1
# Copyright (C) 2005 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
23
from bzrlib.decorators import needs_read_lock, needs_write_lock
24
import bzrlib.errors as errors
25
from bzrlib.errors import InvalidRevisionId
26
import bzrlib.gpg as gpg
27
from bzrlib.lockable_files import LockableFiles
28
from bzrlib.osutils import safe_unicode
29
from bzrlib.revision import NULL_REVISION
30
from bzrlib.store import copy_all
31
from bzrlib.store.weave import WeaveStore
32
from bzrlib.store.text import TextStore
33
from bzrlib.symbol_versioning import *
34
from bzrlib.trace import mutter
35
from bzrlib.tree import RevisionTree
36
from bzrlib.testament import Testament
37
from bzrlib.tree import EmptyTree
42
class Repository(object):
43
"""Repository holding history for one or more branches.
45
The repository holds and retrieves historical information including
46
revisions and file history. It's normally accessed only by the Branch,
47
which views a particular line of development through that history.
49
The Repository builds on top of Stores and a Transport, which respectively
50
describe the disk data format and the way of accessing the (possibly
55
def add_inventory(self, revid, inv, parents):
56
"""Add the inventory inv to the repository as revid.
58
:param parents: The revision ids of the parents that revid
59
is known to have and are in the repository already.
61
returns the sha1 of the serialized inventory.
63
inv_text = bzrlib.xml5.serializer_v5.write_inventory_to_string(inv)
64
inv_sha1 = bzrlib.osutils.sha_string(inv_text)
65
self.control_weaves.add_text('inventory', revid,
66
bzrlib.osutils.split_lines(inv_text), parents,
67
self.get_transaction())
71
def add_revision(self, rev_id, rev, inv=None, config=None):
72
"""Add rev to the revision store as rev_id.
74
:param rev_id: the revision id to use.
75
:param rev: The revision object.
76
:param inv: The inventory for the revision. if None, it will be looked
77
up in the inventory storer
78
:param config: If None no digital signature will be created.
79
If supplied its signature_needed method will be used
80
to determine if a signature should be made.
82
if config is not None and config.signature_needed():
84
inv = self.get_inventory(rev_id)
85
plaintext = Testament(rev, inv).as_short_text()
86
self.store_revision_signature(
87
gpg.GPGStrategy(config), plaintext, rev_id)
88
if not rev_id in self.get_inventory_weave():
90
raise errors.WeaveRevisionNotPresent(rev_id,
91
self.get_inventory_weave())
93
# yes, this is not suitable for adding with ghosts.
94
self.add_inventory(rev_id, inv, rev.parent_ids)
97
bzrlib.xml5.serializer_v5.write_revision(rev, rev_tmp)
99
self.revision_store.add(rev_tmp, rev_id)
100
mutter('added revision_id {%s}', rev_id)
103
def _all_possible_ids(self):
104
"""Return all the possible revisions that we could find."""
105
return self.get_inventory_weave().names()
108
def all_revision_ids(self):
109
"""Returns a list of all the revision ids in the repository.
111
These are in as much topological order as the underlying store can
112
present: for weaves ghosts may lead to a lack of correctness until
113
the reweave updates the parents list.
115
result = self._all_possible_ids()
116
return self._eliminate_revisions_not_present(result)
119
def _eliminate_revisions_not_present(self, revision_ids):
120
"""Check every revision id in revision_ids to see if we have it.
122
Returns a set of the present revisions.
125
for id in revision_ids:
126
if self.has_revision(id):
131
def create(a_bzrdir):
132
"""Construct the current default format repository in a_bzrdir."""
133
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
135
def __init__(self, _format, a_bzrdir, control_files, revision_store):
136
"""instantiate a Repository.
138
:param _format: The format of the repository on disk.
139
:param a_bzrdir: The BzrDir of the repository.
141
In the future we will have a single api for all stores for
142
getting file texts, inventories and revisions, then
143
this construct will accept instances of those things.
145
object.__init__(self)
146
self._format = _format
147
# the following are part of the public API for Repository:
148
self.bzrdir = a_bzrdir
149
self.control_files = control_files
150
self.revision_store = revision_store
152
def lock_write(self):
153
self.control_files.lock_write()
156
self.control_files.lock_read()
159
def missing_revision_ids(self, other, revision_id=None):
160
"""Return the revision ids that other has that this does not.
162
These are returned in topological order.
164
revision_id: only return revision ids included by revision_id.
166
return InterRepository.get(other, self).missing_revision_ids(revision_id)
170
"""Open the repository rooted at base.
172
For instance, if the repository is at URL/.bzr/repository,
173
Repository.open(URL) -> a Repository instance.
175
control = bzrlib.bzrdir.BzrDir.open(base)
176
return control.open_repository()
178
def copy_content_into(self, destination, revision_id=None, basis=None):
179
"""Make a complete copy of the content in self into destination.
181
This is a destructive operation! Do not use it on existing
184
return InterRepository.get(self, destination).copy_content(revision_id, basis)
186
def fetch(self, source, revision_id=None, pb=None):
187
"""Fetch the content required to construct revision_id from source.
189
If revision_id is None all content is copied.
191
return InterRepository.get(source, self).fetch(revision_id=revision_id,
195
self.control_files.unlock()
198
def clone(self, a_bzrdir, revision_id=None, basis=None):
199
"""Clone this repository into a_bzrdir using the current format.
201
Currently no check is made that the format of this repository and
202
the bzrdir format are compatible. FIXME RBC 20060201.
204
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
205
# use target default format.
206
result = a_bzrdir.create_repository()
207
# FIXME RBC 20060209 split out the repository type to avoid this check ?
208
elif isinstance(a_bzrdir._format,
209
(bzrlib.bzrdir.BzrDirFormat4,
210
bzrlib.bzrdir.BzrDirFormat5,
211
bzrlib.bzrdir.BzrDirFormat6)):
212
result = a_bzrdir.open_repository()
214
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
215
self.copy_content_into(result, revision_id, basis)
218
def has_revision(self, revision_id):
219
"""True if this branch has a copy of the revision.
221
This does not necessarily imply the revision is merge
222
or on the mainline."""
223
return (revision_id is None
224
or self.revision_store.has_id(revision_id))
227
def get_revision_xml_file(self, revision_id):
228
"""Return XML file object for revision object."""
229
if not revision_id or not isinstance(revision_id, basestring):
230
raise InvalidRevisionId(revision_id=revision_id, branch=self)
232
return self.revision_store.get(revision_id)
233
except (IndexError, KeyError):
234
raise bzrlib.errors.NoSuchRevision(self, revision_id)
237
def get_revision_xml(self, revision_id):
238
return self.get_revision_xml_file(revision_id).read()
241
def get_revision_reconcile(self, revision_id):
242
"""'reconcile' helper routine that allows access to a revision always.
244
This variant of get_revision does not cross check the weave graph
245
against the revision one as get_revision does: but it should only
246
be used by reconcile, or reconcile-alike commands that are correcting
247
or testing the revision graph.
249
xml_file = self.get_revision_xml_file(revision_id)
252
r = bzrlib.xml5.serializer_v5.read_revision(xml_file)
253
except SyntaxError, e:
254
raise bzrlib.errors.BzrError('failed to unpack revision_xml',
258
assert r.revision_id == revision_id
262
def get_revision(self, revision_id):
263
"""Return the Revision object for a named revision"""
264
r = self.get_revision_reconcile(revision_id)
265
# weave corruption can lead to absent revision markers that should be
267
# the following test is reasonably cheap (it needs a single weave read)
268
# and the weave is cached in read transactions. In write transactions
269
# it is not cached but typically we only read a small number of
270
# revisions. For knits when they are introduced we will probably want
271
# to ensure that caching write transactions are in use.
272
inv = self.get_inventory_weave()
273
weave_parents = inv.parent_names(revision_id)
274
weave_names = inv.names()
275
for parent_id in r.parent_ids:
276
if parent_id in weave_names:
277
# this parent must not be a ghost.
278
if not parent_id in weave_parents:
280
raise errors.CorruptRepository(self)
284
def get_revision_sha1(self, revision_id):
285
"""Hash the stored value of a revision, and return it."""
286
# In the future, revision entries will be signed. At that
287
# point, it is probably best *not* to include the signature
288
# in the revision hash. Because that lets you re-sign
289
# the revision, (add signatures/remove signatures) and still
290
# have all hash pointers stay consistent.
291
# But for now, just hash the contents.
292
return bzrlib.osutils.sha_file(self.get_revision_xml_file(revision_id))
295
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
296
self.revision_store.add(StringIO(gpg_strategy.sign(plaintext)),
299
def fileid_involved_between_revs(self, from_revid, to_revid):
300
"""Find file_id(s) which are involved in the changes between revisions.
302
This determines the set of revisions which are involved, and then
303
finds all file ids affected by those revisions.
305
# TODO: jam 20060119 This code assumes that w.inclusions will
306
# always be correct. But because of the presence of ghosts
307
# it is possible to be wrong.
308
# One specific example from Robert Collins:
309
# Two branches, with revisions ABC, and AD
310
# C is a ghost merge of D.
311
# Inclusions doesn't recognize D as an ancestor.
312
# If D is ever merged in the future, the weave
313
# won't be fixed, because AD never saw revision C
314
# to cause a conflict which would force a reweave.
315
w = self.get_inventory_weave()
316
from_set = set(w.inclusions([w.lookup(from_revid)]))
317
to_set = set(w.inclusions([w.lookup(to_revid)]))
318
included = to_set.difference(from_set)
319
changed = map(w.idx_to_name, included)
320
return self._fileid_involved_by_set(changed)
322
def fileid_involved(self, last_revid=None):
323
"""Find all file_ids modified in the ancestry of last_revid.
325
:param last_revid: If None, last_revision() will be used.
327
w = self.get_inventory_weave()
329
changed = set(w._names)
331
included = w.inclusions([w.lookup(last_revid)])
332
changed = map(w.idx_to_name, included)
333
return self._fileid_involved_by_set(changed)
335
def fileid_involved_by_set(self, changes):
336
"""Find all file_ids modified by the set of revisions passed in.
338
:param changes: A set() of revision ids
340
# TODO: jam 20060119 This line does *nothing*, remove it.
341
# or better yet, change _fileid_involved_by_set so
342
# that it takes the inventory weave, rather than
343
# pulling it out by itself.
344
return self._fileid_involved_by_set(changes)
346
def _fileid_involved_by_set(self, changes):
347
"""Find the set of file-ids affected by the set of revisions.
349
:param changes: A set() of revision ids.
350
:return: A set() of file ids.
352
This peaks at the Weave, interpreting each line, looking to
353
see if it mentions one of the revisions. And if so, includes
354
the file id mentioned.
355
This expects both the Weave format, and the serialization
356
to have a single line per file/directory, and to have
357
fileid="" and revision="" on that line.
359
assert isinstance(self._format, (RepositoryFormat5,
362
RepositoryFormatKnit1)), \
363
"fileid_involved only supported for branches which store inventory as unnested xml"
365
w = self.get_inventory_weave()
367
for line in w._weave:
369
# it is ugly, but it is due to the weave structure
370
if not isinstance(line, basestring): continue
372
start = line.find('file_id="')+9
373
if start < 9: continue
374
end = line.find('"', start)
376
file_id = xml.sax.saxutils.unescape(line[start:end])
378
# check if file_id is already present
379
if file_id in file_ids: continue
381
start = line.find('revision="')+10
382
if start < 10: continue
383
end = line.find('"', start)
385
revision_id = xml.sax.saxutils.unescape(line[start:end])
387
if revision_id in changes:
388
file_ids.add(file_id)
392
def get_inventory_weave(self):
393
return self.control_weaves.get_weave('inventory',
394
self.get_transaction())
397
def get_inventory(self, revision_id):
398
"""Get Inventory object by hash."""
399
xml = self.get_inventory_xml(revision_id)
400
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
403
def get_inventory_xml(self, revision_id):
404
"""Get inventory XML as a file object."""
406
assert isinstance(revision_id, basestring), type(revision_id)
407
iw = self.get_inventory_weave()
408
return iw.get_text(iw.lookup(revision_id))
410
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
413
def get_inventory_sha1(self, revision_id):
414
"""Return the sha1 hash of the inventory entry
416
return self.get_revision(revision_id).inventory_sha1
419
def get_revision_inventory(self, revision_id):
420
"""Return inventory of a past revision."""
421
# TODO: Unify this with get_inventory()
422
# bzr 0.0.6 and later imposes the constraint that the inventory_id
423
# must be the same as its revision, so this is trivial.
424
if revision_id is None:
425
# This does not make sense: if there is no revision,
426
# then it is the current tree inventory surely ?!
427
# and thus get_root_id() is something that looks at the last
428
# commit on the branch, and the get_root_id is an inventory check.
429
raise NotImplementedError
430
# return Inventory(self.get_root_id())
432
return self.get_inventory(revision_id)
436
"""Return True if this repository is flagged as a shared repository."""
437
# FIXME format 4-6 cannot be shared, this is technically faulty.
438
return self.control_files._transport.has('shared-storage')
441
def revision_tree(self, revision_id):
442
"""Return Tree for a revision on this branch.
444
`revision_id` may be None for the null revision, in which case
445
an `EmptyTree` is returned."""
446
# TODO: refactor this to use an existing revision object
447
# so we don't need to read it in twice.
448
if revision_id is None or revision_id == NULL_REVISION:
451
inv = self.get_revision_inventory(revision_id)
452
return RevisionTree(self, inv, revision_id)
455
def get_ancestry(self, revision_id):
456
"""Return a list of revision-ids integrated by a revision.
458
This is topologically sorted.
460
if revision_id is None:
462
if not self.has_revision(revision_id):
463
raise errors.NoSuchRevision(self, revision_id)
464
w = self.get_inventory_weave()
465
return [None] + map(w.idx_to_name,
466
w.inclusions([w.lookup(revision_id)]))
469
def print_file(self, file, revision_id):
470
"""Print `file` to stdout.
472
FIXME RBC 20060125 as John Meinel points out this is a bad api
473
- it writes to stdout, it assumes that that is valid etc. Fix
474
by creating a new more flexible convenience function.
476
tree = self.revision_tree(revision_id)
477
# use inventory as it was in that revision
478
file_id = tree.inventory.path2id(file)
480
raise BzrError("%r is not present in revision %s" % (file, revno))
482
revno = self.revision_id_to_revno(revision_id)
483
except errors.NoSuchRevision:
484
# TODO: This should not be BzrError,
485
# but NoSuchFile doesn't fit either
486
raise BzrError('%r is not present in revision %s'
487
% (file, revision_id))
489
raise BzrError('%r is not present in revision %s'
491
tree.print_file(file_id)
493
def get_transaction(self):
494
return self.control_files.get_transaction()
497
def set_make_working_trees(self, new_value):
498
"""Set the policy flag for making working trees when creating branches.
500
This only applies to branches that use this repository.
502
The default is 'True'.
503
:param new_value: True to restore the default, False to disable making
506
# FIXME: split out into a new class/strategy ?
507
if isinstance(self._format, (RepositoryFormat4,
510
raise NotImplementedError(self.set_make_working_trees)
513
self.control_files._transport.delete('no-working-trees')
514
except errors.NoSuchFile:
517
self.control_files.put_utf8('no-working-trees', '')
519
def make_working_trees(self):
520
"""Returns the policy for making working trees on new branches."""
521
# FIXME: split out into a new class/strategy ?
522
if isinstance(self._format, (RepositoryFormat4,
526
return not self.control_files._transport.has('no-working-trees')
529
def sign_revision(self, revision_id, gpg_strategy):
530
plaintext = Testament.from_revision(self, revision_id).as_short_text()
531
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
534
class AllInOneRepository(Repository):
535
"""Legacy support - the repository behaviour for all-in-one branches."""
537
def __init__(self, _format, a_bzrdir, revision_store):
538
# we reuse one control files instance.
539
dir_mode = a_bzrdir._control_files._dir_mode
540
file_mode = a_bzrdir._control_files._file_mode
542
def get_weave(name, prefixed=False):
544
name = safe_unicode(name)
547
relpath = a_bzrdir._control_files._escape(name)
548
weave_transport = a_bzrdir._control_files._transport.clone(relpath)
549
ws = WeaveStore(weave_transport, prefixed=prefixed,
552
if a_bzrdir._control_files._transport.should_cache():
553
ws.enable_cache = True
556
def get_store(name, compressed=True, prefixed=False):
557
# FIXME: This approach of assuming stores are all entirely compressed
558
# or entirely uncompressed is tidy, but breaks upgrade from
559
# some existing branches where there's a mixture; we probably
560
# still want the option to look for both.
561
relpath = a_bzrdir._control_files._escape(name)
562
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
563
prefixed=prefixed, compressed=compressed,
566
#if self._transport.should_cache():
567
# cache_path = os.path.join(self.cache_root, name)
568
# os.mkdir(cache_path)
569
# store = bzrlib.store.CachedStore(store, cache_path)
572
# not broken out yet because the controlweaves|inventory_store
573
# and text_store | weave_store bits are still different.
574
if isinstance(_format, RepositoryFormat4):
575
self.inventory_store = get_store('inventory-store')
576
self.text_store = get_store('text-store')
577
elif isinstance(_format, RepositoryFormat5):
578
self.control_weaves = get_weave('')
579
self.weave_store = get_weave('weaves')
580
elif isinstance(_format, RepositoryFormat6):
581
self.control_weaves = get_weave('')
582
self.weave_store = get_weave('weaves', prefixed=True)
584
raise errors.BzrError('unreachable code: unexpected repository'
586
revision_store.register_suffix('sig')
587
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, revision_store)
590
class MetaDirRepository(Repository):
591
"""Repositories in the new meta-dir layout."""
593
def __init__(self, _format, a_bzrdir, control_files, revision_store):
594
super(MetaDirRepository, self).__init__(_format,
599
dir_mode = self.control_files._dir_mode
600
file_mode = self.control_files._file_mode
602
def get_weave(name, prefixed=False):
604
name = safe_unicode(name)
607
relpath = self.control_files._escape(name)
608
weave_transport = self.control_files._transport.clone(relpath)
609
ws = WeaveStore(weave_transport, prefixed=prefixed,
612
if self.control_files._transport.should_cache():
613
ws.enable_cache = True
616
if isinstance(self._format, RepositoryFormat7):
617
self.control_weaves = get_weave('')
618
self.weave_store = get_weave('weaves', prefixed=True)
619
elif isinstance(self._format, RepositoryFormatKnit1):
620
self.control_weaves = get_weave('')
621
self.weave_store = get_weave('knits', prefixed=True)
623
raise errors.BzrError('unreachable code: unexpected repository'
627
class RepositoryFormat(object):
628
"""A repository format.
630
Formats provide three things:
631
* An initialization routine to construct repository data on disk.
632
* a format string which is used when the BzrDir supports versioned
634
* an open routine which returns a Repository instance.
636
Formats are placed in an dict by their format string for reference
637
during opening. These should be subclasses of RepositoryFormat
640
Once a format is deprecated, just deprecate the initialize and open
641
methods on the format class. Do not deprecate the object, as the
642
object will be created every system load.
644
Common instance attributes:
645
_matchingbzrdir - the bzrdir format that the repository format was
646
originally written to work with. This can be used if manually
647
constructing a bzrdir and repository, or more commonly for test suite
651
_default_format = None
652
"""The default format used for new repositories."""
655
"""The known formats."""
658
def find_format(klass, a_bzrdir):
659
"""Return the format for the repository object in a_bzrdir."""
661
transport = a_bzrdir.get_repository_transport(None)
662
format_string = transport.get("format").read()
663
return klass._formats[format_string]
664
except errors.NoSuchFile:
665
raise errors.NoRepositoryPresent(a_bzrdir)
667
raise errors.UnknownFormatError(format_string)
670
def get_default_format(klass):
671
"""Return the current default format."""
672
return klass._default_format
674
def get_format_string(self):
675
"""Return the ASCII format string that identifies this format.
677
Note that in pre format ?? repositories the format string is
678
not permitted nor written to disk.
680
raise NotImplementedError(self.get_format_string)
682
def _get_revision_store(self, repo_transport, control_files):
683
"""Return the revision store object for this a_bzrdir."""
684
raise NotImplementedError(self._get_revision_store)
686
def _get_rev_store(self,
692
"""Common logic for getting a revision store for a repository.
694
see self._get_revision_store for the method to
695
get the store for a repository.
698
name = safe_unicode(name)
701
dir_mode = control_files._dir_mode
702
file_mode = control_files._file_mode
703
revision_store =TextStore(transport.clone(name),
705
compressed=compressed,
708
revision_store.register_suffix('sig')
709
return revision_store
711
def initialize(self, a_bzrdir, shared=False):
712
"""Initialize a repository of this format in a_bzrdir.
714
:param a_bzrdir: The bzrdir to put the new repository in it.
715
:param shared: The repository should be initialized as a sharable one.
717
This may raise UninitializableFormat if shared repository are not
718
compatible the a_bzrdir.
721
def is_supported(self):
722
"""Is this format supported?
724
Supported formats must be initializable and openable.
725
Unsupported formats may not support initialization or committing or
726
some other features depending on the reason for not being supported.
730
def open(self, a_bzrdir, _found=False):
731
"""Return an instance of this format for the bzrdir a_bzrdir.
733
_found is a private parameter, do not use it.
735
raise NotImplementedError(self.open)
738
def register_format(klass, format):
739
klass._formats[format.get_format_string()] = format
742
def set_default_format(klass, format):
743
klass._default_format = format
746
def unregister_format(klass, format):
747
assert klass._formats[format.get_format_string()] is format
748
del klass._formats[format.get_format_string()]
751
class PreSplitOutRepositoryFormat(RepositoryFormat):
752
"""Base class for the pre split out repository formats."""
754
def initialize(self, a_bzrdir, shared=False, _internal=False):
755
"""Create a weave repository.
757
TODO: when creating split out bzr branch formats, move this to a common
758
base for Format5, Format6. or something like that.
760
from bzrlib.weavefile import write_weave_v5
761
from bzrlib.weave import Weave
764
raise errors.IncompatibleFormat(self, a_bzrdir._format)
767
# always initialized when the bzrdir is.
768
return self.open(a_bzrdir, _found=True)
770
# Create an empty weave
772
bzrlib.weavefile.write_weave_v5(Weave(), sio)
773
empty_weave = sio.getvalue()
775
mutter('creating repository in %s.', a_bzrdir.transport.base)
776
dirs = ['revision-store', 'weaves']
777
lock_file = 'branch-lock'
778
files = [('inventory.weave', StringIO(empty_weave)),
781
# FIXME: RBC 20060125 dont peek under the covers
782
# NB: no need to escape relative paths that are url safe.
783
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock')
784
control_files.lock_write()
785
control_files._transport.mkdir_multi(dirs,
786
mode=control_files._dir_mode)
788
for file, content in files:
789
control_files.put(file, content)
791
control_files.unlock()
792
return self.open(a_bzrdir, _found=True)
794
def open(self, a_bzrdir, _found=False):
795
"""See RepositoryFormat.open()."""
797
# we are being called directly and must probe.
798
raise NotImplementedError
800
repo_transport = a_bzrdir.get_repository_transport(None)
801
control_files = a_bzrdir._control_files
802
revision_store = self._get_revision_store(repo_transport, control_files)
803
return AllInOneRepository(_format=self,
805
revision_store=revision_store)
808
class RepositoryFormat4(PreSplitOutRepositoryFormat):
809
"""Bzr repository format 4.
811
This repository format has:
813
- TextStores for texts, inventories,revisions.
815
This format is deprecated: it indexes texts using a text id which is
816
removed in format 5; initializationa and write support for this format
821
super(RepositoryFormat4, self).__init__()
822
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
824
def initialize(self, url, shared=False, _internal=False):
825
"""Format 4 branches cannot be created."""
826
raise errors.UninitializableFormat(self)
828
def is_supported(self):
829
"""Format 4 is not supported.
831
It is not supported because the model changed from 4 to 5 and the
832
conversion logic is expensive - so doing it on the fly was not
837
def _get_revision_store(self, repo_transport, control_files):
838
"""See RepositoryFormat._get_revision_store()."""
839
return self._get_rev_store(repo_transport,
844
class RepositoryFormat5(PreSplitOutRepositoryFormat):
845
"""Bzr control format 5.
847
This repository format has:
848
- weaves for file texts and inventory
850
- TextStores for revisions and signatures.
854
super(RepositoryFormat5, self).__init__()
855
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
857
def _get_revision_store(self, repo_transport, control_files):
858
"""See RepositoryFormat._get_revision_store()."""
859
"""Return the revision store object for this a_bzrdir."""
860
return self._get_rev_store(repo_transport,
866
class RepositoryFormat6(PreSplitOutRepositoryFormat):
867
"""Bzr control format 6.
869
This repository format has:
870
- weaves for file texts and inventory
871
- hash subdirectory based stores.
872
- TextStores for revisions and signatures.
876
super(RepositoryFormat6, self).__init__()
877
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
879
def _get_revision_store(self, repo_transport, control_files):
880
"""See RepositoryFormat._get_revision_store()."""
881
return self._get_rev_store(repo_transport,
888
class MetaDirRepositoryFormat(RepositoryFormat):
889
"""Common base class for the new repositories using the metadir layour."""
892
super(MetaDirRepositoryFormat, self).__init__()
893
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
895
def _create_control_files(self, a_bzrdir):
896
"""Create the required files and the initial control_files object."""
897
# FIXME: RBC 20060125 dont peek under the covers
898
# NB: no need to escape relative paths that are url safe.
900
repository_transport = a_bzrdir.get_repository_transport(self)
901
repository_transport.put(lock_file, StringIO()) # TODO get the file mode from the bzrdir lock files., mode=file_mode)
902
control_files = LockableFiles(repository_transport, 'lock')
905
def _get_revision_store(self, repo_transport, control_files):
906
"""See RepositoryFormat._get_revision_store()."""
907
return self._get_rev_store(repo_transport,
914
def open(self, a_bzrdir, _found=False, _override_transport=None):
915
"""See RepositoryFormat.open().
917
:param _override_transport: INTERNAL USE ONLY. Allows opening the
918
repository at a slightly different url
919
than normal. I.e. during 'upgrade'.
922
format = RepositoryFormat.find_format(a_bzrdir)
923
assert format.__class__ == self.__class__
924
if _override_transport is not None:
925
repo_transport = _override_transport
927
repo_transport = a_bzrdir.get_repository_transport(None)
928
control_files = LockableFiles(repo_transport, 'lock')
929
revision_store = self._get_revision_store(repo_transport, control_files)
930
return MetaDirRepository(_format=self,
932
control_files=control_files,
933
revision_store=revision_store)
935
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
936
"""Upload the initial blank content."""
937
control_files = self._create_control_files(a_bzrdir)
938
control_files.lock_write()
939
control_files._transport.mkdir_multi(dirs,
940
mode=control_files._dir_mode)
942
for file, content in files:
943
control_files.put(file, content)
944
for file, content in utf8_files:
945
control_files.put_utf8(file, content)
947
control_files.put_utf8('shared-storage', '')
949
control_files.unlock()
952
class RepositoryFormat7(MetaDirRepositoryFormat):
955
This repository format has:
956
- weaves for file texts and inventory
957
- hash subdirectory based stores.
958
- TextStores for revisions and signatures.
959
- a format marker of its own
960
- an optional 'shared-storage' flag
961
- an optional 'no-working-trees' flag
964
def get_format_string(self):
965
"""See RepositoryFormat.get_format_string()."""
966
return "Bazaar-NG Repository format 7"
968
def initialize(self, a_bzrdir, shared=False):
969
"""Create a weave repository.
971
:param shared: If true the repository will be initialized as a shared
974
from bzrlib.weavefile import write_weave_v5
975
from bzrlib.weave import Weave
977
# Create an empty weave
979
bzrlib.weavefile.write_weave_v5(Weave(), sio)
980
empty_weave = sio.getvalue()
982
mutter('creating repository in %s.', a_bzrdir.transport.base)
983
dirs = ['revision-store', 'weaves']
984
files = [('inventory.weave', StringIO(empty_weave)),
986
utf8_files = [('format', self.get_format_string())]
988
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
989
return self.open(a_bzrdir=a_bzrdir, _found=True)
992
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
993
"""Bzr repository knit format 1.
995
This repository format has:
996
- knits for file texts and inventory
997
- hash subdirectory based stores.
998
- knits for revisions and signatures
999
- TextStores for revisions and signatures.
1000
- a format marker of its own
1001
- an optional 'shared-storage' flag
1002
- an optional 'no-working-trees' flag
1005
def get_format_string(self):
1006
"""See RepositoryFormat.get_format_string()."""
1007
return "Bazaar-NG Knit Repository Format 1"
1009
def initialize(self, a_bzrdir, shared=False):
1010
"""Create a knit format 1 repository.
1012
:param shared: If true the repository will be initialized as a shared
1014
XXX NOTE that this current uses a Weave for testing and will become
1015
A Knit in due course.
1017
from bzrlib.weavefile import write_weave_v5
1018
from bzrlib.weave import Weave
1020
# Create an empty weave
1022
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1023
empty_weave = sio.getvalue()
1025
mutter('creating repository in %s.', a_bzrdir.transport.base)
1026
dirs = ['revision-store', 'knits']
1027
files = [('inventory.weave', StringIO(empty_weave)),
1029
utf8_files = [('format', self.get_format_string())]
1031
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1032
return self.open(a_bzrdir=a_bzrdir, _found=True)
1035
# formats which have no format string are not discoverable
1036
# and not independently creatable, so are not registered.
1037
_default_format = RepositoryFormat7()
1038
RepositoryFormat.register_format(_default_format)
1039
RepositoryFormat.register_format(RepositoryFormatKnit1())
1040
RepositoryFormat.set_default_format(_default_format)
1041
_legacy_formats = [RepositoryFormat4(),
1042
RepositoryFormat5(),
1043
RepositoryFormat6()]
1046
class InterRepository(object):
1047
"""This class represents operations taking place between two repositories.
1049
Its instances have methods like copy_content and fetch, and contain
1050
references to the source and target repositories these operations can be
1053
Often we will provide convenience methods on 'repository' which carry out
1054
operations with another repository - they will always forward to
1055
InterRepository.get(other).method_name(parameters).
1057
# XXX: FIXME: FUTURE: robertc
1058
# testing of these probably requires a factory in optimiser type, and
1059
# then a test adapter to test each type thoroughly.
1063
"""The available optimised InterRepository types."""
1065
def __init__(self, source, target):
1066
"""Construct a default InterRepository instance. Please use 'get'.
1068
Only subclasses of InterRepository should call
1069
InterRepository.__init__ - clients should call InterRepository.get
1070
instead which will create an optimised InterRepository if possible.
1072
self.source = source
1073
self.target = target
1076
def copy_content(self, revision_id=None, basis=None):
1077
"""Make a complete copy of the content in self into destination.
1079
This is a destructive operation! Do not use it on existing
1082
:param revision_id: Only copy the content needed to construct
1083
revision_id and its parents.
1084
:param basis: Copy the needed data preferentially from basis.
1087
self.target.set_make_working_trees(self.source.make_working_trees())
1088
except NotImplementedError:
1090
# grab the basis available data
1091
if basis is not None:
1092
self.target.fetch(basis, revision_id=revision_id)
1093
# but dont both fetching if we have the needed data now.
1094
if (revision_id not in (None, NULL_REVISION) and
1095
self.target.has_revision(revision_id)):
1097
self.target.fetch(self.source, revision_id=revision_id)
1099
def _double_lock(self, lock_source, lock_target):
1100
"""Take out too locks, rolling back the first if the second throws."""
1105
# we want to ensure that we don't leave source locked by mistake.
1106
# and any error on target should not confuse source.
1107
self.source.unlock()
1111
def fetch(self, revision_id=None, pb=None):
1112
"""Fetch the content required to construct revision_id.
1114
The content is copied from source to target.
1116
:param revision_id: if None all content is copied, if NULL_REVISION no
1118
:param pb: optional progress bar to use for progress reports. If not
1119
provided a default one will be created.
1121
Returns the copied revision count and the failed revisions in a tuple:
1124
from bzrlib.fetch import RepoFetcher
1125
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1126
self.source, self.source._format, self.target, self.target._format)
1127
f = RepoFetcher(to_repository=self.target,
1128
from_repository=self.source,
1129
last_revision=revision_id,
1131
return f.count_copied, f.failed_revisions
1134
def get(klass, repository_source, repository_target):
1135
"""Retrieve a InterRepository worker object for these repositories.
1137
:param repository_source: the repository to be the 'source' member of
1138
the InterRepository instance.
1139
:param repository_target: the repository to be the 'target' member of
1140
the InterRepository instance.
1141
If an optimised InterRepository worker exists it will be used otherwise
1142
a default InterRepository instance will be created.
1144
for provider in klass._optimisers:
1145
if provider.is_compatible(repository_source, repository_target):
1146
return provider(repository_source, repository_target)
1147
return InterRepository(repository_source, repository_target)
1149
def lock_read(self):
1150
"""Take out a logical read lock.
1152
This will lock the source branch and the target branch. The source gets
1153
a read lock and the target a read lock.
1155
self._double_lock(self.source.lock_read, self.target.lock_read)
1157
def lock_write(self):
1158
"""Take out a logical write lock.
1160
This will lock the source branch and the target branch. The source gets
1161
a read lock and the target a write lock.
1163
self._double_lock(self.source.lock_read, self.target.lock_write)
1166
def missing_revision_ids(self, revision_id=None):
1167
"""Return the revision ids that source has that target does not.
1169
These are returned in topological order.
1171
:param revision_id: only return revision ids included by this
1174
# generic, possibly worst case, slow code path.
1175
target_ids = set(self.target.all_revision_ids())
1176
if revision_id is not None:
1177
source_ids = self.source.get_ancestry(revision_id)
1178
assert source_ids.pop(0) == None
1180
source_ids = self.source.all_revision_ids()
1181
result_set = set(source_ids).difference(target_ids)
1182
# this may look like a no-op: its not. It preserves the ordering
1183
# other_ids had while only returning the members from other_ids
1184
# that we've decided we need.
1185
return [rev_id for rev_id in source_ids if rev_id in result_set]
1188
def register_optimiser(klass, optimiser):
1189
"""Register an InterRepository optimiser."""
1190
klass._optimisers.add(optimiser)
1193
"""Release the locks on source and target."""
1195
self.target.unlock()
1197
self.source.unlock()
1200
def unregister_optimiser(klass, optimiser):
1201
"""Unregister an InterRepository optimiser."""
1202
klass._optimisers.remove(optimiser)
1205
class InterWeaveRepo(InterRepository):
1206
"""Optimised code paths between Weave based repositories."""
1208
_matching_repo_format = _default_format
1209
"""Repository format for testing with."""
1212
def is_compatible(source, target):
1213
"""Be compatible with known Weave formats.
1215
We dont test for the stores being of specific types becase that
1216
could lead to confusing results, and there is no need to be
1220
return (isinstance(source._format, (RepositoryFormat5,
1222
RepositoryFormat7)) and
1223
isinstance(target._format, (RepositoryFormat5,
1225
RepositoryFormat7)))
1226
except AttributeError:
1230
def copy_content(self, revision_id=None, basis=None):
1231
"""See InterRepository.copy_content()."""
1232
# weave specific optimised path:
1233
if basis is not None:
1234
# copy the basis in, then fetch remaining data.
1235
basis.copy_content_into(self.target, revision_id)
1236
# the basis copy_content_into could misset this.
1238
self.target.set_make_working_trees(self.source.make_working_trees())
1239
except NotImplementedError:
1241
self.target.fetch(self.source, revision_id=revision_id)
1244
self.target.set_make_working_trees(self.source.make_working_trees())
1245
except NotImplementedError:
1247
# FIXME do not peek!
1248
if self.source.control_files._transport.listable():
1249
pb = bzrlib.ui.ui_factory.progress_bar()
1250
copy_all(self.source.weave_store,
1251
self.target.weave_store, pb=pb)
1252
pb.update('copying inventory', 0, 1)
1253
self.target.control_weaves.copy_multi(
1254
self.source.control_weaves, ['inventory'])
1255
copy_all(self.source.revision_store,
1256
self.target.revision_store, pb=pb)
1258
self.target.fetch(self.source, revision_id=revision_id)
1261
def fetch(self, revision_id=None, pb=None):
1262
"""See InterRepository.fetch()."""
1263
from bzrlib.fetch import RepoFetcher
1264
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1265
self.source, self.source._format, self.target, self.target._format)
1266
f = RepoFetcher(to_repository=self.target,
1267
from_repository=self.source,
1268
last_revision=revision_id,
1270
return f.count_copied, f.failed_revisions
1273
def missing_revision_ids(self, revision_id=None):
1274
"""See InterRepository.missing_revision_ids()."""
1275
# we want all revisions to satisfy revision_id in source.
1276
# but we dont want to stat every file here and there.
1277
# we want then, all revisions other needs to satisfy revision_id
1278
# checked, but not those that we have locally.
1279
# so the first thing is to get a subset of the revisions to
1280
# satisfy revision_id in source, and then eliminate those that
1281
# we do already have.
1282
# this is slow on high latency connection to self, but as as this
1283
# disk format scales terribly for push anyway due to rewriting
1284
# inventory.weave, this is considered acceptable.
1286
if revision_id is not None:
1287
source_ids = self.source.get_ancestry(revision_id)
1288
assert source_ids.pop(0) == None
1290
source_ids = self.source._all_possible_ids()
1291
source_ids_set = set(source_ids)
1292
# source_ids is the worst possible case we may need to pull.
1293
# now we want to filter source_ids against what we actually
1294
# have in target, but dont try to check for existence where we know
1295
# we do not have a revision as that would be pointless.
1296
target_ids = set(self.target._all_possible_ids())
1297
possibly_present_revisions = target_ids.intersection(source_ids_set)
1298
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1299
required_revisions = source_ids_set.difference(actually_present_revisions)
1300
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1301
if revision_id is not None:
1302
# we used get_ancestry to determine source_ids then we are assured all
1303
# revisions referenced are present as they are installed in topological order.
1304
# and the tip revision was validated by get_ancestry.
1305
return required_topo_revisions
1307
# if we just grabbed the possibly available ids, then
1308
# we only have an estimate of whats available and need to validate
1309
# that against the revision records.
1310
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1313
InterRepository.register_optimiser(InterWeaveRepo)
1316
class RepositoryTestProviderAdapter(object):
1317
"""A tool to generate a suite testing multiple repository formats at once.
1319
This is done by copying the test once for each transport and injecting
1320
the transport_server, transport_readonly_server, and bzrdir_format and
1321
repository_format classes into each copy. Each copy is also given a new id()
1322
to make it easy to identify.
1325
def __init__(self, transport_server, transport_readonly_server, formats):
1326
self._transport_server = transport_server
1327
self._transport_readonly_server = transport_readonly_server
1328
self._formats = formats
1330
def adapt(self, test):
1331
result = TestSuite()
1332
for repository_format, bzrdir_format in self._formats:
1333
new_test = deepcopy(test)
1334
new_test.transport_server = self._transport_server
1335
new_test.transport_readonly_server = self._transport_readonly_server
1336
new_test.bzrdir_format = bzrdir_format
1337
new_test.repository_format = repository_format
1338
def make_new_test_id():
1339
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1340
return lambda: new_id
1341
new_test.id = make_new_test_id()
1342
result.addTest(new_test)
1346
class InterRepositoryTestProviderAdapter(object):
1347
"""A tool to generate a suite testing multiple inter repository formats.
1349
This is done by copying the test once for each interrepo provider and injecting
1350
the transport_server, transport_readonly_server, repository_format and
1351
repository_to_format classes into each copy.
1352
Each copy is also given a new id() to make it easy to identify.
1355
def __init__(self, transport_server, transport_readonly_server, formats):
1356
self._transport_server = transport_server
1357
self._transport_readonly_server = transport_readonly_server
1358
self._formats = formats
1360
def adapt(self, test):
1361
result = TestSuite()
1362
for interrepo_class, repository_format, repository_format_to in self._formats:
1363
new_test = deepcopy(test)
1364
new_test.transport_server = self._transport_server
1365
new_test.transport_readonly_server = self._transport_readonly_server
1366
new_test.interrepo_class = interrepo_class
1367
new_test.repository_format = repository_format
1368
new_test.repository_format_to = repository_format_to
1369
def make_new_test_id():
1370
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1371
return lambda: new_id
1372
new_test.id = make_new_test_id()
1373
result.addTest(new_test)
1377
def default_test_list():
1378
"""Generate the default list of interrepo permutations to test."""
1380
# test the default InterRepository between format 6 and the current
1382
# XXX: robertc 20060220 reinstate this when there are two supported
1383
# formats which do not have an optimal code path between them.
1384
result.append((InterRepository,
1385
RepositoryFormat6(),
1386
RepositoryFormatKnit1()))
1387
for optimiser in InterRepository._optimisers:
1388
result.append((optimiser,
1389
optimiser._matching_repo_format,
1390
optimiser._matching_repo_format
1392
# if there are specific combinations we want to use, we can add them
1397
class CopyConverter(object):
1398
"""A repository conversion tool which just performs a copy of the content.
1400
This is slow but quite reliable.
1403
def __init__(self, target_format):
1404
"""Create a CopyConverter.
1406
:param target_format: The format the resulting repository should be.
1408
self.target_format = target_format
1410
def convert(self, repo, pb):
1411
"""Perform the conversion of to_convert, giving feedback via pb.
1413
:param to_convert: The disk object to convert.
1414
:param pb: a progress bar to use for progress information.
1419
# this is only useful with metadir layouts - separated repo content.
1420
# trigger an assertion if not such
1421
repo._format.get_format_string()
1422
self.repo_dir = repo.bzrdir
1423
self.step('Moving repository to repository.backup')
1424
self.repo_dir.transport.move('repository', 'repository.backup')
1425
backup_transport = self.repo_dir.transport.clone('repository.backup')
1426
self.source_repo = repo._format.open(self.repo_dir,
1428
_override_transport=backup_transport)
1429
self.step('Creating new repository')
1430
converted = self.target_format.initialize(self.repo_dir,
1431
self.source_repo.is_shared())
1432
converted.lock_write()
1434
self.step('Copying content into repository.')
1435
self.source_repo.copy_content_into(converted)
1438
self.step('Deleting old repository content.')
1439
self.repo_dir.transport.delete_tree('repository.backup')
1440
self.pb.note('repository converted')
1442
def step(self, message):
1443
"""Update the pb by a step."""
1445
self.pb.update(message, self.count, self.total)
added
removed
bzrlib/repository.py
