« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 3665.3.5
- compare with another revision
- download diff
- download tarball
- view history from revision 3665.3.5
- Committer: John Arbash Meinel
- Date: 2008-09-02 17:52:00 UTC
- mto: This revision was merged to the branch mainline in revision 3679.
- Revision ID: john@arbash-meinel.com-20080902175200-nge9qgk0gklkd5ew
Move the point at which we 'buffer_all' if we've read >50% of the index.
We were doing it as soon as you entered 'iter_entries', but often you may already have enough
info to return results. And for small mostly local ops, we don't need to buffer all.
(This happens mostly with moderate size indexes, where the first read of the header
is enough to give you the data you need, but happens to be >50% of the whole file.)
We were doing it as soon as you entered 'iter_entries', but often you may already have enough
info to return results. And for small mostly local ops, we don't need to buffer all.
(This happens mostly with moderate size indexes, where the first read of the header
is enough to give you the data you need, but happens to be >50% of the whole file.)
- files added:
- COPYING.txt
- bzrlib/benchmarks/tree_creator
- bzrlib/help_topics
- bzrlib/help_topics/en
- bzrlib/repofmt
- bzrlib/smart
- bzrlib/tests/commands
- bzrlib/tests/intertree_implementations
- bzrlib/tests/inventory_implementations
- bzrlib/tests/per_lock
- bzrlib/tests/per_repository_reference
- bzrlib/tests/tree_implementations
- bzrlib/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 removed:
- NEWS.developers
- bzrlib/bundle/old
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/util/urlgrabber
- files renamed:
- bzrlib/patiencediff.py => bzrlib/_patiencediff_py.py
- bzrlib/bundle/read_bundle.py => bzrlib/bundle/bundle_data.py
- bzrlib/bundle/serializer/v07.py => bzrlib/bundle/serializer/v08.py
- bzrlib/graph.py => bzrlib/deprecated_graph.py
- doc/configuration.txt => bzrlib/help_topics/en/configuration.txt
- bzrlib/tests/HTTPTestUtil.py => bzrlib/tests/http_utils.py
- bzrlib/tests/test_revprops.py => bzrlib/tests/repository_implementations/test_revision.py
- bzrlib/tests/test_command.py => bzrlib/tests/test_commands.py
- bzrlib/tests/test_graph.py => bzrlib/tests/test_deprecated_graph.py
- bzrlib/tests/test_revisionnamespaces.py => bzrlib/tests/test_revisionspec.py
- bzrlib/transport/ftp.py => bzrlib/transport/ftp/__init__.py
- bzrlib/win32console.py => bzrlib/win32utils.py
- bzrlib/xml5.py => bzrlib/xml8.py
- HACKING => doc/developers/HACKING.txt
- doc/tutorial.txt => doc/en/tutorials/tutorial.txt
- doc/plugins.txt => doc/en/user-guide/plugins.txt
- doc/setting_up_email.txt => doc/en/user-guide/setting_up_email.txt
- doc/specifying_revisions.txt => doc/en/user-guide/specifying_revisions.txt
- doc/using_aliases.txt => doc/en/user-guide/using_aliases.txt
- files modified:
- .bzrignore
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
7
#
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
16
17
from copy import deepcopy
18
17
from cStringIO import StringIO
19
from unittest import TestSuite
20
21
import bzrlib.bzrdir as bzrdir
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
import re
22
import time
23
24
from bzrlib import (
25
bzrdir,
26
check,
27
debug,
28
errors,
29
generate_ids,
30
gpg,
31
graph,
32
lazy_regex,
33
lockable_files,
34
lockdir,
35
lru_cache,
36
osutils,
37
registry,
38
remote,
39
revision as _mod_revision,
40
symbol_versioning,
41
transactions,
42
tsort,
43
ui,
44
)
45
from bzrlib.bundle import serializer
46
from bzrlib.revisiontree import RevisionTree
47
from bzrlib.store.versioned import VersionedFileStore
48
from bzrlib.store.text import TextStore
49
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
51
""")
52
22
53
from bzrlib.decorators import needs_read_lock, needs_write_lock
23
import bzrlib.errors as errors
24
from bzrlib.errors import InvalidRevisionId
25
import bzrlib.gpg as gpg
26
from bzrlib.graph import Graph
27
54
from bzrlib.inter import InterObject
28
from bzrlib.knit import KnitVersionedFile, KnitPlainFactory
29
from bzrlib.lockable_files import LockableFiles, TransportLock
30
from bzrlib.lockdir import LockDir
31
from bzrlib.osutils import safe_unicode
32
from bzrlib.revision import NULL_REVISION
33
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
34
from bzrlib.store.text import TextStore
35
from bzrlib.symbol_versioning import *
36
from bzrlib.trace import mutter, note
37
from bzrlib.tree import RevisionTree
38
from bzrlib.tsort import topo_sort
39
from bzrlib.testament import Testament
40
from bzrlib.tree import EmptyTree
41
import bzrlib.ui
42
from bzrlib.weave import WeaveFile
43
import bzrlib.xml5
44
55
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
56
from bzrlib.symbol_versioning import (
57
deprecated_method,
58
one_one,
59
one_two,
60
one_three,
61
one_six,
62
)
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
64
65
66
# Old formats display a warning, but only once
67
_deprecation_warning_done = False
68
69
70
class CommitBuilder(object):
71
"""Provides an interface to build up a commit.
72
73
This allows describing a tree to be committed without needing to
74
know the internals of the format of the repository.
75
"""
76
77
# all clients should supply tree roots.
78
record_root_entry = True
79
# the default CommitBuilder does not manage trees whose root is versioned.
80
_versioned_root = False
81
82
def __init__(self, repository, parents, config, timestamp=None,
83
timezone=None, committer=None, revprops=None,
84
revision_id=None):
85
"""Initiate a CommitBuilder.
86
87
:param repository: Repository to commit to.
88
:param parents: Revision ids of the parents of the new revision.
89
:param config: Configuration to use.
90
:param timestamp: Optional timestamp recorded for commit.
91
:param timezone: Optional timezone for timestamp.
92
:param committer: Optional committer to set for commit.
93
:param revprops: Optional dictionary of revision properties.
94
:param revision_id: Optional revision id.
95
"""
96
self._config = config
97
98
if committer is None:
99
self._committer = self._config.username()
100
else:
101
self._committer = committer
102
103
self.new_inventory = Inventory(None)
104
self._new_revision_id = revision_id
105
self.parents = parents
106
self.repository = repository
107
108
self._revprops = {}
109
if revprops is not None:
110
self._revprops.update(revprops)
111
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
116
117
if timezone is None:
118
self._timezone = osutils.local_time_offset()
119
else:
120
self._timezone = int(timezone)
121
122
self._generate_revision_if_needed()
123
self.__heads = graph.HeadsCache(repository.get_graph()).heads
124
125
def commit(self, message):
126
"""Make the actual commit.
127
128
:return: The revision id of the recorded revision.
129
"""
130
rev = _mod_revision.Revision(
131
timestamp=self._timestamp,
132
timezone=self._timezone,
133
committer=self._committer,
134
message=message,
135
inventory_sha1=self.inv_sha1,
136
revision_id=self._new_revision_id,
137
properties=self._revprops)
138
rev.parent_ids = self.parents
139
self.repository.add_revision(self._new_revision_id, rev,
140
self.new_inventory, self._config)
141
self.repository.commit_write_group()
142
return self._new_revision_id
143
144
def abort(self):
145
"""Abort the commit that is being built.
146
"""
147
self.repository.abort_write_group()
148
149
def revision_tree(self):
150
"""Return the tree that was just committed.
151
152
After calling commit() this can be called to get a RevisionTree
153
representing the newly committed tree. This is preferred to
154
calling Repository.revision_tree() because that may require
155
deserializing the inventory, while we already have a copy in
156
memory.
157
"""
158
return RevisionTree(self.repository, self.new_inventory,
159
self._new_revision_id)
160
161
def finish_inventory(self):
162
"""Tell the builder that the inventory is finished."""
163
if self.new_inventory.root is None:
164
raise AssertionError('Root entry should be supplied to'
165
' record_entry_contents, as of bzr 0.10.')
166
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
167
self.new_inventory.revision_id = self._new_revision_id
168
self.inv_sha1 = self.repository.add_inventory(
169
self._new_revision_id,
170
self.new_inventory,
171
self.parents
172
)
173
174
def _gen_revision_id(self):
175
"""Return new revision-id."""
176
return generate_ids.gen_revision_id(self._config.username(),
177
self._timestamp)
178
179
def _generate_revision_if_needed(self):
180
"""Create a revision id if None was supplied.
181
182
If the repository can not support user-specified revision ids
183
they should override this function and raise CannotSetRevisionId
184
if _new_revision_id is not None.
185
186
:raises: CannotSetRevisionId
187
"""
188
if self._new_revision_id is None:
189
self._new_revision_id = self._gen_revision_id()
190
self.random_revid = True
191
else:
192
self.random_revid = False
193
194
def _heads(self, file_id, revision_ids):
195
"""Calculate the graph heads for revision_ids in the graph of file_id.
196
197
This can use either a per-file graph or a global revision graph as we
198
have an identity relationship between the two graphs.
199
"""
200
return self.__heads(revision_ids)
201
202
def _check_root(self, ie, parent_invs, tree):
203
"""Helper for record_entry_contents.
204
205
:param ie: An entry being added.
206
:param parent_invs: The inventories of the parent revisions of the
207
commit.
208
:param tree: The tree that is being committed.
209
"""
210
# In this revision format, root entries have no knit or weave When
211
# serializing out to disk and back in root.revision is always
212
# _new_revision_id
213
ie.revision = self._new_revision_id
214
215
def _get_delta(self, ie, basis_inv, path):
216
"""Get a delta against the basis inventory for ie."""
217
if ie.file_id not in basis_inv:
218
# add
219
return (None, path, ie.file_id, ie)
220
elif ie != basis_inv[ie.file_id]:
221
# common but altered
222
# TODO: avoid tis id2path call.
223
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
224
else:
225
# common, unaltered
226
return None
227
228
def record_entry_contents(self, ie, parent_invs, path, tree,
229
content_summary):
230
"""Record the content of ie from tree into the commit if needed.
231
232
Side effect: sets ie.revision when unchanged
233
234
:param ie: An inventory entry present in the commit.
235
:param parent_invs: The inventories of the parent revisions of the
236
commit.
237
:param path: The path the entry is at in the tree.
238
:param tree: The tree which contains this entry and should be used to
239
obtain content.
240
:param content_summary: Summary data from the tree about the paths
241
content - stat, length, exec, sha/link target. This is only
242
accessed when the entry has a revision of None - that is when it is
243
a candidate to commit.
244
:return: A tuple (change_delta, version_recorded). change_delta is
245
an inventory_delta change for this entry against the basis tree of
246
the commit, or None if no change occured against the basis tree.
247
version_recorded is True if a new version of the entry has been
248
recorded. For instance, committing a merge where a file was only
249
changed on the other side will return (delta, False).
250
"""
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
257
else:
258
# ie is carried over from a prior commit
259
kind = ie.kind
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
268
269
# TODO: slow, take it out of the inner loop.
270
try:
271
basis_inv = parent_invs[0]
272
except IndexError:
273
basis_inv = Inventory(root_id=None)
274
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
286
ie.file_id, ie)
287
else:
288
# add
289
delta = (None, path, ie.file_id, ie)
290
return delta, False
291
else:
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
294
# appropriate. If its not being considered for committing then
295
# it and all its parents to the root must be unaltered so
296
# no-change against the basis.
297
if ie.revision == self._new_revision_id:
298
raise AssertionError("Impossible situation, a skipped "
299
"inventory entry (%r) claims to be modified in this "
300
"commit (%r).", (ie, self._new_revision_id))
301
return None, False
302
# XXX: Friction: parent_candidates should return a list not a dict
303
# so that we don't have to walk the inventories again.
304
parent_candiate_entries = ie.parent_candidates(parent_invs)
305
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
306
heads = []
307
for inv in parent_invs:
308
if ie.file_id in inv:
309
old_rev = inv[ie.file_id].revision
310
if old_rev in head_set:
311
heads.append(inv[ie.file_id].revision)
312
head_set.remove(inv[ie.file_id].revision)
313
314
store = False
315
# now we check to see if we need to write a new record to the
316
# file-graph.
317
# We write a new entry unless there is one head to the ancestors, and
318
# the kind-derived content is unchanged.
319
320
# Cheapest check first: no ancestors, or more the one head in the
321
# ancestors, we write a new node.
322
if len(heads) != 1:
323
store = True
324
if not store:
325
# There is a single head, look it up for comparison
326
parent_entry = parent_candiate_entries[heads[0]]
327
# if the non-content specific data has changed, we'll be writing a
328
# node:
329
if (parent_entry.parent_id != ie.parent_id or
330
parent_entry.name != ie.name):
331
store = True
332
# now we need to do content specific checks:
333
if not store:
334
# if the kind changed the content obviously has
335
if kind != parent_entry.kind:
336
store = True
337
if kind == 'file':
338
if content_summary[2] is None:
339
raise ValueError("Files must not have executable = None")
340
if not store:
341
if (# if the file length changed we have to store:
342
parent_entry.text_size != content_summary[1] or
343
# if the exec bit has changed we have to store:
344
parent_entry.executable != content_summary[2]):
345
store = True
346
elif parent_entry.text_sha1 == content_summary[3]:
347
# all meta and content is unchanged (using a hash cache
348
# hit to check the sha)
349
ie.revision = parent_entry.revision
350
ie.text_size = parent_entry.text_size
351
ie.text_sha1 = parent_entry.text_sha1
352
ie.executable = parent_entry.executable
353
return self._get_delta(ie, basis_inv, path), False
354
else:
355
# Either there is only a hash change(no hash cache entry,
356
# or same size content change), or there is no change on
357
# this file at all.
358
# Provide the parent's hash to the store layer, so that the
359
# content is unchanged we will not store a new node.
360
nostore_sha = parent_entry.text_sha1
361
if store:
362
# We want to record a new node regardless of the presence or
363
# absence of a content change in the file.
364
nostore_sha = None
365
ie.executable = content_summary[2]
366
lines = tree.get_file(ie.file_id, path).readlines()
367
try:
368
ie.text_sha1, ie.text_size = self._add_text_to_weave(
369
ie.file_id, lines, heads, nostore_sha)
370
except errors.ExistingContent:
371
# Turns out that the file content was unchanged, and we were
372
# only going to store a new node if it was changed. Carry over
373
# the entry.
374
ie.revision = parent_entry.revision
375
ie.text_size = parent_entry.text_size
376
ie.text_sha1 = parent_entry.text_sha1
377
ie.executable = parent_entry.executable
378
return self._get_delta(ie, basis_inv, path), False
379
elif kind == 'directory':
380
if not store:
381
# all data is meta here, nothing specific to directory, so
382
# carry over:
383
ie.revision = parent_entry.revision
384
return self._get_delta(ie, basis_inv, path), False
385
lines = []
386
self._add_text_to_weave(ie.file_id, lines, heads, None)
387
elif kind == 'symlink':
388
current_link_target = content_summary[3]
389
if not store:
390
# symlink target is not generic metadata, check if it has
391
# changed.
392
if current_link_target != parent_entry.symlink_target:
393
store = True
394
if not store:
395
# unchanged, carry over.
396
ie.revision = parent_entry.revision
397
ie.symlink_target = parent_entry.symlink_target
398
return self._get_delta(ie, basis_inv, path), False
399
ie.symlink_target = current_link_target
400
lines = []
401
self._add_text_to_weave(ie.file_id, lines, heads, None)
402
elif kind == 'tree-reference':
403
if not store:
404
if content_summary[3] != parent_entry.reference_revision:
405
store = True
406
if not store:
407
# unchanged, carry over.
408
ie.reference_revision = parent_entry.reference_revision
409
ie.revision = parent_entry.revision
410
return self._get_delta(ie, basis_inv, path), False
411
ie.reference_revision = content_summary[3]
412
lines = []
413
self._add_text_to_weave(ie.file_id, lines, heads, None)
414
else:
415
raise NotImplementedError('unknown kind')
416
ie.revision = self._new_revision_id
417
return self._get_delta(ie, basis_inv, path), True
418
419
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
420
# Note: as we read the content directly from the tree, we know its not
421
# been turned into unicode or badly split - but a broken tree
422
# implementation could give us bad output from readlines() so this is
423
# not a guarantee of safety. What would be better is always checking
424
# the content during test suite execution. RBC 20070912
425
parent_keys = tuple((file_id, parent) for parent in parents)
426
return self.repository.texts.add_lines(
427
(file_id, self._new_revision_id), parent_keys, new_lines,
428
nostore_sha=nostore_sha, random_id=self.random_revid,
429
check_content=False)[0:2]
430
431
432
class RootCommitBuilder(CommitBuilder):
433
"""This commitbuilder actually records the root id"""
434
435
# the root entry gets versioned properly by this builder.
436
_versioned_root = True
437
438
def _check_root(self, ie, parent_invs, tree):
439
"""Helper for record_entry_contents.
440
441
:param ie: An entry being added.
442
:param parent_invs: The inventories of the parent revisions of the
443
commit.
444
:param tree: The tree that is being committed.
445
"""
446
447
448
######################################################################
449
# Repositories
45
450
46
451
class Repository(object):
47
452
"""Repository holding history for one or more branches.
50
455
revisions and file history. It's normally accessed only by the Branch,
51
456
which views a particular line of development through that history.
52
457
53
The Repository builds on top of Stores and a Transport, which respectively
54
describe the disk data format and the way of accessing the (possibly
458
The Repository builds on top of some byte storage facilies (the revisions,
459
signatures, inventories and texts attributes) and a Transport, which
460
respectively provide byte storage and a means to access the (possibly
55
461
remote) disk.
462
463
The byte storage facilities are addressed via tuples, which we refer to
464
as 'keys' throughout the code base. Revision_keys, inventory_keys and
465
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
466
(file_id, revision_id). We use this interface because it allows low
467
friction with the underlying code that implements disk indices, network
468
encoding and other parts of bzrlib.
469
470
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
471
the serialised revisions for the repository. This can be used to obtain
472
revision graph information or to access raw serialised revisions.
473
The result of trying to insert data into the repository via this store
474
is undefined: it should be considered read-only except for implementors
475
of repositories.
476
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
477
the serialised signatures for the repository. This can be used to
478
obtain access to raw serialised signatures. The result of trying to
479
insert data into the repository via this store is undefined: it should
480
be considered read-only except for implementors of repositories.
481
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
482
the serialised inventories for the repository. This can be used to
483
obtain unserialised inventories. The result of trying to insert data
484
into the repository via this store is undefined: it should be
485
considered read-only except for implementors of repositories.
486
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
487
texts of files and directories for the repository. This can be used to
488
obtain file texts or file graphs. Note that Repository.iter_file_bytes
489
is usually a better interface for accessing file texts.
490
The result of trying to insert data into the repository via this store
491
is undefined: it should be considered read-only except for implementors
492
of repositories.
493
:ivar _transport: Transport for file access to repository, typically
494
pointing to .bzr/repository.
56
495
"""
57
496
58
@needs_write_lock
59
def add_inventory(self, revid, inv, parents):
60
"""Add the inventory inv to the repository as revid.
61
62
:param parents: The revision ids of the parents that revid
497
# What class to use for a CommitBuilder. Often its simpler to change this
498
# in a Repository class subclass rather than to override
499
# get_commit_builder.
500
_commit_builder_class = CommitBuilder
501
# The search regex used by xml based repositories to determine what things
502
# where changed in a single commit.
503
_file_ids_altered_regex = lazy_regex.lazy_compile(
504
r'file_id="(?P<file_id>[^"]+)"'
505
r'.* revision="(?P<revision_id>[^"]+)"'
506
)
507
508
def abort_write_group(self):
509
"""Commit the contents accrued within the current write group.
510
511
:seealso: start_write_group.
512
"""
513
if self._write_group is not self.get_transaction():
514
# has an unlock or relock occured ?
515
raise errors.BzrError('mismatched lock context and write group.')
516
self._abort_write_group()
517
self._write_group = None
518
519
def _abort_write_group(self):
520
"""Template method for per-repository write group cleanup.
521
522
This is called during abort before the write group is considered to be
523
finished and should cleanup any internal state accrued during the write
524
group. There is no requirement that data handed to the repository be
525
*not* made available - this is not a rollback - but neither should any
526
attempt be made to ensure that data added is fully commited. Abort is
527
invoked when an error has occured so futher disk or network operations
528
may not be possible or may error and if possible should not be
529
attempted.
530
"""
531
532
def add_fallback_repository(self, repository):
533
"""Add a repository to use for looking up data not held locally.
534
535
:param repository: A repository.
536
"""
537
if not self._format.supports_external_lookups:
538
raise errors.UnstackableRepositoryFormat(self._format, self.base)
539
self._check_fallback_repository(repository)
540
self._fallback_repositories.append(repository)
541
self.texts.add_fallback_versioned_files(repository.texts)
542
self.inventories.add_fallback_versioned_files(repository.inventories)
543
self.revisions.add_fallback_versioned_files(repository.revisions)
544
self.signatures.add_fallback_versioned_files(repository.signatures)
545
546
def _check_fallback_repository(self, repository):
547
"""Check that this repository can fallback to repository safely.
548
549
Raise an error if not.
550
551
:param repository: A repository to fallback to.
552
"""
553
return InterRepository._assert_same_model(self, repository)
554
555
def add_inventory(self, revision_id, inv, parents):
556
"""Add the inventory inv to the repository as revision_id.
557
558
:param parents: The revision ids of the parents that revision_id
63
559
is known to have and are in the repository already.
64
560
65
returns the sha1 of the serialized inventory.
561
:returns: The validator(which is a sha1 digest, though what is sha'd is
562
repository format specific) of the serialized inventory.
66
563
"""
67
inv_text = bzrlib.xml5.serializer_v5.write_inventory_to_string(inv)
68
inv_sha1 = bzrlib.osutils.sha_string(inv_text)
69
inv_vf = self.control_weaves.get_weave('inventory',
70
self.get_transaction())
71
inv_vf.add_lines(revid, parents, bzrlib.osutils.split_lines(inv_text))
72
return inv_sha1
73
74
@needs_write_lock
75
def add_revision(self, rev_id, rev, inv=None, config=None):
76
"""Add rev to the revision store as rev_id.
77
78
:param rev_id: the revision id to use.
564
if not self.is_in_write_group():
565
raise AssertionError("%r not in write group" % (self,))
566
_mod_revision.check_not_reserved_id(revision_id)
567
if not (inv.revision_id is None or inv.revision_id == revision_id):
568
raise AssertionError(
569
"Mismatch between inventory revision"
570
" id and insertion revid (%r, %r)"
571
% (inv.revision_id, revision_id))
572
if inv.root is None:
573
raise AssertionError()
574
inv_lines = self._serialise_inventory_to_lines(inv)
575
return self._inventory_add_lines(revision_id, parents,
576
inv_lines, check_content=False)
577
578
def _inventory_add_lines(self, revision_id, parents, lines,
579
check_content=True):
580
"""Store lines in inv_vf and return the sha1 of the inventory."""
581
parents = [(parent,) for parent in parents]
582
return self.inventories.add_lines((revision_id,), parents, lines,
583
check_content=check_content)[0]
584
585
def add_revision(self, revision_id, rev, inv=None, config=None):
586
"""Add rev to the revision store as revision_id.
587
588
:param revision_id: the revision id to use.
79
589
:param rev: The revision object.
80
590
:param inv: The inventory for the revision. if None, it will be looked
81
591
up in the inventory storer
83
593
If supplied its signature_needed method will be used
84
594
to determine if a signature should be made.
85
595
"""
596
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
597
# rev.parent_ids?
598
_mod_revision.check_not_reserved_id(revision_id)
86
599
if config is not None and config.signature_needed():
87
600
if inv is None:
88
inv = self.get_inventory(rev_id)
601
inv = self.get_inventory(revision_id)
89
602
plaintext = Testament(rev, inv).as_short_text()
90
603
self.store_revision_signature(
91
gpg.GPGStrategy(config), plaintext, rev_id)
92
if not rev_id in self.get_inventory_weave():
604
gpg.GPGStrategy(config), plaintext, revision_id)
605
# check inventory present
606
if not self.inventories.get_parent_map([(revision_id,)]):
93
607
if inv is None:
94
raise errors.WeaveRevisionNotPresent(rev_id,
95
self.get_inventory_weave())
608
raise errors.WeaveRevisionNotPresent(revision_id,
609
self.inventories)
96
610
else:
97
611
# yes, this is not suitable for adding with ghosts.
98
self.add_inventory(rev_id, inv, rev.parent_ids)
99
self._revision_store.add_revision(rev, self.get_transaction())
100
101
@needs_read_lock
102
def _all_possible_ids(self):
103
"""Return all the possible revisions that we could find."""
104
return self.get_inventory_weave().versions()
105
106
@needs_read_lock
612
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
613
rev.parent_ids)
614
else:
615
key = (revision_id,)
616
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
617
self._add_revision(rev)
618
619
def _add_revision(self, revision):
620
text = self._serializer.write_revision_to_string(revision)
621
key = (revision.revision_id,)
622
parents = tuple((parent,) for parent in revision.parent_ids)
623
self.revisions.add_lines(key, parents, osutils.split_lines(text))
624
107
625
def all_revision_ids(self):
108
626
"""Returns a list of all the revision ids in the repository.
109
627
628
This is conceptually deprecated because code should generally work on
629
the graph reachable from a particular revision, and ignore any other
630
revisions that might be present. There is no direct replacement
631
method.
632
"""
633
if 'evil' in debug.debug_flags:
634
mutter_callsite(2, "all_revision_ids is linear with history.")
635
return self._all_revision_ids()
636
637
def _all_revision_ids(self):
638
"""Returns a list of all the revision ids in the repository.
639
110
640
These are in as much topological order as the underlying store can
111
present: for weaves ghosts may lead to a lack of correctness until
112
the reweave updates the parents list.
641
present.
113
642
"""
114
if self._revision_store.text_store.listable():
115
return self._revision_store.all_revision_ids(self.get_transaction())
116
result = self._all_possible_ids()
117
return self._eliminate_revisions_not_present(result)
643
raise NotImplementedError(self._all_revision_ids)
118
644
119
645
def break_lock(self):
120
646
"""Break a lock if one is present from another instance.
131
657
Returns a set of the present revisions.
132
658
"""
133
659
result = []
134
for id in revision_ids:
135
if self.has_revision(id):
136
result.append(id)
137
return result
660
graph = self.get_graph()
661
parent_map = graph.get_parent_map(revision_ids)
662
# The old API returned a list, should this actually be a set?
663
return parent_map.keys()
138
664
139
665
@staticmethod
140
666
def create(a_bzrdir):
141
667
"""Construct the current default format repository in a_bzrdir."""
142
668
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
143
669
144
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
670
def __init__(self, _format, a_bzrdir, control_files):
145
671
"""instantiate a Repository.
146
672
147
673
:param _format: The format of the repository on disk.
156
682
# the following are part of the public API for Repository:
157
683
self.bzrdir = a_bzrdir
158
684
self.control_files = control_files
159
self._revision_store = _revision_store
160
self.text_store = text_store
161
# backwards compatability
162
self.weave_store = text_store
685
self._transport = control_files._transport
686
self.base = self._transport.base
687
# for tests
688
self._reconcile_does_inventory_gc = True
689
self._reconcile_fixes_text_parents = False
690
self._reconcile_backsup_inventory = True
163
691
# not right yet - should be more semantically clear ?
164
692
#
165
self.control_store = control_store
166
self.control_weaves = control_store
167
693
# TODO: make sure to construct the right store classes, etc, depending
168
694
# on whether escaping is required.
695
self._warn_if_deprecated()
696
self._write_group = None
697
# Additional places to query for data.
698
self._fallback_repositories = []
699
# What order should fetch operations request streams in?
700
# The default is unordered as that is the cheapest for an origin to
701
# provide.
702
self._fetch_order = 'unordered'
703
# Does this repository use deltas that can be fetched as-deltas ?
704
# (E.g. knits, where the knit deltas can be transplanted intact.
705
# We default to False, which will ensure that enough data to get
706
# a full text out of any fetch stream will be grabbed.
707
self._fetch_uses_deltas = False
708
# Should fetch trigger a reconcile after the fetch? Only needed for
709
# some repository formats that can suffer internal inconsistencies.
710
self._fetch_reconcile = False
169
711
170
712
def __repr__(self):
171
return '%s(%r)' % (self.__class__.__name__,
172
self.bzrdir.transport.base)
713
return '%s(%r)' % (self.__class__.__name__,
714
self.base)
715
716
def has_same_location(self, other):
717
"""Returns a boolean indicating if this repository is at the same
718
location as another repository.
719
720
This might return False even when two repository objects are accessing
721
the same physical repository via different URLs.
722
"""
723
if self.__class__ is not other.__class__:
724
return False
725
return (self._transport.base == other._transport.base)
726
727
def is_in_write_group(self):
728
"""Return True if there is an open write group.
729
730
:seealso: start_write_group.
731
"""
732
return self._write_group is not None
173
733
174
734
def is_locked(self):
175
735
return self.control_files.is_locked()
176
736
177
def lock_write(self):
178
self.control_files.lock_write()
737
def is_write_locked(self):
738
"""Return True if this object is write locked."""
739
return self.is_locked() and self.control_files._lock_mode == 'w'
740
741
def lock_write(self, token=None):
742
"""Lock this repository for writing.
743
744
This causes caching within the repository obejct to start accumlating
745
data during reads, and allows a 'write_group' to be obtained. Write
746
groups must be used for actual data insertion.
747
748
:param token: if this is already locked, then lock_write will fail
749
unless the token matches the existing lock.
750
:returns: a token if this instance supports tokens, otherwise None.
751
:raises TokenLockingNotSupported: when a token is given but this
752
instance doesn't support using token locks.
753
:raises MismatchedToken: if the specified token doesn't match the token
754
of the existing lock.
755
:seealso: start_write_group.
756
757
A token should be passed in if you know that you have locked the object
758
some other way, and need to synchronise this object's state with that
759
fact.
760
761
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
762
"""
763
result = self.control_files.lock_write(token=token)
764
for repo in self._fallback_repositories:
765
# Writes don't affect fallback repos
766
repo.lock_read()
767
self._refresh_data()
768
return result
179
769
180
770
def lock_read(self):
181
771
self.control_files.lock_read()
772
for repo in self._fallback_repositories:
773
repo.lock_read()
774
self._refresh_data()
182
775
183
776
def get_physical_lock_status(self):
184
777
return self.control_files.get_physical_lock_status()
185
778
186
@needs_read_lock
187
def missing_revision_ids(self, other, revision_id=None):
188
"""Return the revision ids that other has that this does not.
189
190
These are returned in topological order.
191
192
revision_id: only return revision ids included by revision_id.
193
"""
194
return InterRepository.get(other, self).missing_revision_ids(revision_id)
779
def leave_lock_in_place(self):
780
"""Tell this repository not to release the physical lock when this
781
object is unlocked.
782
783
If lock_write doesn't return a token, then this method is not supported.
784
"""
785
self.control_files.leave_in_place()
786
787
def dont_leave_lock_in_place(self):
788
"""Tell this repository to release the physical lock when this
789
object is unlocked, even if it didn't originally acquire it.
790
791
If lock_write doesn't return a token, then this method is not supported.
792
"""
793
self.control_files.dont_leave_in_place()
794
795
@needs_read_lock
796
def gather_stats(self, revid=None, committers=None):
797
"""Gather statistics from a revision id.
798
799
:param revid: The revision id to gather statistics from, if None, then
800
no revision specific statistics are gathered.
801
:param committers: Optional parameter controlling whether to grab
802
a count of committers from the revision specific statistics.
803
:return: A dictionary of statistics. Currently this contains:
804
committers: The number of committers if requested.
805
firstrev: A tuple with timestamp, timezone for the penultimate left
806
most ancestor of revid, if revid is not the NULL_REVISION.
807
latestrev: A tuple with timestamp, timezone for revid, if revid is
808
not the NULL_REVISION.
809
revisions: The total revision count in the repository.
810
size: An estimate disk size of the repository in bytes.
811
"""
812
result = {}
813
if revid and committers:
814
result['committers'] = 0
815
if revid and revid != _mod_revision.NULL_REVISION:
816
if committers:
817
all_committers = set()
818
revisions = self.get_ancestry(revid)
819
# pop the leading None
820
revisions.pop(0)
821
first_revision = None
822
if not committers:
823
# ignore the revisions in the middle - just grab first and last
824
revisions = revisions[0], revisions[-1]
825
for revision in self.get_revisions(revisions):
826
if not first_revision:
827
first_revision = revision
828
if committers:
829
all_committers.add(revision.committer)
830
last_revision = revision
831
if committers:
832
result['committers'] = len(all_committers)
833
result['firstrev'] = (first_revision.timestamp,
834
first_revision.timezone)
835
result['latestrev'] = (last_revision.timestamp,
836
last_revision.timezone)
837
838
# now gather global repository information
839
# XXX: This is available for many repos regardless of listability.
840
if self.bzrdir.root_transport.listable():
841
# XXX: do we want to __define len__() ?
842
# Maybe the versionedfiles object should provide a different
843
# method to get the number of keys.
844
result['revisions'] = len(self.revisions.keys())
845
# result['size'] = t
846
return result
847
848
def find_branches(self, using=False):
849
"""Find branches underneath this repository.
850
851
This will include branches inside other branches.
852
853
:param using: If True, list only branches using this repository.
854
"""
855
if using and not self.is_shared():
856
try:
857
return [self.bzrdir.open_branch()]
858
except errors.NotBranchError:
859
return []
860
class Evaluator(object):
861
862
def __init__(self):
863
self.first_call = True
864
865
def __call__(self, bzrdir):
866
# On the first call, the parameter is always the bzrdir
867
# containing the current repo.
868
if not self.first_call:
869
try:
870
repository = bzrdir.open_repository()
871
except errors.NoRepositoryPresent:
872
pass
873
else:
874
return False, (None, repository)
875
self.first_call = False
876
try:
877
value = (bzrdir.open_branch(), None)
878
except errors.NotBranchError:
879
value = (None, None)
880
return True, value
881
882
branches = []
883
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
884
self.bzrdir.root_transport, evaluate=Evaluator()):
885
if branch is not None:
886
branches.append(branch)
887
if not using and repository is not None:
888
branches.extend(repository.find_branches())
889
return branches
890
891
@needs_read_lock
892
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
893
"""Return the revision ids that other has that this does not.
894
895
These are returned in topological order.
896
897
revision_id: only return revision ids included by revision_id.
898
"""
899
return InterRepository.get(other, self).search_missing_revision_ids(
900
revision_id, find_ghosts)
901
902
@deprecated_method(one_two)
903
@needs_read_lock
904
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
905
"""Return the revision ids that other has that this does not.
906
907
These are returned in topological order.
908
909
revision_id: only return revision ids included by revision_id.
910
"""
911
keys = self.search_missing_revision_ids(
912
other, revision_id, find_ghosts).get_keys()
913
other.lock_read()
914
try:
915
parents = other.get_graph().get_parent_map(keys)
916
finally:
917
other.unlock()
918
return tsort.topo_sort(parents)
195
919
196
920
@staticmethod
197
921
def open(base):
200
924
For instance, if the repository is at URL/.bzr/repository,
201
925
Repository.open(URL) -> a Repository instance.
202
926
"""
203
control = bzrlib.bzrdir.BzrDir.open(base)
927
control = bzrdir.BzrDir.open(base)
204
928
return control.open_repository()
205
929
206
def copy_content_into(self, destination, revision_id=None, basis=None):
930
def copy_content_into(self, destination, revision_id=None):
207
931
"""Make a complete copy of the content in self into destination.
208
932
209
933
This is a destructive operation! Do not use it on existing
210
934
repositories.
211
935
"""
212
return InterRepository.get(self, destination).copy_content(revision_id, basis)
213
214
def fetch(self, source, revision_id=None, pb=None):
936
return InterRepository.get(self, destination).copy_content(revision_id)
937
938
def commit_write_group(self):
939
"""Commit the contents accrued within the current write group.
940
941
:seealso: start_write_group.
942
"""
943
if self._write_group is not self.get_transaction():
944
# has an unlock or relock occured ?
945
raise errors.BzrError('mismatched lock context %r and '
946
'write group %r.' %
947
(self.get_transaction(), self._write_group))
948
self._commit_write_group()
949
self._write_group = None
950
951
def _commit_write_group(self):
952
"""Template method for per-repository write group cleanup.
953
954
This is called before the write group is considered to be
955
finished and should ensure that all data handed to the repository
956
for writing during the write group is safely committed (to the
957
extent possible considering file system caching etc).
958
"""
959
960
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
215
961
"""Fetch the content required to construct revision_id from source.
216
962
217
963
If revision_id is None all content is copied.
218
"""
219
return InterRepository.get(source, self).fetch(revision_id=revision_id,
220
pb=pb)
964
:param find_ghosts: Find and copy revisions in the source that are
965
ghosts in the target (and not reachable directly by walking out to
966
the first-present revision in target from revision_id).
967
"""
968
# fast path same-url fetch operations
969
if self.has_same_location(source):
970
# check that last_revision is in 'from' and then return a
971
# no-operation.
972
if (revision_id is not None and
973
not _mod_revision.is_null(revision_id)):
974
self.get_revision(revision_id)
975
return 0, []
976
# if there is no specific appropriate InterRepository, this will get
977
# the InterRepository base class, which raises an
978
# IncompatibleRepositories when asked to fetch.
979
inter = InterRepository.get(source, self)
980
return inter.fetch(revision_id=revision_id, pb=pb,
981
find_ghosts=find_ghosts)
982
983
def create_bundle(self, target, base, fileobj, format=None):
984
return serializer.write_bundle(self, target, base, fileobj, format)
985
986
def get_commit_builder(self, branch, parents, config, timestamp=None,
987
timezone=None, committer=None, revprops=None,
988
revision_id=None):
989
"""Obtain a CommitBuilder for this repository.
990
991
:param branch: Branch to commit to.
992
:param parents: Revision ids of the parents of the new revision.
993
:param config: Configuration to use.
994
:param timestamp: Optional timestamp recorded for commit.
995
:param timezone: Optional timezone for timestamp.
996
:param committer: Optional committer to set for commit.
997
:param revprops: Optional dictionary of revision properties.
998
:param revision_id: Optional revision id.
999
"""
1000
result = self._commit_builder_class(self, parents, config,
1001
timestamp, timezone, committer, revprops, revision_id)
1002
self.start_write_group()
1003
return result
221
1004
222
1005
def unlock(self):
1006
if (self.control_files._lock_count == 1 and
1007
self.control_files._lock_mode == 'w'):
1008
if self._write_group is not None:
1009
self.abort_write_group()
1010
self.control_files.unlock()
1011
raise errors.BzrError(
1012
'Must end write groups before releasing write locks.')
223
1013
self.control_files.unlock()
1014
for repo in self._fallback_repositories:
1015
repo.unlock()
224
1016
225
1017
@needs_read_lock
226
def clone(self, a_bzrdir, revision_id=None, basis=None):
1018
def clone(self, a_bzrdir, revision_id=None):
227
1019
"""Clone this repository into a_bzrdir using the current format.
228
1020
229
1021
Currently no check is made that the format of this repository and
230
1022
the bzrdir format are compatible. FIXME RBC 20060201.
231
"""
1023
1024
:return: The newly created destination repository.
1025
"""
1026
# TODO: deprecate after 0.16; cloning this with all its settings is
1027
# probably not very useful -- mbp 20070423
1028
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1029
self.copy_content_into(dest_repo, revision_id)
1030
return dest_repo
1031
1032
def start_write_group(self):
1033
"""Start a write group in the repository.
1034
1035
Write groups are used by repositories which do not have a 1:1 mapping
1036
between file ids and backend store to manage the insertion of data from
1037
both fetch and commit operations.
1038
1039
A write lock is required around the start_write_group/commit_write_group
1040
for the support of lock-requiring repository formats.
1041
1042
One can only insert data into a repository inside a write group.
1043
1044
:return: None.
1045
"""
1046
if not self.is_write_locked():
1047
raise errors.NotWriteLocked(self)
1048
if self._write_group:
1049
raise errors.BzrError('already in a write group')
1050
self._start_write_group()
1051
# so we can detect unlock/relock - the write group is now entered.
1052
self._write_group = self.get_transaction()
1053
1054
def _start_write_group(self):
1055
"""Template method for per-repository write group startup.
1056
1057
This is called before the write group is considered to be
1058
entered.
1059
"""
1060
1061
@needs_read_lock
1062
def sprout(self, to_bzrdir, revision_id=None):
1063
"""Create a descendent repository for new development.
1064
1065
Unlike clone, this does not copy the settings of the repository.
1066
"""
1067
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1068
dest_repo.fetch(self, revision_id=revision_id)
1069
return dest_repo
1070
1071
def _create_sprouting_repo(self, a_bzrdir, shared):
232
1072
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
233
1073
# use target default format.
234
result = a_bzrdir.create_repository()
235
# FIXME RBC 20060209 split out the repository type to avoid this check ?
236
elif isinstance(a_bzrdir._format,
237
(bzrlib.bzrdir.BzrDirFormat4,
238
bzrlib.bzrdir.BzrDirFormat5,
239
bzrlib.bzrdir.BzrDirFormat6)):
240
result = a_bzrdir.open_repository()
1074
dest_repo = a_bzrdir.create_repository()
241
1075
else:
242
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
243
self.copy_content_into(result, revision_id, basis)
244
return result
1076
# Most control formats need the repository to be specifically
1077
# created, but on some old all-in-one formats it's not needed
1078
try:
1079
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1080
except errors.UninitializableFormat:
1081
dest_repo = a_bzrdir.open_repository()
1082
return dest_repo
245
1083
246
1084
@needs_read_lock
247
1085
def has_revision(self, revision_id):
248
1086
"""True if this repository has a copy of the revision."""
249
return self._revision_store.has_revision_id(revision_id,
250
self.get_transaction())
1087
return revision_id in self.has_revisions((revision_id,))
1088
1089
@needs_read_lock
1090
def has_revisions(self, revision_ids):
1091
"""Probe to find out the presence of multiple revisions.
1092
1093
:param revision_ids: An iterable of revision_ids.
1094
:return: A set of the revision_ids that were present.
1095
"""
1096
parent_map = self.revisions.get_parent_map(
1097
[(rev_id,) for rev_id in revision_ids])
1098
result = set()
1099
if _mod_revision.NULL_REVISION in revision_ids:
1100
result.add(_mod_revision.NULL_REVISION)
1101
result.update([key[0] for key in parent_map])
1102
return result
1103
1104
@needs_read_lock
1105
def get_revision(self, revision_id):
1106
"""Return the Revision object for a named revision."""
1107
return self.get_revisions([revision_id])[0]
251
1108
252
1109
@needs_read_lock
253
1110
def get_revision_reconcile(self, revision_id):
258
1115
be used by reconcile, or reconcile-alike commands that are correcting
259
1116
or testing the revision graph.
260
1117
"""
261
if not revision_id or not isinstance(revision_id, basestring):
262
raise InvalidRevisionId(revision_id=revision_id, branch=self)
263
return self._revision_store.get_revision(revision_id,
264
self.get_transaction())
1118
return self._get_revisions([revision_id])[0]
1119
1120
@needs_read_lock
1121
def get_revisions(self, revision_ids):
1122
"""Get many revisions at once."""
1123
return self._get_revisions(revision_ids)
1124
1125
@needs_read_lock
1126
def _get_revisions(self, revision_ids):
1127
"""Core work logic to get many revisions without sanity checks."""
1128
for rev_id in revision_ids:
1129
if not rev_id or not isinstance(rev_id, basestring):
1130
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1131
keys = [(key,) for key in revision_ids]
1132
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1133
revs = {}
1134
for record in stream:
1135
if record.storage_kind == 'absent':
1136
raise errors.NoSuchRevision(self, record.key[0])
1137
text = record.get_bytes_as('fulltext')
1138
rev = self._serializer.read_revision_from_string(text)
1139
revs[record.key[0]] = rev
1140
return [revs[revid] for revid in revision_ids]
265
1141
266
1142
@needs_read_lock
267
1143
def get_revision_xml(self, revision_id):
268
rev = self.get_revision(revision_id)
1144
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1145
# would have already do it.
1146
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1147
rev = self.get_revision(revision_id)
269
1148
rev_tmp = StringIO()
270
1149
# the current serializer..
271
self._revision_store._serializer.write_revision(rev, rev_tmp)
1150
self._serializer.write_revision(rev, rev_tmp)
272
1151
rev_tmp.seek(0)
273
1152
return rev_tmp.getvalue()
274
1153
1154
def get_deltas_for_revisions(self, revisions):
1155
"""Produce a generator of revision deltas.
1156
1157
Note that the input is a sequence of REVISIONS, not revision_ids.
1158
Trees will be held in memory until the generator exits.
1159
Each delta is relative to the revision's lefthand predecessor.
1160
"""
1161
required_trees = set()
1162
for revision in revisions:
1163
required_trees.add(revision.revision_id)
1164
required_trees.update(revision.parent_ids[:1])
1165
trees = dict((t.get_revision_id(), t) for
1166
t in self.revision_trees(required_trees))
1167
for revision in revisions:
1168
if not revision.parent_ids:
1169
old_tree = self.revision_tree(None)
1170
else:
1171
old_tree = trees[revision.parent_ids[0]]
1172
yield trees[revision.revision_id].changes_from(old_tree)
1173
275
1174
@needs_read_lock
276
def get_revision(self, revision_id):
277
"""Return the Revision object for a named revision"""
278
r = self.get_revision_reconcile(revision_id)
279
# weave corruption can lead to absent revision markers that should be
280
# present.
281
# the following test is reasonably cheap (it needs a single weave read)
282
# and the weave is cached in read transactions. In write transactions
283
# it is not cached but typically we only read a small number of
284
# revisions. For knits when they are introduced we will probably want
285
# to ensure that caching write transactions are in use.
286
inv = self.get_inventory_weave()
287
self._check_revision_parents(r, inv)
288
return r
1175
def get_revision_delta(self, revision_id):
1176
"""Return the delta for one revision.
289
1177
290
def _check_revision_parents(self, revision, inventory):
291
"""Private to Repository and Fetch.
292
293
This checks the parentage of revision in an inventory weave for
294
consistency and is only applicable to inventory-weave-for-ancestry
295
using repository formats & fetchers.
1178
The delta is relative to the left-hand predecessor of the
1179
revision.
296
1180
"""
297
weave_parents = inventory.get_parents(revision.revision_id)
298
weave_names = inventory.versions()
299
for parent_id in revision.parent_ids:
300
if parent_id in weave_names:
301
# this parent must not be a ghost.
302
if not parent_id in weave_parents:
303
# but it is a ghost
304
raise errors.CorruptRepository(self)
1181
r = self.get_revision(revision_id)
1182
return list(self.get_deltas_for_revisions([r]))[0]
305
1183
306
1184
@needs_write_lock
307
1185
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
308
1186
signature = gpg_strategy.sign(plaintext)
309
self._revision_store.add_revision_signature_text(revision_id,
310
signature,
311
self.get_transaction())
312
313
def fileids_altered_by_revision_ids(self, revision_ids):
314
"""Find the file ids and versions affected by revisions.
315
316
:param revisions: an iterable containing revision ids.
317
:return: a dictionary mapping altered file-ids to an iterable of
1187
self.add_signature_text(revision_id, signature)
1188
1189
@needs_write_lock
1190
def add_signature_text(self, revision_id, signature):
1191
self.signatures.add_lines((revision_id,), (),
1192
osutils.split_lines(signature))
1193
1194
def find_text_key_references(self):
1195
"""Find the text key references within the repository.
1196
1197
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
318
1198
revision_ids. Each altered file-ids has the exact revision_ids that
319
1199
altered it listed explicitly.
320
"""
321
assert isinstance(self._format, (RepositoryFormat5,
322
RepositoryFormat6,
323
RepositoryFormat7,
324
RepositoryFormatKnit1)), \
325
"fileid_involved only supported for branches which store inventory as unnested xml"
326
selected_revision_ids = set(revision_ids)
327
w = self.get_inventory_weave()
1200
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1201
to whether they were referred to by the inventory of the
1202
revision_id that they contain. The inventory texts from all present
1203
revision ids are assessed to generate this report.
1204
"""
1205
revision_keys = self.revisions.keys()
1206
w = self.inventories
1207
pb = ui.ui_factory.nested_progress_bar()
1208
try:
1209
return self._find_text_key_references_from_xml_inventory_lines(
1210
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1211
finally:
1212
pb.finished()
1213
1214
def _find_text_key_references_from_xml_inventory_lines(self,
1215
line_iterator):
1216
"""Core routine for extracting references to texts from inventories.
1217
1218
This performs the translation of xml lines to revision ids.
1219
1220
:param line_iterator: An iterator of lines, origin_version_id
1221
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1222
to whether they were referred to by the inventory of the
1223
revision_id that they contain. Note that if that revision_id was
1224
not part of the line_iterator's output then False will be given -
1225
even though it may actually refer to that key.
1226
"""
1227
if not self._serializer.support_altered_by_hack:
1228
raise AssertionError(
1229
"_find_text_key_references_from_xml_inventory_lines only "
1230
"supported for branches which store inventory as unnested xml"
1231
", not on %r" % self)
328
1232
result = {}
329
1233
330
1234
# this code needs to read every new line in every inventory for the
331
1235
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
332
# not pesent in one of those inventories is unnecessary but not
1236
# not present in one of those inventories is unnecessary but not
333
1237
# harmful because we are filtering by the revision id marker in the
334
1238
# inventory lines : we only select file ids altered in one of those
335
# revisions. We dont need to see all lines in the inventory because
1239
# revisions. We don't need to see all lines in the inventory because
336
1240
# only those added in an inventory in rev X can contain a revision=X
337
1241
# line.
338
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
339
start = line.find('file_id="')+9
340
if start < 9: continue
341
end = line.find('"', start)
342
assert end>= 0
343
file_id = _unescape_xml(line[start:end])
344
345
start = line.find('revision="')+10
346
if start < 10: continue
347
end = line.find('"', start)
348
assert end>= 0
349
revision_id = _unescape_xml(line[start:end])
350
if revision_id in selected_revision_ids:
351
result.setdefault(file_id, set()).add(revision_id)
352
return result
353
354
@needs_read_lock
355
def get_inventory_weave(self):
356
return self.control_weaves.get_weave('inventory',
357
self.get_transaction())
1242
unescape_revid_cache = {}
1243
unescape_fileid_cache = {}
1244
1245
# jam 20061218 In a big fetch, this handles hundreds of thousands
1246
# of lines, so it has had a lot of inlining and optimizing done.
1247
# Sorry that it is a little bit messy.
1248
# Move several functions to be local variables, since this is a long
1249
# running loop.
1250
search = self._file_ids_altered_regex.search
1251
unescape = _unescape_xml
1252
setdefault = result.setdefault
1253
for line, line_key in line_iterator:
1254
match = search(line)
1255
if match is None:
1256
continue
1257
# One call to match.group() returning multiple items is quite a
1258
# bit faster than 2 calls to match.group() each returning 1
1259
file_id, revision_id = match.group('file_id', 'revision_id')
1260
1261
# Inlining the cache lookups helps a lot when you make 170,000
1262
# lines and 350k ids, versus 8.4 unique ids.
1263
# Using a cache helps in 2 ways:
1264
# 1) Avoids unnecessary decoding calls
1265
# 2) Re-uses cached strings, which helps in future set and
1266
# equality checks.
1267
# (2) is enough that removing encoding entirely along with
1268
# the cache (so we are using plain strings) results in no
1269
# performance improvement.
1270
try:
1271
revision_id = unescape_revid_cache[revision_id]
1272
except KeyError:
1273
unescaped = unescape(revision_id)
1274
unescape_revid_cache[revision_id] = unescaped
1275
revision_id = unescaped
1276
1277
# Note that unconditionally unescaping means that we deserialise
1278
# every fileid, which for general 'pull' is not great, but we don't
1279
# really want to have some many fulltexts that this matters anyway.
1280
# RBC 20071114.
1281
try:
1282
file_id = unescape_fileid_cache[file_id]
1283
except KeyError:
1284
unescaped = unescape(file_id)
1285
unescape_fileid_cache[file_id] = unescaped
1286
file_id = unescaped
1287
1288
key = (file_id, revision_id)
1289
setdefault(key, False)
1290
if revision_id == line_key[-1]:
1291
result[key] = True
1292
return result
1293
1294
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1295
revision_ids):
1296
"""Helper routine for fileids_altered_by_revision_ids.
1297
1298
This performs the translation of xml lines to revision ids.
1299
1300
:param line_iterator: An iterator of lines, origin_version_id
1301
:param revision_ids: The revision ids to filter for. This should be a
1302
set or other type which supports efficient __contains__ lookups, as
1303
the revision id from each parsed line will be looked up in the
1304
revision_ids filter.
1305
:return: a dictionary mapping altered file-ids to an iterable of
1306
revision_ids. Each altered file-ids has the exact revision_ids that
1307
altered it listed explicitly.
1308
"""
1309
result = {}
1310
setdefault = result.setdefault
1311
for key in \
1312
self._find_text_key_references_from_xml_inventory_lines(
1313
line_iterator).iterkeys():
1314
# once data is all ensured-consistent; then this is
1315
# if revision_id == version_id
1316
if key[-1:] in revision_ids:
1317
setdefault(key[0], set()).add(key[-1])
1318
return result
1319
1320
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1321
"""Find the file ids and versions affected by revisions.
1322
1323
:param revisions: an iterable containing revision ids.
1324
:param _inv_weave: The inventory weave from this repository or None.
1325
If None, the inventory weave will be opened automatically.
1326
:return: a dictionary mapping altered file-ids to an iterable of
1327
revision_ids. Each altered file-ids has the exact revision_ids that
1328
altered it listed explicitly.
1329
"""
1330
selected_keys = set((revid,) for revid in revision_ids)
1331
w = _inv_weave or self.inventories
1332
pb = ui.ui_factory.nested_progress_bar()
1333
try:
1334
return self._find_file_ids_from_xml_inventory_lines(
1335
w.iter_lines_added_or_present_in_keys(
1336
selected_keys, pb=pb),
1337
selected_keys)
1338
finally:
1339
pb.finished()
1340
1341
def iter_files_bytes(self, desired_files):
1342
"""Iterate through file versions.
1343
1344
Files will not necessarily be returned in the order they occur in
1345
desired_files. No specific order is guaranteed.
1346
1347
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1348
value supplied by the caller as part of desired_files. It should
1349
uniquely identify the file version in the caller's context. (Examples:
1350
an index number or a TreeTransform trans_id.)
1351
1352
bytes_iterator is an iterable of bytestrings for the file. The
1353
kind of iterable and length of the bytestrings are unspecified, but for
1354
this implementation, it is a list of bytes produced by
1355
VersionedFile.get_record_stream().
1356
1357
:param desired_files: a list of (file_id, revision_id, identifier)
1358
triples
1359
"""
1360
transaction = self.get_transaction()
1361
text_keys = {}
1362
for file_id, revision_id, callable_data in desired_files:
1363
text_keys[(file_id, revision_id)] = callable_data
1364
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1365
if record.storage_kind == 'absent':
1366
raise errors.RevisionNotPresent(record.key, self)
1367
yield text_keys[record.key], record.get_bytes_as('fulltext')
1368
1369
def _generate_text_key_index(self, text_key_references=None,
1370
ancestors=None):
1371
"""Generate a new text key index for the repository.
1372
1373
This is an expensive function that will take considerable time to run.
1374
1375
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1376
list of parents, also text keys. When a given key has no parents,
1377
the parents list will be [NULL_REVISION].
1378
"""
1379
# All revisions, to find inventory parents.
1380
if ancestors is None:
1381
graph = self.get_graph()
1382
ancestors = graph.get_parent_map(self.all_revision_ids())
1383
if text_key_references is None:
1384
text_key_references = self.find_text_key_references()
1385
pb = ui.ui_factory.nested_progress_bar()
1386
try:
1387
return self._do_generate_text_key_index(ancestors,
1388
text_key_references, pb)
1389
finally:
1390
pb.finished()
1391
1392
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1393
"""Helper for _generate_text_key_index to avoid deep nesting."""
1394
revision_order = tsort.topo_sort(ancestors)
1395
invalid_keys = set()
1396
revision_keys = {}
1397
for revision_id in revision_order:
1398
revision_keys[revision_id] = set()
1399
text_count = len(text_key_references)
1400
# a cache of the text keys to allow reuse; costs a dict of all the
1401
# keys, but saves a 2-tuple for every child of a given key.
1402
text_key_cache = {}
1403
for text_key, valid in text_key_references.iteritems():
1404
if not valid:
1405
invalid_keys.add(text_key)
1406
else:
1407
revision_keys[text_key[1]].add(text_key)
1408
text_key_cache[text_key] = text_key
1409
del text_key_references
1410
text_index = {}
1411
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1412
NULL_REVISION = _mod_revision.NULL_REVISION
1413
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1414
# too small for large or very branchy trees. However, for 55K path
1415
# trees, it would be easy to use too much memory trivially. Ideally we
1416
# could gauge this by looking at available real memory etc, but this is
1417
# always a tricky proposition.
1418
inventory_cache = lru_cache.LRUCache(10)
1419
batch_size = 10 # should be ~150MB on a 55K path tree
1420
batch_count = len(revision_order) / batch_size + 1
1421
processed_texts = 0
1422
pb.update("Calculating text parents.", processed_texts, text_count)
1423
for offset in xrange(batch_count):
1424
to_query = revision_order[offset * batch_size:(offset + 1) *
1425
batch_size]
1426
if not to_query:
1427
break
1428
for rev_tree in self.revision_trees(to_query):
1429
revision_id = rev_tree.get_revision_id()
1430
parent_ids = ancestors[revision_id]
1431
for text_key in revision_keys[revision_id]:
1432
pb.update("Calculating text parents.", processed_texts)
1433
processed_texts += 1
1434
candidate_parents = []
1435
for parent_id in parent_ids:
1436
parent_text_key = (text_key[0], parent_id)
1437
try:
1438
check_parent = parent_text_key not in \
1439
revision_keys[parent_id]
1440
except KeyError:
1441
# the parent parent_id is a ghost:
1442
check_parent = False
1443
# truncate the derived graph against this ghost.
1444
parent_text_key = None
1445
if check_parent:
1446
# look at the parent commit details inventories to
1447
# determine possible candidates in the per file graph.
1448
# TODO: cache here.
1449
try:
1450
inv = inventory_cache[parent_id]
1451
except KeyError:
1452
inv = self.revision_tree(parent_id).inventory
1453
inventory_cache[parent_id] = inv
1454
parent_entry = inv._byid.get(text_key[0], None)
1455
if parent_entry is not None:
1456
parent_text_key = (
1457
text_key[0], parent_entry.revision)
1458
else:
1459
parent_text_key = None
1460
if parent_text_key is not None:
1461
candidate_parents.append(
1462
text_key_cache[parent_text_key])
1463
parent_heads = text_graph.heads(candidate_parents)
1464
new_parents = list(parent_heads)
1465
new_parents.sort(key=lambda x:candidate_parents.index(x))
1466
if new_parents == []:
1467
new_parents = [NULL_REVISION]
1468
text_index[text_key] = new_parents
1469
1470
for text_key in invalid_keys:
1471
text_index[text_key] = [NULL_REVISION]
1472
return text_index
1473
1474
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1475
"""Get an iterable listing the keys of all the data introduced by a set
1476
of revision IDs.
1477
1478
The keys will be ordered so that the corresponding items can be safely
1479
fetched and inserted in that order.
1480
1481
:returns: An iterable producing tuples of (knit-kind, file-id,
1482
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1483
'revisions'. file-id is None unless knit-kind is 'file'.
1484
"""
1485
# XXX: it's a bit weird to control the inventory weave caching in this
1486
# generator. Ideally the caching would be done in fetch.py I think. Or
1487
# maybe this generator should explicitly have the contract that it
1488
# should not be iterated until the previously yielded item has been
1489
# processed?
1490
inv_w = self.inventories
1491
1492
# file ids that changed
1493
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1494
count = 0
1495
num_file_ids = len(file_ids)
1496
for file_id, altered_versions in file_ids.iteritems():
1497
if _files_pb is not None:
1498
_files_pb.update("fetch texts", count, num_file_ids)
1499
count += 1
1500
yield ("file", file_id, altered_versions)
1501
# We're done with the files_pb. Note that it finished by the caller,
1502
# just as it was created by the caller.
1503
del _files_pb
1504
1505
# inventory
1506
yield ("inventory", None, revision_ids)
1507
1508
# signatures
1509
revisions_with_signatures = set()
1510
for rev_id in revision_ids:
1511
try:
1512
self.get_signature_text(rev_id)
1513
except errors.NoSuchRevision:
1514
# not signed.
1515
pass
1516
else:
1517
revisions_with_signatures.add(rev_id)
1518
yield ("signatures", None, revisions_with_signatures)
1519
1520
# revisions
1521
yield ("revisions", None, revision_ids)
358
1522
359
1523
@needs_read_lock
360
1524
def get_inventory(self, revision_id):
361
"""Get Inventory object by hash."""
362
xml = self.get_inventory_xml(revision_id)
363
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
1525
"""Get Inventory object by revision id."""
1526
return self.iter_inventories([revision_id]).next()
1527
1528
def iter_inventories(self, revision_ids):
1529
"""Get many inventories by revision_ids.
1530
1531
This will buffer some or all of the texts used in constructing the
1532
inventories in memory, but will only parse a single inventory at a
1533
time.
1534
1535
:return: An iterator of inventories.
1536
"""
1537
if ((None in revision_ids)
1538
or (_mod_revision.NULL_REVISION in revision_ids)):
1539
raise ValueError('cannot get null revision inventory')
1540
return self._iter_inventories(revision_ids)
1541
1542
def _iter_inventories(self, revision_ids):
1543
"""single-document based inventory iteration."""
1544
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1545
yield self.deserialise_inventory(revision_id, text)
1546
1547
def _iter_inventory_xmls(self, revision_ids):
1548
keys = [(revision_id,) for revision_id in revision_ids]
1549
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1550
texts = {}
1551
for record in stream:
1552
if record.storage_kind != 'absent':
1553
texts[record.key] = record.get_bytes_as('fulltext')
1554
else:
1555
raise errors.NoSuchRevision(self, record.key)
1556
for key in keys:
1557
yield texts[key], key[-1]
1558
1559
def deserialise_inventory(self, revision_id, xml):
1560
"""Transform the xml into an inventory object.
1561
1562
:param revision_id: The expected revision id of the inventory.
1563
:param xml: A serialised inventory.
1564
"""
1565
result = self._serializer.read_inventory_from_string(xml, revision_id)
1566
if result.revision_id != revision_id:
1567
raise AssertionError('revision id mismatch %s != %s' % (
1568
result.revision_id, revision_id))
1569
return result
1570
1571
def serialise_inventory(self, inv):
1572
return self._serializer.write_inventory_to_string(inv)
1573
1574
def _serialise_inventory_to_lines(self, inv):
1575
return self._serializer.write_inventory_to_lines(inv)
1576
1577
def get_serializer_format(self):
1578
return self._serializer.format_num
364
1579
365
1580
@needs_read_lock
366
1581
def get_inventory_xml(self, revision_id):
367
1582
"""Get inventory XML as a file object."""
1583
texts = self._iter_inventory_xmls([revision_id])
368
1584
try:
369
assert isinstance(revision_id, basestring), type(revision_id)
370
iw = self.get_inventory_weave()
371
return iw.get_text(revision_id)
372
except IndexError:
373
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
1585
text, revision_id = texts.next()
1586
except StopIteration:
1587
raise errors.HistoryMissing(self, 'inventory', revision_id)
1588
return text
374
1589
375
1590
@needs_read_lock
376
1591
def get_inventory_sha1(self, revision_id):
378
1593
"""
379
1594
return self.get_revision(revision_id).inventory_sha1
380
1595
381
@needs_read_lock
382
def get_revision_graph(self, revision_id=None):
383
"""Return a dictionary containing the revision graph.
384
385
:return: a dictionary of revision_id->revision_parents_list.
386
"""
387
weave = self.get_inventory_weave()
388
all_revisions = self._eliminate_revisions_not_present(weave.versions())
389
entire_graph = dict([(node, weave.get_parents(node)) for
390
node in all_revisions])
391
if revision_id is None:
392
return entire_graph
393
elif revision_id not in entire_graph:
394
raise errors.NoSuchRevision(self, revision_id)
395
else:
396
# add what can be reached from revision_id
397
result = {}
398
pending = set([revision_id])
399
while len(pending) > 0:
400
node = pending.pop()
401
result[node] = entire_graph[node]
402
for revision_id in result[node]:
403
if revision_id not in result:
404
pending.add(revision_id)
405
return result
406
407
@needs_read_lock
408
def get_revision_graph_with_ghosts(self, revision_ids=None):
409
"""Return a graph of the revisions with ghosts marked as applicable.
410
411
:param revision_ids: an iterable of revisions to graph or None for all.
412
:return: a Graph object with the graph reachable from revision_ids.
413
"""
414
result = Graph()
415
if not revision_ids:
416
pending = set(self.all_revision_ids())
417
required = set([])
418
else:
419
pending = set(revision_ids)
420
required = set(revision_ids)
421
done = set([])
422
while len(pending):
423
revision_id = pending.pop()
424
try:
425
rev = self.get_revision(revision_id)
426
except errors.NoSuchRevision:
427
if revision_id in required:
428
raise
429
# a ghost
430
result.add_ghost(revision_id)
431
continue
432
for parent_id in rev.parent_ids:
433
# is this queued or done ?
434
if (parent_id not in pending and
435
parent_id not in done):
436
# no, queue it.
437
pending.add(parent_id)
438
result.add_node(revision_id, rev.parent_ids)
439
done.add(revision_id)
440
return result
1596
def iter_reverse_revision_history(self, revision_id):
1597
"""Iterate backwards through revision ids in the lefthand history
1598
1599
:param revision_id: The revision id to start with. All its lefthand
1600
ancestors will be traversed.
1601
"""
1602
graph = self.get_graph()
1603
next_id = revision_id
1604
while True:
1605
if next_id in (None, _mod_revision.NULL_REVISION):
1606
return
1607
yield next_id
1608
# Note: The following line may raise KeyError in the event of
1609
# truncated history. We decided not to have a try:except:raise
1610
# RevisionNotPresent here until we see a use for it, because of the
1611
# cost in an inner loop that is by its very nature O(history).
1612
# Robert Collins 20080326
1613
parents = graph.get_parent_map([next_id])[next_id]
1614
if len(parents) == 0:
1615
return
1616
else:
1617
next_id = parents[0]
441
1618
442
1619
@needs_read_lock
443
1620
def get_revision_inventory(self, revision_id):
455
1632
else:
456
1633
return self.get_inventory(revision_id)
457
1634
458
@needs_read_lock
459
1635
def is_shared(self):
460
1636
"""Return True if this repository is flagged as a shared repository."""
461
1637
raise NotImplementedError(self.is_shared)
467
1643
reconciler = RepoReconciler(self, thorough=thorough)
468
1644
reconciler.reconcile()
469
1645
return reconciler
470
1646
1647
def _refresh_data(self):
1648
"""Helper called from lock_* to ensure coherency with disk.
1649
1650
The default implementation does nothing; it is however possible
1651
for repositories to maintain loaded indices across multiple locks
1652
by checking inside their implementation of this method to see
1653
whether their indices are still valid. This depends of course on
1654
the disk format being validatable in this manner.
1655
"""
1656
471
1657
@needs_read_lock
472
1658
def revision_tree(self, revision_id):
473
1659
"""Return Tree for a revision on this branch.
474
1660
475
`revision_id` may be None for the null revision, in which case
476
an `EmptyTree` is returned."""
1661
`revision_id` may be None for the empty tree revision.
1662
"""
477
1663
# TODO: refactor this to use an existing revision object
478
1664
# so we don't need to read it in twice.
479
if revision_id is None or revision_id == NULL_REVISION:
480
return EmptyTree()
1665
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1666
return RevisionTree(self, Inventory(root_id=None),
1667
_mod_revision.NULL_REVISION)
481
1668
else:
482
1669
inv = self.get_revision_inventory(revision_id)
483
1670
return RevisionTree(self, inv, revision_id)
484
1671
1672
def revision_trees(self, revision_ids):
1673
"""Return Tree for a revision on this branch.
1674
1675
`revision_id` may not be None or 'null:'"""
1676
inventories = self.iter_inventories(revision_ids)
1677
for inv in inventories:
1678
yield RevisionTree(self, inv, inv.revision_id)
1679
485
1680
@needs_read_lock
486
def get_ancestry(self, revision_id):
1681
def get_ancestry(self, revision_id, topo_sorted=True):
487
1682
"""Return a list of revision-ids integrated by a revision.
1683
1684
The first element of the list is always None, indicating the origin
1685
revision. This might change when we have history horizons, or
1686
perhaps we should have a new API.
488
1687
489
1688
This is topologically sorted.
490
1689
"""
491
if revision_id is None:
1690
if _mod_revision.is_null(revision_id):
492
1691
return [None]
493
1692
if not self.has_revision(revision_id):
494
1693
raise errors.NoSuchRevision(self, revision_id)
495
w = self.get_inventory_weave()
496
candidates = w.get_ancestry(revision_id)
497
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1694
graph = self.get_graph()
1695
keys = set()
1696
search = graph._make_breadth_first_searcher([revision_id])
1697
while True:
1698
try:
1699
found, ghosts = search.next_with_ghosts()
1700
except StopIteration:
1701
break
1702
keys.update(found)
1703
if _mod_revision.NULL_REVISION in keys:
1704
keys.remove(_mod_revision.NULL_REVISION)
1705
if topo_sorted:
1706
parent_map = graph.get_parent_map(keys)
1707
keys = tsort.topo_sort(parent_map)
1708
return [None] + list(keys)
1709
1710
def pack(self):
1711
"""Compress the data within the repository.
1712
1713
This operation only makes sense for some repository types. For other
1714
types it should be a no-op that just returns.
1715
1716
This stub method does not require a lock, but subclasses should use
1717
@needs_write_lock as this is a long running call its reasonable to
1718
implicitly lock for the user.
1719
"""
498
1720
499
1721
@needs_read_lock
1722
@deprecated_method(one_six)
500
1723
def print_file(self, file, revision_id):
501
1724
"""Print `file` to stdout.
502
1725
517
1740
def get_transaction(self):
518
1741
return self.control_files.get_transaction()
519
1742
520
def revision_parents(self, revid):
521
return self.get_inventory_weave().parent_names(revid)
1743
@deprecated_method(one_one)
1744
def get_parents(self, revision_ids):
1745
"""See StackedParentsProvider.get_parents"""
1746
parent_map = self.get_parent_map(revision_ids)
1747
return [parent_map.get(r, None) for r in revision_ids]
1748
1749
def get_parent_map(self, revision_ids):
1750
"""See graph._StackedParentsProvider.get_parent_map"""
1751
# revisions index works in keys; this just works in revisions
1752
# therefore wrap and unwrap
1753
query_keys = []
1754
result = {}
1755
for revision_id in revision_ids:
1756
if revision_id == _mod_revision.NULL_REVISION:
1757
result[revision_id] = ()
1758
elif revision_id is None:
1759
raise ValueError('get_parent_map(None) is not valid')
1760
else:
1761
query_keys.append((revision_id ,))
1762
for ((revision_id,), parent_keys) in \
1763
self.revisions.get_parent_map(query_keys).iteritems():
1764
if parent_keys:
1765
result[revision_id] = tuple(parent_revid
1766
for (parent_revid,) in parent_keys)
1767
else:
1768
result[revision_id] = (_mod_revision.NULL_REVISION,)
1769
return result
1770
1771
def _make_parents_provider(self):
1772
return self
1773
1774
def get_graph(self, other_repository=None):
1775
"""Return the graph walker for this repository format"""
1776
parents_provider = self._make_parents_provider()
1777
if (other_repository is not None and
1778
not self.has_same_location(other_repository)):
1779
parents_provider = graph._StackedParentsProvider(
1780
[parents_provider, other_repository._make_parents_provider()])
1781
return graph.Graph(parents_provider)
1782
1783
def _get_versioned_file_checker(self):
1784
"""Return an object suitable for checking versioned files."""
1785
return _VersionedFileChecker(self)
1786
1787
def revision_ids_to_search_result(self, result_set):
1788
"""Convert a set of revision ids to a graph SearchResult."""
1789
result_parents = set()
1790
for parents in self.get_graph().get_parent_map(
1791
result_set).itervalues():
1792
result_parents.update(parents)
1793
included_keys = result_set.intersection(result_parents)
1794
start_keys = result_set.difference(included_keys)
1795
exclude_keys = result_parents.difference(result_set)
1796
result = graph.SearchResult(start_keys, exclude_keys,
1797
len(result_set), result_set)
1798
return result
522
1799
523
1800
@needs_write_lock
524
1801
def set_make_working_trees(self, new_value):
544
1821
@needs_read_lock
545
1822
def has_signature_for_revision_id(self, revision_id):
546
1823
"""Query for a revision signature for revision_id in the repository."""
547
return self._revision_store.has_signature(revision_id,
548
self.get_transaction())
1824
if not self.has_revision(revision_id):
1825
raise errors.NoSuchRevision(self, revision_id)
1826
sig_present = (1 == len(
1827
self.signatures.get_parent_map([(revision_id,)])))
1828
return sig_present
549
1829
550
1830
@needs_read_lock
551
1831
def get_signature_text(self, revision_id):
552
1832
"""Return the text for a signature."""
553
return self._revision_store.get_signature_text(revision_id,
554
self.get_transaction())
555
556
557
class AllInOneRepository(Repository):
558
"""Legacy support - the repository behaviour for all-in-one branches."""
559
560
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
561
# we reuse one control files instance.
562
dir_mode = a_bzrdir._control_files._dir_mode
563
file_mode = a_bzrdir._control_files._file_mode
564
565
def get_store(name, compressed=True, prefixed=False):
566
# FIXME: This approach of assuming stores are all entirely compressed
567
# or entirely uncompressed is tidy, but breaks upgrade from
568
# some existing branches where there's a mixture; we probably
569
# still want the option to look for both.
570
relpath = a_bzrdir._control_files._escape(name)
571
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
572
prefixed=prefixed, compressed=compressed,
573
dir_mode=dir_mode,
574
file_mode=file_mode)
575
#if self._transport.should_cache():
576
# cache_path = os.path.join(self.cache_root, name)
577
# os.mkdir(cache_path)
578
# store = bzrlib.store.CachedStore(store, cache_path)
579
return store
580
581
# not broken out yet because the controlweaves|inventory_store
582
# and text_store | weave_store bits are still different.
583
if isinstance(_format, RepositoryFormat4):
584
# cannot remove these - there is still no consistent api
585
# which allows access to this old info.
586
self.inventory_store = get_store('inventory-store')
587
text_store = get_store('text-store')
588
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
1833
stream = self.signatures.get_record_stream([(revision_id,)],
1834
'unordered', True)
1835
record = stream.next()
1836
if record.storage_kind == 'absent':
1837
raise errors.NoSuchRevision(self, revision_id)
1838
return record.get_bytes_as('fulltext')
589
1839
590
1840
@needs_read_lock
591
def is_shared(self):
592
"""AllInOne repositories cannot be shared."""
593
return False
594
595
@needs_write_lock
596
def set_make_working_trees(self, new_value):
597
"""Set the policy flag for making working trees when creating branches.
598
599
This only applies to branches that use this repository.
600
601
The default is 'True'.
602
:param new_value: True to restore the default, False to disable making
603
working trees.
1841
def check(self, revision_ids=None):
1842
"""Check consistency of all history of given revision_ids.
1843
1844
Different repository implementations should override _check().
1845
1846
:param revision_ids: A non-empty list of revision_ids whose ancestry
1847
will be checked. Typically the last revision_id of a branch.
604
1848
"""
605
raise NotImplementedError(self.set_make_working_trees)
1849
return self._check(revision_ids)
1850
1851
def _check(self, revision_ids):
1852
result = check.Check(self)
1853
result.check()
1854
return result
1855
1856
def _warn_if_deprecated(self):
1857
global _deprecation_warning_done
1858
if _deprecation_warning_done:
1859
return
1860
_deprecation_warning_done = True
1861
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1862
% (self._format, self.bzrdir.transport.base))
1863
1864
def supports_rich_root(self):
1865
return self._format.rich_root_data
1866
1867
def _check_ascii_revisionid(self, revision_id, method):
1868
"""Private helper for ascii-only repositories."""
1869
# weave repositories refuse to store revisionids that are non-ascii.
1870
if revision_id is not None:
1871
# weaves require ascii revision ids.
1872
if isinstance(revision_id, unicode):
1873
try:
1874
revision_id.encode('ascii')
1875
except UnicodeEncodeError:
1876
raise errors.NonAsciiRevisionId(method, self)
1877
else:
1878
try:
1879
revision_id.decode('ascii')
1880
except UnicodeDecodeError:
1881
raise errors.NonAsciiRevisionId(method, self)
606
1882
607
def make_working_trees(self):
608
"""Returns the policy for making working trees on new branches."""
609
return True
1883
def revision_graph_can_have_wrong_parents(self):
1884
"""Is it possible for this repository to have a revision graph with
1885
incorrect parents?
1886
1887
If True, then this repository must also implement
1888
_find_inconsistent_revision_parents so that check and reconcile can
1889
check for inconsistencies before proceeding with other checks that may
1890
depend on the revision index being consistent.
1891
"""
1892
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1893
1894
1895
# remove these delegates a while after bzr 0.15
1896
def __make_delegated(name, from_module):
1897
def _deprecated_repository_forwarder():
1898
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1899
% (name, from_module),
1900
DeprecationWarning,
1901
stacklevel=2)
1902
m = __import__(from_module, globals(), locals(), [name])
1903
try:
1904
return getattr(m, name)
1905
except AttributeError:
1906
raise AttributeError('module %s has no name %s'
1907
% (m, name))
1908
globals()[name] = _deprecated_repository_forwarder
1909
1910
for _name in [
1911
'AllInOneRepository',
1912
'WeaveMetaDirRepository',
1913
'PreSplitOutRepositoryFormat',
1914
'RepositoryFormat4',
1915
'RepositoryFormat5',
1916
'RepositoryFormat6',
1917
'RepositoryFormat7',
1918
]:
1919
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1920
1921
for _name in [
1922
'KnitRepository',
1923
'RepositoryFormatKnit',
1924
'RepositoryFormatKnit1',
1925
]:
1926
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
610
1927
611
1928
612
1929
def install_revision(repository, rev, revision_tree):
613
1930
"""Install all revision data into a repository."""
1931
install_revisions(repository, [(rev, revision_tree, None)])
1932
1933
1934
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1935
"""Install all revision data into a repository.
1936
1937
Accepts an iterable of revision, tree, signature tuples. The signature
1938
may be None.
1939
"""
1940
repository.start_write_group()
1941
try:
1942
for n, (revision, revision_tree, signature) in enumerate(iterable):
1943
_install_revision(repository, revision, revision_tree, signature)
1944
if pb is not None:
1945
pb.update('Transferring revisions', n + 1, num_revisions)
1946
except:
1947
repository.abort_write_group()
1948
raise
1949
else:
1950
repository.commit_write_group()
1951
1952
1953
def _install_revision(repository, rev, revision_tree, signature):
1954
"""Install all revision data into a repository."""
614
1955
present_parents = []
615
1956
parent_trees = {}
616
1957
for p_id in rev.parent_ids:
618
1959
present_parents.append(p_id)
619
1960
parent_trees[p_id] = repository.revision_tree(p_id)
620
1961
else:
621
parent_trees[p_id] = EmptyTree()
1962
parent_trees[p_id] = repository.revision_tree(None)
622
1963
623
1964
inv = revision_tree.inventory
624
1965
entries = inv.iter_entries()
1966
# backwards compatibility hack: skip the root id.
1967
if not repository.supports_rich_root():
1968
path, root = entries.next()
1969
if root.revision != rev.revision_id:
1970
raise errors.IncompatibleRevision(repr(repository))
1971
text_keys = {}
1972
for path, ie in entries:
1973
text_keys[(ie.file_id, ie.revision)] = ie
1974
text_parent_map = repository.texts.get_parent_map(text_keys)
1975
missing_texts = set(text_keys) - set(text_parent_map)
625
1976
# Add the texts that are not already present
626
for path, ie in inv.iter_entries():
627
w = repository.weave_store.get_weave_or_empty(ie.file_id,
628
repository.get_transaction())
629
if ie.revision not in w:
630
text_parents = []
631
for revision, tree in parent_trees.iteritems():
632
if ie.file_id not in tree:
633
continue
634
parent_id = tree.inventory[ie.file_id].revision
635
if parent_id in text_parents:
636
continue
637
text_parents.append(parent_id)
638
639
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
640
repository.get_transaction())
641
lines = revision_tree.get_file(ie.file_id).readlines()
642
vfile.add_lines(rev.revision_id, text_parents, lines)
1977
for text_key in missing_texts:
1978
ie = text_keys[text_key]
1979
text_parents = []
1980
# FIXME: TODO: The following loop overlaps/duplicates that done by
1981
# commit to determine parents. There is a latent/real bug here where
1982
# the parents inserted are not those commit would do - in particular
1983
# they are not filtered by heads(). RBC, AB
1984
for revision, tree in parent_trees.iteritems():
1985
if ie.file_id not in tree:
1986
continue
1987
parent_id = tree.inventory[ie.file_id].revision
1988
if parent_id in text_parents:
1989
continue
1990
text_parents.append((ie.file_id, parent_id))
1991
lines = revision_tree.get_file(ie.file_id).readlines()
1992
repository.texts.add_lines(text_key, text_parents, lines)
643
1993
try:
644
1994
# install the inventory
645
1995
repository.add_inventory(rev.revision_id, inv, present_parents)
646
1996
except errors.RevisionAlreadyPresent:
647
1997
pass
1998
if signature is not None:
1999
repository.add_signature_text(rev.revision_id, signature)
648
2000
repository.add_revision(rev.revision_id, rev, inv)
649
2001
650
2002
651
2003
class MetaDirRepository(Repository):
652
"""Repositories in the new meta-dir layout."""
653
654
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
655
super(MetaDirRepository, self).__init__(_format,
656
a_bzrdir,
657
control_files,
658
_revision_store,
659
control_store,
660
text_store)
661
662
dir_mode = self.control_files._dir_mode
663
file_mode = self.control_files._file_mode
664
665
@needs_read_lock
2004
"""Repositories in the new meta-dir layout.
2005
2006
:ivar _transport: Transport for access to repository control files,
2007
typically pointing to .bzr/repository.
2008
"""
2009
2010
def __init__(self, _format, a_bzrdir, control_files):
2011
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2012
self._transport = control_files._transport
2013
666
2014
def is_shared(self):
667
2015
"""Return True if this repository is flagged as a shared repository."""
668
return self.control_files._transport.has('shared-storage')
2016
return self._transport.has('shared-storage')
669
2017
670
2018
@needs_write_lock
671
2019
def set_make_working_trees(self, new_value):
679
2027
"""
680
2028
if new_value:
681
2029
try:
682
self.control_files._transport.delete('no-working-trees')
2030
self._transport.delete('no-working-trees')
683
2031
except errors.NoSuchFile:
684
2032
pass
685
2033
else:
686
self.control_files.put_utf8('no-working-trees', '')
2034
self._transport.put_bytes('no-working-trees', '',
2035
mode=self.bzrdir._get_file_mode())
687
2036
688
2037
def make_working_trees(self):
689
2038
"""Returns the policy for making working trees on new branches."""
690
return not self.control_files._transport.has('no-working-trees')
691
692
693
class KnitRepository(MetaDirRepository):
694
"""Knit format repository."""
695
696
@needs_read_lock
697
def all_revision_ids(self):
698
"""See Repository.all_revision_ids()."""
699
return self._revision_store.all_revision_ids(self.get_transaction())
700
701
def fileid_involved_between_revs(self, from_revid, to_revid):
702
"""Find file_id(s) which are involved in the changes between revisions.
703
704
This determines the set of revisions which are involved, and then
705
finds all file ids affected by those revisions.
706
"""
707
vf = self._get_revision_vf()
708
from_set = set(vf.get_ancestry(from_revid))
709
to_set = set(vf.get_ancestry(to_revid))
710
changed = to_set.difference(from_set)
711
return self._fileid_involved_by_set(changed)
712
713
def fileid_involved(self, last_revid=None):
714
"""Find all file_ids modified in the ancestry of last_revid.
715
716
:param last_revid: If None, last_revision() will be used.
717
"""
718
if not last_revid:
719
changed = set(self.all_revision_ids())
720
else:
721
changed = set(self.get_ancestry(last_revid))
722
if None in changed:
723
changed.remove(None)
724
return self._fileid_involved_by_set(changed)
725
726
@needs_read_lock
727
def get_ancestry(self, revision_id):
728
"""Return a list of revision-ids integrated by a revision.
729
730
This is topologically sorted.
731
"""
732
if revision_id is None:
733
return [None]
734
vf = self._get_revision_vf()
735
try:
736
return [None] + vf.get_ancestry(revision_id)
737
except errors.RevisionNotPresent:
738
raise errors.NoSuchRevision(self, revision_id)
739
740
@needs_read_lock
741
def get_revision(self, revision_id):
742
"""Return the Revision object for a named revision"""
743
return self.get_revision_reconcile(revision_id)
744
745
@needs_read_lock
746
def get_revision_graph(self, revision_id=None):
747
"""Return a dictionary containing the revision graph.
748
749
:return: a dictionary of revision_id->revision_parents_list.
750
"""
751
weave = self._get_revision_vf()
752
entire_graph = weave.get_graph()
753
if revision_id is None:
754
return weave.get_graph()
755
elif revision_id not in weave:
756
raise errors.NoSuchRevision(self, revision_id)
757
else:
758
# add what can be reached from revision_id
759
result = {}
760
pending = set([revision_id])
761
while len(pending) > 0:
762
node = pending.pop()
763
result[node] = weave.get_parents(node)
764
for revision_id in result[node]:
765
if revision_id not in result:
766
pending.add(revision_id)
767
return result
768
769
@needs_read_lock
770
def get_revision_graph_with_ghosts(self, revision_ids=None):
771
"""Return a graph of the revisions with ghosts marked as applicable.
772
773
:param revision_ids: an iterable of revisions to graph or None for all.
774
:return: a Graph object with the graph reachable from revision_ids.
775
"""
776
result = Graph()
777
vf = self._get_revision_vf()
778
versions = set(vf.versions())
779
if not revision_ids:
780
pending = set(self.all_revision_ids())
781
required = set([])
782
else:
783
pending = set(revision_ids)
784
required = set(revision_ids)
785
done = set([])
786
while len(pending):
787
revision_id = pending.pop()
788
if not revision_id in versions:
789
if revision_id in required:
790
raise errors.NoSuchRevision(self, revision_id)
791
# a ghost
792
result.add_ghost(revision_id)
793
# mark it as done so we dont try for it again.
794
done.add(revision_id)
795
continue
796
parent_ids = vf.get_parents_with_ghosts(revision_id)
797
for parent_id in parent_ids:
798
# is this queued or done ?
799
if (parent_id not in pending and
800
parent_id not in done):
801
# no, queue it.
802
pending.add(parent_id)
803
result.add_node(revision_id, parent_ids)
804
done.add(revision_id)
805
return result
806
807
def _get_revision_vf(self):
808
""":return: a versioned file containing the revisions."""
809
vf = self._revision_store.get_revision_file(self.get_transaction())
810
return vf
811
812
@needs_write_lock
813
def reconcile(self, other=None, thorough=False):
814
"""Reconcile this repository."""
815
from bzrlib.reconcile import KnitReconciler
816
reconciler = KnitReconciler(self, thorough=thorough)
817
reconciler.reconcile()
818
return reconciler
2039
return not self._transport.has('no-working-trees')
2040
2041
2042
class MetaDirVersionedFileRepository(MetaDirRepository):
2043
"""Repositories in a meta-dir, that work via versioned file objects."""
2044
2045
def __init__(self, _format, a_bzrdir, control_files):
2046
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2047
control_files)
2048
2049
2050
class RepositoryFormatRegistry(registry.Registry):
2051
"""Registry of RepositoryFormats."""
2052
2053
def get(self, format_string):
2054
r = registry.Registry.get(self, format_string)
2055
if callable(r):
2056
r = r()
2057
return r
819
2058
820
def revision_parents(self, revid):
821
return self._get_revision_vf().get_parents(rev_id)
2059
2060
format_registry = RepositoryFormatRegistry()
2061
"""Registry of formats, indexed by their identifying format string.
2062
2063
This can contain either format instances themselves, or classes/factories that
2064
can be called to obtain one.
2065
"""
2066
2067
2068
#####################################################################
2069
# Repository Formats
822
2070
823
2071
class RepositoryFormat(object):
824
2072
"""A repository format.
829
2077
children.
830
2078
* an open routine which returns a Repository instance.
831
2079
2080
There is one and only one Format subclass for each on-disk format. But
2081
there can be one Repository subclass that is used for several different
2082
formats. The _format attribute on a Repository instance can be used to
2083
determine the disk format.
2084
832
2085
Formats are placed in an dict by their format string for reference
833
2086
during opening. These should be subclasses of RepositoryFormat
834
2087
for consistency.
841
2094
_matchingbzrdir - the bzrdir format that the repository format was
842
2095
originally written to work with. This can be used if manually
843
2096
constructing a bzrdir and repository, or more commonly for test suite
844
parameterisation.
2097
parameterization.
845
2098
"""
846
2099
847
_default_format = None
848
"""The default format used for new repositories."""
849
850
_formats = {}
851
"""The known formats."""
2100
# Set to True or False in derived classes. True indicates that the format
2101
# supports ghosts gracefully.
2102
supports_ghosts = None
2103
# Can this repository be given external locations to lookup additional
2104
# data. Set to True or False in derived classes.
2105
supports_external_lookups = None
2106
2107
def __str__(self):
2108
return "<%s>" % self.__class__.__name__
2109
2110
def __eq__(self, other):
2111
# format objects are generally stateless
2112
return isinstance(other, self.__class__)
2113
2114
def __ne__(self, other):
2115
return not self == other
852
2116
853
2117
@classmethod
854
2118
def find_format(klass, a_bzrdir):
855
"""Return the format for the repository object in a_bzrdir."""
2119
"""Return the format for the repository object in a_bzrdir.
2120
2121
This is used by bzr native formats that have a "format" file in
2122
the repository. Other methods may be used by different types of
2123
control directory.
2124
"""
856
2125
try:
857
2126
transport = a_bzrdir.get_repository_transport(None)
858
2127
format_string = transport.get("format").read()
859
return klass._formats[format_string]
2128
return format_registry.get(format_string)
860
2129
except errors.NoSuchFile:
861
2130
raise errors.NoRepositoryPresent(a_bzrdir)
862
2131
except KeyError:
863
raise errors.UnknownFormatError(format_string)
864
865
def _get_control_store(self, repo_transport, control_files):
866
"""Return the control store for this repository."""
867
raise NotImplementedError(self._get_control_store)
2132
raise errors.UnknownFormatError(format=format_string,
2133
kind='repository')
2134
2135
@classmethod
2136
def register_format(klass, format):
2137
format_registry.register(format.get_format_string(), format)
2138
2139
@classmethod
2140
def unregister_format(klass, format):
2141
format_registry.remove(format.get_format_string())
868
2142
869
2143
@classmethod
870
2144
def get_default_format(klass):
871
2145
"""Return the current default format."""
872
return klass._default_format
2146
from bzrlib import bzrdir
2147
return bzrdir.format_registry.make_bzrdir('default').repository_format
873
2148
874
2149
def get_format_string(self):
875
2150
"""Return the ASCII format string that identifies this format.
880
2155
raise NotImplementedError(self.get_format_string)
881
2156
882
2157
def get_format_description(self):
883
"""Return the short desciption for this format."""
2158
"""Return the short description for this format."""
884
2159
raise NotImplementedError(self.get_format_description)
885
2160
886
def _get_revision_store(self, repo_transport, control_files):
887
"""Return the revision store object for this a_bzrdir."""
888
raise NotImplementedError(self._get_revision_store)
889
890
def _get_text_rev_store(self,
891
transport,
892
control_files,
893
name,
894
compressed=True,
895
prefixed=False,
896
serializer=None):
897
"""Common logic for getting a revision store for a repository.
898
899
see self._get_revision_store for the subclass-overridable method to
900
get the store for a repository.
901
"""
902
from bzrlib.store.revision.text import TextRevisionStore
903
dir_mode = control_files._dir_mode
904
file_mode = control_files._file_mode
905
text_store =TextStore(transport.clone(name),
906
prefixed=prefixed,
907
compressed=compressed,
908
dir_mode=dir_mode,
909
file_mode=file_mode)
910
_revision_store = TextRevisionStore(text_store, serializer)
911
return _revision_store
912
2161
# TODO: this shouldn't be in the base class, it's specific to things that
2162
# use weaves or knits -- mbp 20070207
913
2163
def _get_versioned_file_store(self,
914
2164
name,
915
2165
transport,
916
2166
control_files,
917
2167
prefixed=True,
918
versionedfile_class=WeaveFile,
2168
versionedfile_class=None,
2169
versionedfile_kwargs={},
919
2170
escaped=False):
2171
if versionedfile_class is None:
2172
versionedfile_class = self._versionedfile_class
920
2173
weave_transport = control_files._transport.clone(name)
921
2174
dir_mode = control_files._dir_mode
922
2175
file_mode = control_files._file_mode
924
2177
dir_mode=dir_mode,
925
2178
file_mode=file_mode,
926
2179
versionedfile_class=versionedfile_class,
2180
versionedfile_kwargs=versionedfile_kwargs,
927
2181
escaped=escaped)
928
2182
929
2183
def initialize(self, a_bzrdir, shared=False):
931
2185
932
2186
:param a_bzrdir: The bzrdir to put the new repository in it.
933
2187
:param shared: The repository should be initialized as a sharable one.
934
2188
:returns: The new repository object.
2189
935
2190
This may raise UninitializableFormat if shared repository are not
936
2191
compatible the a_bzrdir.
937
2192
"""
2193
raise NotImplementedError(self.initialize)
938
2194
939
2195
def is_supported(self):
940
2196
"""Is this format supported?
945
2201
"""
946
2202
return True
947
2203
2204
def check_conversion_target(self, target_format):
2205
raise NotImplementedError(self.check_conversion_target)
2206
948
2207
def open(self, a_bzrdir, _found=False):
949
2208
"""Return an instance of this format for the bzrdir a_bzrdir.
950
2209
952
2211
"""
953
2212
raise NotImplementedError(self.open)
954
2213
955
@classmethod
956
def register_format(klass, format):
957
klass._formats[format.get_format_string()] = format
958
959
@classmethod
960
def set_default_format(klass, format):
961
klass._default_format = format
962
963
@classmethod
964
def unregister_format(klass, format):
965
assert klass._formats[format.get_format_string()] is format
966
del klass._formats[format.get_format_string()]
967
968
969
class PreSplitOutRepositoryFormat(RepositoryFormat):
970
"""Base class for the pre split out repository formats."""
971
972
def initialize(self, a_bzrdir, shared=False, _internal=False):
973
"""Create a weave repository.
974
975
TODO: when creating split out bzr branch formats, move this to a common
976
base for Format5, Format6. or something like that.
977
"""
978
from bzrlib.weavefile import write_weave_v5
979
from bzrlib.weave import Weave
980
981
if shared:
982
raise errors.IncompatibleFormat(self, a_bzrdir._format)
983
984
if not _internal:
985
# always initialized when the bzrdir is.
986
return self.open(a_bzrdir, _found=True)
987
988
# Create an empty weave
989
sio = StringIO()
990
bzrlib.weavefile.write_weave_v5(Weave(), sio)
991
empty_weave = sio.getvalue()
992
993
mutter('creating repository in %s.', a_bzrdir.transport.base)
994
dirs = ['revision-store', 'weaves']
995
files = [('inventory.weave', StringIO(empty_weave)),
996
]
997
998
# FIXME: RBC 20060125 dont peek under the covers
999
# NB: no need to escape relative paths that are url safe.
1000
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1001
TransportLock)
1002
control_files.create_lock()
1003
control_files.lock_write()
1004
control_files._transport.mkdir_multi(dirs,
1005
mode=control_files._dir_mode)
1006
try:
1007
for file, content in files:
1008
control_files.put(file, content)
1009
finally:
1010
control_files.unlock()
1011
return self.open(a_bzrdir, _found=True)
1012
1013
def _get_control_store(self, repo_transport, control_files):
1014
"""Return the control store for this repository."""
1015
return self._get_versioned_file_store('',
1016
repo_transport,
1017
control_files,
1018
prefixed=False)
1019
1020
def _get_text_store(self, transport, control_files):
1021
"""Get a store for file texts for this format."""
1022
raise NotImplementedError(self._get_text_store)
1023
1024
def open(self, a_bzrdir, _found=False):
1025
"""See RepositoryFormat.open()."""
1026
if not _found:
1027
# we are being called directly and must probe.
1028
raise NotImplementedError
1029
1030
repo_transport = a_bzrdir.get_repository_transport(None)
1031
control_files = a_bzrdir._control_files
1032
text_store = self._get_text_store(repo_transport, control_files)
1033
control_store = self._get_control_store(repo_transport, control_files)
1034
_revision_store = self._get_revision_store(repo_transport, control_files)
1035
return AllInOneRepository(_format=self,
1036
a_bzrdir=a_bzrdir,
1037
_revision_store=_revision_store,
1038
control_store=control_store,
1039
text_store=text_store)
1040
1041
1042
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1043
"""Bzr repository format 4.
1044
1045
This repository format has:
1046
- flat stores
1047
- TextStores for texts, inventories,revisions.
1048
1049
This format is deprecated: it indexes texts using a text id which is
1050
removed in format 5; initializationa and write support for this format
1051
has been removed.
1052
"""
1053
1054
def __init__(self):
1055
super(RepositoryFormat4, self).__init__()
1056
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat4()
1057
1058
def get_format_description(self):
1059
"""See RepositoryFormat.get_format_description()."""
1060
return "Repository format 4"
1061
1062
def initialize(self, url, shared=False, _internal=False):
1063
"""Format 4 branches cannot be created."""
1064
raise errors.UninitializableFormat(self)
1065
1066
def is_supported(self):
1067
"""Format 4 is not supported.
1068
1069
It is not supported because the model changed from 4 to 5 and the
1070
conversion logic is expensive - so doing it on the fly was not
1071
feasible.
1072
"""
1073
return False
1074
1075
def _get_control_store(self, repo_transport, control_files):
1076
"""Format 4 repositories have no formal control store at this point.
1077
1078
This will cause any control-file-needing apis to fail - this is desired.
1079
"""
1080
return None
1081
1082
def _get_revision_store(self, repo_transport, control_files):
1083
"""See RepositoryFormat._get_revision_store()."""
1084
from bzrlib.xml4 import serializer_v4
1085
return self._get_text_rev_store(repo_transport,
1086
control_files,
1087
'revision-store',
1088
serializer=serializer_v4)
1089
1090
def _get_text_store(self, transport, control_files):
1091
"""See RepositoryFormat._get_text_store()."""
1092
1093
1094
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1095
"""Bzr control format 5.
1096
1097
This repository format has:
1098
- weaves for file texts and inventory
1099
- flat stores
1100
- TextStores for revisions and signatures.
1101
"""
1102
1103
def __init__(self):
1104
super(RepositoryFormat5, self).__init__()
1105
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat5()
1106
1107
def get_format_description(self):
1108
"""See RepositoryFormat.get_format_description()."""
1109
return "Weave repository format 5"
1110
1111
def _get_revision_store(self, repo_transport, control_files):
1112
"""See RepositoryFormat._get_revision_store()."""
1113
"""Return the revision store object for this a_bzrdir."""
1114
return self._get_text_rev_store(repo_transport,
1115
control_files,
1116
'revision-store',
1117
compressed=False)
1118
1119
def _get_text_store(self, transport, control_files):
1120
"""See RepositoryFormat._get_text_store()."""
1121
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1122
1123
1124
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1125
"""Bzr control format 6.
1126
1127
This repository format has:
1128
- weaves for file texts and inventory
1129
- hash subdirectory based stores.
1130
- TextStores for revisions and signatures.
1131
"""
1132
1133
def __init__(self):
1134
super(RepositoryFormat6, self).__init__()
1135
self._matchingbzrdir = bzrlib.bzrdir.BzrDirFormat6()
1136
1137
def get_format_description(self):
1138
"""See RepositoryFormat.get_format_description()."""
1139
return "Weave repository format 6"
1140
1141
def _get_revision_store(self, repo_transport, control_files):
1142
"""See RepositoryFormat._get_revision_store()."""
1143
return self._get_text_rev_store(repo_transport,
1144
control_files,
1145
'revision-store',
1146
compressed=False,
1147
prefixed=True)
1148
1149
def _get_text_store(self, transport, control_files):
1150
"""See RepositoryFormat._get_text_store()."""
1151
return self._get_versioned_file_store('weaves', transport, control_files)
1152
1153
2214
1154
2215
class MetaDirRepositoryFormat(RepositoryFormat):
1155
"""Common base class for the new repositories using the metadir layour."""
2216
"""Common base class for the new repositories using the metadir layout."""
2217
2218
rich_root_data = False
2219
supports_tree_reference = False
2220
supports_external_lookups = False
2221
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1156
2222
1157
2223
def __init__(self):
1158
2224
super(MetaDirRepositoryFormat, self).__init__()
1159
self._matchingbzrdir = bzrlib.bzrdir.BzrDirMetaFormat1()
1160
2225
1161
2226
def _create_control_files(self, a_bzrdir):
1162
2227
"""Create the required files and the initial control_files object."""
1163
# FIXME: RBC 20060125 dont peek under the covers
2228
# FIXME: RBC 20060125 don't peek under the covers
1164
2229
# NB: no need to escape relative paths that are url safe.
1165
2230
repository_transport = a_bzrdir.get_repository_transport(self)
1166
control_files = LockableFiles(repository_transport, 'lock', LockDir)
2231
control_files = lockable_files.LockableFiles(repository_transport,
2232
'lock', lockdir.LockDir)
1167
2233
control_files.create_lock()
1168
2234
return control_files
1169
2235
1171
2237
"""Upload the initial blank content."""
1172
2238
control_files = self._create_control_files(a_bzrdir)
1173
2239
control_files.lock_write()
2240
transport = control_files._transport
2241
if shared == True:
2242
utf8_files += [('shared-storage', '')]
1174
2243
try:
1175
control_files._transport.mkdir_multi(dirs,
1176
mode=control_files._dir_mode)
1177
for file, content in files:
1178
control_files.put(file, content)
1179
for file, content in utf8_files:
1180
control_files.put_utf8(file, content)
1181
if shared == True:
1182
control_files.put_utf8('shared-storage', '')
2244
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2245
for (filename, content_stream) in files:
2246
transport.put_file(filename, content_stream,
2247
mode=a_bzrdir._get_file_mode())
2248
for (filename, content_bytes) in utf8_files:
2249
transport.put_bytes_non_atomic(filename, content_bytes,
2250
mode=a_bzrdir._get_file_mode())
1183
2251
finally:
1184
2252
control_files.unlock()
1185
2253
1186
2254
1187
class RepositoryFormat7(MetaDirRepositoryFormat):
1188
"""Bzr repository 7.
1189
1190
This repository format has:
1191
- weaves for file texts and inventory
1192
- hash subdirectory based stores.
1193
- TextStores for revisions and signatures.
1194
- a format marker of its own
1195
- an optional 'shared-storage' flag
1196
- an optional 'no-working-trees' flag
1197
"""
1198
1199
def _get_control_store(self, repo_transport, control_files):
1200
"""Return the control store for this repository."""
1201
return self._get_versioned_file_store('',
1202
repo_transport,
1203
control_files,
1204
prefixed=False)
1205
1206
def get_format_string(self):
1207
"""See RepositoryFormat.get_format_string()."""
1208
return "Bazaar-NG Repository format 7"
1209
1210
def get_format_description(self):
1211
"""See RepositoryFormat.get_format_description()."""
1212
return "Weave repository format 7"
1213
1214
def _get_revision_store(self, repo_transport, control_files):
1215
"""See RepositoryFormat._get_revision_store()."""
1216
return self._get_text_rev_store(repo_transport,
1217
control_files,
1218
'revision-store',
1219
compressed=False,
1220
prefixed=True,
1221
)
1222
1223
def _get_text_store(self, transport, control_files):
1224
"""See RepositoryFormat._get_text_store()."""
1225
return self._get_versioned_file_store('weaves',
1226
transport,
1227
control_files)
1228
1229
def initialize(self, a_bzrdir, shared=False):
1230
"""Create a weave repository.
1231
1232
:param shared: If true the repository will be initialized as a shared
1233
repository.
1234
"""
1235
from bzrlib.weavefile import write_weave_v5
1236
from bzrlib.weave import Weave
1237
1238
# Create an empty weave
1239
sio = StringIO()
1240
bzrlib.weavefile.write_weave_v5(Weave(), sio)
1241
empty_weave = sio.getvalue()
1242
1243
mutter('creating repository in %s.', a_bzrdir.transport.base)
1244
dirs = ['revision-store', 'weaves']
1245
files = [('inventory.weave', StringIO(empty_weave)),
1246
]
1247
utf8_files = [('format', self.get_format_string())]
1248
1249
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1250
return self.open(a_bzrdir=a_bzrdir, _found=True)
1251
1252
def open(self, a_bzrdir, _found=False, _override_transport=None):
1253
"""See RepositoryFormat.open().
1254
1255
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1256
repository at a slightly different url
1257
than normal. I.e. during 'upgrade'.
1258
"""
1259
if not _found:
1260
format = RepositoryFormat.find_format(a_bzrdir)
1261
assert format.__class__ == self.__class__
1262
if _override_transport is not None:
1263
repo_transport = _override_transport
1264
else:
1265
repo_transport = a_bzrdir.get_repository_transport(None)
1266
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1267
text_store = self._get_text_store(repo_transport, control_files)
1268
control_store = self._get_control_store(repo_transport, control_files)
1269
_revision_store = self._get_revision_store(repo_transport, control_files)
1270
return MetaDirRepository(_format=self,
1271
a_bzrdir=a_bzrdir,
1272
control_files=control_files,
1273
_revision_store=_revision_store,
1274
control_store=control_store,
1275
text_store=text_store)
1276
1277
1278
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1279
"""Bzr repository knit format 1.
1280
1281
This repository format has:
1282
- knits for file texts and inventory
1283
- hash subdirectory based stores.
1284
- knits for revisions and signatures
1285
- TextStores for revisions and signatures.
1286
- a format marker of its own
1287
- an optional 'shared-storage' flag
1288
- an optional 'no-working-trees' flag
1289
- a LockDir lock
1290
1291
This format was introduced in bzr 0.8.
1292
"""
1293
1294
def _get_control_store(self, repo_transport, control_files):
1295
"""Return the control store for this repository."""
1296
return VersionedFileStore(
1297
repo_transport,
1298
prefixed=False,
1299
file_mode=control_files._file_mode,
1300
versionedfile_class=KnitVersionedFile,
1301
versionedfile_kwargs={'factory':KnitPlainFactory()},
1302
)
1303
1304
def get_format_string(self):
1305
"""See RepositoryFormat.get_format_string()."""
1306
return "Bazaar-NG Knit Repository Format 1"
1307
1308
def get_format_description(self):
1309
"""See RepositoryFormat.get_format_description()."""
1310
return "Knit repository format 1"
1311
1312
def _get_revision_store(self, repo_transport, control_files):
1313
"""See RepositoryFormat._get_revision_store()."""
1314
from bzrlib.store.revision.knit import KnitRevisionStore
1315
versioned_file_store = VersionedFileStore(
1316
repo_transport,
1317
file_mode=control_files._file_mode,
1318
prefixed=False,
1319
precious=True,
1320
versionedfile_class=KnitVersionedFile,
1321
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1322
escaped=True,
1323
)
1324
return KnitRevisionStore(versioned_file_store)
1325
1326
def _get_text_store(self, transport, control_files):
1327
"""See RepositoryFormat._get_text_store()."""
1328
return self._get_versioned_file_store('knits',
1329
transport,
1330
control_files,
1331
versionedfile_class=KnitVersionedFile,
1332
escaped=True)
1333
1334
def initialize(self, a_bzrdir, shared=False):
1335
"""Create a knit format 1 repository.
1336
1337
:param a_bzrdir: bzrdir to contain the new repository; must already
1338
be initialized.
1339
:param shared: If true the repository will be initialized as a shared
1340
repository.
1341
"""
1342
mutter('creating repository in %s.', a_bzrdir.transport.base)
1343
dirs = ['revision-store', 'knits']
1344
files = []
1345
utf8_files = [('format', self.get_format_string())]
1346
1347
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1348
repo_transport = a_bzrdir.get_repository_transport(None)
1349
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1350
control_store = self._get_control_store(repo_transport, control_files)
1351
transaction = bzrlib.transactions.WriteTransaction()
1352
# trigger a write of the inventory store.
1353
control_store.get_weave_or_empty('inventory', transaction)
1354
_revision_store = self._get_revision_store(repo_transport, control_files)
1355
_revision_store.has_revision_id('A', transaction)
1356
_revision_store.get_signature_file(transaction)
1357
return self.open(a_bzrdir=a_bzrdir, _found=True)
1358
1359
def open(self, a_bzrdir, _found=False, _override_transport=None):
1360
"""See RepositoryFormat.open().
1361
1362
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1363
repository at a slightly different url
1364
than normal. I.e. during 'upgrade'.
1365
"""
1366
if not _found:
1367
format = RepositoryFormat.find_format(a_bzrdir)
1368
assert format.__class__ == self.__class__
1369
if _override_transport is not None:
1370
repo_transport = _override_transport
1371
else:
1372
repo_transport = a_bzrdir.get_repository_transport(None)
1373
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1374
text_store = self._get_text_store(repo_transport, control_files)
1375
control_store = self._get_control_store(repo_transport, control_files)
1376
_revision_store = self._get_revision_store(repo_transport, control_files)
1377
return KnitRepository(_format=self,
1378
a_bzrdir=a_bzrdir,
1379
control_files=control_files,
1380
_revision_store=_revision_store,
1381
control_store=control_store,
1382
text_store=text_store)
1383
1384
1385
2255
# formats which have no format string are not discoverable
1386
# and not independently creatable, so are not registered.
1387
RepositoryFormat.register_format(RepositoryFormat7())
1388
_default_format = RepositoryFormatKnit1()
1389
RepositoryFormat.register_format(_default_format)
1390
RepositoryFormat.set_default_format(_default_format)
1391
_legacy_formats = [RepositoryFormat4(),
1392
RepositoryFormat5(),
1393
RepositoryFormat6()]
2256
# and not independently creatable, so are not registered. They're
2257
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2258
# needed, it's constructed directly by the BzrDir. Non-native formats where
2259
# the repository is not separately opened are similar.
2260
2261
format_registry.register_lazy(
2262
'Bazaar-NG Repository format 7',
2263
'bzrlib.repofmt.weaverepo',
2264
'RepositoryFormat7'
2265
)
2266
2267
format_registry.register_lazy(
2268
'Bazaar-NG Knit Repository Format 1',
2269
'bzrlib.repofmt.knitrepo',
2270
'RepositoryFormatKnit1',
2271
)
2272
2273
format_registry.register_lazy(
2274
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2275
'bzrlib.repofmt.knitrepo',
2276
'RepositoryFormatKnit3',
2277
)
2278
2279
format_registry.register_lazy(
2280
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2281
'bzrlib.repofmt.knitrepo',
2282
'RepositoryFormatKnit4',
2283
)
2284
2285
# Pack-based formats. There is one format for pre-subtrees, and one for
2286
# post-subtrees to allow ease of testing.
2287
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2288
format_registry.register_lazy(
2289
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2290
'bzrlib.repofmt.pack_repo',
2291
'RepositoryFormatKnitPack1',
2292
)
2293
format_registry.register_lazy(
2294
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2295
'bzrlib.repofmt.pack_repo',
2296
'RepositoryFormatKnitPack3',
2297
)
2298
format_registry.register_lazy(
2299
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2300
'bzrlib.repofmt.pack_repo',
2301
'RepositoryFormatKnitPack4',
2302
)
2303
format_registry.register_lazy(
2304
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2305
'bzrlib.repofmt.pack_repo',
2306
'RepositoryFormatKnitPack5',
2307
)
2308
format_registry.register_lazy(
2309
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2310
'bzrlib.repofmt.pack_repo',
2311
'RepositoryFormatKnitPack5RichRoot',
2312
)
2313
2314
# Development formats.
2315
# 1.2->1.3
2316
# development 0 - stub to introduce development versioning scheme.
2317
format_registry.register_lazy(
2318
"Bazaar development format 0 (needs bzr.dev from before 1.3)\n",
2319
'bzrlib.repofmt.pack_repo',
2320
'RepositoryFormatPackDevelopment0',
2321
)
2322
format_registry.register_lazy(
2323
("Bazaar development format 0 with subtree support "
2324
"(needs bzr.dev from before 1.3)\n"),
2325
'bzrlib.repofmt.pack_repo',
2326
'RepositoryFormatPackDevelopment0Subtree',
2327
)
2328
format_registry.register_lazy(
2329
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2330
'bzrlib.repofmt.pack_repo',
2331
'RepositoryFormatPackDevelopment1',
2332
)
2333
format_registry.register_lazy(
2334
("Bazaar development format 1 with subtree support "
2335
"(needs bzr.dev from before 1.6)\n"),
2336
'bzrlib.repofmt.pack_repo',
2337
'RepositoryFormatPackDevelopment1Subtree',
2338
)
2339
# 1.3->1.4 go below here
1394
2340
1395
2341
1396
2342
class InterRepository(InterObject):
1405
2351
InterRepository.get(other).method_name(parameters).
1406
2352
"""
1407
2353
1408
_optimisers = set()
2354
_optimisers = []
1409
2355
"""The available optimised InterRepository types."""
1410
2356
1411
@needs_write_lock
1412
def copy_content(self, revision_id=None, basis=None):
1413
"""Make a complete copy of the content in self into destination.
1414
1415
This is a destructive operation! Do not use it on existing
1416
repositories.
1417
1418
:param revision_id: Only copy the content needed to construct
1419
revision_id and its parents.
1420
:param basis: Copy the needed data preferentially from basis.
1421
"""
1422
try:
1423
self.target.set_make_working_trees(self.source.make_working_trees())
1424
except NotImplementedError:
1425
pass
1426
# grab the basis available data
1427
if basis is not None:
1428
self.target.fetch(basis, revision_id=revision_id)
1429
# but dont bother fetching if we have the needed data now.
1430
if (revision_id not in (None, NULL_REVISION) and
1431
self.target.has_revision(revision_id)):
1432
return
1433
self.target.fetch(self.source, revision_id=revision_id)
1434
1435
def _double_lock(self, lock_source, lock_target):
1436
"""Take out too locks, rolling back the first if the second throws."""
1437
lock_source()
1438
try:
1439
lock_target()
1440
except Exception:
1441
# we want to ensure that we don't leave source locked by mistake.
1442
# and any error on target should not confuse source.
1443
self.source.unlock()
1444
raise
1445
1446
@needs_write_lock
1447
def fetch(self, revision_id=None, pb=None):
2357
def copy_content(self, revision_id=None):
2358
raise NotImplementedError(self.copy_content)
2359
2360
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1448
2361
"""Fetch the content required to construct revision_id.
1449
2362
1450
The content is copied from source to target.
2363
The content is copied from self.source to self.target.
1451
2364
1452
2365
:param revision_id: if None all content is copied, if NULL_REVISION no
1453
2366
content is copied.
1454
2367
:param pb: optional progress bar to use for progress reports. If not
1455
2368
provided a default one will be created.
1456
2369
1457
Returns the copied revision count and the failed revisions in a tuple:
1458
(copied, failures).
1459
"""
1460
from bzrlib.fetch import GenericRepoFetcher
1461
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1462
self.source, self.source._format, self.target, self.target._format)
1463
f = GenericRepoFetcher(to_repository=self.target,
1464
from_repository=self.source,
1465
last_revision=revision_id,
1466
pb=pb)
1467
return f.count_copied, f.failed_revisions
1468
1469
def lock_read(self):
1470
"""Take out a logical read lock.
1471
1472
This will lock the source branch and the target branch. The source gets
1473
a read lock and the target a read lock.
1474
"""
1475
self._double_lock(self.source.lock_read, self.target.lock_read)
1476
1477
def lock_write(self):
1478
"""Take out a logical write lock.
1479
1480
This will lock the source branch and the target branch. The source gets
1481
a read lock and the target a write lock.
1482
"""
1483
self._double_lock(self.source.lock_read, self.target.lock_write)
1484
2370
:returns: (copied_revision_count, failures).
2371
"""
2372
# Normally we should find a specific InterRepository subclass to do
2373
# the fetch; if nothing else then at least InterSameDataRepository.
2374
# If none of them is suitable it looks like fetching is not possible;
2375
# we try to give a good message why. _assert_same_model will probably
2376
# give a helpful message; otherwise a generic one.
2377
self._assert_same_model(self.source, self.target)
2378
raise errors.IncompatibleRepositories(self.source, self.target,
2379
"no suitableInterRepository found")
2380
2381
def _walk_to_common_revisions(self, revision_ids):
2382
"""Walk out from revision_ids in source to revisions target has.
2383
2384
:param revision_ids: The start point for the search.
2385
:return: A set of revision ids.
2386
"""
2387
target_graph = self.target.get_graph()
2388
revision_ids = frozenset(revision_ids)
2389
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2390
return graph.SearchResult(revision_ids, set(), 0, set())
2391
missing_revs = set()
2392
source_graph = self.source.get_graph()
2393
# ensure we don't pay silly lookup costs.
2394
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2395
null_set = frozenset([_mod_revision.NULL_REVISION])
2396
while True:
2397
try:
2398
next_revs, ghosts = searcher.next_with_ghosts()
2399
except StopIteration:
2400
break
2401
if revision_ids.intersection(ghosts):
2402
absent_ids = set(revision_ids.intersection(ghosts))
2403
# If all absent_ids are present in target, no error is needed.
2404
absent_ids.difference_update(
2405
set(target_graph.get_parent_map(absent_ids)))
2406
if absent_ids:
2407
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2408
# we don't care about other ghosts as we can't fetch them and
2409
# haven't been asked to.
2410
next_revs = set(next_revs)
2411
# we always have NULL_REVISION present.
2412
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2413
missing_revs.update(next_revs - have_revs)
2414
searcher.stop_searching_any(have_revs)
2415
return searcher.get_result()
2416
2417
@deprecated_method(one_two)
1485
2418
@needs_read_lock
1486
def missing_revision_ids(self, revision_id=None):
2419
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1487
2420
"""Return the revision ids that source has that target does not.
1488
2421
1489
2422
These are returned in topological order.
1490
2423
1491
2424
:param revision_id: only return revision ids included by this
1492
2425
revision_id.
1493
"""
2426
:param find_ghosts: If True find missing revisions in deep history
2427
rather than just finding the surface difference.
2428
"""
2429
return list(self.search_missing_revision_ids(
2430
revision_id, find_ghosts).get_keys())
2431
2432
@needs_read_lock
2433
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2434
"""Return the revision ids that source has that target does not.
2435
2436
:param revision_id: only return revision ids included by this
2437
revision_id.
2438
:param find_ghosts: If True find missing revisions in deep history
2439
rather than just finding the surface difference.
2440
:return: A bzrlib.graph.SearchResult.
2441
"""
2442
# stop searching at found target revisions.
2443
if not find_ghosts and revision_id is not None:
2444
return self._walk_to_common_revisions([revision_id])
1494
2445
# generic, possibly worst case, slow code path.
1495
2446
target_ids = set(self.target.all_revision_ids())
1496
2447
if revision_id is not None:
1497
2448
source_ids = self.source.get_ancestry(revision_id)
1498
assert source_ids[0] == None
2449
if source_ids[0] is not None:
2450
raise AssertionError()
1499
2451
source_ids.pop(0)
1500
2452
else:
1501
2453
source_ids = self.source.all_revision_ids()
1502
2454
result_set = set(source_ids).difference(target_ids)
1503
# this may look like a no-op: its not. It preserves the ordering
1504
# other_ids had while only returning the members from other_ids
1505
# that we've decided we need.
1506
return [rev_id for rev_id in source_ids if rev_id in result_set]
1507
1508
def unlock(self):
1509
"""Release the locks on source and target."""
1510
try:
1511
self.target.unlock()
1512
finally:
1513
self.source.unlock()
1514
1515
1516
class InterWeaveRepo(InterRepository):
1517
"""Optimised code paths between Weave based repositories."""
1518
1519
_matching_repo_format = RepositoryFormat7()
1520
"""Repository format for testing with."""
2455
return self.source.revision_ids_to_search_result(result_set)
2456
2457
@staticmethod
2458
def _same_model(source, target):
2459
"""True if source and target have the same data representation.
2460
2461
Note: this is always called on the base class; overriding it in a
2462
subclass will have no effect.
2463
"""
2464
try:
2465
InterRepository._assert_same_model(source, target)
2466
return True
2467
except errors.IncompatibleRepositories, e:
2468
return False
2469
2470
@staticmethod
2471
def _assert_same_model(source, target):
2472
"""Raise an exception if two repositories do not use the same model.
2473
"""
2474
if source.supports_rich_root() != target.supports_rich_root():
2475
raise errors.IncompatibleRepositories(source, target,
2476
"different rich-root support")
2477
if source._serializer != target._serializer:
2478
raise errors.IncompatibleRepositories(source, target,
2479
"different serializers")
2480
2481
2482
class InterSameDataRepository(InterRepository):
2483
"""Code for converting between repositories that represent the same data.
2484
2485
Data format and model must match for this to work.
2486
"""
2487
2488
@classmethod
2489
def _get_repo_format_to_test(self):
2490
"""Repository format for testing with.
2491
2492
InterSameData can pull from subtree to subtree and from non-subtree to
2493
non-subtree, so we test this with the richest repository format.
2494
"""
2495
from bzrlib.repofmt import knitrepo
2496
return knitrepo.RepositoryFormatKnit3()
2497
2498
@staticmethod
2499
def is_compatible(source, target):
2500
return InterRepository._same_model(source, target)
2501
2502
@needs_write_lock
2503
def copy_content(self, revision_id=None):
2504
"""Make a complete copy of the content in self into destination.
2505
2506
This copies both the repository's revision data, and configuration information
2507
such as the make_working_trees setting.
2508
2509
This is a destructive operation! Do not use it on existing
2510
repositories.
2511
2512
:param revision_id: Only copy the content needed to construct
2513
revision_id and its parents.
2514
"""
2515
try:
2516
self.target.set_make_working_trees(self.source.make_working_trees())
2517
except NotImplementedError:
2518
pass
2519
# but don't bother fetching if we have the needed data now.
2520
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2521
self.target.has_revision(revision_id)):
2522
return
2523
self.target.fetch(self.source, revision_id=revision_id)
2524
2525
@needs_write_lock
2526
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2527
"""See InterRepository.fetch()."""
2528
from bzrlib.fetch import RepoFetcher
2529
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2530
self.source, self.source._format, self.target,
2531
self.target._format)
2532
f = RepoFetcher(to_repository=self.target,
2533
from_repository=self.source,
2534
last_revision=revision_id,
2535
pb=pb, find_ghosts=find_ghosts)
2536
return f.count_copied, f.failed_revisions
2537
2538
2539
class InterWeaveRepo(InterSameDataRepository):
2540
"""Optimised code paths between Weave based repositories.
2541
2542
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2543
implemented lazy inter-object optimisation.
2544
"""
2545
2546
@classmethod
2547
def _get_repo_format_to_test(self):
2548
from bzrlib.repofmt import weaverepo
2549
return weaverepo.RepositoryFormat7()
1521
2550
1522
2551
@staticmethod
1523
2552
def is_compatible(source, target):
1524
2553
"""Be compatible with known Weave formats.
1525
2554
1526
We dont test for the stores being of specific types becase that
2555
We don't test for the stores being of specific types because that
1527
2556
could lead to confusing results, and there is no need to be
1528
2557
overly general.
1529
2558
"""
2559
from bzrlib.repofmt.weaverepo import (
2560
RepositoryFormat5,
2561
RepositoryFormat6,
2562
RepositoryFormat7,
2563
)
1530
2564
try:
1531
2565
return (isinstance(source._format, (RepositoryFormat5,
1532
2566
RepositoryFormat6,
1538
2572
return False
1539
2573
1540
2574
@needs_write_lock
1541
def copy_content(self, revision_id=None, basis=None):
2575
def copy_content(self, revision_id=None):
1542
2576
"""See InterRepository.copy_content()."""
1543
2577
# weave specific optimised path:
1544
if basis is not None:
1545
# copy the basis in, then fetch remaining data.
1546
basis.copy_content_into(self.target, revision_id)
1547
# the basis copy_content_into could misset this.
2578
try:
2579
self.target.set_make_working_trees(self.source.make_working_trees())
2580
except (errors.RepositoryUpgradeRequired, NotImplemented):
2581
pass
2582
# FIXME do not peek!
2583
if self.source._transport.listable():
2584
pb = ui.ui_factory.nested_progress_bar()
1548
2585
try:
1549
self.target.set_make_working_trees(self.source.make_working_trees())
1550
except NotImplementedError:
1551
pass
2586
self.target.texts.insert_record_stream(
2587
self.source.texts.get_record_stream(
2588
self.source.texts.keys(), 'topological', False))
2589
pb.update('copying inventory', 0, 1)
2590
self.target.inventories.insert_record_stream(
2591
self.source.inventories.get_record_stream(
2592
self.source.inventories.keys(), 'topological', False))
2593
self.target.signatures.insert_record_stream(
2594
self.source.signatures.get_record_stream(
2595
self.source.signatures.keys(),
2596
'unordered', True))
2597
self.target.revisions.insert_record_stream(
2598
self.source.revisions.get_record_stream(
2599
self.source.revisions.keys(),
2600
'topological', True))
2601
finally:
2602
pb.finished()
2603
else:
1552
2604
self.target.fetch(self.source, revision_id=revision_id)
1553
else:
1554
try:
1555
self.target.set_make_working_trees(self.source.make_working_trees())
1556
except NotImplementedError:
1557
pass
1558
# FIXME do not peek!
1559
if self.source.control_files._transport.listable():
1560
pb = bzrlib.ui.ui_factory.nested_progress_bar()
1561
try:
1562
self.target.weave_store.copy_all_ids(
1563
self.source.weave_store,
1564
pb=pb,
1565
from_transaction=self.source.get_transaction(),
1566
to_transaction=self.target.get_transaction())
1567
pb.update('copying inventory', 0, 1)
1568
self.target.control_weaves.copy_multi(
1569
self.source.control_weaves, ['inventory'],
1570
from_transaction=self.source.get_transaction(),
1571
to_transaction=self.target.get_transaction())
1572
self.target._revision_store.text_store.copy_all_ids(
1573
self.source._revision_store.text_store,
1574
pb=pb)
1575
finally:
1576
pb.finished()
1577
else:
1578
self.target.fetch(self.source, revision_id=revision_id)
1579
2605
1580
2606
@needs_write_lock
1581
def fetch(self, revision_id=None, pb=None):
2607
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1582
2608
"""See InterRepository.fetch()."""
1583
from bzrlib.fetch import GenericRepoFetcher
2609
from bzrlib.fetch import RepoFetcher
1584
2610
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1585
2611
self.source, self.source._format, self.target, self.target._format)
1586
f = GenericRepoFetcher(to_repository=self.target,
2612
f = RepoFetcher(to_repository=self.target,
1587
2613
from_repository=self.source,
1588
2614
last_revision=revision_id,
1589
pb=pb)
2615
pb=pb, find_ghosts=find_ghosts)
1590
2616
return f.count_copied, f.failed_revisions
1591
2617
1592
2618
@needs_read_lock
1593
def missing_revision_ids(self, revision_id=None):
2619
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1594
2620
"""See InterRepository.missing_revision_ids()."""
1595
2621
# we want all revisions to satisfy revision_id in source.
1596
# but we dont want to stat every file here and there.
2622
# but we don't want to stat every file here and there.
1597
2623
# we want then, all revisions other needs to satisfy revision_id
1598
2624
# checked, but not those that we have locally.
1599
2625
# so the first thing is to get a subset of the revisions to
1605
2631
# - RBC 20060209
1606
2632
if revision_id is not None:
1607
2633
source_ids = self.source.get_ancestry(revision_id)
1608
assert source_ids[0] == None
2634
if source_ids[0] is not None:
2635
raise AssertionError()
1609
2636
source_ids.pop(0)
1610
2637
else:
1611
2638
source_ids = self.source._all_possible_ids()
1612
2639
source_ids_set = set(source_ids)
1613
2640
# source_ids is the worst possible case we may need to pull.
1614
2641
# now we want to filter source_ids against what we actually
1615
# have in target, but dont try to check for existence where we know
2642
# have in target, but don't try to check for existence where we know
1616
2643
# we do not have a revision as that would be pointless.
1617
2644
target_ids = set(self.target._all_possible_ids())
1618
2645
possibly_present_revisions = target_ids.intersection(source_ids_set)
1619
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2646
actually_present_revisions = set(
2647
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1620
2648
required_revisions = source_ids_set.difference(actually_present_revisions)
1621
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1622
2649
if revision_id is not None:
1623
2650
# we used get_ancestry to determine source_ids then we are assured all
1624
2651
# revisions referenced are present as they are installed in topological order.
1625
2652
# and the tip revision was validated by get_ancestry.
1626
return required_topo_revisions
2653
result_set = required_revisions
1627
2654
else:
1628
2655
# if we just grabbed the possibly available ids, then
1629
2656
# we only have an estimate of whats available and need to validate
1630
2657
# that against the revision records.
1631
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1632
1633
1634
class InterKnitRepo(InterRepository):
2658
result_set = set(
2659
self.source._eliminate_revisions_not_present(required_revisions))
2660
return self.source.revision_ids_to_search_result(result_set)
2661
2662
2663
class InterKnitRepo(InterSameDataRepository):
1635
2664
"""Optimised code paths between Knit based repositories."""
1636
2665
1637
_matching_repo_format = RepositoryFormatKnit1()
1638
"""Repository format for testing with."""
2666
@classmethod
2667
def _get_repo_format_to_test(self):
2668
from bzrlib.repofmt import knitrepo
2669
return knitrepo.RepositoryFormatKnit1()
1639
2670
1640
2671
@staticmethod
1641
2672
def is_compatible(source, target):
1642
2673
"""Be compatible with known Knit formats.
1643
2674
1644
We dont test for the stores being of specific types becase that
2675
We don't test for the stores being of specific types because that
1645
2676
could lead to confusing results, and there is no need to be
1646
2677
overly general.
1647
2678
"""
2679
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
1648
2680
try:
1649
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1650
isinstance(target._format, (RepositoryFormatKnit1)))
2681
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2682
isinstance(target._format, RepositoryFormatKnit))
1651
2683
except AttributeError:
1652
2684
return False
2685
return are_knits and InterRepository._same_model(source, target)
1653
2686
1654
2687
@needs_write_lock
1655
def fetch(self, revision_id=None, pb=None):
2688
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1656
2689
"""See InterRepository.fetch()."""
1657
from bzrlib.fetch import KnitRepoFetcher
2690
from bzrlib.fetch import RepoFetcher
1658
2691
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1659
2692
self.source, self.source._format, self.target, self.target._format)
1660
f = KnitRepoFetcher(to_repository=self.target,
2693
f = RepoFetcher(to_repository=self.target,
1661
2694
from_repository=self.source,
1662
2695
last_revision=revision_id,
1663
pb=pb)
2696
pb=pb, find_ghosts=find_ghosts)
1664
2697
return f.count_copied, f.failed_revisions
1665
2698
1666
2699
@needs_read_lock
1667
def missing_revision_ids(self, revision_id=None):
2700
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1668
2701
"""See InterRepository.missing_revision_ids()."""
1669
2702
if revision_id is not None:
1670
2703
source_ids = self.source.get_ancestry(revision_id)
1671
assert source_ids[0] == None
2704
if source_ids[0] is not None:
2705
raise AssertionError()
1672
2706
source_ids.pop(0)
1673
2707
else:
1674
source_ids = self.source._all_possible_ids()
2708
source_ids = self.source.all_revision_ids()
1675
2709
source_ids_set = set(source_ids)
1676
2710
# source_ids is the worst possible case we may need to pull.
1677
2711
# now we want to filter source_ids against what we actually
1678
# have in target, but dont try to check for existence where we know
2712
# have in target, but don't try to check for existence where we know
1679
2713
# we do not have a revision as that would be pointless.
1680
target_ids = set(self.target._all_possible_ids())
2714
target_ids = set(self.target.all_revision_ids())
1681
2715
possibly_present_revisions = target_ids.intersection(source_ids_set)
1682
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2716
actually_present_revisions = set(
2717
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1683
2718
required_revisions = source_ids_set.difference(actually_present_revisions)
1684
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1685
2719
if revision_id is not None:
1686
2720
# we used get_ancestry to determine source_ids then we are assured all
1687
2721
# revisions referenced are present as they are installed in topological order.
1688
2722
# and the tip revision was validated by get_ancestry.
1689
return required_topo_revisions
2723
result_set = required_revisions
1690
2724
else:
1691
2725
# if we just grabbed the possibly available ids, then
1692
2726
# we only have an estimate of whats available and need to validate
1693
2727
# that against the revision records.
1694
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1695
2728
result_set = set(
2729
self.source._eliminate_revisions_not_present(required_revisions))
2730
return self.source.revision_ids_to_search_result(result_set)
2731
2732
2733
class InterPackRepo(InterSameDataRepository):
2734
"""Optimised code paths between Pack based repositories."""
2735
2736
@classmethod
2737
def _get_repo_format_to_test(self):
2738
from bzrlib.repofmt import pack_repo
2739
return pack_repo.RepositoryFormatKnitPack1()
2740
2741
@staticmethod
2742
def is_compatible(source, target):
2743
"""Be compatible with known Pack formats.
2744
2745
We don't test for the stores being of specific types because that
2746
could lead to confusing results, and there is no need to be
2747
overly general.
2748
"""
2749
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2750
try:
2751
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2752
isinstance(target._format, RepositoryFormatPack))
2753
except AttributeError:
2754
return False
2755
return are_packs and InterRepository._same_model(source, target)
2756
2757
@needs_write_lock
2758
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2759
"""See InterRepository.fetch()."""
2760
if (len(self.source._fallback_repositories) > 0 or
2761
len(self.target._fallback_repositories) > 0):
2762
# The pack layer is not aware of fallback repositories, so when
2763
# fetching from a stacked repository or into a stacked repository
2764
# we use the generic fetch logic which uses the VersionedFiles
2765
# attributes on repository.
2766
from bzrlib.fetch import RepoFetcher
2767
fetcher = RepoFetcher(self.target, self.source, revision_id,
2768
pb, find_ghosts)
2769
return fetcher.count_copied, fetcher.failed_revisions
2770
from bzrlib.repofmt.pack_repo import Packer
2771
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2772
self.source, self.source._format, self.target, self.target._format)
2773
self.count_copied = 0
2774
if revision_id is None:
2775
# TODO:
2776
# everything to do - use pack logic
2777
# to fetch from all packs to one without
2778
# inventory parsing etc, IFF nothing to be copied is in the target.
2779
# till then:
2780
source_revision_ids = frozenset(self.source.all_revision_ids())
2781
revision_ids = source_revision_ids - \
2782
frozenset(self.target.get_parent_map(source_revision_ids))
2783
revision_keys = [(revid,) for revid in revision_ids]
2784
index = self.target._pack_collection.revision_index.combined_index
2785
present_revision_ids = set(item[1][0] for item in
2786
index.iter_entries(revision_keys))
2787
revision_ids = set(revision_ids) - present_revision_ids
2788
# implementing the TODO will involve:
2789
# - detecting when all of a pack is selected
2790
# - avoiding as much as possible pre-selection, so the
2791
# more-core routines such as create_pack_from_packs can filter in
2792
# a just-in-time fashion. (though having a HEADS list on a
2793
# repository might make this a lot easier, because we could
2794
# sensibly detect 'new revisions' without doing a full index scan.
2795
elif _mod_revision.is_null(revision_id):
2796
# nothing to do:
2797
return (0, [])
2798
else:
2799
try:
2800
revision_ids = self.search_missing_revision_ids(revision_id,
2801
find_ghosts=find_ghosts).get_keys()
2802
except errors.NoSuchRevision:
2803
raise errors.InstallFailed([revision_id])
2804
if len(revision_ids) == 0:
2805
return (0, [])
2806
packs = self.source._pack_collection.all_packs()
2807
pack = Packer(self.target._pack_collection, packs, '.fetch',
2808
revision_ids).pack()
2809
if pack is not None:
2810
self.target._pack_collection._save_pack_names()
2811
# Trigger an autopack. This may duplicate effort as we've just done
2812
# a pack creation, but for now it is simpler to think about as
2813
# 'upload data, then repack if needed'.
2814
self.target._pack_collection.autopack()
2815
return (pack.get_revision_count(), [])
2816
else:
2817
return (0, [])
2818
2819
@needs_read_lock
2820
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2821
"""See InterRepository.missing_revision_ids().
2822
2823
:param find_ghosts: Find ghosts throughout the ancestry of
2824
revision_id.
2825
"""
2826
if not find_ghosts and revision_id is not None:
2827
return self._walk_to_common_revisions([revision_id])
2828
elif revision_id is not None:
2829
# Find ghosts: search for revisions pointing from one repository to
2830
# the other, and vice versa, anywhere in the history of revision_id.
2831
graph = self.target.get_graph(other_repository=self.source)
2832
searcher = graph._make_breadth_first_searcher([revision_id])
2833
found_ids = set()
2834
while True:
2835
try:
2836
next_revs, ghosts = searcher.next_with_ghosts()
2837
except StopIteration:
2838
break
2839
if revision_id in ghosts:
2840
raise errors.NoSuchRevision(self.source, revision_id)
2841
found_ids.update(next_revs)
2842
found_ids.update(ghosts)
2843
found_ids = frozenset(found_ids)
2844
# Double query here: should be able to avoid this by changing the
2845
# graph api further.
2846
result_set = found_ids - frozenset(
2847
self.target.get_parent_map(found_ids))
2848
else:
2849
source_ids = self.source.all_revision_ids()
2850
# source_ids is the worst possible case we may need to pull.
2851
# now we want to filter source_ids against what we actually
2852
# have in target, but don't try to check for existence where we know
2853
# we do not have a revision as that would be pointless.
2854
target_ids = set(self.target.all_revision_ids())
2855
result_set = set(source_ids).difference(target_ids)
2856
return self.source.revision_ids_to_search_result(result_set)
2857
2858
2859
class InterModel1and2(InterRepository):
2860
2861
@classmethod
2862
def _get_repo_format_to_test(self):
2863
return None
2864
2865
@staticmethod
2866
def is_compatible(source, target):
2867
if not source.supports_rich_root() and target.supports_rich_root():
2868
return True
2869
else:
2870
return False
2871
2872
@needs_write_lock
2873
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2874
"""See InterRepository.fetch()."""
2875
from bzrlib.fetch import Model1toKnit2Fetcher
2876
f = Model1toKnit2Fetcher(to_repository=self.target,
2877
from_repository=self.source,
2878
last_revision=revision_id,
2879
pb=pb, find_ghosts=find_ghosts)
2880
return f.count_copied, f.failed_revisions
2881
2882
@needs_write_lock
2883
def copy_content(self, revision_id=None):
2884
"""Make a complete copy of the content in self into destination.
2885
2886
This is a destructive operation! Do not use it on existing
2887
repositories.
2888
2889
:param revision_id: Only copy the content needed to construct
2890
revision_id and its parents.
2891
"""
2892
try:
2893
self.target.set_make_working_trees(self.source.make_working_trees())
2894
except NotImplementedError:
2895
pass
2896
# but don't bother fetching if we have the needed data now.
2897
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2898
self.target.has_revision(revision_id)):
2899
return
2900
self.target.fetch(self.source, revision_id=revision_id)
2901
2902
2903
class InterKnit1and2(InterKnitRepo):
2904
2905
@classmethod
2906
def _get_repo_format_to_test(self):
2907
return None
2908
2909
@staticmethod
2910
def is_compatible(source, target):
2911
"""Be compatible with Knit1 source and Knit3 target"""
2912
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2913
try:
2914
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2915
RepositoryFormatKnit3)
2916
from bzrlib.repofmt.pack_repo import (
2917
RepositoryFormatKnitPack1,
2918
RepositoryFormatKnitPack3,
2919
RepositoryFormatPackDevelopment0,
2920
RepositoryFormatPackDevelopment0Subtree,
2921
)
2922
nosubtrees = (
2923
RepositoryFormatKnit1,
2924
RepositoryFormatKnitPack1,
2925
RepositoryFormatPackDevelopment0,
2926
)
2927
subtrees = (
2928
RepositoryFormatKnit3,
2929
RepositoryFormatKnitPack3,
2930
RepositoryFormatPackDevelopment0Subtree,
2931
)
2932
return (isinstance(source._format, nosubtrees) and
2933
isinstance(target._format, subtrees))
2934
except AttributeError:
2935
return False
2936
2937
@needs_write_lock
2938
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2939
"""See InterRepository.fetch()."""
2940
from bzrlib.fetch import Knit1to2Fetcher
2941
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2942
self.source, self.source._format, self.target,
2943
self.target._format)
2944
f = Knit1to2Fetcher(to_repository=self.target,
2945
from_repository=self.source,
2946
last_revision=revision_id,
2947
pb=pb, find_ghosts=find_ghosts)
2948
return f.count_copied, f.failed_revisions
2949
2950
2951
class InterDifferingSerializer(InterKnitRepo):
2952
2953
@classmethod
2954
def _get_repo_format_to_test(self):
2955
return None
2956
2957
@staticmethod
2958
def is_compatible(source, target):
2959
"""Be compatible with Knit2 source and Knit3 target"""
2960
if source.supports_rich_root() != target.supports_rich_root():
2961
return False
2962
# Ideally, we'd support fetching if the source had no tree references
2963
# even if it supported them...
2964
if (getattr(source, '_format.supports_tree_reference', False) and
2965
not getattr(target, '_format.supports_tree_reference', False)):
2966
return False
2967
return True
2968
2969
@needs_write_lock
2970
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2971
"""See InterRepository.fetch()."""
2972
revision_ids = self.target.search_missing_revision_ids(self.source,
2973
revision_id, find_ghosts=find_ghosts).get_keys()
2974
revision_ids = tsort.topo_sort(
2975
self.source.get_graph().get_parent_map(revision_ids))
2976
def revisions_iterator():
2977
for current_revision_id in revision_ids:
2978
revision = self.source.get_revision(current_revision_id)
2979
tree = self.source.revision_tree(current_revision_id)
2980
try:
2981
signature = self.source.get_signature_text(
2982
current_revision_id)
2983
except errors.NoSuchRevision:
2984
signature = None
2985
yield revision, tree, signature
2986
if pb is None:
2987
my_pb = ui.ui_factory.nested_progress_bar()
2988
pb = my_pb
2989
else:
2990
my_pb = None
2991
try:
2992
install_revisions(self.target, revisions_iterator(),
2993
len(revision_ids), pb)
2994
finally:
2995
if my_pb is not None:
2996
my_pb.finished()
2997
return len(revision_ids), 0
2998
2999
3000
class InterOtherToRemote(InterRepository):
3001
3002
def __init__(self, source, target):
3003
InterRepository.__init__(self, source, target)
3004
self._real_inter = None
3005
3006
@staticmethod
3007
def is_compatible(source, target):
3008
if isinstance(target, remote.RemoteRepository):
3009
return True
3010
return False
3011
3012
def _ensure_real_inter(self):
3013
if self._real_inter is None:
3014
self.target._ensure_real()
3015
real_target = self.target._real_repository
3016
self._real_inter = InterRepository.get(self.source, real_target)
3017
3018
def copy_content(self, revision_id=None):
3019
self._ensure_real_inter()
3020
self._real_inter.copy_content(revision_id=revision_id)
3021
3022
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3023
self._ensure_real_inter()
3024
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3025
find_ghosts=find_ghosts)
3026
3027
@classmethod
3028
def _get_repo_format_to_test(self):
3029
return None
3030
3031
3032
class InterRemoteToOther(InterRepository):
3033
3034
def __init__(self, source, target):
3035
InterRepository.__init__(self, source, target)
3036
self._real_inter = None
3037
3038
@staticmethod
3039
def is_compatible(source, target):
3040
if not isinstance(source, remote.RemoteRepository):
3041
return False
3042
# Is source's model compatible with target's model?
3043
source._ensure_real()
3044
real_source = source._real_repository
3045
if isinstance(real_source, remote.RemoteRepository):
3046
raise NotImplementedError(
3047
"We don't support remote repos backed by remote repos yet.")
3048
return InterRepository._same_model(real_source, target)
3049
3050
def _ensure_real_inter(self):
3051
if self._real_inter is None:
3052
self.source._ensure_real()
3053
real_source = self.source._real_repository
3054
self._real_inter = InterRepository.get(real_source, self.target)
3055
3056
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3057
self._ensure_real_inter()
3058
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3059
find_ghosts=find_ghosts)
3060
3061
def copy_content(self, revision_id=None):
3062
self._ensure_real_inter()
3063
self._real_inter.copy_content(revision_id=revision_id)
3064
3065
@classmethod
3066
def _get_repo_format_to_test(self):
3067
return None
3068
3069
3070
3071
InterRepository.register_optimiser(InterDifferingSerializer)
3072
InterRepository.register_optimiser(InterSameDataRepository)
1696
3073
InterRepository.register_optimiser(InterWeaveRepo)
1697
3074
InterRepository.register_optimiser(InterKnitRepo)
1698
1699
1700
class RepositoryTestProviderAdapter(object):
1701
"""A tool to generate a suite testing multiple repository formats at once.
1702
1703
This is done by copying the test once for each transport and injecting
1704
the transport_server, transport_readonly_server, and bzrdir_format and
1705
repository_format classes into each copy. Each copy is also given a new id()
1706
to make it easy to identify.
1707
"""
1708
1709
def __init__(self, transport_server, transport_readonly_server, formats):
1710
self._transport_server = transport_server
1711
self._transport_readonly_server = transport_readonly_server
1712
self._formats = formats
1713
1714
def adapt(self, test):
1715
result = TestSuite()
1716
for repository_format, bzrdir_format in self._formats:
1717
new_test = deepcopy(test)
1718
new_test.transport_server = self._transport_server
1719
new_test.transport_readonly_server = self._transport_readonly_server
1720
new_test.bzrdir_format = bzrdir_format
1721
new_test.repository_format = repository_format
1722
def make_new_test_id():
1723
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1724
return lambda: new_id
1725
new_test.id = make_new_test_id()
1726
result.addTest(new_test)
1727
return result
1728
1729
1730
class InterRepositoryTestProviderAdapter(object):
1731
"""A tool to generate a suite testing multiple inter repository formats.
1732
1733
This is done by copying the test once for each interrepo provider and injecting
1734
the transport_server, transport_readonly_server, repository_format and
1735
repository_to_format classes into each copy.
1736
Each copy is also given a new id() to make it easy to identify.
1737
"""
1738
1739
def __init__(self, transport_server, transport_readonly_server, formats):
1740
self._transport_server = transport_server
1741
self._transport_readonly_server = transport_readonly_server
1742
self._formats = formats
1743
1744
def adapt(self, test):
1745
result = TestSuite()
1746
for interrepo_class, repository_format, repository_format_to in self._formats:
1747
new_test = deepcopy(test)
1748
new_test.transport_server = self._transport_server
1749
new_test.transport_readonly_server = self._transport_readonly_server
1750
new_test.interrepo_class = interrepo_class
1751
new_test.repository_format = repository_format
1752
new_test.repository_format_to = repository_format_to
1753
def make_new_test_id():
1754
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1755
return lambda: new_id
1756
new_test.id = make_new_test_id()
1757
result.addTest(new_test)
1758
return result
1759
1760
@staticmethod
1761
def default_test_list():
1762
"""Generate the default list of interrepo permutations to test."""
1763
result = []
1764
# test the default InterRepository between format 6 and the current
1765
# default format.
1766
# XXX: robertc 20060220 reinstate this when there are two supported
1767
# formats which do not have an optimal code path between them.
1768
result.append((InterRepository,
1769
RepositoryFormat6(),
1770
RepositoryFormatKnit1()))
1771
for optimiser in InterRepository._optimisers:
1772
result.append((optimiser,
1773
optimiser._matching_repo_format,
1774
optimiser._matching_repo_format
1775
))
1776
# if there are specific combinations we want to use, we can add them
1777
# here.
1778
return result
3075
InterRepository.register_optimiser(InterModel1and2)
3076
InterRepository.register_optimiser(InterKnit1and2)
3077
InterRepository.register_optimiser(InterPackRepo)
3078
InterRepository.register_optimiser(InterOtherToRemote)
3079
InterRepository.register_optimiser(InterRemoteToOther)
1779
3080
1780
3081
1781
3082
class CopyConverter(object):
1807
3108
self.step('Moving repository to repository.backup')
1808
3109
self.repo_dir.transport.move('repository', 'repository.backup')
1809
3110
backup_transport = self.repo_dir.transport.clone('repository.backup')
3111
repo._format.check_conversion_target(self.target_format)
1810
3112
self.source_repo = repo._format.open(self.repo_dir,
1811
3113
_found=True,
1812
3114
_override_transport=backup_transport)
1829
3131
self.pb.update(message, self.count, self.total)
1830
3132
1831
3133
1832
# Copied from xml.sax.saxutils
3134
_unescape_map = {
3135
'apos':"'",
3136
'quot':'"',
3137
'amp':'&',
3138
'lt':'<',
3139
'gt':'>'
3140
}
3141
3142
3143
def _unescaper(match, _map=_unescape_map):
3144
code = match.group(1)
3145
try:
3146
return _map[code]
3147
except KeyError:
3148
if not code.startswith('#'):
3149
raise
3150
return unichr(int(code[1:])).encode('utf8')
3151
3152
3153
_unescape_re = None
3154
3155
1833
3156
def _unescape_xml(data):
1834
"""Unescape &, <, and > in a string of data.
1835
"""
1836
data = data.replace("<", "<")
1837
data = data.replace(">", ">")
1838
# must do ampersand last
1839
return data.replace("&", "&")
3157
"""Unescape predefined XML entities in a string of data."""
3158
global _unescape_re
3159
if _unescape_re is None:
3160
_unescape_re = re.compile('\&([^;]*);')
3161
return _unescape_re.sub(_unescaper, data)
3162
3163
3164
class _VersionedFileChecker(object):
3165
3166
def __init__(self, repository):
3167
self.repository = repository
3168
self.text_index = self.repository._generate_text_key_index()
3169
3170
def calculate_file_version_parents(self, text_key):
3171
"""Calculate the correct parents for a file version according to
3172
the inventories.
3173
"""
3174
parent_keys = self.text_index[text_key]
3175
if parent_keys == [_mod_revision.NULL_REVISION]:
3176
return ()
3177
return tuple(parent_keys)
3178
3179
def check_file_version_parents(self, texts, progress_bar=None):
3180
"""Check the parents stored in a versioned file are correct.
3181
3182
It also detects file versions that are not referenced by their
3183
corresponding revision's inventory.
3184
3185
:returns: A tuple of (wrong_parents, dangling_file_versions).
3186
wrong_parents is a dict mapping {revision_id: (stored_parents,
3187
correct_parents)} for each revision_id where the stored parents
3188
are not correct. dangling_file_versions is a set of (file_id,
3189
revision_id) tuples for versions that are present in this versioned
3190
file, but not used by the corresponding inventory.
3191
"""
3192
wrong_parents = {}
3193
self.file_ids = set([file_id for file_id, _ in
3194
self.text_index.iterkeys()])
3195
# text keys is now grouped by file_id
3196
n_weaves = len(self.file_ids)
3197
files_in_revisions = {}
3198
revisions_of_files = {}
3199
n_versions = len(self.text_index)
3200
progress_bar.update('loading text store', 0, n_versions)
3201
parent_map = self.repository.texts.get_parent_map(self.text_index)
3202
# On unlistable transports this could well be empty/error...
3203
text_keys = self.repository.texts.keys()
3204
unused_keys = frozenset(text_keys) - set(self.text_index)
3205
for num, key in enumerate(self.text_index.iterkeys()):
3206
if progress_bar is not None:
3207
progress_bar.update('checking text graph', num, n_versions)
3208
correct_parents = self.calculate_file_version_parents(key)
3209
try:
3210
knit_parents = parent_map[key]
3211
except errors.RevisionNotPresent:
3212
# Missing text!
3213
knit_parents = None
3214
if correct_parents != knit_parents:
3215
wrong_parents[key] = (knit_parents, correct_parents)
3216
return wrong_parents, unused_keys
3217
3218
3219
def _old_get_graph(repository, revision_id):
3220
"""DO NOT USE. That is all. I'm serious."""
3221
graph = repository.get_graph()
3222
revision_graph = dict(((key, value) for key, value in
3223
graph.iter_ancestry([revision_id]) if value is not None))
3224
return _strip_NULL_ghosts(revision_graph)
3225
3226
3227
def _strip_NULL_ghosts(revision_graph):
3228
"""Also don't use this. more compatibility code for unmigrated clients."""
3229
# Filter ghosts, and null:
3230
if _mod_revision.NULL_REVISION in revision_graph:
3231
del revision_graph[_mod_revision.NULL_REVISION]
3232
for key, parents in revision_graph.items():
3233
revision_graph[key] = tuple(parent for parent in parents if parent
3234
in revision_graph)
3235
return revision_graph
Loggerhead is a web-based interface for Breezy
Version: 2.0.1