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, delta, 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.testament import Testament
42
from bzrlib.trace import mutter, note
43
from bzrlib.tree import RevisionTree
44
from bzrlib.tsort import topo_sort
45
from bzrlib.weave import WeaveFile
48
class Repository(object):
49
"""Repository holding history for one or more branches.
51
The repository holds and retrieves historical information including
52
revisions and file history. It's normally accessed only by the Branch,
53
which views a particular line of development through that history.
55
The Repository builds on top of Stores and a Transport, which respectively
56
describe the disk data format and the way of accessing the (possibly
61
def add_inventory(self, revid, inv, parents):
62
"""Add the inventory inv to the repository as revid.
64
:param parents: The revision ids of the parents that revid
65
is known to have and are in the repository already.
67
returns the sha1 of the serialized inventory.
69
assert inv.revision_id is None or inv.revision_id == revid, \
70
"Mismatch between inventory revision" \
71
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
72
inv_text = xml5.serializer_v5.write_inventory_to_string(inv)
73
inv_sha1 = osutils.sha_string(inv_text)
74
inv_vf = self.control_weaves.get_weave('inventory',
75
self.get_transaction())
76
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
79
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
81
for parent in parents:
83
final_parents.append(parent)
85
inv_vf.add_lines(revid, final_parents, lines)
88
def add_revision(self, rev_id, rev, inv=None, config=None):
89
"""Add rev to the revision store as rev_id.
91
:param rev_id: the revision id to use.
92
:param rev: The revision object.
93
:param inv: The inventory for the revision. if None, it will be looked
94
up in the inventory storer
95
:param config: If None no digital signature will be created.
96
If supplied its signature_needed method will be used
97
to determine if a signature should be made.
99
if config is not None and config.signature_needed():
101
inv = self.get_inventory(rev_id)
102
plaintext = Testament(rev, inv).as_short_text()
103
self.store_revision_signature(
104
gpg.GPGStrategy(config), plaintext, rev_id)
105
if not rev_id in self.get_inventory_weave():
107
raise errors.WeaveRevisionNotPresent(rev_id,
108
self.get_inventory_weave())
110
# yes, this is not suitable for adding with ghosts.
111
self.add_inventory(rev_id, inv, rev.parent_ids)
112
self._revision_store.add_revision(rev, self.get_transaction())
115
def _all_possible_ids(self):
116
"""Return all the possible revisions that we could find."""
117
return self.get_inventory_weave().versions()
119
def all_revision_ids(self):
120
"""Returns a list of all the revision ids in the repository.
122
This is deprecated because code should generally work on the graph
123
reachable from a particular revision, and ignore any other revisions
124
that might be present. There is no direct replacement method.
126
return self._all_revision_ids()
129
def _all_revision_ids(self):
130
"""Returns a list of all the revision ids in the repository.
132
These are in as much topological order as the underlying store can
133
present: for weaves ghosts may lead to a lack of correctness until
134
the reweave updates the parents list.
136
if self._revision_store.text_store.listable():
137
return self._revision_store.all_revision_ids(self.get_transaction())
138
result = self._all_possible_ids()
139
return self._eliminate_revisions_not_present(result)
141
def break_lock(self):
142
"""Break a lock if one is present from another instance.
144
Uses the ui factory to ask for confirmation if the lock may be from
147
self.control_files.break_lock()
150
def _eliminate_revisions_not_present(self, revision_ids):
151
"""Check every revision id in revision_ids to see if we have it.
153
Returns a set of the present revisions.
156
for id in revision_ids:
157
if self.has_revision(id):
162
def create(a_bzrdir):
163
"""Construct the current default format repository in a_bzrdir."""
164
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
166
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
167
"""instantiate a Repository.
169
:param _format: The format of the repository on disk.
170
:param a_bzrdir: The BzrDir of the repository.
172
In the future we will have a single api for all stores for
173
getting file texts, inventories and revisions, then
174
this construct will accept instances of those things.
176
super(Repository, self).__init__()
177
self._format = _format
178
# the following are part of the public API for Repository:
179
self.bzrdir = a_bzrdir
180
self.control_files = control_files
181
self._revision_store = _revision_store
182
self.text_store = text_store
183
# backwards compatibility
184
self.weave_store = text_store
185
# not right yet - should be more semantically clear ?
187
self.control_store = control_store
188
self.control_weaves = control_store
189
# TODO: make sure to construct the right store classes, etc, depending
190
# on whether escaping is required.
193
return '%s(%r)' % (self.__class__.__name__,
194
self.bzrdir.transport.base)
197
return self.control_files.is_locked()
199
def lock_write(self):
200
self.control_files.lock_write()
203
self.control_files.lock_read()
205
def get_physical_lock_status(self):
206
return self.control_files.get_physical_lock_status()
209
def missing_revision_ids(self, other, revision_id=None):
210
"""Return the revision ids that other has that this does not.
212
These are returned in topological order.
214
revision_id: only return revision ids included by revision_id.
216
return InterRepository.get(other, self).missing_revision_ids(revision_id)
220
"""Open the repository rooted at base.
222
For instance, if the repository is at URL/.bzr/repository,
223
Repository.open(URL) -> a Repository instance.
225
control = bzrdir.BzrDir.open(base)
226
return control.open_repository()
228
def copy_content_into(self, destination, revision_id=None, basis=None):
229
"""Make a complete copy of the content in self into destination.
231
This is a destructive operation! Do not use it on existing
234
return InterRepository.get(self, destination).copy_content(revision_id, basis)
236
def fetch(self, source, revision_id=None, pb=None):
237
"""Fetch the content required to construct revision_id from source.
239
If revision_id is None all content is copied.
241
return InterRepository.get(source, self).fetch(revision_id=revision_id,
244
def get_commit_builder(self, branch, parents, config, timestamp=None,
245
timezone=None, committer=None, revprops=None,
247
"""Obtain a CommitBuilder for this repository.
249
:param branch: Branch to commit to.
250
:param parents: Revision ids of the parents of the new revision.
251
:param config: Configuration to use.
252
:param timestamp: Optional timestamp recorded for commit.
253
:param timezone: Optional timezone for timestamp.
254
:param committer: Optional committer to set for commit.
255
:param revprops: Optional dictionary of revision properties.
256
:param revision_id: Optional revision id.
258
return CommitBuilder(self, parents, config, timestamp, timezone,
259
committer, revprops, revision_id)
262
self.control_files.unlock()
265
def clone(self, a_bzrdir, revision_id=None, basis=None):
266
"""Clone this repository into a_bzrdir using the current format.
268
Currently no check is made that the format of this repository and
269
the bzrdir format are compatible. FIXME RBC 20060201.
271
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
272
# use target default format.
273
result = a_bzrdir.create_repository()
274
# FIXME RBC 20060209 split out the repository type to avoid this check ?
275
elif isinstance(a_bzrdir._format,
276
(bzrdir.BzrDirFormat4,
277
bzrdir.BzrDirFormat5,
278
bzrdir.BzrDirFormat6)):
279
result = a_bzrdir.open_repository()
281
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
282
self.copy_content_into(result, revision_id, basis)
286
def has_revision(self, revision_id):
287
"""True if this repository has a copy of the revision."""
288
return self._revision_store.has_revision_id(revision_id,
289
self.get_transaction())
292
def get_revision_reconcile(self, revision_id):
293
"""'reconcile' helper routine that allows access to a revision always.
295
This variant of get_revision does not cross check the weave graph
296
against the revision one as get_revision does: but it should only
297
be used by reconcile, or reconcile-alike commands that are correcting
298
or testing the revision graph.
300
if not revision_id or not isinstance(revision_id, basestring):
301
raise InvalidRevisionId(revision_id=revision_id, branch=self)
302
return self._revision_store.get_revisions([revision_id],
303
self.get_transaction())[0]
305
def get_revisions(self, revision_ids):
306
return self._revision_store.get_revisions(revision_ids,
307
self.get_transaction())
310
def get_revision_xml(self, revision_id):
311
rev = self.get_revision(revision_id)
313
# the current serializer..
314
self._revision_store._serializer.write_revision(rev, rev_tmp)
316
return rev_tmp.getvalue()
319
def get_revision(self, revision_id):
320
"""Return the Revision object for a named revision"""
321
r = self.get_revision_reconcile(revision_id)
322
# weave corruption can lead to absent revision markers that should be
324
# the following test is reasonably cheap (it needs a single weave read)
325
# and the weave is cached in read transactions. In write transactions
326
# it is not cached but typically we only read a small number of
327
# revisions. For knits when they are introduced we will probably want
328
# to ensure that caching write transactions are in use.
329
inv = self.get_inventory_weave()
330
self._check_revision_parents(r, inv)
334
def get_deltas_for_revisions(self, revisions):
335
"""Produce a generator of revision deltas.
337
Note that the input is a sequence of REVISIONS, not revision_ids.
338
Trees will be held in memory until the generator exits.
339
Each delta is relative to the revision's lefthand predecessor.
341
required_trees = set()
342
for revision in revisions:
343
required_trees.add(revision.revision_id)
344
required_trees.update(revision.parent_ids[:1])
345
trees = dict((t.get_revision_id(), t) for
346
t in self.revision_trees(required_trees))
347
for revision in revisions:
348
if not revision.parent_ids:
349
old_tree = self.revision_tree(None)
351
old_tree = trees[revision.parent_ids[0]]
352
yield delta.compare_trees(old_tree, trees[revision.revision_id])
355
def get_revision_delta(self, revision_id):
356
"""Return the delta for one revision.
358
The delta is relative to the left-hand predecessor of the
361
r = self.get_revision(revision_id)
362
return list(self.get_deltas_for_revisions([r]))[0]
364
def _check_revision_parents(self, revision, inventory):
365
"""Private to Repository and Fetch.
367
This checks the parentage of revision in an inventory weave for
368
consistency and is only applicable to inventory-weave-for-ancestry
369
using repository formats & fetchers.
371
weave_parents = inventory.get_parents(revision.revision_id)
372
weave_names = inventory.versions()
373
for parent_id in revision.parent_ids:
374
if parent_id in weave_names:
375
# this parent must not be a ghost.
376
if not parent_id in weave_parents:
378
raise errors.CorruptRepository(self)
381
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
382
signature = gpg_strategy.sign(plaintext)
383
self._revision_store.add_revision_signature_text(revision_id,
385
self.get_transaction())
387
def fileids_altered_by_revision_ids(self, revision_ids):
388
"""Find the file ids and versions affected by revisions.
390
:param revisions: an iterable containing revision ids.
391
:return: a dictionary mapping altered file-ids to an iterable of
392
revision_ids. Each altered file-ids has the exact revision_ids that
393
altered it listed explicitly.
395
assert isinstance(self._format, (RepositoryFormat5,
398
RepositoryFormatKnit1)), \
399
("fileids_altered_by_revision_ids only supported for branches "
400
"which store inventory as unnested xml, not on %r" % self)
401
selected_revision_ids = set(revision_ids)
402
w = self.get_inventory_weave()
405
# this code needs to read every new line in every inventory for the
406
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
407
# not present in one of those inventories is unnecessary but not
408
# harmful because we are filtering by the revision id marker in the
409
# inventory lines : we only select file ids altered in one of those
410
# revisions. We don't need to see all lines in the inventory because
411
# only those added in an inventory in rev X can contain a revision=X
413
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
414
start = line.find('file_id="')+9
415
if start < 9: continue
416
end = line.find('"', start)
418
file_id = _unescape_xml(line[start:end])
420
start = line.find('revision="')+10
421
if start < 10: continue
422
end = line.find('"', start)
424
revision_id = _unescape_xml(line[start:end])
425
if revision_id in selected_revision_ids:
426
result.setdefault(file_id, set()).add(revision_id)
430
def get_inventory_weave(self):
431
return self.control_weaves.get_weave('inventory',
432
self.get_transaction())
435
def get_inventory(self, revision_id):
436
"""Get Inventory object by hash."""
437
return self.deserialise_inventory(
438
revision_id, self.get_inventory_xml(revision_id))
440
def deserialise_inventory(self, revision_id, xml):
441
"""Transform the xml into an inventory object.
443
:param revision_id: The expected revision id of the inventory.
444
:param xml: A serialised inventory.
446
return xml5.serializer_v5.read_inventory_from_string(xml)
449
def get_inventory_xml(self, revision_id):
450
"""Get inventory XML as a file object."""
452
assert isinstance(revision_id, basestring), type(revision_id)
453
iw = self.get_inventory_weave()
454
return iw.get_text(revision_id)
456
raise errors.HistoryMissing(self, 'inventory', revision_id)
459
def get_inventory_sha1(self, revision_id):
460
"""Return the sha1 hash of the inventory entry
462
return self.get_revision(revision_id).inventory_sha1
465
def get_revision_graph(self, revision_id=None):
466
"""Return a dictionary containing the revision graph.
468
:param revision_id: The revision_id to get a graph from. If None, then
469
the entire revision graph is returned. This is a deprecated mode of
470
operation and will be removed in the future.
471
:return: a dictionary of revision_id->revision_parents_list.
473
# special case NULL_REVISION
474
if revision_id == NULL_REVISION:
476
weave = self.get_inventory_weave()
477
all_revisions = self._eliminate_revisions_not_present(weave.versions())
478
entire_graph = dict([(node, weave.get_parents(node)) for
479
node in all_revisions])
480
if revision_id is None:
482
elif revision_id not in entire_graph:
483
raise errors.NoSuchRevision(self, revision_id)
485
# add what can be reached from revision_id
487
pending = set([revision_id])
488
while len(pending) > 0:
490
result[node] = entire_graph[node]
491
for revision_id in result[node]:
492
if revision_id not in result:
493
pending.add(revision_id)
497
def get_revision_graph_with_ghosts(self, revision_ids=None):
498
"""Return a graph of the revisions with ghosts marked as applicable.
500
:param revision_ids: an iterable of revisions to graph or None for all.
501
:return: a Graph object with the graph reachable from revision_ids.
505
pending = set(self.all_revision_ids())
508
pending = set(revision_ids)
509
# special case NULL_REVISION
510
if NULL_REVISION in pending:
511
pending.remove(NULL_REVISION)
512
required = set(pending)
515
revision_id = pending.pop()
517
rev = self.get_revision(revision_id)
518
except errors.NoSuchRevision:
519
if revision_id in required:
522
result.add_ghost(revision_id)
524
for parent_id in rev.parent_ids:
525
# is this queued or done ?
526
if (parent_id not in pending and
527
parent_id not in done):
529
pending.add(parent_id)
530
result.add_node(revision_id, rev.parent_ids)
531
done.add(revision_id)
535
def get_revision_inventory(self, revision_id):
536
"""Return inventory of a past revision."""
537
# TODO: Unify this with get_inventory()
538
# bzr 0.0.6 and later imposes the constraint that the inventory_id
539
# must be the same as its revision, so this is trivial.
540
if revision_id is None:
541
# This does not make sense: if there is no revision,
542
# then it is the current tree inventory surely ?!
543
# and thus get_root_id() is something that looks at the last
544
# commit on the branch, and the get_root_id is an inventory check.
545
raise NotImplementedError
546
# return Inventory(self.get_root_id())
548
return self.get_inventory(revision_id)
552
"""Return True if this repository is flagged as a shared repository."""
553
raise NotImplementedError(self.is_shared)
556
def reconcile(self, other=None, thorough=False):
557
"""Reconcile this repository."""
558
from bzrlib.reconcile import RepoReconciler
559
reconciler = RepoReconciler(self, thorough=thorough)
560
reconciler.reconcile()
564
def revision_tree(self, revision_id):
565
"""Return Tree for a revision on this branch.
567
`revision_id` may be None for the empty tree revision.
569
# TODO: refactor this to use an existing revision object
570
# so we don't need to read it in twice.
571
if revision_id is None or revision_id == NULL_REVISION:
572
return RevisionTree(self, Inventory(), NULL_REVISION)
574
inv = self.get_revision_inventory(revision_id)
575
return RevisionTree(self, inv, revision_id)
578
def revision_trees(self, revision_ids):
579
"""Return Tree for a revision on this branch.
581
`revision_id` may not be None or 'null:'"""
582
assert None not in revision_ids
583
assert NULL_REVISION not in revision_ids
584
texts = self.get_inventory_weave().get_texts(revision_ids)
585
for text, revision_id in zip(texts, revision_ids):
586
inv = self.deserialise_inventory(revision_id, text)
587
yield RevisionTree(self, inv, revision_id)
590
def get_ancestry(self, revision_id):
591
"""Return a list of revision-ids integrated by a revision.
593
The first element of the list is always None, indicating the origin
594
revision. This might change when we have history horizons, or
595
perhaps we should have a new API.
597
This is topologically sorted.
599
if revision_id is None:
601
if not self.has_revision(revision_id):
602
raise errors.NoSuchRevision(self, revision_id)
603
w = self.get_inventory_weave()
604
candidates = w.get_ancestry(revision_id)
605
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
608
def print_file(self, file, revision_id):
609
"""Print `file` to stdout.
611
FIXME RBC 20060125 as John Meinel points out this is a bad api
612
- it writes to stdout, it assumes that that is valid etc. Fix
613
by creating a new more flexible convenience function.
615
tree = self.revision_tree(revision_id)
616
# use inventory as it was in that revision
617
file_id = tree.inventory.path2id(file)
619
# TODO: jam 20060427 Write a test for this code path
620
# it had a bug in it, and was raising the wrong
622
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
623
tree.print_file(file_id)
625
def get_transaction(self):
626
return self.control_files.get_transaction()
628
def revision_parents(self, revid):
629
return self.get_inventory_weave().parent_names(revid)
632
def set_make_working_trees(self, new_value):
633
"""Set the policy flag for making working trees when creating branches.
635
This only applies to branches that use this repository.
637
The default is 'True'.
638
:param new_value: True to restore the default, False to disable making
641
raise NotImplementedError(self.set_make_working_trees)
643
def make_working_trees(self):
644
"""Returns the policy for making working trees on new branches."""
645
raise NotImplementedError(self.make_working_trees)
648
def sign_revision(self, revision_id, gpg_strategy):
649
plaintext = Testament.from_revision(self, revision_id).as_short_text()
650
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
653
def has_signature_for_revision_id(self, revision_id):
654
"""Query for a revision signature for revision_id in the repository."""
655
return self._revision_store.has_signature(revision_id,
656
self.get_transaction())
659
def get_signature_text(self, revision_id):
660
"""Return the text for a signature."""
661
return self._revision_store.get_signature_text(revision_id,
662
self.get_transaction())
665
def check(self, revision_ids):
666
"""Check consistency of all history of given revision_ids.
668
Different repository implementations should override _check().
670
:param revision_ids: A non-empty list of revision_ids whose ancestry
671
will be checked. Typically the last revision_id of a branch.
674
raise ValueError("revision_ids must be non-empty in %s.check"
676
return self._check(revision_ids)
678
def _check(self, revision_ids):
679
result = check.Check(self)
684
class AllInOneRepository(Repository):
685
"""Legacy support - the repository behaviour for all-in-one branches."""
687
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
688
# we reuse one control files instance.
689
dir_mode = a_bzrdir._control_files._dir_mode
690
file_mode = a_bzrdir._control_files._file_mode
692
def get_store(name, compressed=True, prefixed=False):
693
# FIXME: This approach of assuming stores are all entirely compressed
694
# or entirely uncompressed is tidy, but breaks upgrade from
695
# some existing branches where there's a mixture; we probably
696
# still want the option to look for both.
697
relpath = a_bzrdir._control_files._escape(name)
698
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
699
prefixed=prefixed, compressed=compressed,
702
#if self._transport.should_cache():
703
# cache_path = os.path.join(self.cache_root, name)
704
# os.mkdir(cache_path)
705
# store = bzrlib.store.CachedStore(store, cache_path)
708
# not broken out yet because the controlweaves|inventory_store
709
# and text_store | weave_store bits are still different.
710
if isinstance(_format, RepositoryFormat4):
711
# cannot remove these - there is still no consistent api
712
# which allows access to this old info.
713
self.inventory_store = get_store('inventory-store')
714
text_store = get_store('text-store')
715
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
719
"""AllInOne repositories cannot be shared."""
723
def set_make_working_trees(self, new_value):
724
"""Set the policy flag for making working trees when creating branches.
726
This only applies to branches that use this repository.
728
The default is 'True'.
729
:param new_value: True to restore the default, False to disable making
732
raise NotImplementedError(self.set_make_working_trees)
734
def make_working_trees(self):
735
"""Returns the policy for making working trees on new branches."""
739
def install_revision(repository, rev, revision_tree):
740
"""Install all revision data into a repository."""
743
for p_id in rev.parent_ids:
744
if repository.has_revision(p_id):
745
present_parents.append(p_id)
746
parent_trees[p_id] = repository.revision_tree(p_id)
748
parent_trees[p_id] = repository.revision_tree(None)
750
inv = revision_tree.inventory
752
# Add the texts that are not already present
753
for path, ie in inv.iter_entries():
754
w = repository.weave_store.get_weave_or_empty(ie.file_id,
755
repository.get_transaction())
756
if ie.revision not in w:
758
# FIXME: TODO: The following loop *may* be overlapping/duplicate
759
# with InventoryEntry.find_previous_heads(). if it is, then there
760
# is a latent bug here where the parents may have ancestors of each
762
for revision, tree in parent_trees.iteritems():
763
if ie.file_id not in tree:
765
parent_id = tree.inventory[ie.file_id].revision
766
if parent_id in text_parents:
768
text_parents.append(parent_id)
770
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
771
repository.get_transaction())
772
lines = revision_tree.get_file(ie.file_id).readlines()
773
vfile.add_lines(rev.revision_id, text_parents, lines)
775
# install the inventory
776
repository.add_inventory(rev.revision_id, inv, present_parents)
777
except errors.RevisionAlreadyPresent:
779
repository.add_revision(rev.revision_id, rev, inv)
782
class MetaDirRepository(Repository):
783
"""Repositories in the new meta-dir layout."""
785
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
786
super(MetaDirRepository, self).__init__(_format,
793
dir_mode = self.control_files._dir_mode
794
file_mode = self.control_files._file_mode
798
"""Return True if this repository is flagged as a shared repository."""
799
return self.control_files._transport.has('shared-storage')
802
def set_make_working_trees(self, new_value):
803
"""Set the policy flag for making working trees when creating branches.
805
This only applies to branches that use this repository.
807
The default is 'True'.
808
:param new_value: True to restore the default, False to disable making
813
self.control_files._transport.delete('no-working-trees')
814
except errors.NoSuchFile:
817
self.control_files.put_utf8('no-working-trees', '')
819
def make_working_trees(self):
820
"""Returns the policy for making working trees on new branches."""
821
return not self.control_files._transport.has('no-working-trees')
824
class KnitRepository(MetaDirRepository):
825
"""Knit format repository."""
827
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
828
inv_vf.add_lines_with_ghosts(revid, parents, lines)
831
def _all_revision_ids(self):
832
"""See Repository.all_revision_ids()."""
833
# Knits get the revision graph from the index of the revision knit, so
834
# it's always possible even if they're on an unlistable transport.
835
return self._revision_store.all_revision_ids(self.get_transaction())
837
def fileid_involved_between_revs(self, from_revid, to_revid):
838
"""Find file_id(s) which are involved in the changes between revisions.
840
This determines the set of revisions which are involved, and then
841
finds all file ids affected by those revisions.
843
vf = self._get_revision_vf()
844
from_set = set(vf.get_ancestry(from_revid))
845
to_set = set(vf.get_ancestry(to_revid))
846
changed = to_set.difference(from_set)
847
return self._fileid_involved_by_set(changed)
849
def fileid_involved(self, last_revid=None):
850
"""Find all file_ids modified in the ancestry of last_revid.
852
:param last_revid: If None, last_revision() will be used.
855
changed = set(self.all_revision_ids())
857
changed = set(self.get_ancestry(last_revid))
860
return self._fileid_involved_by_set(changed)
863
def get_ancestry(self, revision_id):
864
"""Return a list of revision-ids integrated by a revision.
866
This is topologically sorted.
868
if revision_id is None:
870
vf = self._get_revision_vf()
872
return [None] + vf.get_ancestry(revision_id)
873
except errors.RevisionNotPresent:
874
raise errors.NoSuchRevision(self, revision_id)
877
def get_revision(self, revision_id):
878
"""Return the Revision object for a named revision"""
879
return self.get_revision_reconcile(revision_id)
882
def get_revision_graph(self, revision_id=None):
883
"""Return a dictionary containing the revision graph.
885
:param revision_id: The revision_id to get a graph from. If None, then
886
the entire revision graph is returned. This is a deprecated mode of
887
operation and will be removed in the future.
888
:return: a dictionary of revision_id->revision_parents_list.
890
# special case NULL_REVISION
891
if revision_id == NULL_REVISION:
893
weave = self._get_revision_vf()
894
entire_graph = weave.get_graph()
895
if revision_id is None:
896
return weave.get_graph()
897
elif revision_id not in weave:
898
raise errors.NoSuchRevision(self, revision_id)
900
# add what can be reached from revision_id
902
pending = set([revision_id])
903
while len(pending) > 0:
905
result[node] = weave.get_parents(node)
906
for revision_id in result[node]:
907
if revision_id not in result:
908
pending.add(revision_id)
912
def get_revision_graph_with_ghosts(self, revision_ids=None):
913
"""Return a graph of the revisions with ghosts marked as applicable.
915
:param revision_ids: an iterable of revisions to graph or None for all.
916
:return: a Graph object with the graph reachable from revision_ids.
919
vf = self._get_revision_vf()
920
versions = set(vf.versions())
922
pending = set(self.all_revision_ids())
925
pending = set(revision_ids)
926
# special case NULL_REVISION
927
if NULL_REVISION in pending:
928
pending.remove(NULL_REVISION)
929
required = set(pending)
932
revision_id = pending.pop()
933
if not revision_id in versions:
934
if revision_id in required:
935
raise errors.NoSuchRevision(self, revision_id)
937
result.add_ghost(revision_id)
938
# mark it as done so we don't try for it again.
939
done.add(revision_id)
941
parent_ids = vf.get_parents_with_ghosts(revision_id)
942
for parent_id in parent_ids:
943
# is this queued or done ?
944
if (parent_id not in pending and
945
parent_id not in done):
947
pending.add(parent_id)
948
result.add_node(revision_id, parent_ids)
949
done.add(revision_id)
952
def _get_revision_vf(self):
953
""":return: a versioned file containing the revisions."""
954
vf = self._revision_store.get_revision_file(self.get_transaction())
958
def reconcile(self, other=None, thorough=False):
959
"""Reconcile this repository."""
960
from bzrlib.reconcile import KnitReconciler
961
reconciler = KnitReconciler(self, thorough=thorough)
962
reconciler.reconcile()
965
def revision_parents(self, revision_id):
966
return self._get_revision_vf().get_parents(revision_id)
969
class RepositoryFormat(object):
970
"""A repository format.
972
Formats provide three things:
973
* An initialization routine to construct repository data on disk.
974
* a format string which is used when the BzrDir supports versioned
976
* an open routine which returns a Repository instance.
978
Formats are placed in an dict by their format string for reference
979
during opening. These should be subclasses of RepositoryFormat
982
Once a format is deprecated, just deprecate the initialize and open
983
methods on the format class. Do not deprecate the object, as the
984
object will be created every system load.
986
Common instance attributes:
987
_matchingbzrdir - the bzrdir format that the repository format was
988
originally written to work with. This can be used if manually
989
constructing a bzrdir and repository, or more commonly for test suite
993
_default_format = None
994
"""The default format used for new repositories."""
997
"""The known formats."""
1000
def find_format(klass, a_bzrdir):
1001
"""Return the format for the repository object in a_bzrdir."""
1003
transport = a_bzrdir.get_repository_transport(None)
1004
format_string = transport.get("format").read()
1005
return klass._formats[format_string]
1006
except errors.NoSuchFile:
1007
raise errors.NoRepositoryPresent(a_bzrdir)
1009
raise errors.UnknownFormatError(format=format_string)
1011
def _get_control_store(self, repo_transport, control_files):
1012
"""Return the control store for this repository."""
1013
raise NotImplementedError(self._get_control_store)
1016
def get_default_format(klass):
1017
"""Return the current default format."""
1018
return klass._default_format
1020
def get_format_string(self):
1021
"""Return the ASCII format string that identifies this format.
1023
Note that in pre format ?? repositories the format string is
1024
not permitted nor written to disk.
1026
raise NotImplementedError(self.get_format_string)
1028
def get_format_description(self):
1029
"""Return the short description for this format."""
1030
raise NotImplementedError(self.get_format_description)
1032
def _get_revision_store(self, repo_transport, control_files):
1033
"""Return the revision store object for this a_bzrdir."""
1034
raise NotImplementedError(self._get_revision_store)
1036
def _get_text_rev_store(self,
1043
"""Common logic for getting a revision store for a repository.
1045
see self._get_revision_store for the subclass-overridable method to
1046
get the store for a repository.
1048
from bzrlib.store.revision.text import TextRevisionStore
1049
dir_mode = control_files._dir_mode
1050
file_mode = control_files._file_mode
1051
text_store =TextStore(transport.clone(name),
1053
compressed=compressed,
1055
file_mode=file_mode)
1056
_revision_store = TextRevisionStore(text_store, serializer)
1057
return _revision_store
1059
def _get_versioned_file_store(self,
1064
versionedfile_class=WeaveFile,
1066
weave_transport = control_files._transport.clone(name)
1067
dir_mode = control_files._dir_mode
1068
file_mode = control_files._file_mode
1069
return VersionedFileStore(weave_transport, prefixed=prefixed,
1071
file_mode=file_mode,
1072
versionedfile_class=versionedfile_class,
1075
def initialize(self, a_bzrdir, shared=False):
1076
"""Initialize a repository of this format in a_bzrdir.
1078
:param a_bzrdir: The bzrdir to put the new repository in it.
1079
:param shared: The repository should be initialized as a sharable one.
1081
This may raise UninitializableFormat if shared repository are not
1082
compatible the a_bzrdir.
1085
def is_supported(self):
1086
"""Is this format supported?
1088
Supported formats must be initializable and openable.
1089
Unsupported formats may not support initialization or committing or
1090
some other features depending on the reason for not being supported.
1094
def open(self, a_bzrdir, _found=False):
1095
"""Return an instance of this format for the bzrdir a_bzrdir.
1097
_found is a private parameter, do not use it.
1099
raise NotImplementedError(self.open)
1102
def register_format(klass, format):
1103
klass._formats[format.get_format_string()] = format
1106
def set_default_format(klass, format):
1107
klass._default_format = format
1110
def unregister_format(klass, format):
1111
assert klass._formats[format.get_format_string()] is format
1112
del klass._formats[format.get_format_string()]
1115
class PreSplitOutRepositoryFormat(RepositoryFormat):
1116
"""Base class for the pre split out repository formats."""
1118
def initialize(self, a_bzrdir, shared=False, _internal=False):
1119
"""Create a weave repository.
1121
TODO: when creating split out bzr branch formats, move this to a common
1122
base for Format5, Format6. or something like that.
1124
from bzrlib.weavefile import write_weave_v5
1125
from bzrlib.weave import Weave
1128
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1131
# always initialized when the bzrdir is.
1132
return self.open(a_bzrdir, _found=True)
1134
# Create an empty weave
1136
write_weave_v5(Weave(), sio)
1137
empty_weave = sio.getvalue()
1139
mutter('creating repository in %s.', a_bzrdir.transport.base)
1140
dirs = ['revision-store', 'weaves']
1141
files = [('inventory.weave', StringIO(empty_weave)),
1144
# FIXME: RBC 20060125 don't peek under the covers
1145
# NB: no need to escape relative paths that are url safe.
1146
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1148
control_files.create_lock()
1149
control_files.lock_write()
1150
control_files._transport.mkdir_multi(dirs,
1151
mode=control_files._dir_mode)
1153
for file, content in files:
1154
control_files.put(file, content)
1156
control_files.unlock()
1157
return self.open(a_bzrdir, _found=True)
1159
def _get_control_store(self, repo_transport, control_files):
1160
"""Return the control store for this repository."""
1161
return self._get_versioned_file_store('',
1166
def _get_text_store(self, transport, control_files):
1167
"""Get a store for file texts for this format."""
1168
raise NotImplementedError(self._get_text_store)
1170
def open(self, a_bzrdir, _found=False):
1171
"""See RepositoryFormat.open()."""
1173
# we are being called directly and must probe.
1174
raise NotImplementedError
1176
repo_transport = a_bzrdir.get_repository_transport(None)
1177
control_files = a_bzrdir._control_files
1178
text_store = self._get_text_store(repo_transport, control_files)
1179
control_store = self._get_control_store(repo_transport, control_files)
1180
_revision_store = self._get_revision_store(repo_transport, control_files)
1181
return AllInOneRepository(_format=self,
1183
_revision_store=_revision_store,
1184
control_store=control_store,
1185
text_store=text_store)
1188
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1189
"""Bzr repository format 4.
1191
This repository format has:
1193
- TextStores for texts, inventories,revisions.
1195
This format is deprecated: it indexes texts using a text id which is
1196
removed in format 5; initialization and write support for this format
1201
super(RepositoryFormat4, self).__init__()
1202
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1204
def get_format_description(self):
1205
"""See RepositoryFormat.get_format_description()."""
1206
return "Repository format 4"
1208
def initialize(self, url, shared=False, _internal=False):
1209
"""Format 4 branches cannot be created."""
1210
raise errors.UninitializableFormat(self)
1212
def is_supported(self):
1213
"""Format 4 is not supported.
1215
It is not supported because the model changed from 4 to 5 and the
1216
conversion logic is expensive - so doing it on the fly was not
1221
def _get_control_store(self, repo_transport, control_files):
1222
"""Format 4 repositories have no formal control store at this point.
1224
This will cause any control-file-needing apis to fail - this is desired.
1228
def _get_revision_store(self, repo_transport, control_files):
1229
"""See RepositoryFormat._get_revision_store()."""
1230
from bzrlib.xml4 import serializer_v4
1231
return self._get_text_rev_store(repo_transport,
1234
serializer=serializer_v4)
1236
def _get_text_store(self, transport, control_files):
1237
"""See RepositoryFormat._get_text_store()."""
1240
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1241
"""Bzr control format 5.
1243
This repository format has:
1244
- weaves for file texts and inventory
1246
- TextStores for revisions and signatures.
1250
super(RepositoryFormat5, self).__init__()
1251
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1253
def get_format_description(self):
1254
"""See RepositoryFormat.get_format_description()."""
1255
return "Weave repository format 5"
1257
def _get_revision_store(self, repo_transport, control_files):
1258
"""See RepositoryFormat._get_revision_store()."""
1259
"""Return the revision store object for this a_bzrdir."""
1260
return self._get_text_rev_store(repo_transport,
1265
def _get_text_store(self, transport, control_files):
1266
"""See RepositoryFormat._get_text_store()."""
1267
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1270
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1271
"""Bzr control format 6.
1273
This repository format has:
1274
- weaves for file texts and inventory
1275
- hash subdirectory based stores.
1276
- TextStores for revisions and signatures.
1280
super(RepositoryFormat6, self).__init__()
1281
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1283
def get_format_description(self):
1284
"""See RepositoryFormat.get_format_description()."""
1285
return "Weave repository format 6"
1287
def _get_revision_store(self, repo_transport, control_files):
1288
"""See RepositoryFormat._get_revision_store()."""
1289
return self._get_text_rev_store(repo_transport,
1295
def _get_text_store(self, transport, control_files):
1296
"""See RepositoryFormat._get_text_store()."""
1297
return self._get_versioned_file_store('weaves', transport, control_files)
1300
class MetaDirRepositoryFormat(RepositoryFormat):
1301
"""Common base class for the new repositories using the metadir layout."""
1304
super(MetaDirRepositoryFormat, self).__init__()
1305
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1307
def _create_control_files(self, a_bzrdir):
1308
"""Create the required files and the initial control_files object."""
1309
# FIXME: RBC 20060125 don't peek under the covers
1310
# NB: no need to escape relative paths that are url safe.
1311
repository_transport = a_bzrdir.get_repository_transport(self)
1312
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1313
control_files.create_lock()
1314
return control_files
1316
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1317
"""Upload the initial blank content."""
1318
control_files = self._create_control_files(a_bzrdir)
1319
control_files.lock_write()
1321
control_files._transport.mkdir_multi(dirs,
1322
mode=control_files._dir_mode)
1323
for file, content in files:
1324
control_files.put(file, content)
1325
for file, content in utf8_files:
1326
control_files.put_utf8(file, content)
1328
control_files.put_utf8('shared-storage', '')
1330
control_files.unlock()
1333
class RepositoryFormat7(MetaDirRepositoryFormat):
1334
"""Bzr repository 7.
1336
This repository format has:
1337
- weaves for file texts and inventory
1338
- hash subdirectory based stores.
1339
- TextStores for revisions and signatures.
1340
- a format marker of its own
1341
- an optional 'shared-storage' flag
1342
- an optional 'no-working-trees' flag
1345
def _get_control_store(self, repo_transport, control_files):
1346
"""Return the control store for this repository."""
1347
return self._get_versioned_file_store('',
1352
def get_format_string(self):
1353
"""See RepositoryFormat.get_format_string()."""
1354
return "Bazaar-NG Repository format 7"
1356
def get_format_description(self):
1357
"""See RepositoryFormat.get_format_description()."""
1358
return "Weave repository format 7"
1360
def _get_revision_store(self, repo_transport, control_files):
1361
"""See RepositoryFormat._get_revision_store()."""
1362
return self._get_text_rev_store(repo_transport,
1369
def _get_text_store(self, transport, control_files):
1370
"""See RepositoryFormat._get_text_store()."""
1371
return self._get_versioned_file_store('weaves',
1375
def initialize(self, a_bzrdir, shared=False):
1376
"""Create a weave repository.
1378
:param shared: If true the repository will be initialized as a shared
1381
from bzrlib.weavefile import write_weave_v5
1382
from bzrlib.weave import Weave
1384
# Create an empty weave
1386
write_weave_v5(Weave(), sio)
1387
empty_weave = sio.getvalue()
1389
mutter('creating repository in %s.', a_bzrdir.transport.base)
1390
dirs = ['revision-store', 'weaves']
1391
files = [('inventory.weave', StringIO(empty_weave)),
1393
utf8_files = [('format', self.get_format_string())]
1395
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1396
return self.open(a_bzrdir=a_bzrdir, _found=True)
1398
def open(self, a_bzrdir, _found=False, _override_transport=None):
1399
"""See RepositoryFormat.open().
1401
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1402
repository at a slightly different url
1403
than normal. I.e. during 'upgrade'.
1406
format = RepositoryFormat.find_format(a_bzrdir)
1407
assert format.__class__ == self.__class__
1408
if _override_transport is not None:
1409
repo_transport = _override_transport
1411
repo_transport = a_bzrdir.get_repository_transport(None)
1412
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1413
text_store = self._get_text_store(repo_transport, control_files)
1414
control_store = self._get_control_store(repo_transport, control_files)
1415
_revision_store = self._get_revision_store(repo_transport, control_files)
1416
return MetaDirRepository(_format=self,
1418
control_files=control_files,
1419
_revision_store=_revision_store,
1420
control_store=control_store,
1421
text_store=text_store)
1424
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1425
"""Bzr repository knit format 1.
1427
This repository format has:
1428
- knits for file texts and inventory
1429
- hash subdirectory based stores.
1430
- knits for revisions and signatures
1431
- TextStores for revisions and signatures.
1432
- a format marker of its own
1433
- an optional 'shared-storage' flag
1434
- an optional 'no-working-trees' flag
1437
This format was introduced in bzr 0.8.
1440
def _get_control_store(self, repo_transport, control_files):
1441
"""Return the control store for this repository."""
1442
return VersionedFileStore(
1445
file_mode=control_files._file_mode,
1446
versionedfile_class=KnitVersionedFile,
1447
versionedfile_kwargs={'factory':KnitPlainFactory()},
1450
def get_format_string(self):
1451
"""See RepositoryFormat.get_format_string()."""
1452
return "Bazaar-NG Knit Repository Format 1"
1454
def get_format_description(self):
1455
"""See RepositoryFormat.get_format_description()."""
1456
return "Knit repository format 1"
1458
def _get_revision_store(self, repo_transport, control_files):
1459
"""See RepositoryFormat._get_revision_store()."""
1460
from bzrlib.store.revision.knit import KnitRevisionStore
1461
versioned_file_store = VersionedFileStore(
1463
file_mode=control_files._file_mode,
1466
versionedfile_class=KnitVersionedFile,
1467
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1470
return KnitRevisionStore(versioned_file_store)
1472
def _get_text_store(self, transport, control_files):
1473
"""See RepositoryFormat._get_text_store()."""
1474
return self._get_versioned_file_store('knits',
1477
versionedfile_class=KnitVersionedFile,
1480
def initialize(self, a_bzrdir, shared=False):
1481
"""Create a knit format 1 repository.
1483
:param a_bzrdir: bzrdir to contain the new repository; must already
1485
:param shared: If true the repository will be initialized as a shared
1488
mutter('creating repository in %s.', a_bzrdir.transport.base)
1489
dirs = ['revision-store', 'knits']
1491
utf8_files = [('format', self.get_format_string())]
1493
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1494
repo_transport = a_bzrdir.get_repository_transport(None)
1495
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1496
control_store = self._get_control_store(repo_transport, control_files)
1497
transaction = transactions.WriteTransaction()
1498
# trigger a write of the inventory store.
1499
control_store.get_weave_or_empty('inventory', transaction)
1500
_revision_store = self._get_revision_store(repo_transport, control_files)
1501
_revision_store.has_revision_id('A', transaction)
1502
_revision_store.get_signature_file(transaction)
1503
return self.open(a_bzrdir=a_bzrdir, _found=True)
1505
def open(self, a_bzrdir, _found=False, _override_transport=None):
1506
"""See RepositoryFormat.open().
1508
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1509
repository at a slightly different url
1510
than normal. I.e. during 'upgrade'.
1513
format = RepositoryFormat.find_format(a_bzrdir)
1514
assert format.__class__ == self.__class__
1515
if _override_transport is not None:
1516
repo_transport = _override_transport
1518
repo_transport = a_bzrdir.get_repository_transport(None)
1519
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1520
text_store = self._get_text_store(repo_transport, control_files)
1521
control_store = self._get_control_store(repo_transport, control_files)
1522
_revision_store = self._get_revision_store(repo_transport, control_files)
1523
return KnitRepository(_format=self,
1525
control_files=control_files,
1526
_revision_store=_revision_store,
1527
control_store=control_store,
1528
text_store=text_store)
1531
# formats which have no format string are not discoverable
1532
# and not independently creatable, so are not registered.
1533
RepositoryFormat.register_format(RepositoryFormat7())
1534
_default_format = RepositoryFormatKnit1()
1535
RepositoryFormat.register_format(_default_format)
1536
RepositoryFormat.set_default_format(_default_format)
1537
_legacy_formats = [RepositoryFormat4(),
1538
RepositoryFormat5(),
1539
RepositoryFormat6()]
1542
class InterRepository(InterObject):
1543
"""This class represents operations taking place between two repositories.
1545
Its instances have methods like copy_content and fetch, and contain
1546
references to the source and target repositories these operations can be
1549
Often we will provide convenience methods on 'repository' which carry out
1550
operations with another repository - they will always forward to
1551
InterRepository.get(other).method_name(parameters).
1555
"""The available optimised InterRepository types."""
1558
def copy_content(self, revision_id=None, basis=None):
1559
"""Make a complete copy of the content in self into destination.
1561
This is a destructive operation! Do not use it on existing
1564
:param revision_id: Only copy the content needed to construct
1565
revision_id and its parents.
1566
:param basis: Copy the needed data preferentially from basis.
1569
self.target.set_make_working_trees(self.source.make_working_trees())
1570
except NotImplementedError:
1572
# grab the basis available data
1573
if basis is not None:
1574
self.target.fetch(basis, revision_id=revision_id)
1575
# but don't bother fetching if we have the needed data now.
1576
if (revision_id not in (None, NULL_REVISION) and
1577
self.target.has_revision(revision_id)):
1579
self.target.fetch(self.source, revision_id=revision_id)
1581
def _double_lock(self, lock_source, lock_target):
1582
"""Take out too locks, rolling back the first if the second throws."""
1587
# we want to ensure that we don't leave source locked by mistake.
1588
# and any error on target should not confuse source.
1589
self.source.unlock()
1593
def fetch(self, revision_id=None, pb=None):
1594
"""Fetch the content required to construct revision_id.
1596
The content is copied from source to target.
1598
:param revision_id: if None all content is copied, if NULL_REVISION no
1600
:param pb: optional progress bar to use for progress reports. If not
1601
provided a default one will be created.
1603
Returns the copied revision count and the failed revisions in a tuple:
1606
from bzrlib.fetch import GenericRepoFetcher
1607
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1608
self.source, self.source._format, self.target, self.target._format)
1609
f = GenericRepoFetcher(to_repository=self.target,
1610
from_repository=self.source,
1611
last_revision=revision_id,
1613
return f.count_copied, f.failed_revisions
1615
def lock_read(self):
1616
"""Take out a logical read lock.
1618
This will lock the source branch and the target branch. The source gets
1619
a read lock and the target a read lock.
1621
self._double_lock(self.source.lock_read, self.target.lock_read)
1623
def lock_write(self):
1624
"""Take out a logical write lock.
1626
This will lock the source branch and the target branch. The source gets
1627
a read lock and the target a write lock.
1629
self._double_lock(self.source.lock_read, self.target.lock_write)
1632
def missing_revision_ids(self, revision_id=None):
1633
"""Return the revision ids that source has that target does not.
1635
These are returned in topological order.
1637
:param revision_id: only return revision ids included by this
1640
# generic, possibly worst case, slow code path.
1641
target_ids = set(self.target.all_revision_ids())
1642
if revision_id is not None:
1643
source_ids = self.source.get_ancestry(revision_id)
1644
assert source_ids[0] == None
1647
source_ids = self.source.all_revision_ids()
1648
result_set = set(source_ids).difference(target_ids)
1649
# this may look like a no-op: its not. It preserves the ordering
1650
# other_ids had while only returning the members from other_ids
1651
# that we've decided we need.
1652
return [rev_id for rev_id in source_ids if rev_id in result_set]
1655
"""Release the locks on source and target."""
1657
self.target.unlock()
1659
self.source.unlock()
1662
class InterWeaveRepo(InterRepository):
1663
"""Optimised code paths between Weave based repositories."""
1665
_matching_repo_format = RepositoryFormat7()
1666
"""Repository format for testing with."""
1669
def is_compatible(source, target):
1670
"""Be compatible with known Weave formats.
1672
We don't test for the stores being of specific types because that
1673
could lead to confusing results, and there is no need to be
1677
return (isinstance(source._format, (RepositoryFormat5,
1679
RepositoryFormat7)) and
1680
isinstance(target._format, (RepositoryFormat5,
1682
RepositoryFormat7)))
1683
except AttributeError:
1687
def copy_content(self, revision_id=None, basis=None):
1688
"""See InterRepository.copy_content()."""
1689
# weave specific optimised path:
1690
if basis is not None:
1691
# copy the basis in, then fetch remaining data.
1692
basis.copy_content_into(self.target, revision_id)
1693
# the basis copy_content_into could miss-set this.
1695
self.target.set_make_working_trees(self.source.make_working_trees())
1696
except NotImplementedError:
1698
self.target.fetch(self.source, revision_id=revision_id)
1701
self.target.set_make_working_trees(self.source.make_working_trees())
1702
except NotImplementedError:
1704
# FIXME do not peek!
1705
if self.source.control_files._transport.listable():
1706
pb = ui.ui_factory.nested_progress_bar()
1708
self.target.weave_store.copy_all_ids(
1709
self.source.weave_store,
1711
from_transaction=self.source.get_transaction(),
1712
to_transaction=self.target.get_transaction())
1713
pb.update('copying inventory', 0, 1)
1714
self.target.control_weaves.copy_multi(
1715
self.source.control_weaves, ['inventory'],
1716
from_transaction=self.source.get_transaction(),
1717
to_transaction=self.target.get_transaction())
1718
self.target._revision_store.text_store.copy_all_ids(
1719
self.source._revision_store.text_store,
1724
self.target.fetch(self.source, revision_id=revision_id)
1727
def fetch(self, revision_id=None, pb=None):
1728
"""See InterRepository.fetch()."""
1729
from bzrlib.fetch import GenericRepoFetcher
1730
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1731
self.source, self.source._format, self.target, self.target._format)
1732
f = GenericRepoFetcher(to_repository=self.target,
1733
from_repository=self.source,
1734
last_revision=revision_id,
1736
return f.count_copied, f.failed_revisions
1739
def missing_revision_ids(self, revision_id=None):
1740
"""See InterRepository.missing_revision_ids()."""
1741
# we want all revisions to satisfy revision_id in source.
1742
# but we don't want to stat every file here and there.
1743
# we want then, all revisions other needs to satisfy revision_id
1744
# checked, but not those that we have locally.
1745
# so the first thing is to get a subset of the revisions to
1746
# satisfy revision_id in source, and then eliminate those that
1747
# we do already have.
1748
# this is slow on high latency connection to self, but as as this
1749
# disk format scales terribly for push anyway due to rewriting
1750
# inventory.weave, this is considered acceptable.
1752
if revision_id is not None:
1753
source_ids = self.source.get_ancestry(revision_id)
1754
assert source_ids[0] == None
1757
source_ids = self.source._all_possible_ids()
1758
source_ids_set = set(source_ids)
1759
# source_ids is the worst possible case we may need to pull.
1760
# now we want to filter source_ids against what we actually
1761
# have in target, but don't try to check for existence where we know
1762
# we do not have a revision as that would be pointless.
1763
target_ids = set(self.target._all_possible_ids())
1764
possibly_present_revisions = target_ids.intersection(source_ids_set)
1765
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1766
required_revisions = source_ids_set.difference(actually_present_revisions)
1767
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1768
if revision_id is not None:
1769
# we used get_ancestry to determine source_ids then we are assured all
1770
# revisions referenced are present as they are installed in topological order.
1771
# and the tip revision was validated by get_ancestry.
1772
return required_topo_revisions
1774
# if we just grabbed the possibly available ids, then
1775
# we only have an estimate of whats available and need to validate
1776
# that against the revision records.
1777
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1780
class InterKnitRepo(InterRepository):
1781
"""Optimised code paths between Knit based repositories."""
1783
_matching_repo_format = RepositoryFormatKnit1()
1784
"""Repository format for testing with."""
1787
def is_compatible(source, target):
1788
"""Be compatible with known Knit formats.
1790
We don't test for the stores being of specific types because that
1791
could lead to confusing results, and there is no need to be
1795
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1796
isinstance(target._format, (RepositoryFormatKnit1)))
1797
except AttributeError:
1801
def fetch(self, revision_id=None, pb=None):
1802
"""See InterRepository.fetch()."""
1803
from bzrlib.fetch import KnitRepoFetcher
1804
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1805
self.source, self.source._format, self.target, self.target._format)
1806
f = KnitRepoFetcher(to_repository=self.target,
1807
from_repository=self.source,
1808
last_revision=revision_id,
1810
return f.count_copied, f.failed_revisions
1813
def missing_revision_ids(self, revision_id=None):
1814
"""See InterRepository.missing_revision_ids()."""
1815
if revision_id is not None:
1816
source_ids = self.source.get_ancestry(revision_id)
1817
assert source_ids[0] == None
1820
source_ids = self.source._all_possible_ids()
1821
source_ids_set = set(source_ids)
1822
# source_ids is the worst possible case we may need to pull.
1823
# now we want to filter source_ids against what we actually
1824
# have in target, but don't try to check for existence where we know
1825
# we do not have a revision as that would be pointless.
1826
target_ids = set(self.target._all_possible_ids())
1827
possibly_present_revisions = target_ids.intersection(source_ids_set)
1828
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1829
required_revisions = source_ids_set.difference(actually_present_revisions)
1830
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1831
if revision_id is not None:
1832
# we used get_ancestry to determine source_ids then we are assured all
1833
# revisions referenced are present as they are installed in topological order.
1834
# and the tip revision was validated by get_ancestry.
1835
return required_topo_revisions
1837
# if we just grabbed the possibly available ids, then
1838
# we only have an estimate of whats available and need to validate
1839
# that against the revision records.
1840
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1842
InterRepository.register_optimiser(InterWeaveRepo)
1843
InterRepository.register_optimiser(InterKnitRepo)
1846
class RepositoryTestProviderAdapter(object):
1847
"""A tool to generate a suite testing multiple repository formats at once.
1849
This is done by copying the test once for each transport and injecting
1850
the transport_server, transport_readonly_server, and bzrdir_format and
1851
repository_format classes into each copy. Each copy is also given a new id()
1852
to make it easy to identify.
1855
def __init__(self, transport_server, transport_readonly_server, formats):
1856
self._transport_server = transport_server
1857
self._transport_readonly_server = transport_readonly_server
1858
self._formats = formats
1860
def adapt(self, test):
1861
result = TestSuite()
1862
for repository_format, bzrdir_format in self._formats:
1863
new_test = deepcopy(test)
1864
new_test.transport_server = self._transport_server
1865
new_test.transport_readonly_server = self._transport_readonly_server
1866
new_test.bzrdir_format = bzrdir_format
1867
new_test.repository_format = repository_format
1868
def make_new_test_id():
1869
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1870
return lambda: new_id
1871
new_test.id = make_new_test_id()
1872
result.addTest(new_test)
1876
class InterRepositoryTestProviderAdapter(object):
1877
"""A tool to generate a suite testing multiple inter repository formats.
1879
This is done by copying the test once for each interrepo provider and injecting
1880
the transport_server, transport_readonly_server, repository_format and
1881
repository_to_format classes into each copy.
1882
Each copy is also given a new id() to make it easy to identify.
1885
def __init__(self, transport_server, transport_readonly_server, formats):
1886
self._transport_server = transport_server
1887
self._transport_readonly_server = transport_readonly_server
1888
self._formats = formats
1890
def adapt(self, test):
1891
result = TestSuite()
1892
for interrepo_class, repository_format, repository_format_to in self._formats:
1893
new_test = deepcopy(test)
1894
new_test.transport_server = self._transport_server
1895
new_test.transport_readonly_server = self._transport_readonly_server
1896
new_test.interrepo_class = interrepo_class
1897
new_test.repository_format = repository_format
1898
new_test.repository_format_to = repository_format_to
1899
def make_new_test_id():
1900
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1901
return lambda: new_id
1902
new_test.id = make_new_test_id()
1903
result.addTest(new_test)
1907
def default_test_list():
1908
"""Generate the default list of interrepo permutations to test."""
1910
# test the default InterRepository between format 6 and the current
1912
# XXX: robertc 20060220 reinstate this when there are two supported
1913
# formats which do not have an optimal code path between them.
1914
result.append((InterRepository,
1915
RepositoryFormat6(),
1916
RepositoryFormatKnit1()))
1917
for optimiser in InterRepository._optimisers:
1918
result.append((optimiser,
1919
optimiser._matching_repo_format,
1920
optimiser._matching_repo_format
1922
# if there are specific combinations we want to use, we can add them
1927
class CopyConverter(object):
1928
"""A repository conversion tool which just performs a copy of the content.
1930
This is slow but quite reliable.
1933
def __init__(self, target_format):
1934
"""Create a CopyConverter.
1936
:param target_format: The format the resulting repository should be.
1938
self.target_format = target_format
1940
def convert(self, repo, pb):
1941
"""Perform the conversion of to_convert, giving feedback via pb.
1943
:param to_convert: The disk object to convert.
1944
:param pb: a progress bar to use for progress information.
1949
# this is only useful with metadir layouts - separated repo content.
1950
# trigger an assertion if not such
1951
repo._format.get_format_string()
1952
self.repo_dir = repo.bzrdir
1953
self.step('Moving repository to repository.backup')
1954
self.repo_dir.transport.move('repository', 'repository.backup')
1955
backup_transport = self.repo_dir.transport.clone('repository.backup')
1956
self.source_repo = repo._format.open(self.repo_dir,
1958
_override_transport=backup_transport)
1959
self.step('Creating new repository')
1960
converted = self.target_format.initialize(self.repo_dir,
1961
self.source_repo.is_shared())
1962
converted.lock_write()
1964
self.step('Copying content into repository.')
1965
self.source_repo.copy_content_into(converted)
1968
self.step('Deleting old repository content.')
1969
self.repo_dir.transport.delete_tree('repository.backup')
1970
self.pb.note('repository converted')
1972
def step(self, message):
1973
"""Update the pb by a step."""
1975
self.pb.update(message, self.count, self.total)
1978
class CommitBuilder(object):
1979
"""Provides an interface to build up a commit.
1981
This allows describing a tree to be committed without needing to
1982
know the internals of the format of the repository.
1984
def __init__(self, repository, parents, config, timestamp=None,
1985
timezone=None, committer=None, revprops=None,
1987
"""Initiate a CommitBuilder.
1989
:param repository: Repository to commit to.
1990
:param parents: Revision ids of the parents of the new revision.
1991
:param config: Configuration to use.
1992
:param timestamp: Optional timestamp recorded for commit.
1993
:param timezone: Optional timezone for timestamp.
1994
:param committer: Optional committer to set for commit.
1995
:param revprops: Optional dictionary of revision properties.
1996
:param revision_id: Optional revision id.
1998
self._config = config
2000
if committer is None:
2001
self._committer = self._config.username()
2003
assert isinstance(committer, basestring), type(committer)
2004
self._committer = committer
2006
self.new_inventory = Inventory()
2007
self._new_revision_id = revision_id
2008
self.parents = parents
2009
self.repository = repository
2012
if revprops is not None:
2013
self._revprops.update(revprops)
2015
if timestamp is None:
2016
timestamp = time.time()
2017
# Restrict resolution to 1ms
2018
self._timestamp = round(timestamp, 3)
2020
if timezone is None:
2021
self._timezone = local_time_offset()
2023
self._timezone = int(timezone)
2025
self._generate_revision_if_needed()
2027
def commit(self, message):
2028
"""Make the actual commit.
2030
:return: The revision id of the recorded revision.
2032
rev = Revision(timestamp=self._timestamp,
2033
timezone=self._timezone,
2034
committer=self._committer,
2036
inventory_sha1=self.inv_sha1,
2037
revision_id=self._new_revision_id,
2038
properties=self._revprops)
2039
rev.parent_ids = self.parents
2040
self.repository.add_revision(self._new_revision_id, rev,
2041
self.new_inventory, self._config)
2042
return self._new_revision_id
2044
def finish_inventory(self):
2045
"""Tell the builder that the inventory is finished."""
2046
self.new_inventory.revision_id = self._new_revision_id
2047
self.inv_sha1 = self.repository.add_inventory(
2048
self._new_revision_id,
2053
def _gen_revision_id(self):
2054
"""Return new revision-id."""
2055
s = '%s-%s-' % (self._config.user_email(),
2056
compact_date(self._timestamp))
2057
s += hexlify(rand_bytes(8))
2060
def _generate_revision_if_needed(self):
2061
"""Create a revision id if None was supplied.
2063
If the repository can not support user-specified revision ids
2064
they should override this function and raise UnsupportedOperation
2065
if _new_revision_id is not None.
2067
:raises: UnsupportedOperation
2069
if self._new_revision_id is None:
2070
self._new_revision_id = self._gen_revision_id()
2072
def record_entry_contents(self, ie, parent_invs, path, tree):
2073
"""Record the content of ie from tree into the commit if needed.
2075
: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
self.new_inventory.add(ie)
2084
# ie.revision is always None if the InventoryEntry is considered
2085
# for committing. ie.snapshot will record the correct revision
2086
# which may be the sole parent if it is untouched.
2087
if ie.revision is not None:
2089
previous_entries = ie.find_previous_heads(
2091
self.repository.weave_store,
2092
self.repository.get_transaction())
2093
# we are creating a new revision for ie in the history store
2095
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2097
def modified_directory(self, file_id, file_parents):
2098
"""Record the presence of a symbolic link.
2100
:param file_id: The file_id of the link to record.
2101
:param file_parents: The per-file parent revision ids.
2103
self._add_text_to_weave(file_id, [], file_parents.keys())
2105
def modified_file_text(self, file_id, file_parents,
2106
get_content_byte_lines, text_sha1=None,
2108
"""Record the text of file file_id
2110
:param file_id: The file_id of the file to record the text of.
2111
:param file_parents: The per-file parent revision ids.
2112
:param get_content_byte_lines: A callable which will return the byte
2114
:param text_sha1: Optional SHA1 of the file contents.
2115
:param text_size: Optional size of the file contents.
2117
mutter('storing text of file {%s} in revision {%s} into %r',
2118
file_id, self._new_revision_id, self.repository.weave_store)
2119
# special case to avoid diffing on renames or
2121
if (len(file_parents) == 1
2122
and text_sha1 == file_parents.values()[0].text_sha1
2123
and text_size == file_parents.values()[0].text_size):
2124
previous_ie = file_parents.values()[0]
2125
versionedfile = self.repository.weave_store.get_weave(file_id,
2126
self.repository.get_transaction())
2127
versionedfile.clone_text(self._new_revision_id,
2128
previous_ie.revision, file_parents.keys())
2129
return text_sha1, text_size
2131
new_lines = get_content_byte_lines()
2132
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2133
# should return the SHA1 and size
2134
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2135
return osutils.sha_strings(new_lines), \
2136
sum(map(len, new_lines))
2138
def modified_link(self, file_id, file_parents, link_target):
2139
"""Record the presence of a symbolic link.
2141
:param file_id: The file_id of the link to record.
2142
:param file_parents: The per-file parent revision ids.
2143
:param link_target: Target location of this link.
2145
self._add_text_to_weave(file_id, [], file_parents.keys())
2147
def _add_text_to_weave(self, file_id, new_lines, parents):
2148
versionedfile = self.repository.weave_store.get_weave_or_empty(
2149
file_id, self.repository.get_transaction())
2150
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2151
versionedfile.clear_cache()
2163
def _unescaper(match, _map=_unescape_map):
2164
return _map[match.group(1)]
2170
def _unescape_xml(data):
2171
"""Unescape predefined XML entities in a string of data."""
2173
if _unescape_re is None:
2174
_unescape_re = re.compile('\&([^;]*);')
2175
return _unescape_re.sub(_unescaper, data)