1
# Copyright (C) 2005, 2006 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 copy import deepcopy
18
from cStringIO import StringIO
19
from unittest import TestSuite
21
import bzrlib.bzrdir as bzrdir
22
from bzrlib.decorators import needs_read_lock, needs_write_lock
23
import bzrlib.errors as errors
24
from bzrlib.errors import InvalidRevisionId
25
import bzrlib.gpg as gpg
26
from bzrlib.graph import Graph
27
from bzrlib.inter import InterObject
28
from bzrlib.knit import KnitVersionedFile
29
from bzrlib.lockable_files import LockableFiles, TransportLock
30
from bzrlib.lockdir import LockDir
31
from bzrlib.osutils import safe_unicode
32
from bzrlib.revision import NULL_REVISION
33
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
34
from bzrlib.store.text import TextStore
35
from bzrlib.symbol_versioning import *
36
from bzrlib.trace import mutter, note
37
from bzrlib.tree import RevisionTree
38
from bzrlib.tsort import topo_sort
39
from bzrlib.testament import Testament
40
from bzrlib.tree import EmptyTree
42
from bzrlib.weave import WeaveFile
46
class Repository(object):
47
"""Repository holding history for one or more branches.
49
The repository holds and retrieves historical information including
50
revisions and file history. It's normally accessed only by the Branch,
51
which views a particular line of development through that history.
53
The Repository builds on top of Stores and a Transport, which respectively
54
describe the disk data format and the way of accessing the (possibly
59
def add_inventory(self, revid, inv, parents):
60
"""Add the inventory inv to the repository as revid.
62
:param parents: The revision ids of the parents that revid
63
is known to have and are in the repository already.
65
returns the sha1 of the serialized inventory.
67
inv_text = bzrlib.xml5.serializer_v5.write_inventory_to_string(inv)
68
inv_sha1 = bzrlib.osutils.sha_string(inv_text)
69
inv_vf = self.control_weaves.get_weave('inventory',
70
self.get_transaction())
71
inv_vf.add_lines(revid, parents, bzrlib.osutils.split_lines(inv_text))
75
def add_revision(self, rev_id, rev, inv=None, config=None):
76
"""Add rev to the revision store as rev_id.
78
:param rev_id: the revision id to use.
79
:param rev: The revision object.
80
:param inv: The inventory for the revision. if None, it will be looked
81
up in the inventory storer
82
:param config: If None no digital signature will be created.
83
If supplied its signature_needed method will be used
84
to determine if a signature should be made.
86
if config is not None and config.signature_needed():
88
inv = self.get_inventory(rev_id)
89
plaintext = Testament(rev, inv).as_short_text()
90
self.store_revision_signature(
91
gpg.GPGStrategy(config), plaintext, rev_id)
92
if not rev_id in self.get_inventory_weave():
94
raise errors.WeaveRevisionNotPresent(rev_id,
95
self.get_inventory_weave())
97
# yes, this is not suitable for adding with ghosts.
98
self.add_inventory(rev_id, inv, rev.parent_ids)
99
self._revision_store.add_revision(rev, self.get_transaction())
102
def _all_possible_ids(self):
103
"""Return all the possible revisions that we could find."""
104
return self.get_inventory_weave().versions()
107
def all_revision_ids(self):
108
"""Returns a list of all the revision ids in the repository.
110
These are in as much topological order as the underlying store can
111
present: for weaves ghosts may lead to a lack of correctness until
112
the reweave updates the parents list.
114
if self._revision_store.text_store.listable():
115
return self._revision_store.all_revision_ids(self.get_transaction())
116
result = self._all_possible_ids()
117
return self._eliminate_revisions_not_present(result)
120
def _eliminate_revisions_not_present(self, revision_ids):
121
"""Check every revision id in revision_ids to see if we have it.
123
Returns a set of the present revisions.
126
for id in revision_ids:
127
if self.has_revision(id):
132
def create(a_bzrdir):
133
"""Construct the current default format repository in a_bzrdir."""
134
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
136
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
137
"""instantiate a Repository.
139
:param _format: The format of the repository on disk.
140
:param a_bzrdir: The BzrDir of the repository.
142
In the future we will have a single api for all stores for
143
getting file texts, inventories and revisions, then
144
this construct will accept instances of those things.
146
super(Repository, self).__init__()
147
self._format = _format
148
# the following are part of the public API for Repository:
149
self.bzrdir = a_bzrdir
150
self.control_files = control_files
151
self._revision_store = _revision_store
152
self.text_store = text_store
153
# backwards compatability
154
self.weave_store = text_store
155
# not right yet - should be more semantically clear ?
157
self.control_store = control_store
158
self.control_weaves = control_store
159
# TODO: make sure to construct the right store classes, etc, depending
160
# on whether escaping is required.
162
def lock_write(self):
163
self.control_files.lock_write()
166
self.control_files.lock_read()
169
return self.control_files.is_locked()
172
def missing_revision_ids(self, other, revision_id=None):
173
"""Return the revision ids that other has that this does not.
175
These are returned in topological order.
177
revision_id: only return revision ids included by revision_id.
179
return InterRepository.get(other, self).missing_revision_ids(revision_id)
183
"""Open the repository rooted at base.
185
For instance, if the repository is at URL/.bzr/repository,
186
Repository.open(URL) -> a Repository instance.
188
control = bzrlib.bzrdir.BzrDir.open(base)
189
return control.open_repository()
191
def copy_content_into(self, destination, revision_id=None, basis=None):
192
"""Make a complete copy of the content in self into destination.
194
This is a destructive operation! Do not use it on existing
197
return InterRepository.get(self, destination).copy_content(revision_id, basis)
199
def fetch(self, source, revision_id=None, pb=None):
200
"""Fetch the content required to construct revision_id from source.
202
If revision_id is None all content is copied.
204
return InterRepository.get(source, self).fetch(revision_id=revision_id,
208
self.control_files.unlock()
211
def clone(self, a_bzrdir, revision_id=None, basis=None):
212
"""Clone this repository into a_bzrdir using the current format.
214
Currently no check is made that the format of this repository and
215
the bzrdir format are compatible. FIXME RBC 20060201.
217
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
218
# use target default format.
219
result = a_bzrdir.create_repository()
220
# FIXME RBC 20060209 split out the repository type to avoid this check ?
221
elif isinstance(a_bzrdir._format,
222
(bzrlib.bzrdir.BzrDirFormat4,
223
bzrlib.bzrdir.BzrDirFormat5,
224
bzrlib.bzrdir.BzrDirFormat6)):
225
result = a_bzrdir.open_repository()
227
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
228
self.copy_content_into(result, revision_id, basis)
231
# XXX: Is this actually called?
233
def copy(self, destination):
234
raise NotImplementedError("Repository.copy() no longer required?")
235
destination.lock_write()
237
copy_all(self.weave_store, destination.weave_store)
238
note('copying inventories')
239
destination.control_weaves.copy_multi(self.control_weaves,
241
copy_all(self.revision_store, destination.revision_store)
246
def has_revision(self, revision_id):
247
"""True if this repository has a copy of the revision."""
248
return self._revision_store.has_revision_id(revision_id,
249
self.get_transaction())
252
def get_revision_reconcile(self, revision_id):
253
"""'reconcile' helper routine that allows access to a revision always.
255
This variant of get_revision does not cross check the weave graph
256
against the revision one as get_revision does: but it should only
257
be used by reconcile, or reconcile-alike commands that are correcting
258
or testing the revision graph.
260
if not revision_id or not isinstance(revision_id, basestring):
261
raise InvalidRevisionId(revision_id=revision_id, branch=self)
262
return self._revision_store.get_revision(revision_id,
263
self.get_transaction())
266
def get_revision_xml(self, revision_id):
267
rev = self.get_revision(revision_id)
269
# the current serializer..
270
self._revision_store._serializer.write_revision(rev, rev_tmp)
272
return rev_tmp.getvalue()
275
def get_revision(self, revision_id):
276
"""Return the Revision object for a named revision"""
277
r = self.get_revision_reconcile(revision_id)
278
# weave corruption can lead to absent revision markers that should be
280
# the following test is reasonably cheap (it needs a single weave read)
281
# and the weave is cached in read transactions. In write transactions
282
# it is not cached but typically we only read a small number of
283
# revisions. For knits when they are introduced we will probably want
284
# to ensure that caching write transactions are in use.
285
inv = self.get_inventory_weave()
286
self._check_revision_parents(r, inv)
289
def _check_revision_parents(self, revision, inventory):
290
"""Private to Repository and Fetch.
292
This checks the parentage of revision in an inventory weave for
293
consistency and is only applicable to inventory-weave-for-ancestry
294
using repository formats & fetchers.
296
weave_parents = inventory.get_parents(revision.revision_id)
297
weave_names = inventory.versions()
298
for parent_id in revision.parent_ids:
299
if parent_id in weave_names:
300
# this parent must not be a ghost.
301
if not parent_id in weave_parents:
303
raise errors.CorruptRepository(self)
306
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
307
signature = gpg_strategy.sign(plaintext)
308
self._revision_store.add_revision_signature_text(revision_id,
310
self.get_transaction())
312
def fileid_involved_between_revs(self, from_revid, to_revid):
313
"""Find file_id(s) which are involved in the changes between revisions.
315
This determines the set of revisions which are involved, and then
316
finds all file ids affected by those revisions.
318
w = self.get_inventory_weave()
319
from_set = set(w.get_ancestry(from_revid))
320
to_set = set(w.get_ancestry(to_revid))
321
changed = to_set.difference(from_set)
322
return self._fileid_involved_by_set(changed)
324
def fileid_involved(self, last_revid=None):
325
"""Find all file_ids modified in the ancestry of last_revid.
327
:param last_revid: If None, last_revision() will be used.
329
w = self.get_inventory_weave()
331
changed = set(w.versions())
333
changed = set(w.get_ancestry(last_revid))
334
return self._fileid_involved_by_set(changed)
336
def fileid_involved_by_set(self, changes):
337
"""Find all file_ids modified by the set of revisions passed in.
339
:param changes: A set() of revision ids
341
# TODO: jam 20060119 This line does *nothing*, remove it.
342
# or better yet, change _fileid_involved_by_set so
343
# that it takes the inventory weave, rather than
344
# pulling it out by itself.
345
return self._fileid_involved_by_set(changes)
347
def _fileid_involved_by_set(self, changes):
348
"""Find the set of file-ids affected by the set of revisions.
350
:param changes: A set() of revision ids.
351
:return: A set() of file ids.
353
This peaks at the Weave, interpreting each line, looking to
354
see if it mentions one of the revisions. And if so, includes
355
the file id mentioned.
356
This expects both the Weave format, and the serialization
357
to have a single line per file/directory, and to have
358
fileid="" and revision="" on that line.
360
assert isinstance(self._format, (RepositoryFormat5,
363
RepositoryFormatKnit1)), \
364
"fileid_involved only supported for branches which store inventory as unnested xml"
366
w = self.get_inventory_weave()
369
# this code needs to read every line in every inventory for the
370
# inventories [changes]. Seeing a line twice is ok. Seeing a line
371
# not pesent in one of those inventories is unnecessary and not
372
# harmful because we are filtering by the revision id marker in the
373
# inventory lines to only select file ids altered in one of those
374
# revisions. We dont need to see all lines in the inventory because
375
# only those added in an inventory in rev X can contain a revision=X
377
for line in w.iter_lines_added_or_present_in_versions(changes):
378
start = line.find('file_id="')+9
379
if start < 9: continue
380
end = line.find('"', start)
382
file_id = _unescape_xml(line[start:end])
384
# check if file_id is already present
385
if file_id in file_ids: continue
387
start = line.find('revision="')+10
388
if start < 10: continue
389
end = line.find('"', start)
391
revision_id = _unescape_xml(line[start:end])
392
if revision_id in changes:
393
file_ids.add(file_id)
397
def get_inventory_weave(self):
398
return self.control_weaves.get_weave('inventory',
399
self.get_transaction())
402
def get_inventory(self, revision_id):
403
"""Get Inventory object by hash."""
404
xml = self.get_inventory_xml(revision_id)
405
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
408
def get_inventory_xml(self, revision_id):
409
"""Get inventory XML as a file object."""
411
assert isinstance(revision_id, basestring), type(revision_id)
412
iw = self.get_inventory_weave()
413
return iw.get_text(revision_id)
415
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
418
def get_inventory_sha1(self, revision_id):
419
"""Return the sha1 hash of the inventory entry
421
return self.get_revision(revision_id).inventory_sha1
424
def get_revision_graph(self, revision_id=None):
425
"""Return a dictionary containing the revision graph.
427
:return: a dictionary of revision_id->revision_parents_list.
429
weave = self.get_inventory_weave()
430
all_revisions = self._eliminate_revisions_not_present(weave.versions())
431
entire_graph = dict([(node, weave.get_parents(node)) for
432
node in all_revisions])
433
if revision_id is None:
435
elif revision_id not in entire_graph:
436
raise errors.NoSuchRevision(self, revision_id)
438
# add what can be reached from revision_id
440
pending = set([revision_id])
441
while len(pending) > 0:
443
result[node] = entire_graph[node]
444
for revision_id in result[node]:
445
if revision_id not in result:
446
pending.add(revision_id)
450
def get_revision_graph_with_ghosts(self, revision_ids=None):
451
"""Return a graph of the revisions with ghosts marked as applicable.
453
:param revision_ids: an iterable of revisions to graph or None for all.
454
:return: a Graph object with the graph reachable from revision_ids.
458
pending = set(self.all_revision_ids())
461
pending = set(revision_ids)
462
required = set(revision_ids)
465
revision_id = pending.pop()
467
rev = self.get_revision(revision_id)
468
except errors.NoSuchRevision:
469
if revision_id in required:
472
result.add_ghost(revision_id)
474
for parent_id in rev.parent_ids:
475
# is this queued or done ?
476
if (parent_id not in pending and
477
parent_id not in done):
479
pending.add(parent_id)
480
result.add_node(revision_id, rev.parent_ids)
481
done.add(revision_id)
485
def get_revision_inventory(self, revision_id):
486
"""Return inventory of a past revision."""
487
# TODO: Unify this with get_inventory()
488
# bzr 0.0.6 and later imposes the constraint that the inventory_id
489
# must be the same as its revision, so this is trivial.
490
if revision_id is None:
491
# This does not make sense: if there is no revision,
492
# then it is the current tree inventory surely ?!
493
# and thus get_root_id() is something that looks at the last
494
# commit on the branch, and the get_root_id is an inventory check.
495
raise NotImplementedError
496
# return Inventory(self.get_root_id())
498
return self.get_inventory(revision_id)
502
"""Return True if this repository is flagged as a shared repository."""
503
raise NotImplementedError(self.is_shared)
507
"""Reconcile this repository."""
508
from bzrlib.reconcile import RepoReconciler
509
reconciler = RepoReconciler(self)
510
reconciler.reconcile()
514
def revision_tree(self, revision_id):
515
"""Return Tree for a revision on this branch.
517
`revision_id` may be None for the null revision, in which case
518
an `EmptyTree` is returned."""
519
# TODO: refactor this to use an existing revision object
520
# so we don't need to read it in twice.
521
if revision_id is None or revision_id == NULL_REVISION:
524
inv = self.get_revision_inventory(revision_id)
525
return RevisionTree(self, inv, revision_id)
528
def get_ancestry(self, revision_id):
529
"""Return a list of revision-ids integrated by a revision.
531
This is topologically sorted.
533
if revision_id is None:
535
if not self.has_revision(revision_id):
536
raise errors.NoSuchRevision(self, revision_id)
537
w = self.get_inventory_weave()
538
candidates = w.get_ancestry(revision_id)
539
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
542
def print_file(self, file, revision_id):
543
"""Print `file` to stdout.
545
FIXME RBC 20060125 as John Meinel points out this is a bad api
546
- it writes to stdout, it assumes that that is valid etc. Fix
547
by creating a new more flexible convenience function.
549
tree = self.revision_tree(revision_id)
550
# use inventory as it was in that revision
551
file_id = tree.inventory.path2id(file)
553
raise BzrError("%r is not present in revision %s" % (file, revno))
555
revno = self.revision_id_to_revno(revision_id)
556
except errors.NoSuchRevision:
557
# TODO: This should not be BzrError,
558
# but NoSuchFile doesn't fit either
559
raise BzrError('%r is not present in revision %s'
560
% (file, revision_id))
562
raise BzrError('%r is not present in revision %s'
564
tree.print_file(file_id)
566
def get_transaction(self):
567
return self.control_files.get_transaction()
569
def revision_parents(self, revid):
570
return self.get_inventory_weave().parent_names(revid)
573
def set_make_working_trees(self, new_value):
574
"""Set the policy flag for making working trees when creating branches.
576
This only applies to branches that use this repository.
578
The default is 'True'.
579
:param new_value: True to restore the default, False to disable making
582
raise NotImplementedError(self.set_make_working_trees)
584
def make_working_trees(self):
585
"""Returns the policy for making working trees on new branches."""
586
raise NotImplementedError(self.make_working_trees)
589
def sign_revision(self, revision_id, gpg_strategy):
590
plaintext = Testament.from_revision(self, revision_id).as_short_text()
591
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
594
def has_signature_for_revision_id(self, revision_id):
595
"""Query for a revision signature for revision_id in the repository."""
596
return self._revision_store.has_signature(revision_id,
597
self.get_transaction())
600
def get_signature_text(self, revision_id):
601
"""Return the text for a signature."""
602
return self._revision_store.get_signature_text(revision_id,
603
self.get_transaction())
606
class AllInOneRepository(Repository):
607
"""Legacy support - the repository behaviour for all-in-one branches."""
609
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
610
# we reuse one control files instance.
611
dir_mode = a_bzrdir._control_files._dir_mode
612
file_mode = a_bzrdir._control_files._file_mode
614
def get_weave(name, prefixed=False):
616
name = safe_unicode(name)
619
relpath = a_bzrdir._control_files._escape(name)
620
weave_transport = a_bzrdir._control_files._transport.clone(relpath)
621
ws = WeaveStore(weave_transport, prefixed=prefixed,
624
if a_bzrdir._control_files._transport.should_cache():
625
ws.enable_cache = True
628
def get_store(name, compressed=True, prefixed=False):
629
# FIXME: This approach of assuming stores are all entirely compressed
630
# or entirely uncompressed is tidy, but breaks upgrade from
631
# some existing branches where there's a mixture; we probably
632
# still want the option to look for both.
633
relpath = a_bzrdir._control_files._escape(name)
634
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
635
prefixed=prefixed, compressed=compressed,
638
#if self._transport.should_cache():
639
# cache_path = os.path.join(self.cache_root, name)
640
# os.mkdir(cache_path)
641
# store = bzrlib.store.CachedStore(store, cache_path)
644
# not broken out yet because the controlweaves|inventory_store
645
# and text_store | weave_store bits are still different.
646
if isinstance(_format, RepositoryFormat4):
647
# cannot remove these - there is still no consistent api
648
# which allows access to this old info.
649
self.inventory_store = get_store('inventory-store')
650
text_store = get_store('text-store')
651
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
655
"""AllInOne repositories cannot be shared."""
659
def set_make_working_trees(self, new_value):
660
"""Set the policy flag for making working trees when creating branches.
662
This only applies to branches that use this repository.
664
The default is 'True'.
665
:param new_value: True to restore the default, False to disable making
668
raise NotImplementedError(self.set_make_working_trees)
670
def make_working_trees(self):
671
"""Returns the policy for making working trees on new branches."""
675
class MetaDirRepository(Repository):
676
"""Repositories in the new meta-dir layout."""
678
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
679
super(MetaDirRepository, self).__init__(_format,
686
dir_mode = self.control_files._dir_mode
687
file_mode = self.control_files._file_mode
689
def get_weave(name, prefixed=False):
691
name = safe_unicode(name)
694
relpath = self.control_files._escape(name)
695
weave_transport = self.control_files._transport.clone(relpath)
696
ws = WeaveStore(weave_transport, prefixed=prefixed,
699
if self.control_files._transport.should_cache():
700
ws.enable_cache = True
705
"""Return True if this repository is flagged as a shared repository."""
706
return self.control_files._transport.has('shared-storage')
709
def set_make_working_trees(self, new_value):
710
"""Set the policy flag for making working trees when creating branches.
712
This only applies to branches that use this repository.
714
The default is 'True'.
715
:param new_value: True to restore the default, False to disable making
720
self.control_files._transport.delete('no-working-trees')
721
except errors.NoSuchFile:
724
self.control_files.put_utf8('no-working-trees', '')
726
def make_working_trees(self):
727
"""Returns the policy for making working trees on new branches."""
728
return not self.control_files._transport.has('no-working-trees')
731
class KnitRepository(MetaDirRepository):
732
"""Knit format repository."""
735
def all_revision_ids(self):
736
"""See Repository.all_revision_ids()."""
737
return self._revision_store.all_revision_ids(self.get_transaction())
739
def fileid_involved_between_revs(self, from_revid, to_revid):
740
"""Find file_id(s) which are involved in the changes between revisions.
742
This determines the set of revisions which are involved, and then
743
finds all file ids affected by those revisions.
745
vf = self._get_revision_vf()
746
from_set = set(vf.get_ancestry(from_revid))
747
to_set = set(vf.get_ancestry(to_revid))
748
changed = to_set.difference(from_set)
749
return self._fileid_involved_by_set(changed)
751
def fileid_involved(self, last_revid=None):
752
"""Find all file_ids modified in the ancestry of last_revid.
754
:param last_revid: If None, last_revision() will be used.
757
changed = set(self.all_revision_ids())
759
changed = set(self.get_ancestry(last_revid))
762
return self._fileid_involved_by_set(changed)
765
def get_ancestry(self, revision_id):
766
"""Return a list of revision-ids integrated by a revision.
768
This is topologically sorted.
770
if revision_id is None:
772
vf = self._get_revision_vf()
774
return [None] + vf.get_ancestry(revision_id)
775
except errors.RevisionNotPresent:
776
raise errors.NoSuchRevision(self, revision_id)
779
def get_revision(self, revision_id):
780
"""Return the Revision object for a named revision"""
781
return self.get_revision_reconcile(revision_id)
784
def get_revision_graph(self, revision_id=None):
785
"""Return a dictionary containing the revision graph.
787
:return: a dictionary of revision_id->revision_parents_list.
789
weave = self._get_revision_vf()
790
entire_graph = weave.get_graph()
791
if revision_id is None:
792
return weave.get_graph()
793
elif revision_id not in weave:
794
raise errors.NoSuchRevision(self, revision_id)
796
# add what can be reached from revision_id
798
pending = set([revision_id])
799
while len(pending) > 0:
801
result[node] = weave.get_parents(node)
802
for revision_id in result[node]:
803
if revision_id not in result:
804
pending.add(revision_id)
808
def get_revision_graph_with_ghosts(self, revision_ids=None):
809
"""Return a graph of the revisions with ghosts marked as applicable.
811
:param revision_ids: an iterable of revisions to graph or None for all.
812
:return: a Graph object with the graph reachable from revision_ids.
815
vf = self._get_revision_vf()
816
versions = vf.versions()
818
pending = set(self.all_revision_ids())
821
pending = set(revision_ids)
822
required = set(revision_ids)
825
revision_id = pending.pop()
826
if not revision_id in versions:
827
if revision_id in required:
828
raise errors.NoSuchRevision(self, revision_id)
830
result.add_ghost(revision_id)
832
parent_ids = vf.get_parents_with_ghosts(revision_id)
833
for parent_id in parent_ids:
834
# is this queued or done ?
835
if (parent_id not in pending and
836
parent_id not in done):
838
pending.add(parent_id)
839
result.add_node(revision_id, parent_ids)
843
def _get_revision_vf(self):
844
""":return: a versioned file containing the revisions."""
845
vf = self._revision_store.get_revision_file(self.get_transaction())
850
"""Reconcile this repository."""
851
from bzrlib.reconcile import KnitReconciler
852
reconciler = KnitReconciler(self)
853
reconciler.reconcile()
856
def revision_parents(self, revid):
857
return self._get_revision_vf().get_parents(rev_id)
859
class RepositoryFormat(object):
860
"""A repository format.
862
Formats provide three things:
863
* An initialization routine to construct repository data on disk.
864
* a format string which is used when the BzrDir supports versioned
866
* an open routine which returns a Repository instance.
868
Formats are placed in an dict by their format string for reference
869
during opening. These should be subclasses of RepositoryFormat
872
Once a format is deprecated, just deprecate the initialize and open
873
methods on the format class. Do not deprecate the object, as the
874
object will be created every system load.
876
Common instance attributes:
877
_matchingbzrdir - the bzrdir format that the repository format was
878
originally written to work with. This can be used if manually
879
constructing a bzrdir and repository, or more commonly for test suite
883
_default_format = None
884
"""The default format used for new repositories."""
887
"""The known formats."""
890
def find_format(klass, a_bzrdir):
891
"""Return the format for the repository object in a_bzrdir."""
893
transport = a_bzrdir.get_repository_transport(None)
894
format_string = transport.get("format").read()
895
return klass._formats[format_string]
896
except errors.NoSuchFile:
897
raise errors.NoRepositoryPresent(a_bzrdir)
899
raise errors.UnknownFormatError(format_string)
901
def _get_control_store(self, repo_transport, control_files):
902
"""Return the control store for this repository."""
903
raise NotImplementedError(self._get_control_store)
906
def get_default_format(klass):
907
"""Return the current default format."""
908
return klass._default_format
910
def get_format_string(self):
911
"""Return the ASCII format string that identifies this format.
913
Note that in pre format ?? repositories the format string is
914
not permitted nor written to disk.
916
raise NotImplementedError(self.get_format_string)
918
def _get_revision_store(self, repo_transport, control_files):
919
"""Return the revision store object for this a_bzrdir."""
920
raise NotImplementedError(self._get_revision_store)
922
def _get_text_rev_store(self,
929
"""Common logic for getting a revision store for a repository.
931
see self._get_revision_store for the subclass-overridable method to
932
get the store for a repository.
934
from bzrlib.store.revision.text import TextRevisionStore
935
dir_mode = control_files._dir_mode
936
file_mode = control_files._file_mode
937
text_store =TextStore(transport.clone(name),
939
compressed=compressed,
942
_revision_store = TextRevisionStore(text_store, serializer)
943
return _revision_store
945
def _get_versioned_file_store(self,
950
versionedfile_class=WeaveFile):
951
weave_transport = control_files._transport.clone(name)
952
dir_mode = control_files._dir_mode
953
file_mode = control_files._file_mode
954
return VersionedFileStore(weave_transport, prefixed=prefixed,
957
versionedfile_class=versionedfile_class)
959
def initialize(self, a_bzrdir, shared=False):
960
"""Initialize a repository of this format in a_bzrdir.
962
:param a_bzrdir: The bzrdir to put the new repository in it.
963
:param shared: The repository should be initialized as a sharable one.
965
This may raise UninitializableFormat if shared repository are not
966
compatible the a_bzrdir.
969
def is_supported(self):
970
"""Is this format supported?
972
Supported formats must be initializable and openable.
973
Unsupported formats may not support initialization or committing or
974
some other features depending on the reason for not being supported.
978
def open(self, a_bzrdir, _found=False):
979
"""Return an instance of this format for the bzrdir a_bzrdir.
981
_found is a private parameter, do not use it.
983
raise NotImplementedError(self.open)
986
def register_format(klass, format):
987
klass._formats[format.get_format_string()] = format
990
def set_default_format(klass, format):
991
klass._default_format = format
994
def unregister_format(klass, format):
995
assert klass._formats[format.get_format_string()] is format
996
del klass._formats[format.get_format_string()]
999
class PreSplitOutRepositoryFormat(RepositoryFormat):
1000
"""Base class for the pre split out repository formats."""
1002
def initialize(self, a_bzrdir, shared=False, _internal=False):
1003
"""Create a weave repository.
1005
TODO: when creating split out bzr branch formats, move this to a common
1006
base for Format5, Format6. or something like that.
1008
from bzrlib.weavefile import write_weave_v5
1009
from bzrlib.weave import Weave
1012
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1015
# always initialized when the bzrdir is.
1016
return self.open(a_bzrdir, _found=True)
1018
# Create an empty weave
1020
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1021
empty_weave = sio.getvalue()
1023
mutter('creating repository in %s.', a_bzrdir.transport.base)
1024
dirs = ['revision-store', 'weaves']
1025
files = [('inventory.weave', StringIO(empty_weave)),
1028
# FIXME: RBC 20060125 dont peek under the covers
1029
# NB: no need to escape relative paths that are url safe.
1030
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1032
control_files.create_lock()
1033
control_files.lock_write()
1034
control_files._transport.mkdir_multi(dirs,
1035
mode=control_files._dir_mode)
1037
for file, content in files:
1038
control_files.put(file, content)
1040
control_files.unlock()
1041
return self.open(a_bzrdir, _found=True)
1043
def _get_control_store(self, repo_transport, control_files):
1044
"""Return the control store for this repository."""
1045
return self._get_versioned_file_store('',
1050
def _get_text_store(self, transport, control_files):
1051
"""Get a store for file texts for this format."""
1052
raise NotImplementedError(self._get_text_store)
1054
def open(self, a_bzrdir, _found=False):
1055
"""See RepositoryFormat.open()."""
1057
# we are being called directly and must probe.
1058
raise NotImplementedError
1060
repo_transport = a_bzrdir.get_repository_transport(None)
1061
control_files = a_bzrdir._control_files
1062
text_store = self._get_text_store(repo_transport, control_files)
1063
control_store = self._get_control_store(repo_transport, control_files)
1064
_revision_store = self._get_revision_store(repo_transport, control_files)
1065
return AllInOneRepository(_format=self,
1067
_revision_store=_revision_store,
1068
control_store=control_store,
1069
text_store=text_store)
1072
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1073
"""Bzr repository format 4.
1075
This repository format has:
1077
- TextStores for texts, inventories,revisions.
1079
This format is deprecated: it indexes texts using a text id which is
1080
removed in format 5; initializationa and write support for this format
1085
super(RepositoryFormat4, self).__init__()
1086
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1088
def initialize(self, url, shared=False, _internal=False):
1089
"""Format 4 branches cannot be created."""
1090
raise errors.UninitializableFormat(self)
1092
def is_supported(self):
1093
"""Format 4 is not supported.
1095
It is not supported because the model changed from 4 to 5 and the
1096
conversion logic is expensive - so doing it on the fly was not
1101
def _get_control_store(self, repo_transport, control_files):
1102
"""Format 4 repositories have no formal control store at this point.
1104
This will cause any control-file-needing apis to fail - this is desired.
1108
def _get_revision_store(self, repo_transport, control_files):
1109
"""See RepositoryFormat._get_revision_store()."""
1110
from bzrlib.xml4 import serializer_v4
1111
return self._get_text_rev_store(repo_transport,
1114
serializer=serializer_v4)
1116
def _get_text_store(self, transport, control_files):
1117
"""See RepositoryFormat._get_text_store()."""
1120
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1121
"""Bzr control format 5.
1123
This repository format has:
1124
- weaves for file texts and inventory
1126
- TextStores for revisions and signatures.
1130
super(RepositoryFormat5, self).__init__()
1131
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1133
def _get_revision_store(self, repo_transport, control_files):
1134
"""See RepositoryFormat._get_revision_store()."""
1135
"""Return the revision store object for this a_bzrdir."""
1136
return self._get_text_rev_store(repo_transport,
1141
def _get_text_store(self, transport, control_files):
1142
"""See RepositoryFormat._get_text_store()."""
1143
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1146
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1147
"""Bzr control format 6.
1149
This repository format has:
1150
- weaves for file texts and inventory
1151
- hash subdirectory based stores.
1152
- TextStores for revisions and signatures.
1156
super(RepositoryFormat6, self).__init__()
1157
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1159
def _get_revision_store(self, repo_transport, control_files):
1160
"""See RepositoryFormat._get_revision_store()."""
1161
return self._get_text_rev_store(repo_transport,
1167
def _get_text_store(self, transport, control_files):
1168
"""See RepositoryFormat._get_text_store()."""
1169
return self._get_versioned_file_store('weaves', transport, control_files)
1172
class MetaDirRepositoryFormat(RepositoryFormat):
1173
"""Common base class for the new repositories using the metadir layour."""
1176
super(MetaDirRepositoryFormat, self).__init__()
1177
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1179
def _create_control_files(self, a_bzrdir):
1180
"""Create the required files and the initial control_files object."""
1181
# FIXME: RBC 20060125 dont peek under the covers
1182
# NB: no need to escape relative paths that are url safe.
1183
repository_transport = a_bzrdir.get_repository_transport(self)
1184
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1185
control_files.create_lock()
1186
return control_files
1188
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1189
"""Upload the initial blank content."""
1190
control_files = self._create_control_files(a_bzrdir)
1191
control_files.lock_write()
1193
control_files._transport.mkdir_multi(dirs,
1194
mode=control_files._dir_mode)
1195
for file, content in files:
1196
control_files.put(file, content)
1197
for file, content in utf8_files:
1198
control_files.put_utf8(file, content)
1200
control_files.put_utf8('shared-storage', '')
1202
control_files.unlock()
1205
class RepositoryFormat7(MetaDirRepositoryFormat):
1206
"""Bzr repository 7.
1208
This repository format has:
1209
- weaves for file texts and inventory
1210
- hash subdirectory based stores.
1211
- TextStores for revisions and signatures.
1212
- a format marker of its own
1213
- an optional 'shared-storage' flag
1214
- an optional 'no-working-trees' flag
1217
def _get_control_store(self, repo_transport, control_files):
1218
"""Return the control store for this repository."""
1219
return self._get_versioned_file_store('',
1224
def get_format_string(self):
1225
"""See RepositoryFormat.get_format_string()."""
1226
return "Bazaar-NG Repository format 7"
1228
def _get_revision_store(self, repo_transport, control_files):
1229
"""See RepositoryFormat._get_revision_store()."""
1230
return self._get_text_rev_store(repo_transport,
1237
def _get_text_store(self, transport, control_files):
1238
"""See RepositoryFormat._get_text_store()."""
1239
return self._get_versioned_file_store('weaves',
1243
def initialize(self, a_bzrdir, shared=False):
1244
"""Create a weave repository.
1246
:param shared: If true the repository will be initialized as a shared
1249
from bzrlib.weavefile import write_weave_v5
1250
from bzrlib.weave import Weave
1252
# Create an empty weave
1254
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1255
empty_weave = sio.getvalue()
1257
mutter('creating repository in %s.', a_bzrdir.transport.base)
1258
dirs = ['revision-store', 'weaves']
1259
files = [('inventory.weave', StringIO(empty_weave)),
1261
utf8_files = [('format', self.get_format_string())]
1263
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1264
return self.open(a_bzrdir=a_bzrdir, _found=True)
1266
def open(self, a_bzrdir, _found=False, _override_transport=None):
1267
"""See RepositoryFormat.open().
1269
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1270
repository at a slightly different url
1271
than normal. I.e. during 'upgrade'.
1274
format = RepositoryFormat.find_format(a_bzrdir)
1275
assert format.__class__ == self.__class__
1276
if _override_transport is not None:
1277
repo_transport = _override_transport
1279
repo_transport = a_bzrdir.get_repository_transport(None)
1280
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1281
text_store = self._get_text_store(repo_transport, control_files)
1282
control_store = self._get_control_store(repo_transport, control_files)
1283
_revision_store = self._get_revision_store(repo_transport, control_files)
1284
return MetaDirRepository(_format=self,
1286
control_files=control_files,
1287
_revision_store=_revision_store,
1288
control_store=control_store,
1289
text_store=text_store)
1292
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1293
"""Bzr repository knit format 1.
1295
This repository format has:
1296
- knits for file texts and inventory
1297
- hash subdirectory based stores.
1298
- knits for revisions and signatures
1299
- TextStores for revisions and signatures.
1300
- a format marker of its own
1301
- an optional 'shared-storage' flag
1302
- an optional 'no-working-trees' flag
1306
def _get_control_store(self, repo_transport, control_files):
1307
"""Return the control store for this repository."""
1308
return self._get_versioned_file_store('',
1312
versionedfile_class=KnitVersionedFile)
1314
def get_format_string(self):
1315
"""See RepositoryFormat.get_format_string()."""
1316
return "Bazaar-NG Knit Repository Format 1"
1318
def _get_revision_store(self, repo_transport, control_files):
1319
"""See RepositoryFormat._get_revision_store()."""
1320
from bzrlib.store.revision.knit import KnitRevisionStore
1321
versioned_file_store = VersionedFileStore(
1323
file_mode = control_files._file_mode,
1326
versionedfile_class=KnitVersionedFile)
1327
return KnitRevisionStore(versioned_file_store)
1329
def _get_text_store(self, transport, control_files):
1330
"""See RepositoryFormat._get_text_store()."""
1331
return self._get_versioned_file_store('knits',
1334
versionedfile_class=KnitVersionedFile)
1336
def initialize(self, a_bzrdir, shared=False):
1337
"""Create a knit format 1 repository.
1339
:param shared: If true the repository will be initialized as a shared
1341
XXX NOTE that this current uses a Weave for testing and will become
1342
A Knit in due course.
1344
from bzrlib.weavefile import write_weave_v5
1345
from bzrlib.weave import Weave
1347
# Create an empty weave
1349
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1350
empty_weave = sio.getvalue()
1352
mutter('creating repository in %s.', a_bzrdir.transport.base)
1353
dirs = ['revision-store', 'knits', 'control']
1354
files = [('control/inventory.weave', StringIO(empty_weave)),
1356
utf8_files = [('format', self.get_format_string())]
1358
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1359
repo_transport = a_bzrdir.get_repository_transport(None)
1360
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1361
control_store = self._get_control_store(repo_transport, control_files)
1362
transaction = bzrlib.transactions.WriteTransaction()
1363
# trigger a write of the inventory store.
1364
control_store.get_weave_or_empty('inventory', transaction)
1365
_revision_store = self._get_revision_store(repo_transport, control_files)
1366
_revision_store.has_revision_id('A', transaction)
1367
_revision_store.get_signature_file(transaction)
1368
return self.open(a_bzrdir=a_bzrdir, _found=True)
1370
def open(self, a_bzrdir, _found=False, _override_transport=None):
1371
"""See RepositoryFormat.open().
1373
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1374
repository at a slightly different url
1375
than normal. I.e. during 'upgrade'.
1378
format = RepositoryFormat.find_format(a_bzrdir)
1379
assert format.__class__ == self.__class__
1380
if _override_transport is not None:
1381
repo_transport = _override_transport
1383
repo_transport = a_bzrdir.get_repository_transport(None)
1384
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1385
text_store = self._get_text_store(repo_transport, control_files)
1386
control_store = self._get_control_store(repo_transport, control_files)
1387
_revision_store = self._get_revision_store(repo_transport, control_files)
1388
return KnitRepository(_format=self,
1390
control_files=control_files,
1391
_revision_store=_revision_store,
1392
control_store=control_store,
1393
text_store=text_store)
1396
# formats which have no format string are not discoverable
1397
# and not independently creatable, so are not registered.
1398
_default_format = RepositoryFormat7()
1399
RepositoryFormat.register_format(_default_format)
1400
RepositoryFormat.register_format(RepositoryFormatKnit1())
1401
RepositoryFormat.set_default_format(_default_format)
1402
_legacy_formats = [RepositoryFormat4(),
1403
RepositoryFormat5(),
1404
RepositoryFormat6()]
1407
class InterRepository(InterObject):
1408
"""This class represents operations taking place between two repositories.
1410
Its instances have methods like copy_content and fetch, and contain
1411
references to the source and target repositories these operations can be
1414
Often we will provide convenience methods on 'repository' which carry out
1415
operations with another repository - they will always forward to
1416
InterRepository.get(other).method_name(parameters).
1420
"""The available optimised InterRepository types."""
1423
def copy_content(self, revision_id=None, basis=None):
1424
"""Make a complete copy of the content in self into destination.
1426
This is a destructive operation! Do not use it on existing
1429
:param revision_id: Only copy the content needed to construct
1430
revision_id and its parents.
1431
:param basis: Copy the needed data preferentially from basis.
1434
self.target.set_make_working_trees(self.source.make_working_trees())
1435
except NotImplementedError:
1437
# grab the basis available data
1438
if basis is not None:
1439
self.target.fetch(basis, revision_id=revision_id)
1440
# but dont bother fetching if we have the needed data now.
1441
if (revision_id not in (None, NULL_REVISION) and
1442
self.target.has_revision(revision_id)):
1444
self.target.fetch(self.source, revision_id=revision_id)
1446
def _double_lock(self, lock_source, lock_target):
1447
"""Take out too locks, rolling back the first if the second throws."""
1452
# we want to ensure that we don't leave source locked by mistake.
1453
# and any error on target should not confuse source.
1454
self.source.unlock()
1458
def fetch(self, revision_id=None, pb=None):
1459
"""Fetch the content required to construct revision_id.
1461
The content is copied from source to target.
1463
:param revision_id: if None all content is copied, if NULL_REVISION no
1465
:param pb: optional progress bar to use for progress reports. If not
1466
provided a default one will be created.
1468
Returns the copied revision count and the failed revisions in a tuple:
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, self.target._format)
1474
f = GenericRepoFetcher(to_repository=self.target,
1475
from_repository=self.source,
1476
last_revision=revision_id,
1478
return f.count_copied, f.failed_revisions
1480
def lock_read(self):
1481
"""Take out a logical read lock.
1483
This will lock the source branch and the target branch. The source gets
1484
a read lock and the target a read lock.
1486
self._double_lock(self.source.lock_read, self.target.lock_read)
1488
def lock_write(self):
1489
"""Take out a logical write lock.
1491
This will lock the source branch and the target branch. The source gets
1492
a read lock and the target a write lock.
1494
self._double_lock(self.source.lock_read, self.target.lock_write)
1497
def missing_revision_ids(self, revision_id=None):
1498
"""Return the revision ids that source has that target does not.
1500
These are returned in topological order.
1502
:param revision_id: only return revision ids included by this
1505
# generic, possibly worst case, slow code path.
1506
target_ids = set(self.target.all_revision_ids())
1507
if revision_id is not None:
1508
source_ids = self.source.get_ancestry(revision_id)
1509
assert source_ids.pop(0) == None
1511
source_ids = self.source.all_revision_ids()
1512
result_set = set(source_ids).difference(target_ids)
1513
# this may look like a no-op: its not. It preserves the ordering
1514
# other_ids had while only returning the members from other_ids
1515
# that we've decided we need.
1516
return [rev_id for rev_id in source_ids if rev_id in result_set]
1519
"""Release the locks on source and target."""
1521
self.target.unlock()
1523
self.source.unlock()
1526
class InterWeaveRepo(InterRepository):
1527
"""Optimised code paths between Weave based repositories."""
1529
_matching_repo_format = _default_format
1530
"""Repository format for testing with."""
1533
def is_compatible(source, target):
1534
"""Be compatible with known Weave formats.
1536
We dont test for the stores being of specific types becase that
1537
could lead to confusing results, and there is no need to be
1541
return (isinstance(source._format, (RepositoryFormat5,
1543
RepositoryFormat7)) and
1544
isinstance(target._format, (RepositoryFormat5,
1546
RepositoryFormat7)))
1547
except AttributeError:
1551
def copy_content(self, revision_id=None, basis=None):
1552
"""See InterRepository.copy_content()."""
1553
# weave specific optimised path:
1554
if basis is not None:
1555
# copy the basis in, then fetch remaining data.
1556
basis.copy_content_into(self.target, revision_id)
1557
# the basis copy_content_into could misset this.
1559
self.target.set_make_working_trees(self.source.make_working_trees())
1560
except NotImplementedError:
1562
self.target.fetch(self.source, revision_id=revision_id)
1565
self.target.set_make_working_trees(self.source.make_working_trees())
1566
except NotImplementedError:
1568
# FIXME do not peek!
1569
if self.source.control_files._transport.listable():
1570
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1572
self.target.weave_store.copy_all_ids(
1573
self.source.weave_store,
1575
from_transaction=self.source.get_transaction(),
1576
to_transaction=self.target.get_transaction())
1577
pb.update('copying inventory', 0, 1)
1578
self.target.control_weaves.copy_multi(
1579
self.source.control_weaves, ['inventory'],
1580
from_transaction=self.source.get_transaction(),
1581
to_transaction=self.target.get_transaction())
1582
self.target._revision_store.text_store.copy_all_ids(
1583
self.source._revision_store.text_store,
1588
self.target.fetch(self.source, revision_id=revision_id)
1591
def fetch(self, revision_id=None, pb=None):
1592
"""See InterRepository.fetch()."""
1593
from bzrlib.fetch import GenericRepoFetcher
1594
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1595
self.source, self.source._format, self.target, self.target._format)
1596
f = GenericRepoFetcher(to_repository=self.target,
1597
from_repository=self.source,
1598
last_revision=revision_id,
1600
return f.count_copied, f.failed_revisions
1603
def missing_revision_ids(self, revision_id=None):
1604
"""See InterRepository.missing_revision_ids()."""
1605
# we want all revisions to satisfy revision_id in source.
1606
# but we dont want to stat every file here and there.
1607
# we want then, all revisions other needs to satisfy revision_id
1608
# checked, but not those that we have locally.
1609
# so the first thing is to get a subset of the revisions to
1610
# satisfy revision_id in source, and then eliminate those that
1611
# we do already have.
1612
# this is slow on high latency connection to self, but as as this
1613
# disk format scales terribly for push anyway due to rewriting
1614
# inventory.weave, this is considered acceptable.
1616
if revision_id is not None:
1617
source_ids = self.source.get_ancestry(revision_id)
1618
assert source_ids.pop(0) == None
1620
source_ids = self.source._all_possible_ids()
1621
source_ids_set = set(source_ids)
1622
# source_ids is the worst possible case we may need to pull.
1623
# now we want to filter source_ids against what we actually
1624
# have in target, but dont try to check for existence where we know
1625
# we do not have a revision as that would be pointless.
1626
target_ids = set(self.target._all_possible_ids())
1627
possibly_present_revisions = target_ids.intersection(source_ids_set)
1628
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1629
required_revisions = source_ids_set.difference(actually_present_revisions)
1630
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1631
if revision_id is not None:
1632
# we used get_ancestry to determine source_ids then we are assured all
1633
# revisions referenced are present as they are installed in topological order.
1634
# and the tip revision was validated by get_ancestry.
1635
return required_topo_revisions
1637
# if we just grabbed the possibly available ids, then
1638
# we only have an estimate of whats available and need to validate
1639
# that against the revision records.
1640
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1643
class InterKnitRepo(InterRepository):
1644
"""Optimised code paths between Knit based repositories."""
1646
_matching_repo_format = RepositoryFormatKnit1()
1647
"""Repository format for testing with."""
1650
def is_compatible(source, target):
1651
"""Be compatible with known Knit formats.
1653
We dont test for the stores being of specific types becase that
1654
could lead to confusing results, and there is no need to be
1658
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1659
isinstance(target._format, (RepositoryFormatKnit1)))
1660
except AttributeError:
1664
def fetch(self, revision_id=None, pb=None):
1665
"""See InterRepository.fetch()."""
1666
from bzrlib.fetch import KnitRepoFetcher
1667
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1668
self.source, self.source._format, self.target, self.target._format)
1669
f = KnitRepoFetcher(to_repository=self.target,
1670
from_repository=self.source,
1671
last_revision=revision_id,
1673
return f.count_copied, f.failed_revisions
1676
def missing_revision_ids(self, revision_id=None):
1677
"""See InterRepository.missing_revision_ids()."""
1678
if revision_id is not None:
1679
source_ids = self.source.get_ancestry(revision_id)
1680
assert source_ids.pop(0) == None
1682
source_ids = self.source._all_possible_ids()
1683
source_ids_set = set(source_ids)
1684
# source_ids is the worst possible case we may need to pull.
1685
# now we want to filter source_ids against what we actually
1686
# have in target, but dont try to check for existence where we know
1687
# we do not have a revision as that would be pointless.
1688
target_ids = set(self.target._all_possible_ids())
1689
possibly_present_revisions = target_ids.intersection(source_ids_set)
1690
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1691
required_revisions = source_ids_set.difference(actually_present_revisions)
1692
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1693
if revision_id is not None:
1694
# we used get_ancestry to determine source_ids then we are assured all
1695
# revisions referenced are present as they are installed in topological order.
1696
# and the tip revision was validated by get_ancestry.
1697
return required_topo_revisions
1699
# if we just grabbed the possibly available ids, then
1700
# we only have an estimate of whats available and need to validate
1701
# that against the revision records.
1702
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1704
InterRepository.register_optimiser(InterWeaveRepo)
1705
InterRepository.register_optimiser(InterKnitRepo)
1708
class RepositoryTestProviderAdapter(object):
1709
"""A tool to generate a suite testing multiple repository formats at once.
1711
This is done by copying the test once for each transport and injecting
1712
the transport_server, transport_readonly_server, and bzrdir_format and
1713
repository_format classes into each copy. Each copy is also given a new id()
1714
to make it easy to identify.
1717
def __init__(self, transport_server, transport_readonly_server, formats):
1718
self._transport_server = transport_server
1719
self._transport_readonly_server = transport_readonly_server
1720
self._formats = formats
1722
def adapt(self, test):
1723
result = TestSuite()
1724
for repository_format, bzrdir_format in self._formats:
1725
new_test = deepcopy(test)
1726
new_test.transport_server = self._transport_server
1727
new_test.transport_readonly_server = self._transport_readonly_server
1728
new_test.bzrdir_format = bzrdir_format
1729
new_test.repository_format = repository_format
1730
def make_new_test_id():
1731
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1732
return lambda: new_id
1733
new_test.id = make_new_test_id()
1734
result.addTest(new_test)
1738
class InterRepositoryTestProviderAdapter(object):
1739
"""A tool to generate a suite testing multiple inter repository formats.
1741
This is done by copying the test once for each interrepo provider and injecting
1742
the transport_server, transport_readonly_server, repository_format and
1743
repository_to_format classes into each copy.
1744
Each copy is also given a new id() to make it easy to identify.
1747
def __init__(self, transport_server, transport_readonly_server, formats):
1748
self._transport_server = transport_server
1749
self._transport_readonly_server = transport_readonly_server
1750
self._formats = formats
1752
def adapt(self, test):
1753
result = TestSuite()
1754
for interrepo_class, repository_format, repository_format_to in self._formats:
1755
new_test = deepcopy(test)
1756
new_test.transport_server = self._transport_server
1757
new_test.transport_readonly_server = self._transport_readonly_server
1758
new_test.interrepo_class = interrepo_class
1759
new_test.repository_format = repository_format
1760
new_test.repository_format_to = repository_format_to
1761
def make_new_test_id():
1762
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1763
return lambda: new_id
1764
new_test.id = make_new_test_id()
1765
result.addTest(new_test)
1769
def default_test_list():
1770
"""Generate the default list of interrepo permutations to test."""
1772
# test the default InterRepository between format 6 and the current
1774
# XXX: robertc 20060220 reinstate this when there are two supported
1775
# formats which do not have an optimal code path between them.
1776
result.append((InterRepository,
1777
RepositoryFormat6(),
1778
RepositoryFormatKnit1()))
1779
for optimiser in InterRepository._optimisers:
1780
result.append((optimiser,
1781
optimiser._matching_repo_format,
1782
optimiser._matching_repo_format
1784
# if there are specific combinations we want to use, we can add them
1789
class CopyConverter(object):
1790
"""A repository conversion tool which just performs a copy of the content.
1792
This is slow but quite reliable.
1795
def __init__(self, target_format):
1796
"""Create a CopyConverter.
1798
:param target_format: The format the resulting repository should be.
1800
self.target_format = target_format
1802
def convert(self, repo, pb):
1803
"""Perform the conversion of to_convert, giving feedback via pb.
1805
:param to_convert: The disk object to convert.
1806
:param pb: a progress bar to use for progress information.
1811
# this is only useful with metadir layouts - separated repo content.
1812
# trigger an assertion if not such
1813
repo._format.get_format_string()
1814
self.repo_dir = repo.bzrdir
1815
self.step('Moving repository to repository.backup')
1816
self.repo_dir.transport.move('repository', 'repository.backup')
1817
backup_transport = self.repo_dir.transport.clone('repository.backup')
1818
self.source_repo = repo._format.open(self.repo_dir,
1820
_override_transport=backup_transport)
1821
self.step('Creating new repository')
1822
converted = self.target_format.initialize(self.repo_dir,
1823
self.source_repo.is_shared())
1824
converted.lock_write()
1826
self.step('Copying content into repository.')
1827
self.source_repo.copy_content_into(converted)
1830
self.step('Deleting old repository content.')
1831
self.repo_dir.transport.delete_tree('repository.backup')
1832
self.pb.note('repository converted')
1834
def step(self, message):
1835
"""Update the pb by a step."""
1837
self.pb.update(message, self.count, self.total)
1840
# Copied from xml.sax.saxutils
1841
def _unescape_xml(data):
1842
"""Unescape &, <, and > in a string of data.
1844
data = data.replace("<", "<")
1845
data = data.replace(">", ">")
1846
# must do ampersand last
1847
return data.replace("&", "&")
added
removed
bzrlib/repository.py
