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.bundle import serializer
45
from bzrlib.revisiontree import RevisionTree
46
from bzrlib.store.versioned import VersionedFileStore
47
from bzrlib.store.text import TextStore
48
from bzrlib.testament import Testament
52
from bzrlib.decorators import needs_read_lock, needs_write_lock
53
from bzrlib.inter import InterObject
54
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
55
from bzrlib.symbol_versioning import (
59
from bzrlib.trace import mutter, note, warning
62
# Old formats display a warning, but only once
63
_deprecation_warning_done = False
66
######################################################################
69
class Repository(object):
70
"""Repository holding history for one or more branches.
72
The repository holds and retrieves historical information including
73
revisions and file history. It's normally accessed only by the Branch,
74
which views a particular line of development through that history.
76
The Repository builds on top of Stores and a Transport, which respectively
77
describe the disk data format and the way of accessing the (possibly
81
_file_ids_altered_regex = lazy_regex.lazy_compile(
82
r'file_id="(?P<file_id>[^"]+)"'
83
r'.*revision="(?P<revision_id>[^"]+)"'
87
def add_inventory(self, revision_id, inv, parents):
88
"""Add the inventory inv to the repository as revision_id.
90
:param parents: The revision ids of the parents that revision_id
91
is known to have and are in the repository already.
93
returns the sha1 of the serialized inventory.
95
revision_id = osutils.safe_revision_id(revision_id)
96
_mod_revision.check_not_reserved_id(revision_id)
97
assert inv.revision_id is None or inv.revision_id == revision_id, \
98
"Mismatch between inventory revision" \
99
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
100
assert inv.root is not None
101
inv_text = self.serialise_inventory(inv)
102
inv_sha1 = osutils.sha_string(inv_text)
103
inv_vf = self.control_weaves.get_weave('inventory',
104
self.get_transaction())
105
self._inventory_add_lines(inv_vf, revision_id, parents,
106
osutils.split_lines(inv_text))
109
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
111
for parent in parents:
113
final_parents.append(parent)
115
inv_vf.add_lines(revision_id, final_parents, lines)
118
def add_revision(self, revision_id, rev, inv=None, config=None):
119
"""Add rev to the revision store as revision_id.
121
:param revision_id: the revision id to use.
122
:param rev: The revision object.
123
:param inv: The inventory for the revision. if None, it will be looked
124
up in the inventory storer
125
:param config: If None no digital signature will be created.
126
If supplied its signature_needed method will be used
127
to determine if a signature should be made.
129
revision_id = osutils.safe_revision_id(revision_id)
130
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
132
_mod_revision.check_not_reserved_id(revision_id)
133
if config is not None and config.signature_needed():
135
inv = self.get_inventory(revision_id)
136
plaintext = Testament(rev, inv).as_short_text()
137
self.store_revision_signature(
138
gpg.GPGStrategy(config), plaintext, revision_id)
139
if not revision_id in self.get_inventory_weave():
141
raise errors.WeaveRevisionNotPresent(revision_id,
142
self.get_inventory_weave())
144
# yes, this is not suitable for adding with ghosts.
145
self.add_inventory(revision_id, inv, rev.parent_ids)
146
self._revision_store.add_revision(rev, self.get_transaction())
148
def _add_revision_text(self, revision_id, text):
149
revision = self._revision_store._serializer.read_revision_from_string(
151
self._revision_store._add_revision(revision, StringIO(text),
152
self.get_transaction())
155
def _all_possible_ids(self):
156
"""Return all the possible revisions that we could find."""
157
return self.get_inventory_weave().versions()
159
def all_revision_ids(self):
160
"""Returns a list of all the revision ids in the repository.
162
This is deprecated because code should generally work on the graph
163
reachable from a particular revision, and ignore any other revisions
164
that might be present. There is no direct replacement method.
166
return self._all_revision_ids()
169
def _all_revision_ids(self):
170
"""Returns a list of all the revision ids in the repository.
172
These are in as much topological order as the underlying store can
173
present: for weaves ghosts may lead to a lack of correctness until
174
the reweave updates the parents list.
176
if self._revision_store.text_store.listable():
177
return self._revision_store.all_revision_ids(self.get_transaction())
178
result = self._all_possible_ids()
179
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
180
# ids. (It should, since _revision_store's API should change to
181
# return utf8 revision_ids)
182
return self._eliminate_revisions_not_present(result)
184
def break_lock(self):
185
"""Break a lock if one is present from another instance.
187
Uses the ui factory to ask for confirmation if the lock may be from
190
self.control_files.break_lock()
193
def _eliminate_revisions_not_present(self, revision_ids):
194
"""Check every revision id in revision_ids to see if we have it.
196
Returns a set of the present revisions.
199
for id in revision_ids:
200
if self.has_revision(id):
205
def create(a_bzrdir):
206
"""Construct the current default format repository in a_bzrdir."""
207
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
209
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
210
"""instantiate a Repository.
212
:param _format: The format of the repository on disk.
213
:param a_bzrdir: The BzrDir of the repository.
215
In the future we will have a single api for all stores for
216
getting file texts, inventories and revisions, then
217
this construct will accept instances of those things.
219
super(Repository, self).__init__()
220
self._format = _format
221
# the following are part of the public API for Repository:
222
self.bzrdir = a_bzrdir
223
self.control_files = control_files
224
self._revision_store = _revision_store
225
self.text_store = text_store
226
# backwards compatibility
227
self.weave_store = text_store
228
# not right yet - should be more semantically clear ?
230
self.control_store = control_store
231
self.control_weaves = control_store
232
# TODO: make sure to construct the right store classes, etc, depending
233
# on whether escaping is required.
234
self._warn_if_deprecated()
237
return '%s(%r)' % (self.__class__.__name__,
238
self.bzrdir.transport.base)
241
return self.control_files.is_locked()
243
def lock_write(self, token=None):
244
"""Lock this repository for writing.
246
:param token: if this is already locked, then lock_write will fail
247
unless the token matches the existing lock.
248
:returns: a token if this instance supports tokens, otherwise None.
249
:raises TokenLockingNotSupported: when a token is given but this
250
instance doesn't support using token locks.
251
:raises MismatchedToken: if the specified token doesn't match the token
252
of the existing lock.
254
A token should be passed in if you know that you have locked the object
255
some other way, and need to synchronise this object's state with that
258
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
260
return self.control_files.lock_write(token=token)
263
self.control_files.lock_read()
265
def get_physical_lock_status(self):
266
return self.control_files.get_physical_lock_status()
268
def leave_lock_in_place(self):
269
"""Tell this repository not to release the physical lock when this
272
If lock_write doesn't return a token, then this method is not supported.
274
self.control_files.leave_in_place()
276
def dont_leave_lock_in_place(self):
277
"""Tell this repository to release the physical lock when this
278
object is unlocked, even if it didn't originally acquire it.
280
If lock_write doesn't return a token, then this method is not supported.
282
self.control_files.dont_leave_in_place()
285
def gather_stats(self, revid=None, committers=None):
286
"""Gather statistics from a revision id.
288
:param revid: The revision id to gather statistics from, if None, then
289
no revision specific statistics are gathered.
290
:param committers: Optional parameter controlling whether to grab
291
a count of committers from the revision specific statistics.
292
:return: A dictionary of statistics. Currently this contains:
293
committers: The number of committers if requested.
294
firstrev: A tuple with timestamp, timezone for the penultimate left
295
most ancestor of revid, if revid is not the NULL_REVISION.
296
latestrev: A tuple with timestamp, timezone for revid, if revid is
297
not the NULL_REVISION.
298
revisions: The total revision count in the repository.
299
size: An estimate disk size of the repository in bytes.
302
if revid and committers:
303
result['committers'] = 0
304
if revid and revid != _mod_revision.NULL_REVISION:
306
all_committers = set()
307
revisions = self.get_ancestry(revid)
308
# pop the leading None
310
first_revision = None
312
# ignore the revisions in the middle - just grab first and last
313
revisions = revisions[0], revisions[-1]
314
for revision in self.get_revisions(revisions):
315
if not first_revision:
316
first_revision = revision
318
all_committers.add(revision.committer)
319
last_revision = revision
321
result['committers'] = len(all_committers)
322
result['firstrev'] = (first_revision.timestamp,
323
first_revision.timezone)
324
result['latestrev'] = (last_revision.timestamp,
325
last_revision.timezone)
327
# now gather global repository information
328
if self.bzrdir.root_transport.listable():
329
c, t = self._revision_store.total_size(self.get_transaction())
330
result['revisions'] = c
335
def missing_revision_ids(self, other, revision_id=None):
336
"""Return the revision ids that other has that this does not.
338
These are returned in topological order.
340
revision_id: only return revision ids included by revision_id.
342
revision_id = osutils.safe_revision_id(revision_id)
343
return InterRepository.get(other, self).missing_revision_ids(revision_id)
347
"""Open the repository rooted at base.
349
For instance, if the repository is at URL/.bzr/repository,
350
Repository.open(URL) -> a Repository instance.
352
control = bzrdir.BzrDir.open(base)
353
return control.open_repository()
355
def copy_content_into(self, destination, revision_id=None):
356
"""Make a complete copy of the content in self into destination.
358
This is a destructive operation! Do not use it on existing
361
revision_id = osutils.safe_revision_id(revision_id)
362
return InterRepository.get(self, destination).copy_content(revision_id)
364
def fetch(self, source, revision_id=None, pb=None):
365
"""Fetch the content required to construct revision_id from source.
367
If revision_id is None all content is copied.
369
revision_id = osutils.safe_revision_id(revision_id)
370
inter = InterRepository.get(source, self)
372
return inter.fetch(revision_id=revision_id, pb=pb)
373
except NotImplementedError:
374
raise errors.IncompatibleRepositories(source, self)
376
def create_bundle(self, target, base, fileobj, format=None):
377
return serializer.write_bundle(self, target, base, fileobj, format)
379
def get_commit_builder(self, branch, parents, config, timestamp=None,
380
timezone=None, committer=None, revprops=None,
382
"""Obtain a CommitBuilder for this repository.
384
:param branch: Branch to commit to.
385
:param parents: Revision ids of the parents of the new revision.
386
:param config: Configuration to use.
387
:param timestamp: Optional timestamp recorded for commit.
388
:param timezone: Optional timezone for timestamp.
389
:param committer: Optional committer to set for commit.
390
:param revprops: Optional dictionary of revision properties.
391
:param revision_id: Optional revision id.
393
revision_id = osutils.safe_revision_id(revision_id)
394
return _CommitBuilder(self, parents, config, timestamp, timezone,
395
committer, revprops, revision_id)
398
self.control_files.unlock()
401
def clone(self, a_bzrdir, revision_id=None):
402
"""Clone this repository into a_bzrdir using the current format.
404
Currently no check is made that the format of this repository and
405
the bzrdir format are compatible. FIXME RBC 20060201.
407
:return: The newly created destination repository.
409
# TODO: deprecate after 0.16; cloning this with all its settings is
410
# probably not very useful -- mbp 20070423
411
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
412
self.copy_content_into(dest_repo, revision_id)
416
def sprout(self, to_bzrdir, revision_id=None):
417
"""Create a descendent repository for new development.
419
Unlike clone, this does not copy the settings of the repository.
421
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
422
dest_repo.fetch(self, revision_id=revision_id)
425
def _create_sprouting_repo(self, a_bzrdir, shared):
426
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
427
# use target default format.
428
dest_repo = a_bzrdir.create_repository()
430
# Most control formats need the repository to be specifically
431
# created, but on some old all-in-one formats it's not needed
433
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
434
except errors.UninitializableFormat:
435
dest_repo = a_bzrdir.open_repository()
439
def has_revision(self, revision_id):
440
"""True if this repository has a copy of the revision."""
441
revision_id = osutils.safe_revision_id(revision_id)
442
return self._revision_store.has_revision_id(revision_id,
443
self.get_transaction())
446
def get_revision_reconcile(self, revision_id):
447
"""'reconcile' helper routine that allows access to a revision always.
449
This variant of get_revision does not cross check the weave graph
450
against the revision one as get_revision does: but it should only
451
be used by reconcile, or reconcile-alike commands that are correcting
452
or testing the revision graph.
454
if not revision_id or not isinstance(revision_id, basestring):
455
raise errors.InvalidRevisionId(revision_id=revision_id,
457
return self.get_revisions([revision_id])[0]
460
def get_revisions(self, revision_ids):
461
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
462
revs = self._revision_store.get_revisions(revision_ids,
463
self.get_transaction())
465
assert not isinstance(rev.revision_id, unicode)
466
for parent_id in rev.parent_ids:
467
assert not isinstance(parent_id, unicode)
471
def get_revision_xml(self, revision_id):
472
# TODO: jam 20070210 This shouldn't be necessary since get_revision
473
# would have already do it.
474
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
475
revision_id = osutils.safe_revision_id(revision_id)
476
rev = self.get_revision(revision_id)
478
# the current serializer..
479
self._revision_store._serializer.write_revision(rev, rev_tmp)
481
return rev_tmp.getvalue()
484
def get_revision(self, revision_id):
485
"""Return the Revision object for a named revision"""
486
# TODO: jam 20070210 get_revision_reconcile should do this for us
487
revision_id = osutils.safe_revision_id(revision_id)
488
r = self.get_revision_reconcile(revision_id)
489
# weave corruption can lead to absent revision markers that should be
491
# the following test is reasonably cheap (it needs a single weave read)
492
# and the weave is cached in read transactions. In write transactions
493
# it is not cached but typically we only read a small number of
494
# revisions. For knits when they are introduced we will probably want
495
# to ensure that caching write transactions are in use.
496
inv = self.get_inventory_weave()
497
self._check_revision_parents(r, inv)
501
def get_deltas_for_revisions(self, revisions):
502
"""Produce a generator of revision deltas.
504
Note that the input is a sequence of REVISIONS, not revision_ids.
505
Trees will be held in memory until the generator exits.
506
Each delta is relative to the revision's lefthand predecessor.
508
required_trees = set()
509
for revision in revisions:
510
required_trees.add(revision.revision_id)
511
required_trees.update(revision.parent_ids[:1])
512
trees = dict((t.get_revision_id(), t) for
513
t in self.revision_trees(required_trees))
514
for revision in revisions:
515
if not revision.parent_ids:
516
old_tree = self.revision_tree(None)
518
old_tree = trees[revision.parent_ids[0]]
519
yield trees[revision.revision_id].changes_from(old_tree)
522
def get_revision_delta(self, revision_id):
523
"""Return the delta for one revision.
525
The delta is relative to the left-hand predecessor of the
528
r = self.get_revision(revision_id)
529
return list(self.get_deltas_for_revisions([r]))[0]
531
def _check_revision_parents(self, revision, inventory):
532
"""Private to Repository and Fetch.
534
This checks the parentage of revision in an inventory weave for
535
consistency and is only applicable to inventory-weave-for-ancestry
536
using repository formats & fetchers.
538
weave_parents = inventory.get_parents(revision.revision_id)
539
weave_names = inventory.versions()
540
for parent_id in revision.parent_ids:
541
if parent_id in weave_names:
542
# this parent must not be a ghost.
543
if not parent_id in weave_parents:
545
raise errors.CorruptRepository(self)
548
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
549
revision_id = osutils.safe_revision_id(revision_id)
550
signature = gpg_strategy.sign(plaintext)
551
self._revision_store.add_revision_signature_text(revision_id,
553
self.get_transaction())
555
def fileids_altered_by_revision_ids(self, revision_ids):
556
"""Find the file ids and versions affected by revisions.
558
:param revisions: an iterable containing revision ids.
559
:return: a dictionary mapping altered file-ids to an iterable of
560
revision_ids. Each altered file-ids has the exact revision_ids that
561
altered it listed explicitly.
563
assert self._serializer.support_altered_by_hack, \
564
("fileids_altered_by_revision_ids only supported for branches "
565
"which store inventory as unnested xml, not on %r" % self)
566
selected_revision_ids = set(osutils.safe_revision_id(r)
567
for r in revision_ids)
568
w = self.get_inventory_weave()
571
# this code needs to read every new line in every inventory for the
572
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
573
# not present in one of those inventories is unnecessary but not
574
# harmful because we are filtering by the revision id marker in the
575
# inventory lines : we only select file ids altered in one of those
576
# revisions. We don't need to see all lines in the inventory because
577
# only those added in an inventory in rev X can contain a revision=X
579
unescape_revid_cache = {}
580
unescape_fileid_cache = {}
582
# jam 20061218 In a big fetch, this handles hundreds of thousands
583
# of lines, so it has had a lot of inlining and optimizing done.
584
# Sorry that it is a little bit messy.
585
# Move several functions to be local variables, since this is a long
587
search = self._file_ids_altered_regex.search
588
unescape = _unescape_xml
589
setdefault = result.setdefault
590
pb = ui.ui_factory.nested_progress_bar()
592
for line in w.iter_lines_added_or_present_in_versions(
593
selected_revision_ids, pb=pb):
597
# One call to match.group() returning multiple items is quite a
598
# bit faster than 2 calls to match.group() each returning 1
599
file_id, revision_id = match.group('file_id', 'revision_id')
601
# Inlining the cache lookups helps a lot when you make 170,000
602
# lines and 350k ids, versus 8.4 unique ids.
603
# Using a cache helps in 2 ways:
604
# 1) Avoids unnecessary decoding calls
605
# 2) Re-uses cached strings, which helps in future set and
607
# (2) is enough that removing encoding entirely along with
608
# the cache (so we are using plain strings) results in no
609
# performance improvement.
611
revision_id = unescape_revid_cache[revision_id]
613
unescaped = unescape(revision_id)
614
unescape_revid_cache[revision_id] = unescaped
615
revision_id = unescaped
617
if revision_id in selected_revision_ids:
619
file_id = unescape_fileid_cache[file_id]
621
unescaped = unescape(file_id)
622
unescape_fileid_cache[file_id] = unescaped
624
setdefault(file_id, set()).add(revision_id)
630
def get_inventory_weave(self):
631
return self.control_weaves.get_weave('inventory',
632
self.get_transaction())
635
def get_inventory(self, revision_id):
636
"""Get Inventory object by hash."""
637
# TODO: jam 20070210 Technically we don't need to sanitize, since all
638
# called functions must sanitize.
639
revision_id = osutils.safe_revision_id(revision_id)
640
return self.deserialise_inventory(
641
revision_id, self.get_inventory_xml(revision_id))
643
def deserialise_inventory(self, revision_id, xml):
644
"""Transform the xml into an inventory object.
646
:param revision_id: The expected revision id of the inventory.
647
:param xml: A serialised inventory.
649
revision_id = osutils.safe_revision_id(revision_id)
650
result = self._serializer.read_inventory_from_string(xml)
651
result.root.revision = revision_id
654
def serialise_inventory(self, inv):
655
return self._serializer.write_inventory_to_string(inv)
658
def get_inventory_xml(self, revision_id):
659
"""Get inventory XML as a file object."""
660
revision_id = osutils.safe_revision_id(revision_id)
662
assert isinstance(revision_id, str), type(revision_id)
663
iw = self.get_inventory_weave()
664
return iw.get_text(revision_id)
666
raise errors.HistoryMissing(self, 'inventory', revision_id)
669
def get_inventory_sha1(self, revision_id):
670
"""Return the sha1 hash of the inventory entry
672
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
673
revision_id = osutils.safe_revision_id(revision_id)
674
return self.get_revision(revision_id).inventory_sha1
677
def get_revision_graph(self, revision_id=None):
678
"""Return a dictionary containing the revision graph.
680
:param revision_id: The revision_id to get a graph from. If None, then
681
the entire revision graph is returned. This is a deprecated mode of
682
operation and will be removed in the future.
683
:return: a dictionary of revision_id->revision_parents_list.
685
# special case NULL_REVISION
686
if revision_id == _mod_revision.NULL_REVISION:
688
revision_id = osutils.safe_revision_id(revision_id)
689
a_weave = self.get_inventory_weave()
690
all_revisions = self._eliminate_revisions_not_present(
692
entire_graph = dict([(node, a_weave.get_parents(node)) for
693
node in all_revisions])
694
if revision_id is None:
696
elif revision_id not in entire_graph:
697
raise errors.NoSuchRevision(self, revision_id)
699
# add what can be reached from revision_id
701
pending = set([revision_id])
702
while len(pending) > 0:
704
result[node] = entire_graph[node]
705
for revision_id in result[node]:
706
if revision_id not in result:
707
pending.add(revision_id)
711
def get_revision_graph_with_ghosts(self, revision_ids=None):
712
"""Return a graph of the revisions with ghosts marked as applicable.
714
:param revision_ids: an iterable of revisions to graph or None for all.
715
:return: a Graph object with the graph reachable from revision_ids.
717
result = deprecated_graph.Graph()
719
pending = set(self.all_revision_ids())
722
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
723
# special case NULL_REVISION
724
if _mod_revision.NULL_REVISION in pending:
725
pending.remove(_mod_revision.NULL_REVISION)
726
required = set(pending)
729
revision_id = pending.pop()
731
rev = self.get_revision(revision_id)
732
except errors.NoSuchRevision:
733
if revision_id in required:
736
result.add_ghost(revision_id)
738
for parent_id in rev.parent_ids:
739
# is this queued or done ?
740
if (parent_id not in pending and
741
parent_id not in done):
743
pending.add(parent_id)
744
result.add_node(revision_id, rev.parent_ids)
745
done.add(revision_id)
748
def _get_history_vf(self):
749
"""Get a versionedfile whose history graph reflects all revisions.
751
For weave repositories, this is the inventory weave.
753
return self.get_inventory_weave()
755
def iter_reverse_revision_history(self, revision_id):
756
"""Iterate backwards through revision ids in the lefthand history
758
:param revision_id: The revision id to start with. All its lefthand
759
ancestors will be traversed.
761
revision_id = osutils.safe_revision_id(revision_id)
762
if revision_id in (None, _mod_revision.NULL_REVISION):
764
next_id = revision_id
765
versionedfile = self._get_history_vf()
768
parents = versionedfile.get_parents(next_id)
769
if len(parents) == 0:
775
def get_revision_inventory(self, revision_id):
776
"""Return inventory of a past revision."""
777
# TODO: Unify this with get_inventory()
778
# bzr 0.0.6 and later imposes the constraint that the inventory_id
779
# must be the same as its revision, so this is trivial.
780
if revision_id is None:
781
# This does not make sense: if there is no revision,
782
# then it is the current tree inventory surely ?!
783
# and thus get_root_id() is something that looks at the last
784
# commit on the branch, and the get_root_id is an inventory check.
785
raise NotImplementedError
786
# return Inventory(self.get_root_id())
788
return self.get_inventory(revision_id)
792
"""Return True if this repository is flagged as a shared repository."""
793
raise NotImplementedError(self.is_shared)
796
def reconcile(self, other=None, thorough=False):
797
"""Reconcile this repository."""
798
from bzrlib.reconcile import RepoReconciler
799
reconciler = RepoReconciler(self, thorough=thorough)
800
reconciler.reconcile()
804
def revision_tree(self, revision_id):
805
"""Return Tree for a revision on this branch.
807
`revision_id` may be None for the empty tree revision.
809
# TODO: refactor this to use an existing revision object
810
# so we don't need to read it in twice.
811
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
812
return RevisionTree(self, Inventory(root_id=None),
813
_mod_revision.NULL_REVISION)
815
revision_id = osutils.safe_revision_id(revision_id)
816
inv = self.get_revision_inventory(revision_id)
817
return RevisionTree(self, inv, revision_id)
820
def revision_trees(self, revision_ids):
821
"""Return Tree for a revision on this branch.
823
`revision_id` may not be None or 'null:'"""
824
assert None not in revision_ids
825
assert _mod_revision.NULL_REVISION not in revision_ids
826
texts = self.get_inventory_weave().get_texts(revision_ids)
827
for text, revision_id in zip(texts, revision_ids):
828
inv = self.deserialise_inventory(revision_id, text)
829
yield RevisionTree(self, inv, revision_id)
832
def get_ancestry(self, revision_id, topo_sorted=True):
833
"""Return a list of revision-ids integrated by a revision.
835
The first element of the list is always None, indicating the origin
836
revision. This might change when we have history horizons, or
837
perhaps we should have a new API.
839
This is topologically sorted.
841
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
843
revision_id = osutils.safe_revision_id(revision_id)
844
if not self.has_revision(revision_id):
845
raise errors.NoSuchRevision(self, revision_id)
846
w = self.get_inventory_weave()
847
candidates = w.get_ancestry(revision_id, topo_sorted)
848
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
851
def print_file(self, file, revision_id):
852
"""Print `file` to stdout.
854
FIXME RBC 20060125 as John Meinel points out this is a bad api
855
- it writes to stdout, it assumes that that is valid etc. Fix
856
by creating a new more flexible convenience function.
858
revision_id = osutils.safe_revision_id(revision_id)
859
tree = self.revision_tree(revision_id)
860
# use inventory as it was in that revision
861
file_id = tree.inventory.path2id(file)
863
# TODO: jam 20060427 Write a test for this code path
864
# it had a bug in it, and was raising the wrong
866
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
867
tree.print_file(file_id)
869
def get_transaction(self):
870
return self.control_files.get_transaction()
872
def revision_parents(self, revision_id):
873
revision_id = osutils.safe_revision_id(revision_id)
874
return self.get_inventory_weave().parent_names(revision_id)
876
def get_parents(self, revision_ids):
877
"""See StackedParentsProvider.get_parents"""
879
for revision_id in revision_ids:
880
if revision_id == _mod_revision.NULL_REVISION:
884
parents = self.get_revision(revision_id).parent_ids
885
except errors.NoSuchRevision:
888
if len(parents) == 0:
889
parents = [_mod_revision.NULL_REVISION]
890
parents_list.append(parents)
893
def _make_parents_provider(self):
896
def get_graph(self, other_repository=None):
897
"""Return the graph walker for this repository format"""
898
parents_provider = self._make_parents_provider()
899
if (other_repository is not None and
900
other_repository.bzrdir.transport.base !=
901
self.bzrdir.transport.base):
902
parents_provider = graph._StackedParentsProvider(
903
[parents_provider, other_repository._make_parents_provider()])
904
return graph.Graph(parents_provider)
907
def set_make_working_trees(self, new_value):
908
"""Set the policy flag for making working trees when creating branches.
910
This only applies to branches that use this repository.
912
The default is 'True'.
913
:param new_value: True to restore the default, False to disable making
916
raise NotImplementedError(self.set_make_working_trees)
918
def make_working_trees(self):
919
"""Returns the policy for making working trees on new branches."""
920
raise NotImplementedError(self.make_working_trees)
923
def sign_revision(self, revision_id, gpg_strategy):
924
revision_id = osutils.safe_revision_id(revision_id)
925
plaintext = Testament.from_revision(self, revision_id).as_short_text()
926
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
929
def has_signature_for_revision_id(self, revision_id):
930
"""Query for a revision signature for revision_id in the repository."""
931
revision_id = osutils.safe_revision_id(revision_id)
932
return self._revision_store.has_signature(revision_id,
933
self.get_transaction())
936
def get_signature_text(self, revision_id):
937
"""Return the text for a signature."""
938
revision_id = osutils.safe_revision_id(revision_id)
939
return self._revision_store.get_signature_text(revision_id,
940
self.get_transaction())
943
def check(self, revision_ids):
944
"""Check consistency of all history of given revision_ids.
946
Different repository implementations should override _check().
948
:param revision_ids: A non-empty list of revision_ids whose ancestry
949
will be checked. Typically the last revision_id of a branch.
952
raise ValueError("revision_ids must be non-empty in %s.check"
954
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
955
return self._check(revision_ids)
957
def _check(self, revision_ids):
958
result = check.Check(self)
962
def _warn_if_deprecated(self):
963
global _deprecation_warning_done
964
if _deprecation_warning_done:
966
_deprecation_warning_done = True
967
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
968
% (self._format, self.bzrdir.transport.base))
970
def supports_rich_root(self):
971
return self._format.rich_root_data
973
def _check_ascii_revisionid(self, revision_id, method):
974
"""Private helper for ascii-only repositories."""
975
# weave repositories refuse to store revisionids that are non-ascii.
976
if revision_id is not None:
977
# weaves require ascii revision ids.
978
if isinstance(revision_id, unicode):
980
revision_id.encode('ascii')
981
except UnicodeEncodeError:
982
raise errors.NonAsciiRevisionId(method, self)
985
revision_id.decode('ascii')
986
except UnicodeDecodeError:
987
raise errors.NonAsciiRevisionId(method, self)
991
# remove these delegates a while after bzr 0.15
992
def __make_delegated(name, from_module):
993
def _deprecated_repository_forwarder():
994
symbol_versioning.warn('%s moved to %s in bzr 0.15'
995
% (name, from_module),
998
m = __import__(from_module, globals(), locals(), [name])
1000
return getattr(m, name)
1001
except AttributeError:
1002
raise AttributeError('module %s has no name %s'
1004
globals()[name] = _deprecated_repository_forwarder
1007
'AllInOneRepository',
1008
'WeaveMetaDirRepository',
1009
'PreSplitOutRepositoryFormat',
1010
'RepositoryFormat4',
1011
'RepositoryFormat5',
1012
'RepositoryFormat6',
1013
'RepositoryFormat7',
1015
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1019
'RepositoryFormatKnit',
1020
'RepositoryFormatKnit1',
1022
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1025
def install_revision(repository, rev, revision_tree):
1026
"""Install all revision data into a repository."""
1027
present_parents = []
1029
for p_id in rev.parent_ids:
1030
if repository.has_revision(p_id):
1031
present_parents.append(p_id)
1032
parent_trees[p_id] = repository.revision_tree(p_id)
1034
parent_trees[p_id] = repository.revision_tree(None)
1036
inv = revision_tree.inventory
1037
entries = inv.iter_entries()
1038
# backwards compatability hack: skip the root id.
1039
if not repository.supports_rich_root():
1040
path, root = entries.next()
1041
if root.revision != rev.revision_id:
1042
raise errors.IncompatibleRevision(repr(repository))
1043
# Add the texts that are not already present
1044
for path, ie in entries:
1045
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1046
repository.get_transaction())
1047
if ie.revision not in w:
1049
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1050
# with InventoryEntry.find_previous_heads(). if it is, then there
1051
# is a latent bug here where the parents may have ancestors of each
1053
for revision, tree in parent_trees.iteritems():
1054
if ie.file_id not in tree:
1056
parent_id = tree.inventory[ie.file_id].revision
1057
if parent_id in text_parents:
1059
text_parents.append(parent_id)
1061
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1062
repository.get_transaction())
1063
lines = revision_tree.get_file(ie.file_id).readlines()
1064
vfile.add_lines(rev.revision_id, text_parents, lines)
1066
# install the inventory
1067
repository.add_inventory(rev.revision_id, inv, present_parents)
1068
except errors.RevisionAlreadyPresent:
1070
repository.add_revision(rev.revision_id, rev, inv)
1073
class MetaDirRepository(Repository):
1074
"""Repositories in the new meta-dir layout."""
1076
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1077
super(MetaDirRepository, self).__init__(_format,
1083
dir_mode = self.control_files._dir_mode
1084
file_mode = self.control_files._file_mode
1087
def is_shared(self):
1088
"""Return True if this repository is flagged as a shared repository."""
1089
return self.control_files._transport.has('shared-storage')
1092
def set_make_working_trees(self, new_value):
1093
"""Set the policy flag for making working trees when creating branches.
1095
This only applies to branches that use this repository.
1097
The default is 'True'.
1098
:param new_value: True to restore the default, False to disable making
1103
self.control_files._transport.delete('no-working-trees')
1104
except errors.NoSuchFile:
1107
self.control_files.put_utf8('no-working-trees', '')
1109
def make_working_trees(self):
1110
"""Returns the policy for making working trees on new branches."""
1111
return not self.control_files._transport.has('no-working-trees')
1114
class RepositoryFormatRegistry(registry.Registry):
1115
"""Registry of RepositoryFormats.
1118
def get(self, format_string):
1119
r = registry.Registry.get(self, format_string)
1125
format_registry = RepositoryFormatRegistry()
1126
"""Registry of formats, indexed by their identifying format string.
1128
This can contain either format instances themselves, or classes/factories that
1129
can be called to obtain one.
1133
#####################################################################
1134
# Repository Formats
1136
class RepositoryFormat(object):
1137
"""A repository format.
1139
Formats provide three things:
1140
* An initialization routine to construct repository data on disk.
1141
* a format string which is used when the BzrDir supports versioned
1143
* an open routine which returns a Repository instance.
1145
Formats are placed in an dict by their format string for reference
1146
during opening. These should be subclasses of RepositoryFormat
1149
Once a format is deprecated, just deprecate the initialize and open
1150
methods on the format class. Do not deprecate the object, as the
1151
object will be created every system load.
1153
Common instance attributes:
1154
_matchingbzrdir - the bzrdir format that the repository format was
1155
originally written to work with. This can be used if manually
1156
constructing a bzrdir and repository, or more commonly for test suite
1161
return "<%s>" % self.__class__.__name__
1163
def __eq__(self, other):
1164
# format objects are generally stateless
1165
return isinstance(other, self.__class__)
1167
def __ne__(self, other):
1168
return not self == other
1171
def find_format(klass, a_bzrdir):
1172
"""Return the format for the repository object in a_bzrdir.
1174
This is used by bzr native formats that have a "format" file in
1175
the repository. Other methods may be used by different types of
1179
transport = a_bzrdir.get_repository_transport(None)
1180
format_string = transport.get("format").read()
1181
return format_registry.get(format_string)
1182
except errors.NoSuchFile:
1183
raise errors.NoRepositoryPresent(a_bzrdir)
1185
raise errors.UnknownFormatError(format=format_string)
1188
def register_format(klass, format):
1189
format_registry.register(format.get_format_string(), format)
1192
def unregister_format(klass, format):
1193
format_registry.remove(format.get_format_string())
1196
def get_default_format(klass):
1197
"""Return the current default format."""
1198
from bzrlib import bzrdir
1199
return bzrdir.format_registry.make_bzrdir('default').repository_format
1201
def _get_control_store(self, repo_transport, control_files):
1202
"""Return the control store for this repository."""
1203
raise NotImplementedError(self._get_control_store)
1205
def get_format_string(self):
1206
"""Return the ASCII format string that identifies this format.
1208
Note that in pre format ?? repositories the format string is
1209
not permitted nor written to disk.
1211
raise NotImplementedError(self.get_format_string)
1213
def get_format_description(self):
1214
"""Return the short description for this format."""
1215
raise NotImplementedError(self.get_format_description)
1217
def _get_revision_store(self, repo_transport, control_files):
1218
"""Return the revision store object for this a_bzrdir."""
1219
raise NotImplementedError(self._get_revision_store)
1221
def _get_text_rev_store(self,
1228
"""Common logic for getting a revision store for a repository.
1230
see self._get_revision_store for the subclass-overridable method to
1231
get the store for a repository.
1233
from bzrlib.store.revision.text import TextRevisionStore
1234
dir_mode = control_files._dir_mode
1235
file_mode = control_files._file_mode
1236
text_store = TextStore(transport.clone(name),
1238
compressed=compressed,
1240
file_mode=file_mode)
1241
_revision_store = TextRevisionStore(text_store, serializer)
1242
return _revision_store
1244
# TODO: this shouldn't be in the base class, it's specific to things that
1245
# use weaves or knits -- mbp 20070207
1246
def _get_versioned_file_store(self,
1251
versionedfile_class=None,
1252
versionedfile_kwargs={},
1254
if versionedfile_class is None:
1255
versionedfile_class = self._versionedfile_class
1256
weave_transport = control_files._transport.clone(name)
1257
dir_mode = control_files._dir_mode
1258
file_mode = control_files._file_mode
1259
return VersionedFileStore(weave_transport, prefixed=prefixed,
1261
file_mode=file_mode,
1262
versionedfile_class=versionedfile_class,
1263
versionedfile_kwargs=versionedfile_kwargs,
1266
def initialize(self, a_bzrdir, shared=False):
1267
"""Initialize a repository of this format in a_bzrdir.
1269
:param a_bzrdir: The bzrdir to put the new repository in it.
1270
:param shared: The repository should be initialized as a sharable one.
1271
:returns: The new repository object.
1273
This may raise UninitializableFormat if shared repository are not
1274
compatible the a_bzrdir.
1276
raise NotImplementedError(self.initialize)
1278
def is_supported(self):
1279
"""Is this format supported?
1281
Supported formats must be initializable and openable.
1282
Unsupported formats may not support initialization or committing or
1283
some other features depending on the reason for not being supported.
1287
def check_conversion_target(self, target_format):
1288
raise NotImplementedError(self.check_conversion_target)
1290
def open(self, a_bzrdir, _found=False):
1291
"""Return an instance of this format for the bzrdir a_bzrdir.
1293
_found is a private parameter, do not use it.
1295
raise NotImplementedError(self.open)
1298
class MetaDirRepositoryFormat(RepositoryFormat):
1299
"""Common base class for the new repositories using the metadir layout."""
1301
rich_root_data = False
1302
supports_tree_reference = False
1303
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1306
super(MetaDirRepositoryFormat, self).__init__()
1308
def _create_control_files(self, a_bzrdir):
1309
"""Create the required files and the initial control_files object."""
1310
# FIXME: RBC 20060125 don't peek under the covers
1311
# NB: no need to escape relative paths that are url safe.
1312
repository_transport = a_bzrdir.get_repository_transport(self)
1313
control_files = lockable_files.LockableFiles(repository_transport,
1314
'lock', lockdir.LockDir)
1315
control_files.create_lock()
1316
return control_files
1318
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1319
"""Upload the initial blank content."""
1320
control_files = self._create_control_files(a_bzrdir)
1321
control_files.lock_write()
1323
control_files._transport.mkdir_multi(dirs,
1324
mode=control_files._dir_mode)
1325
for file, content in files:
1326
control_files.put(file, content)
1327
for file, content in utf8_files:
1328
control_files.put_utf8(file, content)
1330
control_files.put_utf8('shared-storage', '')
1332
control_files.unlock()
1335
# formats which have no format string are not discoverable
1336
# and not independently creatable, so are not registered. They're
1337
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1338
# needed, it's constructed directly by the BzrDir. Non-native formats where
1339
# the repository is not separately opened are similar.
1341
format_registry.register_lazy(
1342
'Bazaar-NG Repository format 7',
1343
'bzrlib.repofmt.weaverepo',
1346
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1347
# default control directory format
1349
format_registry.register_lazy(
1350
'Bazaar-NG Knit Repository Format 1',
1351
'bzrlib.repofmt.knitrepo',
1352
'RepositoryFormatKnit1',
1354
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1356
format_registry.register_lazy(
1357
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1358
'bzrlib.repofmt.knitrepo',
1359
'RepositoryFormatKnit3',
1363
class InterRepository(InterObject):
1364
"""This class represents operations taking place between two repositories.
1366
Its instances have methods like copy_content and fetch, and contain
1367
references to the source and target repositories these operations can be
1370
Often we will provide convenience methods on 'repository' which carry out
1371
operations with another repository - they will always forward to
1372
InterRepository.get(other).method_name(parameters).
1376
"""The available optimised InterRepository types."""
1378
def copy_content(self, revision_id=None):
1379
raise NotImplementedError(self.copy_content)
1381
def fetch(self, revision_id=None, pb=None):
1382
"""Fetch the content required to construct revision_id.
1384
The content is copied from self.source to self.target.
1386
:param revision_id: if None all content is copied, if NULL_REVISION no
1388
:param pb: optional progress bar to use for progress reports. If not
1389
provided a default one will be created.
1391
Returns the copied revision count and the failed revisions in a tuple:
1394
raise NotImplementedError(self.fetch)
1397
def missing_revision_ids(self, revision_id=None):
1398
"""Return the revision ids that source has that target does not.
1400
These are returned in topological order.
1402
:param revision_id: only return revision ids included by this
1405
# generic, possibly worst case, slow code path.
1406
target_ids = set(self.target.all_revision_ids())
1407
if revision_id is not None:
1408
# TODO: jam 20070210 InterRepository is internal enough that it
1409
# should assume revision_ids are already utf-8
1410
revision_id = osutils.safe_revision_id(revision_id)
1411
source_ids = self.source.get_ancestry(revision_id)
1412
assert source_ids[0] is None
1415
source_ids = self.source.all_revision_ids()
1416
result_set = set(source_ids).difference(target_ids)
1417
# this may look like a no-op: its not. It preserves the ordering
1418
# other_ids had while only returning the members from other_ids
1419
# that we've decided we need.
1420
return [rev_id for rev_id in source_ids if rev_id in result_set]
1423
class InterSameDataRepository(InterRepository):
1424
"""Code for converting between repositories that represent the same data.
1426
Data format and model must match for this to work.
1430
def _get_repo_format_to_test(self):
1431
"""Repository format for testing with."""
1432
return RepositoryFormat.get_default_format()
1435
def is_compatible(source, target):
1436
if source.supports_rich_root() != target.supports_rich_root():
1438
if source._serializer != target._serializer:
1443
def copy_content(self, revision_id=None):
1444
"""Make a complete copy of the content in self into destination.
1446
This copies both the repository's revision data, and configuration information
1447
such as the make_working_trees setting.
1449
This is a destructive operation! Do not use it on existing
1452
:param revision_id: Only copy the content needed to construct
1453
revision_id and its parents.
1456
self.target.set_make_working_trees(self.source.make_working_trees())
1457
except NotImplementedError:
1459
# TODO: jam 20070210 This is fairly internal, so we should probably
1460
# just assert that revision_id is not unicode.
1461
revision_id = osutils.safe_revision_id(revision_id)
1462
# but don't bother fetching if we have the needed data now.
1463
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1464
self.target.has_revision(revision_id)):
1466
self.target.fetch(self.source, revision_id=revision_id)
1469
def fetch(self, revision_id=None, pb=None):
1470
"""See InterRepository.fetch()."""
1471
from bzrlib.fetch import GenericRepoFetcher
1472
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1473
self.source, self.source._format, self.target,
1474
self.target._format)
1475
# TODO: jam 20070210 This should be an assert, not a translate
1476
revision_id = osutils.safe_revision_id(revision_id)
1477
f = GenericRepoFetcher(to_repository=self.target,
1478
from_repository=self.source,
1479
last_revision=revision_id,
1481
return f.count_copied, f.failed_revisions
1484
class InterWeaveRepo(InterSameDataRepository):
1485
"""Optimised code paths between Weave based repositories."""
1488
def _get_repo_format_to_test(self):
1489
from bzrlib.repofmt import weaverepo
1490
return weaverepo.RepositoryFormat7()
1493
def is_compatible(source, target):
1494
"""Be compatible with known Weave formats.
1496
We don't test for the stores being of specific types because that
1497
could lead to confusing results, and there is no need to be
1500
from bzrlib.repofmt.weaverepo import (
1506
return (isinstance(source._format, (RepositoryFormat5,
1508
RepositoryFormat7)) and
1509
isinstance(target._format, (RepositoryFormat5,
1511
RepositoryFormat7)))
1512
except AttributeError:
1516
def copy_content(self, revision_id=None):
1517
"""See InterRepository.copy_content()."""
1518
# weave specific optimised path:
1519
# TODO: jam 20070210 Internal, should be an assert, not translate
1520
revision_id = osutils.safe_revision_id(revision_id)
1522
self.target.set_make_working_trees(self.source.make_working_trees())
1523
except NotImplementedError:
1525
# FIXME do not peek!
1526
if self.source.control_files._transport.listable():
1527
pb = ui.ui_factory.nested_progress_bar()
1529
self.target.weave_store.copy_all_ids(
1530
self.source.weave_store,
1532
from_transaction=self.source.get_transaction(),
1533
to_transaction=self.target.get_transaction())
1534
pb.update('copying inventory', 0, 1)
1535
self.target.control_weaves.copy_multi(
1536
self.source.control_weaves, ['inventory'],
1537
from_transaction=self.source.get_transaction(),
1538
to_transaction=self.target.get_transaction())
1539
self.target._revision_store.text_store.copy_all_ids(
1540
self.source._revision_store.text_store,
1545
self.target.fetch(self.source, revision_id=revision_id)
1548
def fetch(self, revision_id=None, pb=None):
1549
"""See InterRepository.fetch()."""
1550
from bzrlib.fetch import GenericRepoFetcher
1551
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1552
self.source, self.source._format, self.target, self.target._format)
1553
# TODO: jam 20070210 This should be an assert, not a translate
1554
revision_id = osutils.safe_revision_id(revision_id)
1555
f = GenericRepoFetcher(to_repository=self.target,
1556
from_repository=self.source,
1557
last_revision=revision_id,
1559
return f.count_copied, f.failed_revisions
1562
def missing_revision_ids(self, revision_id=None):
1563
"""See InterRepository.missing_revision_ids()."""
1564
# we want all revisions to satisfy revision_id in source.
1565
# but we don't want to stat every file here and there.
1566
# we want then, all revisions other needs to satisfy revision_id
1567
# checked, but not those that we have locally.
1568
# so the first thing is to get a subset of the revisions to
1569
# satisfy revision_id in source, and then eliminate those that
1570
# we do already have.
1571
# this is slow on high latency connection to self, but as as this
1572
# disk format scales terribly for push anyway due to rewriting
1573
# inventory.weave, this is considered acceptable.
1575
if revision_id is not None:
1576
source_ids = self.source.get_ancestry(revision_id)
1577
assert source_ids[0] is None
1580
source_ids = self.source._all_possible_ids()
1581
source_ids_set = set(source_ids)
1582
# source_ids is the worst possible case we may need to pull.
1583
# now we want to filter source_ids against what we actually
1584
# have in target, but don't try to check for existence where we know
1585
# we do not have a revision as that would be pointless.
1586
target_ids = set(self.target._all_possible_ids())
1587
possibly_present_revisions = target_ids.intersection(source_ids_set)
1588
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1589
required_revisions = source_ids_set.difference(actually_present_revisions)
1590
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1591
if revision_id is not None:
1592
# we used get_ancestry to determine source_ids then we are assured all
1593
# revisions referenced are present as they are installed in topological order.
1594
# and the tip revision was validated by get_ancestry.
1595
return required_topo_revisions
1597
# if we just grabbed the possibly available ids, then
1598
# we only have an estimate of whats available and need to validate
1599
# that against the revision records.
1600
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1603
class InterKnitRepo(InterSameDataRepository):
1604
"""Optimised code paths between Knit based repositories."""
1607
def _get_repo_format_to_test(self):
1608
from bzrlib.repofmt import knitrepo
1609
return knitrepo.RepositoryFormatKnit1()
1612
def is_compatible(source, target):
1613
"""Be compatible with known Knit formats.
1615
We don't test for the stores being of specific types because that
1616
could lead to confusing results, and there is no need to be
1619
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1621
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1622
isinstance(target._format, (RepositoryFormatKnit1)))
1623
except AttributeError:
1627
def fetch(self, revision_id=None, pb=None):
1628
"""See InterRepository.fetch()."""
1629
from bzrlib.fetch import KnitRepoFetcher
1630
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1631
self.source, self.source._format, self.target, self.target._format)
1632
# TODO: jam 20070210 This should be an assert, not a translate
1633
revision_id = osutils.safe_revision_id(revision_id)
1634
f = KnitRepoFetcher(to_repository=self.target,
1635
from_repository=self.source,
1636
last_revision=revision_id,
1638
return f.count_copied, f.failed_revisions
1641
def missing_revision_ids(self, revision_id=None):
1642
"""See InterRepository.missing_revision_ids()."""
1643
if revision_id is not None:
1644
source_ids = self.source.get_ancestry(revision_id)
1645
assert source_ids[0] is None
1648
source_ids = self.source._all_possible_ids()
1649
source_ids_set = set(source_ids)
1650
# source_ids is the worst possible case we may need to pull.
1651
# now we want to filter source_ids against what we actually
1652
# have in target, but don't try to check for existence where we know
1653
# we do not have a revision as that would be pointless.
1654
target_ids = set(self.target._all_possible_ids())
1655
possibly_present_revisions = target_ids.intersection(source_ids_set)
1656
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1657
required_revisions = source_ids_set.difference(actually_present_revisions)
1658
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1659
if revision_id is not None:
1660
# we used get_ancestry to determine source_ids then we are assured all
1661
# revisions referenced are present as they are installed in topological order.
1662
# and the tip revision was validated by get_ancestry.
1663
return required_topo_revisions
1665
# if we just grabbed the possibly available ids, then
1666
# we only have an estimate of whats available and need to validate
1667
# that against the revision records.
1668
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1671
class InterModel1and2(InterRepository):
1674
def _get_repo_format_to_test(self):
1678
def is_compatible(source, target):
1679
if not source.supports_rich_root() and target.supports_rich_root():
1685
def fetch(self, revision_id=None, pb=None):
1686
"""See InterRepository.fetch()."""
1687
from bzrlib.fetch import Model1toKnit2Fetcher
1688
# TODO: jam 20070210 This should be an assert, not a translate
1689
revision_id = osutils.safe_revision_id(revision_id)
1690
f = Model1toKnit2Fetcher(to_repository=self.target,
1691
from_repository=self.source,
1692
last_revision=revision_id,
1694
return f.count_copied, f.failed_revisions
1697
def copy_content(self, revision_id=None):
1698
"""Make a complete copy of the content in self into destination.
1700
This is a destructive operation! Do not use it on existing
1703
:param revision_id: Only copy the content needed to construct
1704
revision_id and its parents.
1707
self.target.set_make_working_trees(self.source.make_working_trees())
1708
except NotImplementedError:
1710
# TODO: jam 20070210 Internal, assert, don't translate
1711
revision_id = osutils.safe_revision_id(revision_id)
1712
# but don't bother fetching if we have the needed data now.
1713
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1714
self.target.has_revision(revision_id)):
1716
self.target.fetch(self.source, revision_id=revision_id)
1719
class InterKnit1and2(InterKnitRepo):
1722
def _get_repo_format_to_test(self):
1726
def is_compatible(source, target):
1727
"""Be compatible with Knit1 source and Knit3 target"""
1728
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1730
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1731
RepositoryFormatKnit3
1732
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1733
isinstance(target._format, (RepositoryFormatKnit3)))
1734
except AttributeError:
1738
def fetch(self, revision_id=None, pb=None):
1739
"""See InterRepository.fetch()."""
1740
from bzrlib.fetch import Knit1to2Fetcher
1741
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1742
self.source, self.source._format, self.target,
1743
self.target._format)
1744
# TODO: jam 20070210 This should be an assert, not a translate
1745
revision_id = osutils.safe_revision_id(revision_id)
1746
f = Knit1to2Fetcher(to_repository=self.target,
1747
from_repository=self.source,
1748
last_revision=revision_id,
1750
return f.count_copied, f.failed_revisions
1753
class InterRemoteRepository(InterRepository):
1754
"""Code for converting between RemoteRepository objects.
1756
This just gets an non-remote repository from the RemoteRepository, and calls
1757
InterRepository.get again.
1760
def __init__(self, source, target):
1761
if isinstance(source, remote.RemoteRepository):
1762
source._ensure_real()
1763
real_source = source._real_repository
1765
real_source = source
1766
if isinstance(target, remote.RemoteRepository):
1767
target._ensure_real()
1768
real_target = target._real_repository
1770
real_target = target
1771
self.real_inter = InterRepository.get(real_source, real_target)
1774
def is_compatible(source, target):
1775
if isinstance(source, remote.RemoteRepository):
1777
if isinstance(target, remote.RemoteRepository):
1781
def copy_content(self, revision_id=None):
1782
self.real_inter.copy_content(revision_id=revision_id)
1784
def fetch(self, revision_id=None, pb=None):
1785
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1788
def _get_repo_format_to_test(self):
1792
InterRepository.register_optimiser(InterSameDataRepository)
1793
InterRepository.register_optimiser(InterWeaveRepo)
1794
InterRepository.register_optimiser(InterKnitRepo)
1795
InterRepository.register_optimiser(InterModel1and2)
1796
InterRepository.register_optimiser(InterKnit1and2)
1797
InterRepository.register_optimiser(InterRemoteRepository)
1800
class RepositoryTestProviderAdapter(object):
1801
"""A tool to generate a suite testing multiple repository formats at once.
1803
This is done by copying the test once for each transport and injecting
1804
the transport_server, transport_readonly_server, and bzrdir_format and
1805
repository_format classes into each copy. Each copy is also given a new id()
1806
to make it easy to identify.
1809
def __init__(self, transport_server, transport_readonly_server, formats,
1810
vfs_transport_factory=None):
1811
self._transport_server = transport_server
1812
self._transport_readonly_server = transport_readonly_server
1813
self._vfs_transport_factory = vfs_transport_factory
1814
self._formats = formats
1816
def adapt(self, test):
1817
result = unittest.TestSuite()
1818
for repository_format, bzrdir_format 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
# Only override the test's vfs_transport_factory if one was
1824
# specified, otherwise just leave the default in place.
1825
if self._vfs_transport_factory:
1826
new_test.vfs_transport_factory = self._vfs_transport_factory
1827
new_test.bzrdir_format = bzrdir_format
1828
new_test.repository_format = repository_format
1829
def make_new_test_id():
1830
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1831
return lambda: new_id
1832
new_test.id = make_new_test_id()
1833
result.addTest(new_test)
1837
class InterRepositoryTestProviderAdapter(object):
1838
"""A tool to generate a suite testing multiple inter repository formats.
1840
This is done by copying the test once for each interrepo provider and injecting
1841
the transport_server, transport_readonly_server, repository_format and
1842
repository_to_format classes into each copy.
1843
Each copy is also given a new id() to make it easy to identify.
1846
def __init__(self, transport_server, transport_readonly_server, formats):
1847
self._transport_server = transport_server
1848
self._transport_readonly_server = transport_readonly_server
1849
self._formats = formats
1851
def adapt(self, test):
1852
result = unittest.TestSuite()
1853
for interrepo_class, repository_format, repository_format_to in self._formats:
1854
from copy import deepcopy
1855
new_test = deepcopy(test)
1856
new_test.transport_server = self._transport_server
1857
new_test.transport_readonly_server = self._transport_readonly_server
1858
new_test.interrepo_class = interrepo_class
1859
new_test.repository_format = repository_format
1860
new_test.repository_format_to = repository_format_to
1861
def make_new_test_id():
1862
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1863
return lambda: new_id
1864
new_test.id = make_new_test_id()
1865
result.addTest(new_test)
1869
def default_test_list():
1870
"""Generate the default list of interrepo permutations to test."""
1871
from bzrlib.repofmt import knitrepo, weaverepo
1873
# test the default InterRepository between format 6 and the current
1875
# XXX: robertc 20060220 reinstate this when there are two supported
1876
# formats which do not have an optimal code path between them.
1877
#result.append((InterRepository,
1878
# RepositoryFormat6(),
1879
# RepositoryFormatKnit1()))
1880
for optimiser_class in InterRepository._optimisers:
1881
format_to_test = optimiser_class._get_repo_format_to_test()
1882
if format_to_test is not None:
1883
result.append((optimiser_class,
1884
format_to_test, format_to_test))
1885
# if there are specific combinations we want to use, we can add them
1887
result.append((InterModel1and2,
1888
weaverepo.RepositoryFormat5(),
1889
knitrepo.RepositoryFormatKnit3()))
1890
result.append((InterKnit1and2,
1891
knitrepo.RepositoryFormatKnit1(),
1892
knitrepo.RepositoryFormatKnit3()))
1896
class CopyConverter(object):
1897
"""A repository conversion tool which just performs a copy of the content.
1899
This is slow but quite reliable.
1902
def __init__(self, target_format):
1903
"""Create a CopyConverter.
1905
:param target_format: The format the resulting repository should be.
1907
self.target_format = target_format
1909
def convert(self, repo, pb):
1910
"""Perform the conversion of to_convert, giving feedback via pb.
1912
:param to_convert: The disk object to convert.
1913
:param pb: a progress bar to use for progress information.
1918
# this is only useful with metadir layouts - separated repo content.
1919
# trigger an assertion if not such
1920
repo._format.get_format_string()
1921
self.repo_dir = repo.bzrdir
1922
self.step('Moving repository to repository.backup')
1923
self.repo_dir.transport.move('repository', 'repository.backup')
1924
backup_transport = self.repo_dir.transport.clone('repository.backup')
1925
repo._format.check_conversion_target(self.target_format)
1926
self.source_repo = repo._format.open(self.repo_dir,
1928
_override_transport=backup_transport)
1929
self.step('Creating new repository')
1930
converted = self.target_format.initialize(self.repo_dir,
1931
self.source_repo.is_shared())
1932
converted.lock_write()
1934
self.step('Copying content into repository.')
1935
self.source_repo.copy_content_into(converted)
1938
self.step('Deleting old repository content.')
1939
self.repo_dir.transport.delete_tree('repository.backup')
1940
self.pb.note('repository converted')
1942
def step(self, message):
1943
"""Update the pb by a step."""
1945
self.pb.update(message, self.count, self.total)
1948
class CommitBuilder(object):
1949
"""Provides an interface to build up a commit.
1951
This allows describing a tree to be committed without needing to
1952
know the internals of the format of the repository.
1955
record_root_entry = False
1956
def __init__(self, repository, parents, config, timestamp=None,
1957
timezone=None, committer=None, revprops=None,
1959
"""Initiate a CommitBuilder.
1961
:param repository: Repository to commit to.
1962
:param parents: Revision ids of the parents of the new revision.
1963
:param config: Configuration to use.
1964
:param timestamp: Optional timestamp recorded for commit.
1965
:param timezone: Optional timezone for timestamp.
1966
:param committer: Optional committer to set for commit.
1967
:param revprops: Optional dictionary of revision properties.
1968
:param revision_id: Optional revision id.
1970
self._config = config
1972
if committer is None:
1973
self._committer = self._config.username()
1975
assert isinstance(committer, basestring), type(committer)
1976
self._committer = committer
1978
self.new_inventory = Inventory(None)
1979
self._new_revision_id = osutils.safe_revision_id(revision_id)
1980
self.parents = parents
1981
self.repository = repository
1984
if revprops is not None:
1985
self._revprops.update(revprops)
1987
if timestamp is None:
1988
timestamp = time.time()
1989
# Restrict resolution to 1ms
1990
self._timestamp = round(timestamp, 3)
1992
if timezone is None:
1993
self._timezone = osutils.local_time_offset()
1995
self._timezone = int(timezone)
1997
self._generate_revision_if_needed()
1999
def commit(self, message):
2000
"""Make the actual commit.
2002
:return: The revision id of the recorded revision.
2004
rev = _mod_revision.Revision(
2005
timestamp=self._timestamp,
2006
timezone=self._timezone,
2007
committer=self._committer,
2009
inventory_sha1=self.inv_sha1,
2010
revision_id=self._new_revision_id,
2011
properties=self._revprops)
2012
rev.parent_ids = self.parents
2013
self.repository.add_revision(self._new_revision_id, rev,
2014
self.new_inventory, self._config)
2015
return self._new_revision_id
2017
def revision_tree(self):
2018
"""Return the tree that was just committed.
2020
After calling commit() this can be called to get a RevisionTree
2021
representing the newly committed tree. This is preferred to
2022
calling Repository.revision_tree() because that may require
2023
deserializing the inventory, while we already have a copy in
2026
return RevisionTree(self.repository, self.new_inventory,
2027
self._new_revision_id)
2029
def finish_inventory(self):
2030
"""Tell the builder that the inventory is finished."""
2031
if self.new_inventory.root is None:
2032
symbol_versioning.warn('Root entry should be supplied to'
2033
' record_entry_contents, as of bzr 0.10.',
2034
DeprecationWarning, stacklevel=2)
2035
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2036
self.new_inventory.revision_id = self._new_revision_id
2037
self.inv_sha1 = self.repository.add_inventory(
2038
self._new_revision_id,
2043
def _gen_revision_id(self):
2044
"""Return new revision-id."""
2045
return generate_ids.gen_revision_id(self._config.username(),
2048
def _generate_revision_if_needed(self):
2049
"""Create a revision id if None was supplied.
2051
If the repository can not support user-specified revision ids
2052
they should override this function and raise CannotSetRevisionId
2053
if _new_revision_id is not None.
2055
:raises: CannotSetRevisionId
2057
if self._new_revision_id is None:
2058
self._new_revision_id = self._gen_revision_id()
2060
def record_entry_contents(self, ie, parent_invs, path, tree):
2061
"""Record the content of ie from tree into the commit if needed.
2063
Side effect: sets ie.revision when unchanged
2065
:param ie: An inventory entry present in the commit.
2066
:param parent_invs: The inventories of the parent revisions of the
2068
:param path: The path the entry is at in the tree.
2069
:param tree: The tree which contains this entry and should be used to
2072
if self.new_inventory.root is None and ie.parent_id is not None:
2073
symbol_versioning.warn('Root entry should be supplied to'
2074
' record_entry_contents, as of bzr 0.10.',
2075
DeprecationWarning, stacklevel=2)
2076
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2078
self.new_inventory.add(ie)
2080
# ie.revision is always None if the InventoryEntry is considered
2081
# for committing. ie.snapshot will record the correct revision
2082
# which may be the sole parent if it is untouched.
2083
if ie.revision is not None:
2086
# In this revision format, root entries have no knit or weave
2087
if ie is self.new_inventory.root:
2088
# When serializing out to disk and back in
2089
# root.revision is always _new_revision_id
2090
ie.revision = self._new_revision_id
2092
previous_entries = ie.find_previous_heads(
2094
self.repository.weave_store,
2095
self.repository.get_transaction())
2096
# we are creating a new revision for ie in the history store
2098
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2100
def modified_directory(self, file_id, file_parents):
2101
"""Record the presence of a symbolic link.
2103
:param file_id: The file_id of the link to record.
2104
:param file_parents: The per-file parent revision ids.
2106
self._add_text_to_weave(file_id, [], file_parents.keys())
2108
def modified_reference(self, file_id, file_parents):
2109
"""Record the modification of a reference.
2111
:param file_id: The file_id of the link to record.
2112
:param file_parents: The per-file parent revision ids.
2114
self._add_text_to_weave(file_id, [], file_parents.keys())
2116
def modified_file_text(self, file_id, file_parents,
2117
get_content_byte_lines, text_sha1=None,
2119
"""Record the text of file file_id
2121
:param file_id: The file_id of the file to record the text of.
2122
:param file_parents: The per-file parent revision ids.
2123
:param get_content_byte_lines: A callable which will return the byte
2125
:param text_sha1: Optional SHA1 of the file contents.
2126
:param text_size: Optional size of the file contents.
2128
# mutter('storing text of file {%s} in revision {%s} into %r',
2129
# file_id, self._new_revision_id, self.repository.weave_store)
2130
# special case to avoid diffing on renames or
2132
if (len(file_parents) == 1
2133
and text_sha1 == file_parents.values()[0].text_sha1
2134
and text_size == file_parents.values()[0].text_size):
2135
previous_ie = file_parents.values()[0]
2136
versionedfile = self.repository.weave_store.get_weave(file_id,
2137
self.repository.get_transaction())
2138
versionedfile.clone_text(self._new_revision_id,
2139
previous_ie.revision, file_parents.keys())
2140
return text_sha1, text_size
2142
new_lines = get_content_byte_lines()
2143
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2144
# should return the SHA1 and size
2145
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2146
return osutils.sha_strings(new_lines), \
2147
sum(map(len, new_lines))
2149
def modified_link(self, file_id, file_parents, link_target):
2150
"""Record the presence of a symbolic link.
2152
:param file_id: The file_id of the link to record.
2153
:param file_parents: The per-file parent revision ids.
2154
:param link_target: Target location of this link.
2156
self._add_text_to_weave(file_id, [], file_parents.keys())
2158
def _add_text_to_weave(self, file_id, new_lines, parents):
2159
versionedfile = self.repository.weave_store.get_weave_or_empty(
2160
file_id, self.repository.get_transaction())
2161
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2162
versionedfile.clear_cache()
2165
class _CommitBuilder(CommitBuilder):
2166
"""Temporary class so old CommitBuilders are detected properly
2168
Note: CommitBuilder works whether or not root entry is recorded.
2171
record_root_entry = True
2174
class RootCommitBuilder(CommitBuilder):
2175
"""This commitbuilder actually records the root id"""
2177
record_root_entry = True
2179
def record_entry_contents(self, ie, parent_invs, path, tree):
2180
"""Record the content of ie from tree into the commit if needed.
2182
Side effect: sets ie.revision when unchanged
2184
:param ie: An inventory entry present in the commit.
2185
:param parent_invs: The inventories of the parent revisions of the
2187
:param path: The path the entry is at in the tree.
2188
:param tree: The tree which contains this entry and should be used to
2191
assert self.new_inventory.root is not None or ie.parent_id is None
2192
self.new_inventory.add(ie)
2194
# ie.revision is always None if the InventoryEntry is considered
2195
# for committing. ie.snapshot will record the correct revision
2196
# which may be the sole parent if it is untouched.
2197
if ie.revision is not None:
2200
previous_entries = ie.find_previous_heads(
2202
self.repository.weave_store,
2203
self.repository.get_transaction())
2204
# we are creating a new revision for ie in the history store
2206
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2218
def _unescaper(match, _map=_unescape_map):
2219
code = match.group(1)
2223
if not code.startswith('#'):
2225
return unichr(int(code[1:])).encode('utf8')
2231
def _unescape_xml(data):
2232
"""Unescape predefined XML entities in a string of data."""
2234
if _unescape_re is None:
2235
_unescape_re = re.compile('\&([^;]*);')
2236
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
