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)
734
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
735
"""Get an iterable listing the keys of all the data introduced by a set
738
The keys will be ordered so that the corresponding items can be safely
739
fetched and inserted in that order.
741
:returns: An iterable producing tuples of (knit-kind, file-id,
742
versions). knit-kind is one of 'file', 'inventory', 'signatures',
743
'revisions'. file-id is None unless knit-kind is 'file'.
745
# XXX: it's a bit weird to control the inventory weave caching in this
746
# generator. Ideally the caching would be done in fetch.py I think. Or
747
# maybe this generator should explicitly have the contract that it
748
# should not be iterated until the previously yielded item has been
750
inv_w = self.get_inventory_weave()
753
# file ids that changed
754
file_ids = self.fileids_altered_by_revision_ids(revision_ids)
756
num_file_ids = len(file_ids)
757
for file_id, altered_versions in file_ids.iteritems():
758
if _files_pb is not None:
759
_files_pb.update("fetch texts", count, num_file_ids)
761
yield ("file", file_id, altered_versions)
762
# We're done with the files_pb. Note that it finished by the caller,
763
# just as it was created by the caller.
767
yield ("inventory", None, revision_ids)
771
revisions_with_signatures = set()
772
for rev_id in revision_ids:
774
self.get_signature_text(rev_id)
775
except errors.NoSuchRevision:
779
revisions_with_signatures.add(rev_id)
780
yield ("signatures", None, revisions_with_signatures)
783
yield ("revisions", None, revision_ids)
786
def get_inventory_weave(self):
787
return self.control_weaves.get_weave('inventory',
788
self.get_transaction())
791
def get_inventory(self, revision_id):
792
"""Get Inventory object by hash."""
793
# TODO: jam 20070210 Technically we don't need to sanitize, since all
794
# called functions must sanitize.
795
revision_id = osutils.safe_revision_id(revision_id)
796
return self.deserialise_inventory(
797
revision_id, self.get_inventory_xml(revision_id))
799
def deserialise_inventory(self, revision_id, xml):
800
"""Transform the xml into an inventory object.
802
:param revision_id: The expected revision id of the inventory.
803
:param xml: A serialised inventory.
805
revision_id = osutils.safe_revision_id(revision_id)
806
result = self._serializer.read_inventory_from_string(xml)
807
result.root.revision = revision_id
810
def serialise_inventory(self, inv):
811
return self._serializer.write_inventory_to_string(inv)
813
def get_serializer_format(self):
814
return self._serializer.format_num
817
def get_inventory_xml(self, revision_id):
818
"""Get inventory XML as a file object."""
819
revision_id = osutils.safe_revision_id(revision_id)
821
assert isinstance(revision_id, str), type(revision_id)
822
iw = self.get_inventory_weave()
823
return iw.get_text(revision_id)
825
raise errors.HistoryMissing(self, 'inventory', revision_id)
828
def get_inventory_sha1(self, revision_id):
829
"""Return the sha1 hash of the inventory entry
831
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
832
revision_id = osutils.safe_revision_id(revision_id)
833
return self.get_revision(revision_id).inventory_sha1
836
def get_revision_graph(self, revision_id=None):
837
"""Return a dictionary containing the revision graph.
839
:param revision_id: The revision_id to get a graph from. If None, then
840
the entire revision graph is returned. This is a deprecated mode of
841
operation and will be removed in the future.
842
:return: a dictionary of revision_id->revision_parents_list.
844
# special case NULL_REVISION
845
if revision_id == _mod_revision.NULL_REVISION:
847
revision_id = osutils.safe_revision_id(revision_id)
848
a_weave = self.get_inventory_weave()
849
all_revisions = self._eliminate_revisions_not_present(
851
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
852
node in all_revisions])
853
if revision_id is None:
855
elif revision_id not in entire_graph:
856
raise errors.NoSuchRevision(self, revision_id)
858
# add what can be reached from revision_id
860
pending = set([revision_id])
861
while len(pending) > 0:
863
result[node] = entire_graph[node]
864
for revision_id in result[node]:
865
if revision_id not in result:
866
pending.add(revision_id)
870
def get_revision_graph_with_ghosts(self, revision_ids=None):
871
"""Return a graph of the revisions with ghosts marked as applicable.
873
:param revision_ids: an iterable of revisions to graph or None for all.
874
:return: a Graph object with the graph reachable from revision_ids.
876
result = deprecated_graph.Graph()
878
pending = set(self.all_revision_ids())
881
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
882
# special case NULL_REVISION
883
if _mod_revision.NULL_REVISION in pending:
884
pending.remove(_mod_revision.NULL_REVISION)
885
required = set(pending)
888
revision_id = pending.pop()
890
rev = self.get_revision(revision_id)
891
except errors.NoSuchRevision:
892
if revision_id in required:
895
result.add_ghost(revision_id)
897
for parent_id in rev.parent_ids:
898
# is this queued or done ?
899
if (parent_id not in pending and
900
parent_id not in done):
902
pending.add(parent_id)
903
result.add_node(revision_id, rev.parent_ids)
904
done.add(revision_id)
907
def _get_history_vf(self):
908
"""Get a versionedfile whose history graph reflects all revisions.
910
For weave repositories, this is the inventory weave.
912
return self.get_inventory_weave()
914
def iter_reverse_revision_history(self, revision_id):
915
"""Iterate backwards through revision ids in the lefthand history
917
:param revision_id: The revision id to start with. All its lefthand
918
ancestors will be traversed.
920
revision_id = osutils.safe_revision_id(revision_id)
921
if revision_id in (None, _mod_revision.NULL_REVISION):
923
next_id = revision_id
924
versionedfile = self._get_history_vf()
927
parents = versionedfile.get_parents(next_id)
928
if len(parents) == 0:
934
def get_revision_inventory(self, revision_id):
935
"""Return inventory of a past revision."""
936
# TODO: Unify this with get_inventory()
937
# bzr 0.0.6 and later imposes the constraint that the inventory_id
938
# must be the same as its revision, so this is trivial.
939
if revision_id is None:
940
# This does not make sense: if there is no revision,
941
# then it is the current tree inventory surely ?!
942
# and thus get_root_id() is something that looks at the last
943
# commit on the branch, and the get_root_id is an inventory check.
944
raise NotImplementedError
945
# return Inventory(self.get_root_id())
947
return self.get_inventory(revision_id)
951
"""Return True if this repository is flagged as a shared repository."""
952
raise NotImplementedError(self.is_shared)
955
def reconcile(self, other=None, thorough=False):
956
"""Reconcile this repository."""
957
from bzrlib.reconcile import RepoReconciler
958
reconciler = RepoReconciler(self, thorough=thorough)
959
reconciler.reconcile()
962
def _refresh_data(self):
963
"""Helper called from lock_* to ensure coherency with disk.
965
The default implementation does nothing; it is however possible
966
for repositories to maintain loaded indices across multiple locks
967
by checking inside their implementation of this method to see
968
whether their indices are still valid. This depends of course on
969
the disk format being validatable in this manner.
973
def revision_tree(self, revision_id):
974
"""Return Tree for a revision on this branch.
976
`revision_id` may be None for the empty tree revision.
978
# TODO: refactor this to use an existing revision object
979
# so we don't need to read it in twice.
980
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
981
return RevisionTree(self, Inventory(root_id=None),
982
_mod_revision.NULL_REVISION)
984
revision_id = osutils.safe_revision_id(revision_id)
985
inv = self.get_revision_inventory(revision_id)
986
return RevisionTree(self, inv, revision_id)
989
def revision_trees(self, revision_ids):
990
"""Return Tree for a revision on this branch.
992
`revision_id` may not be None or 'null:'"""
993
assert None not in revision_ids
994
assert _mod_revision.NULL_REVISION not in revision_ids
995
texts = self.get_inventory_weave().get_texts(revision_ids)
996
for text, revision_id in zip(texts, revision_ids):
997
inv = self.deserialise_inventory(revision_id, text)
998
yield RevisionTree(self, inv, revision_id)
1001
def get_ancestry(self, revision_id, topo_sorted=True):
1002
"""Return a list of revision-ids integrated by a revision.
1004
The first element of the list is always None, indicating the origin
1005
revision. This might change when we have history horizons, or
1006
perhaps we should have a new API.
1008
This is topologically sorted.
1010
if _mod_revision.is_null(revision_id):
1012
revision_id = osutils.safe_revision_id(revision_id)
1013
if not self.has_revision(revision_id):
1014
raise errors.NoSuchRevision(self, revision_id)
1015
w = self.get_inventory_weave()
1016
candidates = w.get_ancestry(revision_id, topo_sorted)
1017
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1020
"""Compress the data within the repository.
1022
This operation only makes sense for some repository types. For other
1023
types it should be a no-op that just returns.
1025
This stub method does not require a lock, but subclasses should use
1026
@needs_write_lock as this is a long running call its reasonable to
1027
implicitly lock for the user.
1031
def print_file(self, file, revision_id):
1032
"""Print `file` to stdout.
1034
FIXME RBC 20060125 as John Meinel points out this is a bad api
1035
- it writes to stdout, it assumes that that is valid etc. Fix
1036
by creating a new more flexible convenience function.
1038
revision_id = osutils.safe_revision_id(revision_id)
1039
tree = self.revision_tree(revision_id)
1040
# use inventory as it was in that revision
1041
file_id = tree.inventory.path2id(file)
1043
# TODO: jam 20060427 Write a test for this code path
1044
# it had a bug in it, and was raising the wrong
1046
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1047
tree.print_file(file_id)
1049
def get_transaction(self):
1050
return self.control_files.get_transaction()
1052
def revision_parents(self, revision_id):
1053
revision_id = osutils.safe_revision_id(revision_id)
1054
return self.get_inventory_weave().parent_names(revision_id)
1056
def get_parents(self, revision_ids):
1057
"""See StackedParentsProvider.get_parents"""
1059
for revision_id in revision_ids:
1060
if revision_id == _mod_revision.NULL_REVISION:
1064
parents = self.get_revision(revision_id).parent_ids
1065
except errors.NoSuchRevision:
1068
if len(parents) == 0:
1069
parents = [_mod_revision.NULL_REVISION]
1070
parents_list.append(parents)
1073
def _make_parents_provider(self):
1076
def get_graph(self, other_repository=None):
1077
"""Return the graph walker for this repository format"""
1078
parents_provider = self._make_parents_provider()
1079
if (other_repository is not None and
1080
other_repository.bzrdir.transport.base !=
1081
self.bzrdir.transport.base):
1082
parents_provider = graph._StackedParentsProvider(
1083
[parents_provider, other_repository._make_parents_provider()])
1084
return graph.Graph(parents_provider)
1087
def set_make_working_trees(self, new_value):
1088
"""Set the policy flag for making working trees when creating branches.
1090
This only applies to branches that use this repository.
1092
The default is 'True'.
1093
:param new_value: True to restore the default, False to disable making
1096
raise NotImplementedError(self.set_make_working_trees)
1098
def make_working_trees(self):
1099
"""Returns the policy for making working trees on new branches."""
1100
raise NotImplementedError(self.make_working_trees)
1103
def sign_revision(self, revision_id, gpg_strategy):
1104
revision_id = osutils.safe_revision_id(revision_id)
1105
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1106
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1109
def has_signature_for_revision_id(self, revision_id):
1110
"""Query for a revision signature for revision_id in the repository."""
1111
revision_id = osutils.safe_revision_id(revision_id)
1112
return self._revision_store.has_signature(revision_id,
1113
self.get_transaction())
1116
def get_signature_text(self, revision_id):
1117
"""Return the text for a signature."""
1118
revision_id = osutils.safe_revision_id(revision_id)
1119
return self._revision_store.get_signature_text(revision_id,
1120
self.get_transaction())
1123
def check(self, revision_ids):
1124
"""Check consistency of all history of given revision_ids.
1126
Different repository implementations should override _check().
1128
:param revision_ids: A non-empty list of revision_ids whose ancestry
1129
will be checked. Typically the last revision_id of a branch.
1131
if not revision_ids:
1132
raise ValueError("revision_ids must be non-empty in %s.check"
1134
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1135
return self._check(revision_ids)
1137
def _check(self, revision_ids):
1138
result = check.Check(self)
1142
def _warn_if_deprecated(self):
1143
global _deprecation_warning_done
1144
if _deprecation_warning_done:
1146
_deprecation_warning_done = True
1147
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1148
% (self._format, self.bzrdir.transport.base))
1150
def supports_rich_root(self):
1151
return self._format.rich_root_data
1153
def _check_ascii_revisionid(self, revision_id, method):
1154
"""Private helper for ascii-only repositories."""
1155
# weave repositories refuse to store revisionids that are non-ascii.
1156
if revision_id is not None:
1157
# weaves require ascii revision ids.
1158
if isinstance(revision_id, unicode):
1160
revision_id.encode('ascii')
1161
except UnicodeEncodeError:
1162
raise errors.NonAsciiRevisionId(method, self)
1165
revision_id.decode('ascii')
1166
except UnicodeDecodeError:
1167
raise errors.NonAsciiRevisionId(method, self)
1171
# remove these delegates a while after bzr 0.15
1172
def __make_delegated(name, from_module):
1173
def _deprecated_repository_forwarder():
1174
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1175
% (name, from_module),
1178
m = __import__(from_module, globals(), locals(), [name])
1180
return getattr(m, name)
1181
except AttributeError:
1182
raise AttributeError('module %s has no name %s'
1184
globals()[name] = _deprecated_repository_forwarder
1187
'AllInOneRepository',
1188
'WeaveMetaDirRepository',
1189
'PreSplitOutRepositoryFormat',
1190
'RepositoryFormat4',
1191
'RepositoryFormat5',
1192
'RepositoryFormat6',
1193
'RepositoryFormat7',
1195
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1199
'RepositoryFormatKnit',
1200
'RepositoryFormatKnit1',
1202
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1205
def install_revision(repository, rev, revision_tree):
1206
"""Install all revision data into a repository."""
1207
present_parents = []
1209
for p_id in rev.parent_ids:
1210
if repository.has_revision(p_id):
1211
present_parents.append(p_id)
1212
parent_trees[p_id] = repository.revision_tree(p_id)
1214
parent_trees[p_id] = repository.revision_tree(None)
1216
inv = revision_tree.inventory
1217
entries = inv.iter_entries()
1218
# backwards compatibility hack: skip the root id.
1219
if not repository.supports_rich_root():
1220
path, root = entries.next()
1221
if root.revision != rev.revision_id:
1222
raise errors.IncompatibleRevision(repr(repository))
1223
# Add the texts that are not already present
1224
for path, ie in entries:
1225
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1226
repository.get_transaction())
1227
if ie.revision not in w:
1229
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1230
# with InventoryEntry.find_previous_heads(). if it is, then there
1231
# is a latent bug here where the parents may have ancestors of each
1233
for revision, tree in parent_trees.iteritems():
1234
if ie.file_id not in tree:
1236
parent_id = tree.inventory[ie.file_id].revision
1237
if parent_id in text_parents:
1239
text_parents.append(parent_id)
1241
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1242
repository.get_transaction())
1243
lines = revision_tree.get_file(ie.file_id).readlines()
1244
vfile.add_lines(rev.revision_id, text_parents, lines)
1246
# install the inventory
1247
repository.add_inventory(rev.revision_id, inv, present_parents)
1248
except errors.RevisionAlreadyPresent:
1250
repository.add_revision(rev.revision_id, rev, inv)
1253
class MetaDirRepository(Repository):
1254
"""Repositories in the new meta-dir layout."""
1256
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1257
super(MetaDirRepository, self).__init__(_format,
1263
dir_mode = self.control_files._dir_mode
1264
file_mode = self.control_files._file_mode
1267
def is_shared(self):
1268
"""Return True if this repository is flagged as a shared repository."""
1269
return self.control_files._transport.has('shared-storage')
1272
def set_make_working_trees(self, new_value):
1273
"""Set the policy flag for making working trees when creating branches.
1275
This only applies to branches that use this repository.
1277
The default is 'True'.
1278
:param new_value: True to restore the default, False to disable making
1283
self.control_files._transport.delete('no-working-trees')
1284
except errors.NoSuchFile:
1287
self.control_files.put_utf8('no-working-trees', '')
1289
def make_working_trees(self):
1290
"""Returns the policy for making working trees on new branches."""
1291
return not self.control_files._transport.has('no-working-trees')
1294
class RepositoryFormatRegistry(registry.Registry):
1295
"""Registry of RepositoryFormats.
1298
def get(self, format_string):
1299
r = registry.Registry.get(self, format_string)
1305
format_registry = RepositoryFormatRegistry()
1306
"""Registry of formats, indexed by their identifying format string.
1308
This can contain either format instances themselves, or classes/factories that
1309
can be called to obtain one.
1313
#####################################################################
1314
# Repository Formats
1316
class RepositoryFormat(object):
1317
"""A repository format.
1319
Formats provide three things:
1320
* An initialization routine to construct repository data on disk.
1321
* a format string which is used when the BzrDir supports versioned
1323
* an open routine which returns a Repository instance.
1325
Formats are placed in an dict by their format string for reference
1326
during opening. These should be subclasses of RepositoryFormat
1329
Once a format is deprecated, just deprecate the initialize and open
1330
methods on the format class. Do not deprecate the object, as the
1331
object will be created every system load.
1333
Common instance attributes:
1334
_matchingbzrdir - the bzrdir format that the repository format was
1335
originally written to work with. This can be used if manually
1336
constructing a bzrdir and repository, or more commonly for test suite
1341
return "<%s>" % self.__class__.__name__
1343
def __eq__(self, other):
1344
# format objects are generally stateless
1345
return isinstance(other, self.__class__)
1347
def __ne__(self, other):
1348
return not self == other
1351
def find_format(klass, a_bzrdir):
1352
"""Return the format for the repository object in a_bzrdir.
1354
This is used by bzr native formats that have a "format" file in
1355
the repository. Other methods may be used by different types of
1359
transport = a_bzrdir.get_repository_transport(None)
1360
format_string = transport.get("format").read()
1361
return format_registry.get(format_string)
1362
except errors.NoSuchFile:
1363
raise errors.NoRepositoryPresent(a_bzrdir)
1365
raise errors.UnknownFormatError(format=format_string)
1368
def register_format(klass, format):
1369
format_registry.register(format.get_format_string(), format)
1372
def unregister_format(klass, format):
1373
format_registry.remove(format.get_format_string())
1376
def get_default_format(klass):
1377
"""Return the current default format."""
1378
from bzrlib import bzrdir
1379
return bzrdir.format_registry.make_bzrdir('default').repository_format
1381
def _get_control_store(self, repo_transport, control_files):
1382
"""Return the control store for this repository."""
1383
raise NotImplementedError(self._get_control_store)
1385
def get_format_string(self):
1386
"""Return the ASCII format string that identifies this format.
1388
Note that in pre format ?? repositories the format string is
1389
not permitted nor written to disk.
1391
raise NotImplementedError(self.get_format_string)
1393
def get_format_description(self):
1394
"""Return the short description for this format."""
1395
raise NotImplementedError(self.get_format_description)
1397
def _get_revision_store(self, repo_transport, control_files):
1398
"""Return the revision store object for this a_bzrdir."""
1399
raise NotImplementedError(self._get_revision_store)
1401
def _get_text_rev_store(self,
1408
"""Common logic for getting a revision store for a repository.
1410
see self._get_revision_store for the subclass-overridable method to
1411
get the store for a repository.
1413
from bzrlib.store.revision.text import TextRevisionStore
1414
dir_mode = control_files._dir_mode
1415
file_mode = control_files._file_mode
1416
text_store = TextStore(transport.clone(name),
1418
compressed=compressed,
1420
file_mode=file_mode)
1421
_revision_store = TextRevisionStore(text_store, serializer)
1422
return _revision_store
1424
# TODO: this shouldn't be in the base class, it's specific to things that
1425
# use weaves or knits -- mbp 20070207
1426
def _get_versioned_file_store(self,
1431
versionedfile_class=None,
1432
versionedfile_kwargs={},
1434
if versionedfile_class is None:
1435
versionedfile_class = self._versionedfile_class
1436
weave_transport = control_files._transport.clone(name)
1437
dir_mode = control_files._dir_mode
1438
file_mode = control_files._file_mode
1439
return VersionedFileStore(weave_transport, prefixed=prefixed,
1441
file_mode=file_mode,
1442
versionedfile_class=versionedfile_class,
1443
versionedfile_kwargs=versionedfile_kwargs,
1446
def initialize(self, a_bzrdir, shared=False):
1447
"""Initialize a repository of this format in a_bzrdir.
1449
:param a_bzrdir: The bzrdir to put the new repository in it.
1450
:param shared: The repository should be initialized as a sharable one.
1451
:returns: The new repository object.
1453
This may raise UninitializableFormat if shared repository are not
1454
compatible the a_bzrdir.
1456
raise NotImplementedError(self.initialize)
1458
def is_supported(self):
1459
"""Is this format supported?
1461
Supported formats must be initializable and openable.
1462
Unsupported formats may not support initialization or committing or
1463
some other features depending on the reason for not being supported.
1467
def check_conversion_target(self, target_format):
1468
raise NotImplementedError(self.check_conversion_target)
1470
def open(self, a_bzrdir, _found=False):
1471
"""Return an instance of this format for the bzrdir a_bzrdir.
1473
_found is a private parameter, do not use it.
1475
raise NotImplementedError(self.open)
1478
class MetaDirRepositoryFormat(RepositoryFormat):
1479
"""Common base class for the new repositories using the metadir layout."""
1481
rich_root_data = False
1482
supports_tree_reference = False
1483
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1486
super(MetaDirRepositoryFormat, self).__init__()
1488
def _create_control_files(self, a_bzrdir):
1489
"""Create the required files and the initial control_files object."""
1490
# FIXME: RBC 20060125 don't peek under the covers
1491
# NB: no need to escape relative paths that are url safe.
1492
repository_transport = a_bzrdir.get_repository_transport(self)
1493
control_files = lockable_files.LockableFiles(repository_transport,
1494
'lock', lockdir.LockDir)
1495
control_files.create_lock()
1496
return control_files
1498
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1499
"""Upload the initial blank content."""
1500
control_files = self._create_control_files(a_bzrdir)
1501
control_files.lock_write()
1503
control_files._transport.mkdir_multi(dirs,
1504
mode=control_files._dir_mode)
1505
for file, content in files:
1506
control_files.put(file, content)
1507
for file, content in utf8_files:
1508
control_files.put_utf8(file, content)
1510
control_files.put_utf8('shared-storage', '')
1512
control_files.unlock()
1515
# formats which have no format string are not discoverable
1516
# and not independently creatable, so are not registered. They're
1517
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1518
# needed, it's constructed directly by the BzrDir. Non-native formats where
1519
# the repository is not separately opened are similar.
1521
format_registry.register_lazy(
1522
'Bazaar-NG Repository format 7',
1523
'bzrlib.repofmt.weaverepo',
1526
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1527
# default control directory format
1529
format_registry.register_lazy(
1530
'Bazaar-NG Knit Repository Format 1',
1531
'bzrlib.repofmt.knitrepo',
1532
'RepositoryFormatKnit1',
1534
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1536
format_registry.register_lazy(
1537
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1538
'bzrlib.repofmt.knitrepo',
1539
'RepositoryFormatKnit3',
1543
class InterRepository(InterObject):
1544
"""This class represents operations taking place between two repositories.
1546
Its instances have methods like copy_content and fetch, and contain
1547
references to the source and target repositories these operations can be
1550
Often we will provide convenience methods on 'repository' which carry out
1551
operations with another repository - they will always forward to
1552
InterRepository.get(other).method_name(parameters).
1556
"""The available optimised InterRepository types."""
1558
def copy_content(self, revision_id=None):
1559
raise NotImplementedError(self.copy_content)
1561
def fetch(self, revision_id=None, pb=None):
1562
"""Fetch the content required to construct revision_id.
1564
The content is copied from self.source to self.target.
1566
:param revision_id: if None all content is copied, if NULL_REVISION no
1568
:param pb: optional progress bar to use for progress reports. If not
1569
provided a default one will be created.
1571
Returns the copied revision count and the failed revisions in a tuple:
1574
raise NotImplementedError(self.fetch)
1577
def missing_revision_ids(self, revision_id=None):
1578
"""Return the revision ids that source has that target does not.
1580
These are returned in topological order.
1582
:param revision_id: only return revision ids included by this
1585
# generic, possibly worst case, slow code path.
1586
target_ids = set(self.target.all_revision_ids())
1587
if revision_id is not None:
1588
# TODO: jam 20070210 InterRepository is internal enough that it
1589
# should assume revision_ids are already utf-8
1590
revision_id = osutils.safe_revision_id(revision_id)
1591
source_ids = self.source.get_ancestry(revision_id)
1592
assert source_ids[0] is None
1595
source_ids = self.source.all_revision_ids()
1596
result_set = set(source_ids).difference(target_ids)
1597
# this may look like a no-op: its not. It preserves the ordering
1598
# other_ids had while only returning the members from other_ids
1599
# that we've decided we need.
1600
return [rev_id for rev_id in source_ids if rev_id in result_set]
1603
class InterSameDataRepository(InterRepository):
1604
"""Code for converting between repositories that represent the same data.
1606
Data format and model must match for this to work.
1610
def _get_repo_format_to_test(self):
1611
"""Repository format for testing with."""
1612
return RepositoryFormat.get_default_format()
1615
def is_compatible(source, target):
1616
if source.supports_rich_root() != target.supports_rich_root():
1618
if source._serializer != target._serializer:
1623
def copy_content(self, revision_id=None):
1624
"""Make a complete copy of the content in self into destination.
1626
This copies both the repository's revision data, and configuration information
1627
such as the make_working_trees setting.
1629
This is a destructive operation! Do not use it on existing
1632
:param revision_id: Only copy the content needed to construct
1633
revision_id and its parents.
1636
self.target.set_make_working_trees(self.source.make_working_trees())
1637
except NotImplementedError:
1639
# TODO: jam 20070210 This is fairly internal, so we should probably
1640
# just assert that revision_id is not unicode.
1641
revision_id = osutils.safe_revision_id(revision_id)
1642
# but don't bother fetching if we have the needed data now.
1643
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1644
self.target.has_revision(revision_id)):
1646
self.target.fetch(self.source, revision_id=revision_id)
1649
def fetch(self, revision_id=None, pb=None):
1650
"""See InterRepository.fetch()."""
1651
from bzrlib.fetch import GenericRepoFetcher
1652
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1653
self.source, self.source._format, self.target,
1654
self.target._format)
1655
# TODO: jam 20070210 This should be an assert, not a translate
1656
revision_id = osutils.safe_revision_id(revision_id)
1657
f = GenericRepoFetcher(to_repository=self.target,
1658
from_repository=self.source,
1659
last_revision=revision_id,
1661
return f.count_copied, f.failed_revisions
1664
class InterWeaveRepo(InterSameDataRepository):
1665
"""Optimised code paths between Weave based repositories."""
1668
def _get_repo_format_to_test(self):
1669
from bzrlib.repofmt import weaverepo
1670
return weaverepo.RepositoryFormat7()
1673
def is_compatible(source, target):
1674
"""Be compatible with known Weave formats.
1676
We don't test for the stores being of specific types because that
1677
could lead to confusing results, and there is no need to be
1680
from bzrlib.repofmt.weaverepo import (
1686
return (isinstance(source._format, (RepositoryFormat5,
1688
RepositoryFormat7)) and
1689
isinstance(target._format, (RepositoryFormat5,
1691
RepositoryFormat7)))
1692
except AttributeError:
1696
def copy_content(self, revision_id=None):
1697
"""See InterRepository.copy_content()."""
1698
# weave specific optimised path:
1699
# TODO: jam 20070210 Internal, should be an assert, not translate
1700
revision_id = osutils.safe_revision_id(revision_id)
1702
self.target.set_make_working_trees(self.source.make_working_trees())
1703
except NotImplementedError:
1705
# FIXME do not peek!
1706
if self.source.control_files._transport.listable():
1707
pb = ui.ui_factory.nested_progress_bar()
1709
self.target.weave_store.copy_all_ids(
1710
self.source.weave_store,
1712
from_transaction=self.source.get_transaction(),
1713
to_transaction=self.target.get_transaction())
1714
pb.update('copying inventory', 0, 1)
1715
self.target.control_weaves.copy_multi(
1716
self.source.control_weaves, ['inventory'],
1717
from_transaction=self.source.get_transaction(),
1718
to_transaction=self.target.get_transaction())
1719
self.target._revision_store.text_store.copy_all_ids(
1720
self.source._revision_store.text_store,
1725
self.target.fetch(self.source, revision_id=revision_id)
1728
def fetch(self, revision_id=None, pb=None):
1729
"""See InterRepository.fetch()."""
1730
from bzrlib.fetch import GenericRepoFetcher
1731
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1732
self.source, self.source._format, self.target, self.target._format)
1733
# TODO: jam 20070210 This should be an assert, not a translate
1734
revision_id = osutils.safe_revision_id(revision_id)
1735
f = GenericRepoFetcher(to_repository=self.target,
1736
from_repository=self.source,
1737
last_revision=revision_id,
1739
return f.count_copied, f.failed_revisions
1742
def missing_revision_ids(self, revision_id=None):
1743
"""See InterRepository.missing_revision_ids()."""
1744
# we want all revisions to satisfy revision_id in source.
1745
# but we don't want to stat every file here and there.
1746
# we want then, all revisions other needs to satisfy revision_id
1747
# checked, but not those that we have locally.
1748
# so the first thing is to get a subset of the revisions to
1749
# satisfy revision_id in source, and then eliminate those that
1750
# we do already have.
1751
# this is slow on high latency connection to self, but as as this
1752
# disk format scales terribly for push anyway due to rewriting
1753
# inventory.weave, this is considered acceptable.
1755
if revision_id is not None:
1756
source_ids = self.source.get_ancestry(revision_id)
1757
assert source_ids[0] is None
1760
source_ids = self.source._all_possible_ids()
1761
source_ids_set = set(source_ids)
1762
# source_ids is the worst possible case we may need to pull.
1763
# now we want to filter source_ids against what we actually
1764
# have in target, but don't try to check for existence where we know
1765
# we do not have a revision as that would be pointless.
1766
target_ids = set(self.target._all_possible_ids())
1767
possibly_present_revisions = target_ids.intersection(source_ids_set)
1768
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1769
required_revisions = source_ids_set.difference(actually_present_revisions)
1770
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1771
if revision_id is not None:
1772
# we used get_ancestry to determine source_ids then we are assured all
1773
# revisions referenced are present as they are installed in topological order.
1774
# and the tip revision was validated by get_ancestry.
1775
return required_topo_revisions
1777
# if we just grabbed the possibly available ids, then
1778
# we only have an estimate of whats available and need to validate
1779
# that against the revision records.
1780
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1783
class InterKnitRepo(InterSameDataRepository):
1784
"""Optimised code paths between Knit based repositories."""
1787
def _get_repo_format_to_test(self):
1788
from bzrlib.repofmt import knitrepo
1789
return knitrepo.RepositoryFormatKnit1()
1792
def is_compatible(source, target):
1793
"""Be compatible with known Knit formats.
1795
We don't test for the stores being of specific types because that
1796
could lead to confusing results, and there is no need to be
1799
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1801
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1802
isinstance(target._format, (RepositoryFormatKnit1)))
1803
except AttributeError:
1807
def fetch(self, revision_id=None, pb=None):
1808
"""See InterRepository.fetch()."""
1809
from bzrlib.fetch import KnitRepoFetcher
1810
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1811
self.source, self.source._format, self.target, self.target._format)
1812
# TODO: jam 20070210 This should be an assert, not a translate
1813
revision_id = osutils.safe_revision_id(revision_id)
1814
f = KnitRepoFetcher(to_repository=self.target,
1815
from_repository=self.source,
1816
last_revision=revision_id,
1818
return f.count_copied, f.failed_revisions
1821
def missing_revision_ids(self, revision_id=None):
1822
"""See InterRepository.missing_revision_ids()."""
1823
if revision_id is not None:
1824
source_ids = self.source.get_ancestry(revision_id)
1825
assert source_ids[0] is None
1828
source_ids = self.source._all_possible_ids()
1829
source_ids_set = set(source_ids)
1830
# source_ids is the worst possible case we may need to pull.
1831
# now we want to filter source_ids against what we actually
1832
# have in target, but don't try to check for existence where we know
1833
# we do not have a revision as that would be pointless.
1834
target_ids = set(self.target._all_possible_ids())
1835
possibly_present_revisions = target_ids.intersection(source_ids_set)
1836
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1837
required_revisions = source_ids_set.difference(actually_present_revisions)
1838
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1839
if revision_id is not None:
1840
# we used get_ancestry to determine source_ids then we are assured all
1841
# revisions referenced are present as they are installed in topological order.
1842
# and the tip revision was validated by get_ancestry.
1843
return required_topo_revisions
1845
# if we just grabbed the possibly available ids, then
1846
# we only have an estimate of whats available and need to validate
1847
# that against the revision records.
1848
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1851
class InterModel1and2(InterRepository):
1854
def _get_repo_format_to_test(self):
1858
def is_compatible(source, target):
1859
if not source.supports_rich_root() and target.supports_rich_root():
1865
def fetch(self, revision_id=None, pb=None):
1866
"""See InterRepository.fetch()."""
1867
from bzrlib.fetch import Model1toKnit2Fetcher
1868
# TODO: jam 20070210 This should be an assert, not a translate
1869
revision_id = osutils.safe_revision_id(revision_id)
1870
f = Model1toKnit2Fetcher(to_repository=self.target,
1871
from_repository=self.source,
1872
last_revision=revision_id,
1874
return f.count_copied, f.failed_revisions
1877
def copy_content(self, revision_id=None):
1878
"""Make a complete copy of the content in self into destination.
1880
This is a destructive operation! Do not use it on existing
1883
:param revision_id: Only copy the content needed to construct
1884
revision_id and its parents.
1887
self.target.set_make_working_trees(self.source.make_working_trees())
1888
except NotImplementedError:
1890
# TODO: jam 20070210 Internal, assert, don't translate
1891
revision_id = osutils.safe_revision_id(revision_id)
1892
# but don't bother fetching if we have the needed data now.
1893
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1894
self.target.has_revision(revision_id)):
1896
self.target.fetch(self.source, revision_id=revision_id)
1899
class InterKnit1and2(InterKnitRepo):
1902
def _get_repo_format_to_test(self):
1906
def is_compatible(source, target):
1907
"""Be compatible with Knit1 source and Knit3 target"""
1908
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1910
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1911
RepositoryFormatKnit3
1912
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1913
isinstance(target._format, (RepositoryFormatKnit3)))
1914
except AttributeError:
1918
def fetch(self, revision_id=None, pb=None):
1919
"""See InterRepository.fetch()."""
1920
from bzrlib.fetch import Knit1to2Fetcher
1921
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1922
self.source, self.source._format, self.target,
1923
self.target._format)
1924
# TODO: jam 20070210 This should be an assert, not a translate
1925
revision_id = osutils.safe_revision_id(revision_id)
1926
f = Knit1to2Fetcher(to_repository=self.target,
1927
from_repository=self.source,
1928
last_revision=revision_id,
1930
return f.count_copied, f.failed_revisions
1933
class InterRemoteRepository(InterRepository):
1934
"""Code for converting between RemoteRepository objects.
1936
This just gets an non-remote repository from the RemoteRepository, and calls
1937
InterRepository.get again.
1940
def __init__(self, source, target):
1941
if isinstance(source, remote.RemoteRepository):
1942
source._ensure_real()
1943
real_source = source._real_repository
1945
real_source = source
1946
if isinstance(target, remote.RemoteRepository):
1947
target._ensure_real()
1948
real_target = target._real_repository
1950
real_target = target
1951
self.real_inter = InterRepository.get(real_source, real_target)
1954
def is_compatible(source, target):
1955
if isinstance(source, remote.RemoteRepository):
1957
if isinstance(target, remote.RemoteRepository):
1961
def copy_content(self, revision_id=None):
1962
self.real_inter.copy_content(revision_id=revision_id)
1964
def fetch(self, revision_id=None, pb=None):
1965
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1968
def _get_repo_format_to_test(self):
1972
InterRepository.register_optimiser(InterSameDataRepository)
1973
InterRepository.register_optimiser(InterWeaveRepo)
1974
InterRepository.register_optimiser(InterKnitRepo)
1975
InterRepository.register_optimiser(InterModel1and2)
1976
InterRepository.register_optimiser(InterKnit1and2)
1977
InterRepository.register_optimiser(InterRemoteRepository)
1980
class CopyConverter(object):
1981
"""A repository conversion tool which just performs a copy of the content.
1983
This is slow but quite reliable.
1986
def __init__(self, target_format):
1987
"""Create a CopyConverter.
1989
:param target_format: The format the resulting repository should be.
1991
self.target_format = target_format
1993
def convert(self, repo, pb):
1994
"""Perform the conversion of to_convert, giving feedback via pb.
1996
:param to_convert: The disk object to convert.
1997
:param pb: a progress bar to use for progress information.
2002
# this is only useful with metadir layouts - separated repo content.
2003
# trigger an assertion if not such
2004
repo._format.get_format_string()
2005
self.repo_dir = repo.bzrdir
2006
self.step('Moving repository to repository.backup')
2007
self.repo_dir.transport.move('repository', 'repository.backup')
2008
backup_transport = self.repo_dir.transport.clone('repository.backup')
2009
repo._format.check_conversion_target(self.target_format)
2010
self.source_repo = repo._format.open(self.repo_dir,
2012
_override_transport=backup_transport)
2013
self.step('Creating new repository')
2014
converted = self.target_format.initialize(self.repo_dir,
2015
self.source_repo.is_shared())
2016
converted.lock_write()
2018
self.step('Copying content into repository.')
2019
self.source_repo.copy_content_into(converted)
2022
self.step('Deleting old repository content.')
2023
self.repo_dir.transport.delete_tree('repository.backup')
2024
self.pb.note('repository converted')
2026
def step(self, message):
2027
"""Update the pb by a step."""
2029
self.pb.update(message, self.count, self.total)
2032
class CommitBuilder(object):
2033
"""Provides an interface to build up a commit.
2035
This allows describing a tree to be committed without needing to
2036
know the internals of the format of the repository.
2039
record_root_entry = False
2040
def __init__(self, repository, parents, config, timestamp=None,
2041
timezone=None, committer=None, revprops=None,
2043
"""Initiate a CommitBuilder.
2045
:param repository: Repository to commit to.
2046
:param parents: Revision ids of the parents of the new revision.
2047
:param config: Configuration to use.
2048
:param timestamp: Optional timestamp recorded for commit.
2049
:param timezone: Optional timezone for timestamp.
2050
:param committer: Optional committer to set for commit.
2051
:param revprops: Optional dictionary of revision properties.
2052
:param revision_id: Optional revision id.
2054
self._config = config
2056
if committer is None:
2057
self._committer = self._config.username()
2059
assert isinstance(committer, basestring), type(committer)
2060
self._committer = committer
2062
self.new_inventory = Inventory(None)
2063
self._new_revision_id = osutils.safe_revision_id(revision_id)
2064
self.parents = parents
2065
self.repository = repository
2068
if revprops is not None:
2069
self._revprops.update(revprops)
2071
if timestamp is None:
2072
timestamp = time.time()
2073
# Restrict resolution to 1ms
2074
self._timestamp = round(timestamp, 3)
2076
if timezone is None:
2077
self._timezone = osutils.local_time_offset()
2079
self._timezone = int(timezone)
2081
self._generate_revision_if_needed()
2083
def commit(self, message):
2084
"""Make the actual commit.
2086
:return: The revision id of the recorded revision.
2088
rev = _mod_revision.Revision(
2089
timestamp=self._timestamp,
2090
timezone=self._timezone,
2091
committer=self._committer,
2093
inventory_sha1=self.inv_sha1,
2094
revision_id=self._new_revision_id,
2095
properties=self._revprops)
2096
rev.parent_ids = self.parents
2097
self.repository.add_revision(self._new_revision_id, rev,
2098
self.new_inventory, self._config)
2099
self.repository.commit_write_group()
2100
return self._new_revision_id
2102
def revision_tree(self):
2103
"""Return the tree that was just committed.
2105
After calling commit() this can be called to get a RevisionTree
2106
representing the newly committed tree. This is preferred to
2107
calling Repository.revision_tree() because that may require
2108
deserializing the inventory, while we already have a copy in
2111
return RevisionTree(self.repository, self.new_inventory,
2112
self._new_revision_id)
2114
def finish_inventory(self):
2115
"""Tell the builder that the inventory is finished."""
2116
if self.new_inventory.root is None:
2117
symbol_versioning.warn('Root entry should be supplied to'
2118
' record_entry_contents, as of bzr 0.10.',
2119
DeprecationWarning, stacklevel=2)
2120
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2121
self.new_inventory.revision_id = self._new_revision_id
2122
self.inv_sha1 = self.repository.add_inventory(
2123
self._new_revision_id,
2128
def _gen_revision_id(self):
2129
"""Return new revision-id."""
2130
return generate_ids.gen_revision_id(self._config.username(),
2133
def _generate_revision_if_needed(self):
2134
"""Create a revision id if None was supplied.
2136
If the repository can not support user-specified revision ids
2137
they should override this function and raise CannotSetRevisionId
2138
if _new_revision_id is not None.
2140
:raises: CannotSetRevisionId
2142
if self._new_revision_id is None:
2143
self._new_revision_id = self._gen_revision_id()
2145
def record_entry_contents(self, ie, parent_invs, path, tree):
2146
"""Record the content of ie from tree into the commit if needed.
2148
Side effect: sets ie.revision when unchanged
2150
:param ie: An inventory entry present in the commit.
2151
:param parent_invs: The inventories of the parent revisions of the
2153
:param path: The path the entry is at in the tree.
2154
:param tree: The tree which contains this entry and should be used to
2157
if self.new_inventory.root is None and ie.parent_id is not None:
2158
symbol_versioning.warn('Root entry should be supplied to'
2159
' record_entry_contents, as of bzr 0.10.',
2160
DeprecationWarning, stacklevel=2)
2161
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2163
self.new_inventory.add(ie)
2165
# ie.revision is always None if the InventoryEntry is considered
2166
# for committing. ie.snapshot will record the correct revision
2167
# which may be the sole parent if it is untouched.
2168
if ie.revision is not None:
2171
# In this revision format, root entries have no knit or weave
2172
if ie is self.new_inventory.root:
2173
# When serializing out to disk and back in
2174
# root.revision is always _new_revision_id
2175
ie.revision = self._new_revision_id
2177
previous_entries = ie.find_previous_heads(
2179
self.repository.weave_store,
2180
self.repository.get_transaction())
2181
# we are creating a new revision for ie in the history store
2183
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2185
def modified_directory(self, file_id, file_parents):
2186
"""Record the presence of a symbolic link.
2188
:param file_id: The file_id of the link to record.
2189
:param file_parents: The per-file parent revision ids.
2191
self._add_text_to_weave(file_id, [], file_parents.keys())
2193
def modified_reference(self, file_id, file_parents):
2194
"""Record the modification of a reference.
2196
:param file_id: The file_id of the link to record.
2197
:param file_parents: The per-file parent revision ids.
2199
self._add_text_to_weave(file_id, [], file_parents.keys())
2201
def modified_file_text(self, file_id, file_parents,
2202
get_content_byte_lines, text_sha1=None,
2204
"""Record the text of file file_id
2206
:param file_id: The file_id of the file to record the text of.
2207
:param file_parents: The per-file parent revision ids.
2208
:param get_content_byte_lines: A callable which will return the byte
2210
:param text_sha1: Optional SHA1 of the file contents.
2211
:param text_size: Optional size of the file contents.
2213
# mutter('storing text of file {%s} in revision {%s} into %r',
2214
# file_id, self._new_revision_id, self.repository.weave_store)
2215
# special case to avoid diffing on renames or
2217
if (len(file_parents) == 1
2218
and text_sha1 == file_parents.values()[0].text_sha1
2219
and text_size == file_parents.values()[0].text_size):
2220
previous_ie = file_parents.values()[0]
2221
versionedfile = self.repository.weave_store.get_weave(file_id,
2222
self.repository.get_transaction())
2223
versionedfile.clone_text(self._new_revision_id,
2224
previous_ie.revision, file_parents.keys())
2225
return text_sha1, text_size
2227
new_lines = get_content_byte_lines()
2228
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2229
# should return the SHA1 and size
2230
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2231
return osutils.sha_strings(new_lines), \
2232
sum(map(len, new_lines))
2234
def modified_link(self, file_id, file_parents, link_target):
2235
"""Record the presence of a symbolic link.
2237
:param file_id: The file_id of the link to record.
2238
:param file_parents: The per-file parent revision ids.
2239
:param link_target: Target location of this link.
2241
self._add_text_to_weave(file_id, [], file_parents.keys())
2243
def _add_text_to_weave(self, file_id, new_lines, parents):
2244
versionedfile = self.repository.weave_store.get_weave_or_empty(
2245
file_id, self.repository.get_transaction())
2246
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2247
versionedfile.clear_cache()
2250
class _CommitBuilder(CommitBuilder):
2251
"""Temporary class so old CommitBuilders are detected properly
2253
Note: CommitBuilder works whether or not root entry is recorded.
2256
record_root_entry = True
2259
class RootCommitBuilder(CommitBuilder):
2260
"""This commitbuilder actually records the root id"""
2262
record_root_entry = True
2264
def record_entry_contents(self, ie, parent_invs, path, tree):
2265
"""Record the content of ie from tree into the commit if needed.
2267
Side effect: sets ie.revision when unchanged
2269
:param ie: An inventory entry present in the commit.
2270
:param parent_invs: The inventories of the parent revisions of the
2272
:param path: The path the entry is at in the tree.
2273
:param tree: The tree which contains this entry and should be used to
2276
assert self.new_inventory.root is not None or ie.parent_id is None
2277
self.new_inventory.add(ie)
2279
# ie.revision is always None if the InventoryEntry is considered
2280
# for committing. ie.snapshot will record the correct revision
2281
# which may be the sole parent if it is untouched.
2282
if ie.revision is not None:
2285
previous_entries = ie.find_previous_heads(
2287
self.repository.weave_store,
2288
self.repository.get_transaction())
2289
# we are creating a new revision for ie in the history store
2291
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2303
def _unescaper(match, _map=_unescape_map):
2304
code = match.group(1)
2308
if not code.startswith('#'):
2310
return unichr(int(code[1:])).encode('utf8')
2316
def _unescape_xml(data):
2317
"""Unescape predefined XML entities in a string of data."""
2319
if _unescape_re is None:
2320
_unescape_re = re.compile('\&([^;]*);')
2321
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
