1
# Copyright (C) 2005, 2006 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 binascii import hexlify
18
from copy import deepcopy
19
from cStringIO import StringIO
22
from unittest import TestSuite
24
from bzrlib import bzrdir, check, gpg, errors, xml5, ui, transactions, osutils
25
from bzrlib.decorators import needs_read_lock, needs_write_lock
26
from bzrlib.errors import InvalidRevisionId
27
from bzrlib.graph import Graph
28
from bzrlib.inter import InterObject
29
from bzrlib.inventory import Inventory
30
from bzrlib.knit import KnitVersionedFile, KnitPlainFactory
31
from bzrlib.lockable_files import LockableFiles, TransportLock
32
from bzrlib.lockdir import LockDir
33
from bzrlib.osutils import (safe_unicode, rand_bytes, compact_date,
35
from bzrlib.revision import NULL_REVISION, Revision
36
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
37
from bzrlib.store.text import TextStore
38
from bzrlib.symbol_versioning import (deprecated_method,
41
from bzrlib.trace import mutter, note
42
from bzrlib.tree import RevisionTree
43
from bzrlib.tsort import topo_sort
44
from bzrlib.testament import Testament
45
from bzrlib.tree import EmptyTree
46
from bzrlib.weave import WeaveFile
49
class Repository(object):
50
"""Repository holding history for one or more branches.
52
The repository holds and retrieves historical information including
53
revisions and file history. It's normally accessed only by the Branch,
54
which views a particular line of development through that history.
56
The Repository builds on top of Stores and a Transport, which respectively
57
describe the disk data format and the way of accessing the (possibly
62
def add_inventory(self, revid, inv, parents):
63
"""Add the inventory inv to the repository as revid.
65
:param parents: The revision ids of the parents that revid
66
is known to have and are in the repository already.
68
returns the sha1 of the serialized inventory.
70
assert inv.revision_id is None or inv.revision_id == revid, \
71
"Mismatch between inventory revision" \
72
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
73
inv_text = xml5.serializer_v5.write_inventory_to_string(inv)
74
inv_sha1 = osutils.sha_string(inv_text)
75
inv_vf = self.control_weaves.get_weave('inventory',
76
self.get_transaction())
77
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
80
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
82
for parent in parents:
84
final_parents.append(parent)
86
inv_vf.add_lines(revid, final_parents, lines)
89
def add_revision(self, rev_id, rev, inv=None, config=None):
90
"""Add rev to the revision store as rev_id.
92
:param rev_id: the revision id to use.
93
:param rev: The revision object.
94
:param inv: The inventory for the revision. if None, it will be looked
95
up in the inventory storer
96
:param config: If None no digital signature will be created.
97
If supplied its signature_needed method will be used
98
to determine if a signature should be made.
100
if config is not None and config.signature_needed():
102
inv = self.get_inventory(rev_id)
103
plaintext = Testament(rev, inv).as_short_text()
104
self.store_revision_signature(
105
gpg.GPGStrategy(config), plaintext, rev_id)
106
if not rev_id in self.get_inventory_weave():
108
raise errors.WeaveRevisionNotPresent(rev_id,
109
self.get_inventory_weave())
111
# yes, this is not suitable for adding with ghosts.
112
self.add_inventory(rev_id, inv, rev.parent_ids)
113
self._revision_store.add_revision(rev, self.get_transaction())
116
def _all_possible_ids(self):
117
"""Return all the possible revisions that we could find."""
118
return self.get_inventory_weave().versions()
120
def all_revision_ids(self):
121
"""Returns a list of all the revision ids in the repository.
123
This is deprecated because code should generally work on the graph
124
reachable from a particular revision, and ignore any other revisions
125
that might be present. There is no direct replacement method.
127
return self._all_revision_ids()
130
def _all_revision_ids(self):
131
"""Returns a list of all the revision ids in the repository.
133
These are in as much topological order as the underlying store can
134
present: for weaves ghosts may lead to a lack of correctness until
135
the reweave updates the parents list.
137
if self._revision_store.text_store.listable():
138
return self._revision_store.all_revision_ids(self.get_transaction())
139
result = self._all_possible_ids()
140
return self._eliminate_revisions_not_present(result)
142
def break_lock(self):
143
"""Break a lock if one is present from another instance.
145
Uses the ui factory to ask for confirmation if the lock may be from
148
self.control_files.break_lock()
151
def _eliminate_revisions_not_present(self, revision_ids):
152
"""Check every revision id in revision_ids to see if we have it.
154
Returns a set of the present revisions.
157
for id in revision_ids:
158
if self.has_revision(id):
163
def create(a_bzrdir):
164
"""Construct the current default format repository in a_bzrdir."""
165
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
167
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
168
"""instantiate a Repository.
170
:param _format: The format of the repository on disk.
171
:param a_bzrdir: The BzrDir of the repository.
173
In the future we will have a single api for all stores for
174
getting file texts, inventories and revisions, then
175
this construct will accept instances of those things.
177
super(Repository, self).__init__()
178
self._format = _format
179
# the following are part of the public API for Repository:
180
self.bzrdir = a_bzrdir
181
self.control_files = control_files
182
self._revision_store = _revision_store
183
self.text_store = text_store
184
# backwards compatibility
185
self.weave_store = text_store
186
# not right yet - should be more semantically clear ?
188
self.control_store = control_store
189
self.control_weaves = control_store
190
# TODO: make sure to construct the right store classes, etc, depending
191
# on whether escaping is required.
194
return '%s(%r)' % (self.__class__.__name__,
195
self.bzrdir.transport.base)
198
return self.control_files.is_locked()
200
def lock_write(self):
201
self.control_files.lock_write()
204
self.control_files.lock_read()
206
def get_physical_lock_status(self):
207
return self.control_files.get_physical_lock_status()
210
def missing_revision_ids(self, other, revision_id=None):
211
"""Return the revision ids that other has that this does not.
213
These are returned in topological order.
215
revision_id: only return revision ids included by revision_id.
217
return InterRepository.get(other, self).missing_revision_ids(revision_id)
221
"""Open the repository rooted at base.
223
For instance, if the repository is at URL/.bzr/repository,
224
Repository.open(URL) -> a Repository instance.
226
control = bzrdir.BzrDir.open(base)
227
return control.open_repository()
229
def copy_content_into(self, destination, revision_id=None, basis=None):
230
"""Make a complete copy of the content in self into destination.
232
This is a destructive operation! Do not use it on existing
235
return InterRepository.get(self, destination).copy_content(revision_id, basis)
237
def fetch(self, source, revision_id=None, pb=None):
238
"""Fetch the content required to construct revision_id from source.
240
If revision_id is None all content is copied.
242
return InterRepository.get(source, self).fetch(revision_id=revision_id,
245
def get_commit_builder(self, branch, parents, config, timestamp=None,
246
timezone=None, committer=None, revprops=None,
248
"""Obtain a CommitBuilder for this repository.
250
:param branch: Branch to commit to.
251
:param parents: Revision ids of the parents of the new revision.
252
:param config: Configuration to use.
253
:param timestamp: Optional timestamp recorded for commit.
254
:param timezone: Optional timezone for timestamp.
255
:param committer: Optional committer to set for commit.
256
:param revprops: Optional dictionary of revision properties.
257
:param revision_id: Optional revision id.
259
return CommitBuilder(self, parents, config, timestamp, timezone,
260
committer, revprops, revision_id)
263
self.control_files.unlock()
266
def clone(self, a_bzrdir, revision_id=None, basis=None):
267
"""Clone this repository into a_bzrdir using the current format.
269
Currently no check is made that the format of this repository and
270
the bzrdir format are compatible. FIXME RBC 20060201.
272
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
273
# use target default format.
274
result = a_bzrdir.create_repository()
275
# FIXME RBC 20060209 split out the repository type to avoid this check ?
276
elif isinstance(a_bzrdir._format,
277
(bzrdir.BzrDirFormat4,
278
bzrdir.BzrDirFormat5,
279
bzrdir.BzrDirFormat6)):
280
result = a_bzrdir.open_repository()
282
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
283
self.copy_content_into(result, revision_id, basis)
287
def has_revision(self, revision_id):
288
"""True if this repository has a copy of the revision."""
289
return self._revision_store.has_revision_id(revision_id,
290
self.get_transaction())
293
def get_revision_reconcile(self, revision_id):
294
"""'reconcile' helper routine that allows access to a revision always.
296
This variant of get_revision does not cross check the weave graph
297
against the revision one as get_revision does: but it should only
298
be used by reconcile, or reconcile-alike commands that are correcting
299
or testing the revision graph.
301
if not revision_id or not isinstance(revision_id, basestring):
302
raise InvalidRevisionId(revision_id=revision_id, branch=self)
303
return self._revision_store.get_revision(revision_id,
304
self.get_transaction())
307
def get_revision_xml(self, revision_id):
308
rev = self.get_revision(revision_id)
310
# the current serializer..
311
self._revision_store._serializer.write_revision(rev, rev_tmp)
313
return rev_tmp.getvalue()
316
def get_revision(self, revision_id):
317
"""Return the Revision object for a named revision"""
318
r = self.get_revision_reconcile(revision_id)
319
# weave corruption can lead to absent revision markers that should be
321
# the following test is reasonably cheap (it needs a single weave read)
322
# and the weave is cached in read transactions. In write transactions
323
# it is not cached but typically we only read a small number of
324
# revisions. For knits when they are introduced we will probably want
325
# to ensure that caching write transactions are in use.
326
inv = self.get_inventory_weave()
327
self._check_revision_parents(r, inv)
330
def _check_revision_parents(self, revision, inventory):
331
"""Private to Repository and Fetch.
333
This checks the parentage of revision in an inventory weave for
334
consistency and is only applicable to inventory-weave-for-ancestry
335
using repository formats & fetchers.
337
weave_parents = inventory.get_parents(revision.revision_id)
338
weave_names = inventory.versions()
339
for parent_id in revision.parent_ids:
340
if parent_id in weave_names:
341
# this parent must not be a ghost.
342
if not parent_id in weave_parents:
344
raise errors.CorruptRepository(self)
347
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
348
signature = gpg_strategy.sign(plaintext)
349
self._revision_store.add_revision_signature_text(revision_id,
351
self.get_transaction())
353
def fileids_altered_by_revision_ids(self, revision_ids):
354
"""Find the file ids and versions affected by revisions.
356
:param revisions: an iterable containing revision ids.
357
:return: a dictionary mapping altered file-ids to an iterable of
358
revision_ids. Each altered file-ids has the exact revision_ids that
359
altered it listed explicitly.
361
assert isinstance(self._format, (RepositoryFormat5,
364
RepositoryFormatKnit1)), \
365
("fileids_altered_by_revision_ids only supported for branches "
366
"which store inventory as unnested xml, not on %r" % self)
367
selected_revision_ids = set(revision_ids)
368
w = self.get_inventory_weave()
371
# this code needs to read every new line in every inventory for the
372
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
373
# not present in one of those inventories is unnecessary but not
374
# harmful because we are filtering by the revision id marker in the
375
# inventory lines : we only select file ids altered in one of those
376
# revisions. We don't need to see all lines in the inventory because
377
# only those added in an inventory in rev X can contain a revision=X
379
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
380
start = line.find('file_id="')+9
381
if start < 9: continue
382
end = line.find('"', start)
384
file_id = _unescape_xml(line[start:end])
386
start = line.find('revision="')+10
387
if start < 10: continue
388
end = line.find('"', start)
390
revision_id = _unescape_xml(line[start:end])
391
if revision_id in selected_revision_ids:
392
result.setdefault(file_id, set()).add(revision_id)
396
def get_inventory_weave(self):
397
return self.control_weaves.get_weave('inventory',
398
self.get_transaction())
401
def get_inventory(self, revision_id):
402
"""Get Inventory object by hash."""
403
return self.deserialise_inventory(
404
revision_id, self.get_inventory_xml(revision_id))
406
def deserialise_inventory(self, revision_id, xml):
407
"""Transform the xml into an inventory object.
409
:param revision_id: The expected revision id of the inventory.
410
:param xml: A serialised inventory.
412
return xml5.serializer_v5.read_inventory_from_string(xml)
415
def get_inventory_xml(self, revision_id):
416
"""Get inventory XML as a file object."""
418
assert isinstance(revision_id, basestring), type(revision_id)
419
iw = self.get_inventory_weave()
420
return iw.get_text(revision_id)
422
raise errors.HistoryMissing(self, 'inventory', revision_id)
425
def get_inventory_sha1(self, revision_id):
426
"""Return the sha1 hash of the inventory entry
428
return self.get_revision(revision_id).inventory_sha1
431
def get_revision_graph(self, revision_id=None):
432
"""Return a dictionary containing the revision graph.
434
:return: a dictionary of revision_id->revision_parents_list.
436
weave = self.get_inventory_weave()
437
all_revisions = self._eliminate_revisions_not_present(weave.versions())
438
entire_graph = dict([(node, weave.get_parents(node)) for
439
node in all_revisions])
440
if revision_id is None:
442
elif revision_id not in entire_graph:
443
raise errors.NoSuchRevision(self, revision_id)
445
# add what can be reached from revision_id
447
pending = set([revision_id])
448
while len(pending) > 0:
450
result[node] = entire_graph[node]
451
for revision_id in result[node]:
452
if revision_id not in result:
453
pending.add(revision_id)
457
def get_revision_graph_with_ghosts(self, revision_ids=None):
458
"""Return a graph of the revisions with ghosts marked as applicable.
460
:param revision_ids: an iterable of revisions to graph or None for all.
461
:return: a Graph object with the graph reachable from revision_ids.
465
pending = set(self.all_revision_ids())
468
pending = set(revision_ids)
469
required = set(revision_ids)
472
revision_id = pending.pop()
474
rev = self.get_revision(revision_id)
475
except errors.NoSuchRevision:
476
if revision_id in required:
479
result.add_ghost(revision_id)
481
for parent_id in rev.parent_ids:
482
# is this queued or done ?
483
if (parent_id not in pending and
484
parent_id not in done):
486
pending.add(parent_id)
487
result.add_node(revision_id, rev.parent_ids)
488
done.add(revision_id)
492
def get_revision_inventory(self, revision_id):
493
"""Return inventory of a past revision."""
494
# TODO: Unify this with get_inventory()
495
# bzr 0.0.6 and later imposes the constraint that the inventory_id
496
# must be the same as its revision, so this is trivial.
497
if revision_id is None:
498
# This does not make sense: if there is no revision,
499
# then it is the current tree inventory surely ?!
500
# and thus get_root_id() is something that looks at the last
501
# commit on the branch, and the get_root_id is an inventory check.
502
raise NotImplementedError
503
# return Inventory(self.get_root_id())
505
return self.get_inventory(revision_id)
509
"""Return True if this repository is flagged as a shared repository."""
510
raise NotImplementedError(self.is_shared)
513
def reconcile(self, other=None, thorough=False):
514
"""Reconcile this repository."""
515
from bzrlib.reconcile import RepoReconciler
516
reconciler = RepoReconciler(self, thorough=thorough)
517
reconciler.reconcile()
521
def revision_tree(self, revision_id):
522
"""Return Tree for a revision on this branch.
524
`revision_id` may be None for the null revision, in which case
525
an `EmptyTree` is returned."""
526
# TODO: refactor this to use an existing revision object
527
# so we don't need to read it in twice.
528
if revision_id is None or revision_id == NULL_REVISION:
531
inv = self.get_revision_inventory(revision_id)
532
return RevisionTree(self, inv, revision_id)
535
def get_ancestry(self, revision_id):
536
"""Return a list of revision-ids integrated by a revision.
538
The first element of the list is always None, indicating the origin
539
revision. This might change when we have history horizons, or
540
perhaps we should have a new API.
542
This is topologically sorted.
544
if revision_id is None:
546
if not self.has_revision(revision_id):
547
raise errors.NoSuchRevision(self, revision_id)
548
w = self.get_inventory_weave()
549
candidates = w.get_ancestry(revision_id)
550
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
553
def print_file(self, file, revision_id):
554
"""Print `file` to stdout.
556
FIXME RBC 20060125 as John Meinel points out this is a bad api
557
- it writes to stdout, it assumes that that is valid etc. Fix
558
by creating a new more flexible convenience function.
560
tree = self.revision_tree(revision_id)
561
# use inventory as it was in that revision
562
file_id = tree.inventory.path2id(file)
564
# TODO: jam 20060427 Write a test for this code path
565
# it had a bug in it, and was raising the wrong
567
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
568
tree.print_file(file_id)
570
def get_transaction(self):
571
return self.control_files.get_transaction()
573
def revision_parents(self, revid):
574
return self.get_inventory_weave().parent_names(revid)
577
def set_make_working_trees(self, new_value):
578
"""Set the policy flag for making working trees when creating branches.
580
This only applies to branches that use this repository.
582
The default is 'True'.
583
:param new_value: True to restore the default, False to disable making
586
raise NotImplementedError(self.set_make_working_trees)
588
def make_working_trees(self):
589
"""Returns the policy for making working trees on new branches."""
590
raise NotImplementedError(self.make_working_trees)
593
def sign_revision(self, revision_id, gpg_strategy):
594
plaintext = Testament.from_revision(self, revision_id).as_short_text()
595
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
598
def has_signature_for_revision_id(self, revision_id):
599
"""Query for a revision signature for revision_id in the repository."""
600
return self._revision_store.has_signature(revision_id,
601
self.get_transaction())
604
def get_signature_text(self, revision_id):
605
"""Return the text for a signature."""
606
return self._revision_store.get_signature_text(revision_id,
607
self.get_transaction())
610
def check(self, revision_ids):
611
"""Check consistency of all history of given revision_ids.
613
Different repository implementations should override _check().
615
:param revision_ids: A non-empty list of revision_ids whose ancestry
616
will be checked. Typically the last revision_id of a branch.
619
raise ValueError("revision_ids must be non-empty in %s.check"
621
return self._check(revision_ids)
623
def _check(self, revision_ids):
624
result = check.Check(self)
629
class AllInOneRepository(Repository):
630
"""Legacy support - the repository behaviour for all-in-one branches."""
632
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
633
# we reuse one control files instance.
634
dir_mode = a_bzrdir._control_files._dir_mode
635
file_mode = a_bzrdir._control_files._file_mode
637
def get_store(name, compressed=True, prefixed=False):
638
# FIXME: This approach of assuming stores are all entirely compressed
639
# or entirely uncompressed is tidy, but breaks upgrade from
640
# some existing branches where there's a mixture; we probably
641
# still want the option to look for both.
642
relpath = a_bzrdir._control_files._escape(name)
643
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
644
prefixed=prefixed, compressed=compressed,
647
#if self._transport.should_cache():
648
# cache_path = os.path.join(self.cache_root, name)
649
# os.mkdir(cache_path)
650
# store = bzrlib.store.CachedStore(store, cache_path)
653
# not broken out yet because the controlweaves|inventory_store
654
# and text_store | weave_store bits are still different.
655
if isinstance(_format, RepositoryFormat4):
656
# cannot remove these - there is still no consistent api
657
# which allows access to this old info.
658
self.inventory_store = get_store('inventory-store')
659
text_store = get_store('text-store')
660
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
664
"""AllInOne repositories cannot be shared."""
668
def set_make_working_trees(self, new_value):
669
"""Set the policy flag for making working trees when creating branches.
671
This only applies to branches that use this repository.
673
The default is 'True'.
674
:param new_value: True to restore the default, False to disable making
677
raise NotImplementedError(self.set_make_working_trees)
679
def make_working_trees(self):
680
"""Returns the policy for making working trees on new branches."""
684
def install_revision(repository, rev, revision_tree):
685
"""Install all revision data into a repository."""
688
for p_id in rev.parent_ids:
689
if repository.has_revision(p_id):
690
present_parents.append(p_id)
691
parent_trees[p_id] = repository.revision_tree(p_id)
693
parent_trees[p_id] = EmptyTree()
695
inv = revision_tree.inventory
697
# Add the texts that are not already present
698
for path, ie in inv.iter_entries():
699
w = repository.weave_store.get_weave_or_empty(ie.file_id,
700
repository.get_transaction())
701
if ie.revision not in w:
703
# FIXME: TODO: The following loop *may* be overlapping/duplicate
704
# with InventoryEntry.find_previous_heads(). if it is, then there
705
# is a latent bug here where the parents may have ancestors of each
707
for revision, tree in parent_trees.iteritems():
708
if ie.file_id not in tree:
710
parent_id = tree.inventory[ie.file_id].revision
711
if parent_id in text_parents:
713
text_parents.append(parent_id)
715
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
716
repository.get_transaction())
717
lines = revision_tree.get_file(ie.file_id).readlines()
718
vfile.add_lines(rev.revision_id, text_parents, lines)
720
# install the inventory
721
repository.add_inventory(rev.revision_id, inv, present_parents)
722
except errors.RevisionAlreadyPresent:
724
repository.add_revision(rev.revision_id, rev, inv)
727
class MetaDirRepository(Repository):
728
"""Repositories in the new meta-dir layout."""
730
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
731
super(MetaDirRepository, self).__init__(_format,
738
dir_mode = self.control_files._dir_mode
739
file_mode = self.control_files._file_mode
743
"""Return True if this repository is flagged as a shared repository."""
744
return self.control_files._transport.has('shared-storage')
747
def set_make_working_trees(self, new_value):
748
"""Set the policy flag for making working trees when creating branches.
750
This only applies to branches that use this repository.
752
The default is 'True'.
753
:param new_value: True to restore the default, False to disable making
758
self.control_files._transport.delete('no-working-trees')
759
except errors.NoSuchFile:
762
self.control_files.put_utf8('no-working-trees', '')
764
def make_working_trees(self):
765
"""Returns the policy for making working trees on new branches."""
766
return not self.control_files._transport.has('no-working-trees')
769
class KnitRepository(MetaDirRepository):
770
"""Knit format repository."""
772
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
773
inv_vf.add_lines_with_ghosts(revid, parents, lines)
776
def _all_revision_ids(self):
777
"""See Repository.all_revision_ids()."""
778
# Knits get the revision graph from the index of the revision knit, so
779
# it's always possible even if they're on an unlistable transport.
780
return self._revision_store.all_revision_ids(self.get_transaction())
782
def fileid_involved_between_revs(self, from_revid, to_revid):
783
"""Find file_id(s) which are involved in the changes between revisions.
785
This determines the set of revisions which are involved, and then
786
finds all file ids affected by those revisions.
788
vf = self._get_revision_vf()
789
from_set = set(vf.get_ancestry(from_revid))
790
to_set = set(vf.get_ancestry(to_revid))
791
changed = to_set.difference(from_set)
792
return self._fileid_involved_by_set(changed)
794
def fileid_involved(self, last_revid=None):
795
"""Find all file_ids modified in the ancestry of last_revid.
797
:param last_revid: If None, last_revision() will be used.
800
changed = set(self.all_revision_ids())
802
changed = set(self.get_ancestry(last_revid))
805
return self._fileid_involved_by_set(changed)
808
def get_ancestry(self, revision_id):
809
"""Return a list of revision-ids integrated by a revision.
811
This is topologically sorted.
813
if revision_id is None:
815
vf = self._get_revision_vf()
817
return [None] + vf.get_ancestry(revision_id)
818
except errors.RevisionNotPresent:
819
raise errors.NoSuchRevision(self, revision_id)
822
def get_revision(self, revision_id):
823
"""Return the Revision object for a named revision"""
824
return self.get_revision_reconcile(revision_id)
827
def get_revision_graph(self, revision_id=None):
828
"""Return a dictionary containing the revision graph.
830
:return: a dictionary of revision_id->revision_parents_list.
832
weave = self._get_revision_vf()
833
entire_graph = weave.get_graph()
834
if revision_id is None:
835
return weave.get_graph()
836
elif revision_id not in weave:
837
raise errors.NoSuchRevision(self, revision_id)
839
# add what can be reached from revision_id
841
pending = set([revision_id])
842
while len(pending) > 0:
844
result[node] = weave.get_parents(node)
845
for revision_id in result[node]:
846
if revision_id not in result:
847
pending.add(revision_id)
851
def get_revision_graph_with_ghosts(self, revision_ids=None):
852
"""Return a graph of the revisions with ghosts marked as applicable.
854
:param revision_ids: an iterable of revisions to graph or None for all.
855
:return: a Graph object with the graph reachable from revision_ids.
858
vf = self._get_revision_vf()
859
versions = set(vf.versions())
861
pending = set(self.all_revision_ids())
864
pending = set(revision_ids)
865
required = set(revision_ids)
868
revision_id = pending.pop()
869
if not revision_id in versions:
870
if revision_id in required:
871
raise errors.NoSuchRevision(self, revision_id)
873
result.add_ghost(revision_id)
874
# mark it as done so we don't try for it again.
875
done.add(revision_id)
877
parent_ids = vf.get_parents_with_ghosts(revision_id)
878
for parent_id in parent_ids:
879
# is this queued or done ?
880
if (parent_id not in pending and
881
parent_id not in done):
883
pending.add(parent_id)
884
result.add_node(revision_id, parent_ids)
885
done.add(revision_id)
888
def _get_revision_vf(self):
889
""":return: a versioned file containing the revisions."""
890
vf = self._revision_store.get_revision_file(self.get_transaction())
894
def reconcile(self, other=None, thorough=False):
895
"""Reconcile this repository."""
896
from bzrlib.reconcile import KnitReconciler
897
reconciler = KnitReconciler(self, thorough=thorough)
898
reconciler.reconcile()
901
def revision_parents(self, revision_id):
902
return self._get_revision_vf().get_parents(revision_id)
905
class RepositoryFormat(object):
906
"""A repository format.
908
Formats provide three things:
909
* An initialization routine to construct repository data on disk.
910
* a format string which is used when the BzrDir supports versioned
912
* an open routine which returns a Repository instance.
914
Formats are placed in an dict by their format string for reference
915
during opening. These should be subclasses of RepositoryFormat
918
Once a format is deprecated, just deprecate the initialize and open
919
methods on the format class. Do not deprecate the object, as the
920
object will be created every system load.
922
Common instance attributes:
923
_matchingbzrdir - the bzrdir format that the repository format was
924
originally written to work with. This can be used if manually
925
constructing a bzrdir and repository, or more commonly for test suite
929
_default_format = None
930
"""The default format used for new repositories."""
933
"""The known formats."""
936
def find_format(klass, a_bzrdir):
937
"""Return the format for the repository object in a_bzrdir."""
939
transport = a_bzrdir.get_repository_transport(None)
940
format_string = transport.get("format").read()
941
return klass._formats[format_string]
942
except errors.NoSuchFile:
943
raise errors.NoRepositoryPresent(a_bzrdir)
945
raise errors.UnknownFormatError(format_string)
947
def _get_control_store(self, repo_transport, control_files):
948
"""Return the control store for this repository."""
949
raise NotImplementedError(self._get_control_store)
952
def get_default_format(klass):
953
"""Return the current default format."""
954
return klass._default_format
956
def get_format_string(self):
957
"""Return the ASCII format string that identifies this format.
959
Note that in pre format ?? repositories the format string is
960
not permitted nor written to disk.
962
raise NotImplementedError(self.get_format_string)
964
def get_format_description(self):
965
"""Return the short description for this format."""
966
raise NotImplementedError(self.get_format_description)
968
def _get_revision_store(self, repo_transport, control_files):
969
"""Return the revision store object for this a_bzrdir."""
970
raise NotImplementedError(self._get_revision_store)
972
def _get_text_rev_store(self,
979
"""Common logic for getting a revision store for a repository.
981
see self._get_revision_store for the subclass-overridable method to
982
get the store for a repository.
984
from bzrlib.store.revision.text import TextRevisionStore
985
dir_mode = control_files._dir_mode
986
file_mode = control_files._file_mode
987
text_store =TextStore(transport.clone(name),
989
compressed=compressed,
992
_revision_store = TextRevisionStore(text_store, serializer)
993
return _revision_store
995
def _get_versioned_file_store(self,
1000
versionedfile_class=WeaveFile,
1002
weave_transport = control_files._transport.clone(name)
1003
dir_mode = control_files._dir_mode
1004
file_mode = control_files._file_mode
1005
return VersionedFileStore(weave_transport, prefixed=prefixed,
1007
file_mode=file_mode,
1008
versionedfile_class=versionedfile_class,
1011
def initialize(self, a_bzrdir, shared=False):
1012
"""Initialize a repository of this format in a_bzrdir.
1014
:param a_bzrdir: The bzrdir to put the new repository in it.
1015
:param shared: The repository should be initialized as a sharable one.
1017
This may raise UninitializableFormat if shared repository are not
1018
compatible the a_bzrdir.
1021
def is_supported(self):
1022
"""Is this format supported?
1024
Supported formats must be initializable and openable.
1025
Unsupported formats may not support initialization or committing or
1026
some other features depending on the reason for not being supported.
1030
def open(self, a_bzrdir, _found=False):
1031
"""Return an instance of this format for the bzrdir a_bzrdir.
1033
_found is a private parameter, do not use it.
1035
raise NotImplementedError(self.open)
1038
def register_format(klass, format):
1039
klass._formats[format.get_format_string()] = format
1042
def set_default_format(klass, format):
1043
klass._default_format = format
1046
def unregister_format(klass, format):
1047
assert klass._formats[format.get_format_string()] is format
1048
del klass._formats[format.get_format_string()]
1051
class PreSplitOutRepositoryFormat(RepositoryFormat):
1052
"""Base class for the pre split out repository formats."""
1054
def initialize(self, a_bzrdir, shared=False, _internal=False):
1055
"""Create a weave repository.
1057
TODO: when creating split out bzr branch formats, move this to a common
1058
base for Format5, Format6. or something like that.
1060
from bzrlib.weavefile import write_weave_v5
1061
from bzrlib.weave import Weave
1064
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1067
# always initialized when the bzrdir is.
1068
return self.open(a_bzrdir, _found=True)
1070
# Create an empty weave
1072
write_weave_v5(Weave(), sio)
1073
empty_weave = sio.getvalue()
1075
mutter('creating repository in %s.', a_bzrdir.transport.base)
1076
dirs = ['revision-store', 'weaves']
1077
files = [('inventory.weave', StringIO(empty_weave)),
1080
# FIXME: RBC 20060125 don't peek under the covers
1081
# NB: no need to escape relative paths that are url safe.
1082
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1084
control_files.create_lock()
1085
control_files.lock_write()
1086
control_files._transport.mkdir_multi(dirs,
1087
mode=control_files._dir_mode)
1089
for file, content in files:
1090
control_files.put(file, content)
1092
control_files.unlock()
1093
return self.open(a_bzrdir, _found=True)
1095
def _get_control_store(self, repo_transport, control_files):
1096
"""Return the control store for this repository."""
1097
return self._get_versioned_file_store('',
1102
def _get_text_store(self, transport, control_files):
1103
"""Get a store for file texts for this format."""
1104
raise NotImplementedError(self._get_text_store)
1106
def open(self, a_bzrdir, _found=False):
1107
"""See RepositoryFormat.open()."""
1109
# we are being called directly and must probe.
1110
raise NotImplementedError
1112
repo_transport = a_bzrdir.get_repository_transport(None)
1113
control_files = a_bzrdir._control_files
1114
text_store = self._get_text_store(repo_transport, control_files)
1115
control_store = self._get_control_store(repo_transport, control_files)
1116
_revision_store = self._get_revision_store(repo_transport, control_files)
1117
return AllInOneRepository(_format=self,
1119
_revision_store=_revision_store,
1120
control_store=control_store,
1121
text_store=text_store)
1124
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1125
"""Bzr repository format 4.
1127
This repository format has:
1129
- TextStores for texts, inventories,revisions.
1131
This format is deprecated: it indexes texts using a text id which is
1132
removed in format 5; initialization and write support for this format
1137
super(RepositoryFormat4, self).__init__()
1138
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1140
def get_format_description(self):
1141
"""See RepositoryFormat.get_format_description()."""
1142
return "Repository format 4"
1144
def initialize(self, url, shared=False, _internal=False):
1145
"""Format 4 branches cannot be created."""
1146
raise errors.UninitializableFormat(self)
1148
def is_supported(self):
1149
"""Format 4 is not supported.
1151
It is not supported because the model changed from 4 to 5 and the
1152
conversion logic is expensive - so doing it on the fly was not
1157
def _get_control_store(self, repo_transport, control_files):
1158
"""Format 4 repositories have no formal control store at this point.
1160
This will cause any control-file-needing apis to fail - this is desired.
1164
def _get_revision_store(self, repo_transport, control_files):
1165
"""See RepositoryFormat._get_revision_store()."""
1166
from bzrlib.xml4 import serializer_v4
1167
return self._get_text_rev_store(repo_transport,
1170
serializer=serializer_v4)
1172
def _get_text_store(self, transport, control_files):
1173
"""See RepositoryFormat._get_text_store()."""
1176
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1177
"""Bzr control format 5.
1179
This repository format has:
1180
- weaves for file texts and inventory
1182
- TextStores for revisions and signatures.
1186
super(RepositoryFormat5, self).__init__()
1187
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1189
def get_format_description(self):
1190
"""See RepositoryFormat.get_format_description()."""
1191
return "Weave repository format 5"
1193
def _get_revision_store(self, repo_transport, control_files):
1194
"""See RepositoryFormat._get_revision_store()."""
1195
"""Return the revision store object for this a_bzrdir."""
1196
return self._get_text_rev_store(repo_transport,
1201
def _get_text_store(self, transport, control_files):
1202
"""See RepositoryFormat._get_text_store()."""
1203
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1206
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1207
"""Bzr control format 6.
1209
This repository format has:
1210
- weaves for file texts and inventory
1211
- hash subdirectory based stores.
1212
- TextStores for revisions and signatures.
1216
super(RepositoryFormat6, self).__init__()
1217
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1219
def get_format_description(self):
1220
"""See RepositoryFormat.get_format_description()."""
1221
return "Weave repository format 6"
1223
def _get_revision_store(self, repo_transport, control_files):
1224
"""See RepositoryFormat._get_revision_store()."""
1225
return self._get_text_rev_store(repo_transport,
1231
def _get_text_store(self, transport, control_files):
1232
"""See RepositoryFormat._get_text_store()."""
1233
return self._get_versioned_file_store('weaves', transport, control_files)
1236
class MetaDirRepositoryFormat(RepositoryFormat):
1237
"""Common base class for the new repositories using the metadir layout."""
1240
super(MetaDirRepositoryFormat, self).__init__()
1241
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1243
def _create_control_files(self, a_bzrdir):
1244
"""Create the required files and the initial control_files object."""
1245
# FIXME: RBC 20060125 don't peek under the covers
1246
# NB: no need to escape relative paths that are url safe.
1247
repository_transport = a_bzrdir.get_repository_transport(self)
1248
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1249
control_files.create_lock()
1250
return control_files
1252
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1253
"""Upload the initial blank content."""
1254
control_files = self._create_control_files(a_bzrdir)
1255
control_files.lock_write()
1257
control_files._transport.mkdir_multi(dirs,
1258
mode=control_files._dir_mode)
1259
for file, content in files:
1260
control_files.put(file, content)
1261
for file, content in utf8_files:
1262
control_files.put_utf8(file, content)
1264
control_files.put_utf8('shared-storage', '')
1266
control_files.unlock()
1269
class RepositoryFormat7(MetaDirRepositoryFormat):
1270
"""Bzr repository 7.
1272
This repository format has:
1273
- weaves for file texts and inventory
1274
- hash subdirectory based stores.
1275
- TextStores for revisions and signatures.
1276
- a format marker of its own
1277
- an optional 'shared-storage' flag
1278
- an optional 'no-working-trees' flag
1281
def _get_control_store(self, repo_transport, control_files):
1282
"""Return the control store for this repository."""
1283
return self._get_versioned_file_store('',
1288
def get_format_string(self):
1289
"""See RepositoryFormat.get_format_string()."""
1290
return "Bazaar-NG Repository format 7"
1292
def get_format_description(self):
1293
"""See RepositoryFormat.get_format_description()."""
1294
return "Weave repository format 7"
1296
def _get_revision_store(self, repo_transport, control_files):
1297
"""See RepositoryFormat._get_revision_store()."""
1298
return self._get_text_rev_store(repo_transport,
1305
def _get_text_store(self, transport, control_files):
1306
"""See RepositoryFormat._get_text_store()."""
1307
return self._get_versioned_file_store('weaves',
1311
def initialize(self, a_bzrdir, shared=False):
1312
"""Create a weave repository.
1314
:param shared: If true the repository will be initialized as a shared
1317
from bzrlib.weavefile import write_weave_v5
1318
from bzrlib.weave import Weave
1320
# Create an empty weave
1322
write_weave_v5(Weave(), sio)
1323
empty_weave = sio.getvalue()
1325
mutter('creating repository in %s.', a_bzrdir.transport.base)
1326
dirs = ['revision-store', 'weaves']
1327
files = [('inventory.weave', StringIO(empty_weave)),
1329
utf8_files = [('format', self.get_format_string())]
1331
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1332
return self.open(a_bzrdir=a_bzrdir, _found=True)
1334
def open(self, a_bzrdir, _found=False, _override_transport=None):
1335
"""See RepositoryFormat.open().
1337
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1338
repository at a slightly different url
1339
than normal. I.e. during 'upgrade'.
1342
format = RepositoryFormat.find_format(a_bzrdir)
1343
assert format.__class__ == self.__class__
1344
if _override_transport is not None:
1345
repo_transport = _override_transport
1347
repo_transport = a_bzrdir.get_repository_transport(None)
1348
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1349
text_store = self._get_text_store(repo_transport, control_files)
1350
control_store = self._get_control_store(repo_transport, control_files)
1351
_revision_store = self._get_revision_store(repo_transport, control_files)
1352
return MetaDirRepository(_format=self,
1354
control_files=control_files,
1355
_revision_store=_revision_store,
1356
control_store=control_store,
1357
text_store=text_store)
1360
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1361
"""Bzr repository knit format 1.
1363
This repository format has:
1364
- knits for file texts and inventory
1365
- hash subdirectory based stores.
1366
- knits for revisions and signatures
1367
- TextStores for revisions and signatures.
1368
- a format marker of its own
1369
- an optional 'shared-storage' flag
1370
- an optional 'no-working-trees' flag
1373
This format was introduced in bzr 0.8.
1376
def _get_control_store(self, repo_transport, control_files):
1377
"""Return the control store for this repository."""
1378
return VersionedFileStore(
1381
file_mode=control_files._file_mode,
1382
versionedfile_class=KnitVersionedFile,
1383
versionedfile_kwargs={'factory':KnitPlainFactory()},
1386
def get_format_string(self):
1387
"""See RepositoryFormat.get_format_string()."""
1388
return "Bazaar-NG Knit Repository Format 1"
1390
def get_format_description(self):
1391
"""See RepositoryFormat.get_format_description()."""
1392
return "Knit repository format 1"
1394
def _get_revision_store(self, repo_transport, control_files):
1395
"""See RepositoryFormat._get_revision_store()."""
1396
from bzrlib.store.revision.knit import KnitRevisionStore
1397
versioned_file_store = VersionedFileStore(
1399
file_mode=control_files._file_mode,
1402
versionedfile_class=KnitVersionedFile,
1403
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1406
return KnitRevisionStore(versioned_file_store)
1408
def _get_text_store(self, transport, control_files):
1409
"""See RepositoryFormat._get_text_store()."""
1410
return self._get_versioned_file_store('knits',
1413
versionedfile_class=KnitVersionedFile,
1416
def initialize(self, a_bzrdir, shared=False):
1417
"""Create a knit format 1 repository.
1419
:param a_bzrdir: bzrdir to contain the new repository; must already
1421
:param shared: If true the repository will be initialized as a shared
1424
mutter('creating repository in %s.', a_bzrdir.transport.base)
1425
dirs = ['revision-store', 'knits']
1427
utf8_files = [('format', self.get_format_string())]
1429
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1430
repo_transport = a_bzrdir.get_repository_transport(None)
1431
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1432
control_store = self._get_control_store(repo_transport, control_files)
1433
transaction = transactions.WriteTransaction()
1434
# trigger a write of the inventory store.
1435
control_store.get_weave_or_empty('inventory', transaction)
1436
_revision_store = self._get_revision_store(repo_transport, control_files)
1437
_revision_store.has_revision_id('A', transaction)
1438
_revision_store.get_signature_file(transaction)
1439
return self.open(a_bzrdir=a_bzrdir, _found=True)
1441
def open(self, a_bzrdir, _found=False, _override_transport=None):
1442
"""See RepositoryFormat.open().
1444
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1445
repository at a slightly different url
1446
than normal. I.e. during 'upgrade'.
1449
format = RepositoryFormat.find_format(a_bzrdir)
1450
assert format.__class__ == self.__class__
1451
if _override_transport is not None:
1452
repo_transport = _override_transport
1454
repo_transport = a_bzrdir.get_repository_transport(None)
1455
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1456
text_store = self._get_text_store(repo_transport, control_files)
1457
control_store = self._get_control_store(repo_transport, control_files)
1458
_revision_store = self._get_revision_store(repo_transport, control_files)
1459
return KnitRepository(_format=self,
1461
control_files=control_files,
1462
_revision_store=_revision_store,
1463
control_store=control_store,
1464
text_store=text_store)
1467
# formats which have no format string are not discoverable
1468
# and not independently creatable, so are not registered.
1469
RepositoryFormat.register_format(RepositoryFormat7())
1470
_default_format = RepositoryFormatKnit1()
1471
RepositoryFormat.register_format(_default_format)
1472
RepositoryFormat.set_default_format(_default_format)
1473
_legacy_formats = [RepositoryFormat4(),
1474
RepositoryFormat5(),
1475
RepositoryFormat6()]
1478
class InterRepository(InterObject):
1479
"""This class represents operations taking place between two repositories.
1481
Its instances have methods like copy_content and fetch, and contain
1482
references to the source and target repositories these operations can be
1485
Often we will provide convenience methods on 'repository' which carry out
1486
operations with another repository - they will always forward to
1487
InterRepository.get(other).method_name(parameters).
1491
"""The available optimised InterRepository types."""
1494
def copy_content(self, revision_id=None, basis=None):
1495
"""Make a complete copy of the content in self into destination.
1497
This is a destructive operation! Do not use it on existing
1500
:param revision_id: Only copy the content needed to construct
1501
revision_id and its parents.
1502
:param basis: Copy the needed data preferentially from basis.
1505
self.target.set_make_working_trees(self.source.make_working_trees())
1506
except NotImplementedError:
1508
# grab the basis available data
1509
if basis is not None:
1510
self.target.fetch(basis, revision_id=revision_id)
1511
# but don't bother fetching if we have the needed data now.
1512
if (revision_id not in (None, NULL_REVISION) and
1513
self.target.has_revision(revision_id)):
1515
self.target.fetch(self.source, revision_id=revision_id)
1517
def _double_lock(self, lock_source, lock_target):
1518
"""Take out too locks, rolling back the first if the second throws."""
1523
# we want to ensure that we don't leave source locked by mistake.
1524
# and any error on target should not confuse source.
1525
self.source.unlock()
1529
def fetch(self, revision_id=None, pb=None):
1530
"""Fetch the content required to construct revision_id.
1532
The content is copied from source to target.
1534
:param revision_id: if None all content is copied, if NULL_REVISION no
1536
:param pb: optional progress bar to use for progress reports. If not
1537
provided a default one will be created.
1539
Returns the copied revision count and the failed revisions in a tuple:
1542
from bzrlib.fetch import GenericRepoFetcher
1543
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1544
self.source, self.source._format, self.target, self.target._format)
1545
f = GenericRepoFetcher(to_repository=self.target,
1546
from_repository=self.source,
1547
last_revision=revision_id,
1549
return f.count_copied, f.failed_revisions
1551
def lock_read(self):
1552
"""Take out a logical read lock.
1554
This will lock the source branch and the target branch. The source gets
1555
a read lock and the target a read lock.
1557
self._double_lock(self.source.lock_read, self.target.lock_read)
1559
def lock_write(self):
1560
"""Take out a logical write lock.
1562
This will lock the source branch and the target branch. The source gets
1563
a read lock and the target a write lock.
1565
self._double_lock(self.source.lock_read, self.target.lock_write)
1568
def missing_revision_ids(self, revision_id=None):
1569
"""Return the revision ids that source has that target does not.
1571
These are returned in topological order.
1573
:param revision_id: only return revision ids included by this
1576
# generic, possibly worst case, slow code path.
1577
target_ids = set(self.target.all_revision_ids())
1578
if revision_id is not None:
1579
source_ids = self.source.get_ancestry(revision_id)
1580
assert source_ids[0] == None
1583
source_ids = self.source.all_revision_ids()
1584
result_set = set(source_ids).difference(target_ids)
1585
# this may look like a no-op: its not. It preserves the ordering
1586
# other_ids had while only returning the members from other_ids
1587
# that we've decided we need.
1588
return [rev_id for rev_id in source_ids if rev_id in result_set]
1591
"""Release the locks on source and target."""
1593
self.target.unlock()
1595
self.source.unlock()
1598
class InterWeaveRepo(InterRepository):
1599
"""Optimised code paths between Weave based repositories."""
1601
_matching_repo_format = RepositoryFormat7()
1602
"""Repository format for testing with."""
1605
def is_compatible(source, target):
1606
"""Be compatible with known Weave formats.
1608
We don't test for the stores being of specific types because that
1609
could lead to confusing results, and there is no need to be
1613
return (isinstance(source._format, (RepositoryFormat5,
1615
RepositoryFormat7)) and
1616
isinstance(target._format, (RepositoryFormat5,
1618
RepositoryFormat7)))
1619
except AttributeError:
1623
def copy_content(self, revision_id=None, basis=None):
1624
"""See InterRepository.copy_content()."""
1625
# weave specific optimised path:
1626
if basis is not None:
1627
# copy the basis in, then fetch remaining data.
1628
basis.copy_content_into(self.target, revision_id)
1629
# the basis copy_content_into could miss-set this.
1631
self.target.set_make_working_trees(self.source.make_working_trees())
1632
except NotImplementedError:
1634
self.target.fetch(self.source, revision_id=revision_id)
1637
self.target.set_make_working_trees(self.source.make_working_trees())
1638
except NotImplementedError:
1640
# FIXME do not peek!
1641
if self.source.control_files._transport.listable():
1642
pb = ui.ui_factory.nested_progress_bar()
1644
self.target.weave_store.copy_all_ids(
1645
self.source.weave_store,
1647
from_transaction=self.source.get_transaction(),
1648
to_transaction=self.target.get_transaction())
1649
pb.update('copying inventory', 0, 1)
1650
self.target.control_weaves.copy_multi(
1651
self.source.control_weaves, ['inventory'],
1652
from_transaction=self.source.get_transaction(),
1653
to_transaction=self.target.get_transaction())
1654
self.target._revision_store.text_store.copy_all_ids(
1655
self.source._revision_store.text_store,
1660
self.target.fetch(self.source, revision_id=revision_id)
1663
def fetch(self, revision_id=None, pb=None):
1664
"""See InterRepository.fetch()."""
1665
from bzrlib.fetch import GenericRepoFetcher
1666
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1667
self.source, self.source._format, self.target, self.target._format)
1668
f = GenericRepoFetcher(to_repository=self.target,
1669
from_repository=self.source,
1670
last_revision=revision_id,
1672
return f.count_copied, f.failed_revisions
1675
def missing_revision_ids(self, revision_id=None):
1676
"""See InterRepository.missing_revision_ids()."""
1677
# we want all revisions to satisfy revision_id in source.
1678
# but we don't want to stat every file here and there.
1679
# we want then, all revisions other needs to satisfy revision_id
1680
# checked, but not those that we have locally.
1681
# so the first thing is to get a subset of the revisions to
1682
# satisfy revision_id in source, and then eliminate those that
1683
# we do already have.
1684
# this is slow on high latency connection to self, but as as this
1685
# disk format scales terribly for push anyway due to rewriting
1686
# inventory.weave, this is considered acceptable.
1688
if revision_id is not None:
1689
source_ids = self.source.get_ancestry(revision_id)
1690
assert source_ids[0] == None
1693
source_ids = self.source._all_possible_ids()
1694
source_ids_set = set(source_ids)
1695
# source_ids is the worst possible case we may need to pull.
1696
# now we want to filter source_ids against what we actually
1697
# have in target, but don't try to check for existence where we know
1698
# we do not have a revision as that would be pointless.
1699
target_ids = set(self.target._all_possible_ids())
1700
possibly_present_revisions = target_ids.intersection(source_ids_set)
1701
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1702
required_revisions = source_ids_set.difference(actually_present_revisions)
1703
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1704
if revision_id is not None:
1705
# we used get_ancestry to determine source_ids then we are assured all
1706
# revisions referenced are present as they are installed in topological order.
1707
# and the tip revision was validated by get_ancestry.
1708
return required_topo_revisions
1710
# if we just grabbed the possibly available ids, then
1711
# we only have an estimate of whats available and need to validate
1712
# that against the revision records.
1713
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1716
class InterKnitRepo(InterRepository):
1717
"""Optimised code paths between Knit based repositories."""
1719
_matching_repo_format = RepositoryFormatKnit1()
1720
"""Repository format for testing with."""
1723
def is_compatible(source, target):
1724
"""Be compatible with known Knit formats.
1726
We don't test for the stores being of specific types because that
1727
could lead to confusing results, and there is no need to be
1731
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1732
isinstance(target._format, (RepositoryFormatKnit1)))
1733
except AttributeError:
1737
def fetch(self, revision_id=None, pb=None):
1738
"""See InterRepository.fetch()."""
1739
from bzrlib.fetch import KnitRepoFetcher
1740
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1741
self.source, self.source._format, self.target, self.target._format)
1742
f = KnitRepoFetcher(to_repository=self.target,
1743
from_repository=self.source,
1744
last_revision=revision_id,
1746
return f.count_copied, f.failed_revisions
1749
def missing_revision_ids(self, revision_id=None):
1750
"""See InterRepository.missing_revision_ids()."""
1751
if revision_id is not None:
1752
source_ids = self.source.get_ancestry(revision_id)
1753
assert source_ids[0] == None
1756
source_ids = self.source._all_possible_ids()
1757
source_ids_set = set(source_ids)
1758
# source_ids is the worst possible case we may need to pull.
1759
# now we want to filter source_ids against what we actually
1760
# have in target, but don't try to check for existence where we know
1761
# we do not have a revision as that would be pointless.
1762
target_ids = set(self.target._all_possible_ids())
1763
possibly_present_revisions = target_ids.intersection(source_ids_set)
1764
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1765
required_revisions = source_ids_set.difference(actually_present_revisions)
1766
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1767
if revision_id is not None:
1768
# we used get_ancestry to determine source_ids then we are assured all
1769
# revisions referenced are present as they are installed in topological order.
1770
# and the tip revision was validated by get_ancestry.
1771
return required_topo_revisions
1773
# if we just grabbed the possibly available ids, then
1774
# we only have an estimate of whats available and need to validate
1775
# that against the revision records.
1776
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1778
InterRepository.register_optimiser(InterWeaveRepo)
1779
InterRepository.register_optimiser(InterKnitRepo)
1782
class RepositoryTestProviderAdapter(object):
1783
"""A tool to generate a suite testing multiple repository formats at once.
1785
This is done by copying the test once for each transport and injecting
1786
the transport_server, transport_readonly_server, and bzrdir_format and
1787
repository_format classes into each copy. Each copy is also given a new id()
1788
to make it easy to identify.
1791
def __init__(self, transport_server, transport_readonly_server, formats):
1792
self._transport_server = transport_server
1793
self._transport_readonly_server = transport_readonly_server
1794
self._formats = formats
1796
def adapt(self, test):
1797
result = TestSuite()
1798
for repository_format, bzrdir_format in self._formats:
1799
new_test = deepcopy(test)
1800
new_test.transport_server = self._transport_server
1801
new_test.transport_readonly_server = self._transport_readonly_server
1802
new_test.bzrdir_format = bzrdir_format
1803
new_test.repository_format = repository_format
1804
def make_new_test_id():
1805
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1806
return lambda: new_id
1807
new_test.id = make_new_test_id()
1808
result.addTest(new_test)
1812
class InterRepositoryTestProviderAdapter(object):
1813
"""A tool to generate a suite testing multiple inter repository formats.
1815
This is done by copying the test once for each interrepo provider and injecting
1816
the transport_server, transport_readonly_server, repository_format and
1817
repository_to_format classes into each copy.
1818
Each copy is also given a new id() to make it easy to identify.
1821
def __init__(self, transport_server, transport_readonly_server, formats):
1822
self._transport_server = transport_server
1823
self._transport_readonly_server = transport_readonly_server
1824
self._formats = formats
1826
def adapt(self, test):
1827
result = TestSuite()
1828
for interrepo_class, repository_format, repository_format_to in self._formats:
1829
new_test = deepcopy(test)
1830
new_test.transport_server = self._transport_server
1831
new_test.transport_readonly_server = self._transport_readonly_server
1832
new_test.interrepo_class = interrepo_class
1833
new_test.repository_format = repository_format
1834
new_test.repository_format_to = repository_format_to
1835
def make_new_test_id():
1836
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1837
return lambda: new_id
1838
new_test.id = make_new_test_id()
1839
result.addTest(new_test)
1843
def default_test_list():
1844
"""Generate the default list of interrepo permutations to test."""
1846
# test the default InterRepository between format 6 and the current
1848
# XXX: robertc 20060220 reinstate this when there are two supported
1849
# formats which do not have an optimal code path between them.
1850
result.append((InterRepository,
1851
RepositoryFormat6(),
1852
RepositoryFormatKnit1()))
1853
for optimiser in InterRepository._optimisers:
1854
result.append((optimiser,
1855
optimiser._matching_repo_format,
1856
optimiser._matching_repo_format
1858
# if there are specific combinations we want to use, we can add them
1863
class CopyConverter(object):
1864
"""A repository conversion tool which just performs a copy of the content.
1866
This is slow but quite reliable.
1869
def __init__(self, target_format):
1870
"""Create a CopyConverter.
1872
:param target_format: The format the resulting repository should be.
1874
self.target_format = target_format
1876
def convert(self, repo, pb):
1877
"""Perform the conversion of to_convert, giving feedback via pb.
1879
:param to_convert: The disk object to convert.
1880
:param pb: a progress bar to use for progress information.
1885
# this is only useful with metadir layouts - separated repo content.
1886
# trigger an assertion if not such
1887
repo._format.get_format_string()
1888
self.repo_dir = repo.bzrdir
1889
self.step('Moving repository to repository.backup')
1890
self.repo_dir.transport.move('repository', 'repository.backup')
1891
backup_transport = self.repo_dir.transport.clone('repository.backup')
1892
self.source_repo = repo._format.open(self.repo_dir,
1894
_override_transport=backup_transport)
1895
self.step('Creating new repository')
1896
converted = self.target_format.initialize(self.repo_dir,
1897
self.source_repo.is_shared())
1898
converted.lock_write()
1900
self.step('Copying content into repository.')
1901
self.source_repo.copy_content_into(converted)
1904
self.step('Deleting old repository content.')
1905
self.repo_dir.transport.delete_tree('repository.backup')
1906
self.pb.note('repository converted')
1908
def step(self, message):
1909
"""Update the pb by a step."""
1911
self.pb.update(message, self.count, self.total)
1914
class CommitBuilder(object):
1915
"""Provides an interface to build up a commit.
1917
This allows describing a tree to be committed without needing to
1918
know the internals of the format of the repository.
1920
def __init__(self, repository, parents, config, timestamp=None,
1921
timezone=None, committer=None, revprops=None,
1923
"""Initiate a CommitBuilder.
1925
:param repository: Repository to commit to.
1926
:param parents: Revision ids of the parents of the new revision.
1927
:param config: Configuration to use.
1928
:param timestamp: Optional timestamp recorded for commit.
1929
:param timezone: Optional timezone for timestamp.
1930
:param committer: Optional committer to set for commit.
1931
:param revprops: Optional dictionary of revision properties.
1932
:param revision_id: Optional revision id.
1934
self._config = config
1936
if committer is None:
1937
self._committer = self._config.username()
1939
assert isinstance(committer, basestring), type(committer)
1940
self._committer = committer
1942
self.new_inventory = Inventory()
1943
self._new_revision_id = revision_id
1944
self.parents = parents
1945
self.repository = repository
1948
if revprops is not None:
1949
self._revprops.update(revprops)
1951
if timestamp is None:
1952
self._timestamp = time.time()
1954
self._timestamp = long(timestamp)
1956
if timezone is None:
1957
self._timezone = local_time_offset()
1959
self._timezone = int(timezone)
1961
self._generate_revision_if_needed()
1963
def commit(self, message):
1964
"""Make the actual commit.
1966
:return: The revision id of the recorded revision.
1968
rev = Revision(timestamp=self._timestamp,
1969
timezone=self._timezone,
1970
committer=self._committer,
1972
inventory_sha1=self.inv_sha1,
1973
revision_id=self._new_revision_id,
1974
properties=self._revprops)
1975
rev.parent_ids = self.parents
1976
self.repository.add_revision(self._new_revision_id, rev,
1977
self.new_inventory, self._config)
1978
return self._new_revision_id
1980
def finish_inventory(self):
1981
"""Tell the builder that the inventory is finished."""
1982
self.new_inventory.revision_id = self._new_revision_id
1983
self.inv_sha1 = self.repository.add_inventory(
1984
self._new_revision_id,
1989
def _gen_revision_id(self):
1990
"""Return new revision-id."""
1991
s = '%s-%s-' % (self._config.user_email(),
1992
compact_date(self._timestamp))
1993
s += hexlify(rand_bytes(8))
1996
def _generate_revision_if_needed(self):
1997
"""Create a revision id if None was supplied.
1999
If the repository can not support user-specified revision ids
2000
they should override this function and raise UnsupportedOperation
2001
if _new_revision_id is not None.
2003
:raises: UnsupportedOperation
2005
if self._new_revision_id is None:
2006
self._new_revision_id = self._gen_revision_id()
2008
def record_entry_contents(self, ie, parent_invs, path, tree):
2009
"""Record the content of ie from tree into the commit if needed.
2011
:param ie: An inventory entry present in the commit.
2012
:param parent_invs: The inventories of the parent revisions of the
2014
:param path: The path the entry is at in the tree.
2015
:param tree: The tree which contains this entry and should be used to
2018
self.new_inventory.add(ie)
2020
# ie.revision is always None if the InventoryEntry is considered
2021
# for committing. ie.snapshot will record the correct revision
2022
# which may be the sole parent if it is untouched.
2023
if ie.revision is not None:
2025
previous_entries = ie.find_previous_heads(
2027
self.repository.weave_store,
2028
self.repository.get_transaction())
2029
# we are creating a new revision for ie in the history store
2031
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2033
def modified_directory(self, file_id, file_parents):
2034
"""Record the presence of a symbolic link.
2036
:param file_id: The file_id of the link to record.
2037
:param file_parents: The per-file parent revision ids.
2039
self._add_text_to_weave(file_id, [], file_parents.keys())
2041
def modified_file_text(self, file_id, file_parents,
2042
get_content_byte_lines, text_sha1=None,
2044
"""Record the text of file file_id
2046
:param file_id: The file_id of the file to record the text of.
2047
:param file_parents: The per-file parent revision ids.
2048
:param get_content_byte_lines: A callable which will return the byte
2050
:param text_sha1: Optional SHA1 of the file contents.
2051
:param text_size: Optional size of the file contents.
2053
mutter('storing text of file {%s} in revision {%s} into %r',
2054
file_id, self._new_revision_id, self.repository.weave_store)
2055
# special case to avoid diffing on renames or
2057
if (len(file_parents) == 1
2058
and text_sha1 == file_parents.values()[0].text_sha1
2059
and text_size == file_parents.values()[0].text_size):
2060
previous_ie = file_parents.values()[0]
2061
versionedfile = self.repository.weave_store.get_weave(file_id,
2062
self.repository.get_transaction())
2063
versionedfile.clone_text(self._new_revision_id,
2064
previous_ie.revision, file_parents.keys())
2065
return text_sha1, text_size
2067
new_lines = get_content_byte_lines()
2068
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2069
# should return the SHA1 and size
2070
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2071
return osutils.sha_strings(new_lines), \
2072
sum(map(len, new_lines))
2074
def modified_link(self, file_id, file_parents, link_target):
2075
"""Record the presence of a symbolic link.
2077
:param file_id: The file_id of the link to record.
2078
:param file_parents: The per-file parent revision ids.
2079
:param link_target: Target location of this link.
2081
self._add_text_to_weave(file_id, [], file_parents.keys())
2083
def _add_text_to_weave(self, file_id, new_lines, parents):
2084
versionedfile = self.repository.weave_store.get_weave_or_empty(
2085
file_id, self.repository.get_transaction())
2086
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2087
versionedfile.clear_cache()
2090
# Copied from xml.sax.saxutils
2091
def _unescape_xml(data):
2092
"""Unescape &, <, and > in a string of data.
2094
data = data.replace("<", "<")
2095
data = data.replace(">", ">")
2096
# must do ampersand last
2097
return data.replace("&", "&")