~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/check.py

Exclude more files from dumb-rsync upload

Show diffs side-by-side

added added

removed removed

Lines of Context:
25
25
# TODO: Check revision, inventory and entry objects have all 
26
26
# required fields.
27
27
 
 
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.
28
30
 
29
31
import bzrlib.ui
30
32
from bzrlib.trace import note, warning
37
39
 
38
40
class Check(object):
39
41
    """Check a branch"""
 
42
 
40
43
    def __init__(self, branch):
41
44
        self.branch = branch
42
 
        branch.lock_read()
43
 
        try:
44
 
            branch.weave_store.enable_cache = True
45
 
            branch.control_weaves.enable_cache = True
46
 
            self.run()
47
 
        finally:
48
 
            branch.unlock()
49
 
            branch.weave_store.enable_cache = False
50
 
            branch.control_weaves.enable_cache = False
51
 
 
52
 
 
53
 
    def run(self):
54
 
        branch = self.branch
55
 
 
56
 
 
 
45
        self.inventory_weave = branch._get_inventory_weave()
57
46
        self.checked_text_cnt = 0
58
47
        self.checked_rev_cnt = 0
 
48
        self.ghosts = []
59
49
        self.repeated_text_cnt = 0
 
50
        self.missing_parent_links = {}
60
51
        self.missing_inventory_sha_cnt = 0
61
52
        self.missing_revision_cnt = 0
62
53
        # maps (file-id, version) -> sha1
63
54
        self.checked_texts = {}
64
55
 
65
 
        history = branch.revision_history()
66
 
        revno = 0
67
 
        revcount = len(history)
68
 
 
69
 
        last_rev_id = None
 
56
    def check(self):
 
57
        self.branch.lock_read()
70
58
        self.progress = bzrlib.ui.ui_factory.progress_bar()
71
 
        for rev_id in history:
72
 
            self.progress.update('checking revision', revno, revcount)
73
 
            revno += 1
74
 
            self.check_one_rev(rev_id, last_rev_id)
75
 
            last_rev_id = rev_id
76
 
        self.progress.clear()
77
 
        self.report_results()
78
 
 
79
 
 
80
 
    def report_results(self):
 
59
        try:
 
60
            self.history = self.branch.revision_history()
 
61
            if not len(self.history):
 
62
                # nothing to see here
 
63
                return
 
64
            if not self.branch.revision_store.listable():
 
65
                raise BzrCheckError("Branch must be local")
 
66
            self.planned_revisions = set(self.branch.revision_store)
 
67
            inventoried = set(self.inventory_weave.names())
 
68
            awol = self.planned_revisions - inventoried
 
69
            if len(awol) > 0:
 
70
                raise BzrCheckError('Stored revisions missing from inventory'
 
71
                    '{%s}' % ','.join([f for f in awol]))
 
72
    
 
73
            for revno, rev_id in enumerate(self.planned_revisions):
 
74
                self.progress.update('checking revision', revno+1,
 
75
                                     len(self.planned_revisions))
 
76
                self.check_one_rev(rev_id)
 
77
        finally:
 
78
            self.progress.clear()
 
79
            self.branch.unlock()
 
80
 
 
81
    def report_results(self, verbose):
81
82
        note('checked branch %s format %d',
82
83
             self.branch.base, 
83
84
             self.branch._branch_format)
86
87
        note('%6d unique file texts', self.checked_text_cnt)
87
88
        note('%6d repeated file texts', self.repeated_text_cnt)
88
89
        if self.missing_inventory_sha_cnt:
89
 
            note('%d revisions are missing inventory_sha1',
 
90
            note('%6d revisions are missing inventory_sha1',
90
91
                 self.missing_inventory_sha_cnt)
91
92
        if self.missing_revision_cnt:
92
 
            note('%d revisions are mentioned but not present',
 
93
            note('%6d revisions are mentioned but not present',
93
94
                 self.missing_revision_cnt)
94
 
 
95
 
 
96
 
    def check_one_rev(self, rev_id, last_rev_id):
 
95
        if len(self.ghosts):
 
96
            note('%6d ghost revisions', len(self.ghosts))
 
97
            if verbose:
 
98
                for ghost in self.ghosts:
 
99
                    note('      %s', ghost)
 
100
        if len(self.missing_parent_links):
 
101
            note('%6d revisions missing parents in ancestry', 
 
102
                 len(self.missing_parent_links))
 
103
            if verbose:
 
104
                for link, linkers in self.missing_parent_links.items():
 
105
                    note('      %s should be in the ancestry for:', link)
 
106
                    for linker in linkers:
 
107
                        note('       * %s', linker)
 
108
 
 
109
    def check_one_rev(self, rev_id):
97
110
        """Check one revision.
98
111
 
99
112
        rev_id - the one to check
101
114
        last_rev_id - the previous one on the mainline, if any.
102
115
        """
103
116
 
104
 
        # mutter('    revision {%s}' % rev_id)
 
117
        # mutter('    revision {%s}', rev_id)
105
118
        branch = self.branch
106
 
        rev = branch.get_revision(rev_id)
 
119
        try:
 
120
            rev_history_position = self.history.index(rev_id)
 
121
        except ValueError:
 
122
            rev_history_position = None
 
123
        last_rev_id = None
 
124
        if rev_history_position:
 
125
            rev = branch.get_revision(rev_id)
 
126
            if rev_history_position > 0:
 
127
                last_rev_id = self.history[rev_history_position - 1]
 
128
        else:
 
129
            rev = branch.get_revision(rev_id)
 
130
                
107
131
        if rev.revision_id != rev_id:
108
132
            raise BzrCheckError('wrong internal revision id in revision {%s}'
109
133
                                % rev_id)
110
134
 
111
135
        # check the previous history entry is a parent of this entry
112
136
        if rev.parent_ids:
113
 
            if last_rev_id is None:
114
 
                raise BzrCheckError("revision {%s} has %d parents, but is the "
115
 
                                    "start of the branch"
116
 
                                    % (rev_id, len(rev.parent_ids)))
117
 
            for parent_id in rev.parent_ids:
118
 
                if parent_id == last_rev_id:
119
 
                    break
120
 
            else:
121
 
                raise BzrCheckError("previous revision {%s} not listed among "
122
 
                                    "parents of {%s}"
123
 
                                    % (last_rev_id, rev_id))
 
137
            if last_rev_id is not None:
 
138
                for parent_id in rev.parent_ids:
 
139
                    if parent_id == last_rev_id:
 
140
                        break
 
141
                else:
 
142
                    raise BzrCheckError("previous revision {%s} not listed among "
 
143
                                        "parents of {%s}"
 
144
                                        % (last_rev_id, rev_id))
 
145
            for parent in rev.parent_ids:
 
146
                if not parent in self.planned_revisions:
 
147
                    missing_links = self.missing_parent_links.get(parent, [])
 
148
                    missing_links.append(rev_id)
 
149
                    self.missing_parent_links[parent] = missing_links
 
150
                    # list based so slow, TODO have a planned_revisions list and set.
 
151
                    if self.branch.has_revision(parent):
 
152
                        missing_ancestry = self.branch.get_ancestry(parent)
 
153
                        for missing in missing_ancestry:
 
154
                            if (missing is not None 
 
155
                                and missing not in self.planned_revisions):
 
156
                                self.planned_revisions.append(missing)
 
157
                    else:
 
158
                        self.ghosts.append(rev_id)
124
159
        elif last_rev_id:
125
160
            raise BzrCheckError("revision {%s} has no parents listed "
126
161
                                "but preceded by {%s}"
133
168
                    ' value in revision {%s}' % rev_id)
134
169
        else:
135
170
            missing_inventory_sha_cnt += 1
136
 
            mutter("no inventory_sha1 on revision {%s}" % rev_id)
 
171
            mutter("no inventory_sha1 on revision {%s}", rev_id)
137
172
        self._check_revision_tree(rev_id)
138
173
        self.checked_rev_cnt += 1
139
174
 
159
194
            seen_names[path] = True
160
195
 
161
196
 
162
 
def check(branch):
 
197
def check(branch, verbose):
163
198
    """Run consistency checks on a branch."""
164
 
    Check(branch)
 
199
    checker = Check(branch)
 
200
    checker.check()
 
201
    checker.report_results(verbose)