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.revision_versions = _mod_repository._RevisionTextVersionCache(
61
self.unreferenced_ancestors = set()
62
self.inconsistent_parents = []
65
self.repository.lock_read()
66
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
68
self.progress.update('retrieving inventory', 0, 2)
69
# do not put in init, as it should be done with progess,
70
# and inside the lock.
71
self.inventory_weave = self.repository.get_inventory_weave()
72
self.progress.update('checking revision graph', 1)
73
self.check_revision_graph()
76
while revno < len(self.planned_revisions):
77
rev_id = self.planned_revisions[revno]
78
self.progress.update('checking revision', revno,
79
len(self.planned_revisions))
81
self.check_one_rev(rev_id)
82
# check_weaves is done after the revision scan so that
83
# revision_versions is pre-populated
86
self.progress.finished()
87
self.repository.unlock()
89
def check_revision_graph(self):
90
if not self.repository.revision_graph_can_have_wrong_parents():
91
# This check is not necessary.
92
self.revs_with_bad_parents_in_index = None
94
bad_revisions = self.repository._find_inconsistent_revision_parents()
95
self.revs_with_bad_parents_in_index = list(bad_revisions)
97
def plan_revisions(self):
98
repository = self.repository
99
self.planned_revisions = repository.all_revision_ids()
100
self.progress.clear()
101
inventoried = set(self.inventory_weave.versions())
102
awol = set(self.planned_revisions) - inventoried
104
raise BzrCheckError('Stored revisions missing from inventory'
105
'{%s}' % ','.join([f for f in awol]))
107
def report_results(self, verbose):
108
note('checked repository %s format %s',
109
self.repository.bzrdir.root_transport,
110
self.repository._format)
111
note('%6d revisions', self.checked_rev_cnt)
112
note('%6d file-ids', len(self.checked_weaves))
113
note('%6d unique file texts', self.checked_text_cnt)
114
note('%6d repeated file texts', self.repeated_text_cnt)
115
note('%6d unreferenced text ancestors',
116
len(self.unreferenced_ancestors))
117
if self.missing_inventory_sha_cnt:
118
note('%6d revisions are missing inventory_sha1',
119
self.missing_inventory_sha_cnt)
120
if self.missing_revision_cnt:
121
note('%6d revisions are mentioned but not present',
122
self.missing_revision_cnt)
124
note('%6d ghost revisions', len(self.ghosts))
126
for ghost in self.ghosts:
128
if len(self.missing_parent_links):
129
note('%6d revisions missing parents in ancestry',
130
len(self.missing_parent_links))
132
for link, linkers in self.missing_parent_links.items():
133
note(' %s should be in the ancestry for:', link)
134
for linker in linkers:
135
note(' * %s', linker)
137
for file_id, revision_id in self.unreferenced_ancestors:
138
log_error('unreferenced ancestor: {%s} in %s', revision_id,
140
if len(self.inconsistent_parents):
141
note('%6d inconsistent parents', len(self.inconsistent_parents))
143
for info in self.inconsistent_parents:
144
revision_id, file_id, found_parents, correct_parents = info
145
note(' * %s version %s has parents %r '
147
% (file_id, revision_id, found_parents,
149
if self.revs_with_bad_parents_in_index:
150
note('%6d revisions have incorrect parents in the revision index',
151
len(self.revs_with_bad_parents_in_index))
153
for item in self.revs_with_bad_parents_in_index:
154
revision_id, index_parents, actual_parents = item
156
' %s has wrong parents in index: '
158
revision_id, index_parents, actual_parents)
160
def check_one_rev(self, rev_id):
161
"""Check one revision.
163
rev_id - the one to check
165
rev = self.repository.get_revision(rev_id)
167
if rev.revision_id != rev_id:
168
raise BzrCheckError('wrong internal revision id in revision {%s}'
171
for parent in rev.parent_ids:
172
if not parent in self.planned_revisions:
173
missing_links = self.missing_parent_links.get(parent, [])
174
missing_links.append(rev_id)
175
self.missing_parent_links[parent] = missing_links
176
# list based so somewhat slow,
177
# TODO have a planned_revisions list and set.
178
if self.repository.has_revision(parent):
179
missing_ancestry = self.repository.get_ancestry(parent)
180
for missing in missing_ancestry:
181
if (missing is not None
182
and missing not in self.planned_revisions):
183
self.planned_revisions.append(missing)
185
self.ghosts.append(rev_id)
187
if rev.inventory_sha1:
188
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
189
if inv_sha1 != rev.inventory_sha1:
190
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
191
' value in revision {%s}' % rev_id)
192
self._check_revision_tree(rev_id)
193
self.checked_rev_cnt += 1
195
def check_weaves(self):
196
"""Check all the weaves we can get our hands on.
200
if self.repository.weave_store.listable():
201
weave_ids = list(self.repository.weave_store)
202
n_weaves = len(weave_ids) + 1
203
self.progress.update('checking versionedfile', 0, n_weaves)
204
self.inventory_weave.check(progress_bar=self.progress)
205
files_in_revisions = {}
206
revisions_of_files = {}
207
for i, weave_id in enumerate(weave_ids):
208
self.progress.update('checking versionedfile', i, n_weaves)
209
w = self.repository.weave_store.get_weave(weave_id,
210
self.repository.get_transaction())
211
# No progress here, because it looks ugly.
214
weave_checker = self.repository.get_versioned_file_checker(
215
self.planned_revisions, self.revision_versions)
216
result = weave_checker.check_file_version_parents(w, weave_id)
218
for revision_id, (weave_parents,correct_parents) in result.items():
219
self.inconsistent_parents.append(
220
(revision_id, weave_id, weave_parents, correct_parents))
221
unreferenced_parents = set(weave_parents)-set(correct_parents)
222
for unreferenced_parent in unreferenced_parents:
223
self.unreferenced_ancestors.add(
224
(weave_id, unreferenced_parent))
225
self.checked_weaves[weave_id] = True
227
def _check_revision_tree(self, rev_id):
228
tree = self.repository.revision_tree(rev_id)
229
self.revision_versions.add_revision_text_versions(tree)
77
232
for file_id in inv:
78
233
if file_id in seen_ids:
79
bailout('duplicated file_id {%s} in inventory for revision {%s}'
234
raise BzrCheckError('duplicated file_id {%s} '
235
'in inventory for revision {%s}'
237
seen_ids[file_id] = True
85
238
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))
240
ie.check(self, rev_id, inv, tree)
118
242
for path, ie in inv.iter_entries():
119
243
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))
244
raise BzrCheckError('duplicated path %s '
245
'in inventory for revision {%s}'
247
seen_names[path] = True
250
def check(branch, verbose):
251
"""Run consistency checks on a branch.
253
Results are reported through logging.
255
:raise BzrCheckError: if there's a consistency error.
259
branch_result = branch.check()
260
repo_result = branch.repository.check([branch.last_revision()])
263
branch_result.report_results(verbose)
264
repo_result.report_results(verbose)
added
removed
bzrlib/check.py
