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.revisiontree import RevisionTree
44
from bzrlib.store.versioned import VersionedFileStore
45
from bzrlib.store.text import TextStore
46
from bzrlib.testament import Testament
50
from bzrlib.decorators import needs_read_lock, needs_write_lock
51
from bzrlib.inter import InterObject
52
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
53
from bzrlib.symbol_versioning import (
57
from bzrlib.trace import mutter, note, warning
60
# Old formats display a warning, but only once
61
_deprecation_warning_done = False
64
######################################################################
67
class Repository(object):
68
"""Repository holding history for one or more branches.
70
The repository holds and retrieves historical information including
71
revisions and file history. It's normally accessed only by the Branch,
72
which views a particular line of development through that history.
74
The Repository builds on top of Stores and a Transport, which respectively
75
describe the disk data format and the way of accessing the (possibly
79
_file_ids_altered_regex = lazy_regex.lazy_compile(
80
r'file_id="(?P<file_id>[^"]+)"'
81
r'.*revision="(?P<revision_id>[^"]+)"'
85
def add_inventory(self, revision_id, inv, parents):
86
"""Add the inventory inv to the repository as revision_id.
88
:param parents: The revision ids of the parents that revision_id
89
is known to have and are in the repository already.
91
returns the sha1 of the serialized inventory.
93
revision_id = osutils.safe_revision_id(revision_id)
94
_mod_revision.check_not_reserved_id(revision_id)
95
assert inv.revision_id is None or inv.revision_id == revision_id, \
96
"Mismatch between inventory revision" \
97
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
98
assert inv.root is not None
99
inv_text = self.serialise_inventory(inv)
100
inv_sha1 = osutils.sha_string(inv_text)
101
inv_vf = self.control_weaves.get_weave('inventory',
102
self.get_transaction())
103
self._inventory_add_lines(inv_vf, revision_id, parents,
104
osutils.split_lines(inv_text))
107
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
109
for parent in parents:
111
final_parents.append(parent)
113
inv_vf.add_lines(revision_id, final_parents, lines)
116
def add_revision(self, revision_id, rev, inv=None, config=None):
117
"""Add rev to the revision store as revision_id.
119
:param revision_id: the revision id to use.
120
:param rev: The revision object.
121
:param inv: The inventory for the revision. if None, it will be looked
122
up in the inventory storer
123
:param config: If None no digital signature will be created.
124
If supplied its signature_needed method will be used
125
to determine if a signature should be made.
127
revision_id = osutils.safe_revision_id(revision_id)
128
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
130
_mod_revision.check_not_reserved_id(revision_id)
131
if config is not None and config.signature_needed():
133
inv = self.get_inventory(revision_id)
134
plaintext = Testament(rev, inv).as_short_text()
135
self.store_revision_signature(
136
gpg.GPGStrategy(config), plaintext, revision_id)
137
if not revision_id in self.get_inventory_weave():
139
raise errors.WeaveRevisionNotPresent(revision_id,
140
self.get_inventory_weave())
142
# yes, this is not suitable for adding with ghosts.
143
self.add_inventory(revision_id, inv, rev.parent_ids)
144
self._revision_store.add_revision(rev, self.get_transaction())
147
def _all_possible_ids(self):
148
"""Return all the possible revisions that we could find."""
149
return self.get_inventory_weave().versions()
151
def all_revision_ids(self):
152
"""Returns a list of all the revision ids in the repository.
154
This is deprecated because code should generally work on the graph
155
reachable from a particular revision, and ignore any other revisions
156
that might be present. There is no direct replacement method.
158
return self._all_revision_ids()
161
def _all_revision_ids(self):
162
"""Returns a list of all the revision ids in the repository.
164
These are in as much topological order as the underlying store can
165
present: for weaves ghosts may lead to a lack of correctness until
166
the reweave updates the parents list.
168
if self._revision_store.text_store.listable():
169
return self._revision_store.all_revision_ids(self.get_transaction())
170
result = self._all_possible_ids()
171
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
172
# ids. (It should, since _revision_store's API should change to
173
# return utf8 revision_ids)
174
return self._eliminate_revisions_not_present(result)
176
def break_lock(self):
177
"""Break a lock if one is present from another instance.
179
Uses the ui factory to ask for confirmation if the lock may be from
182
self.control_files.break_lock()
185
def _eliminate_revisions_not_present(self, revision_ids):
186
"""Check every revision id in revision_ids to see if we have it.
188
Returns a set of the present revisions.
191
for id in revision_ids:
192
if self.has_revision(id):
197
def create(a_bzrdir):
198
"""Construct the current default format repository in a_bzrdir."""
199
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
201
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
202
"""instantiate a Repository.
204
:param _format: The format of the repository on disk.
205
:param a_bzrdir: The BzrDir of the repository.
207
In the future we will have a single api for all stores for
208
getting file texts, inventories and revisions, then
209
this construct will accept instances of those things.
211
super(Repository, self).__init__()
212
self._format = _format
213
# the following are part of the public API for Repository:
214
self.bzrdir = a_bzrdir
215
self.control_files = control_files
216
self._revision_store = _revision_store
217
self.text_store = text_store
218
# backwards compatibility
219
self.weave_store = text_store
220
# not right yet - should be more semantically clear ?
222
self.control_store = control_store
223
self.control_weaves = control_store
224
# TODO: make sure to construct the right store classes, etc, depending
225
# on whether escaping is required.
226
self._warn_if_deprecated()
229
return '%s(%r)' % (self.__class__.__name__,
230
self.bzrdir.transport.base)
233
return self.control_files.is_locked()
235
def lock_write(self, token=None):
236
"""Lock this repository for writing.
238
:param token: if this is already locked, then lock_write will fail
239
unless the token matches the existing lock.
240
:returns: a token if this instance supports tokens, otherwise None.
241
:raises TokenLockingNotSupported: when a token is given but this
242
instance doesn't support using token locks.
243
:raises MismatchedToken: if the specified token doesn't match the token
244
of the existing lock.
246
A token should be passed in if you know that you have locked the object
247
some other way, and need to synchronise this object's state with that
250
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
252
return self.control_files.lock_write(token=token)
255
self.control_files.lock_read()
257
def get_physical_lock_status(self):
258
return self.control_files.get_physical_lock_status()
260
def leave_lock_in_place(self):
261
"""Tell this repository not to release the physical lock when this
264
If lock_write doesn't return a token, then this method is not supported.
266
self.control_files.leave_in_place()
268
def dont_leave_lock_in_place(self):
269
"""Tell this repository to release the physical lock when this
270
object is unlocked, even if it didn't originally acquire it.
272
If lock_write doesn't return a token, then this method is not supported.
274
self.control_files.dont_leave_in_place()
277
def gather_stats(self, revid=None, committers=None):
278
"""Gather statistics from a revision id.
280
:param revid: The revision id to gather statistics from, if None, then
281
no revision specific statistics are gathered.
282
:param committers: Optional parameter controlling whether to grab
283
a count of committers from the revision specific statistics.
284
:return: A dictionary of statistics. Currently this contains:
285
committers: The number of committers if requested.
286
firstrev: A tuple with timestamp, timezone for the penultimate left
287
most ancestor of revid, if revid is not the NULL_REVISION.
288
latestrev: A tuple with timestamp, timezone for revid, if revid is
289
not the NULL_REVISION.
290
revisions: The total revision count in the repository.
291
size: An estimate disk size of the repository in bytes.
294
if revid and committers:
295
result['committers'] = 0
296
if revid and revid != _mod_revision.NULL_REVISION:
298
all_committers = set()
299
revisions = self.get_ancestry(revid)
300
# pop the leading None
302
first_revision = None
304
# ignore the revisions in the middle - just grab first and last
305
revisions = revisions[0], revisions[-1]
306
for revision in self.get_revisions(revisions):
307
if not first_revision:
308
first_revision = revision
310
all_committers.add(revision.committer)
311
last_revision = revision
313
result['committers'] = len(all_committers)
314
result['firstrev'] = (first_revision.timestamp,
315
first_revision.timezone)
316
result['latestrev'] = (last_revision.timestamp,
317
last_revision.timezone)
319
# now gather global repository information
320
if self.bzrdir.root_transport.listable():
321
c, t = self._revision_store.total_size(self.get_transaction())
322
result['revisions'] = c
327
def missing_revision_ids(self, other, revision_id=None):
328
"""Return the revision ids that other has that this does not.
330
These are returned in topological order.
332
revision_id: only return revision ids included by revision_id.
334
revision_id = osutils.safe_revision_id(revision_id)
335
return InterRepository.get(other, self).missing_revision_ids(revision_id)
339
"""Open the repository rooted at base.
341
For instance, if the repository is at URL/.bzr/repository,
342
Repository.open(URL) -> a Repository instance.
344
control = bzrdir.BzrDir.open(base)
345
return control.open_repository()
347
def copy_content_into(self, destination, revision_id=None):
348
"""Make a complete copy of the content in self into destination.
350
This is a destructive operation! Do not use it on existing
353
revision_id = osutils.safe_revision_id(revision_id)
354
return InterRepository.get(self, destination).copy_content(revision_id)
356
def fetch(self, source, revision_id=None, pb=None):
357
"""Fetch the content required to construct revision_id from source.
359
If revision_id is None all content is copied.
361
revision_id = osutils.safe_revision_id(revision_id)
362
inter = InterRepository.get(source, self)
364
return inter.fetch(revision_id=revision_id, pb=pb)
365
except NotImplementedError:
366
raise errors.IncompatibleRepositories(source, self)
368
def get_commit_builder(self, branch, parents, config, timestamp=None,
369
timezone=None, committer=None, revprops=None,
371
"""Obtain a CommitBuilder for this repository.
373
:param branch: Branch to commit to.
374
:param parents: Revision ids of the parents of the new revision.
375
:param config: Configuration to use.
376
:param timestamp: Optional timestamp recorded for commit.
377
:param timezone: Optional timezone for timestamp.
378
:param committer: Optional committer to set for commit.
379
:param revprops: Optional dictionary of revision properties.
380
:param revision_id: Optional revision id.
382
revision_id = osutils.safe_revision_id(revision_id)
383
return _CommitBuilder(self, parents, config, timestamp, timezone,
384
committer, revprops, revision_id)
387
self.control_files.unlock()
390
def clone(self, a_bzrdir, revision_id=None):
391
"""Clone this repository into a_bzrdir using the current format.
393
Currently no check is made that the format of this repository and
394
the bzrdir format are compatible. FIXME RBC 20060201.
396
:return: The newly created destination repository.
398
# TODO: deprecate after 0.16; cloning this with all its settings is
399
# probably not very useful -- mbp 20070423
400
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
401
self.copy_content_into(dest_repo, revision_id)
405
def sprout(self, to_bzrdir, revision_id=None):
406
"""Create a descendent repository for new development.
408
Unlike clone, this does not copy the settings of the repository.
410
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
411
dest_repo.fetch(self, revision_id=revision_id)
414
def _create_sprouting_repo(self, a_bzrdir, shared):
415
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
416
# use target default format.
417
dest_repo = a_bzrdir.create_repository()
419
# Most control formats need the repository to be specifically
420
# created, but on some old all-in-one formats it's not needed
422
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
423
except errors.UninitializableFormat:
424
dest_repo = a_bzrdir.open_repository()
428
def has_revision(self, revision_id):
429
"""True if this repository has a copy of the revision."""
430
revision_id = osutils.safe_revision_id(revision_id)
431
return self._revision_store.has_revision_id(revision_id,
432
self.get_transaction())
435
def get_revision_reconcile(self, revision_id):
436
"""'reconcile' helper routine that allows access to a revision always.
438
This variant of get_revision does not cross check the weave graph
439
against the revision one as get_revision does: but it should only
440
be used by reconcile, or reconcile-alike commands that are correcting
441
or testing the revision graph.
443
if not revision_id or not isinstance(revision_id, basestring):
444
raise errors.InvalidRevisionId(revision_id=revision_id,
446
return self.get_revisions([revision_id])[0]
449
def get_revisions(self, revision_ids):
450
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
451
revs = self._revision_store.get_revisions(revision_ids,
452
self.get_transaction())
454
assert not isinstance(rev.revision_id, unicode)
455
for parent_id in rev.parent_ids:
456
assert not isinstance(parent_id, unicode)
460
def get_revision_xml(self, revision_id):
461
# TODO: jam 20070210 This shouldn't be necessary since get_revision
462
# would have already do it.
463
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
464
revision_id = osutils.safe_revision_id(revision_id)
465
rev = self.get_revision(revision_id)
467
# the current serializer..
468
self._revision_store._serializer.write_revision(rev, rev_tmp)
470
return rev_tmp.getvalue()
473
def get_revision(self, revision_id):
474
"""Return the Revision object for a named revision"""
475
# TODO: jam 20070210 get_revision_reconcile should do this for us
476
revision_id = osutils.safe_revision_id(revision_id)
477
r = self.get_revision_reconcile(revision_id)
478
# weave corruption can lead to absent revision markers that should be
480
# the following test is reasonably cheap (it needs a single weave read)
481
# and the weave is cached in read transactions. In write transactions
482
# it is not cached but typically we only read a small number of
483
# revisions. For knits when they are introduced we will probably want
484
# to ensure that caching write transactions are in use.
485
inv = self.get_inventory_weave()
486
self._check_revision_parents(r, inv)
490
def get_deltas_for_revisions(self, revisions):
491
"""Produce a generator of revision deltas.
493
Note that the input is a sequence of REVISIONS, not revision_ids.
494
Trees will be held in memory until the generator exits.
495
Each delta is relative to the revision's lefthand predecessor.
497
required_trees = set()
498
for revision in revisions:
499
required_trees.add(revision.revision_id)
500
required_trees.update(revision.parent_ids[:1])
501
if ('null:' in required_trees):
502
import pdb; pdb.set_trace()
504
trees = dict((t.get_revision_id(), t) for
505
t in self.revision_trees(required_trees))
506
for revision in revisions:
507
if not revision.parent_ids:
508
old_tree = self.revision_tree(None)
510
old_tree = trees[revision.parent_ids[0]]
511
yield trees[revision.revision_id].changes_from(old_tree)
514
def get_revision_delta(self, revision_id):
515
"""Return the delta for one revision.
517
The delta is relative to the left-hand predecessor of the
520
r = self.get_revision(revision_id)
521
return list(self.get_deltas_for_revisions([r]))[0]
523
def _check_revision_parents(self, revision, inventory):
524
"""Private to Repository and Fetch.
526
This checks the parentage of revision in an inventory weave for
527
consistency and is only applicable to inventory-weave-for-ancestry
528
using repository formats & fetchers.
530
weave_parents = inventory.get_parents(revision.revision_id)
531
weave_names = inventory.versions()
532
for parent_id in revision.parent_ids:
533
if parent_id in weave_names:
534
# this parent must not be a ghost.
535
if not parent_id in weave_parents:
537
raise errors.CorruptRepository(self)
540
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
541
revision_id = osutils.safe_revision_id(revision_id)
542
signature = gpg_strategy.sign(plaintext)
543
self._revision_store.add_revision_signature_text(revision_id,
545
self.get_transaction())
547
def fileids_altered_by_revision_ids(self, revision_ids):
548
"""Find the file ids and versions affected by revisions.
550
:param revisions: an iterable containing revision ids.
551
:return: a dictionary mapping altered file-ids to an iterable of
552
revision_ids. Each altered file-ids has the exact revision_ids that
553
altered it listed explicitly.
555
assert self._serializer.support_altered_by_hack, \
556
("fileids_altered_by_revision_ids only supported for branches "
557
"which store inventory as unnested xml, not on %r" % self)
558
selected_revision_ids = set(osutils.safe_revision_id(r)
559
for r in revision_ids)
560
w = self.get_inventory_weave()
563
# this code needs to read every new line in every inventory for the
564
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
565
# not present in one of those inventories is unnecessary but not
566
# harmful because we are filtering by the revision id marker in the
567
# inventory lines : we only select file ids altered in one of those
568
# revisions. We don't need to see all lines in the inventory because
569
# only those added in an inventory in rev X can contain a revision=X
571
unescape_revid_cache = {}
572
unescape_fileid_cache = {}
574
# jam 20061218 In a big fetch, this handles hundreds of thousands
575
# of lines, so it has had a lot of inlining and optimizing done.
576
# Sorry that it is a little bit messy.
577
# Move several functions to be local variables, since this is a long
579
search = self._file_ids_altered_regex.search
580
unescape = _unescape_xml
581
setdefault = result.setdefault
582
pb = ui.ui_factory.nested_progress_bar()
584
for line in w.iter_lines_added_or_present_in_versions(
585
selected_revision_ids, pb=pb):
589
# One call to match.group() returning multiple items is quite a
590
# bit faster than 2 calls to match.group() each returning 1
591
file_id, revision_id = match.group('file_id', 'revision_id')
593
# Inlining the cache lookups helps a lot when you make 170,000
594
# lines and 350k ids, versus 8.4 unique ids.
595
# Using a cache helps in 2 ways:
596
# 1) Avoids unnecessary decoding calls
597
# 2) Re-uses cached strings, which helps in future set and
599
# (2) is enough that removing encoding entirely along with
600
# the cache (so we are using plain strings) results in no
601
# performance improvement.
603
revision_id = unescape_revid_cache[revision_id]
605
unescaped = unescape(revision_id)
606
unescape_revid_cache[revision_id] = unescaped
607
revision_id = unescaped
609
if revision_id in selected_revision_ids:
611
file_id = unescape_fileid_cache[file_id]
613
unescaped = unescape(file_id)
614
unescape_fileid_cache[file_id] = unescaped
616
setdefault(file_id, set()).add(revision_id)
622
def get_inventory_weave(self):
623
return self.control_weaves.get_weave('inventory',
624
self.get_transaction())
627
def get_inventory(self, revision_id):
628
"""Get Inventory object by hash."""
629
# TODO: jam 20070210 Technically we don't need to sanitize, since all
630
# called functions must sanitize.
631
revision_id = osutils.safe_revision_id(revision_id)
632
return self.deserialise_inventory(
633
revision_id, self.get_inventory_xml(revision_id))
635
def deserialise_inventory(self, revision_id, xml):
636
"""Transform the xml into an inventory object.
638
:param revision_id: The expected revision id of the inventory.
639
:param xml: A serialised inventory.
641
revision_id = osutils.safe_revision_id(revision_id)
642
result = self._serializer.read_inventory_from_string(xml)
643
result.root.revision = revision_id
646
def serialise_inventory(self, inv):
647
return self._serializer.write_inventory_to_string(inv)
650
def get_inventory_xml(self, revision_id):
651
"""Get inventory XML as a file object."""
652
revision_id = osutils.safe_revision_id(revision_id)
654
assert isinstance(revision_id, str), type(revision_id)
655
iw = self.get_inventory_weave()
656
return iw.get_text(revision_id)
658
raise errors.HistoryMissing(self, 'inventory', revision_id)
661
def get_inventory_sha1(self, revision_id):
662
"""Return the sha1 hash of the inventory entry
664
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
665
revision_id = osutils.safe_revision_id(revision_id)
666
return self.get_revision(revision_id).inventory_sha1
669
def get_revision_graph(self, revision_id=None):
670
"""Return a dictionary containing the revision graph.
672
:param revision_id: The revision_id to get a graph from. If None, then
673
the entire revision graph is returned. This is a deprecated mode of
674
operation and will be removed in the future.
675
:return: a dictionary of revision_id->revision_parents_list.
677
# special case NULL_REVISION
678
if revision_id == _mod_revision.NULL_REVISION:
680
revision_id = osutils.safe_revision_id(revision_id)
681
a_weave = self.get_inventory_weave()
682
all_revisions = self._eliminate_revisions_not_present(
684
entire_graph = dict([(node, a_weave.get_parents(node)) for
685
node in all_revisions])
686
if revision_id is None:
688
elif revision_id not in entire_graph:
689
raise errors.NoSuchRevision(self, revision_id)
691
# add what can be reached from revision_id
693
pending = set([revision_id])
694
while len(pending) > 0:
696
result[node] = entire_graph[node]
697
for revision_id in result[node]:
698
if revision_id not in result:
699
pending.add(revision_id)
703
def get_revision_graph_with_ghosts(self, revision_ids=None):
704
"""Return a graph of the revisions with ghosts marked as applicable.
706
:param revision_ids: an iterable of revisions to graph or None for all.
707
:return: a Graph object with the graph reachable from revision_ids.
709
result = deprecated_graph.Graph()
711
pending = set(self.all_revision_ids())
714
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
715
# special case NULL_REVISION
716
if _mod_revision.NULL_REVISION in pending:
717
pending.remove(_mod_revision.NULL_REVISION)
718
required = set(pending)
721
revision_id = pending.pop()
723
rev = self.get_revision(revision_id)
724
except errors.NoSuchRevision:
725
if revision_id in required:
728
result.add_ghost(revision_id)
730
for parent_id in rev.parent_ids:
731
# is this queued or done ?
732
if (parent_id not in pending and
733
parent_id not in done):
735
pending.add(parent_id)
736
result.add_node(revision_id, rev.parent_ids)
737
done.add(revision_id)
740
def _get_history_vf(self):
741
"""Get a versionedfile whose history graph reflects all revisions.
743
For weave repositories, this is the inventory weave.
745
return self.get_inventory_weave()
747
def iter_reverse_revision_history(self, revision_id):
748
"""Iterate backwards through revision ids in the lefthand history
750
:param revision_id: The revision id to start with. All its lefthand
751
ancestors will be traversed.
753
revision_id = osutils.safe_revision_id(revision_id)
754
if revision_id in (None, _mod_revision.NULL_REVISION):
756
next_id = revision_id
757
versionedfile = self._get_history_vf()
760
parents = versionedfile.get_parents(next_id)
761
if len(parents) == 0:
767
def get_revision_inventory(self, revision_id):
768
"""Return inventory of a past revision."""
769
# TODO: Unify this with get_inventory()
770
# bzr 0.0.6 and later imposes the constraint that the inventory_id
771
# must be the same as its revision, so this is trivial.
772
if revision_id is None:
773
# This does not make sense: if there is no revision,
774
# then it is the current tree inventory surely ?!
775
# and thus get_root_id() is something that looks at the last
776
# commit on the branch, and the get_root_id is an inventory check.
777
raise NotImplementedError
778
# return Inventory(self.get_root_id())
780
return self.get_inventory(revision_id)
784
"""Return True if this repository is flagged as a shared repository."""
785
raise NotImplementedError(self.is_shared)
788
def reconcile(self, other=None, thorough=False):
789
"""Reconcile this repository."""
790
from bzrlib.reconcile import RepoReconciler
791
reconciler = RepoReconciler(self, thorough=thorough)
792
reconciler.reconcile()
796
def revision_tree(self, revision_id):
797
"""Return Tree for a revision on this branch.
799
`revision_id` may be None for the empty tree revision.
801
# TODO: refactor this to use an existing revision object
802
# so we don't need to read it in twice.
803
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
804
return RevisionTree(self, Inventory(root_id=None),
805
_mod_revision.NULL_REVISION)
807
revision_id = osutils.safe_revision_id(revision_id)
808
inv = self.get_revision_inventory(revision_id)
809
return RevisionTree(self, inv, revision_id)
812
def revision_trees(self, revision_ids):
813
"""Return Tree for a revision on this branch.
815
`revision_id` may not be None or 'null:'"""
816
assert None not in revision_ids
817
assert _mod_revision.NULL_REVISION not in revision_ids
818
texts = self.get_inventory_weave().get_texts(revision_ids)
819
for text, revision_id in zip(texts, revision_ids):
820
inv = self.deserialise_inventory(revision_id, text)
821
yield RevisionTree(self, inv, revision_id)
824
def get_ancestry(self, revision_id, topo_sorted=True):
825
"""Return a list of revision-ids integrated by a revision.
827
The first element of the list is always None, indicating the origin
828
revision. This might change when we have history horizons, or
829
perhaps we should have a new API.
831
This is topologically sorted.
833
if _mod_revision.is_null(revision_id):
835
revision_id = osutils.safe_revision_id(revision_id)
836
if not self.has_revision(revision_id):
837
raise errors.NoSuchRevision(self, revision_id)
838
w = self.get_inventory_weave()
839
candidates = w.get_ancestry(revision_id, topo_sorted)
840
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
843
def print_file(self, file, revision_id):
844
"""Print `file` to stdout.
846
FIXME RBC 20060125 as John Meinel points out this is a bad api
847
- it writes to stdout, it assumes that that is valid etc. Fix
848
by creating a new more flexible convenience function.
850
revision_id = osutils.safe_revision_id(revision_id)
851
tree = self.revision_tree(revision_id)
852
# use inventory as it was in that revision
853
file_id = tree.inventory.path2id(file)
855
# TODO: jam 20060427 Write a test for this code path
856
# it had a bug in it, and was raising the wrong
858
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
859
tree.print_file(file_id)
861
def get_transaction(self):
862
return self.control_files.get_transaction()
864
def revision_parents(self, revision_id):
865
revision_id = osutils.safe_revision_id(revision_id)
866
return self.get_inventory_weave().parent_names(revision_id)
868
def get_parents(self, revision_ids):
869
"""See StackedParentsProvider.get_parents"""
871
for revision_id in revision_ids:
872
if revision_id == _mod_revision.NULL_REVISION:
876
parents = self.get_revision(revision_id).parent_ids
877
except errors.NoSuchRevision:
880
if len(parents) == 0:
881
parents = [_mod_revision.NULL_REVISION]
882
parents_list.append(parents)
885
def _make_parents_provider(self):
888
def get_graph(self, other_repository=None):
889
"""Return the graph walker for this repository format"""
890
parents_provider = self._make_parents_provider()
891
if (other_repository is not None and
892
other_repository.bzrdir.transport.base !=
893
self.bzrdir.transport.base):
894
parents_provider = graph._StackedParentsProvider(
895
[parents_provider, other_repository._make_parents_provider()])
896
return graph.Graph(parents_provider)
899
def set_make_working_trees(self, new_value):
900
"""Set the policy flag for making working trees when creating branches.
902
This only applies to branches that use this repository.
904
The default is 'True'.
905
:param new_value: True to restore the default, False to disable making
908
raise NotImplementedError(self.set_make_working_trees)
910
def make_working_trees(self):
911
"""Returns the policy for making working trees on new branches."""
912
raise NotImplementedError(self.make_working_trees)
915
def sign_revision(self, revision_id, gpg_strategy):
916
revision_id = osutils.safe_revision_id(revision_id)
917
plaintext = Testament.from_revision(self, revision_id).as_short_text()
918
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
921
def has_signature_for_revision_id(self, revision_id):
922
"""Query for a revision signature for revision_id in the repository."""
923
revision_id = osutils.safe_revision_id(revision_id)
924
return self._revision_store.has_signature(revision_id,
925
self.get_transaction())
928
def get_signature_text(self, revision_id):
929
"""Return the text for a signature."""
930
revision_id = osutils.safe_revision_id(revision_id)
931
return self._revision_store.get_signature_text(revision_id,
932
self.get_transaction())
935
def check(self, revision_ids):
936
"""Check consistency of all history of given revision_ids.
938
Different repository implementations should override _check().
940
:param revision_ids: A non-empty list of revision_ids whose ancestry
941
will be checked. Typically the last revision_id of a branch.
944
raise ValueError("revision_ids must be non-empty in %s.check"
946
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
947
return self._check(revision_ids)
949
def _check(self, revision_ids):
950
result = check.Check(self)
954
def _warn_if_deprecated(self):
955
global _deprecation_warning_done
956
if _deprecation_warning_done:
958
_deprecation_warning_done = True
959
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
960
% (self._format, self.bzrdir.transport.base))
962
def supports_rich_root(self):
963
return self._format.rich_root_data
965
def _check_ascii_revisionid(self, revision_id, method):
966
"""Private helper for ascii-only repositories."""
967
# weave repositories refuse to store revisionids that are non-ascii.
968
if revision_id is not None:
969
# weaves require ascii revision ids.
970
if isinstance(revision_id, unicode):
972
revision_id.encode('ascii')
973
except UnicodeEncodeError:
974
raise errors.NonAsciiRevisionId(method, self)
977
revision_id.decode('ascii')
978
except UnicodeDecodeError:
979
raise errors.NonAsciiRevisionId(method, self)
983
# remove these delegates a while after bzr 0.15
984
def __make_delegated(name, from_module):
985
def _deprecated_repository_forwarder():
986
symbol_versioning.warn('%s moved to %s in bzr 0.15'
987
% (name, from_module),
990
m = __import__(from_module, globals(), locals(), [name])
992
return getattr(m, name)
993
except AttributeError:
994
raise AttributeError('module %s has no name %s'
996
globals()[name] = _deprecated_repository_forwarder
999
'AllInOneRepository',
1000
'WeaveMetaDirRepository',
1001
'PreSplitOutRepositoryFormat',
1002
'RepositoryFormat4',
1003
'RepositoryFormat5',
1004
'RepositoryFormat6',
1005
'RepositoryFormat7',
1007
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1011
'RepositoryFormatKnit',
1012
'RepositoryFormatKnit1',
1014
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1017
def install_revision(repository, rev, revision_tree):
1018
"""Install all revision data into a repository."""
1019
present_parents = []
1021
for p_id in rev.parent_ids:
1022
if repository.has_revision(p_id):
1023
present_parents.append(p_id)
1024
parent_trees[p_id] = repository.revision_tree(p_id)
1026
parent_trees[p_id] = repository.revision_tree(None)
1028
inv = revision_tree.inventory
1029
entries = inv.iter_entries()
1030
# backwards compatability hack: skip the root id.
1031
if not repository.supports_rich_root():
1032
path, root = entries.next()
1033
if root.revision != rev.revision_id:
1034
raise errors.IncompatibleRevision(repr(repository))
1035
# Add the texts that are not already present
1036
for path, ie in entries:
1037
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1038
repository.get_transaction())
1039
if ie.revision not in w:
1041
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1042
# with InventoryEntry.find_previous_heads(). if it is, then there
1043
# is a latent bug here where the parents may have ancestors of each
1045
for revision, tree in parent_trees.iteritems():
1046
if ie.file_id not in tree:
1048
parent_id = tree.inventory[ie.file_id].revision
1049
if parent_id in text_parents:
1051
text_parents.append(parent_id)
1053
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1054
repository.get_transaction())
1055
lines = revision_tree.get_file(ie.file_id).readlines()
1056
vfile.add_lines(rev.revision_id, text_parents, lines)
1058
# install the inventory
1059
repository.add_inventory(rev.revision_id, inv, present_parents)
1060
except errors.RevisionAlreadyPresent:
1062
repository.add_revision(rev.revision_id, rev, inv)
1065
class MetaDirRepository(Repository):
1066
"""Repositories in the new meta-dir layout."""
1068
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1069
super(MetaDirRepository, self).__init__(_format,
1075
dir_mode = self.control_files._dir_mode
1076
file_mode = self.control_files._file_mode
1079
def is_shared(self):
1080
"""Return True if this repository is flagged as a shared repository."""
1081
return self.control_files._transport.has('shared-storage')
1084
def set_make_working_trees(self, new_value):
1085
"""Set the policy flag for making working trees when creating branches.
1087
This only applies to branches that use this repository.
1089
The default is 'True'.
1090
:param new_value: True to restore the default, False to disable making
1095
self.control_files._transport.delete('no-working-trees')
1096
except errors.NoSuchFile:
1099
self.control_files.put_utf8('no-working-trees', '')
1101
def make_working_trees(self):
1102
"""Returns the policy for making working trees on new branches."""
1103
return not self.control_files._transport.has('no-working-trees')
1106
class RepositoryFormatRegistry(registry.Registry):
1107
"""Registry of RepositoryFormats.
1110
def get(self, format_string):
1111
r = registry.Registry.get(self, format_string)
1117
format_registry = RepositoryFormatRegistry()
1118
"""Registry of formats, indexed by their identifying format string.
1120
This can contain either format instances themselves, or classes/factories that
1121
can be called to obtain one.
1125
#####################################################################
1126
# Repository Formats
1128
class RepositoryFormat(object):
1129
"""A repository format.
1131
Formats provide three things:
1132
* An initialization routine to construct repository data on disk.
1133
* a format string which is used when the BzrDir supports versioned
1135
* an open routine which returns a Repository instance.
1137
Formats are placed in an dict by their format string for reference
1138
during opening. These should be subclasses of RepositoryFormat
1141
Once a format is deprecated, just deprecate the initialize and open
1142
methods on the format class. Do not deprecate the object, as the
1143
object will be created every system load.
1145
Common instance attributes:
1146
_matchingbzrdir - the bzrdir format that the repository format was
1147
originally written to work with. This can be used if manually
1148
constructing a bzrdir and repository, or more commonly for test suite
1153
return "<%s>" % self.__class__.__name__
1155
def __eq__(self, other):
1156
# format objects are generally stateless
1157
return isinstance(other, self.__class__)
1159
def __ne__(self, other):
1160
return not self == other
1163
def find_format(klass, a_bzrdir):
1164
"""Return the format for the repository object in a_bzrdir.
1166
This is used by bzr native formats that have a "format" file in
1167
the repository. Other methods may be used by different types of
1171
transport = a_bzrdir.get_repository_transport(None)
1172
format_string = transport.get("format").read()
1173
return format_registry.get(format_string)
1174
except errors.NoSuchFile:
1175
raise errors.NoRepositoryPresent(a_bzrdir)
1177
raise errors.UnknownFormatError(format=format_string)
1180
def register_format(klass, format):
1181
format_registry.register(format.get_format_string(), format)
1184
def unregister_format(klass, format):
1185
format_registry.remove(format.get_format_string())
1188
def get_default_format(klass):
1189
"""Return the current default format."""
1190
from bzrlib import bzrdir
1191
return bzrdir.format_registry.make_bzrdir('default').repository_format
1193
def _get_control_store(self, repo_transport, control_files):
1194
"""Return the control store for this repository."""
1195
raise NotImplementedError(self._get_control_store)
1197
def get_format_string(self):
1198
"""Return the ASCII format string that identifies this format.
1200
Note that in pre format ?? repositories the format string is
1201
not permitted nor written to disk.
1203
raise NotImplementedError(self.get_format_string)
1205
def get_format_description(self):
1206
"""Return the short description for this format."""
1207
raise NotImplementedError(self.get_format_description)
1209
def _get_revision_store(self, repo_transport, control_files):
1210
"""Return the revision store object for this a_bzrdir."""
1211
raise NotImplementedError(self._get_revision_store)
1213
def _get_text_rev_store(self,
1220
"""Common logic for getting a revision store for a repository.
1222
see self._get_revision_store for the subclass-overridable method to
1223
get the store for a repository.
1225
from bzrlib.store.revision.text import TextRevisionStore
1226
dir_mode = control_files._dir_mode
1227
file_mode = control_files._file_mode
1228
text_store = TextStore(transport.clone(name),
1230
compressed=compressed,
1232
file_mode=file_mode)
1233
_revision_store = TextRevisionStore(text_store, serializer)
1234
return _revision_store
1236
# TODO: this shouldn't be in the base class, it's specific to things that
1237
# use weaves or knits -- mbp 20070207
1238
def _get_versioned_file_store(self,
1243
versionedfile_class=None,
1244
versionedfile_kwargs={},
1246
if versionedfile_class is None:
1247
versionedfile_class = self._versionedfile_class
1248
weave_transport = control_files._transport.clone(name)
1249
dir_mode = control_files._dir_mode
1250
file_mode = control_files._file_mode
1251
return VersionedFileStore(weave_transport, prefixed=prefixed,
1253
file_mode=file_mode,
1254
versionedfile_class=versionedfile_class,
1255
versionedfile_kwargs=versionedfile_kwargs,
1258
def initialize(self, a_bzrdir, shared=False):
1259
"""Initialize a repository of this format in a_bzrdir.
1261
:param a_bzrdir: The bzrdir to put the new repository in it.
1262
:param shared: The repository should be initialized as a sharable one.
1263
:returns: The new repository object.
1265
This may raise UninitializableFormat if shared repository are not
1266
compatible the a_bzrdir.
1268
raise NotImplementedError(self.initialize)
1270
def is_supported(self):
1271
"""Is this format supported?
1273
Supported formats must be initializable and openable.
1274
Unsupported formats may not support initialization or committing or
1275
some other features depending on the reason for not being supported.
1279
def check_conversion_target(self, target_format):
1280
raise NotImplementedError(self.check_conversion_target)
1282
def open(self, a_bzrdir, _found=False):
1283
"""Return an instance of this format for the bzrdir a_bzrdir.
1285
_found is a private parameter, do not use it.
1287
raise NotImplementedError(self.open)
1290
class MetaDirRepositoryFormat(RepositoryFormat):
1291
"""Common base class for the new repositories using the metadir layout."""
1293
rich_root_data = False
1294
supports_tree_reference = False
1295
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1298
super(MetaDirRepositoryFormat, self).__init__()
1300
def _create_control_files(self, a_bzrdir):
1301
"""Create the required files and the initial control_files object."""
1302
# FIXME: RBC 20060125 don't peek under the covers
1303
# NB: no need to escape relative paths that are url safe.
1304
repository_transport = a_bzrdir.get_repository_transport(self)
1305
control_files = lockable_files.LockableFiles(repository_transport,
1306
'lock', lockdir.LockDir)
1307
control_files.create_lock()
1308
return control_files
1310
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1311
"""Upload the initial blank content."""
1312
control_files = self._create_control_files(a_bzrdir)
1313
control_files.lock_write()
1315
control_files._transport.mkdir_multi(dirs,
1316
mode=control_files._dir_mode)
1317
for file, content in files:
1318
control_files.put(file, content)
1319
for file, content in utf8_files:
1320
control_files.put_utf8(file, content)
1322
control_files.put_utf8('shared-storage', '')
1324
control_files.unlock()
1327
# formats which have no format string are not discoverable
1328
# and not independently creatable, so are not registered. They're
1329
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1330
# needed, it's constructed directly by the BzrDir. Non-native formats where
1331
# the repository is not separately opened are similar.
1333
format_registry.register_lazy(
1334
'Bazaar-NG Repository format 7',
1335
'bzrlib.repofmt.weaverepo',
1338
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1339
# default control directory format
1341
format_registry.register_lazy(
1342
'Bazaar-NG Knit Repository Format 1',
1343
'bzrlib.repofmt.knitrepo',
1344
'RepositoryFormatKnit1',
1346
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1348
format_registry.register_lazy(
1349
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1350
'bzrlib.repofmt.knitrepo',
1351
'RepositoryFormatKnit3',
1355
class InterRepository(InterObject):
1356
"""This class represents operations taking place between two repositories.
1358
Its instances have methods like copy_content and fetch, and contain
1359
references to the source and target repositories these operations can be
1362
Often we will provide convenience methods on 'repository' which carry out
1363
operations with another repository - they will always forward to
1364
InterRepository.get(other).method_name(parameters).
1368
"""The available optimised InterRepository types."""
1370
def copy_content(self, revision_id=None):
1371
raise NotImplementedError(self.copy_content)
1373
def fetch(self, revision_id=None, pb=None):
1374
"""Fetch the content required to construct revision_id.
1376
The content is copied from self.source to self.target.
1378
:param revision_id: if None all content is copied, if NULL_REVISION no
1380
:param pb: optional progress bar to use for progress reports. If not
1381
provided a default one will be created.
1383
Returns the copied revision count and the failed revisions in a tuple:
1386
raise NotImplementedError(self.fetch)
1389
def missing_revision_ids(self, revision_id=None):
1390
"""Return the revision ids that source has that target does not.
1392
These are returned in topological order.
1394
:param revision_id: only return revision ids included by this
1397
# generic, possibly worst case, slow code path.
1398
target_ids = set(self.target.all_revision_ids())
1399
if revision_id is not None:
1400
# TODO: jam 20070210 InterRepository is internal enough that it
1401
# should assume revision_ids are already utf-8
1402
revision_id = osutils.safe_revision_id(revision_id)
1403
source_ids = self.source.get_ancestry(revision_id)
1404
assert source_ids[0] is None
1407
source_ids = self.source.all_revision_ids()
1408
result_set = set(source_ids).difference(target_ids)
1409
# this may look like a no-op: its not. It preserves the ordering
1410
# other_ids had while only returning the members from other_ids
1411
# that we've decided we need.
1412
return [rev_id for rev_id in source_ids if rev_id in result_set]
1415
class InterSameDataRepository(InterRepository):
1416
"""Code for converting between repositories that represent the same data.
1418
Data format and model must match for this to work.
1422
def _get_repo_format_to_test(self):
1423
"""Repository format for testing with."""
1424
return RepositoryFormat.get_default_format()
1427
def is_compatible(source, target):
1428
if source.supports_rich_root() != target.supports_rich_root():
1430
if source._serializer != target._serializer:
1435
def copy_content(self, revision_id=None):
1436
"""Make a complete copy of the content in self into destination.
1438
This copies both the repository's revision data, and configuration information
1439
such as the make_working_trees setting.
1441
This is a destructive operation! Do not use it on existing
1444
:param revision_id: Only copy the content needed to construct
1445
revision_id and its parents.
1448
self.target.set_make_working_trees(self.source.make_working_trees())
1449
except NotImplementedError:
1451
# TODO: jam 20070210 This is fairly internal, so we should probably
1452
# just assert that revision_id is not unicode.
1453
revision_id = osutils.safe_revision_id(revision_id)
1454
# but don't bother fetching if we have the needed data now.
1455
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1456
self.target.has_revision(revision_id)):
1458
self.target.fetch(self.source, revision_id=revision_id)
1461
def fetch(self, revision_id=None, pb=None):
1462
"""See InterRepository.fetch()."""
1463
from bzrlib.fetch import GenericRepoFetcher
1464
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1465
self.source, self.source._format, self.target,
1466
self.target._format)
1467
# TODO: jam 20070210 This should be an assert, not a translate
1468
revision_id = osutils.safe_revision_id(revision_id)
1469
f = GenericRepoFetcher(to_repository=self.target,
1470
from_repository=self.source,
1471
last_revision=revision_id,
1473
return f.count_copied, f.failed_revisions
1476
class InterWeaveRepo(InterSameDataRepository):
1477
"""Optimised code paths between Weave based repositories."""
1480
def _get_repo_format_to_test(self):
1481
from bzrlib.repofmt import weaverepo
1482
return weaverepo.RepositoryFormat7()
1485
def is_compatible(source, target):
1486
"""Be compatible with known Weave formats.
1488
We don't test for the stores being of specific types because that
1489
could lead to confusing results, and there is no need to be
1492
from bzrlib.repofmt.weaverepo import (
1498
return (isinstance(source._format, (RepositoryFormat5,
1500
RepositoryFormat7)) and
1501
isinstance(target._format, (RepositoryFormat5,
1503
RepositoryFormat7)))
1504
except AttributeError:
1508
def copy_content(self, revision_id=None):
1509
"""See InterRepository.copy_content()."""
1510
# weave specific optimised path:
1511
# TODO: jam 20070210 Internal, should be an assert, not translate
1512
revision_id = osutils.safe_revision_id(revision_id)
1514
self.target.set_make_working_trees(self.source.make_working_trees())
1515
except NotImplementedError:
1517
# FIXME do not peek!
1518
if self.source.control_files._transport.listable():
1519
pb = ui.ui_factory.nested_progress_bar()
1521
self.target.weave_store.copy_all_ids(
1522
self.source.weave_store,
1524
from_transaction=self.source.get_transaction(),
1525
to_transaction=self.target.get_transaction())
1526
pb.update('copying inventory', 0, 1)
1527
self.target.control_weaves.copy_multi(
1528
self.source.control_weaves, ['inventory'],
1529
from_transaction=self.source.get_transaction(),
1530
to_transaction=self.target.get_transaction())
1531
self.target._revision_store.text_store.copy_all_ids(
1532
self.source._revision_store.text_store,
1537
self.target.fetch(self.source, revision_id=revision_id)
1540
def fetch(self, revision_id=None, pb=None):
1541
"""See InterRepository.fetch()."""
1542
from bzrlib.fetch import GenericRepoFetcher
1543
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1544
self.source, self.source._format, self.target, self.target._format)
1545
# TODO: jam 20070210 This should be an assert, not a translate
1546
revision_id = osutils.safe_revision_id(revision_id)
1547
f = GenericRepoFetcher(to_repository=self.target,
1548
from_repository=self.source,
1549
last_revision=revision_id,
1551
return f.count_copied, f.failed_revisions
1554
def missing_revision_ids(self, revision_id=None):
1555
"""See InterRepository.missing_revision_ids()."""
1556
# we want all revisions to satisfy revision_id in source.
1557
# but we don't want to stat every file here and there.
1558
# we want then, all revisions other needs to satisfy revision_id
1559
# checked, but not those that we have locally.
1560
# so the first thing is to get a subset of the revisions to
1561
# satisfy revision_id in source, and then eliminate those that
1562
# we do already have.
1563
# this is slow on high latency connection to self, but as as this
1564
# disk format scales terribly for push anyway due to rewriting
1565
# inventory.weave, this is considered acceptable.
1567
if revision_id is not None:
1568
source_ids = self.source.get_ancestry(revision_id)
1569
assert source_ids[0] is None
1572
source_ids = self.source._all_possible_ids()
1573
source_ids_set = set(source_ids)
1574
# source_ids is the worst possible case we may need to pull.
1575
# now we want to filter source_ids against what we actually
1576
# have in target, but don't try to check for existence where we know
1577
# we do not have a revision as that would be pointless.
1578
target_ids = set(self.target._all_possible_ids())
1579
possibly_present_revisions = target_ids.intersection(source_ids_set)
1580
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1581
required_revisions = source_ids_set.difference(actually_present_revisions)
1582
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1583
if revision_id is not None:
1584
# we used get_ancestry to determine source_ids then we are assured all
1585
# revisions referenced are present as they are installed in topological order.
1586
# and the tip revision was validated by get_ancestry.
1587
return required_topo_revisions
1589
# if we just grabbed the possibly available ids, then
1590
# we only have an estimate of whats available and need to validate
1591
# that against the revision records.
1592
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1595
class InterKnitRepo(InterSameDataRepository):
1596
"""Optimised code paths between Knit based repositories."""
1599
def _get_repo_format_to_test(self):
1600
from bzrlib.repofmt import knitrepo
1601
return knitrepo.RepositoryFormatKnit1()
1604
def is_compatible(source, target):
1605
"""Be compatible with known Knit formats.
1607
We don't test for the stores being of specific types because that
1608
could lead to confusing results, and there is no need to be
1611
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1613
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1614
isinstance(target._format, (RepositoryFormatKnit1)))
1615
except AttributeError:
1619
def fetch(self, revision_id=None, pb=None):
1620
"""See InterRepository.fetch()."""
1621
from bzrlib.fetch import KnitRepoFetcher
1622
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1623
self.source, self.source._format, self.target, self.target._format)
1624
# TODO: jam 20070210 This should be an assert, not a translate
1625
revision_id = osutils.safe_revision_id(revision_id)
1626
f = KnitRepoFetcher(to_repository=self.target,
1627
from_repository=self.source,
1628
last_revision=revision_id,
1630
return f.count_copied, f.failed_revisions
1633
def missing_revision_ids(self, revision_id=None):
1634
"""See InterRepository.missing_revision_ids()."""
1635
if revision_id is not None:
1636
source_ids = self.source.get_ancestry(revision_id)
1637
assert source_ids[0] is None
1640
source_ids = self.source._all_possible_ids()
1641
source_ids_set = set(source_ids)
1642
# source_ids is the worst possible case we may need to pull.
1643
# now we want to filter source_ids against what we actually
1644
# have in target, but don't try to check for existence where we know
1645
# we do not have a revision as that would be pointless.
1646
target_ids = set(self.target._all_possible_ids())
1647
possibly_present_revisions = target_ids.intersection(source_ids_set)
1648
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1649
required_revisions = source_ids_set.difference(actually_present_revisions)
1650
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1651
if revision_id is not None:
1652
# we used get_ancestry to determine source_ids then we are assured all
1653
# revisions referenced are present as they are installed in topological order.
1654
# and the tip revision was validated by get_ancestry.
1655
return required_topo_revisions
1657
# if we just grabbed the possibly available ids, then
1658
# we only have an estimate of whats available and need to validate
1659
# that against the revision records.
1660
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1663
class InterModel1and2(InterRepository):
1666
def _get_repo_format_to_test(self):
1670
def is_compatible(source, target):
1671
if not source.supports_rich_root() and target.supports_rich_root():
1677
def fetch(self, revision_id=None, pb=None):
1678
"""See InterRepository.fetch()."""
1679
from bzrlib.fetch import Model1toKnit2Fetcher
1680
# TODO: jam 20070210 This should be an assert, not a translate
1681
revision_id = osutils.safe_revision_id(revision_id)
1682
f = Model1toKnit2Fetcher(to_repository=self.target,
1683
from_repository=self.source,
1684
last_revision=revision_id,
1686
return f.count_copied, f.failed_revisions
1689
def copy_content(self, revision_id=None):
1690
"""Make a complete copy of the content in self into destination.
1692
This is a destructive operation! Do not use it on existing
1695
:param revision_id: Only copy the content needed to construct
1696
revision_id and its parents.
1699
self.target.set_make_working_trees(self.source.make_working_trees())
1700
except NotImplementedError:
1702
# TODO: jam 20070210 Internal, assert, don't translate
1703
revision_id = osutils.safe_revision_id(revision_id)
1704
# but don't bother fetching if we have the needed data now.
1705
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1706
self.target.has_revision(revision_id)):
1708
self.target.fetch(self.source, revision_id=revision_id)
1711
class InterKnit1and2(InterKnitRepo):
1714
def _get_repo_format_to_test(self):
1718
def is_compatible(source, target):
1719
"""Be compatible with Knit1 source and Knit3 target"""
1720
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1722
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1723
RepositoryFormatKnit3
1724
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1725
isinstance(target._format, (RepositoryFormatKnit3)))
1726
except AttributeError:
1730
def fetch(self, revision_id=None, pb=None):
1731
"""See InterRepository.fetch()."""
1732
from bzrlib.fetch import Knit1to2Fetcher
1733
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1734
self.source, self.source._format, self.target,
1735
self.target._format)
1736
# TODO: jam 20070210 This should be an assert, not a translate
1737
revision_id = osutils.safe_revision_id(revision_id)
1738
f = Knit1to2Fetcher(to_repository=self.target,
1739
from_repository=self.source,
1740
last_revision=revision_id,
1742
return f.count_copied, f.failed_revisions
1745
class InterRemoteRepository(InterRepository):
1746
"""Code for converting between RemoteRepository objects.
1748
This just gets an non-remote repository from the RemoteRepository, and calls
1749
InterRepository.get again.
1752
def __init__(self, source, target):
1753
if isinstance(source, remote.RemoteRepository):
1754
source._ensure_real()
1755
real_source = source._real_repository
1757
real_source = source
1758
if isinstance(target, remote.RemoteRepository):
1759
target._ensure_real()
1760
real_target = target._real_repository
1762
real_target = target
1763
self.real_inter = InterRepository.get(real_source, real_target)
1766
def is_compatible(source, target):
1767
if isinstance(source, remote.RemoteRepository):
1769
if isinstance(target, remote.RemoteRepository):
1773
def copy_content(self, revision_id=None):
1774
self.real_inter.copy_content(revision_id=revision_id)
1776
def fetch(self, revision_id=None, pb=None):
1777
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1780
def _get_repo_format_to_test(self):
1784
InterRepository.register_optimiser(InterSameDataRepository)
1785
InterRepository.register_optimiser(InterWeaveRepo)
1786
InterRepository.register_optimiser(InterKnitRepo)
1787
InterRepository.register_optimiser(InterModel1and2)
1788
InterRepository.register_optimiser(InterKnit1and2)
1789
InterRepository.register_optimiser(InterRemoteRepository)
1792
class CopyConverter(object):
1793
"""A repository conversion tool which just performs a copy of the content.
1795
This is slow but quite reliable.
1798
def __init__(self, target_format):
1799
"""Create a CopyConverter.
1801
:param target_format: The format the resulting repository should be.
1803
self.target_format = target_format
1805
def convert(self, repo, pb):
1806
"""Perform the conversion of to_convert, giving feedback via pb.
1808
:param to_convert: The disk object to convert.
1809
:param pb: a progress bar to use for progress information.
1814
# this is only useful with metadir layouts - separated repo content.
1815
# trigger an assertion if not such
1816
repo._format.get_format_string()
1817
self.repo_dir = repo.bzrdir
1818
self.step('Moving repository to repository.backup')
1819
self.repo_dir.transport.move('repository', 'repository.backup')
1820
backup_transport = self.repo_dir.transport.clone('repository.backup')
1821
repo._format.check_conversion_target(self.target_format)
1822
self.source_repo = repo._format.open(self.repo_dir,
1824
_override_transport=backup_transport)
1825
self.step('Creating new repository')
1826
converted = self.target_format.initialize(self.repo_dir,
1827
self.source_repo.is_shared())
1828
converted.lock_write()
1830
self.step('Copying content into repository.')
1831
self.source_repo.copy_content_into(converted)
1834
self.step('Deleting old repository content.')
1835
self.repo_dir.transport.delete_tree('repository.backup')
1836
self.pb.note('repository converted')
1838
def step(self, message):
1839
"""Update the pb by a step."""
1841
self.pb.update(message, self.count, self.total)
1844
class CommitBuilder(object):
1845
"""Provides an interface to build up a commit.
1847
This allows describing a tree to be committed without needing to
1848
know the internals of the format of the repository.
1851
record_root_entry = False
1852
def __init__(self, repository, parents, config, timestamp=None,
1853
timezone=None, committer=None, revprops=None,
1855
"""Initiate a CommitBuilder.
1857
:param repository: Repository to commit to.
1858
:param parents: Revision ids of the parents of the new revision.
1859
:param config: Configuration to use.
1860
:param timestamp: Optional timestamp recorded for commit.
1861
:param timezone: Optional timezone for timestamp.
1862
:param committer: Optional committer to set for commit.
1863
:param revprops: Optional dictionary of revision properties.
1864
:param revision_id: Optional revision id.
1866
self._config = config
1868
if committer is None:
1869
self._committer = self._config.username()
1871
assert isinstance(committer, basestring), type(committer)
1872
self._committer = committer
1874
self.new_inventory = Inventory(None)
1875
self._new_revision_id = osutils.safe_revision_id(revision_id)
1876
self.parents = parents
1877
self.repository = repository
1880
if revprops is not None:
1881
self._revprops.update(revprops)
1883
if timestamp is None:
1884
timestamp = time.time()
1885
# Restrict resolution to 1ms
1886
self._timestamp = round(timestamp, 3)
1888
if timezone is None:
1889
self._timezone = osutils.local_time_offset()
1891
self._timezone = int(timezone)
1893
self._generate_revision_if_needed()
1895
def commit(self, message):
1896
"""Make the actual commit.
1898
:return: The revision id of the recorded revision.
1900
rev = _mod_revision.Revision(
1901
timestamp=self._timestamp,
1902
timezone=self._timezone,
1903
committer=self._committer,
1905
inventory_sha1=self.inv_sha1,
1906
revision_id=self._new_revision_id,
1907
properties=self._revprops)
1908
rev.parent_ids = self.parents
1909
self.repository.add_revision(self._new_revision_id, rev,
1910
self.new_inventory, self._config)
1911
return self._new_revision_id
1913
def revision_tree(self):
1914
"""Return the tree that was just committed.
1916
After calling commit() this can be called to get a RevisionTree
1917
representing the newly committed tree. This is preferred to
1918
calling Repository.revision_tree() because that may require
1919
deserializing the inventory, while we already have a copy in
1922
return RevisionTree(self.repository, self.new_inventory,
1923
self._new_revision_id)
1925
def finish_inventory(self):
1926
"""Tell the builder that the inventory is finished."""
1927
if self.new_inventory.root is None:
1928
symbol_versioning.warn('Root entry should be supplied to'
1929
' record_entry_contents, as of bzr 0.10.',
1930
DeprecationWarning, stacklevel=2)
1931
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1932
self.new_inventory.revision_id = self._new_revision_id
1933
self.inv_sha1 = self.repository.add_inventory(
1934
self._new_revision_id,
1939
def _gen_revision_id(self):
1940
"""Return new revision-id."""
1941
return generate_ids.gen_revision_id(self._config.username(),
1944
def _generate_revision_if_needed(self):
1945
"""Create a revision id if None was supplied.
1947
If the repository can not support user-specified revision ids
1948
they should override this function and raise CannotSetRevisionId
1949
if _new_revision_id is not None.
1951
:raises: CannotSetRevisionId
1953
if self._new_revision_id is None:
1954
self._new_revision_id = self._gen_revision_id()
1956
def record_entry_contents(self, ie, parent_invs, path, tree):
1957
"""Record the content of ie from tree into the commit if needed.
1959
Side effect: sets ie.revision when unchanged
1961
:param ie: An inventory entry present in the commit.
1962
:param parent_invs: The inventories of the parent revisions of the
1964
:param path: The path the entry is at in the tree.
1965
:param tree: The tree which contains this entry and should be used to
1968
if self.new_inventory.root is None and ie.parent_id is not None:
1969
symbol_versioning.warn('Root entry should be supplied to'
1970
' record_entry_contents, as of bzr 0.10.',
1971
DeprecationWarning, stacklevel=2)
1972
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1974
self.new_inventory.add(ie)
1976
# ie.revision is always None if the InventoryEntry is considered
1977
# for committing. ie.snapshot will record the correct revision
1978
# which may be the sole parent if it is untouched.
1979
if ie.revision is not None:
1982
# In this revision format, root entries have no knit or weave
1983
if ie is self.new_inventory.root:
1984
# When serializing out to disk and back in
1985
# root.revision is always _new_revision_id
1986
ie.revision = self._new_revision_id
1988
previous_entries = ie.find_previous_heads(
1990
self.repository.weave_store,
1991
self.repository.get_transaction())
1992
# we are creating a new revision for ie in the history store
1994
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1996
def modified_directory(self, file_id, file_parents):
1997
"""Record the presence of a symbolic link.
1999
:param file_id: The file_id of the link to record.
2000
:param file_parents: The per-file parent revision ids.
2002
self._add_text_to_weave(file_id, [], file_parents.keys())
2004
def modified_reference(self, file_id, file_parents):
2005
"""Record the modification of a reference.
2007
:param file_id: The file_id of the link to record.
2008
:param file_parents: The per-file parent revision ids.
2010
self._add_text_to_weave(file_id, [], file_parents.keys())
2012
def modified_file_text(self, file_id, file_parents,
2013
get_content_byte_lines, text_sha1=None,
2015
"""Record the text of file file_id
2017
:param file_id: The file_id of the file to record the text of.
2018
:param file_parents: The per-file parent revision ids.
2019
:param get_content_byte_lines: A callable which will return the byte
2021
:param text_sha1: Optional SHA1 of the file contents.
2022
:param text_size: Optional size of the file contents.
2024
# mutter('storing text of file {%s} in revision {%s} into %r',
2025
# file_id, self._new_revision_id, self.repository.weave_store)
2026
# special case to avoid diffing on renames or
2028
if (len(file_parents) == 1
2029
and text_sha1 == file_parents.values()[0].text_sha1
2030
and text_size == file_parents.values()[0].text_size):
2031
previous_ie = file_parents.values()[0]
2032
versionedfile = self.repository.weave_store.get_weave(file_id,
2033
self.repository.get_transaction())
2034
versionedfile.clone_text(self._new_revision_id,
2035
previous_ie.revision, file_parents.keys())
2036
return text_sha1, text_size
2038
new_lines = get_content_byte_lines()
2039
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2040
# should return the SHA1 and size
2041
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2042
return osutils.sha_strings(new_lines), \
2043
sum(map(len, new_lines))
2045
def modified_link(self, file_id, file_parents, link_target):
2046
"""Record the presence of a symbolic link.
2048
:param file_id: The file_id of the link to record.
2049
:param file_parents: The per-file parent revision ids.
2050
:param link_target: Target location of this link.
2052
self._add_text_to_weave(file_id, [], file_parents.keys())
2054
def _add_text_to_weave(self, file_id, new_lines, parents):
2055
versionedfile = self.repository.weave_store.get_weave_or_empty(
2056
file_id, self.repository.get_transaction())
2057
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2058
versionedfile.clear_cache()
2061
class _CommitBuilder(CommitBuilder):
2062
"""Temporary class so old CommitBuilders are detected properly
2064
Note: CommitBuilder works whether or not root entry is recorded.
2067
record_root_entry = True
2070
class RootCommitBuilder(CommitBuilder):
2071
"""This commitbuilder actually records the root id"""
2073
record_root_entry = True
2075
def record_entry_contents(self, ie, parent_invs, path, tree):
2076
"""Record the content of ie from tree into the commit if needed.
2078
Side effect: sets ie.revision when unchanged
2080
:param ie: An inventory entry present in the commit.
2081
:param parent_invs: The inventories of the parent revisions of the
2083
:param path: The path the entry is at in the tree.
2084
:param tree: The tree which contains this entry and should be used to
2087
assert self.new_inventory.root is not None or ie.parent_id is None
2088
self.new_inventory.add(ie)
2090
# ie.revision is always None if the InventoryEntry is considered
2091
# for committing. ie.snapshot will record the correct revision
2092
# which may be the sole parent if it is untouched.
2093
if ie.revision is not None:
2096
previous_entries = ie.find_previous_heads(
2098
self.repository.weave_store,
2099
self.repository.get_transaction())
2100
# we are creating a new revision for ie in the history store
2102
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2114
def _unescaper(match, _map=_unescape_map):
2115
code = match.group(1)
2119
if not code.startswith('#'):
2121
return unichr(int(code[1:])).encode('utf8')
2127
def _unescape_xml(data):
2128
"""Unescape predefined XML entities in a string of data."""
2130
if _unescape_re is None:
2131
_unescape_re = re.compile('\&([^;]*);')
2132
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
