1
# Copyright (C) 2005, 2006 Canonical Ltd
4
# Copyright (C) 2004, 2005 by Martin Pool
5
# Copyright (C) 2005 by Canonical Ltd
3
8
# This program is free software; you can redistribute it and/or modify
4
9
# it under the terms of the GNU General Public License as published by
5
10
# the Free Software Foundation; either version 2 of the License, or
6
11
# (at your option) any later version.
8
13
# This program is distributed in the hope that it will be useful,
9
14
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
15
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
16
# GNU General Public License for more details.
13
18
# You should have received a copy of the GNU General Public License
14
19
# along with this program; if not, write to the Free Software
15
20
# 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.errors import BzrCheckError
37
from bzrlib.trace import note
40
"""Check a repository"""
42
# The Check object interacts with InventoryEntry.check, etc.
44
def __init__(self, repository):
45
self.repository = repository
46
self.checked_text_cnt = 0
47
self.checked_rev_cnt = 0
49
self.repeated_text_cnt = 0
50
self.missing_parent_links = {}
51
self.missing_inventory_sha_cnt = 0
52
self.missing_revision_cnt = 0
53
# maps (file-id, version) -> sha1; used by InventoryFile._check
54
self.checked_texts = {}
55
self.checked_weaves = {}
58
self.repository.lock_read()
59
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
61
self.progress.update('retrieving inventory', 0, 0)
62
# do not put in init, as it should be done with progess,
63
# and inside the lock.
64
self.inventory_weave = self.repository.get_inventory_weave()
68
while revno < len(self.planned_revisions):
69
rev_id = self.planned_revisions[revno]
70
self.progress.update('checking revision', revno,
71
len(self.planned_revisions))
73
self.check_one_rev(rev_id)
75
self.progress.finished()
76
self.repository.unlock()
78
def plan_revisions(self):
79
repository = self.repository
80
self.planned_revisions = set(repository.all_revision_ids())
82
inventoried = set(self.inventory_weave.versions())
83
awol = self.planned_revisions - inventoried
85
raise BzrCheckError('Stored revisions missing from inventory'
86
'{%s}' % ','.join([f for f in awol]))
87
self.planned_revisions = list(self.planned_revisions)
89
def report_results(self, verbose):
90
note('checked repository %s format %s',
91
self.repository.bzrdir.root_transport,
92
self.repository._format)
93
note('%6d revisions', self.checked_rev_cnt)
94
note('%6d unique file texts', self.checked_text_cnt)
95
note('%6d repeated file texts', self.repeated_text_cnt)
96
note('%6d weaves', len(self.checked_weaves))
97
if self.missing_inventory_sha_cnt:
98
note('%6d revisions are missing inventory_sha1',
99
self.missing_inventory_sha_cnt)
100
if self.missing_revision_cnt:
101
note('%6d revisions are mentioned but not present',
102
self.missing_revision_cnt)
104
note('%6d ghost revisions', len(self.ghosts))
106
for ghost in self.ghosts:
108
if len(self.missing_parent_links):
109
note('%6d revisions missing parents in ancestry',
110
len(self.missing_parent_links))
112
for link, linkers in self.missing_parent_links.items():
113
note(' %s should be in the ancestry for:', link)
114
for linker in linkers:
115
note(' * %s', linker)
117
def check_one_rev(self, rev_id):
118
"""Check one revision.
120
rev_id - the one to check
122
rev = self.repository.get_revision(rev_id)
124
if rev.revision_id != rev_id:
125
raise BzrCheckError('wrong internal revision id in revision {%s}'
128
for parent in rev.parent_ids:
129
if not parent in self.planned_revisions:
130
missing_links = self.missing_parent_links.get(parent, [])
131
missing_links.append(rev_id)
132
self.missing_parent_links[parent] = missing_links
133
# list based so somewhat slow,
134
# TODO have a planned_revisions list and set.
135
if self.repository.has_revision(parent):
136
missing_ancestry = self.repository.get_ancestry(parent)
137
for missing in missing_ancestry:
138
if (missing is not None
139
and missing not in self.planned_revisions):
140
self.planned_revisions.append(missing)
142
self.ghosts.append(rev_id)
144
if rev.inventory_sha1:
145
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
146
if inv_sha1 != rev.inventory_sha1:
147
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
148
' value in revision {%s}' % rev_id)
149
self._check_revision_tree(rev_id)
150
self.checked_rev_cnt += 1
152
def check_weaves(self):
153
"""Check all the weaves we can get our hands on.
157
if self.repository.weave_store.listable():
158
weave_ids = list(self.repository.weave_store)
159
n_weaves = len(weave_ids)
160
self.progress.update('checking weave', 0, n_weaves)
161
self.inventory_weave.check(progress_bar=self.progress)
162
for i, weave_id in enumerate(weave_ids):
163
self.progress.update('checking weave', i, n_weaves)
164
w = self.repository.weave_store.get_weave(weave_id,
165
self.repository.get_transaction())
166
# No progress here, because it looks ugly.
168
self.checked_weaves[weave_id] = True
170
def _check_revision_tree(self, rev_id):
171
tree = self.repository.revision_tree(rev_id)
175
if file_id in seen_ids:
176
raise BzrCheckError('duplicated file_id {%s} '
177
'in inventory for revision {%s}'
179
seen_ids[file_id] = True
182
ie.check(self, rev_id, inv, tree)
184
for path, ie in inv.iter_entries():
185
if path in seen_names:
186
raise BzrCheckError('duplicated path %s '
187
'in inventory for revision {%s}'
189
seen_names[path] = True
192
def check(branch, verbose):
193
"""Run consistency checks on a branch.
195
Results are reported through logging.
197
:raise BzrCheckError: if there's a consistency error.
201
branch_result = branch.check()
202
repo_result = branch.repository.check([branch.last_revision()])
205
branch_result.report_results(verbose)
206
repo_result.report_results(verbose)
24
######################################################################
28
"""Consistency check of tree."""
30
mutter("checking tree")
32
check_patch_chaining()
33
check_patch_uniqueness()
35
mutter("tree looks OK")
36
## TODO: Check that previous-inventory and previous-manifest
37
## are the same as those stored in the previous changeset.
39
## TODO: Check all patches present in patch directory are
40
## mentioned in patch history; having an orphaned patch only gives
43
## TODO: Check cached data is consistent with data reconstructed
46
## TODO: Check no control files are versioned.
48
## TODO: Check that the before-hash of each file in a later
49
## revision matches the after-hash in the previous revision to
53
def check_inventory():
54
mutter("checking inventory file and ids...")
58
for l in controlfile('inventory').readlines():
61
bailout("malformed inventory line: " + `l`)
64
if file_id in seen_ids:
65
bailout("duplicated file id " + file_id)
68
if name in seen_names:
69
bailout("duplicated file name in inventory: " + quotefn(name))
72
if is_control_file(name):
73
raise BzrError("control file %s present in inventory" % quotefn(name))
76
def check_patches_exist():
77
"""Check constraint of current version: all patches exist"""
78
mutter("checking all patches are present...")
79
for pid in revision_history():
80
read_patch_header(pid)
83
def check_patch_chaining():
84
"""Check ancestry of patches and history file is consistent"""
85
mutter("checking patch chaining...")
87
for pid in revision_history():
88
log_prev = read_patch_header(pid).precursor
90
bailout("inconsistent precursor links on " + pid)
94
def check_patch_uniqueness():
95
"""Make sure no patch is listed twice in the history.
97
This should be implied by having correct ancestry but I'll check it
99
mutter("checking history for duplicates...")
101
for pid in revision_history():
103
bailout("patch " + pid + " appears twice in history")
added
removed
bzrlib/check.py
