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)
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_store(name, compressed=True, prefixed=False):
615
# FIXME: This approach of assuming stores are all entirely compressed
616
# or entirely uncompressed is tidy, but breaks upgrade from
617
# some existing branches where there's a mixture; we probably
618
# still want the option to look for both.
619
relpath = a_bzrdir._control_files._escape(name)
620
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
621
prefixed=prefixed, compressed=compressed,
624
#if self._transport.should_cache():
625
# cache_path = os.path.join(self.cache_root, name)
626
# os.mkdir(cache_path)
627
# store = bzrlib.store.CachedStore(store, cache_path)
630
# not broken out yet because the controlweaves|inventory_store
631
# and text_store | weave_store bits are still different.
632
if isinstance(_format, RepositoryFormat4):
633
# cannot remove these - there is still no consistent api
634
# which allows access to this old info.
635
self.inventory_store = get_store('inventory-store')
636
text_store = get_store('text-store')
637
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
641
"""AllInOne repositories cannot be shared."""
645
def set_make_working_trees(self, new_value):
646
"""Set the policy flag for making working trees when creating branches.
648
This only applies to branches that use this repository.
650
The default is 'True'.
651
:param new_value: True to restore the default, False to disable making
654
raise NotImplementedError(self.set_make_working_trees)
656
def make_working_trees(self):
657
"""Returns the policy for making working trees on new branches."""
661
class MetaDirRepository(Repository):
662
"""Repositories in the new meta-dir layout."""
664
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
665
super(MetaDirRepository, self).__init__(_format,
672
dir_mode = self.control_files._dir_mode
673
file_mode = self.control_files._file_mode
677
"""Return True if this repository is flagged as a shared repository."""
678
return self.control_files._transport.has('shared-storage')
681
def set_make_working_trees(self, new_value):
682
"""Set the policy flag for making working trees when creating branches.
684
This only applies to branches that use this repository.
686
The default is 'True'.
687
:param new_value: True to restore the default, False to disable making
692
self.control_files._transport.delete('no-working-trees')
693
except errors.NoSuchFile:
696
self.control_files.put_utf8('no-working-trees', '')
698
def make_working_trees(self):
699
"""Returns the policy for making working trees on new branches."""
700
return not self.control_files._transport.has('no-working-trees')
703
class KnitRepository(MetaDirRepository):
704
"""Knit format repository."""
707
def all_revision_ids(self):
708
"""See Repository.all_revision_ids()."""
709
return self._revision_store.all_revision_ids(self.get_transaction())
711
def fileid_involved_between_revs(self, from_revid, to_revid):
712
"""Find file_id(s) which are involved in the changes between revisions.
714
This determines the set of revisions which are involved, and then
715
finds all file ids affected by those revisions.
717
vf = self._get_revision_vf()
718
from_set = set(vf.get_ancestry(from_revid))
719
to_set = set(vf.get_ancestry(to_revid))
720
changed = to_set.difference(from_set)
721
return self._fileid_involved_by_set(changed)
723
def fileid_involved(self, last_revid=None):
724
"""Find all file_ids modified in the ancestry of last_revid.
726
:param last_revid: If None, last_revision() will be used.
729
changed = set(self.all_revision_ids())
731
changed = set(self.get_ancestry(last_revid))
734
return self._fileid_involved_by_set(changed)
737
def get_ancestry(self, revision_id):
738
"""Return a list of revision-ids integrated by a revision.
740
This is topologically sorted.
742
if revision_id is None:
744
vf = self._get_revision_vf()
746
return [None] + vf.get_ancestry(revision_id)
747
except errors.RevisionNotPresent:
748
raise errors.NoSuchRevision(self, revision_id)
751
def get_revision(self, revision_id):
752
"""Return the Revision object for a named revision"""
753
return self.get_revision_reconcile(revision_id)
756
def get_revision_graph(self, revision_id=None):
757
"""Return a dictionary containing the revision graph.
759
:return: a dictionary of revision_id->revision_parents_list.
761
weave = self._get_revision_vf()
762
entire_graph = weave.get_graph()
763
if revision_id is None:
764
return weave.get_graph()
765
elif revision_id not in weave:
766
raise errors.NoSuchRevision(self, revision_id)
768
# add what can be reached from revision_id
770
pending = set([revision_id])
771
while len(pending) > 0:
773
result[node] = weave.get_parents(node)
774
for revision_id in result[node]:
775
if revision_id not in result:
776
pending.add(revision_id)
780
def get_revision_graph_with_ghosts(self, revision_ids=None):
781
"""Return a graph of the revisions with ghosts marked as applicable.
783
:param revision_ids: an iterable of revisions to graph or None for all.
784
:return: a Graph object with the graph reachable from revision_ids.
787
vf = self._get_revision_vf()
788
versions = set(vf.versions())
790
pending = set(self.all_revision_ids())
793
pending = set(revision_ids)
794
required = set(revision_ids)
797
revision_id = pending.pop()
798
if not revision_id in versions:
799
if revision_id in required:
800
raise errors.NoSuchRevision(self, revision_id)
802
result.add_ghost(revision_id)
803
# mark it as done so we dont try for it again.
804
done.add(revision_id)
806
parent_ids = vf.get_parents_with_ghosts(revision_id)
807
for parent_id in parent_ids:
808
# is this queued or done ?
809
if (parent_id not in pending and
810
parent_id not in done):
812
pending.add(parent_id)
813
result.add_node(revision_id, parent_ids)
814
done.add(revision_id)
817
def _get_revision_vf(self):
818
""":return: a versioned file containing the revisions."""
819
vf = self._revision_store.get_revision_file(self.get_transaction())
824
"""Reconcile this repository."""
825
from bzrlib.reconcile import KnitReconciler
826
reconciler = KnitReconciler(self)
827
reconciler.reconcile()
830
def revision_parents(self, revid):
831
return self._get_revision_vf().get_parents(rev_id)
833
class RepositoryFormat(object):
834
"""A repository format.
836
Formats provide three things:
837
* An initialization routine to construct repository data on disk.
838
* a format string which is used when the BzrDir supports versioned
840
* an open routine which returns a Repository instance.
842
Formats are placed in an dict by their format string for reference
843
during opening. These should be subclasses of RepositoryFormat
846
Once a format is deprecated, just deprecate the initialize and open
847
methods on the format class. Do not deprecate the object, as the
848
object will be created every system load.
850
Common instance attributes:
851
_matchingbzrdir - the bzrdir format that the repository format was
852
originally written to work with. This can be used if manually
853
constructing a bzrdir and repository, or more commonly for test suite
857
_default_format = None
858
"""The default format used for new repositories."""
861
"""The known formats."""
864
def find_format(klass, a_bzrdir):
865
"""Return the format for the repository object in a_bzrdir."""
867
transport = a_bzrdir.get_repository_transport(None)
868
format_string = transport.get("format").read()
869
return klass._formats[format_string]
870
except errors.NoSuchFile:
871
raise errors.NoRepositoryPresent(a_bzrdir)
873
raise errors.UnknownFormatError(format_string)
875
def _get_control_store(self, repo_transport, control_files):
876
"""Return the control store for this repository."""
877
raise NotImplementedError(self._get_control_store)
880
def get_default_format(klass):
881
"""Return the current default format."""
882
return klass._default_format
884
def get_format_string(self):
885
"""Return the ASCII format string that identifies this format.
887
Note that in pre format ?? repositories the format string is
888
not permitted nor written to disk.
890
raise NotImplementedError(self.get_format_string)
892
def _get_revision_store(self, repo_transport, control_files):
893
"""Return the revision store object for this a_bzrdir."""
894
raise NotImplementedError(self._get_revision_store)
896
def _get_text_rev_store(self,
903
"""Common logic for getting a revision store for a repository.
905
see self._get_revision_store for the subclass-overridable method to
906
get the store for a repository.
908
from bzrlib.store.revision.text import TextRevisionStore
909
dir_mode = control_files._dir_mode
910
file_mode = control_files._file_mode
911
text_store =TextStore(transport.clone(name),
913
compressed=compressed,
916
_revision_store = TextRevisionStore(text_store, serializer)
917
return _revision_store
919
def _get_versioned_file_store(self,
924
versionedfile_class=WeaveFile,
926
weave_transport = control_files._transport.clone(name)
927
dir_mode = control_files._dir_mode
928
file_mode = control_files._file_mode
929
return VersionedFileStore(weave_transport, prefixed=prefixed,
932
versionedfile_class=versionedfile_class,
935
def initialize(self, a_bzrdir, shared=False):
936
"""Initialize a repository of this format in a_bzrdir.
938
:param a_bzrdir: The bzrdir to put the new repository in it.
939
:param shared: The repository should be initialized as a sharable one.
941
This may raise UninitializableFormat if shared repository are not
942
compatible the a_bzrdir.
945
def is_supported(self):
946
"""Is this format supported?
948
Supported formats must be initializable and openable.
949
Unsupported formats may not support initialization or committing or
950
some other features depending on the reason for not being supported.
954
def open(self, a_bzrdir, _found=False):
955
"""Return an instance of this format for the bzrdir a_bzrdir.
957
_found is a private parameter, do not use it.
959
raise NotImplementedError(self.open)
962
def register_format(klass, format):
963
klass._formats[format.get_format_string()] = format
966
def set_default_format(klass, format):
967
klass._default_format = format
970
def unregister_format(klass, format):
971
assert klass._formats[format.get_format_string()] is format
972
del klass._formats[format.get_format_string()]
975
class PreSplitOutRepositoryFormat(RepositoryFormat):
976
"""Base class for the pre split out repository formats."""
978
def initialize(self, a_bzrdir, shared=False, _internal=False):
979
"""Create a weave repository.
981
TODO: when creating split out bzr branch formats, move this to a common
982
base for Format5, Format6. or something like that.
984
from bzrlib.weavefile import write_weave_v5
985
from bzrlib.weave import Weave
988
raise errors.IncompatibleFormat(self, a_bzrdir._format)
991
# always initialized when the bzrdir is.
992
return self.open(a_bzrdir, _found=True)
994
# Create an empty weave
996
bzrlib.weavefile.write_weave_v5(Weave(), sio)
997
empty_weave = sio.getvalue()
999
mutter('creating repository in %s.', a_bzrdir.transport.base)
1000
dirs = ['revision-store', 'weaves']
1001
files = [('inventory.weave', StringIO(empty_weave)),
1004
# FIXME: RBC 20060125 dont peek under the covers
1005
# NB: no need to escape relative paths that are url safe.
1006
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1008
control_files.create_lock()
1009
control_files.lock_write()
1010
control_files._transport.mkdir_multi(dirs,
1011
mode=control_files._dir_mode)
1013
for file, content in files:
1014
control_files.put(file, content)
1016
control_files.unlock()
1017
return self.open(a_bzrdir, _found=True)
1019
def _get_control_store(self, repo_transport, control_files):
1020
"""Return the control store for this repository."""
1021
return self._get_versioned_file_store('',
1026
def _get_text_store(self, transport, control_files):
1027
"""Get a store for file texts for this format."""
1028
raise NotImplementedError(self._get_text_store)
1030
def open(self, a_bzrdir, _found=False):
1031
"""See RepositoryFormat.open()."""
1033
# we are being called directly and must probe.
1034
raise NotImplementedError
1036
repo_transport = a_bzrdir.get_repository_transport(None)
1037
control_files = a_bzrdir._control_files
1038
text_store = self._get_text_store(repo_transport, control_files)
1039
control_store = self._get_control_store(repo_transport, control_files)
1040
_revision_store = self._get_revision_store(repo_transport, control_files)
1041
return AllInOneRepository(_format=self,
1043
_revision_store=_revision_store,
1044
control_store=control_store,
1045
text_store=text_store)
1048
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1049
"""Bzr repository format 4.
1051
This repository format has:
1053
- TextStores for texts, inventories,revisions.
1055
This format is deprecated: it indexes texts using a text id which is
1056
removed in format 5; initializationa and write support for this format
1061
super(RepositoryFormat4, self).__init__()
1062
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1064
def initialize(self, url, shared=False, _internal=False):
1065
"""Format 4 branches cannot be created."""
1066
raise errors.UninitializableFormat(self)
1068
def is_supported(self):
1069
"""Format 4 is not supported.
1071
It is not supported because the model changed from 4 to 5 and the
1072
conversion logic is expensive - so doing it on the fly was not
1077
def _get_control_store(self, repo_transport, control_files):
1078
"""Format 4 repositories have no formal control store at this point.
1080
This will cause any control-file-needing apis to fail - this is desired.
1084
def _get_revision_store(self, repo_transport, control_files):
1085
"""See RepositoryFormat._get_revision_store()."""
1086
from bzrlib.xml4 import serializer_v4
1087
return self._get_text_rev_store(repo_transport,
1090
serializer=serializer_v4)
1092
def _get_text_store(self, transport, control_files):
1093
"""See RepositoryFormat._get_text_store()."""
1096
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1097
"""Bzr control format 5.
1099
This repository format has:
1100
- weaves for file texts and inventory
1102
- TextStores for revisions and signatures.
1106
super(RepositoryFormat5, self).__init__()
1107
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1109
def _get_revision_store(self, repo_transport, control_files):
1110
"""See RepositoryFormat._get_revision_store()."""
1111
"""Return the revision store object for this a_bzrdir."""
1112
return self._get_text_rev_store(repo_transport,
1117
def _get_text_store(self, transport, control_files):
1118
"""See RepositoryFormat._get_text_store()."""
1119
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1122
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1123
"""Bzr control format 6.
1125
This repository format has:
1126
- weaves for file texts and inventory
1127
- hash subdirectory based stores.
1128
- TextStores for revisions and signatures.
1132
super(RepositoryFormat6, self).__init__()
1133
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1135
def _get_revision_store(self, repo_transport, control_files):
1136
"""See RepositoryFormat._get_revision_store()."""
1137
return self._get_text_rev_store(repo_transport,
1143
def _get_text_store(self, transport, control_files):
1144
"""See RepositoryFormat._get_text_store()."""
1145
return self._get_versioned_file_store('weaves', transport, control_files)
1148
class MetaDirRepositoryFormat(RepositoryFormat):
1149
"""Common base class for the new repositories using the metadir layour."""
1152
super(MetaDirRepositoryFormat, self).__init__()
1153
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1155
def _create_control_files(self, a_bzrdir):
1156
"""Create the required files and the initial control_files object."""
1157
# FIXME: RBC 20060125 dont peek under the covers
1158
# NB: no need to escape relative paths that are url safe.
1159
repository_transport = a_bzrdir.get_repository_transport(self)
1160
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1161
control_files.create_lock()
1162
return control_files
1164
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1165
"""Upload the initial blank content."""
1166
control_files = self._create_control_files(a_bzrdir)
1167
control_files.lock_write()
1169
control_files._transport.mkdir_multi(dirs,
1170
mode=control_files._dir_mode)
1171
for file, content in files:
1172
control_files.put(file, content)
1173
for file, content in utf8_files:
1174
control_files.put_utf8(file, content)
1176
control_files.put_utf8('shared-storage', '')
1178
control_files.unlock()
1181
class RepositoryFormat7(MetaDirRepositoryFormat):
1182
"""Bzr repository 7.
1184
This repository format has:
1185
- weaves for file texts and inventory
1186
- hash subdirectory based stores.
1187
- TextStores for revisions and signatures.
1188
- a format marker of its own
1189
- an optional 'shared-storage' flag
1190
- an optional 'no-working-trees' flag
1193
def _get_control_store(self, repo_transport, control_files):
1194
"""Return the control store for this repository."""
1195
return self._get_versioned_file_store('',
1200
def get_format_string(self):
1201
"""See RepositoryFormat.get_format_string()."""
1202
return "Bazaar-NG Repository format 7"
1204
def _get_revision_store(self, repo_transport, control_files):
1205
"""See RepositoryFormat._get_revision_store()."""
1206
return self._get_text_rev_store(repo_transport,
1213
def _get_text_store(self, transport, control_files):
1214
"""See RepositoryFormat._get_text_store()."""
1215
return self._get_versioned_file_store('weaves',
1219
def initialize(self, a_bzrdir, shared=False):
1220
"""Create a weave repository.
1222
:param shared: If true the repository will be initialized as a shared
1225
from bzrlib.weavefile import write_weave_v5
1226
from bzrlib.weave import Weave
1228
# Create an empty weave
1230
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1231
empty_weave = sio.getvalue()
1233
mutter('creating repository in %s.', a_bzrdir.transport.base)
1234
dirs = ['revision-store', 'weaves']
1235
files = [('inventory.weave', StringIO(empty_weave)),
1237
utf8_files = [('format', self.get_format_string())]
1239
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1240
return self.open(a_bzrdir=a_bzrdir, _found=True)
1242
def open(self, a_bzrdir, _found=False, _override_transport=None):
1243
"""See RepositoryFormat.open().
1245
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1246
repository at a slightly different url
1247
than normal. I.e. during 'upgrade'.
1250
format = RepositoryFormat.find_format(a_bzrdir)
1251
assert format.__class__ == self.__class__
1252
if _override_transport is not None:
1253
repo_transport = _override_transport
1255
repo_transport = a_bzrdir.get_repository_transport(None)
1256
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1257
text_store = self._get_text_store(repo_transport, control_files)
1258
control_store = self._get_control_store(repo_transport, control_files)
1259
_revision_store = self._get_revision_store(repo_transport, control_files)
1260
return MetaDirRepository(_format=self,
1262
control_files=control_files,
1263
_revision_store=_revision_store,
1264
control_store=control_store,
1265
text_store=text_store)
1268
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1269
"""Bzr repository knit format 1.
1271
This repository format has:
1272
- knits for file texts and inventory
1273
- hash subdirectory based stores.
1274
- knits for revisions and signatures
1275
- TextStores for revisions and signatures.
1276
- a format marker of its own
1277
- an optional 'shared-storage' flag
1278
- an optional 'no-working-trees' flag
1281
This format was introduced in bzr 0.8.
1284
def _get_control_store(self, repo_transport, control_files):
1285
"""Return the control store for this repository."""
1286
return VersionedFileStore(
1289
file_mode=control_files._file_mode,
1290
versionedfile_class=KnitVersionedFile,
1291
versionedfile_kwargs={'factory':KnitPlainFactory()},
1294
def get_format_string(self):
1295
"""See RepositoryFormat.get_format_string()."""
1296
return "Bazaar-NG Knit Repository Format 1"
1298
def _get_revision_store(self, repo_transport, control_files):
1299
"""See RepositoryFormat._get_revision_store()."""
1300
from bzrlib.store.revision.knit import KnitRevisionStore
1301
versioned_file_store = VersionedFileStore(
1303
file_mode=control_files._file_mode,
1306
versionedfile_class=KnitVersionedFile,
1307
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1310
return KnitRevisionStore(versioned_file_store)
1312
def _get_text_store(self, transport, control_files):
1313
"""See RepositoryFormat._get_text_store()."""
1314
return self._get_versioned_file_store('knits',
1317
versionedfile_class=KnitVersionedFile,
1320
def initialize(self, a_bzrdir, shared=False):
1321
"""Create a knit format 1 repository.
1323
:param shared: If true the repository will be initialized as a shared
1325
XXX NOTE that this current uses a Weave for testing and will become
1326
A Knit in due course.
1328
from bzrlib.weavefile import write_weave_v5
1329
from bzrlib.weave import Weave
1331
# Create an empty weave
1333
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1334
empty_weave = sio.getvalue()
1336
mutter('creating repository in %s.', a_bzrdir.transport.base)
1337
dirs = ['revision-store', 'knits', 'control']
1338
files = [('control/inventory.weave', StringIO(empty_weave)),
1340
utf8_files = [('format', self.get_format_string())]
1342
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1343
repo_transport = a_bzrdir.get_repository_transport(None)
1344
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1345
control_store = self._get_control_store(repo_transport, control_files)
1346
transaction = bzrlib.transactions.WriteTransaction()
1347
# trigger a write of the inventory store.
1348
control_store.get_weave_or_empty('inventory', transaction)
1349
_revision_store = self._get_revision_store(repo_transport, control_files)
1350
_revision_store.has_revision_id('A', transaction)
1351
_revision_store.get_signature_file(transaction)
1352
return self.open(a_bzrdir=a_bzrdir, _found=True)
1354
def open(self, a_bzrdir, _found=False, _override_transport=None):
1355
"""See RepositoryFormat.open().
1357
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1358
repository at a slightly different url
1359
than normal. I.e. during 'upgrade'.
1362
format = RepositoryFormat.find_format(a_bzrdir)
1363
assert format.__class__ == self.__class__
1364
if _override_transport is not None:
1365
repo_transport = _override_transport
1367
repo_transport = a_bzrdir.get_repository_transport(None)
1368
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1369
text_store = self._get_text_store(repo_transport, control_files)
1370
control_store = self._get_control_store(repo_transport, control_files)
1371
_revision_store = self._get_revision_store(repo_transport, control_files)
1372
return KnitRepository(_format=self,
1374
control_files=control_files,
1375
_revision_store=_revision_store,
1376
control_store=control_store,
1377
text_store=text_store)
1380
# formats which have no format string are not discoverable
1381
# and not independently creatable, so are not registered.
1382
_default_format = RepositoryFormat7()
1383
RepositoryFormat.register_format(_default_format)
1384
RepositoryFormat.register_format(RepositoryFormatKnit1())
1385
RepositoryFormat.set_default_format(_default_format)
1386
_legacy_formats = [RepositoryFormat4(),
1387
RepositoryFormat5(),
1388
RepositoryFormat6()]
1391
class InterRepository(InterObject):
1392
"""This class represents operations taking place between two repositories.
1394
Its instances have methods like copy_content and fetch, and contain
1395
references to the source and target repositories these operations can be
1398
Often we will provide convenience methods on 'repository' which carry out
1399
operations with another repository - they will always forward to
1400
InterRepository.get(other).method_name(parameters).
1404
"""The available optimised InterRepository types."""
1407
def copy_content(self, revision_id=None, basis=None):
1408
"""Make a complete copy of the content in self into destination.
1410
This is a destructive operation! Do not use it on existing
1413
:param revision_id: Only copy the content needed to construct
1414
revision_id and its parents.
1415
:param basis: Copy the needed data preferentially from basis.
1418
self.target.set_make_working_trees(self.source.make_working_trees())
1419
except NotImplementedError:
1421
# grab the basis available data
1422
if basis is not None:
1423
self.target.fetch(basis, revision_id=revision_id)
1424
# but dont bother fetching if we have the needed data now.
1425
if (revision_id not in (None, NULL_REVISION) and
1426
self.target.has_revision(revision_id)):
1428
self.target.fetch(self.source, revision_id=revision_id)
1430
def _double_lock(self, lock_source, lock_target):
1431
"""Take out too locks, rolling back the first if the second throws."""
1436
# we want to ensure that we don't leave source locked by mistake.
1437
# and any error on target should not confuse source.
1438
self.source.unlock()
1442
def fetch(self, revision_id=None, pb=None):
1443
"""Fetch the content required to construct revision_id.
1445
The content is copied from source to target.
1447
:param revision_id: if None all content is copied, if NULL_REVISION no
1449
:param pb: optional progress bar to use for progress reports. If not
1450
provided a default one will be created.
1452
Returns the copied revision count and the failed revisions in a tuple:
1455
from bzrlib.fetch import GenericRepoFetcher
1456
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1457
self.source, self.source._format, self.target, self.target._format)
1458
f = GenericRepoFetcher(to_repository=self.target,
1459
from_repository=self.source,
1460
last_revision=revision_id,
1462
return f.count_copied, f.failed_revisions
1464
def lock_read(self):
1465
"""Take out a logical read lock.
1467
This will lock the source branch and the target branch. The source gets
1468
a read lock and the target a read lock.
1470
self._double_lock(self.source.lock_read, self.target.lock_read)
1472
def lock_write(self):
1473
"""Take out a logical write lock.
1475
This will lock the source branch and the target branch. The source gets
1476
a read lock and the target a write lock.
1478
self._double_lock(self.source.lock_read, self.target.lock_write)
1481
def missing_revision_ids(self, revision_id=None):
1482
"""Return the revision ids that source has that target does not.
1484
These are returned in topological order.
1486
:param revision_id: only return revision ids included by this
1489
# generic, possibly worst case, slow code path.
1490
target_ids = set(self.target.all_revision_ids())
1491
if revision_id is not None:
1492
source_ids = self.source.get_ancestry(revision_id)
1493
assert source_ids.pop(0) == None
1495
source_ids = self.source.all_revision_ids()
1496
result_set = set(source_ids).difference(target_ids)
1497
# this may look like a no-op: its not. It preserves the ordering
1498
# other_ids had while only returning the members from other_ids
1499
# that we've decided we need.
1500
return [rev_id for rev_id in source_ids if rev_id in result_set]
1503
"""Release the locks on source and target."""
1505
self.target.unlock()
1507
self.source.unlock()
1510
class InterWeaveRepo(InterRepository):
1511
"""Optimised code paths between Weave based repositories."""
1513
_matching_repo_format = _default_format
1514
"""Repository format for testing with."""
1517
def is_compatible(source, target):
1518
"""Be compatible with known Weave formats.
1520
We dont test for the stores being of specific types becase that
1521
could lead to confusing results, and there is no need to be
1525
return (isinstance(source._format, (RepositoryFormat5,
1527
RepositoryFormat7)) and
1528
isinstance(target._format, (RepositoryFormat5,
1530
RepositoryFormat7)))
1531
except AttributeError:
1535
def copy_content(self, revision_id=None, basis=None):
1536
"""See InterRepository.copy_content()."""
1537
# weave specific optimised path:
1538
if basis is not None:
1539
# copy the basis in, then fetch remaining data.
1540
basis.copy_content_into(self.target, revision_id)
1541
# the basis copy_content_into could misset this.
1543
self.target.set_make_working_trees(self.source.make_working_trees())
1544
except NotImplementedError:
1546
self.target.fetch(self.source, revision_id=revision_id)
1549
self.target.set_make_working_trees(self.source.make_working_trees())
1550
except NotImplementedError:
1552
# FIXME do not peek!
1553
if self.source.control_files._transport.listable():
1554
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1556
self.target.weave_store.copy_all_ids(
1557
self.source.weave_store,
1559
from_transaction=self.source.get_transaction(),
1560
to_transaction=self.target.get_transaction())
1561
pb.update('copying inventory', 0, 1)
1562
self.target.control_weaves.copy_multi(
1563
self.source.control_weaves, ['inventory'],
1564
from_transaction=self.source.get_transaction(),
1565
to_transaction=self.target.get_transaction())
1566
self.target._revision_store.text_store.copy_all_ids(
1567
self.source._revision_store.text_store,
1572
self.target.fetch(self.source, revision_id=revision_id)
1575
def fetch(self, revision_id=None, pb=None):
1576
"""See InterRepository.fetch()."""
1577
from bzrlib.fetch import GenericRepoFetcher
1578
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1579
self.source, self.source._format, self.target, self.target._format)
1580
f = GenericRepoFetcher(to_repository=self.target,
1581
from_repository=self.source,
1582
last_revision=revision_id,
1584
return f.count_copied, f.failed_revisions
1587
def missing_revision_ids(self, revision_id=None):
1588
"""See InterRepository.missing_revision_ids()."""
1589
# we want all revisions to satisfy revision_id in source.
1590
# but we dont want to stat every file here and there.
1591
# we want then, all revisions other needs to satisfy revision_id
1592
# checked, but not those that we have locally.
1593
# so the first thing is to get a subset of the revisions to
1594
# satisfy revision_id in source, and then eliminate those that
1595
# we do already have.
1596
# this is slow on high latency connection to self, but as as this
1597
# disk format scales terribly for push anyway due to rewriting
1598
# inventory.weave, this is considered acceptable.
1600
if revision_id is not None:
1601
source_ids = self.source.get_ancestry(revision_id)
1602
assert source_ids.pop(0) == None
1604
source_ids = self.source._all_possible_ids()
1605
source_ids_set = set(source_ids)
1606
# source_ids is the worst possible case we may need to pull.
1607
# now we want to filter source_ids against what we actually
1608
# have in target, but dont try to check for existence where we know
1609
# we do not have a revision as that would be pointless.
1610
target_ids = set(self.target._all_possible_ids())
1611
possibly_present_revisions = target_ids.intersection(source_ids_set)
1612
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1613
required_revisions = source_ids_set.difference(actually_present_revisions)
1614
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1615
if revision_id is not None:
1616
# we used get_ancestry to determine source_ids then we are assured all
1617
# revisions referenced are present as they are installed in topological order.
1618
# and the tip revision was validated by get_ancestry.
1619
return required_topo_revisions
1621
# if we just grabbed the possibly available ids, then
1622
# we only have an estimate of whats available and need to validate
1623
# that against the revision records.
1624
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1627
class InterKnitRepo(InterRepository):
1628
"""Optimised code paths between Knit based repositories."""
1630
_matching_repo_format = RepositoryFormatKnit1()
1631
"""Repository format for testing with."""
1634
def is_compatible(source, target):
1635
"""Be compatible with known Knit formats.
1637
We dont test for the stores being of specific types becase that
1638
could lead to confusing results, and there is no need to be
1642
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1643
isinstance(target._format, (RepositoryFormatKnit1)))
1644
except AttributeError:
1648
def fetch(self, revision_id=None, pb=None):
1649
"""See InterRepository.fetch()."""
1650
from bzrlib.fetch import KnitRepoFetcher
1651
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1652
self.source, self.source._format, self.target, self.target._format)
1653
f = KnitRepoFetcher(to_repository=self.target,
1654
from_repository=self.source,
1655
last_revision=revision_id,
1657
return f.count_copied, f.failed_revisions
1660
def missing_revision_ids(self, revision_id=None):
1661
"""See InterRepository.missing_revision_ids()."""
1662
if revision_id is not None:
1663
source_ids = self.source.get_ancestry(revision_id)
1664
assert source_ids.pop(0) == None
1666
source_ids = self.source._all_possible_ids()
1667
source_ids_set = set(source_ids)
1668
# source_ids is the worst possible case we may need to pull.
1669
# now we want to filter source_ids against what we actually
1670
# have in target, but dont try to check for existence where we know
1671
# we do not have a revision as that would be pointless.
1672
target_ids = set(self.target._all_possible_ids())
1673
possibly_present_revisions = target_ids.intersection(source_ids_set)
1674
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1675
required_revisions = source_ids_set.difference(actually_present_revisions)
1676
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1677
if revision_id is not None:
1678
# we used get_ancestry to determine source_ids then we are assured all
1679
# revisions referenced are present as they are installed in topological order.
1680
# and the tip revision was validated by get_ancestry.
1681
return required_topo_revisions
1683
# if we just grabbed the possibly available ids, then
1684
# we only have an estimate of whats available and need to validate
1685
# that against the revision records.
1686
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1688
InterRepository.register_optimiser(InterWeaveRepo)
1689
InterRepository.register_optimiser(InterKnitRepo)
1692
class RepositoryTestProviderAdapter(object):
1693
"""A tool to generate a suite testing multiple repository formats at once.
1695
This is done by copying the test once for each transport and injecting
1696
the transport_server, transport_readonly_server, and bzrdir_format and
1697
repository_format classes into each copy. Each copy is also given a new id()
1698
to make it easy to identify.
1701
def __init__(self, transport_server, transport_readonly_server, formats):
1702
self._transport_server = transport_server
1703
self._transport_readonly_server = transport_readonly_server
1704
self._formats = formats
1706
def adapt(self, test):
1707
result = TestSuite()
1708
for repository_format, bzrdir_format in self._formats:
1709
new_test = deepcopy(test)
1710
new_test.transport_server = self._transport_server
1711
new_test.transport_readonly_server = self._transport_readonly_server
1712
new_test.bzrdir_format = bzrdir_format
1713
new_test.repository_format = repository_format
1714
def make_new_test_id():
1715
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1716
return lambda: new_id
1717
new_test.id = make_new_test_id()
1718
result.addTest(new_test)
1722
class InterRepositoryTestProviderAdapter(object):
1723
"""A tool to generate a suite testing multiple inter repository formats.
1725
This is done by copying the test once for each interrepo provider and injecting
1726
the transport_server, transport_readonly_server, repository_format and
1727
repository_to_format classes into each copy.
1728
Each copy is also given a new id() to make it easy to identify.
1731
def __init__(self, transport_server, transport_readonly_server, formats):
1732
self._transport_server = transport_server
1733
self._transport_readonly_server = transport_readonly_server
1734
self._formats = formats
1736
def adapt(self, test):
1737
result = TestSuite()
1738
for interrepo_class, repository_format, repository_format_to in self._formats:
1739
new_test = deepcopy(test)
1740
new_test.transport_server = self._transport_server
1741
new_test.transport_readonly_server = self._transport_readonly_server
1742
new_test.interrepo_class = interrepo_class
1743
new_test.repository_format = repository_format
1744
new_test.repository_format_to = repository_format_to
1745
def make_new_test_id():
1746
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1747
return lambda: new_id
1748
new_test.id = make_new_test_id()
1749
result.addTest(new_test)
1753
def default_test_list():
1754
"""Generate the default list of interrepo permutations to test."""
1756
# test the default InterRepository between format 6 and the current
1758
# XXX: robertc 20060220 reinstate this when there are two supported
1759
# formats which do not have an optimal code path between them.
1760
result.append((InterRepository,
1761
RepositoryFormat6(),
1762
RepositoryFormatKnit1()))
1763
for optimiser in InterRepository._optimisers:
1764
result.append((optimiser,
1765
optimiser._matching_repo_format,
1766
optimiser._matching_repo_format
1768
# if there are specific combinations we want to use, we can add them
1773
class CopyConverter(object):
1774
"""A repository conversion tool which just performs a copy of the content.
1776
This is slow but quite reliable.
1779
def __init__(self, target_format):
1780
"""Create a CopyConverter.
1782
:param target_format: The format the resulting repository should be.
1784
self.target_format = target_format
1786
def convert(self, repo, pb):
1787
"""Perform the conversion of to_convert, giving feedback via pb.
1789
:param to_convert: The disk object to convert.
1790
:param pb: a progress bar to use for progress information.
1795
# this is only useful with metadir layouts - separated repo content.
1796
# trigger an assertion if not such
1797
repo._format.get_format_string()
1798
self.repo_dir = repo.bzrdir
1799
self.step('Moving repository to repository.backup')
1800
self.repo_dir.transport.move('repository', 'repository.backup')
1801
backup_transport = self.repo_dir.transport.clone('repository.backup')
1802
self.source_repo = repo._format.open(self.repo_dir,
1804
_override_transport=backup_transport)
1805
self.step('Creating new repository')
1806
converted = self.target_format.initialize(self.repo_dir,
1807
self.source_repo.is_shared())
1808
converted.lock_write()
1810
self.step('Copying content into repository.')
1811
self.source_repo.copy_content_into(converted)
1814
self.step('Deleting old repository content.')
1815
self.repo_dir.transport.delete_tree('repository.backup')
1816
self.pb.note('repository converted')
1818
def step(self, message):
1819
"""Update the pb by a step."""
1821
self.pb.update(message, self.count, self.total)
1824
# Copied from xml.sax.saxutils
1825
def _unescape_xml(data):
1826
"""Unescape &, <, and > in a string of data.
1828
data = data.replace("<", "<")
1829
data = data.replace(">", ">")
1830
# must do ampersand last
1831
return data.replace("&", "&")