##// END OF EJS Templates
verify: improve handling of empty or missing files...
Matt Mackall -
r5541:ceaa752f default
parent child Browse files
Show More
@@ -1,250 +1,274 b''
1 1 # verify.py - repository integrity checking for Mercurial
2 2 #
3 3 # Copyright 2006, 2007 Matt Mackall <mpm@selenic.com>
4 4 #
5 5 # This software may be used and distributed according to the terms
6 6 # of the GNU General Public License, incorporated herein by reference.
7 7
8 8 from node import *
9 9 from i18n import _
10 10 import revlog
11 11
12 12 def verify(repo):
13 13 lock = repo.lock()
14 14 try:
15 15 return _verify(repo)
16 16 finally:
17 17 del lock
18 18
19 19 def _verify(repo):
20 20 filelinkrevs = {}
21 21 filenodes = {}
22 22 changesets = revisions = files = 0
23 23 firstbad = [None]
24 24 errors = [0]
25 25 warnings = [0]
26 26 neededmanifests = {}
27 27
28 28 def err(linkrev, msg, filename=None):
29 29 if linkrev != None:
30 30 if firstbad[0] != None:
31 31 firstbad[0] = min(firstbad[0], linkrev)
32 32 else:
33 33 firstbad[0] = linkrev
34 34 else:
35 35 linkrev = "?"
36 36 msg = "%s: %s" % (linkrev, msg)
37 37 if filename:
38 38 msg = "%s@%s" % (filename, msg)
39 39 repo.ui.warn(" " + msg + "\n")
40 40 errors[0] += 1
41 41
42 42 def warn(msg):
43 43 repo.ui.warn(msg + "\n")
44 44 warnings[0] += 1
45 45
46 46 def checksize(obj, name):
47 47 d = obj.checksize()
48 48 if d[0]:
49 49 err(None, _("data length off by %d bytes") % d[0], name)
50 50 if d[1]:
51 51 err(None, _("index contains %d extra bytes") % d[1], name)
52 52
53 53 def checkversion(obj, name):
54 54 if obj.version != revlog.REVLOGV0:
55 55 if not revlogv1:
56 56 warn(_("warning: `%s' uses revlog format 1") % name)
57 57 elif revlogv1:
58 58 warn(_("warning: `%s' uses revlog format 0") % name)
59 59
60 60 revlogv1 = repo.changelog.version != revlog.REVLOGV0
61 61 if repo.ui.verbose or not revlogv1:
62 62 repo.ui.status(_("repository uses revlog format %d\n") %
63 63 (revlogv1 and 1 or 0))
64 64
65 havecl = havemf = 1
65 66 seen = {}
66 67 repo.ui.status(_("checking changesets\n"))
67 checksize(repo.changelog, "changelog")
68 if repo.changelog.count() == 0 and repo.manifest.count() > 1:
69 havecl = 0
70 err(0, _("empty or missing 00changelog.i"))
71 else:
72 checksize(repo.changelog, "changelog")
68 73
69 74 for i in xrange(repo.changelog.count()):
70 75 changesets += 1
71 76 n = repo.changelog.node(i)
72 77 l = repo.changelog.linkrev(n)
73 78 if l != i:
74 79 err(i, _("incorrect link (%d) for changeset") %(l))
75 80 if n in seen:
76 81 err(i, _("duplicates changeset at revision %d") % seen[n])
77 82 seen[n] = i
78 83
79 84 for p in repo.changelog.parents(n):
80 85 if p not in repo.changelog.nodemap:
81 86 err(i, _("changeset has unknown parent %s") % short(p))
82 87 try:
83 88 changes = repo.changelog.read(n)
84 89 except KeyboardInterrupt:
85 90 repo.ui.warn(_("interrupted"))
86 91 raise
87 92 except Exception, inst:
88 93 err(i, _("unpacking changeset: %s") % inst)
89 94 continue
90 95
91 96 if changes[0] not in neededmanifests:
92 97 neededmanifests[changes[0]] = i
93 98
94 99 for f in changes[3]:
95 100 filelinkrevs.setdefault(f, []).append(i)
96 101
97 102 seen = {}
98 103 repo.ui.status(_("checking manifests\n"))
99 checkversion(repo.manifest, "manifest")
100 checksize(repo.manifest, "manifest")
104 if repo.changelog.count() > 0 and repo.manifest.count() == 0:
105 havemf = 0
106 err(0, _("empty or missing 00manifest.i"))
107 else:
108 checkversion(repo.manifest, "manifest")
109 checksize(repo.manifest, "manifest")
101 110
102 111 for i in xrange(repo.manifest.count()):
103 112 n = repo.manifest.node(i)
104 113 l = repo.manifest.linkrev(n)
105 114
106 if l < 0 or l >= repo.changelog.count():
115 if l < 0 or (havecl and l >= repo.changelog.count()):
107 116 err(None, _("bad link (%d) at manifest revision %d") % (l, i))
108 117
109 118 if n in neededmanifests:
110 119 del neededmanifests[n]
111 120
112 121 if n in seen:
113 122 err(l, _("duplicates manifest from %d") % seen[n])
114 123
115 124 seen[n] = l
116 125
117 126 for p in repo.manifest.parents(n):
118 127 if p not in repo.manifest.nodemap:
119 128 err(l, _("manifest has unknown parent %s") % short(p))
120 129
121 130 try:
122 131 for f, fn in repo.manifest.readdelta(n).iteritems():
123 132 fns = filenodes.setdefault(f, {})
124 133 if fn not in fns:
125 134 fns[fn] = n
126 135 except KeyboardInterrupt:
127 136 repo.ui.warn(_("interrupted"))
128 137 raise
129 138 except Exception, inst:
130 139 err(l, _("reading manifest delta: %s") % inst)
131 140 continue
132 141
133 142 repo.ui.status(_("crosschecking files in changesets and manifests\n"))
134 143
135 nm = neededmanifests.items()
136 nm.sort()
137 for m, c in nm:
138 err(m, _("changeset refers to unknown manifest %s") % short(c))
139 del neededmanifests, nm
144 if havemf > 0:
145 nm = [(c, m) for m, c in neededmanifests.items()]
146 nm.sort()
147 for c, m in nm:
148 err(c, _("changeset refers to unknown manifest %s") % short(m))
149 del neededmanifests, nm
140 150
141 for f in filenodes:
142 if f not in filelinkrevs:
143 lrs = [repo.manifest.linkrev(n) for n in filenodes[f]]
144 lrs.sort()
145 err(lrs[0], _("in manifest but not in changeset"), f)
151 if havecl:
152 fl = filenodes.keys()
153 fl.sort()
154 for f in fl:
155 if f not in filelinkrevs:
156 lrs = [repo.manifest.linkrev(n) for n in filenodes[f]]
157 lrs.sort()
158 err(lrs[0], _("in manifest but not in changeset"), f)
159 del fl
146 160
147 for f in filelinkrevs:
148 if f not in filenodes:
149 lr = filelinkrevs[f][0]
150 err(lr, _("in changeset but not in manifest"), f)
161 if havemf:
162 fl = filelinkrevs.keys()
163 fl.sort()
164 for f in fl:
165 if f not in filenodes:
166 lr = filelinkrevs[f][0]
167 err(lr, _("in changeset but not in manifest"), f)
168 del fl
151 169
152 170 repo.ui.status(_("checking files\n"))
153 ff = filenodes.keys()
171 ff = dict.fromkeys(filenodes.keys() + filelinkrevs.keys()).keys()
154 172 ff.sort()
155 173 for f in ff:
156 174 if f == "/dev/null":
157 175 continue
158 176 files += 1
159 177 if not f:
160 lr = repo.manifest.linkrev(filenodes[f][0])
161 err(lr, _("file without name in manifest %s") % short(ff[n]))
178 lr = filelinkrevs[f][0]
179 err(lr, _("file without name in manifest"))
162 180 continue
163 181 fl = repo.file(f)
164 182 checkversion(fl, f)
165 183 checksize(fl, f)
166 184
185 if fl.count() == 0:
186 err(filelinkrevs[f][0], _("empty or missing revlog"), f)
187 continue
188
167 189 seen = {}
168 190 nodes = {nullid: 1}
169 191 for i in xrange(fl.count()):
170 192 revisions += 1
171 193 n = fl.node(i)
172 194 flr = fl.linkrev(n)
173 195
174 if flr not in filelinkrevs.get(f, []):
196 if flr < 0 or (havecl and flr not in filelinkrevs.get(f, [])):
175 197 if flr < 0 or flr >= repo.changelog.count():
176 198 err(None, _("rev %d point to nonexistent changeset %d")
177 199 % (i, flr), f)
178 200 else:
179 201 err(None, _("rev %d points to unexpected changeset %d")
180 202 % (i, flr), f)
181 203 if f in filelinkrevs:
182 204 warn(_(" (expected %s)") % filelinkrevs[f][0])
183 205 flr = None # can't be trusted
184 206 else:
185 filelinkrevs[f].remove(flr)
207 if havecl:
208 filelinkrevs[f].remove(flr)
186 209
187 210 if n in seen:
188 211 err(flr, _("duplicate revision %d") % i, f)
189 if n not in filenodes[f]:
190 err(flr, _("%s not in manifests") % (short(n)), f)
191 else:
192 del filenodes[f][n]
212 if f in filenodes:
213 if havemf and n not in filenodes[f]:
214 err(flr, _("%s not in manifests") % (short(n)), f)
215 else:
216 del filenodes[f][n]
193 217
194 218 # verify contents
195 219 try:
196 220 t = fl.read(n)
197 221 except KeyboardInterrupt:
198 222 repo.ui.warn(_("interrupted"))
199 223 raise
200 224 except Exception, inst:
201 225 err(flr, _("unpacking %s: %s") % (short(n), inst), f)
202 226
203 227 # verify parents
204 228 try:
205 229 (p1, p2) = fl.parents(n)
206 230 if p1 not in nodes:
207 231 err(flr, _("unknown parent 1 %s of %s") %
208 232 (short(p1), short(n)), f)
209 233 if p2 not in nodes:
210 234 err(flr, _("unknown parent 2 %s of %s") %
211 235 (short(p2), short(p1)), f)
212 236 except KeyboardInterrupt:
213 237 repo.ui.warn(_("interrupted"))
214 238 raise
215 239 except Exception, inst:
216 240 err(flr, _("checking parents of %s: %s") % (short(n), inst), f)
217 241 nodes[n] = 1
218 242
219 243 # check renames
220 244 try:
221 245 rp = fl.renamed(n)
222 246 if rp:
223 247 fl2 = repo.file(rp[0])
224 248 rev = fl2.rev(rp[1])
225 249 except KeyboardInterrupt:
226 250 repo.ui.warn(_("interrupted"))
227 251 raise
228 252 except Exception, inst:
229 253 err(flr, _("checking rename of %s: %s") %
230 254 (short(n), inst), f)
231 255
232 256 # cross-check
233 fns = [(repo.manifest.linkrev(filenodes[f][n]), n)
234 for n in filenodes[f]]
235 fns.sort()
236 for lr, node in fns:
237 err(lr, _("%s in manifests not found") % short(node), f)
257 if f in filenodes:
258 fns = [(repo.manifest.linkrev(filenodes[f][n]), n)
259 for n in filenodes[f]]
260 fns.sort()
261 for lr, node in fns:
262 err(lr, _("%s in manifests not found") % short(node), f)
238 263
239 264 repo.ui.status(_("%d files, %d changesets, %d total revisions\n") %
240 265 (files, changesets, revisions))
241 266
242 267 if warnings[0]:
243 268 repo.ui.warn(_("%d warnings encountered!\n") % warnings[0])
244 269 if errors[0]:
245 270 repo.ui.warn(_("%d integrity errors encountered!\n") % errors[0])
246 271 if firstbad[0]:
247 272 repo.ui.warn(_("(first damaged changeset appears to be %d)\n")
248 273 % firstbad[0])
249 274 return 1
250
General Comments 0
You need to be logged in to leave comments. Login now