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
30
from bzrlib.trace import note, warning
31
from bzrlib.osutils import rename, sha_string, fingerprint_file
32
from bzrlib.trace import mutter
33
from bzrlib.errors import BzrCheckError, NoSuchRevision
34
from bzrlib.inventory import ROOT_ID
35
from bzrlib.branch import gen_root_id
41
def __init__(self, branch):
43
self.checked_text_cnt = 0
44
self.checked_rev_cnt = 0
46
self.repeated_text_cnt = 0
47
self.missing_parent_links = {}
48
self.missing_inventory_sha_cnt = 0
49
self.missing_revision_cnt = 0
50
# maps (file-id, version) -> sha1
51
self.checked_texts = {}
54
self.branch.lock_read()
56
self.history = self.branch.revision_history()
57
if not len(self.history):
60
self.planned_revisions = self.branch.get_ancestry(self.history[-1])
61
self.planned_revisions.remove(None)
64
self.progress = bzrlib.ui.ui_factory.progress_bar()
65
while revno < len(self.planned_revisions):
66
rev_id = self.planned_revisions[revno]
67
self.progress.update('checking revision', revno,
68
len(self.planned_revisions))
70
self.check_one_rev(rev_id)
75
def report_results(self, verbose):
76
note('checked branch %s format %d',
78
self.branch._branch_format)
80
note('%6d revisions', self.checked_rev_cnt)
81
note('%6d unique file texts', self.checked_text_cnt)
82
note('%6d repeated file texts', self.repeated_text_cnt)
83
if self.missing_inventory_sha_cnt:
84
note('%6d revisions are missing inventory_sha1',
85
self.missing_inventory_sha_cnt)
86
if self.missing_revision_cnt:
87
note('%6d revisions are mentioned but not present',
88
self.missing_revision_cnt)
90
note('%6d ghost revisions', len(self.ghosts))
92
for ghost in self.ghosts:
94
if len(self.missing_parent_links):
95
note('%6d revisions missing parents in ancestry',
96
len(self.missing_parent_links))
98
for link, linkers in self.missing_parent_links.items():
99
note(' %s should be in the ancestry for:', link)
100
for linker in linkers:
101
note(' * %s', linker)
103
def check_one_rev(self, rev_id):
104
"""Check one revision.
106
rev_id - the one to check
108
last_rev_id - the previous one on the mainline, if any.
111
# mutter(' revision {%s}' % rev_id)
114
rev_history_position = self.history.index(rev_id)
116
rev_history_position = None
118
if rev_history_position:
119
rev = branch.get_revision(rev_id)
120
if rev_history_position > 0:
121
last_rev_id = self.history[rev_history_position - 1]
123
rev = branch.get_revision(rev_id)
125
if rev.revision_id != rev_id:
126
raise BzrCheckError('wrong internal revision id in revision {%s}'
129
# check the previous history entry is a parent of this entry
131
if last_rev_id is None and rev_history_position is not None:
132
# what if the start is a ghost ? i.e. conceptually the
134
raise BzrCheckError("revision {%s} has %d parents, but is the "
135
"start of the branch"
136
% (rev_id, len(rev.parent_ids)))
137
if last_rev_id is not None:
138
for parent_id in rev.parent_ids:
139
if parent_id == last_rev_id:
142
raise BzrCheckError("previous revision {%s} not listed among "
144
% (last_rev_id, rev_id))
145
for parent in rev.parent_ids:
146
if not parent in self.planned_revisions:
147
missing_links = self.missing_parent_links.get(parent, [])
148
missing_links.append(rev_id)
149
self.missing_parent_links[parent] = missing_links
150
# list based so slow, TODO have a planned_revisions list and set.
151
if self.branch.has_revision(parent):
152
missing_ancestry = self.branch.get_ancestry(parent)
153
for missing in missing_ancestry:
154
if (missing is not None
155
and missing not in self.planned_revisions):
156
self.planned_revisions.append(missing)
158
self.ghosts.append(rev_id)
160
raise BzrCheckError("revision {%s} has no parents listed "
161
"but preceded by {%s}"
162
% (rev_id, last_rev_id))
164
if rev.inventory_sha1:
165
inv_sha1 = branch.get_inventory_sha1(rev_id)
166
if inv_sha1 != rev.inventory_sha1:
167
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
168
' value in revision {%s}' % rev_id)
170
missing_inventory_sha_cnt += 1
171
mutter("no inventory_sha1 on revision {%s}" % rev_id)
172
self._check_revision_tree(rev_id)
173
self.checked_rev_cnt += 1
175
def _check_revision_tree(self, rev_id):
176
tree = self.branch.revision_tree(rev_id)
20
######################################################################
26
from trace import mutter
27
from errors import bailout
30
def check(branch, progress=True):
35
mutter('checking ' + m)
36
out.write('\rchecking: %-50.50s' % m)
40
mutter('checking ' + m)
42
p('history of %r' % branch.base)
46
history = branch.revision_history()
48
revcount = len(history)
54
p('revision %d/%d' % (revno, revcount))
55
mutter(' revision {%s}' % rid)
56
rev = branch.get_revision(rid)
57
if rev.revision_id != rid:
58
bailout('wrong internal revision id in revision {%s}' % rid)
59
if rev.precursor != last_ptr:
60
bailout('mismatched precursor in revision {%s}' % rid)
62
if rid in checked_revs:
63
bailout('repeated revision {%s}' % rid)
66
## TODO: Check all the required fields are present on the revision.
68
inv = branch.get_inventory(rev.inventory_id)
72
p('revision %d/%d file ids' % (revno, revcount))
179
73
for file_id in inv:
180
74
if file_id in seen_ids:
181
raise BzrCheckError('duplicated file_id {%s} '
182
'in inventory for revision {%s}'
184
seen_ids[file_id] = True
75
bailout('duplicated file_id {%s} in inventory for revision {%s}'
185
81
for file_id in inv:
84
p('revision %d/%d file text %d/%d' % (revno, revcount, i, len_inv))
187
ie.check(self, rev_id, inv, tree)
88
if ie.parent_id != None:
89
if ie.parent_id not in seen_ids:
90
bailout('missing parent {%s} in inventory for revision {%s}'
91
% (ie.parent_id, rid))
94
if ie.text_id in checked_texts:
95
fp = checked_texts[ie.text_id]
97
if not ie.text_id in branch.text_store:
98
bailout('text {%s} not in text_store' % ie.text_id)
100
tf = branch.text_store[ie.text_id]
101
fp = osutils.fingerprint_file(tf)
102
checked_texts[ie.text_id] = fp
104
if ie.text_size != fp['size']:
105
bailout('text {%s} wrong size' % ie.text_id)
106
if ie.text_sha1 != fp['sha1']:
107
bailout('text {%s} wrong sha1' % ie.text_id)
108
elif ie.kind == 'directory':
109
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
110
bailout('directory {%s} has text in revision {%s}'
113
p('revision %d/%d file paths' % (revno, revcount))
189
114
for path, ie in inv.iter_entries():
190
115
if path in seen_names:
191
raise BzrCheckError('duplicated path %s '
192
'in inventory for revision {%s}'
194
seen_names[path] = True
197
def check(branch, verbose):
198
"""Run consistency checks on a branch."""
199
checker = Check(branch)
201
checker.report_results(verbose)
116
bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
added
removed
bzrlib/check.py
