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(self, revision_id):
242
"""Return the Revision object for a named revision"""
243
xml_file = self.get_revision_xml_file(revision_id)
246
r = bzrlib.xml5.serializer_v5.read_revision(xml_file)
247
except SyntaxError, e:
248
raise bzrlib.errors.BzrError('failed to unpack revision_xml',
252
assert r.revision_id == revision_id
256
def get_revision_sha1(self, revision_id):
257
"""Hash the stored value of a revision, and return it."""
258
# In the future, revision entries will be signed. At that
259
# point, it is probably best *not* to include the signature
260
# in the revision hash. Because that lets you re-sign
261
# the revision, (add signatures/remove signatures) and still
262
# have all hash pointers stay consistent.
263
# But for now, just hash the contents.
264
return bzrlib.osutils.sha_file(self.get_revision_xml_file(revision_id))
267
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
268
self.revision_store.add(StringIO(gpg_strategy.sign(plaintext)),
271
def fileid_involved_between_revs(self, from_revid, to_revid):
272
"""Find file_id(s) which are involved in the changes between revisions.
274
This determines the set of revisions which are involved, and then
275
finds all file ids affected by those revisions.
277
# TODO: jam 20060119 This code assumes that w.inclusions will
278
# always be correct. But because of the presence of ghosts
279
# it is possible to be wrong.
280
# One specific example from Robert Collins:
281
# Two branches, with revisions ABC, and AD
282
# C is a ghost merge of D.
283
# Inclusions doesn't recognize D as an ancestor.
284
# If D is ever merged in the future, the weave
285
# won't be fixed, because AD never saw revision C
286
# to cause a conflict which would force a reweave.
287
w = self.get_inventory_weave()
288
from_set = set(w.inclusions([w.lookup(from_revid)]))
289
to_set = set(w.inclusions([w.lookup(to_revid)]))
290
included = to_set.difference(from_set)
291
changed = map(w.idx_to_name, included)
292
return self._fileid_involved_by_set(changed)
294
def fileid_involved(self, last_revid=None):
295
"""Find all file_ids modified in the ancestry of last_revid.
297
:param last_revid: If None, last_revision() will be used.
299
w = self.get_inventory_weave()
301
changed = set(w._names)
303
included = w.inclusions([w.lookup(last_revid)])
304
changed = map(w.idx_to_name, included)
305
return self._fileid_involved_by_set(changed)
307
def fileid_involved_by_set(self, changes):
308
"""Find all file_ids modified by the set of revisions passed in.
310
:param changes: A set() of revision ids
312
# TODO: jam 20060119 This line does *nothing*, remove it.
313
# or better yet, change _fileid_involved_by_set so
314
# that it takes the inventory weave, rather than
315
# pulling it out by itself.
316
return self._fileid_involved_by_set(changes)
318
def _fileid_involved_by_set(self, changes):
319
"""Find the set of file-ids affected by the set of revisions.
321
:param changes: A set() of revision ids.
322
:return: A set() of file ids.
324
This peaks at the Weave, interpreting each line, looking to
325
see if it mentions one of the revisions. And if so, includes
326
the file id mentioned.
327
This expects both the Weave format, and the serialization
328
to have a single line per file/directory, and to have
329
fileid="" and revision="" on that line.
331
assert isinstance(self._format, (RepositoryFormat5,
334
RepositoryFormatKnit1)), \
335
"fileid_involved only supported for branches which store inventory as unnested xml"
337
w = self.get_inventory_weave()
339
for line in w._weave:
341
# it is ugly, but it is due to the weave structure
342
if not isinstance(line, basestring): continue
344
start = line.find('file_id="')+9
345
if start < 9: continue
346
end = line.find('"', start)
348
file_id = xml.sax.saxutils.unescape(line[start:end])
350
# check if file_id is already present
351
if file_id in file_ids: continue
353
start = line.find('revision="')+10
354
if start < 10: continue
355
end = line.find('"', start)
357
revision_id = xml.sax.saxutils.unescape(line[start:end])
359
if revision_id in changes:
360
file_ids.add(file_id)
364
def get_inventory_weave(self):
365
return self.control_weaves.get_weave('inventory',
366
self.get_transaction())
369
def get_inventory(self, revision_id):
370
"""Get Inventory object by hash."""
371
xml = self.get_inventory_xml(revision_id)
372
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
375
def get_inventory_xml(self, revision_id):
376
"""Get inventory XML as a file object."""
378
assert isinstance(revision_id, basestring), type(revision_id)
379
iw = self.get_inventory_weave()
380
return iw.get_text(iw.lookup(revision_id))
382
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
385
def get_inventory_sha1(self, revision_id):
386
"""Return the sha1 hash of the inventory entry
388
return self.get_revision(revision_id).inventory_sha1
391
def get_revision_inventory(self, revision_id):
392
"""Return inventory of a past revision."""
393
# TODO: Unify this with get_inventory()
394
# bzr 0.0.6 and later imposes the constraint that the inventory_id
395
# must be the same as its revision, so this is trivial.
396
if revision_id is None:
397
# This does not make sense: if there is no revision,
398
# then it is the current tree inventory surely ?!
399
# and thus get_root_id() is something that looks at the last
400
# commit on the branch, and the get_root_id is an inventory check.
401
raise NotImplementedError
402
# return Inventory(self.get_root_id())
404
return self.get_inventory(revision_id)
408
"""Return True if this repository is flagged as a shared repository."""
409
# FIXME format 4-6 cannot be shared, this is technically faulty.
410
return self.control_files._transport.has('shared-storage')
413
def revision_tree(self, revision_id):
414
"""Return Tree for a revision on this branch.
416
`revision_id` may be None for the null revision, in which case
417
an `EmptyTree` is returned."""
418
# TODO: refactor this to use an existing revision object
419
# so we don't need to read it in twice.
420
if revision_id is None or revision_id == NULL_REVISION:
423
inv = self.get_revision_inventory(revision_id)
424
return RevisionTree(self, inv, revision_id)
427
def get_ancestry(self, revision_id):
428
"""Return a list of revision-ids integrated by a revision.
430
This is topologically sorted.
432
if revision_id is None:
434
if not self.has_revision(revision_id):
435
raise errors.NoSuchRevision(self, revision_id)
436
w = self.get_inventory_weave()
437
return [None] + map(w.idx_to_name,
438
w.inclusions([w.lookup(revision_id)]))
441
def print_file(self, file, revision_id):
442
"""Print `file` to stdout.
444
FIXME RBC 20060125 as John Meinel points out this is a bad api
445
- it writes to stdout, it assumes that that is valid etc. Fix
446
by creating a new more flexible convenience function.
448
tree = self.revision_tree(revision_id)
449
# use inventory as it was in that revision
450
file_id = tree.inventory.path2id(file)
452
raise BzrError("%r is not present in revision %s" % (file, revno))
454
revno = self.revision_id_to_revno(revision_id)
455
except errors.NoSuchRevision:
456
# TODO: This should not be BzrError,
457
# but NoSuchFile doesn't fit either
458
raise BzrError('%r is not present in revision %s'
459
% (file, revision_id))
461
raise BzrError('%r is not present in revision %s'
463
tree.print_file(file_id)
465
def get_transaction(self):
466
return self.control_files.get_transaction()
469
def set_make_working_trees(self, new_value):
470
"""Set the policy flag for making working trees when creating branches.
472
This only applies to branches that use this repository.
474
The default is 'True'.
475
:param new_value: True to restore the default, False to disable making
478
# FIXME: split out into a new class/strategy ?
479
if isinstance(self._format, (RepositoryFormat4,
482
raise NotImplementedError(self.set_make_working_trees)
485
self.control_files._transport.delete('no-working-trees')
486
except errors.NoSuchFile:
489
self.control_files.put_utf8('no-working-trees', '')
491
def make_working_trees(self):
492
"""Returns the policy for making working trees on new branches."""
493
# FIXME: split out into a new class/strategy ?
494
if isinstance(self._format, (RepositoryFormat4,
498
return not self.control_files._transport.has('no-working-trees')
501
def sign_revision(self, revision_id, gpg_strategy):
502
plaintext = Testament.from_revision(self, revision_id).as_short_text()
503
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
506
class AllInOneRepository(Repository):
507
"""Legacy support - the repository behaviour for all-in-one branches."""
509
def __init__(self, _format, a_bzrdir, revision_store):
510
# we reuse one control files instance.
511
dir_mode = a_bzrdir._control_files._dir_mode
512
file_mode = a_bzrdir._control_files._file_mode
514
def get_weave(name, prefixed=False):
516
name = safe_unicode(name)
519
relpath = a_bzrdir._control_files._escape(name)
520
weave_transport = a_bzrdir._control_files._transport.clone(relpath)
521
ws = WeaveStore(weave_transport, prefixed=prefixed,
524
if a_bzrdir._control_files._transport.should_cache():
525
ws.enable_cache = True
528
def get_store(name, compressed=True, prefixed=False):
529
# FIXME: This approach of assuming stores are all entirely compressed
530
# or entirely uncompressed is tidy, but breaks upgrade from
531
# some existing branches where there's a mixture; we probably
532
# still want the option to look for both.
533
relpath = a_bzrdir._control_files._escape(name)
534
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
535
prefixed=prefixed, compressed=compressed,
538
#if self._transport.should_cache():
539
# cache_path = os.path.join(self.cache_root, name)
540
# os.mkdir(cache_path)
541
# store = bzrlib.store.CachedStore(store, cache_path)
544
# not broken out yet because the controlweaves|inventory_store
545
# and text_store | weave_store bits are still different.
546
if isinstance(_format, RepositoryFormat4):
547
self.inventory_store = get_store('inventory-store')
548
self.text_store = get_store('text-store')
549
elif isinstance(_format, RepositoryFormat5):
550
self.control_weaves = get_weave('')
551
self.weave_store = get_weave('weaves')
552
elif isinstance(_format, RepositoryFormat6):
553
self.control_weaves = get_weave('')
554
self.weave_store = get_weave('weaves', prefixed=True)
556
raise errors.BzrError('unreachable code: unexpected repository'
558
revision_store.register_suffix('sig')
559
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, revision_store)
562
class MetaDirRepository(Repository):
563
"""Repositories in the new meta-dir layout."""
565
def __init__(self, _format, a_bzrdir, control_files, revision_store):
566
super(MetaDirRepository, self).__init__(_format,
571
dir_mode = self.control_files._dir_mode
572
file_mode = self.control_files._file_mode
574
def get_weave(name, prefixed=False):
576
name = safe_unicode(name)
579
relpath = self.control_files._escape(name)
580
weave_transport = self.control_files._transport.clone(relpath)
581
ws = WeaveStore(weave_transport, prefixed=prefixed,
584
if self.control_files._transport.should_cache():
585
ws.enable_cache = True
588
if isinstance(self._format, RepositoryFormat7):
589
self.control_weaves = get_weave('')
590
self.weave_store = get_weave('weaves', prefixed=True)
591
elif isinstance(self._format, RepositoryFormatKnit1):
592
self.control_weaves = get_weave('')
593
self.weave_store = get_weave('knits', prefixed=True)
595
raise errors.BzrError('unreachable code: unexpected repository'
599
class RepositoryFormat(object):
600
"""A repository format.
602
Formats provide three things:
603
* An initialization routine to construct repository data on disk.
604
* a format string which is used when the BzrDir supports versioned
606
* an open routine which returns a Repository instance.
608
Formats are placed in an dict by their format string for reference
609
during opening. These should be subclasses of RepositoryFormat
612
Once a format is deprecated, just deprecate the initialize and open
613
methods on the format class. Do not deprecate the object, as the
614
object will be created every system load.
616
Common instance attributes:
617
_matchingbzrdir - the bzrdir format that the repository format was
618
originally written to work with. This can be used if manually
619
constructing a bzrdir and repository, or more commonly for test suite
623
_default_format = None
624
"""The default format used for new repositories."""
627
"""The known formats."""
630
def find_format(klass, a_bzrdir):
631
"""Return the format for the repository object in a_bzrdir."""
633
transport = a_bzrdir.get_repository_transport(None)
634
format_string = transport.get("format").read()
635
return klass._formats[format_string]
636
except errors.NoSuchFile:
637
raise errors.NoRepositoryPresent(a_bzrdir)
639
raise errors.UnknownFormatError(format_string)
642
def get_default_format(klass):
643
"""Return the current default format."""
644
return klass._default_format
646
def get_format_string(self):
647
"""Return the ASCII format string that identifies this format.
649
Note that in pre format ?? repositories the format string is
650
not permitted nor written to disk.
652
raise NotImplementedError(self.get_format_string)
654
def _get_revision_store(self, repo_transport, control_files):
655
"""Return the revision store object for this a_bzrdir."""
656
raise NotImplementedError(self._get_revision_store)
658
def _get_rev_store(self,
664
"""Common logic for getting a revision store for a repository.
666
see self._get_revision_store for the method to
667
get the store for a repository.
670
name = safe_unicode(name)
673
dir_mode = control_files._dir_mode
674
file_mode = control_files._file_mode
675
revision_store =TextStore(transport.clone(name),
677
compressed=compressed,
680
revision_store.register_suffix('sig')
681
return revision_store
683
def initialize(self, a_bzrdir, shared=False):
684
"""Initialize a repository of this format in a_bzrdir.
686
:param a_bzrdir: The bzrdir to put the new repository in it.
687
:param shared: The repository should be initialized as a sharable one.
689
This may raise UninitializableFormat if shared repository are not
690
compatible the a_bzrdir.
693
def is_supported(self):
694
"""Is this format supported?
696
Supported formats must be initializable and openable.
697
Unsupported formats may not support initialization or committing or
698
some other features depending on the reason for not being supported.
702
def open(self, a_bzrdir, _found=False):
703
"""Return an instance of this format for the bzrdir a_bzrdir.
705
_found is a private parameter, do not use it.
707
raise NotImplementedError(self.open)
710
def register_format(klass, format):
711
klass._formats[format.get_format_string()] = format
714
def set_default_format(klass, format):
715
klass._default_format = format
718
def unregister_format(klass, format):
719
assert klass._formats[format.get_format_string()] is format
720
del klass._formats[format.get_format_string()]
723
class PreSplitOutRepositoryFormat(RepositoryFormat):
724
"""Base class for the pre split out repository formats."""
726
def initialize(self, a_bzrdir, shared=False, _internal=False):
727
"""Create a weave repository.
729
TODO: when creating split out bzr branch formats, move this to a common
730
base for Format5, Format6. or something like that.
732
from bzrlib.weavefile import write_weave_v5
733
from bzrlib.weave import Weave
736
raise errors.IncompatibleFormat(self, a_bzrdir._format)
739
# always initialized when the bzrdir is.
740
return self.open(a_bzrdir, _found=True)
742
# Create an empty weave
744
bzrlib.weavefile.write_weave_v5(Weave(), sio)
745
empty_weave = sio.getvalue()
747
mutter('creating repository in %s.', a_bzrdir.transport.base)
748
dirs = ['revision-store', 'weaves']
749
lock_file = 'branch-lock'
750
files = [('inventory.weave', StringIO(empty_weave)),
753
# FIXME: RBC 20060125 dont peek under the covers
754
# NB: no need to escape relative paths that are url safe.
755
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock')
756
control_files.lock_write()
757
control_files._transport.mkdir_multi(dirs,
758
mode=control_files._dir_mode)
760
for file, content in files:
761
control_files.put(file, content)
763
control_files.unlock()
764
return self.open(a_bzrdir, _found=True)
766
def open(self, a_bzrdir, _found=False):
767
"""See RepositoryFormat.open()."""
769
# we are being called directly and must probe.
770
raise NotImplementedError
772
repo_transport = a_bzrdir.get_repository_transport(None)
773
control_files = a_bzrdir._control_files
774
revision_store = self._get_revision_store(repo_transport, control_files)
775
return AllInOneRepository(_format=self,
777
revision_store=revision_store)
780
class RepositoryFormat4(PreSplitOutRepositoryFormat):
781
"""Bzr repository format 4.
783
This repository format has:
785
- TextStores for texts, inventories,revisions.
787
This format is deprecated: it indexes texts using a text id which is
788
removed in format 5; initializationa and write support for this format
793
super(RepositoryFormat4, self).__init__()
794
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
796
def initialize(self, url, shared=False, _internal=False):
797
"""Format 4 branches cannot be created."""
798
raise errors.UninitializableFormat(self)
800
def is_supported(self):
801
"""Format 4 is not supported.
803
It is not supported because the model changed from 4 to 5 and the
804
conversion logic is expensive - so doing it on the fly was not
809
def _get_revision_store(self, repo_transport, control_files):
810
"""See RepositoryFormat._get_revision_store()."""
811
return self._get_rev_store(repo_transport,
816
class RepositoryFormat5(PreSplitOutRepositoryFormat):
817
"""Bzr control format 5.
819
This repository format has:
820
- weaves for file texts and inventory
822
- TextStores for revisions and signatures.
826
super(RepositoryFormat5, self).__init__()
827
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
829
def _get_revision_store(self, repo_transport, control_files):
830
"""See RepositoryFormat._get_revision_store()."""
831
"""Return the revision store object for this a_bzrdir."""
832
return self._get_rev_store(repo_transport,
838
class RepositoryFormat6(PreSplitOutRepositoryFormat):
839
"""Bzr control format 6.
841
This repository format has:
842
- weaves for file texts and inventory
843
- hash subdirectory based stores.
844
- TextStores for revisions and signatures.
848
super(RepositoryFormat6, self).__init__()
849
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
851
def _get_revision_store(self, repo_transport, control_files):
852
"""See RepositoryFormat._get_revision_store()."""
853
return self._get_rev_store(repo_transport,
860
class MetaDirRepositoryFormat(RepositoryFormat):
861
"""Common base class for the new repositories using the metadir layour."""
864
super(MetaDirRepositoryFormat, self).__init__()
865
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
867
def _create_control_files(self, a_bzrdir):
868
"""Create the required files and the initial control_files object."""
869
# FIXME: RBC 20060125 dont peek under the covers
870
# NB: no need to escape relative paths that are url safe.
872
repository_transport = a_bzrdir.get_repository_transport(self)
873
repository_transport.put(lock_file, StringIO()) # TODO get the file mode from the bzrdir lock files., mode=file_mode)
874
control_files = LockableFiles(repository_transport, 'lock')
877
def _get_revision_store(self, repo_transport, control_files):
878
"""See RepositoryFormat._get_revision_store()."""
879
return self._get_rev_store(repo_transport,
886
def open(self, a_bzrdir, _found=False, _override_transport=None):
887
"""See RepositoryFormat.open().
889
:param _override_transport: INTERNAL USE ONLY. Allows opening the
890
repository at a slightly different url
891
than normal. I.e. during 'upgrade'.
894
format = RepositoryFormat.find_format(a_bzrdir)
895
assert format.__class__ == self.__class__
896
if _override_transport is not None:
897
repo_transport = _override_transport
899
repo_transport = a_bzrdir.get_repository_transport(None)
900
control_files = LockableFiles(repo_transport, 'lock')
901
revision_store = self._get_revision_store(repo_transport, control_files)
902
return MetaDirRepository(_format=self,
904
control_files=control_files,
905
revision_store=revision_store)
907
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
908
"""Upload the initial blank content."""
909
control_files = self._create_control_files(a_bzrdir)
910
control_files.lock_write()
911
control_files._transport.mkdir_multi(dirs,
912
mode=control_files._dir_mode)
914
for file, content in files:
915
control_files.put(file, content)
916
for file, content in utf8_files:
917
control_files.put_utf8(file, content)
919
control_files.put_utf8('shared-storage', '')
921
control_files.unlock()
924
class RepositoryFormat7(MetaDirRepositoryFormat):
927
This repository format has:
928
- weaves for file texts and inventory
929
- hash subdirectory based stores.
930
- TextStores for revisions and signatures.
931
- a format marker of its own
932
- an optional 'shared-storage' flag
933
- an optional 'no-working-trees' flag
936
def get_format_string(self):
937
"""See RepositoryFormat.get_format_string()."""
938
return "Bazaar-NG Repository format 7"
940
def initialize(self, a_bzrdir, shared=False):
941
"""Create a weave repository.
943
:param shared: If true the repository will be initialized as a shared
946
from bzrlib.weavefile import write_weave_v5
947
from bzrlib.weave import Weave
949
# Create an empty weave
951
bzrlib.weavefile.write_weave_v5(Weave(), sio)
952
empty_weave = sio.getvalue()
954
mutter('creating repository in %s.', a_bzrdir.transport.base)
955
dirs = ['revision-store', 'weaves']
956
files = [('inventory.weave', StringIO(empty_weave)),
958
utf8_files = [('format', self.get_format_string())]
960
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
961
return self.open(a_bzrdir=a_bzrdir, _found=True)
964
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
965
"""Bzr repository knit format 1.
967
This repository format has:
968
- knits for file texts and inventory
969
- hash subdirectory based stores.
970
- knits for revisions and signatures
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 Knit Repository Format 1"
981
def initialize(self, a_bzrdir, shared=False):
982
"""Create a knit format 1 repository.
984
:param shared: If true the repository will be initialized as a shared
986
XXX NOTE that this current uses a Weave for testing and will become
987
A Knit in due course.
989
from bzrlib.weavefile import write_weave_v5
990
from bzrlib.weave import Weave
992
# Create an empty weave
994
bzrlib.weavefile.write_weave_v5(Weave(), sio)
995
empty_weave = sio.getvalue()
997
mutter('creating repository in %s.', a_bzrdir.transport.base)
998
dirs = ['revision-store', 'knits']
999
files = [('inventory.weave', StringIO(empty_weave)),
1001
utf8_files = [('format', self.get_format_string())]
1003
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1004
return self.open(a_bzrdir=a_bzrdir, _found=True)
1007
# formats which have no format string are not discoverable
1008
# and not independently creatable, so are not registered.
1009
_default_format = RepositoryFormat7()
1010
RepositoryFormat.register_format(_default_format)
1011
RepositoryFormat.register_format(RepositoryFormatKnit1())
1012
RepositoryFormat.set_default_format(_default_format)
1013
_legacy_formats = [RepositoryFormat4(),
1014
RepositoryFormat5(),
1015
RepositoryFormat6()]
1018
class InterRepository(object):
1019
"""This class represents operations taking place between two repositories.
1021
Its instances have methods like copy_content and fetch, and contain
1022
references to the source and target repositories these operations can be
1025
Often we will provide convenience methods on 'repository' which carry out
1026
operations with another repository - they will always forward to
1027
InterRepository.get(other).method_name(parameters).
1029
# XXX: FIXME: FUTURE: robertc
1030
# testing of these probably requires a factory in optimiser type, and
1031
# then a test adapter to test each type thoroughly.
1035
"""The available optimised InterRepository types."""
1037
def __init__(self, source, target):
1038
"""Construct a default InterRepository instance. Please use 'get'.
1040
Only subclasses of InterRepository should call
1041
InterRepository.__init__ - clients should call InterRepository.get
1042
instead which will create an optimised InterRepository if possible.
1044
self.source = source
1045
self.target = target
1048
def copy_content(self, revision_id=None, basis=None):
1049
"""Make a complete copy of the content in self into destination.
1051
This is a destructive operation! Do not use it on existing
1054
:param revision_id: Only copy the content needed to construct
1055
revision_id and its parents.
1056
:param basis: Copy the needed data preferentially from basis.
1059
self.target.set_make_working_trees(self.source.make_working_trees())
1060
except NotImplementedError:
1062
# grab the basis available data
1063
if basis is not None:
1064
self.target.fetch(basis, revision_id=revision_id)
1065
# but dont both fetching if we have the needed data now.
1066
if (revision_id not in (None, NULL_REVISION) and
1067
self.target.has_revision(revision_id)):
1069
self.target.fetch(self.source, revision_id=revision_id)
1071
def _double_lock(self, lock_source, lock_target):
1072
"""Take out too locks, rolling back the first if the second throws."""
1077
# we want to ensure that we don't leave source locked by mistake.
1078
# and any error on target should not confuse source.
1079
self.source.unlock()
1083
def fetch(self, revision_id=None, pb=None):
1084
"""Fetch the content required to construct revision_id.
1086
The content is copied from source to target.
1088
:param revision_id: if None all content is copied, if NULL_REVISION no
1090
:param pb: optional progress bar to use for progress reports. If not
1091
provided a default one will be created.
1093
Returns the copied revision count and the failed revisions in a tuple:
1096
from bzrlib.fetch import RepoFetcher
1097
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1098
self.source, self.source._format, self.target, self.target._format)
1099
f = RepoFetcher(to_repository=self.target,
1100
from_repository=self.source,
1101
last_revision=revision_id,
1103
return f.count_copied, f.failed_revisions
1106
def get(klass, repository_source, repository_target):
1107
"""Retrieve a InterRepository worker object for these repositories.
1109
:param repository_source: the repository to be the 'source' member of
1110
the InterRepository instance.
1111
:param repository_target: the repository to be the 'target' member of
1112
the InterRepository instance.
1113
If an optimised InterRepository worker exists it will be used otherwise
1114
a default InterRepository instance will be created.
1116
for provider in klass._optimisers:
1117
if provider.is_compatible(repository_source, repository_target):
1118
return provider(repository_source, repository_target)
1119
return InterRepository(repository_source, repository_target)
1121
def lock_read(self):
1122
"""Take out a logical read lock.
1124
This will lock the source branch and the target branch. The source gets
1125
a read lock and the target a read lock.
1127
self._double_lock(self.source.lock_read, self.target.lock_read)
1129
def lock_write(self):
1130
"""Take out a logical write lock.
1132
This will lock the source branch and the target branch. The source gets
1133
a read lock and the target a write lock.
1135
self._double_lock(self.source.lock_read, self.target.lock_write)
1138
def missing_revision_ids(self, revision_id=None):
1139
"""Return the revision ids that source has that target does not.
1141
These are returned in topological order.
1143
:param revision_id: only return revision ids included by this
1146
# generic, possibly worst case, slow code path.
1147
target_ids = set(self.target.all_revision_ids())
1148
if revision_id is not None:
1149
source_ids = self.source.get_ancestry(revision_id)
1150
assert source_ids.pop(0) == None
1152
source_ids = self.source.all_revision_ids()
1153
result_set = set(source_ids).difference(target_ids)
1154
# this may look like a no-op: its not. It preserves the ordering
1155
# other_ids had while only returning the members from other_ids
1156
# that we've decided we need.
1157
return [rev_id for rev_id in source_ids if rev_id in result_set]
1160
def register_optimiser(klass, optimiser):
1161
"""Register an InterRepository optimiser."""
1162
klass._optimisers.add(optimiser)
1165
"""Release the locks on source and target."""
1167
self.target.unlock()
1169
self.source.unlock()
1172
def unregister_optimiser(klass, optimiser):
1173
"""Unregister an InterRepository optimiser."""
1174
klass._optimisers.remove(optimiser)
1177
class InterWeaveRepo(InterRepository):
1178
"""Optimised code paths between Weave based repositories."""
1180
_matching_repo_format = _default_format
1181
"""Repository format for testing with."""
1184
def is_compatible(source, target):
1185
"""Be compatible with known Weave formats.
1187
We dont test for the stores being of specific types becase that
1188
could lead to confusing results, and there is no need to be
1192
return (isinstance(source._format, (RepositoryFormat5,
1194
RepositoryFormat7)) and
1195
isinstance(target._format, (RepositoryFormat5,
1197
RepositoryFormat7)))
1198
except AttributeError:
1202
def copy_content(self, revision_id=None, basis=None):
1203
"""See InterRepository.copy_content()."""
1204
# weave specific optimised path:
1205
if basis is not None:
1206
# copy the basis in, then fetch remaining data.
1207
basis.copy_content_into(self.target, revision_id)
1208
# the basis copy_content_into could misset this.
1210
self.target.set_make_working_trees(self.source.make_working_trees())
1211
except NotImplementedError:
1213
self.target.fetch(self.source, revision_id=revision_id)
1216
self.target.set_make_working_trees(self.source.make_working_trees())
1217
except NotImplementedError:
1219
# FIXME do not peek!
1220
if self.source.control_files._transport.listable():
1221
pb = bzrlib.ui.ui_factory.progress_bar()
1222
copy_all(self.source.weave_store,
1223
self.target.weave_store, pb=pb)
1224
pb.update('copying inventory', 0, 1)
1225
self.target.control_weaves.copy_multi(
1226
self.source.control_weaves, ['inventory'])
1227
copy_all(self.source.revision_store,
1228
self.target.revision_store, pb=pb)
1230
self.target.fetch(self.source, revision_id=revision_id)
1233
def fetch(self, revision_id=None, pb=None):
1234
"""See InterRepository.fetch()."""
1235
from bzrlib.fetch import RepoFetcher
1236
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1237
self.source, self.source._format, self.target, self.target._format)
1238
f = RepoFetcher(to_repository=self.target,
1239
from_repository=self.source,
1240
last_revision=revision_id,
1242
return f.count_copied, f.failed_revisions
1245
def missing_revision_ids(self, revision_id=None):
1246
"""See InterRepository.missing_revision_ids()."""
1247
# we want all revisions to satisfy revision_id in source.
1248
# but we dont want to stat every file here and there.
1249
# we want then, all revisions other needs to satisfy revision_id
1250
# checked, but not those that we have locally.
1251
# so the first thing is to get a subset of the revisions to
1252
# satisfy revision_id in source, and then eliminate those that
1253
# we do already have.
1254
# this is slow on high latency connection to self, but as as this
1255
# disk format scales terribly for push anyway due to rewriting
1256
# inventory.weave, this is considered acceptable.
1258
if revision_id is not None:
1259
source_ids = self.source.get_ancestry(revision_id)
1260
assert source_ids.pop(0) == None
1262
source_ids = self.source._all_possible_ids()
1263
source_ids_set = set(source_ids)
1264
# source_ids is the worst possible case we may need to pull.
1265
# now we want to filter source_ids against what we actually
1266
# have in target, but dont try to check for existence where we know
1267
# we do not have a revision as that would be pointless.
1268
target_ids = set(self.target._all_possible_ids())
1269
possibly_present_revisions = target_ids.intersection(source_ids_set)
1270
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1271
required_revisions = source_ids_set.difference(actually_present_revisions)
1272
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1273
if revision_id is not None:
1274
# we used get_ancestry to determine source_ids then we are assured all
1275
# revisions referenced are present as they are installed in topological order.
1276
# and the tip revision was validated by get_ancestry.
1277
return required_topo_revisions
1279
# if we just grabbed the possibly available ids, then
1280
# we only have an estimate of whats available and need to validate
1281
# that against the revision records.
1282
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1285
InterRepository.register_optimiser(InterWeaveRepo)
1288
class RepositoryTestProviderAdapter(object):
1289
"""A tool to generate a suite testing multiple repository formats at once.
1291
This is done by copying the test once for each transport and injecting
1292
the transport_server, transport_readonly_server, and bzrdir_format and
1293
repository_format classes into each copy. Each copy is also given a new id()
1294
to make it easy to identify.
1297
def __init__(self, transport_server, transport_readonly_server, formats):
1298
self._transport_server = transport_server
1299
self._transport_readonly_server = transport_readonly_server
1300
self._formats = formats
1302
def adapt(self, test):
1303
result = TestSuite()
1304
for repository_format, bzrdir_format in self._formats:
1305
new_test = deepcopy(test)
1306
new_test.transport_server = self._transport_server
1307
new_test.transport_readonly_server = self._transport_readonly_server
1308
new_test.bzrdir_format = bzrdir_format
1309
new_test.repository_format = repository_format
1310
def make_new_test_id():
1311
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1312
return lambda: new_id
1313
new_test.id = make_new_test_id()
1314
result.addTest(new_test)
1318
class InterRepositoryTestProviderAdapter(object):
1319
"""A tool to generate a suite testing multiple inter repository formats.
1321
This is done by copying the test once for each interrepo provider and injecting
1322
the transport_server, transport_readonly_server, repository_format and
1323
repository_to_format classes into each copy.
1324
Each copy is also given a new id() to make it easy to identify.
1327
def __init__(self, transport_server, transport_readonly_server, formats):
1328
self._transport_server = transport_server
1329
self._transport_readonly_server = transport_readonly_server
1330
self._formats = formats
1332
def adapt(self, test):
1333
result = TestSuite()
1334
for interrepo_class, repository_format, repository_format_to in self._formats:
1335
new_test = deepcopy(test)
1336
new_test.transport_server = self._transport_server
1337
new_test.transport_readonly_server = self._transport_readonly_server
1338
new_test.interrepo_class = interrepo_class
1339
new_test.repository_format = repository_format
1340
new_test.repository_format_to = repository_format_to
1341
def make_new_test_id():
1342
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1343
return lambda: new_id
1344
new_test.id = make_new_test_id()
1345
result.addTest(new_test)
1349
def default_test_list():
1350
"""Generate the default list of interrepo permutations to test."""
1352
# test the default InterRepository between format 6 and the current
1354
# XXX: robertc 20060220 reinstate this when there are two supported
1355
# formats which do not have an optimal code path between them.
1356
result.append((InterRepository,
1357
RepositoryFormat6(),
1358
RepositoryFormatKnit1()))
1359
for optimiser in InterRepository._optimisers:
1360
result.append((optimiser,
1361
optimiser._matching_repo_format,
1362
optimiser._matching_repo_format
1364
# if there are specific combinations we want to use, we can add them
1369
class CopyConverter(object):
1370
"""A repository conversion tool which just performs a copy of the content.
1372
This is slow but quite reliable.
1375
def __init__(self, target_format):
1376
"""Create a CopyConverter.
1378
:param target_format: The format the resulting repository should be.
1380
self.target_format = target_format
1382
def convert(self, repo, pb):
1383
"""Perform the conversion of to_convert, giving feedback via pb.
1385
:param to_convert: The disk object to convert.
1386
:param pb: a progress bar to use for progress information.
1391
# this is only useful with metadir layouts - separated repo content.
1392
# trigger an assertion if not such
1393
repo._format.get_format_string()
1394
self.repo_dir = repo.bzrdir
1395
self.step('Moving repository to repository.backup')
1396
self.repo_dir.transport.move('repository', 'repository.backup')
1397
backup_transport = self.repo_dir.transport.clone('repository.backup')
1398
self.source_repo = repo._format.open(self.repo_dir,
1400
_override_transport=backup_transport)
1401
self.step('Creating new repository')
1402
converted = self.target_format.initialize(self.repo_dir,
1403
self.source_repo.is_shared())
1404
converted.lock_write()
1406
self.step('Copying content into repository.')
1407
self.source_repo.copy_content_into(converted)
1410
self.step('Deleting old repository content.')
1411
self.repo_dir.transport.delete_tree('repository.backup')
1412
self.pb.note('repository converted')
1414
def step(self, message):
1415
"""Update the pb by a step."""
1417
self.pb.update(message, self.count, self.total)
added
removed
bzrlib/repository.py
