← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/reconcile.py

  • Committer: John Arbash Meinel
  • Date: 2008-10-04 14:10:13 UTC
  • mto: This revision was merged to the branch mainline in revision 3805.
  • Revision ID: john@arbash-meinel.com-20081004141013-yskxjlwtuy2k18ue
Playing around with expanding requests for btree index nodes into neighboring nodes.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/reconcile.py
1
 
# Copyright (C) 2006-2010 Canonical Ltd
 
1
# Copyright (C) 2005, 2006 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
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
17
17
"""Reconcilers are able to fix some potential data errors in a branch."""
18
18
 
27
27
 
28
28
 
29
29
from bzrlib import (
30
 
    cleanup,
31
30
    errors,
32
 
    revision as _mod_revision,
33
31
    ui,
 
32
    repository,
 
33
    repofmt,
34
34
    )
35
 
from bzrlib.trace import mutter
36
 
from bzrlib.tsort import topo_sort
 
35
from bzrlib.trace import mutter, note
 
36
from bzrlib.tsort import TopoSorter
37
37
from bzrlib.versionedfile import AdapterFactory, FulltextContentFactory
38
38
 
39
39
 
40
 
def reconcile(dir, canonicalize_chks=False):
 
40
def reconcile(dir, other=None):
41
41
    """Reconcile the data in dir.
42
42
 
43
43
    Currently this is limited to a inventory 'reweave'.
47
47
    Directly using Reconciler is recommended for library users that
48
48
    desire fine grained control or analysis of the found issues.
49
49
 
50
 
    :param canonicalize_chks: Make sure CHKs are in canonical form.
 
50
    :param other: another bzrdir to reconcile against.
51
51
    """
52
 
    reconciler = Reconciler(dir, canonicalize_chks=canonicalize_chks)
 
52
    reconciler = Reconciler(dir, other=other)
53
53
    reconciler.reconcile()
54
54
 
55
55
 
56
56
class Reconciler(object):
57
57
    """Reconcilers are used to reconcile existing data."""
58
58
 
59
 
    def __init__(self, dir, other=None, canonicalize_chks=False):
 
59
    def __init__(self, dir, other=None):
60
60
        """Create a Reconciler."""
61
61
        self.bzrdir = dir
62
 
        self.canonicalize_chks = canonicalize_chks
63
62
 
64
63
    def reconcile(self):
65
64
        """Perform reconciliation.
66
 
 
 
65
        
67
66
        After reconciliation the following attributes document found issues:
68
 
 
69
 
        * `inconsistent_parents`: The number of revisions in the repository
70
 
          whose ancestry was being reported incorrectly.
71
 
        * `garbage_inventories`: The number of inventory objects without
72
 
          revisions that were garbage collected.
73
 
        * `fixed_branch_history`: None if there was no branch, False if the
74
 
          branch history was correct, True if the branch history needed to be
75
 
          re-normalized.
 
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.
76
74
        """
77
75
        self.pb = ui.ui_factory.nested_progress_bar()
78
76
        try:
92
90
            # Nothing to check here
93
91
            self.fixed_branch_history = None
94
92
            return
95
 
        ui.ui_factory.note('Reconciling branch %s' % self.branch.base)
 
93
        self.pb.note('Reconciling branch %s',
 
94
                     self.branch.base)
96
95
        branch_reconciler = self.branch.reconcile(thorough=True)
97
96
        self.fixed_branch_history = branch_reconciler.fixed_history
98
97
 
99
98
    def _reconcile_repository(self):
100
99
        self.repo = self.bzrdir.find_repository()
101
 
        ui.ui_factory.note('Reconciling repository %s' %
102
 
            self.repo.user_url)
 
100
        self.pb.note('Reconciling repository %s',
 
101
                     self.repo.bzrdir.root_transport.base)
103
102
        self.pb.update("Reconciling repository", 0, 1)
104
 
        if self.canonicalize_chks:
105
 
            try:
106
 
                self.repo.reconcile_canonicalize_chks
107
 
            except AttributeError:
108
 
                raise errors.BzrError(
109
 
                    "%s cannot canonicalize CHKs." % (self.repo,))
110
 
            repo_reconciler = self.repo.reconcile_canonicalize_chks()
111
 
        else:
112
 
            repo_reconciler = self.repo.reconcile(thorough=True)
 
103
        repo_reconciler = self.repo.reconcile(thorough=True)
113
104
        self.inconsistent_parents = repo_reconciler.inconsistent_parents
114
105
        self.garbage_inventories = repo_reconciler.garbage_inventories
115
106
        if repo_reconciler.aborted:
116
 
            ui.ui_factory.note(
 
107
            self.pb.note(
117
108
                'Reconcile aborted: revision index has inconsistent parents.')
118
 
            ui.ui_factory.note(
 
109
            self.pb.note(
119
110
                'Run "bzr check" for more details.')
120
111
        else:
121
 
            ui.ui_factory.note('Reconciliation complete.')
 
112
            self.pb.note('Reconciliation complete.')
122
113
 
123
114
 
124
115
class BranchReconciler(object):
130
121
        self.branch = a_branch
131
122
 
132
123
    def reconcile(self):
133
 
        operation = cleanup.OperationWithCleanups(self._reconcile)
134
 
        self.add_cleanup = operation.add_cleanup
135
 
        operation.run_simple()
136
 
 
137
 
    def _reconcile(self):
138
124
        self.branch.lock_write()
139
 
        self.add_cleanup(self.branch.unlock)
140
 
        self.pb = ui.ui_factory.nested_progress_bar()
141
 
        self.add_cleanup(self.pb.finished)
142
 
        self._reconcile_steps()
 
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()
143
133
 
144
134
    def _reconcile_steps(self):
145
135
        self._reconcile_revision_history()
146
136
 
147
137
    def _reconcile_revision_history(self):
 
138
        repo = self.branch.repository
148
139
        last_revno, last_revision_id = self.branch.last_revision_info()
149
 
        real_history = []
150
 
        graph = self.branch.repository.get_graph()
151
 
        try:
152
 
            for revid in graph.iter_lefthand_ancestry(
153
 
                    last_revision_id, (_mod_revision.NULL_REVISION,)):
154
 
                real_history.append(revid)
155
 
        except errors.RevisionNotPresent:
156
 
            pass # Hit a ghost left hand parent
 
140
        real_history = list(repo.iter_reverse_revision_history(
 
141
                                last_revision_id))
157
142
        real_history.reverse()
158
143
        if last_revno != len(real_history):
159
144
            self.fixed_history = True
161
146
            # set_revision_history, as this will regenerate it again.
162
147
            # Not really worth a whole BranchReconciler class just for this,
163
148
            # though.
164
 
            ui.ui_factory.note('Fixing last revision info %s => %s' % (
165
 
                 last_revno, len(real_history)))
 
149
            self.pb.note('Fixing last revision info %s => %s',
 
150
                         last_revno, len(real_history))
166
151
            self.branch.set_last_revision_info(len(real_history),
167
152
                                               last_revision_id)
168
153
        else:
169
154
            self.fixed_history = False
170
 
            ui.ui_factory.note('revision_history ok.')
 
155
            self.pb.note('revision_history ok.')
171
156
 
172
157
 
173
158
class RepoReconciler(object):
174
159
    """Reconciler that reconciles a repository.
175
160
 
176
161
    The goal of repository reconciliation is to make any derived data
177
 
    consistent with the core data committed by a user. This can involve
 
162
    consistent with the core data committed by a user. This can involve 
178
163
    reindexing, or removing unreferenced data if that can interfere with
179
164
    queries in a given repository.
180
165
 
196
181
 
197
182
    def reconcile(self):
198
183
        """Perform reconciliation.
199
 
 
 
184
        
200
185
        After reconciliation the following attributes document found issues:
201
 
 
202
 
        * `inconsistent_parents`: The number of revisions in the repository
203
 
          whose ancestry was being reported incorrectly.
204
 
        * `garbage_inventories`: The number of inventory objects without
205
 
          revisions that were garbage collected.
 
186
        inconsistent_parents: The number of revisions in the repository whose
 
187
                              ancestry was being reported incorrectly.
 
188
        garbage_inventories: The number of inventory objects without revisions
 
189
                             that were garbage collected.
206
190
        """
207
 
        operation = cleanup.OperationWithCleanups(self._reconcile)
208
 
        self.add_cleanup = operation.add_cleanup
209
 
        operation.run_simple()
210
 
 
211
 
    def _reconcile(self):
212
191
        self.repo.lock_write()
213
 
        self.add_cleanup(self.repo.unlock)
214
 
        self.pb = ui.ui_factory.nested_progress_bar()
215
 
        self.add_cleanup(self.pb.finished)
216
 
        self._reconcile_steps()
 
192
        try:
 
193
            self.pb = ui.ui_factory.nested_progress_bar()
 
194
            try:
 
195
                self._reconcile_steps()
 
196
            finally:
 
197
                self.pb.finished()
 
198
        finally:
 
199
            self.repo.unlock()
217
200
 
218
201
    def _reconcile_steps(self):
219
202
        """Perform the steps to reconcile this repository."""
221
204
 
222
205
    def _reweave_inventory(self):
223
206
        """Regenerate the inventory weave for the repository from scratch.
224
 
 
225
 
        This is a smart function: it will only do the reweave if doing it
 
207
        
 
208
        This is a smart function: it will only do the reweave if doing it 
226
209
        will correct data issues. The self.thorough flag controls whether
227
210
        only data-loss causing issues (!self.thorough) or all issues
228
211
        (self.thorough) are treated as requiring the reweave.
229
212
        """
 
213
        # local because needing to know about WeaveFile is a wart we want to hide
 
214
        from bzrlib.weave import WeaveFile, Weave
230
215
        transaction = self.repo.get_transaction()
231
 
        self.pb.update('Reading inventory data')
 
216
        self.pb.update('Reading inventory data.')
232
217
        self.inventory = self.repo.inventories
233
218
        self.revisions = self.repo.revisions
234
219
        # the total set of revisions to process
244
229
            # put a revision into the graph.
245
230
            self._graph_revision(rev_id)
246
231
        self._check_garbage_inventories()
247
 
        # if there are no inconsistent_parents and
 
232
        # if there are no inconsistent_parents and 
248
233
        # (no garbage inventories or we are not doing a thorough check)
249
 
        if (not self.inconsistent_parents and
 
234
        if (not self.inconsistent_parents and 
250
235
            (not self.garbage_inventories or not self.thorough)):
251
 
            ui.ui_factory.note('Inventory ok.')
 
236
            self.pb.note('Inventory ok.')
252
237
            return
253
 
        self.pb.update('Backing up inventory', 0, 0)
 
238
        self.pb.update('Backing up inventory...', 0, 0)
254
239
        self.repo._backup_inventory()
255
 
        ui.ui_factory.note('Backup inventory created.')
 
240
        self.pb.note('Backup Inventory created.')
256
241
        new_inventories = self.repo._temp_inventories()
257
242
 
258
243
        # we have topological order of revisions and non ghost parents ready.
259
244
        self._setup_steps(len(self._rev_graph))
260
 
        revision_keys = [(rev_id,) for rev_id in topo_sort(self._rev_graph)]
 
245
        revision_keys = [(rev_id,) for rev_id in
 
246
            TopoSorter(self._rev_graph.items()).iter_topo_order()]
261
247
        stream = self._change_inv_parents(
262
248
            self.inventory.get_record_stream(revision_keys, 'unordered', True),
263
249
            self._new_inv_parents,
271
257
        self.pb.update('Writing weave')
272
258
        self.repo._activate_new_inventory()
273
259
        self.inventory = None
274
 
        ui.ui_factory.note('Inventory regenerated.')
 
260
        self.pb.note('Inventory regenerated.')
275
261
 
276
262
    def _new_inv_parents(self, revision_key):
277
263
        """Lookup ghost-filtered parents for revision_key."""
365
351
    def _load_indexes(self):
366
352
        """Load indexes for the reconciliation."""
367
353
        self.transaction = self.repo.get_transaction()
368
 
        self.pb.update('Reading indexes', 0, 2)
 
354
        self.pb.update('Reading indexes.', 0, 2)
369
355
        self.inventory = self.repo.inventories
370
 
        self.pb.update('Reading indexes', 1, 2)
 
356
        self.pb.update('Reading indexes.', 1, 2)
371
357
        self.repo._check_for_inconsistent_revision_parents()
372
358
        self.revisions = self.repo.revisions
373
 
        self.pb.update('Reading indexes', 2, 2)
 
359
        self.pb.update('Reading indexes.', 2, 2)
374
360
 
375
361
    def _gc_inventory(self):
376
362
        """Remove inventories that are not referenced from the revision store."""
377
 
        self.pb.update('Checking unused inventories', 0, 1)
 
363
        self.pb.update('Checking unused inventories.', 0, 1)
378
364
        self._check_garbage_inventories()
379
 
        self.pb.update('Checking unused inventories', 1, 3)
 
365
        self.pb.update('Checking unused inventories.', 1, 3)
380
366
        if not self.garbage_inventories:
381
 
            ui.ui_factory.note('Inventory ok.')
 
367
            self.pb.note('Inventory ok.')
382
368
            return
383
 
        self.pb.update('Backing up inventory', 0, 0)
 
369
        self.pb.update('Backing up inventory...', 0, 0)
384
370
        self.repo._backup_inventory()
385
 
        ui.ui_factory.note('Backup Inventory created')
 
371
        self.pb.note('Backup Inventory created.')
386
372
        # asking for '' should never return a non-empty weave
387
373
        new_inventories = self.repo._temp_inventories()
388
374
        # we have topological order of revisions and non ghost parents ready.
389
375
        graph = self.revisions.get_parent_map(self.revisions.keys())
390
 
        revision_keys = topo_sort(graph)
 
376
        revision_keys = list(TopoSorter(graph).iter_topo_order())
391
377
        revision_ids = [key[-1] for key in revision_keys]
392
378
        self._setup_steps(len(revision_keys))
393
379
        stream = self._change_inv_parents(
402
388
        self.pb.update('Writing weave')
403
389
        self.repo._activate_new_inventory()
404
390
        self.inventory = None
405
 
        ui.ui_factory.note('Inventory regenerated.')
 
391
        self.pb.note('Inventory regenerated.')
406
392
 
407
393
    def _fix_text_parents(self):
408
394
        """Fix bad versionedfile parent entries.
504
490
    #  - lock the names list
505
491
    #  - perform a customised pack() that regenerates data as needed
506
492
    #  - unlock the names list
507
 
    # https://bugs.launchpad.net/bzr/+bug/154173
508
 
 
509
 
    def __init__(self, repo, other=None, thorough=False,
510
 
            canonicalize_chks=False):
511
 
        super(PackReconciler, self).__init__(repo, other=other,
512
 
            thorough=thorough)
513
 
        self.canonicalize_chks = canonicalize_chks
 
493
    # https://bugs.edge.launchpad.net/bzr/+bug/154173
514
494
 
515
495
    def _reconcile_steps(self):
516
496
        """Perform the steps to reconcile this repository."""
519
499
        collection = self.repo._pack_collection
520
500
        collection.ensure_loaded()
521
501
        collection.lock_names()
522
 
        self.add_cleanup(collection._unlock_names)
523
 
        packs = collection.all_packs()
524
 
        all_revisions = self.repo.all_revision_ids()
525
 
        total_inventories = len(list(
526
 
            collection.inventory_index.combined_index.iter_all_entries()))
527
 
        if len(all_revisions):
528
 
            if self.canonicalize_chks:
529
 
                reconcile_meth = self.repo._canonicalize_chks_pack
 
502
        try:
 
503
            packs = collection.all_packs()
 
504
            all_revisions = self.repo.all_revision_ids()
 
505
            total_inventories = len(list(
 
506
                collection.inventory_index.combined_index.iter_all_entries()))
 
507
            if len(all_revisions):
 
508
                self._packer = repofmt.pack_repo.ReconcilePacker(
 
509
                    collection, packs, ".reconcile", all_revisions)
 
510
                new_pack = self._packer.pack(pb=self.pb)
 
511
                if new_pack is not None:
 
512
                    self._discard_and_save(packs)
530
513
            else:
531
 
                reconcile_meth = self.repo._reconcile_pack
532
 
            new_pack = reconcile_meth(collection, packs, ".reconcile",
533
 
                all_revisions, self.pb)
534
 
            if new_pack is not None:
 
514
                # only make a new pack when there is data to copy.
535
515
                self._discard_and_save(packs)
536
 
        else:
537
 
            # only make a new pack when there is data to copy.
538
 
            self._discard_and_save(packs)
539
 
        self.garbage_inventories = total_inventories - len(list(
540
 
            collection.inventory_index.combined_index.iter_all_entries()))
 
516
            self.garbage_inventories = total_inventories - len(list(
 
517
                collection.inventory_index.combined_index.iter_all_entries()))
 
518
        finally:
 
519
            collection._unlock_names()
541
520
 
542
521
    def _discard_and_save(self, packs):
543
522
        """Discard some packs from the repository.