« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 3606.7.7
- compare with another revision
- download diff
- download tarball
- view history from revision 3606.7.7
- Committer: John Arbash Meinel
- Date: 2008-08-18 22:34:21 UTC
- mto: (3606.5.6 1.6)
- mto: This revision was merged to the branch mainline in revision 3641.
- Revision ID: john@arbash-meinel.com-20080818223421-todjny24vj4faj4t
Add tests for the fetching behavior.
The proper parameter passed is 'unordered' add an assert for it, and
fix callers that were passing 'unsorted' instead.
Add tests that we make the right get_record_stream call based
on the value of _fetch_uses_deltas.
Fix the fetch request for signatures.
The proper parameter passed is 'unordered' add an assert for it, and
fix callers that were passing 'unsorted' instead.
Add tests that we make the right get_record_stream call based
on the value of _fetch_uses_deltas.
Fix the fetch request for signatures.
- files added:
- COPYING.txt
- bzrlib/benchmarks/tree_creator
- bzrlib/help_topics
- bzrlib/help_topics/en
- bzrlib/repofmt
- bzrlib/smart
- bzrlib/tests/commands
- bzrlib/tests/intertree_implementations
- bzrlib/tests/inventory_implementations
- bzrlib/tests/per_lock
- bzrlib/tests/per_repository_reference
- bzrlib/tests/tree_implementations
- bzrlib/util/tests
- bzrlib/version_info_formats
- doc/developers
- doc/en
- doc/en/admin-guide
- doc/en/developer-guide
- doc/en/mini-tutorial
- doc/en/quick-reference
- doc/en/release-notes
- doc/en/tutorials
- doc/en/user-guide
- doc/en/user-guide/images
- doc/en/user-reference
- doc/es
- doc/es/guia-desarrollador
- doc/es/guia-usuario
- doc/es/mini-tutorial
- doc/es/notas-version
- doc/es/referencia
- doc/es/referencia-rapida
- man1
- tools/win32
- files removed:
- NEWS.developers
- bzrlib/bundle/old
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/util/urlgrabber
- files renamed:
- bzrlib/patiencediff.py => bzrlib/_patiencediff_py.py
- bzrlib/bundle/read_bundle.py => bzrlib/bundle/bundle_data.py
- bzrlib/bundle/serializer/v07.py => bzrlib/bundle/serializer/v08.py
- bzrlib/graph.py => bzrlib/deprecated_graph.py
- doc/configuration.txt => bzrlib/help_topics/en/configuration.txt
- bzrlib/tests/HTTPTestUtil.py => bzrlib/tests/http_utils.py
- bzrlib/tests/test_revprops.py => bzrlib/tests/repository_implementations/test_revision.py
- bzrlib/tests/test_command.py => bzrlib/tests/test_commands.py
- bzrlib/tests/test_graph.py => bzrlib/tests/test_deprecated_graph.py
- bzrlib/tests/test_revisionnamespaces.py => bzrlib/tests/test_revisionspec.py
- bzrlib/win32console.py => bzrlib/win32utils.py
- bzrlib/xml5.py => bzrlib/xml8.py
- HACKING => doc/developers/HACKING.txt
- doc/tutorial.txt => doc/en/tutorials/tutorial.txt
- doc/plugins.txt => doc/en/user-guide/plugins.txt
- doc/setting_up_email.txt => doc/en/user-guide/setting_up_email.txt
- doc/specifying_revisions.txt => doc/en/user-guide/specifying_revisions.txt
- doc/using_aliases.txt => doc/en/user-guide/using_aliases.txt
- files modified:
- .bzrignore
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
7
#
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
16
17
from binascii import hexlify
18
from copy import deepcopy
19
17
from cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
20
21
import re
21
22
import time
22
from unittest import TestSuite
23
24
import bzrlib.bzrdir as bzrdir
23
24
from bzrlib import (
25
bzrdir,
26
check,
27
debug,
28
errors,
29
generate_ids,
30
gpg,
31
graph,
32
lazy_regex,
33
lockable_files,
34
lockdir,
35
lru_cache,
36
osutils,
37
registry,
38
remote,
39
revision as _mod_revision,
40
symbol_versioning,
41
transactions,
42
tsort,
43
ui,
44
)
45
from bzrlib.bundle import serializer
46
from bzrlib.revisiontree import RevisionTree
47
from bzrlib.store.versioned import VersionedFileStore
48
from bzrlib.store.text import TextStore
49
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
51
""")
52
25
53
from bzrlib.decorators import needs_read_lock, needs_write_lock
26
import bzrlib.errors as errors
27
from bzrlib.errors import InvalidRevisionId
28
import bzrlib.gpg as gpg
29
from bzrlib.graph import Graph
30
54
from bzrlib.inter import InterObject
31
from bzrlib.inventory import Inventory
32
from bzrlib.knit import KnitVersionedFile, KnitPlainFactory
33
from bzrlib.lockable_files import LockableFiles, TransportLock
34
from bzrlib.lockdir import LockDir
35
from bzrlib.osutils import (safe_unicode, rand_bytes, compact_date,
36
local_time_offset)
37
from bzrlib.revision import NULL_REVISION, Revision
38
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
39
from bzrlib.store.text import TextStore
40
from bzrlib.symbol_versioning import *
41
from bzrlib.trace import mutter, note
42
from bzrlib.tree import RevisionTree
43
from bzrlib.tsort import topo_sort
44
from bzrlib.testament import Testament
45
from bzrlib.tree import EmptyTree
46
import bzrlib.ui
47
from bzrlib.weave import WeaveFile
48
import bzrlib.xml5
49
55
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
56
from bzrlib.symbol_versioning import (
57
deprecated_method,
58
one_one,
59
one_two,
60
one_three,
61
one_six,
62
)
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
64
65
66
# Old formats display a warning, but only once
67
_deprecation_warning_done = False
68
69
70
class CommitBuilder(object):
71
"""Provides an interface to build up a commit.
72
73
This allows describing a tree to be committed without needing to
74
know the internals of the format of the repository.
75
"""
76
77
# all clients should supply tree roots.
78
record_root_entry = True
79
# the default CommitBuilder does not manage trees whose root is versioned.
80
_versioned_root = False
81
82
def __init__(self, repository, parents, config, timestamp=None,
83
timezone=None, committer=None, revprops=None,
84
revision_id=None):
85
"""Initiate a CommitBuilder.
86
87
:param repository: Repository to commit to.
88
:param parents: Revision ids of the parents of the new revision.
89
:param config: Configuration to use.
90
:param timestamp: Optional timestamp recorded for commit.
91
:param timezone: Optional timezone for timestamp.
92
:param committer: Optional committer to set for commit.
93
:param revprops: Optional dictionary of revision properties.
94
:param revision_id: Optional revision id.
95
"""
96
self._config = config
97
98
if committer is None:
99
self._committer = self._config.username()
100
else:
101
self._committer = committer
102
103
self.new_inventory = Inventory(None)
104
self._new_revision_id = revision_id
105
self.parents = parents
106
self.repository = repository
107
108
self._revprops = {}
109
if revprops is not None:
110
self._revprops.update(revprops)
111
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
116
117
if timezone is None:
118
self._timezone = osutils.local_time_offset()
119
else:
120
self._timezone = int(timezone)
121
122
self._generate_revision_if_needed()
123
self.__heads = graph.HeadsCache(repository.get_graph()).heads
124
125
def commit(self, message):
126
"""Make the actual commit.
127
128
:return: The revision id of the recorded revision.
129
"""
130
rev = _mod_revision.Revision(
131
timestamp=self._timestamp,
132
timezone=self._timezone,
133
committer=self._committer,
134
message=message,
135
inventory_sha1=self.inv_sha1,
136
revision_id=self._new_revision_id,
137
properties=self._revprops)
138
rev.parent_ids = self.parents
139
self.repository.add_revision(self._new_revision_id, rev,
140
self.new_inventory, self._config)
141
self.repository.commit_write_group()
142
return self._new_revision_id
143
144
def abort(self):
145
"""Abort the commit that is being built.
146
"""
147
self.repository.abort_write_group()
148
149
def revision_tree(self):
150
"""Return the tree that was just committed.
151
152
After calling commit() this can be called to get a RevisionTree
153
representing the newly committed tree. This is preferred to
154
calling Repository.revision_tree() because that may require
155
deserializing the inventory, while we already have a copy in
156
memory.
157
"""
158
return RevisionTree(self.repository, self.new_inventory,
159
self._new_revision_id)
160
161
def finish_inventory(self):
162
"""Tell the builder that the inventory is finished."""
163
if self.new_inventory.root is None:
164
raise AssertionError('Root entry should be supplied to'
165
' record_entry_contents, as of bzr 0.10.')
166
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
167
self.new_inventory.revision_id = self._new_revision_id
168
self.inv_sha1 = self.repository.add_inventory(
169
self._new_revision_id,
170
self.new_inventory,
171
self.parents
172
)
173
174
def _gen_revision_id(self):
175
"""Return new revision-id."""
176
return generate_ids.gen_revision_id(self._config.username(),
177
self._timestamp)
178
179
def _generate_revision_if_needed(self):
180
"""Create a revision id if None was supplied.
181
182
If the repository can not support user-specified revision ids
183
they should override this function and raise CannotSetRevisionId
184
if _new_revision_id is not None.
185
186
:raises: CannotSetRevisionId
187
"""
188
if self._new_revision_id is None:
189
self._new_revision_id = self._gen_revision_id()
190
self.random_revid = True
191
else:
192
self.random_revid = False
193
194
def _heads(self, file_id, revision_ids):
195
"""Calculate the graph heads for revision_ids in the graph of file_id.
196
197
This can use either a per-file graph or a global revision graph as we
198
have an identity relationship between the two graphs.
199
"""
200
return self.__heads(revision_ids)
201
202
def _check_root(self, ie, parent_invs, tree):
203
"""Helper for record_entry_contents.
204
205
:param ie: An entry being added.
206
:param parent_invs: The inventories of the parent revisions of the
207
commit.
208
:param tree: The tree that is being committed.
209
"""
210
# In this revision format, root entries have no knit or weave When
211
# serializing out to disk and back in root.revision is always
212
# _new_revision_id
213
ie.revision = self._new_revision_id
214
215
def _get_delta(self, ie, basis_inv, path):
216
"""Get a delta against the basis inventory for ie."""
217
if ie.file_id not in basis_inv:
218
# add
219
return (None, path, ie.file_id, ie)
220
elif ie != basis_inv[ie.file_id]:
221
# common but altered
222
# TODO: avoid tis id2path call.
223
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
224
else:
225
# common, unaltered
226
return None
227
228
def record_entry_contents(self, ie, parent_invs, path, tree,
229
content_summary):
230
"""Record the content of ie from tree into the commit if needed.
231
232
Side effect: sets ie.revision when unchanged
233
234
:param ie: An inventory entry present in the commit.
235
:param parent_invs: The inventories of the parent revisions of the
236
commit.
237
:param path: The path the entry is at in the tree.
238
:param tree: The tree which contains this entry and should be used to
239
obtain content.
240
:param content_summary: Summary data from the tree about the paths
241
content - stat, length, exec, sha/link target. This is only
242
accessed when the entry has a revision of None - that is when it is
243
a candidate to commit.
244
:return: A tuple (change_delta, version_recorded). change_delta is
245
an inventory_delta change for this entry against the basis tree of
246
the commit, or None if no change occured against the basis tree.
247
version_recorded is True if a new version of the entry has been
248
recorded. For instance, committing a merge where a file was only
249
changed on the other side will return (delta, False).
250
"""
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
257
else:
258
# ie is carried over from a prior commit
259
kind = ie.kind
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
268
269
# TODO: slow, take it out of the inner loop.
270
try:
271
basis_inv = parent_invs[0]
272
except IndexError:
273
basis_inv = Inventory(root_id=None)
274
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
286
ie.file_id, ie)
287
else:
288
# add
289
delta = (None, path, ie.file_id, ie)
290
return delta, False
291
else:
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
294
# appropriate.
295
return None, (ie.revision == self._new_revision_id)
296
# XXX: Friction: parent_candidates should return a list not a dict
297
# so that we don't have to walk the inventories again.
298
parent_candiate_entries = ie.parent_candidates(parent_invs)
299
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
300
heads = []
301
for inv in parent_invs:
302
if ie.file_id in inv:
303
old_rev = inv[ie.file_id].revision
304
if old_rev in head_set:
305
heads.append(inv[ie.file_id].revision)
306
head_set.remove(inv[ie.file_id].revision)
307
308
store = False
309
# now we check to see if we need to write a new record to the
310
# file-graph.
311
# We write a new entry unless there is one head to the ancestors, and
312
# the kind-derived content is unchanged.
313
314
# Cheapest check first: no ancestors, or more the one head in the
315
# ancestors, we write a new node.
316
if len(heads) != 1:
317
store = True
318
if not store:
319
# There is a single head, look it up for comparison
320
parent_entry = parent_candiate_entries[heads[0]]
321
# if the non-content specific data has changed, we'll be writing a
322
# node:
323
if (parent_entry.parent_id != ie.parent_id or
324
parent_entry.name != ie.name):
325
store = True
326
# now we need to do content specific checks:
327
if not store:
328
# if the kind changed the content obviously has
329
if kind != parent_entry.kind:
330
store = True
331
if kind == 'file':
332
if content_summary[2] is None:
333
raise ValueError("Files must not have executable = None")
334
if not store:
335
if (# if the file length changed we have to store:
336
parent_entry.text_size != content_summary[1] or
337
# if the exec bit has changed we have to store:
338
parent_entry.executable != content_summary[2]):
339
store = True
340
elif parent_entry.text_sha1 == content_summary[3]:
341
# all meta and content is unchanged (using a hash cache
342
# hit to check the sha)
343
ie.revision = parent_entry.revision
344
ie.text_size = parent_entry.text_size
345
ie.text_sha1 = parent_entry.text_sha1
346
ie.executable = parent_entry.executable
347
return self._get_delta(ie, basis_inv, path), False
348
else:
349
# Either there is only a hash change(no hash cache entry,
350
# or same size content change), or there is no change on
351
# this file at all.
352
# Provide the parent's hash to the store layer, so that the
353
# content is unchanged we will not store a new node.
354
nostore_sha = parent_entry.text_sha1
355
if store:
356
# We want to record a new node regardless of the presence or
357
# absence of a content change in the file.
358
nostore_sha = None
359
ie.executable = content_summary[2]
360
lines = tree.get_file(ie.file_id, path).readlines()
361
try:
362
ie.text_sha1, ie.text_size = self._add_text_to_weave(
363
ie.file_id, lines, heads, nostore_sha)
364
except errors.ExistingContent:
365
# Turns out that the file content was unchanged, and we were
366
# only going to store a new node if it was changed. Carry over
367
# the entry.
368
ie.revision = parent_entry.revision
369
ie.text_size = parent_entry.text_size
370
ie.text_sha1 = parent_entry.text_sha1
371
ie.executable = parent_entry.executable
372
return self._get_delta(ie, basis_inv, path), False
373
elif kind == 'directory':
374
if not store:
375
# all data is meta here, nothing specific to directory, so
376
# carry over:
377
ie.revision = parent_entry.revision
378
return self._get_delta(ie, basis_inv, path), False
379
lines = []
380
self._add_text_to_weave(ie.file_id, lines, heads, None)
381
elif kind == 'symlink':
382
current_link_target = content_summary[3]
383
if not store:
384
# symlink target is not generic metadata, check if it has
385
# changed.
386
if current_link_target != parent_entry.symlink_target:
387
store = True
388
if not store:
389
# unchanged, carry over.
390
ie.revision = parent_entry.revision
391
ie.symlink_target = parent_entry.symlink_target
392
return self._get_delta(ie, basis_inv, path), False
393
ie.symlink_target = current_link_target
394
lines = []
395
self._add_text_to_weave(ie.file_id, lines, heads, None)
396
elif kind == 'tree-reference':
397
if not store:
398
if content_summary[3] != parent_entry.reference_revision:
399
store = True
400
if not store:
401
# unchanged, carry over.
402
ie.reference_revision = parent_entry.reference_revision
403
ie.revision = parent_entry.revision
404
return self._get_delta(ie, basis_inv, path), False
405
ie.reference_revision = content_summary[3]
406
lines = []
407
self._add_text_to_weave(ie.file_id, lines, heads, None)
408
else:
409
raise NotImplementedError('unknown kind')
410
ie.revision = self._new_revision_id
411
return self._get_delta(ie, basis_inv, path), True
412
413
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
414
# Note: as we read the content directly from the tree, we know its not
415
# been turned into unicode or badly split - but a broken tree
416
# implementation could give us bad output from readlines() so this is
417
# not a guarantee of safety. What would be better is always checking
418
# the content during test suite execution. RBC 20070912
419
parent_keys = tuple((file_id, parent) for parent in parents)
420
return self.repository.texts.add_lines(
421
(file_id, self._new_revision_id), parent_keys, new_lines,
422
nostore_sha=nostore_sha, random_id=self.random_revid,
423
check_content=False)[0:2]
424
425
426
class RootCommitBuilder(CommitBuilder):
427
"""This commitbuilder actually records the root id"""
428
429
# the root entry gets versioned properly by this builder.
430
_versioned_root = True
431
432
def _check_root(self, ie, parent_invs, tree):
433
"""Helper for record_entry_contents.
434
435
:param ie: An entry being added.
436
:param parent_invs: The inventories of the parent revisions of the
437
commit.
438
:param tree: The tree that is being committed.
439
"""
440
441
442
######################################################################
443
# Repositories
50
444
51
445
class Repository(object):
52
446
"""Repository holding history for one or more branches.
55
449
revisions and file history. It's normally accessed only by the Branch,
56
450
which views a particular line of development through that history.
57
451
58
The Repository builds on top of Stores and a Transport, which respectively
59
describe the disk data format and the way of accessing the (possibly
452
The Repository builds on top of some byte storage facilies (the revisions,
453
signatures, inventories and texts attributes) and a Transport, which
454
respectively provide byte storage and a means to access the (possibly
60
455
remote) disk.
456
457
The byte storage facilities are addressed via tuples, which we refer to
458
as 'keys' throughout the code base. Revision_keys, inventory_keys and
459
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
460
(file_id, revision_id). We use this interface because it allows low
461
friction with the underlying code that implements disk indices, network
462
encoding and other parts of bzrlib.
463
464
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
465
the serialised revisions for the repository. This can be used to obtain
466
revision graph information or to access raw serialised revisions.
467
The result of trying to insert data into the repository via this store
468
is undefined: it should be considered read-only except for implementors
469
of repositories.
470
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
471
the serialised signatures for the repository. This can be used to
472
obtain access to raw serialised signatures. The result of trying to
473
insert data into the repository via this store is undefined: it should
474
be considered read-only except for implementors of repositories.
475
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
476
the serialised inventories for the repository. This can be used to
477
obtain unserialised inventories. The result of trying to insert data
478
into the repository via this store is undefined: it should be
479
considered read-only except for implementors of repositories.
480
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
481
texts of files and directories for the repository. This can be used to
482
obtain file texts or file graphs. Note that Repository.iter_file_bytes
483
is usually a better interface for accessing file texts.
484
The result of trying to insert data into the repository via this store
485
is undefined: it should be considered read-only except for implementors
486
of repositories.
487
:ivar _transport: Transport for file access to repository, typically
488
pointing to .bzr/repository.
61
489
"""
62
490
63
@needs_write_lock
64
def add_inventory(self, revid, inv, parents):
65
"""Add the inventory inv to the repository as revid.
66
67
:param parents: The revision ids of the parents that revid
491
# What class to use for a CommitBuilder. Often its simpler to change this
492
# in a Repository class subclass rather than to override
493
# get_commit_builder.
494
_commit_builder_class = CommitBuilder
495
# The search regex used by xml based repositories to determine what things
496
# where changed in a single commit.
497
_file_ids_altered_regex = lazy_regex.lazy_compile(
498
r'file_id="(?P<file_id>[^"]+)"'
499
r'.* revision="(?P<revision_id>[^"]+)"'
500
)
501
502
def abort_write_group(self):
503
"""Commit the contents accrued within the current write group.
504
505
:seealso: start_write_group.
506
"""
507
if self._write_group is not self.get_transaction():
508
# has an unlock or relock occured ?
509
raise errors.BzrError('mismatched lock context and write group.')
510
self._abort_write_group()
511
self._write_group = None
512
513
def _abort_write_group(self):
514
"""Template method for per-repository write group cleanup.
515
516
This is called during abort before the write group is considered to be
517
finished and should cleanup any internal state accrued during the write
518
group. There is no requirement that data handed to the repository be
519
*not* made available - this is not a rollback - but neither should any
520
attempt be made to ensure that data added is fully commited. Abort is
521
invoked when an error has occured so futher disk or network operations
522
may not be possible or may error and if possible should not be
523
attempted.
524
"""
525
526
def add_fallback_repository(self, repository):
527
"""Add a repository to use for looking up data not held locally.
528
529
:param repository: A repository.
530
"""
531
if not self._format.supports_external_lookups:
532
raise errors.UnstackableRepositoryFormat(self._format, self.base)
533
self._check_fallback_repository(repository)
534
self._fallback_repositories.append(repository)
535
self.texts.add_fallback_versioned_files(repository.texts)
536
self.inventories.add_fallback_versioned_files(repository.inventories)
537
self.revisions.add_fallback_versioned_files(repository.revisions)
538
self.signatures.add_fallback_versioned_files(repository.signatures)
539
540
def _check_fallback_repository(self, repository):
541
"""Check that this repository can fallback to repository safely.
542
543
Raise an error if not.
544
545
:param repository: A repository to fallback to.
546
"""
547
return InterRepository._assert_same_model(self, repository)
548
549
def add_inventory(self, revision_id, inv, parents):
550
"""Add the inventory inv to the repository as revision_id.
551
552
:param parents: The revision ids of the parents that revision_id
68
553
is known to have and are in the repository already.
69
554
70
returns the sha1 of the serialized inventory.
555
:returns: The validator(which is a sha1 digest, though what is sha'd is
556
repository format specific) of the serialized inventory.
71
557
"""
72
assert inv.revision_id is None or inv.revision_id == revid, \
73
"Mismatch between inventory revision" \
74
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
75
inv_text = bzrlib.xml5.serializer_v5.write_inventory_to_string(inv)
76
inv_sha1 = bzrlib.osutils.sha_string(inv_text)
77
inv_vf = self.control_weaves.get_weave('inventory',
78
self.get_transaction())
79
self._inventory_add_lines(inv_vf, revid, parents, bzrlib.osutils.split_lines(inv_text))
80
return inv_sha1
81
82
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
83
final_parents = []
84
for parent in parents:
85
if parent in inv_vf:
86
final_parents.append(parent)
87
88
inv_vf.add_lines(revid, final_parents, lines)
89
90
@needs_write_lock
91
def add_revision(self, rev_id, rev, inv=None, config=None):
92
"""Add rev to the revision store as rev_id.
93
94
:param rev_id: the revision id to use.
558
if not self.is_in_write_group():
559
raise AssertionError("%r not in write group" % (self,))
560
_mod_revision.check_not_reserved_id(revision_id)
561
if not (inv.revision_id is None or inv.revision_id == revision_id):
562
raise AssertionError(
563
"Mismatch between inventory revision"
564
" id and insertion revid (%r, %r)"
565
% (inv.revision_id, revision_id))
566
if inv.root is None:
567
raise AssertionError()
568
inv_lines = self._serialise_inventory_to_lines(inv)
569
return self._inventory_add_lines(revision_id, parents,
570
inv_lines, check_content=False)
571
572
def _inventory_add_lines(self, revision_id, parents, lines,
573
check_content=True):
574
"""Store lines in inv_vf and return the sha1 of the inventory."""
575
parents = [(parent,) for parent in parents]
576
return self.inventories.add_lines((revision_id,), parents, lines,
577
check_content=check_content)[0]
578
579
def add_revision(self, revision_id, rev, inv=None, config=None):
580
"""Add rev to the revision store as revision_id.
581
582
:param revision_id: the revision id to use.
95
583
:param rev: The revision object.
96
584
:param inv: The inventory for the revision. if None, it will be looked
97
585
up in the inventory storer
99
587
If supplied its signature_needed method will be used
100
588
to determine if a signature should be made.
101
589
"""
590
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
591
# rev.parent_ids?
592
_mod_revision.check_not_reserved_id(revision_id)
102
593
if config is not None and config.signature_needed():
103
594
if inv is None:
104
inv = self.get_inventory(rev_id)
595
inv = self.get_inventory(revision_id)
105
596
plaintext = Testament(rev, inv).as_short_text()
106
597
self.store_revision_signature(
107
gpg.GPGStrategy(config), plaintext, rev_id)
108
if not rev_id in self.get_inventory_weave():
598
gpg.GPGStrategy(config), plaintext, revision_id)
599
# check inventory present
600
if not self.inventories.get_parent_map([(revision_id,)]):
109
601
if inv is None:
110
raise errors.WeaveRevisionNotPresent(rev_id,
111
self.get_inventory_weave())
602
raise errors.WeaveRevisionNotPresent(revision_id,
603
self.inventories)
112
604
else:
113
605
# yes, this is not suitable for adding with ghosts.
114
self.add_inventory(rev_id, inv, rev.parent_ids)
115
self._revision_store.add_revision(rev, self.get_transaction())
116
117
@needs_read_lock
118
def _all_possible_ids(self):
119
"""Return all the possible revisions that we could find."""
120
return self.get_inventory_weave().versions()
121
122
@deprecated_method(zero_nine)
606
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
607
rev.parent_ids)
608
else:
609
key = (revision_id,)
610
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
611
self._add_revision(rev)
612
613
def _add_revision(self, revision):
614
text = self._serializer.write_revision_to_string(revision)
615
key = (revision.revision_id,)
616
parents = tuple((parent,) for parent in revision.parent_ids)
617
self.revisions.add_lines(key, parents, osutils.split_lines(text))
618
123
619
def all_revision_ids(self):
124
620
"""Returns a list of all the revision ids in the repository.
125
621
126
This is deprecated because code should generally work on the graph
127
reachable from a particular revision, and ignore any other revisions
128
that might be present. There is no direct replacement method.
622
This is conceptually deprecated because code should generally work on
623
the graph reachable from a particular revision, and ignore any other
624
revisions that might be present. There is no direct replacement
625
method.
129
626
"""
627
if 'evil' in debug.debug_flags:
628
mutter_callsite(2, "all_revision_ids is linear with history.")
130
629
return self._all_revision_ids()
131
630
132
@needs_read_lock
133
631
def _all_revision_ids(self):
134
632
"""Returns a list of all the revision ids in the repository.
135
633
136
634
These are in as much topological order as the underlying store can
137
present: for weaves ghosts may lead to a lack of correctness until
138
the reweave updates the parents list.
635
present.
139
636
"""
140
if self._revision_store.text_store.listable():
141
return self._revision_store.all_revision_ids(self.get_transaction())
142
result = self._all_possible_ids()
143
return self._eliminate_revisions_not_present(result)
637
raise NotImplementedError(self._all_revision_ids)
144
638
145
639
def break_lock(self):
146
640
"""Break a lock if one is present from another instance.
157
651
Returns a set of the present revisions.
158
652
"""
159
653
result = []
160
for id in revision_ids:
161
if self.has_revision(id):
162
result.append(id)
163
return result
654
graph = self.get_graph()
655
parent_map = graph.get_parent_map(revision_ids)
656
# The old API returned a list, should this actually be a set?
657
return parent_map.keys()
164
658
165
659
@staticmethod
166
660
def create(a_bzrdir):
167
661
"""Construct the current default format repository in a_bzrdir."""
168
662
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
169
663
170
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
664
def __init__(self, _format, a_bzrdir, control_files):
171
665
"""instantiate a Repository.
172
666
173
667
:param _format: The format of the repository on disk.
182
676
# the following are part of the public API for Repository:
183
677
self.bzrdir = a_bzrdir
184
678
self.control_files = control_files
185
self._revision_store = _revision_store
186
self.text_store = text_store
187
# backwards compatibility
188
self.weave_store = text_store
679
self._transport = control_files._transport
680
self.base = self._transport.base
681
# for tests
682
self._reconcile_does_inventory_gc = True
683
self._reconcile_fixes_text_parents = False
684
self._reconcile_backsup_inventory = True
189
685
# not right yet - should be more semantically clear ?
190
686
#
191
self.control_store = control_store
192
self.control_weaves = control_store
193
687
# TODO: make sure to construct the right store classes, etc, depending
194
688
# on whether escaping is required.
689
self._warn_if_deprecated()
690
self._write_group = None
691
# Additional places to query for data.
692
self._fallback_repositories = []
693
# What order should fetch operations request streams in?
694
# The default is unordered as that is the cheapest for an origin to
695
# provide.
696
self._fetch_order = 'unordered'
697
# Does this repository use deltas that can be fetched as-deltas ?
698
# (E.g. knits, where the knit deltas can be transplanted intact.
699
# We default to False, which will ensure that enough data to get
700
# a full text out of any fetch stream will be grabbed.
701
self._fetch_uses_deltas = False
702
# Should fetch trigger a reconcile after the fetch? Only needed for
703
# some repository formats that can suffer internal inconsistencies.
704
self._fetch_reconcile = False
195
705
196
706
def __repr__(self):
197
return '%s(%r)' % (self.__class__.__name__,
198
self.bzrdir.transport.base)
707
return '%s(%r)' % (self.__class__.__name__,
708
self.base)
709
710
def has_same_location(self, other):
711
"""Returns a boolean indicating if this repository is at the same
712
location as another repository.
713
714
This might return False even when two repository objects are accessing
715
the same physical repository via different URLs.
716
"""
717
if self.__class__ is not other.__class__:
718
return False
719
return (self._transport.base == other._transport.base)
720
721
def is_in_write_group(self):
722
"""Return True if there is an open write group.
723
724
:seealso: start_write_group.
725
"""
726
return self._write_group is not None
199
727
200
728
def is_locked(self):
201
729
return self.control_files.is_locked()
202
730
203
def lock_write(self):
204
self.control_files.lock_write()
731
def is_write_locked(self):
732
"""Return True if this object is write locked."""
733
return self.is_locked() and self.control_files._lock_mode == 'w'
734
735
def lock_write(self, token=None):
736
"""Lock this repository for writing.
737
738
This causes caching within the repository obejct to start accumlating
739
data during reads, and allows a 'write_group' to be obtained. Write
740
groups must be used for actual data insertion.
741
742
:param token: if this is already locked, then lock_write will fail
743
unless the token matches the existing lock.
744
:returns: a token if this instance supports tokens, otherwise None.
745
:raises TokenLockingNotSupported: when a token is given but this
746
instance doesn't support using token locks.
747
:raises MismatchedToken: if the specified token doesn't match the token
748
of the existing lock.
749
:seealso: start_write_group.
750
751
A token should be passed in if you know that you have locked the object
752
some other way, and need to synchronise this object's state with that
753
fact.
754
755
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
756
"""
757
result = self.control_files.lock_write(token=token)
758
for repo in self._fallback_repositories:
759
# Writes don't affect fallback repos
760
repo.lock_read()
761
self._refresh_data()
762
return result
205
763
206
764
def lock_read(self):
207
765
self.control_files.lock_read()
766
for repo in self._fallback_repositories:
767
repo.lock_read()
768
self._refresh_data()
208
769
209
770
def get_physical_lock_status(self):
210
771
return self.control_files.get_physical_lock_status()
211
772
212
@needs_read_lock
213
def missing_revision_ids(self, other, revision_id=None):
214
"""Return the revision ids that other has that this does not.
215
216
These are returned in topological order.
217
218
revision_id: only return revision ids included by revision_id.
219
"""
220
return InterRepository.get(other, self).missing_revision_ids(revision_id)
773
def leave_lock_in_place(self):
774
"""Tell this repository not to release the physical lock when this
775
object is unlocked.
776
777
If lock_write doesn't return a token, then this method is not supported.
778
"""
779
self.control_files.leave_in_place()
780
781
def dont_leave_lock_in_place(self):
782
"""Tell this repository to release the physical lock when this
783
object is unlocked, even if it didn't originally acquire it.
784
785
If lock_write doesn't return a token, then this method is not supported.
786
"""
787
self.control_files.dont_leave_in_place()
788
789
@needs_read_lock
790
def gather_stats(self, revid=None, committers=None):
791
"""Gather statistics from a revision id.
792
793
:param revid: The revision id to gather statistics from, if None, then
794
no revision specific statistics are gathered.
795
:param committers: Optional parameter controlling whether to grab
796
a count of committers from the revision specific statistics.
797
:return: A dictionary of statistics. Currently this contains:
798
committers: The number of committers if requested.
799
firstrev: A tuple with timestamp, timezone for the penultimate left
800
most ancestor of revid, if revid is not the NULL_REVISION.
801
latestrev: A tuple with timestamp, timezone for revid, if revid is
802
not the NULL_REVISION.
803
revisions: The total revision count in the repository.
804
size: An estimate disk size of the repository in bytes.
805
"""
806
result = {}
807
if revid and committers:
808
result['committers'] = 0
809
if revid and revid != _mod_revision.NULL_REVISION:
810
if committers:
811
all_committers = set()
812
revisions = self.get_ancestry(revid)
813
# pop the leading None
814
revisions.pop(0)
815
first_revision = None
816
if not committers:
817
# ignore the revisions in the middle - just grab first and last
818
revisions = revisions[0], revisions[-1]
819
for revision in self.get_revisions(revisions):
820
if not first_revision:
821
first_revision = revision
822
if committers:
823
all_committers.add(revision.committer)
824
last_revision = revision
825
if committers:
826
result['committers'] = len(all_committers)
827
result['firstrev'] = (first_revision.timestamp,
828
first_revision.timezone)
829
result['latestrev'] = (last_revision.timestamp,
830
last_revision.timezone)
831
832
# now gather global repository information
833
# XXX: This is available for many repos regardless of listability.
834
if self.bzrdir.root_transport.listable():
835
# XXX: do we want to __define len__() ?
836
# Maybe the versionedfiles object should provide a different
837
# method to get the number of keys.
838
result['revisions'] = len(self.revisions.keys())
839
# result['size'] = t
840
return result
841
842
def find_branches(self, using=False):
843
"""Find branches underneath this repository.
844
845
This will include branches inside other branches.
846
847
:param using: If True, list only branches using this repository.
848
"""
849
if using and not self.is_shared():
850
try:
851
return [self.bzrdir.open_branch()]
852
except errors.NotBranchError:
853
return []
854
class Evaluator(object):
855
856
def __init__(self):
857
self.first_call = True
858
859
def __call__(self, bzrdir):
860
# On the first call, the parameter is always the bzrdir
861
# containing the current repo.
862
if not self.first_call:
863
try:
864
repository = bzrdir.open_repository()
865
except errors.NoRepositoryPresent:
866
pass
867
else:
868
return False, (None, repository)
869
self.first_call = False
870
try:
871
value = (bzrdir.open_branch(), None)
872
except errors.NotBranchError:
873
value = (None, None)
874
return True, value
875
876
branches = []
877
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
878
self.bzrdir.root_transport, evaluate=Evaluator()):
879
if branch is not None:
880
branches.append(branch)
881
if not using and repository is not None:
882
branches.extend(repository.find_branches())
883
return branches
884
885
@needs_read_lock
886
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
887
"""Return the revision ids that other has that this does not.
888
889
These are returned in topological order.
890
891
revision_id: only return revision ids included by revision_id.
892
"""
893
return InterRepository.get(other, self).search_missing_revision_ids(
894
revision_id, find_ghosts)
895
896
@deprecated_method(one_two)
897
@needs_read_lock
898
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
899
"""Return the revision ids that other has that this does not.
900
901
These are returned in topological order.
902
903
revision_id: only return revision ids included by revision_id.
904
"""
905
keys = self.search_missing_revision_ids(
906
other, revision_id, find_ghosts).get_keys()
907
other.lock_read()
908
try:
909
parents = other.get_graph().get_parent_map(keys)
910
finally:
911
other.unlock()
912
return tsort.topo_sort(parents)
221
913
222
914
@staticmethod
223
915
def open(base):
226
918
For instance, if the repository is at URL/.bzr/repository,
227
919
Repository.open(URL) -> a Repository instance.
228
920
"""
229
control = bzrlib.bzrdir.BzrDir.open(base)
921
control = bzrdir.BzrDir.open(base)
230
922
return control.open_repository()
231
923
232
def copy_content_into(self, destination, revision_id=None, basis=None):
924
def copy_content_into(self, destination, revision_id=None):
233
925
"""Make a complete copy of the content in self into destination.
234
926
235
927
This is a destructive operation! Do not use it on existing
236
928
repositories.
237
929
"""
238
return InterRepository.get(self, destination).copy_content(revision_id, basis)
239
240
def fetch(self, source, revision_id=None, pb=None):
930
return InterRepository.get(self, destination).copy_content(revision_id)
931
932
def commit_write_group(self):
933
"""Commit the contents accrued within the current write group.
934
935
:seealso: start_write_group.
936
"""
937
if self._write_group is not self.get_transaction():
938
# has an unlock or relock occured ?
939
raise errors.BzrError('mismatched lock context %r and '
940
'write group %r.' %
941
(self.get_transaction(), self._write_group))
942
self._commit_write_group()
943
self._write_group = None
944
945
def _commit_write_group(self):
946
"""Template method for per-repository write group cleanup.
947
948
This is called before the write group is considered to be
949
finished and should ensure that all data handed to the repository
950
for writing during the write group is safely committed (to the
951
extent possible considering file system caching etc).
952
"""
953
954
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
241
955
"""Fetch the content required to construct revision_id from source.
242
956
243
957
If revision_id is None all content is copied.
958
:param find_ghosts: Find and copy revisions in the source that are
959
ghosts in the target (and not reachable directly by walking out to
960
the first-present revision in target from revision_id).
244
961
"""
245
return InterRepository.get(source, self).fetch(revision_id=revision_id,
246
pb=pb)
247
248
def get_commit_builder(self, branch, parents, config, timestamp=None,
249
timezone=None, committer=None, revprops=None,
962
# fast path same-url fetch operations
963
if self.has_same_location(source):
964
# check that last_revision is in 'from' and then return a
965
# no-operation.
966
if (revision_id is not None and
967
not _mod_revision.is_null(revision_id)):
968
self.get_revision(revision_id)
969
return 0, []
970
# if there is no specific appropriate InterRepository, this will get
971
# the InterRepository base class, which raises an
972
# IncompatibleRepositories when asked to fetch.
973
inter = InterRepository.get(source, self)
974
return inter.fetch(revision_id=revision_id, pb=pb,
975
find_ghosts=find_ghosts)
976
977
def create_bundle(self, target, base, fileobj, format=None):
978
return serializer.write_bundle(self, target, base, fileobj, format)
979
980
def get_commit_builder(self, branch, parents, config, timestamp=None,
981
timezone=None, committer=None, revprops=None,
250
982
revision_id=None):
251
983
"""Obtain a CommitBuilder for this repository.
252
984
259
991
:param revprops: Optional dictionary of revision properties.
260
992
:param revision_id: Optional revision id.
261
993
"""
262
return CommitBuilder(self, parents, config, timestamp, timezone,
263
committer, revprops, revision_id)
994
result = self._commit_builder_class(self, parents, config,
995
timestamp, timezone, committer, revprops, revision_id)
996
self.start_write_group()
997
return result
264
998
265
999
def unlock(self):
1000
if (self.control_files._lock_count == 1 and
1001
self.control_files._lock_mode == 'w'):
1002
if self._write_group is not None:
1003
self.abort_write_group()
1004
self.control_files.unlock()
1005
raise errors.BzrError(
1006
'Must end write groups before releasing write locks.')
266
1007
self.control_files.unlock()
1008
for repo in self._fallback_repositories:
1009
repo.unlock()
267
1010
268
1011
@needs_read_lock
269
def clone(self, a_bzrdir, revision_id=None, basis=None):
1012
def clone(self, a_bzrdir, revision_id=None):
270
1013
"""Clone this repository into a_bzrdir using the current format.
271
1014
272
1015
Currently no check is made that the format of this repository and
273
1016
the bzrdir format are compatible. FIXME RBC 20060201.
274
"""
1017
1018
:return: The newly created destination repository.
1019
"""
1020
# TODO: deprecate after 0.16; cloning this with all its settings is
1021
# probably not very useful -- mbp 20070423
1022
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1023
self.copy_content_into(dest_repo, revision_id)
1024
return dest_repo
1025
1026
def start_write_group(self):
1027
"""Start a write group in the repository.
1028
1029
Write groups are used by repositories which do not have a 1:1 mapping
1030
between file ids and backend store to manage the insertion of data from
1031
both fetch and commit operations.
1032
1033
A write lock is required around the start_write_group/commit_write_group
1034
for the support of lock-requiring repository formats.
1035
1036
One can only insert data into a repository inside a write group.
1037
1038
:return: None.
1039
"""
1040
if not self.is_write_locked():
1041
raise errors.NotWriteLocked(self)
1042
if self._write_group:
1043
raise errors.BzrError('already in a write group')
1044
self._start_write_group()
1045
# so we can detect unlock/relock - the write group is now entered.
1046
self._write_group = self.get_transaction()
1047
1048
def _start_write_group(self):
1049
"""Template method for per-repository write group startup.
1050
1051
This is called before the write group is considered to be
1052
entered.
1053
"""
1054
1055
@needs_read_lock
1056
def sprout(self, to_bzrdir, revision_id=None):
1057
"""Create a descendent repository for new development.
1058
1059
Unlike clone, this does not copy the settings of the repository.
1060
"""
1061
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1062
dest_repo.fetch(self, revision_id=revision_id)
1063
return dest_repo
1064
1065
def _create_sprouting_repo(self, a_bzrdir, shared):
275
1066
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
276
1067
# use target default format.
277
result = a_bzrdir.create_repository()
278
# FIXME RBC 20060209 split out the repository type to avoid this check ?
279
elif isinstance(a_bzrdir._format,
280
(bzrlib.bzrdir.BzrDirFormat4,
281
bzrlib.bzrdir.BzrDirFormat5,
282
bzrlib.bzrdir.BzrDirFormat6)):
283
result = a_bzrdir.open_repository()
1068
dest_repo = a_bzrdir.create_repository()
284
1069
else:
285
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
286
self.copy_content_into(result, revision_id, basis)
287
return result
1070
# Most control formats need the repository to be specifically
1071
# created, but on some old all-in-one formats it's not needed
1072
try:
1073
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1074
except errors.UninitializableFormat:
1075
dest_repo = a_bzrdir.open_repository()
1076
return dest_repo
288
1077
289
1078
@needs_read_lock
290
1079
def has_revision(self, revision_id):
291
1080
"""True if this repository has a copy of the revision."""
292
return self._revision_store.has_revision_id(revision_id,
293
self.get_transaction())
1081
return revision_id in self.has_revisions((revision_id,))
1082
1083
@needs_read_lock
1084
def has_revisions(self, revision_ids):
1085
"""Probe to find out the presence of multiple revisions.
1086
1087
:param revision_ids: An iterable of revision_ids.
1088
:return: A set of the revision_ids that were present.
1089
"""
1090
parent_map = self.revisions.get_parent_map(
1091
[(rev_id,) for rev_id in revision_ids])
1092
result = set()
1093
if _mod_revision.NULL_REVISION in revision_ids:
1094
result.add(_mod_revision.NULL_REVISION)
1095
result.update([key[0] for key in parent_map])
1096
return result
1097
1098
@needs_read_lock
1099
def get_revision(self, revision_id):
1100
"""Return the Revision object for a named revision."""
1101
return self.get_revisions([revision_id])[0]
294
1102
295
1103
@needs_read_lock
296
1104
def get_revision_reconcile(self, revision_id):
301
1109
be used by reconcile, or reconcile-alike commands that are correcting
302
1110
or testing the revision graph.
303
1111
"""
304
if not revision_id or not isinstance(revision_id, basestring):
305
raise InvalidRevisionId(revision_id=revision_id, branch=self)
306
return self._revision_store.get_revisions([revision_id],
307
self.get_transaction())[0]
1112
return self._get_revisions([revision_id])[0]
1113
308
1114
@needs_read_lock
309
1115
def get_revisions(self, revision_ids):
310
return self._revision_store.get_revisions(revision_ids,
311
self.get_transaction())
1116
"""Get many revisions at once."""
1117
return self._get_revisions(revision_ids)
1118
1119
@needs_read_lock
1120
def _get_revisions(self, revision_ids):
1121
"""Core work logic to get many revisions without sanity checks."""
1122
for rev_id in revision_ids:
1123
if not rev_id or not isinstance(rev_id, basestring):
1124
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1125
keys = [(key,) for key in revision_ids]
1126
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1127
revs = {}
1128
for record in stream:
1129
if record.storage_kind == 'absent':
1130
raise errors.NoSuchRevision(self, record.key[0])
1131
text = record.get_bytes_as('fulltext')
1132
rev = self._serializer.read_revision_from_string(text)
1133
revs[record.key[0]] = rev
1134
return [revs[revid] for revid in revision_ids]
312
1135
313
1136
@needs_read_lock
314
1137
def get_revision_xml(self, revision_id):
315
rev = self.get_revision(revision_id)
1138
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1139
# would have already do it.
1140
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1141
rev = self.get_revision(revision_id)
316
1142
rev_tmp = StringIO()
317
1143
# the current serializer..
318
self._revision_store._serializer.write_revision(rev, rev_tmp)
1144
self._serializer.write_revision(rev, rev_tmp)
319
1145
rev_tmp.seek(0)
320
1146
return rev_tmp.getvalue()
321
1147
1148
def get_deltas_for_revisions(self, revisions):
1149
"""Produce a generator of revision deltas.
1150
1151
Note that the input is a sequence of REVISIONS, not revision_ids.
1152
Trees will be held in memory until the generator exits.
1153
Each delta is relative to the revision's lefthand predecessor.
1154
"""
1155
required_trees = set()
1156
for revision in revisions:
1157
required_trees.add(revision.revision_id)
1158
required_trees.update(revision.parent_ids[:1])
1159
trees = dict((t.get_revision_id(), t) for
1160
t in self.revision_trees(required_trees))
1161
for revision in revisions:
1162
if not revision.parent_ids:
1163
old_tree = self.revision_tree(None)
1164
else:
1165
old_tree = trees[revision.parent_ids[0]]
1166
yield trees[revision.revision_id].changes_from(old_tree)
1167
322
1168
@needs_read_lock
323
def get_revision(self, revision_id):
324
"""Return the Revision object for a named revision"""
325
r = self.get_revision_reconcile(revision_id)
326
# weave corruption can lead to absent revision markers that should be
327
# present.
328
# the following test is reasonably cheap (it needs a single weave read)
329
# and the weave is cached in read transactions. In write transactions
330
# it is not cached but typically we only read a small number of
331
# revisions. For knits when they are introduced we will probably want
332
# to ensure that caching write transactions are in use.
333
inv = self.get_inventory_weave()
334
self._check_revision_parents(r, inv)
335
return r
1169
def get_revision_delta(self, revision_id):
1170
"""Return the delta for one revision.
336
1171
337
def _check_revision_parents(self, revision, inventory):
338
"""Private to Repository and Fetch.
339
340
This checks the parentage of revision in an inventory weave for
341
consistency and is only applicable to inventory-weave-for-ancestry
342
using repository formats & fetchers.
1172
The delta is relative to the left-hand predecessor of the
1173
revision.
343
1174
"""
344
weave_parents = inventory.get_parents(revision.revision_id)
345
weave_names = inventory.versions()
346
for parent_id in revision.parent_ids:
347
if parent_id in weave_names:
348
# this parent must not be a ghost.
349
if not parent_id in weave_parents:
350
# but it is a ghost
351
raise errors.CorruptRepository(self)
1175
r = self.get_revision(revision_id)
1176
return list(self.get_deltas_for_revisions([r]))[0]
352
1177
353
1178
@needs_write_lock
354
1179
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
355
1180
signature = gpg_strategy.sign(plaintext)
356
self._revision_store.add_revision_signature_text(revision_id,
357
signature,
358
self.get_transaction())
359
360
def fileids_altered_by_revision_ids(self, revision_ids):
361
"""Find the file ids and versions affected by revisions.
362
363
:param revisions: an iterable containing revision ids.
364
:return: a dictionary mapping altered file-ids to an iterable of
1181
self.add_signature_text(revision_id, signature)
1182
1183
@needs_write_lock
1184
def add_signature_text(self, revision_id, signature):
1185
self.signatures.add_lines((revision_id,), (),
1186
osutils.split_lines(signature))
1187
1188
def find_text_key_references(self):
1189
"""Find the text key references within the repository.
1190
1191
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
365
1192
revision_ids. Each altered file-ids has the exact revision_ids that
366
1193
altered it listed explicitly.
367
"""
368
assert isinstance(self._format, (RepositoryFormat5,
369
RepositoryFormat6,
370
RepositoryFormat7,
371
RepositoryFormatKnit1)), \
372
("fileids_altered_by_revision_ids only supported for branches "
373
"which store inventory as unnested xml, not on %r" % self)
374
selected_revision_ids = set(revision_ids)
375
w = self.get_inventory_weave()
1194
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1195
to whether they were referred to by the inventory of the
1196
revision_id that they contain. The inventory texts from all present
1197
revision ids are assessed to generate this report.
1198
"""
1199
revision_keys = self.revisions.keys()
1200
w = self.inventories
1201
pb = ui.ui_factory.nested_progress_bar()
1202
try:
1203
return self._find_text_key_references_from_xml_inventory_lines(
1204
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1205
finally:
1206
pb.finished()
1207
1208
def _find_text_key_references_from_xml_inventory_lines(self,
1209
line_iterator):
1210
"""Core routine for extracting references to texts from inventories.
1211
1212
This performs the translation of xml lines to revision ids.
1213
1214
:param line_iterator: An iterator of lines, origin_version_id
1215
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1216
to whether they were referred to by the inventory of the
1217
revision_id that they contain. Note that if that revision_id was
1218
not part of the line_iterator's output then False will be given -
1219
even though it may actually refer to that key.
1220
"""
1221
if not self._serializer.support_altered_by_hack:
1222
raise AssertionError(
1223
"_find_text_key_references_from_xml_inventory_lines only "
1224
"supported for branches which store inventory as unnested xml"
1225
", not on %r" % self)
376
1226
result = {}
377
1227
378
1228
# this code needs to read every new line in every inventory for the
383
1233
# revisions. We don't need to see all lines in the inventory because
384
1234
# only those added in an inventory in rev X can contain a revision=X
385
1235
# line.
386
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
387
start = line.find('file_id="')+9
388
if start < 9: continue
389
end = line.find('"', start)
390
assert end>= 0
391
file_id = _unescape_xml(line[start:end])
392
393
start = line.find('revision="')+10
394
if start < 10: continue
395
end = line.find('"', start)
396
assert end>= 0
397
revision_id = _unescape_xml(line[start:end])
398
if revision_id in selected_revision_ids:
399
result.setdefault(file_id, set()).add(revision_id)
400
return result
401
402
@needs_read_lock
403
def get_inventory_weave(self):
404
return self.control_weaves.get_weave('inventory',
405
self.get_transaction())
1236
unescape_revid_cache = {}
1237
unescape_fileid_cache = {}
1238
1239
# jam 20061218 In a big fetch, this handles hundreds of thousands
1240
# of lines, so it has had a lot of inlining and optimizing done.
1241
# Sorry that it is a little bit messy.
1242
# Move several functions to be local variables, since this is a long
1243
# running loop.
1244
search = self._file_ids_altered_regex.search
1245
unescape = _unescape_xml
1246
setdefault = result.setdefault
1247
for line, line_key in line_iterator:
1248
match = search(line)
1249
if match is None:
1250
continue
1251
# One call to match.group() returning multiple items is quite a
1252
# bit faster than 2 calls to match.group() each returning 1
1253
file_id, revision_id = match.group('file_id', 'revision_id')
1254
1255
# Inlining the cache lookups helps a lot when you make 170,000
1256
# lines and 350k ids, versus 8.4 unique ids.
1257
# Using a cache helps in 2 ways:
1258
# 1) Avoids unnecessary decoding calls
1259
# 2) Re-uses cached strings, which helps in future set and
1260
# equality checks.
1261
# (2) is enough that removing encoding entirely along with
1262
# the cache (so we are using plain strings) results in no
1263
# performance improvement.
1264
try:
1265
revision_id = unescape_revid_cache[revision_id]
1266
except KeyError:
1267
unescaped = unescape(revision_id)
1268
unescape_revid_cache[revision_id] = unescaped
1269
revision_id = unescaped
1270
1271
# Note that unconditionally unescaping means that we deserialise
1272
# every fileid, which for general 'pull' is not great, but we don't
1273
# really want to have some many fulltexts that this matters anyway.
1274
# RBC 20071114.
1275
try:
1276
file_id = unescape_fileid_cache[file_id]
1277
except KeyError:
1278
unescaped = unescape(file_id)
1279
unescape_fileid_cache[file_id] = unescaped
1280
file_id = unescaped
1281
1282
key = (file_id, revision_id)
1283
setdefault(key, False)
1284
if revision_id == line_key[-1]:
1285
result[key] = True
1286
return result
1287
1288
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1289
revision_ids):
1290
"""Helper routine for fileids_altered_by_revision_ids.
1291
1292
This performs the translation of xml lines to revision ids.
1293
1294
:param line_iterator: An iterator of lines, origin_version_id
1295
:param revision_ids: The revision ids to filter for. This should be a
1296
set or other type which supports efficient __contains__ lookups, as
1297
the revision id from each parsed line will be looked up in the
1298
revision_ids filter.
1299
:return: a dictionary mapping altered file-ids to an iterable of
1300
revision_ids. Each altered file-ids has the exact revision_ids that
1301
altered it listed explicitly.
1302
"""
1303
result = {}
1304
setdefault = result.setdefault
1305
for key in \
1306
self._find_text_key_references_from_xml_inventory_lines(
1307
line_iterator).iterkeys():
1308
# once data is all ensured-consistent; then this is
1309
# if revision_id == version_id
1310
if key[-1:] in revision_ids:
1311
setdefault(key[0], set()).add(key[-1])
1312
return result
1313
1314
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1315
"""Find the file ids and versions affected by revisions.
1316
1317
:param revisions: an iterable containing revision ids.
1318
:param _inv_weave: The inventory weave from this repository or None.
1319
If None, the inventory weave will be opened automatically.
1320
:return: a dictionary mapping altered file-ids to an iterable of
1321
revision_ids. Each altered file-ids has the exact revision_ids that
1322
altered it listed explicitly.
1323
"""
1324
selected_keys = set((revid,) for revid in revision_ids)
1325
w = _inv_weave or self.inventories
1326
pb = ui.ui_factory.nested_progress_bar()
1327
try:
1328
return self._find_file_ids_from_xml_inventory_lines(
1329
w.iter_lines_added_or_present_in_keys(
1330
selected_keys, pb=pb),
1331
selected_keys)
1332
finally:
1333
pb.finished()
1334
1335
def iter_files_bytes(self, desired_files):
1336
"""Iterate through file versions.
1337
1338
Files will not necessarily be returned in the order they occur in
1339
desired_files. No specific order is guaranteed.
1340
1341
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1342
value supplied by the caller as part of desired_files. It should
1343
uniquely identify the file version in the caller's context. (Examples:
1344
an index number or a TreeTransform trans_id.)
1345
1346
bytes_iterator is an iterable of bytestrings for the file. The
1347
kind of iterable and length of the bytestrings are unspecified, but for
1348
this implementation, it is a list of bytes produced by
1349
VersionedFile.get_record_stream().
1350
1351
:param desired_files: a list of (file_id, revision_id, identifier)
1352
triples
1353
"""
1354
transaction = self.get_transaction()
1355
text_keys = {}
1356
for file_id, revision_id, callable_data in desired_files:
1357
text_keys[(file_id, revision_id)] = callable_data
1358
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1359
if record.storage_kind == 'absent':
1360
raise errors.RevisionNotPresent(record.key, self)
1361
yield text_keys[record.key], record.get_bytes_as('fulltext')
1362
1363
def _generate_text_key_index(self, text_key_references=None,
1364
ancestors=None):
1365
"""Generate a new text key index for the repository.
1366
1367
This is an expensive function that will take considerable time to run.
1368
1369
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1370
list of parents, also text keys. When a given key has no parents,
1371
the parents list will be [NULL_REVISION].
1372
"""
1373
# All revisions, to find inventory parents.
1374
if ancestors is None:
1375
graph = self.get_graph()
1376
ancestors = graph.get_parent_map(self.all_revision_ids())
1377
if text_key_references is None:
1378
text_key_references = self.find_text_key_references()
1379
pb = ui.ui_factory.nested_progress_bar()
1380
try:
1381
return self._do_generate_text_key_index(ancestors,
1382
text_key_references, pb)
1383
finally:
1384
pb.finished()
1385
1386
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1387
"""Helper for _generate_text_key_index to avoid deep nesting."""
1388
revision_order = tsort.topo_sort(ancestors)
1389
invalid_keys = set()
1390
revision_keys = {}
1391
for revision_id in revision_order:
1392
revision_keys[revision_id] = set()
1393
text_count = len(text_key_references)
1394
# a cache of the text keys to allow reuse; costs a dict of all the
1395
# keys, but saves a 2-tuple for every child of a given key.
1396
text_key_cache = {}
1397
for text_key, valid in text_key_references.iteritems():
1398
if not valid:
1399
invalid_keys.add(text_key)
1400
else:
1401
revision_keys[text_key[1]].add(text_key)
1402
text_key_cache[text_key] = text_key
1403
del text_key_references
1404
text_index = {}
1405
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1406
NULL_REVISION = _mod_revision.NULL_REVISION
1407
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1408
# too small for large or very branchy trees. However, for 55K path
1409
# trees, it would be easy to use too much memory trivially. Ideally we
1410
# could gauge this by looking at available real memory etc, but this is
1411
# always a tricky proposition.
1412
inventory_cache = lru_cache.LRUCache(10)
1413
batch_size = 10 # should be ~150MB on a 55K path tree
1414
batch_count = len(revision_order) / batch_size + 1
1415
processed_texts = 0
1416
pb.update("Calculating text parents.", processed_texts, text_count)
1417
for offset in xrange(batch_count):
1418
to_query = revision_order[offset * batch_size:(offset + 1) *
1419
batch_size]
1420
if not to_query:
1421
break
1422
for rev_tree in self.revision_trees(to_query):
1423
revision_id = rev_tree.get_revision_id()
1424
parent_ids = ancestors[revision_id]
1425
for text_key in revision_keys[revision_id]:
1426
pb.update("Calculating text parents.", processed_texts)
1427
processed_texts += 1
1428
candidate_parents = []
1429
for parent_id in parent_ids:
1430
parent_text_key = (text_key[0], parent_id)
1431
try:
1432
check_parent = parent_text_key not in \
1433
revision_keys[parent_id]
1434
except KeyError:
1435
# the parent parent_id is a ghost:
1436
check_parent = False
1437
# truncate the derived graph against this ghost.
1438
parent_text_key = None
1439
if check_parent:
1440
# look at the parent commit details inventories to
1441
# determine possible candidates in the per file graph.
1442
# TODO: cache here.
1443
try:
1444
inv = inventory_cache[parent_id]
1445
except KeyError:
1446
inv = self.revision_tree(parent_id).inventory
1447
inventory_cache[parent_id] = inv
1448
parent_entry = inv._byid.get(text_key[0], None)
1449
if parent_entry is not None:
1450
parent_text_key = (
1451
text_key[0], parent_entry.revision)
1452
else:
1453
parent_text_key = None
1454
if parent_text_key is not None:
1455
candidate_parents.append(
1456
text_key_cache[parent_text_key])
1457
parent_heads = text_graph.heads(candidate_parents)
1458
new_parents = list(parent_heads)
1459
new_parents.sort(key=lambda x:candidate_parents.index(x))
1460
if new_parents == []:
1461
new_parents = [NULL_REVISION]
1462
text_index[text_key] = new_parents
1463
1464
for text_key in invalid_keys:
1465
text_index[text_key] = [NULL_REVISION]
1466
return text_index
1467
1468
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1469
"""Get an iterable listing the keys of all the data introduced by a set
1470
of revision IDs.
1471
1472
The keys will be ordered so that the corresponding items can be safely
1473
fetched and inserted in that order.
1474
1475
:returns: An iterable producing tuples of (knit-kind, file-id,
1476
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1477
'revisions'. file-id is None unless knit-kind is 'file'.
1478
"""
1479
# XXX: it's a bit weird to control the inventory weave caching in this
1480
# generator. Ideally the caching would be done in fetch.py I think. Or
1481
# maybe this generator should explicitly have the contract that it
1482
# should not be iterated until the previously yielded item has been
1483
# processed?
1484
inv_w = self.inventories
1485
1486
# file ids that changed
1487
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1488
count = 0
1489
num_file_ids = len(file_ids)
1490
for file_id, altered_versions in file_ids.iteritems():
1491
if _files_pb is not None:
1492
_files_pb.update("fetch texts", count, num_file_ids)
1493
count += 1
1494
yield ("file", file_id, altered_versions)
1495
# We're done with the files_pb. Note that it finished by the caller,
1496
# just as it was created by the caller.
1497
del _files_pb
1498
1499
# inventory
1500
yield ("inventory", None, revision_ids)
1501
1502
# signatures
1503
revisions_with_signatures = set()
1504
for rev_id in revision_ids:
1505
try:
1506
self.get_signature_text(rev_id)
1507
except errors.NoSuchRevision:
1508
# not signed.
1509
pass
1510
else:
1511
revisions_with_signatures.add(rev_id)
1512
yield ("signatures", None, revisions_with_signatures)
1513
1514
# revisions
1515
yield ("revisions", None, revision_ids)
406
1516
407
1517
@needs_read_lock
408
1518
def get_inventory(self, revision_id):
409
"""Get Inventory object by hash."""
410
return self.deserialise_inventory(
411
revision_id, self.get_inventory_xml(revision_id))
1519
"""Get Inventory object by revision id."""
1520
return self.iter_inventories([revision_id]).next()
1521
1522
def iter_inventories(self, revision_ids):
1523
"""Get many inventories by revision_ids.
1524
1525
This will buffer some or all of the texts used in constructing the
1526
inventories in memory, but will only parse a single inventory at a
1527
time.
1528
1529
:return: An iterator of inventories.
1530
"""
1531
if ((None in revision_ids)
1532
or (_mod_revision.NULL_REVISION in revision_ids)):
1533
raise ValueError('cannot get null revision inventory')
1534
return self._iter_inventories(revision_ids)
1535
1536
def _iter_inventories(self, revision_ids):
1537
"""single-document based inventory iteration."""
1538
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1539
yield self.deserialise_inventory(revision_id, text)
1540
1541
def _iter_inventory_xmls(self, revision_ids):
1542
keys = [(revision_id,) for revision_id in revision_ids]
1543
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1544
texts = {}
1545
for record in stream:
1546
if record.storage_kind != 'absent':
1547
texts[record.key] = record.get_bytes_as('fulltext')
1548
else:
1549
raise errors.NoSuchRevision(self, record.key)
1550
for key in keys:
1551
yield texts[key], key[-1]
412
1552
413
1553
def deserialise_inventory(self, revision_id, xml):
414
1554
"""Transform the xml into an inventory object.
416
1556
:param revision_id: The expected revision id of the inventory.
417
1557
:param xml: A serialised inventory.
418
1558
"""
419
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
1559
result = self._serializer.read_inventory_from_string(xml, revision_id)
1560
if result.revision_id != revision_id:
1561
raise AssertionError('revision id mismatch %s != %s' % (
1562
result.revision_id, revision_id))
1563
return result
1564
1565
def serialise_inventory(self, inv):
1566
return self._serializer.write_inventory_to_string(inv)
1567
1568
def _serialise_inventory_to_lines(self, inv):
1569
return self._serializer.write_inventory_to_lines(inv)
1570
1571
def get_serializer_format(self):
1572
return self._serializer.format_num
420
1573
421
1574
@needs_read_lock
422
1575
def get_inventory_xml(self, revision_id):
423
1576
"""Get inventory XML as a file object."""
1577
texts = self._iter_inventory_xmls([revision_id])
424
1578
try:
425
assert isinstance(revision_id, basestring), type(revision_id)
426
iw = self.get_inventory_weave()
427
return iw.get_text(revision_id)
428
except IndexError:
429
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
1579
text, revision_id = texts.next()
1580
except StopIteration:
1581
raise errors.HistoryMissing(self, 'inventory', revision_id)
1582
return text
430
1583
431
1584
@needs_read_lock
432
1585
def get_inventory_sha1(self, revision_id):
434
1587
"""
435
1588
return self.get_revision(revision_id).inventory_sha1
436
1589
437
@needs_read_lock
438
def get_revision_graph(self, revision_id=None):
439
"""Return a dictionary containing the revision graph.
440
441
:return: a dictionary of revision_id->revision_parents_list.
442
"""
443
weave = self.get_inventory_weave()
444
all_revisions = self._eliminate_revisions_not_present(weave.versions())
445
entire_graph = dict([(node, weave.get_parents(node)) for
446
node in all_revisions])
447
if revision_id is None:
448
return entire_graph
449
elif revision_id not in entire_graph:
450
raise errors.NoSuchRevision(self, revision_id)
451
else:
452
# add what can be reached from revision_id
453
result = {}
454
pending = set([revision_id])
455
while len(pending) > 0:
456
node = pending.pop()
457
result[node] = entire_graph[node]
458
for revision_id in result[node]:
459
if revision_id not in result:
460
pending.add(revision_id)
461
return result
462
463
@needs_read_lock
464
def get_revision_graph_with_ghosts(self, revision_ids=None):
465
"""Return a graph of the revisions with ghosts marked as applicable.
466
467
:param revision_ids: an iterable of revisions to graph or None for all.
468
:return: a Graph object with the graph reachable from revision_ids.
469
"""
470
result = Graph()
471
if not revision_ids:
472
pending = set(self.all_revision_ids())
473
required = set([])
474
else:
475
pending = set(revision_ids)
476
required = set(revision_ids)
477
done = set([])
478
while len(pending):
479
revision_id = pending.pop()
480
try:
481
rev = self.get_revision(revision_id)
482
except errors.NoSuchRevision:
483
if revision_id in required:
484
raise
485
# a ghost
486
result.add_ghost(revision_id)
487
continue
488
for parent_id in rev.parent_ids:
489
# is this queued or done ?
490
if (parent_id not in pending and
491
parent_id not in done):
492
# no, queue it.
493
pending.add(parent_id)
494
result.add_node(revision_id, rev.parent_ids)
495
done.add(revision_id)
496
return result
1590
def iter_reverse_revision_history(self, revision_id):
1591
"""Iterate backwards through revision ids in the lefthand history
1592
1593
:param revision_id: The revision id to start with. All its lefthand
1594
ancestors will be traversed.
1595
"""
1596
graph = self.get_graph()
1597
next_id = revision_id
1598
while True:
1599
if next_id in (None, _mod_revision.NULL_REVISION):
1600
return
1601
yield next_id
1602
# Note: The following line may raise KeyError in the event of
1603
# truncated history. We decided not to have a try:except:raise
1604
# RevisionNotPresent here until we see a use for it, because of the
1605
# cost in an inner loop that is by its very nature O(history).
1606
# Robert Collins 20080326
1607
parents = graph.get_parent_map([next_id])[next_id]
1608
if len(parents) == 0:
1609
return
1610
else:
1611
next_id = parents[0]
497
1612
498
1613
@needs_read_lock
499
1614
def get_revision_inventory(self, revision_id):
511
1626
else:
512
1627
return self.get_inventory(revision_id)
513
1628
514
@needs_read_lock
515
1629
def is_shared(self):
516
1630
"""Return True if this repository is flagged as a shared repository."""
517
1631
raise NotImplementedError(self.is_shared)
523
1637
reconciler = RepoReconciler(self, thorough=thorough)
524
1638
reconciler.reconcile()
525
1639
return reconciler
526
1640
1641
def _refresh_data(self):
1642
"""Helper called from lock_* to ensure coherency with disk.
1643
1644
The default implementation does nothing; it is however possible
1645
for repositories to maintain loaded indices across multiple locks
1646
by checking inside their implementation of this method to see
1647
whether their indices are still valid. This depends of course on
1648
the disk format being validatable in this manner.
1649
"""
1650
527
1651
@needs_read_lock
528
1652
def revision_tree(self, revision_id):
529
1653
"""Return Tree for a revision on this branch.
530
1654
531
`revision_id` may be None for the null revision, in which case
532
an `EmptyTree` is returned."""
1655
`revision_id` may be None for the empty tree revision.
1656
"""
533
1657
# TODO: refactor this to use an existing revision object
534
1658
# so we don't need to read it in twice.
535
if revision_id is None or revision_id == NULL_REVISION:
536
return EmptyTree()
1659
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1660
return RevisionTree(self, Inventory(root_id=None),
1661
_mod_revision.NULL_REVISION)
537
1662
else:
538
1663
inv = self.get_revision_inventory(revision_id)
539
1664
return RevisionTree(self, inv, revision_id)
540
1665
1666
def revision_trees(self, revision_ids):
1667
"""Return Tree for a revision on this branch.
1668
1669
`revision_id` may not be None or 'null:'"""
1670
inventories = self.iter_inventories(revision_ids)
1671
for inv in inventories:
1672
yield RevisionTree(self, inv, inv.revision_id)
1673
541
1674
@needs_read_lock
542
def get_ancestry(self, revision_id):
1675
def get_ancestry(self, revision_id, topo_sorted=True):
543
1676
"""Return a list of revision-ids integrated by a revision.
544
1677
545
1678
The first element of the list is always None, indicating the origin
548
1681
549
1682
This is topologically sorted.
550
1683
"""
551
if revision_id is None:
1684
if _mod_revision.is_null(revision_id):
552
1685
return [None]
553
1686
if not self.has_revision(revision_id):
554
1687
raise errors.NoSuchRevision(self, revision_id)
555
w = self.get_inventory_weave()
556
candidates = w.get_ancestry(revision_id)
557
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1688
graph = self.get_graph()
1689
keys = set()
1690
search = graph._make_breadth_first_searcher([revision_id])
1691
while True:
1692
try:
1693
found, ghosts = search.next_with_ghosts()
1694
except StopIteration:
1695
break
1696
keys.update(found)
1697
if _mod_revision.NULL_REVISION in keys:
1698
keys.remove(_mod_revision.NULL_REVISION)
1699
if topo_sorted:
1700
parent_map = graph.get_parent_map(keys)
1701
keys = tsort.topo_sort(parent_map)
1702
return [None] + list(keys)
1703
1704
def pack(self):
1705
"""Compress the data within the repository.
1706
1707
This operation only makes sense for some repository types. For other
1708
types it should be a no-op that just returns.
1709
1710
This stub method does not require a lock, but subclasses should use
1711
@needs_write_lock as this is a long running call its reasonable to
1712
implicitly lock for the user.
1713
"""
558
1714
559
1715
@needs_read_lock
1716
@deprecated_method(one_six)
560
1717
def print_file(self, file, revision_id):
561
1718
"""Print `file` to stdout.
562
1719
577
1734
def get_transaction(self):
578
1735
return self.control_files.get_transaction()
579
1736
580
def revision_parents(self, revid):
581
return self.get_inventory_weave().parent_names(revid)
1737
@deprecated_method(one_one)
1738
def get_parents(self, revision_ids):
1739
"""See StackedParentsProvider.get_parents"""
1740
parent_map = self.get_parent_map(revision_ids)
1741
return [parent_map.get(r, None) for r in revision_ids]
1742
1743
def get_parent_map(self, revision_ids):
1744
"""See graph._StackedParentsProvider.get_parent_map"""
1745
# revisions index works in keys; this just works in revisions
1746
# therefore wrap and unwrap
1747
query_keys = []
1748
result = {}
1749
for revision_id in revision_ids:
1750
if revision_id == _mod_revision.NULL_REVISION:
1751
result[revision_id] = ()
1752
elif revision_id is None:
1753
raise ValueError('get_parent_map(None) is not valid')
1754
else:
1755
query_keys.append((revision_id ,))
1756
for ((revision_id,), parent_keys) in \
1757
self.revisions.get_parent_map(query_keys).iteritems():
1758
if parent_keys:
1759
result[revision_id] = tuple(parent_revid
1760
for (parent_revid,) in parent_keys)
1761
else:
1762
result[revision_id] = (_mod_revision.NULL_REVISION,)
1763
return result
1764
1765
def _make_parents_provider(self):
1766
return self
1767
1768
def get_graph(self, other_repository=None):
1769
"""Return the graph walker for this repository format"""
1770
parents_provider = self._make_parents_provider()
1771
if (other_repository is not None and
1772
not self.has_same_location(other_repository)):
1773
parents_provider = graph._StackedParentsProvider(
1774
[parents_provider, other_repository._make_parents_provider()])
1775
return graph.Graph(parents_provider)
1776
1777
def _get_versioned_file_checker(self):
1778
"""Return an object suitable for checking versioned files."""
1779
return _VersionedFileChecker(self)
1780
1781
def revision_ids_to_search_result(self, result_set):
1782
"""Convert a set of revision ids to a graph SearchResult."""
1783
result_parents = set()
1784
for parents in self.get_graph().get_parent_map(
1785
result_set).itervalues():
1786
result_parents.update(parents)
1787
included_keys = result_set.intersection(result_parents)
1788
start_keys = result_set.difference(included_keys)
1789
exclude_keys = result_parents.difference(result_set)
1790
result = graph.SearchResult(start_keys, exclude_keys,
1791
len(result_set), result_set)
1792
return result
582
1793
583
1794
@needs_write_lock
584
1795
def set_make_working_trees(self, new_value):
604
1815
@needs_read_lock
605
1816
def has_signature_for_revision_id(self, revision_id):
606
1817
"""Query for a revision signature for revision_id in the repository."""
607
return self._revision_store.has_signature(revision_id,
608
self.get_transaction())
1818
if not self.has_revision(revision_id):
1819
raise errors.NoSuchRevision(self, revision_id)
1820
sig_present = (1 == len(
1821
self.signatures.get_parent_map([(revision_id,)])))
1822
return sig_present
609
1823
610
1824
@needs_read_lock
611
1825
def get_signature_text(self, revision_id):
612
1826
"""Return the text for a signature."""
613
return self._revision_store.get_signature_text(revision_id,
614
self.get_transaction())
1827
stream = self.signatures.get_record_stream([(revision_id,)],
1828
'unordered', True)
1829
record = stream.next()
1830
if record.storage_kind == 'absent':
1831
raise errors.NoSuchRevision(self, revision_id)
1832
return record.get_bytes_as('fulltext')
615
1833
616
1834
@needs_read_lock
617
def check(self, revision_ids):
1835
def check(self, revision_ids=None):
618
1836
"""Check consistency of all history of given revision_ids.
619
1837
620
1838
Different repository implementations should override _check().
622
1840
:param revision_ids: A non-empty list of revision_ids whose ancestry
623
1841
will be checked. Typically the last revision_id of a branch.
624
1842
"""
625
if not revision_ids:
626
raise ValueError("revision_ids must be non-empty in %s.check"
627
% (self,))
628
1843
return self._check(revision_ids)
629
1844
630
1845
def _check(self, revision_ids):
631
result = bzrlib.check.Check(self)
1846
result = check.Check(self)
632
1847
result.check()
633
1848
return result
634
1849
635
636
class AllInOneRepository(Repository):
637
"""Legacy support - the repository behaviour for all-in-one branches."""
638
639
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
640
# we reuse one control files instance.
641
dir_mode = a_bzrdir._control_files._dir_mode
642
file_mode = a_bzrdir._control_files._file_mode
643
644
def get_store(name, compressed=True, prefixed=False):
645
# FIXME: This approach of assuming stores are all entirely compressed
646
# or entirely uncompressed is tidy, but breaks upgrade from
647
# some existing branches where there's a mixture; we probably
648
# still want the option to look for both.
649
relpath = a_bzrdir._control_files._escape(name)
650
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
651
prefixed=prefixed, compressed=compressed,
652
dir_mode=dir_mode,
653
file_mode=file_mode)
654
#if self._transport.should_cache():
655
# cache_path = os.path.join(self.cache_root, name)
656
# os.mkdir(cache_path)
657
# store = bzrlib.store.CachedStore(store, cache_path)
658
return store
659
660
# not broken out yet because the controlweaves|inventory_store
661
# and text_store | weave_store bits are still different.
662
if isinstance(_format, RepositoryFormat4):
663
# cannot remove these - there is still no consistent api
664
# which allows access to this old info.
665
self.inventory_store = get_store('inventory-store')
666
text_store = get_store('text-store')
667
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
668
669
@needs_read_lock
670
def is_shared(self):
671
"""AllInOne repositories cannot be shared."""
672
return False
673
674
@needs_write_lock
675
def set_make_working_trees(self, new_value):
676
"""Set the policy flag for making working trees when creating branches.
677
678
This only applies to branches that use this repository.
679
680
The default is 'True'.
681
:param new_value: True to restore the default, False to disable making
682
working trees.
1850
def _warn_if_deprecated(self):
1851
global _deprecation_warning_done
1852
if _deprecation_warning_done:
1853
return
1854
_deprecation_warning_done = True
1855
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1856
% (self._format, self.bzrdir.transport.base))
1857
1858
def supports_rich_root(self):
1859
return self._format.rich_root_data
1860
1861
def _check_ascii_revisionid(self, revision_id, method):
1862
"""Private helper for ascii-only repositories."""
1863
# weave repositories refuse to store revisionids that are non-ascii.
1864
if revision_id is not None:
1865
# weaves require ascii revision ids.
1866
if isinstance(revision_id, unicode):
1867
try:
1868
revision_id.encode('ascii')
1869
except UnicodeEncodeError:
1870
raise errors.NonAsciiRevisionId(method, self)
1871
else:
1872
try:
1873
revision_id.decode('ascii')
1874
except UnicodeDecodeError:
1875
raise errors.NonAsciiRevisionId(method, self)
1876
1877
def revision_graph_can_have_wrong_parents(self):
1878
"""Is it possible for this repository to have a revision graph with
1879
incorrect parents?
1880
1881
If True, then this repository must also implement
1882
_find_inconsistent_revision_parents so that check and reconcile can
1883
check for inconsistencies before proceeding with other checks that may
1884
depend on the revision index being consistent.
683
1885
"""
684
raise NotImplementedError(self.set_make_working_trees)
685
686
def make_working_trees(self):
687
"""Returns the policy for making working trees on new branches."""
688
return True
1886
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1887
1888
1889
# remove these delegates a while after bzr 0.15
1890
def __make_delegated(name, from_module):
1891
def _deprecated_repository_forwarder():
1892
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1893
% (name, from_module),
1894
DeprecationWarning,
1895
stacklevel=2)
1896
m = __import__(from_module, globals(), locals(), [name])
1897
try:
1898
return getattr(m, name)
1899
except AttributeError:
1900
raise AttributeError('module %s has no name %s'
1901
% (m, name))
1902
globals()[name] = _deprecated_repository_forwarder
1903
1904
for _name in [
1905
'AllInOneRepository',
1906
'WeaveMetaDirRepository',
1907
'PreSplitOutRepositoryFormat',
1908
'RepositoryFormat4',
1909
'RepositoryFormat5',
1910
'RepositoryFormat6',
1911
'RepositoryFormat7',
1912
]:
1913
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1914
1915
for _name in [
1916
'KnitRepository',
1917
'RepositoryFormatKnit',
1918
'RepositoryFormatKnit1',
1919
]:
1920
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
689
1921
690
1922
691
1923
def install_revision(repository, rev, revision_tree):
692
1924
"""Install all revision data into a repository."""
1925
install_revisions(repository, [(rev, revision_tree, None)])
1926
1927
1928
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1929
"""Install all revision data into a repository.
1930
1931
Accepts an iterable of revision, tree, signature tuples. The signature
1932
may be None.
1933
"""
1934
repository.start_write_group()
1935
try:
1936
for n, (revision, revision_tree, signature) in enumerate(iterable):
1937
_install_revision(repository, revision, revision_tree, signature)
1938
if pb is not None:
1939
pb.update('Transferring revisions', n + 1, num_revisions)
1940
except:
1941
repository.abort_write_group()
1942
raise
1943
else:
1944
repository.commit_write_group()
1945
1946
1947
def _install_revision(repository, rev, revision_tree, signature):
1948
"""Install all revision data into a repository."""
693
1949
present_parents = []
694
1950
parent_trees = {}
695
1951
for p_id in rev.parent_ids:
697
1953
present_parents.append(p_id)
698
1954
parent_trees[p_id] = repository.revision_tree(p_id)
699
1955
else:
700
parent_trees[p_id] = EmptyTree()
1956
parent_trees[p_id] = repository.revision_tree(None)
701
1957
702
1958
inv = revision_tree.inventory
703
1959
entries = inv.iter_entries()
1960
# backwards compatibility hack: skip the root id.
1961
if not repository.supports_rich_root():
1962
path, root = entries.next()
1963
if root.revision != rev.revision_id:
1964
raise errors.IncompatibleRevision(repr(repository))
1965
text_keys = {}
1966
for path, ie in entries:
1967
text_keys[(ie.file_id, ie.revision)] = ie
1968
text_parent_map = repository.texts.get_parent_map(text_keys)
1969
missing_texts = set(text_keys) - set(text_parent_map)
704
1970
# Add the texts that are not already present
705
for path, ie in inv.iter_entries():
706
w = repository.weave_store.get_weave_or_empty(ie.file_id,
707
repository.get_transaction())
708
if ie.revision not in w:
709
text_parents = []
710
# FIXME: TODO: The following loop *may* be overlapping/duplicate
711
# with InventoryEntry.find_previous_heads(). if it is, then there
712
# is a latent bug here where the parents may have ancestors of each
713
# other. RBC, AB
714
for revision, tree in parent_trees.iteritems():
715
if ie.file_id not in tree:
716
continue
717
parent_id = tree.inventory[ie.file_id].revision
718
if parent_id in text_parents:
719
continue
720
text_parents.append(parent_id)
721
722
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
723
repository.get_transaction())
724
lines = revision_tree.get_file(ie.file_id).readlines()
725
vfile.add_lines(rev.revision_id, text_parents, lines)
1971
for text_key in missing_texts:
1972
ie = text_keys[text_key]
1973
text_parents = []
1974
# FIXME: TODO: The following loop overlaps/duplicates that done by
1975
# commit to determine parents. There is a latent/real bug here where
1976
# the parents inserted are not those commit would do - in particular
1977
# they are not filtered by heads(). RBC, AB
1978
for revision, tree in parent_trees.iteritems():
1979
if ie.file_id not in tree:
1980
continue
1981
parent_id = tree.inventory[ie.file_id].revision
1982
if parent_id in text_parents:
1983
continue
1984
text_parents.append((ie.file_id, parent_id))
1985
lines = revision_tree.get_file(ie.file_id).readlines()
1986
repository.texts.add_lines(text_key, text_parents, lines)
726
1987
try:
727
1988
# install the inventory
728
1989
repository.add_inventory(rev.revision_id, inv, present_parents)
729
1990
except errors.RevisionAlreadyPresent:
730
1991
pass
1992
if signature is not None:
1993
repository.add_signature_text(rev.revision_id, signature)
731
1994
repository.add_revision(rev.revision_id, rev, inv)
732
1995
733
1996
734
1997
class MetaDirRepository(Repository):
735
"""Repositories in the new meta-dir layout."""
736
737
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
738
super(MetaDirRepository, self).__init__(_format,
739
a_bzrdir,
740
control_files,
741
_revision_store,
742
control_store,
743
text_store)
744
745
dir_mode = self.control_files._dir_mode
746
file_mode = self.control_files._file_mode
747
748
@needs_read_lock
1998
"""Repositories in the new meta-dir layout.
1999
2000
:ivar _transport: Transport for access to repository control files,
2001
typically pointing to .bzr/repository.
2002
"""
2003
2004
def __init__(self, _format, a_bzrdir, control_files):
2005
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2006
self._transport = control_files._transport
2007
749
2008
def is_shared(self):
750
2009
"""Return True if this repository is flagged as a shared repository."""
751
return self.control_files._transport.has('shared-storage')
2010
return self._transport.has('shared-storage')
752
2011
753
2012
@needs_write_lock
754
2013
def set_make_working_trees(self, new_value):
762
2021
"""
763
2022
if new_value:
764
2023
try:
765
self.control_files._transport.delete('no-working-trees')
2024
self._transport.delete('no-working-trees')
766
2025
except errors.NoSuchFile:
767
2026
pass
768
2027
else:
769
self.control_files.put_utf8('no-working-trees', '')
2028
self._transport.put_bytes('no-working-trees', '',
2029
mode=self.bzrdir._get_file_mode())
770
2030
771
2031
def make_working_trees(self):
772
2032
"""Returns the policy for making working trees on new branches."""
773
return not self.control_files._transport.has('no-working-trees')
774
775
776
class KnitRepository(MetaDirRepository):
777
"""Knit format repository."""
778
779
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
780
inv_vf.add_lines_with_ghosts(revid, parents, lines)
781
782
@needs_read_lock
783
def _all_revision_ids(self):
784
"""See Repository.all_revision_ids()."""
785
# Knits get the revision graph from the index of the revision knit, so
786
# it's always possible even if they're on an unlistable transport.
787
return self._revision_store.all_revision_ids(self.get_transaction())
788
789
def fileid_involved_between_revs(self, from_revid, to_revid):
790
"""Find file_id(s) which are involved in the changes between revisions.
791
792
This determines the set of revisions which are involved, and then
793
finds all file ids affected by those revisions.
794
"""
795
vf = self._get_revision_vf()
796
from_set = set(vf.get_ancestry(from_revid))
797
to_set = set(vf.get_ancestry(to_revid))
798
changed = to_set.difference(from_set)
799
return self._fileid_involved_by_set(changed)
800
801
def fileid_involved(self, last_revid=None):
802
"""Find all file_ids modified in the ancestry of last_revid.
803
804
:param last_revid: If None, last_revision() will be used.
805
"""
806
if not last_revid:
807
changed = set(self.all_revision_ids())
808
else:
809
changed = set(self.get_ancestry(last_revid))
810
if None in changed:
811
changed.remove(None)
812
return self._fileid_involved_by_set(changed)
813
814
@needs_read_lock
815
def get_ancestry(self, revision_id):
816
"""Return a list of revision-ids integrated by a revision.
817
818
This is topologically sorted.
819
"""
820
if revision_id is None:
821
return [None]
822
vf = self._get_revision_vf()
823
try:
824
return [None] + vf.get_ancestry(revision_id)
825
except errors.RevisionNotPresent:
826
raise errors.NoSuchRevision(self, revision_id)
827
828
@needs_read_lock
829
def get_revision(self, revision_id):
830
"""Return the Revision object for a named revision"""
831
return self.get_revision_reconcile(revision_id)
832
833
@needs_read_lock
834
def get_revision_graph(self, revision_id=None):
835
"""Return a dictionary containing the revision graph.
836
837
:return: a dictionary of revision_id->revision_parents_list.
838
"""
839
weave = self._get_revision_vf()
840
entire_graph = weave.get_graph()
841
if revision_id is None:
842
return weave.get_graph()
843
elif revision_id not in weave:
844
raise errors.NoSuchRevision(self, revision_id)
845
else:
846
# add what can be reached from revision_id
847
result = {}
848
pending = set([revision_id])
849
while len(pending) > 0:
850
node = pending.pop()
851
result[node] = weave.get_parents(node)
852
for revision_id in result[node]:
853
if revision_id not in result:
854
pending.add(revision_id)
855
return result
856
857
@needs_read_lock
858
def get_revision_graph_with_ghosts(self, revision_ids=None):
859
"""Return a graph of the revisions with ghosts marked as applicable.
860
861
:param revision_ids: an iterable of revisions to graph or None for all.
862
:return: a Graph object with the graph reachable from revision_ids.
863
"""
864
result = Graph()
865
vf = self._get_revision_vf()
866
versions = set(vf.versions())
867
if not revision_ids:
868
pending = set(self.all_revision_ids())
869
required = set([])
870
else:
871
pending = set(revision_ids)
872
required = set(revision_ids)
873
done = set([])
874
while len(pending):
875
revision_id = pending.pop()
876
if not revision_id in versions:
877
if revision_id in required:
878
raise errors.NoSuchRevision(self, revision_id)
879
# a ghost
880
result.add_ghost(revision_id)
881
# mark it as done so we don't try for it again.
882
done.add(revision_id)
883
continue
884
parent_ids = vf.get_parents_with_ghosts(revision_id)
885
for parent_id in parent_ids:
886
# is this queued or done ?
887
if (parent_id not in pending and
888
parent_id not in done):
889
# no, queue it.
890
pending.add(parent_id)
891
result.add_node(revision_id, parent_ids)
892
done.add(revision_id)
893
return result
894
895
def _get_revision_vf(self):
896
""":return: a versioned file containing the revisions."""
897
vf = self._revision_store.get_revision_file(self.get_transaction())
898
return vf
899
900
@needs_write_lock
901
def reconcile(self, other=None, thorough=False):
902
"""Reconcile this repository."""
903
from bzrlib.reconcile import KnitReconciler
904
reconciler = KnitReconciler(self, thorough=thorough)
905
reconciler.reconcile()
906
return reconciler
2033
return not self._transport.has('no-working-trees')
2034
2035
2036
class MetaDirVersionedFileRepository(MetaDirRepository):
2037
"""Repositories in a meta-dir, that work via versioned file objects."""
2038
2039
def __init__(self, _format, a_bzrdir, control_files):
2040
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2041
control_files)
2042
2043
2044
class RepositoryFormatRegistry(registry.Registry):
2045
"""Registry of RepositoryFormats."""
2046
2047
def get(self, format_string):
2048
r = registry.Registry.get(self, format_string)
2049
if callable(r):
2050
r = r()
2051
return r
907
2052
908
def revision_parents(self, revid):
909
return self._get_revision_vf().get_parents(rev_id)
2053
2054
format_registry = RepositoryFormatRegistry()
2055
"""Registry of formats, indexed by their identifying format string.
2056
2057
This can contain either format instances themselves, or classes/factories that
2058
can be called to obtain one.
2059
"""
2060
2061
2062
#####################################################################
2063
# Repository Formats
910
2064
911
2065
class RepositoryFormat(object):
912
2066
"""A repository format.
917
2071
children.
918
2072
* an open routine which returns a Repository instance.
919
2073
2074
There is one and only one Format subclass for each on-disk format. But
2075
there can be one Repository subclass that is used for several different
2076
formats. The _format attribute on a Repository instance can be used to
2077
determine the disk format.
2078
920
2079
Formats are placed in an dict by their format string for reference
921
2080
during opening. These should be subclasses of RepositoryFormat
922
2081
for consistency.
929
2088
_matchingbzrdir - the bzrdir format that the repository format was
930
2089
originally written to work with. This can be used if manually
931
2090
constructing a bzrdir and repository, or more commonly for test suite
932
parameterisation.
2091
parameterization.
933
2092
"""
934
2093
935
_default_format = None
936
"""The default format used for new repositories."""
937
938
_formats = {}
939
"""The known formats."""
2094
# Set to True or False in derived classes. True indicates that the format
2095
# supports ghosts gracefully.
2096
supports_ghosts = None
2097
# Can this repository be given external locations to lookup additional
2098
# data. Set to True or False in derived classes.
2099
supports_external_lookups = None
2100
2101
def __str__(self):
2102
return "<%s>" % self.__class__.__name__
2103
2104
def __eq__(self, other):
2105
# format objects are generally stateless
2106
return isinstance(other, self.__class__)
2107
2108
def __ne__(self, other):
2109
return not self == other
940
2110
941
2111
@classmethod
942
2112
def find_format(klass, a_bzrdir):
943
"""Return the format for the repository object in a_bzrdir."""
2113
"""Return the format for the repository object in a_bzrdir.
2114
2115
This is used by bzr native formats that have a "format" file in
2116
the repository. Other methods may be used by different types of
2117
control directory.
2118
"""
944
2119
try:
945
2120
transport = a_bzrdir.get_repository_transport(None)
946
2121
format_string = transport.get("format").read()
947
return klass._formats[format_string]
2122
return format_registry.get(format_string)
948
2123
except errors.NoSuchFile:
949
2124
raise errors.NoRepositoryPresent(a_bzrdir)
950
2125
except KeyError:
951
raise errors.UnknownFormatError(format_string)
952
953
def _get_control_store(self, repo_transport, control_files):
954
"""Return the control store for this repository."""
955
raise NotImplementedError(self._get_control_store)
2126
raise errors.UnknownFormatError(format=format_string,
2127
kind='repository')
2128
2129
@classmethod
2130
def register_format(klass, format):
2131
format_registry.register(format.get_format_string(), format)
2132
2133
@classmethod
2134
def unregister_format(klass, format):
2135
format_registry.remove(format.get_format_string())
956
2136
957
2137
@classmethod
958
2138
def get_default_format(klass):
959
2139
"""Return the current default format."""
960
return klass._default_format
2140
from bzrlib import bzrdir
2141
return bzrdir.format_registry.make_bzrdir('default').repository_format
961
2142
962
2143
def get_format_string(self):
963
2144
"""Return the ASCII format string that identifies this format.
971
2152
"""Return the short description for this format."""
972
2153
raise NotImplementedError(self.get_format_description)
973
2154
974
def _get_revision_store(self, repo_transport, control_files):
975
"""Return the revision store object for this a_bzrdir."""
976
raise NotImplementedError(self._get_revision_store)
977
978
def _get_text_rev_store(self,
979
transport,
980
control_files,
981
name,
982
compressed=True,
983
prefixed=False,
984
serializer=None):
985
"""Common logic for getting a revision store for a repository.
986
987
see self._get_revision_store for the subclass-overridable method to
988
get the store for a repository.
989
"""
990
from bzrlib.store.revision.text import TextRevisionStore
991
dir_mode = control_files._dir_mode
992
file_mode = control_files._file_mode
993
text_store =TextStore(transport.clone(name),
994
prefixed=prefixed,
995
compressed=compressed,
996
dir_mode=dir_mode,
997
file_mode=file_mode)
998
_revision_store = TextRevisionStore(text_store, serializer)
999
return _revision_store
1000
2155
# TODO: this shouldn't be in the base class, it's specific to things that
2156
# use weaves or knits -- mbp 20070207
1001
2157
def _get_versioned_file_store(self,
1002
2158
name,
1003
2159
transport,
1004
2160
control_files,
1005
2161
prefixed=True,
1006
versionedfile_class=WeaveFile,
2162
versionedfile_class=None,
2163
versionedfile_kwargs={},
1007
2164
escaped=False):
2165
if versionedfile_class is None:
2166
versionedfile_class = self._versionedfile_class
1008
2167
weave_transport = control_files._transport.clone(name)
1009
2168
dir_mode = control_files._dir_mode
1010
2169
file_mode = control_files._file_mode
1012
2171
dir_mode=dir_mode,
1013
2172
file_mode=file_mode,
1014
2173
versionedfile_class=versionedfile_class,
2174
versionedfile_kwargs=versionedfile_kwargs,
1015
2175
escaped=escaped)
1016
2176
1017
2177
def initialize(self, a_bzrdir, shared=False):
1019
2179
1020
2180
:param a_bzrdir: The bzrdir to put the new repository in it.
1021
2181
:param shared: The repository should be initialized as a sharable one.
1022
2182
:returns: The new repository object.
2183
1023
2184
This may raise UninitializableFormat if shared repository are not
1024
2185
compatible the a_bzrdir.
1025
2186
"""
2187
raise NotImplementedError(self.initialize)
1026
2188
1027
2189
def is_supported(self):
1028
2190
"""Is this format supported?
1033
2195
"""
1034
2196
return True
1035
2197
2198
def check_conversion_target(self, target_format):
2199
raise NotImplementedError(self.check_conversion_target)
2200
1036
2201
def open(self, a_bzrdir, _found=False):
1037
2202
"""Return an instance of this format for the bzrdir a_bzrdir.
1038
2203
1040
2205
"""
1041
2206
raise NotImplementedError(self.open)
1042
2207
1043
@classmethod
1044
def register_format(klass, format):
1045
klass._formats[format.get_format_string()] = format
1046
1047
@classmethod
1048
def set_default_format(klass, format):
1049
klass._default_format = format
1050
1051
@classmethod
1052
def unregister_format(klass, format):
1053
assert klass._formats[format.get_format_string()] is format
1054
del klass._formats[format.get_format_string()]
1055
1056
1057
class PreSplitOutRepositoryFormat(RepositoryFormat):
1058
"""Base class for the pre split out repository formats."""
1059
1060
def initialize(self, a_bzrdir, shared=False, _internal=False):
1061
"""Create a weave repository.
1062
1063
TODO: when creating split out bzr branch formats, move this to a common
1064
base for Format5, Format6. or something like that.
1065
"""
1066
from bzrlib.weavefile import write_weave_v5
1067
from bzrlib.weave import Weave
1068
1069
if shared:
1070
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1071
1072
if not _internal:
1073
# always initialized when the bzrdir is.
1074
return self.open(a_bzrdir, _found=True)
1075
1076
# Create an empty weave
1077
sio = StringIO()
1078
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1079
empty_weave = sio.getvalue()
1080
1081
mutter('creating repository in %s.', a_bzrdir.transport.base)
1082
dirs = ['revision-store', 'weaves']
1083
files = [('inventory.weave', StringIO(empty_weave)),
1084
]
1085
1086
# FIXME: RBC 20060125 don't peek under the covers
1087
# NB: no need to escape relative paths that are url safe.
1088
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1089
TransportLock)
1090
control_files.create_lock()
1091
control_files.lock_write()
1092
control_files._transport.mkdir_multi(dirs,
1093
mode=control_files._dir_mode)
1094
try:
1095
for file, content in files:
1096
control_files.put(file, content)
1097
finally:
1098
control_files.unlock()
1099
return self.open(a_bzrdir, _found=True)
1100
1101
def _get_control_store(self, repo_transport, control_files):
1102
"""Return the control store for this repository."""
1103
return self._get_versioned_file_store('',
1104
repo_transport,
1105
control_files,
1106
prefixed=False)
1107
1108
def _get_text_store(self, transport, control_files):
1109
"""Get a store for file texts for this format."""
1110
raise NotImplementedError(self._get_text_store)
1111
1112
def open(self, a_bzrdir, _found=False):
1113
"""See RepositoryFormat.open()."""
1114
if not _found:
1115
# we are being called directly and must probe.
1116
raise NotImplementedError
1117
1118
repo_transport = a_bzrdir.get_repository_transport(None)
1119
control_files = a_bzrdir._control_files
1120
text_store = self._get_text_store(repo_transport, control_files)
1121
control_store = self._get_control_store(repo_transport, control_files)
1122
_revision_store = self._get_revision_store(repo_transport, control_files)
1123
return AllInOneRepository(_format=self,
1124
a_bzrdir=a_bzrdir,
1125
_revision_store=_revision_store,
1126
control_store=control_store,
1127
text_store=text_store)
1128
1129
1130
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1131
"""Bzr repository format 4.
1132
1133
This repository format has:
1134
- flat stores
1135
- TextStores for texts, inventories,revisions.
1136
1137
This format is deprecated: it indexes texts using a text id which is
1138
removed in format 5; initialization and write support for this format
1139
has been removed.
1140
"""
1141
1142
def __init__(self):
1143
super(RepositoryFormat4, self).__init__()
1144
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1145
1146
def get_format_description(self):
1147
"""See RepositoryFormat.get_format_description()."""
1148
return "Repository format 4"
1149
1150
def initialize(self, url, shared=False, _internal=False):
1151
"""Format 4 branches cannot be created."""
1152
raise errors.UninitializableFormat(self)
1153
1154
def is_supported(self):
1155
"""Format 4 is not supported.
1156
1157
It is not supported because the model changed from 4 to 5 and the
1158
conversion logic is expensive - so doing it on the fly was not
1159
feasible.
1160
"""
1161
return False
1162
1163
def _get_control_store(self, repo_transport, control_files):
1164
"""Format 4 repositories have no formal control store at this point.
1165
1166
This will cause any control-file-needing apis to fail - this is desired.
1167
"""
1168
return None
1169
1170
def _get_revision_store(self, repo_transport, control_files):
1171
"""See RepositoryFormat._get_revision_store()."""
1172
from bzrlib.xml4 import serializer_v4
1173
return self._get_text_rev_store(repo_transport,
1174
control_files,
1175
'revision-store',
1176
serializer=serializer_v4)
1177
1178
def _get_text_store(self, transport, control_files):
1179
"""See RepositoryFormat._get_text_store()."""
1180
1181
1182
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1183
"""Bzr control format 5.
1184
1185
This repository format has:
1186
- weaves for file texts and inventory
1187
- flat stores
1188
- TextStores for revisions and signatures.
1189
"""
1190
1191
def __init__(self):
1192
super(RepositoryFormat5, self).__init__()
1193
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1194
1195
def get_format_description(self):
1196
"""See RepositoryFormat.get_format_description()."""
1197
return "Weave repository format 5"
1198
1199
def _get_revision_store(self, repo_transport, control_files):
1200
"""See RepositoryFormat._get_revision_store()."""
1201
"""Return the revision store object for this a_bzrdir."""
1202
return self._get_text_rev_store(repo_transport,
1203
control_files,
1204
'revision-store',
1205
compressed=False)
1206
1207
def _get_text_store(self, transport, control_files):
1208
"""See RepositoryFormat._get_text_store()."""
1209
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1210
1211
1212
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1213
"""Bzr control format 6.
1214
1215
This repository format has:
1216
- weaves for file texts and inventory
1217
- hash subdirectory based stores.
1218
- TextStores for revisions and signatures.
1219
"""
1220
1221
def __init__(self):
1222
super(RepositoryFormat6, self).__init__()
1223
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1224
1225
def get_format_description(self):
1226
"""See RepositoryFormat.get_format_description()."""
1227
return "Weave repository format 6"
1228
1229
def _get_revision_store(self, repo_transport, control_files):
1230
"""See RepositoryFormat._get_revision_store()."""
1231
return self._get_text_rev_store(repo_transport,
1232
control_files,
1233
'revision-store',
1234
compressed=False,
1235
prefixed=True)
1236
1237
def _get_text_store(self, transport, control_files):
1238
"""See RepositoryFormat._get_text_store()."""
1239
return self._get_versioned_file_store('weaves', transport, control_files)
1240
1241
2208
1242
2209
class MetaDirRepositoryFormat(RepositoryFormat):
1243
2210
"""Common base class for the new repositories using the metadir layout."""
1244
2211
2212
rich_root_data = False
2213
supports_tree_reference = False
2214
supports_external_lookups = False
2215
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2216
1245
2217
def __init__(self):
1246
2218
super(MetaDirRepositoryFormat, self).__init__()
1247
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1248
2219
1249
2220
def _create_control_files(self, a_bzrdir):
1250
2221
"""Create the required files and the initial control_files object."""
1251
2222
# FIXME: RBC 20060125 don't peek under the covers
1252
2223
# NB: no need to escape relative paths that are url safe.
1253
2224
repository_transport = a_bzrdir.get_repository_transport(self)
1254
control_files = LockableFiles(repository_transport, 'lock', LockDir)
2225
control_files = lockable_files.LockableFiles(repository_transport,
2226
'lock', lockdir.LockDir)
1255
2227
control_files.create_lock()
1256
2228
return control_files
1257
2229
1259
2231
"""Upload the initial blank content."""
1260
2232
control_files = self._create_control_files(a_bzrdir)
1261
2233
control_files.lock_write()
2234
transport = control_files._transport
2235
if shared == True:
2236
utf8_files += [('shared-storage', '')]
1262
2237
try:
1263
control_files._transport.mkdir_multi(dirs,
1264
mode=control_files._dir_mode)
1265
for file, content in files:
1266
control_files.put(file, content)
1267
for file, content in utf8_files:
1268
control_files.put_utf8(file, content)
1269
if shared == True:
1270
control_files.put_utf8('shared-storage', '')
2238
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2239
for (filename, content_stream) in files:
2240
transport.put_file(filename, content_stream,
2241
mode=a_bzrdir._get_file_mode())
2242
for (filename, content_bytes) in utf8_files:
2243
transport.put_bytes_non_atomic(filename, content_bytes,
2244
mode=a_bzrdir._get_file_mode())
1271
2245
finally:
1272
2246
control_files.unlock()
1273
2247
1274
2248
1275
class RepositoryFormat7(MetaDirRepositoryFormat):
1276
"""Bzr repository 7.
1277
1278
This repository format has:
1279
- weaves for file texts and inventory
1280
- hash subdirectory based stores.
1281
- TextStores for revisions and signatures.
1282
- a format marker of its own
1283
- an optional 'shared-storage' flag
1284
- an optional 'no-working-trees' flag
1285
"""
1286
1287
def _get_control_store(self, repo_transport, control_files):
1288
"""Return the control store for this repository."""
1289
return self._get_versioned_file_store('',
1290
repo_transport,
1291
control_files,
1292
prefixed=False)
1293
1294
def get_format_string(self):
1295
"""See RepositoryFormat.get_format_string()."""
1296
return "Bazaar-NG Repository format 7"
1297
1298
def get_format_description(self):
1299
"""See RepositoryFormat.get_format_description()."""
1300
return "Weave repository format 7"
1301
1302
def _get_revision_store(self, repo_transport, control_files):
1303
"""See RepositoryFormat._get_revision_store()."""
1304
return self._get_text_rev_store(repo_transport,
1305
control_files,
1306
'revision-store',
1307
compressed=False,
1308
prefixed=True,
1309
)
1310
1311
def _get_text_store(self, transport, control_files):
1312
"""See RepositoryFormat._get_text_store()."""
1313
return self._get_versioned_file_store('weaves',
1314
transport,
1315
control_files)
1316
1317
def initialize(self, a_bzrdir, shared=False):
1318
"""Create a weave repository.
1319
1320
:param shared: If true the repository will be initialized as a shared
1321
repository.
1322
"""
1323
from bzrlib.weavefile import write_weave_v5
1324
from bzrlib.weave import Weave
1325
1326
# Create an empty weave
1327
sio = StringIO()
1328
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1329
empty_weave = sio.getvalue()
1330
1331
mutter('creating repository in %s.', a_bzrdir.transport.base)
1332
dirs = ['revision-store', 'weaves']
1333
files = [('inventory.weave', StringIO(empty_weave)),
1334
]
1335
utf8_files = [('format', self.get_format_string())]
1336
1337
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1338
return self.open(a_bzrdir=a_bzrdir, _found=True)
1339
1340
def open(self, a_bzrdir, _found=False, _override_transport=None):
1341
"""See RepositoryFormat.open().
1342
1343
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1344
repository at a slightly different url
1345
than normal. I.e. during 'upgrade'.
1346
"""
1347
if not _found:
1348
format = RepositoryFormat.find_format(a_bzrdir)
1349
assert format.__class__ == self.__class__
1350
if _override_transport is not None:
1351
repo_transport = _override_transport
1352
else:
1353
repo_transport = a_bzrdir.get_repository_transport(None)
1354
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1355
text_store = self._get_text_store(repo_transport, control_files)
1356
control_store = self._get_control_store(repo_transport, control_files)
1357
_revision_store = self._get_revision_store(repo_transport, control_files)
1358
return MetaDirRepository(_format=self,
1359
a_bzrdir=a_bzrdir,
1360
control_files=control_files,
1361
_revision_store=_revision_store,
1362
control_store=control_store,
1363
text_store=text_store)
1364
1365
1366
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1367
"""Bzr repository knit format 1.
1368
1369
This repository format has:
1370
- knits for file texts and inventory
1371
- hash subdirectory based stores.
1372
- knits for revisions and signatures
1373
- TextStores for revisions and signatures.
1374
- a format marker of its own
1375
- an optional 'shared-storage' flag
1376
- an optional 'no-working-trees' flag
1377
- a LockDir lock
1378
1379
This format was introduced in bzr 0.8.
1380
"""
1381
1382
def _get_control_store(self, repo_transport, control_files):
1383
"""Return the control store for this repository."""
1384
return VersionedFileStore(
1385
repo_transport,
1386
prefixed=False,
1387
file_mode=control_files._file_mode,
1388
versionedfile_class=KnitVersionedFile,
1389
versionedfile_kwargs={'factory':KnitPlainFactory()},
1390
)
1391
1392
def get_format_string(self):
1393
"""See RepositoryFormat.get_format_string()."""
1394
return "Bazaar-NG Knit Repository Format 1"
1395
1396
def get_format_description(self):
1397
"""See RepositoryFormat.get_format_description()."""
1398
return "Knit repository format 1"
1399
1400
def _get_revision_store(self, repo_transport, control_files):
1401
"""See RepositoryFormat._get_revision_store()."""
1402
from bzrlib.store.revision.knit import KnitRevisionStore
1403
versioned_file_store = VersionedFileStore(
1404
repo_transport,
1405
file_mode=control_files._file_mode,
1406
prefixed=False,
1407
precious=True,
1408
versionedfile_class=KnitVersionedFile,
1409
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1410
escaped=True,
1411
)
1412
return KnitRevisionStore(versioned_file_store)
1413
1414
def _get_text_store(self, transport, control_files):
1415
"""See RepositoryFormat._get_text_store()."""
1416
return self._get_versioned_file_store('knits',
1417
transport,
1418
control_files,
1419
versionedfile_class=KnitVersionedFile,
1420
escaped=True)
1421
1422
def initialize(self, a_bzrdir, shared=False):
1423
"""Create a knit format 1 repository.
1424
1425
:param a_bzrdir: bzrdir to contain the new repository; must already
1426
be initialized.
1427
:param shared: If true the repository will be initialized as a shared
1428
repository.
1429
"""
1430
mutter('creating repository in %s.', a_bzrdir.transport.base)
1431
dirs = ['revision-store', 'knits']
1432
files = []
1433
utf8_files = [('format', self.get_format_string())]
1434
1435
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1436
repo_transport = a_bzrdir.get_repository_transport(None)
1437
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1438
control_store = self._get_control_store(repo_transport, control_files)
1439
transaction = bzrlib.transactions.WriteTransaction()
1440
# trigger a write of the inventory store.
1441
control_store.get_weave_or_empty('inventory', transaction)
1442
_revision_store = self._get_revision_store(repo_transport, control_files)
1443
_revision_store.has_revision_id('A', transaction)
1444
_revision_store.get_signature_file(transaction)
1445
return self.open(a_bzrdir=a_bzrdir, _found=True)
1446
1447
def open(self, a_bzrdir, _found=False, _override_transport=None):
1448
"""See RepositoryFormat.open().
1449
1450
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1451
repository at a slightly different url
1452
than normal. I.e. during 'upgrade'.
1453
"""
1454
if not _found:
1455
format = RepositoryFormat.find_format(a_bzrdir)
1456
assert format.__class__ == self.__class__
1457
if _override_transport is not None:
1458
repo_transport = _override_transport
1459
else:
1460
repo_transport = a_bzrdir.get_repository_transport(None)
1461
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1462
text_store = self._get_text_store(repo_transport, control_files)
1463
control_store = self._get_control_store(repo_transport, control_files)
1464
_revision_store = self._get_revision_store(repo_transport, control_files)
1465
return KnitRepository(_format=self,
1466
a_bzrdir=a_bzrdir,
1467
control_files=control_files,
1468
_revision_store=_revision_store,
1469
control_store=control_store,
1470
text_store=text_store)
1471
1472
1473
2249
# formats which have no format string are not discoverable
1474
# and not independently creatable, so are not registered.
1475
RepositoryFormat.register_format(RepositoryFormat7())
1476
_default_format = RepositoryFormatKnit1()
1477
RepositoryFormat.register_format(_default_format)
1478
RepositoryFormat.set_default_format(_default_format)
1479
_legacy_formats = [RepositoryFormat4(),
1480
RepositoryFormat5(),
1481
RepositoryFormat6()]
2250
# and not independently creatable, so are not registered. They're
2251
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2252
# needed, it's constructed directly by the BzrDir. Non-native formats where
2253
# the repository is not separately opened are similar.
2254
2255
format_registry.register_lazy(
2256
'Bazaar-NG Repository format 7',
2257
'bzrlib.repofmt.weaverepo',
2258
'RepositoryFormat7'
2259
)
2260
2261
format_registry.register_lazy(
2262
'Bazaar-NG Knit Repository Format 1',
2263
'bzrlib.repofmt.knitrepo',
2264
'RepositoryFormatKnit1',
2265
)
2266
2267
format_registry.register_lazy(
2268
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2269
'bzrlib.repofmt.knitrepo',
2270
'RepositoryFormatKnit3',
2271
)
2272
2273
format_registry.register_lazy(
2274
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2275
'bzrlib.repofmt.knitrepo',
2276
'RepositoryFormatKnit4',
2277
)
2278
2279
# Pack-based formats. There is one format for pre-subtrees, and one for
2280
# post-subtrees to allow ease of testing.
2281
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2282
format_registry.register_lazy(
2283
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2284
'bzrlib.repofmt.pack_repo',
2285
'RepositoryFormatKnitPack1',
2286
)
2287
format_registry.register_lazy(
2288
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2289
'bzrlib.repofmt.pack_repo',
2290
'RepositoryFormatKnitPack3',
2291
)
2292
format_registry.register_lazy(
2293
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2294
'bzrlib.repofmt.pack_repo',
2295
'RepositoryFormatKnitPack4',
2296
)
2297
format_registry.register_lazy(
2298
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2299
'bzrlib.repofmt.pack_repo',
2300
'RepositoryFormatKnitPack5',
2301
)
2302
format_registry.register_lazy(
2303
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2304
'bzrlib.repofmt.pack_repo',
2305
'RepositoryFormatKnitPack5RichRoot',
2306
)
2307
2308
# Development formats.
2309
# 1.2->1.3
2310
# development 0 - stub to introduce development versioning scheme.
2311
format_registry.register_lazy(
2312
"Bazaar development format 0 (needs bzr.dev from before 1.3)\n",
2313
'bzrlib.repofmt.pack_repo',
2314
'RepositoryFormatPackDevelopment0',
2315
)
2316
format_registry.register_lazy(
2317
("Bazaar development format 0 with subtree support "
2318
"(needs bzr.dev from before 1.3)\n"),
2319
'bzrlib.repofmt.pack_repo',
2320
'RepositoryFormatPackDevelopment0Subtree',
2321
)
2322
format_registry.register_lazy(
2323
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2324
'bzrlib.repofmt.pack_repo',
2325
'RepositoryFormatPackDevelopment1',
2326
)
2327
format_registry.register_lazy(
2328
("Bazaar development format 1 with subtree support "
2329
"(needs bzr.dev from before 1.6)\n"),
2330
'bzrlib.repofmt.pack_repo',
2331
'RepositoryFormatPackDevelopment1Subtree',
2332
)
2333
# 1.3->1.4 go below here
1482
2334
1483
2335
1484
2336
class InterRepository(InterObject):
1493
2345
InterRepository.get(other).method_name(parameters).
1494
2346
"""
1495
2347
1496
_optimisers = set()
2348
_optimisers = []
1497
2349
"""The available optimised InterRepository types."""
1498
2350
1499
@needs_write_lock
1500
def copy_content(self, revision_id=None, basis=None):
1501
"""Make a complete copy of the content in self into destination.
1502
1503
This is a destructive operation! Do not use it on existing
1504
repositories.
1505
1506
:param revision_id: Only copy the content needed to construct
1507
revision_id and its parents.
1508
:param basis: Copy the needed data preferentially from basis.
1509
"""
1510
try:
1511
self.target.set_make_working_trees(self.source.make_working_trees())
1512
except NotImplementedError:
1513
pass
1514
# grab the basis available data
1515
if basis is not None:
1516
self.target.fetch(basis, revision_id=revision_id)
1517
# but don't bother fetching if we have the needed data now.
1518
if (revision_id not in (None, NULL_REVISION) and
1519
self.target.has_revision(revision_id)):
1520
return
1521
self.target.fetch(self.source, revision_id=revision_id)
1522
1523
def _double_lock(self, lock_source, lock_target):
1524
"""Take out too locks, rolling back the first if the second throws."""
1525
lock_source()
1526
try:
1527
lock_target()
1528
except Exception:
1529
# we want to ensure that we don't leave source locked by mistake.
1530
# and any error on target should not confuse source.
1531
self.source.unlock()
1532
raise
1533
1534
@needs_write_lock
1535
def fetch(self, revision_id=None, pb=None):
2351
def copy_content(self, revision_id=None):
2352
raise NotImplementedError(self.copy_content)
2353
2354
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1536
2355
"""Fetch the content required to construct revision_id.
1537
2356
1538
The content is copied from source to target.
2357
The content is copied from self.source to self.target.
1539
2358
1540
2359
:param revision_id: if None all content is copied, if NULL_REVISION no
1541
2360
content is copied.
1542
2361
:param pb: optional progress bar to use for progress reports. If not
1543
2362
provided a default one will be created.
1544
2363
1545
Returns the copied revision count and the failed revisions in a tuple:
1546
(copied, failures).
1547
"""
1548
from bzrlib.fetch import GenericRepoFetcher
1549
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1550
self.source, self.source._format, self.target, self.target._format)
1551
f = GenericRepoFetcher(to_repository=self.target,
1552
from_repository=self.source,
1553
last_revision=revision_id,
1554
pb=pb)
1555
return f.count_copied, f.failed_revisions
1556
1557
def lock_read(self):
1558
"""Take out a logical read lock.
1559
1560
This will lock the source branch and the target branch. The source gets
1561
a read lock and the target a read lock.
1562
"""
1563
self._double_lock(self.source.lock_read, self.target.lock_read)
1564
1565
def lock_write(self):
1566
"""Take out a logical write lock.
1567
1568
This will lock the source branch and the target branch. The source gets
1569
a read lock and the target a write lock.
1570
"""
1571
self._double_lock(self.source.lock_read, self.target.lock_write)
1572
2364
:returns: (copied_revision_count, failures).
2365
"""
2366
# Normally we should find a specific InterRepository subclass to do
2367
# the fetch; if nothing else then at least InterSameDataRepository.
2368
# If none of them is suitable it looks like fetching is not possible;
2369
# we try to give a good message why. _assert_same_model will probably
2370
# give a helpful message; otherwise a generic one.
2371
self._assert_same_model(self.source, self.target)
2372
raise errors.IncompatibleRepositories(self.source, self.target,
2373
"no suitableInterRepository found")
2374
2375
def _walk_to_common_revisions(self, revision_ids):
2376
"""Walk out from revision_ids in source to revisions target has.
2377
2378
:param revision_ids: The start point for the search.
2379
:return: A set of revision ids.
2380
"""
2381
target_graph = self.target.get_graph()
2382
revision_ids = frozenset(revision_ids)
2383
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2384
return graph.SearchResult(revision_ids, set(), 0, set())
2385
missing_revs = set()
2386
source_graph = self.source.get_graph()
2387
# ensure we don't pay silly lookup costs.
2388
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2389
null_set = frozenset([_mod_revision.NULL_REVISION])
2390
while True:
2391
try:
2392
next_revs, ghosts = searcher.next_with_ghosts()
2393
except StopIteration:
2394
break
2395
if revision_ids.intersection(ghosts):
2396
absent_ids = set(revision_ids.intersection(ghosts))
2397
# If all absent_ids are present in target, no error is needed.
2398
absent_ids.difference_update(
2399
set(target_graph.get_parent_map(absent_ids)))
2400
if absent_ids:
2401
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2402
# we don't care about other ghosts as we can't fetch them and
2403
# haven't been asked to.
2404
next_revs = set(next_revs)
2405
# we always have NULL_REVISION present.
2406
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2407
missing_revs.update(next_revs - have_revs)
2408
searcher.stop_searching_any(have_revs)
2409
return searcher.get_result()
2410
2411
@deprecated_method(one_two)
1573
2412
@needs_read_lock
1574
def missing_revision_ids(self, revision_id=None):
2413
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1575
2414
"""Return the revision ids that source has that target does not.
1576
2415
1577
2416
These are returned in topological order.
1578
2417
1579
2418
:param revision_id: only return revision ids included by this
1580
2419
revision_id.
1581
"""
2420
:param find_ghosts: If True find missing revisions in deep history
2421
rather than just finding the surface difference.
2422
"""
2423
return list(self.search_missing_revision_ids(
2424
revision_id, find_ghosts).get_keys())
2425
2426
@needs_read_lock
2427
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2428
"""Return the revision ids that source has that target does not.
2429
2430
:param revision_id: only return revision ids included by this
2431
revision_id.
2432
:param find_ghosts: If True find missing revisions in deep history
2433
rather than just finding the surface difference.
2434
:return: A bzrlib.graph.SearchResult.
2435
"""
2436
# stop searching at found target revisions.
2437
if not find_ghosts and revision_id is not None:
2438
return self._walk_to_common_revisions([revision_id])
1582
2439
# generic, possibly worst case, slow code path.
1583
2440
target_ids = set(self.target.all_revision_ids())
1584
2441
if revision_id is not None:
1585
2442
source_ids = self.source.get_ancestry(revision_id)
1586
assert source_ids[0] == None
2443
if source_ids[0] is not None:
2444
raise AssertionError()
1587
2445
source_ids.pop(0)
1588
2446
else:
1589
2447
source_ids = self.source.all_revision_ids()
1590
2448
result_set = set(source_ids).difference(target_ids)
1591
# this may look like a no-op: its not. It preserves the ordering
1592
# other_ids had while only returning the members from other_ids
1593
# that we've decided we need.
1594
return [rev_id for rev_id in source_ids if rev_id in result_set]
1595
1596
def unlock(self):
1597
"""Release the locks on source and target."""
1598
try:
1599
self.target.unlock()
1600
finally:
1601
self.source.unlock()
1602
1603
1604
class InterWeaveRepo(InterRepository):
1605
"""Optimised code paths between Weave based repositories."""
1606
1607
_matching_repo_format = RepositoryFormat7()
1608
"""Repository format for testing with."""
2449
return self.source.revision_ids_to_search_result(result_set)
2450
2451
@staticmethod
2452
def _same_model(source, target):
2453
"""True if source and target have the same data representation.
2454
2455
Note: this is always called on the base class; overriding it in a
2456
subclass will have no effect.
2457
"""
2458
try:
2459
InterRepository._assert_same_model(source, target)
2460
return True
2461
except errors.IncompatibleRepositories, e:
2462
return False
2463
2464
@staticmethod
2465
def _assert_same_model(source, target):
2466
"""Raise an exception if two repositories do not use the same model.
2467
"""
2468
if source.supports_rich_root() != target.supports_rich_root():
2469
raise errors.IncompatibleRepositories(source, target,
2470
"different rich-root support")
2471
if source._serializer != target._serializer:
2472
raise errors.IncompatibleRepositories(source, target,
2473
"different serializers")
2474
2475
2476
class InterSameDataRepository(InterRepository):
2477
"""Code for converting between repositories that represent the same data.
2478
2479
Data format and model must match for this to work.
2480
"""
2481
2482
@classmethod
2483
def _get_repo_format_to_test(self):
2484
"""Repository format for testing with.
2485
2486
InterSameData can pull from subtree to subtree and from non-subtree to
2487
non-subtree, so we test this with the richest repository format.
2488
"""
2489
from bzrlib.repofmt import knitrepo
2490
return knitrepo.RepositoryFormatKnit3()
2491
2492
@staticmethod
2493
def is_compatible(source, target):
2494
return InterRepository._same_model(source, target)
2495
2496
@needs_write_lock
2497
def copy_content(self, revision_id=None):
2498
"""Make a complete copy of the content in self into destination.
2499
2500
This copies both the repository's revision data, and configuration information
2501
such as the make_working_trees setting.
2502
2503
This is a destructive operation! Do not use it on existing
2504
repositories.
2505
2506
:param revision_id: Only copy the content needed to construct
2507
revision_id and its parents.
2508
"""
2509
try:
2510
self.target.set_make_working_trees(self.source.make_working_trees())
2511
except NotImplementedError:
2512
pass
2513
# but don't bother fetching if we have the needed data now.
2514
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2515
self.target.has_revision(revision_id)):
2516
return
2517
self.target.fetch(self.source, revision_id=revision_id)
2518
2519
@needs_write_lock
2520
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2521
"""See InterRepository.fetch()."""
2522
from bzrlib.fetch import RepoFetcher
2523
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2524
self.source, self.source._format, self.target,
2525
self.target._format)
2526
f = RepoFetcher(to_repository=self.target,
2527
from_repository=self.source,
2528
last_revision=revision_id,
2529
pb=pb, find_ghosts=find_ghosts)
2530
return f.count_copied, f.failed_revisions
2531
2532
2533
class InterWeaveRepo(InterSameDataRepository):
2534
"""Optimised code paths between Weave based repositories.
2535
2536
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2537
implemented lazy inter-object optimisation.
2538
"""
2539
2540
@classmethod
2541
def _get_repo_format_to_test(self):
2542
from bzrlib.repofmt import weaverepo
2543
return weaverepo.RepositoryFormat7()
1609
2544
1610
2545
@staticmethod
1611
2546
def is_compatible(source, target):
1615
2550
could lead to confusing results, and there is no need to be
1616
2551
overly general.
1617
2552
"""
2553
from bzrlib.repofmt.weaverepo import (
2554
RepositoryFormat5,
2555
RepositoryFormat6,
2556
RepositoryFormat7,
2557
)
1618
2558
try:
1619
2559
return (isinstance(source._format, (RepositoryFormat5,
1620
2560
RepositoryFormat6,
1626
2566
return False
1627
2567
1628
2568
@needs_write_lock
1629
def copy_content(self, revision_id=None, basis=None):
2569
def copy_content(self, revision_id=None):
1630
2570
"""See InterRepository.copy_content()."""
1631
2571
# weave specific optimised path:
1632
if basis is not None:
1633
# copy the basis in, then fetch remaining data.
1634
basis.copy_content_into(self.target, revision_id)
1635
# the basis copy_content_into could miss-set this.
2572
try:
2573
self.target.set_make_working_trees(self.source.make_working_trees())
2574
except (errors.RepositoryUpgradeRequired, NotImplemented):
2575
pass
2576
# FIXME do not peek!
2577
if self.source._transport.listable():
2578
pb = ui.ui_factory.nested_progress_bar()
1636
2579
try:
1637
self.target.set_make_working_trees(self.source.make_working_trees())
1638
except NotImplementedError:
1639
pass
2580
self.target.texts.insert_record_stream(
2581
self.source.texts.get_record_stream(
2582
self.source.texts.keys(), 'topological', False))
2583
pb.update('copying inventory', 0, 1)
2584
self.target.inventories.insert_record_stream(
2585
self.source.inventories.get_record_stream(
2586
self.source.inventories.keys(), 'topological', False))
2587
self.target.signatures.insert_record_stream(
2588
self.source.signatures.get_record_stream(
2589
self.source.signatures.keys(),
2590
'unordered', True))
2591
self.target.revisions.insert_record_stream(
2592
self.source.revisions.get_record_stream(
2593
self.source.revisions.keys(),
2594
'topological', True))
2595
finally:
2596
pb.finished()
2597
else:
1640
2598
self.target.fetch(self.source, revision_id=revision_id)
1641
else:
1642
try:
1643
self.target.set_make_working_trees(self.source.make_working_trees())
1644
except NotImplementedError:
1645
pass
1646
# FIXME do not peek!
1647
if self.source.control_files._transport.listable():
1648
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1649
try:
1650
self.target.weave_store.copy_all_ids(
1651
self.source.weave_store,
1652
pb=pb,
1653
from_transaction=self.source.get_transaction(),
1654
to_transaction=self.target.get_transaction())
1655
pb.update('copying inventory', 0, 1)
1656
self.target.control_weaves.copy_multi(
1657
self.source.control_weaves, ['inventory'],
1658
from_transaction=self.source.get_transaction(),
1659
to_transaction=self.target.get_transaction())
1660
self.target._revision_store.text_store.copy_all_ids(
1661
self.source._revision_store.text_store,
1662
pb=pb)
1663
finally:
1664
pb.finished()
1665
else:
1666
self.target.fetch(self.source, revision_id=revision_id)
1667
2599
1668
2600
@needs_write_lock
1669
def fetch(self, revision_id=None, pb=None):
2601
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1670
2602
"""See InterRepository.fetch()."""
1671
from bzrlib.fetch import GenericRepoFetcher
2603
from bzrlib.fetch import RepoFetcher
1672
2604
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1673
2605
self.source, self.source._format, self.target, self.target._format)
1674
f = GenericRepoFetcher(to_repository=self.target,
2606
f = RepoFetcher(to_repository=self.target,
1675
2607
from_repository=self.source,
1676
2608
last_revision=revision_id,
1677
pb=pb)
2609
pb=pb, find_ghosts=find_ghosts)
1678
2610
return f.count_copied, f.failed_revisions
1679
2611
1680
2612
@needs_read_lock
1681
def missing_revision_ids(self, revision_id=None):
2613
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1682
2614
"""See InterRepository.missing_revision_ids()."""
1683
2615
# we want all revisions to satisfy revision_id in source.
1684
2616
# but we don't want to stat every file here and there.
1693
2625
# - RBC 20060209
1694
2626
if revision_id is not None:
1695
2627
source_ids = self.source.get_ancestry(revision_id)
1696
assert source_ids[0] == None
2628
if source_ids[0] is not None:
2629
raise AssertionError()
1697
2630
source_ids.pop(0)
1698
2631
else:
1699
2632
source_ids = self.source._all_possible_ids()
1704
2637
# we do not have a revision as that would be pointless.
1705
2638
target_ids = set(self.target._all_possible_ids())
1706
2639
possibly_present_revisions = target_ids.intersection(source_ids_set)
1707
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2640
actually_present_revisions = set(
2641
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1708
2642
required_revisions = source_ids_set.difference(actually_present_revisions)
1709
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1710
2643
if revision_id is not None:
1711
2644
# we used get_ancestry to determine source_ids then we are assured all
1712
2645
# revisions referenced are present as they are installed in topological order.
1713
2646
# and the tip revision was validated by get_ancestry.
1714
return required_topo_revisions
2647
result_set = required_revisions
1715
2648
else:
1716
2649
# if we just grabbed the possibly available ids, then
1717
2650
# we only have an estimate of whats available and need to validate
1718
2651
# that against the revision records.
1719
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1720
1721
1722
class InterKnitRepo(InterRepository):
2652
result_set = set(
2653
self.source._eliminate_revisions_not_present(required_revisions))
2654
return self.source.revision_ids_to_search_result(result_set)
2655
2656
2657
class InterKnitRepo(InterSameDataRepository):
1723
2658
"""Optimised code paths between Knit based repositories."""
1724
2659
1725
_matching_repo_format = RepositoryFormatKnit1()
1726
"""Repository format for testing with."""
2660
@classmethod
2661
def _get_repo_format_to_test(self):
2662
from bzrlib.repofmt import knitrepo
2663
return knitrepo.RepositoryFormatKnit1()
1727
2664
1728
2665
@staticmethod
1729
2666
def is_compatible(source, target):
1733
2670
could lead to confusing results, and there is no need to be
1734
2671
overly general.
1735
2672
"""
2673
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
1736
2674
try:
1737
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1738
isinstance(target._format, (RepositoryFormatKnit1)))
2675
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2676
isinstance(target._format, RepositoryFormatKnit))
1739
2677
except AttributeError:
1740
2678
return False
2679
return are_knits and InterRepository._same_model(source, target)
1741
2680
1742
2681
@needs_write_lock
1743
def fetch(self, revision_id=None, pb=None):
2682
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1744
2683
"""See InterRepository.fetch()."""
1745
from bzrlib.fetch import KnitRepoFetcher
2684
from bzrlib.fetch import RepoFetcher
1746
2685
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1747
2686
self.source, self.source._format, self.target, self.target._format)
1748
f = KnitRepoFetcher(to_repository=self.target,
2687
f = RepoFetcher(to_repository=self.target,
1749
2688
from_repository=self.source,
1750
2689
last_revision=revision_id,
1751
pb=pb)
2690
pb=pb, find_ghosts=find_ghosts)
1752
2691
return f.count_copied, f.failed_revisions
1753
2692
1754
2693
@needs_read_lock
1755
def missing_revision_ids(self, revision_id=None):
2694
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1756
2695
"""See InterRepository.missing_revision_ids()."""
1757
2696
if revision_id is not None:
1758
2697
source_ids = self.source.get_ancestry(revision_id)
1759
assert source_ids[0] == None
2698
if source_ids[0] is not None:
2699
raise AssertionError()
1760
2700
source_ids.pop(0)
1761
2701
else:
1762
source_ids = self.source._all_possible_ids()
2702
source_ids = self.source.all_revision_ids()
1763
2703
source_ids_set = set(source_ids)
1764
2704
# source_ids is the worst possible case we may need to pull.
1765
2705
# now we want to filter source_ids against what we actually
1766
2706
# have in target, but don't try to check for existence where we know
1767
2707
# we do not have a revision as that would be pointless.
1768
target_ids = set(self.target._all_possible_ids())
2708
target_ids = set(self.target.all_revision_ids())
1769
2709
possibly_present_revisions = target_ids.intersection(source_ids_set)
1770
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2710
actually_present_revisions = set(
2711
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1771
2712
required_revisions = source_ids_set.difference(actually_present_revisions)
1772
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1773
2713
if revision_id is not None:
1774
2714
# we used get_ancestry to determine source_ids then we are assured all
1775
2715
# revisions referenced are present as they are installed in topological order.
1776
2716
# and the tip revision was validated by get_ancestry.
1777
return required_topo_revisions
2717
result_set = required_revisions
1778
2718
else:
1779
2719
# if we just grabbed the possibly available ids, then
1780
2720
# we only have an estimate of whats available and need to validate
1781
2721
# that against the revision records.
1782
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1783
2722
result_set = set(
2723
self.source._eliminate_revisions_not_present(required_revisions))
2724
return self.source.revision_ids_to_search_result(result_set)
2725
2726
2727
class InterPackRepo(InterSameDataRepository):
2728
"""Optimised code paths between Pack based repositories."""
2729
2730
@classmethod
2731
def _get_repo_format_to_test(self):
2732
from bzrlib.repofmt import pack_repo
2733
return pack_repo.RepositoryFormatKnitPack1()
2734
2735
@staticmethod
2736
def is_compatible(source, target):
2737
"""Be compatible with known Pack formats.
2738
2739
We don't test for the stores being of specific types because that
2740
could lead to confusing results, and there is no need to be
2741
overly general.
2742
"""
2743
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2744
try:
2745
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2746
isinstance(target._format, RepositoryFormatPack))
2747
except AttributeError:
2748
return False
2749
return are_packs and InterRepository._same_model(source, target)
2750
2751
@needs_write_lock
2752
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2753
"""See InterRepository.fetch()."""
2754
if (len(self.source._fallback_repositories) > 0 or
2755
len(self.target._fallback_repositories) > 0):
2756
# The pack layer is not aware of fallback repositories, so when
2757
# fetching from a stacked repository or into a stacked repository
2758
# we use the generic fetch logic which uses the VersionedFiles
2759
# attributes on repository.
2760
from bzrlib.fetch import RepoFetcher
2761
fetcher = RepoFetcher(self.target, self.source, revision_id,
2762
pb, find_ghosts)
2763
return fetcher.count_copied, fetcher.failed_revisions
2764
from bzrlib.repofmt.pack_repo import Packer
2765
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2766
self.source, self.source._format, self.target, self.target._format)
2767
self.count_copied = 0
2768
if revision_id is None:
2769
# TODO:
2770
# everything to do - use pack logic
2771
# to fetch from all packs to one without
2772
# inventory parsing etc, IFF nothing to be copied is in the target.
2773
# till then:
2774
source_revision_ids = frozenset(self.source.all_revision_ids())
2775
revision_ids = source_revision_ids - \
2776
frozenset(self.target.get_parent_map(source_revision_ids))
2777
revision_keys = [(revid,) for revid in revision_ids]
2778
index = self.target._pack_collection.revision_index.combined_index
2779
present_revision_ids = set(item[1][0] for item in
2780
index.iter_entries(revision_keys))
2781
revision_ids = set(revision_ids) - present_revision_ids
2782
# implementing the TODO will involve:
2783
# - detecting when all of a pack is selected
2784
# - avoiding as much as possible pre-selection, so the
2785
# more-core routines such as create_pack_from_packs can filter in
2786
# a just-in-time fashion. (though having a HEADS list on a
2787
# repository might make this a lot easier, because we could
2788
# sensibly detect 'new revisions' without doing a full index scan.
2789
elif _mod_revision.is_null(revision_id):
2790
# nothing to do:
2791
return (0, [])
2792
else:
2793
try:
2794
revision_ids = self.search_missing_revision_ids(revision_id,
2795
find_ghosts=find_ghosts).get_keys()
2796
except errors.NoSuchRevision:
2797
raise errors.InstallFailed([revision_id])
2798
if len(revision_ids) == 0:
2799
return (0, [])
2800
packs = self.source._pack_collection.all_packs()
2801
pack = Packer(self.target._pack_collection, packs, '.fetch',
2802
revision_ids).pack()
2803
if pack is not None:
2804
self.target._pack_collection._save_pack_names()
2805
# Trigger an autopack. This may duplicate effort as we've just done
2806
# a pack creation, but for now it is simpler to think about as
2807
# 'upload data, then repack if needed'.
2808
self.target._pack_collection.autopack()
2809
return (pack.get_revision_count(), [])
2810
else:
2811
return (0, [])
2812
2813
@needs_read_lock
2814
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2815
"""See InterRepository.missing_revision_ids().
2816
2817
:param find_ghosts: Find ghosts throughout the ancestry of
2818
revision_id.
2819
"""
2820
if not find_ghosts and revision_id is not None:
2821
return self._walk_to_common_revisions([revision_id])
2822
elif revision_id is not None:
2823
# Find ghosts: search for revisions pointing from one repository to
2824
# the other, and vice versa, anywhere in the history of revision_id.
2825
graph = self.target.get_graph(other_repository=self.source)
2826
searcher = graph._make_breadth_first_searcher([revision_id])
2827
found_ids = set()
2828
while True:
2829
try:
2830
next_revs, ghosts = searcher.next_with_ghosts()
2831
except StopIteration:
2832
break
2833
if revision_id in ghosts:
2834
raise errors.NoSuchRevision(self.source, revision_id)
2835
found_ids.update(next_revs)
2836
found_ids.update(ghosts)
2837
found_ids = frozenset(found_ids)
2838
# Double query here: should be able to avoid this by changing the
2839
# graph api further.
2840
result_set = found_ids - frozenset(
2841
self.target.get_parent_map(found_ids))
2842
else:
2843
source_ids = self.source.all_revision_ids()
2844
# source_ids is the worst possible case we may need to pull.
2845
# now we want to filter source_ids against what we actually
2846
# have in target, but don't try to check for existence where we know
2847
# we do not have a revision as that would be pointless.
2848
target_ids = set(self.target.all_revision_ids())
2849
result_set = set(source_ids).difference(target_ids)
2850
return self.source.revision_ids_to_search_result(result_set)
2851
2852
2853
class InterModel1and2(InterRepository):
2854
2855
@classmethod
2856
def _get_repo_format_to_test(self):
2857
return None
2858
2859
@staticmethod
2860
def is_compatible(source, target):
2861
if not source.supports_rich_root() and target.supports_rich_root():
2862
return True
2863
else:
2864
return False
2865
2866
@needs_write_lock
2867
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2868
"""See InterRepository.fetch()."""
2869
from bzrlib.fetch import Model1toKnit2Fetcher
2870
f = Model1toKnit2Fetcher(to_repository=self.target,
2871
from_repository=self.source,
2872
last_revision=revision_id,
2873
pb=pb, find_ghosts=find_ghosts)
2874
return f.count_copied, f.failed_revisions
2875
2876
@needs_write_lock
2877
def copy_content(self, revision_id=None):
2878
"""Make a complete copy of the content in self into destination.
2879
2880
This is a destructive operation! Do not use it on existing
2881
repositories.
2882
2883
:param revision_id: Only copy the content needed to construct
2884
revision_id and its parents.
2885
"""
2886
try:
2887
self.target.set_make_working_trees(self.source.make_working_trees())
2888
except NotImplementedError:
2889
pass
2890
# but don't bother fetching if we have the needed data now.
2891
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2892
self.target.has_revision(revision_id)):
2893
return
2894
self.target.fetch(self.source, revision_id=revision_id)
2895
2896
2897
class InterKnit1and2(InterKnitRepo):
2898
2899
@classmethod
2900
def _get_repo_format_to_test(self):
2901
return None
2902
2903
@staticmethod
2904
def is_compatible(source, target):
2905
"""Be compatible with Knit1 source and Knit3 target"""
2906
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2907
try:
2908
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2909
RepositoryFormatKnit3)
2910
from bzrlib.repofmt.pack_repo import (
2911
RepositoryFormatKnitPack1,
2912
RepositoryFormatKnitPack3,
2913
RepositoryFormatPackDevelopment0,
2914
RepositoryFormatPackDevelopment0Subtree,
2915
)
2916
nosubtrees = (
2917
RepositoryFormatKnit1,
2918
RepositoryFormatKnitPack1,
2919
RepositoryFormatPackDevelopment0,
2920
)
2921
subtrees = (
2922
RepositoryFormatKnit3,
2923
RepositoryFormatKnitPack3,
2924
RepositoryFormatPackDevelopment0Subtree,
2925
)
2926
return (isinstance(source._format, nosubtrees) and
2927
isinstance(target._format, subtrees))
2928
except AttributeError:
2929
return False
2930
2931
@needs_write_lock
2932
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2933
"""See InterRepository.fetch()."""
2934
from bzrlib.fetch import Knit1to2Fetcher
2935
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2936
self.source, self.source._format, self.target,
2937
self.target._format)
2938
f = Knit1to2Fetcher(to_repository=self.target,
2939
from_repository=self.source,
2940
last_revision=revision_id,
2941
pb=pb, find_ghosts=find_ghosts)
2942
return f.count_copied, f.failed_revisions
2943
2944
2945
class InterDifferingSerializer(InterKnitRepo):
2946
2947
@classmethod
2948
def _get_repo_format_to_test(self):
2949
return None
2950
2951
@staticmethod
2952
def is_compatible(source, target):
2953
"""Be compatible with Knit2 source and Knit3 target"""
2954
if source.supports_rich_root() != target.supports_rich_root():
2955
return False
2956
# Ideally, we'd support fetching if the source had no tree references
2957
# even if it supported them...
2958
if (getattr(source, '_format.supports_tree_reference', False) and
2959
not getattr(target, '_format.supports_tree_reference', False)):
2960
return False
2961
return True
2962
2963
@needs_write_lock
2964
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2965
"""See InterRepository.fetch()."""
2966
revision_ids = self.target.search_missing_revision_ids(self.source,
2967
revision_id, find_ghosts=find_ghosts).get_keys()
2968
revision_ids = tsort.topo_sort(
2969
self.source.get_graph().get_parent_map(revision_ids))
2970
def revisions_iterator():
2971
for current_revision_id in revision_ids:
2972
revision = self.source.get_revision(current_revision_id)
2973
tree = self.source.revision_tree(current_revision_id)
2974
try:
2975
signature = self.source.get_signature_text(
2976
current_revision_id)
2977
except errors.NoSuchRevision:
2978
signature = None
2979
yield revision, tree, signature
2980
if pb is None:
2981
my_pb = ui.ui_factory.nested_progress_bar()
2982
pb = my_pb
2983
else:
2984
my_pb = None
2985
try:
2986
install_revisions(self.target, revisions_iterator(),
2987
len(revision_ids), pb)
2988
finally:
2989
if my_pb is not None:
2990
my_pb.finished()
2991
return len(revision_ids), 0
2992
2993
2994
class InterOtherToRemote(InterRepository):
2995
2996
def __init__(self, source, target):
2997
InterRepository.__init__(self, source, target)
2998
self._real_inter = None
2999
3000
@staticmethod
3001
def is_compatible(source, target):
3002
if isinstance(target, remote.RemoteRepository):
3003
return True
3004
return False
3005
3006
def _ensure_real_inter(self):
3007
if self._real_inter is None:
3008
self.target._ensure_real()
3009
real_target = self.target._real_repository
3010
self._real_inter = InterRepository.get(self.source, real_target)
3011
3012
def copy_content(self, revision_id=None):
3013
self._ensure_real_inter()
3014
self._real_inter.copy_content(revision_id=revision_id)
3015
3016
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3017
self._ensure_real_inter()
3018
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3019
find_ghosts=find_ghosts)
3020
3021
@classmethod
3022
def _get_repo_format_to_test(self):
3023
return None
3024
3025
3026
class InterRemoteToOther(InterRepository):
3027
3028
def __init__(self, source, target):
3029
InterRepository.__init__(self, source, target)
3030
self._real_inter = None
3031
3032
@staticmethod
3033
def is_compatible(source, target):
3034
if not isinstance(source, remote.RemoteRepository):
3035
return False
3036
# Is source's model compatible with target's model?
3037
source._ensure_real()
3038
real_source = source._real_repository
3039
if isinstance(real_source, remote.RemoteRepository):
3040
raise NotImplementedError(
3041
"We don't support remote repos backed by remote repos yet.")
3042
return InterRepository._same_model(real_source, target)
3043
3044
def _ensure_real_inter(self):
3045
if self._real_inter is None:
3046
self.source._ensure_real()
3047
real_source = self.source._real_repository
3048
self._real_inter = InterRepository.get(real_source, self.target)
3049
3050
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3051
self._ensure_real_inter()
3052
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3053
find_ghosts=find_ghosts)
3054
3055
def copy_content(self, revision_id=None):
3056
self._ensure_real_inter()
3057
self._real_inter.copy_content(revision_id=revision_id)
3058
3059
@classmethod
3060
def _get_repo_format_to_test(self):
3061
return None
3062
3063
3064
3065
InterRepository.register_optimiser(InterDifferingSerializer)
3066
InterRepository.register_optimiser(InterSameDataRepository)
1784
3067
InterRepository.register_optimiser(InterWeaveRepo)
1785
3068
InterRepository.register_optimiser(InterKnitRepo)
1786
1787
1788
class RepositoryTestProviderAdapter(object):
1789
"""A tool to generate a suite testing multiple repository formats at once.
1790
1791
This is done by copying the test once for each transport and injecting
1792
the transport_server, transport_readonly_server, and bzrdir_format and
1793
repository_format classes into each copy. Each copy is also given a new id()
1794
to make it easy to identify.
1795
"""
1796
1797
def __init__(self, transport_server, transport_readonly_server, formats):
1798
self._transport_server = transport_server
1799
self._transport_readonly_server = transport_readonly_server
1800
self._formats = formats
1801
1802
def adapt(self, test):
1803
result = TestSuite()
1804
for repository_format, bzrdir_format in self._formats:
1805
new_test = deepcopy(test)
1806
new_test.transport_server = self._transport_server
1807
new_test.transport_readonly_server = self._transport_readonly_server
1808
new_test.bzrdir_format = bzrdir_format
1809
new_test.repository_format = repository_format
1810
def make_new_test_id():
1811
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1812
return lambda: new_id
1813
new_test.id = make_new_test_id()
1814
result.addTest(new_test)
1815
return result
1816
1817
1818
class InterRepositoryTestProviderAdapter(object):
1819
"""A tool to generate a suite testing multiple inter repository formats.
1820
1821
This is done by copying the test once for each interrepo provider and injecting
1822
the transport_server, transport_readonly_server, repository_format and
1823
repository_to_format classes into each copy.
1824
Each copy is also given a new id() to make it easy to identify.
1825
"""
1826
1827
def __init__(self, transport_server, transport_readonly_server, formats):
1828
self._transport_server = transport_server
1829
self._transport_readonly_server = transport_readonly_server
1830
self._formats = formats
1831
1832
def adapt(self, test):
1833
result = TestSuite()
1834
for interrepo_class, repository_format, repository_format_to in self._formats:
1835
new_test = deepcopy(test)
1836
new_test.transport_server = self._transport_server
1837
new_test.transport_readonly_server = self._transport_readonly_server
1838
new_test.interrepo_class = interrepo_class
1839
new_test.repository_format = repository_format
1840
new_test.repository_format_to = repository_format_to
1841
def make_new_test_id():
1842
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1843
return lambda: new_id
1844
new_test.id = make_new_test_id()
1845
result.addTest(new_test)
1846
return result
1847
1848
@staticmethod
1849
def default_test_list():
1850
"""Generate the default list of interrepo permutations to test."""
1851
result = []
1852
# test the default InterRepository between format 6 and the current
1853
# default format.
1854
# XXX: robertc 20060220 reinstate this when there are two supported
1855
# formats which do not have an optimal code path between them.
1856
result.append((InterRepository,
1857
RepositoryFormat6(),
1858
RepositoryFormatKnit1()))
1859
for optimiser in InterRepository._optimisers:
1860
result.append((optimiser,
1861
optimiser._matching_repo_format,
1862
optimiser._matching_repo_format
1863
))
1864
# if there are specific combinations we want to use, we can add them
1865
# here.
1866
return result
3069
InterRepository.register_optimiser(InterModel1and2)
3070
InterRepository.register_optimiser(InterKnit1and2)
3071
InterRepository.register_optimiser(InterPackRepo)
3072
InterRepository.register_optimiser(InterOtherToRemote)
3073
InterRepository.register_optimiser(InterRemoteToOther)
1867
3074
1868
3075
1869
3076
class CopyConverter(object):
1895
3102
self.step('Moving repository to repository.backup')
1896
3103
self.repo_dir.transport.move('repository', 'repository.backup')
1897
3104
backup_transport = self.repo_dir.transport.clone('repository.backup')
3105
repo._format.check_conversion_target(self.target_format)
1898
3106
self.source_repo = repo._format.open(self.repo_dir,
1899
3107
_found=True,
1900
3108
_override_transport=backup_transport)
1917
3125
self.pb.update(message, self.count, self.total)
1918
3126
1919
3127
1920
class CommitBuilder(object):
1921
"""Provides an interface to build up a commit.
1922
1923
This allows describing a tree to be committed without needing to
1924
know the internals of the format of the repository.
1925
"""
1926
def __init__(self, repository, parents, config, timestamp=None,
1927
timezone=None, committer=None, revprops=None,
1928
revision_id=None):
1929
"""Initiate a CommitBuilder.
1930
1931
:param repository: Repository to commit to.
1932
:param parents: Revision ids of the parents of the new revision.
1933
:param config: Configuration to use.
1934
:param timestamp: Optional timestamp recorded for commit.
1935
:param timezone: Optional timezone for timestamp.
1936
:param committer: Optional committer to set for commit.
1937
:param revprops: Optional dictionary of revision properties.
1938
:param revision_id: Optional revision id.
1939
"""
1940
self._config = config
1941
1942
if committer is None:
1943
self._committer = self._config.username()
1944
else:
1945
assert isinstance(committer, basestring), type(committer)
1946
self._committer = committer
1947
1948
self.new_inventory = Inventory()
1949
self._new_revision_id = revision_id
1950
self.parents = parents
1951
self.repository = repository
1952
1953
self._revprops = {}
1954
if revprops is not None:
1955
self._revprops.update(revprops)
1956
1957
if timestamp is None:
1958
self._timestamp = time.time()
1959
else:
1960
self._timestamp = long(timestamp)
1961
1962
if timezone is None:
1963
self._timezone = local_time_offset()
1964
else:
1965
self._timezone = int(timezone)
1966
1967
self._generate_revision_if_needed()
1968
1969
def commit(self, message):
1970
"""Make the actual commit.
1971
1972
:return: The revision id of the recorded revision.
1973
"""
1974
rev = Revision(timestamp=self._timestamp,
1975
timezone=self._timezone,
1976
committer=self._committer,
1977
message=message,
1978
inventory_sha1=self.inv_sha1,
1979
revision_id=self._new_revision_id,
1980
properties=self._revprops)
1981
rev.parent_ids = self.parents
1982
self.repository.add_revision(self._new_revision_id, rev,
1983
self.new_inventory, self._config)
1984
return self._new_revision_id
1985
1986
def finish_inventory(self):
1987
"""Tell the builder that the inventory is finished."""
1988
self.new_inventory.revision_id = self._new_revision_id
1989
self.inv_sha1 = self.repository.add_inventory(
1990
self._new_revision_id,
1991
self.new_inventory,
1992
self.parents
1993
)
1994
1995
def _gen_revision_id(self):
1996
"""Return new revision-id."""
1997
s = '%s-%s-' % (self._config.user_email(),
1998
compact_date(self._timestamp))
1999
s += hexlify(rand_bytes(8))
2000
return s
2001
2002
def _generate_revision_if_needed(self):
2003
"""Create a revision id if None was supplied.
2004
2005
If the repository can not support user-specified revision ids
2006
they should override this function and raise UnsupportedOperation
2007
if _new_revision_id is not None.
2008
2009
:raises: UnsupportedOperation
2010
"""
2011
if self._new_revision_id is None:
2012
self._new_revision_id = self._gen_revision_id()
2013
2014
def record_entry_contents(self, ie, parent_invs, path, tree):
2015
"""Record the content of ie from tree into the commit if needed.
2016
2017
:param ie: An inventory entry present in the commit.
2018
:param parent_invs: The inventories of the parent revisions of the
2019
commit.
2020
:param path: The path the entry is at in the tree.
2021
:param tree: The tree which contains this entry and should be used to
2022
obtain content.
2023
"""
2024
self.new_inventory.add(ie)
2025
2026
# ie.revision is always None if the InventoryEntry is considered
2027
# for committing. ie.snapshot will record the correct revision
2028
# which may be the sole parent if it is untouched.
2029
if ie.revision is not None:
2030
return
2031
previous_entries = ie.find_previous_heads(
2032
parent_invs,
2033
self.repository.weave_store,
2034
self.repository.get_transaction())
2035
# we are creating a new revision for ie in the history store
2036
# and inventory.
2037
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2038
2039
def modified_directory(self, file_id, file_parents):
2040
"""Record the presence of a symbolic link.
2041
2042
:param file_id: The file_id of the link to record.
2043
:param file_parents: The per-file parent revision ids.
2044
"""
2045
self._add_text_to_weave(file_id, [], file_parents.keys())
2046
2047
def modified_file_text(self, file_id, file_parents,
2048
get_content_byte_lines, text_sha1=None,
2049
text_size=None):
2050
"""Record the text of file file_id
2051
2052
:param file_id: The file_id of the file to record the text of.
2053
:param file_parents: The per-file parent revision ids.
2054
:param get_content_byte_lines: A callable which will return the byte
2055
lines for the file.
2056
:param text_sha1: Optional SHA1 of the file contents.
2057
:param text_size: Optional size of the file contents.
2058
"""
2059
mutter('storing text of file {%s} in revision {%s} into %r',
2060
file_id, self._new_revision_id, self.repository.weave_store)
2061
# special case to avoid diffing on renames or
2062
# reparenting
2063
if (len(file_parents) == 1
2064
and text_sha1 == file_parents.values()[0].text_sha1
2065
and text_size == file_parents.values()[0].text_size):
2066
previous_ie = file_parents.values()[0]
2067
versionedfile = self.repository.weave_store.get_weave(file_id,
2068
self.repository.get_transaction())
2069
versionedfile.clone_text(self._new_revision_id,
2070
previous_ie.revision, file_parents.keys())
2071
return text_sha1, text_size
2072
else:
2073
new_lines = get_content_byte_lines()
2074
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2075
# should return the SHA1 and size
2076
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2077
return bzrlib.osutils.sha_strings(new_lines), \
2078
sum(map(len, new_lines))
2079
2080
def modified_link(self, file_id, file_parents, link_target):
2081
"""Record the presence of a symbolic link.
2082
2083
:param file_id: The file_id of the link to record.
2084
:param file_parents: The per-file parent revision ids.
2085
:param link_target: Target location of this link.
2086
"""
2087
self._add_text_to_weave(file_id, [], file_parents.keys())
2088
2089
def _add_text_to_weave(self, file_id, new_lines, parents):
2090
versionedfile = self.repository.weave_store.get_weave_or_empty(
2091
file_id, self.repository.get_transaction())
2092
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2093
versionedfile.clear_cache()
2094
2095
2096
# Copied from xml.sax.saxutils
3128
_unescape_map = {
3129
'apos':"'",
3130
'quot':'"',
3131
'amp':'&',
3132
'lt':'<',
3133
'gt':'>'
3134
}
3135
3136
3137
def _unescaper(match, _map=_unescape_map):
3138
code = match.group(1)
3139
try:
3140
return _map[code]
3141
except KeyError:
3142
if not code.startswith('#'):
3143
raise
3144
return unichr(int(code[1:])).encode('utf8')
3145
3146
3147
_unescape_re = None
3148
3149
2097
3150
def _unescape_xml(data):
2098
"""Unescape &, <, and > in a string of data.
2099
"""
2100
data = data.replace("<", "<")
2101
data = data.replace(">", ">")
2102
# must do ampersand last
2103
return data.replace("&", "&")
3151
"""Unescape predefined XML entities in a string of data."""
3152
global _unescape_re
3153
if _unescape_re is None:
3154
_unescape_re = re.compile('\&([^;]*);')
3155
return _unescape_re.sub(_unescaper, data)
3156
3157
3158
class _VersionedFileChecker(object):
3159
3160
def __init__(self, repository):
3161
self.repository = repository
3162
self.text_index = self.repository._generate_text_key_index()
3163
3164
def calculate_file_version_parents(self, text_key):
3165
"""Calculate the correct parents for a file version according to
3166
the inventories.
3167
"""
3168
parent_keys = self.text_index[text_key]
3169
if parent_keys == [_mod_revision.NULL_REVISION]:
3170
return ()
3171
return tuple(parent_keys)
3172
3173
def check_file_version_parents(self, texts, progress_bar=None):
3174
"""Check the parents stored in a versioned file are correct.
3175
3176
It also detects file versions that are not referenced by their
3177
corresponding revision's inventory.
3178
3179
:returns: A tuple of (wrong_parents, dangling_file_versions).
3180
wrong_parents is a dict mapping {revision_id: (stored_parents,
3181
correct_parents)} for each revision_id where the stored parents
3182
are not correct. dangling_file_versions is a set of (file_id,
3183
revision_id) tuples for versions that are present in this versioned
3184
file, but not used by the corresponding inventory.
3185
"""
3186
wrong_parents = {}
3187
self.file_ids = set([file_id for file_id, _ in
3188
self.text_index.iterkeys()])
3189
# text keys is now grouped by file_id
3190
n_weaves = len(self.file_ids)
3191
files_in_revisions = {}
3192
revisions_of_files = {}
3193
n_versions = len(self.text_index)
3194
progress_bar.update('loading text store', 0, n_versions)
3195
parent_map = self.repository.texts.get_parent_map(self.text_index)
3196
# On unlistable transports this could well be empty/error...
3197
text_keys = self.repository.texts.keys()
3198
unused_keys = frozenset(text_keys) - set(self.text_index)
3199
for num, key in enumerate(self.text_index.iterkeys()):
3200
if progress_bar is not None:
3201
progress_bar.update('checking text graph', num, n_versions)
3202
correct_parents = self.calculate_file_version_parents(key)
3203
try:
3204
knit_parents = parent_map[key]
3205
except errors.RevisionNotPresent:
3206
# Missing text!
3207
knit_parents = None
3208
if correct_parents != knit_parents:
3209
wrong_parents[key] = (knit_parents, correct_parents)
3210
return wrong_parents, unused_keys
3211
3212
3213
def _old_get_graph(repository, revision_id):
3214
"""DO NOT USE. That is all. I'm serious."""
3215
graph = repository.get_graph()
3216
revision_graph = dict(((key, value) for key, value in
3217
graph.iter_ancestry([revision_id]) if value is not None))
3218
return _strip_NULL_ghosts(revision_graph)
3219
3220
3221
def _strip_NULL_ghosts(revision_graph):
3222
"""Also don't use this. more compatibility code for unmigrated clients."""
3223
# Filter ghosts, and null:
3224
if _mod_revision.NULL_REVISION in revision_graph:
3225
del revision_graph[_mod_revision.NULL_REVISION]
3226
for key, parents in revision_graph.items():
3227
revision_graph[key] = tuple(parent for parent in parents if parent
3228
in revision_graph)
3229
return revision_graph
Loggerhead is a web-based interface for Breezy
Version: 2.0.1