66
81
def _reconcile(self):
67
82
"""Helper function for performing reconciliation."""
83
self._reconcile_branch()
84
self._reconcile_repository()
86
def _reconcile_branch(self):
88
self.branch = self.bzrdir.open_branch()
89
except errors.NotBranchError:
90
# Nothing to check here
91
self.fixed_branch_history = None
93
ui.ui_factory.note('Reconciling branch %s' % self.branch.base)
94
branch_reconciler = self.branch.reconcile(thorough=True)
95
self.fixed_branch_history = branch_reconciler.fixed_history
97
def _reconcile_repository(self):
68
98
self.repo = self.bzrdir.find_repository()
69
self.pb.note('Reconciling repository %s',
70
self.repo.bzrdir.root_transport.base)
99
ui.ui_factory.note('Reconciling repository %s' %
100
self.repo.bzrdir.root_transport.base)
101
self.pb.update("Reconciling repository", 0, 1)
71
102
repo_reconciler = self.repo.reconcile(thorough=True)
72
103
self.inconsistent_parents = repo_reconciler.inconsistent_parents
73
104
self.garbage_inventories = repo_reconciler.garbage_inventories
74
self.pb.note('Reconciliation complete.')
105
if repo_reconciler.aborted:
107
'Reconcile aborted: revision index has inconsistent parents.')
109
'Run "bzr check" for more details.')
111
ui.ui_factory.note('Reconciliation complete.')
114
class BranchReconciler(object):
115
"""Reconciler that works on a branch."""
117
def __init__(self, a_branch, thorough=False):
118
self.fixed_history = None
119
self.thorough = thorough
120
self.branch = a_branch
123
operation = cleanup.OperationWithCleanups(self._reconcile)
124
self.add_cleanup = operation.add_cleanup
125
operation.run_simple()
127
def _reconcile(self):
128
self.branch.lock_write()
129
self.add_cleanup(self.branch.unlock)
130
self.pb = ui.ui_factory.nested_progress_bar()
131
self.add_cleanup(self.pb.finished)
132
self._reconcile_steps()
134
def _reconcile_steps(self):
135
self._reconcile_revision_history()
137
def _reconcile_revision_history(self):
138
repo = self.branch.repository
139
last_revno, last_revision_id = self.branch.last_revision_info()
142
for revid in repo.iter_reverse_revision_history(
144
real_history.append(revid)
145
except errors.RevisionNotPresent:
146
pass # Hit a ghost left hand parent
147
real_history.reverse()
148
if last_revno != len(real_history):
149
self.fixed_history = True
150
# Technically for Branch5 formats, it is more efficient to use
151
# set_revision_history, as this will regenerate it again.
152
# Not really worth a whole BranchReconciler class just for this,
154
ui.ui_factory.note('Fixing last revision info %s => %s' % (
155
last_revno, len(real_history)))
156
self.branch.set_last_revision_info(len(real_history),
159
self.fixed_history = False
160
ui.ui_factory.note('revision_history ok.')
77
163
class RepoReconciler(object):
78
164
"""Reconciler that reconciles a repository.
166
The goal of repository reconciliation is to make any derived data
167
consistent with the core data committed by a user. This can involve
168
reindexing, or removing unreferenced data if that can interfere with
169
queries in a given repository.
80
171
Currently this consists of an inventory reweave with revision cross-checks.
141
235
# put a revision into the graph.
142
236
self._graph_revision(rev_id)
143
237
self._check_garbage_inventories()
144
# if there are no inconsistent_parents and
238
# if there are no inconsistent_parents and
145
239
# (no garbage inventories or we are not doing a thorough check)
146
if (not self.inconsistent_parents and
240
if (not self.inconsistent_parents and
147
241
(not self.garbage_inventories or not self.thorough)):
148
self.pb.note('Inventory ok.')
242
ui.ui_factory.note('Inventory ok.')
150
self.pb.update('Backing up inventory...', 0, 0)
151
self.repo.control_weaves.copy(self.inventory, 'inventory.backup', self.repo.get_transaction())
152
self.pb.note('Backup Inventory created.')
153
# asking for '' should never return a non-empty weave
154
new_inventory_vf = self.repo.control_weaves.get_empty('inventory.new',
155
self.repo.get_transaction())
244
self.pb.update('Backing up inventory', 0, 0)
245
self.repo._backup_inventory()
246
ui.ui_factory.note('Backup inventory created.')
247
new_inventories = self.repo._temp_inventories()
157
249
# we have topological order of revisions and non ghost parents ready.
158
250
self._setup_steps(len(self._rev_graph))
159
for rev_id in TopoSorter(self._rev_graph.items()).iter_topo_order():
160
parents = self._rev_graph[rev_id]
161
# double check this really is in topological order.
162
unavailable = [p for p in parents if p not in new_inventory_vf]
163
assert len(unavailable) == 0
164
# this entry has all the non ghost parents in the inventory
166
self._reweave_step('adding inventories')
167
if isinstance(new_inventory_vf, WeaveFile):
168
# It's really a WeaveFile, but we call straight into the
169
# Weave's add method to disable the auto-write-out behaviour.
170
# This is done to avoid a revision_count * time-to-write additional overhead on
172
new_inventory_vf._check_write_ok()
173
Weave._add_lines(new_inventory_vf, rev_id, parents, self.inventory.get_lines(rev_id),
176
new_inventory_vf.add_lines(rev_id, parents, self.inventory.get_lines(rev_id))
178
if isinstance(new_inventory_vf, WeaveFile):
179
new_inventory_vf._save()
180
# if this worked, the set of new_inventory_vf.names should equal
251
revision_keys = [(rev_id,) for rev_id in topo_sort(self._rev_graph)]
252
stream = self._change_inv_parents(
253
self.inventory.get_record_stream(revision_keys, 'unordered', True),
254
self._new_inv_parents,
256
new_inventories.insert_record_stream(stream)
257
# if this worked, the set of new_inventories.keys should equal
182
assert set(new_inventory_vf.versions()) == self.pending
259
if not (set(new_inventories.keys()) ==
260
set([(revid,) for revid in self.pending])):
261
raise AssertionError()
183
262
self.pb.update('Writing weave')
184
self.repo.control_weaves.copy(new_inventory_vf, 'inventory', self.repo.get_transaction())
185
self.repo.control_weaves.delete('inventory.new', self.repo.get_transaction())
263
self.repo._activate_new_inventory()
186
264
self.inventory = None
187
self.pb.note('Inventory regenerated.')
265
ui.ui_factory.note('Inventory regenerated.')
267
def _new_inv_parents(self, revision_key):
268
"""Lookup ghost-filtered parents for revision_key."""
269
# Use the filtered ghostless parents list:
270
return tuple([(revid,) for revid in self._rev_graph[revision_key[-1]]])
272
def _change_inv_parents(self, stream, get_parents, all_revision_keys):
273
"""Adapt a record stream to reconcile the parents."""
274
for record in stream:
275
wanted_parents = get_parents(record.key)
276
if wanted_parents and wanted_parents[0] not in all_revision_keys:
277
# The check for the left most parent only handles knit
278
# compressors, but this code only applies to knit and weave
279
# repositories anyway.
280
bytes = record.get_bytes_as('fulltext')
281
yield FulltextContentFactory(record.key, wanted_parents, record.sha1, bytes)
283
adapted_record = AdapterFactory(record.key, wanted_parents, record)
285
self._reweave_step('adding inventories')
189
287
def _setup_steps(self, new_total):
190
288
"""Setup the markers we need to control the progress bar."""
267
338
class KnitReconciler(RepoReconciler):
268
339
"""Reconciler that reconciles a knit format repository.
270
This will detect garbage inventories and remove them.
272
Inconsistent parentage is checked for in the revision weave.
341
This will detect garbage inventories and remove them in thorough mode.
275
344
def _reconcile_steps(self):
276
345
"""Perform the steps to reconcile this repository."""
277
346
if self.thorough:
349
except errors.BzrCheckError:
279
352
# knits never suffer this
280
353
self._gc_inventory()
354
self._fix_text_parents()
282
356
def _load_indexes(self):
283
357
"""Load indexes for the reconciliation."""
284
358
self.transaction = self.repo.get_transaction()
285
self.pb.update('Reading indexes.', 0, 2)
286
self.inventory = self.repo.get_inventory_weave()
287
self.pb.update('Reading indexes.', 1, 2)
288
self.revisions = self.repo._revision_store.get_revision_file(self.transaction)
289
self.pb.update('Reading indexes.', 2, 2)
359
self.pb.update('Reading indexes', 0, 2)
360
self.inventory = self.repo.inventories
361
self.pb.update('Reading indexes', 1, 2)
362
self.repo._check_for_inconsistent_revision_parents()
363
self.revisions = self.repo.revisions
364
self.pb.update('Reading indexes', 2, 2)
291
366
def _gc_inventory(self):
292
367
"""Remove inventories that are not referenced from the revision store."""
293
self.pb.update('Checking unused inventories.', 0, 1)
368
self.pb.update('Checking unused inventories', 0, 1)
294
369
self._check_garbage_inventories()
295
self.pb.update('Checking unused inventories.', 1, 3)
370
self.pb.update('Checking unused inventories', 1, 3)
296
371
if not self.garbage_inventories:
297
self.pb.note('Inventory ok.')
372
ui.ui_factory.note('Inventory ok.')
299
self.pb.update('Backing up inventory...', 0, 0)
300
self.repo.control_weaves.copy(self.inventory, 'inventory.backup', self.transaction)
301
self.pb.note('Backup Inventory created.')
374
self.pb.update('Backing up inventory', 0, 0)
375
self.repo._backup_inventory()
376
ui.ui_factory.note('Backup Inventory created')
302
377
# asking for '' should never return a non-empty weave
303
new_inventory_vf = self.repo.control_weaves.get_empty('inventory.new',
378
new_inventories = self.repo._temp_inventories()
306
379
# we have topological order of revisions and non ghost parents ready.
307
self._setup_steps(len(self.revisions))
308
for rev_id in TopoSorter(self.revisions.get_graph().items()).iter_topo_order():
309
parents = self.revisions.get_parents(rev_id)
310
# double check this really is in topological order.
311
unavailable = [p for p in parents if p not in new_inventory_vf]
312
assert len(unavailable) == 0
313
# this entry has all the non ghost parents in the inventory
315
self._reweave_step('adding inventories')
316
# ugly but needed, weaves are just way tooooo slow else.
317
new_inventory_vf.add_lines(rev_id, parents, self.inventory.get_lines(rev_id))
380
graph = self.revisions.get_parent_map(self.revisions.keys())
381
revision_keys = topo_sort(graph)
382
revision_ids = [key[-1] for key in revision_keys]
383
self._setup_steps(len(revision_keys))
384
stream = self._change_inv_parents(
385
self.inventory.get_record_stream(revision_keys, 'unordered', True),
388
new_inventories.insert_record_stream(stream)
319
389
# if this worked, the set of new_inventory_vf.names should equal
321
assert set(new_inventory_vf.versions()) == set(self.revisions.versions())
390
# the revisionds list
391
if not(set(new_inventories.keys()) == set(revision_keys)):
392
raise AssertionError()
322
393
self.pb.update('Writing weave')
323
self.repo.control_weaves.copy(new_inventory_vf, 'inventory', self.transaction)
324
self.repo.control_weaves.delete('inventory.new', self.transaction)
394
self.repo._activate_new_inventory()
325
395
self.inventory = None
326
self.pb.note('Inventory regenerated.')
328
def _check_garbage_inventories(self):
329
"""Check for garbage inventories which we cannot trust
331
We cant trust them because their pre-requisite file data may not
332
be present - all we know is that their revision was not installed.
334
inventories = set(self.inventory.versions())
335
revisions = set(self.revisions.versions())
336
garbage = inventories.difference(revisions)
337
self.garbage_inventories = len(garbage)
338
for revision_id in garbage:
339
mutter('Garbage inventory {%s} found.', revision_id)
396
ui.ui_factory.note('Inventory regenerated.')
398
def _fix_text_parents(self):
399
"""Fix bad versionedfile parent entries.
401
It is possible for the parents entry in a versionedfile entry to be
402
inconsistent with the values in the revision and inventory.
404
This method finds entries with such inconsistencies, corrects their
405
parent lists, and replaces the versionedfile with a corrected version.
407
transaction = self.repo.get_transaction()
408
versions = [key[-1] for key in self.revisions.keys()]
409
mutter('Prepopulating revision text cache with %d revisions',
411
vf_checker = self.repo._get_versioned_file_checker()
412
bad_parents, unused_versions = vf_checker.check_file_version_parents(
413
self.repo.texts, self.pb)
414
text_index = vf_checker.text_index
415
per_id_bad_parents = {}
416
for key in unused_versions:
417
# Ensure that every file with unused versions gets rewritten.
418
# NB: This is really not needed, reconcile != pack.
419
per_id_bad_parents[key[0]] = {}
420
# Generate per-knit/weave data.
421
for key, details in bad_parents.iteritems():
424
knit_parents = tuple([parent[-1] for parent in details[0]])
425
correct_parents = tuple([parent[-1] for parent in details[1]])
426
file_details = per_id_bad_parents.setdefault(file_id, {})
427
file_details[rev_id] = (knit_parents, correct_parents)
428
file_id_versions = {}
429
for text_key in text_index:
430
versions_list = file_id_versions.setdefault(text_key[0], [])
431
versions_list.append(text_key[1])
432
# Do the reconcile of individual weaves.
433
for num, file_id in enumerate(per_id_bad_parents):
434
self.pb.update('Fixing text parents', num,
435
len(per_id_bad_parents))
436
versions_with_bad_parents = per_id_bad_parents[file_id]
437
id_unused_versions = set(key[-1] for key in unused_versions
438
if key[0] == file_id)
439
if file_id in file_id_versions:
440
file_versions = file_id_versions[file_id]
442
# This id was present in the disk store but is not referenced
443
# by any revision at all.
445
self._fix_text_parent(file_id, versions_with_bad_parents,
446
id_unused_versions, file_versions)
448
def _fix_text_parent(self, file_id, versions_with_bad_parents,
449
unused_versions, all_versions):
450
"""Fix bad versionedfile entries in a single versioned file."""
451
mutter('fixing text parent: %r (%d versions)', file_id,
452
len(versions_with_bad_parents))
453
mutter('(%d are unused)', len(unused_versions))
454
new_file_id = 'temp:%s' % file_id
457
for version in all_versions:
458
if version in unused_versions:
460
elif version in versions_with_bad_parents:
461
parents = versions_with_bad_parents[version][1]
463
pmap = self.repo.texts.get_parent_map([(file_id, version)])
464
parents = [key[-1] for key in pmap[(file_id, version)]]
465
new_parents[(new_file_id, version)] = [
466
(new_file_id, parent) for parent in parents]
467
needed_keys.add((file_id, version))
468
def fix_parents(stream):
469
for record in stream:
470
bytes = record.get_bytes_as('fulltext')
471
new_key = (new_file_id, record.key[-1])
472
parents = new_parents[new_key]
473
yield FulltextContentFactory(new_key, parents, record.sha1, bytes)
474
stream = self.repo.texts.get_record_stream(needed_keys, 'topological', True)
475
self.repo._remove_file_id(new_file_id)
476
self.repo.texts.insert_record_stream(fix_parents(stream))
477
self.repo._remove_file_id(file_id)
479
self.repo._move_file_id(new_file_id, file_id)
482
class PackReconciler(RepoReconciler):
483
"""Reconciler that reconciles a pack based repository.
485
Garbage inventories do not affect ancestry queries, and removal is
486
considerably more expensive as there is no separate versioned file for
487
them, so they are not cleaned. In short it is currently a no-op.
489
In future this may be a good place to hook in annotation cache checking,
490
index recreation etc.
493
# XXX: The index corruption that _fix_text_parents performs is needed for
494
# packs, but not yet implemented. The basic approach is to:
495
# - lock the names list
496
# - perform a customised pack() that regenerates data as needed
497
# - unlock the names list
498
# https://bugs.edge.launchpad.net/bzr/+bug/154173
500
def _reconcile_steps(self):
501
"""Perform the steps to reconcile this repository."""
502
if not self.thorough:
504
collection = self.repo._pack_collection
505
collection.ensure_loaded()
506
collection.lock_names()
507
self.add_cleanup(collection._unlock_names)
508
packs = collection.all_packs()
509
all_revisions = self.repo.all_revision_ids()
510
total_inventories = len(list(
511
collection.inventory_index.combined_index.iter_all_entries()))
512
if len(all_revisions):
513
new_pack = self.repo._reconcile_pack(collection, packs,
514
".reconcile", all_revisions, self.pb)
515
if new_pack is not None:
516
self._discard_and_save(packs)
518
# only make a new pack when there is data to copy.
519
self._discard_and_save(packs)
520
self.garbage_inventories = total_inventories - len(list(
521
collection.inventory_index.combined_index.iter_all_entries()))
523
def _discard_and_save(self, packs):
524
"""Discard some packs from the repository.
526
This removes them from the memory index, saves the in-memory index
527
which makes the newly reconciled pack visible and hides the packs to be
528
discarded, and finally renames the packs being discarded into the
529
obsolete packs directory.
531
:param packs: The packs to discard.
534
self.repo._pack_collection._remove_pack_from_memory(pack)
535
self.repo._pack_collection._save_pack_names()
536
self.repo._pack_collection._obsolete_packs(packs)
added
removed
bzrlib/reconcile.py
