← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/check.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2006-04-27 01:14:33 UTC
  • mfrom: (1686.1.1 integration)
  • Revision ID: pqm@pqm.ubuntu.com-20060427011433-95634ee1da8a2049
Merge in faster joins from weave to knit.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/check.py
1
1
# Copyright (C) 2005, 2006 by Canonical Ltd
2
 
#
 
2
 
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
 
#
 
7
 
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
11
11
# GNU General Public License for more details.
12
 
#
 
12
 
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
32
32
# raising them.  If there's more than one exception it'd be good to see them
33
33
# all.
34
34
 
35
 
from bzrlib.errors import BzrCheckError
36
35
import bzrlib.ui
37
 
from bzrlib.trace import note
 
36
from bzrlib.trace import note, warning
 
37
from bzrlib.osutils import rename, sha_string, fingerprint_file
 
38
from bzrlib.trace import mutter
 
39
from bzrlib.errors import BzrCheckError, NoSuchRevision
 
40
from bzrlib.inventory import ROOT_ID
 
41
 
38
42
 
39
43
class Check(object):
40
 
    """Check a repository"""
 
44
    """Check a branch"""
41
45
 
42
46
    # The Check object interacts with InventoryEntry.check, etc.
43
47
 
44
 
    def __init__(self, repository):
45
 
        self.repository = repository
 
48
    def __init__(self, branch):
 
49
        self.branch = branch
 
50
        self.repository = branch.repository
46
51
        self.checked_text_cnt = 0
47
52
        self.checked_rev_cnt = 0
48
53
        self.ghosts = []
55
60
        self.checked_weaves = {}
56
61
 
57
62
    def check(self):
58
 
        self.repository.lock_read()
 
63
        self.branch.lock_read()
59
64
        self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
60
65
        try:
61
66
            self.progress.update('retrieving inventory', 0, 0)
62
67
            # do not put in init, as it should be done with progess,
63
68
            # and inside the lock.
64
 
            self.inventory_weave = self.repository.get_inventory_weave()
 
69
            self.inventory_weave = self.branch.repository.get_inventory_weave()
 
70
            self.history = self.branch.revision_history()
 
71
            if not len(self.history):
 
72
                # nothing to see here
 
73
                return
65
74
            self.plan_revisions()
66
75
            revno = 0
67
76
            self.check_weaves()
73
82
                self.check_one_rev(rev_id)
74
83
        finally:
75
84
            self.progress.finished()
76
 
            self.repository.unlock()
 
85
            self.branch.unlock()
77
86
 
78
87
    def plan_revisions(self):
79
 
        repository = self.repository
 
88
        repository = self.branch.repository
80
89
        self.planned_revisions = set(repository.all_revision_ids())
81
90
        self.progress.clear()
82
91
        inventoried = set(self.inventory_weave.versions())
87
96
        self.planned_revisions = list(self.planned_revisions)
88
97
 
89
98
    def report_results(self, verbose):
90
 
        note('checked repository %s format %s',
91
 
             self.repository.bzrdir.root_transport,
92
 
             self.repository._format)
 
99
        note('checked branch %s format %s',
 
100
             self.branch.base, 
 
101
             self.branch._format)
 
102
 
93
103
        note('%6d revisions', self.checked_rev_cnt)
94
104
        note('%6d unique file texts', self.checked_text_cnt)
95
105
        note('%6d repeated file texts', self.repeated_text_cnt)
106
116
                for ghost in self.ghosts:
107
117
                    note('      %s', ghost)
108
118
        if len(self.missing_parent_links):
109
 
            note('%6d revisions missing parents in ancestry',
 
119
            note('%6d revisions missing parents in ancestry', 
110
120
                 len(self.missing_parent_links))
111
121
            if verbose:
112
122
                for link, linkers in self.missing_parent_links.items():
118
128
        """Check one revision.
119
129
 
120
130
        rev_id - the one to check
 
131
 
 
132
        last_rev_id - the previous one on the mainline, if any.
121
133
        """
122
 
        rev = self.repository.get_revision(rev_id)
 
134
 
 
135
        # mutter('    revision {%s}', rev_id)
 
136
        branch = self.branch
 
137
        try:
 
138
            rev_history_position = self.history.index(rev_id)
 
139
        except ValueError:
 
140
            rev_history_position = None
 
141
        last_rev_id = None
 
142
        if rev_history_position:
 
143
            rev = branch.repository.get_revision(rev_id)
 
144
            if rev_history_position > 0:
 
145
                last_rev_id = self.history[rev_history_position - 1]
 
146
        else:
 
147
            rev = branch.repository.get_revision(rev_id)
123
148
                
124
149
        if rev.revision_id != rev_id:
125
150
            raise BzrCheckError('wrong internal revision id in revision {%s}'
126
151
                                % rev_id)
127
152
 
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)
 
153
        # check the previous history entry is a parent of this entry
 
154
        if rev.parent_ids:
 
155
            if last_rev_id is not None:
 
156
                for parent_id in rev.parent_ids:
 
157
                    if parent_id == last_rev_id:
 
158
                        break
141
159
                else:
142
 
                    self.ghosts.append(rev_id)
 
160
                    raise BzrCheckError("previous revision {%s} not listed among "
 
161
                                        "parents of {%s}"
 
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.repository.has_revision(parent):
 
171
                        missing_ancestry = self.repository.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)
 
176
                    else:
 
177
                        self.ghosts.append(rev_id)
 
178
        elif last_rev_id:
 
179
            raise BzrCheckError("revision {%s} has no parents listed "
 
180
                                "but preceded by {%s}"
 
181
                                % (rev_id, last_rev_id))
143
182
 
144
183
        if rev.inventory_sha1:
145
 
            inv_sha1 = self.repository.get_inventory_sha1(rev_id)
 
184
            inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
146
185
            if inv_sha1 != rev.inventory_sha1:
147
186
                raise BzrCheckError('Inventory sha1 hash doesn\'t match'
148
187
                    ' value in revision {%s}' % rev_id)
 
188
        else:
 
189
            missing_inventory_sha_cnt += 1
 
190
            mutter("no inventory_sha1 on revision {%s}", rev_id)
149
191
        self._check_revision_tree(rev_id)
150
192
        self.checked_rev_cnt += 1
151
193
 
154
196
        """
155
197
        n_weaves = 1
156
198
        weave_ids = []
157
 
        if self.repository.weave_store.listable():
158
 
            weave_ids = list(self.repository.weave_store)
 
199
        if self.branch.repository.weave_store.listable():
 
200
            weave_ids = list(self.branch.repository.weave_store)
159
201
            n_weaves = len(weave_ids)
160
202
        self.progress.update('checking weave', 0, n_weaves)
161
203
        self.inventory_weave.check(progress_bar=self.progress)
162
204
        for i, weave_id in enumerate(weave_ids):
163
205
            self.progress.update('checking weave', i, n_weaves)
164
 
            w = self.repository.weave_store.get_weave(weave_id,
165
 
                    self.repository.get_transaction())
 
206
            w = self.branch.repository.weave_store.get_weave(weave_id,
 
207
                    self.branch.repository.get_transaction())
166
208
            # No progress here, because it looks ugly.
167
209
            w.check()
168
210
            self.checked_weaves[weave_id] = True
169
211
 
170
212
    def _check_revision_tree(self, rev_id):
171
 
        tree = self.repository.revision_tree(rev_id)
 
213
        tree = self.branch.repository.revision_tree(rev_id)
172
214
        inv = tree.inventory
173
215
        seen_ids = {}
174
216
        for file_id in inv:
190
232
 
191
233
 
192
234
def check(branch, verbose):
193
 
    """Run consistency checks on a branch.
194
 
    
195
 
    Results are reported through logging.
196
 
    
197
 
    :raise BzrCheckError: if there's a consistency error.
198
 
    """
199
 
    branch.lock_read()
200
 
    try:
201
 
        branch_result = branch.check()
202
 
        repo_result = branch.repository.check([branch.last_revision()])
203
 
    finally:
204
 
        branch.unlock()
205
 
    branch_result.report_results(verbose)
206
 
    repo_result.report_results(verbose)
 
235
    """Run consistency checks on a branch."""
 
236
    checker = Check(branch)
 
237
    checker.check()
 
238
    checker.report_results(verbose)