To get this branch, use:
bzr branch
http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar
« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 2177
- compare with another revision
- download diff
- download tarball
- view history from revision 2177
- Committer: Canonical.com Patch Queue Manager
- Date: 2006-12-12 02:28:39 UTC
- mfrom: (2163.1.5 fileids_altered_merged)
- Revision ID: pqm@pqm.ubuntu.com-20061212022839-782ee5cf2ddd2647
(John Arbash Meinel) Tune fileids_altered_by_revision_ids
- files added:
- .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 removed:
- .bzrignore
- notes
- tests
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
unescape_revid_cache = {}
443
unescape_fileid_cache = {}
444
445
# revisions. We don't need to see all lines in the inventory because
446
# only those added in an inventory in rev X can contain a revision=X
447
# line.
448
pb = ui.ui_factory.nested_progress_bar()
449
try:
450
for line in w.iter_lines_added_or_present_in_versions(
451
selected_revision_ids, pb=pb):
452
match = self._file_ids_altered_regex.search(line)
453
if match is None:
454
continue
455
file_id, revision_id = match.group('file_id', 'revision_id')
456
revision_id = _unescape_xml_cached(revision_id,
457
unescape_revid_cache)
458
if revision_id in selected_revision_ids:
459
file_id = _unescape_xml_cached(file_id,
460
unescape_fileid_cache)
461
result.setdefault(file_id, set()).add(revision_id)
462
finally:
463
pb.finished()
464
return result
465
466
@needs_read_lock
467
def get_inventory_weave(self):
468
return self.control_weaves.get_weave('inventory',
469
self.get_transaction())
470
471
@needs_read_lock
472
def get_inventory(self, revision_id):
473
"""Get Inventory object by hash."""
474
return self.deserialise_inventory(
475
revision_id, self.get_inventory_xml(revision_id))
476
477
def deserialise_inventory(self, revision_id, xml):
478
"""Transform the xml into an inventory object.
479
480
:param revision_id: The expected revision id of the inventory.
481
:param xml: A serialised inventory.
482
"""
483
result = self._serializer.read_inventory_from_string(xml)
484
result.root.revision = revision_id
485
return result
486
487
def serialise_inventory(self, inv):
488
return self._serializer.write_inventory_to_string(inv)
489
490
@needs_read_lock
491
def get_inventory_xml(self, revision_id):
492
"""Get inventory XML as a file object."""
493
try:
494
assert isinstance(revision_id, basestring), type(revision_id)
495
iw = self.get_inventory_weave()
496
return iw.get_text(revision_id)
497
except IndexError:
498
raise errors.HistoryMissing(self, 'inventory', revision_id)
499
500
@needs_read_lock
501
def get_inventory_sha1(self, revision_id):
502
"""Return the sha1 hash of the inventory entry
503
"""
504
return self.get_revision(revision_id).inventory_sha1
505
506
@needs_read_lock
507
def get_revision_graph(self, revision_id=None):
508
"""Return a dictionary containing the revision graph.
509
510
:param revision_id: The revision_id to get a graph from. If None, then
511
the entire revision graph is returned. This is a deprecated mode of
512
operation and will be removed in the future.
513
:return: a dictionary of revision_id->revision_parents_list.
514
"""
515
# special case NULL_REVISION
516
if revision_id == _mod_revision.NULL_REVISION:
517
return {}
518
a_weave = self.get_inventory_weave()
519
all_revisions = self._eliminate_revisions_not_present(
520
a_weave.versions())
521
entire_graph = dict([(node, a_weave.get_parents(node)) for
522
node in all_revisions])
523
if revision_id is None:
524
return entire_graph
525
elif revision_id not in entire_graph:
526
raise errors.NoSuchRevision(self, revision_id)
527
else:
528
# add what can be reached from revision_id
529
result = {}
530
pending = set([revision_id])
531
while len(pending) > 0:
532
node = pending.pop()
533
result[node] = entire_graph[node]
534
for revision_id in result[node]:
535
if revision_id not in result:
536
pending.add(revision_id)
537
return result
538
539
@needs_read_lock
540
def get_revision_graph_with_ghosts(self, revision_ids=None):
541
"""Return a graph of the revisions with ghosts marked as applicable.
542
543
:param revision_ids: an iterable of revisions to graph or None for all.
544
:return: a Graph object with the graph reachable from revision_ids.
545
"""
546
result = graph.Graph()
547
if not revision_ids:
548
pending = set(self.all_revision_ids())
549
required = set([])
550
else:
551
pending = set(revision_ids)
552
# special case NULL_REVISION
553
if _mod_revision.NULL_REVISION in pending:
554
pending.remove(_mod_revision.NULL_REVISION)
555
required = set(pending)
556
done = set([])
557
while len(pending):
558
revision_id = pending.pop()
559
try:
560
rev = self.get_revision(revision_id)
561
except errors.NoSuchRevision:
562
if revision_id in required:
563
raise
564
# a ghost
565
result.add_ghost(revision_id)
566
continue
567
for parent_id in rev.parent_ids:
568
# is this queued or done ?
569
if (parent_id not in pending and
570
parent_id not in done):
571
# no, queue it.
572
pending.add(parent_id)
573
result.add_node(revision_id, rev.parent_ids)
574
done.add(revision_id)
575
return result
576
577
@needs_read_lock
578
def get_revision_inventory(self, revision_id):
579
"""Return inventory of a past revision."""
580
# TODO: Unify this with get_inventory()
581
# bzr 0.0.6 and later imposes the constraint that the inventory_id
582
# must be the same as its revision, so this is trivial.
583
if revision_id is None:
584
# This does not make sense: if there is no revision,
585
# then it is the current tree inventory surely ?!
586
# and thus get_root_id() is something that looks at the last
587
# commit on the branch, and the get_root_id is an inventory check.
588
raise NotImplementedError
589
# return Inventory(self.get_root_id())
590
else:
591
return self.get_inventory(revision_id)
592
593
@needs_read_lock
594
def is_shared(self):
595
"""Return True if this repository is flagged as a shared repository."""
596
raise NotImplementedError(self.is_shared)
597
598
@needs_write_lock
599
def reconcile(self, other=None, thorough=False):
600
"""Reconcile this repository."""
601
from bzrlib.reconcile import RepoReconciler
602
reconciler = RepoReconciler(self, thorough=thorough)
603
reconciler.reconcile()
604
return reconciler
605
606
@needs_read_lock
607
def revision_tree(self, revision_id):
608
"""Return Tree for a revision on this branch.
609
610
`revision_id` may be None for the empty tree revision.
611
"""
612
# TODO: refactor this to use an existing revision object
613
# so we don't need to read it in twice.
614
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
615
return RevisionTree(self, Inventory(root_id=None),
616
_mod_revision.NULL_REVISION)
617
else:
618
inv = self.get_revision_inventory(revision_id)
619
return RevisionTree(self, inv, revision_id)
620
621
@needs_read_lock
622
def revision_trees(self, revision_ids):
623
"""Return Tree for a revision on this branch.
624
625
`revision_id` may not be None or 'null:'"""
626
assert None not in revision_ids
627
assert _mod_revision.NULL_REVISION not in revision_ids
628
texts = self.get_inventory_weave().get_texts(revision_ids)
629
for text, revision_id in zip(texts, revision_ids):
630
inv = self.deserialise_inventory(revision_id, text)
631
yield RevisionTree(self, inv, revision_id)
632
633
@needs_read_lock
634
def get_ancestry(self, revision_id):
635
"""Return a list of revision-ids integrated by a revision.
636
637
The first element of the list is always None, indicating the origin
638
revision. This might change when we have history horizons, or
639
perhaps we should have a new API.
640
641
This is topologically sorted.
642
"""
643
if revision_id is None:
644
return [None]
645
if not self.has_revision(revision_id):
646
raise errors.NoSuchRevision(self, revision_id)
647
w = self.get_inventory_weave()
648
candidates = w.get_ancestry(revision_id)
649
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
650
651
@needs_read_lock
652
def print_file(self, file, revision_id):
653
"""Print `file` to stdout.
654
655
FIXME RBC 20060125 as John Meinel points out this is a bad api
656
- it writes to stdout, it assumes that that is valid etc. Fix
657
by creating a new more flexible convenience function.
658
"""
659
tree = self.revision_tree(revision_id)
660
# use inventory as it was in that revision
661
file_id = tree.inventory.path2id(file)
662
if not file_id:
663
# TODO: jam 20060427 Write a test for this code path
664
# it had a bug in it, and was raising the wrong
665
# exception.
666
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
667
tree.print_file(file_id)
668
669
def get_transaction(self):
670
return self.control_files.get_transaction()
671
672
def revision_parents(self, revid):
673
return self.get_inventory_weave().parent_names(revid)
674
675
@needs_write_lock
676
def set_make_working_trees(self, new_value):
677
"""Set the policy flag for making working trees when creating branches.
678
679
This only applies to branches that use this repository.
680
681
The default is 'True'.
682
:param new_value: True to restore the default, False to disable making
683
working trees.
684
"""
685
raise NotImplementedError(self.set_make_working_trees)
686
687
def make_working_trees(self):
688
"""Returns the policy for making working trees on new branches."""
689
raise NotImplementedError(self.make_working_trees)
690
691
@needs_write_lock
692
def sign_revision(self, revision_id, gpg_strategy):
693
plaintext = Testament.from_revision(self, revision_id).as_short_text()
694
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
695
696
@needs_read_lock
697
def has_signature_for_revision_id(self, revision_id):
698
"""Query for a revision signature for revision_id in the repository."""
699
return self._revision_store.has_signature(revision_id,
700
self.get_transaction())
701
702
@needs_read_lock
703
def get_signature_text(self, revision_id):
704
"""Return the text for a signature."""
705
return self._revision_store.get_signature_text(revision_id,
706
self.get_transaction())
707
708
@needs_read_lock
709
def check(self, revision_ids):
710
"""Check consistency of all history of given revision_ids.
711
712
Different repository implementations should override _check().
713
714
:param revision_ids: A non-empty list of revision_ids whose ancestry
715
will be checked. Typically the last revision_id of a branch.
716
"""
717
if not revision_ids:
718
raise ValueError("revision_ids must be non-empty in %s.check"
719
% (self,))
720
return self._check(revision_ids)
721
722
def _check(self, revision_ids):
723
result = check.Check(self)
724
result.check()
725
return result
726
727
def _warn_if_deprecated(self):
728
global _deprecation_warning_done
729
if _deprecation_warning_done:
730
return
731
_deprecation_warning_done = True
732
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
733
% (self._format, self.bzrdir.transport.base))
734
735
def supports_rich_root(self):
736
return self._format.rich_root_data
737
738
def _check_ascii_revisionid(self, revision_id, method):
739
"""Private helper for ascii-only repositories."""
740
# weave repositories refuse to store revisionids that are non-ascii.
741
if revision_id is not None:
742
# weaves require ascii revision ids.
743
if isinstance(revision_id, unicode):
744
try:
745
revision_id.encode('ascii')
746
except UnicodeEncodeError:
747
raise errors.NonAsciiRevisionId(method, self)
748
749
750
class AllInOneRepository(Repository):
751
"""Legacy support - the repository behaviour for all-in-one branches."""
752
753
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
754
# we reuse one control files instance.
755
dir_mode = a_bzrdir._control_files._dir_mode
756
file_mode = a_bzrdir._control_files._file_mode
757
758
def get_store(name, compressed=True, prefixed=False):
759
# FIXME: This approach of assuming stores are all entirely compressed
760
# or entirely uncompressed is tidy, but breaks upgrade from
761
# some existing branches where there's a mixture; we probably
762
# still want the option to look for both.
763
relpath = a_bzrdir._control_files._escape(name)
764
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
765
prefixed=prefixed, compressed=compressed,
766
dir_mode=dir_mode,
767
file_mode=file_mode)
768
#if self._transport.should_cache():
769
# cache_path = os.path.join(self.cache_root, name)
770
# os.mkdir(cache_path)
771
# store = bzrlib.store.CachedStore(store, cache_path)
772
return store
773
774
# not broken out yet because the controlweaves|inventory_store
775
# and text_store | weave_store bits are still different.
776
if isinstance(_format, RepositoryFormat4):
777
# cannot remove these - there is still no consistent api
778
# which allows access to this old info.
779
self.inventory_store = get_store('inventory-store')
780
text_store = get_store('text-store')
781
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
782
783
def get_commit_builder(self, branch, parents, config, timestamp=None,
784
timezone=None, committer=None, revprops=None,
785
revision_id=None):
786
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
787
return Repository.get_commit_builder(self, branch, parents, config,
788
timestamp, timezone, committer, revprops, revision_id)
789
790
@needs_read_lock
791
def is_shared(self):
792
"""AllInOne repositories cannot be shared."""
793
return False
794
795
@needs_write_lock
796
def set_make_working_trees(self, new_value):
797
"""Set the policy flag for making working trees when creating branches.
798
799
This only applies to branches that use this repository.
800
801
The default is 'True'.
802
:param new_value: True to restore the default, False to disable making
803
working trees.
804
"""
805
raise NotImplementedError(self.set_make_working_trees)
806
807
def make_working_trees(self):
808
"""Returns the policy for making working trees on new branches."""
809
return True
810
811
812
def install_revision(repository, rev, revision_tree):
813
"""Install all revision data into a repository."""
814
present_parents = []
815
parent_trees = {}
816
for p_id in rev.parent_ids:
817
if repository.has_revision(p_id):
818
present_parents.append(p_id)
819
parent_trees[p_id] = repository.revision_tree(p_id)
820
else:
821
parent_trees[p_id] = repository.revision_tree(None)
822
823
inv = revision_tree.inventory
824
entries = inv.iter_entries()
825
# backwards compatability hack: skip the root id.
826
if not repository.supports_rich_root():
827
path, root = entries.next()
828
if root.revision != rev.revision_id:
829
raise errors.IncompatibleRevision(repr(repository))
830
# Add the texts that are not already present
831
for path, ie in entries:
832
w = repository.weave_store.get_weave_or_empty(ie.file_id,
833
repository.get_transaction())
834
if ie.revision not in w:
835
text_parents = []
836
# FIXME: TODO: The following loop *may* be overlapping/duplicate
837
# with InventoryEntry.find_previous_heads(). if it is, then there
838
# is a latent bug here where the parents may have ancestors of each
839
# other. RBC, AB
840
for revision, tree in parent_trees.iteritems():
841
if ie.file_id not in tree:
842
continue
843
parent_id = tree.inventory[ie.file_id].revision
844
if parent_id in text_parents:
845
continue
846
text_parents.append(parent_id)
847
848
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
849
repository.get_transaction())
850
lines = revision_tree.get_file(ie.file_id).readlines()
851
vfile.add_lines(rev.revision_id, text_parents, lines)
852
try:
853
# install the inventory
854
repository.add_inventory(rev.revision_id, inv, present_parents)
855
except errors.RevisionAlreadyPresent:
856
pass
857
repository.add_revision(rev.revision_id, rev, inv)
858
859
860
class MetaDirRepository(Repository):
861
"""Repositories in the new meta-dir layout."""
862
863
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
864
super(MetaDirRepository, self).__init__(_format,
865
a_bzrdir,
866
control_files,
867
_revision_store,
868
control_store,
869
text_store)
870
dir_mode = self.control_files._dir_mode
871
file_mode = self.control_files._file_mode
872
873
@needs_read_lock
874
def is_shared(self):
875
"""Return True if this repository is flagged as a shared repository."""
876
return self.control_files._transport.has('shared-storage')
877
878
@needs_write_lock
879
def set_make_working_trees(self, new_value):
880
"""Set the policy flag for making working trees when creating branches.
881
882
This only applies to branches that use this repository.
883
884
The default is 'True'.
885
:param new_value: True to restore the default, False to disable making
886
working trees.
887
"""
888
if new_value:
889
try:
890
self.control_files._transport.delete('no-working-trees')
891
except errors.NoSuchFile:
892
pass
893
else:
894
self.control_files.put_utf8('no-working-trees', '')
895
896
def make_working_trees(self):
897
"""Returns the policy for making working trees on new branches."""
898
return not self.control_files._transport.has('no-working-trees')
899
900
901
class WeaveMetaDirRepository(MetaDirRepository):
902
"""A subclass of MetaDirRepository to set weave specific policy."""
903
904
def get_commit_builder(self, branch, parents, config, timestamp=None,
905
timezone=None, committer=None, revprops=None,
906
revision_id=None):
907
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
908
return MetaDirRepository.get_commit_builder(self, branch, parents,
909
config, timestamp, timezone, committer, revprops, revision_id)
910
911
912
class KnitRepository(MetaDirRepository):
913
"""Knit format repository."""
914
915
def _warn_if_deprecated(self):
916
# This class isn't deprecated
917
pass
918
919
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
920
inv_vf.add_lines_with_ghosts(revid, parents, lines)
921
922
@needs_read_lock
923
def _all_revision_ids(self):
924
"""See Repository.all_revision_ids()."""
925
# Knits get the revision graph from the index of the revision knit, so
926
# it's always possible even if they're on an unlistable transport.
927
return self._revision_store.all_revision_ids(self.get_transaction())
928
929
def fileid_involved_between_revs(self, from_revid, to_revid):
930
"""Find file_id(s) which are involved in the changes between revisions.
931
932
This determines the set of revisions which are involved, and then
933
finds all file ids affected by those revisions.
934
"""
935
vf = self._get_revision_vf()
936
from_set = set(vf.get_ancestry(from_revid))
937
to_set = set(vf.get_ancestry(to_revid))
938
changed = to_set.difference(from_set)
939
return self._fileid_involved_by_set(changed)
940
941
def fileid_involved(self, last_revid=None):
942
"""Find all file_ids modified in the ancestry of last_revid.
943
944
:param last_revid: If None, last_revision() will be used.
945
"""
946
if not last_revid:
947
changed = set(self.all_revision_ids())
948
else:
949
changed = set(self.get_ancestry(last_revid))
950
if None in changed:
951
changed.remove(None)
952
return self._fileid_involved_by_set(changed)
953
954
@needs_read_lock
955
def get_ancestry(self, revision_id):
956
"""Return a list of revision-ids integrated by a revision.
957
958
This is topologically sorted.
959
"""
960
if revision_id is None:
961
return [None]
962
vf = self._get_revision_vf()
963
try:
964
return [None] + vf.get_ancestry(revision_id)
965
except errors.RevisionNotPresent:
966
raise errors.NoSuchRevision(self, revision_id)
967
968
@needs_read_lock
969
def get_revision(self, revision_id):
970
"""Return the Revision object for a named revision"""
971
return self.get_revision_reconcile(revision_id)
972
973
@needs_read_lock
974
def get_revision_graph(self, revision_id=None):
975
"""Return a dictionary containing the revision graph.
976
977
:param revision_id: The revision_id to get a graph from. If None, then
978
the entire revision graph is returned. This is a deprecated mode of
979
operation and will be removed in the future.
980
:return: a dictionary of revision_id->revision_parents_list.
981
"""
982
# special case NULL_REVISION
983
if revision_id == _mod_revision.NULL_REVISION:
984
return {}
985
a_weave = self._get_revision_vf()
986
entire_graph = a_weave.get_graph()
987
if revision_id is None:
988
return a_weave.get_graph()
989
elif revision_id not in a_weave:
990
raise errors.NoSuchRevision(self, revision_id)
991
else:
992
# add what can be reached from revision_id
993
result = {}
994
pending = set([revision_id])
995
while len(pending) > 0:
996
node = pending.pop()
997
result[node] = a_weave.get_parents(node)
998
for revision_id in result[node]:
999
if revision_id not in result:
1000
pending.add(revision_id)
1001
return result
1002
1003
@needs_read_lock
1004
def get_revision_graph_with_ghosts(self, revision_ids=None):
1005
"""Return a graph of the revisions with ghosts marked as applicable.
1006
1007
:param revision_ids: an iterable of revisions to graph or None for all.
1008
:return: a Graph object with the graph reachable from revision_ids.
1009
"""
1010
result = graph.Graph()
1011
vf = self._get_revision_vf()
1012
versions = set(vf.versions())
1013
if not revision_ids:
1014
pending = set(self.all_revision_ids())
1015
required = set([])
1016
else:
1017
pending = set(revision_ids)
1018
# special case NULL_REVISION
1019
if _mod_revision.NULL_REVISION in pending:
1020
pending.remove(_mod_revision.NULL_REVISION)
1021
required = set(pending)
1022
done = set([])
1023
while len(pending):
1024
revision_id = pending.pop()
1025
if not revision_id in versions:
1026
if revision_id in required:
1027
raise errors.NoSuchRevision(self, revision_id)
1028
# a ghost
1029
result.add_ghost(revision_id)
1030
# mark it as done so we don't try for it again.
1031
done.add(revision_id)
1032
continue
1033
parent_ids = vf.get_parents_with_ghosts(revision_id)
1034
for parent_id in parent_ids:
1035
# is this queued or done ?
1036
if (parent_id not in pending and
1037
parent_id not in done):
1038
# no, queue it.
1039
pending.add(parent_id)
1040
result.add_node(revision_id, parent_ids)
1041
done.add(revision_id)
1042
return result
1043
1044
def _get_revision_vf(self):
1045
""":return: a versioned file containing the revisions."""
1046
vf = self._revision_store.get_revision_file(self.get_transaction())
1047
return vf
1048
1049
@needs_write_lock
1050
def reconcile(self, other=None, thorough=False):
1051
"""Reconcile this repository."""
1052
from bzrlib.reconcile import KnitReconciler
1053
reconciler = KnitReconciler(self, thorough=thorough)
1054
reconciler.reconcile()
1055
return reconciler
1056
1057
def revision_parents(self, revision_id):
1058
return self._get_revision_vf().get_parents(revision_id)
1059
1060
1061
class KnitRepository2(KnitRepository):
1062
""""""
1063
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1064
control_store, text_store):
1065
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1066
_revision_store, control_store, text_store)
1067
self._serializer = xml6.serializer_v6
1068
1069
def deserialise_inventory(self, revision_id, xml):
1070
"""Transform the xml into an inventory object.
1071
1072
:param revision_id: The expected revision id of the inventory.
1073
:param xml: A serialised inventory.
1074
"""
1075
result = self._serializer.read_inventory_from_string(xml)
1076
assert result.root.revision is not None
1077
return result
1078
1079
def serialise_inventory(self, inv):
1080
"""Transform the inventory object into XML text.
1081
1082
:param revision_id: The expected revision id of the inventory.
1083
:param xml: A serialised inventory.
1084
"""
1085
assert inv.revision_id is not None
1086
assert inv.root.revision is not None
1087
return KnitRepository.serialise_inventory(self, inv)
1088
1089
def get_commit_builder(self, branch, parents, config, timestamp=None,
1090
timezone=None, committer=None, revprops=None,
1091
revision_id=None):
1092
"""Obtain a CommitBuilder for this repository.
1093
1094
:param branch: Branch to commit to.
1095
:param parents: Revision ids of the parents of the new revision.
1096
:param config: Configuration to use.
1097
:param timestamp: Optional timestamp recorded for commit.
1098
:param timezone: Optional timezone for timestamp.
1099
:param committer: Optional committer to set for commit.
1100
:param revprops: Optional dictionary of revision properties.
1101
:param revision_id: Optional revision id.
1102
"""
1103
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1104
committer, revprops, revision_id)
1105
1106
1107
class RepositoryFormat(object):
1108
"""A repository format.
1109
1110
Formats provide three things:
1111
* An initialization routine to construct repository data on disk.
1112
* a format string which is used when the BzrDir supports versioned
1113
children.
1114
* an open routine which returns a Repository instance.
1115
1116
Formats are placed in an dict by their format string for reference
1117
during opening. These should be subclasses of RepositoryFormat
1118
for consistency.
1119
1120
Once a format is deprecated, just deprecate the initialize and open
1121
methods on the format class. Do not deprecate the object, as the
1122
object will be created every system load.
1123
1124
Common instance attributes:
1125
_matchingbzrdir - the bzrdir format that the repository format was
1126
originally written to work with. This can be used if manually
1127
constructing a bzrdir and repository, or more commonly for test suite
1128
parameterisation.
1129
"""
1130
1131
_default_format = None
1132
"""The default format used for new repositories."""
1133
1134
_formats = {}
1135
"""The known formats."""
1136
1137
def __str__(self):
1138
return "<%s>" % self.__class__.__name__
1139
1140
@classmethod
1141
def find_format(klass, a_bzrdir):
1142
"""Return the format for the repository object in a_bzrdir."""
1143
try:
1144
transport = a_bzrdir.get_repository_transport(None)
1145
format_string = transport.get("format").read()
1146
return klass._formats[format_string]
1147
except errors.NoSuchFile:
1148
raise errors.NoRepositoryPresent(a_bzrdir)
1149
except KeyError:
1150
raise errors.UnknownFormatError(format=format_string)
1151
1152
def _get_control_store(self, repo_transport, control_files):
1153
"""Return the control store for this repository."""
1154
raise NotImplementedError(self._get_control_store)
1155
1156
@classmethod
1157
def get_default_format(klass):
1158
"""Return the current default format."""
1159
return klass._default_format
1160
1161
def get_format_string(self):
1162
"""Return the ASCII format string that identifies this format.
1163
1164
Note that in pre format ?? repositories the format string is
1165
not permitted nor written to disk.
1166
"""
1167
raise NotImplementedError(self.get_format_string)
1168
1169
def get_format_description(self):
1170
"""Return the short description for this format."""
1171
raise NotImplementedError(self.get_format_description)
1172
1173
def _get_revision_store(self, repo_transport, control_files):
1174
"""Return the revision store object for this a_bzrdir."""
1175
raise NotImplementedError(self._get_revision_store)
1176
1177
def _get_text_rev_store(self,
1178
transport,
1179
control_files,
1180
name,
1181
compressed=True,
1182
prefixed=False,
1183
serializer=None):
1184
"""Common logic for getting a revision store for a repository.
1185
1186
see self._get_revision_store for the subclass-overridable method to
1187
get the store for a repository.
1188
"""
1189
from bzrlib.store.revision.text import TextRevisionStore
1190
dir_mode = control_files._dir_mode
1191
file_mode = control_files._file_mode
1192
text_store =TextStore(transport.clone(name),
1193
prefixed=prefixed,
1194
compressed=compressed,
1195
dir_mode=dir_mode,
1196
file_mode=file_mode)
1197
_revision_store = TextRevisionStore(text_store, serializer)
1198
return _revision_store
1199
1200
def _get_versioned_file_store(self,
1201
name,
1202
transport,
1203
control_files,
1204
prefixed=True,
1205
versionedfile_class=weave.WeaveFile,
1206
versionedfile_kwargs={},
1207
escaped=False):
1208
weave_transport = control_files._transport.clone(name)
1209
dir_mode = control_files._dir_mode
1210
file_mode = control_files._file_mode
1211
return VersionedFileStore(weave_transport, prefixed=prefixed,
1212
dir_mode=dir_mode,
1213
file_mode=file_mode,
1214
versionedfile_class=versionedfile_class,
1215
versionedfile_kwargs=versionedfile_kwargs,
1216
escaped=escaped)
1217
1218
def initialize(self, a_bzrdir, shared=False):
1219
"""Initialize a repository of this format in a_bzrdir.
1220
1221
:param a_bzrdir: The bzrdir to put the new repository in it.
1222
:param shared: The repository should be initialized as a sharable one.
1223
1224
This may raise UninitializableFormat if shared repository are not
1225
compatible the a_bzrdir.
1226
"""
1227
1228
def is_supported(self):
1229
"""Is this format supported?
1230
1231
Supported formats must be initializable and openable.
1232
Unsupported formats may not support initialization or committing or
1233
some other features depending on the reason for not being supported.
1234
"""
1235
return True
1236
1237
def check_conversion_target(self, target_format):
1238
raise NotImplementedError(self.check_conversion_target)
1239
1240
def open(self, a_bzrdir, _found=False):
1241
"""Return an instance of this format for the bzrdir a_bzrdir.
1242
1243
_found is a private parameter, do not use it.
1244
"""
1245
raise NotImplementedError(self.open)
1246
1247
@classmethod
1248
def register_format(klass, format):
1249
klass._formats[format.get_format_string()] = format
1250
1251
@classmethod
1252
def set_default_format(klass, format):
1253
klass._default_format = format
1254
1255
@classmethod
1256
def unregister_format(klass, format):
1257
assert klass._formats[format.get_format_string()] is format
1258
del klass._formats[format.get_format_string()]
1259
1260
1261
class PreSplitOutRepositoryFormat(RepositoryFormat):
1262
"""Base class for the pre split out repository formats."""
1263
1264
rich_root_data = False
1265
1266
def initialize(self, a_bzrdir, shared=False, _internal=False):
1267
"""Create a weave repository.
1268
1269
TODO: when creating split out bzr branch formats, move this to a common
1270
base for Format5, Format6. or something like that.
1271
"""
1272
if shared:
1273
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1274
1275
if not _internal:
1276
# always initialized when the bzrdir is.
1277
return self.open(a_bzrdir, _found=True)
1278
1279
# Create an empty weave
1280
sio = StringIO()
1281
weavefile.write_weave_v5(weave.Weave(), sio)
1282
empty_weave = sio.getvalue()
1283
1284
mutter('creating repository in %s.', a_bzrdir.transport.base)
1285
dirs = ['revision-store', 'weaves']
1286
files = [('inventory.weave', StringIO(empty_weave)),
1287
]
1288
1289
# FIXME: RBC 20060125 don't peek under the covers
1290
# NB: no need to escape relative paths that are url safe.
1291
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1292
'branch-lock', lockable_files.TransportLock)
1293
control_files.create_lock()
1294
control_files.lock_write()
1295
control_files._transport.mkdir_multi(dirs,
1296
mode=control_files._dir_mode)
1297
try:
1298
for file, content in files:
1299
control_files.put(file, content)
1300
finally:
1301
control_files.unlock()
1302
return self.open(a_bzrdir, _found=True)
1303
1304
def _get_control_store(self, repo_transport, control_files):
1305
"""Return the control store for this repository."""
1306
return self._get_versioned_file_store('',
1307
repo_transport,
1308
control_files,
1309
prefixed=False)
1310
1311
def _get_text_store(self, transport, control_files):
1312
"""Get a store for file texts for this format."""
1313
raise NotImplementedError(self._get_text_store)
1314
1315
def open(self, a_bzrdir, _found=False):
1316
"""See RepositoryFormat.open()."""
1317
if not _found:
1318
# we are being called directly and must probe.
1319
raise NotImplementedError
1320
1321
repo_transport = a_bzrdir.get_repository_transport(None)
1322
control_files = a_bzrdir._control_files
1323
text_store = self._get_text_store(repo_transport, control_files)
1324
control_store = self._get_control_store(repo_transport, control_files)
1325
_revision_store = self._get_revision_store(repo_transport, control_files)
1326
return AllInOneRepository(_format=self,
1327
a_bzrdir=a_bzrdir,
1328
_revision_store=_revision_store,
1329
control_store=control_store,
1330
text_store=text_store)
1331
1332
def check_conversion_target(self, target_format):
1333
pass
1334
1335
1336
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1337
"""Bzr repository format 4.
1338
1339
This repository format has:
1340
- flat stores
1341
- TextStores for texts, inventories,revisions.
1342
1343
This format is deprecated: it indexes texts using a text id which is
1344
removed in format 5; initialization and write support for this format
1345
has been removed.
1346
"""
1347
1348
def __init__(self):
1349
super(RepositoryFormat4, self).__init__()
1350
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1351
1352
def get_format_description(self):
1353
"""See RepositoryFormat.get_format_description()."""
1354
return "Repository format 4"
1355
1356
def initialize(self, url, shared=False, _internal=False):
1357
"""Format 4 branches cannot be created."""
1358
raise errors.UninitializableFormat(self)
1359
1360
def is_supported(self):
1361
"""Format 4 is not supported.
1362
1363
It is not supported because the model changed from 4 to 5 and the
1364
conversion logic is expensive - so doing it on the fly was not
1365
feasible.
1366
"""
1367
return False
1368
1369
def _get_control_store(self, repo_transport, control_files):
1370
"""Format 4 repositories have no formal control store at this point.
1371
1372
This will cause any control-file-needing apis to fail - this is desired.
1373
"""
1374
return None
1375
1376
def _get_revision_store(self, repo_transport, control_files):
1377
"""See RepositoryFormat._get_revision_store()."""
1378
from bzrlib.xml4 import serializer_v4
1379
return self._get_text_rev_store(repo_transport,
1380
control_files,
1381
'revision-store',
1382
serializer=serializer_v4)
1383
1384
def _get_text_store(self, transport, control_files):
1385
"""See RepositoryFormat._get_text_store()."""
1386
1387
1388
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1389
"""Bzr control format 5.
1390
1391
This repository format has:
1392
- weaves for file texts and inventory
1393
- flat stores
1394
- TextStores for revisions and signatures.
1395
"""
1396
1397
def __init__(self):
1398
super(RepositoryFormat5, self).__init__()
1399
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1400
1401
def get_format_description(self):
1402
"""See RepositoryFormat.get_format_description()."""
1403
return "Weave repository format 5"
1404
1405
def _get_revision_store(self, repo_transport, control_files):
1406
"""See RepositoryFormat._get_revision_store()."""
1407
"""Return the revision store object for this a_bzrdir."""
1408
return self._get_text_rev_store(repo_transport,
1409
control_files,
1410
'revision-store',
1411
compressed=False)
1412
1413
def _get_text_store(self, transport, control_files):
1414
"""See RepositoryFormat._get_text_store()."""
1415
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1416
1417
1418
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1419
"""Bzr control format 6.
1420
1421
This repository format has:
1422
- weaves for file texts and inventory
1423
- hash subdirectory based stores.
1424
- TextStores for revisions and signatures.
1425
"""
1426
1427
def __init__(self):
1428
super(RepositoryFormat6, self).__init__()
1429
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1430
1431
def get_format_description(self):
1432
"""See RepositoryFormat.get_format_description()."""
1433
return "Weave repository format 6"
1434
1435
def _get_revision_store(self, repo_transport, control_files):
1436
"""See RepositoryFormat._get_revision_store()."""
1437
return self._get_text_rev_store(repo_transport,
1438
control_files,
1439
'revision-store',
1440
compressed=False,
1441
prefixed=True)
1442
1443
def _get_text_store(self, transport, control_files):
1444
"""See RepositoryFormat._get_text_store()."""
1445
return self._get_versioned_file_store('weaves', transport, control_files)
1446
1447
1448
class MetaDirRepositoryFormat(RepositoryFormat):
1449
"""Common base class for the new repositories using the metadir layout."""
1450
1451
rich_root_data = False
1452
1453
def __init__(self):
1454
super(MetaDirRepositoryFormat, self).__init__()
1455
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1456
1457
def _create_control_files(self, a_bzrdir):
1458
"""Create the required files and the initial control_files object."""
1459
# FIXME: RBC 20060125 don't peek under the covers
1460
# NB: no need to escape relative paths that are url safe.
1461
repository_transport = a_bzrdir.get_repository_transport(self)
1462
control_files = lockable_files.LockableFiles(repository_transport,
1463
'lock', lockdir.LockDir)
1464
control_files.create_lock()
1465
return control_files
1466
1467
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1468
"""Upload the initial blank content."""
1469
control_files = self._create_control_files(a_bzrdir)
1470
control_files.lock_write()
1471
try:
1472
control_files._transport.mkdir_multi(dirs,
1473
mode=control_files._dir_mode)
1474
for file, content in files:
1475
control_files.put(file, content)
1476
for file, content in utf8_files:
1477
control_files.put_utf8(file, content)
1478
if shared == True:
1479
control_files.put_utf8('shared-storage', '')
1480
finally:
1481
control_files.unlock()
1482
1483
1484
class RepositoryFormat7(MetaDirRepositoryFormat):
1485
"""Bzr repository 7.
1486
1487
This repository format has:
1488
- weaves for file texts and inventory
1489
- hash subdirectory based stores.
1490
- TextStores for revisions and signatures.
1491
- a format marker of its own
1492
- an optional 'shared-storage' flag
1493
- an optional 'no-working-trees' flag
1494
"""
1495
1496
def _get_control_store(self, repo_transport, control_files):
1497
"""Return the control store for this repository."""
1498
return self._get_versioned_file_store('',
1499
repo_transport,
1500
control_files,
1501
prefixed=False)
1502
1503
def get_format_string(self):
1504
"""See RepositoryFormat.get_format_string()."""
1505
return "Bazaar-NG Repository format 7"
1506
1507
def get_format_description(self):
1508
"""See RepositoryFormat.get_format_description()."""
1509
return "Weave repository format 7"
1510
1511
def check_conversion_target(self, target_format):
1512
pass
1513
1514
def _get_revision_store(self, repo_transport, control_files):
1515
"""See RepositoryFormat._get_revision_store()."""
1516
return self._get_text_rev_store(repo_transport,
1517
control_files,
1518
'revision-store',
1519
compressed=False,
1520
prefixed=True,
1521
)
1522
1523
def _get_text_store(self, transport, control_files):
1524
"""See RepositoryFormat._get_text_store()."""
1525
return self._get_versioned_file_store('weaves',
1526
transport,
1527
control_files)
1528
1529
def initialize(self, a_bzrdir, shared=False):
1530
"""Create a weave repository.
1531
1532
:param shared: If true the repository will be initialized as a shared
1533
repository.
1534
"""
1535
# Create an empty weave
1536
sio = StringIO()
1537
weavefile.write_weave_v5(weave.Weave(), sio)
1538
empty_weave = sio.getvalue()
1539
1540
mutter('creating repository in %s.', a_bzrdir.transport.base)
1541
dirs = ['revision-store', 'weaves']
1542
files = [('inventory.weave', StringIO(empty_weave)),
1543
]
1544
utf8_files = [('format', self.get_format_string())]
1545
1546
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1547
return self.open(a_bzrdir=a_bzrdir, _found=True)
1548
1549
def open(self, a_bzrdir, _found=False, _override_transport=None):
1550
"""See RepositoryFormat.open().
1551
1552
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1553
repository at a slightly different url
1554
than normal. I.e. during 'upgrade'.
1555
"""
1556
if not _found:
1557
format = RepositoryFormat.find_format(a_bzrdir)
1558
assert format.__class__ == self.__class__
1559
if _override_transport is not None:
1560
repo_transport = _override_transport
1561
else:
1562
repo_transport = a_bzrdir.get_repository_transport(None)
1563
control_files = lockable_files.LockableFiles(repo_transport,
1564
'lock', lockdir.LockDir)
1565
text_store = self._get_text_store(repo_transport, control_files)
1566
control_store = self._get_control_store(repo_transport, control_files)
1567
_revision_store = self._get_revision_store(repo_transport, control_files)
1568
return WeaveMetaDirRepository(_format=self,
1569
a_bzrdir=a_bzrdir,
1570
control_files=control_files,
1571
_revision_store=_revision_store,
1572
control_store=control_store,
1573
text_store=text_store)
1574
1575
1576
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1577
"""Bzr repository knit format (generalized).
1578
1579
This repository format has:
1580
- knits for file texts and inventory
1581
- hash subdirectory based stores.
1582
- knits for revisions and signatures
1583
- TextStores for revisions and signatures.
1584
- a format marker of its own
1585
- an optional 'shared-storage' flag
1586
- an optional 'no-working-trees' flag
1587
- a LockDir lock
1588
"""
1589
1590
def _get_control_store(self, repo_transport, control_files):
1591
"""Return the control store for this repository."""
1592
return VersionedFileStore(
1593
repo_transport,
1594
prefixed=False,
1595
file_mode=control_files._file_mode,
1596
versionedfile_class=knit.KnitVersionedFile,
1597
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1598
)
1599
1600
def _get_revision_store(self, repo_transport, control_files):
1601
"""See RepositoryFormat._get_revision_store()."""
1602
from bzrlib.store.revision.knit import KnitRevisionStore
1603
versioned_file_store = VersionedFileStore(
1604
repo_transport,
1605
file_mode=control_files._file_mode,
1606
prefixed=False,
1607
precious=True,
1608
versionedfile_class=knit.KnitVersionedFile,
1609
versionedfile_kwargs={'delta':False,
1610
'factory':knit.KnitPlainFactory(),
1611
},
1612
escaped=True,
1613
)
1614
return KnitRevisionStore(versioned_file_store)
1615
1616
def _get_text_store(self, transport, control_files):
1617
"""See RepositoryFormat._get_text_store()."""
1618
return self._get_versioned_file_store('knits',
1619
transport,
1620
control_files,
1621
versionedfile_class=knit.KnitVersionedFile,
1622
versionedfile_kwargs={
1623
'create_parent_dir':True,
1624
'delay_create':True,
1625
'dir_mode':control_files._dir_mode,
1626
},
1627
escaped=True)
1628
1629
def initialize(self, a_bzrdir, shared=False):
1630
"""Create a knit format 1 repository.
1631
1632
:param a_bzrdir: bzrdir to contain the new repository; must already
1633
be initialized.
1634
:param shared: If true the repository will be initialized as a shared
1635
repository.
1636
"""
1637
mutter('creating repository in %s.', a_bzrdir.transport.base)
1638
dirs = ['revision-store', 'knits']
1639
files = []
1640
utf8_files = [('format', self.get_format_string())]
1641
1642
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1643
repo_transport = a_bzrdir.get_repository_transport(None)
1644
control_files = lockable_files.LockableFiles(repo_transport,
1645
'lock', lockdir.LockDir)
1646
control_store = self._get_control_store(repo_transport, control_files)
1647
transaction = transactions.WriteTransaction()
1648
# trigger a write of the inventory store.
1649
control_store.get_weave_or_empty('inventory', transaction)
1650
_revision_store = self._get_revision_store(repo_transport, control_files)
1651
# the revision id here is irrelevant: it will not be stored, and cannot
1652
# already exist.
1653
_revision_store.has_revision_id('A', transaction)
1654
_revision_store.get_signature_file(transaction)
1655
return self.open(a_bzrdir=a_bzrdir, _found=True)
1656
1657
def open(self, a_bzrdir, _found=False, _override_transport=None):
1658
"""See RepositoryFormat.open().
1659
1660
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1661
repository at a slightly different url
1662
than normal. I.e. during 'upgrade'.
1663
"""
1664
if not _found:
1665
format = RepositoryFormat.find_format(a_bzrdir)
1666
assert format.__class__ == self.__class__
1667
if _override_transport is not None:
1668
repo_transport = _override_transport
1669
else:
1670
repo_transport = a_bzrdir.get_repository_transport(None)
1671
control_files = lockable_files.LockableFiles(repo_transport,
1672
'lock', lockdir.LockDir)
1673
text_store = self._get_text_store(repo_transport, control_files)
1674
control_store = self._get_control_store(repo_transport, control_files)
1675
_revision_store = self._get_revision_store(repo_transport, control_files)
1676
return KnitRepository(_format=self,
1677
a_bzrdir=a_bzrdir,
1678
control_files=control_files,
1679
_revision_store=_revision_store,
1680
control_store=control_store,
1681
text_store=text_store)
1682
1683
1684
class RepositoryFormatKnit1(RepositoryFormatKnit):
1685
"""Bzr repository knit format 1.
1686
1687
This repository format has:
1688
- knits for file texts and inventory
1689
- hash subdirectory based stores.
1690
- knits for revisions and signatures
1691
- TextStores for revisions and signatures.
1692
- a format marker of its own
1693
- an optional 'shared-storage' flag
1694
- an optional 'no-working-trees' flag
1695
- a LockDir lock
1696
1697
This format was introduced in bzr 0.8.
1698
"""
1699
def get_format_string(self):
1700
"""See RepositoryFormat.get_format_string()."""
1701
return "Bazaar-NG Knit Repository Format 1"
1702
1703
def get_format_description(self):
1704
"""See RepositoryFormat.get_format_description()."""
1705
return "Knit repository format 1"
1706
1707
def check_conversion_target(self, target_format):
1708
pass
1709
1710
1711
class RepositoryFormatKnit2(RepositoryFormatKnit):
1712
"""Bzr repository knit format 2.
1713
1714
THIS FORMAT IS EXPERIMENTAL
1715
This repository format has:
1716
- knits for file texts and inventory
1717
- hash subdirectory based stores.
1718
- knits for revisions and signatures
1719
- TextStores for revisions and signatures.
1720
- a format marker of its own
1721
- an optional 'shared-storage' flag
1722
- an optional 'no-working-trees' flag
1723
- a LockDir lock
1724
- Support for recording full info about the tree root
1725
1726
"""
1727
1728
rich_root_data = True
1729
1730
def get_format_string(self):
1731
"""See RepositoryFormat.get_format_string()."""
1732
return "Bazaar Knit Repository Format 2\n"
1733
1734
def get_format_description(self):
1735
"""See RepositoryFormat.get_format_description()."""
1736
return "Knit repository format 2"
1737
1738
def check_conversion_target(self, target_format):
1739
if not target_format.rich_root_data:
1740
raise errors.BadConversionTarget(
1741
'Does not support rich root data.', target_format)
1742
1743
def open(self, a_bzrdir, _found=False, _override_transport=None):
1744
"""See RepositoryFormat.open().
1745
1746
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1747
repository at a slightly different url
1748
than normal. I.e. during 'upgrade'.
1749
"""
1750
if not _found:
1751
format = RepositoryFormat.find_format(a_bzrdir)
1752
assert format.__class__ == self.__class__
1753
if _override_transport is not None:
1754
repo_transport = _override_transport
1755
else:
1756
repo_transport = a_bzrdir.get_repository_transport(None)
1757
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1758
lockdir.LockDir)
1759
text_store = self._get_text_store(repo_transport, control_files)
1760
control_store = self._get_control_store(repo_transport, control_files)
1761
_revision_store = self._get_revision_store(repo_transport, control_files)
1762
return KnitRepository2(_format=self,
1763
a_bzrdir=a_bzrdir,
1764
control_files=control_files,
1765
_revision_store=_revision_store,
1766
control_store=control_store,
1767
text_store=text_store)
1768
1769
1770
1771
# formats which have no format string are not discoverable
1772
# and not independently creatable, so are not registered.
1773
RepositoryFormat.register_format(RepositoryFormat7())
1774
_default_format = RepositoryFormatKnit1()
1775
RepositoryFormat.register_format(_default_format)
1776
RepositoryFormat.register_format(RepositoryFormatKnit2())
1777
RepositoryFormat.set_default_format(_default_format)
1778
_legacy_formats = [RepositoryFormat4(),
1779
RepositoryFormat5(),
1780
RepositoryFormat6()]
1781
1782
1783
class InterRepository(InterObject):
1784
"""This class represents operations taking place between two repositories.
1785
1786
Its instances have methods like copy_content and fetch, and contain
1787
references to the source and target repositories these operations can be
1788
carried out on.
1789
1790
Often we will provide convenience methods on 'repository' which carry out
1791
operations with another repository - they will always forward to
1792
InterRepository.get(other).method_name(parameters).
1793
"""
1794
1795
_optimisers = []
1796
"""The available optimised InterRepository types."""
1797
1798
def copy_content(self, revision_id=None, basis=None):
1799
raise NotImplementedError(self.copy_content)
1800
1801
def fetch(self, revision_id=None, pb=None):
1802
"""Fetch the content required to construct revision_id.
1803
1804
The content is copied from self.source to self.target.
1805
1806
:param revision_id: if None all content is copied, if NULL_REVISION no
1807
content is copied.
1808
:param pb: optional progress bar to use for progress reports. If not
1809
provided a default one will be created.
1810
1811
Returns the copied revision count and the failed revisions in a tuple:
1812
(copied, failures).
1813
"""
1814
raise NotImplementedError(self.fetch)
1815
1816
@needs_read_lock
1817
def missing_revision_ids(self, revision_id=None):
1818
"""Return the revision ids that source has that target does not.
1819
1820
These are returned in topological order.
1821
1822
:param revision_id: only return revision ids included by this
1823
revision_id.
1824
"""
1825
# generic, possibly worst case, slow code path.
1826
target_ids = set(self.target.all_revision_ids())
1827
if revision_id is not None:
1828
source_ids = self.source.get_ancestry(revision_id)
1829
assert source_ids[0] is None
1830
source_ids.pop(0)
1831
else:
1832
source_ids = self.source.all_revision_ids()
1833
result_set = set(source_ids).difference(target_ids)
1834
# this may look like a no-op: its not. It preserves the ordering
1835
# other_ids had while only returning the members from other_ids
1836
# that we've decided we need.
1837
return [rev_id for rev_id in source_ids if rev_id in result_set]
1838
1839
1840
class InterSameDataRepository(InterRepository):
1841
"""Code for converting between repositories that represent the same data.
1842
1843
Data format and model must match for this to work.
1844
"""
1845
1846
_matching_repo_format = RepositoryFormat4()
1847
"""Repository format for testing with."""
1848
1849
@staticmethod
1850
def is_compatible(source, target):
1851
if not isinstance(source, Repository):
1852
return False
1853
if not isinstance(target, Repository):
1854
return False
1855
if source._format.rich_root_data == target._format.rich_root_data:
1856
return True
1857
else:
1858
return False
1859
1860
@needs_write_lock
1861
def copy_content(self, revision_id=None, basis=None):
1862
"""Make a complete copy of the content in self into destination.
1863
1864
This is a destructive operation! Do not use it on existing
1865
repositories.
1866
1867
:param revision_id: Only copy the content needed to construct
1868
revision_id and its parents.
1869
:param basis: Copy the needed data preferentially from basis.
1870
"""
1871
try:
1872
self.target.set_make_working_trees(self.source.make_working_trees())
1873
except NotImplementedError:
1874
pass
1875
# grab the basis available data
1876
if basis is not None:
1877
self.target.fetch(basis, revision_id=revision_id)
1878
# but don't bother fetching if we have the needed data now.
1879
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1880
self.target.has_revision(revision_id)):
1881
return
1882
self.target.fetch(self.source, revision_id=revision_id)
1883
1884
@needs_write_lock
1885
def fetch(self, revision_id=None, pb=None):
1886
"""See InterRepository.fetch()."""
1887
from bzrlib.fetch import GenericRepoFetcher
1888
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1889
self.source, self.source._format, self.target,
1890
self.target._format)
1891
f = GenericRepoFetcher(to_repository=self.target,
1892
from_repository=self.source,
1893
last_revision=revision_id,
1894
pb=pb)
1895
return f.count_copied, f.failed_revisions
1896
1897
1898
class InterWeaveRepo(InterSameDataRepository):
1899
"""Optimised code paths between Weave based repositories."""
1900
1901
_matching_repo_format = RepositoryFormat7()
1902
"""Repository format for testing with."""
1903
1904
@staticmethod
1905
def is_compatible(source, target):
1906
"""Be compatible with known Weave formats.
1907
1908
We don't test for the stores being of specific types because that
1909
could lead to confusing results, and there is no need to be
1910
overly general.
1911
"""
1912
try:
1913
return (isinstance(source._format, (RepositoryFormat5,
1914
RepositoryFormat6,
1915
RepositoryFormat7)) and
1916
isinstance(target._format, (RepositoryFormat5,
1917
RepositoryFormat6,
1918
RepositoryFormat7)))
1919
except AttributeError:
1920
return False
1921
1922
@needs_write_lock
1923
def copy_content(self, revision_id=None, basis=None):
1924
"""See InterRepository.copy_content()."""
1925
# weave specific optimised path:
1926
if basis is not None:
1927
# copy the basis in, then fetch remaining data.
1928
basis.copy_content_into(self.target, revision_id)
1929
# the basis copy_content_into could miss-set this.
1930
try:
1931
self.target.set_make_working_trees(self.source.make_working_trees())
1932
except NotImplementedError:
1933
pass
1934
self.target.fetch(self.source, revision_id=revision_id)
1935
else:
1936
try:
1937
self.target.set_make_working_trees(self.source.make_working_trees())
1938
except NotImplementedError:
1939
pass
1940
# FIXME do not peek!
1941
if self.source.control_files._transport.listable():
1942
pb = ui.ui_factory.nested_progress_bar()
1943
try:
1944
self.target.weave_store.copy_all_ids(
1945
self.source.weave_store,
1946
pb=pb,
1947
from_transaction=self.source.get_transaction(),
1948
to_transaction=self.target.get_transaction())
1949
pb.update('copying inventory', 0, 1)
1950
self.target.control_weaves.copy_multi(
1951
self.source.control_weaves, ['inventory'],
1952
from_transaction=self.source.get_transaction(),
1953
to_transaction=self.target.get_transaction())
1954
self.target._revision_store.text_store.copy_all_ids(
1955
self.source._revision_store.text_store,
1956
pb=pb)
1957
finally:
1958
pb.finished()
1959
else:
1960
self.target.fetch(self.source, revision_id=revision_id)
1961
1962
@needs_write_lock
1963
def fetch(self, revision_id=None, pb=None):
1964
"""See InterRepository.fetch()."""
1965
from bzrlib.fetch import GenericRepoFetcher
1966
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1967
self.source, self.source._format, self.target, self.target._format)
1968
f = GenericRepoFetcher(to_repository=self.target,
1969
from_repository=self.source,
1970
last_revision=revision_id,
1971
pb=pb)
1972
return f.count_copied, f.failed_revisions
1973
1974
@needs_read_lock
1975
def missing_revision_ids(self, revision_id=None):
1976
"""See InterRepository.missing_revision_ids()."""
1977
# we want all revisions to satisfy revision_id in source.
1978
# but we don't want to stat every file here and there.
1979
# we want then, all revisions other needs to satisfy revision_id
1980
# checked, but not those that we have locally.
1981
# so the first thing is to get a subset of the revisions to
1982
# satisfy revision_id in source, and then eliminate those that
1983
# we do already have.
1984
# this is slow on high latency connection to self, but as as this
1985
# disk format scales terribly for push anyway due to rewriting
1986
# inventory.weave, this is considered acceptable.
1987
# - RBC 20060209
1988
if revision_id is not None:
1989
source_ids = self.source.get_ancestry(revision_id)
1990
assert source_ids[0] is None
1991
source_ids.pop(0)
1992
else:
1993
source_ids = self.source._all_possible_ids()
1994
source_ids_set = set(source_ids)
1995
# source_ids is the worst possible case we may need to pull.
1996
# now we want to filter source_ids against what we actually
1997
# have in target, but don't try to check for existence where we know
1998
# we do not have a revision as that would be pointless.
1999
target_ids = set(self.target._all_possible_ids())
2000
possibly_present_revisions = target_ids.intersection(source_ids_set)
2001
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2002
required_revisions = source_ids_set.difference(actually_present_revisions)
2003
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2004
if revision_id is not None:
2005
# we used get_ancestry to determine source_ids then we are assured all
2006
# revisions referenced are present as they are installed in topological order.
2007
# and the tip revision was validated by get_ancestry.
2008
return required_topo_revisions
2009
else:
2010
# if we just grabbed the possibly available ids, then
2011
# we only have an estimate of whats available and need to validate
2012
# that against the revision records.
2013
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2014
2015
2016
class InterKnitRepo(InterSameDataRepository):
2017
"""Optimised code paths between Knit based repositories."""
2018
2019
_matching_repo_format = RepositoryFormatKnit1()
2020
"""Repository format for testing with."""
2021
2022
@staticmethod
2023
def is_compatible(source, target):
2024
"""Be compatible with known Knit formats.
2025
2026
We don't test for the stores being of specific types because that
2027
could lead to confusing results, and there is no need to be
2028
overly general.
2029
"""
2030
try:
2031
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2032
isinstance(target._format, (RepositoryFormatKnit1)))
2033
except AttributeError:
2034
return False
2035
2036
@needs_write_lock
2037
def fetch(self, revision_id=None, pb=None):
2038
"""See InterRepository.fetch()."""
2039
from bzrlib.fetch import KnitRepoFetcher
2040
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2041
self.source, self.source._format, self.target, self.target._format)
2042
f = KnitRepoFetcher(to_repository=self.target,
2043
from_repository=self.source,
2044
last_revision=revision_id,
2045
pb=pb)
2046
return f.count_copied, f.failed_revisions
2047
2048
@needs_read_lock
2049
def missing_revision_ids(self, revision_id=None):
2050
"""See InterRepository.missing_revision_ids()."""
2051
if revision_id is not None:
2052
source_ids = self.source.get_ancestry(revision_id)
2053
assert source_ids[0] is None
2054
source_ids.pop(0)
2055
else:
2056
source_ids = self.source._all_possible_ids()
2057
source_ids_set = set(source_ids)
2058
# source_ids is the worst possible case we may need to pull.
2059
# now we want to filter source_ids against what we actually
2060
# have in target, but don't try to check for existence where we know
2061
# we do not have a revision as that would be pointless.
2062
target_ids = set(self.target._all_possible_ids())
2063
possibly_present_revisions = target_ids.intersection(source_ids_set)
2064
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2065
required_revisions = source_ids_set.difference(actually_present_revisions)
2066
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2067
if revision_id is not None:
2068
# we used get_ancestry to determine source_ids then we are assured all
2069
# revisions referenced are present as they are installed in topological order.
2070
# and the tip revision was validated by get_ancestry.
2071
return required_topo_revisions
2072
else:
2073
# if we just grabbed the possibly available ids, then
2074
# we only have an estimate of whats available and need to validate
2075
# that against the revision records.
2076
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2077
2078
2079
class InterModel1and2(InterRepository):
2080
2081
_matching_repo_format = None
2082
2083
@staticmethod
2084
def is_compatible(source, target):
2085
if not isinstance(source, Repository):
2086
return False
2087
if not isinstance(target, Repository):
2088
return False
2089
if not source._format.rich_root_data and target._format.rich_root_data:
2090
return True
2091
else:
2092
return False
2093
2094
@needs_write_lock
2095
def fetch(self, revision_id=None, pb=None):
2096
"""See InterRepository.fetch()."""
2097
from bzrlib.fetch import Model1toKnit2Fetcher
2098
f = Model1toKnit2Fetcher(to_repository=self.target,
2099
from_repository=self.source,
2100
last_revision=revision_id,
2101
pb=pb)
2102
return f.count_copied, f.failed_revisions
2103
2104
@needs_write_lock
2105
def copy_content(self, revision_id=None, basis=None):
2106
"""Make a complete copy of the content in self into destination.
2107
2108
This is a destructive operation! Do not use it on existing
2109
repositories.
2110
2111
:param revision_id: Only copy the content needed to construct
2112
revision_id and its parents.
2113
:param basis: Copy the needed data preferentially from basis.
2114
"""
2115
try:
2116
self.target.set_make_working_trees(self.source.make_working_trees())
2117
except NotImplementedError:
2118
pass
2119
# grab the basis available data
2120
if basis is not None:
2121
self.target.fetch(basis, revision_id=revision_id)
2122
# but don't bother fetching if we have the needed data now.
2123
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2124
self.target.has_revision(revision_id)):
2125
return
2126
self.target.fetch(self.source, revision_id=revision_id)
2127
2128
2129
class InterKnit1and2(InterKnitRepo):
2130
2131
_matching_repo_format = None
2132
2133
@staticmethod
2134
def is_compatible(source, target):
2135
"""Be compatible with Knit1 source and Knit2 target"""
2136
try:
2137
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2138
isinstance(target._format, (RepositoryFormatKnit2)))
2139
except AttributeError:
2140
return False
2141
2142
@needs_write_lock
2143
def fetch(self, revision_id=None, pb=None):
2144
"""See InterRepository.fetch()."""
2145
from bzrlib.fetch import Knit1to2Fetcher
2146
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2147
self.source, self.source._format, self.target,
2148
self.target._format)
2149
f = Knit1to2Fetcher(to_repository=self.target,
2150
from_repository=self.source,
2151
last_revision=revision_id,
2152
pb=pb)
2153
return f.count_copied, f.failed_revisions
2154
2155
2156
InterRepository.register_optimiser(InterSameDataRepository)
2157
InterRepository.register_optimiser(InterWeaveRepo)
2158
InterRepository.register_optimiser(InterKnitRepo)
2159
InterRepository.register_optimiser(InterModel1and2)
2160
InterRepository.register_optimiser(InterKnit1and2)
2161
2162
2163
class RepositoryTestProviderAdapter(object):
2164
"""A tool to generate a suite testing multiple repository formats at once.
2165
2166
This is done by copying the test once for each transport and injecting
2167
the transport_server, transport_readonly_server, and bzrdir_format and
2168
repository_format classes into each copy. Each copy is also given a new id()
2169
to make it easy to identify.
2170
"""
2171
2172
def __init__(self, transport_server, transport_readonly_server, formats):
2173
self._transport_server = transport_server
2174
self._transport_readonly_server = transport_readonly_server
2175
self._formats = formats
2176
2177
def adapt(self, test):
2178
result = unittest.TestSuite()
2179
for repository_format, bzrdir_format in self._formats:
2180
new_test = deepcopy(test)
2181
new_test.transport_server = self._transport_server
2182
new_test.transport_readonly_server = self._transport_readonly_server
2183
new_test.bzrdir_format = bzrdir_format
2184
new_test.repository_format = repository_format
2185
def make_new_test_id():
2186
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2187
return lambda: new_id
2188
new_test.id = make_new_test_id()
2189
result.addTest(new_test)
2190
return result
2191
2192
2193
class InterRepositoryTestProviderAdapter(object):
2194
"""A tool to generate a suite testing multiple inter repository formats.
2195
2196
This is done by copying the test once for each interrepo provider and injecting
2197
the transport_server, transport_readonly_server, repository_format and
2198
repository_to_format classes into each copy.
2199
Each copy is also given a new id() 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 interrepo_class, repository_format, repository_format_to 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.interrepo_class = interrepo_class
2214
new_test.repository_format = repository_format
2215
new_test.repository_format_to = repository_format_to
2216
def make_new_test_id():
2217
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2218
return lambda: new_id
2219
new_test.id = make_new_test_id()
2220
result.addTest(new_test)
2221
return result
2222
2223
@staticmethod
2224
def default_test_list():
2225
"""Generate the default list of interrepo permutations to test."""
2226
result = []
2227
# test the default InterRepository between format 6 and the current
2228
# default format.
2229
# XXX: robertc 20060220 reinstate this when there are two supported
2230
# formats which do not have an optimal code path between them.
2231
#result.append((InterRepository,
2232
# RepositoryFormat6(),
2233
# RepositoryFormatKnit1()))
2234
for optimiser in InterRepository._optimisers:
2235
if optimiser._matching_repo_format is not None:
2236
result.append((optimiser,
2237
optimiser._matching_repo_format,
2238
optimiser._matching_repo_format
2239
))
2240
# if there are specific combinations we want to use, we can add them
2241
# here.
2242
result.append((InterModel1and2, RepositoryFormat5(),
2243
RepositoryFormatKnit2()))
2244
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2245
RepositoryFormatKnit2()))
2246
return result
2247
2248
2249
class CopyConverter(object):
2250
"""A repository conversion tool which just performs a copy of the content.
2251
2252
This is slow but quite reliable.
2253
"""
2254
2255
def __init__(self, target_format):
2256
"""Create a CopyConverter.
2257
2258
:param target_format: The format the resulting repository should be.
2259
"""
2260
self.target_format = target_format
2261
2262
def convert(self, repo, pb):
2263
"""Perform the conversion of to_convert, giving feedback via pb.
2264
2265
:param to_convert: The disk object to convert.
2266
:param pb: a progress bar to use for progress information.
2267
"""
2268
self.pb = pb
2269
self.count = 0
2270
self.total = 4
2271
# this is only useful with metadir layouts - separated repo content.
2272
# trigger an assertion if not such
2273
repo._format.get_format_string()
2274
self.repo_dir = repo.bzrdir
2275
self.step('Moving repository to repository.backup')
2276
self.repo_dir.transport.move('repository', 'repository.backup')
2277
backup_transport = self.repo_dir.transport.clone('repository.backup')
2278
repo._format.check_conversion_target(self.target_format)
2279
self.source_repo = repo._format.open(self.repo_dir,
2280
_found=True,
2281
_override_transport=backup_transport)
2282
self.step('Creating new repository')
2283
converted = self.target_format.initialize(self.repo_dir,
2284
self.source_repo.is_shared())
2285
converted.lock_write()
2286
try:
2287
self.step('Copying content into repository.')
2288
self.source_repo.copy_content_into(converted)
2289
finally:
2290
converted.unlock()
2291
self.step('Deleting old repository content.')
2292
self.repo_dir.transport.delete_tree('repository.backup')
2293
self.pb.note('repository converted')
2294
2295
def step(self, message):
2296
"""Update the pb by a step."""
2297
self.count +=1
2298
self.pb.update(message, self.count, self.total)
2299
2300
2301
class CommitBuilder(object):
2302
"""Provides an interface to build up a commit.
2303
2304
This allows describing a tree to be committed without needing to
2305
know the internals of the format of the repository.
2306
"""
2307
2308
record_root_entry = False
2309
def __init__(self, repository, parents, config, timestamp=None,
2310
timezone=None, committer=None, revprops=None,
2311
revision_id=None):
2312
"""Initiate a CommitBuilder.
2313
2314
:param repository: Repository to commit to.
2315
:param parents: Revision ids of the parents of the new revision.
2316
:param config: Configuration to use.
2317
:param timestamp: Optional timestamp recorded for commit.
2318
:param timezone: Optional timezone for timestamp.
2319
:param committer: Optional committer to set for commit.
2320
:param revprops: Optional dictionary of revision properties.
2321
:param revision_id: Optional revision id.
2322
"""
2323
self._config = config
2324
2325
if committer is None:
2326
self._committer = self._config.username()
2327
else:
2328
assert isinstance(committer, basestring), type(committer)
2329
self._committer = committer
2330
2331
self.new_inventory = Inventory(None)
2332
self._new_revision_id = revision_id
2333
self.parents = parents
2334
self.repository = repository
2335
2336
self._revprops = {}
2337
if revprops is not None:
2338
self._revprops.update(revprops)
2339
2340
if timestamp is None:
2341
timestamp = time.time()
2342
# Restrict resolution to 1ms
2343
self._timestamp = round(timestamp, 3)
2344
2345
if timezone is None:
2346
self._timezone = local_time_offset()
2347
else:
2348
self._timezone = int(timezone)
2349
2350
self._generate_revision_if_needed()
2351
2352
def commit(self, message):
2353
"""Make the actual commit.
2354
2355
:return: The revision id of the recorded revision.
2356
"""
2357
rev = _mod_revision.Revision(
2358
timestamp=self._timestamp,
2359
timezone=self._timezone,
2360
committer=self._committer,
2361
message=message,
2362
inventory_sha1=self.inv_sha1,
2363
revision_id=self._new_revision_id,
2364
properties=self._revprops)
2365
rev.parent_ids = self.parents
2366
self.repository.add_revision(self._new_revision_id, rev,
2367
self.new_inventory, self._config)
2368
return self._new_revision_id
2369
2370
def revision_tree(self):
2371
"""Return the tree that was just committed.
2372
2373
After calling commit() this can be called to get a RevisionTree
2374
representing the newly committed tree. This is preferred to
2375
calling Repository.revision_tree() because that may require
2376
deserializing the inventory, while we already have a copy in
2377
memory.
2378
"""
2379
return RevisionTree(self.repository, self.new_inventory,
2380
self._new_revision_id)
2381
2382
def finish_inventory(self):
2383
"""Tell the builder that the inventory is finished."""
2384
if self.new_inventory.root is None:
2385
symbol_versioning.warn('Root entry should be supplied to'
2386
' record_entry_contents, as of bzr 0.10.',
2387
DeprecationWarning, stacklevel=2)
2388
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2389
self.new_inventory.revision_id = self._new_revision_id
2390
self.inv_sha1 = self.repository.add_inventory(
2391
self._new_revision_id,
2392
self.new_inventory,
2393
self.parents
2394
)
2395
2396
def _gen_revision_id(self):
2397
"""Return new revision-id."""
2398
return generate_ids.gen_revision_id(self._config.username(),
2399
self._timestamp)
2400
2401
def _generate_revision_if_needed(self):
2402
"""Create a revision id if None was supplied.
2403
2404
If the repository can not support user-specified revision ids
2405
they should override this function and raise CannotSetRevisionId
2406
if _new_revision_id is not None.
2407
2408
:raises: CannotSetRevisionId
2409
"""
2410
if self._new_revision_id is None:
2411
self._new_revision_id = self._gen_revision_id()
2412
2413
def record_entry_contents(self, ie, parent_invs, path, tree):
2414
"""Record the content of ie from tree into the commit if needed.
2415
2416
Side effect: sets ie.revision when unchanged
2417
2418
:param ie: An inventory entry present in the commit.
2419
:param parent_invs: The inventories of the parent revisions of the
2420
commit.
2421
:param path: The path the entry is at in the tree.
2422
:param tree: The tree which contains this entry and should be used to
2423
obtain content.
2424
"""
2425
if self.new_inventory.root is None and ie.parent_id is not None:
2426
symbol_versioning.warn('Root entry should be supplied to'
2427
' record_entry_contents, as of bzr 0.10.',
2428
DeprecationWarning, stacklevel=2)
2429
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2430
'', tree)
2431
self.new_inventory.add(ie)
2432
2433
# ie.revision is always None if the InventoryEntry is considered
2434
# for committing. ie.snapshot will record the correct revision
2435
# which may be the sole parent if it is untouched.
2436
if ie.revision is not None:
2437
return
2438
2439
# In this revision format, root entries have no knit or weave
2440
if ie is self.new_inventory.root:
2441
# When serializing out to disk and back in
2442
# root.revision is always _new_revision_id
2443
ie.revision = self._new_revision_id
2444
return
2445
previous_entries = ie.find_previous_heads(
2446
parent_invs,
2447
self.repository.weave_store,
2448
self.repository.get_transaction())
2449
# we are creating a new revision for ie in the history store
2450
# and inventory.
2451
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2452
2453
def modified_directory(self, file_id, file_parents):
2454
"""Record the presence of a symbolic link.
2455
2456
:param file_id: The file_id of the link to record.
2457
:param file_parents: The per-file parent revision ids.
2458
"""
2459
self._add_text_to_weave(file_id, [], file_parents.keys())
2460
2461
def modified_file_text(self, file_id, file_parents,
2462
get_content_byte_lines, text_sha1=None,
2463
text_size=None):
2464
"""Record the text of file file_id
2465
2466
:param file_id: The file_id of the file to record the text of.
2467
:param file_parents: The per-file parent revision ids.
2468
:param get_content_byte_lines: A callable which will return the byte
2469
lines for the file.
2470
:param text_sha1: Optional SHA1 of the file contents.
2471
:param text_size: Optional size of the file contents.
2472
"""
2473
# mutter('storing text of file {%s} in revision {%s} into %r',
2474
# file_id, self._new_revision_id, self.repository.weave_store)
2475
# special case to avoid diffing on renames or
2476
# reparenting
2477
if (len(file_parents) == 1
2478
and text_sha1 == file_parents.values()[0].text_sha1
2479
and text_size == file_parents.values()[0].text_size):
2480
previous_ie = file_parents.values()[0]
2481
versionedfile = self.repository.weave_store.get_weave(file_id,
2482
self.repository.get_transaction())
2483
versionedfile.clone_text(self._new_revision_id,
2484
previous_ie.revision, file_parents.keys())
2485
return text_sha1, text_size
2486
else:
2487
new_lines = get_content_byte_lines()
2488
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2489
# should return the SHA1 and size
2490
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2491
return osutils.sha_strings(new_lines), \
2492
sum(map(len, new_lines))
2493
2494
def modified_link(self, file_id, file_parents, link_target):
2495
"""Record the presence of a symbolic link.
2496
2497
:param file_id: The file_id of the link to record.
2498
:param file_parents: The per-file parent revision ids.
2499
:param link_target: Target location of this link.
2500
"""
2501
self._add_text_to_weave(file_id, [], file_parents.keys())
2502
2503
def _add_text_to_weave(self, file_id, new_lines, parents):
2504
versionedfile = self.repository.weave_store.get_weave_or_empty(
2505
file_id, self.repository.get_transaction())
2506
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2507
versionedfile.clear_cache()
2508
2509
2510
class _CommitBuilder(CommitBuilder):
2511
"""Temporary class so old CommitBuilders are detected properly
2512
2513
Note: CommitBuilder works whether or not root entry is recorded.
2514
"""
2515
2516
record_root_entry = True
2517
2518
2519
class RootCommitBuilder(CommitBuilder):
2520
"""This commitbuilder actually records the root id"""
2521
2522
record_root_entry = True
2523
2524
def record_entry_contents(self, ie, parent_invs, path, tree):
2525
"""Record the content of ie from tree into the commit if needed.
2526
2527
Side effect: sets ie.revision when unchanged
2528
2529
:param ie: An inventory entry present in the commit.
2530
:param parent_invs: The inventories of the parent revisions of the
2531
commit.
2532
:param path: The path the entry is at in the tree.
2533
:param tree: The tree which contains this entry and should be used to
2534
obtain content.
2535
"""
2536
assert self.new_inventory.root is not None or ie.parent_id is None
2537
self.new_inventory.add(ie)
2538
2539
# ie.revision is always None if the InventoryEntry is considered
2540
# for committing. ie.snapshot will record the correct revision
2541
# which may be the sole parent if it is untouched.
2542
if ie.revision is not None:
2543
return
2544
2545
previous_entries = ie.find_previous_heads(
2546
parent_invs,
2547
self.repository.weave_store,
2548
self.repository.get_transaction())
2549
# we are creating a new revision for ie in the history store
2550
# and inventory.
2551
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2552
2553
2554
_unescape_map = {
2555
'apos':"'",
2556
'quot':'"',
2557
'amp':'&',
2558
'lt':'<',
2559
'gt':'>'
2560
}
2561
2562
2563
def _unescaper(match, _map=_unescape_map):
2564
return _map[match.group(1)]
2565
2566
2567
_unescape_re = None
2568
2569
2570
def _unescape_xml(data):
2571
"""Unescape predefined XML entities in a string of data."""
2572
global _unescape_re
2573
if _unescape_re is None:
2574
_unescape_re = re.compile('\&([^;]*);')
2575
return _unescape_re.sub(_unescaper, data)
2576
2577
2578
def _unescape_xml_cached(data, cache):
2579
try:
2580
return cache[data]
2581
except KeyError:
2582
unescaped = _unescape_xml(data)
2583
cache[data] = unescaped
2584
return unescaped
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0