← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/reconcile.py

  • Committer: Vincent Ladeuil
  • Date: 2017-01-17 13:48:10 UTC
  • mfrom: (6615.3.6 merges)
  • mto: This revision was merged to the branch mainline in revision 6620.
  • Revision ID: v.ladeuil+lp@free.fr-20170117134810-j9p3lidfy6pfyfsc
Merge 2.7, resolving conflicts

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/reconcile.py
1
 
# Copyright (C) 2005, 2006 Canonical Ltd
 
1
# Copyright (C) 2006-2010 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
16
16
 
17
17
"""Reconcilers are able to fix some potential data errors in a branch."""
18
18
 
 
19
from __future__ import absolute_import
19
20
 
20
21
__all__ = [
21
22
    'KnitReconciler',
27
28
 
28
29
 
29
30
from bzrlib import (
 
31
    cleanup,
30
32
    errors,
 
33
    revision as _mod_revision,
31
34
    ui,
32
 
    repository,
33
 
    repofmt,
34
35
    )
35
 
from bzrlib.trace import mutter, note
36
 
from bzrlib.tsort import TopoSorter
 
36
from bzrlib.trace import mutter
 
37
from bzrlib.tsort import topo_sort
37
38
from bzrlib.versionedfile import AdapterFactory, FulltextContentFactory
38
 
 
39
 
 
40
 
def reconcile(dir, other=None):
 
39
from bzrlib.i18n import gettext
 
40
 
 
41
 
 
42
def reconcile(dir, canonicalize_chks=False):
41
43
    """Reconcile the data in dir.
42
44
 
43
45
    Currently this is limited to a inventory 'reweave'.
47
49
    Directly using Reconciler is recommended for library users that
48
50
    desire fine grained control or analysis of the found issues.
49
51
 
50
 
    :param other: another bzrdir to reconcile against.
 
52
    :param canonicalize_chks: Make sure CHKs are in canonical form.
51
53
    """
52
 
    reconciler = Reconciler(dir, other=other)
 
54
    reconciler = Reconciler(dir, canonicalize_chks=canonicalize_chks)
53
55
    reconciler.reconcile()
54
56
 
55
57
 
56
58
class Reconciler(object):
57
59
    """Reconcilers are used to reconcile existing data."""
58
60
 
59
 
    def __init__(self, dir, other=None):
 
61
    def __init__(self, dir, other=None, canonicalize_chks=False):
60
62
        """Create a Reconciler."""
61
63
        self.bzrdir = dir
 
64
        self.canonicalize_chks = canonicalize_chks
62
65
 
63
66
    def reconcile(self):
64
67
        """Perform reconciliation.
65
68
 
66
69
        After reconciliation the following attributes document found issues:
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.
 
70
 
 
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
 
77
          re-normalized.
74
78
        """
75
79
        self.pb = ui.ui_factory.nested_progress_bar()
76
80
        try:
90
94
            # Nothing to check here
91
95
            self.fixed_branch_history = None
92
96
            return
93
 
        self.pb.note('Reconciling branch %s',
94
 
                     self.branch.base)
 
97
        ui.ui_factory.note(gettext('Reconciling branch %s') % self.branch.base)
95
98
        branch_reconciler = self.branch.reconcile(thorough=True)
96
99
        self.fixed_branch_history = branch_reconciler.fixed_history
97
100
 
98
101
    def _reconcile_repository(self):
99
102
        self.repo = self.bzrdir.find_repository()
100
 
        self.pb.note('Reconciling repository %s',
101
 
                     self.repo.bzrdir.root_transport.base)
102
 
        self.pb.update("Reconciling repository", 0, 1)
103
 
        repo_reconciler = self.repo.reconcile(thorough=True)
 
103
        ui.ui_factory.note(gettext('Reconciling repository %s') %
 
104
            self.repo.user_url)
 
105
        self.pb.update(gettext("Reconciling repository"), 0, 1)
 
106
        if self.canonicalize_chks:
 
107
            try:
 
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()
 
113
        else:
 
114
            repo_reconciler = self.repo.reconcile(thorough=True)
104
115
        self.inconsistent_parents = repo_reconciler.inconsistent_parents
105
116
        self.garbage_inventories = repo_reconciler.garbage_inventories
106
117
        if repo_reconciler.aborted:
107
 
            self.pb.note(
108
 
                'Reconcile aborted: revision index has inconsistent parents.')
109
 
            self.pb.note(
110
 
                'Run "bzr check" for more details.')
 
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.'))
111
122
        else:
112
 
            self.pb.note('Reconciliation complete.')
 
123
            ui.ui_factory.note(gettext('Reconciliation complete.'))
113
124
 
114
125
 
115
126
class BranchReconciler(object):
121
132
        self.branch = a_branch
122
133
 
123
134
    def reconcile(self):
 
135
        operation = cleanup.OperationWithCleanups(self._reconcile)
 
136
        self.add_cleanup = operation.add_cleanup
 
137
        operation.run_simple()
 
138
 
 
139
    def _reconcile(self):
124
140
        self.branch.lock_write()
125
 
        try:
126
 
            self.pb = ui.ui_factory.nested_progress_bar()
127
 
            try:
128
 
                self._reconcile_steps()
129
 
            finally:
130
 
                self.pb.finished()
131
 
        finally:
132
 
            self.branch.unlock()
 
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()
133
145
 
134
146
    def _reconcile_steps(self):
135
147
        self._reconcile_revision_history()
136
148
 
137
149
    def _reconcile_revision_history(self):
138
 
        repo = self.branch.repository
139
150
        last_revno, last_revision_id = self.branch.last_revision_info()
140
151
        real_history = []
 
152
        graph = self.branch.repository.get_graph()
141
153
        try:
142
 
            for revid in repo.iter_reverse_revision_history(
143
 
                    last_revision_id):
 
154
            for revid in graph.iter_lefthand_ancestry(
 
155
                    last_revision_id, (_mod_revision.NULL_REVISION,)):
144
156
                real_history.append(revid)
145
157
        except errors.RevisionNotPresent:
146
158
            pass # Hit a ghost left hand parent
151
163
            # set_revision_history, as this will regenerate it again.
152
164
            # Not really worth a whole BranchReconciler class just for this,
153
165
            # though.
154
 
            self.pb.note('Fixing last revision info %s => %s',
155
 
                         last_revno, len(real_history))
 
166
            ui.ui_factory.note(gettext('Fixing last revision info {0} '\
 
167
                                       ' => {1}').format(
 
168
                                       last_revno, len(real_history)))
156
169
            self.branch.set_last_revision_info(len(real_history),
157
170
                                               last_revision_id)
158
171
        else:
159
172
            self.fixed_history = False
160
 
            self.pb.note('revision_history ok.')
 
173
            ui.ui_factory.note(gettext('revision_history ok.'))
161
174
 
162
175
 
163
176
class RepoReconciler(object):
188
201
        """Perform reconciliation.
189
202
 
190
203
        After reconciliation the following attributes document found issues:
191
 
        inconsistent_parents: The number of revisions in the repository whose
192
 
                              ancestry was being reported incorrectly.
193
 
        garbage_inventories: The number of inventory objects without revisions
194
 
                             that were garbage collected.
 
204
 
 
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.
195
209
        """
 
210
        operation = cleanup.OperationWithCleanups(self._reconcile)
 
211
        self.add_cleanup = operation.add_cleanup
 
212
        operation.run_simple()
 
213
 
 
214
    def _reconcile(self):
196
215
        self.repo.lock_write()
197
 
        try:
198
 
            self.pb = ui.ui_factory.nested_progress_bar()
199
 
            try:
200
 
                self._reconcile_steps()
201
 
            finally:
202
 
                self.pb.finished()
203
 
        finally:
204
 
            self.repo.unlock()
 
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()
205
220
 
206
221
    def _reconcile_steps(self):
207
222
        """Perform the steps to reconcile this repository."""
215
230
        only data-loss causing issues (!self.thorough) or all issues
216
231
        (self.thorough) are treated as requiring the reweave.
217
232
        """
218
 
        # local because needing to know about WeaveFile is a wart we want to hide
219
 
        from bzrlib.weave import WeaveFile, Weave
220
233
        transaction = self.repo.get_transaction()
221
 
        self.pb.update('Reading inventory data')
 
234
        self.pb.update(gettext('Reading inventory data'))
222
235
        self.inventory = self.repo.inventories
223
236
        self.revisions = self.repo.revisions
224
237
        # the total set of revisions to process
238
251
        # (no garbage inventories or we are not doing a thorough check)
239
252
        if (not self.inconsistent_parents and
240
253
            (not self.garbage_inventories or not self.thorough)):
241
 
            self.pb.note('Inventory ok.')
 
254
            ui.ui_factory.note(gettext('Inventory ok.'))
242
255
            return
243
 
        self.pb.update('Backing up inventory', 0, 0)
 
256
        self.pb.update(gettext('Backing up inventory'), 0, 0)
244
257
        self.repo._backup_inventory()
245
 
        self.pb.note('Backup inventory created.')
 
258
        ui.ui_factory.note(gettext('Backup inventory created.'))
246
259
        new_inventories = self.repo._temp_inventories()
247
260
 
248
261
        # we have topological order of revisions and non ghost parents ready.
249
262
        self._setup_steps(len(self._rev_graph))
250
 
        revision_keys = [(rev_id,) for rev_id in
251
 
            TopoSorter(self._rev_graph.items()).iter_topo_order()]
 
263
        revision_keys = [(rev_id,) for rev_id in topo_sort(self._rev_graph)]
252
264
        stream = self._change_inv_parents(
253
265
            self.inventory.get_record_stream(revision_keys, 'unordered', True),
254
266
            self._new_inv_parents,
259
271
        if not (set(new_inventories.keys()) ==
260
272
            set([(revid,) for revid in self.pending])):
261
273
            raise AssertionError()
262
 
        self.pb.update('Writing weave')
 
274
        self.pb.update(gettext('Writing weave'))
263
275
        self.repo._activate_new_inventory()
264
276
        self.inventory = None
265
 
        self.pb.note('Inventory regenerated.')
 
277
        ui.ui_factory.note(gettext('Inventory regenerated.'))
266
278
 
267
279
    def _new_inv_parents(self, revision_key):
268
280
        """Lookup ghost-filtered parents for revision_key."""
356
368
    def _load_indexes(self):
357
369
        """Load indexes for the reconciliation."""
358
370
        self.transaction = self.repo.get_transaction()
359
 
        self.pb.update('Reading indexes', 0, 2)
 
371
        self.pb.update(gettext('Reading indexes'), 0, 2)
360
372
        self.inventory = self.repo.inventories
361
 
        self.pb.update('Reading indexes', 1, 2)
 
373
        self.pb.update(gettext('Reading indexes'), 1, 2)
362
374
        self.repo._check_for_inconsistent_revision_parents()
363
375
        self.revisions = self.repo.revisions
364
 
        self.pb.update('Reading indexes', 2, 2)
 
376
        self.pb.update(gettext('Reading indexes'), 2, 2)
365
377
 
366
378
    def _gc_inventory(self):
367
379
        """Remove inventories that are not referenced from the revision store."""
368
 
        self.pb.update('Checking unused inventories', 0, 1)
 
380
        self.pb.update(gettext('Checking unused inventories'), 0, 1)
369
381
        self._check_garbage_inventories()
370
 
        self.pb.update('Checking unused inventories', 1, 3)
 
382
        self.pb.update(gettext('Checking unused inventories'), 1, 3)
371
383
        if not self.garbage_inventories:
372
 
            self.pb.note('Inventory ok.')
 
384
            ui.ui_factory.note(gettext('Inventory ok.'))
373
385
            return
374
 
        self.pb.update('Backing up inventory', 0, 0)
 
386
        self.pb.update(gettext('Backing up inventory'), 0, 0)
375
387
        self.repo._backup_inventory()
376
 
        self.pb.note('Backup Inventory created')
 
388
        ui.ui_factory.note(gettext('Backup Inventory created'))
377
389
        # asking for '' should never return a non-empty weave
378
390
        new_inventories = self.repo._temp_inventories()
379
391
        # we have topological order of revisions and non ghost parents ready.
380
392
        graph = self.revisions.get_parent_map(self.revisions.keys())
381
 
        revision_keys = list(TopoSorter(graph).iter_topo_order())
 
393
        revision_keys = topo_sort(graph)
382
394
        revision_ids = [key[-1] for key in revision_keys]
383
395
        self._setup_steps(len(revision_keys))
384
396
        stream = self._change_inv_parents(
390
402
        # the revisionds list
391
403
        if not(set(new_inventories.keys()) == set(revision_keys)):
392
404
            raise AssertionError()
393
 
        self.pb.update('Writing weave')
 
405
        self.pb.update(gettext('Writing weave'))
394
406
        self.repo._activate_new_inventory()
395
407
        self.inventory = None
396
 
        self.pb.note('Inventory regenerated.')
 
408
        ui.ui_factory.note(gettext('Inventory regenerated.'))
397
409
 
398
410
    def _fix_text_parents(self):
399
411
        """Fix bad versionedfile parent entries.
431
443
            versions_list.append(text_key[1])
432
444
        # Do the reconcile of individual weaves.
433
445
        for num, file_id in enumerate(per_id_bad_parents):
434
 
            self.pb.update('Fixing text parents', num,
 
446
            self.pb.update(gettext('Fixing text parents'), num,
435
447
                           len(per_id_bad_parents))
436
448
            versions_with_bad_parents = per_id_bad_parents[file_id]
437
449
            id_unused_versions = set(key[-1] for key in unused_versions
495
507
    #  - lock the names list
496
508
    #  - perform a customised pack() that regenerates data as needed
497
509
    #  - unlock the names list
498
 
    # https://bugs.edge.launchpad.net/bzr/+bug/154173
 
510
    # https://bugs.launchpad.net/bzr/+bug/154173
 
511
 
 
512
    def __init__(self, repo, other=None, thorough=False,
 
513
            canonicalize_chks=False):
 
514
        super(PackReconciler, self).__init__(repo, other=other,
 
515
            thorough=thorough)
 
516
        self.canonicalize_chks = canonicalize_chks
499
517
 
500
518
    def _reconcile_steps(self):
501
519
        """Perform the steps to reconcile this repository."""
504
522
        collection = self.repo._pack_collection
505
523
        collection.ensure_loaded()
506
524
        collection.lock_names()
507
 
        try:
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)
 
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
517
533
            else:
518
 
                # only make a new pack when there is data to copy.
 
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:
519
538
                self._discard_and_save(packs)
520
 
            self.garbage_inventories = total_inventories - len(list(
521
 
                collection.inventory_index.combined_index.iter_all_entries()))
522
 
        finally:
523
 
            collection._unlock_names()
 
539
        else:
 
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()))
524
544
 
525
545
    def _discard_and_save(self, packs):
526
546
        """Discard some packs from the repository.