← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/reconcile.py

  • Committer: Martin Pool
  • Date: 2009-01-13 03:11:04 UTC
  • mto: This revision was merged to the branch mainline in revision 3937.
  • Revision ID: mbp@sourcefrog.net-20090113031104-03my054s02i9l2pe
Bump version to 1.12 and add news template

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
31
    ui,
 
32
    repository,
 
33
    repofmt,
33
34
    )
34
 
from bzrlib.trace import mutter
35
 
from bzrlib.tsort import topo_sort
 
35
from bzrlib.trace import mutter, note
 
36
from bzrlib.tsort import TopoSorter
36
37
from bzrlib.versionedfile import AdapterFactory, FulltextContentFactory
37
38
 
38
39
 
61
62
 
62
63
    def reconcile(self):
63
64
        """Perform reconciliation.
64
 
 
 
65
        
65
66
        After reconciliation the following attributes document found issues:
66
67
        inconsistent_parents: The number of revisions in the repository whose
67
68
                              ancestry was being reported incorrectly.
89
90
            # Nothing to check here
90
91
            self.fixed_branch_history = None
91
92
            return
92
 
        ui.ui_factory.note('Reconciling branch %s' % self.branch.base)
 
93
        self.pb.note('Reconciling branch %s',
 
94
                     self.branch.base)
93
95
        branch_reconciler = self.branch.reconcile(thorough=True)
94
96
        self.fixed_branch_history = branch_reconciler.fixed_history
95
97
 
96
98
    def _reconcile_repository(self):
97
99
        self.repo = self.bzrdir.find_repository()
98
 
        ui.ui_factory.note('Reconciling repository %s' %
99
 
            self.repo.user_url)
 
100
        self.pb.note('Reconciling repository %s',
 
101
                     self.repo.bzrdir.root_transport.base)
100
102
        self.pb.update("Reconciling repository", 0, 1)
101
103
        repo_reconciler = self.repo.reconcile(thorough=True)
102
104
        self.inconsistent_parents = repo_reconciler.inconsistent_parents
103
105
        self.garbage_inventories = repo_reconciler.garbage_inventories
104
106
        if repo_reconciler.aborted:
105
 
            ui.ui_factory.note(
 
107
            self.pb.note(
106
108
                'Reconcile aborted: revision index has inconsistent parents.')
107
 
            ui.ui_factory.note(
 
109
            self.pb.note(
108
110
                'Run "bzr check" for more details.')
109
111
        else:
110
 
            ui.ui_factory.note('Reconciliation complete.')
 
112
            self.pb.note('Reconciliation complete.')
111
113
 
112
114
 
113
115
class BranchReconciler(object):
119
121
        self.branch = a_branch
120
122
 
121
123
    def reconcile(self):
122
 
        operation = cleanup.OperationWithCleanups(self._reconcile)
123
 
        self.add_cleanup = operation.add_cleanup
124
 
        operation.run_simple()
125
 
 
126
 
    def _reconcile(self):
127
124
        self.branch.lock_write()
128
 
        self.add_cleanup(self.branch.unlock)
129
 
        self.pb = ui.ui_factory.nested_progress_bar()
130
 
        self.add_cleanup(self.pb.finished)
131
 
        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()
132
133
 
133
134
    def _reconcile_steps(self):
134
135
        self._reconcile_revision_history()
136
137
    def _reconcile_revision_history(self):
137
138
        repo = self.branch.repository
138
139
        last_revno, last_revision_id = self.branch.last_revision_info()
139
 
        real_history = []
140
 
        try:
141
 
            for revid in repo.iter_reverse_revision_history(
142
 
                    last_revision_id):
143
 
                real_history.append(revid)
144
 
        except errors.RevisionNotPresent:
145
 
            pass # Hit a ghost left hand parent
 
140
        real_history = list(repo.iter_reverse_revision_history(
 
141
                                last_revision_id))
146
142
        real_history.reverse()
147
143
        if last_revno != len(real_history):
148
144
            self.fixed_history = True
150
146
            # set_revision_history, as this will regenerate it again.
151
147
            # Not really worth a whole BranchReconciler class just for this,
152
148
            # though.
153
 
            ui.ui_factory.note('Fixing last revision info %s => %s' % (
154
 
                 last_revno, len(real_history)))
 
149
            self.pb.note('Fixing last revision info %s => %s',
 
150
                         last_revno, len(real_history))
155
151
            self.branch.set_last_revision_info(len(real_history),
156
152
                                               last_revision_id)
157
153
        else:
158
154
            self.fixed_history = False
159
 
            ui.ui_factory.note('revision_history ok.')
 
155
            self.pb.note('revision_history ok.')
160
156
 
161
157
 
162
158
class RepoReconciler(object):
163
159
    """Reconciler that reconciles a repository.
164
160
 
165
161
    The goal of repository reconciliation is to make any derived data
166
 
    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 
167
163
    reindexing, or removing unreferenced data if that can interfere with
168
164
    queries in a given repository.
169
165
 
185
181
 
186
182
    def reconcile(self):
187
183
        """Perform reconciliation.
188
 
 
 
184
        
189
185
        After reconciliation the following attributes document found issues:
190
186
        inconsistent_parents: The number of revisions in the repository whose
191
187
                              ancestry was being reported incorrectly.
192
188
        garbage_inventories: The number of inventory objects without revisions
193
189
                             that were garbage collected.
194
190
        """
195
 
        operation = cleanup.OperationWithCleanups(self._reconcile)
196
 
        self.add_cleanup = operation.add_cleanup
197
 
        operation.run_simple()
198
 
 
199
 
    def _reconcile(self):
200
191
        self.repo.lock_write()
201
 
        self.add_cleanup(self.repo.unlock)
202
 
        self.pb = ui.ui_factory.nested_progress_bar()
203
 
        self.add_cleanup(self.pb.finished)
204
 
        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()
205
200
 
206
201
    def _reconcile_steps(self):
207
202
        """Perform the steps to reconcile this repository."""
209
204
 
210
205
    def _reweave_inventory(self):
211
206
        """Regenerate the inventory weave for the repository from scratch.
212
 
 
213
 
        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 
214
209
        will correct data issues. The self.thorough flag controls whether
215
210
        only data-loss causing issues (!self.thorough) or all issues
216
211
        (self.thorough) are treated as requiring the reweave.
218
213
        # local because needing to know about WeaveFile is a wart we want to hide
219
214
        from bzrlib.weave import WeaveFile, Weave
220
215
        transaction = self.repo.get_transaction()
221
 
        self.pb.update('Reading inventory data')
 
216
        self.pb.update('Reading inventory data.')
222
217
        self.inventory = self.repo.inventories
223
218
        self.revisions = self.repo.revisions
224
219
        # the total set of revisions to process
234
229
            # put a revision into the graph.
235
230
            self._graph_revision(rev_id)
236
231
        self._check_garbage_inventories()
237
 
        # if there are no inconsistent_parents and
 
232
        # if there are no inconsistent_parents and 
238
233
        # (no garbage inventories or we are not doing a thorough check)
239
 
        if (not self.inconsistent_parents and
 
234
        if (not self.inconsistent_parents and 
240
235
            (not self.garbage_inventories or not self.thorough)):
241
 
            ui.ui_factory.note('Inventory ok.')
 
236
            self.pb.note('Inventory ok.')
242
237
            return
243
 
        self.pb.update('Backing up inventory', 0, 0)
 
238
        self.pb.update('Backing up inventory...', 0, 0)
244
239
        self.repo._backup_inventory()
245
 
        ui.ui_factory.note('Backup inventory created.')
 
240
        self.pb.note('Backup Inventory created.')
246
241
        new_inventories = self.repo._temp_inventories()
247
242
 
248
243
        # we have topological order of revisions and non ghost parents ready.
249
244
        self._setup_steps(len(self._rev_graph))
250
 
        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()]
251
247
        stream = self._change_inv_parents(
252
248
            self.inventory.get_record_stream(revision_keys, 'unordered', True),
253
249
            self._new_inv_parents,
261
257
        self.pb.update('Writing weave')
262
258
        self.repo._activate_new_inventory()
263
259
        self.inventory = None
264
 
        ui.ui_factory.note('Inventory regenerated.')
 
260
        self.pb.note('Inventory regenerated.')
265
261
 
266
262
    def _new_inv_parents(self, revision_key):
267
263
        """Lookup ghost-filtered parents for revision_key."""
355
351
    def _load_indexes(self):
356
352
        """Load indexes for the reconciliation."""
357
353
        self.transaction = self.repo.get_transaction()
358
 
        self.pb.update('Reading indexes', 0, 2)
 
354
        self.pb.update('Reading indexes.', 0, 2)
359
355
        self.inventory = self.repo.inventories
360
 
        self.pb.update('Reading indexes', 1, 2)
 
356
        self.pb.update('Reading indexes.', 1, 2)
361
357
        self.repo._check_for_inconsistent_revision_parents()
362
358
        self.revisions = self.repo.revisions
363
 
        self.pb.update('Reading indexes', 2, 2)
 
359
        self.pb.update('Reading indexes.', 2, 2)
364
360
 
365
361
    def _gc_inventory(self):
366
362
        """Remove inventories that are not referenced from the revision store."""
367
 
        self.pb.update('Checking unused inventories', 0, 1)
 
363
        self.pb.update('Checking unused inventories.', 0, 1)
368
364
        self._check_garbage_inventories()
369
 
        self.pb.update('Checking unused inventories', 1, 3)
 
365
        self.pb.update('Checking unused inventories.', 1, 3)
370
366
        if not self.garbage_inventories:
371
 
            ui.ui_factory.note('Inventory ok.')
 
367
            self.pb.note('Inventory ok.')
372
368
            return
373
 
        self.pb.update('Backing up inventory', 0, 0)
 
369
        self.pb.update('Backing up inventory...', 0, 0)
374
370
        self.repo._backup_inventory()
375
 
        ui.ui_factory.note('Backup Inventory created')
 
371
        self.pb.note('Backup Inventory created.')
376
372
        # asking for '' should never return a non-empty weave
377
373
        new_inventories = self.repo._temp_inventories()
378
374
        # we have topological order of revisions and non ghost parents ready.
379
375
        graph = self.revisions.get_parent_map(self.revisions.keys())
380
 
        revision_keys = topo_sort(graph)
 
376
        revision_keys = list(TopoSorter(graph).iter_topo_order())
381
377
        revision_ids = [key[-1] for key in revision_keys]
382
378
        self._setup_steps(len(revision_keys))
383
379
        stream = self._change_inv_parents(
392
388
        self.pb.update('Writing weave')
393
389
        self.repo._activate_new_inventory()
394
390
        self.inventory = None
395
 
        ui.ui_factory.note('Inventory regenerated.')
 
391
        self.pb.note('Inventory regenerated.')
396
392
 
397
393
    def _fix_text_parents(self):
398
394
        """Fix bad versionedfile parent entries.
503
499
        collection = self.repo._pack_collection
504
500
        collection.ensure_loaded()
505
501
        collection.lock_names()
506
 
        self.add_cleanup(collection._unlock_names)
507
 
        packs = collection.all_packs()
508
 
        all_revisions = self.repo.all_revision_ids()
509
 
        total_inventories = len(list(
510
 
            collection.inventory_index.combined_index.iter_all_entries()))
511
 
        if len(all_revisions):
512
 
            new_pack =  self.repo._reconcile_pack(collection, packs,
513
 
                ".reconcile", all_revisions, self.pb)
514
 
            if new_pack is not None:
 
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)
 
513
            else:
 
514
                # only make a new pack when there is data to copy.
515
515
                self._discard_and_save(packs)
516
 
        else:
517
 
            # only make a new pack when there is data to copy.
518
 
            self._discard_and_save(packs)
519
 
        self.garbage_inventories = total_inventories - len(list(
520
 
            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()
521
520
 
522
521
    def _discard_and_save(self, packs):
523
522
        """Discard some packs from the repository.