(root)/brz/remove-bazaar
» Revision
2229.2.1
(compared to revision 0.200.1036)
: bzrlib/repository.py
To get this branch, use:
bzr branch
http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar
« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 2229.2.1
- compare with another revision
- download diff
- download tarball
- view history from revision 2229.2.1
- Committer: Aaron Bentley
- Date: 2007-01-11 17:20:25 UTC
- mto: This revision was merged to the branch mainline in revision 2239.
- Revision ID: abentley@panoramicfeedback.com-20070111172025-8rnvvm3jqx7h5uvv
Reject reserved ids in versiondfile, tree, branch and repository
- files added:
- .bzrignore
- bzrlib
- bzrlib/benchmarks
- bzrlib/benchmarks/tree_creator
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/doc
- bzrlib/doc/api
- bzrlib/export
- bzrlib/plugins
- bzrlib/plugins/launchpad
- bzrlib/store
- bzrlib/store/revision
- bzrlib/store/versioned
- bzrlib/tests
- bzrlib/tests/blackbox
- bzrlib/tests/branch_implementations
- bzrlib/tests/bzrdir_implementations
- bzrlib/tests/interrepository_implementations
- bzrlib/tests/intertree_implementations
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/repository_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/tests/test_patches_data
- bzrlib/tests/tree_implementations
- bzrlib/tests/workingtree_implementations
- bzrlib/transport
- bzrlib/transport/http
- bzrlib/ui
- bzrlib/util
- bzrlib/util/configobj
- bzrlib/util/configobj/docs
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- bzrlib/util/elementtree
- bzrlib/version_info_formats
- contrib
- contrib/bash
- contrib/emacs
- contrib/zsh
- doc
- tools
- tools/doc_generate
- tools/win32
- files removed:
- .bzrignore
- notes
- tests
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
#
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
7
#
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
12
#
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
17
from cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
23
import re
24
import time
25
import unittest
26
27
from bzrlib import (
28
bzrdir,
29
check,
30
delta,
31
errors,
32
generate_ids,
33
gpg,
34
graph,
35
knit,
36
lazy_regex,
37
lockable_files,
38
lockdir,
39
osutils,
40
revision as _mod_revision,
41
symbol_versioning,
42
transactions,
43
ui,
44
weave,
45
weavefile,
46
xml5,
47
xml6,
48
)
49
from bzrlib.osutils import (
50
rand_bytes,
51
compact_date,
52
local_time_offset,
53
)
54
from bzrlib.revisiontree import RevisionTree
55
from bzrlib.store.versioned import VersionedFileStore
56
from bzrlib.store.text import TextStore
57
from bzrlib.testament import Testament
58
""")
59
60
from bzrlib.decorators import needs_read_lock, needs_write_lock
61
from bzrlib.inter import InterObject
62
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
63
from bzrlib.symbol_versioning import (
64
deprecated_method,
65
zero_nine,
66
)
67
from bzrlib.trace import mutter, note, warning
68
69
70
# Old formats display a warning, but only once
71
_deprecation_warning_done = False
72
73
74
class Repository(object):
75
"""Repository holding history for one or more branches.
76
77
The repository holds and retrieves historical information including
78
revisions and file history. It's normally accessed only by the Branch,
79
which views a particular line of development through that history.
80
81
The Repository builds on top of Stores and a Transport, which respectively
82
describe the disk data format and the way of accessing the (possibly
83
remote) disk.
84
"""
85
86
_file_ids_altered_regex = lazy_regex.lazy_compile(
87
r'file_id="(?P<file_id>[^"]+)"'
88
r'.*revision="(?P<revision_id>[^"]+)"'
89
)
90
91
@needs_write_lock
92
def add_inventory(self, revid, inv, parents):
93
"""Add the inventory inv to the repository as revid.
94
95
:param parents: The revision ids of the parents that revid
96
is known to have and are in the repository already.
97
98
returns the sha1 of the serialized inventory.
99
"""
100
if _mod_revision.reserved_id(revid):
101
raise errors.ReservedId(revid)
102
assert inv.revision_id is None or inv.revision_id == revid, \
103
"Mismatch between inventory revision" \
104
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
105
assert inv.root is not None
106
inv_text = self.serialise_inventory(inv)
107
inv_sha1 = osutils.sha_string(inv_text)
108
inv_vf = self.control_weaves.get_weave('inventory',
109
self.get_transaction())
110
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
111
return inv_sha1
112
113
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
114
final_parents = []
115
for parent in parents:
116
if parent in inv_vf:
117
final_parents.append(parent)
118
119
inv_vf.add_lines(revid, final_parents, lines)
120
121
@needs_write_lock
122
def add_revision(self, rev_id, rev, inv=None, config=None):
123
"""Add rev to the revision store as rev_id.
124
125
:param rev_id: the revision id to use.
126
:param rev: The revision object.
127
:param inv: The inventory for the revision. if None, it will be looked
128
up in the inventory storer
129
:param config: If None no digital signature will be created.
130
If supplied its signature_needed method will be used
131
to determine if a signature should be made.
132
"""
133
if _mod_revision.reserved_id(rev_id):
134
raise errors.ReservedId(rev_id)
135
if config is not None and config.signature_needed():
136
if inv is None:
137
inv = self.get_inventory(rev_id)
138
plaintext = Testament(rev, inv).as_short_text()
139
self.store_revision_signature(
140
gpg.GPGStrategy(config), plaintext, rev_id)
141
if not rev_id in self.get_inventory_weave():
142
if inv is None:
143
raise errors.WeaveRevisionNotPresent(rev_id,
144
self.get_inventory_weave())
145
else:
146
# yes, this is not suitable for adding with ghosts.
147
self.add_inventory(rev_id, inv, rev.parent_ids)
148
self._revision_store.add_revision(rev, self.get_transaction())
149
150
@needs_read_lock
151
def _all_possible_ids(self):
152
"""Return all the possible revisions that we could find."""
153
return self.get_inventory_weave().versions()
154
155
def all_revision_ids(self):
156
"""Returns a list of all the revision ids in the repository.
157
158
This is deprecated because code should generally work on the graph
159
reachable from a particular revision, and ignore any other revisions
160
that might be present. There is no direct replacement method.
161
"""
162
return self._all_revision_ids()
163
164
@needs_read_lock
165
def _all_revision_ids(self):
166
"""Returns a list of all the revision ids in the repository.
167
168
These are in as much topological order as the underlying store can
169
present: for weaves ghosts may lead to a lack of correctness until
170
the reweave updates the parents list.
171
"""
172
if self._revision_store.text_store.listable():
173
return self._revision_store.all_revision_ids(self.get_transaction())
174
result = self._all_possible_ids()
175
return self._eliminate_revisions_not_present(result)
176
177
def break_lock(self):
178
"""Break a lock if one is present from another instance.
179
180
Uses the ui factory to ask for confirmation if the lock may be from
181
an active process.
182
"""
183
self.control_files.break_lock()
184
185
@needs_read_lock
186
def _eliminate_revisions_not_present(self, revision_ids):
187
"""Check every revision id in revision_ids to see if we have it.
188
189
Returns a set of the present revisions.
190
"""
191
result = []
192
for id in revision_ids:
193
if self.has_revision(id):
194
result.append(id)
195
return result
196
197
@staticmethod
198
def create(a_bzrdir):
199
"""Construct the current default format repository in a_bzrdir."""
200
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
201
202
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
203
"""instantiate a Repository.
204
205
:param _format: The format of the repository on disk.
206
:param a_bzrdir: The BzrDir of the repository.
207
208
In the future we will have a single api for all stores for
209
getting file texts, inventories and revisions, then
210
this construct will accept instances of those things.
211
"""
212
super(Repository, self).__init__()
213
self._format = _format
214
# the following are part of the public API for Repository:
215
self.bzrdir = a_bzrdir
216
self.control_files = control_files
217
self._revision_store = _revision_store
218
self.text_store = text_store
219
# backwards compatibility
220
self.weave_store = text_store
221
# not right yet - should be more semantically clear ?
222
#
223
self.control_store = control_store
224
self.control_weaves = control_store
225
# TODO: make sure to construct the right store classes, etc, depending
226
# on whether escaping is required.
227
self._warn_if_deprecated()
228
self._serializer = xml5.serializer_v5
229
230
def __repr__(self):
231
return '%s(%r)' % (self.__class__.__name__,
232
self.bzrdir.transport.base)
233
234
def is_locked(self):
235
return self.control_files.is_locked()
236
237
def lock_write(self):
238
self.control_files.lock_write()
239
240
def lock_read(self):
241
self.control_files.lock_read()
242
243
def get_physical_lock_status(self):
244
return self.control_files.get_physical_lock_status()
245
246
@needs_read_lock
247
def missing_revision_ids(self, other, revision_id=None):
248
"""Return the revision ids that other has that this does not.
249
250
These are returned in topological order.
251
252
revision_id: only return revision ids included by revision_id.
253
"""
254
return InterRepository.get(other, self).missing_revision_ids(revision_id)
255
256
@staticmethod
257
def open(base):
258
"""Open the repository rooted at base.
259
260
For instance, if the repository is at URL/.bzr/repository,
261
Repository.open(URL) -> a Repository instance.
262
"""
263
control = bzrdir.BzrDir.open(base)
264
return control.open_repository()
265
266
def copy_content_into(self, destination, revision_id=None, basis=None):
267
"""Make a complete copy of the content in self into destination.
268
269
This is a destructive operation! Do not use it on existing
270
repositories.
271
"""
272
return InterRepository.get(self, destination).copy_content(revision_id, basis)
273
274
def fetch(self, source, revision_id=None, pb=None):
275
"""Fetch the content required to construct revision_id from source.
276
277
If revision_id is None all content is copied.
278
"""
279
return InterRepository.get(source, self).fetch(revision_id=revision_id,
280
pb=pb)
281
282
def get_commit_builder(self, branch, parents, config, timestamp=None,
283
timezone=None, committer=None, revprops=None,
284
revision_id=None):
285
"""Obtain a CommitBuilder for this repository.
286
287
:param branch: Branch to commit to.
288
:param parents: Revision ids of the parents of the new revision.
289
:param config: Configuration to use.
290
:param timestamp: Optional timestamp recorded for commit.
291
:param timezone: Optional timezone for timestamp.
292
:param committer: Optional committer to set for commit.
293
:param revprops: Optional dictionary of revision properties.
294
:param revision_id: Optional revision id.
295
"""
296
return _CommitBuilder(self, parents, config, timestamp, timezone,
297
committer, revprops, revision_id)
298
299
def unlock(self):
300
self.control_files.unlock()
301
302
@needs_read_lock
303
def clone(self, a_bzrdir, revision_id=None, basis=None):
304
"""Clone this repository into a_bzrdir using the current format.
305
306
Currently no check is made that the format of this repository and
307
the bzrdir format are compatible. FIXME RBC 20060201.
308
"""
309
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
310
# use target default format.
311
result = a_bzrdir.create_repository()
312
# FIXME RBC 20060209 split out the repository type to avoid this check ?
313
elif isinstance(a_bzrdir._format,
314
(bzrdir.BzrDirFormat4,
315
bzrdir.BzrDirFormat5,
316
bzrdir.BzrDirFormat6)):
317
result = a_bzrdir.open_repository()
318
else:
319
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
320
self.copy_content_into(result, revision_id, basis)
321
return result
322
323
@needs_read_lock
324
def has_revision(self, revision_id):
325
"""True if this repository has a copy of the revision."""
326
return self._revision_store.has_revision_id(revision_id,
327
self.get_transaction())
328
329
@needs_read_lock
330
def get_revision_reconcile(self, revision_id):
331
"""'reconcile' helper routine that allows access to a revision always.
332
333
This variant of get_revision does not cross check the weave graph
334
against the revision one as get_revision does: but it should only
335
be used by reconcile, or reconcile-alike commands that are correcting
336
or testing the revision graph.
337
"""
338
if not revision_id or not isinstance(revision_id, basestring):
339
raise errors.InvalidRevisionId(revision_id=revision_id,
340
branch=self)
341
return self._revision_store.get_revisions([revision_id],
342
self.get_transaction())[0]
343
@needs_read_lock
344
def get_revisions(self, revision_ids):
345
return self._revision_store.get_revisions(revision_ids,
346
self.get_transaction())
347
348
@needs_read_lock
349
def get_revision_xml(self, revision_id):
350
rev = self.get_revision(revision_id)
351
rev_tmp = StringIO()
352
# the current serializer..
353
self._revision_store._serializer.write_revision(rev, rev_tmp)
354
rev_tmp.seek(0)
355
return rev_tmp.getvalue()
356
357
@needs_read_lock
358
def get_revision(self, revision_id):
359
"""Return the Revision object for a named revision"""
360
r = self.get_revision_reconcile(revision_id)
361
# weave corruption can lead to absent revision markers that should be
362
# present.
363
# the following test is reasonably cheap (it needs a single weave read)
364
# and the weave is cached in read transactions. In write transactions
365
# it is not cached but typically we only read a small number of
366
# revisions. For knits when they are introduced we will probably want
367
# to ensure that caching write transactions are in use.
368
inv = self.get_inventory_weave()
369
self._check_revision_parents(r, inv)
370
return r
371
372
@needs_read_lock
373
def get_deltas_for_revisions(self, revisions):
374
"""Produce a generator of revision deltas.
375
376
Note that the input is a sequence of REVISIONS, not revision_ids.
377
Trees will be held in memory until the generator exits.
378
Each delta is relative to the revision's lefthand predecessor.
379
"""
380
required_trees = set()
381
for revision in revisions:
382
required_trees.add(revision.revision_id)
383
required_trees.update(revision.parent_ids[:1])
384
trees = dict((t.get_revision_id(), t) for
385
t in self.revision_trees(required_trees))
386
for revision in revisions:
387
if not revision.parent_ids:
388
old_tree = self.revision_tree(None)
389
else:
390
old_tree = trees[revision.parent_ids[0]]
391
yield trees[revision.revision_id].changes_from(old_tree)
392
393
@needs_read_lock
394
def get_revision_delta(self, revision_id):
395
"""Return the delta for one revision.
396
397
The delta is relative to the left-hand predecessor of the
398
revision.
399
"""
400
r = self.get_revision(revision_id)
401
return list(self.get_deltas_for_revisions([r]))[0]
402
403
def _check_revision_parents(self, revision, inventory):
404
"""Private to Repository and Fetch.
405
406
This checks the parentage of revision in an inventory weave for
407
consistency and is only applicable to inventory-weave-for-ancestry
408
using repository formats & fetchers.
409
"""
410
weave_parents = inventory.get_parents(revision.revision_id)
411
weave_names = inventory.versions()
412
for parent_id in revision.parent_ids:
413
if parent_id in weave_names:
414
# this parent must not be a ghost.
415
if not parent_id in weave_parents:
416
# but it is a ghost
417
raise errors.CorruptRepository(self)
418
419
@needs_write_lock
420
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
421
signature = gpg_strategy.sign(plaintext)
422
self._revision_store.add_revision_signature_text(revision_id,
423
signature,
424
self.get_transaction())
425
426
def fileids_altered_by_revision_ids(self, revision_ids):
427
"""Find the file ids and versions affected by revisions.
428
429
:param revisions: an iterable containing revision ids.
430
:return: a dictionary mapping altered file-ids to an iterable of
431
revision_ids. Each altered file-ids has the exact revision_ids that
432
altered it listed explicitly.
433
"""
434
assert self._serializer.support_altered_by_hack, \
435
("fileids_altered_by_revision_ids only supported for branches "
436
"which store inventory as unnested xml, not on %r" % self)
437
selected_revision_ids = set(revision_ids)
438
w = self.get_inventory_weave()
439
result = {}
440
441
# this code needs to read every new line in every inventory for the
442
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
443
# not present in one of those inventories is unnecessary but not
444
# harmful because we are filtering by the revision id marker in the
445
# inventory lines : we only select file ids altered in one of those
446
# revisions. We don't need to see all lines in the inventory because
447
# only those added in an inventory in rev X can contain a revision=X
448
# line.
449
unescape_revid_cache = {}
450
unescape_fileid_cache = {}
451
452
# jam 20061218 In a big fetch, this handles hundreds of thousands
453
# of lines, so it has had a lot of inlining and optimizing done.
454
# Sorry that it is a little bit messy.
455
# Move several functions to be local variables, since this is a long
456
# running loop.
457
search = self._file_ids_altered_regex.search
458
unescape = _unescape_xml
459
setdefault = result.setdefault
460
pb = ui.ui_factory.nested_progress_bar()
461
try:
462
for line in w.iter_lines_added_or_present_in_versions(
463
selected_revision_ids, pb=pb):
464
match = search(line)
465
if match is None:
466
continue
467
# One call to match.group() returning multiple items is quite a
468
# bit faster than 2 calls to match.group() each returning 1
469
file_id, revision_id = match.group('file_id', 'revision_id')
470
471
# Inlining the cache lookups helps a lot when you make 170,000
472
# lines and 350k ids, versus 8.4 unique ids.
473
# Using a cache helps in 2 ways:
474
# 1) Avoids unnecessary decoding calls
475
# 2) Re-uses cached strings, which helps in future set and
476
# equality checks.
477
# (2) is enough that removing encoding entirely along with
478
# the cache (so we are using plain strings) results in no
479
# performance improvement.
480
try:
481
revision_id = unescape_revid_cache[revision_id]
482
except KeyError:
483
unescaped = unescape(revision_id)
484
unescape_revid_cache[revision_id] = unescaped
485
revision_id = unescaped
486
487
if revision_id in selected_revision_ids:
488
try:
489
file_id = unescape_fileid_cache[file_id]
490
except KeyError:
491
unescaped = unescape(file_id)
492
unescape_fileid_cache[file_id] = unescaped
493
file_id = unescaped
494
setdefault(file_id, set()).add(revision_id)
495
finally:
496
pb.finished()
497
return result
498
499
@needs_read_lock
500
def get_inventory_weave(self):
501
return self.control_weaves.get_weave('inventory',
502
self.get_transaction())
503
504
@needs_read_lock
505
def get_inventory(self, revision_id):
506
"""Get Inventory object by hash."""
507
return self.deserialise_inventory(
508
revision_id, self.get_inventory_xml(revision_id))
509
510
def deserialise_inventory(self, revision_id, xml):
511
"""Transform the xml into an inventory object.
512
513
:param revision_id: The expected revision id of the inventory.
514
:param xml: A serialised inventory.
515
"""
516
result = self._serializer.read_inventory_from_string(xml)
517
result.root.revision = revision_id
518
return result
519
520
def serialise_inventory(self, inv):
521
return self._serializer.write_inventory_to_string(inv)
522
523
@needs_read_lock
524
def get_inventory_xml(self, revision_id):
525
"""Get inventory XML as a file object."""
526
try:
527
assert isinstance(revision_id, basestring), type(revision_id)
528
iw = self.get_inventory_weave()
529
return iw.get_text(revision_id)
530
except IndexError:
531
raise errors.HistoryMissing(self, 'inventory', revision_id)
532
533
@needs_read_lock
534
def get_inventory_sha1(self, revision_id):
535
"""Return the sha1 hash of the inventory entry
536
"""
537
return self.get_revision(revision_id).inventory_sha1
538
539
@needs_read_lock
540
def get_revision_graph(self, revision_id=None):
541
"""Return a dictionary containing the revision graph.
542
543
:param revision_id: The revision_id to get a graph from. If None, then
544
the entire revision graph is returned. This is a deprecated mode of
545
operation and will be removed in the future.
546
:return: a dictionary of revision_id->revision_parents_list.
547
"""
548
# special case NULL_REVISION
549
if revision_id == _mod_revision.NULL_REVISION:
550
return {}
551
a_weave = self.get_inventory_weave()
552
all_revisions = self._eliminate_revisions_not_present(
553
a_weave.versions())
554
entire_graph = dict([(node, a_weave.get_parents(node)) for
555
node in all_revisions])
556
if revision_id is None:
557
return entire_graph
558
elif revision_id not in entire_graph:
559
raise errors.NoSuchRevision(self, revision_id)
560
else:
561
# add what can be reached from revision_id
562
result = {}
563
pending = set([revision_id])
564
while len(pending) > 0:
565
node = pending.pop()
566
result[node] = entire_graph[node]
567
for revision_id in result[node]:
568
if revision_id not in result:
569
pending.add(revision_id)
570
return result
571
572
@needs_read_lock
573
def get_revision_graph_with_ghosts(self, revision_ids=None):
574
"""Return a graph of the revisions with ghosts marked as applicable.
575
576
:param revision_ids: an iterable of revisions to graph or None for all.
577
:return: a Graph object with the graph reachable from revision_ids.
578
"""
579
result = graph.Graph()
580
if not revision_ids:
581
pending = set(self.all_revision_ids())
582
required = set([])
583
else:
584
pending = set(revision_ids)
585
# special case NULL_REVISION
586
if _mod_revision.NULL_REVISION in pending:
587
pending.remove(_mod_revision.NULL_REVISION)
588
required = set(pending)
589
done = set([])
590
while len(pending):
591
revision_id = pending.pop()
592
try:
593
rev = self.get_revision(revision_id)
594
except errors.NoSuchRevision:
595
if revision_id in required:
596
raise
597
# a ghost
598
result.add_ghost(revision_id)
599
continue
600
for parent_id in rev.parent_ids:
601
# is this queued or done ?
602
if (parent_id not in pending and
603
parent_id not in done):
604
# no, queue it.
605
pending.add(parent_id)
606
result.add_node(revision_id, rev.parent_ids)
607
done.add(revision_id)
608
return result
609
610
@needs_read_lock
611
def get_revision_inventory(self, revision_id):
612
"""Return inventory of a past revision."""
613
# TODO: Unify this with get_inventory()
614
# bzr 0.0.6 and later imposes the constraint that the inventory_id
615
# must be the same as its revision, so this is trivial.
616
if revision_id is None:
617
# This does not make sense: if there is no revision,
618
# then it is the current tree inventory surely ?!
619
# and thus get_root_id() is something that looks at the last
620
# commit on the branch, and the get_root_id is an inventory check.
621
raise NotImplementedError
622
# return Inventory(self.get_root_id())
623
else:
624
return self.get_inventory(revision_id)
625
626
@needs_read_lock
627
def is_shared(self):
628
"""Return True if this repository is flagged as a shared repository."""
629
raise NotImplementedError(self.is_shared)
630
631
@needs_write_lock
632
def reconcile(self, other=None, thorough=False):
633
"""Reconcile this repository."""
634
from bzrlib.reconcile import RepoReconciler
635
reconciler = RepoReconciler(self, thorough=thorough)
636
reconciler.reconcile()
637
return reconciler
638
639
@needs_read_lock
640
def revision_tree(self, revision_id):
641
"""Return Tree for a revision on this branch.
642
643
`revision_id` may be None for the empty tree revision.
644
"""
645
# TODO: refactor this to use an existing revision object
646
# so we don't need to read it in twice.
647
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
648
return RevisionTree(self, Inventory(root_id=None),
649
_mod_revision.NULL_REVISION)
650
else:
651
inv = self.get_revision_inventory(revision_id)
652
return RevisionTree(self, inv, revision_id)
653
654
@needs_read_lock
655
def revision_trees(self, revision_ids):
656
"""Return Tree for a revision on this branch.
657
658
`revision_id` may not be None or 'null:'"""
659
assert None not in revision_ids
660
assert _mod_revision.NULL_REVISION not in revision_ids
661
texts = self.get_inventory_weave().get_texts(revision_ids)
662
for text, revision_id in zip(texts, revision_ids):
663
inv = self.deserialise_inventory(revision_id, text)
664
yield RevisionTree(self, inv, revision_id)
665
666
@needs_read_lock
667
def get_ancestry(self, revision_id):
668
"""Return a list of revision-ids integrated by a revision.
669
670
The first element of the list is always None, indicating the origin
671
revision. This might change when we have history horizons, or
672
perhaps we should have a new API.
673
674
This is topologically sorted.
675
"""
676
if revision_id is None:
677
return [None]
678
if not self.has_revision(revision_id):
679
raise errors.NoSuchRevision(self, revision_id)
680
w = self.get_inventory_weave()
681
candidates = w.get_ancestry(revision_id)
682
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
683
684
@needs_read_lock
685
def print_file(self, file, revision_id):
686
"""Print `file` to stdout.
687
688
FIXME RBC 20060125 as John Meinel points out this is a bad api
689
- it writes to stdout, it assumes that that is valid etc. Fix
690
by creating a new more flexible convenience function.
691
"""
692
tree = self.revision_tree(revision_id)
693
# use inventory as it was in that revision
694
file_id = tree.inventory.path2id(file)
695
if not file_id:
696
# TODO: jam 20060427 Write a test for this code path
697
# it had a bug in it, and was raising the wrong
698
# exception.
699
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
700
tree.print_file(file_id)
701
702
def get_transaction(self):
703
return self.control_files.get_transaction()
704
705
def revision_parents(self, revid):
706
return self.get_inventory_weave().parent_names(revid)
707
708
@needs_write_lock
709
def set_make_working_trees(self, new_value):
710
"""Set the policy flag for making working trees when creating branches.
711
712
This only applies to branches that use this repository.
713
714
The default is 'True'.
715
:param new_value: True to restore the default, False to disable making
716
working trees.
717
"""
718
raise NotImplementedError(self.set_make_working_trees)
719
720
def make_working_trees(self):
721
"""Returns the policy for making working trees on new branches."""
722
raise NotImplementedError(self.make_working_trees)
723
724
@needs_write_lock
725
def sign_revision(self, revision_id, gpg_strategy):
726
plaintext = Testament.from_revision(self, revision_id).as_short_text()
727
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
728
729
@needs_read_lock
730
def has_signature_for_revision_id(self, revision_id):
731
"""Query for a revision signature for revision_id in the repository."""
732
return self._revision_store.has_signature(revision_id,
733
self.get_transaction())
734
735
@needs_read_lock
736
def get_signature_text(self, revision_id):
737
"""Return the text for a signature."""
738
return self._revision_store.get_signature_text(revision_id,
739
self.get_transaction())
740
741
@needs_read_lock
742
def check(self, revision_ids):
743
"""Check consistency of all history of given revision_ids.
744
745
Different repository implementations should override _check().
746
747
:param revision_ids: A non-empty list of revision_ids whose ancestry
748
will be checked. Typically the last revision_id of a branch.
749
"""
750
if not revision_ids:
751
raise ValueError("revision_ids must be non-empty in %s.check"
752
% (self,))
753
return self._check(revision_ids)
754
755
def _check(self, revision_ids):
756
result = check.Check(self)
757
result.check()
758
return result
759
760
def _warn_if_deprecated(self):
761
global _deprecation_warning_done
762
if _deprecation_warning_done:
763
return
764
_deprecation_warning_done = True
765
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
766
% (self._format, self.bzrdir.transport.base))
767
768
def supports_rich_root(self):
769
return self._format.rich_root_data
770
771
def _check_ascii_revisionid(self, revision_id, method):
772
"""Private helper for ascii-only repositories."""
773
# weave repositories refuse to store revisionids that are non-ascii.
774
if revision_id is not None:
775
# weaves require ascii revision ids.
776
if isinstance(revision_id, unicode):
777
try:
778
revision_id.encode('ascii')
779
except UnicodeEncodeError:
780
raise errors.NonAsciiRevisionId(method, self)
781
782
783
class AllInOneRepository(Repository):
784
"""Legacy support - the repository behaviour for all-in-one branches."""
785
786
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
787
# we reuse one control files instance.
788
dir_mode = a_bzrdir._control_files._dir_mode
789
file_mode = a_bzrdir._control_files._file_mode
790
791
def get_store(name, compressed=True, prefixed=False):
792
# FIXME: This approach of assuming stores are all entirely compressed
793
# or entirely uncompressed is tidy, but breaks upgrade from
794
# some existing branches where there's a mixture; we probably
795
# still want the option to look for both.
796
relpath = a_bzrdir._control_files._escape(name)
797
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
798
prefixed=prefixed, compressed=compressed,
799
dir_mode=dir_mode,
800
file_mode=file_mode)
801
#if self._transport.should_cache():
802
# cache_path = os.path.join(self.cache_root, name)
803
# os.mkdir(cache_path)
804
# store = bzrlib.store.CachedStore(store, cache_path)
805
return store
806
807
# not broken out yet because the controlweaves|inventory_store
808
# and text_store | weave_store bits are still different.
809
if isinstance(_format, RepositoryFormat4):
810
# cannot remove these - there is still no consistent api
811
# which allows access to this old info.
812
self.inventory_store = get_store('inventory-store')
813
text_store = get_store('text-store')
814
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
815
816
def get_commit_builder(self, branch, parents, config, timestamp=None,
817
timezone=None, committer=None, revprops=None,
818
revision_id=None):
819
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
820
return Repository.get_commit_builder(self, branch, parents, config,
821
timestamp, timezone, committer, revprops, revision_id)
822
823
@needs_read_lock
824
def is_shared(self):
825
"""AllInOne repositories cannot be shared."""
826
return False
827
828
@needs_write_lock
829
def set_make_working_trees(self, new_value):
830
"""Set the policy flag for making working trees when creating branches.
831
832
This only applies to branches that use this repository.
833
834
The default is 'True'.
835
:param new_value: True to restore the default, False to disable making
836
working trees.
837
"""
838
raise NotImplementedError(self.set_make_working_trees)
839
840
def make_working_trees(self):
841
"""Returns the policy for making working trees on new branches."""
842
return True
843
844
845
def install_revision(repository, rev, revision_tree):
846
"""Install all revision data into a repository."""
847
present_parents = []
848
parent_trees = {}
849
for p_id in rev.parent_ids:
850
if repository.has_revision(p_id):
851
present_parents.append(p_id)
852
parent_trees[p_id] = repository.revision_tree(p_id)
853
else:
854
parent_trees[p_id] = repository.revision_tree(None)
855
856
inv = revision_tree.inventory
857
entries = inv.iter_entries()
858
# backwards compatability hack: skip the root id.
859
if not repository.supports_rich_root():
860
path, root = entries.next()
861
if root.revision != rev.revision_id:
862
raise errors.IncompatibleRevision(repr(repository))
863
# Add the texts that are not already present
864
for path, ie in entries:
865
w = repository.weave_store.get_weave_or_empty(ie.file_id,
866
repository.get_transaction())
867
if ie.revision not in w:
868
text_parents = []
869
# FIXME: TODO: The following loop *may* be overlapping/duplicate
870
# with InventoryEntry.find_previous_heads(). if it is, then there
871
# is a latent bug here where the parents may have ancestors of each
872
# other. RBC, AB
873
for revision, tree in parent_trees.iteritems():
874
if ie.file_id not in tree:
875
continue
876
parent_id = tree.inventory[ie.file_id].revision
877
if parent_id in text_parents:
878
continue
879
text_parents.append(parent_id)
880
881
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
882
repository.get_transaction())
883
lines = revision_tree.get_file(ie.file_id).readlines()
884
vfile.add_lines(rev.revision_id, text_parents, lines)
885
try:
886
# install the inventory
887
repository.add_inventory(rev.revision_id, inv, present_parents)
888
except errors.RevisionAlreadyPresent:
889
pass
890
repository.add_revision(rev.revision_id, rev, inv)
891
892
893
class MetaDirRepository(Repository):
894
"""Repositories in the new meta-dir layout."""
895
896
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
897
super(MetaDirRepository, self).__init__(_format,
898
a_bzrdir,
899
control_files,
900
_revision_store,
901
control_store,
902
text_store)
903
dir_mode = self.control_files._dir_mode
904
file_mode = self.control_files._file_mode
905
906
@needs_read_lock
907
def is_shared(self):
908
"""Return True if this repository is flagged as a shared repository."""
909
return self.control_files._transport.has('shared-storage')
910
911
@needs_write_lock
912
def set_make_working_trees(self, new_value):
913
"""Set the policy flag for making working trees when creating branches.
914
915
This only applies to branches that use this repository.
916
917
The default is 'True'.
918
:param new_value: True to restore the default, False to disable making
919
working trees.
920
"""
921
if new_value:
922
try:
923
self.control_files._transport.delete('no-working-trees')
924
except errors.NoSuchFile:
925
pass
926
else:
927
self.control_files.put_utf8('no-working-trees', '')
928
929
def make_working_trees(self):
930
"""Returns the policy for making working trees on new branches."""
931
return not self.control_files._transport.has('no-working-trees')
932
933
934
class WeaveMetaDirRepository(MetaDirRepository):
935
"""A subclass of MetaDirRepository to set weave specific policy."""
936
937
def get_commit_builder(self, branch, parents, config, timestamp=None,
938
timezone=None, committer=None, revprops=None,
939
revision_id=None):
940
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
941
return MetaDirRepository.get_commit_builder(self, branch, parents,
942
config, timestamp, timezone, committer, revprops, revision_id)
943
944
945
class KnitRepository(MetaDirRepository):
946
"""Knit format repository."""
947
948
def _warn_if_deprecated(self):
949
# This class isn't deprecated
950
pass
951
952
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
953
inv_vf.add_lines_with_ghosts(revid, parents, lines)
954
955
@needs_read_lock
956
def _all_revision_ids(self):
957
"""See Repository.all_revision_ids()."""
958
# Knits get the revision graph from the index of the revision knit, so
959
# it's always possible even if they're on an unlistable transport.
960
return self._revision_store.all_revision_ids(self.get_transaction())
961
962
def fileid_involved_between_revs(self, from_revid, to_revid):
963
"""Find file_id(s) which are involved in the changes between revisions.
964
965
This determines the set of revisions which are involved, and then
966
finds all file ids affected by those revisions.
967
"""
968
vf = self._get_revision_vf()
969
from_set = set(vf.get_ancestry(from_revid))
970
to_set = set(vf.get_ancestry(to_revid))
971
changed = to_set.difference(from_set)
972
return self._fileid_involved_by_set(changed)
973
974
def fileid_involved(self, last_revid=None):
975
"""Find all file_ids modified in the ancestry of last_revid.
976
977
:param last_revid: If None, last_revision() will be used.
978
"""
979
if not last_revid:
980
changed = set(self.all_revision_ids())
981
else:
982
changed = set(self.get_ancestry(last_revid))
983
if None in changed:
984
changed.remove(None)
985
return self._fileid_involved_by_set(changed)
986
987
@needs_read_lock
988
def get_ancestry(self, revision_id):
989
"""Return a list of revision-ids integrated by a revision.
990
991
This is topologically sorted.
992
"""
993
if revision_id is None:
994
return [None]
995
vf = self._get_revision_vf()
996
try:
997
return [None] + vf.get_ancestry(revision_id)
998
except errors.RevisionNotPresent:
999
raise errors.NoSuchRevision(self, revision_id)
1000
1001
@needs_read_lock
1002
def get_revision(self, revision_id):
1003
"""Return the Revision object for a named revision"""
1004
return self.get_revision_reconcile(revision_id)
1005
1006
@needs_read_lock
1007
def get_revision_graph(self, revision_id=None):
1008
"""Return a dictionary containing the revision graph.
1009
1010
:param revision_id: The revision_id to get a graph from. If None, then
1011
the entire revision graph is returned. This is a deprecated mode of
1012
operation and will be removed in the future.
1013
:return: a dictionary of revision_id->revision_parents_list.
1014
"""
1015
# special case NULL_REVISION
1016
if revision_id == _mod_revision.NULL_REVISION:
1017
return {}
1018
a_weave = self._get_revision_vf()
1019
entire_graph = a_weave.get_graph()
1020
if revision_id is None:
1021
return a_weave.get_graph()
1022
elif revision_id not in a_weave:
1023
raise errors.NoSuchRevision(self, revision_id)
1024
else:
1025
# add what can be reached from revision_id
1026
result = {}
1027
pending = set([revision_id])
1028
while len(pending) > 0:
1029
node = pending.pop()
1030
result[node] = a_weave.get_parents(node)
1031
for revision_id in result[node]:
1032
if revision_id not in result:
1033
pending.add(revision_id)
1034
return result
1035
1036
@needs_read_lock
1037
def get_revision_graph_with_ghosts(self, revision_ids=None):
1038
"""Return a graph of the revisions with ghosts marked as applicable.
1039
1040
:param revision_ids: an iterable of revisions to graph or None for all.
1041
:return: a Graph object with the graph reachable from revision_ids.
1042
"""
1043
result = graph.Graph()
1044
vf = self._get_revision_vf()
1045
versions = set(vf.versions())
1046
if not revision_ids:
1047
pending = set(self.all_revision_ids())
1048
required = set([])
1049
else:
1050
pending = set(revision_ids)
1051
# special case NULL_REVISION
1052
if _mod_revision.NULL_REVISION in pending:
1053
pending.remove(_mod_revision.NULL_REVISION)
1054
required = set(pending)
1055
done = set([])
1056
while len(pending):
1057
revision_id = pending.pop()
1058
if not revision_id in versions:
1059
if revision_id in required:
1060
raise errors.NoSuchRevision(self, revision_id)
1061
# a ghost
1062
result.add_ghost(revision_id)
1063
# mark it as done so we don't try for it again.
1064
done.add(revision_id)
1065
continue
1066
parent_ids = vf.get_parents_with_ghosts(revision_id)
1067
for parent_id in parent_ids:
1068
# is this queued or done ?
1069
if (parent_id not in pending and
1070
parent_id not in done):
1071
# no, queue it.
1072
pending.add(parent_id)
1073
result.add_node(revision_id, parent_ids)
1074
done.add(revision_id)
1075
return result
1076
1077
def _get_revision_vf(self):
1078
""":return: a versioned file containing the revisions."""
1079
vf = self._revision_store.get_revision_file(self.get_transaction())
1080
return vf
1081
1082
@needs_write_lock
1083
def reconcile(self, other=None, thorough=False):
1084
"""Reconcile this repository."""
1085
from bzrlib.reconcile import KnitReconciler
1086
reconciler = KnitReconciler(self, thorough=thorough)
1087
reconciler.reconcile()
1088
return reconciler
1089
1090
def revision_parents(self, revision_id):
1091
return self._get_revision_vf().get_parents(revision_id)
1092
1093
1094
class KnitRepository2(KnitRepository):
1095
""""""
1096
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1097
control_store, text_store):
1098
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1099
_revision_store, control_store, text_store)
1100
self._serializer = xml6.serializer_v6
1101
1102
def deserialise_inventory(self, revision_id, xml):
1103
"""Transform the xml into an inventory object.
1104
1105
:param revision_id: The expected revision id of the inventory.
1106
:param xml: A serialised inventory.
1107
"""
1108
result = self._serializer.read_inventory_from_string(xml)
1109
assert result.root.revision is not None
1110
return result
1111
1112
def serialise_inventory(self, inv):
1113
"""Transform the inventory object into XML text.
1114
1115
:param revision_id: The expected revision id of the inventory.
1116
:param xml: A serialised inventory.
1117
"""
1118
assert inv.revision_id is not None
1119
assert inv.root.revision is not None
1120
return KnitRepository.serialise_inventory(self, inv)
1121
1122
def get_commit_builder(self, branch, parents, config, timestamp=None,
1123
timezone=None, committer=None, revprops=None,
1124
revision_id=None):
1125
"""Obtain a CommitBuilder for this repository.
1126
1127
:param branch: Branch to commit to.
1128
:param parents: Revision ids of the parents of the new revision.
1129
:param config: Configuration to use.
1130
:param timestamp: Optional timestamp recorded for commit.
1131
:param timezone: Optional timezone for timestamp.
1132
:param committer: Optional committer to set for commit.
1133
:param revprops: Optional dictionary of revision properties.
1134
:param revision_id: Optional revision id.
1135
"""
1136
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1137
committer, revprops, revision_id)
1138
1139
1140
class RepositoryFormat(object):
1141
"""A repository format.
1142
1143
Formats provide three things:
1144
* An initialization routine to construct repository data on disk.
1145
* a format string which is used when the BzrDir supports versioned
1146
children.
1147
* an open routine which returns a Repository instance.
1148
1149
Formats are placed in an dict by their format string for reference
1150
during opening. These should be subclasses of RepositoryFormat
1151
for consistency.
1152
1153
Once a format is deprecated, just deprecate the initialize and open
1154
methods on the format class. Do not deprecate the object, as the
1155
object will be created every system load.
1156
1157
Common instance attributes:
1158
_matchingbzrdir - the bzrdir format that the repository format was
1159
originally written to work with. This can be used if manually
1160
constructing a bzrdir and repository, or more commonly for test suite
1161
parameterisation.
1162
"""
1163
1164
_default_format = None
1165
"""The default format used for new repositories."""
1166
1167
_formats = {}
1168
"""The known formats."""
1169
1170
def __str__(self):
1171
return "<%s>" % self.__class__.__name__
1172
1173
@classmethod
1174
def find_format(klass, a_bzrdir):
1175
"""Return the format for the repository object in a_bzrdir."""
1176
try:
1177
transport = a_bzrdir.get_repository_transport(None)
1178
format_string = transport.get("format").read()
1179
return klass._formats[format_string]
1180
except errors.NoSuchFile:
1181
raise errors.NoRepositoryPresent(a_bzrdir)
1182
except KeyError:
1183
raise errors.UnknownFormatError(format=format_string)
1184
1185
def _get_control_store(self, repo_transport, control_files):
1186
"""Return the control store for this repository."""
1187
raise NotImplementedError(self._get_control_store)
1188
1189
@classmethod
1190
def get_default_format(klass):
1191
"""Return the current default format."""
1192
return klass._default_format
1193
1194
def get_format_string(self):
1195
"""Return the ASCII format string that identifies this format.
1196
1197
Note that in pre format ?? repositories the format string is
1198
not permitted nor written to disk.
1199
"""
1200
raise NotImplementedError(self.get_format_string)
1201
1202
def get_format_description(self):
1203
"""Return the short description for this format."""
1204
raise NotImplementedError(self.get_format_description)
1205
1206
def _get_revision_store(self, repo_transport, control_files):
1207
"""Return the revision store object for this a_bzrdir."""
1208
raise NotImplementedError(self._get_revision_store)
1209
1210
def _get_text_rev_store(self,
1211
transport,
1212
control_files,
1213
name,
1214
compressed=True,
1215
prefixed=False,
1216
serializer=None):
1217
"""Common logic for getting a revision store for a repository.
1218
1219
see self._get_revision_store for the subclass-overridable method to
1220
get the store for a repository.
1221
"""
1222
from bzrlib.store.revision.text import TextRevisionStore
1223
dir_mode = control_files._dir_mode
1224
file_mode = control_files._file_mode
1225
text_store =TextStore(transport.clone(name),
1226
prefixed=prefixed,
1227
compressed=compressed,
1228
dir_mode=dir_mode,
1229
file_mode=file_mode)
1230
_revision_store = TextRevisionStore(text_store, serializer)
1231
return _revision_store
1232
1233
def _get_versioned_file_store(self,
1234
name,
1235
transport,
1236
control_files,
1237
prefixed=True,
1238
versionedfile_class=weave.WeaveFile,
1239
versionedfile_kwargs={},
1240
escaped=False):
1241
weave_transport = control_files._transport.clone(name)
1242
dir_mode = control_files._dir_mode
1243
file_mode = control_files._file_mode
1244
return VersionedFileStore(weave_transport, prefixed=prefixed,
1245
dir_mode=dir_mode,
1246
file_mode=file_mode,
1247
versionedfile_class=versionedfile_class,
1248
versionedfile_kwargs=versionedfile_kwargs,
1249
escaped=escaped)
1250
1251
def initialize(self, a_bzrdir, shared=False):
1252
"""Initialize a repository of this format in a_bzrdir.
1253
1254
:param a_bzrdir: The bzrdir to put the new repository in it.
1255
:param shared: The repository should be initialized as a sharable one.
1256
1257
This may raise UninitializableFormat if shared repository are not
1258
compatible the a_bzrdir.
1259
"""
1260
1261
def is_supported(self):
1262
"""Is this format supported?
1263
1264
Supported formats must be initializable and openable.
1265
Unsupported formats may not support initialization or committing or
1266
some other features depending on the reason for not being supported.
1267
"""
1268
return True
1269
1270
def check_conversion_target(self, target_format):
1271
raise NotImplementedError(self.check_conversion_target)
1272
1273
def open(self, a_bzrdir, _found=False):
1274
"""Return an instance of this format for the bzrdir a_bzrdir.
1275
1276
_found is a private parameter, do not use it.
1277
"""
1278
raise NotImplementedError(self.open)
1279
1280
@classmethod
1281
def register_format(klass, format):
1282
klass._formats[format.get_format_string()] = format
1283
1284
@classmethod
1285
@deprecated_method(symbol_versioning.zero_fourteen)
1286
def set_default_format(klass, format):
1287
klass._set_default_format(format)
1288
1289
@classmethod
1290
def _set_default_format(klass, format):
1291
klass._default_format = format
1292
1293
@classmethod
1294
def unregister_format(klass, format):
1295
assert klass._formats[format.get_format_string()] is format
1296
del klass._formats[format.get_format_string()]
1297
1298
1299
class PreSplitOutRepositoryFormat(RepositoryFormat):
1300
"""Base class for the pre split out repository formats."""
1301
1302
rich_root_data = False
1303
1304
def initialize(self, a_bzrdir, shared=False, _internal=False):
1305
"""Create a weave repository.
1306
1307
TODO: when creating split out bzr branch formats, move this to a common
1308
base for Format5, Format6. or something like that.
1309
"""
1310
if shared:
1311
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1312
1313
if not _internal:
1314
# always initialized when the bzrdir is.
1315
return self.open(a_bzrdir, _found=True)
1316
1317
# Create an empty weave
1318
sio = StringIO()
1319
weavefile.write_weave_v5(weave.Weave(), sio)
1320
empty_weave = sio.getvalue()
1321
1322
mutter('creating repository in %s.', a_bzrdir.transport.base)
1323
dirs = ['revision-store', 'weaves']
1324
files = [('inventory.weave', StringIO(empty_weave)),
1325
]
1326
1327
# FIXME: RBC 20060125 don't peek under the covers
1328
# NB: no need to escape relative paths that are url safe.
1329
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1330
'branch-lock', lockable_files.TransportLock)
1331
control_files.create_lock()
1332
control_files.lock_write()
1333
control_files._transport.mkdir_multi(dirs,
1334
mode=control_files._dir_mode)
1335
try:
1336
for file, content in files:
1337
control_files.put(file, content)
1338
finally:
1339
control_files.unlock()
1340
return self.open(a_bzrdir, _found=True)
1341
1342
def _get_control_store(self, repo_transport, control_files):
1343
"""Return the control store for this repository."""
1344
return self._get_versioned_file_store('',
1345
repo_transport,
1346
control_files,
1347
prefixed=False)
1348
1349
def _get_text_store(self, transport, control_files):
1350
"""Get a store for file texts for this format."""
1351
raise NotImplementedError(self._get_text_store)
1352
1353
def open(self, a_bzrdir, _found=False):
1354
"""See RepositoryFormat.open()."""
1355
if not _found:
1356
# we are being called directly and must probe.
1357
raise NotImplementedError
1358
1359
repo_transport = a_bzrdir.get_repository_transport(None)
1360
control_files = a_bzrdir._control_files
1361
text_store = self._get_text_store(repo_transport, control_files)
1362
control_store = self._get_control_store(repo_transport, control_files)
1363
_revision_store = self._get_revision_store(repo_transport, control_files)
1364
return AllInOneRepository(_format=self,
1365
a_bzrdir=a_bzrdir,
1366
_revision_store=_revision_store,
1367
control_store=control_store,
1368
text_store=text_store)
1369
1370
def check_conversion_target(self, target_format):
1371
pass
1372
1373
1374
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1375
"""Bzr repository format 4.
1376
1377
This repository format has:
1378
- flat stores
1379
- TextStores for texts, inventories,revisions.
1380
1381
This format is deprecated: it indexes texts using a text id which is
1382
removed in format 5; initialization and write support for this format
1383
has been removed.
1384
"""
1385
1386
def __init__(self):
1387
super(RepositoryFormat4, self).__init__()
1388
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1389
1390
def get_format_description(self):
1391
"""See RepositoryFormat.get_format_description()."""
1392
return "Repository format 4"
1393
1394
def initialize(self, url, shared=False, _internal=False):
1395
"""Format 4 branches cannot be created."""
1396
raise errors.UninitializableFormat(self)
1397
1398
def is_supported(self):
1399
"""Format 4 is not supported.
1400
1401
It is not supported because the model changed from 4 to 5 and the
1402
conversion logic is expensive - so doing it on the fly was not
1403
feasible.
1404
"""
1405
return False
1406
1407
def _get_control_store(self, repo_transport, control_files):
1408
"""Format 4 repositories have no formal control store at this point.
1409
1410
This will cause any control-file-needing apis to fail - this is desired.
1411
"""
1412
return None
1413
1414
def _get_revision_store(self, repo_transport, control_files):
1415
"""See RepositoryFormat._get_revision_store()."""
1416
from bzrlib.xml4 import serializer_v4
1417
return self._get_text_rev_store(repo_transport,
1418
control_files,
1419
'revision-store',
1420
serializer=serializer_v4)
1421
1422
def _get_text_store(self, transport, control_files):
1423
"""See RepositoryFormat._get_text_store()."""
1424
1425
1426
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1427
"""Bzr control format 5.
1428
1429
This repository format has:
1430
- weaves for file texts and inventory
1431
- flat stores
1432
- TextStores for revisions and signatures.
1433
"""
1434
1435
def __init__(self):
1436
super(RepositoryFormat5, self).__init__()
1437
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1438
1439
def get_format_description(self):
1440
"""See RepositoryFormat.get_format_description()."""
1441
return "Weave repository format 5"
1442
1443
def _get_revision_store(self, repo_transport, control_files):
1444
"""See RepositoryFormat._get_revision_store()."""
1445
"""Return the revision store object for this a_bzrdir."""
1446
return self._get_text_rev_store(repo_transport,
1447
control_files,
1448
'revision-store',
1449
compressed=False)
1450
1451
def _get_text_store(self, transport, control_files):
1452
"""See RepositoryFormat._get_text_store()."""
1453
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1454
1455
1456
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1457
"""Bzr control format 6.
1458
1459
This repository format has:
1460
- weaves for file texts and inventory
1461
- hash subdirectory based stores.
1462
- TextStores for revisions and signatures.
1463
"""
1464
1465
def __init__(self):
1466
super(RepositoryFormat6, self).__init__()
1467
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1468
1469
def get_format_description(self):
1470
"""See RepositoryFormat.get_format_description()."""
1471
return "Weave repository format 6"
1472
1473
def _get_revision_store(self, repo_transport, control_files):
1474
"""See RepositoryFormat._get_revision_store()."""
1475
return self._get_text_rev_store(repo_transport,
1476
control_files,
1477
'revision-store',
1478
compressed=False,
1479
prefixed=True)
1480
1481
def _get_text_store(self, transport, control_files):
1482
"""See RepositoryFormat._get_text_store()."""
1483
return self._get_versioned_file_store('weaves', transport, control_files)
1484
1485
1486
class MetaDirRepositoryFormat(RepositoryFormat):
1487
"""Common base class for the new repositories using the metadir layout."""
1488
1489
rich_root_data = False
1490
1491
def __init__(self):
1492
super(MetaDirRepositoryFormat, self).__init__()
1493
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1494
1495
def _create_control_files(self, a_bzrdir):
1496
"""Create the required files and the initial control_files object."""
1497
# FIXME: RBC 20060125 don't peek under the covers
1498
# NB: no need to escape relative paths that are url safe.
1499
repository_transport = a_bzrdir.get_repository_transport(self)
1500
control_files = lockable_files.LockableFiles(repository_transport,
1501
'lock', lockdir.LockDir)
1502
control_files.create_lock()
1503
return control_files
1504
1505
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1506
"""Upload the initial blank content."""
1507
control_files = self._create_control_files(a_bzrdir)
1508
control_files.lock_write()
1509
try:
1510
control_files._transport.mkdir_multi(dirs,
1511
mode=control_files._dir_mode)
1512
for file, content in files:
1513
control_files.put(file, content)
1514
for file, content in utf8_files:
1515
control_files.put_utf8(file, content)
1516
if shared == True:
1517
control_files.put_utf8('shared-storage', '')
1518
finally:
1519
control_files.unlock()
1520
1521
1522
class RepositoryFormat7(MetaDirRepositoryFormat):
1523
"""Bzr repository 7.
1524
1525
This repository format has:
1526
- weaves for file texts and inventory
1527
- hash subdirectory based stores.
1528
- TextStores for revisions and signatures.
1529
- a format marker of its own
1530
- an optional 'shared-storage' flag
1531
- an optional 'no-working-trees' flag
1532
"""
1533
1534
def _get_control_store(self, repo_transport, control_files):
1535
"""Return the control store for this repository."""
1536
return self._get_versioned_file_store('',
1537
repo_transport,
1538
control_files,
1539
prefixed=False)
1540
1541
def get_format_string(self):
1542
"""See RepositoryFormat.get_format_string()."""
1543
return "Bazaar-NG Repository format 7"
1544
1545
def get_format_description(self):
1546
"""See RepositoryFormat.get_format_description()."""
1547
return "Weave repository format 7"
1548
1549
def check_conversion_target(self, target_format):
1550
pass
1551
1552
def _get_revision_store(self, repo_transport, control_files):
1553
"""See RepositoryFormat._get_revision_store()."""
1554
return self._get_text_rev_store(repo_transport,
1555
control_files,
1556
'revision-store',
1557
compressed=False,
1558
prefixed=True,
1559
)
1560
1561
def _get_text_store(self, transport, control_files):
1562
"""See RepositoryFormat._get_text_store()."""
1563
return self._get_versioned_file_store('weaves',
1564
transport,
1565
control_files)
1566
1567
def initialize(self, a_bzrdir, shared=False):
1568
"""Create a weave repository.
1569
1570
:param shared: If true the repository will be initialized as a shared
1571
repository.
1572
"""
1573
# Create an empty weave
1574
sio = StringIO()
1575
weavefile.write_weave_v5(weave.Weave(), sio)
1576
empty_weave = sio.getvalue()
1577
1578
mutter('creating repository in %s.', a_bzrdir.transport.base)
1579
dirs = ['revision-store', 'weaves']
1580
files = [('inventory.weave', StringIO(empty_weave)),
1581
]
1582
utf8_files = [('format', self.get_format_string())]
1583
1584
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1585
return self.open(a_bzrdir=a_bzrdir, _found=True)
1586
1587
def open(self, a_bzrdir, _found=False, _override_transport=None):
1588
"""See RepositoryFormat.open().
1589
1590
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1591
repository at a slightly different url
1592
than normal. I.e. during 'upgrade'.
1593
"""
1594
if not _found:
1595
format = RepositoryFormat.find_format(a_bzrdir)
1596
assert format.__class__ == self.__class__
1597
if _override_transport is not None:
1598
repo_transport = _override_transport
1599
else:
1600
repo_transport = a_bzrdir.get_repository_transport(None)
1601
control_files = lockable_files.LockableFiles(repo_transport,
1602
'lock', lockdir.LockDir)
1603
text_store = self._get_text_store(repo_transport, control_files)
1604
control_store = self._get_control_store(repo_transport, control_files)
1605
_revision_store = self._get_revision_store(repo_transport, control_files)
1606
return WeaveMetaDirRepository(_format=self,
1607
a_bzrdir=a_bzrdir,
1608
control_files=control_files,
1609
_revision_store=_revision_store,
1610
control_store=control_store,
1611
text_store=text_store)
1612
1613
1614
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1615
"""Bzr repository knit format (generalized).
1616
1617
This repository format has:
1618
- knits for file texts and inventory
1619
- hash subdirectory based stores.
1620
- knits for revisions and signatures
1621
- TextStores for revisions and signatures.
1622
- a format marker of its own
1623
- an optional 'shared-storage' flag
1624
- an optional 'no-working-trees' flag
1625
- a LockDir lock
1626
"""
1627
1628
def _get_control_store(self, repo_transport, control_files):
1629
"""Return the control store for this repository."""
1630
return VersionedFileStore(
1631
repo_transport,
1632
prefixed=False,
1633
file_mode=control_files._file_mode,
1634
versionedfile_class=knit.KnitVersionedFile,
1635
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1636
)
1637
1638
def _get_revision_store(self, repo_transport, control_files):
1639
"""See RepositoryFormat._get_revision_store()."""
1640
from bzrlib.store.revision.knit import KnitRevisionStore
1641
versioned_file_store = VersionedFileStore(
1642
repo_transport,
1643
file_mode=control_files._file_mode,
1644
prefixed=False,
1645
precious=True,
1646
versionedfile_class=knit.KnitVersionedFile,
1647
versionedfile_kwargs={'delta':False,
1648
'factory':knit.KnitPlainFactory(),
1649
},
1650
escaped=True,
1651
)
1652
return KnitRevisionStore(versioned_file_store)
1653
1654
def _get_text_store(self, transport, control_files):
1655
"""See RepositoryFormat._get_text_store()."""
1656
return self._get_versioned_file_store('knits',
1657
transport,
1658
control_files,
1659
versionedfile_class=knit.KnitVersionedFile,
1660
versionedfile_kwargs={
1661
'create_parent_dir':True,
1662
'delay_create':True,
1663
'dir_mode':control_files._dir_mode,
1664
},
1665
escaped=True)
1666
1667
def initialize(self, a_bzrdir, shared=False):
1668
"""Create a knit format 1 repository.
1669
1670
:param a_bzrdir: bzrdir to contain the new repository; must already
1671
be initialized.
1672
:param shared: If true the repository will be initialized as a shared
1673
repository.
1674
"""
1675
mutter('creating repository in %s.', a_bzrdir.transport.base)
1676
dirs = ['revision-store', 'knits']
1677
files = []
1678
utf8_files = [('format', self.get_format_string())]
1679
1680
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1681
repo_transport = a_bzrdir.get_repository_transport(None)
1682
control_files = lockable_files.LockableFiles(repo_transport,
1683
'lock', lockdir.LockDir)
1684
control_store = self._get_control_store(repo_transport, control_files)
1685
transaction = transactions.WriteTransaction()
1686
# trigger a write of the inventory store.
1687
control_store.get_weave_or_empty('inventory', transaction)
1688
_revision_store = self._get_revision_store(repo_transport, control_files)
1689
# the revision id here is irrelevant: it will not be stored, and cannot
1690
# already exist.
1691
_revision_store.has_revision_id('A', transaction)
1692
_revision_store.get_signature_file(transaction)
1693
return self.open(a_bzrdir=a_bzrdir, _found=True)
1694
1695
def open(self, a_bzrdir, _found=False, _override_transport=None):
1696
"""See RepositoryFormat.open().
1697
1698
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1699
repository at a slightly different url
1700
than normal. I.e. during 'upgrade'.
1701
"""
1702
if not _found:
1703
format = RepositoryFormat.find_format(a_bzrdir)
1704
assert format.__class__ == self.__class__
1705
if _override_transport is not None:
1706
repo_transport = _override_transport
1707
else:
1708
repo_transport = a_bzrdir.get_repository_transport(None)
1709
control_files = lockable_files.LockableFiles(repo_transport,
1710
'lock', lockdir.LockDir)
1711
text_store = self._get_text_store(repo_transport, control_files)
1712
control_store = self._get_control_store(repo_transport, control_files)
1713
_revision_store = self._get_revision_store(repo_transport, control_files)
1714
return KnitRepository(_format=self,
1715
a_bzrdir=a_bzrdir,
1716
control_files=control_files,
1717
_revision_store=_revision_store,
1718
control_store=control_store,
1719
text_store=text_store)
1720
1721
1722
class RepositoryFormatKnit1(RepositoryFormatKnit):
1723
"""Bzr repository knit format 1.
1724
1725
This repository format has:
1726
- knits for file texts and inventory
1727
- hash subdirectory based stores.
1728
- knits for revisions and signatures
1729
- TextStores for revisions and signatures.
1730
- a format marker of its own
1731
- an optional 'shared-storage' flag
1732
- an optional 'no-working-trees' flag
1733
- a LockDir lock
1734
1735
This format was introduced in bzr 0.8.
1736
"""
1737
def get_format_string(self):
1738
"""See RepositoryFormat.get_format_string()."""
1739
return "Bazaar-NG Knit Repository Format 1"
1740
1741
def get_format_description(self):
1742
"""See RepositoryFormat.get_format_description()."""
1743
return "Knit repository format 1"
1744
1745
def check_conversion_target(self, target_format):
1746
pass
1747
1748
1749
class RepositoryFormatKnit2(RepositoryFormatKnit):
1750
"""Bzr repository knit format 2.
1751
1752
THIS FORMAT IS EXPERIMENTAL
1753
This repository format has:
1754
- knits for file texts and inventory
1755
- hash subdirectory based stores.
1756
- knits for revisions and signatures
1757
- TextStores for revisions and signatures.
1758
- a format marker of its own
1759
- an optional 'shared-storage' flag
1760
- an optional 'no-working-trees' flag
1761
- a LockDir lock
1762
- Support for recording full info about the tree root
1763
1764
"""
1765
1766
rich_root_data = True
1767
1768
def get_format_string(self):
1769
"""See RepositoryFormat.get_format_string()."""
1770
return "Bazaar Knit Repository Format 2\n"
1771
1772
def get_format_description(self):
1773
"""See RepositoryFormat.get_format_description()."""
1774
return "Knit repository format 2"
1775
1776
def check_conversion_target(self, target_format):
1777
if not target_format.rich_root_data:
1778
raise errors.BadConversionTarget(
1779
'Does not support rich root data.', target_format)
1780
1781
def open(self, a_bzrdir, _found=False, _override_transport=None):
1782
"""See RepositoryFormat.open().
1783
1784
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1785
repository at a slightly different url
1786
than normal. I.e. during 'upgrade'.
1787
"""
1788
if not _found:
1789
format = RepositoryFormat.find_format(a_bzrdir)
1790
assert format.__class__ == self.__class__
1791
if _override_transport is not None:
1792
repo_transport = _override_transport
1793
else:
1794
repo_transport = a_bzrdir.get_repository_transport(None)
1795
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1796
lockdir.LockDir)
1797
text_store = self._get_text_store(repo_transport, control_files)
1798
control_store = self._get_control_store(repo_transport, control_files)
1799
_revision_store = self._get_revision_store(repo_transport, control_files)
1800
return KnitRepository2(_format=self,
1801
a_bzrdir=a_bzrdir,
1802
control_files=control_files,
1803
_revision_store=_revision_store,
1804
control_store=control_store,
1805
text_store=text_store)
1806
1807
1808
1809
# formats which have no format string are not discoverable
1810
# and not independently creatable, so are not registered.
1811
RepositoryFormat.register_format(RepositoryFormat7())
1812
# KEEP in sync with bzrdir.format_registry default
1813
_default_format = RepositoryFormatKnit1()
1814
RepositoryFormat.register_format(_default_format)
1815
RepositoryFormat.register_format(RepositoryFormatKnit2())
1816
RepositoryFormat._set_default_format(_default_format)
1817
_legacy_formats = [RepositoryFormat4(),
1818
RepositoryFormat5(),
1819
RepositoryFormat6()]
1820
1821
1822
class InterRepository(InterObject):
1823
"""This class represents operations taking place between two repositories.
1824
1825
Its instances have methods like copy_content and fetch, and contain
1826
references to the source and target repositories these operations can be
1827
carried out on.
1828
1829
Often we will provide convenience methods on 'repository' which carry out
1830
operations with another repository - they will always forward to
1831
InterRepository.get(other).method_name(parameters).
1832
"""
1833
1834
_optimisers = []
1835
"""The available optimised InterRepository types."""
1836
1837
def copy_content(self, revision_id=None, basis=None):
1838
raise NotImplementedError(self.copy_content)
1839
1840
def fetch(self, revision_id=None, pb=None):
1841
"""Fetch the content required to construct revision_id.
1842
1843
The content is copied from self.source to self.target.
1844
1845
:param revision_id: if None all content is copied, if NULL_REVISION no
1846
content is copied.
1847
:param pb: optional progress bar to use for progress reports. If not
1848
provided a default one will be created.
1849
1850
Returns the copied revision count and the failed revisions in a tuple:
1851
(copied, failures).
1852
"""
1853
raise NotImplementedError(self.fetch)
1854
1855
@needs_read_lock
1856
def missing_revision_ids(self, revision_id=None):
1857
"""Return the revision ids that source has that target does not.
1858
1859
These are returned in topological order.
1860
1861
:param revision_id: only return revision ids included by this
1862
revision_id.
1863
"""
1864
# generic, possibly worst case, slow code path.
1865
target_ids = set(self.target.all_revision_ids())
1866
if revision_id is not None:
1867
source_ids = self.source.get_ancestry(revision_id)
1868
assert source_ids[0] is None
1869
source_ids.pop(0)
1870
else:
1871
source_ids = self.source.all_revision_ids()
1872
result_set = set(source_ids).difference(target_ids)
1873
# this may look like a no-op: its not. It preserves the ordering
1874
# other_ids had while only returning the members from other_ids
1875
# that we've decided we need.
1876
return [rev_id for rev_id in source_ids if rev_id in result_set]
1877
1878
1879
class InterSameDataRepository(InterRepository):
1880
"""Code for converting between repositories that represent the same data.
1881
1882
Data format and model must match for this to work.
1883
"""
1884
1885
_matching_repo_format = RepositoryFormat4()
1886
"""Repository format for testing with."""
1887
1888
@staticmethod
1889
def is_compatible(source, target):
1890
if not isinstance(source, Repository):
1891
return False
1892
if not isinstance(target, Repository):
1893
return False
1894
if source._format.rich_root_data == target._format.rich_root_data:
1895
return True
1896
else:
1897
return False
1898
1899
@needs_write_lock
1900
def copy_content(self, revision_id=None, basis=None):
1901
"""Make a complete copy of the content in self into destination.
1902
1903
This is a destructive operation! Do not use it on existing
1904
repositories.
1905
1906
:param revision_id: Only copy the content needed to construct
1907
revision_id and its parents.
1908
:param basis: Copy the needed data preferentially from basis.
1909
"""
1910
try:
1911
self.target.set_make_working_trees(self.source.make_working_trees())
1912
except NotImplementedError:
1913
pass
1914
# grab the basis available data
1915
if basis is not None:
1916
self.target.fetch(basis, revision_id=revision_id)
1917
# but don't bother fetching if we have the needed data now.
1918
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1919
self.target.has_revision(revision_id)):
1920
return
1921
self.target.fetch(self.source, revision_id=revision_id)
1922
1923
@needs_write_lock
1924
def fetch(self, revision_id=None, pb=None):
1925
"""See InterRepository.fetch()."""
1926
from bzrlib.fetch import GenericRepoFetcher
1927
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1928
self.source, self.source._format, self.target,
1929
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
1937
class InterWeaveRepo(InterSameDataRepository):
1938
"""Optimised code paths between Weave based repositories."""
1939
1940
_matching_repo_format = RepositoryFormat7()
1941
"""Repository format for testing with."""
1942
1943
@staticmethod
1944
def is_compatible(source, target):
1945
"""Be compatible with known Weave formats.
1946
1947
We don't test for the stores being of specific types because that
1948
could lead to confusing results, and there is no need to be
1949
overly general.
1950
"""
1951
try:
1952
return (isinstance(source._format, (RepositoryFormat5,
1953
RepositoryFormat6,
1954
RepositoryFormat7)) and
1955
isinstance(target._format, (RepositoryFormat5,
1956
RepositoryFormat6,
1957
RepositoryFormat7)))
1958
except AttributeError:
1959
return False
1960
1961
@needs_write_lock
1962
def copy_content(self, revision_id=None, basis=None):
1963
"""See InterRepository.copy_content()."""
1964
# weave specific optimised path:
1965
if basis is not None:
1966
# copy the basis in, then fetch remaining data.
1967
basis.copy_content_into(self.target, revision_id)
1968
# the basis copy_content_into could miss-set this.
1969
try:
1970
self.target.set_make_working_trees(self.source.make_working_trees())
1971
except NotImplementedError:
1972
pass
1973
self.target.fetch(self.source, revision_id=revision_id)
1974
else:
1975
try:
1976
self.target.set_make_working_trees(self.source.make_working_trees())
1977
except NotImplementedError:
1978
pass
1979
# FIXME do not peek!
1980
if self.source.control_files._transport.listable():
1981
pb = ui.ui_factory.nested_progress_bar()
1982
try:
1983
self.target.weave_store.copy_all_ids(
1984
self.source.weave_store,
1985
pb=pb,
1986
from_transaction=self.source.get_transaction(),
1987
to_transaction=self.target.get_transaction())
1988
pb.update('copying inventory', 0, 1)
1989
self.target.control_weaves.copy_multi(
1990
self.source.control_weaves, ['inventory'],
1991
from_transaction=self.source.get_transaction(),
1992
to_transaction=self.target.get_transaction())
1993
self.target._revision_store.text_store.copy_all_ids(
1994
self.source._revision_store.text_store,
1995
pb=pb)
1996
finally:
1997
pb.finished()
1998
else:
1999
self.target.fetch(self.source, revision_id=revision_id)
2000
2001
@needs_write_lock
2002
def fetch(self, revision_id=None, pb=None):
2003
"""See InterRepository.fetch()."""
2004
from bzrlib.fetch import GenericRepoFetcher
2005
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2006
self.source, self.source._format, self.target, self.target._format)
2007
f = GenericRepoFetcher(to_repository=self.target,
2008
from_repository=self.source,
2009
last_revision=revision_id,
2010
pb=pb)
2011
return f.count_copied, f.failed_revisions
2012
2013
@needs_read_lock
2014
def missing_revision_ids(self, revision_id=None):
2015
"""See InterRepository.missing_revision_ids()."""
2016
# we want all revisions to satisfy revision_id in source.
2017
# but we don't want to stat every file here and there.
2018
# we want then, all revisions other needs to satisfy revision_id
2019
# checked, but not those that we have locally.
2020
# so the first thing is to get a subset of the revisions to
2021
# satisfy revision_id in source, and then eliminate those that
2022
# we do already have.
2023
# this is slow on high latency connection to self, but as as this
2024
# disk format scales terribly for push anyway due to rewriting
2025
# inventory.weave, this is considered acceptable.
2026
# - RBC 20060209
2027
if revision_id is not None:
2028
source_ids = self.source.get_ancestry(revision_id)
2029
assert source_ids[0] is None
2030
source_ids.pop(0)
2031
else:
2032
source_ids = self.source._all_possible_ids()
2033
source_ids_set = set(source_ids)
2034
# source_ids is the worst possible case we may need to pull.
2035
# now we want to filter source_ids against what we actually
2036
# have in target, but don't try to check for existence where we know
2037
# we do not have a revision as that would be pointless.
2038
target_ids = set(self.target._all_possible_ids())
2039
possibly_present_revisions = target_ids.intersection(source_ids_set)
2040
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2041
required_revisions = source_ids_set.difference(actually_present_revisions)
2042
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2043
if revision_id is not None:
2044
# we used get_ancestry to determine source_ids then we are assured all
2045
# revisions referenced are present as they are installed in topological order.
2046
# and the tip revision was validated by get_ancestry.
2047
return required_topo_revisions
2048
else:
2049
# if we just grabbed the possibly available ids, then
2050
# we only have an estimate of whats available and need to validate
2051
# that against the revision records.
2052
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2053
2054
2055
class InterKnitRepo(InterSameDataRepository):
2056
"""Optimised code paths between Knit based repositories."""
2057
2058
_matching_repo_format = RepositoryFormatKnit1()
2059
"""Repository format for testing with."""
2060
2061
@staticmethod
2062
def is_compatible(source, target):
2063
"""Be compatible with known Knit formats.
2064
2065
We don't test for the stores being of specific types because that
2066
could lead to confusing results, and there is no need to be
2067
overly general.
2068
"""
2069
try:
2070
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2071
isinstance(target._format, (RepositoryFormatKnit1)))
2072
except AttributeError:
2073
return False
2074
2075
@needs_write_lock
2076
def fetch(self, revision_id=None, pb=None):
2077
"""See InterRepository.fetch()."""
2078
from bzrlib.fetch import KnitRepoFetcher
2079
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2080
self.source, self.source._format, self.target, self.target._format)
2081
f = KnitRepoFetcher(to_repository=self.target,
2082
from_repository=self.source,
2083
last_revision=revision_id,
2084
pb=pb)
2085
return f.count_copied, f.failed_revisions
2086
2087
@needs_read_lock
2088
def missing_revision_ids(self, revision_id=None):
2089
"""See InterRepository.missing_revision_ids()."""
2090
if revision_id is not None:
2091
source_ids = self.source.get_ancestry(revision_id)
2092
assert source_ids[0] is None
2093
source_ids.pop(0)
2094
else:
2095
source_ids = self.source._all_possible_ids()
2096
source_ids_set = set(source_ids)
2097
# source_ids is the worst possible case we may need to pull.
2098
# now we want to filter source_ids against what we actually
2099
# have in target, but don't try to check for existence where we know
2100
# we do not have a revision as that would be pointless.
2101
target_ids = set(self.target._all_possible_ids())
2102
possibly_present_revisions = target_ids.intersection(source_ids_set)
2103
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2104
required_revisions = source_ids_set.difference(actually_present_revisions)
2105
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2106
if revision_id is not None:
2107
# we used get_ancestry to determine source_ids then we are assured all
2108
# revisions referenced are present as they are installed in topological order.
2109
# and the tip revision was validated by get_ancestry.
2110
return required_topo_revisions
2111
else:
2112
# if we just grabbed the possibly available ids, then
2113
# we only have an estimate of whats available and need to validate
2114
# that against the revision records.
2115
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2116
2117
2118
class InterModel1and2(InterRepository):
2119
2120
_matching_repo_format = None
2121
2122
@staticmethod
2123
def is_compatible(source, target):
2124
if not isinstance(source, Repository):
2125
return False
2126
if not isinstance(target, Repository):
2127
return False
2128
if not source._format.rich_root_data and target._format.rich_root_data:
2129
return True
2130
else:
2131
return False
2132
2133
@needs_write_lock
2134
def fetch(self, revision_id=None, pb=None):
2135
"""See InterRepository.fetch()."""
2136
from bzrlib.fetch import Model1toKnit2Fetcher
2137
f = Model1toKnit2Fetcher(to_repository=self.target,
2138
from_repository=self.source,
2139
last_revision=revision_id,
2140
pb=pb)
2141
return f.count_copied, f.failed_revisions
2142
2143
@needs_write_lock
2144
def copy_content(self, revision_id=None, basis=None):
2145
"""Make a complete copy of the content in self into destination.
2146
2147
This is a destructive operation! Do not use it on existing
2148
repositories.
2149
2150
:param revision_id: Only copy the content needed to construct
2151
revision_id and its parents.
2152
:param basis: Copy the needed data preferentially from basis.
2153
"""
2154
try:
2155
self.target.set_make_working_trees(self.source.make_working_trees())
2156
except NotImplementedError:
2157
pass
2158
# grab the basis available data
2159
if basis is not None:
2160
self.target.fetch(basis, revision_id=revision_id)
2161
# but don't bother fetching if we have the needed data now.
2162
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2163
self.target.has_revision(revision_id)):
2164
return
2165
self.target.fetch(self.source, revision_id=revision_id)
2166
2167
2168
class InterKnit1and2(InterKnitRepo):
2169
2170
_matching_repo_format = None
2171
2172
@staticmethod
2173
def is_compatible(source, target):
2174
"""Be compatible with Knit1 source and Knit2 target"""
2175
try:
2176
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2177
isinstance(target._format, (RepositoryFormatKnit2)))
2178
except AttributeError:
2179
return False
2180
2181
@needs_write_lock
2182
def fetch(self, revision_id=None, pb=None):
2183
"""See InterRepository.fetch()."""
2184
from bzrlib.fetch import Knit1to2Fetcher
2185
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2186
self.source, self.source._format, self.target,
2187
self.target._format)
2188
f = Knit1to2Fetcher(to_repository=self.target,
2189
from_repository=self.source,
2190
last_revision=revision_id,
2191
pb=pb)
2192
return f.count_copied, f.failed_revisions
2193
2194
2195
InterRepository.register_optimiser(InterSameDataRepository)
2196
InterRepository.register_optimiser(InterWeaveRepo)
2197
InterRepository.register_optimiser(InterKnitRepo)
2198
InterRepository.register_optimiser(InterModel1and2)
2199
InterRepository.register_optimiser(InterKnit1and2)
2200
2201
2202
class RepositoryTestProviderAdapter(object):
2203
"""A tool to generate a suite testing multiple repository formats at once.
2204
2205
This is done by copying the test once for each transport and injecting
2206
the transport_server, transport_readonly_server, and bzrdir_format and
2207
repository_format classes into each copy. Each copy is also given a new id()
2208
to make it easy to identify.
2209
"""
2210
2211
def __init__(self, transport_server, transport_readonly_server, formats):
2212
self._transport_server = transport_server
2213
self._transport_readonly_server = transport_readonly_server
2214
self._formats = formats
2215
2216
def adapt(self, test):
2217
result = unittest.TestSuite()
2218
for repository_format, bzrdir_format in self._formats:
2219
new_test = deepcopy(test)
2220
new_test.transport_server = self._transport_server
2221
new_test.transport_readonly_server = self._transport_readonly_server
2222
new_test.bzrdir_format = bzrdir_format
2223
new_test.repository_format = repository_format
2224
def make_new_test_id():
2225
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2226
return lambda: new_id
2227
new_test.id = make_new_test_id()
2228
result.addTest(new_test)
2229
return result
2230
2231
2232
class InterRepositoryTestProviderAdapter(object):
2233
"""A tool to generate a suite testing multiple inter repository formats.
2234
2235
This is done by copying the test once for each interrepo provider and injecting
2236
the transport_server, transport_readonly_server, repository_format and
2237
repository_to_format classes into each copy.
2238
Each copy is also given a new id() to make it easy to identify.
2239
"""
2240
2241
def __init__(self, transport_server, transport_readonly_server, formats):
2242
self._transport_server = transport_server
2243
self._transport_readonly_server = transport_readonly_server
2244
self._formats = formats
2245
2246
def adapt(self, test):
2247
result = unittest.TestSuite()
2248
for interrepo_class, repository_format, repository_format_to in self._formats:
2249
new_test = deepcopy(test)
2250
new_test.transport_server = self._transport_server
2251
new_test.transport_readonly_server = self._transport_readonly_server
2252
new_test.interrepo_class = interrepo_class
2253
new_test.repository_format = repository_format
2254
new_test.repository_format_to = repository_format_to
2255
def make_new_test_id():
2256
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2257
return lambda: new_id
2258
new_test.id = make_new_test_id()
2259
result.addTest(new_test)
2260
return result
2261
2262
@staticmethod
2263
def default_test_list():
2264
"""Generate the default list of interrepo permutations to test."""
2265
result = []
2266
# test the default InterRepository between format 6 and the current
2267
# default format.
2268
# XXX: robertc 20060220 reinstate this when there are two supported
2269
# formats which do not have an optimal code path between them.
2270
#result.append((InterRepository,
2271
# RepositoryFormat6(),
2272
# RepositoryFormatKnit1()))
2273
for optimiser in InterRepository._optimisers:
2274
if optimiser._matching_repo_format is not None:
2275
result.append((optimiser,
2276
optimiser._matching_repo_format,
2277
optimiser._matching_repo_format
2278
))
2279
# if there are specific combinations we want to use, we can add them
2280
# here.
2281
result.append((InterModel1and2, RepositoryFormat5(),
2282
RepositoryFormatKnit2()))
2283
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2284
RepositoryFormatKnit2()))
2285
return result
2286
2287
2288
class CopyConverter(object):
2289
"""A repository conversion tool which just performs a copy of the content.
2290
2291
This is slow but quite reliable.
2292
"""
2293
2294
def __init__(self, target_format):
2295
"""Create a CopyConverter.
2296
2297
:param target_format: The format the resulting repository should be.
2298
"""
2299
self.target_format = target_format
2300
2301
def convert(self, repo, pb):
2302
"""Perform the conversion of to_convert, giving feedback via pb.
2303
2304
:param to_convert: The disk object to convert.
2305
:param pb: a progress bar to use for progress information.
2306
"""
2307
self.pb = pb
2308
self.count = 0
2309
self.total = 4
2310
# this is only useful with metadir layouts - separated repo content.
2311
# trigger an assertion if not such
2312
repo._format.get_format_string()
2313
self.repo_dir = repo.bzrdir
2314
self.step('Moving repository to repository.backup')
2315
self.repo_dir.transport.move('repository', 'repository.backup')
2316
backup_transport = self.repo_dir.transport.clone('repository.backup')
2317
repo._format.check_conversion_target(self.target_format)
2318
self.source_repo = repo._format.open(self.repo_dir,
2319
_found=True,
2320
_override_transport=backup_transport)
2321
self.step('Creating new repository')
2322
converted = self.target_format.initialize(self.repo_dir,
2323
self.source_repo.is_shared())
2324
converted.lock_write()
2325
try:
2326
self.step('Copying content into repository.')
2327
self.source_repo.copy_content_into(converted)
2328
finally:
2329
converted.unlock()
2330
self.step('Deleting old repository content.')
2331
self.repo_dir.transport.delete_tree('repository.backup')
2332
self.pb.note('repository converted')
2333
2334
def step(self, message):
2335
"""Update the pb by a step."""
2336
self.count +=1
2337
self.pb.update(message, self.count, self.total)
2338
2339
2340
class CommitBuilder(object):
2341
"""Provides an interface to build up a commit.
2342
2343
This allows describing a tree to be committed without needing to
2344
know the internals of the format of the repository.
2345
"""
2346
2347
record_root_entry = False
2348
def __init__(self, repository, parents, config, timestamp=None,
2349
timezone=None, committer=None, revprops=None,
2350
revision_id=None):
2351
"""Initiate a CommitBuilder.
2352
2353
:param repository: Repository to commit to.
2354
:param parents: Revision ids of the parents of the new revision.
2355
:param config: Configuration to use.
2356
:param timestamp: Optional timestamp recorded for commit.
2357
:param timezone: Optional timezone for timestamp.
2358
:param committer: Optional committer to set for commit.
2359
:param revprops: Optional dictionary of revision properties.
2360
:param revision_id: Optional revision id.
2361
"""
2362
self._config = config
2363
2364
if committer is None:
2365
self._committer = self._config.username()
2366
else:
2367
assert isinstance(committer, basestring), type(committer)
2368
self._committer = committer
2369
2370
self.new_inventory = Inventory(None)
2371
self._new_revision_id = revision_id
2372
self.parents = parents
2373
self.repository = repository
2374
2375
self._revprops = {}
2376
if revprops is not None:
2377
self._revprops.update(revprops)
2378
2379
if timestamp is None:
2380
timestamp = time.time()
2381
# Restrict resolution to 1ms
2382
self._timestamp = round(timestamp, 3)
2383
2384
if timezone is None:
2385
self._timezone = local_time_offset()
2386
else:
2387
self._timezone = int(timezone)
2388
2389
self._generate_revision_if_needed()
2390
2391
def commit(self, message):
2392
"""Make the actual commit.
2393
2394
:return: The revision id of the recorded revision.
2395
"""
2396
rev = _mod_revision.Revision(
2397
timestamp=self._timestamp,
2398
timezone=self._timezone,
2399
committer=self._committer,
2400
message=message,
2401
inventory_sha1=self.inv_sha1,
2402
revision_id=self._new_revision_id,
2403
properties=self._revprops)
2404
rev.parent_ids = self.parents
2405
self.repository.add_revision(self._new_revision_id, rev,
2406
self.new_inventory, self._config)
2407
return self._new_revision_id
2408
2409
def revision_tree(self):
2410
"""Return the tree that was just committed.
2411
2412
After calling commit() this can be called to get a RevisionTree
2413
representing the newly committed tree. This is preferred to
2414
calling Repository.revision_tree() because that may require
2415
deserializing the inventory, while we already have a copy in
2416
memory.
2417
"""
2418
return RevisionTree(self.repository, self.new_inventory,
2419
self._new_revision_id)
2420
2421
def finish_inventory(self):
2422
"""Tell the builder that the inventory is finished."""
2423
if self.new_inventory.root is None:
2424
symbol_versioning.warn('Root entry should be supplied to'
2425
' record_entry_contents, as of bzr 0.10.',
2426
DeprecationWarning, stacklevel=2)
2427
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2428
self.new_inventory.revision_id = self._new_revision_id
2429
self.inv_sha1 = self.repository.add_inventory(
2430
self._new_revision_id,
2431
self.new_inventory,
2432
self.parents
2433
)
2434
2435
def _gen_revision_id(self):
2436
"""Return new revision-id."""
2437
return generate_ids.gen_revision_id(self._config.username(),
2438
self._timestamp)
2439
2440
def _generate_revision_if_needed(self):
2441
"""Create a revision id if None was supplied.
2442
2443
If the repository can not support user-specified revision ids
2444
they should override this function and raise CannotSetRevisionId
2445
if _new_revision_id is not None.
2446
2447
:raises: CannotSetRevisionId
2448
"""
2449
if self._new_revision_id is None:
2450
self._new_revision_id = self._gen_revision_id()
2451
2452
def record_entry_contents(self, ie, parent_invs, path, tree):
2453
"""Record the content of ie from tree into the commit if needed.
2454
2455
Side effect: sets ie.revision when unchanged
2456
2457
:param ie: An inventory entry present in the commit.
2458
:param parent_invs: The inventories of the parent revisions of the
2459
commit.
2460
:param path: The path the entry is at in the tree.
2461
:param tree: The tree which contains this entry and should be used to
2462
obtain content.
2463
"""
2464
if self.new_inventory.root is None and ie.parent_id is not None:
2465
symbol_versioning.warn('Root entry should be supplied to'
2466
' record_entry_contents, as of bzr 0.10.',
2467
DeprecationWarning, stacklevel=2)
2468
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2469
'', tree)
2470
self.new_inventory.add(ie)
2471
2472
# ie.revision is always None if the InventoryEntry is considered
2473
# for committing. ie.snapshot will record the correct revision
2474
# which may be the sole parent if it is untouched.
2475
if ie.revision is not None:
2476
return
2477
2478
# In this revision format, root entries have no knit or weave
2479
if ie is self.new_inventory.root:
2480
# When serializing out to disk and back in
2481
# root.revision is always _new_revision_id
2482
ie.revision = self._new_revision_id
2483
return
2484
previous_entries = ie.find_previous_heads(
2485
parent_invs,
2486
self.repository.weave_store,
2487
self.repository.get_transaction())
2488
# we are creating a new revision for ie in the history store
2489
# and inventory.
2490
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2491
2492
def modified_directory(self, file_id, file_parents):
2493
"""Record the presence of a symbolic link.
2494
2495
:param file_id: The file_id of the link to record.
2496
:param file_parents: The per-file parent revision ids.
2497
"""
2498
self._add_text_to_weave(file_id, [], file_parents.keys())
2499
2500
def modified_file_text(self, file_id, file_parents,
2501
get_content_byte_lines, text_sha1=None,
2502
text_size=None):
2503
"""Record the text of file file_id
2504
2505
:param file_id: The file_id of the file to record the text of.
2506
:param file_parents: The per-file parent revision ids.
2507
:param get_content_byte_lines: A callable which will return the byte
2508
lines for the file.
2509
:param text_sha1: Optional SHA1 of the file contents.
2510
:param text_size: Optional size of the file contents.
2511
"""
2512
# mutter('storing text of file {%s} in revision {%s} into %r',
2513
# file_id, self._new_revision_id, self.repository.weave_store)
2514
# special case to avoid diffing on renames or
2515
# reparenting
2516
if (len(file_parents) == 1
2517
and text_sha1 == file_parents.values()[0].text_sha1
2518
and text_size == file_parents.values()[0].text_size):
2519
previous_ie = file_parents.values()[0]
2520
versionedfile = self.repository.weave_store.get_weave(file_id,
2521
self.repository.get_transaction())
2522
versionedfile.clone_text(self._new_revision_id,
2523
previous_ie.revision, file_parents.keys())
2524
return text_sha1, text_size
2525
else:
2526
new_lines = get_content_byte_lines()
2527
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2528
# should return the SHA1 and size
2529
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2530
return osutils.sha_strings(new_lines), \
2531
sum(map(len, new_lines))
2532
2533
def modified_link(self, file_id, file_parents, link_target):
2534
"""Record the presence of a symbolic link.
2535
2536
:param file_id: The file_id of the link to record.
2537
:param file_parents: The per-file parent revision ids.
2538
:param link_target: Target location of this link.
2539
"""
2540
self._add_text_to_weave(file_id, [], file_parents.keys())
2541
2542
def _add_text_to_weave(self, file_id, new_lines, parents):
2543
versionedfile = self.repository.weave_store.get_weave_or_empty(
2544
file_id, self.repository.get_transaction())
2545
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2546
versionedfile.clear_cache()
2547
2548
2549
class _CommitBuilder(CommitBuilder):
2550
"""Temporary class so old CommitBuilders are detected properly
2551
2552
Note: CommitBuilder works whether or not root entry is recorded.
2553
"""
2554
2555
record_root_entry = True
2556
2557
2558
class RootCommitBuilder(CommitBuilder):
2559
"""This commitbuilder actually records the root id"""
2560
2561
record_root_entry = True
2562
2563
def record_entry_contents(self, ie, parent_invs, path, tree):
2564
"""Record the content of ie from tree into the commit if needed.
2565
2566
Side effect: sets ie.revision when unchanged
2567
2568
:param ie: An inventory entry present in the commit.
2569
:param parent_invs: The inventories of the parent revisions of the
2570
commit.
2571
:param path: The path the entry is at in the tree.
2572
:param tree: The tree which contains this entry and should be used to
2573
obtain content.
2574
"""
2575
assert self.new_inventory.root is not None or ie.parent_id is None
2576
self.new_inventory.add(ie)
2577
2578
# ie.revision is always None if the InventoryEntry is considered
2579
# for committing. ie.snapshot will record the correct revision
2580
# which may be the sole parent if it is untouched.
2581
if ie.revision is not None:
2582
return
2583
2584
previous_entries = ie.find_previous_heads(
2585
parent_invs,
2586
self.repository.weave_store,
2587
self.repository.get_transaction())
2588
# we are creating a new revision for ie in the history store
2589
# and inventory.
2590
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2591
2592
2593
_unescape_map = {
2594
'apos':"'",
2595
'quot':'"',
2596
'amp':'&',
2597
'lt':'<',
2598
'gt':'>'
2599
}
2600
2601
2602
def _unescaper(match, _map=_unescape_map):
2603
return _map[match.group(1)]
2604
2605
2606
_unescape_re = None
2607
2608
2609
def _unescape_xml(data):
2610
"""Unescape predefined XML entities in a string of data."""
2611
global _unescape_re
2612
if _unescape_re is None:
2613
_unescape_re = re.compile('\&([^;]*);')
2614
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0