« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 3697
- compare with another revision
- download diff
- download tarball
- view history from revision 3697
- Committer: Canonical.com Patch Queue Manager
- Date: 2008-09-09 17:33:48 UTC
- mfrom: (3688.3.3 jam-integration)
- Revision ID: pqm@pqm.ubuntu.com-20080909173348-5y4vo2tga9d19fpa
(Mark Hammond) Update bzr.ico, and include TBZR icons in installer.
- files added:
- bzrlib/_btree_serializer_c.pyx
- bzrlib/help_topics
- bzrlib/help_topics/en
- bzrlib/repofmt
- bzrlib/smart
- bzrlib/tests/commands
- bzrlib/tests/inventory_implementations
- bzrlib/tests/per_lock
- bzrlib/tests/per_repository_reference
- 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
- files removed:
- NEWS.developers
- bzrlib/bundle/old
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/revisionstore_implementations
- files renamed:
- bzrlib/patiencediff.py => bzrlib/_patiencediff_py.py
- bzrlib/graph.py => bzrlib/deprecated_graph.py
- doc/configuration.txt => bzrlib/help_topics/en/configuration.txt
- bzrlib/tests/blackbox/test_bundle.py => bzrlib/tests/blackbox/test_send.py
- bzrlib/tests/HTTPTestUtil.py => bzrlib/tests/http_utils.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/index.txt => doc/en/user-guide/index.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
1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
2
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
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
16
17
from binascii import hexlify
18
from copy import deepcopy
19
17
from cStringIO import StringIO
18
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
20
21
import re
21
22
import time
22
from unittest import TestSuite
23
24
from bzrlib import bzrdir, check, delta, gpg, errors, xml5, ui, transactions, osutils
23
24
from bzrlib import (
25
bzrdir,
26
check,
27
debug,
28
errors,
29
generate_ids,
30
gpg,
31
graph,
32
lazy_regex,
33
lockable_files,
34
lockdir,
35
lru_cache,
36
osutils,
37
registry,
38
remote,
39
revision as _mod_revision,
40
symbol_versioning,
41
transactions,
42
tsort,
43
ui,
44
)
45
from bzrlib.bundle import serializer
46
from bzrlib.revisiontree import RevisionTree
47
from bzrlib.store.versioned import VersionedFileStore
48
from bzrlib.store.text import TextStore
49
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
51
""")
52
25
53
from bzrlib.decorators import needs_read_lock, needs_write_lock
26
from bzrlib.errors import InvalidRevisionId
27
from bzrlib.graph import Graph
28
54
from bzrlib.inter import InterObject
29
55
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
30
from bzrlib.knit import KnitVersionedFile, KnitPlainFactory
31
from bzrlib.lockable_files import LockableFiles, TransportLock
32
from bzrlib.lockdir import LockDir
33
from bzrlib.osutils import (safe_unicode, rand_bytes, compact_date,
34
local_time_offset)
35
from bzrlib.revision import NULL_REVISION, Revision
36
from bzrlib.revisiontree import RevisionTree
37
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
38
from bzrlib.store.text import TextStore
39
from bzrlib import symbol_versioning
40
from bzrlib.symbol_versioning import (deprecated_method,
41
zero_nine,
56
from bzrlib.symbol_versioning import (
57
deprecated_method,
58
one_one,
59
one_two,
60
one_three,
61
one_six,
42
62
)
43
from bzrlib.testament import Testament
44
from bzrlib.trace import mutter, note, warning
45
from bzrlib.tsort import topo_sort
46
from bzrlib.weave import WeaveFile
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
47
64
48
65
49
66
# Old formats display a warning, but only once
50
67
_deprecation_warning_done = False
51
68
52
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
450
53
451
class Repository(object):
54
452
"""Repository holding history for one or more branches.
55
453
57
455
revisions and file history. It's normally accessed only by the Branch,
58
456
which views a particular line of development through that history.
59
457
60
The Repository builds on top of Stores and a Transport, which respectively
61
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
62
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.
63
495
"""
64
496
65
@needs_write_lock
66
def add_inventory(self, revid, inv, parents):
67
"""Add the inventory inv to the repository as revid.
68
69
: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
70
559
is known to have and are in the repository already.
71
560
72
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.
73
563
"""
74
assert inv.revision_id is None or inv.revision_id == revid, \
75
"Mismatch between inventory revision" \
76
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
77
assert inv.root is not None
78
inv_text = xml5.serializer_v5.write_inventory_to_string(inv)
79
inv_sha1 = osutils.sha_string(inv_text)
80
inv_vf = self.control_weaves.get_weave('inventory',
81
self.get_transaction())
82
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
83
return inv_sha1
84
85
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
86
final_parents = []
87
for parent in parents:
88
if parent in inv_vf:
89
final_parents.append(parent)
90
91
inv_vf.add_lines(revid, final_parents, lines)
92
93
@needs_write_lock
94
def add_revision(self, rev_id, rev, inv=None, config=None):
95
"""Add rev to the revision store as rev_id.
96
97
: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.
98
589
:param rev: The revision object.
99
590
:param inv: The inventory for the revision. if None, it will be looked
100
591
up in the inventory storer
102
593
If supplied its signature_needed method will be used
103
594
to determine if a signature should be made.
104
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)
105
599
if config is not None and config.signature_needed():
106
600
if inv is None:
107
inv = self.get_inventory(rev_id)
601
inv = self.get_inventory(revision_id)
108
602
plaintext = Testament(rev, inv).as_short_text()
109
603
self.store_revision_signature(
110
gpg.GPGStrategy(config), plaintext, rev_id)
111
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,)]):
112
607
if inv is None:
113
raise errors.WeaveRevisionNotPresent(rev_id,
114
self.get_inventory_weave())
608
raise errors.WeaveRevisionNotPresent(revision_id,
609
self.inventories)
115
610
else:
116
611
# yes, this is not suitable for adding with ghosts.
117
self.add_inventory(rev_id, inv, rev.parent_ids)
118
self._revision_store.add_revision(rev, self.get_transaction())
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)
119
618
120
@needs_read_lock
121
def _all_possible_ids(self):
122
"""Return all the possible revisions that we could find."""
123
return self.get_inventory_weave().versions()
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))
124
624
125
625
def all_revision_ids(self):
126
626
"""Returns a list of all the revision ids in the repository.
127
627
128
This is deprecated because code should generally work on the graph
129
reachable from a particular revision, and ignore any other revisions
130
that might be present. There is no direct replacement method.
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.
131
632
"""
633
if 'evil' in debug.debug_flags:
634
mutter_callsite(2, "all_revision_ids is linear with history.")
132
635
return self._all_revision_ids()
133
636
134
@needs_read_lock
135
637
def _all_revision_ids(self):
136
638
"""Returns a list of all the revision ids in the repository.
137
639
138
640
These are in as much topological order as the underlying store can
139
present: for weaves ghosts may lead to a lack of correctness until
140
the reweave updates the parents list.
641
present.
141
642
"""
142
if self._revision_store.text_store.listable():
143
return self._revision_store.all_revision_ids(self.get_transaction())
144
result = self._all_possible_ids()
145
return self._eliminate_revisions_not_present(result)
643
raise NotImplementedError(self._all_revision_ids)
146
644
147
645
def break_lock(self):
148
646
"""Break a lock if one is present from another instance.
159
657
Returns a set of the present revisions.
160
658
"""
161
659
result = []
162
for id in revision_ids:
163
if self.has_revision(id):
164
result.append(id)
165
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()
166
664
167
665
@staticmethod
168
666
def create(a_bzrdir):
169
667
"""Construct the current default format repository in a_bzrdir."""
170
668
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
171
669
172
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
670
def __init__(self, _format, a_bzrdir, control_files):
173
671
"""instantiate a Repository.
174
672
175
673
:param _format: The format of the repository on disk.
184
682
# the following are part of the public API for Repository:
185
683
self.bzrdir = a_bzrdir
186
684
self.control_files = control_files
187
self._revision_store = _revision_store
188
self.text_store = text_store
189
# backwards compatibility
190
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
191
691
# not right yet - should be more semantically clear ?
192
692
#
193
self.control_store = control_store
194
self.control_weaves = control_store
195
693
# TODO: make sure to construct the right store classes, etc, depending
196
694
# on whether escaping is required.
197
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
198
711
199
712
def __repr__(self):
200
return '%s(%r)' % (self.__class__.__name__,
201
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
202
733
203
734
def is_locked(self):
204
735
return self.control_files.is_locked()
205
736
206
def lock_write(self):
207
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
208
769
209
770
def lock_read(self):
210
771
self.control_files.lock_read()
772
for repo in self._fallback_repositories:
773
repo.lock_read()
774
self._refresh_data()
211
775
212
776
def get_physical_lock_status(self):
213
777
return self.control_files.get_physical_lock_status()
214
778
215
@needs_read_lock
216
def missing_revision_ids(self, other, revision_id=None):
217
"""Return the revision ids that other has that this does not.
218
219
These are returned in topological order.
220
221
revision_id: only return revision ids included by revision_id.
222
"""
223
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)
224
919
225
920
@staticmethod
226
921
def open(base):
232
927
control = bzrdir.BzrDir.open(base)
233
928
return control.open_repository()
234
929
235
def copy_content_into(self, destination, revision_id=None, basis=None):
930
def copy_content_into(self, destination, revision_id=None):
236
931
"""Make a complete copy of the content in self into destination.
237
932
238
933
This is a destructive operation! Do not use it on existing
239
934
repositories.
240
935
"""
241
return InterRepository.get(self, destination).copy_content(revision_id, basis)
242
243
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):
244
961
"""Fetch the content required to construct revision_id from source.
245
962
246
963
If revision_id is None all content is copied.
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).
247
967
"""
248
return InterRepository.get(source, self).fetch(revision_id=revision_id,
249
pb=pb)
250
251
def get_commit_builder(self, branch, parents, config, timestamp=None,
252
timezone=None, committer=None, revprops=None,
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,
253
988
revision_id=None):
254
989
"""Obtain a CommitBuilder for this repository.
255
990
262
997
:param revprops: Optional dictionary of revision properties.
263
998
:param revision_id: Optional revision id.
264
999
"""
265
return _CommitBuilder(self, parents, config, timestamp, timezone,
266
committer, revprops, revision_id)
1000
result = self._commit_builder_class(self, parents, config,
1001
timestamp, timezone, committer, revprops, revision_id)
1002
self.start_write_group()
1003
return result
267
1004
268
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.')
269
1013
self.control_files.unlock()
1014
for repo in self._fallback_repositories:
1015
repo.unlock()
270
1016
271
1017
@needs_read_lock
272
def clone(self, a_bzrdir, revision_id=None, basis=None):
1018
def clone(self, a_bzrdir, revision_id=None):
273
1019
"""Clone this repository into a_bzrdir using the current format.
274
1020
275
1021
Currently no check is made that the format of this repository and
276
1022
the bzrdir format are compatible. FIXME RBC 20060201.
277
"""
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):
278
1072
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
279
1073
# use target default format.
280
result = a_bzrdir.create_repository()
281
# FIXME RBC 20060209 split out the repository type to avoid this check ?
282
elif isinstance(a_bzrdir._format,
283
(bzrdir.BzrDirFormat4,
284
bzrdir.BzrDirFormat5,
285
bzrdir.BzrDirFormat6)):
286
result = a_bzrdir.open_repository()
1074
dest_repo = a_bzrdir.create_repository()
287
1075
else:
288
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
289
self.copy_content_into(result, revision_id, basis)
290
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
291
1083
292
1084
@needs_read_lock
293
1085
def has_revision(self, revision_id):
294
1086
"""True if this repository has a copy of the revision."""
295
return self._revision_store.has_revision_id(revision_id,
296
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]
297
1108
298
1109
@needs_read_lock
299
1110
def get_revision_reconcile(self, revision_id):
304
1115
be used by reconcile, or reconcile-alike commands that are correcting
305
1116
or testing the revision graph.
306
1117
"""
307
if not revision_id or not isinstance(revision_id, basestring):
308
raise InvalidRevisionId(revision_id=revision_id, branch=self)
309
return self._revision_store.get_revisions([revision_id],
310
self.get_transaction())[0]
1118
return self._get_revisions([revision_id])[0]
1119
311
1120
@needs_read_lock
312
1121
def get_revisions(self, revision_ids):
313
return self._revision_store.get_revisions(revision_ids,
314
self.get_transaction())
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]
315
1141
316
1142
@needs_read_lock
317
1143
def get_revision_xml(self, revision_id):
318
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)
319
1148
rev_tmp = StringIO()
320
1149
# the current serializer..
321
self._revision_store._serializer.write_revision(rev, rev_tmp)
1150
self._serializer.write_revision(rev, rev_tmp)
322
1151
rev_tmp.seek(0)
323
1152
return rev_tmp.getvalue()
324
1153
325
@needs_read_lock
326
def get_revision(self, revision_id):
327
"""Return the Revision object for a named revision"""
328
r = self.get_revision_reconcile(revision_id)
329
# weave corruption can lead to absent revision markers that should be
330
# present.
331
# the following test is reasonably cheap (it needs a single weave read)
332
# and the weave is cached in read transactions. In write transactions
333
# it is not cached but typically we only read a small number of
334
# revisions. For knits when they are introduced we will probably want
335
# to ensure that caching write transactions are in use.
336
inv = self.get_inventory_weave()
337
self._check_revision_parents(r, inv)
338
return r
339
340
@needs_read_lock
341
1154
def get_deltas_for_revisions(self, revisions):
342
1155
"""Produce a generator of revision deltas.
343
1156
353
1166
t in self.revision_trees(required_trees))
354
1167
for revision in revisions:
355
1168
if not revision.parent_ids:
356
old_tree = self.revision_tree(None)
1169
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
357
1170
else:
358
1171
old_tree = trees[revision.parent_ids[0]]
359
1172
yield trees[revision.revision_id].changes_from(old_tree)
368
1181
r = self.get_revision(revision_id)
369
1182
return list(self.get_deltas_for_revisions([r]))[0]
370
1183
371
def _check_revision_parents(self, revision, inventory):
372
"""Private to Repository and Fetch.
373
374
This checks the parentage of revision in an inventory weave for
375
consistency and is only applicable to inventory-weave-for-ancestry
376
using repository formats & fetchers.
377
"""
378
weave_parents = inventory.get_parents(revision.revision_id)
379
weave_names = inventory.versions()
380
for parent_id in revision.parent_ids:
381
if parent_id in weave_names:
382
# this parent must not be a ghost.
383
if not parent_id in weave_parents:
384
# but it is a ghost
385
raise errors.CorruptRepository(self)
386
387
1184
@needs_write_lock
388
1185
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
389
1186
signature = gpg_strategy.sign(plaintext)
390
self._revision_store.add_revision_signature_text(revision_id,
391
signature,
392
self.get_transaction())
393
394
def fileids_altered_by_revision_ids(self, revision_ids):
395
"""Find the file ids and versions affected by revisions.
396
397
:param revisions: an iterable containing revision ids.
398
: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
399
1198
revision_ids. Each altered file-ids has the exact revision_ids that
400
1199
altered it listed explicitly.
401
"""
402
assert isinstance(self._format, (RepositoryFormat5,
403
RepositoryFormat6,
404
RepositoryFormat7,
405
RepositoryFormatKnit1)), \
406
("fileids_altered_by_revision_ids only supported for branches "
407
"which store inventory as unnested xml, not on %r" % self)
408
selected_revision_ids = set(revision_ids)
409
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)
410
1232
result = {}
411
1233
412
1234
# this code needs to read every new line in every inventory for the
417
1239
# revisions. We don't need to see all lines in the inventory because
418
1240
# only those added in an inventory in rev X can contain a revision=X
419
1241
# line.
420
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
421
start = line.find('file_id="')+9
422
if start < 9: continue
423
end = line.find('"', start)
424
assert end>= 0
425
file_id = _unescape_xml(line[start:end])
426
427
start = line.find('revision="')+10
428
if start < 10: continue
429
end = line.find('"', start)
430
assert end>= 0
431
revision_id = _unescape_xml(line[start:end])
432
if revision_id in selected_revision_ids:
433
result.setdefault(file_id, set()).add(revision_id)
434
return result
435
436
@needs_read_lock
437
def get_inventory_weave(self):
438
return self.control_weaves.get_weave('inventory',
439
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)
440
1522
441
1523
@needs_read_lock
442
1524
def get_inventory(self, revision_id):
443
"""Get Inventory object by hash."""
444
return self.deserialise_inventory(
445
revision_id, self.get_inventory_xml(revision_id))
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]
446
1558
447
1559
def deserialise_inventory(self, revision_id, xml):
448
1560
"""Transform the xml into an inventory object.
450
1562
:param revision_id: The expected revision id of the inventory.
451
1563
:param xml: A serialised inventory.
452
1564
"""
453
result = xml5.serializer_v5.read_inventory_from_string(xml)
454
result.root.revision = revision_id
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))
455
1569
return result
456
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
1579
457
1580
@needs_read_lock
458
1581
def get_inventory_xml(self, revision_id):
459
1582
"""Get inventory XML as a file object."""
1583
texts = self._iter_inventory_xmls([revision_id])
460
1584
try:
461
assert isinstance(revision_id, basestring), type(revision_id)
462
iw = self.get_inventory_weave()
463
return iw.get_text(revision_id)
464
except IndexError:
1585
text, revision_id = texts.next()
1586
except StopIteration:
465
1587
raise errors.HistoryMissing(self, 'inventory', revision_id)
1588
return text
466
1589
467
1590
@needs_read_lock
468
1591
def get_inventory_sha1(self, revision_id):
470
1593
"""
471
1594
return self.get_revision(revision_id).inventory_sha1
472
1595
473
@needs_read_lock
474
def get_revision_graph(self, revision_id=None):
475
"""Return a dictionary containing the revision graph.
476
477
:param revision_id: The revision_id to get a graph from. If None, then
478
the entire revision graph is returned. This is a deprecated mode of
479
operation and will be removed in the future.
480
:return: a dictionary of revision_id->revision_parents_list.
481
"""
482
# special case NULL_REVISION
483
if revision_id == NULL_REVISION:
484
return {}
485
weave = self.get_inventory_weave()
486
all_revisions = self._eliminate_revisions_not_present(weave.versions())
487
entire_graph = dict([(node, weave.get_parents(node)) for
488
node in all_revisions])
489
if revision_id is None:
490
return entire_graph
491
elif revision_id not in entire_graph:
492
raise errors.NoSuchRevision(self, revision_id)
493
else:
494
# add what can be reached from revision_id
495
result = {}
496
pending = set([revision_id])
497
while len(pending) > 0:
498
node = pending.pop()
499
result[node] = entire_graph[node]
500
for revision_id in result[node]:
501
if revision_id not in result:
502
pending.add(revision_id)
503
return result
504
505
@needs_read_lock
506
def get_revision_graph_with_ghosts(self, revision_ids=None):
507
"""Return a graph of the revisions with ghosts marked as applicable.
508
509
:param revision_ids: an iterable of revisions to graph or None for all.
510
:return: a Graph object with the graph reachable from revision_ids.
511
"""
512
result = Graph()
513
if not revision_ids:
514
pending = set(self.all_revision_ids())
515
required = set([])
516
else:
517
pending = set(revision_ids)
518
# special case NULL_REVISION
519
if NULL_REVISION in pending:
520
pending.remove(NULL_REVISION)
521
required = set(pending)
522
done = set([])
523
while len(pending):
524
revision_id = pending.pop()
525
try:
526
rev = self.get_revision(revision_id)
527
except errors.NoSuchRevision:
528
if revision_id in required:
529
raise
530
# a ghost
531
result.add_ghost(revision_id)
532
continue
533
for parent_id in rev.parent_ids:
534
# is this queued or done ?
535
if (parent_id not in pending and
536
parent_id not in done):
537
# no, queue it.
538
pending.add(parent_id)
539
result.add_node(revision_id, rev.parent_ids)
540
done.add(revision_id)
541
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]
542
1618
543
1619
@needs_read_lock
544
1620
def get_revision_inventory(self, revision_id):
556
1632
else:
557
1633
return self.get_inventory(revision_id)
558
1634
559
@needs_read_lock
560
1635
def is_shared(self):
561
1636
"""Return True if this repository is flagged as a shared repository."""
562
1637
raise NotImplementedError(self.is_shared)
568
1643
reconciler = RepoReconciler(self, thorough=thorough)
569
1644
reconciler.reconcile()
570
1645
return reconciler
571
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
572
1657
@needs_read_lock
573
1658
def revision_tree(self, revision_id):
574
1659
"""Return Tree for a revision on this branch.
575
1660
576
`revision_id` may be None for the empty tree revision.
1661
`revision_id` may be NULL_REVISION for the empty tree revision.
577
1662
"""
1663
revision_id = _mod_revision.ensure_null(revision_id)
578
1664
# TODO: refactor this to use an existing revision object
579
1665
# so we don't need to read it in twice.
580
if revision_id is None or revision_id == NULL_REVISION:
581
return RevisionTree(self, Inventory(), NULL_REVISION)
1666
if revision_id == _mod_revision.NULL_REVISION:
1667
return RevisionTree(self, Inventory(root_id=None),
1668
_mod_revision.NULL_REVISION)
582
1669
else:
583
1670
inv = self.get_revision_inventory(revision_id)
584
1671
return RevisionTree(self, inv, revision_id)
585
1672
586
@needs_read_lock
587
1673
def revision_trees(self, revision_ids):
588
1674
"""Return Tree for a revision on this branch.
589
1675
590
1676
`revision_id` may not be None or 'null:'"""
591
assert None not in revision_ids
592
assert NULL_REVISION not in revision_ids
593
texts = self.get_inventory_weave().get_texts(revision_ids)
594
for text, revision_id in zip(texts, revision_ids):
595
inv = self.deserialise_inventory(revision_id, text)
596
yield RevisionTree(self, inv, revision_id)
1677
inventories = self.iter_inventories(revision_ids)
1678
for inv in inventories:
1679
yield RevisionTree(self, inv, inv.revision_id)
597
1680
598
1681
@needs_read_lock
599
def get_ancestry(self, revision_id):
1682
def get_ancestry(self, revision_id, topo_sorted=True):
600
1683
"""Return a list of revision-ids integrated by a revision.
601
1684
602
1685
The first element of the list is always None, indicating the origin
605
1688
606
1689
This is topologically sorted.
607
1690
"""
608
if revision_id is None:
1691
if _mod_revision.is_null(revision_id):
609
1692
return [None]
610
1693
if not self.has_revision(revision_id):
611
1694
raise errors.NoSuchRevision(self, revision_id)
612
w = self.get_inventory_weave()
613
candidates = w.get_ancestry(revision_id)
614
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1695
graph = self.get_graph()
1696
keys = set()
1697
search = graph._make_breadth_first_searcher([revision_id])
1698
while True:
1699
try:
1700
found, ghosts = search.next_with_ghosts()
1701
except StopIteration:
1702
break
1703
keys.update(found)
1704
if _mod_revision.NULL_REVISION in keys:
1705
keys.remove(_mod_revision.NULL_REVISION)
1706
if topo_sorted:
1707
parent_map = graph.get_parent_map(keys)
1708
keys = tsort.topo_sort(parent_map)
1709
return [None] + list(keys)
1710
1711
def pack(self):
1712
"""Compress the data within the repository.
1713
1714
This operation only makes sense for some repository types. For other
1715
types it should be a no-op that just returns.
1716
1717
This stub method does not require a lock, but subclasses should use
1718
@needs_write_lock as this is a long running call its reasonable to
1719
implicitly lock for the user.
1720
"""
615
1721
616
1722
@needs_read_lock
1723
@deprecated_method(one_six)
617
1724
def print_file(self, file, revision_id):
618
1725
"""Print `file` to stdout.
619
1726
634
1741
def get_transaction(self):
635
1742
return self.control_files.get_transaction()
636
1743
637
def revision_parents(self, revid):
638
return self.get_inventory_weave().parent_names(revid)
1744
@deprecated_method(one_one)
1745
def get_parents(self, revision_ids):
1746
"""See StackedParentsProvider.get_parents"""
1747
parent_map = self.get_parent_map(revision_ids)
1748
return [parent_map.get(r, None) for r in revision_ids]
1749
1750
def get_parent_map(self, revision_ids):
1751
"""See graph._StackedParentsProvider.get_parent_map"""
1752
# revisions index works in keys; this just works in revisions
1753
# therefore wrap and unwrap
1754
query_keys = []
1755
result = {}
1756
for revision_id in revision_ids:
1757
if revision_id == _mod_revision.NULL_REVISION:
1758
result[revision_id] = ()
1759
elif revision_id is None:
1760
raise ValueError('get_parent_map(None) is not valid')
1761
else:
1762
query_keys.append((revision_id ,))
1763
for ((revision_id,), parent_keys) in \
1764
self.revisions.get_parent_map(query_keys).iteritems():
1765
if parent_keys:
1766
result[revision_id] = tuple(parent_revid
1767
for (parent_revid,) in parent_keys)
1768
else:
1769
result[revision_id] = (_mod_revision.NULL_REVISION,)
1770
return result
1771
1772
def _make_parents_provider(self):
1773
return self
1774
1775
def get_graph(self, other_repository=None):
1776
"""Return the graph walker for this repository format"""
1777
parents_provider = self._make_parents_provider()
1778
if (other_repository is not None and
1779
not self.has_same_location(other_repository)):
1780
parents_provider = graph._StackedParentsProvider(
1781
[parents_provider, other_repository._make_parents_provider()])
1782
return graph.Graph(parents_provider)
1783
1784
def _get_versioned_file_checker(self):
1785
"""Return an object suitable for checking versioned files."""
1786
return _VersionedFileChecker(self)
1787
1788
def revision_ids_to_search_result(self, result_set):
1789
"""Convert a set of revision ids to a graph SearchResult."""
1790
result_parents = set()
1791
for parents in self.get_graph().get_parent_map(
1792
result_set).itervalues():
1793
result_parents.update(parents)
1794
included_keys = result_set.intersection(result_parents)
1795
start_keys = result_set.difference(included_keys)
1796
exclude_keys = result_parents.difference(result_set)
1797
result = graph.SearchResult(start_keys, exclude_keys,
1798
len(result_set), result_set)
1799
return result
639
1800
640
1801
@needs_write_lock
641
1802
def set_make_working_trees(self, new_value):
661
1822
@needs_read_lock
662
1823
def has_signature_for_revision_id(self, revision_id):
663
1824
"""Query for a revision signature for revision_id in the repository."""
664
return self._revision_store.has_signature(revision_id,
665
self.get_transaction())
1825
if not self.has_revision(revision_id):
1826
raise errors.NoSuchRevision(self, revision_id)
1827
sig_present = (1 == len(
1828
self.signatures.get_parent_map([(revision_id,)])))
1829
return sig_present
666
1830
667
1831
@needs_read_lock
668
1832
def get_signature_text(self, revision_id):
669
1833
"""Return the text for a signature."""
670
return self._revision_store.get_signature_text(revision_id,
671
self.get_transaction())
1834
stream = self.signatures.get_record_stream([(revision_id,)],
1835
'unordered', True)
1836
record = stream.next()
1837
if record.storage_kind == 'absent':
1838
raise errors.NoSuchRevision(self, revision_id)
1839
return record.get_bytes_as('fulltext')
672
1840
673
1841
@needs_read_lock
674
def check(self, revision_ids):
1842
def check(self, revision_ids=None):
675
1843
"""Check consistency of all history of given revision_ids.
676
1844
677
1845
Different repository implementations should override _check().
679
1847
:param revision_ids: A non-empty list of revision_ids whose ancestry
680
1848
will be checked. Typically the last revision_id of a branch.
681
1849
"""
682
if not revision_ids:
683
raise ValueError("revision_ids must be non-empty in %s.check"
684
% (self,))
685
1850
return self._check(revision_ids)
686
1851
687
1852
def _check(self, revision_ids):
697
1862
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
698
1863
% (self._format, self.bzrdir.transport.base))
699
1864
700
701
class AllInOneRepository(Repository):
702
"""Legacy support - the repository behaviour for all-in-one branches."""
703
704
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
705
# we reuse one control files instance.
706
dir_mode = a_bzrdir._control_files._dir_mode
707
file_mode = a_bzrdir._control_files._file_mode
708
709
def get_store(name, compressed=True, prefixed=False):
710
# FIXME: This approach of assuming stores are all entirely compressed
711
# or entirely uncompressed is tidy, but breaks upgrade from
712
# some existing branches where there's a mixture; we probably
713
# still want the option to look for both.
714
relpath = a_bzrdir._control_files._escape(name)
715
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
716
prefixed=prefixed, compressed=compressed,
717
dir_mode=dir_mode,
718
file_mode=file_mode)
719
#if self._transport.should_cache():
720
# cache_path = os.path.join(self.cache_root, name)
721
# os.mkdir(cache_path)
722
# store = bzrlib.store.CachedStore(store, cache_path)
723
return store
724
725
# not broken out yet because the controlweaves|inventory_store
726
# and text_store | weave_store bits are still different.
727
if isinstance(_format, RepositoryFormat4):
728
# cannot remove these - there is still no consistent api
729
# which allows access to this old info.
730
self.inventory_store = get_store('inventory-store')
731
text_store = get_store('text-store')
732
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
733
734
@needs_read_lock
735
def is_shared(self):
736
"""AllInOne repositories cannot be shared."""
737
return False
738
739
@needs_write_lock
740
def set_make_working_trees(self, new_value):
741
"""Set the policy flag for making working trees when creating branches.
742
743
This only applies to branches that use this repository.
744
745
The default is 'True'.
746
:param new_value: True to restore the default, False to disable making
747
working trees.
1865
def supports_rich_root(self):
1866
return self._format.rich_root_data
1867
1868
def _check_ascii_revisionid(self, revision_id, method):
1869
"""Private helper for ascii-only repositories."""
1870
# weave repositories refuse to store revisionids that are non-ascii.
1871
if revision_id is not None:
1872
# weaves require ascii revision ids.
1873
if isinstance(revision_id, unicode):
1874
try:
1875
revision_id.encode('ascii')
1876
except UnicodeEncodeError:
1877
raise errors.NonAsciiRevisionId(method, self)
1878
else:
1879
try:
1880
revision_id.decode('ascii')
1881
except UnicodeDecodeError:
1882
raise errors.NonAsciiRevisionId(method, self)
1883
1884
def revision_graph_can_have_wrong_parents(self):
1885
"""Is it possible for this repository to have a revision graph with
1886
incorrect parents?
1887
1888
If True, then this repository must also implement
1889
_find_inconsistent_revision_parents so that check and reconcile can
1890
check for inconsistencies before proceeding with other checks that may
1891
depend on the revision index being consistent.
748
1892
"""
749
raise NotImplementedError(self.set_make_working_trees)
750
751
def make_working_trees(self):
752
"""Returns the policy for making working trees on new branches."""
753
return True
1893
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1894
1895
1896
# remove these delegates a while after bzr 0.15
1897
def __make_delegated(name, from_module):
1898
def _deprecated_repository_forwarder():
1899
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1900
% (name, from_module),
1901
DeprecationWarning,
1902
stacklevel=2)
1903
m = __import__(from_module, globals(), locals(), [name])
1904
try:
1905
return getattr(m, name)
1906
except AttributeError:
1907
raise AttributeError('module %s has no name %s'
1908
% (m, name))
1909
globals()[name] = _deprecated_repository_forwarder
1910
1911
for _name in [
1912
'AllInOneRepository',
1913
'WeaveMetaDirRepository',
1914
'PreSplitOutRepositoryFormat',
1915
'RepositoryFormat4',
1916
'RepositoryFormat5',
1917
'RepositoryFormat6',
1918
'RepositoryFormat7',
1919
]:
1920
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1921
1922
for _name in [
1923
'KnitRepository',
1924
'RepositoryFormatKnit',
1925
'RepositoryFormatKnit1',
1926
]:
1927
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
754
1928
755
1929
756
1930
def install_revision(repository, rev, revision_tree):
757
1931
"""Install all revision data into a repository."""
1932
install_revisions(repository, [(rev, revision_tree, None)])
1933
1934
1935
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1936
"""Install all revision data into a repository.
1937
1938
Accepts an iterable of revision, tree, signature tuples. The signature
1939
may be None.
1940
"""
1941
repository.start_write_group()
1942
try:
1943
for n, (revision, revision_tree, signature) in enumerate(iterable):
1944
_install_revision(repository, revision, revision_tree, signature)
1945
if pb is not None:
1946
pb.update('Transferring revisions', n + 1, num_revisions)
1947
except:
1948
repository.abort_write_group()
1949
raise
1950
else:
1951
repository.commit_write_group()
1952
1953
1954
def _install_revision(repository, rev, revision_tree, signature):
1955
"""Install all revision data into a repository."""
758
1956
present_parents = []
759
1957
parent_trees = {}
760
1958
for p_id in rev.parent_ids:
762
1960
present_parents.append(p_id)
763
1961
parent_trees[p_id] = repository.revision_tree(p_id)
764
1962
else:
765
parent_trees[p_id] = repository.revision_tree(None)
1963
parent_trees[p_id] = repository.revision_tree(
1964
_mod_revision.NULL_REVISION)
766
1965
767
1966
inv = revision_tree.inventory
768
769
# backwards compatability hack: skip the root id.
770
1967
entries = inv.iter_entries()
771
entries.next()
1968
# backwards compatibility hack: skip the root id.
1969
if not repository.supports_rich_root():
1970
path, root = entries.next()
1971
if root.revision != rev.revision_id:
1972
raise errors.IncompatibleRevision(repr(repository))
1973
text_keys = {}
1974
for path, ie in entries:
1975
text_keys[(ie.file_id, ie.revision)] = ie
1976
text_parent_map = repository.texts.get_parent_map(text_keys)
1977
missing_texts = set(text_keys) - set(text_parent_map)
772
1978
# Add the texts that are not already present
773
for path, ie in entries:
774
w = repository.weave_store.get_weave_or_empty(ie.file_id,
775
repository.get_transaction())
776
if ie.revision not in w:
777
text_parents = []
778
# FIXME: TODO: The following loop *may* be overlapping/duplicate
779
# with InventoryEntry.find_previous_heads(). if it is, then there
780
# is a latent bug here where the parents may have ancestors of each
781
# other. RBC, AB
782
for revision, tree in parent_trees.iteritems():
783
if ie.file_id not in tree:
784
continue
785
parent_id = tree.inventory[ie.file_id].revision
786
if parent_id in text_parents:
787
continue
788
text_parents.append(parent_id)
789
790
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
791
repository.get_transaction())
792
lines = revision_tree.get_file(ie.file_id).readlines()
793
vfile.add_lines(rev.revision_id, text_parents, lines)
1979
for text_key in missing_texts:
1980
ie = text_keys[text_key]
1981
text_parents = []
1982
# FIXME: TODO: The following loop overlaps/duplicates that done by
1983
# commit to determine parents. There is a latent/real bug here where
1984
# the parents inserted are not those commit would do - in particular
1985
# they are not filtered by heads(). RBC, AB
1986
for revision, tree in parent_trees.iteritems():
1987
if ie.file_id not in tree:
1988
continue
1989
parent_id = tree.inventory[ie.file_id].revision
1990
if parent_id in text_parents:
1991
continue
1992
text_parents.append((ie.file_id, parent_id))
1993
lines = revision_tree.get_file(ie.file_id).readlines()
1994
repository.texts.add_lines(text_key, text_parents, lines)
794
1995
try:
795
1996
# install the inventory
796
1997
repository.add_inventory(rev.revision_id, inv, present_parents)
797
1998
except errors.RevisionAlreadyPresent:
798
1999
pass
2000
if signature is not None:
2001
repository.add_signature_text(rev.revision_id, signature)
799
2002
repository.add_revision(rev.revision_id, rev, inv)
800
2003
801
2004
802
2005
class MetaDirRepository(Repository):
803
"""Repositories in the new meta-dir layout."""
804
805
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
806
super(MetaDirRepository, self).__init__(_format,
807
a_bzrdir,
808
control_files,
809
_revision_store,
810
control_store,
811
text_store)
812
dir_mode = self.control_files._dir_mode
813
file_mode = self.control_files._file_mode
814
815
@needs_read_lock
2006
"""Repositories in the new meta-dir layout.
2007
2008
:ivar _transport: Transport for access to repository control files,
2009
typically pointing to .bzr/repository.
2010
"""
2011
2012
def __init__(self, _format, a_bzrdir, control_files):
2013
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2014
self._transport = control_files._transport
2015
816
2016
def is_shared(self):
817
2017
"""Return True if this repository is flagged as a shared repository."""
818
return self.control_files._transport.has('shared-storage')
2018
return self._transport.has('shared-storage')
819
2019
820
2020
@needs_write_lock
821
2021
def set_make_working_trees(self, new_value):
829
2029
"""
830
2030
if new_value:
831
2031
try:
832
self.control_files._transport.delete('no-working-trees')
2032
self._transport.delete('no-working-trees')
833
2033
except errors.NoSuchFile:
834
2034
pass
835
2035
else:
836
self.control_files.put_utf8('no-working-trees', '')
2036
self._transport.put_bytes('no-working-trees', '',
2037
mode=self.bzrdir._get_file_mode())
837
2038
838
2039
def make_working_trees(self):
839
2040
"""Returns the policy for making working trees on new branches."""
840
return not self.control_files._transport.has('no-working-trees')
841
842
843
class KnitRepository(MetaDirRepository):
844
"""Knit format repository."""
845
846
def _warn_if_deprecated(self):
847
# This class isn't deprecated
848
pass
849
850
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
851
inv_vf.add_lines_with_ghosts(revid, parents, lines)
852
853
@needs_read_lock
854
def _all_revision_ids(self):
855
"""See Repository.all_revision_ids()."""
856
# Knits get the revision graph from the index of the revision knit, so
857
# it's always possible even if they're on an unlistable transport.
858
return self._revision_store.all_revision_ids(self.get_transaction())
859
860
def fileid_involved_between_revs(self, from_revid, to_revid):
861
"""Find file_id(s) which are involved in the changes between revisions.
862
863
This determines the set of revisions which are involved, and then
864
finds all file ids affected by those revisions.
865
"""
866
vf = self._get_revision_vf()
867
from_set = set(vf.get_ancestry(from_revid))
868
to_set = set(vf.get_ancestry(to_revid))
869
changed = to_set.difference(from_set)
870
return self._fileid_involved_by_set(changed)
871
872
def fileid_involved(self, last_revid=None):
873
"""Find all file_ids modified in the ancestry of last_revid.
874
875
:param last_revid: If None, last_revision() will be used.
876
"""
877
if not last_revid:
878
changed = set(self.all_revision_ids())
879
else:
880
changed = set(self.get_ancestry(last_revid))
881
if None in changed:
882
changed.remove(None)
883
return self._fileid_involved_by_set(changed)
884
885
@needs_read_lock
886
def get_ancestry(self, revision_id):
887
"""Return a list of revision-ids integrated by a revision.
888
889
This is topologically sorted.
890
"""
891
if revision_id is None:
892
return [None]
893
vf = self._get_revision_vf()
894
try:
895
return [None] + vf.get_ancestry(revision_id)
896
except errors.RevisionNotPresent:
897
raise errors.NoSuchRevision(self, revision_id)
898
899
@needs_read_lock
900
def get_revision(self, revision_id):
901
"""Return the Revision object for a named revision"""
902
return self.get_revision_reconcile(revision_id)
903
904
@needs_read_lock
905
def get_revision_graph(self, revision_id=None):
906
"""Return a dictionary containing the revision graph.
907
908
:param revision_id: The revision_id to get a graph from. If None, then
909
the entire revision graph is returned. This is a deprecated mode of
910
operation and will be removed in the future.
911
:return: a dictionary of revision_id->revision_parents_list.
912
"""
913
# special case NULL_REVISION
914
if revision_id == NULL_REVISION:
915
return {}
916
weave = self._get_revision_vf()
917
entire_graph = weave.get_graph()
918
if revision_id is None:
919
return weave.get_graph()
920
elif revision_id not in weave:
921
raise errors.NoSuchRevision(self, revision_id)
922
else:
923
# add what can be reached from revision_id
924
result = {}
925
pending = set([revision_id])
926
while len(pending) > 0:
927
node = pending.pop()
928
result[node] = weave.get_parents(node)
929
for revision_id in result[node]:
930
if revision_id not in result:
931
pending.add(revision_id)
932
return result
933
934
@needs_read_lock
935
def get_revision_graph_with_ghosts(self, revision_ids=None):
936
"""Return a graph of the revisions with ghosts marked as applicable.
937
938
:param revision_ids: an iterable of revisions to graph or None for all.
939
:return: a Graph object with the graph reachable from revision_ids.
940
"""
941
result = Graph()
942
vf = self._get_revision_vf()
943
versions = set(vf.versions())
944
if not revision_ids:
945
pending = set(self.all_revision_ids())
946
required = set([])
947
else:
948
pending = set(revision_ids)
949
# special case NULL_REVISION
950
if NULL_REVISION in pending:
951
pending.remove(NULL_REVISION)
952
required = set(pending)
953
done = set([])
954
while len(pending):
955
revision_id = pending.pop()
956
if not revision_id in versions:
957
if revision_id in required:
958
raise errors.NoSuchRevision(self, revision_id)
959
# a ghost
960
result.add_ghost(revision_id)
961
# mark it as done so we don't try for it again.
962
done.add(revision_id)
963
continue
964
parent_ids = vf.get_parents_with_ghosts(revision_id)
965
for parent_id in parent_ids:
966
# is this queued or done ?
967
if (parent_id not in pending and
968
parent_id not in done):
969
# no, queue it.
970
pending.add(parent_id)
971
result.add_node(revision_id, parent_ids)
972
done.add(revision_id)
973
return result
974
975
def _get_revision_vf(self):
976
""":return: a versioned file containing the revisions."""
977
vf = self._revision_store.get_revision_file(self.get_transaction())
978
return vf
979
980
@needs_write_lock
981
def reconcile(self, other=None, thorough=False):
982
"""Reconcile this repository."""
983
from bzrlib.reconcile import KnitReconciler
984
reconciler = KnitReconciler(self, thorough=thorough)
985
reconciler.reconcile()
986
return reconciler
2041
return not self._transport.has('no-working-trees')
2042
2043
2044
class MetaDirVersionedFileRepository(MetaDirRepository):
2045
"""Repositories in a meta-dir, that work via versioned file objects."""
2046
2047
def __init__(self, _format, a_bzrdir, control_files):
2048
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2049
control_files)
2050
2051
2052
class RepositoryFormatRegistry(registry.Registry):
2053
"""Registry of RepositoryFormats."""
2054
2055
def get(self, format_string):
2056
r = registry.Registry.get(self, format_string)
2057
if callable(r):
2058
r = r()
2059
return r
987
2060
988
def revision_parents(self, revision_id):
989
return self._get_revision_vf().get_parents(revision_id)
990
2061
2062
format_registry = RepositoryFormatRegistry()
2063
"""Registry of formats, indexed by their identifying format string.
2064
2065
This can contain either format instances themselves, or classes/factories that
2066
can be called to obtain one.
2067
"""
2068
2069
2070
#####################################################################
2071
# Repository Formats
991
2072
992
2073
class RepositoryFormat(object):
993
2074
"""A repository format.
998
2079
children.
999
2080
* an open routine which returns a Repository instance.
1000
2081
2082
There is one and only one Format subclass for each on-disk format. But
2083
there can be one Repository subclass that is used for several different
2084
formats. The _format attribute on a Repository instance can be used to
2085
determine the disk format.
2086
1001
2087
Formats are placed in an dict by their format string for reference
1002
2088
during opening. These should be subclasses of RepositoryFormat
1003
2089
for consistency.
1010
2096
_matchingbzrdir - the bzrdir format that the repository format was
1011
2097
originally written to work with. This can be used if manually
1012
2098
constructing a bzrdir and repository, or more commonly for test suite
1013
parameterisation.
2099
parameterization.
1014
2100
"""
1015
2101
1016
_default_format = None
1017
"""The default format used for new repositories."""
1018
1019
_formats = {}
1020
"""The known formats."""
2102
# Set to True or False in derived classes. True indicates that the format
2103
# supports ghosts gracefully.
2104
supports_ghosts = None
2105
# Can this repository be given external locations to lookup additional
2106
# data. Set to True or False in derived classes.
2107
supports_external_lookups = None
1021
2108
1022
2109
def __str__(self):
1023
2110
return "<%s>" % self.__class__.__name__
1024
2111
2112
def __eq__(self, other):
2113
# format objects are generally stateless
2114
return isinstance(other, self.__class__)
2115
2116
def __ne__(self, other):
2117
return not self == other
2118
1025
2119
@classmethod
1026
2120
def find_format(klass, a_bzrdir):
1027
"""Return the format for the repository object in a_bzrdir."""
2121
"""Return the format for the repository object in a_bzrdir.
2122
2123
This is used by bzr native formats that have a "format" file in
2124
the repository. Other methods may be used by different types of
2125
control directory.
2126
"""
1028
2127
try:
1029
2128
transport = a_bzrdir.get_repository_transport(None)
1030
2129
format_string = transport.get("format").read()
1031
return klass._formats[format_string]
2130
return format_registry.get(format_string)
1032
2131
except errors.NoSuchFile:
1033
2132
raise errors.NoRepositoryPresent(a_bzrdir)
1034
2133
except KeyError:
1035
raise errors.UnknownFormatError(format=format_string)
1036
1037
def _get_control_store(self, repo_transport, control_files):
1038
"""Return the control store for this repository."""
1039
raise NotImplementedError(self._get_control_store)
2134
raise errors.UnknownFormatError(format=format_string,
2135
kind='repository')
2136
2137
@classmethod
2138
def register_format(klass, format):
2139
format_registry.register(format.get_format_string(), format)
2140
2141
@classmethod
2142
def unregister_format(klass, format):
2143
format_registry.remove(format.get_format_string())
1040
2144
1041
2145
@classmethod
1042
2146
def get_default_format(klass):
1043
2147
"""Return the current default format."""
1044
return klass._default_format
2148
from bzrlib import bzrdir
2149
return bzrdir.format_registry.make_bzrdir('default').repository_format
1045
2150
1046
2151
def get_format_string(self):
1047
2152
"""Return the ASCII format string that identifies this format.
1055
2160
"""Return the short description for this format."""
1056
2161
raise NotImplementedError(self.get_format_description)
1057
2162
1058
def _get_revision_store(self, repo_transport, control_files):
1059
"""Return the revision store object for this a_bzrdir."""
1060
raise NotImplementedError(self._get_revision_store)
1061
1062
def _get_text_rev_store(self,
1063
transport,
1064
control_files,
1065
name,
1066
compressed=True,
1067
prefixed=False,
1068
serializer=None):
1069
"""Common logic for getting a revision store for a repository.
1070
1071
see self._get_revision_store for the subclass-overridable method to
1072
get the store for a repository.
1073
"""
1074
from bzrlib.store.revision.text import TextRevisionStore
1075
dir_mode = control_files._dir_mode
1076
file_mode = control_files._file_mode
1077
text_store =TextStore(transport.clone(name),
1078
prefixed=prefixed,
1079
compressed=compressed,
1080
dir_mode=dir_mode,
1081
file_mode=file_mode)
1082
_revision_store = TextRevisionStore(text_store, serializer)
1083
return _revision_store
1084
2163
# TODO: this shouldn't be in the base class, it's specific to things that
2164
# use weaves or knits -- mbp 20070207
1085
2165
def _get_versioned_file_store(self,
1086
2166
name,
1087
2167
transport,
1088
2168
control_files,
1089
2169
prefixed=True,
1090
versionedfile_class=WeaveFile,
2170
versionedfile_class=None,
2171
versionedfile_kwargs={},
1091
2172
escaped=False):
2173
if versionedfile_class is None:
2174
versionedfile_class = self._versionedfile_class
1092
2175
weave_transport = control_files._transport.clone(name)
1093
2176
dir_mode = control_files._dir_mode
1094
2177
file_mode = control_files._file_mode
1096
2179
dir_mode=dir_mode,
1097
2180
file_mode=file_mode,
1098
2181
versionedfile_class=versionedfile_class,
2182
versionedfile_kwargs=versionedfile_kwargs,
1099
2183
escaped=escaped)
1100
2184
1101
2185
def initialize(self, a_bzrdir, shared=False):
1103
2187
1104
2188
:param a_bzrdir: The bzrdir to put the new repository in it.
1105
2189
:param shared: The repository should be initialized as a sharable one.
1106
2190
:returns: The new repository object.
2191
1107
2192
This may raise UninitializableFormat if shared repository are not
1108
2193
compatible the a_bzrdir.
1109
2194
"""
2195
raise NotImplementedError(self.initialize)
1110
2196
1111
2197
def is_supported(self):
1112
2198
"""Is this format supported?
1117
2203
"""
1118
2204
return True
1119
2205
2206
def check_conversion_target(self, target_format):
2207
raise NotImplementedError(self.check_conversion_target)
2208
1120
2209
def open(self, a_bzrdir, _found=False):
1121
2210
"""Return an instance of this format for the bzrdir a_bzrdir.
1122
2211
1124
2213
"""
1125
2214
raise NotImplementedError(self.open)
1126
2215
1127
@classmethod
1128
def register_format(klass, format):
1129
klass._formats[format.get_format_string()] = format
1130
1131
@classmethod
1132
def set_default_format(klass, format):
1133
klass._default_format = format
1134
1135
@classmethod
1136
def unregister_format(klass, format):
1137
assert klass._formats[format.get_format_string()] is format
1138
del klass._formats[format.get_format_string()]
1139
1140
1141
class PreSplitOutRepositoryFormat(RepositoryFormat):
1142
"""Base class for the pre split out repository formats."""
1143
1144
def initialize(self, a_bzrdir, shared=False, _internal=False):
1145
"""Create a weave repository.
1146
1147
TODO: when creating split out bzr branch formats, move this to a common
1148
base for Format5, Format6. or something like that.
1149
"""
1150
from bzrlib.weavefile import write_weave_v5
1151
from bzrlib.weave import Weave
1152
1153
if shared:
1154
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1155
1156
if not _internal:
1157
# always initialized when the bzrdir is.
1158
return self.open(a_bzrdir, _found=True)
1159
1160
# Create an empty weave
1161
sio = StringIO()
1162
write_weave_v5(Weave(), sio)
1163
empty_weave = sio.getvalue()
1164
1165
mutter('creating repository in %s.', a_bzrdir.transport.base)
1166
dirs = ['revision-store', 'weaves']
1167
files = [('inventory.weave', StringIO(empty_weave)),
1168
]
1169
1170
# FIXME: RBC 20060125 don't peek under the covers
1171
# NB: no need to escape relative paths that are url safe.
1172
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1173
TransportLock)
1174
control_files.create_lock()
1175
control_files.lock_write()
1176
control_files._transport.mkdir_multi(dirs,
1177
mode=control_files._dir_mode)
1178
try:
1179
for file, content in files:
1180
control_files.put(file, content)
1181
finally:
1182
control_files.unlock()
1183
return self.open(a_bzrdir, _found=True)
1184
1185
def _get_control_store(self, repo_transport, control_files):
1186
"""Return the control store for this repository."""
1187
return self._get_versioned_file_store('',
1188
repo_transport,
1189
control_files,
1190
prefixed=False)
1191
1192
def _get_text_store(self, transport, control_files):
1193
"""Get a store for file texts for this format."""
1194
raise NotImplementedError(self._get_text_store)
1195
1196
def open(self, a_bzrdir, _found=False):
1197
"""See RepositoryFormat.open()."""
1198
if not _found:
1199
# we are being called directly and must probe.
1200
raise NotImplementedError
1201
1202
repo_transport = a_bzrdir.get_repository_transport(None)
1203
control_files = a_bzrdir._control_files
1204
text_store = self._get_text_store(repo_transport, control_files)
1205
control_store = self._get_control_store(repo_transport, control_files)
1206
_revision_store = self._get_revision_store(repo_transport, control_files)
1207
return AllInOneRepository(_format=self,
1208
a_bzrdir=a_bzrdir,
1209
_revision_store=_revision_store,
1210
control_store=control_store,
1211
text_store=text_store)
1212
1213
1214
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1215
"""Bzr repository format 4.
1216
1217
This repository format has:
1218
- flat stores
1219
- TextStores for texts, inventories,revisions.
1220
1221
This format is deprecated: it indexes texts using a text id which is
1222
removed in format 5; initialization and write support for this format
1223
has been removed.
1224
"""
1225
1226
def __init__(self):
1227
super(RepositoryFormat4, self).__init__()
1228
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1229
1230
def get_format_description(self):
1231
"""See RepositoryFormat.get_format_description()."""
1232
return "Repository format 4"
1233
1234
def initialize(self, url, shared=False, _internal=False):
1235
"""Format 4 branches cannot be created."""
1236
raise errors.UninitializableFormat(self)
1237
1238
def is_supported(self):
1239
"""Format 4 is not supported.
1240
1241
It is not supported because the model changed from 4 to 5 and the
1242
conversion logic is expensive - so doing it on the fly was not
1243
feasible.
1244
"""
1245
return False
1246
1247
def _get_control_store(self, repo_transport, control_files):
1248
"""Format 4 repositories have no formal control store at this point.
1249
1250
This will cause any control-file-needing apis to fail - this is desired.
1251
"""
1252
return None
1253
1254
def _get_revision_store(self, repo_transport, control_files):
1255
"""See RepositoryFormat._get_revision_store()."""
1256
from bzrlib.xml4 import serializer_v4
1257
return self._get_text_rev_store(repo_transport,
1258
control_files,
1259
'revision-store',
1260
serializer=serializer_v4)
1261
1262
def _get_text_store(self, transport, control_files):
1263
"""See RepositoryFormat._get_text_store()."""
1264
1265
1266
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1267
"""Bzr control format 5.
1268
1269
This repository format has:
1270
- weaves for file texts and inventory
1271
- flat stores
1272
- TextStores for revisions and signatures.
1273
"""
1274
1275
def __init__(self):
1276
super(RepositoryFormat5, self).__init__()
1277
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1278
1279
def get_format_description(self):
1280
"""See RepositoryFormat.get_format_description()."""
1281
return "Weave repository format 5"
1282
1283
def _get_revision_store(self, repo_transport, control_files):
1284
"""See RepositoryFormat._get_revision_store()."""
1285
"""Return the revision store object for this a_bzrdir."""
1286
return self._get_text_rev_store(repo_transport,
1287
control_files,
1288
'revision-store',
1289
compressed=False)
1290
1291
def _get_text_store(self, transport, control_files):
1292
"""See RepositoryFormat._get_text_store()."""
1293
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1294
1295
1296
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1297
"""Bzr control format 6.
1298
1299
This repository format has:
1300
- weaves for file texts and inventory
1301
- hash subdirectory based stores.
1302
- TextStores for revisions and signatures.
1303
"""
1304
1305
def __init__(self):
1306
super(RepositoryFormat6, self).__init__()
1307
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1308
1309
def get_format_description(self):
1310
"""See RepositoryFormat.get_format_description()."""
1311
return "Weave repository format 6"
1312
1313
def _get_revision_store(self, repo_transport, control_files):
1314
"""See RepositoryFormat._get_revision_store()."""
1315
return self._get_text_rev_store(repo_transport,
1316
control_files,
1317
'revision-store',
1318
compressed=False,
1319
prefixed=True)
1320
1321
def _get_text_store(self, transport, control_files):
1322
"""See RepositoryFormat._get_text_store()."""
1323
return self._get_versioned_file_store('weaves', transport, control_files)
1324
1325
2216
1326
2217
class MetaDirRepositoryFormat(RepositoryFormat):
1327
2218
"""Common base class for the new repositories using the metadir layout."""
1328
2219
2220
rich_root_data = False
2221
supports_tree_reference = False
2222
supports_external_lookups = False
2223
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2224
1329
2225
def __init__(self):
1330
2226
super(MetaDirRepositoryFormat, self).__init__()
1331
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1332
2227
1333
2228
def _create_control_files(self, a_bzrdir):
1334
2229
"""Create the required files and the initial control_files object."""
1335
2230
# FIXME: RBC 20060125 don't peek under the covers
1336
2231
# NB: no need to escape relative paths that are url safe.
1337
2232
repository_transport = a_bzrdir.get_repository_transport(self)
1338
control_files = LockableFiles(repository_transport, 'lock', LockDir)
2233
control_files = lockable_files.LockableFiles(repository_transport,
2234
'lock', lockdir.LockDir)
1339
2235
control_files.create_lock()
1340
2236
return control_files
1341
2237
1343
2239
"""Upload the initial blank content."""
1344
2240
control_files = self._create_control_files(a_bzrdir)
1345
2241
control_files.lock_write()
2242
transport = control_files._transport
2243
if shared == True:
2244
utf8_files += [('shared-storage', '')]
1346
2245
try:
1347
control_files._transport.mkdir_multi(dirs,
1348
mode=control_files._dir_mode)
1349
for file, content in files:
1350
control_files.put(file, content)
1351
for file, content in utf8_files:
1352
control_files.put_utf8(file, content)
1353
if shared == True:
1354
control_files.put_utf8('shared-storage', '')
2246
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2247
for (filename, content_stream) in files:
2248
transport.put_file(filename, content_stream,
2249
mode=a_bzrdir._get_file_mode())
2250
for (filename, content_bytes) in utf8_files:
2251
transport.put_bytes_non_atomic(filename, content_bytes,
2252
mode=a_bzrdir._get_file_mode())
1355
2253
finally:
1356
2254
control_files.unlock()
1357
2255
1358
2256
1359
class RepositoryFormat7(MetaDirRepositoryFormat):
1360
"""Bzr repository 7.
1361
1362
This repository format has:
1363
- weaves for file texts and inventory
1364
- hash subdirectory based stores.
1365
- TextStores for revisions and signatures.
1366
- a format marker of its own
1367
- an optional 'shared-storage' flag
1368
- an optional 'no-working-trees' flag
1369
"""
1370
1371
def _get_control_store(self, repo_transport, control_files):
1372
"""Return the control store for this repository."""
1373
return self._get_versioned_file_store('',
1374
repo_transport,
1375
control_files,
1376
prefixed=False)
1377
1378
def get_format_string(self):
1379
"""See RepositoryFormat.get_format_string()."""
1380
return "Bazaar-NG Repository format 7"
1381
1382
def get_format_description(self):
1383
"""See RepositoryFormat.get_format_description()."""
1384
return "Weave repository format 7"
1385
1386
def _get_revision_store(self, repo_transport, control_files):
1387
"""See RepositoryFormat._get_revision_store()."""
1388
return self._get_text_rev_store(repo_transport,
1389
control_files,
1390
'revision-store',
1391
compressed=False,
1392
prefixed=True,
1393
)
1394
1395
def _get_text_store(self, transport, control_files):
1396
"""See RepositoryFormat._get_text_store()."""
1397
return self._get_versioned_file_store('weaves',
1398
transport,
1399
control_files)
1400
1401
def initialize(self, a_bzrdir, shared=False):
1402
"""Create a weave repository.
1403
1404
:param shared: If true the repository will be initialized as a shared
1405
repository.
1406
"""
1407
from bzrlib.weavefile import write_weave_v5
1408
from bzrlib.weave import Weave
1409
1410
# Create an empty weave
1411
sio = StringIO()
1412
write_weave_v5(Weave(), sio)
1413
empty_weave = sio.getvalue()
1414
1415
mutter('creating repository in %s.', a_bzrdir.transport.base)
1416
dirs = ['revision-store', 'weaves']
1417
files = [('inventory.weave', StringIO(empty_weave)),
1418
]
1419
utf8_files = [('format', self.get_format_string())]
1420
1421
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1422
return self.open(a_bzrdir=a_bzrdir, _found=True)
1423
1424
def open(self, a_bzrdir, _found=False, _override_transport=None):
1425
"""See RepositoryFormat.open().
1426
1427
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1428
repository at a slightly different url
1429
than normal. I.e. during 'upgrade'.
1430
"""
1431
if not _found:
1432
format = RepositoryFormat.find_format(a_bzrdir)
1433
assert format.__class__ == self.__class__
1434
if _override_transport is not None:
1435
repo_transport = _override_transport
1436
else:
1437
repo_transport = a_bzrdir.get_repository_transport(None)
1438
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1439
text_store = self._get_text_store(repo_transport, control_files)
1440
control_store = self._get_control_store(repo_transport, control_files)
1441
_revision_store = self._get_revision_store(repo_transport, control_files)
1442
return MetaDirRepository(_format=self,
1443
a_bzrdir=a_bzrdir,
1444
control_files=control_files,
1445
_revision_store=_revision_store,
1446
control_store=control_store,
1447
text_store=text_store)
1448
1449
1450
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1451
"""Bzr repository knit format 1.
1452
1453
This repository format has:
1454
- knits for file texts and inventory
1455
- hash subdirectory based stores.
1456
- knits for revisions and signatures
1457
- TextStores for revisions and signatures.
1458
- a format marker of its own
1459
- an optional 'shared-storage' flag
1460
- an optional 'no-working-trees' flag
1461
- a LockDir lock
1462
1463
This format was introduced in bzr 0.8.
1464
"""
1465
1466
def _get_control_store(self, repo_transport, control_files):
1467
"""Return the control store for this repository."""
1468
return VersionedFileStore(
1469
repo_transport,
1470
prefixed=False,
1471
file_mode=control_files._file_mode,
1472
versionedfile_class=KnitVersionedFile,
1473
versionedfile_kwargs={'factory':KnitPlainFactory()},
1474
)
1475
1476
def get_format_string(self):
1477
"""See RepositoryFormat.get_format_string()."""
1478
return "Bazaar-NG Knit Repository Format 1"
1479
1480
def get_format_description(self):
1481
"""See RepositoryFormat.get_format_description()."""
1482
return "Knit repository format 1"
1483
1484
def _get_revision_store(self, repo_transport, control_files):
1485
"""See RepositoryFormat._get_revision_store()."""
1486
from bzrlib.store.revision.knit import KnitRevisionStore
1487
versioned_file_store = VersionedFileStore(
1488
repo_transport,
1489
file_mode=control_files._file_mode,
1490
prefixed=False,
1491
precious=True,
1492
versionedfile_class=KnitVersionedFile,
1493
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1494
escaped=True,
1495
)
1496
return KnitRevisionStore(versioned_file_store)
1497
1498
def _get_text_store(self, transport, control_files):
1499
"""See RepositoryFormat._get_text_store()."""
1500
return self._get_versioned_file_store('knits',
1501
transport,
1502
control_files,
1503
versionedfile_class=KnitVersionedFile,
1504
escaped=True)
1505
1506
def initialize(self, a_bzrdir, shared=False):
1507
"""Create a knit format 1 repository.
1508
1509
:param a_bzrdir: bzrdir to contain the new repository; must already
1510
be initialized.
1511
:param shared: If true the repository will be initialized as a shared
1512
repository.
1513
"""
1514
mutter('creating repository in %s.', a_bzrdir.transport.base)
1515
dirs = ['revision-store', 'knits']
1516
files = []
1517
utf8_files = [('format', self.get_format_string())]
1518
1519
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1520
repo_transport = a_bzrdir.get_repository_transport(None)
1521
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1522
control_store = self._get_control_store(repo_transport, control_files)
1523
transaction = transactions.WriteTransaction()
1524
# trigger a write of the inventory store.
1525
control_store.get_weave_or_empty('inventory', transaction)
1526
_revision_store = self._get_revision_store(repo_transport, control_files)
1527
_revision_store.has_revision_id('A', transaction)
1528
_revision_store.get_signature_file(transaction)
1529
return self.open(a_bzrdir=a_bzrdir, _found=True)
1530
1531
def open(self, a_bzrdir, _found=False, _override_transport=None):
1532
"""See RepositoryFormat.open().
1533
1534
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1535
repository at a slightly different url
1536
than normal. I.e. during 'upgrade'.
1537
"""
1538
if not _found:
1539
format = RepositoryFormat.find_format(a_bzrdir)
1540
assert format.__class__ == self.__class__
1541
if _override_transport is not None:
1542
repo_transport = _override_transport
1543
else:
1544
repo_transport = a_bzrdir.get_repository_transport(None)
1545
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1546
text_store = self._get_text_store(repo_transport, control_files)
1547
control_store = self._get_control_store(repo_transport, control_files)
1548
_revision_store = self._get_revision_store(repo_transport, control_files)
1549
return KnitRepository(_format=self,
1550
a_bzrdir=a_bzrdir,
1551
control_files=control_files,
1552
_revision_store=_revision_store,
1553
control_store=control_store,
1554
text_store=text_store)
1555
1556
1557
2257
# formats which have no format string are not discoverable
1558
# and not independently creatable, so are not registered.
1559
RepositoryFormat.register_format(RepositoryFormat7())
1560
_default_format = RepositoryFormatKnit1()
1561
RepositoryFormat.register_format(_default_format)
1562
RepositoryFormat.set_default_format(_default_format)
1563
_legacy_formats = [RepositoryFormat4(),
1564
RepositoryFormat5(),
1565
RepositoryFormat6()]
2258
# and not independently creatable, so are not registered. They're
2259
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2260
# needed, it's constructed directly by the BzrDir. Non-native formats where
2261
# the repository is not separately opened are similar.
2262
2263
format_registry.register_lazy(
2264
'Bazaar-NG Repository format 7',
2265
'bzrlib.repofmt.weaverepo',
2266
'RepositoryFormat7'
2267
)
2268
2269
format_registry.register_lazy(
2270
'Bazaar-NG Knit Repository Format 1',
2271
'bzrlib.repofmt.knitrepo',
2272
'RepositoryFormatKnit1',
2273
)
2274
2275
format_registry.register_lazy(
2276
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2277
'bzrlib.repofmt.knitrepo',
2278
'RepositoryFormatKnit3',
2279
)
2280
2281
format_registry.register_lazy(
2282
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2283
'bzrlib.repofmt.knitrepo',
2284
'RepositoryFormatKnit4',
2285
)
2286
2287
# Pack-based formats. There is one format for pre-subtrees, and one for
2288
# post-subtrees to allow ease of testing.
2289
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2290
format_registry.register_lazy(
2291
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2292
'bzrlib.repofmt.pack_repo',
2293
'RepositoryFormatKnitPack1',
2294
)
2295
format_registry.register_lazy(
2296
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2297
'bzrlib.repofmt.pack_repo',
2298
'RepositoryFormatKnitPack3',
2299
)
2300
format_registry.register_lazy(
2301
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2302
'bzrlib.repofmt.pack_repo',
2303
'RepositoryFormatKnitPack4',
2304
)
2305
format_registry.register_lazy(
2306
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2307
'bzrlib.repofmt.pack_repo',
2308
'RepositoryFormatKnitPack5',
2309
)
2310
format_registry.register_lazy(
2311
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2312
'bzrlib.repofmt.pack_repo',
2313
'RepositoryFormatKnitPack5RichRoot',
2314
)
2315
format_registry.register_lazy(
2316
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2317
'bzrlib.repofmt.pack_repo',
2318
'RepositoryFormatKnitPack5RichRootBroken',
2319
)
2320
2321
# Development formats.
2322
# 1.5->1.6
2323
format_registry.register_lazy(
2324
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2325
'bzrlib.repofmt.pack_repo',
2326
'RepositoryFormatPackDevelopment1',
2327
)
2328
format_registry.register_lazy(
2329
("Bazaar development format 1 with subtree support "
2330
"(needs bzr.dev from before 1.6)\n"),
2331
'bzrlib.repofmt.pack_repo',
2332
'RepositoryFormatPackDevelopment1Subtree',
2333
)
2334
# 1.6->1.7 go below here
1566
2335
1567
2336
1568
2337
class InterRepository(InterObject):
1577
2346
InterRepository.get(other).method_name(parameters).
1578
2347
"""
1579
2348
1580
_optimisers = set()
2349
_optimisers = []
1581
2350
"""The available optimised InterRepository types."""
1582
2351
1583
@needs_write_lock
1584
def copy_content(self, revision_id=None, basis=None):
1585
"""Make a complete copy of the content in self into destination.
1586
1587
This is a destructive operation! Do not use it on existing
1588
repositories.
1589
1590
:param revision_id: Only copy the content needed to construct
1591
revision_id and its parents.
1592
:param basis: Copy the needed data preferentially from basis.
1593
"""
1594
try:
1595
self.target.set_make_working_trees(self.source.make_working_trees())
1596
except NotImplementedError:
1597
pass
1598
# grab the basis available data
1599
if basis is not None:
1600
self.target.fetch(basis, revision_id=revision_id)
1601
# but don't bother fetching if we have the needed data now.
1602
if (revision_id not in (None, NULL_REVISION) and
1603
self.target.has_revision(revision_id)):
1604
return
1605
self.target.fetch(self.source, revision_id=revision_id)
1606
1607
@needs_write_lock
1608
def fetch(self, revision_id=None, pb=None):
2352
def copy_content(self, revision_id=None):
2353
raise NotImplementedError(self.copy_content)
2354
2355
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1609
2356
"""Fetch the content required to construct revision_id.
1610
2357
1611
The content is copied from source to target.
2358
The content is copied from self.source to self.target.
1612
2359
1613
2360
:param revision_id: if None all content is copied, if NULL_REVISION no
1614
2361
content is copied.
1615
2362
:param pb: optional progress bar to use for progress reports. If not
1616
2363
provided a default one will be created.
1617
2364
1618
Returns the copied revision count and the failed revisions in a tuple:
1619
(copied, failures).
1620
"""
1621
from bzrlib.fetch import GenericRepoFetcher
1622
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1623
self.source, self.source._format, self.target, self.target._format)
1624
f = GenericRepoFetcher(to_repository=self.target,
1625
from_repository=self.source,
1626
last_revision=revision_id,
1627
pb=pb)
1628
return f.count_copied, f.failed_revisions
1629
2365
:returns: (copied_revision_count, failures).
2366
"""
2367
# Normally we should find a specific InterRepository subclass to do
2368
# the fetch; if nothing else then at least InterSameDataRepository.
2369
# If none of them is suitable it looks like fetching is not possible;
2370
# we try to give a good message why. _assert_same_model will probably
2371
# give a helpful message; otherwise a generic one.
2372
self._assert_same_model(self.source, self.target)
2373
raise errors.IncompatibleRepositories(self.source, self.target,
2374
"no suitableInterRepository found")
2375
2376
def _walk_to_common_revisions(self, revision_ids):
2377
"""Walk out from revision_ids in source to revisions target has.
2378
2379
:param revision_ids: The start point for the search.
2380
:return: A set of revision ids.
2381
"""
2382
target_graph = self.target.get_graph()
2383
revision_ids = frozenset(revision_ids)
2384
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2385
return graph.SearchResult(revision_ids, set(), 0, set())
2386
missing_revs = set()
2387
source_graph = self.source.get_graph()
2388
# ensure we don't pay silly lookup costs.
2389
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2390
null_set = frozenset([_mod_revision.NULL_REVISION])
2391
while True:
2392
try:
2393
next_revs, ghosts = searcher.next_with_ghosts()
2394
except StopIteration:
2395
break
2396
if revision_ids.intersection(ghosts):
2397
absent_ids = set(revision_ids.intersection(ghosts))
2398
# If all absent_ids are present in target, no error is needed.
2399
absent_ids.difference_update(
2400
set(target_graph.get_parent_map(absent_ids)))
2401
if absent_ids:
2402
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2403
# we don't care about other ghosts as we can't fetch them and
2404
# haven't been asked to.
2405
next_revs = set(next_revs)
2406
# we always have NULL_REVISION present.
2407
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2408
missing_revs.update(next_revs - have_revs)
2409
searcher.stop_searching_any(have_revs)
2410
return searcher.get_result()
2411
2412
@deprecated_method(one_two)
1630
2413
@needs_read_lock
1631
def missing_revision_ids(self, revision_id=None):
2414
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1632
2415
"""Return the revision ids that source has that target does not.
1633
2416
1634
2417
These are returned in topological order.
1635
2418
1636
2419
:param revision_id: only return revision ids included by this
1637
2420
revision_id.
1638
"""
2421
:param find_ghosts: If True find missing revisions in deep history
2422
rather than just finding the surface difference.
2423
"""
2424
return list(self.search_missing_revision_ids(
2425
revision_id, find_ghosts).get_keys())
2426
2427
@needs_read_lock
2428
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2429
"""Return the revision ids that source has that target does not.
2430
2431
:param revision_id: only return revision ids included by this
2432
revision_id.
2433
:param find_ghosts: If True find missing revisions in deep history
2434
rather than just finding the surface difference.
2435
:return: A bzrlib.graph.SearchResult.
2436
"""
2437
# stop searching at found target revisions.
2438
if not find_ghosts and revision_id is not None:
2439
return self._walk_to_common_revisions([revision_id])
1639
2440
# generic, possibly worst case, slow code path.
1640
2441
target_ids = set(self.target.all_revision_ids())
1641
2442
if revision_id is not None:
1642
2443
source_ids = self.source.get_ancestry(revision_id)
1643
assert source_ids[0] == None
2444
if source_ids[0] is not None:
2445
raise AssertionError()
1644
2446
source_ids.pop(0)
1645
2447
else:
1646
2448
source_ids = self.source.all_revision_ids()
1647
2449
result_set = set(source_ids).difference(target_ids)
1648
# this may look like a no-op: its not. It preserves the ordering
1649
# other_ids had while only returning the members from other_ids
1650
# that we've decided we need.
1651
return [rev_id for rev_id in source_ids if rev_id in result_set]
1652
1653
1654
class InterWeaveRepo(InterRepository):
1655
"""Optimised code paths between Weave based repositories."""
1656
1657
_matching_repo_format = RepositoryFormat7()
1658
"""Repository format for testing with."""
2450
return self.source.revision_ids_to_search_result(result_set)
2451
2452
@staticmethod
2453
def _same_model(source, target):
2454
"""True if source and target have the same data representation.
2455
2456
Note: this is always called on the base class; overriding it in a
2457
subclass will have no effect.
2458
"""
2459
try:
2460
InterRepository._assert_same_model(source, target)
2461
return True
2462
except errors.IncompatibleRepositories, e:
2463
return False
2464
2465
@staticmethod
2466
def _assert_same_model(source, target):
2467
"""Raise an exception if two repositories do not use the same model.
2468
"""
2469
if source.supports_rich_root() != target.supports_rich_root():
2470
raise errors.IncompatibleRepositories(source, target,
2471
"different rich-root support")
2472
if source._serializer != target._serializer:
2473
raise errors.IncompatibleRepositories(source, target,
2474
"different serializers")
2475
2476
2477
class InterSameDataRepository(InterRepository):
2478
"""Code for converting between repositories that represent the same data.
2479
2480
Data format and model must match for this to work.
2481
"""
2482
2483
@classmethod
2484
def _get_repo_format_to_test(self):
2485
"""Repository format for testing with.
2486
2487
InterSameData can pull from subtree to subtree and from non-subtree to
2488
non-subtree, so we test this with the richest repository format.
2489
"""
2490
from bzrlib.repofmt import knitrepo
2491
return knitrepo.RepositoryFormatKnit3()
2492
2493
@staticmethod
2494
def is_compatible(source, target):
2495
return InterRepository._same_model(source, target)
2496
2497
@needs_write_lock
2498
def copy_content(self, revision_id=None):
2499
"""Make a complete copy of the content in self into destination.
2500
2501
This copies both the repository's revision data, and configuration information
2502
such as the make_working_trees setting.
2503
2504
This is a destructive operation! Do not use it on existing
2505
repositories.
2506
2507
:param revision_id: Only copy the content needed to construct
2508
revision_id and its parents.
2509
"""
2510
try:
2511
self.target.set_make_working_trees(self.source.make_working_trees())
2512
except NotImplementedError:
2513
pass
2514
# but don't bother fetching if we have the needed data now.
2515
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2516
self.target.has_revision(revision_id)):
2517
return
2518
self.target.fetch(self.source, revision_id=revision_id)
2519
2520
@needs_write_lock
2521
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2522
"""See InterRepository.fetch()."""
2523
from bzrlib.fetch import RepoFetcher
2524
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2525
self.source, self.source._format, self.target,
2526
self.target._format)
2527
f = RepoFetcher(to_repository=self.target,
2528
from_repository=self.source,
2529
last_revision=revision_id,
2530
pb=pb, find_ghosts=find_ghosts)
2531
return f.count_copied, f.failed_revisions
2532
2533
2534
class InterWeaveRepo(InterSameDataRepository):
2535
"""Optimised code paths between Weave based repositories.
2536
2537
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2538
implemented lazy inter-object optimisation.
2539
"""
2540
2541
@classmethod
2542
def _get_repo_format_to_test(self):
2543
from bzrlib.repofmt import weaverepo
2544
return weaverepo.RepositoryFormat7()
1659
2545
1660
2546
@staticmethod
1661
2547
def is_compatible(source, target):
1665
2551
could lead to confusing results, and there is no need to be
1666
2552
overly general.
1667
2553
"""
2554
from bzrlib.repofmt.weaverepo import (
2555
RepositoryFormat5,
2556
RepositoryFormat6,
2557
RepositoryFormat7,
2558
)
1668
2559
try:
1669
2560
return (isinstance(source._format, (RepositoryFormat5,
1670
2561
RepositoryFormat6,
1676
2567
return False
1677
2568
1678
2569
@needs_write_lock
1679
def copy_content(self, revision_id=None, basis=None):
2570
def copy_content(self, revision_id=None):
1680
2571
"""See InterRepository.copy_content()."""
1681
2572
# weave specific optimised path:
1682
if basis is not None:
1683
# copy the basis in, then fetch remaining data.
1684
basis.copy_content_into(self.target, revision_id)
1685
# the basis copy_content_into could miss-set this.
2573
try:
2574
self.target.set_make_working_trees(self.source.make_working_trees())
2575
except (errors.RepositoryUpgradeRequired, NotImplemented):
2576
pass
2577
# FIXME do not peek!
2578
if self.source._transport.listable():
2579
pb = ui.ui_factory.nested_progress_bar()
1686
2580
try:
1687
self.target.set_make_working_trees(self.source.make_working_trees())
1688
except NotImplementedError:
1689
pass
2581
self.target.texts.insert_record_stream(
2582
self.source.texts.get_record_stream(
2583
self.source.texts.keys(), 'topological', False))
2584
pb.update('copying inventory', 0, 1)
2585
self.target.inventories.insert_record_stream(
2586
self.source.inventories.get_record_stream(
2587
self.source.inventories.keys(), 'topological', False))
2588
self.target.signatures.insert_record_stream(
2589
self.source.signatures.get_record_stream(
2590
self.source.signatures.keys(),
2591
'unordered', True))
2592
self.target.revisions.insert_record_stream(
2593
self.source.revisions.get_record_stream(
2594
self.source.revisions.keys(),
2595
'topological', True))
2596
finally:
2597
pb.finished()
2598
else:
1690
2599
self.target.fetch(self.source, revision_id=revision_id)
1691
else:
1692
try:
1693
self.target.set_make_working_trees(self.source.make_working_trees())
1694
except NotImplementedError:
1695
pass
1696
# FIXME do not peek!
1697
if self.source.control_files._transport.listable():
1698
pb = ui.ui_factory.nested_progress_bar()
1699
try:
1700
self.target.weave_store.copy_all_ids(
1701
self.source.weave_store,
1702
pb=pb,
1703
from_transaction=self.source.get_transaction(),
1704
to_transaction=self.target.get_transaction())
1705
pb.update('copying inventory', 0, 1)
1706
self.target.control_weaves.copy_multi(
1707
self.source.control_weaves, ['inventory'],
1708
from_transaction=self.source.get_transaction(),
1709
to_transaction=self.target.get_transaction())
1710
self.target._revision_store.text_store.copy_all_ids(
1711
self.source._revision_store.text_store,
1712
pb=pb)
1713
finally:
1714
pb.finished()
1715
else:
1716
self.target.fetch(self.source, revision_id=revision_id)
1717
2600
1718
2601
@needs_write_lock
1719
def fetch(self, revision_id=None, pb=None):
2602
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1720
2603
"""See InterRepository.fetch()."""
1721
from bzrlib.fetch import GenericRepoFetcher
2604
from bzrlib.fetch import RepoFetcher
1722
2605
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1723
2606
self.source, self.source._format, self.target, self.target._format)
1724
f = GenericRepoFetcher(to_repository=self.target,
2607
f = RepoFetcher(to_repository=self.target,
1725
2608
from_repository=self.source,
1726
2609
last_revision=revision_id,
1727
pb=pb)
2610
pb=pb, find_ghosts=find_ghosts)
1728
2611
return f.count_copied, f.failed_revisions
1729
2612
1730
2613
@needs_read_lock
1731
def missing_revision_ids(self, revision_id=None):
2614
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1732
2615
"""See InterRepository.missing_revision_ids()."""
1733
2616
# we want all revisions to satisfy revision_id in source.
1734
2617
# but we don't want to stat every file here and there.
1743
2626
# - RBC 20060209
1744
2627
if revision_id is not None:
1745
2628
source_ids = self.source.get_ancestry(revision_id)
1746
assert source_ids[0] == None
2629
if source_ids[0] is not None:
2630
raise AssertionError()
1747
2631
source_ids.pop(0)
1748
2632
else:
1749
2633
source_ids = self.source._all_possible_ids()
1754
2638
# we do not have a revision as that would be pointless.
1755
2639
target_ids = set(self.target._all_possible_ids())
1756
2640
possibly_present_revisions = target_ids.intersection(source_ids_set)
1757
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2641
actually_present_revisions = set(
2642
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1758
2643
required_revisions = source_ids_set.difference(actually_present_revisions)
1759
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1760
2644
if revision_id is not None:
1761
2645
# we used get_ancestry to determine source_ids then we are assured all
1762
2646
# revisions referenced are present as they are installed in topological order.
1763
2647
# and the tip revision was validated by get_ancestry.
1764
return required_topo_revisions
2648
result_set = required_revisions
1765
2649
else:
1766
2650
# if we just grabbed the possibly available ids, then
1767
2651
# we only have an estimate of whats available and need to validate
1768
2652
# that against the revision records.
1769
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1770
1771
1772
class InterKnitRepo(InterRepository):
2653
result_set = set(
2654
self.source._eliminate_revisions_not_present(required_revisions))
2655
return self.source.revision_ids_to_search_result(result_set)
2656
2657
2658
class InterKnitRepo(InterSameDataRepository):
1773
2659
"""Optimised code paths between Knit based repositories."""
1774
2660
1775
_matching_repo_format = RepositoryFormatKnit1()
1776
"""Repository format for testing with."""
2661
@classmethod
2662
def _get_repo_format_to_test(self):
2663
from bzrlib.repofmt import knitrepo
2664
return knitrepo.RepositoryFormatKnit1()
1777
2665
1778
2666
@staticmethod
1779
2667
def is_compatible(source, target):
1783
2671
could lead to confusing results, and there is no need to be
1784
2672
overly general.
1785
2673
"""
2674
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
1786
2675
try:
1787
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1788
isinstance(target._format, (RepositoryFormatKnit1)))
2676
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2677
isinstance(target._format, RepositoryFormatKnit))
1789
2678
except AttributeError:
1790
2679
return False
2680
return are_knits and InterRepository._same_model(source, target)
1791
2681
1792
2682
@needs_write_lock
1793
def fetch(self, revision_id=None, pb=None):
2683
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1794
2684
"""See InterRepository.fetch()."""
1795
from bzrlib.fetch import KnitRepoFetcher
2685
from bzrlib.fetch import RepoFetcher
1796
2686
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1797
2687
self.source, self.source._format, self.target, self.target._format)
1798
f = KnitRepoFetcher(to_repository=self.target,
2688
f = RepoFetcher(to_repository=self.target,
1799
2689
from_repository=self.source,
1800
2690
last_revision=revision_id,
1801
pb=pb)
2691
pb=pb, find_ghosts=find_ghosts)
1802
2692
return f.count_copied, f.failed_revisions
1803
2693
1804
2694
@needs_read_lock
1805
def missing_revision_ids(self, revision_id=None):
2695
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1806
2696
"""See InterRepository.missing_revision_ids()."""
1807
2697
if revision_id is not None:
1808
2698
source_ids = self.source.get_ancestry(revision_id)
1809
assert source_ids[0] == None
2699
if source_ids[0] is not None:
2700
raise AssertionError()
1810
2701
source_ids.pop(0)
1811
2702
else:
1812
source_ids = self.source._all_possible_ids()
2703
source_ids = self.source.all_revision_ids()
1813
2704
source_ids_set = set(source_ids)
1814
2705
# source_ids is the worst possible case we may need to pull.
1815
2706
# now we want to filter source_ids against what we actually
1816
2707
# have in target, but don't try to check for existence where we know
1817
2708
# we do not have a revision as that would be pointless.
1818
target_ids = set(self.target._all_possible_ids())
2709
target_ids = set(self.target.all_revision_ids())
1819
2710
possibly_present_revisions = target_ids.intersection(source_ids_set)
1820
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2711
actually_present_revisions = set(
2712
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1821
2713
required_revisions = source_ids_set.difference(actually_present_revisions)
1822
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1823
2714
if revision_id is not None:
1824
2715
# we used get_ancestry to determine source_ids then we are assured all
1825
2716
# revisions referenced are present as they are installed in topological order.
1826
2717
# and the tip revision was validated by get_ancestry.
1827
return required_topo_revisions
2718
result_set = required_revisions
1828
2719
else:
1829
2720
# if we just grabbed the possibly available ids, then
1830
2721
# we only have an estimate of whats available and need to validate
1831
2722
# that against the revision records.
1832
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1833
2723
result_set = set(
2724
self.source._eliminate_revisions_not_present(required_revisions))
2725
return self.source.revision_ids_to_search_result(result_set)
2726
2727
2728
class InterPackRepo(InterSameDataRepository):
2729
"""Optimised code paths between Pack based repositories."""
2730
2731
@classmethod
2732
def _get_repo_format_to_test(self):
2733
from bzrlib.repofmt import pack_repo
2734
return pack_repo.RepositoryFormatKnitPack1()
2735
2736
@staticmethod
2737
def is_compatible(source, target):
2738
"""Be compatible with known Pack formats.
2739
2740
We don't test for the stores being of specific types because that
2741
could lead to confusing results, and there is no need to be
2742
overly general.
2743
"""
2744
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2745
try:
2746
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2747
isinstance(target._format, RepositoryFormatPack))
2748
except AttributeError:
2749
return False
2750
return are_packs and InterRepository._same_model(source, target)
2751
2752
@needs_write_lock
2753
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2754
"""See InterRepository.fetch()."""
2755
if (len(self.source._fallback_repositories) > 0 or
2756
len(self.target._fallback_repositories) > 0):
2757
# The pack layer is not aware of fallback repositories, so when
2758
# fetching from a stacked repository or into a stacked repository
2759
# we use the generic fetch logic which uses the VersionedFiles
2760
# attributes on repository.
2761
from bzrlib.fetch import RepoFetcher
2762
fetcher = RepoFetcher(self.target, self.source, revision_id,
2763
pb, find_ghosts)
2764
return fetcher.count_copied, fetcher.failed_revisions
2765
from bzrlib.repofmt.pack_repo import Packer
2766
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2767
self.source, self.source._format, self.target, self.target._format)
2768
self.count_copied = 0
2769
if revision_id is None:
2770
# TODO:
2771
# everything to do - use pack logic
2772
# to fetch from all packs to one without
2773
# inventory parsing etc, IFF nothing to be copied is in the target.
2774
# till then:
2775
source_revision_ids = frozenset(self.source.all_revision_ids())
2776
revision_ids = source_revision_ids - \
2777
frozenset(self.target.get_parent_map(source_revision_ids))
2778
revision_keys = [(revid,) for revid in revision_ids]
2779
index = self.target._pack_collection.revision_index.combined_index
2780
present_revision_ids = set(item[1][0] for item in
2781
index.iter_entries(revision_keys))
2782
revision_ids = set(revision_ids) - present_revision_ids
2783
# implementing the TODO will involve:
2784
# - detecting when all of a pack is selected
2785
# - avoiding as much as possible pre-selection, so the
2786
# more-core routines such as create_pack_from_packs can filter in
2787
# a just-in-time fashion. (though having a HEADS list on a
2788
# repository might make this a lot easier, because we could
2789
# sensibly detect 'new revisions' without doing a full index scan.
2790
elif _mod_revision.is_null(revision_id):
2791
# nothing to do:
2792
return (0, [])
2793
else:
2794
try:
2795
revision_ids = self.search_missing_revision_ids(revision_id,
2796
find_ghosts=find_ghosts).get_keys()
2797
except errors.NoSuchRevision:
2798
raise errors.InstallFailed([revision_id])
2799
if len(revision_ids) == 0:
2800
return (0, [])
2801
packs = self.source._pack_collection.all_packs()
2802
pack = Packer(self.target._pack_collection, packs, '.fetch',
2803
revision_ids).pack()
2804
if pack is not None:
2805
self.target._pack_collection._save_pack_names()
2806
# Trigger an autopack. This may duplicate effort as we've just done
2807
# a pack creation, but for now it is simpler to think about as
2808
# 'upload data, then repack if needed'.
2809
self.target._pack_collection.autopack()
2810
return (pack.get_revision_count(), [])
2811
else:
2812
return (0, [])
2813
2814
@needs_read_lock
2815
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2816
"""See InterRepository.missing_revision_ids().
2817
2818
:param find_ghosts: Find ghosts throughout the ancestry of
2819
revision_id.
2820
"""
2821
if not find_ghosts and revision_id is not None:
2822
return self._walk_to_common_revisions([revision_id])
2823
elif revision_id is not None:
2824
# Find ghosts: search for revisions pointing from one repository to
2825
# the other, and vice versa, anywhere in the history of revision_id.
2826
graph = self.target.get_graph(other_repository=self.source)
2827
searcher = graph._make_breadth_first_searcher([revision_id])
2828
found_ids = set()
2829
while True:
2830
try:
2831
next_revs, ghosts = searcher.next_with_ghosts()
2832
except StopIteration:
2833
break
2834
if revision_id in ghosts:
2835
raise errors.NoSuchRevision(self.source, revision_id)
2836
found_ids.update(next_revs)
2837
found_ids.update(ghosts)
2838
found_ids = frozenset(found_ids)
2839
# Double query here: should be able to avoid this by changing the
2840
# graph api further.
2841
result_set = found_ids - frozenset(
2842
self.target.get_parent_map(found_ids))
2843
else:
2844
source_ids = self.source.all_revision_ids()
2845
# source_ids is the worst possible case we may need to pull.
2846
# now we want to filter source_ids against what we actually
2847
# have in target, but don't try to check for existence where we know
2848
# we do not have a revision as that would be pointless.
2849
target_ids = set(self.target.all_revision_ids())
2850
result_set = set(source_ids).difference(target_ids)
2851
return self.source.revision_ids_to_search_result(result_set)
2852
2853
2854
class InterModel1and2(InterRepository):
2855
2856
@classmethod
2857
def _get_repo_format_to_test(self):
2858
return None
2859
2860
@staticmethod
2861
def is_compatible(source, target):
2862
if not source.supports_rich_root() and target.supports_rich_root():
2863
return True
2864
else:
2865
return False
2866
2867
@needs_write_lock
2868
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2869
"""See InterRepository.fetch()."""
2870
from bzrlib.fetch import Model1toKnit2Fetcher
2871
f = Model1toKnit2Fetcher(to_repository=self.target,
2872
from_repository=self.source,
2873
last_revision=revision_id,
2874
pb=pb, find_ghosts=find_ghosts)
2875
return f.count_copied, f.failed_revisions
2876
2877
@needs_write_lock
2878
def copy_content(self, revision_id=None):
2879
"""Make a complete copy of the content in self into destination.
2880
2881
This is a destructive operation! Do not use it on existing
2882
repositories.
2883
2884
:param revision_id: Only copy the content needed to construct
2885
revision_id and its parents.
2886
"""
2887
try:
2888
self.target.set_make_working_trees(self.source.make_working_trees())
2889
except NotImplementedError:
2890
pass
2891
# but don't bother fetching if we have the needed data now.
2892
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2893
self.target.has_revision(revision_id)):
2894
return
2895
self.target.fetch(self.source, revision_id=revision_id)
2896
2897
2898
class InterKnit1and2(InterKnitRepo):
2899
2900
@classmethod
2901
def _get_repo_format_to_test(self):
2902
return None
2903
2904
@staticmethod
2905
def is_compatible(source, target):
2906
"""Be compatible with Knit1 source and Knit3 target"""
2907
try:
2908
from bzrlib.repofmt.knitrepo import (
2909
RepositoryFormatKnit1,
2910
RepositoryFormatKnit3,
2911
)
2912
from bzrlib.repofmt.pack_repo import (
2913
RepositoryFormatKnitPack1,
2914
RepositoryFormatKnitPack3,
2915
RepositoryFormatKnitPack4,
2916
RepositoryFormatKnitPack5,
2917
RepositoryFormatKnitPack5RichRoot,
2918
RepositoryFormatPackDevelopment1,
2919
RepositoryFormatPackDevelopment1Subtree,
2920
)
2921
norichroot = (
2922
RepositoryFormatKnit1, # no rr, no subtree
2923
RepositoryFormatKnitPack1, # no rr, no subtree
2924
RepositoryFormatPackDevelopment1, # no rr, no subtree
2925
RepositoryFormatKnitPack5, # no rr, no subtree
2926
)
2927
richroot = (
2928
RepositoryFormatKnit3, # rr, subtree
2929
RepositoryFormatKnitPack3, # rr, subtree
2930
RepositoryFormatKnitPack4, # rr, no subtree
2931
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2932
RepositoryFormatPackDevelopment1Subtree, # rr, subtree
2933
)
2934
for format in norichroot:
2935
if format.rich_root_data:
2936
raise AssertionError('Format %s is a rich-root format'
2937
' but is included in the non-rich-root list'
2938
% (format,))
2939
for format in richroot:
2940
if not format.rich_root_data:
2941
raise AssertionError('Format %s is not a rich-root format'
2942
' but is included in the rich-root list'
2943
% (format,))
2944
# TODO: One alternative is to just check format.rich_root_data,
2945
# instead of keeping membership lists. However, the formats
2946
# *also* have to use the same 'Knit' style of storage
2947
# (line-deltas, fulltexts, etc.)
2948
return (isinstance(source._format, norichroot) and
2949
isinstance(target._format, richroot))
2950
except AttributeError:
2951
return False
2952
2953
@needs_write_lock
2954
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2955
"""See InterRepository.fetch()."""
2956
from bzrlib.fetch import Knit1to2Fetcher
2957
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2958
self.source, self.source._format, self.target,
2959
self.target._format)
2960
f = Knit1to2Fetcher(to_repository=self.target,
2961
from_repository=self.source,
2962
last_revision=revision_id,
2963
pb=pb, find_ghosts=find_ghosts)
2964
return f.count_copied, f.failed_revisions
2965
2966
2967
class InterDifferingSerializer(InterKnitRepo):
2968
2969
@classmethod
2970
def _get_repo_format_to_test(self):
2971
return None
2972
2973
@staticmethod
2974
def is_compatible(source, target):
2975
"""Be compatible with Knit2 source and Knit3 target"""
2976
if source.supports_rich_root() != target.supports_rich_root():
2977
return False
2978
# Ideally, we'd support fetching if the source had no tree references
2979
# even if it supported them...
2980
if (getattr(source, '_format.supports_tree_reference', False) and
2981
not getattr(target, '_format.supports_tree_reference', False)):
2982
return False
2983
return True
2984
2985
@needs_write_lock
2986
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2987
"""See InterRepository.fetch()."""
2988
revision_ids = self.target.search_missing_revision_ids(self.source,
2989
revision_id, find_ghosts=find_ghosts).get_keys()
2990
revision_ids = tsort.topo_sort(
2991
self.source.get_graph().get_parent_map(revision_ids))
2992
def revisions_iterator():
2993
for current_revision_id in revision_ids:
2994
revision = self.source.get_revision(current_revision_id)
2995
tree = self.source.revision_tree(current_revision_id)
2996
try:
2997
signature = self.source.get_signature_text(
2998
current_revision_id)
2999
except errors.NoSuchRevision:
3000
signature = None
3001
yield revision, tree, signature
3002
if pb is None:
3003
my_pb = ui.ui_factory.nested_progress_bar()
3004
pb = my_pb
3005
else:
3006
my_pb = None
3007
try:
3008
install_revisions(self.target, revisions_iterator(),
3009
len(revision_ids), pb)
3010
finally:
3011
if my_pb is not None:
3012
my_pb.finished()
3013
return len(revision_ids), 0
3014
3015
3016
class InterOtherToRemote(InterRepository):
3017
3018
def __init__(self, source, target):
3019
InterRepository.__init__(self, source, target)
3020
self._real_inter = None
3021
3022
@staticmethod
3023
def is_compatible(source, target):
3024
if isinstance(target, remote.RemoteRepository):
3025
return True
3026
return False
3027
3028
def _ensure_real_inter(self):
3029
if self._real_inter is None:
3030
self.target._ensure_real()
3031
real_target = self.target._real_repository
3032
self._real_inter = InterRepository.get(self.source, real_target)
3033
3034
def copy_content(self, revision_id=None):
3035
self._ensure_real_inter()
3036
self._real_inter.copy_content(revision_id=revision_id)
3037
3038
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3039
self._ensure_real_inter()
3040
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3041
find_ghosts=find_ghosts)
3042
3043
@classmethod
3044
def _get_repo_format_to_test(self):
3045
return None
3046
3047
3048
class InterRemoteToOther(InterRepository):
3049
3050
def __init__(self, source, target):
3051
InterRepository.__init__(self, source, target)
3052
self._real_inter = None
3053
3054
@staticmethod
3055
def is_compatible(source, target):
3056
if not isinstance(source, remote.RemoteRepository):
3057
return False
3058
# Is source's model compatible with target's model?
3059
source._ensure_real()
3060
real_source = source._real_repository
3061
if isinstance(real_source, remote.RemoteRepository):
3062
raise NotImplementedError(
3063
"We don't support remote repos backed by remote repos yet.")
3064
return InterRepository._same_model(real_source, target)
3065
3066
def _ensure_real_inter(self):
3067
if self._real_inter is None:
3068
self.source._ensure_real()
3069
real_source = self.source._real_repository
3070
self._real_inter = InterRepository.get(real_source, self.target)
3071
3072
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3073
self._ensure_real_inter()
3074
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3075
find_ghosts=find_ghosts)
3076
3077
def copy_content(self, revision_id=None):
3078
self._ensure_real_inter()
3079
self._real_inter.copy_content(revision_id=revision_id)
3080
3081
@classmethod
3082
def _get_repo_format_to_test(self):
3083
return None
3084
3085
3086
3087
InterRepository.register_optimiser(InterDifferingSerializer)
3088
InterRepository.register_optimiser(InterSameDataRepository)
1834
3089
InterRepository.register_optimiser(InterWeaveRepo)
1835
3090
InterRepository.register_optimiser(InterKnitRepo)
1836
1837
1838
class RepositoryTestProviderAdapter(object):
1839
"""A tool to generate a suite testing multiple repository formats at once.
1840
1841
This is done by copying the test once for each transport and injecting
1842
the transport_server, transport_readonly_server, and bzrdir_format and
1843
repository_format classes into each copy. Each copy is also given a new id()
1844
to make it easy to identify.
1845
"""
1846
1847
def __init__(self, transport_server, transport_readonly_server, formats):
1848
self._transport_server = transport_server
1849
self._transport_readonly_server = transport_readonly_server
1850
self._formats = formats
1851
1852
def adapt(self, test):
1853
result = TestSuite()
1854
for repository_format, bzrdir_format in self._formats:
1855
new_test = deepcopy(test)
1856
new_test.transport_server = self._transport_server
1857
new_test.transport_readonly_server = self._transport_readonly_server
1858
new_test.bzrdir_format = bzrdir_format
1859
new_test.repository_format = repository_format
1860
def make_new_test_id():
1861
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1862
return lambda: new_id
1863
new_test.id = make_new_test_id()
1864
result.addTest(new_test)
1865
return result
1866
1867
1868
class InterRepositoryTestProviderAdapter(object):
1869
"""A tool to generate a suite testing multiple inter repository formats.
1870
1871
This is done by copying the test once for each interrepo provider and injecting
1872
the transport_server, transport_readonly_server, repository_format and
1873
repository_to_format classes into each copy.
1874
Each copy is also given a new id() to make it easy to identify.
1875
"""
1876
1877
def __init__(self, transport_server, transport_readonly_server, formats):
1878
self._transport_server = transport_server
1879
self._transport_readonly_server = transport_readonly_server
1880
self._formats = formats
1881
1882
def adapt(self, test):
1883
result = TestSuite()
1884
for interrepo_class, repository_format, repository_format_to in self._formats:
1885
new_test = deepcopy(test)
1886
new_test.transport_server = self._transport_server
1887
new_test.transport_readonly_server = self._transport_readonly_server
1888
new_test.interrepo_class = interrepo_class
1889
new_test.repository_format = repository_format
1890
new_test.repository_format_to = repository_format_to
1891
def make_new_test_id():
1892
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1893
return lambda: new_id
1894
new_test.id = make_new_test_id()
1895
result.addTest(new_test)
1896
return result
1897
1898
@staticmethod
1899
def default_test_list():
1900
"""Generate the default list of interrepo permutations to test."""
1901
result = []
1902
# test the default InterRepository between format 6 and the current
1903
# default format.
1904
# XXX: robertc 20060220 reinstate this when there are two supported
1905
# formats which do not have an optimal code path between them.
1906
result.append((InterRepository,
1907
RepositoryFormat6(),
1908
RepositoryFormatKnit1()))
1909
for optimiser in InterRepository._optimisers:
1910
result.append((optimiser,
1911
optimiser._matching_repo_format,
1912
optimiser._matching_repo_format
1913
))
1914
# if there are specific combinations we want to use, we can add them
1915
# here.
1916
return result
3091
InterRepository.register_optimiser(InterModel1and2)
3092
InterRepository.register_optimiser(InterKnit1and2)
3093
InterRepository.register_optimiser(InterPackRepo)
3094
InterRepository.register_optimiser(InterOtherToRemote)
3095
InterRepository.register_optimiser(InterRemoteToOther)
1917
3096
1918
3097
1919
3098
class CopyConverter(object):
1945
3124
self.step('Moving repository to repository.backup')
1946
3125
self.repo_dir.transport.move('repository', 'repository.backup')
1947
3126
backup_transport = self.repo_dir.transport.clone('repository.backup')
3127
repo._format.check_conversion_target(self.target_format)
1948
3128
self.source_repo = repo._format.open(self.repo_dir,
1949
3129
_found=True,
1950
3130
_override_transport=backup_transport)
1967
3147
self.pb.update(message, self.count, self.total)
1968
3148
1969
3149
1970
class CommitBuilder(object):
1971
"""Provides an interface to build up a commit.
1972
1973
This allows describing a tree to be committed without needing to
1974
know the internals of the format of the repository.
1975
"""
1976
1977
record_root_entry = False
1978
def __init__(self, repository, parents, config, timestamp=None,
1979
timezone=None, committer=None, revprops=None,
1980
revision_id=None):
1981
"""Initiate a CommitBuilder.
1982
1983
:param repository: Repository to commit to.
1984
:param parents: Revision ids of the parents of the new revision.
1985
:param config: Configuration to use.
1986
:param timestamp: Optional timestamp recorded for commit.
1987
:param timezone: Optional timezone for timestamp.
1988
:param committer: Optional committer to set for commit.
1989
:param revprops: Optional dictionary of revision properties.
1990
:param revision_id: Optional revision id.
1991
"""
1992
self._config = config
1993
1994
if committer is None:
1995
self._committer = self._config.username()
1996
else:
1997
assert isinstance(committer, basestring), type(committer)
1998
self._committer = committer
1999
2000
self.new_inventory = Inventory(None)
2001
self._new_revision_id = revision_id
2002
self.parents = parents
2003
self.repository = repository
2004
2005
self._revprops = {}
2006
if revprops is not None:
2007
self._revprops.update(revprops)
2008
2009
if timestamp is None:
2010
timestamp = time.time()
2011
# Restrict resolution to 1ms
2012
self._timestamp = round(timestamp, 3)
2013
2014
if timezone is None:
2015
self._timezone = local_time_offset()
2016
else:
2017
self._timezone = int(timezone)
2018
2019
self._generate_revision_if_needed()
2020
2021
def commit(self, message):
2022
"""Make the actual commit.
2023
2024
:return: The revision id of the recorded revision.
2025
"""
2026
rev = Revision(timestamp=self._timestamp,
2027
timezone=self._timezone,
2028
committer=self._committer,
2029
message=message,
2030
inventory_sha1=self.inv_sha1,
2031
revision_id=self._new_revision_id,
2032
properties=self._revprops)
2033
rev.parent_ids = self.parents
2034
self.repository.add_revision(self._new_revision_id, rev,
2035
self.new_inventory, self._config)
2036
return self._new_revision_id
2037
2038
def finish_inventory(self):
2039
"""Tell the builder that the inventory is finished."""
2040
if self.new_inventory.root is None:
2041
symbol_versioning.warn('Root entry should be supplied to'
2042
' record_entry_contents, as of bzr 0.10.',
2043
DeprecationWarning, stacklevel=2)
2044
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2045
self.new_inventory.revision_id = self._new_revision_id
2046
self.inv_sha1 = self.repository.add_inventory(
2047
self._new_revision_id,
2048
self.new_inventory,
2049
self.parents
2050
)
2051
2052
def _gen_revision_id(self):
2053
"""Return new revision-id."""
2054
s = '%s-%s-' % (self._config.user_email(),
2055
compact_date(self._timestamp))
2056
s += hexlify(rand_bytes(8))
2057
return s
2058
2059
def _generate_revision_if_needed(self):
2060
"""Create a revision id if None was supplied.
2061
2062
If the repository can not support user-specified revision ids
2063
they should override this function and raise UnsupportedOperation
2064
if _new_revision_id is not None.
2065
2066
:raises: UnsupportedOperation
2067
"""
2068
if self._new_revision_id is None:
2069
self._new_revision_id = self._gen_revision_id()
2070
2071
def record_entry_contents(self, ie, parent_invs, path, tree):
2072
"""Record the content of ie from tree into the commit if needed.
2073
2074
Side effect: sets ie.revision when unchanged
2075
2076
:param ie: An inventory entry present in the commit.
2077
:param parent_invs: The inventories of the parent revisions of the
2078
commit.
2079
:param path: The path the entry is at in the tree.
2080
:param tree: The tree which contains this entry and should be used to
2081
obtain content.
2082
"""
2083
if self.new_inventory.root is None and ie.parent_id is not None:
2084
symbol_versioning.warn('Root entry should be supplied to'
2085
' record_entry_contents, as of bzr 0.10.',
2086
DeprecationWarning, stacklevel=2)
2087
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2088
'', tree)
2089
self.new_inventory.add(ie)
2090
2091
# ie.revision is always None if the InventoryEntry is considered
2092
# for committing. ie.snapshot will record the correct revision
2093
# which may be the sole parent if it is untouched.
2094
if ie.revision is not None:
2095
return
2096
2097
# In this revision format, root entries have no knit or weave
2098
if ie is self.new_inventory.root:
2099
if len(parent_invs):
2100
ie.revision = parent_invs[0].root.revision
2101
else:
2102
ie.revision = None
2103
return
2104
previous_entries = ie.find_previous_heads(
2105
parent_invs,
2106
self.repository.weave_store,
2107
self.repository.get_transaction())
2108
# we are creating a new revision for ie in the history store
2109
# and inventory.
2110
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2111
2112
def modified_directory(self, file_id, file_parents):
2113
"""Record the presence of a symbolic link.
2114
2115
:param file_id: The file_id of the link to record.
2116
:param file_parents: The per-file parent revision ids.
2117
"""
2118
self._add_text_to_weave(file_id, [], file_parents.keys())
2119
2120
def modified_file_text(self, file_id, file_parents,
2121
get_content_byte_lines, text_sha1=None,
2122
text_size=None):
2123
"""Record the text of file file_id
2124
2125
:param file_id: The file_id of the file to record the text of.
2126
:param file_parents: The per-file parent revision ids.
2127
:param get_content_byte_lines: A callable which will return the byte
2128
lines for the file.
2129
:param text_sha1: Optional SHA1 of the file contents.
2130
:param text_size: Optional size of the file contents.
2131
"""
2132
# mutter('storing text of file {%s} in revision {%s} into %r',
2133
# file_id, self._new_revision_id, self.repository.weave_store)
2134
# special case to avoid diffing on renames or
2135
# reparenting
2136
if (len(file_parents) == 1
2137
and text_sha1 == file_parents.values()[0].text_sha1
2138
and text_size == file_parents.values()[0].text_size):
2139
previous_ie = file_parents.values()[0]
2140
versionedfile = self.repository.weave_store.get_weave(file_id,
2141
self.repository.get_transaction())
2142
versionedfile.clone_text(self._new_revision_id,
2143
previous_ie.revision, file_parents.keys())
2144
return text_sha1, text_size
2145
else:
2146
new_lines = get_content_byte_lines()
2147
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2148
# should return the SHA1 and size
2149
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2150
return osutils.sha_strings(new_lines), \
2151
sum(map(len, new_lines))
2152
2153
def modified_link(self, file_id, file_parents, link_target):
2154
"""Record the presence of a symbolic link.
2155
2156
:param file_id: The file_id of the link to record.
2157
:param file_parents: The per-file parent revision ids.
2158
:param link_target: Target location of this link.
2159
"""
2160
self._add_text_to_weave(file_id, [], file_parents.keys())
2161
2162
def _add_text_to_weave(self, file_id, new_lines, parents):
2163
versionedfile = self.repository.weave_store.get_weave_or_empty(
2164
file_id, self.repository.get_transaction())
2165
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2166
versionedfile.clear_cache()
2167
2168
2169
class _CommitBuilder(CommitBuilder):
2170
"""Temporary class so old CommitBuilders are detected properly
2171
2172
Note: CommitBuilder works whether or not root entry is recorded.
2173
"""
2174
2175
record_root_entry = True
2176
2177
2178
3150
_unescape_map = {
2179
3151
'apos':"'",
2180
3152
'quot':'"',
2185
3157
2186
3158
2187
3159
def _unescaper(match, _map=_unescape_map):
2188
return _map[match.group(1)]
3160
code = match.group(1)
3161
try:
3162
return _map[code]
3163
except KeyError:
3164
if not code.startswith('#'):
3165
raise
3166
return unichr(int(code[1:])).encode('utf8')
2189
3167
2190
3168
2191
3169
_unescape_re = None
2195
3173
"""Unescape predefined XML entities in a string of data."""
2196
3174
global _unescape_re
2197
3175
if _unescape_re is None:
2198
_unescape_re = re.compile('\&([^;]*);')
3176
_unescape_re = re.compile('\&([^;]*);')
2199
3177
return _unescape_re.sub(_unescaper, data)
3178
3179
3180
class _VersionedFileChecker(object):
3181
3182
def __init__(self, repository):
3183
self.repository = repository
3184
self.text_index = self.repository._generate_text_key_index()
3185
3186
def calculate_file_version_parents(self, text_key):
3187
"""Calculate the correct parents for a file version according to
3188
the inventories.
3189
"""
3190
parent_keys = self.text_index[text_key]
3191
if parent_keys == [_mod_revision.NULL_REVISION]:
3192
return ()
3193
return tuple(parent_keys)
3194
3195
def check_file_version_parents(self, texts, progress_bar=None):
3196
"""Check the parents stored in a versioned file are correct.
3197
3198
It also detects file versions that are not referenced by their
3199
corresponding revision's inventory.
3200
3201
:returns: A tuple of (wrong_parents, dangling_file_versions).
3202
wrong_parents is a dict mapping {revision_id: (stored_parents,
3203
correct_parents)} for each revision_id where the stored parents
3204
are not correct. dangling_file_versions is a set of (file_id,
3205
revision_id) tuples for versions that are present in this versioned
3206
file, but not used by the corresponding inventory.
3207
"""
3208
wrong_parents = {}
3209
self.file_ids = set([file_id for file_id, _ in
3210
self.text_index.iterkeys()])
3211
# text keys is now grouped by file_id
3212
n_weaves = len(self.file_ids)
3213
files_in_revisions = {}
3214
revisions_of_files = {}
3215
n_versions = len(self.text_index)
3216
progress_bar.update('loading text store', 0, n_versions)
3217
parent_map = self.repository.texts.get_parent_map(self.text_index)
3218
# On unlistable transports this could well be empty/error...
3219
text_keys = self.repository.texts.keys()
3220
unused_keys = frozenset(text_keys) - set(self.text_index)
3221
for num, key in enumerate(self.text_index.iterkeys()):
3222
if progress_bar is not None:
3223
progress_bar.update('checking text graph', num, n_versions)
3224
correct_parents = self.calculate_file_version_parents(key)
3225
try:
3226
knit_parents = parent_map[key]
3227
except errors.RevisionNotPresent:
3228
# Missing text!
3229
knit_parents = None
3230
if correct_parents != knit_parents:
3231
wrong_parents[key] = (knit_parents, correct_parents)
3232
return wrong_parents, unused_keys
3233
3234
3235
def _old_get_graph(repository, revision_id):
3236
"""DO NOT USE. That is all. I'm serious."""
3237
graph = repository.get_graph()
3238
revision_graph = dict(((key, value) for key, value in
3239
graph.iter_ancestry([revision_id]) if value is not None))
3240
return _strip_NULL_ghosts(revision_graph)
3241
3242
3243
def _strip_NULL_ghosts(revision_graph):
3244
"""Also don't use this. more compatibility code for unmigrated clients."""
3245
# Filter ghosts, and null:
3246
if _mod_revision.NULL_REVISION in revision_graph:
3247
del revision_graph[_mod_revision.NULL_REVISION]
3248
for key, parents in revision_graph.items():
3249
revision_graph[key] = tuple(parent for parent in parents if parent
3250
in revision_graph)
3251
return revision_graph
Loggerhead is a web-based interface for Breezy
Version: 2.0.1