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
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
object.__init__(self)
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
160
def lock_write(self):
161
self.control_files.lock_write()
164
self.control_files.lock_read()
167
return self.control_files.is_locked()
170
def missing_revision_ids(self, other, revision_id=None):
171
"""Return the revision ids that other has that this does not.
173
These are returned in topological order.
175
revision_id: only return revision ids included by revision_id.
177
return InterRepository.get(other, self).missing_revision_ids(revision_id)
181
"""Open the repository rooted at base.
183
For instance, if the repository is at URL/.bzr/repository,
184
Repository.open(URL) -> a Repository instance.
186
control = bzrlib.bzrdir.BzrDir.open(base)
187
return control.open_repository()
189
def copy_content_into(self, destination, revision_id=None, basis=None):
190
"""Make a complete copy of the content in self into destination.
192
This is a destructive operation! Do not use it on existing
195
return InterRepository.get(self, destination).copy_content(revision_id, basis)
197
def fetch(self, source, revision_id=None, pb=None):
198
"""Fetch the content required to construct revision_id from source.
200
If revision_id is None all content is copied.
202
return InterRepository.get(source, self).fetch(revision_id=revision_id,
206
self.control_files.unlock()
209
def clone(self, a_bzrdir, revision_id=None, basis=None):
210
"""Clone this repository into a_bzrdir using the current format.
212
Currently no check is made that the format of this repository and
213
the bzrdir format are compatible. FIXME RBC 20060201.
215
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
216
# use target default format.
217
result = a_bzrdir.create_repository()
218
# FIXME RBC 20060209 split out the repository type to avoid this check ?
219
elif isinstance(a_bzrdir._format,
220
(bzrlib.bzrdir.BzrDirFormat4,
221
bzrlib.bzrdir.BzrDirFormat5,
222
bzrlib.bzrdir.BzrDirFormat6)):
223
result = a_bzrdir.open_repository()
225
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
226
self.copy_content_into(result, revision_id, basis)
230
def has_revision(self, revision_id):
231
"""True if this repository has a copy of the revision."""
232
return self._revision_store.has_revision_id(revision_id,
233
self.get_transaction())
236
def get_revision_reconcile(self, revision_id):
237
"""'reconcile' helper routine that allows access to a revision always.
239
This variant of get_revision does not cross check the weave graph
240
against the revision one as get_revision does: but it should only
241
be used by reconcile, or reconcile-alike commands that are correcting
242
or testing the revision graph.
244
if not revision_id or not isinstance(revision_id, basestring):
245
raise InvalidRevisionId(revision_id=revision_id, branch=self)
246
return self._revision_store.get_revision(revision_id,
247
self.get_transaction())
250
def get_revision_xml(self, revision_id):
251
rev = self.get_revision(revision_id)
253
# the current serializer..
254
self._revision_store._serializer.write_revision(rev, rev_tmp)
256
return rev_tmp.getvalue()
259
def get_revision(self, revision_id):
260
"""Return the Revision object for a named revision"""
261
r = self.get_revision_reconcile(revision_id)
262
# weave corruption can lead to absent revision markers that should be
264
# the following test is reasonably cheap (it needs a single weave read)
265
# and the weave is cached in read transactions. In write transactions
266
# it is not cached but typically we only read a small number of
267
# revisions. For knits when they are introduced we will probably want
268
# to ensure that caching write transactions are in use.
269
inv = self.get_inventory_weave()
270
self._check_revision_parents(r, inv)
273
def _check_revision_parents(self, revision, inventory):
274
"""Private to Repository and Fetch.
276
This checks the parentage of revision in an inventory weave for
277
consistency and is only applicable to inventory-weave-for-ancestry
278
using repository formats & fetchers.
280
weave_parents = inventory.get_parents(revision.revision_id)
281
weave_names = inventory.versions()
282
for parent_id in revision.parent_ids:
283
if parent_id in weave_names:
284
# this parent must not be a ghost.
285
if not parent_id in weave_parents:
287
raise errors.CorruptRepository(self)
290
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
291
signature = gpg_strategy.sign(plaintext)
292
self._revision_store.add_revision_signature_text(revision_id,
294
self.get_transaction())
296
def fileid_involved_between_revs(self, from_revid, to_revid):
297
"""Find file_id(s) which are involved in the changes between revisions.
299
This determines the set of revisions which are involved, and then
300
finds all file ids affected by those revisions.
302
w = self.get_inventory_weave()
303
from_set = set(w.get_ancestry(from_revid))
304
to_set = set(w.get_ancestry(to_revid))
305
changed = to_set.difference(from_set)
306
return self._fileid_involved_by_set(changed)
308
def fileid_involved(self, last_revid=None):
309
"""Find all file_ids modified in the ancestry of last_revid.
311
:param last_revid: If None, last_revision() will be used.
313
w = self.get_inventory_weave()
315
changed = set(w.versions())
317
changed = set(w.get_ancestry(last_revid))
318
return self._fileid_involved_by_set(changed)
320
def fileid_involved_by_set(self, changes):
321
"""Find all file_ids modified by the set of revisions passed in.
323
:param changes: A set() of revision ids
325
# TODO: jam 20060119 This line does *nothing*, remove it.
326
# or better yet, change _fileid_involved_by_set so
327
# that it takes the inventory weave, rather than
328
# pulling it out by itself.
329
return self._fileid_involved_by_set(changes)
331
def _fileid_involved_by_set(self, changes):
332
"""Find the set of file-ids affected by the set of revisions.
334
:param changes: A set() of revision ids.
335
:return: A set() of file ids.
337
This peaks at the Weave, interpreting each line, looking to
338
see if it mentions one of the revisions. And if so, includes
339
the file id mentioned.
340
This expects both the Weave format, and the serialization
341
to have a single line per file/directory, and to have
342
fileid="" and revision="" on that line.
344
assert isinstance(self._format, (RepositoryFormat5,
347
RepositoryFormatKnit1)), \
348
"fileid_involved only supported for branches which store inventory as unnested xml"
350
w = self.get_inventory_weave()
353
# this code needs to read every line in every inventory for the
354
# inventories [changes]. Seeing a line twice is ok. Seeing a line
355
# not pesent in one of those inventories is unnecessary and not
356
# harmful because we are filtering by the revision id marker in the
357
# inventory lines to only select file ids altered in one of those
358
# revisions. We dont need to see all lines in the inventory because
359
# only those added in an inventory in rev X can contain a revision=X
361
for line in w.iter_lines_added_or_present_in_versions(changes):
362
start = line.find('file_id="')+9
363
if start < 9: continue
364
end = line.find('"', start)
366
file_id = _unescape_xml(line[start:end])
368
# check if file_id is already present
369
if file_id in file_ids: continue
371
start = line.find('revision="')+10
372
if start < 10: continue
373
end = line.find('"', start)
375
revision_id = _unescape_xml(line[start:end])
376
if revision_id in changes:
377
file_ids.add(file_id)
381
def get_inventory_weave(self):
382
return self.control_weaves.get_weave('inventory',
383
self.get_transaction())
386
def get_inventory(self, revision_id):
387
"""Get Inventory object by hash."""
388
xml = self.get_inventory_xml(revision_id)
389
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
392
def get_inventory_xml(self, revision_id):
393
"""Get inventory XML as a file object."""
395
assert isinstance(revision_id, basestring), type(revision_id)
396
iw = self.get_inventory_weave()
397
return iw.get_text(revision_id)
399
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
402
def get_inventory_sha1(self, revision_id):
403
"""Return the sha1 hash of the inventory entry
405
return self.get_revision(revision_id).inventory_sha1
408
def get_revision_graph(self, revision_id=None):
409
"""Return a dictionary containing the revision graph.
411
:return: a dictionary of revision_id->revision_parents_list.
413
weave = self.get_inventory_weave()
414
all_revisions = self._eliminate_revisions_not_present(weave.versions())
415
entire_graph = dict([(node, weave.get_parents(node)) for
416
node in all_revisions])
417
if revision_id is None:
419
elif revision_id not in entire_graph:
420
raise errors.NoSuchRevision(self, revision_id)
422
# add what can be reached from revision_id
424
pending = set([revision_id])
425
while len(pending) > 0:
427
result[node] = entire_graph[node]
428
for revision_id in result[node]:
429
if revision_id not in result:
430
pending.add(revision_id)
434
def get_revision_graph_with_ghosts(self, revision_ids=None):
435
"""Return a graph of the revisions with ghosts marked as applicable.
437
:param revision_ids: an iterable of revisions to graph or None for all.
438
:return: a Graph object with the graph reachable from revision_ids.
442
pending = set(self.all_revision_ids())
445
pending = set(revision_ids)
446
required = set(revision_ids)
449
revision_id = pending.pop()
451
rev = self.get_revision(revision_id)
452
except errors.NoSuchRevision:
453
if revision_id in required:
456
result.add_ghost(revision_id)
458
for parent_id in rev.parent_ids:
459
# is this queued or done ?
460
if (parent_id not in pending and
461
parent_id not in done):
463
pending.add(parent_id)
464
result.add_node(revision_id, rev.parent_ids)
465
done.add(revision_id)
469
def get_revision_inventory(self, revision_id):
470
"""Return inventory of a past revision."""
471
# TODO: Unify this with get_inventory()
472
# bzr 0.0.6 and later imposes the constraint that the inventory_id
473
# must be the same as its revision, so this is trivial.
474
if revision_id is None:
475
# This does not make sense: if there is no revision,
476
# then it is the current tree inventory surely ?!
477
# and thus get_root_id() is something that looks at the last
478
# commit on the branch, and the get_root_id is an inventory check.
479
raise NotImplementedError
480
# return Inventory(self.get_root_id())
482
return self.get_inventory(revision_id)
486
"""Return True if this repository is flagged as a shared repository."""
487
raise NotImplementedError(self.is_shared)
491
"""Reconcile this repository."""
492
from bzrlib.reconcile import RepoReconciler
493
reconciler = RepoReconciler(self)
494
reconciler.reconcile()
498
def revision_tree(self, revision_id):
499
"""Return Tree for a revision on this branch.
501
`revision_id` may be None for the null revision, in which case
502
an `EmptyTree` is returned."""
503
# TODO: refactor this to use an existing revision object
504
# so we don't need to read it in twice.
505
if revision_id is None or revision_id == NULL_REVISION:
508
inv = self.get_revision_inventory(revision_id)
509
return RevisionTree(self, inv, revision_id)
512
def get_ancestry(self, revision_id):
513
"""Return a list of revision-ids integrated by a revision.
515
This is topologically sorted.
517
if revision_id is None:
519
if not self.has_revision(revision_id):
520
raise errors.NoSuchRevision(self, revision_id)
521
w = self.get_inventory_weave()
522
candidates = w.get_ancestry(revision_id)
523
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
526
def print_file(self, file, revision_id):
527
"""Print `file` to stdout.
529
FIXME RBC 20060125 as John Meinel points out this is a bad api
530
- it writes to stdout, it assumes that that is valid etc. Fix
531
by creating a new more flexible convenience function.
533
tree = self.revision_tree(revision_id)
534
# use inventory as it was in that revision
535
file_id = tree.inventory.path2id(file)
537
raise BzrError("%r is not present in revision %s" % (file, revno))
539
revno = self.revision_id_to_revno(revision_id)
540
except errors.NoSuchRevision:
541
# TODO: This should not be BzrError,
542
# but NoSuchFile doesn't fit either
543
raise BzrError('%r is not present in revision %s'
544
% (file, revision_id))
546
raise BzrError('%r is not present in revision %s'
548
tree.print_file(file_id)
550
def get_transaction(self):
551
return self.control_files.get_transaction()
553
def revision_parents(self, revid):
554
return self.get_inventory_weave().parent_names(revid)
557
def set_make_working_trees(self, new_value):
558
"""Set the policy flag for making working trees when creating branches.
560
This only applies to branches that use this repository.
562
The default is 'True'.
563
:param new_value: True to restore the default, False to disable making
566
raise NotImplementedError(self.set_make_working_trees)
568
def make_working_trees(self):
569
"""Returns the policy for making working trees on new branches."""
570
raise NotImplementedError(self.make_working_trees)
573
def sign_revision(self, revision_id, gpg_strategy):
574
plaintext = Testament.from_revision(self, revision_id).as_short_text()
575
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
578
def has_signature_for_revision_id(self, revision_id):
579
"""Query for a revision signature for revision_id in the repository."""
580
return self._revision_store.has_signature(revision_id,
581
self.get_transaction())
584
def get_signature_text(self, revision_id):
585
"""Return the text for a signature."""
586
return self._revision_store.get_signature_text(revision_id,
587
self.get_transaction())
590
class AllInOneRepository(Repository):
591
"""Legacy support - the repository behaviour for all-in-one branches."""
593
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
594
# we reuse one control files instance.
595
dir_mode = a_bzrdir._control_files._dir_mode
596
file_mode = a_bzrdir._control_files._file_mode
598
def get_store(name, compressed=True, prefixed=False):
599
# FIXME: This approach of assuming stores are all entirely compressed
600
# or entirely uncompressed is tidy, but breaks upgrade from
601
# some existing branches where there's a mixture; we probably
602
# still want the option to look for both.
603
relpath = a_bzrdir._control_files._escape(name)
604
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
605
prefixed=prefixed, compressed=compressed,
608
#if self._transport.should_cache():
609
# cache_path = os.path.join(self.cache_root, name)
610
# os.mkdir(cache_path)
611
# store = bzrlib.store.CachedStore(store, cache_path)
614
# not broken out yet because the controlweaves|inventory_store
615
# and text_store | weave_store bits are still different.
616
if isinstance(_format, RepositoryFormat4):
617
# cannot remove these - there is still no consistent api
618
# which allows access to this old info.
619
self.inventory_store = get_store('inventory-store')
620
text_store = get_store('text-store')
621
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
625
"""AllInOne repositories cannot be shared."""
629
def set_make_working_trees(self, new_value):
630
"""Set the policy flag for making working trees when creating branches.
632
This only applies to branches that use this repository.
634
The default is 'True'.
635
:param new_value: True to restore the default, False to disable making
638
raise NotImplementedError(self.set_make_working_trees)
640
def make_working_trees(self):
641
"""Returns the policy for making working trees on new branches."""
645
class MetaDirRepository(Repository):
646
"""Repositories in the new meta-dir layout."""
648
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
649
super(MetaDirRepository, self).__init__(_format,
656
dir_mode = self.control_files._dir_mode
657
file_mode = self.control_files._file_mode
661
"""Return True if this repository is flagged as a shared repository."""
662
return self.control_files._transport.has('shared-storage')
665
def set_make_working_trees(self, new_value):
666
"""Set the policy flag for making working trees when creating branches.
668
This only applies to branches that use this repository.
670
The default is 'True'.
671
:param new_value: True to restore the default, False to disable making
676
self.control_files._transport.delete('no-working-trees')
677
except errors.NoSuchFile:
680
self.control_files.put_utf8('no-working-trees', '')
682
def make_working_trees(self):
683
"""Returns the policy for making working trees on new branches."""
684
return not self.control_files._transport.has('no-working-trees')
687
class KnitRepository(MetaDirRepository):
688
"""Knit format repository."""
691
def all_revision_ids(self):
692
"""See Repository.all_revision_ids()."""
693
return self._revision_store.all_revision_ids(self.get_transaction())
695
def fileid_involved_between_revs(self, from_revid, to_revid):
696
"""Find file_id(s) which are involved in the changes between revisions.
698
This determines the set of revisions which are involved, and then
699
finds all file ids affected by those revisions.
701
vf = self._get_revision_vf()
702
from_set = set(vf.get_ancestry(from_revid))
703
to_set = set(vf.get_ancestry(to_revid))
704
changed = to_set.difference(from_set)
705
return self._fileid_involved_by_set(changed)
707
def fileid_involved(self, last_revid=None):
708
"""Find all file_ids modified in the ancestry of last_revid.
710
:param last_revid: If None, last_revision() will be used.
713
changed = set(self.all_revision_ids())
715
changed = set(self.get_ancestry(last_revid))
718
return self._fileid_involved_by_set(changed)
721
def get_ancestry(self, revision_id):
722
"""Return a list of revision-ids integrated by a revision.
724
This is topologically sorted.
726
if revision_id is None:
728
vf = self._get_revision_vf()
730
return [None] + vf.get_ancestry(revision_id)
731
except errors.RevisionNotPresent:
732
raise errors.NoSuchRevision(self, revision_id)
735
def get_revision(self, revision_id):
736
"""Return the Revision object for a named revision"""
737
return self.get_revision_reconcile(revision_id)
740
def get_revision_graph(self, revision_id=None):
741
"""Return a dictionary containing the revision graph.
743
:return: a dictionary of revision_id->revision_parents_list.
745
weave = self._get_revision_vf()
746
entire_graph = weave.get_graph()
747
if revision_id is None:
748
return weave.get_graph()
749
elif revision_id not in weave:
750
raise errors.NoSuchRevision(self, revision_id)
752
# add what can be reached from revision_id
754
pending = set([revision_id])
755
while len(pending) > 0:
757
result[node] = weave.get_parents(node)
758
for revision_id in result[node]:
759
if revision_id not in result:
760
pending.add(revision_id)
764
def get_revision_graph_with_ghosts(self, revision_ids=None):
765
"""Return a graph of the revisions with ghosts marked as applicable.
767
:param revision_ids: an iterable of revisions to graph or None for all.
768
:return: a Graph object with the graph reachable from revision_ids.
771
vf = self._get_revision_vf()
772
versions = set(vf.versions())
774
pending = set(self.all_revision_ids())
777
pending = set(revision_ids)
778
required = set(revision_ids)
781
revision_id = pending.pop()
782
if not revision_id in versions:
783
if revision_id in required:
784
raise errors.NoSuchRevision(self, revision_id)
786
result.add_ghost(revision_id)
787
# mark it as done so we dont try for it again.
788
done.add(revision_id)
790
parent_ids = vf.get_parents_with_ghosts(revision_id)
791
for parent_id in parent_ids:
792
# is this queued or done ?
793
if (parent_id not in pending and
794
parent_id not in done):
796
pending.add(parent_id)
797
result.add_node(revision_id, parent_ids)
798
done.add(revision_id)
801
def _get_revision_vf(self):
802
""":return: a versioned file containing the revisions."""
803
vf = self._revision_store.get_revision_file(self.get_transaction())
808
"""Reconcile this repository."""
809
from bzrlib.reconcile import KnitReconciler
810
reconciler = KnitReconciler(self)
811
reconciler.reconcile()
814
def revision_parents(self, revid):
815
return self._get_revision_vf().get_parents(rev_id)
817
class RepositoryFormat(object):
818
"""A repository format.
820
Formats provide three things:
821
* An initialization routine to construct repository data on disk.
822
* a format string which is used when the BzrDir supports versioned
824
* an open routine which returns a Repository instance.
826
Formats are placed in an dict by their format string for reference
827
during opening. These should be subclasses of RepositoryFormat
830
Once a format is deprecated, just deprecate the initialize and open
831
methods on the format class. Do not deprecate the object, as the
832
object will be created every system load.
834
Common instance attributes:
835
_matchingbzrdir - the bzrdir format that the repository format was
836
originally written to work with. This can be used if manually
837
constructing a bzrdir and repository, or more commonly for test suite
841
_default_format = None
842
"""The default format used for new repositories."""
845
"""The known formats."""
848
def find_format(klass, a_bzrdir):
849
"""Return the format for the repository object in a_bzrdir."""
851
transport = a_bzrdir.get_repository_transport(None)
852
format_string = transport.get("format").read()
853
return klass._formats[format_string]
854
except errors.NoSuchFile:
855
raise errors.NoRepositoryPresent(a_bzrdir)
857
raise errors.UnknownFormatError(format_string)
859
def _get_control_store(self, repo_transport, control_files):
860
"""Return the control store for this repository."""
861
raise NotImplementedError(self._get_control_store)
864
def get_default_format(klass):
865
"""Return the current default format."""
866
return klass._default_format
868
def get_format_string(self):
869
"""Return the ASCII format string that identifies this format.
871
Note that in pre format ?? repositories the format string is
872
not permitted nor written to disk.
874
raise NotImplementedError(self.get_format_string)
876
def _get_revision_store(self, repo_transport, control_files):
877
"""Return the revision store object for this a_bzrdir."""
878
raise NotImplementedError(self._get_revision_store)
880
def _get_text_rev_store(self,
887
"""Common logic for getting a revision store for a repository.
889
see self._get_revision_store for the subclass-overridable method to
890
get the store for a repository.
892
from bzrlib.store.revision.text import TextRevisionStore
893
dir_mode = control_files._dir_mode
894
file_mode = control_files._file_mode
895
text_store =TextStore(transport.clone(name),
897
compressed=compressed,
900
_revision_store = TextRevisionStore(text_store, serializer)
901
return _revision_store
903
def _get_versioned_file_store(self,
908
versionedfile_class=WeaveFile):
909
weave_transport = control_files._transport.clone(name)
910
dir_mode = control_files._dir_mode
911
file_mode = control_files._file_mode
912
return VersionedFileStore(weave_transport, prefixed=prefixed,
915
versionedfile_class=versionedfile_class)
917
def initialize(self, a_bzrdir, shared=False):
918
"""Initialize a repository of this format in a_bzrdir.
920
:param a_bzrdir: The bzrdir to put the new repository in it.
921
:param shared: The repository should be initialized as a sharable one.
923
This may raise UninitializableFormat if shared repository are not
924
compatible the a_bzrdir.
927
def is_supported(self):
928
"""Is this format supported?
930
Supported formats must be initializable and openable.
931
Unsupported formats may not support initialization or committing or
932
some other features depending on the reason for not being supported.
936
def open(self, a_bzrdir, _found=False):
937
"""Return an instance of this format for the bzrdir a_bzrdir.
939
_found is a private parameter, do not use it.
941
raise NotImplementedError(self.open)
944
def register_format(klass, format):
945
klass._formats[format.get_format_string()] = format
948
def set_default_format(klass, format):
949
klass._default_format = format
952
def unregister_format(klass, format):
953
assert klass._formats[format.get_format_string()] is format
954
del klass._formats[format.get_format_string()]
957
class PreSplitOutRepositoryFormat(RepositoryFormat):
958
"""Base class for the pre split out repository formats."""
960
def initialize(self, a_bzrdir, shared=False, _internal=False):
961
"""Create a weave repository.
963
TODO: when creating split out bzr branch formats, move this to a common
964
base for Format5, Format6. or something like that.
966
from bzrlib.weavefile import write_weave_v5
967
from bzrlib.weave import Weave
970
raise errors.IncompatibleFormat(self, a_bzrdir._format)
973
# always initialized when the bzrdir is.
974
return self.open(a_bzrdir, _found=True)
976
# Create an empty weave
978
bzrlib.weavefile.write_weave_v5(Weave(), sio)
979
empty_weave = sio.getvalue()
981
mutter('creating repository in %s.', a_bzrdir.transport.base)
982
dirs = ['revision-store', 'weaves']
983
files = [('inventory.weave', StringIO(empty_weave)),
986
# FIXME: RBC 20060125 dont peek under the covers
987
# NB: no need to escape relative paths that are url safe.
988
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
990
control_files.create_lock()
991
control_files.lock_write()
992
control_files._transport.mkdir_multi(dirs,
993
mode=control_files._dir_mode)
995
for file, content in files:
996
control_files.put(file, content)
998
control_files.unlock()
999
return self.open(a_bzrdir, _found=True)
1001
def _get_control_store(self, repo_transport, control_files):
1002
"""Return the control store for this repository."""
1003
return self._get_versioned_file_store('',
1008
def _get_text_store(self, transport, control_files):
1009
"""Get a store for file texts for this format."""
1010
raise NotImplementedError(self._get_text_store)
1012
def open(self, a_bzrdir, _found=False):
1013
"""See RepositoryFormat.open()."""
1015
# we are being called directly and must probe.
1016
raise NotImplementedError
1018
repo_transport = a_bzrdir.get_repository_transport(None)
1019
control_files = a_bzrdir._control_files
1020
text_store = self._get_text_store(repo_transport, control_files)
1021
control_store = self._get_control_store(repo_transport, control_files)
1022
_revision_store = self._get_revision_store(repo_transport, control_files)
1023
return AllInOneRepository(_format=self,
1025
_revision_store=_revision_store,
1026
control_store=control_store,
1027
text_store=text_store)
1030
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1031
"""Bzr repository format 4.
1033
This repository format has:
1035
- TextStores for texts, inventories,revisions.
1037
This format is deprecated: it indexes texts using a text id which is
1038
removed in format 5; initializationa and write support for this format
1043
super(RepositoryFormat4, self).__init__()
1044
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1046
def initialize(self, url, shared=False, _internal=False):
1047
"""Format 4 branches cannot be created."""
1048
raise errors.UninitializableFormat(self)
1050
def is_supported(self):
1051
"""Format 4 is not supported.
1053
It is not supported because the model changed from 4 to 5 and the
1054
conversion logic is expensive - so doing it on the fly was not
1059
def _get_control_store(self, repo_transport, control_files):
1060
"""Format 4 repositories have no formal control store at this point.
1062
This will cause any control-file-needing apis to fail - this is desired.
1066
def _get_revision_store(self, repo_transport, control_files):
1067
"""See RepositoryFormat._get_revision_store()."""
1068
from bzrlib.xml4 import serializer_v4
1069
return self._get_text_rev_store(repo_transport,
1072
serializer=serializer_v4)
1074
def _get_text_store(self, transport, control_files):
1075
"""See RepositoryFormat._get_text_store()."""
1078
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1079
"""Bzr control format 5.
1081
This repository format has:
1082
- weaves for file texts and inventory
1084
- TextStores for revisions and signatures.
1088
super(RepositoryFormat5, self).__init__()
1089
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1091
def _get_revision_store(self, repo_transport, control_files):
1092
"""See RepositoryFormat._get_revision_store()."""
1093
"""Return the revision store object for this a_bzrdir."""
1094
return self._get_text_rev_store(repo_transport,
1099
def _get_text_store(self, transport, control_files):
1100
"""See RepositoryFormat._get_text_store()."""
1101
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1104
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1105
"""Bzr control format 6.
1107
This repository format has:
1108
- weaves for file texts and inventory
1109
- hash subdirectory based stores.
1110
- TextStores for revisions and signatures.
1114
super(RepositoryFormat6, self).__init__()
1115
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1117
def _get_revision_store(self, repo_transport, control_files):
1118
"""See RepositoryFormat._get_revision_store()."""
1119
return self._get_text_rev_store(repo_transport,
1125
def _get_text_store(self, transport, control_files):
1126
"""See RepositoryFormat._get_text_store()."""
1127
return self._get_versioned_file_store('weaves', transport, control_files)
1130
class MetaDirRepositoryFormat(RepositoryFormat):
1131
"""Common base class for the new repositories using the metadir layour."""
1134
super(MetaDirRepositoryFormat, self).__init__()
1135
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1137
def _create_control_files(self, a_bzrdir):
1138
"""Create the required files and the initial control_files object."""
1139
# FIXME: RBC 20060125 dont peek under the covers
1140
# NB: no need to escape relative paths that are url safe.
1141
repository_transport = a_bzrdir.get_repository_transport(self)
1142
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1143
control_files.create_lock()
1144
return control_files
1146
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1147
"""Upload the initial blank content."""
1148
control_files = self._create_control_files(a_bzrdir)
1149
control_files.lock_write()
1151
control_files._transport.mkdir_multi(dirs,
1152
mode=control_files._dir_mode)
1153
for file, content in files:
1154
control_files.put(file, content)
1155
for file, content in utf8_files:
1156
control_files.put_utf8(file, content)
1158
control_files.put_utf8('shared-storage', '')
1160
control_files.unlock()
1163
class RepositoryFormat7(MetaDirRepositoryFormat):
1164
"""Bzr repository 7.
1166
This repository format has:
1167
- weaves for file texts and inventory
1168
- hash subdirectory based stores.
1169
- TextStores for revisions and signatures.
1170
- a format marker of its own
1171
- an optional 'shared-storage' flag
1172
- an optional 'no-working-trees' flag
1175
def _get_control_store(self, repo_transport, control_files):
1176
"""Return the control store for this repository."""
1177
return self._get_versioned_file_store('',
1182
def get_format_string(self):
1183
"""See RepositoryFormat.get_format_string()."""
1184
return "Bazaar-NG Repository format 7"
1186
def _get_revision_store(self, repo_transport, control_files):
1187
"""See RepositoryFormat._get_revision_store()."""
1188
return self._get_text_rev_store(repo_transport,
1195
def _get_text_store(self, transport, control_files):
1196
"""See RepositoryFormat._get_text_store()."""
1197
return self._get_versioned_file_store('weaves',
1201
def initialize(self, a_bzrdir, shared=False):
1202
"""Create a weave repository.
1204
:param shared: If true the repository will be initialized as a shared
1207
from bzrlib.weavefile import write_weave_v5
1208
from bzrlib.weave import Weave
1210
# Create an empty weave
1212
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1213
empty_weave = sio.getvalue()
1215
mutter('creating repository in %s.', a_bzrdir.transport.base)
1216
dirs = ['revision-store', 'weaves']
1217
files = [('inventory.weave', StringIO(empty_weave)),
1219
utf8_files = [('format', self.get_format_string())]
1221
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1222
return self.open(a_bzrdir=a_bzrdir, _found=True)
1224
def open(self, a_bzrdir, _found=False, _override_transport=None):
1225
"""See RepositoryFormat.open().
1227
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1228
repository at a slightly different url
1229
than normal. I.e. during 'upgrade'.
1232
format = RepositoryFormat.find_format(a_bzrdir)
1233
assert format.__class__ == self.__class__
1234
if _override_transport is not None:
1235
repo_transport = _override_transport
1237
repo_transport = a_bzrdir.get_repository_transport(None)
1238
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1239
text_store = self._get_text_store(repo_transport, control_files)
1240
control_store = self._get_control_store(repo_transport, control_files)
1241
_revision_store = self._get_revision_store(repo_transport, control_files)
1242
return MetaDirRepository(_format=self,
1244
control_files=control_files,
1245
_revision_store=_revision_store,
1246
control_store=control_store,
1247
text_store=text_store)
1250
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1251
"""Bzr repository knit format 1.
1253
This repository format has:
1254
- knits for file texts and inventory
1255
- hash subdirectory based stores.
1256
- knits for revisions and signatures
1257
- TextStores for revisions and signatures.
1258
- a format marker of its own
1259
- an optional 'shared-storage' flag
1260
- an optional 'no-working-trees' flag
1264
def _get_control_store(self, repo_transport, control_files):
1265
"""Return the control store for this repository."""
1266
return VersionedFileStore(
1269
file_mode=control_files._file_mode,
1270
versionedfile_class=KnitVersionedFile,
1271
versionedfile_kwargs={'factory':KnitPlainFactory()},
1274
def get_format_string(self):
1275
"""See RepositoryFormat.get_format_string()."""
1276
return "Bazaar-NG Knit Repository Format 1"
1278
def _get_revision_store(self, repo_transport, control_files):
1279
"""See RepositoryFormat._get_revision_store()."""
1280
from bzrlib.store.revision.knit import KnitRevisionStore
1281
versioned_file_store = VersionedFileStore(
1283
file_mode = control_files._file_mode,
1286
versionedfile_class=KnitVersionedFile,
1287
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()})
1288
return KnitRevisionStore(versioned_file_store)
1290
def _get_text_store(self, transport, control_files):
1291
"""See RepositoryFormat._get_text_store()."""
1292
return self._get_versioned_file_store('knits',
1295
versionedfile_class=KnitVersionedFile)
1297
def initialize(self, a_bzrdir, shared=False):
1298
"""Create a knit format 1 repository.
1300
:param shared: If true the repository will be initialized as a shared
1302
XXX NOTE that this current uses a Weave for testing and will become
1303
A Knit in due course.
1305
from bzrlib.weavefile import write_weave_v5
1306
from bzrlib.weave import Weave
1308
# Create an empty weave
1310
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1311
empty_weave = sio.getvalue()
1313
mutter('creating repository in %s.', a_bzrdir.transport.base)
1314
dirs = ['revision-store', 'knits', 'control']
1315
files = [('control/inventory.weave', StringIO(empty_weave)),
1317
utf8_files = [('format', self.get_format_string())]
1319
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1320
repo_transport = a_bzrdir.get_repository_transport(None)
1321
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1322
control_store = self._get_control_store(repo_transport, control_files)
1323
transaction = bzrlib.transactions.WriteTransaction()
1324
# trigger a write of the inventory store.
1325
control_store.get_weave_or_empty('inventory', transaction)
1326
_revision_store = self._get_revision_store(repo_transport, control_files)
1327
_revision_store.has_revision_id('A', transaction)
1328
_revision_store.get_signature_file(transaction)
1329
return self.open(a_bzrdir=a_bzrdir, _found=True)
1331
def open(self, a_bzrdir, _found=False, _override_transport=None):
1332
"""See RepositoryFormat.open().
1334
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1335
repository at a slightly different url
1336
than normal. I.e. during 'upgrade'.
1339
format = RepositoryFormat.find_format(a_bzrdir)
1340
assert format.__class__ == self.__class__
1341
if _override_transport is not None:
1342
repo_transport = _override_transport
1344
repo_transport = a_bzrdir.get_repository_transport(None)
1345
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1346
text_store = self._get_text_store(repo_transport, control_files)
1347
control_store = self._get_control_store(repo_transport, control_files)
1348
_revision_store = self._get_revision_store(repo_transport, control_files)
1349
return KnitRepository(_format=self,
1351
control_files=control_files,
1352
_revision_store=_revision_store,
1353
control_store=control_store,
1354
text_store=text_store)
1357
# formats which have no format string are not discoverable
1358
# and not independently creatable, so are not registered.
1359
_default_format = RepositoryFormat7()
1360
RepositoryFormat.register_format(_default_format)
1361
RepositoryFormat.register_format(RepositoryFormatKnit1())
1362
RepositoryFormat.set_default_format(_default_format)
1363
_legacy_formats = [RepositoryFormat4(),
1364
RepositoryFormat5(),
1365
RepositoryFormat6()]
1368
class InterRepository(InterObject):
1369
"""This class represents operations taking place between two repositories.
1371
Its instances have methods like copy_content and fetch, and contain
1372
references to the source and target repositories these operations can be
1375
Often we will provide convenience methods on 'repository' which carry out
1376
operations with another repository - they will always forward to
1377
InterRepository.get(other).method_name(parameters).
1381
"""The available optimised InterRepository types."""
1384
def copy_content(self, revision_id=None, basis=None):
1385
"""Make a complete copy of the content in self into destination.
1387
This is a destructive operation! Do not use it on existing
1390
:param revision_id: Only copy the content needed to construct
1391
revision_id and its parents.
1392
:param basis: Copy the needed data preferentially from basis.
1395
self.target.set_make_working_trees(self.source.make_working_trees())
1396
except NotImplementedError:
1398
# grab the basis available data
1399
if basis is not None:
1400
self.target.fetch(basis, revision_id=revision_id)
1401
# but dont bother fetching if we have the needed data now.
1402
if (revision_id not in (None, NULL_REVISION) and
1403
self.target.has_revision(revision_id)):
1405
self.target.fetch(self.source, revision_id=revision_id)
1407
def _double_lock(self, lock_source, lock_target):
1408
"""Take out too locks, rolling back the first if the second throws."""
1413
# we want to ensure that we don't leave source locked by mistake.
1414
# and any error on target should not confuse source.
1415
self.source.unlock()
1419
def fetch(self, revision_id=None, pb=None):
1420
"""Fetch the content required to construct revision_id.
1422
The content is copied from source to target.
1424
:param revision_id: if None all content is copied, if NULL_REVISION no
1426
:param pb: optional progress bar to use for progress reports. If not
1427
provided a default one will be created.
1429
Returns the copied revision count and the failed revisions in a tuple:
1432
from bzrlib.fetch import GenericRepoFetcher
1433
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1434
self.source, self.source._format, self.target, self.target._format)
1435
f = GenericRepoFetcher(to_repository=self.target,
1436
from_repository=self.source,
1437
last_revision=revision_id,
1439
return f.count_copied, f.failed_revisions
1441
def lock_read(self):
1442
"""Take out a logical read lock.
1444
This will lock the source branch and the target branch. The source gets
1445
a read lock and the target a read lock.
1447
self._double_lock(self.source.lock_read, self.target.lock_read)
1449
def lock_write(self):
1450
"""Take out a logical write lock.
1452
This will lock the source branch and the target branch. The source gets
1453
a read lock and the target a write lock.
1455
self._double_lock(self.source.lock_read, self.target.lock_write)
1458
def missing_revision_ids(self, revision_id=None):
1459
"""Return the revision ids that source has that target does not.
1461
These are returned in topological order.
1463
:param revision_id: only return revision ids included by this
1466
# generic, possibly worst case, slow code path.
1467
target_ids = set(self.target.all_revision_ids())
1468
if revision_id is not None:
1469
source_ids = self.source.get_ancestry(revision_id)
1470
assert source_ids.pop(0) == None
1472
source_ids = self.source.all_revision_ids()
1473
result_set = set(source_ids).difference(target_ids)
1474
# this may look like a no-op: its not. It preserves the ordering
1475
# other_ids had while only returning the members from other_ids
1476
# that we've decided we need.
1477
return [rev_id for rev_id in source_ids if rev_id in result_set]
1480
"""Release the locks on source and target."""
1482
self.target.unlock()
1484
self.source.unlock()
1487
class InterWeaveRepo(InterRepository):
1488
"""Optimised code paths between Weave based repositories."""
1490
_matching_repo_format = _default_format
1491
"""Repository format for testing with."""
1494
def is_compatible(source, target):
1495
"""Be compatible with known Weave formats.
1497
We dont test for the stores being of specific types becase that
1498
could lead to confusing results, and there is no need to be
1502
return (isinstance(source._format, (RepositoryFormat5,
1504
RepositoryFormat7)) and
1505
isinstance(target._format, (RepositoryFormat5,
1507
RepositoryFormat7)))
1508
except AttributeError:
1512
def copy_content(self, revision_id=None, basis=None):
1513
"""See InterRepository.copy_content()."""
1514
# weave specific optimised path:
1515
if basis is not None:
1516
# copy the basis in, then fetch remaining data.
1517
basis.copy_content_into(self.target, revision_id)
1518
# the basis copy_content_into could misset this.
1520
self.target.set_make_working_trees(self.source.make_working_trees())
1521
except NotImplementedError:
1523
self.target.fetch(self.source, revision_id=revision_id)
1526
self.target.set_make_working_trees(self.source.make_working_trees())
1527
except NotImplementedError:
1529
# FIXME do not peek!
1530
if self.source.control_files._transport.listable():
1531
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1533
self.target.weave_store.copy_all_ids(
1534
self.source.weave_store,
1536
from_transaction=self.source.get_transaction(),
1537
to_transaction=self.target.get_transaction())
1538
pb.update('copying inventory', 0, 1)
1539
self.target.control_weaves.copy_multi(
1540
self.source.control_weaves, ['inventory'],
1541
from_transaction=self.source.get_transaction(),
1542
to_transaction=self.target.get_transaction())
1543
self.target._revision_store.text_store.copy_all_ids(
1544
self.source._revision_store.text_store,
1549
self.target.fetch(self.source, revision_id=revision_id)
1552
def fetch(self, revision_id=None, pb=None):
1553
"""See InterRepository.fetch()."""
1554
from bzrlib.fetch import GenericRepoFetcher
1555
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1556
self.source, self.source._format, self.target, self.target._format)
1557
f = GenericRepoFetcher(to_repository=self.target,
1558
from_repository=self.source,
1559
last_revision=revision_id,
1561
return f.count_copied, f.failed_revisions
1564
def missing_revision_ids(self, revision_id=None):
1565
"""See InterRepository.missing_revision_ids()."""
1566
# we want all revisions to satisfy revision_id in source.
1567
# but we dont want to stat every file here and there.
1568
# we want then, all revisions other needs to satisfy revision_id
1569
# checked, but not those that we have locally.
1570
# so the first thing is to get a subset of the revisions to
1571
# satisfy revision_id in source, and then eliminate those that
1572
# we do already have.
1573
# this is slow on high latency connection to self, but as as this
1574
# disk format scales terribly for push anyway due to rewriting
1575
# inventory.weave, this is considered acceptable.
1577
if revision_id is not None:
1578
source_ids = self.source.get_ancestry(revision_id)
1579
assert source_ids.pop(0) == None
1581
source_ids = self.source._all_possible_ids()
1582
source_ids_set = set(source_ids)
1583
# source_ids is the worst possible case we may need to pull.
1584
# now we want to filter source_ids against what we actually
1585
# have in target, but dont try to check for existence where we know
1586
# we do not have a revision as that would be pointless.
1587
target_ids = set(self.target._all_possible_ids())
1588
possibly_present_revisions = target_ids.intersection(source_ids_set)
1589
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1590
required_revisions = source_ids_set.difference(actually_present_revisions)
1591
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1592
if revision_id is not None:
1593
# we used get_ancestry to determine source_ids then we are assured all
1594
# revisions referenced are present as they are installed in topological order.
1595
# and the tip revision was validated by get_ancestry.
1596
return required_topo_revisions
1598
# if we just grabbed the possibly available ids, then
1599
# we only have an estimate of whats available and need to validate
1600
# that against the revision records.
1601
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1604
class InterKnitRepo(InterRepository):
1605
"""Optimised code paths between Knit based repositories."""
1607
_matching_repo_format = RepositoryFormatKnit1()
1608
"""Repository format for testing with."""
1611
def is_compatible(source, target):
1612
"""Be compatible with known Knit formats.
1614
We dont test for the stores being of specific types becase that
1615
could lead to confusing results, and there is no need to be
1619
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1620
isinstance(target._format, (RepositoryFormatKnit1)))
1621
except AttributeError:
1625
def fetch(self, revision_id=None, pb=None):
1626
"""See InterRepository.fetch()."""
1627
from bzrlib.fetch import KnitRepoFetcher
1628
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1629
self.source, self.source._format, self.target, self.target._format)
1630
f = KnitRepoFetcher(to_repository=self.target,
1631
from_repository=self.source,
1632
last_revision=revision_id,
1634
return f.count_copied, f.failed_revisions
1637
def missing_revision_ids(self, revision_id=None):
1638
"""See InterRepository.missing_revision_ids()."""
1639
if revision_id is not None:
1640
source_ids = self.source.get_ancestry(revision_id)
1641
assert source_ids.pop(0) == None
1643
source_ids = self.source._all_possible_ids()
1644
source_ids_set = set(source_ids)
1645
# source_ids is the worst possible case we may need to pull.
1646
# now we want to filter source_ids against what we actually
1647
# have in target, but dont try to check for existence where we know
1648
# we do not have a revision as that would be pointless.
1649
target_ids = set(self.target._all_possible_ids())
1650
possibly_present_revisions = target_ids.intersection(source_ids_set)
1651
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1652
required_revisions = source_ids_set.difference(actually_present_revisions)
1653
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1654
if revision_id is not None:
1655
# we used get_ancestry to determine source_ids then we are assured all
1656
# revisions referenced are present as they are installed in topological order.
1657
# and the tip revision was validated by get_ancestry.
1658
return required_topo_revisions
1660
# if we just grabbed the possibly available ids, then
1661
# we only have an estimate of whats available and need to validate
1662
# that against the revision records.
1663
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1665
InterRepository.register_optimiser(InterWeaveRepo)
1666
InterRepository.register_optimiser(InterKnitRepo)
1669
class RepositoryTestProviderAdapter(object):
1670
"""A tool to generate a suite testing multiple repository formats at once.
1672
This is done by copying the test once for each transport and injecting
1673
the transport_server, transport_readonly_server, and bzrdir_format and
1674
repository_format classes into each copy. Each copy is also given a new id()
1675
to make it easy to identify.
1678
def __init__(self, transport_server, transport_readonly_server, formats):
1679
self._transport_server = transport_server
1680
self._transport_readonly_server = transport_readonly_server
1681
self._formats = formats
1683
def adapt(self, test):
1684
result = TestSuite()
1685
for repository_format, bzrdir_format in self._formats:
1686
new_test = deepcopy(test)
1687
new_test.transport_server = self._transport_server
1688
new_test.transport_readonly_server = self._transport_readonly_server
1689
new_test.bzrdir_format = bzrdir_format
1690
new_test.repository_format = repository_format
1691
def make_new_test_id():
1692
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1693
return lambda: new_id
1694
new_test.id = make_new_test_id()
1695
result.addTest(new_test)
1699
class InterRepositoryTestProviderAdapter(object):
1700
"""A tool to generate a suite testing multiple inter repository formats.
1702
This is done by copying the test once for each interrepo provider and injecting
1703
the transport_server, transport_readonly_server, repository_format and
1704
repository_to_format classes into each copy.
1705
Each copy is also given a new id() to make it easy to identify.
1708
def __init__(self, transport_server, transport_readonly_server, formats):
1709
self._transport_server = transport_server
1710
self._transport_readonly_server = transport_readonly_server
1711
self._formats = formats
1713
def adapt(self, test):
1714
result = TestSuite()
1715
for interrepo_class, repository_format, repository_format_to in self._formats:
1716
new_test = deepcopy(test)
1717
new_test.transport_server = self._transport_server
1718
new_test.transport_readonly_server = self._transport_readonly_server
1719
new_test.interrepo_class = interrepo_class
1720
new_test.repository_format = repository_format
1721
new_test.repository_format_to = repository_format_to
1722
def make_new_test_id():
1723
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1724
return lambda: new_id
1725
new_test.id = make_new_test_id()
1726
result.addTest(new_test)
1730
def default_test_list():
1731
"""Generate the default list of interrepo permutations to test."""
1733
# test the default InterRepository between format 6 and the current
1735
# XXX: robertc 20060220 reinstate this when there are two supported
1736
# formats which do not have an optimal code path between them.
1737
result.append((InterRepository,
1738
RepositoryFormat6(),
1739
RepositoryFormatKnit1()))
1740
for optimiser in InterRepository._optimisers:
1741
result.append((optimiser,
1742
optimiser._matching_repo_format,
1743
optimiser._matching_repo_format
1745
# if there are specific combinations we want to use, we can add them
1750
class CopyConverter(object):
1751
"""A repository conversion tool which just performs a copy of the content.
1753
This is slow but quite reliable.
1756
def __init__(self, target_format):
1757
"""Create a CopyConverter.
1759
:param target_format: The format the resulting repository should be.
1761
self.target_format = target_format
1763
def convert(self, repo, pb):
1764
"""Perform the conversion of to_convert, giving feedback via pb.
1766
:param to_convert: The disk object to convert.
1767
:param pb: a progress bar to use for progress information.
1772
# this is only useful with metadir layouts - separated repo content.
1773
# trigger an assertion if not such
1774
repo._format.get_format_string()
1775
self.repo_dir = repo.bzrdir
1776
self.step('Moving repository to repository.backup')
1777
self.repo_dir.transport.move('repository', 'repository.backup')
1778
backup_transport = self.repo_dir.transport.clone('repository.backup')
1779
self.source_repo = repo._format.open(self.repo_dir,
1781
_override_transport=backup_transport)
1782
self.step('Creating new repository')
1783
converted = self.target_format.initialize(self.repo_dir,
1784
self.source_repo.is_shared())
1785
converted.lock_write()
1787
self.step('Copying content into repository.')
1788
self.source_repo.copy_content_into(converted)
1791
self.step('Deleting old repository content.')
1792
self.repo_dir.transport.delete_tree('repository.backup')
1793
self.pb.note('repository converted')
1795
def step(self, message):
1796
"""Update the pb by a step."""
1798
self.pb.update(message, self.count, self.total)
1801
# Copied from xml.sax.saxutils
1802
def _unescape_xml(data):
1803
"""Unescape &, <, and > in a string of data.
1805
data = data.replace("<", "<")
1806
data = data.replace(">", ">")
1807
# must do ampersand last
1808
return data.replace("&", "&")