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
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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, 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
45
from bzrlib.workingtree import WorkingTree
48
"""Check a repository"""
50
# The Check object interacts with InventoryEntry.check, etc.
52
def __init__(self, repository):
53
self.repository = repository
54
self.checked_text_cnt = 0
55
self.checked_rev_cnt = 0
57
self.repeated_text_cnt = 0
58
self.missing_parent_links = {}
59
self.missing_inventory_sha_cnt = 0
60
self.missing_revision_cnt = 0
61
# maps (file-id, version) -> sha1; used by InventoryFile._check
62
self.checked_texts = {}
63
self.checked_weaves = set()
64
self.unreferenced_versions = set()
65
self.inconsistent_parents = []
66
self.rich_roots = repository.supports_rich_root()
67
self.text_key_references = {}
70
self.repository.lock_read()
71
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
def _update_store_entry(obj, obj_id, branch, store_name, store):
20
"""This is just a meta-function, which handles both revision entries
21
and inventory entries.
23
from bzrlib.trace import mutter
24
import tempfile, os, errno
25
from osutils import rename
26
obj_tmp = tempfile.TemporaryFile()
27
obj.write_xml(obj_tmp)
30
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
31
dir=branch.controlfilename(store_name))
34
orig_obj_path = branch.controlfilename([store_name, obj_id+'.gz'])
35
# Remove the old entry out of the way
36
rename(orig_obj_path, tmp_path)
73
self.progress.update('retrieving inventory', 0, 2)
74
# do not put in init, as it should be done with progess,
75
# and inside the lock.
76
self.inventory_weave = self.repository.inventories
77
self.progress.update('checking revision graph', 1)
78
self.check_revision_graph()
81
while revno < len(self.planned_revisions):
82
rev_id = self.planned_revisions[revno]
83
self.progress.update('checking revision', revno,
84
len(self.planned_revisions))
86
self.check_one_rev(rev_id)
87
# check_weaves is done after the revision scan so that
88
# revision index is known to be valid.
91
self.progress.finished()
92
self.repository.unlock()
94
def check_revision_graph(self):
95
if not self.repository.revision_graph_can_have_wrong_parents():
96
# This check is not necessary.
97
self.revs_with_bad_parents_in_index = None
99
bad_revisions = self.repository._find_inconsistent_revision_parents()
100
self.revs_with_bad_parents_in_index = list(bad_revisions)
102
def plan_revisions(self):
103
repository = self.repository
104
self.planned_revisions = repository.all_revision_ids()
105
self.progress.clear()
106
inventoried = set(key[-1] for key in self.inventory_weave.keys())
107
awol = set(self.planned_revisions) - inventoried
109
raise BzrCheckError('Stored revisions missing from inventory'
110
'{%s}' % ','.join([f for f in awol]))
112
def report_results(self, verbose):
113
note('checked repository %s format %s',
114
self.repository.bzrdir.root_transport,
115
self.repository._format)
116
note('%6d revisions', self.checked_rev_cnt)
117
note('%6d file-ids', len(self.checked_weaves))
118
note('%6d unique file texts', self.checked_text_cnt)
119
note('%6d repeated file texts', self.repeated_text_cnt)
120
note('%6d unreferenced text versions',
121
len(self.unreferenced_versions))
122
if self.missing_inventory_sha_cnt:
123
note('%6d revisions are missing inventory_sha1',
124
self.missing_inventory_sha_cnt)
125
if self.missing_revision_cnt:
126
note('%6d revisions are mentioned but not present',
127
self.missing_revision_cnt)
129
note('%6d ghost revisions', len(self.ghosts))
131
for ghost in self.ghosts:
133
if len(self.missing_parent_links):
134
note('%6d revisions missing parents in ancestry',
135
len(self.missing_parent_links))
137
for link, linkers in self.missing_parent_links.items():
138
note(' %s should be in the ancestry for:', link)
139
for linker in linkers:
140
note(' * %s', linker)
142
for file_id, revision_id in self.unreferenced_versions:
143
log_error('unreferenced version: {%s} in %s', revision_id,
145
if len(self.inconsistent_parents):
146
note('%6d inconsistent parents', len(self.inconsistent_parents))
148
for info in self.inconsistent_parents:
149
revision_id, file_id, found_parents, correct_parents = info
150
note(' * %s version %s has parents %r '
152
% (file_id, revision_id, found_parents,
154
if self.revs_with_bad_parents_in_index:
155
note('%6d revisions have incorrect parents in the revision index',
156
len(self.revs_with_bad_parents_in_index))
158
for item in self.revs_with_bad_parents_in_index:
159
revision_id, index_parents, actual_parents = item
161
' %s has wrong parents in index: '
163
revision_id, index_parents, actual_parents)
165
def check_one_rev(self, rev_id):
166
"""Check one revision.
168
rev_id - the one to check
170
rev = self.repository.get_revision(rev_id)
172
if rev.revision_id != rev_id:
173
raise BzrCheckError('wrong internal revision id in revision {%s}'
176
for parent in rev.parent_ids:
177
if not parent in self.planned_revisions:
178
# rev has a parent we didn't know about.
179
missing_links = self.missing_parent_links.get(parent, [])
180
missing_links.append(rev_id)
181
self.missing_parent_links[parent] = missing_links
182
# list based so somewhat slow,
183
# TODO have a planned_revisions list and set.
184
if self.repository.has_revision(parent):
185
missing_ancestry = self.repository.get_ancestry(parent)
186
for missing in missing_ancestry:
187
if (missing is not None
188
and missing not in self.planned_revisions):
189
self.planned_revisions.append(missing)
191
self.ghosts.append(rev_id)
193
if rev.inventory_sha1:
194
# Loopback - this is currently circular logic as the
195
# knit get_inventory_sha1 call returns rev.inventory_sha1.
196
# Repository.py's get_inventory_sha1 should instead return
197
# inventories.get_record_stream([(revid,)]).next().sha1 or
199
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
200
if inv_sha1 != rev.inventory_sha1:
201
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
202
' value in revision {%s}' % rev_id)
203
self._check_revision_tree(rev_id)
204
self.checked_rev_cnt += 1
206
def check_weaves(self):
207
"""Check all the weaves we can get our hands on.
210
self.progress.update('checking inventory', 0, 2)
211
self.inventory_weave.check(progress_bar=self.progress)
212
self.progress.update('checking text storage', 1, 2)
213
self.repository.texts.check(progress_bar=self.progress)
214
weave_checker = self.repository._get_versioned_file_checker(
215
text_key_references=self.text_key_references)
216
result = weave_checker.check_file_version_parents(
217
self.repository.texts, progress_bar=self.progress)
218
self.checked_weaves = weave_checker.file_ids
219
bad_parents, unused_versions = result
220
bad_parents = bad_parents.items()
221
for text_key, (stored_parents, correct_parents) in bad_parents:
222
# XXX not ready for id join/split operations.
223
weave_id = text_key[0]
224
revision_id = text_key[-1]
225
weave_parents = tuple([parent[-1] for parent in stored_parents])
226
correct_parents = tuple([parent[-1] for parent in correct_parents])
227
self.inconsistent_parents.append(
228
(revision_id, weave_id, weave_parents, correct_parents))
229
self.unreferenced_versions.update(unused_versions)
231
def _check_revision_tree(self, rev_id):
232
tree = self.repository.revision_tree(rev_id)
236
for path, ie in inv.iter_entries():
237
self._add_entry_to_text_key_references(inv, ie)
239
if file_id in seen_ids:
240
raise BzrCheckError('duplicated file_id {%s} '
241
'in inventory for revision {%s}'
243
seen_ids.add(file_id)
244
ie.check(self, rev_id, inv, tree)
245
if path in seen_names:
246
raise BzrCheckError('duplicated path %s '
247
'in inventory for revision {%s}'
251
def _add_entry_to_text_key_references(self, inv, entry):
252
if not self.rich_roots and entry == inv.root:
254
key = (entry.file_id, entry.revision)
255
self.text_key_references.setdefault(key, False)
256
if entry.revision == inv.revision_id:
257
self.text_key_references[key] = True
260
@deprecated_function(deprecated_in((1,6,0)))
261
def check(branch, verbose):
262
"""Run consistency checks on a branch.
264
Results are reported through logging.
266
Deprecated in 1.6. Please use check_branch instead.
268
:raise BzrCheckError: if there's a consistency error.
270
check_branch(branch, verbose)
273
def check_branch(branch, verbose):
274
"""Run consistency checks on a branch.
276
Results are reported through logging.
278
:raise BzrCheckError: if there's a consistency error.
38
# TODO: We may need to handle the case where the old
39
# entry was not compressed (and thus did not end with .gz)
41
store.add(obj_tmp, obj_id) # Add the new one
42
os.remove(tmp_path) # Remove the old name
43
mutter(' Updated %s entry {%s}' % (store_name, obj_id))
45
# On any exception, restore the old entry
46
rename(tmp_path, orig_obj_path)
49
if os.path.exists(tmp_path):
50
# Unfortunately, the next command might throw
51
# an exception, which will mask a previous exception.
55
def _update_revision_entry(rev, branch):
56
"""After updating the values in a revision, make sure to
57
write out the data, but try to do it in an atomic manner.
59
:param rev: The Revision object to store
60
:param branch: The Branch object where this Revision is to be stored.
62
_update_store_entry(rev, rev.revision_id, branch,
63
'revision-store', branch.revision_store)
65
def _update_inventory_entry(inv, inv_id, branch):
66
"""When an inventory has been modified (such as by adding a unique tree root)
67
this atomically re-generates the file.
69
:param inv: The Inventory
70
:param inv_id: The inventory id for this inventory
71
:param branch: The Branch where this entry will be stored.
73
_update_store_entry(inv, inv_id, branch,
74
'inventory-store', branch.inventory_store)
77
"""Run consistency checks on a branch.
79
TODO: Also check non-mainline revisions mentioned as parents.
81
TODO: Check for extra files in the control directory.
83
from bzrlib.trace import mutter
84
from bzrlib.errors import BzrCheckError
85
from bzrlib.osutils import fingerprint_file
86
from bzrlib.progress import ProgressBar
87
from bzrlib.inventory import ROOT_ID
88
from bzrlib.branch import gen_root_id
280
90
branch.lock_read()
282
branch_result = branch.check()
93
pb = ProgressBar(show_spinner=True)
96
missing_inventory_sha_cnt = 0
97
missing_revision_sha_cnt = 0
99
history = branch.revision_history()
101
revcount = len(history)
104
# for all texts checked, text_id -> sha1
107
for rev_id in history:
109
pb.update('checking revision', revno, revcount)
110
mutter(' revision {%s}' % rev_id)
111
rev = branch.get_revision(rev_id)
112
if rev.revision_id != rev_id:
113
raise BzrCheckError('wrong internal revision id in revision {%s}'
116
# check the previous history entry is a parent of this entry
118
if last_rev_id is None:
119
raise BzrCheckError("revision {%s} has %d parents, but is the "
120
"start of the branch"
121
% (rev_id, len(rev.parents)))
122
for prr in rev.parents:
123
if prr.revision_id == last_rev_id:
126
raise BzrCheckError("previous revision {%s} not listed among "
128
% (last_rev_id, rev_id))
130
for prr in rev.parents:
131
if prr.revision_sha1 is None:
132
missing_revision_sha_cnt += 1
134
prid = prr.revision_id
135
actual_sha = branch.get_revision_sha1(prid)
136
if prr.revision_sha1 != actual_sha:
137
raise BzrCheckError("mismatched revision sha1 for "
138
"parent {%s} of {%s}: %s vs %s"
140
prr.revision_sha1, actual_sha))
142
raise BzrCheckError("revision {%s} has no parents listed but preceded "
144
% (rev_id, last_rev_id))
146
if rev.inventory_id != rev_id:
147
mismatch_inv_id.append(rev_id)
149
## TODO: Check all the required fields are present on the revision.
151
if rev.inventory_sha1:
152
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
153
if inv_sha1 != rev.inventory_sha1:
154
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
155
' value in revision {%s}' % rev_id)
157
missing_inventory_sha_cnt += 1
158
mutter("no inventory_sha1 on revision {%s}" % rev_id)
160
inv = branch.get_inventory(rev.inventory_id)
164
## p('revision %d/%d file ids' % (revno, revcount))
166
if file_id in seen_ids:
167
raise BzrCheckError('duplicated file_id {%s} '
168
'in inventory for revision {%s}'
170
seen_ids[file_id] = True
180
if ie.parent_id != None:
181
if ie.parent_id not in seen_ids:
182
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
183
% (ie.parent_id, rev_id))
185
if ie.kind == 'file':
186
if ie.text_id in checked_texts:
187
fp = checked_texts[ie.text_id]
189
if not ie.text_id in branch.text_store:
190
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
192
tf = branch.text_store[ie.text_id]
193
fp = fingerprint_file(tf)
194
checked_texts[ie.text_id] = fp
196
if ie.text_size != fp['size']:
197
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
198
if ie.text_sha1 != fp['sha1']:
199
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
200
elif ie.kind == 'directory':
201
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
202
raise BzrCheckError('directory {%s} has text in revision {%s}'
206
for path, ie in inv.iter_entries():
207
if path in seen_names:
208
raise BzrCheckError('duplicated path %s '
209
'in inventory for revision {%s}'
211
seen_names[path] = True
285
branch_result.report_results(verbose)
288
def check_dwim(path, verbose, do_branch=False, do_repo=False, do_tree=False):
290
tree, branch, repo, relpath = \
291
BzrDir.open_containing_tree_branch_or_repository(path)
292
except errors.NotBranchError:
293
tree = branch = repo = None
297
note("Checking working tree at '%s'."
298
% (tree.bzrdir.root_transport.base,))
301
log_error("No working tree found at specified location.")
303
if branch is not None:
306
# The branch is in a shared repository
307
repo = branch.repository
309
elif repo is not None:
310
branches = repo.find_branches(using=True)
316
note("Checking repository at '%s'."
317
% (repo.bzrdir.root_transport.base,))
318
result = repo.check()
319
result.report_results(verbose)
322
log_error("No branch found at specified location.")
324
for branch in branches:
325
note("Checking branch at '%s'."
326
% (branch.bzrdir.root_transport.base,))
327
check_branch(branch, verbose)
332
log_error("No branch found at specified location.")
334
log_error("No repository found at specified location.")
219
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
221
if missing_inventory_sha_cnt:
222
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
224
if missing_revision_sha_cnt:
225
print '%d parent links are missing revision_sha1' % missing_revision_sha_cnt
227
# stub this out for now because the main bzr branch has references
228
# to revisions that aren't present in the store -- mbp 20050804
229
# if (missing_inventory_sha_cnt
230
# or missing_revision_sha_cnt):
231
# print ' (use "bzr upgrade" to fix them)'
234
print '%d revisions have mismatched inventory ids:' % len(mismatch_inv_id)
235
for rev_id in mismatch_inv_id: