« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 2520.4.129
- compare with another revision
- download diff
- download tarball
- view history from revision 2520.4.129
- Committer: Aaron Bentley
- Date: 2007-07-04 23:38:26 UTC
- mto: This revision was merged to the branch mainline in revision 2631.
- Revision ID: aaron.bentley@utoronto.ca-20070704233826-jkp63376wi1n96mm
update docs
- files added:
- bzrlib/plugins/multiparent.py
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/revisionstore_implementations
- files removed:
- bzrlib/_btree_serializer_c.pyx
- bzrlib/help_topics
- bzrlib/help_topics/en
- bzrlib/tests/commands
- bzrlib/tests/inventory_implementations
- bzrlib/tests/per_repository_reference
- bzrlib/transport/ftp
- bzrlib/util/tests
- 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
- tools/packaging
- files renamed:
- bzrlib/help_topics/__init__.py => bzrlib/help_topics.py
- bzrlib/_patiencediff_py.py => bzrlib/patiencediff.py
- bzrlib/plugins/launchpad/lp_directory.py => bzrlib/plugins/launchpad/lp_indirect.py
- bzrlib/plugins/launchpad/test_lp_directory.py => bzrlib/plugins/launchpad/test_lp_indirect.py
- bzrlib/tests/http_utils.py => bzrlib/tests/HTTPTestUtil.py
- bzrlib/tests/http_server.py => bzrlib/tests/HttpServer.py
- bzrlib/tests/blackbox/test_send.py => bzrlib/tests/blackbox/test_submit.py
- bzrlib/tests/test_revisionspec.py => bzrlib/tests/test_revisionnamespaces.py
- bzrlib/transport/ftp/__init__.py => bzrlib/transport/ftp.py
- bzrlib/xml8.py => bzrlib/xml5.py
- doc/en/tutorials/centralized_workflow.txt => doc/centralized_workflow.txt
- bzrlib/help_topics/en/configuration.txt => doc/configuration.txt
- doc/developers/HACKING.txt => doc/developers/HACKING
- doc/en/user-guide/http_smart_server.txt => doc/http_smart_server.txt
- doc/en/user-guide/index.txt => doc/index.txt
- doc/en/user-guide/plugins.txt => doc/plugins.txt
- doc/en/user-guide/server.txt => doc/server.txt
- doc/en/user-guide/setting_up_email.txt => doc/setting_up_email.txt
- doc/en/user-guide/shared_repository_layouts.txt => doc/shared_repository_layouts.txt
- doc/en/user-guide/specifying_revisions.txt => doc/specifying_revisions.txt
- doc/en/tutorials/tutorial.txt => doc/tutorial.txt
- doc/en/user-guide/using_aliases.txt => doc/using_aliases.txt
- doc/en/user-guide/version_info.txt => doc/version_info.txt
- files modified:
- .bzrignore
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
1
# Copyright (C) 2005, 2006, 2007 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
24
24
from bzrlib import (
25
25
bzrdir,
26
26
check,
27
debug,
27
deprecated_graph,
28
28
errors,
29
29
generate_ids,
30
30
gpg,
32
32
lazy_regex,
33
33
lockable_files,
34
34
lockdir,
35
lru_cache,
36
35
osutils,
37
36
registry,
38
37
remote,
39
38
revision as _mod_revision,
40
39
symbol_versioning,
41
40
transactions,
42
tsort,
43
41
ui,
44
42
)
45
43
from bzrlib.bundle import serializer
47
45
from bzrlib.store.versioned import VersionedFileStore
48
46
from bzrlib.store.text import TextStore
49
47
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
48
51
49
""")
52
50
53
51
from bzrlib.decorators import needs_read_lock, needs_write_lock
55
53
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
56
54
from bzrlib.symbol_versioning import (
57
55
deprecated_method,
58
one_one,
59
one_two,
60
one_three,
61
one_six,
56
zero_nine,
62
57
)
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
58
from bzrlib.trace import mutter, note, warning
64
59
65
60
66
61
# Old formats display a warning, but only once
67
62
_deprecation_warning_done = False
68
63
69
64
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
65
######################################################################
449
66
# Repositories
450
67
455
72
revisions and file history. It's normally accessed only by the Branch,
456
73
which views a particular line of development through that history.
457
74
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
75
The Repository builds on top of Stores and a Transport, which respectively
76
describe the disk data format and the way of accessing the (possibly
461
77
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.
495
78
"""
496
79
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
80
_file_ids_altered_regex = lazy_regex.lazy_compile(
504
81
r'file_id="(?P<file_id>[^"]+)"'
505
r'.* revision="(?P<revision_id>[^"]+)"'
82
r'.*revision="(?P<revision_id>[^"]+)"'
506
83
)
507
84
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
85
@needs_write_lock
555
86
def add_inventory(self, revision_id, inv, parents):
556
87
"""Add the inventory inv to the repository as revision_id.
557
88
558
89
:param parents: The revision ids of the parents that revision_id
559
90
is known to have and are in the repository already.
560
91
561
:returns: The validator(which is a sha1 digest, though what is sha'd is
562
repository format specific) of the serialized inventory.
92
returns the sha1 of the serialized inventory.
563
93
"""
564
if not self.is_in_write_group():
565
raise AssertionError("%r not in write group" % (self,))
94
revision_id = osutils.safe_revision_id(revision_id)
566
95
_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
96
assert inv.revision_id is None or inv.revision_id == revision_id, \
97
"Mismatch between inventory revision" \
98
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
99
assert inv.root is not None
100
inv_text = self.serialise_inventory(inv)
101
inv_sha1 = osutils.sha_string(inv_text)
102
inv_vf = self.control_weaves.get_weave('inventory',
103
self.get_transaction())
104
self._inventory_add_lines(inv_vf, revision_id, parents,
105
osutils.split_lines(inv_text))
106
return inv_sha1
107
108
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
109
final_parents = []
110
for parent in parents:
111
if parent in inv_vf:
112
final_parents.append(parent)
113
114
inv_vf.add_lines(revision_id, final_parents, lines)
115
116
@needs_write_lock
585
117
def add_revision(self, revision_id, rev, inv=None, config=None):
586
118
"""Add rev to the revision store as revision_id.
587
119
593
125
If supplied its signature_needed method will be used
594
126
to determine if a signature should be made.
595
127
"""
128
revision_id = osutils.safe_revision_id(revision_id)
596
129
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
597
130
# rev.parent_ids?
598
131
_mod_revision.check_not_reserved_id(revision_id)
602
135
plaintext = Testament(rev, inv).as_short_text()
603
136
self.store_revision_signature(
604
137
gpg.GPGStrategy(config), plaintext, revision_id)
605
# check inventory present
606
if not self.inventories.get_parent_map([(revision_id,)]):
138
if not revision_id in self.get_inventory_weave():
607
139
if inv is None:
608
140
raise errors.WeaveRevisionNotPresent(revision_id,
609
self.inventories)
141
self.get_inventory_weave())
610
142
else:
611
143
# yes, this is not suitable for adding with ghosts.
612
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
613
rev.parent_ids)
614
else:
615
key = (revision_id,)
616
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
617
self._add_revision(rev)
618
619
def _add_revision(self, revision):
620
text = self._serializer.write_revision_to_string(revision)
621
key = (revision.revision_id,)
622
parents = tuple((parent,) for parent in revision.parent_ids)
623
self.revisions.add_lines(key, parents, osutils.split_lines(text))
144
self.add_inventory(revision_id, inv, rev.parent_ids)
145
self._revision_store.add_revision(rev, self.get_transaction())
146
147
def _add_revision_text(self, revision_id, text):
148
revision = self._revision_store._serializer.read_revision_from_string(
149
text)
150
self._revision_store._add_revision(revision, StringIO(text),
151
self.get_transaction())
152
153
@needs_read_lock
154
def _all_possible_ids(self):
155
"""Return all the possible revisions that we could find."""
156
return self.get_inventory_weave().versions()
624
157
625
158
def all_revision_ids(self):
626
159
"""Returns a list of all the revision ids in the repository.
627
160
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.
161
This is deprecated because code should generally work on the graph
162
reachable from a particular revision, and ignore any other revisions
163
that might be present. There is no direct replacement method.
632
164
"""
633
if 'evil' in debug.debug_flags:
634
mutter_callsite(2, "all_revision_ids is linear with history.")
635
165
return self._all_revision_ids()
636
166
167
@needs_read_lock
637
168
def _all_revision_ids(self):
638
169
"""Returns a list of all the revision ids in the repository.
639
170
640
171
These are in as much topological order as the underlying store can
641
present.
172
present: for weaves ghosts may lead to a lack of correctness until
173
the reweave updates the parents list.
642
174
"""
643
raise NotImplementedError(self._all_revision_ids)
175
if self._revision_store.text_store.listable():
176
return self._revision_store.all_revision_ids(self.get_transaction())
177
result = self._all_possible_ids()
178
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
179
# ids. (It should, since _revision_store's API should change to
180
# return utf8 revision_ids)
181
return self._eliminate_revisions_not_present(result)
644
182
645
183
def break_lock(self):
646
184
"""Break a lock if one is present from another instance.
657
195
Returns a set of the present revisions.
658
196
"""
659
197
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()
198
for id in revision_ids:
199
if self.has_revision(id):
200
result.append(id)
201
return result
664
202
665
203
@staticmethod
666
204
def create(a_bzrdir):
667
205
"""Construct the current default format repository in a_bzrdir."""
668
206
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
669
207
670
def __init__(self, _format, a_bzrdir, control_files):
208
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
671
209
"""instantiate a Repository.
672
210
673
211
:param _format: The format of the repository on disk.
682
220
# the following are part of the public API for Repository:
683
221
self.bzrdir = a_bzrdir
684
222
self.control_files = control_files
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
223
self._revision_store = _revision_store
224
self.text_store = text_store
225
# backwards compatibility
226
self.weave_store = text_store
691
227
# not right yet - should be more semantically clear ?
692
228
#
229
self.control_store = control_store
230
self.control_weaves = control_store
693
231
# TODO: make sure to construct the right store classes, etc, depending
694
232
# on whether escaping is required.
695
233
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
711
234
712
235
def __repr__(self):
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
236
return '%s(%r)' % (self.__class__.__name__,
237
self.bzrdir.transport.base)
733
238
734
239
def is_locked(self):
735
240
return self.control_files.is_locked()
736
241
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
242
def lock_write(self, token=None):
742
243
"""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
244
748
245
:param token: if this is already locked, then lock_write will fail
749
246
unless the token matches the existing lock.
752
249
instance doesn't support using token locks.
753
250
:raises MismatchedToken: if the specified token doesn't match the token
754
251
of the existing lock.
755
:seealso: start_write_group.
756
252
757
253
A token should be passed in if you know that you have locked the object
758
254
some other way, and need to synchronise this object's state with that
760
256
761
257
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
762
258
"""
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
259
return self.control_files.lock_write(token=token)
769
260
770
261
def lock_read(self):
771
262
self.control_files.lock_read()
772
for repo in self._fallback_repositories:
773
repo.lock_read()
774
self._refresh_data()
775
263
776
264
def get_physical_lock_status(self):
777
265
return self.control_files.get_physical_lock_status()
836
324
last_revision.timezone)
837
325
838
326
# now gather global repository information
839
# XXX: This is available for many repos regardless of listability.
840
327
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
328
c, t = self._revision_store.total_size(self.get_transaction())
329
result['revisions'] = c
330
result['size'] = t
846
331
return result
847
332
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)
333
@needs_read_lock
334
def missing_revision_ids(self, other, revision_id=None):
335
"""Return the revision ids that other has that this does not.
336
337
These are returned in topological order.
338
339
revision_id: only return revision ids included by revision_id.
340
"""
341
revision_id = osutils.safe_revision_id(revision_id)
342
return InterRepository.get(other, self).missing_revision_ids(revision_id)
919
343
920
344
@staticmethod
921
345
def open(base):
933
357
This is a destructive operation! Do not use it on existing
934
358
repositories.
935
359
"""
360
revision_id = osutils.safe_revision_id(revision_id)
936
361
return InterRepository.get(self, destination).copy_content(revision_id)
937
362
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):
363
def fetch(self, source, revision_id=None, pb=None):
961
364
"""Fetch the content required to construct revision_id from source.
962
365
963
366
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).
967
367
"""
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.
368
revision_id = osutils.safe_revision_id(revision_id)
979
369
inter = InterRepository.get(source, self)
980
return inter.fetch(revision_id=revision_id, pb=pb,
981
find_ghosts=find_ghosts)
370
try:
371
return inter.fetch(revision_id=revision_id, pb=pb)
372
except NotImplementedError:
373
raise errors.IncompatibleRepositories(source, self)
982
374
983
375
def create_bundle(self, target, base, fileobj, format=None):
984
376
return serializer.write_bundle(self, target, base, fileobj, format)
985
377
986
def get_commit_builder(self, branch, parents, config, timestamp=None,
987
timezone=None, committer=None, revprops=None,
378
def get_commit_builder(self, branch, parents, config, timestamp=None,
379
timezone=None, committer=None, revprops=None,
988
380
revision_id=None):
989
381
"""Obtain a CommitBuilder for this repository.
990
382
997
389
:param revprops: Optional dictionary of revision properties.
998
390
:param revision_id: Optional revision id.
999
391
"""
1000
result = self._commit_builder_class(self, parents, config,
1001
timestamp, timezone, committer, revprops, revision_id)
1002
self.start_write_group()
1003
return result
392
revision_id = osutils.safe_revision_id(revision_id)
393
return _CommitBuilder(self, parents, config, timestamp, timezone,
394
committer, revprops, revision_id)
1004
395
1005
396
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.')
1013
397
self.control_files.unlock()
1014
for repo in self._fallback_repositories:
1015
repo.unlock()
1016
398
1017
399
@needs_read_lock
1018
400
def clone(self, a_bzrdir, revision_id=None):
1029
411
self.copy_content_into(dest_repo, revision_id)
1030
412
return dest_repo
1031
413
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
414
@needs_read_lock
1062
415
def sprout(self, to_bzrdir, revision_id=None):
1063
416
"""Create a descendent repository for new development.
1084
437
@needs_read_lock
1085
438
def has_revision(self, revision_id):
1086
439
"""True if this repository has a copy of the revision."""
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]
440
revision_id = osutils.safe_revision_id(revision_id)
441
return self._revision_store.has_revision_id(revision_id,
442
self.get_transaction())
1108
443
1109
444
@needs_read_lock
1110
445
def get_revision_reconcile(self, revision_id):
1115
450
be used by reconcile, or reconcile-alike commands that are correcting
1116
451
or testing the revision graph.
1117
452
"""
1118
return self._get_revisions([revision_id])[0]
453
if not revision_id or not isinstance(revision_id, basestring):
454
raise errors.InvalidRevisionId(revision_id=revision_id,
455
branch=self)
456
return self.get_revisions([revision_id])[0]
1119
457
1120
458
@needs_read_lock
1121
459
def get_revisions(self, revision_ids):
1122
"""Get many revisions at once."""
1123
return self._get_revisions(revision_ids)
1124
1125
@needs_read_lock
1126
def _get_revisions(self, revision_ids):
1127
"""Core work logic to get many revisions without sanity checks."""
1128
for rev_id in revision_ids:
1129
if not rev_id or not isinstance(rev_id, basestring):
1130
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1131
keys = [(key,) for key in revision_ids]
1132
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1133
revs = {}
1134
for record in stream:
1135
if record.storage_kind == 'absent':
1136
raise errors.NoSuchRevision(self, record.key[0])
1137
text = record.get_bytes_as('fulltext')
1138
rev = self._serializer.read_revision_from_string(text)
1139
revs[record.key[0]] = rev
1140
return [revs[revid] for revid in revision_ids]
460
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
461
revs = self._revision_store.get_revisions(revision_ids,
462
self.get_transaction())
463
for rev in revs:
464
assert not isinstance(rev.revision_id, unicode)
465
for parent_id in rev.parent_ids:
466
assert not isinstance(parent_id, unicode)
467
return revs
1141
468
1142
469
@needs_read_lock
1143
470
def get_revision_xml(self, revision_id):
1144
471
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1145
472
# would have already do it.
1146
473
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
474
revision_id = osutils.safe_revision_id(revision_id)
1147
475
rev = self.get_revision(revision_id)
1148
476
rev_tmp = StringIO()
1149
477
# the current serializer..
1150
self._serializer.write_revision(rev, rev_tmp)
478
self._revision_store._serializer.write_revision(rev, rev_tmp)
1151
479
rev_tmp.seek(0)
1152
480
return rev_tmp.getvalue()
1153
481
482
@needs_read_lock
483
def get_revision(self, revision_id):
484
"""Return the Revision object for a named revision"""
485
# TODO: jam 20070210 get_revision_reconcile should do this for us
486
revision_id = osutils.safe_revision_id(revision_id)
487
r = self.get_revision_reconcile(revision_id)
488
# weave corruption can lead to absent revision markers that should be
489
# present.
490
# the following test is reasonably cheap (it needs a single weave read)
491
# and the weave is cached in read transactions. In write transactions
492
# it is not cached but typically we only read a small number of
493
# revisions. For knits when they are introduced we will probably want
494
# to ensure that caching write transactions are in use.
495
inv = self.get_inventory_weave()
496
self._check_revision_parents(r, inv)
497
return r
498
499
@needs_read_lock
1154
500
def get_deltas_for_revisions(self, revisions):
1155
501
"""Produce a generator of revision deltas.
1156
502
1181
527
r = self.get_revision(revision_id)
1182
528
return list(self.get_deltas_for_revisions([r]))[0]
1183
529
530
def _check_revision_parents(self, revision, inventory):
531
"""Private to Repository and Fetch.
532
533
This checks the parentage of revision in an inventory weave for
534
consistency and is only applicable to inventory-weave-for-ancestry
535
using repository formats & fetchers.
536
"""
537
weave_parents = inventory.get_parents(revision.revision_id)
538
weave_names = inventory.versions()
539
for parent_id in revision.parent_ids:
540
if parent_id in weave_names:
541
# this parent must not be a ghost.
542
if not parent_id in weave_parents:
543
# but it is a ghost
544
raise errors.CorruptRepository(self)
545
1184
546
@needs_write_lock
1185
547
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
548
revision_id = osutils.safe_revision_id(revision_id)
1186
549
signature = gpg_strategy.sign(plaintext)
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
550
self._revision_store.add_revision_signature_text(revision_id,
551
signature,
552
self.get_transaction())
553
554
def fileids_altered_by_revision_ids(self, revision_ids):
555
"""Find the file ids and versions affected by revisions.
556
557
:param revisions: an iterable containing revision ids.
558
:return: a dictionary mapping altered file-ids to an iterable of
1198
559
revision_ids. Each altered file-ids has the exact revision_ids that
1199
560
altered it listed explicitly.
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)
561
"""
562
assert self._serializer.support_altered_by_hack, \
563
("fileids_altered_by_revision_ids only supported for branches "
564
"which store inventory as unnested xml, not on %r" % self)
565
selected_revision_ids = set(osutils.safe_revision_id(r)
566
for r in revision_ids)
567
w = self.get_inventory_weave()
1232
568
result = {}
1233
569
1234
570
# this code needs to read every new line in every inventory for the
1250
586
search = self._file_ids_altered_regex.search
1251
587
unescape = _unescape_xml
1252
588
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)
589
pb = ui.ui_factory.nested_progress_bar()
590
try:
591
for line in w.iter_lines_added_or_present_in_versions(
592
selected_revision_ids, pb=pb):
593
match = search(line)
594
if match is None:
595
continue
596
# One call to match.group() returning multiple items is quite a
597
# bit faster than 2 calls to match.group() each returning 1
598
file_id, revision_id = match.group('file_id', 'revision_id')
599
600
# Inlining the cache lookups helps a lot when you make 170,000
601
# lines and 350k ids, versus 8.4 unique ids.
602
# Using a cache helps in 2 ways:
603
# 1) Avoids unnecessary decoding calls
604
# 2) Re-uses cached strings, which helps in future set and
605
# equality checks.
606
# (2) is enough that removing encoding entirely along with
607
# the cache (so we are using plain strings) results in no
608
# performance improvement.
609
try:
610
revision_id = unescape_revid_cache[revision_id]
611
except KeyError:
612
unescaped = unescape(revision_id)
613
unescape_revid_cache[revision_id] = unescaped
614
revision_id = unescaped
615
616
if revision_id in selected_revision_ids:
617
try:
618
file_id = unescape_fileid_cache[file_id]
619
except KeyError:
620
unescaped = unescape(file_id)
621
unescape_fileid_cache[file_id] = unescaped
622
file_id = unescaped
623
setdefault(file_id, set()).add(revision_id)
624
finally:
625
pb.finished()
626
return result
627
628
@needs_read_lock
629
def get_inventory_weave(self):
630
return self.control_weaves.get_weave('inventory',
631
self.get_transaction())
1522
632
1523
633
@needs_read_lock
1524
634
def get_inventory(self, 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]
635
"""Get Inventory object by hash."""
636
# TODO: jam 20070210 Technically we don't need to sanitize, since all
637
# called functions must sanitize.
638
revision_id = osutils.safe_revision_id(revision_id)
639
return self.deserialise_inventory(
640
revision_id, self.get_inventory_xml(revision_id))
1558
641
1559
642
def deserialise_inventory(self, revision_id, xml):
1560
643
"""Transform the xml into an inventory object.
1562
645
:param revision_id: The expected revision id of the inventory.
1563
646
:param xml: A serialised inventory.
1564
647
"""
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))
648
revision_id = osutils.safe_revision_id(revision_id)
649
result = self._serializer.read_inventory_from_string(xml)
650
result.root.revision = revision_id
1569
651
return result
1570
652
1571
653
def serialise_inventory(self, inv):
1572
654
return self._serializer.write_inventory_to_string(inv)
1573
655
1574
def _serialise_inventory_to_lines(self, inv):
1575
return self._serializer.write_inventory_to_lines(inv)
1576
1577
656
def get_serializer_format(self):
1578
657
return self._serializer.format_num
1579
658
1580
659
@needs_read_lock
1581
660
def get_inventory_xml(self, revision_id):
1582
661
"""Get inventory XML as a file object."""
1583
texts = self._iter_inventory_xmls([revision_id])
662
revision_id = osutils.safe_revision_id(revision_id)
1584
663
try:
1585
text, revision_id = texts.next()
1586
except StopIteration:
664
assert isinstance(revision_id, str), type(revision_id)
665
iw = self.get_inventory_weave()
666
return iw.get_text(revision_id)
667
except IndexError:
1587
668
raise errors.HistoryMissing(self, 'inventory', revision_id)
1588
return text
1589
669
1590
670
@needs_read_lock
1591
671
def get_inventory_sha1(self, revision_id):
1592
672
"""Return the sha1 hash of the inventory entry
1593
673
"""
674
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
675
revision_id = osutils.safe_revision_id(revision_id)
1594
676
return self.get_revision(revision_id).inventory_sha1
1595
677
678
@needs_read_lock
679
def get_revision_graph(self, revision_id=None):
680
"""Return a dictionary containing the revision graph.
681
682
:param revision_id: The revision_id to get a graph from. If None, then
683
the entire revision graph is returned. This is a deprecated mode of
684
operation and will be removed in the future.
685
:return: a dictionary of revision_id->revision_parents_list.
686
"""
687
# special case NULL_REVISION
688
if revision_id == _mod_revision.NULL_REVISION:
689
return {}
690
revision_id = osutils.safe_revision_id(revision_id)
691
a_weave = self.get_inventory_weave()
692
all_revisions = self._eliminate_revisions_not_present(
693
a_weave.versions())
694
entire_graph = dict([(node, a_weave.get_parents(node)) for
695
node in all_revisions])
696
if revision_id is None:
697
return entire_graph
698
elif revision_id not in entire_graph:
699
raise errors.NoSuchRevision(self, revision_id)
700
else:
701
# add what can be reached from revision_id
702
result = {}
703
pending = set([revision_id])
704
while len(pending) > 0:
705
node = pending.pop()
706
result[node] = entire_graph[node]
707
for revision_id in result[node]:
708
if revision_id not in result:
709
pending.add(revision_id)
710
return result
711
712
@needs_read_lock
713
def get_revision_graph_with_ghosts(self, revision_ids=None):
714
"""Return a graph of the revisions with ghosts marked as applicable.
715
716
:param revision_ids: an iterable of revisions to graph or None for all.
717
:return: a Graph object with the graph reachable from revision_ids.
718
"""
719
result = deprecated_graph.Graph()
720
if not revision_ids:
721
pending = set(self.all_revision_ids())
722
required = set([])
723
else:
724
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
725
# special case NULL_REVISION
726
if _mod_revision.NULL_REVISION in pending:
727
pending.remove(_mod_revision.NULL_REVISION)
728
required = set(pending)
729
done = set([])
730
while len(pending):
731
revision_id = pending.pop()
732
try:
733
rev = self.get_revision(revision_id)
734
except errors.NoSuchRevision:
735
if revision_id in required:
736
raise
737
# a ghost
738
result.add_ghost(revision_id)
739
continue
740
for parent_id in rev.parent_ids:
741
# is this queued or done ?
742
if (parent_id not in pending and
743
parent_id not in done):
744
# no, queue it.
745
pending.add(parent_id)
746
result.add_node(revision_id, rev.parent_ids)
747
done.add(revision_id)
748
return result
749
750
def _get_history_vf(self):
751
"""Get a versionedfile whose history graph reflects all revisions.
752
753
For weave repositories, this is the inventory weave.
754
"""
755
return self.get_inventory_weave()
756
1596
757
def iter_reverse_revision_history(self, revision_id):
1597
758
"""Iterate backwards through revision ids in the lefthand history
1598
759
1599
760
:param revision_id: The revision id to start with. All its lefthand
1600
761
ancestors will be traversed.
1601
762
"""
1602
graph = self.get_graph()
763
revision_id = osutils.safe_revision_id(revision_id)
764
if revision_id in (None, _mod_revision.NULL_REVISION):
765
return
1603
766
next_id = revision_id
767
versionedfile = self._get_history_vf()
1604
768
while True:
1605
if next_id in (None, _mod_revision.NULL_REVISION):
1606
return
1607
769
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]
770
parents = versionedfile.get_parents(next_id)
1614
771
if len(parents) == 0:
1615
772
return
1616
773
else:
1632
789
else:
1633
790
return self.get_inventory(revision_id)
1634
791
792
@needs_read_lock
1635
793
def is_shared(self):
1636
794
"""Return True if this repository is flagged as a shared repository."""
1637
795
raise NotImplementedError(self.is_shared)
1644
802
reconciler.reconcile()
1645
803
return reconciler
1646
804
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
1657
805
@needs_read_lock
1658
806
def revision_tree(self, revision_id):
1659
807
"""Return Tree for a revision on this branch.
1666
814
return RevisionTree(self, Inventory(root_id=None),
1667
815
_mod_revision.NULL_REVISION)
1668
816
else:
817
revision_id = osutils.safe_revision_id(revision_id)
1669
818
inv = self.get_revision_inventory(revision_id)
1670
819
return RevisionTree(self, inv, revision_id)
1671
820
821
@needs_read_lock
1672
822
def revision_trees(self, revision_ids):
1673
823
"""Return Tree for a revision on this branch.
1674
824
1675
825
`revision_id` may not be None or 'null:'"""
1676
inventories = self.iter_inventories(revision_ids)
1677
for inv in inventories:
1678
yield RevisionTree(self, inv, inv.revision_id)
826
assert None not in revision_ids
827
assert _mod_revision.NULL_REVISION not in revision_ids
828
texts = self.get_inventory_weave().get_texts(revision_ids)
829
for text, revision_id in zip(texts, revision_ids):
830
inv = self.deserialise_inventory(revision_id, text)
831
yield RevisionTree(self, inv, revision_id)
1679
832
1680
833
@needs_read_lock
1681
834
def get_ancestry(self, revision_id, topo_sorted=True):
1687
840
1688
841
This is topologically sorted.
1689
842
"""
1690
if _mod_revision.is_null(revision_id):
843
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1691
844
return [None]
845
revision_id = osutils.safe_revision_id(revision_id)
1692
846
if not self.has_revision(revision_id):
1693
847
raise errors.NoSuchRevision(self, revision_id)
1694
graph = self.get_graph()
1695
keys = set()
1696
search = graph._make_breadth_first_searcher([revision_id])
1697
while True:
1698
try:
1699
found, ghosts = search.next_with_ghosts()
1700
except StopIteration:
1701
break
1702
keys.update(found)
1703
if _mod_revision.NULL_REVISION in keys:
1704
keys.remove(_mod_revision.NULL_REVISION)
1705
if topo_sorted:
1706
parent_map = graph.get_parent_map(keys)
1707
keys = tsort.topo_sort(parent_map)
1708
return [None] + list(keys)
1709
1710
def pack(self):
1711
"""Compress the data within the repository.
1712
1713
This operation only makes sense for some repository types. For other
1714
types it should be a no-op that just returns.
1715
1716
This stub method does not require a lock, but subclasses should use
1717
@needs_write_lock as this is a long running call its reasonable to
1718
implicitly lock for the user.
1719
"""
848
w = self.get_inventory_weave()
849
candidates = w.get_ancestry(revision_id, topo_sorted)
850
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1720
851
1721
852
@needs_read_lock
1722
@deprecated_method(one_six)
1723
853
def print_file(self, file, revision_id):
1724
854
"""Print `file` to stdout.
1725
855
1727
857
- it writes to stdout, it assumes that that is valid etc. Fix
1728
858
by creating a new more flexible convenience function.
1729
859
"""
860
revision_id = osutils.safe_revision_id(revision_id)
1730
861
tree = self.revision_tree(revision_id)
1731
862
# use inventory as it was in that revision
1732
863
file_id = tree.inventory.path2id(file)
1740
871
def get_transaction(self):
1741
872
return self.control_files.get_transaction()
1742
873
1743
@deprecated_method(one_one)
874
def revision_parents(self, revision_id):
875
revision_id = osutils.safe_revision_id(revision_id)
876
return self.get_inventory_weave().parent_names(revision_id)
877
1744
878
def get_parents(self, revision_ids):
1745
879
"""See StackedParentsProvider.get_parents"""
1746
parent_map = self.get_parent_map(revision_ids)
1747
return [parent_map.get(r, None) for r in revision_ids]
1748
1749
def get_parent_map(self, revision_ids):
1750
"""See graph._StackedParentsProvider.get_parent_map"""
1751
# revisions index works in keys; this just works in revisions
1752
# therefore wrap and unwrap
1753
query_keys = []
1754
result = {}
880
parents_list = []
1755
881
for revision_id in revision_ids:
1756
882
if revision_id == _mod_revision.NULL_REVISION:
1757
result[revision_id] = ()
1758
elif revision_id is None:
1759
raise ValueError('get_parent_map(None) is not valid')
1760
else:
1761
query_keys.append((revision_id ,))
1762
for ((revision_id,), parent_keys) in \
1763
self.revisions.get_parent_map(query_keys).iteritems():
1764
if parent_keys:
1765
result[revision_id] = tuple(parent_revid
1766
for (parent_revid,) in parent_keys)
1767
else:
1768
result[revision_id] = (_mod_revision.NULL_REVISION,)
1769
return result
883
parents = []
884
else:
885
try:
886
parents = self.get_revision(revision_id).parent_ids
887
except errors.NoSuchRevision:
888
parents = None
889
else:
890
if len(parents) == 0:
891
parents = [_mod_revision.NULL_REVISION]
892
parents_list.append(parents)
893
return parents_list
1770
894
1771
895
def _make_parents_provider(self):
1772
896
return self
1775
899
"""Return the graph walker for this repository format"""
1776
900
parents_provider = self._make_parents_provider()
1777
901
if (other_repository is not None and
1778
not self.has_same_location(other_repository)):
902
other_repository.bzrdir.transport.base !=
903
self.bzrdir.transport.base):
1779
904
parents_provider = graph._StackedParentsProvider(
1780
905
[parents_provider, other_repository._make_parents_provider()])
1781
906
return graph.Graph(parents_provider)
1782
907
1783
def _get_versioned_file_checker(self):
1784
"""Return an object suitable for checking versioned files."""
1785
return _VersionedFileChecker(self)
1786
1787
def revision_ids_to_search_result(self, result_set):
1788
"""Convert a set of revision ids to a graph SearchResult."""
1789
result_parents = set()
1790
for parents in self.get_graph().get_parent_map(
1791
result_set).itervalues():
1792
result_parents.update(parents)
1793
included_keys = result_set.intersection(result_parents)
1794
start_keys = result_set.difference(included_keys)
1795
exclude_keys = result_parents.difference(result_set)
1796
result = graph.SearchResult(start_keys, exclude_keys,
1797
len(result_set), result_set)
1798
return result
1799
1800
908
@needs_write_lock
1801
909
def set_make_working_trees(self, new_value):
1802
910
"""Set the policy flag for making working trees when creating branches.
1815
923
1816
924
@needs_write_lock
1817
925
def sign_revision(self, revision_id, gpg_strategy):
926
revision_id = osutils.safe_revision_id(revision_id)
1818
927
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1819
928
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1820
929
1821
930
@needs_read_lock
1822
931
def has_signature_for_revision_id(self, revision_id):
1823
932
"""Query for a revision signature for revision_id in the repository."""
1824
if not self.has_revision(revision_id):
1825
raise errors.NoSuchRevision(self, revision_id)
1826
sig_present = (1 == len(
1827
self.signatures.get_parent_map([(revision_id,)])))
1828
return sig_present
933
revision_id = osutils.safe_revision_id(revision_id)
934
return self._revision_store.has_signature(revision_id,
935
self.get_transaction())
1829
936
1830
937
@needs_read_lock
1831
938
def get_signature_text(self, revision_id):
1832
939
"""Return the text for a signature."""
1833
stream = self.signatures.get_record_stream([(revision_id,)],
1834
'unordered', True)
1835
record = stream.next()
1836
if record.storage_kind == 'absent':
1837
raise errors.NoSuchRevision(self, revision_id)
1838
return record.get_bytes_as('fulltext')
940
revision_id = osutils.safe_revision_id(revision_id)
941
return self._revision_store.get_signature_text(revision_id,
942
self.get_transaction())
1839
943
1840
944
@needs_read_lock
1841
def check(self, revision_ids=None):
945
def check(self, revision_ids):
1842
946
"""Check consistency of all history of given revision_ids.
1843
947
1844
948
Different repository implementations should override _check().
1846
950
:param revision_ids: A non-empty list of revision_ids whose ancestry
1847
951
will be checked. Typically the last revision_id of a branch.
1848
952
"""
953
if not revision_ids:
954
raise ValueError("revision_ids must be non-empty in %s.check"
955
% (self,))
956
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1849
957
return self._check(revision_ids)
1850
958
1851
959
def _check(self, revision_ids):
1879
987
revision_id.decode('ascii')
1880
988
except UnicodeDecodeError:
1881
989
raise errors.NonAsciiRevisionId(method, self)
1882
1883
def revision_graph_can_have_wrong_parents(self):
1884
"""Is it possible for this repository to have a revision graph with
1885
incorrect parents?
1886
990
1887
If True, then this repository must also implement
1888
_find_inconsistent_revision_parents so that check and reconcile can
1889
check for inconsistencies before proceeding with other checks that may
1890
depend on the revision index being consistent.
1891
"""
1892
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1893
991
1894
992
1895
993
# remove these delegates a while after bzr 0.15
1928
1026
1929
1027
def install_revision(repository, rev, revision_tree):
1930
1028
"""Install all revision data into a repository."""
1931
install_revisions(repository, [(rev, revision_tree, None)])
1932
1933
1934
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1935
"""Install all revision data into a repository.
1936
1937
Accepts an iterable of revision, tree, signature tuples. The signature
1938
may be None.
1939
"""
1940
repository.start_write_group()
1941
try:
1942
for n, (revision, revision_tree, signature) in enumerate(iterable):
1943
_install_revision(repository, revision, revision_tree, signature)
1944
if pb is not None:
1945
pb.update('Transferring revisions', n + 1, num_revisions)
1946
except:
1947
repository.abort_write_group()
1948
raise
1949
else:
1950
repository.commit_write_group()
1951
1952
1953
def _install_revision(repository, rev, revision_tree, signature):
1954
"""Install all revision data into a repository."""
1955
1029
present_parents = []
1956
1030
parent_trees = {}
1957
1031
for p_id in rev.parent_ids:
1963
1037
1964
1038
inv = revision_tree.inventory
1965
1039
entries = inv.iter_entries()
1966
# backwards compatibility hack: skip the root id.
1040
# backwards compatability hack: skip the root id.
1967
1041
if not repository.supports_rich_root():
1968
1042
path, root = entries.next()
1969
1043
if root.revision != rev.revision_id:
1970
1044
raise errors.IncompatibleRevision(repr(repository))
1971
text_keys = {}
1045
# Add the texts that are not already present
1972
1046
for path, ie in entries:
1973
text_keys[(ie.file_id, ie.revision)] = ie
1974
text_parent_map = repository.texts.get_parent_map(text_keys)
1975
missing_texts = set(text_keys) - set(text_parent_map)
1976
# Add the texts that are not already present
1977
for text_key in missing_texts:
1978
ie = text_keys[text_key]
1979
text_parents = []
1980
# FIXME: TODO: The following loop overlaps/duplicates that done by
1981
# commit to determine parents. There is a latent/real bug here where
1982
# the parents inserted are not those commit would do - in particular
1983
# they are not filtered by heads(). RBC, AB
1984
for revision, tree in parent_trees.iteritems():
1985
if ie.file_id not in tree:
1986
continue
1987
parent_id = tree.inventory[ie.file_id].revision
1988
if parent_id in text_parents:
1989
continue
1990
text_parents.append((ie.file_id, parent_id))
1991
lines = revision_tree.get_file(ie.file_id).readlines()
1992
repository.texts.add_lines(text_key, text_parents, lines)
1047
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1048
repository.get_transaction())
1049
if ie.revision not in w:
1050
text_parents = []
1051
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1052
# with InventoryEntry.find_previous_heads(). if it is, then there
1053
# is a latent bug here where the parents may have ancestors of each
1054
# other. RBC, AB
1055
for revision, tree in parent_trees.iteritems():
1056
if ie.file_id not in tree:
1057
continue
1058
parent_id = tree.inventory[ie.file_id].revision
1059
if parent_id in text_parents:
1060
continue
1061
text_parents.append(parent_id)
1062
1063
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1064
repository.get_transaction())
1065
lines = revision_tree.get_file(ie.file_id).readlines()
1066
vfile.add_lines(rev.revision_id, text_parents, lines)
1993
1067
try:
1994
1068
# install the inventory
1995
1069
repository.add_inventory(rev.revision_id, inv, present_parents)
1996
1070
except errors.RevisionAlreadyPresent:
1997
1071
pass
1998
if signature is not None:
1999
repository.add_signature_text(rev.revision_id, signature)
2000
1072
repository.add_revision(rev.revision_id, rev, inv)
2001
1073
2002
1074
2003
1075
class MetaDirRepository(Repository):
2004
"""Repositories in the new meta-dir layout.
2005
2006
:ivar _transport: Transport for access to repository control files,
2007
typically pointing to .bzr/repository.
2008
"""
2009
2010
def __init__(self, _format, a_bzrdir, control_files):
2011
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2012
self._transport = control_files._transport
2013
1076
"""Repositories in the new meta-dir layout."""
1077
1078
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1079
super(MetaDirRepository, self).__init__(_format,
1080
a_bzrdir,
1081
control_files,
1082
_revision_store,
1083
control_store,
1084
text_store)
1085
dir_mode = self.control_files._dir_mode
1086
file_mode = self.control_files._file_mode
1087
1088
@needs_read_lock
2014
1089
def is_shared(self):
2015
1090
"""Return True if this repository is flagged as a shared repository."""
2016
return self._transport.has('shared-storage')
1091
return self.control_files._transport.has('shared-storage')
2017
1092
2018
1093
@needs_write_lock
2019
1094
def set_make_working_trees(self, new_value):
2027
1102
"""
2028
1103
if new_value:
2029
1104
try:
2030
self._transport.delete('no-working-trees')
1105
self.control_files._transport.delete('no-working-trees')
2031
1106
except errors.NoSuchFile:
2032
1107
pass
2033
1108
else:
2034
self._transport.put_bytes('no-working-trees', '',
2035
mode=self.bzrdir._get_file_mode())
1109
self.control_files.put_utf8('no-working-trees', '')
2036
1110
2037
1111
def make_working_trees(self):
2038
1112
"""Returns the policy for making working trees on new branches."""
2039
return not self._transport.has('no-working-trees')
2040
2041
2042
class MetaDirVersionedFileRepository(MetaDirRepository):
2043
"""Repositories in a meta-dir, that work via versioned file objects."""
2044
2045
def __init__(self, _format, a_bzrdir, control_files):
2046
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2047
control_files)
1113
return not self.control_files._transport.has('no-working-trees')
2048
1114
2049
1115
2050
1116
class RepositoryFormatRegistry(registry.Registry):
2051
"""Registry of RepositoryFormats."""
1117
"""Registry of RepositoryFormats.
1118
"""
2052
1119
2053
1120
def get(self, format_string):
2054
1121
r = registry.Registry.get(self, format_string)
2077
1144
children.
2078
1145
* an open routine which returns a Repository instance.
2079
1146
2080
There is one and only one Format subclass for each on-disk format. But
2081
there can be one Repository subclass that is used for several different
2082
formats. The _format attribute on a Repository instance can be used to
2083
determine the disk format.
2084
2085
1147
Formats are placed in an dict by their format string for reference
2086
1148
during opening. These should be subclasses of RepositoryFormat
2087
1149
for consistency.
2094
1156
_matchingbzrdir - the bzrdir format that the repository format was
2095
1157
originally written to work with. This can be used if manually
2096
1158
constructing a bzrdir and repository, or more commonly for test suite
2097
parameterization.
1159
parameterisation.
2098
1160
"""
2099
1161
2100
# Set to True or False in derived classes. True indicates that the format
2101
# supports ghosts gracefully.
2102
supports_ghosts = None
2103
# Can this repository be given external locations to lookup additional
2104
# data. Set to True or False in derived classes.
2105
supports_external_lookups = None
2106
2107
1162
def __str__(self):
2108
1163
return "<%s>" % self.__class__.__name__
2109
1164
2129
1184
except errors.NoSuchFile:
2130
1185
raise errors.NoRepositoryPresent(a_bzrdir)
2131
1186
except KeyError:
2132
raise errors.UnknownFormatError(format=format_string,
2133
kind='repository')
1187
raise errors.UnknownFormatError(format=format_string)
2134
1188
2135
1189
@classmethod
2136
1190
def register_format(klass, format):
2146
1200
from bzrlib import bzrdir
2147
1201
return bzrdir.format_registry.make_bzrdir('default').repository_format
2148
1202
1203
def _get_control_store(self, repo_transport, control_files):
1204
"""Return the control store for this repository."""
1205
raise NotImplementedError(self._get_control_store)
1206
2149
1207
def get_format_string(self):
2150
1208
"""Return the ASCII format string that identifies this format.
2151
1209
2158
1216
"""Return the short description for this format."""
2159
1217
raise NotImplementedError(self.get_format_description)
2160
1218
1219
def _get_revision_store(self, repo_transport, control_files):
1220
"""Return the revision store object for this a_bzrdir."""
1221
raise NotImplementedError(self._get_revision_store)
1222
1223
def _get_text_rev_store(self,
1224
transport,
1225
control_files,
1226
name,
1227
compressed=True,
1228
prefixed=False,
1229
serializer=None):
1230
"""Common logic for getting a revision store for a repository.
1231
1232
see self._get_revision_store for the subclass-overridable method to
1233
get the store for a repository.
1234
"""
1235
from bzrlib.store.revision.text import TextRevisionStore
1236
dir_mode = control_files._dir_mode
1237
file_mode = control_files._file_mode
1238
text_store = TextStore(transport.clone(name),
1239
prefixed=prefixed,
1240
compressed=compressed,
1241
dir_mode=dir_mode,
1242
file_mode=file_mode)
1243
_revision_store = TextRevisionStore(text_store, serializer)
1244
return _revision_store
1245
2161
1246
# TODO: this shouldn't be in the base class, it's specific to things that
2162
1247
# use weaves or knits -- mbp 20070207
2163
1248
def _get_versioned_file_store(self,
2217
1302
2218
1303
rich_root_data = False
2219
1304
supports_tree_reference = False
2220
supports_external_lookups = False
2221
1305
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2222
1306
2223
1307
def __init__(self):
2237
1321
"""Upload the initial blank content."""
2238
1322
control_files = self._create_control_files(a_bzrdir)
2239
1323
control_files.lock_write()
2240
transport = control_files._transport
2241
if shared == True:
2242
utf8_files += [('shared-storage', '')]
2243
1324
try:
2244
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2245
for (filename, content_stream) in files:
2246
transport.put_file(filename, content_stream,
2247
mode=a_bzrdir._get_file_mode())
2248
for (filename, content_bytes) in utf8_files:
2249
transport.put_bytes_non_atomic(filename, content_bytes,
2250
mode=a_bzrdir._get_file_mode())
1325
control_files._transport.mkdir_multi(dirs,
1326
mode=control_files._dir_mode)
1327
for file, content in files:
1328
control_files.put(file, content)
1329
for file, content in utf8_files:
1330
control_files.put_utf8(file, content)
1331
if shared == True:
1332
control_files.put_utf8('shared-storage', '')
2251
1333
finally:
2252
1334
control_files.unlock()
2253
1335
2263
1345
'bzrlib.repofmt.weaverepo',
2264
1346
'RepositoryFormat7'
2265
1347
)
1348
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1349
# default control directory format
2266
1350
2267
1351
format_registry.register_lazy(
2268
1352
'Bazaar-NG Knit Repository Format 1',
2269
1353
'bzrlib.repofmt.knitrepo',
2270
1354
'RepositoryFormatKnit1',
2271
1355
)
1356
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
2272
1357
2273
1358
format_registry.register_lazy(
2274
1359
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2276
1361
'RepositoryFormatKnit3',
2277
1362
)
2278
1363
2279
format_registry.register_lazy(
2280
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2281
'bzrlib.repofmt.knitrepo',
2282
'RepositoryFormatKnit4',
2283
)
2284
2285
# Pack-based formats. There is one format for pre-subtrees, and one for
2286
# post-subtrees to allow ease of testing.
2287
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2288
format_registry.register_lazy(
2289
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2290
'bzrlib.repofmt.pack_repo',
2291
'RepositoryFormatKnitPack1',
2292
)
2293
format_registry.register_lazy(
2294
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2295
'bzrlib.repofmt.pack_repo',
2296
'RepositoryFormatKnitPack3',
2297
)
2298
format_registry.register_lazy(
2299
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2300
'bzrlib.repofmt.pack_repo',
2301
'RepositoryFormatKnitPack4',
2302
)
2303
format_registry.register_lazy(
2304
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2305
'bzrlib.repofmt.pack_repo',
2306
'RepositoryFormatKnitPack5',
2307
)
2308
format_registry.register_lazy(
2309
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2310
'bzrlib.repofmt.pack_repo',
2311
'RepositoryFormatKnitPack5RichRoot',
2312
)
2313
format_registry.register_lazy(
2314
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2315
'bzrlib.repofmt.pack_repo',
2316
'RepositoryFormatKnitPack5RichRootBroken',
2317
)
2318
2319
# Development formats.
2320
# 1.5->1.6
2321
format_registry.register_lazy(
2322
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2323
'bzrlib.repofmt.pack_repo',
2324
'RepositoryFormatPackDevelopment1',
2325
)
2326
format_registry.register_lazy(
2327
("Bazaar development format 1 with subtree support "
2328
"(needs bzr.dev from before 1.6)\n"),
2329
'bzrlib.repofmt.pack_repo',
2330
'RepositoryFormatPackDevelopment1Subtree',
2331
)
2332
# 1.6->1.7 go below here
2333
2334
1364
2335
1365
class InterRepository(InterObject):
2336
1366
"""This class represents operations taking place between two repositories.
2350
1380
def copy_content(self, revision_id=None):
2351
1381
raise NotImplementedError(self.copy_content)
2352
1382
2353
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1383
def fetch(self, revision_id=None, pb=None):
2354
1384
"""Fetch the content required to construct revision_id.
2355
1385
2356
1386
The content is copied from self.source to self.target.
2360
1390
:param pb: optional progress bar to use for progress reports. If not
2361
1391
provided a default one will be created.
2362
1392
2363
:returns: (copied_revision_count, failures).
2364
"""
2365
# Normally we should find a specific InterRepository subclass to do
2366
# the fetch; if nothing else then at least InterSameDataRepository.
2367
# If none of them is suitable it looks like fetching is not possible;
2368
# we try to give a good message why. _assert_same_model will probably
2369
# give a helpful message; otherwise a generic one.
2370
self._assert_same_model(self.source, self.target)
2371
raise errors.IncompatibleRepositories(self.source, self.target,
2372
"no suitableInterRepository found")
2373
2374
def _walk_to_common_revisions(self, revision_ids):
2375
"""Walk out from revision_ids in source to revisions target has.
2376
2377
:param revision_ids: The start point for the search.
2378
:return: A set of revision ids.
2379
"""
2380
target_graph = self.target.get_graph()
2381
revision_ids = frozenset(revision_ids)
2382
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2383
return graph.SearchResult(revision_ids, set(), 0, set())
2384
missing_revs = set()
2385
source_graph = self.source.get_graph()
2386
# ensure we don't pay silly lookup costs.
2387
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2388
null_set = frozenset([_mod_revision.NULL_REVISION])
2389
while True:
2390
try:
2391
next_revs, ghosts = searcher.next_with_ghosts()
2392
except StopIteration:
2393
break
2394
if revision_ids.intersection(ghosts):
2395
absent_ids = set(revision_ids.intersection(ghosts))
2396
# If all absent_ids are present in target, no error is needed.
2397
absent_ids.difference_update(
2398
set(target_graph.get_parent_map(absent_ids)))
2399
if absent_ids:
2400
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2401
# we don't care about other ghosts as we can't fetch them and
2402
# haven't been asked to.
2403
next_revs = set(next_revs)
2404
# we always have NULL_REVISION present.
2405
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2406
missing_revs.update(next_revs - have_revs)
2407
searcher.stop_searching_any(have_revs)
2408
return searcher.get_result()
1393
Returns the copied revision count and the failed revisions in a tuple:
1394
(copied, failures).
1395
"""
1396
raise NotImplementedError(self.fetch)
2409
1397
2410
@deprecated_method(one_two)
2411
1398
@needs_read_lock
2412
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1399
def missing_revision_ids(self, revision_id=None):
2413
1400
"""Return the revision ids that source has that target does not.
2414
1401
2415
1402
These are returned in topological order.
2416
1403
2417
1404
:param revision_id: only return revision ids included by this
2418
1405
revision_id.
2419
:param find_ghosts: If True find missing revisions in deep history
2420
rather than just finding the surface difference.
2421
"""
2422
return list(self.search_missing_revision_ids(
2423
revision_id, find_ghosts).get_keys())
2424
2425
@needs_read_lock
2426
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2427
"""Return the revision ids that source has that target does not.
2428
2429
:param revision_id: only return revision ids included by this
2430
revision_id.
2431
:param find_ghosts: If True find missing revisions in deep history
2432
rather than just finding the surface difference.
2433
:return: A bzrlib.graph.SearchResult.
2434
"""
2435
# stop searching at found target revisions.
2436
if not find_ghosts and revision_id is not None:
2437
return self._walk_to_common_revisions([revision_id])
1406
"""
2438
1407
# generic, possibly worst case, slow code path.
2439
1408
target_ids = set(self.target.all_revision_ids())
2440
1409
if revision_id is not None:
1410
# TODO: jam 20070210 InterRepository is internal enough that it
1411
# should assume revision_ids are already utf-8
1412
revision_id = osutils.safe_revision_id(revision_id)
2441
1413
source_ids = self.source.get_ancestry(revision_id)
2442
if source_ids[0] is not None:
2443
raise AssertionError()
1414
assert source_ids[0] is None
2444
1415
source_ids.pop(0)
2445
1416
else:
2446
1417
source_ids = self.source.all_revision_ids()
2447
1418
result_set = set(source_ids).difference(target_ids)
2448
return self.source.revision_ids_to_search_result(result_set)
2449
2450
@staticmethod
2451
def _same_model(source, target):
2452
"""True if source and target have the same data representation.
2453
2454
Note: this is always called on the base class; overriding it in a
2455
subclass will have no effect.
2456
"""
2457
try:
2458
InterRepository._assert_same_model(source, target)
2459
return True
2460
except errors.IncompatibleRepositories, e:
2461
return False
2462
2463
@staticmethod
2464
def _assert_same_model(source, target):
2465
"""Raise an exception if two repositories do not use the same model.
2466
"""
2467
if source.supports_rich_root() != target.supports_rich_root():
2468
raise errors.IncompatibleRepositories(source, target,
2469
"different rich-root support")
2470
if source._serializer != target._serializer:
2471
raise errors.IncompatibleRepositories(source, target,
2472
"different serializers")
1419
# this may look like a no-op: its not. It preserves the ordering
1420
# other_ids had while only returning the members from other_ids
1421
# that we've decided we need.
1422
return [rev_id for rev_id in source_ids if rev_id in result_set]
2473
1423
2474
1424
2475
1425
class InterSameDataRepository(InterRepository):
2480
1430
2481
1431
@classmethod
2482
1432
def _get_repo_format_to_test(self):
2483
"""Repository format for testing with.
2484
2485
InterSameData can pull from subtree to subtree and from non-subtree to
2486
non-subtree, so we test this with the richest repository format.
2487
"""
2488
from bzrlib.repofmt import knitrepo
2489
return knitrepo.RepositoryFormatKnit3()
1433
"""Repository format for testing with."""
1434
return RepositoryFormat.get_default_format()
2490
1435
2491
1436
@staticmethod
2492
1437
def is_compatible(source, target):
2493
return InterRepository._same_model(source, target)
1438
if source.supports_rich_root() != target.supports_rich_root():
1439
return False
1440
if source._serializer != target._serializer:
1441
return False
1442
return True
2494
1443
2495
1444
@needs_write_lock
2496
1445
def copy_content(self, revision_id=None):
2509
1458
self.target.set_make_working_trees(self.source.make_working_trees())
2510
1459
except NotImplementedError:
2511
1460
pass
1461
# TODO: jam 20070210 This is fairly internal, so we should probably
1462
# just assert that revision_id is not unicode.
1463
revision_id = osutils.safe_revision_id(revision_id)
2512
1464
# but don't bother fetching if we have the needed data now.
2513
1465
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2514
1466
self.target.has_revision(revision_id)):
2516
1468
self.target.fetch(self.source, revision_id=revision_id)
2517
1469
2518
1470
@needs_write_lock
2519
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1471
def fetch(self, revision_id=None, pb=None):
2520
1472
"""See InterRepository.fetch()."""
2521
from bzrlib.fetch import RepoFetcher
1473
from bzrlib.fetch import GenericRepoFetcher
2522
1474
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2523
self.source, self.source._format, self.target,
1475
self.source, self.source._format, self.target,
2524
1476
self.target._format)
2525
f = RepoFetcher(to_repository=self.target,
1477
# TODO: jam 20070210 This should be an assert, not a translate
1478
revision_id = osutils.safe_revision_id(revision_id)
1479
f = GenericRepoFetcher(to_repository=self.target,
2526
1480
from_repository=self.source,
2527
1481
last_revision=revision_id,
2528
pb=pb, find_ghosts=find_ghosts)
1482
pb=pb)
2529
1483
return f.count_copied, f.failed_revisions
2530
1484
2531
1485
2532
1486
class InterWeaveRepo(InterSameDataRepository):
2533
"""Optimised code paths between Weave based repositories.
2534
2535
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2536
implemented lazy inter-object optimisation.
2537
"""
1487
"""Optimised code paths between Weave based repositories."""
2538
1488
2539
1489
@classmethod
2540
1490
def _get_repo_format_to_test(self):
2568
1518
def copy_content(self, revision_id=None):
2569
1519
"""See InterRepository.copy_content()."""
2570
1520
# weave specific optimised path:
1521
# TODO: jam 20070210 Internal, should be an assert, not translate
1522
revision_id = osutils.safe_revision_id(revision_id)
2571
1523
try:
2572
1524
self.target.set_make_working_trees(self.source.make_working_trees())
2573
except (errors.RepositoryUpgradeRequired, NotImplemented):
1525
except NotImplementedError:
2574
1526
pass
2575
1527
# FIXME do not peek!
2576
if self.source._transport.listable():
1528
if self.source.control_files._transport.listable():
2577
1529
pb = ui.ui_factory.nested_progress_bar()
2578
1530
try:
2579
self.target.texts.insert_record_stream(
2580
self.source.texts.get_record_stream(
2581
self.source.texts.keys(), 'topological', False))
1531
self.target.weave_store.copy_all_ids(
1532
self.source.weave_store,
1533
pb=pb,
1534
from_transaction=self.source.get_transaction(),
1535
to_transaction=self.target.get_transaction())
2582
1536
pb.update('copying inventory', 0, 1)
2583
self.target.inventories.insert_record_stream(
2584
self.source.inventories.get_record_stream(
2585
self.source.inventories.keys(), 'topological', False))
2586
self.target.signatures.insert_record_stream(
2587
self.source.signatures.get_record_stream(
2588
self.source.signatures.keys(),
2589
'unordered', True))
2590
self.target.revisions.insert_record_stream(
2591
self.source.revisions.get_record_stream(
2592
self.source.revisions.keys(),
2593
'topological', True))
1537
self.target.control_weaves.copy_multi(
1538
self.source.control_weaves, ['inventory'],
1539
from_transaction=self.source.get_transaction(),
1540
to_transaction=self.target.get_transaction())
1541
self.target._revision_store.text_store.copy_all_ids(
1542
self.source._revision_store.text_store,
1543
pb=pb)
2594
1544
finally:
2595
1545
pb.finished()
2596
1546
else:
2597
1547
self.target.fetch(self.source, revision_id=revision_id)
2598
1548
2599
1549
@needs_write_lock
2600
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1550
def fetch(self, revision_id=None, pb=None):
2601
1551
"""See InterRepository.fetch()."""
2602
from bzrlib.fetch import RepoFetcher
1552
from bzrlib.fetch import GenericRepoFetcher
2603
1553
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2604
1554
self.source, self.source._format, self.target, self.target._format)
2605
f = RepoFetcher(to_repository=self.target,
1555
# TODO: jam 20070210 This should be an assert, not a translate
1556
revision_id = osutils.safe_revision_id(revision_id)
1557
f = GenericRepoFetcher(to_repository=self.target,
2606
1558
from_repository=self.source,
2607
1559
last_revision=revision_id,
2608
pb=pb, find_ghosts=find_ghosts)
1560
pb=pb)
2609
1561
return f.count_copied, f.failed_revisions
2610
1562
2611
1563
@needs_read_lock
2612
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1564
def missing_revision_ids(self, revision_id=None):
2613
1565
"""See InterRepository.missing_revision_ids()."""
2614
1566
# we want all revisions to satisfy revision_id in source.
2615
1567
# but we don't want to stat every file here and there.
2624
1576
# - RBC 20060209
2625
1577
if revision_id is not None:
2626
1578
source_ids = self.source.get_ancestry(revision_id)
2627
if source_ids[0] is not None:
2628
raise AssertionError()
1579
assert source_ids[0] is None
2629
1580
source_ids.pop(0)
2630
1581
else:
2631
1582
source_ids = self.source._all_possible_ids()
2636
1587
# we do not have a revision as that would be pointless.
2637
1588
target_ids = set(self.target._all_possible_ids())
2638
1589
possibly_present_revisions = target_ids.intersection(source_ids_set)
2639
actually_present_revisions = set(
2640
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1590
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2641
1591
required_revisions = source_ids_set.difference(actually_present_revisions)
1592
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2642
1593
if revision_id is not None:
2643
1594
# we used get_ancestry to determine source_ids then we are assured all
2644
1595
# revisions referenced are present as they are installed in topological order.
2645
1596
# and the tip revision was validated by get_ancestry.
2646
result_set = required_revisions
1597
return required_topo_revisions
2647
1598
else:
2648
1599
# if we just grabbed the possibly available ids, then
2649
1600
# we only have an estimate of whats available and need to validate
2650
1601
# that against the revision records.
2651
result_set = set(
2652
self.source._eliminate_revisions_not_present(required_revisions))
2653
return self.source.revision_ids_to_search_result(result_set)
1602
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2654
1603
2655
1604
2656
1605
class InterKnitRepo(InterSameDataRepository):
2669
1618
could lead to confusing results, and there is no need to be
2670
1619
overly general.
2671
1620
"""
2672
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
1621
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
2673
1622
try:
2674
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2675
isinstance(target._format, RepositoryFormatKnit))
1623
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1624
isinstance(target._format, (RepositoryFormatKnit1)))
2676
1625
except AttributeError:
2677
1626
return False
2678
return are_knits and InterRepository._same_model(source, target)
2679
1627
2680
1628
@needs_write_lock
2681
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1629
def fetch(self, revision_id=None, pb=None):
2682
1630
"""See InterRepository.fetch()."""
2683
from bzrlib.fetch import RepoFetcher
1631
from bzrlib.fetch import KnitRepoFetcher
2684
1632
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2685
1633
self.source, self.source._format, self.target, self.target._format)
2686
f = RepoFetcher(to_repository=self.target,
1634
# TODO: jam 20070210 This should be an assert, not a translate
1635
revision_id = osutils.safe_revision_id(revision_id)
1636
f = KnitRepoFetcher(to_repository=self.target,
2687
1637
from_repository=self.source,
2688
1638
last_revision=revision_id,
2689
pb=pb, find_ghosts=find_ghosts)
1639
pb=pb)
2690
1640
return f.count_copied, f.failed_revisions
2691
1641
2692
1642
@needs_read_lock
2693
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1643
def missing_revision_ids(self, revision_id=None):
2694
1644
"""See InterRepository.missing_revision_ids()."""
2695
1645
if revision_id is not None:
2696
1646
source_ids = self.source.get_ancestry(revision_id)
2697
if source_ids[0] is not None:
2698
raise AssertionError()
1647
assert source_ids[0] is None
2699
1648
source_ids.pop(0)
2700
1649
else:
2701
source_ids = self.source.all_revision_ids()
1650
source_ids = self.source._all_possible_ids()
2702
1651
source_ids_set = set(source_ids)
2703
1652
# source_ids is the worst possible case we may need to pull.
2704
1653
# now we want to filter source_ids against what we actually
2705
1654
# have in target, but don't try to check for existence where we know
2706
1655
# we do not have a revision as that would be pointless.
2707
target_ids = set(self.target.all_revision_ids())
1656
target_ids = set(self.target._all_possible_ids())
2708
1657
possibly_present_revisions = target_ids.intersection(source_ids_set)
2709
actually_present_revisions = set(
2710
self.target._eliminate_revisions_not_present(possibly_present_revisions))
1658
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2711
1659
required_revisions = source_ids_set.difference(actually_present_revisions)
1660
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2712
1661
if revision_id is not None:
2713
1662
# we used get_ancestry to determine source_ids then we are assured all
2714
1663
# revisions referenced are present as they are installed in topological order.
2715
1664
# and the tip revision was validated by get_ancestry.
2716
result_set = required_revisions
1665
return required_topo_revisions
2717
1666
else:
2718
1667
# if we just grabbed the possibly available ids, then
2719
1668
# we only have an estimate of whats available and need to validate
2720
1669
# that against the revision records.
2721
result_set = set(
2722
self.source._eliminate_revisions_not_present(required_revisions))
2723
return self.source.revision_ids_to_search_result(result_set)
2724
2725
2726
class InterPackRepo(InterSameDataRepository):
2727
"""Optimised code paths between Pack based repositories."""
2728
2729
@classmethod
2730
def _get_repo_format_to_test(self):
2731
from bzrlib.repofmt import pack_repo
2732
return pack_repo.RepositoryFormatKnitPack1()
2733
2734
@staticmethod
2735
def is_compatible(source, target):
2736
"""Be compatible with known Pack formats.
2737
2738
We don't test for the stores being of specific types because that
2739
could lead to confusing results, and there is no need to be
2740
overly general.
2741
"""
2742
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2743
try:
2744
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2745
isinstance(target._format, RepositoryFormatPack))
2746
except AttributeError:
2747
return False
2748
return are_packs and InterRepository._same_model(source, target)
2749
2750
@needs_write_lock
2751
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2752
"""See InterRepository.fetch()."""
2753
if (len(self.source._fallback_repositories) > 0 or
2754
len(self.target._fallback_repositories) > 0):
2755
# The pack layer is not aware of fallback repositories, so when
2756
# fetching from a stacked repository or into a stacked repository
2757
# we use the generic fetch logic which uses the VersionedFiles
2758
# attributes on repository.
2759
from bzrlib.fetch import RepoFetcher
2760
fetcher = RepoFetcher(self.target, self.source, revision_id,
2761
pb, find_ghosts)
2762
return fetcher.count_copied, fetcher.failed_revisions
2763
from bzrlib.repofmt.pack_repo import Packer
2764
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2765
self.source, self.source._format, self.target, self.target._format)
2766
self.count_copied = 0
2767
if revision_id is None:
2768
# TODO:
2769
# everything to do - use pack logic
2770
# to fetch from all packs to one without
2771
# inventory parsing etc, IFF nothing to be copied is in the target.
2772
# till then:
2773
source_revision_ids = frozenset(self.source.all_revision_ids())
2774
revision_ids = source_revision_ids - \
2775
frozenset(self.target.get_parent_map(source_revision_ids))
2776
revision_keys = [(revid,) for revid in revision_ids]
2777
index = self.target._pack_collection.revision_index.combined_index
2778
present_revision_ids = set(item[1][0] for item in
2779
index.iter_entries(revision_keys))
2780
revision_ids = set(revision_ids) - present_revision_ids
2781
# implementing the TODO will involve:
2782
# - detecting when all of a pack is selected
2783
# - avoiding as much as possible pre-selection, so the
2784
# more-core routines such as create_pack_from_packs can filter in
2785
# a just-in-time fashion. (though having a HEADS list on a
2786
# repository might make this a lot easier, because we could
2787
# sensibly detect 'new revisions' without doing a full index scan.
2788
elif _mod_revision.is_null(revision_id):
2789
# nothing to do:
2790
return (0, [])
2791
else:
2792
try:
2793
revision_ids = self.search_missing_revision_ids(revision_id,
2794
find_ghosts=find_ghosts).get_keys()
2795
except errors.NoSuchRevision:
2796
raise errors.InstallFailed([revision_id])
2797
if len(revision_ids) == 0:
2798
return (0, [])
2799
packs = self.source._pack_collection.all_packs()
2800
pack = Packer(self.target._pack_collection, packs, '.fetch',
2801
revision_ids).pack()
2802
if pack is not None:
2803
self.target._pack_collection._save_pack_names()
2804
# Trigger an autopack. This may duplicate effort as we've just done
2805
# a pack creation, but for now it is simpler to think about as
2806
# 'upload data, then repack if needed'.
2807
self.target._pack_collection.autopack()
2808
return (pack.get_revision_count(), [])
2809
else:
2810
return (0, [])
2811
2812
@needs_read_lock
2813
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2814
"""See InterRepository.missing_revision_ids().
2815
2816
:param find_ghosts: Find ghosts throughout the ancestry of
2817
revision_id.
2818
"""
2819
if not find_ghosts and revision_id is not None:
2820
return self._walk_to_common_revisions([revision_id])
2821
elif revision_id is not None:
2822
# Find ghosts: search for revisions pointing from one repository to
2823
# the other, and vice versa, anywhere in the history of revision_id.
2824
graph = self.target.get_graph(other_repository=self.source)
2825
searcher = graph._make_breadth_first_searcher([revision_id])
2826
found_ids = set()
2827
while True:
2828
try:
2829
next_revs, ghosts = searcher.next_with_ghosts()
2830
except StopIteration:
2831
break
2832
if revision_id in ghosts:
2833
raise errors.NoSuchRevision(self.source, revision_id)
2834
found_ids.update(next_revs)
2835
found_ids.update(ghosts)
2836
found_ids = frozenset(found_ids)
2837
# Double query here: should be able to avoid this by changing the
2838
# graph api further.
2839
result_set = found_ids - frozenset(
2840
self.target.get_parent_map(found_ids))
2841
else:
2842
source_ids = self.source.all_revision_ids()
2843
# source_ids is the worst possible case we may need to pull.
2844
# now we want to filter source_ids against what we actually
2845
# have in target, but don't try to check for existence where we know
2846
# we do not have a revision as that would be pointless.
2847
target_ids = set(self.target.all_revision_ids())
2848
result_set = set(source_ids).difference(target_ids)
2849
return self.source.revision_ids_to_search_result(result_set)
1670
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2850
1671
2851
1672
2852
1673
class InterModel1and2(InterRepository):
2863
1684
return False
2864
1685
2865
1686
@needs_write_lock
2866
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1687
def fetch(self, revision_id=None, pb=None):
2867
1688
"""See InterRepository.fetch()."""
2868
1689
from bzrlib.fetch import Model1toKnit2Fetcher
1690
# TODO: jam 20070210 This should be an assert, not a translate
1691
revision_id = osutils.safe_revision_id(revision_id)
2869
1692
f = Model1toKnit2Fetcher(to_repository=self.target,
2870
1693
from_repository=self.source,
2871
1694
last_revision=revision_id,
2872
pb=pb, find_ghosts=find_ghosts)
1695
pb=pb)
2873
1696
return f.count_copied, f.failed_revisions
2874
1697
2875
1698
@needs_write_lock
2886
1709
self.target.set_make_working_trees(self.source.make_working_trees())
2887
1710
except NotImplementedError:
2888
1711
pass
1712
# TODO: jam 20070210 Internal, assert, don't translate
1713
revision_id = osutils.safe_revision_id(revision_id)
2889
1714
# but don't bother fetching if we have the needed data now.
2890
1715
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2891
1716
self.target.has_revision(revision_id)):
2902
1727
@staticmethod
2903
1728
def is_compatible(source, target):
2904
1729
"""Be compatible with Knit1 source and Knit3 target"""
1730
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2905
1731
try:
2906
from bzrlib.repofmt.knitrepo import (
2907
RepositoryFormatKnit1,
2908
RepositoryFormatKnit3,
2909
)
2910
from bzrlib.repofmt.pack_repo import (
2911
RepositoryFormatKnitPack1,
2912
RepositoryFormatKnitPack3,
2913
RepositoryFormatKnitPack4,
2914
RepositoryFormatKnitPack5,
2915
RepositoryFormatKnitPack5RichRoot,
2916
RepositoryFormatPackDevelopment1,
2917
RepositoryFormatPackDevelopment1Subtree,
2918
)
2919
norichroot = (
2920
RepositoryFormatKnit1, # no rr, no subtree
2921
RepositoryFormatKnitPack1, # no rr, no subtree
2922
RepositoryFormatPackDevelopment1, # no rr, no subtree
2923
RepositoryFormatKnitPack5, # no rr, no subtree
2924
)
2925
richroot = (
2926
RepositoryFormatKnit3, # rr, subtree
2927
RepositoryFormatKnitPack3, # rr, subtree
2928
RepositoryFormatKnitPack4, # rr, no subtree
2929
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2930
RepositoryFormatPackDevelopment1Subtree, # rr, subtree
2931
)
2932
for format in norichroot:
2933
if format.rich_root_data:
2934
raise AssertionError('Format %s is a rich-root format'
2935
' but is included in the non-rich-root list'
2936
% (format,))
2937
for format in richroot:
2938
if not format.rich_root_data:
2939
raise AssertionError('Format %s is not a rich-root format'
2940
' but is included in the rich-root list'
2941
% (format,))
2942
# TODO: One alternative is to just check format.rich_root_data,
2943
# instead of keeping membership lists. However, the formats
2944
# *also* have to use the same 'Knit' style of storage
2945
# (line-deltas, fulltexts, etc.)
2946
return (isinstance(source._format, norichroot) and
2947
isinstance(target._format, richroot))
1732
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1733
RepositoryFormatKnit3
1734
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1735
isinstance(target._format, (RepositoryFormatKnit3)))
2948
1736
except AttributeError:
2949
1737
return False
2950
1738
2951
1739
@needs_write_lock
2952
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1740
def fetch(self, revision_id=None, pb=None):
2953
1741
"""See InterRepository.fetch()."""
2954
1742
from bzrlib.fetch import Knit1to2Fetcher
2955
1743
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2956
1744
self.source, self.source._format, self.target,
2957
1745
self.target._format)
1746
# TODO: jam 20070210 This should be an assert, not a translate
1747
revision_id = osutils.safe_revision_id(revision_id)
2958
1748
f = Knit1to2Fetcher(to_repository=self.target,
2959
1749
from_repository=self.source,
2960
1750
last_revision=revision_id,
2961
pb=pb, find_ghosts=find_ghosts)
1751
pb=pb)
2962
1752
return f.count_copied, f.failed_revisions
2963
1753
2964
1754
2965
class InterDifferingSerializer(InterKnitRepo):
2966
2967
@classmethod
2968
def _get_repo_format_to_test(self):
2969
return None
2970
2971
@staticmethod
2972
def is_compatible(source, target):
2973
"""Be compatible with Knit2 source and Knit3 target"""
2974
if source.supports_rich_root() != target.supports_rich_root():
2975
return False
2976
# Ideally, we'd support fetching if the source had no tree references
2977
# even if it supported them...
2978
if (getattr(source, '_format.supports_tree_reference', False) and
2979
not getattr(target, '_format.supports_tree_reference', False)):
2980
return False
2981
return True
2982
2983
@needs_write_lock
2984
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2985
"""See InterRepository.fetch()."""
2986
revision_ids = self.target.search_missing_revision_ids(self.source,
2987
revision_id, find_ghosts=find_ghosts).get_keys()
2988
revision_ids = tsort.topo_sort(
2989
self.source.get_graph().get_parent_map(revision_ids))
2990
def revisions_iterator():
2991
for current_revision_id in revision_ids:
2992
revision = self.source.get_revision(current_revision_id)
2993
tree = self.source.revision_tree(current_revision_id)
2994
try:
2995
signature = self.source.get_signature_text(
2996
current_revision_id)
2997
except errors.NoSuchRevision:
2998
signature = None
2999
yield revision, tree, signature
3000
if pb is None:
3001
my_pb = ui.ui_factory.nested_progress_bar()
3002
pb = my_pb
3003
else:
3004
my_pb = None
3005
try:
3006
install_revisions(self.target, revisions_iterator(),
3007
len(revision_ids), pb)
3008
finally:
3009
if my_pb is not None:
3010
my_pb.finished()
3011
return len(revision_ids), 0
3012
3013
3014
class InterOtherToRemote(InterRepository):
1755
class InterRemoteRepository(InterRepository):
1756
"""Code for converting between RemoteRepository objects.
1757
1758
This just gets an non-remote repository from the RemoteRepository, and calls
1759
InterRepository.get again.
1760
"""
3015
1761
3016
1762
def __init__(self, source, target):
3017
InterRepository.__init__(self, source, target)
3018
self._real_inter = None
1763
if isinstance(source, remote.RemoteRepository):
1764
source._ensure_real()
1765
real_source = source._real_repository
1766
else:
1767
real_source = source
1768
if isinstance(target, remote.RemoteRepository):
1769
target._ensure_real()
1770
real_target = target._real_repository
1771
else:
1772
real_target = target
1773
self.real_inter = InterRepository.get(real_source, real_target)
3019
1774
3020
1775
@staticmethod
3021
1776
def is_compatible(source, target):
1777
if isinstance(source, remote.RemoteRepository):
1778
return True
3022
1779
if isinstance(target, remote.RemoteRepository):
3023
1780
return True
3024
1781
return False
3025
1782
3026
def _ensure_real_inter(self):
3027
if self._real_inter is None:
3028
self.target._ensure_real()
3029
real_target = self.target._real_repository
3030
self._real_inter = InterRepository.get(self.source, real_target)
3031
3032
def copy_content(self, revision_id=None):
3033
self._ensure_real_inter()
3034
self._real_inter.copy_content(revision_id=revision_id)
3035
3036
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3037
self._ensure_real_inter()
3038
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3039
find_ghosts=find_ghosts)
3040
3041
@classmethod
3042
def _get_repo_format_to_test(self):
3043
return None
3044
3045
3046
class InterRemoteToOther(InterRepository):
3047
3048
def __init__(self, source, target):
3049
InterRepository.__init__(self, source, target)
3050
self._real_inter = None
3051
3052
@staticmethod
3053
def is_compatible(source, target):
3054
if not isinstance(source, remote.RemoteRepository):
3055
return False
3056
# Is source's model compatible with target's model?
3057
source._ensure_real()
3058
real_source = source._real_repository
3059
if isinstance(real_source, remote.RemoteRepository):
3060
raise NotImplementedError(
3061
"We don't support remote repos backed by remote repos yet.")
3062
return InterRepository._same_model(real_source, target)
3063
3064
def _ensure_real_inter(self):
3065
if self._real_inter is None:
3066
self.source._ensure_real()
3067
real_source = self.source._real_repository
3068
self._real_inter = InterRepository.get(real_source, self.target)
3069
3070
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3071
self._ensure_real_inter()
3072
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3073
find_ghosts=find_ghosts)
3074
3075
def copy_content(self, revision_id=None):
3076
self._ensure_real_inter()
3077
self._real_inter.copy_content(revision_id=revision_id)
3078
3079
@classmethod
3080
def _get_repo_format_to_test(self):
3081
return None
3082
3083
3084
3085
InterRepository.register_optimiser(InterDifferingSerializer)
1783
def copy_content(self, revision_id=None):
1784
self.real_inter.copy_content(revision_id=revision_id)
1785
1786
def fetch(self, revision_id=None, pb=None):
1787
self.real_inter.fetch(revision_id=revision_id, pb=pb)
1788
1789
@classmethod
1790
def _get_repo_format_to_test(self):
1791
return None
1792
1793
3086
1794
InterRepository.register_optimiser(InterSameDataRepository)
3087
1795
InterRepository.register_optimiser(InterWeaveRepo)
3088
1796
InterRepository.register_optimiser(InterKnitRepo)
3089
1797
InterRepository.register_optimiser(InterModel1and2)
3090
1798
InterRepository.register_optimiser(InterKnit1and2)
3091
InterRepository.register_optimiser(InterPackRepo)
3092
InterRepository.register_optimiser(InterOtherToRemote)
3093
InterRepository.register_optimiser(InterRemoteToOther)
1799
InterRepository.register_optimiser(InterRemoteRepository)
3094
1800
3095
1801
3096
1802
class CopyConverter(object):
3145
1851
self.pb.update(message, self.count, self.total)
3146
1852
3147
1853
1854
class CommitBuilder(object):
1855
"""Provides an interface to build up a commit.
1856
1857
This allows describing a tree to be committed without needing to
1858
know the internals of the format of the repository.
1859
"""
1860
1861
record_root_entry = False
1862
def __init__(self, repository, parents, config, timestamp=None,
1863
timezone=None, committer=None, revprops=None,
1864
revision_id=None):
1865
"""Initiate a CommitBuilder.
1866
1867
:param repository: Repository to commit to.
1868
:param parents: Revision ids of the parents of the new revision.
1869
:param config: Configuration to use.
1870
:param timestamp: Optional timestamp recorded for commit.
1871
:param timezone: Optional timezone for timestamp.
1872
:param committer: Optional committer to set for commit.
1873
:param revprops: Optional dictionary of revision properties.
1874
:param revision_id: Optional revision id.
1875
"""
1876
self._config = config
1877
1878
if committer is None:
1879
self._committer = self._config.username()
1880
else:
1881
assert isinstance(committer, basestring), type(committer)
1882
self._committer = committer
1883
1884
self.new_inventory = Inventory(None)
1885
self._new_revision_id = osutils.safe_revision_id(revision_id)
1886
self.parents = parents
1887
self.repository = repository
1888
1889
self._revprops = {}
1890
if revprops is not None:
1891
self._revprops.update(revprops)
1892
1893
if timestamp is None:
1894
timestamp = time.time()
1895
# Restrict resolution to 1ms
1896
self._timestamp = round(timestamp, 3)
1897
1898
if timezone is None:
1899
self._timezone = osutils.local_time_offset()
1900
else:
1901
self._timezone = int(timezone)
1902
1903
self._generate_revision_if_needed()
1904
1905
def commit(self, message):
1906
"""Make the actual commit.
1907
1908
:return: The revision id of the recorded revision.
1909
"""
1910
rev = _mod_revision.Revision(
1911
timestamp=self._timestamp,
1912
timezone=self._timezone,
1913
committer=self._committer,
1914
message=message,
1915
inventory_sha1=self.inv_sha1,
1916
revision_id=self._new_revision_id,
1917
properties=self._revprops)
1918
rev.parent_ids = self.parents
1919
self.repository.add_revision(self._new_revision_id, rev,
1920
self.new_inventory, self._config)
1921
return self._new_revision_id
1922
1923
def revision_tree(self):
1924
"""Return the tree that was just committed.
1925
1926
After calling commit() this can be called to get a RevisionTree
1927
representing the newly committed tree. This is preferred to
1928
calling Repository.revision_tree() because that may require
1929
deserializing the inventory, while we already have a copy in
1930
memory.
1931
"""
1932
return RevisionTree(self.repository, self.new_inventory,
1933
self._new_revision_id)
1934
1935
def finish_inventory(self):
1936
"""Tell the builder that the inventory is finished."""
1937
if self.new_inventory.root is None:
1938
symbol_versioning.warn('Root entry should be supplied to'
1939
' record_entry_contents, as of bzr 0.10.',
1940
DeprecationWarning, stacklevel=2)
1941
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
1942
self.new_inventory.revision_id = self._new_revision_id
1943
self.inv_sha1 = self.repository.add_inventory(
1944
self._new_revision_id,
1945
self.new_inventory,
1946
self.parents
1947
)
1948
1949
def _gen_revision_id(self):
1950
"""Return new revision-id."""
1951
return generate_ids.gen_revision_id(self._config.username(),
1952
self._timestamp)
1953
1954
def _generate_revision_if_needed(self):
1955
"""Create a revision id if None was supplied.
1956
1957
If the repository can not support user-specified revision ids
1958
they should override this function and raise CannotSetRevisionId
1959
if _new_revision_id is not None.
1960
1961
:raises: CannotSetRevisionId
1962
"""
1963
if self._new_revision_id is None:
1964
self._new_revision_id = self._gen_revision_id()
1965
1966
def record_entry_contents(self, ie, parent_invs, path, tree):
1967
"""Record the content of ie from tree into the commit if needed.
1968
1969
Side effect: sets ie.revision when unchanged
1970
1971
:param ie: An inventory entry present in the commit.
1972
:param parent_invs: The inventories of the parent revisions of the
1973
commit.
1974
:param path: The path the entry is at in the tree.
1975
:param tree: The tree which contains this entry and should be used to
1976
obtain content.
1977
"""
1978
if self.new_inventory.root is None and ie.parent_id is not None:
1979
symbol_versioning.warn('Root entry should be supplied to'
1980
' record_entry_contents, as of bzr 0.10.',
1981
DeprecationWarning, stacklevel=2)
1982
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
1983
'', tree)
1984
self.new_inventory.add(ie)
1985
1986
# ie.revision is always None if the InventoryEntry is considered
1987
# for committing. ie.snapshot will record the correct revision
1988
# which may be the sole parent if it is untouched.
1989
if ie.revision is not None:
1990
return
1991
1992
# In this revision format, root entries have no knit or weave
1993
if ie is self.new_inventory.root:
1994
# When serializing out to disk and back in
1995
# root.revision is always _new_revision_id
1996
ie.revision = self._new_revision_id
1997
return
1998
previous_entries = ie.find_previous_heads(
1999
parent_invs,
2000
self.repository.weave_store,
2001
self.repository.get_transaction())
2002
# we are creating a new revision for ie in the history store
2003
# and inventory.
2004
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2005
2006
def modified_directory(self, file_id, file_parents):
2007
"""Record the presence of a symbolic link.
2008
2009
:param file_id: The file_id of the link to record.
2010
:param file_parents: The per-file parent revision ids.
2011
"""
2012
self._add_text_to_weave(file_id, [], file_parents.keys())
2013
2014
def modified_reference(self, file_id, file_parents):
2015
"""Record the modification of a reference.
2016
2017
:param file_id: The file_id of the link to record.
2018
:param file_parents: The per-file parent revision ids.
2019
"""
2020
self._add_text_to_weave(file_id, [], file_parents.keys())
2021
2022
def modified_file_text(self, file_id, file_parents,
2023
get_content_byte_lines, text_sha1=None,
2024
text_size=None):
2025
"""Record the text of file file_id
2026
2027
:param file_id: The file_id of the file to record the text of.
2028
:param file_parents: The per-file parent revision ids.
2029
:param get_content_byte_lines: A callable which will return the byte
2030
lines for the file.
2031
:param text_sha1: Optional SHA1 of the file contents.
2032
:param text_size: Optional size of the file contents.
2033
"""
2034
# mutter('storing text of file {%s} in revision {%s} into %r',
2035
# file_id, self._new_revision_id, self.repository.weave_store)
2036
# special case to avoid diffing on renames or
2037
# reparenting
2038
if (len(file_parents) == 1
2039
and text_sha1 == file_parents.values()[0].text_sha1
2040
and text_size == file_parents.values()[0].text_size):
2041
previous_ie = file_parents.values()[0]
2042
versionedfile = self.repository.weave_store.get_weave(file_id,
2043
self.repository.get_transaction())
2044
versionedfile.clone_text(self._new_revision_id,
2045
previous_ie.revision, file_parents.keys())
2046
return text_sha1, text_size
2047
else:
2048
new_lines = get_content_byte_lines()
2049
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2050
# should return the SHA1 and size
2051
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2052
return osutils.sha_strings(new_lines), \
2053
sum(map(len, new_lines))
2054
2055
def modified_link(self, file_id, file_parents, link_target):
2056
"""Record the presence of a symbolic link.
2057
2058
:param file_id: The file_id of the link to record.
2059
:param file_parents: The per-file parent revision ids.
2060
:param link_target: Target location of this link.
2061
"""
2062
self._add_text_to_weave(file_id, [], file_parents.keys())
2063
2064
def _add_text_to_weave(self, file_id, new_lines, parents):
2065
versionedfile = self.repository.weave_store.get_weave_or_empty(
2066
file_id, self.repository.get_transaction())
2067
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2068
versionedfile.clear_cache()
2069
2070
2071
class _CommitBuilder(CommitBuilder):
2072
"""Temporary class so old CommitBuilders are detected properly
2073
2074
Note: CommitBuilder works whether or not root entry is recorded.
2075
"""
2076
2077
record_root_entry = True
2078
2079
2080
class RootCommitBuilder(CommitBuilder):
2081
"""This commitbuilder actually records the root id"""
2082
2083
record_root_entry = True
2084
2085
def record_entry_contents(self, ie, parent_invs, path, tree):
2086
"""Record the content of ie from tree into the commit if needed.
2087
2088
Side effect: sets ie.revision when unchanged
2089
2090
:param ie: An inventory entry present in the commit.
2091
:param parent_invs: The inventories of the parent revisions of the
2092
commit.
2093
:param path: The path the entry is at in the tree.
2094
:param tree: The tree which contains this entry and should be used to
2095
obtain content.
2096
"""
2097
assert self.new_inventory.root is not None or ie.parent_id is None
2098
self.new_inventory.add(ie)
2099
2100
# ie.revision is always None if the InventoryEntry is considered
2101
# for committing. ie.snapshot will record the correct revision
2102
# which may be the sole parent if it is untouched.
2103
if ie.revision is not None:
2104
return
2105
2106
previous_entries = ie.find_previous_heads(
2107
parent_invs,
2108
self.repository.weave_store,
2109
self.repository.get_transaction())
2110
# we are creating a new revision for ie in the history store
2111
# and inventory.
2112
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2113
2114
3148
2115
_unescape_map = {
3149
2116
'apos':"'",
3150
2117
'quot':'"',
3173
2140
if _unescape_re is None:
3174
2141
_unescape_re = re.compile('\&([^;]*);')
3175
2142
return _unescape_re.sub(_unescaper, data)
3176
3177
3178
class _VersionedFileChecker(object):
3179
3180
def __init__(self, repository):
3181
self.repository = repository
3182
self.text_index = self.repository._generate_text_key_index()
3183
3184
def calculate_file_version_parents(self, text_key):
3185
"""Calculate the correct parents for a file version according to
3186
the inventories.
3187
"""
3188
parent_keys = self.text_index[text_key]
3189
if parent_keys == [_mod_revision.NULL_REVISION]:
3190
return ()
3191
return tuple(parent_keys)
3192
3193
def check_file_version_parents(self, texts, progress_bar=None):
3194
"""Check the parents stored in a versioned file are correct.
3195
3196
It also detects file versions that are not referenced by their
3197
corresponding revision's inventory.
3198
3199
:returns: A tuple of (wrong_parents, dangling_file_versions).
3200
wrong_parents is a dict mapping {revision_id: (stored_parents,
3201
correct_parents)} for each revision_id where the stored parents
3202
are not correct. dangling_file_versions is a set of (file_id,
3203
revision_id) tuples for versions that are present in this versioned
3204
file, but not used by the corresponding inventory.
3205
"""
3206
wrong_parents = {}
3207
self.file_ids = set([file_id for file_id, _ in
3208
self.text_index.iterkeys()])
3209
# text keys is now grouped by file_id
3210
n_weaves = len(self.file_ids)
3211
files_in_revisions = {}
3212
revisions_of_files = {}
3213
n_versions = len(self.text_index)
3214
progress_bar.update('loading text store', 0, n_versions)
3215
parent_map = self.repository.texts.get_parent_map(self.text_index)
3216
# On unlistable transports this could well be empty/error...
3217
text_keys = self.repository.texts.keys()
3218
unused_keys = frozenset(text_keys) - set(self.text_index)
3219
for num, key in enumerate(self.text_index.iterkeys()):
3220
if progress_bar is not None:
3221
progress_bar.update('checking text graph', num, n_versions)
3222
correct_parents = self.calculate_file_version_parents(key)
3223
try:
3224
knit_parents = parent_map[key]
3225
except errors.RevisionNotPresent:
3226
# Missing text!
3227
knit_parents = None
3228
if correct_parents != knit_parents:
3229
wrong_parents[key] = (knit_parents, correct_parents)
3230
return wrong_parents, unused_keys
3231
3232
3233
def _old_get_graph(repository, revision_id):
3234
"""DO NOT USE. That is all. I'm serious."""
3235
graph = repository.get_graph()
3236
revision_graph = dict(((key, value) for key, value in
3237
graph.iter_ancestry([revision_id]) if value is not None))
3238
return _strip_NULL_ghosts(revision_graph)
3239
3240
3241
def _strip_NULL_ghosts(revision_graph):
3242
"""Also don't use this. more compatibility code for unmigrated clients."""
3243
# Filter ghosts, and null:
3244
if _mod_revision.NULL_REVISION in revision_graph:
3245
del revision_graph[_mod_revision.NULL_REVISION]
3246
for key, parents in revision_graph.items():
3247
revision_graph[key] = tuple(parent for parent in parents if parent
3248
in revision_graph)
3249
return revision_graph
Loggerhead is a web-based interface for Breezy
Version: 2.0.1