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
inter = InterRepository.get(source, self)
331
return inter.fetch(revision_id=revision_id, pb=pb)
332
except NotImplementedError:
333
raise errors.IncompatibleRepositories(source, self)
335
def get_commit_builder(self, branch, parents, config, timestamp=None,
336
timezone=None, committer=None, revprops=None,
338
"""Obtain a CommitBuilder for this repository.
340
:param branch: Branch to commit to.
341
:param parents: Revision ids of the parents of the new revision.
342
:param config: Configuration to use.
343
:param timestamp: Optional timestamp recorded for commit.
344
:param timezone: Optional timezone for timestamp.
345
:param committer: Optional committer to set for commit.
346
:param revprops: Optional dictionary of revision properties.
347
:param revision_id: Optional revision id.
349
revision_id = osutils.safe_revision_id(revision_id)
350
return _CommitBuilder(self, parents, config, timestamp, timezone,
351
committer, revprops, revision_id)
354
self.control_files.unlock()
357
def clone(self, a_bzrdir, revision_id=None, basis=None):
358
"""Clone this repository into a_bzrdir using the current format.
360
Currently no check is made that the format of this repository and
361
the bzrdir format are compatible. FIXME RBC 20060201.
363
:return: The newly created destination repository.
365
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
366
# use target default format.
367
dest_repo = a_bzrdir.create_repository()
369
# Most control formats need the repository to be specifically
370
# created, but on some old all-in-one formats it's not needed
372
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
373
except errors.UninitializableFormat:
374
dest_repo = a_bzrdir.open_repository()
375
self.copy_content_into(dest_repo, revision_id, basis)
379
def has_revision(self, revision_id):
380
"""True if this repository has a copy of the revision."""
381
revision_id = osutils.safe_revision_id(revision_id)
382
return self._revision_store.has_revision_id(revision_id,
383
self.get_transaction())
386
def get_revision_reconcile(self, revision_id):
387
"""'reconcile' helper routine that allows access to a revision always.
389
This variant of get_revision does not cross check the weave graph
390
against the revision one as get_revision does: but it should only
391
be used by reconcile, or reconcile-alike commands that are correcting
392
or testing the revision graph.
394
if not revision_id or not isinstance(revision_id, basestring):
395
raise errors.InvalidRevisionId(revision_id=revision_id,
397
return self.get_revisions([revision_id])[0]
400
def get_revisions(self, revision_ids):
401
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
402
revs = self._revision_store.get_revisions(revision_ids,
403
self.get_transaction())
405
assert not isinstance(rev.revision_id, unicode)
406
for parent_id in rev.parent_ids:
407
assert not isinstance(parent_id, unicode)
411
def get_revision_xml(self, revision_id):
412
# TODO: jam 20070210 This shouldn't be necessary since get_revision
413
# would have already do it.
414
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
415
revision_id = osutils.safe_revision_id(revision_id)
416
rev = self.get_revision(revision_id)
418
# the current serializer..
419
self._revision_store._serializer.write_revision(rev, rev_tmp)
421
return rev_tmp.getvalue()
424
def get_revision(self, revision_id):
425
"""Return the Revision object for a named revision"""
426
# TODO: jam 20070210 get_revision_reconcile should do this for us
427
revision_id = osutils.safe_revision_id(revision_id)
428
r = self.get_revision_reconcile(revision_id)
429
# weave corruption can lead to absent revision markers that should be
431
# the following test is reasonably cheap (it needs a single weave read)
432
# and the weave is cached in read transactions. In write transactions
433
# it is not cached but typically we only read a small number of
434
# revisions. For knits when they are introduced we will probably want
435
# to ensure that caching write transactions are in use.
436
inv = self.get_inventory_weave()
437
self._check_revision_parents(r, inv)
441
def get_deltas_for_revisions(self, revisions):
442
"""Produce a generator of revision deltas.
444
Note that the input is a sequence of REVISIONS, not revision_ids.
445
Trees will be held in memory until the generator exits.
446
Each delta is relative to the revision's lefthand predecessor.
448
required_trees = set()
449
for revision in revisions:
450
required_trees.add(revision.revision_id)
451
required_trees.update(revision.parent_ids[:1])
452
trees = dict((t.get_revision_id(), t) for
453
t in self.revision_trees(required_trees))
454
for revision in revisions:
455
if not revision.parent_ids:
456
old_tree = self.revision_tree(None)
458
old_tree = trees[revision.parent_ids[0]]
459
yield trees[revision.revision_id].changes_from(old_tree)
462
def get_revision_delta(self, revision_id):
463
"""Return the delta for one revision.
465
The delta is relative to the left-hand predecessor of the
468
r = self.get_revision(revision_id)
469
return list(self.get_deltas_for_revisions([r]))[0]
471
def _check_revision_parents(self, revision, inventory):
472
"""Private to Repository and Fetch.
474
This checks the parentage of revision in an inventory weave for
475
consistency and is only applicable to inventory-weave-for-ancestry
476
using repository formats & fetchers.
478
weave_parents = inventory.get_parents(revision.revision_id)
479
weave_names = inventory.versions()
480
for parent_id in revision.parent_ids:
481
if parent_id in weave_names:
482
# this parent must not be a ghost.
483
if not parent_id in weave_parents:
485
raise errors.CorruptRepository(self)
488
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
489
revision_id = osutils.safe_revision_id(revision_id)
490
signature = gpg_strategy.sign(plaintext)
491
self._revision_store.add_revision_signature_text(revision_id,
493
self.get_transaction())
495
def fileids_altered_by_revision_ids(self, revision_ids):
496
"""Find the file ids and versions affected by revisions.
498
:param revisions: an iterable containing revision ids.
499
:return: a dictionary mapping altered file-ids to an iterable of
500
revision_ids. Each altered file-ids has the exact revision_ids that
501
altered it listed explicitly.
503
assert self._serializer.support_altered_by_hack, \
504
("fileids_altered_by_revision_ids only supported for branches "
505
"which store inventory as unnested xml, not on %r" % self)
506
selected_revision_ids = set(osutils.safe_revision_id(r)
507
for r in revision_ids)
508
w = self.get_inventory_weave()
511
# this code needs to read every new line in every inventory for the
512
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
513
# not present in one of those inventories is unnecessary but not
514
# harmful because we are filtering by the revision id marker in the
515
# inventory lines : we only select file ids altered in one of those
516
# revisions. We don't need to see all lines in the inventory because
517
# only those added in an inventory in rev X can contain a revision=X
519
unescape_revid_cache = {}
520
unescape_fileid_cache = {}
522
# jam 20061218 In a big fetch, this handles hundreds of thousands
523
# of lines, so it has had a lot of inlining and optimizing done.
524
# Sorry that it is a little bit messy.
525
# Move several functions to be local variables, since this is a long
527
search = self._file_ids_altered_regex.search
528
unescape = _unescape_xml
529
setdefault = result.setdefault
530
pb = ui.ui_factory.nested_progress_bar()
532
for line in w.iter_lines_added_or_present_in_versions(
533
selected_revision_ids, pb=pb):
537
# One call to match.group() returning multiple items is quite a
538
# bit faster than 2 calls to match.group() each returning 1
539
file_id, revision_id = match.group('file_id', 'revision_id')
541
# Inlining the cache lookups helps a lot when you make 170,000
542
# lines and 350k ids, versus 8.4 unique ids.
543
# Using a cache helps in 2 ways:
544
# 1) Avoids unnecessary decoding calls
545
# 2) Re-uses cached strings, which helps in future set and
547
# (2) is enough that removing encoding entirely along with
548
# the cache (so we are using plain strings) results in no
549
# performance improvement.
551
revision_id = unescape_revid_cache[revision_id]
553
unescaped = unescape(revision_id)
554
unescape_revid_cache[revision_id] = unescaped
555
revision_id = unescaped
557
if revision_id in selected_revision_ids:
559
file_id = unescape_fileid_cache[file_id]
561
unescaped = unescape(file_id)
562
unescape_fileid_cache[file_id] = unescaped
564
setdefault(file_id, set()).add(revision_id)
570
def get_inventory_weave(self):
571
return self.control_weaves.get_weave('inventory',
572
self.get_transaction())
575
def get_inventory(self, revision_id):
576
"""Get Inventory object by hash."""
577
# TODO: jam 20070210 Technically we don't need to sanitize, since all
578
# called functions must sanitize.
579
revision_id = osutils.safe_revision_id(revision_id)
580
return self.deserialise_inventory(
581
revision_id, self.get_inventory_xml(revision_id))
583
def deserialise_inventory(self, revision_id, xml):
584
"""Transform the xml into an inventory object.
586
:param revision_id: The expected revision id of the inventory.
587
:param xml: A serialised inventory.
589
revision_id = osutils.safe_revision_id(revision_id)
590
result = self._serializer.read_inventory_from_string(xml)
591
result.root.revision = revision_id
594
def serialise_inventory(self, inv):
595
return self._serializer.write_inventory_to_string(inv)
598
def get_inventory_xml(self, revision_id):
599
"""Get inventory XML as a file object."""
600
revision_id = osutils.safe_revision_id(revision_id)
602
assert isinstance(revision_id, str), type(revision_id)
603
iw = self.get_inventory_weave()
604
return iw.get_text(revision_id)
606
raise errors.HistoryMissing(self, 'inventory', revision_id)
609
def get_inventory_sha1(self, revision_id):
610
"""Return the sha1 hash of the inventory entry
612
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
613
revision_id = osutils.safe_revision_id(revision_id)
614
return self.get_revision(revision_id).inventory_sha1
617
def get_revision_graph(self, revision_id=None):
618
"""Return a dictionary containing the revision graph.
620
:param revision_id: The revision_id to get a graph from. If None, then
621
the entire revision graph is returned. This is a deprecated mode of
622
operation and will be removed in the future.
623
:return: a dictionary of revision_id->revision_parents_list.
625
# special case NULL_REVISION
626
if revision_id == _mod_revision.NULL_REVISION:
628
revision_id = osutils.safe_revision_id(revision_id)
629
a_weave = self.get_inventory_weave()
630
all_revisions = self._eliminate_revisions_not_present(
632
entire_graph = dict([(node, a_weave.get_parents(node)) for
633
node in all_revisions])
634
if revision_id is None:
636
elif revision_id not in entire_graph:
637
raise errors.NoSuchRevision(self, revision_id)
639
# add what can be reached from revision_id
641
pending = set([revision_id])
642
while len(pending) > 0:
644
result[node] = entire_graph[node]
645
for revision_id in result[node]:
646
if revision_id not in result:
647
pending.add(revision_id)
651
def get_revision_graph_with_ghosts(self, revision_ids=None):
652
"""Return a graph of the revisions with ghosts marked as applicable.
654
:param revision_ids: an iterable of revisions to graph or None for all.
655
:return: a Graph object with the graph reachable from revision_ids.
657
result = graph.Graph()
659
pending = set(self.all_revision_ids())
662
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
663
# special case NULL_REVISION
664
if _mod_revision.NULL_REVISION in pending:
665
pending.remove(_mod_revision.NULL_REVISION)
666
required = set(pending)
669
revision_id = pending.pop()
671
rev = self.get_revision(revision_id)
672
except errors.NoSuchRevision:
673
if revision_id in required:
676
result.add_ghost(revision_id)
678
for parent_id in rev.parent_ids:
679
# is this queued or done ?
680
if (parent_id not in pending and
681
parent_id not in done):
683
pending.add(parent_id)
684
result.add_node(revision_id, rev.parent_ids)
685
done.add(revision_id)
688
def _get_history_vf(self):
689
"""Get a versionedfile whose history graph reflects all revisions.
691
For weave repositories, this is the inventory weave.
693
return self.get_inventory_weave()
695
def iter_reverse_revision_history(self, revision_id):
696
"""Iterate backwards through revision ids in the lefthand history
698
:param revision_id: The revision id to start with. All its lefthand
699
ancestors will be traversed.
701
revision_id = osutils.safe_revision_id(revision_id)
702
if revision_id in (None, _mod_revision.NULL_REVISION):
704
next_id = revision_id
705
versionedfile = self._get_history_vf()
708
parents = versionedfile.get_parents(next_id)
709
if len(parents) == 0:
715
def get_revision_inventory(self, revision_id):
716
"""Return inventory of a past revision."""
717
# TODO: Unify this with get_inventory()
718
# bzr 0.0.6 and later imposes the constraint that the inventory_id
719
# must be the same as its revision, so this is trivial.
720
if revision_id is None:
721
# This does not make sense: if there is no revision,
722
# then it is the current tree inventory surely ?!
723
# and thus get_root_id() is something that looks at the last
724
# commit on the branch, and the get_root_id is an inventory check.
725
raise NotImplementedError
726
# return Inventory(self.get_root_id())
728
return self.get_inventory(revision_id)
732
"""Return True if this repository is flagged as a shared repository."""
733
raise NotImplementedError(self.is_shared)
736
def reconcile(self, other=None, thorough=False):
737
"""Reconcile this repository."""
738
from bzrlib.reconcile import RepoReconciler
739
reconciler = RepoReconciler(self, thorough=thorough)
740
reconciler.reconcile()
744
def revision_tree(self, revision_id):
745
"""Return Tree for a revision on this branch.
747
`revision_id` may be None for the empty tree revision.
749
# TODO: refactor this to use an existing revision object
750
# so we don't need to read it in twice.
751
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
752
return RevisionTree(self, Inventory(root_id=None),
753
_mod_revision.NULL_REVISION)
755
revision_id = osutils.safe_revision_id(revision_id)
756
inv = self.get_revision_inventory(revision_id)
757
return RevisionTree(self, inv, revision_id)
760
def revision_trees(self, revision_ids):
761
"""Return Tree for a revision on this branch.
763
`revision_id` may not be None or 'null:'"""
764
assert None not in revision_ids
765
assert _mod_revision.NULL_REVISION not in revision_ids
766
texts = self.get_inventory_weave().get_texts(revision_ids)
767
for text, revision_id in zip(texts, revision_ids):
768
inv = self.deserialise_inventory(revision_id, text)
769
yield RevisionTree(self, inv, revision_id)
772
def get_ancestry(self, revision_id):
773
"""Return a list of revision-ids integrated by a revision.
775
The first element of the list is always None, indicating the origin
776
revision. This might change when we have history horizons, or
777
perhaps we should have a new API.
779
This is topologically sorted.
781
if revision_id is None:
783
revision_id = osutils.safe_revision_id(revision_id)
784
if not self.has_revision(revision_id):
785
raise errors.NoSuchRevision(self, revision_id)
786
w = self.get_inventory_weave()
787
candidates = w.get_ancestry(revision_id)
788
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
791
def print_file(self, file, revision_id):
792
"""Print `file` to stdout.
794
FIXME RBC 20060125 as John Meinel points out this is a bad api
795
- it writes to stdout, it assumes that that is valid etc. Fix
796
by creating a new more flexible convenience function.
798
revision_id = osutils.safe_revision_id(revision_id)
799
tree = self.revision_tree(revision_id)
800
# use inventory as it was in that revision
801
file_id = tree.inventory.path2id(file)
803
# TODO: jam 20060427 Write a test for this code path
804
# it had a bug in it, and was raising the wrong
806
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
807
tree.print_file(file_id)
809
def get_transaction(self):
810
return self.control_files.get_transaction()
812
def revision_parents(self, revision_id):
813
revision_id = osutils.safe_revision_id(revision_id)
814
return self.get_inventory_weave().parent_names(revision_id)
817
def set_make_working_trees(self, new_value):
818
"""Set the policy flag for making working trees when creating branches.
820
This only applies to branches that use this repository.
822
The default is 'True'.
823
:param new_value: True to restore the default, False to disable making
826
raise NotImplementedError(self.set_make_working_trees)
828
def make_working_trees(self):
829
"""Returns the policy for making working trees on new branches."""
830
raise NotImplementedError(self.make_working_trees)
833
def sign_revision(self, revision_id, gpg_strategy):
834
revision_id = osutils.safe_revision_id(revision_id)
835
plaintext = Testament.from_revision(self, revision_id).as_short_text()
836
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
839
def has_signature_for_revision_id(self, revision_id):
840
"""Query for a revision signature for revision_id in the repository."""
841
revision_id = osutils.safe_revision_id(revision_id)
842
return self._revision_store.has_signature(revision_id,
843
self.get_transaction())
846
def get_signature_text(self, revision_id):
847
"""Return the text for a signature."""
848
revision_id = osutils.safe_revision_id(revision_id)
849
return self._revision_store.get_signature_text(revision_id,
850
self.get_transaction())
853
def check(self, revision_ids):
854
"""Check consistency of all history of given revision_ids.
856
Different repository implementations should override _check().
858
:param revision_ids: A non-empty list of revision_ids whose ancestry
859
will be checked. Typically the last revision_id of a branch.
862
raise ValueError("revision_ids must be non-empty in %s.check"
864
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
865
return self._check(revision_ids)
867
def _check(self, revision_ids):
868
result = check.Check(self)
872
def _warn_if_deprecated(self):
873
global _deprecation_warning_done
874
if _deprecation_warning_done:
876
_deprecation_warning_done = True
877
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
878
% (self._format, self.bzrdir.transport.base))
880
def supports_rich_root(self):
881
return self._format.rich_root_data
883
def _check_ascii_revisionid(self, revision_id, method):
884
"""Private helper for ascii-only repositories."""
885
# weave repositories refuse to store revisionids that are non-ascii.
886
if revision_id is not None:
887
# weaves require ascii revision ids.
888
if isinstance(revision_id, unicode):
890
revision_id.encode('ascii')
891
except UnicodeEncodeError:
892
raise errors.NonAsciiRevisionId(method, self)
895
revision_id.decode('ascii')
896
except UnicodeDecodeError:
897
raise errors.NonAsciiRevisionId(method, self)
901
# remove these delegates a while after bzr 0.15
902
def __make_delegated(name, from_module):
903
def _deprecated_repository_forwarder():
904
symbol_versioning.warn('%s moved to %s in bzr 0.15'
905
% (name, from_module),
908
m = __import__(from_module, globals(), locals(), [name])
910
return getattr(m, name)
911
except AttributeError:
912
raise AttributeError('module %s has no name %s'
914
globals()[name] = _deprecated_repository_forwarder
917
'AllInOneRepository',
918
'WeaveMetaDirRepository',
919
'PreSplitOutRepositoryFormat',
925
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
929
'RepositoryFormatKnit',
930
'RepositoryFormatKnit1',
932
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
935
def install_revision(repository, rev, revision_tree):
936
"""Install all revision data into a repository."""
939
for p_id in rev.parent_ids:
940
if repository.has_revision(p_id):
941
present_parents.append(p_id)
942
parent_trees[p_id] = repository.revision_tree(p_id)
944
parent_trees[p_id] = repository.revision_tree(None)
946
inv = revision_tree.inventory
947
entries = inv.iter_entries()
948
# backwards compatability hack: skip the root id.
949
if not repository.supports_rich_root():
950
path, root = entries.next()
951
if root.revision != rev.revision_id:
952
raise errors.IncompatibleRevision(repr(repository))
953
# Add the texts that are not already present
954
for path, ie in entries:
955
w = repository.weave_store.get_weave_or_empty(ie.file_id,
956
repository.get_transaction())
957
if ie.revision not in w:
959
# FIXME: TODO: The following loop *may* be overlapping/duplicate
960
# with InventoryEntry.find_previous_heads(). if it is, then there
961
# is a latent bug here where the parents may have ancestors of each
963
for revision, tree in parent_trees.iteritems():
964
if ie.file_id not in tree:
966
parent_id = tree.inventory[ie.file_id].revision
967
if parent_id in text_parents:
969
text_parents.append(parent_id)
971
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
972
repository.get_transaction())
973
lines = revision_tree.get_file(ie.file_id).readlines()
974
vfile.add_lines(rev.revision_id, text_parents, lines)
976
# install the inventory
977
repository.add_inventory(rev.revision_id, inv, present_parents)
978
except errors.RevisionAlreadyPresent:
980
repository.add_revision(rev.revision_id, rev, inv)
983
class MetaDirRepository(Repository):
984
"""Repositories in the new meta-dir layout."""
986
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
987
super(MetaDirRepository, self).__init__(_format,
993
dir_mode = self.control_files._dir_mode
994
file_mode = self.control_files._file_mode
998
"""Return True if this repository is flagged as a shared repository."""
999
return self.control_files._transport.has('shared-storage')
1002
def set_make_working_trees(self, new_value):
1003
"""Set the policy flag for making working trees when creating branches.
1005
This only applies to branches that use this repository.
1007
The default is 'True'.
1008
:param new_value: True to restore the default, False to disable making
1013
self.control_files._transport.delete('no-working-trees')
1014
except errors.NoSuchFile:
1017
self.control_files.put_utf8('no-working-trees', '')
1019
def make_working_trees(self):
1020
"""Returns the policy for making working trees on new branches."""
1021
return not self.control_files._transport.has('no-working-trees')
1024
class RepositoryFormatRegistry(registry.Registry):
1025
"""Registry of RepositoryFormats.
1028
def get(self, format_string):
1029
r = registry.Registry.get(self, format_string)
1035
format_registry = RepositoryFormatRegistry()
1036
"""Registry of formats, indexed by their identifying format string.
1038
This can contain either format instances themselves, or classes/factories that
1039
can be called to obtain one.
1043
#####################################################################
1044
# Repository Formats
1046
class RepositoryFormat(object):
1047
"""A repository format.
1049
Formats provide three things:
1050
* An initialization routine to construct repository data on disk.
1051
* a format string which is used when the BzrDir supports versioned
1053
* an open routine which returns a Repository instance.
1055
Formats are placed in an dict by their format string for reference
1056
during opening. These should be subclasses of RepositoryFormat
1059
Once a format is deprecated, just deprecate the initialize and open
1060
methods on the format class. Do not deprecate the object, as the
1061
object will be created every system load.
1063
Common instance attributes:
1064
_matchingbzrdir - the bzrdir format that the repository format was
1065
originally written to work with. This can be used if manually
1066
constructing a bzrdir and repository, or more commonly for test suite
1071
return "<%s>" % self.__class__.__name__
1073
def __eq__(self, other):
1074
# format objects are generally stateless
1075
return isinstance(other, self.__class__)
1077
def __ne__(self, other):
1078
return not self == other
1081
def find_format(klass, a_bzrdir):
1082
"""Return the format for the repository object in a_bzrdir.
1084
This is used by bzr native formats that have a "format" file in
1085
the repository. Other methods may be used by different types of
1089
transport = a_bzrdir.get_repository_transport(None)
1090
format_string = transport.get("format").read()
1091
return format_registry.get(format_string)
1092
except errors.NoSuchFile:
1093
raise errors.NoRepositoryPresent(a_bzrdir)
1095
raise errors.UnknownFormatError(format=format_string)
1098
def register_format(klass, format):
1099
format_registry.register(format.get_format_string(), format)
1102
def unregister_format(klass, format):
1103
format_registry.remove(format.get_format_string())
1106
def get_default_format(klass):
1107
"""Return the current default format."""
1108
from bzrlib import bzrdir
1109
return bzrdir.format_registry.make_bzrdir('default').repository_format
1111
def _get_control_store(self, repo_transport, control_files):
1112
"""Return the control store for this repository."""
1113
raise NotImplementedError(self._get_control_store)
1115
def get_format_string(self):
1116
"""Return the ASCII format string that identifies this format.
1118
Note that in pre format ?? repositories the format string is
1119
not permitted nor written to disk.
1121
raise NotImplementedError(self.get_format_string)
1123
def get_format_description(self):
1124
"""Return the short description for this format."""
1125
raise NotImplementedError(self.get_format_description)
1127
def _get_revision_store(self, repo_transport, control_files):
1128
"""Return the revision store object for this a_bzrdir."""
1129
raise NotImplementedError(self._get_revision_store)
1131
def _get_text_rev_store(self,
1138
"""Common logic for getting a revision store for a repository.
1140
see self._get_revision_store for the subclass-overridable method to
1141
get the store for a repository.
1143
from bzrlib.store.revision.text import TextRevisionStore
1144
dir_mode = control_files._dir_mode
1145
file_mode = control_files._file_mode
1146
text_store = TextStore(transport.clone(name),
1148
compressed=compressed,
1150
file_mode=file_mode)
1151
_revision_store = TextRevisionStore(text_store, serializer)
1152
return _revision_store
1154
# TODO: this shouldn't be in the base class, it's specific to things that
1155
# use weaves or knits -- mbp 20070207
1156
def _get_versioned_file_store(self,
1161
versionedfile_class=None,
1162
versionedfile_kwargs={},
1164
if versionedfile_class is None:
1165
versionedfile_class = self._versionedfile_class
1166
weave_transport = control_files._transport.clone(name)
1167
dir_mode = control_files._dir_mode
1168
file_mode = control_files._file_mode
1169
return VersionedFileStore(weave_transport, prefixed=prefixed,
1171
file_mode=file_mode,
1172
versionedfile_class=versionedfile_class,
1173
versionedfile_kwargs=versionedfile_kwargs,
1176
def initialize(self, a_bzrdir, shared=False):
1177
"""Initialize a repository of this format in a_bzrdir.
1179
:param a_bzrdir: The bzrdir to put the new repository in it.
1180
:param shared: The repository should be initialized as a sharable one.
1182
This may raise UninitializableFormat if shared repository are not
1183
compatible the a_bzrdir.
1186
def is_supported(self):
1187
"""Is this format supported?
1189
Supported formats must be initializable and openable.
1190
Unsupported formats may not support initialization or committing or
1191
some other features depending on the reason for not being supported.
1195
def check_conversion_target(self, target_format):
1196
raise NotImplementedError(self.check_conversion_target)
1198
def open(self, a_bzrdir, _found=False):
1199
"""Return an instance of this format for the bzrdir a_bzrdir.
1201
_found is a private parameter, do not use it.
1203
raise NotImplementedError(self.open)
1206
class MetaDirRepositoryFormat(RepositoryFormat):
1207
"""Common base class for the new repositories using the metadir layout."""
1209
rich_root_data = False
1210
supports_tree_reference = False
1211
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1214
super(MetaDirRepositoryFormat, self).__init__()
1216
def _create_control_files(self, a_bzrdir):
1217
"""Create the required files and the initial control_files object."""
1218
# FIXME: RBC 20060125 don't peek under the covers
1219
# NB: no need to escape relative paths that are url safe.
1220
repository_transport = a_bzrdir.get_repository_transport(self)
1221
control_files = lockable_files.LockableFiles(repository_transport,
1222
'lock', lockdir.LockDir)
1223
control_files.create_lock()
1224
return control_files
1226
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1227
"""Upload the initial blank content."""
1228
control_files = self._create_control_files(a_bzrdir)
1229
control_files.lock_write()
1231
control_files._transport.mkdir_multi(dirs,
1232
mode=control_files._dir_mode)
1233
for file, content in files:
1234
control_files.put(file, content)
1235
for file, content in utf8_files:
1236
control_files.put_utf8(file, content)
1238
control_files.put_utf8('shared-storage', '')
1240
control_files.unlock()
1243
# formats which have no format string are not discoverable
1244
# and not independently creatable, so are not registered. They're
1245
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1246
# needed, it's constructed directly by the BzrDir. Non-native formats where
1247
# the repository is not separately opened are similar.
1249
format_registry.register_lazy(
1250
'Bazaar-NG Repository format 7',
1251
'bzrlib.repofmt.weaverepo',
1254
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1255
# default control directory format
1257
format_registry.register_lazy(
1258
'Bazaar-NG Knit Repository Format 1',
1259
'bzrlib.repofmt.knitrepo',
1260
'RepositoryFormatKnit1',
1262
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1264
format_registry.register_lazy(
1265
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1266
'bzrlib.repofmt.knitrepo',
1267
'RepositoryFormatKnit3',
1271
class InterRepository(InterObject):
1272
"""This class represents operations taking place between two repositories.
1274
Its instances have methods like copy_content and fetch, and contain
1275
references to the source and target repositories these operations can be
1278
Often we will provide convenience methods on 'repository' which carry out
1279
operations with another repository - they will always forward to
1280
InterRepository.get(other).method_name(parameters).
1284
"""The available optimised InterRepository types."""
1286
def copy_content(self, revision_id=None, basis=None):
1287
raise NotImplementedError(self.copy_content)
1289
def fetch(self, revision_id=None, pb=None):
1290
"""Fetch the content required to construct revision_id.
1292
The content is copied from self.source to self.target.
1294
:param revision_id: if None all content is copied, if NULL_REVISION no
1296
:param pb: optional progress bar to use for progress reports. If not
1297
provided a default one will be created.
1299
Returns the copied revision count and the failed revisions in a tuple:
1302
raise NotImplementedError(self.fetch)
1305
def missing_revision_ids(self, revision_id=None):
1306
"""Return the revision ids that source has that target does not.
1308
These are returned in topological order.
1310
:param revision_id: only return revision ids included by this
1313
# generic, possibly worst case, slow code path.
1314
target_ids = set(self.target.all_revision_ids())
1315
if revision_id is not None:
1316
# TODO: jam 20070210 InterRepository is internal enough that it
1317
# should assume revision_ids are already utf-8
1318
revision_id = osutils.safe_revision_id(revision_id)
1319
source_ids = self.source.get_ancestry(revision_id)
1320
assert source_ids[0] is None
1323
source_ids = self.source.all_revision_ids()
1324
result_set = set(source_ids).difference(target_ids)
1325
# this may look like a no-op: its not. It preserves the ordering
1326
# other_ids had while only returning the members from other_ids
1327
# that we've decided we need.
1328
return [rev_id for rev_id in source_ids if rev_id in result_set]
1331
class InterSameDataRepository(InterRepository):
1332
"""Code for converting between repositories that represent the same data.
1334
Data format and model must match for this to work.
1338
def _get_repo_format_to_test(self):
1339
"""Repository format for testing with."""
1340
return RepositoryFormat.get_default_format()
1343
def is_compatible(source, target):
1344
if not isinstance(source, Repository):
1346
if not isinstance(target, Repository):
1348
if source._format.rich_root_data != target._format.rich_root_data:
1350
if source._serializer != target._serializer:
1356
def copy_content(self, revision_id=None, basis=None):
1357
"""Make a complete copy of the content in self into destination.
1359
This is a destructive operation! Do not use it on existing
1362
:param revision_id: Only copy the content needed to construct
1363
revision_id and its parents.
1364
:param basis: Copy the needed data preferentially from basis.
1367
self.target.set_make_working_trees(self.source.make_working_trees())
1368
except NotImplementedError:
1370
# TODO: jam 20070210 This is fairly internal, so we should probably
1371
# just assert that revision_id is not unicode.
1372
revision_id = osutils.safe_revision_id(revision_id)
1373
# grab the basis available data
1374
if basis is not None:
1375
self.target.fetch(basis, revision_id=revision_id)
1376
# but don't bother fetching if we have the needed data now.
1377
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1378
self.target.has_revision(revision_id)):
1380
self.target.fetch(self.source, revision_id=revision_id)
1383
def fetch(self, revision_id=None, pb=None):
1384
"""See InterRepository.fetch()."""
1385
from bzrlib.fetch import GenericRepoFetcher
1386
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1387
self.source, self.source._format, self.target,
1388
self.target._format)
1389
# TODO: jam 20070210 This should be an assert, not a translate
1390
revision_id = osutils.safe_revision_id(revision_id)
1391
f = GenericRepoFetcher(to_repository=self.target,
1392
from_repository=self.source,
1393
last_revision=revision_id,
1395
return f.count_copied, f.failed_revisions
1398
class InterWeaveRepo(InterSameDataRepository):
1399
"""Optimised code paths between Weave based repositories."""
1402
def _get_repo_format_to_test(self):
1403
from bzrlib.repofmt import weaverepo
1404
return weaverepo.RepositoryFormat7()
1407
def is_compatible(source, target):
1408
"""Be compatible with known Weave formats.
1410
We don't test for the stores being of specific types because that
1411
could lead to confusing results, and there is no need to be
1414
from bzrlib.repofmt.weaverepo import (
1420
return (isinstance(source._format, (RepositoryFormat5,
1422
RepositoryFormat7)) and
1423
isinstance(target._format, (RepositoryFormat5,
1425
RepositoryFormat7)))
1426
except AttributeError:
1430
def copy_content(self, revision_id=None, basis=None):
1431
"""See InterRepository.copy_content()."""
1432
# weave specific optimised path:
1433
# TODO: jam 20070210 Internal, should be an assert, not translate
1434
revision_id = osutils.safe_revision_id(revision_id)
1435
if basis is not None:
1436
# copy the basis in, then fetch remaining data.
1437
basis.copy_content_into(self.target, revision_id)
1438
# the basis copy_content_into could miss-set this.
1440
self.target.set_make_working_trees(self.source.make_working_trees())
1441
except NotImplementedError:
1443
self.target.fetch(self.source, revision_id=revision_id)
1446
self.target.set_make_working_trees(self.source.make_working_trees())
1447
except NotImplementedError:
1449
# FIXME do not peek!
1450
if self.source.control_files._transport.listable():
1451
pb = ui.ui_factory.nested_progress_bar()
1453
self.target.weave_store.copy_all_ids(
1454
self.source.weave_store,
1456
from_transaction=self.source.get_transaction(),
1457
to_transaction=self.target.get_transaction())
1458
pb.update('copying inventory', 0, 1)
1459
self.target.control_weaves.copy_multi(
1460
self.source.control_weaves, ['inventory'],
1461
from_transaction=self.source.get_transaction(),
1462
to_transaction=self.target.get_transaction())
1463
self.target._revision_store.text_store.copy_all_ids(
1464
self.source._revision_store.text_store,
1469
self.target.fetch(self.source, revision_id=revision_id)
1472
def fetch(self, revision_id=None, pb=None):
1473
"""See InterRepository.fetch()."""
1474
from bzrlib.fetch import GenericRepoFetcher
1475
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1476
self.source, self.source._format, self.target, self.target._format)
1477
# TODO: jam 20070210 This should be an assert, not a translate
1478
revision_id = osutils.safe_revision_id(revision_id)
1479
f = GenericRepoFetcher(to_repository=self.target,
1480
from_repository=self.source,
1481
last_revision=revision_id,
1483
return f.count_copied, f.failed_revisions
1486
def missing_revision_ids(self, revision_id=None):
1487
"""See InterRepository.missing_revision_ids()."""
1488
# we want all revisions to satisfy revision_id in source.
1489
# but we don't want to stat every file here and there.
1490
# we want then, all revisions other needs to satisfy revision_id
1491
# checked, but not those that we have locally.
1492
# so the first thing is to get a subset of the revisions to
1493
# satisfy revision_id in source, and then eliminate those that
1494
# we do already have.
1495
# this is slow on high latency connection to self, but as as this
1496
# disk format scales terribly for push anyway due to rewriting
1497
# inventory.weave, this is considered acceptable.
1499
if revision_id is not None:
1500
source_ids = self.source.get_ancestry(revision_id)
1501
assert source_ids[0] is None
1504
source_ids = self.source._all_possible_ids()
1505
source_ids_set = set(source_ids)
1506
# source_ids is the worst possible case we may need to pull.
1507
# now we want to filter source_ids against what we actually
1508
# have in target, but don't try to check for existence where we know
1509
# we do not have a revision as that would be pointless.
1510
target_ids = set(self.target._all_possible_ids())
1511
possibly_present_revisions = target_ids.intersection(source_ids_set)
1512
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1513
required_revisions = source_ids_set.difference(actually_present_revisions)
1514
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1515
if revision_id is not None:
1516
# we used get_ancestry to determine source_ids then we are assured all
1517
# revisions referenced are present as they are installed in topological order.
1518
# and the tip revision was validated by get_ancestry.
1519
return required_topo_revisions
1521
# if we just grabbed the possibly available ids, then
1522
# we only have an estimate of whats available and need to validate
1523
# that against the revision records.
1524
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1527
class InterKnitRepo(InterSameDataRepository):
1528
"""Optimised code paths between Knit based repositories."""
1531
def _get_repo_format_to_test(self):
1532
from bzrlib.repofmt import knitrepo
1533
return knitrepo.RepositoryFormatKnit1()
1536
def is_compatible(source, target):
1537
"""Be compatible with known Knit formats.
1539
We don't test for the stores being of specific types because that
1540
could lead to confusing results, and there is no need to be
1543
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1545
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1546
isinstance(target._format, (RepositoryFormatKnit1)))
1547
except AttributeError:
1551
def fetch(self, revision_id=None, pb=None):
1552
"""See InterRepository.fetch()."""
1553
from bzrlib.fetch import KnitRepoFetcher
1554
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1555
self.source, self.source._format, self.target, self.target._format)
1556
# TODO: jam 20070210 This should be an assert, not a translate
1557
revision_id = osutils.safe_revision_id(revision_id)
1558
f = KnitRepoFetcher(to_repository=self.target,
1559
from_repository=self.source,
1560
last_revision=revision_id,
1562
return f.count_copied, f.failed_revisions
1565
def missing_revision_ids(self, revision_id=None):
1566
"""See InterRepository.missing_revision_ids()."""
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 InterModel1and2(InterRepository):
1598
def _get_repo_format_to_test(self):
1602
def is_compatible(source, target):
1603
if not isinstance(source, Repository):
1605
if not isinstance(target, Repository):
1607
if not source._format.rich_root_data and target._format.rich_root_data:
1613
def fetch(self, revision_id=None, pb=None):
1614
"""See InterRepository.fetch()."""
1615
from bzrlib.fetch import Model1toKnit2Fetcher
1616
# TODO: jam 20070210 This should be an assert, not a translate
1617
revision_id = osutils.safe_revision_id(revision_id)
1618
f = Model1toKnit2Fetcher(to_repository=self.target,
1619
from_repository=self.source,
1620
last_revision=revision_id,
1622
return f.count_copied, f.failed_revisions
1625
def copy_content(self, revision_id=None, basis=None):
1626
"""Make a complete copy of the content in self into destination.
1628
This is a destructive operation! Do not use it on existing
1631
:param revision_id: Only copy the content needed to construct
1632
revision_id and its parents.
1633
:param basis: Copy the needed data preferentially from basis.
1636
self.target.set_make_working_trees(self.source.make_working_trees())
1637
except NotImplementedError:
1639
# TODO: jam 20070210 Internal, assert, don't translate
1640
revision_id = osutils.safe_revision_id(revision_id)
1641
# grab the basis available data
1642
if basis is not None:
1643
self.target.fetch(basis, revision_id=revision_id)
1644
# but don't bother fetching if we have the needed data now.
1645
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1646
self.target.has_revision(revision_id)):
1648
self.target.fetch(self.source, revision_id=revision_id)
1651
class InterKnit1and2(InterKnitRepo):
1654
def _get_repo_format_to_test(self):
1658
def is_compatible(source, target):
1659
"""Be compatible with Knit1 source and Knit3 target"""
1660
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1662
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1663
RepositoryFormatKnit3
1664
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1665
isinstance(target._format, (RepositoryFormatKnit3)))
1666
except AttributeError:
1670
def fetch(self, revision_id=None, pb=None):
1671
"""See InterRepository.fetch()."""
1672
from bzrlib.fetch import Knit1to2Fetcher
1673
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1674
self.source, self.source._format, self.target,
1675
self.target._format)
1676
# TODO: jam 20070210 This should be an assert, not a translate
1677
revision_id = osutils.safe_revision_id(revision_id)
1678
f = Knit1to2Fetcher(to_repository=self.target,
1679
from_repository=self.source,
1680
last_revision=revision_id,
1682
return f.count_copied, f.failed_revisions
1685
InterRepository.register_optimiser(InterSameDataRepository)
1686
InterRepository.register_optimiser(InterWeaveRepo)
1687
InterRepository.register_optimiser(InterKnitRepo)
1688
InterRepository.register_optimiser(InterModel1and2)
1689
InterRepository.register_optimiser(InterKnit1and2)
1692
class RepositoryTestProviderAdapter(object):
1693
"""A tool to generate a suite testing multiple repository formats at once.
1695
This is done by copying the test once for each transport and injecting
1696
the transport_server, transport_readonly_server, and bzrdir_format and
1697
repository_format classes into each copy. Each copy is also given a new id()
1698
to make it easy to identify.
1701
def __init__(self, transport_server, transport_readonly_server, formats):
1702
self._transport_server = transport_server
1703
self._transport_readonly_server = transport_readonly_server
1704
self._formats = formats
1706
def adapt(self, test):
1707
result = unittest.TestSuite()
1708
for repository_format, bzrdir_format in self._formats:
1709
from copy import deepcopy
1710
new_test = deepcopy(test)
1711
new_test.transport_server = self._transport_server
1712
new_test.transport_readonly_server = self._transport_readonly_server
1713
new_test.bzrdir_format = bzrdir_format
1714
new_test.repository_format = repository_format
1715
def make_new_test_id():
1716
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1717
return lambda: new_id
1718
new_test.id = make_new_test_id()
1719
result.addTest(new_test)
1723
class InterRepositoryTestProviderAdapter(object):
1724
"""A tool to generate a suite testing multiple inter repository formats.
1726
This is done by copying the test once for each interrepo provider and injecting
1727
the transport_server, transport_readonly_server, repository_format and
1728
repository_to_format classes into each copy.
1729
Each copy is also given a new id() to make it easy to identify.
1732
def __init__(self, transport_server, transport_readonly_server, formats):
1733
self._transport_server = transport_server
1734
self._transport_readonly_server = transport_readonly_server
1735
self._formats = formats
1737
def adapt(self, test):
1738
result = unittest.TestSuite()
1739
for interrepo_class, repository_format, repository_format_to in self._formats:
1740
from copy import deepcopy
1741
new_test = deepcopy(test)
1742
new_test.transport_server = self._transport_server
1743
new_test.transport_readonly_server = self._transport_readonly_server
1744
new_test.interrepo_class = interrepo_class
1745
new_test.repository_format = repository_format
1746
new_test.repository_format_to = repository_format_to
1747
def make_new_test_id():
1748
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1749
return lambda: new_id
1750
new_test.id = make_new_test_id()
1751
result.addTest(new_test)
1755
def default_test_list():
1756
"""Generate the default list of interrepo permutations to test."""
1757
from bzrlib.repofmt import knitrepo, weaverepo
1759
# test the default InterRepository between format 6 and the current
1761
# XXX: robertc 20060220 reinstate this when there are two supported
1762
# formats which do not have an optimal code path between them.
1763
#result.append((InterRepository,
1764
# RepositoryFormat6(),
1765
# RepositoryFormatKnit1()))
1766
for optimiser_class in InterRepository._optimisers:
1767
format_to_test = optimiser_class._get_repo_format_to_test()
1768
if format_to_test is not None:
1769
result.append((optimiser_class,
1770
format_to_test, format_to_test))
1771
# if there are specific combinations we want to use, we can add them
1773
result.append((InterModel1and2,
1774
weaverepo.RepositoryFormat5(),
1775
knitrepo.RepositoryFormatKnit3()))
1776
result.append((InterKnit1and2,
1777
knitrepo.RepositoryFormatKnit1(),
1778
knitrepo.RepositoryFormatKnit3()))
1782
class CopyConverter(object):
1783
"""A repository conversion tool which just performs a copy of the content.
1785
This is slow but quite reliable.
1788
def __init__(self, target_format):
1789
"""Create a CopyConverter.
1791
:param target_format: The format the resulting repository should be.
1793
self.target_format = target_format
1795
def convert(self, repo, pb):
1796
"""Perform the conversion of to_convert, giving feedback via pb.
1798
:param to_convert: The disk object to convert.
1799
:param pb: a progress bar to use for progress information.
1804
# this is only useful with metadir layouts - separated repo content.
1805
# trigger an assertion if not such
1806
repo._format.get_format_string()
1807
self.repo_dir = repo.bzrdir
1808
self.step('Moving repository to repository.backup')
1809
self.repo_dir.transport.move('repository', 'repository.backup')
1810
backup_transport = self.repo_dir.transport.clone('repository.backup')
1811
repo._format.check_conversion_target(self.target_format)
1812
self.source_repo = repo._format.open(self.repo_dir,
1814
_override_transport=backup_transport)
1815
self.step('Creating new repository')
1816
converted = self.target_format.initialize(self.repo_dir,
1817
self.source_repo.is_shared())
1818
converted.lock_write()
1820
self.step('Copying content into repository.')
1821
self.source_repo.copy_content_into(converted)
1824
self.step('Deleting old repository content.')
1825
self.repo_dir.transport.delete_tree('repository.backup')
1826
self.pb.note('repository converted')
1828
def step(self, message):
1829
"""Update the pb by a step."""
1831
self.pb.update(message, self.count, self.total)
1834
class CommitBuilder(object):
1835
"""Provides an interface to build up a commit.
1837
This allows describing a tree to be committed without needing to
1838
know the internals of the format of the repository.
1841
record_root_entry = False
1842
def __init__(self, repository, parents, config, timestamp=None,
1843
timezone=None, committer=None, revprops=None,
1845
"""Initiate a CommitBuilder.
1847
:param repository: Repository to commit to.
1848
:param parents: Revision ids of the parents of the new revision.
1849
:param config: Configuration to use.
1850
:param timestamp: Optional timestamp recorded for commit.
1851
:param timezone: Optional timezone for timestamp.
1852
:param committer: Optional committer to set for commit.
1853
:param revprops: Optional dictionary of revision properties.
1854
:param revision_id: Optional revision id.
1856
self._config = config
1858
if committer is None:
1859
self._committer = self._config.username()
1861
assert isinstance(committer, basestring), type(committer)
1862
self._committer = committer
1864
self.new_inventory = Inventory(None)
1865
self._new_revision_id = osutils.safe_revision_id(revision_id)
1866
self.parents = parents
1867
self.repository = repository
1870
if revprops is not None:
1871
self._revprops.update(revprops)
1873
if timestamp is None:
1874
timestamp = time.time()
1875
# Restrict resolution to 1ms
1876
self._timestamp = round(timestamp, 3)
1878
if timezone is None:
1879
self._timezone = osutils.local_time_offset()
1881
self._timezone = int(timezone)
1883
self._generate_revision_if_needed()
1885
def commit(self, message):
1886
"""Make the actual commit.
1888
:return: The revision id of the recorded revision.
1890
rev = _mod_revision.Revision(
1891
timestamp=self._timestamp,
1892
timezone=self._timezone,
1893
committer=self._committer,
1895
inventory_sha1=self.inv_sha1,
1896
revision_id=self._new_revision_id,
1897
properties=self._revprops)
1898
rev.parent_ids = self.parents
1899
self.repository.add_revision(self._new_revision_id, rev,
1900
self.new_inventory, self._config)
1901
return self._new_revision_id
1903
def revision_tree(self):
1904
"""Return the tree that was just committed.
1906
After calling commit() this can be called to get a RevisionTree
1907
representing the newly committed tree. This is preferred to
1908
calling Repository.revision_tree() because that may require
1909
deserializing the inventory, while we already have a copy in
1912
return RevisionTree(self.repository, self.new_inventory,
1913
self._new_revision_id)
1915
def finish_inventory(self):
1916
"""Tell the builder that the inventory is finished."""
1917
if self.new_inventory.root is None:
1918
symbol_versioning.warn('Root entry should be supplied to'
1919
' record_entry_contents, as of bzr 0.10.',
1920
DeprecationWarning, stacklevel=2)
1921
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1922
self.new_inventory.revision_id = self._new_revision_id
1923
self.inv_sha1 = self.repository.add_inventory(
1924
self._new_revision_id,
1929
def _gen_revision_id(self):
1930
"""Return new revision-id."""
1931
return generate_ids.gen_revision_id(self._config.username(),
1934
def _generate_revision_if_needed(self):
1935
"""Create a revision id if None was supplied.
1937
If the repository can not support user-specified revision ids
1938
they should override this function and raise CannotSetRevisionId
1939
if _new_revision_id is not None.
1941
:raises: CannotSetRevisionId
1943
if self._new_revision_id is None:
1944
self._new_revision_id = self._gen_revision_id()
1946
def record_entry_contents(self, ie, parent_invs, path, tree):
1947
"""Record the content of ie from tree into the commit if needed.
1949
Side effect: sets ie.revision when unchanged
1951
:param ie: An inventory entry present in the commit.
1952
:param parent_invs: The inventories of the parent revisions of the
1954
:param path: The path the entry is at in the tree.
1955
:param tree: The tree which contains this entry and should be used to
1958
if self.new_inventory.root is None and ie.parent_id is not None:
1959
symbol_versioning.warn('Root entry should be supplied to'
1960
' record_entry_contents, as of bzr 0.10.',
1961
DeprecationWarning, stacklevel=2)
1962
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1964
self.new_inventory.add(ie)
1966
# ie.revision is always None if the InventoryEntry is considered
1967
# for committing. ie.snapshot will record the correct revision
1968
# which may be the sole parent if it is untouched.
1969
if ie.revision is not None:
1972
# In this revision format, root entries have no knit or weave
1973
if ie is self.new_inventory.root:
1974
# When serializing out to disk and back in
1975
# root.revision is always _new_revision_id
1976
ie.revision = self._new_revision_id
1978
previous_entries = ie.find_previous_heads(
1980
self.repository.weave_store,
1981
self.repository.get_transaction())
1982
# we are creating a new revision for ie in the history store
1984
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1986
def modified_directory(self, file_id, file_parents):
1987
"""Record the presence of a symbolic link.
1989
:param file_id: The file_id of the link to record.
1990
:param file_parents: The per-file parent revision ids.
1992
self._add_text_to_weave(file_id, [], file_parents.keys())
1994
def modified_reference(self, file_id, file_parents):
1995
"""Record the modification of a reference.
1997
:param file_id: The file_id of the link to record.
1998
:param file_parents: The per-file parent revision ids.
2000
self._add_text_to_weave(file_id, [], file_parents.keys())
2002
def modified_file_text(self, file_id, file_parents,
2003
get_content_byte_lines, text_sha1=None,
2005
"""Record the text of file file_id
2007
:param file_id: The file_id of the file to record the text of.
2008
:param file_parents: The per-file parent revision ids.
2009
:param get_content_byte_lines: A callable which will return the byte
2011
:param text_sha1: Optional SHA1 of the file contents.
2012
:param text_size: Optional size of the file contents.
2014
# mutter('storing text of file {%s} in revision {%s} into %r',
2015
# file_id, self._new_revision_id, self.repository.weave_store)
2016
# special case to avoid diffing on renames or
2018
if (len(file_parents) == 1
2019
and text_sha1 == file_parents.values()[0].text_sha1
2020
and text_size == file_parents.values()[0].text_size):
2021
previous_ie = file_parents.values()[0]
2022
versionedfile = self.repository.weave_store.get_weave(file_id,
2023
self.repository.get_transaction())
2024
versionedfile.clone_text(self._new_revision_id,
2025
previous_ie.revision, file_parents.keys())
2026
return text_sha1, text_size
2028
new_lines = get_content_byte_lines()
2029
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2030
# should return the SHA1 and size
2031
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2032
return osutils.sha_strings(new_lines), \
2033
sum(map(len, new_lines))
2035
def modified_link(self, file_id, file_parents, link_target):
2036
"""Record the presence of a symbolic link.
2038
:param file_id: The file_id of the link to record.
2039
:param file_parents: The per-file parent revision ids.
2040
:param link_target: Target location of this link.
2042
self._add_text_to_weave(file_id, [], file_parents.keys())
2044
def _add_text_to_weave(self, file_id, new_lines, parents):
2045
versionedfile = self.repository.weave_store.get_weave_or_empty(
2046
file_id, self.repository.get_transaction())
2047
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2048
versionedfile.clear_cache()
2051
class _CommitBuilder(CommitBuilder):
2052
"""Temporary class so old CommitBuilders are detected properly
2054
Note: CommitBuilder works whether or not root entry is recorded.
2057
record_root_entry = True
2060
class RootCommitBuilder(CommitBuilder):
2061
"""This commitbuilder actually records the root id"""
2063
record_root_entry = True
2065
def record_entry_contents(self, ie, parent_invs, path, tree):
2066
"""Record the content of ie from tree into the commit if needed.
2068
Side effect: sets ie.revision when unchanged
2070
:param ie: An inventory entry present in the commit.
2071
:param parent_invs: The inventories of the parent revisions of the
2073
:param path: The path the entry is at in the tree.
2074
:param tree: The tree which contains this entry and should be used to
2077
assert self.new_inventory.root is not None or ie.parent_id is None
2078
self.new_inventory.add(ie)
2080
# ie.revision is always None if the InventoryEntry is considered
2081
# for committing. ie.snapshot will record the correct revision
2082
# which may be the sole parent if it is untouched.
2083
if ie.revision is not None:
2086
previous_entries = ie.find_previous_heads(
2088
self.repository.weave_store,
2089
self.repository.get_transaction())
2090
# we are creating a new revision for ie in the history store
2092
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2104
def _unescaper(match, _map=_unescape_map):
2105
code = match.group(1)
2109
if not code.startswith('#'):
2111
return unichr(int(code[1:])).encode('utf8')
2117
def _unescape_xml(data):
2118
"""Unescape predefined XML entities in a string of data."""
2120
if _unescape_re is None:
2121
_unescape_re = re.compile('\&([^;]*);')
2122
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
