1
# Copyright (C) 2006-2010 Canonical Ltd
1
# Copyright (C) 2005, 2006 Canonical Ltd
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
30
29
from bzrlib import (
33
revision as _mod_revision,
36
from bzrlib.trace import mutter
35
from bzrlib.trace import mutter, note
37
36
from bzrlib.tsort import topo_sort
38
37
from bzrlib.versionedfile import AdapterFactory, FulltextContentFactory
39
from bzrlib.i18n import gettext
42
def reconcile(dir, canonicalize_chks=False):
40
def reconcile(dir, other=None):
43
41
"""Reconcile the data in dir.
45
43
Currently this is limited to a inventory 'reweave'.
49
47
Directly using Reconciler is recommended for library users that
50
48
desire fine grained control or analysis of the found issues.
52
:param canonicalize_chks: Make sure CHKs are in canonical form.
50
:param other: another bzrdir to reconcile against.
54
reconciler = Reconciler(dir, canonicalize_chks=canonicalize_chks)
52
reconciler = Reconciler(dir, other=other)
55
53
reconciler.reconcile()
58
56
class Reconciler(object):
59
57
"""Reconcilers are used to reconcile existing data."""
61
def __init__(self, dir, other=None, canonicalize_chks=False):
59
def __init__(self, dir, other=None):
62
60
"""Create a Reconciler."""
64
self.canonicalize_chks = canonicalize_chks
66
63
def reconcile(self):
67
64
"""Perform reconciliation.
69
66
After reconciliation the following attributes document found issues:
71
* `inconsistent_parents`: The number of revisions in the repository
72
whose ancestry was being reported incorrectly.
73
* `garbage_inventories`: The number of inventory objects without
74
revisions that were garbage collected.
75
* `fixed_branch_history`: None if there was no branch, False if the
76
branch history was correct, True if the branch history needed to be
67
inconsistent_parents: The number of revisions in the repository whose
68
ancestry was being reported incorrectly.
69
garbage_inventories: The number of inventory objects without revisions
70
that were garbage collected.
71
fixed_branch_history: None if there was no branch, False if the branch
72
history was correct, True if the branch history
73
needed to be re-normalized.
79
75
self.pb = ui.ui_factory.nested_progress_bar()
94
90
# Nothing to check here
95
91
self.fixed_branch_history = None
97
ui.ui_factory.note(gettext('Reconciling branch %s') % self.branch.base)
93
ui.ui_factory.note('Reconciling branch %s' % self.branch.base)
98
94
branch_reconciler = self.branch.reconcile(thorough=True)
99
95
self.fixed_branch_history = branch_reconciler.fixed_history
101
97
def _reconcile_repository(self):
102
98
self.repo = self.bzrdir.find_repository()
103
ui.ui_factory.note(gettext('Reconciling repository %s') %
105
self.pb.update(gettext("Reconciling repository"), 0, 1)
106
if self.canonicalize_chks:
108
self.repo.reconcile_canonicalize_chks
109
except AttributeError:
110
raise errors.BzrError(
111
gettext("%s cannot canonicalize CHKs.") % (self.repo,))
112
repo_reconciler = self.repo.reconcile_canonicalize_chks()
114
repo_reconciler = self.repo.reconcile(thorough=True)
99
ui.ui_factory.note('Reconciling repository %s' %
100
self.repo.bzrdir.root_transport.base)
101
self.pb.update("Reconciling repository", 0, 1)
102
repo_reconciler = self.repo.reconcile(thorough=True)
115
103
self.inconsistent_parents = repo_reconciler.inconsistent_parents
116
104
self.garbage_inventories = repo_reconciler.garbage_inventories
117
105
if repo_reconciler.aborted:
118
ui.ui_factory.note(gettext(
119
'Reconcile aborted: revision index has inconsistent parents.'))
120
ui.ui_factory.note(gettext(
121
'Run "bzr check" for more details.'))
107
'Reconcile aborted: revision index has inconsistent parents.')
109
'Run "bzr check" for more details.')
123
ui.ui_factory.note(gettext('Reconciliation complete.'))
111
ui.ui_factory.note('Reconciliation complete.')
126
114
class BranchReconciler(object):
132
120
self.branch = a_branch
134
122
def reconcile(self):
135
operation = cleanup.OperationWithCleanups(self._reconcile)
136
self.add_cleanup = operation.add_cleanup
137
operation.run_simple()
139
def _reconcile(self):
140
123
self.branch.lock_write()
141
self.add_cleanup(self.branch.unlock)
142
self.pb = ui.ui_factory.nested_progress_bar()
143
self.add_cleanup(self.pb.finished)
144
self._reconcile_steps()
125
self.pb = ui.ui_factory.nested_progress_bar()
127
self._reconcile_steps()
146
133
def _reconcile_steps(self):
147
134
self._reconcile_revision_history()
149
136
def _reconcile_revision_history(self):
137
repo = self.branch.repository
150
138
last_revno, last_revision_id = self.branch.last_revision_info()
151
139
real_history = []
152
graph = self.branch.repository.get_graph()
154
for revid in graph.iter_lefthand_ancestry(
155
last_revision_id, (_mod_revision.NULL_REVISION,)):
141
for revid in repo.iter_reverse_revision_history(
156
143
real_history.append(revid)
157
144
except errors.RevisionNotPresent:
158
145
pass # Hit a ghost left hand parent
163
150
# set_revision_history, as this will regenerate it again.
164
151
# Not really worth a whole BranchReconciler class just for this,
166
ui.ui_factory.note(gettext('Fixing last revision info {0} '\
168
last_revno, len(real_history)))
153
ui.ui_factory.note('Fixing last revision info %s => %s' % (
154
last_revno, len(real_history)))
169
155
self.branch.set_last_revision_info(len(real_history),
170
156
last_revision_id)
172
158
self.fixed_history = False
173
ui.ui_factory.note(gettext('revision_history ok.'))
159
ui.ui_factory.note('revision_history ok.')
176
162
class RepoReconciler(object):
201
187
"""Perform reconciliation.
203
189
After reconciliation the following attributes document found issues:
205
* `inconsistent_parents`: The number of revisions in the repository
206
whose ancestry was being reported incorrectly.
207
* `garbage_inventories`: The number of inventory objects without
208
revisions that were garbage collected.
190
inconsistent_parents: The number of revisions in the repository whose
191
ancestry was being reported incorrectly.
192
garbage_inventories: The number of inventory objects without revisions
193
that were garbage collected.
210
operation = cleanup.OperationWithCleanups(self._reconcile)
211
self.add_cleanup = operation.add_cleanup
212
operation.run_simple()
214
def _reconcile(self):
215
195
self.repo.lock_write()
216
self.add_cleanup(self.repo.unlock)
217
self.pb = ui.ui_factory.nested_progress_bar()
218
self.add_cleanup(self.pb.finished)
219
self._reconcile_steps()
197
self.pb = ui.ui_factory.nested_progress_bar()
199
self._reconcile_steps()
221
205
def _reconcile_steps(self):
222
206
"""Perform the steps to reconcile this repository."""
230
214
only data-loss causing issues (!self.thorough) or all issues
231
215
(self.thorough) are treated as requiring the reweave.
217
# local because needing to know about WeaveFile is a wart we want to hide
218
from bzrlib.weave import WeaveFile, Weave
233
219
transaction = self.repo.get_transaction()
234
self.pb.update(gettext('Reading inventory data'))
220
self.pb.update('Reading inventory data')
235
221
self.inventory = self.repo.inventories
236
222
self.revisions = self.repo.revisions
237
223
# the total set of revisions to process
251
237
# (no garbage inventories or we are not doing a thorough check)
252
238
if (not self.inconsistent_parents and
253
239
(not self.garbage_inventories or not self.thorough)):
254
ui.ui_factory.note(gettext('Inventory ok.'))
240
ui.ui_factory.note('Inventory ok.')
256
self.pb.update(gettext('Backing up inventory'), 0, 0)
242
self.pb.update('Backing up inventory', 0, 0)
257
243
self.repo._backup_inventory()
258
ui.ui_factory.note(gettext('Backup inventory created.'))
244
ui.ui_factory.note('Backup inventory created.')
259
245
new_inventories = self.repo._temp_inventories()
261
247
# we have topological order of revisions and non ghost parents ready.
271
257
if not (set(new_inventories.keys()) ==
272
258
set([(revid,) for revid in self.pending])):
273
259
raise AssertionError()
274
self.pb.update(gettext('Writing weave'))
260
self.pb.update('Writing weave')
275
261
self.repo._activate_new_inventory()
276
262
self.inventory = None
277
ui.ui_factory.note(gettext('Inventory regenerated.'))
263
ui.ui_factory.note('Inventory regenerated.')
279
265
def _new_inv_parents(self, revision_key):
280
266
"""Lookup ghost-filtered parents for revision_key."""
368
354
def _load_indexes(self):
369
355
"""Load indexes for the reconciliation."""
370
356
self.transaction = self.repo.get_transaction()
371
self.pb.update(gettext('Reading indexes'), 0, 2)
357
self.pb.update('Reading indexes', 0, 2)
372
358
self.inventory = self.repo.inventories
373
self.pb.update(gettext('Reading indexes'), 1, 2)
359
self.pb.update('Reading indexes', 1, 2)
374
360
self.repo._check_for_inconsistent_revision_parents()
375
361
self.revisions = self.repo.revisions
376
self.pb.update(gettext('Reading indexes'), 2, 2)
362
self.pb.update('Reading indexes', 2, 2)
378
364
def _gc_inventory(self):
379
365
"""Remove inventories that are not referenced from the revision store."""
380
self.pb.update(gettext('Checking unused inventories'), 0, 1)
366
self.pb.update('Checking unused inventories', 0, 1)
381
367
self._check_garbage_inventories()
382
self.pb.update(gettext('Checking unused inventories'), 1, 3)
368
self.pb.update('Checking unused inventories', 1, 3)
383
369
if not self.garbage_inventories:
384
ui.ui_factory.note(gettext('Inventory ok.'))
370
ui.ui_factory.note('Inventory ok.')
386
self.pb.update(gettext('Backing up inventory'), 0, 0)
372
self.pb.update('Backing up inventory', 0, 0)
387
373
self.repo._backup_inventory()
388
ui.ui_factory.note(gettext('Backup Inventory created'))
374
ui.ui_factory.note('Backup Inventory created')
389
375
# asking for '' should never return a non-empty weave
390
376
new_inventories = self.repo._temp_inventories()
391
377
# we have topological order of revisions and non ghost parents ready.
402
388
# the revisionds list
403
389
if not(set(new_inventories.keys()) == set(revision_keys)):
404
390
raise AssertionError()
405
self.pb.update(gettext('Writing weave'))
391
self.pb.update('Writing weave')
406
392
self.repo._activate_new_inventory()
407
393
self.inventory = None
408
ui.ui_factory.note(gettext('Inventory regenerated.'))
394
ui.ui_factory.note('Inventory regenerated.')
410
396
def _fix_text_parents(self):
411
397
"""Fix bad versionedfile parent entries.
443
429
versions_list.append(text_key[1])
444
430
# Do the reconcile of individual weaves.
445
431
for num, file_id in enumerate(per_id_bad_parents):
446
self.pb.update(gettext('Fixing text parents'), num,
432
self.pb.update('Fixing text parents', num,
447
433
len(per_id_bad_parents))
448
434
versions_with_bad_parents = per_id_bad_parents[file_id]
449
435
id_unused_versions = set(key[-1] for key in unused_versions
507
493
# - lock the names list
508
494
# - perform a customised pack() that regenerates data as needed
509
495
# - unlock the names list
510
# https://bugs.launchpad.net/bzr/+bug/154173
512
def __init__(self, repo, other=None, thorough=False,
513
canonicalize_chks=False):
514
super(PackReconciler, self).__init__(repo, other=other,
516
self.canonicalize_chks = canonicalize_chks
496
# https://bugs.edge.launchpad.net/bzr/+bug/154173
518
498
def _reconcile_steps(self):
519
499
"""Perform the steps to reconcile this repository."""
522
502
collection = self.repo._pack_collection
523
503
collection.ensure_loaded()
524
504
collection.lock_names()
525
self.add_cleanup(collection._unlock_names)
526
packs = collection.all_packs()
527
all_revisions = self.repo.all_revision_ids()
528
total_inventories = len(list(
529
collection.inventory_index.combined_index.iter_all_entries()))
530
if len(all_revisions):
531
if self.canonicalize_chks:
532
reconcile_meth = self.repo._canonicalize_chks_pack
506
packs = collection.all_packs()
507
all_revisions = self.repo.all_revision_ids()
508
total_inventories = len(list(
509
collection.inventory_index.combined_index.iter_all_entries()))
510
if len(all_revisions):
511
new_pack = self.repo._reconcile_pack(collection, packs,
512
".reconcile", all_revisions, self.pb)
513
if new_pack is not None:
514
self._discard_and_save(packs)
534
reconcile_meth = self.repo._reconcile_pack
535
new_pack = reconcile_meth(collection, packs, ".reconcile",
536
all_revisions, self.pb)
537
if new_pack is not None:
516
# only make a new pack when there is data to copy.
538
517
self._discard_and_save(packs)
540
# only make a new pack when there is data to copy.
541
self._discard_and_save(packs)
542
self.garbage_inventories = total_inventories - len(list(
543
collection.inventory_index.combined_index.iter_all_entries()))
518
self.garbage_inventories = total_inventories - len(list(
519
collection.inventory_index.combined_index.iter_all_entries()))
521
collection._unlock_names()
545
523
def _discard_and_save(self, packs):
546
524
"""Discard some packs from the repository.
added
removed
bzrlib/reconcile.py
