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.nested_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)
78
self.progress.finished()
81
def plan_revisions(self):
82
repository = self.branch.repository
83
self.planned_revisions = set(repository.all_revision_ids())
85
inventoried = set(self.inventory_weave.versions())
86
awol = self.planned_revisions - inventoried
88
raise BzrCheckError('Stored revisions missing from inventory'
89
'{%s}' % ','.join([f for f in awol]))
90
self.planned_revisions = list(self.planned_revisions)
92
def report_results(self, verbose):
93
note('checked branch %s format %s',
97
note('%6d revisions', self.checked_rev_cnt)
98
note('%6d unique file texts', self.checked_text_cnt)
99
note('%6d repeated file texts', self.repeated_text_cnt)
100
note('%6d weaves', len(self.checked_weaves))
101
if self.missing_inventory_sha_cnt:
102
note('%6d revisions are missing inventory_sha1',
103
self.missing_inventory_sha_cnt)
104
if self.missing_revision_cnt:
105
note('%6d revisions are mentioned but not present',
106
self.missing_revision_cnt)
108
note('%6d ghost revisions', len(self.ghosts))
110
for ghost in self.ghosts:
112
if len(self.missing_parent_links):
113
note('%6d revisions missing parents in ancestry',
114
len(self.missing_parent_links))
116
for link, linkers in self.missing_parent_links.items():
117
note(' %s should be in the ancestry for:', link)
118
for linker in linkers:
119
note(' * %s', linker)
121
def check_one_rev(self, rev_id):
122
"""Check one revision.
124
rev_id - the one to check
126
last_rev_id - the previous one on the mainline, if any.
129
# mutter(' revision {%s}', rev_id)
132
rev_history_position = self.history.index(rev_id)
134
rev_history_position = None
136
if rev_history_position:
137
rev = branch.repository.get_revision(rev_id)
138
if rev_history_position > 0:
139
last_rev_id = self.history[rev_history_position - 1]
141
rev = branch.repository.get_revision(rev_id)
143
if rev.revision_id != rev_id:
144
raise BzrCheckError('wrong internal revision id in revision {%s}'
147
# check the previous history entry is a parent of this entry
149
if last_rev_id is not None:
150
for parent_id in rev.parent_ids:
151
if parent_id == last_rev_id:
154
raise BzrCheckError("previous revision {%s} not listed among "
156
% (last_rev_id, rev_id))
157
for parent in rev.parent_ids:
158
if not parent in self.planned_revisions:
159
missing_links = self.missing_parent_links.get(parent, [])
160
missing_links.append(rev_id)
161
self.missing_parent_links[parent] = missing_links
162
# list based so somewhat slow,
163
# TODO have a planned_revisions list and set.
164
if self.branch.repository.has_revision(parent):
165
missing_ancestry = self.repository.get_ancestry(parent)
166
for missing in missing_ancestry:
167
if (missing is not None
168
and missing not in self.planned_revisions):
169
self.planned_revisions.append(missing)
171
self.ghosts.append(rev_id)
173
raise BzrCheckError("revision {%s} has no parents listed "
174
"but preceded by {%s}"
175
% (rev_id, last_rev_id))
177
if rev.inventory_sha1:
178
inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
179
if inv_sha1 != rev.inventory_sha1:
180
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
181
' value in revision {%s}' % rev_id)
183
missing_inventory_sha_cnt += 1
184
mutter("no inventory_sha1 on revision {%s}", rev_id)
185
self._check_revision_tree(rev_id)
186
self.checked_rev_cnt += 1
188
def check_weaves(self):
189
"""Check all the weaves we can get our hands on.
193
if self.branch.repository.weave_store.listable():
194
weave_ids = list(self.branch.repository.weave_store)
195
n_weaves = len(weave_ids)
196
self.progress.update('checking weave', 0, n_weaves)
197
self.inventory_weave.check(progress_bar=self.progress)
198
for i, weave_id in enumerate(weave_ids):
199
self.progress.update('checking weave', i, n_weaves)
200
w = self.branch.repository.weave_store.get_weave(weave_id,
201
self.branch.repository.get_transaction())
202
# No progress here, because it looks ugly.
204
self.checked_weaves[weave_id] = True
206
def _check_revision_tree(self, rev_id):
207
tree = self.branch.repository.revision_tree(rev_id)
20
######################################################################
26
from trace import mutter
27
from errors import bailout
30
def check(branch, progress=True):
34
if not (hasattr(out, 'isatty') and out.isatty()):
39
mutter('checking ' + m)
40
out.write('\rchecking: %-50.50s' % m)
44
mutter('checking ' + m)
46
p('history of %r' % branch.base)
50
history = branch.revision_history()
52
revcount = len(history)
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)
66
if rid in checked_revs:
67
bailout('repeated revision {%s}' % rid)
70
## TODO: Check all the required fields are present on the revision.
72
inv = branch.get_inventory(rev.inventory_id)
76
p('revision %d/%d file ids' % (revno, revcount))
210
77
for file_id in inv:
211
78
if file_id in seen_ids:
212
raise BzrCheckError('duplicated file_id {%s} '
213
'in inventory for revision {%s}'
215
seen_ids[file_id] = True
79
bailout('duplicated file_id {%s} in inventory for revision {%s}'
216
85
for file_id in inv:
88
p('revision %d/%d file text %d/%d' % (revno, revcount, i, len_inv))
218
ie.check(self, rev_id, inv, tree)
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))
98
if ie.text_id in checked_texts:
99
fp = checked_texts[ie.text_id]
101
if not ie.text_id in branch.text_store:
102
bailout('text {%s} not in text_store' % ie.text_id)
104
tf = branch.text_store[ie.text_id]
105
fp = osutils.fingerprint_file(tf)
106
checked_texts[ie.text_id] = fp
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}'
117
p('revision %d/%d file paths' % (revno, revcount))
220
118
for path, ie in inv.iter_entries():
221
119
if path in seen_names:
222
raise BzrCheckError('duplicated path %s '
223
'in inventory for revision {%s}'
225
seen_names[path] = True
228
def check(branch, verbose):
229
"""Run consistency checks on a branch."""
230
checker = Check(branch)
232
checker.report_results(verbose)
120
bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
127
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
added
removed
bzrlib/check.py
