← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/check.py

  • Committer: Robert Collins
  • Date: 2005-10-18 05:26:22 UTC
  • mto: This revision was merged to the branch mainline in revision 1463.
  • Revision ID: robertc@robertcollins.net-20051018052622-653d638c9e26fde4
fix broken tests

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
24
25
# TODO: Check revision, inventory and entry objects have all 
25
26
# required fields.
26
27
 
27
 
# TODO: Get every revision in the revision-store even if they're not
28
 
# referenced by history and make sure they're all valid.
29
 
 
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
28
 
35
29
import bzrlib.ui
36
30
from bzrlib.trace import note, warning
38
32
from bzrlib.trace import mutter
39
33
from bzrlib.errors import BzrCheckError, NoSuchRevision
40
34
from bzrlib.inventory import ROOT_ID
 
35
from bzrlib.branch import gen_root_id
41
36
 
42
37
 
43
38
class Check(object):
44
39
    """Check a branch"""
45
40
 
46
 
    # The Check object interacts with InventoryEntry.check, etc.
47
 
 
48
41
    def __init__(self, branch):
49
42
        self.branch = branch
50
 
        self.repository = branch.repository
51
43
        self.checked_text_cnt = 0
52
44
        self.checked_rev_cnt = 0
53
45
        self.ghosts = []
55
47
        self.missing_parent_links = {}
56
48
        self.missing_inventory_sha_cnt = 0
57
49
        self.missing_revision_cnt = 0
58
 
        # maps (file-id, version) -> sha1; used by InventoryFile._check
 
50
        # maps (file-id, version) -> sha1
59
51
        self.checked_texts = {}
60
 
        self.checked_weaves = {}
61
52
 
62
53
    def check(self):
63
54
        self.branch.lock_read()
64
 
        self.progress = bzrlib.ui.ui_factory.nested_progress_bar()
65
55
        try:
66
 
            self.progress.update('retrieving inventory', 0, 0)
67
 
            # do not put in init, as it should be done with progess,
68
 
            # and inside the lock.
69
 
            self.inventory_weave = self.branch.repository.get_inventory_weave()
70
56
            self.history = self.branch.revision_history()
71
57
            if not len(self.history):
72
58
                # nothing to see here
73
59
                return
74
 
            self.plan_revisions()
 
60
            self.planned_revisions = self.branch.get_ancestry(self.history[-1])
 
61
            self.planned_revisions.remove(None)
75
62
            revno = 0
76
 
            self.check_weaves()
 
63
    
 
64
            self.progress = bzrlib.ui.ui_factory.progress_bar()
77
65
            while revno < len(self.planned_revisions):
78
66
                rev_id = self.planned_revisions[revno]
79
67
                self.progress.update('checking revision', revno,
80
68
                                     len(self.planned_revisions))
81
69
                revno += 1
82
70
                self.check_one_rev(rev_id)
 
71
            self.progress.clear()
83
72
        finally:
84
 
            self.progress.finished()
85
73
            self.branch.unlock()
86
74
 
87
 
    def plan_revisions(self):
88
 
        repository = self.branch.repository
89
 
        self.planned_revisions = set(repository.all_revision_ids())
90
 
        self.progress.clear()
91
 
        inventoried = set(self.inventory_weave.versions())
92
 
        awol = self.planned_revisions - inventoried
93
 
        if len(awol) > 0:
94
 
            raise BzrCheckError('Stored revisions missing from inventory'
95
 
                '{%s}' % ','.join([f for f in awol]))
96
 
        self.planned_revisions = list(self.planned_revisions)
97
 
 
98
75
    def report_results(self, verbose):
99
 
        note('checked branch %s format %s',
 
76
        note('checked branch %s format %d',
100
77
             self.branch.base, 
101
 
             self.branch._format)
 
78
             self.branch._branch_format)
102
79
 
103
80
        note('%6d revisions', self.checked_rev_cnt)
104
81
        note('%6d unique file texts', self.checked_text_cnt)
105
82
        note('%6d repeated file texts', self.repeated_text_cnt)
106
 
        note('%6d weaves', len(self.checked_weaves))
107
83
        if self.missing_inventory_sha_cnt:
108
84
            note('%6d revisions are missing inventory_sha1',
109
85
                 self.missing_inventory_sha_cnt)
132
108
        last_rev_id - the previous one on the mainline, if any.
133
109
        """
134
110
 
135
 
        # mutter('    revision {%s}', rev_id)
 
111
        # mutter('    revision {%s}' % rev_id)
136
112
        branch = self.branch
137
113
        try:
138
114
            rev_history_position = self.history.index(rev_id)
140
116
            rev_history_position = None
141
117
        last_rev_id = None
142
118
        if rev_history_position:
143
 
            rev = branch.repository.get_revision(rev_id)
 
119
            rev = branch.get_revision(rev_id)
144
120
            if rev_history_position > 0:
145
121
                last_rev_id = self.history[rev_history_position - 1]
146
122
        else:
147
 
            rev = branch.repository.get_revision(rev_id)
 
123
            rev = branch.get_revision(rev_id)
148
124
                
149
125
        if rev.revision_id != rev_id:
150
126
            raise BzrCheckError('wrong internal revision id in revision {%s}'
152
128
 
153
129
        # check the previous history entry is a parent of this entry
154
130
        if rev.parent_ids:
 
131
            if last_rev_id is None and rev_history_position is not None:
 
132
                # what if the start is a ghost ? i.e. conceptually the 
 
133
                # baz imports
 
134
                raise BzrCheckError("revision {%s} has %d parents, but is the "
 
135
                                    "start of the branch"
 
136
                                    % (rev_id, len(rev.parent_ids)))
155
137
            if last_rev_id is not None:
156
138
                for parent_id in rev.parent_ids:
157
139
                    if parent_id == last_rev_id:
165
147
                    missing_links = self.missing_parent_links.get(parent, [])
166
148
                    missing_links.append(rev_id)
167
149
                    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)
 
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)
172
153
                        for missing in missing_ancestry:
173
154
                            if (missing is not None 
174
155
                                and missing not in self.planned_revisions):
181
162
                                % (rev_id, last_rev_id))
182
163
 
183
164
        if rev.inventory_sha1:
184
 
            inv_sha1 = branch.repository.get_inventory_sha1(rev_id)
 
165
            inv_sha1 = branch.get_inventory_sha1(rev_id)
185
166
            if inv_sha1 != rev.inventory_sha1:
186
167
                raise BzrCheckError('Inventory sha1 hash doesn\'t match'
187
168
                    ' value in revision {%s}' % rev_id)
188
169
        else:
189
170
            missing_inventory_sha_cnt += 1
190
 
            mutter("no inventory_sha1 on revision {%s}", rev_id)
 
171
            mutter("no inventory_sha1 on revision {%s}" % rev_id)
191
172
        self._check_revision_tree(rev_id)
192
173
        self.checked_rev_cnt += 1
193
174
 
194
 
    def check_weaves(self):
195
 
        """Check all the weaves we can get our hands on.
196
 
        """
197
 
        n_weaves = 1
198
 
        weave_ids = []
199
 
        if self.branch.repository.weave_store.listable():
200
 
            weave_ids = list(self.branch.repository.weave_store)
201
 
            n_weaves = len(weave_ids)
202
 
        self.progress.update('checking weave', 0, n_weaves)
203
 
        self.inventory_weave.check(progress_bar=self.progress)
204
 
        for i, weave_id in enumerate(weave_ids):
205
 
            self.progress.update('checking weave', i, n_weaves)
206
 
            w = self.branch.repository.weave_store.get_weave(weave_id,
207
 
                    self.branch.repository.get_transaction())
208
 
            # No progress here, because it looks ugly.
209
 
            w.check()
210
 
            self.checked_weaves[weave_id] = True
211
 
 
212
175
    def _check_revision_tree(self, rev_id):
213
 
        tree = self.branch.repository.revision_tree(rev_id)
 
176
        tree = self.branch.revision_tree(rev_id)
214
177
        inv = tree.inventory
215
178
        seen_ids = {}
216
179
        for file_id in inv: