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
19
def _update_store_entry(obj, obj_id, branch, store_name, store):
20
"""This is just a meta-function, which handles both revision entries
21
and inventory entries.
23
from bzrlib.trace import mutter
24
import tempfile, os, errno
25
from osutils import rename
26
obj_tmp = tempfile.TemporaryFile()
27
obj.write_xml(obj_tmp)
30
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
31
dir=branch.controlfilename(store_name))
34
orig_obj_path = branch.controlfilename([store_name, obj_id+'.gz'])
35
# Remove the old entry out of the way
36
rename(orig_obj_path, tmp_path)
38
# TODO: We may need to handle the case where the old
39
# entry was not compressed (and thus did not end with .gz)
41
store.add(obj_tmp, obj_id) # Add the new one
42
os.remove(tmp_path) # Remove the old name
43
mutter(' Updated %s entry {%s}' % (store_name, obj_id))
45
# On any exception, restore the old entry
46
rename(tmp_path, orig_obj_path)
49
if os.path.exists(tmp_path):
50
# Unfortunately, the next command might throw
51
# an exception, which will mask a previous exception.
55
def _update_revision_entry(rev, branch):
56
"""After updating the values in a revision, make sure to
57
write out the data, but try to do it in an atomic manner.
59
:param rev: The Revision object to store
60
:param branch: The Branch object where this Revision is to be stored.
62
_update_store_entry(rev, rev.revision_id, branch,
63
'revision-store', branch.revision_store)
65
def _update_inventory_entry(inv, inv_id, branch):
66
"""When an inventory has been modified (such as by adding a unique tree root)
67
this atomically re-generates the file.
69
:param inv: The Inventory
70
:param inv_id: The inventory id for this inventory
71
:param branch: The Branch where this entry will be stored.
73
_update_store_entry(inv, inv_id, branch,
74
'inventory-store', branch.inventory_store)
77
"""Run consistency checks on a branch.
79
TODO: Also check non-mainline revisions mentioned as parents.
81
TODO: Check for extra files in the control directory.
83
from bzrlib.trace import mutter
84
from bzrlib.errors import BzrCheckError
85
from bzrlib.osutils import fingerprint_file
86
from bzrlib.progress import ProgressBar
87
from bzrlib.inventory import ROOT_ID
88
from bzrlib.branch import gen_root_id
93
pb = ProgressBar(show_spinner=True)
138
94
last_rev_id = 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:
96
missing_inventory_sha_cnt = 0
97
missing_revision_sha_cnt = 0
99
history = branch.revision_history()
101
revcount = len(history)
104
# for all texts checked, text_id -> sha1
107
for rev_id in history:
109
pb.update('checking revision', revno, revcount)
110
mutter(' revision {%s}' % rev_id)
111
rev = branch.get_revision(rev_id)
112
if rev.revision_id != rev_id:
113
raise BzrCheckError('wrong internal revision id in revision {%s}'
116
# check the previous history entry is a parent of this entry
118
if last_rev_id is None:
119
raise BzrCheckError("revision {%s} has %d parents, but is the "
120
"start of the branch"
121
% (rev_id, len(rev.parents)))
122
for prr in rev.parents:
123
if prr.revision_id == last_rev_id:
157
126
raise BzrCheckError("previous revision {%s} not listed among "
158
127
"parents of {%s}"
159
128
% (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)
130
for prr in rev.parents:
131
if prr.revision_sha1 is None:
132
missing_revision_sha_cnt += 1
134
prid = prr.revision_id
135
actual_sha = branch.get_revision_sha1(prid)
136
if prr.revision_sha1 != actual_sha:
137
raise BzrCheckError("mismatched revision sha1 for "
138
"parent {%s} of {%s}: %s vs %s"
140
prr.revision_sha1, actual_sha))
142
raise BzrCheckError("revision {%s} has no parents listed but preceded "
144
% (rev_id, last_rev_id))
146
if rev.inventory_id != rev_id:
147
mismatch_inv_id.append(rev_id)
149
## TODO: Check all the required fields are present on the revision.
151
if rev.inventory_sha1:
152
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
153
if inv_sha1 != rev.inventory_sha1:
154
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
155
' value in revision {%s}' % rev_id)
157
missing_inventory_sha_cnt += 1
158
mutter("no inventory_sha1 on revision {%s}" % rev_id)
160
inv = branch.get_inventory(rev.inventory_id)
164
## p('revision %d/%d file ids' % (revno, revcount))
166
if file_id in seen_ids:
167
raise BzrCheckError('duplicated file_id {%s} '
168
'in inventory for revision {%s}'
170
seen_ids[file_id] = True
180
if ie.parent_id != None:
181
if ie.parent_id not in seen_ids:
182
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
183
% (ie.parent_id, rev_id))
185
if ie.kind == 'file':
186
if ie.text_id in checked_texts:
187
fp = checked_texts[ie.text_id]
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)
196
if file_id in seen_ids:
197
raise BzrCheckError('duplicated file_id {%s} '
198
'in inventory for revision {%s}'
200
seen_ids[file_id] = True
203
ie.check(self, rev_id, inv, tree)
205
for path, ie in inv.iter_entries():
206
if path in seen_names:
207
raise BzrCheckError('duplicated path %s '
208
'in inventory for revision {%s}'
189
if not ie.text_id in branch.text_store:
190
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
192
tf = branch.text_store[ie.text_id]
193
fp = fingerprint_file(tf)
194
checked_texts[ie.text_id] = fp
196
if ie.text_size != fp['size']:
197
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
198
if ie.text_sha1 != fp['sha1']:
199
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
200
elif ie.kind == 'directory':
201
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
202
raise BzrCheckError('directory {%s} has text in revision {%s}'
206
for path, ie in inv.iter_entries():
207
if path in seen_names:
208
raise BzrCheckError('duplicated path %s '
209
'in inventory for revision {%s}'
210
211
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)
219
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
221
if missing_inventory_sha_cnt:
222
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
224
if missing_revision_sha_cnt:
225
print '%d parent links are missing revision_sha1' % missing_revision_sha_cnt
227
# stub this out for now because the main bzr branch has references
228
# to revisions that aren't present in the store -- mbp 20050804
229
# if (missing_inventory_sha_cnt
230
# or missing_revision_sha_cnt):
231
# print ' (use "bzr upgrade" to fix them)'
234
print '%d revisions have mismatched inventory ids:' % len(mismatch_inv_id)
235
for rev_id in mismatch_inv_id:
added
removed
bzrlib/check.py
