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.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):
21
def check(branch, update=False):
193
22
"""Run consistency checks on a branch.
195
Results are reported through logging.
197
:raise BzrCheckError: if there's a consistency error.
24
If update is True, for revisions missing certain information
25
(right now this is inventory_sha1 and revision_sha1),
26
update them to include the appropriate values.
30
from bzrlib.trace import mutter
31
from bzrlib.errors import BzrCheckError
32
from bzrlib.osutils import fingerprint_file
33
from bzrlib.progress import ProgressBar
201
branch_result = branch.check()
202
repo_result = branch.repository.check([branch.last_revision()])
44
pb = ProgressBar(show_spinner=True)
48
missing_inventory_sha_cnt = 0
50
history = branch.revision_history()
52
revcount = len(history)
56
updated_revisions = []
58
# Set to True in the case that the previous revision entry
59
# was updated, since this will affect future revision entries
60
updated_previous_revision = False
64
pb.update('checking revision', revno, revcount)
65
mutter(' revision {%s}' % rid)
66
rev = branch.get_revision(rid)
67
if rev.revision_id != rid:
68
raise BzrCheckError('wrong internal revision id in revision {%s}' % rid)
69
if rev.precursor != last_ptr:
70
raise BzrCheckError('mismatched precursor in revision {%s}' % rid)
72
if rid in checked_revs:
73
raise BzrCheckError('repeated revision {%s}' % rid)
74
checked_revs[rid] = True
76
## TODO: Check all the required fields are present on the revision.
79
if rev.inventory_sha1:
80
#mutter(' checking inventory hash {%s}' % rev.inventory_sha1)
81
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
82
if inv_sha1 != rev.inventory_sha1:
83
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
84
' value in revision {%s}' % rid)
86
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
87
rev.inventory_sha1 = inv_sha1
90
missing_inventory_sha_cnt += 1
91
mutter("no inventory_sha1 on revision {%s}" % rid)
94
if rev.precursor_sha1:
95
precursor_sha1 = branch.get_revision_sha1(rev.precursor)
96
if updated_previous_revision:
97
# we don't expect the hashes to match, because
98
# we had to modify the previous revision_history entry.
99
rev.precursor_sha1 = precursor_sha1
102
#mutter(' checking precursor hash {%s}' % rev.precursor_sha1)
103
if rev.precursor_sha1 != precursor_sha1:
104
raise BzrCheckError('Precursor sha1 hash doesn\'t match'
105
' value in revision {%s}' % rid)
107
precursor_sha1 = branch.get_revision_sha1(rev.precursor)
108
rev.precursor_sha1 = precursor_sha1
112
updated_previous_revision = True
113
# We had to update this revision entries hashes
114
# Now we need to write out a new value
115
# This is a little bit invasive, since we are *rewriting* a
116
# revision entry. I'm not supremely happy about it, but
117
# there should be *some* way of making old entries have
118
# the full meta information.
119
import tempfile, os, errno
120
rev_tmp = tempfile.TemporaryFile()
121
rev.write_xml(rev_tmp)
124
tmpfd, tmp_path = tempfile.mkstemp(prefix=rid, suffix='.gz',
125
dir=branch.controlfilename('revision-store'))
127
def special_rename(p1, p2):
128
if sys.platform == 'win32':
132
if e.errno != e.ENOENT:
137
# TODO: We may need to handle the case where the old revision
138
# entry was not compressed (and thus did not end with .gz)
140
# Remove the old revision entry out of the way
141
rev_path = branch.controlfilename(['revision-store', rid+'.gz'])
142
special_rename(rev_path, tmp_path)
143
branch.revision_store.add(rev_tmp, rid) # Add the new one
144
os.remove(tmp_path) # Remove the old name
145
mutter(' Updated revision entry {%s}' % rid)
147
# On any exception, restore the old entry
148
special_rename(tmp_path, rev_path)
151
updated_revisions.append(rid)
153
updated_previous_revision = False
155
inv = branch.get_inventory(rev.inventory_id)
159
## p('revision %d/%d file ids' % (revno, revcount))
161
if file_id in seen_ids:
162
raise BzrCheckError('duplicated file_id {%s} '
163
'in inventory for revision {%s}'
165
seen_ids[file_id] = True
176
if ie.parent_id != None:
177
if ie.parent_id not in seen_ids:
178
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
179
% (ie.parent_id, rid))
181
if ie.kind == 'file':
182
if ie.text_id in checked_texts:
183
fp = checked_texts[ie.text_id]
185
if not ie.text_id in branch.text_store:
186
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
188
tf = branch.text_store[ie.text_id]
189
fp = fingerprint_file(tf)
190
checked_texts[ie.text_id] = fp
192
if ie.text_size != fp['size']:
193
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
194
if ie.text_sha1 != fp['sha1']:
195
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
196
elif ie.kind == 'directory':
197
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
198
raise BzrCheckError('directory {%s} has text in revision {%s}'
202
for path, ie in inv.iter_entries():
203
if path in seen_names:
204
raise BzrCheckError('duplicated path %r '
205
'in inventory for revision {%s}'
207
seen_names[path] = True
205
branch_result.report_results(verbose)
206
repo_result.report_results(verbose)
214
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
215
if updated_revisions:
216
print '%d revisions updated to current format' % len(updated_revisions)
217
if missing_inventory_sha_cnt:
218
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
219
print ' (use bzr check --update to fix them)'
added
removed
bzrlib/check.py
