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 0.200.1036
- compare with another revision
- download diff
- download tarball
- view history from revision 0.200.1036
- Committer: Jelmer Vernooij
- Date: 2010-09-11 17:48:45 UTC
- mto: (0.312.1 master) (6883.23.1 bundle-git)
- mto: This revision was merged to the branch mainline in revision 6960.
- Revision ID: jelmer@samba.org-20100911174845-ro06bb6gg6ws8jit
More work on roundtrip push support.
- files added:
- .bzrignore
- notes
- tests
- files removed:
- .bzrignore
- bzrlib
- bzrlib/benchmarks
- bzrlib/benchmarks/tree_creator
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/doc
- bzrlib/doc/api
- bzrlib/export
- bzrlib/plugins
- bzrlib/plugins/launchpad
- bzrlib/repofmt
- 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
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
2
#
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
7
#
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
12
#
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
17
from cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
23
import re
24
import time
25
import unittest
26
27
from bzrlib import (
28
bzrdir,
29
check,
30
delta,
31
errors,
32
generate_ids,
33
gpg,
34
graph,
35
knit,
36
lazy_regex,
37
lockable_files,
38
lockdir,
39
osutils,
40
registry,
41
revision as _mod_revision,
42
symbol_versioning,
43
transactions,
44
ui,
45
weave,
46
weavefile,
47
xml5,
48
xml6,
49
)
50
from bzrlib.osutils import (
51
rand_bytes,
52
compact_date,
53
local_time_offset,
54
)
55
from bzrlib.revisiontree import RevisionTree
56
from bzrlib.store.versioned import VersionedFileStore
57
from bzrlib.store.text import TextStore
58
from bzrlib.testament import Testament
59
""")
60
61
from bzrlib.decorators import needs_read_lock, needs_write_lock
62
from bzrlib.inter import InterObject
63
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
64
from bzrlib.symbol_versioning import (
65
deprecated_method,
66
zero_nine,
67
)
68
from bzrlib.trace import mutter, note, warning
69
70
71
# Old formats display a warning, but only once
72
_deprecation_warning_done = False
73
74
75
class Repository(object):
76
"""Repository holding history for one or more branches.
77
78
The repository holds and retrieves historical information including
79
revisions and file history. It's normally accessed only by the Branch,
80
which views a particular line of development through that history.
81
82
The Repository builds on top of Stores and a Transport, which respectively
83
describe the disk data format and the way of accessing the (possibly
84
remote) disk.
85
"""
86
87
_file_ids_altered_regex = lazy_regex.lazy_compile(
88
r'file_id="(?P<file_id>[^"]+)"'
89
r'.*revision="(?P<revision_id>[^"]+)"'
90
)
91
92
@needs_write_lock
93
def add_inventory(self, revid, inv, parents):
94
"""Add the inventory inv to the repository as revid.
95
96
:param parents: The revision ids of the parents that revid
97
is known to have and are in the repository already.
98
99
returns the sha1 of the serialized inventory.
100
"""
101
_mod_revision.check_not_reserved_id(revid)
102
assert inv.revision_id is None or inv.revision_id == revid, \
103
"Mismatch between inventory revision" \
104
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
105
assert inv.root is not None
106
inv_text = self.serialise_inventory(inv)
107
inv_sha1 = osutils.sha_string(inv_text)
108
inv_vf = self.control_weaves.get_weave('inventory',
109
self.get_transaction())
110
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
111
return inv_sha1
112
113
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
114
final_parents = []
115
for parent in parents:
116
if parent in inv_vf:
117
final_parents.append(parent)
118
119
inv_vf.add_lines(revid, final_parents, lines)
120
121
@needs_write_lock
122
def add_revision(self, rev_id, rev, inv=None, config=None):
123
"""Add rev to the revision store as rev_id.
124
125
:param rev_id: the revision id to use.
126
:param rev: The revision object.
127
:param inv: The inventory for the revision. if None, it will be looked
128
up in the inventory storer
129
:param config: If None no digital signature will be created.
130
If supplied its signature_needed method will be used
131
to determine if a signature should be made.
132
"""
133
_mod_revision.check_not_reserved_id(rev_id)
134
if config is not None and config.signature_needed():
135
if inv is None:
136
inv = self.get_inventory(rev_id)
137
plaintext = Testament(rev, inv).as_short_text()
138
self.store_revision_signature(
139
gpg.GPGStrategy(config), plaintext, rev_id)
140
if not rev_id in self.get_inventory_weave():
141
if inv is None:
142
raise errors.WeaveRevisionNotPresent(rev_id,
143
self.get_inventory_weave())
144
else:
145
# yes, this is not suitable for adding with ghosts.
146
self.add_inventory(rev_id, inv, rev.parent_ids)
147
self._revision_store.add_revision(rev, self.get_transaction())
148
149
@needs_read_lock
150
def _all_possible_ids(self):
151
"""Return all the possible revisions that we could find."""
152
return self.get_inventory_weave().versions()
153
154
def all_revision_ids(self):
155
"""Returns a list of all the revision ids in the repository.
156
157
This is deprecated because code should generally work on the graph
158
reachable from a particular revision, and ignore any other revisions
159
that might be present. There is no direct replacement method.
160
"""
161
return self._all_revision_ids()
162
163
@needs_read_lock
164
def _all_revision_ids(self):
165
"""Returns a list of all the revision ids in the repository.
166
167
These are in as much topological order as the underlying store can
168
present: for weaves ghosts may lead to a lack of correctness until
169
the reweave updates the parents list.
170
"""
171
if self._revision_store.text_store.listable():
172
return self._revision_store.all_revision_ids(self.get_transaction())
173
result = self._all_possible_ids()
174
return self._eliminate_revisions_not_present(result)
175
176
def break_lock(self):
177
"""Break a lock if one is present from another instance.
178
179
Uses the ui factory to ask for confirmation if the lock may be from
180
an active process.
181
"""
182
self.control_files.break_lock()
183
184
@needs_read_lock
185
def _eliminate_revisions_not_present(self, revision_ids):
186
"""Check every revision id in revision_ids to see if we have it.
187
188
Returns a set of the present revisions.
189
"""
190
result = []
191
for id in revision_ids:
192
if self.has_revision(id):
193
result.append(id)
194
return result
195
196
@staticmethod
197
def create(a_bzrdir):
198
"""Construct the current default format repository in a_bzrdir."""
199
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
200
201
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
202
"""instantiate a Repository.
203
204
:param _format: The format of the repository on disk.
205
:param a_bzrdir: The BzrDir of the repository.
206
207
In the future we will have a single api for all stores for
208
getting file texts, inventories and revisions, then
209
this construct will accept instances of those things.
210
"""
211
super(Repository, self).__init__()
212
self._format = _format
213
# the following are part of the public API for Repository:
214
self.bzrdir = a_bzrdir
215
self.control_files = control_files
216
self._revision_store = _revision_store
217
self.text_store = text_store
218
# backwards compatibility
219
self.weave_store = text_store
220
# not right yet - should be more semantically clear ?
221
#
222
self.control_store = control_store
223
self.control_weaves = control_store
224
# TODO: make sure to construct the right store classes, etc, depending
225
# on whether escaping is required.
226
self._warn_if_deprecated()
227
self._serializer = xml5.serializer_v5
228
229
def __repr__(self):
230
return '%s(%r)' % (self.__class__.__name__,
231
self.bzrdir.transport.base)
232
233
def is_locked(self):
234
return self.control_files.is_locked()
235
236
def lock_write(self):
237
self.control_files.lock_write()
238
239
def lock_read(self):
240
self.control_files.lock_read()
241
242
def get_physical_lock_status(self):
243
return self.control_files.get_physical_lock_status()
244
245
@needs_read_lock
246
def missing_revision_ids(self, other, revision_id=None):
247
"""Return the revision ids that other has that this does not.
248
249
These are returned in topological order.
250
251
revision_id: only return revision ids included by revision_id.
252
"""
253
return InterRepository.get(other, self).missing_revision_ids(revision_id)
254
255
@staticmethod
256
def open(base):
257
"""Open the repository rooted at base.
258
259
For instance, if the repository is at URL/.bzr/repository,
260
Repository.open(URL) -> a Repository instance.
261
"""
262
control = bzrdir.BzrDir.open(base)
263
return control.open_repository()
264
265
def copy_content_into(self, destination, revision_id=None, basis=None):
266
"""Make a complete copy of the content in self into destination.
267
268
This is a destructive operation! Do not use it on existing
269
repositories.
270
"""
271
return InterRepository.get(self, destination).copy_content(revision_id, basis)
272
273
def fetch(self, source, revision_id=None, pb=None):
274
"""Fetch the content required to construct revision_id from source.
275
276
If revision_id is None all content is copied.
277
"""
278
return InterRepository.get(source, self).fetch(revision_id=revision_id,
279
pb=pb)
280
281
def get_commit_builder(self, branch, parents, config, timestamp=None,
282
timezone=None, committer=None, revprops=None,
283
revision_id=None):
284
"""Obtain a CommitBuilder for this repository.
285
286
:param branch: Branch to commit to.
287
:param parents: Revision ids of the parents of the new revision.
288
:param config: Configuration to use.
289
:param timestamp: Optional timestamp recorded for commit.
290
:param timezone: Optional timezone for timestamp.
291
:param committer: Optional committer to set for commit.
292
:param revprops: Optional dictionary of revision properties.
293
:param revision_id: Optional revision id.
294
"""
295
return _CommitBuilder(self, parents, config, timestamp, timezone,
296
committer, revprops, revision_id)
297
298
def unlock(self):
299
self.control_files.unlock()
300
301
@needs_read_lock
302
def clone(self, a_bzrdir, revision_id=None, basis=None):
303
"""Clone this repository into a_bzrdir using the current format.
304
305
Currently no check is made that the format of this repository and
306
the bzrdir format are compatible. FIXME RBC 20060201.
307
"""
308
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
309
# use target default format.
310
result = a_bzrdir.create_repository()
311
# FIXME RBC 20060209 split out the repository type to avoid this check ?
312
elif isinstance(a_bzrdir._format,
313
(bzrdir.BzrDirFormat4,
314
bzrdir.BzrDirFormat5,
315
bzrdir.BzrDirFormat6)):
316
result = a_bzrdir.open_repository()
317
else:
318
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
319
self.copy_content_into(result, revision_id, basis)
320
return result
321
322
@needs_read_lock
323
def has_revision(self, revision_id):
324
"""True if this repository has a copy of the revision."""
325
return self._revision_store.has_revision_id(revision_id,
326
self.get_transaction())
327
328
@needs_read_lock
329
def get_revision_reconcile(self, revision_id):
330
"""'reconcile' helper routine that allows access to a revision always.
331
332
This variant of get_revision does not cross check the weave graph
333
against the revision one as get_revision does: but it should only
334
be used by reconcile, or reconcile-alike commands that are correcting
335
or testing the revision graph.
336
"""
337
if not revision_id or not isinstance(revision_id, basestring):
338
raise errors.InvalidRevisionId(revision_id=revision_id,
339
branch=self)
340
return self._revision_store.get_revisions([revision_id],
341
self.get_transaction())[0]
342
@needs_read_lock
343
def get_revisions(self, revision_ids):
344
return self._revision_store.get_revisions(revision_ids,
345
self.get_transaction())
346
347
@needs_read_lock
348
def get_revision_xml(self, revision_id):
349
rev = self.get_revision(revision_id)
350
rev_tmp = StringIO()
351
# the current serializer..
352
self._revision_store._serializer.write_revision(rev, rev_tmp)
353
rev_tmp.seek(0)
354
return rev_tmp.getvalue()
355
356
@needs_read_lock
357
def get_revision(self, revision_id):
358
"""Return the Revision object for a named revision"""
359
r = self.get_revision_reconcile(revision_id)
360
# weave corruption can lead to absent revision markers that should be
361
# present.
362
# the following test is reasonably cheap (it needs a single weave read)
363
# and the weave is cached in read transactions. In write transactions
364
# it is not cached but typically we only read a small number of
365
# revisions. For knits when they are introduced we will probably want
366
# to ensure that caching write transactions are in use.
367
inv = self.get_inventory_weave()
368
self._check_revision_parents(r, inv)
369
return r
370
371
@needs_read_lock
372
def get_deltas_for_revisions(self, revisions):
373
"""Produce a generator of revision deltas.
374
375
Note that the input is a sequence of REVISIONS, not revision_ids.
376
Trees will be held in memory until the generator exits.
377
Each delta is relative to the revision's lefthand predecessor.
378
"""
379
required_trees = set()
380
for revision in revisions:
381
required_trees.add(revision.revision_id)
382
required_trees.update(revision.parent_ids[:1])
383
trees = dict((t.get_revision_id(), t) for
384
t in self.revision_trees(required_trees))
385
for revision in revisions:
386
if not revision.parent_ids:
387
old_tree = self.revision_tree(None)
388
else:
389
old_tree = trees[revision.parent_ids[0]]
390
yield trees[revision.revision_id].changes_from(old_tree)
391
392
@needs_read_lock
393
def get_revision_delta(self, revision_id):
394
"""Return the delta for one revision.
395
396
The delta is relative to the left-hand predecessor of the
397
revision.
398
"""
399
r = self.get_revision(revision_id)
400
return list(self.get_deltas_for_revisions([r]))[0]
401
402
def _check_revision_parents(self, revision, inventory):
403
"""Private to Repository and Fetch.
404
405
This checks the parentage of revision in an inventory weave for
406
consistency and is only applicable to inventory-weave-for-ancestry
407
using repository formats & fetchers.
408
"""
409
weave_parents = inventory.get_parents(revision.revision_id)
410
weave_names = inventory.versions()
411
for parent_id in revision.parent_ids:
412
if parent_id in weave_names:
413
# this parent must not be a ghost.
414
if not parent_id in weave_parents:
415
# but it is a ghost
416
raise errors.CorruptRepository(self)
417
418
@needs_write_lock
419
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
420
signature = gpg_strategy.sign(plaintext)
421
self._revision_store.add_revision_signature_text(revision_id,
422
signature,
423
self.get_transaction())
424
425
def fileids_altered_by_revision_ids(self, revision_ids):
426
"""Find the file ids and versions affected by revisions.
427
428
:param revisions: an iterable containing revision ids.
429
:return: a dictionary mapping altered file-ids to an iterable of
430
revision_ids. Each altered file-ids has the exact revision_ids that
431
altered it listed explicitly.
432
"""
433
assert self._serializer.support_altered_by_hack, \
434
("fileids_altered_by_revision_ids only supported for branches "
435
"which store inventory as unnested xml, not on %r" % self)
436
selected_revision_ids = set(revision_ids)
437
w = self.get_inventory_weave()
438
result = {}
439
440
# this code needs to read every new line in every inventory for the
441
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
442
# not present in one of those inventories is unnecessary but not
443
# harmful because we are filtering by the revision id marker in the
444
# inventory lines : we only select file ids altered in one of those
445
# revisions. We don't need to see all lines in the inventory because
446
# only those added in an inventory in rev X can contain a revision=X
447
# line.
448
unescape_revid_cache = {}
449
unescape_fileid_cache = {}
450
451
# jam 20061218 In a big fetch, this handles hundreds of thousands
452
# of lines, so it has had a lot of inlining and optimizing done.
453
# Sorry that it is a little bit messy.
454
# Move several functions to be local variables, since this is a long
455
# running loop.
456
search = self._file_ids_altered_regex.search
457
unescape = _unescape_xml
458
setdefault = result.setdefault
459
pb = ui.ui_factory.nested_progress_bar()
460
try:
461
for line in w.iter_lines_added_or_present_in_versions(
462
selected_revision_ids, pb=pb):
463
match = search(line)
464
if match is None:
465
continue
466
# One call to match.group() returning multiple items is quite a
467
# bit faster than 2 calls to match.group() each returning 1
468
file_id, revision_id = match.group('file_id', 'revision_id')
469
470
# Inlining the cache lookups helps a lot when you make 170,000
471
# lines and 350k ids, versus 8.4 unique ids.
472
# Using a cache helps in 2 ways:
473
# 1) Avoids unnecessary decoding calls
474
# 2) Re-uses cached strings, which helps in future set and
475
# equality checks.
476
# (2) is enough that removing encoding entirely along with
477
# the cache (so we are using plain strings) results in no
478
# performance improvement.
479
try:
480
revision_id = unescape_revid_cache[revision_id]
481
except KeyError:
482
unescaped = unescape(revision_id)
483
unescape_revid_cache[revision_id] = unescaped
484
revision_id = unescaped
485
486
if revision_id in selected_revision_ids:
487
try:
488
file_id = unescape_fileid_cache[file_id]
489
except KeyError:
490
unescaped = unescape(file_id)
491
unescape_fileid_cache[file_id] = unescaped
492
file_id = unescaped
493
setdefault(file_id, set()).add(revision_id)
494
finally:
495
pb.finished()
496
return result
497
498
@needs_read_lock
499
def get_inventory_weave(self):
500
return self.control_weaves.get_weave('inventory',
501
self.get_transaction())
502
503
@needs_read_lock
504
def get_inventory(self, revision_id):
505
"""Get Inventory object by hash."""
506
return self.deserialise_inventory(
507
revision_id, self.get_inventory_xml(revision_id))
508
509
def deserialise_inventory(self, revision_id, xml):
510
"""Transform the xml into an inventory object.
511
512
:param revision_id: The expected revision id of the inventory.
513
:param xml: A serialised inventory.
514
"""
515
result = self._serializer.read_inventory_from_string(xml)
516
result.root.revision = revision_id
517
return result
518
519
def serialise_inventory(self, inv):
520
return self._serializer.write_inventory_to_string(inv)
521
522
@needs_read_lock
523
def get_inventory_xml(self, revision_id):
524
"""Get inventory XML as a file object."""
525
try:
526
assert isinstance(revision_id, basestring), type(revision_id)
527
iw = self.get_inventory_weave()
528
return iw.get_text(revision_id)
529
except IndexError:
530
raise errors.HistoryMissing(self, 'inventory', revision_id)
531
532
@needs_read_lock
533
def get_inventory_sha1(self, revision_id):
534
"""Return the sha1 hash of the inventory entry
535
"""
536
return self.get_revision(revision_id).inventory_sha1
537
538
@needs_read_lock
539
def get_revision_graph(self, revision_id=None):
540
"""Return a dictionary containing the revision graph.
541
542
:param revision_id: The revision_id to get a graph from. If None, then
543
the entire revision graph is returned. This is a deprecated mode of
544
operation and will be removed in the future.
545
:return: a dictionary of revision_id->revision_parents_list.
546
"""
547
# special case NULL_REVISION
548
if revision_id == _mod_revision.NULL_REVISION:
549
return {}
550
a_weave = self.get_inventory_weave()
551
all_revisions = self._eliminate_revisions_not_present(
552
a_weave.versions())
553
entire_graph = dict([(node, a_weave.get_parents(node)) for
554
node in all_revisions])
555
if revision_id is None:
556
return entire_graph
557
elif revision_id not in entire_graph:
558
raise errors.NoSuchRevision(self, revision_id)
559
else:
560
# add what can be reached from revision_id
561
result = {}
562
pending = set([revision_id])
563
while len(pending) > 0:
564
node = pending.pop()
565
result[node] = entire_graph[node]
566
for revision_id in result[node]:
567
if revision_id not in result:
568
pending.add(revision_id)
569
return result
570
571
@needs_read_lock
572
def get_revision_graph_with_ghosts(self, revision_ids=None):
573
"""Return a graph of the revisions with ghosts marked as applicable.
574
575
:param revision_ids: an iterable of revisions to graph or None for all.
576
:return: a Graph object with the graph reachable from revision_ids.
577
"""
578
result = graph.Graph()
579
if not revision_ids:
580
pending = set(self.all_revision_ids())
581
required = set([])
582
else:
583
pending = set(revision_ids)
584
# special case NULL_REVISION
585
if _mod_revision.NULL_REVISION in pending:
586
pending.remove(_mod_revision.NULL_REVISION)
587
required = set(pending)
588
done = set([])
589
while len(pending):
590
revision_id = pending.pop()
591
try:
592
rev = self.get_revision(revision_id)
593
except errors.NoSuchRevision:
594
if revision_id in required:
595
raise
596
# a ghost
597
result.add_ghost(revision_id)
598
continue
599
for parent_id in rev.parent_ids:
600
# is this queued or done ?
601
if (parent_id not in pending and
602
parent_id not in done):
603
# no, queue it.
604
pending.add(parent_id)
605
result.add_node(revision_id, rev.parent_ids)
606
done.add(revision_id)
607
return result
608
609
@needs_read_lock
610
def get_revision_inventory(self, revision_id):
611
"""Return inventory of a past revision."""
612
# TODO: Unify this with get_inventory()
613
# bzr 0.0.6 and later imposes the constraint that the inventory_id
614
# must be the same as its revision, so this is trivial.
615
if revision_id is None:
616
# This does not make sense: if there is no revision,
617
# then it is the current tree inventory surely ?!
618
# and thus get_root_id() is something that looks at the last
619
# commit on the branch, and the get_root_id is an inventory check.
620
raise NotImplementedError
621
# return Inventory(self.get_root_id())
622
else:
623
return self.get_inventory(revision_id)
624
625
@needs_read_lock
626
def is_shared(self):
627
"""Return True if this repository is flagged as a shared repository."""
628
raise NotImplementedError(self.is_shared)
629
630
@needs_write_lock
631
def reconcile(self, other=None, thorough=False):
632
"""Reconcile this repository."""
633
from bzrlib.reconcile import RepoReconciler
634
reconciler = RepoReconciler(self, thorough=thorough)
635
reconciler.reconcile()
636
return reconciler
637
638
@needs_read_lock
639
def revision_tree(self, revision_id):
640
"""Return Tree for a revision on this branch.
641
642
`revision_id` may be None for the empty tree revision.
643
"""
644
# TODO: refactor this to use an existing revision object
645
# so we don't need to read it in twice.
646
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
647
return RevisionTree(self, Inventory(root_id=None),
648
_mod_revision.NULL_REVISION)
649
else:
650
inv = self.get_revision_inventory(revision_id)
651
return RevisionTree(self, inv, revision_id)
652
653
@needs_read_lock
654
def revision_trees(self, revision_ids):
655
"""Return Tree for a revision on this branch.
656
657
`revision_id` may not be None or 'null:'"""
658
assert None not in revision_ids
659
assert _mod_revision.NULL_REVISION not in revision_ids
660
texts = self.get_inventory_weave().get_texts(revision_ids)
661
for text, revision_id in zip(texts, revision_ids):
662
inv = self.deserialise_inventory(revision_id, text)
663
yield RevisionTree(self, inv, revision_id)
664
665
@needs_read_lock
666
def get_ancestry(self, revision_id):
667
"""Return a list of revision-ids integrated by a revision.
668
669
The first element of the list is always None, indicating the origin
670
revision. This might change when we have history horizons, or
671
perhaps we should have a new API.
672
673
This is topologically sorted.
674
"""
675
if revision_id is None:
676
return [None]
677
if not self.has_revision(revision_id):
678
raise errors.NoSuchRevision(self, revision_id)
679
w = self.get_inventory_weave()
680
candidates = w.get_ancestry(revision_id)
681
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
682
683
@needs_read_lock
684
def print_file(self, file, revision_id):
685
"""Print `file` to stdout.
686
687
FIXME RBC 20060125 as John Meinel points out this is a bad api
688
- it writes to stdout, it assumes that that is valid etc. Fix
689
by creating a new more flexible convenience function.
690
"""
691
tree = self.revision_tree(revision_id)
692
# use inventory as it was in that revision
693
file_id = tree.inventory.path2id(file)
694
if not file_id:
695
# TODO: jam 20060427 Write a test for this code path
696
# it had a bug in it, and was raising the wrong
697
# exception.
698
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
699
tree.print_file(file_id)
700
701
def get_transaction(self):
702
return self.control_files.get_transaction()
703
704
def revision_parents(self, revid):
705
return self.get_inventory_weave().parent_names(revid)
706
707
@needs_write_lock
708
def set_make_working_trees(self, new_value):
709
"""Set the policy flag for making working trees when creating branches.
710
711
This only applies to branches that use this repository.
712
713
The default is 'True'.
714
:param new_value: True to restore the default, False to disable making
715
working trees.
716
"""
717
raise NotImplementedError(self.set_make_working_trees)
718
719
def make_working_trees(self):
720
"""Returns the policy for making working trees on new branches."""
721
raise NotImplementedError(self.make_working_trees)
722
723
@needs_write_lock
724
def sign_revision(self, revision_id, gpg_strategy):
725
plaintext = Testament.from_revision(self, revision_id).as_short_text()
726
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
727
728
@needs_read_lock
729
def has_signature_for_revision_id(self, revision_id):
730
"""Query for a revision signature for revision_id in the repository."""
731
return self._revision_store.has_signature(revision_id,
732
self.get_transaction())
733
734
@needs_read_lock
735
def get_signature_text(self, revision_id):
736
"""Return the text for a signature."""
737
return self._revision_store.get_signature_text(revision_id,
738
self.get_transaction())
739
740
@needs_read_lock
741
def check(self, revision_ids):
742
"""Check consistency of all history of given revision_ids.
743
744
Different repository implementations should override _check().
745
746
:param revision_ids: A non-empty list of revision_ids whose ancestry
747
will be checked. Typically the last revision_id of a branch.
748
"""
749
if not revision_ids:
750
raise ValueError("revision_ids must be non-empty in %s.check"
751
% (self,))
752
return self._check(revision_ids)
753
754
def _check(self, revision_ids):
755
result = check.Check(self)
756
result.check()
757
return result
758
759
def _warn_if_deprecated(self):
760
global _deprecation_warning_done
761
if _deprecation_warning_done:
762
return
763
_deprecation_warning_done = True
764
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
765
% (self._format, self.bzrdir.transport.base))
766
767
def supports_rich_root(self):
768
return self._format.rich_root_data
769
770
def _check_ascii_revisionid(self, revision_id, method):
771
"""Private helper for ascii-only repositories."""
772
# weave repositories refuse to store revisionids that are non-ascii.
773
if revision_id is not None:
774
# weaves require ascii revision ids.
775
if isinstance(revision_id, unicode):
776
try:
777
revision_id.encode('ascii')
778
except UnicodeEncodeError:
779
raise errors.NonAsciiRevisionId(method, self)
780
781
782
class AllInOneRepository(Repository):
783
"""Legacy support - the repository behaviour for all-in-one branches."""
784
785
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
786
# we reuse one control files instance.
787
dir_mode = a_bzrdir._control_files._dir_mode
788
file_mode = a_bzrdir._control_files._file_mode
789
790
def get_store(name, compressed=True, prefixed=False):
791
# FIXME: This approach of assuming stores are all entirely compressed
792
# or entirely uncompressed is tidy, but breaks upgrade from
793
# some existing branches where there's a mixture; we probably
794
# still want the option to look for both.
795
relpath = a_bzrdir._control_files._escape(name)
796
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
797
prefixed=prefixed, compressed=compressed,
798
dir_mode=dir_mode,
799
file_mode=file_mode)
800
#if self._transport.should_cache():
801
# cache_path = os.path.join(self.cache_root, name)
802
# os.mkdir(cache_path)
803
# store = bzrlib.store.CachedStore(store, cache_path)
804
return store
805
806
# not broken out yet because the controlweaves|inventory_store
807
# and text_store | weave_store bits are still different.
808
if isinstance(_format, RepositoryFormat4):
809
# cannot remove these - there is still no consistent api
810
# which allows access to this old info.
811
self.inventory_store = get_store('inventory-store')
812
text_store = get_store('text-store')
813
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
814
815
def get_commit_builder(self, branch, parents, config, timestamp=None,
816
timezone=None, committer=None, revprops=None,
817
revision_id=None):
818
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
819
return Repository.get_commit_builder(self, branch, parents, config,
820
timestamp, timezone, committer, revprops, revision_id)
821
822
@needs_read_lock
823
def is_shared(self):
824
"""AllInOne repositories cannot be shared."""
825
return False
826
827
@needs_write_lock
828
def set_make_working_trees(self, new_value):
829
"""Set the policy flag for making working trees when creating branches.
830
831
This only applies to branches that use this repository.
832
833
The default is 'True'.
834
:param new_value: True to restore the default, False to disable making
835
working trees.
836
"""
837
raise NotImplementedError(self.set_make_working_trees)
838
839
def make_working_trees(self):
840
"""Returns the policy for making working trees on new branches."""
841
return True
842
843
844
def install_revision(repository, rev, revision_tree):
845
"""Install all revision data into a repository."""
846
present_parents = []
847
parent_trees = {}
848
for p_id in rev.parent_ids:
849
if repository.has_revision(p_id):
850
present_parents.append(p_id)
851
parent_trees[p_id] = repository.revision_tree(p_id)
852
else:
853
parent_trees[p_id] = repository.revision_tree(None)
854
855
inv = revision_tree.inventory
856
entries = inv.iter_entries()
857
# backwards compatability hack: skip the root id.
858
if not repository.supports_rich_root():
859
path, root = entries.next()
860
if root.revision != rev.revision_id:
861
raise errors.IncompatibleRevision(repr(repository))
862
# Add the texts that are not already present
863
for path, ie in entries:
864
w = repository.weave_store.get_weave_or_empty(ie.file_id,
865
repository.get_transaction())
866
if ie.revision not in w:
867
text_parents = []
868
# FIXME: TODO: The following loop *may* be overlapping/duplicate
869
# with InventoryEntry.find_previous_heads(). if it is, then there
870
# is a latent bug here where the parents may have ancestors of each
871
# other. RBC, AB
872
for revision, tree in parent_trees.iteritems():
873
if ie.file_id not in tree:
874
continue
875
parent_id = tree.inventory[ie.file_id].revision
876
if parent_id in text_parents:
877
continue
878
text_parents.append(parent_id)
879
880
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
881
repository.get_transaction())
882
lines = revision_tree.get_file(ie.file_id).readlines()
883
vfile.add_lines(rev.revision_id, text_parents, lines)
884
try:
885
# install the inventory
886
repository.add_inventory(rev.revision_id, inv, present_parents)
887
except errors.RevisionAlreadyPresent:
888
pass
889
repository.add_revision(rev.revision_id, rev, inv)
890
891
892
class MetaDirRepository(Repository):
893
"""Repositories in the new meta-dir layout."""
894
895
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
896
super(MetaDirRepository, self).__init__(_format,
897
a_bzrdir,
898
control_files,
899
_revision_store,
900
control_store,
901
text_store)
902
dir_mode = self.control_files._dir_mode
903
file_mode = self.control_files._file_mode
904
905
@needs_read_lock
906
def is_shared(self):
907
"""Return True if this repository is flagged as a shared repository."""
908
return self.control_files._transport.has('shared-storage')
909
910
@needs_write_lock
911
def set_make_working_trees(self, new_value):
912
"""Set the policy flag for making working trees when creating branches.
913
914
This only applies to branches that use this repository.
915
916
The default is 'True'.
917
:param new_value: True to restore the default, False to disable making
918
working trees.
919
"""
920
if new_value:
921
try:
922
self.control_files._transport.delete('no-working-trees')
923
except errors.NoSuchFile:
924
pass
925
else:
926
self.control_files.put_utf8('no-working-trees', '')
927
928
def make_working_trees(self):
929
"""Returns the policy for making working trees on new branches."""
930
return not self.control_files._transport.has('no-working-trees')
931
932
933
class WeaveMetaDirRepository(MetaDirRepository):
934
"""A subclass of MetaDirRepository to set weave specific policy."""
935
936
def get_commit_builder(self, branch, parents, config, timestamp=None,
937
timezone=None, committer=None, revprops=None,
938
revision_id=None):
939
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
940
return MetaDirRepository.get_commit_builder(self, branch, parents,
941
config, timestamp, timezone, committer, revprops, revision_id)
942
943
944
class KnitRepository(MetaDirRepository):
945
"""Knit format repository."""
946
947
def _warn_if_deprecated(self):
948
# This class isn't deprecated
949
pass
950
951
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
952
inv_vf.add_lines_with_ghosts(revid, parents, lines)
953
954
@needs_read_lock
955
def _all_revision_ids(self):
956
"""See Repository.all_revision_ids()."""
957
# Knits get the revision graph from the index of the revision knit, so
958
# it's always possible even if they're on an unlistable transport.
959
return self._revision_store.all_revision_ids(self.get_transaction())
960
961
def fileid_involved_between_revs(self, from_revid, to_revid):
962
"""Find file_id(s) which are involved in the changes between revisions.
963
964
This determines the set of revisions which are involved, and then
965
finds all file ids affected by those revisions.
966
"""
967
vf = self._get_revision_vf()
968
from_set = set(vf.get_ancestry(from_revid))
969
to_set = set(vf.get_ancestry(to_revid))
970
changed = to_set.difference(from_set)
971
return self._fileid_involved_by_set(changed)
972
973
def fileid_involved(self, last_revid=None):
974
"""Find all file_ids modified in the ancestry of last_revid.
975
976
:param last_revid: If None, last_revision() will be used.
977
"""
978
if not last_revid:
979
changed = set(self.all_revision_ids())
980
else:
981
changed = set(self.get_ancestry(last_revid))
982
if None in changed:
983
changed.remove(None)
984
return self._fileid_involved_by_set(changed)
985
986
@needs_read_lock
987
def get_ancestry(self, revision_id):
988
"""Return a list of revision-ids integrated by a revision.
989
990
This is topologically sorted.
991
"""
992
if revision_id is None:
993
return [None]
994
vf = self._get_revision_vf()
995
try:
996
return [None] + vf.get_ancestry(revision_id)
997
except errors.RevisionNotPresent:
998
raise errors.NoSuchRevision(self, revision_id)
999
1000
@needs_read_lock
1001
def get_revision(self, revision_id):
1002
"""Return the Revision object for a named revision"""
1003
return self.get_revision_reconcile(revision_id)
1004
1005
@needs_read_lock
1006
def get_revision_graph(self, revision_id=None):
1007
"""Return a dictionary containing the revision graph.
1008
1009
:param revision_id: The revision_id to get a graph from. If None, then
1010
the entire revision graph is returned. This is a deprecated mode of
1011
operation and will be removed in the future.
1012
:return: a dictionary of revision_id->revision_parents_list.
1013
"""
1014
# special case NULL_REVISION
1015
if revision_id == _mod_revision.NULL_REVISION:
1016
return {}
1017
a_weave = self._get_revision_vf()
1018
entire_graph = a_weave.get_graph()
1019
if revision_id is None:
1020
return a_weave.get_graph()
1021
elif revision_id not in a_weave:
1022
raise errors.NoSuchRevision(self, revision_id)
1023
else:
1024
# add what can be reached from revision_id
1025
result = {}
1026
pending = set([revision_id])
1027
while len(pending) > 0:
1028
node = pending.pop()
1029
result[node] = a_weave.get_parents(node)
1030
for revision_id in result[node]:
1031
if revision_id not in result:
1032
pending.add(revision_id)
1033
return result
1034
1035
@needs_read_lock
1036
def get_revision_graph_with_ghosts(self, revision_ids=None):
1037
"""Return a graph of the revisions with ghosts marked as applicable.
1038
1039
:param revision_ids: an iterable of revisions to graph or None for all.
1040
:return: a Graph object with the graph reachable from revision_ids.
1041
"""
1042
result = graph.Graph()
1043
vf = self._get_revision_vf()
1044
versions = set(vf.versions())
1045
if not revision_ids:
1046
pending = set(self.all_revision_ids())
1047
required = set([])
1048
else:
1049
pending = set(revision_ids)
1050
# special case NULL_REVISION
1051
if _mod_revision.NULL_REVISION in pending:
1052
pending.remove(_mod_revision.NULL_REVISION)
1053
required = set(pending)
1054
done = set([])
1055
while len(pending):
1056
revision_id = pending.pop()
1057
if not revision_id in versions:
1058
if revision_id in required:
1059
raise errors.NoSuchRevision(self, revision_id)
1060
# a ghost
1061
result.add_ghost(revision_id)
1062
# mark it as done so we don't try for it again.
1063
done.add(revision_id)
1064
continue
1065
parent_ids = vf.get_parents_with_ghosts(revision_id)
1066
for parent_id in parent_ids:
1067
# is this queued or done ?
1068
if (parent_id not in pending and
1069
parent_id not in done):
1070
# no, queue it.
1071
pending.add(parent_id)
1072
result.add_node(revision_id, parent_ids)
1073
done.add(revision_id)
1074
return result
1075
1076
def _get_revision_vf(self):
1077
""":return: a versioned file containing the revisions."""
1078
vf = self._revision_store.get_revision_file(self.get_transaction())
1079
return vf
1080
1081
@needs_write_lock
1082
def reconcile(self, other=None, thorough=False):
1083
"""Reconcile this repository."""
1084
from bzrlib.reconcile import KnitReconciler
1085
reconciler = KnitReconciler(self, thorough=thorough)
1086
reconciler.reconcile()
1087
return reconciler
1088
1089
def revision_parents(self, revision_id):
1090
return self._get_revision_vf().get_parents(revision_id)
1091
1092
1093
class KnitRepository2(KnitRepository):
1094
""""""
1095
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1096
control_store, text_store):
1097
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1098
_revision_store, control_store, text_store)
1099
self._serializer = xml6.serializer_v6
1100
1101
def deserialise_inventory(self, revision_id, xml):
1102
"""Transform the xml into an inventory object.
1103
1104
:param revision_id: The expected revision id of the inventory.
1105
:param xml: A serialised inventory.
1106
"""
1107
result = self._serializer.read_inventory_from_string(xml)
1108
assert result.root.revision is not None
1109
return result
1110
1111
def serialise_inventory(self, inv):
1112
"""Transform the inventory object into XML text.
1113
1114
:param revision_id: The expected revision id of the inventory.
1115
:param xml: A serialised inventory.
1116
"""
1117
assert inv.revision_id is not None
1118
assert inv.root.revision is not None
1119
return KnitRepository.serialise_inventory(self, inv)
1120
1121
def get_commit_builder(self, branch, parents, config, timestamp=None,
1122
timezone=None, committer=None, revprops=None,
1123
revision_id=None):
1124
"""Obtain a CommitBuilder for this repository.
1125
1126
:param branch: Branch to commit to.
1127
:param parents: Revision ids of the parents of the new revision.
1128
:param config: Configuration to use.
1129
:param timestamp: Optional timestamp recorded for commit.
1130
:param timezone: Optional timezone for timestamp.
1131
:param committer: Optional committer to set for commit.
1132
:param revprops: Optional dictionary of revision properties.
1133
:param revision_id: Optional revision id.
1134
"""
1135
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1136
committer, revprops, revision_id)
1137
1138
1139
class RepositoryFormatRegistry(registry.Registry):
1140
"""Registry of RepositoryFormats.
1141
"""
1142
1143
1144
format_registry = RepositoryFormatRegistry()
1145
"""Registry of formats, indexed by their identifying format string."""
1146
1147
1148
class RepositoryFormat(object):
1149
"""A repository format.
1150
1151
Formats provide three things:
1152
* An initialization routine to construct repository data on disk.
1153
* a format string which is used when the BzrDir supports versioned
1154
children.
1155
* an open routine which returns a Repository instance.
1156
1157
Formats are placed in an dict by their format string for reference
1158
during opening. These should be subclasses of RepositoryFormat
1159
for consistency.
1160
1161
Once a format is deprecated, just deprecate the initialize and open
1162
methods on the format class. Do not deprecate the object, as the
1163
object will be created every system load.
1164
1165
Common instance attributes:
1166
_matchingbzrdir - the bzrdir format that the repository format was
1167
originally written to work with. This can be used if manually
1168
constructing a bzrdir and repository, or more commonly for test suite
1169
parameterisation.
1170
"""
1171
1172
def __str__(self):
1173
return "<%s>" % self.__class__.__name__
1174
1175
@classmethod
1176
def find_format(klass, a_bzrdir):
1177
"""Return the format for the repository object in a_bzrdir.
1178
1179
This is used by bzr native formats that have a "format" file in
1180
the repository. Other methods may be used by different types of
1181
control directory.
1182
"""
1183
try:
1184
transport = a_bzrdir.get_repository_transport(None)
1185
format_string = transport.get("format").read()
1186
return format_registry.get(format_string)
1187
except errors.NoSuchFile:
1188
raise errors.NoRepositoryPresent(a_bzrdir)
1189
except KeyError:
1190
raise errors.UnknownFormatError(format=format_string)
1191
1192
@classmethod
1193
@deprecated_method(symbol_versioning.zero_fourteen)
1194
def set_default_format(klass, format):
1195
klass._set_default_format(format)
1196
1197
@classmethod
1198
def _set_default_format(klass, format):
1199
"""Set the default format for new Repository creation.
1200
1201
The format must already be registered.
1202
"""
1203
format_registry.default_key = format.get_format_string()
1204
1205
@classmethod
1206
def register_format(klass, format):
1207
format_registry.register(format.get_format_string(), format)
1208
1209
@classmethod
1210
def unregister_format(klass, format):
1211
format_registry.remove(format.get_format_string())
1212
1213
@classmethod
1214
def get_default_format(klass):
1215
"""Return the current default format."""
1216
return format_registry.get(format_registry.default_key)
1217
1218
def _get_control_store(self, repo_transport, control_files):
1219
"""Return the control store for this repository."""
1220
raise NotImplementedError(self._get_control_store)
1221
1222
def get_format_string(self):
1223
"""Return the ASCII format string that identifies this format.
1224
1225
Note that in pre format ?? repositories the format string is
1226
not permitted nor written to disk.
1227
"""
1228
raise NotImplementedError(self.get_format_string)
1229
1230
def get_format_description(self):
1231
"""Return the short description for this format."""
1232
raise NotImplementedError(self.get_format_description)
1233
1234
def _get_revision_store(self, repo_transport, control_files):
1235
"""Return the revision store object for this a_bzrdir."""
1236
raise NotImplementedError(self._get_revision_store)
1237
1238
def _get_text_rev_store(self,
1239
transport,
1240
control_files,
1241
name,
1242
compressed=True,
1243
prefixed=False,
1244
serializer=None):
1245
"""Common logic for getting a revision store for a repository.
1246
1247
see self._get_revision_store for the subclass-overridable method to
1248
get the store for a repository.
1249
"""
1250
from bzrlib.store.revision.text import TextRevisionStore
1251
dir_mode = control_files._dir_mode
1252
file_mode = control_files._file_mode
1253
text_store =TextStore(transport.clone(name),
1254
prefixed=prefixed,
1255
compressed=compressed,
1256
dir_mode=dir_mode,
1257
file_mode=file_mode)
1258
_revision_store = TextRevisionStore(text_store, serializer)
1259
return _revision_store
1260
1261
def _get_versioned_file_store(self,
1262
name,
1263
transport,
1264
control_files,
1265
prefixed=True,
1266
versionedfile_class=weave.WeaveFile,
1267
versionedfile_kwargs={},
1268
escaped=False):
1269
weave_transport = control_files._transport.clone(name)
1270
dir_mode = control_files._dir_mode
1271
file_mode = control_files._file_mode
1272
return VersionedFileStore(weave_transport, prefixed=prefixed,
1273
dir_mode=dir_mode,
1274
file_mode=file_mode,
1275
versionedfile_class=versionedfile_class,
1276
versionedfile_kwargs=versionedfile_kwargs,
1277
escaped=escaped)
1278
1279
def initialize(self, a_bzrdir, shared=False):
1280
"""Initialize a repository of this format in a_bzrdir.
1281
1282
:param a_bzrdir: The bzrdir to put the new repository in it.
1283
:param shared: The repository should be initialized as a sharable one.
1284
1285
This may raise UninitializableFormat if shared repository are not
1286
compatible the a_bzrdir.
1287
"""
1288
1289
def is_supported(self):
1290
"""Is this format supported?
1291
1292
Supported formats must be initializable and openable.
1293
Unsupported formats may not support initialization or committing or
1294
some other features depending on the reason for not being supported.
1295
"""
1296
return True
1297
1298
def check_conversion_target(self, target_format):
1299
raise NotImplementedError(self.check_conversion_target)
1300
1301
def open(self, a_bzrdir, _found=False):
1302
"""Return an instance of this format for the bzrdir a_bzrdir.
1303
1304
_found is a private parameter, do not use it.
1305
"""
1306
raise NotImplementedError(self.open)
1307
1308
1309
class PreSplitOutRepositoryFormat(RepositoryFormat):
1310
"""Base class for the pre split out repository formats."""
1311
1312
rich_root_data = False
1313
1314
def initialize(self, a_bzrdir, shared=False, _internal=False):
1315
"""Create a weave repository.
1316
1317
TODO: when creating split out bzr branch formats, move this to a common
1318
base for Format5, Format6. or something like that.
1319
"""
1320
if shared:
1321
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1322
1323
if not _internal:
1324
# always initialized when the bzrdir is.
1325
return self.open(a_bzrdir, _found=True)
1326
1327
# Create an empty weave
1328
sio = StringIO()
1329
weavefile.write_weave_v5(weave.Weave(), sio)
1330
empty_weave = sio.getvalue()
1331
1332
mutter('creating repository in %s.', a_bzrdir.transport.base)
1333
dirs = ['revision-store', 'weaves']
1334
files = [('inventory.weave', StringIO(empty_weave)),
1335
]
1336
1337
# FIXME: RBC 20060125 don't peek under the covers
1338
# NB: no need to escape relative paths that are url safe.
1339
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1340
'branch-lock', lockable_files.TransportLock)
1341
control_files.create_lock()
1342
control_files.lock_write()
1343
control_files._transport.mkdir_multi(dirs,
1344
mode=control_files._dir_mode)
1345
try:
1346
for file, content in files:
1347
control_files.put(file, content)
1348
finally:
1349
control_files.unlock()
1350
return self.open(a_bzrdir, _found=True)
1351
1352
def _get_control_store(self, repo_transport, control_files):
1353
"""Return the control store for this repository."""
1354
return self._get_versioned_file_store('',
1355
repo_transport,
1356
control_files,
1357
prefixed=False)
1358
1359
def _get_text_store(self, transport, control_files):
1360
"""Get a store for file texts for this format."""
1361
raise NotImplementedError(self._get_text_store)
1362
1363
def open(self, a_bzrdir, _found=False):
1364
"""See RepositoryFormat.open()."""
1365
if not _found:
1366
# we are being called directly and must probe.
1367
raise NotImplementedError
1368
1369
repo_transport = a_bzrdir.get_repository_transport(None)
1370
control_files = a_bzrdir._control_files
1371
text_store = self._get_text_store(repo_transport, control_files)
1372
control_store = self._get_control_store(repo_transport, control_files)
1373
_revision_store = self._get_revision_store(repo_transport, control_files)
1374
return AllInOneRepository(_format=self,
1375
a_bzrdir=a_bzrdir,
1376
_revision_store=_revision_store,
1377
control_store=control_store,
1378
text_store=text_store)
1379
1380
def check_conversion_target(self, target_format):
1381
pass
1382
1383
1384
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1385
"""Bzr repository format 4.
1386
1387
This repository format has:
1388
- flat stores
1389
- TextStores for texts, inventories,revisions.
1390
1391
This format is deprecated: it indexes texts using a text id which is
1392
removed in format 5; initialization and write support for this format
1393
has been removed.
1394
"""
1395
1396
def __init__(self):
1397
super(RepositoryFormat4, self).__init__()
1398
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1399
1400
def get_format_description(self):
1401
"""See RepositoryFormat.get_format_description()."""
1402
return "Repository format 4"
1403
1404
def initialize(self, url, shared=False, _internal=False):
1405
"""Format 4 branches cannot be created."""
1406
raise errors.UninitializableFormat(self)
1407
1408
def is_supported(self):
1409
"""Format 4 is not supported.
1410
1411
It is not supported because the model changed from 4 to 5 and the
1412
conversion logic is expensive - so doing it on the fly was not
1413
feasible.
1414
"""
1415
return False
1416
1417
def _get_control_store(self, repo_transport, control_files):
1418
"""Format 4 repositories have no formal control store at this point.
1419
1420
This will cause any control-file-needing apis to fail - this is desired.
1421
"""
1422
return None
1423
1424
def _get_revision_store(self, repo_transport, control_files):
1425
"""See RepositoryFormat._get_revision_store()."""
1426
from bzrlib.xml4 import serializer_v4
1427
return self._get_text_rev_store(repo_transport,
1428
control_files,
1429
'revision-store',
1430
serializer=serializer_v4)
1431
1432
def _get_text_store(self, transport, control_files):
1433
"""See RepositoryFormat._get_text_store()."""
1434
1435
1436
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1437
"""Bzr control format 5.
1438
1439
This repository format has:
1440
- weaves for file texts and inventory
1441
- flat stores
1442
- TextStores for revisions and signatures.
1443
"""
1444
1445
def __init__(self):
1446
super(RepositoryFormat5, self).__init__()
1447
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1448
1449
def get_format_description(self):
1450
"""See RepositoryFormat.get_format_description()."""
1451
return "Weave repository format 5"
1452
1453
def _get_revision_store(self, repo_transport, control_files):
1454
"""See RepositoryFormat._get_revision_store()."""
1455
"""Return the revision store object for this a_bzrdir."""
1456
return self._get_text_rev_store(repo_transport,
1457
control_files,
1458
'revision-store',
1459
compressed=False)
1460
1461
def _get_text_store(self, transport, control_files):
1462
"""See RepositoryFormat._get_text_store()."""
1463
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1464
1465
1466
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1467
"""Bzr control format 6.
1468
1469
This repository format has:
1470
- weaves for file texts and inventory
1471
- hash subdirectory based stores.
1472
- TextStores for revisions and signatures.
1473
"""
1474
1475
def __init__(self):
1476
super(RepositoryFormat6, self).__init__()
1477
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1478
1479
def get_format_description(self):
1480
"""See RepositoryFormat.get_format_description()."""
1481
return "Weave repository format 6"
1482
1483
def _get_revision_store(self, repo_transport, control_files):
1484
"""See RepositoryFormat._get_revision_store()."""
1485
return self._get_text_rev_store(repo_transport,
1486
control_files,
1487
'revision-store',
1488
compressed=False,
1489
prefixed=True)
1490
1491
def _get_text_store(self, transport, control_files):
1492
"""See RepositoryFormat._get_text_store()."""
1493
return self._get_versioned_file_store('weaves', transport, control_files)
1494
1495
1496
class MetaDirRepositoryFormat(RepositoryFormat):
1497
"""Common base class for the new repositories using the metadir layout."""
1498
1499
rich_root_data = False
1500
1501
def __init__(self):
1502
super(MetaDirRepositoryFormat, self).__init__()
1503
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1504
1505
def _create_control_files(self, a_bzrdir):
1506
"""Create the required files and the initial control_files object."""
1507
# FIXME: RBC 20060125 don't peek under the covers
1508
# NB: no need to escape relative paths that are url safe.
1509
repository_transport = a_bzrdir.get_repository_transport(self)
1510
control_files = lockable_files.LockableFiles(repository_transport,
1511
'lock', lockdir.LockDir)
1512
control_files.create_lock()
1513
return control_files
1514
1515
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1516
"""Upload the initial blank content."""
1517
control_files = self._create_control_files(a_bzrdir)
1518
control_files.lock_write()
1519
try:
1520
control_files._transport.mkdir_multi(dirs,
1521
mode=control_files._dir_mode)
1522
for file, content in files:
1523
control_files.put(file, content)
1524
for file, content in utf8_files:
1525
control_files.put_utf8(file, content)
1526
if shared == True:
1527
control_files.put_utf8('shared-storage', '')
1528
finally:
1529
control_files.unlock()
1530
1531
1532
class RepositoryFormat7(MetaDirRepositoryFormat):
1533
"""Bzr repository 7.
1534
1535
This repository format has:
1536
- weaves for file texts and inventory
1537
- hash subdirectory based stores.
1538
- TextStores for revisions and signatures.
1539
- a format marker of its own
1540
- an optional 'shared-storage' flag
1541
- an optional 'no-working-trees' flag
1542
"""
1543
1544
def _get_control_store(self, repo_transport, control_files):
1545
"""Return the control store for this repository."""
1546
return self._get_versioned_file_store('',
1547
repo_transport,
1548
control_files,
1549
prefixed=False)
1550
1551
def get_format_string(self):
1552
"""See RepositoryFormat.get_format_string()."""
1553
return "Bazaar-NG Repository format 7"
1554
1555
def get_format_description(self):
1556
"""See RepositoryFormat.get_format_description()."""
1557
return "Weave repository format 7"
1558
1559
def check_conversion_target(self, target_format):
1560
pass
1561
1562
def _get_revision_store(self, repo_transport, control_files):
1563
"""See RepositoryFormat._get_revision_store()."""
1564
return self._get_text_rev_store(repo_transport,
1565
control_files,
1566
'revision-store',
1567
compressed=False,
1568
prefixed=True,
1569
)
1570
1571
def _get_text_store(self, transport, control_files):
1572
"""See RepositoryFormat._get_text_store()."""
1573
return self._get_versioned_file_store('weaves',
1574
transport,
1575
control_files)
1576
1577
def initialize(self, a_bzrdir, shared=False):
1578
"""Create a weave repository.
1579
1580
:param shared: If true the repository will be initialized as a shared
1581
repository.
1582
"""
1583
# Create an empty weave
1584
sio = StringIO()
1585
weavefile.write_weave_v5(weave.Weave(), sio)
1586
empty_weave = sio.getvalue()
1587
1588
mutter('creating repository in %s.', a_bzrdir.transport.base)
1589
dirs = ['revision-store', 'weaves']
1590
files = [('inventory.weave', StringIO(empty_weave)),
1591
]
1592
utf8_files = [('format', self.get_format_string())]
1593
1594
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1595
return self.open(a_bzrdir=a_bzrdir, _found=True)
1596
1597
def open(self, a_bzrdir, _found=False, _override_transport=None):
1598
"""See RepositoryFormat.open().
1599
1600
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1601
repository at a slightly different url
1602
than normal. I.e. during 'upgrade'.
1603
"""
1604
if not _found:
1605
format = RepositoryFormat.find_format(a_bzrdir)
1606
assert format.__class__ == self.__class__
1607
if _override_transport is not None:
1608
repo_transport = _override_transport
1609
else:
1610
repo_transport = a_bzrdir.get_repository_transport(None)
1611
control_files = lockable_files.LockableFiles(repo_transport,
1612
'lock', lockdir.LockDir)
1613
text_store = self._get_text_store(repo_transport, control_files)
1614
control_store = self._get_control_store(repo_transport, control_files)
1615
_revision_store = self._get_revision_store(repo_transport, control_files)
1616
return WeaveMetaDirRepository(_format=self,
1617
a_bzrdir=a_bzrdir,
1618
control_files=control_files,
1619
_revision_store=_revision_store,
1620
control_store=control_store,
1621
text_store=text_store)
1622
1623
1624
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1625
"""Bzr repository knit format (generalized).
1626
1627
This repository format has:
1628
- knits for file texts and inventory
1629
- hash subdirectory based stores.
1630
- knits for revisions and signatures
1631
- TextStores for revisions and signatures.
1632
- a format marker of its own
1633
- an optional 'shared-storage' flag
1634
- an optional 'no-working-trees' flag
1635
- a LockDir lock
1636
"""
1637
1638
def _get_control_store(self, repo_transport, control_files):
1639
"""Return the control store for this repository."""
1640
return VersionedFileStore(
1641
repo_transport,
1642
prefixed=False,
1643
file_mode=control_files._file_mode,
1644
versionedfile_class=knit.KnitVersionedFile,
1645
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1646
)
1647
1648
def _get_revision_store(self, repo_transport, control_files):
1649
"""See RepositoryFormat._get_revision_store()."""
1650
from bzrlib.store.revision.knit import KnitRevisionStore
1651
versioned_file_store = VersionedFileStore(
1652
repo_transport,
1653
file_mode=control_files._file_mode,
1654
prefixed=False,
1655
precious=True,
1656
versionedfile_class=knit.KnitVersionedFile,
1657
versionedfile_kwargs={'delta':False,
1658
'factory':knit.KnitPlainFactory(),
1659
},
1660
escaped=True,
1661
)
1662
return KnitRevisionStore(versioned_file_store)
1663
1664
def _get_text_store(self, transport, control_files):
1665
"""See RepositoryFormat._get_text_store()."""
1666
return self._get_versioned_file_store('knits',
1667
transport,
1668
control_files,
1669
versionedfile_class=knit.KnitVersionedFile,
1670
versionedfile_kwargs={
1671
'create_parent_dir':True,
1672
'delay_create':True,
1673
'dir_mode':control_files._dir_mode,
1674
},
1675
escaped=True)
1676
1677
def initialize(self, a_bzrdir, shared=False):
1678
"""Create a knit format 1 repository.
1679
1680
:param a_bzrdir: bzrdir to contain the new repository; must already
1681
be initialized.
1682
:param shared: If true the repository will be initialized as a shared
1683
repository.
1684
"""
1685
mutter('creating repository in %s.', a_bzrdir.transport.base)
1686
dirs = ['revision-store', 'knits']
1687
files = []
1688
utf8_files = [('format', self.get_format_string())]
1689
1690
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1691
repo_transport = a_bzrdir.get_repository_transport(None)
1692
control_files = lockable_files.LockableFiles(repo_transport,
1693
'lock', lockdir.LockDir)
1694
control_store = self._get_control_store(repo_transport, control_files)
1695
transaction = transactions.WriteTransaction()
1696
# trigger a write of the inventory store.
1697
control_store.get_weave_or_empty('inventory', transaction)
1698
_revision_store = self._get_revision_store(repo_transport, control_files)
1699
# the revision id here is irrelevant: it will not be stored, and cannot
1700
# already exist.
1701
_revision_store.has_revision_id('A', transaction)
1702
_revision_store.get_signature_file(transaction)
1703
return self.open(a_bzrdir=a_bzrdir, _found=True)
1704
1705
def open(self, a_bzrdir, _found=False, _override_transport=None):
1706
"""See RepositoryFormat.open().
1707
1708
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1709
repository at a slightly different url
1710
than normal. I.e. during 'upgrade'.
1711
"""
1712
if not _found:
1713
format = RepositoryFormat.find_format(a_bzrdir)
1714
assert format.__class__ == self.__class__
1715
if _override_transport is not None:
1716
repo_transport = _override_transport
1717
else:
1718
repo_transport = a_bzrdir.get_repository_transport(None)
1719
control_files = lockable_files.LockableFiles(repo_transport,
1720
'lock', lockdir.LockDir)
1721
text_store = self._get_text_store(repo_transport, control_files)
1722
control_store = self._get_control_store(repo_transport, control_files)
1723
_revision_store = self._get_revision_store(repo_transport, control_files)
1724
return KnitRepository(_format=self,
1725
a_bzrdir=a_bzrdir,
1726
control_files=control_files,
1727
_revision_store=_revision_store,
1728
control_store=control_store,
1729
text_store=text_store)
1730
1731
1732
class RepositoryFormatKnit1(RepositoryFormatKnit):
1733
"""Bzr repository knit format 1.
1734
1735
This repository format has:
1736
- knits for file texts and inventory
1737
- hash subdirectory based stores.
1738
- knits for revisions and signatures
1739
- TextStores for revisions and signatures.
1740
- a format marker of its own
1741
- an optional 'shared-storage' flag
1742
- an optional 'no-working-trees' flag
1743
- a LockDir lock
1744
1745
This format was introduced in bzr 0.8.
1746
"""
1747
def get_format_string(self):
1748
"""See RepositoryFormat.get_format_string()."""
1749
return "Bazaar-NG Knit Repository Format 1"
1750
1751
def get_format_description(self):
1752
"""See RepositoryFormat.get_format_description()."""
1753
return "Knit repository format 1"
1754
1755
def check_conversion_target(self, target_format):
1756
pass
1757
1758
1759
class RepositoryFormatKnit2(RepositoryFormatKnit):
1760
"""Bzr repository knit format 2.
1761
1762
THIS FORMAT IS EXPERIMENTAL
1763
This repository format has:
1764
- knits for file texts and inventory
1765
- hash subdirectory based stores.
1766
- knits for revisions and signatures
1767
- TextStores for revisions and signatures.
1768
- a format marker of its own
1769
- an optional 'shared-storage' flag
1770
- an optional 'no-working-trees' flag
1771
- a LockDir lock
1772
- Support for recording full info about the tree root
1773
1774
"""
1775
1776
rich_root_data = True
1777
1778
def get_format_string(self):
1779
"""See RepositoryFormat.get_format_string()."""
1780
return "Bazaar Knit Repository Format 2\n"
1781
1782
def get_format_description(self):
1783
"""See RepositoryFormat.get_format_description()."""
1784
return "Knit repository format 2"
1785
1786
def check_conversion_target(self, target_format):
1787
if not target_format.rich_root_data:
1788
raise errors.BadConversionTarget(
1789
'Does not support rich root data.', target_format)
1790
1791
def open(self, a_bzrdir, _found=False, _override_transport=None):
1792
"""See RepositoryFormat.open().
1793
1794
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1795
repository at a slightly different url
1796
than normal. I.e. during 'upgrade'.
1797
"""
1798
if not _found:
1799
format = RepositoryFormat.find_format(a_bzrdir)
1800
assert format.__class__ == self.__class__
1801
if _override_transport is not None:
1802
repo_transport = _override_transport
1803
else:
1804
repo_transport = a_bzrdir.get_repository_transport(None)
1805
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1806
lockdir.LockDir)
1807
text_store = self._get_text_store(repo_transport, control_files)
1808
control_store = self._get_control_store(repo_transport, control_files)
1809
_revision_store = self._get_revision_store(repo_transport, control_files)
1810
return KnitRepository2(_format=self,
1811
a_bzrdir=a_bzrdir,
1812
control_files=control_files,
1813
_revision_store=_revision_store,
1814
control_store=control_store,
1815
text_store=text_store)
1816
1817
1818
1819
# formats which have no format string are not discoverable
1820
# and not independently creatable, so are not registered.
1821
RepositoryFormat.register_format(RepositoryFormat7())
1822
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1823
# default control directory format
1824
_default_format = RepositoryFormatKnit1()
1825
RepositoryFormat.register_format(_default_format)
1826
RepositoryFormat.register_format(RepositoryFormatKnit2())
1827
RepositoryFormat._set_default_format(_default_format)
1828
_legacy_formats = [RepositoryFormat4(),
1829
RepositoryFormat5(),
1830
RepositoryFormat6()]
1831
1832
1833
class InterRepository(InterObject):
1834
"""This class represents operations taking place between two repositories.
1835
1836
Its instances have methods like copy_content and fetch, and contain
1837
references to the source and target repositories these operations can be
1838
carried out on.
1839
1840
Often we will provide convenience methods on 'repository' which carry out
1841
operations with another repository - they will always forward to
1842
InterRepository.get(other).method_name(parameters).
1843
"""
1844
1845
_optimisers = []
1846
"""The available optimised InterRepository types."""
1847
1848
def copy_content(self, revision_id=None, basis=None):
1849
raise NotImplementedError(self.copy_content)
1850
1851
def fetch(self, revision_id=None, pb=None):
1852
"""Fetch the content required to construct revision_id.
1853
1854
The content is copied from self.source to self.target.
1855
1856
:param revision_id: if None all content is copied, if NULL_REVISION no
1857
content is copied.
1858
:param pb: optional progress bar to use for progress reports. If not
1859
provided a default one will be created.
1860
1861
Returns the copied revision count and the failed revisions in a tuple:
1862
(copied, failures).
1863
"""
1864
raise NotImplementedError(self.fetch)
1865
1866
@needs_read_lock
1867
def missing_revision_ids(self, revision_id=None):
1868
"""Return the revision ids that source has that target does not.
1869
1870
These are returned in topological order.
1871
1872
:param revision_id: only return revision ids included by this
1873
revision_id.
1874
"""
1875
# generic, possibly worst case, slow code path.
1876
target_ids = set(self.target.all_revision_ids())
1877
if revision_id is not None:
1878
source_ids = self.source.get_ancestry(revision_id)
1879
assert source_ids[0] is None
1880
source_ids.pop(0)
1881
else:
1882
source_ids = self.source.all_revision_ids()
1883
result_set = set(source_ids).difference(target_ids)
1884
# this may look like a no-op: its not. It preserves the ordering
1885
# other_ids had while only returning the members from other_ids
1886
# that we've decided we need.
1887
return [rev_id for rev_id in source_ids if rev_id in result_set]
1888
1889
1890
class InterSameDataRepository(InterRepository):
1891
"""Code for converting between repositories that represent the same data.
1892
1893
Data format and model must match for this to work.
1894
"""
1895
1896
_matching_repo_format = RepositoryFormat4()
1897
"""Repository format for testing with."""
1898
1899
@staticmethod
1900
def is_compatible(source, target):
1901
if not isinstance(source, Repository):
1902
return False
1903
if not isinstance(target, Repository):
1904
return False
1905
if source._format.rich_root_data == target._format.rich_root_data:
1906
return True
1907
else:
1908
return False
1909
1910
@needs_write_lock
1911
def copy_content(self, revision_id=None, basis=None):
1912
"""Make a complete copy of the content in self into destination.
1913
1914
This is a destructive operation! Do not use it on existing
1915
repositories.
1916
1917
:param revision_id: Only copy the content needed to construct
1918
revision_id and its parents.
1919
:param basis: Copy the needed data preferentially from basis.
1920
"""
1921
try:
1922
self.target.set_make_working_trees(self.source.make_working_trees())
1923
except NotImplementedError:
1924
pass
1925
# grab the basis available data
1926
if basis is not None:
1927
self.target.fetch(basis, revision_id=revision_id)
1928
# but don't bother fetching if we have the needed data now.
1929
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1930
self.target.has_revision(revision_id)):
1931
return
1932
self.target.fetch(self.source, revision_id=revision_id)
1933
1934
@needs_write_lock
1935
def fetch(self, revision_id=None, pb=None):
1936
"""See InterRepository.fetch()."""
1937
from bzrlib.fetch import GenericRepoFetcher
1938
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1939
self.source, self.source._format, self.target,
1940
self.target._format)
1941
f = GenericRepoFetcher(to_repository=self.target,
1942
from_repository=self.source,
1943
last_revision=revision_id,
1944
pb=pb)
1945
return f.count_copied, f.failed_revisions
1946
1947
1948
class InterWeaveRepo(InterSameDataRepository):
1949
"""Optimised code paths between Weave based repositories."""
1950
1951
_matching_repo_format = RepositoryFormat7()
1952
"""Repository format for testing with."""
1953
1954
@staticmethod
1955
def is_compatible(source, target):
1956
"""Be compatible with known Weave formats.
1957
1958
We don't test for the stores being of specific types because that
1959
could lead to confusing results, and there is no need to be
1960
overly general.
1961
"""
1962
try:
1963
return (isinstance(source._format, (RepositoryFormat5,
1964
RepositoryFormat6,
1965
RepositoryFormat7)) and
1966
isinstance(target._format, (RepositoryFormat5,
1967
RepositoryFormat6,
1968
RepositoryFormat7)))
1969
except AttributeError:
1970
return False
1971
1972
@needs_write_lock
1973
def copy_content(self, revision_id=None, basis=None):
1974
"""See InterRepository.copy_content()."""
1975
# weave specific optimised path:
1976
if basis is not None:
1977
# copy the basis in, then fetch remaining data.
1978
basis.copy_content_into(self.target, revision_id)
1979
# the basis copy_content_into could miss-set this.
1980
try:
1981
self.target.set_make_working_trees(self.source.make_working_trees())
1982
except NotImplementedError:
1983
pass
1984
self.target.fetch(self.source, revision_id=revision_id)
1985
else:
1986
try:
1987
self.target.set_make_working_trees(self.source.make_working_trees())
1988
except NotImplementedError:
1989
pass
1990
# FIXME do not peek!
1991
if self.source.control_files._transport.listable():
1992
pb = ui.ui_factory.nested_progress_bar()
1993
try:
1994
self.target.weave_store.copy_all_ids(
1995
self.source.weave_store,
1996
pb=pb,
1997
from_transaction=self.source.get_transaction(),
1998
to_transaction=self.target.get_transaction())
1999
pb.update('copying inventory', 0, 1)
2000
self.target.control_weaves.copy_multi(
2001
self.source.control_weaves, ['inventory'],
2002
from_transaction=self.source.get_transaction(),
2003
to_transaction=self.target.get_transaction())
2004
self.target._revision_store.text_store.copy_all_ids(
2005
self.source._revision_store.text_store,
2006
pb=pb)
2007
finally:
2008
pb.finished()
2009
else:
2010
self.target.fetch(self.source, revision_id=revision_id)
2011
2012
@needs_write_lock
2013
def fetch(self, revision_id=None, pb=None):
2014
"""See InterRepository.fetch()."""
2015
from bzrlib.fetch import GenericRepoFetcher
2016
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2017
self.source, self.source._format, self.target, self.target._format)
2018
f = GenericRepoFetcher(to_repository=self.target,
2019
from_repository=self.source,
2020
last_revision=revision_id,
2021
pb=pb)
2022
return f.count_copied, f.failed_revisions
2023
2024
@needs_read_lock
2025
def missing_revision_ids(self, revision_id=None):
2026
"""See InterRepository.missing_revision_ids()."""
2027
# we want all revisions to satisfy revision_id in source.
2028
# but we don't want to stat every file here and there.
2029
# we want then, all revisions other needs to satisfy revision_id
2030
# checked, but not those that we have locally.
2031
# so the first thing is to get a subset of the revisions to
2032
# satisfy revision_id in source, and then eliminate those that
2033
# we do already have.
2034
# this is slow on high latency connection to self, but as as this
2035
# disk format scales terribly for push anyway due to rewriting
2036
# inventory.weave, this is considered acceptable.
2037
# - RBC 20060209
2038
if revision_id is not None:
2039
source_ids = self.source.get_ancestry(revision_id)
2040
assert source_ids[0] is None
2041
source_ids.pop(0)
2042
else:
2043
source_ids = self.source._all_possible_ids()
2044
source_ids_set = set(source_ids)
2045
# source_ids is the worst possible case we may need to pull.
2046
# now we want to filter source_ids against what we actually
2047
# have in target, but don't try to check for existence where we know
2048
# we do not have a revision as that would be pointless.
2049
target_ids = set(self.target._all_possible_ids())
2050
possibly_present_revisions = target_ids.intersection(source_ids_set)
2051
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2052
required_revisions = source_ids_set.difference(actually_present_revisions)
2053
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2054
if revision_id is not None:
2055
# we used get_ancestry to determine source_ids then we are assured all
2056
# revisions referenced are present as they are installed in topological order.
2057
# and the tip revision was validated by get_ancestry.
2058
return required_topo_revisions
2059
else:
2060
# if we just grabbed the possibly available ids, then
2061
# we only have an estimate of whats available and need to validate
2062
# that against the revision records.
2063
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2064
2065
2066
class InterKnitRepo(InterSameDataRepository):
2067
"""Optimised code paths between Knit based repositories."""
2068
2069
_matching_repo_format = RepositoryFormatKnit1()
2070
"""Repository format for testing with."""
2071
2072
@staticmethod
2073
def is_compatible(source, target):
2074
"""Be compatible with known Knit formats.
2075
2076
We don't test for the stores being of specific types because that
2077
could lead to confusing results, and there is no need to be
2078
overly general.
2079
"""
2080
try:
2081
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2082
isinstance(target._format, (RepositoryFormatKnit1)))
2083
except AttributeError:
2084
return False
2085
2086
@needs_write_lock
2087
def fetch(self, revision_id=None, pb=None):
2088
"""See InterRepository.fetch()."""
2089
from bzrlib.fetch import KnitRepoFetcher
2090
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2091
self.source, self.source._format, self.target, self.target._format)
2092
f = KnitRepoFetcher(to_repository=self.target,
2093
from_repository=self.source,
2094
last_revision=revision_id,
2095
pb=pb)
2096
return f.count_copied, f.failed_revisions
2097
2098
@needs_read_lock
2099
def missing_revision_ids(self, revision_id=None):
2100
"""See InterRepository.missing_revision_ids()."""
2101
if revision_id is not None:
2102
source_ids = self.source.get_ancestry(revision_id)
2103
assert source_ids[0] is None
2104
source_ids.pop(0)
2105
else:
2106
source_ids = self.source._all_possible_ids()
2107
source_ids_set = set(source_ids)
2108
# source_ids is the worst possible case we may need to pull.
2109
# now we want to filter source_ids against what we actually
2110
# have in target, but don't try to check for existence where we know
2111
# we do not have a revision as that would be pointless.
2112
target_ids = set(self.target._all_possible_ids())
2113
possibly_present_revisions = target_ids.intersection(source_ids_set)
2114
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2115
required_revisions = source_ids_set.difference(actually_present_revisions)
2116
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2117
if revision_id is not None:
2118
# we used get_ancestry to determine source_ids then we are assured all
2119
# revisions referenced are present as they are installed in topological order.
2120
# and the tip revision was validated by get_ancestry.
2121
return required_topo_revisions
2122
else:
2123
# if we just grabbed the possibly available ids, then
2124
# we only have an estimate of whats available and need to validate
2125
# that against the revision records.
2126
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2127
2128
2129
class InterModel1and2(InterRepository):
2130
2131
_matching_repo_format = None
2132
2133
@staticmethod
2134
def is_compatible(source, target):
2135
if not isinstance(source, Repository):
2136
return False
2137
if not isinstance(target, Repository):
2138
return False
2139
if not source._format.rich_root_data and target._format.rich_root_data:
2140
return True
2141
else:
2142
return False
2143
2144
@needs_write_lock
2145
def fetch(self, revision_id=None, pb=None):
2146
"""See InterRepository.fetch()."""
2147
from bzrlib.fetch import Model1toKnit2Fetcher
2148
f = Model1toKnit2Fetcher(to_repository=self.target,
2149
from_repository=self.source,
2150
last_revision=revision_id,
2151
pb=pb)
2152
return f.count_copied, f.failed_revisions
2153
2154
@needs_write_lock
2155
def copy_content(self, revision_id=None, basis=None):
2156
"""Make a complete copy of the content in self into destination.
2157
2158
This is a destructive operation! Do not use it on existing
2159
repositories.
2160
2161
:param revision_id: Only copy the content needed to construct
2162
revision_id and its parents.
2163
:param basis: Copy the needed data preferentially from basis.
2164
"""
2165
try:
2166
self.target.set_make_working_trees(self.source.make_working_trees())
2167
except NotImplementedError:
2168
pass
2169
# grab the basis available data
2170
if basis is not None:
2171
self.target.fetch(basis, revision_id=revision_id)
2172
# but don't bother fetching if we have the needed data now.
2173
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2174
self.target.has_revision(revision_id)):
2175
return
2176
self.target.fetch(self.source, revision_id=revision_id)
2177
2178
2179
class InterKnit1and2(InterKnitRepo):
2180
2181
_matching_repo_format = None
2182
2183
@staticmethod
2184
def is_compatible(source, target):
2185
"""Be compatible with Knit1 source and Knit2 target"""
2186
try:
2187
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2188
isinstance(target._format, (RepositoryFormatKnit2)))
2189
except AttributeError:
2190
return False
2191
2192
@needs_write_lock
2193
def fetch(self, revision_id=None, pb=None):
2194
"""See InterRepository.fetch()."""
2195
from bzrlib.fetch import Knit1to2Fetcher
2196
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2197
self.source, self.source._format, self.target,
2198
self.target._format)
2199
f = Knit1to2Fetcher(to_repository=self.target,
2200
from_repository=self.source,
2201
last_revision=revision_id,
2202
pb=pb)
2203
return f.count_copied, f.failed_revisions
2204
2205
2206
InterRepository.register_optimiser(InterSameDataRepository)
2207
InterRepository.register_optimiser(InterWeaveRepo)
2208
InterRepository.register_optimiser(InterKnitRepo)
2209
InterRepository.register_optimiser(InterModel1and2)
2210
InterRepository.register_optimiser(InterKnit1and2)
2211
2212
2213
class RepositoryTestProviderAdapter(object):
2214
"""A tool to generate a suite testing multiple repository formats at once.
2215
2216
This is done by copying the test once for each transport and injecting
2217
the transport_server, transport_readonly_server, and bzrdir_format and
2218
repository_format classes into each copy. Each copy is also given a new id()
2219
to make it easy to identify.
2220
"""
2221
2222
def __init__(self, transport_server, transport_readonly_server, formats):
2223
self._transport_server = transport_server
2224
self._transport_readonly_server = transport_readonly_server
2225
self._formats = formats
2226
2227
def adapt(self, test):
2228
result = unittest.TestSuite()
2229
for repository_format, bzrdir_format in self._formats:
2230
new_test = deepcopy(test)
2231
new_test.transport_server = self._transport_server
2232
new_test.transport_readonly_server = self._transport_readonly_server
2233
new_test.bzrdir_format = bzrdir_format
2234
new_test.repository_format = repository_format
2235
def make_new_test_id():
2236
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2237
return lambda: new_id
2238
new_test.id = make_new_test_id()
2239
result.addTest(new_test)
2240
return result
2241
2242
2243
class InterRepositoryTestProviderAdapter(object):
2244
"""A tool to generate a suite testing multiple inter repository formats.
2245
2246
This is done by copying the test once for each interrepo provider and injecting
2247
the transport_server, transport_readonly_server, repository_format and
2248
repository_to_format classes into each copy.
2249
Each copy is also given a new id() to make it easy to identify.
2250
"""
2251
2252
def __init__(self, transport_server, transport_readonly_server, formats):
2253
self._transport_server = transport_server
2254
self._transport_readonly_server = transport_readonly_server
2255
self._formats = formats
2256
2257
def adapt(self, test):
2258
result = unittest.TestSuite()
2259
for interrepo_class, repository_format, repository_format_to in self._formats:
2260
new_test = deepcopy(test)
2261
new_test.transport_server = self._transport_server
2262
new_test.transport_readonly_server = self._transport_readonly_server
2263
new_test.interrepo_class = interrepo_class
2264
new_test.repository_format = repository_format
2265
new_test.repository_format_to = repository_format_to
2266
def make_new_test_id():
2267
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2268
return lambda: new_id
2269
new_test.id = make_new_test_id()
2270
result.addTest(new_test)
2271
return result
2272
2273
@staticmethod
2274
def default_test_list():
2275
"""Generate the default list of interrepo permutations to test."""
2276
result = []
2277
# test the default InterRepository between format 6 and the current
2278
# default format.
2279
# XXX: robertc 20060220 reinstate this when there are two supported
2280
# formats which do not have an optimal code path between them.
2281
#result.append((InterRepository,
2282
# RepositoryFormat6(),
2283
# RepositoryFormatKnit1()))
2284
for optimiser in InterRepository._optimisers:
2285
if optimiser._matching_repo_format is not None:
2286
result.append((optimiser,
2287
optimiser._matching_repo_format,
2288
optimiser._matching_repo_format
2289
))
2290
# if there are specific combinations we want to use, we can add them
2291
# here.
2292
result.append((InterModel1and2, RepositoryFormat5(),
2293
RepositoryFormatKnit2()))
2294
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2295
RepositoryFormatKnit2()))
2296
return result
2297
2298
2299
class CopyConverter(object):
2300
"""A repository conversion tool which just performs a copy of the content.
2301
2302
This is slow but quite reliable.
2303
"""
2304
2305
def __init__(self, target_format):
2306
"""Create a CopyConverter.
2307
2308
:param target_format: The format the resulting repository should be.
2309
"""
2310
self.target_format = target_format
2311
2312
def convert(self, repo, pb):
2313
"""Perform the conversion of to_convert, giving feedback via pb.
2314
2315
:param to_convert: The disk object to convert.
2316
:param pb: a progress bar to use for progress information.
2317
"""
2318
self.pb = pb
2319
self.count = 0
2320
self.total = 4
2321
# this is only useful with metadir layouts - separated repo content.
2322
# trigger an assertion if not such
2323
repo._format.get_format_string()
2324
self.repo_dir = repo.bzrdir
2325
self.step('Moving repository to repository.backup')
2326
self.repo_dir.transport.move('repository', 'repository.backup')
2327
backup_transport = self.repo_dir.transport.clone('repository.backup')
2328
repo._format.check_conversion_target(self.target_format)
2329
self.source_repo = repo._format.open(self.repo_dir,
2330
_found=True,
2331
_override_transport=backup_transport)
2332
self.step('Creating new repository')
2333
converted = self.target_format.initialize(self.repo_dir,
2334
self.source_repo.is_shared())
2335
converted.lock_write()
2336
try:
2337
self.step('Copying content into repository.')
2338
self.source_repo.copy_content_into(converted)
2339
finally:
2340
converted.unlock()
2341
self.step('Deleting old repository content.')
2342
self.repo_dir.transport.delete_tree('repository.backup')
2343
self.pb.note('repository converted')
2344
2345
def step(self, message):
2346
"""Update the pb by a step."""
2347
self.count +=1
2348
self.pb.update(message, self.count, self.total)
2349
2350
2351
class CommitBuilder(object):
2352
"""Provides an interface to build up a commit.
2353
2354
This allows describing a tree to be committed without needing to
2355
know the internals of the format of the repository.
2356
"""
2357
2358
record_root_entry = False
2359
def __init__(self, repository, parents, config, timestamp=None,
2360
timezone=None, committer=None, revprops=None,
2361
revision_id=None):
2362
"""Initiate a CommitBuilder.
2363
2364
:param repository: Repository to commit to.
2365
:param parents: Revision ids of the parents of the new revision.
2366
:param config: Configuration to use.
2367
:param timestamp: Optional timestamp recorded for commit.
2368
:param timezone: Optional timezone for timestamp.
2369
:param committer: Optional committer to set for commit.
2370
:param revprops: Optional dictionary of revision properties.
2371
:param revision_id: Optional revision id.
2372
"""
2373
self._config = config
2374
2375
if committer is None:
2376
self._committer = self._config.username()
2377
else:
2378
assert isinstance(committer, basestring), type(committer)
2379
self._committer = committer
2380
2381
self.new_inventory = Inventory(None)
2382
self._new_revision_id = revision_id
2383
self.parents = parents
2384
self.repository = repository
2385
2386
self._revprops = {}
2387
if revprops is not None:
2388
self._revprops.update(revprops)
2389
2390
if timestamp is None:
2391
timestamp = time.time()
2392
# Restrict resolution to 1ms
2393
self._timestamp = round(timestamp, 3)
2394
2395
if timezone is None:
2396
self._timezone = local_time_offset()
2397
else:
2398
self._timezone = int(timezone)
2399
2400
self._generate_revision_if_needed()
2401
2402
def commit(self, message):
2403
"""Make the actual commit.
2404
2405
:return: The revision id of the recorded revision.
2406
"""
2407
rev = _mod_revision.Revision(
2408
timestamp=self._timestamp,
2409
timezone=self._timezone,
2410
committer=self._committer,
2411
message=message,
2412
inventory_sha1=self.inv_sha1,
2413
revision_id=self._new_revision_id,
2414
properties=self._revprops)
2415
rev.parent_ids = self.parents
2416
self.repository.add_revision(self._new_revision_id, rev,
2417
self.new_inventory, self._config)
2418
return self._new_revision_id
2419
2420
def revision_tree(self):
2421
"""Return the tree that was just committed.
2422
2423
After calling commit() this can be called to get a RevisionTree
2424
representing the newly committed tree. This is preferred to
2425
calling Repository.revision_tree() because that may require
2426
deserializing the inventory, while we already have a copy in
2427
memory.
2428
"""
2429
return RevisionTree(self.repository, self.new_inventory,
2430
self._new_revision_id)
2431
2432
def finish_inventory(self):
2433
"""Tell the builder that the inventory is finished."""
2434
if self.new_inventory.root is None:
2435
symbol_versioning.warn('Root entry should be supplied to'
2436
' record_entry_contents, as of bzr 0.10.',
2437
DeprecationWarning, stacklevel=2)
2438
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2439
self.new_inventory.revision_id = self._new_revision_id
2440
self.inv_sha1 = self.repository.add_inventory(
2441
self._new_revision_id,
2442
self.new_inventory,
2443
self.parents
2444
)
2445
2446
def _gen_revision_id(self):
2447
"""Return new revision-id."""
2448
return generate_ids.gen_revision_id(self._config.username(),
2449
self._timestamp)
2450
2451
def _generate_revision_if_needed(self):
2452
"""Create a revision id if None was supplied.
2453
2454
If the repository can not support user-specified revision ids
2455
they should override this function and raise CannotSetRevisionId
2456
if _new_revision_id is not None.
2457
2458
:raises: CannotSetRevisionId
2459
"""
2460
if self._new_revision_id is None:
2461
self._new_revision_id = self._gen_revision_id()
2462
2463
def record_entry_contents(self, ie, parent_invs, path, tree):
2464
"""Record the content of ie from tree into the commit if needed.
2465
2466
Side effect: sets ie.revision when unchanged
2467
2468
:param ie: An inventory entry present in the commit.
2469
:param parent_invs: The inventories of the parent revisions of the
2470
commit.
2471
:param path: The path the entry is at in the tree.
2472
:param tree: The tree which contains this entry and should be used to
2473
obtain content.
2474
"""
2475
if self.new_inventory.root is None and ie.parent_id is not None:
2476
symbol_versioning.warn('Root entry should be supplied to'
2477
' record_entry_contents, as of bzr 0.10.',
2478
DeprecationWarning, stacklevel=2)
2479
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2480
'', tree)
2481
self.new_inventory.add(ie)
2482
2483
# ie.revision is always None if the InventoryEntry is considered
2484
# for committing. ie.snapshot will record the correct revision
2485
# which may be the sole parent if it is untouched.
2486
if ie.revision is not None:
2487
return
2488
2489
# In this revision format, root entries have no knit or weave
2490
if ie is self.new_inventory.root:
2491
# When serializing out to disk and back in
2492
# root.revision is always _new_revision_id
2493
ie.revision = self._new_revision_id
2494
return
2495
previous_entries = ie.find_previous_heads(
2496
parent_invs,
2497
self.repository.weave_store,
2498
self.repository.get_transaction())
2499
# we are creating a new revision for ie in the history store
2500
# and inventory.
2501
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2502
2503
def modified_directory(self, file_id, file_parents):
2504
"""Record the presence of a symbolic link.
2505
2506
:param file_id: The file_id of the link to record.
2507
:param file_parents: The per-file parent revision ids.
2508
"""
2509
self._add_text_to_weave(file_id, [], file_parents.keys())
2510
2511
def modified_file_text(self, file_id, file_parents,
2512
get_content_byte_lines, text_sha1=None,
2513
text_size=None):
2514
"""Record the text of file file_id
2515
2516
:param file_id: The file_id of the file to record the text of.
2517
:param file_parents: The per-file parent revision ids.
2518
:param get_content_byte_lines: A callable which will return the byte
2519
lines for the file.
2520
:param text_sha1: Optional SHA1 of the file contents.
2521
:param text_size: Optional size of the file contents.
2522
"""
2523
# mutter('storing text of file {%s} in revision {%s} into %r',
2524
# file_id, self._new_revision_id, self.repository.weave_store)
2525
# special case to avoid diffing on renames or
2526
# reparenting
2527
if (len(file_parents) == 1
2528
and text_sha1 == file_parents.values()[0].text_sha1
2529
and text_size == file_parents.values()[0].text_size):
2530
previous_ie = file_parents.values()[0]
2531
versionedfile = self.repository.weave_store.get_weave(file_id,
2532
self.repository.get_transaction())
2533
versionedfile.clone_text(self._new_revision_id,
2534
previous_ie.revision, file_parents.keys())
2535
return text_sha1, text_size
2536
else:
2537
new_lines = get_content_byte_lines()
2538
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2539
# should return the SHA1 and size
2540
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2541
return osutils.sha_strings(new_lines), \
2542
sum(map(len, new_lines))
2543
2544
def modified_link(self, file_id, file_parents, link_target):
2545
"""Record the presence of a symbolic link.
2546
2547
:param file_id: The file_id of the link to record.
2548
:param file_parents: The per-file parent revision ids.
2549
:param link_target: Target location of this link.
2550
"""
2551
self._add_text_to_weave(file_id, [], file_parents.keys())
2552
2553
def _add_text_to_weave(self, file_id, new_lines, parents):
2554
versionedfile = self.repository.weave_store.get_weave_or_empty(
2555
file_id, self.repository.get_transaction())
2556
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2557
versionedfile.clear_cache()
2558
2559
2560
class _CommitBuilder(CommitBuilder):
2561
"""Temporary class so old CommitBuilders are detected properly
2562
2563
Note: CommitBuilder works whether or not root entry is recorded.
2564
"""
2565
2566
record_root_entry = True
2567
2568
2569
class RootCommitBuilder(CommitBuilder):
2570
"""This commitbuilder actually records the root id"""
2571
2572
record_root_entry = True
2573
2574
def record_entry_contents(self, ie, parent_invs, path, tree):
2575
"""Record the content of ie from tree into the commit if needed.
2576
2577
Side effect: sets ie.revision when unchanged
2578
2579
:param ie: An inventory entry present in the commit.
2580
:param parent_invs: The inventories of the parent revisions of the
2581
commit.
2582
:param path: The path the entry is at in the tree.
2583
:param tree: The tree which contains this entry and should be used to
2584
obtain content.
2585
"""
2586
assert self.new_inventory.root is not None or ie.parent_id is None
2587
self.new_inventory.add(ie)
2588
2589
# ie.revision is always None if the InventoryEntry is considered
2590
# for committing. ie.snapshot will record the correct revision
2591
# which may be the sole parent if it is untouched.
2592
if ie.revision is not None:
2593
return
2594
2595
previous_entries = ie.find_previous_heads(
2596
parent_invs,
2597
self.repository.weave_store,
2598
self.repository.get_transaction())
2599
# we are creating a new revision for ie in the history store
2600
# and inventory.
2601
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2602
2603
2604
_unescape_map = {
2605
'apos':"'",
2606
'quot':'"',
2607
'amp':'&',
2608
'lt':'<',
2609
'gt':'>'
2610
}
2611
2612
2613
def _unescaper(match, _map=_unescape_map):
2614
return _map[match.group(1)]
2615
2616
2617
_unescape_re = None
2618
2619
2620
def _unescape_xml(data):
2621
"""Unescape predefined XML entities in a string of data."""
2622
global _unescape_re
2623
if _unescape_re is None:
2624
_unescape_re = re.compile('\&([^;]*);')
2625
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0