1
# Copyright (C) 2004, 2005 by Martin Pool
2
# Copyright (C) 2005 by Canonical Ltd
1
# Copyright (C) 2005, 2006 Canonical Ltd
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20
######################################################################
26
from trace import mutter
27
from errors import bailout
30
def check(branch, progress=True):
34
if not (hasattr(out, 'isatty') and out.isatty()):
39
mutter('checking ' + m)
40
out.write('\rchecking: %-50.50s' % m)
44
mutter('checking ' + m)
46
p('history of %r' % branch.base)
50
history = branch.revision_history()
52
revcount = len(history)
58
p('revision %d/%d' % (revno, revcount))
59
mutter(' revision {%s}' % rid)
60
rev = branch.get_revision(rid)
61
if rev.revision_id != rid:
62
bailout('wrong internal revision id in revision {%s}' % rid)
63
if rev.precursor != last_ptr:
64
bailout('mismatched precursor in revision {%s}' % rid)
66
if rid in checked_revs:
67
bailout('repeated revision {%s}' % rid)
70
## TODO: Check all the required fields are present on the revision.
72
inv = branch.get_inventory(rev.inventory_id)
76
p('revision %d/%d file ids' % (revno, revcount))
17
# TODO: Check ancestries are correct for every revision: includes
18
# every committed so far, and in a reasonable order.
20
# TODO: Also check non-mainline revisions mentioned as parents.
22
# TODO: Check for extra files in the control directory.
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
36
from bzrlib import repository as _mod_repository
37
from bzrlib import revision
38
from bzrlib.errors import BzrCheckError
40
from bzrlib.trace import log_error, note
43
"""Check a repository"""
45
# The Check object interacts with InventoryEntry.check, etc.
47
def __init__(self, repository):
48
self.repository = repository
49
self.checked_text_cnt = 0
50
self.checked_rev_cnt = 0
52
self.repeated_text_cnt = 0
53
self.missing_parent_links = {}
54
self.missing_inventory_sha_cnt = 0
55
self.missing_revision_cnt = 0
56
# maps (file-id, version) -> sha1; used by InventoryFile._check
57
self.checked_texts = {}
58
self.checked_weaves = {}
59
self.unreferenced_versions = set()
60
self.inconsistent_parents = []
63
self.repository.lock_read()
64
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
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.
84
self.progress.finished()
85
self.repository.unlock()
87
def check_revision_graph(self):
88
if not self.repository.revision_graph_can_have_wrong_parents():
89
# This check is not necessary.
90
self.revs_with_bad_parents_in_index = None
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]))
105
def report_results(self, verbose):
106
note('checked repository %s format %s',
107
self.repository.bzrdir.root_transport,
108
self.repository._format)
109
note('%6d revisions', self.checked_rev_cnt)
110
note('%6d file-ids', len(self.checked_weaves))
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))
115
if self.missing_inventory_sha_cnt:
116
note('%6d revisions are missing inventory_sha1',
117
self.missing_inventory_sha_cnt)
118
if self.missing_revision_cnt:
119
note('%6d revisions are mentioned but not present',
120
self.missing_revision_cnt)
122
note('%6d ghost revisions', len(self.ghosts))
124
for ghost in self.ghosts:
126
if len(self.missing_parent_links):
127
note('%6d revisions missing parents in ancestry',
128
len(self.missing_parent_links))
130
for link, linkers in self.missing_parent_links.items():
131
note(' %s should be in the ancestry for:', link)
132
for linker in linkers:
133
note(' * %s', linker)
135
for file_id, revision_id in self.unreferenced_versions:
136
log_error('unreferenced version: {%s} in %s', revision_id,
138
if len(self.inconsistent_parents):
139
note('%6d inconsistent parents', len(self.inconsistent_parents))
141
for info in self.inconsistent_parents:
142
revision_id, file_id, found_parents, correct_parents = info
143
note(' * %s version %s has parents %r '
145
% (file_id, revision_id, found_parents,
147
if self.revs_with_bad_parents_in_index:
148
note('%6d revisions have incorrect parents in the revision index',
149
len(self.revs_with_bad_parents_in_index))
151
for item in self.revs_with_bad_parents_in_index:
152
revision_id, index_parents, actual_parents = item
154
' %s has wrong parents in index: '
156
revision_id, index_parents, actual_parents)
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)
165
if rev.revision_id != rev_id:
166
raise BzrCheckError('wrong internal revision id in revision {%s}'
169
for parent in rev.parent_ids:
170
if not parent in self.planned_revisions:
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)
191
self.checked_rev_cnt += 1
193
def check_weaves(self):
194
"""Check all the weaves we can get our hands on.
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)
77
226
for file_id in inv:
78
227
if file_id in seen_ids:
79
bailout('duplicated file_id {%s} in inventory for revision {%s}'
228
raise BzrCheckError('duplicated file_id {%s} '
229
'in inventory for revision {%s}'
231
seen_ids[file_id] = True
85
232
for file_id in inv:
88
p('revision %d/%d file text %d/%d' % (revno, revcount, i, len_inv))
92
if ie.parent_id != None:
93
if ie.parent_id not in seen_ids:
94
bailout('missing parent {%s} in inventory for revision {%s}'
95
% (ie.parent_id, rid))
98
if ie.text_id in checked_texts:
99
fp = checked_texts[ie.text_id]
101
if not ie.text_id in branch.text_store:
102
bailout('text {%s} not in text_store' % ie.text_id)
104
tf = branch.text_store[ie.text_id]
105
fp = osutils.fingerprint_file(tf)
106
checked_texts[ie.text_id] = fp
108
if ie.text_size != fp['size']:
109
bailout('text {%s} wrong size' % ie.text_id)
110
if ie.text_sha1 != fp['sha1']:
111
bailout('text {%s} wrong sha1' % ie.text_id)
112
elif ie.kind == 'directory':
113
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
114
bailout('directory {%s} has text in revision {%s}'
117
p('revision %d/%d file paths' % (revno, revcount))
234
ie.check(self, rev_id, inv, tree)
118
236
for path, ie in inv.iter_entries():
119
237
if path in seen_names:
120
bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
127
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
238
raise BzrCheckError('duplicated path %s '
239
'in inventory for revision {%s}'
241
seen_names[path] = True
244
def check(branch, verbose):
245
"""Run consistency checks on a branch.
247
Results are reported through logging.
249
:raise BzrCheckError: if there's a consistency error.
253
branch_result = branch.check()
254
repo_result = branch.repository.check([branch.last_revision()])
257
branch_result.report_results(verbose)
258
repo_result.report_results(verbose)
added
removed
bzrlib/check.py
