« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 1770.2.1
- compare with another revision
- download diff
- download tarball
- view history from revision 1770.2.1
- Committer: Aaron Bentley
- Date: 2006-06-14 19:45:57 UTC
- mto: This revision was merged to the branch mainline in revision 1777.
- Revision ID: abentley@panoramicfeedback.com-20060614194557-6b499aa1cf03f7e6
Use create_signature for signing policy, deprecate check_signatures for this
- files added:
- NEWS.developers
- bzrlib/bundle/old
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/util/urlgrabber
- files removed:
- 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/transport/ftp
- 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/packaging
- tools/win32
- files renamed:
- doc/developers/HACKING.txt => HACKING
- bzrlib/bundle/bundle_data.py => bzrlib/bundle/read_bundle.py
- bzrlib/bundle/serializer/v08.py => bzrlib/bundle/serializer/v07.py
- bzrlib/deprecated_graph.py => bzrlib/graph.py
- bzrlib/_patiencediff_py.py => bzrlib/patiencediff.py
- bzrlib/tests/http_utils.py => bzrlib/tests/HTTPTestUtil.py
- bzrlib/tests/per_repository/ => bzrlib/tests/repository_implementations/
- bzrlib/tests/test_commands.py => bzrlib/tests/test_command.py
- bzrlib/tests/test_deprecated_graph.py => bzrlib/tests/test_graph.py
- bzrlib/tests/test_revisionspec.py => bzrlib/tests/test_revisionnamespaces.py
- bzrlib/tests/per_repository/test_revision.py => bzrlib/tests/test_revprops.py
- bzrlib/transport/ftp/__init__.py => bzrlib/transport/ftp.py
- bzrlib/win32utils.py => bzrlib/win32console.py
- bzrlib/xml8.py => bzrlib/xml5.py
- bzrlib/help_topics/en/configuration.txt => doc/configuration.txt
- doc/en/user-guide/plugins.txt => doc/plugins.txt
- doc/en/user-guide/setting_up_email.txt => doc/setting_up_email.txt
- doc/en/user-guide/specifying_revisions.txt => doc/specifying_revisions.txt
- doc/en/tutorials/tutorial.txt => doc/tutorial.txt
- doc/en/user-guide/using_aliases.txt => doc/using_aliases.txt
- files modified:
- .bzrignore
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
2
#
1
# Copyright (C) 2005, 2006 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 bzrlib.lazy_import import lazy_import
18
lazy_import(globals(), """
19
import cStringIO
17
from binascii import hexlify
18
from copy import deepcopy
19
from cStringIO import StringIO
20
20
import re
21
21
import time
22
23
from bzrlib import (
24
bzrdir,
25
check,
26
debug,
27
errors,
28
generate_ids,
29
gpg,
30
graph,
31
lazy_regex,
32
lockable_files,
33
lockdir,
34
lru_cache,
35
osutils,
36
remote,
37
revision as _mod_revision,
38
symbol_versioning,
39
tsort,
40
ui,
41
)
42
from bzrlib.bundle import serializer
43
from bzrlib.revisiontree import RevisionTree
44
from bzrlib.store.versioned import VersionedFileStore
45
from bzrlib.testament import Testament
46
""")
47
48
from bzrlib import registry
22
from unittest import TestSuite
23
24
import bzrlib.bzrdir as bzrdir
49
25
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
50
30
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
53
deprecated_method,
54
one_one,
55
one_two,
56
one_six,
57
)
58
from bzrlib.trace import mutter, mutter_callsite, warning
59
60
61
# Old formats display a warning, but only once
62
_deprecation_warning_done = False
63
64
65
class CommitBuilder(object):
66
"""Provides an interface to build up a commit.
67
68
This allows describing a tree to be committed without needing to
69
know the internals of the format of the repository.
70
"""
71
72
# all clients should supply tree roots.
73
record_root_entry = True
74
# the default CommitBuilder does not manage trees whose root is versioned.
75
_versioned_root = False
76
77
def __init__(self, repository, parents, config, timestamp=None,
78
timezone=None, committer=None, revprops=None,
79
revision_id=None):
80
"""Initiate a CommitBuilder.
81
82
:param repository: Repository to commit to.
83
:param parents: Revision ids of the parents of the new revision.
84
:param config: Configuration to use.
85
:param timestamp: Optional timestamp recorded for commit.
86
:param timezone: Optional timezone for timestamp.
87
:param committer: Optional committer to set for commit.
88
:param revprops: Optional dictionary of revision properties.
89
:param revision_id: Optional revision id.
90
"""
91
self._config = config
92
93
if committer is None:
94
self._committer = self._config.username()
95
else:
96
self._committer = committer
97
98
self.new_inventory = Inventory(None)
99
self._new_revision_id = revision_id
100
self.parents = parents
101
self.repository = repository
102
103
self._revprops = {}
104
if revprops is not None:
105
self._revprops.update(revprops)
106
107
if timestamp is None:
108
timestamp = time.time()
109
# Restrict resolution to 1ms
110
self._timestamp = round(timestamp, 3)
111
112
if timezone is None:
113
self._timezone = osutils.local_time_offset()
114
else:
115
self._timezone = int(timezone)
116
117
self._generate_revision_if_needed()
118
self.__heads = graph.HeadsCache(repository.get_graph()).heads
119
120
def commit(self, message):
121
"""Make the actual commit.
122
123
:return: The revision id of the recorded revision.
124
"""
125
rev = _mod_revision.Revision(
126
timestamp=self._timestamp,
127
timezone=self._timezone,
128
committer=self._committer,
129
message=message,
130
inventory_sha1=self.inv_sha1,
131
revision_id=self._new_revision_id,
132
properties=self._revprops)
133
rev.parent_ids = self.parents
134
self.repository.add_revision(self._new_revision_id, rev,
135
self.new_inventory, self._config)
136
self.repository.commit_write_group()
137
return self._new_revision_id
138
139
def abort(self):
140
"""Abort the commit that is being built.
141
"""
142
self.repository.abort_write_group()
143
144
def revision_tree(self):
145
"""Return the tree that was just committed.
146
147
After calling commit() this can be called to get a RevisionTree
148
representing the newly committed tree. This is preferred to
149
calling Repository.revision_tree() because that may require
150
deserializing the inventory, while we already have a copy in
151
memory.
152
"""
153
return RevisionTree(self.repository, self.new_inventory,
154
self._new_revision_id)
155
156
def finish_inventory(self):
157
"""Tell the builder that the inventory is finished."""
158
if self.new_inventory.root is None:
159
raise AssertionError('Root entry should be supplied to'
160
' record_entry_contents, as of bzr 0.10.')
161
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
162
self.new_inventory.revision_id = self._new_revision_id
163
self.inv_sha1 = self.repository.add_inventory(
164
self._new_revision_id,
165
self.new_inventory,
166
self.parents
167
)
168
169
def _gen_revision_id(self):
170
"""Return new revision-id."""
171
return generate_ids.gen_revision_id(self._config.username(),
172
self._timestamp)
173
174
def _generate_revision_if_needed(self):
175
"""Create a revision id if None was supplied.
176
177
If the repository can not support user-specified revision ids
178
they should override this function and raise CannotSetRevisionId
179
if _new_revision_id is not None.
180
181
:raises: CannotSetRevisionId
182
"""
183
if self._new_revision_id is None:
184
self._new_revision_id = self._gen_revision_id()
185
self.random_revid = True
186
else:
187
self.random_revid = False
188
189
def _heads(self, file_id, revision_ids):
190
"""Calculate the graph heads for revision_ids in the graph of file_id.
191
192
This can use either a per-file graph or a global revision graph as we
193
have an identity relationship between the two graphs.
194
"""
195
return self.__heads(revision_ids)
196
197
def _check_root(self, ie, parent_invs, tree):
198
"""Helper for record_entry_contents.
199
200
:param ie: An entry being added.
201
:param parent_invs: The inventories of the parent revisions of the
202
commit.
203
:param tree: The tree that is being committed.
204
"""
205
# In this revision format, root entries have no knit or weave When
206
# serializing out to disk and back in root.revision is always
207
# _new_revision_id
208
ie.revision = self._new_revision_id
209
210
def _get_delta(self, ie, basis_inv, path):
211
"""Get a delta against the basis inventory for ie."""
212
if ie.file_id not in basis_inv:
213
# add
214
return (None, path, ie.file_id, ie)
215
elif ie != basis_inv[ie.file_id]:
216
# common but altered
217
# TODO: avoid tis id2path call.
218
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
219
else:
220
# common, unaltered
221
return None
222
223
def record_entry_contents(self, ie, parent_invs, path, tree,
224
content_summary):
225
"""Record the content of ie from tree into the commit if needed.
226
227
Side effect: sets ie.revision when unchanged
228
229
:param ie: An inventory entry present in the commit.
230
:param parent_invs: The inventories of the parent revisions of the
231
commit.
232
:param path: The path the entry is at in the tree.
233
:param tree: The tree which contains this entry and should be used to
234
obtain content.
235
:param content_summary: Summary data from the tree about the paths
236
content - stat, length, exec, sha/link target. This is only
237
accessed when the entry has a revision of None - that is when it is
238
a candidate to commit.
239
:return: A tuple (change_delta, version_recorded, fs_hash).
240
change_delta is an inventory_delta change for this entry against
241
the basis tree of the commit, or None if no change occured against
242
the basis tree.
243
version_recorded is True if a new version of the entry has been
244
recorded. For instance, committing a merge where a file was only
245
changed on the other side will return (delta, False).
246
fs_hash is either None, or the hash details for the path (currently
247
a tuple of the contents sha1 and the statvalue returned by
248
tree.get_file_with_stat()).
249
"""
250
if self.new_inventory.root is None:
251
if ie.parent_id is not None:
252
raise errors.RootMissing()
253
self._check_root(ie, parent_invs, tree)
254
if ie.revision is None:
255
kind = content_summary[0]
256
else:
257
# ie is carried over from a prior commit
258
kind = ie.kind
259
# XXX: repository specific check for nested tree support goes here - if
260
# the repo doesn't want nested trees we skip it ?
261
if (kind == 'tree-reference' and
262
not self.repository._format.supports_tree_reference):
263
# mismatch between commit builder logic and repository:
264
# this needs the entry creation pushed down into the builder.
265
raise NotImplementedError('Missing repository subtree support.')
266
self.new_inventory.add(ie)
267
268
# TODO: slow, take it out of the inner loop.
269
try:
270
basis_inv = parent_invs[0]
271
except IndexError:
272
basis_inv = Inventory(root_id=None)
273
274
# ie.revision is always None if the InventoryEntry is considered
275
# for committing. We may record the previous parents revision if the
276
# content is actually unchanged against a sole head.
277
if ie.revision is not None:
278
if not self._versioned_root and path == '':
279
# repositories that do not version the root set the root's
280
# revision to the new commit even when no change occurs, and
281
# this masks when a change may have occurred against the basis,
282
# so calculate if one happened.
283
if ie.file_id in basis_inv:
284
delta = (basis_inv.id2path(ie.file_id), path,
285
ie.file_id, ie)
286
else:
287
# add
288
delta = (None, path, ie.file_id, ie)
289
return delta, False, None
290
else:
291
# we don't need to commit this, because the caller already
292
# determined that an existing revision of this file is
293
# appropriate. If its not being considered for committing then
294
# it and all its parents to the root must be unaltered so
295
# no-change against the basis.
296
if ie.revision == self._new_revision_id:
297
raise AssertionError("Impossible situation, a skipped "
298
"inventory entry (%r) claims to be modified in this "
299
"commit (%r).", (ie, self._new_revision_id))
300
return None, False, None
301
# XXX: Friction: parent_candidates should return a list not a dict
302
# so that we don't have to walk the inventories again.
303
parent_candiate_entries = ie.parent_candidates(parent_invs)
304
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
305
heads = []
306
for inv in parent_invs:
307
if ie.file_id in inv:
308
old_rev = inv[ie.file_id].revision
309
if old_rev in head_set:
310
heads.append(inv[ie.file_id].revision)
311
head_set.remove(inv[ie.file_id].revision)
312
313
store = False
314
# now we check to see if we need to write a new record to the
315
# file-graph.
316
# We write a new entry unless there is one head to the ancestors, and
317
# the kind-derived content is unchanged.
318
319
# Cheapest check first: no ancestors, or more the one head in the
320
# ancestors, we write a new node.
321
if len(heads) != 1:
322
store = True
323
if not store:
324
# There is a single head, look it up for comparison
325
parent_entry = parent_candiate_entries[heads[0]]
326
# if the non-content specific data has changed, we'll be writing a
327
# node:
328
if (parent_entry.parent_id != ie.parent_id or
329
parent_entry.name != ie.name):
330
store = True
331
# now we need to do content specific checks:
332
if not store:
333
# if the kind changed the content obviously has
334
if kind != parent_entry.kind:
335
store = True
336
# Stat cache fingerprint feedback for the caller - None as we usually
337
# don't generate one.
338
fingerprint = None
339
if kind == 'file':
340
if content_summary[2] is None:
341
raise ValueError("Files must not have executable = None")
342
if not store:
343
if (# if the file length changed we have to store:
344
parent_entry.text_size != content_summary[1] or
345
# if the exec bit has changed we have to store:
346
parent_entry.executable != content_summary[2]):
347
store = True
348
elif parent_entry.text_sha1 == content_summary[3]:
349
# all meta and content is unchanged (using a hash cache
350
# hit to check the sha)
351
ie.revision = parent_entry.revision
352
ie.text_size = parent_entry.text_size
353
ie.text_sha1 = parent_entry.text_sha1
354
ie.executable = parent_entry.executable
355
return self._get_delta(ie, basis_inv, path), False, None
356
else:
357
# Either there is only a hash change(no hash cache entry,
358
# or same size content change), or there is no change on
359
# this file at all.
360
# Provide the parent's hash to the store layer, so that the
361
# content is unchanged we will not store a new node.
362
nostore_sha = parent_entry.text_sha1
363
if store:
364
# We want to record a new node regardless of the presence or
365
# absence of a content change in the file.
366
nostore_sha = None
367
ie.executable = content_summary[2]
368
file_obj, stat_value = tree.get_file_with_stat(ie.file_id, path)
369
try:
370
lines = file_obj.readlines()
371
finally:
372
file_obj.close()
373
try:
374
ie.text_sha1, ie.text_size = self._add_text_to_weave(
375
ie.file_id, lines, heads, nostore_sha)
376
# Let the caller know we generated a stat fingerprint.
377
fingerprint = (ie.text_sha1, stat_value)
378
except errors.ExistingContent:
379
# Turns out that the file content was unchanged, and we were
380
# only going to store a new node if it was changed. Carry over
381
# the entry.
382
ie.revision = parent_entry.revision
383
ie.text_size = parent_entry.text_size
384
ie.text_sha1 = parent_entry.text_sha1
385
ie.executable = parent_entry.executable
386
return self._get_delta(ie, basis_inv, path), False, None
387
elif kind == 'directory':
388
if not store:
389
# all data is meta here, nothing specific to directory, so
390
# carry over:
391
ie.revision = parent_entry.revision
392
return self._get_delta(ie, basis_inv, path), False, None
393
lines = []
394
self._add_text_to_weave(ie.file_id, lines, heads, None)
395
elif kind == 'symlink':
396
current_link_target = content_summary[3]
397
if not store:
398
# symlink target is not generic metadata, check if it has
399
# changed.
400
if current_link_target != parent_entry.symlink_target:
401
store = True
402
if not store:
403
# unchanged, carry over.
404
ie.revision = parent_entry.revision
405
ie.symlink_target = parent_entry.symlink_target
406
return self._get_delta(ie, basis_inv, path), False, None
407
ie.symlink_target = current_link_target
408
lines = []
409
self._add_text_to_weave(ie.file_id, lines, heads, None)
410
elif kind == 'tree-reference':
411
if not store:
412
if content_summary[3] != parent_entry.reference_revision:
413
store = True
414
if not store:
415
# unchanged, carry over.
416
ie.reference_revision = parent_entry.reference_revision
417
ie.revision = parent_entry.revision
418
return self._get_delta(ie, basis_inv, path), False, None
419
ie.reference_revision = content_summary[3]
420
lines = []
421
self._add_text_to_weave(ie.file_id, lines, heads, None)
422
else:
423
raise NotImplementedError('unknown kind')
424
ie.revision = self._new_revision_id
425
return self._get_delta(ie, basis_inv, path), True, fingerprint
426
427
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
428
# Note: as we read the content directly from the tree, we know its not
429
# been turned into unicode or badly split - but a broken tree
430
# implementation could give us bad output from readlines() so this is
431
# not a guarantee of safety. What would be better is always checking
432
# the content during test suite execution. RBC 20070912
433
parent_keys = tuple((file_id, parent) for parent in parents)
434
return self.repository.texts.add_lines(
435
(file_id, self._new_revision_id), parent_keys, new_lines,
436
nostore_sha=nostore_sha, random_id=self.random_revid,
437
check_content=False)[0:2]
438
439
440
class RootCommitBuilder(CommitBuilder):
441
"""This commitbuilder actually records the root id"""
442
443
# the root entry gets versioned properly by this builder.
444
_versioned_root = True
445
446
def _check_root(self, ie, parent_invs, tree):
447
"""Helper for record_entry_contents.
448
449
:param ie: An entry being added.
450
:param parent_invs: The inventories of the parent revisions of the
451
commit.
452
:param tree: The tree that is being committed.
453
"""
454
455
456
######################################################################
457
# Repositories
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
458
50
459
51
class Repository(object):
460
52
"""Repository holding history for one or more branches.
463
55
revisions and file history. It's normally accessed only by the Branch,
464
56
which views a particular line of development through that history.
465
57
466
The Repository builds on top of some byte storage facilies (the revisions,
467
signatures, inventories and texts attributes) and a Transport, which
468
respectively provide byte storage and a means to access the (possibly
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
469
60
remote) disk.
470
471
The byte storage facilities are addressed via tuples, which we refer to
472
as 'keys' throughout the code base. Revision_keys, inventory_keys and
473
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
474
(file_id, revision_id). We use this interface because it allows low
475
friction with the underlying code that implements disk indices, network
476
encoding and other parts of bzrlib.
477
478
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
479
the serialised revisions for the repository. This can be used to obtain
480
revision graph information or to access raw serialised revisions.
481
The result of trying to insert data into the repository via this store
482
is undefined: it should be considered read-only except for implementors
483
of repositories.
484
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
485
the serialised signatures for the repository. This can be used to
486
obtain access to raw serialised signatures. The result of trying to
487
insert data into the repository via this store is undefined: it should
488
be considered read-only except for implementors of repositories.
489
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
490
the serialised inventories for the repository. This can be used to
491
obtain unserialised inventories. The result of trying to insert data
492
into the repository via this store is undefined: it should be
493
considered read-only except for implementors of repositories.
494
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
495
texts of files and directories for the repository. This can be used to
496
obtain file texts or file graphs. Note that Repository.iter_file_bytes
497
is usually a better interface for accessing file texts.
498
The result of trying to insert data into the repository via this store
499
is undefined: it should be considered read-only except for implementors
500
of repositories.
501
:ivar _transport: Transport for file access to repository, typically
502
pointing to .bzr/repository.
503
61
"""
504
62
505
# What class to use for a CommitBuilder. Often its simpler to change this
506
# in a Repository class subclass rather than to override
507
# get_commit_builder.
508
_commit_builder_class = CommitBuilder
509
# The search regex used by xml based repositories to determine what things
510
# where changed in a single commit.
511
_file_ids_altered_regex = lazy_regex.lazy_compile(
512
r'file_id="(?P<file_id>[^"]+)"'
513
r'.* revision="(?P<revision_id>[^"]+)"'
514
)
515
516
def abort_write_group(self):
517
"""Commit the contents accrued within the current write group.
518
519
:seealso: start_write_group.
520
"""
521
if self._write_group is not self.get_transaction():
522
# has an unlock or relock occured ?
523
raise errors.BzrError('mismatched lock context and write group.')
524
self._abort_write_group()
525
self._write_group = None
526
527
def _abort_write_group(self):
528
"""Template method for per-repository write group cleanup.
529
530
This is called during abort before the write group is considered to be
531
finished and should cleanup any internal state accrued during the write
532
group. There is no requirement that data handed to the repository be
533
*not* made available - this is not a rollback - but neither should any
534
attempt be made to ensure that data added is fully commited. Abort is
535
invoked when an error has occured so futher disk or network operations
536
may not be possible or may error and if possible should not be
537
attempted.
538
"""
539
540
def add_fallback_repository(self, repository):
541
"""Add a repository to use for looking up data not held locally.
542
543
:param repository: A repository.
544
"""
545
if not self._format.supports_external_lookups:
546
raise errors.UnstackableRepositoryFormat(self._format, self.base)
547
self._check_fallback_repository(repository)
548
self._fallback_repositories.append(repository)
549
self.texts.add_fallback_versioned_files(repository.texts)
550
self.inventories.add_fallback_versioned_files(repository.inventories)
551
self.revisions.add_fallback_versioned_files(repository.revisions)
552
self.signatures.add_fallback_versioned_files(repository.signatures)
553
554
def _check_fallback_repository(self, repository):
555
"""Check that this repository can fallback to repository safely.
556
557
Raise an error if not.
558
559
:param repository: A repository to fallback to.
560
"""
561
return InterRepository._assert_same_model(self, repository)
562
563
def add_inventory(self, revision_id, inv, parents):
564
"""Add the inventory inv to the repository as revision_id.
565
566
:param parents: The revision ids of the parents that revision_id
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
567
68
is known to have and are in the repository already.
568
69
569
:returns: The validator(which is a sha1 digest, though what is sha'd is
570
repository format specific) of the serialized inventory.
70
returns the sha1 of the serialized inventory.
571
71
"""
572
if not self.is_in_write_group():
573
raise AssertionError("%r not in write group" % (self,))
574
_mod_revision.check_not_reserved_id(revision_id)
575
if not (inv.revision_id is None or inv.revision_id == revision_id):
576
raise AssertionError(
577
"Mismatch between inventory revision"
578
" id and insertion revid (%r, %r)"
579
% (inv.revision_id, revision_id))
580
if inv.root is None:
581
raise AssertionError()
582
inv_lines = self._serialise_inventory_to_lines(inv)
583
return self._inventory_add_lines(revision_id, parents,
584
inv_lines, check_content=False)
585
586
def _inventory_add_lines(self, revision_id, parents, lines,
587
check_content=True):
588
"""Store lines in inv_vf and return the sha1 of the inventory."""
589
parents = [(parent,) for parent in parents]
590
return self.inventories.add_lines((revision_id,), parents, lines,
591
check_content=check_content)[0]
592
593
def add_revision(self, revision_id, rev, inv=None, config=None):
594
"""Add rev to the revision store as revision_id.
595
596
:param revision_id: the revision id to use.
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.
597
95
:param rev: The revision object.
598
96
:param inv: The inventory for the revision. if None, it will be looked
599
97
up in the inventory storer
601
99
If supplied its signature_needed method will be used
602
100
to determine if a signature should be made.
603
101
"""
604
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
605
# rev.parent_ids?
606
_mod_revision.check_not_reserved_id(revision_id)
607
102
if config is not None and config.signature_needed():
608
103
if inv is None:
609
inv = self.get_inventory(revision_id)
104
inv = self.get_inventory(rev_id)
610
105
plaintext = Testament(rev, inv).as_short_text()
611
106
self.store_revision_signature(
612
gpg.GPGStrategy(config), plaintext, revision_id)
613
# check inventory present
614
if not self.inventories.get_parent_map([(revision_id,)]):
107
gpg.GPGStrategy(config), plaintext, rev_id)
108
if not rev_id in self.get_inventory_weave():
615
109
if inv is None:
616
raise errors.WeaveRevisionNotPresent(revision_id,
617
self.inventories)
110
raise errors.WeaveRevisionNotPresent(rev_id,
111
self.get_inventory_weave())
618
112
else:
619
113
# yes, this is not suitable for adding with ghosts.
620
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
621
rev.parent_ids)
622
else:
623
key = (revision_id,)
624
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
625
self._add_revision(rev)
626
627
def _add_revision(self, revision):
628
text = self._serializer.write_revision_to_string(revision)
629
key = (revision.revision_id,)
630
parents = tuple((parent,) for parent in revision.parent_ids)
631
self.revisions.add_lines(key, parents, osutils.split_lines(text))
632
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)
633
123
def all_revision_ids(self):
634
124
"""Returns a list of all the revision ids in the repository.
635
125
636
This is conceptually deprecated because code should generally work on
637
the graph reachable from a particular revision, and ignore any other
638
revisions that might be present. There is no direct replacement
639
method.
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.
640
129
"""
641
if 'evil' in debug.debug_flags:
642
mutter_callsite(2, "all_revision_ids is linear with history.")
643
130
return self._all_revision_ids()
644
131
132
@needs_read_lock
645
133
def _all_revision_ids(self):
646
134
"""Returns a list of all the revision ids in the repository.
647
135
648
136
These are in as much topological order as the underlying store can
649
present.
137
present: for weaves ghosts may lead to a lack of correctness until
138
the reweave updates the parents list.
650
139
"""
651
raise NotImplementedError(self._all_revision_ids)
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)
652
144
653
145
def break_lock(self):
654
146
"""Break a lock if one is present from another instance.
665
157
Returns a set of the present revisions.
666
158
"""
667
159
result = []
668
graph = self.get_graph()
669
parent_map = graph.get_parent_map(revision_ids)
670
# The old API returned a list, should this actually be a set?
671
return parent_map.keys()
160
for id in revision_ids:
161
if self.has_revision(id):
162
result.append(id)
163
return result
672
164
673
165
@staticmethod
674
166
def create(a_bzrdir):
675
167
"""Construct the current default format repository in a_bzrdir."""
676
168
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
677
169
678
def __init__(self, _format, a_bzrdir, control_files):
170
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
679
171
"""instantiate a Repository.
680
172
681
173
:param _format: The format of the repository on disk.
690
182
# the following are part of the public API for Repository:
691
183
self.bzrdir = a_bzrdir
692
184
self.control_files = control_files
693
self._transport = control_files._transport
694
self.base = self._transport.base
695
# for tests
696
self._reconcile_does_inventory_gc = True
697
self._reconcile_fixes_text_parents = False
698
self._reconcile_backsup_inventory = True
185
self._revision_store = _revision_store
186
self.text_store = text_store
187
# backwards compatibility
188
self.weave_store = text_store
699
189
# not right yet - should be more semantically clear ?
700
190
#
191
self.control_store = control_store
192
self.control_weaves = control_store
701
193
# TODO: make sure to construct the right store classes, etc, depending
702
194
# on whether escaping is required.
703
self._warn_if_deprecated()
704
self._write_group = None
705
# Additional places to query for data.
706
self._fallback_repositories = []
707
# What order should fetch operations request streams in?
708
# The default is unordered as that is the cheapest for an origin to
709
# provide.
710
self._fetch_order = 'unordered'
711
# Does this repository use deltas that can be fetched as-deltas ?
712
# (E.g. knits, where the knit deltas can be transplanted intact.
713
# We default to False, which will ensure that enough data to get
714
# a full text out of any fetch stream will be grabbed.
715
self._fetch_uses_deltas = False
716
# Should fetch trigger a reconcile after the fetch? Only needed for
717
# some repository formats that can suffer internal inconsistencies.
718
self._fetch_reconcile = False
719
195
720
196
def __repr__(self):
721
return '%s(%r)' % (self.__class__.__name__,
722
self.base)
723
724
def has_same_location(self, other):
725
"""Returns a boolean indicating if this repository is at the same
726
location as another repository.
727
728
This might return False even when two repository objects are accessing
729
the same physical repository via different URLs.
730
"""
731
if self.__class__ is not other.__class__:
732
return False
733
return (self._transport.base == other._transport.base)
734
735
def is_in_write_group(self):
736
"""Return True if there is an open write group.
737
738
:seealso: start_write_group.
739
"""
740
return self._write_group is not None
197
return '%s(%r)' % (self.__class__.__name__,
198
self.bzrdir.transport.base)
741
199
742
200
def is_locked(self):
743
201
return self.control_files.is_locked()
744
202
745
def is_write_locked(self):
746
"""Return True if this object is write locked."""
747
return self.is_locked() and self.control_files._lock_mode == 'w'
748
749
def lock_write(self, token=None):
750
"""Lock this repository for writing.
751
752
This causes caching within the repository obejct to start accumlating
753
data during reads, and allows a 'write_group' to be obtained. Write
754
groups must be used for actual data insertion.
755
756
:param token: if this is already locked, then lock_write will fail
757
unless the token matches the existing lock.
758
:returns: a token if this instance supports tokens, otherwise None.
759
:raises TokenLockingNotSupported: when a token is given but this
760
instance doesn't support using token locks.
761
:raises MismatchedToken: if the specified token doesn't match the token
762
of the existing lock.
763
:seealso: start_write_group.
764
765
A token should be passed in if you know that you have locked the object
766
some other way, and need to synchronise this object's state with that
767
fact.
768
769
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
770
"""
771
result = self.control_files.lock_write(token=token)
772
for repo in self._fallback_repositories:
773
# Writes don't affect fallback repos
774
repo.lock_read()
775
self._refresh_data()
776
return result
203
def lock_write(self):
204
self.control_files.lock_write()
777
205
778
206
def lock_read(self):
779
207
self.control_files.lock_read()
780
for repo in self._fallback_repositories:
781
repo.lock_read()
782
self._refresh_data()
783
208
784
209
def get_physical_lock_status(self):
785
210
return self.control_files.get_physical_lock_status()
786
211
787
def leave_lock_in_place(self):
788
"""Tell this repository not to release the physical lock when this
789
object is unlocked.
790
791
If lock_write doesn't return a token, then this method is not supported.
792
"""
793
self.control_files.leave_in_place()
794
795
def dont_leave_lock_in_place(self):
796
"""Tell this repository to release the physical lock when this
797
object is unlocked, even if it didn't originally acquire it.
798
799
If lock_write doesn't return a token, then this method is not supported.
800
"""
801
self.control_files.dont_leave_in_place()
802
803
@needs_read_lock
804
def gather_stats(self, revid=None, committers=None):
805
"""Gather statistics from a revision id.
806
807
:param revid: The revision id to gather statistics from, if None, then
808
no revision specific statistics are gathered.
809
:param committers: Optional parameter controlling whether to grab
810
a count of committers from the revision specific statistics.
811
:return: A dictionary of statistics. Currently this contains:
812
committers: The number of committers if requested.
813
firstrev: A tuple with timestamp, timezone for the penultimate left
814
most ancestor of revid, if revid is not the NULL_REVISION.
815
latestrev: A tuple with timestamp, timezone for revid, if revid is
816
not the NULL_REVISION.
817
revisions: The total revision count in the repository.
818
size: An estimate disk size of the repository in bytes.
819
"""
820
result = {}
821
if revid and committers:
822
result['committers'] = 0
823
if revid and revid != _mod_revision.NULL_REVISION:
824
if committers:
825
all_committers = set()
826
revisions = self.get_ancestry(revid)
827
# pop the leading None
828
revisions.pop(0)
829
first_revision = None
830
if not committers:
831
# ignore the revisions in the middle - just grab first and last
832
revisions = revisions[0], revisions[-1]
833
for revision in self.get_revisions(revisions):
834
if not first_revision:
835
first_revision = revision
836
if committers:
837
all_committers.add(revision.committer)
838
last_revision = revision
839
if committers:
840
result['committers'] = len(all_committers)
841
result['firstrev'] = (first_revision.timestamp,
842
first_revision.timezone)
843
result['latestrev'] = (last_revision.timestamp,
844
last_revision.timezone)
845
846
# now gather global repository information
847
# XXX: This is available for many repos regardless of listability.
848
if self.bzrdir.root_transport.listable():
849
# XXX: do we want to __define len__() ?
850
# Maybe the versionedfiles object should provide a different
851
# method to get the number of keys.
852
result['revisions'] = len(self.revisions.keys())
853
# result['size'] = t
854
return result
855
856
def find_branches(self, using=False):
857
"""Find branches underneath this repository.
858
859
This will include branches inside other branches.
860
861
:param using: If True, list only branches using this repository.
862
"""
863
if using and not self.is_shared():
864
try:
865
return [self.bzrdir.open_branch()]
866
except errors.NotBranchError:
867
return []
868
class Evaluator(object):
869
870
def __init__(self):
871
self.first_call = True
872
873
def __call__(self, bzrdir):
874
# On the first call, the parameter is always the bzrdir
875
# containing the current repo.
876
if not self.first_call:
877
try:
878
repository = bzrdir.open_repository()
879
except errors.NoRepositoryPresent:
880
pass
881
else:
882
return False, (None, repository)
883
self.first_call = False
884
try:
885
value = (bzrdir.open_branch(), None)
886
except errors.NotBranchError:
887
value = (None, None)
888
return True, value
889
890
branches = []
891
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
892
self.bzrdir.root_transport, evaluate=Evaluator()):
893
if branch is not None:
894
branches.append(branch)
895
if not using and repository is not None:
896
branches.extend(repository.find_branches())
897
return branches
898
899
@needs_read_lock
900
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
901
"""Return the revision ids that other has that this does not.
902
903
These are returned in topological order.
904
905
revision_id: only return revision ids included by revision_id.
906
"""
907
return InterRepository.get(other, self).search_missing_revision_ids(
908
revision_id, find_ghosts)
909
910
@deprecated_method(one_two)
911
@needs_read_lock
912
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
913
"""Return the revision ids that other has that this does not.
914
915
These are returned in topological order.
916
917
revision_id: only return revision ids included by revision_id.
918
"""
919
keys = self.search_missing_revision_ids(
920
other, revision_id, find_ghosts).get_keys()
921
other.lock_read()
922
try:
923
parents = other.get_graph().get_parent_map(keys)
924
finally:
925
other.unlock()
926
return tsort.topo_sort(parents)
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)
927
221
928
222
@staticmethod
929
223
def open(base):
932
226
For instance, if the repository is at URL/.bzr/repository,
933
227
Repository.open(URL) -> a Repository instance.
934
228
"""
935
control = bzrdir.BzrDir.open(base)
229
control = bzrlib.bzrdir.BzrDir.open(base)
936
230
return control.open_repository()
937
231
938
def copy_content_into(self, destination, revision_id=None):
232
def copy_content_into(self, destination, revision_id=None, basis=None):
939
233
"""Make a complete copy of the content in self into destination.
940
234
941
235
This is a destructive operation! Do not use it on existing
942
236
repositories.
943
237
"""
944
return InterRepository.get(self, destination).copy_content(revision_id)
945
946
def commit_write_group(self):
947
"""Commit the contents accrued within the current write group.
948
949
:seealso: start_write_group.
950
"""
951
if self._write_group is not self.get_transaction():
952
# has an unlock or relock occured ?
953
raise errors.BzrError('mismatched lock context %r and '
954
'write group %r.' %
955
(self.get_transaction(), self._write_group))
956
self._commit_write_group()
957
self._write_group = None
958
959
def _commit_write_group(self):
960
"""Template method for per-repository write group cleanup.
961
962
This is called before the write group is considered to be
963
finished and should ensure that all data handed to the repository
964
for writing during the write group is safely committed (to the
965
extent possible considering file system caching etc).
966
"""
967
968
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
238
return InterRepository.get(self, destination).copy_content(revision_id, basis)
239
240
def fetch(self, source, revision_id=None, pb=None):
969
241
"""Fetch the content required to construct revision_id from source.
970
242
971
243
If revision_id is None all content is copied.
972
:param find_ghosts: Find and copy revisions in the source that are
973
ghosts in the target (and not reachable directly by walking out to
974
the first-present revision in target from revision_id).
975
244
"""
976
# fast path same-url fetch operations
977
if self.has_same_location(source):
978
# check that last_revision is in 'from' and then return a
979
# no-operation.
980
if (revision_id is not None and
981
not _mod_revision.is_null(revision_id)):
982
self.get_revision(revision_id)
983
return 0, []
984
# if there is no specific appropriate InterRepository, this will get
985
# the InterRepository base class, which raises an
986
# IncompatibleRepositories when asked to fetch.
987
inter = InterRepository.get(source, self)
988
return inter.fetch(revision_id=revision_id, pb=pb,
989
find_ghosts=find_ghosts)
990
991
def create_bundle(self, target, base, fileobj, format=None):
992
return serializer.write_bundle(self, target, base, fileobj, format)
993
994
def get_commit_builder(self, branch, parents, config, timestamp=None,
995
timezone=None, committer=None, revprops=None,
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,
996
250
revision_id=None):
997
251
"""Obtain a CommitBuilder for this repository.
998
252
1005
259
:param revprops: Optional dictionary of revision properties.
1006
260
:param revision_id: Optional revision id.
1007
261
"""
1008
result = self._commit_builder_class(self, parents, config,
1009
timestamp, timezone, committer, revprops, revision_id)
1010
self.start_write_group()
1011
return result
262
return CommitBuilder(self, parents, config, timestamp, timezone,
263
committer, revprops, revision_id)
1012
264
1013
265
def unlock(self):
1014
if (self.control_files._lock_count == 1 and
1015
self.control_files._lock_mode == 'w'):
1016
if self._write_group is not None:
1017
self.abort_write_group()
1018
self.control_files.unlock()
1019
raise errors.BzrError(
1020
'Must end write groups before releasing write locks.')
1021
266
self.control_files.unlock()
1022
for repo in self._fallback_repositories:
1023
repo.unlock()
1024
267
1025
268
@needs_read_lock
1026
def clone(self, a_bzrdir, revision_id=None):
269
def clone(self, a_bzrdir, revision_id=None, basis=None):
1027
270
"""Clone this repository into a_bzrdir using the current format.
1028
271
1029
272
Currently no check is made that the format of this repository and
1030
273
the bzrdir format are compatible. FIXME RBC 20060201.
1031
1032
:return: The newly created destination repository.
1033
"""
1034
# TODO: deprecate after 0.16; cloning this with all its settings is
1035
# probably not very useful -- mbp 20070423
1036
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1037
self.copy_content_into(dest_repo, revision_id)
1038
return dest_repo
1039
1040
def start_write_group(self):
1041
"""Start a write group in the repository.
1042
1043
Write groups are used by repositories which do not have a 1:1 mapping
1044
between file ids and backend store to manage the insertion of data from
1045
both fetch and commit operations.
1046
1047
A write lock is required around the start_write_group/commit_write_group
1048
for the support of lock-requiring repository formats.
1049
1050
One can only insert data into a repository inside a write group.
1051
1052
:return: None.
1053
"""
1054
if not self.is_write_locked():
1055
raise errors.NotWriteLocked(self)
1056
if self._write_group:
1057
raise errors.BzrError('already in a write group')
1058
self._start_write_group()
1059
# so we can detect unlock/relock - the write group is now entered.
1060
self._write_group = self.get_transaction()
1061
1062
def _start_write_group(self):
1063
"""Template method for per-repository write group startup.
1064
1065
This is called before the write group is considered to be
1066
entered.
1067
"""
1068
1069
@needs_read_lock
1070
def sprout(self, to_bzrdir, revision_id=None):
1071
"""Create a descendent repository for new development.
1072
1073
Unlike clone, this does not copy the settings of the repository.
1074
"""
1075
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1076
dest_repo.fetch(self, revision_id=revision_id)
1077
return dest_repo
1078
1079
def _create_sprouting_repo(self, a_bzrdir, shared):
274
"""
1080
275
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1081
276
# use target default format.
1082
dest_repo = a_bzrdir.create_repository()
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()
1083
284
else:
1084
# Most control formats need the repository to be specifically
1085
# created, but on some old all-in-one formats it's not needed
1086
try:
1087
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1088
except errors.UninitializableFormat:
1089
dest_repo = a_bzrdir.open_repository()
1090
return dest_repo
285
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
286
self.copy_content_into(result, revision_id, basis)
287
return result
1091
288
1092
289
@needs_read_lock
1093
290
def has_revision(self, revision_id):
1094
291
"""True if this repository has a copy of the revision."""
1095
return revision_id in self.has_revisions((revision_id,))
1096
1097
@needs_read_lock
1098
def has_revisions(self, revision_ids):
1099
"""Probe to find out the presence of multiple revisions.
1100
1101
:param revision_ids: An iterable of revision_ids.
1102
:return: A set of the revision_ids that were present.
1103
"""
1104
parent_map = self.revisions.get_parent_map(
1105
[(rev_id,) for rev_id in revision_ids])
1106
result = set()
1107
if _mod_revision.NULL_REVISION in revision_ids:
1108
result.add(_mod_revision.NULL_REVISION)
1109
result.update([key[0] for key in parent_map])
1110
return result
1111
1112
@needs_read_lock
1113
def get_revision(self, revision_id):
1114
"""Return the Revision object for a named revision."""
1115
return self.get_revisions([revision_id])[0]
292
return self._revision_store.has_revision_id(revision_id,
293
self.get_transaction())
1116
294
1117
295
@needs_read_lock
1118
296
def get_revision_reconcile(self, revision_id):
1123
301
be used by reconcile, or reconcile-alike commands that are correcting
1124
302
or testing the revision graph.
1125
303
"""
1126
return self._get_revisions([revision_id])[0]
1127
1128
@needs_read_lock
1129
def get_revisions(self, revision_ids):
1130
"""Get many revisions at once."""
1131
return self._get_revisions(revision_ids)
1132
1133
@needs_read_lock
1134
def _get_revisions(self, revision_ids):
1135
"""Core work logic to get many revisions without sanity checks."""
1136
for rev_id in revision_ids:
1137
if not rev_id or not isinstance(rev_id, basestring):
1138
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1139
keys = [(key,) for key in revision_ids]
1140
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1141
revs = {}
1142
for record in stream:
1143
if record.storage_kind == 'absent':
1144
raise errors.NoSuchRevision(self, record.key[0])
1145
text = record.get_bytes_as('fulltext')
1146
rev = self._serializer.read_revision_from_string(text)
1147
revs[record.key[0]] = rev
1148
return [revs[revid] for revid in revision_ids]
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_revision(revision_id,
307
self.get_transaction())
1149
308
1150
309
@needs_read_lock
1151
310
def get_revision_xml(self, revision_id):
1152
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1153
# would have already do it.
1154
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1155
rev = self.get_revision(revision_id)
1156
rev_tmp = cStringIO.StringIO()
311
rev = self.get_revision(revision_id)
312
rev_tmp = StringIO()
1157
313
# the current serializer..
1158
self._serializer.write_revision(rev, rev_tmp)
314
self._revision_store._serializer.write_revision(rev, rev_tmp)
1159
315
rev_tmp.seek(0)
1160
316
return rev_tmp.getvalue()
1161
317
1162
def get_deltas_for_revisions(self, revisions):
1163
"""Produce a generator of revision deltas.
318
@needs_read_lock
319
def get_revision(self, revision_id):
320
"""Return the Revision object for a named revision"""
321
r = self.get_revision_reconcile(revision_id)
322
# weave corruption can lead to absent revision markers that should be
323
# present.
324
# the following test is reasonably cheap (it needs a single weave read)
325
# and the weave is cached in read transactions. In write transactions
326
# it is not cached but typically we only read a small number of
327
# revisions. For knits when they are introduced we will probably want
328
# to ensure that caching write transactions are in use.
329
inv = self.get_inventory_weave()
330
self._check_revision_parents(r, inv)
331
return r
332
333
def _check_revision_parents(self, revision, inventory):
334
"""Private to Repository and Fetch.
1164
335
1165
Note that the input is a sequence of REVISIONS, not revision_ids.
1166
Trees will be held in memory until the generator exits.
1167
Each delta is relative to the revision's lefthand predecessor.
1168
"""
1169
required_trees = set()
1170
for revision in revisions:
1171
required_trees.add(revision.revision_id)
1172
required_trees.update(revision.parent_ids[:1])
1173
trees = dict((t.get_revision_id(), t) for
1174
t in self.revision_trees(required_trees))
1175
for revision in revisions:
1176
if not revision.parent_ids:
1177
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1178
else:
1179
old_tree = trees[revision.parent_ids[0]]
1180
yield trees[revision.revision_id].changes_from(old_tree)
1181
1182
@needs_read_lock
1183
def get_revision_delta(self, revision_id):
1184
"""Return the delta for one revision.
1185
1186
The delta is relative to the left-hand predecessor of the
1187
revision.
1188
"""
1189
r = self.get_revision(revision_id)
1190
return list(self.get_deltas_for_revisions([r]))[0]
336
This checks the parentage of revision in an inventory weave for
337
consistency and is only applicable to inventory-weave-for-ancestry
338
using repository formats & fetchers.
339
"""
340
weave_parents = inventory.get_parents(revision.revision_id)
341
weave_names = inventory.versions()
342
for parent_id in revision.parent_ids:
343
if parent_id in weave_names:
344
# this parent must not be a ghost.
345
if not parent_id in weave_parents:
346
# but it is a ghost
347
raise errors.CorruptRepository(self)
1191
348
1192
349
@needs_write_lock
1193
350
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1194
351
signature = gpg_strategy.sign(plaintext)
1195
self.add_signature_text(revision_id, signature)
1196
1197
@needs_write_lock
1198
def add_signature_text(self, revision_id, signature):
1199
self.signatures.add_lines((revision_id,), (),
1200
osutils.split_lines(signature))
1201
1202
def find_text_key_references(self):
1203
"""Find the text key references within the repository.
1204
1205
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
352
self._revision_store.add_revision_signature_text(revision_id,
353
signature,
354
self.get_transaction())
355
356
def fileids_altered_by_revision_ids(self, revision_ids):
357
"""Find the file ids and versions affected by revisions.
358
359
:param revisions: an iterable containing revision ids.
360
:return: a dictionary mapping altered file-ids to an iterable of
1206
361
revision_ids. Each altered file-ids has the exact revision_ids that
1207
362
altered it listed explicitly.
1208
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1209
to whether they were referred to by the inventory of the
1210
revision_id that they contain. The inventory texts from all present
1211
revision ids are assessed to generate this report.
1212
"""
1213
revision_keys = self.revisions.keys()
1214
w = self.inventories
1215
pb = ui.ui_factory.nested_progress_bar()
1216
try:
1217
return self._find_text_key_references_from_xml_inventory_lines(
1218
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1219
finally:
1220
pb.finished()
1221
1222
def _find_text_key_references_from_xml_inventory_lines(self,
1223
line_iterator):
1224
"""Core routine for extracting references to texts from inventories.
1225
1226
This performs the translation of xml lines to revision ids.
1227
1228
:param line_iterator: An iterator of lines, origin_version_id
1229
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1230
to whether they were referred to by the inventory of the
1231
revision_id that they contain. Note that if that revision_id was
1232
not part of the line_iterator's output then False will be given -
1233
even though it may actually refer to that key.
1234
"""
1235
if not self._serializer.support_altered_by_hack:
1236
raise AssertionError(
1237
"_find_text_key_references_from_xml_inventory_lines only "
1238
"supported for branches which store inventory as unnested xml"
1239
", not on %r" % self)
363
"""
364
assert isinstance(self._format, (RepositoryFormat5,
365
RepositoryFormat6,
366
RepositoryFormat7,
367
RepositoryFormatKnit1)), \
368
("fileids_altered_by_revision_ids only supported for branches "
369
"which store inventory as unnested xml, not on %r" % self)
370
selected_revision_ids = set(revision_ids)
371
w = self.get_inventory_weave()
1240
372
result = {}
1241
373
1242
374
# this code needs to read every new line in every inventory for the
1247
379
# revisions. We don't need to see all lines in the inventory because
1248
380
# only those added in an inventory in rev X can contain a revision=X
1249
381
# line.
1250
unescape_revid_cache = {}
1251
unescape_fileid_cache = {}
1252
1253
# jam 20061218 In a big fetch, this handles hundreds of thousands
1254
# of lines, so it has had a lot of inlining and optimizing done.
1255
# Sorry that it is a little bit messy.
1256
# Move several functions to be local variables, since this is a long
1257
# running loop.
1258
search = self._file_ids_altered_regex.search
1259
unescape = _unescape_xml
1260
setdefault = result.setdefault
1261
for line, line_key in line_iterator:
1262
match = search(line)
1263
if match is None:
1264
continue
1265
# One call to match.group() returning multiple items is quite a
1266
# bit faster than 2 calls to match.group() each returning 1
1267
file_id, revision_id = match.group('file_id', 'revision_id')
1268
1269
# Inlining the cache lookups helps a lot when you make 170,000
1270
# lines and 350k ids, versus 8.4 unique ids.
1271
# Using a cache helps in 2 ways:
1272
# 1) Avoids unnecessary decoding calls
1273
# 2) Re-uses cached strings, which helps in future set and
1274
# equality checks.
1275
# (2) is enough that removing encoding entirely along with
1276
# the cache (so we are using plain strings) results in no
1277
# performance improvement.
1278
try:
1279
revision_id = unescape_revid_cache[revision_id]
1280
except KeyError:
1281
unescaped = unescape(revision_id)
1282
unescape_revid_cache[revision_id] = unescaped
1283
revision_id = unescaped
1284
1285
# Note that unconditionally unescaping means that we deserialise
1286
# every fileid, which for general 'pull' is not great, but we don't
1287
# really want to have some many fulltexts that this matters anyway.
1288
# RBC 20071114.
1289
try:
1290
file_id = unescape_fileid_cache[file_id]
1291
except KeyError:
1292
unescaped = unescape(file_id)
1293
unescape_fileid_cache[file_id] = unescaped
1294
file_id = unescaped
1295
1296
key = (file_id, revision_id)
1297
setdefault(key, False)
1298
if revision_id == line_key[-1]:
1299
result[key] = True
1300
return result
1301
1302
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1303
revision_ids):
1304
"""Helper routine for fileids_altered_by_revision_ids.
1305
1306
This performs the translation of xml lines to revision ids.
1307
1308
:param line_iterator: An iterator of lines, origin_version_id
1309
:param revision_ids: The revision ids to filter for. This should be a
1310
set or other type which supports efficient __contains__ lookups, as
1311
the revision id from each parsed line will be looked up in the
1312
revision_ids filter.
1313
:return: a dictionary mapping altered file-ids to an iterable of
1314
revision_ids. Each altered file-ids has the exact revision_ids that
1315
altered it listed explicitly.
1316
"""
1317
result = {}
1318
setdefault = result.setdefault
1319
for key in \
1320
self._find_text_key_references_from_xml_inventory_lines(
1321
line_iterator).iterkeys():
1322
# once data is all ensured-consistent; then this is
1323
# if revision_id == version_id
1324
if key[-1:] in revision_ids:
1325
setdefault(key[0], set()).add(key[-1])
1326
return result
1327
1328
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1329
"""Find the file ids and versions affected by revisions.
1330
1331
:param revisions: an iterable containing revision ids.
1332
:param _inv_weave: The inventory weave from this repository or None.
1333
If None, the inventory weave will be opened automatically.
1334
:return: a dictionary mapping altered file-ids to an iterable of
1335
revision_ids. Each altered file-ids has the exact revision_ids that
1336
altered it listed explicitly.
1337
"""
1338
selected_keys = set((revid,) for revid in revision_ids)
1339
w = _inv_weave or self.inventories
1340
pb = ui.ui_factory.nested_progress_bar()
1341
try:
1342
return self._find_file_ids_from_xml_inventory_lines(
1343
w.iter_lines_added_or_present_in_keys(
1344
selected_keys, pb=pb),
1345
selected_keys)
1346
finally:
1347
pb.finished()
1348
1349
def iter_files_bytes(self, desired_files):
1350
"""Iterate through file versions.
1351
1352
Files will not necessarily be returned in the order they occur in
1353
desired_files. No specific order is guaranteed.
1354
1355
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1356
value supplied by the caller as part of desired_files. It should
1357
uniquely identify the file version in the caller's context. (Examples:
1358
an index number or a TreeTransform trans_id.)
1359
1360
bytes_iterator is an iterable of bytestrings for the file. The
1361
kind of iterable and length of the bytestrings are unspecified, but for
1362
this implementation, it is a list of bytes produced by
1363
VersionedFile.get_record_stream().
1364
1365
:param desired_files: a list of (file_id, revision_id, identifier)
1366
triples
1367
"""
1368
transaction = self.get_transaction()
1369
text_keys = {}
1370
for file_id, revision_id, callable_data in desired_files:
1371
text_keys[(file_id, revision_id)] = callable_data
1372
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1373
if record.storage_kind == 'absent':
1374
raise errors.RevisionNotPresent(record.key, self)
1375
yield text_keys[record.key], record.get_bytes_as('fulltext')
1376
1377
def _generate_text_key_index(self, text_key_references=None,
1378
ancestors=None):
1379
"""Generate a new text key index for the repository.
1380
1381
This is an expensive function that will take considerable time to run.
1382
1383
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1384
list of parents, also text keys. When a given key has no parents,
1385
the parents list will be [NULL_REVISION].
1386
"""
1387
# All revisions, to find inventory parents.
1388
if ancestors is None:
1389
graph = self.get_graph()
1390
ancestors = graph.get_parent_map(self.all_revision_ids())
1391
if text_key_references is None:
1392
text_key_references = self.find_text_key_references()
1393
pb = ui.ui_factory.nested_progress_bar()
1394
try:
1395
return self._do_generate_text_key_index(ancestors,
1396
text_key_references, pb)
1397
finally:
1398
pb.finished()
1399
1400
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1401
"""Helper for _generate_text_key_index to avoid deep nesting."""
1402
revision_order = tsort.topo_sort(ancestors)
1403
invalid_keys = set()
1404
revision_keys = {}
1405
for revision_id in revision_order:
1406
revision_keys[revision_id] = set()
1407
text_count = len(text_key_references)
1408
# a cache of the text keys to allow reuse; costs a dict of all the
1409
# keys, but saves a 2-tuple for every child of a given key.
1410
text_key_cache = {}
1411
for text_key, valid in text_key_references.iteritems():
1412
if not valid:
1413
invalid_keys.add(text_key)
1414
else:
1415
revision_keys[text_key[1]].add(text_key)
1416
text_key_cache[text_key] = text_key
1417
del text_key_references
1418
text_index = {}
1419
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1420
NULL_REVISION = _mod_revision.NULL_REVISION
1421
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1422
# too small for large or very branchy trees. However, for 55K path
1423
# trees, it would be easy to use too much memory trivially. Ideally we
1424
# could gauge this by looking at available real memory etc, but this is
1425
# always a tricky proposition.
1426
inventory_cache = lru_cache.LRUCache(10)
1427
batch_size = 10 # should be ~150MB on a 55K path tree
1428
batch_count = len(revision_order) / batch_size + 1
1429
processed_texts = 0
1430
pb.update("Calculating text parents.", processed_texts, text_count)
1431
for offset in xrange(batch_count):
1432
to_query = revision_order[offset * batch_size:(offset + 1) *
1433
batch_size]
1434
if not to_query:
1435
break
1436
for rev_tree in self.revision_trees(to_query):
1437
revision_id = rev_tree.get_revision_id()
1438
parent_ids = ancestors[revision_id]
1439
for text_key in revision_keys[revision_id]:
1440
pb.update("Calculating text parents.", processed_texts)
1441
processed_texts += 1
1442
candidate_parents = []
1443
for parent_id in parent_ids:
1444
parent_text_key = (text_key[0], parent_id)
1445
try:
1446
check_parent = parent_text_key not in \
1447
revision_keys[parent_id]
1448
except KeyError:
1449
# the parent parent_id is a ghost:
1450
check_parent = False
1451
# truncate the derived graph against this ghost.
1452
parent_text_key = None
1453
if check_parent:
1454
# look at the parent commit details inventories to
1455
# determine possible candidates in the per file graph.
1456
# TODO: cache here.
1457
try:
1458
inv = inventory_cache[parent_id]
1459
except KeyError:
1460
inv = self.revision_tree(parent_id).inventory
1461
inventory_cache[parent_id] = inv
1462
parent_entry = inv._byid.get(text_key[0], None)
1463
if parent_entry is not None:
1464
parent_text_key = (
1465
text_key[0], parent_entry.revision)
1466
else:
1467
parent_text_key = None
1468
if parent_text_key is not None:
1469
candidate_parents.append(
1470
text_key_cache[parent_text_key])
1471
parent_heads = text_graph.heads(candidate_parents)
1472
new_parents = list(parent_heads)
1473
new_parents.sort(key=lambda x:candidate_parents.index(x))
1474
if new_parents == []:
1475
new_parents = [NULL_REVISION]
1476
text_index[text_key] = new_parents
1477
1478
for text_key in invalid_keys:
1479
text_index[text_key] = [NULL_REVISION]
1480
return text_index
1481
1482
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1483
"""Get an iterable listing the keys of all the data introduced by a set
1484
of revision IDs.
1485
1486
The keys will be ordered so that the corresponding items can be safely
1487
fetched and inserted in that order.
1488
1489
:returns: An iterable producing tuples of (knit-kind, file-id,
1490
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1491
'revisions'. file-id is None unless knit-kind is 'file'.
1492
"""
1493
# XXX: it's a bit weird to control the inventory weave caching in this
1494
# generator. Ideally the caching would be done in fetch.py I think. Or
1495
# maybe this generator should explicitly have the contract that it
1496
# should not be iterated until the previously yielded item has been
1497
# processed?
1498
inv_w = self.inventories
1499
1500
# file ids that changed
1501
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1502
count = 0
1503
num_file_ids = len(file_ids)
1504
for file_id, altered_versions in file_ids.iteritems():
1505
if _files_pb is not None:
1506
_files_pb.update("fetch texts", count, num_file_ids)
1507
count += 1
1508
yield ("file", file_id, altered_versions)
1509
# We're done with the files_pb. Note that it finished by the caller,
1510
# just as it was created by the caller.
1511
del _files_pb
1512
1513
# inventory
1514
yield ("inventory", None, revision_ids)
1515
1516
# signatures
1517
revisions_with_signatures = set()
1518
for rev_id in revision_ids:
1519
try:
1520
self.get_signature_text(rev_id)
1521
except errors.NoSuchRevision:
1522
# not signed.
1523
pass
1524
else:
1525
revisions_with_signatures.add(rev_id)
1526
yield ("signatures", None, revisions_with_signatures)
1527
1528
# revisions
1529
yield ("revisions", None, revision_ids)
382
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
383
start = line.find('file_id="')+9
384
if start < 9: continue
385
end = line.find('"', start)
386
assert end>= 0
387
file_id = _unescape_xml(line[start:end])
388
389
start = line.find('revision="')+10
390
if start < 10: continue
391
end = line.find('"', start)
392
assert end>= 0
393
revision_id = _unescape_xml(line[start:end])
394
if revision_id in selected_revision_ids:
395
result.setdefault(file_id, set()).add(revision_id)
396
return result
397
398
@needs_read_lock
399
def get_inventory_weave(self):
400
return self.control_weaves.get_weave('inventory',
401
self.get_transaction())
1530
402
1531
403
@needs_read_lock
1532
404
def get_inventory(self, revision_id):
1533
"""Get Inventory object by revision id."""
1534
return self.iter_inventories([revision_id]).next()
1535
1536
def iter_inventories(self, revision_ids):
1537
"""Get many inventories by revision_ids.
1538
1539
This will buffer some or all of the texts used in constructing the
1540
inventories in memory, but will only parse a single inventory at a
1541
time.
1542
1543
:return: An iterator of inventories.
1544
"""
1545
if ((None in revision_ids)
1546
or (_mod_revision.NULL_REVISION in revision_ids)):
1547
raise ValueError('cannot get null revision inventory')
1548
return self._iter_inventories(revision_ids)
1549
1550
def _iter_inventories(self, revision_ids):
1551
"""single-document based inventory iteration."""
1552
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1553
yield self.deserialise_inventory(revision_id, text)
1554
1555
def _iter_inventory_xmls(self, revision_ids):
1556
keys = [(revision_id,) for revision_id in revision_ids]
1557
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1558
texts = {}
1559
for record in stream:
1560
if record.storage_kind != 'absent':
1561
texts[record.key] = record.get_bytes_as('fulltext')
1562
else:
1563
raise errors.NoSuchRevision(self, record.key)
1564
for key in keys:
1565
yield texts[key], key[-1]
405
"""Get Inventory object by hash."""
406
return self.deserialise_inventory(
407
revision_id, self.get_inventory_xml(revision_id))
1566
408
1567
409
def deserialise_inventory(self, revision_id, xml):
1568
410
"""Transform the xml into an inventory object.
1570
412
:param revision_id: The expected revision id of the inventory.
1571
413
:param xml: A serialised inventory.
1572
414
"""
1573
result = self._serializer.read_inventory_from_string(xml, revision_id)
1574
if result.revision_id != revision_id:
1575
raise AssertionError('revision id mismatch %s != %s' % (
1576
result.revision_id, revision_id))
1577
return result
1578
1579
def serialise_inventory(self, inv):
1580
return self._serializer.write_inventory_to_string(inv)
1581
1582
def _serialise_inventory_to_lines(self, inv):
1583
return self._serializer.write_inventory_to_lines(inv)
1584
1585
def get_serializer_format(self):
1586
return self._serializer.format_num
415
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
1587
416
1588
417
@needs_read_lock
1589
418
def get_inventory_xml(self, revision_id):
1590
419
"""Get inventory XML as a file object."""
1591
texts = self._iter_inventory_xmls([revision_id])
1592
420
try:
1593
text, revision_id = texts.next()
1594
except StopIteration:
1595
raise errors.HistoryMissing(self, 'inventory', revision_id)
1596
return text
421
assert isinstance(revision_id, basestring), type(revision_id)
422
iw = self.get_inventory_weave()
423
return iw.get_text(revision_id)
424
except IndexError:
425
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
1597
426
1598
427
@needs_read_lock
1599
428
def get_inventory_sha1(self, revision_id):
1601
430
"""
1602
431
return self.get_revision(revision_id).inventory_sha1
1603
432
1604
def iter_reverse_revision_history(self, revision_id):
1605
"""Iterate backwards through revision ids in the lefthand history
1606
1607
:param revision_id: The revision id to start with. All its lefthand
1608
ancestors will be traversed.
1609
"""
1610
graph = self.get_graph()
1611
next_id = revision_id
1612
while True:
1613
if next_id in (None, _mod_revision.NULL_REVISION):
1614
return
1615
yield next_id
1616
# Note: The following line may raise KeyError in the event of
1617
# truncated history. We decided not to have a try:except:raise
1618
# RevisionNotPresent here until we see a use for it, because of the
1619
# cost in an inner loop that is by its very nature O(history).
1620
# Robert Collins 20080326
1621
parents = graph.get_parent_map([next_id])[next_id]
1622
if len(parents) == 0:
1623
return
1624
else:
1625
next_id = parents[0]
433
@needs_read_lock
434
def get_revision_graph(self, revision_id=None):
435
"""Return a dictionary containing the revision graph.
436
437
:return: a dictionary of revision_id->revision_parents_list.
438
"""
439
weave = self.get_inventory_weave()
440
all_revisions = self._eliminate_revisions_not_present(weave.versions())
441
entire_graph = dict([(node, weave.get_parents(node)) for
442
node in all_revisions])
443
if revision_id is None:
444
return entire_graph
445
elif revision_id not in entire_graph:
446
raise errors.NoSuchRevision(self, revision_id)
447
else:
448
# add what can be reached from revision_id
449
result = {}
450
pending = set([revision_id])
451
while len(pending) > 0:
452
node = pending.pop()
453
result[node] = entire_graph[node]
454
for revision_id in result[node]:
455
if revision_id not in result:
456
pending.add(revision_id)
457
return result
458
459
@needs_read_lock
460
def get_revision_graph_with_ghosts(self, revision_ids=None):
461
"""Return a graph of the revisions with ghosts marked as applicable.
462
463
:param revision_ids: an iterable of revisions to graph or None for all.
464
:return: a Graph object with the graph reachable from revision_ids.
465
"""
466
result = Graph()
467
if not revision_ids:
468
pending = set(self.all_revision_ids())
469
required = set([])
470
else:
471
pending = set(revision_ids)
472
required = set(revision_ids)
473
done = set([])
474
while len(pending):
475
revision_id = pending.pop()
476
try:
477
rev = self.get_revision(revision_id)
478
except errors.NoSuchRevision:
479
if revision_id in required:
480
raise
481
# a ghost
482
result.add_ghost(revision_id)
483
continue
484
for parent_id in rev.parent_ids:
485
# is this queued or done ?
486
if (parent_id not in pending and
487
parent_id not in done):
488
# no, queue it.
489
pending.add(parent_id)
490
result.add_node(revision_id, rev.parent_ids)
491
done.add(revision_id)
492
return result
1626
493
1627
494
@needs_read_lock
1628
495
def get_revision_inventory(self, revision_id):
1640
507
else:
1641
508
return self.get_inventory(revision_id)
1642
509
510
@needs_read_lock
1643
511
def is_shared(self):
1644
512
"""Return True if this repository is flagged as a shared repository."""
1645
513
raise NotImplementedError(self.is_shared)
1651
519
reconciler = RepoReconciler(self, thorough=thorough)
1652
520
reconciler.reconcile()
1653
521
return reconciler
1654
1655
def _refresh_data(self):
1656
"""Helper called from lock_* to ensure coherency with disk.
1657
1658
The default implementation does nothing; it is however possible
1659
for repositories to maintain loaded indices across multiple locks
1660
by checking inside their implementation of this method to see
1661
whether their indices are still valid. This depends of course on
1662
the disk format being validatable in this manner.
1663
"""
1664
522
1665
523
@needs_read_lock
1666
524
def revision_tree(self, revision_id):
1667
525
"""Return Tree for a revision on this branch.
1668
526
1669
`revision_id` may be NULL_REVISION for the empty tree revision.
1670
"""
1671
revision_id = _mod_revision.ensure_null(revision_id)
527
`revision_id` may be None for the null revision, in which case
528
an `EmptyTree` is returned."""
1672
529
# TODO: refactor this to use an existing revision object
1673
530
# so we don't need to read it in twice.
1674
if revision_id == _mod_revision.NULL_REVISION:
1675
return RevisionTree(self, Inventory(root_id=None),
1676
_mod_revision.NULL_REVISION)
531
if revision_id is None or revision_id == NULL_REVISION:
532
return EmptyTree()
1677
533
else:
1678
534
inv = self.get_revision_inventory(revision_id)
1679
535
return RevisionTree(self, inv, revision_id)
1680
536
1681
def revision_trees(self, revision_ids):
1682
"""Return Tree for a revision on this branch.
1683
1684
`revision_id` may not be None or 'null:'"""
1685
inventories = self.iter_inventories(revision_ids)
1686
for inv in inventories:
1687
yield RevisionTree(self, inv, inv.revision_id)
1688
1689
537
@needs_read_lock
1690
def get_ancestry(self, revision_id, topo_sorted=True):
538
def get_ancestry(self, revision_id):
1691
539
"""Return a list of revision-ids integrated by a revision.
1692
540
1693
541
The first element of the list is always None, indicating the origin
1696
544
1697
545
This is topologically sorted.
1698
546
"""
1699
if _mod_revision.is_null(revision_id):
547
if revision_id is None:
1700
548
return [None]
1701
549
if not self.has_revision(revision_id):
1702
550
raise errors.NoSuchRevision(self, revision_id)
1703
graph = self.get_graph()
1704
keys = set()
1705
search = graph._make_breadth_first_searcher([revision_id])
1706
while True:
1707
try:
1708
found, ghosts = search.next_with_ghosts()
1709
except StopIteration:
1710
break
1711
keys.update(found)
1712
if _mod_revision.NULL_REVISION in keys:
1713
keys.remove(_mod_revision.NULL_REVISION)
1714
if topo_sorted:
1715
parent_map = graph.get_parent_map(keys)
1716
keys = tsort.topo_sort(parent_map)
1717
return [None] + list(keys)
1718
1719
def pack(self):
1720
"""Compress the data within the repository.
1721
1722
This operation only makes sense for some repository types. For other
1723
types it should be a no-op that just returns.
1724
1725
This stub method does not require a lock, but subclasses should use
1726
@needs_write_lock as this is a long running call its reasonable to
1727
implicitly lock for the user.
1728
"""
551
w = self.get_inventory_weave()
552
candidates = w.get_ancestry(revision_id)
553
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1729
554
1730
555
@needs_read_lock
1731
@deprecated_method(one_six)
1732
556
def print_file(self, file, revision_id):
1733
557
"""Print `file` to stdout.
1734
558
1749
573
def get_transaction(self):
1750
574
return self.control_files.get_transaction()
1751
575
1752
@deprecated_method(one_one)
1753
def get_parents(self, revision_ids):
1754
"""See StackedParentsProvider.get_parents"""
1755
parent_map = self.get_parent_map(revision_ids)
1756
return [parent_map.get(r, None) for r in revision_ids]
1757
1758
def get_parent_map(self, revision_ids):
1759
"""See graph._StackedParentsProvider.get_parent_map"""
1760
# revisions index works in keys; this just works in revisions
1761
# therefore wrap and unwrap
1762
query_keys = []
1763
result = {}
1764
for revision_id in revision_ids:
1765
if revision_id == _mod_revision.NULL_REVISION:
1766
result[revision_id] = ()
1767
elif revision_id is None:
1768
raise ValueError('get_parent_map(None) is not valid')
1769
else:
1770
query_keys.append((revision_id ,))
1771
for ((revision_id,), parent_keys) in \
1772
self.revisions.get_parent_map(query_keys).iteritems():
1773
if parent_keys:
1774
result[revision_id] = tuple(parent_revid
1775
for (parent_revid,) in parent_keys)
1776
else:
1777
result[revision_id] = (_mod_revision.NULL_REVISION,)
1778
return result
1779
1780
def _make_parents_provider(self):
1781
return self
1782
1783
def get_graph(self, other_repository=None):
1784
"""Return the graph walker for this repository format"""
1785
parents_provider = self._make_parents_provider()
1786
if (other_repository is not None and
1787
not self.has_same_location(other_repository)):
1788
parents_provider = graph._StackedParentsProvider(
1789
[parents_provider, other_repository._make_parents_provider()])
1790
return graph.Graph(parents_provider)
1791
1792
def _get_versioned_file_checker(self):
1793
"""Return an object suitable for checking versioned files."""
1794
return _VersionedFileChecker(self)
1795
1796
def revision_ids_to_search_result(self, result_set):
1797
"""Convert a set of revision ids to a graph SearchResult."""
1798
result_parents = set()
1799
for parents in self.get_graph().get_parent_map(
1800
result_set).itervalues():
1801
result_parents.update(parents)
1802
included_keys = result_set.intersection(result_parents)
1803
start_keys = result_set.difference(included_keys)
1804
exclude_keys = result_parents.difference(result_set)
1805
result = graph.SearchResult(start_keys, exclude_keys,
1806
len(result_set), result_set)
1807
return result
576
def revision_parents(self, revid):
577
return self.get_inventory_weave().parent_names(revid)
1808
578
1809
579
@needs_write_lock
1810
580
def set_make_working_trees(self, new_value):
1830
600
@needs_read_lock
1831
601
def has_signature_for_revision_id(self, revision_id):
1832
602
"""Query for a revision signature for revision_id in the repository."""
1833
if not self.has_revision(revision_id):
1834
raise errors.NoSuchRevision(self, revision_id)
1835
sig_present = (1 == len(
1836
self.signatures.get_parent_map([(revision_id,)])))
1837
return sig_present
603
return self._revision_store.has_signature(revision_id,
604
self.get_transaction())
1838
605
1839
606
@needs_read_lock
1840
607
def get_signature_text(self, revision_id):
1841
608
"""Return the text for a signature."""
1842
stream = self.signatures.get_record_stream([(revision_id,)],
1843
'unordered', True)
1844
record = stream.next()
1845
if record.storage_kind == 'absent':
1846
raise errors.NoSuchRevision(self, revision_id)
1847
return record.get_bytes_as('fulltext')
609
return self._revision_store.get_signature_text(revision_id,
610
self.get_transaction())
1848
611
1849
612
@needs_read_lock
1850
def check(self, revision_ids=None):
613
def check(self, revision_ids):
1851
614
"""Check consistency of all history of given revision_ids.
1852
615
1853
616
Different repository implementations should override _check().
1855
618
:param revision_ids: A non-empty list of revision_ids whose ancestry
1856
619
will be checked. Typically the last revision_id of a branch.
1857
620
"""
621
if not revision_ids:
622
raise ValueError("revision_ids must be non-empty in %s.check"
623
% (self,))
1858
624
return self._check(revision_ids)
1859
625
1860
626
def _check(self, revision_ids):
1861
result = check.Check(self)
627
result = bzrlib.check.Check(self)
1862
628
result.check()
1863
629
return result
1864
630
1865
def _warn_if_deprecated(self):
1866
global _deprecation_warning_done
1867
if _deprecation_warning_done:
1868
return
1869
_deprecation_warning_done = True
1870
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1871
% (self._format, self.bzrdir.transport.base))
1872
1873
def supports_rich_root(self):
1874
return self._format.rich_root_data
1875
1876
def _check_ascii_revisionid(self, revision_id, method):
1877
"""Private helper for ascii-only repositories."""
1878
# weave repositories refuse to store revisionids that are non-ascii.
1879
if revision_id is not None:
1880
# weaves require ascii revision ids.
1881
if isinstance(revision_id, unicode):
1882
try:
1883
revision_id.encode('ascii')
1884
except UnicodeEncodeError:
1885
raise errors.NonAsciiRevisionId(method, self)
1886
else:
1887
try:
1888
revision_id.decode('ascii')
1889
except UnicodeDecodeError:
1890
raise errors.NonAsciiRevisionId(method, self)
631
632
class AllInOneRepository(Repository):
633
"""Legacy support - the repository behaviour for all-in-one branches."""
634
635
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
636
# we reuse one control files instance.
637
dir_mode = a_bzrdir._control_files._dir_mode
638
file_mode = a_bzrdir._control_files._file_mode
639
640
def get_store(name, compressed=True, prefixed=False):
641
# FIXME: This approach of assuming stores are all entirely compressed
642
# or entirely uncompressed is tidy, but breaks upgrade from
643
# some existing branches where there's a mixture; we probably
644
# still want the option to look for both.
645
relpath = a_bzrdir._control_files._escape(name)
646
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
647
prefixed=prefixed, compressed=compressed,
648
dir_mode=dir_mode,
649
file_mode=file_mode)
650
#if self._transport.should_cache():
651
# cache_path = os.path.join(self.cache_root, name)
652
# os.mkdir(cache_path)
653
# store = bzrlib.store.CachedStore(store, cache_path)
654
return store
655
656
# not broken out yet because the controlweaves|inventory_store
657
# and text_store | weave_store bits are still different.
658
if isinstance(_format, RepositoryFormat4):
659
# cannot remove these - there is still no consistent api
660
# which allows access to this old info.
661
self.inventory_store = get_store('inventory-store')
662
text_store = get_store('text-store')
663
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
664
665
@needs_read_lock
666
def is_shared(self):
667
"""AllInOne repositories cannot be shared."""
668
return False
669
670
@needs_write_lock
671
def set_make_working_trees(self, new_value):
672
"""Set the policy flag for making working trees when creating branches.
673
674
This only applies to branches that use this repository.
675
676
The default is 'True'.
677
:param new_value: True to restore the default, False to disable making
678
working trees.
679
"""
680
raise NotImplementedError(self.set_make_working_trees)
1891
681
1892
def revision_graph_can_have_wrong_parents(self):
1893
"""Is it possible for this repository to have a revision graph with
1894
incorrect parents?
1895
1896
If True, then this repository must also implement
1897
_find_inconsistent_revision_parents so that check and reconcile can
1898
check for inconsistencies before proceeding with other checks that may
1899
depend on the revision index being consistent.
1900
"""
1901
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1902
1903
1904
# remove these delegates a while after bzr 0.15
1905
def __make_delegated(name, from_module):
1906
def _deprecated_repository_forwarder():
1907
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1908
% (name, from_module),
1909
DeprecationWarning,
1910
stacklevel=2)
1911
m = __import__(from_module, globals(), locals(), [name])
1912
try:
1913
return getattr(m, name)
1914
except AttributeError:
1915
raise AttributeError('module %s has no name %s'
1916
% (m, name))
1917
globals()[name] = _deprecated_repository_forwarder
1918
1919
for _name in [
1920
'AllInOneRepository',
1921
'WeaveMetaDirRepository',
1922
'PreSplitOutRepositoryFormat',
1923
'RepositoryFormat4',
1924
'RepositoryFormat5',
1925
'RepositoryFormat6',
1926
'RepositoryFormat7',
1927
]:
1928
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1929
1930
for _name in [
1931
'KnitRepository',
1932
'RepositoryFormatKnit',
1933
'RepositoryFormatKnit1',
1934
]:
1935
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
682
def make_working_trees(self):
683
"""Returns the policy for making working trees on new branches."""
684
return True
1936
685
1937
686
1938
687
def install_revision(repository, rev, revision_tree):
1939
688
"""Install all revision data into a repository."""
1940
install_revisions(repository, [(rev, revision_tree, None)])
1941
1942
1943
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1944
"""Install all revision data into a repository.
1945
1946
Accepts an iterable of revision, tree, signature tuples. The signature
1947
may be None.
1948
"""
1949
repository.start_write_group()
1950
try:
1951
for n, (revision, revision_tree, signature) in enumerate(iterable):
1952
_install_revision(repository, revision, revision_tree, signature)
1953
if pb is not None:
1954
pb.update('Transferring revisions', n + 1, num_revisions)
1955
except:
1956
repository.abort_write_group()
1957
raise
1958
else:
1959
repository.commit_write_group()
1960
1961
1962
def _install_revision(repository, rev, revision_tree, signature):
1963
"""Install all revision data into a repository."""
1964
689
present_parents = []
1965
690
parent_trees = {}
1966
691
for p_id in rev.parent_ids:
1968
693
present_parents.append(p_id)
1969
694
parent_trees[p_id] = repository.revision_tree(p_id)
1970
695
else:
1971
parent_trees[p_id] = repository.revision_tree(
1972
_mod_revision.NULL_REVISION)
696
parent_trees[p_id] = EmptyTree()
1973
697
1974
698
inv = revision_tree.inventory
1975
entries = inv.iter_entries()
1976
# backwards compatibility hack: skip the root id.
1977
if not repository.supports_rich_root():
1978
path, root = entries.next()
1979
if root.revision != rev.revision_id:
1980
raise errors.IncompatibleRevision(repr(repository))
1981
text_keys = {}
1982
for path, ie in entries:
1983
text_keys[(ie.file_id, ie.revision)] = ie
1984
text_parent_map = repository.texts.get_parent_map(text_keys)
1985
missing_texts = set(text_keys) - set(text_parent_map)
699
1986
700
# Add the texts that are not already present
1987
for text_key in missing_texts:
1988
ie = text_keys[text_key]
1989
text_parents = []
1990
# FIXME: TODO: The following loop overlaps/duplicates that done by
1991
# commit to determine parents. There is a latent/real bug here where
1992
# the parents inserted are not those commit would do - in particular
1993
# they are not filtered by heads(). RBC, AB
1994
for revision, tree in parent_trees.iteritems():
1995
if ie.file_id not in tree:
1996
continue
1997
parent_id = tree.inventory[ie.file_id].revision
1998
if parent_id in text_parents:
1999
continue
2000
text_parents.append((ie.file_id, parent_id))
2001
lines = revision_tree.get_file(ie.file_id).readlines()
2002
repository.texts.add_lines(text_key, text_parents, lines)
701
for path, ie in inv.iter_entries():
702
w = repository.weave_store.get_weave_or_empty(ie.file_id,
703
repository.get_transaction())
704
if ie.revision not in w:
705
text_parents = []
706
# FIXME: TODO: The following loop *may* be overlapping/duplicate
707
# with InventoryEntry.find_previous_heads(). if it is, then there
708
# is a latent bug here where the parents may have ancestors of each
709
# other. RBC, AB
710
for revision, tree in parent_trees.iteritems():
711
if ie.file_id not in tree:
712
continue
713
parent_id = tree.inventory[ie.file_id].revision
714
if parent_id in text_parents:
715
continue
716
text_parents.append(parent_id)
717
718
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
719
repository.get_transaction())
720
lines = revision_tree.get_file(ie.file_id).readlines()
721
vfile.add_lines(rev.revision_id, text_parents, lines)
2003
722
try:
2004
723
# install the inventory
2005
724
repository.add_inventory(rev.revision_id, inv, present_parents)
2006
725
except errors.RevisionAlreadyPresent:
2007
726
pass
2008
if signature is not None:
2009
repository.add_signature_text(rev.revision_id, signature)
2010
727
repository.add_revision(rev.revision_id, rev, inv)
2011
728
2012
729
2013
730
class MetaDirRepository(Repository):
2014
"""Repositories in the new meta-dir layout.
2015
2016
:ivar _transport: Transport for access to repository control files,
2017
typically pointing to .bzr/repository.
2018
"""
2019
2020
def __init__(self, _format, a_bzrdir, control_files):
2021
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2022
self._transport = control_files._transport
2023
731
"""Repositories in the new meta-dir layout."""
732
733
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
734
super(MetaDirRepository, self).__init__(_format,
735
a_bzrdir,
736
control_files,
737
_revision_store,
738
control_store,
739
text_store)
740
741
dir_mode = self.control_files._dir_mode
742
file_mode = self.control_files._file_mode
743
744
@needs_read_lock
2024
745
def is_shared(self):
2025
746
"""Return True if this repository is flagged as a shared repository."""
2026
return self._transport.has('shared-storage')
747
return self.control_files._transport.has('shared-storage')
2027
748
2028
749
@needs_write_lock
2029
750
def set_make_working_trees(self, new_value):
2037
758
"""
2038
759
if new_value:
2039
760
try:
2040
self._transport.delete('no-working-trees')
761
self.control_files._transport.delete('no-working-trees')
2041
762
except errors.NoSuchFile:
2042
763
pass
2043
764
else:
2044
self._transport.put_bytes('no-working-trees', '',
2045
mode=self.bzrdir._get_file_mode())
765
self.control_files.put_utf8('no-working-trees', '')
2046
766
2047
767
def make_working_trees(self):
2048
768
"""Returns the policy for making working trees on new branches."""
2049
return not self._transport.has('no-working-trees')
2050
2051
2052
class MetaDirVersionedFileRepository(MetaDirRepository):
2053
"""Repositories in a meta-dir, that work via versioned file objects."""
2054
2055
def __init__(self, _format, a_bzrdir, control_files):
2056
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2057
control_files)
2058
2059
2060
class RepositoryFormatRegistry(registry.Registry):
2061
"""Registry of RepositoryFormats."""
2062
2063
def get(self, format_string):
2064
r = registry.Registry.get(self, format_string)
2065
if callable(r):
2066
r = r()
2067
return r
769
return not self.control_files._transport.has('no-working-trees')
770
771
772
class KnitRepository(MetaDirRepository):
773
"""Knit format repository."""
774
775
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
776
inv_vf.add_lines_with_ghosts(revid, parents, lines)
777
778
@needs_read_lock
779
def _all_revision_ids(self):
780
"""See Repository.all_revision_ids()."""
781
# Knits get the revision graph from the index of the revision knit, so
782
# it's always possible even if they're on an unlistable transport.
783
return self._revision_store.all_revision_ids(self.get_transaction())
784
785
def fileid_involved_between_revs(self, from_revid, to_revid):
786
"""Find file_id(s) which are involved in the changes between revisions.
787
788
This determines the set of revisions which are involved, and then
789
finds all file ids affected by those revisions.
790
"""
791
vf = self._get_revision_vf()
792
from_set = set(vf.get_ancestry(from_revid))
793
to_set = set(vf.get_ancestry(to_revid))
794
changed = to_set.difference(from_set)
795
return self._fileid_involved_by_set(changed)
796
797
def fileid_involved(self, last_revid=None):
798
"""Find all file_ids modified in the ancestry of last_revid.
799
800
:param last_revid: If None, last_revision() will be used.
801
"""
802
if not last_revid:
803
changed = set(self.all_revision_ids())
804
else:
805
changed = set(self.get_ancestry(last_revid))
806
if None in changed:
807
changed.remove(None)
808
return self._fileid_involved_by_set(changed)
809
810
@needs_read_lock
811
def get_ancestry(self, revision_id):
812
"""Return a list of revision-ids integrated by a revision.
813
814
This is topologically sorted.
815
"""
816
if revision_id is None:
817
return [None]
818
vf = self._get_revision_vf()
819
try:
820
return [None] + vf.get_ancestry(revision_id)
821
except errors.RevisionNotPresent:
822
raise errors.NoSuchRevision(self, revision_id)
823
824
@needs_read_lock
825
def get_revision(self, revision_id):
826
"""Return the Revision object for a named revision"""
827
return self.get_revision_reconcile(revision_id)
828
829
@needs_read_lock
830
def get_revision_graph(self, revision_id=None):
831
"""Return a dictionary containing the revision graph.
832
833
:return: a dictionary of revision_id->revision_parents_list.
834
"""
835
weave = self._get_revision_vf()
836
entire_graph = weave.get_graph()
837
if revision_id is None:
838
return weave.get_graph()
839
elif revision_id not in weave:
840
raise errors.NoSuchRevision(self, revision_id)
841
else:
842
# add what can be reached from revision_id
843
result = {}
844
pending = set([revision_id])
845
while len(pending) > 0:
846
node = pending.pop()
847
result[node] = weave.get_parents(node)
848
for revision_id in result[node]:
849
if revision_id not in result:
850
pending.add(revision_id)
851
return result
852
853
@needs_read_lock
854
def get_revision_graph_with_ghosts(self, revision_ids=None):
855
"""Return a graph of the revisions with ghosts marked as applicable.
856
857
:param revision_ids: an iterable of revisions to graph or None for all.
858
:return: a Graph object with the graph reachable from revision_ids.
859
"""
860
result = Graph()
861
vf = self._get_revision_vf()
862
versions = set(vf.versions())
863
if not revision_ids:
864
pending = set(self.all_revision_ids())
865
required = set([])
866
else:
867
pending = set(revision_ids)
868
required = set(revision_ids)
869
done = set([])
870
while len(pending):
871
revision_id = pending.pop()
872
if not revision_id in versions:
873
if revision_id in required:
874
raise errors.NoSuchRevision(self, revision_id)
875
# a ghost
876
result.add_ghost(revision_id)
877
# mark it as done so we don't try for it again.
878
done.add(revision_id)
879
continue
880
parent_ids = vf.get_parents_with_ghosts(revision_id)
881
for parent_id in parent_ids:
882
# is this queued or done ?
883
if (parent_id not in pending and
884
parent_id not in done):
885
# no, queue it.
886
pending.add(parent_id)
887
result.add_node(revision_id, parent_ids)
888
done.add(revision_id)
889
return result
890
891
def _get_revision_vf(self):
892
""":return: a versioned file containing the revisions."""
893
vf = self._revision_store.get_revision_file(self.get_transaction())
894
return vf
895
896
@needs_write_lock
897
def reconcile(self, other=None, thorough=False):
898
"""Reconcile this repository."""
899
from bzrlib.reconcile import KnitReconciler
900
reconciler = KnitReconciler(self, thorough=thorough)
901
reconciler.reconcile()
902
return reconciler
2068
903
2069
2070
format_registry = RepositoryFormatRegistry()
2071
"""Registry of formats, indexed by their identifying format string.
2072
2073
This can contain either format instances themselves, or classes/factories that
2074
can be called to obtain one.
2075
"""
2076
2077
2078
#####################################################################
2079
# Repository Formats
904
def revision_parents(self, revid):
905
return self._get_revision_vf().get_parents(rev_id)
2080
906
2081
907
class RepositoryFormat(object):
2082
908
"""A repository format.
2087
913
children.
2088
914
* an open routine which returns a Repository instance.
2089
915
2090
There is one and only one Format subclass for each on-disk format. But
2091
there can be one Repository subclass that is used for several different
2092
formats. The _format attribute on a Repository instance can be used to
2093
determine the disk format.
2094
2095
916
Formats are placed in an dict by their format string for reference
2096
917
during opening. These should be subclasses of RepositoryFormat
2097
918
for consistency.
2104
925
_matchingbzrdir - the bzrdir format that the repository format was
2105
926
originally written to work with. This can be used if manually
2106
927
constructing a bzrdir and repository, or more commonly for test suite
2107
parameterization.
928
parameterisation.
2108
929
"""
2109
930
2110
# Set to True or False in derived classes. True indicates that the format
2111
# supports ghosts gracefully.
2112
supports_ghosts = None
2113
# Can this repository be given external locations to lookup additional
2114
# data. Set to True or False in derived classes.
2115
supports_external_lookups = None
2116
2117
def __str__(self):
2118
return "<%s>" % self.__class__.__name__
2119
2120
def __eq__(self, other):
2121
# format objects are generally stateless
2122
return isinstance(other, self.__class__)
2123
2124
def __ne__(self, other):
2125
return not self == other
931
_default_format = None
932
"""The default format used for new repositories."""
933
934
_formats = {}
935
"""The known formats."""
2126
936
2127
937
@classmethod
2128
938
def find_format(klass, a_bzrdir):
2129
"""Return the format for the repository object in a_bzrdir.
2130
2131
This is used by bzr native formats that have a "format" file in
2132
the repository. Other methods may be used by different types of
2133
control directory.
2134
"""
939
"""Return the format for the repository object in a_bzrdir."""
2135
940
try:
2136
941
transport = a_bzrdir.get_repository_transport(None)
2137
942
format_string = transport.get("format").read()
2138
return format_registry.get(format_string)
943
return klass._formats[format_string]
2139
944
except errors.NoSuchFile:
2140
945
raise errors.NoRepositoryPresent(a_bzrdir)
2141
946
except KeyError:
2142
raise errors.UnknownFormatError(format=format_string,
2143
kind='repository')
2144
2145
@classmethod
2146
def register_format(klass, format):
2147
format_registry.register(format.get_format_string(), format)
2148
2149
@classmethod
2150
def unregister_format(klass, format):
2151
format_registry.remove(format.get_format_string())
947
raise errors.UnknownFormatError(format_string)
948
949
def _get_control_store(self, repo_transport, control_files):
950
"""Return the control store for this repository."""
951
raise NotImplementedError(self._get_control_store)
2152
952
2153
953
@classmethod
2154
954
def get_default_format(klass):
2155
955
"""Return the current default format."""
2156
from bzrlib import bzrdir
2157
return bzrdir.format_registry.make_bzrdir('default').repository_format
956
return klass._default_format
2158
957
2159
958
def get_format_string(self):
2160
959
"""Return the ASCII format string that identifies this format.
2168
967
"""Return the short description for this format."""
2169
968
raise NotImplementedError(self.get_format_description)
2170
969
2171
# TODO: this shouldn't be in the base class, it's specific to things that
2172
# use weaves or knits -- mbp 20070207
970
def _get_revision_store(self, repo_transport, control_files):
971
"""Return the revision store object for this a_bzrdir."""
972
raise NotImplementedError(self._get_revision_store)
973
974
def _get_text_rev_store(self,
975
transport,
976
control_files,
977
name,
978
compressed=True,
979
prefixed=False,
980
serializer=None):
981
"""Common logic for getting a revision store for a repository.
982
983
see self._get_revision_store for the subclass-overridable method to
984
get the store for a repository.
985
"""
986
from bzrlib.store.revision.text import TextRevisionStore
987
dir_mode = control_files._dir_mode
988
file_mode = control_files._file_mode
989
text_store =TextStore(transport.clone(name),
990
prefixed=prefixed,
991
compressed=compressed,
992
dir_mode=dir_mode,
993
file_mode=file_mode)
994
_revision_store = TextRevisionStore(text_store, serializer)
995
return _revision_store
996
2173
997
def _get_versioned_file_store(self,
2174
998
name,
2175
999
transport,
2176
1000
control_files,
2177
1001
prefixed=True,
2178
versionedfile_class=None,
2179
versionedfile_kwargs={},
1002
versionedfile_class=WeaveFile,
2180
1003
escaped=False):
2181
if versionedfile_class is None:
2182
versionedfile_class = self._versionedfile_class
2183
1004
weave_transport = control_files._transport.clone(name)
2184
1005
dir_mode = control_files._dir_mode
2185
1006
file_mode = control_files._file_mode
2187
1008
dir_mode=dir_mode,
2188
1009
file_mode=file_mode,
2189
1010
versionedfile_class=versionedfile_class,
2190
versionedfile_kwargs=versionedfile_kwargs,
2191
1011
escaped=escaped)
2192
1012
2193
1013
def initialize(self, a_bzrdir, shared=False):
2195
1015
2196
1016
:param a_bzrdir: The bzrdir to put the new repository in it.
2197
1017
:param shared: The repository should be initialized as a sharable one.
2198
:returns: The new repository object.
2199
1018
2200
1019
This may raise UninitializableFormat if shared repository are not
2201
1020
compatible the a_bzrdir.
2202
1021
"""
2203
raise NotImplementedError(self.initialize)
2204
1022
2205
1023
def is_supported(self):
2206
1024
"""Is this format supported?
2211
1029
"""
2212
1030
return True
2213
1031
2214
def check_conversion_target(self, target_format):
2215
raise NotImplementedError(self.check_conversion_target)
2216
2217
1032
def open(self, a_bzrdir, _found=False):
2218
1033
"""Return an instance of this format for the bzrdir a_bzrdir.
2219
1034
2221
1036
"""
2222
1037
raise NotImplementedError(self.open)
2223
1038
1039
@classmethod
1040
def register_format(klass, format):
1041
klass._formats[format.get_format_string()] = format
1042
1043
@classmethod
1044
def set_default_format(klass, format):
1045
klass._default_format = format
1046
1047
@classmethod
1048
def unregister_format(klass, format):
1049
assert klass._formats[format.get_format_string()] is format
1050
del klass._formats[format.get_format_string()]
1051
1052
1053
class PreSplitOutRepositoryFormat(RepositoryFormat):
1054
"""Base class for the pre split out repository formats."""
1055
1056
def initialize(self, a_bzrdir, shared=False, _internal=False):
1057
"""Create a weave repository.
1058
1059
TODO: when creating split out bzr branch formats, move this to a common
1060
base for Format5, Format6. or something like that.
1061
"""
1062
from bzrlib.weavefile import write_weave_v5
1063
from bzrlib.weave import Weave
1064
1065
if shared:
1066
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1067
1068
if not _internal:
1069
# always initialized when the bzrdir is.
1070
return self.open(a_bzrdir, _found=True)
1071
1072
# Create an empty weave
1073
sio = StringIO()
1074
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1075
empty_weave = sio.getvalue()
1076
1077
mutter('creating repository in %s.', a_bzrdir.transport.base)
1078
dirs = ['revision-store', 'weaves']
1079
files = [('inventory.weave', StringIO(empty_weave)),
1080
]
1081
1082
# FIXME: RBC 20060125 don't peek under the covers
1083
# NB: no need to escape relative paths that are url safe.
1084
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1085
TransportLock)
1086
control_files.create_lock()
1087
control_files.lock_write()
1088
control_files._transport.mkdir_multi(dirs,
1089
mode=control_files._dir_mode)
1090
try:
1091
for file, content in files:
1092
control_files.put(file, content)
1093
finally:
1094
control_files.unlock()
1095
return self.open(a_bzrdir, _found=True)
1096
1097
def _get_control_store(self, repo_transport, control_files):
1098
"""Return the control store for this repository."""
1099
return self._get_versioned_file_store('',
1100
repo_transport,
1101
control_files,
1102
prefixed=False)
1103
1104
def _get_text_store(self, transport, control_files):
1105
"""Get a store for file texts for this format."""
1106
raise NotImplementedError(self._get_text_store)
1107
1108
def open(self, a_bzrdir, _found=False):
1109
"""See RepositoryFormat.open()."""
1110
if not _found:
1111
# we are being called directly and must probe.
1112
raise NotImplementedError
1113
1114
repo_transport = a_bzrdir.get_repository_transport(None)
1115
control_files = a_bzrdir._control_files
1116
text_store = self._get_text_store(repo_transport, control_files)
1117
control_store = self._get_control_store(repo_transport, control_files)
1118
_revision_store = self._get_revision_store(repo_transport, control_files)
1119
return AllInOneRepository(_format=self,
1120
a_bzrdir=a_bzrdir,
1121
_revision_store=_revision_store,
1122
control_store=control_store,
1123
text_store=text_store)
1124
1125
1126
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1127
"""Bzr repository format 4.
1128
1129
This repository format has:
1130
- flat stores
1131
- TextStores for texts, inventories,revisions.
1132
1133
This format is deprecated: it indexes texts using a text id which is
1134
removed in format 5; initialization and write support for this format
1135
has been removed.
1136
"""
1137
1138
def __init__(self):
1139
super(RepositoryFormat4, self).__init__()
1140
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1141
1142
def get_format_description(self):
1143
"""See RepositoryFormat.get_format_description()."""
1144
return "Repository format 4"
1145
1146
def initialize(self, url, shared=False, _internal=False):
1147
"""Format 4 branches cannot be created."""
1148
raise errors.UninitializableFormat(self)
1149
1150
def is_supported(self):
1151
"""Format 4 is not supported.
1152
1153
It is not supported because the model changed from 4 to 5 and the
1154
conversion logic is expensive - so doing it on the fly was not
1155
feasible.
1156
"""
1157
return False
1158
1159
def _get_control_store(self, repo_transport, control_files):
1160
"""Format 4 repositories have no formal control store at this point.
1161
1162
This will cause any control-file-needing apis to fail - this is desired.
1163
"""
1164
return None
1165
1166
def _get_revision_store(self, repo_transport, control_files):
1167
"""See RepositoryFormat._get_revision_store()."""
1168
from bzrlib.xml4 import serializer_v4
1169
return self._get_text_rev_store(repo_transport,
1170
control_files,
1171
'revision-store',
1172
serializer=serializer_v4)
1173
1174
def _get_text_store(self, transport, control_files):
1175
"""See RepositoryFormat._get_text_store()."""
1176
1177
1178
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1179
"""Bzr control format 5.
1180
1181
This repository format has:
1182
- weaves for file texts and inventory
1183
- flat stores
1184
- TextStores for revisions and signatures.
1185
"""
1186
1187
def __init__(self):
1188
super(RepositoryFormat5, self).__init__()
1189
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1190
1191
def get_format_description(self):
1192
"""See RepositoryFormat.get_format_description()."""
1193
return "Weave repository format 5"
1194
1195
def _get_revision_store(self, repo_transport, control_files):
1196
"""See RepositoryFormat._get_revision_store()."""
1197
"""Return the revision store object for this a_bzrdir."""
1198
return self._get_text_rev_store(repo_transport,
1199
control_files,
1200
'revision-store',
1201
compressed=False)
1202
1203
def _get_text_store(self, transport, control_files):
1204
"""See RepositoryFormat._get_text_store()."""
1205
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1206
1207
1208
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1209
"""Bzr control format 6.
1210
1211
This repository format has:
1212
- weaves for file texts and inventory
1213
- hash subdirectory based stores.
1214
- TextStores for revisions and signatures.
1215
"""
1216
1217
def __init__(self):
1218
super(RepositoryFormat6, self).__init__()
1219
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1220
1221
def get_format_description(self):
1222
"""See RepositoryFormat.get_format_description()."""
1223
return "Weave repository format 6"
1224
1225
def _get_revision_store(self, repo_transport, control_files):
1226
"""See RepositoryFormat._get_revision_store()."""
1227
return self._get_text_rev_store(repo_transport,
1228
control_files,
1229
'revision-store',
1230
compressed=False,
1231
prefixed=True)
1232
1233
def _get_text_store(self, transport, control_files):
1234
"""See RepositoryFormat._get_text_store()."""
1235
return self._get_versioned_file_store('weaves', transport, control_files)
1236
2224
1237
2225
1238
class MetaDirRepositoryFormat(RepositoryFormat):
2226
1239
"""Common base class for the new repositories using the metadir layout."""
2227
1240
2228
rich_root_data = False
2229
supports_tree_reference = False
2230
supports_external_lookups = False
2231
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2232
2233
1241
def __init__(self):
2234
1242
super(MetaDirRepositoryFormat, self).__init__()
1243
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
2235
1244
2236
1245
def _create_control_files(self, a_bzrdir):
2237
1246
"""Create the required files and the initial control_files object."""
2238
1247
# FIXME: RBC 20060125 don't peek under the covers
2239
1248
# NB: no need to escape relative paths that are url safe.
2240
1249
repository_transport = a_bzrdir.get_repository_transport(self)
2241
control_files = lockable_files.LockableFiles(repository_transport,
2242
'lock', lockdir.LockDir)
1250
control_files = LockableFiles(repository_transport, 'lock', LockDir)
2243
1251
control_files.create_lock()
2244
1252
return control_files
2245
1253
2247
1255
"""Upload the initial blank content."""
2248
1256
control_files = self._create_control_files(a_bzrdir)
2249
1257
control_files.lock_write()
2250
transport = control_files._transport
2251
if shared == True:
2252
utf8_files += [('shared-storage', '')]
2253
1258
try:
2254
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2255
for (filename, content_stream) in files:
2256
transport.put_file(filename, content_stream,
2257
mode=a_bzrdir._get_file_mode())
2258
for (filename, content_bytes) in utf8_files:
2259
transport.put_bytes_non_atomic(filename, content_bytes,
2260
mode=a_bzrdir._get_file_mode())
1259
control_files._transport.mkdir_multi(dirs,
1260
mode=control_files._dir_mode)
1261
for file, content in files:
1262
control_files.put(file, content)
1263
for file, content in utf8_files:
1264
control_files.put_utf8(file, content)
1265
if shared == True:
1266
control_files.put_utf8('shared-storage', '')
2261
1267
finally:
2262
1268
control_files.unlock()
2263
1269
2264
1270
1271
class RepositoryFormat7(MetaDirRepositoryFormat):
1272
"""Bzr repository 7.
1273
1274
This repository format has:
1275
- weaves for file texts and inventory
1276
- hash subdirectory based stores.
1277
- TextStores for revisions and signatures.
1278
- a format marker of its own
1279
- an optional 'shared-storage' flag
1280
- an optional 'no-working-trees' flag
1281
"""
1282
1283
def _get_control_store(self, repo_transport, control_files):
1284
"""Return the control store for this repository."""
1285
return self._get_versioned_file_store('',
1286
repo_transport,
1287
control_files,
1288
prefixed=False)
1289
1290
def get_format_string(self):
1291
"""See RepositoryFormat.get_format_string()."""
1292
return "Bazaar-NG Repository format 7"
1293
1294
def get_format_description(self):
1295
"""See RepositoryFormat.get_format_description()."""
1296
return "Weave repository format 7"
1297
1298
def _get_revision_store(self, repo_transport, control_files):
1299
"""See RepositoryFormat._get_revision_store()."""
1300
return self._get_text_rev_store(repo_transport,
1301
control_files,
1302
'revision-store',
1303
compressed=False,
1304
prefixed=True,
1305
)
1306
1307
def _get_text_store(self, transport, control_files):
1308
"""See RepositoryFormat._get_text_store()."""
1309
return self._get_versioned_file_store('weaves',
1310
transport,
1311
control_files)
1312
1313
def initialize(self, a_bzrdir, shared=False):
1314
"""Create a weave repository.
1315
1316
:param shared: If true the repository will be initialized as a shared
1317
repository.
1318
"""
1319
from bzrlib.weavefile import write_weave_v5
1320
from bzrlib.weave import Weave
1321
1322
# Create an empty weave
1323
sio = StringIO()
1324
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1325
empty_weave = sio.getvalue()
1326
1327
mutter('creating repository in %s.', a_bzrdir.transport.base)
1328
dirs = ['revision-store', 'weaves']
1329
files = [('inventory.weave', StringIO(empty_weave)),
1330
]
1331
utf8_files = [('format', self.get_format_string())]
1332
1333
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1334
return self.open(a_bzrdir=a_bzrdir, _found=True)
1335
1336
def open(self, a_bzrdir, _found=False, _override_transport=None):
1337
"""See RepositoryFormat.open().
1338
1339
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1340
repository at a slightly different url
1341
than normal. I.e. during 'upgrade'.
1342
"""
1343
if not _found:
1344
format = RepositoryFormat.find_format(a_bzrdir)
1345
assert format.__class__ == self.__class__
1346
if _override_transport is not None:
1347
repo_transport = _override_transport
1348
else:
1349
repo_transport = a_bzrdir.get_repository_transport(None)
1350
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1351
text_store = self._get_text_store(repo_transport, control_files)
1352
control_store = self._get_control_store(repo_transport, control_files)
1353
_revision_store = self._get_revision_store(repo_transport, control_files)
1354
return MetaDirRepository(_format=self,
1355
a_bzrdir=a_bzrdir,
1356
control_files=control_files,
1357
_revision_store=_revision_store,
1358
control_store=control_store,
1359
text_store=text_store)
1360
1361
1362
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1363
"""Bzr repository knit format 1.
1364
1365
This repository format has:
1366
- knits for file texts and inventory
1367
- hash subdirectory based stores.
1368
- knits for revisions and signatures
1369
- TextStores for revisions and signatures.
1370
- a format marker of its own
1371
- an optional 'shared-storage' flag
1372
- an optional 'no-working-trees' flag
1373
- a LockDir lock
1374
1375
This format was introduced in bzr 0.8.
1376
"""
1377
1378
def _get_control_store(self, repo_transport, control_files):
1379
"""Return the control store for this repository."""
1380
return VersionedFileStore(
1381
repo_transport,
1382
prefixed=False,
1383
file_mode=control_files._file_mode,
1384
versionedfile_class=KnitVersionedFile,
1385
versionedfile_kwargs={'factory':KnitPlainFactory()},
1386
)
1387
1388
def get_format_string(self):
1389
"""See RepositoryFormat.get_format_string()."""
1390
return "Bazaar-NG Knit Repository Format 1"
1391
1392
def get_format_description(self):
1393
"""See RepositoryFormat.get_format_description()."""
1394
return "Knit repository format 1"
1395
1396
def _get_revision_store(self, repo_transport, control_files):
1397
"""See RepositoryFormat._get_revision_store()."""
1398
from bzrlib.store.revision.knit import KnitRevisionStore
1399
versioned_file_store = VersionedFileStore(
1400
repo_transport,
1401
file_mode=control_files._file_mode,
1402
prefixed=False,
1403
precious=True,
1404
versionedfile_class=KnitVersionedFile,
1405
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1406
escaped=True,
1407
)
1408
return KnitRevisionStore(versioned_file_store)
1409
1410
def _get_text_store(self, transport, control_files):
1411
"""See RepositoryFormat._get_text_store()."""
1412
return self._get_versioned_file_store('knits',
1413
transport,
1414
control_files,
1415
versionedfile_class=KnitVersionedFile,
1416
escaped=True)
1417
1418
def initialize(self, a_bzrdir, shared=False):
1419
"""Create a knit format 1 repository.
1420
1421
:param a_bzrdir: bzrdir to contain the new repository; must already
1422
be initialized.
1423
:param shared: If true the repository will be initialized as a shared
1424
repository.
1425
"""
1426
mutter('creating repository in %s.', a_bzrdir.transport.base)
1427
dirs = ['revision-store', 'knits']
1428
files = []
1429
utf8_files = [('format', self.get_format_string())]
1430
1431
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1432
repo_transport = a_bzrdir.get_repository_transport(None)
1433
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1434
control_store = self._get_control_store(repo_transport, control_files)
1435
transaction = bzrlib.transactions.WriteTransaction()
1436
# trigger a write of the inventory store.
1437
control_store.get_weave_or_empty('inventory', transaction)
1438
_revision_store = self._get_revision_store(repo_transport, control_files)
1439
_revision_store.has_revision_id('A', transaction)
1440
_revision_store.get_signature_file(transaction)
1441
return self.open(a_bzrdir=a_bzrdir, _found=True)
1442
1443
def open(self, a_bzrdir, _found=False, _override_transport=None):
1444
"""See RepositoryFormat.open().
1445
1446
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1447
repository at a slightly different url
1448
than normal. I.e. during 'upgrade'.
1449
"""
1450
if not _found:
1451
format = RepositoryFormat.find_format(a_bzrdir)
1452
assert format.__class__ == self.__class__
1453
if _override_transport is not None:
1454
repo_transport = _override_transport
1455
else:
1456
repo_transport = a_bzrdir.get_repository_transport(None)
1457
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1458
text_store = self._get_text_store(repo_transport, control_files)
1459
control_store = self._get_control_store(repo_transport, control_files)
1460
_revision_store = self._get_revision_store(repo_transport, control_files)
1461
return KnitRepository(_format=self,
1462
a_bzrdir=a_bzrdir,
1463
control_files=control_files,
1464
_revision_store=_revision_store,
1465
control_store=control_store,
1466
text_store=text_store)
1467
1468
2265
1469
# formats which have no format string are not discoverable
2266
# and not independently creatable, so are not registered. They're
2267
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2268
# needed, it's constructed directly by the BzrDir. Non-native formats where
2269
# the repository is not separately opened are similar.
2270
2271
format_registry.register_lazy(
2272
'Bazaar-NG Repository format 7',
2273
'bzrlib.repofmt.weaverepo',
2274
'RepositoryFormat7'
2275
)
2276
2277
format_registry.register_lazy(
2278
'Bazaar-NG Knit Repository Format 1',
2279
'bzrlib.repofmt.knitrepo',
2280
'RepositoryFormatKnit1',
2281
)
2282
2283
format_registry.register_lazy(
2284
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2285
'bzrlib.repofmt.knitrepo',
2286
'RepositoryFormatKnit3',
2287
)
2288
2289
format_registry.register_lazy(
2290
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2291
'bzrlib.repofmt.knitrepo',
2292
'RepositoryFormatKnit4',
2293
)
2294
2295
# Pack-based formats. There is one format for pre-subtrees, and one for
2296
# post-subtrees to allow ease of testing.
2297
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2298
format_registry.register_lazy(
2299
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2300
'bzrlib.repofmt.pack_repo',
2301
'RepositoryFormatKnitPack1',
2302
)
2303
format_registry.register_lazy(
2304
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2305
'bzrlib.repofmt.pack_repo',
2306
'RepositoryFormatKnitPack3',
2307
)
2308
format_registry.register_lazy(
2309
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2310
'bzrlib.repofmt.pack_repo',
2311
'RepositoryFormatKnitPack4',
2312
)
2313
format_registry.register_lazy(
2314
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2315
'bzrlib.repofmt.pack_repo',
2316
'RepositoryFormatKnitPack5',
2317
)
2318
format_registry.register_lazy(
2319
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2320
'bzrlib.repofmt.pack_repo',
2321
'RepositoryFormatKnitPack5RichRoot',
2322
)
2323
format_registry.register_lazy(
2324
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2325
'bzrlib.repofmt.pack_repo',
2326
'RepositoryFormatKnitPack5RichRootBroken',
2327
)
2328
2329
# Development formats.
2330
# 1.7->1.8 go below here
2331
format_registry.register_lazy(
2332
"Bazaar development format 2 (needs bzr.dev from before 1.8)\n",
2333
'bzrlib.repofmt.pack_repo',
2334
'RepositoryFormatPackDevelopment2',
2335
)
2336
format_registry.register_lazy(
2337
("Bazaar development format 2 with subtree support "
2338
"(needs bzr.dev from before 1.8)\n"),
2339
'bzrlib.repofmt.pack_repo',
2340
'RepositoryFormatPackDevelopment2Subtree',
2341
)
1470
# and not independently creatable, so are not registered.
1471
RepositoryFormat.register_format(RepositoryFormat7())
1472
_default_format = RepositoryFormatKnit1()
1473
RepositoryFormat.register_format(_default_format)
1474
RepositoryFormat.set_default_format(_default_format)
1475
_legacy_formats = [RepositoryFormat4(),
1476
RepositoryFormat5(),
1477
RepositoryFormat6()]
2342
1478
2343
1479
2344
1480
class InterRepository(InterObject):
2353
1489
InterRepository.get(other).method_name(parameters).
2354
1490
"""
2355
1491
2356
_optimisers = []
1492
_optimisers = set()
2357
1493
"""The available optimised InterRepository types."""
2358
1494
2359
def copy_content(self, revision_id=None):
2360
raise NotImplementedError(self.copy_content)
2361
2362
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1495
@needs_write_lock
1496
def copy_content(self, revision_id=None, basis=None):
1497
"""Make a complete copy of the content in self into destination.
1498
1499
This is a destructive operation! Do not use it on existing
1500
repositories.
1501
1502
:param revision_id: Only copy the content needed to construct
1503
revision_id and its parents.
1504
:param basis: Copy the needed data preferentially from basis.
1505
"""
1506
try:
1507
self.target.set_make_working_trees(self.source.make_working_trees())
1508
except NotImplementedError:
1509
pass
1510
# grab the basis available data
1511
if basis is not None:
1512
self.target.fetch(basis, revision_id=revision_id)
1513
# but don't bother fetching if we have the needed data now.
1514
if (revision_id not in (None, NULL_REVISION) and
1515
self.target.has_revision(revision_id)):
1516
return
1517
self.target.fetch(self.source, revision_id=revision_id)
1518
1519
def _double_lock(self, lock_source, lock_target):
1520
"""Take out too locks, rolling back the first if the second throws."""
1521
lock_source()
1522
try:
1523
lock_target()
1524
except Exception:
1525
# we want to ensure that we don't leave source locked by mistake.
1526
# and any error on target should not confuse source.
1527
self.source.unlock()
1528
raise
1529
1530
@needs_write_lock
1531
def fetch(self, revision_id=None, pb=None):
2363
1532
"""Fetch the content required to construct revision_id.
2364
1533
2365
The content is copied from self.source to self.target.
1534
The content is copied from source to target.
2366
1535
2367
1536
:param revision_id: if None all content is copied, if NULL_REVISION no
2368
1537
content is copied.
2369
1538
:param pb: optional progress bar to use for progress reports. If not
2370
1539
provided a default one will be created.
2371
1540
2372
:returns: (copied_revision_count, failures).
2373
"""
2374
# Normally we should find a specific InterRepository subclass to do
2375
# the fetch; if nothing else then at least InterSameDataRepository.
2376
# If none of them is suitable it looks like fetching is not possible;
2377
# we try to give a good message why. _assert_same_model will probably
2378
# give a helpful message; otherwise a generic one.
2379
self._assert_same_model(self.source, self.target)
2380
raise errors.IncompatibleRepositories(self.source, self.target,
2381
"no suitableInterRepository found")
2382
2383
def _walk_to_common_revisions(self, revision_ids):
2384
"""Walk out from revision_ids in source to revisions target has.
2385
2386
:param revision_ids: The start point for the search.
2387
:return: A set of revision ids.
2388
"""
2389
target_graph = self.target.get_graph()
2390
revision_ids = frozenset(revision_ids)
2391
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2392
return graph.SearchResult(revision_ids, set(), 0, set())
2393
missing_revs = set()
2394
source_graph = self.source.get_graph()
2395
# ensure we don't pay silly lookup costs.
2396
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2397
null_set = frozenset([_mod_revision.NULL_REVISION])
2398
while True:
2399
try:
2400
next_revs, ghosts = searcher.next_with_ghosts()
2401
except StopIteration:
2402
break
2403
if revision_ids.intersection(ghosts):
2404
absent_ids = set(revision_ids.intersection(ghosts))
2405
# If all absent_ids are present in target, no error is needed.
2406
absent_ids.difference_update(
2407
set(target_graph.get_parent_map(absent_ids)))
2408
if absent_ids:
2409
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2410
# we don't care about other ghosts as we can't fetch them and
2411
# haven't been asked to.
2412
next_revs = set(next_revs)
2413
# we always have NULL_REVISION present.
2414
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2415
missing_revs.update(next_revs - have_revs)
2416
searcher.stop_searching_any(have_revs)
2417
return searcher.get_result()
2418
2419
@deprecated_method(one_two)
1541
Returns the copied revision count and the failed revisions in a tuple:
1542
(copied, failures).
1543
"""
1544
from bzrlib.fetch import GenericRepoFetcher
1545
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1546
self.source, self.source._format, self.target, self.target._format)
1547
f = GenericRepoFetcher(to_repository=self.target,
1548
from_repository=self.source,
1549
last_revision=revision_id,
1550
pb=pb)
1551
return f.count_copied, f.failed_revisions
1552
1553
def lock_read(self):
1554
"""Take out a logical read lock.
1555
1556
This will lock the source branch and the target branch. The source gets
1557
a read lock and the target a read lock.
1558
"""
1559
self._double_lock(self.source.lock_read, self.target.lock_read)
1560
1561
def lock_write(self):
1562
"""Take out a logical write lock.
1563
1564
This will lock the source branch and the target branch. The source gets
1565
a read lock and the target a write lock.
1566
"""
1567
self._double_lock(self.source.lock_read, self.target.lock_write)
1568
2420
1569
@needs_read_lock
2421
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1570
def missing_revision_ids(self, revision_id=None):
2422
1571
"""Return the revision ids that source has that target does not.
2423
1572
2424
1573
These are returned in topological order.
2425
1574
2426
1575
:param revision_id: only return revision ids included by this
2427
1576
revision_id.
2428
:param find_ghosts: If True find missing revisions in deep history
2429
rather than just finding the surface difference.
2430
"""
2431
return list(self.search_missing_revision_ids(
2432
revision_id, find_ghosts).get_keys())
2433
2434
@needs_read_lock
2435
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2436
"""Return the revision ids that source has that target does not.
2437
2438
:param revision_id: only return revision ids included by this
2439
revision_id.
2440
:param find_ghosts: If True find missing revisions in deep history
2441
rather than just finding the surface difference.
2442
:return: A bzrlib.graph.SearchResult.
2443
"""
2444
# stop searching at found target revisions.
2445
if not find_ghosts and revision_id is not None:
2446
return self._walk_to_common_revisions([revision_id])
1577
"""
2447
1578
# generic, possibly worst case, slow code path.
2448
1579
target_ids = set(self.target.all_revision_ids())
2449
1580
if revision_id is not None:
2450
1581
source_ids = self.source.get_ancestry(revision_id)
2451
if source_ids[0] is not None:
2452
raise AssertionError()
1582
assert source_ids[0] == None
2453
1583
source_ids.pop(0)
2454
1584
else:
2455
1585
source_ids = self.source.all_revision_ids()
2456
1586
result_set = set(source_ids).difference(target_ids)
2457
return self.source.revision_ids_to_search_result(result_set)
2458
2459
@staticmethod
2460
def _same_model(source, target):
2461
"""True if source and target have the same data representation.
2462
2463
Note: this is always called on the base class; overriding it in a
2464
subclass will have no effect.
2465
"""
2466
try:
2467
InterRepository._assert_same_model(source, target)
2468
return True
2469
except errors.IncompatibleRepositories, e:
2470
return False
2471
2472
@staticmethod
2473
def _assert_same_model(source, target):
2474
"""Raise an exception if two repositories do not use the same model.
2475
"""
2476
if source.supports_rich_root() != target.supports_rich_root():
2477
raise errors.IncompatibleRepositories(source, target,
2478
"different rich-root support")
2479
if source._serializer != target._serializer:
2480
raise errors.IncompatibleRepositories(source, target,
2481
"different serializers")
2482
2483
2484
class InterSameDataRepository(InterRepository):
2485
"""Code for converting between repositories that represent the same data.
2486
2487
Data format and model must match for this to work.
2488
"""
2489
2490
@classmethod
2491
def _get_repo_format_to_test(self):
2492
"""Repository format for testing with.
2493
2494
InterSameData can pull from subtree to subtree and from non-subtree to
2495
non-subtree, so we test this with the richest repository format.
2496
"""
2497
from bzrlib.repofmt import knitrepo
2498
return knitrepo.RepositoryFormatKnit3()
2499
2500
@staticmethod
2501
def is_compatible(source, target):
2502
return InterRepository._same_model(source, target)
2503
2504
@needs_write_lock
2505
def copy_content(self, revision_id=None):
2506
"""Make a complete copy of the content in self into destination.
2507
2508
This copies both the repository's revision data, and configuration information
2509
such as the make_working_trees setting.
2510
2511
This is a destructive operation! Do not use it on existing
2512
repositories.
2513
2514
:param revision_id: Only copy the content needed to construct
2515
revision_id and its parents.
2516
"""
2517
try:
2518
self.target.set_make_working_trees(self.source.make_working_trees())
2519
except NotImplementedError:
2520
pass
2521
# but don't bother fetching if we have the needed data now.
2522
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2523
self.target.has_revision(revision_id)):
2524
return
2525
self.target.fetch(self.source, revision_id=revision_id)
2526
2527
@needs_write_lock
2528
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2529
"""See InterRepository.fetch()."""
2530
from bzrlib.fetch import RepoFetcher
2531
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2532
self.source, self.source._format, self.target,
2533
self.target._format)
2534
f = RepoFetcher(to_repository=self.target,
2535
from_repository=self.source,
2536
last_revision=revision_id,
2537
pb=pb, find_ghosts=find_ghosts)
2538
return f.count_copied, f.failed_revisions
2539
2540
2541
class InterWeaveRepo(InterSameDataRepository):
2542
"""Optimised code paths between Weave based repositories.
2543
2544
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2545
implemented lazy inter-object optimisation.
2546
"""
2547
2548
@classmethod
2549
def _get_repo_format_to_test(self):
2550
from bzrlib.repofmt import weaverepo
2551
return weaverepo.RepositoryFormat7()
1587
# this may look like a no-op: its not. It preserves the ordering
1588
# other_ids had while only returning the members from other_ids
1589
# that we've decided we need.
1590
return [rev_id for rev_id in source_ids if rev_id in result_set]
1591
1592
def unlock(self):
1593
"""Release the locks on source and target."""
1594
try:
1595
self.target.unlock()
1596
finally:
1597
self.source.unlock()
1598
1599
1600
class InterWeaveRepo(InterRepository):
1601
"""Optimised code paths between Weave based repositories."""
1602
1603
_matching_repo_format = RepositoryFormat7()
1604
"""Repository format for testing with."""
2552
1605
2553
1606
@staticmethod
2554
1607
def is_compatible(source, target):
2558
1611
could lead to confusing results, and there is no need to be
2559
1612
overly general.
2560
1613
"""
2561
from bzrlib.repofmt.weaverepo import (
2562
RepositoryFormat5,
2563
RepositoryFormat6,
2564
RepositoryFormat7,
2565
)
2566
1614
try:
2567
1615
return (isinstance(source._format, (RepositoryFormat5,
2568
1616
RepositoryFormat6,
2574
1622
return False
2575
1623
2576
1624
@needs_write_lock
2577
def copy_content(self, revision_id=None):
1625
def copy_content(self, revision_id=None, basis=None):
2578
1626
"""See InterRepository.copy_content()."""
2579
1627
# weave specific optimised path:
2580
try:
2581
self.target.set_make_working_trees(self.source.make_working_trees())
2582
except (errors.RepositoryUpgradeRequired, NotImplemented):
2583
pass
2584
# FIXME do not peek!
2585
if self.source._transport.listable():
2586
pb = ui.ui_factory.nested_progress_bar()
1628
if basis is not None:
1629
# copy the basis in, then fetch remaining data.
1630
basis.copy_content_into(self.target, revision_id)
1631
# the basis copy_content_into could miss-set this.
2587
1632
try:
2588
self.target.texts.insert_record_stream(
2589
self.source.texts.get_record_stream(
2590
self.source.texts.keys(), 'topological', False))
2591
pb.update('copying inventory', 0, 1)
2592
self.target.inventories.insert_record_stream(
2593
self.source.inventories.get_record_stream(
2594
self.source.inventories.keys(), 'topological', False))
2595
self.target.signatures.insert_record_stream(
2596
self.source.signatures.get_record_stream(
2597
self.source.signatures.keys(),
2598
'unordered', True))
2599
self.target.revisions.insert_record_stream(
2600
self.source.revisions.get_record_stream(
2601
self.source.revisions.keys(),
2602
'topological', True))
2603
finally:
2604
pb.finished()
2605
else:
1633
self.target.set_make_working_trees(self.source.make_working_trees())
1634
except NotImplementedError:
1635
pass
2606
1636
self.target.fetch(self.source, revision_id=revision_id)
1637
else:
1638
try:
1639
self.target.set_make_working_trees(self.source.make_working_trees())
1640
except NotImplementedError:
1641
pass
1642
# FIXME do not peek!
1643
if self.source.control_files._transport.listable():
1644
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1645
try:
1646
self.target.weave_store.copy_all_ids(
1647
self.source.weave_store,
1648
pb=pb,
1649
from_transaction=self.source.get_transaction(),
1650
to_transaction=self.target.get_transaction())
1651
pb.update('copying inventory', 0, 1)
1652
self.target.control_weaves.copy_multi(
1653
self.source.control_weaves, ['inventory'],
1654
from_transaction=self.source.get_transaction(),
1655
to_transaction=self.target.get_transaction())
1656
self.target._revision_store.text_store.copy_all_ids(
1657
self.source._revision_store.text_store,
1658
pb=pb)
1659
finally:
1660
pb.finished()
1661
else:
1662
self.target.fetch(self.source, revision_id=revision_id)
2607
1663
2608
1664
@needs_write_lock
2609
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1665
def fetch(self, revision_id=None, pb=None):
2610
1666
"""See InterRepository.fetch()."""
2611
from bzrlib.fetch import RepoFetcher
1667
from bzrlib.fetch import GenericRepoFetcher
2612
1668
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2613
1669
self.source, self.source._format, self.target, self.target._format)
2614
f = RepoFetcher(to_repository=self.target,
1670
f = GenericRepoFetcher(to_repository=self.target,
2615
1671
from_repository=self.source,
2616
1672
last_revision=revision_id,
2617
pb=pb, find_ghosts=find_ghosts)
1673
pb=pb)
2618
1674
return f.count_copied, f.failed_revisions
2619
1675
2620
1676
@needs_read_lock
2621
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1677
def missing_revision_ids(self, revision_id=None):
2622
1678
"""See InterRepository.missing_revision_ids()."""
2623
1679
# we want all revisions to satisfy revision_id in source.
2624
1680
# but we don't want to stat every file here and there.
2633
1689
# - RBC 20060209
2634
1690
if revision_id is not None:
2635
1691
source_ids = self.source.get_ancestry(revision_id)
2636
if source_ids[0] is not None:
2637
raise AssertionError()
1692
assert source_ids[0] == None
2638
1693
source_ids.pop(0)
2639
1694
else:
2640
1695
source_ids = self.source._all_possible_ids()
2645
1700
# we do not have a revision as that would be pointless.
2646
1701
target_ids = set(self.target._all_possible_ids())
2647
1702
possibly_present_revisions = target_ids.intersection(source_ids_set)
2648
actually_present_revisions = set(
2649
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1703
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2650
1704
required_revisions = source_ids_set.difference(actually_present_revisions)
1705
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2651
1706
if revision_id is not None:
2652
1707
# we used get_ancestry to determine source_ids then we are assured all
2653
1708
# revisions referenced are present as they are installed in topological order.
2654
1709
# and the tip revision was validated by get_ancestry.
2655
result_set = required_revisions
1710
return required_topo_revisions
2656
1711
else:
2657
1712
# if we just grabbed the possibly available ids, then
2658
1713
# we only have an estimate of whats available and need to validate
2659
1714
# that against the revision records.
2660
result_set = set(
2661
self.source._eliminate_revisions_not_present(required_revisions))
2662
return self.source.revision_ids_to_search_result(result_set)
2663
2664
2665
class InterKnitRepo(InterSameDataRepository):
1715
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1716
1717
1718
class InterKnitRepo(InterRepository):
2666
1719
"""Optimised code paths between Knit based repositories."""
2667
1720
2668
@classmethod
2669
def _get_repo_format_to_test(self):
2670
from bzrlib.repofmt import knitrepo
2671
return knitrepo.RepositoryFormatKnit1()
1721
_matching_repo_format = RepositoryFormatKnit1()
1722
"""Repository format for testing with."""
2672
1723
2673
1724
@staticmethod
2674
1725
def is_compatible(source, target):
2678
1729
could lead to confusing results, and there is no need to be
2679
1730
overly general.
2680
1731
"""
2681
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2682
1732
try:
2683
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2684
isinstance(target._format, RepositoryFormatKnit))
1733
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1734
isinstance(target._format, (RepositoryFormatKnit1)))
2685
1735
except AttributeError:
2686
1736
return False
2687
return are_knits and InterRepository._same_model(source, target)
2688
1737
2689
1738
@needs_write_lock
2690
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1739
def fetch(self, revision_id=None, pb=None):
2691
1740
"""See InterRepository.fetch()."""
2692
from bzrlib.fetch import RepoFetcher
1741
from bzrlib.fetch import KnitRepoFetcher
2693
1742
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2694
1743
self.source, self.source._format, self.target, self.target._format)
2695
f = RepoFetcher(to_repository=self.target,
1744
f = KnitRepoFetcher(to_repository=self.target,
2696
1745
from_repository=self.source,
2697
1746
last_revision=revision_id,
2698
pb=pb, find_ghosts=find_ghosts)
1747
pb=pb)
2699
1748
return f.count_copied, f.failed_revisions
2700
1749
2701
1750
@needs_read_lock
2702
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1751
def missing_revision_ids(self, revision_id=None):
2703
1752
"""See InterRepository.missing_revision_ids()."""
2704
1753
if revision_id is not None:
2705
1754
source_ids = self.source.get_ancestry(revision_id)
2706
if source_ids[0] is not None:
2707
raise AssertionError()
1755
assert source_ids[0] == None
2708
1756
source_ids.pop(0)
2709
1757
else:
2710
source_ids = self.source.all_revision_ids()
1758
source_ids = self.source._all_possible_ids()
2711
1759
source_ids_set = set(source_ids)
2712
1760
# source_ids is the worst possible case we may need to pull.
2713
1761
# now we want to filter source_ids against what we actually
2714
1762
# have in target, but don't try to check for existence where we know
2715
1763
# we do not have a revision as that would be pointless.
2716
target_ids = set(self.target.all_revision_ids())
1764
target_ids = set(self.target._all_possible_ids())
2717
1765
possibly_present_revisions = target_ids.intersection(source_ids_set)
2718
actually_present_revisions = set(
2719
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1766
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2720
1767
required_revisions = source_ids_set.difference(actually_present_revisions)
1768
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2721
1769
if revision_id is not None:
2722
1770
# we used get_ancestry to determine source_ids then we are assured all
2723
1771
# revisions referenced are present as they are installed in topological order.
2724
1772
# and the tip revision was validated by get_ancestry.
2725
result_set = required_revisions
1773
return required_topo_revisions
2726
1774
else:
2727
1775
# if we just grabbed the possibly available ids, then
2728
1776
# we only have an estimate of whats available and need to validate
2729
1777
# that against the revision records.
2730
result_set = set(
2731
self.source._eliminate_revisions_not_present(required_revisions))
2732
return self.source.revision_ids_to_search_result(result_set)
2733
2734
2735
class InterPackRepo(InterSameDataRepository):
2736
"""Optimised code paths between Pack based repositories."""
2737
2738
@classmethod
2739
def _get_repo_format_to_test(self):
2740
from bzrlib.repofmt import pack_repo
2741
return pack_repo.RepositoryFormatKnitPack1()
2742
2743
@staticmethod
2744
def is_compatible(source, target):
2745
"""Be compatible with known Pack formats.
2746
2747
We don't test for the stores being of specific types because that
2748
could lead to confusing results, and there is no need to be
2749
overly general.
2750
"""
2751
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2752
try:
2753
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2754
isinstance(target._format, RepositoryFormatPack))
2755
except AttributeError:
2756
return False
2757
return are_packs and InterRepository._same_model(source, target)
2758
2759
@needs_write_lock
2760
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2761
"""See InterRepository.fetch()."""
2762
if (len(self.source._fallback_repositories) > 0 or
2763
len(self.target._fallback_repositories) > 0):
2764
# The pack layer is not aware of fallback repositories, so when
2765
# fetching from a stacked repository or into a stacked repository
2766
# we use the generic fetch logic which uses the VersionedFiles
2767
# attributes on repository.
2768
from bzrlib.fetch import RepoFetcher
2769
fetcher = RepoFetcher(self.target, self.source, revision_id,
2770
pb, find_ghosts)
2771
return fetcher.count_copied, fetcher.failed_revisions
2772
from bzrlib.repofmt.pack_repo import Packer
2773
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2774
self.source, self.source._format, self.target, self.target._format)
2775
self.count_copied = 0
2776
if revision_id is None:
2777
# TODO:
2778
# everything to do - use pack logic
2779
# to fetch from all packs to one without
2780
# inventory parsing etc, IFF nothing to be copied is in the target.
2781
# till then:
2782
source_revision_ids = frozenset(self.source.all_revision_ids())
2783
revision_ids = source_revision_ids - \
2784
frozenset(self.target.get_parent_map(source_revision_ids))
2785
revision_keys = [(revid,) for revid in revision_ids]
2786
index = self.target._pack_collection.revision_index.combined_index
2787
present_revision_ids = set(item[1][0] for item in
2788
index.iter_entries(revision_keys))
2789
revision_ids = set(revision_ids) - present_revision_ids
2790
# implementing the TODO will involve:
2791
# - detecting when all of a pack is selected
2792
# - avoiding as much as possible pre-selection, so the
2793
# more-core routines such as create_pack_from_packs can filter in
2794
# a just-in-time fashion. (though having a HEADS list on a
2795
# repository might make this a lot easier, because we could
2796
# sensibly detect 'new revisions' without doing a full index scan.
2797
elif _mod_revision.is_null(revision_id):
2798
# nothing to do:
2799
return (0, [])
2800
else:
2801
try:
2802
revision_ids = self.search_missing_revision_ids(revision_id,
2803
find_ghosts=find_ghosts).get_keys()
2804
except errors.NoSuchRevision:
2805
raise errors.InstallFailed([revision_id])
2806
if len(revision_ids) == 0:
2807
return (0, [])
2808
packs = self.source._pack_collection.all_packs()
2809
pack = Packer(self.target._pack_collection, packs, '.fetch',
2810
revision_ids).pack()
2811
if pack is not None:
2812
self.target._pack_collection._save_pack_names()
2813
# Trigger an autopack. This may duplicate effort as we've just done
2814
# a pack creation, but for now it is simpler to think about as
2815
# 'upload data, then repack if needed'.
2816
self.target._pack_collection.autopack()
2817
return (pack.get_revision_count(), [])
2818
else:
2819
return (0, [])
2820
2821
@needs_read_lock
2822
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2823
"""See InterRepository.missing_revision_ids().
2824
2825
:param find_ghosts: Find ghosts throughout the ancestry of
2826
revision_id.
2827
"""
2828
if not find_ghosts and revision_id is not None:
2829
return self._walk_to_common_revisions([revision_id])
2830
elif revision_id is not None:
2831
# Find ghosts: search for revisions pointing from one repository to
2832
# the other, and vice versa, anywhere in the history of revision_id.
2833
graph = self.target.get_graph(other_repository=self.source)
2834
searcher = graph._make_breadth_first_searcher([revision_id])
2835
found_ids = set()
2836
while True:
2837
try:
2838
next_revs, ghosts = searcher.next_with_ghosts()
2839
except StopIteration:
2840
break
2841
if revision_id in ghosts:
2842
raise errors.NoSuchRevision(self.source, revision_id)
2843
found_ids.update(next_revs)
2844
found_ids.update(ghosts)
2845
found_ids = frozenset(found_ids)
2846
# Double query here: should be able to avoid this by changing the
2847
# graph api further.
2848
result_set = found_ids - frozenset(
2849
self.target.get_parent_map(found_ids))
2850
else:
2851
source_ids = self.source.all_revision_ids()
2852
# source_ids is the worst possible case we may need to pull.
2853
# now we want to filter source_ids against what we actually
2854
# have in target, but don't try to check for existence where we know
2855
# we do not have a revision as that would be pointless.
2856
target_ids = set(self.target.all_revision_ids())
2857
result_set = set(source_ids).difference(target_ids)
2858
return self.source.revision_ids_to_search_result(result_set)
2859
2860
2861
class InterModel1and2(InterRepository):
2862
2863
@classmethod
2864
def _get_repo_format_to_test(self):
2865
return None
2866
2867
@staticmethod
2868
def is_compatible(source, target):
2869
if not source.supports_rich_root() and target.supports_rich_root():
2870
return True
2871
else:
2872
return False
2873
2874
@needs_write_lock
2875
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2876
"""See InterRepository.fetch()."""
2877
from bzrlib.fetch import Model1toKnit2Fetcher
2878
f = Model1toKnit2Fetcher(to_repository=self.target,
2879
from_repository=self.source,
2880
last_revision=revision_id,
2881
pb=pb, find_ghosts=find_ghosts)
2882
return f.count_copied, f.failed_revisions
2883
2884
@needs_write_lock
2885
def copy_content(self, revision_id=None):
2886
"""Make a complete copy of the content in self into destination.
2887
2888
This is a destructive operation! Do not use it on existing
2889
repositories.
2890
2891
:param revision_id: Only copy the content needed to construct
2892
revision_id and its parents.
2893
"""
2894
try:
2895
self.target.set_make_working_trees(self.source.make_working_trees())
2896
except NotImplementedError:
2897
pass
2898
# but don't bother fetching if we have the needed data now.
2899
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2900
self.target.has_revision(revision_id)):
2901
return
2902
self.target.fetch(self.source, revision_id=revision_id)
2903
2904
2905
class InterKnit1and2(InterKnitRepo):
2906
2907
@classmethod
2908
def _get_repo_format_to_test(self):
2909
return None
2910
2911
@staticmethod
2912
def is_compatible(source, target):
2913
"""Be compatible with Knit1 source and Knit3 target"""
2914
try:
2915
from bzrlib.repofmt.knitrepo import (
2916
RepositoryFormatKnit1,
2917
RepositoryFormatKnit3,
2918
)
2919
from bzrlib.repofmt.pack_repo import (
2920
RepositoryFormatKnitPack1,
2921
RepositoryFormatKnitPack3,
2922
RepositoryFormatKnitPack4,
2923
RepositoryFormatKnitPack5,
2924
RepositoryFormatKnitPack5RichRoot,
2925
RepositoryFormatPackDevelopment2,
2926
RepositoryFormatPackDevelopment2Subtree,
2927
)
2928
norichroot = (
2929
RepositoryFormatKnit1, # no rr, no subtree
2930
RepositoryFormatKnitPack1, # no rr, no subtree
2931
RepositoryFormatPackDevelopment2, # no rr, no subtree
2932
RepositoryFormatKnitPack5, # no rr, no subtree
2933
)
2934
richroot = (
2935
RepositoryFormatKnit3, # rr, subtree
2936
RepositoryFormatKnitPack3, # rr, subtree
2937
RepositoryFormatKnitPack4, # rr, no subtree
2938
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2939
RepositoryFormatPackDevelopment2Subtree, # rr, subtree
2940
)
2941
for format in norichroot:
2942
if format.rich_root_data:
2943
raise AssertionError('Format %s is a rich-root format'
2944
' but is included in the non-rich-root list'
2945
% (format,))
2946
for format in richroot:
2947
if not format.rich_root_data:
2948
raise AssertionError('Format %s is not a rich-root format'
2949
' but is included in the rich-root list'
2950
% (format,))
2951
# TODO: One alternative is to just check format.rich_root_data,
2952
# instead of keeping membership lists. However, the formats
2953
# *also* have to use the same 'Knit' style of storage
2954
# (line-deltas, fulltexts, etc.)
2955
return (isinstance(source._format, norichroot) and
2956
isinstance(target._format, richroot))
2957
except AttributeError:
2958
return False
2959
2960
@needs_write_lock
2961
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2962
"""See InterRepository.fetch()."""
2963
from bzrlib.fetch import Knit1to2Fetcher
2964
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2965
self.source, self.source._format, self.target,
2966
self.target._format)
2967
f = Knit1to2Fetcher(to_repository=self.target,
2968
from_repository=self.source,
2969
last_revision=revision_id,
2970
pb=pb, find_ghosts=find_ghosts)
2971
return f.count_copied, f.failed_revisions
2972
2973
2974
class InterDifferingSerializer(InterKnitRepo):
2975
2976
@classmethod
2977
def _get_repo_format_to_test(self):
2978
return None
2979
2980
@staticmethod
2981
def is_compatible(source, target):
2982
"""Be compatible with Knit2 source and Knit3 target"""
2983
if source.supports_rich_root() != target.supports_rich_root():
2984
return False
2985
# Ideally, we'd support fetching if the source had no tree references
2986
# even if it supported them...
2987
if (getattr(source, '_format.supports_tree_reference', False) and
2988
not getattr(target, '_format.supports_tree_reference', False)):
2989
return False
2990
return True
2991
2992
@needs_write_lock
2993
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2994
"""See InterRepository.fetch()."""
2995
revision_ids = self.target.search_missing_revision_ids(self.source,
2996
revision_id, find_ghosts=find_ghosts).get_keys()
2997
revision_ids = tsort.topo_sort(
2998
self.source.get_graph().get_parent_map(revision_ids))
2999
def revisions_iterator():
3000
for current_revision_id in revision_ids:
3001
revision = self.source.get_revision(current_revision_id)
3002
tree = self.source.revision_tree(current_revision_id)
3003
try:
3004
signature = self.source.get_signature_text(
3005
current_revision_id)
3006
except errors.NoSuchRevision:
3007
signature = None
3008
yield revision, tree, signature
3009
if pb is None:
3010
my_pb = ui.ui_factory.nested_progress_bar()
3011
pb = my_pb
3012
else:
3013
my_pb = None
3014
try:
3015
install_revisions(self.target, revisions_iterator(),
3016
len(revision_ids), pb)
3017
finally:
3018
if my_pb is not None:
3019
my_pb.finished()
3020
return len(revision_ids), 0
3021
3022
3023
class InterOtherToRemote(InterRepository):
3024
3025
def __init__(self, source, target):
3026
InterRepository.__init__(self, source, target)
3027
self._real_inter = None
3028
3029
@staticmethod
3030
def is_compatible(source, target):
3031
if isinstance(target, remote.RemoteRepository):
3032
return True
3033
return False
3034
3035
def _ensure_real_inter(self):
3036
if self._real_inter is None:
3037
self.target._ensure_real()
3038
real_target = self.target._real_repository
3039
self._real_inter = InterRepository.get(self.source, real_target)
3040
3041
def copy_content(self, revision_id=None):
3042
self._ensure_real_inter()
3043
self._real_inter.copy_content(revision_id=revision_id)
3044
3045
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3046
self._ensure_real_inter()
3047
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3048
find_ghosts=find_ghosts)
3049
3050
@classmethod
3051
def _get_repo_format_to_test(self):
3052
return None
3053
3054
3055
class InterRemoteToOther(InterRepository):
3056
3057
def __init__(self, source, target):
3058
InterRepository.__init__(self, source, target)
3059
self._real_inter = None
3060
3061
@staticmethod
3062
def is_compatible(source, target):
3063
if not isinstance(source, remote.RemoteRepository):
3064
return False
3065
# Is source's model compatible with target's model?
3066
source._ensure_real()
3067
real_source = source._real_repository
3068
if isinstance(real_source, remote.RemoteRepository):
3069
raise NotImplementedError(
3070
"We don't support remote repos backed by remote repos yet.")
3071
return InterRepository._same_model(real_source, target)
3072
3073
def _ensure_real_inter(self):
3074
if self._real_inter is None:
3075
self.source._ensure_real()
3076
real_source = self.source._real_repository
3077
self._real_inter = InterRepository.get(real_source, self.target)
3078
3079
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3080
self._ensure_real_inter()
3081
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3082
find_ghosts=find_ghosts)
3083
3084
def copy_content(self, revision_id=None):
3085
self._ensure_real_inter()
3086
self._real_inter.copy_content(revision_id=revision_id)
3087
3088
@classmethod
3089
def _get_repo_format_to_test(self):
3090
return None
3091
3092
3093
3094
InterRepository.register_optimiser(InterDifferingSerializer)
3095
InterRepository.register_optimiser(InterSameDataRepository)
1778
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1779
3096
1780
InterRepository.register_optimiser(InterWeaveRepo)
3097
1781
InterRepository.register_optimiser(InterKnitRepo)
3098
InterRepository.register_optimiser(InterModel1and2)
3099
InterRepository.register_optimiser(InterKnit1and2)
3100
InterRepository.register_optimiser(InterPackRepo)
3101
InterRepository.register_optimiser(InterOtherToRemote)
3102
InterRepository.register_optimiser(InterRemoteToOther)
1782
1783
1784
class RepositoryTestProviderAdapter(object):
1785
"""A tool to generate a suite testing multiple repository formats at once.
1786
1787
This is done by copying the test once for each transport and injecting
1788
the transport_server, transport_readonly_server, and bzrdir_format and
1789
repository_format classes into each copy. Each copy is also given a new id()
1790
to make it easy to identify.
1791
"""
1792
1793
def __init__(self, transport_server, transport_readonly_server, formats):
1794
self._transport_server = transport_server
1795
self._transport_readonly_server = transport_readonly_server
1796
self._formats = formats
1797
1798
def adapt(self, test):
1799
result = TestSuite()
1800
for repository_format, bzrdir_format in self._formats:
1801
new_test = deepcopy(test)
1802
new_test.transport_server = self._transport_server
1803
new_test.transport_readonly_server = self._transport_readonly_server
1804
new_test.bzrdir_format = bzrdir_format
1805
new_test.repository_format = repository_format
1806
def make_new_test_id():
1807
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1808
return lambda: new_id
1809
new_test.id = make_new_test_id()
1810
result.addTest(new_test)
1811
return result
1812
1813
1814
class InterRepositoryTestProviderAdapter(object):
1815
"""A tool to generate a suite testing multiple inter repository formats.
1816
1817
This is done by copying the test once for each interrepo provider and injecting
1818
the transport_server, transport_readonly_server, repository_format and
1819
repository_to_format classes into each copy.
1820
Each copy is also given a new id() to make it easy to identify.
1821
"""
1822
1823
def __init__(self, transport_server, transport_readonly_server, formats):
1824
self._transport_server = transport_server
1825
self._transport_readonly_server = transport_readonly_server
1826
self._formats = formats
1827
1828
def adapt(self, test):
1829
result = TestSuite()
1830
for interrepo_class, repository_format, repository_format_to in self._formats:
1831
new_test = deepcopy(test)
1832
new_test.transport_server = self._transport_server
1833
new_test.transport_readonly_server = self._transport_readonly_server
1834
new_test.interrepo_class = interrepo_class
1835
new_test.repository_format = repository_format
1836
new_test.repository_format_to = repository_format_to
1837
def make_new_test_id():
1838
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1839
return lambda: new_id
1840
new_test.id = make_new_test_id()
1841
result.addTest(new_test)
1842
return result
1843
1844
@staticmethod
1845
def default_test_list():
1846
"""Generate the default list of interrepo permutations to test."""
1847
result = []
1848
# test the default InterRepository between format 6 and the current
1849
# default format.
1850
# XXX: robertc 20060220 reinstate this when there are two supported
1851
# formats which do not have an optimal code path between them.
1852
result.append((InterRepository,
1853
RepositoryFormat6(),
1854
RepositoryFormatKnit1()))
1855
for optimiser in InterRepository._optimisers:
1856
result.append((optimiser,
1857
optimiser._matching_repo_format,
1858
optimiser._matching_repo_format
1859
))
1860
# if there are specific combinations we want to use, we can add them
1861
# here.
1862
return result
3103
1863
3104
1864
3105
1865
class CopyConverter(object):
3131
1891
self.step('Moving repository to repository.backup')
3132
1892
self.repo_dir.transport.move('repository', 'repository.backup')
3133
1893
backup_transport = self.repo_dir.transport.clone('repository.backup')
3134
repo._format.check_conversion_target(self.target_format)
3135
1894
self.source_repo = repo._format.open(self.repo_dir,
3136
1895
_found=True,
3137
1896
_override_transport=backup_transport)
3154
1913
self.pb.update(message, self.count, self.total)
3155
1914
3156
1915
3157
_unescape_map = {
3158
'apos':"'",
3159
'quot':'"',
3160
'amp':'&',
3161
'lt':'<',
3162
'gt':'>'
3163
}
3164
3165
3166
def _unescaper(match, _map=_unescape_map):
3167
code = match.group(1)
3168
try:
3169
return _map[code]
3170
except KeyError:
3171
if not code.startswith('#'):
3172
raise
3173
return unichr(int(code[1:])).encode('utf8')
3174
3175
3176
_unescape_re = None
3177
3178
1916
class CommitBuilder(object):
1917
"""Provides an interface to build up a commit.
1918
1919
This allows describing a tree to be committed without needing to
1920
know the internals of the format of the repository.
1921
"""
1922
def __init__(self, repository, parents, config, timestamp=None,
1923
timezone=None, committer=None, revprops=None,
1924
revision_id=None):
1925
"""Initiate a CommitBuilder.
1926
1927
:param repository: Repository to commit to.
1928
:param parents: Revision ids of the parents of the new revision.
1929
:param config: Configuration to use.
1930
:param timestamp: Optional timestamp recorded for commit.
1931
:param timezone: Optional timezone for timestamp.
1932
:param committer: Optional committer to set for commit.
1933
:param revprops: Optional dictionary of revision properties.
1934
:param revision_id: Optional revision id.
1935
"""
1936
self._config = config
1937
1938
if committer is None:
1939
self._committer = self._config.username()
1940
else:
1941
assert isinstance(committer, basestring), type(committer)
1942
self._committer = committer
1943
1944
self.new_inventory = Inventory()
1945
self._new_revision_id = revision_id
1946
self.parents = parents
1947
self.repository = repository
1948
1949
self._revprops = {}
1950
if revprops is not None:
1951
self._revprops.update(revprops)
1952
1953
if timestamp is None:
1954
self._timestamp = time.time()
1955
else:
1956
self._timestamp = long(timestamp)
1957
1958
if timezone is None:
1959
self._timezone = local_time_offset()
1960
else:
1961
self._timezone = int(timezone)
1962
1963
self._generate_revision_if_needed()
1964
1965
def commit(self, message):
1966
"""Make the actual commit.
1967
1968
:return: The revision id of the recorded revision.
1969
"""
1970
rev = Revision(timestamp=self._timestamp,
1971
timezone=self._timezone,
1972
committer=self._committer,
1973
message=message,
1974
inventory_sha1=self.inv_sha1,
1975
revision_id=self._new_revision_id,
1976
properties=self._revprops)
1977
rev.parent_ids = self.parents
1978
self.repository.add_revision(self._new_revision_id, rev,
1979
self.new_inventory, self._config)
1980
return self._new_revision_id
1981
1982
def finish_inventory(self):
1983
"""Tell the builder that the inventory is finished."""
1984
self.new_inventory.revision_id = self._new_revision_id
1985
self.inv_sha1 = self.repository.add_inventory(
1986
self._new_revision_id,
1987
self.new_inventory,
1988
self.parents
1989
)
1990
1991
def _gen_revision_id(self):
1992
"""Return new revision-id."""
1993
s = '%s-%s-' % (self._config.user_email(),
1994
compact_date(self._timestamp))
1995
s += hexlify(rand_bytes(8))
1996
return s
1997
1998
def _generate_revision_if_needed(self):
1999
"""Create a revision id if None was supplied.
2000
2001
If the repository can not support user-specified revision ids
2002
they should override this function and raise UnsupportedOperation
2003
if _new_revision_id is not None.
2004
2005
:raises: UnsupportedOperation
2006
"""
2007
if self._new_revision_id is None:
2008
self._new_revision_id = self._gen_revision_id()
2009
2010
def record_entry_contents(self, ie, parent_invs, path, tree):
2011
"""Record the content of ie from tree into the commit if needed.
2012
2013
:param ie: An inventory entry present in the commit.
2014
:param parent_invs: The inventories of the parent revisions of the
2015
commit.
2016
:param path: The path the entry is at in the tree.
2017
:param tree: The tree which contains this entry and should be used to
2018
obtain content.
2019
"""
2020
self.new_inventory.add(ie)
2021
2022
# ie.revision is always None if the InventoryEntry is considered
2023
# for committing. ie.snapshot will record the correct revision
2024
# which may be the sole parent if it is untouched.
2025
if ie.revision is not None:
2026
return
2027
previous_entries = ie.find_previous_heads(
2028
parent_invs,
2029
self.repository.weave_store,
2030
self.repository.get_transaction())
2031
# we are creating a new revision for ie in the history store
2032
# and inventory.
2033
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2034
2035
def modified_directory(self, file_id, file_parents):
2036
"""Record the presence of a symbolic link.
2037
2038
:param file_id: The file_id of the link to record.
2039
:param file_parents: The per-file parent revision ids.
2040
"""
2041
self._add_text_to_weave(file_id, [], file_parents.keys())
2042
2043
def modified_file_text(self, file_id, file_parents,
2044
get_content_byte_lines, text_sha1=None,
2045
text_size=None):
2046
"""Record the text of file file_id
2047
2048
:param file_id: The file_id of the file to record the text of.
2049
:param file_parents: The per-file parent revision ids.
2050
:param get_content_byte_lines: A callable which will return the byte
2051
lines for the file.
2052
:param text_sha1: Optional SHA1 of the file contents.
2053
:param text_size: Optional size of the file contents.
2054
"""
2055
mutter('storing text of file {%s} in revision {%s} into %r',
2056
file_id, self._new_revision_id, self.repository.weave_store)
2057
# special case to avoid diffing on renames or
2058
# reparenting
2059
if (len(file_parents) == 1
2060
and text_sha1 == file_parents.values()[0].text_sha1
2061
and text_size == file_parents.values()[0].text_size):
2062
previous_ie = file_parents.values()[0]
2063
versionedfile = self.repository.weave_store.get_weave(file_id,
2064
self.repository.get_transaction())
2065
versionedfile.clone_text(self._new_revision_id,
2066
previous_ie.revision, file_parents.keys())
2067
return text_sha1, text_size
2068
else:
2069
new_lines = get_content_byte_lines()
2070
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2071
# should return the SHA1 and size
2072
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2073
return bzrlib.osutils.sha_strings(new_lines), \
2074
sum(map(len, new_lines))
2075
2076
def modified_link(self, file_id, file_parents, link_target):
2077
"""Record the presence of a symbolic link.
2078
2079
:param file_id: The file_id of the link to record.
2080
:param file_parents: The per-file parent revision ids.
2081
:param link_target: Target location of this link.
2082
"""
2083
self._add_text_to_weave(file_id, [], file_parents.keys())
2084
2085
def _add_text_to_weave(self, file_id, new_lines, parents):
2086
versionedfile = self.repository.weave_store.get_weave_or_empty(
2087
file_id, self.repository.get_transaction())
2088
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2089
versionedfile.clear_cache()
2090
2091
2092
# Copied from xml.sax.saxutils
3179
2093
def _unescape_xml(data):
3180
"""Unescape predefined XML entities in a string of data."""
3181
global _unescape_re
3182
if _unescape_re is None:
3183
_unescape_re = re.compile('\&([^;]*);')
3184
return _unescape_re.sub(_unescaper, data)
3185
3186
3187
class _VersionedFileChecker(object):
3188
3189
def __init__(self, repository):
3190
self.repository = repository
3191
self.text_index = self.repository._generate_text_key_index()
3192
3193
def calculate_file_version_parents(self, text_key):
3194
"""Calculate the correct parents for a file version according to
3195
the inventories.
3196
"""
3197
parent_keys = self.text_index[text_key]
3198
if parent_keys == [_mod_revision.NULL_REVISION]:
3199
return ()
3200
return tuple(parent_keys)
3201
3202
def check_file_version_parents(self, texts, progress_bar=None):
3203
"""Check the parents stored in a versioned file are correct.
3204
3205
It also detects file versions that are not referenced by their
3206
corresponding revision's inventory.
3207
3208
:returns: A tuple of (wrong_parents, dangling_file_versions).
3209
wrong_parents is a dict mapping {revision_id: (stored_parents,
3210
correct_parents)} for each revision_id where the stored parents
3211
are not correct. dangling_file_versions is a set of (file_id,
3212
revision_id) tuples for versions that are present in this versioned
3213
file, but not used by the corresponding inventory.
3214
"""
3215
wrong_parents = {}
3216
self.file_ids = set([file_id for file_id, _ in
3217
self.text_index.iterkeys()])
3218
# text keys is now grouped by file_id
3219
n_weaves = len(self.file_ids)
3220
files_in_revisions = {}
3221
revisions_of_files = {}
3222
n_versions = len(self.text_index)
3223
progress_bar.update('loading text store', 0, n_versions)
3224
parent_map = self.repository.texts.get_parent_map(self.text_index)
3225
# On unlistable transports this could well be empty/error...
3226
text_keys = self.repository.texts.keys()
3227
unused_keys = frozenset(text_keys) - set(self.text_index)
3228
for num, key in enumerate(self.text_index.iterkeys()):
3229
if progress_bar is not None:
3230
progress_bar.update('checking text graph', num, n_versions)
3231
correct_parents = self.calculate_file_version_parents(key)
3232
try:
3233
knit_parents = parent_map[key]
3234
except errors.RevisionNotPresent:
3235
# Missing text!
3236
knit_parents = None
3237
if correct_parents != knit_parents:
3238
wrong_parents[key] = (knit_parents, correct_parents)
3239
return wrong_parents, unused_keys
3240
3241
3242
def _old_get_graph(repository, revision_id):
3243
"""DO NOT USE. That is all. I'm serious."""
3244
graph = repository.get_graph()
3245
revision_graph = dict(((key, value) for key, value in
3246
graph.iter_ancestry([revision_id]) if value is not None))
3247
return _strip_NULL_ghosts(revision_graph)
3248
3249
3250
def _strip_NULL_ghosts(revision_graph):
3251
"""Also don't use this. more compatibility code for unmigrated clients."""
3252
# Filter ghosts, and null:
3253
if _mod_revision.NULL_REVISION in revision_graph:
3254
del revision_graph[_mod_revision.NULL_REVISION]
3255
for key, parents in revision_graph.items():
3256
revision_graph[key] = tuple(parent for parent in parents if parent
3257
in revision_graph)
3258
return revision_graph
2094
"""Unescape &, <, and > in a string of data.
2095
"""
2096
data = data.replace("<", "<")
2097
data = data.replace(">", ">")
2098
# must do ampersand last
2099
return data.replace("&", "&")
Loggerhead is a web-based interface for Breezy
Version: 2.0.1