« back to all changes in this revision
Viewing changes to bzrlib/check.py
- browse files at revision 706
- compare with another revision
- download diff
- download tarball
- view history from revision 706
- Committer: Martin Pool
- Date: 2005-06-20 03:29:21 UTC
- Revision ID: mbp@sourcefrog.net-20050620032921-240d0ade23545055
- remove redundant precursor check
- files added:
- bzr-man.py
- patches
- files removed:
- bzrlib/tests.py
- files renamed:
- contrib/bash/bzr => contrib/bash/bzr.simple
- files modified:
- .bzrignore
added
removed
bzrlib/check.py
17
17
18
18
19
19
20
######################################################################
21
# consistency checks
22
23
import sys
24
from sets import Set
25
26
from trace import mutter
27
from errors import bailout
28
import osutils
29
30
def check(branch, progress=True):
31
out = sys.stdout
32
33
# TODO: factor out
34
if not (hasattr(out, 'isatty') and out.isatty()):
35
progress=False
36
37
if progress:
38
def p(m):
39
mutter('checking ' + m)
40
out.write('\rchecking: %-50.50s' % m)
41
out.flush()
42
else:
43
def p(m):
44
mutter('checking ' + m)
45
46
p('history of %r' % branch.base)
47
last_ptr = None
48
checked_revs = Set()
49
50
history = branch.revision_history()
51
revno = 0
52
revcount = len(history)
53
54
checked_texts = {}
55
56
for rid in history:
57
revno += 1
58
p('revision %d/%d' % (revno, revcount))
59
mutter(' revision {%s}' % rid)
60
rev = branch.get_revision(rid)
61
if rev.revision_id != rid:
62
bailout('wrong internal revision id in revision {%s}' % rid)
63
if rev.precursor != last_ptr:
64
bailout('mismatched precursor in revision {%s}' % rid)
65
last_ptr = rid
66
if rid in checked_revs:
67
bailout('repeated revision {%s}' % rid)
68
checked_revs.add(rid)
69
70
## TODO: Check all the required fields are present on the revision.
71
72
inv = branch.get_inventory(rev.inventory_id)
73
seen_ids = Set()
74
seen_names = Set()
75
76
p('revision %d/%d file ids' % (revno, revcount))
77
for file_id in inv:
78
if file_id in seen_ids:
79
bailout('duplicated file_id {%s} in inventory for revision {%s}'
80
% (file_id, rid))
81
seen_ids.add(file_id)
82
83
i = 0
84
len_inv = len(inv)
85
for file_id in inv:
86
i += 1
87
if (i % 100) == 0:
88
p('revision %d/%d file text %d/%d' % (revno, revcount, i, len_inv))
89
90
ie = inv[file_id]
91
92
if ie.parent_id != None:
93
if ie.parent_id not in seen_ids:
94
bailout('missing parent {%s} in inventory for revision {%s}'
95
% (ie.parent_id, rid))
96
97
if ie.kind == 'file':
98
if ie.text_id in checked_texts:
99
fp = checked_texts[ie.text_id]
20
21
def check(branch):
22
"""Run consistency checks on a branch.
23
24
TODO: Also check non-mailine revisions mentioned as parents.
25
"""
26
from bzrlib.trace import mutter
27
from bzrlib.errors import BzrCheckError
28
from bzrlib.osutils import fingerprint_file
29
from bzrlib.progress import ProgressBar
30
31
branch.lock_read()
32
33
try:
34
pb = ProgressBar(show_spinner=True)
35
last_rev_id = None
36
37
missing_inventory_sha_cnt = 0
38
missing_revision_sha_cnt = 0
39
40
history = branch.revision_history()
41
revno = 0
42
revcount = len(history)
43
44
# for all texts checked, text_id -> sha1
45
checked_texts = {}
46
47
for rev_id in history:
48
revno += 1
49
pb.update('checking revision', revno, revcount)
50
mutter(' revision {%s}' % rev_id)
51
rev = branch.get_revision(rev_id)
52
if rev.revision_id != rev_id:
53
raise BzrCheckError('wrong internal revision id in revision {%s}'
54
% rev_id)
55
56
# check the previous history entry is a parent of this entry
57
if rev.parents:
58
if last_rev_id is None:
59
raise BzrCheckError("revision {%s} has %d parents, but is the "
60
"start of the branch"
61
% (rev_id, len(rev.parents)))
62
for prr in rev.parents:
63
if prr.revision_id == last_rev_id:
64
break
100
65
else:
101
if not ie.text_id in branch.text_store:
102
bailout('text {%s} not in text_store' % ie.text_id)
103
104
tf = branch.text_store[ie.text_id]
105
fp = osutils.fingerprint_file(tf)
106
checked_texts[ie.text_id] = fp
107
108
if ie.text_size != fp['size']:
109
bailout('text {%s} wrong size' % ie.text_id)
110
if ie.text_sha1 != fp['sha1']:
111
bailout('text {%s} wrong sha1' % ie.text_id)
112
elif ie.kind == 'directory':
113
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
114
bailout('directory {%s} has text in revision {%s}'
115
% (file_id, rid))
116
117
p('revision %d/%d file paths' % (revno, revcount))
118
for path, ie in inv.iter_entries():
119
if path in seen_names:
120
bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
121
seen_names.add(path)
122
123
124
p('done')
125
if progress:
126
print
66
raise BzrCheckError("previous revision {%s} not listed among "
67
"parents of {%s}"
68
% (last_rev_id, rev_id))
69
70
for prr in rev.parents:
71
if prr.revision_sha1 is None:
72
missing_revision_sha_cnt += 1
73
continue
74
prid = prr.revision_id
75
actual_sha = branch.get_revision_sha1(prid)
76
if prr.revision_sha1 != actual_sha:
77
raise BzrCheckError("mismatched revision sha1 for "
78
"parent {%s} of {%s}: %s vs %s"
79
% (prid, rev_id,
80
prr.revision_sha1, actual_sha))
81
elif last_rev_id:
82
raise BzrCheckError("revision {%s} has no parents listed but preceded "
83
"by {%s}"
84
% (rev_id, last_rev_id))
85
86
## TODO: Check all the required fields are present on the revision.
87
88
if rev.inventory_sha1:
89
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
90
if inv_sha1 != rev.inventory_sha1:
91
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
92
' value in revision {%s}' % rev_id)
93
else:
94
missing_inventory_sha_cnt += 1
95
mutter("no inventory_sha1 on revision {%s}" % rev_id)
96
97
inv = branch.get_inventory(rev.inventory_id)
98
seen_ids = {}
99
seen_names = {}
100
101
## p('revision %d/%d file ids' % (revno, revcount))
102
for file_id in inv:
103
if file_id in seen_ids:
104
raise BzrCheckError('duplicated file_id {%s} '
105
'in inventory for revision {%s}'
106
% (file_id, rev_id))
107
seen_ids[file_id] = True
108
109
i = 0
110
for file_id in inv:
111
i += 1
112
if i & 31 == 0:
113
pb.tick()
114
115
ie = inv[file_id]
116
117
if ie.parent_id != None:
118
if ie.parent_id not in seen_ids:
119
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
120
% (ie.parent_id, rev_id))
121
122
if ie.kind == 'file':
123
if ie.text_id in checked_texts:
124
fp = checked_texts[ie.text_id]
125
else:
126
if not ie.text_id in branch.text_store:
127
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
128
129
tf = branch.text_store[ie.text_id]
130
fp = fingerprint_file(tf)
131
checked_texts[ie.text_id] = fp
132
133
if ie.text_size != fp['size']:
134
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
135
if ie.text_sha1 != fp['sha1']:
136
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
137
elif ie.kind == 'directory':
138
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
139
raise BzrCheckError('directory {%s} has text in revision {%s}'
140
% (file_id, rev_id))
141
142
pb.tick()
143
for path, ie in inv.iter_entries():
144
if path in seen_names:
145
raise BzrCheckError('duplicated path %s '
146
'in inventory for revision {%s}'
147
% (path, rev_id))
148
seen_names[path] = True
149
last_rev_id = rev_id
150
151
finally:
152
branch.unlock()
153
154
pb.clear()
155
127
156
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
128
157
158
if missing_inventory_sha_cnt:
159
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
160
161
if missing_revision_sha_cnt:
162
print '%d parent links are missing revision_sha1' % missing_revision_sha_cnt
163
164
if (missing_inventory_sha_cnt
165
or missing_revision_sha_cnt):
166
print ' (use "bzr upgrade" to fix them)'
Loggerhead is a web-based interface for Breezy
Version: 2.0.1