← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/check.py

Merge trunk.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/check.py
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 import errors
 
36
from bzrlib import repository as _mod_repository
 
37
from bzrlib import revision
35
38
from bzrlib.errors import BzrCheckError
36
39
import bzrlib.ui
37
 
from bzrlib.trace import note
 
40
from bzrlib.trace import log_error, note
38
41
 
39
42
class Check(object):
40
43
    """Check a repository"""
53
56
        # maps (file-id, version) -> sha1; used by InventoryFile._check
54
57
        self.checked_texts = {}
55
58
        self.checked_weaves = {}
 
59
        self.unreferenced_versions = set()
 
60
        self.inconsistent_parents = []
56
61
 
57
62
    def check(self):
58
63
        self.repository.lock_read()
59
64
        self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
60
65
        try:
61
 
            self.progress.update('retrieving inventory', 0, 0)
 
66
            self.progress.update('retrieving inventory', 0, 2)
62
67
            # do not put in init, as it should be done with progess,
63
68
            # and inside the lock.
64
69
            self.inventory_weave = self.repository.get_inventory_weave()
 
70
            self.progress.update('checking revision graph', 1)
 
71
            self.check_revision_graph()
65
72
            self.plan_revisions()
66
73
            revno = 0
67
 
            self.check_weaves()
68
74
            while revno < len(self.planned_revisions):
69
75
                rev_id = self.planned_revisions[revno]
70
76
                self.progress.update('checking revision', revno,
71
77
                                     len(self.planned_revisions))
72
78
                revno += 1
73
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.
 
82
            self.check_weaves()
74
83
        finally:
75
84
            self.progress.finished()
76
85
            self.repository.unlock()
77
86
 
 
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
 
91
            return
 
92
        bad_revisions = self.repository._find_inconsistent_revision_parents()
 
93
        self.revs_with_bad_parents_in_index = list(bad_revisions)
 
94
 
78
95
    def plan_revisions(self):
79
96
        repository = self.repository
80
 
        self.planned_revisions = set(repository.all_revision_ids())
 
97
        self.planned_revisions = repository.all_revision_ids()
81
98
        self.progress.clear()
82
99
        inventoried = set(self.inventory_weave.versions())
83
 
        awol = self.planned_revisions - inventoried
 
100
        awol = set(self.planned_revisions) - inventoried
84
101
        if len(awol) > 0:
85
102
            raise BzrCheckError('Stored revisions missing from inventory'
86
103
                '{%s}' % ','.join([f for f in awol]))
87
 
        self.planned_revisions = list(self.planned_revisions)
88
104
 
89
105
    def report_results(self, verbose):
90
106
        note('checked repository %s format %s',
91
107
             self.repository.bzrdir.root_transport,
92
108
             self.repository._format)
93
109
        note('%6d revisions', self.checked_rev_cnt)
 
110
        note('%6d file-ids', len(self.checked_weaves))
94
111
        note('%6d unique file texts', self.checked_text_cnt)
95
112
        note('%6d repeated file texts', self.repeated_text_cnt)
96
 
        note('%6d weaves', len(self.checked_weaves))
 
113
        note('%6d unreferenced text versions',
 
114
             len(self.unreferenced_versions))
97
115
        if self.missing_inventory_sha_cnt:
98
116
            note('%6d revisions are missing inventory_sha1',
99
117
                 self.missing_inventory_sha_cnt)
113
131
                    note('      %s should be in the ancestry for:', link)
114
132
                    for linker in linkers:
115
133
                        note('       * %s', linker)
 
134
            if verbose:
 
135
                for file_id, revision_id in self.unreferenced_versions:
 
136
                    log_error('unreferenced version: {%s} in %s', revision_id,
 
137
                        file_id)
 
138
        if len(self.inconsistent_parents):
 
139
            note('%6d inconsistent parents', len(self.inconsistent_parents))
 
140
            if verbose:
 
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 '
 
144
                         'but should have %r'
 
145
                         % (file_id, revision_id, found_parents,
 
146
                             correct_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))
 
150
            if verbose:
 
151
                for item in self.revs_with_bad_parents_in_index:
 
152
                    revision_id, index_parents, actual_parents = item
 
153
                    note(
 
154
                        '       %s has wrong parents in index: '
 
155
                        '%r should be %r',
 
156
                        revision_id, index_parents, actual_parents)
116
157
 
117
158
    def check_one_rev(self, rev_id):
118
159
        """Check one revision.
157
198
        if self.repository.weave_store.listable():
158
199
            weave_ids = list(self.repository.weave_store)
159
200
            n_weaves = len(weave_ids) + 1
160
 
        self.progress.update('checking weave', 0, n_weaves)
 
201
        self.progress.update('checking versionedfile', 0, n_weaves)
161
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()
162
206
        for i, weave_id in enumerate(weave_ids):
163
 
            self.progress.update('checking weave', i, n_weaves)
 
207
            self.progress.update('checking versionedfile', i, n_weaves)
164
208
            w = self.repository.weave_store.get_weave(weave_id,
165
209
                    self.repository.get_transaction())
166
210
            # No progress here, because it looks ugly.
167
211
            w.check()
 
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))
168
220
            self.checked_weaves[weave_id] = True
169
221
 
170
222
    def _check_revision_tree(self, rev_id):
204
256
        branch.unlock()
205
257
    branch_result.report_results(verbose)
206
258
    repo_result.report_results(verbose)
 
259
 
 
260
 
 
261