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
20
import xml.sax.saxutils
22
import bzrlib.bzrdir as bzrdir
23
from bzrlib.decorators import needs_read_lock, needs_write_lock
24
from bzrlib.errors import InvalidRevisionId
25
from bzrlib.lockable_files import LockableFiles, TransportLock
26
from bzrlib.lockdir import LockDir
27
from bzrlib.osutils import safe_unicode
28
from bzrlib.revision import NULL_REVISION
29
import bzrlib.errors as errors
30
import bzrlib.gpg as gpg
31
from bzrlib.store import copy_all
32
from bzrlib.store.weave import WeaveStore
33
from bzrlib.store.text import TextStore
34
from bzrlib.symbol_versioning import *
35
from bzrlib.trace import mutter
36
from bzrlib.tree import RevisionTree
37
from bzrlib.testament import Testament
38
from bzrlib.tree import EmptyTree
43
class Repository(object):
44
"""Repository holding history for one or more branches.
46
The repository holds and retrieves historical information including
47
revisions and file history. It's normally accessed only by the Branch,
48
which views a particular line of development through that history.
50
The Repository builds on top of Stores and a Transport, which respectively
51
describe the disk data format and the way of accessing the (possibly
56
def add_inventory(self, revid, inv, parents):
57
"""Add the inventory inv to the repository as revid.
59
:param parents: The revision ids of the parents that revid
60
is known to have and are in the repository already.
62
returns the sha1 of the serialized inventory.
64
inv_text = bzrlib.xml5.serializer_v5.write_inventory_to_string(inv)
65
inv_sha1 = bzrlib.osutils.sha_string(inv_text)
66
self.control_weaves.add_text('inventory', revid,
67
bzrlib.osutils.split_lines(inv_text), parents,
68
self.get_transaction())
72
def add_revision(self, rev_id, rev, inv=None, config=None):
73
"""Add rev to the revision store as rev_id.
75
:param rev_id: the revision id to use.
76
:param rev: The revision object.
77
:param inv: The inventory for the revision. if None, it will be looked
78
up in the inventory storer
79
:param config: If None no digital signature will be created.
80
If supplied its signature_needed method will be used
81
to determine if a signature should be made.
83
if config is not None and config.signature_needed():
85
inv = self.get_inventory(rev_id)
86
plaintext = Testament(rev, inv).as_short_text()
87
self.store_revision_signature(
88
gpg.GPGStrategy(config), plaintext, rev_id)
89
if not rev_id in self.get_inventory_weave():
91
raise errors.WeaveRevisionNotPresent(rev_id,
92
self.get_inventory_weave())
94
# yes, this is not suitable for adding with ghosts.
95
self.add_inventory(rev_id, inv, rev.parent_ids)
98
bzrlib.xml5.serializer_v5.write_revision(rev, rev_tmp)
100
self.revision_store.add(rev_tmp, rev_id)
101
mutter('added revision_id {%s}', rev_id)
104
def _all_possible_ids(self):
105
"""Return all the possible revisions that we could find."""
106
return self.get_inventory_weave().names()
109
def all_revision_ids(self):
110
"""Returns a list of all the revision ids in the repository.
112
These are in as much topological order as the underlying store can
113
present: for weaves ghosts may lead to a lack of correctness until
114
the reweave updates the parents list.
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):
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
153
def lock_write(self):
154
self.control_files.lock_write()
157
self.control_files.lock_read()
160
return self.control_files.is_locked()
163
def missing_revision_ids(self, other, revision_id=None):
164
"""Return the revision ids that other has that this does not.
166
These are returned in topological order.
168
revision_id: only return revision ids included by revision_id.
170
return InterRepository.get(other, self).missing_revision_ids(revision_id)
174
"""Open the repository rooted at base.
176
For instance, if the repository is at URL/.bzr/repository,
177
Repository.open(URL) -> a Repository instance.
179
control = bzrlib.bzrdir.BzrDir.open(base)
180
return control.open_repository()
182
def copy_content_into(self, destination, revision_id=None, basis=None):
183
"""Make a complete copy of the content in self into destination.
185
This is a destructive operation! Do not use it on existing
188
return InterRepository.get(self, destination).copy_content(revision_id, basis)
190
def fetch(self, source, revision_id=None, pb=None):
191
"""Fetch the content required to construct revision_id from source.
193
If revision_id is None all content is copied.
195
return InterRepository.get(source, self).fetch(revision_id=revision_id,
199
self.control_files.unlock()
202
def clone(self, a_bzrdir, revision_id=None, basis=None):
203
"""Clone this repository into a_bzrdir using the current format.
205
Currently no check is made that the format of this repository and
206
the bzrdir format are compatible. FIXME RBC 20060201.
208
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
209
# use target default format.
210
result = a_bzrdir.create_repository()
211
# FIXME RBC 20060209 split out the repository type to avoid this check ?
212
elif isinstance(a_bzrdir._format,
213
(bzrlib.bzrdir.BzrDirFormat4,
214
bzrlib.bzrdir.BzrDirFormat5,
215
bzrlib.bzrdir.BzrDirFormat6)):
216
result = a_bzrdir.open_repository()
218
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
219
self.copy_content_into(result, revision_id, basis)
222
def has_revision(self, revision_id):
223
"""True if this branch has a copy of the revision.
225
This does not necessarily imply the revision is merge
226
or on the mainline."""
227
return (revision_id is None
228
or self.revision_store.has_id(revision_id))
231
def get_revision_xml_file(self, revision_id):
232
"""Return XML file object for revision object."""
233
if not revision_id or not isinstance(revision_id, basestring):
234
raise InvalidRevisionId(revision_id=revision_id, branch=self)
236
return self.revision_store.get(revision_id)
237
except (IndexError, KeyError):
238
raise bzrlib.errors.NoSuchRevision(self, revision_id)
241
def get_revision_xml(self, revision_id):
242
return self.get_revision_xml_file(revision_id).read()
245
def get_revision_reconcile(self, revision_id):
246
"""'reconcile' helper routine that allows access to a revision always.
248
This variant of get_revision does not cross check the weave graph
249
against the revision one as get_revision does: but it should only
250
be used by reconcile, or reconcile-alike commands that are correcting
251
or testing the revision graph.
253
xml_file = self.get_revision_xml_file(revision_id)
256
r = bzrlib.xml5.serializer_v5.read_revision(xml_file)
257
except SyntaxError, e:
258
raise bzrlib.errors.BzrError('failed to unpack revision_xml',
262
assert r.revision_id == revision_id
266
def get_revision(self, revision_id):
267
"""Return the Revision object for a named revision"""
268
r = self.get_revision_reconcile(revision_id)
269
# weave corruption can lead to absent revision markers that should be
271
# the following test is reasonably cheap (it needs a single weave read)
272
# and the weave is cached in read transactions. In write transactions
273
# it is not cached but typically we only read a small number of
274
# revisions. For knits when they are introduced we will probably want
275
# to ensure that caching write transactions are in use.
276
inv = self.get_inventory_weave()
277
self._check_revision_parents(r, inv)
280
def _check_revision_parents(self, revision, inventory):
281
"""Private to Repository and Fetch.
283
This checks the parentage of revision in an inventory weave for
284
consistency and is only applicable to inventory-weave-for-ancestry
285
using repository formats & fetchers.
287
weave_parents = inventory.parent_names(revision.revision_id)
288
weave_names = inventory.names()
289
for parent_id in revision.parent_ids:
290
if parent_id in weave_names:
291
# this parent must not be a ghost.
292
if not parent_id in weave_parents:
294
raise errors.CorruptRepository(self)
297
def get_revision_sha1(self, revision_id):
298
"""Hash the stored value of a revision, and return it."""
299
# In the future, revision entries will be signed. At that
300
# point, it is probably best *not* to include the signature
301
# in the revision hash. Because that lets you re-sign
302
# the revision, (add signatures/remove signatures) and still
303
# have all hash pointers stay consistent.
304
# But for now, just hash the contents.
305
return bzrlib.osutils.sha_file(self.get_revision_xml_file(revision_id))
308
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
309
self.revision_store.add(StringIO(gpg_strategy.sign(plaintext)),
312
def fileid_involved_between_revs(self, from_revid, to_revid):
313
"""Find file_id(s) which are involved in the changes between revisions.
315
This determines the set of revisions which are involved, and then
316
finds all file ids affected by those revisions.
318
# TODO: jam 20060119 This code assumes that w.inclusions will
319
# always be correct. But because of the presence of ghosts
320
# it is possible to be wrong.
321
# One specific example from Robert Collins:
322
# Two branches, with revisions ABC, and AD
323
# C is a ghost merge of D.
324
# Inclusions doesn't recognize D as an ancestor.
325
# If D is ever merged in the future, the weave
326
# won't be fixed, because AD never saw revision C
327
# to cause a conflict which would force a reweave.
328
w = self.get_inventory_weave()
329
from_set = set(w.inclusions([w.lookup(from_revid)]))
330
to_set = set(w.inclusions([w.lookup(to_revid)]))
331
included = to_set.difference(from_set)
332
changed = map(w.idx_to_name, included)
333
return self._fileid_involved_by_set(changed)
335
def fileid_involved(self, last_revid=None):
336
"""Find all file_ids modified in the ancestry of last_revid.
338
:param last_revid: If None, last_revision() will be used.
340
w = self.get_inventory_weave()
342
changed = set(w._names)
344
included = w.inclusions([w.lookup(last_revid)])
345
changed = map(w.idx_to_name, included)
346
return self._fileid_involved_by_set(changed)
348
def fileid_involved_by_set(self, changes):
349
"""Find all file_ids modified by the set of revisions passed in.
351
:param changes: A set() of revision ids
353
# TODO: jam 20060119 This line does *nothing*, remove it.
354
# or better yet, change _fileid_involved_by_set so
355
# that it takes the inventory weave, rather than
356
# pulling it out by itself.
357
return self._fileid_involved_by_set(changes)
359
def _fileid_involved_by_set(self, changes):
360
"""Find the set of file-ids affected by the set of revisions.
362
:param changes: A set() of revision ids.
363
:return: A set() of file ids.
365
This peaks at the Weave, interpreting each line, looking to
366
see if it mentions one of the revisions. And if so, includes
367
the file id mentioned.
368
This expects both the Weave format, and the serialization
369
to have a single line per file/directory, and to have
370
fileid="" and revision="" on that line.
372
assert isinstance(self._format, (RepositoryFormat5,
375
RepositoryFormatKnit1)), \
376
"fileid_involved only supported for branches which store inventory as unnested xml"
378
w = self.get_inventory_weave()
380
for line in w._weave:
382
# it is ugly, but it is due to the weave structure
383
if not isinstance(line, basestring): continue
385
start = line.find('file_id="')+9
386
if start < 9: continue
387
end = line.find('"', start)
389
file_id = xml.sax.saxutils.unescape(line[start:end])
391
# check if file_id is already present
392
if file_id in file_ids: continue
394
start = line.find('revision="')+10
395
if start < 10: continue
396
end = line.find('"', start)
398
revision_id = xml.sax.saxutils.unescape(line[start:end])
400
if revision_id in changes:
401
file_ids.add(file_id)
405
def get_inventory_weave(self):
406
return self.control_weaves.get_weave('inventory',
407
self.get_transaction())
410
def get_inventory(self, revision_id):
411
"""Get Inventory object by hash."""
412
xml = self.get_inventory_xml(revision_id)
413
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
416
def get_inventory_xml(self, revision_id):
417
"""Get inventory XML as a file object."""
419
assert isinstance(revision_id, basestring), type(revision_id)
420
iw = self.get_inventory_weave()
421
return iw.get_text(iw.lookup(revision_id))
423
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
426
def get_inventory_sha1(self, revision_id):
427
"""Return the sha1 hash of the inventory entry
429
return self.get_revision(revision_id).inventory_sha1
432
def get_revision_inventory(self, revision_id):
433
"""Return inventory of a past revision."""
434
# TODO: Unify this with get_inventory()
435
# bzr 0.0.6 and later imposes the constraint that the inventory_id
436
# must be the same as its revision, so this is trivial.
437
if revision_id is None:
438
# This does not make sense: if there is no revision,
439
# then it is the current tree inventory surely ?!
440
# and thus get_root_id() is something that looks at the last
441
# commit on the branch, and the get_root_id is an inventory check.
442
raise NotImplementedError
443
# return Inventory(self.get_root_id())
445
return self.get_inventory(revision_id)
449
"""Return True if this repository is flagged as a shared repository."""
450
# FIXME format 4-6 cannot be shared, this is technically faulty.
451
return self.control_files._transport.has('shared-storage')
454
def revision_tree(self, revision_id):
455
"""Return Tree for a revision on this branch.
457
`revision_id` may be None for the null revision, in which case
458
an `EmptyTree` is returned."""
459
# TODO: refactor this to use an existing revision object
460
# so we don't need to read it in twice.
461
if revision_id is None or revision_id == NULL_REVISION:
464
inv = self.get_revision_inventory(revision_id)
465
return RevisionTree(self, inv, revision_id)
468
def get_ancestry(self, revision_id):
469
"""Return a list of revision-ids integrated by a revision.
471
This is topologically sorted.
473
if revision_id is None:
475
if not self.has_revision(revision_id):
476
raise errors.NoSuchRevision(self, revision_id)
477
w = self.get_inventory_weave()
478
return [None] + map(w.idx_to_name,
479
w.inclusions([w.lookup(revision_id)]))
482
def print_file(self, file, revision_id):
483
"""Print `file` to stdout.
485
FIXME RBC 20060125 as John Meinel points out this is a bad api
486
- it writes to stdout, it assumes that that is valid etc. Fix
487
by creating a new more flexible convenience function.
489
tree = self.revision_tree(revision_id)
490
# use inventory as it was in that revision
491
file_id = tree.inventory.path2id(file)
493
raise BzrError("%r is not present in revision %s" % (file, revno))
495
revno = self.revision_id_to_revno(revision_id)
496
except errors.NoSuchRevision:
497
# TODO: This should not be BzrError,
498
# but NoSuchFile doesn't fit either
499
raise BzrError('%r is not present in revision %s'
500
% (file, revision_id))
502
raise BzrError('%r is not present in revision %s'
504
tree.print_file(file_id)
506
def get_transaction(self):
507
return self.control_files.get_transaction()
510
def set_make_working_trees(self, new_value):
511
"""Set the policy flag for making working trees when creating branches.
513
This only applies to branches that use this repository.
515
The default is 'True'.
516
:param new_value: True to restore the default, False to disable making
519
# FIXME: split out into a new class/strategy ?
520
if isinstance(self._format, (RepositoryFormat4,
523
raise NotImplementedError(self.set_make_working_trees)
526
self.control_files._transport.delete('no-working-trees')
527
except errors.NoSuchFile:
530
self.control_files.put_utf8('no-working-trees', '')
532
def make_working_trees(self):
533
"""Returns the policy for making working trees on new branches."""
534
# FIXME: split out into a new class/strategy ?
535
if isinstance(self._format, (RepositoryFormat4,
539
return not self.control_files._transport.has('no-working-trees')
542
def sign_revision(self, revision_id, gpg_strategy):
543
plaintext = Testament.from_revision(self, revision_id).as_short_text()
544
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
547
class AllInOneRepository(Repository):
548
"""Legacy support - the repository behaviour for all-in-one branches."""
550
def __init__(self, _format, a_bzrdir, revision_store):
551
# we reuse one control files instance.
552
dir_mode = a_bzrdir._control_files._dir_mode
553
file_mode = a_bzrdir._control_files._file_mode
555
def get_weave(name, prefixed=False):
557
name = safe_unicode(name)
560
relpath = a_bzrdir._control_files._escape(name)
561
weave_transport = a_bzrdir._control_files._transport.clone(relpath)
562
ws = WeaveStore(weave_transport, prefixed=prefixed,
565
if a_bzrdir._control_files._transport.should_cache():
566
ws.enable_cache = True
569
def get_store(name, compressed=True, prefixed=False):
570
# FIXME: This approach of assuming stores are all entirely compressed
571
# or entirely uncompressed is tidy, but breaks upgrade from
572
# some existing branches where there's a mixture; we probably
573
# still want the option to look for both.
574
relpath = a_bzrdir._control_files._escape(name)
575
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
576
prefixed=prefixed, compressed=compressed,
579
#if self._transport.should_cache():
580
# cache_path = os.path.join(self.cache_root, name)
581
# os.mkdir(cache_path)
582
# store = bzrlib.store.CachedStore(store, cache_path)
585
# not broken out yet because the controlweaves|inventory_store
586
# and text_store | weave_store bits are still different.
587
if isinstance(_format, RepositoryFormat4):
588
self.inventory_store = get_store('inventory-store')
589
self.text_store = get_store('text-store')
590
elif isinstance(_format, RepositoryFormat5):
591
self.control_weaves = get_weave('')
592
self.weave_store = get_weave('weaves')
593
elif isinstance(_format, RepositoryFormat6):
594
self.control_weaves = get_weave('')
595
self.weave_store = get_weave('weaves', prefixed=True)
597
raise errors.BzrError('unreachable code: unexpected repository'
599
revision_store.register_suffix('sig')
600
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, revision_store)
603
class MetaDirRepository(Repository):
604
"""Repositories in the new meta-dir layout."""
606
def __init__(self, _format, a_bzrdir, control_files, revision_store):
607
super(MetaDirRepository, self).__init__(_format,
612
dir_mode = self.control_files._dir_mode
613
file_mode = self.control_files._file_mode
615
def get_weave(name, prefixed=False):
617
name = safe_unicode(name)
620
relpath = self.control_files._escape(name)
621
weave_transport = self.control_files._transport.clone(relpath)
622
ws = WeaveStore(weave_transport, prefixed=prefixed,
625
if self.control_files._transport.should_cache():
626
ws.enable_cache = True
629
if isinstance(self._format, RepositoryFormat7):
630
self.control_weaves = get_weave('')
631
self.weave_store = get_weave('weaves', prefixed=True)
632
elif isinstance(self._format, RepositoryFormatKnit1):
633
self.control_weaves = get_weave('')
634
self.weave_store = get_weave('knits', prefixed=True)
636
raise errors.BzrError('unreachable code: unexpected repository'
640
class RepositoryFormat(object):
641
"""A repository format.
643
Formats provide three things:
644
* An initialization routine to construct repository data on disk.
645
* a format string which is used when the BzrDir supports versioned
647
* an open routine which returns a Repository instance.
649
Formats are placed in an dict by their format string for reference
650
during opening. These should be subclasses of RepositoryFormat
653
Once a format is deprecated, just deprecate the initialize and open
654
methods on the format class. Do not deprecate the object, as the
655
object will be created every system load.
657
Common instance attributes:
658
_matchingbzrdir - the bzrdir format that the repository format was
659
originally written to work with. This can be used if manually
660
constructing a bzrdir and repository, or more commonly for test suite
664
_default_format = None
665
"""The default format used for new repositories."""
668
"""The known formats."""
671
def find_format(klass, a_bzrdir):
672
"""Return the format for the repository object in a_bzrdir."""
674
transport = a_bzrdir.get_repository_transport(None)
675
format_string = transport.get("format").read()
676
return klass._formats[format_string]
677
except errors.NoSuchFile:
678
raise errors.NoRepositoryPresent(a_bzrdir)
680
raise errors.UnknownFormatError(format_string)
683
def get_default_format(klass):
684
"""Return the current default format."""
685
return klass._default_format
687
def get_format_string(self):
688
"""Return the ASCII format string that identifies this format.
690
Note that in pre format ?? repositories the format string is
691
not permitted nor written to disk.
693
raise NotImplementedError(self.get_format_string)
695
def _get_revision_store(self, repo_transport, control_files):
696
"""Return the revision store object for this a_bzrdir."""
697
raise NotImplementedError(self._get_revision_store)
699
def _get_rev_store(self,
705
"""Common logic for getting a revision store for a repository.
707
see self._get_revision_store for the method to
708
get the store for a repository.
711
name = safe_unicode(name)
714
dir_mode = control_files._dir_mode
715
file_mode = control_files._file_mode
716
revision_store =TextStore(transport.clone(name),
718
compressed=compressed,
721
revision_store.register_suffix('sig')
722
return revision_store
724
def initialize(self, a_bzrdir, shared=False):
725
"""Initialize a repository of this format in a_bzrdir.
727
:param a_bzrdir: The bzrdir to put the new repository in it.
728
:param shared: The repository should be initialized as a sharable one.
730
This may raise UninitializableFormat if shared repository are not
731
compatible the a_bzrdir.
734
def is_supported(self):
735
"""Is this format supported?
737
Supported formats must be initializable and openable.
738
Unsupported formats may not support initialization or committing or
739
some other features depending on the reason for not being supported.
743
def open(self, a_bzrdir, _found=False):
744
"""Return an instance of this format for the bzrdir a_bzrdir.
746
_found is a private parameter, do not use it.
748
raise NotImplementedError(self.open)
751
def register_format(klass, format):
752
klass._formats[format.get_format_string()] = format
755
def set_default_format(klass, format):
756
klass._default_format = format
759
def unregister_format(klass, format):
760
assert klass._formats[format.get_format_string()] is format
761
del klass._formats[format.get_format_string()]
764
class PreSplitOutRepositoryFormat(RepositoryFormat):
765
"""Base class for the pre split out repository formats."""
767
def initialize(self, a_bzrdir, shared=False, _internal=False):
768
"""Create a weave repository.
770
TODO: when creating split out bzr branch formats, move this to a common
771
base for Format5, Format6. or something like that.
773
from bzrlib.weavefile import write_weave_v5
774
from bzrlib.weave import Weave
777
raise errors.IncompatibleFormat(self, a_bzrdir._format)
780
# always initialized when the bzrdir is.
781
return self.open(a_bzrdir, _found=True)
783
# Create an empty weave
785
bzrlib.weavefile.write_weave_v5(Weave(), sio)
786
empty_weave = sio.getvalue()
788
mutter('creating repository in %s.', a_bzrdir.transport.base)
789
dirs = ['revision-store', 'weaves']
790
files = [('inventory.weave', StringIO(empty_weave)),
793
# FIXME: RBC 20060125 dont peek under the covers
794
# NB: no need to escape relative paths that are url safe.
795
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
797
control_files.create_lock()
798
control_files.lock_write()
799
control_files._transport.mkdir_multi(dirs,
800
mode=control_files._dir_mode)
802
for file, content in files:
803
control_files.put(file, content)
805
control_files.unlock()
806
return self.open(a_bzrdir, _found=True)
808
def open(self, a_bzrdir, _found=False):
809
"""See RepositoryFormat.open()."""
811
# we are being called directly and must probe.
812
raise NotImplementedError
814
repo_transport = a_bzrdir.get_repository_transport(None)
815
control_files = a_bzrdir._control_files
816
revision_store = self._get_revision_store(repo_transport, control_files)
817
return AllInOneRepository(_format=self,
819
revision_store=revision_store)
822
class RepositoryFormat4(PreSplitOutRepositoryFormat):
823
"""Bzr repository format 4.
825
This repository format has:
827
- TextStores for texts, inventories,revisions.
829
This format is deprecated: it indexes texts using a text id which is
830
removed in format 5; initializationa and write support for this format
835
super(RepositoryFormat4, self).__init__()
836
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
838
def initialize(self, url, shared=False, _internal=False):
839
"""Format 4 branches cannot be created."""
840
raise errors.UninitializableFormat(self)
842
def is_supported(self):
843
"""Format 4 is not supported.
845
It is not supported because the model changed from 4 to 5 and the
846
conversion logic is expensive - so doing it on the fly was not
851
def _get_revision_store(self, repo_transport, control_files):
852
"""See RepositoryFormat._get_revision_store()."""
853
return self._get_rev_store(repo_transport,
858
class RepositoryFormat5(PreSplitOutRepositoryFormat):
859
"""Bzr control format 5.
861
This repository format has:
862
- weaves for file texts and inventory
864
- TextStores for revisions and signatures.
868
super(RepositoryFormat5, self).__init__()
869
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
871
def _get_revision_store(self, repo_transport, control_files):
872
"""See RepositoryFormat._get_revision_store()."""
873
"""Return the revision store object for this a_bzrdir."""
874
return self._get_rev_store(repo_transport,
880
class RepositoryFormat6(PreSplitOutRepositoryFormat):
881
"""Bzr control format 6.
883
This repository format has:
884
- weaves for file texts and inventory
885
- hash subdirectory based stores.
886
- TextStores for revisions and signatures.
890
super(RepositoryFormat6, self).__init__()
891
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
893
def _get_revision_store(self, repo_transport, control_files):
894
"""See RepositoryFormat._get_revision_store()."""
895
return self._get_rev_store(repo_transport,
902
class MetaDirRepositoryFormat(RepositoryFormat):
903
"""Common base class for the new repositories using the metadir layour."""
906
super(MetaDirRepositoryFormat, self).__init__()
907
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
909
def _create_control_files(self, a_bzrdir):
910
"""Create the required files and the initial control_files object."""
911
# FIXME: RBC 20060125 dont peek under the covers
912
# NB: no need to escape relative paths that are url safe.
913
repository_transport = a_bzrdir.get_repository_transport(self)
914
control_files = LockableFiles(repository_transport, 'lock', LockDir)
915
control_files.create_lock()
918
def _get_revision_store(self, repo_transport, control_files):
919
"""See RepositoryFormat._get_revision_store()."""
920
return self._get_rev_store(repo_transport,
927
def open(self, a_bzrdir, _found=False, _override_transport=None):
928
"""See RepositoryFormat.open().
930
:param _override_transport: INTERNAL USE ONLY. Allows opening the
931
repository at a slightly different url
932
than normal. I.e. during 'upgrade'.
935
format = RepositoryFormat.find_format(a_bzrdir)
936
assert format.__class__ == self.__class__
937
if _override_transport is not None:
938
repo_transport = _override_transport
940
repo_transport = a_bzrdir.get_repository_transport(None)
941
control_files = LockableFiles(repo_transport, 'lock', LockDir)
942
revision_store = self._get_revision_store(repo_transport, control_files)
943
return MetaDirRepository(_format=self,
945
control_files=control_files,
946
revision_store=revision_store)
948
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
949
"""Upload the initial blank content."""
950
control_files = self._create_control_files(a_bzrdir)
951
control_files.lock_write()
953
control_files._transport.mkdir_multi(dirs,
954
mode=control_files._dir_mode)
955
for file, content in files:
956
control_files.put(file, content)
957
for file, content in utf8_files:
958
control_files.put_utf8(file, content)
960
control_files.put_utf8('shared-storage', '')
962
control_files.unlock()
965
class RepositoryFormat7(MetaDirRepositoryFormat):
968
This repository format has:
969
- weaves for file texts and inventory
970
- hash subdirectory based stores.
971
- TextStores for revisions and signatures.
972
- a format marker of its own
973
- an optional 'shared-storage' flag
974
- an optional 'no-working-trees' flag
977
def get_format_string(self):
978
"""See RepositoryFormat.get_format_string()."""
979
return "Bazaar-NG Repository format 7"
981
def initialize(self, a_bzrdir, shared=False):
982
"""Create a weave repository.
984
:param shared: If true the repository will be initialized as a shared
987
from bzrlib.weavefile import write_weave_v5
988
from bzrlib.weave import Weave
990
# Create an empty weave
992
bzrlib.weavefile.write_weave_v5(Weave(), sio)
993
empty_weave = sio.getvalue()
995
mutter('creating repository in %s.', a_bzrdir.transport.base)
996
dirs = ['revision-store', 'weaves']
997
files = [('inventory.weave', StringIO(empty_weave)),
999
utf8_files = [('format', self.get_format_string())]
1001
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1002
return self.open(a_bzrdir=a_bzrdir, _found=True)
1005
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1006
"""Bzr repository knit format 1.
1008
This repository format has:
1009
- knits for file texts and inventory
1010
- hash subdirectory based stores.
1011
- knits for revisions and signatures
1012
- TextStores for revisions and signatures.
1013
- a format marker of its own
1014
- an optional 'shared-storage' flag
1015
- an optional 'no-working-trees' flag
1019
def get_format_string(self):
1020
"""See RepositoryFormat.get_format_string()."""
1021
return "Bazaar-NG Knit Repository Format 1"
1023
def initialize(self, a_bzrdir, shared=False):
1024
"""Create a knit format 1 repository.
1026
:param shared: If true the repository will be initialized as a shared
1028
XXX NOTE that this current uses a Weave for testing and will become
1029
A Knit in due course.
1031
from bzrlib.weavefile import write_weave_v5
1032
from bzrlib.weave import Weave
1034
# Create an empty weave
1036
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1037
empty_weave = sio.getvalue()
1039
mutter('creating repository in %s.', a_bzrdir.transport.base)
1040
dirs = ['revision-store', 'knits']
1041
files = [('inventory.weave', StringIO(empty_weave)),
1043
utf8_files = [('format', self.get_format_string())]
1045
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1046
return self.open(a_bzrdir=a_bzrdir, _found=True)
1049
# formats which have no format string are not discoverable
1050
# and not independently creatable, so are not registered.
1051
_default_format = RepositoryFormat7()
1052
RepositoryFormat.register_format(_default_format)
1053
RepositoryFormat.register_format(RepositoryFormatKnit1())
1054
RepositoryFormat.set_default_format(_default_format)
1055
_legacy_formats = [RepositoryFormat4(),
1056
RepositoryFormat5(),
1057
RepositoryFormat6()]
1060
class InterRepository(object):
1061
"""This class represents operations taking place between two repositories.
1063
Its instances have methods like copy_content and fetch, and contain
1064
references to the source and target repositories these operations can be
1067
Often we will provide convenience methods on 'repository' which carry out
1068
operations with another repository - they will always forward to
1069
InterRepository.get(other).method_name(parameters).
1071
# XXX: FIXME: FUTURE: robertc
1072
# testing of these probably requires a factory in optimiser type, and
1073
# then a test adapter to test each type thoroughly.
1077
"""The available optimised InterRepository types."""
1079
def __init__(self, source, target):
1080
"""Construct a default InterRepository instance. Please use 'get'.
1082
Only subclasses of InterRepository should call
1083
InterRepository.__init__ - clients should call InterRepository.get
1084
instead which will create an optimised InterRepository if possible.
1086
self.source = source
1087
self.target = target
1090
def copy_content(self, revision_id=None, basis=None):
1091
"""Make a complete copy of the content in self into destination.
1093
This is a destructive operation! Do not use it on existing
1096
:param revision_id: Only copy the content needed to construct
1097
revision_id and its parents.
1098
:param basis: Copy the needed data preferentially from basis.
1101
self.target.set_make_working_trees(self.source.make_working_trees())
1102
except NotImplementedError:
1104
# grab the basis available data
1105
if basis is not None:
1106
self.target.fetch(basis, revision_id=revision_id)
1107
# but dont both fetching if we have the needed data now.
1108
if (revision_id not in (None, NULL_REVISION) and
1109
self.target.has_revision(revision_id)):
1111
self.target.fetch(self.source, revision_id=revision_id)
1113
def _double_lock(self, lock_source, lock_target):
1114
"""Take out too locks, rolling back the first if the second throws."""
1119
# we want to ensure that we don't leave source locked by mistake.
1120
# and any error on target should not confuse source.
1121
self.source.unlock()
1125
def fetch(self, revision_id=None, pb=None):
1126
"""Fetch the content required to construct revision_id.
1128
The content is copied from source to target.
1130
:param revision_id: if None all content is copied, if NULL_REVISION no
1132
:param pb: optional progress bar to use for progress reports. If not
1133
provided a default one will be created.
1135
Returns the copied revision count and the failed revisions in a tuple:
1138
from bzrlib.fetch import RepoFetcher
1139
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1140
self.source, self.source._format, self.target, self.target._format)
1141
f = RepoFetcher(to_repository=self.target,
1142
from_repository=self.source,
1143
last_revision=revision_id,
1145
return f.count_copied, f.failed_revisions
1148
def get(klass, repository_source, repository_target):
1149
"""Retrieve a InterRepository worker object for these repositories.
1151
:param repository_source: the repository to be the 'source' member of
1152
the InterRepository instance.
1153
:param repository_target: the repository to be the 'target' member of
1154
the InterRepository instance.
1155
If an optimised InterRepository worker exists it will be used otherwise
1156
a default InterRepository instance will be created.
1158
for provider in klass._optimisers:
1159
if provider.is_compatible(repository_source, repository_target):
1160
return provider(repository_source, repository_target)
1161
return InterRepository(repository_source, repository_target)
1163
def lock_read(self):
1164
"""Take out a logical read lock.
1166
This will lock the source branch and the target branch. The source gets
1167
a read lock and the target a read lock.
1169
self._double_lock(self.source.lock_read, self.target.lock_read)
1171
def lock_write(self):
1172
"""Take out a logical write lock.
1174
This will lock the source branch and the target branch. The source gets
1175
a read lock and the target a write lock.
1177
self._double_lock(self.source.lock_read, self.target.lock_write)
1180
def missing_revision_ids(self, revision_id=None):
1181
"""Return the revision ids that source has that target does not.
1183
These are returned in topological order.
1185
:param revision_id: only return revision ids included by this
1188
# generic, possibly worst case, slow code path.
1189
target_ids = set(self.target.all_revision_ids())
1190
if revision_id is not None:
1191
source_ids = self.source.get_ancestry(revision_id)
1192
assert source_ids.pop(0) == None
1194
source_ids = self.source.all_revision_ids()
1195
result_set = set(source_ids).difference(target_ids)
1196
# this may look like a no-op: its not. It preserves the ordering
1197
# other_ids had while only returning the members from other_ids
1198
# that we've decided we need.
1199
return [rev_id for rev_id in source_ids if rev_id in result_set]
1202
def register_optimiser(klass, optimiser):
1203
"""Register an InterRepository optimiser."""
1204
klass._optimisers.add(optimiser)
1207
"""Release the locks on source and target."""
1209
self.target.unlock()
1211
self.source.unlock()
1214
def unregister_optimiser(klass, optimiser):
1215
"""Unregister an InterRepository optimiser."""
1216
klass._optimisers.remove(optimiser)
1219
class InterWeaveRepo(InterRepository):
1220
"""Optimised code paths between Weave based repositories."""
1222
_matching_repo_format = _default_format
1223
"""Repository format for testing with."""
1226
def is_compatible(source, target):
1227
"""Be compatible with known Weave formats.
1229
We dont test for the stores being of specific types becase that
1230
could lead to confusing results, and there is no need to be
1234
return (isinstance(source._format, (RepositoryFormat5,
1236
RepositoryFormat7)) and
1237
isinstance(target._format, (RepositoryFormat5,
1239
RepositoryFormat7)))
1240
except AttributeError:
1244
def copy_content(self, revision_id=None, basis=None):
1245
"""See InterRepository.copy_content()."""
1246
# weave specific optimised path:
1247
if basis is not None:
1248
# copy the basis in, then fetch remaining data.
1249
basis.copy_content_into(self.target, revision_id)
1250
# the basis copy_content_into could misset this.
1252
self.target.set_make_working_trees(self.source.make_working_trees())
1253
except NotImplementedError:
1255
self.target.fetch(self.source, revision_id=revision_id)
1258
self.target.set_make_working_trees(self.source.make_working_trees())
1259
except NotImplementedError:
1261
# FIXME do not peek!
1262
if self.source.control_files._transport.listable():
1263
pb = bzrlib.ui.ui_factory.progress_bar()
1264
copy_all(self.source.weave_store,
1265
self.target.weave_store, pb=pb)
1266
pb.update('copying inventory', 0, 1)
1267
self.target.control_weaves.copy_multi(
1268
self.source.control_weaves, ['inventory'])
1269
copy_all(self.source.revision_store,
1270
self.target.revision_store, pb=pb)
1272
self.target.fetch(self.source, revision_id=revision_id)
1275
def fetch(self, revision_id=None, pb=None):
1276
"""See InterRepository.fetch()."""
1277
from bzrlib.fetch import RepoFetcher
1278
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1279
self.source, self.source._format, self.target, self.target._format)
1280
f = RepoFetcher(to_repository=self.target,
1281
from_repository=self.source,
1282
last_revision=revision_id,
1284
return f.count_copied, f.failed_revisions
1287
def missing_revision_ids(self, revision_id=None):
1288
"""See InterRepository.missing_revision_ids()."""
1289
# we want all revisions to satisfy revision_id in source.
1290
# but we dont want to stat every file here and there.
1291
# we want then, all revisions other needs to satisfy revision_id
1292
# checked, but not those that we have locally.
1293
# so the first thing is to get a subset of the revisions to
1294
# satisfy revision_id in source, and then eliminate those that
1295
# we do already have.
1296
# this is slow on high latency connection to self, but as as this
1297
# disk format scales terribly for push anyway due to rewriting
1298
# inventory.weave, this is considered acceptable.
1300
if revision_id is not None:
1301
source_ids = self.source.get_ancestry(revision_id)
1302
assert source_ids.pop(0) == None
1304
source_ids = self.source._all_possible_ids()
1305
source_ids_set = set(source_ids)
1306
# source_ids is the worst possible case we may need to pull.
1307
# now we want to filter source_ids against what we actually
1308
# have in target, but dont try to check for existence where we know
1309
# we do not have a revision as that would be pointless.
1310
target_ids = set(self.target._all_possible_ids())
1311
possibly_present_revisions = target_ids.intersection(source_ids_set)
1312
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1313
required_revisions = source_ids_set.difference(actually_present_revisions)
1314
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1315
if revision_id is not None:
1316
# we used get_ancestry to determine source_ids then we are assured all
1317
# revisions referenced are present as they are installed in topological order.
1318
# and the tip revision was validated by get_ancestry.
1319
return required_topo_revisions
1321
# if we just grabbed the possibly available ids, then
1322
# we only have an estimate of whats available and need to validate
1323
# that against the revision records.
1324
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1327
InterRepository.register_optimiser(InterWeaveRepo)
1330
class RepositoryTestProviderAdapter(object):
1331
"""A tool to generate a suite testing multiple repository formats at once.
1333
This is done by copying the test once for each transport and injecting
1334
the transport_server, transport_readonly_server, and bzrdir_format and
1335
repository_format classes into each copy. Each copy is also given a new id()
1336
to make it easy to identify.
1339
def __init__(self, transport_server, transport_readonly_server, formats):
1340
self._transport_server = transport_server
1341
self._transport_readonly_server = transport_readonly_server
1342
self._formats = formats
1344
def adapt(self, test):
1345
result = TestSuite()
1346
for repository_format, bzrdir_format in self._formats:
1347
new_test = deepcopy(test)
1348
new_test.transport_server = self._transport_server
1349
new_test.transport_readonly_server = self._transport_readonly_server
1350
new_test.bzrdir_format = bzrdir_format
1351
new_test.repository_format = repository_format
1352
def make_new_test_id():
1353
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1354
return lambda: new_id
1355
new_test.id = make_new_test_id()
1356
result.addTest(new_test)
1360
class InterRepositoryTestProviderAdapter(object):
1361
"""A tool to generate a suite testing multiple inter repository formats.
1363
This is done by copying the test once for each interrepo provider and injecting
1364
the transport_server, transport_readonly_server, repository_format and
1365
repository_to_format classes into each copy.
1366
Each copy is also given a new id() to make it easy to identify.
1369
def __init__(self, transport_server, transport_readonly_server, formats):
1370
self._transport_server = transport_server
1371
self._transport_readonly_server = transport_readonly_server
1372
self._formats = formats
1374
def adapt(self, test):
1375
result = TestSuite()
1376
for interrepo_class, repository_format, repository_format_to in self._formats:
1377
new_test = deepcopy(test)
1378
new_test.transport_server = self._transport_server
1379
new_test.transport_readonly_server = self._transport_readonly_server
1380
new_test.interrepo_class = interrepo_class
1381
new_test.repository_format = repository_format
1382
new_test.repository_format_to = repository_format_to
1383
def make_new_test_id():
1384
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1385
return lambda: new_id
1386
new_test.id = make_new_test_id()
1387
result.addTest(new_test)
1391
def default_test_list():
1392
"""Generate the default list of interrepo permutations to test."""
1394
# test the default InterRepository between format 6 and the current
1396
# XXX: robertc 20060220 reinstate this when there are two supported
1397
# formats which do not have an optimal code path between them.
1398
result.append((InterRepository,
1399
RepositoryFormat6(),
1400
RepositoryFormatKnit1()))
1401
for optimiser in InterRepository._optimisers:
1402
result.append((optimiser,
1403
optimiser._matching_repo_format,
1404
optimiser._matching_repo_format
1406
# if there are specific combinations we want to use, we can add them
1411
class CopyConverter(object):
1412
"""A repository conversion tool which just performs a copy of the content.
1414
This is slow but quite reliable.
1417
def __init__(self, target_format):
1418
"""Create a CopyConverter.
1420
:param target_format: The format the resulting repository should be.
1422
self.target_format = target_format
1424
def convert(self, repo, pb):
1425
"""Perform the conversion of to_convert, giving feedback via pb.
1427
:param to_convert: The disk object to convert.
1428
:param pb: a progress bar to use for progress information.
1433
# this is only useful with metadir layouts - separated repo content.
1434
# trigger an assertion if not such
1435
repo._format.get_format_string()
1436
self.repo_dir = repo.bzrdir
1437
self.step('Moving repository to repository.backup')
1438
self.repo_dir.transport.move('repository', 'repository.backup')
1439
backup_transport = self.repo_dir.transport.clone('repository.backup')
1440
self.source_repo = repo._format.open(self.repo_dir,
1442
_override_transport=backup_transport)
1443
self.step('Creating new repository')
1444
converted = self.target_format.initialize(self.repo_dir,
1445
self.source_repo.is_shared())
1446
converted.lock_write()
1448
self.step('Copying content into repository.')
1449
self.source_repo.copy_content_into(converted)
1452
self.step('Deleting old repository content.')
1453
self.repo_dir.transport.delete_tree('repository.backup')
1454
self.pb.note('repository converted')
1456
def step(self, message):
1457
"""Update the pb by a step."""
1459
self.pb.update(message, self.count, self.total)