1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
38
revision as _mod_revision,
43
from bzrlib.bundle import serializer
44
from bzrlib.revisiontree import RevisionTree
45
from bzrlib.store.versioned import VersionedFileStore
46
from bzrlib.store.text import TextStore
47
from bzrlib.testament import Testament
51
from bzrlib.decorators import needs_read_lock, needs_write_lock
52
from bzrlib.inter import InterObject
53
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
54
from bzrlib.symbol_versioning import (
58
from bzrlib.trace import mutter, note, warning
61
# Old formats display a warning, but only once
62
_deprecation_warning_done = False
65
######################################################################
68
class Repository(object):
69
"""Repository holding history for one or more branches.
71
The repository holds and retrieves historical information including
72
revisions and file history. It's normally accessed only by the Branch,
73
which views a particular line of development through that history.
75
The Repository builds on top of Stores and a Transport, which respectively
76
describe the disk data format and the way of accessing the (possibly
80
_file_ids_altered_regex = lazy_regex.lazy_compile(
81
r'file_id="(?P<file_id>[^"]+)"'
82
r'.*revision="(?P<revision_id>[^"]+)"'
85
def abort_write_group(self):
86
"""Commit the contents accrued within the current write group.
88
:seealso: start_write_group.
90
if self._write_group is not self.get_transaction():
91
# has an unlock or relock occured ?
92
raise errors.BzrError('mismatched lock context and write group.')
93
self._abort_write_group()
94
self._write_group = None
96
def _abort_write_group(self):
97
"""Template method for per-repository write group cleanup.
99
This is called during abort before the write group is considered to be
100
finished and should cleanup any internal state accrued during the write
101
group. There is no requirement that data handed to the repository be
102
*not* made available - this is not a rollback - but neither should any
103
attempt be made to ensure that data added is fully commited. Abort is
104
invoked when an error has occured so futher disk or network operations
105
may not be possible or may error and if possible should not be
110
def add_inventory(self, revision_id, inv, parents):
111
"""Add the inventory inv to the repository as revision_id.
113
:param parents: The revision ids of the parents that revision_id
114
is known to have and are in the repository already.
116
returns the sha1 of the serialized inventory.
118
revision_id = osutils.safe_revision_id(revision_id)
119
_mod_revision.check_not_reserved_id(revision_id)
120
assert inv.revision_id is None or inv.revision_id == revision_id, \
121
"Mismatch between inventory revision" \
122
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
123
assert inv.root is not None
124
inv_text = self.serialise_inventory(inv)
125
inv_sha1 = osutils.sha_string(inv_text)
126
inv_vf = self.control_weaves.get_weave('inventory',
127
self.get_transaction())
128
self._inventory_add_lines(inv_vf, revision_id, parents,
129
osutils.split_lines(inv_text))
132
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
134
for parent in parents:
136
final_parents.append(parent)
138
inv_vf.add_lines(revision_id, final_parents, lines)
141
def add_revision(self, revision_id, rev, inv=None, config=None):
142
"""Add rev to the revision store as revision_id.
144
:param revision_id: the revision id to use.
145
:param rev: The revision object.
146
:param inv: The inventory for the revision. if None, it will be looked
147
up in the inventory storer
148
:param config: If None no digital signature will be created.
149
If supplied its signature_needed method will be used
150
to determine if a signature should be made.
152
revision_id = osutils.safe_revision_id(revision_id)
153
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
155
_mod_revision.check_not_reserved_id(revision_id)
156
if config is not None and config.signature_needed():
158
inv = self.get_inventory(revision_id)
159
plaintext = Testament(rev, inv).as_short_text()
160
self.store_revision_signature(
161
gpg.GPGStrategy(config), plaintext, revision_id)
162
if not revision_id in self.get_inventory_weave():
164
raise errors.WeaveRevisionNotPresent(revision_id,
165
self.get_inventory_weave())
167
# yes, this is not suitable for adding with ghosts.
168
self.add_inventory(revision_id, inv, rev.parent_ids)
169
self._revision_store.add_revision(rev, self.get_transaction())
171
def _add_revision_text(self, revision_id, text):
172
revision = self._revision_store._serializer.read_revision_from_string(
174
self._revision_store._add_revision(revision, StringIO(text),
175
self.get_transaction())
178
def _all_possible_ids(self):
179
"""Return all the possible revisions that we could find."""
180
return self.get_inventory_weave().versions()
182
def all_revision_ids(self):
183
"""Returns a list of all the revision ids in the repository.
185
This is deprecated because code should generally work on the graph
186
reachable from a particular revision, and ignore any other revisions
187
that might be present. There is no direct replacement method.
189
return self._all_revision_ids()
192
def _all_revision_ids(self):
193
"""Returns a list of all the revision ids in the repository.
195
These are in as much topological order as the underlying store can
196
present: for weaves ghosts may lead to a lack of correctness until
197
the reweave updates the parents list.
199
if self._revision_store.text_store.listable():
200
return self._revision_store.all_revision_ids(self.get_transaction())
201
result = self._all_possible_ids()
202
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
203
# ids. (It should, since _revision_store's API should change to
204
# return utf8 revision_ids)
205
return self._eliminate_revisions_not_present(result)
207
def break_lock(self):
208
"""Break a lock if one is present from another instance.
210
Uses the ui factory to ask for confirmation if the lock may be from
213
self.control_files.break_lock()
216
def _eliminate_revisions_not_present(self, revision_ids):
217
"""Check every revision id in revision_ids to see if we have it.
219
Returns a set of the present revisions.
222
for id in revision_ids:
223
if self.has_revision(id):
228
def create(a_bzrdir):
229
"""Construct the current default format repository in a_bzrdir."""
230
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
232
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
233
"""instantiate a Repository.
235
:param _format: The format of the repository on disk.
236
:param a_bzrdir: The BzrDir of the repository.
238
In the future we will have a single api for all stores for
239
getting file texts, inventories and revisions, then
240
this construct will accept instances of those things.
242
super(Repository, self).__init__()
243
self._format = _format
244
# the following are part of the public API for Repository:
245
self.bzrdir = a_bzrdir
246
self.control_files = control_files
247
self._revision_store = _revision_store
248
# backwards compatibility
249
self.weave_store = text_store
250
# not right yet - should be more semantically clear ?
252
self.control_store = control_store
253
self.control_weaves = control_store
254
# TODO: make sure to construct the right store classes, etc, depending
255
# on whether escaping is required.
256
self._warn_if_deprecated()
257
self._write_group = None
260
return '%s(%r)' % (self.__class__.__name__,
261
self.bzrdir.transport.base)
263
def has_same_location(self, other):
264
"""Returns a boolean indicating if this repository is at the same
265
location as another repository.
267
This might return False even when two repository objects are accessing
268
the same physical repository via different URLs.
270
if self.__class__ is not other.__class__:
272
return (self.control_files._transport.base ==
273
other.control_files._transport.base)
275
def is_in_write_group(self):
276
"""Return True if there is an open write group.
278
:seealso: start_write_group.
280
return self._write_group is not None
283
return self.control_files.is_locked()
285
def lock_write(self, token=None):
286
"""Lock this repository for writing.
288
This causes caching within the repository obejct to start accumlating
289
data during reads, and allows a 'write_group' to be obtained. Write
290
groups must be used for actual data insertion.
292
:param token: if this is already locked, then lock_write will fail
293
unless the token matches the existing lock.
294
:returns: a token if this instance supports tokens, otherwise None.
295
:raises TokenLockingNotSupported: when a token is given but this
296
instance doesn't support using token locks.
297
:raises MismatchedToken: if the specified token doesn't match the token
298
of the existing lock.
299
:seealso: start_write_group.
301
A token should be passed in if you know that you have locked the object
302
some other way, and need to synchronise this object's state with that
305
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
307
result = self.control_files.lock_write(token=token)
312
self.control_files.lock_read()
315
def get_physical_lock_status(self):
316
return self.control_files.get_physical_lock_status()
318
def leave_lock_in_place(self):
319
"""Tell this repository not to release the physical lock when this
322
If lock_write doesn't return a token, then this method is not supported.
324
self.control_files.leave_in_place()
326
def dont_leave_lock_in_place(self):
327
"""Tell this repository to release the physical lock when this
328
object is unlocked, even if it didn't originally acquire it.
330
If lock_write doesn't return a token, then this method is not supported.
332
self.control_files.dont_leave_in_place()
335
def gather_stats(self, revid=None, committers=None):
336
"""Gather statistics from a revision id.
338
:param revid: The revision id to gather statistics from, if None, then
339
no revision specific statistics are gathered.
340
:param committers: Optional parameter controlling whether to grab
341
a count of committers from the revision specific statistics.
342
:return: A dictionary of statistics. Currently this contains:
343
committers: The number of committers if requested.
344
firstrev: A tuple with timestamp, timezone for the penultimate left
345
most ancestor of revid, if revid is not the NULL_REVISION.
346
latestrev: A tuple with timestamp, timezone for revid, if revid is
347
not the NULL_REVISION.
348
revisions: The total revision count in the repository.
349
size: An estimate disk size of the repository in bytes.
352
if revid and committers:
353
result['committers'] = 0
354
if revid and revid != _mod_revision.NULL_REVISION:
356
all_committers = set()
357
revisions = self.get_ancestry(revid)
358
# pop the leading None
360
first_revision = None
362
# ignore the revisions in the middle - just grab first and last
363
revisions = revisions[0], revisions[-1]
364
for revision in self.get_revisions(revisions):
365
if not first_revision:
366
first_revision = revision
368
all_committers.add(revision.committer)
369
last_revision = revision
371
result['committers'] = len(all_committers)
372
result['firstrev'] = (first_revision.timestamp,
373
first_revision.timezone)
374
result['latestrev'] = (last_revision.timestamp,
375
last_revision.timezone)
377
# now gather global repository information
378
if self.bzrdir.root_transport.listable():
379
c, t = self._revision_store.total_size(self.get_transaction())
380
result['revisions'] = c
385
def missing_revision_ids(self, other, revision_id=None):
386
"""Return the revision ids that other has that this does not.
388
These are returned in topological order.
390
revision_id: only return revision ids included by revision_id.
392
revision_id = osutils.safe_revision_id(revision_id)
393
return InterRepository.get(other, self).missing_revision_ids(revision_id)
397
"""Open the repository rooted at base.
399
For instance, if the repository is at URL/.bzr/repository,
400
Repository.open(URL) -> a Repository instance.
402
control = bzrdir.BzrDir.open(base)
403
return control.open_repository()
405
def copy_content_into(self, destination, revision_id=None):
406
"""Make a complete copy of the content in self into destination.
408
This is a destructive operation! Do not use it on existing
411
revision_id = osutils.safe_revision_id(revision_id)
412
return InterRepository.get(self, destination).copy_content(revision_id)
414
def commit_write_group(self):
415
"""Commit the contents accrued within the current write group.
417
:seealso: start_write_group.
419
if self._write_group is not self.get_transaction():
420
# has an unlock or relock occured ?
421
raise errors.BzrError('mismatched lock context and write group.')
422
self._commit_write_group()
423
self._write_group = None
425
def _commit_write_group(self):
426
"""Template method for per-repository write group cleanup.
428
This is called before the write group is considered to be
429
finished and should ensure that all data handed to the repository
430
for writing during the write group is safely committed (to the
431
extent possible considering file system caching etc).
434
def fetch(self, source, revision_id=None, pb=None):
435
"""Fetch the content required to construct revision_id from source.
437
If revision_id is None all content is copied.
439
revision_id = osutils.safe_revision_id(revision_id)
440
inter = InterRepository.get(source, self)
442
return inter.fetch(revision_id=revision_id, pb=pb)
443
except NotImplementedError:
444
raise errors.IncompatibleRepositories(source, self)
446
def create_bundle(self, target, base, fileobj, format=None):
447
return serializer.write_bundle(self, target, base, fileobj, format)
449
def get_commit_builder(self, branch, parents, config, timestamp=None,
450
timezone=None, committer=None, revprops=None,
452
"""Obtain a CommitBuilder for this repository.
454
:param branch: Branch to commit to.
455
:param parents: Revision ids of the parents of the new revision.
456
:param config: Configuration to use.
457
:param timestamp: Optional timestamp recorded for commit.
458
:param timezone: Optional timezone for timestamp.
459
:param committer: Optional committer to set for commit.
460
:param revprops: Optional dictionary of revision properties.
461
:param revision_id: Optional revision id.
463
revision_id = osutils.safe_revision_id(revision_id)
464
result =_CommitBuilder(self, parents, config, timestamp, timezone,
465
committer, revprops, revision_id)
466
self.start_write_group()
470
if (self.control_files._lock_count == 1 and
471
self.control_files._lock_mode == 'w'):
472
if self._write_group is not None:
473
raise errors.BzrError(
474
'Must end write groups before releasing write locks.')
475
self.control_files.unlock()
478
def clone(self, a_bzrdir, revision_id=None):
479
"""Clone this repository into a_bzrdir using the current format.
481
Currently no check is made that the format of this repository and
482
the bzrdir format are compatible. FIXME RBC 20060201.
484
:return: The newly created destination repository.
486
# TODO: deprecate after 0.16; cloning this with all its settings is
487
# probably not very useful -- mbp 20070423
488
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
489
self.copy_content_into(dest_repo, revision_id)
492
def start_write_group(self):
493
"""Start a write group in the repository.
495
Write groups are used by repositories which do not have a 1:1 mapping
496
between file ids and backend store to manage the insertion of data from
497
both fetch and commit operations.
499
A write lock is required around the start_write_group/commit_write_group
500
for the support of lock-requiring repository formats.
502
One can only insert data into a repository inside a write group.
506
if not self.is_locked() or self.control_files._lock_mode != 'w':
507
raise errors.NotWriteLocked(self)
508
if self._write_group:
509
raise errors.BzrError('already in a write group')
510
self._start_write_group()
511
# so we can detect unlock/relock - the write group is now entered.
512
self._write_group = self.get_transaction()
514
def _start_write_group(self):
515
"""Template method for per-repository write group startup.
517
This is called before the write group is considered to be
522
def sprout(self, to_bzrdir, revision_id=None):
523
"""Create a descendent repository for new development.
525
Unlike clone, this does not copy the settings of the repository.
527
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
528
dest_repo.fetch(self, revision_id=revision_id)
531
def _create_sprouting_repo(self, a_bzrdir, shared):
532
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
533
# use target default format.
534
dest_repo = a_bzrdir.create_repository()
536
# Most control formats need the repository to be specifically
537
# created, but on some old all-in-one formats it's not needed
539
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
540
except errors.UninitializableFormat:
541
dest_repo = a_bzrdir.open_repository()
545
def has_revision(self, revision_id):
546
"""True if this repository has a copy of the revision."""
547
revision_id = osutils.safe_revision_id(revision_id)
548
return self._revision_store.has_revision_id(revision_id,
549
self.get_transaction())
552
def get_revision_reconcile(self, revision_id):
553
"""'reconcile' helper routine that allows access to a revision always.
555
This variant of get_revision does not cross check the weave graph
556
against the revision one as get_revision does: but it should only
557
be used by reconcile, or reconcile-alike commands that are correcting
558
or testing the revision graph.
560
if not revision_id or not isinstance(revision_id, basestring):
561
raise errors.InvalidRevisionId(revision_id=revision_id,
563
return self.get_revisions([revision_id])[0]
566
def get_revisions(self, revision_ids):
567
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
568
revs = self._revision_store.get_revisions(revision_ids,
569
self.get_transaction())
571
assert not isinstance(rev.revision_id, unicode)
572
for parent_id in rev.parent_ids:
573
assert not isinstance(parent_id, unicode)
577
def get_revision_xml(self, revision_id):
578
# TODO: jam 20070210 This shouldn't be necessary since get_revision
579
# would have already do it.
580
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
581
revision_id = osutils.safe_revision_id(revision_id)
582
rev = self.get_revision(revision_id)
584
# the current serializer..
585
self._revision_store._serializer.write_revision(rev, rev_tmp)
587
return rev_tmp.getvalue()
590
def get_revision(self, revision_id):
591
"""Return the Revision object for a named revision"""
592
# TODO: jam 20070210 get_revision_reconcile should do this for us
593
revision_id = osutils.safe_revision_id(revision_id)
594
r = self.get_revision_reconcile(revision_id)
595
# weave corruption can lead to absent revision markers that should be
597
# the following test is reasonably cheap (it needs a single weave read)
598
# and the weave is cached in read transactions. In write transactions
599
# it is not cached but typically we only read a small number of
600
# revisions. For knits when they are introduced we will probably want
601
# to ensure that caching write transactions are in use.
602
inv = self.get_inventory_weave()
603
self._check_revision_parents(r, inv)
607
def get_deltas_for_revisions(self, revisions):
608
"""Produce a generator of revision deltas.
610
Note that the input is a sequence of REVISIONS, not revision_ids.
611
Trees will be held in memory until the generator exits.
612
Each delta is relative to the revision's lefthand predecessor.
614
required_trees = set()
615
for revision in revisions:
616
required_trees.add(revision.revision_id)
617
required_trees.update(revision.parent_ids[:1])
618
trees = dict((t.get_revision_id(), t) for
619
t in self.revision_trees(required_trees))
620
for revision in revisions:
621
if not revision.parent_ids:
622
old_tree = self.revision_tree(None)
624
old_tree = trees[revision.parent_ids[0]]
625
yield trees[revision.revision_id].changes_from(old_tree)
628
def get_revision_delta(self, revision_id):
629
"""Return the delta for one revision.
631
The delta is relative to the left-hand predecessor of the
634
r = self.get_revision(revision_id)
635
return list(self.get_deltas_for_revisions([r]))[0]
637
def _check_revision_parents(self, revision, inventory):
638
"""Private to Repository and Fetch.
640
This checks the parentage of revision in an inventory weave for
641
consistency and is only applicable to inventory-weave-for-ancestry
642
using repository formats & fetchers.
644
weave_parents = inventory.get_parents(revision.revision_id)
645
weave_names = inventory.versions()
646
for parent_id in revision.parent_ids:
647
if parent_id in weave_names:
648
# this parent must not be a ghost.
649
if not parent_id in weave_parents:
651
raise errors.CorruptRepository(self)
654
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
655
revision_id = osutils.safe_revision_id(revision_id)
656
signature = gpg_strategy.sign(plaintext)
657
self._revision_store.add_revision_signature_text(revision_id,
659
self.get_transaction())
661
def fileids_altered_by_revision_ids(self, revision_ids):
662
"""Find the file ids and versions affected by revisions.
664
:param revisions: an iterable containing revision ids.
665
:return: a dictionary mapping altered file-ids to an iterable of
666
revision_ids. Each altered file-ids has the exact revision_ids that
667
altered it listed explicitly.
669
assert self._serializer.support_altered_by_hack, \
670
("fileids_altered_by_revision_ids only supported for branches "
671
"which store inventory as unnested xml, not on %r" % self)
672
selected_revision_ids = set(osutils.safe_revision_id(r)
673
for r in revision_ids)
674
w = self.get_inventory_weave()
677
# this code needs to read every new line in every inventory for the
678
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
679
# not present in one of those inventories is unnecessary but not
680
# harmful because we are filtering by the revision id marker in the
681
# inventory lines : we only select file ids altered in one of those
682
# revisions. We don't need to see all lines in the inventory because
683
# only those added in an inventory in rev X can contain a revision=X
685
unescape_revid_cache = {}
686
unescape_fileid_cache = {}
688
# jam 20061218 In a big fetch, this handles hundreds of thousands
689
# of lines, so it has had a lot of inlining and optimizing done.
690
# Sorry that it is a little bit messy.
691
# Move several functions to be local variables, since this is a long
693
search = self._file_ids_altered_regex.search
694
unescape = _unescape_xml
695
setdefault = result.setdefault
696
pb = ui.ui_factory.nested_progress_bar()
698
for line in w.iter_lines_added_or_present_in_versions(
699
selected_revision_ids, pb=pb):
703
# One call to match.group() returning multiple items is quite a
704
# bit faster than 2 calls to match.group() each returning 1
705
file_id, revision_id = match.group('file_id', 'revision_id')
707
# Inlining the cache lookups helps a lot when you make 170,000
708
# lines and 350k ids, versus 8.4 unique ids.
709
# Using a cache helps in 2 ways:
710
# 1) Avoids unnecessary decoding calls
711
# 2) Re-uses cached strings, which helps in future set and
713
# (2) is enough that removing encoding entirely along with
714
# the cache (so we are using plain strings) results in no
715
# performance improvement.
717
revision_id = unescape_revid_cache[revision_id]
719
unescaped = unescape(revision_id)
720
unescape_revid_cache[revision_id] = unescaped
721
revision_id = unescaped
723
if revision_id in selected_revision_ids:
725
file_id = unescape_fileid_cache[file_id]
727
unescaped = unescape(file_id)
728
unescape_fileid_cache[file_id] = unescaped
730
setdefault(file_id, set()).add(revision_id)
736
def get_inventory_weave(self):
737
return self.control_weaves.get_weave('inventory',
738
self.get_transaction())
741
def get_inventory(self, revision_id):
742
"""Get Inventory object by hash."""
743
# TODO: jam 20070210 Technically we don't need to sanitize, since all
744
# called functions must sanitize.
745
revision_id = osutils.safe_revision_id(revision_id)
746
return self.deserialise_inventory(
747
revision_id, self.get_inventory_xml(revision_id))
749
def deserialise_inventory(self, revision_id, xml):
750
"""Transform the xml into an inventory object.
752
:param revision_id: The expected revision id of the inventory.
753
:param xml: A serialised inventory.
755
revision_id = osutils.safe_revision_id(revision_id)
756
result = self._serializer.read_inventory_from_string(xml)
757
result.root.revision = revision_id
760
def serialise_inventory(self, inv):
761
return self._serializer.write_inventory_to_string(inv)
763
def get_serializer_format(self):
764
return self._serializer.format_num
767
def get_inventory_xml(self, revision_id):
768
"""Get inventory XML as a file object."""
769
revision_id = osutils.safe_revision_id(revision_id)
771
assert isinstance(revision_id, str), type(revision_id)
772
iw = self.get_inventory_weave()
773
return iw.get_text(revision_id)
775
raise errors.HistoryMissing(self, 'inventory', revision_id)
778
def get_inventory_sha1(self, revision_id):
779
"""Return the sha1 hash of the inventory entry
781
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
782
revision_id = osutils.safe_revision_id(revision_id)
783
return self.get_revision(revision_id).inventory_sha1
786
def get_revision_graph(self, revision_id=None):
787
"""Return a dictionary containing the revision graph.
789
:param revision_id: The revision_id to get a graph from. If None, then
790
the entire revision graph is returned. This is a deprecated mode of
791
operation and will be removed in the future.
792
:return: a dictionary of revision_id->revision_parents_list.
794
# special case NULL_REVISION
795
if revision_id == _mod_revision.NULL_REVISION:
797
revision_id = osutils.safe_revision_id(revision_id)
798
a_weave = self.get_inventory_weave()
799
all_revisions = self._eliminate_revisions_not_present(
801
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
802
node in all_revisions])
803
if revision_id is None:
805
elif revision_id not in entire_graph:
806
raise errors.NoSuchRevision(self, revision_id)
808
# add what can be reached from revision_id
810
pending = set([revision_id])
811
while len(pending) > 0:
813
result[node] = entire_graph[node]
814
for revision_id in result[node]:
815
if revision_id not in result:
816
pending.add(revision_id)
820
def get_revision_graph_with_ghosts(self, revision_ids=None):
821
"""Return a graph of the revisions with ghosts marked as applicable.
823
:param revision_ids: an iterable of revisions to graph or None for all.
824
:return: a Graph object with the graph reachable from revision_ids.
826
result = deprecated_graph.Graph()
828
pending = set(self.all_revision_ids())
831
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
832
# special case NULL_REVISION
833
if _mod_revision.NULL_REVISION in pending:
834
pending.remove(_mod_revision.NULL_REVISION)
835
required = set(pending)
838
revision_id = pending.pop()
840
rev = self.get_revision(revision_id)
841
except errors.NoSuchRevision:
842
if revision_id in required:
845
result.add_ghost(revision_id)
847
for parent_id in rev.parent_ids:
848
# is this queued or done ?
849
if (parent_id not in pending and
850
parent_id not in done):
852
pending.add(parent_id)
853
result.add_node(revision_id, rev.parent_ids)
854
done.add(revision_id)
857
def _get_history_vf(self):
858
"""Get a versionedfile whose history graph reflects all revisions.
860
For weave repositories, this is the inventory weave.
862
return self.get_inventory_weave()
864
def iter_reverse_revision_history(self, revision_id):
865
"""Iterate backwards through revision ids in the lefthand history
867
:param revision_id: The revision id to start with. All its lefthand
868
ancestors will be traversed.
870
revision_id = osutils.safe_revision_id(revision_id)
871
if revision_id in (None, _mod_revision.NULL_REVISION):
873
next_id = revision_id
874
versionedfile = self._get_history_vf()
877
parents = versionedfile.get_parents(next_id)
878
if len(parents) == 0:
884
def get_revision_inventory(self, revision_id):
885
"""Return inventory of a past revision."""
886
# TODO: Unify this with get_inventory()
887
# bzr 0.0.6 and later imposes the constraint that the inventory_id
888
# must be the same as its revision, so this is trivial.
889
if revision_id is None:
890
# This does not make sense: if there is no revision,
891
# then it is the current tree inventory surely ?!
892
# and thus get_root_id() is something that looks at the last
893
# commit on the branch, and the get_root_id is an inventory check.
894
raise NotImplementedError
895
# return Inventory(self.get_root_id())
897
return self.get_inventory(revision_id)
901
"""Return True if this repository is flagged as a shared repository."""
902
raise NotImplementedError(self.is_shared)
905
def reconcile(self, other=None, thorough=False):
906
"""Reconcile this repository."""
907
from bzrlib.reconcile import RepoReconciler
908
reconciler = RepoReconciler(self, thorough=thorough)
909
reconciler.reconcile()
912
def _refresh_data(self):
913
"""Helper called from lock_* to ensure coherency with disk.
915
The default implementation does nothing; it is however possible
916
for repositories to maintain loaded indices across multiple locks
917
by checking inside their implementation of this method to see
918
whether their indices are still valid. This depends of course on
919
the disk format being validatable in this manner.
923
def revision_tree(self, revision_id):
924
"""Return Tree for a revision on this branch.
926
`revision_id` may be None for the empty tree revision.
928
# TODO: refactor this to use an existing revision object
929
# so we don't need to read it in twice.
930
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
931
return RevisionTree(self, Inventory(root_id=None),
932
_mod_revision.NULL_REVISION)
934
revision_id = osutils.safe_revision_id(revision_id)
935
inv = self.get_revision_inventory(revision_id)
936
return RevisionTree(self, inv, revision_id)
939
def revision_trees(self, revision_ids):
940
"""Return Tree for a revision on this branch.
942
`revision_id` may not be None or 'null:'"""
943
assert None not in revision_ids
944
assert _mod_revision.NULL_REVISION not in revision_ids
945
texts = self.get_inventory_weave().get_texts(revision_ids)
946
for text, revision_id in zip(texts, revision_ids):
947
inv = self.deserialise_inventory(revision_id, text)
948
yield RevisionTree(self, inv, revision_id)
951
def get_ancestry(self, revision_id, topo_sorted=True):
952
"""Return a list of revision-ids integrated by a revision.
954
The first element of the list is always None, indicating the origin
955
revision. This might change when we have history horizons, or
956
perhaps we should have a new API.
958
This is topologically sorted.
960
if _mod_revision.is_null(revision_id):
962
revision_id = osutils.safe_revision_id(revision_id)
963
if not self.has_revision(revision_id):
964
raise errors.NoSuchRevision(self, revision_id)
965
w = self.get_inventory_weave()
966
candidates = w.get_ancestry(revision_id, topo_sorted)
967
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
970
"""Compress the data within the repository.
972
This operation only makes sense for some repository types. For other
973
types it should be a no-op that just returns.
975
This stub method does not require a lock, but subclasses should use
976
@needs_write_lock as this is a long running call its reasonable to
977
implicitly lock for the user.
981
def print_file(self, file, revision_id):
982
"""Print `file` to stdout.
984
FIXME RBC 20060125 as John Meinel points out this is a bad api
985
- it writes to stdout, it assumes that that is valid etc. Fix
986
by creating a new more flexible convenience function.
988
revision_id = osutils.safe_revision_id(revision_id)
989
tree = self.revision_tree(revision_id)
990
# use inventory as it was in that revision
991
file_id = tree.inventory.path2id(file)
993
# TODO: jam 20060427 Write a test for this code path
994
# it had a bug in it, and was raising the wrong
996
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
997
tree.print_file(file_id)
999
def get_transaction(self):
1000
return self.control_files.get_transaction()
1002
def revision_parents(self, revision_id):
1003
revision_id = osutils.safe_revision_id(revision_id)
1004
return self.get_inventory_weave().parent_names(revision_id)
1006
def get_parents(self, revision_ids):
1007
"""See StackedParentsProvider.get_parents"""
1009
for revision_id in revision_ids:
1010
if revision_id == _mod_revision.NULL_REVISION:
1014
parents = self.get_revision(revision_id).parent_ids
1015
except errors.NoSuchRevision:
1018
if len(parents) == 0:
1019
parents = [_mod_revision.NULL_REVISION]
1020
parents_list.append(parents)
1023
def _make_parents_provider(self):
1026
def get_graph(self, other_repository=None):
1027
"""Return the graph walker for this repository format"""
1028
parents_provider = self._make_parents_provider()
1029
if (other_repository is not None and
1030
other_repository.bzrdir.transport.base !=
1031
self.bzrdir.transport.base):
1032
parents_provider = graph._StackedParentsProvider(
1033
[parents_provider, other_repository._make_parents_provider()])
1034
return graph.Graph(parents_provider)
1037
def set_make_working_trees(self, new_value):
1038
"""Set the policy flag for making working trees when creating branches.
1040
This only applies to branches that use this repository.
1042
The default is 'True'.
1043
:param new_value: True to restore the default, False to disable making
1046
raise NotImplementedError(self.set_make_working_trees)
1048
def make_working_trees(self):
1049
"""Returns the policy for making working trees on new branches."""
1050
raise NotImplementedError(self.make_working_trees)
1053
def sign_revision(self, revision_id, gpg_strategy):
1054
revision_id = osutils.safe_revision_id(revision_id)
1055
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1056
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1059
def has_signature_for_revision_id(self, revision_id):
1060
"""Query for a revision signature for revision_id in the repository."""
1061
revision_id = osutils.safe_revision_id(revision_id)
1062
return self._revision_store.has_signature(revision_id,
1063
self.get_transaction())
1066
def get_signature_text(self, revision_id):
1067
"""Return the text for a signature."""
1068
revision_id = osutils.safe_revision_id(revision_id)
1069
return self._revision_store.get_signature_text(revision_id,
1070
self.get_transaction())
1073
def check(self, revision_ids):
1074
"""Check consistency of all history of given revision_ids.
1076
Different repository implementations should override _check().
1078
:param revision_ids: A non-empty list of revision_ids whose ancestry
1079
will be checked. Typically the last revision_id of a branch.
1081
if not revision_ids:
1082
raise ValueError("revision_ids must be non-empty in %s.check"
1084
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1085
return self._check(revision_ids)
1087
def _check(self, revision_ids):
1088
result = check.Check(self)
1092
def _warn_if_deprecated(self):
1093
global _deprecation_warning_done
1094
if _deprecation_warning_done:
1096
_deprecation_warning_done = True
1097
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1098
% (self._format, self.bzrdir.transport.base))
1100
def supports_rich_root(self):
1101
return self._format.rich_root_data
1103
def _check_ascii_revisionid(self, revision_id, method):
1104
"""Private helper for ascii-only repositories."""
1105
# weave repositories refuse to store revisionids that are non-ascii.
1106
if revision_id is not None:
1107
# weaves require ascii revision ids.
1108
if isinstance(revision_id, unicode):
1110
revision_id.encode('ascii')
1111
except UnicodeEncodeError:
1112
raise errors.NonAsciiRevisionId(method, self)
1115
revision_id.decode('ascii')
1116
except UnicodeDecodeError:
1117
raise errors.NonAsciiRevisionId(method, self)
1121
# remove these delegates a while after bzr 0.15
1122
def __make_delegated(name, from_module):
1123
def _deprecated_repository_forwarder():
1124
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1125
% (name, from_module),
1128
m = __import__(from_module, globals(), locals(), [name])
1130
return getattr(m, name)
1131
except AttributeError:
1132
raise AttributeError('module %s has no name %s'
1134
globals()[name] = _deprecated_repository_forwarder
1137
'AllInOneRepository',
1138
'WeaveMetaDirRepository',
1139
'PreSplitOutRepositoryFormat',
1140
'RepositoryFormat4',
1141
'RepositoryFormat5',
1142
'RepositoryFormat6',
1143
'RepositoryFormat7',
1145
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1149
'RepositoryFormatKnit',
1150
'RepositoryFormatKnit1',
1152
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1155
def install_revision(repository, rev, revision_tree):
1156
"""Install all revision data into a repository."""
1157
present_parents = []
1159
for p_id in rev.parent_ids:
1160
if repository.has_revision(p_id):
1161
present_parents.append(p_id)
1162
parent_trees[p_id] = repository.revision_tree(p_id)
1164
parent_trees[p_id] = repository.revision_tree(None)
1166
inv = revision_tree.inventory
1167
entries = inv.iter_entries()
1168
# backwards compatibility hack: skip the root id.
1169
if not repository.supports_rich_root():
1170
path, root = entries.next()
1171
if root.revision != rev.revision_id:
1172
raise errors.IncompatibleRevision(repr(repository))
1173
# Add the texts that are not already present
1174
for path, ie in entries:
1175
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1176
repository.get_transaction())
1177
if ie.revision not in w:
1179
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1180
# with InventoryEntry.find_previous_heads(). if it is, then there
1181
# is a latent bug here where the parents may have ancestors of each
1183
for revision, tree in parent_trees.iteritems():
1184
if ie.file_id not in tree:
1186
parent_id = tree.inventory[ie.file_id].revision
1187
if parent_id in text_parents:
1189
text_parents.append(parent_id)
1191
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1192
repository.get_transaction())
1193
lines = revision_tree.get_file(ie.file_id).readlines()
1194
vfile.add_lines(rev.revision_id, text_parents, lines)
1196
# install the inventory
1197
repository.add_inventory(rev.revision_id, inv, present_parents)
1198
except errors.RevisionAlreadyPresent:
1200
repository.add_revision(rev.revision_id, rev, inv)
1203
class MetaDirRepository(Repository):
1204
"""Repositories in the new meta-dir layout."""
1206
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1207
super(MetaDirRepository, self).__init__(_format,
1213
dir_mode = self.control_files._dir_mode
1214
file_mode = self.control_files._file_mode
1217
def is_shared(self):
1218
"""Return True if this repository is flagged as a shared repository."""
1219
return self.control_files._transport.has('shared-storage')
1222
def set_make_working_trees(self, new_value):
1223
"""Set the policy flag for making working trees when creating branches.
1225
This only applies to branches that use this repository.
1227
The default is 'True'.
1228
:param new_value: True to restore the default, False to disable making
1233
self.control_files._transport.delete('no-working-trees')
1234
except errors.NoSuchFile:
1237
self.control_files.put_utf8('no-working-trees', '')
1239
def make_working_trees(self):
1240
"""Returns the policy for making working trees on new branches."""
1241
return not self.control_files._transport.has('no-working-trees')
1244
class RepositoryFormatRegistry(registry.Registry):
1245
"""Registry of RepositoryFormats.
1248
def get(self, format_string):
1249
r = registry.Registry.get(self, format_string)
1255
format_registry = RepositoryFormatRegistry()
1256
"""Registry of formats, indexed by their identifying format string.
1258
This can contain either format instances themselves, or classes/factories that
1259
can be called to obtain one.
1263
#####################################################################
1264
# Repository Formats
1266
class RepositoryFormat(object):
1267
"""A repository format.
1269
Formats provide three things:
1270
* An initialization routine to construct repository data on disk.
1271
* a format string which is used when the BzrDir supports versioned
1273
* an open routine which returns a Repository instance.
1275
Formats are placed in an dict by their format string for reference
1276
during opening. These should be subclasses of RepositoryFormat
1279
Once a format is deprecated, just deprecate the initialize and open
1280
methods on the format class. Do not deprecate the object, as the
1281
object will be created every system load.
1283
Common instance attributes:
1284
_matchingbzrdir - the bzrdir format that the repository format was
1285
originally written to work with. This can be used if manually
1286
constructing a bzrdir and repository, or more commonly for test suite
1291
return "<%s>" % self.__class__.__name__
1293
def __eq__(self, other):
1294
# format objects are generally stateless
1295
return isinstance(other, self.__class__)
1297
def __ne__(self, other):
1298
return not self == other
1301
def find_format(klass, a_bzrdir):
1302
"""Return the format for the repository object in a_bzrdir.
1304
This is used by bzr native formats that have a "format" file in
1305
the repository. Other methods may be used by different types of
1309
transport = a_bzrdir.get_repository_transport(None)
1310
format_string = transport.get("format").read()
1311
return format_registry.get(format_string)
1312
except errors.NoSuchFile:
1313
raise errors.NoRepositoryPresent(a_bzrdir)
1315
raise errors.UnknownFormatError(format=format_string)
1318
def register_format(klass, format):
1319
format_registry.register(format.get_format_string(), format)
1322
def unregister_format(klass, format):
1323
format_registry.remove(format.get_format_string())
1326
def get_default_format(klass):
1327
"""Return the current default format."""
1328
from bzrlib import bzrdir
1329
return bzrdir.format_registry.make_bzrdir('default').repository_format
1331
def _get_control_store(self, repo_transport, control_files):
1332
"""Return the control store for this repository."""
1333
raise NotImplementedError(self._get_control_store)
1335
def get_format_string(self):
1336
"""Return the ASCII format string that identifies this format.
1338
Note that in pre format ?? repositories the format string is
1339
not permitted nor written to disk.
1341
raise NotImplementedError(self.get_format_string)
1343
def get_format_description(self):
1344
"""Return the short description for this format."""
1345
raise NotImplementedError(self.get_format_description)
1347
def _get_revision_store(self, repo_transport, control_files):
1348
"""Return the revision store object for this a_bzrdir."""
1349
raise NotImplementedError(self._get_revision_store)
1351
def _get_text_rev_store(self,
1358
"""Common logic for getting a revision store for a repository.
1360
see self._get_revision_store for the subclass-overridable method to
1361
get the store for a repository.
1363
from bzrlib.store.revision.text import TextRevisionStore
1364
dir_mode = control_files._dir_mode
1365
file_mode = control_files._file_mode
1366
text_store = TextStore(transport.clone(name),
1368
compressed=compressed,
1370
file_mode=file_mode)
1371
_revision_store = TextRevisionStore(text_store, serializer)
1372
return _revision_store
1374
# TODO: this shouldn't be in the base class, it's specific to things that
1375
# use weaves or knits -- mbp 20070207
1376
def _get_versioned_file_store(self,
1381
versionedfile_class=None,
1382
versionedfile_kwargs={},
1384
if versionedfile_class is None:
1385
versionedfile_class = self._versionedfile_class
1386
weave_transport = control_files._transport.clone(name)
1387
dir_mode = control_files._dir_mode
1388
file_mode = control_files._file_mode
1389
return VersionedFileStore(weave_transport, prefixed=prefixed,
1391
file_mode=file_mode,
1392
versionedfile_class=versionedfile_class,
1393
versionedfile_kwargs=versionedfile_kwargs,
1396
def initialize(self, a_bzrdir, shared=False):
1397
"""Initialize a repository of this format in a_bzrdir.
1399
:param a_bzrdir: The bzrdir to put the new repository in it.
1400
:param shared: The repository should be initialized as a sharable one.
1401
:returns: The new repository object.
1403
This may raise UninitializableFormat if shared repository are not
1404
compatible the a_bzrdir.
1406
raise NotImplementedError(self.initialize)
1408
def is_supported(self):
1409
"""Is this format supported?
1411
Supported formats must be initializable and openable.
1412
Unsupported formats may not support initialization or committing or
1413
some other features depending on the reason for not being supported.
1417
def check_conversion_target(self, target_format):
1418
raise NotImplementedError(self.check_conversion_target)
1420
def open(self, a_bzrdir, _found=False):
1421
"""Return an instance of this format for the bzrdir a_bzrdir.
1423
_found is a private parameter, do not use it.
1425
raise NotImplementedError(self.open)
1428
class MetaDirRepositoryFormat(RepositoryFormat):
1429
"""Common base class for the new repositories using the metadir layout."""
1431
rich_root_data = False
1432
supports_tree_reference = False
1433
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1436
super(MetaDirRepositoryFormat, self).__init__()
1438
def _create_control_files(self, a_bzrdir):
1439
"""Create the required files and the initial control_files object."""
1440
# FIXME: RBC 20060125 don't peek under the covers
1441
# NB: no need to escape relative paths that are url safe.
1442
repository_transport = a_bzrdir.get_repository_transport(self)
1443
control_files = lockable_files.LockableFiles(repository_transport,
1444
'lock', lockdir.LockDir)
1445
control_files.create_lock()
1446
return control_files
1448
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1449
"""Upload the initial blank content."""
1450
control_files = self._create_control_files(a_bzrdir)
1451
control_files.lock_write()
1453
control_files._transport.mkdir_multi(dirs,
1454
mode=control_files._dir_mode)
1455
for file, content in files:
1456
control_files.put(file, content)
1457
for file, content in utf8_files:
1458
control_files.put_utf8(file, content)
1460
control_files.put_utf8('shared-storage', '')
1462
control_files.unlock()
1465
# formats which have no format string are not discoverable
1466
# and not independently creatable, so are not registered. They're
1467
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1468
# needed, it's constructed directly by the BzrDir. Non-native formats where
1469
# the repository is not separately opened are similar.
1471
format_registry.register_lazy(
1472
'Bazaar-NG Repository format 7',
1473
'bzrlib.repofmt.weaverepo',
1476
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1477
# default control directory format
1479
format_registry.register_lazy(
1480
'Bazaar-NG Knit Repository Format 1',
1481
'bzrlib.repofmt.knitrepo',
1482
'RepositoryFormatKnit1',
1484
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1486
format_registry.register_lazy(
1487
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1488
'bzrlib.repofmt.knitrepo',
1489
'RepositoryFormatKnit3',
1493
class InterRepository(InterObject):
1494
"""This class represents operations taking place between two repositories.
1496
Its instances have methods like copy_content and fetch, and contain
1497
references to the source and target repositories these operations can be
1500
Often we will provide convenience methods on 'repository' which carry out
1501
operations with another repository - they will always forward to
1502
InterRepository.get(other).method_name(parameters).
1506
"""The available optimised InterRepository types."""
1508
def copy_content(self, revision_id=None):
1509
raise NotImplementedError(self.copy_content)
1511
def fetch(self, revision_id=None, pb=None):
1512
"""Fetch the content required to construct revision_id.
1514
The content is copied from self.source to self.target.
1516
:param revision_id: if None all content is copied, if NULL_REVISION no
1518
:param pb: optional progress bar to use for progress reports. If not
1519
provided a default one will be created.
1521
Returns the copied revision count and the failed revisions in a tuple:
1524
raise NotImplementedError(self.fetch)
1527
def missing_revision_ids(self, revision_id=None):
1528
"""Return the revision ids that source has that target does not.
1530
These are returned in topological order.
1532
:param revision_id: only return revision ids included by this
1535
# generic, possibly worst case, slow code path.
1536
target_ids = set(self.target.all_revision_ids())
1537
if revision_id is not None:
1538
# TODO: jam 20070210 InterRepository is internal enough that it
1539
# should assume revision_ids are already utf-8
1540
revision_id = osutils.safe_revision_id(revision_id)
1541
source_ids = self.source.get_ancestry(revision_id)
1542
assert source_ids[0] is None
1545
source_ids = self.source.all_revision_ids()
1546
result_set = set(source_ids).difference(target_ids)
1547
# this may look like a no-op: its not. It preserves the ordering
1548
# other_ids had while only returning the members from other_ids
1549
# that we've decided we need.
1550
return [rev_id for rev_id in source_ids if rev_id in result_set]
1553
class InterSameDataRepository(InterRepository):
1554
"""Code for converting between repositories that represent the same data.
1556
Data format and model must match for this to work.
1560
def _get_repo_format_to_test(self):
1561
"""Repository format for testing with."""
1562
return RepositoryFormat.get_default_format()
1565
def is_compatible(source, target):
1566
if source.supports_rich_root() != target.supports_rich_root():
1568
if source._serializer != target._serializer:
1573
def copy_content(self, revision_id=None):
1574
"""Make a complete copy of the content in self into destination.
1576
This copies both the repository's revision data, and configuration information
1577
such as the make_working_trees setting.
1579
This is a destructive operation! Do not use it on existing
1582
:param revision_id: Only copy the content needed to construct
1583
revision_id and its parents.
1586
self.target.set_make_working_trees(self.source.make_working_trees())
1587
except NotImplementedError:
1589
# TODO: jam 20070210 This is fairly internal, so we should probably
1590
# just assert that revision_id is not unicode.
1591
revision_id = osutils.safe_revision_id(revision_id)
1592
# but don't bother fetching if we have the needed data now.
1593
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1594
self.target.has_revision(revision_id)):
1596
self.target.fetch(self.source, revision_id=revision_id)
1599
def fetch(self, revision_id=None, pb=None):
1600
"""See InterRepository.fetch()."""
1601
from bzrlib.fetch import GenericRepoFetcher
1602
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1603
self.source, self.source._format, self.target,
1604
self.target._format)
1605
# TODO: jam 20070210 This should be an assert, not a translate
1606
revision_id = osutils.safe_revision_id(revision_id)
1607
f = GenericRepoFetcher(to_repository=self.target,
1608
from_repository=self.source,
1609
last_revision=revision_id,
1611
return f.count_copied, f.failed_revisions
1614
class InterWeaveRepo(InterSameDataRepository):
1615
"""Optimised code paths between Weave based repositories."""
1618
def _get_repo_format_to_test(self):
1619
from bzrlib.repofmt import weaverepo
1620
return weaverepo.RepositoryFormat7()
1623
def is_compatible(source, target):
1624
"""Be compatible with known Weave formats.
1626
We don't test for the stores being of specific types because that
1627
could lead to confusing results, and there is no need to be
1630
from bzrlib.repofmt.weaverepo import (
1636
return (isinstance(source._format, (RepositoryFormat5,
1638
RepositoryFormat7)) and
1639
isinstance(target._format, (RepositoryFormat5,
1641
RepositoryFormat7)))
1642
except AttributeError:
1646
def copy_content(self, revision_id=None):
1647
"""See InterRepository.copy_content()."""
1648
# weave specific optimised path:
1649
# TODO: jam 20070210 Internal, should be an assert, not translate
1650
revision_id = osutils.safe_revision_id(revision_id)
1652
self.target.set_make_working_trees(self.source.make_working_trees())
1653
except NotImplementedError:
1655
# FIXME do not peek!
1656
if self.source.control_files._transport.listable():
1657
pb = ui.ui_factory.nested_progress_bar()
1659
self.target.weave_store.copy_all_ids(
1660
self.source.weave_store,
1662
from_transaction=self.source.get_transaction(),
1663
to_transaction=self.target.get_transaction())
1664
pb.update('copying inventory', 0, 1)
1665
self.target.control_weaves.copy_multi(
1666
self.source.control_weaves, ['inventory'],
1667
from_transaction=self.source.get_transaction(),
1668
to_transaction=self.target.get_transaction())
1669
self.target._revision_store.text_store.copy_all_ids(
1670
self.source._revision_store.text_store,
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, self.target._format)
1683
# TODO: jam 20070210 This should be an assert, not a translate
1684
revision_id = osutils.safe_revision_id(revision_id)
1685
f = GenericRepoFetcher(to_repository=self.target,
1686
from_repository=self.source,
1687
last_revision=revision_id,
1689
return f.count_copied, f.failed_revisions
1692
def missing_revision_ids(self, revision_id=None):
1693
"""See InterRepository.missing_revision_ids()."""
1694
# we want all revisions to satisfy revision_id in source.
1695
# but we don't want to stat every file here and there.
1696
# we want then, all revisions other needs to satisfy revision_id
1697
# checked, but not those that we have locally.
1698
# so the first thing is to get a subset of the revisions to
1699
# satisfy revision_id in source, and then eliminate those that
1700
# we do already have.
1701
# this is slow on high latency connection to self, but as as this
1702
# disk format scales terribly for push anyway due to rewriting
1703
# inventory.weave, this is considered acceptable.
1705
if revision_id is not None:
1706
source_ids = self.source.get_ancestry(revision_id)
1707
assert source_ids[0] is None
1710
source_ids = self.source._all_possible_ids()
1711
source_ids_set = set(source_ids)
1712
# source_ids is the worst possible case we may need to pull.
1713
# now we want to filter source_ids against what we actually
1714
# have in target, but don't try to check for existence where we know
1715
# we do not have a revision as that would be pointless.
1716
target_ids = set(self.target._all_possible_ids())
1717
possibly_present_revisions = target_ids.intersection(source_ids_set)
1718
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1719
required_revisions = source_ids_set.difference(actually_present_revisions)
1720
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1721
if revision_id is not None:
1722
# we used get_ancestry to determine source_ids then we are assured all
1723
# revisions referenced are present as they are installed in topological order.
1724
# and the tip revision was validated by get_ancestry.
1725
return required_topo_revisions
1727
# if we just grabbed the possibly available ids, then
1728
# we only have an estimate of whats available and need to validate
1729
# that against the revision records.
1730
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1733
class InterKnitRepo(InterSameDataRepository):
1734
"""Optimised code paths between Knit based repositories."""
1737
def _get_repo_format_to_test(self):
1738
from bzrlib.repofmt import knitrepo
1739
return knitrepo.RepositoryFormatKnit1()
1742
def is_compatible(source, target):
1743
"""Be compatible with known Knit formats.
1745
We don't test for the stores being of specific types because that
1746
could lead to confusing results, and there is no need to be
1749
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1751
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1752
isinstance(target._format, (RepositoryFormatKnit1)))
1753
except AttributeError:
1757
def fetch(self, revision_id=None, pb=None):
1758
"""See InterRepository.fetch()."""
1759
from bzrlib.fetch import KnitRepoFetcher
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 = KnitRepoFetcher(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
if revision_id is not None:
1774
source_ids = self.source.get_ancestry(revision_id)
1775
assert source_ids[0] is None
1778
source_ids = self.source._all_possible_ids()
1779
source_ids_set = set(source_ids)
1780
# source_ids is the worst possible case we may need to pull.
1781
# now we want to filter source_ids against what we actually
1782
# have in target, but don't try to check for existence where we know
1783
# we do not have a revision as that would be pointless.
1784
target_ids = set(self.target._all_possible_ids())
1785
possibly_present_revisions = target_ids.intersection(source_ids_set)
1786
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1787
required_revisions = source_ids_set.difference(actually_present_revisions)
1788
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1789
if revision_id is not None:
1790
# we used get_ancestry to determine source_ids then we are assured all
1791
# revisions referenced are present as they are installed in topological order.
1792
# and the tip revision was validated by get_ancestry.
1793
return required_topo_revisions
1795
# if we just grabbed the possibly available ids, then
1796
# we only have an estimate of whats available and need to validate
1797
# that against the revision records.
1798
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1801
class InterModel1and2(InterRepository):
1804
def _get_repo_format_to_test(self):
1808
def is_compatible(source, target):
1809
if not source.supports_rich_root() and target.supports_rich_root():
1815
def fetch(self, revision_id=None, pb=None):
1816
"""See InterRepository.fetch()."""
1817
from bzrlib.fetch import Model1toKnit2Fetcher
1818
# TODO: jam 20070210 This should be an assert, not a translate
1819
revision_id = osutils.safe_revision_id(revision_id)
1820
f = Model1toKnit2Fetcher(to_repository=self.target,
1821
from_repository=self.source,
1822
last_revision=revision_id,
1824
return f.count_copied, f.failed_revisions
1827
def copy_content(self, revision_id=None):
1828
"""Make a complete copy of the content in self into destination.
1830
This is a destructive operation! Do not use it on existing
1833
:param revision_id: Only copy the content needed to construct
1834
revision_id and its parents.
1837
self.target.set_make_working_trees(self.source.make_working_trees())
1838
except NotImplementedError:
1840
# TODO: jam 20070210 Internal, assert, don't translate
1841
revision_id = osutils.safe_revision_id(revision_id)
1842
# but don't bother fetching if we have the needed data now.
1843
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1844
self.target.has_revision(revision_id)):
1846
self.target.fetch(self.source, revision_id=revision_id)
1849
class InterKnit1and2(InterKnitRepo):
1852
def _get_repo_format_to_test(self):
1856
def is_compatible(source, target):
1857
"""Be compatible with Knit1 source and Knit3 target"""
1858
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1860
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1861
RepositoryFormatKnit3
1862
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1863
isinstance(target._format, (RepositoryFormatKnit3)))
1864
except AttributeError:
1868
def fetch(self, revision_id=None, pb=None):
1869
"""See InterRepository.fetch()."""
1870
from bzrlib.fetch import Knit1to2Fetcher
1871
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1872
self.source, self.source._format, self.target,
1873
self.target._format)
1874
# TODO: jam 20070210 This should be an assert, not a translate
1875
revision_id = osutils.safe_revision_id(revision_id)
1876
f = Knit1to2Fetcher(to_repository=self.target,
1877
from_repository=self.source,
1878
last_revision=revision_id,
1880
return f.count_copied, f.failed_revisions
1883
class InterRemoteRepository(InterRepository):
1884
"""Code for converting between RemoteRepository objects.
1886
This just gets an non-remote repository from the RemoteRepository, and calls
1887
InterRepository.get again.
1890
def __init__(self, source, target):
1891
if isinstance(source, remote.RemoteRepository):
1892
source._ensure_real()
1893
real_source = source._real_repository
1895
real_source = source
1896
if isinstance(target, remote.RemoteRepository):
1897
target._ensure_real()
1898
real_target = target._real_repository
1900
real_target = target
1901
self.real_inter = InterRepository.get(real_source, real_target)
1904
def is_compatible(source, target):
1905
if isinstance(source, remote.RemoteRepository):
1907
if isinstance(target, remote.RemoteRepository):
1911
def copy_content(self, revision_id=None):
1912
self.real_inter.copy_content(revision_id=revision_id)
1914
def fetch(self, revision_id=None, pb=None):
1915
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1918
def _get_repo_format_to_test(self):
1922
InterRepository.register_optimiser(InterSameDataRepository)
1923
InterRepository.register_optimiser(InterWeaveRepo)
1924
InterRepository.register_optimiser(InterKnitRepo)
1925
InterRepository.register_optimiser(InterModel1and2)
1926
InterRepository.register_optimiser(InterKnit1and2)
1927
InterRepository.register_optimiser(InterRemoteRepository)
1930
class CopyConverter(object):
1931
"""A repository conversion tool which just performs a copy of the content.
1933
This is slow but quite reliable.
1936
def __init__(self, target_format):
1937
"""Create a CopyConverter.
1939
:param target_format: The format the resulting repository should be.
1941
self.target_format = target_format
1943
def convert(self, repo, pb):
1944
"""Perform the conversion of to_convert, giving feedback via pb.
1946
:param to_convert: The disk object to convert.
1947
:param pb: a progress bar to use for progress information.
1952
# this is only useful with metadir layouts - separated repo content.
1953
# trigger an assertion if not such
1954
repo._format.get_format_string()
1955
self.repo_dir = repo.bzrdir
1956
self.step('Moving repository to repository.backup')
1957
self.repo_dir.transport.move('repository', 'repository.backup')
1958
backup_transport = self.repo_dir.transport.clone('repository.backup')
1959
repo._format.check_conversion_target(self.target_format)
1960
self.source_repo = repo._format.open(self.repo_dir,
1962
_override_transport=backup_transport)
1963
self.step('Creating new repository')
1964
converted = self.target_format.initialize(self.repo_dir,
1965
self.source_repo.is_shared())
1966
converted.lock_write()
1968
self.step('Copying content into repository.')
1969
self.source_repo.copy_content_into(converted)
1972
self.step('Deleting old repository content.')
1973
self.repo_dir.transport.delete_tree('repository.backup')
1974
self.pb.note('repository converted')
1976
def step(self, message):
1977
"""Update the pb by a step."""
1979
self.pb.update(message, self.count, self.total)
1982
class CommitBuilder(object):
1983
"""Provides an interface to build up a commit.
1985
This allows describing a tree to be committed without needing to
1986
know the internals of the format of the repository.
1989
record_root_entry = False
1990
def __init__(self, repository, parents, config, timestamp=None,
1991
timezone=None, committer=None, revprops=None,
1993
"""Initiate a CommitBuilder.
1995
:param repository: Repository to commit to.
1996
:param parents: Revision ids of the parents of the new revision.
1997
:param config: Configuration to use.
1998
:param timestamp: Optional timestamp recorded for commit.
1999
:param timezone: Optional timezone for timestamp.
2000
:param committer: Optional committer to set for commit.
2001
:param revprops: Optional dictionary of revision properties.
2002
:param revision_id: Optional revision id.
2004
self._config = config
2006
if committer is None:
2007
self._committer = self._config.username()
2009
assert isinstance(committer, basestring), type(committer)
2010
self._committer = committer
2012
self.new_inventory = Inventory(None)
2013
self._new_revision_id = osutils.safe_revision_id(revision_id)
2014
self.parents = parents
2015
self.repository = repository
2018
if revprops is not None:
2019
self._revprops.update(revprops)
2021
if timestamp is None:
2022
timestamp = time.time()
2023
# Restrict resolution to 1ms
2024
self._timestamp = round(timestamp, 3)
2026
if timezone is None:
2027
self._timezone = osutils.local_time_offset()
2029
self._timezone = int(timezone)
2031
self._generate_revision_if_needed()
2033
def commit(self, message):
2034
"""Make the actual commit.
2036
:return: The revision id of the recorded revision.
2038
rev = _mod_revision.Revision(
2039
timestamp=self._timestamp,
2040
timezone=self._timezone,
2041
committer=self._committer,
2043
inventory_sha1=self.inv_sha1,
2044
revision_id=self._new_revision_id,
2045
properties=self._revprops)
2046
rev.parent_ids = self.parents
2047
self.repository.add_revision(self._new_revision_id, rev,
2048
self.new_inventory, self._config)
2049
self.repository.commit_write_group()
2050
return self._new_revision_id
2052
def revision_tree(self):
2053
"""Return the tree that was just committed.
2055
After calling commit() this can be called to get a RevisionTree
2056
representing the newly committed tree. This is preferred to
2057
calling Repository.revision_tree() because that may require
2058
deserializing the inventory, while we already have a copy in
2061
return RevisionTree(self.repository, self.new_inventory,
2062
self._new_revision_id)
2064
def finish_inventory(self):
2065
"""Tell the builder that the inventory is finished."""
2066
if self.new_inventory.root is None:
2067
symbol_versioning.warn('Root entry should be supplied to'
2068
' record_entry_contents, as of bzr 0.10.',
2069
DeprecationWarning, stacklevel=2)
2070
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2071
self.new_inventory.revision_id = self._new_revision_id
2072
self.inv_sha1 = self.repository.add_inventory(
2073
self._new_revision_id,
2078
def _gen_revision_id(self):
2079
"""Return new revision-id."""
2080
return generate_ids.gen_revision_id(self._config.username(),
2083
def _generate_revision_if_needed(self):
2084
"""Create a revision id if None was supplied.
2086
If the repository can not support user-specified revision ids
2087
they should override this function and raise CannotSetRevisionId
2088
if _new_revision_id is not None.
2090
:raises: CannotSetRevisionId
2092
if self._new_revision_id is None:
2093
self._new_revision_id = self._gen_revision_id()
2095
def record_entry_contents(self, ie, parent_invs, path, tree):
2096
"""Record the content of ie from tree into the commit if needed.
2098
Side effect: sets ie.revision when unchanged
2100
:param ie: An inventory entry present in the commit.
2101
:param parent_invs: The inventories of the parent revisions of the
2103
:param path: The path the entry is at in the tree.
2104
:param tree: The tree which contains this entry and should be used to
2107
if self.new_inventory.root is None and ie.parent_id is not None:
2108
symbol_versioning.warn('Root entry should be supplied to'
2109
' record_entry_contents, as of bzr 0.10.',
2110
DeprecationWarning, stacklevel=2)
2111
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2113
self.new_inventory.add(ie)
2115
# ie.revision is always None if the InventoryEntry is considered
2116
# for committing. ie.snapshot will record the correct revision
2117
# which may be the sole parent if it is untouched.
2118
if ie.revision is not None:
2121
# In this revision format, root entries have no knit or weave
2122
if ie is self.new_inventory.root:
2123
# When serializing out to disk and back in
2124
# root.revision is always _new_revision_id
2125
ie.revision = self._new_revision_id
2127
previous_entries = ie.find_previous_heads(
2129
self.repository.weave_store,
2130
self.repository.get_transaction())
2131
# we are creating a new revision for ie in the history store
2133
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2135
def modified_directory(self, file_id, file_parents):
2136
"""Record the presence of a symbolic link.
2138
:param file_id: The file_id of the link to record.
2139
:param file_parents: The per-file parent revision ids.
2141
self._add_text_to_weave(file_id, [], file_parents.keys())
2143
def modified_reference(self, file_id, file_parents):
2144
"""Record the modification of a reference.
2146
:param file_id: The file_id of the link to record.
2147
:param file_parents: The per-file parent revision ids.
2149
self._add_text_to_weave(file_id, [], file_parents.keys())
2151
def modified_file_text(self, file_id, file_parents,
2152
get_content_byte_lines, text_sha1=None,
2154
"""Record the text of file file_id
2156
:param file_id: The file_id of the file to record the text of.
2157
:param file_parents: The per-file parent revision ids.
2158
:param get_content_byte_lines: A callable which will return the byte
2160
:param text_sha1: Optional SHA1 of the file contents.
2161
:param text_size: Optional size of the file contents.
2163
# mutter('storing text of file {%s} in revision {%s} into %r',
2164
# file_id, self._new_revision_id, self.repository.weave_store)
2165
# special case to avoid diffing on renames or
2167
if (len(file_parents) == 1
2168
and text_sha1 == file_parents.values()[0].text_sha1
2169
and text_size == file_parents.values()[0].text_size):
2170
previous_ie = file_parents.values()[0]
2171
versionedfile = self.repository.weave_store.get_weave(file_id,
2172
self.repository.get_transaction())
2173
versionedfile.clone_text(self._new_revision_id,
2174
previous_ie.revision, file_parents.keys())
2175
return text_sha1, text_size
2177
new_lines = get_content_byte_lines()
2178
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2179
# should return the SHA1 and size
2180
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2181
return osutils.sha_strings(new_lines), \
2182
sum(map(len, new_lines))
2184
def modified_link(self, file_id, file_parents, link_target):
2185
"""Record the presence of a symbolic link.
2187
:param file_id: The file_id of the link to record.
2188
:param file_parents: The per-file parent revision ids.
2189
:param link_target: Target location of this link.
2191
self._add_text_to_weave(file_id, [], file_parents.keys())
2193
def _add_text_to_weave(self, file_id, new_lines, parents):
2194
versionedfile = self.repository.weave_store.get_weave_or_empty(
2195
file_id, self.repository.get_transaction())
2196
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2197
versionedfile.clear_cache()
2200
class _CommitBuilder(CommitBuilder):
2201
"""Temporary class so old CommitBuilders are detected properly
2203
Note: CommitBuilder works whether or not root entry is recorded.
2206
record_root_entry = True
2209
class RootCommitBuilder(CommitBuilder):
2210
"""This commitbuilder actually records the root id"""
2212
record_root_entry = True
2214
def record_entry_contents(self, ie, parent_invs, path, tree):
2215
"""Record the content of ie from tree into the commit if needed.
2217
Side effect: sets ie.revision when unchanged
2219
:param ie: An inventory entry present in the commit.
2220
:param parent_invs: The inventories of the parent revisions of the
2222
:param path: The path the entry is at in the tree.
2223
:param tree: The tree which contains this entry and should be used to
2226
assert self.new_inventory.root is not None or ie.parent_id is None
2227
self.new_inventory.add(ie)
2229
# ie.revision is always None if the InventoryEntry is considered
2230
# for committing. ie.snapshot will record the correct revision
2231
# which may be the sole parent if it is untouched.
2232
if ie.revision is not None:
2235
previous_entries = ie.find_previous_heads(
2237
self.repository.weave_store,
2238
self.repository.get_transaction())
2239
# we are creating a new revision for ie in the history store
2241
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2253
def _unescaper(match, _map=_unescape_map):
2254
code = match.group(1)
2258
if not code.startswith('#'):
2260
return unichr(int(code[1:])).encode('utf8')
2266
def _unescape_xml(data):
2267
"""Unescape predefined XML entities in a string of data."""
2269
if _unescape_re is None:
2270
_unescape_re = re.compile('\&([^;]*);')
2271
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
