32
32
# raising them. If there's more than one exception it'd be good to see them
35
"""Checking of bzr objects.
37
check_refs is a concept used for optimising check. Objects that depend on other
38
objects (e.g. tree on repository) can list the objects they would be requesting
39
so that when the dependent object is checked, matches can be pulled out and
40
evaluated in-line rather than re-reading the same data many times.
41
check_refs are tuples (kind, value). Currently defined kinds are:
42
* 'trees', where value is a revid and the looked up objects are revision trees.
43
* 'lefthand-distance', where value is a revid and the looked up objects are the
44
distance along the lefthand path to NULL for that revid.
45
* 'revision-existence', where value is a revid, and the result is True or False
46
indicating that the revision was found/not found.
49
from bzrlib import errors, osutils
35
from bzrlib import errors
50
36
from bzrlib import repository as _mod_repository
51
37
from bzrlib import revision
52
from bzrlib.branch import Branch
53
from bzrlib.bzrdir import BzrDir
54
38
from bzrlib.errors import BzrCheckError
55
from bzrlib.repository import Repository
56
from bzrlib.revision import NULL_REVISION
57
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
58
from bzrlib.trace import note
60
from bzrlib.workingtree import WorkingTree
40
from bzrlib.trace import log_error, note
62
42
class Check(object):
63
43
"""Check a repository"""
65
45
# The Check object interacts with InventoryEntry.check, etc.
67
def __init__(self, repository, check_repo=True):
47
def __init__(self, repository):
68
48
self.repository = repository
49
self.checked_text_cnt = 0
69
50
self.checked_rev_cnt = 0
52
self.repeated_text_cnt = 0
71
53
self.missing_parent_links = {}
72
54
self.missing_inventory_sha_cnt = 0
73
55
self.missing_revision_cnt = 0
74
self.checked_weaves = set()
56
# maps (file-id, version) -> sha1; used by InventoryFile._check
57
self.checked_texts = {}
58
self.checked_weaves = {}
75
59
self.unreferenced_versions = set()
76
60
self.inconsistent_parents = []
77
self.rich_roots = repository.supports_rich_root()
78
self.text_key_references = {}
79
self.check_repo = check_repo
80
self.other_results = []
81
# Plain text lines to include in the report
82
self._report_items = []
83
# Keys we are looking for; may be large and need spilling to disk.
84
# key->(type(revision/inventory/text/signature/map), sha1, first-referer)
85
self.pending_keys = {}
86
# Ancestors map for all of revisions being checked; while large helper
87
# functions we call would create it anyway, so better to have once and
91
def check(self, callback_refs=None, check_repo=True):
92
if callback_refs is None:
94
63
self.repository.lock_read()
95
64
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
97
self.progress.update('check', 0, 4)
99
self.progress.update('checking revisions', 0)
100
self.check_revisions()
101
self.progress.update('checking commit contents', 1)
102
self.repository._check_inventories(self)
103
self.progress.update('checking file graphs', 2)
104
# check_weaves is done after the revision scan so that
105
# revision index is known to be valid.
107
self.progress.update('checking branches and trees', 3)
109
repo = self.repository
110
# calculate all refs, and callback the objects requesting them.
112
wanting_items = set()
113
# Current crude version calculates everything and calls
114
# everything at once. Doing a queue and popping as things are
115
# satisfied would be cheaper on memory [but few people have
116
# huge numbers of working trees today. TODO: fix before
120
for ref, wantlist in callback_refs.iteritems():
121
wanting_items.update(wantlist)
124
refs[ref] = repo.revision_tree(value)
125
elif kind == 'lefthand-distance':
127
elif kind == 'revision-existence':
128
existences.add(value)
130
raise AssertionError(
131
'unknown ref kind for ref %s' % ref)
132
node_distances = repo.get_graph().find_lefthand_distances(distances)
133
for key, distance in node_distances.iteritems():
134
refs[('lefthand-distance', key)] = distance
135
if key in existences and distance > 0:
136
refs[('revision-existence', key)] = True
137
existences.remove(key)
138
parent_map = repo.get_graph().get_parent_map(existences)
139
for key in parent_map:
140
refs[('revision-existence', key)] = True
141
existences.remove(key)
142
for key in existences:
143
refs[('revision-existence', key)] = False
144
for item in wanting_items:
145
if isinstance(item, WorkingTree):
147
if isinstance(item, Branch):
148
self.other_results.append(item.check(refs))
66
self.progress.update('retrieving inventory', 0, 2)
67
# do not put in init, as it should be done with progess,
68
# and inside the lock.
69
self.inventory_weave = self.repository.get_inventory_weave()
70
self.progress.update('checking revision graph', 1)
71
self.check_revision_graph()
74
while revno < len(self.planned_revisions):
75
rev_id = self.planned_revisions[revno]
76
self.progress.update('checking revision', revno,
77
len(self.planned_revisions))
79
self.check_one_rev(rev_id)
80
# check_weaves is done after the revision scan so that
81
# revision index is known to be valid.
150
84
self.progress.finished()
151
85
self.repository.unlock()
153
def _check_revisions(self, revisions_iterator):
154
"""Check revision objects by decorating a generator.
156
:param revisions_iterator: An iterator of(revid, Revision-or-None).
157
:return: A generator of the contents of revisions_iterator.
159
self.planned_revisions = set()
160
for revid, revision in revisions_iterator:
161
yield revid, revision
162
self._check_one_rev(revid, revision)
163
# Flatten the revisions we found to guarantee consistent later
165
self.planned_revisions = list(self.planned_revisions)
166
# TODO: extract digital signatures as items to callback on too.
168
def check_revisions(self):
169
"""Scan revisions, checking data directly available as we go."""
170
revision_iterator = self.repository._iter_revisions(None)
171
revision_iterator = self._check_revisions(revision_iterator)
172
# We read the all revisions here:
173
# - doing this allows later code to depend on the revision index.
174
# - we can fill out existence flags at this point
175
# - we can read the revision inventory sha at this point
176
# - we can check properties and serialisers etc.
87
def check_revision_graph(self):
177
88
if not self.repository.revision_graph_can_have_wrong_parents():
178
# The check against the index isn't needed.
89
# This check is not necessary.
179
90
self.revs_with_bad_parents_in_index = None
180
for thing in revision_iterator:
183
bad_revisions = self.repository._find_inconsistent_revision_parents(
185
self.revs_with_bad_parents_in_index = list(bad_revisions)
92
bad_revisions = self.repository._find_inconsistent_revision_parents()
93
self.revs_with_bad_parents_in_index = list(bad_revisions)
95
def plan_revisions(self):
96
repository = self.repository
97
self.planned_revisions = repository.all_revision_ids()
99
inventoried = set(self.inventory_weave.versions())
100
awol = set(self.planned_revisions) - inventoried
102
raise BzrCheckError('Stored revisions missing from inventory'
103
'{%s}' % ','.join([f for f in awol]))
187
105
def report_results(self, verbose):
189
self._report_repo_results(verbose)
190
for result in self.other_results:
191
result.report_results(verbose)
193
def _report_repo_results(self, verbose):
194
106
note('checked repository %s format %s',
195
107
self.repository.bzrdir.root_transport,
196
108
self.repository._format)
197
109
note('%6d revisions', self.checked_rev_cnt)
198
110
note('%6d file-ids', len(self.checked_weaves))
200
note('%6d unreferenced text versions',
201
len(self.unreferenced_versions))
202
if verbose and len(self.unreferenced_versions):
203
for file_id, revision_id in self.unreferenced_versions:
204
note('unreferenced version: {%s} in %s', revision_id,
111
note('%6d unique file texts', self.checked_text_cnt)
112
note('%6d repeated file texts', self.repeated_text_cnt)
113
note('%6d unreferenced text versions',
114
len(self.unreferenced_versions))
206
115
if self.missing_inventory_sha_cnt:
207
116
note('%6d revisions are missing inventory_sha1',
208
117
self.missing_inventory_sha_cnt)
241
154
' %s has wrong parents in index: '
242
155
'%r should be %r',
243
156
revision_id, index_parents, actual_parents)
244
for item in self._report_items:
247
def _check_one_rev(self, rev_id, rev):
248
"""Cross-check one revision.
250
:param rev_id: A revision id to check.
251
:param rev: A revision or None to indicate a missing revision.
158
def check_one_rev(self, rev_id):
159
"""Check one revision.
161
rev_id - the one to check
163
rev = self.repository.get_revision(rev_id)
253
165
if rev.revision_id != rev_id:
254
self._report_items.append(
255
'Mismatched internal revid {%s} and index revid {%s}' % (
256
rev.revision_id, rev_id))
257
rev_id = rev.revision_id
258
# Check this revision tree etc, and count as seen when we encounter a
260
self.planned_revisions.add(rev_id)
262
self.ghosts.discard(rev_id)
263
# Count all parents as ghosts if we haven't seen them yet.
166
raise BzrCheckError('wrong internal revision id in revision {%s}'
264
169
for parent in rev.parent_ids:
265
170
if not parent in self.planned_revisions:
266
self.ghosts.add(parent)
268
self.ancestors[rev_id] = tuple(rev.parent_ids) or (NULL_REVISION,)
269
self.add_pending_item(rev_id, ('inventories', rev_id), 'inventory',
171
missing_links = self.missing_parent_links.get(parent, [])
172
missing_links.append(rev_id)
173
self.missing_parent_links[parent] = missing_links
174
# list based so somewhat slow,
175
# TODO have a planned_revisions list and set.
176
if self.repository.has_revision(parent):
177
missing_ancestry = self.repository.get_ancestry(parent)
178
for missing in missing_ancestry:
179
if (missing is not None
180
and missing not in self.planned_revisions):
181
self.planned_revisions.append(missing)
183
self.ghosts.append(rev_id)
185
if rev.inventory_sha1:
186
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
187
if inv_sha1 != rev.inventory_sha1:
188
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
189
' value in revision {%s}' % rev_id)
190
self._check_revision_tree(rev_id)
271
191
self.checked_rev_cnt += 1
273
def add_pending_item(self, referer, key, kind, sha1):
274
"""Add a reference to a sha1 to be cross checked against a key.
276
:param referer: The referer that expects key to have sha1.
277
:param key: A storage key e.g. ('texts', 'foo@bar-20040504-1234')
278
:param kind: revision/inventory/text/map/signature
279
:param sha1: A hex sha1 or None if no sha1 is known.
281
existing = self.pending_keys.get(key)
283
if sha1 != existing[1]:
284
self._report_items.append('Multiple expected sha1s for %s. {%s}'
285
' expects {%s}, {%s} expects {%s}', (
286
key, referer, sha1, existing[1], existing[0]))
288
self.pending_keys[key] = (kind, sha1, referer)
290
193
def check_weaves(self):
291
194
"""Check all the weaves we can get our hands on.
294
storebar = bzrlib.ui.ui_factory.nested_progress_bar()
296
self._check_weaves(storebar)
300
def _check_weaves(self, storebar):
301
storebar.update('text-index', 0, 2)
302
if self.repository._format.fast_deltas:
303
# We haven't considered every fileid instance so far.
304
weave_checker = self.repository._get_versioned_file_checker(
305
ancestors=self.ancestors)
307
weave_checker = self.repository._get_versioned_file_checker(
308
text_key_references=self.text_key_references,
309
ancestors=self.ancestors)
310
storebar.update('file-graph', 1)
311
result = weave_checker.check_file_version_parents(
312
self.repository.texts)
313
self.checked_weaves = weave_checker.file_ids
314
bad_parents, unused_versions = result
315
bad_parents = bad_parents.items()
316
for text_key, (stored_parents, correct_parents) in bad_parents:
317
# XXX not ready for id join/split operations.
318
weave_id = text_key[0]
319
revision_id = text_key[-1]
320
weave_parents = tuple([parent[-1] for parent in stored_parents])
321
correct_parents = tuple([parent[-1] for parent in correct_parents])
322
self.inconsistent_parents.append(
323
(revision_id, weave_id, weave_parents, correct_parents))
324
self.unreferenced_versions.update(unused_versions)
326
def _add_entry_to_text_key_references(self, inv, entry):
327
if not self.rich_roots and entry.name == '':
329
key = (entry.file_id, entry.revision)
330
self.text_key_references.setdefault(key, False)
331
if entry.revision == inv.revision_id:
332
self.text_key_references[key] = True
335
@deprecated_function(deprecated_in((1,6,0)))
198
if self.repository.weave_store.listable():
199
weave_ids = list(self.repository.weave_store)
200
n_weaves = len(weave_ids) + 1
201
self.progress.update('checking versionedfile', 0, n_weaves)
202
self.inventory_weave.check(progress_bar=self.progress)
203
files_in_revisions = {}
204
revisions_of_files = {}
205
weave_checker = self.repository._get_versioned_file_checker()
206
for i, weave_id in enumerate(weave_ids):
207
self.progress.update('checking versionedfile', i, n_weaves)
208
w = self.repository.weave_store.get_weave(weave_id,
209
self.repository.get_transaction())
210
# No progress here, because it looks ugly.
212
result = weave_checker.check_file_version_parents(w, weave_id)
213
bad_parents, unused_versions = result
214
bad_parents = bad_parents.items()
215
for revision_id, (weave_parents, correct_parents) in bad_parents:
216
self.inconsistent_parents.append(
217
(revision_id, weave_id, weave_parents, correct_parents))
218
for revision_id in unused_versions:
219
self.unreferenced_versions.add((weave_id, revision_id))
220
self.checked_weaves[weave_id] = True
222
def _check_revision_tree(self, rev_id):
223
tree = self.repository.revision_tree(rev_id)
227
if file_id in seen_ids:
228
raise BzrCheckError('duplicated file_id {%s} '
229
'in inventory for revision {%s}'
231
seen_ids[file_id] = True
234
ie.check(self, rev_id, inv, tree)
236
for path, ie in inv.iter_entries():
237
if path in seen_names:
238
raise BzrCheckError('duplicated path %s '
239
'in inventory for revision {%s}'
241
seen_names[path] = True
336
244
def check(branch, verbose):
337
245
"""Run consistency checks on a branch.
339
Results are reported through logging.
341
Deprecated in 1.6. Please use check_dwim instead.
343
:raise BzrCheckError: if there's a consistency error.
345
check_branch(branch, verbose)
348
@deprecated_function(deprecated_in((1,16,0)))
349
def check_branch(branch, verbose):
350
"""Run consistency checks on a branch.
352
Results are reported through logging.
247
Results are reported through logging.
354
249
:raise BzrCheckError: if there's a consistency error.
356
251
branch.lock_read()
359
for ref in branch._get_check_refs():
360
needed_refs.setdefault(ref, []).append(branch)
361
result = branch.repository.check([branch.last_revision()], needed_refs)
362
branch_result = result.other_results[0]
253
branch_result = branch.check()
254
repo_result = branch.repository.check([branch.last_revision()])
365
257
branch_result.report_results(verbose)
368
def scan_branch(branch, needed_refs, to_unlock):
369
"""Scan a branch for refs.
371
:param branch: The branch to schedule for checking.
372
:param needed_refs: Refs we are accumulating.
373
:param to_unlock: The unlock list accumulating.
375
note("Checking branch at '%s'." % (branch.base,))
377
to_unlock.append(branch)
378
branch_refs = branch._get_check_refs()
379
for ref in branch_refs:
380
reflist = needed_refs.setdefault(ref, [])
381
reflist.append(branch)
384
def scan_tree(base_tree, tree, needed_refs, to_unlock):
385
"""Scan a tree for refs.
387
:param base_tree: The original tree check opened, used to detect duplicate
389
:param tree: The tree to schedule for checking.
390
:param needed_refs: Refs we are accumulating.
391
:param to_unlock: The unlock list accumulating.
393
if base_tree is not None and tree.basedir == base_tree.basedir:
395
note("Checking working tree at '%s'." % (tree.basedir,))
397
to_unlock.append(tree)
398
tree_refs = tree._get_check_refs()
399
for ref in tree_refs:
400
reflist = needed_refs.setdefault(ref, [])
404
def check_dwim(path, verbose, do_branch=False, do_repo=False, do_tree=False):
405
"""Check multiple objects.
407
If errors occur they are accumulated and reported as far as possible, and
408
an exception raised at the end of the process.
411
base_tree, branch, repo, relpath = \
412
BzrDir.open_containing_tree_branch_or_repository(path)
413
except errors.NotBranchError:
414
base_tree = branch = repo = None
419
if base_tree is not None:
420
# If the tree is a lightweight checkout we won't see it in
421
# repo.find_branches - add now.
423
scan_tree(None, base_tree, needed_refs, to_unlock)
424
branch = base_tree.branch
425
if branch is not None:
428
# The branch is in a shared repository
429
repo = branch.repository
432
to_unlock.append(repo)
433
branches = repo.find_branches(using=True)
435
if do_branch or do_tree:
436
for branch in branches:
439
tree = branch.bzrdir.open_workingtree()
441
except (errors.NotLocalUrl, errors.NoWorkingTree):
444
scan_tree(base_tree, tree, needed_refs, to_unlock)
446
scan_branch(branch, needed_refs, to_unlock)
447
if do_branch and not branches:
448
note("No branch found at specified location.")
449
if do_tree and base_tree is None and not saw_tree:
450
note("No working tree found at specified location.")
451
if do_repo or do_branch or do_tree:
453
note("Checking repository at '%s'."
454
% (repo.bzrdir.root_transport.base,))
455
result = repo.check(None, callback_refs=needed_refs,
457
result.report_results(verbose)
460
note("No working tree found at specified location.")
462
note("No branch found at specified location.")
464
note("No repository found at specified location.")
466
for thing in to_unlock:
258
repo_result.report_results(verbose)
added
removed
bzrlib/check.py
