1
# Copyright (C) 2004, 2005 by Martin Pool
2
# Copyright (C) 2005 by Canonical Ltd
1
# Copyright (C) 2005, 2006 Canonical Ltd
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
from bzrlib.trace import note, warning
21
def _update_store_entry(obj, obj_id, branch, store_name, store):
22
"""This is just a meta-function, which handles both revision entries
23
and inventory entries.
25
from bzrlib.trace import mutter
26
import tempfile, os, errno
27
from osutils import rename
28
obj_tmp = tempfile.TemporaryFile()
29
obj.write_xml(obj_tmp)
32
tmpfd, tmp_path = tempfile.mkstemp(prefix=obj_id, suffix='.gz',
33
dir=branch.controlfilename(store_name))
36
orig_obj_path = branch.controlfilename([store_name, obj_id+'.gz'])
37
# Remove the old entry out of the way
38
rename(orig_obj_path, tmp_path)
40
# TODO: We may need to handle the case where the old
41
# entry was not compressed (and thus did not end with .gz)
43
store.add(obj_tmp, obj_id) # Add the new one
44
os.remove(tmp_path) # Remove the old name
45
mutter(' Updated %s entry {%s}' % (store_name, obj_id))
47
# On any exception, restore the old entry
48
rename(tmp_path, orig_obj_path)
51
if os.path.exists(tmp_path):
52
# Unfortunately, the next command might throw
53
# an exception, which will mask a previous exception.
57
def _update_revision_entry(rev, branch):
58
"""After updating the values in a revision, make sure to
59
write out the data, but try to do it in an atomic manner.
61
:param rev: The Revision object to store
62
:param branch: The Branch object where this Revision is to be stored.
64
_update_store_entry(rev, rev.revision_id, branch,
65
'revision-store', branch.revision_store)
67
def _update_inventory_entry(inv, inv_id, branch):
68
"""When an inventory has been modified (such as by adding a unique tree root)
69
this atomically re-generates the file.
71
:param inv: The Inventory
72
:param inv_id: The inventory id for this inventory
73
:param branch: The Branch where this entry will be stored.
75
_update_store_entry(inv, inv_id, branch,
76
'inventory-store', branch.inventory_store)
80
"""Run consistency checks on a branch.
82
TODO: Also check non-mainline revisions mentioned as parents.
84
TODO: Check for extra files in the control directory.
86
from bzrlib.trace import mutter
87
from bzrlib.errors import BzrCheckError, NoSuchRevision
88
from bzrlib.osutils import fingerprint_file
89
from bzrlib.inventory import ROOT_ID
90
from bzrlib.branch import gen_root_id
97
missing_inventory_sha_cnt = 0
98
missing_revision_sha_cnt = 0
99
missing_revision_cnt = 0
101
history = branch.revision_history()
103
revcount = len(history)
106
# for all texts checked, text_id -> sha1
109
progress = bzrlib.ui.ui_factory.progress_bar()
111
for rev_id in history:
113
progress.update('checking revision', revno, revcount)
114
# mutter(' revision {%s}' % rev_id)
115
rev = branch.get_revision(rev_id)
116
if rev.revision_id != rev_id:
117
raise BzrCheckError('wrong internal revision id in revision {%s}'
120
# check the previous history entry is a parent of this entry
122
if last_rev_id is None:
123
raise BzrCheckError("revision {%s} has %d parents, but is the "
124
"start of the branch"
125
% (rev_id, len(rev.parents)))
126
for prr in rev.parents:
127
if prr.revision_id == last_rev_id:
130
raise BzrCheckError("previous revision {%s} not listed among "
132
% (last_rev_id, rev_id))
134
for prr in rev.parents:
135
if prr.revision_sha1 is None:
136
missing_revision_sha_cnt += 1
138
prid = prr.revision_id
141
actual_sha = branch.get_revision_sha1(prid)
142
except NoSuchRevision:
143
missing_revision_cnt += 1
144
mutter("parent {%s} of {%s} not present in store",
148
if prr.revision_sha1 != actual_sha:
149
raise BzrCheckError("mismatched revision sha1 for "
150
"parent {%s} of {%s}: %s vs %s"
152
prr.revision_sha1, actual_sha))
154
raise BzrCheckError("revision {%s} has no parents listed but preceded "
156
% (rev_id, last_rev_id))
158
if rev.inventory_id != rev_id:
159
mismatch_inv_id.append(rev_id)
161
## TODO: Check all the required fields are present on the revision.
163
if rev.inventory_sha1:
164
inv_sha1 = branch.get_inventory_sha1(rev.inventory_id)
165
if inv_sha1 != rev.inventory_sha1:
166
raise BzrCheckError('Inventory sha1 hash doesn\'t match'
167
' value in revision {%s}' % rev_id)
169
missing_inventory_sha_cnt += 1
170
mutter("no inventory_sha1 on revision {%s}" % rev_id)
172
inv = branch.get_inventory(rev.inventory_id)
176
## p('revision %d/%d file ids' % (revno, revcount))
178
if file_id in seen_ids:
179
raise BzrCheckError('duplicated file_id {%s} '
180
'in inventory for revision {%s}'
182
seen_ids[file_id] = True
192
if ie.parent_id != None:
193
if ie.parent_id not in seen_ids:
194
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
195
% (ie.parent_id, rev_id))
197
if ie.kind == 'file':
198
if ie.text_id in checked_texts:
199
fp = checked_texts[ie.text_id]
201
if not ie.text_id in branch.text_store:
202
raise BzrCheckError('text {%s} not in text_store' % ie.text_id)
204
tf = branch.text_store[ie.text_id]
205
fp = fingerprint_file(tf)
206
checked_texts[ie.text_id] = fp
208
if ie.text_size != fp['size']:
209
raise BzrCheckError('text {%s} wrong size' % ie.text_id)
210
if ie.text_sha1 != fp['sha1']:
211
raise BzrCheckError('text {%s} wrong sha1' % ie.text_id)
212
elif ie.kind == 'directory':
213
if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
214
raise BzrCheckError('directory {%s} has text in revision {%s}'
218
for path, ie in inv.iter_entries():
219
if path in seen_names:
220
raise BzrCheckError('duplicated path %s '
221
'in inventory for revision {%s}'
223
seen_names[path] = True
231
note('checked %d revisions, %d file texts' % (revcount, len(checked_texts)))
233
if missing_inventory_sha_cnt:
234
note('%d revisions are missing inventory_sha1' % missing_inventory_sha_cnt)
236
if missing_revision_sha_cnt:
237
note('%d parent links are missing revision_sha1' % missing_revision_sha_cnt)
239
if missing_revision_cnt:
240
note('%d revisions are mentioned but not present' % missing_revision_cnt)
242
if missing_revision_cnt:
243
print '%d revisions are mentioned but not present' % missing_revision_cnt
245
# stub this out for now because the main bzr branch has references
246
# to revisions that aren't present in the store -- mbp 20050804
247
# if (missing_inventory_sha_cnt
248
# or missing_revision_sha_cnt):
249
# print ' (use "bzr upgrade" to fix them)'
252
warning('%d revisions have mismatched inventory ids:' % len(mismatch_inv_id))
253
for rev_id in mismatch_inv_id:
254
warning(' %s', rev_id)
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
# TODO: Perhaps have a way to record errors other than by raising exceptions;
18
# would perhaps be enough to accumulate exception objects in a list without
19
# raising them. If there's more than one exception it'd be good to see them
22
"""Checking of bzr objects.
24
check_refs is a concept used for optimising check. Objects that depend on other
25
objects (e.g. tree on repository) can list the objects they would be requesting
26
so that when the dependent object is checked, matches can be pulled out and
27
evaluated in-line rather than re-reading the same data many times.
28
check_refs are tuples (kind, value). Currently defined kinds are:
30
* 'trees', where value is a revid and the looked up objects are revision trees.
31
* 'lefthand-distance', where value is a revid and the looked up objects are the
32
distance along the lefthand path to NULL for that revid.
33
* 'revision-existence', where value is a revid, and the result is True or False
34
indicating that the revision was found/not found.
42
from .controldir import ControlDir
43
from .trace import note
44
from .i18n import gettext
48
"""Check a repository"""
50
def __init__(self, repository, check_repo=True):
51
self.repository = repository
53
def report_results(self, verbose):
54
raise NotImplementedError(self.report_results)
57
def scan_branch(branch, needed_refs, exit_stack):
58
"""Scan a branch for refs.
60
:param branch: The branch to schedule for checking.
61
:param needed_refs: Refs we are accumulating.
62
:param exit_stack: The exit stack accumulating.
64
note(gettext("Checking branch at '%s'.") % (branch.base,))
65
exit_stack.enter_context(branch.lock_read())
66
branch_refs = branch._get_check_refs()
67
for ref in branch_refs:
68
reflist = needed_refs.setdefault(ref, [])
69
reflist.append(branch)
72
def scan_tree(base_tree, tree, needed_refs, exit_stack):
73
"""Scan a tree for refs.
75
:param base_tree: The original tree check opened, used to detect duplicate
77
:param tree: The tree to schedule for checking.
78
:param needed_refs: Refs we are accumulating.
79
:param exit_stack: The exit stack accumulating.
81
if base_tree is not None and tree.basedir == base_tree.basedir:
83
note(gettext("Checking working tree at '%s'.") % (tree.basedir,))
84
exit_stack.enter_context(tree.lock_read())
85
tree_refs = tree._get_check_refs()
87
reflist = needed_refs.setdefault(ref, [])
91
def check_dwim(path, verbose, do_branch=False, do_repo=False, do_tree=False):
92
"""Check multiple objects.
94
If errors occur they are accumulated and reported as far as possible, and
95
an exception raised at the end of the process.
98
base_tree, branch, repo, relpath = \
99
ControlDir.open_containing_tree_branch_or_repository(path)
100
except errors.NotBranchError:
101
base_tree = branch = repo = None
103
with contextlib.ExitStack() as exit_stack:
105
if base_tree is not None:
106
# If the tree is a lightweight checkout we won't see it in
107
# repo.find_branches - add now.
109
scan_tree(None, base_tree, needed_refs, exit_stack)
110
branch = base_tree.branch
111
if branch is not None:
114
# The branch is in a shared repository
115
repo = branch.repository
117
exit_stack.enter_context(repo.lock_read())
118
branches = list(repo.find_branches(using=True))
120
if do_branch or do_tree:
121
for branch in branches:
124
tree = branch.controldir.open_workingtree()
126
except (errors.NotLocalUrl, errors.NoWorkingTree):
129
scan_tree(base_tree, tree, needed_refs, exit_stack)
131
scan_branch(branch, needed_refs, exit_stack)
132
if do_branch and not branches:
133
note(gettext("No branch found at specified location."))
134
if do_tree and base_tree is None and not saw_tree:
135
note(gettext("No working tree found at specified location."))
136
if do_repo or do_branch or do_tree:
138
note(gettext("Checking repository at '%s'.")
140
result = repo.check(None, callback_refs=needed_refs,
142
result.report_results(verbose)
145
note(gettext("No working tree found at specified location."))
147
note(gettext("No branch found at specified location."))
149
note(gettext("No repository found at specified location."))