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, token=None):
235
"""Lock this repository for writing.
237
:param token: if this is already locked, then lock_write will fail
238
unless the token matches the existing lock.
239
:returns: a token if this instance supports tokens, otherwise None.
240
:raises TokenLockingNotSupported: when a token is given but this
241
instance doesn't support using token locks.
242
:raises MismatchedToken: if the specified token doesn't match the token
243
of the existing lock.
245
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
247
return self.control_files.lock_write(token=token)
250
self.control_files.lock_read()
252
def get_physical_lock_status(self):
253
return self.control_files.get_physical_lock_status()
255
def leave_lock_in_place(self):
256
"""Tell this repository not to release the physical lock when this
259
If lock_write doesn't return a token, then this method is not supported.
261
self.control_files.leave_in_place()
263
def dont_leave_lock_in_place(self):
264
"""Tell this repository to release the physical lock when this
265
object is unlocked, even if it didn't originally acquire it.
267
If lock_write doesn't return a token, then this method is not supported.
269
self.control_files.dont_leave_in_place()
272
def gather_stats(self, revid=None, committers=None):
273
"""Gather statistics from a revision id.
275
:param revid: The revision id to gather statistics from, if None, then
276
no revision specific statistics are gathered.
277
:param committers: Optional parameter controlling whether to grab
278
a count of committers from the revision specific statistics.
279
:return: A dictionary of statistics. Currently this contains:
280
committers: The number of committers if requested.
281
firstrev: A tuple with timestamp, timezone for the penultimate left
282
most ancestor of revid, if revid is not the NULL_REVISION.
283
latestrev: A tuple with timestamp, timezone for revid, if revid is
284
not the NULL_REVISION.
285
revisions: The total revision count in the repository.
286
size: An estimate disk size of the repository in bytes.
289
if revid and committers:
290
result['committers'] = 0
291
if revid and revid != _mod_revision.NULL_REVISION:
293
all_committers = set()
294
revisions = self.get_ancestry(revid)
295
# pop the leading None
297
first_revision = None
299
# ignore the revisions in the middle - just grab first and last
300
revisions = revisions[0], revisions[-1]
301
for revision in self.get_revisions(revisions):
302
if not first_revision:
303
first_revision = revision
305
all_committers.add(revision.committer)
306
last_revision = revision
308
result['committers'] = len(all_committers)
309
result['firstrev'] = (first_revision.timestamp,
310
first_revision.timezone)
311
result['latestrev'] = (last_revision.timestamp,
312
last_revision.timezone)
314
# now gather global repository information
315
if self.bzrdir.root_transport.listable():
316
c, t = self._revision_store.total_size(self.get_transaction())
317
result['revisions'] = c
322
def missing_revision_ids(self, other, revision_id=None):
323
"""Return the revision ids that other has that this does not.
325
These are returned in topological order.
327
revision_id: only return revision ids included by revision_id.
329
revision_id = osutils.safe_revision_id(revision_id)
330
return InterRepository.get(other, self).missing_revision_ids(revision_id)
334
"""Open the repository rooted at base.
336
For instance, if the repository is at URL/.bzr/repository,
337
Repository.open(URL) -> a Repository instance.
339
control = bzrdir.BzrDir.open(base)
340
return control.open_repository()
342
def copy_content_into(self, destination, revision_id=None):
343
"""Make a complete copy of the content in self into destination.
345
This is a destructive operation! Do not use it on existing
348
revision_id = osutils.safe_revision_id(revision_id)
349
return InterRepository.get(self, destination).copy_content(revision_id)
351
def fetch(self, source, revision_id=None, pb=None):
352
"""Fetch the content required to construct revision_id from source.
354
If revision_id is None all content is copied.
356
revision_id = osutils.safe_revision_id(revision_id)
357
inter = InterRepository.get(source, self)
359
return inter.fetch(revision_id=revision_id, pb=pb)
360
except NotImplementedError:
361
raise errors.IncompatibleRepositories(source, self)
363
def get_commit_builder(self, branch, parents, config, timestamp=None,
364
timezone=None, committer=None, revprops=None,
366
"""Obtain a CommitBuilder for this repository.
368
:param branch: Branch to commit to.
369
:param parents: Revision ids of the parents of the new revision.
370
:param config: Configuration to use.
371
:param timestamp: Optional timestamp recorded for commit.
372
:param timezone: Optional timezone for timestamp.
373
:param committer: Optional committer to set for commit.
374
:param revprops: Optional dictionary of revision properties.
375
:param revision_id: Optional revision id.
377
revision_id = osutils.safe_revision_id(revision_id)
378
return _CommitBuilder(self, parents, config, timestamp, timezone,
379
committer, revprops, revision_id)
382
self.control_files.unlock()
385
def clone(self, a_bzrdir, revision_id=None):
386
"""Clone this repository into a_bzrdir using the current format.
388
Currently no check is made that the format of this repository and
389
the bzrdir format are compatible. FIXME RBC 20060201.
391
:return: The newly created destination repository.
393
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
394
# use target default format.
395
dest_repo = a_bzrdir.create_repository()
397
# Most control formats need the repository to be specifically
398
# created, but on some old all-in-one formats it's not needed
400
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
401
except errors.UninitializableFormat:
402
dest_repo = a_bzrdir.open_repository()
403
self.copy_content_into(dest_repo, revision_id)
407
def has_revision(self, revision_id):
408
"""True if this repository has a copy of the revision."""
409
revision_id = osutils.safe_revision_id(revision_id)
410
return self._revision_store.has_revision_id(revision_id,
411
self.get_transaction())
414
def get_revision_reconcile(self, revision_id):
415
"""'reconcile' helper routine that allows access to a revision always.
417
This variant of get_revision does not cross check the weave graph
418
against the revision one as get_revision does: but it should only
419
be used by reconcile, or reconcile-alike commands that are correcting
420
or testing the revision graph.
422
if not revision_id or not isinstance(revision_id, basestring):
423
raise errors.InvalidRevisionId(revision_id=revision_id,
425
return self.get_revisions([revision_id])[0]
428
def get_revisions(self, revision_ids):
429
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
430
revs = self._revision_store.get_revisions(revision_ids,
431
self.get_transaction())
433
assert not isinstance(rev.revision_id, unicode)
434
for parent_id in rev.parent_ids:
435
assert not isinstance(parent_id, unicode)
439
def get_revision_xml(self, revision_id):
440
# TODO: jam 20070210 This shouldn't be necessary since get_revision
441
# would have already do it.
442
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
443
revision_id = osutils.safe_revision_id(revision_id)
444
rev = self.get_revision(revision_id)
446
# the current serializer..
447
self._revision_store._serializer.write_revision(rev, rev_tmp)
449
return rev_tmp.getvalue()
452
def get_revision(self, revision_id):
453
"""Return the Revision object for a named revision"""
454
# TODO: jam 20070210 get_revision_reconcile should do this for us
455
revision_id = osutils.safe_revision_id(revision_id)
456
r = self.get_revision_reconcile(revision_id)
457
# weave corruption can lead to absent revision markers that should be
459
# the following test is reasonably cheap (it needs a single weave read)
460
# and the weave is cached in read transactions. In write transactions
461
# it is not cached but typically we only read a small number of
462
# revisions. For knits when they are introduced we will probably want
463
# to ensure that caching write transactions are in use.
464
inv = self.get_inventory_weave()
465
self._check_revision_parents(r, inv)
469
def get_deltas_for_revisions(self, revisions):
470
"""Produce a generator of revision deltas.
472
Note that the input is a sequence of REVISIONS, not revision_ids.
473
Trees will be held in memory until the generator exits.
474
Each delta is relative to the revision's lefthand predecessor.
476
required_trees = set()
477
for revision in revisions:
478
required_trees.add(revision.revision_id)
479
required_trees.update(revision.parent_ids[:1])
480
trees = dict((t.get_revision_id(), t) for
481
t in self.revision_trees(required_trees))
482
for revision in revisions:
483
if not revision.parent_ids:
484
old_tree = self.revision_tree(None)
486
old_tree = trees[revision.parent_ids[0]]
487
yield trees[revision.revision_id].changes_from(old_tree)
490
def get_revision_delta(self, revision_id):
491
"""Return the delta for one revision.
493
The delta is relative to the left-hand predecessor of the
496
r = self.get_revision(revision_id)
497
return list(self.get_deltas_for_revisions([r]))[0]
499
def _check_revision_parents(self, revision, inventory):
500
"""Private to Repository and Fetch.
502
This checks the parentage of revision in an inventory weave for
503
consistency and is only applicable to inventory-weave-for-ancestry
504
using repository formats & fetchers.
506
weave_parents = inventory.get_parents(revision.revision_id)
507
weave_names = inventory.versions()
508
for parent_id in revision.parent_ids:
509
if parent_id in weave_names:
510
# this parent must not be a ghost.
511
if not parent_id in weave_parents:
513
raise errors.CorruptRepository(self)
516
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
517
revision_id = osutils.safe_revision_id(revision_id)
518
signature = gpg_strategy.sign(plaintext)
519
self._revision_store.add_revision_signature_text(revision_id,
521
self.get_transaction())
523
def fileids_altered_by_revision_ids(self, revision_ids):
524
"""Find the file ids and versions affected by revisions.
526
:param revisions: an iterable containing revision ids.
527
:return: a dictionary mapping altered file-ids to an iterable of
528
revision_ids. Each altered file-ids has the exact revision_ids that
529
altered it listed explicitly.
531
assert self._serializer.support_altered_by_hack, \
532
("fileids_altered_by_revision_ids only supported for branches "
533
"which store inventory as unnested xml, not on %r" % self)
534
selected_revision_ids = set(osutils.safe_revision_id(r)
535
for r in revision_ids)
536
w = self.get_inventory_weave()
539
# this code needs to read every new line in every inventory for the
540
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
541
# not present in one of those inventories is unnecessary but not
542
# harmful because we are filtering by the revision id marker in the
543
# inventory lines : we only select file ids altered in one of those
544
# revisions. We don't need to see all lines in the inventory because
545
# only those added in an inventory in rev X can contain a revision=X
547
unescape_revid_cache = {}
548
unescape_fileid_cache = {}
550
# jam 20061218 In a big fetch, this handles hundreds of thousands
551
# of lines, so it has had a lot of inlining and optimizing done.
552
# Sorry that it is a little bit messy.
553
# Move several functions to be local variables, since this is a long
555
search = self._file_ids_altered_regex.search
556
unescape = _unescape_xml
557
setdefault = result.setdefault
558
pb = ui.ui_factory.nested_progress_bar()
560
for line in w.iter_lines_added_or_present_in_versions(
561
selected_revision_ids, pb=pb):
565
# One call to match.group() returning multiple items is quite a
566
# bit faster than 2 calls to match.group() each returning 1
567
file_id, revision_id = match.group('file_id', 'revision_id')
569
# Inlining the cache lookups helps a lot when you make 170,000
570
# lines and 350k ids, versus 8.4 unique ids.
571
# Using a cache helps in 2 ways:
572
# 1) Avoids unnecessary decoding calls
573
# 2) Re-uses cached strings, which helps in future set and
575
# (2) is enough that removing encoding entirely along with
576
# the cache (so we are using plain strings) results in no
577
# performance improvement.
579
revision_id = unescape_revid_cache[revision_id]
581
unescaped = unescape(revision_id)
582
unescape_revid_cache[revision_id] = unescaped
583
revision_id = unescaped
585
if revision_id in selected_revision_ids:
587
file_id = unescape_fileid_cache[file_id]
589
unescaped = unescape(file_id)
590
unescape_fileid_cache[file_id] = unescaped
592
setdefault(file_id, set()).add(revision_id)
598
def get_inventory_weave(self):
599
return self.control_weaves.get_weave('inventory',
600
self.get_transaction())
603
def get_inventory(self, revision_id):
604
"""Get Inventory object by hash."""
605
# TODO: jam 20070210 Technically we don't need to sanitize, since all
606
# called functions must sanitize.
607
revision_id = osutils.safe_revision_id(revision_id)
608
return self.deserialise_inventory(
609
revision_id, self.get_inventory_xml(revision_id))
611
def deserialise_inventory(self, revision_id, xml):
612
"""Transform the xml into an inventory object.
614
:param revision_id: The expected revision id of the inventory.
615
:param xml: A serialised inventory.
617
revision_id = osutils.safe_revision_id(revision_id)
618
result = self._serializer.read_inventory_from_string(xml)
619
result.root.revision = revision_id
622
def serialise_inventory(self, inv):
623
return self._serializer.write_inventory_to_string(inv)
626
def get_inventory_xml(self, revision_id):
627
"""Get inventory XML as a file object."""
628
revision_id = osutils.safe_revision_id(revision_id)
630
assert isinstance(revision_id, str), type(revision_id)
631
iw = self.get_inventory_weave()
632
return iw.get_text(revision_id)
634
raise errors.HistoryMissing(self, 'inventory', revision_id)
637
def get_inventory_sha1(self, revision_id):
638
"""Return the sha1 hash of the inventory entry
640
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
641
revision_id = osutils.safe_revision_id(revision_id)
642
return self.get_revision(revision_id).inventory_sha1
645
def get_revision_graph(self, revision_id=None):
646
"""Return a dictionary containing the revision graph.
648
:param revision_id: The revision_id to get a graph from. If None, then
649
the entire revision graph is returned. This is a deprecated mode of
650
operation and will be removed in the future.
651
:return: a dictionary of revision_id->revision_parents_list.
653
# special case NULL_REVISION
654
if revision_id == _mod_revision.NULL_REVISION:
656
revision_id = osutils.safe_revision_id(revision_id)
657
a_weave = self.get_inventory_weave()
658
all_revisions = self._eliminate_revisions_not_present(
660
entire_graph = dict([(node, a_weave.get_parents(node)) for
661
node in all_revisions])
662
if revision_id is None:
664
elif revision_id not in entire_graph:
665
raise errors.NoSuchRevision(self, revision_id)
667
# add what can be reached from revision_id
669
pending = set([revision_id])
670
while len(pending) > 0:
672
result[node] = entire_graph[node]
673
for revision_id in result[node]:
674
if revision_id not in result:
675
pending.add(revision_id)
679
def get_revision_graph_with_ghosts(self, revision_ids=None):
680
"""Return a graph of the revisions with ghosts marked as applicable.
682
:param revision_ids: an iterable of revisions to graph or None for all.
683
:return: a Graph object with the graph reachable from revision_ids.
685
result = graph.Graph()
687
pending = set(self.all_revision_ids())
690
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
691
# special case NULL_REVISION
692
if _mod_revision.NULL_REVISION in pending:
693
pending.remove(_mod_revision.NULL_REVISION)
694
required = set(pending)
697
revision_id = pending.pop()
699
rev = self.get_revision(revision_id)
700
except errors.NoSuchRevision:
701
if revision_id in required:
704
result.add_ghost(revision_id)
706
for parent_id in rev.parent_ids:
707
# is this queued or done ?
708
if (parent_id not in pending and
709
parent_id not in done):
711
pending.add(parent_id)
712
result.add_node(revision_id, rev.parent_ids)
713
done.add(revision_id)
716
def _get_history_vf(self):
717
"""Get a versionedfile whose history graph reflects all revisions.
719
For weave repositories, this is the inventory weave.
721
return self.get_inventory_weave()
723
def iter_reverse_revision_history(self, revision_id):
724
"""Iterate backwards through revision ids in the lefthand history
726
:param revision_id: The revision id to start with. All its lefthand
727
ancestors will be traversed.
729
revision_id = osutils.safe_revision_id(revision_id)
730
if revision_id in (None, _mod_revision.NULL_REVISION):
732
next_id = revision_id
733
versionedfile = self._get_history_vf()
736
parents = versionedfile.get_parents(next_id)
737
if len(parents) == 0:
743
def get_revision_inventory(self, revision_id):
744
"""Return inventory of a past revision."""
745
# TODO: Unify this with get_inventory()
746
# bzr 0.0.6 and later imposes the constraint that the inventory_id
747
# must be the same as its revision, so this is trivial.
748
if revision_id is None:
749
# This does not make sense: if there is no revision,
750
# then it is the current tree inventory surely ?!
751
# and thus get_root_id() is something that looks at the last
752
# commit on the branch, and the get_root_id is an inventory check.
753
raise NotImplementedError
754
# return Inventory(self.get_root_id())
756
return self.get_inventory(revision_id)
760
"""Return True if this repository is flagged as a shared repository."""
761
raise NotImplementedError(self.is_shared)
764
def reconcile(self, other=None, thorough=False):
765
"""Reconcile this repository."""
766
from bzrlib.reconcile import RepoReconciler
767
reconciler = RepoReconciler(self, thorough=thorough)
768
reconciler.reconcile()
772
def revision_tree(self, revision_id):
773
"""Return Tree for a revision on this branch.
775
`revision_id` may be None for the empty tree revision.
777
# TODO: refactor this to use an existing revision object
778
# so we don't need to read it in twice.
779
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
780
return RevisionTree(self, Inventory(root_id=None),
781
_mod_revision.NULL_REVISION)
783
revision_id = osutils.safe_revision_id(revision_id)
784
inv = self.get_revision_inventory(revision_id)
785
return RevisionTree(self, inv, revision_id)
788
def revision_trees(self, revision_ids):
789
"""Return Tree for a revision on this branch.
791
`revision_id` may not be None or 'null:'"""
792
assert None not in revision_ids
793
assert _mod_revision.NULL_REVISION not in revision_ids
794
texts = self.get_inventory_weave().get_texts(revision_ids)
795
for text, revision_id in zip(texts, revision_ids):
796
inv = self.deserialise_inventory(revision_id, text)
797
yield RevisionTree(self, inv, revision_id)
800
def get_ancestry(self, revision_id):
801
"""Return a list of revision-ids integrated by a revision.
803
The first element of the list is always None, indicating the origin
804
revision. This might change when we have history horizons, or
805
perhaps we should have a new API.
807
This is topologically sorted.
809
if revision_id is None:
811
revision_id = osutils.safe_revision_id(revision_id)
812
if not self.has_revision(revision_id):
813
raise errors.NoSuchRevision(self, revision_id)
814
w = self.get_inventory_weave()
815
candidates = w.get_ancestry(revision_id)
816
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
819
def print_file(self, file, revision_id):
820
"""Print `file` to stdout.
822
FIXME RBC 20060125 as John Meinel points out this is a bad api
823
- it writes to stdout, it assumes that that is valid etc. Fix
824
by creating a new more flexible convenience function.
826
revision_id = osutils.safe_revision_id(revision_id)
827
tree = self.revision_tree(revision_id)
828
# use inventory as it was in that revision
829
file_id = tree.inventory.path2id(file)
831
# TODO: jam 20060427 Write a test for this code path
832
# it had a bug in it, and was raising the wrong
834
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
835
tree.print_file(file_id)
837
def get_transaction(self):
838
return self.control_files.get_transaction()
840
def revision_parents(self, revision_id):
841
revision_id = osutils.safe_revision_id(revision_id)
842
return self.get_inventory_weave().parent_names(revision_id)
845
def set_make_working_trees(self, new_value):
846
"""Set the policy flag for making working trees when creating branches.
848
This only applies to branches that use this repository.
850
The default is 'True'.
851
:param new_value: True to restore the default, False to disable making
854
raise NotImplementedError(self.set_make_working_trees)
856
def make_working_trees(self):
857
"""Returns the policy for making working trees on new branches."""
858
raise NotImplementedError(self.make_working_trees)
861
def sign_revision(self, revision_id, gpg_strategy):
862
revision_id = osutils.safe_revision_id(revision_id)
863
plaintext = Testament.from_revision(self, revision_id).as_short_text()
864
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
867
def has_signature_for_revision_id(self, revision_id):
868
"""Query for a revision signature for revision_id in the repository."""
869
revision_id = osutils.safe_revision_id(revision_id)
870
return self._revision_store.has_signature(revision_id,
871
self.get_transaction())
874
def get_signature_text(self, revision_id):
875
"""Return the text for a signature."""
876
revision_id = osutils.safe_revision_id(revision_id)
877
return self._revision_store.get_signature_text(revision_id,
878
self.get_transaction())
881
def check(self, revision_ids):
882
"""Check consistency of all history of given revision_ids.
884
Different repository implementations should override _check().
886
:param revision_ids: A non-empty list of revision_ids whose ancestry
887
will be checked. Typically the last revision_id of a branch.
890
raise ValueError("revision_ids must be non-empty in %s.check"
892
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
893
return self._check(revision_ids)
895
def _check(self, revision_ids):
896
result = check.Check(self)
900
def _warn_if_deprecated(self):
901
global _deprecation_warning_done
902
if _deprecation_warning_done:
904
_deprecation_warning_done = True
905
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
906
% (self._format, self.bzrdir.transport.base))
908
def supports_rich_root(self):
909
return self._format.rich_root_data
911
def _check_ascii_revisionid(self, revision_id, method):
912
"""Private helper for ascii-only repositories."""
913
# weave repositories refuse to store revisionids that are non-ascii.
914
if revision_id is not None:
915
# weaves require ascii revision ids.
916
if isinstance(revision_id, unicode):
918
revision_id.encode('ascii')
919
except UnicodeEncodeError:
920
raise errors.NonAsciiRevisionId(method, self)
923
revision_id.decode('ascii')
924
except UnicodeDecodeError:
925
raise errors.NonAsciiRevisionId(method, self)
929
# remove these delegates a while after bzr 0.15
930
def __make_delegated(name, from_module):
931
def _deprecated_repository_forwarder():
932
symbol_versioning.warn('%s moved to %s in bzr 0.15'
933
% (name, from_module),
936
m = __import__(from_module, globals(), locals(), [name])
938
return getattr(m, name)
939
except AttributeError:
940
raise AttributeError('module %s has no name %s'
942
globals()[name] = _deprecated_repository_forwarder
945
'AllInOneRepository',
946
'WeaveMetaDirRepository',
947
'PreSplitOutRepositoryFormat',
953
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
957
'RepositoryFormatKnit',
958
'RepositoryFormatKnit1',
960
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
963
def install_revision(repository, rev, revision_tree):
964
"""Install all revision data into a repository."""
967
for p_id in rev.parent_ids:
968
if repository.has_revision(p_id):
969
present_parents.append(p_id)
970
parent_trees[p_id] = repository.revision_tree(p_id)
972
parent_trees[p_id] = repository.revision_tree(None)
974
inv = revision_tree.inventory
975
entries = inv.iter_entries()
976
# backwards compatability hack: skip the root id.
977
if not repository.supports_rich_root():
978
path, root = entries.next()
979
if root.revision != rev.revision_id:
980
raise errors.IncompatibleRevision(repr(repository))
981
# Add the texts that are not already present
982
for path, ie in entries:
983
w = repository.weave_store.get_weave_or_empty(ie.file_id,
984
repository.get_transaction())
985
if ie.revision not in w:
987
# FIXME: TODO: The following loop *may* be overlapping/duplicate
988
# with InventoryEntry.find_previous_heads(). if it is, then there
989
# is a latent bug here where the parents may have ancestors of each
991
for revision, tree in parent_trees.iteritems():
992
if ie.file_id not in tree:
994
parent_id = tree.inventory[ie.file_id].revision
995
if parent_id in text_parents:
997
text_parents.append(parent_id)
999
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1000
repository.get_transaction())
1001
lines = revision_tree.get_file(ie.file_id).readlines()
1002
vfile.add_lines(rev.revision_id, text_parents, lines)
1004
# install the inventory
1005
repository.add_inventory(rev.revision_id, inv, present_parents)
1006
except errors.RevisionAlreadyPresent:
1008
repository.add_revision(rev.revision_id, rev, inv)
1011
class MetaDirRepository(Repository):
1012
"""Repositories in the new meta-dir layout."""
1014
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1015
super(MetaDirRepository, self).__init__(_format,
1021
dir_mode = self.control_files._dir_mode
1022
file_mode = self.control_files._file_mode
1025
def is_shared(self):
1026
"""Return True if this repository is flagged as a shared repository."""
1027
return self.control_files._transport.has('shared-storage')
1030
def set_make_working_trees(self, new_value):
1031
"""Set the policy flag for making working trees when creating branches.
1033
This only applies to branches that use this repository.
1035
The default is 'True'.
1036
:param new_value: True to restore the default, False to disable making
1041
self.control_files._transport.delete('no-working-trees')
1042
except errors.NoSuchFile:
1045
self.control_files.put_utf8('no-working-trees', '')
1047
def make_working_trees(self):
1048
"""Returns the policy for making working trees on new branches."""
1049
return not self.control_files._transport.has('no-working-trees')
1052
class RepositoryFormatRegistry(registry.Registry):
1053
"""Registry of RepositoryFormats.
1056
def get(self, format_string):
1057
r = registry.Registry.get(self, format_string)
1063
format_registry = RepositoryFormatRegistry()
1064
"""Registry of formats, indexed by their identifying format string.
1066
This can contain either format instances themselves, or classes/factories that
1067
can be called to obtain one.
1071
#####################################################################
1072
# Repository Formats
1074
class RepositoryFormat(object):
1075
"""A repository format.
1077
Formats provide three things:
1078
* An initialization routine to construct repository data on disk.
1079
* a format string which is used when the BzrDir supports versioned
1081
* an open routine which returns a Repository instance.
1083
Formats are placed in an dict by their format string for reference
1084
during opening. These should be subclasses of RepositoryFormat
1087
Once a format is deprecated, just deprecate the initialize and open
1088
methods on the format class. Do not deprecate the object, as the
1089
object will be created every system load.
1091
Common instance attributes:
1092
_matchingbzrdir - the bzrdir format that the repository format was
1093
originally written to work with. This can be used if manually
1094
constructing a bzrdir and repository, or more commonly for test suite
1099
return "<%s>" % self.__class__.__name__
1101
def __eq__(self, other):
1102
# format objects are generally stateless
1103
return isinstance(other, self.__class__)
1105
def __ne__(self, other):
1106
return not self == other
1109
def find_format(klass, a_bzrdir):
1110
"""Return the format for the repository object in a_bzrdir.
1112
This is used by bzr native formats that have a "format" file in
1113
the repository. Other methods may be used by different types of
1117
transport = a_bzrdir.get_repository_transport(None)
1118
format_string = transport.get("format").read()
1119
return format_registry.get(format_string)
1120
except errors.NoSuchFile:
1121
raise errors.NoRepositoryPresent(a_bzrdir)
1123
raise errors.UnknownFormatError(format=format_string)
1126
def register_format(klass, format):
1127
format_registry.register(format.get_format_string(), format)
1130
def unregister_format(klass, format):
1131
format_registry.remove(format.get_format_string())
1134
def get_default_format(klass):
1135
"""Return the current default format."""
1136
from bzrlib import bzrdir
1137
return bzrdir.format_registry.make_bzrdir('default').repository_format
1139
def _get_control_store(self, repo_transport, control_files):
1140
"""Return the control store for this repository."""
1141
raise NotImplementedError(self._get_control_store)
1143
def get_format_string(self):
1144
"""Return the ASCII format string that identifies this format.
1146
Note that in pre format ?? repositories the format string is
1147
not permitted nor written to disk.
1149
raise NotImplementedError(self.get_format_string)
1151
def get_format_description(self):
1152
"""Return the short description for this format."""
1153
raise NotImplementedError(self.get_format_description)
1155
def _get_revision_store(self, repo_transport, control_files):
1156
"""Return the revision store object for this a_bzrdir."""
1157
raise NotImplementedError(self._get_revision_store)
1159
def _get_text_rev_store(self,
1166
"""Common logic for getting a revision store for a repository.
1168
see self._get_revision_store for the subclass-overridable method to
1169
get the store for a repository.
1171
from bzrlib.store.revision.text import TextRevisionStore
1172
dir_mode = control_files._dir_mode
1173
file_mode = control_files._file_mode
1174
text_store = TextStore(transport.clone(name),
1176
compressed=compressed,
1178
file_mode=file_mode)
1179
_revision_store = TextRevisionStore(text_store, serializer)
1180
return _revision_store
1182
# TODO: this shouldn't be in the base class, it's specific to things that
1183
# use weaves or knits -- mbp 20070207
1184
def _get_versioned_file_store(self,
1189
versionedfile_class=None,
1190
versionedfile_kwargs={},
1192
if versionedfile_class is None:
1193
versionedfile_class = self._versionedfile_class
1194
weave_transport = control_files._transport.clone(name)
1195
dir_mode = control_files._dir_mode
1196
file_mode = control_files._file_mode
1197
return VersionedFileStore(weave_transport, prefixed=prefixed,
1199
file_mode=file_mode,
1200
versionedfile_class=versionedfile_class,
1201
versionedfile_kwargs=versionedfile_kwargs,
1204
def initialize(self, a_bzrdir, shared=False):
1205
"""Initialize a repository of this format in a_bzrdir.
1207
:param a_bzrdir: The bzrdir to put the new repository in it.
1208
:param shared: The repository should be initialized as a sharable one.
1209
:returns: The new repository object.
1211
This may raise UninitializableFormat if shared repository are not
1212
compatible the a_bzrdir.
1214
raise NotImplementedError(self.initialize)
1216
def is_supported(self):
1217
"""Is this format supported?
1219
Supported formats must be initializable and openable.
1220
Unsupported formats may not support initialization or committing or
1221
some other features depending on the reason for not being supported.
1225
def check_conversion_target(self, target_format):
1226
raise NotImplementedError(self.check_conversion_target)
1228
def open(self, a_bzrdir, _found=False):
1229
"""Return an instance of this format for the bzrdir a_bzrdir.
1231
_found is a private parameter, do not use it.
1233
raise NotImplementedError(self.open)
1236
class MetaDirRepositoryFormat(RepositoryFormat):
1237
"""Common base class for the new repositories using the metadir layout."""
1239
rich_root_data = False
1240
supports_tree_reference = False
1241
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1244
super(MetaDirRepositoryFormat, self).__init__()
1246
def _create_control_files(self, a_bzrdir):
1247
"""Create the required files and the initial control_files object."""
1248
# FIXME: RBC 20060125 don't peek under the covers
1249
# NB: no need to escape relative paths that are url safe.
1250
repository_transport = a_bzrdir.get_repository_transport(self)
1251
control_files = lockable_files.LockableFiles(repository_transport,
1252
'lock', lockdir.LockDir)
1253
control_files.create_lock()
1254
return control_files
1256
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1257
"""Upload the initial blank content."""
1258
control_files = self._create_control_files(a_bzrdir)
1259
control_files.lock_write()
1261
control_files._transport.mkdir_multi(dirs,
1262
mode=control_files._dir_mode)
1263
for file, content in files:
1264
control_files.put(file, content)
1265
for file, content in utf8_files:
1266
control_files.put_utf8(file, content)
1268
control_files.put_utf8('shared-storage', '')
1270
control_files.unlock()
1273
# formats which have no format string are not discoverable
1274
# and not independently creatable, so are not registered. They're
1275
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1276
# needed, it's constructed directly by the BzrDir. Non-native formats where
1277
# the repository is not separately opened are similar.
1279
format_registry.register_lazy(
1280
'Bazaar-NG Repository format 7',
1281
'bzrlib.repofmt.weaverepo',
1284
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1285
# default control directory format
1287
format_registry.register_lazy(
1288
'Bazaar-NG Knit Repository Format 1',
1289
'bzrlib.repofmt.knitrepo',
1290
'RepositoryFormatKnit1',
1292
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1294
format_registry.register_lazy(
1295
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1296
'bzrlib.repofmt.knitrepo',
1297
'RepositoryFormatKnit3',
1301
class InterRepository(InterObject):
1302
"""This class represents operations taking place between two repositories.
1304
Its instances have methods like copy_content and fetch, and contain
1305
references to the source and target repositories these operations can be
1308
Often we will provide convenience methods on 'repository' which carry out
1309
operations with another repository - they will always forward to
1310
InterRepository.get(other).method_name(parameters).
1314
"""The available optimised InterRepository types."""
1316
def copy_content(self, revision_id=None):
1317
raise NotImplementedError(self.copy_content)
1319
def fetch(self, revision_id=None, pb=None):
1320
"""Fetch the content required to construct revision_id.
1322
The content is copied from self.source to self.target.
1324
:param revision_id: if None all content is copied, if NULL_REVISION no
1326
:param pb: optional progress bar to use for progress reports. If not
1327
provided a default one will be created.
1329
Returns the copied revision count and the failed revisions in a tuple:
1332
raise NotImplementedError(self.fetch)
1335
def missing_revision_ids(self, revision_id=None):
1336
"""Return the revision ids that source has that target does not.
1338
These are returned in topological order.
1340
:param revision_id: only return revision ids included by this
1343
# generic, possibly worst case, slow code path.
1344
target_ids = set(self.target.all_revision_ids())
1345
if revision_id is not None:
1346
# TODO: jam 20070210 InterRepository is internal enough that it
1347
# should assume revision_ids are already utf-8
1348
revision_id = osutils.safe_revision_id(revision_id)
1349
source_ids = self.source.get_ancestry(revision_id)
1350
assert source_ids[0] is None
1353
source_ids = self.source.all_revision_ids()
1354
result_set = set(source_ids).difference(target_ids)
1355
# this may look like a no-op: its not. It preserves the ordering
1356
# other_ids had while only returning the members from other_ids
1357
# that we've decided we need.
1358
return [rev_id for rev_id in source_ids if rev_id in result_set]
1361
class InterSameDataRepository(InterRepository):
1362
"""Code for converting between repositories that represent the same data.
1364
Data format and model must match for this to work.
1368
def _get_repo_format_to_test(self):
1369
"""Repository format for testing with."""
1370
return RepositoryFormat.get_default_format()
1373
def is_compatible(source, target):
1374
if source.supports_rich_root() != target.supports_rich_root():
1376
if source._serializer != target._serializer:
1381
def copy_content(self, revision_id=None):
1382
"""Make a complete copy of the content in self into destination.
1384
This is a destructive operation! Do not use it on existing
1387
:param revision_id: Only copy the content needed to construct
1388
revision_id and its parents.
1391
self.target.set_make_working_trees(self.source.make_working_trees())
1392
except NotImplementedError:
1394
# TODO: jam 20070210 This is fairly internal, so we should probably
1395
# just assert that revision_id is not unicode.
1396
revision_id = osutils.safe_revision_id(revision_id)
1397
# but don't bother fetching if we have the needed data now.
1398
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1399
self.target.has_revision(revision_id)):
1401
self.target.fetch(self.source, revision_id=revision_id)
1404
def fetch(self, revision_id=None, pb=None):
1405
"""See InterRepository.fetch()."""
1406
from bzrlib.fetch import GenericRepoFetcher
1407
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1408
self.source, self.source._format, self.target,
1409
self.target._format)
1410
# TODO: jam 20070210 This should be an assert, not a translate
1411
revision_id = osutils.safe_revision_id(revision_id)
1412
f = GenericRepoFetcher(to_repository=self.target,
1413
from_repository=self.source,
1414
last_revision=revision_id,
1416
return f.count_copied, f.failed_revisions
1419
class InterWeaveRepo(InterSameDataRepository):
1420
"""Optimised code paths between Weave based repositories."""
1423
def _get_repo_format_to_test(self):
1424
from bzrlib.repofmt import weaverepo
1425
return weaverepo.RepositoryFormat7()
1428
def is_compatible(source, target):
1429
"""Be compatible with known Weave formats.
1431
We don't test for the stores being of specific types because that
1432
could lead to confusing results, and there is no need to be
1435
from bzrlib.repofmt.weaverepo import (
1441
return (isinstance(source._format, (RepositoryFormat5,
1443
RepositoryFormat7)) and
1444
isinstance(target._format, (RepositoryFormat5,
1446
RepositoryFormat7)))
1447
except AttributeError:
1451
def copy_content(self, revision_id=None):
1452
"""See InterRepository.copy_content()."""
1453
# weave specific optimised path:
1454
# TODO: jam 20070210 Internal, should be an assert, not translate
1455
revision_id = osutils.safe_revision_id(revision_id)
1457
self.target.set_make_working_trees(self.source.make_working_trees())
1458
except NotImplementedError:
1460
# FIXME do not peek!
1461
if self.source.control_files._transport.listable():
1462
pb = ui.ui_factory.nested_progress_bar()
1464
self.target.weave_store.copy_all_ids(
1465
self.source.weave_store,
1467
from_transaction=self.source.get_transaction(),
1468
to_transaction=self.target.get_transaction())
1469
pb.update('copying inventory', 0, 1)
1470
self.target.control_weaves.copy_multi(
1471
self.source.control_weaves, ['inventory'],
1472
from_transaction=self.source.get_transaction(),
1473
to_transaction=self.target.get_transaction())
1474
self.target._revision_store.text_store.copy_all_ids(
1475
self.source._revision_store.text_store,
1480
self.target.fetch(self.source, revision_id=revision_id)
1483
def fetch(self, revision_id=None, pb=None):
1484
"""See InterRepository.fetch()."""
1485
from bzrlib.fetch import GenericRepoFetcher
1486
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1487
self.source, self.source._format, self.target, self.target._format)
1488
# TODO: jam 20070210 This should be an assert, not a translate
1489
revision_id = osutils.safe_revision_id(revision_id)
1490
f = GenericRepoFetcher(to_repository=self.target,
1491
from_repository=self.source,
1492
last_revision=revision_id,
1494
return f.count_copied, f.failed_revisions
1497
def missing_revision_ids(self, revision_id=None):
1498
"""See InterRepository.missing_revision_ids()."""
1499
# we want all revisions to satisfy revision_id in source.
1500
# but we don't want to stat every file here and there.
1501
# we want then, all revisions other needs to satisfy revision_id
1502
# checked, but not those that we have locally.
1503
# so the first thing is to get a subset of the revisions to
1504
# satisfy revision_id in source, and then eliminate those that
1505
# we do already have.
1506
# this is slow on high latency connection to self, but as as this
1507
# disk format scales terribly for push anyway due to rewriting
1508
# inventory.weave, this is considered acceptable.
1510
if revision_id is not None:
1511
source_ids = self.source.get_ancestry(revision_id)
1512
assert source_ids[0] is None
1515
source_ids = self.source._all_possible_ids()
1516
source_ids_set = set(source_ids)
1517
# source_ids is the worst possible case we may need to pull.
1518
# now we want to filter source_ids against what we actually
1519
# have in target, but don't try to check for existence where we know
1520
# we do not have a revision as that would be pointless.
1521
target_ids = set(self.target._all_possible_ids())
1522
possibly_present_revisions = target_ids.intersection(source_ids_set)
1523
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1524
required_revisions = source_ids_set.difference(actually_present_revisions)
1525
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1526
if revision_id is not None:
1527
# we used get_ancestry to determine source_ids then we are assured all
1528
# revisions referenced are present as they are installed in topological order.
1529
# and the tip revision was validated by get_ancestry.
1530
return required_topo_revisions
1532
# if we just grabbed the possibly available ids, then
1533
# we only have an estimate of whats available and need to validate
1534
# that against the revision records.
1535
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1538
class InterKnitRepo(InterSameDataRepository):
1539
"""Optimised code paths between Knit based repositories."""
1542
def _get_repo_format_to_test(self):
1543
from bzrlib.repofmt import knitrepo
1544
return knitrepo.RepositoryFormatKnit1()
1547
def is_compatible(source, target):
1548
"""Be compatible with known Knit formats.
1550
We don't test for the stores being of specific types because that
1551
could lead to confusing results, and there is no need to be
1554
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1556
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1557
isinstance(target._format, (RepositoryFormatKnit1)))
1558
except AttributeError:
1562
def fetch(self, revision_id=None, pb=None):
1563
"""See InterRepository.fetch()."""
1564
from bzrlib.fetch import KnitRepoFetcher
1565
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1566
self.source, self.source._format, self.target, self.target._format)
1567
# TODO: jam 20070210 This should be an assert, not a translate
1568
revision_id = osutils.safe_revision_id(revision_id)
1569
f = KnitRepoFetcher(to_repository=self.target,
1570
from_repository=self.source,
1571
last_revision=revision_id,
1573
return f.count_copied, f.failed_revisions
1576
def missing_revision_ids(self, revision_id=None):
1577
"""See InterRepository.missing_revision_ids()."""
1578
if revision_id is not None:
1579
source_ids = self.source.get_ancestry(revision_id)
1580
assert source_ids[0] is None
1583
source_ids = self.source._all_possible_ids()
1584
source_ids_set = set(source_ids)
1585
# source_ids is the worst possible case we may need to pull.
1586
# now we want to filter source_ids against what we actually
1587
# have in target, but don't try to check for existence where we know
1588
# we do not have a revision as that would be pointless.
1589
target_ids = set(self.target._all_possible_ids())
1590
possibly_present_revisions = target_ids.intersection(source_ids_set)
1591
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1592
required_revisions = source_ids_set.difference(actually_present_revisions)
1593
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1594
if revision_id is not None:
1595
# we used get_ancestry to determine source_ids then we are assured all
1596
# revisions referenced are present as they are installed in topological order.
1597
# and the tip revision was validated by get_ancestry.
1598
return required_topo_revisions
1600
# if we just grabbed the possibly available ids, then
1601
# we only have an estimate of whats available and need to validate
1602
# that against the revision records.
1603
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1606
class InterModel1and2(InterRepository):
1609
def _get_repo_format_to_test(self):
1613
def is_compatible(source, target):
1614
if not source.supports_rich_root() and target.supports_rich_root():
1620
def fetch(self, revision_id=None, pb=None):
1621
"""See InterRepository.fetch()."""
1622
from bzrlib.fetch import Model1toKnit2Fetcher
1623
# TODO: jam 20070210 This should be an assert, not a translate
1624
revision_id = osutils.safe_revision_id(revision_id)
1625
f = Model1toKnit2Fetcher(to_repository=self.target,
1626
from_repository=self.source,
1627
last_revision=revision_id,
1629
return f.count_copied, f.failed_revisions
1632
def copy_content(self, revision_id=None):
1633
"""Make a complete copy of the content in self into destination.
1635
This is a destructive operation! Do not use it on existing
1638
:param revision_id: Only copy the content needed to construct
1639
revision_id and its parents.
1642
self.target.set_make_working_trees(self.source.make_working_trees())
1643
except NotImplementedError:
1645
# TODO: jam 20070210 Internal, assert, don't translate
1646
revision_id = osutils.safe_revision_id(revision_id)
1647
# but don't bother fetching if we have the needed data now.
1648
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1649
self.target.has_revision(revision_id)):
1651
self.target.fetch(self.source, revision_id=revision_id)
1654
class InterKnit1and2(InterKnitRepo):
1657
def _get_repo_format_to_test(self):
1661
def is_compatible(source, target):
1662
"""Be compatible with Knit1 source and Knit3 target"""
1663
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1665
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1666
RepositoryFormatKnit3
1667
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1668
isinstance(target._format, (RepositoryFormatKnit3)))
1669
except AttributeError:
1673
def fetch(self, revision_id=None, pb=None):
1674
"""See InterRepository.fetch()."""
1675
from bzrlib.fetch import Knit1to2Fetcher
1676
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1677
self.source, self.source._format, self.target,
1678
self.target._format)
1679
# TODO: jam 20070210 This should be an assert, not a translate
1680
revision_id = osutils.safe_revision_id(revision_id)
1681
f = Knit1to2Fetcher(to_repository=self.target,
1682
from_repository=self.source,
1683
last_revision=revision_id,
1685
return f.count_copied, f.failed_revisions
1688
InterRepository.register_optimiser(InterSameDataRepository)
1689
InterRepository.register_optimiser(InterWeaveRepo)
1690
InterRepository.register_optimiser(InterKnitRepo)
1691
InterRepository.register_optimiser(InterModel1and2)
1692
InterRepository.register_optimiser(InterKnit1and2)
1695
class RepositoryTestProviderAdapter(object):
1696
"""A tool to generate a suite testing multiple repository formats at once.
1698
This is done by copying the test once for each transport and injecting
1699
the transport_server, transport_readonly_server, and bzrdir_format and
1700
repository_format classes into each copy. Each copy is also given a new id()
1701
to make it easy to identify.
1704
def __init__(self, transport_server, transport_readonly_server, formats,
1705
vfs_transport_factory=None):
1706
self._transport_server = transport_server
1707
self._transport_readonly_server = transport_readonly_server
1708
self._vfs_transport_factory = vfs_transport_factory
1709
self._formats = formats
1711
def adapt(self, test):
1712
result = unittest.TestSuite()
1713
for repository_format, bzrdir_format in self._formats:
1714
from copy import deepcopy
1715
new_test = deepcopy(test)
1716
new_test.transport_server = self._transport_server
1717
new_test.transport_readonly_server = self._transport_readonly_server
1718
if self._vfs_transport_factory:
1719
new_test.vfs_transport_factory = self._vfs_transport_factory
1720
new_test.bzrdir_format = bzrdir_format
1721
new_test.repository_format = repository_format
1722
def make_new_test_id():
1723
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1724
return lambda: new_id
1725
new_test.id = make_new_test_id()
1726
result.addTest(new_test)
1730
class InterRepositoryTestProviderAdapter(object):
1731
"""A tool to generate a suite testing multiple inter repository formats.
1733
This is done by copying the test once for each interrepo provider and injecting
1734
the transport_server, transport_readonly_server, repository_format and
1735
repository_to_format classes into each copy.
1736
Each copy is also given a new id() to make it easy to identify.
1739
def __init__(self, transport_server, transport_readonly_server, formats):
1740
self._transport_server = transport_server
1741
self._transport_readonly_server = transport_readonly_server
1742
self._formats = formats
1744
def adapt(self, test):
1745
result = unittest.TestSuite()
1746
for interrepo_class, repository_format, repository_format_to in self._formats:
1747
from copy import deepcopy
1748
new_test = deepcopy(test)
1749
new_test.transport_server = self._transport_server
1750
new_test.transport_readonly_server = self._transport_readonly_server
1751
new_test.interrepo_class = interrepo_class
1752
new_test.repository_format = repository_format
1753
new_test.repository_format_to = repository_format_to
1754
def make_new_test_id():
1755
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1756
return lambda: new_id
1757
new_test.id = make_new_test_id()
1758
result.addTest(new_test)
1762
def default_test_list():
1763
"""Generate the default list of interrepo permutations to test."""
1764
from bzrlib.repofmt import knitrepo, weaverepo
1766
# test the default InterRepository between format 6 and the current
1768
# XXX: robertc 20060220 reinstate this when there are two supported
1769
# formats which do not have an optimal code path between them.
1770
#result.append((InterRepository,
1771
# RepositoryFormat6(),
1772
# RepositoryFormatKnit1()))
1773
for optimiser_class in InterRepository._optimisers:
1774
format_to_test = optimiser_class._get_repo_format_to_test()
1775
if format_to_test is not None:
1776
result.append((optimiser_class,
1777
format_to_test, format_to_test))
1778
# if there are specific combinations we want to use, we can add them
1780
result.append((InterModel1and2,
1781
weaverepo.RepositoryFormat5(),
1782
knitrepo.RepositoryFormatKnit3()))
1783
result.append((InterKnit1and2,
1784
knitrepo.RepositoryFormatKnit1(),
1785
knitrepo.RepositoryFormatKnit3()))
1789
class CopyConverter(object):
1790
"""A repository conversion tool which just performs a copy of the content.
1792
This is slow but quite reliable.
1795
def __init__(self, target_format):
1796
"""Create a CopyConverter.
1798
:param target_format: The format the resulting repository should be.
1800
self.target_format = target_format
1802
def convert(self, repo, pb):
1803
"""Perform the conversion of to_convert, giving feedback via pb.
1805
:param to_convert: The disk object to convert.
1806
:param pb: a progress bar to use for progress information.
1811
# this is only useful with metadir layouts - separated repo content.
1812
# trigger an assertion if not such
1813
repo._format.get_format_string()
1814
self.repo_dir = repo.bzrdir
1815
self.step('Moving repository to repository.backup')
1816
self.repo_dir.transport.move('repository', 'repository.backup')
1817
backup_transport = self.repo_dir.transport.clone('repository.backup')
1818
repo._format.check_conversion_target(self.target_format)
1819
self.source_repo = repo._format.open(self.repo_dir,
1821
_override_transport=backup_transport)
1822
self.step('Creating new repository')
1823
converted = self.target_format.initialize(self.repo_dir,
1824
self.source_repo.is_shared())
1825
converted.lock_write()
1827
self.step('Copying content into repository.')
1828
self.source_repo.copy_content_into(converted)
1831
self.step('Deleting old repository content.')
1832
self.repo_dir.transport.delete_tree('repository.backup')
1833
self.pb.note('repository converted')
1835
def step(self, message):
1836
"""Update the pb by a step."""
1838
self.pb.update(message, self.count, self.total)
1841
class CommitBuilder(object):
1842
"""Provides an interface to build up a commit.
1844
This allows describing a tree to be committed without needing to
1845
know the internals of the format of the repository.
1848
record_root_entry = False
1849
def __init__(self, repository, parents, config, timestamp=None,
1850
timezone=None, committer=None, revprops=None,
1852
"""Initiate a CommitBuilder.
1854
:param repository: Repository to commit to.
1855
:param parents: Revision ids of the parents of the new revision.
1856
:param config: Configuration to use.
1857
:param timestamp: Optional timestamp recorded for commit.
1858
:param timezone: Optional timezone for timestamp.
1859
:param committer: Optional committer to set for commit.
1860
:param revprops: Optional dictionary of revision properties.
1861
:param revision_id: Optional revision id.
1863
self._config = config
1865
if committer is None:
1866
self._committer = self._config.username()
1868
assert isinstance(committer, basestring), type(committer)
1869
self._committer = committer
1871
self.new_inventory = Inventory(None)
1872
self._new_revision_id = osutils.safe_revision_id(revision_id)
1873
self.parents = parents
1874
self.repository = repository
1877
if revprops is not None:
1878
self._revprops.update(revprops)
1880
if timestamp is None:
1881
timestamp = time.time()
1882
# Restrict resolution to 1ms
1883
self._timestamp = round(timestamp, 3)
1885
if timezone is None:
1886
self._timezone = osutils.local_time_offset()
1888
self._timezone = int(timezone)
1890
self._generate_revision_if_needed()
1892
def commit(self, message):
1893
"""Make the actual commit.
1895
:return: The revision id of the recorded revision.
1897
rev = _mod_revision.Revision(
1898
timestamp=self._timestamp,
1899
timezone=self._timezone,
1900
committer=self._committer,
1902
inventory_sha1=self.inv_sha1,
1903
revision_id=self._new_revision_id,
1904
properties=self._revprops)
1905
rev.parent_ids = self.parents
1906
self.repository.add_revision(self._new_revision_id, rev,
1907
self.new_inventory, self._config)
1908
return self._new_revision_id
1910
def revision_tree(self):
1911
"""Return the tree that was just committed.
1913
After calling commit() this can be called to get a RevisionTree
1914
representing the newly committed tree. This is preferred to
1915
calling Repository.revision_tree() because that may require
1916
deserializing the inventory, while we already have a copy in
1919
return RevisionTree(self.repository, self.new_inventory,
1920
self._new_revision_id)
1922
def finish_inventory(self):
1923
"""Tell the builder that the inventory is finished."""
1924
if self.new_inventory.root is None:
1925
symbol_versioning.warn('Root entry should be supplied to'
1926
' record_entry_contents, as of bzr 0.10.',
1927
DeprecationWarning, stacklevel=2)
1928
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1929
self.new_inventory.revision_id = self._new_revision_id
1930
self.inv_sha1 = self.repository.add_inventory(
1931
self._new_revision_id,
1936
def _gen_revision_id(self):
1937
"""Return new revision-id."""
1938
return generate_ids.gen_revision_id(self._config.username(),
1941
def _generate_revision_if_needed(self):
1942
"""Create a revision id if None was supplied.
1944
If the repository can not support user-specified revision ids
1945
they should override this function and raise CannotSetRevisionId
1946
if _new_revision_id is not None.
1948
:raises: CannotSetRevisionId
1950
if self._new_revision_id is None:
1951
self._new_revision_id = self._gen_revision_id()
1953
def record_entry_contents(self, ie, parent_invs, path, tree):
1954
"""Record the content of ie from tree into the commit if needed.
1956
Side effect: sets ie.revision when unchanged
1958
:param ie: An inventory entry present in the commit.
1959
:param parent_invs: The inventories of the parent revisions of the
1961
:param path: The path the entry is at in the tree.
1962
:param tree: The tree which contains this entry and should be used to
1965
if self.new_inventory.root is None and ie.parent_id is not None:
1966
symbol_versioning.warn('Root entry should be supplied to'
1967
' record_entry_contents, as of bzr 0.10.',
1968
DeprecationWarning, stacklevel=2)
1969
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1971
self.new_inventory.add(ie)
1973
# ie.revision is always None if the InventoryEntry is considered
1974
# for committing. ie.snapshot will record the correct revision
1975
# which may be the sole parent if it is untouched.
1976
if ie.revision is not None:
1979
# In this revision format, root entries have no knit or weave
1980
if ie is self.new_inventory.root:
1981
# When serializing out to disk and back in
1982
# root.revision is always _new_revision_id
1983
ie.revision = self._new_revision_id
1985
previous_entries = ie.find_previous_heads(
1987
self.repository.weave_store,
1988
self.repository.get_transaction())
1989
# we are creating a new revision for ie in the history store
1991
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1993
def modified_directory(self, file_id, file_parents):
1994
"""Record the presence of a symbolic link.
1996
:param file_id: The file_id of the link to record.
1997
:param file_parents: The per-file parent revision ids.
1999
self._add_text_to_weave(file_id, [], file_parents.keys())
2001
def modified_reference(self, file_id, file_parents):
2002
"""Record the modification of a reference.
2004
:param file_id: The file_id of the link to record.
2005
:param file_parents: The per-file parent revision ids.
2007
self._add_text_to_weave(file_id, [], file_parents.keys())
2009
def modified_file_text(self, file_id, file_parents,
2010
get_content_byte_lines, text_sha1=None,
2012
"""Record the text of file file_id
2014
:param file_id: The file_id of the file to record the text of.
2015
:param file_parents: The per-file parent revision ids.
2016
:param get_content_byte_lines: A callable which will return the byte
2018
:param text_sha1: Optional SHA1 of the file contents.
2019
:param text_size: Optional size of the file contents.
2021
# mutter('storing text of file {%s} in revision {%s} into %r',
2022
# file_id, self._new_revision_id, self.repository.weave_store)
2023
# special case to avoid diffing on renames or
2025
if (len(file_parents) == 1
2026
and text_sha1 == file_parents.values()[0].text_sha1
2027
and text_size == file_parents.values()[0].text_size):
2028
previous_ie = file_parents.values()[0]
2029
versionedfile = self.repository.weave_store.get_weave(file_id,
2030
self.repository.get_transaction())
2031
versionedfile.clone_text(self._new_revision_id,
2032
previous_ie.revision, file_parents.keys())
2033
return text_sha1, text_size
2035
new_lines = get_content_byte_lines()
2036
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2037
# should return the SHA1 and size
2038
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2039
return osutils.sha_strings(new_lines), \
2040
sum(map(len, new_lines))
2042
def modified_link(self, file_id, file_parents, link_target):
2043
"""Record the presence of a symbolic link.
2045
:param file_id: The file_id of the link to record.
2046
:param file_parents: The per-file parent revision ids.
2047
:param link_target: Target location of this link.
2049
self._add_text_to_weave(file_id, [], file_parents.keys())
2051
def _add_text_to_weave(self, file_id, new_lines, parents):
2052
versionedfile = self.repository.weave_store.get_weave_or_empty(
2053
file_id, self.repository.get_transaction())
2054
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2055
versionedfile.clear_cache()
2058
class _CommitBuilder(CommitBuilder):
2059
"""Temporary class so old CommitBuilders are detected properly
2061
Note: CommitBuilder works whether or not root entry is recorded.
2064
record_root_entry = True
2067
class RootCommitBuilder(CommitBuilder):
2068
"""This commitbuilder actually records the root id"""
2070
record_root_entry = True
2072
def record_entry_contents(self, ie, parent_invs, path, tree):
2073
"""Record the content of ie from tree into the commit if needed.
2075
Side effect: sets ie.revision when unchanged
2077
:param ie: An inventory entry present in the commit.
2078
:param parent_invs: The inventories of the parent revisions of the
2080
:param path: The path the entry is at in the tree.
2081
:param tree: The tree which contains this entry and should be used to
2084
assert self.new_inventory.root is not None or ie.parent_id is None
2085
self.new_inventory.add(ie)
2087
# ie.revision is always None if the InventoryEntry is considered
2088
# for committing. ie.snapshot will record the correct revision
2089
# which may be the sole parent if it is untouched.
2090
if ie.revision is not None:
2093
previous_entries = ie.find_previous_heads(
2095
self.repository.weave_store,
2096
self.repository.get_transaction())
2097
# we are creating a new revision for ie in the history store
2099
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2111
def _unescaper(match, _map=_unescape_map):
2112
code = match.group(1)
2116
if not code.startswith('#'):
2118
return unichr(int(code[1:])).encode('utf8')
2124
def _unescape_xml(data):
2125
"""Unescape predefined XML entities in a string of data."""
2127
if _unescape_re is None:
2128
_unescape_re = re.compile('\&([^;]*);')
2129
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
