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 (
57
from bzrlib.trace import mutter, note, warning
60
# Old formats display a warning, but only once
61
_deprecation_warning_done = False
64
######################################################################
67
class Repository(object):
68
"""Repository holding history for one or more branches.
70
The repository holds and retrieves historical information including
71
revisions and file history. It's normally accessed only by the Branch,
72
which views a particular line of development through that history.
74
The Repository builds on top of Stores and a Transport, which respectively
75
describe the disk data format and the way of accessing the (possibly
79
_file_ids_altered_regex = lazy_regex.lazy_compile(
80
r'file_id="(?P<file_id>[^"]+)"'
81
r'.*revision="(?P<revision_id>[^"]+)"'
84
def abort_write_group(self):
85
"""Commit the contents accrued within the current write group.
87
:seealso: start_write_group.
89
if self._write_group is not self.get_transaction():
90
# has an unlock or relock occured ?
91
raise errors.BzrError('mismatched lock context and write group.')
92
self._abort_write_group()
93
self._write_group = None
95
def _abort_write_group(self):
96
"""Template method for per-repository write group cleanup.
98
This is called during abort before the write group is considered to be
99
finished and should cleanup any internal state accrued during the write
100
group. There is no requirement that data handed to the repository be
101
*not* made available - this is not a rollback - but neither should any
102
attempt be made to ensure that data added is fully commited. Abort is
103
invoked when an error has occured so futher disk or network operations
104
may not be possible or may error and if possible should not be
109
def add_inventory(self, revision_id, inv, parents):
110
"""Add the inventory inv to the repository as revision_id.
112
:param parents: The revision ids of the parents that revision_id
113
is known to have and are in the repository already.
115
returns the sha1 of the serialized inventory.
117
revision_id = osutils.safe_revision_id(revision_id)
118
_mod_revision.check_not_reserved_id(revision_id)
119
assert inv.revision_id is None or inv.revision_id == revision_id, \
120
"Mismatch between inventory revision" \
121
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
122
assert inv.root is not None
123
inv_text = self.serialise_inventory(inv)
124
inv_sha1 = osutils.sha_string(inv_text)
125
inv_vf = self.control_weaves.get_weave('inventory',
126
self.get_transaction())
127
self._inventory_add_lines(inv_vf, revision_id, parents,
128
osutils.split_lines(inv_text))
131
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
133
for parent in parents:
135
final_parents.append(parent)
137
inv_vf.add_lines(revision_id, final_parents, lines)
140
def add_revision(self, revision_id, rev, inv=None, config=None):
141
"""Add rev to the revision store as revision_id.
143
:param revision_id: the revision id to use.
144
:param rev: The revision object.
145
:param inv: The inventory for the revision. if None, it will be looked
146
up in the inventory storer
147
:param config: If None no digital signature will be created.
148
If supplied its signature_needed method will be used
149
to determine if a signature should be made.
151
revision_id = osutils.safe_revision_id(revision_id)
152
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
154
_mod_revision.check_not_reserved_id(revision_id)
155
if config is not None and config.signature_needed():
157
inv = self.get_inventory(revision_id)
158
plaintext = Testament(rev, inv).as_short_text()
159
self.store_revision_signature(
160
gpg.GPGStrategy(config), plaintext, revision_id)
161
if not revision_id in self.get_inventory_weave():
163
raise errors.WeaveRevisionNotPresent(revision_id,
164
self.get_inventory_weave())
166
# yes, this is not suitable for adding with ghosts.
167
self.add_inventory(revision_id, inv, rev.parent_ids)
168
self._revision_store.add_revision(rev, self.get_transaction())
170
def _add_revision_text(self, revision_id, text):
171
revision = self._revision_store._serializer.read_revision_from_string(
173
self._revision_store._add_revision(revision, StringIO(text),
174
self.get_transaction())
177
def _all_possible_ids(self):
178
"""Return all the possible revisions that we could find."""
179
return self.get_inventory_weave().versions()
181
def all_revision_ids(self):
182
"""Returns a list of all the revision ids in the repository.
184
This is deprecated because code should generally work on the graph
185
reachable from a particular revision, and ignore any other revisions
186
that might be present. There is no direct replacement method.
188
return self._all_revision_ids()
191
def _all_revision_ids(self):
192
"""Returns a list of all the revision ids in the repository.
194
These are in as much topological order as the underlying store can
195
present: for weaves ghosts may lead to a lack of correctness until
196
the reweave updates the parents list.
198
if self._revision_store.text_store.listable():
199
return self._revision_store.all_revision_ids(self.get_transaction())
200
result = self._all_possible_ids()
201
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
202
# ids. (It should, since _revision_store's API should change to
203
# return utf8 revision_ids)
204
return self._eliminate_revisions_not_present(result)
206
def break_lock(self):
207
"""Break a lock if one is present from another instance.
209
Uses the ui factory to ask for confirmation if the lock may be from
212
self.control_files.break_lock()
215
def _eliminate_revisions_not_present(self, revision_ids):
216
"""Check every revision id in revision_ids to see if we have it.
218
Returns a set of the present revisions.
221
for id in revision_ids:
222
if self.has_revision(id):
227
def create(a_bzrdir):
228
"""Construct the current default format repository in a_bzrdir."""
229
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
231
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
232
"""instantiate a Repository.
234
:param _format: The format of the repository on disk.
235
:param a_bzrdir: The BzrDir of the repository.
237
In the future we will have a single api for all stores for
238
getting file texts, inventories and revisions, then
239
this construct will accept instances of those things.
241
super(Repository, self).__init__()
242
self._format = _format
243
# the following are part of the public API for Repository:
244
self.bzrdir = a_bzrdir
245
self.control_files = control_files
246
self._revision_store = _revision_store
247
# backwards compatibility
248
self.weave_store = text_store
249
# not right yet - should be more semantically clear ?
251
self.control_store = control_store
252
self.control_weaves = control_store
253
# TODO: make sure to construct the right store classes, etc, depending
254
# on whether escaping is required.
255
self._warn_if_deprecated()
256
self._write_group = None
259
return '%s(%r)' % (self.__class__.__name__,
260
self.bzrdir.transport.base)
262
def has_same_location(self, other):
263
"""Returns a boolean indicating if this repository is at the same
264
location as another repository.
266
This might return False even when two repository objects are accessing
267
the same physical repository via different URLs.
269
if self.__class__ is not other.__class__:
271
return (self.control_files._transport.base ==
272
other.control_files._transport.base)
274
def is_in_write_group(self):
275
"""Return True if there is an open write group.
277
:seealso: start_write_group.
279
return self._write_group is not None
282
return self.control_files.is_locked()
284
def lock_write(self, token=None):
285
"""Lock this repository for writing.
287
This causes caching within the repository obejct to start accumlating
288
data during reads, and allows a 'write_group' to be obtained. Write
289
groups must be used for actual data insertion.
291
:param token: if this is already locked, then lock_write will fail
292
unless the token matches the existing lock.
293
:returns: a token if this instance supports tokens, otherwise None.
294
:raises TokenLockingNotSupported: when a token is given but this
295
instance doesn't support using token locks.
296
:raises MismatchedToken: if the specified token doesn't match the token
297
of the existing lock.
298
:seealso: start_write_group.
300
A token should be passed in if you know that you have locked the object
301
some other way, and need to synchronise this object's state with that
304
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
306
result = self.control_files.lock_write(token=token)
311
self.control_files.lock_read()
314
def get_physical_lock_status(self):
315
return self.control_files.get_physical_lock_status()
317
def leave_lock_in_place(self):
318
"""Tell this repository not to release the physical lock when this
321
If lock_write doesn't return a token, then this method is not supported.
323
self.control_files.leave_in_place()
325
def dont_leave_lock_in_place(self):
326
"""Tell this repository to release the physical lock when this
327
object is unlocked, even if it didn't originally acquire it.
329
If lock_write doesn't return a token, then this method is not supported.
331
self.control_files.dont_leave_in_place()
334
def gather_stats(self, revid=None, committers=None):
335
"""Gather statistics from a revision id.
337
:param revid: The revision id to gather statistics from, if None, then
338
no revision specific statistics are gathered.
339
:param committers: Optional parameter controlling whether to grab
340
a count of committers from the revision specific statistics.
341
:return: A dictionary of statistics. Currently this contains:
342
committers: The number of committers if requested.
343
firstrev: A tuple with timestamp, timezone for the penultimate left
344
most ancestor of revid, if revid is not the NULL_REVISION.
345
latestrev: A tuple with timestamp, timezone for revid, if revid is
346
not the NULL_REVISION.
347
revisions: The total revision count in the repository.
348
size: An estimate disk size of the repository in bytes.
351
if revid and committers:
352
result['committers'] = 0
353
if revid and revid != _mod_revision.NULL_REVISION:
355
all_committers = set()
356
revisions = self.get_ancestry(revid)
357
# pop the leading None
359
first_revision = None
361
# ignore the revisions in the middle - just grab first and last
362
revisions = revisions[0], revisions[-1]
363
for revision in self.get_revisions(revisions):
364
if not first_revision:
365
first_revision = revision
367
all_committers.add(revision.committer)
368
last_revision = revision
370
result['committers'] = len(all_committers)
371
result['firstrev'] = (first_revision.timestamp,
372
first_revision.timezone)
373
result['latestrev'] = (last_revision.timestamp,
374
last_revision.timezone)
376
# now gather global repository information
377
if self.bzrdir.root_transport.listable():
378
c, t = self._revision_store.total_size(self.get_transaction())
379
result['revisions'] = c
384
def missing_revision_ids(self, other, revision_id=None):
385
"""Return the revision ids that other has that this does not.
387
These are returned in topological order.
389
revision_id: only return revision ids included by revision_id.
391
revision_id = osutils.safe_revision_id(revision_id)
392
return InterRepository.get(other, self).missing_revision_ids(revision_id)
396
"""Open the repository rooted at base.
398
For instance, if the repository is at URL/.bzr/repository,
399
Repository.open(URL) -> a Repository instance.
401
control = bzrdir.BzrDir.open(base)
402
return control.open_repository()
404
def copy_content_into(self, destination, revision_id=None):
405
"""Make a complete copy of the content in self into destination.
407
This is a destructive operation! Do not use it on existing
410
revision_id = osutils.safe_revision_id(revision_id)
411
return InterRepository.get(self, destination).copy_content(revision_id)
413
def commit_write_group(self):
414
"""Commit the contents accrued within the current write group.
416
:seealso: start_write_group.
418
if self._write_group is not self.get_transaction():
419
# has an unlock or relock occured ?
420
raise errors.BzrError('mismatched lock context and write group.')
421
self._commit_write_group()
422
self._write_group = None
424
def _commit_write_group(self):
425
"""Template method for per-repository write group cleanup.
427
This is called before the write group is considered to be
428
finished and should ensure that all data handed to the repository
429
for writing during the write group is safely committed (to the
430
extent possible considering file system caching etc).
433
def fetch(self, source, revision_id=None, pb=None):
434
"""Fetch the content required to construct revision_id from source.
436
If revision_id is None all content is copied.
438
revision_id = osutils.safe_revision_id(revision_id)
439
inter = InterRepository.get(source, self)
441
return inter.fetch(revision_id=revision_id, pb=pb)
442
except NotImplementedError:
443
raise errors.IncompatibleRepositories(source, self)
445
def create_bundle(self, target, base, fileobj, format=None):
446
return serializer.write_bundle(self, target, base, fileobj, format)
448
def get_commit_builder(self, branch, parents, config, timestamp=None,
449
timezone=None, committer=None, revprops=None,
451
"""Obtain a CommitBuilder for this repository.
453
:param branch: Branch to commit to.
454
:param parents: Revision ids of the parents of the new revision.
455
:param config: Configuration to use.
456
:param timestamp: Optional timestamp recorded for commit.
457
:param timezone: Optional timezone for timestamp.
458
:param committer: Optional committer to set for commit.
459
:param revprops: Optional dictionary of revision properties.
460
:param revision_id: Optional revision id.
462
revision_id = osutils.safe_revision_id(revision_id)
463
result =_CommitBuilder(self, parents, config, timestamp, timezone,
464
committer, revprops, revision_id)
465
self.start_write_group()
469
if (self.control_files._lock_count == 1 and
470
self.control_files._lock_mode == 'w'):
471
if self._write_group is not None:
472
raise errors.BzrError(
473
'Must end write groups before releasing write locks.')
474
self.control_files.unlock()
477
def clone(self, a_bzrdir, revision_id=None):
478
"""Clone this repository into a_bzrdir using the current format.
480
Currently no check is made that the format of this repository and
481
the bzrdir format are compatible. FIXME RBC 20060201.
483
:return: The newly created destination repository.
485
# TODO: deprecate after 0.16; cloning this with all its settings is
486
# probably not very useful -- mbp 20070423
487
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
488
self.copy_content_into(dest_repo, revision_id)
491
def start_write_group(self):
492
"""Start a write group in the repository.
494
Write groups are used by repositories which do not have a 1:1 mapping
495
between file ids and backend store to manage the insertion of data from
496
both fetch and commit operations.
498
A write lock is required around the start_write_group/commit_write_group
499
for the support of lock-requiring repository formats.
501
One can only insert data into a repository inside a write group.
505
if not self.is_locked() or self.control_files._lock_mode != 'w':
506
raise errors.NotWriteLocked(self)
507
if self._write_group:
508
raise errors.BzrError('already in a write group')
509
self._start_write_group()
510
# so we can detect unlock/relock - the write group is now entered.
511
self._write_group = self.get_transaction()
513
def _start_write_group(self):
514
"""Template method for per-repository write group startup.
516
This is called before the write group is considered to be
521
def sprout(self, to_bzrdir, revision_id=None):
522
"""Create a descendent repository for new development.
524
Unlike clone, this does not copy the settings of the repository.
526
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
527
dest_repo.fetch(self, revision_id=revision_id)
530
def _create_sprouting_repo(self, a_bzrdir, shared):
531
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
532
# use target default format.
533
dest_repo = a_bzrdir.create_repository()
535
# Most control formats need the repository to be specifically
536
# created, but on some old all-in-one formats it's not needed
538
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
539
except errors.UninitializableFormat:
540
dest_repo = a_bzrdir.open_repository()
544
def has_revision(self, revision_id):
545
"""True if this repository has a copy of the revision."""
546
revision_id = osutils.safe_revision_id(revision_id)
547
return self._revision_store.has_revision_id(revision_id,
548
self.get_transaction())
551
def get_revision_reconcile(self, revision_id):
552
"""'reconcile' helper routine that allows access to a revision always.
554
This variant of get_revision does not cross check the weave graph
555
against the revision one as get_revision does: but it should only
556
be used by reconcile, or reconcile-alike commands that are correcting
557
or testing the revision graph.
559
if not revision_id or not isinstance(revision_id, basestring):
560
raise errors.InvalidRevisionId(revision_id=revision_id,
562
return self.get_revisions([revision_id])[0]
565
def get_revisions(self, revision_ids):
566
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
567
revs = self._revision_store.get_revisions(revision_ids,
568
self.get_transaction())
570
assert not isinstance(rev.revision_id, unicode)
571
for parent_id in rev.parent_ids:
572
assert not isinstance(parent_id, unicode)
576
def get_revision_xml(self, revision_id):
577
# TODO: jam 20070210 This shouldn't be necessary since get_revision
578
# would have already do it.
579
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
580
revision_id = osutils.safe_revision_id(revision_id)
581
rev = self.get_revision(revision_id)
583
# the current serializer..
584
self._revision_store._serializer.write_revision(rev, rev_tmp)
586
return rev_tmp.getvalue()
589
def get_revision(self, revision_id):
590
"""Return the Revision object for a named revision"""
591
# TODO: jam 20070210 get_revision_reconcile should do this for us
592
revision_id = osutils.safe_revision_id(revision_id)
593
r = self.get_revision_reconcile(revision_id)
594
# weave corruption can lead to absent revision markers that should be
596
# the following test is reasonably cheap (it needs a single weave read)
597
# and the weave is cached in read transactions. In write transactions
598
# it is not cached but typically we only read a small number of
599
# revisions. For knits when they are introduced we will probably want
600
# to ensure that caching write transactions are in use.
601
inv = self.get_inventory_weave()
602
self._check_revision_parents(r, inv)
606
def get_deltas_for_revisions(self, revisions):
607
"""Produce a generator of revision deltas.
609
Note that the input is a sequence of REVISIONS, not revision_ids.
610
Trees will be held in memory until the generator exits.
611
Each delta is relative to the revision's lefthand predecessor.
613
required_trees = set()
614
for revision in revisions:
615
required_trees.add(revision.revision_id)
616
required_trees.update(revision.parent_ids[:1])
617
trees = dict((t.get_revision_id(), t) for
618
t in self.revision_trees(required_trees))
619
for revision in revisions:
620
if not revision.parent_ids:
621
old_tree = self.revision_tree(None)
623
old_tree = trees[revision.parent_ids[0]]
624
yield trees[revision.revision_id].changes_from(old_tree)
627
def get_revision_delta(self, revision_id):
628
"""Return the delta for one revision.
630
The delta is relative to the left-hand predecessor of the
633
r = self.get_revision(revision_id)
634
return list(self.get_deltas_for_revisions([r]))[0]
636
def _check_revision_parents(self, revision, inventory):
637
"""Private to Repository and Fetch.
639
This checks the parentage of revision in an inventory weave for
640
consistency and is only applicable to inventory-weave-for-ancestry
641
using repository formats & fetchers.
643
weave_parents = inventory.get_parents(revision.revision_id)
644
weave_names = inventory.versions()
645
for parent_id in revision.parent_ids:
646
if parent_id in weave_names:
647
# this parent must not be a ghost.
648
if not parent_id in weave_parents:
650
raise errors.CorruptRepository(self)
653
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
654
revision_id = osutils.safe_revision_id(revision_id)
655
signature = gpg_strategy.sign(plaintext)
656
self._revision_store.add_revision_signature_text(revision_id,
658
self.get_transaction())
660
def fileids_altered_by_revision_ids(self, revision_ids):
661
"""Find the file ids and versions affected by revisions.
663
:param revisions: an iterable containing revision ids.
664
:return: a dictionary mapping altered file-ids to an iterable of
665
revision_ids. Each altered file-ids has the exact revision_ids that
666
altered it listed explicitly.
668
assert self._serializer.support_altered_by_hack, \
669
("fileids_altered_by_revision_ids only supported for branches "
670
"which store inventory as unnested xml, not on %r" % self)
671
selected_revision_ids = set(osutils.safe_revision_id(r)
672
for r in revision_ids)
673
w = self.get_inventory_weave()
676
# this code needs to read every new line in every inventory for the
677
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
678
# not present in one of those inventories is unnecessary but not
679
# harmful because we are filtering by the revision id marker in the
680
# inventory lines : we only select file ids altered in one of those
681
# revisions. We don't need to see all lines in the inventory because
682
# only those added in an inventory in rev X can contain a revision=X
684
unescape_revid_cache = {}
685
unescape_fileid_cache = {}
687
# jam 20061218 In a big fetch, this handles hundreds of thousands
688
# of lines, so it has had a lot of inlining and optimizing done.
689
# Sorry that it is a little bit messy.
690
# Move several functions to be local variables, since this is a long
692
search = self._file_ids_altered_regex.search
693
unescape = _unescape_xml
694
setdefault = result.setdefault
695
pb = ui.ui_factory.nested_progress_bar()
697
for line in w.iter_lines_added_or_present_in_versions(
698
selected_revision_ids, pb=pb):
702
# One call to match.group() returning multiple items is quite a
703
# bit faster than 2 calls to match.group() each returning 1
704
file_id, revision_id = match.group('file_id', 'revision_id')
706
# Inlining the cache lookups helps a lot when you make 170,000
707
# lines and 350k ids, versus 8.4 unique ids.
708
# Using a cache helps in 2 ways:
709
# 1) Avoids unnecessary decoding calls
710
# 2) Re-uses cached strings, which helps in future set and
712
# (2) is enough that removing encoding entirely along with
713
# the cache (so we are using plain strings) results in no
714
# performance improvement.
716
revision_id = unescape_revid_cache[revision_id]
718
unescaped = unescape(revision_id)
719
unescape_revid_cache[revision_id] = unescaped
720
revision_id = unescaped
722
if revision_id in selected_revision_ids:
724
file_id = unescape_fileid_cache[file_id]
726
unescaped = unescape(file_id)
727
unescape_fileid_cache[file_id] = unescaped
729
setdefault(file_id, set()).add(revision_id)
735
def get_inventory_weave(self):
736
return self.control_weaves.get_weave('inventory',
737
self.get_transaction())
740
def get_inventory(self, revision_id):
741
"""Get Inventory object by hash."""
742
# TODO: jam 20070210 Technically we don't need to sanitize, since all
743
# called functions must sanitize.
744
revision_id = osutils.safe_revision_id(revision_id)
745
return self.deserialise_inventory(
746
revision_id, self.get_inventory_xml(revision_id))
748
def deserialise_inventory(self, revision_id, xml):
749
"""Transform the xml into an inventory object.
751
:param revision_id: The expected revision id of the inventory.
752
:param xml: A serialised inventory.
754
revision_id = osutils.safe_revision_id(revision_id)
755
result = self._serializer.read_inventory_from_string(xml)
756
result.root.revision = revision_id
759
def serialise_inventory(self, inv):
760
return self._serializer.write_inventory_to_string(inv)
762
def get_serializer_format(self):
763
return self._serializer.format_num
766
def get_inventory_xml(self, revision_id):
767
"""Get inventory XML as a file object."""
768
revision_id = osutils.safe_revision_id(revision_id)
770
assert isinstance(revision_id, str), type(revision_id)
771
iw = self.get_inventory_weave()
772
return iw.get_text(revision_id)
774
raise errors.HistoryMissing(self, 'inventory', revision_id)
777
def get_inventory_sha1(self, revision_id):
778
"""Return the sha1 hash of the inventory entry
780
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
781
revision_id = osutils.safe_revision_id(revision_id)
782
return self.get_revision(revision_id).inventory_sha1
785
def get_revision_graph(self, revision_id=None):
786
"""Return a dictionary containing the revision graph.
788
:param revision_id: The revision_id to get a graph from. If None, then
789
the entire revision graph is returned. This is a deprecated mode of
790
operation and will be removed in the future.
791
:return: a dictionary of revision_id->revision_parents_list.
793
# special case NULL_REVISION
794
if revision_id == _mod_revision.NULL_REVISION:
796
revision_id = osutils.safe_revision_id(revision_id)
797
a_weave = self.get_inventory_weave()
798
all_revisions = self._eliminate_revisions_not_present(
800
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
801
node in all_revisions])
802
if revision_id is None:
804
elif revision_id not in entire_graph:
805
raise errors.NoSuchRevision(self, revision_id)
807
# add what can be reached from revision_id
809
pending = set([revision_id])
810
while len(pending) > 0:
812
result[node] = entire_graph[node]
813
for revision_id in result[node]:
814
if revision_id not in result:
815
pending.add(revision_id)
819
def get_revision_graph_with_ghosts(self, revision_ids=None):
820
"""Return a graph of the revisions with ghosts marked as applicable.
822
:param revision_ids: an iterable of revisions to graph or None for all.
823
:return: a Graph object with the graph reachable from revision_ids.
825
result = deprecated_graph.Graph()
827
pending = set(self.all_revision_ids())
830
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
831
# special case NULL_REVISION
832
if _mod_revision.NULL_REVISION in pending:
833
pending.remove(_mod_revision.NULL_REVISION)
834
required = set(pending)
837
revision_id = pending.pop()
839
rev = self.get_revision(revision_id)
840
except errors.NoSuchRevision:
841
if revision_id in required:
844
result.add_ghost(revision_id)
846
for parent_id in rev.parent_ids:
847
# is this queued or done ?
848
if (parent_id not in pending and
849
parent_id not in done):
851
pending.add(parent_id)
852
result.add_node(revision_id, rev.parent_ids)
853
done.add(revision_id)
856
def _get_history_vf(self):
857
"""Get a versionedfile whose history graph reflects all revisions.
859
For weave repositories, this is the inventory weave.
861
return self.get_inventory_weave()
863
def iter_reverse_revision_history(self, revision_id):
864
"""Iterate backwards through revision ids in the lefthand history
866
:param revision_id: The revision id to start with. All its lefthand
867
ancestors will be traversed.
869
revision_id = osutils.safe_revision_id(revision_id)
870
if revision_id in (None, _mod_revision.NULL_REVISION):
872
next_id = revision_id
873
versionedfile = self._get_history_vf()
876
parents = versionedfile.get_parents(next_id)
877
if len(parents) == 0:
883
def get_revision_inventory(self, revision_id):
884
"""Return inventory of a past revision."""
885
# TODO: Unify this with get_inventory()
886
# bzr 0.0.6 and later imposes the constraint that the inventory_id
887
# must be the same as its revision, so this is trivial.
888
if revision_id is None:
889
# This does not make sense: if there is no revision,
890
# then it is the current tree inventory surely ?!
891
# and thus get_root_id() is something that looks at the last
892
# commit on the branch, and the get_root_id is an inventory check.
893
raise NotImplementedError
894
# return Inventory(self.get_root_id())
896
return self.get_inventory(revision_id)
900
"""Return True if this repository is flagged as a shared repository."""
901
raise NotImplementedError(self.is_shared)
904
def reconcile(self, other=None, thorough=False):
905
"""Reconcile this repository."""
906
from bzrlib.reconcile import RepoReconciler
907
reconciler = RepoReconciler(self, thorough=thorough)
908
reconciler.reconcile()
911
def _refresh_data(self):
912
"""Helper called from lock_* to ensure coherency with disk.
914
The default implementation does nothing; it is however possible
915
for repositories to maintain loaded indices across multiple locks
916
by checking inside their implementation of this method to see
917
whether their indices are still valid. This depends of course on
918
the disk format being validatable in this manner.
922
def revision_tree(self, revision_id):
923
"""Return Tree for a revision on this branch.
925
`revision_id` may be None for the empty tree revision.
927
# TODO: refactor this to use an existing revision object
928
# so we don't need to read it in twice.
929
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
930
return RevisionTree(self, Inventory(root_id=None),
931
_mod_revision.NULL_REVISION)
933
revision_id = osutils.safe_revision_id(revision_id)
934
inv = self.get_revision_inventory(revision_id)
935
return RevisionTree(self, inv, revision_id)
938
def revision_trees(self, revision_ids):
939
"""Return Tree for a revision on this branch.
941
`revision_id` may not be None or 'null:'"""
942
assert None not in revision_ids
943
assert _mod_revision.NULL_REVISION not in revision_ids
944
texts = self.get_inventory_weave().get_texts(revision_ids)
945
for text, revision_id in zip(texts, revision_ids):
946
inv = self.deserialise_inventory(revision_id, text)
947
yield RevisionTree(self, inv, revision_id)
950
def get_ancestry(self, revision_id, topo_sorted=True):
951
"""Return a list of revision-ids integrated by a revision.
953
The first element of the list is always None, indicating the origin
954
revision. This might change when we have history horizons, or
955
perhaps we should have a new API.
957
This is topologically sorted.
959
if _mod_revision.is_null(revision_id):
961
revision_id = osutils.safe_revision_id(revision_id)
962
if not self.has_revision(revision_id):
963
raise errors.NoSuchRevision(self, revision_id)
964
w = self.get_inventory_weave()
965
candidates = w.get_ancestry(revision_id, topo_sorted)
966
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
969
"""Compress the data within the repository.
971
This operation only makes sense for some repository types. For other
972
types it should be a no-op that just returns.
974
This stub method does not require a lock, but subclasses should use
975
@needs_write_lock as this is a long running call its reasonable to
976
implicitly lock for the user.
980
def print_file(self, file, revision_id):
981
"""Print `file` to stdout.
983
FIXME RBC 20060125 as John Meinel points out this is a bad api
984
- it writes to stdout, it assumes that that is valid etc. Fix
985
by creating a new more flexible convenience function.
987
revision_id = osutils.safe_revision_id(revision_id)
988
tree = self.revision_tree(revision_id)
989
# use inventory as it was in that revision
990
file_id = tree.inventory.path2id(file)
992
# TODO: jam 20060427 Write a test for this code path
993
# it had a bug in it, and was raising the wrong
995
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
996
tree.print_file(file_id)
998
def get_transaction(self):
999
return self.control_files.get_transaction()
1001
def revision_parents(self, revision_id):
1002
revision_id = osutils.safe_revision_id(revision_id)
1003
return self.get_inventory_weave().parent_names(revision_id)
1005
def get_parents(self, revision_ids):
1006
"""See StackedParentsProvider.get_parents"""
1008
for revision_id in revision_ids:
1009
if revision_id == _mod_revision.NULL_REVISION:
1013
parents = self.get_revision(revision_id).parent_ids
1014
except errors.NoSuchRevision:
1017
if len(parents) == 0:
1018
parents = [_mod_revision.NULL_REVISION]
1019
parents_list.append(parents)
1022
def _make_parents_provider(self):
1025
def get_graph(self, other_repository=None):
1026
"""Return the graph walker for this repository format"""
1027
parents_provider = self._make_parents_provider()
1028
if (other_repository is not None and
1029
other_repository.bzrdir.transport.base !=
1030
self.bzrdir.transport.base):
1031
parents_provider = graph._StackedParentsProvider(
1032
[parents_provider, other_repository._make_parents_provider()])
1033
return graph.Graph(parents_provider)
1036
def set_make_working_trees(self, new_value):
1037
"""Set the policy flag for making working trees when creating branches.
1039
This only applies to branches that use this repository.
1041
The default is 'True'.
1042
:param new_value: True to restore the default, False to disable making
1045
raise NotImplementedError(self.set_make_working_trees)
1047
def make_working_trees(self):
1048
"""Returns the policy for making working trees on new branches."""
1049
raise NotImplementedError(self.make_working_trees)
1052
def sign_revision(self, revision_id, gpg_strategy):
1053
revision_id = osutils.safe_revision_id(revision_id)
1054
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1055
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1058
def has_signature_for_revision_id(self, revision_id):
1059
"""Query for a revision signature for revision_id in the repository."""
1060
revision_id = osutils.safe_revision_id(revision_id)
1061
return self._revision_store.has_signature(revision_id,
1062
self.get_transaction())
1065
def get_signature_text(self, revision_id):
1066
"""Return the text for a signature."""
1067
revision_id = osutils.safe_revision_id(revision_id)
1068
return self._revision_store.get_signature_text(revision_id,
1069
self.get_transaction())
1072
def check(self, revision_ids):
1073
"""Check consistency of all history of given revision_ids.
1075
Different repository implementations should override _check().
1077
:param revision_ids: A non-empty list of revision_ids whose ancestry
1078
will be checked. Typically the last revision_id of a branch.
1080
if not revision_ids:
1081
raise ValueError("revision_ids must be non-empty in %s.check"
1083
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1084
return self._check(revision_ids)
1086
def _check(self, revision_ids):
1087
result = check.Check(self)
1091
def _warn_if_deprecated(self):
1092
global _deprecation_warning_done
1093
if _deprecation_warning_done:
1095
_deprecation_warning_done = True
1096
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1097
% (self._format, self.bzrdir.transport.base))
1099
def supports_rich_root(self):
1100
return self._format.rich_root_data
1102
def _check_ascii_revisionid(self, revision_id, method):
1103
"""Private helper for ascii-only repositories."""
1104
# weave repositories refuse to store revisionids that are non-ascii.
1105
if revision_id is not None:
1106
# weaves require ascii revision ids.
1107
if isinstance(revision_id, unicode):
1109
revision_id.encode('ascii')
1110
except UnicodeEncodeError:
1111
raise errors.NonAsciiRevisionId(method, self)
1114
revision_id.decode('ascii')
1115
except UnicodeDecodeError:
1116
raise errors.NonAsciiRevisionId(method, self)
1120
# remove these delegates a while after bzr 0.15
1121
def __make_delegated(name, from_module):
1122
def _deprecated_repository_forwarder():
1123
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1124
% (name, from_module),
1127
m = __import__(from_module, globals(), locals(), [name])
1129
return getattr(m, name)
1130
except AttributeError:
1131
raise AttributeError('module %s has no name %s'
1133
globals()[name] = _deprecated_repository_forwarder
1136
'AllInOneRepository',
1137
'WeaveMetaDirRepository',
1138
'PreSplitOutRepositoryFormat',
1139
'RepositoryFormat4',
1140
'RepositoryFormat5',
1141
'RepositoryFormat6',
1142
'RepositoryFormat7',
1144
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1148
'RepositoryFormatKnit',
1149
'RepositoryFormatKnit1',
1151
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1154
def install_revision(repository, rev, revision_tree):
1155
"""Install all revision data into a repository."""
1156
present_parents = []
1158
for p_id in rev.parent_ids:
1159
if repository.has_revision(p_id):
1160
present_parents.append(p_id)
1161
parent_trees[p_id] = repository.revision_tree(p_id)
1163
parent_trees[p_id] = repository.revision_tree(None)
1165
inv = revision_tree.inventory
1166
entries = inv.iter_entries()
1167
# backwards compatibility hack: skip the root id.
1168
if not repository.supports_rich_root():
1169
path, root = entries.next()
1170
if root.revision != rev.revision_id:
1171
raise errors.IncompatibleRevision(repr(repository))
1172
# Add the texts that are not already present
1173
for path, ie in entries:
1174
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1175
repository.get_transaction())
1176
if ie.revision not in w:
1178
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1179
# with InventoryEntry.find_previous_heads(). if it is, then there
1180
# is a latent bug here where the parents may have ancestors of each
1182
for revision, tree in parent_trees.iteritems():
1183
if ie.file_id not in tree:
1185
parent_id = tree.inventory[ie.file_id].revision
1186
if parent_id in text_parents:
1188
text_parents.append(parent_id)
1190
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1191
repository.get_transaction())
1192
lines = revision_tree.get_file(ie.file_id).readlines()
1193
vfile.add_lines(rev.revision_id, text_parents, lines)
1195
# install the inventory
1196
repository.add_inventory(rev.revision_id, inv, present_parents)
1197
except errors.RevisionAlreadyPresent:
1199
repository.add_revision(rev.revision_id, rev, inv)
1202
class MetaDirRepository(Repository):
1203
"""Repositories in the new meta-dir layout."""
1205
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1206
super(MetaDirRepository, self).__init__(_format,
1212
dir_mode = self.control_files._dir_mode
1213
file_mode = self.control_files._file_mode
1216
def is_shared(self):
1217
"""Return True if this repository is flagged as a shared repository."""
1218
return self.control_files._transport.has('shared-storage')
1221
def set_make_working_trees(self, new_value):
1222
"""Set the policy flag for making working trees when creating branches.
1224
This only applies to branches that use this repository.
1226
The default is 'True'.
1227
:param new_value: True to restore the default, False to disable making
1232
self.control_files._transport.delete('no-working-trees')
1233
except errors.NoSuchFile:
1236
self.control_files.put_utf8('no-working-trees', '')
1238
def make_working_trees(self):
1239
"""Returns the policy for making working trees on new branches."""
1240
return not self.control_files._transport.has('no-working-trees')
1243
class RepositoryFormatRegistry(registry.Registry):
1244
"""Registry of RepositoryFormats.
1247
def get(self, format_string):
1248
r = registry.Registry.get(self, format_string)
1254
format_registry = RepositoryFormatRegistry()
1255
"""Registry of formats, indexed by their identifying format string.
1257
This can contain either format instances themselves, or classes/factories that
1258
can be called to obtain one.
1262
#####################################################################
1263
# Repository Formats
1265
class RepositoryFormat(object):
1266
"""A repository format.
1268
Formats provide three things:
1269
* An initialization routine to construct repository data on disk.
1270
* a format string which is used when the BzrDir supports versioned
1272
* an open routine which returns a Repository instance.
1274
Formats are placed in an dict by their format string for reference
1275
during opening. These should be subclasses of RepositoryFormat
1278
Once a format is deprecated, just deprecate the initialize and open
1279
methods on the format class. Do not deprecate the object, as the
1280
object will be created every system load.
1282
Common instance attributes:
1283
_matchingbzrdir - the bzrdir format that the repository format was
1284
originally written to work with. This can be used if manually
1285
constructing a bzrdir and repository, or more commonly for test suite
1290
return "<%s>" % self.__class__.__name__
1292
def __eq__(self, other):
1293
# format objects are generally stateless
1294
return isinstance(other, self.__class__)
1296
def __ne__(self, other):
1297
return not self == other
1300
def find_format(klass, a_bzrdir):
1301
"""Return the format for the repository object in a_bzrdir.
1303
This is used by bzr native formats that have a "format" file in
1304
the repository. Other methods may be used by different types of
1308
transport = a_bzrdir.get_repository_transport(None)
1309
format_string = transport.get("format").read()
1310
return format_registry.get(format_string)
1311
except errors.NoSuchFile:
1312
raise errors.NoRepositoryPresent(a_bzrdir)
1314
raise errors.UnknownFormatError(format=format_string)
1317
def register_format(klass, format):
1318
format_registry.register(format.get_format_string(), format)
1321
def unregister_format(klass, format):
1322
format_registry.remove(format.get_format_string())
1325
def get_default_format(klass):
1326
"""Return the current default format."""
1327
from bzrlib import bzrdir
1328
return bzrdir.format_registry.make_bzrdir('default').repository_format
1330
def _get_control_store(self, repo_transport, control_files):
1331
"""Return the control store for this repository."""
1332
raise NotImplementedError(self._get_control_store)
1334
def get_format_string(self):
1335
"""Return the ASCII format string that identifies this format.
1337
Note that in pre format ?? repositories the format string is
1338
not permitted nor written to disk.
1340
raise NotImplementedError(self.get_format_string)
1342
def get_format_description(self):
1343
"""Return the short description for this format."""
1344
raise NotImplementedError(self.get_format_description)
1346
def _get_revision_store(self, repo_transport, control_files):
1347
"""Return the revision store object for this a_bzrdir."""
1348
raise NotImplementedError(self._get_revision_store)
1350
def _get_text_rev_store(self,
1357
"""Common logic for getting a revision store for a repository.
1359
see self._get_revision_store for the subclass-overridable method to
1360
get the store for a repository.
1362
from bzrlib.store.revision.text import TextRevisionStore
1363
dir_mode = control_files._dir_mode
1364
file_mode = control_files._file_mode
1365
text_store = TextStore(transport.clone(name),
1367
compressed=compressed,
1369
file_mode=file_mode)
1370
_revision_store = TextRevisionStore(text_store, serializer)
1371
return _revision_store
1373
# TODO: this shouldn't be in the base class, it's specific to things that
1374
# use weaves or knits -- mbp 20070207
1375
def _get_versioned_file_store(self,
1380
versionedfile_class=None,
1381
versionedfile_kwargs={},
1383
if versionedfile_class is None:
1384
versionedfile_class = self._versionedfile_class
1385
weave_transport = control_files._transport.clone(name)
1386
dir_mode = control_files._dir_mode
1387
file_mode = control_files._file_mode
1388
return VersionedFileStore(weave_transport, prefixed=prefixed,
1390
file_mode=file_mode,
1391
versionedfile_class=versionedfile_class,
1392
versionedfile_kwargs=versionedfile_kwargs,
1395
def initialize(self, a_bzrdir, shared=False):
1396
"""Initialize a repository of this format in a_bzrdir.
1398
:param a_bzrdir: The bzrdir to put the new repository in it.
1399
:param shared: The repository should be initialized as a sharable one.
1400
:returns: The new repository object.
1402
This may raise UninitializableFormat if shared repository are not
1403
compatible the a_bzrdir.
1405
raise NotImplementedError(self.initialize)
1407
def is_supported(self):
1408
"""Is this format supported?
1410
Supported formats must be initializable and openable.
1411
Unsupported formats may not support initialization or committing or
1412
some other features depending on the reason for not being supported.
1416
def check_conversion_target(self, target_format):
1417
raise NotImplementedError(self.check_conversion_target)
1419
def open(self, a_bzrdir, _found=False):
1420
"""Return an instance of this format for the bzrdir a_bzrdir.
1422
_found is a private parameter, do not use it.
1424
raise NotImplementedError(self.open)
1427
class MetaDirRepositoryFormat(RepositoryFormat):
1428
"""Common base class for the new repositories using the metadir layout."""
1430
rich_root_data = False
1431
supports_tree_reference = False
1432
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1435
super(MetaDirRepositoryFormat, self).__init__()
1437
def _create_control_files(self, a_bzrdir):
1438
"""Create the required files and the initial control_files object."""
1439
# FIXME: RBC 20060125 don't peek under the covers
1440
# NB: no need to escape relative paths that are url safe.
1441
repository_transport = a_bzrdir.get_repository_transport(self)
1442
control_files = lockable_files.LockableFiles(repository_transport,
1443
'lock', lockdir.LockDir)
1444
control_files.create_lock()
1445
return control_files
1447
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1448
"""Upload the initial blank content."""
1449
control_files = self._create_control_files(a_bzrdir)
1450
control_files.lock_write()
1452
control_files._transport.mkdir_multi(dirs,
1453
mode=control_files._dir_mode)
1454
for file, content in files:
1455
control_files.put(file, content)
1456
for file, content in utf8_files:
1457
control_files.put_utf8(file, content)
1459
control_files.put_utf8('shared-storage', '')
1461
control_files.unlock()
1464
# formats which have no format string are not discoverable
1465
# and not independently creatable, so are not registered. They're
1466
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1467
# needed, it's constructed directly by the BzrDir. Non-native formats where
1468
# the repository is not separately opened are similar.
1470
format_registry.register_lazy(
1471
'Bazaar-NG Repository format 7',
1472
'bzrlib.repofmt.weaverepo',
1475
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1476
# default control directory format
1478
format_registry.register_lazy(
1479
'Bazaar-NG Knit Repository Format 1',
1480
'bzrlib.repofmt.knitrepo',
1481
'RepositoryFormatKnit1',
1483
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1485
format_registry.register_lazy(
1486
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1487
'bzrlib.repofmt.knitrepo',
1488
'RepositoryFormatKnit3',
1492
class InterRepository(InterObject):
1493
"""This class represents operations taking place between two repositories.
1495
Its instances have methods like copy_content and fetch, and contain
1496
references to the source and target repositories these operations can be
1499
Often we will provide convenience methods on 'repository' which carry out
1500
operations with another repository - they will always forward to
1501
InterRepository.get(other).method_name(parameters).
1505
"""The available optimised InterRepository types."""
1507
def copy_content(self, revision_id=None):
1508
raise NotImplementedError(self.copy_content)
1510
def fetch(self, revision_id=None, pb=None):
1511
"""Fetch the content required to construct revision_id.
1513
The content is copied from self.source to self.target.
1515
:param revision_id: if None all content is copied, if NULL_REVISION no
1517
:param pb: optional progress bar to use for progress reports. If not
1518
provided a default one will be created.
1520
Returns the copied revision count and the failed revisions in a tuple:
1523
raise NotImplementedError(self.fetch)
1526
def missing_revision_ids(self, revision_id=None):
1527
"""Return the revision ids that source has that target does not.
1529
These are returned in topological order.
1531
:param revision_id: only return revision ids included by this
1534
# generic, possibly worst case, slow code path.
1535
target_ids = set(self.target.all_revision_ids())
1536
if revision_id is not None:
1537
# TODO: jam 20070210 InterRepository is internal enough that it
1538
# should assume revision_ids are already utf-8
1539
revision_id = osutils.safe_revision_id(revision_id)
1540
source_ids = self.source.get_ancestry(revision_id)
1541
assert source_ids[0] is None
1544
source_ids = self.source.all_revision_ids()
1545
result_set = set(source_ids).difference(target_ids)
1546
# this may look like a no-op: its not. It preserves the ordering
1547
# other_ids had while only returning the members from other_ids
1548
# that we've decided we need.
1549
return [rev_id for rev_id in source_ids if rev_id in result_set]
1552
class InterSameDataRepository(InterRepository):
1553
"""Code for converting between repositories that represent the same data.
1555
Data format and model must match for this to work.
1559
def _get_repo_format_to_test(self):
1560
"""Repository format for testing with."""
1561
return RepositoryFormat.get_default_format()
1564
def is_compatible(source, target):
1565
if source.supports_rich_root() != target.supports_rich_root():
1567
if source._serializer != target._serializer:
1572
def copy_content(self, revision_id=None):
1573
"""Make a complete copy of the content in self into destination.
1575
This copies both the repository's revision data, and configuration information
1576
such as the make_working_trees setting.
1578
This is a destructive operation! Do not use it on existing
1581
:param revision_id: Only copy the content needed to construct
1582
revision_id and its parents.
1585
self.target.set_make_working_trees(self.source.make_working_trees())
1586
except NotImplementedError:
1588
# TODO: jam 20070210 This is fairly internal, so we should probably
1589
# just assert that revision_id is not unicode.
1590
revision_id = osutils.safe_revision_id(revision_id)
1591
# but don't bother fetching if we have the needed data now.
1592
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1593
self.target.has_revision(revision_id)):
1595
self.target.fetch(self.source, revision_id=revision_id)
1598
def fetch(self, revision_id=None, pb=None):
1599
"""See InterRepository.fetch()."""
1600
from bzrlib.fetch import GenericRepoFetcher
1601
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1602
self.source, self.source._format, self.target,
1603
self.target._format)
1604
# TODO: jam 20070210 This should be an assert, not a translate
1605
revision_id = osutils.safe_revision_id(revision_id)
1606
f = GenericRepoFetcher(to_repository=self.target,
1607
from_repository=self.source,
1608
last_revision=revision_id,
1610
return f.count_copied, f.failed_revisions
1613
class InterWeaveRepo(InterSameDataRepository):
1614
"""Optimised code paths between Weave based repositories."""
1617
def _get_repo_format_to_test(self):
1618
from bzrlib.repofmt import weaverepo
1619
return weaverepo.RepositoryFormat7()
1622
def is_compatible(source, target):
1623
"""Be compatible with known Weave formats.
1625
We don't test for the stores being of specific types because that
1626
could lead to confusing results, and there is no need to be
1629
from bzrlib.repofmt.weaverepo import (
1635
return (isinstance(source._format, (RepositoryFormat5,
1637
RepositoryFormat7)) and
1638
isinstance(target._format, (RepositoryFormat5,
1640
RepositoryFormat7)))
1641
except AttributeError:
1645
def copy_content(self, revision_id=None):
1646
"""See InterRepository.copy_content()."""
1647
# weave specific optimised path:
1648
# TODO: jam 20070210 Internal, should be an assert, not translate
1649
revision_id = osutils.safe_revision_id(revision_id)
1651
self.target.set_make_working_trees(self.source.make_working_trees())
1652
except NotImplementedError:
1654
# FIXME do not peek!
1655
if self.source.control_files._transport.listable():
1656
pb = ui.ui_factory.nested_progress_bar()
1658
self.target.weave_store.copy_all_ids(
1659
self.source.weave_store,
1661
from_transaction=self.source.get_transaction(),
1662
to_transaction=self.target.get_transaction())
1663
pb.update('copying inventory', 0, 1)
1664
self.target.control_weaves.copy_multi(
1665
self.source.control_weaves, ['inventory'],
1666
from_transaction=self.source.get_transaction(),
1667
to_transaction=self.target.get_transaction())
1668
self.target._revision_store.text_store.copy_all_ids(
1669
self.source._revision_store.text_store,
1674
self.target.fetch(self.source, revision_id=revision_id)
1677
def fetch(self, revision_id=None, pb=None):
1678
"""See InterRepository.fetch()."""
1679
from bzrlib.fetch import GenericRepoFetcher
1680
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1681
self.source, self.source._format, self.target, self.target._format)
1682
# TODO: jam 20070210 This should be an assert, not a translate
1683
revision_id = osutils.safe_revision_id(revision_id)
1684
f = GenericRepoFetcher(to_repository=self.target,
1685
from_repository=self.source,
1686
last_revision=revision_id,
1688
return f.count_copied, f.failed_revisions
1691
def missing_revision_ids(self, revision_id=None):
1692
"""See InterRepository.missing_revision_ids()."""
1693
# we want all revisions to satisfy revision_id in source.
1694
# but we don't want to stat every file here and there.
1695
# we want then, all revisions other needs to satisfy revision_id
1696
# checked, but not those that we have locally.
1697
# so the first thing is to get a subset of the revisions to
1698
# satisfy revision_id in source, and then eliminate those that
1699
# we do already have.
1700
# this is slow on high latency connection to self, but as as this
1701
# disk format scales terribly for push anyway due to rewriting
1702
# inventory.weave, this is considered acceptable.
1704
if revision_id is not None:
1705
source_ids = self.source.get_ancestry(revision_id)
1706
assert source_ids[0] is None
1709
source_ids = self.source._all_possible_ids()
1710
source_ids_set = set(source_ids)
1711
# source_ids is the worst possible case we may need to pull.
1712
# now we want to filter source_ids against what we actually
1713
# have in target, but don't try to check for existence where we know
1714
# we do not have a revision as that would be pointless.
1715
target_ids = set(self.target._all_possible_ids())
1716
possibly_present_revisions = target_ids.intersection(source_ids_set)
1717
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1718
required_revisions = source_ids_set.difference(actually_present_revisions)
1719
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1720
if revision_id is not None:
1721
# we used get_ancestry to determine source_ids then we are assured all
1722
# revisions referenced are present as they are installed in topological order.
1723
# and the tip revision was validated by get_ancestry.
1724
return required_topo_revisions
1726
# if we just grabbed the possibly available ids, then
1727
# we only have an estimate of whats available and need to validate
1728
# that against the revision records.
1729
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1732
class InterKnitRepo(InterSameDataRepository):
1733
"""Optimised code paths between Knit based repositories."""
1736
def _get_repo_format_to_test(self):
1737
from bzrlib.repofmt import knitrepo
1738
return knitrepo.RepositoryFormatKnit1()
1741
def is_compatible(source, target):
1742
"""Be compatible with known Knit formats.
1744
We don't test for the stores being of specific types because that
1745
could lead to confusing results, and there is no need to be
1748
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1750
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1751
isinstance(target._format, (RepositoryFormatKnit1)))
1752
except AttributeError:
1756
def fetch(self, revision_id=None, pb=None):
1757
"""See InterRepository.fetch()."""
1758
from bzrlib.fetch import KnitRepoFetcher
1759
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1760
self.source, self.source._format, self.target, self.target._format)
1761
# TODO: jam 20070210 This should be an assert, not a translate
1762
revision_id = osutils.safe_revision_id(revision_id)
1763
f = KnitRepoFetcher(to_repository=self.target,
1764
from_repository=self.source,
1765
last_revision=revision_id,
1767
return f.count_copied, f.failed_revisions
1770
def missing_revision_ids(self, revision_id=None):
1771
"""See InterRepository.missing_revision_ids()."""
1772
if revision_id is not None:
1773
source_ids = self.source.get_ancestry(revision_id)
1774
assert source_ids[0] is None
1777
source_ids = self.source._all_possible_ids()
1778
source_ids_set = set(source_ids)
1779
# source_ids is the worst possible case we may need to pull.
1780
# now we want to filter source_ids against what we actually
1781
# have in target, but don't try to check for existence where we know
1782
# we do not have a revision as that would be pointless.
1783
target_ids = set(self.target._all_possible_ids())
1784
possibly_present_revisions = target_ids.intersection(source_ids_set)
1785
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1786
required_revisions = source_ids_set.difference(actually_present_revisions)
1787
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1788
if revision_id is not None:
1789
# we used get_ancestry to determine source_ids then we are assured all
1790
# revisions referenced are present as they are installed in topological order.
1791
# and the tip revision was validated by get_ancestry.
1792
return required_topo_revisions
1794
# if we just grabbed the possibly available ids, then
1795
# we only have an estimate of whats available and need to validate
1796
# that against the revision records.
1797
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1800
class InterModel1and2(InterRepository):
1803
def _get_repo_format_to_test(self):
1807
def is_compatible(source, target):
1808
if not source.supports_rich_root() and target.supports_rich_root():
1814
def fetch(self, revision_id=None, pb=None):
1815
"""See InterRepository.fetch()."""
1816
from bzrlib.fetch import Model1toKnit2Fetcher
1817
# TODO: jam 20070210 This should be an assert, not a translate
1818
revision_id = osutils.safe_revision_id(revision_id)
1819
f = Model1toKnit2Fetcher(to_repository=self.target,
1820
from_repository=self.source,
1821
last_revision=revision_id,
1823
return f.count_copied, f.failed_revisions
1826
def copy_content(self, revision_id=None):
1827
"""Make a complete copy of the content in self into destination.
1829
This is a destructive operation! Do not use it on existing
1832
:param revision_id: Only copy the content needed to construct
1833
revision_id and its parents.
1836
self.target.set_make_working_trees(self.source.make_working_trees())
1837
except NotImplementedError:
1839
# TODO: jam 20070210 Internal, assert, don't translate
1840
revision_id = osutils.safe_revision_id(revision_id)
1841
# but don't bother fetching if we have the needed data now.
1842
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1843
self.target.has_revision(revision_id)):
1845
self.target.fetch(self.source, revision_id=revision_id)
1848
class InterKnit1and2(InterKnitRepo):
1851
def _get_repo_format_to_test(self):
1855
def is_compatible(source, target):
1856
"""Be compatible with Knit1 source and Knit3 target"""
1857
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1859
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1860
RepositoryFormatKnit3
1861
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1862
isinstance(target._format, (RepositoryFormatKnit3)))
1863
except AttributeError:
1867
def fetch(self, revision_id=None, pb=None):
1868
"""See InterRepository.fetch()."""
1869
from bzrlib.fetch import Knit1to2Fetcher
1870
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1871
self.source, self.source._format, self.target,
1872
self.target._format)
1873
# TODO: jam 20070210 This should be an assert, not a translate
1874
revision_id = osutils.safe_revision_id(revision_id)
1875
f = Knit1to2Fetcher(to_repository=self.target,
1876
from_repository=self.source,
1877
last_revision=revision_id,
1879
return f.count_copied, f.failed_revisions
1882
class InterRemoteRepository(InterRepository):
1883
"""Code for converting between RemoteRepository objects.
1885
This just gets an non-remote repository from the RemoteRepository, and calls
1886
InterRepository.get again.
1889
def __init__(self, source, target):
1890
if isinstance(source, remote.RemoteRepository):
1891
source._ensure_real()
1892
real_source = source._real_repository
1894
real_source = source
1895
if isinstance(target, remote.RemoteRepository):
1896
target._ensure_real()
1897
real_target = target._real_repository
1899
real_target = target
1900
self.real_inter = InterRepository.get(real_source, real_target)
1903
def is_compatible(source, target):
1904
if isinstance(source, remote.RemoteRepository):
1906
if isinstance(target, remote.RemoteRepository):
1910
def copy_content(self, revision_id=None):
1911
self.real_inter.copy_content(revision_id=revision_id)
1913
def fetch(self, revision_id=None, pb=None):
1914
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1917
def _get_repo_format_to_test(self):
1921
InterRepository.register_optimiser(InterSameDataRepository)
1922
InterRepository.register_optimiser(InterWeaveRepo)
1923
InterRepository.register_optimiser(InterKnitRepo)
1924
InterRepository.register_optimiser(InterModel1and2)
1925
InterRepository.register_optimiser(InterKnit1and2)
1926
InterRepository.register_optimiser(InterRemoteRepository)
1929
class CopyConverter(object):
1930
"""A repository conversion tool which just performs a copy of the content.
1932
This is slow but quite reliable.
1935
def __init__(self, target_format):
1936
"""Create a CopyConverter.
1938
:param target_format: The format the resulting repository should be.
1940
self.target_format = target_format
1942
def convert(self, repo, pb):
1943
"""Perform the conversion of to_convert, giving feedback via pb.
1945
:param to_convert: The disk object to convert.
1946
:param pb: a progress bar to use for progress information.
1951
# this is only useful with metadir layouts - separated repo content.
1952
# trigger an assertion if not such
1953
repo._format.get_format_string()
1954
self.repo_dir = repo.bzrdir
1955
self.step('Moving repository to repository.backup')
1956
self.repo_dir.transport.move('repository', 'repository.backup')
1957
backup_transport = self.repo_dir.transport.clone('repository.backup')
1958
repo._format.check_conversion_target(self.target_format)
1959
self.source_repo = repo._format.open(self.repo_dir,
1961
_override_transport=backup_transport)
1962
self.step('Creating new repository')
1963
converted = self.target_format.initialize(self.repo_dir,
1964
self.source_repo.is_shared())
1965
converted.lock_write()
1967
self.step('Copying content into repository.')
1968
self.source_repo.copy_content_into(converted)
1971
self.step('Deleting old repository content.')
1972
self.repo_dir.transport.delete_tree('repository.backup')
1973
self.pb.note('repository converted')
1975
def step(self, message):
1976
"""Update the pb by a step."""
1978
self.pb.update(message, self.count, self.total)
1981
class CommitBuilder(object):
1982
"""Provides an interface to build up a commit.
1984
This allows describing a tree to be committed without needing to
1985
know the internals of the format of the repository.
1988
record_root_entry = False
1989
def __init__(self, repository, parents, config, timestamp=None,
1990
timezone=None, committer=None, revprops=None,
1992
"""Initiate a CommitBuilder.
1994
:param repository: Repository to commit to.
1995
:param parents: Revision ids of the parents of the new revision.
1996
:param config: Configuration to use.
1997
:param timestamp: Optional timestamp recorded for commit.
1998
:param timezone: Optional timezone for timestamp.
1999
:param committer: Optional committer to set for commit.
2000
:param revprops: Optional dictionary of revision properties.
2001
:param revision_id: Optional revision id.
2003
self._config = config
2005
if committer is None:
2006
self._committer = self._config.username()
2008
assert isinstance(committer, basestring), type(committer)
2009
self._committer = committer
2011
self.new_inventory = Inventory(None)
2012
self._new_revision_id = osutils.safe_revision_id(revision_id)
2013
self.parents = parents
2014
self.repository = repository
2017
if revprops is not None:
2018
self._revprops.update(revprops)
2020
if timestamp is None:
2021
timestamp = time.time()
2022
# Restrict resolution to 1ms
2023
self._timestamp = round(timestamp, 3)
2025
if timezone is None:
2026
self._timezone = osutils.local_time_offset()
2028
self._timezone = int(timezone)
2030
self._generate_revision_if_needed()
2032
def commit(self, message):
2033
"""Make the actual commit.
2035
:return: The revision id of the recorded revision.
2037
rev = _mod_revision.Revision(
2038
timestamp=self._timestamp,
2039
timezone=self._timezone,
2040
committer=self._committer,
2042
inventory_sha1=self.inv_sha1,
2043
revision_id=self._new_revision_id,
2044
properties=self._revprops)
2045
rev.parent_ids = self.parents
2046
self.repository.add_revision(self._new_revision_id, rev,
2047
self.new_inventory, self._config)
2048
self.repository.commit_write_group()
2049
return self._new_revision_id
2051
def revision_tree(self):
2052
"""Return the tree that was just committed.
2054
After calling commit() this can be called to get a RevisionTree
2055
representing the newly committed tree. This is preferred to
2056
calling Repository.revision_tree() because that may require
2057
deserializing the inventory, while we already have a copy in
2060
return RevisionTree(self.repository, self.new_inventory,
2061
self._new_revision_id)
2063
def finish_inventory(self):
2064
"""Tell the builder that the inventory is finished."""
2065
if self.new_inventory.root is None:
2066
symbol_versioning.warn('Root entry should be supplied to'
2067
' record_entry_contents, as of bzr 0.10.',
2068
DeprecationWarning, stacklevel=2)
2069
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2070
self.new_inventory.revision_id = self._new_revision_id
2071
self.inv_sha1 = self.repository.add_inventory(
2072
self._new_revision_id,
2077
def _gen_revision_id(self):
2078
"""Return new revision-id."""
2079
return generate_ids.gen_revision_id(self._config.username(),
2082
def _generate_revision_if_needed(self):
2083
"""Create a revision id if None was supplied.
2085
If the repository can not support user-specified revision ids
2086
they should override this function and raise CannotSetRevisionId
2087
if _new_revision_id is not None.
2089
:raises: CannotSetRevisionId
2091
if self._new_revision_id is None:
2092
self._new_revision_id = self._gen_revision_id()
2094
def record_entry_contents(self, ie, parent_invs, path, tree):
2095
"""Record the content of ie from tree into the commit if needed.
2097
Side effect: sets ie.revision when unchanged
2099
:param ie: An inventory entry present in the commit.
2100
:param parent_invs: The inventories of the parent revisions of the
2102
:param path: The path the entry is at in the tree.
2103
:param tree: The tree which contains this entry and should be used to
2106
if self.new_inventory.root is None and ie.parent_id is not None:
2107
symbol_versioning.warn('Root entry should be supplied to'
2108
' record_entry_contents, as of bzr 0.10.',
2109
DeprecationWarning, stacklevel=2)
2110
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2112
self.new_inventory.add(ie)
2114
# ie.revision is always None if the InventoryEntry is considered
2115
# for committing. ie.snapshot will record the correct revision
2116
# which may be the sole parent if it is untouched.
2117
if ie.revision is not None:
2120
# In this revision format, root entries have no knit or weave
2121
if ie is self.new_inventory.root:
2122
# When serializing out to disk and back in
2123
# root.revision is always _new_revision_id
2124
ie.revision = self._new_revision_id
2126
previous_entries = ie.find_previous_heads(
2128
self.repository.weave_store,
2129
self.repository.get_transaction())
2130
# we are creating a new revision for ie in the history store
2132
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2134
def modified_directory(self, file_id, file_parents):
2135
"""Record the presence of a symbolic link.
2137
:param file_id: The file_id of the link to record.
2138
:param file_parents: The per-file parent revision ids.
2140
self._add_text_to_weave(file_id, [], file_parents.keys())
2142
def modified_reference(self, file_id, file_parents):
2143
"""Record the modification of a reference.
2145
:param file_id: The file_id of the link to record.
2146
:param file_parents: The per-file parent revision ids.
2148
self._add_text_to_weave(file_id, [], file_parents.keys())
2150
def modified_file_text(self, file_id, file_parents,
2151
get_content_byte_lines, text_sha1=None,
2153
"""Record the text of file file_id
2155
:param file_id: The file_id of the file to record the text of.
2156
:param file_parents: The per-file parent revision ids.
2157
:param get_content_byte_lines: A callable which will return the byte
2159
:param text_sha1: Optional SHA1 of the file contents.
2160
:param text_size: Optional size of the file contents.
2162
# mutter('storing text of file {%s} in revision {%s} into %r',
2163
# file_id, self._new_revision_id, self.repository.weave_store)
2164
# special case to avoid diffing on renames or
2166
if (len(file_parents) == 1
2167
and text_sha1 == file_parents.values()[0].text_sha1
2168
and text_size == file_parents.values()[0].text_size):
2169
previous_ie = file_parents.values()[0]
2170
versionedfile = self.repository.weave_store.get_weave(file_id,
2171
self.repository.get_transaction())
2172
versionedfile.clone_text(self._new_revision_id,
2173
previous_ie.revision, file_parents.keys())
2174
return text_sha1, text_size
2176
new_lines = get_content_byte_lines()
2177
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2178
# should return the SHA1 and size
2179
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2180
return osutils.sha_strings(new_lines), \
2181
sum(map(len, new_lines))
2183
def modified_link(self, file_id, file_parents, link_target):
2184
"""Record the presence of a symbolic link.
2186
:param file_id: The file_id of the link to record.
2187
:param file_parents: The per-file parent revision ids.
2188
:param link_target: Target location of this link.
2190
self._add_text_to_weave(file_id, [], file_parents.keys())
2192
def _add_text_to_weave(self, file_id, new_lines, parents):
2193
versionedfile = self.repository.weave_store.get_weave_or_empty(
2194
file_id, self.repository.get_transaction())
2195
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2196
versionedfile.clear_cache()
2199
class _CommitBuilder(CommitBuilder):
2200
"""Temporary class so old CommitBuilders are detected properly
2202
Note: CommitBuilder works whether or not root entry is recorded.
2205
record_root_entry = True
2208
class RootCommitBuilder(CommitBuilder):
2209
"""This commitbuilder actually records the root id"""
2211
record_root_entry = True
2213
def record_entry_contents(self, ie, parent_invs, path, tree):
2214
"""Record the content of ie from tree into the commit if needed.
2216
Side effect: sets ie.revision when unchanged
2218
:param ie: An inventory entry present in the commit.
2219
:param parent_invs: The inventories of the parent revisions of the
2221
:param path: The path the entry is at in the tree.
2222
:param tree: The tree which contains this entry and should be used to
2225
assert self.new_inventory.root is not None or ie.parent_id is None
2226
self.new_inventory.add(ie)
2228
# ie.revision is always None if the InventoryEntry is considered
2229
# for committing. ie.snapshot will record the correct revision
2230
# which may be the sole parent if it is untouched.
2231
if ie.revision is not None:
2234
previous_entries = ie.find_previous_heads(
2236
self.repository.weave_store,
2237
self.repository.get_transaction())
2238
# we are creating a new revision for ie in the history store
2240
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2252
def _unescaper(match, _map=_unescape_map):
2253
code = match.group(1)
2257
if not code.startswith('#'):
2259
return unichr(int(code[1:])).encode('utf8')
2265
def _unescape_xml(data):
2266
"""Unescape predefined XML entities in a string of data."""
2268
if _unescape_re is None:
2269
_unescape_re = re.compile('\&([^;]*);')
2270
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
