25
24
# 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 import errors
36
from bzrlib import repository as _mod_repository
37
from bzrlib import revision
38
from bzrlib.errors import BzrCheckError
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
40
from bzrlib.trace import log_error, note
38
42
class Check(object):
41
def __init__(self, branch):
43
"""Check a repository"""
45
# The Check object interacts with InventoryEntry.check, etc.
47
def __init__(self, repository):
48
self.repository = repository
43
49
self.checked_text_cnt = 0
44
50
self.checked_rev_cnt = 0
47
53
self.missing_parent_links = {}
48
54
self.missing_inventory_sha_cnt = 0
49
55
self.missing_revision_cnt = 0
50
# maps (file-id, version) -> sha1
56
# maps (file-id, version) -> sha1; used by InventoryFile._check
51
57
self.checked_texts = {}
58
self.checked_weaves = {}
59
self.unreferenced_versions = set()
60
self.inconsistent_parents = []
54
self.branch.lock_read()
63
self.repository.lock_read()
64
self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
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)
66
self.progress.update('retrieving inventory', 0, 2)
67
# do not put in init, as it should be done with progess,
68
# and inside the lock.
69
self.inventory_weave = self.repository.get_inventory_weave()
70
self.progress.update('checking revision graph', 1)
71
self.check_revision_graph()
64
self.progress = bzrlib.ui.ui_factory.progress_bar()
65
74
while revno < len(self.planned_revisions):
66
75
rev_id = self.planned_revisions[revno]
67
76
self.progress.update('checking revision', revno,
68
77
len(self.planned_revisions))
70
79
self.check_one_rev(rev_id)
80
# check_weaves is done after the revision scan so that
81
# revision index is known to be valid.
84
self.progress.finished()
85
self.repository.unlock()
87
def check_revision_graph(self):
88
if not self.repository.revision_graph_can_have_wrong_parents():
89
# This check is not necessary.
90
self.revs_with_bad_parents_in_index = None
92
bad_revisions = self.repository._find_inconsistent_revision_parents()
93
self.revs_with_bad_parents_in_index = list(bad_revisions)
95
def plan_revisions(self):
96
repository = self.repository
97
self.planned_revisions = repository.all_revision_ids()
99
inventoried = set(self.inventory_weave.versions())
100
awol = set(self.planned_revisions) - inventoried
102
raise BzrCheckError('Stored revisions missing from inventory'
103
'{%s}' % ','.join([f for f in awol]))
75
105
def report_results(self, verbose):
76
note('checked branch %s format %d',
78
self.branch._branch_format)
106
note('checked repository %s format %s',
107
self.repository.bzrdir.root_transport,
108
self.repository._format)
80
109
note('%6d revisions', self.checked_rev_cnt)
110
note('%6d file-ids', len(self.checked_weaves))
81
111
note('%6d unique file texts', self.checked_text_cnt)
82
112
note('%6d repeated file texts', self.repeated_text_cnt)
113
note('%6d unreferenced text versions',
114
len(self.unreferenced_versions))
83
115
if self.missing_inventory_sha_cnt:
84
116
note('%6d revisions are missing inventory_sha1',
85
117
self.missing_inventory_sha_cnt)
92
124
for ghost in self.ghosts:
93
125
note(' %s', ghost)
94
126
if len(self.missing_parent_links):
95
note('%6d revisions missing parents in ancestry',
127
note('%6d revisions missing parents in ancestry',
96
128
len(self.missing_parent_links))
98
130
for link, linkers in self.missing_parent_links.items():
99
131
note(' %s should be in the ancestry for:', link)
100
132
for linker in linkers:
101
133
note(' * %s', linker)
135
for file_id, revision_id in self.unreferenced_versions:
136
log_error('unreferenced version: {%s} in %s', revision_id,
138
if len(self.inconsistent_parents):
139
note('%6d inconsistent parents', len(self.inconsistent_parents))
141
for info in self.inconsistent_parents:
142
revision_id, file_id, found_parents, correct_parents = info
143
note(' * %s version %s has parents %r '
145
% (file_id, revision_id, found_parents,
147
if self.revs_with_bad_parents_in_index:
148
note('%6d revisions have incorrect parents in the revision index',
149
len(self.revs_with_bad_parents_in_index))
151
for item in self.revs_with_bad_parents_in_index:
152
revision_id, index_parents, actual_parents = item
154
' %s has wrong parents in index: '
156
revision_id, index_parents, actual_parents)
103
158
def check_one_rev(self, rev_id):
104
159
"""Check one revision.
106
161
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)
163
rev = self.repository.get_revision(rev_id)
125
165
if rev.revision_id != rev_id:
126
166
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:
169
for parent in rev.parent_ids:
170
if not parent in self.planned_revisions:
171
missing_links = self.missing_parent_links.get(parent, [])
172
missing_links.append(rev_id)
173
self.missing_parent_links[parent] = missing_links
174
# list based so somewhat slow,
175
# TODO have a planned_revisions list and set.
176
if self.repository.has_revision(parent):
177
missing_ancestry = self.repository.get_ancestry(parent)
178
for missing in missing_ancestry:
179
if (missing is not None
180
and missing not in self.planned_revisions):
181
self.planned_revisions.append(missing)
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))
183
self.ghosts.append(rev_id)
164
185
if rev.inventory_sha1:
165
inv_sha1 = branch.get_inventory_sha1(rev_id)
186
inv_sha1 = self.repository.get_inventory_sha1(rev_id)
166
187
if inv_sha1 != rev.inventory_sha1:
167
188
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
168
189
' value in revision {%s}' % rev_id)
170
missing_inventory_sha_cnt += 1
171
mutter("no inventory_sha1 on revision {%s}" % rev_id)
172
190
self._check_revision_tree(rev_id)
173
191
self.checked_rev_cnt += 1
193
def check_weaves(self):
194
"""Check all the weaves we can get our hands on.
198
if self.repository.weave_store.listable():
199
weave_ids = list(self.repository.weave_store)
200
n_weaves = len(weave_ids) + 1
201
self.progress.update('checking versionedfile', 0, n_weaves)
202
self.inventory_weave.check(progress_bar=self.progress)
203
files_in_revisions = {}
204
revisions_of_files = {}
205
weave_checker = self.repository._get_versioned_file_checker()
206
for i, weave_id in enumerate(weave_ids):
207
self.progress.update('checking versionedfile', i, n_weaves)
208
w = self.repository.weave_store.get_weave(weave_id,
209
self.repository.get_transaction())
210
# No progress here, because it looks ugly.
212
result = weave_checker.check_file_version_parents(w, weave_id)
213
bad_parents, unused_versions = result
214
bad_parents = bad_parents.items()
215
for revision_id, (weave_parents, correct_parents) in bad_parents:
216
self.inconsistent_parents.append(
217
(revision_id, weave_id, weave_parents, correct_parents))
218
for revision_id in unused_versions:
219
self.unreferenced_versions.add((weave_id, revision_id))
220
self.checked_weaves[weave_id] = True
175
222
def _check_revision_tree(self, rev_id):
176
tree = self.branch.revision_tree(rev_id)
223
tree = self.repository.revision_tree(rev_id)
177
224
inv = tree.inventory
179
226
for file_id in inv:
added
removed
bzrlib/check.py
