1
# Copyright (C) 2005, 2006, 2007 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 cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
38
revision as _mod_revision,
43
from bzrlib.bundle import serializer
44
from bzrlib.revisiontree import RevisionTree
45
from bzrlib.store.versioned import VersionedFileStore
46
from bzrlib.store.text import TextStore
47
from bzrlib.testament import Testament
51
from bzrlib.decorators import needs_read_lock, needs_write_lock
52
from bzrlib.inter import InterObject
53
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
54
from bzrlib.symbol_versioning import (
58
from bzrlib.trace import mutter, note, warning
61
# Old formats display a warning, but only once
62
_deprecation_warning_done = False
65
######################################################################
68
class Repository(object):
69
"""Repository holding history for one or more branches.
71
The repository holds and retrieves historical information including
72
revisions and file history. It's normally accessed only by the Branch,
73
which views a particular line of development through that history.
75
The Repository builds on top of Stores and a Transport, which respectively
76
describe the disk data format and the way of accessing the (possibly
80
_file_ids_altered_regex = lazy_regex.lazy_compile(
81
r'file_id="(?P<file_id>[^"]+)"'
82
r'.*revision="(?P<revision_id>[^"]+)"'
85
def abort_write_group(self):
86
"""Commit the contents accrued within the current write group.
88
:seealso: start_write_group.
90
if self._write_group is not self.get_transaction():
91
# has an unlock or relock occured ?
92
raise errors.BzrError('mismatched lock context and write group.')
93
self._abort_write_group()
94
self._write_group = None
96
def _abort_write_group(self):
97
"""Template method for per-repository write group cleanup.
99
This is called during abort before the write group is considered to be
100
finished and should cleanup any internal state accrued during the write
101
group. There is no requirement that data handed to the repository be
102
*not* made available - this is not a rollback - but neither should any
103
attempt be made to ensure that data added is fully commited. Abort is
104
invoked when an error has occured so futher disk or network operations
105
may not be possible or may error and if possible should not be
110
def add_inventory(self, revision_id, inv, parents):
111
"""Add the inventory inv to the repository as revision_id.
113
:param parents: The revision ids of the parents that revision_id
114
is known to have and are in the repository already.
116
returns the sha1 of the serialized inventory.
118
revision_id = osutils.safe_revision_id(revision_id)
119
_mod_revision.check_not_reserved_id(revision_id)
120
assert inv.revision_id is None or inv.revision_id == revision_id, \
121
"Mismatch between inventory revision" \
122
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
123
assert inv.root is not None
124
inv_text = self.serialise_inventory(inv)
125
inv_sha1 = osutils.sha_string(inv_text)
126
inv_vf = self.control_weaves.get_weave('inventory',
127
self.get_transaction())
128
self._inventory_add_lines(inv_vf, revision_id, parents,
129
osutils.split_lines(inv_text))
132
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
134
for parent in parents:
136
final_parents.append(parent)
138
inv_vf.add_lines(revision_id, final_parents, lines)
141
def add_revision(self, revision_id, rev, inv=None, config=None):
142
"""Add rev to the revision store as revision_id.
144
:param revision_id: the revision id to use.
145
:param rev: The revision object.
146
:param inv: The inventory for the revision. if None, it will be looked
147
up in the inventory storer
148
:param config: If None no digital signature will be created.
149
If supplied its signature_needed method will be used
150
to determine if a signature should be made.
152
revision_id = osutils.safe_revision_id(revision_id)
153
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
155
_mod_revision.check_not_reserved_id(revision_id)
156
if config is not None and config.signature_needed():
158
inv = self.get_inventory(revision_id)
159
plaintext = Testament(rev, inv).as_short_text()
160
self.store_revision_signature(
161
gpg.GPGStrategy(config), plaintext, revision_id)
162
if not revision_id in self.get_inventory_weave():
164
raise errors.WeaveRevisionNotPresent(revision_id,
165
self.get_inventory_weave())
167
# yes, this is not suitable for adding with ghosts.
168
self.add_inventory(revision_id, inv, rev.parent_ids)
169
self._revision_store.add_revision(rev, self.get_transaction())
171
def _add_revision_text(self, revision_id, text):
172
revision = self._revision_store._serializer.read_revision_from_string(
174
self._revision_store._add_revision(revision, StringIO(text),
175
self.get_transaction())
178
def _all_possible_ids(self):
179
"""Return all the possible revisions that we could find."""
180
return self.get_inventory_weave().versions()
182
def all_revision_ids(self):
183
"""Returns a list of all the revision ids in the repository.
185
This is deprecated because code should generally work on the graph
186
reachable from a particular revision, and ignore any other revisions
187
that might be present. There is no direct replacement method.
189
return self._all_revision_ids()
192
def _all_revision_ids(self):
193
"""Returns a list of all the revision ids in the repository.
195
These are in as much topological order as the underlying store can
196
present: for weaves ghosts may lead to a lack of correctness until
197
the reweave updates the parents list.
199
if self._revision_store.text_store.listable():
200
return self._revision_store.all_revision_ids(self.get_transaction())
201
result = self._all_possible_ids()
202
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
203
# ids. (It should, since _revision_store's API should change to
204
# return utf8 revision_ids)
205
return self._eliminate_revisions_not_present(result)
207
def break_lock(self):
208
"""Break a lock if one is present from another instance.
210
Uses the ui factory to ask for confirmation if the lock may be from
213
self.control_files.break_lock()
216
def _eliminate_revisions_not_present(self, revision_ids):
217
"""Check every revision id in revision_ids to see if we have it.
219
Returns a set of the present revisions.
222
for id in revision_ids:
223
if self.has_revision(id):
228
def create(a_bzrdir):
229
"""Construct the current default format repository in a_bzrdir."""
230
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
232
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
233
"""instantiate a Repository.
235
:param _format: The format of the repository on disk.
236
:param a_bzrdir: The BzrDir of the repository.
238
In the future we will have a single api for all stores for
239
getting file texts, inventories and revisions, then
240
this construct will accept instances of those things.
242
super(Repository, self).__init__()
243
self._format = _format
244
# the following are part of the public API for Repository:
245
self.bzrdir = a_bzrdir
246
self.control_files = control_files
247
self._revision_store = _revision_store
248
self.text_store = text_store
249
# backwards compatibility
250
self.weave_store = text_store
251
# not right yet - should be more semantically clear ?
253
self.control_store = control_store
254
self.control_weaves = control_store
255
# TODO: make sure to construct the right store classes, etc, depending
256
# on whether escaping is required.
257
self._warn_if_deprecated()
258
self._write_group = None
261
return '%s(%r)' % (self.__class__.__name__,
262
self.bzrdir.transport.base)
264
def is_in_write_group(self):
265
"""Return True if there is an open write group.
267
:seealso: start_write_group.
269
return self._write_group is not None
272
return self.control_files.is_locked()
274
def lock_write(self, token=None):
275
"""Lock this repository for writing.
277
This causes caching within the repository obejct to start accumlating
278
data during reads, and allows a 'write_group' to be obtained. Write
279
groups must be used for actual data insertion.
281
:param token: if this is already locked, then lock_write will fail
282
unless the token matches the existing lock.
283
:returns: a token if this instance supports tokens, otherwise None.
284
:raises TokenLockingNotSupported: when a token is given but this
285
instance doesn't support using token locks.
286
:raises MismatchedToken: if the specified token doesn't match the token
287
of the existing lock.
288
:seealso: start_write_group.
290
A token should be passed in if you know that you have locked the object
291
some other way, and need to synchronise this object's state with that
294
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
296
result = self.control_files.lock_write(token=token)
301
self.control_files.lock_read()
304
def get_physical_lock_status(self):
305
return self.control_files.get_physical_lock_status()
307
def leave_lock_in_place(self):
308
"""Tell this repository not to release the physical lock when this
311
If lock_write doesn't return a token, then this method is not supported.
313
self.control_files.leave_in_place()
315
def dont_leave_lock_in_place(self):
316
"""Tell this repository to release the physical lock when this
317
object is unlocked, even if it didn't originally acquire it.
319
If lock_write doesn't return a token, then this method is not supported.
321
self.control_files.dont_leave_in_place()
324
def gather_stats(self, revid=None, committers=None):
325
"""Gather statistics from a revision id.
327
:param revid: The revision id to gather statistics from, if None, then
328
no revision specific statistics are gathered.
329
:param committers: Optional parameter controlling whether to grab
330
a count of committers from the revision specific statistics.
331
:return: A dictionary of statistics. Currently this contains:
332
committers: The number of committers if requested.
333
firstrev: A tuple with timestamp, timezone for the penultimate left
334
most ancestor of revid, if revid is not the NULL_REVISION.
335
latestrev: A tuple with timestamp, timezone for revid, if revid is
336
not the NULL_REVISION.
337
revisions: The total revision count in the repository.
338
size: An estimate disk size of the repository in bytes.
341
if revid and committers:
342
result['committers'] = 0
343
if revid and revid != _mod_revision.NULL_REVISION:
345
all_committers = set()
346
revisions = self.get_ancestry(revid)
347
# pop the leading None
349
first_revision = None
351
# ignore the revisions in the middle - just grab first and last
352
revisions = revisions[0], revisions[-1]
353
for revision in self.get_revisions(revisions):
354
if not first_revision:
355
first_revision = revision
357
all_committers.add(revision.committer)
358
last_revision = revision
360
result['committers'] = len(all_committers)
361
result['firstrev'] = (first_revision.timestamp,
362
first_revision.timezone)
363
result['latestrev'] = (last_revision.timestamp,
364
last_revision.timezone)
366
# now gather global repository information
367
if self.bzrdir.root_transport.listable():
368
c, t = self._revision_store.total_size(self.get_transaction())
369
result['revisions'] = c
374
def missing_revision_ids(self, other, revision_id=None):
375
"""Return the revision ids that other has that this does not.
377
These are returned in topological order.
379
revision_id: only return revision ids included by revision_id.
381
revision_id = osutils.safe_revision_id(revision_id)
382
return InterRepository.get(other, self).missing_revision_ids(revision_id)
386
"""Open the repository rooted at base.
388
For instance, if the repository is at URL/.bzr/repository,
389
Repository.open(URL) -> a Repository instance.
391
control = bzrdir.BzrDir.open(base)
392
return control.open_repository()
394
def copy_content_into(self, destination, revision_id=None):
395
"""Make a complete copy of the content in self into destination.
397
This is a destructive operation! Do not use it on existing
400
revision_id = osutils.safe_revision_id(revision_id)
401
return InterRepository.get(self, destination).copy_content(revision_id)
403
def commit_write_group(self):
404
"""Commit the contents accrued within the current write group.
406
:seealso: start_write_group.
408
if self._write_group is not self.get_transaction():
409
# has an unlock or relock occured ?
410
raise errors.BzrError('mismatched lock context and write group.')
411
self._commit_write_group()
412
self._write_group = None
414
def _commit_write_group(self):
415
"""Template method for per-repository write group cleanup.
417
This is called before the write group is considered to be
418
finished and should ensure that all data handed to the repository
419
for writing during the write group is safely committed (to the
420
extent possible considering file system caching etc).
423
def fetch(self, source, revision_id=None, pb=None):
424
"""Fetch the content required to construct revision_id from source.
426
If revision_id is None all content is copied.
428
revision_id = osutils.safe_revision_id(revision_id)
429
inter = InterRepository.get(source, self)
431
return inter.fetch(revision_id=revision_id, pb=pb)
432
except NotImplementedError:
433
raise errors.IncompatibleRepositories(source, self)
435
def create_bundle(self, target, base, fileobj, format=None):
436
return serializer.write_bundle(self, target, base, fileobj, format)
438
def get_commit_builder(self, branch, parents, config, timestamp=None,
439
timezone=None, committer=None, revprops=None,
441
"""Obtain a CommitBuilder for this repository.
443
:param branch: Branch to commit to.
444
:param parents: Revision ids of the parents of the new revision.
445
:param config: Configuration to use.
446
:param timestamp: Optional timestamp recorded for commit.
447
:param timezone: Optional timezone for timestamp.
448
:param committer: Optional committer to set for commit.
449
:param revprops: Optional dictionary of revision properties.
450
:param revision_id: Optional revision id.
452
revision_id = osutils.safe_revision_id(revision_id)
453
result =_CommitBuilder(self, parents, config, timestamp, timezone,
454
committer, revprops, revision_id)
455
self.start_write_group()
459
if (self.control_files._lock_count == 1 and
460
self.control_files._lock_mode == 'w'):
461
if self._write_group is not None:
462
raise errors.BzrError(
463
'Must end write groups before releasing write locks.')
464
self.control_files.unlock()
467
def clone(self, a_bzrdir, revision_id=None):
468
"""Clone this repository into a_bzrdir using the current format.
470
Currently no check is made that the format of this repository and
471
the bzrdir format are compatible. FIXME RBC 20060201.
473
:return: The newly created destination repository.
475
# TODO: deprecate after 0.16; cloning this with all its settings is
476
# probably not very useful -- mbp 20070423
477
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
478
self.copy_content_into(dest_repo, revision_id)
481
def start_write_group(self):
482
"""Start a write group in the repository.
484
Write groups are used by repositories which do not have a 1:1 mapping
485
between file ids and backend store to manage the insertion of data from
486
both fetch and commit operations.
488
A write lock is required around the start_write_group/commit_write_group
489
for the support of lock-requiring repository formats.
491
One can only insert data into a repository inside a write group.
495
if not self.is_locked() or self.control_files._lock_mode != 'w':
496
raise errors.NotWriteLocked(self)
497
if self._write_group:
498
raise errors.BzrError('already in a write group')
499
self._start_write_group()
500
# so we can detect unlock/relock - the write group is now entered.
501
self._write_group = self.get_transaction()
503
def _start_write_group(self):
504
"""Template method for per-repository write group startup.
506
This is called before the write group is considered to be
511
def sprout(self, to_bzrdir, revision_id=None):
512
"""Create a descendent repository for new development.
514
Unlike clone, this does not copy the settings of the repository.
516
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
517
dest_repo.fetch(self, revision_id=revision_id)
520
def _create_sprouting_repo(self, a_bzrdir, shared):
521
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
522
# use target default format.
523
dest_repo = a_bzrdir.create_repository()
525
# Most control formats need the repository to be specifically
526
# created, but on some old all-in-one formats it's not needed
528
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
529
except errors.UninitializableFormat:
530
dest_repo = a_bzrdir.open_repository()
534
def has_revision(self, revision_id):
535
"""True if this repository has a copy of the revision."""
536
revision_id = osutils.safe_revision_id(revision_id)
537
return self._revision_store.has_revision_id(revision_id,
538
self.get_transaction())
541
def get_revision_reconcile(self, revision_id):
542
"""'reconcile' helper routine that allows access to a revision always.
544
This variant of get_revision does not cross check the weave graph
545
against the revision one as get_revision does: but it should only
546
be used by reconcile, or reconcile-alike commands that are correcting
547
or testing the revision graph.
549
if not revision_id or not isinstance(revision_id, basestring):
550
raise errors.InvalidRevisionId(revision_id=revision_id,
552
return self.get_revisions([revision_id])[0]
555
def get_revisions(self, revision_ids):
556
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
557
revs = self._revision_store.get_revisions(revision_ids,
558
self.get_transaction())
560
assert not isinstance(rev.revision_id, unicode)
561
for parent_id in rev.parent_ids:
562
assert not isinstance(parent_id, unicode)
566
def get_revision_xml(self, revision_id):
567
# TODO: jam 20070210 This shouldn't be necessary since get_revision
568
# would have already do it.
569
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
570
revision_id = osutils.safe_revision_id(revision_id)
571
rev = self.get_revision(revision_id)
573
# the current serializer..
574
self._revision_store._serializer.write_revision(rev, rev_tmp)
576
return rev_tmp.getvalue()
579
def get_revision(self, revision_id):
580
"""Return the Revision object for a named revision"""
581
# TODO: jam 20070210 get_revision_reconcile should do this for us
582
revision_id = osutils.safe_revision_id(revision_id)
583
r = self.get_revision_reconcile(revision_id)
584
# weave corruption can lead to absent revision markers that should be
586
# the following test is reasonably cheap (it needs a single weave read)
587
# and the weave is cached in read transactions. In write transactions
588
# it is not cached but typically we only read a small number of
589
# revisions. For knits when they are introduced we will probably want
590
# to ensure that caching write transactions are in use.
591
inv = self.get_inventory_weave()
592
self._check_revision_parents(r, inv)
596
def get_deltas_for_revisions(self, revisions):
597
"""Produce a generator of revision deltas.
599
Note that the input is a sequence of REVISIONS, not revision_ids.
600
Trees will be held in memory until the generator exits.
601
Each delta is relative to the revision's lefthand predecessor.
603
required_trees = set()
604
for revision in revisions:
605
required_trees.add(revision.revision_id)
606
required_trees.update(revision.parent_ids[:1])
607
trees = dict((t.get_revision_id(), t) for
608
t in self.revision_trees(required_trees))
609
for revision in revisions:
610
if not revision.parent_ids:
611
old_tree = self.revision_tree(None)
613
old_tree = trees[revision.parent_ids[0]]
614
yield trees[revision.revision_id].changes_from(old_tree)
617
def get_revision_delta(self, revision_id):
618
"""Return the delta for one revision.
620
The delta is relative to the left-hand predecessor of the
623
r = self.get_revision(revision_id)
624
return list(self.get_deltas_for_revisions([r]))[0]
626
def _check_revision_parents(self, revision, inventory):
627
"""Private to Repository and Fetch.
629
This checks the parentage of revision in an inventory weave for
630
consistency and is only applicable to inventory-weave-for-ancestry
631
using repository formats & fetchers.
633
weave_parents = inventory.get_parents(revision.revision_id)
634
weave_names = inventory.versions()
635
for parent_id in revision.parent_ids:
636
if parent_id in weave_names:
637
# this parent must not be a ghost.
638
if not parent_id in weave_parents:
640
raise errors.CorruptRepository(self)
643
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
644
revision_id = osutils.safe_revision_id(revision_id)
645
signature = gpg_strategy.sign(plaintext)
646
self._revision_store.add_revision_signature_text(revision_id,
648
self.get_transaction())
650
def fileids_altered_by_revision_ids(self, revision_ids):
651
"""Find the file ids and versions affected by revisions.
653
:param revisions: an iterable containing revision ids.
654
:return: a dictionary mapping altered file-ids to an iterable of
655
revision_ids. Each altered file-ids has the exact revision_ids that
656
altered it listed explicitly.
658
assert self._serializer.support_altered_by_hack, \
659
("fileids_altered_by_revision_ids only supported for branches "
660
"which store inventory as unnested xml, not on %r" % self)
661
selected_revision_ids = set(osutils.safe_revision_id(r)
662
for r in revision_ids)
663
w = self.get_inventory_weave()
666
# this code needs to read every new line in every inventory for the
667
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
668
# not present in one of those inventories is unnecessary but not
669
# harmful because we are filtering by the revision id marker in the
670
# inventory lines : we only select file ids altered in one of those
671
# revisions. We don't need to see all lines in the inventory because
672
# only those added in an inventory in rev X can contain a revision=X
674
unescape_revid_cache = {}
675
unescape_fileid_cache = {}
677
# jam 20061218 In a big fetch, this handles hundreds of thousands
678
# of lines, so it has had a lot of inlining and optimizing done.
679
# Sorry that it is a little bit messy.
680
# Move several functions to be local variables, since this is a long
682
search = self._file_ids_altered_regex.search
683
unescape = _unescape_xml
684
setdefault = result.setdefault
685
pb = ui.ui_factory.nested_progress_bar()
687
for line in w.iter_lines_added_or_present_in_versions(
688
selected_revision_ids, pb=pb):
692
# One call to match.group() returning multiple items is quite a
693
# bit faster than 2 calls to match.group() each returning 1
694
file_id, revision_id = match.group('file_id', 'revision_id')
696
# Inlining the cache lookups helps a lot when you make 170,000
697
# lines and 350k ids, versus 8.4 unique ids.
698
# Using a cache helps in 2 ways:
699
# 1) Avoids unnecessary decoding calls
700
# 2) Re-uses cached strings, which helps in future set and
702
# (2) is enough that removing encoding entirely along with
703
# the cache (so we are using plain strings) results in no
704
# performance improvement.
706
revision_id = unescape_revid_cache[revision_id]
708
unescaped = unescape(revision_id)
709
unescape_revid_cache[revision_id] = unescaped
710
revision_id = unescaped
712
if revision_id in selected_revision_ids:
714
file_id = unescape_fileid_cache[file_id]
716
unescaped = unescape(file_id)
717
unescape_fileid_cache[file_id] = unescaped
719
setdefault(file_id, set()).add(revision_id)
725
def get_inventory_weave(self):
726
return self.control_weaves.get_weave('inventory',
727
self.get_transaction())
730
def get_inventory(self, revision_id):
731
"""Get Inventory object by hash."""
732
# TODO: jam 20070210 Technically we don't need to sanitize, since all
733
# called functions must sanitize.
734
revision_id = osutils.safe_revision_id(revision_id)
735
return self.deserialise_inventory(
736
revision_id, self.get_inventory_xml(revision_id))
738
def deserialise_inventory(self, revision_id, xml):
739
"""Transform the xml into an inventory object.
741
:param revision_id: The expected revision id of the inventory.
742
:param xml: A serialised inventory.
744
revision_id = osutils.safe_revision_id(revision_id)
745
result = self._serializer.read_inventory_from_string(xml)
746
result.root.revision = revision_id
749
def serialise_inventory(self, inv):
750
return self._serializer.write_inventory_to_string(inv)
752
def get_serializer_format(self):
753
return self._serializer.format_num
756
def get_inventory_xml(self, revision_id):
757
"""Get inventory XML as a file object."""
758
revision_id = osutils.safe_revision_id(revision_id)
760
assert isinstance(revision_id, str), type(revision_id)
761
iw = self.get_inventory_weave()
762
return iw.get_text(revision_id)
764
raise errors.HistoryMissing(self, 'inventory', revision_id)
767
def get_inventory_sha1(self, revision_id):
768
"""Return the sha1 hash of the inventory entry
770
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
771
revision_id = osutils.safe_revision_id(revision_id)
772
return self.get_revision(revision_id).inventory_sha1
775
def get_revision_graph(self, revision_id=None):
776
"""Return a dictionary containing the revision graph.
778
:param revision_id: The revision_id to get a graph from. If None, then
779
the entire revision graph is returned. This is a deprecated mode of
780
operation and will be removed in the future.
781
:return: a dictionary of revision_id->revision_parents_list.
783
# special case NULL_REVISION
784
if revision_id == _mod_revision.NULL_REVISION:
786
revision_id = osutils.safe_revision_id(revision_id)
787
a_weave = self.get_inventory_weave()
788
all_revisions = self._eliminate_revisions_not_present(
790
entire_graph = dict([(node, a_weave.get_parents(node)) for
791
node in all_revisions])
792
if revision_id is None:
794
elif revision_id not in entire_graph:
795
raise errors.NoSuchRevision(self, revision_id)
797
# add what can be reached from revision_id
799
pending = set([revision_id])
800
while len(pending) > 0:
802
result[node] = entire_graph[node]
803
for revision_id in result[node]:
804
if revision_id not in result:
805
pending.add(revision_id)
809
def get_revision_graph_with_ghosts(self, revision_ids=None):
810
"""Return a graph of the revisions with ghosts marked as applicable.
812
:param revision_ids: an iterable of revisions to graph or None for all.
813
:return: a Graph object with the graph reachable from revision_ids.
815
result = deprecated_graph.Graph()
817
pending = set(self.all_revision_ids())
820
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
821
# special case NULL_REVISION
822
if _mod_revision.NULL_REVISION in pending:
823
pending.remove(_mod_revision.NULL_REVISION)
824
required = set(pending)
827
revision_id = pending.pop()
829
rev = self.get_revision(revision_id)
830
except errors.NoSuchRevision:
831
if revision_id in required:
834
result.add_ghost(revision_id)
836
for parent_id in rev.parent_ids:
837
# is this queued or done ?
838
if (parent_id not in pending and
839
parent_id not in done):
841
pending.add(parent_id)
842
result.add_node(revision_id, rev.parent_ids)
843
done.add(revision_id)
846
def _get_history_vf(self):
847
"""Get a versionedfile whose history graph reflects all revisions.
849
For weave repositories, this is the inventory weave.
851
return self.get_inventory_weave()
853
def iter_reverse_revision_history(self, revision_id):
854
"""Iterate backwards through revision ids in the lefthand history
856
:param revision_id: The revision id to start with. All its lefthand
857
ancestors will be traversed.
859
revision_id = osutils.safe_revision_id(revision_id)
860
if revision_id in (None, _mod_revision.NULL_REVISION):
862
next_id = revision_id
863
versionedfile = self._get_history_vf()
866
parents = versionedfile.get_parents(next_id)
867
if len(parents) == 0:
873
def get_revision_inventory(self, revision_id):
874
"""Return inventory of a past revision."""
875
# TODO: Unify this with get_inventory()
876
# bzr 0.0.6 and later imposes the constraint that the inventory_id
877
# must be the same as its revision, so this is trivial.
878
if revision_id is None:
879
# This does not make sense: if there is no revision,
880
# then it is the current tree inventory surely ?!
881
# and thus get_root_id() is something that looks at the last
882
# commit on the branch, and the get_root_id is an inventory check.
883
raise NotImplementedError
884
# return Inventory(self.get_root_id())
886
return self.get_inventory(revision_id)
890
"""Return True if this repository is flagged as a shared repository."""
891
raise NotImplementedError(self.is_shared)
894
def reconcile(self, other=None, thorough=False):
895
"""Reconcile this repository."""
896
from bzrlib.reconcile import RepoReconciler
897
reconciler = RepoReconciler(self, thorough=thorough)
898
reconciler.reconcile()
901
def _refresh_data(self):
902
"""Helper called from lock_* to ensure coherency with disk.
904
The default implementation does nothing; it is however possible
905
for repositories to maintain loaded indices across multiple locks
906
by checking inside their implementation of this method to see
907
whether their indices are still valid. This depends of course on
908
the disk format being validatable in this manner.
912
def revision_tree(self, revision_id):
913
"""Return Tree for a revision on this branch.
915
`revision_id` may be None for the empty tree revision.
917
# TODO: refactor this to use an existing revision object
918
# so we don't need to read it in twice.
919
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
920
return RevisionTree(self, Inventory(root_id=None),
921
_mod_revision.NULL_REVISION)
923
revision_id = osutils.safe_revision_id(revision_id)
924
inv = self.get_revision_inventory(revision_id)
925
return RevisionTree(self, inv, revision_id)
928
def revision_trees(self, revision_ids):
929
"""Return Tree for a revision on this branch.
931
`revision_id` may not be None or 'null:'"""
932
assert None not in revision_ids
933
assert _mod_revision.NULL_REVISION not in revision_ids
934
texts = self.get_inventory_weave().get_texts(revision_ids)
935
for text, revision_id in zip(texts, revision_ids):
936
inv = self.deserialise_inventory(revision_id, text)
937
yield RevisionTree(self, inv, revision_id)
940
def get_ancestry(self, revision_id, topo_sorted=True):
941
"""Return a list of revision-ids integrated by a revision.
943
The first element of the list is always None, indicating the origin
944
revision. This might change when we have history horizons, or
945
perhaps we should have a new API.
947
This is topologically sorted.
949
if _mod_revision.is_null(revision_id):
951
revision_id = osutils.safe_revision_id(revision_id)
952
if not self.has_revision(revision_id):
953
raise errors.NoSuchRevision(self, revision_id)
954
w = self.get_inventory_weave()
955
candidates = w.get_ancestry(revision_id, topo_sorted)
956
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
959
"""Compress the data within the repository.
961
This operation only makes sense for some repository types. For other
962
types it should be a no-op that just returns.
964
This stub method does not require a lock, but subclasses should use
965
@needs_write_lock as this is a long running call its reasonable to
966
implicitly lock for the user.
970
def print_file(self, file, revision_id):
971
"""Print `file` to stdout.
973
FIXME RBC 20060125 as John Meinel points out this is a bad api
974
- it writes to stdout, it assumes that that is valid etc. Fix
975
by creating a new more flexible convenience function.
977
revision_id = osutils.safe_revision_id(revision_id)
978
tree = self.revision_tree(revision_id)
979
# use inventory as it was in that revision
980
file_id = tree.inventory.path2id(file)
982
# TODO: jam 20060427 Write a test for this code path
983
# it had a bug in it, and was raising the wrong
985
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
986
tree.print_file(file_id)
988
def get_transaction(self):
989
return self.control_files.get_transaction()
991
def revision_parents(self, revision_id):
992
revision_id = osutils.safe_revision_id(revision_id)
993
return self.get_inventory_weave().parent_names(revision_id)
995
def get_parents(self, revision_ids):
996
"""See StackedParentsProvider.get_parents"""
998
for revision_id in revision_ids:
999
if revision_id == _mod_revision.NULL_REVISION:
1003
parents = self.get_revision(revision_id).parent_ids
1004
except errors.NoSuchRevision:
1007
if len(parents) == 0:
1008
parents = [_mod_revision.NULL_REVISION]
1009
parents_list.append(parents)
1012
def _make_parents_provider(self):
1015
def get_graph(self, other_repository=None):
1016
"""Return the graph walker for this repository format"""
1017
parents_provider = self._make_parents_provider()
1018
if (other_repository is not None and
1019
other_repository.bzrdir.transport.base !=
1020
self.bzrdir.transport.base):
1021
parents_provider = graph._StackedParentsProvider(
1022
[parents_provider, other_repository._make_parents_provider()])
1023
return graph.Graph(parents_provider)
1026
def set_make_working_trees(self, new_value):
1027
"""Set the policy flag for making working trees when creating branches.
1029
This only applies to branches that use this repository.
1031
The default is 'True'.
1032
:param new_value: True to restore the default, False to disable making
1035
raise NotImplementedError(self.set_make_working_trees)
1037
def make_working_trees(self):
1038
"""Returns the policy for making working trees on new branches."""
1039
raise NotImplementedError(self.make_working_trees)
1042
def sign_revision(self, revision_id, gpg_strategy):
1043
revision_id = osutils.safe_revision_id(revision_id)
1044
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1045
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1048
def has_signature_for_revision_id(self, revision_id):
1049
"""Query for a revision signature for revision_id in the repository."""
1050
revision_id = osutils.safe_revision_id(revision_id)
1051
return self._revision_store.has_signature(revision_id,
1052
self.get_transaction())
1055
def get_signature_text(self, revision_id):
1056
"""Return the text for a signature."""
1057
revision_id = osutils.safe_revision_id(revision_id)
1058
return self._revision_store.get_signature_text(revision_id,
1059
self.get_transaction())
1062
def check(self, revision_ids):
1063
"""Check consistency of all history of given revision_ids.
1065
Different repository implementations should override _check().
1067
:param revision_ids: A non-empty list of revision_ids whose ancestry
1068
will be checked. Typically the last revision_id of a branch.
1070
if not revision_ids:
1071
raise ValueError("revision_ids must be non-empty in %s.check"
1073
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1074
return self._check(revision_ids)
1076
def _check(self, revision_ids):
1077
result = check.Check(self)
1081
def _warn_if_deprecated(self):
1082
global _deprecation_warning_done
1083
if _deprecation_warning_done:
1085
_deprecation_warning_done = True
1086
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1087
% (self._format, self.bzrdir.transport.base))
1089
def supports_rich_root(self):
1090
return self._format.rich_root_data
1092
def _check_ascii_revisionid(self, revision_id, method):
1093
"""Private helper for ascii-only repositories."""
1094
# weave repositories refuse to store revisionids that are non-ascii.
1095
if revision_id is not None:
1096
# weaves require ascii revision ids.
1097
if isinstance(revision_id, unicode):
1099
revision_id.encode('ascii')
1100
except UnicodeEncodeError:
1101
raise errors.NonAsciiRevisionId(method, self)
1104
revision_id.decode('ascii')
1105
except UnicodeDecodeError:
1106
raise errors.NonAsciiRevisionId(method, self)
1110
# remove these delegates a while after bzr 0.15
1111
def __make_delegated(name, from_module):
1112
def _deprecated_repository_forwarder():
1113
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1114
% (name, from_module),
1117
m = __import__(from_module, globals(), locals(), [name])
1119
return getattr(m, name)
1120
except AttributeError:
1121
raise AttributeError('module %s has no name %s'
1123
globals()[name] = _deprecated_repository_forwarder
1126
'AllInOneRepository',
1127
'WeaveMetaDirRepository',
1128
'PreSplitOutRepositoryFormat',
1129
'RepositoryFormat4',
1130
'RepositoryFormat5',
1131
'RepositoryFormat6',
1132
'RepositoryFormat7',
1134
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1138
'RepositoryFormatKnit',
1139
'RepositoryFormatKnit1',
1141
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1144
def install_revision(repository, rev, revision_tree):
1145
"""Install all revision data into a repository."""
1146
present_parents = []
1148
for p_id in rev.parent_ids:
1149
if repository.has_revision(p_id):
1150
present_parents.append(p_id)
1151
parent_trees[p_id] = repository.revision_tree(p_id)
1153
parent_trees[p_id] = repository.revision_tree(None)
1155
inv = revision_tree.inventory
1156
entries = inv.iter_entries()
1157
# backwards compatibility hack: skip the root id.
1158
if not repository.supports_rich_root():
1159
path, root = entries.next()
1160
if root.revision != rev.revision_id:
1161
raise errors.IncompatibleRevision(repr(repository))
1162
# Add the texts that are not already present
1163
for path, ie in entries:
1164
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1165
repository.get_transaction())
1166
if ie.revision not in w:
1168
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1169
# with InventoryEntry.find_previous_heads(). if it is, then there
1170
# is a latent bug here where the parents may have ancestors of each
1172
for revision, tree in parent_trees.iteritems():
1173
if ie.file_id not in tree:
1175
parent_id = tree.inventory[ie.file_id].revision
1176
if parent_id in text_parents:
1178
text_parents.append(parent_id)
1180
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1181
repository.get_transaction())
1182
lines = revision_tree.get_file(ie.file_id).readlines()
1183
vfile.add_lines(rev.revision_id, text_parents, lines)
1185
# install the inventory
1186
repository.add_inventory(rev.revision_id, inv, present_parents)
1187
except errors.RevisionAlreadyPresent:
1189
repository.add_revision(rev.revision_id, rev, inv)
1192
class MetaDirRepository(Repository):
1193
"""Repositories in the new meta-dir layout."""
1195
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1196
super(MetaDirRepository, self).__init__(_format,
1202
dir_mode = self.control_files._dir_mode
1203
file_mode = self.control_files._file_mode
1206
def is_shared(self):
1207
"""Return True if this repository is flagged as a shared repository."""
1208
return self.control_files._transport.has('shared-storage')
1211
def set_make_working_trees(self, new_value):
1212
"""Set the policy flag for making working trees when creating branches.
1214
This only applies to branches that use this repository.
1216
The default is 'True'.
1217
:param new_value: True to restore the default, False to disable making
1222
self.control_files._transport.delete('no-working-trees')
1223
except errors.NoSuchFile:
1226
self.control_files.put_utf8('no-working-trees', '')
1228
def make_working_trees(self):
1229
"""Returns the policy for making working trees on new branches."""
1230
return not self.control_files._transport.has('no-working-trees')
1233
class RepositoryFormatRegistry(registry.Registry):
1234
"""Registry of RepositoryFormats.
1237
def get(self, format_string):
1238
r = registry.Registry.get(self, format_string)
1244
format_registry = RepositoryFormatRegistry()
1245
"""Registry of formats, indexed by their identifying format string.
1247
This can contain either format instances themselves, or classes/factories that
1248
can be called to obtain one.
1252
#####################################################################
1253
# Repository Formats
1255
class RepositoryFormat(object):
1256
"""A repository format.
1258
Formats provide three things:
1259
* An initialization routine to construct repository data on disk.
1260
* a format string which is used when the BzrDir supports versioned
1262
* an open routine which returns a Repository instance.
1264
Formats are placed in an dict by their format string for reference
1265
during opening. These should be subclasses of RepositoryFormat
1268
Once a format is deprecated, just deprecate the initialize and open
1269
methods on the format class. Do not deprecate the object, as the
1270
object will be created every system load.
1272
Common instance attributes:
1273
_matchingbzrdir - the bzrdir format that the repository format was
1274
originally written to work with. This can be used if manually
1275
constructing a bzrdir and repository, or more commonly for test suite
1280
return "<%s>" % self.__class__.__name__
1282
def __eq__(self, other):
1283
# format objects are generally stateless
1284
return isinstance(other, self.__class__)
1286
def __ne__(self, other):
1287
return not self == other
1290
def find_format(klass, a_bzrdir):
1291
"""Return the format for the repository object in a_bzrdir.
1293
This is used by bzr native formats that have a "format" file in
1294
the repository. Other methods may be used by different types of
1298
transport = a_bzrdir.get_repository_transport(None)
1299
format_string = transport.get("format").read()
1300
return format_registry.get(format_string)
1301
except errors.NoSuchFile:
1302
raise errors.NoRepositoryPresent(a_bzrdir)
1304
raise errors.UnknownFormatError(format=format_string)
1307
def register_format(klass, format):
1308
format_registry.register(format.get_format_string(), format)
1311
def unregister_format(klass, format):
1312
format_registry.remove(format.get_format_string())
1315
def get_default_format(klass):
1316
"""Return the current default format."""
1317
from bzrlib import bzrdir
1318
return bzrdir.format_registry.make_bzrdir('default').repository_format
1320
def _get_control_store(self, repo_transport, control_files):
1321
"""Return the control store for this repository."""
1322
raise NotImplementedError(self._get_control_store)
1324
def get_format_string(self):
1325
"""Return the ASCII format string that identifies this format.
1327
Note that in pre format ?? repositories the format string is
1328
not permitted nor written to disk.
1330
raise NotImplementedError(self.get_format_string)
1332
def get_format_description(self):
1333
"""Return the short description for this format."""
1334
raise NotImplementedError(self.get_format_description)
1336
def _get_revision_store(self, repo_transport, control_files):
1337
"""Return the revision store object for this a_bzrdir."""
1338
raise NotImplementedError(self._get_revision_store)
1340
def _get_text_rev_store(self,
1347
"""Common logic for getting a revision store for a repository.
1349
see self._get_revision_store for the subclass-overridable method to
1350
get the store for a repository.
1352
from bzrlib.store.revision.text import TextRevisionStore
1353
dir_mode = control_files._dir_mode
1354
file_mode = control_files._file_mode
1355
text_store = TextStore(transport.clone(name),
1357
compressed=compressed,
1359
file_mode=file_mode)
1360
_revision_store = TextRevisionStore(text_store, serializer)
1361
return _revision_store
1363
# TODO: this shouldn't be in the base class, it's specific to things that
1364
# use weaves or knits -- mbp 20070207
1365
def _get_versioned_file_store(self,
1370
versionedfile_class=None,
1371
versionedfile_kwargs={},
1373
if versionedfile_class is None:
1374
versionedfile_class = self._versionedfile_class
1375
weave_transport = control_files._transport.clone(name)
1376
dir_mode = control_files._dir_mode
1377
file_mode = control_files._file_mode
1378
return VersionedFileStore(weave_transport, prefixed=prefixed,
1380
file_mode=file_mode,
1381
versionedfile_class=versionedfile_class,
1382
versionedfile_kwargs=versionedfile_kwargs,
1385
def initialize(self, a_bzrdir, shared=False):
1386
"""Initialize a repository of this format in a_bzrdir.
1388
:param a_bzrdir: The bzrdir to put the new repository in it.
1389
:param shared: The repository should be initialized as a sharable one.
1390
:returns: The new repository object.
1392
This may raise UninitializableFormat if shared repository are not
1393
compatible the a_bzrdir.
1395
raise NotImplementedError(self.initialize)
1397
def is_supported(self):
1398
"""Is this format supported?
1400
Supported formats must be initializable and openable.
1401
Unsupported formats may not support initialization or committing or
1402
some other features depending on the reason for not being supported.
1406
def check_conversion_target(self, target_format):
1407
raise NotImplementedError(self.check_conversion_target)
1409
def open(self, a_bzrdir, _found=False):
1410
"""Return an instance of this format for the bzrdir a_bzrdir.
1412
_found is a private parameter, do not use it.
1414
raise NotImplementedError(self.open)
1417
class MetaDirRepositoryFormat(RepositoryFormat):
1418
"""Common base class for the new repositories using the metadir layout."""
1420
rich_root_data = False
1421
supports_tree_reference = False
1422
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1425
super(MetaDirRepositoryFormat, self).__init__()
1427
def _create_control_files(self, a_bzrdir):
1428
"""Create the required files and the initial control_files object."""
1429
# FIXME: RBC 20060125 don't peek under the covers
1430
# NB: no need to escape relative paths that are url safe.
1431
repository_transport = a_bzrdir.get_repository_transport(self)
1432
control_files = lockable_files.LockableFiles(repository_transport,
1433
'lock', lockdir.LockDir)
1434
control_files.create_lock()
1435
return control_files
1437
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1438
"""Upload the initial blank content."""
1439
control_files = self._create_control_files(a_bzrdir)
1440
control_files.lock_write()
1442
control_files._transport.mkdir_multi(dirs,
1443
mode=control_files._dir_mode)
1444
for file, content in files:
1445
control_files.put(file, content)
1446
for file, content in utf8_files:
1447
control_files.put_utf8(file, content)
1449
control_files.put_utf8('shared-storage', '')
1451
control_files.unlock()
1454
# formats which have no format string are not discoverable
1455
# and not independently creatable, so are not registered. They're
1456
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1457
# needed, it's constructed directly by the BzrDir. Non-native formats where
1458
# the repository is not separately opened are similar.
1460
format_registry.register_lazy(
1461
'Bazaar-NG Repository format 7',
1462
'bzrlib.repofmt.weaverepo',
1465
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1466
# default control directory format
1468
format_registry.register_lazy(
1469
'Bazaar-NG Knit Repository Format 1',
1470
'bzrlib.repofmt.knitrepo',
1471
'RepositoryFormatKnit1',
1473
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1475
format_registry.register_lazy(
1476
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1477
'bzrlib.repofmt.knitrepo',
1478
'RepositoryFormatKnit3',
1482
class InterRepository(InterObject):
1483
"""This class represents operations taking place between two repositories.
1485
Its instances have methods like copy_content and fetch, and contain
1486
references to the source and target repositories these operations can be
1489
Often we will provide convenience methods on 'repository' which carry out
1490
operations with another repository - they will always forward to
1491
InterRepository.get(other).method_name(parameters).
1495
"""The available optimised InterRepository types."""
1497
def copy_content(self, revision_id=None):
1498
raise NotImplementedError(self.copy_content)
1500
def fetch(self, revision_id=None, pb=None):
1501
"""Fetch the content required to construct revision_id.
1503
The content is copied from self.source to self.target.
1505
:param revision_id: if None all content is copied, if NULL_REVISION no
1507
:param pb: optional progress bar to use for progress reports. If not
1508
provided a default one will be created.
1510
Returns the copied revision count and the failed revisions in a tuple:
1513
raise NotImplementedError(self.fetch)
1516
def missing_revision_ids(self, revision_id=None):
1517
"""Return the revision ids that source has that target does not.
1519
These are returned in topological order.
1521
:param revision_id: only return revision ids included by this
1524
# generic, possibly worst case, slow code path.
1525
target_ids = set(self.target.all_revision_ids())
1526
if revision_id is not None:
1527
# TODO: jam 20070210 InterRepository is internal enough that it
1528
# should assume revision_ids are already utf-8
1529
revision_id = osutils.safe_revision_id(revision_id)
1530
source_ids = self.source.get_ancestry(revision_id)
1531
assert source_ids[0] is None
1534
source_ids = self.source.all_revision_ids()
1535
result_set = set(source_ids).difference(target_ids)
1536
# this may look like a no-op: its not. It preserves the ordering
1537
# other_ids had while only returning the members from other_ids
1538
# that we've decided we need.
1539
return [rev_id for rev_id in source_ids if rev_id in result_set]
1542
class InterSameDataRepository(InterRepository):
1543
"""Code for converting between repositories that represent the same data.
1545
Data format and model must match for this to work.
1549
def _get_repo_format_to_test(self):
1550
"""Repository format for testing with."""
1551
return RepositoryFormat.get_default_format()
1554
def is_compatible(source, target):
1555
if source.supports_rich_root() != target.supports_rich_root():
1557
if source._serializer != target._serializer:
1562
def copy_content(self, revision_id=None):
1563
"""Make a complete copy of the content in self into destination.
1565
This copies both the repository's revision data, and configuration information
1566
such as the make_working_trees setting.
1568
This is a destructive operation! Do not use it on existing
1571
:param revision_id: Only copy the content needed to construct
1572
revision_id and its parents.
1575
self.target.set_make_working_trees(self.source.make_working_trees())
1576
except NotImplementedError:
1578
# TODO: jam 20070210 This is fairly internal, so we should probably
1579
# just assert that revision_id is not unicode.
1580
revision_id = osutils.safe_revision_id(revision_id)
1581
# but don't bother fetching if we have the needed data now.
1582
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1583
self.target.has_revision(revision_id)):
1585
self.target.fetch(self.source, revision_id=revision_id)
1588
def fetch(self, revision_id=None, pb=None):
1589
"""See InterRepository.fetch()."""
1590
from bzrlib.fetch import GenericRepoFetcher
1591
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1592
self.source, self.source._format, self.target,
1593
self.target._format)
1594
# TODO: jam 20070210 This should be an assert, not a translate
1595
revision_id = osutils.safe_revision_id(revision_id)
1596
f = GenericRepoFetcher(to_repository=self.target,
1597
from_repository=self.source,
1598
last_revision=revision_id,
1600
return f.count_copied, f.failed_revisions
1603
class InterWeaveRepo(InterSameDataRepository):
1604
"""Optimised code paths between Weave based repositories."""
1607
def _get_repo_format_to_test(self):
1608
from bzrlib.repofmt import weaverepo
1609
return weaverepo.RepositoryFormat7()
1612
def is_compatible(source, target):
1613
"""Be compatible with known Weave formats.
1615
We don't test for the stores being of specific types because that
1616
could lead to confusing results, and there is no need to be
1619
from bzrlib.repofmt.weaverepo import (
1625
return (isinstance(source._format, (RepositoryFormat5,
1627
RepositoryFormat7)) and
1628
isinstance(target._format, (RepositoryFormat5,
1630
RepositoryFormat7)))
1631
except AttributeError:
1635
def copy_content(self, revision_id=None):
1636
"""See InterRepository.copy_content()."""
1637
# weave specific optimised path:
1638
# TODO: jam 20070210 Internal, should be an assert, not translate
1639
revision_id = osutils.safe_revision_id(revision_id)
1641
self.target.set_make_working_trees(self.source.make_working_trees())
1642
except NotImplementedError:
1644
# FIXME do not peek!
1645
if self.source.control_files._transport.listable():
1646
pb = ui.ui_factory.nested_progress_bar()
1648
self.target.weave_store.copy_all_ids(
1649
self.source.weave_store,
1651
from_transaction=self.source.get_transaction(),
1652
to_transaction=self.target.get_transaction())
1653
pb.update('copying inventory', 0, 1)
1654
self.target.control_weaves.copy_multi(
1655
self.source.control_weaves, ['inventory'],
1656
from_transaction=self.source.get_transaction(),
1657
to_transaction=self.target.get_transaction())
1658
self.target._revision_store.text_store.copy_all_ids(
1659
self.source._revision_store.text_store,
1664
self.target.fetch(self.source, revision_id=revision_id)
1667
def fetch(self, revision_id=None, pb=None):
1668
"""See InterRepository.fetch()."""
1669
from bzrlib.fetch import GenericRepoFetcher
1670
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1671
self.source, self.source._format, self.target, self.target._format)
1672
# TODO: jam 20070210 This should be an assert, not a translate
1673
revision_id = osutils.safe_revision_id(revision_id)
1674
f = GenericRepoFetcher(to_repository=self.target,
1675
from_repository=self.source,
1676
last_revision=revision_id,
1678
return f.count_copied, f.failed_revisions
1681
def missing_revision_ids(self, revision_id=None):
1682
"""See InterRepository.missing_revision_ids()."""
1683
# we want all revisions to satisfy revision_id in source.
1684
# but we don't want to stat every file here and there.
1685
# we want then, all revisions other needs to satisfy revision_id
1686
# checked, but not those that we have locally.
1687
# so the first thing is to get a subset of the revisions to
1688
# satisfy revision_id in source, and then eliminate those that
1689
# we do already have.
1690
# this is slow on high latency connection to self, but as as this
1691
# disk format scales terribly for push anyway due to rewriting
1692
# inventory.weave, this is considered acceptable.
1694
if revision_id is not None:
1695
source_ids = self.source.get_ancestry(revision_id)
1696
assert source_ids[0] is None
1699
source_ids = self.source._all_possible_ids()
1700
source_ids_set = set(source_ids)
1701
# source_ids is the worst possible case we may need to pull.
1702
# now we want to filter source_ids against what we actually
1703
# have in target, but don't try to check for existence where we know
1704
# we do not have a revision as that would be pointless.
1705
target_ids = set(self.target._all_possible_ids())
1706
possibly_present_revisions = target_ids.intersection(source_ids_set)
1707
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1708
required_revisions = source_ids_set.difference(actually_present_revisions)
1709
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1710
if revision_id is not None:
1711
# we used get_ancestry to determine source_ids then we are assured all
1712
# revisions referenced are present as they are installed in topological order.
1713
# and the tip revision was validated by get_ancestry.
1714
return required_topo_revisions
1716
# if we just grabbed the possibly available ids, then
1717
# we only have an estimate of whats available and need to validate
1718
# that against the revision records.
1719
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1722
class InterKnitRepo(InterSameDataRepository):
1723
"""Optimised code paths between Knit based repositories."""
1726
def _get_repo_format_to_test(self):
1727
from bzrlib.repofmt import knitrepo
1728
return knitrepo.RepositoryFormatKnit1()
1731
def is_compatible(source, target):
1732
"""Be compatible with known Knit formats.
1734
We don't test for the stores being of specific types because that
1735
could lead to confusing results, and there is no need to be
1738
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1740
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1741
isinstance(target._format, (RepositoryFormatKnit1)))
1742
except AttributeError:
1746
def fetch(self, revision_id=None, pb=None):
1747
"""See InterRepository.fetch()."""
1748
from bzrlib.fetch import KnitRepoFetcher
1749
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1750
self.source, self.source._format, self.target, self.target._format)
1751
# TODO: jam 20070210 This should be an assert, not a translate
1752
revision_id = osutils.safe_revision_id(revision_id)
1753
f = KnitRepoFetcher(to_repository=self.target,
1754
from_repository=self.source,
1755
last_revision=revision_id,
1757
return f.count_copied, f.failed_revisions
1760
def missing_revision_ids(self, revision_id=None):
1761
"""See InterRepository.missing_revision_ids()."""
1762
if revision_id is not None:
1763
source_ids = self.source.get_ancestry(revision_id)
1764
assert source_ids[0] is None
1767
source_ids = self.source._all_possible_ids()
1768
source_ids_set = set(source_ids)
1769
# source_ids is the worst possible case we may need to pull.
1770
# now we want to filter source_ids against what we actually
1771
# have in target, but don't try to check for existence where we know
1772
# we do not have a revision as that would be pointless.
1773
target_ids = set(self.target._all_possible_ids())
1774
possibly_present_revisions = target_ids.intersection(source_ids_set)
1775
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1776
required_revisions = source_ids_set.difference(actually_present_revisions)
1777
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1778
if revision_id is not None:
1779
# we used get_ancestry to determine source_ids then we are assured all
1780
# revisions referenced are present as they are installed in topological order.
1781
# and the tip revision was validated by get_ancestry.
1782
return required_topo_revisions
1784
# if we just grabbed the possibly available ids, then
1785
# we only have an estimate of whats available and need to validate
1786
# that against the revision records.
1787
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1790
class InterModel1and2(InterRepository):
1793
def _get_repo_format_to_test(self):
1797
def is_compatible(source, target):
1798
if not source.supports_rich_root() and target.supports_rich_root():
1804
def fetch(self, revision_id=None, pb=None):
1805
"""See InterRepository.fetch()."""
1806
from bzrlib.fetch import Model1toKnit2Fetcher
1807
# TODO: jam 20070210 This should be an assert, not a translate
1808
revision_id = osutils.safe_revision_id(revision_id)
1809
f = Model1toKnit2Fetcher(to_repository=self.target,
1810
from_repository=self.source,
1811
last_revision=revision_id,
1813
return f.count_copied, f.failed_revisions
1816
def copy_content(self, revision_id=None):
1817
"""Make a complete copy of the content in self into destination.
1819
This is a destructive operation! Do not use it on existing
1822
:param revision_id: Only copy the content needed to construct
1823
revision_id and its parents.
1826
self.target.set_make_working_trees(self.source.make_working_trees())
1827
except NotImplementedError:
1829
# TODO: jam 20070210 Internal, assert, don't translate
1830
revision_id = osutils.safe_revision_id(revision_id)
1831
# but don't bother fetching if we have the needed data now.
1832
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1833
self.target.has_revision(revision_id)):
1835
self.target.fetch(self.source, revision_id=revision_id)
1838
class InterKnit1and2(InterKnitRepo):
1841
def _get_repo_format_to_test(self):
1845
def is_compatible(source, target):
1846
"""Be compatible with Knit1 source and Knit3 target"""
1847
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1849
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1850
RepositoryFormatKnit3
1851
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1852
isinstance(target._format, (RepositoryFormatKnit3)))
1853
except AttributeError:
1857
def fetch(self, revision_id=None, pb=None):
1858
"""See InterRepository.fetch()."""
1859
from bzrlib.fetch import Knit1to2Fetcher
1860
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1861
self.source, self.source._format, self.target,
1862
self.target._format)
1863
# TODO: jam 20070210 This should be an assert, not a translate
1864
revision_id = osutils.safe_revision_id(revision_id)
1865
f = Knit1to2Fetcher(to_repository=self.target,
1866
from_repository=self.source,
1867
last_revision=revision_id,
1869
return f.count_copied, f.failed_revisions
1872
class InterRemoteRepository(InterRepository):
1873
"""Code for converting between RemoteRepository objects.
1875
This just gets an non-remote repository from the RemoteRepository, and calls
1876
InterRepository.get again.
1879
def __init__(self, source, target):
1880
if isinstance(source, remote.RemoteRepository):
1881
source._ensure_real()
1882
real_source = source._real_repository
1884
real_source = source
1885
if isinstance(target, remote.RemoteRepository):
1886
target._ensure_real()
1887
real_target = target._real_repository
1889
real_target = target
1890
self.real_inter = InterRepository.get(real_source, real_target)
1893
def is_compatible(source, target):
1894
if isinstance(source, remote.RemoteRepository):
1896
if isinstance(target, remote.RemoteRepository):
1900
def copy_content(self, revision_id=None):
1901
self.real_inter.copy_content(revision_id=revision_id)
1903
def fetch(self, revision_id=None, pb=None):
1904
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1907
def _get_repo_format_to_test(self):
1911
InterRepository.register_optimiser(InterSameDataRepository)
1912
InterRepository.register_optimiser(InterWeaveRepo)
1913
InterRepository.register_optimiser(InterKnitRepo)
1914
InterRepository.register_optimiser(InterModel1and2)
1915
InterRepository.register_optimiser(InterKnit1and2)
1916
InterRepository.register_optimiser(InterRemoteRepository)
1919
class CopyConverter(object):
1920
"""A repository conversion tool which just performs a copy of the content.
1922
This is slow but quite reliable.
1925
def __init__(self, target_format):
1926
"""Create a CopyConverter.
1928
:param target_format: The format the resulting repository should be.
1930
self.target_format = target_format
1932
def convert(self, repo, pb):
1933
"""Perform the conversion of to_convert, giving feedback via pb.
1935
:param to_convert: The disk object to convert.
1936
:param pb: a progress bar to use for progress information.
1941
# this is only useful with metadir layouts - separated repo content.
1942
# trigger an assertion if not such
1943
repo._format.get_format_string()
1944
self.repo_dir = repo.bzrdir
1945
self.step('Moving repository to repository.backup')
1946
self.repo_dir.transport.move('repository', 'repository.backup')
1947
backup_transport = self.repo_dir.transport.clone('repository.backup')
1948
repo._format.check_conversion_target(self.target_format)
1949
self.source_repo = repo._format.open(self.repo_dir,
1951
_override_transport=backup_transport)
1952
self.step('Creating new repository')
1953
converted = self.target_format.initialize(self.repo_dir,
1954
self.source_repo.is_shared())
1955
converted.lock_write()
1957
self.step('Copying content into repository.')
1958
self.source_repo.copy_content_into(converted)
1961
self.step('Deleting old repository content.')
1962
self.repo_dir.transport.delete_tree('repository.backup')
1963
self.pb.note('repository converted')
1965
def step(self, message):
1966
"""Update the pb by a step."""
1968
self.pb.update(message, self.count, self.total)
1971
class CommitBuilder(object):
1972
"""Provides an interface to build up a commit.
1974
This allows describing a tree to be committed without needing to
1975
know the internals of the format of the repository.
1978
record_root_entry = False
1979
def __init__(self, repository, parents, config, timestamp=None,
1980
timezone=None, committer=None, revprops=None,
1982
"""Initiate a CommitBuilder.
1984
:param repository: Repository to commit to.
1985
:param parents: Revision ids of the parents of the new revision.
1986
:param config: Configuration to use.
1987
:param timestamp: Optional timestamp recorded for commit.
1988
:param timezone: Optional timezone for timestamp.
1989
:param committer: Optional committer to set for commit.
1990
:param revprops: Optional dictionary of revision properties.
1991
:param revision_id: Optional revision id.
1993
self._config = config
1995
if committer is None:
1996
self._committer = self._config.username()
1998
assert isinstance(committer, basestring), type(committer)
1999
self._committer = committer
2001
self.new_inventory = Inventory(None)
2002
self._new_revision_id = osutils.safe_revision_id(revision_id)
2003
self.parents = parents
2004
self.repository = repository
2007
if revprops is not None:
2008
self._revprops.update(revprops)
2010
if timestamp is None:
2011
timestamp = time.time()
2012
# Restrict resolution to 1ms
2013
self._timestamp = round(timestamp, 3)
2015
if timezone is None:
2016
self._timezone = osutils.local_time_offset()
2018
self._timezone = int(timezone)
2020
self._generate_revision_if_needed()
2022
def commit(self, message):
2023
"""Make the actual commit.
2025
:return: The revision id of the recorded revision.
2027
rev = _mod_revision.Revision(
2028
timestamp=self._timestamp,
2029
timezone=self._timezone,
2030
committer=self._committer,
2032
inventory_sha1=self.inv_sha1,
2033
revision_id=self._new_revision_id,
2034
properties=self._revprops)
2035
rev.parent_ids = self.parents
2036
self.repository.add_revision(self._new_revision_id, rev,
2037
self.new_inventory, self._config)
2038
self.repository.commit_write_group()
2039
return self._new_revision_id
2041
def revision_tree(self):
2042
"""Return the tree that was just committed.
2044
After calling commit() this can be called to get a RevisionTree
2045
representing the newly committed tree. This is preferred to
2046
calling Repository.revision_tree() because that may require
2047
deserializing the inventory, while we already have a copy in
2050
return RevisionTree(self.repository, self.new_inventory,
2051
self._new_revision_id)
2053
def finish_inventory(self):
2054
"""Tell the builder that the inventory is finished."""
2055
if self.new_inventory.root is None:
2056
symbol_versioning.warn('Root entry should be supplied to'
2057
' record_entry_contents, as of bzr 0.10.',
2058
DeprecationWarning, stacklevel=2)
2059
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2060
self.new_inventory.revision_id = self._new_revision_id
2061
self.inv_sha1 = self.repository.add_inventory(
2062
self._new_revision_id,
2067
def _gen_revision_id(self):
2068
"""Return new revision-id."""
2069
return generate_ids.gen_revision_id(self._config.username(),
2072
def _generate_revision_if_needed(self):
2073
"""Create a revision id if None was supplied.
2075
If the repository can not support user-specified revision ids
2076
they should override this function and raise CannotSetRevisionId
2077
if _new_revision_id is not None.
2079
:raises: CannotSetRevisionId
2081
if self._new_revision_id is None:
2082
self._new_revision_id = self._gen_revision_id()
2084
def record_entry_contents(self, ie, parent_invs, path, tree):
2085
"""Record the content of ie from tree into the commit if needed.
2087
Side effect: sets ie.revision when unchanged
2089
:param ie: An inventory entry present in the commit.
2090
:param parent_invs: The inventories of the parent revisions of the
2092
:param path: The path the entry is at in the tree.
2093
:param tree: The tree which contains this entry and should be used to
2096
if self.new_inventory.root is None and ie.parent_id is not None:
2097
symbol_versioning.warn('Root entry should be supplied to'
2098
' record_entry_contents, as of bzr 0.10.',
2099
DeprecationWarning, stacklevel=2)
2100
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2102
self.new_inventory.add(ie)
2104
# ie.revision is always None if the InventoryEntry is considered
2105
# for committing. ie.snapshot will record the correct revision
2106
# which may be the sole parent if it is untouched.
2107
if ie.revision is not None:
2110
# In this revision format, root entries have no knit or weave
2111
if ie is self.new_inventory.root:
2112
# When serializing out to disk and back in
2113
# root.revision is always _new_revision_id
2114
ie.revision = self._new_revision_id
2116
previous_entries = ie.find_previous_heads(
2118
self.repository.weave_store,
2119
self.repository.get_transaction())
2120
# we are creating a new revision for ie in the history store
2122
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2124
def modified_directory(self, file_id, file_parents):
2125
"""Record the presence of a symbolic link.
2127
:param file_id: The file_id of the link to record.
2128
:param file_parents: The per-file parent revision ids.
2130
self._add_text_to_weave(file_id, [], file_parents.keys())
2132
def modified_reference(self, file_id, file_parents):
2133
"""Record the modification of a reference.
2135
:param file_id: The file_id of the link to record.
2136
:param file_parents: The per-file parent revision ids.
2138
self._add_text_to_weave(file_id, [], file_parents.keys())
2140
def modified_file_text(self, file_id, file_parents,
2141
get_content_byte_lines, text_sha1=None,
2143
"""Record the text of file file_id
2145
:param file_id: The file_id of the file to record the text of.
2146
:param file_parents: The per-file parent revision ids.
2147
:param get_content_byte_lines: A callable which will return the byte
2149
:param text_sha1: Optional SHA1 of the file contents.
2150
:param text_size: Optional size of the file contents.
2152
# mutter('storing text of file {%s} in revision {%s} into %r',
2153
# file_id, self._new_revision_id, self.repository.weave_store)
2154
# special case to avoid diffing on renames or
2156
if (len(file_parents) == 1
2157
and text_sha1 == file_parents.values()[0].text_sha1
2158
and text_size == file_parents.values()[0].text_size):
2159
previous_ie = file_parents.values()[0]
2160
versionedfile = self.repository.weave_store.get_weave(file_id,
2161
self.repository.get_transaction())
2162
versionedfile.clone_text(self._new_revision_id,
2163
previous_ie.revision, file_parents.keys())
2164
return text_sha1, text_size
2166
new_lines = get_content_byte_lines()
2167
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2168
# should return the SHA1 and size
2169
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2170
return osutils.sha_strings(new_lines), \
2171
sum(map(len, new_lines))
2173
def modified_link(self, file_id, file_parents, link_target):
2174
"""Record the presence of a symbolic link.
2176
:param file_id: The file_id of the link to record.
2177
:param file_parents: The per-file parent revision ids.
2178
:param link_target: Target location of this link.
2180
self._add_text_to_weave(file_id, [], file_parents.keys())
2182
def _add_text_to_weave(self, file_id, new_lines, parents):
2183
versionedfile = self.repository.weave_store.get_weave_or_empty(
2184
file_id, self.repository.get_transaction())
2185
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2186
versionedfile.clear_cache()
2189
class _CommitBuilder(CommitBuilder):
2190
"""Temporary class so old CommitBuilders are detected properly
2192
Note: CommitBuilder works whether or not root entry is recorded.
2195
record_root_entry = True
2198
class RootCommitBuilder(CommitBuilder):
2199
"""This commitbuilder actually records the root id"""
2201
record_root_entry = True
2203
def record_entry_contents(self, ie, parent_invs, path, tree):
2204
"""Record the content of ie from tree into the commit if needed.
2206
Side effect: sets ie.revision when unchanged
2208
:param ie: An inventory entry present in the commit.
2209
:param parent_invs: The inventories of the parent revisions of the
2211
:param path: The path the entry is at in the tree.
2212
:param tree: The tree which contains this entry and should be used to
2215
assert self.new_inventory.root is not None or ie.parent_id is None
2216
self.new_inventory.add(ie)
2218
# ie.revision is always None if the InventoryEntry is considered
2219
# for committing. ie.snapshot will record the correct revision
2220
# which may be the sole parent if it is untouched.
2221
if ie.revision is not None:
2224
previous_entries = ie.find_previous_heads(
2226
self.repository.weave_store,
2227
self.repository.get_transaction())
2228
# we are creating a new revision for ie in the history store
2230
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2242
def _unescaper(match, _map=_unescape_map):
2243
code = match.group(1)
2247
if not code.startswith('#'):
2249
return unichr(int(code[1:])).encode('utf8')
2255
def _unescape_xml(data):
2256
"""Unescape predefined XML entities in a string of data."""
2258
if _unescape_re is None:
2259
_unescape_re = re.compile('\&([^;]*);')
2260
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
