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 = []
63
self.dangling_versions = set()
66
self.repository.lock_read()
67
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
69
self.progress.update('retrieving inventory', 0, 2)
70
# do not put in init, as it should be done with progess,
71
# and inside the lock.
72
self.inventory_weave = self.repository.get_inventory_weave()
73
self.progress.update('checking revision graph', 1)
74
self.check_revision_graph()
77
while revno < len(self.planned_revisions):
78
rev_id = self.planned_revisions[revno]
79
self.progress.update('checking revision', revno,
80
len(self.planned_revisions))
82
self.check_one_rev(rev_id)
83
# check_weaves is done after the revision scan so that
84
# revision_versions is pre-populated
87
self.progress.finished()
88
self.repository.unlock()
90
def check_revision_graph(self):
91
if not self.repository.revision_graph_can_have_wrong_parents():
92
# This check is not necessary.
93
self.revs_with_bad_parents_in_index = None
95
bad_revisions = self.repository._find_inconsistent_revision_parents()
96
self.revs_with_bad_parents_in_index = list(bad_revisions)
98
def plan_revisions(self):
99
repository = self.repository
100
self.planned_revisions = repository.all_revision_ids()
101
self.progress.clear()
102
inventoried = set(self.inventory_weave.versions())
103
awol = set(self.planned_revisions) - inventoried
105
raise BzrCheckError('Stored revisions missing from inventory'
106
'{%s}' % ','.join([f for f in awol]))
108
def report_results(self, verbose):
109
note('checked repository %s format %s',
110
self.repository.bzrdir.root_transport,
111
self.repository._format)
112
note('%6d revisions', self.checked_rev_cnt)
113
note('%6d file-ids', len(self.checked_weaves))
114
note('%6d unique file texts', self.checked_text_cnt)
115
note('%6d repeated file texts', self.repeated_text_cnt)
116
note('%6d unreferenced text ancestors',
117
len(self.unreferenced_ancestors))
118
if self.missing_inventory_sha_cnt:
119
note('%6d revisions are missing inventory_sha1',
120
self.missing_inventory_sha_cnt)
121
if self.missing_revision_cnt:
122
note('%6d revisions are mentioned but not present',
123
self.missing_revision_cnt)
125
note('%6d ghost revisions', len(self.ghosts))
127
for ghost in self.ghosts:
129
if len(self.missing_parent_links):
130
note('%6d revisions missing parents in ancestry',
131
len(self.missing_parent_links))
133
for link, linkers in self.missing_parent_links.items():
134
note(' %s should be in the ancestry for:', link)
135
for linker in linkers:
136
note(' * %s', linker)
138
for file_id, revision_id in self.unreferenced_ancestors:
139
log_error('unreferenced ancestor: {%s} in %s', revision_id,
141
if len(self.inconsistent_parents):
142
note('%6d inconsistent parents', len(self.inconsistent_parents))
144
for info in self.inconsistent_parents:
145
revision_id, file_id, found_parents, correct_parents = info
146
note(' * %s version %s has parents %r '
148
% (file_id, revision_id, found_parents,
150
if self.revs_with_bad_parents_in_index:
151
note('%6d revisions have incorrect parents in the revision index',
152
len(self.revs_with_bad_parents_in_index))
154
for item in self.revs_with_bad_parents_in_index:
155
revision_id, index_parents, actual_parents = item
157
' %s has wrong parents in index: '
159
revision_id, index_parents, actual_parents)
160
if self.dangling_versions:
161
note('%6d file versions are not referenced by their inventory',
162
len(self.dangling_versions))
164
def check_one_rev(self, rev_id):
165
"""Check one revision.
167
rev_id - the one to check
169
rev = self.repository.get_revision(rev_id)
171
if rev.revision_id != rev_id:
172
raise BzrCheckError('wrong internal revision id in revision {%s}'
175
for parent in rev.parent_ids:
176
if not parent in self.planned_revisions:
177
missing_links = self.missing_parent_links.get(parent, [])
178
missing_links.append(rev_id)
179
self.missing_parent_links[parent] = missing_links
180
# list based so somewhat slow,
181
# TODO have a planned_revisions list and set.
182
if self.repository.has_revision(parent):
183
missing_ancestry = self.repository.get_ancestry(parent)
184
for missing in missing_ancestry:
185
if (missing is not None
186
and missing not in self.planned_revisions):
187
self.planned_revisions.append(missing)
189
self.ghosts.append(rev_id)
191
if rev.inventory_sha1:
192
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
193
if inv_sha1 != rev.inventory_sha1:
194
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
195
' value in revision {%s}' % rev_id)
196
self._check_revision_tree(rev_id)
197
self.checked_rev_cnt += 1
199
def check_weaves(self):
200
"""Check all the weaves we can get our hands on.
204
if self.repository.weave_store.listable():
205
weave_ids = list(self.repository.weave_store)
206
n_weaves = len(weave_ids) + 1
207
self.progress.update('checking versionedfile', 0, n_weaves)
208
self.inventory_weave.check(progress_bar=self.progress)
209
files_in_revisions = {}
210
revisions_of_files = {}
211
for i, weave_id in enumerate(weave_ids):
212
self.progress.update('checking versionedfile', i, n_weaves)
213
w = self.repository.weave_store.get_weave(weave_id,
214
self.repository.get_transaction())
215
# No progress here, because it looks ugly.
218
weave_checker = self.repository.get_versioned_file_checker(
219
self.planned_revisions, self.revision_versions)
220
result = weave_checker.check_file_version_parents(w, weave_id)
221
bad_parents, dangling_versions = result
222
bad_parents = bad_parents.items()
223
for revision_id, (weave_parents,correct_parents) in bad_parents:
224
self.inconsistent_parents.append(
225
(revision_id, weave_id, weave_parents, correct_parents))
226
if weave_parents is None:
228
unreferenced_parents = set(weave_parents)-set(correct_parents)
229
for unreferenced_parent in unreferenced_parents:
230
self.unreferenced_ancestors.add(
231
(weave_id, unreferenced_parent))
232
self.dangling_versions.update(dangling_versions)
233
self.checked_weaves[weave_id] = True
235
def _check_revision_tree(self, rev_id):
236
tree = self.repository.revision_tree(rev_id)
237
self.revision_versions.add_revision_text_versions(tree)
77
240
for file_id in inv:
78
241
if file_id in seen_ids:
79
bailout('duplicated file_id {%s} in inventory for revision {%s}'
242
raise BzrCheckError('duplicated file_id {%s} '
243
'in inventory for revision {%s}'
245
seen_ids[file_id] = True
85
246
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))
248
ie.check(self, rev_id, inv, tree)
118
250
for path, ie in inv.iter_entries():
119
251
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))
252
raise BzrCheckError('duplicated path %s '
253
'in inventory for revision {%s}'
255
seen_names[path] = True
258
def check(branch, verbose):
259
"""Run consistency checks on a branch.
261
Results are reported through logging.
263
:raise BzrCheckError: if there's a consistency error.
267
branch_result = branch.check()
268
repo_result = branch.repository.check([branch.last_revision()])
271
branch_result.report_results(verbose)
272
repo_result.report_results(verbose)
added
removed
bzrlib/check.py
