress reports. If not
1354
provided a default one will be created.
1356
Returns the copied revision count and the failed revisions in a tuple:
1359
raise NotImplementedError(self.fetch)
1362
def missing_revision_ids(self, revision_id=None):
1363
"""Return the revision ids that source has that target does not.
1365
These are returned in topological order.
1367
:param revision_id: only return revision ids included by this
1370
# generic, possibly worst case, slow code path.
1371
target_ids = set(self.target.all_revision_ids())
1372
if revision_id is not None:
1373
# TODO: jam 20070210 InterRepository is internal enough that it
1374
# should assume revision_ids are already utf-8
1375
revision_id = osutils.safe_revision_id(revision_id)
1376
source_ids = self.source.get_ancestry(revision_id)
1377
assert source_ids[0] is None
1380
source_ids = self.source.all_revision_ids()
1381
result_set = set(source_ids).difference(target_ids)
1382
# this may look like a no-op: its not. It preserves the ordering
1383
# other_ids had while only returning the members from other_ids
1384
# that we've decided we need.
1385
return [rev_id for rev_id in source_ids if rev_id in result_set]
1388
class InterSameDataRepository(InterRepository):
1389
"""Code for converting between repositories that represent the same data.
1391
Data format and model must match for this to work.
1395
def _get_repo_format_to_test(self):
1396
"""Repository format for testing with."""
1397
return RepositoryFormat.get_default_format()
1400
def is_compatible(source, target):
1401
if source.supports_rich_root() != target.supports_rich_root():
1403
if source._serializer != target._serializer:
1408
def copy_content(self, revision_id=None):
1409
"""Make a complete copy of the content in self into destination.
1411
This copies both the repository's revision data, and configuration information
1412
such as the make_working_trees setting.
1414
This is a destructive operation! Do not use it on existing
1417
:param revision_id: Only copy the content needed to construct
1418
revision_id and its parents.
1421
self.target.set_make_working_trees(self.source.make_working_trees())
1422
except NotImplementedError:
1424
# TODO: jam 20070210 This is fairly internal, so we should probably
1425
# just assert that revision_id is not unicode.
1426
revision_id = osutils.safe_revision_id(revision_id)
1427
# but don't bother fetching if we have the needed data now.
1428
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1429
self.target.has_revision(revision_id)):
1431
self.target.fetch(self.source, revision_id=revision_id)
1434
def fetch(self, revision_id=None, pb=None):
1435
"""See InterRepository.fetch()."""
1436
from bzrlib.fetch import GenericRepoFetcher
1437
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1438
self.source, self.source._format, self.target,
1439
self.target._format)
1440
# TODO: jam 20070210 This should be an assert, not a translate
1441
revision_id = osutils.safe_revision_id(revision_id)
1442
f = GenericRepoFetcher(to_repository=self.target,
1443
from_repository=self.source,
1444
last_revision=revision_id,
1446
return f.count_copied, f.failed_revisions
1449
class InterWeaveRepo(InterSameDataRepository):
1450
"""Optimised code paths between Weave based repositories."""
1453
def _get_repo_format_to_test(self):
1454
from bzrlib.repofmt import weaverepo
1455
return weaverepo.RepositoryFormat7()
1458
def is_compatible(source, target):
1459
"""Be compatible with known Weave formats.
1461
We don't test for the stores being of specific types because that
1462
could lead to confusing results, and there is no need to be
1465
from bzrlib.repofmt.weaverepo import (
1471
return (isinstance(source._format, (RepositoryFormat5,
1473
RepositoryFormat7)) and
1474
isinstance(target._format, (RepositoryFormat5,
1476
RepositoryFormat7)))
1477
except AttributeError:
1481
def copy_content(self, revision_id=None):
1482
"""See InterRepository.copy_content()."""
1483
# weave specific optimised path:
1484
# TODO: jam 20070210 Internal, should be an assert, not translate
1485
revision_id = osutils.safe_revision_id(revision_id)
1487
self.target.set_make_working_trees(self.source.make_working_trees())
1488
except NotImplementedError:
1490
# FIXME do not peek!
1491
if self.source.control_files._transport.listable():
1492
pb = ui.ui_factory.nested_progress_bar()
1494
self.target.weave_store.copy_all_ids(
1495
self.source.weave_store,
1497
from_transaction=self.source.get_transaction(),
1498
to_transaction=self.target.get_transaction())
1499
pb.update('copying inventory', 0, 1)
1500
self.target.control_weaves.copy_multi(
1501
self.source.control_weaves, ['inventory'],
1502
from_transaction=self.source.get_transaction(),
1503
to_transaction=self.target.get_transaction())
1504
self.target._revision_store.text_store.copy_all_ids(
1505
self.source._revision_store.text_store,
1510
self.target.fetch(self.source, revision_id=revision_id)
1513
def fetch(self, revision_id=None, pb=None):
1514
"""See InterRepository.fetch()."""
1515
from bzrlib.fetch import GenericRepoFetcher
1516
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1517
self.source, self.source._format, self.target, self.target._format)
1518
# TODO: jam 20070210 This should be an assert, not a translate
1519
revision_id = osutils.safe_revision_id(revision_id)
1520
f = GenericRepoFetcher(to_repository=self.target,
1521
from_repository=self.source,
1522
last_revision=revision_id,
1524
return f.count_copied, f.failed_revisions
1527
def missing_revision_ids(self, revision_id=None):
1528
"""See InterRepository.missing_revision_ids()."""
1529
# we want all revisions to satisfy revision_id in source.
1530
# but we don't want to stat every file here and there.
1531
# we want then, all revisions other needs to satisfy revision_id
1532
# checked, but not those that we have locally.
1533
# so the first thing is to get a subset of the revisions to
1534
# satisfy revision_id in source, and then eliminate those that
1535
# we do already have.
1536
# this is slow on high latency connection to self, but as as this
1537
# disk format scales terribly for push anyway due to rewriting
1538
# inventory.weave, this is considered acceptable.
1540
if revision_id is not None:
1541
source_ids = self.source.get_ancestry(revision_id)
1542
assert source_ids[0] is None
1545
source_ids = self.source._all_possible_ids()
1546
source_ids_set = set(source_ids)
1547
# source_ids is the worst possible case we may need to pull.
1548
# now we want to filter source_ids against what we actually
1549
# have in target, but don't try to check for existence where we know
1550
# we do not have a revision as that would be pointless.
1551
target_ids = set(self.target._all_possible_ids())
1552
possibly_present_revisions = target_ids.intersection(source_ids_set)
1553
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1554
required_revisions = source_ids_set.difference(actually_present_revisions)
1555
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1556
if revision_id is not None:
1557
# we used get_ancestry to determine source_ids then we are assured all
1558
# revisions referenced are present as they are installed in topological order.
1559
# and the tip revision was validated by get_ancestry.
1560
return required_topo_revisions
1562
# if we just grabbed the possibly available ids, then
1563
# we only have an estimate of whats available and need to validate
1564
# that against the revision records.
1565
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1568
class InterKnitRepo(InterSameDataRepository):
1569
"""Optimised code paths between Knit based repositories."""
1572
def _get_repo_format_to_test(self):
1573
from bzrlib.repofmt import knitrepo
1574
return knitrepo.RepositoryFormatKnit1()
1577
def is_compatible(source, target):
1578
"""Be compatible with known Knit formats.
1580
We don't test for the stores being of specific types because that
1581
could lead to confusing results, and there is no need to be
1584
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1586
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1587
isinstance(target._format, (RepositoryFormatKnit1)))
1588
except AttributeError:
1592
def fetch(self, revision_id=None, pb=None):
1593
"""See InterRepository.fetch()."""
1594
from bzrlib.fetch import KnitRepoFetcher
1595
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1596
self.source, self.source._format, self.target, self.target._format)
1597
# TODO: jam 20070210 This should be an assert, not a translate
1598
revision_id = osutils.safe_revision_id(revision_id)
1599
f = KnitRepoFetcher(to_repository=self.target,
1600
from_repository=self.source,
1601
last_revision=revision_id,
1603
return f.count_copied, f.failed_revisions
1606
def missing_revision_ids(self, revision_id=None):
1607
"""See InterRepository.missing_revision_ids()."""
1608
if revision_id is not None:
1609
source_ids = self.source.get_ancestry(revision_id)
1610
assert source_ids[0] is None
1613
source_ids = self.source._all_possible_ids()
1614
source_ids_set = set(source_ids)
1615
# source_ids is the worst possible case we may need to pull.
1616
# now we want to filter source_ids against what we actually
1617
# have in target, but don't try to check for existence where we know
1618
# we do not have a revision as that would be pointless.
1619
target_ids = set(self.target._all_possible_ids())
1620
possibly_present_revisions = target_ids.intersection(source_ids_set)
1621
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1622
required_revisions = source_ids_set.difference(actually_present_revisions)
1623
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1624
if revision_id is not None:
1625
# we used get_ancestry to determine source_ids then we are assured all
1626
# revisions referenced are present as they are installed in topological order.
1627
# and the tip revision was validated by get_ancestry.
1628
return required_topo_revisions
1630
# if we just grabbed the possibly available ids, then
1631
# we only have an estimate of whats available and need to validate
1632
# that against the revision records.
1633
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1636
class InterModel1and2(InterRepository):
1639
def _get_repo_format_to_test(self):
1643
def is_compatible(source, target):
1644
if not source.supports_rich_root() and target.supports_rich_root():
1650
def fetch(self, revision_id=None, pb=None):
1651
"""See InterRepository.fetch()."""
1652
from bzrlib.fetch import Model1toKnit2Fetcher
1653
# TODO: jam 20070210 This should be an assert, not a translate
1654
revision_id = osutils.safe_revision_id(revision_id)
1655
f = Model1toKnit2Fetcher(to_repository=self.target,
1656
from_repository=self.source,
1657
last_revision=revision_id,
1659
return f.count_copied, f.failed_revisions
1662
def copy_content(self, revision_id=None):
1663
"""Make a complete copy of the content in self into destination.
1665
This is a destructive operation! Do not use it on existing
1668
:param revision_id: Only copy the content needed to construct
1669
revision_id and its parents.
1672
self.target.set_make_working_trees(self.source.make_working_trees())
1673
except NotImplementedError:
1675
# TODO: jam 20070210 Internal, assert, don't translate
1676
revision_id = osutils.safe_revision_id(revision_id)
1677
# but don't bother fetching if we have the needed data now.
1678
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1679
self.target.has_revision(revision_id)):
1681
self.target.fetch(self.source, revision_id=revision_id)
1684
class InterKnit1and2(InterKnitRepo):
1687
def _get_repo_format_to_test(self):
1691
def is_compatible(source, target):
1692
"""Be compatible with Knit1 source and Knit3 target"""
1693
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1695
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1696
RepositoryFormatKnit3
1697
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1698
isinstance(target._format, (RepositoryFormatKnit3)))
1699
except AttributeError:
1703
def fetch(self, revision_id=None, pb=None):
1704
"""See InterRepository.fetch()."""
1705
from bzrlib.fetch import Knit1to2Fetcher
1706
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1707
self.source, self.source._format, self.target,
1708
self.target._format)
1709
# TODO: jam 20070210 This should be an assert, not a translate
1710
revision_id = osutils.safe_revision_id(revision_id)
1711
f = Knit1to2Fetcher(to_repository=self.target,
1712
from_repository=self.source,
1713
last_revision=revision_id,
1715
return f.count_copied, f.failed_revisions
1718
class InterRemoteRepository(InterRepository):
1719
"""Code for converting between RemoteRepository objects.
1721
This just gets an non-remote repository from the RemoteRepository, and calls
1722
InterRepository.get again.
1725
def __init__(self, source, target):
1726
if isinstance(source, remote.RemoteRepository):
1727
source._ensure_real()
1728
real_source = source._real_repository
1730
real_source = source
1731
if isinstance(target, remote.RemoteRepository):
1732
target._ensure_real()
1733
real_target = target._real_repository
1735
real_target = target
1736
self.real_inter = InterRepository.get(real_source, real_target)
1739
def is_compatible(source, target):
1740
if isinstance(source, remote.RemoteRepository):
1742
if isinstance(target, remote.RemoteRepository):
1746
def copy_content(self, revision_id=None):
1747
self.real_inter.copy_content(revision_id=revision_id)
1749
def fetch(self, revision_id=None, pb=None):
1750
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1753
def _get_repo_format_to_test(self):
1757
InterRepository.register_optimiser(InterSameDataRepository)
1758
InterRepository.register_optimiser(InterWeaveRepo)
1759
InterRepository.register_optimiser(InterKnitRepo)
1760
InterRepository.register_optimiser(InterModel1and2)
1761
InterRepository.register_optimiser(InterKnit1and2)
1762
InterRepository.register_optimiser(InterRemoteRepository)
1765
class RepositoryTestProviderAdapter(object):
1766
"""A tool to generate a suite testing multiple repository formats at once.
1768
This is done by copying the test once for each transport and injecting
1769
the transport_server, transport_readonly_server, and bzrdir_format and
1770
repository_format classes into each copy. Each copy is also given a new id()
1771
to make it easy to identify.
1774
def __init__(self, transport_server, transport_readonly_server, formats,
1775
vfs_transport_factory=None):
1776
self._transport_server = transport_server
1777
self._transport_readonly_server = transport_readonly_server
1778
self._vfs_transport_factory = vfs_transport_factory
1779
self._formats = formats
1781
def adapt(self, test):
1782
result = unittest.TestSuite()
1783
for repository_format, bzrdir_format in self._formats:
1784
from copy import deepcopy
1785
new_test = deepcopy(test)
1786
new_test.transport_server = self._transport_server
1787
new_test.transport_readonly_server = self._transport_readonly_server
1788
# Only override the test's vfs_transport_factory if one was
1789
# specified, otherwise just leave the default in place.
1790
if self._vfs_transport_factory:
1791
new_test.vfs_transport_factory = self._vfs_transport_factory
1792
new_test.bzrdir_format = bzrdir_format
1793
new_test.repository_format = repository_format
1794
def make_new_test_id():
1795
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1796
return lambda: new_id
1797
new_test.id = make_new_test_id()
1798
result.addTest(new_test)
1802
class InterRepositoryTestProviderAdapter(object):
1803
"""A tool to generate a suite testing multiple inter repository formats.
1805
This is done by copying the test once for each interrepo provider and injecting
1806
the transport_server, transport_readonly_server, repository_format and
1807
repository_to_format classes into each copy.
1808
Each copy is also given a new id() to make it easy to identify.
1811
def __init__(self, transport_server, transport_readonly_server, formats):
1812
self._transport_server = transport_server
1813
self._transport_readonly_server = transport_readonly_server
1814
self._formats = formats
1816
def adapt(self, test):
1817
result = unittest.TestSuite()
1818
for interrepo_class, repository_format, repository_format_to in self._formats:
1819
from copy import deepcopy
1820
new_test = deepcopy(test)
1821
new_test.transport_server = self._transport_server
1822
new_test.transport_readonly_server = self._transport_readonly_server
1823
new_test.interrepo_class = interrepo_class
1824
new_test.repository_format = repository_format
1825
new_test.repository_format_to = repository_format_to
1826
def make_new_test_id():
1827
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1828
return lambda: new_id
1829
new_test.id = make_new_test_id()
1830
result.addTest(new_test)
1834
def default_test_list():
1835
"""Generate the default list of interrepo permutations to test."""
1836
from bzrlib.repofmt import knitrepo, weaverepo
1838
# test the default InterRepository between format 6 and the current
1840
# XXX: robertc 20060220 reinstate this when there are two supported
1841
# formats which do not have an optimal code path between them.
1842
#result.append((InterRepository,
1843
# RepositoryFormat6(),
1844
# RepositoryFormatKnit1()))
1845
for optimiser_class in InterRepository._optimisers:
1846
format_to_test = optimiser_class._get_repo_format_to_test()
1847
if format_to_test is not None:
1848
result.append((optimiser_class,
1849
format_to_test, format_to_test))
1850
# if there are specific combinations we want to use, we can add them
1852
result.append((InterModel1and2,
1853
weaverepo.RepositoryFormat5(),
1854
knitrepo.RepositoryFormatKnit3()))
1855
result.append((InterKnit1and2,
1856
knitrepo.RepositoryFormatKnit1(),
1857
knitrepo.RepositoryFormatKnit3()))
1861
class CopyConverter(object):
1862
"""A repository conversion tool which just performs a copy of the content.
1864
This is slow but quite reliable.
1867
def __init__(self, target_format):
1868
"""Create a CopyConverter.
1870
:param target_format: The format the resulting repository should be.
1872
self.target_format = target_format
1874
def convert(self, repo, pb):
1875
"""Perform the conversion of to_convert, giving feedback via pb.
1877
:param to_convert: The disk object to convert.
1878
:param pb: a progress bar to use for progress information.
1883
# this is only useful with metadir layouts - separated repo content.
1884
# trigger an assertion if not such
1885
repo._format.get_format_string()
1886
self.repo_dir = repo.bzrdir
1887
self.step('Moving repository to repository.backup')
1888
self.repo_dir.transport.move('repository', 'repository.backup')
1889
backup_transport = self.repo_dir.transport.clone('repository.backup')
1890
repo._format.check_conversion_target(self.target_format)
1891
self.source_repo = repo._format.open(self.repo_dir,
1893
_override_transport=backup_transport)
1894
self.step('Creating new repository')
1895
converted = self.target_format.initialize(self.repo_dir,
1896
self.source_repo.is_shared())
1897
converted.lock_write()
1899
self.step('Copying content into repository.')
1900
self.source_repo.copy_content_into(converted)
1903
self.step('Deleting old repository content.')
1904
self.repo_dir.transport.delete_tree('repository.backup')
1905
self.pb.note('repository converted')
1907
def step(self, message):
1908
"""Update the pb by a step."""
1910
self.pb.update(message, self.count, self.total)
1913
class CommitBuilder(object):
1914
"""Provides an interface to build up a commit.
1916
This allows describing a tree to be committed without needing to
1917
know the internals of the format of the repository.
1920
record_root_entry = False
1921
def __init__(self, repository, parents, config, timestamp=None,
1922
timezone=None, committer=None, revprops=None,
1924
"""Initiate a CommitBuilder.
1926
:param repository: Repository to commit to.
1927
:param parents: Revision ids of the parents of the new revision.
1928
:param config: Configuration to use.
1929
:param timestamp: Optional timestamp recorded for commit.
1930
:param timezone: Optional timezone for timestamp.
1931
:param committer: Optional committer to set for commit.
1932
:param revprops: Optional dictionary of revision properties.
1933
:param revision_id: Optional revision id.
1935
self._config = config
1937
if committer is None:
1938
self._committer = self._config.username()
1940
assert isinstance(committer, basestring), type(committer)
1941
self._committer = committer
1943
self.new_inventory = Inventory(None)
1944
self._new_revision_id = osutils.safe_revision_id(revision_id)
1945
self.parents = parents
1946
self.repository = repository
1949
if revprops is not None:
1950
self._revprops.update(revprops)
1952
if timestamp is None:
1953
timestamp = time.time()
1954
# Restrict resolution to 1ms
1955
self._timestamp = round(timestamp, 3)
1957
if timezone is None:
1958
self._timezone = osutils.local_time_offset()
1960
self._timezone = int(timezone)
1962
self._generate_revision_if_needed()
1964
def commit(self, message):
1965
"""Make the actual commit.
1967
:return: The revision id of the recorded revision.
1969
rev = _mod_revision.Revision(
1970
timestamp=self._timestamp,
1971
timezone=self._timezone,
1972
committer=self._committer,
1974
inventory_sha1=self.inv_sha1,
1975
revision_id=self._new_revision_id,
1976
properties=self._revprops)
1977
rev.parent_ids = self.parents
1978
self.repository.add_revision(self._new_revision_id, rev,
1979
self.new_inventory, self._config)
1980
return self._new_revision_id
1982
def revision_tree(self):
1983
"""Return the tree that was just committed.
1985
After calling commit() this can be called to get a RevisionTree
1986
representing the newly committed tree. This is preferred to
1987
calling Repository.revision_tree() because that may require
1988
deserializing the inventory, while we already have a copy in
1991
return RevisionTree(self.repository, self.new_inventory,
1992
self._new_revision_id)
1994
def finish_inventory(self):
1995
"""Tell the builder that the inventory is finished."""
1996
if self.new_inventory.root is None:
1997
symbol_versioning.warn('Root entry should be supplied to'
1998
' record_entry_contents, as of bzr 0.10.',
1999
DeprecationWarning, stacklevel=2)
2000
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2001
self.new_inventory.revision_id = self._new_revision_id
2002
self.inv_sha1 = self.repository.add_inventory(
2003
self._new_revision_id,
2008
def _gen_revision_id(self):
2009
"""Return new revision-id."""
2010
return generate_ids.gen_revision_id(self._config.username(),
2013
def _generate_revision_if_needed(self):
2014
"""Create a revision id if None was supplied.
2016
If the repository can not support user-specified revision ids
2017
they should override this function and raise CannotSetRevisionId
2018
if _new_revision_id is not None.
2020
:raises: CannotSetRevisionId
2022
if self._new_revision_id is None:
2023
self._new_revision_id = self._gen_revision_id()
2025
def record_entry_contents(self, ie, parent_invs, path, tree):
2026
"""Record the content of ie from tree into the commit if needed.
2028
Side effect: sets ie.revision when unchanged
2030
:param ie: An inventory entry present in the commit.
2031
:param parent_invs: The inventories of the parent revisions of the
2033
:param path: The path the entry is at in the tree.
2034
:param tree: The tree which contains this entry and should be used to
2037
if self.new_inventory.root is None and ie.parent_id is not None:
2038
symbol_versioning.warn('Root entry should be supplied to'
2039
' record_entry_contents, as of bzr 0.10.',
2040
DeprecationWarning, stacklevel=2)
2041
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2043
self.new_inventory.add(ie)
2045
# ie.revision is always None if the InventoryEntry is considered
2046
# for committing. ie.snapshot will record the correct revision
2047
# which may be the sole parent if it is untouched.
2048
if ie.revision is not None:
2051
# In this revision format, root entries have no knit or weave
2052
if ie is self.new_inventory.root:
2053
# When serializing out to disk and back in
2054
# root.revision is always _new_revision_id
2055
ie.revision = self._new_revision_id
2057
previous_entries = ie.find_previous_heads(
2059
self.repository.weave_store,
2060
self.repository.get_transaction())
2061
# we are creating a new revision for ie in the history store
2063
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2065
def modified_directory(self, file_id, file_parents):
2066
"""Record the presence of a symbolic link.
2068
:param file_id: The file_id of the link to record.
2069
:param file_parents: The per-file parent revision ids.
2071
self._add_text_to_weave(file_id, [], file_parents.keys())
2073
def modified_reference(self, file_id, file_parents):
2074
"""Record the modification of a reference.
2076
:param file_id: The file_id of the link to record.
2077
:param file_parents: The per-file parent revision ids.
2079
self._add_text_to_weave(file_id, [], file_parents.keys())
2081
def modified_file_text(self, file_id, file_parents,
2082
get_content_byte_lines, text_sha1=None,
2084
"""Record the text of file file_id
2086
:param file_id: The file_id of the file to record the text of.
2087
:param file_parents: The per-file parent revision ids.
2088
:param get_content_byte_lines: A callable which will return the byte
2090
:param text_sha1: Optional SHA1 of the file contents.
2091
:param text_size: Optional size of the file contents.
2093
# mutter('storing text of file {%s} in revision {%s} into %r',
2094
# file_id, self._new_revision_id, self.repository.weave_store)
2095
# special case to avoid diffing on renames or
2097
if (len(file_parents) == 1
2098
and text_sha1 == file_parents.values()[0].text_sha1
2099
and text_size == file_parents.values()[0].text_size):
2100
previous_ie = file_parents.values()[0]
2101
versionedfile = self.repository.weave_store.get_weave(file_id,
2102
self.repository.get_transaction())
2103
versionedfile.clone_text(self._new_revision_id,
2104
previous_ie.revision, file_parents.keys())
2105
return text_sha1, text_size
2107
new_lines = get_content_byte_lines()
2108
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2109
# should return the SHA1 and size
2110
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2111
return osutils.sha_strings(new_lines), \
2112
sum(map(len, new_lines))
2114
def modified_link(self, file_id, file_parents, link_target):
2115
"""Record the presence of a symbolic link.
2117
:param file_id: The file_id of the link to record.
2118
:param file_parents: The per-file parent revision ids.
2119
:param link_target: Target location of this link.
2121
self._add_text_to_weave(file_id, [], file_parents.keys())
2123
def _add_text_to_weave(self, file_id, new_lines, parents):
2124
versionedfile = self.repository.weave_store.get_weave_or_empty(
2125
file_id, self.repository.get_transaction())
2126
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2127
versionedfile.clear_cache()
2130
class _CommitBuilder(CommitBuilder):
2131
"""Temporary class so old CommitBuilders are detected properly
2133
Note: CommitBuilder works whether or not root entry is recorded.
2136
record_root_entry = True
2139
class RootCommitBuilder(CommitBuilder):
2140
"""This commitbuilder actually records the root id"""
2142
record_root_entry = True
2144
def record_entry_contents(self, ie, parent_invs, path, tree):
2145
"""Record the content of ie from tree into the commit if needed.
2147
Side effect: sets ie.revision when unchanged
2149
:param ie: An inventory entry present in the commit.
2150
:param parent_invs: The inventories of the parent revisions of the
2152
:param path: The path the entry is at in the tree.
2153
:param tree: The tree which contains this entry and should be used to
2156
assert self.new_inventory.root is not None or ie.parent_id is None
2157
self.new_inventory.add(ie)
2159
# ie.revision is always None if the InventoryEntry is considered
2160
# for committing. ie.snapshot will record the correct revision
2161
# which may be the sole parent if it is untouched.
2162
if ie.revision is not None:
2165
previous_entries = ie.find_previous_heads(
2167
self.repository.weave_store,
2168
self.repository.get_transaction())
2169
# we are creating a new revision for ie in the history store
2171
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2183
def _unescaper(match, _map=_unescape_map):
2184
code = match.group(1)
2188
if not code.startswith('#'):
2190
return unichr(int(code[1:])).encode('utf8')
2196
def _unescape_xml(data):
2197
"""Unescape predefined XML entities in a string of data."""
2199
if _unescape_re is None:
2200
_unescape_re = re.compile('\&([^;]*);')
2201
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
