23
22
# TODO: Check for extra files in the control directory.
25
# TODO: Check revision, inventory and entry objects have all
24
# TODO: Check revision, inventory and entry objects have all
27
# TODO: Get every revision in the revision-store even if they're not
28
# referenced by history and make sure they're all valid.
30
# TODO: Perhaps have a way to record errors other than by raising exceptions;
31
# would perhaps be enough to accumulate exception objects in a list without
32
# raising them. If there's more than one exception it'd be good to see them
35
from bzrlib import errors, osutils
36
from bzrlib import repository as _mod_repository
37
from bzrlib import revision
38
from bzrlib.branch import Branch
39
from bzrlib.bzrdir import BzrDir
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
30
from bzrlib.trace import note, warning
31
from bzrlib.osutils import rename, sha_string, fingerprint_file
32
from bzrlib.trace import mutter
33
from bzrlib.errors import BzrCheckError, NoSuchRevision
34
from bzrlib.inventory import ROOT_ID
35
from bzrlib.branch import gen_root_id
45
from bzrlib.workingtree import WorkingTree
38
47
class Check(object):
41
def __init__(self, branch):
48
"""Check a repository"""
50
# The Check object interacts with InventoryEntry.check, etc.
52
def __init__(self, repository):
53
self.repository = repository
43
54
self.checked_text_cnt = 0
44
55
self.checked_rev_cnt = 0
47
58
self.missing_parent_links = {}
48
59
self.missing_inventory_sha_cnt = 0
49
60
self.missing_revision_cnt = 0
50
# maps (file-id, version) -> sha1
61
# maps (file-id, version) -> sha1; used by InventoryFile._check
51
62
self.checked_texts = {}
63
self.checked_weaves = set()
64
self.unreferenced_versions = set()
65
self.inconsistent_parents = []
66
self.rich_roots = repository.supports_rich_root()
67
self.text_key_references = {}
54
self.branch.lock_read()
70
self.repository.lock_read()
71
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
56
self.history = self.branch.revision_history()
57
if not len(self.history):
60
self.planned_revisions = self.branch.get_ancestry(self.history[-1])
61
self.planned_revisions.remove(None)
73
self.progress.update('retrieving inventory', 0, 2)
74
# do not put in init, as it should be done with progess,
75
# and inside the lock.
76
self.inventory_weave = self.repository.inventories
77
self.progress.update('checking revision graph', 1)
78
self.check_revision_graph()
64
self.progress = bzrlib.ui.ui_factory.progress_bar()
65
81
while revno < len(self.planned_revisions):
66
82
rev_id = self.planned_revisions[revno]
67
83
self.progress.update('checking revision', revno,
68
84
len(self.planned_revisions))
70
86
self.check_one_rev(rev_id)
87
# check_weaves is done after the revision scan so that
88
# revision index is known to be valid.
91
self.progress.finished()
92
self.repository.unlock()
94
def check_revision_graph(self):
95
if not self.repository.revision_graph_can_have_wrong_parents():
96
# This check is not necessary.
97
self.revs_with_bad_parents_in_index = None
99
bad_revisions = self.repository._find_inconsistent_revision_parents()
100
self.revs_with_bad_parents_in_index = list(bad_revisions)
102
def plan_revisions(self):
103
repository = self.repository
104
self.planned_revisions = repository.all_revision_ids()
105
self.progress.clear()
106
inventoried = set(key[-1] for key in self.inventory_weave.keys())
107
awol = set(self.planned_revisions) - inventoried
109
raise BzrCheckError('Stored revisions missing from inventory'
110
'{%s}' % ','.join([f for f in awol]))
75
112
def report_results(self, verbose):
76
note('checked branch %s format %d',
78
self.branch._branch_format)
113
note('checked repository %s format %s',
114
self.repository.bzrdir.root_transport,
115
self.repository._format)
80
116
note('%6d revisions', self.checked_rev_cnt)
117
note('%6d file-ids', len(self.checked_weaves))
81
118
note('%6d unique file texts', self.checked_text_cnt)
82
119
note('%6d repeated file texts', self.repeated_text_cnt)
120
note('%6d unreferenced text versions',
121
len(self.unreferenced_versions))
83
122
if self.missing_inventory_sha_cnt:
84
123
note('%6d revisions are missing inventory_sha1',
85
124
self.missing_inventory_sha_cnt)
92
131
for ghost in self.ghosts:
93
132
note(' %s', ghost)
94
133
if len(self.missing_parent_links):
95
note('%6d revisions missing parents in ancestry',
134
note('%6d revisions missing parents in ancestry',
96
135
len(self.missing_parent_links))
98
137
for link, linkers in self.missing_parent_links.items():
99
138
note(' %s should be in the ancestry for:', link)
100
139
for linker in linkers:
101
140
note(' * %s', linker)
142
for file_id, revision_id in self.unreferenced_versions:
143
log_error('unreferenced version: {%s} in %s', revision_id,
145
if len(self.inconsistent_parents):
146
note('%6d inconsistent parents', len(self.inconsistent_parents))
148
for info in self.inconsistent_parents:
149
revision_id, file_id, found_parents, correct_parents = info
150
note(' * %s version %s has parents %r '
152
% (file_id, revision_id, found_parents,
154
if self.revs_with_bad_parents_in_index:
155
note('%6d revisions have incorrect parents in the revision index',
156
len(self.revs_with_bad_parents_in_index))
158
for item in self.revs_with_bad_parents_in_index:
159
revision_id, index_parents, actual_parents = item
161
' %s has wrong parents in index: '
163
revision_id, index_parents, actual_parents)
103
165
def check_one_rev(self, rev_id):
104
166
"""Check one revision.
106
168
rev_id - the one to check
108
last_rev_id - the previous one on the mainline, if any.
170
rev = self.repository.get_revision(rev_id)
111
# mutter(' revision {%s}' % rev_id)
114
rev_history_position = self.history.index(rev_id)
116
rev_history_position = None
118
if rev_history_position:
119
rev = branch.get_revision(rev_id)
120
if rev_history_position > 0:
121
last_rev_id = self.history[rev_history_position - 1]
123
rev = branch.get_revision(rev_id)
125
172
if rev.revision_id != rev_id:
126
173
raise BzrCheckError('wrong internal revision id in revision {%s}'
129
# check the previous history entry is a parent of this entry
131
if last_rev_id is None and rev_history_position is not None:
132
# what if the start is a ghost ? i.e. conceptually the
134
raise BzrCheckError("revision {%s} has %d parents, but is the "
135
"start of the branch"
136
% (rev_id, len(rev.parent_ids)))
137
if last_rev_id is not None:
138
for parent_id in rev.parent_ids:
139
if parent_id == last_rev_id:
176
for parent in rev.parent_ids:
177
if not parent in self.planned_revisions:
178
# rev has a parent we didn't know about.
179
missing_links = self.missing_parent_links.get(parent, [])
180
missing_links.append(rev_id)
181
self.missing_parent_links[parent] = missing_links
182
# list based so somewhat slow,
183
# TODO have a planned_revisions list and set.
184
if self.repository.has_revision(parent):
185
missing_ancestry = self.repository.get_ancestry(parent)
186
for missing in missing_ancestry:
187
if (missing is not None
188
and missing not in self.planned_revisions):
189
self.planned_revisions.append(missing)
142
raise BzrCheckError("previous revision {%s} not listed among "
144
% (last_rev_id, rev_id))
145
for parent in rev.parent_ids:
146
if not parent in self.planned_revisions:
147
missing_links = self.missing_parent_links.get(parent, [])
148
missing_links.append(rev_id)
149
self.missing_parent_links[parent] = missing_links
150
# list based so slow, TODO have a planned_revisions list and set.
151
if self.branch.has_revision(parent):
152
missing_ancestry = self.branch.get_ancestry(parent)
153
for missing in missing_ancestry:
154
if (missing is not None
155
and missing not in self.planned_revisions):
156
self.planned_revisions.append(missing)
158
self.ghosts.append(rev_id)
160
raise BzrCheckError("revision {%s} has no parents listed "
161
"but preceded by {%s}"
162
% (rev_id, last_rev_id))
191
self.ghosts.append(rev_id)
164
193
if rev.inventory_sha1:
165
inv_sha1 = branch.get_inventory_sha1(rev_id)
194
# Loopback - this is currently circular logic as the
195
# knit get_inventory_sha1 call returns rev.inventory_sha1.
196
# Repository.py's get_inventory_sha1 should instead return
197
# inventories.get_record_stream([(revid,)]).next().sha1 or
199
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
166
200
if inv_sha1 != rev.inventory_sha1:
167
201
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
168
202
' value in revision {%s}' % rev_id)
170
missing_inventory_sha_cnt += 1
171
mutter("no inventory_sha1 on revision {%s}" % rev_id)
172
203
self._check_revision_tree(rev_id)
173
204
self.checked_rev_cnt += 1
206
def check_weaves(self):
207
"""Check all the weaves we can get our hands on.
210
self.progress.update('checking inventory', 0, 2)
211
self.inventory_weave.check(progress_bar=self.progress)
212
self.progress.update('checking text storage', 1, 2)
213
self.repository.texts.check(progress_bar=self.progress)
214
weave_checker = self.repository._get_versioned_file_checker(
215
text_key_references=self.text_key_references)
216
result = weave_checker.check_file_version_parents(
217
self.repository.texts, progress_bar=self.progress)
218
self.checked_weaves = weave_checker.file_ids
219
bad_parents, unused_versions = result
220
bad_parents = bad_parents.items()
221
for text_key, (stored_parents, correct_parents) in bad_parents:
222
# XXX not ready for id join/split operations.
223
weave_id = text_key[0]
224
revision_id = text_key[-1]
225
weave_parents = tuple([parent[-1] for parent in stored_parents])
226
correct_parents = tuple([parent[-1] for parent in correct_parents])
227
self.inconsistent_parents.append(
228
(revision_id, weave_id, weave_parents, correct_parents))
229
self.unreferenced_versions.update(unused_versions)
175
231
def _check_revision_tree(self, rev_id):
176
tree = self.branch.revision_tree(rev_id)
232
tree = self.repository.revision_tree(rev_id)
177
233
inv = tree.inventory
236
for path, ie in inv.iter_entries():
237
self._add_entry_to_text_key_references(inv, ie)
180
239
if file_id in seen_ids:
181
240
raise BzrCheckError('duplicated file_id {%s} '
182
241
'in inventory for revision {%s}'
183
242
% (file_id, rev_id))
184
seen_ids[file_id] = True
243
seen_ids.add(file_id)
187
244
ie.check(self, rev_id, inv, tree)
189
for path, ie in inv.iter_entries():
190
245
if path in seen_names:
191
246
raise BzrCheckError('duplicated path %s '
192
247
'in inventory for revision {%s}'
193
248
% (path, rev_id))
194
seen_names[path] = True
251
def _add_entry_to_text_key_references(self, inv, entry):
252
if not self.rich_roots and entry == inv.root:
254
key = (entry.file_id, entry.revision)
255
self.text_key_references.setdefault(key, False)
256
if entry.revision == inv.revision_id:
257
self.text_key_references[key] = True
260
@deprecated_function(deprecated_in((1,6,0)))
197
261
def check(branch, verbose):
198
"""Run consistency checks on a branch."""
199
checker = Check(branch)
201
checker.report_results(verbose)
262
"""Run consistency checks on a branch.
264
Results are reported through logging.
266
Deprecated in 1.6. Please use check_branch instead.
268
:raise BzrCheckError: if there's a consistency error.
270
check_branch(branch, verbose)
273
def check_branch(branch, verbose):
274
"""Run consistency checks on a branch.
276
Results are reported through logging.
278
:raise BzrCheckError: if there's a consistency error.
282
branch_result = branch.check()
285
branch_result.report_results(verbose)
288
def check_dwim(path, verbose, do_branch=False, do_repo=False, do_tree=False):
290
tree, branch, repo, relpath = \
291
BzrDir.open_containing_tree_branch_or_repository(path)
292
except errors.NotBranchError:
293
tree = branch = repo = None
297
note("Checking working tree at '%s'."
298
% (tree.bzrdir.root_transport.base,))
301
log_error("No working tree found at specified location.")
303
if branch is not None:
306
# The branch is in a shared repository
307
repo = branch.repository
309
elif repo is not None:
310
branches = repo.find_branches(using=True)
316
note("Checking repository at '%s'."
317
% (repo.bzrdir.root_transport.base,))
318
result = repo.check()
319
result.report_results(verbose)
322
log_error("No branch found at specified location.")
324
for branch in branches:
325
note("Checking branch at '%s'."
326
% (branch.bzrdir.root_transport.base,))
327
check_branch(branch, verbose)
332
log_error("No branch found at specified location.")
334
log_error("No repository found at specified location.")
added
removed
bzrlib/check.py
