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)
735
def iter_files_bytes(self, desired_files):
736
transaction = self.get_transaction()
737
for file_id, revision_id, callable_data in desired_files:
738
weave = self.weave_store.get_weave(file_id, transaction)
739
yield callable_data, weave.get_lines(revision_id)
742
def get_inventory_weave(self):
743
return self.control_weaves.get_weave('inventory',
744
self.get_transaction())
747
def get_inventory(self, revision_id):
748
"""Get Inventory object by hash."""
749
# TODO: jam 20070210 Technically we don't need to sanitize, since all
750
# called functions must sanitize.
751
revision_id = osutils.safe_revision_id(revision_id)
752
return self.deserialise_inventory(
753
revision_id, self.get_inventory_xml(revision_id))
755
def deserialise_inventory(self, revision_id, xml):
756
"""Transform the xml into an inventory object.
758
:param revision_id: The expected revision id of the inventory.
759
:param xml: A serialised inventory.
761
revision_id = osutils.safe_revision_id(revision_id)
762
result = self._serializer.read_inventory_from_string(xml)
763
result.root.revision = revision_id
766
def serialise_inventory(self, inv):
767
return self._serializer.write_inventory_to_string(inv)
769
def get_serializer_format(self):
770
return self._serializer.format_num
773
def get_inventory_xml(self, revision_id):
774
"""Get inventory XML as a file object."""
775
revision_id = osutils.safe_revision_id(revision_id)
777
assert isinstance(revision_id, str), type(revision_id)
778
iw = self.get_inventory_weave()
779
return iw.get_text(revision_id)
781
raise errors.HistoryMissing(self, 'inventory', revision_id)
784
def get_inventory_sha1(self, revision_id):
785
"""Return the sha1 hash of the inventory entry
787
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
788
revision_id = osutils.safe_revision_id(revision_id)
789
return self.get_revision(revision_id).inventory_sha1
792
def get_revision_graph(self, revision_id=None):
793
"""Return a dictionary containing the revision graph.
795
:param revision_id: The revision_id to get a graph from. If None, then
796
the entire revision graph is returned. This is a deprecated mode of
797
operation and will be removed in the future.
798
:return: a dictionary of revision_id->revision_parents_list.
800
# special case NULL_REVISION
801
if revision_id == _mod_revision.NULL_REVISION:
803
revision_id = osutils.safe_revision_id(revision_id)
804
a_weave = self.get_inventory_weave()
805
all_revisions = self._eliminate_revisions_not_present(
807
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
808
node in all_revisions])
809
if revision_id is None:
811
elif revision_id not in entire_graph:
812
raise errors.NoSuchRevision(self, revision_id)
814
# add what can be reached from revision_id
816
pending = set([revision_id])
817
while len(pending) > 0:
819
result[node] = entire_graph[node]
820
for revision_id in result[node]:
821
if revision_id not in result:
822
pending.add(revision_id)
826
def get_revision_graph_with_ghosts(self, revision_ids=None):
827
"""Return a graph of the revisions with ghosts marked as applicable.
829
:param revision_ids: an iterable of revisions to graph or None for all.
830
:return: a Graph object with the graph reachable from revision_ids.
832
result = deprecated_graph.Graph()
834
pending = set(self.all_revision_ids())
837
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
838
# special case NULL_REVISION
839
if _mod_revision.NULL_REVISION in pending:
840
pending.remove(_mod_revision.NULL_REVISION)
841
required = set(pending)
844
revision_id = pending.pop()
846
rev = self.get_revision(revision_id)
847
except errors.NoSuchRevision:
848
if revision_id in required:
851
result.add_ghost(revision_id)
853
for parent_id in rev.parent_ids:
854
# is this queued or done ?
855
if (parent_id not in pending and
856
parent_id not in done):
858
pending.add(parent_id)
859
result.add_node(revision_id, rev.parent_ids)
860
done.add(revision_id)
863
def _get_history_vf(self):
864
"""Get a versionedfile whose history graph reflects all revisions.
866
For weave repositories, this is the inventory weave.
868
return self.get_inventory_weave()
870
def iter_reverse_revision_history(self, revision_id):
871
"""Iterate backwards through revision ids in the lefthand history
873
:param revision_id: The revision id to start with. All its lefthand
874
ancestors will be traversed.
876
revision_id = osutils.safe_revision_id(revision_id)
877
if revision_id in (None, _mod_revision.NULL_REVISION):
879
next_id = revision_id
880
versionedfile = self._get_history_vf()
883
parents = versionedfile.get_parents(next_id)
884
if len(parents) == 0:
890
def get_revision_inventory(self, revision_id):
891
"""Return inventory of a past revision."""
892
# TODO: Unify this with get_inventory()
893
# bzr 0.0.6 and later imposes the constraint that the inventory_id
894
# must be the same as its revision, so this is trivial.
895
if revision_id is None:
896
# This does not make sense: if there is no revision,
897
# then it is the current tree inventory surely ?!
898
# and thus get_root_id() is something that looks at the last
899
# commit on the branch, and the get_root_id is an inventory check.
900
raise NotImplementedError
901
# return Inventory(self.get_root_id())
903
return self.get_inventory(revision_id)
907
"""Return True if this repository is flagged as a shared repository."""
908
raise NotImplementedError(self.is_shared)
911
def reconcile(self, other=None, thorough=False):
912
"""Reconcile this repository."""
913
from bzrlib.reconcile import RepoReconciler
914
reconciler = RepoReconciler(self, thorough=thorough)
915
reconciler.reconcile()
918
def _refresh_data(self):
919
"""Helper called from lock_* to ensure coherency with disk.
921
The default implementation does nothing; it is however possible
922
for repositories to maintain loaded indices across multiple locks
923
by checking inside their implementation of this method to see
924
whether their indices are still valid. This depends of course on
925
the disk format being validatable in this manner.
929
def revision_tree(self, revision_id):
930
"""Return Tree for a revision on this branch.
932
`revision_id` may be None for the empty tree revision.
934
# TODO: refactor this to use an existing revision object
935
# so we don't need to read it in twice.
936
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
937
return RevisionTree(self, Inventory(root_id=None),
938
_mod_revision.NULL_REVISION)
940
revision_id = osutils.safe_revision_id(revision_id)
941
inv = self.get_revision_inventory(revision_id)
942
return RevisionTree(self, inv, revision_id)
945
def revision_trees(self, revision_ids):
946
"""Return Tree for a revision on this branch.
948
`revision_id` may not be None or 'null:'"""
949
assert None not in revision_ids
950
assert _mod_revision.NULL_REVISION not in revision_ids
951
texts = self.get_inventory_weave().get_texts(revision_ids)
952
for text, revision_id in zip(texts, revision_ids):
953
inv = self.deserialise_inventory(revision_id, text)
954
yield RevisionTree(self, inv, revision_id)
957
def get_ancestry(self, revision_id, topo_sorted=True):
958
"""Return a list of revision-ids integrated by a revision.
960
The first element of the list is always None, indicating the origin
961
revision. This might change when we have history horizons, or
962
perhaps we should have a new API.
964
This is topologically sorted.
966
if _mod_revision.is_null(revision_id):
968
revision_id = osutils.safe_revision_id(revision_id)
969
if not self.has_revision(revision_id):
970
raise errors.NoSuchRevision(self, revision_id)
971
w = self.get_inventory_weave()
972
candidates = w.get_ancestry(revision_id, topo_sorted)
973
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
976
"""Compress the data within the repository.
978
This operation only makes sense for some repository types. For other
979
types it should be a no-op that just returns.
981
This stub method does not require a lock, but subclasses should use
982
@needs_write_lock as this is a long running call its reasonable to
983
implicitly lock for the user.
987
def print_file(self, file, revision_id):
988
"""Print `file` to stdout.
990
FIXME RBC 20060125 as John Meinel points out this is a bad api
991
- it writes to stdout, it assumes that that is valid etc. Fix
992
by creating a new more flexible convenience function.
994
revision_id = osutils.safe_revision_id(revision_id)
995
tree = self.revision_tree(revision_id)
996
# use inventory as it was in that revision
997
file_id = tree.inventory.path2id(file)
999
# TODO: jam 20060427 Write a test for this code path
1000
# it had a bug in it, and was raising the wrong
1002
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1003
tree.print_file(file_id)
1005
def get_transaction(self):
1006
return self.control_files.get_transaction()
1008
def revision_parents(self, revision_id):
1009
revision_id = osutils.safe_revision_id(revision_id)
1010
return self.get_inventory_weave().parent_names(revision_id)
1012
def get_parents(self, revision_ids):
1013
"""See StackedParentsProvider.get_parents"""
1015
for revision_id in revision_ids:
1016
if revision_id == _mod_revision.NULL_REVISION:
1020
parents = self.get_revision(revision_id).parent_ids
1021
except errors.NoSuchRevision:
1024
if len(parents) == 0:
1025
parents = [_mod_revision.NULL_REVISION]
1026
parents_list.append(parents)
1029
def _make_parents_provider(self):
1032
def get_graph(self, other_repository=None):
1033
"""Return the graph walker for this repository format"""
1034
parents_provider = self._make_parents_provider()
1035
if (other_repository is not None and
1036
other_repository.bzrdir.transport.base !=
1037
self.bzrdir.transport.base):
1038
parents_provider = graph._StackedParentsProvider(
1039
[parents_provider, other_repository._make_parents_provider()])
1040
return graph.Graph(parents_provider)
1043
def set_make_working_trees(self, new_value):
1044
"""Set the policy flag for making working trees when creating branches.
1046
This only applies to branches that use this repository.
1048
The default is 'True'.
1049
:param new_value: True to restore the default, False to disable making
1052
raise NotImplementedError(self.set_make_working_trees)
1054
def make_working_trees(self):
1055
"""Returns the policy for making working trees on new branches."""
1056
raise NotImplementedError(self.make_working_trees)
1059
def sign_revision(self, revision_id, gpg_strategy):
1060
revision_id = osutils.safe_revision_id(revision_id)
1061
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1062
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1065
def has_signature_for_revision_id(self, revision_id):
1066
"""Query for a revision signature for revision_id in the repository."""
1067
revision_id = osutils.safe_revision_id(revision_id)
1068
return self._revision_store.has_signature(revision_id,
1069
self.get_transaction())
1072
def get_signature_text(self, revision_id):
1073
"""Return the text for a signature."""
1074
revision_id = osutils.safe_revision_id(revision_id)
1075
return self._revision_store.get_signature_text(revision_id,
1076
self.get_transaction())
1079
def check(self, revision_ids):
1080
"""Check consistency of all history of given revision_ids.
1082
Different repository implementations should override _check().
1084
:param revision_ids: A non-empty list of revision_ids whose ancestry
1085
will be checked. Typically the last revision_id of a branch.
1087
if not revision_ids:
1088
raise ValueError("revision_ids must be non-empty in %s.check"
1090
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1091
return self._check(revision_ids)
1093
def _check(self, revision_ids):
1094
result = check.Check(self)
1098
def _warn_if_deprecated(self):
1099
global _deprecation_warning_done
1100
if _deprecation_warning_done:
1102
_deprecation_warning_done = True
1103
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1104
% (self._format, self.bzrdir.transport.base))
1106
def supports_rich_root(self):
1107
return self._format.rich_root_data
1109
def _check_ascii_revisionid(self, revision_id, method):
1110
"""Private helper for ascii-only repositories."""
1111
# weave repositories refuse to store revisionids that are non-ascii.
1112
if revision_id is not None:
1113
# weaves require ascii revision ids.
1114
if isinstance(revision_id, unicode):
1116
revision_id.encode('ascii')
1117
except UnicodeEncodeError:
1118
raise errors.NonAsciiRevisionId(method, self)
1121
revision_id.decode('ascii')
1122
except UnicodeDecodeError:
1123
raise errors.NonAsciiRevisionId(method, self)
1127
# remove these delegates a while after bzr 0.15
1128
def __make_delegated(name, from_module):
1129
def _deprecated_repository_forwarder():
1130
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1131
% (name, from_module),
1134
m = __import__(from_module, globals(), locals(), [name])
1136
return getattr(m, name)
1137
except AttributeError:
1138
raise AttributeError('module %s has no name %s'
1140
globals()[name] = _deprecated_repository_forwarder
1143
'AllInOneRepository',
1144
'WeaveMetaDirRepository',
1145
'PreSplitOutRepositoryFormat',
1146
'RepositoryFormat4',
1147
'RepositoryFormat5',
1148
'RepositoryFormat6',
1149
'RepositoryFormat7',
1151
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1155
'RepositoryFormatKnit',
1156
'RepositoryFormatKnit1',
1158
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1161
def install_revision(repository, rev, revision_tree):
1162
"""Install all revision data into a repository."""
1163
present_parents = []
1165
for p_id in rev.parent_ids:
1166
if repository.has_revision(p_id):
1167
present_parents.append(p_id)
1168
parent_trees[p_id] = repository.revision_tree(p_id)
1170
parent_trees[p_id] = repository.revision_tree(None)
1172
inv = revision_tree.inventory
1173
entries = inv.iter_entries()
1174
# backwards compatibility hack: skip the root id.
1175
if not repository.supports_rich_root():
1176
path, root = entries.next()
1177
if root.revision != rev.revision_id:
1178
raise errors.IncompatibleRevision(repr(repository))
1179
# Add the texts that are not already present
1180
for path, ie in entries:
1181
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1182
repository.get_transaction())
1183
if ie.revision not in w:
1185
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1186
# with InventoryEntry.find_previous_heads(). if it is, then there
1187
# is a latent bug here where the parents may have ancestors of each
1189
for revision, tree in parent_trees.iteritems():
1190
if ie.file_id not in tree:
1192
parent_id = tree.inventory[ie.file_id].revision
1193
if parent_id in text_parents:
1195
text_parents.append(parent_id)
1197
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1198
repository.get_transaction())
1199
lines = revision_tree.get_file(ie.file_id).readlines()
1200
vfile.add_lines(rev.revision_id, text_parents, lines)
1202
# install the inventory
1203
repository.add_inventory(rev.revision_id, inv, present_parents)
1204
except errors.RevisionAlreadyPresent:
1206
repository.add_revision(rev.revision_id, rev, inv)
1209
class MetaDirRepository(Repository):
1210
"""Repositories in the new meta-dir layout."""
1212
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1213
super(MetaDirRepository, self).__init__(_format,
1219
dir_mode = self.control_files._dir_mode
1220
file_mode = self.control_files._file_mode
1223
def is_shared(self):
1224
"""Return True if this repository is flagged as a shared repository."""
1225
return self.control_files._transport.has('shared-storage')
1228
def set_make_working_trees(self, new_value):
1229
"""Set the policy flag for making working trees when creating branches.
1231
This only applies to branches that use this repository.
1233
The default is 'True'.
1234
:param new_value: True to restore the default, False to disable making
1239
self.control_files._transport.delete('no-working-trees')
1240
except errors.NoSuchFile:
1243
self.control_files.put_utf8('no-working-trees', '')
1245
def make_working_trees(self):
1246
"""Returns the policy for making working trees on new branches."""
1247
return not self.control_files._transport.has('no-working-trees')
1250
class RepositoryFormatRegistry(registry.Registry):
1251
"""Registry of RepositoryFormats.
1254
def get(self, format_string):
1255
r = registry.Registry.get(self, format_string)
1261
format_registry = RepositoryFormatRegistry()
1262
"""Registry of formats, indexed by their identifying format string.
1264
This can contain either format instances themselves, or classes/factories that
1265
can be called to obtain one.
1269
#####################################################################
1270
# Repository Formats
1272
class RepositoryFormat(object):
1273
"""A repository format.
1275
Formats provide three things:
1276
* An initialization routine to construct repository data on disk.
1277
* a format string which is used when the BzrDir supports versioned
1279
* an open routine which returns a Repository instance.
1281
Formats are placed in an dict by their format string for reference
1282
during opening. These should be subclasses of RepositoryFormat
1285
Once a format is deprecated, just deprecate the initialize and open
1286
methods on the format class. Do not deprecate the object, as the
1287
object will be created every system load.
1289
Common instance attributes:
1290
_matchingbzrdir - the bzrdir format that the repository format was
1291
originally written to work with. This can be used if manually
1292
constructing a bzrdir and repository, or more commonly for test suite
1297
return "<%s>" % self.__class__.__name__
1299
def __eq__(self, other):
1300
# format objects are generally stateless
1301
return isinstance(other, self.__class__)
1303
def __ne__(self, other):
1304
return not self == other
1307
def find_format(klass, a_bzrdir):
1308
"""Return the format for the repository object in a_bzrdir.
1310
This is used by bzr native formats that have a "format" file in
1311
the repository. Other methods may be used by different types of
1315
transport = a_bzrdir.get_repository_transport(None)
1316
format_string = transport.get("format").read()
1317
return format_registry.get(format_string)
1318
except errors.NoSuchFile:
1319
raise errors.NoRepositoryPresent(a_bzrdir)
1321
raise errors.UnknownFormatError(format=format_string)
1324
def register_format(klass, format):
1325
format_registry.register(format.get_format_string(), format)
1328
def unregister_format(klass, format):
1329
format_registry.remove(format.get_format_string())
1332
def get_default_format(klass):
1333
"""Return the current default format."""
1334
from bzrlib import bzrdir
1335
return bzrdir.format_registry.make_bzrdir('default').repository_format
1337
def _get_control_store(self, repo_transport, control_files):
1338
"""Return the control store for this repository."""
1339
raise NotImplementedError(self._get_control_store)
1341
def get_format_string(self):
1342
"""Return the ASCII format string that identifies this format.
1344
Note that in pre format ?? repositories the format string is
1345
not permitted nor written to disk.
1347
raise NotImplementedError(self.get_format_string)
1349
def get_format_description(self):
1350
"""Return the short description for this format."""
1351
raise NotImplementedError(self.get_format_description)
1353
def _get_revision_store(self, repo_transport, control_files):
1354
"""Return the revision store object for this a_bzrdir."""
1355
raise NotImplementedError(self._get_revision_store)
1357
def _get_text_rev_store(self,
1364
"""Common logic for getting a revision store for a repository.
1366
see self._get_revision_store for the subclass-overridable method to
1367
get the store for a repository.
1369
from bzrlib.store.revision.text import TextRevisionStore
1370
dir_mode = control_files._dir_mode
1371
file_mode = control_files._file_mode
1372
text_store = TextStore(transport.clone(name),
1374
compressed=compressed,
1376
file_mode=file_mode)
1377
_revision_store = TextRevisionStore(text_store, serializer)
1378
return _revision_store
1380
# TODO: this shouldn't be in the base class, it's specific to things that
1381
# use weaves or knits -- mbp 20070207
1382
def _get_versioned_file_store(self,
1387
versionedfile_class=None,
1388
versionedfile_kwargs={},
1390
if versionedfile_class is None:
1391
versionedfile_class = self._versionedfile_class
1392
weave_transport = control_files._transport.clone(name)
1393
dir_mode = control_files._dir_mode
1394
file_mode = control_files._file_mode
1395
return VersionedFileStore(weave_transport, prefixed=prefixed,
1397
file_mode=file_mode,
1398
versionedfile_class=versionedfile_class,
1399
versionedfile_kwargs=versionedfile_kwargs,
1402
def initialize(self, a_bzrdir, shared=False):
1403
"""Initialize a repository of this format in a_bzrdir.
1405
:param a_bzrdir: The bzrdir to put the new repository in it.
1406
:param shared: The repository should be initialized as a sharable one.
1407
:returns: The new repository object.
1409
This may raise UninitializableFormat if shared repository are not
1410
compatible the a_bzrdir.
1412
raise NotImplementedError(self.initialize)
1414
def is_supported(self):
1415
"""Is this format supported?
1417
Supported formats must be initializable and openable.
1418
Unsupported formats may not support initialization or committing or
1419
some other features depending on the reason for not being supported.
1423
def check_conversion_target(self, target_format):
1424
raise NotImplementedError(self.check_conversion_target)
1426
def open(self, a_bzrdir, _found=False):
1427
"""Return an instance of this format for the bzrdir a_bzrdir.
1429
_found is a private parameter, do not use it.
1431
raise NotImplementedError(self.open)
1434
class MetaDirRepositoryFormat(RepositoryFormat):
1435
"""Common base class for the new repositories using the metadir layout."""
1437
rich_root_data = False
1438
supports_tree_reference = False
1439
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1442
super(MetaDirRepositoryFormat, self).__init__()
1444
def _create_control_files(self, a_bzrdir):
1445
"""Create the required files and the initial control_files object."""
1446
# FIXME: RBC 20060125 don't peek under the covers
1447
# NB: no need to escape relative paths that are url safe.
1448
repository_transport = a_bzrdir.get_repository_transport(self)
1449
control_files = lockable_files.LockableFiles(repository_transport,
1450
'lock', lockdir.LockDir)
1451
control_files.create_lock()
1452
return control_files
1454
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1455
"""Upload the initial blank content."""
1456
control_files = self._create_control_files(a_bzrdir)
1457
control_files.lock_write()
1459
control_files._transport.mkdir_multi(dirs,
1460
mode=control_files._dir_mode)
1461
for file, content in files:
1462
control_files.put(file, content)
1463
for file, content in utf8_files:
1464
control_files.put_utf8(file, content)
1466
control_files.put_utf8('shared-storage', '')
1468
control_files.unlock()
1471
# formats which have no format string are not discoverable
1472
# and not independently creatable, so are not registered. They're
1473
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1474
# needed, it's constructed directly by the BzrDir. Non-native formats where
1475
# the repository is not separately opened are similar.
1477
format_registry.register_lazy(
1478
'Bazaar-NG Repository format 7',
1479
'bzrlib.repofmt.weaverepo',
1482
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1483
# default control directory format
1485
format_registry.register_lazy(
1486
'Bazaar-NG Knit Repository Format 1',
1487
'bzrlib.repofmt.knitrepo',
1488
'RepositoryFormatKnit1',
1490
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1492
format_registry.register_lazy(
1493
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1494
'bzrlib.repofmt.knitrepo',
1495
'RepositoryFormatKnit3',
1499
class InterRepository(InterObject):
1500
"""This class represents operations taking place between two repositories.
1502
Its instances have methods like copy_content and fetch, and contain
1503
references to the source and target repositories these operations can be
1506
Often we will provide convenience methods on 'repository' which carry out
1507
operations with another repository - they will always forward to
1508
InterRepository.get(other).method_name(parameters).
1512
"""The available optimised InterRepository types."""
1514
def copy_content(self, revision_id=None):
1515
raise NotImplementedError(self.copy_content)
1517
def fetch(self, revision_id=None, pb=None):
1518
"""Fetch the content required to construct revision_id.
1520
The content is copied from self.source to self.target.
1522
:param revision_id: if None all content is copied, if NULL_REVISION no
1524
:param pb: optional progress bar to use for progress reports. If not
1525
provided a default one will be created.
1527
Returns the copied revision count and the failed revisions in a tuple:
1530
raise NotImplementedError(self.fetch)
1533
def missing_revision_ids(self, revision_id=None):
1534
"""Return the revision ids that source has that target does not.
1536
These are returned in topological order.
1538
:param revision_id: only return revision ids included by this
1541
# generic, possibly worst case, slow code path.
1542
target_ids = set(self.target.all_revision_ids())
1543
if revision_id is not None:
1544
# TODO: jam 20070210 InterRepository is internal enough that it
1545
# should assume revision_ids are already utf-8
1546
revision_id = osutils.safe_revision_id(revision_id)
1547
source_ids = self.source.get_ancestry(revision_id)
1548
assert source_ids[0] is None
1551
source_ids = self.source.all_revision_ids()
1552
result_set = set(source_ids).difference(target_ids)
1553
# this may look like a no-op: its not. It preserves the ordering
1554
# other_ids had while only returning the members from other_ids
1555
# that we've decided we need.
1556
return [rev_id for rev_id in source_ids if rev_id in result_set]
1559
class InterSameDataRepository(InterRepository):
1560
"""Code for converting between repositories that represent the same data.
1562
Data format and model must match for this to work.
1566
def _get_repo_format_to_test(self):
1567
"""Repository format for testing with."""
1568
return RepositoryFormat.get_default_format()
1571
def is_compatible(source, target):
1572
if source.supports_rich_root() != target.supports_rich_root():
1574
if source._serializer != target._serializer:
1579
def copy_content(self, revision_id=None):
1580
"""Make a complete copy of the content in self into destination.
1582
This copies both the repository's revision data, and configuration information
1583
such as the make_working_trees setting.
1585
This is a destructive operation! Do not use it on existing
1588
:param revision_id: Only copy the content needed to construct
1589
revision_id and its parents.
1592
self.target.set_make_working_trees(self.source.make_working_trees())
1593
except NotImplementedError:
1595
# TODO: jam 20070210 This is fairly internal, so we should probably
1596
# just assert that revision_id is not unicode.
1597
revision_id = osutils.safe_revision_id(revision_id)
1598
# but don't bother fetching if we have the needed data now.
1599
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1600
self.target.has_revision(revision_id)):
1602
self.target.fetch(self.source, revision_id=revision_id)
1605
def fetch(self, revision_id=None, pb=None):
1606
"""See InterRepository.fetch()."""
1607
from bzrlib.fetch import GenericRepoFetcher
1608
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1609
self.source, self.source._format, self.target,
1610
self.target._format)
1611
# TODO: jam 20070210 This should be an assert, not a translate
1612
revision_id = osutils.safe_revision_id(revision_id)
1613
f = GenericRepoFetcher(to_repository=self.target,
1614
from_repository=self.source,
1615
last_revision=revision_id,
1617
return f.count_copied, f.failed_revisions
1620
class InterWeaveRepo(InterSameDataRepository):
1621
"""Optimised code paths between Weave based repositories."""
1624
def _get_repo_format_to_test(self):
1625
from bzrlib.repofmt import weaverepo
1626
return weaverepo.RepositoryFormat7()
1629
def is_compatible(source, target):
1630
"""Be compatible with known Weave formats.
1632
We don't test for the stores being of specific types because that
1633
could lead to confusing results, and there is no need to be
1636
from bzrlib.repofmt.weaverepo import (
1642
return (isinstance(source._format, (RepositoryFormat5,
1644
RepositoryFormat7)) and
1645
isinstance(target._format, (RepositoryFormat5,
1647
RepositoryFormat7)))
1648
except AttributeError:
1652
def copy_content(self, revision_id=None):
1653
"""See InterRepository.copy_content()."""
1654
# weave specific optimised path:
1655
# TODO: jam 20070210 Internal, should be an assert, not translate
1656
revision_id = osutils.safe_revision_id(revision_id)
1658
self.target.set_make_working_trees(self.source.make_working_trees())
1659
except NotImplementedError:
1661
# FIXME do not peek!
1662
if self.source.control_files._transport.listable():
1663
pb = ui.ui_factory.nested_progress_bar()
1665
self.target.weave_store.copy_all_ids(
1666
self.source.weave_store,
1668
from_transaction=self.source.get_transaction(),
1669
to_transaction=self.target.get_transaction())
1670
pb.update('copying inventory', 0, 1)
1671
self.target.control_weaves.copy_multi(
1672
self.source.control_weaves, ['inventory'],
1673
from_transaction=self.source.get_transaction(),
1674
to_transaction=self.target.get_transaction())
1675
self.target._revision_store.text_store.copy_all_ids(
1676
self.source._revision_store.text_store,
1681
self.target.fetch(self.source, revision_id=revision_id)
1684
def fetch(self, revision_id=None, pb=None):
1685
"""See InterRepository.fetch()."""
1686
from bzrlib.fetch import GenericRepoFetcher
1687
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1688
self.source, self.source._format, self.target, self.target._format)
1689
# TODO: jam 20070210 This should be an assert, not a translate
1690
revision_id = osutils.safe_revision_id(revision_id)
1691
f = GenericRepoFetcher(to_repository=self.target,
1692
from_repository=self.source,
1693
last_revision=revision_id,
1695
return f.count_copied, f.failed_revisions
1698
def missing_revision_ids(self, revision_id=None):
1699
"""See InterRepository.missing_revision_ids()."""
1700
# we want all revisions to satisfy revision_id in source.
1701
# but we don't want to stat every file here and there.
1702
# we want then, all revisions other needs to satisfy revision_id
1703
# checked, but not those that we have locally.
1704
# so the first thing is to get a subset of the revisions to
1705
# satisfy revision_id in source, and then eliminate those that
1706
# we do already have.
1707
# this is slow on high latency connection to self, but as as this
1708
# disk format scales terribly for push anyway due to rewriting
1709
# inventory.weave, this is considered acceptable.
1711
if revision_id is not None:
1712
source_ids = self.source.get_ancestry(revision_id)
1713
assert source_ids[0] is None
1716
source_ids = self.source._all_possible_ids()
1717
source_ids_set = set(source_ids)
1718
# source_ids is the worst possible case we may need to pull.
1719
# now we want to filter source_ids against what we actually
1720
# have in target, but don't try to check for existence where we know
1721
# we do not have a revision as that would be pointless.
1722
target_ids = set(self.target._all_possible_ids())
1723
possibly_present_revisions = target_ids.intersection(source_ids_set)
1724
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1725
required_revisions = source_ids_set.difference(actually_present_revisions)
1726
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1727
if revision_id is not None:
1728
# we used get_ancestry to determine source_ids then we are assured all
1729
# revisions referenced are present as they are installed in topological order.
1730
# and the tip revision was validated by get_ancestry.
1731
return required_topo_revisions
1733
# if we just grabbed the possibly available ids, then
1734
# we only have an estimate of whats available and need to validate
1735
# that against the revision records.
1736
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1739
class InterKnitRepo(InterSameDataRepository):
1740
"""Optimised code paths between Knit based repositories."""
1743
def _get_repo_format_to_test(self):
1744
from bzrlib.repofmt import knitrepo
1745
return knitrepo.RepositoryFormatKnit1()
1748
def is_compatible(source, target):
1749
"""Be compatible with known Knit formats.
1751
We don't test for the stores being of specific types because that
1752
could lead to confusing results, and there is no need to be
1755
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1757
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1758
isinstance(target._format, (RepositoryFormatKnit1)))
1759
except AttributeError:
1763
def fetch(self, revision_id=None, pb=None):
1764
"""See InterRepository.fetch()."""
1765
from bzrlib.fetch import KnitRepoFetcher
1766
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1767
self.source, self.source._format, self.target, self.target._format)
1768
# TODO: jam 20070210 This should be an assert, not a translate
1769
revision_id = osutils.safe_revision_id(revision_id)
1770
f = KnitRepoFetcher(to_repository=self.target,
1771
from_repository=self.source,
1772
last_revision=revision_id,
1774
return f.count_copied, f.failed_revisions
1777
def missing_revision_ids(self, revision_id=None):
1778
"""See InterRepository.missing_revision_ids()."""
1779
if revision_id is not None:
1780
source_ids = self.source.get_ancestry(revision_id)
1781
assert source_ids[0] is None
1784
source_ids = self.source._all_possible_ids()
1785
source_ids_set = set(source_ids)
1786
# source_ids is the worst possible case we may need to pull.
1787
# now we want to filter source_ids against what we actually
1788
# have in target, but don't try to check for existence where we know
1789
# we do not have a revision as that would be pointless.
1790
target_ids = set(self.target._all_possible_ids())
1791
possibly_present_revisions = target_ids.intersection(source_ids_set)
1792
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1793
required_revisions = source_ids_set.difference(actually_present_revisions)
1794
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1795
if revision_id is not None:
1796
# we used get_ancestry to determine source_ids then we are assured all
1797
# revisions referenced are present as they are installed in topological order.
1798
# and the tip revision was validated by get_ancestry.
1799
return required_topo_revisions
1801
# if we just grabbed the possibly available ids, then
1802
# we only have an estimate of whats available and need to validate
1803
# that against the revision records.
1804
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1807
class InterModel1and2(InterRepository):
1810
def _get_repo_format_to_test(self):
1814
def is_compatible(source, target):
1815
if not source.supports_rich_root() and target.supports_rich_root():
1821
def fetch(self, revision_id=None, pb=None):
1822
"""See InterRepository.fetch()."""
1823
from bzrlib.fetch import Model1toKnit2Fetcher
1824
# TODO: jam 20070210 This should be an assert, not a translate
1825
revision_id = osutils.safe_revision_id(revision_id)
1826
f = Model1toKnit2Fetcher(to_repository=self.target,
1827
from_repository=self.source,
1828
last_revision=revision_id,
1830
return f.count_copied, f.failed_revisions
1833
def copy_content(self, revision_id=None):
1834
"""Make a complete copy of the content in self into destination.
1836
This is a destructive operation! Do not use it on existing
1839
:param revision_id: Only copy the content needed to construct
1840
revision_id and its parents.
1843
self.target.set_make_working_trees(self.source.make_working_trees())
1844
except NotImplementedError:
1846
# TODO: jam 20070210 Internal, assert, don't translate
1847
revision_id = osutils.safe_revision_id(revision_id)
1848
# but don't bother fetching if we have the needed data now.
1849
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1850
self.target.has_revision(revision_id)):
1852
self.target.fetch(self.source, revision_id=revision_id)
1855
class InterKnit1and2(InterKnitRepo):
1858
def _get_repo_format_to_test(self):
1862
def is_compatible(source, target):
1863
"""Be compatible with Knit1 source and Knit3 target"""
1864
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1866
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1867
RepositoryFormatKnit3
1868
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1869
isinstance(target._format, (RepositoryFormatKnit3)))
1870
except AttributeError:
1874
def fetch(self, revision_id=None, pb=None):
1875
"""See InterRepository.fetch()."""
1876
from bzrlib.fetch import Knit1to2Fetcher
1877
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1878
self.source, self.source._format, self.target,
1879
self.target._format)
1880
# TODO: jam 20070210 This should be an assert, not a translate
1881
revision_id = osutils.safe_revision_id(revision_id)
1882
f = Knit1to2Fetcher(to_repository=self.target,
1883
from_repository=self.source,
1884
last_revision=revision_id,
1886
return f.count_copied, f.failed_revisions
1889
class InterRemoteRepository(InterRepository):
1890
"""Code for converting between RemoteRepository objects.
1892
This just gets an non-remote repository from the RemoteRepository, and calls
1893
InterRepository.get again.
1896
def __init__(self, source, target):
1897
if isinstance(source, remote.RemoteRepository):
1898
source._ensure_real()
1899
real_source = source._real_repository
1901
real_source = source
1902
if isinstance(target, remote.RemoteRepository):
1903
target._ensure_real()
1904
real_target = target._real_repository
1906
real_target = target
1907
self.real_inter = InterRepository.get(real_source, real_target)
1910
def is_compatible(source, target):
1911
if isinstance(source, remote.RemoteRepository):
1913
if isinstance(target, remote.RemoteRepository):
1917
def copy_content(self, revision_id=None):
1918
self.real_inter.copy_content(revision_id=revision_id)
1920
def fetch(self, revision_id=None, pb=None):
1921
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1924
def _get_repo_format_to_test(self):
1928
InterRepository.register_optimiser(InterSameDataRepository)
1929
InterRepository.register_optimiser(InterWeaveRepo)
1930
InterRepository.register_optimiser(InterKnitRepo)
1931
InterRepository.register_optimiser(InterModel1and2)
1932
InterRepository.register_optimiser(InterKnit1and2)
1933
InterRepository.register_optimiser(InterRemoteRepository)
1936
class CopyConverter(object):
1937
"""A repository conversion tool which just performs a copy of the content.
1939
This is slow but quite reliable.
1942
def __init__(self, target_format):
1943
"""Create a CopyConverter.
1945
:param target_format: The format the resulting repository should be.
1947
self.target_format = target_format
1949
def convert(self, repo, pb):
1950
"""Perform the conversion of to_convert, giving feedback via pb.
1952
:param to_convert: The disk object to convert.
1953
:param pb: a progress bar to use for progress information.
1958
# this is only useful with metadir layouts - separated repo content.
1959
# trigger an assertion if not such
1960
repo._format.get_format_string()
1961
self.repo_dir = repo.bzrdir
1962
self.step('Moving repository to repository.backup')
1963
self.repo_dir.transport.move('repository', 'repository.backup')
1964
backup_transport = self.repo_dir.transport.clone('repository.backup')
1965
repo._format.check_conversion_target(self.target_format)
1966
self.source_repo = repo._format.open(self.repo_dir,
1968
_override_transport=backup_transport)
1969
self.step('Creating new repository')
1970
converted = self.target_format.initialize(self.repo_dir,
1971
self.source_repo.is_shared())
1972
converted.lock_write()
1974
self.step('Copying content into repository.')
1975
self.source_repo.copy_content_into(converted)
1978
self.step('Deleting old repository content.')
1979
self.repo_dir.transport.delete_tree('repository.backup')
1980
self.pb.note('repository converted')
1982
def step(self, message):
1983
"""Update the pb by a step."""
1985
self.pb.update(message, self.count, self.total)
1988
class CommitBuilder(object):
1989
"""Provides an interface to build up a commit.
1991
This allows describing a tree to be committed without needing to
1992
know the internals of the format of the repository.
1995
record_root_entry = False
1996
def __init__(self, repository, parents, config, timestamp=None,
1997
timezone=None, committer=None, revprops=None,
1999
"""Initiate a CommitBuilder.
2001
:param repository: Repository to commit to.
2002
:param parents: Revision ids of the parents of the new revision.
2003
:param config: Configuration to use.
2004
:param timestamp: Optional timestamp recorded for commit.
2005
:param timezone: Optional timezone for timestamp.
2006
:param committer: Optional committer to set for commit.
2007
:param revprops: Optional dictionary of revision properties.
2008
:param revision_id: Optional revision id.
2010
self._config = config
2012
if committer is None:
2013
self._committer = self._config.username()
2015
assert isinstance(committer, basestring), type(committer)
2016
self._committer = committer
2018
self.new_inventory = Inventory(None)
2019
self._new_revision_id = osutils.safe_revision_id(revision_id)
2020
self.parents = parents
2021
self.repository = repository
2024
if revprops is not None:
2025
self._revprops.update(revprops)
2027
if timestamp is None:
2028
timestamp = time.time()
2029
# Restrict resolution to 1ms
2030
self._timestamp = round(timestamp, 3)
2032
if timezone is None:
2033
self._timezone = osutils.local_time_offset()
2035
self._timezone = int(timezone)
2037
self._generate_revision_if_needed()
2039
def commit(self, message):
2040
"""Make the actual commit.
2042
:return: The revision id of the recorded revision.
2044
rev = _mod_revision.Revision(
2045
timestamp=self._timestamp,
2046
timezone=self._timezone,
2047
committer=self._committer,
2049
inventory_sha1=self.inv_sha1,
2050
revision_id=self._new_revision_id,
2051
properties=self._revprops)
2052
rev.parent_ids = self.parents
2053
self.repository.add_revision(self._new_revision_id, rev,
2054
self.new_inventory, self._config)
2055
self.repository.commit_write_group()
2056
return self._new_revision_id
2058
def revision_tree(self):
2059
"""Return the tree that was just committed.
2061
After calling commit() this can be called to get a RevisionTree
2062
representing the newly committed tree. This is preferred to
2063
calling Repository.revision_tree() because that may require
2064
deserializing the inventory, while we already have a copy in
2067
return RevisionTree(self.repository, self.new_inventory,
2068
self._new_revision_id)
2070
def finish_inventory(self):
2071
"""Tell the builder that the inventory is finished."""
2072
if self.new_inventory.root is None:
2073
symbol_versioning.warn('Root entry should be supplied to'
2074
' record_entry_contents, as of bzr 0.10.',
2075
DeprecationWarning, stacklevel=2)
2076
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2077
self.new_inventory.revision_id = self._new_revision_id
2078
self.inv_sha1 = self.repository.add_inventory(
2079
self._new_revision_id,
2084
def _gen_revision_id(self):
2085
"""Return new revision-id."""
2086
return generate_ids.gen_revision_id(self._config.username(),
2089
def _generate_revision_if_needed(self):
2090
"""Create a revision id if None was supplied.
2092
If the repository can not support user-specified revision ids
2093
they should override this function and raise CannotSetRevisionId
2094
if _new_revision_id is not None.
2096
:raises: CannotSetRevisionId
2098
if self._new_revision_id is None:
2099
self._new_revision_id = self._gen_revision_id()
2101
def record_entry_contents(self, ie, parent_invs, path, tree):
2102
"""Record the content of ie from tree into the commit if needed.
2104
Side effect: sets ie.revision when unchanged
2106
:param ie: An inventory entry present in the commit.
2107
:param parent_invs: The inventories of the parent revisions of the
2109
:param path: The path the entry is at in the tree.
2110
:param tree: The tree which contains this entry and should be used to
2113
if self.new_inventory.root is None and ie.parent_id is not None:
2114
symbol_versioning.warn('Root entry should be supplied to'
2115
' record_entry_contents, as of bzr 0.10.',
2116
DeprecationWarning, stacklevel=2)
2117
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2119
self.new_inventory.add(ie)
2121
# ie.revision is always None if the InventoryEntry is considered
2122
# for committing. ie.snapshot will record the correct revision
2123
# which may be the sole parent if it is untouched.
2124
if ie.revision is not None:
2127
# In this revision format, root entries have no knit or weave
2128
if ie is self.new_inventory.root:
2129
# When serializing out to disk and back in
2130
# root.revision is always _new_revision_id
2131
ie.revision = self._new_revision_id
2133
previous_entries = ie.find_previous_heads(
2135
self.repository.weave_store,
2136
self.repository.get_transaction())
2137
# we are creating a new revision for ie in the history store
2139
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2141
def modified_directory(self, file_id, file_parents):
2142
"""Record the presence of a symbolic link.
2144
:param file_id: The file_id of the link to record.
2145
:param file_parents: The per-file parent revision ids.
2147
self._add_text_to_weave(file_id, [], file_parents.keys())
2149
def modified_reference(self, file_id, file_parents):
2150
"""Record the modification of a reference.
2152
:param file_id: The file_id of the link to record.
2153
:param file_parents: The per-file parent revision ids.
2155
self._add_text_to_weave(file_id, [], file_parents.keys())
2157
def modified_file_text(self, file_id, file_parents,
2158
get_content_byte_lines, text_sha1=None,
2160
"""Record the text of file file_id
2162
:param file_id: The file_id of the file to record the text of.
2163
:param file_parents: The per-file parent revision ids.
2164
:param get_content_byte_lines: A callable which will return the byte
2166
:param text_sha1: Optional SHA1 of the file contents.
2167
:param text_size: Optional size of the file contents.
2169
# mutter('storing text of file {%s} in revision {%s} into %r',
2170
# file_id, self._new_revision_id, self.repository.weave_store)
2171
# special case to avoid diffing on renames or
2173
if (len(file_parents) == 1
2174
and text_sha1 == file_parents.values()[0].text_sha1
2175
and text_size == file_parents.values()[0].text_size):
2176
previous_ie = file_parents.values()[0]
2177
versionedfile = self.repository.weave_store.get_weave(file_id,
2178
self.repository.get_transaction())
2179
versionedfile.clone_text(self._new_revision_id,
2180
previous_ie.revision, file_parents.keys())
2181
return text_sha1, text_size
2183
new_lines = get_content_byte_lines()
2184
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2185
# should return the SHA1 and size
2186
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2187
return osutils.sha_strings(new_lines), \
2188
sum(map(len, new_lines))
2190
def modified_link(self, file_id, file_parents, link_target):
2191
"""Record the presence of a symbolic link.
2193
:param file_id: The file_id of the link to record.
2194
:param file_parents: The per-file parent revision ids.
2195
:param link_target: Target location of this link.
2197
self._add_text_to_weave(file_id, [], file_parents.keys())
2199
def _add_text_to_weave(self, file_id, new_lines, parents):
2200
versionedfile = self.repository.weave_store.get_weave_or_empty(
2201
file_id, self.repository.get_transaction())
2202
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2203
versionedfile.clear_cache()
2206
class _CommitBuilder(CommitBuilder):
2207
"""Temporary class so old CommitBuilders are detected properly
2209
Note: CommitBuilder works whether or not root entry is recorded.
2212
record_root_entry = True
2215
class RootCommitBuilder(CommitBuilder):
2216
"""This commitbuilder actually records the root id"""
2218
record_root_entry = True
2220
def record_entry_contents(self, ie, parent_invs, path, tree):
2221
"""Record the content of ie from tree into the commit if needed.
2223
Side effect: sets ie.revision when unchanged
2225
:param ie: An inventory entry present in the commit.
2226
:param parent_invs: The inventories of the parent revisions of the
2228
:param path: The path the entry is at in the tree.
2229
:param tree: The tree which contains this entry and should be used to
2232
assert self.new_inventory.root is not None or ie.parent_id is None
2233
self.new_inventory.add(ie)
2235
# ie.revision is always None if the InventoryEntry is considered
2236
# for committing. ie.snapshot will record the correct revision
2237
# which may be the sole parent if it is untouched.
2238
if ie.revision is not None:
2241
previous_entries = ie.find_previous_heads(
2243
self.repository.weave_store,
2244
self.repository.get_transaction())
2245
# we are creating a new revision for ie in the history store
2247
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2259
def _unescaper(match, _map=_unescape_map):
2260
code = match.group(1)
2264
if not code.startswith('#'):
2266
return unichr(int(code[1:])).encode('utf8')
2272
def _unescape_xml(data):
2273
"""Unescape predefined XML entities in a string of data."""
2275
if _unescape_re is None:
2276
_unescape_re = re.compile('\&([^;]*);')
2277
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
