1
# Copyright (C) 2005, 2006 Canonical Ltd
1
# Copyright (C) 2004, 2005 by Martin Pool
2
# Copyright (C) 2005 by Canonical Ltd
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
5
6
# the Free Software Foundation; either version 2 of the License, or
6
7
# (at your option) any later version.
8
9
# This program is distributed in the hope that it will be useful,
9
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
12
# GNU General Public License for more details.
13
14
# You should have received a copy of the GNU General Public License
14
15
# along with this program; if not, write to the Free Software
15
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
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()
19
from bzrlib.trace import note, warning
21
def _update_store_entry(obj, obj_id, branch, store_name, store):
22
"""This is just a meta-function, which handles both revision entries
23
and inventory entries.
25
from bzrlib.trace import mutter
26
import tempfile, os, errno
27
from osutils import rename
28
obj_tmp = tempfile.TemporaryFile()
29
obj.write_xml(obj_tmp)
32
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
33
dir=branch.controlfilename(store_name))
36
orig_obj_path = branch.controlfilename([store_name, obj_id+'.gz'])
37
# Remove the old entry out of the way
38
rename(orig_obj_path, tmp_path)
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)
40
# TODO: We may need to handle the case where the old
41
# entry was not compressed (and thus did not end with .gz)
43
store.add(obj_tmp, obj_id) # Add the new one
44
os.remove(tmp_path) # Remove the old name
45
mutter(' Updated %s entry {%s}' % (store_name, obj_id))
47
# On any exception, restore the old entry
48
rename(tmp_path, orig_obj_path)
51
if os.path.exists(tmp_path):
52
# Unfortunately, the next command might throw
53
# an exception, which will mask a previous exception.
57
def _update_revision_entry(rev, branch):
58
"""After updating the values in a revision, make sure to
59
write out the data, but try to do it in an atomic manner.
61
:param rev: The Revision object to store
62
:param branch: The Branch object where this Revision is to be stored.
64
_update_store_entry(rev, rev.revision_id, branch,
65
'revision-store', branch.revision_store)
67
def _update_inventory_entry(inv, inv_id, branch):
68
"""When an inventory has been modified (such as by adding a unique tree root)
69
this atomically re-generates the file.
71
:param inv: The Inventory
72
:param inv_id: The inventory id for this inventory
73
:param branch: The Branch where this entry will be stored.
75
_update_store_entry(inv, inv_id, branch,
76
'inventory-store', branch.inventory_store)
80
"""Run consistency checks on a branch.
82
TODO: Also check non-mainline revisions mentioned as parents.
84
TODO: Check for extra files in the control directory.
86
from bzrlib.trace import mutter
87
from bzrlib.errors import BzrCheckError, NoSuchRevision
88
from bzrlib.osutils import fingerprint_file
89
from bzrlib.inventory import ROOT_ID
90
from bzrlib.branch import gen_root_id
97
missing_inventory_sha_cnt = 0
98
missing_revision_sha_cnt = 0
99
missing_revision_cnt = 0
101
history = branch.revision_history()
103
revcount = len(history)
106
# for all texts checked, text_id -> sha1
109
progress = bzrlib.ui.ui_factory.progress_bar()
111
for rev_id in history:
113
progress.update('checking revision', revno, revcount)
114
# mutter(' revision {%s}' % rev_id)
115
rev = branch.get_revision(rev_id)
116
if rev.revision_id != rev_id:
117
raise BzrCheckError('wrong internal revision id in revision {%s}'
120
# check the previous history entry is a parent of this entry
122
if last_rev_id is None:
123
raise BzrCheckError("revision {%s} has %d parents, but is the "
124
"start of the branch"
125
% (rev_id, len(rev.parents)))
126
for prr in rev.parents:
127
if prr.revision_id == last_rev_id:
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
# XXX: for pack repositories, this iteration seems a bit redundant
207
# with iteration also done within the versioned file checker. also
208
# note that the inventory/revision/signature data is not included in
209
# this enumeration -- mbp 20080528
210
for i, weave_id in enumerate(weave_ids):
211
self.progress.update('checking versionedfile', i, n_weaves)
212
w = self.repository.weave_store.get_weave(weave_id,
213
self.repository.get_transaction())
214
# No progress here, because it looks ugly.
216
result = weave_checker.check_file_version_parents(w, weave_id)
217
bad_parents, unused_versions = result
218
bad_parents = bad_parents.items()
219
for revision_id, (weave_parents, correct_parents) in bad_parents:
220
self.inconsistent_parents.append(
221
(revision_id, weave_id, weave_parents, correct_parents))
222
for revision_id in unused_versions:
223
self.unreferenced_versions.add((weave_id, revision_id))
224
self.checked_weaves[weave_id] = True
226
def _check_revision_tree(self, rev_id):
227
tree = self.repository.revision_tree(rev_id)
231
if file_id in seen_ids:
232
raise BzrCheckError('duplicated file_id {%s} '
233
'in inventory for revision {%s}'
235
seen_ids[file_id] = True
238
ie.check(self, rev_id, inv, tree)
240
for path, ie in inv.iter_entries():
241
if path in seen_names:
242
raise BzrCheckError('duplicated path %s '
243
'in inventory for revision {%s}'
130
raise BzrCheckError("previous revision {%s} not listed among "
132
% (last_rev_id, rev_id))
134
for prr in rev.parents:
135
if prr.revision_sha1 is None:
136
missing_revision_sha_cnt += 1
138
prid = prr.revision_id
141
actual_sha = branch.get_revision_sha1(prid)
142
except NoSuchRevision:
143
missing_revision_cnt += 1
144
mutter("parent {%s} of {%s} not present in store",
148
if prr.revision_sha1 != actual_sha:
149
raise BzrCheckError("mismatched revision sha1 for "
150
"parent {%s} of {%s}: %s vs %s"
152
prr.revision_sha1, actual_sha))
154
raise BzrCheckError("revision {%s} has no parents listed but preceded "
156
% (rev_id, last_rev_id))
158
if rev.inventory_id != rev_id:
159
mismatch_inv_id.append(rev_id)
161
## TODO: Check all the required fields are present on the revision.
163
if rev.inventory_sha1:
164
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
165
if inv_sha1 != rev.inventory_sha1:
166
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
167
' value in revision {%s}' % rev_id)
169
missing_inventory_sha_cnt += 1
170
mutter("no inventory_sha1 on revision {%s}" % rev_id)
172
inv = branch.get_inventory(rev.inventory_id)
176
## p('revision %d/%d file ids' % (revno, revcount))
178
if file_id in seen_ids:
179
raise BzrCheckError('duplicated file_id {%s} '
180
'in inventory for revision {%s}'
182
seen_ids[file_id] = True
192
if ie.parent_id != None:
193
if ie.parent_id not in seen_ids:
194
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
195
% (ie.parent_id, rev_id))
197
if ie.kind == 'file':
198
if ie.text_id in checked_texts:
199
fp = checked_texts[ie.text_id]
201
if not ie.text_id in branch.text_store:
202
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
204
tf = branch.text_store[ie.text_id]
205
fp = fingerprint_file(tf)
206
checked_texts[ie.text_id] = fp
208
if ie.text_size != fp['size']:
209
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
210
if ie.text_sha1 != fp['sha1']:
211
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
212
elif ie.kind == 'directory':
213
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
214
raise BzrCheckError('directory {%s} has text in revision {%s}'
218
for path, ie in inv.iter_entries():
219
if path in seen_names:
220
raise BzrCheckError('duplicated path %s '
221
'in inventory for revision {%s}'
245
223
seen_names[path] = True
248
def check(branch, verbose):
249
"""Run consistency checks on a branch.
251
Results are reported through logging.
253
:raise BzrCheckError: if there's a consistency error.
257
branch_result = branch.check()
258
repo_result = branch.repository.check([branch.last_revision()])
261
branch_result.report_results(verbose)
262
repo_result.report_results(verbose)
231
note('checked %d revisions, %d file texts' % (revcount, len(checked_texts)))
233
if missing_inventory_sha_cnt:
234
note('%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt)
236
if missing_revision_sha_cnt:
237
note('%d parent links are missing revision_sha1' % missing_revision_sha_cnt)
239
if missing_revision_cnt:
240
note('%d revisions are mentioned but not present' % missing_revision_cnt)
242
if missing_revision_cnt:
243
print '%d revisions are mentioned but not present' % missing_revision_cnt
245
# stub this out for now because the main bzr branch has references
246
# to revisions that aren't present in the store -- mbp 20050804
247
# if (missing_inventory_sha_cnt
248
# or missing_revision_sha_cnt):
249
# print ' (use "bzr upgrade" to fix them)'
252
warning('%d revisions have mismatched inventory ids:' % len(mismatch_inv_id))
253
for rev_id in mismatch_inv_id:
254
warning(' %s', rev_id)
added
removed
bzrlib/check.py
