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
class Repository(object):
64
"""Repository holding history for one or more branches.
66
The repository holds and retrieves historical information including
67
revisions and file history. It's normally accessed only by the Branch,
68
which views a particular line of development through that history.
70
The Repository builds on top of Stores and a Transport, which respectively
71
describe the disk data format and the way of accessing the (possibly
75
_file_ids_altered_regex = lazy_regex.lazy_compile(
76
r'file_id="(?P<file_id>[^"]+)"'
77
r'.*revision="(?P<revision_id>[^"]+)"'
81
def add_inventory(self, revision_id, inv, parents):
82
"""Add the inventory inv to the repository as revision_id.
84
:param parents: The revision ids of the parents that revision_id
85
is known to have and are in the repository already.
87
returns the sha1 of the serialized inventory.
89
revision_id = osutils.safe_revision_id(revision_id)
90
_mod_revision.check_not_reserved_id(revision_id)
91
assert inv.revision_id is None or inv.revision_id == revision_id, \
92
"Mismatch between inventory revision" \
93
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
94
assert inv.root is not None
95
inv_text = self.serialise_inventory(inv)
96
inv_sha1 = osutils.sha_string(inv_text)
97
inv_vf = self.control_weaves.get_weave('inventory',
98
self.get_transaction())
99
self._inventory_add_lines(inv_vf, revision_id, parents,
100
osutils.split_lines(inv_text))
103
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
105
for parent in parents:
107
final_parents.append(parent)
109
inv_vf.add_lines(revision_id, final_parents, lines)
112
def add_revision(self, revision_id, rev, inv=None, config=None):
113
"""Add rev to the revision store as revision_id.
115
:param revision_id: the revision id to use.
116
:param rev: The revision object.
117
:param inv: The inventory for the revision. if None, it will be looked
118
up in the inventory storer
119
:param config: If None no digital signature will be created.
120
If supplied its signature_needed method will be used
121
to determine if a signature should be made.
123
revision_id = osutils.safe_revision_id(revision_id)
124
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
126
_mod_revision.check_not_reserved_id(revision_id)
127
if config is not None and config.signature_needed():
129
inv = self.get_inventory(revision_id)
130
plaintext = Testament(rev, inv).as_short_text()
131
self.store_revision_signature(
132
gpg.GPGStrategy(config), plaintext, revision_id)
133
if not revision_id in self.get_inventory_weave():
135
raise errors.WeaveRevisionNotPresent(revision_id,
136
self.get_inventory_weave())
138
# yes, this is not suitable for adding with ghosts.
139
self.add_inventory(revision_id, inv, rev.parent_ids)
140
self._revision_store.add_revision(rev, self.get_transaction())
143
def _all_possible_ids(self):
144
"""Return all the possible revisions that we could find."""
145
return self.get_inventory_weave().versions()
147
def all_revision_ids(self):
148
"""Returns a list of all the revision ids in the repository.
150
This is deprecated because code should generally work on the graph
151
reachable from a particular revision, and ignore any other revisions
152
that might be present. There is no direct replacement method.
154
return self._all_revision_ids()
157
def _all_revision_ids(self):
158
"""Returns a list of all the revision ids in the repository.
160
These are in as much topological order as the underlying store can
161
present: for weaves ghosts may lead to a lack of correctness until
162
the reweave updates the parents list.
164
if self._revision_store.text_store.listable():
165
return self._revision_store.all_revision_ids(self.get_transaction())
166
result = self._all_possible_ids()
167
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
168
# ids. (It should, since _revision_store's API should change to
169
# return utf8 revision_ids)
170
return self._eliminate_revisions_not_present(result)
172
def break_lock(self):
173
"""Break a lock if one is present from another instance.
175
Uses the ui factory to ask for confirmation if the lock may be from
178
self.control_files.break_lock()
181
def _eliminate_revisions_not_present(self, revision_ids):
182
"""Check every revision id in revision_ids to see if we have it.
184
Returns a set of the present revisions.
187
for id in revision_ids:
188
if self.has_revision(id):
193
def create(a_bzrdir):
194
"""Construct the current default format repository in a_bzrdir."""
195
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
197
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
198
"""instantiate a Repository.
200
:param _format: The format of the repository on disk.
201
:param a_bzrdir: The BzrDir of the repository.
203
In the future we will have a single api for all stores for
204
getting file texts, inventories and revisions, then
205
this construct will accept instances of those things.
207
super(Repository, self).__init__()
208
self._format = _format
209
# the following are part of the public API for Repository:
210
self.bzrdir = a_bzrdir
211
self.control_files = control_files
212
self._revision_store = _revision_store
213
self.text_store = text_store
214
# backwards compatibility
215
self.weave_store = text_store
216
# not right yet - should be more semantically clear ?
218
self.control_store = control_store
219
self.control_weaves = control_store
220
# TODO: make sure to construct the right store classes, etc, depending
221
# on whether escaping is required.
222
self._warn_if_deprecated()
225
return '%s(%r)' % (self.__class__.__name__,
226
self.bzrdir.transport.base)
229
return self.control_files.is_locked()
231
def lock_write(self, token=None):
232
"""Lock this repository for writing.
234
:param token: if this is already locked, then lock_write will fail
235
unless the token matches the existing lock.
236
:returns: a token if this instance supports tokens, otherwise None.
237
:raises TokenLockingNotSupported: when a token is given but this
238
instance doesn't support using token locks.
239
:raises MismatchedToken: if the specified token doesn't match the token
240
of the existing lock.
242
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
244
return self.control_files.lock_write(token=token)
247
self.control_files.lock_read()
249
def get_physical_lock_status(self):
250
return self.control_files.get_physical_lock_status()
252
def leave_lock_in_place(self):
253
"""Tell this repository not to release the physical lock when this
256
If lock_write doesn't return a token, then this method is not supported.
258
self.control_files.leave_in_place()
260
def dont_leave_lock_in_place(self):
261
"""Tell this repository to release the physical lock when this
262
object is unlocked, even if it didn't originally acquire it.
264
If lock_write doesn't return a token, then this method is not supported.
266
self.control_files.dont_leave_in_place()
269
def gather_stats(self, revid=None, committers=None):
270
"""Gather statistics from a revision id.
272
:param revid: The revision id to gather statistics from, if None, then
273
no revision specific statistics are gathered.
274
:param committers: Optional parameter controlling whether to grab
275
a count of committers from the revision specific statistics.
276
:return: A dictionary of statistics. Currently this contains:
277
committers: The number of committers if requested.
278
firstrev: A tuple with timestamp, timezone for the penultimate left
279
most ancestor of revid, if revid is not the NULL_REVISION.
280
latestrev: A tuple with timestamp, timezone for revid, if revid is
281
not the NULL_REVISION.
282
revisions: The total revision count in the repository.
283
size: An estimate disk size of the repository in bytes.
286
if revid and committers:
287
result['committers'] = 0
288
if revid and revid != _mod_revision.NULL_REVISION:
290
all_committers = set()
291
revisions = self.get_ancestry(revid)
292
# pop the leading None
294
first_revision = None
296
# ignore the revisions in the middle - just grab first and last
297
revisions = revisions[0], revisions[-1]
298
for revision in self.get_revisions(revisions):
299
if not first_revision:
300
first_revision = revision
302
all_committers.add(revision.committer)
303
last_revision = revision
305
result['committers'] = len(all_committers)
306
result['firstrev'] = (first_revision.timestamp,
307
first_revision.timezone)
308
result['latestrev'] = (last_revision.timestamp,
309
last_revision.timezone)
311
# now gather global repository information
312
if self.bzrdir.root_transport.listable():
313
c, t = self._revision_store.total_size(self.get_transaction())
314
result['revisions'] = c
319
def missing_revision_ids(self, other, revision_id=None):
320
"""Return the revision ids that other has that this does not.
322
These are returned in topological order.
324
revision_id: only return revision ids included by revision_id.
326
revision_id = osutils.safe_revision_id(revision_id)
327
return InterRepository.get(other, self).missing_revision_ids(revision_id)
331
"""Open the repository rooted at base.
333
For instance, if the repository is at URL/.bzr/repository,
334
Repository.open(URL) -> a Repository instance.
336
control = bzrdir.BzrDir.open(base)
337
return control.open_repository()
339
def copy_content_into(self, destination, revision_id=None, basis=None):
340
"""Make a complete copy of the content in self into destination.
342
This is a destructive operation! Do not use it on existing
345
revision_id = osutils.safe_revision_id(revision_id)
346
return InterRepository.get(self, destination).copy_content(revision_id, basis)
348
def fetch(self, source, revision_id=None, pb=None):
349
"""Fetch the content required to construct revision_id from source.
351
If revision_id is None all content is copied.
353
revision_id = osutils.safe_revision_id(revision_id)
354
return InterRepository.get(source, self).fetch(revision_id=revision_id,
357
def get_commit_builder(self, branch, parents, config, timestamp=None,
358
timezone=None, committer=None, revprops=None,
360
"""Obtain a CommitBuilder for this repository.
362
:param branch: Branch to commit to.
363
:param parents: Revision ids of the parents of the new revision.
364
:param config: Configuration to use.
365
:param timestamp: Optional timestamp recorded for commit.
366
:param timezone: Optional timezone for timestamp.
367
:param committer: Optional committer to set for commit.
368
:param revprops: Optional dictionary of revision properties.
369
:param revision_id: Optional revision id.
371
revision_id = osutils.safe_revision_id(revision_id)
372
return _CommitBuilder(self, parents, config, timestamp, timezone,
373
committer, revprops, revision_id)
376
self.control_files.unlock()
379
def clone(self, a_bzrdir, revision_id=None, basis=None):
380
"""Clone this repository into a_bzrdir using the current format.
382
Currently no check is made that the format of this repository and
383
the bzrdir format are compatible. FIXME RBC 20060201.
385
:return: The newly created destination repository.
387
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
388
# use target default format.
389
dest_repo = a_bzrdir.create_repository()
391
# Most control formats need the repository to be specifically
392
# created, but on some old all-in-one formats it's not needed
394
dest_repo = self._format.initialize(a_bzrdir, shared=self.is_shared())
395
except errors.UninitializableFormat:
396
dest_repo = a_bzrdir.open_repository()
397
self.copy_content_into(dest_repo, revision_id, basis)
401
def has_revision(self, revision_id):
402
"""True if this repository has a copy of the revision."""
403
revision_id = osutils.safe_revision_id(revision_id)
404
return self._revision_store.has_revision_id(revision_id,
405
self.get_transaction())
408
def get_revision_reconcile(self, revision_id):
409
"""'reconcile' helper routine that allows access to a revision always.
411
This variant of get_revision does not cross check the weave graph
412
against the revision one as get_revision does: but it should only
413
be used by reconcile, or reconcile-alike commands that are correcting
414
or testing the revision graph.
416
if not revision_id or not isinstance(revision_id, basestring):
417
raise errors.InvalidRevisionId(revision_id=revision_id,
419
return self.get_revisions([revision_id])[0]
422
def get_revisions(self, revision_ids):
423
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
424
revs = self._revision_store.get_revisions(revision_ids,
425
self.get_transaction())
427
assert not isinstance(rev.revision_id, unicode)
428
for parent_id in rev.parent_ids:
429
assert not isinstance(parent_id, unicode)
433
def get_revision_xml(self, revision_id):
434
# TODO: jam 20070210 This shouldn't be necessary since get_revision
435
# would have already do it.
436
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
437
revision_id = osutils.safe_revision_id(revision_id)
438
rev = self.get_revision(revision_id)
440
# the current serializer..
441
self._revision_store._serializer.write_revision(rev, rev_tmp)
443
return rev_tmp.getvalue()
446
def get_revision(self, revision_id):
447
"""Return the Revision object for a named revision"""
448
# TODO: jam 20070210 get_revision_reconcile should do this for us
449
revision_id = osutils.safe_revision_id(revision_id)
450
r = self.get_revision_reconcile(revision_id)
451
# weave corruption can lead to absent revision markers that should be
453
# the following test is reasonably cheap (it needs a single weave read)
454
# and the weave is cached in read transactions. In write transactions
455
# it is not cached but typically we only read a small number of
456
# revisions. For knits when they are introduced we will probably want
457
# to ensure that caching write transactions are in use.
458
inv = self.get_inventory_weave()
459
self._check_revision_parents(r, inv)
463
def get_deltas_for_revisions(self, revisions):
464
"""Produce a generator of revision deltas.
466
Note that the input is a sequence of REVISIONS, not revision_ids.
467
Trees will be held in memory until the generator exits.
468
Each delta is relative to the revision's lefthand predecessor.
470
required_trees = set()
471
for revision in revisions:
472
required_trees.add(revision.revision_id)
473
required_trees.update(revision.parent_ids[:1])
474
trees = dict((t.get_revision_id(), t) for
475
t in self.revision_trees(required_trees))
476
for revision in revisions:
477
if not revision.parent_ids:
478
old_tree = self.revision_tree(None)
480
old_tree = trees[revision.parent_ids[0]]
481
yield trees[revision.revision_id].changes_from(old_tree)
484
def get_revision_delta(self, revision_id):
485
"""Return the delta for one revision.
487
The delta is relative to the left-hand predecessor of the
490
r = self.get_revision(revision_id)
491
return list(self.get_deltas_for_revisions([r]))[0]
493
def _check_revision_parents(self, revision, inventory):
494
"""Private to Repository and Fetch.
496
This checks the parentage of revision in an inventory weave for
497
consistency and is only applicable to inventory-weave-for-ancestry
498
using repository formats & fetchers.
500
weave_parents = inventory.get_parents(revision.revision_id)
501
weave_names = inventory.versions()
502
for parent_id in revision.parent_ids:
503
if parent_id in weave_names:
504
# this parent must not be a ghost.
505
if not parent_id in weave_parents:
507
raise errors.CorruptRepository(self)
510
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
511
revision_id = osutils.safe_revision_id(revision_id)
512
signature = gpg_strategy.sign(plaintext)
513
self._revision_store.add_revision_signature_text(revision_id,
515
self.get_transaction())
517
def fileids_altered_by_revision_ids(self, revision_ids):
518
"""Find the file ids and versions affected by revisions.
520
:param revisions: an iterable containing revision ids.
521
:return: a dictionary mapping altered file-ids to an iterable of
522
revision_ids. Each altered file-ids has the exact revision_ids that
523
altered it listed explicitly.
525
assert self._serializer.support_altered_by_hack, \
526
("fileids_altered_by_revision_ids only supported for branches "
527
"which store inventory as unnested xml, not on %r" % self)
528
selected_revision_ids = set(osutils.safe_revision_id(r)
529
for r in revision_ids)
530
w = self.get_inventory_weave()
533
# this code needs to read every new line in every inventory for the
534
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
535
# not present in one of those inventories is unnecessary but not
536
# harmful because we are filtering by the revision id marker in the
537
# inventory lines : we only select file ids altered in one of those
538
# revisions. We don't need to see all lines in the inventory because
539
# only those added in an inventory in rev X can contain a revision=X
541
unescape_revid_cache = {}
542
unescape_fileid_cache = {}
544
# jam 20061218 In a big fetch, this handles hundreds of thousands
545
# of lines, so it has had a lot of inlining and optimizing done.
546
# Sorry that it is a little bit messy.
547
# Move several functions to be local variables, since this is a long
549
search = self._file_ids_altered_regex.search
550
unescape = _unescape_xml
551
setdefault = result.setdefault
552
pb = ui.ui_factory.nested_progress_bar()
554
for line in w.iter_lines_added_or_present_in_versions(
555
selected_revision_ids, pb=pb):
559
# One call to match.group() returning multiple items is quite a
560
# bit faster than 2 calls to match.group() each returning 1
561
file_id, revision_id = match.group('file_id', 'revision_id')
563
# Inlining the cache lookups helps a lot when you make 170,000
564
# lines and 350k ids, versus 8.4 unique ids.
565
# Using a cache helps in 2 ways:
566
# 1) Avoids unnecessary decoding calls
567
# 2) Re-uses cached strings, which helps in future set and
569
# (2) is enough that removing encoding entirely along with
570
# the cache (so we are using plain strings) results in no
571
# performance improvement.
573
revision_id = unescape_revid_cache[revision_id]
575
unescaped = unescape(revision_id)
576
unescape_revid_cache[revision_id] = unescaped
577
revision_id = unescaped
579
if revision_id in selected_revision_ids:
581
file_id = unescape_fileid_cache[file_id]
583
unescaped = unescape(file_id)
584
unescape_fileid_cache[file_id] = unescaped
586
setdefault(file_id, set()).add(revision_id)
592
def get_inventory_weave(self):
593
return self.control_weaves.get_weave('inventory',
594
self.get_transaction())
597
def get_inventory(self, revision_id):
598
"""Get Inventory object by hash."""
599
# TODO: jam 20070210 Technically we don't need to sanitize, since all
600
# called functions must sanitize.
601
revision_id = osutils.safe_revision_id(revision_id)
602
return self.deserialise_inventory(
603
revision_id, self.get_inventory_xml(revision_id))
605
def deserialise_inventory(self, revision_id, xml):
606
"""Transform the xml into an inventory object.
608
:param revision_id: The expected revision id of the inventory.
609
:param xml: A serialised inventory.
611
revision_id = osutils.safe_revision_id(revision_id)
612
result = self._serializer.read_inventory_from_string(xml)
613
result.root.revision = revision_id
616
def serialise_inventory(self, inv):
617
return self._serializer.write_inventory_to_string(inv)
620
def get_inventory_xml(self, revision_id):
621
"""Get inventory XML as a file object."""
622
revision_id = osutils.safe_revision_id(revision_id)
624
assert isinstance(revision_id, str), type(revision_id)
625
iw = self.get_inventory_weave()
626
return iw.get_text(revision_id)
628
raise errors.HistoryMissing(self, 'inventory', revision_id)
631
def get_inventory_sha1(self, revision_id):
632
"""Return the sha1 hash of the inventory entry
634
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
635
revision_id = osutils.safe_revision_id(revision_id)
636
return self.get_revision(revision_id).inventory_sha1
639
def get_revision_graph(self, revision_id=None):
640
"""Return a dictionary containing the revision graph.
642
:param revision_id: The revision_id to get a graph from. If None, then
643
the entire revision graph is returned. This is a deprecated mode of
644
operation and will be removed in the future.
645
:return: a dictionary of revision_id->revision_parents_list.
647
# special case NULL_REVISION
648
if revision_id == _mod_revision.NULL_REVISION:
650
revision_id = osutils.safe_revision_id(revision_id)
651
a_weave = self.get_inventory_weave()
652
all_revisions = self._eliminate_revisions_not_present(
654
entire_graph = dict([(node, a_weave.get_parents(node)) for
655
node in all_revisions])
656
if revision_id is None:
658
elif revision_id not in entire_graph:
659
raise errors.NoSuchRevision(self, revision_id)
661
# add what can be reached from revision_id
663
pending = set([revision_id])
664
while len(pending) > 0:
666
result[node] = entire_graph[node]
667
for revision_id in result[node]:
668
if revision_id not in result:
669
pending.add(revision_id)
673
def get_revision_graph_with_ghosts(self, revision_ids=None):
674
"""Return a graph of the revisions with ghosts marked as applicable.
676
:param revision_ids: an iterable of revisions to graph or None for all.
677
:return: a Graph object with the graph reachable from revision_ids.
679
result = graph.Graph()
681
pending = set(self.all_revision_ids())
684
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
685
# special case NULL_REVISION
686
if _mod_revision.NULL_REVISION in pending:
687
pending.remove(_mod_revision.NULL_REVISION)
688
required = set(pending)
691
revision_id = pending.pop()
693
rev = self.get_revision(revision_id)
694
except errors.NoSuchRevision:
695
if revision_id in required:
698
result.add_ghost(revision_id)
700
for parent_id in rev.parent_ids:
701
# is this queued or done ?
702
if (parent_id not in pending and
703
parent_id not in done):
705
pending.add(parent_id)
706
result.add_node(revision_id, rev.parent_ids)
707
done.add(revision_id)
710
def _get_history_vf(self):
711
"""Get a versionedfile whose history graph reflects all revisions.
713
For weave repositories, this is the inventory weave.
715
return self.get_inventory_weave()
717
def iter_reverse_revision_history(self, revision_id):
718
"""Iterate backwards through revision ids in the lefthand history
720
:param revision_id: The revision id to start with. All its lefthand
721
ancestors will be traversed.
723
revision_id = osutils.safe_revision_id(revision_id)
724
if revision_id in (None, _mod_revision.NULL_REVISION):
726
next_id = revision_id
727
versionedfile = self._get_history_vf()
730
parents = versionedfile.get_parents(next_id)
731
if len(parents) == 0:
737
def get_revision_inventory(self, revision_id):
738
"""Return inventory of a past revision."""
739
# TODO: Unify this with get_inventory()
740
# bzr 0.0.6 and later imposes the constraint that the inventory_id
741
# must be the same as its revision, so this is trivial.
742
if revision_id is None:
743
# This does not make sense: if there is no revision,
744
# then it is the current tree inventory surely ?!
745
# and thus get_root_id() is something that looks at the last
746
# commit on the branch, and the get_root_id is an inventory check.
747
raise NotImplementedError
748
# return Inventory(self.get_root_id())
750
return self.get_inventory(revision_id)
754
"""Return True if this repository is flagged as a shared repository."""
755
raise NotImplementedError(self.is_shared)
758
def reconcile(self, other=None, thorough=False):
759
"""Reconcile this repository."""
760
from bzrlib.reconcile import RepoReconciler
761
reconciler = RepoReconciler(self, thorough=thorough)
762
reconciler.reconcile()
766
def revision_tree(self, revision_id):
767
"""Return Tree for a revision on this branch.
769
`revision_id` may be None for the empty tree revision.
771
# TODO: refactor this to use an existing revision object
772
# so we don't need to read it in twice.
773
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
774
return RevisionTree(self, Inventory(root_id=None),
775
_mod_revision.NULL_REVISION)
777
revision_id = osutils.safe_revision_id(revision_id)
778
inv = self.get_revision_inventory(revision_id)
779
return RevisionTree(self, inv, revision_id)
782
def revision_trees(self, revision_ids):
783
"""Return Tree for a revision on this branch.
785
`revision_id` may not be None or 'null:'"""
786
assert None not in revision_ids
787
assert _mod_revision.NULL_REVISION not in revision_ids
788
texts = self.get_inventory_weave().get_texts(revision_ids)
789
for text, revision_id in zip(texts, revision_ids):
790
inv = self.deserialise_inventory(revision_id, text)
791
yield RevisionTree(self, inv, revision_id)
794
def get_ancestry(self, revision_id):
795
"""Return a list of revision-ids integrated by a revision.
797
The first element of the list is always None, indicating the origin
798
revision. This might change when we have history horizons, or
799
perhaps we should have a new API.
801
This is topologically sorted.
803
if revision_id is None:
805
revision_id = osutils.safe_revision_id(revision_id)
806
if not self.has_revision(revision_id):
807
raise errors.NoSuchRevision(self, revision_id)
808
w = self.get_inventory_weave()
809
candidates = w.get_ancestry(revision_id)
810
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
813
def print_file(self, file, revision_id):
814
"""Print `file` to stdout.
816
FIXME RBC 20060125 as John Meinel points out this is a bad api
817
- it writes to stdout, it assumes that that is valid etc. Fix
818
by creating a new more flexible convenience function.
820
revision_id = osutils.safe_revision_id(revision_id)
821
tree = self.revision_tree(revision_id)
822
# use inventory as it was in that revision
823
file_id = tree.inventory.path2id(file)
825
# TODO: jam 20060427 Write a test for this code path
826
# it had a bug in it, and was raising the wrong
828
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
829
tree.print_file(file_id)
831
def get_transaction(self):
832
return self.control_files.get_transaction()
834
def revision_parents(self, revision_id):
835
revision_id = osutils.safe_revision_id(revision_id)
836
return self.get_inventory_weave().parent_names(revision_id)
839
def set_make_working_trees(self, new_value):
840
"""Set the policy flag for making working trees when creating branches.
842
This only applies to branches that use this repository.
844
The default is 'True'.
845
:param new_value: True to restore the default, False to disable making
848
raise NotImplementedError(self.set_make_working_trees)
850
def make_working_trees(self):
851
"""Returns the policy for making working trees on new branches."""
852
raise NotImplementedError(self.make_working_trees)
855
def sign_revision(self, revision_id, gpg_strategy):
856
revision_id = osutils.safe_revision_id(revision_id)
857
plaintext = Testament.from_revision(self, revision_id).as_short_text()
858
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
861
def has_signature_for_revision_id(self, revision_id):
862
"""Query for a revision signature for revision_id in the repository."""
863
revision_id = osutils.safe_revision_id(revision_id)
864
return self._revision_store.has_signature(revision_id,
865
self.get_transaction())
868
def get_signature_text(self, revision_id):
869
"""Return the text for a signature."""
870
revision_id = osutils.safe_revision_id(revision_id)
871
return self._revision_store.get_signature_text(revision_id,
872
self.get_transaction())
875
def check(self, revision_ids):
876
"""Check consistency of all history of given revision_ids.
878
Different repository implementations should override _check().
880
:param revision_ids: A non-empty list of revision_ids whose ancestry
881
will be checked. Typically the last revision_id of a branch.
884
raise ValueError("revision_ids must be non-empty in %s.check"
886
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
887
return self._check(revision_ids)
889
def _check(self, revision_ids):
890
result = check.Check(self)
894
def _warn_if_deprecated(self):
895
global _deprecation_warning_done
896
if _deprecation_warning_done:
898
_deprecation_warning_done = True
899
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
900
% (self._format, self.bzrdir.transport.base))
902
def supports_rich_root(self):
903
return self._format.rich_root_data
905
def _check_ascii_revisionid(self, revision_id, method):
906
"""Private helper for ascii-only repositories."""
907
# weave repositories refuse to store revisionids that are non-ascii.
908
if revision_id is not None:
909
# weaves require ascii revision ids.
910
if isinstance(revision_id, unicode):
912
revision_id.encode('ascii')
913
except UnicodeEncodeError:
914
raise errors.NonAsciiRevisionId(method, self)
917
revision_id.decode('ascii')
918
except UnicodeDecodeError:
919
raise errors.NonAsciiRevisionId(method, self)
923
# remove these delegates a while after bzr 0.15
924
def __make_delegated(name, from_module):
925
def _deprecated_repository_forwarder():
926
symbol_versioning.warn('%s moved to %s in bzr 0.15'
927
% (name, from_module),
930
m = __import__(from_module, globals(), locals(), [name])
932
return getattr(m, name)
933
except AttributeError:
934
raise AttributeError('module %s has no name %s'
936
globals()[name] = _deprecated_repository_forwarder
939
'AllInOneRepository',
940
'WeaveMetaDirRepository',
941
'PreSplitOutRepositoryFormat',
947
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
952
'RepositoryFormatKnit',
953
'RepositoryFormatKnit1',
954
'RepositoryFormatKnit2',
956
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
959
def install_revision(repository, rev, revision_tree):
960
"""Install all revision data into a repository."""
963
for p_id in rev.parent_ids:
964
if repository.has_revision(p_id):
965
present_parents.append(p_id)
966
parent_trees[p_id] = repository.revision_tree(p_id)
968
parent_trees[p_id] = repository.revision_tree(None)
970
inv = revision_tree.inventory
971
entries = inv.iter_entries()
972
# backwards compatability hack: skip the root id.
973
if not repository.supports_rich_root():
974
path, root = entries.next()
975
if root.revision != rev.revision_id:
976
raise errors.IncompatibleRevision(repr(repository))
977
# Add the texts that are not already present
978
for path, ie in entries:
979
w = repository.weave_store.get_weave_or_empty(ie.file_id,
980
repository.get_transaction())
981
if ie.revision not in w:
983
# FIXME: TODO: The following loop *may* be overlapping/duplicate
984
# with InventoryEntry.find_previous_heads(). if it is, then there
985
# is a latent bug here where the parents may have ancestors of each
987
for revision, tree in parent_trees.iteritems():
988
if ie.file_id not in tree:
990
parent_id = tree.inventory[ie.file_id].revision
991
if parent_id in text_parents:
993
text_parents.append(parent_id)
995
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
996
repository.get_transaction())
997
lines = revision_tree.get_file(ie.file_id).readlines()
998
vfile.add_lines(rev.revision_id, text_parents, lines)
1000
# install the inventory
1001
repository.add_inventory(rev.revision_id, inv, present_parents)
1002
except errors.RevisionAlreadyPresent:
1004
repository.add_revision(rev.revision_id, rev, inv)
1007
class MetaDirRepository(Repository):
1008
"""Repositories in the new meta-dir layout."""
1010
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1011
super(MetaDirRepository, self).__init__(_format,
1017
dir_mode = self.control_files._dir_mode
1018
file_mode = self.control_files._file_mode
1021
def is_shared(self):
1022
"""Return True if this repository is flagged as a shared repository."""
1023
return self.control_files._transport.has('shared-storage')
1026
def set_make_working_trees(self, new_value):
1027
"""Set the policy flag for making working trees when creating branches.
1029
This only applies to branches that use this repository.
1031
The default is 'True'.
1032
:param new_value: True to restore the default, False to disable making
1037
self.control_files._transport.delete('no-working-trees')
1038
except errors.NoSuchFile:
1041
self.control_files.put_utf8('no-working-trees', '')
1043
def make_working_trees(self):
1044
"""Returns the policy for making working trees on new branches."""
1045
return not self.control_files._transport.has('no-working-trees')
1048
class RepositoryFormatRegistry(registry.Registry):
1049
"""Registry of RepositoryFormats.
1052
def get(self, format_string):
1053
r = registry.Registry.get(self, format_string)
1059
format_registry = RepositoryFormatRegistry()
1060
"""Registry of formats, indexed by their identifying format string.
1062
This can contain either format instances themselves, or classes/factories that
1063
can be called to obtain one.
1067
class RepositoryFormat(object):
1068
"""A repository format.
1070
Formats provide three things:
1071
* An initialization routine to construct repository data on disk.
1072
* a format string which is used when the BzrDir supports versioned
1074
* an open routine which returns a Repository instance.
1076
Formats are placed in an dict by their format string for reference
1077
during opening. These should be subclasses of RepositoryFormat
1080
Once a format is deprecated, just deprecate the initialize and open
1081
methods on the format class. Do not deprecate the object, as the
1082
object will be created every system load.
1084
Common instance attributes:
1085
_matchingbzrdir - the bzrdir format that the repository format was
1086
originally written to work with. This can be used if manually
1087
constructing a bzrdir and repository, or more commonly for test suite
1092
return "<%s>" % self.__class__.__name__
1094
def __eq__(self, other):
1095
# format objects are generally stateless
1096
return isinstance(other, self.__class__)
1099
def find_format(klass, a_bzrdir):
1100
"""Return the format for the repository object in a_bzrdir.
1102
This is used by bzr native formats that have a "format" file in
1103
the repository. Other methods may be used by different types of
1107
transport = a_bzrdir.get_repository_transport(None)
1108
format_string = transport.get("format").read()
1109
return format_registry.get(format_string)
1110
except errors.NoSuchFile:
1111
raise errors.NoRepositoryPresent(a_bzrdir)
1113
raise errors.UnknownFormatError(format=format_string)
1116
def register_format(klass, format):
1117
format_registry.register(format.get_format_string(), format)
1120
def unregister_format(klass, format):
1121
format_registry.remove(format.get_format_string())
1124
def get_default_format(klass):
1125
"""Return the current default format."""
1126
from bzrlib import bzrdir
1127
return bzrdir.format_registry.make_bzrdir('default').repository_format
1129
def _get_control_store(self, repo_transport, control_files):
1130
"""Return the control store for this repository."""
1131
raise NotImplementedError(self._get_control_store)
1133
def get_format_string(self):
1134
"""Return the ASCII format string that identifies this format.
1136
Note that in pre format ?? repositories the format string is
1137
not permitted nor written to disk.
1139
raise NotImplementedError(self.get_format_string)
1141
def get_format_description(self):
1142
"""Return the short description for this format."""
1143
raise NotImplementedError(self.get_format_description)
1145
def _get_revision_store(self, repo_transport, control_files):
1146
"""Return the revision store object for this a_bzrdir."""
1147
raise NotImplementedError(self._get_revision_store)
1149
def _get_text_rev_store(self,
1156
"""Common logic for getting a revision store for a repository.
1158
see self._get_revision_store for the subclass-overridable method to
1159
get the store for a repository.
1161
from bzrlib.store.revision.text import TextRevisionStore
1162
dir_mode = control_files._dir_mode
1163
file_mode = control_files._file_mode
1164
text_store =TextStore(transport.clone(name),
1166
compressed=compressed,
1168
file_mode=file_mode)
1169
_revision_store = TextRevisionStore(text_store, serializer)
1170
return _revision_store
1172
# TODO: this shouldn't be in the base class, it's specific to things that
1173
# use weaves or knits -- mbp 20070207
1174
def _get_versioned_file_store(self,
1179
versionedfile_class=None,
1180
versionedfile_kwargs={},
1182
if versionedfile_class is None:
1183
versionedfile_class = self._versionedfile_class
1184
weave_transport = control_files._transport.clone(name)
1185
dir_mode = control_files._dir_mode
1186
file_mode = control_files._file_mode
1187
return VersionedFileStore(weave_transport, prefixed=prefixed,
1189
file_mode=file_mode,
1190
versionedfile_class=versionedfile_class,
1191
versionedfile_kwargs=versionedfile_kwargs,
1194
def initialize(self, a_bzrdir, shared=False):
1195
"""Initialize a repository of this format in a_bzrdir.
1197
:param a_bzrdir: The bzrdir to put the new repository in it.
1198
:param shared: The repository should be initialized as a sharable one.
1199
:returns: The new repository object.
1201
This may raise UninitializableFormat if shared repository are not
1202
compatible the a_bzrdir.
1204
raise NotImplementedError(self.initialize)
1206
def is_supported(self):
1207
"""Is this format supported?
1209
Supported formats must be initializable and openable.
1210
Unsupported formats may not support initialization or committing or
1211
some other features depending on the reason for not being supported.
1215
def check_conversion_target(self, target_format):
1216
raise NotImplementedError(self.check_conversion_target)
1218
def open(self, a_bzrdir, _found=False):
1219
"""Return an instance of this format for the bzrdir a_bzrdir.
1221
_found is a private parameter, do not use it.
1223
raise NotImplementedError(self.open)
1226
class MetaDirRepositoryFormat(RepositoryFormat):
1227
"""Common base class for the new repositories using the metadir layout."""
1229
rich_root_data = False
1230
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1233
super(MetaDirRepositoryFormat, self).__init__()
1235
def _create_control_files(self, a_bzrdir):
1236
"""Create the required files and the initial control_files object."""
1237
# FIXME: RBC 20060125 don't peek under the covers
1238
# NB: no need to escape relative paths that are url safe.
1239
repository_transport = a_bzrdir.get_repository_transport(self)
1240
control_files = lockable_files.LockableFiles(repository_transport,
1241
'lock', lockdir.LockDir)
1242
control_files.create_lock()
1243
return control_files
1245
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1246
"""Upload the initial blank content."""
1247
control_files = self._create_control_files(a_bzrdir)
1248
control_files.lock_write()
1250
control_files._transport.mkdir_multi(dirs,
1251
mode=control_files._dir_mode)
1252
for file, content in files:
1253
control_files.put(file, content)
1254
for file, content in utf8_files:
1255
control_files.put_utf8(file, content)
1257
control_files.put_utf8('shared-storage', '')
1259
control_files.unlock()
1262
# formats which have no format string are not discoverable
1263
# and not independently creatable, so are not registered. They're
1264
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1265
# needed, it's constructed directly by the BzrDir. Non-native formats where
1266
# the repository is not separately opened are similar.
1268
format_registry.register_lazy(
1269
'Bazaar-NG Repository format 7',
1270
'bzrlib.repofmt.weaverepo',
1273
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1274
# default control directory format
1276
format_registry.register_lazy(
1277
'Bazaar-NG Knit Repository Format 1',
1278
'bzrlib.repofmt.knitrepo',
1279
'RepositoryFormatKnit1',
1281
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1283
format_registry.register_lazy(
1284
'Bazaar Knit Repository Format 2\n',
1285
'bzrlib.repofmt.knitrepo',
1286
'RepositoryFormatKnit2',
1290
class InterRepository(InterObject):
1291
"""This class represents operations taking place between two repositories.
1293
Its instances have methods like copy_content and fetch, and contain
1294
references to the source and target repositories these operations can be
1297
Often we will provide convenience methods on 'repository' which carry out
1298
operations with another repository - they will always forward to
1299
InterRepository.get(other).method_name(parameters).
1303
"""The available optimised InterRepository types."""
1305
def copy_content(self, revision_id=None, basis=None):
1306
raise NotImplementedError(self.copy_content)
1308
def fetch(self, revision_id=None, pb=None):
1309
"""Fetch the content required to construct revision_id.
1311
The content is copied from self.source to self.target.
1313
:param revision_id: if None all content is copied, if NULL_REVISION no
1315
:param pb: optional progress bar to use for progress reports. If not
1316
provided a default one will be created.
1318
Returns the copied revision count and the failed revisions in a tuple:
1321
raise NotImplementedError(self.fetch)
1324
def missing_revision_ids(self, revision_id=None):
1325
"""Return the revision ids that source has that target does not.
1327
These are returned in topological order.
1329
:param revision_id: only return revision ids included by this
1332
# generic, possibly worst case, slow code path.
1333
target_ids = set(self.target.all_revision_ids())
1334
if revision_id is not None:
1335
# TODO: jam 20070210 InterRepository is internal enough that it
1336
# should assume revision_ids are already utf-8
1337
revision_id = osutils.safe_revision_id(revision_id)
1338
source_ids = self.source.get_ancestry(revision_id)
1339
assert source_ids[0] is None
1342
source_ids = self.source.all_revision_ids()
1343
result_set = set(source_ids).difference(target_ids)
1344
# this may look like a no-op: its not. It preserves the ordering
1345
# other_ids had while only returning the members from other_ids
1346
# that we've decided we need.
1347
return [rev_id for rev_id in source_ids if rev_id in result_set]
1350
class InterSameDataRepository(InterRepository):
1351
"""Code for converting between repositories that represent the same data.
1353
Data format and model must match for this to work.
1357
def _get_repo_format_to_test(self):
1358
"""Repository format for testing with."""
1359
return RepositoryFormat.get_default_format()
1362
def is_compatible(source, target):
1363
if source.supports_rich_root() == target.supports_rich_root():
1369
def copy_content(self, revision_id=None, basis=None):
1370
"""Make a complete copy of the content in self into destination.
1372
This is a destructive operation! Do not use it on existing
1375
:param revision_id: Only copy the content needed to construct
1376
revision_id and its parents.
1377
:param basis: Copy the needed data preferentially from basis.
1380
self.target.set_make_working_trees(self.source.make_working_trees())
1381
except NotImplementedError:
1383
# TODO: jam 20070210 This is fairly internal, so we should probably
1384
# just assert that revision_id is not unicode.
1385
revision_id = osutils.safe_revision_id(revision_id)
1386
# grab the basis available data
1387
if basis is not None:
1388
self.target.fetch(basis, revision_id=revision_id)
1389
# but don't bother fetching if we have the needed data now.
1390
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1391
self.target.has_revision(revision_id)):
1393
self.target.fetch(self.source, revision_id=revision_id)
1396
def fetch(self, revision_id=None, pb=None):
1397
"""See InterRepository.fetch()."""
1398
from bzrlib.fetch import GenericRepoFetcher
1399
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1400
self.source, self.source._format, self.target,
1401
self.target._format)
1402
# TODO: jam 20070210 This should be an assert, not a translate
1403
revision_id = osutils.safe_revision_id(revision_id)
1404
f = GenericRepoFetcher(to_repository=self.target,
1405
from_repository=self.source,
1406
last_revision=revision_id,
1408
return f.count_copied, f.failed_revisions
1411
class InterWeaveRepo(InterSameDataRepository):
1412
"""Optimised code paths between Weave based repositories."""
1415
def _get_repo_format_to_test(self):
1416
from bzrlib.repofmt import weaverepo
1417
return weaverepo.RepositoryFormat7()
1420
def is_compatible(source, target):
1421
"""Be compatible with known Weave formats.
1423
We don't test for the stores being of specific types because that
1424
could lead to confusing results, and there is no need to be
1427
from bzrlib.repofmt.weaverepo import (
1433
return (isinstance(source._format, (RepositoryFormat5,
1435
RepositoryFormat7)) and
1436
isinstance(target._format, (RepositoryFormat5,
1438
RepositoryFormat7)))
1439
except AttributeError:
1443
def copy_content(self, revision_id=None, basis=None):
1444
"""See InterRepository.copy_content()."""
1445
# weave specific optimised path:
1446
# TODO: jam 20070210 Internal, should be an assert, not translate
1447
revision_id = osutils.safe_revision_id(revision_id)
1448
if basis is not None:
1449
# copy the basis in, then fetch remaining data.
1450
basis.copy_content_into(self.target, revision_id)
1451
# the basis copy_content_into could miss-set this.
1453
self.target.set_make_working_trees(self.source.make_working_trees())
1454
except NotImplementedError:
1456
self.target.fetch(self.source, revision_id=revision_id)
1459
self.target.set_make_working_trees(self.source.make_working_trees())
1460
except NotImplementedError:
1462
# FIXME do not peek!
1463
if self.source.control_files._transport.listable():
1464
pb = ui.ui_factory.nested_progress_bar()
1466
self.target.weave_store.copy_all_ids(
1467
self.source.weave_store,
1469
from_transaction=self.source.get_transaction(),
1470
to_transaction=self.target.get_transaction())
1471
pb.update('copying inventory', 0, 1)
1472
self.target.control_weaves.copy_multi(
1473
self.source.control_weaves, ['inventory'],
1474
from_transaction=self.source.get_transaction(),
1475
to_transaction=self.target.get_transaction())
1476
self.target._revision_store.text_store.copy_all_ids(
1477
self.source._revision_store.text_store,
1482
self.target.fetch(self.source, revision_id=revision_id)
1485
def fetch(self, revision_id=None, pb=None):
1486
"""See InterRepository.fetch()."""
1487
from bzrlib.fetch import GenericRepoFetcher
1488
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1489
self.source, self.source._format, self.target, self.target._format)
1490
# TODO: jam 20070210 This should be an assert, not a translate
1491
revision_id = osutils.safe_revision_id(revision_id)
1492
f = GenericRepoFetcher(to_repository=self.target,
1493
from_repository=self.source,
1494
last_revision=revision_id,
1496
return f.count_copied, f.failed_revisions
1499
def missing_revision_ids(self, revision_id=None):
1500
"""See InterRepository.missing_revision_ids()."""
1501
# we want all revisions to satisfy revision_id in source.
1502
# but we don't want to stat every file here and there.
1503
# we want then, all revisions other needs to satisfy revision_id
1504
# checked, but not those that we have locally.
1505
# so the first thing is to get a subset of the revisions to
1506
# satisfy revision_id in source, and then eliminate those that
1507
# we do already have.
1508
# this is slow on high latency connection to self, but as as this
1509
# disk format scales terribly for push anyway due to rewriting
1510
# inventory.weave, this is considered acceptable.
1512
if revision_id is not None:
1513
source_ids = self.source.get_ancestry(revision_id)
1514
assert source_ids[0] is None
1517
source_ids = self.source._all_possible_ids()
1518
source_ids_set = set(source_ids)
1519
# source_ids is the worst possible case we may need to pull.
1520
# now we want to filter source_ids against what we actually
1521
# have in target, but don't try to check for existence where we know
1522
# we do not have a revision as that would be pointless.
1523
target_ids = set(self.target._all_possible_ids())
1524
possibly_present_revisions = target_ids.intersection(source_ids_set)
1525
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1526
required_revisions = source_ids_set.difference(actually_present_revisions)
1527
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1528
if revision_id is not None:
1529
# we used get_ancestry to determine source_ids then we are assured all
1530
# revisions referenced are present as they are installed in topological order.
1531
# and the tip revision was validated by get_ancestry.
1532
return required_topo_revisions
1534
# if we just grabbed the possibly available ids, then
1535
# we only have an estimate of whats available and need to validate
1536
# that against the revision records.
1537
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1540
class InterKnitRepo(InterSameDataRepository):
1541
"""Optimised code paths between Knit based repositories."""
1544
def _get_repo_format_to_test(self):
1545
from bzrlib.repofmt import knitrepo
1546
return knitrepo.RepositoryFormatKnit1()
1549
def is_compatible(source, target):
1550
"""Be compatible with known Knit formats.
1552
We don't test for the stores being of specific types because that
1553
could lead to confusing results, and there is no need to be
1556
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1558
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1559
isinstance(target._format, (RepositoryFormatKnit1)))
1560
except AttributeError:
1564
def fetch(self, revision_id=None, pb=None):
1565
"""See InterRepository.fetch()."""
1566
from bzrlib.fetch import KnitRepoFetcher
1567
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1568
self.source, self.source._format, self.target, self.target._format)
1569
# TODO: jam 20070210 This should be an assert, not a translate
1570
revision_id = osutils.safe_revision_id(revision_id)
1571
f = KnitRepoFetcher(to_repository=self.target,
1572
from_repository=self.source,
1573
last_revision=revision_id,
1575
return f.count_copied, f.failed_revisions
1578
def missing_revision_ids(self, revision_id=None):
1579
"""See InterRepository.missing_revision_ids()."""
1580
if revision_id is not None:
1581
source_ids = self.source.get_ancestry(revision_id)
1582
assert source_ids[0] is None
1585
source_ids = self.source._all_possible_ids()
1586
source_ids_set = set(source_ids)
1587
# source_ids is the worst possible case we may need to pull.
1588
# now we want to filter source_ids against what we actually
1589
# have in target, but don't try to check for existence where we know
1590
# we do not have a revision as that would be pointless.
1591
target_ids = set(self.target._all_possible_ids())
1592
possibly_present_revisions = target_ids.intersection(source_ids_set)
1593
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1594
required_revisions = source_ids_set.difference(actually_present_revisions)
1595
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1596
if revision_id is not None:
1597
# we used get_ancestry to determine source_ids then we are assured all
1598
# revisions referenced are present as they are installed in topological order.
1599
# and the tip revision was validated by get_ancestry.
1600
return required_topo_revisions
1602
# if we just grabbed the possibly available ids, then
1603
# we only have an estimate of whats available and need to validate
1604
# that against the revision records.
1605
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1608
class InterModel1and2(InterRepository):
1611
def _get_repo_format_to_test(self):
1615
def is_compatible(source, target):
1616
if not source.supports_rich_root() and target.supports_rich_root():
1622
def fetch(self, revision_id=None, pb=None):
1623
"""See InterRepository.fetch()."""
1624
from bzrlib.fetch import Model1toKnit2Fetcher
1625
# TODO: jam 20070210 This should be an assert, not a translate
1626
revision_id = osutils.safe_revision_id(revision_id)
1627
f = Model1toKnit2Fetcher(to_repository=self.target,
1628
from_repository=self.source,
1629
last_revision=revision_id,
1631
return f.count_copied, f.failed_revisions
1634
def copy_content(self, revision_id=None, basis=None):
1635
"""Make a complete copy of the content in self into destination.
1637
This is a destructive operation! Do not use it on existing
1640
:param revision_id: Only copy the content needed to construct
1641
revision_id and its parents.
1642
:param basis: Copy the needed data preferentially from basis.
1645
self.target.set_make_working_trees(self.source.make_working_trees())
1646
except NotImplementedError:
1648
# TODO: jam 20070210 Internal, assert, don't translate
1649
revision_id = osutils.safe_revision_id(revision_id)
1650
# grab the basis available data
1651
if basis is not None:
1652
self.target.fetch(basis, revision_id=revision_id)
1653
# but don't bother fetching if we have the needed data now.
1654
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1655
self.target.has_revision(revision_id)):
1657
self.target.fetch(self.source, revision_id=revision_id)
1660
class InterKnit1and2(InterKnitRepo):
1663
def _get_repo_format_to_test(self):
1667
def is_compatible(source, target):
1668
"""Be compatible with Knit1 source and Knit2 target"""
1669
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit2
1671
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1672
RepositoryFormatKnit2
1673
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1674
isinstance(target._format, (RepositoryFormatKnit2)))
1675
except AttributeError:
1679
def fetch(self, revision_id=None, pb=None):
1680
"""See InterRepository.fetch()."""
1681
from bzrlib.fetch import Knit1to2Fetcher
1682
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1683
self.source, self.source._format, self.target,
1684
self.target._format)
1685
# TODO: jam 20070210 This should be an assert, not a translate
1686
revision_id = osutils.safe_revision_id(revision_id)
1687
f = Knit1to2Fetcher(to_repository=self.target,
1688
from_repository=self.source,
1689
last_revision=revision_id,
1691
return f.count_copied, f.failed_revisions
1694
InterRepository.register_optimiser(InterSameDataRepository)
1695
InterRepository.register_optimiser(InterWeaveRepo)
1696
InterRepository.register_optimiser(InterKnitRepo)
1697
InterRepository.register_optimiser(InterModel1and2)
1698
InterRepository.register_optimiser(InterKnit1and2)
1701
class RepositoryTestProviderAdapter(object):
1702
"""A tool to generate a suite testing multiple repository formats at once.
1704
This is done by copying the test once for each transport and injecting
1705
the transport_server, transport_readonly_server, and bzrdir_format and
1706
repository_format classes into each copy. Each copy is also given a new id()
1707
to make it easy to identify.
1710
def __init__(self, transport_server, transport_readonly_server, formats,
1711
vfs_transport_factory=None):
1712
self._transport_server = transport_server
1713
self._transport_readonly_server = transport_readonly_server
1714
self._vfs_transport_factory = vfs_transport_factory
1715
self._formats = formats
1717
def adapt(self, test):
1718
result = unittest.TestSuite()
1719
for repository_format, bzrdir_format in self._formats:
1720
from copy import deepcopy
1721
new_test = deepcopy(test)
1722
new_test.transport_server = self._transport_server
1723
new_test.transport_readonly_server = self._transport_readonly_server
1724
if self._vfs_transport_factory:
1725
new_test.vfs_transport_factory = self._vfs_transport_factory
1726
new_test.bzrdir_format = bzrdir_format
1727
new_test.repository_format = repository_format
1728
def make_new_test_id():
1729
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1730
return lambda: new_id
1731
new_test.id = make_new_test_id()
1732
result.addTest(new_test)
1736
class InterRepositoryTestProviderAdapter(object):
1737
"""A tool to generate a suite testing multiple inter repository formats.
1739
This is done by copying the test once for each interrepo provider and injecting
1740
the transport_server, transport_readonly_server, repository_format and
1741
repository_to_format classes into each copy.
1742
Each copy is also given a new id() to make it easy to identify.
1745
def __init__(self, transport_server, transport_readonly_server, formats):
1746
self._transport_server = transport_server
1747
self._transport_readonly_server = transport_readonly_server
1748
self._formats = formats
1750
def adapt(self, test):
1751
result = unittest.TestSuite()
1752
for interrepo_class, repository_format, repository_format_to in self._formats:
1753
from copy import deepcopy
1754
new_test = deepcopy(test)
1755
new_test.transport_server = self._transport_server
1756
new_test.transport_readonly_server = self._transport_readonly_server
1757
new_test.interrepo_class = interrepo_class
1758
new_test.repository_format = repository_format
1759
new_test.repository_format_to = repository_format_to
1760
def make_new_test_id():
1761
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1762
return lambda: new_id
1763
new_test.id = make_new_test_id()
1764
result.addTest(new_test)
1768
def default_test_list():
1769
"""Generate the default list of interrepo permutations to test."""
1770
from bzrlib.repofmt import knitrepo, weaverepo
1772
# test the default InterRepository between format 6 and the current
1774
# XXX: robertc 20060220 reinstate this when there are two supported
1775
# formats which do not have an optimal code path between them.
1776
#result.append((InterRepository,
1777
# RepositoryFormat6(),
1778
# RepositoryFormatKnit1()))
1779
for optimiser_class in InterRepository._optimisers:
1780
format_to_test = optimiser_class._get_repo_format_to_test()
1781
if format_to_test is not None:
1782
result.append((optimiser_class,
1783
format_to_test, format_to_test))
1784
# if there are specific combinations we want to use, we can add them
1786
result.append((InterModel1and2,
1787
weaverepo.RepositoryFormat5(),
1788
knitrepo.RepositoryFormatKnit2()))
1789
result.append((InterKnit1and2,
1790
knitrepo.RepositoryFormatKnit1(),
1791
knitrepo.RepositoryFormatKnit2()))
1795
class CopyConverter(object):
1796
"""A repository conversion tool which just performs a copy of the content.
1798
This is slow but quite reliable.
1801
def __init__(self, target_format):
1802
"""Create a CopyConverter.
1804
:param target_format: The format the resulting repository should be.
1806
self.target_format = target_format
1808
def convert(self, repo, pb):
1809
"""Perform the conversion of to_convert, giving feedback via pb.
1811
:param to_convert: The disk object to convert.
1812
:param pb: a progress bar to use for progress information.
1817
# this is only useful with metadir layouts - separated repo content.
1818
# trigger an assertion if not such
1819
repo._format.get_format_string()
1820
self.repo_dir = repo.bzrdir
1821
self.step('Moving repository to repository.backup')
1822
self.repo_dir.transport.move('repository', 'repository.backup')
1823
backup_transport = self.repo_dir.transport.clone('repository.backup')
1824
repo._format.check_conversion_target(self.target_format)
1825
self.source_repo = repo._format.open(self.repo_dir,
1827
_override_transport=backup_transport)
1828
self.step('Creating new repository')
1829
converted = self.target_format.initialize(self.repo_dir,
1830
self.source_repo.is_shared())
1831
converted.lock_write()
1833
self.step('Copying content into repository.')
1834
self.source_repo.copy_content_into(converted)
1837
self.step('Deleting old repository content.')
1838
self.repo_dir.transport.delete_tree('repository.backup')
1839
self.pb.note('repository converted')
1841
def step(self, message):
1842
"""Update the pb by a step."""
1844
self.pb.update(message, self.count, self.total)
1847
class CommitBuilder(object):
1848
"""Provides an interface to build up a commit.
1850
This allows describing a tree to be committed without needing to
1851
know the internals of the format of the repository.
1854
record_root_entry = False
1855
def __init__(self, repository, parents, config, timestamp=None,
1856
timezone=None, committer=None, revprops=None,
1858
"""Initiate a CommitBuilder.
1860
:param repository: Repository to commit to.
1861
:param parents: Revision ids of the parents of the new revision.
1862
:param config: Configuration to use.
1863
:param timestamp: Optional timestamp recorded for commit.
1864
:param timezone: Optional timezone for timestamp.
1865
:param committer: Optional committer to set for commit.
1866
:param revprops: Optional dictionary of revision properties.
1867
:param revision_id: Optional revision id.
1869
self._config = config
1871
if committer is None:
1872
self._committer = self._config.username()
1874
assert isinstance(committer, basestring), type(committer)
1875
self._committer = committer
1877
self.new_inventory = Inventory(None)
1878
self._new_revision_id = osutils.safe_revision_id(revision_id)
1879
self.parents = parents
1880
self.repository = repository
1883
if revprops is not None:
1884
self._revprops.update(revprops)
1886
if timestamp is None:
1887
timestamp = time.time()
1888
# Restrict resolution to 1ms
1889
self._timestamp = round(timestamp, 3)
1891
if timezone is None:
1892
self._timezone = osutils.local_time_offset()
1894
self._timezone = int(timezone)
1896
self._generate_revision_if_needed()
1898
def commit(self, message):
1899
"""Make the actual commit.
1901
:return: The revision id of the recorded revision.
1903
rev = _mod_revision.Revision(
1904
timestamp=self._timestamp,
1905
timezone=self._timezone,
1906
committer=self._committer,
1908
inventory_sha1=self.inv_sha1,
1909
revision_id=self._new_revision_id,
1910
properties=self._revprops)
1911
rev.parent_ids = self.parents
1912
self.repository.add_revision(self._new_revision_id, rev,
1913
self.new_inventory, self._config)
1914
return self._new_revision_id
1916
def revision_tree(self):
1917
"""Return the tree that was just committed.
1919
After calling commit() this can be called to get a RevisionTree
1920
representing the newly committed tree. This is preferred to
1921
calling Repository.revision_tree() because that may require
1922
deserializing the inventory, while we already have a copy in
1925
return RevisionTree(self.repository, self.new_inventory,
1926
self._new_revision_id)
1928
def finish_inventory(self):
1929
"""Tell the builder that the inventory is finished."""
1930
if self.new_inventory.root is None:
1931
symbol_versioning.warn('Root entry should be supplied to'
1932
' record_entry_contents, as of bzr 0.10.',
1933
DeprecationWarning, stacklevel=2)
1934
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1935
self.new_inventory.revision_id = self._new_revision_id
1936
self.inv_sha1 = self.repository.add_inventory(
1937
self._new_revision_id,
1942
def _gen_revision_id(self):
1943
"""Return new revision-id."""
1944
return generate_ids.gen_revision_id(self._config.username(),
1947
def _generate_revision_if_needed(self):
1948
"""Create a revision id if None was supplied.
1950
If the repository can not support user-specified revision ids
1951
they should override this function and raise CannotSetRevisionId
1952
if _new_revision_id is not None.
1954
:raises: CannotSetRevisionId
1956
if self._new_revision_id is None:
1957
self._new_revision_id = self._gen_revision_id()
1959
def record_entry_contents(self, ie, parent_invs, path, tree):
1960
"""Record the content of ie from tree into the commit if needed.
1962
Side effect: sets ie.revision when unchanged
1964
:param ie: An inventory entry present in the commit.
1965
:param parent_invs: The inventories of the parent revisions of the
1967
:param path: The path the entry is at in the tree.
1968
:param tree: The tree which contains this entry and should be used to
1971
if self.new_inventory.root is None and ie.parent_id is not None:
1972
symbol_versioning.warn('Root entry should be supplied to'
1973
' record_entry_contents, as of bzr 0.10.',
1974
DeprecationWarning, stacklevel=2)
1975
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1977
self.new_inventory.add(ie)
1979
# ie.revision is always None if the InventoryEntry is considered
1980
# for committing. ie.snapshot will record the correct revision
1981
# which may be the sole parent if it is untouched.
1982
if ie.revision is not None:
1985
# In this revision format, root entries have no knit or weave
1986
if ie is self.new_inventory.root:
1987
# When serializing out to disk and back in
1988
# root.revision is always _new_revision_id
1989
ie.revision = self._new_revision_id
1991
previous_entries = ie.find_previous_heads(
1993
self.repository.weave_store,
1994
self.repository.get_transaction())
1995
# we are creating a new revision for ie in the history store
1997
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
1999
def modified_directory(self, file_id, file_parents):
2000
"""Record the presence of a symbolic link.
2002
:param file_id: The file_id of the link to record.
2003
:param file_parents: The per-file parent revision ids.
2005
self._add_text_to_weave(file_id, [], file_parents.keys())
2007
def modified_file_text(self, file_id, file_parents,
2008
get_content_byte_lines, text_sha1=None,
2010
"""Record the text of file file_id
2012
:param file_id: The file_id of the file to record the text of.
2013
:param file_parents: The per-file parent revision ids.
2014
:param get_content_byte_lines: A callable which will return the byte
2016
:param text_sha1: Optional SHA1 of the file contents.
2017
:param text_size: Optional size of the file contents.
2019
# mutter('storing text of file {%s} in revision {%s} into %r',
2020
# file_id, self._new_revision_id, self.repository.weave_store)
2021
# special case to avoid diffing on renames or
2023
if (len(file_parents) == 1
2024
and text_sha1 == file_parents.values()[0].text_sha1
2025
and text_size == file_parents.values()[0].text_size):
2026
previous_ie = file_parents.values()[0]
2027
versionedfile = self.repository.weave_store.get_weave(file_id,
2028
self.repository.get_transaction())
2029
versionedfile.clone_text(self._new_revision_id,
2030
previous_ie.revision, file_parents.keys())
2031
return text_sha1, text_size
2033
new_lines = get_content_byte_lines()
2034
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2035
# should return the SHA1 and size
2036
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2037
return osutils.sha_strings(new_lines), \
2038
sum(map(len, new_lines))
2040
def modified_link(self, file_id, file_parents, link_target):
2041
"""Record the presence of a symbolic link.
2043
:param file_id: The file_id of the link to record.
2044
:param file_parents: The per-file parent revision ids.
2045
:param link_target: Target location of this link.
2047
self._add_text_to_weave(file_id, [], file_parents.keys())
2049
def _add_text_to_weave(self, file_id, new_lines, parents):
2050
versionedfile = self.repository.weave_store.get_weave_or_empty(
2051
file_id, self.repository.get_transaction())
2052
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2053
versionedfile.clear_cache()
2056
class _CommitBuilder(CommitBuilder):
2057
"""Temporary class so old CommitBuilders are detected properly
2059
Note: CommitBuilder works whether or not root entry is recorded.
2062
record_root_entry = True
2065
class RootCommitBuilder(CommitBuilder):
2066
"""This commitbuilder actually records the root id"""
2068
record_root_entry = True
2070
def record_entry_contents(self, ie, parent_invs, path, tree):
2071
"""Record the content of ie from tree into the commit if needed.
2073
Side effect: sets ie.revision when unchanged
2075
:param ie: An inventory entry present in the commit.
2076
:param parent_invs: The inventories of the parent revisions of the
2078
:param path: The path the entry is at in the tree.
2079
:param tree: The tree which contains this entry and should be used to
2082
assert self.new_inventory.root is not None or ie.parent_id is None
2083
self.new_inventory.add(ie)
2085
# ie.revision is always None if the InventoryEntry is considered
2086
# for committing. ie.snapshot will record the correct revision
2087
# which may be the sole parent if it is untouched.
2088
if ie.revision is not None:
2091
previous_entries = ie.find_previous_heads(
2093
self.repository.weave_store,
2094
self.repository.get_transaction())
2095
# we are creating a new revision for ie in the history store
2097
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2109
def _unescaper(match, _map=_unescape_map):
2110
code = match.group(1)
2114
if not code.startswith('#'):
2116
return unichr(int(code[1:])).encode('utf8')
2122
def _unescape_xml(data):
2123
"""Unescape predefined XML entities in a string of data."""
2125
if _unescape_re is None:
2126
_unescape_re = re.compile('\&([^;]*);')
2127
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
