← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/check.py

  • Committer: Michael Ellerman
  • Date: 2006-02-28 14:45:51 UTC
  • mto: (1558.1.18 Aaron's integration)
  • mto: This revision was merged to the branch mainline in revision 1586.
  • Revision ID: michael@ellerman.id.au-20060228144551-3d9941ecde4a0b0a
Update contrib/pwk for -p1 diffs from bzr

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/check.py
1
 
# Copyright (C) 2005, 2006 by Canonical Ltd
 
1
# Copyright (C) 2004, 2005 by Martin Pool
 
2
# Copyright (C) 2005 by Canonical Ltd
2
3
 
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
27
28
# TODO: Get every revision in the revision-store even if they're not
28
29
# referenced by history and make sure they're all valid.
29
30
 
30
 
# TODO: Perhaps have a way to record errors other than by raising exceptions;
31
 
# would perhaps be enough to accumulate exception objects in a list without
32
 
# raising them.  If there's more than one exception it'd be good to see them
33
 
# all.
34
 
 
35
 
from bzrlib.errors import BzrCheckError
36
31
import bzrlib.ui
37
 
from bzrlib.trace import note
 
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
 
37
 
38
38
 
39
39
class Check(object):
40
 
    """Check a repository"""
41
 
 
42
 
    # The Check object interacts with InventoryEntry.check, etc.
43
 
 
44
 
    def __init__(self, repository):
45
 
        self.repository = repository
 
40
    """Check a branch"""
 
41
 
 
42
    def __init__(self, branch):
 
43
        self.branch = branch
 
44
        self.repository = branch.repository
46
45
        self.checked_text_cnt = 0
47
46
        self.checked_rev_cnt = 0
48
47
        self.ghosts = []
50
49
        self.missing_parent_links = {}
51
50
        self.missing_inventory_sha_cnt = 0
52
51
        self.missing_revision_cnt = 0
53
 
        # maps (file-id, version) -> sha1; used by InventoryFile._check
 
52
        # maps (file-id, version) -> sha1
54
53
        self.checked_texts = {}
55
54
        self.checked_weaves = {}
56
55
 
57
56
    def check(self):
58
 
        self.repository.lock_read()
59
 
        self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
 
57
        self.branch.lock_read()
 
58
        self.progress = bzrlib.ui.ui_factory.progress_bar()
60
59
        try:
61
60
            self.progress.update('retrieving inventory', 0, 0)
62
61
            # do not put in init, as it should be done with progess,
63
62
            # and inside the lock.
64
 
            self.inventory_weave = self.repository.get_inventory_weave()
 
63
            self.inventory_weave = self.branch.repository.get_inventory_weave()
 
64
            self.history = self.branch.revision_history()
 
65
            if not len(self.history):
 
66
                # nothing to see here
 
67
                return
65
68
            self.plan_revisions()
66
69
            revno = 0
67
70
            self.check_weaves()
72
75
                revno += 1
73
76
                self.check_one_rev(rev_id)
74
77
        finally:
75
 
            self.progress.finished()
76
 
            self.repository.unlock()
 
78
            self.progress.clear()
 
79
            self.branch.unlock()
77
80
 
78
81
    def plan_revisions(self):
79
 
        repository = self.repository
80
 
        self.planned_revisions = set(repository.all_revision_ids())
81
 
        self.progress.clear()
82
 
        inventoried = set(self.inventory_weave.versions())
 
82
        repository = self.branch.repository
 
83
        if not repository.revision_store.listable():
 
84
            self.planned_revisions = repository.get_ancestry(self.history[-1])
 
85
            self.planned_revisions.remove(None)
 
86
            # FIXME progress bars should support this more nicely.
 
87
            self.progress.clear()
 
88
            print ("Checking reachable history -"
 
89
                   " for a complete check use a local branch.")
 
90
            return
 
91
        
 
92
        self.planned_revisions = set(repository.revision_store)
 
93
        inventoried = set(self.inventory_weave.names())
83
94
        awol = self.planned_revisions - inventoried
84
95
        if len(awol) > 0:
85
96
            raise BzrCheckError('Stored revisions missing from inventory'
87
98
        self.planned_revisions = list(self.planned_revisions)
88
99
 
89
100
    def report_results(self, verbose):
90
 
        note('checked repository %s format %s',
91
 
             self.repository.bzrdir.root_transport,
92
 
             self.repository._format)
 
101
        note('checked branch %s format %s',
 
102
             self.branch.base, 
 
103
             self.branch._format)
 
104
 
93
105
        note('%6d revisions', self.checked_rev_cnt)
94
106
        note('%6d unique file texts', self.checked_text_cnt)
95
107
        note('%6d repeated file texts', self.repeated_text_cnt)
106
118
                for ghost in self.ghosts:
107
119
                    note('      %s', ghost)
108
120
        if len(self.missing_parent_links):
109
 
            note('%6d revisions missing parents in ancestry',
 
121
            note('%6d revisions missing parents in ancestry', 
110
122
                 len(self.missing_parent_links))
111
123
            if verbose:
112
124
                for link, linkers in self.missing_parent_links.items():
118
130
        """Check one revision.
119
131
 
120
132
        rev_id - the one to check
 
133
 
 
134
        last_rev_id - the previous one on the mainline, if any.
121
135
        """
122
 
        rev = self.repository.get_revision(rev_id)
 
136
 
 
137
        # mutter('    revision {%s}', rev_id)
 
138
        branch = self.branch
 
139
        try:
 
140
            rev_history_position = self.history.index(rev_id)
 
141
        except ValueError:
 
142
            rev_history_position = None
 
143
        last_rev_id = None
 
144
        if rev_history_position:
 
145
            rev = branch.repository.get_revision(rev_id)
 
146
            if rev_history_position > 0:
 
147
                last_rev_id = self.history[rev_history_position - 1]
 
148
        else:
 
149
            rev = branch.repository.get_revision(rev_id)
123
150
                
124
151
        if rev.revision_id != rev_id:
125
152
            raise BzrCheckError('wrong internal revision id in revision {%s}'
126
153
                                % rev_id)
127
154
 
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)
 
155
        # check the previous history entry is a parent of this entry
 
156
        if rev.parent_ids:
 
157
            if last_rev_id is not None:
 
158
                for parent_id in rev.parent_ids:
 
159
                    if parent_id == last_rev_id:
 
160
                        break
141
161
                else:
142
 
                    self.ghosts.append(rev_id)
 
162
                    raise BzrCheckError("previous revision {%s} not listed among "
 
163
                                        "parents of {%s}"
 
164
                                        % (last_rev_id, rev_id))
 
165
            for parent in rev.parent_ids:
 
166
                if not parent in self.planned_revisions:
 
167
                    missing_links = self.missing_parent_links.get(parent, [])
 
168
                    missing_links.append(rev_id)
 
169
                    self.missing_parent_links[parent] = missing_links
 
170
                    # list based so somewhat slow,
 
171
                    # TODO have a planned_revisions list and set.
 
172
                    if self.branch.has_revision(parent):
 
173
                        missing_ancestry = self.repository.get_ancestry(parent)
 
174
                        for missing in missing_ancestry:
 
175
                            if (missing is not None 
 
176
                                and missing not in self.planned_revisions):
 
177
                                self.planned_revisions.append(missing)
 
178
                    else:
 
179
                        self.ghosts.append(rev_id)
 
180
        elif last_rev_id:
 
181
            raise BzrCheckError("revision {%s} has no parents listed "
 
182
                                "but preceded by {%s}"
 
183
                                % (rev_id, last_rev_id))
143
184
 
144
185
        if rev.inventory_sha1:
145
 
            inv_sha1 = self.repository.get_inventory_sha1(rev_id)
 
186
            inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
146
187
            if inv_sha1 != rev.inventory_sha1:
147
188
                raise BzrCheckError('Inventory sha1 hash doesn\'t match'
148
189
                    ' value in revision {%s}' % rev_id)
 
190
        else:
 
191
            missing_inventory_sha_cnt += 1
 
192
            mutter("no inventory_sha1 on revision {%s}", rev_id)
149
193
        self._check_revision_tree(rev_id)
150
194
        self.checked_rev_cnt += 1
151
195
 
154
198
        """
155
199
        n_weaves = 1
156
200
        weave_ids = []
157
 
        if self.repository.weave_store.listable():
158
 
            weave_ids = list(self.repository.weave_store)
 
201
        if self.branch.repository.weave_store.listable():
 
202
            weave_ids = list(self.branch.repository.weave_store)
159
203
            n_weaves = len(weave_ids)
160
204
        self.progress.update('checking weave', 0, n_weaves)
161
205
        self.inventory_weave.check(progress_bar=self.progress)
162
206
        for i, weave_id in enumerate(weave_ids):
163
207
            self.progress.update('checking weave', i, n_weaves)
164
 
            w = self.repository.weave_store.get_weave(weave_id,
165
 
                    self.repository.get_transaction())
 
208
            w = self.branch.repository.weave_store.get_weave(weave_id,
 
209
                    self.branch.repository.get_transaction())
166
210
            # No progress here, because it looks ugly.
167
211
            w.check()
168
212
            self.checked_weaves[weave_id] = True
169
213
 
170
214
    def _check_revision_tree(self, rev_id):
171
 
        tree = self.repository.revision_tree(rev_id)
 
215
        tree = self.branch.repository.revision_tree(rev_id)
172
216
        inv = tree.inventory
173
217
        seen_ids = {}
174
218
        for file_id in inv:
190
234
 
191
235
 
192
236
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)
 
237
    """Run consistency checks on a branch."""
 
238
    checker = Check(branch)
 
239
    checker.check()
 
240
    checker.report_results(verbose)