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.checked_text_cnt = 0
45
self.checked_rev_cnt = 0
47
self.repeated_text_cnt = 0
48
self.missing_parent_links = {}
49
self.missing_inventory_sha_cnt = 0
50
self.missing_revision_cnt = 0
51
# maps (file-id, version) -> sha1
52
self.checked_texts = {}
55
self.branch.lock_read()
56
self.progress = bzrlib.ui.ui_factory.progress_bar()
58
self.progress.update('retrieving inventory', 0, 0)
59
# do not put in init, as it should be done with progess,
60
# and inside the lock.
61
self.inventory_weave = self.branch._get_inventory_weave()
62
self.history = self.branch.revision_history()
63
if not len(self.history):
68
while revno < len(self.planned_revisions):
69
rev_id = self.planned_revisions[revno]
70
self.progress.update('checking revision', revno,
71
len(self.planned_revisions))
73
self.check_one_rev(rev_id)
78
def plan_revisions(self):
79
if not self.branch.revision_store.listable():
80
self.planned_revisions = self.branch.get_ancestry(self.history[-1])
81
self.planned_revisions.remove(None)
82
# FIXME progress bars should support this more nicely.
84
print ("Checking reachable history -"
85
" for a complete check use a local branch.")
88
self.planned_revisions = set(self.branch.revision_store)
89
inventoried = set(self.inventory_weave.names())
90
awol = self.planned_revisions - inventoried
92
raise BzrCheckError('Stored revisions missing from inventory'
93
'{%s}' % ','.join([f for f in awol]))
94
self.planned_revisions = list(self.planned_revisions)
96
def report_results(self, verbose):
97
note('checked branch %s format %d',
99
self.branch._branch_format)
101
note('%6d revisions', self.checked_rev_cnt)
102
note('%6d unique file texts', self.checked_text_cnt)
103
note('%6d repeated file texts', self.repeated_text_cnt)
104
if self.missing_inventory_sha_cnt:
105
note('%6d revisions are missing inventory_sha1',
106
self.missing_inventory_sha_cnt)
107
if self.missing_revision_cnt:
108
note('%6d revisions are mentioned but not present',
109
self.missing_revision_cnt)
111
note('%6d ghost revisions', len(self.ghosts))
113
for ghost in self.ghosts:
115
if len(self.missing_parent_links):
116
note('%6d revisions missing parents in ancestry',
117
len(self.missing_parent_links))
119
for link, linkers in self.missing_parent_links.items():
120
note(' %s should be in the ancestry for:', link)
121
for linker in linkers:
122
note(' * %s', linker)
124
def check_one_rev(self, rev_id):
125
"""Check one revision.
127
rev_id - the one to check
129
last_rev_id - the previous one on the mainline, if any.
132
# mutter(' revision {%s}', rev_id)
135
rev_history_position = self.history.index(rev_id)
137
rev_history_position = None
139
if rev_history_position:
140
rev = branch.get_revision(rev_id)
141
if rev_history_position > 0:
142
last_rev_id = self.history[rev_history_position - 1]
144
rev = branch.get_revision(rev_id)
146
if rev.revision_id != rev_id:
147
raise BzrCheckError('wrong internal revision id in revision {%s}'
150
# check the previous history entry is a parent of this entry
152
if last_rev_id is not None:
153
for parent_id in rev.parent_ids:
154
if parent_id == last_rev_id:
157
raise BzrCheckError("previous revision {%s} not listed among "
159
% (last_rev_id, rev_id))
160
for parent in rev.parent_ids:
161
if not parent in self.planned_revisions:
162
missing_links = self.missing_parent_links.get(parent, [])
163
missing_links.append(rev_id)
164
self.missing_parent_links[parent] = missing_links
165
# list based so somewhat slow,
166
# TODO have a planned_revisions list and set.
167
if self.branch.has_revision(parent):
168
missing_ancestry = self.branch.get_ancestry(parent)
169
for missing in missing_ancestry:
170
if (missing is not None
171
and missing not in self.planned_revisions):
172
self.planned_revisions.append(missing)
174
self.ghosts.append(rev_id)
176
raise BzrCheckError("revision {%s} has no parents listed "
177
"but preceded by {%s}"
178
% (rev_id, last_rev_id))
180
if rev.inventory_sha1:
181
inv_sha1 = branch.get_inventory_sha1(rev_id)
182
if inv_sha1 != rev.inventory_sha1:
183
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
184
' value in revision {%s}' % rev_id)
186
missing_inventory_sha_cnt += 1
187
mutter("no inventory_sha1 on revision {%s}", rev_id)
188
self._check_revision_tree(rev_id)
189
self.checked_rev_cnt += 1
191
def _check_revision_tree(self, rev_id):
192
tree = self.branch.revision_tree(rev_id)
20
######################################################################
25
from trace import mutter
26
from errors import bailout
29
def check(branch, progress=True):
33
if not (hasattr(out, 'isatty') and out.isatty()):
38
mutter('checking ' + m)
39
out.write('\rchecking: %-50.50s' % m)
43
mutter('checking ' + m)
45
p('history of %r' % branch.base)
49
history = branch.revision_history()
51
revcount = len(history)
57
p('revision %d/%d' % (revno, revcount))
58
mutter(' revision {%s}' % rid)
59
rev = branch.get_revision(rid)
60
if rev.revision_id != rid:
61
bailout('wrong internal revision id in revision {%s}' % rid)
62
if rev.precursor != last_ptr:
63
bailout('mismatched precursor in revision {%s}' % rid)
65
if rid in checked_revs:
66
bailout('repeated revision {%s}' % rid)
67
checked_revs[rid] = True
69
## TODO: Check all the required fields are present on the revision.
71
inv = branch.get_inventory(rev.inventory_id)
75
p('revision %d/%d file ids' % (revno, revcount))
195
76
for file_id in inv:
196
77
if file_id in seen_ids:
197
raise BzrCheckError('duplicated file_id {%s} '
198
'in inventory for revision {%s}'
78
bailout('duplicated file_id {%s} in inventory for revision {%s}'
200
80
seen_ids[file_id] = True
201
84
for file_id in inv:
87
p('revision %d/%d file text %d/%d' % (revno, revcount, i, len_inv))
203
ie.check(self, rev_id, inv, tree)
91
if ie.parent_id != None:
92
if ie.parent_id not in seen_ids:
93
bailout('missing parent {%s} in inventory for revision {%s}'
94
% (ie.parent_id, rid))
97
if ie.text_id in checked_texts:
98
fp = checked_texts[ie.text_id]
100
if not ie.text_id in branch.text_store:
101
bailout('text {%s} not in text_store' % ie.text_id)
103
tf = branch.text_store[ie.text_id]
104
fp = osutils.fingerprint_file(tf)
105
checked_texts[ie.text_id] = fp
107
if ie.text_size != fp['size']:
108
bailout('text {%s} wrong size' % ie.text_id)
109
if ie.text_sha1 != fp['sha1']:
110
bailout('text {%s} wrong sha1' % ie.text_id)
111
elif ie.kind == 'directory':
112
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
113
bailout('directory {%s} has text in revision {%s}'
116
p('revision %d/%d file paths' % (revno, revcount))
205
117
for path, ie in inv.iter_entries():
206
118
if path in seen_names:
207
raise BzrCheckError('duplicated path %s '
208
'in inventory for revision {%s}'
119
bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
210
120
seen_names[path] = True
213
def check(branch, verbose):
214
"""Run consistency checks on a branch."""
215
checker = Check(branch)
217
checker.report_results(verbose)
126
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
added
removed
bzrlib/check.py
