27
28
# TODO: Get every revision in the revision-store even if they're not
28
29
# 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
45
from bzrlib.workingtree import WorkingTree
32
from bzrlib.trace import note, warning
33
from bzrlib.osutils import rename, sha_string, fingerprint_file
34
from bzrlib.trace import mutter
35
from bzrlib.errors import BzrCheckError, NoSuchRevision
36
from bzrlib.inventory import ROOT_ID
47
39
class Check(object):
48
"""Check a repository"""
50
# The Check object interacts with InventoryEntry.check, etc.
52
def __init__(self, repository):
53
self.repository = repository
42
def __init__(self, branch):
44
self.repository = branch.repository
54
45
self.checked_text_cnt = 0
55
46
self.checked_rev_cnt = 0
58
49
self.missing_parent_links = {}
59
50
self.missing_inventory_sha_cnt = 0
60
51
self.missing_revision_cnt = 0
61
# maps (file-id, version) -> sha1; used by InventoryFile._check
52
# maps (file-id, version) -> sha1
62
53
self.checked_texts = {}
63
self.checked_weaves = set()
64
self.unreferenced_versions = set()
65
self.inconsistent_parents = []
54
self.checked_weaves = {}
68
self.repository.lock_read()
69
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
57
self.branch.lock_read()
58
self.progress = bzrlib.ui.ui_factory.progress_bar()
71
self.progress.update('retrieving inventory', 0, 2)
60
self.progress.update('retrieving inventory', 0, 0)
72
61
# do not put in init, as it should be done with progess,
73
62
# and inside the lock.
74
self.inventory_weave = self.repository.inventories
75
self.progress.update('checking revision graph', 1)
76
self.check_revision_graph()
63
self.inventory_weave = self.branch.repository.get_inventory_weave()
64
self.history = self.branch.revision_history()
65
if not len(self.history):
77
68
self.plan_revisions()
79
71
while revno < len(self.planned_revisions):
80
72
rev_id = self.planned_revisions[revno]
81
73
self.progress.update('checking revision', revno,
82
74
len(self.planned_revisions))
84
76
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.
89
self.progress.finished()
90
self.repository.unlock()
92
def check_revision_graph(self):
93
if not self.repository.revision_graph_can_have_wrong_parents():
94
# This check is not necessary.
95
self.revs_with_bad_parents_in_index = None
81
def plan_revisions(self):
82
repository = self.branch.repository
83
if not repository.revision_store.listable():
84
self.planned_revisions = repository.get_ancestry(self.history[-1])
85
self.planned_revisions.remove(None)
86
# FIXME progress bars should support this more nicely.
88
print ("Checking reachable history -"
89
" for a complete check use a local branch.")
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
92
self.planned_revisions = set(repository.revision_store)
93
inventoried = set(self.inventory_weave.names())
94
awol = self.planned_revisions - inventoried
107
96
raise BzrCheckError('Stored revisions missing from inventory'
108
97
'{%s}' % ','.join([f for f in awol]))
98
self.planned_revisions = list(self.planned_revisions)
110
100
def report_results(self, verbose):
111
note('checked repository %s format %s',
112
self.repository.bzrdir.root_transport,
113
self.repository._format)
101
note('checked branch %s format %s',
114
105
note('%6d revisions', self.checked_rev_cnt)
115
note('%6d file-ids', len(self.checked_weaves))
116
106
note('%6d unique file texts', self.checked_text_cnt)
117
107
note('%6d repeated file texts', self.repeated_text_cnt)
118
note('%6d unreferenced text versions',
119
len(self.unreferenced_versions))
108
note('%6d weaves', len(self.checked_weaves))
120
109
if self.missing_inventory_sha_cnt:
121
110
note('%6d revisions are missing inventory_sha1',
122
111
self.missing_inventory_sha_cnt)
129
118
for ghost in self.ghosts:
130
119
note(' %s', ghost)
131
120
if len(self.missing_parent_links):
132
note('%6d revisions missing parents in ancestry',
121
note('%6d revisions missing parents in ancestry',
133
122
len(self.missing_parent_links))
135
124
for link, linkers in self.missing_parent_links.items():
136
125
note(' %s should be in the ancestry for:', link)
137
126
for linker in linkers:
138
127
note(' * %s', linker)
140
for file_id, revision_id in self.unreferenced_versions:
141
log_error('unreferenced version: {%s} in %s', revision_id,
143
if len(self.inconsistent_parents):
144
note('%6d inconsistent parents', len(self.inconsistent_parents))
146
for info in self.inconsistent_parents:
147
revision_id, file_id, found_parents, correct_parents = info
148
note(' * %s version %s has parents %r '
150
% (file_id, revision_id, found_parents,
152
if self.revs_with_bad_parents_in_index:
153
note('%6d revisions have incorrect parents in the revision index',
154
len(self.revs_with_bad_parents_in_index))
156
for item in self.revs_with_bad_parents_in_index:
157
revision_id, index_parents, actual_parents = item
159
' %s has wrong parents in index: '
161
revision_id, index_parents, actual_parents)
163
129
def check_one_rev(self, rev_id):
164
130
"""Check one revision.
166
132
rev_id - the one to check
134
last_rev_id - the previous one on the mainline, if any.
168
rev = self.repository.get_revision(rev_id)
137
# mutter(' revision {%s}', rev_id)
140
rev_history_position = self.history.index(rev_id)
142
rev_history_position = None
144
if rev_history_position:
145
rev = branch.repository.get_revision(rev_id)
146
if rev_history_position > 0:
147
last_rev_id = self.history[rev_history_position - 1]
149
rev = branch.repository.get_revision(rev_id)
170
151
if rev.revision_id != rev_id:
171
152
raise BzrCheckError('wrong internal revision id in revision {%s}'
174
for parent in rev.parent_ids:
175
if not parent in self.planned_revisions:
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)
155
# check the previous history entry is a parent of this entry
157
if last_rev_id is not None:
158
for parent_id in rev.parent_ids:
159
if parent_id == last_rev_id:
188
self.ghosts.append(rev_id)
162
raise BzrCheckError("previous revision {%s} not listed among "
164
% (last_rev_id, rev_id))
165
for parent in rev.parent_ids:
166
if not parent in self.planned_revisions:
167
missing_links = self.missing_parent_links.get(parent, [])
168
missing_links.append(rev_id)
169
self.missing_parent_links[parent] = missing_links
170
# list based so somewhat slow,
171
# TODO have a planned_revisions list and set.
172
if self.branch.has_revision(parent):
173
missing_ancestry = self.repository.get_ancestry(parent)
174
for missing in missing_ancestry:
175
if (missing is not None
176
and missing not in self.planned_revisions):
177
self.planned_revisions.append(missing)
179
self.ghosts.append(rev_id)
181
raise BzrCheckError("revision {%s} has no parents listed "
182
"but preceded by {%s}"
183
% (rev_id, last_rev_id))
190
185
if rev.inventory_sha1:
191
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
186
inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
192
187
if inv_sha1 != rev.inventory_sha1:
193
188
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
194
189
' value in revision {%s}' % rev_id)
191
missing_inventory_sha_cnt += 1
192
mutter("no inventory_sha1 on revision {%s}", rev_id)
195
193
self._check_revision_tree(rev_id)
196
194
self.checked_rev_cnt += 1
198
196
def check_weaves(self):
199
197
"""Check all the weaves we can get our hands on.
202
self.progress.update('checking inventory', 0, 2)
201
if self.branch.repository.weave_store.listable():
202
weave_ids = list(self.branch.repository.weave_store)
203
n_weaves = len(weave_ids)
204
self.progress.update('checking weave', 0, n_weaves)
203
205
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()
207
result = weave_checker.check_file_version_parents(
208
self.repository.texts, progress_bar=self.progress)
209
self.checked_weaves = weave_checker.file_ids
210
bad_parents, unused_versions = result
211
bad_parents = bad_parents.items()
212
for text_key, (stored_parents, correct_parents) in bad_parents:
213
# XXX not ready for id join/split operations.
214
weave_id = text_key[0]
215
revision_id = text_key[-1]
216
weave_parents = tuple([parent[-1] for parent in stored_parents])
217
correct_parents = tuple([parent[-1] for parent in correct_parents])
218
self.inconsistent_parents.append(
219
(revision_id, weave_id, weave_parents, correct_parents))
220
self.unreferenced_versions.update(unused_versions)
206
for i, weave_id in enumerate(weave_ids):
207
self.progress.update('checking weave', i, n_weaves)
208
w = self.branch.repository.weave_store.get_weave(weave_id,
209
self.branch.repository.get_transaction())
210
# No progress here, because it looks ugly.
212
self.checked_weaves[weave_id] = True
222
214
def _check_revision_tree(self, rev_id):
223
tree = self.repository.revision_tree(rev_id)
215
tree = self.branch.repository.revision_tree(rev_id)
224
216
inv = tree.inventory
226
218
for file_id in inv:
241
233
seen_names[path] = True
244
@deprecated_function(deprecated_in((1,6,0)))
245
236
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.
266
branch_result = branch.check()
269
branch_result.report_results(verbose)
272
def check_dwim(path, verbose, do_branch=False, do_repo=False, do_tree=False):
274
tree, branch, repo, relpath = \
275
BzrDir.open_containing_tree_branch_or_repository(path)
276
except errors.NotBranchError:
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()
303
result.report_results(verbose)
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.")
237
"""Run consistency checks on a branch."""
238
checker = Check(branch)
240
checker.report_results(verbose)
added
removed
bzrlib/check.py
