24
25
# TODO: Check revision, inventory and entry objects have all
27
# TODO: Get every revision in the revision-store even if they're not
28
# referenced by history and make sure they're all valid.
30
# TODO: Perhaps have a way to record errors other than by raising exceptions;
31
# would perhaps be enough to accumulate exception objects in a list without
32
# raising them. If there's more than one exception it'd be good to see them
35
from bzrlib.errors import BzrCheckError
37
from bzrlib.trace import note
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
39
38
class Check(object):
40
"""Check a repository"""
42
# The Check object interacts with InventoryEntry.check, etc.
44
def __init__(self, repository):
45
self.repository = repository
46
self.checked_text_cnt = 0
47
self.checked_rev_cnt = 0
49
self.repeated_text_cnt = 0
50
self.missing_parent_links = {}
40
def __init__(self, branch):
48
self.checked_text_cnt = 0
49
self.checked_rev_cnt = 0
51
50
self.missing_inventory_sha_cnt = 0
52
51
self.missing_revision_cnt = 0
53
# maps (file-id, version) -> sha1; used by InventoryFile._check
54
self.checked_texts = {}
55
self.checked_weaves = {}
58
self.repository.lock_read()
59
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
61
self.progress.update('retrieving inventory', 0, 0)
62
# do not put in init, as it should be done with progess,
63
# and inside the lock.
64
self.inventory_weave = self.repository.get_inventory_weave()
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)
75
self.progress.finished()
76
self.repository.unlock()
78
def plan_revisions(self):
79
repository = self.repository
80
self.planned_revisions = set(repository.all_revision_ids())
53
history = branch.revision_history()
55
revcount = len(history)
58
self.progress = bzrlib.ui.ui_factory.progress_bar()
59
for rev_id in history:
60
self.progress.update('checking revision', revno, revcount)
62
self.check_one_rev(rev_id, last_rev_id)
81
64
self.progress.clear()
82
inventoried = set(self.inventory_weave.versions())
83
awol = self.planned_revisions - inventoried
85
raise BzrCheckError('Stored revisions missing from inventory'
86
'{%s}' % ','.join([f for f in awol]))
87
self.planned_revisions = list(self.planned_revisions)
89
def report_results(self, verbose):
90
note('checked repository %s format %s',
91
self.repository.bzrdir.root_transport,
92
self.repository._format)
93
note('%6d revisions', self.checked_rev_cnt)
94
note('%6d unique file texts', self.checked_text_cnt)
95
note('%6d repeated file texts', self.repeated_text_cnt)
96
note('%6d weaves', len(self.checked_weaves))
68
def report_results(self):
69
note('checked branch %s format %d',
71
self.branch._branch_format)
73
note('checked %d revisions, %d file texts',
75
self.checked_text_cnt)
97
77
if self.missing_inventory_sha_cnt:
98
note('%6d revisions are missing inventory_sha1',
99
self.missing_inventory_sha_cnt)
78
note('%d revisions are missing inventory_sha1',
79
self.missing_inventory_sha_cnt)
100
81
if self.missing_revision_cnt:
101
note('%6d revisions are mentioned but not present',
102
self.missing_revision_cnt)
104
note('%6d ghost revisions', len(self.ghosts))
106
for ghost in self.ghosts:
108
if len(self.missing_parent_links):
109
note('%6d revisions missing parents in ancestry',
110
len(self.missing_parent_links))
112
for link, linkers in self.missing_parent_links.items():
113
note(' %s should be in the ancestry for:', link)
114
for linker in linkers:
115
note(' * %s', linker)
117
def check_one_rev(self, rev_id):
118
"""Check one revision.
120
rev_id - the one to check
122
rev = self.repository.get_revision(rev_id)
124
if rev.revision_id != rev_id:
125
raise BzrCheckError('wrong internal revision id in revision {%s}'
128
for parent in rev.parent_ids:
129
if not parent in self.planned_revisions:
130
missing_links = self.missing_parent_links.get(parent, [])
131
missing_links.append(rev_id)
132
self.missing_parent_links[parent] = missing_links
133
# list based so somewhat slow,
134
# TODO have a planned_revisions list and set.
135
if self.repository.has_revision(parent):
136
missing_ancestry = self.repository.get_ancestry(parent)
137
for missing in missing_ancestry:
138
if (missing is not None
139
and missing not in self.planned_revisions):
140
self.planned_revisions.append(missing)
142
self.ghosts.append(rev_id)
144
if rev.inventory_sha1:
145
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
146
if inv_sha1 != rev.inventory_sha1:
147
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
148
' value in revision {%s}' % rev_id)
149
self._check_revision_tree(rev_id)
150
self.checked_rev_cnt += 1
152
def check_weaves(self):
153
"""Check all the weaves we can get our hands on.
157
if self.repository.weave_store.listable():
158
weave_ids = list(self.repository.weave_store)
159
n_weaves = len(weave_ids)
160
self.progress.update('checking weave', 0, n_weaves)
161
self.inventory_weave.check(progress_bar=self.progress)
162
for i, weave_id in enumerate(weave_ids):
163
self.progress.update('checking weave', i, n_weaves)
164
w = self.repository.weave_store.get_weave(weave_id,
165
self.repository.get_transaction())
166
# No progress here, because it looks ugly.
168
self.checked_weaves[weave_id] = True
82
note('%d revisions are mentioned but not present',
83
self.missing_revision_cnt)
86
def check_one_rev(self, rev_id, last_rev_id):
87
"""Check one revision.
89
rev_id - the one to check
91
last_rev_id - the previous one on the mainline, if any.
94
# mutter(' revision {%s}' % rev_id)
96
rev = branch.get_revision(rev_id)
97
if rev.revision_id != rev_id:
98
raise BzrCheckError('wrong internal revision id in revision {%s}'
101
# check the previous history entry is a parent of this entry
103
if last_rev_id is None:
104
raise BzrCheckError("revision {%s} has %d parents, but is the "
105
"start of the branch"
106
% (rev_id, len(rev.parent_ids)))
107
for parent_id in rev.parent_ids:
108
if parent_id == last_rev_id:
111
raise BzrCheckError("previous revision {%s} not listed among "
113
% (last_rev_id, rev_id))
115
raise BzrCheckError("revision {%s} has no parents listed "
116
"but preceded by {%s}"
117
% (rev_id, last_rev_id))
119
if rev.inventory_sha1:
120
inv_sha1 = branch.get_inventory_sha1(rev_id)
121
if inv_sha1 != rev.inventory_sha1:
122
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
123
' value in revision {%s}' % rev_id)
125
missing_inventory_sha_cnt += 1
126
mutter("no inventory_sha1 on revision {%s}" % rev_id)
128
self._check_revision_tree(rev_id)
170
130
def _check_revision_tree(self, rev_id):
171
tree = self.repository.revision_tree(rev_id)
175
if file_id in seen_ids:
176
raise BzrCheckError('duplicated file_id {%s} '
177
'in inventory for revision {%s}'
179
seen_ids[file_id] = True
182
ie.check(self, rev_id, inv, tree)
184
for path, ie in inv.iter_entries():
185
if path in seen_names:
186
raise BzrCheckError('duplicated path %s '
187
'in inventory for revision {%s}'
189
seen_names[path] = True
192
def check(branch, verbose):
193
"""Run consistency checks on a branch.
195
Results are reported through logging.
197
:raise BzrCheckError: if there's a consistency error.
201
branch_result = branch.check()
202
repo_result = branch.repository.check([branch.last_revision()])
205
branch_result.report_results(verbose)
206
repo_result.report_results(verbose)
131
tree = self.branch.revision_tree(rev_id)
135
if file_id in seen_ids:
136
raise BzrCheckError('duplicated file_id {%s} '
137
'in inventory for revision {%s}'
139
seen_ids[file_id] = True
141
self._check_one_entry(rev_id, inv, tree, file_id)
143
for path, ie in inv.iter_entries():
144
if path in seen_names:
145
raise BzrCheckError('duplicated path %s '
146
'in inventory for revision {%s}'
148
seen_names[path] = True
151
def _check_one_entry(self, rev_id, inv, tree, file_id):
153
if ie.parent_id != None:
154
if not inv.has_id(ie.parent_id):
155
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
156
% (ie.parent_id, rev_id))
157
if ie.kind == 'file':
158
text = tree.get_file_text(file_id)
159
self.checked_text_cnt += 1
160
if ie.text_size != len(text):
161
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
162
if ie.text_sha1 != sha_string(text):
163
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
164
elif ie.kind == 'directory':
165
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
166
raise BzrCheckError('directory {%s} has text in revision {%s}'
168
elif ie.kind == 'root_directory':
171
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
176
"""Run consistency checks on a branch."""
added
removed
bzrlib/check.py
