« back to all changes in this revision
Viewing changes to bzrlib/check.py
- browse files at revision 958
- compare with another revision
- download diff
- download tarball
- view history from revision 958
- Committer: Martin Pool
- Date: 2005-07-22 19:44:09 UTC
- Revision ID: mbp@sourcefrog.net-20050722194409-9f99ffe3d62cf8c0
- tidy up imports to use full module names
- files added:
- bzrlib/hashcache.py
- bzrlib/selftest
- effbot
- effbot/org
- plugins
- plugins/changeset
- tools
- files removed:
- bzrlib/statcache.py
- files renamed:
- bzrlib/selftest.py => bzrlib/selftest/__init__.py
- bzrlib/whitebox.py => bzrlib/selftest/whitebox.py
- files modified:
- .bzrignore
added
removed
bzrlib/check.py
16
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
17
18
18
19
20
21
def check(branch, update=False):
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.
22
"""
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)
28
obj_tmp.seek(0)
29
30
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
31
dir=branch.controlfilename(store_name))
32
os.close(tmpfd)
33
try:
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)
37
try:
38
# TODO: We may need to handle the case where the old
39
# entry was not compressed (and thus did not end with .gz)
40
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))
44
except:
45
# On any exception, restore the old entry
46
rename(tmp_path, orig_obj_path)
47
raise
48
finally:
49
if os.path.exists(tmp_path):
50
# Unfortunately, the next command might throw
51
# an exception, which will mask a previous exception.
52
os.remove(tmp_path)
53
obj_tmp.close()
54
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.
58
59
:param rev: The Revision object to store
60
:param branch: The Branch object where this Revision is to be stored.
61
"""
62
_update_store_entry(rev, rev.revision_id, branch,
63
'revision-store', branch.revision_store)
64
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.
68
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.
72
"""
73
_update_store_entry(inv, inv_id, branch,
74
'inventory-store', branch.inventory_store)
75
76
def check(branch):
22
77
"""Run consistency checks on a branch.
23
78
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.
79
TODO: Also check non-mainline revisions mentioned as parents.
80
81
TODO: Check for extra files in the control directory.
27
82
"""
28
import sys
29
30
83
from bzrlib.trace import mutter
31
84
from bzrlib.errors import BzrCheckError
32
85
from bzrlib.osutils import fingerprint_file
33
86
from bzrlib.progress import ProgressBar
87
from bzrlib.inventory import ROOT_ID
88
from bzrlib.branch import gen_root_id
34
89
35
if update:
36
branch.lock_write()
37
else:
38
branch.lock_read()
90
branch.lock_read()
39
91
40
92
try:
41
42
out = sys.stdout
43
44
93
pb = ProgressBar(show_spinner=True)
45
last_ptr = None
46
checked_revs = {}
94
last_rev_id = None
47
95
48
96
missing_inventory_sha_cnt = 0
97
missing_revision_sha_cnt = 0
49
98
50
99
history = branch.revision_history()
51
100
revno = 0
52
101
revcount = len(history)
102
mismatch_inv_id = []
53
103
104
# for all texts checked, text_id -> sha1
54
105
checked_texts = {}
55
106
56
updated_revisions = []
57
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
61
62
for rid in history:
107
for rev_id in history:
63
108
revno += 1
64
109
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)
71
last_ptr = rid
72
if rid in checked_revs:
73
raise BzrCheckError('repeated revision {%s}' % rid)
74
checked_revs[rid] = True
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}'
114
% rev_id)
115
116
# check the previous history entry is a parent of this entry
117
if rev.parents:
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:
124
break
125
else:
126
raise BzrCheckError("previous revision {%s} not listed among "
127
"parents of {%s}"
128
% (last_rev_id, rev_id))
129
130
for prr in rev.parents:
131
if prr.revision_sha1 is None:
132
missing_revision_sha_cnt += 1
133
continue
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"
139
% (prid, rev_id,
140
prr.revision_sha1, actual_sha))
141
elif last_rev_id:
142
raise BzrCheckError("revision {%s} has no parents listed but preceded "
143
"by {%s}"
144
% (rev_id, last_rev_id))
145
146
if rev.inventory_id != rev_id:
147
mismatch_inv_id.append(rev_id)
75
148
76
149
## TODO: Check all the required fields are present on the revision.
77
150
78
updated = False
79
151
if rev.inventory_sha1:
80
#mutter(' checking inventory hash {%s}' % rev.inventory_sha1)
81
152
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
82
153
if inv_sha1 != rev.inventory_sha1:
83
154
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
84
' value in revision {%s}' % rid)
85
elif update:
86
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
87
rev.inventory_sha1 = inv_sha1
88
updated = True
155
' value in revision {%s}' % rev_id)
89
156
else:
90
157
missing_inventory_sha_cnt += 1
91
mutter("no inventory_sha1 on revision {%s}" % rid)
92
93
if rev.precursor:
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
100
updated = True
101
else:
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)
106
elif update:
107
precursor_sha1 = branch.get_revision_sha1(rev.precursor)
108
rev.precursor_sha1 = precursor_sha1
109
updated = True
110
111
if updated:
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)
122
rev_tmp.seek(0)
123
124
tmpfd, tmp_path = tempfile.mkstemp(prefix=rid, suffix='.gz',
125
dir=branch.controlfilename('revision-store'))
126
os.close(tmpfd)
127
def special_rename(p1, p2):
128
if sys.platform == 'win32':
129
try:
130
os.remove(p2)
131
except OSError, e:
132
if e.errno != e.ENOENT:
133
raise
134
os.rename(p1, p2)
135
136
try:
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)
139
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)
146
except:
147
# On any exception, restore the old entry
148
special_rename(tmp_path, rev_path)
149
raise
150
rev_tmp.close()
151
updated_revisions.append(rid)
152
else:
153
updated_previous_revision = False
158
mutter("no inventory_sha1 on revision {%s}" % rev_id)
154
159
155
160
inv = branch.get_inventory(rev.inventory_id)
156
161
seen_ids = {}
161
166
if file_id in seen_ids:
162
167
raise BzrCheckError('duplicated file_id {%s} '
163
168
'in inventory for revision {%s}'
164
% (file_id, rid))
169
% (file_id, rev_id))
165
170
seen_ids[file_id] = True
166
171
167
172
i = 0
168
len_inv = len(inv)
169
173
for file_id in inv:
170
174
i += 1
171
175
if i & 31 == 0:
176
180
if ie.parent_id != None:
177
181
if ie.parent_id not in seen_ids:
178
182
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
179
% (ie.parent_id, rid))
183
% (ie.parent_id, rev_id))
180
184
181
185
if ie.kind == 'file':
182
186
if ie.text_id in checked_texts:
196
200
elif ie.kind == 'directory':
197
201
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
198
202
raise BzrCheckError('directory {%s} has text in revision {%s}'
199
% (file_id, rid))
203
% (file_id, rev_id))
200
204
201
205
pb.tick()
202
206
for path, ie in inv.iter_entries():
203
207
if path in seen_names:
204
raise BzrCheckError('duplicated path %r '
208
raise BzrCheckError('duplicated path %s '
205
209
'in inventory for revision {%s}'
206
% (path, revid))
210
% (path, rev_id))
207
211
seen_names[path] = True
212
last_rev_id = rev_id
208
213
209
214
finally:
210
215
branch.unlock()
212
217
pb.clear()
213
218
214
219
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)
220
217
221
if missing_inventory_sha_cnt:
218
222
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
219
print ' (use bzr check --update to fix them)'
223
224
if missing_revision_sha_cnt:
225
print '%d parent links are missing revision_sha1' % missing_revision_sha_cnt
226
227
if (missing_inventory_sha_cnt
228
or missing_revision_sha_cnt):
229
print ' (use "bzr upgrade" to fix them)'
230
231
if mismatch_inv_id:
232
print '%d revisions have mismatched inventory ids:' % len(mismatch_inv_id)
233
for rev_id in mismatch_inv_id:
234
print ' ', rev_id
Loggerhead is a web-based interface for Breezy
Version: 2.0.1