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
250
self._reconcile_does_inventory_gc = True
251
# not right yet - should be more semantically clear ?
253
self.control_store = control_store
254
self.control_weaves = control_store
255
# TODO: make sure to construct the right store classes, etc, depending
256
# on whether escaping is required.
257
self._warn_if_deprecated()
258
self._write_group = None
261
return '%s(%r)' % (self.__class__.__name__,
262
self.bzrdir.transport.base)
264
def has_same_location(self, other):
265
"""Returns a boolean indicating if this repository is at the same
266
location as another repository.
268
This might return False even when two repository objects are accessing
269
the same physical repository via different URLs.
271
if self.__class__ is not other.__class__:
273
return (self.control_files._transport.base ==
274
other.control_files._transport.base)
276
def is_in_write_group(self):
277
"""Return True if there is an open write group.
279
:seealso: start_write_group.
281
return self._write_group is not None
284
return self.control_files.is_locked()
286
def lock_write(self, token=None):
287
"""Lock this repository for writing.
289
This causes caching within the repository obejct to start accumlating
290
data during reads, and allows a 'write_group' to be obtained. Write
291
groups must be used for actual data insertion.
293
:param token: if this is already locked, then lock_write will fail
294
unless the token matches the existing lock.
295
:returns: a token if this instance supports tokens, otherwise None.
296
:raises TokenLockingNotSupported: when a token is given but this
297
instance doesn't support using token locks.
298
:raises MismatchedToken: if the specified token doesn't match the token
299
of the existing lock.
300
:seealso: start_write_group.
302
A token should be passed in if you know that you have locked the object
303
some other way, and need to synchronise this object's state with that
306
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
308
result = self.control_files.lock_write(token=token)
313
self.control_files.lock_read()
316
def get_physical_lock_status(self):
317
return self.control_files.get_physical_lock_status()
319
def leave_lock_in_place(self):
320
"""Tell this repository not to release the physical lock when this
323
If lock_write doesn't return a token, then this method is not supported.
325
self.control_files.leave_in_place()
327
def dont_leave_lock_in_place(self):
328
"""Tell this repository to release the physical lock when this
329
object is unlocked, even if it didn't originally acquire it.
331
If lock_write doesn't return a token, then this method is not supported.
333
self.control_files.dont_leave_in_place()
336
def gather_stats(self, revid=None, committers=None):
337
"""Gather statistics from a revision id.
339
:param revid: The revision id to gather statistics from, if None, then
340
no revision specific statistics are gathered.
341
:param committers: Optional parameter controlling whether to grab
342
a count of committers from the revision specific statistics.
343
:return: A dictionary of statistics. Currently this contains:
344
committers: The number of committers if requested.
345
firstrev: A tuple with timestamp, timezone for the penultimate left
346
most ancestor of revid, if revid is not the NULL_REVISION.
347
latestrev: A tuple with timestamp, timezone for revid, if revid is
348
not the NULL_REVISION.
349
revisions: The total revision count in the repository.
350
size: An estimate disk size of the repository in bytes.
353
if revid and committers:
354
result['committers'] = 0
355
if revid and revid != _mod_revision.NULL_REVISION:
357
all_committers = set()
358
revisions = self.get_ancestry(revid)
359
# pop the leading None
361
first_revision = None
363
# ignore the revisions in the middle - just grab first and last
364
revisions = revisions[0], revisions[-1]
365
for revision in self.get_revisions(revisions):
366
if not first_revision:
367
first_revision = revision
369
all_committers.add(revision.committer)
370
last_revision = revision
372
result['committers'] = len(all_committers)
373
result['firstrev'] = (first_revision.timestamp,
374
first_revision.timezone)
375
result['latestrev'] = (last_revision.timestamp,
376
last_revision.timezone)
378
# now gather global repository information
379
if self.bzrdir.root_transport.listable():
380
c, t = self._revision_store.total_size(self.get_transaction())
381
result['revisions'] = c
386
def missing_revision_ids(self, other, revision_id=None):
387
"""Return the revision ids that other has that this does not.
389
These are returned in topological order.
391
revision_id: only return revision ids included by revision_id.
393
revision_id = osutils.safe_revision_id(revision_id)
394
return InterRepository.get(other, self).missing_revision_ids(revision_id)
398
"""Open the repository rooted at base.
400
For instance, if the repository is at URL/.bzr/repository,
401
Repository.open(URL) -> a Repository instance.
403
control = bzrdir.BzrDir.open(base)
404
return control.open_repository()
406
def copy_content_into(self, destination, revision_id=None):
407
"""Make a complete copy of the content in self into destination.
409
This is a destructive operation! Do not use it on existing
412
revision_id = osutils.safe_revision_id(revision_id)
413
return InterRepository.get(self, destination).copy_content(revision_id)
415
def commit_write_group(self):
416
"""Commit the contents accrued within the current write group.
418
:seealso: start_write_group.
420
if self._write_group is not self.get_transaction():
421
# has an unlock or relock occured ?
422
raise errors.BzrError('mismatched lock context and write group.')
423
self._commit_write_group()
424
self._write_group = None
426
def _commit_write_group(self):
427
"""Template method for per-repository write group cleanup.
429
This is called before the write group is considered to be
430
finished and should ensure that all data handed to the repository
431
for writing during the write group is safely committed (to the
432
extent possible considering file system caching etc).
435
def fetch(self, source, revision_id=None, pb=None):
436
"""Fetch the content required to construct revision_id from source.
438
If revision_id is None all content is copied.
440
revision_id = osutils.safe_revision_id(revision_id)
441
inter = InterRepository.get(source, self)
443
return inter.fetch(revision_id=revision_id, pb=pb)
444
except NotImplementedError:
445
raise errors.IncompatibleRepositories(source, self)
447
def create_bundle(self, target, base, fileobj, format=None):
448
return serializer.write_bundle(self, target, base, fileobj, format)
450
def get_commit_builder(self, branch, parents, config, timestamp=None,
451
timezone=None, committer=None, revprops=None,
453
"""Obtain a CommitBuilder for this repository.
455
:param branch: Branch to commit to.
456
:param parents: Revision ids of the parents of the new revision.
457
:param config: Configuration to use.
458
:param timestamp: Optional timestamp recorded for commit.
459
:param timezone: Optional timezone for timestamp.
460
:param committer: Optional committer to set for commit.
461
:param revprops: Optional dictionary of revision properties.
462
:param revision_id: Optional revision id.
464
revision_id = osutils.safe_revision_id(revision_id)
465
result =_CommitBuilder(self, parents, config, timestamp, timezone,
466
committer, revprops, revision_id)
467
self.start_write_group()
471
if (self.control_files._lock_count == 1 and
472
self.control_files._lock_mode == 'w'):
473
if self._write_group is not None:
474
raise errors.BzrError(
475
'Must end write groups before releasing write locks.')
476
self.control_files.unlock()
479
def clone(self, a_bzrdir, revision_id=None):
480
"""Clone this repository into a_bzrdir using the current format.
482
Currently no check is made that the format of this repository and
483
the bzrdir format are compatible. FIXME RBC 20060201.
485
:return: The newly created destination repository.
487
# TODO: deprecate after 0.16; cloning this with all its settings is
488
# probably not very useful -- mbp 20070423
489
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
490
self.copy_content_into(dest_repo, revision_id)
493
def start_write_group(self):
494
"""Start a write group in the repository.
496
Write groups are used by repositories which do not have a 1:1 mapping
497
between file ids and backend store to manage the insertion of data from
498
both fetch and commit operations.
500
A write lock is required around the start_write_group/commit_write_group
501
for the support of lock-requiring repository formats.
503
One can only insert data into a repository inside a write group.
507
if not self.is_locked() or self.control_files._lock_mode != 'w':
508
raise errors.NotWriteLocked(self)
509
if self._write_group:
510
raise errors.BzrError('already in a write group')
511
self._start_write_group()
512
# so we can detect unlock/relock - the write group is now entered.
513
self._write_group = self.get_transaction()
515
def _start_write_group(self):
516
"""Template method for per-repository write group startup.
518
This is called before the write group is considered to be
523
def sprout(self, to_bzrdir, revision_id=None):
524
"""Create a descendent repository for new development.
526
Unlike clone, this does not copy the settings of the repository.
528
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
529
dest_repo.fetch(self, revision_id=revision_id)
532
def _create_sprouting_repo(self, a_bzrdir, shared):
533
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
534
# use target default format.
535
dest_repo = a_bzrdir.create_repository()
537
# Most control formats need the repository to be specifically
538
# created, but on some old all-in-one formats it's not needed
540
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
541
except errors.UninitializableFormat:
542
dest_repo = a_bzrdir.open_repository()
546
def has_revision(self, revision_id):
547
"""True if this repository has a copy of the revision."""
548
revision_id = osutils.safe_revision_id(revision_id)
549
return self._revision_store.has_revision_id(revision_id,
550
self.get_transaction())
553
def get_revision_reconcile(self, revision_id):
554
"""'reconcile' helper routine that allows access to a revision always.
556
This variant of get_revision does not cross check the weave graph
557
against the revision one as get_revision does: but it should only
558
be used by reconcile, or reconcile-alike commands that are correcting
559
or testing the revision graph.
561
if not revision_id or not isinstance(revision_id, basestring):
562
raise errors.InvalidRevisionId(revision_id=revision_id,
564
return self.get_revisions([revision_id])[0]
567
def get_revisions(self, revision_ids):
568
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
569
revs = self._revision_store.get_revisions(revision_ids,
570
self.get_transaction())
572
assert not isinstance(rev.revision_id, unicode)
573
for parent_id in rev.parent_ids:
574
assert not isinstance(parent_id, unicode)
578
def get_revision_xml(self, revision_id):
579
# TODO: jam 20070210 This shouldn't be necessary since get_revision
580
# would have already do it.
581
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
582
revision_id = osutils.safe_revision_id(revision_id)
583
rev = self.get_revision(revision_id)
585
# the current serializer..
586
self._revision_store._serializer.write_revision(rev, rev_tmp)
588
return rev_tmp.getvalue()
591
def get_revision(self, revision_id):
592
"""Return the Revision object for a named revision"""
593
# TODO: jam 20070210 get_revision_reconcile should do this for us
594
revision_id = osutils.safe_revision_id(revision_id)
595
r = self.get_revision_reconcile(revision_id)
596
# weave corruption can lead to absent revision markers that should be
598
# the following test is reasonably cheap (it needs a single weave read)
599
# and the weave is cached in read transactions. In write transactions
600
# it is not cached but typically we only read a small number of
601
# revisions. For knits when they are introduced we will probably want
602
# to ensure that caching write transactions are in use.
603
inv = self.get_inventory_weave()
604
self._check_revision_parents(r, inv)
608
def get_deltas_for_revisions(self, revisions):
609
"""Produce a generator of revision deltas.
611
Note that the input is a sequence of REVISIONS, not revision_ids.
612
Trees will be held in memory until the generator exits.
613
Each delta is relative to the revision's lefthand predecessor.
615
required_trees = set()
616
for revision in revisions:
617
required_trees.add(revision.revision_id)
618
required_trees.update(revision.parent_ids[:1])
619
trees = dict((t.get_revision_id(), t) for
620
t in self.revision_trees(required_trees))
621
for revision in revisions:
622
if not revision.parent_ids:
623
old_tree = self.revision_tree(None)
625
old_tree = trees[revision.parent_ids[0]]
626
yield trees[revision.revision_id].changes_from(old_tree)
629
def get_revision_delta(self, revision_id):
630
"""Return the delta for one revision.
632
The delta is relative to the left-hand predecessor of the
635
r = self.get_revision(revision_id)
636
return list(self.get_deltas_for_revisions([r]))[0]
638
def _check_revision_parents(self, revision, inventory):
639
"""Private to Repository and Fetch.
641
This checks the parentage of revision in an inventory weave for
642
consistency and is only applicable to inventory-weave-for-ancestry
643
using repository formats & fetchers.
645
weave_parents = inventory.get_parents(revision.revision_id)
646
weave_names = inventory.versions()
647
for parent_id in revision.parent_ids:
648
if parent_id in weave_names:
649
# this parent must not be a ghost.
650
if not parent_id in weave_parents:
652
raise errors.CorruptRepository(self)
655
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
656
revision_id = osutils.safe_revision_id(revision_id)
657
signature = gpg_strategy.sign(plaintext)
658
self._revision_store.add_revision_signature_text(revision_id,
660
self.get_transaction())
662
def fileids_altered_by_revision_ids(self, revision_ids):
663
"""Find the file ids and versions affected by revisions.
665
:param revisions: an iterable containing revision ids.
666
:return: a dictionary mapping altered file-ids to an iterable of
667
revision_ids. Each altered file-ids has the exact revision_ids that
668
altered it listed explicitly.
670
assert self._serializer.support_altered_by_hack, \
671
("fileids_altered_by_revision_ids only supported for branches "
672
"which store inventory as unnested xml, not on %r" % self)
673
selected_revision_ids = set(osutils.safe_revision_id(r)
674
for r in revision_ids)
675
w = self.get_inventory_weave()
678
# this code needs to read every new line in every inventory for the
679
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
680
# not present in one of those inventories is unnecessary but not
681
# harmful because we are filtering by the revision id marker in the
682
# inventory lines : we only select file ids altered in one of those
683
# revisions. We don't need to see all lines in the inventory because
684
# only those added in an inventory in rev X can contain a revision=X
686
unescape_revid_cache = {}
687
unescape_fileid_cache = {}
689
# jam 20061218 In a big fetch, this handles hundreds of thousands
690
# of lines, so it has had a lot of inlining and optimizing done.
691
# Sorry that it is a little bit messy.
692
# Move several functions to be local variables, since this is a long
694
search = self._file_ids_altered_regex.search
695
unescape = _unescape_xml
696
setdefault = result.setdefault
697
pb = ui.ui_factory.nested_progress_bar()
699
for line in w.iter_lines_added_or_present_in_versions(
700
selected_revision_ids, pb=pb):
704
# One call to match.group() returning multiple items is quite a
705
# bit faster than 2 calls to match.group() each returning 1
706
file_id, revision_id = match.group('file_id', 'revision_id')
708
# Inlining the cache lookups helps a lot when you make 170,000
709
# lines and 350k ids, versus 8.4 unique ids.
710
# Using a cache helps in 2 ways:
711
# 1) Avoids unnecessary decoding calls
712
# 2) Re-uses cached strings, which helps in future set and
714
# (2) is enough that removing encoding entirely along with
715
# the cache (so we are using plain strings) results in no
716
# performance improvement.
718
revision_id = unescape_revid_cache[revision_id]
720
unescaped = unescape(revision_id)
721
unescape_revid_cache[revision_id] = unescaped
722
revision_id = unescaped
724
if revision_id in selected_revision_ids:
726
file_id = unescape_fileid_cache[file_id]
728
unescaped = unescape(file_id)
729
unescape_fileid_cache[file_id] = unescaped
731
setdefault(file_id, set()).add(revision_id)
736
def iter_files_bytes(self, desired_files):
737
"""Iterate through file versions.
739
Files will not necessarily be returned in the order they occur in
740
desired_files. No specific order is guaranteed.
742
Yields pairs of identifier, bytes_iterator. identifier is an opaque
743
value supplied by the caller as part of desired_files. It should
744
uniquely identify the file version in the caller's context. (Examples:
745
an index number or a TreeTransform trans_id.)
747
bytes_iterator is an iterable of bytestrings for the file. The
748
kind of iterable and length of the bytestrings are unspecified, but for
749
this implementation, it is a list of lines produced by
750
VersionedFile.get_lines().
752
:param desired_files: a list of (file_id, revision_id, identifier)
755
transaction = self.get_transaction()
756
for file_id, revision_id, callable_data in desired_files:
758
weave = self.weave_store.get_weave(file_id, transaction)
759
except errors.NoSuchFile:
760
raise errors.NoSuchIdInRepository(self, file_id)
761
yield callable_data, weave.get_lines(revision_id)
763
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
764
"""Get an iterable listing the keys of all the data introduced by a set
767
The keys will be ordered so that the corresponding items can be safely
768
fetched and inserted in that order.
770
:returns: An iterable producing tuples of (knit-kind, file-id,
771
versions). knit-kind is one of 'file', 'inventory', 'signatures',
772
'revisions'. file-id is None unless knit-kind is 'file'.
774
# XXX: it's a bit weird to control the inventory weave caching in this
775
# generator. Ideally the caching would be done in fetch.py I think. Or
776
# maybe this generator should explicitly have the contract that it
777
# should not be iterated until the previously yielded item has been
779
inv_w = self.get_inventory_weave()
782
# file ids that changed
783
file_ids = self.fileids_altered_by_revision_ids(revision_ids)
785
num_file_ids = len(file_ids)
786
for file_id, altered_versions in file_ids.iteritems():
787
if _files_pb is not None:
788
_files_pb.update("fetch texts", count, num_file_ids)
790
yield ("file", file_id, altered_versions)
791
# We're done with the files_pb. Note that it finished by the caller,
792
# just as it was created by the caller.
796
yield ("inventory", None, revision_ids)
800
revisions_with_signatures = set()
801
for rev_id in revision_ids:
803
self.get_signature_text(rev_id)
804
except errors.NoSuchRevision:
808
revisions_with_signatures.add(rev_id)
809
yield ("signatures", None, revisions_with_signatures)
812
yield ("revisions", None, revision_ids)
815
def get_inventory_weave(self):
816
return self.control_weaves.get_weave('inventory',
817
self.get_transaction())
820
def get_inventory(self, revision_id):
821
"""Get Inventory object by hash."""
822
# TODO: jam 20070210 Technically we don't need to sanitize, since all
823
# called functions must sanitize.
824
revision_id = osutils.safe_revision_id(revision_id)
825
return self.deserialise_inventory(
826
revision_id, self.get_inventory_xml(revision_id))
828
def deserialise_inventory(self, revision_id, xml):
829
"""Transform the xml into an inventory object.
831
:param revision_id: The expected revision id of the inventory.
832
:param xml: A serialised inventory.
834
revision_id = osutils.safe_revision_id(revision_id)
835
result = self._serializer.read_inventory_from_string(xml)
836
result.root.revision = revision_id
839
def serialise_inventory(self, inv):
840
return self._serializer.write_inventory_to_string(inv)
842
def get_serializer_format(self):
843
return self._serializer.format_num
846
def get_inventory_xml(self, revision_id):
847
"""Get inventory XML as a file object."""
848
revision_id = osutils.safe_revision_id(revision_id)
850
assert isinstance(revision_id, str), type(revision_id)
851
iw = self.get_inventory_weave()
852
return iw.get_text(revision_id)
854
raise errors.HistoryMissing(self, 'inventory', revision_id)
857
def get_inventory_sha1(self, revision_id):
858
"""Return the sha1 hash of the inventory entry
860
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
861
revision_id = osutils.safe_revision_id(revision_id)
862
return self.get_revision(revision_id).inventory_sha1
865
def get_revision_graph(self, revision_id=None):
866
"""Return a dictionary containing the revision graph.
868
:param revision_id: The revision_id to get a graph from. If None, then
869
the entire revision graph is returned. This is a deprecated mode of
870
operation and will be removed in the future.
871
:return: a dictionary of revision_id->revision_parents_list.
873
# special case NULL_REVISION
874
if revision_id == _mod_revision.NULL_REVISION:
876
revision_id = osutils.safe_revision_id(revision_id)
877
a_weave = self.get_inventory_weave()
878
all_revisions = self._eliminate_revisions_not_present(
880
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
881
node in all_revisions])
882
if revision_id is None:
884
elif revision_id not in entire_graph:
885
raise errors.NoSuchRevision(self, revision_id)
887
# add what can be reached from revision_id
889
pending = set([revision_id])
890
while len(pending) > 0:
892
result[node] = entire_graph[node]
893
for revision_id in result[node]:
894
if revision_id not in result:
895
pending.add(revision_id)
899
def get_revision_graph_with_ghosts(self, revision_ids=None):
900
"""Return a graph of the revisions with ghosts marked as applicable.
902
:param revision_ids: an iterable of revisions to graph or None for all.
903
:return: a Graph object with the graph reachable from revision_ids.
905
result = deprecated_graph.Graph()
907
pending = set(self.all_revision_ids())
910
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
911
# special case NULL_REVISION
912
if _mod_revision.NULL_REVISION in pending:
913
pending.remove(_mod_revision.NULL_REVISION)
914
required = set(pending)
917
revision_id = pending.pop()
919
rev = self.get_revision(revision_id)
920
except errors.NoSuchRevision:
921
if revision_id in required:
924
result.add_ghost(revision_id)
926
for parent_id in rev.parent_ids:
927
# is this queued or done ?
928
if (parent_id not in pending and
929
parent_id not in done):
931
pending.add(parent_id)
932
result.add_node(revision_id, rev.parent_ids)
933
done.add(revision_id)
936
def _get_history_vf(self):
937
"""Get a versionedfile whose history graph reflects all revisions.
939
For weave repositories, this is the inventory weave.
941
return self.get_inventory_weave()
943
def iter_reverse_revision_history(self, revision_id):
944
"""Iterate backwards through revision ids in the lefthand history
946
:param revision_id: The revision id to start with. All its lefthand
947
ancestors will be traversed.
949
revision_id = osutils.safe_revision_id(revision_id)
950
if revision_id in (None, _mod_revision.NULL_REVISION):
952
next_id = revision_id
953
versionedfile = self._get_history_vf()
956
parents = versionedfile.get_parents(next_id)
957
if len(parents) == 0:
963
def get_revision_inventory(self, revision_id):
964
"""Return inventory of a past revision."""
965
# TODO: Unify this with get_inventory()
966
# bzr 0.0.6 and later imposes the constraint that the inventory_id
967
# must be the same as its revision, so this is trivial.
968
if revision_id is None:
969
# This does not make sense: if there is no revision,
970
# then it is the current tree inventory surely ?!
971
# and thus get_root_id() is something that looks at the last
972
# commit on the branch, and the get_root_id is an inventory check.
973
raise NotImplementedError
974
# return Inventory(self.get_root_id())
976
return self.get_inventory(revision_id)
980
"""Return True if this repository is flagged as a shared repository."""
981
raise NotImplementedError(self.is_shared)
984
def reconcile(self, other=None, thorough=False):
985
"""Reconcile this repository."""
986
from bzrlib.reconcile import RepoReconciler
987
reconciler = RepoReconciler(self, thorough=thorough)
988
reconciler.reconcile()
991
def _refresh_data(self):
992
"""Helper called from lock_* to ensure coherency with disk.
994
The default implementation does nothing; it is however possible
995
for repositories to maintain loaded indices across multiple locks
996
by checking inside their implementation of this method to see
997
whether their indices are still valid. This depends of course on
998
the disk format being validatable in this manner.
1002
def revision_tree(self, revision_id):
1003
"""Return Tree for a revision on this branch.
1005
`revision_id` may be None for the empty tree revision.
1007
# TODO: refactor this to use an existing revision object
1008
# so we don't need to read it in twice.
1009
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1010
return RevisionTree(self, Inventory(root_id=None),
1011
_mod_revision.NULL_REVISION)
1013
revision_id = osutils.safe_revision_id(revision_id)
1014
inv = self.get_revision_inventory(revision_id)
1015
return RevisionTree(self, inv, revision_id)
1018
def revision_trees(self, revision_ids):
1019
"""Return Tree for a revision on this branch.
1021
`revision_id` may not be None or 'null:'"""
1022
assert None not in revision_ids
1023
assert _mod_revision.NULL_REVISION not in revision_ids
1024
texts = self.get_inventory_weave().get_texts(revision_ids)
1025
for text, revision_id in zip(texts, revision_ids):
1026
inv = self.deserialise_inventory(revision_id, text)
1027
yield RevisionTree(self, inv, revision_id)
1030
def get_ancestry(self, revision_id, topo_sorted=True):
1031
"""Return a list of revision-ids integrated by a revision.
1033
The first element of the list is always None, indicating the origin
1034
revision. This might change when we have history horizons, or
1035
perhaps we should have a new API.
1037
This is topologically sorted.
1039
if _mod_revision.is_null(revision_id):
1041
revision_id = osutils.safe_revision_id(revision_id)
1042
if not self.has_revision(revision_id):
1043
raise errors.NoSuchRevision(self, revision_id)
1044
w = self.get_inventory_weave()
1045
candidates = w.get_ancestry(revision_id, topo_sorted)
1046
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1049
"""Compress the data within the repository.
1051
This operation only makes sense for some repository types. For other
1052
types it should be a no-op that just returns.
1054
This stub method does not require a lock, but subclasses should use
1055
@needs_write_lock as this is a long running call its reasonable to
1056
implicitly lock for the user.
1060
def print_file(self, file, revision_id):
1061
"""Print `file` to stdout.
1063
FIXME RBC 20060125 as John Meinel points out this is a bad api
1064
- it writes to stdout, it assumes that that is valid etc. Fix
1065
by creating a new more flexible convenience function.
1067
revision_id = osutils.safe_revision_id(revision_id)
1068
tree = self.revision_tree(revision_id)
1069
# use inventory as it was in that revision
1070
file_id = tree.inventory.path2id(file)
1072
# TODO: jam 20060427 Write a test for this code path
1073
# it had a bug in it, and was raising the wrong
1075
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1076
tree.print_file(file_id)
1078
def get_transaction(self):
1079
return self.control_files.get_transaction()
1081
def revision_parents(self, revision_id):
1082
revision_id = osutils.safe_revision_id(revision_id)
1083
return self.get_inventory_weave().parent_names(revision_id)
1085
def get_parents(self, revision_ids):
1086
"""See StackedParentsProvider.get_parents"""
1088
for revision_id in revision_ids:
1089
if revision_id == _mod_revision.NULL_REVISION:
1093
parents = self.get_revision(revision_id).parent_ids
1094
except errors.NoSuchRevision:
1097
if len(parents) == 0:
1098
parents = [_mod_revision.NULL_REVISION]
1099
parents_list.append(parents)
1102
def _make_parents_provider(self):
1105
def get_graph(self, other_repository=None):
1106
"""Return the graph walker for this repository format"""
1107
parents_provider = self._make_parents_provider()
1108
if (other_repository is not None and
1109
other_repository.bzrdir.transport.base !=
1110
self.bzrdir.transport.base):
1111
parents_provider = graph._StackedParentsProvider(
1112
[parents_provider, other_repository._make_parents_provider()])
1113
return graph.Graph(parents_provider)
1116
def set_make_working_trees(self, new_value):
1117
"""Set the policy flag for making working trees when creating branches.
1119
This only applies to branches that use this repository.
1121
The default is 'True'.
1122
:param new_value: True to restore the default, False to disable making
1125
raise NotImplementedError(self.set_make_working_trees)
1127
def make_working_trees(self):
1128
"""Returns the policy for making working trees on new branches."""
1129
raise NotImplementedError(self.make_working_trees)
1132
def sign_revision(self, revision_id, gpg_strategy):
1133
revision_id = osutils.safe_revision_id(revision_id)
1134
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1135
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1138
def has_signature_for_revision_id(self, revision_id):
1139
"""Query for a revision signature for revision_id in the repository."""
1140
revision_id = osutils.safe_revision_id(revision_id)
1141
return self._revision_store.has_signature(revision_id,
1142
self.get_transaction())
1145
def get_signature_text(self, revision_id):
1146
"""Return the text for a signature."""
1147
revision_id = osutils.safe_revision_id(revision_id)
1148
return self._revision_store.get_signature_text(revision_id,
1149
self.get_transaction())
1152
def check(self, revision_ids):
1153
"""Check consistency of all history of given revision_ids.
1155
Different repository implementations should override _check().
1157
:param revision_ids: A non-empty list of revision_ids whose ancestry
1158
will be checked. Typically the last revision_id of a branch.
1160
if not revision_ids:
1161
raise ValueError("revision_ids must be non-empty in %s.check"
1163
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1164
return self._check(revision_ids)
1166
def _check(self, revision_ids):
1167
result = check.Check(self)
1171
def _warn_if_deprecated(self):
1172
global _deprecation_warning_done
1173
if _deprecation_warning_done:
1175
_deprecation_warning_done = True
1176
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1177
% (self._format, self.bzrdir.transport.base))
1179
def supports_rich_root(self):
1180
return self._format.rich_root_data
1182
def _check_ascii_revisionid(self, revision_id, method):
1183
"""Private helper for ascii-only repositories."""
1184
# weave repositories refuse to store revisionids that are non-ascii.
1185
if revision_id is not None:
1186
# weaves require ascii revision ids.
1187
if isinstance(revision_id, unicode):
1189
revision_id.encode('ascii')
1190
except UnicodeEncodeError:
1191
raise errors.NonAsciiRevisionId(method, self)
1194
revision_id.decode('ascii')
1195
except UnicodeDecodeError:
1196
raise errors.NonAsciiRevisionId(method, self)
1200
# remove these delegates a while after bzr 0.15
1201
def __make_delegated(name, from_module):
1202
def _deprecated_repository_forwarder():
1203
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1204
% (name, from_module),
1207
m = __import__(from_module, globals(), locals(), [name])
1209
return getattr(m, name)
1210
except AttributeError:
1211
raise AttributeError('module %s has no name %s'
1213
globals()[name] = _deprecated_repository_forwarder
1216
'AllInOneRepository',
1217
'WeaveMetaDirRepository',
1218
'PreSplitOutRepositoryFormat',
1219
'RepositoryFormat4',
1220
'RepositoryFormat5',
1221
'RepositoryFormat6',
1222
'RepositoryFormat7',
1224
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1228
'RepositoryFormatKnit',
1229
'RepositoryFormatKnit1',
1231
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1234
def install_revision(repository, rev, revision_tree):
1235
"""Install all revision data into a repository."""
1236
present_parents = []
1238
for p_id in rev.parent_ids:
1239
if repository.has_revision(p_id):
1240
present_parents.append(p_id)
1241
parent_trees[p_id] = repository.revision_tree(p_id)
1243
parent_trees[p_id] = repository.revision_tree(None)
1245
inv = revision_tree.inventory
1246
entries = inv.iter_entries()
1247
# backwards compatibility hack: skip the root id.
1248
if not repository.supports_rich_root():
1249
path, root = entries.next()
1250
if root.revision != rev.revision_id:
1251
raise errors.IncompatibleRevision(repr(repository))
1252
# Add the texts that are not already present
1253
for path, ie in entries:
1254
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1255
repository.get_transaction())
1256
if ie.revision not in w:
1258
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1259
# with InventoryEntry.find_previous_heads(). if it is, then there
1260
# is a latent bug here where the parents may have ancestors of each
1262
for revision, tree in parent_trees.iteritems():
1263
if ie.file_id not in tree:
1265
parent_id = tree.inventory[ie.file_id].revision
1266
if parent_id in text_parents:
1268
text_parents.append(parent_id)
1270
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1271
repository.get_transaction())
1272
lines = revision_tree.get_file(ie.file_id).readlines()
1273
vfile.add_lines(rev.revision_id, text_parents, lines)
1275
# install the inventory
1276
repository.add_inventory(rev.revision_id, inv, present_parents)
1277
except errors.RevisionAlreadyPresent:
1279
repository.add_revision(rev.revision_id, rev, inv)
1282
class MetaDirRepository(Repository):
1283
"""Repositories in the new meta-dir layout."""
1285
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1286
super(MetaDirRepository, self).__init__(_format,
1292
dir_mode = self.control_files._dir_mode
1293
file_mode = self.control_files._file_mode
1296
def is_shared(self):
1297
"""Return True if this repository is flagged as a shared repository."""
1298
return self.control_files._transport.has('shared-storage')
1301
def set_make_working_trees(self, new_value):
1302
"""Set the policy flag for making working trees when creating branches.
1304
This only applies to branches that use this repository.
1306
The default is 'True'.
1307
:param new_value: True to restore the default, False to disable making
1312
self.control_files._transport.delete('no-working-trees')
1313
except errors.NoSuchFile:
1316
self.control_files.put_utf8('no-working-trees', '')
1318
def make_working_trees(self):
1319
"""Returns the policy for making working trees on new branches."""
1320
return not self.control_files._transport.has('no-working-trees')
1323
class RepositoryFormatRegistry(registry.Registry):
1324
"""Registry of RepositoryFormats.
1327
def get(self, format_string):
1328
r = registry.Registry.get(self, format_string)
1334
format_registry = RepositoryFormatRegistry()
1335
"""Registry of formats, indexed by their identifying format string.
1337
This can contain either format instances themselves, or classes/factories that
1338
can be called to obtain one.
1342
#####################################################################
1343
# Repository Formats
1345
class RepositoryFormat(object):
1346
"""A repository format.
1348
Formats provide three things:
1349
* An initialization routine to construct repository data on disk.
1350
* a format string which is used when the BzrDir supports versioned
1352
* an open routine which returns a Repository instance.
1354
Formats are placed in an dict by their format string for reference
1355
during opening. These should be subclasses of RepositoryFormat
1358
Once a format is deprecated, just deprecate the initialize and open
1359
methods on the format class. Do not deprecate the object, as the
1360
object will be created every system load.
1362
Common instance attributes:
1363
_matchingbzrdir - the bzrdir format that the repository format was
1364
originally written to work with. This can be used if manually
1365
constructing a bzrdir and repository, or more commonly for test suite
1370
return "<%s>" % self.__class__.__name__
1372
def __eq__(self, other):
1373
# format objects are generally stateless
1374
return isinstance(other, self.__class__)
1376
def __ne__(self, other):
1377
return not self == other
1380
def find_format(klass, a_bzrdir):
1381
"""Return the format for the repository object in a_bzrdir.
1383
This is used by bzr native formats that have a "format" file in
1384
the repository. Other methods may be used by different types of
1388
transport = a_bzrdir.get_repository_transport(None)
1389
format_string = transport.get("format").read()
1390
return format_registry.get(format_string)
1391
except errors.NoSuchFile:
1392
raise errors.NoRepositoryPresent(a_bzrdir)
1394
raise errors.UnknownFormatError(format=format_string)
1397
def register_format(klass, format):
1398
format_registry.register(format.get_format_string(), format)
1401
def unregister_format(klass, format):
1402
format_registry.remove(format.get_format_string())
1405
def get_default_format(klass):
1406
"""Return the current default format."""
1407
from bzrlib import bzrdir
1408
return bzrdir.format_registry.make_bzrdir('default').repository_format
1410
def _get_control_store(self, repo_transport, control_files):
1411
"""Return the control store for this repository."""
1412
raise NotImplementedError(self._get_control_store)
1414
def get_format_string(self):
1415
"""Return the ASCII format string that identifies this format.
1417
Note that in pre format ?? repositories the format string is
1418
not permitted nor written to disk.
1420
raise NotImplementedError(self.get_format_string)
1422
def get_format_description(self):
1423
"""Return the short description for this format."""
1424
raise NotImplementedError(self.get_format_description)
1426
def _get_revision_store(self, repo_transport, control_files):
1427
"""Return the revision store object for this a_bzrdir."""
1428
raise NotImplementedError(self._get_revision_store)
1430
def _get_text_rev_store(self,
1437
"""Common logic for getting a revision store for a repository.
1439
see self._get_revision_store for the subclass-overridable method to
1440
get the store for a repository.
1442
from bzrlib.store.revision.text import TextRevisionStore
1443
dir_mode = control_files._dir_mode
1444
file_mode = control_files._file_mode
1445
text_store = TextStore(transport.clone(name),
1447
compressed=compressed,
1449
file_mode=file_mode)
1450
_revision_store = TextRevisionStore(text_store, serializer)
1451
return _revision_store
1453
# TODO: this shouldn't be in the base class, it's specific to things that
1454
# use weaves or knits -- mbp 20070207
1455
def _get_versioned_file_store(self,
1460
versionedfile_class=None,
1461
versionedfile_kwargs={},
1463
if versionedfile_class is None:
1464
versionedfile_class = self._versionedfile_class
1465
weave_transport = control_files._transport.clone(name)
1466
dir_mode = control_files._dir_mode
1467
file_mode = control_files._file_mode
1468
return VersionedFileStore(weave_transport, prefixed=prefixed,
1470
file_mode=file_mode,
1471
versionedfile_class=versionedfile_class,
1472
versionedfile_kwargs=versionedfile_kwargs,
1475
def initialize(self, a_bzrdir, shared=False):
1476
"""Initialize a repository of this format in a_bzrdir.
1478
:param a_bzrdir: The bzrdir to put the new repository in it.
1479
:param shared: The repository should be initialized as a sharable one.
1480
:returns: The new repository object.
1482
This may raise UninitializableFormat if shared repository are not
1483
compatible the a_bzrdir.
1485
raise NotImplementedError(self.initialize)
1487
def is_supported(self):
1488
"""Is this format supported?
1490
Supported formats must be initializable and openable.
1491
Unsupported formats may not support initialization or committing or
1492
some other features depending on the reason for not being supported.
1496
def check_conversion_target(self, target_format):
1497
raise NotImplementedError(self.check_conversion_target)
1499
def open(self, a_bzrdir, _found=False):
1500
"""Return an instance of this format for the bzrdir a_bzrdir.
1502
_found is a private parameter, do not use it.
1504
raise NotImplementedError(self.open)
1507
class MetaDirRepositoryFormat(RepositoryFormat):
1508
"""Common base class for the new repositories using the metadir layout."""
1510
rich_root_data = False
1511
supports_tree_reference = False
1512
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1515
super(MetaDirRepositoryFormat, self).__init__()
1517
def _create_control_files(self, a_bzrdir):
1518
"""Create the required files and the initial control_files object."""
1519
# FIXME: RBC 20060125 don't peek under the covers
1520
# NB: no need to escape relative paths that are url safe.
1521
repository_transport = a_bzrdir.get_repository_transport(self)
1522
control_files = lockable_files.LockableFiles(repository_transport,
1523
'lock', lockdir.LockDir)
1524
control_files.create_lock()
1525
return control_files
1527
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1528
"""Upload the initial blank content."""
1529
control_files = self._create_control_files(a_bzrdir)
1530
control_files.lock_write()
1532
control_files._transport.mkdir_multi(dirs,
1533
mode=control_files._dir_mode)
1534
for file, content in files:
1535
control_files.put(file, content)
1536
for file, content in utf8_files:
1537
control_files.put_utf8(file, content)
1539
control_files.put_utf8('shared-storage', '')
1541
control_files.unlock()
1544
# formats which have no format string are not discoverable
1545
# and not independently creatable, so are not registered. They're
1546
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1547
# needed, it's constructed directly by the BzrDir. Non-native formats where
1548
# the repository is not separately opened are similar.
1550
format_registry.register_lazy(
1551
'Bazaar-NG Repository format 7',
1552
'bzrlib.repofmt.weaverepo',
1555
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1556
# default control directory format
1558
format_registry.register_lazy(
1559
'Bazaar-NG Knit Repository Format 1',
1560
'bzrlib.repofmt.knitrepo',
1561
'RepositoryFormatKnit1',
1563
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1565
format_registry.register_lazy(
1566
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1567
'bzrlib.repofmt.knitrepo',
1568
'RepositoryFormatKnit3',
1572
class InterRepository(InterObject):
1573
"""This class represents operations taking place between two repositories.
1575
Its instances have methods like copy_content and fetch, and contain
1576
references to the source and target repositories these operations can be
1579
Often we will provide convenience methods on 'repository' which carry out
1580
operations with another repository - they will always forward to
1581
InterRepository.get(other).method_name(parameters).
1585
"""The available optimised InterRepository types."""
1587
def copy_content(self, revision_id=None):
1588
raise NotImplementedError(self.copy_content)
1590
def fetch(self, revision_id=None, pb=None):
1591
"""Fetch the content required to construct revision_id.
1593
The content is copied from self.source to self.target.
1595
:param revision_id: if None all content is copied, if NULL_REVISION no
1597
:param pb: optional progress bar to use for progress reports. If not
1598
provided a default one will be created.
1600
Returns the copied revision count and the failed revisions in a tuple:
1603
raise NotImplementedError(self.fetch)
1606
def missing_revision_ids(self, revision_id=None):
1607
"""Return the revision ids that source has that target does not.
1609
These are returned in topological order.
1611
:param revision_id: only return revision ids included by this
1614
# generic, possibly worst case, slow code path.
1615
target_ids = set(self.target.all_revision_ids())
1616
if revision_id is not None:
1617
# TODO: jam 20070210 InterRepository is internal enough that it
1618
# should assume revision_ids are already utf-8
1619
revision_id = osutils.safe_revision_id(revision_id)
1620
source_ids = self.source.get_ancestry(revision_id)
1621
assert source_ids[0] is None
1624
source_ids = self.source.all_revision_ids()
1625
result_set = set(source_ids).difference(target_ids)
1626
# this may look like a no-op: its not. It preserves the ordering
1627
# other_ids had while only returning the members from other_ids
1628
# that we've decided we need.
1629
return [rev_id for rev_id in source_ids if rev_id in result_set]
1632
class InterSameDataRepository(InterRepository):
1633
"""Code for converting between repositories that represent the same data.
1635
Data format and model must match for this to work.
1639
def _get_repo_format_to_test(self):
1640
"""Repository format for testing with."""
1641
return RepositoryFormat.get_default_format()
1644
def is_compatible(source, target):
1645
if source.supports_rich_root() != target.supports_rich_root():
1647
if source._serializer != target._serializer:
1652
def copy_content(self, revision_id=None):
1653
"""Make a complete copy of the content in self into destination.
1655
This copies both the repository's revision data, and configuration information
1656
such as the make_working_trees setting.
1658
This is a destructive operation! Do not use it on existing
1661
:param revision_id: Only copy the content needed to construct
1662
revision_id and its parents.
1665
self.target.set_make_working_trees(self.source.make_working_trees())
1666
except NotImplementedError:
1668
# TODO: jam 20070210 This is fairly internal, so we should probably
1669
# just assert that revision_id is not unicode.
1670
revision_id = osutils.safe_revision_id(revision_id)
1671
# but don't bother fetching if we have the needed data now.
1672
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1673
self.target.has_revision(revision_id)):
1675
self.target.fetch(self.source, revision_id=revision_id)
1678
def fetch(self, revision_id=None, pb=None):
1679
"""See InterRepository.fetch()."""
1680
from bzrlib.fetch import GenericRepoFetcher
1681
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1682
self.source, self.source._format, self.target,
1683
self.target._format)
1684
# TODO: jam 20070210 This should be an assert, not a translate
1685
revision_id = osutils.safe_revision_id(revision_id)
1686
f = GenericRepoFetcher(to_repository=self.target,
1687
from_repository=self.source,
1688
last_revision=revision_id,
1690
return f.count_copied, f.failed_revisions
1693
class InterWeaveRepo(InterSameDataRepository):
1694
"""Optimised code paths between Weave based repositories."""
1697
def _get_repo_format_to_test(self):
1698
from bzrlib.repofmt import weaverepo
1699
return weaverepo.RepositoryFormat7()
1702
def is_compatible(source, target):
1703
"""Be compatible with known Weave formats.
1705
We don't test for the stores being of specific types because that
1706
could lead to confusing results, and there is no need to be
1709
from bzrlib.repofmt.weaverepo import (
1715
return (isinstance(source._format, (RepositoryFormat5,
1717
RepositoryFormat7)) and
1718
isinstance(target._format, (RepositoryFormat5,
1720
RepositoryFormat7)))
1721
except AttributeError:
1725
def copy_content(self, revision_id=None):
1726
"""See InterRepository.copy_content()."""
1727
# weave specific optimised path:
1728
# TODO: jam 20070210 Internal, should be an assert, not translate
1729
revision_id = osutils.safe_revision_id(revision_id)
1731
self.target.set_make_working_trees(self.source.make_working_trees())
1732
except NotImplementedError:
1734
# FIXME do not peek!
1735
if self.source.control_files._transport.listable():
1736
pb = ui.ui_factory.nested_progress_bar()
1738
self.target.weave_store.copy_all_ids(
1739
self.source.weave_store,
1741
from_transaction=self.source.get_transaction(),
1742
to_transaction=self.target.get_transaction())
1743
pb.update('copying inventory', 0, 1)
1744
self.target.control_weaves.copy_multi(
1745
self.source.control_weaves, ['inventory'],
1746
from_transaction=self.source.get_transaction(),
1747
to_transaction=self.target.get_transaction())
1748
self.target._revision_store.text_store.copy_all_ids(
1749
self.source._revision_store.text_store,
1754
self.target.fetch(self.source, revision_id=revision_id)
1757
def fetch(self, revision_id=None, pb=None):
1758
"""See InterRepository.fetch()."""
1759
from bzrlib.fetch import GenericRepoFetcher
1760
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1761
self.source, self.source._format, self.target, self.target._format)
1762
# TODO: jam 20070210 This should be an assert, not a translate
1763
revision_id = osutils.safe_revision_id(revision_id)
1764
f = GenericRepoFetcher(to_repository=self.target,
1765
from_repository=self.source,
1766
last_revision=revision_id,
1768
return f.count_copied, f.failed_revisions
1771
def missing_revision_ids(self, revision_id=None):
1772
"""See InterRepository.missing_revision_ids()."""
1773
# we want all revisions to satisfy revision_id in source.
1774
# but we don't want to stat every file here and there.
1775
# we want then, all revisions other needs to satisfy revision_id
1776
# checked, but not those that we have locally.
1777
# so the first thing is to get a subset of the revisions to
1778
# satisfy revision_id in source, and then eliminate those that
1779
# we do already have.
1780
# this is slow on high latency connection to self, but as as this
1781
# disk format scales terribly for push anyway due to rewriting
1782
# inventory.weave, this is considered acceptable.
1784
if revision_id is not None:
1785
source_ids = self.source.get_ancestry(revision_id)
1786
assert source_ids[0] is None
1789
source_ids = self.source._all_possible_ids()
1790
source_ids_set = set(source_ids)
1791
# source_ids is the worst possible case we may need to pull.
1792
# now we want to filter source_ids against what we actually
1793
# have in target, but don't try to check for existence where we know
1794
# we do not have a revision as that would be pointless.
1795
target_ids = set(self.target._all_possible_ids())
1796
possibly_present_revisions = target_ids.intersection(source_ids_set)
1797
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1798
required_revisions = source_ids_set.difference(actually_present_revisions)
1799
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1800
if revision_id is not None:
1801
# we used get_ancestry to determine source_ids then we are assured all
1802
# revisions referenced are present as they are installed in topological order.
1803
# and the tip revision was validated by get_ancestry.
1804
return required_topo_revisions
1806
# if we just grabbed the possibly available ids, then
1807
# we only have an estimate of whats available and need to validate
1808
# that against the revision records.
1809
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1812
class InterKnitRepo(InterSameDataRepository):
1813
"""Optimised code paths between Knit based repositories."""
1816
def _get_repo_format_to_test(self):
1817
from bzrlib.repofmt import knitrepo
1818
return knitrepo.RepositoryFormatKnit1()
1821
def is_compatible(source, target):
1822
"""Be compatible with known Knit formats.
1824
We don't test for the stores being of specific types because that
1825
could lead to confusing results, and there is no need to be
1828
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1830
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1831
isinstance(target._format, (RepositoryFormatKnit1)))
1832
except AttributeError:
1836
def fetch(self, revision_id=None, pb=None):
1837
"""See InterRepository.fetch()."""
1838
from bzrlib.fetch import KnitRepoFetcher
1839
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1840
self.source, self.source._format, self.target, self.target._format)
1841
# TODO: jam 20070210 This should be an assert, not a translate
1842
revision_id = osutils.safe_revision_id(revision_id)
1843
f = KnitRepoFetcher(to_repository=self.target,
1844
from_repository=self.source,
1845
last_revision=revision_id,
1847
return f.count_copied, f.failed_revisions
1850
def missing_revision_ids(self, revision_id=None):
1851
"""See InterRepository.missing_revision_ids()."""
1852
if revision_id is not None:
1853
source_ids = self.source.get_ancestry(revision_id)
1854
assert source_ids[0] is None
1857
source_ids = self.source._all_possible_ids()
1858
source_ids_set = set(source_ids)
1859
# source_ids is the worst possible case we may need to pull.
1860
# now we want to filter source_ids against what we actually
1861
# have in target, but don't try to check for existence where we know
1862
# we do not have a revision as that would be pointless.
1863
target_ids = set(self.target._all_possible_ids())
1864
possibly_present_revisions = target_ids.intersection(source_ids_set)
1865
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1866
required_revisions = source_ids_set.difference(actually_present_revisions)
1867
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1868
if revision_id is not None:
1869
# we used get_ancestry to determine source_ids then we are assured all
1870
# revisions referenced are present as they are installed in topological order.
1871
# and the tip revision was validated by get_ancestry.
1872
return required_topo_revisions
1874
# if we just grabbed the possibly available ids, then
1875
# we only have an estimate of whats available and need to validate
1876
# that against the revision records.
1877
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1880
class InterModel1and2(InterRepository):
1883
def _get_repo_format_to_test(self):
1887
def is_compatible(source, target):
1888
if not source.supports_rich_root() and target.supports_rich_root():
1894
def fetch(self, revision_id=None, pb=None):
1895
"""See InterRepository.fetch()."""
1896
from bzrlib.fetch import Model1toKnit2Fetcher
1897
# TODO: jam 20070210 This should be an assert, not a translate
1898
revision_id = osutils.safe_revision_id(revision_id)
1899
f = Model1toKnit2Fetcher(to_repository=self.target,
1900
from_repository=self.source,
1901
last_revision=revision_id,
1903
return f.count_copied, f.failed_revisions
1906
def copy_content(self, revision_id=None):
1907
"""Make a complete copy of the content in self into destination.
1909
This is a destructive operation! Do not use it on existing
1912
:param revision_id: Only copy the content needed to construct
1913
revision_id and its parents.
1916
self.target.set_make_working_trees(self.source.make_working_trees())
1917
except NotImplementedError:
1919
# TODO: jam 20070210 Internal, assert, don't translate
1920
revision_id = osutils.safe_revision_id(revision_id)
1921
# but don't bother fetching if we have the needed data now.
1922
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1923
self.target.has_revision(revision_id)):
1925
self.target.fetch(self.source, revision_id=revision_id)
1928
class InterKnit1and2(InterKnitRepo):
1931
def _get_repo_format_to_test(self):
1935
def is_compatible(source, target):
1936
"""Be compatible with Knit1 source and Knit3 target"""
1937
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1939
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1940
RepositoryFormatKnit3
1941
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1942
isinstance(target._format, (RepositoryFormatKnit3)))
1943
except AttributeError:
1947
def fetch(self, revision_id=None, pb=None):
1948
"""See InterRepository.fetch()."""
1949
from bzrlib.fetch import Knit1to2Fetcher
1950
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1951
self.source, self.source._format, self.target,
1952
self.target._format)
1953
# TODO: jam 20070210 This should be an assert, not a translate
1954
revision_id = osutils.safe_revision_id(revision_id)
1955
f = Knit1to2Fetcher(to_repository=self.target,
1956
from_repository=self.source,
1957
last_revision=revision_id,
1959
return f.count_copied, f.failed_revisions
1962
class InterRemoteRepository(InterRepository):
1963
"""Code for converting between RemoteRepository objects.
1965
This just gets an non-remote repository from the RemoteRepository, and calls
1966
InterRepository.get again.
1969
def __init__(self, source, target):
1970
if isinstance(source, remote.RemoteRepository):
1971
source._ensure_real()
1972
real_source = source._real_repository
1974
real_source = source
1975
if isinstance(target, remote.RemoteRepository):
1976
target._ensure_real()
1977
real_target = target._real_repository
1979
real_target = target
1980
self.real_inter = InterRepository.get(real_source, real_target)
1983
def is_compatible(source, target):
1984
if isinstance(source, remote.RemoteRepository):
1986
if isinstance(target, remote.RemoteRepository):
1990
def copy_content(self, revision_id=None):
1991
self.real_inter.copy_content(revision_id=revision_id)
1993
def fetch(self, revision_id=None, pb=None):
1994
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1997
def _get_repo_format_to_test(self):
2001
InterRepository.register_optimiser(InterSameDataRepository)
2002
InterRepository.register_optimiser(InterWeaveRepo)
2003
InterRepository.register_optimiser(InterKnitRepo)
2004
InterRepository.register_optimiser(InterModel1and2)
2005
InterRepository.register_optimiser(InterKnit1and2)
2006
InterRepository.register_optimiser(InterRemoteRepository)
2009
class CopyConverter(object):
2010
"""A repository conversion tool which just performs a copy of the content.
2012
This is slow but quite reliable.
2015
def __init__(self, target_format):
2016
"""Create a CopyConverter.
2018
:param target_format: The format the resulting repository should be.
2020
self.target_format = target_format
2022
def convert(self, repo, pb):
2023
"""Perform the conversion of to_convert, giving feedback via pb.
2025
:param to_convert: The disk object to convert.
2026
:param pb: a progress bar to use for progress information.
2031
# this is only useful with metadir layouts - separated repo content.
2032
# trigger an assertion if not such
2033
repo._format.get_format_string()
2034
self.repo_dir = repo.bzrdir
2035
self.step('Moving repository to repository.backup')
2036
self.repo_dir.transport.move('repository', 'repository.backup')
2037
backup_transport = self.repo_dir.transport.clone('repository.backup')
2038
repo._format.check_conversion_target(self.target_format)
2039
self.source_repo = repo._format.open(self.repo_dir,
2041
_override_transport=backup_transport)
2042
self.step('Creating new repository')
2043
converted = self.target_format.initialize(self.repo_dir,
2044
self.source_repo.is_shared())
2045
converted.lock_write()
2047
self.step('Copying content into repository.')
2048
self.source_repo.copy_content_into(converted)
2051
self.step('Deleting old repository content.')
2052
self.repo_dir.transport.delete_tree('repository.backup')
2053
self.pb.note('repository converted')
2055
def step(self, message):
2056
"""Update the pb by a step."""
2058
self.pb.update(message, self.count, self.total)
2061
class CommitBuilder(object):
2062
"""Provides an interface to build up a commit.
2064
This allows describing a tree to be committed without needing to
2065
know the internals of the format of the repository.
2068
record_root_entry = False
2069
def __init__(self, repository, parents, config, timestamp=None,
2070
timezone=None, committer=None, revprops=None,
2072
"""Initiate a CommitBuilder.
2074
:param repository: Repository to commit to.
2075
:param parents: Revision ids of the parents of the new revision.
2076
:param config: Configuration to use.
2077
:param timestamp: Optional timestamp recorded for commit.
2078
:param timezone: Optional timezone for timestamp.
2079
:param committer: Optional committer to set for commit.
2080
:param revprops: Optional dictionary of revision properties.
2081
:param revision_id: Optional revision id.
2083
self._config = config
2085
if committer is None:
2086
self._committer = self._config.username()
2088
assert isinstance(committer, basestring), type(committer)
2089
self._committer = committer
2091
self.new_inventory = Inventory(None)
2092
self._new_revision_id = osutils.safe_revision_id(revision_id)
2093
self.parents = parents
2094
self.repository = repository
2097
if revprops is not None:
2098
self._revprops.update(revprops)
2100
if timestamp is None:
2101
timestamp = time.time()
2102
# Restrict resolution to 1ms
2103
self._timestamp = round(timestamp, 3)
2105
if timezone is None:
2106
self._timezone = osutils.local_time_offset()
2108
self._timezone = int(timezone)
2110
self._generate_revision_if_needed()
2112
def commit(self, message):
2113
"""Make the actual commit.
2115
:return: The revision id of the recorded revision.
2117
rev = _mod_revision.Revision(
2118
timestamp=self._timestamp,
2119
timezone=self._timezone,
2120
committer=self._committer,
2122
inventory_sha1=self.inv_sha1,
2123
revision_id=self._new_revision_id,
2124
properties=self._revprops)
2125
rev.parent_ids = self.parents
2126
self.repository.add_revision(self._new_revision_id, rev,
2127
self.new_inventory, self._config)
2128
self.repository.commit_write_group()
2129
return self._new_revision_id
2131
def revision_tree(self):
2132
"""Return the tree that was just committed.
2134
After calling commit() this can be called to get a RevisionTree
2135
representing the newly committed tree. This is preferred to
2136
calling Repository.revision_tree() because that may require
2137
deserializing the inventory, while we already have a copy in
2140
return RevisionTree(self.repository, self.new_inventory,
2141
self._new_revision_id)
2143
def finish_inventory(self):
2144
"""Tell the builder that the inventory is finished."""
2145
if self.new_inventory.root is None:
2146
symbol_versioning.warn('Root entry should be supplied to'
2147
' record_entry_contents, as of bzr 0.10.',
2148
DeprecationWarning, stacklevel=2)
2149
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2150
self.new_inventory.revision_id = self._new_revision_id
2151
self.inv_sha1 = self.repository.add_inventory(
2152
self._new_revision_id,
2157
def _gen_revision_id(self):
2158
"""Return new revision-id."""
2159
return generate_ids.gen_revision_id(self._config.username(),
2162
def _generate_revision_if_needed(self):
2163
"""Create a revision id if None was supplied.
2165
If the repository can not support user-specified revision ids
2166
they should override this function and raise CannotSetRevisionId
2167
if _new_revision_id is not None.
2169
:raises: CannotSetRevisionId
2171
if self._new_revision_id is None:
2172
self._new_revision_id = self._gen_revision_id()
2174
def record_entry_contents(self, ie, parent_invs, path, tree):
2175
"""Record the content of ie from tree into the commit if needed.
2177
Side effect: sets ie.revision when unchanged
2179
:param ie: An inventory entry present in the commit.
2180
:param parent_invs: The inventories of the parent revisions of the
2182
:param path: The path the entry is at in the tree.
2183
:param tree: The tree which contains this entry and should be used to
2186
if self.new_inventory.root is None and ie.parent_id is not None:
2187
symbol_versioning.warn('Root entry should be supplied to'
2188
' record_entry_contents, as of bzr 0.10.',
2189
DeprecationWarning, stacklevel=2)
2190
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2192
self.new_inventory.add(ie)
2194
# ie.revision is always None if the InventoryEntry is considered
2195
# for committing. ie.snapshot will record the correct revision
2196
# which may be the sole parent if it is untouched.
2197
if ie.revision is not None:
2200
# In this revision format, root entries have no knit or weave
2201
if ie is self.new_inventory.root:
2202
# When serializing out to disk and back in
2203
# root.revision is always _new_revision_id
2204
ie.revision = self._new_revision_id
2206
previous_entries = ie.find_previous_heads(
2208
self.repository.weave_store,
2209
self.repository.get_transaction())
2210
# we are creating a new revision for ie in the history store
2212
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2214
def modified_directory(self, file_id, file_parents):
2215
"""Record the presence of a symbolic link.
2217
:param file_id: The file_id of the link to record.
2218
:param file_parents: The per-file parent revision ids.
2220
self._add_text_to_weave(file_id, [], file_parents.keys())
2222
def modified_reference(self, file_id, file_parents):
2223
"""Record the modification of a reference.
2225
:param file_id: The file_id of the link to record.
2226
:param file_parents: The per-file parent revision ids.
2228
self._add_text_to_weave(file_id, [], file_parents.keys())
2230
def modified_file_text(self, file_id, file_parents,
2231
get_content_byte_lines, text_sha1=None,
2233
"""Record the text of file file_id
2235
:param file_id: The file_id of the file to record the text of.
2236
:param file_parents: The per-file parent revision ids.
2237
:param get_content_byte_lines: A callable which will return the byte
2239
:param text_sha1: Optional SHA1 of the file contents.
2240
:param text_size: Optional size of the file contents.
2242
# mutter('storing text of file {%s} in revision {%s} into %r',
2243
# file_id, self._new_revision_id, self.repository.weave_store)
2244
# special case to avoid diffing on renames or
2246
if (len(file_parents) == 1
2247
and text_sha1 == file_parents.values()[0].text_sha1
2248
and text_size == file_parents.values()[0].text_size):
2249
previous_ie = file_parents.values()[0]
2250
versionedfile = self.repository.weave_store.get_weave(file_id,
2251
self.repository.get_transaction())
2252
versionedfile.clone_text(self._new_revision_id,
2253
previous_ie.revision, file_parents.keys())
2254
return text_sha1, text_size
2256
new_lines = get_content_byte_lines()
2257
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2258
# should return the SHA1 and size
2259
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2260
return osutils.sha_strings(new_lines), \
2261
sum(map(len, new_lines))
2263
def modified_link(self, file_id, file_parents, link_target):
2264
"""Record the presence of a symbolic link.
2266
:param file_id: The file_id of the link to record.
2267
:param file_parents: The per-file parent revision ids.
2268
:param link_target: Target location of this link.
2270
self._add_text_to_weave(file_id, [], file_parents.keys())
2272
def _add_text_to_weave(self, file_id, new_lines, parents):
2273
versionedfile = self.repository.weave_store.get_weave_or_empty(
2274
file_id, self.repository.get_transaction())
2275
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2276
versionedfile.clear_cache()
2279
class _CommitBuilder(CommitBuilder):
2280
"""Temporary class so old CommitBuilders are detected properly
2282
Note: CommitBuilder works whether or not root entry is recorded.
2285
record_root_entry = True
2288
class RootCommitBuilder(CommitBuilder):
2289
"""This commitbuilder actually records the root id"""
2291
record_root_entry = True
2293
def record_entry_contents(self, ie, parent_invs, path, tree):
2294
"""Record the content of ie from tree into the commit if needed.
2296
Side effect: sets ie.revision when unchanged
2298
:param ie: An inventory entry present in the commit.
2299
:param parent_invs: The inventories of the parent revisions of the
2301
:param path: The path the entry is at in the tree.
2302
:param tree: The tree which contains this entry and should be used to
2305
assert self.new_inventory.root is not None or ie.parent_id is None
2306
self.new_inventory.add(ie)
2308
# ie.revision is always None if the InventoryEntry is considered
2309
# for committing. ie.snapshot will record the correct revision
2310
# which may be the sole parent if it is untouched.
2311
if ie.revision is not None:
2314
previous_entries = ie.find_previous_heads(
2316
self.repository.weave_store,
2317
self.repository.get_transaction())
2318
# we are creating a new revision for ie in the history store
2320
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2332
def _unescaper(match, _map=_unescape_map):
2333
code = match.group(1)
2337
if not code.startswith('#'):
2339
return unichr(int(code[1:])).encode('utf8')
2345
def _unescape_xml(data):
2346
"""Unescape predefined XML entities in a string of data."""
2348
if _unescape_re is None:
2349
_unescape_re = re.compile('\&([^;]*);')
2350
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
