236
234
self.pb.note('Inventory ok.')
238
236
self.pb.update('Backing up inventory...', 0, 0)
239
self.repo._backup_inventory()
237
self.repo.control_weaves.copy(self.inventory, 'inventory.backup', self.repo.get_transaction())
240
238
self.pb.note('Backup Inventory created.')
241
new_inventories = self.repo._temp_inventories()
239
# asking for '' should never return a non-empty weave
240
new_inventory_vf = self.repo.control_weaves.get_empty('inventory.new',
241
self.repo.get_transaction())
243
243
# we have topological order of revisions and non ghost parents ready.
244
244
self._setup_steps(len(self._rev_graph))
245
revision_keys = [(rev_id,) for rev_id in
246
TopoSorter(self._rev_graph.items()).iter_topo_order()]
247
stream = self._change_inv_parents(
248
self.inventory.get_record_stream(revision_keys, 'unordered', True),
249
self._new_inv_parents,
251
new_inventories.insert_record_stream(stream)
252
# if this worked, the set of new_inventories.keys should equal
245
for rev_id in TopoSorter(self._rev_graph.items()).iter_topo_order():
246
parents = self._rev_graph[rev_id]
247
# double check this really is in topological order.
248
unavailable = [p for p in parents if p not in new_inventory_vf]
250
raise AssertionError('unavailable parents: %r'
252
# this entry has all the non ghost parents in the inventory
254
self._reweave_step('adding inventories')
255
if isinstance(new_inventory_vf, WeaveFile):
256
# It's really a WeaveFile, but we call straight into the
257
# Weave's add method to disable the auto-write-out behaviour.
258
# This is done to avoid a revision_count * time-to-write additional overhead on
260
new_inventory_vf._check_write_ok()
261
Weave._add_lines(new_inventory_vf, rev_id, parents,
262
self.inventory.get_lines(rev_id), None, None, None, False, True)
264
new_inventory_vf.add_lines(rev_id, parents, self.inventory.get_lines(rev_id))
266
if isinstance(new_inventory_vf, WeaveFile):
267
new_inventory_vf._save()
268
# if this worked, the set of new_inventory_vf.names should equal
254
if not (set(new_inventories.keys()) ==
255
set([(revid,) for revid in self.pending])):
270
if not (set(new_inventory_vf.versions()) == self.pending):
256
271
raise AssertionError()
257
272
self.pb.update('Writing weave')
258
self.repo._activate_new_inventory()
273
self.repo.control_weaves.copy(new_inventory_vf, 'inventory', self.repo.get_transaction())
274
self.repo.control_weaves.delete('inventory.new', self.repo.get_transaction())
259
275
self.inventory = None
260
276
self.pb.note('Inventory regenerated.')
262
def _new_inv_parents(self, revision_key):
263
"""Lookup ghost-filtered parents for revision_key."""
264
# Use the filtered ghostless parents list:
265
return tuple([(revid,) for revid in self._rev_graph[revision_key[-1]]])
267
def _change_inv_parents(self, stream, get_parents, all_revision_keys):
268
"""Adapt a record stream to reconcile the parents."""
269
for record in stream:
270
wanted_parents = get_parents(record.key)
271
if wanted_parents and wanted_parents[0] not in all_revision_keys:
272
# The check for the left most parent only handles knit
273
# compressors, but this code only applies to knit and weave
274
# repositories anyway.
275
bytes = record.get_bytes_as('fulltext')
276
yield FulltextContentFactory(record.key, wanted_parents, record.sha1, bytes)
278
adapted_record = AdapterFactory(record.key, wanted_parents, record)
280
self._reweave_step('adding inventories')
282
278
def _setup_steps(self, new_total):
283
279
"""Setup the markers we need to control the progress bar."""
284
280
self.total = new_total
298
294
mutter('found ghost %s', parent)
299
295
self._rev_graph[rev_id] = parents
296
if self._parents_are_inconsistent(rev_id, parents):
297
self.inconsistent_parents += 1
298
mutter('Inconsistent inventory parents: id {%s} '
299
'inventory claims %r, '
300
'available parents are %r, '
301
'unavailable parents are %r',
303
set(self.inventory.get_parent_map([rev_id])[rev_id]),
305
set(rev.parent_ids).difference(set(parents)))
307
def _parents_are_inconsistent(self, rev_id, parents):
308
"""Return True if the parents list of rev_id does not match the weave.
310
This detects inconsistencies based on the self.thorough value:
311
if thorough is on, the first parent value is checked as well as ghost
313
Otherwise only the ghost differences are evaluated.
315
weave_parents = self.inventory.get_parent_map([rev_id])[rev_id]
316
weave_missing_old_ghosts = set(weave_parents) != set(parents)
317
first_parent_is_wrong = (
318
len(weave_parents) and len(parents) and
319
parents[0] != weave_parents[0])
321
return weave_missing_old_ghosts or first_parent_is_wrong
323
return weave_missing_old_ghosts
301
325
def _check_garbage_inventories(self):
302
326
"""Check for garbage inventories which we cannot trust
367
389
self.pb.note('Inventory ok.')
369
391
self.pb.update('Backing up inventory...', 0, 0)
370
self.repo._backup_inventory()
392
self.repo.control_weaves.copy(self.inventory, 'inventory.backup', self.transaction)
371
393
self.pb.note('Backup Inventory created.')
372
394
# asking for '' should never return a non-empty weave
373
new_inventories = self.repo._temp_inventories()
395
new_inventory_vf = self.repo.control_weaves.get_empty('inventory.new',
374
398
# we have topological order of revisions and non ghost parents ready.
375
graph = self.revisions.get_parent_map(self.revisions.keys())
376
revision_keys = list(TopoSorter(graph).iter_topo_order())
377
revision_ids = [key[-1] for key in revision_keys]
378
self._setup_steps(len(revision_keys))
379
stream = self._change_inv_parents(
380
self.inventory.get_record_stream(revision_keys, 'unordered', True),
383
new_inventories.insert_record_stream(stream)
399
self._setup_steps(len(self.revisions))
400
revision_ids = self.revisions.versions()
401
graph = self.revisions.get_parent_map(revision_ids)
402
for rev_id in TopoSorter(graph.items()).iter_topo_order():
403
parents = graph[rev_id]
404
# double check this really is in topological order, ignoring existing ghosts.
405
unavailable = [p for p in parents if p not in new_inventory_vf and
408
raise AssertionError(
409
'unavailable parents: %r' % (unavailable,))
410
# this entry has all the non ghost parents in the inventory
412
self._reweave_step('adding inventories')
413
# ugly but needed, weaves are just way tooooo slow else.
414
new_inventory_vf.add_lines_with_ghosts(rev_id, parents,
415
self.inventory.get_lines(rev_id))
384
417
# if this worked, the set of new_inventory_vf.names should equal
385
# the revisionds list
386
if not(set(new_inventories.keys()) == set(revision_keys)):
419
if not(set(new_inventory_vf.versions()) == set(self.revisions.versions())):
387
420
raise AssertionError()
388
421
self.pb.update('Writing weave')
389
self.repo._activate_new_inventory()
422
self.repo.control_weaves.copy(new_inventory_vf, 'inventory', self.transaction)
423
self.repo.control_weaves.delete('inventory.new', self.transaction)
390
424
self.inventory = None
391
425
self.pb.note('Inventory regenerated.')
427
def _check_garbage_inventories(self):
428
"""Check for garbage inventories which we cannot trust
430
We cant trust them because their pre-requisite file data may not
431
be present - all we know is that their revision was not installed.
433
inventories = set(self.inventory.versions())
434
revisions = set(self.revisions.versions())
435
garbage = inventories.difference(revisions)
436
self.garbage_inventories = len(garbage)
437
for revision_id in garbage:
438
mutter('Garbage inventory {%s} found.', revision_id)
393
440
def _fix_text_parents(self):
394
441
"""Fix bad versionedfile parent entries.
400
447
parent lists, and replaces the versionedfile with a corrected version.
402
449
transaction = self.repo.get_transaction()
403
versions = [key[-1] for key in self.revisions.keys()]
450
versions = self.revisions.versions()
404
451
mutter('Prepopulating revision text cache with %d revisions',
406
453
vf_checker = self.repo._get_versioned_file_checker()
407
bad_parents, unused_versions = vf_checker.check_file_version_parents(
408
self.repo.texts, self.pb)
409
text_index = vf_checker.text_index
410
per_id_bad_parents = {}
411
for key in unused_versions:
412
# Ensure that every file with unused versions gets rewritten.
413
# NB: This is really not needed, reconcile != pack.
414
per_id_bad_parents[key[0]] = {}
415
# Generate per-knit/weave data.
416
for key, details in bad_parents.iteritems():
419
knit_parents = tuple([parent[-1] for parent in details[0]])
420
correct_parents = tuple([parent[-1] for parent in details[1]])
421
file_details = per_id_bad_parents.setdefault(file_id, {})
422
file_details[rev_id] = (knit_parents, correct_parents)
423
file_id_versions = {}
424
for text_key in text_index:
425
versions_list = file_id_versions.setdefault(text_key[0], [])
426
versions_list.append(text_key[1])
427
# Do the reconcile of individual weaves.
428
for num, file_id in enumerate(per_id_bad_parents):
454
# List all weaves before altering, to avoid race conditions when we
455
# delete unused weaves.
456
weaves = list(enumerate(self.repo.weave_store))
457
for num, file_id in weaves:
429
458
self.pb.update('Fixing text parents', num,
430
len(per_id_bad_parents))
431
versions_with_bad_parents = per_id_bad_parents[file_id]
432
id_unused_versions = set(key[-1] for key in unused_versions
433
if key[0] == file_id)
434
if file_id in file_id_versions:
435
file_versions = file_id_versions[file_id]
437
# This id was present in the disk store but is not referenced
438
# by any revision at all.
440
self._fix_text_parent(file_id, versions_with_bad_parents,
441
id_unused_versions, file_versions)
459
len(self.repo.weave_store))
460
vf = self.repo.weave_store.get_weave(file_id, transaction)
461
versions_with_bad_parents, unused_versions = \
462
vf_checker.check_file_version_parents(vf, file_id)
463
if (len(versions_with_bad_parents) == 0 and
464
len(unused_versions) == 0):
466
full_text_versions = set()
467
self._fix_text_parent(file_id, vf, versions_with_bad_parents,
468
full_text_versions, unused_versions)
443
def _fix_text_parent(self, file_id, versions_with_bad_parents,
444
unused_versions, all_versions):
470
def _fix_text_parent(self, file_id, vf, versions_with_bad_parents,
471
full_text_versions, unused_versions):
445
472
"""Fix bad versionedfile entries in a single versioned file."""
446
473
mutter('fixing text parent: %r (%d versions)', file_id,
447
474
len(versions_with_bad_parents))
448
mutter('(%d are unused)', len(unused_versions))
449
new_file_id = 'temp:%s' % file_id
475
mutter('(%d need to be full texts, %d are unused)',
476
len(full_text_versions), len(unused_versions))
477
new_vf = self.repo.weave_store.get_empty('temp:%s' % file_id,
452
for version in all_versions:
480
for version in vf.versions():
453
481
if version in unused_versions:
455
483
elif version in versions_with_bad_parents:
456
484
parents = versions_with_bad_parents[version][1]
458
pmap = self.repo.texts.get_parent_map([(file_id, version)])
459
parents = [key[-1] for key in pmap[(file_id, version)]]
460
new_parents[(new_file_id, version)] = [
461
(new_file_id, parent) for parent in parents]
462
needed_keys.add((file_id, version))
463
def fix_parents(stream):
464
for record in stream:
465
bytes = record.get_bytes_as('fulltext')
466
new_key = (new_file_id, record.key[-1])
467
parents = new_parents[new_key]
468
yield FulltextContentFactory(new_key, parents, record.sha1, bytes)
469
stream = self.repo.texts.get_record_stream(needed_keys, 'topological', True)
470
self.repo._remove_file_id(new_file_id)
471
self.repo.texts.insert_record_stream(fix_parents(stream))
472
self.repo._remove_file_id(file_id)
474
self.repo._move_file_id(new_file_id, file_id)
486
parents = vf.get_parent_map([version])[version]
487
new_parents[version] = parents
488
if not len(new_parents):
489
# No used versions, remove the VF.
490
self.repo.weave_store.delete(file_id, self.transaction)
492
for version in TopoSorter(new_parents.items()).iter_topo_order():
493
lines = vf.get_lines(version)
494
parents = new_parents[version]
495
if parents and (parents[0] in full_text_versions):
496
# Force this record to be a fulltext, not a delta.
497
new_vf._add(version, lines, parents, False,
498
None, None, None, False)
500
new_vf.add_lines(version, parents, lines)
501
self.repo.weave_store.copy(new_vf, file_id, self.transaction)
502
self.repo.weave_store.delete('temp:%s' % file_id, self.transaction)
477
505
class PackReconciler(RepoReconciler):