32
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
36
from bzrlib.trace import note, warning
37
from bzrlib.osutils import rename, sha_string, fingerprint_file
38
from bzrlib.trace import mutter
39
from bzrlib.errors import BzrCheckError, NoSuchRevision
40
from bzrlib.inventory import ROOT_ID
45
from bzrlib.workingtree import WorkingTree
43
47
class Check(object):
48
"""Check a repository"""
46
50
# The Check object interacts with InventoryEntry.check, etc.
48
def __init__(self, branch):
50
self.repository = branch.repository
52
def __init__(self, repository):
53
self.repository = repository
51
54
self.checked_text_cnt = 0
52
55
self.checked_rev_cnt = 0
57
60
self.missing_revision_cnt = 0
58
61
# maps (file-id, version) -> sha1; used by InventoryFile._check
59
62
self.checked_texts = {}
60
self.checked_weaves = {}
63
self.checked_weaves = set()
64
self.unreferenced_versions = set()
65
self.inconsistent_parents = []
63
self.branch.lock_read()
68
self.repository.lock_read()
64
69
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
66
self.progress.update('retrieving inventory', 0, 0)
71
self.progress.update('retrieving inventory', 0, 2)
67
72
# do not put in init, as it should be done with progess,
68
73
# and inside the lock.
69
self.inventory_weave = self.branch.repository.get_inventory_weave()
70
self.history = self.branch.revision_history()
71
if not len(self.history):
74
self.inventory_weave = self.repository.inventories
75
self.progress.update('checking revision graph', 1)
76
self.check_revision_graph()
74
77
self.plan_revisions()
77
79
while revno < len(self.planned_revisions):
78
80
rev_id = self.planned_revisions[revno]
79
81
self.progress.update('checking revision', revno,
80
82
len(self.planned_revisions))
82
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.
84
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
97
bad_revisions = self.repository._find_inconsistent_revision_parents()
98
self.revs_with_bad_parents_in_index = list(bad_revisions)
87
100
def plan_revisions(self):
88
repository = self.branch.repository
89
self.planned_revisions = set(repository.all_revision_ids())
101
repository = self.repository
102
self.planned_revisions = repository.all_revision_ids()
90
103
self.progress.clear()
91
inventoried = set(self.inventory_weave.versions())
92
awol = self.planned_revisions - inventoried
104
inventoried = set(key[-1] for key in self.inventory_weave.keys())
105
awol = set(self.planned_revisions) - inventoried
94
107
raise BzrCheckError('Stored revisions missing from inventory'
95
108
'{%s}' % ','.join([f for f in awol]))
96
self.planned_revisions = list(self.planned_revisions)
98
110
def report_results(self, verbose):
99
note('checked branch %s format %s',
111
note('checked repository %s format %s',
112
self.repository.bzrdir.root_transport,
113
self.repository._format)
103
114
note('%6d revisions', self.checked_rev_cnt)
115
note('%6d file-ids', len(self.checked_weaves))
104
116
note('%6d unique file texts', self.checked_text_cnt)
105
117
note('%6d repeated file texts', self.repeated_text_cnt)
106
note('%6d weaves', len(self.checked_weaves))
118
note('%6d unreferenced text versions',
119
len(self.unreferenced_versions))
107
120
if self.missing_inventory_sha_cnt:
108
121
note('%6d revisions are missing inventory_sha1',
109
122
self.missing_inventory_sha_cnt)
116
129
for ghost in self.ghosts:
117
130
note(' %s', ghost)
118
131
if len(self.missing_parent_links):
119
note('%6d revisions missing parents in ancestry',
132
note('%6d revisions missing parents in ancestry',
120
133
len(self.missing_parent_links))
122
135
for link, linkers in self.missing_parent_links.items():
123
136
note(' %s should be in the ancestry for:', link)
124
137
for linker in linkers:
125
138
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)
127
163
def check_one_rev(self, rev_id):
128
164
"""Check one revision.
130
166
rev_id - the one to check
132
last_rev_id - the previous one on the mainline, if any.
135
# mutter(' revision {%s}', rev_id)
138
rev_history_position = self.history.index(rev_id)
140
rev_history_position = None
142
if rev_history_position:
143
rev = branch.repository.get_revision(rev_id)
144
if rev_history_position > 0:
145
last_rev_id = self.history[rev_history_position - 1]
147
rev = branch.repository.get_revision(rev_id)
168
rev = self.repository.get_revision(rev_id)
149
170
if rev.revision_id != rev_id:
150
171
raise BzrCheckError('wrong internal revision id in revision {%s}'
153
# check the previous history entry is a parent of this entry
155
if last_rev_id is not None:
156
for parent_id in rev.parent_ids:
157
if parent_id == last_rev_id:
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)
160
raise BzrCheckError("previous revision {%s} not listed among "
162
% (last_rev_id, rev_id))
163
for parent in rev.parent_ids:
164
if not parent in self.planned_revisions:
165
missing_links = self.missing_parent_links.get(parent, [])
166
missing_links.append(rev_id)
167
self.missing_parent_links[parent] = missing_links
168
# list based so somewhat slow,
169
# TODO have a planned_revisions list and set.
170
if self.branch.repository.has_revision(parent):
171
missing_ancestry = self.repository.get_ancestry(parent)
172
for missing in missing_ancestry:
173
if (missing is not None
174
and missing not in self.planned_revisions):
175
self.planned_revisions.append(missing)
177
self.ghosts.append(rev_id)
179
raise BzrCheckError("revision {%s} has no parents listed "
180
"but preceded by {%s}"
181
% (rev_id, last_rev_id))
188
self.ghosts.append(rev_id)
183
190
if rev.inventory_sha1:
184
inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
191
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
185
192
if inv_sha1 != rev.inventory_sha1:
186
193
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
187
194
' value in revision {%s}' % rev_id)
189
self.missing_inventory_sha_cnt += 1
190
mutter("no inventory_sha1 on revision {%s}", rev_id)
191
195
self._check_revision_tree(rev_id)
192
196
self.checked_rev_cnt += 1
194
198
def check_weaves(self):
195
199
"""Check all the weaves we can get our hands on.
199
if self.branch.repository.weave_store.listable():
200
weave_ids = list(self.branch.repository.weave_store)
201
n_weaves = len(weave_ids)
202
self.progress.update('checking weave', 0, n_weaves)
202
self.progress.update('checking inventory', 0, 2)
203
203
self.inventory_weave.check(progress_bar=self.progress)
204
for i, weave_id in enumerate(weave_ids):
205
self.progress.update('checking weave', i, n_weaves)
206
w = self.branch.repository.weave_store.get_weave(weave_id,
207
self.branch.repository.get_transaction())
208
# No progress here, because it looks ugly.
210
self.checked_weaves[weave_id] = True
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)
212
222
def _check_revision_tree(self, rev_id):
213
tree = self.branch.repository.revision_tree(rev_id)
223
tree = self.repository.revision_tree(rev_id)
214
224
inv = tree.inventory
216
226
for file_id in inv:
231
241
seen_names[path] = True
244
@deprecated_function(deprecated_in((1,6,0)))
234
245
def check(branch, verbose):
235
"""Run consistency checks on a branch."""
236
checker = Check(branch)
238
checker.report_results(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.")
added
removed
bzrlib/check.py
