« back to all changes in this revision
Viewing changes to bzrlib/check.py
- browse files at revision 1185.12.2
- compare with another revision
- compare with revision 294
- download diff
- download tarball
- view history from revision 1185.12.2
- Committer: Aaron Bentley
- Date: 2005-09-29 21:07:17 UTC
- mfrom: (1393.1.6)
- mto: (1185.25.1)
- mto: This revision was merged to the branch mainline in revision 1419.
- Revision ID: abentley@panoramicfeedback.com-20050929210717-cd73981590f17017
Merged the weave changes
- files added:
- NEWS.developers
- files removed:
- bzrlib/upgrade.py
- files renamed:
- tools/testweave.py => bzrlib/selftest/test_weave.py
- tools/history2weaves.py => bzrlib/upgrade.py
- bzrlib/newinventory.py => contrib/newinventory.py
- files modified:
- .bzrignore
added
removed
bzrlib/check.py
15
15
# along with this program; if not, write to the Free Software
16
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
17
18
# TODO: Check ancestries are correct for every revision: includes
19
# every committed so far, and in a reasonable order.
20
21
# TODO: Also check non-mainline revisions mentioned as parents.
22
23
# TODO: Check for extra files in the control directory.
24
25
# TODO: Check revision, inventory and entry objects have all
26
# required fields.
27
28
18
29
import bzrlib.ui
19
30
from bzrlib.trace import note, warning
20
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.
24
"""
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)
30
obj_tmp.seek(0)
31
32
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
33
dir=branch.controlfilename(store_name))
34
os.close(tmpfd)
35
try:
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)
31
from bzrlib.osutils import rename, sha_string, fingerprint_file, sha_strings
32
from bzrlib.trace import mutter
33
from bzrlib.errors import BzrCheckError, NoSuchRevision
34
from bzrlib.inventory import ROOT_ID
35
from bzrlib.branch import gen_root_id
36
37
38
class Check(object):
39
"""Check a branch"""
40
def __init__(self, branch):
41
self.branch = branch
42
branch.lock_read()
39
43
try:
40
# TODO: We may need to handle the case where the old
41
# entry was not compressed (and thus did not end with .gz)
42
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))
46
except:
47
# On any exception, restore the old entry
48
rename(tmp_path, orig_obj_path)
49
raise
50
finally:
51
if os.path.exists(tmp_path):
52
# Unfortunately, the next command might throw
53
# an exception, which will mask a previous exception.
54
os.remove(tmp_path)
55
obj_tmp.close()
56
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.
60
61
:param rev: The Revision object to store
62
:param branch: The Branch object where this Revision is to be stored.
63
"""
64
_update_store_entry(rev, rev.revision_id, branch,
65
'revision-store', branch.revision_store)
66
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.
70
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.
74
"""
75
_update_store_entry(inv, inv_id, branch,
76
'inventory-store', branch.inventory_store)
77
78
79
def check(branch):
80
"""Run consistency checks on a branch.
81
82
TODO: Also check non-mainline revisions mentioned as parents.
83
84
TODO: Check for extra files in the control directory.
85
"""
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
91
92
branch.lock_read()
93
94
try:
95
last_rev_id = None
96
97
missing_inventory_sha_cnt = 0
98
missing_revision_sha_cnt = 0
99
missing_revision_cnt = 0
44
branch.weave_store.enable_cache = True
45
branch.control_weaves.enable_cache = True
46
self.run()
47
finally:
48
branch.unlock()
49
branch.weave_store.enable_cache = False
50
branch.control_weaves.enable_cache = False
51
52
53
def run(self):
54
branch = self.branch
55
56
57
self.checked_text_cnt = 0
58
self.checked_rev_cnt = 0
59
self.repeated_text_cnt = 0
60
self.missing_inventory_sha_cnt = 0
61
self.missing_revision_cnt = 0
62
# maps (file-id, version) -> sha1
63
self.checked_texts = {}
100
64
101
65
history = branch.revision_history()
102
66
revno = 0
103
67
revcount = len(history)
104
mismatch_inv_id = []
105
106
# for all texts checked, text_id -> sha1
107
checked_texts = {}
108
109
progress = bzrlib.ui.ui_factory.progress_bar()
110
68
69
last_rev_id = None
70
self.progress = bzrlib.ui.ui_factory.progress_bar()
111
71
for rev_id in history:
72
self.progress.update('checking revision', revno, revcount)
112
73
revno += 1
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}'
118
% rev_id)
119
120
# check the previous history entry is a parent of this entry
121
if rev.parents:
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:
128
break
129
else:
130
raise BzrCheckError("previous revision {%s} not listed among "
131
"parents of {%s}"
132
% (last_rev_id, rev_id))
133
134
for prr in rev.parents:
135
if prr.revision_sha1 is None:
136
missing_revision_sha_cnt += 1
137
continue
138
prid = prr.revision_id
139
140
try:
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",
145
prid, rev_id)
146
continue
147
148
if prr.revision_sha1 != actual_sha:
149
raise BzrCheckError("mismatched revision sha1 for "
150
"parent {%s} of {%s}: %s vs %s"
151
% (prid, rev_id,
152
prr.revision_sha1, actual_sha))
153
elif last_rev_id:
154
raise BzrCheckError("revision {%s} has no parents listed but preceded "
155
"by {%s}"
156
% (rev_id, last_rev_id))
157
158
if rev.inventory_id != rev_id:
159
mismatch_inv_id.append(rev_id)
160
161
## TODO: Check all the required fields are present on the revision.
162
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)
74
self.check_one_rev(rev_id, last_rev_id)
75
last_rev_id = rev_id
76
self.progress.clear()
77
self.report_results()
78
79
80
def report_results(self):
81
note('checked branch %s format %d',
82
self.branch.base,
83
self.branch._branch_format)
84
85
note('%6d revisions', self.checked_rev_cnt)
86
note('%6d unique file texts', self.checked_text_cnt)
87
note('%6d repeated file texts', self.repeated_text_cnt)
88
if self.missing_inventory_sha_cnt:
89
note('%d revisions are missing inventory_sha1',
90
self.missing_inventory_sha_cnt)
91
if self.missing_revision_cnt:
92
note('%d revisions are mentioned but not present',
93
self.missing_revision_cnt)
94
95
96
def check_one_rev(self, rev_id, last_rev_id):
97
"""Check one revision.
98
99
rev_id - the one to check
100
101
last_rev_id - the previous one on the mainline, if any.
102
"""
103
104
# mutter(' revision {%s}' % rev_id)
105
branch = self.branch
106
rev = branch.get_revision(rev_id)
107
if rev.revision_id != rev_id:
108
raise BzrCheckError('wrong internal revision id in revision {%s}'
109
% rev_id)
110
111
# check the previous history entry is a parent of this entry
112
if rev.parent_ids:
113
if last_rev_id is None:
114
raise BzrCheckError("revision {%s} has %d parents, but is the "
115
"start of the branch"
116
% (rev_id, len(rev.parent_ids)))
117
for parent_id in rev.parent_ids:
118
if parent_id == last_rev_id:
119
break
168
120
else:
169
missing_inventory_sha_cnt += 1
170
mutter("no inventory_sha1 on revision {%s}" % rev_id)
171
172
inv = branch.get_inventory(rev.inventory_id)
173
seen_ids = {}
174
seen_names = {}
175
176
## p('revision %d/%d file ids' % (revno, revcount))
177
for file_id in inv:
178
if file_id in seen_ids:
179
raise BzrCheckError('duplicated file_id {%s} '
180
'in inventory for revision {%s}'
181
% (file_id, rev_id))
182
seen_ids[file_id] = True
183
184
i = 0
185
for file_id in inv:
186
i += 1
187
if i & 31 == 0:
188
progress.tick()
189
190
ie = inv[file_id]
191
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))
196
197
if ie.kind == 'file':
198
if ie.text_id in checked_texts:
199
fp = checked_texts[ie.text_id]
200
else:
201
if not ie.text_id in branch.text_store:
202
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
203
204
tf = branch.text_store[ie.text_id]
205
fp = fingerprint_file(tf)
206
checked_texts[ie.text_id] = fp
207
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}'
215
% (file_id, rev_id))
216
217
progress.tick()
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}'
222
% (path, rev_id))
121
raise BzrCheckError("previous revision {%s} not listed among "
122
"parents of {%s}"
123
% (last_rev_id, rev_id))
124
elif last_rev_id:
125
raise BzrCheckError("revision {%s} has no parents listed "
126
"but preceded by {%s}"
127
% (rev_id, last_rev_id))
128
129
if rev.inventory_sha1:
130
inv_sha1 = branch.get_inventory_sha1(rev_id)
131
if inv_sha1 != rev.inventory_sha1:
132
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
133
' value in revision {%s}' % rev_id)
134
else:
135
missing_inventory_sha_cnt += 1
136
mutter("no inventory_sha1 on revision {%s}" % rev_id)
137
self._check_revision_tree(rev_id)
138
self.checked_rev_cnt += 1
139
140
def _check_revision_tree(self, rev_id):
141
tree = self.branch.revision_tree(rev_id)
142
inv = tree.inventory
143
seen_ids = {}
144
for file_id in inv:
145
if file_id in seen_ids:
146
raise BzrCheckError('duplicated file_id {%s} '
147
'in inventory for revision {%s}'
148
% (file_id, rev_id))
149
seen_ids[file_id] = True
150
for file_id in inv:
151
self._check_one_entry(rev_id, inv, tree, file_id)
152
seen_names = {}
153
for path, ie in inv.iter_entries():
154
if path in seen_names:
155
raise BzrCheckError('duplicated path %s '
156
'in inventory for revision {%s}'
157
% (path, rev_id))
223
158
seen_names[path] = True
224
last_rev_id = rev_id
225
226
finally:
227
branch.unlock()
228
229
progress.clear()
230
231
note('checked %d revisions, %d file texts' % (revcount, len(checked_texts)))
232
233
if missing_inventory_sha_cnt:
234
note('%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt)
235
236
if missing_revision_sha_cnt:
237
note('%d parent links are missing revision_sha1' % missing_revision_sha_cnt)
238
239
if missing_revision_cnt:
240
note('%d revisions are mentioned but not present' % missing_revision_cnt)
241
242
if missing_revision_cnt:
243
print '%d revisions are mentioned but not present' % missing_revision_cnt
244
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)'
250
251
if mismatch_inv_id:
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)
159
160
161
def _check_one_entry(self, rev_id, inv, tree, file_id):
162
ie = inv[file_id]
163
if ie.parent_id != None:
164
if not inv.has_id(ie.parent_id):
165
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
166
% (ie.parent_id, rev_id))
167
if ie.kind == 'file':
168
text_version = ie.text_version
169
t = (file_id, text_version)
170
if t in self.checked_texts:
171
prev_sha = self.checked_texts[t]
172
if prev_sha != ie.text_sha1:
173
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
174
(file_id, rev_id))
175
else:
176
self.repeated_text_cnt += 1
177
return
178
mutter('check version {%s} of {%s}', rev_id, file_id)
179
file_lines = tree.get_file_lines(file_id)
180
self.checked_text_cnt += 1
181
if ie.text_size != sum(map(len, file_lines)):
182
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
183
if ie.text_sha1 != sha_strings(file_lines):
184
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
185
self.checked_texts[t] = ie.text_sha1
186
elif ie.kind == 'directory':
187
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
188
raise BzrCheckError('directory {%s} has text in revision {%s}'
189
% (file_id, rev_id))
190
elif ie.kind == 'root_directory':
191
pass
192
else:
193
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
194
(ie.kind, rev_id))
195
196
197
def check(branch):
198
"""Run consistency checks on a branch."""
199
Check(branch)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1