##// END OF EJS Templates
verify: don't trip over binary files starting with 01 0a
Matt Mackall -
r7675:011e69b9 default
parent child Browse files
Show More
@@ -1,239 +1,239 b''
1 # verify.py - repository integrity checking for Mercurial
1 # verify.py - repository integrity checking for Mercurial
2 #
2 #
3 # Copyright 2006, 2007 Matt Mackall <mpm@selenic.com>
3 # Copyright 2006, 2007 Matt Mackall <mpm@selenic.com>
4 #
4 #
5 # This software may be used and distributed according to the terms
5 # This software may be used and distributed according to the terms
6 # of the GNU General Public License, incorporated herein by reference.
6 # of the GNU General Public License, incorporated herein by reference.
7
7
8 from node import nullid, short
8 from node import nullid, short
9 from i18n import _
9 from i18n import _
10 import revlog, util
10 import revlog, util
11
11
12 def verify(repo):
12 def verify(repo):
13 lock = repo.lock()
13 lock = repo.lock()
14 try:
14 try:
15 return _verify(repo)
15 return _verify(repo)
16 finally:
16 finally:
17 del lock
17 del lock
18
18
19 def _verify(repo):
19 def _verify(repo):
20 mflinkrevs = {}
20 mflinkrevs = {}
21 filelinkrevs = {}
21 filelinkrevs = {}
22 filenodes = {}
22 filenodes = {}
23 revisions = 0
23 revisions = 0
24 badrevs = {}
24 badrevs = {}
25 errors = [0]
25 errors = [0]
26 warnings = [0]
26 warnings = [0]
27 ui = repo.ui
27 ui = repo.ui
28 cl = repo.changelog
28 cl = repo.changelog
29 mf = repo.manifest
29 mf = repo.manifest
30
30
31 if not repo.cancopy():
31 if not repo.cancopy():
32 raise util.Abort(_("cannot verify bundle or remote repos"))
32 raise util.Abort(_("cannot verify bundle or remote repos"))
33
33
34 def err(linkrev, msg, filename=None):
34 def err(linkrev, msg, filename=None):
35 if linkrev != None:
35 if linkrev != None:
36 badrevs[linkrev] = True
36 badrevs[linkrev] = True
37 else:
37 else:
38 linkrev = '?'
38 linkrev = '?'
39 msg = "%s: %s" % (linkrev, msg)
39 msg = "%s: %s" % (linkrev, msg)
40 if filename:
40 if filename:
41 msg = "%s@%s" % (filename, msg)
41 msg = "%s@%s" % (filename, msg)
42 ui.warn(" " + msg + "\n")
42 ui.warn(" " + msg + "\n")
43 errors[0] += 1
43 errors[0] += 1
44
44
45 def exc(linkrev, msg, inst, filename=None):
45 def exc(linkrev, msg, inst, filename=None):
46 if isinstance(inst, KeyboardInterrupt):
46 if isinstance(inst, KeyboardInterrupt):
47 ui.warn(_("interrupted"))
47 ui.warn(_("interrupted"))
48 raise
48 raise
49 err(linkrev, "%s: %s" % (msg, inst), filename)
49 err(linkrev, "%s: %s" % (msg, inst), filename)
50
50
51 def warn(msg):
51 def warn(msg):
52 ui.warn(msg + "\n")
52 ui.warn(msg + "\n")
53 warnings[0] += 1
53 warnings[0] += 1
54
54
55 def checklog(obj, name):
55 def checklog(obj, name):
56 if not len(obj) and (havecl or havemf):
56 if not len(obj) and (havecl or havemf):
57 err(0, _("empty or missing %s") % name)
57 err(0, _("empty or missing %s") % name)
58 return
58 return
59
59
60 d = obj.checksize()
60 d = obj.checksize()
61 if d[0]:
61 if d[0]:
62 err(None, _("data length off by %d bytes") % d[0], name)
62 err(None, _("data length off by %d bytes") % d[0], name)
63 if d[1]:
63 if d[1]:
64 err(None, _("index contains %d extra bytes") % d[1], name)
64 err(None, _("index contains %d extra bytes") % d[1], name)
65
65
66 if obj.version != revlog.REVLOGV0:
66 if obj.version != revlog.REVLOGV0:
67 if not revlogv1:
67 if not revlogv1:
68 warn(_("warning: `%s' uses revlog format 1") % name)
68 warn(_("warning: `%s' uses revlog format 1") % name)
69 elif revlogv1:
69 elif revlogv1:
70 warn(_("warning: `%s' uses revlog format 0") % name)
70 warn(_("warning: `%s' uses revlog format 0") % name)
71
71
72 def checkentry(obj, i, node, seen, linkrevs, f):
72 def checkentry(obj, i, node, seen, linkrevs, f):
73 lr = obj.linkrev(obj.rev(node))
73 lr = obj.linkrev(obj.rev(node))
74 if lr < 0 or (havecl and lr not in linkrevs):
74 if lr < 0 or (havecl and lr not in linkrevs):
75 t = "unexpected"
75 t = "unexpected"
76 if lr < 0 or lr >= len(cl):
76 if lr < 0 or lr >= len(cl):
77 t = "nonexistent"
77 t = "nonexistent"
78 err(None, _("rev %d point to %s changeset %d") % (i, t, lr), f)
78 err(None, _("rev %d point to %s changeset %d") % (i, t, lr), f)
79 if linkrevs:
79 if linkrevs:
80 warn(_(" (expected %s)") % " ".join(map(str,linkrevs)))
80 warn(_(" (expected %s)") % " ".join(map(str,linkrevs)))
81 lr = None # can't be trusted
81 lr = None # can't be trusted
82
82
83 try:
83 try:
84 p1, p2 = obj.parents(node)
84 p1, p2 = obj.parents(node)
85 if p1 not in seen and p1 != nullid:
85 if p1 not in seen and p1 != nullid:
86 err(lr, _("unknown parent 1 %s of %s") %
86 err(lr, _("unknown parent 1 %s of %s") %
87 (short(p1), short(n)), f)
87 (short(p1), short(n)), f)
88 if p2 not in seen and p2 != nullid:
88 if p2 not in seen and p2 != nullid:
89 err(lr, _("unknown parent 2 %s of %s") %
89 err(lr, _("unknown parent 2 %s of %s") %
90 (short(p2), short(p1)), f)
90 (short(p2), short(p1)), f)
91 except Exception, inst:
91 except Exception, inst:
92 exc(lr, _("checking parents of %s") % short(node), inst, f)
92 exc(lr, _("checking parents of %s") % short(node), inst, f)
93
93
94 if node in seen:
94 if node in seen:
95 err(lr, _("duplicate revision %d (%d)") % (i, seen[n]), f)
95 err(lr, _("duplicate revision %d (%d)") % (i, seen[n]), f)
96 seen[n] = i
96 seen[n] = i
97 return lr
97 return lr
98
98
99 revlogv1 = cl.version != revlog.REVLOGV0
99 revlogv1 = cl.version != revlog.REVLOGV0
100 if ui.verbose or not revlogv1:
100 if ui.verbose or not revlogv1:
101 ui.status(_("repository uses revlog format %d\n") %
101 ui.status(_("repository uses revlog format %d\n") %
102 (revlogv1 and 1 or 0))
102 (revlogv1 and 1 or 0))
103
103
104 havecl = len(cl) > 0
104 havecl = len(cl) > 0
105 havemf = len(mf) > 0
105 havemf = len(mf) > 0
106
106
107 ui.status(_("checking changesets\n"))
107 ui.status(_("checking changesets\n"))
108 seen = {}
108 seen = {}
109 checklog(cl, "changelog")
109 checklog(cl, "changelog")
110 for i in repo:
110 for i in repo:
111 n = cl.node(i)
111 n = cl.node(i)
112 checkentry(cl, i, n, seen, [i], "changelog")
112 checkentry(cl, i, n, seen, [i], "changelog")
113
113
114 try:
114 try:
115 changes = cl.read(n)
115 changes = cl.read(n)
116 mflinkrevs.setdefault(changes[0], []).append(i)
116 mflinkrevs.setdefault(changes[0], []).append(i)
117 for f in changes[3]:
117 for f in changes[3]:
118 filelinkrevs.setdefault(f, []).append(i)
118 filelinkrevs.setdefault(f, []).append(i)
119 except Exception, inst:
119 except Exception, inst:
120 exc(i, _("unpacking changeset %s") % short(n), inst)
120 exc(i, _("unpacking changeset %s") % short(n), inst)
121
121
122 ui.status(_("checking manifests\n"))
122 ui.status(_("checking manifests\n"))
123 seen = {}
123 seen = {}
124 checklog(mf, "manifest")
124 checklog(mf, "manifest")
125 for i in mf:
125 for i in mf:
126 n = mf.node(i)
126 n = mf.node(i)
127 lr = checkentry(mf, i, n, seen, mflinkrevs.get(n, []), "manifest")
127 lr = checkentry(mf, i, n, seen, mflinkrevs.get(n, []), "manifest")
128 if n in mflinkrevs:
128 if n in mflinkrevs:
129 del mflinkrevs[n]
129 del mflinkrevs[n]
130
130
131 try:
131 try:
132 for f, fn in mf.readdelta(n).iteritems():
132 for f, fn in mf.readdelta(n).iteritems():
133 if not f:
133 if not f:
134 err(lr, _("file without name in manifest"))
134 err(lr, _("file without name in manifest"))
135 elif f != "/dev/null":
135 elif f != "/dev/null":
136 fns = filenodes.setdefault(f, {})
136 fns = filenodes.setdefault(f, {})
137 if fn not in fns:
137 if fn not in fns:
138 fns[fn] = i
138 fns[fn] = i
139 except Exception, inst:
139 except Exception, inst:
140 exc(lr, _("reading manifest delta %s") % short(n), inst)
140 exc(lr, _("reading manifest delta %s") % short(n), inst)
141
141
142 ui.status(_("crosschecking files in changesets and manifests\n"))
142 ui.status(_("crosschecking files in changesets and manifests\n"))
143
143
144 if havemf:
144 if havemf:
145 for c, m in util.sort([(c, m) for m in mflinkrevs for c in mflinkrevs[m]]):
145 for c, m in util.sort([(c, m) for m in mflinkrevs for c in mflinkrevs[m]]):
146 err(c, _("changeset refers to unknown manifest %s") % short(m))
146 err(c, _("changeset refers to unknown manifest %s") % short(m))
147 del mflinkrevs
147 del mflinkrevs
148
148
149 for f in util.sort(filelinkrevs):
149 for f in util.sort(filelinkrevs):
150 if f not in filenodes:
150 if f not in filenodes:
151 lr = filelinkrevs[f][0]
151 lr = filelinkrevs[f][0]
152 err(lr, _("in changeset but not in manifest"), f)
152 err(lr, _("in changeset but not in manifest"), f)
153
153
154 if havecl:
154 if havecl:
155 for f in util.sort(filenodes):
155 for f in util.sort(filenodes):
156 if f not in filelinkrevs:
156 if f not in filelinkrevs:
157 try:
157 try:
158 fl = repo.file(f)
158 fl = repo.file(f)
159 lr = min([fl.linkrev(fl.rev(n)) for n in filenodes[f]])
159 lr = min([fl.linkrev(fl.rev(n)) for n in filenodes[f]])
160 except:
160 except:
161 lr = None
161 lr = None
162 err(lr, _("in manifest but not in changeset"), f)
162 err(lr, _("in manifest but not in changeset"), f)
163
163
164 ui.status(_("checking files\n"))
164 ui.status(_("checking files\n"))
165
165
166 storefiles = {}
166 storefiles = {}
167 for f, f2, size in repo.store.datafiles():
167 for f, f2, size in repo.store.datafiles():
168 if not f:
168 if not f:
169 err(None, _("cannot decode filename '%s'") % f2)
169 err(None, _("cannot decode filename '%s'") % f2)
170 elif size > 0:
170 elif size > 0:
171 storefiles[f] = True
171 storefiles[f] = True
172
172
173 files = util.sort(util.unique(filenodes.keys() + filelinkrevs.keys()))
173 files = util.sort(util.unique(filenodes.keys() + filelinkrevs.keys()))
174 for f in files:
174 for f in files:
175 fl = repo.file(f)
175 fl = repo.file(f)
176
176
177 for ff in fl.files():
177 for ff in fl.files():
178 try:
178 try:
179 del storefiles[ff]
179 del storefiles[ff]
180 except KeyError:
180 except KeyError:
181 err(0, _("missing revlog!"), ff)
181 err(0, _("missing revlog!"), ff)
182
182
183 checklog(fl, f)
183 checklog(fl, f)
184 seen = {}
184 seen = {}
185 for i in fl:
185 for i in fl:
186 revisions += 1
186 revisions += 1
187 n = fl.node(i)
187 n = fl.node(i)
188 lr = checkentry(fl, i, n, seen, filelinkrevs.get(f, []), f)
188 lr = checkentry(fl, i, n, seen, filelinkrevs.get(f, []), f)
189 if f in filenodes:
189 if f in filenodes:
190 if havemf and n not in filenodes[f]:
190 if havemf and n not in filenodes[f]:
191 err(lr, _("%s not in manifests") % (short(n)), f)
191 err(lr, _("%s not in manifests") % (short(n)), f)
192 else:
192 else:
193 del filenodes[f][n]
193 del filenodes[f][n]
194
194
195 # verify contents
195 # verify contents
196 try:
196 try:
197 t = fl.read(n)
197 t = fl.read(n)
198 rp = fl.renamed(n)
198 rp = fl.renamed(n)
199 if len(t) != fl.size(i):
199 if len(t) != fl.size(i):
200 if not fl._readmeta(n): # ancient copy?
200 if len(fl.revision(n)) != fl.size(i):
201 err(lr, _("unpacked size is %s, %s expected") %
201 err(lr, _("unpacked size is %s, %s expected") %
202 (len(t), fl.size(i)), f)
202 (len(t), fl.size(i)), f)
203 except Exception, inst:
203 except Exception, inst:
204 exc(lr, _("unpacking %s") % short(n), inst, f)
204 exc(lr, _("unpacking %s") % short(n), inst, f)
205
205
206 # check renames
206 # check renames
207 try:
207 try:
208 if rp:
208 if rp:
209 fl2 = repo.file(rp[0])
209 fl2 = repo.file(rp[0])
210 if not len(fl2):
210 if not len(fl2):
211 err(lr, _("empty or missing copy source revlog %s:%s")
211 err(lr, _("empty or missing copy source revlog %s:%s")
212 % (rp[0], short(rp[1])), f)
212 % (rp[0], short(rp[1])), f)
213 elif rp[1] == nullid:
213 elif rp[1] == nullid:
214 warn(_("warning: %s@%s: copy source revision is nullid %s:%s")
214 warn(_("warning: %s@%s: copy source revision is nullid %s:%s")
215 % (f, lr, rp[0], short(rp[1])))
215 % (f, lr, rp[0], short(rp[1])))
216 else:
216 else:
217 rev = fl2.rev(rp[1])
217 rev = fl2.rev(rp[1])
218 except Exception, inst:
218 except Exception, inst:
219 exc(lr, _("checking rename of %s") % short(n), inst, f)
219 exc(lr, _("checking rename of %s") % short(n), inst, f)
220
220
221 # cross-check
221 # cross-check
222 if f in filenodes:
222 if f in filenodes:
223 fns = [(mf.linkrev(l), n) for n,l in filenodes[f].items()]
223 fns = [(mf.linkrev(l), n) for n,l in filenodes[f].items()]
224 for lr, node in util.sort(fns):
224 for lr, node in util.sort(fns):
225 err(lr, _("%s in manifests not found") % short(node), f)
225 err(lr, _("%s in manifests not found") % short(node), f)
226
226
227 for f in storefiles:
227 for f in storefiles:
228 warn(_("warning: orphan revlog '%s'") % f)
228 warn(_("warning: orphan revlog '%s'") % f)
229
229
230 ui.status(_("%d files, %d changesets, %d total revisions\n") %
230 ui.status(_("%d files, %d changesets, %d total revisions\n") %
231 (len(files), len(cl), revisions))
231 (len(files), len(cl), revisions))
232 if warnings[0]:
232 if warnings[0]:
233 ui.warn(_("%d warnings encountered!\n") % warnings[0])
233 ui.warn(_("%d warnings encountered!\n") % warnings[0])
234 if errors[0]:
234 if errors[0]:
235 ui.warn(_("%d integrity errors encountered!\n") % errors[0])
235 ui.warn(_("%d integrity errors encountered!\n") % errors[0])
236 if badrevs:
236 if badrevs:
237 ui.warn(_("(first damaged changeset appears to be %d)\n")
237 ui.warn(_("(first damaged changeset appears to be %d)\n")
238 % min(badrevs))
238 % min(badrevs))
239 return 1
239 return 1
General Comments 0
You need to be logged in to leave comments. Login now