(root)/brz/remove-bazaar
» Revision
2230.2.6
(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 2230.2.6
- compare with another revision
- download diff
- download tarball
- view history from revision 2230.2.6
- Committer: John Arbash Meinel
- Date: 2007-01-25 15:54:29 UTC
- mto: This revision was merged to the branch mainline in revision 2245.
- Revision ID: john@arbash-meinel.com-20070125155429-x53892xqx4r17l8h
Clean up the documentation and imports for decorators (per Martin's suggestions)
- files added:
- .bzrignore
- bzrlib
- bzrlib/benchmarks
- bzrlib/benchmarks/tree_creator
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/doc
- bzrlib/doc/api
- bzrlib/export
- bzrlib/plugins
- bzrlib/plugins/launchpad
- bzrlib/store
- bzrlib/store/revision
- bzrlib/store/versioned
- bzrlib/tests
- bzrlib/tests/blackbox
- bzrlib/tests/branch_implementations
- bzrlib/tests/bzrdir_implementations
- bzrlib/tests/interrepository_implementations
- bzrlib/tests/intertree_implementations
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/repository_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/tests/test_patches_data
- bzrlib/tests/tree_implementations
- bzrlib/tests/workingtree_implementations
- bzrlib/transport
- bzrlib/transport/http
- bzrlib/ui
- bzrlib/util
- bzrlib/util/configobj
- bzrlib/util/configobj/docs
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- bzrlib/util/elementtree
- bzrlib/version_info_formats
- contrib
- contrib/bash
- contrib/emacs
- contrib/zsh
- doc
- tools
- tools/doc_generate
- tools/win32
- files removed:
- .bzrignore
- notes
- tests
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
#
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
7
#
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
12
#
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
17
from cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
23
import re
24
import time
25
import unittest
26
27
from bzrlib import (
28
bzrdir,
29
check,
30
delta,
31
errors,
32
generate_ids,
33
gpg,
34
graph,
35
knit,
36
lazy_regex,
37
lockable_files,
38
lockdir,
39
osutils,
40
revision as _mod_revision,
41
symbol_versioning,
42
transactions,
43
ui,
44
weave,
45
weavefile,
46
xml5,
47
xml6,
48
)
49
from bzrlib.osutils import (
50
rand_bytes,
51
compact_date,
52
local_time_offset,
53
)
54
from bzrlib.revisiontree import RevisionTree
55
from bzrlib.store.versioned import VersionedFileStore
56
from bzrlib.store.text import TextStore
57
from bzrlib.testament import Testament
58
""")
59
60
from bzrlib.decorators import needs_read_lock, needs_write_lock
61
from bzrlib.inter import InterObject
62
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
63
from bzrlib.symbol_versioning import (
64
deprecated_method,
65
zero_nine,
66
)
67
from bzrlib.trace import mutter, note, warning
68
69
70
# Old formats display a warning, but only once
71
_deprecation_warning_done = False
72
73
74
class Repository(object):
75
"""Repository holding history for one or more branches.
76
77
The repository holds and retrieves historical information including
78
revisions and file history. It's normally accessed only by the Branch,
79
which views a particular line of development through that history.
80
81
The Repository builds on top of Stores and a Transport, which respectively
82
describe the disk data format and the way of accessing the (possibly
83
remote) disk.
84
"""
85
86
_file_ids_altered_regex = lazy_regex.lazy_compile(
87
r'file_id="(?P<file_id>[^"]+)"'
88
r'.*revision="(?P<revision_id>[^"]+)"'
89
)
90
91
@needs_write_lock
92
def add_inventory(self, revid, inv, parents):
93
"""Add the inventory inv to the repository as revid.
94
95
:param parents: The revision ids of the parents that revid
96
is known to have and are in the repository already.
97
98
returns the sha1 of the serialized inventory.
99
"""
100
_mod_revision.check_not_reserved_id(revid)
101
assert inv.revision_id is None or inv.revision_id == revid, \
102
"Mismatch between inventory revision" \
103
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
104
assert inv.root is not None
105
inv_text = self.serialise_inventory(inv)
106
inv_sha1 = osutils.sha_string(inv_text)
107
inv_vf = self.control_weaves.get_weave('inventory',
108
self.get_transaction())
109
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
110
return inv_sha1
111
112
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
113
final_parents = []
114
for parent in parents:
115
if parent in inv_vf:
116
final_parents.append(parent)
117
118
inv_vf.add_lines(revid, final_parents, lines)
119
120
@needs_write_lock
121
def add_revision(self, rev_id, rev, inv=None, config=None):
122
"""Add rev to the revision store as rev_id.
123
124
:param rev_id: the revision id to use.
125
:param rev: The revision object.
126
:param inv: The inventory for the revision. if None, it will be looked
127
up in the inventory storer
128
:param config: If None no digital signature will be created.
129
If supplied its signature_needed method will be used
130
to determine if a signature should be made.
131
"""
132
_mod_revision.check_not_reserved_id(rev_id)
133
if config is not None and config.signature_needed():
134
if inv is None:
135
inv = self.get_inventory(rev_id)
136
plaintext = Testament(rev, inv).as_short_text()
137
self.store_revision_signature(
138
gpg.GPGStrategy(config), plaintext, rev_id)
139
if not rev_id in self.get_inventory_weave():
140
if inv is None:
141
raise errors.WeaveRevisionNotPresent(rev_id,
142
self.get_inventory_weave())
143
else:
144
# yes, this is not suitable for adding with ghosts.
145
self.add_inventory(rev_id, inv, rev.parent_ids)
146
self._revision_store.add_revision(rev, self.get_transaction())
147
148
@needs_read_lock
149
def _all_possible_ids(self):
150
"""Return all the possible revisions that we could find."""
151
return self.get_inventory_weave().versions()
152
153
def all_revision_ids(self):
154
"""Returns a list of all the revision ids in the repository.
155
156
This is deprecated because code should generally work on the graph
157
reachable from a particular revision, and ignore any other revisions
158
that might be present. There is no direct replacement method.
159
"""
160
return self._all_revision_ids()
161
162
@needs_read_lock
163
def _all_revision_ids(self):
164
"""Returns a list of all the revision ids in the repository.
165
166
These are in as much topological order as the underlying store can
167
present: for weaves ghosts may lead to a lack of correctness until
168
the reweave updates the parents list.
169
"""
170
if self._revision_store.text_store.listable():
171
return self._revision_store.all_revision_ids(self.get_transaction())
172
result = self._all_possible_ids()
173
return self._eliminate_revisions_not_present(result)
174
175
def break_lock(self):
176
"""Break a lock if one is present from another instance.
177
178
Uses the ui factory to ask for confirmation if the lock may be from
179
an active process.
180
"""
181
self.control_files.break_lock()
182
183
@needs_read_lock
184
def _eliminate_revisions_not_present(self, revision_ids):
185
"""Check every revision id in revision_ids to see if we have it.
186
187
Returns a set of the present revisions.
188
"""
189
result = []
190
for id in revision_ids:
191
if self.has_revision(id):
192
result.append(id)
193
return result
194
195
@staticmethod
196
def create(a_bzrdir):
197
"""Construct the current default format repository in a_bzrdir."""
198
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
199
200
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
201
"""instantiate a Repository.
202
203
:param _format: The format of the repository on disk.
204
:param a_bzrdir: The BzrDir of the repository.
205
206
In the future we will have a single api for all stores for
207
getting file texts, inventories and revisions, then
208
this construct will accept instances of those things.
209
"""
210
super(Repository, self).__init__()
211
self._format = _format
212
# the following are part of the public API for Repository:
213
self.bzrdir = a_bzrdir
214
self.control_files = control_files
215
self._revision_store = _revision_store
216
self.text_store = text_store
217
# backwards compatibility
218
self.weave_store = text_store
219
# not right yet - should be more semantically clear ?
220
#
221
self.control_store = control_store
222
self.control_weaves = control_store
223
# TODO: make sure to construct the right store classes, etc, depending
224
# on whether escaping is required.
225
self._warn_if_deprecated()
226
self._serializer = xml5.serializer_v5
227
228
def __repr__(self):
229
return '%s(%r)' % (self.__class__.__name__,
230
self.bzrdir.transport.base)
231
232
def is_locked(self):
233
return self.control_files.is_locked()
234
235
def lock_write(self):
236
self.control_files.lock_write()
237
238
def lock_read(self):
239
self.control_files.lock_read()
240
241
def get_physical_lock_status(self):
242
return self.control_files.get_physical_lock_status()
243
244
@needs_read_lock
245
def missing_revision_ids(self, other, revision_id=None):
246
"""Return the revision ids that other has that this does not.
247
248
These are returned in topological order.
249
250
revision_id: only return revision ids included by revision_id.
251
"""
252
return InterRepository.get(other, self).missing_revision_ids(revision_id)
253
254
@staticmethod
255
def open(base):
256
"""Open the repository rooted at base.
257
258
For instance, if the repository is at URL/.bzr/repository,
259
Repository.open(URL) -> a Repository instance.
260
"""
261
control = bzrdir.BzrDir.open(base)
262
return control.open_repository()
263
264
def copy_content_into(self, destination, revision_id=None, basis=None):
265
"""Make a complete copy of the content in self into destination.
266
267
This is a destructive operation! Do not use it on existing
268
repositories.
269
"""
270
return InterRepository.get(self, destination).copy_content(revision_id, basis)
271
272
def fetch(self, source, revision_id=None, pb=None):
273
"""Fetch the content required to construct revision_id from source.
274
275
If revision_id is None all content is copied.
276
"""
277
return InterRepository.get(source, self).fetch(revision_id=revision_id,
278
pb=pb)
279
280
def get_commit_builder(self, branch, parents, config, timestamp=None,
281
timezone=None, committer=None, revprops=None,
282
revision_id=None):
283
"""Obtain a CommitBuilder for this repository.
284
285
:param branch: Branch to commit to.
286
:param parents: Revision ids of the parents of the new revision.
287
:param config: Configuration to use.
288
:param timestamp: Optional timestamp recorded for commit.
289
:param timezone: Optional timezone for timestamp.
290
:param committer: Optional committer to set for commit.
291
:param revprops: Optional dictionary of revision properties.
292
:param revision_id: Optional revision id.
293
"""
294
return _CommitBuilder(self, parents, config, timestamp, timezone,
295
committer, revprops, revision_id)
296
297
def unlock(self):
298
self.control_files.unlock()
299
300
@needs_read_lock
301
def clone(self, a_bzrdir, revision_id=None, basis=None):
302
"""Clone this repository into a_bzrdir using the current format.
303
304
Currently no check is made that the format of this repository and
305
the bzrdir format are compatible. FIXME RBC 20060201.
306
"""
307
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
308
# use target default format.
309
result = a_bzrdir.create_repository()
310
# FIXME RBC 20060209 split out the repository type to avoid this check ?
311
elif isinstance(a_bzrdir._format,
312
(bzrdir.BzrDirFormat4,
313
bzrdir.BzrDirFormat5,
314
bzrdir.BzrDirFormat6)):
315
result = a_bzrdir.open_repository()
316
else:
317
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
318
self.copy_content_into(result, revision_id, basis)
319
return result
320
321
@needs_read_lock
322
def has_revision(self, revision_id):
323
"""True if this repository has a copy of the revision."""
324
return self._revision_store.has_revision_id(revision_id,
325
self.get_transaction())
326
327
@needs_read_lock
328
def get_revision_reconcile(self, revision_id):
329
"""'reconcile' helper routine that allows access to a revision always.
330
331
This variant of get_revision does not cross check the weave graph
332
against the revision one as get_revision does: but it should only
333
be used by reconcile, or reconcile-alike commands that are correcting
334
or testing the revision graph.
335
"""
336
if not revision_id or not isinstance(revision_id, basestring):
337
raise errors.InvalidRevisionId(revision_id=revision_id,
338
branch=self)
339
return self._revision_store.get_revisions([revision_id],
340
self.get_transaction())[0]
341
@needs_read_lock
342
def get_revisions(self, revision_ids):
343
return self._revision_store.get_revisions(revision_ids,
344
self.get_transaction())
345
346
@needs_read_lock
347
def get_revision_xml(self, revision_id):
348
rev = self.get_revision(revision_id)
349
rev_tmp = StringIO()
350
# the current serializer..
351
self._revision_store._serializer.write_revision(rev, rev_tmp)
352
rev_tmp.seek(0)
353
return rev_tmp.getvalue()
354
355
@needs_read_lock
356
def get_revision(self, revision_id):
357
"""Return the Revision object for a named revision"""
358
r = self.get_revision_reconcile(revision_id)
359
# weave corruption can lead to absent revision markers that should be
360
# present.
361
# the following test is reasonably cheap (it needs a single weave read)
362
# and the weave is cached in read transactions. In write transactions
363
# it is not cached but typically we only read a small number of
364
# revisions. For knits when they are introduced we will probably want
365
# to ensure that caching write transactions are in use.
366
inv = self.get_inventory_weave()
367
self._check_revision_parents(r, inv)
368
return r
369
370
@needs_read_lock
371
def get_deltas_for_revisions(self, revisions):
372
"""Produce a generator of revision deltas.
373
374
Note that the input is a sequence of REVISIONS, not revision_ids.
375
Trees will be held in memory until the generator exits.
376
Each delta is relative to the revision's lefthand predecessor.
377
"""
378
required_trees = set()
379
for revision in revisions:
380
required_trees.add(revision.revision_id)
381
required_trees.update(revision.parent_ids[:1])
382
trees = dict((t.get_revision_id(), t) for
383
t in self.revision_trees(required_trees))
384
for revision in revisions:
385
if not revision.parent_ids:
386
old_tree = self.revision_tree(None)
387
else:
388
old_tree = trees[revision.parent_ids[0]]
389
yield trees[revision.revision_id].changes_from(old_tree)
390
391
@needs_read_lock
392
def get_revision_delta(self, revision_id):
393
"""Return the delta for one revision.
394
395
The delta is relative to the left-hand predecessor of the
396
revision.
397
"""
398
r = self.get_revision(revision_id)
399
return list(self.get_deltas_for_revisions([r]))[0]
400
401
def _check_revision_parents(self, revision, inventory):
402
"""Private to Repository and Fetch.
403
404
This checks the parentage of revision in an inventory weave for
405
consistency and is only applicable to inventory-weave-for-ancestry
406
using repository formats & fetchers.
407
"""
408
weave_parents = inventory.get_parents(revision.revision_id)
409
weave_names = inventory.versions()
410
for parent_id in revision.parent_ids:
411
if parent_id in weave_names:
412
# this parent must not be a ghost.
413
if not parent_id in weave_parents:
414
# but it is a ghost
415
raise errors.CorruptRepository(self)
416
417
@needs_write_lock
418
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
419
signature = gpg_strategy.sign(plaintext)
420
self._revision_store.add_revision_signature_text(revision_id,
421
signature,
422
self.get_transaction())
423
424
def fileids_altered_by_revision_ids(self, revision_ids):
425
"""Find the file ids and versions affected by revisions.
426
427
:param revisions: an iterable containing revision ids.
428
:return: a dictionary mapping altered file-ids to an iterable of
429
revision_ids. Each altered file-ids has the exact revision_ids that
430
altered it listed explicitly.
431
"""
432
assert self._serializer.support_altered_by_hack, \
433
("fileids_altered_by_revision_ids only supported for branches "
434
"which store inventory as unnested xml, not on %r" % self)
435
selected_revision_ids = set(revision_ids)
436
w = self.get_inventory_weave()
437
result = {}
438
439
# this code needs to read every new line in every inventory for the
440
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
441
# not present in one of those inventories is unnecessary but not
442
# harmful because we are filtering by the revision id marker in the
443
# inventory lines : we only select file ids altered in one of those
444
# revisions. We don't need to see all lines in the inventory because
445
# only those added in an inventory in rev X can contain a revision=X
446
# line.
447
unescape_revid_cache = {}
448
unescape_fileid_cache = {}
449
450
# jam 20061218 In a big fetch, this handles hundreds of thousands
451
# of lines, so it has had a lot of inlining and optimizing done.
452
# Sorry that it is a little bit messy.
453
# Move several functions to be local variables, since this is a long
454
# running loop.
455
search = self._file_ids_altered_regex.search
456
unescape = _unescape_xml
457
setdefault = result.setdefault
458
pb = ui.ui_factory.nested_progress_bar()
459
try:
460
for line in w.iter_lines_added_or_present_in_versions(
461
selected_revision_ids, pb=pb):
462
match = search(line)
463
if match is None:
464
continue
465
# One call to match.group() returning multiple items is quite a
466
# bit faster than 2 calls to match.group() each returning 1
467
file_id, revision_id = match.group('file_id', 'revision_id')
468
469
# Inlining the cache lookups helps a lot when you make 170,000
470
# lines and 350k ids, versus 8.4 unique ids.
471
# Using a cache helps in 2 ways:
472
# 1) Avoids unnecessary decoding calls
473
# 2) Re-uses cached strings, which helps in future set and
474
# equality checks.
475
# (2) is enough that removing encoding entirely along with
476
# the cache (so we are using plain strings) results in no
477
# performance improvement.
478
try:
479
revision_id = unescape_revid_cache[revision_id]
480
except KeyError:
481
unescaped = unescape(revision_id)
482
unescape_revid_cache[revision_id] = unescaped
483
revision_id = unescaped
484
485
if revision_id in selected_revision_ids:
486
try:
487
file_id = unescape_fileid_cache[file_id]
488
except KeyError:
489
unescaped = unescape(file_id)
490
unescape_fileid_cache[file_id] = unescaped
491
file_id = unescaped
492
setdefault(file_id, set()).add(revision_id)
493
finally:
494
pb.finished()
495
return result
496
497
@needs_read_lock
498
def get_inventory_weave(self):
499
return self.control_weaves.get_weave('inventory',
500
self.get_transaction())
501
502
@needs_read_lock
503
def get_inventory(self, revision_id):
504
"""Get Inventory object by hash."""
505
return self.deserialise_inventory(
506
revision_id, self.get_inventory_xml(revision_id))
507
508
def deserialise_inventory(self, revision_id, xml):
509
"""Transform the xml into an inventory object.
510
511
:param revision_id: The expected revision id of the inventory.
512
:param xml: A serialised inventory.
513
"""
514
result = self._serializer.read_inventory_from_string(xml)
515
result.root.revision = revision_id
516
return result
517
518
def serialise_inventory(self, inv):
519
return self._serializer.write_inventory_to_string(inv)
520
521
@needs_read_lock
522
def get_inventory_xml(self, revision_id):
523
"""Get inventory XML as a file object."""
524
try:
525
assert isinstance(revision_id, basestring), type(revision_id)
526
iw = self.get_inventory_weave()
527
return iw.get_text(revision_id)
528
except IndexError:
529
raise errors.HistoryMissing(self, 'inventory', revision_id)
530
531
@needs_read_lock
532
def get_inventory_sha1(self, revision_id):
533
"""Return the sha1 hash of the inventory entry
534
"""
535
return self.get_revision(revision_id).inventory_sha1
536
537
@needs_read_lock
538
def get_revision_graph(self, revision_id=None):
539
"""Return a dictionary containing the revision graph.
540
541
:param revision_id: The revision_id to get a graph from. If None, then
542
the entire revision graph is returned. This is a deprecated mode of
543
operation and will be removed in the future.
544
:return: a dictionary of revision_id->revision_parents_list.
545
"""
546
# special case NULL_REVISION
547
if revision_id == _mod_revision.NULL_REVISION:
548
return {}
549
a_weave = self.get_inventory_weave()
550
all_revisions = self._eliminate_revisions_not_present(
551
a_weave.versions())
552
entire_graph = dict([(node, a_weave.get_parents(node)) for
553
node in all_revisions])
554
if revision_id is None:
555
return entire_graph
556
elif revision_id not in entire_graph:
557
raise errors.NoSuchRevision(self, revision_id)
558
else:
559
# add what can be reached from revision_id
560
result = {}
561
pending = set([revision_id])
562
while len(pending) > 0:
563
node = pending.pop()
564
result[node] = entire_graph[node]
565
for revision_id in result[node]:
566
if revision_id not in result:
567
pending.add(revision_id)
568
return result
569
570
@needs_read_lock
571
def get_revision_graph_with_ghosts(self, revision_ids=None):
572
"""Return a graph of the revisions with ghosts marked as applicable.
573
574
:param revision_ids: an iterable of revisions to graph or None for all.
575
:return: a Graph object with the graph reachable from revision_ids.
576
"""
577
result = graph.Graph()
578
if not revision_ids:
579
pending = set(self.all_revision_ids())
580
required = set([])
581
else:
582
pending = set(revision_ids)
583
# special case NULL_REVISION
584
if _mod_revision.NULL_REVISION in pending:
585
pending.remove(_mod_revision.NULL_REVISION)
586
required = set(pending)
587
done = set([])
588
while len(pending):
589
revision_id = pending.pop()
590
try:
591
rev = self.get_revision(revision_id)
592
except errors.NoSuchRevision:
593
if revision_id in required:
594
raise
595
# a ghost
596
result.add_ghost(revision_id)
597
continue
598
for parent_id in rev.parent_ids:
599
# is this queued or done ?
600
if (parent_id not in pending and
601
parent_id not in done):
602
# no, queue it.
603
pending.add(parent_id)
604
result.add_node(revision_id, rev.parent_ids)
605
done.add(revision_id)
606
return result
607
608
@needs_read_lock
609
def get_revision_inventory(self, revision_id):
610
"""Return inventory of a past revision."""
611
# TODO: Unify this with get_inventory()
612
# bzr 0.0.6 and later imposes the constraint that the inventory_id
613
# must be the same as its revision, so this is trivial.
614
if revision_id is None:
615
# This does not make sense: if there is no revision,
616
# then it is the current tree inventory surely ?!
617
# and thus get_root_id() is something that looks at the last
618
# commit on the branch, and the get_root_id is an inventory check.
619
raise NotImplementedError
620
# return Inventory(self.get_root_id())
621
else:
622
return self.get_inventory(revision_id)
623
624
@needs_read_lock
625
def is_shared(self):
626
"""Return True if this repository is flagged as a shared repository."""
627
raise NotImplementedError(self.is_shared)
628
629
@needs_write_lock
630
def reconcile(self, other=None, thorough=False):
631
"""Reconcile this repository."""
632
from bzrlib.reconcile import RepoReconciler
633
reconciler = RepoReconciler(self, thorough=thorough)
634
reconciler.reconcile()
635
return reconciler
636
637
@needs_read_lock
638
def revision_tree(self, revision_id):
639
"""Return Tree for a revision on this branch.
640
641
`revision_id` may be None for the empty tree revision.
642
"""
643
# TODO: refactor this to use an existing revision object
644
# so we don't need to read it in twice.
645
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
646
return RevisionTree(self, Inventory(root_id=None),
647
_mod_revision.NULL_REVISION)
648
else:
649
inv = self.get_revision_inventory(revision_id)
650
return RevisionTree(self, inv, revision_id)
651
652
@needs_read_lock
653
def revision_trees(self, revision_ids):
654
"""Return Tree for a revision on this branch.
655
656
`revision_id` may not be None or 'null:'"""
657
assert None not in revision_ids
658
assert _mod_revision.NULL_REVISION not in revision_ids
659
texts = self.get_inventory_weave().get_texts(revision_ids)
660
for text, revision_id in zip(texts, revision_ids):
661
inv = self.deserialise_inventory(revision_id, text)
662
yield RevisionTree(self, inv, revision_id)
663
664
@needs_read_lock
665
def get_ancestry(self, revision_id):
666
"""Return a list of revision-ids integrated by a revision.
667
668
The first element of the list is always None, indicating the origin
669
revision. This might change when we have history horizons, or
670
perhaps we should have a new API.
671
672
This is topologically sorted.
673
"""
674
if revision_id is None:
675
return [None]
676
if not self.has_revision(revision_id):
677
raise errors.NoSuchRevision(self, revision_id)
678
w = self.get_inventory_weave()
679
candidates = w.get_ancestry(revision_id)
680
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
681
682
@needs_read_lock
683
def print_file(self, file, revision_id):
684
"""Print `file` to stdout.
685
686
FIXME RBC 20060125 as John Meinel points out this is a bad api
687
- it writes to stdout, it assumes that that is valid etc. Fix
688
by creating a new more flexible convenience function.
689
"""
690
tree = self.revision_tree(revision_id)
691
# use inventory as it was in that revision
692
file_id = tree.inventory.path2id(file)
693
if not file_id:
694
# TODO: jam 20060427 Write a test for this code path
695
# it had a bug in it, and was raising the wrong
696
# exception.
697
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
698
tree.print_file(file_id)
699
700
def get_transaction(self):
701
return self.control_files.get_transaction()
702
703
def revision_parents(self, revid):
704
return self.get_inventory_weave().parent_names(revid)
705
706
@needs_write_lock
707
def set_make_working_trees(self, new_value):
708
"""Set the policy flag for making working trees when creating branches.
709
710
This only applies to branches that use this repository.
711
712
The default is 'True'.
713
:param new_value: True to restore the default, False to disable making
714
working trees.
715
"""
716
raise NotImplementedError(self.set_make_working_trees)
717
718
def make_working_trees(self):
719
"""Returns the policy for making working trees on new branches."""
720
raise NotImplementedError(self.make_working_trees)
721
722
@needs_write_lock
723
def sign_revision(self, revision_id, gpg_strategy):
724
plaintext = Testament.from_revision(self, revision_id).as_short_text()
725
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
726
727
@needs_read_lock
728
def has_signature_for_revision_id(self, revision_id):
729
"""Query for a revision signature for revision_id in the repository."""
730
return self._revision_store.has_signature(revision_id,
731
self.get_transaction())
732
733
@needs_read_lock
734
def get_signature_text(self, revision_id):
735
"""Return the text for a signature."""
736
return self._revision_store.get_signature_text(revision_id,
737
self.get_transaction())
738
739
@needs_read_lock
740
def check(self, revision_ids):
741
"""Check consistency of all history of given revision_ids.
742
743
Different repository implementations should override _check().
744
745
:param revision_ids: A non-empty list of revision_ids whose ancestry
746
will be checked. Typically the last revision_id of a branch.
747
"""
748
if not revision_ids:
749
raise ValueError("revision_ids must be non-empty in %s.check"
750
% (self,))
751
return self._check(revision_ids)
752
753
def _check(self, revision_ids):
754
result = check.Check(self)
755
result.check()
756
return result
757
758
def _warn_if_deprecated(self):
759
global _deprecation_warning_done
760
if _deprecation_warning_done:
761
return
762
_deprecation_warning_done = True
763
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
764
% (self._format, self.bzrdir.transport.base))
765
766
def supports_rich_root(self):
767
return self._format.rich_root_data
768
769
def _check_ascii_revisionid(self, revision_id, method):
770
"""Private helper for ascii-only repositories."""
771
# weave repositories refuse to store revisionids that are non-ascii.
772
if revision_id is not None:
773
# weaves require ascii revision ids.
774
if isinstance(revision_id, unicode):
775
try:
776
revision_id.encode('ascii')
777
except UnicodeEncodeError:
778
raise errors.NonAsciiRevisionId(method, self)
779
780
781
class AllInOneRepository(Repository):
782
"""Legacy support - the repository behaviour for all-in-one branches."""
783
784
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
785
# we reuse one control files instance.
786
dir_mode = a_bzrdir._control_files._dir_mode
787
file_mode = a_bzrdir._control_files._file_mode
788
789
def get_store(name, compressed=True, prefixed=False):
790
# FIXME: This approach of assuming stores are all entirely compressed
791
# or entirely uncompressed is tidy, but breaks upgrade from
792
# some existing branches where there's a mixture; we probably
793
# still want the option to look for both.
794
relpath = a_bzrdir._control_files._escape(name)
795
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
796
prefixed=prefixed, compressed=compressed,
797
dir_mode=dir_mode,
798
file_mode=file_mode)
799
#if self._transport.should_cache():
800
# cache_path = os.path.join(self.cache_root, name)
801
# os.mkdir(cache_path)
802
# store = bzrlib.store.CachedStore(store, cache_path)
803
return store
804
805
# not broken out yet because the controlweaves|inventory_store
806
# and text_store | weave_store bits are still different.
807
if isinstance(_format, RepositoryFormat4):
808
# cannot remove these - there is still no consistent api
809
# which allows access to this old info.
810
self.inventory_store = get_store('inventory-store')
811
text_store = get_store('text-store')
812
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
813
814
def get_commit_builder(self, branch, parents, config, timestamp=None,
815
timezone=None, committer=None, revprops=None,
816
revision_id=None):
817
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
818
return Repository.get_commit_builder(self, branch, parents, config,
819
timestamp, timezone, committer, revprops, revision_id)
820
821
@needs_read_lock
822
def is_shared(self):
823
"""AllInOne repositories cannot be shared."""
824
return False
825
826
@needs_write_lock
827
def set_make_working_trees(self, new_value):
828
"""Set the policy flag for making working trees when creating branches.
829
830
This only applies to branches that use this repository.
831
832
The default is 'True'.
833
:param new_value: True to restore the default, False to disable making
834
working trees.
835
"""
836
raise NotImplementedError(self.set_make_working_trees)
837
838
def make_working_trees(self):
839
"""Returns the policy for making working trees on new branches."""
840
return True
841
842
843
def install_revision(repository, rev, revision_tree):
844
"""Install all revision data into a repository."""
845
present_parents = []
846
parent_trees = {}
847
for p_id in rev.parent_ids:
848
if repository.has_revision(p_id):
849
present_parents.append(p_id)
850
parent_trees[p_id] = repository.revision_tree(p_id)
851
else:
852
parent_trees[p_id] = repository.revision_tree(None)
853
854
inv = revision_tree.inventory
855
entries = inv.iter_entries()
856
# backwards compatability hack: skip the root id.
857
if not repository.supports_rich_root():
858
path, root = entries.next()
859
if root.revision != rev.revision_id:
860
raise errors.IncompatibleRevision(repr(repository))
861
# Add the texts that are not already present
862
for path, ie in entries:
863
w = repository.weave_store.get_weave_or_empty(ie.file_id,
864
repository.get_transaction())
865
if ie.revision not in w:
866
text_parents = []
867
# FIXME: TODO: The following loop *may* be overlapping/duplicate
868
# with InventoryEntry.find_previous_heads(). if it is, then there
869
# is a latent bug here where the parents may have ancestors of each
870
# other. RBC, AB
871
for revision, tree in parent_trees.iteritems():
872
if ie.file_id not in tree:
873
continue
874
parent_id = tree.inventory[ie.file_id].revision
875
if parent_id in text_parents:
876
continue
877
text_parents.append(parent_id)
878
879
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
880
repository.get_transaction())
881
lines = revision_tree.get_file(ie.file_id).readlines()
882
vfile.add_lines(rev.revision_id, text_parents, lines)
883
try:
884
# install the inventory
885
repository.add_inventory(rev.revision_id, inv, present_parents)
886
except errors.RevisionAlreadyPresent:
887
pass
888
repository.add_revision(rev.revision_id, rev, inv)
889
890
891
class MetaDirRepository(Repository):
892
"""Repositories in the new meta-dir layout."""
893
894
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
895
super(MetaDirRepository, self).__init__(_format,
896
a_bzrdir,
897
control_files,
898
_revision_store,
899
control_store,
900
text_store)
901
dir_mode = self.control_files._dir_mode
902
file_mode = self.control_files._file_mode
903
904
@needs_read_lock
905
def is_shared(self):
906
"""Return True if this repository is flagged as a shared repository."""
907
return self.control_files._transport.has('shared-storage')
908
909
@needs_write_lock
910
def set_make_working_trees(self, new_value):
911
"""Set the policy flag for making working trees when creating branches.
912
913
This only applies to branches that use this repository.
914
915
The default is 'True'.
916
:param new_value: True to restore the default, False to disable making
917
working trees.
918
"""
919
if new_value:
920
try:
921
self.control_files._transport.delete('no-working-trees')
922
except errors.NoSuchFile:
923
pass
924
else:
925
self.control_files.put_utf8('no-working-trees', '')
926
927
def make_working_trees(self):
928
"""Returns the policy for making working trees on new branches."""
929
return not self.control_files._transport.has('no-working-trees')
930
931
932
class WeaveMetaDirRepository(MetaDirRepository):
933
"""A subclass of MetaDirRepository to set weave specific policy."""
934
935
def get_commit_builder(self, branch, parents, config, timestamp=None,
936
timezone=None, committer=None, revprops=None,
937
revision_id=None):
938
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
939
return MetaDirRepository.get_commit_builder(self, branch, parents,
940
config, timestamp, timezone, committer, revprops, revision_id)
941
942
943
class KnitRepository(MetaDirRepository):
944
"""Knit format repository."""
945
946
def _warn_if_deprecated(self):
947
# This class isn't deprecated
948
pass
949
950
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
951
inv_vf.add_lines_with_ghosts(revid, parents, lines)
952
953
@needs_read_lock
954
def _all_revision_ids(self):
955
"""See Repository.all_revision_ids()."""
956
# Knits get the revision graph from the index of the revision knit, so
957
# it's always possible even if they're on an unlistable transport.
958
return self._revision_store.all_revision_ids(self.get_transaction())
959
960
def fileid_involved_between_revs(self, from_revid, to_revid):
961
"""Find file_id(s) which are involved in the changes between revisions.
962
963
This determines the set of revisions which are involved, and then
964
finds all file ids affected by those revisions.
965
"""
966
vf = self._get_revision_vf()
967
from_set = set(vf.get_ancestry(from_revid))
968
to_set = set(vf.get_ancestry(to_revid))
969
changed = to_set.difference(from_set)
970
return self._fileid_involved_by_set(changed)
971
972
def fileid_involved(self, last_revid=None):
973
"""Find all file_ids modified in the ancestry of last_revid.
974
975
:param last_revid: If None, last_revision() will be used.
976
"""
977
if not last_revid:
978
changed = set(self.all_revision_ids())
979
else:
980
changed = set(self.get_ancestry(last_revid))
981
if None in changed:
982
changed.remove(None)
983
return self._fileid_involved_by_set(changed)
984
985
@needs_read_lock
986
def get_ancestry(self, revision_id):
987
"""Return a list of revision-ids integrated by a revision.
988
989
This is topologically sorted.
990
"""
991
if revision_id is None:
992
return [None]
993
vf = self._get_revision_vf()
994
try:
995
return [None] + vf.get_ancestry(revision_id)
996
except errors.RevisionNotPresent:
997
raise errors.NoSuchRevision(self, revision_id)
998
999
@needs_read_lock
1000
def get_revision(self, revision_id):
1001
"""Return the Revision object for a named revision"""
1002
return self.get_revision_reconcile(revision_id)
1003
1004
@needs_read_lock
1005
def get_revision_graph(self, revision_id=None):
1006
"""Return a dictionary containing the revision graph.
1007
1008
:param revision_id: The revision_id to get a graph from. If None, then
1009
the entire revision graph is returned. This is a deprecated mode of
1010
operation and will be removed in the future.
1011
:return: a dictionary of revision_id->revision_parents_list.
1012
"""
1013
# special case NULL_REVISION
1014
if revision_id == _mod_revision.NULL_REVISION:
1015
return {}
1016
a_weave = self._get_revision_vf()
1017
entire_graph = a_weave.get_graph()
1018
if revision_id is None:
1019
return a_weave.get_graph()
1020
elif revision_id not in a_weave:
1021
raise errors.NoSuchRevision(self, revision_id)
1022
else:
1023
# add what can be reached from revision_id
1024
result = {}
1025
pending = set([revision_id])
1026
while len(pending) > 0:
1027
node = pending.pop()
1028
result[node] = a_weave.get_parents(node)
1029
for revision_id in result[node]:
1030
if revision_id not in result:
1031
pending.add(revision_id)
1032
return result
1033
1034
@needs_read_lock
1035
def get_revision_graph_with_ghosts(self, revision_ids=None):
1036
"""Return a graph of the revisions with ghosts marked as applicable.
1037
1038
:param revision_ids: an iterable of revisions to graph or None for all.
1039
:return: a Graph object with the graph reachable from revision_ids.
1040
"""
1041
result = graph.Graph()
1042
vf = self._get_revision_vf()
1043
versions = set(vf.versions())
1044
if not revision_ids:
1045
pending = set(self.all_revision_ids())
1046
required = set([])
1047
else:
1048
pending = set(revision_ids)
1049
# special case NULL_REVISION
1050
if _mod_revision.NULL_REVISION in pending:
1051
pending.remove(_mod_revision.NULL_REVISION)
1052
required = set(pending)
1053
done = set([])
1054
while len(pending):
1055
revision_id = pending.pop()
1056
if not revision_id in versions:
1057
if revision_id in required:
1058
raise errors.NoSuchRevision(self, revision_id)
1059
# a ghost
1060
result.add_ghost(revision_id)
1061
# mark it as done so we don't try for it again.
1062
done.add(revision_id)
1063
continue
1064
parent_ids = vf.get_parents_with_ghosts(revision_id)
1065
for parent_id in parent_ids:
1066
# is this queued or done ?
1067
if (parent_id not in pending and
1068
parent_id not in done):
1069
# no, queue it.
1070
pending.add(parent_id)
1071
result.add_node(revision_id, parent_ids)
1072
done.add(revision_id)
1073
return result
1074
1075
def _get_revision_vf(self):
1076
""":return: a versioned file containing the revisions."""
1077
vf = self._revision_store.get_revision_file(self.get_transaction())
1078
return vf
1079
1080
@needs_write_lock
1081
def reconcile(self, other=None, thorough=False):
1082
"""Reconcile this repository."""
1083
from bzrlib.reconcile import KnitReconciler
1084
reconciler = KnitReconciler(self, thorough=thorough)
1085
reconciler.reconcile()
1086
return reconciler
1087
1088
def revision_parents(self, revision_id):
1089
return self._get_revision_vf().get_parents(revision_id)
1090
1091
1092
class KnitRepository2(KnitRepository):
1093
""""""
1094
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1095
control_store, text_store):
1096
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1097
_revision_store, control_store, text_store)
1098
self._serializer = xml6.serializer_v6
1099
1100
def deserialise_inventory(self, revision_id, xml):
1101
"""Transform the xml into an inventory object.
1102
1103
:param revision_id: The expected revision id of the inventory.
1104
:param xml: A serialised inventory.
1105
"""
1106
result = self._serializer.read_inventory_from_string(xml)
1107
assert result.root.revision is not None
1108
return result
1109
1110
def serialise_inventory(self, inv):
1111
"""Transform the inventory object into XML text.
1112
1113
:param revision_id: The expected revision id of the inventory.
1114
:param xml: A serialised inventory.
1115
"""
1116
assert inv.revision_id is not None
1117
assert inv.root.revision is not None
1118
return KnitRepository.serialise_inventory(self, inv)
1119
1120
def get_commit_builder(self, branch, parents, config, timestamp=None,
1121
timezone=None, committer=None, revprops=None,
1122
revision_id=None):
1123
"""Obtain a CommitBuilder for this repository.
1124
1125
:param branch: Branch to commit to.
1126
:param parents: Revision ids of the parents of the new revision.
1127
:param config: Configuration to use.
1128
:param timestamp: Optional timestamp recorded for commit.
1129
:param timezone: Optional timezone for timestamp.
1130
:param committer: Optional committer to set for commit.
1131
:param revprops: Optional dictionary of revision properties.
1132
:param revision_id: Optional revision id.
1133
"""
1134
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1135
committer, revprops, revision_id)
1136
1137
1138
class RepositoryFormat(object):
1139
"""A repository format.
1140
1141
Formats provide three things:
1142
* An initialization routine to construct repository data on disk.
1143
* a format string which is used when the BzrDir supports versioned
1144
children.
1145
* an open routine which returns a Repository instance.
1146
1147
Formats are placed in an dict by their format string for reference
1148
during opening. These should be subclasses of RepositoryFormat
1149
for consistency.
1150
1151
Once a format is deprecated, just deprecate the initialize and open
1152
methods on the format class. Do not deprecate the object, as the
1153
object will be created every system load.
1154
1155
Common instance attributes:
1156
_matchingbzrdir - the bzrdir format that the repository format was
1157
originally written to work with. This can be used if manually
1158
constructing a bzrdir and repository, or more commonly for test suite
1159
parameterisation.
1160
"""
1161
1162
_default_format = None
1163
"""The default format used for new repositories."""
1164
1165
_formats = {}
1166
"""The known formats."""
1167
1168
def __str__(self):
1169
return "<%s>" % self.__class__.__name__
1170
1171
@classmethod
1172
def find_format(klass, a_bzrdir):
1173
"""Return the format for the repository object in a_bzrdir."""
1174
try:
1175
transport = a_bzrdir.get_repository_transport(None)
1176
format_string = transport.get("format").read()
1177
return klass._formats[format_string]
1178
except errors.NoSuchFile:
1179
raise errors.NoRepositoryPresent(a_bzrdir)
1180
except KeyError:
1181
raise errors.UnknownFormatError(format=format_string)
1182
1183
def _get_control_store(self, repo_transport, control_files):
1184
"""Return the control store for this repository."""
1185
raise NotImplementedError(self._get_control_store)
1186
1187
@classmethod
1188
def get_default_format(klass):
1189
"""Return the current default format."""
1190
return klass._default_format
1191
1192
def get_format_string(self):
1193
"""Return the ASCII format string that identifies this format.
1194
1195
Note that in pre format ?? repositories the format string is
1196
not permitted nor written to disk.
1197
"""
1198
raise NotImplementedError(self.get_format_string)
1199
1200
def get_format_description(self):
1201
"""Return the short description for this format."""
1202
raise NotImplementedError(self.get_format_description)
1203
1204
def _get_revision_store(self, repo_transport, control_files):
1205
"""Return the revision store object for this a_bzrdir."""
1206
raise NotImplementedError(self._get_revision_store)
1207
1208
def _get_text_rev_store(self,
1209
transport,
1210
control_files,
1211
name,
1212
compressed=True,
1213
prefixed=False,
1214
serializer=None):
1215
"""Common logic for getting a revision store for a repository.
1216
1217
see self._get_revision_store for the subclass-overridable method to
1218
get the store for a repository.
1219
"""
1220
from bzrlib.store.revision.text import TextRevisionStore
1221
dir_mode = control_files._dir_mode
1222
file_mode = control_files._file_mode
1223
text_store =TextStore(transport.clone(name),
1224
prefixed=prefixed,
1225
compressed=compressed,
1226
dir_mode=dir_mode,
1227
file_mode=file_mode)
1228
_revision_store = TextRevisionStore(text_store, serializer)
1229
return _revision_store
1230
1231
def _get_versioned_file_store(self,
1232
name,
1233
transport,
1234
control_files,
1235
prefixed=True,
1236
versionedfile_class=weave.WeaveFile,
1237
versionedfile_kwargs={},
1238
escaped=False):
1239
weave_transport = control_files._transport.clone(name)
1240
dir_mode = control_files._dir_mode
1241
file_mode = control_files._file_mode
1242
return VersionedFileStore(weave_transport, prefixed=prefixed,
1243
dir_mode=dir_mode,
1244
file_mode=file_mode,
1245
versionedfile_class=versionedfile_class,
1246
versionedfile_kwargs=versionedfile_kwargs,
1247
escaped=escaped)
1248
1249
def initialize(self, a_bzrdir, shared=False):
1250
"""Initialize a repository of this format in a_bzrdir.
1251
1252
:param a_bzrdir: The bzrdir to put the new repository in it.
1253
:param shared: The repository should be initialized as a sharable one.
1254
1255
This may raise UninitializableFormat if shared repository are not
1256
compatible the a_bzrdir.
1257
"""
1258
1259
def is_supported(self):
1260
"""Is this format supported?
1261
1262
Supported formats must be initializable and openable.
1263
Unsupported formats may not support initialization or committing or
1264
some other features depending on the reason for not being supported.
1265
"""
1266
return True
1267
1268
def check_conversion_target(self, target_format):
1269
raise NotImplementedError(self.check_conversion_target)
1270
1271
def open(self, a_bzrdir, _found=False):
1272
"""Return an instance of this format for the bzrdir a_bzrdir.
1273
1274
_found is a private parameter, do not use it.
1275
"""
1276
raise NotImplementedError(self.open)
1277
1278
@classmethod
1279
def register_format(klass, format):
1280
klass._formats[format.get_format_string()] = format
1281
1282
@classmethod
1283
@deprecated_method(symbol_versioning.zero_fourteen)
1284
def set_default_format(klass, format):
1285
klass._set_default_format(format)
1286
1287
@classmethod
1288
def _set_default_format(klass, format):
1289
klass._default_format = format
1290
1291
@classmethod
1292
def unregister_format(klass, format):
1293
assert klass._formats[format.get_format_string()] is format
1294
del klass._formats[format.get_format_string()]
1295
1296
1297
class PreSplitOutRepositoryFormat(RepositoryFormat):
1298
"""Base class for the pre split out repository formats."""
1299
1300
rich_root_data = False
1301
1302
def initialize(self, a_bzrdir, shared=False, _internal=False):
1303
"""Create a weave repository.
1304
1305
TODO: when creating split out bzr branch formats, move this to a common
1306
base for Format5, Format6. or something like that.
1307
"""
1308
if shared:
1309
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1310
1311
if not _internal:
1312
# always initialized when the bzrdir is.
1313
return self.open(a_bzrdir, _found=True)
1314
1315
# Create an empty weave
1316
sio = StringIO()
1317
weavefile.write_weave_v5(weave.Weave(), sio)
1318
empty_weave = sio.getvalue()
1319
1320
mutter('creating repository in %s.', a_bzrdir.transport.base)
1321
dirs = ['revision-store', 'weaves']
1322
files = [('inventory.weave', StringIO(empty_weave)),
1323
]
1324
1325
# FIXME: RBC 20060125 don't peek under the covers
1326
# NB: no need to escape relative paths that are url safe.
1327
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1328
'branch-lock', lockable_files.TransportLock)
1329
control_files.create_lock()
1330
control_files.lock_write()
1331
control_files._transport.mkdir_multi(dirs,
1332
mode=control_files._dir_mode)
1333
try:
1334
for file, content in files:
1335
control_files.put(file, content)
1336
finally:
1337
control_files.unlock()
1338
return self.open(a_bzrdir, _found=True)
1339
1340
def _get_control_store(self, repo_transport, control_files):
1341
"""Return the control store for this repository."""
1342
return self._get_versioned_file_store('',
1343
repo_transport,
1344
control_files,
1345
prefixed=False)
1346
1347
def _get_text_store(self, transport, control_files):
1348
"""Get a store for file texts for this format."""
1349
raise NotImplementedError(self._get_text_store)
1350
1351
def open(self, a_bzrdir, _found=False):
1352
"""See RepositoryFormat.open()."""
1353
if not _found:
1354
# we are being called directly and must probe.
1355
raise NotImplementedError
1356
1357
repo_transport = a_bzrdir.get_repository_transport(None)
1358
control_files = a_bzrdir._control_files
1359
text_store = self._get_text_store(repo_transport, control_files)
1360
control_store = self._get_control_store(repo_transport, control_files)
1361
_revision_store = self._get_revision_store(repo_transport, control_files)
1362
return AllInOneRepository(_format=self,
1363
a_bzrdir=a_bzrdir,
1364
_revision_store=_revision_store,
1365
control_store=control_store,
1366
text_store=text_store)
1367
1368
def check_conversion_target(self, target_format):
1369
pass
1370
1371
1372
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1373
"""Bzr repository format 4.
1374
1375
This repository format has:
1376
- flat stores
1377
- TextStores for texts, inventories,revisions.
1378
1379
This format is deprecated: it indexes texts using a text id which is
1380
removed in format 5; initialization and write support for this format
1381
has been removed.
1382
"""
1383
1384
def __init__(self):
1385
super(RepositoryFormat4, self).__init__()
1386
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1387
1388
def get_format_description(self):
1389
"""See RepositoryFormat.get_format_description()."""
1390
return "Repository format 4"
1391
1392
def initialize(self, url, shared=False, _internal=False):
1393
"""Format 4 branches cannot be created."""
1394
raise errors.UninitializableFormat(self)
1395
1396
def is_supported(self):
1397
"""Format 4 is not supported.
1398
1399
It is not supported because the model changed from 4 to 5 and the
1400
conversion logic is expensive - so doing it on the fly was not
1401
feasible.
1402
"""
1403
return False
1404
1405
def _get_control_store(self, repo_transport, control_files):
1406
"""Format 4 repositories have no formal control store at this point.
1407
1408
This will cause any control-file-needing apis to fail - this is desired.
1409
"""
1410
return None
1411
1412
def _get_revision_store(self, repo_transport, control_files):
1413
"""See RepositoryFormat._get_revision_store()."""
1414
from bzrlib.xml4 import serializer_v4
1415
return self._get_text_rev_store(repo_transport,
1416
control_files,
1417
'revision-store',
1418
serializer=serializer_v4)
1419
1420
def _get_text_store(self, transport, control_files):
1421
"""See RepositoryFormat._get_text_store()."""
1422
1423
1424
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1425
"""Bzr control format 5.
1426
1427
This repository format has:
1428
- weaves for file texts and inventory
1429
- flat stores
1430
- TextStores for revisions and signatures.
1431
"""
1432
1433
def __init__(self):
1434
super(RepositoryFormat5, self).__init__()
1435
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1436
1437
def get_format_description(self):
1438
"""See RepositoryFormat.get_format_description()."""
1439
return "Weave repository format 5"
1440
1441
def _get_revision_store(self, repo_transport, control_files):
1442
"""See RepositoryFormat._get_revision_store()."""
1443
"""Return the revision store object for this a_bzrdir."""
1444
return self._get_text_rev_store(repo_transport,
1445
control_files,
1446
'revision-store',
1447
compressed=False)
1448
1449
def _get_text_store(self, transport, control_files):
1450
"""See RepositoryFormat._get_text_store()."""
1451
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1452
1453
1454
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1455
"""Bzr control format 6.
1456
1457
This repository format has:
1458
- weaves for file texts and inventory
1459
- hash subdirectory based stores.
1460
- TextStores for revisions and signatures.
1461
"""
1462
1463
def __init__(self):
1464
super(RepositoryFormat6, self).__init__()
1465
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1466
1467
def get_format_description(self):
1468
"""See RepositoryFormat.get_format_description()."""
1469
return "Weave repository format 6"
1470
1471
def _get_revision_store(self, repo_transport, control_files):
1472
"""See RepositoryFormat._get_revision_store()."""
1473
return self._get_text_rev_store(repo_transport,
1474
control_files,
1475
'revision-store',
1476
compressed=False,
1477
prefixed=True)
1478
1479
def _get_text_store(self, transport, control_files):
1480
"""See RepositoryFormat._get_text_store()."""
1481
return self._get_versioned_file_store('weaves', transport, control_files)
1482
1483
1484
class MetaDirRepositoryFormat(RepositoryFormat):
1485
"""Common base class for the new repositories using the metadir layout."""
1486
1487
rich_root_data = False
1488
1489
def __init__(self):
1490
super(MetaDirRepositoryFormat, self).__init__()
1491
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1492
1493
def _create_control_files(self, a_bzrdir):
1494
"""Create the required files and the initial control_files object."""
1495
# FIXME: RBC 20060125 don't peek under the covers
1496
# NB: no need to escape relative paths that are url safe.
1497
repository_transport = a_bzrdir.get_repository_transport(self)
1498
control_files = lockable_files.LockableFiles(repository_transport,
1499
'lock', lockdir.LockDir)
1500
control_files.create_lock()
1501
return control_files
1502
1503
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1504
"""Upload the initial blank content."""
1505
control_files = self._create_control_files(a_bzrdir)
1506
control_files.lock_write()
1507
try:
1508
control_files._transport.mkdir_multi(dirs,
1509
mode=control_files._dir_mode)
1510
for file, content in files:
1511
control_files.put(file, content)
1512
for file, content in utf8_files:
1513
control_files.put_utf8(file, content)
1514
if shared == True:
1515
control_files.put_utf8('shared-storage', '')
1516
finally:
1517
control_files.unlock()
1518
1519
1520
class RepositoryFormat7(MetaDirRepositoryFormat):
1521
"""Bzr repository 7.
1522
1523
This repository format has:
1524
- weaves for file texts and inventory
1525
- hash subdirectory based stores.
1526
- TextStores for revisions and signatures.
1527
- a format marker of its own
1528
- an optional 'shared-storage' flag
1529
- an optional 'no-working-trees' flag
1530
"""
1531
1532
def _get_control_store(self, repo_transport, control_files):
1533
"""Return the control store for this repository."""
1534
return self._get_versioned_file_store('',
1535
repo_transport,
1536
control_files,
1537
prefixed=False)
1538
1539
def get_format_string(self):
1540
"""See RepositoryFormat.get_format_string()."""
1541
return "Bazaar-NG Repository format 7"
1542
1543
def get_format_description(self):
1544
"""See RepositoryFormat.get_format_description()."""
1545
return "Weave repository format 7"
1546
1547
def check_conversion_target(self, target_format):
1548
pass
1549
1550
def _get_revision_store(self, repo_transport, control_files):
1551
"""See RepositoryFormat._get_revision_store()."""
1552
return self._get_text_rev_store(repo_transport,
1553
control_files,
1554
'revision-store',
1555
compressed=False,
1556
prefixed=True,
1557
)
1558
1559
def _get_text_store(self, transport, control_files):
1560
"""See RepositoryFormat._get_text_store()."""
1561
return self._get_versioned_file_store('weaves',
1562
transport,
1563
control_files)
1564
1565
def initialize(self, a_bzrdir, shared=False):
1566
"""Create a weave repository.
1567
1568
:param shared: If true the repository will be initialized as a shared
1569
repository.
1570
"""
1571
# Create an empty weave
1572
sio = StringIO()
1573
weavefile.write_weave_v5(weave.Weave(), sio)
1574
empty_weave = sio.getvalue()
1575
1576
mutter('creating repository in %s.', a_bzrdir.transport.base)
1577
dirs = ['revision-store', 'weaves']
1578
files = [('inventory.weave', StringIO(empty_weave)),
1579
]
1580
utf8_files = [('format', self.get_format_string())]
1581
1582
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1583
return self.open(a_bzrdir=a_bzrdir, _found=True)
1584
1585
def open(self, a_bzrdir, _found=False, _override_transport=None):
1586
"""See RepositoryFormat.open().
1587
1588
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1589
repository at a slightly different url
1590
than normal. I.e. during 'upgrade'.
1591
"""
1592
if not _found:
1593
format = RepositoryFormat.find_format(a_bzrdir)
1594
assert format.__class__ == self.__class__
1595
if _override_transport is not None:
1596
repo_transport = _override_transport
1597
else:
1598
repo_transport = a_bzrdir.get_repository_transport(None)
1599
control_files = lockable_files.LockableFiles(repo_transport,
1600
'lock', lockdir.LockDir)
1601
text_store = self._get_text_store(repo_transport, control_files)
1602
control_store = self._get_control_store(repo_transport, control_files)
1603
_revision_store = self._get_revision_store(repo_transport, control_files)
1604
return WeaveMetaDirRepository(_format=self,
1605
a_bzrdir=a_bzrdir,
1606
control_files=control_files,
1607
_revision_store=_revision_store,
1608
control_store=control_store,
1609
text_store=text_store)
1610
1611
1612
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1613
"""Bzr repository knit format (generalized).
1614
1615
This repository format has:
1616
- knits for file texts and inventory
1617
- hash subdirectory based stores.
1618
- knits for revisions and signatures
1619
- TextStores for revisions and signatures.
1620
- a format marker of its own
1621
- an optional 'shared-storage' flag
1622
- an optional 'no-working-trees' flag
1623
- a LockDir lock
1624
"""
1625
1626
def _get_control_store(self, repo_transport, control_files):
1627
"""Return the control store for this repository."""
1628
return VersionedFileStore(
1629
repo_transport,
1630
prefixed=False,
1631
file_mode=control_files._file_mode,
1632
versionedfile_class=knit.KnitVersionedFile,
1633
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1634
)
1635
1636
def _get_revision_store(self, repo_transport, control_files):
1637
"""See RepositoryFormat._get_revision_store()."""
1638
from bzrlib.store.revision.knit import KnitRevisionStore
1639
versioned_file_store = VersionedFileStore(
1640
repo_transport,
1641
file_mode=control_files._file_mode,
1642
prefixed=False,
1643
precious=True,
1644
versionedfile_class=knit.KnitVersionedFile,
1645
versionedfile_kwargs={'delta':False,
1646
'factory':knit.KnitPlainFactory(),
1647
},
1648
escaped=True,
1649
)
1650
return KnitRevisionStore(versioned_file_store)
1651
1652
def _get_text_store(self, transport, control_files):
1653
"""See RepositoryFormat._get_text_store()."""
1654
return self._get_versioned_file_store('knits',
1655
transport,
1656
control_files,
1657
versionedfile_class=knit.KnitVersionedFile,
1658
versionedfile_kwargs={
1659
'create_parent_dir':True,
1660
'delay_create':True,
1661
'dir_mode':control_files._dir_mode,
1662
},
1663
escaped=True)
1664
1665
def initialize(self, a_bzrdir, shared=False):
1666
"""Create a knit format 1 repository.
1667
1668
:param a_bzrdir: bzrdir to contain the new repository; must already
1669
be initialized.
1670
:param shared: If true the repository will be initialized as a shared
1671
repository.
1672
"""
1673
mutter('creating repository in %s.', a_bzrdir.transport.base)
1674
dirs = ['revision-store', 'knits']
1675
files = []
1676
utf8_files = [('format', self.get_format_string())]
1677
1678
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1679
repo_transport = a_bzrdir.get_repository_transport(None)
1680
control_files = lockable_files.LockableFiles(repo_transport,
1681
'lock', lockdir.LockDir)
1682
control_store = self._get_control_store(repo_transport, control_files)
1683
transaction = transactions.WriteTransaction()
1684
# trigger a write of the inventory store.
1685
control_store.get_weave_or_empty('inventory', transaction)
1686
_revision_store = self._get_revision_store(repo_transport, control_files)
1687
# the revision id here is irrelevant: it will not be stored, and cannot
1688
# already exist.
1689
_revision_store.has_revision_id('A', transaction)
1690
_revision_store.get_signature_file(transaction)
1691
return self.open(a_bzrdir=a_bzrdir, _found=True)
1692
1693
def open(self, a_bzrdir, _found=False, _override_transport=None):
1694
"""See RepositoryFormat.open().
1695
1696
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1697
repository at a slightly different url
1698
than normal. I.e. during 'upgrade'.
1699
"""
1700
if not _found:
1701
format = RepositoryFormat.find_format(a_bzrdir)
1702
assert format.__class__ == self.__class__
1703
if _override_transport is not None:
1704
repo_transport = _override_transport
1705
else:
1706
repo_transport = a_bzrdir.get_repository_transport(None)
1707
control_files = lockable_files.LockableFiles(repo_transport,
1708
'lock', lockdir.LockDir)
1709
text_store = self._get_text_store(repo_transport, control_files)
1710
control_store = self._get_control_store(repo_transport, control_files)
1711
_revision_store = self._get_revision_store(repo_transport, control_files)
1712
return KnitRepository(_format=self,
1713
a_bzrdir=a_bzrdir,
1714
control_files=control_files,
1715
_revision_store=_revision_store,
1716
control_store=control_store,
1717
text_store=text_store)
1718
1719
1720
class RepositoryFormatKnit1(RepositoryFormatKnit):
1721
"""Bzr repository knit format 1.
1722
1723
This repository format has:
1724
- knits for file texts and inventory
1725
- hash subdirectory based stores.
1726
- knits for revisions and signatures
1727
- TextStores for revisions and signatures.
1728
- a format marker of its own
1729
- an optional 'shared-storage' flag
1730
- an optional 'no-working-trees' flag
1731
- a LockDir lock
1732
1733
This format was introduced in bzr 0.8.
1734
"""
1735
def get_format_string(self):
1736
"""See RepositoryFormat.get_format_string()."""
1737
return "Bazaar-NG Knit Repository Format 1"
1738
1739
def get_format_description(self):
1740
"""See RepositoryFormat.get_format_description()."""
1741
return "Knit repository format 1"
1742
1743
def check_conversion_target(self, target_format):
1744
pass
1745
1746
1747
class RepositoryFormatKnit2(RepositoryFormatKnit):
1748
"""Bzr repository knit format 2.
1749
1750
THIS FORMAT IS EXPERIMENTAL
1751
This repository format has:
1752
- knits for file texts and inventory
1753
- hash subdirectory based stores.
1754
- knits for revisions and signatures
1755
- TextStores for revisions and signatures.
1756
- a format marker of its own
1757
- an optional 'shared-storage' flag
1758
- an optional 'no-working-trees' flag
1759
- a LockDir lock
1760
- Support for recording full info about the tree root
1761
1762
"""
1763
1764
rich_root_data = True
1765
1766
def get_format_string(self):
1767
"""See RepositoryFormat.get_format_string()."""
1768
return "Bazaar Knit Repository Format 2\n"
1769
1770
def get_format_description(self):
1771
"""See RepositoryFormat.get_format_description()."""
1772
return "Knit repository format 2"
1773
1774
def check_conversion_target(self, target_format):
1775
if not target_format.rich_root_data:
1776
raise errors.BadConversionTarget(
1777
'Does not support rich root data.', target_format)
1778
1779
def open(self, a_bzrdir, _found=False, _override_transport=None):
1780
"""See RepositoryFormat.open().
1781
1782
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1783
repository at a slightly different url
1784
than normal. I.e. during 'upgrade'.
1785
"""
1786
if not _found:
1787
format = RepositoryFormat.find_format(a_bzrdir)
1788
assert format.__class__ == self.__class__
1789
if _override_transport is not None:
1790
repo_transport = _override_transport
1791
else:
1792
repo_transport = a_bzrdir.get_repository_transport(None)
1793
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1794
lockdir.LockDir)
1795
text_store = self._get_text_store(repo_transport, control_files)
1796
control_store = self._get_control_store(repo_transport, control_files)
1797
_revision_store = self._get_revision_store(repo_transport, control_files)
1798
return KnitRepository2(_format=self,
1799
a_bzrdir=a_bzrdir,
1800
control_files=control_files,
1801
_revision_store=_revision_store,
1802
control_store=control_store,
1803
text_store=text_store)
1804
1805
1806
1807
# formats which have no format string are not discoverable
1808
# and not independently creatable, so are not registered.
1809
RepositoryFormat.register_format(RepositoryFormat7())
1810
# KEEP in sync with bzrdir.format_registry default
1811
_default_format = RepositoryFormatKnit1()
1812
RepositoryFormat.register_format(_default_format)
1813
RepositoryFormat.register_format(RepositoryFormatKnit2())
1814
RepositoryFormat._set_default_format(_default_format)
1815
_legacy_formats = [RepositoryFormat4(),
1816
RepositoryFormat5(),
1817
RepositoryFormat6()]
1818
1819
1820
class InterRepository(InterObject):
1821
"""This class represents operations taking place between two repositories.
1822
1823
Its instances have methods like copy_content and fetch, and contain
1824
references to the source and target repositories these operations can be
1825
carried out on.
1826
1827
Often we will provide convenience methods on 'repository' which carry out
1828
operations with another repository - they will always forward to
1829
InterRepository.get(other).method_name(parameters).
1830
"""
1831
1832
_optimisers = []
1833
"""The available optimised InterRepository types."""
1834
1835
def copy_content(self, revision_id=None, basis=None):
1836
raise NotImplementedError(self.copy_content)
1837
1838
def fetch(self, revision_id=None, pb=None):
1839
"""Fetch the content required to construct revision_id.
1840
1841
The content is copied from self.source to self.target.
1842
1843
:param revision_id: if None all content is copied, if NULL_REVISION no
1844
content is copied.
1845
:param pb: optional progress bar to use for progress reports. If not
1846
provided a default one will be created.
1847
1848
Returns the copied revision count and the failed revisions in a tuple:
1849
(copied, failures).
1850
"""
1851
raise NotImplementedError(self.fetch)
1852
1853
@needs_read_lock
1854
def missing_revision_ids(self, revision_id=None):
1855
"""Return the revision ids that source has that target does not.
1856
1857
These are returned in topological order.
1858
1859
:param revision_id: only return revision ids included by this
1860
revision_id.
1861
"""
1862
# generic, possibly worst case, slow code path.
1863
target_ids = set(self.target.all_revision_ids())
1864
if revision_id is not None:
1865
source_ids = self.source.get_ancestry(revision_id)
1866
assert source_ids[0] is None
1867
source_ids.pop(0)
1868
else:
1869
source_ids = self.source.all_revision_ids()
1870
result_set = set(source_ids).difference(target_ids)
1871
# this may look like a no-op: its not. It preserves the ordering
1872
# other_ids had while only returning the members from other_ids
1873
# that we've decided we need.
1874
return [rev_id for rev_id in source_ids if rev_id in result_set]
1875
1876
1877
class InterSameDataRepository(InterRepository):
1878
"""Code for converting between repositories that represent the same data.
1879
1880
Data format and model must match for this to work.
1881
"""
1882
1883
_matching_repo_format = RepositoryFormat4()
1884
"""Repository format for testing with."""
1885
1886
@staticmethod
1887
def is_compatible(source, target):
1888
if not isinstance(source, Repository):
1889
return False
1890
if not isinstance(target, Repository):
1891
return False
1892
if source._format.rich_root_data == target._format.rich_root_data:
1893
return True
1894
else:
1895
return False
1896
1897
@needs_write_lock
1898
def copy_content(self, revision_id=None, basis=None):
1899
"""Make a complete copy of the content in self into destination.
1900
1901
This is a destructive operation! Do not use it on existing
1902
repositories.
1903
1904
:param revision_id: Only copy the content needed to construct
1905
revision_id and its parents.
1906
:param basis: Copy the needed data preferentially from basis.
1907
"""
1908
try:
1909
self.target.set_make_working_trees(self.source.make_working_trees())
1910
except NotImplementedError:
1911
pass
1912
# grab the basis available data
1913
if basis is not None:
1914
self.target.fetch(basis, revision_id=revision_id)
1915
# but don't bother fetching if we have the needed data now.
1916
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1917
self.target.has_revision(revision_id)):
1918
return
1919
self.target.fetch(self.source, revision_id=revision_id)
1920
1921
@needs_write_lock
1922
def fetch(self, revision_id=None, pb=None):
1923
"""See InterRepository.fetch()."""
1924
from bzrlib.fetch import GenericRepoFetcher
1925
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1926
self.source, self.source._format, self.target,
1927
self.target._format)
1928
f = GenericRepoFetcher(to_repository=self.target,
1929
from_repository=self.source,
1930
last_revision=revision_id,
1931
pb=pb)
1932
return f.count_copied, f.failed_revisions
1933
1934
1935
class InterWeaveRepo(InterSameDataRepository):
1936
"""Optimised code paths between Weave based repositories."""
1937
1938
_matching_repo_format = RepositoryFormat7()
1939
"""Repository format for testing with."""
1940
1941
@staticmethod
1942
def is_compatible(source, target):
1943
"""Be compatible with known Weave formats.
1944
1945
We don't test for the stores being of specific types because that
1946
could lead to confusing results, and there is no need to be
1947
overly general.
1948
"""
1949
try:
1950
return (isinstance(source._format, (RepositoryFormat5,
1951
RepositoryFormat6,
1952
RepositoryFormat7)) and
1953
isinstance(target._format, (RepositoryFormat5,
1954
RepositoryFormat6,
1955
RepositoryFormat7)))
1956
except AttributeError:
1957
return False
1958
1959
@needs_write_lock
1960
def copy_content(self, revision_id=None, basis=None):
1961
"""See InterRepository.copy_content()."""
1962
# weave specific optimised path:
1963
if basis is not None:
1964
# copy the basis in, then fetch remaining data.
1965
basis.copy_content_into(self.target, revision_id)
1966
# the basis copy_content_into could miss-set this.
1967
try:
1968
self.target.set_make_working_trees(self.source.make_working_trees())
1969
except NotImplementedError:
1970
pass
1971
self.target.fetch(self.source, revision_id=revision_id)
1972
else:
1973
try:
1974
self.target.set_make_working_trees(self.source.make_working_trees())
1975
except NotImplementedError:
1976
pass
1977
# FIXME do not peek!
1978
if self.source.control_files._transport.listable():
1979
pb = ui.ui_factory.nested_progress_bar()
1980
try:
1981
self.target.weave_store.copy_all_ids(
1982
self.source.weave_store,
1983
pb=pb,
1984
from_transaction=self.source.get_transaction(),
1985
to_transaction=self.target.get_transaction())
1986
pb.update('copying inventory', 0, 1)
1987
self.target.control_weaves.copy_multi(
1988
self.source.control_weaves, ['inventory'],
1989
from_transaction=self.source.get_transaction(),
1990
to_transaction=self.target.get_transaction())
1991
self.target._revision_store.text_store.copy_all_ids(
1992
self.source._revision_store.text_store,
1993
pb=pb)
1994
finally:
1995
pb.finished()
1996
else:
1997
self.target.fetch(self.source, revision_id=revision_id)
1998
1999
@needs_write_lock
2000
def fetch(self, revision_id=None, pb=None):
2001
"""See InterRepository.fetch()."""
2002
from bzrlib.fetch import GenericRepoFetcher
2003
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2004
self.source, self.source._format, self.target, self.target._format)
2005
f = GenericRepoFetcher(to_repository=self.target,
2006
from_repository=self.source,
2007
last_revision=revision_id,
2008
pb=pb)
2009
return f.count_copied, f.failed_revisions
2010
2011
@needs_read_lock
2012
def missing_revision_ids(self, revision_id=None):
2013
"""See InterRepository.missing_revision_ids()."""
2014
# we want all revisions to satisfy revision_id in source.
2015
# but we don't want to stat every file here and there.
2016
# we want then, all revisions other needs to satisfy revision_id
2017
# checked, but not those that we have locally.
2018
# so the first thing is to get a subset of the revisions to
2019
# satisfy revision_id in source, and then eliminate those that
2020
# we do already have.
2021
# this is slow on high latency connection to self, but as as this
2022
# disk format scales terribly for push anyway due to rewriting
2023
# inventory.weave, this is considered acceptable.
2024
# - RBC 20060209
2025
if revision_id is not None:
2026
source_ids = self.source.get_ancestry(revision_id)
2027
assert source_ids[0] is None
2028
source_ids.pop(0)
2029
else:
2030
source_ids = self.source._all_possible_ids()
2031
source_ids_set = set(source_ids)
2032
# source_ids is the worst possible case we may need to pull.
2033
# now we want to filter source_ids against what we actually
2034
# have in target, but don't try to check for existence where we know
2035
# we do not have a revision as that would be pointless.
2036
target_ids = set(self.target._all_possible_ids())
2037
possibly_present_revisions = target_ids.intersection(source_ids_set)
2038
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2039
required_revisions = source_ids_set.difference(actually_present_revisions)
2040
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2041
if revision_id is not None:
2042
# we used get_ancestry to determine source_ids then we are assured all
2043
# revisions referenced are present as they are installed in topological order.
2044
# and the tip revision was validated by get_ancestry.
2045
return required_topo_revisions
2046
else:
2047
# if we just grabbed the possibly available ids, then
2048
# we only have an estimate of whats available and need to validate
2049
# that against the revision records.
2050
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2051
2052
2053
class InterKnitRepo(InterSameDataRepository):
2054
"""Optimised code paths between Knit based repositories."""
2055
2056
_matching_repo_format = RepositoryFormatKnit1()
2057
"""Repository format for testing with."""
2058
2059
@staticmethod
2060
def is_compatible(source, target):
2061
"""Be compatible with known Knit formats.
2062
2063
We don't test for the stores being of specific types because that
2064
could lead to confusing results, and there is no need to be
2065
overly general.
2066
"""
2067
try:
2068
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2069
isinstance(target._format, (RepositoryFormatKnit1)))
2070
except AttributeError:
2071
return False
2072
2073
@needs_write_lock
2074
def fetch(self, revision_id=None, pb=None):
2075
"""See InterRepository.fetch()."""
2076
from bzrlib.fetch import KnitRepoFetcher
2077
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2078
self.source, self.source._format, self.target, self.target._format)
2079
f = KnitRepoFetcher(to_repository=self.target,
2080
from_repository=self.source,
2081
last_revision=revision_id,
2082
pb=pb)
2083
return f.count_copied, f.failed_revisions
2084
2085
@needs_read_lock
2086
def missing_revision_ids(self, revision_id=None):
2087
"""See InterRepository.missing_revision_ids()."""
2088
if revision_id is not None:
2089
source_ids = self.source.get_ancestry(revision_id)
2090
assert source_ids[0] is None
2091
source_ids.pop(0)
2092
else:
2093
source_ids = self.source._all_possible_ids()
2094
source_ids_set = set(source_ids)
2095
# source_ids is the worst possible case we may need to pull.
2096
# now we want to filter source_ids against what we actually
2097
# have in target, but don't try to check for existence where we know
2098
# we do not have a revision as that would be pointless.
2099
target_ids = set(self.target._all_possible_ids())
2100
possibly_present_revisions = target_ids.intersection(source_ids_set)
2101
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2102
required_revisions = source_ids_set.difference(actually_present_revisions)
2103
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2104
if revision_id is not None:
2105
# we used get_ancestry to determine source_ids then we are assured all
2106
# revisions referenced are present as they are installed in topological order.
2107
# and the tip revision was validated by get_ancestry.
2108
return required_topo_revisions
2109
else:
2110
# if we just grabbed the possibly available ids, then
2111
# we only have an estimate of whats available and need to validate
2112
# that against the revision records.
2113
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2114
2115
2116
class InterModel1and2(InterRepository):
2117
2118
_matching_repo_format = None
2119
2120
@staticmethod
2121
def is_compatible(source, target):
2122
if not isinstance(source, Repository):
2123
return False
2124
if not isinstance(target, Repository):
2125
return False
2126
if not source._format.rich_root_data and target._format.rich_root_data:
2127
return True
2128
else:
2129
return False
2130
2131
@needs_write_lock
2132
def fetch(self, revision_id=None, pb=None):
2133
"""See InterRepository.fetch()."""
2134
from bzrlib.fetch import Model1toKnit2Fetcher
2135
f = Model1toKnit2Fetcher(to_repository=self.target,
2136
from_repository=self.source,
2137
last_revision=revision_id,
2138
pb=pb)
2139
return f.count_copied, f.failed_revisions
2140
2141
@needs_write_lock
2142
def copy_content(self, revision_id=None, basis=None):
2143
"""Make a complete copy of the content in self into destination.
2144
2145
This is a destructive operation! Do not use it on existing
2146
repositories.
2147
2148
:param revision_id: Only copy the content needed to construct
2149
revision_id and its parents.
2150
:param basis: Copy the needed data preferentially from basis.
2151
"""
2152
try:
2153
self.target.set_make_working_trees(self.source.make_working_trees())
2154
except NotImplementedError:
2155
pass
2156
# grab the basis available data
2157
if basis is not None:
2158
self.target.fetch(basis, revision_id=revision_id)
2159
# but don't bother fetching if we have the needed data now.
2160
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2161
self.target.has_revision(revision_id)):
2162
return
2163
self.target.fetch(self.source, revision_id=revision_id)
2164
2165
2166
class InterKnit1and2(InterKnitRepo):
2167
2168
_matching_repo_format = None
2169
2170
@staticmethod
2171
def is_compatible(source, target):
2172
"""Be compatible with Knit1 source and Knit2 target"""
2173
try:
2174
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2175
isinstance(target._format, (RepositoryFormatKnit2)))
2176
except AttributeError:
2177
return False
2178
2179
@needs_write_lock
2180
def fetch(self, revision_id=None, pb=None):
2181
"""See InterRepository.fetch()."""
2182
from bzrlib.fetch import Knit1to2Fetcher
2183
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2184
self.source, self.source._format, self.target,
2185
self.target._format)
2186
f = Knit1to2Fetcher(to_repository=self.target,
2187
from_repository=self.source,
2188
last_revision=revision_id,
2189
pb=pb)
2190
return f.count_copied, f.failed_revisions
2191
2192
2193
InterRepository.register_optimiser(InterSameDataRepository)
2194
InterRepository.register_optimiser(InterWeaveRepo)
2195
InterRepository.register_optimiser(InterKnitRepo)
2196
InterRepository.register_optimiser(InterModel1and2)
2197
InterRepository.register_optimiser(InterKnit1and2)
2198
2199
2200
class RepositoryTestProviderAdapter(object):
2201
"""A tool to generate a suite testing multiple repository formats at once.
2202
2203
This is done by copying the test once for each transport and injecting
2204
the transport_server, transport_readonly_server, and bzrdir_format and
2205
repository_format classes into each copy. Each copy is also given a new id()
2206
to make it easy to identify.
2207
"""
2208
2209
def __init__(self, transport_server, transport_readonly_server, formats):
2210
self._transport_server = transport_server
2211
self._transport_readonly_server = transport_readonly_server
2212
self._formats = formats
2213
2214
def adapt(self, test):
2215
result = unittest.TestSuite()
2216
for repository_format, bzrdir_format in self._formats:
2217
new_test = deepcopy(test)
2218
new_test.transport_server = self._transport_server
2219
new_test.transport_readonly_server = self._transport_readonly_server
2220
new_test.bzrdir_format = bzrdir_format
2221
new_test.repository_format = repository_format
2222
def make_new_test_id():
2223
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2224
return lambda: new_id
2225
new_test.id = make_new_test_id()
2226
result.addTest(new_test)
2227
return result
2228
2229
2230
class InterRepositoryTestProviderAdapter(object):
2231
"""A tool to generate a suite testing multiple inter repository formats.
2232
2233
This is done by copying the test once for each interrepo provider and injecting
2234
the transport_server, transport_readonly_server, repository_format and
2235
repository_to_format classes into each copy.
2236
Each copy is also given a new id() to make it easy to identify.
2237
"""
2238
2239
def __init__(self, transport_server, transport_readonly_server, formats):
2240
self._transport_server = transport_server
2241
self._transport_readonly_server = transport_readonly_server
2242
self._formats = formats
2243
2244
def adapt(self, test):
2245
result = unittest.TestSuite()
2246
for interrepo_class, repository_format, repository_format_to in self._formats:
2247
new_test = deepcopy(test)
2248
new_test.transport_server = self._transport_server
2249
new_test.transport_readonly_server = self._transport_readonly_server
2250
new_test.interrepo_class = interrepo_class
2251
new_test.repository_format = repository_format
2252
new_test.repository_format_to = repository_format_to
2253
def make_new_test_id():
2254
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2255
return lambda: new_id
2256
new_test.id = make_new_test_id()
2257
result.addTest(new_test)
2258
return result
2259
2260
@staticmethod
2261
def default_test_list():
2262
"""Generate the default list of interrepo permutations to test."""
2263
result = []
2264
# test the default InterRepository between format 6 and the current
2265
# default format.
2266
# XXX: robertc 20060220 reinstate this when there are two supported
2267
# formats which do not have an optimal code path between them.
2268
#result.append((InterRepository,
2269
# RepositoryFormat6(),
2270
# RepositoryFormatKnit1()))
2271
for optimiser in InterRepository._optimisers:
2272
if optimiser._matching_repo_format is not None:
2273
result.append((optimiser,
2274
optimiser._matching_repo_format,
2275
optimiser._matching_repo_format
2276
))
2277
# if there are specific combinations we want to use, we can add them
2278
# here.
2279
result.append((InterModel1and2, RepositoryFormat5(),
2280
RepositoryFormatKnit2()))
2281
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2282
RepositoryFormatKnit2()))
2283
return result
2284
2285
2286
class CopyConverter(object):
2287
"""A repository conversion tool which just performs a copy of the content.
2288
2289
This is slow but quite reliable.
2290
"""
2291
2292
def __init__(self, target_format):
2293
"""Create a CopyConverter.
2294
2295
:param target_format: The format the resulting repository should be.
2296
"""
2297
self.target_format = target_format
2298
2299
def convert(self, repo, pb):
2300
"""Perform the conversion of to_convert, giving feedback via pb.
2301
2302
:param to_convert: The disk object to convert.
2303
:param pb: a progress bar to use for progress information.
2304
"""
2305
self.pb = pb
2306
self.count = 0
2307
self.total = 4
2308
# this is only useful with metadir layouts - separated repo content.
2309
# trigger an assertion if not such
2310
repo._format.get_format_string()
2311
self.repo_dir = repo.bzrdir
2312
self.step('Moving repository to repository.backup')
2313
self.repo_dir.transport.move('repository', 'repository.backup')
2314
backup_transport = self.repo_dir.transport.clone('repository.backup')
2315
repo._format.check_conversion_target(self.target_format)
2316
self.source_repo = repo._format.open(self.repo_dir,
2317
_found=True,
2318
_override_transport=backup_transport)
2319
self.step('Creating new repository')
2320
converted = self.target_format.initialize(self.repo_dir,
2321
self.source_repo.is_shared())
2322
converted.lock_write()
2323
try:
2324
self.step('Copying content into repository.')
2325
self.source_repo.copy_content_into(converted)
2326
finally:
2327
converted.unlock()
2328
self.step('Deleting old repository content.')
2329
self.repo_dir.transport.delete_tree('repository.backup')
2330
self.pb.note('repository converted')
2331
2332
def step(self, message):
2333
"""Update the pb by a step."""
2334
self.count +=1
2335
self.pb.update(message, self.count, self.total)
2336
2337
2338
class CommitBuilder(object):
2339
"""Provides an interface to build up a commit.
2340
2341
This allows describing a tree to be committed without needing to
2342
know the internals of the format of the repository.
2343
"""
2344
2345
record_root_entry = False
2346
def __init__(self, repository, parents, config, timestamp=None,
2347
timezone=None, committer=None, revprops=None,
2348
revision_id=None):
2349
"""Initiate a CommitBuilder.
2350
2351
:param repository: Repository to commit to.
2352
:param parents: Revision ids of the parents of the new revision.
2353
:param config: Configuration to use.
2354
:param timestamp: Optional timestamp recorded for commit.
2355
:param timezone: Optional timezone for timestamp.
2356
:param committer: Optional committer to set for commit.
2357
:param revprops: Optional dictionary of revision properties.
2358
:param revision_id: Optional revision id.
2359
"""
2360
self._config = config
2361
2362
if committer is None:
2363
self._committer = self._config.username()
2364
else:
2365
assert isinstance(committer, basestring), type(committer)
2366
self._committer = committer
2367
2368
self.new_inventory = Inventory(None)
2369
self._new_revision_id = revision_id
2370
self.parents = parents
2371
self.repository = repository
2372
2373
self._revprops = {}
2374
if revprops is not None:
2375
self._revprops.update(revprops)
2376
2377
if timestamp is None:
2378
timestamp = time.time()
2379
# Restrict resolution to 1ms
2380
self._timestamp = round(timestamp, 3)
2381
2382
if timezone is None:
2383
self._timezone = local_time_offset()
2384
else:
2385
self._timezone = int(timezone)
2386
2387
self._generate_revision_if_needed()
2388
2389
def commit(self, message):
2390
"""Make the actual commit.
2391
2392
:return: The revision id of the recorded revision.
2393
"""
2394
rev = _mod_revision.Revision(
2395
timestamp=self._timestamp,
2396
timezone=self._timezone,
2397
committer=self._committer,
2398
message=message,
2399
inventory_sha1=self.inv_sha1,
2400
revision_id=self._new_revision_id,
2401
properties=self._revprops)
2402
rev.parent_ids = self.parents
2403
self.repository.add_revision(self._new_revision_id, rev,
2404
self.new_inventory, self._config)
2405
return self._new_revision_id
2406
2407
def revision_tree(self):
2408
"""Return the tree that was just committed.
2409
2410
After calling commit() this can be called to get a RevisionTree
2411
representing the newly committed tree. This is preferred to
2412
calling Repository.revision_tree() because that may require
2413
deserializing the inventory, while we already have a copy in
2414
memory.
2415
"""
2416
return RevisionTree(self.repository, self.new_inventory,
2417
self._new_revision_id)
2418
2419
def finish_inventory(self):
2420
"""Tell the builder that the inventory is finished."""
2421
if self.new_inventory.root is None:
2422
symbol_versioning.warn('Root entry should be supplied to'
2423
' record_entry_contents, as of bzr 0.10.',
2424
DeprecationWarning, stacklevel=2)
2425
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2426
self.new_inventory.revision_id = self._new_revision_id
2427
self.inv_sha1 = self.repository.add_inventory(
2428
self._new_revision_id,
2429
self.new_inventory,
2430
self.parents
2431
)
2432
2433
def _gen_revision_id(self):
2434
"""Return new revision-id."""
2435
return generate_ids.gen_revision_id(self._config.username(),
2436
self._timestamp)
2437
2438
def _generate_revision_if_needed(self):
2439
"""Create a revision id if None was supplied.
2440
2441
If the repository can not support user-specified revision ids
2442
they should override this function and raise CannotSetRevisionId
2443
if _new_revision_id is not None.
2444
2445
:raises: CannotSetRevisionId
2446
"""
2447
if self._new_revision_id is None:
2448
self._new_revision_id = self._gen_revision_id()
2449
2450
def record_entry_contents(self, ie, parent_invs, path, tree):
2451
"""Record the content of ie from tree into the commit if needed.
2452
2453
Side effect: sets ie.revision when unchanged
2454
2455
:param ie: An inventory entry present in the commit.
2456
:param parent_invs: The inventories of the parent revisions of the
2457
commit.
2458
:param path: The path the entry is at in the tree.
2459
:param tree: The tree which contains this entry and should be used to
2460
obtain content.
2461
"""
2462
if self.new_inventory.root is None and ie.parent_id is not None:
2463
symbol_versioning.warn('Root entry should be supplied to'
2464
' record_entry_contents, as of bzr 0.10.',
2465
DeprecationWarning, stacklevel=2)
2466
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2467
'', tree)
2468
self.new_inventory.add(ie)
2469
2470
# ie.revision is always None if the InventoryEntry is considered
2471
# for committing. ie.snapshot will record the correct revision
2472
# which may be the sole parent if it is untouched.
2473
if ie.revision is not None:
2474
return
2475
2476
# In this revision format, root entries have no knit or weave
2477
if ie is self.new_inventory.root:
2478
# When serializing out to disk and back in
2479
# root.revision is always _new_revision_id
2480
ie.revision = self._new_revision_id
2481
return
2482
previous_entries = ie.find_previous_heads(
2483
parent_invs,
2484
self.repository.weave_store,
2485
self.repository.get_transaction())
2486
# we are creating a new revision for ie in the history store
2487
# and inventory.
2488
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2489
2490
def modified_directory(self, file_id, file_parents):
2491
"""Record the presence of a symbolic link.
2492
2493
:param file_id: The file_id of the link to record.
2494
:param file_parents: The per-file parent revision ids.
2495
"""
2496
self._add_text_to_weave(file_id, [], file_parents.keys())
2497
2498
def modified_file_text(self, file_id, file_parents,
2499
get_content_byte_lines, text_sha1=None,
2500
text_size=None):
2501
"""Record the text of file file_id
2502
2503
:param file_id: The file_id of the file to record the text of.
2504
:param file_parents: The per-file parent revision ids.
2505
:param get_content_byte_lines: A callable which will return the byte
2506
lines for the file.
2507
:param text_sha1: Optional SHA1 of the file contents.
2508
:param text_size: Optional size of the file contents.
2509
"""
2510
# mutter('storing text of file {%s} in revision {%s} into %r',
2511
# file_id, self._new_revision_id, self.repository.weave_store)
2512
# special case to avoid diffing on renames or
2513
# reparenting
2514
if (len(file_parents) == 1
2515
and text_sha1 == file_parents.values()[0].text_sha1
2516
and text_size == file_parents.values()[0].text_size):
2517
previous_ie = file_parents.values()[0]
2518
versionedfile = self.repository.weave_store.get_weave(file_id,
2519
self.repository.get_transaction())
2520
versionedfile.clone_text(self._new_revision_id,
2521
previous_ie.revision, file_parents.keys())
2522
return text_sha1, text_size
2523
else:
2524
new_lines = get_content_byte_lines()
2525
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2526
# should return the SHA1 and size
2527
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2528
return osutils.sha_strings(new_lines), \
2529
sum(map(len, new_lines))
2530
2531
def modified_link(self, file_id, file_parents, link_target):
2532
"""Record the presence of a symbolic link.
2533
2534
:param file_id: The file_id of the link to record.
2535
:param file_parents: The per-file parent revision ids.
2536
:param link_target: Target location of this link.
2537
"""
2538
self._add_text_to_weave(file_id, [], file_parents.keys())
2539
2540
def _add_text_to_weave(self, file_id, new_lines, parents):
2541
versionedfile = self.repository.weave_store.get_weave_or_empty(
2542
file_id, self.repository.get_transaction())
2543
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2544
versionedfile.clear_cache()
2545
2546
2547
class _CommitBuilder(CommitBuilder):
2548
"""Temporary class so old CommitBuilders are detected properly
2549
2550
Note: CommitBuilder works whether or not root entry is recorded.
2551
"""
2552
2553
record_root_entry = True
2554
2555
2556
class RootCommitBuilder(CommitBuilder):
2557
"""This commitbuilder actually records the root id"""
2558
2559
record_root_entry = True
2560
2561
def record_entry_contents(self, ie, parent_invs, path, tree):
2562
"""Record the content of ie from tree into the commit if needed.
2563
2564
Side effect: sets ie.revision when unchanged
2565
2566
:param ie: An inventory entry present in the commit.
2567
:param parent_invs: The inventories of the parent revisions of the
2568
commit.
2569
:param path: The path the entry is at in the tree.
2570
:param tree: The tree which contains this entry and should be used to
2571
obtain content.
2572
"""
2573
assert self.new_inventory.root is not None or ie.parent_id is None
2574
self.new_inventory.add(ie)
2575
2576
# ie.revision is always None if the InventoryEntry is considered
2577
# for committing. ie.snapshot will record the correct revision
2578
# which may be the sole parent if it is untouched.
2579
if ie.revision is not None:
2580
return
2581
2582
previous_entries = ie.find_previous_heads(
2583
parent_invs,
2584
self.repository.weave_store,
2585
self.repository.get_transaction())
2586
# we are creating a new revision for ie in the history store
2587
# and inventory.
2588
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2589
2590
2591
_unescape_map = {
2592
'apos':"'",
2593
'quot':'"',
2594
'amp':'&',
2595
'lt':'<',
2596
'gt':'>'
2597
}
2598
2599
2600
def _unescaper(match, _map=_unescape_map):
2601
return _map[match.group(1)]
2602
2603
2604
_unescape_re = None
2605
2606
2607
def _unescape_xml(data):
2608
"""Unescape predefined XML entities in a string of data."""
2609
global _unescape_re
2610
if _unescape_re is None:
2611
_unescape_re = re.compile('\&([^;]*);')
2612
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0