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, KnitPlainFactory
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)
232
def has_revision(self, revision_id):
233
"""True if this repository has a copy of the revision."""
234
return self._revision_store.has_revision_id(revision_id,
235
self.get_transaction())
238
def get_revision_reconcile(self, revision_id):
239
"""'reconcile' helper routine that allows access to a revision always.
241
This variant of get_revision does not cross check the weave graph
242
against the revision one as get_revision does: but it should only
243
be used by reconcile, or reconcile-alike commands that are correcting
244
or testing the revision graph.
246
if not revision_id or not isinstance(revision_id, basestring):
247
raise InvalidRevisionId(revision_id=revision_id, branch=self)
248
return self._revision_store.get_revision(revision_id,
249
self.get_transaction())
252
def get_revision_xml(self, revision_id):
253
rev = self.get_revision(revision_id)
255
# the current serializer..
256
self._revision_store._serializer.write_revision(rev, rev_tmp)
258
return rev_tmp.getvalue()
261
def get_revision(self, revision_id):
262
"""Return the Revision object for a named revision"""
263
r = self.get_revision_reconcile(revision_id)
264
# weave corruption can lead to absent revision markers that should be
266
# the following test is reasonably cheap (it needs a single weave read)
267
# and the weave is cached in read transactions. In write transactions
268
# it is not cached but typically we only read a small number of
269
# revisions. For knits when they are introduced we will probably want
270
# to ensure that caching write transactions are in use.
271
inv = self.get_inventory_weave()
272
self._check_revision_parents(r, inv)
275
def _check_revision_parents(self, revision, inventory):
276
"""Private to Repository and Fetch.
278
This checks the parentage of revision in an inventory weave for
279
consistency and is only applicable to inventory-weave-for-ancestry
280
using repository formats & fetchers.
282
weave_parents = inventory.get_parents(revision.revision_id)
283
weave_names = inventory.versions()
284
for parent_id in revision.parent_ids:
285
if parent_id in weave_names:
286
# this parent must not be a ghost.
287
if not parent_id in weave_parents:
289
raise errors.CorruptRepository(self)
292
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
293
signature = gpg_strategy.sign(plaintext)
294
self._revision_store.add_revision_signature_text(revision_id,
296
self.get_transaction())
298
def fileid_involved_between_revs(self, from_revid, to_revid):
299
"""Find file_id(s) which are involved in the changes between revisions.
301
This determines the set of revisions which are involved, and then
302
finds all file ids affected by those revisions.
304
w = self.get_inventory_weave()
305
from_set = set(w.get_ancestry(from_revid))
306
to_set = set(w.get_ancestry(to_revid))
307
changed = to_set.difference(from_set)
308
return self._fileid_involved_by_set(changed)
310
def fileid_involved(self, last_revid=None):
311
"""Find all file_ids modified in the ancestry of last_revid.
313
:param last_revid: If None, last_revision() will be used.
315
w = self.get_inventory_weave()
317
changed = set(w.versions())
319
changed = set(w.get_ancestry(last_revid))
320
return self._fileid_involved_by_set(changed)
322
def fileid_involved_by_set(self, changes):
323
"""Find all file_ids modified by the set of revisions passed in.
325
:param changes: A set() of revision ids
327
# TODO: jam 20060119 This line does *nothing*, remove it.
328
# or better yet, change _fileid_involved_by_set so
329
# that it takes the inventory weave, rather than
330
# pulling it out by itself.
331
return self._fileid_involved_by_set(changes)
333
def _fileid_involved_by_set(self, changes):
334
"""Find the set of file-ids affected by the set of revisions.
336
:param changes: A set() of revision ids.
337
:return: A set() of file ids.
339
This peaks at the Weave, interpreting each line, looking to
340
see if it mentions one of the revisions. And if so, includes
341
the file id mentioned.
342
This expects both the Weave format, and the serialization
343
to have a single line per file/directory, and to have
344
fileid="" and revision="" on that line.
346
assert isinstance(self._format, (RepositoryFormat5,
349
RepositoryFormatKnit1)), \
350
"fileid_involved only supported for branches which store inventory as unnested xml"
352
w = self.get_inventory_weave()
355
# this code needs to read every line in every inventory for the
356
# inventories [changes]. Seeing a line twice is ok. Seeing a line
357
# not pesent in one of those inventories is unnecessary and not
358
# harmful because we are filtering by the revision id marker in the
359
# inventory lines to only select file ids altered in one of those
360
# revisions. We dont need to see all lines in the inventory because
361
# only those added in an inventory in rev X can contain a revision=X
363
for line in w.iter_lines_added_or_present_in_versions(changes):
364
start = line.find('file_id="')+9
365
if start < 9: continue
366
end = line.find('"', start)
368
file_id = _unescape_xml(line[start:end])
370
# check if file_id is already present
371
if file_id in file_ids: continue
373
start = line.find('revision="')+10
374
if start < 10: continue
375
end = line.find('"', start)
377
revision_id = _unescape_xml(line[start:end])
378
if revision_id in changes:
379
file_ids.add(file_id)
383
def get_inventory_weave(self):
384
return self.control_weaves.get_weave('inventory',
385
self.get_transaction())
388
def get_inventory(self, revision_id):
389
"""Get Inventory object by hash."""
390
xml = self.get_inventory_xml(revision_id)
391
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
394
def get_inventory_xml(self, revision_id):
395
"""Get inventory XML as a file object."""
397
assert isinstance(revision_id, basestring), type(revision_id)
398
iw = self.get_inventory_weave()
399
return iw.get_text(revision_id)
401
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
404
def get_inventory_sha1(self, revision_id):
405
"""Return the sha1 hash of the inventory entry
407
return self.get_revision(revision_id).inventory_sha1
410
def get_revision_graph(self, revision_id=None):
411
"""Return a dictionary containing the revision graph.
413
:return: a dictionary of revision_id->revision_parents_list.
415
weave = self.get_inventory_weave()
416
all_revisions = self._eliminate_revisions_not_present(weave.versions())
417
entire_graph = dict([(node, weave.get_parents(node)) for
418
node in all_revisions])
419
if revision_id is None:
421
elif revision_id not in entire_graph:
422
raise errors.NoSuchRevision(self, revision_id)
424
# add what can be reached from revision_id
426
pending = set([revision_id])
427
while len(pending) > 0:
429
result[node] = entire_graph[node]
430
for revision_id in result[node]:
431
if revision_id not in result:
432
pending.add(revision_id)
436
def get_revision_graph_with_ghosts(self, revision_ids=None):
437
"""Return a graph of the revisions with ghosts marked as applicable.
439
:param revision_ids: an iterable of revisions to graph or None for all.
440
:return: a Graph object with the graph reachable from revision_ids.
444
pending = set(self.all_revision_ids())
447
pending = set(revision_ids)
448
required = set(revision_ids)
451
revision_id = pending.pop()
453
rev = self.get_revision(revision_id)
454
except errors.NoSuchRevision:
455
if revision_id in required:
458
result.add_ghost(revision_id)
460
for parent_id in rev.parent_ids:
461
# is this queued or done ?
462
if (parent_id not in pending and
463
parent_id not in done):
465
pending.add(parent_id)
466
result.add_node(revision_id, rev.parent_ids)
467
done.add(revision_id)
471
def get_revision_inventory(self, revision_id):
472
"""Return inventory of a past revision."""
473
# TODO: Unify this with get_inventory()
474
# bzr 0.0.6 and later imposes the constraint that the inventory_id
475
# must be the same as its revision, so this is trivial.
476
if revision_id is None:
477
# This does not make sense: if there is no revision,
478
# then it is the current tree inventory surely ?!
479
# and thus get_root_id() is something that looks at the last
480
# commit on the branch, and the get_root_id is an inventory check.
481
raise NotImplementedError
482
# return Inventory(self.get_root_id())
484
return self.get_inventory(revision_id)
488
"""Return True if this repository is flagged as a shared repository."""
489
raise NotImplementedError(self.is_shared)
493
"""Reconcile this repository."""
494
from bzrlib.reconcile import RepoReconciler
495
reconciler = RepoReconciler(self)
496
reconciler.reconcile()
500
def revision_tree(self, revision_id):
501
"""Return Tree for a revision on this branch.
503
`revision_id` may be None for the null revision, in which case
504
an `EmptyTree` is returned."""
505
# TODO: refactor this to use an existing revision object
506
# so we don't need to read it in twice.
507
if revision_id is None or revision_id == NULL_REVISION:
510
inv = self.get_revision_inventory(revision_id)
511
return RevisionTree(self, inv, revision_id)
514
def get_ancestry(self, revision_id):
515
"""Return a list of revision-ids integrated by a revision.
517
This is topologically sorted.
519
if revision_id is None:
521
if not self.has_revision(revision_id):
522
raise errors.NoSuchRevision(self, revision_id)
523
w = self.get_inventory_weave()
524
candidates = w.get_ancestry(revision_id)
525
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
528
def print_file(self, file, revision_id):
529
"""Print `file` to stdout.
531
FIXME RBC 20060125 as John Meinel points out this is a bad api
532
- it writes to stdout, it assumes that that is valid etc. Fix
533
by creating a new more flexible convenience function.
535
tree = self.revision_tree(revision_id)
536
# use inventory as it was in that revision
537
file_id = tree.inventory.path2id(file)
539
# TODO: jam 20060427 Write a test for this code path
540
# it had a bug in it, and was raising the wrong
542
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
543
tree.print_file(file_id)
545
def get_transaction(self):
546
return self.control_files.get_transaction()
548
def revision_parents(self, revid):
549
return self.get_inventory_weave().parent_names(revid)
552
def set_make_working_trees(self, new_value):
553
"""Set the policy flag for making working trees when creating branches.
555
This only applies to branches that use this repository.
557
The default is 'True'.
558
:param new_value: True to restore the default, False to disable making
561
raise NotImplementedError(self.set_make_working_trees)
563
def make_working_trees(self):
564
"""Returns the policy for making working trees on new branches."""
565
raise NotImplementedError(self.make_working_trees)
568
def sign_revision(self, revision_id, gpg_strategy):
569
plaintext = Testament.from_revision(self, revision_id).as_short_text()
570
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
573
def has_signature_for_revision_id(self, revision_id):
574
"""Query for a revision signature for revision_id in the repository."""
575
return self._revision_store.has_signature(revision_id,
576
self.get_transaction())
579
def get_signature_text(self, revision_id):
580
"""Return the text for a signature."""
581
return self._revision_store.get_signature_text(revision_id,
582
self.get_transaction())
585
class AllInOneRepository(Repository):
586
"""Legacy support - the repository behaviour for all-in-one branches."""
588
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
589
# we reuse one control files instance.
590
dir_mode = a_bzrdir._control_files._dir_mode
591
file_mode = a_bzrdir._control_files._file_mode
593
def get_store(name, compressed=True, prefixed=False):
594
# FIXME: This approach of assuming stores are all entirely compressed
595
# or entirely uncompressed is tidy, but breaks upgrade from
596
# some existing branches where there's a mixture; we probably
597
# still want the option to look for both.
598
relpath = a_bzrdir._control_files._escape(name)
599
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
600
prefixed=prefixed, compressed=compressed,
603
#if self._transport.should_cache():
604
# cache_path = os.path.join(self.cache_root, name)
605
# os.mkdir(cache_path)
606
# store = bzrlib.store.CachedStore(store, cache_path)
609
# not broken out yet because the controlweaves|inventory_store
610
# and text_store | weave_store bits are still different.
611
if isinstance(_format, RepositoryFormat4):
612
# cannot remove these - there is still no consistent api
613
# which allows access to this old info.
614
self.inventory_store = get_store('inventory-store')
615
text_store = get_store('text-store')
616
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
620
"""AllInOne repositories cannot be shared."""
624
def set_make_working_trees(self, new_value):
625
"""Set the policy flag for making working trees when creating branches.
627
This only applies to branches that use this repository.
629
The default is 'True'.
630
:param new_value: True to restore the default, False to disable making
633
raise NotImplementedError(self.set_make_working_trees)
635
def make_working_trees(self):
636
"""Returns the policy for making working trees on new branches."""
640
class MetaDirRepository(Repository):
641
"""Repositories in the new meta-dir layout."""
643
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
644
super(MetaDirRepository, self).__init__(_format,
651
dir_mode = self.control_files._dir_mode
652
file_mode = self.control_files._file_mode
656
"""Return True if this repository is flagged as a shared repository."""
657
return self.control_files._transport.has('shared-storage')
660
def set_make_working_trees(self, new_value):
661
"""Set the policy flag for making working trees when creating branches.
663
This only applies to branches that use this repository.
665
The default is 'True'.
666
:param new_value: True to restore the default, False to disable making
671
self.control_files._transport.delete('no-working-trees')
672
except errors.NoSuchFile:
675
self.control_files.put_utf8('no-working-trees', '')
677
def make_working_trees(self):
678
"""Returns the policy for making working trees on new branches."""
679
return not self.control_files._transport.has('no-working-trees')
682
class KnitRepository(MetaDirRepository):
683
"""Knit format repository."""
686
def all_revision_ids(self):
687
"""See Repository.all_revision_ids()."""
688
return self._revision_store.all_revision_ids(self.get_transaction())
690
def fileid_involved_between_revs(self, from_revid, to_revid):
691
"""Find file_id(s) which are involved in the changes between revisions.
693
This determines the set of revisions which are involved, and then
694
finds all file ids affected by those revisions.
696
vf = self._get_revision_vf()
697
from_set = set(vf.get_ancestry(from_revid))
698
to_set = set(vf.get_ancestry(to_revid))
699
changed = to_set.difference(from_set)
700
return self._fileid_involved_by_set(changed)
702
def fileid_involved(self, last_revid=None):
703
"""Find all file_ids modified in the ancestry of last_revid.
705
:param last_revid: If None, last_revision() will be used.
708
changed = set(self.all_revision_ids())
710
changed = set(self.get_ancestry(last_revid))
713
return self._fileid_involved_by_set(changed)
716
def get_ancestry(self, revision_id):
717
"""Return a list of revision-ids integrated by a revision.
719
This is topologically sorted.
721
if revision_id is None:
723
vf = self._get_revision_vf()
725
return [None] + vf.get_ancestry(revision_id)
726
except errors.RevisionNotPresent:
727
raise errors.NoSuchRevision(self, revision_id)
730
def get_revision(self, revision_id):
731
"""Return the Revision object for a named revision"""
732
return self.get_revision_reconcile(revision_id)
735
def get_revision_graph(self, revision_id=None):
736
"""Return a dictionary containing the revision graph.
738
:return: a dictionary of revision_id->revision_parents_list.
740
weave = self._get_revision_vf()
741
entire_graph = weave.get_graph()
742
if revision_id is None:
743
return weave.get_graph()
744
elif revision_id not in weave:
745
raise errors.NoSuchRevision(self, revision_id)
747
# add what can be reached from revision_id
749
pending = set([revision_id])
750
while len(pending) > 0:
752
result[node] = weave.get_parents(node)
753
for revision_id in result[node]:
754
if revision_id not in result:
755
pending.add(revision_id)
759
def get_revision_graph_with_ghosts(self, revision_ids=None):
760
"""Return a graph of the revisions with ghosts marked as applicable.
762
:param revision_ids: an iterable of revisions to graph or None for all.
763
:return: a Graph object with the graph reachable from revision_ids.
766
vf = self._get_revision_vf()
767
versions = set(vf.versions())
769
pending = set(self.all_revision_ids())
772
pending = set(revision_ids)
773
required = set(revision_ids)
776
revision_id = pending.pop()
777
if not revision_id in versions:
778
if revision_id in required:
779
raise errors.NoSuchRevision(self, revision_id)
781
result.add_ghost(revision_id)
782
# mark it as done so we dont try for it again.
783
done.add(revision_id)
785
parent_ids = vf.get_parents_with_ghosts(revision_id)
786
for parent_id in parent_ids:
787
# is this queued or done ?
788
if (parent_id not in pending and
789
parent_id not in done):
791
pending.add(parent_id)
792
result.add_node(revision_id, parent_ids)
793
done.add(revision_id)
796
def _get_revision_vf(self):
797
""":return: a versioned file containing the revisions."""
798
vf = self._revision_store.get_revision_file(self.get_transaction())
803
"""Reconcile this repository."""
804
from bzrlib.reconcile import KnitReconciler
805
reconciler = KnitReconciler(self)
806
reconciler.reconcile()
809
def revision_parents(self, revid):
810
return self._get_revision_vf().get_parents(rev_id)
812
class RepositoryFormat(object):
813
"""A repository format.
815
Formats provide three things:
816
* An initialization routine to construct repository data on disk.
817
* a format string which is used when the BzrDir supports versioned
819
* an open routine which returns a Repository instance.
821
Formats are placed in an dict by their format string for reference
822
during opening. These should be subclasses of RepositoryFormat
825
Once a format is deprecated, just deprecate the initialize and open
826
methods on the format class. Do not deprecate the object, as the
827
object will be created every system load.
829
Common instance attributes:
830
_matchingbzrdir - the bzrdir format that the repository format was
831
originally written to work with. This can be used if manually
832
constructing a bzrdir and repository, or more commonly for test suite
836
_default_format = None
837
"""The default format used for new repositories."""
840
"""The known formats."""
843
def find_format(klass, a_bzrdir):
844
"""Return the format for the repository object in a_bzrdir."""
846
transport = a_bzrdir.get_repository_transport(None)
847
format_string = transport.get("format").read()
848
return klass._formats[format_string]
849
except errors.NoSuchFile:
850
raise errors.NoRepositoryPresent(a_bzrdir)
852
raise errors.UnknownFormatError(format_string)
854
def _get_control_store(self, repo_transport, control_files):
855
"""Return the control store for this repository."""
856
raise NotImplementedError(self._get_control_store)
859
def get_default_format(klass):
860
"""Return the current default format."""
861
return klass._default_format
863
def get_format_string(self):
864
"""Return the ASCII format string that identifies this format.
866
Note that in pre format ?? repositories the format string is
867
not permitted nor written to disk.
869
raise NotImplementedError(self.get_format_string)
871
def get_format_description(self):
872
"""Return the short desciption for this format."""
873
raise NotImplementedError(self.get_format_description)
875
def _get_revision_store(self, repo_transport, control_files):
876
"""Return the revision store object for this a_bzrdir."""
877
raise NotImplementedError(self._get_revision_store)
879
def _get_text_rev_store(self,
886
"""Common logic for getting a revision store for a repository.
888
see self._get_revision_store for the subclass-overridable method to
889
get the store for a repository.
891
from bzrlib.store.revision.text import TextRevisionStore
892
dir_mode = control_files._dir_mode
893
file_mode = control_files._file_mode
894
text_store =TextStore(transport.clone(name),
896
compressed=compressed,
899
_revision_store = TextRevisionStore(text_store, serializer)
900
return _revision_store
902
def _get_versioned_file_store(self,
907
versionedfile_class=WeaveFile,
909
weave_transport = control_files._transport.clone(name)
910
dir_mode = control_files._dir_mode
911
file_mode = control_files._file_mode
912
return VersionedFileStore(weave_transport, prefixed=prefixed,
915
versionedfile_class=versionedfile_class,
918
def initialize(self, a_bzrdir, shared=False):
919
"""Initialize a repository of this format in a_bzrdir.
921
:param a_bzrdir: The bzrdir to put the new repository in it.
922
:param shared: The repository should be initialized as a sharable one.
924
This may raise UninitializableFormat if shared repository are not
925
compatible the a_bzrdir.
928
def is_supported(self):
929
"""Is this format supported?
931
Supported formats must be initializable and openable.
932
Unsupported formats may not support initialization or committing or
933
some other features depending on the reason for not being supported.
937
def open(self, a_bzrdir, _found=False):
938
"""Return an instance of this format for the bzrdir a_bzrdir.
940
_found is a private parameter, do not use it.
942
raise NotImplementedError(self.open)
945
def register_format(klass, format):
946
klass._formats[format.get_format_string()] = format
949
def set_default_format(klass, format):
950
klass._default_format = format
953
def unregister_format(klass, format):
954
assert klass._formats[format.get_format_string()] is format
955
del klass._formats[format.get_format_string()]
958
class PreSplitOutRepositoryFormat(RepositoryFormat):
959
"""Base class for the pre split out repository formats."""
961
def initialize(self, a_bzrdir, shared=False, _internal=False):
962
"""Create a weave repository.
964
TODO: when creating split out bzr branch formats, move this to a common
965
base for Format5, Format6. or something like that.
967
from bzrlib.weavefile import write_weave_v5
968
from bzrlib.weave import Weave
971
raise errors.IncompatibleFormat(self, a_bzrdir._format)
974
# always initialized when the bzrdir is.
975
return self.open(a_bzrdir, _found=True)
977
# Create an empty weave
979
bzrlib.weavefile.write_weave_v5(Weave(), sio)
980
empty_weave = sio.getvalue()
982
mutter('creating repository in %s.', a_bzrdir.transport.base)
983
dirs = ['revision-store', 'weaves']
984
files = [('inventory.weave', StringIO(empty_weave)),
987
# FIXME: RBC 20060125 dont peek under the covers
988
# NB: no need to escape relative paths that are url safe.
989
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
991
control_files.create_lock()
992
control_files.lock_write()
993
control_files._transport.mkdir_multi(dirs,
994
mode=control_files._dir_mode)
996
for file, content in files:
997
control_files.put(file, content)
999
control_files.unlock()
1000
return self.open(a_bzrdir, _found=True)
1002
def _get_control_store(self, repo_transport, control_files):
1003
"""Return the control store for this repository."""
1004
return self._get_versioned_file_store('',
1009
def _get_text_store(self, transport, control_files):
1010
"""Get a store for file texts for this format."""
1011
raise NotImplementedError(self._get_text_store)
1013
def open(self, a_bzrdir, _found=False):
1014
"""See RepositoryFormat.open()."""
1016
# we are being called directly and must probe.
1017
raise NotImplementedError
1019
repo_transport = a_bzrdir.get_repository_transport(None)
1020
control_files = a_bzrdir._control_files
1021
text_store = self._get_text_store(repo_transport, control_files)
1022
control_store = self._get_control_store(repo_transport, control_files)
1023
_revision_store = self._get_revision_store(repo_transport, control_files)
1024
return AllInOneRepository(_format=self,
1026
_revision_store=_revision_store,
1027
control_store=control_store,
1028
text_store=text_store)
1031
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1032
"""Bzr repository format 4.
1034
This repository format has:
1036
- TextStores for texts, inventories,revisions.
1038
This format is deprecated: it indexes texts using a text id which is
1039
removed in format 5; initializationa and write support for this format
1044
super(RepositoryFormat4, self).__init__()
1045
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1047
def get_format_description(self):
1048
"""See RepositoryFormat.get_format_description()."""
1049
return "Repository format 4"
1051
def initialize(self, url, shared=False, _internal=False):
1052
"""Format 4 branches cannot be created."""
1053
raise errors.UninitializableFormat(self)
1055
def is_supported(self):
1056
"""Format 4 is not supported.
1058
It is not supported because the model changed from 4 to 5 and the
1059
conversion logic is expensive - so doing it on the fly was not
1064
def _get_control_store(self, repo_transport, control_files):
1065
"""Format 4 repositories have no formal control store at this point.
1067
This will cause any control-file-needing apis to fail - this is desired.
1071
def _get_revision_store(self, repo_transport, control_files):
1072
"""See RepositoryFormat._get_revision_store()."""
1073
from bzrlib.xml4 import serializer_v4
1074
return self._get_text_rev_store(repo_transport,
1077
serializer=serializer_v4)
1079
def _get_text_store(self, transport, control_files):
1080
"""See RepositoryFormat._get_text_store()."""
1083
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1084
"""Bzr control format 5.
1086
This repository format has:
1087
- weaves for file texts and inventory
1089
- TextStores for revisions and signatures.
1093
super(RepositoryFormat5, self).__init__()
1094
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1096
def get_format_description(self):
1097
"""See RepositoryFormat.get_format_description()."""
1098
return "Weave repository format 5"
1100
def _get_revision_store(self, repo_transport, control_files):
1101
"""See RepositoryFormat._get_revision_store()."""
1102
"""Return the revision store object for this a_bzrdir."""
1103
return self._get_text_rev_store(repo_transport,
1108
def _get_text_store(self, transport, control_files):
1109
"""See RepositoryFormat._get_text_store()."""
1110
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1113
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1114
"""Bzr control format 6.
1116
This repository format has:
1117
- weaves for file texts and inventory
1118
- hash subdirectory based stores.
1119
- TextStores for revisions and signatures.
1123
super(RepositoryFormat6, self).__init__()
1124
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1126
def get_format_description(self):
1127
"""See RepositoryFormat.get_format_description()."""
1128
return "Weave repository format 6"
1130
def _get_revision_store(self, repo_transport, control_files):
1131
"""See RepositoryFormat._get_revision_store()."""
1132
return self._get_text_rev_store(repo_transport,
1138
def _get_text_store(self, transport, control_files):
1139
"""See RepositoryFormat._get_text_store()."""
1140
return self._get_versioned_file_store('weaves', transport, control_files)
1143
class MetaDirRepositoryFormat(RepositoryFormat):
1144
"""Common base class for the new repositories using the metadir layour."""
1147
super(MetaDirRepositoryFormat, self).__init__()
1148
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1150
def _create_control_files(self, a_bzrdir):
1151
"""Create the required files and the initial control_files object."""
1152
# FIXME: RBC 20060125 dont peek under the covers
1153
# NB: no need to escape relative paths that are url safe.
1154
repository_transport = a_bzrdir.get_repository_transport(self)
1155
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1156
control_files.create_lock()
1157
return control_files
1159
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1160
"""Upload the initial blank content."""
1161
control_files = self._create_control_files(a_bzrdir)
1162
control_files.lock_write()
1164
control_files._transport.mkdir_multi(dirs,
1165
mode=control_files._dir_mode)
1166
for file, content in files:
1167
control_files.put(file, content)
1168
for file, content in utf8_files:
1169
control_files.put_utf8(file, content)
1171
control_files.put_utf8('shared-storage', '')
1173
control_files.unlock()
1176
class RepositoryFormat7(MetaDirRepositoryFormat):
1177
"""Bzr repository 7.
1179
This repository format has:
1180
- weaves for file texts and inventory
1181
- hash subdirectory based stores.
1182
- TextStores for revisions and signatures.
1183
- a format marker of its own
1184
- an optional 'shared-storage' flag
1185
- an optional 'no-working-trees' flag
1188
def _get_control_store(self, repo_transport, control_files):
1189
"""Return the control store for this repository."""
1190
return self._get_versioned_file_store('',
1195
def get_format_string(self):
1196
"""See RepositoryFormat.get_format_string()."""
1197
return "Bazaar-NG Repository format 7"
1199
def get_format_description(self):
1200
"""See RepositoryFormat.get_format_description()."""
1201
return "Weave repository format 7"
1203
def _get_revision_store(self, repo_transport, control_files):
1204
"""See RepositoryFormat._get_revision_store()."""
1205
return self._get_text_rev_store(repo_transport,
1212
def _get_text_store(self, transport, control_files):
1213
"""See RepositoryFormat._get_text_store()."""
1214
return self._get_versioned_file_store('weaves',
1218
def initialize(self, a_bzrdir, shared=False):
1219
"""Create a weave repository.
1221
:param shared: If true the repository will be initialized as a shared
1224
from bzrlib.weavefile import write_weave_v5
1225
from bzrlib.weave import Weave
1227
# Create an empty weave
1229
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1230
empty_weave = sio.getvalue()
1232
mutter('creating repository in %s.', a_bzrdir.transport.base)
1233
dirs = ['revision-store', 'weaves']
1234
files = [('inventory.weave', StringIO(empty_weave)),
1236
utf8_files = [('format', self.get_format_string())]
1238
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1239
return self.open(a_bzrdir=a_bzrdir, _found=True)
1241
def open(self, a_bzrdir, _found=False, _override_transport=None):
1242
"""See RepositoryFormat.open().
1244
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1245
repository at a slightly different url
1246
than normal. I.e. during 'upgrade'.
1249
format = RepositoryFormat.find_format(a_bzrdir)
1250
assert format.__class__ == self.__class__
1251
if _override_transport is not None:
1252
repo_transport = _override_transport
1254
repo_transport = a_bzrdir.get_repository_transport(None)
1255
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1256
text_store = self._get_text_store(repo_transport, control_files)
1257
control_store = self._get_control_store(repo_transport, control_files)
1258
_revision_store = self._get_revision_store(repo_transport, control_files)
1259
return MetaDirRepository(_format=self,
1261
control_files=control_files,
1262
_revision_store=_revision_store,
1263
control_store=control_store,
1264
text_store=text_store)
1267
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1268
"""Bzr repository knit format 1.
1270
This repository format has:
1271
- knits for file texts and inventory
1272
- hash subdirectory based stores.
1273
- knits for revisions and signatures
1274
- TextStores for revisions and signatures.
1275
- a format marker of its own
1276
- an optional 'shared-storage' flag
1277
- an optional 'no-working-trees' flag
1280
This format was introduced in bzr 0.8.
1283
def _get_control_store(self, repo_transport, control_files):
1284
"""Return the control store for this repository."""
1285
return VersionedFileStore(
1288
file_mode=control_files._file_mode,
1289
versionedfile_class=KnitVersionedFile,
1290
versionedfile_kwargs={'factory':KnitPlainFactory()},
1293
def get_format_string(self):
1294
"""See RepositoryFormat.get_format_string()."""
1295
return "Bazaar-NG Knit Repository Format 1"
1297
def get_format_description(self):
1298
"""See RepositoryFormat.get_format_description()."""
1299
return "Knit repository format 1"
1301
def _get_revision_store(self, repo_transport, control_files):
1302
"""See RepositoryFormat._get_revision_store()."""
1303
from bzrlib.store.revision.knit import KnitRevisionStore
1304
versioned_file_store = VersionedFileStore(
1306
file_mode=control_files._file_mode,
1309
versionedfile_class=KnitVersionedFile,
1310
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1313
return KnitRevisionStore(versioned_file_store)
1315
def _get_text_store(self, transport, control_files):
1316
"""See RepositoryFormat._get_text_store()."""
1317
return self._get_versioned_file_store('knits',
1320
versionedfile_class=KnitVersionedFile,
1323
def initialize(self, a_bzrdir, shared=False):
1324
"""Create a knit format 1 repository.
1326
:param a_bzrdir: bzrdir to contain the new repository; must already
1328
:param shared: If true the repository will be initialized as a shared
1331
mutter('creating repository in %s.', a_bzrdir.transport.base)
1332
dirs = ['revision-store', 'knits']
1334
utf8_files = [('format', self.get_format_string())]
1336
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1337
repo_transport = a_bzrdir.get_repository_transport(None)
1338
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1339
control_store = self._get_control_store(repo_transport, control_files)
1340
transaction = bzrlib.transactions.WriteTransaction()
1341
# trigger a write of the inventory store.
1342
control_store.get_weave_or_empty('inventory', transaction)
1343
_revision_store = self._get_revision_store(repo_transport, control_files)
1344
_revision_store.has_revision_id('A', transaction)
1345
_revision_store.get_signature_file(transaction)
1346
return self.open(a_bzrdir=a_bzrdir, _found=True)
1348
def open(self, a_bzrdir, _found=False, _override_transport=None):
1349
"""See RepositoryFormat.open().
1351
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1352
repository at a slightly different url
1353
than normal. I.e. during 'upgrade'.
1356
format = RepositoryFormat.find_format(a_bzrdir)
1357
assert format.__class__ == self.__class__
1358
if _override_transport is not None:
1359
repo_transport = _override_transport
1361
repo_transport = a_bzrdir.get_repository_transport(None)
1362
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1363
text_store = self._get_text_store(repo_transport, control_files)
1364
control_store = self._get_control_store(repo_transport, control_files)
1365
_revision_store = self._get_revision_store(repo_transport, control_files)
1366
return KnitRepository(_format=self,
1368
control_files=control_files,
1369
_revision_store=_revision_store,
1370
control_store=control_store,
1371
text_store=text_store)
1374
# formats which have no format string are not discoverable
1375
# and not independently creatable, so are not registered.
1376
RepositoryFormat.register_format(RepositoryFormat7())
1377
_default_format = RepositoryFormatKnit1()
1378
RepositoryFormat.register_format(_default_format)
1379
RepositoryFormat.set_default_format(_default_format)
1380
_legacy_formats = [RepositoryFormat4(),
1381
RepositoryFormat5(),
1382
RepositoryFormat6()]
1385
class InterRepository(InterObject):
1386
"""This class represents operations taking place between two repositories.
1388
Its instances have methods like copy_content and fetch, and contain
1389
references to the source and target repositories these operations can be
1392
Often we will provide convenience methods on 'repository' which carry out
1393
operations with another repository - they will always forward to
1394
InterRepository.get(other).method_name(parameters).
1398
"""The available optimised InterRepository types."""
1401
def copy_content(self, revision_id=None, basis=None):
1402
"""Make a complete copy of the content in self into destination.
1404
This is a destructive operation! Do not use it on existing
1407
:param revision_id: Only copy the content needed to construct
1408
revision_id and its parents.
1409
:param basis: Copy the needed data preferentially from basis.
1412
self.target.set_make_working_trees(self.source.make_working_trees())
1413
except NotImplementedError:
1415
# grab the basis available data
1416
if basis is not None:
1417
self.target.fetch(basis, revision_id=revision_id)
1418
# but dont bother fetching if we have the needed data now.
1419
if (revision_id not in (None, NULL_REVISION) and
1420
self.target.has_revision(revision_id)):
1422
self.target.fetch(self.source, revision_id=revision_id)
1424
def _double_lock(self, lock_source, lock_target):
1425
"""Take out too locks, rolling back the first if the second throws."""
1430
# we want to ensure that we don't leave source locked by mistake.
1431
# and any error on target should not confuse source.
1432
self.source.unlock()
1436
def fetch(self, revision_id=None, pb=None):
1437
"""Fetch the content required to construct revision_id.
1439
The content is copied from source to target.
1441
:param revision_id: if None all content is copied, if NULL_REVISION no
1443
:param pb: optional progress bar to use for progress reports. If not
1444
provided a default one will be created.
1446
Returns the copied revision count and the failed revisions in a tuple:
1449
from bzrlib.fetch import GenericRepoFetcher
1450
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1451
self.source, self.source._format, self.target, self.target._format)
1452
f = GenericRepoFetcher(to_repository=self.target,
1453
from_repository=self.source,
1454
last_revision=revision_id,
1456
return f.count_copied, f.failed_revisions
1458
def lock_read(self):
1459
"""Take out a logical read lock.
1461
This will lock the source branch and the target branch. The source gets
1462
a read lock and the target a read lock.
1464
self._double_lock(self.source.lock_read, self.target.lock_read)
1466
def lock_write(self):
1467
"""Take out a logical write lock.
1469
This will lock the source branch and the target branch. The source gets
1470
a read lock and the target a write lock.
1472
self._double_lock(self.source.lock_read, self.target.lock_write)
1475
def missing_revision_ids(self, revision_id=None):
1476
"""Return the revision ids that source has that target does not.
1478
These are returned in topological order.
1480
:param revision_id: only return revision ids included by this
1483
# generic, possibly worst case, slow code path.
1484
target_ids = set(self.target.all_revision_ids())
1485
if revision_id is not None:
1486
source_ids = self.source.get_ancestry(revision_id)
1487
assert source_ids.pop(0) == None
1489
source_ids = self.source.all_revision_ids()
1490
result_set = set(source_ids).difference(target_ids)
1491
# this may look like a no-op: its not. It preserves the ordering
1492
# other_ids had while only returning the members from other_ids
1493
# that we've decided we need.
1494
return [rev_id for rev_id in source_ids if rev_id in result_set]
1497
"""Release the locks on source and target."""
1499
self.target.unlock()
1501
self.source.unlock()
1504
class InterWeaveRepo(InterRepository):
1505
"""Optimised code paths between Weave based repositories."""
1507
_matching_repo_format = RepositoryFormat7()
1508
"""Repository format for testing with."""
1511
def is_compatible(source, target):
1512
"""Be compatible with known Weave formats.
1514
We dont test for the stores being of specific types becase that
1515
could lead to confusing results, and there is no need to be
1519
return (isinstance(source._format, (RepositoryFormat5,
1521
RepositoryFormat7)) and
1522
isinstance(target._format, (RepositoryFormat5,
1524
RepositoryFormat7)))
1525
except AttributeError:
1529
def copy_content(self, revision_id=None, basis=None):
1530
"""See InterRepository.copy_content()."""
1531
# weave specific optimised path:
1532
if basis is not None:
1533
# copy the basis in, then fetch remaining data.
1534
basis.copy_content_into(self.target, revision_id)
1535
# the basis copy_content_into could misset this.
1537
self.target.set_make_working_trees(self.source.make_working_trees())
1538
except NotImplementedError:
1540
self.target.fetch(self.source, revision_id=revision_id)
1543
self.target.set_make_working_trees(self.source.make_working_trees())
1544
except NotImplementedError:
1546
# FIXME do not peek!
1547
if self.source.control_files._transport.listable():
1548
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1550
self.target.weave_store.copy_all_ids(
1551
self.source.weave_store,
1553
from_transaction=self.source.get_transaction(),
1554
to_transaction=self.target.get_transaction())
1555
pb.update('copying inventory', 0, 1)
1556
self.target.control_weaves.copy_multi(
1557
self.source.control_weaves, ['inventory'],
1558
from_transaction=self.source.get_transaction(),
1559
to_transaction=self.target.get_transaction())
1560
self.target._revision_store.text_store.copy_all_ids(
1561
self.source._revision_store.text_store,
1566
self.target.fetch(self.source, revision_id=revision_id)
1569
def fetch(self, revision_id=None, pb=None):
1570
"""See InterRepository.fetch()."""
1571
from bzrlib.fetch import GenericRepoFetcher
1572
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1573
self.source, self.source._format, self.target, self.target._format)
1574
f = GenericRepoFetcher(to_repository=self.target,
1575
from_repository=self.source,
1576
last_revision=revision_id,
1578
return f.count_copied, f.failed_revisions
1581
def missing_revision_ids(self, revision_id=None):
1582
"""See InterRepository.missing_revision_ids()."""
1583
# we want all revisions to satisfy revision_id in source.
1584
# but we dont want to stat every file here and there.
1585
# we want then, all revisions other needs to satisfy revision_id
1586
# checked, but not those that we have locally.
1587
# so the first thing is to get a subset of the revisions to
1588
# satisfy revision_id in source, and then eliminate those that
1589
# we do already have.
1590
# this is slow on high latency connection to self, but as as this
1591
# disk format scales terribly for push anyway due to rewriting
1592
# inventory.weave, this is considered acceptable.
1594
if revision_id is not None:
1595
source_ids = self.source.get_ancestry(revision_id)
1596
assert source_ids.pop(0) == None
1598
source_ids = self.source._all_possible_ids()
1599
source_ids_set = set(source_ids)
1600
# source_ids is the worst possible case we may need to pull.
1601
# now we want to filter source_ids against what we actually
1602
# have in target, but dont try to check for existence where we know
1603
# we do not have a revision as that would be pointless.
1604
target_ids = set(self.target._all_possible_ids())
1605
possibly_present_revisions = target_ids.intersection(source_ids_set)
1606
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1607
required_revisions = source_ids_set.difference(actually_present_revisions)
1608
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1609
if revision_id is not None:
1610
# we used get_ancestry to determine source_ids then we are assured all
1611
# revisions referenced are present as they are installed in topological order.
1612
# and the tip revision was validated by get_ancestry.
1613
return required_topo_revisions
1615
# if we just grabbed the possibly available ids, then
1616
# we only have an estimate of whats available and need to validate
1617
# that against the revision records.
1618
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1621
class InterKnitRepo(InterRepository):
1622
"""Optimised code paths between Knit based repositories."""
1624
_matching_repo_format = RepositoryFormatKnit1()
1625
"""Repository format for testing with."""
1628
def is_compatible(source, target):
1629
"""Be compatible with known Knit formats.
1631
We dont test for the stores being of specific types becase that
1632
could lead to confusing results, and there is no need to be
1636
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1637
isinstance(target._format, (RepositoryFormatKnit1)))
1638
except AttributeError:
1642
def fetch(self, revision_id=None, pb=None):
1643
"""See InterRepository.fetch()."""
1644
from bzrlib.fetch import KnitRepoFetcher
1645
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1646
self.source, self.source._format, self.target, self.target._format)
1647
f = KnitRepoFetcher(to_repository=self.target,
1648
from_repository=self.source,
1649
last_revision=revision_id,
1651
return f.count_copied, f.failed_revisions
1654
def missing_revision_ids(self, revision_id=None):
1655
"""See InterRepository.missing_revision_ids()."""
1656
if revision_id is not None:
1657
source_ids = self.source.get_ancestry(revision_id)
1658
assert source_ids.pop(0) == None
1660
source_ids = self.source._all_possible_ids()
1661
source_ids_set = set(source_ids)
1662
# source_ids is the worst possible case we may need to pull.
1663
# now we want to filter source_ids against what we actually
1664
# have in target, but dont try to check for existence where we know
1665
# we do not have a revision as that would be pointless.
1666
target_ids = set(self.target._all_possible_ids())
1667
possibly_present_revisions = target_ids.intersection(source_ids_set)
1668
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1669
required_revisions = source_ids_set.difference(actually_present_revisions)
1670
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1671
if revision_id is not None:
1672
# we used get_ancestry to determine source_ids then we are assured all
1673
# revisions referenced are present as they are installed in topological order.
1674
# and the tip revision was validated by get_ancestry.
1675
return required_topo_revisions
1677
# if we just grabbed the possibly available ids, then
1678
# we only have an estimate of whats available and need to validate
1679
# that against the revision records.
1680
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1682
InterRepository.register_optimiser(InterWeaveRepo)
1683
InterRepository.register_optimiser(InterKnitRepo)
1686
class RepositoryTestProviderAdapter(object):
1687
"""A tool to generate a suite testing multiple repository formats at once.
1689
This is done by copying the test once for each transport and injecting
1690
the transport_server, transport_readonly_server, and bzrdir_format and
1691
repository_format classes into each copy. Each copy is also given a new id()
1692
to make it easy to identify.
1695
def __init__(self, transport_server, transport_readonly_server, formats):
1696
self._transport_server = transport_server
1697
self._transport_readonly_server = transport_readonly_server
1698
self._formats = formats
1700
def adapt(self, test):
1701
result = TestSuite()
1702
for repository_format, bzrdir_format in self._formats:
1703
new_test = deepcopy(test)
1704
new_test.transport_server = self._transport_server
1705
new_test.transport_readonly_server = self._transport_readonly_server
1706
new_test.bzrdir_format = bzrdir_format
1707
new_test.repository_format = repository_format
1708
def make_new_test_id():
1709
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1710
return lambda: new_id
1711
new_test.id = make_new_test_id()
1712
result.addTest(new_test)
1716
class InterRepositoryTestProviderAdapter(object):
1717
"""A tool to generate a suite testing multiple inter repository formats.
1719
This is done by copying the test once for each interrepo provider and injecting
1720
the transport_server, transport_readonly_server, repository_format and
1721
repository_to_format classes into each copy.
1722
Each copy is also given a new id() to make it easy to identify.
1725
def __init__(self, transport_server, transport_readonly_server, formats):
1726
self._transport_server = transport_server
1727
self._transport_readonly_server = transport_readonly_server
1728
self._formats = formats
1730
def adapt(self, test):
1731
result = TestSuite()
1732
for interrepo_class, repository_format, repository_format_to in self._formats:
1733
new_test = deepcopy(test)
1734
new_test.transport_server = self._transport_server
1735
new_test.transport_readonly_server = self._transport_readonly_server
1736
new_test.interrepo_class = interrepo_class
1737
new_test.repository_format = repository_format
1738
new_test.repository_format_to = repository_format_to
1739
def make_new_test_id():
1740
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1741
return lambda: new_id
1742
new_test.id = make_new_test_id()
1743
result.addTest(new_test)
1747
def default_test_list():
1748
"""Generate the default list of interrepo permutations to test."""
1750
# test the default InterRepository between format 6 and the current
1752
# XXX: robertc 20060220 reinstate this when there are two supported
1753
# formats which do not have an optimal code path between them.
1754
result.append((InterRepository,
1755
RepositoryFormat6(),
1756
RepositoryFormatKnit1()))
1757
for optimiser in InterRepository._optimisers:
1758
result.append((optimiser,
1759
optimiser._matching_repo_format,
1760
optimiser._matching_repo_format
1762
# if there are specific combinations we want to use, we can add them
1767
class CopyConverter(object):
1768
"""A repository conversion tool which just performs a copy of the content.
1770
This is slow but quite reliable.
1773
def __init__(self, target_format):
1774
"""Create a CopyConverter.
1776
:param target_format: The format the resulting repository should be.
1778
self.target_format = target_format
1780
def convert(self, repo, pb):
1781
"""Perform the conversion of to_convert, giving feedback via pb.
1783
:param to_convert: The disk object to convert.
1784
:param pb: a progress bar to use for progress information.
1789
# this is only useful with metadir layouts - separated repo content.
1790
# trigger an assertion if not such
1791
repo._format.get_format_string()
1792
self.repo_dir = repo.bzrdir
1793
self.step('Moving repository to repository.backup')
1794
self.repo_dir.transport.move('repository', 'repository.backup')
1795
backup_transport = self.repo_dir.transport.clone('repository.backup')
1796
self.source_repo = repo._format.open(self.repo_dir,
1798
_override_transport=backup_transport)
1799
self.step('Creating new repository')
1800
converted = self.target_format.initialize(self.repo_dir,
1801
self.source_repo.is_shared())
1802
converted.lock_write()
1804
self.step('Copying content into repository.')
1805
self.source_repo.copy_content_into(converted)
1808
self.step('Deleting old repository content.')
1809
self.repo_dir.transport.delete_tree('repository.backup')
1810
self.pb.note('repository converted')
1812
def step(self, message):
1813
"""Update the pb by a step."""
1815
self.pb.update(message, self.count, self.total)
1818
# Copied from xml.sax.saxutils
1819
def _unescape_xml(data):
1820
"""Unescape &, <, and > in a string of data.
1822
data = data.replace("<", "<")
1823
data = data.replace(">", ">")
1824
# must do ampersand last
1825
return data.replace("&", "&")