« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 2067.3.4
- compare with another revision
- download diff
- download tarball
- view history from revision 2067.3.4
- Committer: Martin Pool
- Date: 2006-11-03 01:51:26 UTC
- mto: This revision was merged to the branch mainline in revision 2119.
- Revision ID: mbp@sourcefrog.net-20061103015126-427cb97a48404f42
Error deprecations will come in for 0.13
- 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/mdiff.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
gpg,
33
graph,
34
knit,
35
lockable_files,
36
lockdir,
37
osutils,
38
revision as _mod_revision,
39
symbol_versioning,
40
transactions,
41
ui,
42
weave,
43
weavefile,
44
xml5,
45
xml6,
46
)
47
from bzrlib.osutils import (
48
rand_bytes,
49
compact_date,
50
local_time_offset,
51
)
52
from bzrlib.revisiontree import RevisionTree
53
from bzrlib.store.versioned import VersionedFileStore
54
from bzrlib.store.text import TextStore
55
from bzrlib.testament import Testament
56
""")
57
58
from bzrlib.decorators import needs_read_lock, needs_write_lock
59
from bzrlib.inter import InterObject
60
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
61
from bzrlib.symbol_versioning import (
62
deprecated_method,
63
zero_nine,
64
)
65
from bzrlib.trace import mutter, note, warning
66
67
68
# Old formats display a warning, but only once
69
_deprecation_warning_done = False
70
71
72
class Repository(object):
73
"""Repository holding history for one or more branches.
74
75
The repository holds and retrieves historical information including
76
revisions and file history. It's normally accessed only by the Branch,
77
which views a particular line of development through that history.
78
79
The Repository builds on top of Stores and a Transport, which respectively
80
describe the disk data format and the way of accessing the (possibly
81
remote) disk.
82
"""
83
84
@needs_write_lock
85
def add_inventory(self, revid, inv, parents):
86
"""Add the inventory inv to the repository as revid.
87
88
:param parents: The revision ids of the parents that revid
89
is known to have and are in the repository already.
90
91
returns the sha1 of the serialized inventory.
92
"""
93
assert inv.revision_id is None or inv.revision_id == revid, \
94
"Mismatch between inventory revision" \
95
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
96
assert inv.root is not None
97
inv_text = self.serialise_inventory(inv)
98
inv_sha1 = osutils.sha_string(inv_text)
99
inv_vf = self.control_weaves.get_weave('inventory',
100
self.get_transaction())
101
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
102
return inv_sha1
103
104
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
105
final_parents = []
106
for parent in parents:
107
if parent in inv_vf:
108
final_parents.append(parent)
109
110
inv_vf.add_lines(revid, final_parents, lines)
111
112
@needs_write_lock
113
def add_revision(self, rev_id, rev, inv=None, config=None):
114
"""Add rev to the revision store as rev_id.
115
116
:param rev_id: the revision id to use.
117
:param rev: The revision object.
118
:param inv: The inventory for the revision. if None, it will be looked
119
up in the inventory storer
120
:param config: If None no digital signature will be created.
121
If supplied its signature_needed method will be used
122
to determine if a signature should be made.
123
"""
124
if config is not None and config.signature_needed():
125
if inv is None:
126
inv = self.get_inventory(rev_id)
127
plaintext = Testament(rev, inv).as_short_text()
128
self.store_revision_signature(
129
gpg.GPGStrategy(config), plaintext, rev_id)
130
if not rev_id in self.get_inventory_weave():
131
if inv is None:
132
raise errors.WeaveRevisionNotPresent(rev_id,
133
self.get_inventory_weave())
134
else:
135
# yes, this is not suitable for adding with ghosts.
136
self.add_inventory(rev_id, inv, rev.parent_ids)
137
self._revision_store.add_revision(rev, self.get_transaction())
138
139
@needs_read_lock
140
def _all_possible_ids(self):
141
"""Return all the possible revisions that we could find."""
142
return self.get_inventory_weave().versions()
143
144
def all_revision_ids(self):
145
"""Returns a list of all the revision ids in the repository.
146
147
This is deprecated because code should generally work on the graph
148
reachable from a particular revision, and ignore any other revisions
149
that might be present. There is no direct replacement method.
150
"""
151
return self._all_revision_ids()
152
153
@needs_read_lock
154
def _all_revision_ids(self):
155
"""Returns a list of all the revision ids in the repository.
156
157
These are in as much topological order as the underlying store can
158
present: for weaves ghosts may lead to a lack of correctness until
159
the reweave updates the parents list.
160
"""
161
if self._revision_store.text_store.listable():
162
return self._revision_store.all_revision_ids(self.get_transaction())
163
result = self._all_possible_ids()
164
return self._eliminate_revisions_not_present(result)
165
166
def break_lock(self):
167
"""Break a lock if one is present from another instance.
168
169
Uses the ui factory to ask for confirmation if the lock may be from
170
an active process.
171
"""
172
self.control_files.break_lock()
173
174
@needs_read_lock
175
def _eliminate_revisions_not_present(self, revision_ids):
176
"""Check every revision id in revision_ids to see if we have it.
177
178
Returns a set of the present revisions.
179
"""
180
result = []
181
for id in revision_ids:
182
if self.has_revision(id):
183
result.append(id)
184
return result
185
186
@staticmethod
187
def create(a_bzrdir):
188
"""Construct the current default format repository in a_bzrdir."""
189
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
190
191
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
192
"""instantiate a Repository.
193
194
:param _format: The format of the repository on disk.
195
:param a_bzrdir: The BzrDir of the repository.
196
197
In the future we will have a single api for all stores for
198
getting file texts, inventories and revisions, then
199
this construct will accept instances of those things.
200
"""
201
super(Repository, self).__init__()
202
self._format = _format
203
# the following are part of the public API for Repository:
204
self.bzrdir = a_bzrdir
205
self.control_files = control_files
206
self._revision_store = _revision_store
207
self.text_store = text_store
208
# backwards compatibility
209
self.weave_store = text_store
210
# not right yet - should be more semantically clear ?
211
#
212
self.control_store = control_store
213
self.control_weaves = control_store
214
# TODO: make sure to construct the right store classes, etc, depending
215
# on whether escaping is required.
216
self._warn_if_deprecated()
217
self._serializer = xml5.serializer_v5
218
219
def __repr__(self):
220
return '%s(%r)' % (self.__class__.__name__,
221
self.bzrdir.transport.base)
222
223
def is_locked(self):
224
return self.control_files.is_locked()
225
226
def lock_write(self):
227
self.control_files.lock_write()
228
229
def lock_read(self):
230
self.control_files.lock_read()
231
232
def get_physical_lock_status(self):
233
return self.control_files.get_physical_lock_status()
234
235
@needs_read_lock
236
def missing_revision_ids(self, other, revision_id=None):
237
"""Return the revision ids that other has that this does not.
238
239
These are returned in topological order.
240
241
revision_id: only return revision ids included by revision_id.
242
"""
243
return InterRepository.get(other, self).missing_revision_ids(revision_id)
244
245
@staticmethod
246
def open(base):
247
"""Open the repository rooted at base.
248
249
For instance, if the repository is at URL/.bzr/repository,
250
Repository.open(URL) -> a Repository instance.
251
"""
252
control = bzrdir.BzrDir.open(base)
253
return control.open_repository()
254
255
def copy_content_into(self, destination, revision_id=None, basis=None):
256
"""Make a complete copy of the content in self into destination.
257
258
This is a destructive operation! Do not use it on existing
259
repositories.
260
"""
261
return InterRepository.get(self, destination).copy_content(revision_id, basis)
262
263
def fetch(self, source, revision_id=None, pb=None):
264
"""Fetch the content required to construct revision_id from source.
265
266
If revision_id is None all content is copied.
267
"""
268
return InterRepository.get(source, self).fetch(revision_id=revision_id,
269
pb=pb)
270
271
def get_commit_builder(self, branch, parents, config, timestamp=None,
272
timezone=None, committer=None, revprops=None,
273
revision_id=None):
274
"""Obtain a CommitBuilder for this repository.
275
276
:param branch: Branch to commit to.
277
:param parents: Revision ids of the parents of the new revision.
278
:param config: Configuration to use.
279
:param timestamp: Optional timestamp recorded for commit.
280
:param timezone: Optional timezone for timestamp.
281
:param committer: Optional committer to set for commit.
282
:param revprops: Optional dictionary of revision properties.
283
:param revision_id: Optional revision id.
284
"""
285
return _CommitBuilder(self, parents, config, timestamp, timezone,
286
committer, revprops, revision_id)
287
288
def unlock(self):
289
self.control_files.unlock()
290
291
@needs_read_lock
292
def clone(self, a_bzrdir, revision_id=None, basis=None):
293
"""Clone this repository into a_bzrdir using the current format.
294
295
Currently no check is made that the format of this repository and
296
the bzrdir format are compatible. FIXME RBC 20060201.
297
"""
298
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
299
# use target default format.
300
result = a_bzrdir.create_repository()
301
# FIXME RBC 20060209 split out the repository type to avoid this check ?
302
elif isinstance(a_bzrdir._format,
303
(bzrdir.BzrDirFormat4,
304
bzrdir.BzrDirFormat5,
305
bzrdir.BzrDirFormat6)):
306
result = a_bzrdir.open_repository()
307
else:
308
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
309
self.copy_content_into(result, revision_id, basis)
310
return result
311
312
@needs_read_lock
313
def has_revision(self, revision_id):
314
"""True if this repository has a copy of the revision."""
315
return self._revision_store.has_revision_id(revision_id,
316
self.get_transaction())
317
318
@needs_read_lock
319
def get_revision_reconcile(self, revision_id):
320
"""'reconcile' helper routine that allows access to a revision always.
321
322
This variant of get_revision does not cross check the weave graph
323
against the revision one as get_revision does: but it should only
324
be used by reconcile, or reconcile-alike commands that are correcting
325
or testing the revision graph.
326
"""
327
if not revision_id or not isinstance(revision_id, basestring):
328
raise errors.InvalidRevisionId(revision_id=revision_id,
329
branch=self)
330
return self._revision_store.get_revisions([revision_id],
331
self.get_transaction())[0]
332
@needs_read_lock
333
def get_revisions(self, revision_ids):
334
return self._revision_store.get_revisions(revision_ids,
335
self.get_transaction())
336
337
@needs_read_lock
338
def get_revision_xml(self, revision_id):
339
rev = self.get_revision(revision_id)
340
rev_tmp = StringIO()
341
# the current serializer..
342
self._revision_store._serializer.write_revision(rev, rev_tmp)
343
rev_tmp.seek(0)
344
return rev_tmp.getvalue()
345
346
@needs_read_lock
347
def get_revision(self, revision_id):
348
"""Return the Revision object for a named revision"""
349
r = self.get_revision_reconcile(revision_id)
350
# weave corruption can lead to absent revision markers that should be
351
# present.
352
# the following test is reasonably cheap (it needs a single weave read)
353
# and the weave is cached in read transactions. In write transactions
354
# it is not cached but typically we only read a small number of
355
# revisions. For knits when they are introduced we will probably want
356
# to ensure that caching write transactions are in use.
357
inv = self.get_inventory_weave()
358
self._check_revision_parents(r, inv)
359
return r
360
361
@needs_read_lock
362
def get_deltas_for_revisions(self, revisions):
363
"""Produce a generator of revision deltas.
364
365
Note that the input is a sequence of REVISIONS, not revision_ids.
366
Trees will be held in memory until the generator exits.
367
Each delta is relative to the revision's lefthand predecessor.
368
"""
369
required_trees = set()
370
for revision in revisions:
371
required_trees.add(revision.revision_id)
372
required_trees.update(revision.parent_ids[:1])
373
trees = dict((t.get_revision_id(), t) for
374
t in self.revision_trees(required_trees))
375
for revision in revisions:
376
if not revision.parent_ids:
377
old_tree = self.revision_tree(None)
378
else:
379
old_tree = trees[revision.parent_ids[0]]
380
yield trees[revision.revision_id].changes_from(old_tree)
381
382
@needs_read_lock
383
def get_revision_delta(self, revision_id):
384
"""Return the delta for one revision.
385
386
The delta is relative to the left-hand predecessor of the
387
revision.
388
"""
389
r = self.get_revision(revision_id)
390
return list(self.get_deltas_for_revisions([r]))[0]
391
392
def _check_revision_parents(self, revision, inventory):
393
"""Private to Repository and Fetch.
394
395
This checks the parentage of revision in an inventory weave for
396
consistency and is only applicable to inventory-weave-for-ancestry
397
using repository formats & fetchers.
398
"""
399
weave_parents = inventory.get_parents(revision.revision_id)
400
weave_names = inventory.versions()
401
for parent_id in revision.parent_ids:
402
if parent_id in weave_names:
403
# this parent must not be a ghost.
404
if not parent_id in weave_parents:
405
# but it is a ghost
406
raise errors.CorruptRepository(self)
407
408
@needs_write_lock
409
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
410
signature = gpg_strategy.sign(plaintext)
411
self._revision_store.add_revision_signature_text(revision_id,
412
signature,
413
self.get_transaction())
414
415
def fileids_altered_by_revision_ids(self, revision_ids):
416
"""Find the file ids and versions affected by revisions.
417
418
:param revisions: an iterable containing revision ids.
419
:return: a dictionary mapping altered file-ids to an iterable of
420
revision_ids. Each altered file-ids has the exact revision_ids that
421
altered it listed explicitly.
422
"""
423
assert self._serializer.support_altered_by_hack, \
424
("fileids_altered_by_revision_ids only supported for branches "
425
"which store inventory as unnested xml, not on %r" % self)
426
selected_revision_ids = set(revision_ids)
427
w = self.get_inventory_weave()
428
result = {}
429
430
# this code needs to read every new line in every inventory for the
431
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
432
# not present in one of those inventories is unnecessary but not
433
# harmful because we are filtering by the revision id marker in the
434
# inventory lines : we only select file ids altered in one of those
435
# revisions. We don't need to see all lines in the inventory because
436
# only those added in an inventory in rev X can contain a revision=X
437
# line.
438
pb = ui.ui_factory.nested_progress_bar()
439
try:
440
for line in w.iter_lines_added_or_present_in_versions(
441
selected_revision_ids, pb=pb):
442
start = line.find('file_id="')+9
443
if start < 9: continue
444
end = line.find('"', start)
445
assert end>= 0
446
file_id = _unescape_xml(line[start:end])
447
448
start = line.find('revision="')+10
449
if start < 10: continue
450
end = line.find('"', start)
451
assert end>= 0
452
revision_id = _unescape_xml(line[start:end])
453
if revision_id in selected_revision_ids:
454
result.setdefault(file_id, set()).add(revision_id)
455
finally:
456
pb.finished()
457
return result
458
459
@needs_read_lock
460
def get_inventory_weave(self):
461
return self.control_weaves.get_weave('inventory',
462
self.get_transaction())
463
464
@needs_read_lock
465
def get_inventory(self, revision_id):
466
"""Get Inventory object by hash."""
467
return self.deserialise_inventory(
468
revision_id, self.get_inventory_xml(revision_id))
469
470
def deserialise_inventory(self, revision_id, xml):
471
"""Transform the xml into an inventory object.
472
473
:param revision_id: The expected revision id of the inventory.
474
:param xml: A serialised inventory.
475
"""
476
result = self._serializer.read_inventory_from_string(xml)
477
result.root.revision = revision_id
478
return result
479
480
def serialise_inventory(self, inv):
481
return self._serializer.write_inventory_to_string(inv)
482
483
@needs_read_lock
484
def get_inventory_xml(self, revision_id):
485
"""Get inventory XML as a file object."""
486
try:
487
assert isinstance(revision_id, basestring), type(revision_id)
488
iw = self.get_inventory_weave()
489
return iw.get_text(revision_id)
490
except IndexError:
491
raise errors.HistoryMissing(self, 'inventory', revision_id)
492
493
@needs_read_lock
494
def get_inventory_sha1(self, revision_id):
495
"""Return the sha1 hash of the inventory entry
496
"""
497
return self.get_revision(revision_id).inventory_sha1
498
499
@needs_read_lock
500
def get_revision_graph(self, revision_id=None):
501
"""Return a dictionary containing the revision graph.
502
503
:param revision_id: The revision_id to get a graph from. If None, then
504
the entire revision graph is returned. This is a deprecated mode of
505
operation and will be removed in the future.
506
:return: a dictionary of revision_id->revision_parents_list.
507
"""
508
# special case NULL_REVISION
509
if revision_id == _mod_revision.NULL_REVISION:
510
return {}
511
a_weave = self.get_inventory_weave()
512
all_revisions = self._eliminate_revisions_not_present(
513
a_weave.versions())
514
entire_graph = dict([(node, a_weave.get_parents(node)) for
515
node in all_revisions])
516
if revision_id is None:
517
return entire_graph
518
elif revision_id not in entire_graph:
519
raise errors.NoSuchRevision(self, revision_id)
520
else:
521
# add what can be reached from revision_id
522
result = {}
523
pending = set([revision_id])
524
while len(pending) > 0:
525
node = pending.pop()
526
result[node] = entire_graph[node]
527
for revision_id in result[node]:
528
if revision_id not in result:
529
pending.add(revision_id)
530
return result
531
532
@needs_read_lock
533
def get_revision_graph_with_ghosts(self, revision_ids=None):
534
"""Return a graph of the revisions with ghosts marked as applicable.
535
536
:param revision_ids: an iterable of revisions to graph or None for all.
537
:return: a Graph object with the graph reachable from revision_ids.
538
"""
539
result = graph.Graph()
540
if not revision_ids:
541
pending = set(self.all_revision_ids())
542
required = set([])
543
else:
544
pending = set(revision_ids)
545
# special case NULL_REVISION
546
if _mod_revision.NULL_REVISION in pending:
547
pending.remove(_mod_revision.NULL_REVISION)
548
required = set(pending)
549
done = set([])
550
while len(pending):
551
revision_id = pending.pop()
552
try:
553
rev = self.get_revision(revision_id)
554
except errors.NoSuchRevision:
555
if revision_id in required:
556
raise
557
# a ghost
558
result.add_ghost(revision_id)
559
continue
560
for parent_id in rev.parent_ids:
561
# is this queued or done ?
562
if (parent_id not in pending and
563
parent_id not in done):
564
# no, queue it.
565
pending.add(parent_id)
566
result.add_node(revision_id, rev.parent_ids)
567
done.add(revision_id)
568
return result
569
570
@needs_read_lock
571
def get_revision_inventory(self, revision_id):
572
"""Return inventory of a past revision."""
573
# TODO: Unify this with get_inventory()
574
# bzr 0.0.6 and later imposes the constraint that the inventory_id
575
# must be the same as its revision, so this is trivial.
576
if revision_id is None:
577
# This does not make sense: if there is no revision,
578
# then it is the current tree inventory surely ?!
579
# and thus get_root_id() is something that looks at the last
580
# commit on the branch, and the get_root_id is an inventory check.
581
raise NotImplementedError
582
# return Inventory(self.get_root_id())
583
else:
584
return self.get_inventory(revision_id)
585
586
@needs_read_lock
587
def is_shared(self):
588
"""Return True if this repository is flagged as a shared repository."""
589
raise NotImplementedError(self.is_shared)
590
591
@needs_write_lock
592
def reconcile(self, other=None, thorough=False):
593
"""Reconcile this repository."""
594
from bzrlib.reconcile import RepoReconciler
595
reconciler = RepoReconciler(self, thorough=thorough)
596
reconciler.reconcile()
597
return reconciler
598
599
@needs_read_lock
600
def revision_tree(self, revision_id):
601
"""Return Tree for a revision on this branch.
602
603
`revision_id` may be None for the empty tree revision.
604
"""
605
# TODO: refactor this to use an existing revision object
606
# so we don't need to read it in twice.
607
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
608
return RevisionTree(self, Inventory(root_id=None),
609
_mod_revision.NULL_REVISION)
610
else:
611
inv = self.get_revision_inventory(revision_id)
612
return RevisionTree(self, inv, revision_id)
613
614
@needs_read_lock
615
def revision_trees(self, revision_ids):
616
"""Return Tree for a revision on this branch.
617
618
`revision_id` may not be None or 'null:'"""
619
assert None not in revision_ids
620
assert _mod_revision.NULL_REVISION not in revision_ids
621
texts = self.get_inventory_weave().get_texts(revision_ids)
622
for text, revision_id in zip(texts, revision_ids):
623
inv = self.deserialise_inventory(revision_id, text)
624
yield RevisionTree(self, inv, revision_id)
625
626
@needs_read_lock
627
def get_ancestry(self, revision_id):
628
"""Return a list of revision-ids integrated by a revision.
629
630
The first element of the list is always None, indicating the origin
631
revision. This might change when we have history horizons, or
632
perhaps we should have a new API.
633
634
This is topologically sorted.
635
"""
636
if revision_id is None:
637
return [None]
638
if not self.has_revision(revision_id):
639
raise errors.NoSuchRevision(self, revision_id)
640
w = self.get_inventory_weave()
641
candidates = w.get_ancestry(revision_id)
642
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
643
644
@needs_read_lock
645
def print_file(self, file, revision_id):
646
"""Print `file` to stdout.
647
648
FIXME RBC 20060125 as John Meinel points out this is a bad api
649
- it writes to stdout, it assumes that that is valid etc. Fix
650
by creating a new more flexible convenience function.
651
"""
652
tree = self.revision_tree(revision_id)
653
# use inventory as it was in that revision
654
file_id = tree.inventory.path2id(file)
655
if not file_id:
656
# TODO: jam 20060427 Write a test for this code path
657
# it had a bug in it, and was raising the wrong
658
# exception.
659
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
660
tree.print_file(file_id)
661
662
def get_transaction(self):
663
return self.control_files.get_transaction()
664
665
def revision_parents(self, revid):
666
return self.get_inventory_weave().parent_names(revid)
667
668
@needs_write_lock
669
def set_make_working_trees(self, new_value):
670
"""Set the policy flag for making working trees when creating branches.
671
672
This only applies to branches that use this repository.
673
674
The default is 'True'.
675
:param new_value: True to restore the default, False to disable making
676
working trees.
677
"""
678
raise NotImplementedError(self.set_make_working_trees)
679
680
def make_working_trees(self):
681
"""Returns the policy for making working trees on new branches."""
682
raise NotImplementedError(self.make_working_trees)
683
684
@needs_write_lock
685
def sign_revision(self, revision_id, gpg_strategy):
686
plaintext = Testament.from_revision(self, revision_id).as_short_text()
687
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
688
689
@needs_read_lock
690
def has_signature_for_revision_id(self, revision_id):
691
"""Query for a revision signature for revision_id in the repository."""
692
return self._revision_store.has_signature(revision_id,
693
self.get_transaction())
694
695
@needs_read_lock
696
def get_signature_text(self, revision_id):
697
"""Return the text for a signature."""
698
return self._revision_store.get_signature_text(revision_id,
699
self.get_transaction())
700
701
@needs_read_lock
702
def check(self, revision_ids):
703
"""Check consistency of all history of given revision_ids.
704
705
Different repository implementations should override _check().
706
707
:param revision_ids: A non-empty list of revision_ids whose ancestry
708
will be checked. Typically the last revision_id of a branch.
709
"""
710
if not revision_ids:
711
raise ValueError("revision_ids must be non-empty in %s.check"
712
% (self,))
713
return self._check(revision_ids)
714
715
def _check(self, revision_ids):
716
result = check.Check(self)
717
result.check()
718
return result
719
720
def _warn_if_deprecated(self):
721
global _deprecation_warning_done
722
if _deprecation_warning_done:
723
return
724
_deprecation_warning_done = True
725
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
726
% (self._format, self.bzrdir.transport.base))
727
728
def supports_rich_root(self):
729
return self._format.rich_root_data
730
731
732
class AllInOneRepository(Repository):
733
"""Legacy support - the repository behaviour for all-in-one branches."""
734
735
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
736
# we reuse one control files instance.
737
dir_mode = a_bzrdir._control_files._dir_mode
738
file_mode = a_bzrdir._control_files._file_mode
739
740
def get_store(name, compressed=True, prefixed=False):
741
# FIXME: This approach of assuming stores are all entirely compressed
742
# or entirely uncompressed is tidy, but breaks upgrade from
743
# some existing branches where there's a mixture; we probably
744
# still want the option to look for both.
745
relpath = a_bzrdir._control_files._escape(name)
746
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
747
prefixed=prefixed, compressed=compressed,
748
dir_mode=dir_mode,
749
file_mode=file_mode)
750
#if self._transport.should_cache():
751
# cache_path = os.path.join(self.cache_root, name)
752
# os.mkdir(cache_path)
753
# store = bzrlib.store.CachedStore(store, cache_path)
754
return store
755
756
# not broken out yet because the controlweaves|inventory_store
757
# and text_store | weave_store bits are still different.
758
if isinstance(_format, RepositoryFormat4):
759
# cannot remove these - there is still no consistent api
760
# which allows access to this old info.
761
self.inventory_store = get_store('inventory-store')
762
text_store = get_store('text-store')
763
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
764
765
@needs_read_lock
766
def is_shared(self):
767
"""AllInOne repositories cannot be shared."""
768
return False
769
770
@needs_write_lock
771
def set_make_working_trees(self, new_value):
772
"""Set the policy flag for making working trees when creating branches.
773
774
This only applies to branches that use this repository.
775
776
The default is 'True'.
777
:param new_value: True to restore the default, False to disable making
778
working trees.
779
"""
780
raise NotImplementedError(self.set_make_working_trees)
781
782
def make_working_trees(self):
783
"""Returns the policy for making working trees on new branches."""
784
return True
785
786
787
def install_revision(repository, rev, revision_tree):
788
"""Install all revision data into a repository."""
789
present_parents = []
790
parent_trees = {}
791
for p_id in rev.parent_ids:
792
if repository.has_revision(p_id):
793
present_parents.append(p_id)
794
parent_trees[p_id] = repository.revision_tree(p_id)
795
else:
796
parent_trees[p_id] = repository.revision_tree(None)
797
798
inv = revision_tree.inventory
799
entries = inv.iter_entries()
800
# backwards compatability hack: skip the root id.
801
if not repository.supports_rich_root():
802
path, root = entries.next()
803
if root.revision != rev.revision_id:
804
raise errors.IncompatibleRevision(repr(repository))
805
# Add the texts that are not already present
806
for path, ie in entries:
807
w = repository.weave_store.get_weave_or_empty(ie.file_id,
808
repository.get_transaction())
809
if ie.revision not in w:
810
text_parents = []
811
# FIXME: TODO: The following loop *may* be overlapping/duplicate
812
# with InventoryEntry.find_previous_heads(). if it is, then there
813
# is a latent bug here where the parents may have ancestors of each
814
# other. RBC, AB
815
for revision, tree in parent_trees.iteritems():
816
if ie.file_id not in tree:
817
continue
818
parent_id = tree.inventory[ie.file_id].revision
819
if parent_id in text_parents:
820
continue
821
text_parents.append(parent_id)
822
823
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
824
repository.get_transaction())
825
lines = revision_tree.get_file(ie.file_id).readlines()
826
vfile.add_lines(rev.revision_id, text_parents, lines)
827
try:
828
# install the inventory
829
repository.add_inventory(rev.revision_id, inv, present_parents)
830
except errors.RevisionAlreadyPresent:
831
pass
832
repository.add_revision(rev.revision_id, rev, inv)
833
834
835
class MetaDirRepository(Repository):
836
"""Repositories in the new meta-dir layout."""
837
838
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
839
super(MetaDirRepository, self).__init__(_format,
840
a_bzrdir,
841
control_files,
842
_revision_store,
843
control_store,
844
text_store)
845
dir_mode = self.control_files._dir_mode
846
file_mode = self.control_files._file_mode
847
848
@needs_read_lock
849
def is_shared(self):
850
"""Return True if this repository is flagged as a shared repository."""
851
return self.control_files._transport.has('shared-storage')
852
853
@needs_write_lock
854
def set_make_working_trees(self, new_value):
855
"""Set the policy flag for making working trees when creating branches.
856
857
This only applies to branches that use this repository.
858
859
The default is 'True'.
860
:param new_value: True to restore the default, False to disable making
861
working trees.
862
"""
863
if new_value:
864
try:
865
self.control_files._transport.delete('no-working-trees')
866
except errors.NoSuchFile:
867
pass
868
else:
869
self.control_files.put_utf8('no-working-trees', '')
870
871
def make_working_trees(self):
872
"""Returns the policy for making working trees on new branches."""
873
return not self.control_files._transport.has('no-working-trees')
874
875
876
class KnitRepository(MetaDirRepository):
877
"""Knit format repository."""
878
879
def _warn_if_deprecated(self):
880
# This class isn't deprecated
881
pass
882
883
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
884
inv_vf.add_lines_with_ghosts(revid, parents, lines)
885
886
@needs_read_lock
887
def _all_revision_ids(self):
888
"""See Repository.all_revision_ids()."""
889
# Knits get the revision graph from the index of the revision knit, so
890
# it's always possible even if they're on an unlistable transport.
891
return self._revision_store.all_revision_ids(self.get_transaction())
892
893
def fileid_involved_between_revs(self, from_revid, to_revid):
894
"""Find file_id(s) which are involved in the changes between revisions.
895
896
This determines the set of revisions which are involved, and then
897
finds all file ids affected by those revisions.
898
"""
899
vf = self._get_revision_vf()
900
from_set = set(vf.get_ancestry(from_revid))
901
to_set = set(vf.get_ancestry(to_revid))
902
changed = to_set.difference(from_set)
903
return self._fileid_involved_by_set(changed)
904
905
def fileid_involved(self, last_revid=None):
906
"""Find all file_ids modified in the ancestry of last_revid.
907
908
:param last_revid: If None, last_revision() will be used.
909
"""
910
if not last_revid:
911
changed = set(self.all_revision_ids())
912
else:
913
changed = set(self.get_ancestry(last_revid))
914
if None in changed:
915
changed.remove(None)
916
return self._fileid_involved_by_set(changed)
917
918
@needs_read_lock
919
def get_ancestry(self, revision_id):
920
"""Return a list of revision-ids integrated by a revision.
921
922
This is topologically sorted.
923
"""
924
if revision_id is None:
925
return [None]
926
vf = self._get_revision_vf()
927
try:
928
return [None] + vf.get_ancestry(revision_id)
929
except errors.RevisionNotPresent:
930
raise errors.NoSuchRevision(self, revision_id)
931
932
@needs_read_lock
933
def get_revision(self, revision_id):
934
"""Return the Revision object for a named revision"""
935
return self.get_revision_reconcile(revision_id)
936
937
@needs_read_lock
938
def get_revision_graph(self, revision_id=None):
939
"""Return a dictionary containing the revision graph.
940
941
:param revision_id: The revision_id to get a graph from. If None, then
942
the entire revision graph is returned. This is a deprecated mode of
943
operation and will be removed in the future.
944
:return: a dictionary of revision_id->revision_parents_list.
945
"""
946
# special case NULL_REVISION
947
if revision_id == _mod_revision.NULL_REVISION:
948
return {}
949
a_weave = self._get_revision_vf()
950
entire_graph = a_weave.get_graph()
951
if revision_id is None:
952
return a_weave.get_graph()
953
elif revision_id not in a_weave:
954
raise errors.NoSuchRevision(self, revision_id)
955
else:
956
# add what can be reached from revision_id
957
result = {}
958
pending = set([revision_id])
959
while len(pending) > 0:
960
node = pending.pop()
961
result[node] = a_weave.get_parents(node)
962
for revision_id in result[node]:
963
if revision_id not in result:
964
pending.add(revision_id)
965
return result
966
967
@needs_read_lock
968
def get_revision_graph_with_ghosts(self, revision_ids=None):
969
"""Return a graph of the revisions with ghosts marked as applicable.
970
971
:param revision_ids: an iterable of revisions to graph or None for all.
972
:return: a Graph object with the graph reachable from revision_ids.
973
"""
974
result = graph.Graph()
975
vf = self._get_revision_vf()
976
versions = set(vf.versions())
977
if not revision_ids:
978
pending = set(self.all_revision_ids())
979
required = set([])
980
else:
981
pending = set(revision_ids)
982
# special case NULL_REVISION
983
if _mod_revision.NULL_REVISION in pending:
984
pending.remove(_mod_revision.NULL_REVISION)
985
required = set(pending)
986
done = set([])
987
while len(pending):
988
revision_id = pending.pop()
989
if not revision_id in versions:
990
if revision_id in required:
991
raise errors.NoSuchRevision(self, revision_id)
992
# a ghost
993
result.add_ghost(revision_id)
994
# mark it as done so we don't try for it again.
995
done.add(revision_id)
996
continue
997
parent_ids = vf.get_parents_with_ghosts(revision_id)
998
for parent_id in parent_ids:
999
# is this queued or done ?
1000
if (parent_id not in pending and
1001
parent_id not in done):
1002
# no, queue it.
1003
pending.add(parent_id)
1004
result.add_node(revision_id, parent_ids)
1005
done.add(revision_id)
1006
return result
1007
1008
def _get_revision_vf(self):
1009
""":return: a versioned file containing the revisions."""
1010
vf = self._revision_store.get_revision_file(self.get_transaction())
1011
return vf
1012
1013
@needs_write_lock
1014
def reconcile(self, other=None, thorough=False):
1015
"""Reconcile this repository."""
1016
from bzrlib.reconcile import KnitReconciler
1017
reconciler = KnitReconciler(self, thorough=thorough)
1018
reconciler.reconcile()
1019
return reconciler
1020
1021
def revision_parents(self, revision_id):
1022
return self._get_revision_vf().get_parents(revision_id)
1023
1024
1025
class KnitRepository2(KnitRepository):
1026
""""""
1027
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1028
control_store, text_store):
1029
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1030
_revision_store, control_store, text_store)
1031
self._serializer = xml6.serializer_v6
1032
1033
def deserialise_inventory(self, revision_id, xml):
1034
"""Transform the xml into an inventory object.
1035
1036
:param revision_id: The expected revision id of the inventory.
1037
:param xml: A serialised inventory.
1038
"""
1039
result = self._serializer.read_inventory_from_string(xml)
1040
assert result.root.revision is not None
1041
return result
1042
1043
def serialise_inventory(self, inv):
1044
"""Transform the inventory object into XML text.
1045
1046
:param revision_id: The expected revision id of the inventory.
1047
:param xml: A serialised inventory.
1048
"""
1049
assert inv.revision_id is not None
1050
assert inv.root.revision is not None
1051
return KnitRepository.serialise_inventory(self, inv)
1052
1053
def get_commit_builder(self, branch, parents, config, timestamp=None,
1054
timezone=None, committer=None, revprops=None,
1055
revision_id=None):
1056
"""Obtain a CommitBuilder for this repository.
1057
1058
:param branch: Branch to commit to.
1059
:param parents: Revision ids of the parents of the new revision.
1060
:param config: Configuration to use.
1061
:param timestamp: Optional timestamp recorded for commit.
1062
:param timezone: Optional timezone for timestamp.
1063
:param committer: Optional committer to set for commit.
1064
:param revprops: Optional dictionary of revision properties.
1065
:param revision_id: Optional revision id.
1066
"""
1067
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1068
committer, revprops, revision_id)
1069
1070
1071
class RepositoryFormat(object):
1072
"""A repository format.
1073
1074
Formats provide three things:
1075
* An initialization routine to construct repository data on disk.
1076
* a format string which is used when the BzrDir supports versioned
1077
children.
1078
* an open routine which returns a Repository instance.
1079
1080
Formats are placed in an dict by their format string for reference
1081
during opening. These should be subclasses of RepositoryFormat
1082
for consistency.
1083
1084
Once a format is deprecated, just deprecate the initialize and open
1085
methods on the format class. Do not deprecate the object, as the
1086
object will be created every system load.
1087
1088
Common instance attributes:
1089
_matchingbzrdir - the bzrdir format that the repository format was
1090
originally written to work with. This can be used if manually
1091
constructing a bzrdir and repository, or more commonly for test suite
1092
parameterisation.
1093
"""
1094
1095
_default_format = None
1096
"""The default format used for new repositories."""
1097
1098
_formats = {}
1099
"""The known formats."""
1100
1101
def __str__(self):
1102
return "<%s>" % self.__class__.__name__
1103
1104
@classmethod
1105
def find_format(klass, a_bzrdir):
1106
"""Return the format for the repository object in a_bzrdir."""
1107
try:
1108
transport = a_bzrdir.get_repository_transport(None)
1109
format_string = transport.get("format").read()
1110
return klass._formats[format_string]
1111
except errors.NoSuchFile:
1112
raise errors.NoRepositoryPresent(a_bzrdir)
1113
except KeyError:
1114
raise errors.UnknownFormatError(format=format_string)
1115
1116
def _get_control_store(self, repo_transport, control_files):
1117
"""Return the control store for this repository."""
1118
raise NotImplementedError(self._get_control_store)
1119
1120
@classmethod
1121
def get_default_format(klass):
1122
"""Return the current default format."""
1123
return klass._default_format
1124
1125
def get_format_string(self):
1126
"""Return the ASCII format string that identifies this format.
1127
1128
Note that in pre format ?? repositories the format string is
1129
not permitted nor written to disk.
1130
"""
1131
raise NotImplementedError(self.get_format_string)
1132
1133
def get_format_description(self):
1134
"""Return the short description for this format."""
1135
raise NotImplementedError(self.get_format_description)
1136
1137
def _get_revision_store(self, repo_transport, control_files):
1138
"""Return the revision store object for this a_bzrdir."""
1139
raise NotImplementedError(self._get_revision_store)
1140
1141
def _get_text_rev_store(self,
1142
transport,
1143
control_files,
1144
name,
1145
compressed=True,
1146
prefixed=False,
1147
serializer=None):
1148
"""Common logic for getting a revision store for a repository.
1149
1150
see self._get_revision_store for the subclass-overridable method to
1151
get the store for a repository.
1152
"""
1153
from bzrlib.store.revision.text import TextRevisionStore
1154
dir_mode = control_files._dir_mode
1155
file_mode = control_files._file_mode
1156
text_store =TextStore(transport.clone(name),
1157
prefixed=prefixed,
1158
compressed=compressed,
1159
dir_mode=dir_mode,
1160
file_mode=file_mode)
1161
_revision_store = TextRevisionStore(text_store, serializer)
1162
return _revision_store
1163
1164
def _get_versioned_file_store(self,
1165
name,
1166
transport,
1167
control_files,
1168
prefixed=True,
1169
versionedfile_class=weave.WeaveFile,
1170
versionedfile_kwargs={},
1171
escaped=False):
1172
weave_transport = control_files._transport.clone(name)
1173
dir_mode = control_files._dir_mode
1174
file_mode = control_files._file_mode
1175
return VersionedFileStore(weave_transport, prefixed=prefixed,
1176
dir_mode=dir_mode,
1177
file_mode=file_mode,
1178
versionedfile_class=versionedfile_class,
1179
versionedfile_kwargs=versionedfile_kwargs,
1180
escaped=escaped)
1181
1182
def initialize(self, a_bzrdir, shared=False):
1183
"""Initialize a repository of this format in a_bzrdir.
1184
1185
:param a_bzrdir: The bzrdir to put the new repository in it.
1186
:param shared: The repository should be initialized as a sharable one.
1187
1188
This may raise UninitializableFormat if shared repository are not
1189
compatible the a_bzrdir.
1190
"""
1191
1192
def is_supported(self):
1193
"""Is this format supported?
1194
1195
Supported formats must be initializable and openable.
1196
Unsupported formats may not support initialization or committing or
1197
some other features depending on the reason for not being supported.
1198
"""
1199
return True
1200
1201
def check_conversion_target(self, target_format):
1202
raise NotImplementedError(self.check_conversion_target)
1203
1204
def open(self, a_bzrdir, _found=False):
1205
"""Return an instance of this format for the bzrdir a_bzrdir.
1206
1207
_found is a private parameter, do not use it.
1208
"""
1209
raise NotImplementedError(self.open)
1210
1211
@classmethod
1212
def register_format(klass, format):
1213
klass._formats[format.get_format_string()] = format
1214
1215
@classmethod
1216
def set_default_format(klass, format):
1217
klass._default_format = format
1218
1219
@classmethod
1220
def unregister_format(klass, format):
1221
assert klass._formats[format.get_format_string()] is format
1222
del klass._formats[format.get_format_string()]
1223
1224
1225
class PreSplitOutRepositoryFormat(RepositoryFormat):
1226
"""Base class for the pre split out repository formats."""
1227
1228
rich_root_data = False
1229
1230
def initialize(self, a_bzrdir, shared=False, _internal=False):
1231
"""Create a weave repository.
1232
1233
TODO: when creating split out bzr branch formats, move this to a common
1234
base for Format5, Format6. or something like that.
1235
"""
1236
if shared:
1237
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1238
1239
if not _internal:
1240
# always initialized when the bzrdir is.
1241
return self.open(a_bzrdir, _found=True)
1242
1243
# Create an empty weave
1244
sio = StringIO()
1245
weavefile.write_weave_v5(weave.Weave(), sio)
1246
empty_weave = sio.getvalue()
1247
1248
mutter('creating repository in %s.', a_bzrdir.transport.base)
1249
dirs = ['revision-store', 'weaves']
1250
files = [('inventory.weave', StringIO(empty_weave)),
1251
]
1252
1253
# FIXME: RBC 20060125 don't peek under the covers
1254
# NB: no need to escape relative paths that are url safe.
1255
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1256
'branch-lock', lockable_files.TransportLock)
1257
control_files.create_lock()
1258
control_files.lock_write()
1259
control_files._transport.mkdir_multi(dirs,
1260
mode=control_files._dir_mode)
1261
try:
1262
for file, content in files:
1263
control_files.put(file, content)
1264
finally:
1265
control_files.unlock()
1266
return self.open(a_bzrdir, _found=True)
1267
1268
def _get_control_store(self, repo_transport, control_files):
1269
"""Return the control store for this repository."""
1270
return self._get_versioned_file_store('',
1271
repo_transport,
1272
control_files,
1273
prefixed=False)
1274
1275
def _get_text_store(self, transport, control_files):
1276
"""Get a store for file texts for this format."""
1277
raise NotImplementedError(self._get_text_store)
1278
1279
def open(self, a_bzrdir, _found=False):
1280
"""See RepositoryFormat.open()."""
1281
if not _found:
1282
# we are being called directly and must probe.
1283
raise NotImplementedError
1284
1285
repo_transport = a_bzrdir.get_repository_transport(None)
1286
control_files = a_bzrdir._control_files
1287
text_store = self._get_text_store(repo_transport, control_files)
1288
control_store = self._get_control_store(repo_transport, control_files)
1289
_revision_store = self._get_revision_store(repo_transport, control_files)
1290
return AllInOneRepository(_format=self,
1291
a_bzrdir=a_bzrdir,
1292
_revision_store=_revision_store,
1293
control_store=control_store,
1294
text_store=text_store)
1295
1296
def check_conversion_target(self, target_format):
1297
pass
1298
1299
1300
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1301
"""Bzr repository format 4.
1302
1303
This repository format has:
1304
- flat stores
1305
- TextStores for texts, inventories,revisions.
1306
1307
This format is deprecated: it indexes texts using a text id which is
1308
removed in format 5; initialization and write support for this format
1309
has been removed.
1310
"""
1311
1312
def __init__(self):
1313
super(RepositoryFormat4, self).__init__()
1314
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1315
1316
def get_format_description(self):
1317
"""See RepositoryFormat.get_format_description()."""
1318
return "Repository format 4"
1319
1320
def initialize(self, url, shared=False, _internal=False):
1321
"""Format 4 branches cannot be created."""
1322
raise errors.UninitializableFormat(self)
1323
1324
def is_supported(self):
1325
"""Format 4 is not supported.
1326
1327
It is not supported because the model changed from 4 to 5 and the
1328
conversion logic is expensive - so doing it on the fly was not
1329
feasible.
1330
"""
1331
return False
1332
1333
def _get_control_store(self, repo_transport, control_files):
1334
"""Format 4 repositories have no formal control store at this point.
1335
1336
This will cause any control-file-needing apis to fail - this is desired.
1337
"""
1338
return None
1339
1340
def _get_revision_store(self, repo_transport, control_files):
1341
"""See RepositoryFormat._get_revision_store()."""
1342
from bzrlib.xml4 import serializer_v4
1343
return self._get_text_rev_store(repo_transport,
1344
control_files,
1345
'revision-store',
1346
serializer=serializer_v4)
1347
1348
def _get_text_store(self, transport, control_files):
1349
"""See RepositoryFormat._get_text_store()."""
1350
1351
1352
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1353
"""Bzr control format 5.
1354
1355
This repository format has:
1356
- weaves for file texts and inventory
1357
- flat stores
1358
- TextStores for revisions and signatures.
1359
"""
1360
1361
def __init__(self):
1362
super(RepositoryFormat5, self).__init__()
1363
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1364
1365
def get_format_description(self):
1366
"""See RepositoryFormat.get_format_description()."""
1367
return "Weave repository format 5"
1368
1369
def _get_revision_store(self, repo_transport, control_files):
1370
"""See RepositoryFormat._get_revision_store()."""
1371
"""Return the revision store object for this a_bzrdir."""
1372
return self._get_text_rev_store(repo_transport,
1373
control_files,
1374
'revision-store',
1375
compressed=False)
1376
1377
def _get_text_store(self, transport, control_files):
1378
"""See RepositoryFormat._get_text_store()."""
1379
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1380
1381
1382
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1383
"""Bzr control format 6.
1384
1385
This repository format has:
1386
- weaves for file texts and inventory
1387
- hash subdirectory based stores.
1388
- TextStores for revisions and signatures.
1389
"""
1390
1391
def __init__(self):
1392
super(RepositoryFormat6, self).__init__()
1393
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1394
1395
def get_format_description(self):
1396
"""See RepositoryFormat.get_format_description()."""
1397
return "Weave repository format 6"
1398
1399
def _get_revision_store(self, repo_transport, control_files):
1400
"""See RepositoryFormat._get_revision_store()."""
1401
return self._get_text_rev_store(repo_transport,
1402
control_files,
1403
'revision-store',
1404
compressed=False,
1405
prefixed=True)
1406
1407
def _get_text_store(self, transport, control_files):
1408
"""See RepositoryFormat._get_text_store()."""
1409
return self._get_versioned_file_store('weaves', transport, control_files)
1410
1411
1412
class MetaDirRepositoryFormat(RepositoryFormat):
1413
"""Common base class for the new repositories using the metadir layout."""
1414
1415
rich_root_data = False
1416
1417
def __init__(self):
1418
super(MetaDirRepositoryFormat, self).__init__()
1419
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1420
1421
def _create_control_files(self, a_bzrdir):
1422
"""Create the required files and the initial control_files object."""
1423
# FIXME: RBC 20060125 don't peek under the covers
1424
# NB: no need to escape relative paths that are url safe.
1425
repository_transport = a_bzrdir.get_repository_transport(self)
1426
control_files = lockable_files.LockableFiles(repository_transport,
1427
'lock', lockdir.LockDir)
1428
control_files.create_lock()
1429
return control_files
1430
1431
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1432
"""Upload the initial blank content."""
1433
control_files = self._create_control_files(a_bzrdir)
1434
control_files.lock_write()
1435
try:
1436
control_files._transport.mkdir_multi(dirs,
1437
mode=control_files._dir_mode)
1438
for file, content in files:
1439
control_files.put(file, content)
1440
for file, content in utf8_files:
1441
control_files.put_utf8(file, content)
1442
if shared == True:
1443
control_files.put_utf8('shared-storage', '')
1444
finally:
1445
control_files.unlock()
1446
1447
1448
class RepositoryFormat7(MetaDirRepositoryFormat):
1449
"""Bzr repository 7.
1450
1451
This repository format has:
1452
- weaves for file texts and inventory
1453
- hash subdirectory based stores.
1454
- TextStores for revisions and signatures.
1455
- a format marker of its own
1456
- an optional 'shared-storage' flag
1457
- an optional 'no-working-trees' flag
1458
"""
1459
1460
def _get_control_store(self, repo_transport, control_files):
1461
"""Return the control store for this repository."""
1462
return self._get_versioned_file_store('',
1463
repo_transport,
1464
control_files,
1465
prefixed=False)
1466
1467
def get_format_string(self):
1468
"""See RepositoryFormat.get_format_string()."""
1469
return "Bazaar-NG Repository format 7"
1470
1471
def get_format_description(self):
1472
"""See RepositoryFormat.get_format_description()."""
1473
return "Weave repository format 7"
1474
1475
def check_conversion_target(self, target_format):
1476
pass
1477
1478
def _get_revision_store(self, repo_transport, control_files):
1479
"""See RepositoryFormat._get_revision_store()."""
1480
return self._get_text_rev_store(repo_transport,
1481
control_files,
1482
'revision-store',
1483
compressed=False,
1484
prefixed=True,
1485
)
1486
1487
def _get_text_store(self, transport, control_files):
1488
"""See RepositoryFormat._get_text_store()."""
1489
return self._get_versioned_file_store('weaves',
1490
transport,
1491
control_files)
1492
1493
def initialize(self, a_bzrdir, shared=False):
1494
"""Create a weave repository.
1495
1496
:param shared: If true the repository will be initialized as a shared
1497
repository.
1498
"""
1499
# Create an empty weave
1500
sio = StringIO()
1501
weavefile.write_weave_v5(weave.Weave(), sio)
1502
empty_weave = sio.getvalue()
1503
1504
mutter('creating repository in %s.', a_bzrdir.transport.base)
1505
dirs = ['revision-store', 'weaves']
1506
files = [('inventory.weave', StringIO(empty_weave)),
1507
]
1508
utf8_files = [('format', self.get_format_string())]
1509
1510
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1511
return self.open(a_bzrdir=a_bzrdir, _found=True)
1512
1513
def open(self, a_bzrdir, _found=False, _override_transport=None):
1514
"""See RepositoryFormat.open().
1515
1516
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1517
repository at a slightly different url
1518
than normal. I.e. during 'upgrade'.
1519
"""
1520
if not _found:
1521
format = RepositoryFormat.find_format(a_bzrdir)
1522
assert format.__class__ == self.__class__
1523
if _override_transport is not None:
1524
repo_transport = _override_transport
1525
else:
1526
repo_transport = a_bzrdir.get_repository_transport(None)
1527
control_files = lockable_files.LockableFiles(repo_transport,
1528
'lock', lockdir.LockDir)
1529
text_store = self._get_text_store(repo_transport, control_files)
1530
control_store = self._get_control_store(repo_transport, control_files)
1531
_revision_store = self._get_revision_store(repo_transport, control_files)
1532
return MetaDirRepository(_format=self,
1533
a_bzrdir=a_bzrdir,
1534
control_files=control_files,
1535
_revision_store=_revision_store,
1536
control_store=control_store,
1537
text_store=text_store)
1538
1539
1540
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1541
"""Bzr repository knit format (generalized).
1542
1543
This repository format has:
1544
- knits for file texts and inventory
1545
- hash subdirectory based stores.
1546
- knits for revisions and signatures
1547
- TextStores for revisions and signatures.
1548
- a format marker of its own
1549
- an optional 'shared-storage' flag
1550
- an optional 'no-working-trees' flag
1551
- a LockDir lock
1552
"""
1553
1554
def _get_control_store(self, repo_transport, control_files):
1555
"""Return the control store for this repository."""
1556
return VersionedFileStore(
1557
repo_transport,
1558
prefixed=False,
1559
file_mode=control_files._file_mode,
1560
versionedfile_class=knit.KnitVersionedFile,
1561
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1562
)
1563
1564
def _get_revision_store(self, repo_transport, control_files):
1565
"""See RepositoryFormat._get_revision_store()."""
1566
from bzrlib.store.revision.knit import KnitRevisionStore
1567
versioned_file_store = VersionedFileStore(
1568
repo_transport,
1569
file_mode=control_files._file_mode,
1570
prefixed=False,
1571
precious=True,
1572
versionedfile_class=knit.KnitVersionedFile,
1573
versionedfile_kwargs={'delta':False,
1574
'factory':knit.KnitPlainFactory(),
1575
},
1576
escaped=True,
1577
)
1578
return KnitRevisionStore(versioned_file_store)
1579
1580
def _get_text_store(self, transport, control_files):
1581
"""See RepositoryFormat._get_text_store()."""
1582
return self._get_versioned_file_store('knits',
1583
transport,
1584
control_files,
1585
versionedfile_class=knit.KnitVersionedFile,
1586
versionedfile_kwargs={
1587
'create_parent_dir':True,
1588
'delay_create':True,
1589
'dir_mode':control_files._dir_mode,
1590
},
1591
escaped=True)
1592
1593
def initialize(self, a_bzrdir, shared=False):
1594
"""Create a knit format 1 repository.
1595
1596
:param a_bzrdir: bzrdir to contain the new repository; must already
1597
be initialized.
1598
:param shared: If true the repository will be initialized as a shared
1599
repository.
1600
"""
1601
mutter('creating repository in %s.', a_bzrdir.transport.base)
1602
dirs = ['revision-store', 'knits']
1603
files = []
1604
utf8_files = [('format', self.get_format_string())]
1605
1606
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1607
repo_transport = a_bzrdir.get_repository_transport(None)
1608
control_files = lockable_files.LockableFiles(repo_transport,
1609
'lock', lockdir.LockDir)
1610
control_store = self._get_control_store(repo_transport, control_files)
1611
transaction = transactions.WriteTransaction()
1612
# trigger a write of the inventory store.
1613
control_store.get_weave_or_empty('inventory', transaction)
1614
_revision_store = self._get_revision_store(repo_transport, control_files)
1615
_revision_store.has_revision_id('A', transaction)
1616
_revision_store.get_signature_file(transaction)
1617
return self.open(a_bzrdir=a_bzrdir, _found=True)
1618
1619
def open(self, a_bzrdir, _found=False, _override_transport=None):
1620
"""See RepositoryFormat.open().
1621
1622
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1623
repository at a slightly different url
1624
than normal. I.e. during 'upgrade'.
1625
"""
1626
if not _found:
1627
format = RepositoryFormat.find_format(a_bzrdir)
1628
assert format.__class__ == self.__class__
1629
if _override_transport is not None:
1630
repo_transport = _override_transport
1631
else:
1632
repo_transport = a_bzrdir.get_repository_transport(None)
1633
control_files = lockable_files.LockableFiles(repo_transport,
1634
'lock', lockdir.LockDir)
1635
text_store = self._get_text_store(repo_transport, control_files)
1636
control_store = self._get_control_store(repo_transport, control_files)
1637
_revision_store = self._get_revision_store(repo_transport, control_files)
1638
return KnitRepository(_format=self,
1639
a_bzrdir=a_bzrdir,
1640
control_files=control_files,
1641
_revision_store=_revision_store,
1642
control_store=control_store,
1643
text_store=text_store)
1644
1645
1646
class RepositoryFormatKnit1(RepositoryFormatKnit):
1647
"""Bzr repository knit format 1.
1648
1649
This repository format has:
1650
- knits for file texts and inventory
1651
- hash subdirectory based stores.
1652
- knits for revisions and signatures
1653
- TextStores for revisions and signatures.
1654
- a format marker of its own
1655
- an optional 'shared-storage' flag
1656
- an optional 'no-working-trees' flag
1657
- a LockDir lock
1658
1659
This format was introduced in bzr 0.8.
1660
"""
1661
def get_format_string(self):
1662
"""See RepositoryFormat.get_format_string()."""
1663
return "Bazaar-NG Knit Repository Format 1"
1664
1665
def get_format_description(self):
1666
"""See RepositoryFormat.get_format_description()."""
1667
return "Knit repository format 1"
1668
1669
def check_conversion_target(self, target_format):
1670
pass
1671
1672
1673
class RepositoryFormatKnit2(RepositoryFormatKnit):
1674
"""Bzr repository knit format 2.
1675
1676
THIS FORMAT IS EXPERIMENTAL
1677
This repository format has:
1678
- knits for file texts and inventory
1679
- hash subdirectory based stores.
1680
- knits for revisions and signatures
1681
- TextStores for revisions and signatures.
1682
- a format marker of its own
1683
- an optional 'shared-storage' flag
1684
- an optional 'no-working-trees' flag
1685
- a LockDir lock
1686
- Support for recording full info about the tree root
1687
1688
"""
1689
1690
rich_root_data = True
1691
1692
def get_format_string(self):
1693
"""See RepositoryFormat.get_format_string()."""
1694
return "Bazaar Knit Repository Format 2\n"
1695
1696
def get_format_description(self):
1697
"""See RepositoryFormat.get_format_description()."""
1698
return "Knit repository format 2"
1699
1700
def check_conversion_target(self, target_format):
1701
if not target_format.rich_root_data:
1702
raise errors.BadConversionTarget(
1703
'Does not support rich root data.', target_format)
1704
1705
def open(self, a_bzrdir, _found=False, _override_transport=None):
1706
"""See RepositoryFormat.open().
1707
1708
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1709
repository at a slightly different url
1710
than normal. I.e. during 'upgrade'.
1711
"""
1712
if not _found:
1713
format = RepositoryFormat.find_format(a_bzrdir)
1714
assert format.__class__ == self.__class__
1715
if _override_transport is not None:
1716
repo_transport = _override_transport
1717
else:
1718
repo_transport = a_bzrdir.get_repository_transport(None)
1719
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1720
lockdir.LockDir)
1721
text_store = self._get_text_store(repo_transport, control_files)
1722
control_store = self._get_control_store(repo_transport, control_files)
1723
_revision_store = self._get_revision_store(repo_transport, control_files)
1724
return KnitRepository2(_format=self,
1725
a_bzrdir=a_bzrdir,
1726
control_files=control_files,
1727
_revision_store=_revision_store,
1728
control_store=control_store,
1729
text_store=text_store)
1730
1731
1732
1733
# formats which have no format string are not discoverable
1734
# and not independently creatable, so are not registered.
1735
RepositoryFormat.register_format(RepositoryFormat7())
1736
_default_format = RepositoryFormatKnit1()
1737
RepositoryFormat.register_format(_default_format)
1738
RepositoryFormat.register_format(RepositoryFormatKnit2())
1739
RepositoryFormat.set_default_format(_default_format)
1740
_legacy_formats = [RepositoryFormat4(),
1741
RepositoryFormat5(),
1742
RepositoryFormat6()]
1743
1744
1745
class InterRepository(InterObject):
1746
"""This class represents operations taking place between two repositories.
1747
1748
Its instances have methods like copy_content and fetch, and contain
1749
references to the source and target repositories these operations can be
1750
carried out on.
1751
1752
Often we will provide convenience methods on 'repository' which carry out
1753
operations with another repository - they will always forward to
1754
InterRepository.get(other).method_name(parameters).
1755
"""
1756
1757
_optimisers = []
1758
"""The available optimised InterRepository types."""
1759
1760
def copy_content(self, revision_id=None, basis=None):
1761
raise NotImplementedError(self.copy_content)
1762
1763
def fetch(self, revision_id=None, pb=None):
1764
"""Fetch the content required to construct revision_id.
1765
1766
The content is copied from self.source to self.target.
1767
1768
:param revision_id: if None all content is copied, if NULL_REVISION no
1769
content is copied.
1770
:param pb: optional progress bar to use for progress reports. If not
1771
provided a default one will be created.
1772
1773
Returns the copied revision count and the failed revisions in a tuple:
1774
(copied, failures).
1775
"""
1776
raise NotImplementedError(self.fetch)
1777
1778
@needs_read_lock
1779
def missing_revision_ids(self, revision_id=None):
1780
"""Return the revision ids that source has that target does not.
1781
1782
These are returned in topological order.
1783
1784
:param revision_id: only return revision ids included by this
1785
revision_id.
1786
"""
1787
# generic, possibly worst case, slow code path.
1788
target_ids = set(self.target.all_revision_ids())
1789
if revision_id is not None:
1790
source_ids = self.source.get_ancestry(revision_id)
1791
assert source_ids[0] is None
1792
source_ids.pop(0)
1793
else:
1794
source_ids = self.source.all_revision_ids()
1795
result_set = set(source_ids).difference(target_ids)
1796
# this may look like a no-op: its not. It preserves the ordering
1797
# other_ids had while only returning the members from other_ids
1798
# that we've decided we need.
1799
return [rev_id for rev_id in source_ids if rev_id in result_set]
1800
1801
1802
class InterSameDataRepository(InterRepository):
1803
"""Code for converting between repositories that represent the same data.
1804
1805
Data format and model must match for this to work.
1806
"""
1807
1808
_matching_repo_format = RepositoryFormat4()
1809
"""Repository format for testing with."""
1810
1811
@staticmethod
1812
def is_compatible(source, target):
1813
if not isinstance(source, Repository):
1814
return False
1815
if not isinstance(target, Repository):
1816
return False
1817
if source._format.rich_root_data == target._format.rich_root_data:
1818
return True
1819
else:
1820
return False
1821
1822
@needs_write_lock
1823
def copy_content(self, revision_id=None, basis=None):
1824
"""Make a complete copy of the content in self into destination.
1825
1826
This is a destructive operation! Do not use it on existing
1827
repositories.
1828
1829
:param revision_id: Only copy the content needed to construct
1830
revision_id and its parents.
1831
:param basis: Copy the needed data preferentially from basis.
1832
"""
1833
try:
1834
self.target.set_make_working_trees(self.source.make_working_trees())
1835
except NotImplementedError:
1836
pass
1837
# grab the basis available data
1838
if basis is not None:
1839
self.target.fetch(basis, revision_id=revision_id)
1840
# but don't bother fetching if we have the needed data now.
1841
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1842
self.target.has_revision(revision_id)):
1843
return
1844
self.target.fetch(self.source, revision_id=revision_id)
1845
1846
@needs_write_lock
1847
def fetch(self, revision_id=None, pb=None):
1848
"""See InterRepository.fetch()."""
1849
from bzrlib.fetch import GenericRepoFetcher
1850
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1851
self.source, self.source._format, self.target,
1852
self.target._format)
1853
f = GenericRepoFetcher(to_repository=self.target,
1854
from_repository=self.source,
1855
last_revision=revision_id,
1856
pb=pb)
1857
return f.count_copied, f.failed_revisions
1858
1859
1860
class InterWeaveRepo(InterSameDataRepository):
1861
"""Optimised code paths between Weave based repositories."""
1862
1863
_matching_repo_format = RepositoryFormat7()
1864
"""Repository format for testing with."""
1865
1866
@staticmethod
1867
def is_compatible(source, target):
1868
"""Be compatible with known Weave formats.
1869
1870
We don't test for the stores being of specific types because that
1871
could lead to confusing results, and there is no need to be
1872
overly general.
1873
"""
1874
try:
1875
return (isinstance(source._format, (RepositoryFormat5,
1876
RepositoryFormat6,
1877
RepositoryFormat7)) and
1878
isinstance(target._format, (RepositoryFormat5,
1879
RepositoryFormat6,
1880
RepositoryFormat7)))
1881
except AttributeError:
1882
return False
1883
1884
@needs_write_lock
1885
def copy_content(self, revision_id=None, basis=None):
1886
"""See InterRepository.copy_content()."""
1887
# weave specific optimised path:
1888
if basis is not None:
1889
# copy the basis in, then fetch remaining data.
1890
basis.copy_content_into(self.target, revision_id)
1891
# the basis copy_content_into could miss-set this.
1892
try:
1893
self.target.set_make_working_trees(self.source.make_working_trees())
1894
except NotImplementedError:
1895
pass
1896
self.target.fetch(self.source, revision_id=revision_id)
1897
else:
1898
try:
1899
self.target.set_make_working_trees(self.source.make_working_trees())
1900
except NotImplementedError:
1901
pass
1902
# FIXME do not peek!
1903
if self.source.control_files._transport.listable():
1904
pb = ui.ui_factory.nested_progress_bar()
1905
try:
1906
self.target.weave_store.copy_all_ids(
1907
self.source.weave_store,
1908
pb=pb,
1909
from_transaction=self.source.get_transaction(),
1910
to_transaction=self.target.get_transaction())
1911
pb.update('copying inventory', 0, 1)
1912
self.target.control_weaves.copy_multi(
1913
self.source.control_weaves, ['inventory'],
1914
from_transaction=self.source.get_transaction(),
1915
to_transaction=self.target.get_transaction())
1916
self.target._revision_store.text_store.copy_all_ids(
1917
self.source._revision_store.text_store,
1918
pb=pb)
1919
finally:
1920
pb.finished()
1921
else:
1922
self.target.fetch(self.source, revision_id=revision_id)
1923
1924
@needs_write_lock
1925
def fetch(self, revision_id=None, pb=None):
1926
"""See InterRepository.fetch()."""
1927
from bzrlib.fetch import GenericRepoFetcher
1928
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1929
self.source, self.source._format, self.target, self.target._format)
1930
f = GenericRepoFetcher(to_repository=self.target,
1931
from_repository=self.source,
1932
last_revision=revision_id,
1933
pb=pb)
1934
return f.count_copied, f.failed_revisions
1935
1936
@needs_read_lock
1937
def missing_revision_ids(self, revision_id=None):
1938
"""See InterRepository.missing_revision_ids()."""
1939
# we want all revisions to satisfy revision_id in source.
1940
# but we don't want to stat every file here and there.
1941
# we want then, all revisions other needs to satisfy revision_id
1942
# checked, but not those that we have locally.
1943
# so the first thing is to get a subset of the revisions to
1944
# satisfy revision_id in source, and then eliminate those that
1945
# we do already have.
1946
# this is slow on high latency connection to self, but as as this
1947
# disk format scales terribly for push anyway due to rewriting
1948
# inventory.weave, this is considered acceptable.
1949
# - RBC 20060209
1950
if revision_id is not None:
1951
source_ids = self.source.get_ancestry(revision_id)
1952
assert source_ids[0] is None
1953
source_ids.pop(0)
1954
else:
1955
source_ids = self.source._all_possible_ids()
1956
source_ids_set = set(source_ids)
1957
# source_ids is the worst possible case we may need to pull.
1958
# now we want to filter source_ids against what we actually
1959
# have in target, but don't try to check for existence where we know
1960
# we do not have a revision as that would be pointless.
1961
target_ids = set(self.target._all_possible_ids())
1962
possibly_present_revisions = target_ids.intersection(source_ids_set)
1963
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1964
required_revisions = source_ids_set.difference(actually_present_revisions)
1965
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1966
if revision_id is not None:
1967
# we used get_ancestry to determine source_ids then we are assured all
1968
# revisions referenced are present as they are installed in topological order.
1969
# and the tip revision was validated by get_ancestry.
1970
return required_topo_revisions
1971
else:
1972
# if we just grabbed the possibly available ids, then
1973
# we only have an estimate of whats available and need to validate
1974
# that against the revision records.
1975
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1976
1977
1978
class InterKnitRepo(InterSameDataRepository):
1979
"""Optimised code paths between Knit based repositories."""
1980
1981
_matching_repo_format = RepositoryFormatKnit1()
1982
"""Repository format for testing with."""
1983
1984
@staticmethod
1985
def is_compatible(source, target):
1986
"""Be compatible with known Knit formats.
1987
1988
We don't test for the stores being of specific types because that
1989
could lead to confusing results, and there is no need to be
1990
overly general.
1991
"""
1992
try:
1993
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1994
isinstance(target._format, (RepositoryFormatKnit1)))
1995
except AttributeError:
1996
return False
1997
1998
@needs_write_lock
1999
def fetch(self, revision_id=None, pb=None):
2000
"""See InterRepository.fetch()."""
2001
from bzrlib.fetch import KnitRepoFetcher
2002
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2003
self.source, self.source._format, self.target, self.target._format)
2004
f = KnitRepoFetcher(to_repository=self.target,
2005
from_repository=self.source,
2006
last_revision=revision_id,
2007
pb=pb)
2008
return f.count_copied, f.failed_revisions
2009
2010
@needs_read_lock
2011
def missing_revision_ids(self, revision_id=None):
2012
"""See InterRepository.missing_revision_ids()."""
2013
if revision_id is not None:
2014
source_ids = self.source.get_ancestry(revision_id)
2015
assert source_ids[0] is None
2016
source_ids.pop(0)
2017
else:
2018
source_ids = self.source._all_possible_ids()
2019
source_ids_set = set(source_ids)
2020
# source_ids is the worst possible case we may need to pull.
2021
# now we want to filter source_ids against what we actually
2022
# have in target, but don't try to check for existence where we know
2023
# we do not have a revision as that would be pointless.
2024
target_ids = set(self.target._all_possible_ids())
2025
possibly_present_revisions = target_ids.intersection(source_ids_set)
2026
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2027
required_revisions = source_ids_set.difference(actually_present_revisions)
2028
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2029
if revision_id is not None:
2030
# we used get_ancestry to determine source_ids then we are assured all
2031
# revisions referenced are present as they are installed in topological order.
2032
# and the tip revision was validated by get_ancestry.
2033
return required_topo_revisions
2034
else:
2035
# if we just grabbed the possibly available ids, then
2036
# we only have an estimate of whats available and need to validate
2037
# that against the revision records.
2038
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2039
2040
2041
class InterModel1and2(InterRepository):
2042
2043
_matching_repo_format = None
2044
2045
@staticmethod
2046
def is_compatible(source, target):
2047
if not isinstance(source, Repository):
2048
return False
2049
if not isinstance(target, Repository):
2050
return False
2051
if not source._format.rich_root_data and target._format.rich_root_data:
2052
return True
2053
else:
2054
return False
2055
2056
@needs_write_lock
2057
def fetch(self, revision_id=None, pb=None):
2058
"""See InterRepository.fetch()."""
2059
from bzrlib.fetch import Model1toKnit2Fetcher
2060
f = Model1toKnit2Fetcher(to_repository=self.target,
2061
from_repository=self.source,
2062
last_revision=revision_id,
2063
pb=pb)
2064
return f.count_copied, f.failed_revisions
2065
2066
@needs_write_lock
2067
def copy_content(self, revision_id=None, basis=None):
2068
"""Make a complete copy of the content in self into destination.
2069
2070
This is a destructive operation! Do not use it on existing
2071
repositories.
2072
2073
:param revision_id: Only copy the content needed to construct
2074
revision_id and its parents.
2075
:param basis: Copy the needed data preferentially from basis.
2076
"""
2077
try:
2078
self.target.set_make_working_trees(self.source.make_working_trees())
2079
except NotImplementedError:
2080
pass
2081
# grab the basis available data
2082
if basis is not None:
2083
self.target.fetch(basis, revision_id=revision_id)
2084
# but don't bother fetching if we have the needed data now.
2085
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2086
self.target.has_revision(revision_id)):
2087
return
2088
self.target.fetch(self.source, revision_id=revision_id)
2089
2090
2091
class InterKnit1and2(InterKnitRepo):
2092
2093
_matching_repo_format = None
2094
2095
@staticmethod
2096
def is_compatible(source, target):
2097
"""Be compatible with Knit1 source and Knit2 target"""
2098
try:
2099
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2100
isinstance(target._format, (RepositoryFormatKnit2)))
2101
except AttributeError:
2102
return False
2103
2104
@needs_write_lock
2105
def fetch(self, revision_id=None, pb=None):
2106
"""See InterRepository.fetch()."""
2107
from bzrlib.fetch import Knit1to2Fetcher
2108
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2109
self.source, self.source._format, self.target,
2110
self.target._format)
2111
f = Knit1to2Fetcher(to_repository=self.target,
2112
from_repository=self.source,
2113
last_revision=revision_id,
2114
pb=pb)
2115
return f.count_copied, f.failed_revisions
2116
2117
2118
InterRepository.register_optimiser(InterSameDataRepository)
2119
InterRepository.register_optimiser(InterWeaveRepo)
2120
InterRepository.register_optimiser(InterKnitRepo)
2121
InterRepository.register_optimiser(InterModel1and2)
2122
InterRepository.register_optimiser(InterKnit1and2)
2123
2124
2125
class RepositoryTestProviderAdapter(object):
2126
"""A tool to generate a suite testing multiple repository formats at once.
2127
2128
This is done by copying the test once for each transport and injecting
2129
the transport_server, transport_readonly_server, and bzrdir_format and
2130
repository_format classes into each copy. Each copy is also given a new id()
2131
to make it easy to identify.
2132
"""
2133
2134
def __init__(self, transport_server, transport_readonly_server, formats):
2135
self._transport_server = transport_server
2136
self._transport_readonly_server = transport_readonly_server
2137
self._formats = formats
2138
2139
def adapt(self, test):
2140
result = unittest.TestSuite()
2141
for repository_format, bzrdir_format in self._formats:
2142
new_test = deepcopy(test)
2143
new_test.transport_server = self._transport_server
2144
new_test.transport_readonly_server = self._transport_readonly_server
2145
new_test.bzrdir_format = bzrdir_format
2146
new_test.repository_format = repository_format
2147
def make_new_test_id():
2148
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2149
return lambda: new_id
2150
new_test.id = make_new_test_id()
2151
result.addTest(new_test)
2152
return result
2153
2154
2155
class InterRepositoryTestProviderAdapter(object):
2156
"""A tool to generate a suite testing multiple inter repository formats.
2157
2158
This is done by copying the test once for each interrepo provider and injecting
2159
the transport_server, transport_readonly_server, repository_format and
2160
repository_to_format classes into each copy.
2161
Each copy is also given a new id() to make it easy to identify.
2162
"""
2163
2164
def __init__(self, transport_server, transport_readonly_server, formats):
2165
self._transport_server = transport_server
2166
self._transport_readonly_server = transport_readonly_server
2167
self._formats = formats
2168
2169
def adapt(self, test):
2170
result = unittest.TestSuite()
2171
for interrepo_class, repository_format, repository_format_to in self._formats:
2172
new_test = deepcopy(test)
2173
new_test.transport_server = self._transport_server
2174
new_test.transport_readonly_server = self._transport_readonly_server
2175
new_test.interrepo_class = interrepo_class
2176
new_test.repository_format = repository_format
2177
new_test.repository_format_to = repository_format_to
2178
def make_new_test_id():
2179
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2180
return lambda: new_id
2181
new_test.id = make_new_test_id()
2182
result.addTest(new_test)
2183
return result
2184
2185
@staticmethod
2186
def default_test_list():
2187
"""Generate the default list of interrepo permutations to test."""
2188
result = []
2189
# test the default InterRepository between format 6 and the current
2190
# default format.
2191
# XXX: robertc 20060220 reinstate this when there are two supported
2192
# formats which do not have an optimal code path between them.
2193
#result.append((InterRepository,
2194
# RepositoryFormat6(),
2195
# RepositoryFormatKnit1()))
2196
for optimiser in InterRepository._optimisers:
2197
if optimiser._matching_repo_format is not None:
2198
result.append((optimiser,
2199
optimiser._matching_repo_format,
2200
optimiser._matching_repo_format
2201
))
2202
# if there are specific combinations we want to use, we can add them
2203
# here.
2204
result.append((InterModel1and2, RepositoryFormat5(),
2205
RepositoryFormatKnit2()))
2206
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2207
RepositoryFormatKnit2()))
2208
return result
2209
2210
2211
class CopyConverter(object):
2212
"""A repository conversion tool which just performs a copy of the content.
2213
2214
This is slow but quite reliable.
2215
"""
2216
2217
def __init__(self, target_format):
2218
"""Create a CopyConverter.
2219
2220
:param target_format: The format the resulting repository should be.
2221
"""
2222
self.target_format = target_format
2223
2224
def convert(self, repo, pb):
2225
"""Perform the conversion of to_convert, giving feedback via pb.
2226
2227
:param to_convert: The disk object to convert.
2228
:param pb: a progress bar to use for progress information.
2229
"""
2230
self.pb = pb
2231
self.count = 0
2232
self.total = 4
2233
# this is only useful with metadir layouts - separated repo content.
2234
# trigger an assertion if not such
2235
repo._format.get_format_string()
2236
self.repo_dir = repo.bzrdir
2237
self.step('Moving repository to repository.backup')
2238
self.repo_dir.transport.move('repository', 'repository.backup')
2239
backup_transport = self.repo_dir.transport.clone('repository.backup')
2240
repo._format.check_conversion_target(self.target_format)
2241
self.source_repo = repo._format.open(self.repo_dir,
2242
_found=True,
2243
_override_transport=backup_transport)
2244
self.step('Creating new repository')
2245
converted = self.target_format.initialize(self.repo_dir,
2246
self.source_repo.is_shared())
2247
converted.lock_write()
2248
try:
2249
self.step('Copying content into repository.')
2250
self.source_repo.copy_content_into(converted)
2251
finally:
2252
converted.unlock()
2253
self.step('Deleting old repository content.')
2254
self.repo_dir.transport.delete_tree('repository.backup')
2255
self.pb.note('repository converted')
2256
2257
def step(self, message):
2258
"""Update the pb by a step."""
2259
self.count +=1
2260
self.pb.update(message, self.count, self.total)
2261
2262
2263
class CommitBuilder(object):
2264
"""Provides an interface to build up a commit.
2265
2266
This allows describing a tree to be committed without needing to
2267
know the internals of the format of the repository.
2268
"""
2269
2270
record_root_entry = False
2271
def __init__(self, repository, parents, config, timestamp=None,
2272
timezone=None, committer=None, revprops=None,
2273
revision_id=None):
2274
"""Initiate a CommitBuilder.
2275
2276
:param repository: Repository to commit to.
2277
:param parents: Revision ids of the parents of the new revision.
2278
:param config: Configuration to use.
2279
:param timestamp: Optional timestamp recorded for commit.
2280
:param timezone: Optional timezone for timestamp.
2281
:param committer: Optional committer to set for commit.
2282
:param revprops: Optional dictionary of revision properties.
2283
:param revision_id: Optional revision id.
2284
"""
2285
self._config = config
2286
2287
if committer is None:
2288
self._committer = self._config.username()
2289
else:
2290
assert isinstance(committer, basestring), type(committer)
2291
self._committer = committer
2292
2293
self.new_inventory = Inventory(None)
2294
self._new_revision_id = revision_id
2295
self.parents = parents
2296
self.repository = repository
2297
2298
self._revprops = {}
2299
if revprops is not None:
2300
self._revprops.update(revprops)
2301
2302
if timestamp is None:
2303
timestamp = time.time()
2304
# Restrict resolution to 1ms
2305
self._timestamp = round(timestamp, 3)
2306
2307
if timezone is None:
2308
self._timezone = local_time_offset()
2309
else:
2310
self._timezone = int(timezone)
2311
2312
self._generate_revision_if_needed()
2313
2314
def commit(self, message):
2315
"""Make the actual commit.
2316
2317
:return: The revision id of the recorded revision.
2318
"""
2319
rev = _mod_revision.Revision(
2320
timestamp=self._timestamp,
2321
timezone=self._timezone,
2322
committer=self._committer,
2323
message=message,
2324
inventory_sha1=self.inv_sha1,
2325
revision_id=self._new_revision_id,
2326
properties=self._revprops)
2327
rev.parent_ids = self.parents
2328
self.repository.add_revision(self._new_revision_id, rev,
2329
self.new_inventory, self._config)
2330
return self._new_revision_id
2331
2332
def revision_tree(self):
2333
"""Return the tree that was just committed.
2334
2335
After calling commit() this can be called to get a RevisionTree
2336
representing the newly committed tree. This is preferred to
2337
calling Repository.revision_tree() because that may require
2338
deserializing the inventory, while we already have a copy in
2339
memory.
2340
"""
2341
return RevisionTree(self.repository, self.new_inventory,
2342
self._new_revision_id)
2343
2344
def finish_inventory(self):
2345
"""Tell the builder that the inventory is finished."""
2346
if self.new_inventory.root is None:
2347
symbol_versioning.warn('Root entry should be supplied to'
2348
' record_entry_contents, as of bzr 0.10.',
2349
DeprecationWarning, stacklevel=2)
2350
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2351
self.new_inventory.revision_id = self._new_revision_id
2352
self.inv_sha1 = self.repository.add_inventory(
2353
self._new_revision_id,
2354
self.new_inventory,
2355
self.parents
2356
)
2357
2358
def _gen_revision_id(self):
2359
"""Return new revision-id."""
2360
s = '%s-%s-' % (self._config.user_email(),
2361
compact_date(self._timestamp))
2362
s += hexlify(rand_bytes(8))
2363
return s
2364
2365
def _generate_revision_if_needed(self):
2366
"""Create a revision id if None was supplied.
2367
2368
If the repository can not support user-specified revision ids
2369
they should override this function and raise UnsupportedOperation
2370
if _new_revision_id is not None.
2371
2372
:raises: UnsupportedOperation
2373
"""
2374
if self._new_revision_id is None:
2375
self._new_revision_id = self._gen_revision_id()
2376
2377
def record_entry_contents(self, ie, parent_invs, path, tree):
2378
"""Record the content of ie from tree into the commit if needed.
2379
2380
Side effect: sets ie.revision when unchanged
2381
2382
:param ie: An inventory entry present in the commit.
2383
:param parent_invs: The inventories of the parent revisions of the
2384
commit.
2385
:param path: The path the entry is at in the tree.
2386
:param tree: The tree which contains this entry and should be used to
2387
obtain content.
2388
"""
2389
if self.new_inventory.root is None and ie.parent_id is not None:
2390
symbol_versioning.warn('Root entry should be supplied to'
2391
' record_entry_contents, as of bzr 0.10.',
2392
DeprecationWarning, stacklevel=2)
2393
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2394
'', tree)
2395
self.new_inventory.add(ie)
2396
2397
# ie.revision is always None if the InventoryEntry is considered
2398
# for committing. ie.snapshot will record the correct revision
2399
# which may be the sole parent if it is untouched.
2400
if ie.revision is not None:
2401
return
2402
2403
# In this revision format, root entries have no knit or weave
2404
if ie is self.new_inventory.root:
2405
# When serializing out to disk and back in
2406
# root.revision is always _new_revision_id
2407
ie.revision = self._new_revision_id
2408
return
2409
previous_entries = ie.find_previous_heads(
2410
parent_invs,
2411
self.repository.weave_store,
2412
self.repository.get_transaction())
2413
# we are creating a new revision for ie in the history store
2414
# and inventory.
2415
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2416
2417
def modified_directory(self, file_id, file_parents):
2418
"""Record the presence of a symbolic link.
2419
2420
:param file_id: The file_id of the link to record.
2421
:param file_parents: The per-file parent revision ids.
2422
"""
2423
self._add_text_to_weave(file_id, [], file_parents.keys())
2424
2425
def modified_file_text(self, file_id, file_parents,
2426
get_content_byte_lines, text_sha1=None,
2427
text_size=None):
2428
"""Record the text of file file_id
2429
2430
:param file_id: The file_id of the file to record the text of.
2431
:param file_parents: The per-file parent revision ids.
2432
:param get_content_byte_lines: A callable which will return the byte
2433
lines for the file.
2434
:param text_sha1: Optional SHA1 of the file contents.
2435
:param text_size: Optional size of the file contents.
2436
"""
2437
# mutter('storing text of file {%s} in revision {%s} into %r',
2438
# file_id, self._new_revision_id, self.repository.weave_store)
2439
# special case to avoid diffing on renames or
2440
# reparenting
2441
if (len(file_parents) == 1
2442
and text_sha1 == file_parents.values()[0].text_sha1
2443
and text_size == file_parents.values()[0].text_size):
2444
previous_ie = file_parents.values()[0]
2445
versionedfile = self.repository.weave_store.get_weave(file_id,
2446
self.repository.get_transaction())
2447
versionedfile.clone_text(self._new_revision_id,
2448
previous_ie.revision, file_parents.keys())
2449
return text_sha1, text_size
2450
else:
2451
new_lines = get_content_byte_lines()
2452
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2453
# should return the SHA1 and size
2454
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2455
return osutils.sha_strings(new_lines), \
2456
sum(map(len, new_lines))
2457
2458
def modified_link(self, file_id, file_parents, link_target):
2459
"""Record the presence of a symbolic link.
2460
2461
:param file_id: The file_id of the link to record.
2462
:param file_parents: The per-file parent revision ids.
2463
:param link_target: Target location of this link.
2464
"""
2465
self._add_text_to_weave(file_id, [], file_parents.keys())
2466
2467
def _add_text_to_weave(self, file_id, new_lines, parents):
2468
versionedfile = self.repository.weave_store.get_weave_or_empty(
2469
file_id, self.repository.get_transaction())
2470
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2471
versionedfile.clear_cache()
2472
2473
2474
class _CommitBuilder(CommitBuilder):
2475
"""Temporary class so old CommitBuilders are detected properly
2476
2477
Note: CommitBuilder works whether or not root entry is recorded.
2478
"""
2479
2480
record_root_entry = True
2481
2482
2483
class RootCommitBuilder(CommitBuilder):
2484
"""This commitbuilder actually records the root id"""
2485
2486
record_root_entry = True
2487
2488
def record_entry_contents(self, ie, parent_invs, path, tree):
2489
"""Record the content of ie from tree into the commit if needed.
2490
2491
Side effect: sets ie.revision when unchanged
2492
2493
:param ie: An inventory entry present in the commit.
2494
:param parent_invs: The inventories of the parent revisions of the
2495
commit.
2496
:param path: The path the entry is at in the tree.
2497
:param tree: The tree which contains this entry and should be used to
2498
obtain content.
2499
"""
2500
assert self.new_inventory.root is not None or ie.parent_id is None
2501
self.new_inventory.add(ie)
2502
2503
# ie.revision is always None if the InventoryEntry is considered
2504
# for committing. ie.snapshot will record the correct revision
2505
# which may be the sole parent if it is untouched.
2506
if ie.revision is not None:
2507
return
2508
2509
previous_entries = ie.find_previous_heads(
2510
parent_invs,
2511
self.repository.weave_store,
2512
self.repository.get_transaction())
2513
# we are creating a new revision for ie in the history store
2514
# and inventory.
2515
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2516
2517
2518
_unescape_map = {
2519
'apos':"'",
2520
'quot':'"',
2521
'amp':'&',
2522
'lt':'<',
2523
'gt':'>'
2524
}
2525
2526
2527
def _unescaper(match, _map=_unescape_map):
2528
return _map[match.group(1)]
2529
2530
2531
_unescape_re = None
2532
2533
2534
def _unescape_xml(data):
2535
"""Unescape predefined XML entities in a string of data."""
2536
global _unescape_re
2537
if _unescape_re is None:
2538
_unescape_re = re.compile('\&([^;]*);')
2539
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1