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:
43
* 'trees', where value is a revid and the looked up objects are revision trees.
44
* 'lefthand-distance', where value is a revid and the looked up objects are the
45
distance along the lefthand path to NULL for that revid.
46
* 'revision-existence', where value is a revid, and the result is True or False
47
indicating that the revision was found/not found.
35
from bzrlib import errors, osutils
36
from bzrlib import repository as _mod_repository
37
from bzrlib import revision
54
38
from bzrlib.branch import Branch
55
39
from bzrlib.bzrdir import BzrDir
56
from bzrlib.revision import NULL_REVISION
57
from bzrlib.trace import note
40
from bzrlib.errors import BzrCheckError
41
from bzrlib.repository import Repository
42
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
43
from bzrlib.trace import log_error, note
58
45
from bzrlib.workingtree import WorkingTree
61
47
class Check(object):
62
48
"""Check a repository"""
64
def __init__(self, repository, check_repo=True):
65
self.repository = repository
67
def report_results(self, verbose):
68
raise NotImplementedError(self.report_results)
71
class VersionedFileCheck(Check):
72
"""Check a versioned file repository"""
74
50
# The Check object interacts with InventoryEntry.check, etc.
76
def __init__(self, repository, check_repo=True):
52
def __init__(self, repository):
77
53
self.repository = repository
54
self.checked_text_cnt = 0
78
55
self.checked_rev_cnt = 0
57
self.repeated_text_cnt = 0
80
58
self.missing_parent_links = {}
81
59
self.missing_inventory_sha_cnt = 0
82
60
self.missing_revision_cnt = 0
61
# maps (file-id, version) -> sha1; used by InventoryFile._check
62
self.checked_texts = {}
83
63
self.checked_weaves = set()
84
64
self.unreferenced_versions = set()
85
65
self.inconsistent_parents = []
86
self.rich_roots = repository.supports_rich_root()
87
self.text_key_references = {}
88
self.check_repo = check_repo
89
self.other_results = []
90
# Plain text lines to include in the report
91
self._report_items = []
92
# Keys we are looking for; may be large and need spilling to disk.
93
# key->(type(revision/inventory/text/signature/map), sha1, first-referer)
94
self.pending_keys = {}
95
# Ancestors map for all of revisions being checked; while large helper
96
# functions we call would create it anyway, so better to have once and
100
def check(self, callback_refs=None, check_repo=True):
101
if callback_refs is None:
103
68
self.repository.lock_read()
104
self.progress = ui.ui_factory.nested_progress_bar()
69
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
106
self.progress.update('check', 0, 4)
108
self.progress.update('checking revisions', 0)
109
self.check_revisions()
110
self.progress.update('checking commit contents', 1)
111
self.repository._check_inventories(self)
112
self.progress.update('checking file graphs', 2)
113
# check_weaves is done after the revision scan so that
114
# revision index is known to be valid.
116
self.progress.update('checking branches and trees', 3)
118
repo = self.repository
119
# calculate all refs, and callback the objects requesting them.
121
wanting_items = set()
122
# Current crude version calculates everything and calls
123
# everything at once. Doing a queue and popping as things are
124
# satisfied would be cheaper on memory [but few people have
125
# huge numbers of working trees today. TODO: fix before
129
for ref, wantlist in callback_refs.iteritems():
130
wanting_items.update(wantlist)
133
refs[ref] = repo.revision_tree(value)
134
elif kind == 'lefthand-distance':
136
elif kind == 'revision-existence':
137
existences.add(value)
139
raise AssertionError(
140
'unknown ref kind for ref %s' % ref)
141
node_distances = repo.get_graph().find_lefthand_distances(distances)
142
for key, distance in node_distances.iteritems():
143
refs[('lefthand-distance', key)] = distance
144
if key in existences and distance > 0:
145
refs[('revision-existence', key)] = True
146
existences.remove(key)
147
parent_map = repo.get_graph().get_parent_map(existences)
148
for key in parent_map:
149
refs[('revision-existence', key)] = True
150
existences.remove(key)
151
for key in existences:
152
refs[('revision-existence', key)] = False
153
for item in wanting_items:
154
if isinstance(item, WorkingTree):
156
if isinstance(item, Branch):
157
self.other_results.append(item.check(refs))
71
self.progress.update('retrieving inventory', 0, 2)
72
# do not put in init, as it should be done with progess,
73
# and inside the lock.
74
self.inventory_weave = self.repository.inventories
75
self.progress.update('checking revision graph', 1)
76
self.check_revision_graph()
79
while revno < len(self.planned_revisions):
80
rev_id = self.planned_revisions[revno]
81
self.progress.update('checking revision', revno,
82
len(self.planned_revisions))
84
self.check_one_rev(rev_id)
85
# check_weaves is done after the revision scan so that
86
# revision index is known to be valid.
159
89
self.progress.finished()
160
90
self.repository.unlock()
162
def _check_revisions(self, revisions_iterator):
163
"""Check revision objects by decorating a generator.
165
:param revisions_iterator: An iterator of(revid, Revision-or-None).
166
:return: A generator of the contents of revisions_iterator.
168
self.planned_revisions = set()
169
for revid, revision in revisions_iterator:
170
yield revid, revision
171
self._check_one_rev(revid, revision)
172
# Flatten the revisions we found to guarantee consistent later
174
self.planned_revisions = list(self.planned_revisions)
175
# TODO: extract digital signatures as items to callback on too.
177
def check_revisions(self):
178
"""Scan revisions, checking data directly available as we go."""
179
revision_iterator = self.repository._iter_revisions(None)
180
revision_iterator = self._check_revisions(revision_iterator)
181
# We read the all revisions here:
182
# - doing this allows later code to depend on the revision index.
183
# - we can fill out existence flags at this point
184
# - we can read the revision inventory sha at this point
185
# - we can check properties and serialisers etc.
186
if not self.repository._format.revision_graph_can_have_wrong_parents:
187
# The check against the index isn't needed.
92
def check_revision_graph(self):
93
if not self.repository.revision_graph_can_have_wrong_parents():
94
# This check is not necessary.
188
95
self.revs_with_bad_parents_in_index = None
189
for thing in revision_iterator:
192
bad_revisions = self.repository._find_inconsistent_revision_parents(
194
self.revs_with_bad_parents_in_index = list(bad_revisions)
97
bad_revisions = self.repository._find_inconsistent_revision_parents()
98
self.revs_with_bad_parents_in_index = list(bad_revisions)
100
def plan_revisions(self):
101
repository = self.repository
102
self.planned_revisions = repository.all_revision_ids()
103
self.progress.clear()
104
inventoried = set(key[-1] for key in self.inventory_weave.keys())
105
awol = set(self.planned_revisions) - inventoried
107
raise BzrCheckError('Stored revisions missing from inventory'
108
'{%s}' % ','.join([f for f in awol]))
196
110
def report_results(self, verbose):
198
self._report_repo_results(verbose)
199
for result in self.other_results:
200
result.report_results(verbose)
202
def _report_repo_results(self, verbose):
203
111
note('checked repository %s format %s',
204
self.repository.user_url,
205
self.repository._format)
112
self.repository.bzrdir.root_transport,
113
self.repository._format)
206
114
note('%6d revisions', self.checked_rev_cnt)
207
115
note('%6d file-ids', len(self.checked_weaves))
209
note('%6d unreferenced text versions',
210
len(self.unreferenced_versions))
211
if verbose and len(self.unreferenced_versions):
212
for file_id, revision_id in self.unreferenced_versions:
213
note('unreferenced version: {%s} in %s', revision_id,
116
note('%6d unique file texts', self.checked_text_cnt)
117
note('%6d repeated file texts', self.repeated_text_cnt)
118
note('%6d unreferenced text versions',
119
len(self.unreferenced_versions))
215
120
if self.missing_inventory_sha_cnt:
216
121
note('%6d revisions are missing inventory_sha1',
217
122
self.missing_inventory_sha_cnt)
250
159
' %s has wrong parents in index: '
251
160
'%r should be %r',
252
161
revision_id, index_parents, actual_parents)
253
for item in self._report_items:
256
def _check_one_rev(self, rev_id, rev):
257
"""Cross-check one revision.
259
:param rev_id: A revision id to check.
260
:param rev: A revision or None to indicate a missing revision.
163
def check_one_rev(self, rev_id):
164
"""Check one revision.
166
rev_id - the one to check
168
rev = self.repository.get_revision(rev_id)
262
170
if rev.revision_id != rev_id:
263
self._report_items.append(
264
'Mismatched internal revid {%s} and index revid {%s}' % (
265
rev.revision_id, rev_id))
266
rev_id = rev.revision_id
267
# Check this revision tree etc, and count as seen when we encounter a
269
self.planned_revisions.add(rev_id)
271
self.ghosts.discard(rev_id)
272
# Count all parents as ghosts if we haven't seen them yet.
171
raise BzrCheckError('wrong internal revision id in revision {%s}'
273
174
for parent in rev.parent_ids:
274
175
if not parent in self.planned_revisions:
275
self.ghosts.add(parent)
277
self.ancestors[rev_id] = tuple(rev.parent_ids) or (NULL_REVISION,)
278
self.add_pending_item(rev_id, ('inventories', rev_id), 'inventory',
176
missing_links = self.missing_parent_links.get(parent, [])
177
missing_links.append(rev_id)
178
self.missing_parent_links[parent] = missing_links
179
# list based so somewhat slow,
180
# TODO have a planned_revisions list and set.
181
if self.repository.has_revision(parent):
182
missing_ancestry = self.repository.get_ancestry(parent)
183
for missing in missing_ancestry:
184
if (missing is not None
185
and missing not in self.planned_revisions):
186
self.planned_revisions.append(missing)
188
self.ghosts.append(rev_id)
190
if rev.inventory_sha1:
191
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
192
if inv_sha1 != rev.inventory_sha1:
193
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
194
' value in revision {%s}' % rev_id)
195
self._check_revision_tree(rev_id)
280
196
self.checked_rev_cnt += 1
282
def add_pending_item(self, referer, key, kind, sha1):
283
"""Add a reference to a sha1 to be cross checked against a key.
285
:param referer: The referer that expects key to have sha1.
286
:param key: A storage key e.g. ('texts', 'foo@bar-20040504-1234')
287
:param kind: revision/inventory/text/map/signature
288
:param sha1: A hex sha1 or None if no sha1 is known.
290
existing = self.pending_keys.get(key)
292
if sha1 != existing[1]:
293
self._report_items.append('Multiple expected sha1s for %s. {%s}'
294
' expects {%s}, {%s} expects {%s}', (
295
key, referer, sha1, existing[1], existing[0]))
297
self.pending_keys[key] = (kind, sha1, referer)
299
198
def check_weaves(self):
300
199
"""Check all the weaves we can get our hands on.
303
storebar = ui.ui_factory.nested_progress_bar()
305
self._check_weaves(storebar)
309
def _check_weaves(self, storebar):
310
storebar.update('text-index', 0, 2)
311
if self.repository._format.fast_deltas:
312
# We haven't considered every fileid instance so far.
313
weave_checker = self.repository._get_versioned_file_checker(
314
ancestors=self.ancestors)
316
weave_checker = self.repository._get_versioned_file_checker(
317
text_key_references=self.text_key_references,
318
ancestors=self.ancestors)
319
storebar.update('file-graph', 1)
202
self.progress.update('checking inventory', 0, 2)
203
self.inventory_weave.check(progress_bar=self.progress)
204
self.progress.update('checking text storage', 1, 2)
205
self.repository.texts.check(progress_bar=self.progress)
206
weave_checker = self.repository._get_versioned_file_checker()
320
207
result = weave_checker.check_file_version_parents(
321
self.repository.texts)
208
self.repository.texts, progress_bar=self.progress)
322
209
self.checked_weaves = weave_checker.file_ids
323
210
bad_parents, unused_versions = result
324
211
bad_parents = bad_parents.items()
332
219
(revision_id, weave_id, weave_parents, correct_parents))
333
220
self.unreferenced_versions.update(unused_versions)
335
def _add_entry_to_text_key_references(self, inv, entry):
336
if not self.rich_roots and entry.name == '':
338
key = (entry.file_id, entry.revision)
339
self.text_key_references.setdefault(key, False)
340
if entry.revision == inv.revision_id:
341
self.text_key_references[key] = True
344
def scan_branch(branch, needed_refs, to_unlock):
345
"""Scan a branch for refs.
347
:param branch: The branch to schedule for checking.
348
:param needed_refs: Refs we are accumulating.
349
:param to_unlock: The unlock list accumulating.
351
note("Checking branch at '%s'." % (branch.base,))
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
244
@deprecated_function(deprecated_in((1,6,0)))
245
def check(branch, verbose):
246
"""Run consistency checks on a branch.
248
Results are reported through logging.
250
Deprecated in 1.6. Please use check_branch instead.
252
:raise BzrCheckError: if there's a consistency error.
254
check_branch(branch, verbose)
257
def check_branch(branch, verbose):
258
"""Run consistency checks on a branch.
260
Results are reported through logging.
262
:raise BzrCheckError: if there's a consistency error.
352
264
branch.lock_read()
353
to_unlock.append(branch)
354
branch_refs = branch._get_check_refs()
355
for ref in branch_refs:
356
reflist = needed_refs.setdefault(ref, [])
357
reflist.append(branch)
360
def scan_tree(base_tree, tree, needed_refs, to_unlock):
361
"""Scan a tree for refs.
363
:param base_tree: The original tree check opened, used to detect duplicate
365
:param tree: The tree to schedule for checking.
366
:param needed_refs: Refs we are accumulating.
367
:param to_unlock: The unlock list accumulating.
369
if base_tree is not None and tree.basedir == base_tree.basedir:
371
note("Checking working tree at '%s'." % (tree.basedir,))
373
to_unlock.append(tree)
374
tree_refs = tree._get_check_refs()
375
for ref in tree_refs:
376
reflist = needed_refs.setdefault(ref, [])
266
branch_result = branch.check()
269
branch_result.report_results(verbose)
380
272
def check_dwim(path, verbose, do_branch=False, do_repo=False, do_tree=False):
381
"""Check multiple objects.
383
If errors occur they are accumulated and reported as far as possible, and
384
an exception raised at the end of the process.
387
base_tree, branch, repo, relpath = \
274
tree, branch, repo, relpath = \
388
275
BzrDir.open_containing_tree_branch_or_repository(path)
389
276
except errors.NotBranchError:
390
base_tree = branch = repo = None
395
if base_tree is not None:
396
# If the tree is a lightweight checkout we won't see it in
397
# repo.find_branches - add now.
399
scan_tree(None, base_tree, needed_refs, to_unlock)
400
branch = base_tree.branch
401
if branch is not None:
404
# The branch is in a shared repository
405
repo = branch.repository
408
to_unlock.append(repo)
409
branches = repo.find_branches(using=True)
411
if do_branch or do_tree:
412
for branch in branches:
415
tree = branch.bzrdir.open_workingtree()
417
except (errors.NotLocalUrl, errors.NoWorkingTree):
420
scan_tree(base_tree, tree, needed_refs, to_unlock)
422
scan_branch(branch, needed_refs, to_unlock)
423
if do_branch and not branches:
424
note("No branch found at specified location.")
425
if do_tree and base_tree is None and not saw_tree:
426
note("No working tree found at specified location.")
427
if do_repo or do_branch or do_tree:
429
note("Checking repository at '%s'."
431
result = repo.check(None, callback_refs=needed_refs,
277
tree = branch = repo = None
281
note("Checking working tree at '%s'."
282
% (tree.bzrdir.root_transport.base,))
285
log_error("No working tree found at specified location.")
287
if branch is not None:
290
# The branch is in a shared repository
291
repo = branch.repository
293
elif repo is not None:
294
branches = repo.find_branches(using=True)
300
note("Checking repository at '%s'."
301
% (repo.bzrdir.root_transport.base,))
302
result = repo.check()
433
303
result.report_results(verbose)
436
note("No working tree found at specified location.")
438
note("No branch found at specified location.")
440
note("No repository found at specified location.")
442
for thing in to_unlock:
306
log_error("No branch found at specified location.")
308
for branch in branches:
309
note("Checking branch at '%s'."
310
% (branch.bzrdir.root_transport.base,))
311
check_branch(branch, verbose)
316
log_error("No branch found at specified location.")
318
log_error("No repository found at specified location.")
added
removed
bzrlib/check.py
