« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 2227.3.1
- compare with another revision
- download diff
- download tarball
- view history from revision 2227.3.1
- Committer: John Arbash Meinel
- Date: 2007-01-09 19:54:08 UTC
- mto: This revision was merged to the branch mainline in revision 2271.
- Revision ID: john@arbash-meinel.com-20070109195408-1spuyx876hjat226
Allow push to create Branch when necessary, and add --use-existing
- files added:
- BRANCH.TODO
- bzrlib/benchmarks
- bzrlib/benchmarks/tree_creator
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/doc
- bzrlib/doc/api
- bzrlib/export
- bzrlib/plugins
- bzrlib/plugins/launchpad
- bzrlib/store
- bzrlib/store/revision
- bzrlib/store/versioned
- bzrlib/tests
- bzrlib/tests/blackbox
- bzrlib/tests/branch_implementations
- bzrlib/tests/bzrdir_implementations
- bzrlib/tests/interrepository_implementations
- bzrlib/tests/intertree_implementations
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/repository_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/tests/test_patches_data
- bzrlib/tests/tree_implementations
- bzrlib/tests/workingtree_implementations
- bzrlib/transport
- bzrlib/transport/http
- bzrlib/ui
- bzrlib/util
- bzrlib/util/configobj
- bzrlib/util/configobj/docs
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- bzrlib/version_info_formats
- contrib/emacs
- tools
- tools/doc_generate
- tools/win32
- files removed:
- bzrlib/cache.py
- notes
- urlgrabber
- files renamed:
- bzrlib/store.py => bzrlib/store/__init__.py
- elementtree/ => bzrlib/util/elementtree/
- bzrlib/xml.py => bzrlib/xml_serializer.py
- contrib/bash/bzr => contrib/bash/bzr.simple
- bzrlib/newinventory.py => contrib/newinventory.py
- files modified:
- .bzrignore
- bzr *
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
#
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
7
#
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
12
#
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
17
from cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
23
import re
24
import time
25
import unittest
26
27
from bzrlib import (
28
bzrdir,
29
check,
30
delta,
31
errors,
32
generate_ids,
33
gpg,
34
graph,
35
knit,
36
lazy_regex,
37
lockable_files,
38
lockdir,
39
osutils,
40
revision as _mod_revision,
41
symbol_versioning,
42
transactions,
43
ui,
44
weave,
45
weavefile,
46
xml5,
47
xml6,
48
)
49
from bzrlib.osutils import (
50
rand_bytes,
51
compact_date,
52
local_time_offset,
53
)
54
from bzrlib.revisiontree import RevisionTree
55
from bzrlib.store.versioned import VersionedFileStore
56
from bzrlib.store.text import TextStore
57
from bzrlib.testament import Testament
58
""")
59
60
from bzrlib.decorators import needs_read_lock, needs_write_lock
61
from bzrlib.inter import InterObject
62
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
63
from bzrlib.symbol_versioning import (
64
deprecated_method,
65
zero_nine,
66
)
67
from bzrlib.trace import mutter, note, warning
68
69
70
# Old formats display a warning, but only once
71
_deprecation_warning_done = False
72
73
74
class Repository(object):
75
"""Repository holding history for one or more branches.
76
77
The repository holds and retrieves historical information including
78
revisions and file history. It's normally accessed only by the Branch,
79
which views a particular line of development through that history.
80
81
The Repository builds on top of Stores and a Transport, which respectively
82
describe the disk data format and the way of accessing the (possibly
83
remote) disk.
84
"""
85
86
_file_ids_altered_regex = lazy_regex.lazy_compile(
87
r'file_id="(?P<file_id>[^"]+)"'
88
r'.*revision="(?P<revision_id>[^"]+)"'
89
)
90
91
@needs_write_lock
92
def add_inventory(self, revid, inv, parents):
93
"""Add the inventory inv to the repository as revid.
94
95
:param parents: The revision ids of the parents that revid
96
is known to have and are in the repository already.
97
98
returns the sha1 of the serialized inventory.
99
"""
100
assert inv.revision_id is None or inv.revision_id == revid, \
101
"Mismatch between inventory revision" \
102
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
103
assert inv.root is not None
104
inv_text = self.serialise_inventory(inv)
105
inv_sha1 = osutils.sha_string(inv_text)
106
inv_vf = self.control_weaves.get_weave('inventory',
107
self.get_transaction())
108
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
109
return inv_sha1
110
111
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
112
final_parents = []
113
for parent in parents:
114
if parent in inv_vf:
115
final_parents.append(parent)
116
117
inv_vf.add_lines(revid, final_parents, lines)
118
119
@needs_write_lock
120
def add_revision(self, rev_id, rev, inv=None, config=None):
121
"""Add rev to the revision store as rev_id.
122
123
:param rev_id: the revision id to use.
124
:param rev: The revision object.
125
:param inv: The inventory for the revision. if None, it will be looked
126
up in the inventory storer
127
:param config: If None no digital signature will be created.
128
If supplied its signature_needed method will be used
129
to determine if a signature should be made.
130
"""
131
if config is not None and config.signature_needed():
132
if inv is None:
133
inv = self.get_inventory(rev_id)
134
plaintext = Testament(rev, inv).as_short_text()
135
self.store_revision_signature(
136
gpg.GPGStrategy(config), plaintext, rev_id)
137
if not rev_id in self.get_inventory_weave():
138
if inv is None:
139
raise errors.WeaveRevisionNotPresent(rev_id,
140
self.get_inventory_weave())
141
else:
142
# yes, this is not suitable for adding with ghosts.
143
self.add_inventory(rev_id, inv, rev.parent_ids)
144
self._revision_store.add_revision(rev, self.get_transaction())
145
146
@needs_read_lock
147
def _all_possible_ids(self):
148
"""Return all the possible revisions that we could find."""
149
return self.get_inventory_weave().versions()
150
151
def all_revision_ids(self):
152
"""Returns a list of all the revision ids in the repository.
153
154
This is deprecated because code should generally work on the graph
155
reachable from a particular revision, and ignore any other revisions
156
that might be present. There is no direct replacement method.
157
"""
158
return self._all_revision_ids()
159
160
@needs_read_lock
161
def _all_revision_ids(self):
162
"""Returns a list of all the revision ids in the repository.
163
164
These are in as much topological order as the underlying store can
165
present: for weaves ghosts may lead to a lack of correctness until
166
the reweave updates the parents list.
167
"""
168
if self._revision_store.text_store.listable():
169
return self._revision_store.all_revision_ids(self.get_transaction())
170
result = self._all_possible_ids()
171
return self._eliminate_revisions_not_present(result)
172
173
def break_lock(self):
174
"""Break a lock if one is present from another instance.
175
176
Uses the ui factory to ask for confirmation if the lock may be from
177
an active process.
178
"""
179
self.control_files.break_lock()
180
181
@needs_read_lock
182
def _eliminate_revisions_not_present(self, revision_ids):
183
"""Check every revision id in revision_ids to see if we have it.
184
185
Returns a set of the present revisions.
186
"""
187
result = []
188
for id in revision_ids:
189
if self.has_revision(id):
190
result.append(id)
191
return result
192
193
@staticmethod
194
def create(a_bzrdir):
195
"""Construct the current default format repository in a_bzrdir."""
196
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
197
198
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
199
"""instantiate a Repository.
200
201
:param _format: The format of the repository on disk.
202
:param a_bzrdir: The BzrDir of the repository.
203
204
In the future we will have a single api for all stores for
205
getting file texts, inventories and revisions, then
206
this construct will accept instances of those things.
207
"""
208
super(Repository, self).__init__()
209
self._format = _format
210
# the following are part of the public API for Repository:
211
self.bzrdir = a_bzrdir
212
self.control_files = control_files
213
self._revision_store = _revision_store
214
self.text_store = text_store
215
# backwards compatibility
216
self.weave_store = text_store
217
# not right yet - should be more semantically clear ?
218
#
219
self.control_store = control_store
220
self.control_weaves = control_store
221
# TODO: make sure to construct the right store classes, etc, depending
222
# on whether escaping is required.
223
self._warn_if_deprecated()
224
self._serializer = xml5.serializer_v5
225
226
def __repr__(self):
227
return '%s(%r)' % (self.__class__.__name__,
228
self.bzrdir.transport.base)
229
230
def is_locked(self):
231
return self.control_files.is_locked()
232
233
def lock_write(self):
234
self.control_files.lock_write()
235
236
def lock_read(self):
237
self.control_files.lock_read()
238
239
def get_physical_lock_status(self):
240
return self.control_files.get_physical_lock_status()
241
242
@needs_read_lock
243
def missing_revision_ids(self, other, revision_id=None):
244
"""Return the revision ids that other has that this does not.
245
246
These are returned in topological order.
247
248
revision_id: only return revision ids included by revision_id.
249
"""
250
return InterRepository.get(other, self).missing_revision_ids(revision_id)
251
252
@staticmethod
253
def open(base):
254
"""Open the repository rooted at base.
255
256
For instance, if the repository is at URL/.bzr/repository,
257
Repository.open(URL) -> a Repository instance.
258
"""
259
control = bzrdir.BzrDir.open(base)
260
return control.open_repository()
261
262
def copy_content_into(self, destination, revision_id=None, basis=None):
263
"""Make a complete copy of the content in self into destination.
264
265
This is a destructive operation! Do not use it on existing
266
repositories.
267
"""
268
return InterRepository.get(self, destination).copy_content(revision_id, basis)
269
270
def fetch(self, source, revision_id=None, pb=None):
271
"""Fetch the content required to construct revision_id from source.
272
273
If revision_id is None all content is copied.
274
"""
275
return InterRepository.get(source, self).fetch(revision_id=revision_id,
276
pb=pb)
277
278
def get_commit_builder(self, branch, parents, config, timestamp=None,
279
timezone=None, committer=None, revprops=None,
280
revision_id=None):
281
"""Obtain a CommitBuilder for this repository.
282
283
:param branch: Branch to commit to.
284
:param parents: Revision ids of the parents of the new revision.
285
:param config: Configuration to use.
286
:param timestamp: Optional timestamp recorded for commit.
287
:param timezone: Optional timezone for timestamp.
288
:param committer: Optional committer to set for commit.
289
:param revprops: Optional dictionary of revision properties.
290
:param revision_id: Optional revision id.
291
"""
292
return _CommitBuilder(self, parents, config, timestamp, timezone,
293
committer, revprops, revision_id)
294
295
def unlock(self):
296
self.control_files.unlock()
297
298
@needs_read_lock
299
def clone(self, a_bzrdir, revision_id=None, basis=None):
300
"""Clone this repository into a_bzrdir using the current format.
301
302
Currently no check is made that the format of this repository and
303
the bzrdir format are compatible. FIXME RBC 20060201.
304
"""
305
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
306
# use target default format.
307
result = a_bzrdir.create_repository()
308
# FIXME RBC 20060209 split out the repository type to avoid this check ?
309
elif isinstance(a_bzrdir._format,
310
(bzrdir.BzrDirFormat4,
311
bzrdir.BzrDirFormat5,
312
bzrdir.BzrDirFormat6)):
313
result = a_bzrdir.open_repository()
314
else:
315
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
316
self.copy_content_into(result, revision_id, basis)
317
return result
318
319
@needs_read_lock
320
def has_revision(self, revision_id):
321
"""True if this repository has a copy of the revision."""
322
return self._revision_store.has_revision_id(revision_id,
323
self.get_transaction())
324
325
@needs_read_lock
326
def get_revision_reconcile(self, revision_id):
327
"""'reconcile' helper routine that allows access to a revision always.
328
329
This variant of get_revision does not cross check the weave graph
330
against the revision one as get_revision does: but it should only
331
be used by reconcile, or reconcile-alike commands that are correcting
332
or testing the revision graph.
333
"""
334
if not revision_id or not isinstance(revision_id, basestring):
335
raise errors.InvalidRevisionId(revision_id=revision_id,
336
branch=self)
337
return self._revision_store.get_revisions([revision_id],
338
self.get_transaction())[0]
339
@needs_read_lock
340
def get_revisions(self, revision_ids):
341
return self._revision_store.get_revisions(revision_ids,
342
self.get_transaction())
343
344
@needs_read_lock
345
def get_revision_xml(self, revision_id):
346
rev = self.get_revision(revision_id)
347
rev_tmp = StringIO()
348
# the current serializer..
349
self._revision_store._serializer.write_revision(rev, rev_tmp)
350
rev_tmp.seek(0)
351
return rev_tmp.getvalue()
352
353
@needs_read_lock
354
def get_revision(self, revision_id):
355
"""Return the Revision object for a named revision"""
356
r = self.get_revision_reconcile(revision_id)
357
# weave corruption can lead to absent revision markers that should be
358
# present.
359
# the following test is reasonably cheap (it needs a single weave read)
360
# and the weave is cached in read transactions. In write transactions
361
# it is not cached but typically we only read a small number of
362
# revisions. For knits when they are introduced we will probably want
363
# to ensure that caching write transactions are in use.
364
inv = self.get_inventory_weave()
365
self._check_revision_parents(r, inv)
366
return r
367
368
@needs_read_lock
369
def get_deltas_for_revisions(self, revisions):
370
"""Produce a generator of revision deltas.
371
372
Note that the input is a sequence of REVISIONS, not revision_ids.
373
Trees will be held in memory until the generator exits.
374
Each delta is relative to the revision's lefthand predecessor.
375
"""
376
required_trees = set()
377
for revision in revisions:
378
required_trees.add(revision.revision_id)
379
required_trees.update(revision.parent_ids[:1])
380
trees = dict((t.get_revision_id(), t) for
381
t in self.revision_trees(required_trees))
382
for revision in revisions:
383
if not revision.parent_ids:
384
old_tree = self.revision_tree(None)
385
else:
386
old_tree = trees[revision.parent_ids[0]]
387
yield trees[revision.revision_id].changes_from(old_tree)
388
389
@needs_read_lock
390
def get_revision_delta(self, revision_id):
391
"""Return the delta for one revision.
392
393
The delta is relative to the left-hand predecessor of the
394
revision.
395
"""
396
r = self.get_revision(revision_id)
397
return list(self.get_deltas_for_revisions([r]))[0]
398
399
def _check_revision_parents(self, revision, inventory):
400
"""Private to Repository and Fetch.
401
402
This checks the parentage of revision in an inventory weave for
403
consistency and is only applicable to inventory-weave-for-ancestry
404
using repository formats & fetchers.
405
"""
406
weave_parents = inventory.get_parents(revision.revision_id)
407
weave_names = inventory.versions()
408
for parent_id in revision.parent_ids:
409
if parent_id in weave_names:
410
# this parent must not be a ghost.
411
if not parent_id in weave_parents:
412
# but it is a ghost
413
raise errors.CorruptRepository(self)
414
415
@needs_write_lock
416
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
417
signature = gpg_strategy.sign(plaintext)
418
self._revision_store.add_revision_signature_text(revision_id,
419
signature,
420
self.get_transaction())
421
422
def fileids_altered_by_revision_ids(self, revision_ids):
423
"""Find the file ids and versions affected by revisions.
424
425
:param revisions: an iterable containing revision ids.
426
:return: a dictionary mapping altered file-ids to an iterable of
427
revision_ids. Each altered file-ids has the exact revision_ids that
428
altered it listed explicitly.
429
"""
430
assert self._serializer.support_altered_by_hack, \
431
("fileids_altered_by_revision_ids only supported for branches "
432
"which store inventory as unnested xml, not on %r" % self)
433
selected_revision_ids = set(revision_ids)
434
w = self.get_inventory_weave()
435
result = {}
436
437
# this code needs to read every new line in every inventory for the
438
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
439
# not present in one of those inventories is unnecessary but not
440
# harmful because we are filtering by the revision id marker in the
441
# inventory lines : we only select file ids altered in one of those
442
# revisions. We don't need to see all lines in the inventory because
443
# only those added in an inventory in rev X can contain a revision=X
444
# line.
445
unescape_revid_cache = {}
446
unescape_fileid_cache = {}
447
448
# jam 20061218 In a big fetch, this handles hundreds of thousands
449
# of lines, so it has had a lot of inlining and optimizing done.
450
# Sorry that it is a little bit messy.
451
# Move several functions to be local variables, since this is a long
452
# running loop.
453
search = self._file_ids_altered_regex.search
454
unescape = _unescape_xml
455
setdefault = result.setdefault
456
pb = ui.ui_factory.nested_progress_bar()
457
try:
458
for line in w.iter_lines_added_or_present_in_versions(
459
selected_revision_ids, pb=pb):
460
match = search(line)
461
if match is None:
462
continue
463
# One call to match.group() returning multiple items is quite a
464
# bit faster than 2 calls to match.group() each returning 1
465
file_id, revision_id = match.group('file_id', 'revision_id')
466
467
# Inlining the cache lookups helps a lot when you make 170,000
468
# lines and 350k ids, versus 8.4 unique ids.
469
# Using a cache helps in 2 ways:
470
# 1) Avoids unnecessary decoding calls
471
# 2) Re-uses cached strings, which helps in future set and
472
# equality checks.
473
# (2) is enough that removing encoding entirely along with
474
# the cache (so we are using plain strings) results in no
475
# performance improvement.
476
try:
477
revision_id = unescape_revid_cache[revision_id]
478
except KeyError:
479
unescaped = unescape(revision_id)
480
unescape_revid_cache[revision_id] = unescaped
481
revision_id = unescaped
482
483
if revision_id in selected_revision_ids:
484
try:
485
file_id = unescape_fileid_cache[file_id]
486
except KeyError:
487
unescaped = unescape(file_id)
488
unescape_fileid_cache[file_id] = unescaped
489
file_id = unescaped
490
setdefault(file_id, set()).add(revision_id)
491
finally:
492
pb.finished()
493
return result
494
495
@needs_read_lock
496
def get_inventory_weave(self):
497
return self.control_weaves.get_weave('inventory',
498
self.get_transaction())
499
500
@needs_read_lock
501
def get_inventory(self, revision_id):
502
"""Get Inventory object by hash."""
503
return self.deserialise_inventory(
504
revision_id, self.get_inventory_xml(revision_id))
505
506
def deserialise_inventory(self, revision_id, xml):
507
"""Transform the xml into an inventory object.
508
509
:param revision_id: The expected revision id of the inventory.
510
:param xml: A serialised inventory.
511
"""
512
result = self._serializer.read_inventory_from_string(xml)
513
result.root.revision = revision_id
514
return result
515
516
def serialise_inventory(self, inv):
517
return self._serializer.write_inventory_to_string(inv)
518
519
@needs_read_lock
520
def get_inventory_xml(self, revision_id):
521
"""Get inventory XML as a file object."""
522
try:
523
assert isinstance(revision_id, basestring), type(revision_id)
524
iw = self.get_inventory_weave()
525
return iw.get_text(revision_id)
526
except IndexError:
527
raise errors.HistoryMissing(self, 'inventory', revision_id)
528
529
@needs_read_lock
530
def get_inventory_sha1(self, revision_id):
531
"""Return the sha1 hash of the inventory entry
532
"""
533
return self.get_revision(revision_id).inventory_sha1
534
535
@needs_read_lock
536
def get_revision_graph(self, revision_id=None):
537
"""Return a dictionary containing the revision graph.
538
539
:param revision_id: The revision_id to get a graph from. If None, then
540
the entire revision graph is returned. This is a deprecated mode of
541
operation and will be removed in the future.
542
:return: a dictionary of revision_id->revision_parents_list.
543
"""
544
# special case NULL_REVISION
545
if revision_id == _mod_revision.NULL_REVISION:
546
return {}
547
a_weave = self.get_inventory_weave()
548
all_revisions = self._eliminate_revisions_not_present(
549
a_weave.versions())
550
entire_graph = dict([(node, a_weave.get_parents(node)) for
551
node in all_revisions])
552
if revision_id is None:
553
return entire_graph
554
elif revision_id not in entire_graph:
555
raise errors.NoSuchRevision(self, revision_id)
556
else:
557
# add what can be reached from revision_id
558
result = {}
559
pending = set([revision_id])
560
while len(pending) > 0:
561
node = pending.pop()
562
result[node] = entire_graph[node]
563
for revision_id in result[node]:
564
if revision_id not in result:
565
pending.add(revision_id)
566
return result
567
568
@needs_read_lock
569
def get_revision_graph_with_ghosts(self, revision_ids=None):
570
"""Return a graph of the revisions with ghosts marked as applicable.
571
572
:param revision_ids: an iterable of revisions to graph or None for all.
573
:return: a Graph object with the graph reachable from revision_ids.
574
"""
575
result = graph.Graph()
576
if not revision_ids:
577
pending = set(self.all_revision_ids())
578
required = set([])
579
else:
580
pending = set(revision_ids)
581
# special case NULL_REVISION
582
if _mod_revision.NULL_REVISION in pending:
583
pending.remove(_mod_revision.NULL_REVISION)
584
required = set(pending)
585
done = set([])
586
while len(pending):
587
revision_id = pending.pop()
588
try:
589
rev = self.get_revision(revision_id)
590
except errors.NoSuchRevision:
591
if revision_id in required:
592
raise
593
# a ghost
594
result.add_ghost(revision_id)
595
continue
596
for parent_id in rev.parent_ids:
597
# is this queued or done ?
598
if (parent_id not in pending and
599
parent_id not in done):
600
# no, queue it.
601
pending.add(parent_id)
602
result.add_node(revision_id, rev.parent_ids)
603
done.add(revision_id)
604
return result
605
606
@needs_read_lock
607
def get_revision_inventory(self, revision_id):
608
"""Return inventory of a past revision."""
609
# TODO: Unify this with get_inventory()
610
# bzr 0.0.6 and later imposes the constraint that the inventory_id
611
# must be the same as its revision, so this is trivial.
612
if revision_id is None:
613
# This does not make sense: if there is no revision,
614
# then it is the current tree inventory surely ?!
615
# and thus get_root_id() is something that looks at the last
616
# commit on the branch, and the get_root_id is an inventory check.
617
raise NotImplementedError
618
# return Inventory(self.get_root_id())
619
else:
620
return self.get_inventory(revision_id)
621
622
@needs_read_lock
623
def is_shared(self):
624
"""Return True if this repository is flagged as a shared repository."""
625
raise NotImplementedError(self.is_shared)
626
627
@needs_write_lock
628
def reconcile(self, other=None, thorough=False):
629
"""Reconcile this repository."""
630
from bzrlib.reconcile import RepoReconciler
631
reconciler = RepoReconciler(self, thorough=thorough)
632
reconciler.reconcile()
633
return reconciler
634
635
@needs_read_lock
636
def revision_tree(self, revision_id):
637
"""Return Tree for a revision on this branch.
638
639
`revision_id` may be None for the empty tree revision.
640
"""
641
# TODO: refactor this to use an existing revision object
642
# so we don't need to read it in twice.
643
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
644
return RevisionTree(self, Inventory(root_id=None),
645
_mod_revision.NULL_REVISION)
646
else:
647
inv = self.get_revision_inventory(revision_id)
648
return RevisionTree(self, inv, revision_id)
649
650
@needs_read_lock
651
def revision_trees(self, revision_ids):
652
"""Return Tree for a revision on this branch.
653
654
`revision_id` may not be None or 'null:'"""
655
assert None not in revision_ids
656
assert _mod_revision.NULL_REVISION not in revision_ids
657
texts = self.get_inventory_weave().get_texts(revision_ids)
658
for text, revision_id in zip(texts, revision_ids):
659
inv = self.deserialise_inventory(revision_id, text)
660
yield RevisionTree(self, inv, revision_id)
661
662
@needs_read_lock
663
def get_ancestry(self, revision_id):
664
"""Return a list of revision-ids integrated by a revision.
665
666
The first element of the list is always None, indicating the origin
667
revision. This might change when we have history horizons, or
668
perhaps we should have a new API.
669
670
This is topologically sorted.
671
"""
672
if revision_id is None:
673
return [None]
674
if not self.has_revision(revision_id):
675
raise errors.NoSuchRevision(self, revision_id)
676
w = self.get_inventory_weave()
677
candidates = w.get_ancestry(revision_id)
678
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
679
680
@needs_read_lock
681
def print_file(self, file, revision_id):
682
"""Print `file` to stdout.
683
684
FIXME RBC 20060125 as John Meinel points out this is a bad api
685
- it writes to stdout, it assumes that that is valid etc. Fix
686
by creating a new more flexible convenience function.
687
"""
688
tree = self.revision_tree(revision_id)
689
# use inventory as it was in that revision
690
file_id = tree.inventory.path2id(file)
691
if not file_id:
692
# TODO: jam 20060427 Write a test for this code path
693
# it had a bug in it, and was raising the wrong
694
# exception.
695
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
696
tree.print_file(file_id)
697
698
def get_transaction(self):
699
return self.control_files.get_transaction()
700
701
def revision_parents(self, revid):
702
return self.get_inventory_weave().parent_names(revid)
703
704
@needs_write_lock
705
def set_make_working_trees(self, new_value):
706
"""Set the policy flag for making working trees when creating branches.
707
708
This only applies to branches that use this repository.
709
710
The default is 'True'.
711
:param new_value: True to restore the default, False to disable making
712
working trees.
713
"""
714
raise NotImplementedError(self.set_make_working_trees)
715
716
def make_working_trees(self):
717
"""Returns the policy for making working trees on new branches."""
718
raise NotImplementedError(self.make_working_trees)
719
720
@needs_write_lock
721
def sign_revision(self, revision_id, gpg_strategy):
722
plaintext = Testament.from_revision(self, revision_id).as_short_text()
723
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
724
725
@needs_read_lock
726
def has_signature_for_revision_id(self, revision_id):
727
"""Query for a revision signature for revision_id in the repository."""
728
return self._revision_store.has_signature(revision_id,
729
self.get_transaction())
730
731
@needs_read_lock
732
def get_signature_text(self, revision_id):
733
"""Return the text for a signature."""
734
return self._revision_store.get_signature_text(revision_id,
735
self.get_transaction())
736
737
@needs_read_lock
738
def check(self, revision_ids):
739
"""Check consistency of all history of given revision_ids.
740
741
Different repository implementations should override _check().
742
743
:param revision_ids: A non-empty list of revision_ids whose ancestry
744
will be checked. Typically the last revision_id of a branch.
745
"""
746
if not revision_ids:
747
raise ValueError("revision_ids must be non-empty in %s.check"
748
% (self,))
749
return self._check(revision_ids)
750
751
def _check(self, revision_ids):
752
result = check.Check(self)
753
result.check()
754
return result
755
756
def _warn_if_deprecated(self):
757
global _deprecation_warning_done
758
if _deprecation_warning_done:
759
return
760
_deprecation_warning_done = True
761
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
762
% (self._format, self.bzrdir.transport.base))
763
764
def supports_rich_root(self):
765
return self._format.rich_root_data
766
767
def _check_ascii_revisionid(self, revision_id, method):
768
"""Private helper for ascii-only repositories."""
769
# weave repositories refuse to store revisionids that are non-ascii.
770
if revision_id is not None:
771
# weaves require ascii revision ids.
772
if isinstance(revision_id, unicode):
773
try:
774
revision_id.encode('ascii')
775
except UnicodeEncodeError:
776
raise errors.NonAsciiRevisionId(method, self)
777
778
779
class AllInOneRepository(Repository):
780
"""Legacy support - the repository behaviour for all-in-one branches."""
781
782
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
783
# we reuse one control files instance.
784
dir_mode = a_bzrdir._control_files._dir_mode
785
file_mode = a_bzrdir._control_files._file_mode
786
787
def get_store(name, compressed=True, prefixed=False):
788
# FIXME: This approach of assuming stores are all entirely compressed
789
# or entirely uncompressed is tidy, but breaks upgrade from
790
# some existing branches where there's a mixture; we probably
791
# still want the option to look for both.
792
relpath = a_bzrdir._control_files._escape(name)
793
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
794
prefixed=prefixed, compressed=compressed,
795
dir_mode=dir_mode,
796
file_mode=file_mode)
797
#if self._transport.should_cache():
798
# cache_path = os.path.join(self.cache_root, name)
799
# os.mkdir(cache_path)
800
# store = bzrlib.store.CachedStore(store, cache_path)
801
return store
802
803
# not broken out yet because the controlweaves|inventory_store
804
# and text_store | weave_store bits are still different.
805
if isinstance(_format, RepositoryFormat4):
806
# cannot remove these - there is still no consistent api
807
# which allows access to this old info.
808
self.inventory_store = get_store('inventory-store')
809
text_store = get_store('text-store')
810
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
811
812
def get_commit_builder(self, branch, parents, config, timestamp=None,
813
timezone=None, committer=None, revprops=None,
814
revision_id=None):
815
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
816
return Repository.get_commit_builder(self, branch, parents, config,
817
timestamp, timezone, committer, revprops, revision_id)
818
819
@needs_read_lock
820
def is_shared(self):
821
"""AllInOne repositories cannot be shared."""
822
return False
823
824
@needs_write_lock
825
def set_make_working_trees(self, new_value):
826
"""Set the policy flag for making working trees when creating branches.
827
828
This only applies to branches that use this repository.
829
830
The default is 'True'.
831
:param new_value: True to restore the default, False to disable making
832
working trees.
833
"""
834
raise NotImplementedError(self.set_make_working_trees)
835
836
def make_working_trees(self):
837
"""Returns the policy for making working trees on new branches."""
838
return True
839
840
841
def install_revision(repository, rev, revision_tree):
842
"""Install all revision data into a repository."""
843
present_parents = []
844
parent_trees = {}
845
for p_id in rev.parent_ids:
846
if repository.has_revision(p_id):
847
present_parents.append(p_id)
848
parent_trees[p_id] = repository.revision_tree(p_id)
849
else:
850
parent_trees[p_id] = repository.revision_tree(None)
851
852
inv = revision_tree.inventory
853
entries = inv.iter_entries()
854
# backwards compatability hack: skip the root id.
855
if not repository.supports_rich_root():
856
path, root = entries.next()
857
if root.revision != rev.revision_id:
858
raise errors.IncompatibleRevision(repr(repository))
859
# Add the texts that are not already present
860
for path, ie in entries:
861
w = repository.weave_store.get_weave_or_empty(ie.file_id,
862
repository.get_transaction())
863
if ie.revision not in w:
864
text_parents = []
865
# FIXME: TODO: The following loop *may* be overlapping/duplicate
866
# with InventoryEntry.find_previous_heads(). if it is, then there
867
# is a latent bug here where the parents may have ancestors of each
868
# other. RBC, AB
869
for revision, tree in parent_trees.iteritems():
870
if ie.file_id not in tree:
871
continue
872
parent_id = tree.inventory[ie.file_id].revision
873
if parent_id in text_parents:
874
continue
875
text_parents.append(parent_id)
876
877
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
878
repository.get_transaction())
879
lines = revision_tree.get_file(ie.file_id).readlines()
880
vfile.add_lines(rev.revision_id, text_parents, lines)
881
try:
882
# install the inventory
883
repository.add_inventory(rev.revision_id, inv, present_parents)
884
except errors.RevisionAlreadyPresent:
885
pass
886
repository.add_revision(rev.revision_id, rev, inv)
887
888
889
class MetaDirRepository(Repository):
890
"""Repositories in the new meta-dir layout."""
891
892
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
893
super(MetaDirRepository, self).__init__(_format,
894
a_bzrdir,
895
control_files,
896
_revision_store,
897
control_store,
898
text_store)
899
dir_mode = self.control_files._dir_mode
900
file_mode = self.control_files._file_mode
901
902
@needs_read_lock
903
def is_shared(self):
904
"""Return True if this repository is flagged as a shared repository."""
905
return self.control_files._transport.has('shared-storage')
906
907
@needs_write_lock
908
def set_make_working_trees(self, new_value):
909
"""Set the policy flag for making working trees when creating branches.
910
911
This only applies to branches that use this repository.
912
913
The default is 'True'.
914
:param new_value: True to restore the default, False to disable making
915
working trees.
916
"""
917
if new_value:
918
try:
919
self.control_files._transport.delete('no-working-trees')
920
except errors.NoSuchFile:
921
pass
922
else:
923
self.control_files.put_utf8('no-working-trees', '')
924
925
def make_working_trees(self):
926
"""Returns the policy for making working trees on new branches."""
927
return not self.control_files._transport.has('no-working-trees')
928
929
930
class WeaveMetaDirRepository(MetaDirRepository):
931
"""A subclass of MetaDirRepository to set weave specific policy."""
932
933
def get_commit_builder(self, branch, parents, config, timestamp=None,
934
timezone=None, committer=None, revprops=None,
935
revision_id=None):
936
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
937
return MetaDirRepository.get_commit_builder(self, branch, parents,
938
config, timestamp, timezone, committer, revprops, revision_id)
939
940
941
class KnitRepository(MetaDirRepository):
942
"""Knit format repository."""
943
944
def _warn_if_deprecated(self):
945
# This class isn't deprecated
946
pass
947
948
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
949
inv_vf.add_lines_with_ghosts(revid, parents, lines)
950
951
@needs_read_lock
952
def _all_revision_ids(self):
953
"""See Repository.all_revision_ids()."""
954
# Knits get the revision graph from the index of the revision knit, so
955
# it's always possible even if they're on an unlistable transport.
956
return self._revision_store.all_revision_ids(self.get_transaction())
957
958
def fileid_involved_between_revs(self, from_revid, to_revid):
959
"""Find file_id(s) which are involved in the changes between revisions.
960
961
This determines the set of revisions which are involved, and then
962
finds all file ids affected by those revisions.
963
"""
964
vf = self._get_revision_vf()
965
from_set = set(vf.get_ancestry(from_revid))
966
to_set = set(vf.get_ancestry(to_revid))
967
changed = to_set.difference(from_set)
968
return self._fileid_involved_by_set(changed)
969
970
def fileid_involved(self, last_revid=None):
971
"""Find all file_ids modified in the ancestry of last_revid.
972
973
:param last_revid: If None, last_revision() will be used.
974
"""
975
if not last_revid:
976
changed = set(self.all_revision_ids())
977
else:
978
changed = set(self.get_ancestry(last_revid))
979
if None in changed:
980
changed.remove(None)
981
return self._fileid_involved_by_set(changed)
982
983
@needs_read_lock
984
def get_ancestry(self, revision_id):
985
"""Return a list of revision-ids integrated by a revision.
986
987
This is topologically sorted.
988
"""
989
if revision_id is None:
990
return [None]
991
vf = self._get_revision_vf()
992
try:
993
return [None] + vf.get_ancestry(revision_id)
994
except errors.RevisionNotPresent:
995
raise errors.NoSuchRevision(self, revision_id)
996
997
@needs_read_lock
998
def get_revision(self, revision_id):
999
"""Return the Revision object for a named revision"""
1000
return self.get_revision_reconcile(revision_id)
1001
1002
@needs_read_lock
1003
def get_revision_graph(self, revision_id=None):
1004
"""Return a dictionary containing the revision graph.
1005
1006
:param revision_id: The revision_id to get a graph from. If None, then
1007
the entire revision graph is returned. This is a deprecated mode of
1008
operation and will be removed in the future.
1009
:return: a dictionary of revision_id->revision_parents_list.
1010
"""
1011
# special case NULL_REVISION
1012
if revision_id == _mod_revision.NULL_REVISION:
1013
return {}
1014
a_weave = self._get_revision_vf()
1015
entire_graph = a_weave.get_graph()
1016
if revision_id is None:
1017
return a_weave.get_graph()
1018
elif revision_id not in a_weave:
1019
raise errors.NoSuchRevision(self, revision_id)
1020
else:
1021
# add what can be reached from revision_id
1022
result = {}
1023
pending = set([revision_id])
1024
while len(pending) > 0:
1025
node = pending.pop()
1026
result[node] = a_weave.get_parents(node)
1027
for revision_id in result[node]:
1028
if revision_id not in result:
1029
pending.add(revision_id)
1030
return result
1031
1032
@needs_read_lock
1033
def get_revision_graph_with_ghosts(self, revision_ids=None):
1034
"""Return a graph of the revisions with ghosts marked as applicable.
1035
1036
:param revision_ids: an iterable of revisions to graph or None for all.
1037
:return: a Graph object with the graph reachable from revision_ids.
1038
"""
1039
result = graph.Graph()
1040
vf = self._get_revision_vf()
1041
versions = set(vf.versions())
1042
if not revision_ids:
1043
pending = set(self.all_revision_ids())
1044
required = set([])
1045
else:
1046
pending = set(revision_ids)
1047
# special case NULL_REVISION
1048
if _mod_revision.NULL_REVISION in pending:
1049
pending.remove(_mod_revision.NULL_REVISION)
1050
required = set(pending)
1051
done = set([])
1052
while len(pending):
1053
revision_id = pending.pop()
1054
if not revision_id in versions:
1055
if revision_id in required:
1056
raise errors.NoSuchRevision(self, revision_id)
1057
# a ghost
1058
result.add_ghost(revision_id)
1059
# mark it as done so we don't try for it again.
1060
done.add(revision_id)
1061
continue
1062
parent_ids = vf.get_parents_with_ghosts(revision_id)
1063
for parent_id in parent_ids:
1064
# is this queued or done ?
1065
if (parent_id not in pending and
1066
parent_id not in done):
1067
# no, queue it.
1068
pending.add(parent_id)
1069
result.add_node(revision_id, parent_ids)
1070
done.add(revision_id)
1071
return result
1072
1073
def _get_revision_vf(self):
1074
""":return: a versioned file containing the revisions."""
1075
vf = self._revision_store.get_revision_file(self.get_transaction())
1076
return vf
1077
1078
@needs_write_lock
1079
def reconcile(self, other=None, thorough=False):
1080
"""Reconcile this repository."""
1081
from bzrlib.reconcile import KnitReconciler
1082
reconciler = KnitReconciler(self, thorough=thorough)
1083
reconciler.reconcile()
1084
return reconciler
1085
1086
def revision_parents(self, revision_id):
1087
return self._get_revision_vf().get_parents(revision_id)
1088
1089
1090
class KnitRepository2(KnitRepository):
1091
""""""
1092
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1093
control_store, text_store):
1094
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1095
_revision_store, control_store, text_store)
1096
self._serializer = xml6.serializer_v6
1097
1098
def deserialise_inventory(self, revision_id, xml):
1099
"""Transform the xml into an inventory object.
1100
1101
:param revision_id: The expected revision id of the inventory.
1102
:param xml: A serialised inventory.
1103
"""
1104
result = self._serializer.read_inventory_from_string(xml)
1105
assert result.root.revision is not None
1106
return result
1107
1108
def serialise_inventory(self, inv):
1109
"""Transform the inventory object into XML text.
1110
1111
:param revision_id: The expected revision id of the inventory.
1112
:param xml: A serialised inventory.
1113
"""
1114
assert inv.revision_id is not None
1115
assert inv.root.revision is not None
1116
return KnitRepository.serialise_inventory(self, inv)
1117
1118
def get_commit_builder(self, branch, parents, config, timestamp=None,
1119
timezone=None, committer=None, revprops=None,
1120
revision_id=None):
1121
"""Obtain a CommitBuilder for this repository.
1122
1123
:param branch: Branch to commit to.
1124
:param parents: Revision ids of the parents of the new revision.
1125
:param config: Configuration to use.
1126
:param timestamp: Optional timestamp recorded for commit.
1127
:param timezone: Optional timezone for timestamp.
1128
:param committer: Optional committer to set for commit.
1129
:param revprops: Optional dictionary of revision properties.
1130
:param revision_id: Optional revision id.
1131
"""
1132
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1133
committer, revprops, revision_id)
1134
1135
1136
class RepositoryFormat(object):
1137
"""A repository format.
1138
1139
Formats provide three things:
1140
* An initialization routine to construct repository data on disk.
1141
* a format string which is used when the BzrDir supports versioned
1142
children.
1143
* an open routine which returns a Repository instance.
1144
1145
Formats are placed in an dict by their format string for reference
1146
during opening. These should be subclasses of RepositoryFormat
1147
for consistency.
1148
1149
Once a format is deprecated, just deprecate the initialize and open
1150
methods on the format class. Do not deprecate the object, as the
1151
object will be created every system load.
1152
1153
Common instance attributes:
1154
_matchingbzrdir - the bzrdir format that the repository format was
1155
originally written to work with. This can be used if manually
1156
constructing a bzrdir and repository, or more commonly for test suite
1157
parameterisation.
1158
"""
1159
1160
_default_format = None
1161
"""The default format used for new repositories."""
1162
1163
_formats = {}
1164
"""The known formats."""
1165
1166
def __str__(self):
1167
return "<%s>" % self.__class__.__name__
1168
1169
@classmethod
1170
def find_format(klass, a_bzrdir):
1171
"""Return the format for the repository object in a_bzrdir."""
1172
try:
1173
transport = a_bzrdir.get_repository_transport(None)
1174
format_string = transport.get("format").read()
1175
return klass._formats[format_string]
1176
except errors.NoSuchFile:
1177
raise errors.NoRepositoryPresent(a_bzrdir)
1178
except KeyError:
1179
raise errors.UnknownFormatError(format=format_string)
1180
1181
def _get_control_store(self, repo_transport, control_files):
1182
"""Return the control store for this repository."""
1183
raise NotImplementedError(self._get_control_store)
1184
1185
@classmethod
1186
def get_default_format(klass):
1187
"""Return the current default format."""
1188
return klass._default_format
1189
1190
def get_format_string(self):
1191
"""Return the ASCII format string that identifies this format.
1192
1193
Note that in pre format ?? repositories the format string is
1194
not permitted nor written to disk.
1195
"""
1196
raise NotImplementedError(self.get_format_string)
1197
1198
def get_format_description(self):
1199
"""Return the short description for this format."""
1200
raise NotImplementedError(self.get_format_description)
1201
1202
def _get_revision_store(self, repo_transport, control_files):
1203
"""Return the revision store object for this a_bzrdir."""
1204
raise NotImplementedError(self._get_revision_store)
1205
1206
def _get_text_rev_store(self,
1207
transport,
1208
control_files,
1209
name,
1210
compressed=True,
1211
prefixed=False,
1212
serializer=None):
1213
"""Common logic for getting a revision store for a repository.
1214
1215
see self._get_revision_store for the subclass-overridable method to
1216
get the store for a repository.
1217
"""
1218
from bzrlib.store.revision.text import TextRevisionStore
1219
dir_mode = control_files._dir_mode
1220
file_mode = control_files._file_mode
1221
text_store =TextStore(transport.clone(name),
1222
prefixed=prefixed,
1223
compressed=compressed,
1224
dir_mode=dir_mode,
1225
file_mode=file_mode)
1226
_revision_store = TextRevisionStore(text_store, serializer)
1227
return _revision_store
1228
1229
def _get_versioned_file_store(self,
1230
name,
1231
transport,
1232
control_files,
1233
prefixed=True,
1234
versionedfile_class=weave.WeaveFile,
1235
versionedfile_kwargs={},
1236
escaped=False):
1237
weave_transport = control_files._transport.clone(name)
1238
dir_mode = control_files._dir_mode
1239
file_mode = control_files._file_mode
1240
return VersionedFileStore(weave_transport, prefixed=prefixed,
1241
dir_mode=dir_mode,
1242
file_mode=file_mode,
1243
versionedfile_class=versionedfile_class,
1244
versionedfile_kwargs=versionedfile_kwargs,
1245
escaped=escaped)
1246
1247
def initialize(self, a_bzrdir, shared=False):
1248
"""Initialize a repository of this format in a_bzrdir.
1249
1250
:param a_bzrdir: The bzrdir to put the new repository in it.
1251
:param shared: The repository should be initialized as a sharable one.
1252
1253
This may raise UninitializableFormat if shared repository are not
1254
compatible the a_bzrdir.
1255
"""
1256
1257
def is_supported(self):
1258
"""Is this format supported?
1259
1260
Supported formats must be initializable and openable.
1261
Unsupported formats may not support initialization or committing or
1262
some other features depending on the reason for not being supported.
1263
"""
1264
return True
1265
1266
def check_conversion_target(self, target_format):
1267
raise NotImplementedError(self.check_conversion_target)
1268
1269
def open(self, a_bzrdir, _found=False):
1270
"""Return an instance of this format for the bzrdir a_bzrdir.
1271
1272
_found is a private parameter, do not use it.
1273
"""
1274
raise NotImplementedError(self.open)
1275
1276
@classmethod
1277
def register_format(klass, format):
1278
klass._formats[format.get_format_string()] = format
1279
1280
@classmethod
1281
@deprecated_method(symbol_versioning.zero_fourteen)
1282
def set_default_format(klass, format):
1283
klass._set_default_format(format)
1284
1285
@classmethod
1286
def _set_default_format(klass, format):
1287
klass._default_format = format
1288
1289
@classmethod
1290
def unregister_format(klass, format):
1291
assert klass._formats[format.get_format_string()] is format
1292
del klass._formats[format.get_format_string()]
1293
1294
1295
class PreSplitOutRepositoryFormat(RepositoryFormat):
1296
"""Base class for the pre split out repository formats."""
1297
1298
rich_root_data = False
1299
1300
def initialize(self, a_bzrdir, shared=False, _internal=False):
1301
"""Create a weave repository.
1302
1303
TODO: when creating split out bzr branch formats, move this to a common
1304
base for Format5, Format6. or something like that.
1305
"""
1306
if shared:
1307
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1308
1309
if not _internal:
1310
# always initialized when the bzrdir is.
1311
return self.open(a_bzrdir, _found=True)
1312
1313
# Create an empty weave
1314
sio = StringIO()
1315
weavefile.write_weave_v5(weave.Weave(), sio)
1316
empty_weave = sio.getvalue()
1317
1318
mutter('creating repository in %s.', a_bzrdir.transport.base)
1319
dirs = ['revision-store', 'weaves']
1320
files = [('inventory.weave', StringIO(empty_weave)),
1321
]
1322
1323
# FIXME: RBC 20060125 don't peek under the covers
1324
# NB: no need to escape relative paths that are url safe.
1325
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1326
'branch-lock', lockable_files.TransportLock)
1327
control_files.create_lock()
1328
control_files.lock_write()
1329
control_files._transport.mkdir_multi(dirs,
1330
mode=control_files._dir_mode)
1331
try:
1332
for file, content in files:
1333
control_files.put(file, content)
1334
finally:
1335
control_files.unlock()
1336
return self.open(a_bzrdir, _found=True)
1337
1338
def _get_control_store(self, repo_transport, control_files):
1339
"""Return the control store for this repository."""
1340
return self._get_versioned_file_store('',
1341
repo_transport,
1342
control_files,
1343
prefixed=False)
1344
1345
def _get_text_store(self, transport, control_files):
1346
"""Get a store for file texts for this format."""
1347
raise NotImplementedError(self._get_text_store)
1348
1349
def open(self, a_bzrdir, _found=False):
1350
"""See RepositoryFormat.open()."""
1351
if not _found:
1352
# we are being called directly and must probe.
1353
raise NotImplementedError
1354
1355
repo_transport = a_bzrdir.get_repository_transport(None)
1356
control_files = a_bzrdir._control_files
1357
text_store = self._get_text_store(repo_transport, control_files)
1358
control_store = self._get_control_store(repo_transport, control_files)
1359
_revision_store = self._get_revision_store(repo_transport, control_files)
1360
return AllInOneRepository(_format=self,
1361
a_bzrdir=a_bzrdir,
1362
_revision_store=_revision_store,
1363
control_store=control_store,
1364
text_store=text_store)
1365
1366
def check_conversion_target(self, target_format):
1367
pass
1368
1369
1370
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1371
"""Bzr repository format 4.
1372
1373
This repository format has:
1374
- flat stores
1375
- TextStores for texts, inventories,revisions.
1376
1377
This format is deprecated: it indexes texts using a text id which is
1378
removed in format 5; initialization and write support for this format
1379
has been removed.
1380
"""
1381
1382
def __init__(self):
1383
super(RepositoryFormat4, self).__init__()
1384
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1385
1386
def get_format_description(self):
1387
"""See RepositoryFormat.get_format_description()."""
1388
return "Repository format 4"
1389
1390
def initialize(self, url, shared=False, _internal=False):
1391
"""Format 4 branches cannot be created."""
1392
raise errors.UninitializableFormat(self)
1393
1394
def is_supported(self):
1395
"""Format 4 is not supported.
1396
1397
It is not supported because the model changed from 4 to 5 and the
1398
conversion logic is expensive - so doing it on the fly was not
1399
feasible.
1400
"""
1401
return False
1402
1403
def _get_control_store(self, repo_transport, control_files):
1404
"""Format 4 repositories have no formal control store at this point.
1405
1406
This will cause any control-file-needing apis to fail - this is desired.
1407
"""
1408
return None
1409
1410
def _get_revision_store(self, repo_transport, control_files):
1411
"""See RepositoryFormat._get_revision_store()."""
1412
from bzrlib.xml4 import serializer_v4
1413
return self._get_text_rev_store(repo_transport,
1414
control_files,
1415
'revision-store',
1416
serializer=serializer_v4)
1417
1418
def _get_text_store(self, transport, control_files):
1419
"""See RepositoryFormat._get_text_store()."""
1420
1421
1422
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1423
"""Bzr control format 5.
1424
1425
This repository format has:
1426
- weaves for file texts and inventory
1427
- flat stores
1428
- TextStores for revisions and signatures.
1429
"""
1430
1431
def __init__(self):
1432
super(RepositoryFormat5, self).__init__()
1433
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1434
1435
def get_format_description(self):
1436
"""See RepositoryFormat.get_format_description()."""
1437
return "Weave repository format 5"
1438
1439
def _get_revision_store(self, repo_transport, control_files):
1440
"""See RepositoryFormat._get_revision_store()."""
1441
"""Return the revision store object for this a_bzrdir."""
1442
return self._get_text_rev_store(repo_transport,
1443
control_files,
1444
'revision-store',
1445
compressed=False)
1446
1447
def _get_text_store(self, transport, control_files):
1448
"""See RepositoryFormat._get_text_store()."""
1449
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1450
1451
1452
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1453
"""Bzr control format 6.
1454
1455
This repository format has:
1456
- weaves for file texts and inventory
1457
- hash subdirectory based stores.
1458
- TextStores for revisions and signatures.
1459
"""
1460
1461
def __init__(self):
1462
super(RepositoryFormat6, self).__init__()
1463
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1464
1465
def get_format_description(self):
1466
"""See RepositoryFormat.get_format_description()."""
1467
return "Weave repository format 6"
1468
1469
def _get_revision_store(self, repo_transport, control_files):
1470
"""See RepositoryFormat._get_revision_store()."""
1471
return self._get_text_rev_store(repo_transport,
1472
control_files,
1473
'revision-store',
1474
compressed=False,
1475
prefixed=True)
1476
1477
def _get_text_store(self, transport, control_files):
1478
"""See RepositoryFormat._get_text_store()."""
1479
return self._get_versioned_file_store('weaves', transport, control_files)
1480
1481
1482
class MetaDirRepositoryFormat(RepositoryFormat):
1483
"""Common base class for the new repositories using the metadir layout."""
1484
1485
rich_root_data = False
1486
1487
def __init__(self):
1488
super(MetaDirRepositoryFormat, self).__init__()
1489
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1490
1491
def _create_control_files(self, a_bzrdir):
1492
"""Create the required files and the initial control_files object."""
1493
# FIXME: RBC 20060125 don't peek under the covers
1494
# NB: no need to escape relative paths that are url safe.
1495
repository_transport = a_bzrdir.get_repository_transport(self)
1496
control_files = lockable_files.LockableFiles(repository_transport,
1497
'lock', lockdir.LockDir)
1498
control_files.create_lock()
1499
return control_files
1500
1501
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1502
"""Upload the initial blank content."""
1503
control_files = self._create_control_files(a_bzrdir)
1504
control_files.lock_write()
1505
try:
1506
control_files._transport.mkdir_multi(dirs,
1507
mode=control_files._dir_mode)
1508
for file, content in files:
1509
control_files.put(file, content)
1510
for file, content in utf8_files:
1511
control_files.put_utf8(file, content)
1512
if shared == True:
1513
control_files.put_utf8('shared-storage', '')
1514
finally:
1515
control_files.unlock()
1516
1517
1518
class RepositoryFormat7(MetaDirRepositoryFormat):
1519
"""Bzr repository 7.
1520
1521
This repository format has:
1522
- weaves for file texts and inventory
1523
- hash subdirectory based stores.
1524
- TextStores for revisions and signatures.
1525
- a format marker of its own
1526
- an optional 'shared-storage' flag
1527
- an optional 'no-working-trees' flag
1528
"""
1529
1530
def _get_control_store(self, repo_transport, control_files):
1531
"""Return the control store for this repository."""
1532
return self._get_versioned_file_store('',
1533
repo_transport,
1534
control_files,
1535
prefixed=False)
1536
1537
def get_format_string(self):
1538
"""See RepositoryFormat.get_format_string()."""
1539
return "Bazaar-NG Repository format 7"
1540
1541
def get_format_description(self):
1542
"""See RepositoryFormat.get_format_description()."""
1543
return "Weave repository format 7"
1544
1545
def check_conversion_target(self, target_format):
1546
pass
1547
1548
def _get_revision_store(self, repo_transport, control_files):
1549
"""See RepositoryFormat._get_revision_store()."""
1550
return self._get_text_rev_store(repo_transport,
1551
control_files,
1552
'revision-store',
1553
compressed=False,
1554
prefixed=True,
1555
)
1556
1557
def _get_text_store(self, transport, control_files):
1558
"""See RepositoryFormat._get_text_store()."""
1559
return self._get_versioned_file_store('weaves',
1560
transport,
1561
control_files)
1562
1563
def initialize(self, a_bzrdir, shared=False):
1564
"""Create a weave repository.
1565
1566
:param shared: If true the repository will be initialized as a shared
1567
repository.
1568
"""
1569
# Create an empty weave
1570
sio = StringIO()
1571
weavefile.write_weave_v5(weave.Weave(), sio)
1572
empty_weave = sio.getvalue()
1573
1574
mutter('creating repository in %s.', a_bzrdir.transport.base)
1575
dirs = ['revision-store', 'weaves']
1576
files = [('inventory.weave', StringIO(empty_weave)),
1577
]
1578
utf8_files = [('format', self.get_format_string())]
1579
1580
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1581
return self.open(a_bzrdir=a_bzrdir, _found=True)
1582
1583
def open(self, a_bzrdir, _found=False, _override_transport=None):
1584
"""See RepositoryFormat.open().
1585
1586
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1587
repository at a slightly different url
1588
than normal. I.e. during 'upgrade'.
1589
"""
1590
if not _found:
1591
format = RepositoryFormat.find_format(a_bzrdir)
1592
assert format.__class__ == self.__class__
1593
if _override_transport is not None:
1594
repo_transport = _override_transport
1595
else:
1596
repo_transport = a_bzrdir.get_repository_transport(None)
1597
control_files = lockable_files.LockableFiles(repo_transport,
1598
'lock', lockdir.LockDir)
1599
text_store = self._get_text_store(repo_transport, control_files)
1600
control_store = self._get_control_store(repo_transport, control_files)
1601
_revision_store = self._get_revision_store(repo_transport, control_files)
1602
return WeaveMetaDirRepository(_format=self,
1603
a_bzrdir=a_bzrdir,
1604
control_files=control_files,
1605
_revision_store=_revision_store,
1606
control_store=control_store,
1607
text_store=text_store)
1608
1609
1610
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1611
"""Bzr repository knit format (generalized).
1612
1613
This repository format has:
1614
- knits for file texts and inventory
1615
- hash subdirectory based stores.
1616
- knits for revisions and signatures
1617
- TextStores for revisions and signatures.
1618
- a format marker of its own
1619
- an optional 'shared-storage' flag
1620
- an optional 'no-working-trees' flag
1621
- a LockDir lock
1622
"""
1623
1624
def _get_control_store(self, repo_transport, control_files):
1625
"""Return the control store for this repository."""
1626
return VersionedFileStore(
1627
repo_transport,
1628
prefixed=False,
1629
file_mode=control_files._file_mode,
1630
versionedfile_class=knit.KnitVersionedFile,
1631
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1632
)
1633
1634
def _get_revision_store(self, repo_transport, control_files):
1635
"""See RepositoryFormat._get_revision_store()."""
1636
from bzrlib.store.revision.knit import KnitRevisionStore
1637
versioned_file_store = VersionedFileStore(
1638
repo_transport,
1639
file_mode=control_files._file_mode,
1640
prefixed=False,
1641
precious=True,
1642
versionedfile_class=knit.KnitVersionedFile,
1643
versionedfile_kwargs={'delta':False,
1644
'factory':knit.KnitPlainFactory(),
1645
},
1646
escaped=True,
1647
)
1648
return KnitRevisionStore(versioned_file_store)
1649
1650
def _get_text_store(self, transport, control_files):
1651
"""See RepositoryFormat._get_text_store()."""
1652
return self._get_versioned_file_store('knits',
1653
transport,
1654
control_files,
1655
versionedfile_class=knit.KnitVersionedFile,
1656
versionedfile_kwargs={
1657
'create_parent_dir':True,
1658
'delay_create':True,
1659
'dir_mode':control_files._dir_mode,
1660
},
1661
escaped=True)
1662
1663
def initialize(self, a_bzrdir, shared=False):
1664
"""Create a knit format 1 repository.
1665
1666
:param a_bzrdir: bzrdir to contain the new repository; must already
1667
be initialized.
1668
:param shared: If true the repository will be initialized as a shared
1669
repository.
1670
"""
1671
mutter('creating repository in %s.', a_bzrdir.transport.base)
1672
dirs = ['revision-store', 'knits']
1673
files = []
1674
utf8_files = [('format', self.get_format_string())]
1675
1676
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1677
repo_transport = a_bzrdir.get_repository_transport(None)
1678
control_files = lockable_files.LockableFiles(repo_transport,
1679
'lock', lockdir.LockDir)
1680
control_store = self._get_control_store(repo_transport, control_files)
1681
transaction = transactions.WriteTransaction()
1682
# trigger a write of the inventory store.
1683
control_store.get_weave_or_empty('inventory', transaction)
1684
_revision_store = self._get_revision_store(repo_transport, control_files)
1685
# the revision id here is irrelevant: it will not be stored, and cannot
1686
# already exist.
1687
_revision_store.has_revision_id('A', transaction)
1688
_revision_store.get_signature_file(transaction)
1689
return self.open(a_bzrdir=a_bzrdir, _found=True)
1690
1691
def open(self, a_bzrdir, _found=False, _override_transport=None):
1692
"""See RepositoryFormat.open().
1693
1694
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1695
repository at a slightly different url
1696
than normal. I.e. during 'upgrade'.
1697
"""
1698
if not _found:
1699
format = RepositoryFormat.find_format(a_bzrdir)
1700
assert format.__class__ == self.__class__
1701
if _override_transport is not None:
1702
repo_transport = _override_transport
1703
else:
1704
repo_transport = a_bzrdir.get_repository_transport(None)
1705
control_files = lockable_files.LockableFiles(repo_transport,
1706
'lock', lockdir.LockDir)
1707
text_store = self._get_text_store(repo_transport, control_files)
1708
control_store = self._get_control_store(repo_transport, control_files)
1709
_revision_store = self._get_revision_store(repo_transport, control_files)
1710
return KnitRepository(_format=self,
1711
a_bzrdir=a_bzrdir,
1712
control_files=control_files,
1713
_revision_store=_revision_store,
1714
control_store=control_store,
1715
text_store=text_store)
1716
1717
1718
class RepositoryFormatKnit1(RepositoryFormatKnit):
1719
"""Bzr repository knit format 1.
1720
1721
This repository format has:
1722
- knits for file texts and inventory
1723
- hash subdirectory based stores.
1724
- knits for revisions and signatures
1725
- TextStores for revisions and signatures.
1726
- a format marker of its own
1727
- an optional 'shared-storage' flag
1728
- an optional 'no-working-trees' flag
1729
- a LockDir lock
1730
1731
This format was introduced in bzr 0.8.
1732
"""
1733
def get_format_string(self):
1734
"""See RepositoryFormat.get_format_string()."""
1735
return "Bazaar-NG Knit Repository Format 1"
1736
1737
def get_format_description(self):
1738
"""See RepositoryFormat.get_format_description()."""
1739
return "Knit repository format 1"
1740
1741
def check_conversion_target(self, target_format):
1742
pass
1743
1744
1745
class RepositoryFormatKnit2(RepositoryFormatKnit):
1746
"""Bzr repository knit format 2.
1747
1748
THIS FORMAT IS EXPERIMENTAL
1749
This repository format has:
1750
- knits for file texts and inventory
1751
- hash subdirectory based stores.
1752
- knits for revisions and signatures
1753
- TextStores for revisions and signatures.
1754
- a format marker of its own
1755
- an optional 'shared-storage' flag
1756
- an optional 'no-working-trees' flag
1757
- a LockDir lock
1758
- Support for recording full info about the tree root
1759
1760
"""
1761
1762
rich_root_data = True
1763
1764
def get_format_string(self):
1765
"""See RepositoryFormat.get_format_string()."""
1766
return "Bazaar Knit Repository Format 2\n"
1767
1768
def get_format_description(self):
1769
"""See RepositoryFormat.get_format_description()."""
1770
return "Knit repository format 2"
1771
1772
def check_conversion_target(self, target_format):
1773
if not target_format.rich_root_data:
1774
raise errors.BadConversionTarget(
1775
'Does not support rich root data.', target_format)
1776
1777
def open(self, a_bzrdir, _found=False, _override_transport=None):
1778
"""See RepositoryFormat.open().
1779
1780
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1781
repository at a slightly different url
1782
than normal. I.e. during 'upgrade'.
1783
"""
1784
if not _found:
1785
format = RepositoryFormat.find_format(a_bzrdir)
1786
assert format.__class__ == self.__class__
1787
if _override_transport is not None:
1788
repo_transport = _override_transport
1789
else:
1790
repo_transport = a_bzrdir.get_repository_transport(None)
1791
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1792
lockdir.LockDir)
1793
text_store = self._get_text_store(repo_transport, control_files)
1794
control_store = self._get_control_store(repo_transport, control_files)
1795
_revision_store = self._get_revision_store(repo_transport, control_files)
1796
return KnitRepository2(_format=self,
1797
a_bzrdir=a_bzrdir,
1798
control_files=control_files,
1799
_revision_store=_revision_store,
1800
control_store=control_store,
1801
text_store=text_store)
1802
1803
1804
1805
# formats which have no format string are not discoverable
1806
# and not independently creatable, so are not registered.
1807
RepositoryFormat.register_format(RepositoryFormat7())
1808
# KEEP in sync with bzrdir.format_registry default
1809
_default_format = RepositoryFormatKnit1()
1810
RepositoryFormat.register_format(_default_format)
1811
RepositoryFormat.register_format(RepositoryFormatKnit2())
1812
RepositoryFormat._set_default_format(_default_format)
1813
_legacy_formats = [RepositoryFormat4(),
1814
RepositoryFormat5(),
1815
RepositoryFormat6()]
1816
1817
1818
class InterRepository(InterObject):
1819
"""This class represents operations taking place between two repositories.
1820
1821
Its instances have methods like copy_content and fetch, and contain
1822
references to the source and target repositories these operations can be
1823
carried out on.
1824
1825
Often we will provide convenience methods on 'repository' which carry out
1826
operations with another repository - they will always forward to
1827
InterRepository.get(other).method_name(parameters).
1828
"""
1829
1830
_optimisers = []
1831
"""The available optimised InterRepository types."""
1832
1833
def copy_content(self, revision_id=None, basis=None):
1834
raise NotImplementedError(self.copy_content)
1835
1836
def fetch(self, revision_id=None, pb=None):
1837
"""Fetch the content required to construct revision_id.
1838
1839
The content is copied from self.source to self.target.
1840
1841
:param revision_id: if None all content is copied, if NULL_REVISION no
1842
content is copied.
1843
:param pb: optional progress bar to use for progress reports. If not
1844
provided a default one will be created.
1845
1846
Returns the copied revision count and the failed revisions in a tuple:
1847
(copied, failures).
1848
"""
1849
raise NotImplementedError(self.fetch)
1850
1851
@needs_read_lock
1852
def missing_revision_ids(self, revision_id=None):
1853
"""Return the revision ids that source has that target does not.
1854
1855
These are returned in topological order.
1856
1857
:param revision_id: only return revision ids included by this
1858
revision_id.
1859
"""
1860
# generic, possibly worst case, slow code path.
1861
target_ids = set(self.target.all_revision_ids())
1862
if revision_id is not None:
1863
source_ids = self.source.get_ancestry(revision_id)
1864
assert source_ids[0] is None
1865
source_ids.pop(0)
1866
else:
1867
source_ids = self.source.all_revision_ids()
1868
result_set = set(source_ids).difference(target_ids)
1869
# this may look like a no-op: its not. It preserves the ordering
1870
# other_ids had while only returning the members from other_ids
1871
# that we've decided we need.
1872
return [rev_id for rev_id in source_ids if rev_id in result_set]
1873
1874
1875
class InterSameDataRepository(InterRepository):
1876
"""Code for converting between repositories that represent the same data.
1877
1878
Data format and model must match for this to work.
1879
"""
1880
1881
_matching_repo_format = RepositoryFormat4()
1882
"""Repository format for testing with."""
1883
1884
@staticmethod
1885
def is_compatible(source, target):
1886
if not isinstance(source, Repository):
1887
return False
1888
if not isinstance(target, Repository):
1889
return False
1890
if source._format.rich_root_data == target._format.rich_root_data:
1891
return True
1892
else:
1893
return False
1894
1895
@needs_write_lock
1896
def copy_content(self, revision_id=None, basis=None):
1897
"""Make a complete copy of the content in self into destination.
1898
1899
This is a destructive operation! Do not use it on existing
1900
repositories.
1901
1902
:param revision_id: Only copy the content needed to construct
1903
revision_id and its parents.
1904
:param basis: Copy the needed data preferentially from basis.
1905
"""
1906
try:
1907
self.target.set_make_working_trees(self.source.make_working_trees())
1908
except NotImplementedError:
1909
pass
1910
# grab the basis available data
1911
if basis is not None:
1912
self.target.fetch(basis, revision_id=revision_id)
1913
# but don't bother fetching if we have the needed data now.
1914
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1915
self.target.has_revision(revision_id)):
1916
return
1917
self.target.fetch(self.source, revision_id=revision_id)
1918
1919
@needs_write_lock
1920
def fetch(self, revision_id=None, pb=None):
1921
"""See InterRepository.fetch()."""
1922
from bzrlib.fetch import GenericRepoFetcher
1923
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1924
self.source, self.source._format, self.target,
1925
self.target._format)
1926
f = GenericRepoFetcher(to_repository=self.target,
1927
from_repository=self.source,
1928
last_revision=revision_id,
1929
pb=pb)
1930
return f.count_copied, f.failed_revisions
1931
1932
1933
class InterWeaveRepo(InterSameDataRepository):
1934
"""Optimised code paths between Weave based repositories."""
1935
1936
_matching_repo_format = RepositoryFormat7()
1937
"""Repository format for testing with."""
1938
1939
@staticmethod
1940
def is_compatible(source, target):
1941
"""Be compatible with known Weave formats.
1942
1943
We don't test for the stores being of specific types because that
1944
could lead to confusing results, and there is no need to be
1945
overly general.
1946
"""
1947
try:
1948
return (isinstance(source._format, (RepositoryFormat5,
1949
RepositoryFormat6,
1950
RepositoryFormat7)) and
1951
isinstance(target._format, (RepositoryFormat5,
1952
RepositoryFormat6,
1953
RepositoryFormat7)))
1954
except AttributeError:
1955
return False
1956
1957
@needs_write_lock
1958
def copy_content(self, revision_id=None, basis=None):
1959
"""See InterRepository.copy_content()."""
1960
# weave specific optimised path:
1961
if basis is not None:
1962
# copy the basis in, then fetch remaining data.
1963
basis.copy_content_into(self.target, revision_id)
1964
# the basis copy_content_into could miss-set this.
1965
try:
1966
self.target.set_make_working_trees(self.source.make_working_trees())
1967
except NotImplementedError:
1968
pass
1969
self.target.fetch(self.source, revision_id=revision_id)
1970
else:
1971
try:
1972
self.target.set_make_working_trees(self.source.make_working_trees())
1973
except NotImplementedError:
1974
pass
1975
# FIXME do not peek!
1976
if self.source.control_files._transport.listable():
1977
pb = ui.ui_factory.nested_progress_bar()
1978
try:
1979
self.target.weave_store.copy_all_ids(
1980
self.source.weave_store,
1981
pb=pb,
1982
from_transaction=self.source.get_transaction(),
1983
to_transaction=self.target.get_transaction())
1984
pb.update('copying inventory', 0, 1)
1985
self.target.control_weaves.copy_multi(
1986
self.source.control_weaves, ['inventory'],
1987
from_transaction=self.source.get_transaction(),
1988
to_transaction=self.target.get_transaction())
1989
self.target._revision_store.text_store.copy_all_ids(
1990
self.source._revision_store.text_store,
1991
pb=pb)
1992
finally:
1993
pb.finished()
1994
else:
1995
self.target.fetch(self.source, revision_id=revision_id)
1996
1997
@needs_write_lock
1998
def fetch(self, revision_id=None, pb=None):
1999
"""See InterRepository.fetch()."""
2000
from bzrlib.fetch import GenericRepoFetcher
2001
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2002
self.source, self.source._format, self.target, self.target._format)
2003
f = GenericRepoFetcher(to_repository=self.target,
2004
from_repository=self.source,
2005
last_revision=revision_id,
2006
pb=pb)
2007
return f.count_copied, f.failed_revisions
2008
2009
@needs_read_lock
2010
def missing_revision_ids(self, revision_id=None):
2011
"""See InterRepository.missing_revision_ids()."""
2012
# we want all revisions to satisfy revision_id in source.
2013
# but we don't want to stat every file here and there.
2014
# we want then, all revisions other needs to satisfy revision_id
2015
# checked, but not those that we have locally.
2016
# so the first thing is to get a subset of the revisions to
2017
# satisfy revision_id in source, and then eliminate those that
2018
# we do already have.
2019
# this is slow on high latency connection to self, but as as this
2020
# disk format scales terribly for push anyway due to rewriting
2021
# inventory.weave, this is considered acceptable.
2022
# - RBC 20060209
2023
if revision_id is not None:
2024
source_ids = self.source.get_ancestry(revision_id)
2025
assert source_ids[0] is None
2026
source_ids.pop(0)
2027
else:
2028
source_ids = self.source._all_possible_ids()
2029
source_ids_set = set(source_ids)
2030
# source_ids is the worst possible case we may need to pull.
2031
# now we want to filter source_ids against what we actually
2032
# have in target, but don't try to check for existence where we know
2033
# we do not have a revision as that would be pointless.
2034
target_ids = set(self.target._all_possible_ids())
2035
possibly_present_revisions = target_ids.intersection(source_ids_set)
2036
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2037
required_revisions = source_ids_set.difference(actually_present_revisions)
2038
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2039
if revision_id is not None:
2040
# we used get_ancestry to determine source_ids then we are assured all
2041
# revisions referenced are present as they are installed in topological order.
2042
# and the tip revision was validated by get_ancestry.
2043
return required_topo_revisions
2044
else:
2045
# if we just grabbed the possibly available ids, then
2046
# we only have an estimate of whats available and need to validate
2047
# that against the revision records.
2048
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2049
2050
2051
class InterKnitRepo(InterSameDataRepository):
2052
"""Optimised code paths between Knit based repositories."""
2053
2054
_matching_repo_format = RepositoryFormatKnit1()
2055
"""Repository format for testing with."""
2056
2057
@staticmethod
2058
def is_compatible(source, target):
2059
"""Be compatible with known Knit formats.
2060
2061
We don't test for the stores being of specific types because that
2062
could lead to confusing results, and there is no need to be
2063
overly general.
2064
"""
2065
try:
2066
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2067
isinstance(target._format, (RepositoryFormatKnit1)))
2068
except AttributeError:
2069
return False
2070
2071
@needs_write_lock
2072
def fetch(self, revision_id=None, pb=None):
2073
"""See InterRepository.fetch()."""
2074
from bzrlib.fetch import KnitRepoFetcher
2075
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2076
self.source, self.source._format, self.target, self.target._format)
2077
f = KnitRepoFetcher(to_repository=self.target,
2078
from_repository=self.source,
2079
last_revision=revision_id,
2080
pb=pb)
2081
return f.count_copied, f.failed_revisions
2082
2083
@needs_read_lock
2084
def missing_revision_ids(self, revision_id=None):
2085
"""See InterRepository.missing_revision_ids()."""
2086
if revision_id is not None:
2087
source_ids = self.source.get_ancestry(revision_id)
2088
assert source_ids[0] is None
2089
source_ids.pop(0)
2090
else:
2091
source_ids = self.source._all_possible_ids()
2092
source_ids_set = set(source_ids)
2093
# source_ids is the worst possible case we may need to pull.
2094
# now we want to filter source_ids against what we actually
2095
# have in target, but don't try to check for existence where we know
2096
# we do not have a revision as that would be pointless.
2097
target_ids = set(self.target._all_possible_ids())
2098
possibly_present_revisions = target_ids.intersection(source_ids_set)
2099
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2100
required_revisions = source_ids_set.difference(actually_present_revisions)
2101
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2102
if revision_id is not None:
2103
# we used get_ancestry to determine source_ids then we are assured all
2104
# revisions referenced are present as they are installed in topological order.
2105
# and the tip revision was validated by get_ancestry.
2106
return required_topo_revisions
2107
else:
2108
# if we just grabbed the possibly available ids, then
2109
# we only have an estimate of whats available and need to validate
2110
# that against the revision records.
2111
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2112
2113
2114
class InterModel1and2(InterRepository):
2115
2116
_matching_repo_format = None
2117
2118
@staticmethod
2119
def is_compatible(source, target):
2120
if not isinstance(source, Repository):
2121
return False
2122
if not isinstance(target, Repository):
2123
return False
2124
if not source._format.rich_root_data and target._format.rich_root_data:
2125
return True
2126
else:
2127
return False
2128
2129
@needs_write_lock
2130
def fetch(self, revision_id=None, pb=None):
2131
"""See InterRepository.fetch()."""
2132
from bzrlib.fetch import Model1toKnit2Fetcher
2133
f = Model1toKnit2Fetcher(to_repository=self.target,
2134
from_repository=self.source,
2135
last_revision=revision_id,
2136
pb=pb)
2137
return f.count_copied, f.failed_revisions
2138
2139
@needs_write_lock
2140
def copy_content(self, revision_id=None, basis=None):
2141
"""Make a complete copy of the content in self into destination.
2142
2143
This is a destructive operation! Do not use it on existing
2144
repositories.
2145
2146
:param revision_id: Only copy the content needed to construct
2147
revision_id and its parents.
2148
:param basis: Copy the needed data preferentially from basis.
2149
"""
2150
try:
2151
self.target.set_make_working_trees(self.source.make_working_trees())
2152
except NotImplementedError:
2153
pass
2154
# grab the basis available data
2155
if basis is not None:
2156
self.target.fetch(basis, revision_id=revision_id)
2157
# but don't bother fetching if we have the needed data now.
2158
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2159
self.target.has_revision(revision_id)):
2160
return
2161
self.target.fetch(self.source, revision_id=revision_id)
2162
2163
2164
class InterKnit1and2(InterKnitRepo):
2165
2166
_matching_repo_format = None
2167
2168
@staticmethod
2169
def is_compatible(source, target):
2170
"""Be compatible with Knit1 source and Knit2 target"""
2171
try:
2172
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2173
isinstance(target._format, (RepositoryFormatKnit2)))
2174
except AttributeError:
2175
return False
2176
2177
@needs_write_lock
2178
def fetch(self, revision_id=None, pb=None):
2179
"""See InterRepository.fetch()."""
2180
from bzrlib.fetch import Knit1to2Fetcher
2181
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2182
self.source, self.source._format, self.target,
2183
self.target._format)
2184
f = Knit1to2Fetcher(to_repository=self.target,
2185
from_repository=self.source,
2186
last_revision=revision_id,
2187
pb=pb)
2188
return f.count_copied, f.failed_revisions
2189
2190
2191
InterRepository.register_optimiser(InterSameDataRepository)
2192
InterRepository.register_optimiser(InterWeaveRepo)
2193
InterRepository.register_optimiser(InterKnitRepo)
2194
InterRepository.register_optimiser(InterModel1and2)
2195
InterRepository.register_optimiser(InterKnit1and2)
2196
2197
2198
class RepositoryTestProviderAdapter(object):
2199
"""A tool to generate a suite testing multiple repository formats at once.
2200
2201
This is done by copying the test once for each transport and injecting
2202
the transport_server, transport_readonly_server, and bzrdir_format and
2203
repository_format classes into each copy. Each copy is also given a new id()
2204
to make it easy to identify.
2205
"""
2206
2207
def __init__(self, transport_server, transport_readonly_server, formats):
2208
self._transport_server = transport_server
2209
self._transport_readonly_server = transport_readonly_server
2210
self._formats = formats
2211
2212
def adapt(self, test):
2213
result = unittest.TestSuite()
2214
for repository_format, bzrdir_format in self._formats:
2215
new_test = deepcopy(test)
2216
new_test.transport_server = self._transport_server
2217
new_test.transport_readonly_server = self._transport_readonly_server
2218
new_test.bzrdir_format = bzrdir_format
2219
new_test.repository_format = repository_format
2220
def make_new_test_id():
2221
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2222
return lambda: new_id
2223
new_test.id = make_new_test_id()
2224
result.addTest(new_test)
2225
return result
2226
2227
2228
class InterRepositoryTestProviderAdapter(object):
2229
"""A tool to generate a suite testing multiple inter repository formats.
2230
2231
This is done by copying the test once for each interrepo provider and injecting
2232
the transport_server, transport_readonly_server, repository_format and
2233
repository_to_format classes into each copy.
2234
Each copy is also given a new id() to make it easy to identify.
2235
"""
2236
2237
def __init__(self, transport_server, transport_readonly_server, formats):
2238
self._transport_server = transport_server
2239
self._transport_readonly_server = transport_readonly_server
2240
self._formats = formats
2241
2242
def adapt(self, test):
2243
result = unittest.TestSuite()
2244
for interrepo_class, repository_format, repository_format_to in self._formats:
2245
new_test = deepcopy(test)
2246
new_test.transport_server = self._transport_server
2247
new_test.transport_readonly_server = self._transport_readonly_server
2248
new_test.interrepo_class = interrepo_class
2249
new_test.repository_format = repository_format
2250
new_test.repository_format_to = repository_format_to
2251
def make_new_test_id():
2252
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2253
return lambda: new_id
2254
new_test.id = make_new_test_id()
2255
result.addTest(new_test)
2256
return result
2257
2258
@staticmethod
2259
def default_test_list():
2260
"""Generate the default list of interrepo permutations to test."""
2261
result = []
2262
# test the default InterRepository between format 6 and the current
2263
# default format.
2264
# XXX: robertc 20060220 reinstate this when there are two supported
2265
# formats which do not have an optimal code path between them.
2266
#result.append((InterRepository,
2267
# RepositoryFormat6(),
2268
# RepositoryFormatKnit1()))
2269
for optimiser in InterRepository._optimisers:
2270
if optimiser._matching_repo_format is not None:
2271
result.append((optimiser,
2272
optimiser._matching_repo_format,
2273
optimiser._matching_repo_format
2274
))
2275
# if there are specific combinations we want to use, we can add them
2276
# here.
2277
result.append((InterModel1and2, RepositoryFormat5(),
2278
RepositoryFormatKnit2()))
2279
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2280
RepositoryFormatKnit2()))
2281
return result
2282
2283
2284
class CopyConverter(object):
2285
"""A repository conversion tool which just performs a copy of the content.
2286
2287
This is slow but quite reliable.
2288
"""
2289
2290
def __init__(self, target_format):
2291
"""Create a CopyConverter.
2292
2293
:param target_format: The format the resulting repository should be.
2294
"""
2295
self.target_format = target_format
2296
2297
def convert(self, repo, pb):
2298
"""Perform the conversion of to_convert, giving feedback via pb.
2299
2300
:param to_convert: The disk object to convert.
2301
:param pb: a progress bar to use for progress information.
2302
"""
2303
self.pb = pb
2304
self.count = 0
2305
self.total = 4
2306
# this is only useful with metadir layouts - separated repo content.
2307
# trigger an assertion if not such
2308
repo._format.get_format_string()
2309
self.repo_dir = repo.bzrdir
2310
self.step('Moving repository to repository.backup')
2311
self.repo_dir.transport.move('repository', 'repository.backup')
2312
backup_transport = self.repo_dir.transport.clone('repository.backup')
2313
repo._format.check_conversion_target(self.target_format)
2314
self.source_repo = repo._format.open(self.repo_dir,
2315
_found=True,
2316
_override_transport=backup_transport)
2317
self.step('Creating new repository')
2318
converted = self.target_format.initialize(self.repo_dir,
2319
self.source_repo.is_shared())
2320
converted.lock_write()
2321
try:
2322
self.step('Copying content into repository.')
2323
self.source_repo.copy_content_into(converted)
2324
finally:
2325
converted.unlock()
2326
self.step('Deleting old repository content.')
2327
self.repo_dir.transport.delete_tree('repository.backup')
2328
self.pb.note('repository converted')
2329
2330
def step(self, message):
2331
"""Update the pb by a step."""
2332
self.count +=1
2333
self.pb.update(message, self.count, self.total)
2334
2335
2336
class CommitBuilder(object):
2337
"""Provides an interface to build up a commit.
2338
2339
This allows describing a tree to be committed without needing to
2340
know the internals of the format of the repository.
2341
"""
2342
2343
record_root_entry = False
2344
def __init__(self, repository, parents, config, timestamp=None,
2345
timezone=None, committer=None, revprops=None,
2346
revision_id=None):
2347
"""Initiate a CommitBuilder.
2348
2349
:param repository: Repository to commit to.
2350
:param parents: Revision ids of the parents of the new revision.
2351
:param config: Configuration to use.
2352
:param timestamp: Optional timestamp recorded for commit.
2353
:param timezone: Optional timezone for timestamp.
2354
:param committer: Optional committer to set for commit.
2355
:param revprops: Optional dictionary of revision properties.
2356
:param revision_id: Optional revision id.
2357
"""
2358
self._config = config
2359
2360
if committer is None:
2361
self._committer = self._config.username()
2362
else:
2363
assert isinstance(committer, basestring), type(committer)
2364
self._committer = committer
2365
2366
self.new_inventory = Inventory(None)
2367
self._new_revision_id = revision_id
2368
self.parents = parents
2369
self.repository = repository
2370
2371
self._revprops = {}
2372
if revprops is not None:
2373
self._revprops.update(revprops)
2374
2375
if timestamp is None:
2376
timestamp = time.time()
2377
# Restrict resolution to 1ms
2378
self._timestamp = round(timestamp, 3)
2379
2380
if timezone is None:
2381
self._timezone = local_time_offset()
2382
else:
2383
self._timezone = int(timezone)
2384
2385
self._generate_revision_if_needed()
2386
2387
def commit(self, message):
2388
"""Make the actual commit.
2389
2390
:return: The revision id of the recorded revision.
2391
"""
2392
rev = _mod_revision.Revision(
2393
timestamp=self._timestamp,
2394
timezone=self._timezone,
2395
committer=self._committer,
2396
message=message,
2397
inventory_sha1=self.inv_sha1,
2398
revision_id=self._new_revision_id,
2399
properties=self._revprops)
2400
rev.parent_ids = self.parents
2401
self.repository.add_revision(self._new_revision_id, rev,
2402
self.new_inventory, self._config)
2403
return self._new_revision_id
2404
2405
def revision_tree(self):
2406
"""Return the tree that was just committed.
2407
2408
After calling commit() this can be called to get a RevisionTree
2409
representing the newly committed tree. This is preferred to
2410
calling Repository.revision_tree() because that may require
2411
deserializing the inventory, while we already have a copy in
2412
memory.
2413
"""
2414
return RevisionTree(self.repository, self.new_inventory,
2415
self._new_revision_id)
2416
2417
def finish_inventory(self):
2418
"""Tell the builder that the inventory is finished."""
2419
if self.new_inventory.root is None:
2420
symbol_versioning.warn('Root entry should be supplied to'
2421
' record_entry_contents, as of bzr 0.10.',
2422
DeprecationWarning, stacklevel=2)
2423
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2424
self.new_inventory.revision_id = self._new_revision_id
2425
self.inv_sha1 = self.repository.add_inventory(
2426
self._new_revision_id,
2427
self.new_inventory,
2428
self.parents
2429
)
2430
2431
def _gen_revision_id(self):
2432
"""Return new revision-id."""
2433
return generate_ids.gen_revision_id(self._config.username(),
2434
self._timestamp)
2435
2436
def _generate_revision_if_needed(self):
2437
"""Create a revision id if None was supplied.
2438
2439
If the repository can not support user-specified revision ids
2440
they should override this function and raise CannotSetRevisionId
2441
if _new_revision_id is not None.
2442
2443
:raises: CannotSetRevisionId
2444
"""
2445
if self._new_revision_id is None:
2446
self._new_revision_id = self._gen_revision_id()
2447
2448
def record_entry_contents(self, ie, parent_invs, path, tree):
2449
"""Record the content of ie from tree into the commit if needed.
2450
2451
Side effect: sets ie.revision when unchanged
2452
2453
:param ie: An inventory entry present in the commit.
2454
:param parent_invs: The inventories of the parent revisions of the
2455
commit.
2456
:param path: The path the entry is at in the tree.
2457
:param tree: The tree which contains this entry and should be used to
2458
obtain content.
2459
"""
2460
if self.new_inventory.root is None and ie.parent_id is not None:
2461
symbol_versioning.warn('Root entry should be supplied to'
2462
' record_entry_contents, as of bzr 0.10.',
2463
DeprecationWarning, stacklevel=2)
2464
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2465
'', tree)
2466
self.new_inventory.add(ie)
2467
2468
# ie.revision is always None if the InventoryEntry is considered
2469
# for committing. ie.snapshot will record the correct revision
2470
# which may be the sole parent if it is untouched.
2471
if ie.revision is not None:
2472
return
2473
2474
# In this revision format, root entries have no knit or weave
2475
if ie is self.new_inventory.root:
2476
# When serializing out to disk and back in
2477
# root.revision is always _new_revision_id
2478
ie.revision = self._new_revision_id
2479
return
2480
previous_entries = ie.find_previous_heads(
2481
parent_invs,
2482
self.repository.weave_store,
2483
self.repository.get_transaction())
2484
# we are creating a new revision for ie in the history store
2485
# and inventory.
2486
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2487
2488
def modified_directory(self, file_id, file_parents):
2489
"""Record the presence of a symbolic link.
2490
2491
:param file_id: The file_id of the link to record.
2492
:param file_parents: The per-file parent revision ids.
2493
"""
2494
self._add_text_to_weave(file_id, [], file_parents.keys())
2495
2496
def modified_file_text(self, file_id, file_parents,
2497
get_content_byte_lines, text_sha1=None,
2498
text_size=None):
2499
"""Record the text of file file_id
2500
2501
:param file_id: The file_id of the file to record the text of.
2502
:param file_parents: The per-file parent revision ids.
2503
:param get_content_byte_lines: A callable which will return the byte
2504
lines for the file.
2505
:param text_sha1: Optional SHA1 of the file contents.
2506
:param text_size: Optional size of the file contents.
2507
"""
2508
# mutter('storing text of file {%s} in revision {%s} into %r',
2509
# file_id, self._new_revision_id, self.repository.weave_store)
2510
# special case to avoid diffing on renames or
2511
# reparenting
2512
if (len(file_parents) == 1
2513
and text_sha1 == file_parents.values()[0].text_sha1
2514
and text_size == file_parents.values()[0].text_size):
2515
previous_ie = file_parents.values()[0]
2516
versionedfile = self.repository.weave_store.get_weave(file_id,
2517
self.repository.get_transaction())
2518
versionedfile.clone_text(self._new_revision_id,
2519
previous_ie.revision, file_parents.keys())
2520
return text_sha1, text_size
2521
else:
2522
new_lines = get_content_byte_lines()
2523
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2524
# should return the SHA1 and size
2525
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2526
return osutils.sha_strings(new_lines), \
2527
sum(map(len, new_lines))
2528
2529
def modified_link(self, file_id, file_parents, link_target):
2530
"""Record the presence of a symbolic link.
2531
2532
:param file_id: The file_id of the link to record.
2533
:param file_parents: The per-file parent revision ids.
2534
:param link_target: Target location of this link.
2535
"""
2536
self._add_text_to_weave(file_id, [], file_parents.keys())
2537
2538
def _add_text_to_weave(self, file_id, new_lines, parents):
2539
versionedfile = self.repository.weave_store.get_weave_or_empty(
2540
file_id, self.repository.get_transaction())
2541
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2542
versionedfile.clear_cache()
2543
2544
2545
class _CommitBuilder(CommitBuilder):
2546
"""Temporary class so old CommitBuilders are detected properly
2547
2548
Note: CommitBuilder works whether or not root entry is recorded.
2549
"""
2550
2551
record_root_entry = True
2552
2553
2554
class RootCommitBuilder(CommitBuilder):
2555
"""This commitbuilder actually records the root id"""
2556
2557
record_root_entry = True
2558
2559
def record_entry_contents(self, ie, parent_invs, path, tree):
2560
"""Record the content of ie from tree into the commit if needed.
2561
2562
Side effect: sets ie.revision when unchanged
2563
2564
:param ie: An inventory entry present in the commit.
2565
:param parent_invs: The inventories of the parent revisions of the
2566
commit.
2567
:param path: The path the entry is at in the tree.
2568
:param tree: The tree which contains this entry and should be used to
2569
obtain content.
2570
"""
2571
assert self.new_inventory.root is not None or ie.parent_id is None
2572
self.new_inventory.add(ie)
2573
2574
# ie.revision is always None if the InventoryEntry is considered
2575
# for committing. ie.snapshot will record the correct revision
2576
# which may be the sole parent if it is untouched.
2577
if ie.revision is not None:
2578
return
2579
2580
previous_entries = ie.find_previous_heads(
2581
parent_invs,
2582
self.repository.weave_store,
2583
self.repository.get_transaction())
2584
# we are creating a new revision for ie in the history store
2585
# and inventory.
2586
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2587
2588
2589
_unescape_map = {
2590
'apos':"'",
2591
'quot':'"',
2592
'amp':'&',
2593
'lt':'<',
2594
'gt':'>'
2595
}
2596
2597
2598
def _unescaper(match, _map=_unescape_map):
2599
return _map[match.group(1)]
2600
2601
2602
_unescape_re = None
2603
2604
2605
def _unescape_xml(data):
2606
"""Unescape predefined XML entities in a string of data."""
2607
global _unescape_re
2608
if _unescape_re is None:
2609
_unescape_re = re.compile('\&([^;]*);')
2610
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1