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(), """
37
revision as _mod_revision,
42
from bzrlib.revisiontree import RevisionTree
43
from bzrlib.store.versioned import VersionedFileStore
44
from bzrlib.store.text import TextStore
45
from bzrlib.testament import Testament
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
56
from bzrlib.trace import mutter, note, warning
59
# Old formats display a warning, but only once
60
_deprecation_warning_done = False
63
######################################################################
66
class Repository(object):
67
"""Repository holding history for one or more branches.
69
The repository holds and retrieves historical information including
70
revisions and file history. It's normally accessed only by the Branch,
71
which views a particular line of development through that history.
73
The Repository builds on top of Stores and a Transport, which respectively
74
describe the disk data format and the way of accessing the (possibly
78
_file_ids_altered_regex = lazy_regex.lazy_compile(
79
r'file_id="(?P<file_id>[^"]+)"'
80
r'.*revision="(?P<revision_id>[^"]+)"'
84
def add_inventory(self, revision_id, inv, parents):
85
"""Add the inventory inv to the repository as revision_id.
87
:param parents: The revision ids of the parents that revision_id
88
is known to have and are in the repository already.
90
returns the sha1 of the serialized inventory.
92
revision_id = osutils.safe_revision_id(revision_id)
93
_mod_revision.check_not_reserved_id(revision_id)
94
assert inv.revision_id is None or inv.revision_id == revision_id, \
95
"Mismatch between inventory revision" \
96
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
97
assert inv.root is not None
98
inv_text = self.serialise_inventory(inv)
99
inv_sha1 = osutils.sha_string(inv_text)
100
inv_vf = self.control_weaves.get_weave('inventory',
101
self.get_transaction())
102
self._inventory_add_lines(inv_vf, revision_id, parents,
103
osutils.split_lines(inv_text))
106
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
108
for parent in parents:
110
final_parents.append(parent)
112
inv_vf.add_lines(revision_id, final_parents, lines)
115
def add_revision(self, revision_id, rev, inv=None, config=None):
116
"""Add rev to the revision store as revision_id.
118
:param revision_id: the revision id to use.
119
:param rev: The revision object.
120
:param inv: The inventory for the revision. if None, it will be looked
121
up in the inventory storer
122
:param config: If None no digital signature will be created.
123
If supplied its signature_needed method will be used
124
to determine if a signature should be made.
126
revision_id = osutils.safe_revision_id(revision_id)
127
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
129
_mod_revision.check_not_reserved_id(revision_id)
130
if config is not None and config.signature_needed():
132
inv = self.get_inventory(revision_id)
133
plaintext = Testament(rev, inv).as_short_text()
134
self.store_revision_signature(
135
gpg.GPGStrategy(config), plaintext, revision_id)
136
if not revision_id in self.get_inventory_weave():
138
raise errors.WeaveRevisionNotPresent(revision_id,
139
self.get_inventory_weave())
141
# yes, this is not suitable for adding with ghosts.
142
self.add_inventory(revision_id, inv, rev.parent_ids)
143
self._revision_store.add_revision(rev, self.get_transaction())
146
def _all_possible_ids(self):
147
"""Return all the possible revisions that we could find."""
148
return self.get_inventory_weave().versions()
150
def all_revision_ids(self):
151
"""Returns a list of all the revision ids in the repository.
153
This is deprecated because code should generally work on the graph
154
reachable from a particular revision, and ignore any other revisions
155
that might be present. There is no direct replacement method.
157
return self._all_revision_ids()
160
def _all_revision_ids(self):
161
"""Returns a list of all the revision ids in the repository.
163
These are in as much topological order as the underlying store can
164
present: for weaves ghosts may lead to a lack of correctness until
165
the reweave updates the parents list.
167
if self._revision_store.text_store.listable():
168
return self._revision_store.all_revision_ids(self.get_transaction())
169
result = self._all_possible_ids()
170
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
171
# ids. (It should, since _revision_store's API should change to
172
# return utf8 revision_ids)
173
return self._eliminate_revisions_not_present(result)
175
def break_lock(self):
176
"""Break a lock if one is present from another instance.
178
Uses the ui factory to ask for confirmation if the lock may be from
181
self.control_files.break_lock()
184
def _eliminate_revisions_not_present(self, revision_ids):
185
"""Check every revision id in revision_ids to see if we have it.
187
Returns a set of the present revisions.
190
for id in revision_ids:
191
if self.has_revision(id):
196
def create(a_bzrdir):
197
"""Construct the current default format repository in a_bzrdir."""
198
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
200
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
201
"""instantiate a Repository.
203
:param _format: The format of the repository on disk.
204
:param a_bzrdir: The BzrDir of the repository.
206
In the future we will have a single api for all stores for
207
getting file texts, inventories and revisions, then
208
this construct will accept instances of those things.
210
super(Repository, self).__init__()
211
self._format = _format
212
# the following are part of the public API for Repository:
213
self.bzrdir = a_bzrdir
214
self.control_files = control_files
215
self._revision_store = _revision_store
216
self.text_store = text_store
217
# backwards compatibility
218
self.weave_store = text_store
219
# not right yet - should be more semantically clear ?
221
self.control_store = control_store
222
self.control_weaves = control_store
223
# TODO: make sure to construct the right store classes, etc, depending
224
# on whether escaping is required.
225
self._warn_if_deprecated()
228
return '%s(%r)' % (self.__class__.__name__,
229
self.bzrdir.transport.base)
232
return self.control_files.is_locked()
234
def lock_write(self):
235
self.control_files.lock_write()
238
self.control_files.lock_read()
240
def get_physical_lock_status(self):
241
return self.control_files.get_physical_lock_status()
244
def gather_stats(self, revid=None, committers=None):
245
"""Gather statistics from a revision id.
247
:param revid: The revision id to gather statistics from, if None, then
248
no revision specific statistics are gathered.
249
:param committers: Optional parameter controlling whether to grab
250
a count of committers from the revision specific statistics.
251
:return: A dictionary of statistics. Currently this contains:
252
committers: The number of committers if requested.
253
firstrev: A tuple with timestamp, timezone for the penultimate left
254
most ancestor of revid, if revid is not the NULL_REVISION.
255
latestrev: A tuple with timestamp, timezone for revid, if revid is
256
not the NULL_REVISION.
257
revisions: The total revision count in the repository.
258
size: An estimate disk size of the repository in bytes.
261
if revid and committers:
262
result['committers'] = 0
263
if revid and revid != _mod_revision.NULL_REVISION:
265
all_committers = set()
266
revisions = self.get_ancestry(revid)
267
# pop the leading None
269
first_revision = None
271
# ignore the revisions in the middle - just grab first and last
272
revisions = revisions[0], revisions[-1]
273
for revision in self.get_revisions(revisions):
274
if not first_revision:
275
first_revision = revision
277
all_committers.add(revision.committer)
278
last_revision = revision
280
result['committers'] = len(all_committers)
281
result['firstrev'] = (first_revision.timestamp,
282
first_revision.timezone)
283
result['latestrev'] = (last_revision.timestamp,
284
last_revision.timezone)
286
# now gather global repository information
287
if self.bzrdir.root_transport.listable():
288
c, t = self._revision_store.total_size(self.get_transaction())
289
result['revisions'] = c
294
def missing_revision_ids(self, other, revision_id=None):
295
"""Return the revision ids that other has that this does not.
297
These are returned in topological order.
299
revision_id: only return revision ids included by revision_id.
301
revision_id = osutils.safe_revision_id(revision_id)
302
return InterRepository.get(other, self).missing_revision_ids(revision_id)
306
"""Open the repository rooted at base.
308
For instance, if the repository is at URL/.bzr/repository,
309
Repository.open(URL) -> a Repository instance.
311
control = bzrdir.BzrDir.open(base)
312
return control.open_repository()
314
def copy_content_into(self, destination, revision_id=None, basis=None):
315
"""Make a complete copy of the content in self into destination.
317
This is a destructive operation! Do not use it on existing
320
revision_id = osutils.safe_revision_id(revision_id)
321
return InterRepository.get(self, destination).copy_content(revision_id, basis)
323
def fetch(self, source, revision_id=None, pb=None):
324
"""Fetch the content required to construct revision_id from source.
326
If revision_id is None all content is copied.
328
revision_id = osutils.safe_revision_id(revision_id)
329
return InterRepository.get(source, self).fetch(revision_id=revision_id,
332
def get_commit_builder(self, branch, parents, config, timestamp=None,
333
timezone=None, committer=None, revprops=None,
335
"""Obtain a CommitBuilder for this repository.
337
:param branch: Branch to commit to.
338
:param parents: Revision ids of the parents of the new revision.
339
:param config: Configuration to use.
340
:param timestamp: Optional timestamp recorded for commit.
341
:param timezone: Optional timezone for timestamp.
342
:param committer: Optional committer to set for commit.
343
:param revprops: Optional dictionary of revision properties.
344
:param revision_id: Optional revision id.
346
revision_id = osutils.safe_revision_id(revision_id)
347
return _CommitBuilder(self, parents, config, timestamp, timezone,
348
committer, revprops, revision_id)
351
self.control_files.unlock()
354
def clone(self, a_bzrdir, revision_id=None, basis=None):
355
"""Clone this repository into a_bzrdir using the current format.
357
Currently no check is made that the format of this repository and
358
the bzrdir format are compatible. FIXME RBC 20060201.
360
:return: The newly created destination repository.
362
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
363
# use target default format.
364
dest_repo = a_bzrdir.create_repository()
366
# Most control formats need the repository to be specifically
367
# created, but on some old all-in-one formats it's not needed
369
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
370
except errors.UninitializableFormat:
371
dest_repo = a_bzrdir.open_repository()
372
self.copy_content_into(dest_repo, revision_id, basis)
376
def has_revision(self, revision_id):
377
"""True if this repository has a copy of the revision."""
378
revision_id = osutils.safe_revision_id(revision_id)
379
return self._revision_store.has_revision_id(revision_id,
380
self.get_transaction())
383
def get_revision_reconcile(self, revision_id):
384
"""'reconcile' helper routine that allows access to a revision always.
386
This variant of get_revision does not cross check the weave graph
387
against the revision one as get_revision does: but it should only
388
be used by reconcile, or reconcile-alike commands that are correcting
389
or testing the revision graph.
391
if not revision_id or not isinstance(revision_id, basestring):
392
raise errors.InvalidRevisionId(revision_id=revision_id,
394
return self.get_revisions([revision_id])[0]
397
def get_revisions(self, revision_ids):
398
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
399
revs = self._revision_store.get_revisions(revision_ids,
400
self.get_transaction())
402
assert not isinstance(rev.revision_id, unicode)
403
for parent_id in rev.parent_ids:
404
assert not isinstance(parent_id, unicode)
408
def get_revision_xml(self, revision_id):
409
# TODO: jam 20070210 This shouldn't be necessary since get_revision
410
# would have already do it.
411
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
412
revision_id = osutils.safe_revision_id(revision_id)
413
rev = self.get_revision(revision_id)
415
# the current serializer..
416
self._revision_store._serializer.write_revision(rev, rev_tmp)
418
return rev_tmp.getvalue()
421
def get_revision(self, revision_id):
422
"""Return the Revision object for a named revision"""
423
# TODO: jam 20070210 get_revision_reconcile should do this for us
424
revision_id = osutils.safe_revision_id(revision_id)
425
r = self.get_revision_reconcile(revision_id)
426
# weave corruption can lead to absent revision markers that should be
428
# the following test is reasonably cheap (it needs a single weave read)
429
# and the weave is cached in read transactions. In write transactions
430
# it is not cached but typically we only read a small number of
431
# revisions. For knits when they are introduced we will probably want
432
# to ensure that caching write transactions are in use.
433
inv = self.get_inventory_weave()
434
self._check_revision_parents(r, inv)
438
def get_deltas_for_revisions(self, revisions):
439
"""Produce a generator of revision deltas.
441
Note that the input is a sequence of REVISIONS, not revision_ids.
442
Trees will be held in memory until the generator exits.
443
Each delta is relative to the revision's lefthand predecessor.
445
required_trees = set()
446
for revision in revisions:
447
required_trees.add(revision.revision_id)
448
required_trees.update(revision.parent_ids[:1])
449
trees = dict((t.get_revision_id(), t) for
450
t in self.revision_trees(required_trees))
451
for revision in revisions:
452
if not revision.parent_ids:
453
old_tree = self.revision_tree(None)
455
old_tree = trees[revision.parent_ids[0]]
456
yield trees[revision.revision_id].changes_from(old_tree)
459
def get_revision_delta(self, revision_id):
460
"""Return the delta for one revision.
462
The delta is relative to the left-hand predecessor of the
465
r = self.get_revision(revision_id)
466
return list(self.get_deltas_for_revisions([r]))[0]
468
def _check_revision_parents(self, revision, inventory):
469
"""Private to Repository and Fetch.
471
This checks the parentage of revision in an inventory weave for
472
consistency and is only applicable to inventory-weave-for-ancestry
473
using repository formats & fetchers.
475
weave_parents = inventory.get_parents(revision.revision_id)
476
weave_names = inventory.versions()
477
for parent_id in revision.parent_ids:
478
if parent_id in weave_names:
479
# this parent must not be a ghost.
480
if not parent_id in weave_parents:
482
raise errors.CorruptRepository(self)
485
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
486
revision_id = osutils.safe_revision_id(revision_id)
487
signature = gpg_strategy.sign(plaintext)
488
self._revision_store.add_revision_signature_text(revision_id,
490
self.get_transaction())
492
def fileids_altered_by_revision_ids(self, revision_ids):
493
"""Find the file ids and versions affected by revisions.
495
:param revisions: an iterable containing revision ids.
496
:return: a dictionary mapping altered file-ids to an iterable of
497
revision_ids. Each altered file-ids has the exact revision_ids that
498
altered it listed explicitly.
500
assert self._serializer.support_altered_by_hack, \
501
("fileids_altered_by_revision_ids only supported for branches "
502
"which store inventory as unnested xml, not on %r" % self)
503
selected_revision_ids = set(osutils.safe_revision_id(r)
504
for r in revision_ids)
505
w = self.get_inventory_weave()
508
# this code needs to read every new line in every inventory for the
509
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
510
# not present in one of those inventories is unnecessary but not
511
# harmful because we are filtering by the revision id marker in the
512
# inventory lines : we only select file ids altered in one of those
513
# revisions. We don't need to see all lines in the inventory because
514
# only those added in an inventory in rev X can contain a revision=X
516
unescape_revid_cache = {}
517
unescape_fileid_cache = {}
519
# jam 20061218 In a big fetch, this handles hundreds of thousands
520
# of lines, so it has had a lot of inlining and optimizing done.
521
# Sorry that it is a little bit messy.
522
# Move several functions to be local variables, since this is a long
524
search = self._file_ids_altered_regex.search
525
unescape = _unescape_xml
526
setdefault = result.setdefault
527
pb = ui.ui_factory.nested_progress_bar()
529
for line in w.iter_lines_added_or_present_in_versions(
530
selected_revision_ids, pb=pb):
534
# One call to match.group() returning multiple items is quite a
535
# bit faster than 2 calls to match.group() each returning 1
536
file_id, revision_id = match.group('file_id', 'revision_id')
538
# Inlining the cache lookups helps a lot when you make 170,000
539
# lines and 350k ids, versus 8.4 unique ids.
540
# Using a cache helps in 2 ways:
541
# 1) Avoids unnecessary decoding calls
542
# 2) Re-uses cached strings, which helps in future set and
544
# (2) is enough that removing encoding entirely along with
545
# the cache (so we are using plain strings) results in no
546
# performance improvement.
548
revision_id = unescape_revid_cache[revision_id]
550
unescaped = unescape(revision_id)
551
unescape_revid_cache[revision_id] = unescaped
552
revision_id = unescaped
554
if revision_id in selected_revision_ids:
556
file_id = unescape_fileid_cache[file_id]
558
unescaped = unescape(file_id)
559
unescape_fileid_cache[file_id] = unescaped
561
setdefault(file_id, set()).add(revision_id)
567
def get_inventory_weave(self):
568
return self.control_weaves.get_weave('inventory',
569
self.get_transaction())
572
def get_inventory(self, revision_id):
573
"""Get Inventory object by hash."""
574
# TODO: jam 20070210 Technically we don't need to sanitize, since all
575
# called functions must sanitize.
576
revision_id = osutils.safe_revision_id(revision_id)
577
return self.deserialise_inventory(
578
revision_id, self.get_inventory_xml(revision_id))
580
def deserialise_inventory(self, revision_id, xml):
581
"""Transform the xml into an inventory object.
583
:param revision_id: The expected revision id of the inventory.
584
:param xml: A serialised inventory.
586
revision_id = osutils.safe_revision_id(revision_id)
587
result = self._serializer.read_inventory_from_string(xml)
588
result.root.revision = revision_id
591
def serialise_inventory(self, inv):
592
return self._serializer.write_inventory_to_string(inv)
595
def get_inventory_xml(self, revision_id):
596
"""Get inventory XML as a file object."""
597
revision_id = osutils.safe_revision_id(revision_id)
599
assert isinstance(revision_id, str), type(revision_id)
600
iw = self.get_inventory_weave()
601
return iw.get_text(revision_id)
603
raise errors.HistoryMissing(self, 'inventory', revision_id)
606
def get_inventory_sha1(self, revision_id):
607
"""Return the sha1 hash of the inventory entry
609
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
610
revision_id = osutils.safe_revision_id(revision_id)
611
return self.get_revision(revision_id).inventory_sha1
614
def get_revision_graph(self, revision_id=None):
615
"""Return a dictionary containing the revision graph.
617
:param revision_id: The revision_id to get a graph from. If None, then
618
the entire revision graph is returned. This is a deprecated mode of
619
operation and will be removed in the future.
620
:return: a dictionary of revision_id->revision_parents_list.
622
# special case NULL_REVISION
623
if revision_id == _mod_revision.NULL_REVISION:
625
revision_id = osutils.safe_revision_id(revision_id)
626
a_weave = self.get_inventory_weave()
627
all_revisions = self._eliminate_revisions_not_present(
629
entire_graph = dict([(node, a_weave.get_parents(node)) for
630
node in all_revisions])
631
if revision_id is None:
633
elif revision_id not in entire_graph:
634
raise errors.NoSuchRevision(self, revision_id)
636
# add what can be reached from revision_id
638
pending = set([revision_id])
639
while len(pending) > 0:
641
result[node] = entire_graph[node]
642
for revision_id in result[node]:
643
if revision_id not in result:
644
pending.add(revision_id)
648
def get_revision_graph_with_ghosts(self, revision_ids=None):
649
"""Return a graph of the revisions with ghosts marked as applicable.
651
:param revision_ids: an iterable of revisions to graph or None for all.
652
:return: a Graph object with the graph reachable from revision_ids.
654
result = graph.Graph()
656
pending = set(self.all_revision_ids())
659
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
660
# special case NULL_REVISION
661
if _mod_revision.NULL_REVISION in pending:
662
pending.remove(_mod_revision.NULL_REVISION)
663
required = set(pending)
666
revision_id = pending.pop()
668
rev = self.get_revision(revision_id)
669
except errors.NoSuchRevision:
670
if revision_id in required:
673
result.add_ghost(revision_id)
675
for parent_id in rev.parent_ids:
676
# is this queued or done ?
677
if (parent_id not in pending and
678
parent_id not in done):
680
pending.add(parent_id)
681
result.add_node(revision_id, rev.parent_ids)
682
done.add(revision_id)
685
def _get_history_vf(self):
686
"""Get a versionedfile whose history graph reflects all revisions.
688
For weave repositories, this is the inventory weave.
690
return self.get_inventory_weave()
692
def iter_reverse_revision_history(self, revision_id):
693
"""Iterate backwards through revision ids in the lefthand history
695
:param revision_id: The revision id to start with. All its lefthand
696
ancestors will be traversed.
698
revision_id = osutils.safe_revision_id(revision_id)
699
if revision_id in (None, _mod_revision.NULL_REVISION):
701
next_id = revision_id
702
versionedfile = self._get_history_vf()
705
parents = versionedfile.get_parents(next_id)
706
if len(parents) == 0:
712
def get_revision_inventory(self, revision_id):
713
"""Return inventory of a past revision."""
714
# TODO: Unify this with get_inventory()
715
# bzr 0.0.6 and later imposes the constraint that the inventory_id
716
# must be the same as its revision, so this is trivial.
717
if revision_id is None:
718
# This does not make sense: if there is no revision,
719
# then it is the current tree inventory surely ?!
720
# and thus get_root_id() is something that looks at the last
721
# commit on the branch, and the get_root_id is an inventory check.
722
raise NotImplementedError
723
# return Inventory(self.get_root_id())
725
return self.get_inventory(revision_id)
729
"""Return True if this repository is flagged as a shared repository."""
730
raise NotImplementedError(self.is_shared)
733
def reconcile(self, other=None, thorough=False):
734
"""Reconcile this repository."""
735
from bzrlib.reconcile import RepoReconciler
736
reconciler = RepoReconciler(self, thorough=thorough)
737
reconciler.reconcile()
741
def revision_tree(self, revision_id):
742
"""Return Tree for a revision on this branch.
744
`revision_id` may be None for the empty tree revision.
746
# TODO: refactor this to use an existing revision object
747
# so we don't need to read it in twice.
748
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
749
return RevisionTree(self, Inventory(root_id=None),
750
_mod_revision.NULL_REVISION)
752
revision_id = osutils.safe_revision_id(revision_id)
753
inv = self.get_revision_inventory(revision_id)
754
return RevisionTree(self, inv, revision_id)
757
def revision_trees(self, revision_ids):
758
"""Return Tree for a revision on this branch.
760
`revision_id` may not be None or 'null:'"""
761
assert None not in revision_ids
762
assert _mod_revision.NULL_REVISION not in revision_ids
763
texts = self.get_inventory_weave().get_texts(revision_ids)
764
for text, revision_id in zip(texts, revision_ids):
765
inv = self.deserialise_inventory(revision_id, text)
766
yield RevisionTree(self, inv, revision_id)
769
def get_ancestry(self, revision_id):
770
"""Return a list of revision-ids integrated by a revision.
772
The first element of the list is always None, indicating the origin
773
revision. This might change when we have history horizons, or
774
perhaps we should have a new API.
776
This is topologically sorted.
778
if revision_id is None:
780
revision_id = osutils.safe_revision_id(revision_id)
781
if not self.has_revision(revision_id):
782
raise errors.NoSuchRevision(self, revision_id)
783
w = self.get_inventory_weave()
784
candidates = w.get_ancestry(revision_id)
785
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
788
def print_file(self, file, revision_id):
789
"""Print `file` to stdout.
791
FIXME RBC 20060125 as John Meinel points out this is a bad api
792
- it writes to stdout, it assumes that that is valid etc. Fix
793
by creating a new more flexible convenience function.
795
revision_id = osutils.safe_revision_id(revision_id)
796
tree = self.revision_tree(revision_id)
797
# use inventory as it was in that revision
798
file_id = tree.inventory.path2id(file)
800
# TODO: jam 20060427 Write a test for this code path
801
# it had a bug in it, and was raising the wrong
803
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
804
tree.print_file(file_id)
806
def get_transaction(self):
807
return self.control_files.get_transaction()
809
def revision_parents(self, revision_id):
810
revision_id = osutils.safe_revision_id(revision_id)
811
return self.get_inventory_weave().parent_names(revision_id)
814
def set_make_working_trees(self, new_value):
815
"""Set the policy flag for making working trees when creating branches.
817
This only applies to branches that use this repository.
819
The default is 'True'.
820
:param new_value: True to restore the default, False to disable making
823
raise NotImplementedError(self.set_make_working_trees)
825
def make_working_trees(self):
826
"""Returns the policy for making working trees on new branches."""
827
raise NotImplementedError(self.make_working_trees)
830
def sign_revision(self, revision_id, gpg_strategy):
831
revision_id = osutils.safe_revision_id(revision_id)
832
plaintext = Testament.from_revision(self, revision_id).as_short_text()
833
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
836
def has_signature_for_revision_id(self, revision_id):
837
"""Query for a revision signature for revision_id in the repository."""
838
revision_id = osutils.safe_revision_id(revision_id)
839
return self._revision_store.has_signature(revision_id,
840
self.get_transaction())
843
def get_signature_text(self, revision_id):
844
"""Return the text for a signature."""
845
revision_id = osutils.safe_revision_id(revision_id)
846
return self._revision_store.get_signature_text(revision_id,
847
self.get_transaction())
850
def check(self, revision_ids):
851
"""Check consistency of all history of given revision_ids.
853
Different repository implementations should override _check().
855
:param revision_ids: A non-empty list of revision_ids whose ancestry
856
will be checked. Typically the last revision_id of a branch.
859
raise ValueError("revision_ids must be non-empty in %s.check"
861
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
862
return self._check(revision_ids)
864
def _check(self, revision_ids):
865
result = check.Check(self)
869
def _warn_if_deprecated(self):
870
global _deprecation_warning_done
871
if _deprecation_warning_done:
873
_deprecation_warning_done = True
874
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
875
% (self._format, self.bzrdir.transport.base))
877
def supports_rich_root(self):
878
return self._format.rich_root_data
880
def _check_ascii_revisionid(self, revision_id, method):
881
"""Private helper for ascii-only repositories."""
882
# weave repositories refuse to store revisionids that are non-ascii.
883
if revision_id is not None:
884
# weaves require ascii revision ids.
885
if isinstance(revision_id, unicode):
887
revision_id.encode('ascii')
888
except UnicodeEncodeError:
889
raise errors.NonAsciiRevisionId(method, self)
892
revision_id.decode('ascii')
893
except UnicodeDecodeError:
894
raise errors.NonAsciiRevisionId(method, self)
898
# remove these delegates a while after bzr 0.15
899
def __make_delegated(name, from_module):
900
def _deprecated_repository_forwarder():
901
symbol_versioning.warn('%s moved to %s in bzr 0.15'
902
% (name, from_module),
905
m = __import__(from_module, globals(), locals(), [name])
907
return getattr(m, name)
908
except AttributeError:
909
raise AttributeError('module %s has no name %s'
911
globals()[name] = _deprecated_repository_forwarder
914
'AllInOneRepository',
915
'WeaveMetaDirRepository',
916
'PreSplitOutRepositoryFormat',
922
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
927
'RepositoryFormatKnit',
928
'RepositoryFormatKnit1',
929
'RepositoryFormatKnit2',
931
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
934
def install_revision(repository, rev, revision_tree):
935
"""Install all revision data into a repository."""
938
for p_id in rev.parent_ids:
939
if repository.has_revision(p_id):
940
present_parents.append(p_id)
941
parent_trees[p_id] = repository.revision_tree(p_id)
943
parent_trees[p_id] = repository.revision_tree(None)
945
inv = revision_tree.inventory
946
entries = inv.iter_entries()
947
# backwards compatability hack: skip the root id.
948
if not repository.supports_rich_root():
949
path, root = entries.next()
950
if root.revision != rev.revision_id:
951
raise errors.IncompatibleRevision(repr(repository))
952
# Add the texts that are not already present
953
for path, ie in entries:
954
w = repository.weave_store.get_weave_or_empty(ie.file_id,
955
repository.get_transaction())
956
if ie.revision not in w:
958
# FIXME: TODO: The following loop *may* be overlapping/duplicate
959
# with InventoryEntry.find_previous_heads(). if it is, then there
960
# is a latent bug here where the parents may have ancestors of each
962
for revision, tree in parent_trees.iteritems():
963
if ie.file_id not in tree:
965
parent_id = tree.inventory[ie.file_id].revision
966
if parent_id in text_parents:
968
text_parents.append(parent_id)
970
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
971
repository.get_transaction())
972
lines = revision_tree.get_file(ie.file_id).readlines()
973
vfile.add_lines(rev.revision_id, text_parents, lines)
975
# install the inventory
976
repository.add_inventory(rev.revision_id, inv, present_parents)
977
except errors.RevisionAlreadyPresent:
979
repository.add_revision(rev.revision_id, rev, inv)
982
class MetaDirRepository(Repository):
983
"""Repositories in the new meta-dir layout."""
985
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
986
super(MetaDirRepository, self).__init__(_format,
992
dir_mode = self.control_files._dir_mode
993
file_mode = self.control_files._file_mode
997
"""Return True if this repository is flagged as a shared repository."""
998
return self.control_files._transport.has('shared-storage')
1001
def set_make_working_trees(self, new_value):
1002
"""Set the policy flag for making working trees when creating branches.
1004
This only applies to branches that use this repository.
1006
The default is 'True'.
1007
:param new_value: True to restore the default, False to disable making
1012
self.control_files._transport.delete('no-working-trees')
1013
except errors.NoSuchFile:
1016
self.control_files.put_utf8('no-working-trees', '')
1018
def make_working_trees(self):
1019
"""Returns the policy for making working trees on new branches."""
1020
return not self.control_files._transport.has('no-working-trees')
1023
class RepositoryFormatRegistry(registry.Registry):
1024
"""Registry of RepositoryFormats.
1027
def get(self, format_string):
1028
r = registry.Registry.get(self, format_string)
1034
format_registry = RepositoryFormatRegistry()
1035
"""Registry of formats, indexed by their identifying format string.
1037
This can contain either format instances themselves, or classes/factories that
1038
can be called to obtain one.
1042
#####################################################################
1043
# Repository Formats
1045
class RepositoryFormat(object):
1046
"""A repository format.
1048
Formats provide three things:
1049
* An initialization routine to construct repository data on disk.
1050
* a format string which is used when the BzrDir supports versioned
1052
* an open routine which returns a Repository instance.
1054
Formats are placed in an dict by their format string for reference
1055
during opening. These should be subclasses of RepositoryFormat
1058
Once a format is deprecated, just deprecate the initialize and open
1059
methods on the format class. Do not deprecate the object, as the
1060
object will be created every system load.
1062
Common instance attributes:
1063
_matchingbzrdir - the bzrdir format that the repository format was
1064
originally written to work with. This can be used if manually
1065
constructing a bzrdir and repository, or more commonly for test suite
1070
return "<%s>" % self.__class__.__name__
1072
def __eq__(self, other):
1073
# format objects are generally stateless
1074
return isinstance(other, self.__class__)
1076
def __ne__(self, other):
1077
return not self == other
1080
def find_format(klass, a_bzrdir):
1081
"""Return the format for the repository object in a_bzrdir.
1083
This is used by bzr native formats that have a "format" file in
1084
the repository. Other methods may be used by different types of
1088
transport = a_bzrdir.get_repository_transport(None)
1089
format_string = transport.get("format").read()
1090
return format_registry.get(format_string)
1091
except errors.NoSuchFile:
1092
raise errors.NoRepositoryPresent(a_bzrdir)
1094
raise errors.UnknownFormatError(format=format_string)
1097
def register_format(klass, format):
1098
format_registry.register(format.get_format_string(), format)
1101
def unregister_format(klass, format):
1102
format_registry.remove(format.get_format_string())
1105
def get_default_format(klass):
1106
"""Return the current default format."""
1107
from bzrlib import bzrdir
1108
return bzrdir.format_registry.make_bzrdir('default').repository_format
1110
def _get_control_store(self, repo_transport, control_files):
1111
"""Return the control store for this repository."""
1112
raise NotImplementedError(self._get_control_store)
1114
def get_format_string(self):
1115
"""Return the ASCII format string that identifies this format.
1117
Note that in pre format ?? repositories the format string is
1118
not permitted nor written to disk.
1120
raise NotImplementedError(self.get_format_string)
1122
def get_format_description(self):
1123
"""Return the short description for this format."""
1124
raise NotImplementedError(self.get_format_description)
1126
def _get_revision_store(self, repo_transport, control_files):
1127
"""Return the revision store object for this a_bzrdir."""
1128
raise NotImplementedError(self._get_revision_store)
1130
def _get_text_rev_store(self,
1137
"""Common logic for getting a revision store for a repository.
1139
see self._get_revision_store for the subclass-overridable method to
1140
get the store for a repository.
1142
from bzrlib.store.revision.text import TextRevisionStore
1143
dir_mode = control_files._dir_mode
1144
file_mode = control_files._file_mode
1145
text_store = TextStore(transport.clone(name),
1147
compressed=compressed,
1149
file_mode=file_mode)
1150
_revision_store = TextRevisionStore(text_store, serializer)
1151
return _revision_store
1153
# TODO: this shouldn't be in the base class, it's specific to things that
1154
# use weaves or knits -- mbp 20070207
1155
def _get_versioned_file_store(self,
1160
versionedfile_class=None,
1161
versionedfile_kwargs={},
1163
if versionedfile_class is None:
1164
versionedfile_class = self._versionedfile_class
1165
weave_transport = control_files._transport.clone(name)
1166
dir_mode = control_files._dir_mode
1167
file_mode = control_files._file_mode
1168
return VersionedFileStore(weave_transport, prefixed=prefixed,
1170
file_mode=file_mode,
1171
versionedfile_class=versionedfile_class,
1172
versionedfile_kwargs=versionedfile_kwargs,
1175
def initialize(self, a_bzrdir, shared=False):
1176
"""Initialize a repository of this format in a_bzrdir.
1178
:param a_bzrdir: The bzrdir to put the new repository in it.
1179
:param shared: The repository should be initialized as a sharable one.
1181
This may raise UninitializableFormat if shared repository are not
1182
compatible the a_bzrdir.
1185
def is_supported(self):
1186
"""Is this format supported?
1188
Supported formats must be initializable and openable.
1189
Unsupported formats may not support initialization or committing or
1190
some other features depending on the reason for not being supported.
1194
def check_conversion_target(self, target_format):
1195
raise NotImplementedError(self.check_conversion_target)
1197
def open(self, a_bzrdir, _found=False):
1198
"""Return an instance of this format for the bzrdir a_bzrdir.
1200
_found is a private parameter, do not use it.
1202
raise NotImplementedError(self.open)
1205
class MetaDirRepositoryFormat(RepositoryFormat):
1206
"""Common base class for the new repositories using the metadir layout."""
1208
rich_root_data = False
1209
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1212
super(MetaDirRepositoryFormat, self).__init__()
1214
def _create_control_files(self, a_bzrdir):
1215
"""Create the required files and the initial control_files object."""
1216
# FIXME: RBC 20060125 don't peek under the covers
1217
# NB: no need to escape relative paths that are url safe.
1218
repository_transport = a_bzrdir.get_repository_transport(self)
1219
control_files = lockable_files.LockableFiles(repository_transport,
1220
'lock', lockdir.LockDir)
1221
control_files.create_lock()
1222
return control_files
1224
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1225
"""Upload the initial blank content."""
1226
control_files = self._create_control_files(a_bzrdir)
1227
control_files.lock_write()
1229
control_files._transport.mkdir_multi(dirs,
1230
mode=control_files._dir_mode)
1231
for file, content in files:
1232
control_files.put(file, content)
1233
for file, content in utf8_files:
1234
control_files.put_utf8(file, content)
1236
control_files.put_utf8('shared-storage', '')
1238
control_files.unlock()
1241
# formats which have no format string are not discoverable
1242
# and not independently creatable, so are not registered. They're
1243
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1244
# needed, it's constructed directly by the BzrDir. Non-native formats where
1245
# the repository is not separately opened are similar.
1247
format_registry.register_lazy(
1248
'Bazaar-NG Repository format 7',
1249
'bzrlib.repofmt.weaverepo',
1252
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1253
# default control directory format
1255
format_registry.register_lazy(
1256
'Bazaar-NG Knit Repository Format 1',
1257
'bzrlib.repofmt.knitrepo',
1258
'RepositoryFormatKnit1',
1260
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1262
format_registry.register_lazy(
1263
'Bazaar Knit Repository Format 2\n',
1264
'bzrlib.repofmt.knitrepo',
1265
'RepositoryFormatKnit2',
1268
format_registry.register_lazy(
1269
'Bazaar Knit Repository Format 3\n',
1270
'bzrlib.repofmt.knitrepo',
1271
'RepositoryFormatKnit3',
1275
class InterRepository(InterObject):
1276
"""This class represents operations taking place between two repositories.
1278
Its instances have methods like copy_content and fetch, and contain
1279
references to the source and target repositories these operations can be
1282
Often we will provide convenience methods on 'repository' which carry out
1283
operations with another repository - they will always forward to
1284
InterRepository.get(other).method_name(parameters).
1288
"""The available optimised InterRepository types."""
1290
def copy_content(self, revision_id=None, basis=None):
1291
raise NotImplementedError(self.copy_content)
1293
def fetch(self, revision_id=None, pb=None):
1294
"""Fetch the content required to construct revision_id.
1296
The content is copied from self.source to self.target.
1298
:param revision_id: if None all content is copied, if NULL_REVISION no
1300
:param pb: optional progress bar to use for progress reports. If not
1301
provided a default one will be created.
1303
Returns the copied revision count and the failed revisions in a tuple:
1306
raise NotImplementedError(self.fetch)
1309
def missing_revision_ids(self, revision_id=None):
1310
"""Return the revision ids that source has that target does not.
1312
These are returned in topological order.
1314
:param revision_id: only return revision ids included by this
1317
# generic, possibly worst case, slow code path.
1318
target_ids = set(self.target.all_revision_ids())
1319
if revision_id is not None:
1320
# TODO: jam 20070210 InterRepository is internal enough that it
1321
# should assume revision_ids are already utf-8
1322
revision_id = osutils.safe_revision_id(revision_id)
1323
source_ids = self.source.get_ancestry(revision_id)
1324
assert source_ids[0] is None
1327
source_ids = self.source.all_revision_ids()
1328
result_set = set(source_ids).difference(target_ids)
1329
# this may look like a no-op: its not. It preserves the ordering
1330
# other_ids had while only returning the members from other_ids
1331
# that we've decided we need.
1332
return [rev_id for rev_id in source_ids if rev_id in result_set]
1335
class InterSameDataRepository(InterRepository):
1336
"""Code for converting between repositories that represent the same data.
1338
Data format and model must match for this to work.
1342
def _get_repo_format_to_test(self):
1343
"""Repository format for testing with."""
1344
return RepositoryFormat.get_default_format()
1347
def is_compatible(source, target):
1348
if not isinstance(source, Repository):
1350
if not isinstance(target, Repository):
1352
if source._format.rich_root_data != target._format.rich_root_data:
1354
if source._serializer != target._serializer:
1360
def copy_content(self, revision_id=None, basis=None):
1361
"""Make a complete copy of the content in self into destination.
1363
This is a destructive operation! Do not use it on existing
1366
:param revision_id: Only copy the content needed to construct
1367
revision_id and its parents.
1368
:param basis: Copy the needed data preferentially from basis.
1371
self.target.set_make_working_trees(self.source.make_working_trees())
1372
except NotImplementedError:
1374
# TODO: jam 20070210 This is fairly internal, so we should probably
1375
# just assert that revision_id is not unicode.
1376
revision_id = osutils.safe_revision_id(revision_id)
1377
# grab the basis available data
1378
if basis is not None:
1379
self.target.fetch(basis, revision_id=revision_id)
1380
# but don't bother fetching if we have the needed data now.
1381
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1382
self.target.has_revision(revision_id)):
1384
self.target.fetch(self.source, revision_id=revision_id)
1387
def fetch(self, revision_id=None, pb=None):
1388
"""See InterRepository.fetch()."""
1389
from bzrlib.fetch import GenericRepoFetcher
1390
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1391
self.source, self.source._format, self.target,
1392
self.target._format)
1393
# TODO: jam 20070210 This should be an assert, not a translate
1394
revision_id = osutils.safe_revision_id(revision_id)
1395
f = GenericRepoFetcher(to_repository=self.target,
1396
from_repository=self.source,
1397
last_revision=revision_id,
1399
return f.count_copied, f.failed_revisions
1402
class InterWeaveRepo(InterSameDataRepository):
1403
"""Optimised code paths between Weave based repositories."""
1406
def _get_repo_format_to_test(self):
1407
from bzrlib.repofmt import weaverepo
1408
return weaverepo.RepositoryFormat7()
1411
def is_compatible(source, target):
1412
"""Be compatible with known Weave formats.
1414
We don't test for the stores being of specific types because that
1415
could lead to confusing results, and there is no need to be
1418
from bzrlib.repofmt.weaverepo import (
1424
return (isinstance(source._format, (RepositoryFormat5,
1426
RepositoryFormat7)) and
1427
isinstance(target._format, (RepositoryFormat5,
1429
RepositoryFormat7)))
1430
except AttributeError:
1434
def copy_content(self, revision_id=None, basis=None):
1435
"""See InterRepository.copy_content()."""
1436
# weave specific optimised path:
1437
# TODO: jam 20070210 Internal, should be an assert, not translate
1438
revision_id = osutils.safe_revision_id(revision_id)
1439
if basis is not None:
1440
# copy the basis in, then fetch remaining data.
1441
basis.copy_content_into(self.target, revision_id)
1442
# the basis copy_content_into could miss-set this.
1444
self.target.set_make_working_trees(self.source.make_working_trees())
1445
except NotImplementedError:
1447
self.target.fetch(self.source, revision_id=revision_id)
1450
self.target.set_make_working_trees(self.source.make_working_trees())
1451
except NotImplementedError:
1453
# FIXME do not peek!
1454
if self.source.control_files._transport.listable():
1455
pb = ui.ui_factory.nested_progress_bar()
1457
self.target.weave_store.copy_all_ids(
1458
self.source.weave_store,
1460
from_transaction=self.source.get_transaction(),
1461
to_transaction=self.target.get_transaction())
1462
pb.update('copying inventory', 0, 1)
1463
self.target.control_weaves.copy_multi(
1464
self.source.control_weaves, ['inventory'],
1465
from_transaction=self.source.get_transaction(),
1466
to_transaction=self.target.get_transaction())
1467
self.target._revision_store.text_store.copy_all_ids(
1468
self.source._revision_store.text_store,
1473
self.target.fetch(self.source, revision_id=revision_id)
1476
def fetch(self, revision_id=None, pb=None):
1477
"""See InterRepository.fetch()."""
1478
from bzrlib.fetch import GenericRepoFetcher
1479
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1480
self.source, self.source._format, self.target, self.target._format)
1481
# TODO: jam 20070210 This should be an assert, not a translate
1482
revision_id = osutils.safe_revision_id(revision_id)
1483
f = GenericRepoFetcher(to_repository=self.target,
1484
from_repository=self.source,
1485
last_revision=revision_id,
1487
return f.count_copied, f.failed_revisions
1490
def missing_revision_ids(self, revision_id=None):
1491
"""See InterRepository.missing_revision_ids()."""
1492
# we want all revisions to satisfy revision_id in source.
1493
# but we don't want to stat every file here and there.
1494
# we want then, all revisions other needs to satisfy revision_id
1495
# checked, but not those that we have locally.
1496
# so the first thing is to get a subset of the revisions to
1497
# satisfy revision_id in source, and then eliminate those that
1498
# we do already have.
1499
# this is slow on high latency connection to self, but as as this
1500
# disk format scales terribly for push anyway due to rewriting
1501
# inventory.weave, this is considered acceptable.
1503
if revision_id is not None:
1504
source_ids = self.source.get_ancestry(revision_id)
1505
assert source_ids[0] is None
1508
source_ids = self.source._all_possible_ids()
1509
source_ids_set = set(source_ids)
1510
# source_ids is the worst possible case we may need to pull.
1511
# now we want to filter source_ids against what we actually
1512
# have in target, but don't try to check for existence where we know
1513
# we do not have a revision as that would be pointless.
1514
target_ids = set(self.target._all_possible_ids())
1515
possibly_present_revisions = target_ids.intersection(source_ids_set)
1516
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1517
required_revisions = source_ids_set.difference(actually_present_revisions)
1518
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1519
if revision_id is not None:
1520
# we used get_ancestry to determine source_ids then we are assured all
1521
# revisions referenced are present as they are installed in topological order.
1522
# and the tip revision was validated by get_ancestry.
1523
return required_topo_revisions
1525
# if we just grabbed the possibly available ids, then
1526
# we only have an estimate of whats available and need to validate
1527
# that against the revision records.
1528
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1531
class InterKnitRepo(InterSameDataRepository):
1532
"""Optimised code paths between Knit based repositories."""
1535
def _get_repo_format_to_test(self):
1536
from bzrlib.repofmt import knitrepo
1537
return knitrepo.RepositoryFormatKnit1()
1540
def is_compatible(source, target):
1541
"""Be compatible with known Knit formats.
1543
We don't test for the stores being of specific types because that
1544
could lead to confusing results, and there is no need to be
1547
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1549
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1550
isinstance(target._format, (RepositoryFormatKnit1)))
1551
except AttributeError:
1555
def fetch(self, revision_id=None, pb=None):
1556
"""See InterRepository.fetch()."""
1557
from bzrlib.fetch import KnitRepoFetcher
1558
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1559
self.source, self.source._format, self.target, self.target._format)
1560
# TODO: jam 20070210 This should be an assert, not a translate
1561
revision_id = osutils.safe_revision_id(revision_id)
1562
f = KnitRepoFetcher(to_repository=self.target,
1563
from_repository=self.source,
1564
last_revision=revision_id,
1566
return f.count_copied, f.failed_revisions
1569
def missing_revision_ids(self, revision_id=None):
1570
"""See InterRepository.missing_revision_ids()."""
1571
if revision_id is not None:
1572
source_ids = self.source.get_ancestry(revision_id)
1573
assert source_ids[0] is None
1576
source_ids = self.source._all_possible_ids()
1577
source_ids_set = set(source_ids)
1578
# source_ids is the worst possible case we may need to pull.
1579
# now we want to filter source_ids against what we actually
1580
# have in target, but don't try to check for existence where we know
1581
# we do not have a revision as that would be pointless.
1582
target_ids = set(self.target._all_possible_ids())
1583
possibly_present_revisions = target_ids.intersection(source_ids_set)
1584
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1585
required_revisions = source_ids_set.difference(actually_present_revisions)
1586
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1587
if revision_id is not None:
1588
# we used get_ancestry to determine source_ids then we are assured all
1589
# revisions referenced are present as they are installed in topological order.
1590
# and the tip revision was validated by get_ancestry.
1591
return required_topo_revisions
1593
# if we just grabbed the possibly available ids, then
1594
# we only have an estimate of whats available and need to validate
1595
# that against the revision records.
1596
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1599
class InterModel1and2(InterRepository):
1602
def _get_repo_format_to_test(self):
1606
def is_compatible(source, target):
1607
if not isinstance(source, Repository):
1609
if not isinstance(target, Repository):
1611
if not source._format.rich_root_data and target._format.rich_root_data:
1617
def fetch(self, revision_id=None, pb=None):
1618
"""See InterRepository.fetch()."""
1619
from bzrlib.fetch import Model1toKnit2Fetcher
1620
# TODO: jam 20070210 This should be an assert, not a translate
1621
revision_id = osutils.safe_revision_id(revision_id)
1622
f = Model1toKnit2Fetcher(to_repository=self.target,
1623
from_repository=self.source,
1624
last_revision=revision_id,
1626
return f.count_copied, f.failed_revisions
1629
def copy_content(self, revision_id=None, basis=None):
1630
"""Make a complete copy of the content in self into destination.
1632
This is a destructive operation! Do not use it on existing
1635
:param revision_id: Only copy the content needed to construct
1636
revision_id and its parents.
1637
:param basis: Copy the needed data preferentially from basis.
1640
self.target.set_make_working_trees(self.source.make_working_trees())
1641
except NotImplementedError:
1643
# TODO: jam 20070210 Internal, assert, don't translate
1644
revision_id = osutils.safe_revision_id(revision_id)
1645
# grab the basis available data
1646
if basis is not None:
1647
self.target.fetch(basis, revision_id=revision_id)
1648
# but don't bother fetching if we have the needed data now.
1649
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1650
self.target.has_revision(revision_id)):
1652
self.target.fetch(self.source, revision_id=revision_id)
1655
class InterKnit1and2(InterKnitRepo):
1658
def _get_repo_format_to_test(self):
1662
def is_compatible(source, target):
1663
"""Be compatible with Knit1 source and Knit2 target"""
1664
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit2
1666
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1667
RepositoryFormatKnit2
1668
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1669
isinstance(target._format, (RepositoryFormatKnit2)))
1670
except AttributeError:
1674
def fetch(self, revision_id=None, pb=None):
1675
"""See InterRepository.fetch()."""
1676
from bzrlib.fetch import Knit1to2Fetcher
1677
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1678
self.source, self.source._format, self.target,
1679
self.target._format)
1680
# TODO: jam 20070210 This should be an assert, not a translate
1681
revision_id = osutils.safe_revision_id(revision_id)
1682
f = Knit1to2Fetcher(to_repository=self.target,
1683
from_repository=self.source,
1684
last_revision=revision_id,
1686
return f.count_copied, f.failed_revisions
1689
InterRepository.register_optimiser(InterSameDataRepository)
1690
InterRepository.register_optimiser(InterWeaveRepo)
1691
InterRepository.register_optimiser(InterKnitRepo)
1692
InterRepository.register_optimiser(InterModel1and2)
1693
InterRepository.register_optimiser(InterKnit1and2)
1696
class RepositoryTestProviderAdapter(object):
1697
"""A tool to generate a suite testing multiple repository formats at once.
1699
This is done by copying the test once for each transport and injecting
1700
the transport_server, transport_readonly_server, and bzrdir_format and
1701
repository_format classes into each copy. Each copy is also given a new id()
1702
to make it easy to identify.
1705
def __init__(self, transport_server, transport_readonly_server, formats):
1706
self._transport_server = transport_server
1707
self._transport_readonly_server = transport_readonly_server
1708
self._formats = formats
1710
def adapt(self, test):
1711
result = unittest.TestSuite()
1712
for repository_format, bzrdir_format in self._formats:
1713
from copy import deepcopy
1714
new_test = deepcopy(test)
1715
new_test.transport_server = self._transport_server
1716
new_test.transport_readonly_server = self._transport_readonly_server
1717
new_test.bzrdir_format = bzrdir_format
1718
new_test.repository_format = repository_format
1719
def make_new_test_id():
1720
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1721
return lambda: new_id
1722
new_test.id = make_new_test_id()
1723
result.addTest(new_test)
1727
class InterRepositoryTestProviderAdapter(object):
1728
"""A tool to generate a suite testing multiple inter repository formats.
1730
This is done by copying the test once for each interrepo provider and injecting
1731
the transport_server, transport_readonly_server, repository_format and
1732
repository_to_format classes into each copy.
1733
Each copy is also given a new id() to make it easy to identify.
1736
def __init__(self, transport_server, transport_readonly_server, formats):
1737
self._transport_server = transport_server
1738
self._transport_readonly_server = transport_readonly_server
1739
self._formats = formats
1741
def adapt(self, test):
1742
result = unittest.TestSuite()
1743
for interrepo_class, repository_format, repository_format_to in self._formats:
1744
from copy import deepcopy
1745
new_test = deepcopy(test)
1746
new_test.transport_server = self._transport_server
1747
new_test.transport_readonly_server = self._transport_readonly_server
1748
new_test.interrepo_class = interrepo_class
1749
new_test.repository_format = repository_format
1750
new_test.repository_format_to = repository_format_to
1751
def make_new_test_id():
1752
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1753
return lambda: new_id
1754
new_test.id = make_new_test_id()
1755
result.addTest(new_test)
1759
def default_test_list():
1760
"""Generate the default list of interrepo permutations to test."""
1761
from bzrlib.repofmt import knitrepo, weaverepo
1763
# test the default InterRepository between format 6 and the current
1765
# XXX: robertc 20060220 reinstate this when there are two supported
1766
# formats which do not have an optimal code path between them.
1767
#result.append((InterRepository,
1768
# RepositoryFormat6(),
1769
# RepositoryFormatKnit1()))
1770
for optimiser_class in InterRepository._optimisers:
1771
format_to_test = optimiser_class._get_repo_format_to_test()
1772
if format_to_test is not None:
1773
result.append((optimiser_class,
1774
format_to_test, format_to_test))
1775
# if there are specific combinations we want to use, we can add them
1777
result.append((InterModel1and2,
1778
weaverepo.RepositoryFormat5(),
1779
knitrepo.RepositoryFormatKnit2()))
1780
result.append((InterKnit1and2,
1781
knitrepo.RepositoryFormatKnit1(),
1782
knitrepo.RepositoryFormatKnit2()))
1786
class CopyConverter(object):
1787
"""A repository conversion tool which just performs a copy of the content.
1789
This is slow but quite reliable.
1792
def __init__(self, target_format):
1793
"""Create a CopyConverter.
1795
:param target_format: The format the resulting repository should be.
1797
self.target_format = target_format
1799
def convert(self, repo, pb):
1800
"""Perform the conversion of to_convert, giving feedback via pb.
1802
:param to_convert: The disk object to convert.
1803
:param pb: a progress bar to use for progress information.
1808
# this is only useful with metadir layouts - separated repo content.
1809
# trigger an assertion if not such
1810
repo._format.get_format_string()
1811
self.repo_dir = repo.bzrdir
1812
self.step('Moving repository to repository.backup')
1813
self.repo_dir.transport.move('repository', 'repository.backup')
1814
backup_transport = self.repo_dir.transport.clone('repository.backup')
1815
repo._format.check_conversion_target(self.target_format)
1816
self.source_repo = repo._format.open(self.repo_dir,
1818
_override_transport=backup_transport)
1819
self.step('Creating new repository')
1820
converted = self.target_format.initialize(self.repo_dir,
1821
self.source_repo.is_shared())
1822
converted.lock_write()
1824
self.step('Copying content into repository.')
1825
self.source_repo.copy_content_into(converted)
1828
self.step('Deleting old repository content.')
1829
self.repo_dir.transport.delete_tree('repository.backup')
1830
self.pb.note('repository converted')
1832
def step(self, message):
1833
"""Update the pb by a step."""
1835
self.pb.update(message, self.count, self.total)
1838
class CommitBuilder(object):
1839
"""Provides an interface to build up a commit.
1841
This allows describing a tree to be committed without needing to
1842
know the internals of the format of the repository.
1845
record_root_entry = False
1846
def __init__(self, repository, parents, config, timestamp=None,
1847
timezone=None, committer=None, revprops=None,
1849
"""Initiate a CommitBuilder.
1851
:param repository: Repository to commit to.
1852
:param parents: Revision ids of the parents of the new revision.
1853
:param config: Configuration to use.
1854
:param timestamp: Optional timestamp recorded for commit.
1855
:param timezone: Optional timezone for timestamp.
1856
:param committer: Optional committer to set for commit.
1857
:param revprops: Optional dictionary of revision properties.
1858
:param revision_id: Optional revision id.
1860
self._config = config
1862
if committer is None:
1863
self._committer = self._config.username()
1865
assert isinstance(committer, basestring), type(committer)
1866
self._committer = committer
1868
self.new_inventory = Inventory(None)
1869
self._new_revision_id = osutils.safe_revision_id(revision_id)
1870
self.parents = parents
1871
self.repository = repository
1874
if revprops is not None:
1875
self._revprops.update(revprops)
1877
if timestamp is None:
1878
timestamp = time.time()
1879
# Restrict resolution to 1ms
1880
self._timestamp = round(timestamp, 3)
1882
if timezone is None:
1883
self._timezone = osutils.local_time_offset()
1885
self._timezone = int(timezone)
1887
self._generate_revision_if_needed()
1889
def commit(self, message):
1890
"""Make the actual commit.
1892
:return: The revision id of the recorded revision.
1894
rev = _mod_revision.Revision(
1895
timestamp=self._timestamp,
1896
timezone=self._timezone,
1897
committer=self._committer,
1899
inventory_sha1=self.inv_sha1,
1900
revision_id=self._new_revision_id,
1901
properties=self._revprops)
1902
rev.parent_ids = self.parents
1903
self.repository.add_revision(self._new_revision_id, rev,
1904
self.new_inventory, self._config)
1905
return self._new_revision_id
1907
def revision_tree(self):
1908
"""Return the tree that was just committed.
1910
After calling commit() this can be called to get a RevisionTree
1911
representing the newly committed tree. This is preferred to
1912
calling Repository.revision_tree() because that may require
1913
deserializing the inventory, while we already have a copy in
1916
return RevisionTree(self.repository, self.new_inventory,
1917
self._new_revision_id)
1919
def finish_inventory(self):
1920
"""Tell the builder that the inventory is finished."""
1921
if self.new_inventory.root is None:
1922
symbol_versioning.warn('Root entry should be supplied to'
1923
' record_entry_contents, as of bzr 0.10.',
1924
DeprecationWarning, stacklevel=2)
1925
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1926
self.new_inventory.revision_id = self._new_revision_id
1927
self.inv_sha1 = self.repository.add_inventory(
1928
self._new_revision_id,
1933
def _gen_revision_id(self):
1934
"""Return new revision-id."""
1935
return generate_ids.gen_revision_id(self._config.username(),
1938
def _generate_revision_if_needed(self):
1939
"""Create a revision id if None was supplied.
1941
If the repository can not support user-specified revision ids
1942
they should override this function and raise CannotSetRevisionId
1943
if _new_revision_id is not None.
1945
:raises: CannotSetRevisionId
1947
if self._new_revision_id is None:
1948
self._new_revision_id = self._gen_revision_id()
1950
def record_entry_contents(self, ie, parent_invs, path, tree):
1951
"""Record the content of ie from tree into the commit if needed.
1953
Side effect: sets ie.revision when unchanged
1955
:param ie: An inventory entry present in the commit.
1956
:param parent_invs: The inventories of the parent revisions of the
1958
:param path: The path the entry is at in the tree.
1959
:param tree: The tree which contains this entry and should be used to
1962
if self.new_inventory.root is None and ie.parent_id is not None:
1963
symbol_versioning.warn('Root entry should be supplied to'
1964
' record_entry_contents, as of bzr 0.10.',
1965
DeprecationWarning, stacklevel=2)
1966
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1968
self.new_inventory.add(ie)
1970
# ie.revision is always None if the InventoryEntry is considered
1971
# for committing. ie.snapshot will record the correct revision
1972
# which may be the sole parent if it is untouched.
1973
if ie.revision is not None:
1976
# In this revision format, root entries have no knit or weave
1977
if ie is self.new_inventory.root:
1978
# When serializing out to disk and back in
1979
# root.revision is always _new_revision_id
1980
ie.revision = self._new_revision_id
1982
previous_entries = ie.find_previous_heads(
1984
self.repository.weave_store,
1985
self.repository.get_transaction())
1986
# we are creating a new revision for ie in the history store
1988
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1990
def modified_directory(self, file_id, file_parents):
1991
"""Record the presence of a symbolic link.
1993
:param file_id: The file_id of the link to record.
1994
:param file_parents: The per-file parent revision ids.
1996
self._add_text_to_weave(file_id, [], file_parents.keys())
1998
def modified_reference(self, file_id, file_parents):
1999
"""Record the modification of a reference.
2001
:param file_id: The file_id of the link to record.
2002
:param file_parents: The per-file parent revision ids.
2004
self._add_text_to_weave(file_id, [], file_parents.keys())
2006
def modified_file_text(self, file_id, file_parents,
2007
get_content_byte_lines, text_sha1=None,
2009
"""Record the text of file file_id
2011
:param file_id: The file_id of the file to record the text of.
2012
:param file_parents: The per-file parent revision ids.
2013
:param get_content_byte_lines: A callable which will return the byte
2015
:param text_sha1: Optional SHA1 of the file contents.
2016
:param text_size: Optional size of the file contents.
2018
# mutter('storing text of file {%s} in revision {%s} into %r',
2019
# file_id, self._new_revision_id, self.repository.weave_store)
2020
# special case to avoid diffing on renames or
2022
if (len(file_parents) == 1
2023
and text_sha1 == file_parents.values()[0].text_sha1
2024
and text_size == file_parents.values()[0].text_size):
2025
previous_ie = file_parents.values()[0]
2026
versionedfile = self.repository.weave_store.get_weave(file_id,
2027
self.repository.get_transaction())
2028
versionedfile.clone_text(self._new_revision_id,
2029
previous_ie.revision, file_parents.keys())
2030
return text_sha1, text_size
2032
new_lines = get_content_byte_lines()
2033
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2034
# should return the SHA1 and size
2035
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2036
return osutils.sha_strings(new_lines), \
2037
sum(map(len, new_lines))
2039
def modified_link(self, file_id, file_parents, link_target):
2040
"""Record the presence of a symbolic link.
2042
:param file_id: The file_id of the link to record.
2043
:param file_parents: The per-file parent revision ids.
2044
:param link_target: Target location of this link.
2046
self._add_text_to_weave(file_id, [], file_parents.keys())
2048
def _add_text_to_weave(self, file_id, new_lines, parents):
2049
versionedfile = self.repository.weave_store.get_weave_or_empty(
2050
file_id, self.repository.get_transaction())
2051
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2052
versionedfile.clear_cache()
2055
class _CommitBuilder(CommitBuilder):
2056
"""Temporary class so old CommitBuilders are detected properly
2058
Note: CommitBuilder works whether or not root entry is recorded.
2061
record_root_entry = True
2064
class RootCommitBuilder(CommitBuilder):
2065
"""This commitbuilder actually records the root id"""
2067
record_root_entry = True
2069
def record_entry_contents(self, ie, parent_invs, path, tree):
2070
"""Record the content of ie from tree into the commit if needed.
2072
Side effect: sets ie.revision when unchanged
2074
:param ie: An inventory entry present in the commit.
2075
:param parent_invs: The inventories of the parent revisions of the
2077
:param path: The path the entry is at in the tree.
2078
:param tree: The tree which contains this entry and should be used to
2081
assert self.new_inventory.root is not None or ie.parent_id is None
2082
self.new_inventory.add(ie)
2084
# ie.revision is always None if the InventoryEntry is considered
2085
# for committing. ie.snapshot will record the correct revision
2086
# which may be the sole parent if it is untouched.
2087
if ie.revision is not None:
2090
previous_entries = ie.find_previous_heads(
2092
self.repository.weave_store,
2093
self.repository.get_transaction())
2094
# we are creating a new revision for ie in the history store
2096
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2108
def _unescaper(match, _map=_unescape_map):
2109
code = match.group(1)
2113
if not code.startswith('#'):
2115
return unichr(int(code[1:])).encode('utf8')
2121
def _unescape_xml(data):
2122
"""Unescape predefined XML entities in a string of data."""
2124
if _unescape_re is None:
2125
_unescape_re = re.compile('\&([^;]*);')
2126
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
