« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 1055
- compare with another revision
- download diff
- download tarball
- view history from revision 1055
- Committer: Martin Pool
- Date: 2005-08-12 15:41:44 UTC
- Revision ID: mbp@sourcefrog.net-20050812154144-bc98570a78b8f633
- merge in deferred revfile work
- files added:
- bzrlib/changeset.py
- patches
- plugins/changeset
- files removed:
- BRANCH.TODO
- bzrlib/doc
- bzrlib/doc/api
- bzrlib/export
- bzrlib/store
- bzrlib/store/revision
- bzrlib/store/versioned
- bzrlib/tests/blackbox
- bzrlib/tests/branch_implementations
- bzrlib/tests/bzrdir_implementations
- bzrlib/tests/interrepository_implementations
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/repository_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/tests/workingtree_implementations
- bzrlib/transport
- bzrlib/transport/http
- bzrlib/ui
- bzrlib/util
- bzrlib/util/configobj
- bzrlib/util/configobj/docs
- tools/doc_generate
- files renamed:
- tools/doc_generate/autodoc_man.py => bzr-man.py
- contrib/newinventory.py => bzrlib/newinventory.py
- bzrlib/tests/ => bzrlib/selftest/
- bzrlib/tests/blackbox/test_too_much.py => bzrlib/selftest/blackbox.py
- bzrlib/tests/test_plugins.py => bzrlib/selftest/plugins.py
- bzrlib/tests/branch_implementations/test_branch.py => bzrlib/selftest/testbranch.py
- bzrlib/tests/test_diff.py => bzrlib/selftest/testdiff.py
- bzrlib/tests/test_hashcache.py => bzrlib/selftest/testhashcache.py
- bzrlib/tests/test_inv.py => bzrlib/selftest/testinv.py
- bzrlib/tests/test_log.py => bzrlib/selftest/testlog.py
- bzrlib/tests/test_merge3.py => bzrlib/selftest/testmerge3.py
- bzrlib/tests/test_revision.py => bzrlib/selftest/testrevision.py
- bzrlib/tests/test_revisionnamespaces.py => bzrlib/selftest/testrevisionnamespaces.py
- bzrlib/tests/blackbox/test_status.py => bzrlib/selftest/teststatus.py
- bzrlib/tests/blackbox/test_versioning.py => bzrlib/selftest/versioning.py
- bzrlib/tests/test_whitebox.py => bzrlib/selftest/whitebox.py
- bzrlib/store/__init__.py => bzrlib/store.py
- bzrlib/xml_serializer.py => bzrlib/xml.py
- bzrlib/util/effbot/ => effbot/
- bzrlib/util/elementtree/ => elementtree/
- bzrlib/plugins/ => plugins/
- bzrlib/tests/test_weave.py => tools/testweave.py
- bzrlib/util/urlgrabber/ => urlgrabber/
- files modified:
- .bzrignore
- bzr *
- setup.py *
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
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.
7
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.
12
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
16
17
from copy import deepcopy
18
from cStringIO import StringIO
19
from unittest import TestSuite
20
21
import bzrlib.bzrdir as bzrdir
22
from bzrlib.decorators import needs_read_lock, needs_write_lock
23
import bzrlib.errors as errors
24
from bzrlib.errors import InvalidRevisionId
25
import bzrlib.gpg as gpg
26
from bzrlib.graph import Graph
27
from bzrlib.inter import InterObject
28
from bzrlib.knit import KnitVersionedFile
29
from bzrlib.lockable_files import LockableFiles, TransportLock
30
from bzrlib.lockdir import LockDir
31
from bzrlib.osutils import safe_unicode
32
from bzrlib.revision import NULL_REVISION
33
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
34
from bzrlib.store.text import TextStore
35
from bzrlib.symbol_versioning import *
36
from bzrlib.trace import mutter
37
from bzrlib.tree import RevisionTree
38
from bzrlib.tsort import topo_sort
39
from bzrlib.testament import Testament
40
from bzrlib.tree import EmptyTree
41
import bzrlib.ui
42
from bzrlib.weave import WeaveFile
43
import bzrlib.xml5
44
45
46
class Repository(object):
47
"""Repository holding history for one or more branches.
48
49
The repository holds and retrieves historical information including
50
revisions and file history. It's normally accessed only by the Branch,
51
which views a particular line of development through that history.
52
53
The Repository builds on top of Stores and a Transport, which respectively
54
describe the disk data format and the way of accessing the (possibly
55
remote) disk.
56
"""
57
58
@needs_write_lock
59
def add_inventory(self, revid, inv, parents):
60
"""Add the inventory inv to the repository as revid.
61
62
:param parents: The revision ids of the parents that revid
63
is known to have and are in the repository already.
64
65
returns the sha1 of the serialized inventory.
66
"""
67
inv_text = bzrlib.xml5.serializer_v5.write_inventory_to_string(inv)
68
inv_sha1 = bzrlib.osutils.sha_string(inv_text)
69
inv_vf = self.control_weaves.get_weave('inventory',
70
self.get_transaction())
71
inv_vf.add_lines(revid, parents, bzrlib.osutils.split_lines(inv_text))
72
return inv_sha1
73
74
@needs_write_lock
75
def add_revision(self, rev_id, rev, inv=None, config=None):
76
"""Add rev to the revision store as rev_id.
77
78
:param rev_id: the revision id to use.
79
:param rev: The revision object.
80
:param inv: The inventory for the revision. if None, it will be looked
81
up in the inventory storer
82
:param config: If None no digital signature will be created.
83
If supplied its signature_needed method will be used
84
to determine if a signature should be made.
85
"""
86
if config is not None and config.signature_needed():
87
if inv is None:
88
inv = self.get_inventory(rev_id)
89
plaintext = Testament(rev, inv).as_short_text()
90
self.store_revision_signature(
91
gpg.GPGStrategy(config), plaintext, rev_id)
92
if not rev_id in self.get_inventory_weave():
93
if inv is None:
94
raise errors.WeaveRevisionNotPresent(rev_id,
95
self.get_inventory_weave())
96
else:
97
# yes, this is not suitable for adding with ghosts.
98
self.add_inventory(rev_id, inv, rev.parent_ids)
99
self._revision_store.add_revision(rev, self.get_transaction())
100
101
@needs_read_lock
102
def _all_possible_ids(self):
103
"""Return all the possible revisions that we could find."""
104
return self.get_inventory_weave().versions()
105
106
@needs_read_lock
107
def all_revision_ids(self):
108
"""Returns a list of all the revision ids in the repository.
109
110
These are in as much topological order as the underlying store can
111
present: for weaves ghosts may lead to a lack of correctness until
112
the reweave updates the parents list.
113
"""
114
if self._revision_store.text_store.listable():
115
return self._revision_store.all_revision_ids(self.get_transaction())
116
result = self._all_possible_ids()
117
return self._eliminate_revisions_not_present(result)
118
119
@needs_read_lock
120
def _eliminate_revisions_not_present(self, revision_ids):
121
"""Check every revision id in revision_ids to see if we have it.
122
123
Returns a set of the present revisions.
124
"""
125
result = []
126
for id in revision_ids:
127
if self.has_revision(id):
128
result.append(id)
129
return result
130
131
@staticmethod
132
def create(a_bzrdir):
133
"""Construct the current default format repository in a_bzrdir."""
134
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
135
136
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
137
"""instantiate a Repository.
138
139
:param _format: The format of the repository on disk.
140
:param a_bzrdir: The BzrDir of the repository.
141
142
In the future we will have a single api for all stores for
143
getting file texts, inventories and revisions, then
144
this construct will accept instances of those things.
145
"""
146
object.__init__(self)
147
self._format = _format
148
# the following are part of the public API for Repository:
149
self.bzrdir = a_bzrdir
150
self.control_files = control_files
151
self._revision_store = _revision_store
152
self.text_store = text_store
153
# backwards compatability
154
self.weave_store = text_store
155
# not right yet - should be more semantically clear ?
156
#
157
self.control_store = control_store
158
self.control_weaves = control_store
159
160
def lock_write(self):
161
self.control_files.lock_write()
162
163
def lock_read(self):
164
self.control_files.lock_read()
165
166
def is_locked(self):
167
return self.control_files.is_locked()
168
169
@needs_read_lock
170
def missing_revision_ids(self, other, revision_id=None):
171
"""Return the revision ids that other has that this does not.
172
173
These are returned in topological order.
174
175
revision_id: only return revision ids included by revision_id.
176
"""
177
return InterRepository.get(other, self).missing_revision_ids(revision_id)
178
179
@staticmethod
180
def open(base):
181
"""Open the repository rooted at base.
182
183
For instance, if the repository is at URL/.bzr/repository,
184
Repository.open(URL) -> a Repository instance.
185
"""
186
control = bzrlib.bzrdir.BzrDir.open(base)
187
return control.open_repository()
188
189
def copy_content_into(self, destination, revision_id=None, basis=None):
190
"""Make a complete copy of the content in self into destination.
191
192
This is a destructive operation! Do not use it on existing
193
repositories.
194
"""
195
return InterRepository.get(self, destination).copy_content(revision_id, basis)
196
197
def fetch(self, source, revision_id=None, pb=None):
198
"""Fetch the content required to construct revision_id from source.
199
200
If revision_id is None all content is copied.
201
"""
202
return InterRepository.get(source, self).fetch(revision_id=revision_id,
203
pb=pb)
204
205
def unlock(self):
206
self.control_files.unlock()
207
208
@needs_read_lock
209
def clone(self, a_bzrdir, revision_id=None, basis=None):
210
"""Clone this repository into a_bzrdir using the current format.
211
212
Currently no check is made that the format of this repository and
213
the bzrdir format are compatible. FIXME RBC 20060201.
214
"""
215
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
216
# use target default format.
217
result = a_bzrdir.create_repository()
218
# FIXME RBC 20060209 split out the repository type to avoid this check ?
219
elif isinstance(a_bzrdir._format,
220
(bzrlib.bzrdir.BzrDirFormat4,
221
bzrlib.bzrdir.BzrDirFormat5,
222
bzrlib.bzrdir.BzrDirFormat6)):
223
result = a_bzrdir.open_repository()
224
else:
225
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
226
self.copy_content_into(result, revision_id, basis)
227
return result
228
229
@needs_read_lock
230
def has_revision(self, revision_id):
231
"""True if this repository has a copy of the revision."""
232
return self._revision_store.has_revision_id(revision_id,
233
self.get_transaction())
234
235
@needs_read_lock
236
def get_revision_reconcile(self, revision_id):
237
"""'reconcile' helper routine that allows access to a revision always.
238
239
This variant of get_revision does not cross check the weave graph
240
against the revision one as get_revision does: but it should only
241
be used by reconcile, or reconcile-alike commands that are correcting
242
or testing the revision graph.
243
"""
244
if not revision_id or not isinstance(revision_id, basestring):
245
raise InvalidRevisionId(revision_id=revision_id, branch=self)
246
return self._revision_store.get_revision(revision_id,
247
self.get_transaction())
248
249
@needs_read_lock
250
def get_revision_xml(self, revision_id):
251
rev = self.get_revision(revision_id)
252
rev_tmp = StringIO()
253
# the current serializer..
254
self._revision_store._serializer.write_revision(rev, rev_tmp)
255
rev_tmp.seek(0)
256
return rev_tmp.getvalue()
257
258
@needs_read_lock
259
def get_revision(self, revision_id):
260
"""Return the Revision object for a named revision"""
261
r = self.get_revision_reconcile(revision_id)
262
# weave corruption can lead to absent revision markers that should be
263
# present.
264
# the following test is reasonably cheap (it needs a single weave read)
265
# and the weave is cached in read transactions. In write transactions
266
# it is not cached but typically we only read a small number of
267
# revisions. For knits when they are introduced we will probably want
268
# to ensure that caching write transactions are in use.
269
inv = self.get_inventory_weave()
270
self._check_revision_parents(r, inv)
271
return r
272
273
def _check_revision_parents(self, revision, inventory):
274
"""Private to Repository and Fetch.
275
276
This checks the parentage of revision in an inventory weave for
277
consistency and is only applicable to inventory-weave-for-ancestry
278
using repository formats & fetchers.
279
"""
280
weave_parents = inventory.get_parents(revision.revision_id)
281
weave_names = inventory.versions()
282
for parent_id in revision.parent_ids:
283
if parent_id in weave_names:
284
# this parent must not be a ghost.
285
if not parent_id in weave_parents:
286
# but it is a ghost
287
raise errors.CorruptRepository(self)
288
289
@needs_write_lock
290
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
291
signature = gpg_strategy.sign(plaintext)
292
self._revision_store.add_revision_signature_text(revision_id,
293
signature,
294
self.get_transaction())
295
296
def fileid_involved_between_revs(self, from_revid, to_revid):
297
"""Find file_id(s) which are involved in the changes between revisions.
298
299
This determines the set of revisions which are involved, and then
300
finds all file ids affected by those revisions.
301
"""
302
w = self.get_inventory_weave()
303
from_set = set(w.get_ancestry(from_revid))
304
to_set = set(w.get_ancestry(to_revid))
305
changed = to_set.difference(from_set)
306
return self._fileid_involved_by_set(changed)
307
308
def fileid_involved(self, last_revid=None):
309
"""Find all file_ids modified in the ancestry of last_revid.
310
311
:param last_revid: If None, last_revision() will be used.
312
"""
313
w = self.get_inventory_weave()
314
if not last_revid:
315
changed = set(w.versions())
316
else:
317
changed = set(w.get_ancestry(last_revid))
318
return self._fileid_involved_by_set(changed)
319
320
def fileid_involved_by_set(self, changes):
321
"""Find all file_ids modified by the set of revisions passed in.
322
323
:param changes: A set() of revision ids
324
"""
325
# TODO: jam 20060119 This line does *nothing*, remove it.
326
# or better yet, change _fileid_involved_by_set so
327
# that it takes the inventory weave, rather than
328
# pulling it out by itself.
329
return self._fileid_involved_by_set(changes)
330
331
def _fileid_involved_by_set(self, changes):
332
"""Find the set of file-ids affected by the set of revisions.
333
334
:param changes: A set() of revision ids.
335
:return: A set() of file ids.
336
337
This peaks at the Weave, interpreting each line, looking to
338
see if it mentions one of the revisions. And if so, includes
339
the file id mentioned.
340
This expects both the Weave format, and the serialization
341
to have a single line per file/directory, and to have
342
fileid="" and revision="" on that line.
343
"""
344
assert isinstance(self._format, (RepositoryFormat5,
345
RepositoryFormat6,
346
RepositoryFormat7,
347
RepositoryFormatKnit1)), \
348
"fileid_involved only supported for branches which store inventory as unnested xml"
349
350
w = self.get_inventory_weave()
351
file_ids = set()
352
353
# this code needs to read every line in every inventory for the
354
# inventories [changes]. Seeing a line twice is ok. Seeing a line
355
# not pesent in one of those inventories is unnecessary and not
356
# harmful because we are filtering by the revision id marker in the
357
# inventory lines to only select file ids altered in one of those
358
# revisions. We dont need to see all lines in the inventory because
359
# only those added in an inventory in rev X can contain a revision=X
360
# line.
361
for line in w.iter_lines_added_or_present_in_versions(changes):
362
start = line.find('file_id="')+9
363
if start < 9: continue
364
end = line.find('"', start)
365
assert end>= 0
366
file_id = _unescape_xml(line[start:end])
367
368
# check if file_id is already present
369
if file_id in file_ids: continue
370
371
start = line.find('revision="')+10
372
if start < 10: continue
373
end = line.find('"', start)
374
assert end>= 0
375
revision_id = _unescape_xml(line[start:end])
376
if revision_id in changes:
377
file_ids.add(file_id)
378
return file_ids
379
380
@needs_read_lock
381
def get_inventory_weave(self):
382
return self.control_weaves.get_weave('inventory',
383
self.get_transaction())
384
385
@needs_read_lock
386
def get_inventory(self, revision_id):
387
"""Get Inventory object by hash."""
388
xml = self.get_inventory_xml(revision_id)
389
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
390
391
@needs_read_lock
392
def get_inventory_xml(self, revision_id):
393
"""Get inventory XML as a file object."""
394
try:
395
assert isinstance(revision_id, basestring), type(revision_id)
396
iw = self.get_inventory_weave()
397
return iw.get_text(revision_id)
398
except IndexError:
399
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
400
401
@needs_read_lock
402
def get_inventory_sha1(self, revision_id):
403
"""Return the sha1 hash of the inventory entry
404
"""
405
return self.get_revision(revision_id).inventory_sha1
406
407
@needs_read_lock
408
def get_revision_graph(self, revision_id=None):
409
"""Return a dictionary containing the revision graph.
410
411
:return: a dictionary of revision_id->revision_parents_list.
412
"""
413
weave = self.get_inventory_weave()
414
all_revisions = self._eliminate_revisions_not_present(weave.versions())
415
entire_graph = dict([(node, weave.get_parents(node)) for
416
node in all_revisions])
417
if revision_id is None:
418
return entire_graph
419
elif revision_id not in entire_graph:
420
raise errors.NoSuchRevision(self, revision_id)
421
else:
422
# add what can be reached from revision_id
423
result = {}
424
pending = set([revision_id])
425
while len(pending) > 0:
426
node = pending.pop()
427
result[node] = entire_graph[node]
428
for revision_id in result[node]:
429
if revision_id not in result:
430
pending.add(revision_id)
431
return result
432
433
@needs_read_lock
434
def get_revision_graph_with_ghosts(self, revision_ids=None):
435
"""Return a graph of the revisions with ghosts marked as applicable.
436
437
:param revision_ids: an iterable of revisions to graph or None for all.
438
:return: a Graph object with the graph reachable from revision_ids.
439
"""
440
result = Graph()
441
if not revision_ids:
442
pending = set(self.all_revision_ids())
443
required = set([])
444
else:
445
pending = set(revision_ids)
446
required = set(revision_ids)
447
done = set([])
448
while len(pending):
449
revision_id = pending.pop()
450
try:
451
rev = self.get_revision(revision_id)
452
except errors.NoSuchRevision:
453
if revision_id in required:
454
raise
455
# a ghost
456
result.add_ghost(revision_id)
457
continue
458
for parent_id in rev.parent_ids:
459
# is this queued or done ?
460
if (parent_id not in pending and
461
parent_id not in done):
462
# no, queue it.
463
pending.add(parent_id)
464
result.add_node(revision_id, rev.parent_ids)
465
done.add(revision_id)
466
return result
467
468
@needs_read_lock
469
def get_revision_inventory(self, revision_id):
470
"""Return inventory of a past revision."""
471
# TODO: Unify this with get_inventory()
472
# bzr 0.0.6 and later imposes the constraint that the inventory_id
473
# must be the same as its revision, so this is trivial.
474
if revision_id is None:
475
# This does not make sense: if there is no revision,
476
# then it is the current tree inventory surely ?!
477
# and thus get_root_id() is something that looks at the last
478
# commit on the branch, and the get_root_id is an inventory check.
479
raise NotImplementedError
480
# return Inventory(self.get_root_id())
481
else:
482
return self.get_inventory(revision_id)
483
484
@needs_read_lock
485
def is_shared(self):
486
"""Return True if this repository is flagged as a shared repository."""
487
raise NotImplementedError(self.is_shared)
488
489
@needs_write_lock
490
def reconcile(self):
491
"""Reconcile this repository."""
492
from bzrlib.reconcile import RepoReconciler
493
reconciler = RepoReconciler(self)
494
reconciler.reconcile()
495
return reconciler
496
497
@needs_read_lock
498
def revision_tree(self, revision_id):
499
"""Return Tree for a revision on this branch.
500
501
`revision_id` may be None for the null revision, in which case
502
an `EmptyTree` is returned."""
503
# TODO: refactor this to use an existing revision object
504
# so we don't need to read it in twice.
505
if revision_id is None or revision_id == NULL_REVISION:
506
return EmptyTree()
507
else:
508
inv = self.get_revision_inventory(revision_id)
509
return RevisionTree(self, inv, revision_id)
510
511
@needs_read_lock
512
def get_ancestry(self, revision_id):
513
"""Return a list of revision-ids integrated by a revision.
514
515
This is topologically sorted.
516
"""
517
if revision_id is None:
518
return [None]
519
if not self.has_revision(revision_id):
520
raise errors.NoSuchRevision(self, revision_id)
521
w = self.get_inventory_weave()
522
candidates = w.get_ancestry(revision_id)
523
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
524
525
@needs_read_lock
526
def print_file(self, file, revision_id):
527
"""Print `file` to stdout.
528
529
FIXME RBC 20060125 as John Meinel points out this is a bad api
530
- it writes to stdout, it assumes that that is valid etc. Fix
531
by creating a new more flexible convenience function.
532
"""
533
tree = self.revision_tree(revision_id)
534
# use inventory as it was in that revision
535
file_id = tree.inventory.path2id(file)
536
if not file_id:
537
raise BzrError("%r is not present in revision %s" % (file, revno))
538
try:
539
revno = self.revision_id_to_revno(revision_id)
540
except errors.NoSuchRevision:
541
# TODO: This should not be BzrError,
542
# but NoSuchFile doesn't fit either
543
raise BzrError('%r is not present in revision %s'
544
% (file, revision_id))
545
else:
546
raise BzrError('%r is not present in revision %s'
547
% (file, revno))
548
tree.print_file(file_id)
549
550
def get_transaction(self):
551
return self.control_files.get_transaction()
552
553
def revision_parents(self, revid):
554
return self.get_inventory_weave().parent_names(revid)
555
556
@needs_write_lock
557
def set_make_working_trees(self, new_value):
558
"""Set the policy flag for making working trees when creating branches.
559
560
This only applies to branches that use this repository.
561
562
The default is 'True'.
563
:param new_value: True to restore the default, False to disable making
564
working trees.
565
"""
566
raise NotImplementedError(self.set_make_working_trees)
567
568
def make_working_trees(self):
569
"""Returns the policy for making working trees on new branches."""
570
raise NotImplementedError(self.make_working_trees)
571
572
@needs_write_lock
573
def sign_revision(self, revision_id, gpg_strategy):
574
plaintext = Testament.from_revision(self, revision_id).as_short_text()
575
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
576
577
@needs_read_lock
578
def has_signature_for_revision_id(self, revision_id):
579
"""Query for a revision signature for revision_id in the repository."""
580
return self._revision_store.has_signature(revision_id,
581
self.get_transaction())
582
583
@needs_read_lock
584
def get_signature_text(self, revision_id):
585
"""Return the text for a signature."""
586
return self._revision_store.get_signature_text(revision_id,
587
self.get_transaction())
588
589
590
class AllInOneRepository(Repository):
591
"""Legacy support - the repository behaviour for all-in-one branches."""
592
593
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
594
# we reuse one control files instance.
595
dir_mode = a_bzrdir._control_files._dir_mode
596
file_mode = a_bzrdir._control_files._file_mode
597
598
def get_store(name, compressed=True, prefixed=False):
599
# FIXME: This approach of assuming stores are all entirely compressed
600
# or entirely uncompressed is tidy, but breaks upgrade from
601
# some existing branches where there's a mixture; we probably
602
# still want the option to look for both.
603
relpath = a_bzrdir._control_files._escape(name)
604
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
605
prefixed=prefixed, compressed=compressed,
606
dir_mode=dir_mode,
607
file_mode=file_mode)
608
#if self._transport.should_cache():
609
# cache_path = os.path.join(self.cache_root, name)
610
# os.mkdir(cache_path)
611
# store = bzrlib.store.CachedStore(store, cache_path)
612
return store
613
614
# not broken out yet because the controlweaves|inventory_store
615
# and text_store | weave_store bits are still different.
616
if isinstance(_format, RepositoryFormat4):
617
# cannot remove these - there is still no consistent api
618
# which allows access to this old info.
619
self.inventory_store = get_store('inventory-store')
620
text_store = get_store('text-store')
621
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
622
623
@needs_read_lock
624
def is_shared(self):
625
"""AllInOne repositories cannot be shared."""
626
return False
627
628
@needs_write_lock
629
def set_make_working_trees(self, new_value):
630
"""Set the policy flag for making working trees when creating branches.
631
632
This only applies to branches that use this repository.
633
634
The default is 'True'.
635
:param new_value: True to restore the default, False to disable making
636
working trees.
637
"""
638
raise NotImplementedError(self.set_make_working_trees)
639
640
def make_working_trees(self):
641
"""Returns the policy for making working trees on new branches."""
642
return True
643
644
645
class MetaDirRepository(Repository):
646
"""Repositories in the new meta-dir layout."""
647
648
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
649
super(MetaDirRepository, self).__init__(_format,
650
a_bzrdir,
651
control_files,
652
_revision_store,
653
control_store,
654
text_store)
655
656
dir_mode = self.control_files._dir_mode
657
file_mode = self.control_files._file_mode
658
659
@needs_read_lock
660
def is_shared(self):
661
"""Return True if this repository is flagged as a shared repository."""
662
return self.control_files._transport.has('shared-storage')
663
664
@needs_write_lock
665
def set_make_working_trees(self, new_value):
666
"""Set the policy flag for making working trees when creating branches.
667
668
This only applies to branches that use this repository.
669
670
The default is 'True'.
671
:param new_value: True to restore the default, False to disable making
672
working trees.
673
"""
674
if new_value:
675
try:
676
self.control_files._transport.delete('no-working-trees')
677
except errors.NoSuchFile:
678
pass
679
else:
680
self.control_files.put_utf8('no-working-trees', '')
681
682
def make_working_trees(self):
683
"""Returns the policy for making working trees on new branches."""
684
return not self.control_files._transport.has('no-working-trees')
685
686
687
class KnitRepository(MetaDirRepository):
688
"""Knit format repository."""
689
690
@needs_read_lock
691
def all_revision_ids(self):
692
"""See Repository.all_revision_ids()."""
693
return self._revision_store.all_revision_ids(self.get_transaction())
694
695
def fileid_involved_between_revs(self, from_revid, to_revid):
696
"""Find file_id(s) which are involved in the changes between revisions.
697
698
This determines the set of revisions which are involved, and then
699
finds all file ids affected by those revisions.
700
"""
701
vf = self._get_revision_vf()
702
from_set = set(vf.get_ancestry(from_revid))
703
to_set = set(vf.get_ancestry(to_revid))
704
changed = to_set.difference(from_set)
705
return self._fileid_involved_by_set(changed)
706
707
def fileid_involved(self, last_revid=None):
708
"""Find all file_ids modified in the ancestry of last_revid.
709
710
:param last_revid: If None, last_revision() will be used.
711
"""
712
if not last_revid:
713
changed = set(self.all_revision_ids())
714
else:
715
changed = set(self.get_ancestry(last_revid))
716
if None in changed:
717
changed.remove(None)
718
return self._fileid_involved_by_set(changed)
719
720
@needs_read_lock
721
def get_ancestry(self, revision_id):
722
"""Return a list of revision-ids integrated by a revision.
723
724
This is topologically sorted.
725
"""
726
if revision_id is None:
727
return [None]
728
vf = self._get_revision_vf()
729
try:
730
return [None] + vf.get_ancestry(revision_id)
731
except errors.RevisionNotPresent:
732
raise errors.NoSuchRevision(self, revision_id)
733
734
@needs_read_lock
735
def get_revision(self, revision_id):
736
"""Return the Revision object for a named revision"""
737
return self.get_revision_reconcile(revision_id)
738
739
@needs_read_lock
740
def get_revision_graph(self, revision_id=None):
741
"""Return a dictionary containing the revision graph.
742
743
:return: a dictionary of revision_id->revision_parents_list.
744
"""
745
weave = self._get_revision_vf()
746
entire_graph = weave.get_graph()
747
if revision_id is None:
748
return weave.get_graph()
749
elif revision_id not in weave:
750
raise errors.NoSuchRevision(self, revision_id)
751
else:
752
# add what can be reached from revision_id
753
result = {}
754
pending = set([revision_id])
755
while len(pending) > 0:
756
node = pending.pop()
757
result[node] = weave.get_parents(node)
758
for revision_id in result[node]:
759
if revision_id not in result:
760
pending.add(revision_id)
761
return result
762
763
@needs_read_lock
764
def get_revision_graph_with_ghosts(self, revision_ids=None):
765
"""Return a graph of the revisions with ghosts marked as applicable.
766
767
:param revision_ids: an iterable of revisions to graph or None for all.
768
:return: a Graph object with the graph reachable from revision_ids.
769
"""
770
result = Graph()
771
vf = self._get_revision_vf()
772
versions = vf.versions()
773
if not revision_ids:
774
pending = set(self.all_revision_ids())
775
required = set([])
776
else:
777
pending = set(revision_ids)
778
required = set(revision_ids)
779
done = set([])
780
while len(pending):
781
revision_id = pending.pop()
782
if not revision_id in versions:
783
if revision_id in required:
784
raise errors.NoSuchRevision(self, revision_id)
785
# a ghost
786
result.add_ghost(revision_id)
787
continue
788
parent_ids = vf.get_parents_with_ghosts(revision_id)
789
for parent_id in parent_ids:
790
# is this queued or done ?
791
if (parent_id not in pending and
792
parent_id not in done):
793
# no, queue it.
794
pending.add(parent_id)
795
result.add_node(revision_id, parent_ids)
796
done.add(result)
797
return result
798
799
def _get_revision_vf(self):
800
""":return: a versioned file containing the revisions."""
801
vf = self._revision_store.get_revision_file(self.get_transaction())
802
return vf
803
804
@needs_write_lock
805
def reconcile(self):
806
"""Reconcile this repository."""
807
from bzrlib.reconcile import KnitReconciler
808
reconciler = KnitReconciler(self)
809
reconciler.reconcile()
810
return reconciler
811
812
def revision_parents(self, revid):
813
return self._get_revision_vf().get_parents(rev_id)
814
815
class RepositoryFormat(object):
816
"""A repository format.
817
818
Formats provide three things:
819
* An initialization routine to construct repository data on disk.
820
* a format string which is used when the BzrDir supports versioned
821
children.
822
* an open routine which returns a Repository instance.
823
824
Formats are placed in an dict by their format string for reference
825
during opening. These should be subclasses of RepositoryFormat
826
for consistency.
827
828
Once a format is deprecated, just deprecate the initialize and open
829
methods on the format class. Do not deprecate the object, as the
830
object will be created every system load.
831
832
Common instance attributes:
833
_matchingbzrdir - the bzrdir format that the repository format was
834
originally written to work with. This can be used if manually
835
constructing a bzrdir and repository, or more commonly for test suite
836
parameterisation.
837
"""
838
839
_default_format = None
840
"""The default format used for new repositories."""
841
842
_formats = {}
843
"""The known formats."""
844
845
@classmethod
846
def find_format(klass, a_bzrdir):
847
"""Return the format for the repository object in a_bzrdir."""
848
try:
849
transport = a_bzrdir.get_repository_transport(None)
850
format_string = transport.get("format").read()
851
return klass._formats[format_string]
852
except errors.NoSuchFile:
853
raise errors.NoRepositoryPresent(a_bzrdir)
854
except KeyError:
855
raise errors.UnknownFormatError(format_string)
856
857
def _get_control_store(self, repo_transport, control_files):
858
"""Return the control store for this repository."""
859
raise NotImplementedError(self._get_control_store)
860
861
@classmethod
862
def get_default_format(klass):
863
"""Return the current default format."""
864
return klass._default_format
865
866
def get_format_string(self):
867
"""Return the ASCII format string that identifies this format.
868
869
Note that in pre format ?? repositories the format string is
870
not permitted nor written to disk.
871
"""
872
raise NotImplementedError(self.get_format_string)
873
874
def _get_revision_store(self, repo_transport, control_files):
875
"""Return the revision store object for this a_bzrdir."""
876
raise NotImplementedError(self._get_revision_store)
877
878
def _get_text_rev_store(self,
879
transport,
880
control_files,
881
name,
882
compressed=True,
883
prefixed=False,
884
serializer=None):
885
"""Common logic for getting a revision store for a repository.
886
887
see self._get_revision_store for the subclass-overridable method to
888
get the store for a repository.
889
"""
890
from bzrlib.store.revision.text import TextRevisionStore
891
dir_mode = control_files._dir_mode
892
file_mode = control_files._file_mode
893
text_store =TextStore(transport.clone(name),
894
prefixed=prefixed,
895
compressed=compressed,
896
dir_mode=dir_mode,
897
file_mode=file_mode)
898
_revision_store = TextRevisionStore(text_store, serializer)
899
return _revision_store
900
901
def _get_versioned_file_store(self,
902
name,
903
transport,
904
control_files,
905
prefixed=True,
906
versionedfile_class=WeaveFile):
907
weave_transport = control_files._transport.clone(name)
908
dir_mode = control_files._dir_mode
909
file_mode = control_files._file_mode
910
return VersionedFileStore(weave_transport, prefixed=prefixed,
911
dir_mode=dir_mode,
912
file_mode=file_mode,
913
versionedfile_class=versionedfile_class)
914
915
def initialize(self, a_bzrdir, shared=False):
916
"""Initialize a repository of this format in a_bzrdir.
917
918
:param a_bzrdir: The bzrdir to put the new repository in it.
919
:param shared: The repository should be initialized as a sharable one.
920
921
This may raise UninitializableFormat if shared repository are not
922
compatible the a_bzrdir.
923
"""
924
925
def is_supported(self):
926
"""Is this format supported?
927
928
Supported formats must be initializable and openable.
929
Unsupported formats may not support initialization or committing or
930
some other features depending on the reason for not being supported.
931
"""
932
return True
933
934
def open(self, a_bzrdir, _found=False):
935
"""Return an instance of this format for the bzrdir a_bzrdir.
936
937
_found is a private parameter, do not use it.
938
"""
939
raise NotImplementedError(self.open)
940
941
@classmethod
942
def register_format(klass, format):
943
klass._formats[format.get_format_string()] = format
944
945
@classmethod
946
def set_default_format(klass, format):
947
klass._default_format = format
948
949
@classmethod
950
def unregister_format(klass, format):
951
assert klass._formats[format.get_format_string()] is format
952
del klass._formats[format.get_format_string()]
953
954
955
class PreSplitOutRepositoryFormat(RepositoryFormat):
956
"""Base class for the pre split out repository formats."""
957
958
def initialize(self, a_bzrdir, shared=False, _internal=False):
959
"""Create a weave repository.
960
961
TODO: when creating split out bzr branch formats, move this to a common
962
base for Format5, Format6. or something like that.
963
"""
964
from bzrlib.weavefile import write_weave_v5
965
from bzrlib.weave import Weave
966
967
if shared:
968
raise errors.IncompatibleFormat(self, a_bzrdir._format)
969
970
if not _internal:
971
# always initialized when the bzrdir is.
972
return self.open(a_bzrdir, _found=True)
973
974
# Create an empty weave
975
sio = StringIO()
976
bzrlib.weavefile.write_weave_v5(Weave(), sio)
977
empty_weave = sio.getvalue()
978
979
mutter('creating repository in %s.', a_bzrdir.transport.base)
980
dirs = ['revision-store', 'weaves']
981
files = [('inventory.weave', StringIO(empty_weave)),
982
]
983
984
# FIXME: RBC 20060125 dont peek under the covers
985
# NB: no need to escape relative paths that are url safe.
986
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
987
TransportLock)
988
control_files.create_lock()
989
control_files.lock_write()
990
control_files._transport.mkdir_multi(dirs,
991
mode=control_files._dir_mode)
992
try:
993
for file, content in files:
994
control_files.put(file, content)
995
finally:
996
control_files.unlock()
997
return self.open(a_bzrdir, _found=True)
998
999
def _get_control_store(self, repo_transport, control_files):
1000
"""Return the control store for this repository."""
1001
return self._get_versioned_file_store('',
1002
repo_transport,
1003
control_files,
1004
prefixed=False)
1005
1006
def _get_text_store(self, transport, control_files):
1007
"""Get a store for file texts for this format."""
1008
raise NotImplementedError(self._get_text_store)
1009
1010
def open(self, a_bzrdir, _found=False):
1011
"""See RepositoryFormat.open()."""
1012
if not _found:
1013
# we are being called directly and must probe.
1014
raise NotImplementedError
1015
1016
repo_transport = a_bzrdir.get_repository_transport(None)
1017
control_files = a_bzrdir._control_files
1018
text_store = self._get_text_store(repo_transport, control_files)
1019
control_store = self._get_control_store(repo_transport, control_files)
1020
_revision_store = self._get_revision_store(repo_transport, control_files)
1021
return AllInOneRepository(_format=self,
1022
a_bzrdir=a_bzrdir,
1023
_revision_store=_revision_store,
1024
control_store=control_store,
1025
text_store=text_store)
1026
1027
1028
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1029
"""Bzr repository format 4.
1030
1031
This repository format has:
1032
- flat stores
1033
- TextStores for texts, inventories,revisions.
1034
1035
This format is deprecated: it indexes texts using a text id which is
1036
removed in format 5; initializationa and write support for this format
1037
has been removed.
1038
"""
1039
1040
def __init__(self):
1041
super(RepositoryFormat4, self).__init__()
1042
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1043
1044
def initialize(self, url, shared=False, _internal=False):
1045
"""Format 4 branches cannot be created."""
1046
raise errors.UninitializableFormat(self)
1047
1048
def is_supported(self):
1049
"""Format 4 is not supported.
1050
1051
It is not supported because the model changed from 4 to 5 and the
1052
conversion logic is expensive - so doing it on the fly was not
1053
feasible.
1054
"""
1055
return False
1056
1057
def _get_control_store(self, repo_transport, control_files):
1058
"""Format 4 repositories have no formal control store at this point.
1059
1060
This will cause any control-file-needing apis to fail - this is desired.
1061
"""
1062
return None
1063
1064
def _get_revision_store(self, repo_transport, control_files):
1065
"""See RepositoryFormat._get_revision_store()."""
1066
from bzrlib.xml4 import serializer_v4
1067
return self._get_text_rev_store(repo_transport,
1068
control_files,
1069
'revision-store',
1070
serializer=serializer_v4)
1071
1072
def _get_text_store(self, transport, control_files):
1073
"""See RepositoryFormat._get_text_store()."""
1074
1075
1076
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1077
"""Bzr control format 5.
1078
1079
This repository format has:
1080
- weaves for file texts and inventory
1081
- flat stores
1082
- TextStores for revisions and signatures.
1083
"""
1084
1085
def __init__(self):
1086
super(RepositoryFormat5, self).__init__()
1087
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1088
1089
def _get_revision_store(self, repo_transport, control_files):
1090
"""See RepositoryFormat._get_revision_store()."""
1091
"""Return the revision store object for this a_bzrdir."""
1092
return self._get_text_rev_store(repo_transport,
1093
control_files,
1094
'revision-store',
1095
compressed=False)
1096
1097
def _get_text_store(self, transport, control_files):
1098
"""See RepositoryFormat._get_text_store()."""
1099
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1100
1101
1102
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1103
"""Bzr control format 6.
1104
1105
This repository format has:
1106
- weaves for file texts and inventory
1107
- hash subdirectory based stores.
1108
- TextStores for revisions and signatures.
1109
"""
1110
1111
def __init__(self):
1112
super(RepositoryFormat6, self).__init__()
1113
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1114
1115
def _get_revision_store(self, repo_transport, control_files):
1116
"""See RepositoryFormat._get_revision_store()."""
1117
return self._get_text_rev_store(repo_transport,
1118
control_files,
1119
'revision-store',
1120
compressed=False,
1121
prefixed=True)
1122
1123
def _get_text_store(self, transport, control_files):
1124
"""See RepositoryFormat._get_text_store()."""
1125
return self._get_versioned_file_store('weaves', transport, control_files)
1126
1127
1128
class MetaDirRepositoryFormat(RepositoryFormat):
1129
"""Common base class for the new repositories using the metadir layour."""
1130
1131
def __init__(self):
1132
super(MetaDirRepositoryFormat, self).__init__()
1133
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1134
1135
def _create_control_files(self, a_bzrdir):
1136
"""Create the required files and the initial control_files object."""
1137
# FIXME: RBC 20060125 dont peek under the covers
1138
# NB: no need to escape relative paths that are url safe.
1139
repository_transport = a_bzrdir.get_repository_transport(self)
1140
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1141
control_files.create_lock()
1142
return control_files
1143
1144
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1145
"""Upload the initial blank content."""
1146
control_files = self._create_control_files(a_bzrdir)
1147
control_files.lock_write()
1148
try:
1149
control_files._transport.mkdir_multi(dirs,
1150
mode=control_files._dir_mode)
1151
for file, content in files:
1152
control_files.put(file, content)
1153
for file, content in utf8_files:
1154
control_files.put_utf8(file, content)
1155
if shared == True:
1156
control_files.put_utf8('shared-storage', '')
1157
finally:
1158
control_files.unlock()
1159
1160
1161
class RepositoryFormat7(MetaDirRepositoryFormat):
1162
"""Bzr repository 7.
1163
1164
This repository format has:
1165
- weaves for file texts and inventory
1166
- hash subdirectory based stores.
1167
- TextStores for revisions and signatures.
1168
- a format marker of its own
1169
- an optional 'shared-storage' flag
1170
- an optional 'no-working-trees' flag
1171
"""
1172
1173
def _get_control_store(self, repo_transport, control_files):
1174
"""Return the control store for this repository."""
1175
return self._get_versioned_file_store('',
1176
repo_transport,
1177
control_files,
1178
prefixed=False)
1179
1180
def get_format_string(self):
1181
"""See RepositoryFormat.get_format_string()."""
1182
return "Bazaar-NG Repository format 7"
1183
1184
def _get_revision_store(self, repo_transport, control_files):
1185
"""See RepositoryFormat._get_revision_store()."""
1186
return self._get_text_rev_store(repo_transport,
1187
control_files,
1188
'revision-store',
1189
compressed=False,
1190
prefixed=True,
1191
)
1192
1193
def _get_text_store(self, transport, control_files):
1194
"""See RepositoryFormat._get_text_store()."""
1195
return self._get_versioned_file_store('weaves',
1196
transport,
1197
control_files)
1198
1199
def initialize(self, a_bzrdir, shared=False):
1200
"""Create a weave repository.
1201
1202
:param shared: If true the repository will be initialized as a shared
1203
repository.
1204
"""
1205
from bzrlib.weavefile import write_weave_v5
1206
from bzrlib.weave import Weave
1207
1208
# Create an empty weave
1209
sio = StringIO()
1210
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1211
empty_weave = sio.getvalue()
1212
1213
mutter('creating repository in %s.', a_bzrdir.transport.base)
1214
dirs = ['revision-store', 'weaves']
1215
files = [('inventory.weave', StringIO(empty_weave)),
1216
]
1217
utf8_files = [('format', self.get_format_string())]
1218
1219
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1220
return self.open(a_bzrdir=a_bzrdir, _found=True)
1221
1222
def open(self, a_bzrdir, _found=False, _override_transport=None):
1223
"""See RepositoryFormat.open().
1224
1225
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1226
repository at a slightly different url
1227
than normal. I.e. during 'upgrade'.
1228
"""
1229
if not _found:
1230
format = RepositoryFormat.find_format(a_bzrdir)
1231
assert format.__class__ == self.__class__
1232
if _override_transport is not None:
1233
repo_transport = _override_transport
1234
else:
1235
repo_transport = a_bzrdir.get_repository_transport(None)
1236
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1237
text_store = self._get_text_store(repo_transport, control_files)
1238
control_store = self._get_control_store(repo_transport, control_files)
1239
_revision_store = self._get_revision_store(repo_transport, control_files)
1240
return MetaDirRepository(_format=self,
1241
a_bzrdir=a_bzrdir,
1242
control_files=control_files,
1243
_revision_store=_revision_store,
1244
control_store=control_store,
1245
text_store=text_store)
1246
1247
1248
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1249
"""Bzr repository knit format 1.
1250
1251
This repository format has:
1252
- knits for file texts and inventory
1253
- hash subdirectory based stores.
1254
- knits for revisions and signatures
1255
- TextStores for revisions and signatures.
1256
- a format marker of its own
1257
- an optional 'shared-storage' flag
1258
- an optional 'no-working-trees' flag
1259
- a LockDir lock
1260
"""
1261
1262
def _get_control_store(self, repo_transport, control_files):
1263
"""Return the control store for this repository."""
1264
return self._get_versioned_file_store('',
1265
repo_transport,
1266
control_files,
1267
prefixed=False,
1268
versionedfile_class=KnitVersionedFile)
1269
1270
def get_format_string(self):
1271
"""See RepositoryFormat.get_format_string()."""
1272
return "Bazaar-NG Knit Repository Format 1"
1273
1274
def _get_revision_store(self, repo_transport, control_files):
1275
"""See RepositoryFormat._get_revision_store()."""
1276
from bzrlib.store.revision.knit import KnitRevisionStore
1277
versioned_file_store = VersionedFileStore(
1278
repo_transport,
1279
file_mode = control_files._file_mode,
1280
prefixed=False,
1281
precious=True,
1282
versionedfile_class=KnitVersionedFile)
1283
return KnitRevisionStore(versioned_file_store)
1284
1285
def _get_text_store(self, transport, control_files):
1286
"""See RepositoryFormat._get_text_store()."""
1287
return self._get_versioned_file_store('knits',
1288
transport,
1289
control_files,
1290
versionedfile_class=KnitVersionedFile)
1291
1292
def initialize(self, a_bzrdir, shared=False):
1293
"""Create a knit format 1 repository.
1294
1295
:param shared: If true the repository will be initialized as a shared
1296
repository.
1297
XXX NOTE that this current uses a Weave for testing and will become
1298
A Knit in due course.
1299
"""
1300
from bzrlib.weavefile import write_weave_v5
1301
from bzrlib.weave import Weave
1302
1303
# Create an empty weave
1304
sio = StringIO()
1305
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1306
empty_weave = sio.getvalue()
1307
1308
mutter('creating repository in %s.', a_bzrdir.transport.base)
1309
dirs = ['revision-store', 'knits', 'control']
1310
files = [('control/inventory.weave', StringIO(empty_weave)),
1311
]
1312
utf8_files = [('format', self.get_format_string())]
1313
1314
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1315
repo_transport = a_bzrdir.get_repository_transport(None)
1316
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1317
control_store = self._get_control_store(repo_transport, control_files)
1318
transaction = bzrlib.transactions.WriteTransaction()
1319
# trigger a write of the inventory store.
1320
control_store.get_weave_or_empty('inventory', transaction)
1321
_revision_store = self._get_revision_store(repo_transport, control_files)
1322
_revision_store.has_revision_id('A', transaction)
1323
_revision_store.get_signature_file(transaction)
1324
return self.open(a_bzrdir=a_bzrdir, _found=True)
1325
1326
def open(self, a_bzrdir, _found=False, _override_transport=None):
1327
"""See RepositoryFormat.open().
1328
1329
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1330
repository at a slightly different url
1331
than normal. I.e. during 'upgrade'.
1332
"""
1333
if not _found:
1334
format = RepositoryFormat.find_format(a_bzrdir)
1335
assert format.__class__ == self.__class__
1336
if _override_transport is not None:
1337
repo_transport = _override_transport
1338
else:
1339
repo_transport = a_bzrdir.get_repository_transport(None)
1340
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1341
text_store = self._get_text_store(repo_transport, control_files)
1342
control_store = self._get_control_store(repo_transport, control_files)
1343
_revision_store = self._get_revision_store(repo_transport, control_files)
1344
return KnitRepository(_format=self,
1345
a_bzrdir=a_bzrdir,
1346
control_files=control_files,
1347
_revision_store=_revision_store,
1348
control_store=control_store,
1349
text_store=text_store)
1350
1351
1352
# formats which have no format string are not discoverable
1353
# and not independently creatable, so are not registered.
1354
_default_format = RepositoryFormat7()
1355
RepositoryFormat.register_format(_default_format)
1356
RepositoryFormat.register_format(RepositoryFormatKnit1())
1357
RepositoryFormat.set_default_format(_default_format)
1358
_legacy_formats = [RepositoryFormat4(),
1359
RepositoryFormat5(),
1360
RepositoryFormat6()]
1361
1362
1363
class InterRepository(InterObject):
1364
"""This class represents operations taking place between two repositories.
1365
1366
Its instances have methods like copy_content and fetch, and contain
1367
references to the source and target repositories these operations can be
1368
carried out on.
1369
1370
Often we will provide convenience methods on 'repository' which carry out
1371
operations with another repository - they will always forward to
1372
InterRepository.get(other).method_name(parameters).
1373
"""
1374
1375
_optimisers = set()
1376
"""The available optimised InterRepository types."""
1377
1378
@needs_write_lock
1379
def copy_content(self, revision_id=None, basis=None):
1380
"""Make a complete copy of the content in self into destination.
1381
1382
This is a destructive operation! Do not use it on existing
1383
repositories.
1384
1385
:param revision_id: Only copy the content needed to construct
1386
revision_id and its parents.
1387
:param basis: Copy the needed data preferentially from basis.
1388
"""
1389
try:
1390
self.target.set_make_working_trees(self.source.make_working_trees())
1391
except NotImplementedError:
1392
pass
1393
# grab the basis available data
1394
if basis is not None:
1395
self.target.fetch(basis, revision_id=revision_id)
1396
# but dont bother fetching if we have the needed data now.
1397
if (revision_id not in (None, NULL_REVISION) and
1398
self.target.has_revision(revision_id)):
1399
return
1400
self.target.fetch(self.source, revision_id=revision_id)
1401
1402
def _double_lock(self, lock_source, lock_target):
1403
"""Take out too locks, rolling back the first if the second throws."""
1404
lock_source()
1405
try:
1406
lock_target()
1407
except Exception:
1408
# we want to ensure that we don't leave source locked by mistake.
1409
# and any error on target should not confuse source.
1410
self.source.unlock()
1411
raise
1412
1413
@needs_write_lock
1414
def fetch(self, revision_id=None, pb=None):
1415
"""Fetch the content required to construct revision_id.
1416
1417
The content is copied from source to target.
1418
1419
:param revision_id: if None all content is copied, if NULL_REVISION no
1420
content is copied.
1421
:param pb: optional progress bar to use for progress reports. If not
1422
provided a default one will be created.
1423
1424
Returns the copied revision count and the failed revisions in a tuple:
1425
(copied, failures).
1426
"""
1427
from bzrlib.fetch import GenericRepoFetcher
1428
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1429
self.source, self.source._format, self.target, self.target._format)
1430
f = GenericRepoFetcher(to_repository=self.target,
1431
from_repository=self.source,
1432
last_revision=revision_id,
1433
pb=pb)
1434
return f.count_copied, f.failed_revisions
1435
1436
def lock_read(self):
1437
"""Take out a logical read lock.
1438
1439
This will lock the source branch and the target branch. The source gets
1440
a read lock and the target a read lock.
1441
"""
1442
self._double_lock(self.source.lock_read, self.target.lock_read)
1443
1444
def lock_write(self):
1445
"""Take out a logical write lock.
1446
1447
This will lock the source branch and the target branch. The source gets
1448
a read lock and the target a write lock.
1449
"""
1450
self._double_lock(self.source.lock_read, self.target.lock_write)
1451
1452
@needs_read_lock
1453
def missing_revision_ids(self, revision_id=None):
1454
"""Return the revision ids that source has that target does not.
1455
1456
These are returned in topological order.
1457
1458
:param revision_id: only return revision ids included by this
1459
revision_id.
1460
"""
1461
# generic, possibly worst case, slow code path.
1462
target_ids = set(self.target.all_revision_ids())
1463
if revision_id is not None:
1464
source_ids = self.source.get_ancestry(revision_id)
1465
assert source_ids.pop(0) == None
1466
else:
1467
source_ids = self.source.all_revision_ids()
1468
result_set = set(source_ids).difference(target_ids)
1469
# this may look like a no-op: its not. It preserves the ordering
1470
# other_ids had while only returning the members from other_ids
1471
# that we've decided we need.
1472
return [rev_id for rev_id in source_ids if rev_id in result_set]
1473
1474
def unlock(self):
1475
"""Release the locks on source and target."""
1476
try:
1477
self.target.unlock()
1478
finally:
1479
self.source.unlock()
1480
1481
1482
class InterWeaveRepo(InterRepository):
1483
"""Optimised code paths between Weave based repositories."""
1484
1485
_matching_repo_format = _default_format
1486
"""Repository format for testing with."""
1487
1488
@staticmethod
1489
def is_compatible(source, target):
1490
"""Be compatible with known Weave formats.
1491
1492
We dont test for the stores being of specific types becase that
1493
could lead to confusing results, and there is no need to be
1494
overly general.
1495
"""
1496
try:
1497
return (isinstance(source._format, (RepositoryFormat5,
1498
RepositoryFormat6,
1499
RepositoryFormat7)) and
1500
isinstance(target._format, (RepositoryFormat5,
1501
RepositoryFormat6,
1502
RepositoryFormat7)))
1503
except AttributeError:
1504
return False
1505
1506
@needs_write_lock
1507
def copy_content(self, revision_id=None, basis=None):
1508
"""See InterRepository.copy_content()."""
1509
# weave specific optimised path:
1510
if basis is not None:
1511
# copy the basis in, then fetch remaining data.
1512
basis.copy_content_into(self.target, revision_id)
1513
# the basis copy_content_into could misset this.
1514
try:
1515
self.target.set_make_working_trees(self.source.make_working_trees())
1516
except NotImplementedError:
1517
pass
1518
self.target.fetch(self.source, revision_id=revision_id)
1519
else:
1520
try:
1521
self.target.set_make_working_trees(self.source.make_working_trees())
1522
except NotImplementedError:
1523
pass
1524
# FIXME do not peek!
1525
if self.source.control_files._transport.listable():
1526
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1527
try:
1528
self.target.weave_store.copy_all_ids(
1529
self.source.weave_store,
1530
pb=pb,
1531
from_transaction=self.source.get_transaction(),
1532
to_transaction=self.target.get_transaction())
1533
pb.update('copying inventory', 0, 1)
1534
self.target.control_weaves.copy_multi(
1535
self.source.control_weaves, ['inventory'],
1536
from_transaction=self.source.get_transaction(),
1537
to_transaction=self.target.get_transaction())
1538
self.target._revision_store.text_store.copy_all_ids(
1539
self.source._revision_store.text_store,
1540
pb=pb)
1541
finally:
1542
pb.finished()
1543
else:
1544
self.target.fetch(self.source, revision_id=revision_id)
1545
1546
@needs_write_lock
1547
def fetch(self, revision_id=None, pb=None):
1548
"""See InterRepository.fetch()."""
1549
from bzrlib.fetch import GenericRepoFetcher
1550
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1551
self.source, self.source._format, self.target, self.target._format)
1552
f = GenericRepoFetcher(to_repository=self.target,
1553
from_repository=self.source,
1554
last_revision=revision_id,
1555
pb=pb)
1556
return f.count_copied, f.failed_revisions
1557
1558
@needs_read_lock
1559
def missing_revision_ids(self, revision_id=None):
1560
"""See InterRepository.missing_revision_ids()."""
1561
# we want all revisions to satisfy revision_id in source.
1562
# but we dont want to stat every file here and there.
1563
# we want then, all revisions other needs to satisfy revision_id
1564
# checked, but not those that we have locally.
1565
# so the first thing is to get a subset of the revisions to
1566
# satisfy revision_id in source, and then eliminate those that
1567
# we do already have.
1568
# this is slow on high latency connection to self, but as as this
1569
# disk format scales terribly for push anyway due to rewriting
1570
# inventory.weave, this is considered acceptable.
1571
# - RBC 20060209
1572
if revision_id is not None:
1573
source_ids = self.source.get_ancestry(revision_id)
1574
assert source_ids.pop(0) == None
1575
else:
1576
source_ids = self.source._all_possible_ids()
1577
source_ids_set = set(source_ids)
1578
# source_ids is the worst possible case we may need to pull.
1579
# now we want to filter source_ids against what we actually
1580
# have in target, but dont try to check for existence where we know
1581
# we do not have a revision as that would be pointless.
1582
target_ids = set(self.target._all_possible_ids())
1583
possibly_present_revisions = target_ids.intersection(source_ids_set)
1584
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1585
required_revisions = source_ids_set.difference(actually_present_revisions)
1586
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1587
if revision_id is not None:
1588
# we used get_ancestry to determine source_ids then we are assured all
1589
# revisions referenced are present as they are installed in topological order.
1590
# and the tip revision was validated by get_ancestry.
1591
return required_topo_revisions
1592
else:
1593
# if we just grabbed the possibly available ids, then
1594
# we only have an estimate of whats available and need to validate
1595
# that against the revision records.
1596
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1597
1598
1599
class InterKnitRepo(InterRepository):
1600
"""Optimised code paths between Knit based repositories."""
1601
1602
_matching_repo_format = RepositoryFormatKnit1()
1603
"""Repository format for testing with."""
1604
1605
@staticmethod
1606
def is_compatible(source, target):
1607
"""Be compatible with known Knit formats.
1608
1609
We dont test for the stores being of specific types becase that
1610
could lead to confusing results, and there is no need to be
1611
overly general.
1612
"""
1613
try:
1614
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1615
isinstance(target._format, (RepositoryFormatKnit1)))
1616
except AttributeError:
1617
return False
1618
1619
@needs_write_lock
1620
def fetch(self, revision_id=None, pb=None):
1621
"""See InterRepository.fetch()."""
1622
from bzrlib.fetch import KnitRepoFetcher
1623
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1624
self.source, self.source._format, self.target, self.target._format)
1625
f = KnitRepoFetcher(to_repository=self.target,
1626
from_repository=self.source,
1627
last_revision=revision_id,
1628
pb=pb)
1629
return f.count_copied, f.failed_revisions
1630
1631
@needs_read_lock
1632
def missing_revision_ids(self, revision_id=None):
1633
"""See InterRepository.missing_revision_ids()."""
1634
if revision_id is not None:
1635
source_ids = self.source.get_ancestry(revision_id)
1636
assert source_ids.pop(0) == None
1637
else:
1638
source_ids = self.source._all_possible_ids()
1639
source_ids_set = set(source_ids)
1640
# source_ids is the worst possible case we may need to pull.
1641
# now we want to filter source_ids against what we actually
1642
# have in target, but dont try to check for existence where we know
1643
# we do not have a revision as that would be pointless.
1644
target_ids = set(self.target._all_possible_ids())
1645
possibly_present_revisions = target_ids.intersection(source_ids_set)
1646
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1647
required_revisions = source_ids_set.difference(actually_present_revisions)
1648
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1649
if revision_id is not None:
1650
# we used get_ancestry to determine source_ids then we are assured all
1651
# revisions referenced are present as they are installed in topological order.
1652
# and the tip revision was validated by get_ancestry.
1653
return required_topo_revisions
1654
else:
1655
# if we just grabbed the possibly available ids, then
1656
# we only have an estimate of whats available and need to validate
1657
# that against the revision records.
1658
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1659
1660
InterRepository.register_optimiser(InterWeaveRepo)
1661
InterRepository.register_optimiser(InterKnitRepo)
1662
1663
1664
class RepositoryTestProviderAdapter(object):
1665
"""A tool to generate a suite testing multiple repository formats at once.
1666
1667
This is done by copying the test once for each transport and injecting
1668
the transport_server, transport_readonly_server, and bzrdir_format and
1669
repository_format classes into each copy. Each copy is also given a new id()
1670
to make it easy to identify.
1671
"""
1672
1673
def __init__(self, transport_server, transport_readonly_server, formats):
1674
self._transport_server = transport_server
1675
self._transport_readonly_server = transport_readonly_server
1676
self._formats = formats
1677
1678
def adapt(self, test):
1679
result = TestSuite()
1680
for repository_format, bzrdir_format in self._formats:
1681
new_test = deepcopy(test)
1682
new_test.transport_server = self._transport_server
1683
new_test.transport_readonly_server = self._transport_readonly_server
1684
new_test.bzrdir_format = bzrdir_format
1685
new_test.repository_format = repository_format
1686
def make_new_test_id():
1687
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1688
return lambda: new_id
1689
new_test.id = make_new_test_id()
1690
result.addTest(new_test)
1691
return result
1692
1693
1694
class InterRepositoryTestProviderAdapter(object):
1695
"""A tool to generate a suite testing multiple inter repository formats.
1696
1697
This is done by copying the test once for each interrepo provider and injecting
1698
the transport_server, transport_readonly_server, repository_format and
1699
repository_to_format classes into each copy.
1700
Each copy is also given a new id() to make it easy to identify.
1701
"""
1702
1703
def __init__(self, transport_server, transport_readonly_server, formats):
1704
self._transport_server = transport_server
1705
self._transport_readonly_server = transport_readonly_server
1706
self._formats = formats
1707
1708
def adapt(self, test):
1709
result = TestSuite()
1710
for interrepo_class, repository_format, repository_format_to in self._formats:
1711
new_test = deepcopy(test)
1712
new_test.transport_server = self._transport_server
1713
new_test.transport_readonly_server = self._transport_readonly_server
1714
new_test.interrepo_class = interrepo_class
1715
new_test.repository_format = repository_format
1716
new_test.repository_format_to = repository_format_to
1717
def make_new_test_id():
1718
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1719
return lambda: new_id
1720
new_test.id = make_new_test_id()
1721
result.addTest(new_test)
1722
return result
1723
1724
@staticmethod
1725
def default_test_list():
1726
"""Generate the default list of interrepo permutations to test."""
1727
result = []
1728
# test the default InterRepository between format 6 and the current
1729
# default format.
1730
# XXX: robertc 20060220 reinstate this when there are two supported
1731
# formats which do not have an optimal code path between them.
1732
result.append((InterRepository,
1733
RepositoryFormat6(),
1734
RepositoryFormatKnit1()))
1735
for optimiser in InterRepository._optimisers:
1736
result.append((optimiser,
1737
optimiser._matching_repo_format,
1738
optimiser._matching_repo_format
1739
))
1740
# if there are specific combinations we want to use, we can add them
1741
# here.
1742
return result
1743
1744
1745
class CopyConverter(object):
1746
"""A repository conversion tool which just performs a copy of the content.
1747
1748
This is slow but quite reliable.
1749
"""
1750
1751
def __init__(self, target_format):
1752
"""Create a CopyConverter.
1753
1754
:param target_format: The format the resulting repository should be.
1755
"""
1756
self.target_format = target_format
1757
1758
def convert(self, repo, pb):
1759
"""Perform the conversion of to_convert, giving feedback via pb.
1760
1761
:param to_convert: The disk object to convert.
1762
:param pb: a progress bar to use for progress information.
1763
"""
1764
self.pb = pb
1765
self.count = 0
1766
self.total = 4
1767
# this is only useful with metadir layouts - separated repo content.
1768
# trigger an assertion if not such
1769
repo._format.get_format_string()
1770
self.repo_dir = repo.bzrdir
1771
self.step('Moving repository to repository.backup')
1772
self.repo_dir.transport.move('repository', 'repository.backup')
1773
backup_transport = self.repo_dir.transport.clone('repository.backup')
1774
self.source_repo = repo._format.open(self.repo_dir,
1775
_found=True,
1776
_override_transport=backup_transport)
1777
self.step('Creating new repository')
1778
converted = self.target_format.initialize(self.repo_dir,
1779
self.source_repo.is_shared())
1780
converted.lock_write()
1781
try:
1782
self.step('Copying content into repository.')
1783
self.source_repo.copy_content_into(converted)
1784
finally:
1785
converted.unlock()
1786
self.step('Deleting old repository content.')
1787
self.repo_dir.transport.delete_tree('repository.backup')
1788
self.pb.note('repository converted')
1789
1790
def step(self, message):
1791
"""Update the pb by a step."""
1792
self.count +=1
1793
self.pb.update(message, self.count, self.total)
1794
1795
1796
# Copied from xml.sax.saxutils
1797
def _unescape_xml(data):
1798
"""Unescape &, <, and > in a string of data.
1799
"""
1800
data = data.replace("<", "<")
1801
data = data.replace(">", ">")
1802
# must do ampersand last
1803
return data.replace("&", "&")
Loggerhead is a web-based interface for Breezy
Version: 2.0.1