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
18
# TODO: Check ancestries are correct for every revision: includes
19
# every committed so far, and in a reasonable order.
21
# TODO: Also check non-mainline revisions mentioned as parents.
23
# TODO: Check for extra files in the control directory.
25
# TODO: Check revision, inventory and entry objects have all
28
# TODO: Get every revision in the revision-store even if they're not
29
# referenced by history and make sure they're all valid.
32
from bzrlib.trace import note, warning
33
from bzrlib.osutils import rename, sha_string, fingerprint_file
34
from bzrlib.trace import mutter
35
from bzrlib.errors import BzrCheckError, NoSuchRevision
36
from bzrlib.inventory import ROOT_ID
42
def __init__(self, branch):
44
self.repository = branch.repository
45
self.checked_text_cnt = 0
46
self.checked_rev_cnt = 0
48
self.repeated_text_cnt = 0
49
self.missing_parent_links = {}
50
self.missing_inventory_sha_cnt = 0
51
self.missing_revision_cnt = 0
52
# maps (file-id, version) -> sha1
53
self.checked_texts = {}
54
self.checked_weaves = {}
57
self.branch.lock_read()
58
self.progress = bzrlib.ui.ui_factory.progress_bar()
60
self.progress.update('retrieving inventory', 0, 0)
61
# do not put in init, as it should be done with progess,
62
# and inside the lock.
63
self.inventory_weave = self.branch.repository.get_inventory_weave()
64
self.history = self.branch.revision_history()
65
if not len(self.history):
71
while revno < len(self.planned_revisions):
72
rev_id = self.planned_revisions[revno]
73
self.progress.update('checking revision', revno,
74
len(self.planned_revisions))
76
self.check_one_rev(rev_id)
81
def plan_revisions(self):
82
repository = self.branch.repository
83
if not repository.revision_store.listable():
84
self.planned_revisions = repository.get_ancestry(self.history[-1])
85
self.planned_revisions.remove(None)
86
# FIXME progress bars should support this more nicely.
88
print ("Checking reachable history -"
89
" for a complete check use a local branch.")
92
self.planned_revisions = set(repository.revision_store)
93
inventoried = set(self.inventory_weave.names())
94
awol = self.planned_revisions - inventoried
96
raise BzrCheckError('Stored revisions missing from inventory'
97
'{%s}' % ','.join([f for f in awol]))
98
self.planned_revisions = list(self.planned_revisions)
100
def report_results(self, verbose):
101
note('checked branch %s format %s',
105
note('%6d revisions', self.checked_rev_cnt)
106
note('%6d unique file texts', self.checked_text_cnt)
107
note('%6d repeated file texts', self.repeated_text_cnt)
108
note('%6d weaves', len(self.checked_weaves))
109
if self.missing_inventory_sha_cnt:
110
note('%6d revisions are missing inventory_sha1',
111
self.missing_inventory_sha_cnt)
112
if self.missing_revision_cnt:
113
note('%6d revisions are mentioned but not present',
114
self.missing_revision_cnt)
116
note('%6d ghost revisions', len(self.ghosts))
118
for ghost in self.ghosts:
120
if len(self.missing_parent_links):
121
note('%6d revisions missing parents in ancestry',
122
len(self.missing_parent_links))
124
for link, linkers in self.missing_parent_links.items():
125
note(' %s should be in the ancestry for:', link)
126
for linker in linkers:
127
note(' * %s', linker)
129
def check_one_rev(self, rev_id):
130
"""Check one revision.
132
rev_id - the one to check
134
last_rev_id - the previous one on the mainline, if any.
137
# mutter(' revision {%s}', rev_id)
140
rev_history_position = self.history.index(rev_id)
142
rev_history_position = None
20
def _update_store_entry(obj, obj_id, branch, store_name, store):
21
"""This is just a meta-function, which handles both revision entries
22
and inventory entries.
24
from bzrlib.trace import mutter
25
import tempfile, os, errno
26
from osutils import rename
27
obj_tmp = tempfile.TemporaryFile()
28
obj.write_xml(obj_tmp)
31
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
32
dir=branch.controlfilename(store_name))
35
orig_obj_path = branch.controlfilename([store_name, obj_id+'.gz'])
36
# Remove the old entry out of the way
37
rename(orig_obj_path, tmp_path)
39
# TODO: We may need to handle the case where the old
40
# entry was not compressed (and thus did not end with .gz)
42
store.add(obj_tmp, obj_id) # Add the new one
43
os.remove(tmp_path) # Remove the old name
44
mutter(' Updated %s entry {%s}' % (store_name, obj_id))
46
# On any exception, restore the old entry
47
rename(tmp_path, orig_obj_path)
50
if os.path.exists(tmp_path):
51
# Unfortunately, the next command might throw
52
# an exception, which will mask a previous exception.
56
def _update_revision_entry(rev, branch):
57
"""After updating the values in a revision, make sure to
58
write out the data, but try to do it in an atomic manner.
60
:param rev: The Revision object to store
61
:param branch: The Branch object where this Revision is to be stored.
63
_update_store_entry(rev, rev.revision_id, branch,
64
'revision-store', branch.revision_store)
66
def _update_inventory_entry(inv, inv_id, branch):
67
"""When an inventory has been modified (such as by adding a unique tree root)
68
this atomically re-generates the file.
70
:param inv: The Inventory
71
:param inv_id: The inventory id for this inventory
72
:param branch: The Branch where this entry will be stored.
74
_update_store_entry(inv, inv_id, branch,
75
'inventory-store', branch.inventory_store)
79
"""Run consistency checks on a branch.
81
TODO: Also check non-mainline revisions mentioned as parents.
83
TODO: Check for extra files in the control directory.
85
from bzrlib.trace import mutter
86
from bzrlib.errors import BzrCheckError
87
from bzrlib.osutils import fingerprint_file
88
from bzrlib.inventory import ROOT_ID
89
from bzrlib.branch import gen_root_id
143
94
last_rev_id = None
144
if rev_history_position:
145
rev = branch.repository.get_revision(rev_id)
146
if rev_history_position > 0:
147
last_rev_id = self.history[rev_history_position - 1]
149
rev = branch.repository.get_revision(rev_id)
151
if rev.revision_id != rev_id:
152
raise BzrCheckError('wrong internal revision id in revision {%s}'
155
# check the previous history entry is a parent of this entry
157
if last_rev_id is not None:
158
for parent_id in rev.parent_ids:
159
if parent_id == last_rev_id:
96
missing_inventory_sha_cnt = 0
97
missing_revision_sha_cnt = 0
99
history = branch.revision_history()
101
revcount = len(history)
104
# for all texts checked, text_id -> sha1
107
progress = bzrlib.ui.ui_factory.progress_bar()
109
for rev_id in history:
111
progress.update('checking revision', revno, revcount)
112
# mutter(' revision {%s}' % rev_id)
113
rev = branch.get_revision(rev_id)
114
if rev.revision_id != rev_id:
115
raise BzrCheckError('wrong internal revision id in revision {%s}'
118
# check the previous history entry is a parent of this entry
120
if last_rev_id is None:
121
raise BzrCheckError("revision {%s} has %d parents, but is the "
122
"start of the branch"
123
% (rev_id, len(rev.parents)))
124
for prr in rev.parents:
125
if prr.revision_id == last_rev_id:
162
128
raise BzrCheckError("previous revision {%s} not listed among "
163
129
"parents of {%s}"
164
130
% (last_rev_id, rev_id))
165
for parent in rev.parent_ids:
166
if not parent in self.planned_revisions:
167
missing_links = self.missing_parent_links.get(parent, [])
168
missing_links.append(rev_id)
169
self.missing_parent_links[parent] = missing_links
170
# list based so somewhat slow,
171
# TODO have a planned_revisions list and set.
172
if self.branch.has_revision(parent):
173
missing_ancestry = self.repository.get_ancestry(parent)
174
for missing in missing_ancestry:
175
if (missing is not None
176
and missing not in self.planned_revisions):
177
self.planned_revisions.append(missing)
132
for prr in rev.parents:
133
if prr.revision_sha1 is None:
134
missing_revision_sha_cnt += 1
136
prid = prr.revision_id
137
actual_sha = branch.get_revision_sha1(prid)
138
if prr.revision_sha1 != actual_sha:
139
raise BzrCheckError("mismatched revision sha1 for "
140
"parent {%s} of {%s}: %s vs %s"
142
prr.revision_sha1, actual_sha))
144
raise BzrCheckError("revision {%s} has no parents listed but preceded "
146
% (rev_id, last_rev_id))
148
if rev.inventory_id != rev_id:
149
mismatch_inv_id.append(rev_id)
151
## TODO: Check all the required fields are present on the revision.
153
if rev.inventory_sha1:
154
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
155
if inv_sha1 != rev.inventory_sha1:
156
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
157
' value in revision {%s}' % rev_id)
159
missing_inventory_sha_cnt += 1
160
mutter("no inventory_sha1 on revision {%s}" % rev_id)
162
inv = branch.get_inventory(rev.inventory_id)
166
## p('revision %d/%d file ids' % (revno, revcount))
168
if file_id in seen_ids:
169
raise BzrCheckError('duplicated file_id {%s} '
170
'in inventory for revision {%s}'
172
seen_ids[file_id] = True
182
if ie.parent_id != None:
183
if ie.parent_id not in seen_ids:
184
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
185
% (ie.parent_id, rev_id))
187
if ie.kind == 'file':
188
if ie.text_id in checked_texts:
189
fp = checked_texts[ie.text_id]
179
self.ghosts.append(rev_id)
181
raise BzrCheckError("revision {%s} has no parents listed "
182
"but preceded by {%s}"
183
% (rev_id, last_rev_id))
185
if rev.inventory_sha1:
186
inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
187
if inv_sha1 != rev.inventory_sha1:
188
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
189
' value in revision {%s}' % rev_id)
191
missing_inventory_sha_cnt += 1
192
mutter("no inventory_sha1 on revision {%s}", rev_id)
193
self._check_revision_tree(rev_id)
194
self.checked_rev_cnt += 1
196
def check_weaves(self):
197
"""Check all the weaves we can get our hands on.
201
if self.branch.repository.weave_store.listable():
202
weave_ids = list(self.branch.repository.weave_store)
203
n_weaves = len(weave_ids)
204
self.progress.update('checking weave', 0, n_weaves)
205
self.inventory_weave.check(progress_bar=self.progress)
206
for i, weave_id in enumerate(weave_ids):
207
self.progress.update('checking weave', i, n_weaves)
208
w = self.branch.repository.weave_store.get_weave(weave_id,
209
self.branch.repository.get_transaction())
210
# No progress here, because it looks ugly.
212
self.checked_weaves[weave_id] = True
214
def _check_revision_tree(self, rev_id):
215
tree = self.branch.repository.revision_tree(rev_id)
219
if file_id in seen_ids:
220
raise BzrCheckError('duplicated file_id {%s} '
221
'in inventory for revision {%s}'
223
seen_ids[file_id] = True
226
ie.check(self, rev_id, inv, tree)
228
for path, ie in inv.iter_entries():
229
if path in seen_names:
230
raise BzrCheckError('duplicated path %s '
231
'in inventory for revision {%s}'
191
if not ie.text_id in branch.text_store:
192
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
194
tf = branch.text_store[ie.text_id]
195
fp = fingerprint_file(tf)
196
checked_texts[ie.text_id] = fp
198
if ie.text_size != fp['size']:
199
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
200
if ie.text_sha1 != fp['sha1']:
201
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
202
elif ie.kind == 'directory':
203
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
204
raise BzrCheckError('directory {%s} has text in revision {%s}'
208
for path, ie in inv.iter_entries():
209
if path in seen_names:
210
raise BzrCheckError('duplicated path %s '
211
'in inventory for revision {%s}'
233
213
seen_names[path] = True
236
def check(branch, verbose):
237
"""Run consistency checks on a branch."""
238
checker = Check(branch)
240
checker.report_results(verbose)
221
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
223
if missing_inventory_sha_cnt:
224
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
226
if missing_revision_sha_cnt:
227
print '%d parent links are missing revision_sha1' % missing_revision_sha_cnt
229
# stub this out for now because the main bzr branch has references
230
# to revisions that aren't present in the store -- mbp 20050804
231
# if (missing_inventory_sha_cnt
232
# or missing_revision_sha_cnt):
233
# print ' (use "bzr upgrade" to fix them)'
236
print '%d revisions have mismatched inventory ids:' % len(mismatch_inv_id)
237
for rev_id in mismatch_inv_id:
added
removed
bzrlib/check.py
