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(), """
37
revision as _mod_revision,
42
from bzrlib.revisiontree import RevisionTree
43
from bzrlib.store.versioned import VersionedFileStore
44
from bzrlib.store.text import TextStore
45
from bzrlib.testament import Testament
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
56
from bzrlib.trace import mutter, note, warning
59
# Old formats display a warning, but only once
60
_deprecation_warning_done = False
63
######################################################################
66
class Repository(object):
67
"""Repository holding history for one or more branches.
69
The repository holds and retrieves historical information including
70
revisions and file history. It's normally accessed only by the Branch,
71
which views a particular line of development through that history.
73
The Repository builds on top of Stores and a Transport, which respectively
74
describe the disk data format and the way of accessing the (possibly
78
_file_ids_altered_regex = lazy_regex.lazy_compile(
79
r'file_id="(?P<file_id>[^"]+)"'
80
r'.*revision="(?P<revision_id>[^"]+)"'
84
def add_inventory(self, revision_id, inv, parents):
85
"""Add the inventory inv to the repository as revision_id.
87
:param parents: The revision ids of the parents that revision_id
88
is known to have and are in the repository already.
90
returns the sha1 of the serialized inventory.
92
revision_id = osutils.safe_revision_id(revision_id)
93
_mod_revision.check_not_reserved_id(revision_id)
94
assert inv.revision_id is None or inv.revision_id == revision_id, \
95
"Mismatch between inventory revision" \
96
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
97
assert inv.root is not None
98
inv_text = self.serialise_inventory(inv)
99
inv_sha1 = osutils.sha_string(inv_text)
100
inv_vf = self.control_weaves.get_weave('inventory',
101
self.get_transaction())
102
self._inventory_add_lines(inv_vf, revision_id, parents,
103
osutils.split_lines(inv_text))
106
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
108
for parent in parents:
110
final_parents.append(parent)
112
inv_vf.add_lines(revision_id, final_parents, lines)
115
def add_revision(self, revision_id, rev, inv=None, config=None):
116
"""Add rev to the revision store as revision_id.
118
:param revision_id: the revision id to use.
119
:param rev: The revision object.
120
:param inv: The inventory for the revision. if None, it will be looked
121
up in the inventory storer
122
:param config: If None no digital signature will be created.
123
If supplied its signature_needed method will be used
124
to determine if a signature should be made.
126
revision_id = osutils.safe_revision_id(revision_id)
127
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
129
_mod_revision.check_not_reserved_id(revision_id)
130
if config is not None and config.signature_needed():
132
inv = self.get_inventory(revision_id)
133
plaintext = Testament(rev, inv).as_short_text()
134
self.store_revision_signature(
135
gpg.GPGStrategy(config), plaintext, revision_id)
136
if not revision_id in self.get_inventory_weave():
138
raise errors.WeaveRevisionNotPresent(revision_id,
139
self.get_inventory_weave())
141
# yes, this is not suitable for adding with ghosts.
142
self.add_inventory(revision_id, inv, rev.parent_ids)
143
self._revision_store.add_revision(rev, self.get_transaction())
146
def _all_possible_ids(self):
147
"""Return all the possible revisions that we could find."""
148
return self.get_inventory_weave().versions()
150
def all_revision_ids(self):
151
"""Returns a list of all the revision ids in the repository.
153
This is deprecated because code should generally work on the graph
154
reachable from a particular revision, and ignore any other revisions
155
that might be present. There is no direct replacement method.
157
return self._all_revision_ids()
160
def _all_revision_ids(self):
161
"""Returns a list of all the revision ids in the repository.
163
These are in as much topological order as the underlying store can
164
present: for weaves ghosts may lead to a lack of correctness until
165
the reweave updates the parents list.
167
if self._revision_store.text_store.listable():
168
return self._revision_store.all_revision_ids(self.get_transaction())
169
result = self._all_possible_ids()
170
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
171
# ids. (It should, since _revision_store's API should change to
172
# return utf8 revision_ids)
173
return self._eliminate_revisions_not_present(result)
175
def break_lock(self):
176
"""Break a lock if one is present from another instance.
178
Uses the ui factory to ask for confirmation if the lock may be from
181
self.control_files.break_lock()
184
def _eliminate_revisions_not_present(self, revision_ids):
185
"""Check every revision id in revision_ids to see if we have it.
187
Returns a set of the present revisions.
190
for id in revision_ids:
191
if self.has_revision(id):
196
def create(a_bzrdir):
197
"""Construct the current default format repository in a_bzrdir."""
198
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
200
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
201
"""instantiate a Repository.
203
:param _format: The format of the repository on disk.
204
:param a_bzrdir: The BzrDir of the repository.
206
In the future we will have a single api for all stores for
207
getting file texts, inventories and revisions, then
208
this construct will accept instances of those things.
210
super(Repository, self).__init__()
211
self._format = _format
212
# the following are part of the public API for Repository:
213
self.bzrdir = a_bzrdir
214
self.control_files = control_files
215
self._revision_store = _revision_store
216
self.text_store = text_store
217
# backwards compatibility
218
self.weave_store = text_store
219
# not right yet - should be more semantically clear ?
221
self.control_store = control_store
222
self.control_weaves = control_store
223
# TODO: make sure to construct the right store classes, etc, depending
224
# on whether escaping is required.
225
self._warn_if_deprecated()
228
return '%s(%r)' % (self.__class__.__name__,
229
self.bzrdir.transport.base)
232
return self.control_files.is_locked()
234
def lock_write(self, token=None):
235
"""Lock this repository for writing.
237
:param token: if this is already locked, then lock_write will fail
238
unless the token matches the existing lock.
239
:returns: a token if this instance supports tokens, otherwise None.
240
:raises TokenLockingNotSupported: when a token is given but this
241
instance doesn't support using token locks.
242
:raises MismatchedToken: if the specified token doesn't match the token
243
of the existing lock.
245
A token should be passed in if you know that you have locked the object
246
some other way, and need to synchronise this object's state with that
249
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
251
return self.control_files.lock_write(token=token)
254
self.control_files.lock_read()
256
def get_physical_lock_status(self):
257
return self.control_files.get_physical_lock_status()
259
def leave_lock_in_place(self):
260
"""Tell this repository not to release the physical lock when this
263
If lock_write doesn't return a token, then this method is not supported.
265
self.control_files.leave_in_place()
267
def dont_leave_lock_in_place(self):
268
"""Tell this repository to release the physical lock when this
269
object is unlocked, even if it didn't originally acquire it.
271
If lock_write doesn't return a token, then this method is not supported.
273
self.control_files.dont_leave_in_place()
276
def gather_stats(self, revid=None, committers=None):
277
"""Gather statistics from a revision id.
279
:param revid: The revision id to gather statistics from, if None, then
280
no revision specific statistics are gathered.
281
:param committers: Optional parameter controlling whether to grab
282
a count of committers from the revision specific statistics.
283
:return: A dictionary of statistics. Currently this contains:
284
committers: The number of committers if requested.
285
firstrev: A tuple with timestamp, timezone for the penultimate left
286
most ancestor of revid, if revid is not the NULL_REVISION.
287
latestrev: A tuple with timestamp, timezone for revid, if revid is
288
not the NULL_REVISION.
289
revisions: The total revision count in the repository.
290
size: An estimate disk size of the repository in bytes.
293
if revid and committers:
294
result['committers'] = 0
295
if revid and revid != _mod_revision.NULL_REVISION:
297
all_committers = set()
298
revisions = self.get_ancestry(revid)
299
# pop the leading None
301
first_revision = None
303
# ignore the revisions in the middle - just grab first and last
304
revisions = revisions[0], revisions[-1]
305
for revision in self.get_revisions(revisions):
306
if not first_revision:
307
first_revision = revision
309
all_committers.add(revision.committer)
310
last_revision = revision
312
result['committers'] = len(all_committers)
313
result['firstrev'] = (first_revision.timestamp,
314
first_revision.timezone)
315
result['latestrev'] = (last_revision.timestamp,
316
last_revision.timezone)
318
# now gather global repository information
319
if self.bzrdir.root_transport.listable():
320
c, t = self._revision_store.total_size(self.get_transaction())
321
result['revisions'] = c
326
def missing_revision_ids(self, other, revision_id=None):
327
"""Return the revision ids that other has that this does not.
329
These are returned in topological order.
331
revision_id: only return revision ids included by revision_id.
333
revision_id = osutils.safe_revision_id(revision_id)
334
return InterRepository.get(other, self).missing_revision_ids(revision_id)
338
"""Open the repository rooted at base.
340
For instance, if the repository is at URL/.bzr/repository,
341
Repository.open(URL) -> a Repository instance.
343
control = bzrdir.BzrDir.open(base)
344
return control.open_repository()
346
def copy_content_into(self, destination, revision_id=None):
347
"""Make a complete copy of the content in self into destination.
349
This is a destructive operation! Do not use it on existing
352
revision_id = osutils.safe_revision_id(revision_id)
353
return InterRepository.get(self, destination).copy_content(revision_id)
355
def fetch(self, source, revision_id=None, pb=None):
356
"""Fetch the content required to construct revision_id from source.
358
If revision_id is None all content is copied.
360
revision_id = osutils.safe_revision_id(revision_id)
361
inter = InterRepository.get(source, self)
363
return inter.fetch(revision_id=revision_id, pb=pb)
364
except NotImplementedError:
365
raise errors.IncompatibleRepositories(source, self)
367
def get_commit_builder(self, branch, parents, config, timestamp=None,
368
timezone=None, committer=None, revprops=None,
370
"""Obtain a CommitBuilder for this repository.
372
:param branch: Branch to commit to.
373
:param parents: Revision ids of the parents of the new revision.
374
:param config: Configuration to use.
375
:param timestamp: Optional timestamp recorded for commit.
376
:param timezone: Optional timezone for timestamp.
377
:param committer: Optional committer to set for commit.
378
:param revprops: Optional dictionary of revision properties.
379
:param revision_id: Optional revision id.
381
revision_id = osutils.safe_revision_id(revision_id)
382
return _CommitBuilder(self, parents, config, timestamp, timezone,
383
committer, revprops, revision_id)
386
self.control_files.unlock()
389
def clone(self, a_bzrdir, revision_id=None):
390
"""Clone this repository into a_bzrdir using the current format.
392
Currently no check is made that the format of this repository and
393
the bzrdir format are compatible. FIXME RBC 20060201.
395
:return: The newly created destination repository.
397
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
398
# use target default format.
399
dest_repo = a_bzrdir.create_repository()
401
# Most control formats need the repository to be specifically
402
# created, but on some old all-in-one formats it's not needed
404
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
405
except errors.UninitializableFormat:
406
dest_repo = a_bzrdir.open_repository()
407
self.copy_content_into(dest_repo, revision_id)
411
def has_revision(self, revision_id):
412
"""True if this repository has a copy of the revision."""
413
revision_id = osutils.safe_revision_id(revision_id)
414
return self._revision_store.has_revision_id(revision_id,
415
self.get_transaction())
418
def get_revision_reconcile(self, revision_id):
419
"""'reconcile' helper routine that allows access to a revision always.
421
This variant of get_revision does not cross check the weave graph
422
against the revision one as get_revision does: but it should only
423
be used by reconcile, or reconcile-alike commands that are correcting
424
or testing the revision graph.
426
if not revision_id or not isinstance(revision_id, basestring):
427
raise errors.InvalidRevisionId(revision_id=revision_id,
429
return self.get_revisions([revision_id])[0]
432
def get_revisions(self, revision_ids):
433
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
434
revs = self._revision_store.get_revisions(revision_ids,
435
self.get_transaction())
437
assert not isinstance(rev.revision_id, unicode)
438
for parent_id in rev.parent_ids:
439
assert not isinstance(parent_id, unicode)
443
def get_revision_xml(self, revision_id):
444
# TODO: jam 20070210 This shouldn't be necessary since get_revision
445
# would have already do it.
446
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
447
revision_id = osutils.safe_revision_id(revision_id)
448
rev = self.get_revision(revision_id)
450
# the current serializer..
451
self._revision_store._serializer.write_revision(rev, rev_tmp)
453
return rev_tmp.getvalue()
456
def get_revision(self, revision_id):
457
"""Return the Revision object for a named revision"""
458
# TODO: jam 20070210 get_revision_reconcile should do this for us
459
revision_id = osutils.safe_revision_id(revision_id)
460
r = self.get_revision_reconcile(revision_id)
461
# weave corruption can lead to absent revision markers that should be
463
# the following test is reasonably cheap (it needs a single weave read)
464
# and the weave is cached in read transactions. In write transactions
465
# it is not cached but typically we only read a small number of
466
# revisions. For knits when they are introduced we will probably want
467
# to ensure that caching write transactions are in use.
468
inv = self.get_inventory_weave()
469
self._check_revision_parents(r, inv)
473
def get_deltas_for_revisions(self, revisions):
474
"""Produce a generator of revision deltas.
476
Note that the input is a sequence of REVISIONS, not revision_ids.
477
Trees will be held in memory until the generator exits.
478
Each delta is relative to the revision's lefthand predecessor.
480
required_trees = set()
481
for revision in revisions:
482
required_trees.add(revision.revision_id)
483
required_trees.update(revision.parent_ids[:1])
484
trees = dict((t.get_revision_id(), t) for
485
t in self.revision_trees(required_trees))
486
for revision in revisions:
487
if not revision.parent_ids:
488
old_tree = self.revision_tree(None)
490
old_tree = trees[revision.parent_ids[0]]
491
yield trees[revision.revision_id].changes_from(old_tree)
494
def get_revision_delta(self, revision_id):
495
"""Return the delta for one revision.
497
The delta is relative to the left-hand predecessor of the
500
r = self.get_revision(revision_id)
501
return list(self.get_deltas_for_revisions([r]))[0]
503
def _check_revision_parents(self, revision, inventory):
504
"""Private to Repository and Fetch.
506
This checks the parentage of revision in an inventory weave for
507
consistency and is only applicable to inventory-weave-for-ancestry
508
using repository formats & fetchers.
510
weave_parents = inventory.get_parents(revision.revision_id)
511
weave_names = inventory.versions()
512
for parent_id in revision.parent_ids:
513
if parent_id in weave_names:
514
# this parent must not be a ghost.
515
if not parent_id in weave_parents:
517
raise errors.CorruptRepository(self)
520
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
521
revision_id = osutils.safe_revision_id(revision_id)
522
signature = gpg_strategy.sign(plaintext)
523
self._revision_store.add_revision_signature_text(revision_id,
525
self.get_transaction())
527
def fileids_altered_by_revision_ids(self, revision_ids):
528
"""Find the file ids and versions affected by revisions.
530
:param revisions: an iterable containing revision ids.
531
:return: a dictionary mapping altered file-ids to an iterable of
532
revision_ids. Each altered file-ids has the exact revision_ids that
533
altered it listed explicitly.
535
assert self._serializer.support_altered_by_hack, \
536
("fileids_altered_by_revision_ids only supported for branches "
537
"which store inventory as unnested xml, not on %r" % self)
538
selected_revision_ids = set(osutils.safe_revision_id(r)
539
for r in revision_ids)
540
w = self.get_inventory_weave()
543
# this code needs to read every new line in every inventory for the
544
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
545
# not present in one of those inventories is unnecessary but not
546
# harmful because we are filtering by the revision id marker in the
547
# inventory lines : we only select file ids altered in one of those
548
# revisions. We don't need to see all lines in the inventory because
549
# only those added in an inventory in rev X can contain a revision=X
551
unescape_revid_cache = {}
552
unescape_fileid_cache = {}
554
# jam 20061218 In a big fetch, this handles hundreds of thousands
555
# of lines, so it has had a lot of inlining and optimizing done.
556
# Sorry that it is a little bit messy.
557
# Move several functions to be local variables, since this is a long
559
search = self._file_ids_altered_regex.search
560
unescape = _unescape_xml
561
setdefault = result.setdefault
562
pb = ui.ui_factory.nested_progress_bar()
564
for line in w.iter_lines_added_or_present_in_versions(
565
selected_revision_ids, pb=pb):
569
# One call to match.group() returning multiple items is quite a
570
# bit faster than 2 calls to match.group() each returning 1
571
file_id, revision_id = match.group('file_id', 'revision_id')
573
# Inlining the cache lookups helps a lot when you make 170,000
574
# lines and 350k ids, versus 8.4 unique ids.
575
# Using a cache helps in 2 ways:
576
# 1) Avoids unnecessary decoding calls
577
# 2) Re-uses cached strings, which helps in future set and
579
# (2) is enough that removing encoding entirely along with
580
# the cache (so we are using plain strings) results in no
581
# performance improvement.
583
revision_id = unescape_revid_cache[revision_id]
585
unescaped = unescape(revision_id)
586
unescape_revid_cache[revision_id] = unescaped
587
revision_id = unescaped
589
if revision_id in selected_revision_ids:
591
file_id = unescape_fileid_cache[file_id]
593
unescaped = unescape(file_id)
594
unescape_fileid_cache[file_id] = unescaped
596
setdefault(file_id, set()).add(revision_id)
602
def get_inventory_weave(self):
603
return self.control_weaves.get_weave('inventory',
604
self.get_transaction())
607
def get_inventory(self, revision_id):
608
"""Get Inventory object by hash."""
609
# TODO: jam 20070210 Technically we don't need to sanitize, since all
610
# called functions must sanitize.
611
revision_id = osutils.safe_revision_id(revision_id)
612
return self.deserialise_inventory(
613
revision_id, self.get_inventory_xml(revision_id))
615
def deserialise_inventory(self, revision_id, xml):
616
"""Transform the xml into an inventory object.
618
:param revision_id: The expected revision id of the inventory.
619
:param xml: A serialised inventory.
621
revision_id = osutils.safe_revision_id(revision_id)
622
result = self._serializer.read_inventory_from_string(xml)
623
result.root.revision = revision_id
626
def serialise_inventory(self, inv):
627
return self._serializer.write_inventory_to_string(inv)
630
def get_inventory_xml(self, revision_id):
631
"""Get inventory XML as a file object."""
632
revision_id = osutils.safe_revision_id(revision_id)
634
assert isinstance(revision_id, str), type(revision_id)
635
iw = self.get_inventory_weave()
636
return iw.get_text(revision_id)
638
raise errors.HistoryMissing(self, 'inventory', revision_id)
641
def get_inventory_sha1(self, revision_id):
642
"""Return the sha1 hash of the inventory entry
644
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
645
revision_id = osutils.safe_revision_id(revision_id)
646
return self.get_revision(revision_id).inventory_sha1
649
def get_revision_graph(self, revision_id=None):
650
"""Return a dictionary containing the revision graph.
652
:param revision_id: The revision_id to get a graph from. If None, then
653
the entire revision graph is returned. This is a deprecated mode of
654
operation and will be removed in the future.
655
:return: a dictionary of revision_id->revision_parents_list.
657
# special case NULL_REVISION
658
if revision_id == _mod_revision.NULL_REVISION:
660
revision_id = osutils.safe_revision_id(revision_id)
661
a_weave = self.get_inventory_weave()
662
all_revisions = self._eliminate_revisions_not_present(
664
entire_graph = dict([(node, a_weave.get_parents(node)) for
665
node in all_revisions])
666
if revision_id is None:
668
elif revision_id not in entire_graph:
669
raise errors.NoSuchRevision(self, revision_id)
671
# add what can be reached from revision_id
673
pending = set([revision_id])
674
while len(pending) > 0:
676
result[node] = entire_graph[node]
677
for revision_id in result[node]:
678
if revision_id not in result:
679
pending.add(revision_id)
683
def get_revision_graph_with_ghosts(self, revision_ids=None):
684
"""Return a graph of the revisions with ghosts marked as applicable.
686
:param revision_ids: an iterable of revisions to graph or None for all.
687
:return: a Graph object with the graph reachable from revision_ids.
689
result = graph.Graph()
691
pending = set(self.all_revision_ids())
694
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
695
# special case NULL_REVISION
696
if _mod_revision.NULL_REVISION in pending:
697
pending.remove(_mod_revision.NULL_REVISION)
698
required = set(pending)
701
revision_id = pending.pop()
703
rev = self.get_revision(revision_id)
704
except errors.NoSuchRevision:
705
if revision_id in required:
708
result.add_ghost(revision_id)
710
for parent_id in rev.parent_ids:
711
# is this queued or done ?
712
if (parent_id not in pending and
713
parent_id not in done):
715
pending.add(parent_id)
716
result.add_node(revision_id, rev.parent_ids)
717
done.add(revision_id)
720
def _get_history_vf(self):
721
"""Get a versionedfile whose history graph reflects all revisions.
723
For weave repositories, this is the inventory weave.
725
return self.get_inventory_weave()
727
def iter_reverse_revision_history(self, revision_id):
728
"""Iterate backwards through revision ids in the lefthand history
730
:param revision_id: The revision id to start with. All its lefthand
731
ancestors will be traversed.
733
revision_id = osutils.safe_revision_id(revision_id)
734
if revision_id in (None, _mod_revision.NULL_REVISION):
736
next_id = revision_id
737
versionedfile = self._get_history_vf()
740
parents = versionedfile.get_parents(next_id)
741
if len(parents) == 0:
747
def get_revision_inventory(self, revision_id):
748
"""Return inventory of a past revision."""
749
# TODO: Unify this with get_inventory()
750
# bzr 0.0.6 and later imposes the constraint that the inventory_id
751
# must be the same as its revision, so this is trivial.
752
if revision_id is None:
753
# This does not make sense: if there is no revision,
754
# then it is the current tree inventory surely ?!
755
# and thus get_root_id() is something that looks at the last
756
# commit on the branch, and the get_root_id is an inventory check.
757
raise NotImplementedError
758
# return Inventory(self.get_root_id())
760
return self.get_inventory(revision_id)
764
"""Return True if this repository is flagged as a shared repository."""
765
raise NotImplementedError(self.is_shared)
768
def reconcile(self, other=None, thorough=False):
769
"""Reconcile this repository."""
770
from bzrlib.reconcile import RepoReconciler
771
reconciler = RepoReconciler(self, thorough=thorough)
772
reconciler.reconcile()
776
def revision_tree(self, revision_id):
777
"""Return Tree for a revision on this branch.
779
`revision_id` may be None for the empty tree revision.
781
# TODO: refactor this to use an existing revision object
782
# so we don't need to read it in twice.
783
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
784
return RevisionTree(self, Inventory(root_id=None),
785
_mod_revision.NULL_REVISION)
787
revision_id = osutils.safe_revision_id(revision_id)
788
inv = self.get_revision_inventory(revision_id)
789
return RevisionTree(self, inv, revision_id)
792
def revision_trees(self, revision_ids):
793
"""Return Tree for a revision on this branch.
795
`revision_id` may not be None or 'null:'"""
796
assert None not in revision_ids
797
assert _mod_revision.NULL_REVISION not in revision_ids
798
texts = self.get_inventory_weave().get_texts(revision_ids)
799
for text, revision_id in zip(texts, revision_ids):
800
inv = self.deserialise_inventory(revision_id, text)
801
yield RevisionTree(self, inv, revision_id)
804
def get_ancestry(self, revision_id):
805
"""Return a list of revision-ids integrated by a revision.
807
The first element of the list is always None, indicating the origin
808
revision. This might change when we have history horizons, or
809
perhaps we should have a new API.
811
This is topologically sorted.
813
if revision_id is None:
815
revision_id = osutils.safe_revision_id(revision_id)
816
if not self.has_revision(revision_id):
817
raise errors.NoSuchRevision(self, revision_id)
818
w = self.get_inventory_weave()
819
candidates = w.get_ancestry(revision_id)
820
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
823
def print_file(self, file, revision_id):
824
"""Print `file` to stdout.
826
FIXME RBC 20060125 as John Meinel points out this is a bad api
827
- it writes to stdout, it assumes that that is valid etc. Fix
828
by creating a new more flexible convenience function.
830
revision_id = osutils.safe_revision_id(revision_id)
831
tree = self.revision_tree(revision_id)
832
# use inventory as it was in that revision
833
file_id = tree.inventory.path2id(file)
835
# TODO: jam 20060427 Write a test for this code path
836
# it had a bug in it, and was raising the wrong
838
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
839
tree.print_file(file_id)
841
def get_transaction(self):
842
return self.control_files.get_transaction()
844
def revision_parents(self, revision_id):
845
revision_id = osutils.safe_revision_id(revision_id)
846
return self.get_inventory_weave().parent_names(revision_id)
849
def set_make_working_trees(self, new_value):
850
"""Set the policy flag for making working trees when creating branches.
852
This only applies to branches that use this repository.
854
The default is 'True'.
855
:param new_value: True to restore the default, False to disable making
858
raise NotImplementedError(self.set_make_working_trees)
860
def make_working_trees(self):
861
"""Returns the policy for making working trees on new branches."""
862
raise NotImplementedError(self.make_working_trees)
865
def sign_revision(self, revision_id, gpg_strategy):
866
revision_id = osutils.safe_revision_id(revision_id)
867
plaintext = Testament.from_revision(self, revision_id).as_short_text()
868
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
871
def has_signature_for_revision_id(self, revision_id):
872
"""Query for a revision signature for revision_id in the repository."""
873
revision_id = osutils.safe_revision_id(revision_id)
874
return self._revision_store.has_signature(revision_id,
875
self.get_transaction())
878
def get_signature_text(self, revision_id):
879
"""Return the text for a signature."""
880
revision_id = osutils.safe_revision_id(revision_id)
881
return self._revision_store.get_signature_text(revision_id,
882
self.get_transaction())
885
def check(self, revision_ids):
886
"""Check consistency of all history of given revision_ids.
888
Different repository implementations should override _check().
890
:param revision_ids: A non-empty list of revision_ids whose ancestry
891
will be checked. Typically the last revision_id of a branch.
894
raise ValueError("revision_ids must be non-empty in %s.check"
896
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
897
return self._check(revision_ids)
899
def _check(self, revision_ids):
900
result = check.Check(self)
904
def _warn_if_deprecated(self):
905
global _deprecation_warning_done
906
if _deprecation_warning_done:
908
_deprecation_warning_done = True
909
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
910
% (self._format, self.bzrdir.transport.base))
912
def supports_rich_root(self):
913
return self._format.rich_root_data
915
def _check_ascii_revisionid(self, revision_id, method):
916
"""Private helper for ascii-only repositories."""
917
# weave repositories refuse to store revisionids that are non-ascii.
918
if revision_id is not None:
919
# weaves require ascii revision ids.
920
if isinstance(revision_id, unicode):
922
revision_id.encode('ascii')
923
except UnicodeEncodeError:
924
raise errors.NonAsciiRevisionId(method, self)
927
revision_id.decode('ascii')
928
except UnicodeDecodeError:
929
raise errors.NonAsciiRevisionId(method, self)
933
# remove these delegates a while after bzr 0.15
934
def __make_delegated(name, from_module):
935
def _deprecated_repository_forwarder():
936
symbol_versioning.warn('%s moved to %s in bzr 0.15'
937
% (name, from_module),
940
m = __import__(from_module, globals(), locals(), [name])
942
return getattr(m, name)
943
except AttributeError:
944
raise AttributeError('module %s has no name %s'
946
globals()[name] = _deprecated_repository_forwarder
949
'AllInOneRepository',
950
'WeaveMetaDirRepository',
951
'PreSplitOutRepositoryFormat',
957
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
961
'RepositoryFormatKnit',
962
'RepositoryFormatKnit1',
964
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
967
def install_revision(repository, rev, revision_tree):
968
"""Install all revision data into a repository."""
971
for p_id in rev.parent_ids:
972
if repository.has_revision(p_id):
973
present_parents.append(p_id)
974
parent_trees[p_id] = repository.revision_tree(p_id)
976
parent_trees[p_id] = repository.revision_tree(None)
978
inv = revision_tree.inventory
979
entries = inv.iter_entries()
980
# backwards compatability hack: skip the root id.
981
if not repository.supports_rich_root():
982
path, root = entries.next()
983
if root.revision != rev.revision_id:
984
raise errors.IncompatibleRevision(repr(repository))
985
# Add the texts that are not already present
986
for path, ie in entries:
987
w = repository.weave_store.get_weave_or_empty(ie.file_id,
988
repository.get_transaction())
989
if ie.revision not in w:
991
# FIXME: TODO: The following loop *may* be overlapping/duplicate
992
# with InventoryEntry.find_previous_heads(). if it is, then there
993
# is a latent bug here where the parents may have ancestors of each
995
for revision, tree in parent_trees.iteritems():
996
if ie.file_id not in tree:
998
parent_id = tree.inventory[ie.file_id].revision
999
if parent_id in text_parents:
1001
text_parents.append(parent_id)
1003
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1004
repository.get_transaction())
1005
lines = revision_tree.get_file(ie.file_id).readlines()
1006
vfile.add_lines(rev.revision_id, text_parents, lines)
1008
# install the inventory
1009
repository.add_inventory(rev.revision_id, inv, present_parents)
1010
except errors.RevisionAlreadyPresent:
1012
repository.add_revision(rev.revision_id, rev, inv)
1015
class MetaDirRepository(Repository):
1016
"""Repositories in the new meta-dir layout."""
1018
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1019
super(MetaDirRepository, self).__init__(_format,
1025
dir_mode = self.control_files._dir_mode
1026
file_mode = self.control_files._file_mode
1029
def is_shared(self):
1030
"""Return True if this repository is flagged as a shared repository."""
1031
return self.control_files._transport.has('shared-storage')
1034
def set_make_working_trees(self, new_value):
1035
"""Set the policy flag for making working trees when creating branches.
1037
This only applies to branches that use this repository.
1039
The default is 'True'.
1040
:param new_value: True to restore the default, False to disable making
1045
self.control_files._transport.delete('no-working-trees')
1046
except errors.NoSuchFile:
1049
self.control_files.put_utf8('no-working-trees', '')
1051
def make_working_trees(self):
1052
"""Returns the policy for making working trees on new branches."""
1053
return not self.control_files._transport.has('no-working-trees')
1056
class RepositoryFormatRegistry(registry.Registry):
1057
"""Registry of RepositoryFormats.
1060
def get(self, format_string):
1061
r = registry.Registry.get(self, format_string)
1067
format_registry = RepositoryFormatRegistry()
1068
"""Registry of formats, indexed by their identifying format string.
1070
This can contain either format instances themselves, or classes/factories that
1071
can be called to obtain one.
1075
#####################################################################
1076
# Repository Formats
1078
class RepositoryFormat(object):
1079
"""A repository format.
1081
Formats provide three things:
1082
* An initialization routine to construct repository data on disk.
1083
* a format string which is used when the BzrDir supports versioned
1085
* an open routine which returns a Repository instance.
1087
Formats are placed in an dict by their format string for reference
1088
during opening. These should be subclasses of RepositoryFormat
1091
Once a format is deprecated, just deprecate the initialize and open
1092
methods on the format class. Do not deprecate the object, as the
1093
object will be created every system load.
1095
Common instance attributes:
1096
_matchingbzrdir - the bzrdir format that the repository format was
1097
originally written to work with. This can be used if manually
1098
constructing a bzrdir and repository, or more commonly for test suite
1103
return "<%s>" % self.__class__.__name__
1105
def __eq__(self, other):
1106
# format objects are generally stateless
1107
return isinstance(other, self.__class__)
1109
def __ne__(self, other):
1110
return not self == other
1113
def find_format(klass, a_bzrdir):
1114
"""Return the format for the repository object in a_bzrdir.
1116
This is used by bzr native formats that have a "format" file in
1117
the repository. Other methods may be used by different types of
1121
transport = a_bzrdir.get_repository_transport(None)
1122
format_string = transport.get("format").read()
1123
return format_registry.get(format_string)
1124
except errors.NoSuchFile:
1125
raise errors.NoRepositoryPresent(a_bzrdir)
1127
raise errors.UnknownFormatError(format=format_string)
1130
def register_format(klass, format):
1131
format_registry.register(format.get_format_string(), format)
1134
def unregister_format(klass, format):
1135
format_registry.remove(format.get_format_string())
1138
def get_default_format(klass):
1139
"""Return the current default format."""
1140
from bzrlib import bzrdir
1141
return bzrdir.format_registry.make_bzrdir('default').repository_format
1143
def _get_control_store(self, repo_transport, control_files):
1144
"""Return the control store for this repository."""
1145
raise NotImplementedError(self._get_control_store)
1147
def get_format_string(self):
1148
"""Return the ASCII format string that identifies this format.
1150
Note that in pre format ?? repositories the format string is
1151
not permitted nor written to disk.
1153
raise NotImplementedError(self.get_format_string)
1155
def get_format_description(self):
1156
"""Return the short description for this format."""
1157
raise NotImplementedError(self.get_format_description)
1159
def _get_revision_store(self, repo_transport, control_files):
1160
"""Return the revision store object for this a_bzrdir."""
1161
raise NotImplementedError(self._get_revision_store)
1163
def _get_text_rev_store(self,
1170
"""Common logic for getting a revision store for a repository.
1172
see self._get_revision_store for the subclass-overridable method to
1173
get the store for a repository.
1175
from bzrlib.store.revision.text import TextRevisionStore
1176
dir_mode = control_files._dir_mode
1177
file_mode = control_files._file_mode
1178
text_store = TextStore(transport.clone(name),
1180
compressed=compressed,
1182
file_mode=file_mode)
1183
_revision_store = TextRevisionStore(text_store, serializer)
1184
return _revision_store
1186
# TODO: this shouldn't be in the base class, it's specific to things that
1187
# use weaves or knits -- mbp 20070207
1188
def _get_versioned_file_store(self,
1193
versionedfile_class=None,
1194
versionedfile_kwargs={},
1196
if versionedfile_class is None:
1197
versionedfile_class = self._versionedfile_class
1198
weave_transport = control_files._transport.clone(name)
1199
dir_mode = control_files._dir_mode
1200
file_mode = control_files._file_mode
1201
return VersionedFileStore(weave_transport, prefixed=prefixed,
1203
file_mode=file_mode,
1204
versionedfile_class=versionedfile_class,
1205
versionedfile_kwargs=versionedfile_kwargs,
1208
def initialize(self, a_bzrdir, shared=False):
1209
"""Initialize a repository of this format in a_bzrdir.
1211
:param a_bzrdir: The bzrdir to put the new repository in it.
1212
:param shared: The repository should be initialized as a sharable one.
1213
:returns: The new repository object.
1215
This may raise UninitializableFormat if shared repository are not
1216
compatible the a_bzrdir.
1218
raise NotImplementedError(self.initialize)
1220
def is_supported(self):
1221
"""Is this format supported?
1223
Supported formats must be initializable and openable.
1224
Unsupported formats may not support initialization or committing or
1225
some other features depending on the reason for not being supported.
1229
def check_conversion_target(self, target_format):
1230
raise NotImplementedError(self.check_conversion_target)
1232
def open(self, a_bzrdir, _found=False):
1233
"""Return an instance of this format for the bzrdir a_bzrdir.
1235
_found is a private parameter, do not use it.
1237
raise NotImplementedError(self.open)
1240
class MetaDirRepositoryFormat(RepositoryFormat):
1241
"""Common base class for the new repositories using the metadir layout."""
1243
rich_root_data = False
1244
supports_tree_reference = False
1245
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1248
super(MetaDirRepositoryFormat, self).__init__()
1250
def _create_control_files(self, a_bzrdir):
1251
"""Create the required files and the initial control_files object."""
1252
# FIXME: RBC 20060125 don't peek under the covers
1253
# NB: no need to escape relative paths that are url safe.
1254
repository_transport = a_bzrdir.get_repository_transport(self)
1255
control_files = lockable_files.LockableFiles(repository_transport,
1256
'lock', lockdir.LockDir)
1257
control_files.create_lock()
1258
return control_files
1260
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1261
"""Upload the initial blank content."""
1262
control_files = self._create_control_files(a_bzrdir)
1263
control_files.lock_write()
1265
control_files._transport.mkdir_multi(dirs,
1266
mode=control_files._dir_mode)
1267
for file, content in files:
1268
control_files.put(file, content)
1269
for file, content in utf8_files:
1270
control_files.put_utf8(file, content)
1272
control_files.put_utf8('shared-storage', '')
1274
control_files.unlock()
1277
# formats which have no format string are not discoverable
1278
# and not independently creatable, so are not registered. They're
1279
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1280
# needed, it's constructed directly by the BzrDir. Non-native formats where
1281
# the repository is not separately opened are similar.
1283
format_registry.register_lazy(
1284
'Bazaar-NG Repository format 7',
1285
'bzrlib.repofmt.weaverepo',
1288
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1289
# default control directory format
1291
format_registry.register_lazy(
1292
'Bazaar-NG Knit Repository Format 1',
1293
'bzrlib.repofmt.knitrepo',
1294
'RepositoryFormatKnit1',
1296
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1298
format_registry.register_lazy(
1299
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1300
'bzrlib.repofmt.knitrepo',
1301
'RepositoryFormatKnit3',
1305
class InterRepository(InterObject):
1306
"""This class represents operations taking place between two repositories.
1308
Its instances have methods like copy_content and fetch, and contain
1309
references to the source and target repositories these operations can be
1312
Often we will provide convenience methods on 'repository' which carry out
1313
operations with another repository - they will always forward to
1314
InterRepository.get(other).method_name(parameters).
1318
"""The available optimised InterRepository types."""
1320
def copy_content(self, revision_id=None):
1321
raise NotImplementedError(self.copy_content)
1323
def fetch(self, revision_id=None, pb=None):
1324
"""Fetch the content required to construct revision_id.
1326
The content is copied from self.source to self.target.
1328
:param revision_id: if None all content is copied, if NULL_REVISION no
1330
:param pb: optional progress bar to use for progress reports. If not
1331
provided a default one will be created.
1333
Returns the copied revision count and the failed revisions in a tuple:
1336
raise NotImplementedError(self.fetch)
1339
def missing_revision_ids(self, revision_id=None):
1340
"""Return the revision ids that source has that target does not.
1342
These are returned in topological order.
1344
:param revision_id: only return revision ids included by this
1347
# generic, possibly worst case, slow code path.
1348
target_ids = set(self.target.all_revision_ids())
1349
if revision_id is not None:
1350
# TODO: jam 20070210 InterRepository is internal enough that it
1351
# should assume revision_ids are already utf-8
1352
revision_id = osutils.safe_revision_id(revision_id)
1353
source_ids = self.source.get_ancestry(revision_id)
1354
assert source_ids[0] is None
1357
source_ids = self.source.all_revision_ids()
1358
result_set = set(source_ids).difference(target_ids)
1359
# this may look like a no-op: its not. It preserves the ordering
1360
# other_ids had while only returning the members from other_ids
1361
# that we've decided we need.
1362
return [rev_id for rev_id in source_ids if rev_id in result_set]
1365
class InterSameDataRepository(InterRepository):
1366
"""Code for converting between repositories that represent the same data.
1368
Data format and model must match for this to work.
1372
def _get_repo_format_to_test(self):
1373
"""Repository format for testing with."""
1374
return RepositoryFormat.get_default_format()
1377
def is_compatible(source, target):
1378
if source.supports_rich_root() != target.supports_rich_root():
1380
if source._serializer != target._serializer:
1385
def copy_content(self, revision_id=None):
1386
"""Make a complete copy of the content in self into destination.
1388
This is a destructive operation! Do not use it on existing
1391
:param revision_id: Only copy the content needed to construct
1392
revision_id and its parents.
1395
self.target.set_make_working_trees(self.source.make_working_trees())
1396
except NotImplementedError:
1398
# TODO: jam 20070210 This is fairly internal, so we should probably
1399
# just assert that revision_id is not unicode.
1400
revision_id = osutils.safe_revision_id(revision_id)
1401
# but don't bother fetching if we have the needed data now.
1402
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1403
self.target.has_revision(revision_id)):
1405
self.target.fetch(self.source, revision_id=revision_id)
1408
def fetch(self, revision_id=None, pb=None):
1409
"""See InterRepository.fetch()."""
1410
from bzrlib.fetch import GenericRepoFetcher
1411
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1412
self.source, self.source._format, self.target,
1413
self.target._format)
1414
# TODO: jam 20070210 This should be an assert, not a translate
1415
revision_id = osutils.safe_revision_id(revision_id)
1416
f = GenericRepoFetcher(to_repository=self.target,
1417
from_repository=self.source,
1418
last_revision=revision_id,
1420
return f.count_copied, f.failed_revisions
1423
class InterWeaveRepo(InterSameDataRepository):
1424
"""Optimised code paths between Weave based repositories."""
1427
def _get_repo_format_to_test(self):
1428
from bzrlib.repofmt import weaverepo
1429
return weaverepo.RepositoryFormat7()
1432
def is_compatible(source, target):
1433
"""Be compatible with known Weave formats.
1435
We don't test for the stores being of specific types because that
1436
could lead to confusing results, and there is no need to be
1439
from bzrlib.repofmt.weaverepo import (
1445
return (isinstance(source._format, (RepositoryFormat5,
1447
RepositoryFormat7)) and
1448
isinstance(target._format, (RepositoryFormat5,
1450
RepositoryFormat7)))
1451
except AttributeError:
1455
def copy_content(self, revision_id=None):
1456
"""See InterRepository.copy_content()."""
1457
# weave specific optimised path:
1458
# TODO: jam 20070210 Internal, should be an assert, not translate
1459
revision_id = osutils.safe_revision_id(revision_id)
1461
self.target.set_make_working_trees(self.source.make_working_trees())
1462
except NotImplementedError:
1464
# FIXME do not peek!
1465
if self.source.control_files._transport.listable():
1466
pb = ui.ui_factory.nested_progress_bar()
1468
self.target.weave_store.copy_all_ids(
1469
self.source.weave_store,
1471
from_transaction=self.source.get_transaction(),
1472
to_transaction=self.target.get_transaction())
1473
pb.update('copying inventory', 0, 1)
1474
self.target.control_weaves.copy_multi(
1475
self.source.control_weaves, ['inventory'],
1476
from_transaction=self.source.get_transaction(),
1477
to_transaction=self.target.get_transaction())
1478
self.target._revision_store.text_store.copy_all_ids(
1479
self.source._revision_store.text_store,
1484
self.target.fetch(self.source, revision_id=revision_id)
1487
def fetch(self, revision_id=None, pb=None):
1488
"""See InterRepository.fetch()."""
1489
from bzrlib.fetch import GenericRepoFetcher
1490
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1491
self.source, self.source._format, self.target, self.target._format)
1492
# TODO: jam 20070210 This should be an assert, not a translate
1493
revision_id = osutils.safe_revision_id(revision_id)
1494
f = GenericRepoFetcher(to_repository=self.target,
1495
from_repository=self.source,
1496
last_revision=revision_id,
1498
return f.count_copied, f.failed_revisions
1501
def missing_revision_ids(self, revision_id=None):
1502
"""See InterRepository.missing_revision_ids()."""
1503
# we want all revisions to satisfy revision_id in source.
1504
# but we don't want to stat every file here and there.
1505
# we want then, all revisions other needs to satisfy revision_id
1506
# checked, but not those that we have locally.
1507
# so the first thing is to get a subset of the revisions to
1508
# satisfy revision_id in source, and then eliminate those that
1509
# we do already have.
1510
# this is slow on high latency connection to self, but as as this
1511
# disk format scales terribly for push anyway due to rewriting
1512
# inventory.weave, this is considered acceptable.
1514
if revision_id is not None:
1515
source_ids = self.source.get_ancestry(revision_id)
1516
assert source_ids[0] is None
1519
source_ids = self.source._all_possible_ids()
1520
source_ids_set = set(source_ids)
1521
# source_ids is the worst possible case we may need to pull.
1522
# now we want to filter source_ids against what we actually
1523
# have in target, but don't try to check for existence where we know
1524
# we do not have a revision as that would be pointless.
1525
target_ids = set(self.target._all_possible_ids())
1526
possibly_present_revisions = target_ids.intersection(source_ids_set)
1527
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1528
required_revisions = source_ids_set.difference(actually_present_revisions)
1529
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1530
if revision_id is not None:
1531
# we used get_ancestry to determine source_ids then we are assured all
1532
# revisions referenced are present as they are installed in topological order.
1533
# and the tip revision was validated by get_ancestry.
1534
return required_topo_revisions
1536
# if we just grabbed the possibly available ids, then
1537
# we only have an estimate of whats available and need to validate
1538
# that against the revision records.
1539
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1542
class InterKnitRepo(InterSameDataRepository):
1543
"""Optimised code paths between Knit based repositories."""
1546
def _get_repo_format_to_test(self):
1547
from bzrlib.repofmt import knitrepo
1548
return knitrepo.RepositoryFormatKnit1()
1551
def is_compatible(source, target):
1552
"""Be compatible with known Knit formats.
1554
We don't test for the stores being of specific types because that
1555
could lead to confusing results, and there is no need to be
1558
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1560
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1561
isinstance(target._format, (RepositoryFormatKnit1)))
1562
except AttributeError:
1566
def fetch(self, revision_id=None, pb=None):
1567
"""See InterRepository.fetch()."""
1568
from bzrlib.fetch import KnitRepoFetcher
1569
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1570
self.source, self.source._format, self.target, self.target._format)
1571
# TODO: jam 20070210 This should be an assert, not a translate
1572
revision_id = osutils.safe_revision_id(revision_id)
1573
f = KnitRepoFetcher(to_repository=self.target,
1574
from_repository=self.source,
1575
last_revision=revision_id,
1577
return f.count_copied, f.failed_revisions
1580
def missing_revision_ids(self, revision_id=None):
1581
"""See InterRepository.missing_revision_ids()."""
1582
if revision_id is not None:
1583
source_ids = self.source.get_ancestry(revision_id)
1584
assert source_ids[0] is None
1587
source_ids = self.source._all_possible_ids()
1588
source_ids_set = set(source_ids)
1589
# source_ids is the worst possible case we may need to pull.
1590
# now we want to filter source_ids against what we actually
1591
# have in target, but don't try to check for existence where we know
1592
# we do not have a revision as that would be pointless.
1593
target_ids = set(self.target._all_possible_ids())
1594
possibly_present_revisions = target_ids.intersection(source_ids_set)
1595
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1596
required_revisions = source_ids_set.difference(actually_present_revisions)
1597
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1598
if revision_id is not None:
1599
# we used get_ancestry to determine source_ids then we are assured all
1600
# revisions referenced are present as they are installed in topological order.
1601
# and the tip revision was validated by get_ancestry.
1602
return required_topo_revisions
1604
# if we just grabbed the possibly available ids, then
1605
# we only have an estimate of whats available and need to validate
1606
# that against the revision records.
1607
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1610
class InterModel1and2(InterRepository):
1613
def _get_repo_format_to_test(self):
1617
def is_compatible(source, target):
1618
if not source.supports_rich_root() and target.supports_rich_root():
1624
def fetch(self, revision_id=None, pb=None):
1625
"""See InterRepository.fetch()."""
1626
from bzrlib.fetch import Model1toKnit2Fetcher
1627
# TODO: jam 20070210 This should be an assert, not a translate
1628
revision_id = osutils.safe_revision_id(revision_id)
1629
f = Model1toKnit2Fetcher(to_repository=self.target,
1630
from_repository=self.source,
1631
last_revision=revision_id,
1633
return f.count_copied, f.failed_revisions
1636
def copy_content(self, revision_id=None):
1637
"""Make a complete copy of the content in self into destination.
1639
This is a destructive operation! Do not use it on existing
1642
:param revision_id: Only copy the content needed to construct
1643
revision_id and its parents.
1646
self.target.set_make_working_trees(self.source.make_working_trees())
1647
except NotImplementedError:
1649
# TODO: jam 20070210 Internal, assert, don't translate
1650
revision_id = osutils.safe_revision_id(revision_id)
1651
# but don't bother fetching if we have the needed data now.
1652
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1653
self.target.has_revision(revision_id)):
1655
self.target.fetch(self.source, revision_id=revision_id)
1658
class InterKnit1and2(InterKnitRepo):
1661
def _get_repo_format_to_test(self):
1665
def is_compatible(source, target):
1666
"""Be compatible with Knit1 source and Knit3 target"""
1667
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1669
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1670
RepositoryFormatKnit3
1671
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1672
isinstance(target._format, (RepositoryFormatKnit3)))
1673
except AttributeError:
1677
def fetch(self, revision_id=None, pb=None):
1678
"""See InterRepository.fetch()."""
1679
from bzrlib.fetch import Knit1to2Fetcher
1680
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1681
self.source, self.source._format, self.target,
1682
self.target._format)
1683
# TODO: jam 20070210 This should be an assert, not a translate
1684
revision_id = osutils.safe_revision_id(revision_id)
1685
f = Knit1to2Fetcher(to_repository=self.target,
1686
from_repository=self.source,
1687
last_revision=revision_id,
1689
return f.count_copied, f.failed_revisions
1692
InterRepository.register_optimiser(InterSameDataRepository)
1693
InterRepository.register_optimiser(InterWeaveRepo)
1694
InterRepository.register_optimiser(InterKnitRepo)
1695
InterRepository.register_optimiser(InterModel1and2)
1696
InterRepository.register_optimiser(InterKnit1and2)
1699
class RepositoryTestProviderAdapter(object):
1700
"""A tool to generate a suite testing multiple repository formats at once.
1702
This is done by copying the test once for each transport and injecting
1703
the transport_server, transport_readonly_server, and bzrdir_format and
1704
repository_format classes into each copy. Each copy is also given a new id()
1705
to make it easy to identify.
1708
def __init__(self, transport_server, transport_readonly_server, formats,
1709
vfs_transport_factory=None):
1710
self._transport_server = transport_server
1711
self._transport_readonly_server = transport_readonly_server
1712
self._vfs_transport_factory = vfs_transport_factory
1713
self._formats = formats
1715
def adapt(self, test):
1716
result = unittest.TestSuite()
1717
for repository_format, bzrdir_format in self._formats:
1718
from copy import deepcopy
1719
new_test = deepcopy(test)
1720
new_test.transport_server = self._transport_server
1721
new_test.transport_readonly_server = self._transport_readonly_server
1722
if self._vfs_transport_factory:
1723
new_test.vfs_transport_factory = self._vfs_transport_factory
1724
new_test.bzrdir_format = bzrdir_format
1725
new_test.repository_format = repository_format
1726
def make_new_test_id():
1727
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1728
return lambda: new_id
1729
new_test.id = make_new_test_id()
1730
result.addTest(new_test)
1734
class InterRepositoryTestProviderAdapter(object):
1735
"""A tool to generate a suite testing multiple inter repository formats.
1737
This is done by copying the test once for each interrepo provider and injecting
1738
the transport_server, transport_readonly_server, repository_format and
1739
repository_to_format classes into each copy.
1740
Each copy is also given a new id() to make it easy to identify.
1743
def __init__(self, transport_server, transport_readonly_server, formats):
1744
self._transport_server = transport_server
1745
self._transport_readonly_server = transport_readonly_server
1746
self._formats = formats
1748
def adapt(self, test):
1749
result = unittest.TestSuite()
1750
for interrepo_class, repository_format, repository_format_to in self._formats:
1751
from copy import deepcopy
1752
new_test = deepcopy(test)
1753
new_test.transport_server = self._transport_server
1754
new_test.transport_readonly_server = self._transport_readonly_server
1755
new_test.interrepo_class = interrepo_class
1756
new_test.repository_format = repository_format
1757
new_test.repository_format_to = repository_format_to
1758
def make_new_test_id():
1759
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1760
return lambda: new_id
1761
new_test.id = make_new_test_id()
1762
result.addTest(new_test)
1766
def default_test_list():
1767
"""Generate the default list of interrepo permutations to test."""
1768
from bzrlib.repofmt import knitrepo, weaverepo
1770
# test the default InterRepository between format 6 and the current
1772
# XXX: robertc 20060220 reinstate this when there are two supported
1773
# formats which do not have an optimal code path between them.
1774
#result.append((InterRepository,
1775
# RepositoryFormat6(),
1776
# RepositoryFormatKnit1()))
1777
for optimiser_class in InterRepository._optimisers:
1778
format_to_test = optimiser_class._get_repo_format_to_test()
1779
if format_to_test is not None:
1780
result.append((optimiser_class,
1781
format_to_test, format_to_test))
1782
# if there are specific combinations we want to use, we can add them
1784
result.append((InterModel1and2,
1785
weaverepo.RepositoryFormat5(),
1786
knitrepo.RepositoryFormatKnit3()))
1787
result.append((InterKnit1and2,
1788
knitrepo.RepositoryFormatKnit1(),
1789
knitrepo.RepositoryFormatKnit3()))
1793
class CopyConverter(object):
1794
"""A repository conversion tool which just performs a copy of the content.
1796
This is slow but quite reliable.
1799
def __init__(self, target_format):
1800
"""Create a CopyConverter.
1802
:param target_format: The format the resulting repository should be.
1804
self.target_format = target_format
1806
def convert(self, repo, pb):
1807
"""Perform the conversion of to_convert, giving feedback via pb.
1809
:param to_convert: The disk object to convert.
1810
:param pb: a progress bar to use for progress information.
1815
# this is only useful with metadir layouts - separated repo content.
1816
# trigger an assertion if not such
1817
repo._format.get_format_string()
1818
self.repo_dir = repo.bzrdir
1819
self.step('Moving repository to repository.backup')
1820
self.repo_dir.transport.move('repository', 'repository.backup')
1821
backup_transport = self.repo_dir.transport.clone('repository.backup')
1822
repo._format.check_conversion_target(self.target_format)
1823
self.source_repo = repo._format.open(self.repo_dir,
1825
_override_transport=backup_transport)
1826
self.step('Creating new repository')
1827
converted = self.target_format.initialize(self.repo_dir,
1828
self.source_repo.is_shared())
1829
converted.lock_write()
1831
self.step('Copying content into repository.')
1832
self.source_repo.copy_content_into(converted)
1835
self.step('Deleting old repository content.')
1836
self.repo_dir.transport.delete_tree('repository.backup')
1837
self.pb.note('repository converted')
1839
def step(self, message):
1840
"""Update the pb by a step."""
1842
self.pb.update(message, self.count, self.total)
1845
class CommitBuilder(object):
1846
"""Provides an interface to build up a commit.
1848
This allows describing a tree to be committed without needing to
1849
know the internals of the format of the repository.
1852
record_root_entry = False
1853
def __init__(self, repository, parents, config, timestamp=None,
1854
timezone=None, committer=None, revprops=None,
1856
"""Initiate a CommitBuilder.
1858
:param repository: Repository to commit to.
1859
:param parents: Revision ids of the parents of the new revision.
1860
:param config: Configuration to use.
1861
:param timestamp: Optional timestamp recorded for commit.
1862
:param timezone: Optional timezone for timestamp.
1863
:param committer: Optional committer to set for commit.
1864
:param revprops: Optional dictionary of revision properties.
1865
:param revision_id: Optional revision id.
1867
self._config = config
1869
if committer is None:
1870
self._committer = self._config.username()
1872
assert isinstance(committer, basestring), type(committer)
1873
self._committer = committer
1875
self.new_inventory = Inventory(None)
1876
self._new_revision_id = osutils.safe_revision_id(revision_id)
1877
self.parents = parents
1878
self.repository = repository
1881
if revprops is not None:
1882
self._revprops.update(revprops)
1884
if timestamp is None:
1885
timestamp = time.time()
1886
# Restrict resolution to 1ms
1887
self._timestamp = round(timestamp, 3)
1889
if timezone is None:
1890
self._timezone = osutils.local_time_offset()
1892
self._timezone = int(timezone)
1894
self._generate_revision_if_needed()
1896
def commit(self, message):
1897
"""Make the actual commit.
1899
:return: The revision id of the recorded revision.
1901
rev = _mod_revision.Revision(
1902
timestamp=self._timestamp,
1903
timezone=self._timezone,
1904
committer=self._committer,
1906
inventory_sha1=self.inv_sha1,
1907
revision_id=self._new_revision_id,
1908
properties=self._revprops)
1909
rev.parent_ids = self.parents
1910
self.repository.add_revision(self._new_revision_id, rev,
1911
self.new_inventory, self._config)
1912
return self._new_revision_id
1914
def revision_tree(self):
1915
"""Return the tree that was just committed.
1917
After calling commit() this can be called to get a RevisionTree
1918
representing the newly committed tree. This is preferred to
1919
calling Repository.revision_tree() because that may require
1920
deserializing the inventory, while we already have a copy in
1923
return RevisionTree(self.repository, self.new_inventory,
1924
self._new_revision_id)
1926
def finish_inventory(self):
1927
"""Tell the builder that the inventory is finished."""
1928
if self.new_inventory.root is None:
1929
symbol_versioning.warn('Root entry should be supplied to'
1930
' record_entry_contents, as of bzr 0.10.',
1931
DeprecationWarning, stacklevel=2)
1932
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1933
self.new_inventory.revision_id = self._new_revision_id
1934
self.inv_sha1 = self.repository.add_inventory(
1935
self._new_revision_id,
1940
def _gen_revision_id(self):
1941
"""Return new revision-id."""
1942
return generate_ids.gen_revision_id(self._config.username(),
1945
def _generate_revision_if_needed(self):
1946
"""Create a revision id if None was supplied.
1948
If the repository can not support user-specified revision ids
1949
they should override this function and raise CannotSetRevisionId
1950
if _new_revision_id is not None.
1952
:raises: CannotSetRevisionId
1954
if self._new_revision_id is None:
1955
self._new_revision_id = self._gen_revision_id()
1957
def record_entry_contents(self, ie, parent_invs, path, tree):
1958
"""Record the content of ie from tree into the commit if needed.
1960
Side effect: sets ie.revision when unchanged
1962
:param ie: An inventory entry present in the commit.
1963
:param parent_invs: The inventories of the parent revisions of the
1965
:param path: The path the entry is at in the tree.
1966
:param tree: The tree which contains this entry and should be used to
1969
if self.new_inventory.root is None and ie.parent_id is not None:
1970
symbol_versioning.warn('Root entry should be supplied to'
1971
' record_entry_contents, as of bzr 0.10.',
1972
DeprecationWarning, stacklevel=2)
1973
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1975
self.new_inventory.add(ie)
1977
# ie.revision is always None if the InventoryEntry is considered
1978
# for committing. ie.snapshot will record the correct revision
1979
# which may be the sole parent if it is untouched.
1980
if ie.revision is not None:
1983
# In this revision format, root entries have no knit or weave
1984
if ie is self.new_inventory.root:
1985
# When serializing out to disk and back in
1986
# root.revision is always _new_revision_id
1987
ie.revision = self._new_revision_id
1989
previous_entries = ie.find_previous_heads(
1991
self.repository.weave_store,
1992
self.repository.get_transaction())
1993
# we are creating a new revision for ie in the history store
1995
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1997
def modified_directory(self, file_id, file_parents):
1998
"""Record the presence of a symbolic link.
2000
:param file_id: The file_id of the link to record.
2001
:param file_parents: The per-file parent revision ids.
2003
self._add_text_to_weave(file_id, [], file_parents.keys())
2005
def modified_reference(self, file_id, file_parents):
2006
"""Record the modification of a reference.
2008
:param file_id: The file_id of the link to record.
2009
:param file_parents: The per-file parent revision ids.
2011
self._add_text_to_weave(file_id, [], file_parents.keys())
2013
def modified_file_text(self, file_id, file_parents,
2014
get_content_byte_lines, text_sha1=None,
2016
"""Record the text of file file_id
2018
:param file_id: The file_id of the file to record the text of.
2019
:param file_parents: The per-file parent revision ids.
2020
:param get_content_byte_lines: A callable which will return the byte
2022
:param text_sha1: Optional SHA1 of the file contents.
2023
:param text_size: Optional size of the file contents.
2025
# mutter('storing text of file {%s} in revision {%s} into %r',
2026
# file_id, self._new_revision_id, self.repository.weave_store)
2027
# special case to avoid diffing on renames or
2029
if (len(file_parents) == 1
2030
and text_sha1 == file_parents.values()[0].text_sha1
2031
and text_size == file_parents.values()[0].text_size):
2032
previous_ie = file_parents.values()[0]
2033
versionedfile = self.repository.weave_store.get_weave(file_id,
2034
self.repository.get_transaction())
2035
versionedfile.clone_text(self._new_revision_id,
2036
previous_ie.revision, file_parents.keys())
2037
return text_sha1, text_size
2039
new_lines = get_content_byte_lines()
2040
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2041
# should return the SHA1 and size
2042
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2043
return osutils.sha_strings(new_lines), \
2044
sum(map(len, new_lines))
2046
def modified_link(self, file_id, file_parents, link_target):
2047
"""Record the presence of a symbolic link.
2049
:param file_id: The file_id of the link to record.
2050
:param file_parents: The per-file parent revision ids.
2051
:param link_target: Target location of this link.
2053
self._add_text_to_weave(file_id, [], file_parents.keys())
2055
def _add_text_to_weave(self, file_id, new_lines, parents):
2056
versionedfile = self.repository.weave_store.get_weave_or_empty(
2057
file_id, self.repository.get_transaction())
2058
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2059
versionedfile.clear_cache()
2062
class _CommitBuilder(CommitBuilder):
2063
"""Temporary class so old CommitBuilders are detected properly
2065
Note: CommitBuilder works whether or not root entry is recorded.
2068
record_root_entry = True
2071
class RootCommitBuilder(CommitBuilder):
2072
"""This commitbuilder actually records the root id"""
2074
record_root_entry = True
2076
def record_entry_contents(self, ie, parent_invs, path, tree):
2077
"""Record the content of ie from tree into the commit if needed.
2079
Side effect: sets ie.revision when unchanged
2081
:param ie: An inventory entry present in the commit.
2082
:param parent_invs: The inventories of the parent revisions of the
2084
:param path: The path the entry is at in the tree.
2085
:param tree: The tree which contains this entry and should be used to
2088
assert self.new_inventory.root is not None or ie.parent_id is None
2089
self.new_inventory.add(ie)
2091
# ie.revision is always None if the InventoryEntry is considered
2092
# for committing. ie.snapshot will record the correct revision
2093
# which may be the sole parent if it is untouched.
2094
if ie.revision is not None:
2097
previous_entries = ie.find_previous_heads(
2099
self.repository.weave_store,
2100
self.repository.get_transaction())
2101
# we are creating a new revision for ie in the history store
2103
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2115
def _unescaper(match, _map=_unescape_map):
2116
code = match.group(1)
2120
if not code.startswith('#'):
2122
return unichr(int(code[1:])).encode('utf8')
2128
def _unescape_xml(data):
2129
"""Unescape predefined XML entities in a string of data."""
2131
if _unescape_re is None:
2132
_unescape_re = re.compile('\&([^;]*);')
2133
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
