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.trace import mutter, note
42
from bzrlib.tree import RevisionTree, EmptyTree
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_revisions([revision_id],
304
self.get_transaction())[0]
306
def get_revisions(self, revision_ids):
307
return self._revision_store.get_revisions(revision_ids,
308
self.get_transaction())
311
def get_revision_xml(self, revision_id):
312
rev = self.get_revision(revision_id)
314
# the current serializer..
315
self._revision_store._serializer.write_revision(rev, rev_tmp)
317
return rev_tmp.getvalue()
320
def get_revision(self, revision_id):
321
"""Return the Revision object for a named revision"""
322
r = self.get_revision_reconcile(revision_id)
323
# weave corruption can lead to absent revision markers that should be
325
# the following test is reasonably cheap (it needs a single weave read)
326
# and the weave is cached in read transactions. In write transactions
327
# it is not cached but typically we only read a small number of
328
# revisions. For knits when they are introduced we will probably want
329
# to ensure that caching write transactions are in use.
330
inv = self.get_inventory_weave()
331
self._check_revision_parents(r, inv)
334
def get_revision_delta(self, revision_id):
335
"""Return the delta for one revision.
337
The delta is relative to the left-hand predecessor of the
340
revision = self.get_revision(revision_id)
341
new_tree = self.revision_tree(revision_id)
342
if not revision.parent_ids:
343
old_tree = EmptyTree()
345
old_tree = self.revision_tree(revision.parent_ids[0])
346
return delta.compare_trees(old_tree, new_tree)
348
def _check_revision_parents(self, revision, inventory):
349
"""Private to Repository and Fetch.
351
This checks the parentage of revision in an inventory weave for
352
consistency and is only applicable to inventory-weave-for-ancestry
353
using repository formats & fetchers.
355
weave_parents = inventory.get_parents(revision.revision_id)
356
weave_names = inventory.versions()
357
for parent_id in revision.parent_ids:
358
if parent_id in weave_names:
359
# this parent must not be a ghost.
360
if not parent_id in weave_parents:
362
raise errors.CorruptRepository(self)
365
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
366
signature = gpg_strategy.sign(plaintext)
367
self._revision_store.add_revision_signature_text(revision_id,
369
self.get_transaction())
371
def fileids_altered_by_revision_ids(self, revision_ids):
372
"""Find the file ids and versions affected by revisions.
374
:param revisions: an iterable containing revision ids.
375
:return: a dictionary mapping altered file-ids to an iterable of
376
revision_ids. Each altered file-ids has the exact revision_ids that
377
altered it listed explicitly.
379
assert isinstance(self._format, (RepositoryFormat5,
382
RepositoryFormatKnit1)), \
383
("fileids_altered_by_revision_ids only supported for branches "
384
"which store inventory as unnested xml, not on %r" % self)
385
selected_revision_ids = set(revision_ids)
386
w = self.get_inventory_weave()
389
# this code needs to read every new line in every inventory for the
390
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
391
# not present in one of those inventories is unnecessary but not
392
# harmful because we are filtering by the revision id marker in the
393
# inventory lines : we only select file ids altered in one of those
394
# revisions. We don't need to see all lines in the inventory because
395
# only those added in an inventory in rev X can contain a revision=X
397
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
398
start = line.find('file_id="')+9
399
if start < 9: continue
400
end = line.find('"', start)
402
file_id = _unescape_xml(line[start:end])
404
start = line.find('revision="')+10
405
if start < 10: continue
406
end = line.find('"', start)
408
revision_id = _unescape_xml(line[start:end])
409
if revision_id in selected_revision_ids:
410
result.setdefault(file_id, set()).add(revision_id)
414
def get_inventory_weave(self):
415
return self.control_weaves.get_weave('inventory',
416
self.get_transaction())
419
def get_inventory(self, revision_id):
420
"""Get Inventory object by hash."""
421
return self.deserialise_inventory(
422
revision_id, self.get_inventory_xml(revision_id))
424
def deserialise_inventory(self, revision_id, xml):
425
"""Transform the xml into an inventory object.
427
:param revision_id: The expected revision id of the inventory.
428
:param xml: A serialised inventory.
430
return xml5.serializer_v5.read_inventory_from_string(xml)
433
def get_inventory_xml(self, revision_id):
434
"""Get inventory XML as a file object."""
436
assert isinstance(revision_id, basestring), type(revision_id)
437
iw = self.get_inventory_weave()
438
return iw.get_text(revision_id)
440
raise errors.HistoryMissing(self, 'inventory', revision_id)
443
def get_inventory_sha1(self, revision_id):
444
"""Return the sha1 hash of the inventory entry
446
return self.get_revision(revision_id).inventory_sha1
449
def get_revision_graph(self, revision_id=None):
450
"""Return a dictionary containing the revision graph.
452
:return: a dictionary of revision_id->revision_parents_list.
454
weave = self.get_inventory_weave()
455
all_revisions = self._eliminate_revisions_not_present(weave.versions())
456
entire_graph = dict([(node, weave.get_parents(node)) for
457
node in all_revisions])
458
if revision_id is None:
460
elif revision_id not in entire_graph:
461
raise errors.NoSuchRevision(self, revision_id)
463
# add what can be reached from revision_id
465
pending = set([revision_id])
466
while len(pending) > 0:
468
result[node] = entire_graph[node]
469
for revision_id in result[node]:
470
if revision_id not in result:
471
pending.add(revision_id)
475
def get_revision_graph_with_ghosts(self, revision_ids=None):
476
"""Return a graph of the revisions with ghosts marked as applicable.
478
:param revision_ids: an iterable of revisions to graph or None for all.
479
:return: a Graph object with the graph reachable from revision_ids.
483
pending = set(self.all_revision_ids())
486
pending = set(revision_ids)
487
required = set(revision_ids)
490
revision_id = pending.pop()
492
rev = self.get_revision(revision_id)
493
except errors.NoSuchRevision:
494
if revision_id in required:
497
result.add_ghost(revision_id)
499
for parent_id in rev.parent_ids:
500
# is this queued or done ?
501
if (parent_id not in pending and
502
parent_id not in done):
504
pending.add(parent_id)
505
result.add_node(revision_id, rev.parent_ids)
506
done.add(revision_id)
510
def get_revision_inventory(self, revision_id):
511
"""Return inventory of a past revision."""
512
# TODO: Unify this with get_inventory()
513
# bzr 0.0.6 and later imposes the constraint that the inventory_id
514
# must be the same as its revision, so this is trivial.
515
if revision_id is None:
516
# This does not make sense: if there is no revision,
517
# then it is the current tree inventory surely ?!
518
# and thus get_root_id() is something that looks at the last
519
# commit on the branch, and the get_root_id is an inventory check.
520
raise NotImplementedError
521
# return Inventory(self.get_root_id())
523
return self.get_inventory(revision_id)
527
"""Return True if this repository is flagged as a shared repository."""
528
raise NotImplementedError(self.is_shared)
531
def reconcile(self, other=None, thorough=False):
532
"""Reconcile this repository."""
533
from bzrlib.reconcile import RepoReconciler
534
reconciler = RepoReconciler(self, thorough=thorough)
535
reconciler.reconcile()
539
def revision_tree(self, revision_id):
540
"""Return Tree for a revision on this branch.
542
`revision_id` may be None for the null revision, in which case
543
an `EmptyTree` is returned."""
544
# TODO: refactor this to use an existing revision object
545
# so we don't need to read it in twice.
546
if revision_id is None or revision_id == NULL_REVISION:
549
inv = self.get_revision_inventory(revision_id)
550
return RevisionTree(self, inv, revision_id)
553
def get_ancestry(self, revision_id):
554
"""Return a list of revision-ids integrated by a revision.
556
The first element of the list is always None, indicating the origin
557
revision. This might change when we have history horizons, or
558
perhaps we should have a new API.
560
This is topologically sorted.
562
if revision_id is None:
564
if not self.has_revision(revision_id):
565
raise errors.NoSuchRevision(self, revision_id)
566
w = self.get_inventory_weave()
567
candidates = w.get_ancestry(revision_id)
568
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
571
def print_file(self, file, revision_id):
572
"""Print `file` to stdout.
574
FIXME RBC 20060125 as John Meinel points out this is a bad api
575
- it writes to stdout, it assumes that that is valid etc. Fix
576
by creating a new more flexible convenience function.
578
tree = self.revision_tree(revision_id)
579
# use inventory as it was in that revision
580
file_id = tree.inventory.path2id(file)
582
# TODO: jam 20060427 Write a test for this code path
583
# it had a bug in it, and was raising the wrong
585
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
586
tree.print_file(file_id)
588
def get_transaction(self):
589
return self.control_files.get_transaction()
591
def revision_parents(self, revid):
592
return self.get_inventory_weave().parent_names(revid)
595
def set_make_working_trees(self, new_value):
596
"""Set the policy flag for making working trees when creating branches.
598
This only applies to branches that use this repository.
600
The default is 'True'.
601
:param new_value: True to restore the default, False to disable making
604
raise NotImplementedError(self.set_make_working_trees)
606
def make_working_trees(self):
607
"""Returns the policy for making working trees on new branches."""
608
raise NotImplementedError(self.make_working_trees)
611
def sign_revision(self, revision_id, gpg_strategy):
612
plaintext = Testament.from_revision(self, revision_id).as_short_text()
613
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
616
def has_signature_for_revision_id(self, revision_id):
617
"""Query for a revision signature for revision_id in the repository."""
618
return self._revision_store.has_signature(revision_id,
619
self.get_transaction())
622
def get_signature_text(self, revision_id):
623
"""Return the text for a signature."""
624
return self._revision_store.get_signature_text(revision_id,
625
self.get_transaction())
628
def check(self, revision_ids):
629
"""Check consistency of all history of given revision_ids.
631
Different repository implementations should override _check().
633
:param revision_ids: A non-empty list of revision_ids whose ancestry
634
will be checked. Typically the last revision_id of a branch.
637
raise ValueError("revision_ids must be non-empty in %s.check"
639
return self._check(revision_ids)
641
def _check(self, revision_ids):
642
result = check.Check(self)
647
class AllInOneRepository(Repository):
648
"""Legacy support - the repository behaviour for all-in-one branches."""
650
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
651
# we reuse one control files instance.
652
dir_mode = a_bzrdir._control_files._dir_mode
653
file_mode = a_bzrdir._control_files._file_mode
655
def get_store(name, compressed=True, prefixed=False):
656
# FIXME: This approach of assuming stores are all entirely compressed
657
# or entirely uncompressed is tidy, but breaks upgrade from
658
# some existing branches where there's a mixture; we probably
659
# still want the option to look for both.
660
relpath = a_bzrdir._control_files._escape(name)
661
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
662
prefixed=prefixed, compressed=compressed,
665
#if self._transport.should_cache():
666
# cache_path = os.path.join(self.cache_root, name)
667
# os.mkdir(cache_path)
668
# store = bzrlib.store.CachedStore(store, cache_path)
671
# not broken out yet because the controlweaves|inventory_store
672
# and text_store | weave_store bits are still different.
673
if isinstance(_format, RepositoryFormat4):
674
# cannot remove these - there is still no consistent api
675
# which allows access to this old info.
676
self.inventory_store = get_store('inventory-store')
677
text_store = get_store('text-store')
678
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
682
"""AllInOne repositories cannot be shared."""
686
def set_make_working_trees(self, new_value):
687
"""Set the policy flag for making working trees when creating branches.
689
This only applies to branches that use this repository.
691
The default is 'True'.
692
:param new_value: True to restore the default, False to disable making
695
raise NotImplementedError(self.set_make_working_trees)
697
def make_working_trees(self):
698
"""Returns the policy for making working trees on new branches."""
702
def install_revision(repository, rev, revision_tree):
703
"""Install all revision data into a repository."""
706
for p_id in rev.parent_ids:
707
if repository.has_revision(p_id):
708
present_parents.append(p_id)
709
parent_trees[p_id] = repository.revision_tree(p_id)
711
parent_trees[p_id] = EmptyTree()
713
inv = revision_tree.inventory
715
# Add the texts that are not already present
716
for path, ie in inv.iter_entries():
717
w = repository.weave_store.get_weave_or_empty(ie.file_id,
718
repository.get_transaction())
719
if ie.revision not in w:
721
# FIXME: TODO: The following loop *may* be overlapping/duplicate
722
# with InventoryEntry.find_previous_heads(). if it is, then there
723
# is a latent bug here where the parents may have ancestors of each
725
for revision, tree in parent_trees.iteritems():
726
if ie.file_id not in tree:
728
parent_id = tree.inventory[ie.file_id].revision
729
if parent_id in text_parents:
731
text_parents.append(parent_id)
733
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
734
repository.get_transaction())
735
lines = revision_tree.get_file(ie.file_id).readlines()
736
vfile.add_lines(rev.revision_id, text_parents, lines)
738
# install the inventory
739
repository.add_inventory(rev.revision_id, inv, present_parents)
740
except errors.RevisionAlreadyPresent:
742
repository.add_revision(rev.revision_id, rev, inv)
745
class MetaDirRepository(Repository):
746
"""Repositories in the new meta-dir layout."""
748
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
749
super(MetaDirRepository, self).__init__(_format,
756
dir_mode = self.control_files._dir_mode
757
file_mode = self.control_files._file_mode
761
"""Return True if this repository is flagged as a shared repository."""
762
return self.control_files._transport.has('shared-storage')
765
def set_make_working_trees(self, new_value):
766
"""Set the policy flag for making working trees when creating branches.
768
This only applies to branches that use this repository.
770
The default is 'True'.
771
:param new_value: True to restore the default, False to disable making
776
self.control_files._transport.delete('no-working-trees')
777
except errors.NoSuchFile:
780
self.control_files.put_utf8('no-working-trees', '')
782
def make_working_trees(self):
783
"""Returns the policy for making working trees on new branches."""
784
return not self.control_files._transport.has('no-working-trees')
787
class KnitRepository(MetaDirRepository):
788
"""Knit format repository."""
790
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
791
inv_vf.add_lines_with_ghosts(revid, parents, lines)
794
def _all_revision_ids(self):
795
"""See Repository.all_revision_ids()."""
796
# Knits get the revision graph from the index of the revision knit, so
797
# it's always possible even if they're on an unlistable transport.
798
return self._revision_store.all_revision_ids(self.get_transaction())
800
def fileid_involved_between_revs(self, from_revid, to_revid):
801
"""Find file_id(s) which are involved in the changes between revisions.
803
This determines the set of revisions which are involved, and then
804
finds all file ids affected by those revisions.
806
vf = self._get_revision_vf()
807
from_set = set(vf.get_ancestry(from_revid))
808
to_set = set(vf.get_ancestry(to_revid))
809
changed = to_set.difference(from_set)
810
return self._fileid_involved_by_set(changed)
812
def fileid_involved(self, last_revid=None):
813
"""Find all file_ids modified in the ancestry of last_revid.
815
:param last_revid: If None, last_revision() will be used.
818
changed = set(self.all_revision_ids())
820
changed = set(self.get_ancestry(last_revid))
823
return self._fileid_involved_by_set(changed)
826
def get_ancestry(self, revision_id):
827
"""Return a list of revision-ids integrated by a revision.
829
This is topologically sorted.
831
if revision_id is None:
833
vf = self._get_revision_vf()
835
return [None] + vf.get_ancestry(revision_id)
836
except errors.RevisionNotPresent:
837
raise errors.NoSuchRevision(self, revision_id)
840
def get_revision(self, revision_id):
841
"""Return the Revision object for a named revision"""
842
return self.get_revision_reconcile(revision_id)
845
def get_revision_graph(self, revision_id=None):
846
"""Return a dictionary containing the revision graph.
848
:return: a dictionary of revision_id->revision_parents_list.
850
weave = self._get_revision_vf()
851
entire_graph = weave.get_graph()
852
if revision_id is None:
853
return weave.get_graph()
854
elif revision_id not in weave:
855
raise errors.NoSuchRevision(self, revision_id)
857
# add what can be reached from revision_id
859
pending = set([revision_id])
860
while len(pending) > 0:
862
result[node] = weave.get_parents(node)
863
for revision_id in result[node]:
864
if revision_id not in result:
865
pending.add(revision_id)
869
def get_revision_graph_with_ghosts(self, revision_ids=None):
870
"""Return a graph of the revisions with ghosts marked as applicable.
872
:param revision_ids: an iterable of revisions to graph or None for all.
873
:return: a Graph object with the graph reachable from revision_ids.
876
vf = self._get_revision_vf()
877
versions = set(vf.versions())
879
pending = set(self.all_revision_ids())
882
pending = set(revision_ids)
883
required = set(revision_ids)
886
revision_id = pending.pop()
887
if not revision_id in versions:
888
if revision_id in required:
889
raise errors.NoSuchRevision(self, revision_id)
891
result.add_ghost(revision_id)
892
# mark it as done so we don't try for it again.
893
done.add(revision_id)
895
parent_ids = vf.get_parents_with_ghosts(revision_id)
896
for parent_id in parent_ids:
897
# is this queued or done ?
898
if (parent_id not in pending and
899
parent_id not in done):
901
pending.add(parent_id)
902
result.add_node(revision_id, parent_ids)
903
done.add(revision_id)
906
def _get_revision_vf(self):
907
""":return: a versioned file containing the revisions."""
908
vf = self._revision_store.get_revision_file(self.get_transaction())
912
def reconcile(self, other=None, thorough=False):
913
"""Reconcile this repository."""
914
from bzrlib.reconcile import KnitReconciler
915
reconciler = KnitReconciler(self, thorough=thorough)
916
reconciler.reconcile()
919
def revision_parents(self, revision_id):
920
return self._get_revision_vf().get_parents(revision_id)
923
class RepositoryFormat(object):
924
"""A repository format.
926
Formats provide three things:
927
* An initialization routine to construct repository data on disk.
928
* a format string which is used when the BzrDir supports versioned
930
* an open routine which returns a Repository instance.
932
Formats are placed in an dict by their format string for reference
933
during opening. These should be subclasses of RepositoryFormat
936
Once a format is deprecated, just deprecate the initialize and open
937
methods on the format class. Do not deprecate the object, as the
938
object will be created every system load.
940
Common instance attributes:
941
_matchingbzrdir - the bzrdir format that the repository format was
942
originally written to work with. This can be used if manually
943
constructing a bzrdir and repository, or more commonly for test suite
947
_default_format = None
948
"""The default format used for new repositories."""
951
"""The known formats."""
954
def find_format(klass, a_bzrdir):
955
"""Return the format for the repository object in a_bzrdir."""
957
transport = a_bzrdir.get_repository_transport(None)
958
format_string = transport.get("format").read()
959
return klass._formats[format_string]
960
except errors.NoSuchFile:
961
raise errors.NoRepositoryPresent(a_bzrdir)
963
raise errors.UnknownFormatError(format=format_string)
965
def _get_control_store(self, repo_transport, control_files):
966
"""Return the control store for this repository."""
967
raise NotImplementedError(self._get_control_store)
970
def get_default_format(klass):
971
"""Return the current default format."""
972
return klass._default_format
974
def get_format_string(self):
975
"""Return the ASCII format string that identifies this format.
977
Note that in pre format ?? repositories the format string is
978
not permitted nor written to disk.
980
raise NotImplementedError(self.get_format_string)
982
def get_format_description(self):
983
"""Return the short description for this format."""
984
raise NotImplementedError(self.get_format_description)
986
def _get_revision_store(self, repo_transport, control_files):
987
"""Return the revision store object for this a_bzrdir."""
988
raise NotImplementedError(self._get_revision_store)
990
def _get_text_rev_store(self,
997
"""Common logic for getting a revision store for a repository.
999
see self._get_revision_store for the subclass-overridable method to
1000
get the store for a repository.
1002
from bzrlib.store.revision.text import TextRevisionStore
1003
dir_mode = control_files._dir_mode
1004
file_mode = control_files._file_mode
1005
text_store =TextStore(transport.clone(name),
1007
compressed=compressed,
1009
file_mode=file_mode)
1010
_revision_store = TextRevisionStore(text_store, serializer)
1011
return _revision_store
1013
def _get_versioned_file_store(self,
1018
versionedfile_class=WeaveFile,
1020
weave_transport = control_files._transport.clone(name)
1021
dir_mode = control_files._dir_mode
1022
file_mode = control_files._file_mode
1023
return VersionedFileStore(weave_transport, prefixed=prefixed,
1025
file_mode=file_mode,
1026
versionedfile_class=versionedfile_class,
1029
def initialize(self, a_bzrdir, shared=False):
1030
"""Initialize a repository of this format in a_bzrdir.
1032
:param a_bzrdir: The bzrdir to put the new repository in it.
1033
:param shared: The repository should be initialized as a sharable one.
1035
This may raise UninitializableFormat if shared repository are not
1036
compatible the a_bzrdir.
1039
def is_supported(self):
1040
"""Is this format supported?
1042
Supported formats must be initializable and openable.
1043
Unsupported formats may not support initialization or committing or
1044
some other features depending on the reason for not being supported.
1048
def open(self, a_bzrdir, _found=False):
1049
"""Return an instance of this format for the bzrdir a_bzrdir.
1051
_found is a private parameter, do not use it.
1053
raise NotImplementedError(self.open)
1056
def register_format(klass, format):
1057
klass._formats[format.get_format_string()] = format
1060
def set_default_format(klass, format):
1061
klass._default_format = format
1064
def unregister_format(klass, format):
1065
assert klass._formats[format.get_format_string()] is format
1066
del klass._formats[format.get_format_string()]
1069
class PreSplitOutRepositoryFormat(RepositoryFormat):
1070
"""Base class for the pre split out repository formats."""
1072
def initialize(self, a_bzrdir, shared=False, _internal=False):
1073
"""Create a weave repository.
1075
TODO: when creating split out bzr branch formats, move this to a common
1076
base for Format5, Format6. or something like that.
1078
from bzrlib.weavefile import write_weave_v5
1079
from bzrlib.weave import Weave
1082
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1085
# always initialized when the bzrdir is.
1086
return self.open(a_bzrdir, _found=True)
1088
# Create an empty weave
1090
write_weave_v5(Weave(), sio)
1091
empty_weave = sio.getvalue()
1093
mutter('creating repository in %s.', a_bzrdir.transport.base)
1094
dirs = ['revision-store', 'weaves']
1095
files = [('inventory.weave', StringIO(empty_weave)),
1098
# FIXME: RBC 20060125 don't peek under the covers
1099
# NB: no need to escape relative paths that are url safe.
1100
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1102
control_files.create_lock()
1103
control_files.lock_write()
1104
control_files._transport.mkdir_multi(dirs,
1105
mode=control_files._dir_mode)
1107
for file, content in files:
1108
control_files.put(file, content)
1110
control_files.unlock()
1111
return self.open(a_bzrdir, _found=True)
1113
def _get_control_store(self, repo_transport, control_files):
1114
"""Return the control store for this repository."""
1115
return self._get_versioned_file_store('',
1120
def _get_text_store(self, transport, control_files):
1121
"""Get a store for file texts for this format."""
1122
raise NotImplementedError(self._get_text_store)
1124
def open(self, a_bzrdir, _found=False):
1125
"""See RepositoryFormat.open()."""
1127
# we are being called directly and must probe.
1128
raise NotImplementedError
1130
repo_transport = a_bzrdir.get_repository_transport(None)
1131
control_files = a_bzrdir._control_files
1132
text_store = self._get_text_store(repo_transport, control_files)
1133
control_store = self._get_control_store(repo_transport, control_files)
1134
_revision_store = self._get_revision_store(repo_transport, control_files)
1135
return AllInOneRepository(_format=self,
1137
_revision_store=_revision_store,
1138
control_store=control_store,
1139
text_store=text_store)
1142
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1143
"""Bzr repository format 4.
1145
This repository format has:
1147
- TextStores for texts, inventories,revisions.
1149
This format is deprecated: it indexes texts using a text id which is
1150
removed in format 5; initialization and write support for this format
1155
super(RepositoryFormat4, self).__init__()
1156
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1158
def get_format_description(self):
1159
"""See RepositoryFormat.get_format_description()."""
1160
return "Repository format 4"
1162
def initialize(self, url, shared=False, _internal=False):
1163
"""Format 4 branches cannot be created."""
1164
raise errors.UninitializableFormat(self)
1166
def is_supported(self):
1167
"""Format 4 is not supported.
1169
It is not supported because the model changed from 4 to 5 and the
1170
conversion logic is expensive - so doing it on the fly was not
1175
def _get_control_store(self, repo_transport, control_files):
1176
"""Format 4 repositories have no formal control store at this point.
1178
This will cause any control-file-needing apis to fail - this is desired.
1182
def _get_revision_store(self, repo_transport, control_files):
1183
"""See RepositoryFormat._get_revision_store()."""
1184
from bzrlib.xml4 import serializer_v4
1185
return self._get_text_rev_store(repo_transport,
1188
serializer=serializer_v4)
1190
def _get_text_store(self, transport, control_files):
1191
"""See RepositoryFormat._get_text_store()."""
1194
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1195
"""Bzr control format 5.
1197
This repository format has:
1198
- weaves for file texts and inventory
1200
- TextStores for revisions and signatures.
1204
super(RepositoryFormat5, self).__init__()
1205
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1207
def get_format_description(self):
1208
"""See RepositoryFormat.get_format_description()."""
1209
return "Weave repository format 5"
1211
def _get_revision_store(self, repo_transport, control_files):
1212
"""See RepositoryFormat._get_revision_store()."""
1213
"""Return the revision store object for this a_bzrdir."""
1214
return self._get_text_rev_store(repo_transport,
1219
def _get_text_store(self, transport, control_files):
1220
"""See RepositoryFormat._get_text_store()."""
1221
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1224
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1225
"""Bzr control format 6.
1227
This repository format has:
1228
- weaves for file texts and inventory
1229
- hash subdirectory based stores.
1230
- TextStores for revisions and signatures.
1234
super(RepositoryFormat6, self).__init__()
1235
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1237
def get_format_description(self):
1238
"""See RepositoryFormat.get_format_description()."""
1239
return "Weave repository format 6"
1241
def _get_revision_store(self, repo_transport, control_files):
1242
"""See RepositoryFormat._get_revision_store()."""
1243
return self._get_text_rev_store(repo_transport,
1249
def _get_text_store(self, transport, control_files):
1250
"""See RepositoryFormat._get_text_store()."""
1251
return self._get_versioned_file_store('weaves', transport, control_files)
1254
class MetaDirRepositoryFormat(RepositoryFormat):
1255
"""Common base class for the new repositories using the metadir layout."""
1258
super(MetaDirRepositoryFormat, self).__init__()
1259
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1261
def _create_control_files(self, a_bzrdir):
1262
"""Create the required files and the initial control_files object."""
1263
# FIXME: RBC 20060125 don't peek under the covers
1264
# NB: no need to escape relative paths that are url safe.
1265
repository_transport = a_bzrdir.get_repository_transport(self)
1266
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1267
control_files.create_lock()
1268
return control_files
1270
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1271
"""Upload the initial blank content."""
1272
control_files = self._create_control_files(a_bzrdir)
1273
control_files.lock_write()
1275
control_files._transport.mkdir_multi(dirs,
1276
mode=control_files._dir_mode)
1277
for file, content in files:
1278
control_files.put(file, content)
1279
for file, content in utf8_files:
1280
control_files.put_utf8(file, content)
1282
control_files.put_utf8('shared-storage', '')
1284
control_files.unlock()
1287
class RepositoryFormat7(MetaDirRepositoryFormat):
1288
"""Bzr repository 7.
1290
This repository format has:
1291
- weaves for file texts and inventory
1292
- hash subdirectory based stores.
1293
- TextStores for revisions and signatures.
1294
- a format marker of its own
1295
- an optional 'shared-storage' flag
1296
- an optional 'no-working-trees' flag
1299
def _get_control_store(self, repo_transport, control_files):
1300
"""Return the control store for this repository."""
1301
return self._get_versioned_file_store('',
1306
def get_format_string(self):
1307
"""See RepositoryFormat.get_format_string()."""
1308
return "Bazaar-NG Repository format 7"
1310
def get_format_description(self):
1311
"""See RepositoryFormat.get_format_description()."""
1312
return "Weave repository format 7"
1314
def _get_revision_store(self, repo_transport, control_files):
1315
"""See RepositoryFormat._get_revision_store()."""
1316
return self._get_text_rev_store(repo_transport,
1323
def _get_text_store(self, transport, control_files):
1324
"""See RepositoryFormat._get_text_store()."""
1325
return self._get_versioned_file_store('weaves',
1329
def initialize(self, a_bzrdir, shared=False):
1330
"""Create a weave repository.
1332
:param shared: If true the repository will be initialized as a shared
1335
from bzrlib.weavefile import write_weave_v5
1336
from bzrlib.weave import Weave
1338
# Create an empty weave
1340
write_weave_v5(Weave(), sio)
1341
empty_weave = sio.getvalue()
1343
mutter('creating repository in %s.', a_bzrdir.transport.base)
1344
dirs = ['revision-store', 'weaves']
1345
files = [('inventory.weave', StringIO(empty_weave)),
1347
utf8_files = [('format', self.get_format_string())]
1349
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1350
return self.open(a_bzrdir=a_bzrdir, _found=True)
1352
def open(self, a_bzrdir, _found=False, _override_transport=None):
1353
"""See RepositoryFormat.open().
1355
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1356
repository at a slightly different url
1357
than normal. I.e. during 'upgrade'.
1360
format = RepositoryFormat.find_format(a_bzrdir)
1361
assert format.__class__ == self.__class__
1362
if _override_transport is not None:
1363
repo_transport = _override_transport
1365
repo_transport = a_bzrdir.get_repository_transport(None)
1366
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1367
text_store = self._get_text_store(repo_transport, control_files)
1368
control_store = self._get_control_store(repo_transport, control_files)
1369
_revision_store = self._get_revision_store(repo_transport, control_files)
1370
return MetaDirRepository(_format=self,
1372
control_files=control_files,
1373
_revision_store=_revision_store,
1374
control_store=control_store,
1375
text_store=text_store)
1378
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1379
"""Bzr repository knit format 1.
1381
This repository format has:
1382
- knits for file texts and inventory
1383
- hash subdirectory based stores.
1384
- knits for revisions and signatures
1385
- TextStores for revisions and signatures.
1386
- a format marker of its own
1387
- an optional 'shared-storage' flag
1388
- an optional 'no-working-trees' flag
1391
This format was introduced in bzr 0.8.
1394
def _get_control_store(self, repo_transport, control_files):
1395
"""Return the control store for this repository."""
1396
return VersionedFileStore(
1399
file_mode=control_files._file_mode,
1400
versionedfile_class=KnitVersionedFile,
1401
versionedfile_kwargs={'factory':KnitPlainFactory()},
1404
def get_format_string(self):
1405
"""See RepositoryFormat.get_format_string()."""
1406
return "Bazaar-NG Knit Repository Format 1"
1408
def get_format_description(self):
1409
"""See RepositoryFormat.get_format_description()."""
1410
return "Knit repository format 1"
1412
def _get_revision_store(self, repo_transport, control_files):
1413
"""See RepositoryFormat._get_revision_store()."""
1414
from bzrlib.store.revision.knit import KnitRevisionStore
1415
versioned_file_store = VersionedFileStore(
1417
file_mode=control_files._file_mode,
1420
versionedfile_class=KnitVersionedFile,
1421
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1424
return KnitRevisionStore(versioned_file_store)
1426
def _get_text_store(self, transport, control_files):
1427
"""See RepositoryFormat._get_text_store()."""
1428
return self._get_versioned_file_store('knits',
1431
versionedfile_class=KnitVersionedFile,
1434
def initialize(self, a_bzrdir, shared=False):
1435
"""Create a knit format 1 repository.
1437
:param a_bzrdir: bzrdir to contain the new repository; must already
1439
:param shared: If true the repository will be initialized as a shared
1442
mutter('creating repository in %s.', a_bzrdir.transport.base)
1443
dirs = ['revision-store', 'knits']
1445
utf8_files = [('format', self.get_format_string())]
1447
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1448
repo_transport = a_bzrdir.get_repository_transport(None)
1449
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1450
control_store = self._get_control_store(repo_transport, control_files)
1451
transaction = transactions.WriteTransaction()
1452
# trigger a write of the inventory store.
1453
control_store.get_weave_or_empty('inventory', transaction)
1454
_revision_store = self._get_revision_store(repo_transport, control_files)
1455
_revision_store.has_revision_id('A', transaction)
1456
_revision_store.get_signature_file(transaction)
1457
return self.open(a_bzrdir=a_bzrdir, _found=True)
1459
def open(self, a_bzrdir, _found=False, _override_transport=None):
1460
"""See RepositoryFormat.open().
1462
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1463
repository at a slightly different url
1464
than normal. I.e. during 'upgrade'.
1467
format = RepositoryFormat.find_format(a_bzrdir)
1468
assert format.__class__ == self.__class__
1469
if _override_transport is not None:
1470
repo_transport = _override_transport
1472
repo_transport = a_bzrdir.get_repository_transport(None)
1473
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1474
text_store = self._get_text_store(repo_transport, control_files)
1475
control_store = self._get_control_store(repo_transport, control_files)
1476
_revision_store = self._get_revision_store(repo_transport, control_files)
1477
return KnitRepository(_format=self,
1479
control_files=control_files,
1480
_revision_store=_revision_store,
1481
control_store=control_store,
1482
text_store=text_store)
1485
# formats which have no format string are not discoverable
1486
# and not independently creatable, so are not registered.
1487
RepositoryFormat.register_format(RepositoryFormat7())
1488
_default_format = RepositoryFormatKnit1()
1489
RepositoryFormat.register_format(_default_format)
1490
RepositoryFormat.set_default_format(_default_format)
1491
_legacy_formats = [RepositoryFormat4(),
1492
RepositoryFormat5(),
1493
RepositoryFormat6()]
1496
class InterRepository(InterObject):
1497
"""This class represents operations taking place between two repositories.
1499
Its instances have methods like copy_content and fetch, and contain
1500
references to the source and target repositories these operations can be
1503
Often we will provide convenience methods on 'repository' which carry out
1504
operations with another repository - they will always forward to
1505
InterRepository.get(other).method_name(parameters).
1509
"""The available optimised InterRepository types."""
1512
def copy_content(self, revision_id=None, basis=None):
1513
"""Make a complete copy of the content in self into destination.
1515
This is a destructive operation! Do not use it on existing
1518
:param revision_id: Only copy the content needed to construct
1519
revision_id and its parents.
1520
:param basis: Copy the needed data preferentially from basis.
1523
self.target.set_make_working_trees(self.source.make_working_trees())
1524
except NotImplementedError:
1526
# grab the basis available data
1527
if basis is not None:
1528
self.target.fetch(basis, revision_id=revision_id)
1529
# but don't bother fetching if we have the needed data now.
1530
if (revision_id not in (None, NULL_REVISION) and
1531
self.target.has_revision(revision_id)):
1533
self.target.fetch(self.source, revision_id=revision_id)
1535
def _double_lock(self, lock_source, lock_target):
1536
"""Take out too locks, rolling back the first if the second throws."""
1541
# we want to ensure that we don't leave source locked by mistake.
1542
# and any error on target should not confuse source.
1543
self.source.unlock()
1547
def fetch(self, revision_id=None, pb=None):
1548
"""Fetch the content required to construct revision_id.
1550
The content is copied from source to target.
1552
:param revision_id: if None all content is copied, if NULL_REVISION no
1554
:param pb: optional progress bar to use for progress reports. If not
1555
provided a default one will be created.
1557
Returns the copied revision count and the failed revisions in a tuple:
1560
from bzrlib.fetch import GenericRepoFetcher
1561
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1562
self.source, self.source._format, self.target, self.target._format)
1563
f = GenericRepoFetcher(to_repository=self.target,
1564
from_repository=self.source,
1565
last_revision=revision_id,
1567
return f.count_copied, f.failed_revisions
1569
def lock_read(self):
1570
"""Take out a logical read lock.
1572
This will lock the source branch and the target branch. The source gets
1573
a read lock and the target a read lock.
1575
self._double_lock(self.source.lock_read, self.target.lock_read)
1577
def lock_write(self):
1578
"""Take out a logical write lock.
1580
This will lock the source branch and the target branch. The source gets
1581
a read lock and the target a write lock.
1583
self._double_lock(self.source.lock_read, self.target.lock_write)
1586
def missing_revision_ids(self, revision_id=None):
1587
"""Return the revision ids that source has that target does not.
1589
These are returned in topological order.
1591
:param revision_id: only return revision ids included by this
1594
# generic, possibly worst case, slow code path.
1595
target_ids = set(self.target.all_revision_ids())
1596
if revision_id is not None:
1597
source_ids = self.source.get_ancestry(revision_id)
1598
assert source_ids[0] == None
1601
source_ids = self.source.all_revision_ids()
1602
result_set = set(source_ids).difference(target_ids)
1603
# this may look like a no-op: its not. It preserves the ordering
1604
# other_ids had while only returning the members from other_ids
1605
# that we've decided we need.
1606
return [rev_id for rev_id in source_ids if rev_id in result_set]
1609
"""Release the locks on source and target."""
1611
self.target.unlock()
1613
self.source.unlock()
1616
class InterWeaveRepo(InterRepository):
1617
"""Optimised code paths between Weave based repositories."""
1619
_matching_repo_format = RepositoryFormat7()
1620
"""Repository format for testing with."""
1623
def is_compatible(source, target):
1624
"""Be compatible with known Weave formats.
1626
We don't test for the stores being of specific types because that
1627
could lead to confusing results, and there is no need to be
1631
return (isinstance(source._format, (RepositoryFormat5,
1633
RepositoryFormat7)) and
1634
isinstance(target._format, (RepositoryFormat5,
1636
RepositoryFormat7)))
1637
except AttributeError:
1641
def copy_content(self, revision_id=None, basis=None):
1642
"""See InterRepository.copy_content()."""
1643
# weave specific optimised path:
1644
if basis is not None:
1645
# copy the basis in, then fetch remaining data.
1646
basis.copy_content_into(self.target, revision_id)
1647
# the basis copy_content_into could miss-set this.
1649
self.target.set_make_working_trees(self.source.make_working_trees())
1650
except NotImplementedError:
1652
self.target.fetch(self.source, revision_id=revision_id)
1655
self.target.set_make_working_trees(self.source.make_working_trees())
1656
except NotImplementedError:
1658
# FIXME do not peek!
1659
if self.source.control_files._transport.listable():
1660
pb = ui.ui_factory.nested_progress_bar()
1662
self.target.weave_store.copy_all_ids(
1663
self.source.weave_store,
1665
from_transaction=self.source.get_transaction(),
1666
to_transaction=self.target.get_transaction())
1667
pb.update('copying inventory', 0, 1)
1668
self.target.control_weaves.copy_multi(
1669
self.source.control_weaves, ['inventory'],
1670
from_transaction=self.source.get_transaction(),
1671
to_transaction=self.target.get_transaction())
1672
self.target._revision_store.text_store.copy_all_ids(
1673
self.source._revision_store.text_store,
1678
self.target.fetch(self.source, revision_id=revision_id)
1681
def fetch(self, revision_id=None, pb=None):
1682
"""See InterRepository.fetch()."""
1683
from bzrlib.fetch import GenericRepoFetcher
1684
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1685
self.source, self.source._format, self.target, self.target._format)
1686
f = GenericRepoFetcher(to_repository=self.target,
1687
from_repository=self.source,
1688
last_revision=revision_id,
1690
return f.count_copied, f.failed_revisions
1693
def missing_revision_ids(self, revision_id=None):
1694
"""See InterRepository.missing_revision_ids()."""
1695
# we want all revisions to satisfy revision_id in source.
1696
# but we don't want to stat every file here and there.
1697
# we want then, all revisions other needs to satisfy revision_id
1698
# checked, but not those that we have locally.
1699
# so the first thing is to get a subset of the revisions to
1700
# satisfy revision_id in source, and then eliminate those that
1701
# we do already have.
1702
# this is slow on high latency connection to self, but as as this
1703
# disk format scales terribly for push anyway due to rewriting
1704
# inventory.weave, this is considered acceptable.
1706
if revision_id is not None:
1707
source_ids = self.source.get_ancestry(revision_id)
1708
assert source_ids[0] == None
1711
source_ids = self.source._all_possible_ids()
1712
source_ids_set = set(source_ids)
1713
# source_ids is the worst possible case we may need to pull.
1714
# now we want to filter source_ids against what we actually
1715
# have in target, but don't try to check for existence where we know
1716
# we do not have a revision as that would be pointless.
1717
target_ids = set(self.target._all_possible_ids())
1718
possibly_present_revisions = target_ids.intersection(source_ids_set)
1719
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1720
required_revisions = source_ids_set.difference(actually_present_revisions)
1721
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1722
if revision_id is not None:
1723
# we used get_ancestry to determine source_ids then we are assured all
1724
# revisions referenced are present as they are installed in topological order.
1725
# and the tip revision was validated by get_ancestry.
1726
return required_topo_revisions
1728
# if we just grabbed the possibly available ids, then
1729
# we only have an estimate of whats available and need to validate
1730
# that against the revision records.
1731
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1734
class InterKnitRepo(InterRepository):
1735
"""Optimised code paths between Knit based repositories."""
1737
_matching_repo_format = RepositoryFormatKnit1()
1738
"""Repository format for testing with."""
1741
def is_compatible(source, target):
1742
"""Be compatible with known Knit formats.
1744
We don't test for the stores being of specific types because that
1745
could lead to confusing results, and there is no need to be
1749
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1750
isinstance(target._format, (RepositoryFormatKnit1)))
1751
except AttributeError:
1755
def fetch(self, revision_id=None, pb=None):
1756
"""See InterRepository.fetch()."""
1757
from bzrlib.fetch import KnitRepoFetcher
1758
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1759
self.source, self.source._format, self.target, self.target._format)
1760
f = KnitRepoFetcher(to_repository=self.target,
1761
from_repository=self.source,
1762
last_revision=revision_id,
1764
return f.count_copied, f.failed_revisions
1767
def missing_revision_ids(self, revision_id=None):
1768
"""See InterRepository.missing_revision_ids()."""
1769
if revision_id is not None:
1770
source_ids = self.source.get_ancestry(revision_id)
1771
assert source_ids[0] == None
1774
source_ids = self.source._all_possible_ids()
1775
source_ids_set = set(source_ids)
1776
# source_ids is the worst possible case we may need to pull.
1777
# now we want to filter source_ids against what we actually
1778
# have in target, but don't try to check for existence where we know
1779
# we do not have a revision as that would be pointless.
1780
target_ids = set(self.target._all_possible_ids())
1781
possibly_present_revisions = target_ids.intersection(source_ids_set)
1782
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1783
required_revisions = source_ids_set.difference(actually_present_revisions)
1784
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1785
if revision_id is not None:
1786
# we used get_ancestry to determine source_ids then we are assured all
1787
# revisions referenced are present as they are installed in topological order.
1788
# and the tip revision was validated by get_ancestry.
1789
return required_topo_revisions
1791
# if we just grabbed the possibly available ids, then
1792
# we only have an estimate of whats available and need to validate
1793
# that against the revision records.
1794
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1796
InterRepository.register_optimiser(InterWeaveRepo)
1797
InterRepository.register_optimiser(InterKnitRepo)
1800
class RepositoryTestProviderAdapter(object):
1801
"""A tool to generate a suite testing multiple repository formats at once.
1803
This is done by copying the test once for each transport and injecting
1804
the transport_server, transport_readonly_server, and bzrdir_format and
1805
repository_format classes into each copy. Each copy is also given a new id()
1806
to make it easy to identify.
1809
def __init__(self, transport_server, transport_readonly_server, formats):
1810
self._transport_server = transport_server
1811
self._transport_readonly_server = transport_readonly_server
1812
self._formats = formats
1814
def adapt(self, test):
1815
result = TestSuite()
1816
for repository_format, bzrdir_format in self._formats:
1817
new_test = deepcopy(test)
1818
new_test.transport_server = self._transport_server
1819
new_test.transport_readonly_server = self._transport_readonly_server
1820
new_test.bzrdir_format = bzrdir_format
1821
new_test.repository_format = repository_format
1822
def make_new_test_id():
1823
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1824
return lambda: new_id
1825
new_test.id = make_new_test_id()
1826
result.addTest(new_test)
1830
class InterRepositoryTestProviderAdapter(object):
1831
"""A tool to generate a suite testing multiple inter repository formats.
1833
This is done by copying the test once for each interrepo provider and injecting
1834
the transport_server, transport_readonly_server, repository_format and
1835
repository_to_format classes into each copy.
1836
Each copy is also given a new id() to make it easy to identify.
1839
def __init__(self, transport_server, transport_readonly_server, formats):
1840
self._transport_server = transport_server
1841
self._transport_readonly_server = transport_readonly_server
1842
self._formats = formats
1844
def adapt(self, test):
1845
result = TestSuite()
1846
for interrepo_class, repository_format, repository_format_to in self._formats:
1847
new_test = deepcopy(test)
1848
new_test.transport_server = self._transport_server
1849
new_test.transport_readonly_server = self._transport_readonly_server
1850
new_test.interrepo_class = interrepo_class
1851
new_test.repository_format = repository_format
1852
new_test.repository_format_to = repository_format_to
1853
def make_new_test_id():
1854
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1855
return lambda: new_id
1856
new_test.id = make_new_test_id()
1857
result.addTest(new_test)
1861
def default_test_list():
1862
"""Generate the default list of interrepo permutations to test."""
1864
# test the default InterRepository between format 6 and the current
1866
# XXX: robertc 20060220 reinstate this when there are two supported
1867
# formats which do not have an optimal code path between them.
1868
result.append((InterRepository,
1869
RepositoryFormat6(),
1870
RepositoryFormatKnit1()))
1871
for optimiser in InterRepository._optimisers:
1872
result.append((optimiser,
1873
optimiser._matching_repo_format,
1874
optimiser._matching_repo_format
1876
# if there are specific combinations we want to use, we can add them
1881
class CopyConverter(object):
1882
"""A repository conversion tool which just performs a copy of the content.
1884
This is slow but quite reliable.
1887
def __init__(self, target_format):
1888
"""Create a CopyConverter.
1890
:param target_format: The format the resulting repository should be.
1892
self.target_format = target_format
1894
def convert(self, repo, pb):
1895
"""Perform the conversion of to_convert, giving feedback via pb.
1897
:param to_convert: The disk object to convert.
1898
:param pb: a progress bar to use for progress information.
1903
# this is only useful with metadir layouts - separated repo content.
1904
# trigger an assertion if not such
1905
repo._format.get_format_string()
1906
self.repo_dir = repo.bzrdir
1907
self.step('Moving repository to repository.backup')
1908
self.repo_dir.transport.move('repository', 'repository.backup')
1909
backup_transport = self.repo_dir.transport.clone('repository.backup')
1910
self.source_repo = repo._format.open(self.repo_dir,
1912
_override_transport=backup_transport)
1913
self.step('Creating new repository')
1914
converted = self.target_format.initialize(self.repo_dir,
1915
self.source_repo.is_shared())
1916
converted.lock_write()
1918
self.step('Copying content into repository.')
1919
self.source_repo.copy_content_into(converted)
1922
self.step('Deleting old repository content.')
1923
self.repo_dir.transport.delete_tree('repository.backup')
1924
self.pb.note('repository converted')
1926
def step(self, message):
1927
"""Update the pb by a step."""
1929
self.pb.update(message, self.count, self.total)
1932
class CommitBuilder(object):
1933
"""Provides an interface to build up a commit.
1935
This allows describing a tree to be committed without needing to
1936
know the internals of the format of the repository.
1938
def __init__(self, repository, parents, config, timestamp=None,
1939
timezone=None, committer=None, revprops=None,
1941
"""Initiate a CommitBuilder.
1943
:param repository: Repository to commit to.
1944
:param parents: Revision ids of the parents of the new revision.
1945
:param config: Configuration to use.
1946
:param timestamp: Optional timestamp recorded for commit.
1947
:param timezone: Optional timezone for timestamp.
1948
:param committer: Optional committer to set for commit.
1949
:param revprops: Optional dictionary of revision properties.
1950
:param revision_id: Optional revision id.
1952
self._config = config
1954
if committer is None:
1955
self._committer = self._config.username()
1957
assert isinstance(committer, basestring), type(committer)
1958
self._committer = committer
1960
self.new_inventory = Inventory()
1961
self._new_revision_id = revision_id
1962
self.parents = parents
1963
self.repository = repository
1966
if revprops is not None:
1967
self._revprops.update(revprops)
1969
if timestamp is None:
1970
self._timestamp = time.time()
1972
self._timestamp = long(timestamp)
1974
if timezone is None:
1975
self._timezone = local_time_offset()
1977
self._timezone = int(timezone)
1979
self._generate_revision_if_needed()
1981
def commit(self, message):
1982
"""Make the actual commit.
1984
:return: The revision id of the recorded revision.
1986
rev = Revision(timestamp=self._timestamp,
1987
timezone=self._timezone,
1988
committer=self._committer,
1990
inventory_sha1=self.inv_sha1,
1991
revision_id=self._new_revision_id,
1992
properties=self._revprops)
1993
rev.parent_ids = self.parents
1994
self.repository.add_revision(self._new_revision_id, rev,
1995
self.new_inventory, self._config)
1996
return self._new_revision_id
1998
def finish_inventory(self):
1999
"""Tell the builder that the inventory is finished."""
2000
self.new_inventory.revision_id = self._new_revision_id
2001
self.inv_sha1 = self.repository.add_inventory(
2002
self._new_revision_id,
2007
def _gen_revision_id(self):
2008
"""Return new revision-id."""
2009
s = '%s-%s-' % (self._config.user_email(),
2010
compact_date(self._timestamp))
2011
s += hexlify(rand_bytes(8))
2014
def _generate_revision_if_needed(self):
2015
"""Create a revision id if None was supplied.
2017
If the repository can not support user-specified revision ids
2018
they should override this function and raise UnsupportedOperation
2019
if _new_revision_id is not None.
2021
:raises: UnsupportedOperation
2023
if self._new_revision_id is None:
2024
self._new_revision_id = self._gen_revision_id()
2026
def record_entry_contents(self, ie, parent_invs, path, tree):
2027
"""Record the content of ie from tree into the commit if needed.
2029
:param ie: An inventory entry present in the commit.
2030
:param parent_invs: The inventories of the parent revisions of the
2032
:param path: The path the entry is at in the tree.
2033
:param tree: The tree which contains this entry and should be used to
2036
self.new_inventory.add(ie)
2038
# ie.revision is always None if the InventoryEntry is considered
2039
# for committing. ie.snapshot will record the correct revision
2040
# which may be the sole parent if it is untouched.
2041
if ie.revision is not None:
2043
previous_entries = ie.find_previous_heads(
2045
self.repository.weave_store,
2046
self.repository.get_transaction())
2047
# we are creating a new revision for ie in the history store
2049
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2051
def modified_directory(self, file_id, file_parents):
2052
"""Record the presence of a symbolic link.
2054
:param file_id: The file_id of the link to record.
2055
:param file_parents: The per-file parent revision ids.
2057
self._add_text_to_weave(file_id, [], file_parents.keys())
2059
def modified_file_text(self, file_id, file_parents,
2060
get_content_byte_lines, text_sha1=None,
2062
"""Record the text of file file_id
2064
:param file_id: The file_id of the file to record the text of.
2065
:param file_parents: The per-file parent revision ids.
2066
:param get_content_byte_lines: A callable which will return the byte
2068
:param text_sha1: Optional SHA1 of the file contents.
2069
:param text_size: Optional size of the file contents.
2071
mutter('storing text of file {%s} in revision {%s} into %r',
2072
file_id, self._new_revision_id, self.repository.weave_store)
2073
# special case to avoid diffing on renames or
2075
if (len(file_parents) == 1
2076
and text_sha1 == file_parents.values()[0].text_sha1
2077
and text_size == file_parents.values()[0].text_size):
2078
previous_ie = file_parents.values()[0]
2079
versionedfile = self.repository.weave_store.get_weave(file_id,
2080
self.repository.get_transaction())
2081
versionedfile.clone_text(self._new_revision_id,
2082
previous_ie.revision, file_parents.keys())
2083
return text_sha1, text_size
2085
new_lines = get_content_byte_lines()
2086
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2087
# should return the SHA1 and size
2088
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2089
return osutils.sha_strings(new_lines), \
2090
sum(map(len, new_lines))
2092
def modified_link(self, file_id, file_parents, link_target):
2093
"""Record the presence of a symbolic link.
2095
:param file_id: The file_id of the link to record.
2096
:param file_parents: The per-file parent revision ids.
2097
:param link_target: Target location of this link.
2099
self._add_text_to_weave(file_id, [], file_parents.keys())
2101
def _add_text_to_weave(self, file_id, new_lines, parents):
2102
versionedfile = self.repository.weave_store.get_weave_or_empty(
2103
file_id, self.repository.get_transaction())
2104
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2105
versionedfile.clear_cache()
2108
# Copied from xml.sax.saxutils
2109
def _unescape_xml(data):
2110
"""Unescape &, <, and > in a string of data.
2112
data = data.replace("<", "<")
2113
data = data.replace(">", ">")
2114
# must do ampersand last
2115
return data.replace("&", "&")