← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/reconcile.py

  • Committer: Andrew Bennetts
  • Date: 2008-07-03 07:56:02 UTC
  • mto: This revision was merged to the branch mainline in revision 3520.
  • Revision ID: andrew.bennetts@canonical.com-20080703075602-8n055qsfkjijcz6i
Better tests for {pre,post}_change_branch_tip hooks.

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
 
39
 
def reconcile(dir, canonicalize_chks=False):
 
40
def reconcile(dir, other=None):
40
41
    """Reconcile the data in dir.
41
42
 
42
43
    Currently this is limited to a inventory 'reweave'.
46
47
    Directly using Reconciler is recommended for library users that
47
48
    desire fine grained control or analysis of the found issues.
48
49
 
49
 
    :param canonicalize_chks: Make sure CHKs are in canonical form.
 
50
    :param other: another bzrdir to reconcile against.
50
51
    """
51
 
    reconciler = Reconciler(dir, canonicalize_chks=canonicalize_chks)
 
52
    reconciler = Reconciler(dir, other=other)
52
53
    reconciler.reconcile()
53
54
 
54
55
 
55
56
class Reconciler(object):
56
57
    """Reconcilers are used to reconcile existing data."""
57
58
 
58
 
    def __init__(self, dir, other=None, canonicalize_chks=False):
 
59
    def __init__(self, dir, other=None):
59
60
        """Create a Reconciler."""
60
61
        self.bzrdir = dir
61
 
        self.canonicalize_chks = canonicalize_chks
62
62
 
63
63
    def reconcile(self):
64
64
        """Perform reconciliation.
65
 
 
 
65
        
66
66
        After reconciliation the following attributes document found issues:
67
67
        inconsistent_parents: The number of revisions in the repository whose
68
68
                              ancestry was being reported incorrectly.
90
90
            # Nothing to check here
91
91
            self.fixed_branch_history = None
92
92
            return
93
 
        ui.ui_factory.note('Reconciling branch %s' % self.branch.base)
 
93
        self.pb.note('Reconciling branch %s',
 
94
                     self.branch.base)
94
95
        branch_reconciler = self.branch.reconcile(thorough=True)
95
96
        self.fixed_branch_history = branch_reconciler.fixed_history
96
97
 
97
98
    def _reconcile_repository(self):
98
99
        self.repo = self.bzrdir.find_repository()
99
 
        ui.ui_factory.note('Reconciling repository %s' %
100
 
            self.repo.user_url)
 
100
        self.pb.note('Reconciling repository %s',
 
101
                     self.repo.bzrdir.root_transport.base)
101
102
        self.pb.update("Reconciling repository", 0, 1)
102
 
        if self.canonicalize_chks:
103
 
            try:
104
 
                self.repo.reconcile_canonicalize_chks
105
 
            except AttributeError:
106
 
                raise errors.BzrError(
107
 
                    "%s cannot canonicalize CHKs." % (self.repo,))
108
 
            repo_reconciler = self.repo.reconcile_canonicalize_chks()
109
 
        else:
110
 
            repo_reconciler = self.repo.reconcile(thorough=True)
 
103
        repo_reconciler = self.repo.reconcile(thorough=True)
111
104
        self.inconsistent_parents = repo_reconciler.inconsistent_parents
112
105
        self.garbage_inventories = repo_reconciler.garbage_inventories
113
106
        if repo_reconciler.aborted:
114
 
            ui.ui_factory.note(
 
107
            self.pb.note(
115
108
                'Reconcile aborted: revision index has inconsistent parents.')
116
 
            ui.ui_factory.note(
 
109
            self.pb.note(
117
110
                'Run "bzr check" for more details.')
118
111
        else:
119
 
            ui.ui_factory.note('Reconciliation complete.')
 
112
            self.pb.note('Reconciliation complete.')
120
113
 
121
114
 
122
115
class BranchReconciler(object):
128
121
        self.branch = a_branch
129
122
 
130
123
    def reconcile(self):
131
 
        operation = cleanup.OperationWithCleanups(self._reconcile)
132
 
        self.add_cleanup = operation.add_cleanup
133
 
        operation.run_simple()
134
 
 
135
 
    def _reconcile(self):
136
124
        self.branch.lock_write()
137
 
        self.add_cleanup(self.branch.unlock)
138
 
        self.pb = ui.ui_factory.nested_progress_bar()
139
 
        self.add_cleanup(self.pb.finished)
140
 
        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()
141
133
 
142
134
    def _reconcile_steps(self):
143
135
        self._reconcile_revision_history()
145
137
    def _reconcile_revision_history(self):
146
138
        repo = self.branch.repository
147
139
        last_revno, last_revision_id = self.branch.last_revision_info()
148
 
        real_history = []
149
 
        try:
150
 
            for revid in repo.iter_reverse_revision_history(
151
 
                    last_revision_id):
152
 
                real_history.append(revid)
153
 
        except errors.RevisionNotPresent:
154
 
            pass # Hit a ghost left hand parent
 
140
        real_history = list(repo.iter_reverse_revision_history(
 
141
                                last_revision_id))
155
142
        real_history.reverse()
156
143
        if last_revno != len(real_history):
157
144
            self.fixed_history = True
159
146
            # set_revision_history, as this will regenerate it again.
160
147
            # Not really worth a whole BranchReconciler class just for this,
161
148
            # though.
162
 
            ui.ui_factory.note('Fixing last revision info %s => %s' % (
163
 
                 last_revno, len(real_history)))
 
149
            self.pb.note('Fixing last revision info %s => %s',
 
150
                         last_revno, len(real_history))
164
151
            self.branch.set_last_revision_info(len(real_history),
165
152
                                               last_revision_id)
166
153
        else:
167
154
            self.fixed_history = False
168
 
            ui.ui_factory.note('revision_history ok.')
 
155
            self.pb.note('revision_history ok.')
169
156
 
170
157
 
171
158
class RepoReconciler(object):
172
159
    """Reconciler that reconciles a repository.
173
160
 
174
161
    The goal of repository reconciliation is to make any derived data
175
 
    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 
176
163
    reindexing, or removing unreferenced data if that can interfere with
177
164
    queries in a given repository.
178
165
 
194
181
 
195
182
    def reconcile(self):
196
183
        """Perform reconciliation.
197
 
 
 
184
        
198
185
        After reconciliation the following attributes document found issues:
199
186
        inconsistent_parents: The number of revisions in the repository whose
200
187
                              ancestry was being reported incorrectly.
201
188
        garbage_inventories: The number of inventory objects without revisions
202
189
                             that were garbage collected.
203
190
        """
204
 
        operation = cleanup.OperationWithCleanups(self._reconcile)
205
 
        self.add_cleanup = operation.add_cleanup
206
 
        operation.run_simple()
207
 
 
208
 
    def _reconcile(self):
209
191
        self.repo.lock_write()
210
 
        self.add_cleanup(self.repo.unlock)
211
 
        self.pb = ui.ui_factory.nested_progress_bar()
212
 
        self.add_cleanup(self.pb.finished)
213
 
        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()
214
200
 
215
201
    def _reconcile_steps(self):
216
202
        """Perform the steps to reconcile this repository."""
218
204
 
219
205
    def _reweave_inventory(self):
220
206
        """Regenerate the inventory weave for the repository from scratch.
221
 
 
222
 
        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 
223
209
        will correct data issues. The self.thorough flag controls whether
224
210
        only data-loss causing issues (!self.thorough) or all issues
225
211
        (self.thorough) are treated as requiring the reweave.
226
212
        """
 
213
        # local because needing to know about WeaveFile is a wart we want to hide
 
214
        from bzrlib.weave import WeaveFile, Weave
227
215
        transaction = self.repo.get_transaction()
228
 
        self.pb.update('Reading inventory data')
 
216
        self.pb.update('Reading inventory data.')
229
217
        self.inventory = self.repo.inventories
230
218
        self.revisions = self.repo.revisions
231
219
        # the total set of revisions to process
241
229
            # put a revision into the graph.
242
230
            self._graph_revision(rev_id)
243
231
        self._check_garbage_inventories()
244
 
        # if there are no inconsistent_parents and
 
232
        # if there are no inconsistent_parents and 
245
233
        # (no garbage inventories or we are not doing a thorough check)
246
 
        if (not self.inconsistent_parents and
 
234
        if (not self.inconsistent_parents and 
247
235
            (not self.garbage_inventories or not self.thorough)):
248
 
            ui.ui_factory.note('Inventory ok.')
 
236
            self.pb.note('Inventory ok.')
249
237
            return
250
 
        self.pb.update('Backing up inventory', 0, 0)
 
238
        self.pb.update('Backing up inventory...', 0, 0)
251
239
        self.repo._backup_inventory()
252
 
        ui.ui_factory.note('Backup inventory created.')
 
240
        self.pb.note('Backup Inventory created.')
253
241
        new_inventories = self.repo._temp_inventories()
254
242
 
255
243
        # we have topological order of revisions and non ghost parents ready.
256
244
        self._setup_steps(len(self._rev_graph))
257
 
        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()]
258
247
        stream = self._change_inv_parents(
259
 
            self.inventory.get_record_stream(revision_keys, 'unordered', True),
 
248
            self.inventory.get_record_stream(revision_keys, 'unsorted', True),
260
249
            self._new_inv_parents,
261
250
            set(revision_keys))
262
251
        new_inventories.insert_record_stream(stream)
268
257
        self.pb.update('Writing weave')
269
258
        self.repo._activate_new_inventory()
270
259
        self.inventory = None
271
 
        ui.ui_factory.note('Inventory regenerated.')
 
260
        self.pb.note('Inventory regenerated.')
272
261
 
273
262
    def _new_inv_parents(self, revision_key):
274
263
        """Lookup ghost-filtered parents for revision_key."""
362
351
    def _load_indexes(self):
363
352
        """Load indexes for the reconciliation."""
364
353
        self.transaction = self.repo.get_transaction()
365
 
        self.pb.update('Reading indexes', 0, 2)
 
354
        self.pb.update('Reading indexes.', 0, 2)
366
355
        self.inventory = self.repo.inventories
367
 
        self.pb.update('Reading indexes', 1, 2)
 
356
        self.pb.update('Reading indexes.', 1, 2)
368
357
        self.repo._check_for_inconsistent_revision_parents()
369
358
        self.revisions = self.repo.revisions
370
 
        self.pb.update('Reading indexes', 2, 2)
 
359
        self.pb.update('Reading indexes.', 2, 2)
371
360
 
372
361
    def _gc_inventory(self):
373
362
        """Remove inventories that are not referenced from the revision store."""
374
 
        self.pb.update('Checking unused inventories', 0, 1)
 
363
        self.pb.update('Checking unused inventories.', 0, 1)
375
364
        self._check_garbage_inventories()
376
 
        self.pb.update('Checking unused inventories', 1, 3)
 
365
        self.pb.update('Checking unused inventories.', 1, 3)
377
366
        if not self.garbage_inventories:
378
 
            ui.ui_factory.note('Inventory ok.')
 
367
            self.pb.note('Inventory ok.')
379
368
            return
380
 
        self.pb.update('Backing up inventory', 0, 0)
 
369
        self.pb.update('Backing up inventory...', 0, 0)
381
370
        self.repo._backup_inventory()
382
 
        ui.ui_factory.note('Backup Inventory created')
 
371
        self.pb.note('Backup Inventory created.')
383
372
        # asking for '' should never return a non-empty weave
384
373
        new_inventories = self.repo._temp_inventories()
385
374
        # we have topological order of revisions and non ghost parents ready.
386
375
        graph = self.revisions.get_parent_map(self.revisions.keys())
387
 
        revision_keys = topo_sort(graph)
 
376
        revision_keys = list(TopoSorter(graph).iter_topo_order())
388
377
        revision_ids = [key[-1] for key in revision_keys]
389
378
        self._setup_steps(len(revision_keys))
390
379
        stream = self._change_inv_parents(
391
 
            self.inventory.get_record_stream(revision_keys, 'unordered', True),
 
380
            self.inventory.get_record_stream(revision_keys, 'unsorted', True),
392
381
            graph.__getitem__,
393
382
            set(revision_keys))
394
383
        new_inventories.insert_record_stream(stream)
399
388
        self.pb.update('Writing weave')
400
389
        self.repo._activate_new_inventory()
401
390
        self.inventory = None
402
 
        ui.ui_factory.note('Inventory regenerated.')
 
391
        self.pb.note('Inventory regenerated.')
403
392
 
404
393
    def _fix_text_parents(self):
405
394
        """Fix bad versionedfile parent entries.
501
490
    #  - lock the names list
502
491
    #  - perform a customised pack() that regenerates data as needed
503
492
    #  - unlock the names list
504
 
    # https://bugs.launchpad.net/bzr/+bug/154173
505
 
 
506
 
    def __init__(self, repo, other=None, thorough=False,
507
 
            canonicalize_chks=False):
508
 
        super(PackReconciler, self).__init__(repo, other=other,
509
 
            thorough=thorough)
510
 
        self.canonicalize_chks = canonicalize_chks
 
493
    # https://bugs.edge.launchpad.net/bzr/+bug/154173
511
494
 
512
495
    def _reconcile_steps(self):
513
496
        """Perform the steps to reconcile this repository."""
516
499
        collection = self.repo._pack_collection
517
500
        collection.ensure_loaded()
518
501
        collection.lock_names()
519
 
        self.add_cleanup(collection._unlock_names)
520
 
        packs = collection.all_packs()
521
 
        all_revisions = self.repo.all_revision_ids()
522
 
        total_inventories = len(list(
523
 
            collection.inventory_index.combined_index.iter_all_entries()))
524
 
        if len(all_revisions):
525
 
            if self.canonicalize_chks:
526
 
                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)
527
513
            else:
528
 
                reconcile_meth = self.repo._reconcile_pack
529
 
            new_pack = reconcile_meth(collection, packs, ".reconcile",
530
 
                all_revisions, self.pb)
531
 
            if new_pack is not None:
 
514
                # only make a new pack when there is data to copy.
532
515
                self._discard_and_save(packs)
533
 
        else:
534
 
            # only make a new pack when there is data to copy.
535
 
            self._discard_and_save(packs)
536
 
        self.garbage_inventories = total_inventories - len(list(
537
 
            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()
538
520
 
539
521
    def _discard_and_save(self, packs):
540
522
        """Discard some packs from the repository.