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 iter_files_bytes(self, desired_files):
735
"""Iterate through file versions.
737
Files will not necessarily be returned in the order they occur in
738
desired_files. No specific order is guaranteed.
740
Yields pairs of identifier, bytes_iterator. identifier is an opaque
741
value supplied by the caller as part of desired_files. It should
742
uniquely identify the file version in the caller's context. (Examples:
743
an index number or a TreeTransform trans_id.)
745
bytes_iterator is an iterable of bytestrings for the file. The
746
kind of iterable and length of the bytestrings are unspecified, but for
747
this implementation, it is a list of lines produced by
748
VersionedFile.get_lines().
750
:param desired_files: a list of (file_id, revision_id, identifier)
753
transaction = self.get_transaction()
754
for file_id, revision_id, callable_data in desired_files:
756
weave = self.weave_store.get_weave(file_id, transaction)
757
except errors.NoSuchFile:
758
raise errors.NoSuchIdInRepository(self, file_id)
759
yield callable_data, weave.get_lines(revision_id)
761
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
762
"""Get an iterable listing the keys of all the data introduced by a set
765
The keys will be ordered so that the corresponding items can be safely
766
fetched and inserted in that order.
768
:returns: An iterable producing tuples of (knit-kind, file-id,
769
versions). knit-kind is one of 'file', 'inventory', 'signatures',
770
'revisions'. file-id is None unless knit-kind is 'file'.
772
# XXX: it's a bit weird to control the inventory weave caching in this
773
# generator. Ideally the caching would be done in fetch.py I think. Or
774
# maybe this generator should explicitly have the contract that it
775
# should not be iterated until the previously yielded item has been
777
inv_w = self.get_inventory_weave()
780
# file ids that changed
781
file_ids = self.fileids_altered_by_revision_ids(revision_ids)
783
num_file_ids = len(file_ids)
784
for file_id, altered_versions in file_ids.iteritems():
785
if _files_pb is not None:
786
_files_pb.update("fetch texts", count, num_file_ids)
788
yield ("file", file_id, altered_versions)
789
# We're done with the files_pb. Note that it finished by the caller,
790
# just as it was created by the caller.
794
yield ("inventory", None, revision_ids)
798
revisions_with_signatures = set()
799
for rev_id in revision_ids:
801
self.get_signature_text(rev_id)
802
except errors.NoSuchRevision:
806
revisions_with_signatures.add(rev_id)
807
yield ("signatures", None, revisions_with_signatures)
810
yield ("revisions", None, revision_ids)
813
def get_inventory_weave(self):
814
return self.control_weaves.get_weave('inventory',
815
self.get_transaction())
818
def get_inventory(self, revision_id):
819
"""Get Inventory object by hash."""
820
# TODO: jam 20070210 Technically we don't need to sanitize, since all
821
# called functions must sanitize.
822
revision_id = osutils.safe_revision_id(revision_id)
823
return self.deserialise_inventory(
824
revision_id, self.get_inventory_xml(revision_id))
826
def deserialise_inventory(self, revision_id, xml):
827
"""Transform the xml into an inventory object.
829
:param revision_id: The expected revision id of the inventory.
830
:param xml: A serialised inventory.
832
revision_id = osutils.safe_revision_id(revision_id)
833
result = self._serializer.read_inventory_from_string(xml)
834
result.root.revision = revision_id
837
def serialise_inventory(self, inv):
838
return self._serializer.write_inventory_to_string(inv)
840
def get_serializer_format(self):
841
return self._serializer.format_num
844
def get_inventory_xml(self, revision_id):
845
"""Get inventory XML as a file object."""
846
revision_id = osutils.safe_revision_id(revision_id)
848
assert isinstance(revision_id, str), type(revision_id)
849
iw = self.get_inventory_weave()
850
return iw.get_text(revision_id)
852
raise errors.HistoryMissing(self, 'inventory', revision_id)
855
def get_inventory_sha1(self, revision_id):
856
"""Return the sha1 hash of the inventory entry
858
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
859
revision_id = osutils.safe_revision_id(revision_id)
860
return self.get_revision(revision_id).inventory_sha1
863
def get_revision_graph(self, revision_id=None):
864
"""Return a dictionary containing the revision graph.
866
:param revision_id: The revision_id to get a graph from. If None, then
867
the entire revision graph is returned. This is a deprecated mode of
868
operation and will be removed in the future.
869
:return: a dictionary of revision_id->revision_parents_list.
871
# special case NULL_REVISION
872
if revision_id == _mod_revision.NULL_REVISION:
874
revision_id = osutils.safe_revision_id(revision_id)
875
a_weave = self.get_inventory_weave()
876
all_revisions = self._eliminate_revisions_not_present(
878
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
879
node in all_revisions])
880
if revision_id is None:
882
elif revision_id not in entire_graph:
883
raise errors.NoSuchRevision(self, revision_id)
885
# add what can be reached from revision_id
887
pending = set([revision_id])
888
while len(pending) > 0:
890
result[node] = entire_graph[node]
891
for revision_id in result[node]:
892
if revision_id not in result:
893
pending.add(revision_id)
897
def get_revision_graph_with_ghosts(self, revision_ids=None):
898
"""Return a graph of the revisions with ghosts marked as applicable.
900
:param revision_ids: an iterable of revisions to graph or None for all.
901
:return: a Graph object with the graph reachable from revision_ids.
903
result = deprecated_graph.Graph()
905
pending = set(self.all_revision_ids())
908
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
909
# special case NULL_REVISION
910
if _mod_revision.NULL_REVISION in pending:
911
pending.remove(_mod_revision.NULL_REVISION)
912
required = set(pending)
915
revision_id = pending.pop()
917
rev = self.get_revision(revision_id)
918
except errors.NoSuchRevision:
919
if revision_id in required:
922
result.add_ghost(revision_id)
924
for parent_id in rev.parent_ids:
925
# is this queued or done ?
926
if (parent_id not in pending and
927
parent_id not in done):
929
pending.add(parent_id)
930
result.add_node(revision_id, rev.parent_ids)
931
done.add(revision_id)
934
def _get_history_vf(self):
935
"""Get a versionedfile whose history graph reflects all revisions.
937
For weave repositories, this is the inventory weave.
939
return self.get_inventory_weave()
941
def iter_reverse_revision_history(self, revision_id):
942
"""Iterate backwards through revision ids in the lefthand history
944
:param revision_id: The revision id to start with. All its lefthand
945
ancestors will be traversed.
947
revision_id = osutils.safe_revision_id(revision_id)
948
if revision_id in (None, _mod_revision.NULL_REVISION):
950
next_id = revision_id
951
versionedfile = self._get_history_vf()
954
parents = versionedfile.get_parents(next_id)
955
if len(parents) == 0:
961
def get_revision_inventory(self, revision_id):
962
"""Return inventory of a past revision."""
963
# TODO: Unify this with get_inventory()
964
# bzr 0.0.6 and later imposes the constraint that the inventory_id
965
# must be the same as its revision, so this is trivial.
966
if revision_id is None:
967
# This does not make sense: if there is no revision,
968
# then it is the current tree inventory surely ?!
969
# and thus get_root_id() is something that looks at the last
970
# commit on the branch, and the get_root_id is an inventory check.
971
raise NotImplementedError
972
# return Inventory(self.get_root_id())
974
return self.get_inventory(revision_id)
978
"""Return True if this repository is flagged as a shared repository."""
979
raise NotImplementedError(self.is_shared)
982
def reconcile(self, other=None, thorough=False):
983
"""Reconcile this repository."""
984
from bzrlib.reconcile import RepoReconciler
985
reconciler = RepoReconciler(self, thorough=thorough)
986
reconciler.reconcile()
989
def _refresh_data(self):
990
"""Helper called from lock_* to ensure coherency with disk.
992
The default implementation does nothing; it is however possible
993
for repositories to maintain loaded indices across multiple locks
994
by checking inside their implementation of this method to see
995
whether their indices are still valid. This depends of course on
996
the disk format being validatable in this manner.
1000
def revision_tree(self, revision_id):
1001
"""Return Tree for a revision on this branch.
1003
`revision_id` may be None for the empty tree revision.
1005
# TODO: refactor this to use an existing revision object
1006
# so we don't need to read it in twice.
1007
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1008
return RevisionTree(self, Inventory(root_id=None),
1009
_mod_revision.NULL_REVISION)
1011
revision_id = osutils.safe_revision_id(revision_id)
1012
inv = self.get_revision_inventory(revision_id)
1013
return RevisionTree(self, inv, revision_id)
1016
def revision_trees(self, revision_ids):
1017
"""Return Tree for a revision on this branch.
1019
`revision_id` may not be None or 'null:'"""
1020
assert None not in revision_ids
1021
assert _mod_revision.NULL_REVISION not in revision_ids
1022
texts = self.get_inventory_weave().get_texts(revision_ids)
1023
for text, revision_id in zip(texts, revision_ids):
1024
inv = self.deserialise_inventory(revision_id, text)
1025
yield RevisionTree(self, inv, revision_id)
1028
def get_ancestry(self, revision_id, topo_sorted=True):
1029
"""Return a list of revision-ids integrated by a revision.
1031
The first element of the list is always None, indicating the origin
1032
revision. This might change when we have history horizons, or
1033
perhaps we should have a new API.
1035
This is topologically sorted.
1037
if _mod_revision.is_null(revision_id):
1039
revision_id = osutils.safe_revision_id(revision_id)
1040
if not self.has_revision(revision_id):
1041
raise errors.NoSuchRevision(self, revision_id)
1042
w = self.get_inventory_weave()
1043
candidates = w.get_ancestry(revision_id, topo_sorted)
1044
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1047
"""Compress the data within the repository.
1049
This operation only makes sense for some repository types. For other
1050
types it should be a no-op that just returns.
1052
This stub method does not require a lock, but subclasses should use
1053
@needs_write_lock as this is a long running call its reasonable to
1054
implicitly lock for the user.
1058
def print_file(self, file, revision_id):
1059
"""Print `file` to stdout.
1061
FIXME RBC 20060125 as John Meinel points out this is a bad api
1062
- it writes to stdout, it assumes that that is valid etc. Fix
1063
by creating a new more flexible convenience function.
1065
revision_id = osutils.safe_revision_id(revision_id)
1066
tree = self.revision_tree(revision_id)
1067
# use inventory as it was in that revision
1068
file_id = tree.inventory.path2id(file)
1070
# TODO: jam 20060427 Write a test for this code path
1071
# it had a bug in it, and was raising the wrong
1073
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1074
tree.print_file(file_id)
1076
def get_transaction(self):
1077
return self.control_files.get_transaction()
1079
def revision_parents(self, revision_id):
1080
revision_id = osutils.safe_revision_id(revision_id)
1081
return self.get_inventory_weave().parent_names(revision_id)
1083
def get_parents(self, revision_ids):
1084
"""See StackedParentsProvider.get_parents"""
1086
for revision_id in revision_ids:
1087
if revision_id == _mod_revision.NULL_REVISION:
1091
parents = self.get_revision(revision_id).parent_ids
1092
except errors.NoSuchRevision:
1095
if len(parents) == 0:
1096
parents = [_mod_revision.NULL_REVISION]
1097
parents_list.append(parents)
1100
def _make_parents_provider(self):
1103
def get_graph(self, other_repository=None):
1104
"""Return the graph walker for this repository format"""
1105
parents_provider = self._make_parents_provider()
1106
if (other_repository is not None and
1107
other_repository.bzrdir.transport.base !=
1108
self.bzrdir.transport.base):
1109
parents_provider = graph._StackedParentsProvider(
1110
[parents_provider, other_repository._make_parents_provider()])
1111
return graph.Graph(parents_provider)
1114
def set_make_working_trees(self, new_value):
1115
"""Set the policy flag for making working trees when creating branches.
1117
This only applies to branches that use this repository.
1119
The default is 'True'.
1120
:param new_value: True to restore the default, False to disable making
1123
raise NotImplementedError(self.set_make_working_trees)
1125
def make_working_trees(self):
1126
"""Returns the policy for making working trees on new branches."""
1127
raise NotImplementedError(self.make_working_trees)
1130
def sign_revision(self, revision_id, gpg_strategy):
1131
revision_id = osutils.safe_revision_id(revision_id)
1132
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1133
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1136
def has_signature_for_revision_id(self, revision_id):
1137
"""Query for a revision signature for revision_id in the repository."""
1138
revision_id = osutils.safe_revision_id(revision_id)
1139
return self._revision_store.has_signature(revision_id,
1140
self.get_transaction())
1143
def get_signature_text(self, revision_id):
1144
"""Return the text for a signature."""
1145
revision_id = osutils.safe_revision_id(revision_id)
1146
return self._revision_store.get_signature_text(revision_id,
1147
self.get_transaction())
1150
def check(self, revision_ids):
1151
"""Check consistency of all history of given revision_ids.
1153
Different repository implementations should override _check().
1155
:param revision_ids: A non-empty list of revision_ids whose ancestry
1156
will be checked. Typically the last revision_id of a branch.
1158
if not revision_ids:
1159
raise ValueError("revision_ids must be non-empty in %s.check"
1161
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1162
return self._check(revision_ids)
1164
def _check(self, revision_ids):
1165
result = check.Check(self)
1169
def _warn_if_deprecated(self):
1170
global _deprecation_warning_done
1171
if _deprecation_warning_done:
1173
_deprecation_warning_done = True
1174
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1175
% (self._format, self.bzrdir.transport.base))
1177
def supports_rich_root(self):
1178
return self._format.rich_root_data
1180
def _check_ascii_revisionid(self, revision_id, method):
1181
"""Private helper for ascii-only repositories."""
1182
# weave repositories refuse to store revisionids that are non-ascii.
1183
if revision_id is not None:
1184
# weaves require ascii revision ids.
1185
if isinstance(revision_id, unicode):
1187
revision_id.encode('ascii')
1188
except UnicodeEncodeError:
1189
raise errors.NonAsciiRevisionId(method, self)
1192
revision_id.decode('ascii')
1193
except UnicodeDecodeError:
1194
raise errors.NonAsciiRevisionId(method, self)
1198
# remove these delegates a while after bzr 0.15
1199
def __make_delegated(name, from_module):
1200
def _deprecated_repository_forwarder():
1201
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1202
% (name, from_module),
1205
m = __import__(from_module, globals(), locals(), [name])
1207
return getattr(m, name)
1208
except AttributeError:
1209
raise AttributeError('module %s has no name %s'
1211
globals()[name] = _deprecated_repository_forwarder
1214
'AllInOneRepository',
1215
'WeaveMetaDirRepository',
1216
'PreSplitOutRepositoryFormat',
1217
'RepositoryFormat4',
1218
'RepositoryFormat5',
1219
'RepositoryFormat6',
1220
'RepositoryFormat7',
1222
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1226
'RepositoryFormatKnit',
1227
'RepositoryFormatKnit1',
1229
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1232
def install_revision(repository, rev, revision_tree):
1233
"""Install all revision data into a repository."""
1234
present_parents = []
1236
for p_id in rev.parent_ids:
1237
if repository.has_revision(p_id):
1238
present_parents.append(p_id)
1239
parent_trees[p_id] = repository.revision_tree(p_id)
1241
parent_trees[p_id] = repository.revision_tree(None)
1243
inv = revision_tree.inventory
1244
entries = inv.iter_entries()
1245
# backwards compatibility hack: skip the root id.
1246
if not repository.supports_rich_root():
1247
path, root = entries.next()
1248
if root.revision != rev.revision_id:
1249
raise errors.IncompatibleRevision(repr(repository))
1250
# Add the texts that are not already present
1251
for path, ie in entries:
1252
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1253
repository.get_transaction())
1254
if ie.revision not in w:
1256
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1257
# with InventoryEntry.find_previous_heads(). if it is, then there
1258
# is a latent bug here where the parents may have ancestors of each
1260
for revision, tree in parent_trees.iteritems():
1261
if ie.file_id not in tree:
1263
parent_id = tree.inventory[ie.file_id].revision
1264
if parent_id in text_parents:
1266
text_parents.append(parent_id)
1268
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1269
repository.get_transaction())
1270
lines = revision_tree.get_file(ie.file_id).readlines()
1271
vfile.add_lines(rev.revision_id, text_parents, lines)
1273
# install the inventory
1274
repository.add_inventory(rev.revision_id, inv, present_parents)
1275
except errors.RevisionAlreadyPresent:
1277
repository.add_revision(rev.revision_id, rev, inv)
1280
class MetaDirRepository(Repository):
1281
"""Repositories in the new meta-dir layout."""
1283
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1284
super(MetaDirRepository, self).__init__(_format,
1290
dir_mode = self.control_files._dir_mode
1291
file_mode = self.control_files._file_mode
1294
def is_shared(self):
1295
"""Return True if this repository is flagged as a shared repository."""
1296
return self.control_files._transport.has('shared-storage')
1299
def set_make_working_trees(self, new_value):
1300
"""Set the policy flag for making working trees when creating branches.
1302
This only applies to branches that use this repository.
1304
The default is 'True'.
1305
:param new_value: True to restore the default, False to disable making
1310
self.control_files._transport.delete('no-working-trees')
1311
except errors.NoSuchFile:
1314
self.control_files.put_utf8('no-working-trees', '')
1316
def make_working_trees(self):
1317
"""Returns the policy for making working trees on new branches."""
1318
return not self.control_files._transport.has('no-working-trees')
1321
class RepositoryFormatRegistry(registry.Registry):
1322
"""Registry of RepositoryFormats.
1325
def get(self, format_string):
1326
r = registry.Registry.get(self, format_string)
1332
format_registry = RepositoryFormatRegistry()
1333
"""Registry of formats, indexed by their identifying format string.
1335
This can contain either format instances themselves, or classes/factories that
1336
can be called to obtain one.
1340
#####################################################################
1341
# Repository Formats
1343
class RepositoryFormat(object):
1344
"""A repository format.
1346
Formats provide three things:
1347
* An initialization routine to construct repository data on disk.
1348
* a format string which is used when the BzrDir supports versioned
1350
* an open routine which returns a Repository instance.
1352
Formats are placed in an dict by their format string for reference
1353
during opening. These should be subclasses of RepositoryFormat
1356
Once a format is deprecated, just deprecate the initialize and open
1357
methods on the format class. Do not deprecate the object, as the
1358
object will be created every system load.
1360
Common instance attributes:
1361
_matchingbzrdir - the bzrdir format that the repository format was
1362
originally written to work with. This can be used if manually
1363
constructing a bzrdir and repository, or more commonly for test suite
1368
return "<%s>" % self.__class__.__name__
1370
def __eq__(self, other):
1371
# format objects are generally stateless
1372
return isinstance(other, self.__class__)
1374
def __ne__(self, other):
1375
return not self == other
1378
def find_format(klass, a_bzrdir):
1379
"""Return the format for the repository object in a_bzrdir.
1381
This is used by bzr native formats that have a "format" file in
1382
the repository. Other methods may be used by different types of
1386
transport = a_bzrdir.get_repository_transport(None)
1387
format_string = transport.get("format").read()
1388
return format_registry.get(format_string)
1389
except errors.NoSuchFile:
1390
raise errors.NoRepositoryPresent(a_bzrdir)
1392
raise errors.UnknownFormatError(format=format_string)
1395
def register_format(klass, format):
1396
format_registry.register(format.get_format_string(), format)
1399
def unregister_format(klass, format):
1400
format_registry.remove(format.get_format_string())
1403
def get_default_format(klass):
1404
"""Return the current default format."""
1405
from bzrlib import bzrdir
1406
return bzrdir.format_registry.make_bzrdir('default').repository_format
1408
def _get_control_store(self, repo_transport, control_files):
1409
"""Return the control store for this repository."""
1410
raise NotImplementedError(self._get_control_store)
1412
def get_format_string(self):
1413
"""Return the ASCII format string that identifies this format.
1415
Note that in pre format ?? repositories the format string is
1416
not permitted nor written to disk.
1418
raise NotImplementedError(self.get_format_string)
1420
def get_format_description(self):
1421
"""Return the short description for this format."""
1422
raise NotImplementedError(self.get_format_description)
1424
def _get_revision_store(self, repo_transport, control_files):
1425
"""Return the revision store object for this a_bzrdir."""
1426
raise NotImplementedError(self._get_revision_store)
1428
def _get_text_rev_store(self,
1435
"""Common logic for getting a revision store for a repository.
1437
see self._get_revision_store for the subclass-overridable method to
1438
get the store for a repository.
1440
from bzrlib.store.revision.text import TextRevisionStore
1441
dir_mode = control_files._dir_mode
1442
file_mode = control_files._file_mode
1443
text_store = TextStore(transport.clone(name),
1445
compressed=compressed,
1447
file_mode=file_mode)
1448
_revision_store = TextRevisionStore(text_store, serializer)
1449
return _revision_store
1451
# TODO: this shouldn't be in the base class, it's specific to things that
1452
# use weaves or knits -- mbp 20070207
1453
def _get_versioned_file_store(self,
1458
versionedfile_class=None,
1459
versionedfile_kwargs={},
1461
if versionedfile_class is None:
1462
versionedfile_class = self._versionedfile_class
1463
weave_transport = control_files._transport.clone(name)
1464
dir_mode = control_files._dir_mode
1465
file_mode = control_files._file_mode
1466
return VersionedFileStore(weave_transport, prefixed=prefixed,
1468
file_mode=file_mode,
1469
versionedfile_class=versionedfile_class,
1470
versionedfile_kwargs=versionedfile_kwargs,
1473
def initialize(self, a_bzrdir, shared=False):
1474
"""Initialize a repository of this format in a_bzrdir.
1476
:param a_bzrdir: The bzrdir to put the new repository in it.
1477
:param shared: The repository should be initialized as a sharable one.
1478
:returns: The new repository object.
1480
This may raise UninitializableFormat if shared repository are not
1481
compatible the a_bzrdir.
1483
raise NotImplementedError(self.initialize)
1485
def is_supported(self):
1486
"""Is this format supported?
1488
Supported formats must be initializable and openable.
1489
Unsupported formats may not support initialization or committing or
1490
some other features depending on the reason for not being supported.
1494
def check_conversion_target(self, target_format):
1495
raise NotImplementedError(self.check_conversion_target)
1497
def open(self, a_bzrdir, _found=False):
1498
"""Return an instance of this format for the bzrdir a_bzrdir.
1500
_found is a private parameter, do not use it.
1502
raise NotImplementedError(self.open)
1505
class MetaDirRepositoryFormat(RepositoryFormat):
1506
"""Common base class for the new repositories using the metadir layout."""
1508
rich_root_data = False
1509
supports_tree_reference = False
1510
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1513
super(MetaDirRepositoryFormat, self).__init__()
1515
def _create_control_files(self, a_bzrdir):
1516
"""Create the required files and the initial control_files object."""
1517
# FIXME: RBC 20060125 don't peek under the covers
1518
# NB: no need to escape relative paths that are url safe.
1519
repository_transport = a_bzrdir.get_repository_transport(self)
1520
control_files = lockable_files.LockableFiles(repository_transport,
1521
'lock', lockdir.LockDir)
1522
control_files.create_lock()
1523
return control_files
1525
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1526
"""Upload the initial blank content."""
1527
control_files = self._create_control_files(a_bzrdir)
1528
control_files.lock_write()
1530
control_files._transport.mkdir_multi(dirs,
1531
mode=control_files._dir_mode)
1532
for file, content in files:
1533
control_files.put(file, content)
1534
for file, content in utf8_files:
1535
control_files.put_utf8(file, content)
1537
control_files.put_utf8('shared-storage', '')
1539
control_files.unlock()
1542
# formats which have no format string are not discoverable
1543
# and not independently creatable, so are not registered. They're
1544
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1545
# needed, it's constructed directly by the BzrDir. Non-native formats where
1546
# the repository is not separately opened are similar.
1548
format_registry.register_lazy(
1549
'Bazaar-NG Repository format 7',
1550
'bzrlib.repofmt.weaverepo',
1553
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1554
# default control directory format
1556
format_registry.register_lazy(
1557
'Bazaar-NG Knit Repository Format 1',
1558
'bzrlib.repofmt.knitrepo',
1559
'RepositoryFormatKnit1',
1561
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1563
format_registry.register_lazy(
1564
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1565
'bzrlib.repofmt.knitrepo',
1566
'RepositoryFormatKnit3',
1570
class InterRepository(InterObject):
1571
"""This class represents operations taking place between two repositories.
1573
Its instances have methods like copy_content and fetch, and contain
1574
references to the source and target repositories these operations can be
1577
Often we will provide convenience methods on 'repository' which carry out
1578
operations with another repository - they will always forward to
1579
InterRepository.get(other).method_name(parameters).
1583
"""The available optimised InterRepository types."""
1585
def copy_content(self, revision_id=None):
1586
raise NotImplementedError(self.copy_content)
1588
def fetch(self, revision_id=None, pb=None):
1589
"""Fetch the content required to construct revision_id.
1591
The content is copied from self.source to self.target.
1593
:param revision_id: if None all content is copied, if NULL_REVISION no
1595
:param pb: optional progress bar to use for progress reports. If not
1596
provided a default one will be created.
1598
Returns the copied revision count and the failed revisions in a tuple:
1601
raise NotImplementedError(self.fetch)
1604
def missing_revision_ids(self, revision_id=None):
1605
"""Return the revision ids that source has that target does not.
1607
These are returned in topological order.
1609
:param revision_id: only return revision ids included by this
1612
# generic, possibly worst case, slow code path.
1613
target_ids = set(self.target.all_revision_ids())
1614
if revision_id is not None:
1615
# TODO: jam 20070210 InterRepository is internal enough that it
1616
# should assume revision_ids are already utf-8
1617
revision_id = osutils.safe_revision_id(revision_id)
1618
source_ids = self.source.get_ancestry(revision_id)
1619
assert source_ids[0] is None
1622
source_ids = self.source.all_revision_ids()
1623
result_set = set(source_ids).difference(target_ids)
1624
# this may look like a no-op: its not. It preserves the ordering
1625
# other_ids had while only returning the members from other_ids
1626
# that we've decided we need.
1627
return [rev_id for rev_id in source_ids if rev_id in result_set]
1630
class InterSameDataRepository(InterRepository):
1631
"""Code for converting between repositories that represent the same data.
1633
Data format and model must match for this to work.
1637
def _get_repo_format_to_test(self):
1638
"""Repository format for testing with."""
1639
return RepositoryFormat.get_default_format()
1642
def is_compatible(source, target):
1643
if source.supports_rich_root() != target.supports_rich_root():
1645
if source._serializer != target._serializer:
1650
def copy_content(self, revision_id=None):
1651
"""Make a complete copy of the content in self into destination.
1653
This copies both the repository's revision data, and configuration information
1654
such as the make_working_trees setting.
1656
This is a destructive operation! Do not use it on existing
1659
:param revision_id: Only copy the content needed to construct
1660
revision_id and its parents.
1663
self.target.set_make_working_trees(self.source.make_working_trees())
1664
except NotImplementedError:
1666
# TODO: jam 20070210 This is fairly internal, so we should probably
1667
# just assert that revision_id is not unicode.
1668
revision_id = osutils.safe_revision_id(revision_id)
1669
# but don't bother fetching if we have the needed data now.
1670
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1671
self.target.has_revision(revision_id)):
1673
self.target.fetch(self.source, revision_id=revision_id)
1676
def fetch(self, revision_id=None, pb=None):
1677
"""See InterRepository.fetch()."""
1678
from bzrlib.fetch import GenericRepoFetcher
1679
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1680
self.source, self.source._format, self.target,
1681
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
class InterWeaveRepo(InterSameDataRepository):
1692
"""Optimised code paths between Weave based repositories."""
1695
def _get_repo_format_to_test(self):
1696
from bzrlib.repofmt import weaverepo
1697
return weaverepo.RepositoryFormat7()
1700
def is_compatible(source, target):
1701
"""Be compatible with known Weave formats.
1703
We don't test for the stores being of specific types because that
1704
could lead to confusing results, and there is no need to be
1707
from bzrlib.repofmt.weaverepo import (
1713
return (isinstance(source._format, (RepositoryFormat5,
1715
RepositoryFormat7)) and
1716
isinstance(target._format, (RepositoryFormat5,
1718
RepositoryFormat7)))
1719
except AttributeError:
1723
def copy_content(self, revision_id=None):
1724
"""See InterRepository.copy_content()."""
1725
# weave specific optimised path:
1726
# TODO: jam 20070210 Internal, should be an assert, not translate
1727
revision_id = osutils.safe_revision_id(revision_id)
1729
self.target.set_make_working_trees(self.source.make_working_trees())
1730
except NotImplementedError:
1732
# FIXME do not peek!
1733
if self.source.control_files._transport.listable():
1734
pb = ui.ui_factory.nested_progress_bar()
1736
self.target.weave_store.copy_all_ids(
1737
self.source.weave_store,
1739
from_transaction=self.source.get_transaction(),
1740
to_transaction=self.target.get_transaction())
1741
pb.update('copying inventory', 0, 1)
1742
self.target.control_weaves.copy_multi(
1743
self.source.control_weaves, ['inventory'],
1744
from_transaction=self.source.get_transaction(),
1745
to_transaction=self.target.get_transaction())
1746
self.target._revision_store.text_store.copy_all_ids(
1747
self.source._revision_store.text_store,
1752
self.target.fetch(self.source, revision_id=revision_id)
1755
def fetch(self, revision_id=None, pb=None):
1756
"""See InterRepository.fetch()."""
1757
from bzrlib.fetch import GenericRepoFetcher
1758
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1759
self.source, self.source._format, self.target, self.target._format)
1760
# TODO: jam 20070210 This should be an assert, not a translate
1761
revision_id = osutils.safe_revision_id(revision_id)
1762
f = GenericRepoFetcher(to_repository=self.target,
1763
from_repository=self.source,
1764
last_revision=revision_id,
1766
return f.count_copied, f.failed_revisions
1769
def missing_revision_ids(self, revision_id=None):
1770
"""See InterRepository.missing_revision_ids()."""
1771
# we want all revisions to satisfy revision_id in source.
1772
# but we don't want to stat every file here and there.
1773
# we want then, all revisions other needs to satisfy revision_id
1774
# checked, but not those that we have locally.
1775
# so the first thing is to get a subset of the revisions to
1776
# satisfy revision_id in source, and then eliminate those that
1777
# we do already have.
1778
# this is slow on high latency connection to self, but as as this
1779
# disk format scales terribly for push anyway due to rewriting
1780
# inventory.weave, this is considered acceptable.
1782
if revision_id is not None:
1783
source_ids = self.source.get_ancestry(revision_id)
1784
assert source_ids[0] is None
1787
source_ids = self.source._all_possible_ids()
1788
source_ids_set = set(source_ids)
1789
# source_ids is the worst possible case we may need to pull.
1790
# now we want to filter source_ids against what we actually
1791
# have in target, but don't try to check for existence where we know
1792
# we do not have a revision as that would be pointless.
1793
target_ids = set(self.target._all_possible_ids())
1794
possibly_present_revisions = target_ids.intersection(source_ids_set)
1795
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1796
required_revisions = source_ids_set.difference(actually_present_revisions)
1797
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1798
if revision_id is not None:
1799
# we used get_ancestry to determine source_ids then we are assured all
1800
# revisions referenced are present as they are installed in topological order.
1801
# and the tip revision was validated by get_ancestry.
1802
return required_topo_revisions
1804
# if we just grabbed the possibly available ids, then
1805
# we only have an estimate of whats available and need to validate
1806
# that against the revision records.
1807
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1810
class InterKnitRepo(InterSameDataRepository):
1811
"""Optimised code paths between Knit based repositories."""
1814
def _get_repo_format_to_test(self):
1815
from bzrlib.repofmt import knitrepo
1816
return knitrepo.RepositoryFormatKnit1()
1819
def is_compatible(source, target):
1820
"""Be compatible with known Knit formats.
1822
We don't test for the stores being of specific types because that
1823
could lead to confusing results, and there is no need to be
1826
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1828
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1829
isinstance(target._format, (RepositoryFormatKnit1)))
1830
except AttributeError:
1834
def fetch(self, revision_id=None, pb=None):
1835
"""See InterRepository.fetch()."""
1836
from bzrlib.fetch import KnitRepoFetcher
1837
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1838
self.source, self.source._format, self.target, self.target._format)
1839
# TODO: jam 20070210 This should be an assert, not a translate
1840
revision_id = osutils.safe_revision_id(revision_id)
1841
f = KnitRepoFetcher(to_repository=self.target,
1842
from_repository=self.source,
1843
last_revision=revision_id,
1845
return f.count_copied, f.failed_revisions
1848
def missing_revision_ids(self, revision_id=None):
1849
"""See InterRepository.missing_revision_ids()."""
1850
if revision_id is not None:
1851
source_ids = self.source.get_ancestry(revision_id)
1852
assert source_ids[0] is None
1855
source_ids = self.source._all_possible_ids()
1856
source_ids_set = set(source_ids)
1857
# source_ids is the worst possible case we may need to pull.
1858
# now we want to filter source_ids against what we actually
1859
# have in target, but don't try to check for existence where we know
1860
# we do not have a revision as that would be pointless.
1861
target_ids = set(self.target._all_possible_ids())
1862
possibly_present_revisions = target_ids.intersection(source_ids_set)
1863
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1864
required_revisions = source_ids_set.difference(actually_present_revisions)
1865
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1866
if revision_id is not None:
1867
# we used get_ancestry to determine source_ids then we are assured all
1868
# revisions referenced are present as they are installed in topological order.
1869
# and the tip revision was validated by get_ancestry.
1870
return required_topo_revisions
1872
# if we just grabbed the possibly available ids, then
1873
# we only have an estimate of whats available and need to validate
1874
# that against the revision records.
1875
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1878
class InterModel1and2(InterRepository):
1881
def _get_repo_format_to_test(self):
1885
def is_compatible(source, target):
1886
if not source.supports_rich_root() and target.supports_rich_root():
1892
def fetch(self, revision_id=None, pb=None):
1893
"""See InterRepository.fetch()."""
1894
from bzrlib.fetch import Model1toKnit2Fetcher
1895
# TODO: jam 20070210 This should be an assert, not a translate
1896
revision_id = osutils.safe_revision_id(revision_id)
1897
f = Model1toKnit2Fetcher(to_repository=self.target,
1898
from_repository=self.source,
1899
last_revision=revision_id,
1901
return f.count_copied, f.failed_revisions
1904
def copy_content(self, revision_id=None):
1905
"""Make a complete copy of the content in self into destination.
1907
This is a destructive operation! Do not use it on existing
1910
:param revision_id: Only copy the content needed to construct
1911
revision_id and its parents.
1914
self.target.set_make_working_trees(self.source.make_working_trees())
1915
except NotImplementedError:
1917
# TODO: jam 20070210 Internal, assert, don't translate
1918
revision_id = osutils.safe_revision_id(revision_id)
1919
# but don't bother fetching if we have the needed data now.
1920
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1921
self.target.has_revision(revision_id)):
1923
self.target.fetch(self.source, revision_id=revision_id)
1926
class InterKnit1and2(InterKnitRepo):
1929
def _get_repo_format_to_test(self):
1933
def is_compatible(source, target):
1934
"""Be compatible with Knit1 source and Knit3 target"""
1935
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1937
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1938
RepositoryFormatKnit3
1939
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1940
isinstance(target._format, (RepositoryFormatKnit3)))
1941
except AttributeError:
1945
def fetch(self, revision_id=None, pb=None):
1946
"""See InterRepository.fetch()."""
1947
from bzrlib.fetch import Knit1to2Fetcher
1948
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1949
self.source, self.source._format, self.target,
1950
self.target._format)
1951
# TODO: jam 20070210 This should be an assert, not a translate
1952
revision_id = osutils.safe_revision_id(revision_id)
1953
f = Knit1to2Fetcher(to_repository=self.target,
1954
from_repository=self.source,
1955
last_revision=revision_id,
1957
return f.count_copied, f.failed_revisions
1960
class InterRemoteRepository(InterRepository):
1961
"""Code for converting between RemoteRepository objects.
1963
This just gets an non-remote repository from the RemoteRepository, and calls
1964
InterRepository.get again.
1967
def __init__(self, source, target):
1968
if isinstance(source, remote.RemoteRepository):
1969
source._ensure_real()
1970
real_source = source._real_repository
1972
real_source = source
1973
if isinstance(target, remote.RemoteRepository):
1974
target._ensure_real()
1975
real_target = target._real_repository
1977
real_target = target
1978
self.real_inter = InterRepository.get(real_source, real_target)
1981
def is_compatible(source, target):
1982
if isinstance(source, remote.RemoteRepository):
1984
if isinstance(target, remote.RemoteRepository):
1988
def copy_content(self, revision_id=None):
1989
self.real_inter.copy_content(revision_id=revision_id)
1991
def fetch(self, revision_id=None, pb=None):
1992
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1995
def _get_repo_format_to_test(self):
1999
InterRepository.register_optimiser(InterSameDataRepository)
2000
InterRepository.register_optimiser(InterWeaveRepo)
2001
InterRepository.register_optimiser(InterKnitRepo)
2002
InterRepository.register_optimiser(InterModel1and2)
2003
InterRepository.register_optimiser(InterKnit1and2)
2004
InterRepository.register_optimiser(InterRemoteRepository)
2007
class CopyConverter(object):
2008
"""A repository conversion tool which just performs a copy of the content.
2010
This is slow but quite reliable.
2013
def __init__(self, target_format):
2014
"""Create a CopyConverter.
2016
:param target_format: The format the resulting repository should be.
2018
self.target_format = target_format
2020
def convert(self, repo, pb):
2021
"""Perform the conversion of to_convert, giving feedback via pb.
2023
:param to_convert: The disk object to convert.
2024
:param pb: a progress bar to use for progress information.
2029
# this is only useful with metadir layouts - separated repo content.
2030
# trigger an assertion if not such
2031
repo._format.get_format_string()
2032
self.repo_dir = repo.bzrdir
2033
self.step('Moving repository to repository.backup')
2034
self.repo_dir.transport.move('repository', 'repository.backup')
2035
backup_transport = self.repo_dir.transport.clone('repository.backup')
2036
repo._format.check_conversion_target(self.target_format)
2037
self.source_repo = repo._format.open(self.repo_dir,
2039
_override_transport=backup_transport)
2040
self.step('Creating new repository')
2041
converted = self.target_format.initialize(self.repo_dir,
2042
self.source_repo.is_shared())
2043
converted.lock_write()
2045
self.step('Copying content into repository.')
2046
self.source_repo.copy_content_into(converted)
2049
self.step('Deleting old repository content.')
2050
self.repo_dir.transport.delete_tree('repository.backup')
2051
self.pb.note('repository converted')
2053
def step(self, message):
2054
"""Update the pb by a step."""
2056
self.pb.update(message, self.count, self.total)
2059
class CommitBuilder(object):
2060
"""Provides an interface to build up a commit.
2062
This allows describing a tree to be committed without needing to
2063
know the internals of the format of the repository.
2066
record_root_entry = False
2067
def __init__(self, repository, parents, config, timestamp=None,
2068
timezone=None, committer=None, revprops=None,
2070
"""Initiate a CommitBuilder.
2072
:param repository: Repository to commit to.
2073
:param parents: Revision ids of the parents of the new revision.
2074
:param config: Configuration to use.
2075
:param timestamp: Optional timestamp recorded for commit.
2076
:param timezone: Optional timezone for timestamp.
2077
:param committer: Optional committer to set for commit.
2078
:param revprops: Optional dictionary of revision properties.
2079
:param revision_id: Optional revision id.
2081
self._config = config
2083
if committer is None:
2084
self._committer = self._config.username()
2086
assert isinstance(committer, basestring), type(committer)
2087
self._committer = committer
2089
self.new_inventory = Inventory(None)
2090
self._new_revision_id = osutils.safe_revision_id(revision_id)
2091
self.parents = parents
2092
self.repository = repository
2095
if revprops is not None:
2096
self._revprops.update(revprops)
2098
if timestamp is None:
2099
timestamp = time.time()
2100
# Restrict resolution to 1ms
2101
self._timestamp = round(timestamp, 3)
2103
if timezone is None:
2104
self._timezone = osutils.local_time_offset()
2106
self._timezone = int(timezone)
2108
self._generate_revision_if_needed()
2110
def commit(self, message):
2111
"""Make the actual commit.
2113
:return: The revision id of the recorded revision.
2115
rev = _mod_revision.Revision(
2116
timestamp=self._timestamp,
2117
timezone=self._timezone,
2118
committer=self._committer,
2120
inventory_sha1=self.inv_sha1,
2121
revision_id=self._new_revision_id,
2122
properties=self._revprops)
2123
rev.parent_ids = self.parents
2124
self.repository.add_revision(self._new_revision_id, rev,
2125
self.new_inventory, self._config)
2126
self.repository.commit_write_group()
2127
return self._new_revision_id
2129
def revision_tree(self):
2130
"""Return the tree that was just committed.
2132
After calling commit() this can be called to get a RevisionTree
2133
representing the newly committed tree. This is preferred to
2134
calling Repository.revision_tree() because that may require
2135
deserializing the inventory, while we already have a copy in
2138
return RevisionTree(self.repository, self.new_inventory,
2139
self._new_revision_id)
2141
def finish_inventory(self):
2142
"""Tell the builder that the inventory is finished."""
2143
if self.new_inventory.root is None:
2144
symbol_versioning.warn('Root entry should be supplied to'
2145
' record_entry_contents, as of bzr 0.10.',
2146
DeprecationWarning, stacklevel=2)
2147
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2148
self.new_inventory.revision_id = self._new_revision_id
2149
self.inv_sha1 = self.repository.add_inventory(
2150
self._new_revision_id,
2155
def _gen_revision_id(self):
2156
"""Return new revision-id."""
2157
return generate_ids.gen_revision_id(self._config.username(),
2160
def _generate_revision_if_needed(self):
2161
"""Create a revision id if None was supplied.
2163
If the repository can not support user-specified revision ids
2164
they should override this function and raise CannotSetRevisionId
2165
if _new_revision_id is not None.
2167
:raises: CannotSetRevisionId
2169
if self._new_revision_id is None:
2170
self._new_revision_id = self._gen_revision_id()
2172
def record_entry_contents(self, ie, parent_invs, path, tree):
2173
"""Record the content of ie from tree into the commit if needed.
2175
Side effect: sets ie.revision when unchanged
2177
:param ie: An inventory entry present in the commit.
2178
:param parent_invs: The inventories of the parent revisions of the
2180
:param path: The path the entry is at in the tree.
2181
:param tree: The tree which contains this entry and should be used to
2184
if self.new_inventory.root is None and ie.parent_id is not None:
2185
symbol_versioning.warn('Root entry should be supplied to'
2186
' record_entry_contents, as of bzr 0.10.',
2187
DeprecationWarning, stacklevel=2)
2188
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2190
self.new_inventory.add(ie)
2192
# ie.revision is always None if the InventoryEntry is considered
2193
# for committing. ie.snapshot will record the correct revision
2194
# which may be the sole parent if it is untouched.
2195
if ie.revision is not None:
2198
# In this revision format, root entries have no knit or weave
2199
if ie is self.new_inventory.root:
2200
# When serializing out to disk and back in
2201
# root.revision is always _new_revision_id
2202
ie.revision = self._new_revision_id
2204
previous_entries = ie.find_previous_heads(
2206
self.repository.weave_store,
2207
self.repository.get_transaction())
2208
# we are creating a new revision for ie in the history store
2210
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2212
def modified_directory(self, file_id, file_parents):
2213
"""Record the presence of a symbolic link.
2215
:param file_id: The file_id of the link to record.
2216
:param file_parents: The per-file parent revision ids.
2218
self._add_text_to_weave(file_id, [], file_parents.keys())
2220
def modified_reference(self, file_id, file_parents):
2221
"""Record the modification of a reference.
2223
:param file_id: The file_id of the link to record.
2224
:param file_parents: The per-file parent revision ids.
2226
self._add_text_to_weave(file_id, [], file_parents.keys())
2228
def modified_file_text(self, file_id, file_parents,
2229
get_content_byte_lines, text_sha1=None,
2231
"""Record the text of file file_id
2233
:param file_id: The file_id of the file to record the text of.
2234
:param file_parents: The per-file parent revision ids.
2235
:param get_content_byte_lines: A callable which will return the byte
2237
:param text_sha1: Optional SHA1 of the file contents.
2238
:param text_size: Optional size of the file contents.
2240
# mutter('storing text of file {%s} in revision {%s} into %r',
2241
# file_id, self._new_revision_id, self.repository.weave_store)
2242
# special case to avoid diffing on renames or
2244
if (len(file_parents) == 1
2245
and text_sha1 == file_parents.values()[0].text_sha1
2246
and text_size == file_parents.values()[0].text_size):
2247
previous_ie = file_parents.values()[0]
2248
versionedfile = self.repository.weave_store.get_weave(file_id,
2249
self.repository.get_transaction())
2250
versionedfile.clone_text(self._new_revision_id,
2251
previous_ie.revision, file_parents.keys())
2252
return text_sha1, text_size
2254
new_lines = get_content_byte_lines()
2255
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2256
# should return the SHA1 and size
2257
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2258
return osutils.sha_strings(new_lines), \
2259
sum(map(len, new_lines))
2261
def modified_link(self, file_id, file_parents, link_target):
2262
"""Record the presence of a symbolic link.
2264
:param file_id: The file_id of the link to record.
2265
:param file_parents: The per-file parent revision ids.
2266
:param link_target: Target location of this link.
2268
self._add_text_to_weave(file_id, [], file_parents.keys())
2270
def _add_text_to_weave(self, file_id, new_lines, parents):
2271
versionedfile = self.repository.weave_store.get_weave_or_empty(
2272
file_id, self.repository.get_transaction())
2273
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2274
versionedfile.clear_cache()
2277
class _CommitBuilder(CommitBuilder):
2278
"""Temporary class so old CommitBuilders are detected properly
2280
Note: CommitBuilder works whether or not root entry is recorded.
2283
record_root_entry = True
2286
class RootCommitBuilder(CommitBuilder):
2287
"""This commitbuilder actually records the root id"""
2289
record_root_entry = True
2291
def record_entry_contents(self, ie, parent_invs, path, tree):
2292
"""Record the content of ie from tree into the commit if needed.
2294
Side effect: sets ie.revision when unchanged
2296
:param ie: An inventory entry present in the commit.
2297
:param parent_invs: The inventories of the parent revisions of the
2299
:param path: The path the entry is at in the tree.
2300
:param tree: The tree which contains this entry and should be used to
2303
assert self.new_inventory.root is not None or ie.parent_id is None
2304
self.new_inventory.add(ie)
2306
# ie.revision is always None if the InventoryEntry is considered
2307
# for committing. ie.snapshot will record the correct revision
2308
# which may be the sole parent if it is untouched.
2309
if ie.revision is not None:
2312
previous_entries = ie.find_previous_heads(
2314
self.repository.weave_store,
2315
self.repository.get_transaction())
2316
# we are creating a new revision for ie in the history store
2318
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2330
def _unescaper(match, _map=_unescape_map):
2331
code = match.group(1)
2335
if not code.startswith('#'):
2337
return unichr(int(code[1:])).encode('utf8')
2343
def _unescape_xml(data):
2344
"""Unescape predefined XML entities in a string of data."""
2346
if _unescape_re is None:
2347
_unescape_re = re.compile('\&([^;]*);')
2348
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
