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
raise BzrError("%r is not present in revision %s" % (file, revno))
541
revno = self.revision_id_to_revno(revision_id)
542
except errors.NoSuchRevision:
543
# TODO: This should not be BzrError,
544
# but NoSuchFile doesn't fit either
545
raise BzrError('%r is not present in revision %s'
546
% (file, revision_id))
548
raise BzrError('%r is not present in revision %s'
550
tree.print_file(file_id)
552
def get_transaction(self):
553
return self.control_files.get_transaction()
555
def revision_parents(self, revid):
556
return self.get_inventory_weave().parent_names(revid)
559
def set_make_working_trees(self, new_value):
560
"""Set the policy flag for making working trees when creating branches.
562
This only applies to branches that use this repository.
564
The default is 'True'.
565
:param new_value: True to restore the default, False to disable making
568
raise NotImplementedError(self.set_make_working_trees)
570
def make_working_trees(self):
571
"""Returns the policy for making working trees on new branches."""
572
raise NotImplementedError(self.make_working_trees)
575
def sign_revision(self, revision_id, gpg_strategy):
576
plaintext = Testament.from_revision(self, revision_id).as_short_text()
577
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
580
def has_signature_for_revision_id(self, revision_id):
581
"""Query for a revision signature for revision_id in the repository."""
582
return self._revision_store.has_signature(revision_id,
583
self.get_transaction())
586
def get_signature_text(self, revision_id):
587
"""Return the text for a signature."""
588
return self._revision_store.get_signature_text(revision_id,
589
self.get_transaction())
592
class AllInOneRepository(Repository):
593
"""Legacy support - the repository behaviour for all-in-one branches."""
595
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
596
# we reuse one control files instance.
597
dir_mode = a_bzrdir._control_files._dir_mode
598
file_mode = a_bzrdir._control_files._file_mode
600
def get_store(name, compressed=True, prefixed=False):
601
# FIXME: This approach of assuming stores are all entirely compressed
602
# or entirely uncompressed is tidy, but breaks upgrade from
603
# some existing branches where there's a mixture; we probably
604
# still want the option to look for both.
605
relpath = a_bzrdir._control_files._escape(name)
606
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
607
prefixed=prefixed, compressed=compressed,
610
#if self._transport.should_cache():
611
# cache_path = os.path.join(self.cache_root, name)
612
# os.mkdir(cache_path)
613
# store = bzrlib.store.CachedStore(store, cache_path)
616
# not broken out yet because the controlweaves|inventory_store
617
# and text_store | weave_store bits are still different.
618
if isinstance(_format, RepositoryFormat4):
619
# cannot remove these - there is still no consistent api
620
# which allows access to this old info.
621
self.inventory_store = get_store('inventory-store')
622
text_store = get_store('text-store')
623
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
627
"""AllInOne repositories cannot be shared."""
631
def set_make_working_trees(self, new_value):
632
"""Set the policy flag for making working trees when creating branches.
634
This only applies to branches that use this repository.
636
The default is 'True'.
637
:param new_value: True to restore the default, False to disable making
640
raise NotImplementedError(self.set_make_working_trees)
642
def make_working_trees(self):
643
"""Returns the policy for making working trees on new branches."""
647
class MetaDirRepository(Repository):
648
"""Repositories in the new meta-dir layout."""
650
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
651
super(MetaDirRepository, self).__init__(_format,
658
dir_mode = self.control_files._dir_mode
659
file_mode = self.control_files._file_mode
663
"""Return True if this repository is flagged as a shared repository."""
664
return self.control_files._transport.has('shared-storage')
667
def set_make_working_trees(self, new_value):
668
"""Set the policy flag for making working trees when creating branches.
670
This only applies to branches that use this repository.
672
The default is 'True'.
673
:param new_value: True to restore the default, False to disable making
678
self.control_files._transport.delete('no-working-trees')
679
except errors.NoSuchFile:
682
self.control_files.put_utf8('no-working-trees', '')
684
def make_working_trees(self):
685
"""Returns the policy for making working trees on new branches."""
686
return not self.control_files._transport.has('no-working-trees')
689
class KnitRepository(MetaDirRepository):
690
"""Knit format repository."""
693
def all_revision_ids(self):
694
"""See Repository.all_revision_ids()."""
695
return self._revision_store.all_revision_ids(self.get_transaction())
697
def fileid_involved_between_revs(self, from_revid, to_revid):
698
"""Find file_id(s) which are involved in the changes between revisions.
700
This determines the set of revisions which are involved, and then
701
finds all file ids affected by those revisions.
703
vf = self._get_revision_vf()
704
from_set = set(vf.get_ancestry(from_revid))
705
to_set = set(vf.get_ancestry(to_revid))
706
changed = to_set.difference(from_set)
707
return self._fileid_involved_by_set(changed)
709
def fileid_involved(self, last_revid=None):
710
"""Find all file_ids modified in the ancestry of last_revid.
712
:param last_revid: If None, last_revision() will be used.
715
changed = set(self.all_revision_ids())
717
changed = set(self.get_ancestry(last_revid))
720
return self._fileid_involved_by_set(changed)
723
def get_ancestry(self, revision_id):
724
"""Return a list of revision-ids integrated by a revision.
726
This is topologically sorted.
728
if revision_id is None:
730
vf = self._get_revision_vf()
732
return [None] + vf.get_ancestry(revision_id)
733
except errors.RevisionNotPresent:
734
raise errors.NoSuchRevision(self, revision_id)
737
def get_revision(self, revision_id):
738
"""Return the Revision object for a named revision"""
739
return self.get_revision_reconcile(revision_id)
742
def get_revision_graph(self, revision_id=None):
743
"""Return a dictionary containing the revision graph.
745
:return: a dictionary of revision_id->revision_parents_list.
747
weave = self._get_revision_vf()
748
entire_graph = weave.get_graph()
749
if revision_id is None:
750
return weave.get_graph()
751
elif revision_id not in weave:
752
raise errors.NoSuchRevision(self, revision_id)
754
# add what can be reached from revision_id
756
pending = set([revision_id])
757
while len(pending) > 0:
759
result[node] = weave.get_parents(node)
760
for revision_id in result[node]:
761
if revision_id not in result:
762
pending.add(revision_id)
766
def get_revision_graph_with_ghosts(self, revision_ids=None):
767
"""Return a graph of the revisions with ghosts marked as applicable.
769
:param revision_ids: an iterable of revisions to graph or None for all.
770
:return: a Graph object with the graph reachable from revision_ids.
773
vf = self._get_revision_vf()
774
versions = set(vf.versions())
776
pending = set(self.all_revision_ids())
779
pending = set(revision_ids)
780
required = set(revision_ids)
783
revision_id = pending.pop()
784
if not revision_id in versions:
785
if revision_id in required:
786
raise errors.NoSuchRevision(self, revision_id)
788
result.add_ghost(revision_id)
789
# mark it as done so we dont try for it again.
790
done.add(revision_id)
792
parent_ids = vf.get_parents_with_ghosts(revision_id)
793
for parent_id in parent_ids:
794
# is this queued or done ?
795
if (parent_id not in pending and
796
parent_id not in done):
798
pending.add(parent_id)
799
result.add_node(revision_id, parent_ids)
800
done.add(revision_id)
803
def _get_revision_vf(self):
804
""":return: a versioned file containing the revisions."""
805
vf = self._revision_store.get_revision_file(self.get_transaction())
810
"""Reconcile this repository."""
811
from bzrlib.reconcile import KnitReconciler
812
reconciler = KnitReconciler(self)
813
reconciler.reconcile()
816
def revision_parents(self, revid):
817
return self._get_revision_vf().get_parents(rev_id)
819
class RepositoryFormat(object):
820
"""A repository format.
822
Formats provide three things:
823
* An initialization routine to construct repository data on disk.
824
* a format string which is used when the BzrDir supports versioned
826
* an open routine which returns a Repository instance.
828
Formats are placed in an dict by their format string for reference
829
during opening. These should be subclasses of RepositoryFormat
832
Once a format is deprecated, just deprecate the initialize and open
833
methods on the format class. Do not deprecate the object, as the
834
object will be created every system load.
836
Common instance attributes:
837
_matchingbzrdir - the bzrdir format that the repository format was
838
originally written to work with. This can be used if manually
839
constructing a bzrdir and repository, or more commonly for test suite
843
_default_format = None
844
"""The default format used for new repositories."""
847
"""The known formats."""
850
def find_format(klass, a_bzrdir):
851
"""Return the format for the repository object in a_bzrdir."""
853
transport = a_bzrdir.get_repository_transport(None)
854
format_string = transport.get("format").read()
855
return klass._formats[format_string]
856
except errors.NoSuchFile:
857
raise errors.NoRepositoryPresent(a_bzrdir)
859
raise errors.UnknownFormatError(format_string)
861
def _get_control_store(self, repo_transport, control_files):
862
"""Return the control store for this repository."""
863
raise NotImplementedError(self._get_control_store)
866
def get_default_format(klass):
867
"""Return the current default format."""
868
return klass._default_format
870
def get_format_string(self):
871
"""Return the ASCII format string that identifies this format.
873
Note that in pre format ?? repositories the format string is
874
not permitted nor written to disk.
876
raise NotImplementedError(self.get_format_string)
878
def _get_revision_store(self, repo_transport, control_files):
879
"""Return the revision store object for this a_bzrdir."""
880
raise NotImplementedError(self._get_revision_store)
882
def _get_text_rev_store(self,
889
"""Common logic for getting a revision store for a repository.
891
see self._get_revision_store for the subclass-overridable method to
892
get the store for a repository.
894
from bzrlib.store.revision.text import TextRevisionStore
895
dir_mode = control_files._dir_mode
896
file_mode = control_files._file_mode
897
text_store =TextStore(transport.clone(name),
899
compressed=compressed,
902
_revision_store = TextRevisionStore(text_store, serializer)
903
return _revision_store
905
def _get_versioned_file_store(self,
910
versionedfile_class=WeaveFile,
912
weave_transport = control_files._transport.clone(name)
913
dir_mode = control_files._dir_mode
914
file_mode = control_files._file_mode
915
return VersionedFileStore(weave_transport, prefixed=prefixed,
918
versionedfile_class=versionedfile_class,
921
def initialize(self, a_bzrdir, shared=False):
922
"""Initialize a repository of this format in a_bzrdir.
924
:param a_bzrdir: The bzrdir to put the new repository in it.
925
:param shared: The repository should be initialized as a sharable one.
927
This may raise UninitializableFormat if shared repository are not
928
compatible the a_bzrdir.
931
def is_supported(self):
932
"""Is this format supported?
934
Supported formats must be initializable and openable.
935
Unsupported formats may not support initialization or committing or
936
some other features depending on the reason for not being supported.
940
def open(self, a_bzrdir, _found=False):
941
"""Return an instance of this format for the bzrdir a_bzrdir.
943
_found is a private parameter, do not use it.
945
raise NotImplementedError(self.open)
948
def register_format(klass, format):
949
klass._formats[format.get_format_string()] = format
952
def set_default_format(klass, format):
953
klass._default_format = format
956
def unregister_format(klass, format):
957
assert klass._formats[format.get_format_string()] is format
958
del klass._formats[format.get_format_string()]
961
class PreSplitOutRepositoryFormat(RepositoryFormat):
962
"""Base class for the pre split out repository formats."""
964
def initialize(self, a_bzrdir, shared=False, _internal=False):
965
"""Create a weave repository.
967
TODO: when creating split out bzr branch formats, move this to a common
968
base for Format5, Format6. or something like that.
970
from bzrlib.weavefile import write_weave_v5
971
from bzrlib.weave import Weave
974
raise errors.IncompatibleFormat(self, a_bzrdir._format)
977
# always initialized when the bzrdir is.
978
return self.open(a_bzrdir, _found=True)
980
# Create an empty weave
982
bzrlib.weavefile.write_weave_v5(Weave(), sio)
983
empty_weave = sio.getvalue()
985
mutter('creating repository in %s.', a_bzrdir.transport.base)
986
dirs = ['revision-store', 'weaves']
987
files = [('inventory.weave', StringIO(empty_weave)),
990
# FIXME: RBC 20060125 dont peek under the covers
991
# NB: no need to escape relative paths that are url safe.
992
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
994
control_files.create_lock()
995
control_files.lock_write()
996
control_files._transport.mkdir_multi(dirs,
997
mode=control_files._dir_mode)
999
for file, content in files:
1000
control_files.put(file, content)
1002
control_files.unlock()
1003
return self.open(a_bzrdir, _found=True)
1005
def _get_control_store(self, repo_transport, control_files):
1006
"""Return the control store for this repository."""
1007
return self._get_versioned_file_store('',
1012
def _get_text_store(self, transport, control_files):
1013
"""Get a store for file texts for this format."""
1014
raise NotImplementedError(self._get_text_store)
1016
def open(self, a_bzrdir, _found=False):
1017
"""See RepositoryFormat.open()."""
1019
# we are being called directly and must probe.
1020
raise NotImplementedError
1022
repo_transport = a_bzrdir.get_repository_transport(None)
1023
control_files = a_bzrdir._control_files
1024
text_store = self._get_text_store(repo_transport, control_files)
1025
control_store = self._get_control_store(repo_transport, control_files)
1026
_revision_store = self._get_revision_store(repo_transport, control_files)
1027
return AllInOneRepository(_format=self,
1029
_revision_store=_revision_store,
1030
control_store=control_store,
1031
text_store=text_store)
1034
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1035
"""Bzr repository format 4.
1037
This repository format has:
1039
- TextStores for texts, inventories,revisions.
1041
This format is deprecated: it indexes texts using a text id which is
1042
removed in format 5; initializationa and write support for this format
1047
super(RepositoryFormat4, self).__init__()
1048
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1050
def initialize(self, url, shared=False, _internal=False):
1051
"""Format 4 branches cannot be created."""
1052
raise errors.UninitializableFormat(self)
1054
def is_supported(self):
1055
"""Format 4 is not supported.
1057
It is not supported because the model changed from 4 to 5 and the
1058
conversion logic is expensive - so doing it on the fly was not
1063
def _get_control_store(self, repo_transport, control_files):
1064
"""Format 4 repositories have no formal control store at this point.
1066
This will cause any control-file-needing apis to fail - this is desired.
1070
def _get_revision_store(self, repo_transport, control_files):
1071
"""See RepositoryFormat._get_revision_store()."""
1072
from bzrlib.xml4 import serializer_v4
1073
return self._get_text_rev_store(repo_transport,
1076
serializer=serializer_v4)
1078
def _get_text_store(self, transport, control_files):
1079
"""See RepositoryFormat._get_text_store()."""
1082
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1083
"""Bzr control format 5.
1085
This repository format has:
1086
- weaves for file texts and inventory
1088
- TextStores for revisions and signatures.
1092
super(RepositoryFormat5, self).__init__()
1093
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1095
def _get_revision_store(self, repo_transport, control_files):
1096
"""See RepositoryFormat._get_revision_store()."""
1097
"""Return the revision store object for this a_bzrdir."""
1098
return self._get_text_rev_store(repo_transport,
1103
def _get_text_store(self, transport, control_files):
1104
"""See RepositoryFormat._get_text_store()."""
1105
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1108
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1109
"""Bzr control format 6.
1111
This repository format has:
1112
- weaves for file texts and inventory
1113
- hash subdirectory based stores.
1114
- TextStores for revisions and signatures.
1118
super(RepositoryFormat6, self).__init__()
1119
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1121
def _get_revision_store(self, repo_transport, control_files):
1122
"""See RepositoryFormat._get_revision_store()."""
1123
return self._get_text_rev_store(repo_transport,
1129
def _get_text_store(self, transport, control_files):
1130
"""See RepositoryFormat._get_text_store()."""
1131
return self._get_versioned_file_store('weaves', transport, control_files)
1134
class MetaDirRepositoryFormat(RepositoryFormat):
1135
"""Common base class for the new repositories using the metadir layour."""
1138
super(MetaDirRepositoryFormat, self).__init__()
1139
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1141
def _create_control_files(self, a_bzrdir):
1142
"""Create the required files and the initial control_files object."""
1143
# FIXME: RBC 20060125 dont peek under the covers
1144
# NB: no need to escape relative paths that are url safe.
1145
repository_transport = a_bzrdir.get_repository_transport(self)
1146
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1147
control_files.create_lock()
1148
return control_files
1150
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1151
"""Upload the initial blank content."""
1152
control_files = self._create_control_files(a_bzrdir)
1153
control_files.lock_write()
1155
control_files._transport.mkdir_multi(dirs,
1156
mode=control_files._dir_mode)
1157
for file, content in files:
1158
control_files.put(file, content)
1159
for file, content in utf8_files:
1160
control_files.put_utf8(file, content)
1162
control_files.put_utf8('shared-storage', '')
1164
control_files.unlock()
1167
class RepositoryFormat7(MetaDirRepositoryFormat):
1168
"""Bzr repository 7.
1170
This repository format has:
1171
- weaves for file texts and inventory
1172
- hash subdirectory based stores.
1173
- TextStores for revisions and signatures.
1174
- a format marker of its own
1175
- an optional 'shared-storage' flag
1176
- an optional 'no-working-trees' flag
1179
def _get_control_store(self, repo_transport, control_files):
1180
"""Return the control store for this repository."""
1181
return self._get_versioned_file_store('',
1186
def get_format_string(self):
1187
"""See RepositoryFormat.get_format_string()."""
1188
return "Bazaar-NG Repository format 7"
1190
def _get_revision_store(self, repo_transport, control_files):
1191
"""See RepositoryFormat._get_revision_store()."""
1192
return self._get_text_rev_store(repo_transport,
1199
def _get_text_store(self, transport, control_files):
1200
"""See RepositoryFormat._get_text_store()."""
1201
return self._get_versioned_file_store('weaves',
1205
def initialize(self, a_bzrdir, shared=False):
1206
"""Create a weave repository.
1208
:param shared: If true the repository will be initialized as a shared
1211
from bzrlib.weavefile import write_weave_v5
1212
from bzrlib.weave import Weave
1214
# Create an empty weave
1216
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1217
empty_weave = sio.getvalue()
1219
mutter('creating repository in %s.', a_bzrdir.transport.base)
1220
dirs = ['revision-store', 'weaves']
1221
files = [('inventory.weave', StringIO(empty_weave)),
1223
utf8_files = [('format', self.get_format_string())]
1225
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1226
return self.open(a_bzrdir=a_bzrdir, _found=True)
1228
def open(self, a_bzrdir, _found=False, _override_transport=None):
1229
"""See RepositoryFormat.open().
1231
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1232
repository at a slightly different url
1233
than normal. I.e. during 'upgrade'.
1236
format = RepositoryFormat.find_format(a_bzrdir)
1237
assert format.__class__ == self.__class__
1238
if _override_transport is not None:
1239
repo_transport = _override_transport
1241
repo_transport = a_bzrdir.get_repository_transport(None)
1242
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1243
text_store = self._get_text_store(repo_transport, control_files)
1244
control_store = self._get_control_store(repo_transport, control_files)
1245
_revision_store = self._get_revision_store(repo_transport, control_files)
1246
return MetaDirRepository(_format=self,
1248
control_files=control_files,
1249
_revision_store=_revision_store,
1250
control_store=control_store,
1251
text_store=text_store)
1254
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1255
"""Bzr repository knit format 1.
1257
This repository format has:
1258
- knits for file texts and inventory
1259
- hash subdirectory based stores.
1260
- knits for revisions and signatures
1261
- TextStores for revisions and signatures.
1262
- a format marker of its own
1263
- an optional 'shared-storage' flag
1264
- an optional 'no-working-trees' flag
1267
This format was introduced in bzr 0.8.
1270
def _get_control_store(self, repo_transport, control_files):
1271
"""Return the control store for this repository."""
1272
return VersionedFileStore(
1275
file_mode=control_files._file_mode,
1276
versionedfile_class=KnitVersionedFile,
1277
versionedfile_kwargs={'factory':KnitPlainFactory()},
1280
def get_format_string(self):
1281
"""See RepositoryFormat.get_format_string()."""
1282
return "Bazaar-NG Knit Repository Format 1"
1284
def _get_revision_store(self, repo_transport, control_files):
1285
"""See RepositoryFormat._get_revision_store()."""
1286
from bzrlib.store.revision.knit import KnitRevisionStore
1287
versioned_file_store = VersionedFileStore(
1289
file_mode=control_files._file_mode,
1292
versionedfile_class=KnitVersionedFile,
1293
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1296
return KnitRevisionStore(versioned_file_store)
1298
def _get_text_store(self, transport, control_files):
1299
"""See RepositoryFormat._get_text_store()."""
1300
return self._get_versioned_file_store('knits',
1303
versionedfile_class=KnitVersionedFile,
1306
def initialize(self, a_bzrdir, shared=False):
1307
"""Create a knit format 1 repository.
1309
:param shared: If true the repository will be initialized as a shared
1311
XXX NOTE that this current uses a Weave for testing and will become
1312
A Knit in due course.
1314
from bzrlib.weavefile import write_weave_v5
1315
from bzrlib.weave import Weave
1317
# Create an empty weave
1319
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1320
empty_weave = sio.getvalue()
1322
mutter('creating repository in %s.', a_bzrdir.transport.base)
1323
dirs = ['revision-store', 'knits', 'control']
1324
files = [('control/inventory.weave', StringIO(empty_weave)),
1326
utf8_files = [('format', self.get_format_string())]
1328
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1329
repo_transport = a_bzrdir.get_repository_transport(None)
1330
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1331
control_store = self._get_control_store(repo_transport, control_files)
1332
transaction = bzrlib.transactions.WriteTransaction()
1333
# trigger a write of the inventory store.
1334
control_store.get_weave_or_empty('inventory', transaction)
1335
_revision_store = self._get_revision_store(repo_transport, control_files)
1336
_revision_store.has_revision_id('A', transaction)
1337
_revision_store.get_signature_file(transaction)
1338
return self.open(a_bzrdir=a_bzrdir, _found=True)
1340
def open(self, a_bzrdir, _found=False, _override_transport=None):
1341
"""See RepositoryFormat.open().
1343
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1344
repository at a slightly different url
1345
than normal. I.e. during 'upgrade'.
1348
format = RepositoryFormat.find_format(a_bzrdir)
1349
assert format.__class__ == self.__class__
1350
if _override_transport is not None:
1351
repo_transport = _override_transport
1353
repo_transport = a_bzrdir.get_repository_transport(None)
1354
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1355
text_store = self._get_text_store(repo_transport, control_files)
1356
control_store = self._get_control_store(repo_transport, control_files)
1357
_revision_store = self._get_revision_store(repo_transport, control_files)
1358
return KnitRepository(_format=self,
1360
control_files=control_files,
1361
_revision_store=_revision_store,
1362
control_store=control_store,
1363
text_store=text_store)
1366
# formats which have no format string are not discoverable
1367
# and not independently creatable, so are not registered.
1368
_default_format = RepositoryFormat7()
1369
RepositoryFormat.register_format(_default_format)
1370
RepositoryFormat.register_format(RepositoryFormatKnit1())
1371
RepositoryFormat.set_default_format(_default_format)
1372
_legacy_formats = [RepositoryFormat4(),
1373
RepositoryFormat5(),
1374
RepositoryFormat6()]
1377
class InterRepository(InterObject):
1378
"""This class represents operations taking place between two repositories.
1380
Its instances have methods like copy_content and fetch, and contain
1381
references to the source and target repositories these operations can be
1384
Often we will provide convenience methods on 'repository' which carry out
1385
operations with another repository - they will always forward to
1386
InterRepository.get(other).method_name(parameters).
1390
"""The available optimised InterRepository types."""
1393
def copy_content(self, revision_id=None, basis=None):
1394
"""Make a complete copy of the content in self into destination.
1396
This is a destructive operation! Do not use it on existing
1399
:param revision_id: Only copy the content needed to construct
1400
revision_id and its parents.
1401
:param basis: Copy the needed data preferentially from basis.
1404
self.target.set_make_working_trees(self.source.make_working_trees())
1405
except NotImplementedError:
1407
# grab the basis available data
1408
if basis is not None:
1409
self.target.fetch(basis, revision_id=revision_id)
1410
# but dont bother fetching if we have the needed data now.
1411
if (revision_id not in (None, NULL_REVISION) and
1412
self.target.has_revision(revision_id)):
1414
self.target.fetch(self.source, revision_id=revision_id)
1416
def _double_lock(self, lock_source, lock_target):
1417
"""Take out too locks, rolling back the first if the second throws."""
1422
# we want to ensure that we don't leave source locked by mistake.
1423
# and any error on target should not confuse source.
1424
self.source.unlock()
1428
def fetch(self, revision_id=None, pb=None):
1429
"""Fetch the content required to construct revision_id.
1431
The content is copied from source to target.
1433
:param revision_id: if None all content is copied, if NULL_REVISION no
1435
:param pb: optional progress bar to use for progress reports. If not
1436
provided a default one will be created.
1438
Returns the copied revision count and the failed revisions in a tuple:
1441
from bzrlib.fetch import GenericRepoFetcher
1442
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1443
self.source, self.source._format, self.target, self.target._format)
1444
f = GenericRepoFetcher(to_repository=self.target,
1445
from_repository=self.source,
1446
last_revision=revision_id,
1448
return f.count_copied, f.failed_revisions
1450
def lock_read(self):
1451
"""Take out a logical read lock.
1453
This will lock the source branch and the target branch. The source gets
1454
a read lock and the target a read lock.
1456
self._double_lock(self.source.lock_read, self.target.lock_read)
1458
def lock_write(self):
1459
"""Take out a logical write lock.
1461
This will lock the source branch and the target branch. The source gets
1462
a read lock and the target a write lock.
1464
self._double_lock(self.source.lock_read, self.target.lock_write)
1467
def missing_revision_ids(self, revision_id=None):
1468
"""Return the revision ids that source has that target does not.
1470
These are returned in topological order.
1472
:param revision_id: only return revision ids included by this
1475
# generic, possibly worst case, slow code path.
1476
target_ids = set(self.target.all_revision_ids())
1477
if revision_id is not None:
1478
source_ids = self.source.get_ancestry(revision_id)
1479
assert source_ids.pop(0) == None
1481
source_ids = self.source.all_revision_ids()
1482
result_set = set(source_ids).difference(target_ids)
1483
# this may look like a no-op: its not. It preserves the ordering
1484
# other_ids had while only returning the members from other_ids
1485
# that we've decided we need.
1486
return [rev_id for rev_id in source_ids if rev_id in result_set]
1489
"""Release the locks on source and target."""
1491
self.target.unlock()
1493
self.source.unlock()
1496
class InterWeaveRepo(InterRepository):
1497
"""Optimised code paths between Weave based repositories."""
1499
_matching_repo_format = _default_format
1500
"""Repository format for testing with."""
1503
def is_compatible(source, target):
1504
"""Be compatible with known Weave formats.
1506
We dont test for the stores being of specific types becase that
1507
could lead to confusing results, and there is no need to be
1511
return (isinstance(source._format, (RepositoryFormat5,
1513
RepositoryFormat7)) and
1514
isinstance(target._format, (RepositoryFormat5,
1516
RepositoryFormat7)))
1517
except AttributeError:
1521
def copy_content(self, revision_id=None, basis=None):
1522
"""See InterRepository.copy_content()."""
1523
# weave specific optimised path:
1524
if basis is not None:
1525
# copy the basis in, then fetch remaining data.
1526
basis.copy_content_into(self.target, revision_id)
1527
# the basis copy_content_into could misset this.
1529
self.target.set_make_working_trees(self.source.make_working_trees())
1530
except NotImplementedError:
1532
self.target.fetch(self.source, revision_id=revision_id)
1535
self.target.set_make_working_trees(self.source.make_working_trees())
1536
except NotImplementedError:
1538
# FIXME do not peek!
1539
if self.source.control_files._transport.listable():
1540
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1542
self.target.weave_store.copy_all_ids(
1543
self.source.weave_store,
1545
from_transaction=self.source.get_transaction(),
1546
to_transaction=self.target.get_transaction())
1547
pb.update('copying inventory', 0, 1)
1548
self.target.control_weaves.copy_multi(
1549
self.source.control_weaves, ['inventory'],
1550
from_transaction=self.source.get_transaction(),
1551
to_transaction=self.target.get_transaction())
1552
self.target._revision_store.text_store.copy_all_ids(
1553
self.source._revision_store.text_store,
1558
self.target.fetch(self.source, revision_id=revision_id)
1561
def fetch(self, revision_id=None, pb=None):
1562
"""See InterRepository.fetch()."""
1563
from bzrlib.fetch import GenericRepoFetcher
1564
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1565
self.source, self.source._format, self.target, self.target._format)
1566
f = GenericRepoFetcher(to_repository=self.target,
1567
from_repository=self.source,
1568
last_revision=revision_id,
1570
return f.count_copied, f.failed_revisions
1573
def missing_revision_ids(self, revision_id=None):
1574
"""See InterRepository.missing_revision_ids()."""
1575
# we want all revisions to satisfy revision_id in source.
1576
# but we dont want to stat every file here and there.
1577
# we want then, all revisions other needs to satisfy revision_id
1578
# checked, but not those that we have locally.
1579
# so the first thing is to get a subset of the revisions to
1580
# satisfy revision_id in source, and then eliminate those that
1581
# we do already have.
1582
# this is slow on high latency connection to self, but as as this
1583
# disk format scales terribly for push anyway due to rewriting
1584
# inventory.weave, this is considered acceptable.
1586
if revision_id is not None:
1587
source_ids = self.source.get_ancestry(revision_id)
1588
assert source_ids.pop(0) == None
1590
source_ids = self.source._all_possible_ids()
1591
source_ids_set = set(source_ids)
1592
# source_ids is the worst possible case we may need to pull.
1593
# now we want to filter source_ids against what we actually
1594
# have in target, but dont try to check for existence where we know
1595
# we do not have a revision as that would be pointless.
1596
target_ids = set(self.target._all_possible_ids())
1597
possibly_present_revisions = target_ids.intersection(source_ids_set)
1598
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1599
required_revisions = source_ids_set.difference(actually_present_revisions)
1600
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1601
if revision_id is not None:
1602
# we used get_ancestry to determine source_ids then we are assured all
1603
# revisions referenced are present as they are installed in topological order.
1604
# and the tip revision was validated by get_ancestry.
1605
return required_topo_revisions
1607
# if we just grabbed the possibly available ids, then
1608
# we only have an estimate of whats available and need to validate
1609
# that against the revision records.
1610
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1613
class InterKnitRepo(InterRepository):
1614
"""Optimised code paths between Knit based repositories."""
1616
_matching_repo_format = RepositoryFormatKnit1()
1617
"""Repository format for testing with."""
1620
def is_compatible(source, target):
1621
"""Be compatible with known Knit formats.
1623
We dont test for the stores being of specific types becase that
1624
could lead to confusing results, and there is no need to be
1628
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1629
isinstance(target._format, (RepositoryFormatKnit1)))
1630
except AttributeError:
1634
def fetch(self, revision_id=None, pb=None):
1635
"""See InterRepository.fetch()."""
1636
from bzrlib.fetch import KnitRepoFetcher
1637
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1638
self.source, self.source._format, self.target, self.target._format)
1639
f = KnitRepoFetcher(to_repository=self.target,
1640
from_repository=self.source,
1641
last_revision=revision_id,
1643
return f.count_copied, f.failed_revisions
1646
def missing_revision_ids(self, revision_id=None):
1647
"""See InterRepository.missing_revision_ids()."""
1648
if revision_id is not None:
1649
source_ids = self.source.get_ancestry(revision_id)
1650
assert source_ids.pop(0) == None
1652
source_ids = self.source._all_possible_ids()
1653
source_ids_set = set(source_ids)
1654
# source_ids is the worst possible case we may need to pull.
1655
# now we want to filter source_ids against what we actually
1656
# have in target, but dont try to check for existence where we know
1657
# we do not have a revision as that would be pointless.
1658
target_ids = set(self.target._all_possible_ids())
1659
possibly_present_revisions = target_ids.intersection(source_ids_set)
1660
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1661
required_revisions = source_ids_set.difference(actually_present_revisions)
1662
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1663
if revision_id is not None:
1664
# we used get_ancestry to determine source_ids then we are assured all
1665
# revisions referenced are present as they are installed in topological order.
1666
# and the tip revision was validated by get_ancestry.
1667
return required_topo_revisions
1669
# if we just grabbed the possibly available ids, then
1670
# we only have an estimate of whats available and need to validate
1671
# that against the revision records.
1672
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1674
InterRepository.register_optimiser(InterWeaveRepo)
1675
InterRepository.register_optimiser(InterKnitRepo)
1678
class RepositoryTestProviderAdapter(object):
1679
"""A tool to generate a suite testing multiple repository formats at once.
1681
This is done by copying the test once for each transport and injecting
1682
the transport_server, transport_readonly_server, and bzrdir_format and
1683
repository_format classes into each copy. Each copy is also given a new id()
1684
to make it easy to identify.
1687
def __init__(self, transport_server, transport_readonly_server, formats):
1688
self._transport_server = transport_server
1689
self._transport_readonly_server = transport_readonly_server
1690
self._formats = formats
1692
def adapt(self, test):
1693
result = TestSuite()
1694
for repository_format, bzrdir_format in self._formats:
1695
new_test = deepcopy(test)
1696
new_test.transport_server = self._transport_server
1697
new_test.transport_readonly_server = self._transport_readonly_server
1698
new_test.bzrdir_format = bzrdir_format
1699
new_test.repository_format = repository_format
1700
def make_new_test_id():
1701
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1702
return lambda: new_id
1703
new_test.id = make_new_test_id()
1704
result.addTest(new_test)
1708
class InterRepositoryTestProviderAdapter(object):
1709
"""A tool to generate a suite testing multiple inter repository formats.
1711
This is done by copying the test once for each interrepo provider and injecting
1712
the transport_server, transport_readonly_server, repository_format and
1713
repository_to_format classes into each copy.
1714
Each copy is also given a new id() to make it easy to identify.
1717
def __init__(self, transport_server, transport_readonly_server, formats):
1718
self._transport_server = transport_server
1719
self._transport_readonly_server = transport_readonly_server
1720
self._formats = formats
1722
def adapt(self, test):
1723
result = TestSuite()
1724
for interrepo_class, repository_format, repository_format_to in self._formats:
1725
new_test = deepcopy(test)
1726
new_test.transport_server = self._transport_server
1727
new_test.transport_readonly_server = self._transport_readonly_server
1728
new_test.interrepo_class = interrepo_class
1729
new_test.repository_format = repository_format
1730
new_test.repository_format_to = repository_format_to
1731
def make_new_test_id():
1732
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1733
return lambda: new_id
1734
new_test.id = make_new_test_id()
1735
result.addTest(new_test)
1739
def default_test_list():
1740
"""Generate the default list of interrepo permutations to test."""
1742
# test the default InterRepository between format 6 and the current
1744
# XXX: robertc 20060220 reinstate this when there are two supported
1745
# formats which do not have an optimal code path between them.
1746
result.append((InterRepository,
1747
RepositoryFormat6(),
1748
RepositoryFormatKnit1()))
1749
for optimiser in InterRepository._optimisers:
1750
result.append((optimiser,
1751
optimiser._matching_repo_format,
1752
optimiser._matching_repo_format
1754
# if there are specific combinations we want to use, we can add them
1759
class CopyConverter(object):
1760
"""A repository conversion tool which just performs a copy of the content.
1762
This is slow but quite reliable.
1765
def __init__(self, target_format):
1766
"""Create a CopyConverter.
1768
:param target_format: The format the resulting repository should be.
1770
self.target_format = target_format
1772
def convert(self, repo, pb):
1773
"""Perform the conversion of to_convert, giving feedback via pb.
1775
:param to_convert: The disk object to convert.
1776
:param pb: a progress bar to use for progress information.
1781
# this is only useful with metadir layouts - separated repo content.
1782
# trigger an assertion if not such
1783
repo._format.get_format_string()
1784
self.repo_dir = repo.bzrdir
1785
self.step('Moving repository to repository.backup')
1786
self.repo_dir.transport.move('repository', 'repository.backup')
1787
backup_transport = self.repo_dir.transport.clone('repository.backup')
1788
self.source_repo = repo._format.open(self.repo_dir,
1790
_override_transport=backup_transport)
1791
self.step('Creating new repository')
1792
converted = self.target_format.initialize(self.repo_dir,
1793
self.source_repo.is_shared())
1794
converted.lock_write()
1796
self.step('Copying content into repository.')
1797
self.source_repo.copy_content_into(converted)
1800
self.step('Deleting old repository content.')
1801
self.repo_dir.transport.delete_tree('repository.backup')
1802
self.pb.note('repository converted')
1804
def step(self, message):
1805
"""Update the pb by a step."""
1807
self.pb.update(message, self.count, self.total)
1810
# Copied from xml.sax.saxutils
1811
def _unescape_xml(data):
1812
"""Unescape &, <, and > in a string of data.
1814
data = data.replace("<", "<")
1815
data = data.replace(">", ">")
1816
# must do ampersand last
1817
return data.replace("&", "&")