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
21
def check(branch, update=False):
22
"""Run consistency checks on a branch.
24
If update is True, for revisions missing certain information
25
(right now this is inventory_sha1 and revision_sha1),
26
update them to include the appropriate values.
30
from bzrlib.trace import mutter
31
from bzrlib.errors import BzrCheckError
32
from bzrlib.osutils import fingerprint_file
33
from bzrlib.progress import ProgressBar
44
pb = ProgressBar(show_spinner=True)
48
missing_inventory_sha_cnt = 0
50
history = branch.revision_history()
52
revcount = len(history)
56
updated_revisions = []
58
# Set to True in the case that the previous revision entry
59
# was updated, since this will affect future revision entries
60
updated_previous_revision = False
64
pb.update('checking revision', revno, revcount)
65
mutter(' revision {%s}' % rid)
66
rev = branch.get_revision(rid)
67
if rev.revision_id != rid:
68
raise BzrCheckError('wrong internal revision id in revision {%s}' % rid)
69
if rev.precursor != last_ptr:
70
raise BzrCheckError('mismatched precursor in revision {%s}' % rid)
72
if rid in checked_revs:
73
raise BzrCheckError('repeated revision {%s}' % rid)
74
checked_revs[rid] = True
76
## TODO: Check all the required fields are present on the revision.
79
if rev.inventory_sha1:
80
#mutter(' checking inventory hash {%s}' % rev.inventory_sha1)
81
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
82
if inv_sha1 != rev.inventory_sha1:
83
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
84
' value in revision {%s}' % rid)
86
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
87
rev.inventory_sha1 = inv_sha1
90
missing_inventory_sha_cnt += 1
91
mutter("no inventory_sha1 on revision {%s}" % rid)
94
if rev.precursor_sha1:
95
precursor_sha1 = branch.get_revision_sha1(rev.precursor)
96
if updated_previous_revision:
97
# we don't expect the hashes to match, because
98
# we had to modify the previous revision_history entry.
99
rev.precursor_sha1 = precursor_sha1
102
#mutter(' checking precursor hash {%s}' % rev.precursor_sha1)
103
if rev.precursor_sha1 != precursor_sha1:
104
raise BzrCheckError('Precursor sha1 hash doesn\'t match'
105
' value in revision {%s}' % rid)
107
precursor_sha1 = branch.get_revision_sha1(rev.precursor)
108
rev.precursor_sha1 = precursor_sha1
112
updated_previous_revision = True
113
# We had to update this revision entries hashes
114
# Now we need to write out a new value
115
# This is a little bit invasive, since we are *rewriting* a
116
# revision entry. I'm not supremely happy about it, but
117
# there should be *some* way of making old entries have
118
# the full meta information.
119
import tempfile, os, errno
120
rev_tmp = tempfile.TemporaryFile()
121
rev.write_xml(rev_tmp)
124
tmpfd, tmp_path = tempfile.mkstemp(prefix=rid, suffix='.gz',
125
dir=branch.controlfilename('revision-store'))
127
def special_rename(p1, p2):
128
if sys.platform == 'win32':
132
if e.errno != e.ENOENT:
137
# TODO: We may need to handle the case where the old revision
138
# entry was not compressed (and thus did not end with .gz)
140
# Remove the old revision entry out of the way
141
rev_path = branch.controlfilename(['revision-store', rid+'.gz'])
142
special_rename(rev_path, tmp_path)
143
branch.revision_store.add(rev_tmp, rid) # Add the new one
144
os.remove(tmp_path) # Remove the old name
145
mutter(' Updated revision entry {%s}' % rid)
147
# On any exception, restore the old entry
148
special_rename(tmp_path, rev_path)
151
updated_revisions.append(rid)
153
updated_previous_revision = False
155
inv = branch.get_inventory(rev.inventory_id)
159
## p('revision %d/%d file ids' % (revno, revcount))
161
if file_id in seen_ids:
162
raise BzrCheckError('duplicated file_id {%s} '
163
'in inventory for revision {%s}'
165
seen_ids[file_id] = True
176
if ie.parent_id != None:
177
if ie.parent_id not in seen_ids:
178
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
179
% (ie.parent_id, rid))
181
if ie.kind == 'file':
182
if ie.text_id in checked_texts:
183
fp = checked_texts[ie.text_id]
185
if not ie.text_id in branch.text_store:
186
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
188
tf = branch.text_store[ie.text_id]
189
fp = fingerprint_file(tf)
190
checked_texts[ie.text_id] = fp
192
if ie.text_size != fp['size']:
193
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
194
if ie.text_sha1 != fp['sha1']:
195
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
196
elif ie.kind == 'directory':
197
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
198
raise BzrCheckError('directory {%s} has text in revision {%s}'
202
for path, ie in inv.iter_entries():
203
if path in seen_names:
204
raise BzrCheckError('duplicated path %r '
205
'in inventory for revision {%s}'
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 = {}
53
self.checked_weaves = {}
56
self.branch.lock_read()
57
self.progress = bzrlib.ui.ui_factory.progress_bar()
59
self.progress.update('retrieving inventory', 0, 0)
60
# do not put in init, as it should be done with progess,
61
# and inside the lock.
62
self.inventory_weave = self.branch._get_inventory_weave()
63
self.history = self.branch.revision_history()
64
if not len(self.history):
70
while revno < len(self.planned_revisions):
71
rev_id = self.planned_revisions[revno]
72
self.progress.update('checking revision', revno,
73
len(self.planned_revisions))
75
self.check_one_rev(rev_id)
80
def plan_revisions(self):
81
if not self.branch.revision_store.listable():
82
self.planned_revisions = self.branch.get_ancestry(self.history[-1])
83
self.planned_revisions.remove(None)
84
# FIXME progress bars should support this more nicely.
86
print ("Checking reachable history -"
87
" for a complete check use a local branch.")
90
self.planned_revisions = set(self.branch.revision_store)
91
inventoried = set(self.inventory_weave.names())
92
awol = self.planned_revisions - inventoried
94
raise BzrCheckError('Stored revisions missing from inventory'
95
'{%s}' % ','.join([f for f in awol]))
96
self.planned_revisions = list(self.planned_revisions)
98
def report_results(self, verbose):
99
note('checked branch %s format %d',
101
self.branch._branch_format)
103
note('%6d revisions', self.checked_rev_cnt)
104
note('%6d unique file texts', self.checked_text_cnt)
105
note('%6d repeated file texts', self.repeated_text_cnt)
106
note('%6d weaves', len(self.checked_weaves))
107
if self.missing_inventory_sha_cnt:
108
note('%6d revisions are missing inventory_sha1',
109
self.missing_inventory_sha_cnt)
110
if self.missing_revision_cnt:
111
note('%6d revisions are mentioned but not present',
112
self.missing_revision_cnt)
114
note('%6d ghost revisions', len(self.ghosts))
116
for ghost in self.ghosts:
118
if len(self.missing_parent_links):
119
note('%6d revisions missing parents in ancestry',
120
len(self.missing_parent_links))
122
for link, linkers in self.missing_parent_links.items():
123
note(' %s should be in the ancestry for:', link)
124
for linker in linkers:
125
note(' * %s', linker)
127
def check_one_rev(self, rev_id):
128
"""Check one revision.
130
rev_id - the one to check
132
last_rev_id - the previous one on the mainline, if any.
135
# mutter(' revision {%s}', rev_id)
138
rev_history_position = self.history.index(rev_id)
140
rev_history_position = None
142
if rev_history_position:
143
rev = branch.get_revision(rev_id)
144
if rev_history_position > 0:
145
last_rev_id = self.history[rev_history_position - 1]
147
rev = branch.get_revision(rev_id)
149
if rev.revision_id != rev_id:
150
raise BzrCheckError('wrong internal revision id in revision {%s}'
153
# check the previous history entry is a parent of this entry
155
if last_rev_id is not None:
156
for parent_id in rev.parent_ids:
157
if parent_id == last_rev_id:
160
raise BzrCheckError("previous revision {%s} not listed among "
162
% (last_rev_id, rev_id))
163
for parent in rev.parent_ids:
164
if not parent in self.planned_revisions:
165
missing_links = self.missing_parent_links.get(parent, [])
166
missing_links.append(rev_id)
167
self.missing_parent_links[parent] = missing_links
168
# list based so somewhat slow,
169
# TODO have a planned_revisions list and set.
170
if self.branch.has_revision(parent):
171
missing_ancestry = self.branch.get_ancestry(parent)
172
for missing in missing_ancestry:
173
if (missing is not None
174
and missing not in self.planned_revisions):
175
self.planned_revisions.append(missing)
177
self.ghosts.append(rev_id)
179
raise BzrCheckError("revision {%s} has no parents listed "
180
"but preceded by {%s}"
181
% (rev_id, last_rev_id))
183
if rev.inventory_sha1:
184
inv_sha1 = branch.get_inventory_sha1(rev_id)
185
if inv_sha1 != rev.inventory_sha1:
186
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
187
' value in revision {%s}' % rev_id)
189
missing_inventory_sha_cnt += 1
190
mutter("no inventory_sha1 on revision {%s}", rev_id)
191
self._check_revision_tree(rev_id)
192
self.checked_rev_cnt += 1
194
def check_weaves(self):
195
"""Check all the weaves we can get our hands on.
199
if self.branch.weave_store.listable():
200
weave_ids = list(self.branch.weave_store)
201
n_weaves = len(weave_ids)
202
self.progress.update('checking weave', 0, n_weaves)
203
self.inventory_weave.check(progress_bar=self.progress)
204
for i, weave_id in enumerate(weave_ids):
205
self.progress.update('checking weave', i, n_weaves)
206
w = self.branch.weave_store.get_weave(weave_id,
207
self.branch.get_transaction())
208
# No progress here, because it looks ugly.
210
self.checked_weaves[weave_id] = True
212
def _check_revision_tree(self, rev_id):
213
tree = self.branch.revision_tree(rev_id)
217
if file_id in seen_ids:
218
raise BzrCheckError('duplicated file_id {%s} '
219
'in inventory for revision {%s}'
221
seen_ids[file_id] = True
224
ie.check(self, rev_id, inv, tree)
226
for path, ie in inv.iter_entries():
227
if path in seen_names:
228
raise BzrCheckError('duplicated path %s '
229
'in inventory for revision {%s}'
207
231
seen_names[path] = True
214
print 'checked %d revisions, %d file texts' % (revcount, len(checked_texts))
215
if updated_revisions:
216
print '%d revisions updated to current format' % len(updated_revisions)
217
if missing_inventory_sha_cnt:
218
print '%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt
219
print ' (use bzr check --update to fix them)'
234
def check(branch, verbose):
235
"""Run consistency checks on a branch."""
236
checker = Check(branch)
238
checker.report_results(verbose)
added
removed
bzrlib/check.py
