« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 0.1.28
- compare with another revision
- download diff
- download tarball
- view history from revision 0.1.28
- Committer: Martin Pool
- Date: 2005-06-27 08:18:07 UTC
- mto: This revision was merged to the branch mainline in revision 852.
- Revision ID: mbp@sourcefrog.net-20050627081807-dc3ff5726c88b247
More tests for insertion of lines in new versions.
- files added:
- .bzrignore
- files removed:
- .bzrignore
- bzrlib
- bzrlib/benchmarks
- bzrlib/benchmarks/tree_creator
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/doc
- bzrlib/doc/api
- bzrlib/export
- bzrlib/plugins
- bzrlib/plugins/launchpad
- bzrlib/store
- bzrlib/store/revision
- bzrlib/store/versioned
- bzrlib/tests
- bzrlib/tests/blackbox
- bzrlib/tests/branch_implementations
- bzrlib/tests/bzrdir_implementations
- bzrlib/tests/interrepository_implementations
- bzrlib/tests/intertree_implementations
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/repository_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/tests/test_patches_data
- bzrlib/tests/tree_implementations
- bzrlib/tests/workingtree_implementations
- bzrlib/transport
- bzrlib/transport/http
- bzrlib/ui
- bzrlib/util
- bzrlib/util/configobj
- bzrlib/util/configobj/docs
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- bzrlib/util/elementtree
- bzrlib/version_info_formats
- contrib
- contrib/bash
- contrib/emacs
- contrib/zsh
- doc
- tools
- tools/doc_generate
- tools/win32
- files renamed:
- bzrlib/weave.py => knit.py *
- bzrlib/tests/test_weave.py => testknit.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 cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
23
import re
24
import time
25
import unittest
26
27
from bzrlib import (
28
bzrdir,
29
check,
30
delta,
31
errors,
32
generate_ids,
33
gpg,
34
graph,
35
knit,
36
lazy_regex,
37
lockable_files,
38
lockdir,
39
osutils,
40
revision as _mod_revision,
41
symbol_versioning,
42
transactions,
43
ui,
44
weave,
45
weavefile,
46
xml5,
47
xml6,
48
)
49
from bzrlib.osutils import (
50
rand_bytes,
51
compact_date,
52
local_time_offset,
53
)
54
from bzrlib.revisiontree import RevisionTree
55
from bzrlib.store.versioned import VersionedFileStore
56
from bzrlib.store.text import TextStore
57
from bzrlib.testament import Testament
58
""")
59
60
from bzrlib.decorators import needs_read_lock, needs_write_lock
61
from bzrlib.inter import InterObject
62
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
63
from bzrlib.symbol_versioning import (
64
deprecated_method,
65
zero_nine,
66
)
67
from bzrlib.trace import mutter, note, warning
68
69
70
# Old formats display a warning, but only once
71
_deprecation_warning_done = False
72
73
74
class Repository(object):
75
"""Repository holding history for one or more branches.
76
77
The repository holds and retrieves historical information including
78
revisions and file history. It's normally accessed only by the Branch,
79
which views a particular line of development through that history.
80
81
The Repository builds on top of Stores and a Transport, which respectively
82
describe the disk data format and the way of accessing the (possibly
83
remote) disk.
84
"""
85
86
_file_ids_altered_regex = lazy_regex.lazy_compile(
87
r'file_id="(?P<file_id>[^"]+)"'
88
r'.*revision="(?P<revision_id>[^"]+)"'
89
)
90
91
@needs_write_lock
92
def add_inventory(self, revid, inv, parents):
93
"""Add the inventory inv to the repository as revid.
94
95
:param parents: The revision ids of the parents that revid
96
is known to have and are in the repository already.
97
98
returns the sha1 of the serialized inventory.
99
"""
100
assert inv.revision_id is None or inv.revision_id == revid, \
101
"Mismatch between inventory revision" \
102
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
103
assert inv.root is not None
104
inv_text = self.serialise_inventory(inv)
105
inv_sha1 = osutils.sha_string(inv_text)
106
inv_vf = self.control_weaves.get_weave('inventory',
107
self.get_transaction())
108
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
109
return inv_sha1
110
111
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
112
final_parents = []
113
for parent in parents:
114
if parent in inv_vf:
115
final_parents.append(parent)
116
117
inv_vf.add_lines(revid, final_parents, lines)
118
119
@needs_write_lock
120
def add_revision(self, rev_id, rev, inv=None, config=None):
121
"""Add rev to the revision store as rev_id.
122
123
:param rev_id: the revision id to use.
124
:param rev: The revision object.
125
:param inv: The inventory for the revision. if None, it will be looked
126
up in the inventory storer
127
:param config: If None no digital signature will be created.
128
If supplied its signature_needed method will be used
129
to determine if a signature should be made.
130
"""
131
if config is not None and config.signature_needed():
132
if inv is None:
133
inv = self.get_inventory(rev_id)
134
plaintext = Testament(rev, inv).as_short_text()
135
self.store_revision_signature(
136
gpg.GPGStrategy(config), plaintext, rev_id)
137
if not rev_id in self.get_inventory_weave():
138
if inv is None:
139
raise errors.WeaveRevisionNotPresent(rev_id,
140
self.get_inventory_weave())
141
else:
142
# yes, this is not suitable for adding with ghosts.
143
self.add_inventory(rev_id, inv, rev.parent_ids)
144
self._revision_store.add_revision(rev, self.get_transaction())
145
146
@needs_read_lock
147
def _all_possible_ids(self):
148
"""Return all the possible revisions that we could find."""
149
return self.get_inventory_weave().versions()
150
151
def all_revision_ids(self):
152
"""Returns a list of all the revision ids in the repository.
153
154
This is deprecated because code should generally work on the graph
155
reachable from a particular revision, and ignore any other revisions
156
that might be present. There is no direct replacement method.
157
"""
158
return self._all_revision_ids()
159
160
@needs_read_lock
161
def _all_revision_ids(self):
162
"""Returns a list of all the revision ids in the repository.
163
164
These are in as much topological order as the underlying store can
165
present: for weaves ghosts may lead to a lack of correctness until
166
the reweave updates the parents list.
167
"""
168
if self._revision_store.text_store.listable():
169
return self._revision_store.all_revision_ids(self.get_transaction())
170
result = self._all_possible_ids()
171
return self._eliminate_revisions_not_present(result)
172
173
def break_lock(self):
174
"""Break a lock if one is present from another instance.
175
176
Uses the ui factory to ask for confirmation if the lock may be from
177
an active process.
178
"""
179
self.control_files.break_lock()
180
181
@needs_read_lock
182
def _eliminate_revisions_not_present(self, revision_ids):
183
"""Check every revision id in revision_ids to see if we have it.
184
185
Returns a set of the present revisions.
186
"""
187
result = []
188
for id in revision_ids:
189
if self.has_revision(id):
190
result.append(id)
191
return result
192
193
@staticmethod
194
def create(a_bzrdir):
195
"""Construct the current default format repository in a_bzrdir."""
196
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
197
198
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
199
"""instantiate a Repository.
200
201
:param _format: The format of the repository on disk.
202
:param a_bzrdir: The BzrDir of the repository.
203
204
In the future we will have a single api for all stores for
205
getting file texts, inventories and revisions, then
206
this construct will accept instances of those things.
207
"""
208
super(Repository, self).__init__()
209
self._format = _format
210
# the following are part of the public API for Repository:
211
self.bzrdir = a_bzrdir
212
self.control_files = control_files
213
self._revision_store = _revision_store
214
self.text_store = text_store
215
# backwards compatibility
216
self.weave_store = text_store
217
# not right yet - should be more semantically clear ?
218
#
219
self.control_store = control_store
220
self.control_weaves = control_store
221
# TODO: make sure to construct the right store classes, etc, depending
222
# on whether escaping is required.
223
self._warn_if_deprecated()
224
self._serializer = xml5.serializer_v5
225
226
def __repr__(self):
227
return '%s(%r)' % (self.__class__.__name__,
228
self.bzrdir.transport.base)
229
230
def is_locked(self):
231
return self.control_files.is_locked()
232
233
def lock_write(self):
234
self.control_files.lock_write()
235
236
def lock_read(self):
237
self.control_files.lock_read()
238
239
def get_physical_lock_status(self):
240
return self.control_files.get_physical_lock_status()
241
242
@needs_read_lock
243
def missing_revision_ids(self, other, revision_id=None):
244
"""Return the revision ids that other has that this does not.
245
246
These are returned in topological order.
247
248
revision_id: only return revision ids included by revision_id.
249
"""
250
return InterRepository.get(other, self).missing_revision_ids(revision_id)
251
252
@staticmethod
253
def open(base):
254
"""Open the repository rooted at base.
255
256
For instance, if the repository is at URL/.bzr/repository,
257
Repository.open(URL) -> a Repository instance.
258
"""
259
control = bzrdir.BzrDir.open(base)
260
return control.open_repository()
261
262
def copy_content_into(self, destination, revision_id=None, basis=None):
263
"""Make a complete copy of the content in self into destination.
264
265
This is a destructive operation! Do not use it on existing
266
repositories.
267
"""
268
return InterRepository.get(self, destination).copy_content(revision_id, basis)
269
270
def fetch(self, source, revision_id=None, pb=None):
271
"""Fetch the content required to construct revision_id from source.
272
273
If revision_id is None all content is copied.
274
"""
275
return InterRepository.get(source, self).fetch(revision_id=revision_id,
276
pb=pb)
277
278
def get_commit_builder(self, branch, parents, config, timestamp=None,
279
timezone=None, committer=None, revprops=None,
280
revision_id=None):
281
"""Obtain a CommitBuilder for this repository.
282
283
:param branch: Branch to commit to.
284
:param parents: Revision ids of the parents of the new revision.
285
:param config: Configuration to use.
286
:param timestamp: Optional timestamp recorded for commit.
287
:param timezone: Optional timezone for timestamp.
288
:param committer: Optional committer to set for commit.
289
:param revprops: Optional dictionary of revision properties.
290
:param revision_id: Optional revision id.
291
"""
292
return _CommitBuilder(self, parents, config, timestamp, timezone,
293
committer, revprops, revision_id)
294
295
def unlock(self):
296
self.control_files.unlock()
297
298
@needs_read_lock
299
def clone(self, a_bzrdir, revision_id=None, basis=None):
300
"""Clone this repository into a_bzrdir using the current format.
301
302
Currently no check is made that the format of this repository and
303
the bzrdir format are compatible. FIXME RBC 20060201.
304
"""
305
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
306
# use target default format.
307
result = a_bzrdir.create_repository()
308
# FIXME RBC 20060209 split out the repository type to avoid this check ?
309
elif isinstance(a_bzrdir._format,
310
(bzrdir.BzrDirFormat4,
311
bzrdir.BzrDirFormat5,
312
bzrdir.BzrDirFormat6)):
313
result = a_bzrdir.open_repository()
314
else:
315
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
316
self.copy_content_into(result, revision_id, basis)
317
return result
318
319
@needs_read_lock
320
def has_revision(self, revision_id):
321
"""True if this repository has a copy of the revision."""
322
return self._revision_store.has_revision_id(revision_id,
323
self.get_transaction())
324
325
@needs_read_lock
326
def get_revision_reconcile(self, revision_id):
327
"""'reconcile' helper routine that allows access to a revision always.
328
329
This variant of get_revision does not cross check the weave graph
330
against the revision one as get_revision does: but it should only
331
be used by reconcile, or reconcile-alike commands that are correcting
332
or testing the revision graph.
333
"""
334
if not revision_id or not isinstance(revision_id, basestring):
335
raise errors.InvalidRevisionId(revision_id=revision_id,
336
branch=self)
337
return self._revision_store.get_revisions([revision_id],
338
self.get_transaction())[0]
339
@needs_read_lock
340
def get_revisions(self, revision_ids):
341
return self._revision_store.get_revisions(revision_ids,
342
self.get_transaction())
343
344
@needs_read_lock
345
def get_revision_xml(self, revision_id):
346
rev = self.get_revision(revision_id)
347
rev_tmp = StringIO()
348
# the current serializer..
349
self._revision_store._serializer.write_revision(rev, rev_tmp)
350
rev_tmp.seek(0)
351
return rev_tmp.getvalue()
352
353
@needs_read_lock
354
def get_revision(self, revision_id):
355
"""Return the Revision object for a named revision"""
356
r = self.get_revision_reconcile(revision_id)
357
# weave corruption can lead to absent revision markers that should be
358
# present.
359
# the following test is reasonably cheap (it needs a single weave read)
360
# and the weave is cached in read transactions. In write transactions
361
# it is not cached but typically we only read a small number of
362
# revisions. For knits when they are introduced we will probably want
363
# to ensure that caching write transactions are in use.
364
inv = self.get_inventory_weave()
365
self._check_revision_parents(r, inv)
366
return r
367
368
@needs_read_lock
369
def get_deltas_for_revisions(self, revisions):
370
"""Produce a generator of revision deltas.
371
372
Note that the input is a sequence of REVISIONS, not revision_ids.
373
Trees will be held in memory until the generator exits.
374
Each delta is relative to the revision's lefthand predecessor.
375
"""
376
required_trees = set()
377
for revision in revisions:
378
required_trees.add(revision.revision_id)
379
required_trees.update(revision.parent_ids[:1])
380
trees = dict((t.get_revision_id(), t) for
381
t in self.revision_trees(required_trees))
382
for revision in revisions:
383
if not revision.parent_ids:
384
old_tree = self.revision_tree(None)
385
else:
386
old_tree = trees[revision.parent_ids[0]]
387
yield trees[revision.revision_id].changes_from(old_tree)
388
389
@needs_read_lock
390
def get_revision_delta(self, revision_id):
391
"""Return the delta for one revision.
392
393
The delta is relative to the left-hand predecessor of the
394
revision.
395
"""
396
r = self.get_revision(revision_id)
397
return list(self.get_deltas_for_revisions([r]))[0]
398
399
def _check_revision_parents(self, revision, inventory):
400
"""Private to Repository and Fetch.
401
402
This checks the parentage of revision in an inventory weave for
403
consistency and is only applicable to inventory-weave-for-ancestry
404
using repository formats & fetchers.
405
"""
406
weave_parents = inventory.get_parents(revision.revision_id)
407
weave_names = inventory.versions()
408
for parent_id in revision.parent_ids:
409
if parent_id in weave_names:
410
# this parent must not be a ghost.
411
if not parent_id in weave_parents:
412
# but it is a ghost
413
raise errors.CorruptRepository(self)
414
415
@needs_write_lock
416
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
417
signature = gpg_strategy.sign(plaintext)
418
self._revision_store.add_revision_signature_text(revision_id,
419
signature,
420
self.get_transaction())
421
422
def fileids_altered_by_revision_ids(self, revision_ids):
423
"""Find the file ids and versions affected by revisions.
424
425
:param revisions: an iterable containing revision ids.
426
:return: a dictionary mapping altered file-ids to an iterable of
427
revision_ids. Each altered file-ids has the exact revision_ids that
428
altered it listed explicitly.
429
"""
430
assert self._serializer.support_altered_by_hack, \
431
("fileids_altered_by_revision_ids only supported for branches "
432
"which store inventory as unnested xml, not on %r" % self)
433
selected_revision_ids = set(revision_ids)
434
w = self.get_inventory_weave()
435
result = {}
436
437
# this code needs to read every new line in every inventory for the
438
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
439
# not present in one of those inventories is unnecessary but not
440
# harmful because we are filtering by the revision id marker in the
441
# inventory lines : we only select file ids altered in one of those
442
# revisions. We don't need to see all lines in the inventory because
443
# only those added in an inventory in rev X can contain a revision=X
444
# line.
445
unescape_revid_cache = {}
446
unescape_fileid_cache = {}
447
448
# jam 20061218 In a big fetch, this handles hundreds of thousands
449
# of lines, so it has had a lot of inlining and optimizing done.
450
# Sorry that it is a little bit messy.
451
# Move several functions to be local variables, since this is a long
452
# running loop.
453
search = self._file_ids_altered_regex.search
454
unescape = _unescape_xml
455
setdefault = result.setdefault
456
pb = ui.ui_factory.nested_progress_bar()
457
try:
458
for line in w.iter_lines_added_or_present_in_versions(
459
selected_revision_ids, pb=pb):
460
match = search(line)
461
if match is None:
462
continue
463
# One call to match.group() returning multiple items is quite a
464
# bit faster than 2 calls to match.group() each returning 1
465
file_id, revision_id = match.group('file_id', 'revision_id')
466
467
# Inlining the cache lookups helps a lot when you make 170,000
468
# lines and 350k ids, versus 8.4 unique ids.
469
# Using a cache helps in 2 ways:
470
# 1) Avoids unnecessary decoding calls
471
# 2) Re-uses cached strings, which helps in future set and
472
# equality checks.
473
# (2) is enough that removing encoding entirely along with
474
# the cache (so we are using plain strings) results in no
475
# performance improvement.
476
try:
477
revision_id = unescape_revid_cache[revision_id]
478
except KeyError:
479
unescaped = unescape(revision_id)
480
unescape_revid_cache[revision_id] = unescaped
481
revision_id = unescaped
482
483
if revision_id in selected_revision_ids:
484
try:
485
file_id = unescape_fileid_cache[file_id]
486
except KeyError:
487
unescaped = unescape(file_id)
488
unescape_fileid_cache[file_id] = unescaped
489
file_id = unescaped
490
setdefault(file_id, set()).add(revision_id)
491
finally:
492
pb.finished()
493
return result
494
495
@needs_read_lock
496
def get_inventory_weave(self):
497
return self.control_weaves.get_weave('inventory',
498
self.get_transaction())
499
500
@needs_read_lock
501
def get_inventory(self, revision_id):
502
"""Get Inventory object by hash."""
503
return self.deserialise_inventory(
504
revision_id, self.get_inventory_xml(revision_id))
505
506
def deserialise_inventory(self, revision_id, xml):
507
"""Transform the xml into an inventory object.
508
509
:param revision_id: The expected revision id of the inventory.
510
:param xml: A serialised inventory.
511
"""
512
result = self._serializer.read_inventory_from_string(xml)
513
result.root.revision = revision_id
514
return result
515
516
def serialise_inventory(self, inv):
517
return self._serializer.write_inventory_to_string(inv)
518
519
@needs_read_lock
520
def get_inventory_xml(self, revision_id):
521
"""Get inventory XML as a file object."""
522
try:
523
assert isinstance(revision_id, basestring), type(revision_id)
524
iw = self.get_inventory_weave()
525
return iw.get_text(revision_id)
526
except IndexError:
527
raise errors.HistoryMissing(self, 'inventory', revision_id)
528
529
@needs_read_lock
530
def get_inventory_sha1(self, revision_id):
531
"""Return the sha1 hash of the inventory entry
532
"""
533
return self.get_revision(revision_id).inventory_sha1
534
535
@needs_read_lock
536
def get_revision_graph(self, revision_id=None):
537
"""Return a dictionary containing the revision graph.
538
539
:param revision_id: The revision_id to get a graph from. If None, then
540
the entire revision graph is returned. This is a deprecated mode of
541
operation and will be removed in the future.
542
:return: a dictionary of revision_id->revision_parents_list.
543
"""
544
# special case NULL_REVISION
545
if revision_id == _mod_revision.NULL_REVISION:
546
return {}
547
a_weave = self.get_inventory_weave()
548
all_revisions = self._eliminate_revisions_not_present(
549
a_weave.versions())
550
entire_graph = dict([(node, a_weave.get_parents(node)) for
551
node in all_revisions])
552
if revision_id is None:
553
return entire_graph
554
elif revision_id not in entire_graph:
555
raise errors.NoSuchRevision(self, revision_id)
556
else:
557
# add what can be reached from revision_id
558
result = {}
559
pending = set([revision_id])
560
while len(pending) > 0:
561
node = pending.pop()
562
result[node] = entire_graph[node]
563
for revision_id in result[node]:
564
if revision_id not in result:
565
pending.add(revision_id)
566
return result
567
568
@needs_read_lock
569
def get_revision_graph_with_ghosts(self, revision_ids=None):
570
"""Return a graph of the revisions with ghosts marked as applicable.
571
572
:param revision_ids: an iterable of revisions to graph or None for all.
573
:return: a Graph object with the graph reachable from revision_ids.
574
"""
575
result = graph.Graph()
576
if not revision_ids:
577
pending = set(self.all_revision_ids())
578
required = set([])
579
else:
580
pending = set(revision_ids)
581
# special case NULL_REVISION
582
if _mod_revision.NULL_REVISION in pending:
583
pending.remove(_mod_revision.NULL_REVISION)
584
required = set(pending)
585
done = set([])
586
while len(pending):
587
revision_id = pending.pop()
588
try:
589
rev = self.get_revision(revision_id)
590
except errors.NoSuchRevision:
591
if revision_id in required:
592
raise
593
# a ghost
594
result.add_ghost(revision_id)
595
continue
596
for parent_id in rev.parent_ids:
597
# is this queued or done ?
598
if (parent_id not in pending and
599
parent_id not in done):
600
# no, queue it.
601
pending.add(parent_id)
602
result.add_node(revision_id, rev.parent_ids)
603
done.add(revision_id)
604
return result
605
606
@needs_read_lock
607
def get_revision_inventory(self, revision_id):
608
"""Return inventory of a past revision."""
609
# TODO: Unify this with get_inventory()
610
# bzr 0.0.6 and later imposes the constraint that the inventory_id
611
# must be the same as its revision, so this is trivial.
612
if revision_id is None:
613
# This does not make sense: if there is no revision,
614
# then it is the current tree inventory surely ?!
615
# and thus get_root_id() is something that looks at the last
616
# commit on the branch, and the get_root_id is an inventory check.
617
raise NotImplementedError
618
# return Inventory(self.get_root_id())
619
else:
620
return self.get_inventory(revision_id)
621
622
@needs_read_lock
623
def is_shared(self):
624
"""Return True if this repository is flagged as a shared repository."""
625
raise NotImplementedError(self.is_shared)
626
627
@needs_write_lock
628
def reconcile(self, other=None, thorough=False):
629
"""Reconcile this repository."""
630
from bzrlib.reconcile import RepoReconciler
631
reconciler = RepoReconciler(self, thorough=thorough)
632
reconciler.reconcile()
633
return reconciler
634
635
@needs_read_lock
636
def revision_tree(self, revision_id):
637
"""Return Tree for a revision on this branch.
638
639
`revision_id` may be None for the empty tree revision.
640
"""
641
# TODO: refactor this to use an existing revision object
642
# so we don't need to read it in twice.
643
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
644
return RevisionTree(self, Inventory(root_id=None),
645
_mod_revision.NULL_REVISION)
646
else:
647
inv = self.get_revision_inventory(revision_id)
648
return RevisionTree(self, inv, revision_id)
649
650
@needs_read_lock
651
def revision_trees(self, revision_ids):
652
"""Return Tree for a revision on this branch.
653
654
`revision_id` may not be None or 'null:'"""
655
assert None not in revision_ids
656
assert _mod_revision.NULL_REVISION not in revision_ids
657
texts = self.get_inventory_weave().get_texts(revision_ids)
658
for text, revision_id in zip(texts, revision_ids):
659
inv = self.deserialise_inventory(revision_id, text)
660
yield RevisionTree(self, inv, revision_id)
661
662
@needs_read_lock
663
def get_ancestry(self, revision_id):
664
"""Return a list of revision-ids integrated by a revision.
665
666
The first element of the list is always None, indicating the origin
667
revision. This might change when we have history horizons, or
668
perhaps we should have a new API.
669
670
This is topologically sorted.
671
"""
672
if revision_id is None:
673
return [None]
674
if not self.has_revision(revision_id):
675
raise errors.NoSuchRevision(self, revision_id)
676
w = self.get_inventory_weave()
677
candidates = w.get_ancestry(revision_id)
678
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
679
680
@needs_read_lock
681
def print_file(self, file, revision_id):
682
"""Print `file` to stdout.
683
684
FIXME RBC 20060125 as John Meinel points out this is a bad api
685
- it writes to stdout, it assumes that that is valid etc. Fix
686
by creating a new more flexible convenience function.
687
"""
688
tree = self.revision_tree(revision_id)
689
# use inventory as it was in that revision
690
file_id = tree.inventory.path2id(file)
691
if not file_id:
692
# TODO: jam 20060427 Write a test for this code path
693
# it had a bug in it, and was raising the wrong
694
# exception.
695
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
696
tree.print_file(file_id)
697
698
def get_transaction(self):
699
return self.control_files.get_transaction()
700
701
def revision_parents(self, revid):
702
return self.get_inventory_weave().parent_names(revid)
703
704
@needs_write_lock
705
def set_make_working_trees(self, new_value):
706
"""Set the policy flag for making working trees when creating branches.
707
708
This only applies to branches that use this repository.
709
710
The default is 'True'.
711
:param new_value: True to restore the default, False to disable making
712
working trees.
713
"""
714
raise NotImplementedError(self.set_make_working_trees)
715
716
def make_working_trees(self):
717
"""Returns the policy for making working trees on new branches."""
718
raise NotImplementedError(self.make_working_trees)
719
720
@needs_write_lock
721
def sign_revision(self, revision_id, gpg_strategy):
722
plaintext = Testament.from_revision(self, revision_id).as_short_text()
723
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
724
725
@needs_read_lock
726
def has_signature_for_revision_id(self, revision_id):
727
"""Query for a revision signature for revision_id in the repository."""
728
return self._revision_store.has_signature(revision_id,
729
self.get_transaction())
730
731
@needs_read_lock
732
def get_signature_text(self, revision_id):
733
"""Return the text for a signature."""
734
return self._revision_store.get_signature_text(revision_id,
735
self.get_transaction())
736
737
@needs_read_lock
738
def check(self, revision_ids):
739
"""Check consistency of all history of given revision_ids.
740
741
Different repository implementations should override _check().
742
743
:param revision_ids: A non-empty list of revision_ids whose ancestry
744
will be checked. Typically the last revision_id of a branch.
745
"""
746
if not revision_ids:
747
raise ValueError("revision_ids must be non-empty in %s.check"
748
% (self,))
749
return self._check(revision_ids)
750
751
def _check(self, revision_ids):
752
result = check.Check(self)
753
result.check()
754
return result
755
756
def _warn_if_deprecated(self):
757
global _deprecation_warning_done
758
if _deprecation_warning_done:
759
return
760
_deprecation_warning_done = True
761
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
762
% (self._format, self.bzrdir.transport.base))
763
764
def supports_rich_root(self):
765
return self._format.rich_root_data
766
767
def _check_ascii_revisionid(self, revision_id, method):
768
"""Private helper for ascii-only repositories."""
769
# weave repositories refuse to store revisionids that are non-ascii.
770
if revision_id is not None:
771
# weaves require ascii revision ids.
772
if isinstance(revision_id, unicode):
773
try:
774
revision_id.encode('ascii')
775
except UnicodeEncodeError:
776
raise errors.NonAsciiRevisionId(method, self)
777
778
779
class AllInOneRepository(Repository):
780
"""Legacy support - the repository behaviour for all-in-one branches."""
781
782
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
783
# we reuse one control files instance.
784
dir_mode = a_bzrdir._control_files._dir_mode
785
file_mode = a_bzrdir._control_files._file_mode
786
787
def get_store(name, compressed=True, prefixed=False):
788
# FIXME: This approach of assuming stores are all entirely compressed
789
# or entirely uncompressed is tidy, but breaks upgrade from
790
# some existing branches where there's a mixture; we probably
791
# still want the option to look for both.
792
relpath = a_bzrdir._control_files._escape(name)
793
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
794
prefixed=prefixed, compressed=compressed,
795
dir_mode=dir_mode,
796
file_mode=file_mode)
797
#if self._transport.should_cache():
798
# cache_path = os.path.join(self.cache_root, name)
799
# os.mkdir(cache_path)
800
# store = bzrlib.store.CachedStore(store, cache_path)
801
return store
802
803
# not broken out yet because the controlweaves|inventory_store
804
# and text_store | weave_store bits are still different.
805
if isinstance(_format, RepositoryFormat4):
806
# cannot remove these - there is still no consistent api
807
# which allows access to this old info.
808
self.inventory_store = get_store('inventory-store')
809
text_store = get_store('text-store')
810
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
811
812
def get_commit_builder(self, branch, parents, config, timestamp=None,
813
timezone=None, committer=None, revprops=None,
814
revision_id=None):
815
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
816
return Repository.get_commit_builder(self, branch, parents, config,
817
timestamp, timezone, committer, revprops, revision_id)
818
819
@needs_read_lock
820
def is_shared(self):
821
"""AllInOne repositories cannot be shared."""
822
return False
823
824
@needs_write_lock
825
def set_make_working_trees(self, new_value):
826
"""Set the policy flag for making working trees when creating branches.
827
828
This only applies to branches that use this repository.
829
830
The default is 'True'.
831
:param new_value: True to restore the default, False to disable making
832
working trees.
833
"""
834
raise NotImplementedError(self.set_make_working_trees)
835
836
def make_working_trees(self):
837
"""Returns the policy for making working trees on new branches."""
838
return True
839
840
841
def install_revision(repository, rev, revision_tree):
842
"""Install all revision data into a repository."""
843
present_parents = []
844
parent_trees = {}
845
for p_id in rev.parent_ids:
846
if repository.has_revision(p_id):
847
present_parents.append(p_id)
848
parent_trees[p_id] = repository.revision_tree(p_id)
849
else:
850
parent_trees[p_id] = repository.revision_tree(None)
851
852
inv = revision_tree.inventory
853
entries = inv.iter_entries()
854
# backwards compatability hack: skip the root id.
855
if not repository.supports_rich_root():
856
path, root = entries.next()
857
if root.revision != rev.revision_id:
858
raise errors.IncompatibleRevision(repr(repository))
859
# Add the texts that are not already present
860
for path, ie in entries:
861
w = repository.weave_store.get_weave_or_empty(ie.file_id,
862
repository.get_transaction())
863
if ie.revision not in w:
864
text_parents = []
865
# FIXME: TODO: The following loop *may* be overlapping/duplicate
866
# with InventoryEntry.find_previous_heads(). if it is, then there
867
# is a latent bug here where the parents may have ancestors of each
868
# other. RBC, AB
869
for revision, tree in parent_trees.iteritems():
870
if ie.file_id not in tree:
871
continue
872
parent_id = tree.inventory[ie.file_id].revision
873
if parent_id in text_parents:
874
continue
875
text_parents.append(parent_id)
876
877
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
878
repository.get_transaction())
879
lines = revision_tree.get_file(ie.file_id).readlines()
880
vfile.add_lines(rev.revision_id, text_parents, lines)
881
try:
882
# install the inventory
883
repository.add_inventory(rev.revision_id, inv, present_parents)
884
except errors.RevisionAlreadyPresent:
885
pass
886
repository.add_revision(rev.revision_id, rev, inv)
887
888
889
class MetaDirRepository(Repository):
890
"""Repositories in the new meta-dir layout."""
891
892
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
893
super(MetaDirRepository, self).__init__(_format,
894
a_bzrdir,
895
control_files,
896
_revision_store,
897
control_store,
898
text_store)
899
dir_mode = self.control_files._dir_mode
900
file_mode = self.control_files._file_mode
901
902
@needs_read_lock
903
def is_shared(self):
904
"""Return True if this repository is flagged as a shared repository."""
905
return self.control_files._transport.has('shared-storage')
906
907
@needs_write_lock
908
def set_make_working_trees(self, new_value):
909
"""Set the policy flag for making working trees when creating branches.
910
911
This only applies to branches that use this repository.
912
913
The default is 'True'.
914
:param new_value: True to restore the default, False to disable making
915
working trees.
916
"""
917
if new_value:
918
try:
919
self.control_files._transport.delete('no-working-trees')
920
except errors.NoSuchFile:
921
pass
922
else:
923
self.control_files.put_utf8('no-working-trees', '')
924
925
def make_working_trees(self):
926
"""Returns the policy for making working trees on new branches."""
927
return not self.control_files._transport.has('no-working-trees')
928
929
930
class WeaveMetaDirRepository(MetaDirRepository):
931
"""A subclass of MetaDirRepository to set weave specific policy."""
932
933
def get_commit_builder(self, branch, parents, config, timestamp=None,
934
timezone=None, committer=None, revprops=None,
935
revision_id=None):
936
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
937
return MetaDirRepository.get_commit_builder(self, branch, parents,
938
config, timestamp, timezone, committer, revprops, revision_id)
939
940
941
class KnitRepository(MetaDirRepository):
942
"""Knit format repository."""
943
944
def _warn_if_deprecated(self):
945
# This class isn't deprecated
946
pass
947
948
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
949
inv_vf.add_lines_with_ghosts(revid, parents, lines)
950
951
@needs_read_lock
952
def _all_revision_ids(self):
953
"""See Repository.all_revision_ids()."""
954
# Knits get the revision graph from the index of the revision knit, so
955
# it's always possible even if they're on an unlistable transport.
956
return self._revision_store.all_revision_ids(self.get_transaction())
957
958
def fileid_involved_between_revs(self, from_revid, to_revid):
959
"""Find file_id(s) which are involved in the changes between revisions.
960
961
This determines the set of revisions which are involved, and then
962
finds all file ids affected by those revisions.
963
"""
964
vf = self._get_revision_vf()
965
from_set = set(vf.get_ancestry(from_revid))
966
to_set = set(vf.get_ancestry(to_revid))
967
changed = to_set.difference(from_set)
968
return self._fileid_involved_by_set(changed)
969
970
def fileid_involved(self, last_revid=None):
971
"""Find all file_ids modified in the ancestry of last_revid.
972
973
:param last_revid: If None, last_revision() will be used.
974
"""
975
if not last_revid:
976
changed = set(self.all_revision_ids())
977
else:
978
changed = set(self.get_ancestry(last_revid))
979
if None in changed:
980
changed.remove(None)
981
return self._fileid_involved_by_set(changed)
982
983
@needs_read_lock
984
def get_ancestry(self, revision_id):
985
"""Return a list of revision-ids integrated by a revision.
986
987
This is topologically sorted.
988
"""
989
if revision_id is None:
990
return [None]
991
vf = self._get_revision_vf()
992
try:
993
return [None] + vf.get_ancestry(revision_id)
994
except errors.RevisionNotPresent:
995
raise errors.NoSuchRevision(self, revision_id)
996
997
@needs_read_lock
998
def get_revision(self, revision_id):
999
"""Return the Revision object for a named revision"""
1000
return self.get_revision_reconcile(revision_id)
1001
1002
@needs_read_lock
1003
def get_revision_graph(self, revision_id=None):
1004
"""Return a dictionary containing the revision graph.
1005
1006
:param revision_id: The revision_id to get a graph from. If None, then
1007
the entire revision graph is returned. This is a deprecated mode of
1008
operation and will be removed in the future.
1009
:return: a dictionary of revision_id->revision_parents_list.
1010
"""
1011
# special case NULL_REVISION
1012
if revision_id == _mod_revision.NULL_REVISION:
1013
return {}
1014
a_weave = self._get_revision_vf()
1015
entire_graph = a_weave.get_graph()
1016
if revision_id is None:
1017
return a_weave.get_graph()
1018
elif revision_id not in a_weave:
1019
raise errors.NoSuchRevision(self, revision_id)
1020
else:
1021
# add what can be reached from revision_id
1022
result = {}
1023
pending = set([revision_id])
1024
while len(pending) > 0:
1025
node = pending.pop()
1026
result[node] = a_weave.get_parents(node)
1027
for revision_id in result[node]:
1028
if revision_id not in result:
1029
pending.add(revision_id)
1030
return result
1031
1032
@needs_read_lock
1033
def get_revision_graph_with_ghosts(self, revision_ids=None):
1034
"""Return a graph of the revisions with ghosts marked as applicable.
1035
1036
:param revision_ids: an iterable of revisions to graph or None for all.
1037
:return: a Graph object with the graph reachable from revision_ids.
1038
"""
1039
result = graph.Graph()
1040
vf = self._get_revision_vf()
1041
versions = set(vf.versions())
1042
if not revision_ids:
1043
pending = set(self.all_revision_ids())
1044
required = set([])
1045
else:
1046
pending = set(revision_ids)
1047
# special case NULL_REVISION
1048
if _mod_revision.NULL_REVISION in pending:
1049
pending.remove(_mod_revision.NULL_REVISION)
1050
required = set(pending)
1051
done = set([])
1052
while len(pending):
1053
revision_id = pending.pop()
1054
if not revision_id in versions:
1055
if revision_id in required:
1056
raise errors.NoSuchRevision(self, revision_id)
1057
# a ghost
1058
result.add_ghost(revision_id)
1059
# mark it as done so we don't try for it again.
1060
done.add(revision_id)
1061
continue
1062
parent_ids = vf.get_parents_with_ghosts(revision_id)
1063
for parent_id in parent_ids:
1064
# is this queued or done ?
1065
if (parent_id not in pending and
1066
parent_id not in done):
1067
# no, queue it.
1068
pending.add(parent_id)
1069
result.add_node(revision_id, parent_ids)
1070
done.add(revision_id)
1071
return result
1072
1073
def _get_revision_vf(self):
1074
""":return: a versioned file containing the revisions."""
1075
vf = self._revision_store.get_revision_file(self.get_transaction())
1076
return vf
1077
1078
@needs_write_lock
1079
def reconcile(self, other=None, thorough=False):
1080
"""Reconcile this repository."""
1081
from bzrlib.reconcile import KnitReconciler
1082
reconciler = KnitReconciler(self, thorough=thorough)
1083
reconciler.reconcile()
1084
return reconciler
1085
1086
def revision_parents(self, revision_id):
1087
return self._get_revision_vf().get_parents(revision_id)
1088
1089
1090
class KnitRepository2(KnitRepository):
1091
""""""
1092
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1093
control_store, text_store):
1094
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1095
_revision_store, control_store, text_store)
1096
self._serializer = xml6.serializer_v6
1097
1098
def deserialise_inventory(self, revision_id, xml):
1099
"""Transform the xml into an inventory object.
1100
1101
:param revision_id: The expected revision id of the inventory.
1102
:param xml: A serialised inventory.
1103
"""
1104
result = self._serializer.read_inventory_from_string(xml)
1105
assert result.root.revision is not None
1106
return result
1107
1108
def serialise_inventory(self, inv):
1109
"""Transform the inventory object into XML text.
1110
1111
:param revision_id: The expected revision id of the inventory.
1112
:param xml: A serialised inventory.
1113
"""
1114
assert inv.revision_id is not None
1115
assert inv.root.revision is not None
1116
return KnitRepository.serialise_inventory(self, inv)
1117
1118
def get_commit_builder(self, branch, parents, config, timestamp=None,
1119
timezone=None, committer=None, revprops=None,
1120
revision_id=None):
1121
"""Obtain a CommitBuilder for this repository.
1122
1123
:param branch: Branch to commit to.
1124
:param parents: Revision ids of the parents of the new revision.
1125
:param config: Configuration to use.
1126
:param timestamp: Optional timestamp recorded for commit.
1127
:param timezone: Optional timezone for timestamp.
1128
:param committer: Optional committer to set for commit.
1129
:param revprops: Optional dictionary of revision properties.
1130
:param revision_id: Optional revision id.
1131
"""
1132
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1133
committer, revprops, revision_id)
1134
1135
1136
class RepositoryFormat(object):
1137
"""A repository format.
1138
1139
Formats provide three things:
1140
* An initialization routine to construct repository data on disk.
1141
* a format string which is used when the BzrDir supports versioned
1142
children.
1143
* an open routine which returns a Repository instance.
1144
1145
Formats are placed in an dict by their format string for reference
1146
during opening. These should be subclasses of RepositoryFormat
1147
for consistency.
1148
1149
Once a format is deprecated, just deprecate the initialize and open
1150
methods on the format class. Do not deprecate the object, as the
1151
object will be created every system load.
1152
1153
Common instance attributes:
1154
_matchingbzrdir - the bzrdir format that the repository format was
1155
originally written to work with. This can be used if manually
1156
constructing a bzrdir and repository, or more commonly for test suite
1157
parameterisation.
1158
"""
1159
1160
_default_format = None
1161
"""The default format used for new repositories."""
1162
1163
_formats = {}
1164
"""The known formats."""
1165
1166
def __str__(self):
1167
return "<%s>" % self.__class__.__name__
1168
1169
@classmethod
1170
def find_format(klass, a_bzrdir):
1171
"""Return the format for the repository object in a_bzrdir."""
1172
try:
1173
transport = a_bzrdir.get_repository_transport(None)
1174
format_string = transport.get("format").read()
1175
return klass._formats[format_string]
1176
except errors.NoSuchFile:
1177
raise errors.NoRepositoryPresent(a_bzrdir)
1178
except KeyError:
1179
raise errors.UnknownFormatError(format=format_string)
1180
1181
def _get_control_store(self, repo_transport, control_files):
1182
"""Return the control store for this repository."""
1183
raise NotImplementedError(self._get_control_store)
1184
1185
@classmethod
1186
def get_default_format(klass):
1187
"""Return the current default format."""
1188
return klass._default_format
1189
1190
def get_format_string(self):
1191
"""Return the ASCII format string that identifies this format.
1192
1193
Note that in pre format ?? repositories the format string is
1194
not permitted nor written to disk.
1195
"""
1196
raise NotImplementedError(self.get_format_string)
1197
1198
def get_format_description(self):
1199
"""Return the short description for this format."""
1200
raise NotImplementedError(self.get_format_description)
1201
1202
def _get_revision_store(self, repo_transport, control_files):
1203
"""Return the revision store object for this a_bzrdir."""
1204
raise NotImplementedError(self._get_revision_store)
1205
1206
def _get_text_rev_store(self,
1207
transport,
1208
control_files,
1209
name,
1210
compressed=True,
1211
prefixed=False,
1212
serializer=None):
1213
"""Common logic for getting a revision store for a repository.
1214
1215
see self._get_revision_store for the subclass-overridable method to
1216
get the store for a repository.
1217
"""
1218
from bzrlib.store.revision.text import TextRevisionStore
1219
dir_mode = control_files._dir_mode
1220
file_mode = control_files._file_mode
1221
text_store =TextStore(transport.clone(name),
1222
prefixed=prefixed,
1223
compressed=compressed,
1224
dir_mode=dir_mode,
1225
file_mode=file_mode)
1226
_revision_store = TextRevisionStore(text_store, serializer)
1227
return _revision_store
1228
1229
def _get_versioned_file_store(self,
1230
name,
1231
transport,
1232
control_files,
1233
prefixed=True,
1234
versionedfile_class=weave.WeaveFile,
1235
versionedfile_kwargs={},
1236
escaped=False):
1237
weave_transport = control_files._transport.clone(name)
1238
dir_mode = control_files._dir_mode
1239
file_mode = control_files._file_mode
1240
return VersionedFileStore(weave_transport, prefixed=prefixed,
1241
dir_mode=dir_mode,
1242
file_mode=file_mode,
1243
versionedfile_class=versionedfile_class,
1244
versionedfile_kwargs=versionedfile_kwargs,
1245
escaped=escaped)
1246
1247
def initialize(self, a_bzrdir, shared=False):
1248
"""Initialize a repository of this format in a_bzrdir.
1249
1250
:param a_bzrdir: The bzrdir to put the new repository in it.
1251
:param shared: The repository should be initialized as a sharable one.
1252
1253
This may raise UninitializableFormat if shared repository are not
1254
compatible the a_bzrdir.
1255
"""
1256
1257
def is_supported(self):
1258
"""Is this format supported?
1259
1260
Supported formats must be initializable and openable.
1261
Unsupported formats may not support initialization or committing or
1262
some other features depending on the reason for not being supported.
1263
"""
1264
return True
1265
1266
def check_conversion_target(self, target_format):
1267
raise NotImplementedError(self.check_conversion_target)
1268
1269
def open(self, a_bzrdir, _found=False):
1270
"""Return an instance of this format for the bzrdir a_bzrdir.
1271
1272
_found is a private parameter, do not use it.
1273
"""
1274
raise NotImplementedError(self.open)
1275
1276
@classmethod
1277
def register_format(klass, format):
1278
klass._formats[format.get_format_string()] = format
1279
1280
@classmethod
1281
def set_default_format(klass, format):
1282
klass._default_format = format
1283
1284
@classmethod
1285
def unregister_format(klass, format):
1286
assert klass._formats[format.get_format_string()] is format
1287
del klass._formats[format.get_format_string()]
1288
1289
1290
class PreSplitOutRepositoryFormat(RepositoryFormat):
1291
"""Base class for the pre split out repository formats."""
1292
1293
rich_root_data = False
1294
1295
def initialize(self, a_bzrdir, shared=False, _internal=False):
1296
"""Create a weave repository.
1297
1298
TODO: when creating split out bzr branch formats, move this to a common
1299
base for Format5, Format6. or something like that.
1300
"""
1301
if shared:
1302
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1303
1304
if not _internal:
1305
# always initialized when the bzrdir is.
1306
return self.open(a_bzrdir, _found=True)
1307
1308
# Create an empty weave
1309
sio = StringIO()
1310
weavefile.write_weave_v5(weave.Weave(), sio)
1311
empty_weave = sio.getvalue()
1312
1313
mutter('creating repository in %s.', a_bzrdir.transport.base)
1314
dirs = ['revision-store', 'weaves']
1315
files = [('inventory.weave', StringIO(empty_weave)),
1316
]
1317
1318
# FIXME: RBC 20060125 don't peek under the covers
1319
# NB: no need to escape relative paths that are url safe.
1320
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1321
'branch-lock', lockable_files.TransportLock)
1322
control_files.create_lock()
1323
control_files.lock_write()
1324
control_files._transport.mkdir_multi(dirs,
1325
mode=control_files._dir_mode)
1326
try:
1327
for file, content in files:
1328
control_files.put(file, content)
1329
finally:
1330
control_files.unlock()
1331
return self.open(a_bzrdir, _found=True)
1332
1333
def _get_control_store(self, repo_transport, control_files):
1334
"""Return the control store for this repository."""
1335
return self._get_versioned_file_store('',
1336
repo_transport,
1337
control_files,
1338
prefixed=False)
1339
1340
def _get_text_store(self, transport, control_files):
1341
"""Get a store for file texts for this format."""
1342
raise NotImplementedError(self._get_text_store)
1343
1344
def open(self, a_bzrdir, _found=False):
1345
"""See RepositoryFormat.open()."""
1346
if not _found:
1347
# we are being called directly and must probe.
1348
raise NotImplementedError
1349
1350
repo_transport = a_bzrdir.get_repository_transport(None)
1351
control_files = a_bzrdir._control_files
1352
text_store = self._get_text_store(repo_transport, control_files)
1353
control_store = self._get_control_store(repo_transport, control_files)
1354
_revision_store = self._get_revision_store(repo_transport, control_files)
1355
return AllInOneRepository(_format=self,
1356
a_bzrdir=a_bzrdir,
1357
_revision_store=_revision_store,
1358
control_store=control_store,
1359
text_store=text_store)
1360
1361
def check_conversion_target(self, target_format):
1362
pass
1363
1364
1365
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1366
"""Bzr repository format 4.
1367
1368
This repository format has:
1369
- flat stores
1370
- TextStores for texts, inventories,revisions.
1371
1372
This format is deprecated: it indexes texts using a text id which is
1373
removed in format 5; initialization and write support for this format
1374
has been removed.
1375
"""
1376
1377
def __init__(self):
1378
super(RepositoryFormat4, self).__init__()
1379
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1380
1381
def get_format_description(self):
1382
"""See RepositoryFormat.get_format_description()."""
1383
return "Repository format 4"
1384
1385
def initialize(self, url, shared=False, _internal=False):
1386
"""Format 4 branches cannot be created."""
1387
raise errors.UninitializableFormat(self)
1388
1389
def is_supported(self):
1390
"""Format 4 is not supported.
1391
1392
It is not supported because the model changed from 4 to 5 and the
1393
conversion logic is expensive - so doing it on the fly was not
1394
feasible.
1395
"""
1396
return False
1397
1398
def _get_control_store(self, repo_transport, control_files):
1399
"""Format 4 repositories have no formal control store at this point.
1400
1401
This will cause any control-file-needing apis to fail - this is desired.
1402
"""
1403
return None
1404
1405
def _get_revision_store(self, repo_transport, control_files):
1406
"""See RepositoryFormat._get_revision_store()."""
1407
from bzrlib.xml4 import serializer_v4
1408
return self._get_text_rev_store(repo_transport,
1409
control_files,
1410
'revision-store',
1411
serializer=serializer_v4)
1412
1413
def _get_text_store(self, transport, control_files):
1414
"""See RepositoryFormat._get_text_store()."""
1415
1416
1417
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1418
"""Bzr control format 5.
1419
1420
This repository format has:
1421
- weaves for file texts and inventory
1422
- flat stores
1423
- TextStores for revisions and signatures.
1424
"""
1425
1426
def __init__(self):
1427
super(RepositoryFormat5, self).__init__()
1428
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1429
1430
def get_format_description(self):
1431
"""See RepositoryFormat.get_format_description()."""
1432
return "Weave repository format 5"
1433
1434
def _get_revision_store(self, repo_transport, control_files):
1435
"""See RepositoryFormat._get_revision_store()."""
1436
"""Return the revision store object for this a_bzrdir."""
1437
return self._get_text_rev_store(repo_transport,
1438
control_files,
1439
'revision-store',
1440
compressed=False)
1441
1442
def _get_text_store(self, transport, control_files):
1443
"""See RepositoryFormat._get_text_store()."""
1444
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1445
1446
1447
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1448
"""Bzr control format 6.
1449
1450
This repository format has:
1451
- weaves for file texts and inventory
1452
- hash subdirectory based stores.
1453
- TextStores for revisions and signatures.
1454
"""
1455
1456
def __init__(self):
1457
super(RepositoryFormat6, self).__init__()
1458
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1459
1460
def get_format_description(self):
1461
"""See RepositoryFormat.get_format_description()."""
1462
return "Weave repository format 6"
1463
1464
def _get_revision_store(self, repo_transport, control_files):
1465
"""See RepositoryFormat._get_revision_store()."""
1466
return self._get_text_rev_store(repo_transport,
1467
control_files,
1468
'revision-store',
1469
compressed=False,
1470
prefixed=True)
1471
1472
def _get_text_store(self, transport, control_files):
1473
"""See RepositoryFormat._get_text_store()."""
1474
return self._get_versioned_file_store('weaves', transport, control_files)
1475
1476
1477
class MetaDirRepositoryFormat(RepositoryFormat):
1478
"""Common base class for the new repositories using the metadir layout."""
1479
1480
rich_root_data = False
1481
1482
def __init__(self):
1483
super(MetaDirRepositoryFormat, self).__init__()
1484
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1485
1486
def _create_control_files(self, a_bzrdir):
1487
"""Create the required files and the initial control_files object."""
1488
# FIXME: RBC 20060125 don't peek under the covers
1489
# NB: no need to escape relative paths that are url safe.
1490
repository_transport = a_bzrdir.get_repository_transport(self)
1491
control_files = lockable_files.LockableFiles(repository_transport,
1492
'lock', lockdir.LockDir)
1493
control_files.create_lock()
1494
return control_files
1495
1496
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1497
"""Upload the initial blank content."""
1498
control_files = self._create_control_files(a_bzrdir)
1499
control_files.lock_write()
1500
try:
1501
control_files._transport.mkdir_multi(dirs,
1502
mode=control_files._dir_mode)
1503
for file, content in files:
1504
control_files.put(file, content)
1505
for file, content in utf8_files:
1506
control_files.put_utf8(file, content)
1507
if shared == True:
1508
control_files.put_utf8('shared-storage', '')
1509
finally:
1510
control_files.unlock()
1511
1512
1513
class RepositoryFormat7(MetaDirRepositoryFormat):
1514
"""Bzr repository 7.
1515
1516
This repository format has:
1517
- weaves for file texts and inventory
1518
- hash subdirectory based stores.
1519
- TextStores for revisions and signatures.
1520
- a format marker of its own
1521
- an optional 'shared-storage' flag
1522
- an optional 'no-working-trees' flag
1523
"""
1524
1525
def _get_control_store(self, repo_transport, control_files):
1526
"""Return the control store for this repository."""
1527
return self._get_versioned_file_store('',
1528
repo_transport,
1529
control_files,
1530
prefixed=False)
1531
1532
def get_format_string(self):
1533
"""See RepositoryFormat.get_format_string()."""
1534
return "Bazaar-NG Repository format 7"
1535
1536
def get_format_description(self):
1537
"""See RepositoryFormat.get_format_description()."""
1538
return "Weave repository format 7"
1539
1540
def check_conversion_target(self, target_format):
1541
pass
1542
1543
def _get_revision_store(self, repo_transport, control_files):
1544
"""See RepositoryFormat._get_revision_store()."""
1545
return self._get_text_rev_store(repo_transport,
1546
control_files,
1547
'revision-store',
1548
compressed=False,
1549
prefixed=True,
1550
)
1551
1552
def _get_text_store(self, transport, control_files):
1553
"""See RepositoryFormat._get_text_store()."""
1554
return self._get_versioned_file_store('weaves',
1555
transport,
1556
control_files)
1557
1558
def initialize(self, a_bzrdir, shared=False):
1559
"""Create a weave repository.
1560
1561
:param shared: If true the repository will be initialized as a shared
1562
repository.
1563
"""
1564
# Create an empty weave
1565
sio = StringIO()
1566
weavefile.write_weave_v5(weave.Weave(), sio)
1567
empty_weave = sio.getvalue()
1568
1569
mutter('creating repository in %s.', a_bzrdir.transport.base)
1570
dirs = ['revision-store', 'weaves']
1571
files = [('inventory.weave', StringIO(empty_weave)),
1572
]
1573
utf8_files = [('format', self.get_format_string())]
1574
1575
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1576
return self.open(a_bzrdir=a_bzrdir, _found=True)
1577
1578
def open(self, a_bzrdir, _found=False, _override_transport=None):
1579
"""See RepositoryFormat.open().
1580
1581
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1582
repository at a slightly different url
1583
than normal. I.e. during 'upgrade'.
1584
"""
1585
if not _found:
1586
format = RepositoryFormat.find_format(a_bzrdir)
1587
assert format.__class__ == self.__class__
1588
if _override_transport is not None:
1589
repo_transport = _override_transport
1590
else:
1591
repo_transport = a_bzrdir.get_repository_transport(None)
1592
control_files = lockable_files.LockableFiles(repo_transport,
1593
'lock', lockdir.LockDir)
1594
text_store = self._get_text_store(repo_transport, control_files)
1595
control_store = self._get_control_store(repo_transport, control_files)
1596
_revision_store = self._get_revision_store(repo_transport, control_files)
1597
return WeaveMetaDirRepository(_format=self,
1598
a_bzrdir=a_bzrdir,
1599
control_files=control_files,
1600
_revision_store=_revision_store,
1601
control_store=control_store,
1602
text_store=text_store)
1603
1604
1605
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1606
"""Bzr repository knit format (generalized).
1607
1608
This repository format has:
1609
- knits for file texts and inventory
1610
- hash subdirectory based stores.
1611
- knits for revisions and signatures
1612
- TextStores for revisions and signatures.
1613
- a format marker of its own
1614
- an optional 'shared-storage' flag
1615
- an optional 'no-working-trees' flag
1616
- a LockDir lock
1617
"""
1618
1619
def _get_control_store(self, repo_transport, control_files):
1620
"""Return the control store for this repository."""
1621
return VersionedFileStore(
1622
repo_transport,
1623
prefixed=False,
1624
file_mode=control_files._file_mode,
1625
versionedfile_class=knit.KnitVersionedFile,
1626
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1627
)
1628
1629
def _get_revision_store(self, repo_transport, control_files):
1630
"""See RepositoryFormat._get_revision_store()."""
1631
from bzrlib.store.revision.knit import KnitRevisionStore
1632
versioned_file_store = VersionedFileStore(
1633
repo_transport,
1634
file_mode=control_files._file_mode,
1635
prefixed=False,
1636
precious=True,
1637
versionedfile_class=knit.KnitVersionedFile,
1638
versionedfile_kwargs={'delta':False,
1639
'factory':knit.KnitPlainFactory(),
1640
},
1641
escaped=True,
1642
)
1643
return KnitRevisionStore(versioned_file_store)
1644
1645
def _get_text_store(self, transport, control_files):
1646
"""See RepositoryFormat._get_text_store()."""
1647
return self._get_versioned_file_store('knits',
1648
transport,
1649
control_files,
1650
versionedfile_class=knit.KnitVersionedFile,
1651
versionedfile_kwargs={
1652
'create_parent_dir':True,
1653
'delay_create':True,
1654
'dir_mode':control_files._dir_mode,
1655
},
1656
escaped=True)
1657
1658
def initialize(self, a_bzrdir, shared=False):
1659
"""Create a knit format 1 repository.
1660
1661
:param a_bzrdir: bzrdir to contain the new repository; must already
1662
be initialized.
1663
:param shared: If true the repository will be initialized as a shared
1664
repository.
1665
"""
1666
mutter('creating repository in %s.', a_bzrdir.transport.base)
1667
dirs = ['revision-store', 'knits']
1668
files = []
1669
utf8_files = [('format', self.get_format_string())]
1670
1671
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1672
repo_transport = a_bzrdir.get_repository_transport(None)
1673
control_files = lockable_files.LockableFiles(repo_transport,
1674
'lock', lockdir.LockDir)
1675
control_store = self._get_control_store(repo_transport, control_files)
1676
transaction = transactions.WriteTransaction()
1677
# trigger a write of the inventory store.
1678
control_store.get_weave_or_empty('inventory', transaction)
1679
_revision_store = self._get_revision_store(repo_transport, control_files)
1680
# the revision id here is irrelevant: it will not be stored, and cannot
1681
# already exist.
1682
_revision_store.has_revision_id('A', transaction)
1683
_revision_store.get_signature_file(transaction)
1684
return self.open(a_bzrdir=a_bzrdir, _found=True)
1685
1686
def open(self, a_bzrdir, _found=False, _override_transport=None):
1687
"""See RepositoryFormat.open().
1688
1689
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1690
repository at a slightly different url
1691
than normal. I.e. during 'upgrade'.
1692
"""
1693
if not _found:
1694
format = RepositoryFormat.find_format(a_bzrdir)
1695
assert format.__class__ == self.__class__
1696
if _override_transport is not None:
1697
repo_transport = _override_transport
1698
else:
1699
repo_transport = a_bzrdir.get_repository_transport(None)
1700
control_files = lockable_files.LockableFiles(repo_transport,
1701
'lock', lockdir.LockDir)
1702
text_store = self._get_text_store(repo_transport, control_files)
1703
control_store = self._get_control_store(repo_transport, control_files)
1704
_revision_store = self._get_revision_store(repo_transport, control_files)
1705
return KnitRepository(_format=self,
1706
a_bzrdir=a_bzrdir,
1707
control_files=control_files,
1708
_revision_store=_revision_store,
1709
control_store=control_store,
1710
text_store=text_store)
1711
1712
1713
class RepositoryFormatKnit1(RepositoryFormatKnit):
1714
"""Bzr repository knit format 1.
1715
1716
This repository format has:
1717
- knits for file texts and inventory
1718
- hash subdirectory based stores.
1719
- knits for revisions and signatures
1720
- TextStores for revisions and signatures.
1721
- a format marker of its own
1722
- an optional 'shared-storage' flag
1723
- an optional 'no-working-trees' flag
1724
- a LockDir lock
1725
1726
This format was introduced in bzr 0.8.
1727
"""
1728
def get_format_string(self):
1729
"""See RepositoryFormat.get_format_string()."""
1730
return "Bazaar-NG Knit Repository Format 1"
1731
1732
def get_format_description(self):
1733
"""See RepositoryFormat.get_format_description()."""
1734
return "Knit repository format 1"
1735
1736
def check_conversion_target(self, target_format):
1737
pass
1738
1739
1740
class RepositoryFormatKnit2(RepositoryFormatKnit):
1741
"""Bzr repository knit format 2.
1742
1743
THIS FORMAT IS EXPERIMENTAL
1744
This repository format has:
1745
- knits for file texts and inventory
1746
- hash subdirectory based stores.
1747
- knits for revisions and signatures
1748
- TextStores for revisions and signatures.
1749
- a format marker of its own
1750
- an optional 'shared-storage' flag
1751
- an optional 'no-working-trees' flag
1752
- a LockDir lock
1753
- Support for recording full info about the tree root
1754
1755
"""
1756
1757
rich_root_data = True
1758
1759
def get_format_string(self):
1760
"""See RepositoryFormat.get_format_string()."""
1761
return "Bazaar Knit Repository Format 2\n"
1762
1763
def get_format_description(self):
1764
"""See RepositoryFormat.get_format_description()."""
1765
return "Knit repository format 2"
1766
1767
def check_conversion_target(self, target_format):
1768
if not target_format.rich_root_data:
1769
raise errors.BadConversionTarget(
1770
'Does not support rich root data.', target_format)
1771
1772
def open(self, a_bzrdir, _found=False, _override_transport=None):
1773
"""See RepositoryFormat.open().
1774
1775
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1776
repository at a slightly different url
1777
than normal. I.e. during 'upgrade'.
1778
"""
1779
if not _found:
1780
format = RepositoryFormat.find_format(a_bzrdir)
1781
assert format.__class__ == self.__class__
1782
if _override_transport is not None:
1783
repo_transport = _override_transport
1784
else:
1785
repo_transport = a_bzrdir.get_repository_transport(None)
1786
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1787
lockdir.LockDir)
1788
text_store = self._get_text_store(repo_transport, control_files)
1789
control_store = self._get_control_store(repo_transport, control_files)
1790
_revision_store = self._get_revision_store(repo_transport, control_files)
1791
return KnitRepository2(_format=self,
1792
a_bzrdir=a_bzrdir,
1793
control_files=control_files,
1794
_revision_store=_revision_store,
1795
control_store=control_store,
1796
text_store=text_store)
1797
1798
1799
1800
# formats which have no format string are not discoverable
1801
# and not independently creatable, so are not registered.
1802
RepositoryFormat.register_format(RepositoryFormat7())
1803
# KEEP in sync with bzrdir.format_registry default
1804
_default_format = RepositoryFormatKnit1()
1805
RepositoryFormat.register_format(_default_format)
1806
RepositoryFormat.register_format(RepositoryFormatKnit2())
1807
RepositoryFormat.set_default_format(_default_format)
1808
_legacy_formats = [RepositoryFormat4(),
1809
RepositoryFormat5(),
1810
RepositoryFormat6()]
1811
1812
1813
class InterRepository(InterObject):
1814
"""This class represents operations taking place between two repositories.
1815
1816
Its instances have methods like copy_content and fetch, and contain
1817
references to the source and target repositories these operations can be
1818
carried out on.
1819
1820
Often we will provide convenience methods on 'repository' which carry out
1821
operations with another repository - they will always forward to
1822
InterRepository.get(other).method_name(parameters).
1823
"""
1824
1825
_optimisers = []
1826
"""The available optimised InterRepository types."""
1827
1828
def copy_content(self, revision_id=None, basis=None):
1829
raise NotImplementedError(self.copy_content)
1830
1831
def fetch(self, revision_id=None, pb=None):
1832
"""Fetch the content required to construct revision_id.
1833
1834
The content is copied from self.source to self.target.
1835
1836
:param revision_id: if None all content is copied, if NULL_REVISION no
1837
content is copied.
1838
:param pb: optional progress bar to use for progress reports. If not
1839
provided a default one will be created.
1840
1841
Returns the copied revision count and the failed revisions in a tuple:
1842
(copied, failures).
1843
"""
1844
raise NotImplementedError(self.fetch)
1845
1846
@needs_read_lock
1847
def missing_revision_ids(self, revision_id=None):
1848
"""Return the revision ids that source has that target does not.
1849
1850
These are returned in topological order.
1851
1852
:param revision_id: only return revision ids included by this
1853
revision_id.
1854
"""
1855
# generic, possibly worst case, slow code path.
1856
target_ids = set(self.target.all_revision_ids())
1857
if revision_id is not None:
1858
source_ids = self.source.get_ancestry(revision_id)
1859
assert source_ids[0] is None
1860
source_ids.pop(0)
1861
else:
1862
source_ids = self.source.all_revision_ids()
1863
result_set = set(source_ids).difference(target_ids)
1864
# this may look like a no-op: its not. It preserves the ordering
1865
# other_ids had while only returning the members from other_ids
1866
# that we've decided we need.
1867
return [rev_id for rev_id in source_ids if rev_id in result_set]
1868
1869
1870
class InterSameDataRepository(InterRepository):
1871
"""Code for converting between repositories that represent the same data.
1872
1873
Data format and model must match for this to work.
1874
"""
1875
1876
_matching_repo_format = RepositoryFormat4()
1877
"""Repository format for testing with."""
1878
1879
@staticmethod
1880
def is_compatible(source, target):
1881
if not isinstance(source, Repository):
1882
return False
1883
if not isinstance(target, Repository):
1884
return False
1885
if source._format.rich_root_data == target._format.rich_root_data:
1886
return True
1887
else:
1888
return False
1889
1890
@needs_write_lock
1891
def copy_content(self, revision_id=None, basis=None):
1892
"""Make a complete copy of the content in self into destination.
1893
1894
This is a destructive operation! Do not use it on existing
1895
repositories.
1896
1897
:param revision_id: Only copy the content needed to construct
1898
revision_id and its parents.
1899
:param basis: Copy the needed data preferentially from basis.
1900
"""
1901
try:
1902
self.target.set_make_working_trees(self.source.make_working_trees())
1903
except NotImplementedError:
1904
pass
1905
# grab the basis available data
1906
if basis is not None:
1907
self.target.fetch(basis, revision_id=revision_id)
1908
# but don't bother fetching if we have the needed data now.
1909
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1910
self.target.has_revision(revision_id)):
1911
return
1912
self.target.fetch(self.source, revision_id=revision_id)
1913
1914
@needs_write_lock
1915
def fetch(self, revision_id=None, pb=None):
1916
"""See InterRepository.fetch()."""
1917
from bzrlib.fetch import GenericRepoFetcher
1918
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1919
self.source, self.source._format, self.target,
1920
self.target._format)
1921
f = GenericRepoFetcher(to_repository=self.target,
1922
from_repository=self.source,
1923
last_revision=revision_id,
1924
pb=pb)
1925
return f.count_copied, f.failed_revisions
1926
1927
1928
class InterWeaveRepo(InterSameDataRepository):
1929
"""Optimised code paths between Weave based repositories."""
1930
1931
_matching_repo_format = RepositoryFormat7()
1932
"""Repository format for testing with."""
1933
1934
@staticmethod
1935
def is_compatible(source, target):
1936
"""Be compatible with known Weave formats.
1937
1938
We don't test for the stores being of specific types because that
1939
could lead to confusing results, and there is no need to be
1940
overly general.
1941
"""
1942
try:
1943
return (isinstance(source._format, (RepositoryFormat5,
1944
RepositoryFormat6,
1945
RepositoryFormat7)) and
1946
isinstance(target._format, (RepositoryFormat5,
1947
RepositoryFormat6,
1948
RepositoryFormat7)))
1949
except AttributeError:
1950
return False
1951
1952
@needs_write_lock
1953
def copy_content(self, revision_id=None, basis=None):
1954
"""See InterRepository.copy_content()."""
1955
# weave specific optimised path:
1956
if basis is not None:
1957
# copy the basis in, then fetch remaining data.
1958
basis.copy_content_into(self.target, revision_id)
1959
# the basis copy_content_into could miss-set this.
1960
try:
1961
self.target.set_make_working_trees(self.source.make_working_trees())
1962
except NotImplementedError:
1963
pass
1964
self.target.fetch(self.source, revision_id=revision_id)
1965
else:
1966
try:
1967
self.target.set_make_working_trees(self.source.make_working_trees())
1968
except NotImplementedError:
1969
pass
1970
# FIXME do not peek!
1971
if self.source.control_files._transport.listable():
1972
pb = ui.ui_factory.nested_progress_bar()
1973
try:
1974
self.target.weave_store.copy_all_ids(
1975
self.source.weave_store,
1976
pb=pb,
1977
from_transaction=self.source.get_transaction(),
1978
to_transaction=self.target.get_transaction())
1979
pb.update('copying inventory', 0, 1)
1980
self.target.control_weaves.copy_multi(
1981
self.source.control_weaves, ['inventory'],
1982
from_transaction=self.source.get_transaction(),
1983
to_transaction=self.target.get_transaction())
1984
self.target._revision_store.text_store.copy_all_ids(
1985
self.source._revision_store.text_store,
1986
pb=pb)
1987
finally:
1988
pb.finished()
1989
else:
1990
self.target.fetch(self.source, revision_id=revision_id)
1991
1992
@needs_write_lock
1993
def fetch(self, revision_id=None, pb=None):
1994
"""See InterRepository.fetch()."""
1995
from bzrlib.fetch import GenericRepoFetcher
1996
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1997
self.source, self.source._format, self.target, self.target._format)
1998
f = GenericRepoFetcher(to_repository=self.target,
1999
from_repository=self.source,
2000
last_revision=revision_id,
2001
pb=pb)
2002
return f.count_copied, f.failed_revisions
2003
2004
@needs_read_lock
2005
def missing_revision_ids(self, revision_id=None):
2006
"""See InterRepository.missing_revision_ids()."""
2007
# we want all revisions to satisfy revision_id in source.
2008
# but we don't want to stat every file here and there.
2009
# we want then, all revisions other needs to satisfy revision_id
2010
# checked, but not those that we have locally.
2011
# so the first thing is to get a subset of the revisions to
2012
# satisfy revision_id in source, and then eliminate those that
2013
# we do already have.
2014
# this is slow on high latency connection to self, but as as this
2015
# disk format scales terribly for push anyway due to rewriting
2016
# inventory.weave, this is considered acceptable.
2017
# - RBC 20060209
2018
if revision_id is not None:
2019
source_ids = self.source.get_ancestry(revision_id)
2020
assert source_ids[0] is None
2021
source_ids.pop(0)
2022
else:
2023
source_ids = self.source._all_possible_ids()
2024
source_ids_set = set(source_ids)
2025
# source_ids is the worst possible case we may need to pull.
2026
# now we want to filter source_ids against what we actually
2027
# have in target, but don't try to check for existence where we know
2028
# we do not have a revision as that would be pointless.
2029
target_ids = set(self.target._all_possible_ids())
2030
possibly_present_revisions = target_ids.intersection(source_ids_set)
2031
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2032
required_revisions = source_ids_set.difference(actually_present_revisions)
2033
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2034
if revision_id is not None:
2035
# we used get_ancestry to determine source_ids then we are assured all
2036
# revisions referenced are present as they are installed in topological order.
2037
# and the tip revision was validated by get_ancestry.
2038
return required_topo_revisions
2039
else:
2040
# if we just grabbed the possibly available ids, then
2041
# we only have an estimate of whats available and need to validate
2042
# that against the revision records.
2043
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2044
2045
2046
class InterKnitRepo(InterSameDataRepository):
2047
"""Optimised code paths between Knit based repositories."""
2048
2049
_matching_repo_format = RepositoryFormatKnit1()
2050
"""Repository format for testing with."""
2051
2052
@staticmethod
2053
def is_compatible(source, target):
2054
"""Be compatible with known Knit formats.
2055
2056
We don't test for the stores being of specific types because that
2057
could lead to confusing results, and there is no need to be
2058
overly general.
2059
"""
2060
try:
2061
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2062
isinstance(target._format, (RepositoryFormatKnit1)))
2063
except AttributeError:
2064
return False
2065
2066
@needs_write_lock
2067
def fetch(self, revision_id=None, pb=None):
2068
"""See InterRepository.fetch()."""
2069
from bzrlib.fetch import KnitRepoFetcher
2070
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2071
self.source, self.source._format, self.target, self.target._format)
2072
f = KnitRepoFetcher(to_repository=self.target,
2073
from_repository=self.source,
2074
last_revision=revision_id,
2075
pb=pb)
2076
return f.count_copied, f.failed_revisions
2077
2078
@needs_read_lock
2079
def missing_revision_ids(self, revision_id=None):
2080
"""See InterRepository.missing_revision_ids()."""
2081
if revision_id is not None:
2082
source_ids = self.source.get_ancestry(revision_id)
2083
assert source_ids[0] is None
2084
source_ids.pop(0)
2085
else:
2086
source_ids = self.source._all_possible_ids()
2087
source_ids_set = set(source_ids)
2088
# source_ids is the worst possible case we may need to pull.
2089
# now we want to filter source_ids against what we actually
2090
# have in target, but don't try to check for existence where we know
2091
# we do not have a revision as that would be pointless.
2092
target_ids = set(self.target._all_possible_ids())
2093
possibly_present_revisions = target_ids.intersection(source_ids_set)
2094
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2095
required_revisions = source_ids_set.difference(actually_present_revisions)
2096
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2097
if revision_id is not None:
2098
# we used get_ancestry to determine source_ids then we are assured all
2099
# revisions referenced are present as they are installed in topological order.
2100
# and the tip revision was validated by get_ancestry.
2101
return required_topo_revisions
2102
else:
2103
# if we just grabbed the possibly available ids, then
2104
# we only have an estimate of whats available and need to validate
2105
# that against the revision records.
2106
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2107
2108
2109
class InterModel1and2(InterRepository):
2110
2111
_matching_repo_format = None
2112
2113
@staticmethod
2114
def is_compatible(source, target):
2115
if not isinstance(source, Repository):
2116
return False
2117
if not isinstance(target, Repository):
2118
return False
2119
if not source._format.rich_root_data and target._format.rich_root_data:
2120
return True
2121
else:
2122
return False
2123
2124
@needs_write_lock
2125
def fetch(self, revision_id=None, pb=None):
2126
"""See InterRepository.fetch()."""
2127
from bzrlib.fetch import Model1toKnit2Fetcher
2128
f = Model1toKnit2Fetcher(to_repository=self.target,
2129
from_repository=self.source,
2130
last_revision=revision_id,
2131
pb=pb)
2132
return f.count_copied, f.failed_revisions
2133
2134
@needs_write_lock
2135
def copy_content(self, revision_id=None, basis=None):
2136
"""Make a complete copy of the content in self into destination.
2137
2138
This is a destructive operation! Do not use it on existing
2139
repositories.
2140
2141
:param revision_id: Only copy the content needed to construct
2142
revision_id and its parents.
2143
:param basis: Copy the needed data preferentially from basis.
2144
"""
2145
try:
2146
self.target.set_make_working_trees(self.source.make_working_trees())
2147
except NotImplementedError:
2148
pass
2149
# grab the basis available data
2150
if basis is not None:
2151
self.target.fetch(basis, revision_id=revision_id)
2152
# but don't bother fetching if we have the needed data now.
2153
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2154
self.target.has_revision(revision_id)):
2155
return
2156
self.target.fetch(self.source, revision_id=revision_id)
2157
2158
2159
class InterKnit1and2(InterKnitRepo):
2160
2161
_matching_repo_format = None
2162
2163
@staticmethod
2164
def is_compatible(source, target):
2165
"""Be compatible with Knit1 source and Knit2 target"""
2166
try:
2167
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2168
isinstance(target._format, (RepositoryFormatKnit2)))
2169
except AttributeError:
2170
return False
2171
2172
@needs_write_lock
2173
def fetch(self, revision_id=None, pb=None):
2174
"""See InterRepository.fetch()."""
2175
from bzrlib.fetch import Knit1to2Fetcher
2176
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2177
self.source, self.source._format, self.target,
2178
self.target._format)
2179
f = Knit1to2Fetcher(to_repository=self.target,
2180
from_repository=self.source,
2181
last_revision=revision_id,
2182
pb=pb)
2183
return f.count_copied, f.failed_revisions
2184
2185
2186
InterRepository.register_optimiser(InterSameDataRepository)
2187
InterRepository.register_optimiser(InterWeaveRepo)
2188
InterRepository.register_optimiser(InterKnitRepo)
2189
InterRepository.register_optimiser(InterModel1and2)
2190
InterRepository.register_optimiser(InterKnit1and2)
2191
2192
2193
class RepositoryTestProviderAdapter(object):
2194
"""A tool to generate a suite testing multiple repository formats at once.
2195
2196
This is done by copying the test once for each transport and injecting
2197
the transport_server, transport_readonly_server, and bzrdir_format and
2198
repository_format classes into each copy. Each copy is also given a new id()
2199
to make it easy to identify.
2200
"""
2201
2202
def __init__(self, transport_server, transport_readonly_server, formats):
2203
self._transport_server = transport_server
2204
self._transport_readonly_server = transport_readonly_server
2205
self._formats = formats
2206
2207
def adapt(self, test):
2208
result = unittest.TestSuite()
2209
for repository_format, bzrdir_format in self._formats:
2210
new_test = deepcopy(test)
2211
new_test.transport_server = self._transport_server
2212
new_test.transport_readonly_server = self._transport_readonly_server
2213
new_test.bzrdir_format = bzrdir_format
2214
new_test.repository_format = repository_format
2215
def make_new_test_id():
2216
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2217
return lambda: new_id
2218
new_test.id = make_new_test_id()
2219
result.addTest(new_test)
2220
return result
2221
2222
2223
class InterRepositoryTestProviderAdapter(object):
2224
"""A tool to generate a suite testing multiple inter repository formats.
2225
2226
This is done by copying the test once for each interrepo provider and injecting
2227
the transport_server, transport_readonly_server, repository_format and
2228
repository_to_format classes into each copy.
2229
Each copy is also given a new id() to make it easy to identify.
2230
"""
2231
2232
def __init__(self, transport_server, transport_readonly_server, formats):
2233
self._transport_server = transport_server
2234
self._transport_readonly_server = transport_readonly_server
2235
self._formats = formats
2236
2237
def adapt(self, test):
2238
result = unittest.TestSuite()
2239
for interrepo_class, repository_format, repository_format_to in self._formats:
2240
new_test = deepcopy(test)
2241
new_test.transport_server = self._transport_server
2242
new_test.transport_readonly_server = self._transport_readonly_server
2243
new_test.interrepo_class = interrepo_class
2244
new_test.repository_format = repository_format
2245
new_test.repository_format_to = repository_format_to
2246
def make_new_test_id():
2247
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2248
return lambda: new_id
2249
new_test.id = make_new_test_id()
2250
result.addTest(new_test)
2251
return result
2252
2253
@staticmethod
2254
def default_test_list():
2255
"""Generate the default list of interrepo permutations to test."""
2256
result = []
2257
# test the default InterRepository between format 6 and the current
2258
# default format.
2259
# XXX: robertc 20060220 reinstate this when there are two supported
2260
# formats which do not have an optimal code path between them.
2261
#result.append((InterRepository,
2262
# RepositoryFormat6(),
2263
# RepositoryFormatKnit1()))
2264
for optimiser in InterRepository._optimisers:
2265
if optimiser._matching_repo_format is not None:
2266
result.append((optimiser,
2267
optimiser._matching_repo_format,
2268
optimiser._matching_repo_format
2269
))
2270
# if there are specific combinations we want to use, we can add them
2271
# here.
2272
result.append((InterModel1and2, RepositoryFormat5(),
2273
RepositoryFormatKnit2()))
2274
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2275
RepositoryFormatKnit2()))
2276
return result
2277
2278
2279
class CopyConverter(object):
2280
"""A repository conversion tool which just performs a copy of the content.
2281
2282
This is slow but quite reliable.
2283
"""
2284
2285
def __init__(self, target_format):
2286
"""Create a CopyConverter.
2287
2288
:param target_format: The format the resulting repository should be.
2289
"""
2290
self.target_format = target_format
2291
2292
def convert(self, repo, pb):
2293
"""Perform the conversion of to_convert, giving feedback via pb.
2294
2295
:param to_convert: The disk object to convert.
2296
:param pb: a progress bar to use for progress information.
2297
"""
2298
self.pb = pb
2299
self.count = 0
2300
self.total = 4
2301
# this is only useful with metadir layouts - separated repo content.
2302
# trigger an assertion if not such
2303
repo._format.get_format_string()
2304
self.repo_dir = repo.bzrdir
2305
self.step('Moving repository to repository.backup')
2306
self.repo_dir.transport.move('repository', 'repository.backup')
2307
backup_transport = self.repo_dir.transport.clone('repository.backup')
2308
repo._format.check_conversion_target(self.target_format)
2309
self.source_repo = repo._format.open(self.repo_dir,
2310
_found=True,
2311
_override_transport=backup_transport)
2312
self.step('Creating new repository')
2313
converted = self.target_format.initialize(self.repo_dir,
2314
self.source_repo.is_shared())
2315
converted.lock_write()
2316
try:
2317
self.step('Copying content into repository.')
2318
self.source_repo.copy_content_into(converted)
2319
finally:
2320
converted.unlock()
2321
self.step('Deleting old repository content.')
2322
self.repo_dir.transport.delete_tree('repository.backup')
2323
self.pb.note('repository converted')
2324
2325
def step(self, message):
2326
"""Update the pb by a step."""
2327
self.count +=1
2328
self.pb.update(message, self.count, self.total)
2329
2330
2331
class CommitBuilder(object):
2332
"""Provides an interface to build up a commit.
2333
2334
This allows describing a tree to be committed without needing to
2335
know the internals of the format of the repository.
2336
"""
2337
2338
record_root_entry = False
2339
def __init__(self, repository, parents, config, timestamp=None,
2340
timezone=None, committer=None, revprops=None,
2341
revision_id=None):
2342
"""Initiate a CommitBuilder.
2343
2344
:param repository: Repository to commit to.
2345
:param parents: Revision ids of the parents of the new revision.
2346
:param config: Configuration to use.
2347
:param timestamp: Optional timestamp recorded for commit.
2348
:param timezone: Optional timezone for timestamp.
2349
:param committer: Optional committer to set for commit.
2350
:param revprops: Optional dictionary of revision properties.
2351
:param revision_id: Optional revision id.
2352
"""
2353
self._config = config
2354
2355
if committer is None:
2356
self._committer = self._config.username()
2357
else:
2358
assert isinstance(committer, basestring), type(committer)
2359
self._committer = committer
2360
2361
self.new_inventory = Inventory(None)
2362
self._new_revision_id = revision_id
2363
self.parents = parents
2364
self.repository = repository
2365
2366
self._revprops = {}
2367
if revprops is not None:
2368
self._revprops.update(revprops)
2369
2370
if timestamp is None:
2371
timestamp = time.time()
2372
# Restrict resolution to 1ms
2373
self._timestamp = round(timestamp, 3)
2374
2375
if timezone is None:
2376
self._timezone = local_time_offset()
2377
else:
2378
self._timezone = int(timezone)
2379
2380
self._generate_revision_if_needed()
2381
2382
def commit(self, message):
2383
"""Make the actual commit.
2384
2385
:return: The revision id of the recorded revision.
2386
"""
2387
rev = _mod_revision.Revision(
2388
timestamp=self._timestamp,
2389
timezone=self._timezone,
2390
committer=self._committer,
2391
message=message,
2392
inventory_sha1=self.inv_sha1,
2393
revision_id=self._new_revision_id,
2394
properties=self._revprops)
2395
rev.parent_ids = self.parents
2396
self.repository.add_revision(self._new_revision_id, rev,
2397
self.new_inventory, self._config)
2398
return self._new_revision_id
2399
2400
def revision_tree(self):
2401
"""Return the tree that was just committed.
2402
2403
After calling commit() this can be called to get a RevisionTree
2404
representing the newly committed tree. This is preferred to
2405
calling Repository.revision_tree() because that may require
2406
deserializing the inventory, while we already have a copy in
2407
memory.
2408
"""
2409
return RevisionTree(self.repository, self.new_inventory,
2410
self._new_revision_id)
2411
2412
def finish_inventory(self):
2413
"""Tell the builder that the inventory is finished."""
2414
if self.new_inventory.root is None:
2415
symbol_versioning.warn('Root entry should be supplied to'
2416
' record_entry_contents, as of bzr 0.10.',
2417
DeprecationWarning, stacklevel=2)
2418
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2419
self.new_inventory.revision_id = self._new_revision_id
2420
self.inv_sha1 = self.repository.add_inventory(
2421
self._new_revision_id,
2422
self.new_inventory,
2423
self.parents
2424
)
2425
2426
def _gen_revision_id(self):
2427
"""Return new revision-id."""
2428
return generate_ids.gen_revision_id(self._config.username(),
2429
self._timestamp)
2430
2431
def _generate_revision_if_needed(self):
2432
"""Create a revision id if None was supplied.
2433
2434
If the repository can not support user-specified revision ids
2435
they should override this function and raise CannotSetRevisionId
2436
if _new_revision_id is not None.
2437
2438
:raises: CannotSetRevisionId
2439
"""
2440
if self._new_revision_id is None:
2441
self._new_revision_id = self._gen_revision_id()
2442
2443
def record_entry_contents(self, ie, parent_invs, path, tree):
2444
"""Record the content of ie from tree into the commit if needed.
2445
2446
Side effect: sets ie.revision when unchanged
2447
2448
:param ie: An inventory entry present in the commit.
2449
:param parent_invs: The inventories of the parent revisions of the
2450
commit.
2451
:param path: The path the entry is at in the tree.
2452
:param tree: The tree which contains this entry and should be used to
2453
obtain content.
2454
"""
2455
if self.new_inventory.root is None and ie.parent_id is not None:
2456
symbol_versioning.warn('Root entry should be supplied to'
2457
' record_entry_contents, as of bzr 0.10.',
2458
DeprecationWarning, stacklevel=2)
2459
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2460
'', tree)
2461
self.new_inventory.add(ie)
2462
2463
# ie.revision is always None if the InventoryEntry is considered
2464
# for committing. ie.snapshot will record the correct revision
2465
# which may be the sole parent if it is untouched.
2466
if ie.revision is not None:
2467
return
2468
2469
# In this revision format, root entries have no knit or weave
2470
if ie is self.new_inventory.root:
2471
# When serializing out to disk and back in
2472
# root.revision is always _new_revision_id
2473
ie.revision = self._new_revision_id
2474
return
2475
previous_entries = ie.find_previous_heads(
2476
parent_invs,
2477
self.repository.weave_store,
2478
self.repository.get_transaction())
2479
# we are creating a new revision for ie in the history store
2480
# and inventory.
2481
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2482
2483
def modified_directory(self, file_id, file_parents):
2484
"""Record the presence of a symbolic link.
2485
2486
:param file_id: The file_id of the link to record.
2487
:param file_parents: The per-file parent revision ids.
2488
"""
2489
self._add_text_to_weave(file_id, [], file_parents.keys())
2490
2491
def modified_file_text(self, file_id, file_parents,
2492
get_content_byte_lines, text_sha1=None,
2493
text_size=None):
2494
"""Record the text of file file_id
2495
2496
:param file_id: The file_id of the file to record the text of.
2497
:param file_parents: The per-file parent revision ids.
2498
:param get_content_byte_lines: A callable which will return the byte
2499
lines for the file.
2500
:param text_sha1: Optional SHA1 of the file contents.
2501
:param text_size: Optional size of the file contents.
2502
"""
2503
# mutter('storing text of file {%s} in revision {%s} into %r',
2504
# file_id, self._new_revision_id, self.repository.weave_store)
2505
# special case to avoid diffing on renames or
2506
# reparenting
2507
if (len(file_parents) == 1
2508
and text_sha1 == file_parents.values()[0].text_sha1
2509
and text_size == file_parents.values()[0].text_size):
2510
previous_ie = file_parents.values()[0]
2511
versionedfile = self.repository.weave_store.get_weave(file_id,
2512
self.repository.get_transaction())
2513
versionedfile.clone_text(self._new_revision_id,
2514
previous_ie.revision, file_parents.keys())
2515
return text_sha1, text_size
2516
else:
2517
new_lines = get_content_byte_lines()
2518
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2519
# should return the SHA1 and size
2520
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2521
return osutils.sha_strings(new_lines), \
2522
sum(map(len, new_lines))
2523
2524
def modified_link(self, file_id, file_parents, link_target):
2525
"""Record the presence of a symbolic link.
2526
2527
:param file_id: The file_id of the link to record.
2528
:param file_parents: The per-file parent revision ids.
2529
:param link_target: Target location of this link.
2530
"""
2531
self._add_text_to_weave(file_id, [], file_parents.keys())
2532
2533
def _add_text_to_weave(self, file_id, new_lines, parents):
2534
versionedfile = self.repository.weave_store.get_weave_or_empty(
2535
file_id, self.repository.get_transaction())
2536
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2537
versionedfile.clear_cache()
2538
2539
2540
class _CommitBuilder(CommitBuilder):
2541
"""Temporary class so old CommitBuilders are detected properly
2542
2543
Note: CommitBuilder works whether or not root entry is recorded.
2544
"""
2545
2546
record_root_entry = True
2547
2548
2549
class RootCommitBuilder(CommitBuilder):
2550
"""This commitbuilder actually records the root id"""
2551
2552
record_root_entry = True
2553
2554
def record_entry_contents(self, ie, parent_invs, path, tree):
2555
"""Record the content of ie from tree into the commit if needed.
2556
2557
Side effect: sets ie.revision when unchanged
2558
2559
:param ie: An inventory entry present in the commit.
2560
:param parent_invs: The inventories of the parent revisions of the
2561
commit.
2562
:param path: The path the entry is at in the tree.
2563
:param tree: The tree which contains this entry and should be used to
2564
obtain content.
2565
"""
2566
assert self.new_inventory.root is not None or ie.parent_id is None
2567
self.new_inventory.add(ie)
2568
2569
# ie.revision is always None if the InventoryEntry is considered
2570
# for committing. ie.snapshot will record the correct revision
2571
# which may be the sole parent if it is untouched.
2572
if ie.revision is not None:
2573
return
2574
2575
previous_entries = ie.find_previous_heads(
2576
parent_invs,
2577
self.repository.weave_store,
2578
self.repository.get_transaction())
2579
# we are creating a new revision for ie in the history store
2580
# and inventory.
2581
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2582
2583
2584
_unescape_map = {
2585
'apos':"'",
2586
'quot':'"',
2587
'amp':'&',
2588
'lt':'<',
2589
'gt':'>'
2590
}
2591
2592
2593
def _unescaper(match, _map=_unescape_map):
2594
return _map[match.group(1)]
2595
2596
2597
_unescape_re = None
2598
2599
2600
def _unescape_xml(data):
2601
"""Unescape predefined XML entities in a string of data."""
2602
global _unescape_re
2603
if _unescape_re is None:
2604
_unescape_re = re.compile('\&([^;]*);')
2605
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1