(root)/brz/remove-bazaar
» Revision
2220.3.1
(compared to revision 0.200.1036)
: bzrlib/repository.py
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 2220.3.1
- compare with another revision
- download diff
- download tarball
- view history from revision 2220.3.1
- Committer: Martin Pool
- Date: 2007-02-20 04:47:54 UTC
- mto: (2220.2.19 tags)
- mto: This revision was merged to the branch mainline in revision 2309.
- Revision ID: mbp@sourcefrog.net-20070220044754-ptovhd212hs4vkkf
add bencode utility
- 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, 2007 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
######################################################################
75
# Repositories
76
77
class Repository(object):
78
"""Repository holding history for one or more branches.
79
80
The repository holds and retrieves historical information including
81
revisions and file history. It's normally accessed only by the Branch,
82
which views a particular line of development through that history.
83
84
The Repository builds on top of Stores and a Transport, which respectively
85
describe the disk data format and the way of accessing the (possibly
86
remote) disk.
87
"""
88
89
_file_ids_altered_regex = lazy_regex.lazy_compile(
90
r'file_id="(?P<file_id>[^"]+)"'
91
r'.*revision="(?P<revision_id>[^"]+)"'
92
)
93
94
@needs_write_lock
95
def add_inventory(self, revid, inv, parents):
96
"""Add the inventory inv to the repository as revid.
97
98
:param parents: The revision ids of the parents that revid
99
is known to have and are in the repository already.
100
101
returns the sha1 of the serialized inventory.
102
"""
103
assert inv.revision_id is None or inv.revision_id == revid, \
104
"Mismatch between inventory revision" \
105
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
106
assert inv.root is not None
107
inv_text = self.serialise_inventory(inv)
108
inv_sha1 = osutils.sha_string(inv_text)
109
inv_vf = self.control_weaves.get_weave('inventory',
110
self.get_transaction())
111
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
112
return inv_sha1
113
114
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
115
final_parents = []
116
for parent in parents:
117
if parent in inv_vf:
118
final_parents.append(parent)
119
120
inv_vf.add_lines(revid, final_parents, lines)
121
122
@needs_write_lock
123
def add_revision(self, rev_id, rev, inv=None, config=None):
124
"""Add rev to the revision store as rev_id.
125
126
:param rev_id: the revision id to use.
127
:param rev: The revision object.
128
:param inv: The inventory for the revision. if None, it will be looked
129
up in the inventory storer
130
:param config: If None no digital signature will be created.
131
If supplied its signature_needed method will be used
132
to determine if a signature should be made.
133
"""
134
if config is not None and config.signature_needed():
135
if inv is None:
136
inv = self.get_inventory(rev_id)
137
plaintext = Testament(rev, inv).as_short_text()
138
self.store_revision_signature(
139
gpg.GPGStrategy(config), plaintext, rev_id)
140
if not rev_id in self.get_inventory_weave():
141
if inv is None:
142
raise errors.WeaveRevisionNotPresent(rev_id,
143
self.get_inventory_weave())
144
else:
145
# yes, this is not suitable for adding with ghosts.
146
self.add_inventory(rev_id, inv, rev.parent_ids)
147
self._revision_store.add_revision(rev, self.get_transaction())
148
149
@needs_read_lock
150
def _all_possible_ids(self):
151
"""Return all the possible revisions that we could find."""
152
return self.get_inventory_weave().versions()
153
154
def all_revision_ids(self):
155
"""Returns a list of all the revision ids in the repository.
156
157
This is deprecated because code should generally work on the graph
158
reachable from a particular revision, and ignore any other revisions
159
that might be present. There is no direct replacement method.
160
"""
161
return self._all_revision_ids()
162
163
@needs_read_lock
164
def _all_revision_ids(self):
165
"""Returns a list of all the revision ids in the repository.
166
167
These are in as much topological order as the underlying store can
168
present: for weaves ghosts may lead to a lack of correctness until
169
the reweave updates the parents list.
170
"""
171
if self._revision_store.text_store.listable():
172
return self._revision_store.all_revision_ids(self.get_transaction())
173
result = self._all_possible_ids()
174
return self._eliminate_revisions_not_present(result)
175
176
def break_lock(self):
177
"""Break a lock if one is present from another instance.
178
179
Uses the ui factory to ask for confirmation if the lock may be from
180
an active process.
181
"""
182
self.control_files.break_lock()
183
184
@needs_read_lock
185
def _eliminate_revisions_not_present(self, revision_ids):
186
"""Check every revision id in revision_ids to see if we have it.
187
188
Returns a set of the present revisions.
189
"""
190
result = []
191
for id in revision_ids:
192
if self.has_revision(id):
193
result.append(id)
194
return result
195
196
@staticmethod
197
def create(a_bzrdir):
198
"""Construct the current default format repository in a_bzrdir."""
199
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
200
201
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
202
"""instantiate a Repository.
203
204
:param _format: The format of the repository on disk.
205
:param a_bzrdir: The BzrDir of the repository.
206
207
In the future we will have a single api for all stores for
208
getting file texts, inventories and revisions, then
209
this construct will accept instances of those things.
210
"""
211
super(Repository, self).__init__()
212
self._format = _format
213
# the following are part of the public API for Repository:
214
self.bzrdir = a_bzrdir
215
self.control_files = control_files
216
self._revision_store = _revision_store
217
self.text_store = text_store
218
# backwards compatibility
219
self.weave_store = text_store
220
# not right yet - should be more semantically clear ?
221
#
222
self.control_store = control_store
223
self.control_weaves = control_store
224
# TODO: make sure to construct the right store classes, etc, depending
225
# on whether escaping is required.
226
self._warn_if_deprecated()
227
self._serializer = xml5.serializer_v5
228
229
def __repr__(self):
230
return '%s(%r)' % (self.__class__.__name__,
231
self.bzrdir.transport.base)
232
233
def is_locked(self):
234
return self.control_files.is_locked()
235
236
def lock_write(self):
237
self.control_files.lock_write()
238
239
def lock_read(self):
240
self.control_files.lock_read()
241
242
def get_physical_lock_status(self):
243
return self.control_files.get_physical_lock_status()
244
245
@needs_read_lock
246
def missing_revision_ids(self, other, revision_id=None):
247
"""Return the revision ids that other has that this does not.
248
249
These are returned in topological order.
250
251
revision_id: only return revision ids included by revision_id.
252
"""
253
return InterRepository.get(other, self).missing_revision_ids(revision_id)
254
255
@staticmethod
256
def open(base):
257
"""Open the repository rooted at base.
258
259
For instance, if the repository is at URL/.bzr/repository,
260
Repository.open(URL) -> a Repository instance.
261
"""
262
control = bzrdir.BzrDir.open(base)
263
return control.open_repository()
264
265
def copy_content_into(self, destination, revision_id=None, basis=None):
266
"""Make a complete copy of the content in self into destination.
267
268
This is a destructive operation! Do not use it on existing
269
repositories.
270
"""
271
return InterRepository.get(self, destination).copy_content(revision_id, basis)
272
273
def fetch(self, source, revision_id=None, pb=None):
274
"""Fetch the content required to construct revision_id from source.
275
276
If revision_id is None all content is copied.
277
"""
278
return InterRepository.get(source, self).fetch(revision_id=revision_id,
279
pb=pb)
280
281
def get_commit_builder(self, branch, parents, config, timestamp=None,
282
timezone=None, committer=None, revprops=None,
283
revision_id=None):
284
"""Obtain a CommitBuilder for this repository.
285
286
:param branch: Branch to commit to.
287
:param parents: Revision ids of the parents of the new revision.
288
:param config: Configuration to use.
289
:param timestamp: Optional timestamp recorded for commit.
290
:param timezone: Optional timezone for timestamp.
291
:param committer: Optional committer to set for commit.
292
:param revprops: Optional dictionary of revision properties.
293
:param revision_id: Optional revision id.
294
"""
295
return _CommitBuilder(self, parents, config, timestamp, timezone,
296
committer, revprops, revision_id)
297
298
def unlock(self):
299
self.control_files.unlock()
300
301
@needs_read_lock
302
def clone(self, a_bzrdir, revision_id=None, basis=None):
303
"""Clone this repository into a_bzrdir using the current format.
304
305
Currently no check is made that the format of this repository and
306
the bzrdir format are compatible. FIXME RBC 20060201.
307
"""
308
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
309
# use target default format.
310
result = a_bzrdir.create_repository()
311
# FIXME RBC 20060209 split out the repository type to avoid this check ?
312
elif isinstance(a_bzrdir._format,
313
(bzrdir.BzrDirFormat4,
314
bzrdir.BzrDirFormat5,
315
bzrdir.BzrDirFormat6)):
316
result = a_bzrdir.open_repository()
317
else:
318
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
319
self.copy_content_into(result, revision_id, basis)
320
return result
321
322
@needs_read_lock
323
def has_revision(self, revision_id):
324
"""True if this repository has a copy of the revision."""
325
return self._revision_store.has_revision_id(revision_id,
326
self.get_transaction())
327
328
@needs_read_lock
329
def get_revision_reconcile(self, revision_id):
330
"""'reconcile' helper routine that allows access to a revision always.
331
332
This variant of get_revision does not cross check the weave graph
333
against the revision one as get_revision does: but it should only
334
be used by reconcile, or reconcile-alike commands that are correcting
335
or testing the revision graph.
336
"""
337
if not revision_id or not isinstance(revision_id, basestring):
338
raise errors.InvalidRevisionId(revision_id=revision_id,
339
branch=self)
340
return self._revision_store.get_revisions([revision_id],
341
self.get_transaction())[0]
342
@needs_read_lock
343
def get_revisions(self, revision_ids):
344
return self._revision_store.get_revisions(revision_ids,
345
self.get_transaction())
346
347
@needs_read_lock
348
def get_revision_xml(self, revision_id):
349
rev = self.get_revision(revision_id)
350
rev_tmp = StringIO()
351
# the current serializer..
352
self._revision_store._serializer.write_revision(rev, rev_tmp)
353
rev_tmp.seek(0)
354
return rev_tmp.getvalue()
355
356
@needs_read_lock
357
def get_revision(self, revision_id):
358
"""Return the Revision object for a named revision"""
359
r = self.get_revision_reconcile(revision_id)
360
# weave corruption can lead to absent revision markers that should be
361
# present.
362
# the following test is reasonably cheap (it needs a single weave read)
363
# and the weave is cached in read transactions. In write transactions
364
# it is not cached but typically we only read a small number of
365
# revisions. For knits when they are introduced we will probably want
366
# to ensure that caching write transactions are in use.
367
inv = self.get_inventory_weave()
368
self._check_revision_parents(r, inv)
369
return r
370
371
@needs_read_lock
372
def get_deltas_for_revisions(self, revisions):
373
"""Produce a generator of revision deltas.
374
375
Note that the input is a sequence of REVISIONS, not revision_ids.
376
Trees will be held in memory until the generator exits.
377
Each delta is relative to the revision's lefthand predecessor.
378
"""
379
required_trees = set()
380
for revision in revisions:
381
required_trees.add(revision.revision_id)
382
required_trees.update(revision.parent_ids[:1])
383
trees = dict((t.get_revision_id(), t) for
384
t in self.revision_trees(required_trees))
385
for revision in revisions:
386
if not revision.parent_ids:
387
old_tree = self.revision_tree(None)
388
else:
389
old_tree = trees[revision.parent_ids[0]]
390
yield trees[revision.revision_id].changes_from(old_tree)
391
392
@needs_read_lock
393
def get_revision_delta(self, revision_id):
394
"""Return the delta for one revision.
395
396
The delta is relative to the left-hand predecessor of the
397
revision.
398
"""
399
r = self.get_revision(revision_id)
400
return list(self.get_deltas_for_revisions([r]))[0]
401
402
def _check_revision_parents(self, revision, inventory):
403
"""Private to Repository and Fetch.
404
405
This checks the parentage of revision in an inventory weave for
406
consistency and is only applicable to inventory-weave-for-ancestry
407
using repository formats & fetchers.
408
"""
409
weave_parents = inventory.get_parents(revision.revision_id)
410
weave_names = inventory.versions()
411
for parent_id in revision.parent_ids:
412
if parent_id in weave_names:
413
# this parent must not be a ghost.
414
if not parent_id in weave_parents:
415
# but it is a ghost
416
raise errors.CorruptRepository(self)
417
418
@needs_write_lock
419
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
420
signature = gpg_strategy.sign(plaintext)
421
self._revision_store.add_revision_signature_text(revision_id,
422
signature,
423
self.get_transaction())
424
425
def fileids_altered_by_revision_ids(self, revision_ids):
426
"""Find the file ids and versions affected by revisions.
427
428
:param revisions: an iterable containing revision ids.
429
:return: a dictionary mapping altered file-ids to an iterable of
430
revision_ids. Each altered file-ids has the exact revision_ids that
431
altered it listed explicitly.
432
"""
433
assert self._serializer.support_altered_by_hack, \
434
("fileids_altered_by_revision_ids only supported for branches "
435
"which store inventory as unnested xml, not on %r" % self)
436
selected_revision_ids = set(revision_ids)
437
w = self.get_inventory_weave()
438
result = {}
439
440
# this code needs to read every new line in every inventory for the
441
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
442
# not present in one of those inventories is unnecessary but not
443
# harmful because we are filtering by the revision id marker in the
444
# inventory lines : we only select file ids altered in one of those
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
unescape_revid_cache = {}
449
unescape_fileid_cache = {}
450
451
# jam 20061218 In a big fetch, this handles hundreds of thousands
452
# of lines, so it has had a lot of inlining and optimizing done.
453
# Sorry that it is a little bit messy.
454
# Move several functions to be local variables, since this is a long
455
# running loop.
456
search = self._file_ids_altered_regex.search
457
unescape = _unescape_xml
458
setdefault = result.setdefault
459
pb = ui.ui_factory.nested_progress_bar()
460
try:
461
for line in w.iter_lines_added_or_present_in_versions(
462
selected_revision_ids, pb=pb):
463
match = search(line)
464
if match is None:
465
continue
466
# One call to match.group() returning multiple items is quite a
467
# bit faster than 2 calls to match.group() each returning 1
468
file_id, revision_id = match.group('file_id', 'revision_id')
469
470
# Inlining the cache lookups helps a lot when you make 170,000
471
# lines and 350k ids, versus 8.4 unique ids.
472
# Using a cache helps in 2 ways:
473
# 1) Avoids unnecessary decoding calls
474
# 2) Re-uses cached strings, which helps in future set and
475
# equality checks.
476
# (2) is enough that removing encoding entirely along with
477
# the cache (so we are using plain strings) results in no
478
# performance improvement.
479
try:
480
revision_id = unescape_revid_cache[revision_id]
481
except KeyError:
482
unescaped = unescape(revision_id)
483
unescape_revid_cache[revision_id] = unescaped
484
revision_id = unescaped
485
486
if revision_id in selected_revision_ids:
487
try:
488
file_id = unescape_fileid_cache[file_id]
489
except KeyError:
490
unescaped = unescape(file_id)
491
unescape_fileid_cache[file_id] = unescaped
492
file_id = unescaped
493
setdefault(file_id, set()).add(revision_id)
494
finally:
495
pb.finished()
496
return result
497
498
@needs_read_lock
499
def get_inventory_weave(self):
500
return self.control_weaves.get_weave('inventory',
501
self.get_transaction())
502
503
@needs_read_lock
504
def get_inventory(self, revision_id):
505
"""Get Inventory object by hash."""
506
return self.deserialise_inventory(
507
revision_id, self.get_inventory_xml(revision_id))
508
509
def deserialise_inventory(self, revision_id, xml):
510
"""Transform the xml into an inventory object.
511
512
:param revision_id: The expected revision id of the inventory.
513
:param xml: A serialised inventory.
514
"""
515
result = self._serializer.read_inventory_from_string(xml)
516
result.root.revision = revision_id
517
return result
518
519
def serialise_inventory(self, inv):
520
return self._serializer.write_inventory_to_string(inv)
521
522
@needs_read_lock
523
def get_inventory_xml(self, revision_id):
524
"""Get inventory XML as a file object."""
525
try:
526
assert isinstance(revision_id, basestring), type(revision_id)
527
iw = self.get_inventory_weave()
528
return iw.get_text(revision_id)
529
except IndexError:
530
raise errors.HistoryMissing(self, 'inventory', revision_id)
531
532
@needs_read_lock
533
def get_inventory_sha1(self, revision_id):
534
"""Return the sha1 hash of the inventory entry
535
"""
536
return self.get_revision(revision_id).inventory_sha1
537
538
@needs_read_lock
539
def get_revision_graph(self, revision_id=None):
540
"""Return a dictionary containing the revision graph.
541
542
:param revision_id: The revision_id to get a graph from. If None, then
543
the entire revision graph is returned. This is a deprecated mode of
544
operation and will be removed in the future.
545
:return: a dictionary of revision_id->revision_parents_list.
546
"""
547
# special case NULL_REVISION
548
if revision_id == _mod_revision.NULL_REVISION:
549
return {}
550
a_weave = self.get_inventory_weave()
551
all_revisions = self._eliminate_revisions_not_present(
552
a_weave.versions())
553
entire_graph = dict([(node, a_weave.get_parents(node)) for
554
node in all_revisions])
555
if revision_id is None:
556
return entire_graph
557
elif revision_id not in entire_graph:
558
raise errors.NoSuchRevision(self, revision_id)
559
else:
560
# add what can be reached from revision_id
561
result = {}
562
pending = set([revision_id])
563
while len(pending) > 0:
564
node = pending.pop()
565
result[node] = entire_graph[node]
566
for revision_id in result[node]:
567
if revision_id not in result:
568
pending.add(revision_id)
569
return result
570
571
@needs_read_lock
572
def get_revision_graph_with_ghosts(self, revision_ids=None):
573
"""Return a graph of the revisions with ghosts marked as applicable.
574
575
:param revision_ids: an iterable of revisions to graph or None for all.
576
:return: a Graph object with the graph reachable from revision_ids.
577
"""
578
result = graph.Graph()
579
if not revision_ids:
580
pending = set(self.all_revision_ids())
581
required = set([])
582
else:
583
pending = set(revision_ids)
584
# special case NULL_REVISION
585
if _mod_revision.NULL_REVISION in pending:
586
pending.remove(_mod_revision.NULL_REVISION)
587
required = set(pending)
588
done = set([])
589
while len(pending):
590
revision_id = pending.pop()
591
try:
592
rev = self.get_revision(revision_id)
593
except errors.NoSuchRevision:
594
if revision_id in required:
595
raise
596
# a ghost
597
result.add_ghost(revision_id)
598
continue
599
for parent_id in rev.parent_ids:
600
# is this queued or done ?
601
if (parent_id not in pending and
602
parent_id not in done):
603
# no, queue it.
604
pending.add(parent_id)
605
result.add_node(revision_id, rev.parent_ids)
606
done.add(revision_id)
607
return result
608
609
@needs_read_lock
610
def get_revision_inventory(self, revision_id):
611
"""Return inventory of a past revision."""
612
# TODO: Unify this with get_inventory()
613
# bzr 0.0.6 and later imposes the constraint that the inventory_id
614
# must be the same as its revision, so this is trivial.
615
if revision_id is None:
616
# This does not make sense: if there is no revision,
617
# then it is the current tree inventory surely ?!
618
# and thus get_root_id() is something that looks at the last
619
# commit on the branch, and the get_root_id is an inventory check.
620
raise NotImplementedError
621
# return Inventory(self.get_root_id())
622
else:
623
return self.get_inventory(revision_id)
624
625
@needs_read_lock
626
def is_shared(self):
627
"""Return True if this repository is flagged as a shared repository."""
628
raise NotImplementedError(self.is_shared)
629
630
@needs_write_lock
631
def reconcile(self, other=None, thorough=False):
632
"""Reconcile this repository."""
633
from bzrlib.reconcile import RepoReconciler
634
reconciler = RepoReconciler(self, thorough=thorough)
635
reconciler.reconcile()
636
return reconciler
637
638
@needs_read_lock
639
def revision_tree(self, revision_id):
640
"""Return Tree for a revision on this branch.
641
642
`revision_id` may be None for the empty tree revision.
643
"""
644
# TODO: refactor this to use an existing revision object
645
# so we don't need to read it in twice.
646
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
647
return RevisionTree(self, Inventory(root_id=None),
648
_mod_revision.NULL_REVISION)
649
else:
650
inv = self.get_revision_inventory(revision_id)
651
return RevisionTree(self, inv, revision_id)
652
653
@needs_read_lock
654
def revision_trees(self, revision_ids):
655
"""Return Tree for a revision on this branch.
656
657
`revision_id` may not be None or 'null:'"""
658
assert None not in revision_ids
659
assert _mod_revision.NULL_REVISION not in revision_ids
660
texts = self.get_inventory_weave().get_texts(revision_ids)
661
for text, revision_id in zip(texts, revision_ids):
662
inv = self.deserialise_inventory(revision_id, text)
663
yield RevisionTree(self, inv, revision_id)
664
665
@needs_read_lock
666
def get_ancestry(self, revision_id):
667
"""Return a list of revision-ids integrated by a revision.
668
669
The first element of the list is always None, indicating the origin
670
revision. This might change when we have history horizons, or
671
perhaps we should have a new API.
672
673
This is topologically sorted.
674
"""
675
if revision_id is None:
676
return [None]
677
if not self.has_revision(revision_id):
678
raise errors.NoSuchRevision(self, revision_id)
679
w = self.get_inventory_weave()
680
candidates = w.get_ancestry(revision_id)
681
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
682
683
@needs_read_lock
684
def print_file(self, file, revision_id):
685
"""Print `file` to stdout.
686
687
FIXME RBC 20060125 as John Meinel points out this is a bad api
688
- it writes to stdout, it assumes that that is valid etc. Fix
689
by creating a new more flexible convenience function.
690
"""
691
tree = self.revision_tree(revision_id)
692
# use inventory as it was in that revision
693
file_id = tree.inventory.path2id(file)
694
if not file_id:
695
# TODO: jam 20060427 Write a test for this code path
696
# it had a bug in it, and was raising the wrong
697
# exception.
698
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
699
tree.print_file(file_id)
700
701
def get_transaction(self):
702
return self.control_files.get_transaction()
703
704
def revision_parents(self, revid):
705
return self.get_inventory_weave().parent_names(revid)
706
707
@needs_write_lock
708
def set_make_working_trees(self, new_value):
709
"""Set the policy flag for making working trees when creating branches.
710
711
This only applies to branches that use this repository.
712
713
The default is 'True'.
714
:param new_value: True to restore the default, False to disable making
715
working trees.
716
"""
717
raise NotImplementedError(self.set_make_working_trees)
718
719
def make_working_trees(self):
720
"""Returns the policy for making working trees on new branches."""
721
raise NotImplementedError(self.make_working_trees)
722
723
@needs_write_lock
724
def sign_revision(self, revision_id, gpg_strategy):
725
plaintext = Testament.from_revision(self, revision_id).as_short_text()
726
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
727
728
@needs_read_lock
729
def has_signature_for_revision_id(self, revision_id):
730
"""Query for a revision signature for revision_id in the repository."""
731
return self._revision_store.has_signature(revision_id,
732
self.get_transaction())
733
734
@needs_read_lock
735
def get_signature_text(self, revision_id):
736
"""Return the text for a signature."""
737
return self._revision_store.get_signature_text(revision_id,
738
self.get_transaction())
739
740
@needs_read_lock
741
def check(self, revision_ids):
742
"""Check consistency of all history of given revision_ids.
743
744
Different repository implementations should override _check().
745
746
:param revision_ids: A non-empty list of revision_ids whose ancestry
747
will be checked. Typically the last revision_id of a branch.
748
"""
749
if not revision_ids:
750
raise ValueError("revision_ids must be non-empty in %s.check"
751
% (self,))
752
return self._check(revision_ids)
753
754
def _check(self, revision_ids):
755
result = check.Check(self)
756
result.check()
757
return result
758
759
def _warn_if_deprecated(self):
760
global _deprecation_warning_done
761
if _deprecation_warning_done:
762
return
763
_deprecation_warning_done = True
764
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
765
% (self._format, self.bzrdir.transport.base))
766
767
def supports_rich_root(self):
768
return self._format.rich_root_data
769
770
def _check_ascii_revisionid(self, revision_id, method):
771
"""Private helper for ascii-only repositories."""
772
# weave repositories refuse to store revisionids that are non-ascii.
773
if revision_id is not None:
774
# weaves require ascii revision ids.
775
if isinstance(revision_id, unicode):
776
try:
777
revision_id.encode('ascii')
778
except UnicodeEncodeError:
779
raise errors.NonAsciiRevisionId(method, self)
780
781
782
class AllInOneRepository(Repository):
783
"""Legacy support - the repository behaviour for all-in-one branches."""
784
785
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
786
# we reuse one control files instance.
787
dir_mode = a_bzrdir._control_files._dir_mode
788
file_mode = a_bzrdir._control_files._file_mode
789
790
def get_store(name, compressed=True, prefixed=False):
791
# FIXME: This approach of assuming stores are all entirely compressed
792
# or entirely uncompressed is tidy, but breaks upgrade from
793
# some existing branches where there's a mixture; we probably
794
# still want the option to look for both.
795
relpath = a_bzrdir._control_files._escape(name)
796
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
797
prefixed=prefixed, compressed=compressed,
798
dir_mode=dir_mode,
799
file_mode=file_mode)
800
#if self._transport.should_cache():
801
# cache_path = os.path.join(self.cache_root, name)
802
# os.mkdir(cache_path)
803
# store = bzrlib.store.CachedStore(store, cache_path)
804
return store
805
806
# not broken out yet because the controlweaves|inventory_store
807
# and text_store | weave_store bits are still different.
808
if isinstance(_format, RepositoryFormat4):
809
# cannot remove these - there is still no consistent api
810
# which allows access to this old info.
811
self.inventory_store = get_store('inventory-store')
812
text_store = get_store('text-store')
813
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
814
815
def get_commit_builder(self, branch, parents, config, timestamp=None,
816
timezone=None, committer=None, revprops=None,
817
revision_id=None):
818
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
819
return Repository.get_commit_builder(self, branch, parents, config,
820
timestamp, timezone, committer, revprops, revision_id)
821
822
@needs_read_lock
823
def is_shared(self):
824
"""AllInOne repositories cannot be shared."""
825
return False
826
827
@needs_write_lock
828
def set_make_working_trees(self, new_value):
829
"""Set the policy flag for making working trees when creating branches.
830
831
This only applies to branches that use this repository.
832
833
The default is 'True'.
834
:param new_value: True to restore the default, False to disable making
835
working trees.
836
"""
837
raise NotImplementedError(self.set_make_working_trees)
838
839
def make_working_trees(self):
840
"""Returns the policy for making working trees on new branches."""
841
return True
842
843
844
def install_revision(repository, rev, revision_tree):
845
"""Install all revision data into a repository."""
846
present_parents = []
847
parent_trees = {}
848
for p_id in rev.parent_ids:
849
if repository.has_revision(p_id):
850
present_parents.append(p_id)
851
parent_trees[p_id] = repository.revision_tree(p_id)
852
else:
853
parent_trees[p_id] = repository.revision_tree(None)
854
855
inv = revision_tree.inventory
856
entries = inv.iter_entries()
857
# backwards compatability hack: skip the root id.
858
if not repository.supports_rich_root():
859
path, root = entries.next()
860
if root.revision != rev.revision_id:
861
raise errors.IncompatibleRevision(repr(repository))
862
# Add the texts that are not already present
863
for path, ie in entries:
864
w = repository.weave_store.get_weave_or_empty(ie.file_id,
865
repository.get_transaction())
866
if ie.revision not in w:
867
text_parents = []
868
# FIXME: TODO: The following loop *may* be overlapping/duplicate
869
# with InventoryEntry.find_previous_heads(). if it is, then there
870
# is a latent bug here where the parents may have ancestors of each
871
# other. RBC, AB
872
for revision, tree in parent_trees.iteritems():
873
if ie.file_id not in tree:
874
continue
875
parent_id = tree.inventory[ie.file_id].revision
876
if parent_id in text_parents:
877
continue
878
text_parents.append(parent_id)
879
880
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
881
repository.get_transaction())
882
lines = revision_tree.get_file(ie.file_id).readlines()
883
vfile.add_lines(rev.revision_id, text_parents, lines)
884
try:
885
# install the inventory
886
repository.add_inventory(rev.revision_id, inv, present_parents)
887
except errors.RevisionAlreadyPresent:
888
pass
889
repository.add_revision(rev.revision_id, rev, inv)
890
891
892
class MetaDirRepository(Repository):
893
"""Repositories in the new meta-dir layout."""
894
895
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
896
super(MetaDirRepository, self).__init__(_format,
897
a_bzrdir,
898
control_files,
899
_revision_store,
900
control_store,
901
text_store)
902
dir_mode = self.control_files._dir_mode
903
file_mode = self.control_files._file_mode
904
905
@needs_read_lock
906
def is_shared(self):
907
"""Return True if this repository is flagged as a shared repository."""
908
return self.control_files._transport.has('shared-storage')
909
910
@needs_write_lock
911
def set_make_working_trees(self, new_value):
912
"""Set the policy flag for making working trees when creating branches.
913
914
This only applies to branches that use this repository.
915
916
The default is 'True'.
917
:param new_value: True to restore the default, False to disable making
918
working trees.
919
"""
920
if new_value:
921
try:
922
self.control_files._transport.delete('no-working-trees')
923
except errors.NoSuchFile:
924
pass
925
else:
926
self.control_files.put_utf8('no-working-trees', '')
927
928
def make_working_trees(self):
929
"""Returns the policy for making working trees on new branches."""
930
return not self.control_files._transport.has('no-working-trees')
931
932
933
class WeaveMetaDirRepository(MetaDirRepository):
934
"""A subclass of MetaDirRepository to set weave specific policy."""
935
936
def get_commit_builder(self, branch, parents, config, timestamp=None,
937
timezone=None, committer=None, revprops=None,
938
revision_id=None):
939
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
940
return MetaDirRepository.get_commit_builder(self, branch, parents,
941
config, timestamp, timezone, committer, revprops, revision_id)
942
943
944
class KnitRepository(MetaDirRepository):
945
"""Knit format repository."""
946
947
def _warn_if_deprecated(self):
948
# This class isn't deprecated
949
pass
950
951
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
952
inv_vf.add_lines_with_ghosts(revid, parents, lines)
953
954
@needs_read_lock
955
def _all_revision_ids(self):
956
"""See Repository.all_revision_ids()."""
957
# Knits get the revision graph from the index of the revision knit, so
958
# it's always possible even if they're on an unlistable transport.
959
return self._revision_store.all_revision_ids(self.get_transaction())
960
961
def fileid_involved_between_revs(self, from_revid, to_revid):
962
"""Find file_id(s) which are involved in the changes between revisions.
963
964
This determines the set of revisions which are involved, and then
965
finds all file ids affected by those revisions.
966
"""
967
vf = self._get_revision_vf()
968
from_set = set(vf.get_ancestry(from_revid))
969
to_set = set(vf.get_ancestry(to_revid))
970
changed = to_set.difference(from_set)
971
return self._fileid_involved_by_set(changed)
972
973
def fileid_involved(self, last_revid=None):
974
"""Find all file_ids modified in the ancestry of last_revid.
975
976
:param last_revid: If None, last_revision() will be used.
977
"""
978
if not last_revid:
979
changed = set(self.all_revision_ids())
980
else:
981
changed = set(self.get_ancestry(last_revid))
982
if None in changed:
983
changed.remove(None)
984
return self._fileid_involved_by_set(changed)
985
986
@needs_read_lock
987
def get_ancestry(self, revision_id):
988
"""Return a list of revision-ids integrated by a revision.
989
990
This is topologically sorted.
991
"""
992
if revision_id is None:
993
return [None]
994
vf = self._get_revision_vf()
995
try:
996
return [None] + vf.get_ancestry(revision_id)
997
except errors.RevisionNotPresent:
998
raise errors.NoSuchRevision(self, revision_id)
999
1000
@needs_read_lock
1001
def get_revision(self, revision_id):
1002
"""Return the Revision object for a named revision"""
1003
return self.get_revision_reconcile(revision_id)
1004
1005
@needs_read_lock
1006
def get_revision_graph(self, revision_id=None):
1007
"""Return a dictionary containing the revision graph.
1008
1009
:param revision_id: The revision_id to get a graph from. If None, then
1010
the entire revision graph is returned. This is a deprecated mode of
1011
operation and will be removed in the future.
1012
:return: a dictionary of revision_id->revision_parents_list.
1013
"""
1014
# special case NULL_REVISION
1015
if revision_id == _mod_revision.NULL_REVISION:
1016
return {}
1017
a_weave = self._get_revision_vf()
1018
entire_graph = a_weave.get_graph()
1019
if revision_id is None:
1020
return a_weave.get_graph()
1021
elif revision_id not in a_weave:
1022
raise errors.NoSuchRevision(self, revision_id)
1023
else:
1024
# add what can be reached from revision_id
1025
result = {}
1026
pending = set([revision_id])
1027
while len(pending) > 0:
1028
node = pending.pop()
1029
result[node] = a_weave.get_parents(node)
1030
for revision_id in result[node]:
1031
if revision_id not in result:
1032
pending.add(revision_id)
1033
return result
1034
1035
@needs_read_lock
1036
def get_revision_graph_with_ghosts(self, revision_ids=None):
1037
"""Return a graph of the revisions with ghosts marked as applicable.
1038
1039
:param revision_ids: an iterable of revisions to graph or None for all.
1040
:return: a Graph object with the graph reachable from revision_ids.
1041
"""
1042
result = graph.Graph()
1043
vf = self._get_revision_vf()
1044
versions = set(vf.versions())
1045
if not revision_ids:
1046
pending = set(self.all_revision_ids())
1047
required = set([])
1048
else:
1049
pending = set(revision_ids)
1050
# special case NULL_REVISION
1051
if _mod_revision.NULL_REVISION in pending:
1052
pending.remove(_mod_revision.NULL_REVISION)
1053
required = set(pending)
1054
done = set([])
1055
while len(pending):
1056
revision_id = pending.pop()
1057
if not revision_id in versions:
1058
if revision_id in required:
1059
raise errors.NoSuchRevision(self, revision_id)
1060
# a ghost
1061
result.add_ghost(revision_id)
1062
# mark it as done so we don't try for it again.
1063
done.add(revision_id)
1064
continue
1065
parent_ids = vf.get_parents_with_ghosts(revision_id)
1066
for parent_id in parent_ids:
1067
# is this queued or done ?
1068
if (parent_id not in pending and
1069
parent_id not in done):
1070
# no, queue it.
1071
pending.add(parent_id)
1072
result.add_node(revision_id, parent_ids)
1073
done.add(revision_id)
1074
return result
1075
1076
def _get_revision_vf(self):
1077
""":return: a versioned file containing the revisions."""
1078
vf = self._revision_store.get_revision_file(self.get_transaction())
1079
return vf
1080
1081
@needs_write_lock
1082
def reconcile(self, other=None, thorough=False):
1083
"""Reconcile this repository."""
1084
from bzrlib.reconcile import KnitReconciler
1085
reconciler = KnitReconciler(self, thorough=thorough)
1086
reconciler.reconcile()
1087
return reconciler
1088
1089
def revision_parents(self, revision_id):
1090
return self._get_revision_vf().get_parents(revision_id)
1091
1092
1093
class KnitRepository2(KnitRepository):
1094
"""Experimental enhanced knit format"""
1095
1096
# corresponds to RepositoryFormatKnit2
1097
1098
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1099
control_store, text_store):
1100
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1101
_revision_store, control_store, text_store)
1102
self._transport = control_files._transport
1103
self._serializer = xml6.serializer_v6
1104
1105
def deserialise_inventory(self, revision_id, xml):
1106
"""Transform the xml into an inventory object.
1107
1108
:param revision_id: The expected revision id of the inventory.
1109
:param xml: A serialised inventory.
1110
"""
1111
result = self._serializer.read_inventory_from_string(xml)
1112
assert result.root.revision is not None
1113
return result
1114
1115
def serialise_inventory(self, inv):
1116
"""Transform the inventory object into XML text.
1117
1118
:param revision_id: The expected revision id of the inventory.
1119
:param xml: A serialised inventory.
1120
"""
1121
assert inv.revision_id is not None
1122
assert inv.root.revision is not None
1123
return KnitRepository.serialise_inventory(self, inv)
1124
1125
def get_commit_builder(self, branch, parents, config, timestamp=None,
1126
timezone=None, committer=None, revprops=None,
1127
revision_id=None):
1128
"""Obtain a CommitBuilder for this repository.
1129
1130
:param branch: Branch to commit to.
1131
:param parents: Revision ids of the parents of the new revision.
1132
:param config: Configuration to use.
1133
:param timestamp: Optional timestamp recorded for commit.
1134
:param timezone: Optional timezone for timestamp.
1135
:param committer: Optional committer to set for commit.
1136
:param revprops: Optional dictionary of revision properties.
1137
:param revision_id: Optional revision id.
1138
"""
1139
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1140
committer, revprops, revision_id)
1141
1142
1143
#####################################################################
1144
# Repository Formats
1145
1146
class RepositoryFormat(object):
1147
"""A repository format.
1148
1149
Formats provide three things:
1150
* An initialization routine to construct repository data on disk.
1151
* a format string which is used when the BzrDir supports versioned
1152
children.
1153
* an open routine which returns a Repository instance.
1154
1155
Formats are placed in an dict by their format string for reference
1156
during opening. These should be subclasses of RepositoryFormat
1157
for consistency.
1158
1159
Once a format is deprecated, just deprecate the initialize and open
1160
methods on the format class. Do not deprecate the object, as the
1161
object will be created every system load.
1162
1163
Common instance attributes:
1164
_matchingbzrdir - the bzrdir format that the repository format was
1165
originally written to work with. This can be used if manually
1166
constructing a bzrdir and repository, or more commonly for test suite
1167
parameterisation.
1168
"""
1169
1170
_default_format = None
1171
"""The default format used for new repositories."""
1172
1173
_formats = {}
1174
"""The known formats."""
1175
1176
def __str__(self):
1177
return "<%s>" % self.__class__.__name__
1178
1179
@classmethod
1180
def find_format(klass, a_bzrdir):
1181
"""Return the format for the repository object in a_bzrdir."""
1182
try:
1183
transport = a_bzrdir.get_repository_transport(None)
1184
format_string = transport.get("format").read()
1185
return klass._formats[format_string]
1186
except errors.NoSuchFile:
1187
raise errors.NoRepositoryPresent(a_bzrdir)
1188
except KeyError:
1189
raise errors.UnknownFormatError(format=format_string)
1190
1191
def _get_control_store(self, repo_transport, control_files):
1192
"""Return the control store for this repository."""
1193
raise NotImplementedError(self._get_control_store)
1194
1195
@classmethod
1196
def get_default_format(klass):
1197
"""Return the current default format."""
1198
return klass._default_format
1199
1200
def get_format_string(self):
1201
"""Return the ASCII format string that identifies this format.
1202
1203
Note that in pre format ?? repositories the format string is
1204
not permitted nor written to disk.
1205
"""
1206
raise NotImplementedError(self.get_format_string)
1207
1208
def get_format_description(self):
1209
"""Return the short description for this format."""
1210
raise NotImplementedError(self.get_format_description)
1211
1212
def _get_revision_store(self, repo_transport, control_files):
1213
"""Return the revision store object for this a_bzrdir."""
1214
raise NotImplementedError(self._get_revision_store)
1215
1216
def _get_text_rev_store(self,
1217
transport,
1218
control_files,
1219
name,
1220
compressed=True,
1221
prefixed=False,
1222
serializer=None):
1223
"""Common logic for getting a revision store for a repository.
1224
1225
see self._get_revision_store for the subclass-overridable method to
1226
get the store for a repository.
1227
"""
1228
from bzrlib.store.revision.text import TextRevisionStore
1229
dir_mode = control_files._dir_mode
1230
file_mode = control_files._file_mode
1231
text_store = TextStore(transport.clone(name),
1232
prefixed=prefixed,
1233
compressed=compressed,
1234
dir_mode=dir_mode,
1235
file_mode=file_mode)
1236
_revision_store = TextRevisionStore(text_store, serializer)
1237
return _revision_store
1238
1239
def _get_versioned_file_store(self,
1240
name,
1241
transport,
1242
control_files,
1243
prefixed=True,
1244
versionedfile_class=weave.WeaveFile,
1245
versionedfile_kwargs={},
1246
escaped=False):
1247
weave_transport = control_files._transport.clone(name)
1248
dir_mode = control_files._dir_mode
1249
file_mode = control_files._file_mode
1250
return VersionedFileStore(weave_transport, prefixed=prefixed,
1251
dir_mode=dir_mode,
1252
file_mode=file_mode,
1253
versionedfile_class=versionedfile_class,
1254
versionedfile_kwargs=versionedfile_kwargs,
1255
escaped=escaped)
1256
1257
def initialize(self, a_bzrdir, shared=False):
1258
"""Initialize a repository of this format in a_bzrdir.
1259
1260
:param a_bzrdir: The bzrdir to put the new repository in it.
1261
:param shared: The repository should be initialized as a sharable one.
1262
1263
This may raise UninitializableFormat if shared repository are not
1264
compatible the a_bzrdir.
1265
"""
1266
1267
def is_supported(self):
1268
"""Is this format supported?
1269
1270
Supported formats must be initializable and openable.
1271
Unsupported formats may not support initialization or committing or
1272
some other features depending on the reason for not being supported.
1273
"""
1274
return True
1275
1276
def check_conversion_target(self, target_format):
1277
raise NotImplementedError(self.check_conversion_target)
1278
1279
def open(self, a_bzrdir, _found=False):
1280
"""Return an instance of this format for the bzrdir a_bzrdir.
1281
1282
_found is a private parameter, do not use it.
1283
"""
1284
raise NotImplementedError(self.open)
1285
1286
@classmethod
1287
def register_format(klass, format):
1288
klass._formats[format.get_format_string()] = format
1289
1290
@classmethod
1291
def set_default_format(klass, format):
1292
klass._default_format = format
1293
1294
@classmethod
1295
def unregister_format(klass, format):
1296
assert klass._formats[format.get_format_string()] is format
1297
del klass._formats[format.get_format_string()]
1298
1299
1300
class PreSplitOutRepositoryFormat(RepositoryFormat):
1301
"""Base class for the pre split out repository formats."""
1302
1303
rich_root_data = False
1304
1305
def initialize(self, a_bzrdir, shared=False, _internal=False):
1306
"""Create a weave repository.
1307
1308
TODO: when creating split out bzr branch formats, move this to a common
1309
base for Format5, Format6. or something like that.
1310
"""
1311
if shared:
1312
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1313
1314
if not _internal:
1315
# always initialized when the bzrdir is.
1316
return self.open(a_bzrdir, _found=True)
1317
1318
# Create an empty weave
1319
sio = StringIO()
1320
weavefile.write_weave_v5(weave.Weave(), sio)
1321
empty_weave = sio.getvalue()
1322
1323
mutter('creating repository in %s.', a_bzrdir.transport.base)
1324
dirs = ['revision-store', 'weaves']
1325
files = [('inventory.weave', StringIO(empty_weave)),
1326
]
1327
1328
# FIXME: RBC 20060125 don't peek under the covers
1329
# NB: no need to escape relative paths that are url safe.
1330
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1331
'branch-lock', lockable_files.TransportLock)
1332
control_files.create_lock()
1333
control_files.lock_write()
1334
control_files._transport.mkdir_multi(dirs,
1335
mode=control_files._dir_mode)
1336
try:
1337
for file, content in files:
1338
control_files.put(file, content)
1339
finally:
1340
control_files.unlock()
1341
return self.open(a_bzrdir, _found=True)
1342
1343
def _get_control_store(self, repo_transport, control_files):
1344
"""Return the control store for this repository."""
1345
return self._get_versioned_file_store('',
1346
repo_transport,
1347
control_files,
1348
prefixed=False)
1349
1350
def _get_text_store(self, transport, control_files):
1351
"""Get a store for file texts for this format."""
1352
raise NotImplementedError(self._get_text_store)
1353
1354
def open(self, a_bzrdir, _found=False):
1355
"""See RepositoryFormat.open()."""
1356
if not _found:
1357
# we are being called directly and must probe.
1358
raise NotImplementedError
1359
1360
repo_transport = a_bzrdir.get_repository_transport(None)
1361
control_files = a_bzrdir._control_files
1362
text_store = self._get_text_store(repo_transport, control_files)
1363
control_store = self._get_control_store(repo_transport, control_files)
1364
_revision_store = self._get_revision_store(repo_transport, control_files)
1365
return AllInOneRepository(_format=self,
1366
a_bzrdir=a_bzrdir,
1367
_revision_store=_revision_store,
1368
control_store=control_store,
1369
text_store=text_store)
1370
1371
def check_conversion_target(self, target_format):
1372
pass
1373
1374
1375
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1376
"""Bzr repository format 4.
1377
1378
This repository format has:
1379
- flat stores
1380
- TextStores for texts, inventories,revisions.
1381
1382
This format is deprecated: it indexes texts using a text id which is
1383
removed in format 5; initialization and write support for this format
1384
has been removed.
1385
"""
1386
1387
def __init__(self):
1388
super(RepositoryFormat4, self).__init__()
1389
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1390
1391
def get_format_description(self):
1392
"""See RepositoryFormat.get_format_description()."""
1393
return "Repository format 4"
1394
1395
def initialize(self, url, shared=False, _internal=False):
1396
"""Format 4 branches cannot be created."""
1397
raise errors.UninitializableFormat(self)
1398
1399
def is_supported(self):
1400
"""Format 4 is not supported.
1401
1402
It is not supported because the model changed from 4 to 5 and the
1403
conversion logic is expensive - so doing it on the fly was not
1404
feasible.
1405
"""
1406
return False
1407
1408
def _get_control_store(self, repo_transport, control_files):
1409
"""Format 4 repositories have no formal control store at this point.
1410
1411
This will cause any control-file-needing apis to fail - this is desired.
1412
"""
1413
return None
1414
1415
def _get_revision_store(self, repo_transport, control_files):
1416
"""See RepositoryFormat._get_revision_store()."""
1417
from bzrlib.xml4 import serializer_v4
1418
return self._get_text_rev_store(repo_transport,
1419
control_files,
1420
'revision-store',
1421
serializer=serializer_v4)
1422
1423
def _get_text_store(self, transport, control_files):
1424
"""See RepositoryFormat._get_text_store()."""
1425
1426
1427
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1428
"""Bzr control format 5.
1429
1430
This repository format has:
1431
- weaves for file texts and inventory
1432
- flat stores
1433
- TextStores for revisions and signatures.
1434
"""
1435
1436
def __init__(self):
1437
super(RepositoryFormat5, self).__init__()
1438
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1439
1440
def get_format_description(self):
1441
"""See RepositoryFormat.get_format_description()."""
1442
return "Weave repository format 5"
1443
1444
def _get_revision_store(self, repo_transport, control_files):
1445
"""See RepositoryFormat._get_revision_store()."""
1446
"""Return the revision store object for this a_bzrdir."""
1447
return self._get_text_rev_store(repo_transport,
1448
control_files,
1449
'revision-store',
1450
compressed=False)
1451
1452
def _get_text_store(self, transport, control_files):
1453
"""See RepositoryFormat._get_text_store()."""
1454
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1455
1456
1457
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1458
"""Bzr control format 6.
1459
1460
This repository format has:
1461
- weaves for file texts and inventory
1462
- hash subdirectory based stores.
1463
- TextStores for revisions and signatures.
1464
"""
1465
1466
def __init__(self):
1467
super(RepositoryFormat6, self).__init__()
1468
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1469
1470
def get_format_description(self):
1471
"""See RepositoryFormat.get_format_description()."""
1472
return "Weave repository format 6"
1473
1474
def _get_revision_store(self, repo_transport, control_files):
1475
"""See RepositoryFormat._get_revision_store()."""
1476
return self._get_text_rev_store(repo_transport,
1477
control_files,
1478
'revision-store',
1479
compressed=False,
1480
prefixed=True)
1481
1482
def _get_text_store(self, transport, control_files):
1483
"""See RepositoryFormat._get_text_store()."""
1484
return self._get_versioned_file_store('weaves', transport, control_files)
1485
1486
1487
class MetaDirRepositoryFormat(RepositoryFormat):
1488
"""Common base class for the new repositories using the metadir layout."""
1489
1490
rich_root_data = False
1491
1492
def __init__(self):
1493
super(MetaDirRepositoryFormat, self).__init__()
1494
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1495
1496
def _create_control_files(self, a_bzrdir):
1497
"""Create the required files and the initial control_files object."""
1498
# FIXME: RBC 20060125 don't peek under the covers
1499
# NB: no need to escape relative paths that are url safe.
1500
repository_transport = a_bzrdir.get_repository_transport(self)
1501
control_files = lockable_files.LockableFiles(repository_transport,
1502
'lock', lockdir.LockDir)
1503
control_files.create_lock()
1504
return control_files
1505
1506
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1507
"""Upload the initial blank content."""
1508
control_files = self._create_control_files(a_bzrdir)
1509
control_files.lock_write()
1510
try:
1511
control_files._transport.mkdir_multi(dirs,
1512
mode=control_files._dir_mode)
1513
for file, content in files:
1514
control_files.put(file, content)
1515
for file, content in utf8_files:
1516
control_files.put_utf8(file, content)
1517
if shared == True:
1518
control_files.put_utf8('shared-storage', '')
1519
finally:
1520
control_files.unlock()
1521
1522
1523
class RepositoryFormat7(MetaDirRepositoryFormat):
1524
"""Bzr repository 7.
1525
1526
This repository format has:
1527
- weaves for file texts and inventory
1528
- hash subdirectory based stores.
1529
- TextStores for revisions and signatures.
1530
- a format marker of its own
1531
- an optional 'shared-storage' flag
1532
- an optional 'no-working-trees' flag
1533
"""
1534
1535
def _get_control_store(self, repo_transport, control_files):
1536
"""Return the control store for this repository."""
1537
return self._get_versioned_file_store('',
1538
repo_transport,
1539
control_files,
1540
prefixed=False)
1541
1542
def get_format_string(self):
1543
"""See RepositoryFormat.get_format_string()."""
1544
return "Bazaar-NG Repository format 7"
1545
1546
def get_format_description(self):
1547
"""See RepositoryFormat.get_format_description()."""
1548
return "Weave repository format 7"
1549
1550
def check_conversion_target(self, target_format):
1551
pass
1552
1553
def _get_revision_store(self, repo_transport, control_files):
1554
"""See RepositoryFormat._get_revision_store()."""
1555
return self._get_text_rev_store(repo_transport,
1556
control_files,
1557
'revision-store',
1558
compressed=False,
1559
prefixed=True,
1560
)
1561
1562
def _get_text_store(self, transport, control_files):
1563
"""See RepositoryFormat._get_text_store()."""
1564
return self._get_versioned_file_store('weaves',
1565
transport,
1566
control_files)
1567
1568
def initialize(self, a_bzrdir, shared=False):
1569
"""Create a weave repository.
1570
1571
:param shared: If true the repository will be initialized as a shared
1572
repository.
1573
"""
1574
# Create an empty weave
1575
sio = StringIO()
1576
weavefile.write_weave_v5(weave.Weave(), sio)
1577
empty_weave = sio.getvalue()
1578
1579
mutter('creating repository in %s.', a_bzrdir.transport.base)
1580
dirs = ['revision-store', 'weaves']
1581
files = [('inventory.weave', StringIO(empty_weave)),
1582
]
1583
utf8_files = [('format', self.get_format_string())]
1584
1585
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1586
return self.open(a_bzrdir=a_bzrdir, _found=True)
1587
1588
def open(self, a_bzrdir, _found=False, _override_transport=None):
1589
"""See RepositoryFormat.open().
1590
1591
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1592
repository at a slightly different url
1593
than normal. I.e. during 'upgrade'.
1594
"""
1595
if not _found:
1596
format = RepositoryFormat.find_format(a_bzrdir)
1597
assert format.__class__ == self.__class__
1598
if _override_transport is not None:
1599
repo_transport = _override_transport
1600
else:
1601
repo_transport = a_bzrdir.get_repository_transport(None)
1602
control_files = lockable_files.LockableFiles(repo_transport,
1603
'lock', lockdir.LockDir)
1604
text_store = self._get_text_store(repo_transport, control_files)
1605
control_store = self._get_control_store(repo_transport, control_files)
1606
_revision_store = self._get_revision_store(repo_transport, control_files)
1607
return WeaveMetaDirRepository(_format=self,
1608
a_bzrdir=a_bzrdir,
1609
control_files=control_files,
1610
_revision_store=_revision_store,
1611
control_store=control_store,
1612
text_store=text_store)
1613
1614
1615
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1616
"""Bzr repository knit format (generalized).
1617
1618
This repository format has:
1619
- knits for file texts and inventory
1620
- hash subdirectory based stores.
1621
- knits for revisions and signatures
1622
- TextStores for revisions and signatures.
1623
- a format marker of its own
1624
- an optional 'shared-storage' flag
1625
- an optional 'no-working-trees' flag
1626
- a LockDir lock
1627
"""
1628
1629
def _get_control_store(self, repo_transport, control_files):
1630
"""Return the control store for this repository."""
1631
return VersionedFileStore(
1632
repo_transport,
1633
prefixed=False,
1634
file_mode=control_files._file_mode,
1635
versionedfile_class=knit.KnitVersionedFile,
1636
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1637
)
1638
1639
def _get_revision_store(self, repo_transport, control_files):
1640
"""See RepositoryFormat._get_revision_store()."""
1641
from bzrlib.store.revision.knit import KnitRevisionStore
1642
versioned_file_store = VersionedFileStore(
1643
repo_transport,
1644
file_mode=control_files._file_mode,
1645
prefixed=False,
1646
precious=True,
1647
versionedfile_class=knit.KnitVersionedFile,
1648
versionedfile_kwargs={'delta':False,
1649
'factory':knit.KnitPlainFactory(),
1650
},
1651
escaped=True,
1652
)
1653
return KnitRevisionStore(versioned_file_store)
1654
1655
def _get_text_store(self, transport, control_files):
1656
"""See RepositoryFormat._get_text_store()."""
1657
return self._get_versioned_file_store('knits',
1658
transport,
1659
control_files,
1660
versionedfile_class=knit.KnitVersionedFile,
1661
versionedfile_kwargs={
1662
'create_parent_dir':True,
1663
'delay_create':True,
1664
'dir_mode':control_files._dir_mode,
1665
},
1666
escaped=True)
1667
1668
def initialize(self, a_bzrdir, shared=False):
1669
"""Create a knit format 1 repository.
1670
1671
:param a_bzrdir: bzrdir to contain the new repository; must already
1672
be initialized.
1673
:param shared: If true the repository will be initialized as a shared
1674
repository.
1675
"""
1676
mutter('creating repository in %s.', a_bzrdir.transport.base)
1677
dirs = ['revision-store', 'knits']
1678
files = []
1679
utf8_files = [('format', self.get_format_string())]
1680
1681
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1682
repo_transport = a_bzrdir.get_repository_transport(None)
1683
control_files = lockable_files.LockableFiles(repo_transport,
1684
'lock', lockdir.LockDir)
1685
control_store = self._get_control_store(repo_transport, control_files)
1686
transaction = transactions.WriteTransaction()
1687
# trigger a write of the inventory store.
1688
control_store.get_weave_or_empty('inventory', transaction)
1689
_revision_store = self._get_revision_store(repo_transport, control_files)
1690
# the revision id here is irrelevant: it will not be stored, and cannot
1691
# already exist.
1692
_revision_store.has_revision_id('A', transaction)
1693
_revision_store.get_signature_file(transaction)
1694
return self.open(a_bzrdir=a_bzrdir, _found=True)
1695
1696
def open(self, a_bzrdir, _found=False, _override_transport=None):
1697
"""See RepositoryFormat.open().
1698
1699
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1700
repository at a slightly different url
1701
than normal. I.e. during 'upgrade'.
1702
"""
1703
if not _found:
1704
format = RepositoryFormat.find_format(a_bzrdir)
1705
assert format.__class__ == self.__class__
1706
if _override_transport is not None:
1707
repo_transport = _override_transport
1708
else:
1709
repo_transport = a_bzrdir.get_repository_transport(None)
1710
control_files = lockable_files.LockableFiles(repo_transport,
1711
'lock', lockdir.LockDir)
1712
text_store = self._get_text_store(repo_transport, control_files)
1713
control_store = self._get_control_store(repo_transport, control_files)
1714
_revision_store = self._get_revision_store(repo_transport, control_files)
1715
return KnitRepository(_format=self,
1716
a_bzrdir=a_bzrdir,
1717
control_files=control_files,
1718
_revision_store=_revision_store,
1719
control_store=control_store,
1720
text_store=text_store)
1721
1722
1723
class RepositoryFormatKnit1(RepositoryFormatKnit):
1724
"""Bzr repository knit format 1.
1725
1726
This repository format has:
1727
- knits for file texts and inventory
1728
- hash subdirectory based stores.
1729
- knits for revisions and signatures
1730
- TextStores for revisions and signatures.
1731
- a format marker of its own
1732
- an optional 'shared-storage' flag
1733
- an optional 'no-working-trees' flag
1734
- a LockDir lock
1735
1736
This format was introduced in bzr 0.8.
1737
"""
1738
def get_format_string(self):
1739
"""See RepositoryFormat.get_format_string()."""
1740
return "Bazaar-NG Knit Repository Format 1"
1741
1742
def get_format_description(self):
1743
"""See RepositoryFormat.get_format_description()."""
1744
return "Knit repository format 1"
1745
1746
def check_conversion_target(self, target_format):
1747
pass
1748
1749
1750
class RepositoryFormatKnit2(RepositoryFormatKnit):
1751
"""Bzr repository knit format 2.
1752
1753
THIS FORMAT IS EXPERIMENTAL
1754
This repository format has:
1755
- knits for file texts and inventory
1756
- hash subdirectory based stores.
1757
- knits for revisions and signatures
1758
- TextStores for revisions and signatures.
1759
- a format marker of its own
1760
- an optional 'shared-storage' flag
1761
- an optional 'no-working-trees' flag
1762
- a LockDir lock
1763
- Support for recording full info about the tree root
1764
"""
1765
1766
rich_root_data = True
1767
1768
def get_format_string(self):
1769
"""See RepositoryFormat.get_format_string()."""
1770
return "Bazaar Knit Repository Format 2\n"
1771
1772
def get_format_description(self):
1773
"""See RepositoryFormat.get_format_description()."""
1774
return "Knit repository format 2"
1775
1776
def check_conversion_target(self, target_format):
1777
if not target_format.rich_root_data:
1778
raise errors.BadConversionTarget(
1779
'Does not support rich root data.', target_format)
1780
1781
def open(self, a_bzrdir, _found=False, _override_transport=None):
1782
"""See RepositoryFormat.open().
1783
1784
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1785
repository at a slightly different url
1786
than normal. I.e. during 'upgrade'.
1787
"""
1788
if not _found:
1789
format = RepositoryFormat.find_format(a_bzrdir)
1790
assert format.__class__ == self.__class__
1791
if _override_transport is not None:
1792
repo_transport = _override_transport
1793
else:
1794
repo_transport = a_bzrdir.get_repository_transport(None)
1795
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1796
lockdir.LockDir)
1797
text_store = self._get_text_store(repo_transport, control_files)
1798
control_store = self._get_control_store(repo_transport, control_files)
1799
_revision_store = self._get_revision_store(repo_transport, control_files)
1800
return KnitRepository2(_format=self,
1801
a_bzrdir=a_bzrdir,
1802
control_files=control_files,
1803
_revision_store=_revision_store,
1804
control_store=control_store,
1805
text_store=text_store)
1806
1807
def initialize(self, a_bzrdir, shared=False):
1808
repo = super(RepositoryFormatKnit2, self).initialize(a_bzrdir, shared)
1809
repo._transport.put_bytes_non_atomic('tags', '')
1810
return repo
1811
1812
def supports_tags(self):
1813
return True
1814
1815
1816
1817
# formats which have no format string are not discoverable
1818
# and not independently creatable, so are not registered.
1819
RepositoryFormat.register_format(RepositoryFormat7())
1820
# KEEP in sync with bzrdir.format_registry default
1821
_default_format = RepositoryFormatKnit1()
1822
RepositoryFormat.register_format(_default_format)
1823
RepositoryFormat.register_format(RepositoryFormatKnit2())
1824
RepositoryFormat.set_default_format(_default_format)
1825
_legacy_formats = [RepositoryFormat4(),
1826
RepositoryFormat5(),
1827
RepositoryFormat6()]
1828
1829
1830
class InterRepository(InterObject):
1831
"""This class represents operations taking place between two repositories.
1832
1833
Its instances have methods like copy_content and fetch, and contain
1834
references to the source and target repositories these operations can be
1835
carried out on.
1836
1837
Often we will provide convenience methods on 'repository' which carry out
1838
operations with another repository - they will always forward to
1839
InterRepository.get(other).method_name(parameters).
1840
"""
1841
1842
_optimisers = []
1843
"""The available optimised InterRepository types."""
1844
1845
def copy_content(self, revision_id=None, basis=None):
1846
raise NotImplementedError(self.copy_content)
1847
1848
def fetch(self, revision_id=None, pb=None):
1849
"""Fetch the content required to construct revision_id.
1850
1851
The content is copied from self.source to self.target.
1852
1853
:param revision_id: if None all content is copied, if NULL_REVISION no
1854
content is copied.
1855
:param pb: optional progress bar to use for progress reports. If not
1856
provided a default one will be created.
1857
1858
Returns the copied revision count and the failed revisions in a tuple:
1859
(copied, failures).
1860
"""
1861
raise NotImplementedError(self.fetch)
1862
1863
@needs_read_lock
1864
def missing_revision_ids(self, revision_id=None):
1865
"""Return the revision ids that source has that target does not.
1866
1867
These are returned in topological order.
1868
1869
:param revision_id: only return revision ids included by this
1870
revision_id.
1871
"""
1872
# generic, possibly worst case, slow code path.
1873
target_ids = set(self.target.all_revision_ids())
1874
if revision_id is not None:
1875
source_ids = self.source.get_ancestry(revision_id)
1876
assert source_ids[0] is None
1877
source_ids.pop(0)
1878
else:
1879
source_ids = self.source.all_revision_ids()
1880
result_set = set(source_ids).difference(target_ids)
1881
# this may look like a no-op: its not. It preserves the ordering
1882
# other_ids had while only returning the members from other_ids
1883
# that we've decided we need.
1884
return [rev_id for rev_id in source_ids if rev_id in result_set]
1885
1886
def copy_tags(self):
1887
"""Copy all tags from source to the target."""
1888
# A default implementation is provided even though not all
1889
# Repositories will support tags... we'll just get an error back from
1890
# the underlying method.
1891
if self.target == self.source:
1892
return
1893
self.target.lock_write()
1894
try:
1895
self.target._set_tag_dict(self.source.get_tag_dict())
1896
finally:
1897
self.target.unlock()
1898
1899
1900
class InterSameDataRepository(InterRepository):
1901
"""Code for converting between repositories that represent the same data.
1902
1903
Data format and model must match for this to work.
1904
"""
1905
1906
_matching_repo_format = RepositoryFormat4()
1907
"""Repository format for testing with."""
1908
1909
@staticmethod
1910
def is_compatible(source, target):
1911
if not isinstance(source, Repository):
1912
return False
1913
if not isinstance(target, Repository):
1914
return False
1915
if source._format.rich_root_data == target._format.rich_root_data:
1916
return True
1917
else:
1918
return False
1919
1920
@needs_write_lock
1921
def copy_content(self, revision_id=None, basis=None):
1922
"""Make a complete copy of the content in self into destination.
1923
1924
This is a destructive operation! Do not use it on existing
1925
repositories.
1926
1927
:param revision_id: Only copy the content needed to construct
1928
revision_id and its parents.
1929
:param basis: Copy the needed data preferentially from basis.
1930
"""
1931
try:
1932
self.target.set_make_working_trees(self.source.make_working_trees())
1933
except NotImplementedError:
1934
pass
1935
# grab the basis available data
1936
if basis is not None:
1937
self.target.fetch(basis, revision_id=revision_id)
1938
# but don't bother fetching if we have the needed data now.
1939
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1940
self.target.has_revision(revision_id)):
1941
return
1942
self.target.fetch(self.source, revision_id=revision_id)
1943
1944
@needs_write_lock
1945
def fetch(self, revision_id=None, pb=None):
1946
"""See InterRepository.fetch()."""
1947
from bzrlib.fetch import GenericRepoFetcher
1948
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1949
self.source, self.source._format, self.target,
1950
self.target._format)
1951
f = GenericRepoFetcher(to_repository=self.target,
1952
from_repository=self.source,
1953
last_revision=revision_id,
1954
pb=pb)
1955
return f.count_copied, f.failed_revisions
1956
1957
1958
class InterWeaveRepo(InterSameDataRepository):
1959
"""Optimised code paths between Weave based repositories."""
1960
1961
_matching_repo_format = RepositoryFormat7()
1962
"""Repository format for testing with."""
1963
1964
@staticmethod
1965
def is_compatible(source, target):
1966
"""Be compatible with known Weave formats.
1967
1968
We don't test for the stores being of specific types because that
1969
could lead to confusing results, and there is no need to be
1970
overly general.
1971
"""
1972
try:
1973
return (isinstance(source._format, (RepositoryFormat5,
1974
RepositoryFormat6,
1975
RepositoryFormat7)) and
1976
isinstance(target._format, (RepositoryFormat5,
1977
RepositoryFormat6,
1978
RepositoryFormat7)))
1979
except AttributeError:
1980
return False
1981
1982
@needs_write_lock
1983
def copy_content(self, revision_id=None, basis=None):
1984
"""See InterRepository.copy_content()."""
1985
# weave specific optimised path:
1986
if basis is not None:
1987
# copy the basis in, then fetch remaining data.
1988
basis.copy_content_into(self.target, revision_id)
1989
# the basis copy_content_into could miss-set this.
1990
try:
1991
self.target.set_make_working_trees(self.source.make_working_trees())
1992
except NotImplementedError:
1993
pass
1994
self.target.fetch(self.source, revision_id=revision_id)
1995
else:
1996
try:
1997
self.target.set_make_working_trees(self.source.make_working_trees())
1998
except NotImplementedError:
1999
pass
2000
# FIXME do not peek!
2001
if self.source.control_files._transport.listable():
2002
pb = ui.ui_factory.nested_progress_bar()
2003
try:
2004
self.target.weave_store.copy_all_ids(
2005
self.source.weave_store,
2006
pb=pb,
2007
from_transaction=self.source.get_transaction(),
2008
to_transaction=self.target.get_transaction())
2009
pb.update('copying inventory', 0, 1)
2010
self.target.control_weaves.copy_multi(
2011
self.source.control_weaves, ['inventory'],
2012
from_transaction=self.source.get_transaction(),
2013
to_transaction=self.target.get_transaction())
2014
self.target._revision_store.text_store.copy_all_ids(
2015
self.source._revision_store.text_store,
2016
pb=pb)
2017
finally:
2018
pb.finished()
2019
else:
2020
self.target.fetch(self.source, revision_id=revision_id)
2021
2022
@needs_write_lock
2023
def fetch(self, revision_id=None, pb=None):
2024
"""See InterRepository.fetch()."""
2025
from bzrlib.fetch import GenericRepoFetcher
2026
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2027
self.source, self.source._format, self.target, self.target._format)
2028
f = GenericRepoFetcher(to_repository=self.target,
2029
from_repository=self.source,
2030
last_revision=revision_id,
2031
pb=pb)
2032
return f.count_copied, f.failed_revisions
2033
2034
@needs_read_lock
2035
def missing_revision_ids(self, revision_id=None):
2036
"""See InterRepository.missing_revision_ids()."""
2037
# we want all revisions to satisfy revision_id in source.
2038
# but we don't want to stat every file here and there.
2039
# we want then, all revisions other needs to satisfy revision_id
2040
# checked, but not those that we have locally.
2041
# so the first thing is to get a subset of the revisions to
2042
# satisfy revision_id in source, and then eliminate those that
2043
# we do already have.
2044
# this is slow on high latency connection to self, but as as this
2045
# disk format scales terribly for push anyway due to rewriting
2046
# inventory.weave, this is considered acceptable.
2047
# - RBC 20060209
2048
if revision_id is not None:
2049
source_ids = self.source.get_ancestry(revision_id)
2050
assert source_ids[0] is None
2051
source_ids.pop(0)
2052
else:
2053
source_ids = self.source._all_possible_ids()
2054
source_ids_set = set(source_ids)
2055
# source_ids is the worst possible case we may need to pull.
2056
# now we want to filter source_ids against what we actually
2057
# have in target, but don't try to check for existence where we know
2058
# we do not have a revision as that would be pointless.
2059
target_ids = set(self.target._all_possible_ids())
2060
possibly_present_revisions = target_ids.intersection(source_ids_set)
2061
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2062
required_revisions = source_ids_set.difference(actually_present_revisions)
2063
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2064
if revision_id is not None:
2065
# we used get_ancestry to determine source_ids then we are assured all
2066
# revisions referenced are present as they are installed in topological order.
2067
# and the tip revision was validated by get_ancestry.
2068
return required_topo_revisions
2069
else:
2070
# if we just grabbed the possibly available ids, then
2071
# we only have an estimate of whats available and need to validate
2072
# that against the revision records.
2073
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2074
2075
2076
class InterKnitRepo(InterSameDataRepository):
2077
"""Optimised code paths between Knit based repositories."""
2078
2079
_matching_repo_format = RepositoryFormatKnit1()
2080
"""Repository format for testing with."""
2081
2082
@staticmethod
2083
def is_compatible(source, target):
2084
"""Be compatible with known Knit formats.
2085
2086
We don't test for the stores being of specific types because that
2087
could lead to confusing results, and there is no need to be
2088
overly general.
2089
"""
2090
try:
2091
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2092
isinstance(target._format, (RepositoryFormatKnit1)))
2093
except AttributeError:
2094
return False
2095
2096
@needs_write_lock
2097
def fetch(self, revision_id=None, pb=None):
2098
"""See InterRepository.fetch()."""
2099
from bzrlib.fetch import KnitRepoFetcher
2100
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2101
self.source, self.source._format, self.target, self.target._format)
2102
f = KnitRepoFetcher(to_repository=self.target,
2103
from_repository=self.source,
2104
last_revision=revision_id,
2105
pb=pb)
2106
return f.count_copied, f.failed_revisions
2107
2108
@needs_read_lock
2109
def missing_revision_ids(self, revision_id=None):
2110
"""See InterRepository.missing_revision_ids()."""
2111
if revision_id is not None:
2112
source_ids = self.source.get_ancestry(revision_id)
2113
assert source_ids[0] is None
2114
source_ids.pop(0)
2115
else:
2116
source_ids = self.source._all_possible_ids()
2117
source_ids_set = set(source_ids)
2118
# source_ids is the worst possible case we may need to pull.
2119
# now we want to filter source_ids against what we actually
2120
# have in target, but don't try to check for existence where we know
2121
# we do not have a revision as that would be pointless.
2122
target_ids = set(self.target._all_possible_ids())
2123
possibly_present_revisions = target_ids.intersection(source_ids_set)
2124
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2125
required_revisions = source_ids_set.difference(actually_present_revisions)
2126
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2127
if revision_id is not None:
2128
# we used get_ancestry to determine source_ids then we are assured all
2129
# revisions referenced are present as they are installed in topological order.
2130
# and the tip revision was validated by get_ancestry.
2131
return required_topo_revisions
2132
else:
2133
# if we just grabbed the possibly available ids, then
2134
# we only have an estimate of whats available and need to validate
2135
# that against the revision records.
2136
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2137
2138
2139
class InterModel1and2(InterRepository):
2140
2141
_matching_repo_format = None
2142
2143
@staticmethod
2144
def is_compatible(source, target):
2145
if not isinstance(source, Repository):
2146
return False
2147
if not isinstance(target, Repository):
2148
return False
2149
if not source._format.rich_root_data and target._format.rich_root_data:
2150
return True
2151
else:
2152
return False
2153
2154
@needs_write_lock
2155
def fetch(self, revision_id=None, pb=None):
2156
"""See InterRepository.fetch()."""
2157
from bzrlib.fetch import Model1toKnit2Fetcher
2158
f = Model1toKnit2Fetcher(to_repository=self.target,
2159
from_repository=self.source,
2160
last_revision=revision_id,
2161
pb=pb)
2162
return f.count_copied, f.failed_revisions
2163
2164
@needs_write_lock
2165
def copy_content(self, revision_id=None, basis=None):
2166
"""Make a complete copy of the content in self into destination.
2167
2168
This is a destructive operation! Do not use it on existing
2169
repositories.
2170
2171
:param revision_id: Only copy the content needed to construct
2172
revision_id and its parents.
2173
:param basis: Copy the needed data preferentially from basis.
2174
"""
2175
try:
2176
self.target.set_make_working_trees(self.source.make_working_trees())
2177
except NotImplementedError:
2178
pass
2179
# grab the basis available data
2180
if basis is not None:
2181
self.target.fetch(basis, revision_id=revision_id)
2182
# but don't bother fetching if we have the needed data now.
2183
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2184
self.target.has_revision(revision_id)):
2185
return
2186
self.target.fetch(self.source, revision_id=revision_id)
2187
2188
2189
class InterKnit1and2(InterKnitRepo):
2190
2191
_matching_repo_format = None
2192
2193
@staticmethod
2194
def is_compatible(source, target):
2195
"""Be compatible with Knit1 source and Knit2 target"""
2196
try:
2197
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2198
isinstance(target._format, (RepositoryFormatKnit2)))
2199
except AttributeError:
2200
return False
2201
2202
@needs_write_lock
2203
def fetch(self, revision_id=None, pb=None):
2204
"""See InterRepository.fetch()."""
2205
from bzrlib.fetch import Knit1to2Fetcher
2206
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2207
self.source, self.source._format, self.target,
2208
self.target._format)
2209
f = Knit1to2Fetcher(to_repository=self.target,
2210
from_repository=self.source,
2211
last_revision=revision_id,
2212
pb=pb)
2213
return f.count_copied, f.failed_revisions
2214
2215
2216
InterRepository.register_optimiser(InterSameDataRepository)
2217
InterRepository.register_optimiser(InterWeaveRepo)
2218
InterRepository.register_optimiser(InterKnitRepo)
2219
InterRepository.register_optimiser(InterModel1and2)
2220
InterRepository.register_optimiser(InterKnit1and2)
2221
2222
2223
class RepositoryTestProviderAdapter(object):
2224
"""A tool to generate a suite testing multiple repository formats at once.
2225
2226
This is done by copying the test once for each transport and injecting
2227
the transport_server, transport_readonly_server, and bzrdir_format and
2228
repository_format classes into each copy. Each copy is also given a new id()
2229
to make it easy to identify.
2230
"""
2231
2232
def __init__(self, transport_server, transport_readonly_server, formats):
2233
self._transport_server = transport_server
2234
self._transport_readonly_server = transport_readonly_server
2235
self._formats = formats
2236
2237
def adapt(self, test):
2238
result = unittest.TestSuite()
2239
for repository_format, bzrdir_format in self._formats:
2240
new_test = deepcopy(test)
2241
new_test.transport_server = self._transport_server
2242
new_test.transport_readonly_server = self._transport_readonly_server
2243
new_test.bzrdir_format = bzrdir_format
2244
new_test.repository_format = repository_format
2245
def make_new_test_id():
2246
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2247
return lambda: new_id
2248
new_test.id = make_new_test_id()
2249
result.addTest(new_test)
2250
return result
2251
2252
2253
class InterRepositoryTestProviderAdapter(object):
2254
"""A tool to generate a suite testing multiple inter repository formats.
2255
2256
This is done by copying the test once for each interrepo provider and injecting
2257
the transport_server, transport_readonly_server, repository_format and
2258
repository_to_format classes into each copy.
2259
Each copy is also given a new id() to make it easy to identify.
2260
"""
2261
2262
def __init__(self, transport_server, transport_readonly_server, formats):
2263
self._transport_server = transport_server
2264
self._transport_readonly_server = transport_readonly_server
2265
self._formats = formats
2266
2267
def adapt(self, test):
2268
result = unittest.TestSuite()
2269
for interrepo_class, repository_format, repository_format_to in self._formats:
2270
new_test = deepcopy(test)
2271
new_test.transport_server = self._transport_server
2272
new_test.transport_readonly_server = self._transport_readonly_server
2273
new_test.interrepo_class = interrepo_class
2274
new_test.repository_format = repository_format
2275
new_test.repository_format_to = repository_format_to
2276
def make_new_test_id():
2277
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2278
return lambda: new_id
2279
new_test.id = make_new_test_id()
2280
result.addTest(new_test)
2281
return result
2282
2283
@staticmethod
2284
def default_test_list():
2285
"""Generate the default list of interrepo permutations to test."""
2286
result = []
2287
# test the default InterRepository between format 6 and the current
2288
# default format.
2289
# XXX: robertc 20060220 reinstate this when there are two supported
2290
# formats which do not have an optimal code path between them.
2291
#result.append((InterRepository,
2292
# RepositoryFormat6(),
2293
# RepositoryFormatKnit1()))
2294
for optimiser in InterRepository._optimisers:
2295
if optimiser._matching_repo_format is not None:
2296
result.append((optimiser,
2297
optimiser._matching_repo_format,
2298
optimiser._matching_repo_format
2299
))
2300
# if there are specific combinations we want to use, we can add them
2301
# here.
2302
result.append((InterModel1and2, RepositoryFormat5(),
2303
RepositoryFormatKnit2()))
2304
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2305
RepositoryFormatKnit2()))
2306
return result
2307
2308
2309
class CopyConverter(object):
2310
"""A repository conversion tool which just performs a copy of the content.
2311
2312
This is slow but quite reliable.
2313
"""
2314
2315
def __init__(self, target_format):
2316
"""Create a CopyConverter.
2317
2318
:param target_format: The format the resulting repository should be.
2319
"""
2320
self.target_format = target_format
2321
2322
def convert(self, repo, pb):
2323
"""Perform the conversion of to_convert, giving feedback via pb.
2324
2325
:param to_convert: The disk object to convert.
2326
:param pb: a progress bar to use for progress information.
2327
"""
2328
self.pb = pb
2329
self.count = 0
2330
self.total = 4
2331
# this is only useful with metadir layouts - separated repo content.
2332
# trigger an assertion if not such
2333
repo._format.get_format_string()
2334
self.repo_dir = repo.bzrdir
2335
self.step('Moving repository to repository.backup')
2336
self.repo_dir.transport.move('repository', 'repository.backup')
2337
backup_transport = self.repo_dir.transport.clone('repository.backup')
2338
repo._format.check_conversion_target(self.target_format)
2339
self.source_repo = repo._format.open(self.repo_dir,
2340
_found=True,
2341
_override_transport=backup_transport)
2342
self.step('Creating new repository')
2343
converted = self.target_format.initialize(self.repo_dir,
2344
self.source_repo.is_shared())
2345
converted.lock_write()
2346
try:
2347
self.step('Copying content into repository.')
2348
self.source_repo.copy_content_into(converted)
2349
finally:
2350
converted.unlock()
2351
self.step('Deleting old repository content.')
2352
self.repo_dir.transport.delete_tree('repository.backup')
2353
self.pb.note('repository converted')
2354
2355
def step(self, message):
2356
"""Update the pb by a step."""
2357
self.count +=1
2358
self.pb.update(message, self.count, self.total)
2359
2360
2361
class CommitBuilder(object):
2362
"""Provides an interface to build up a commit.
2363
2364
This allows describing a tree to be committed without needing to
2365
know the internals of the format of the repository.
2366
"""
2367
2368
record_root_entry = False
2369
def __init__(self, repository, parents, config, timestamp=None,
2370
timezone=None, committer=None, revprops=None,
2371
revision_id=None):
2372
"""Initiate a CommitBuilder.
2373
2374
:param repository: Repository to commit to.
2375
:param parents: Revision ids of the parents of the new revision.
2376
:param config: Configuration to use.
2377
:param timestamp: Optional timestamp recorded for commit.
2378
:param timezone: Optional timezone for timestamp.
2379
:param committer: Optional committer to set for commit.
2380
:param revprops: Optional dictionary of revision properties.
2381
:param revision_id: Optional revision id.
2382
"""
2383
self._config = config
2384
2385
if committer is None:
2386
self._committer = self._config.username()
2387
else:
2388
assert isinstance(committer, basestring), type(committer)
2389
self._committer = committer
2390
2391
self.new_inventory = Inventory(None)
2392
self._new_revision_id = revision_id
2393
self.parents = parents
2394
self.repository = repository
2395
2396
self._revprops = {}
2397
if revprops is not None:
2398
self._revprops.update(revprops)
2399
2400
if timestamp is None:
2401
timestamp = time.time()
2402
# Restrict resolution to 1ms
2403
self._timestamp = round(timestamp, 3)
2404
2405
if timezone is None:
2406
self._timezone = local_time_offset()
2407
else:
2408
self._timezone = int(timezone)
2409
2410
self._generate_revision_if_needed()
2411
2412
def commit(self, message):
2413
"""Make the actual commit.
2414
2415
:return: The revision id of the recorded revision.
2416
"""
2417
rev = _mod_revision.Revision(
2418
timestamp=self._timestamp,
2419
timezone=self._timezone,
2420
committer=self._committer,
2421
message=message,
2422
inventory_sha1=self.inv_sha1,
2423
revision_id=self._new_revision_id,
2424
properties=self._revprops)
2425
rev.parent_ids = self.parents
2426
self.repository.add_revision(self._new_revision_id, rev,
2427
self.new_inventory, self._config)
2428
return self._new_revision_id
2429
2430
def revision_tree(self):
2431
"""Return the tree that was just committed.
2432
2433
After calling commit() this can be called to get a RevisionTree
2434
representing the newly committed tree. This is preferred to
2435
calling Repository.revision_tree() because that may require
2436
deserializing the inventory, while we already have a copy in
2437
memory.
2438
"""
2439
return RevisionTree(self.repository, self.new_inventory,
2440
self._new_revision_id)
2441
2442
def finish_inventory(self):
2443
"""Tell the builder that the inventory is finished."""
2444
if self.new_inventory.root is None:
2445
symbol_versioning.warn('Root entry should be supplied to'
2446
' record_entry_contents, as of bzr 0.10.',
2447
DeprecationWarning, stacklevel=2)
2448
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2449
self.new_inventory.revision_id = self._new_revision_id
2450
self.inv_sha1 = self.repository.add_inventory(
2451
self._new_revision_id,
2452
self.new_inventory,
2453
self.parents
2454
)
2455
2456
def _gen_revision_id(self):
2457
"""Return new revision-id."""
2458
return generate_ids.gen_revision_id(self._config.username(),
2459
self._timestamp)
2460
2461
def _generate_revision_if_needed(self):
2462
"""Create a revision id if None was supplied.
2463
2464
If the repository can not support user-specified revision ids
2465
they should override this function and raise CannotSetRevisionId
2466
if _new_revision_id is not None.
2467
2468
:raises: CannotSetRevisionId
2469
"""
2470
if self._new_revision_id is None:
2471
self._new_revision_id = self._gen_revision_id()
2472
2473
def record_entry_contents(self, ie, parent_invs, path, tree):
2474
"""Record the content of ie from tree into the commit if needed.
2475
2476
Side effect: sets ie.revision when unchanged
2477
2478
:param ie: An inventory entry present in the commit.
2479
:param parent_invs: The inventories of the parent revisions of the
2480
commit.
2481
:param path: The path the entry is at in the tree.
2482
:param tree: The tree which contains this entry and should be used to
2483
obtain content.
2484
"""
2485
if self.new_inventory.root is None and ie.parent_id is not None:
2486
symbol_versioning.warn('Root entry should be supplied to'
2487
' record_entry_contents, as of bzr 0.10.',
2488
DeprecationWarning, stacklevel=2)
2489
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2490
'', tree)
2491
self.new_inventory.add(ie)
2492
2493
# ie.revision is always None if the InventoryEntry is considered
2494
# for committing. ie.snapshot will record the correct revision
2495
# which may be the sole parent if it is untouched.
2496
if ie.revision is not None:
2497
return
2498
2499
# In this revision format, root entries have no knit or weave
2500
if ie is self.new_inventory.root:
2501
# When serializing out to disk and back in
2502
# root.revision is always _new_revision_id
2503
ie.revision = self._new_revision_id
2504
return
2505
previous_entries = ie.find_previous_heads(
2506
parent_invs,
2507
self.repository.weave_store,
2508
self.repository.get_transaction())
2509
# we are creating a new revision for ie in the history store
2510
# and inventory.
2511
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2512
2513
def modified_directory(self, file_id, file_parents):
2514
"""Record the presence of a symbolic link.
2515
2516
:param file_id: The file_id of the link to record.
2517
:param file_parents: The per-file parent revision ids.
2518
"""
2519
self._add_text_to_weave(file_id, [], file_parents.keys())
2520
2521
def modified_file_text(self, file_id, file_parents,
2522
get_content_byte_lines, text_sha1=None,
2523
text_size=None):
2524
"""Record the text of file file_id
2525
2526
:param file_id: The file_id of the file to record the text of.
2527
:param file_parents: The per-file parent revision ids.
2528
:param get_content_byte_lines: A callable which will return the byte
2529
lines for the file.
2530
:param text_sha1: Optional SHA1 of the file contents.
2531
:param text_size: Optional size of the file contents.
2532
"""
2533
# mutter('storing text of file {%s} in revision {%s} into %r',
2534
# file_id, self._new_revision_id, self.repository.weave_store)
2535
# special case to avoid diffing on renames or
2536
# reparenting
2537
if (len(file_parents) == 1
2538
and text_sha1 == file_parents.values()[0].text_sha1
2539
and text_size == file_parents.values()[0].text_size):
2540
previous_ie = file_parents.values()[0]
2541
versionedfile = self.repository.weave_store.get_weave(file_id,
2542
self.repository.get_transaction())
2543
versionedfile.clone_text(self._new_revision_id,
2544
previous_ie.revision, file_parents.keys())
2545
return text_sha1, text_size
2546
else:
2547
new_lines = get_content_byte_lines()
2548
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2549
# should return the SHA1 and size
2550
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2551
return osutils.sha_strings(new_lines), \
2552
sum(map(len, new_lines))
2553
2554
def modified_link(self, file_id, file_parents, link_target):
2555
"""Record the presence of a symbolic link.
2556
2557
:param file_id: The file_id of the link to record.
2558
:param file_parents: The per-file parent revision ids.
2559
:param link_target: Target location of this link.
2560
"""
2561
self._add_text_to_weave(file_id, [], file_parents.keys())
2562
2563
def _add_text_to_weave(self, file_id, new_lines, parents):
2564
versionedfile = self.repository.weave_store.get_weave_or_empty(
2565
file_id, self.repository.get_transaction())
2566
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2567
versionedfile.clear_cache()
2568
2569
2570
class _CommitBuilder(CommitBuilder):
2571
"""Temporary class so old CommitBuilders are detected properly
2572
2573
Note: CommitBuilder works whether or not root entry is recorded.
2574
"""
2575
2576
record_root_entry = True
2577
2578
2579
class RootCommitBuilder(CommitBuilder):
2580
"""This commitbuilder actually records the root id"""
2581
2582
record_root_entry = True
2583
2584
def record_entry_contents(self, ie, parent_invs, path, tree):
2585
"""Record the content of ie from tree into the commit if needed.
2586
2587
Side effect: sets ie.revision when unchanged
2588
2589
:param ie: An inventory entry present in the commit.
2590
:param parent_invs: The inventories of the parent revisions of the
2591
commit.
2592
:param path: The path the entry is at in the tree.
2593
:param tree: The tree which contains this entry and should be used to
2594
obtain content.
2595
"""
2596
assert self.new_inventory.root is not None or ie.parent_id is None
2597
self.new_inventory.add(ie)
2598
2599
# ie.revision is always None if the InventoryEntry is considered
2600
# for committing. ie.snapshot will record the correct revision
2601
# which may be the sole parent if it is untouched.
2602
if ie.revision is not None:
2603
return
2604
2605
previous_entries = ie.find_previous_heads(
2606
parent_invs,
2607
self.repository.weave_store,
2608
self.repository.get_transaction())
2609
# we are creating a new revision for ie in the history store
2610
# and inventory.
2611
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2612
2613
2614
_unescape_map = {
2615
'apos':"'",
2616
'quot':'"',
2617
'amp':'&',
2618
'lt':'<',
2619
'gt':'>'
2620
}
2621
2622
2623
def _unescaper(match, _map=_unescape_map):
2624
return _map[match.group(1)]
2625
2626
2627
_unescape_re = None
2628
2629
2630
def _unescape_xml(data):
2631
"""Unescape predefined XML entities in a string of data."""
2632
global _unescape_re
2633
if _unescape_re is None:
2634
_unescape_re = re.compile('\&([^;]*);')
2635
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0