annotate metacheck.pl @ 13:f33fb7f1898e

added mapping for spanish lang
author casties
date Wed, 01 Jun 2005 14:24:19 +0200
parents b7259a1c85aa
children fdf4ceb36db1
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
1 #!/usr/local/bin/perl -w
30497c6a3eca Initial revision
casties
parents:
diff changeset
2
30497c6a3eca Initial revision
casties
parents:
diff changeset
3 use strict;
30497c6a3eca Initial revision
casties
parents:
diff changeset
4 use XML::LibXML;
30497c6a3eca Initial revision
casties
parents:
diff changeset
5
30497c6a3eca Initial revision
casties
parents:
diff changeset
6 use lib '/usr/local/mpiwg/archive';
30497c6a3eca Initial revision
casties
parents:
diff changeset
7 use MPIWGStor;
30497c6a3eca Initial revision
casties
parents:
diff changeset
8
30497c6a3eca Initial revision
casties
parents:
diff changeset
9 # make output unbuffered
30497c6a3eca Initial revision
casties
parents:
diff changeset
10 $|=1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
11
30497c6a3eca Initial revision
casties
parents:
diff changeset
12 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
13 # internal parameters
30497c6a3eca Initial revision
casties
parents:
diff changeset
14 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
15
30497c6a3eca Initial revision
casties
parents:
diff changeset
16 # program version
2
b7259a1c85aa version number upped
casties
parents: 1
diff changeset
17 my $version = "0.5.3 (8.7.2004)";
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
18
30497c6a3eca Initial revision
casties
parents:
diff changeset
19 # read command line parameters
30497c6a3eca Initial revision
casties
parents:
diff changeset
20 my $args = MPIWGStor::parseargs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
21
30497c6a3eca Initial revision
casties
parents:
diff changeset
22 # debug level
30497c6a3eca Initial revision
casties
parents:
diff changeset
23 $debug = (exists $$args{'debug'}) ? $$args{'debug'} : 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
24
30497c6a3eca Initial revision
casties
parents:
diff changeset
25 # check only or fix index file also
30497c6a3eca Initial revision
casties
parents:
diff changeset
26 my $check_only = (exists $$args{'checkonly'}) ? $$args{'checkonly'} : 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
27 # add file tags for missing files
30497c6a3eca Initial revision
casties
parents:
diff changeset
28 my $fix_files = ! $check_only;
30497c6a3eca Initial revision
casties
parents:
diff changeset
29 # add tags for file size and date
30497c6a3eca Initial revision
casties
parents:
diff changeset
30 my $fix_fs_meta = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
31 # add dir tags for missing directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
32 my $fix_dirs = ! $check_only;
30497c6a3eca Initial revision
casties
parents:
diff changeset
33 # rewrite XML file (necessary for fix_files and fix_dirs)
30497c6a3eca Initial revision
casties
parents:
diff changeset
34 my $fix_xml = ! $check_only;
30497c6a3eca Initial revision
casties
parents:
diff changeset
35 my $xml_changed = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
36 # XML namespace (not really implemented!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
37 my $namespace = "";
30497c6a3eca Initial revision
casties
parents:
diff changeset
38
30497c6a3eca Initial revision
casties
parents:
diff changeset
39
30497c6a3eca Initial revision
casties
parents:
diff changeset
40 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
41 # check parameters that were passed to the program
30497c6a3eca Initial revision
casties
parents:
diff changeset
42 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
43 my $docdir = $$args{'path'};
30497c6a3eca Initial revision
casties
parents:
diff changeset
44 if (! $docdir) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
45 logger("ABORT", "no document directory given!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
46 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
47 }
1
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
48 # strip double slashes
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
49 $docdir =~ s/\/\//\//;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
50 # strip trailing slashes
30497c6a3eca Initial revision
casties
parents:
diff changeset
51 $docdir =~ s/\/$//;
30497c6a3eca Initial revision
casties
parents:
diff changeset
52 if (! -d $docdir) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
53 logger("ABORT", "document directory \'$docdir\' doesn't exist!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
54 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
55 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
56
30497c6a3eca Initial revision
casties
parents:
diff changeset
57 my $metafile = "$docdir/index.meta";
30497c6a3eca Initial revision
casties
parents:
diff changeset
58 if (! -f $metafile) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
59 logger("ABORT", "metadata index file \'$metafile\' doesn't exist!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
60 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
61 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
62
30497c6a3eca Initial revision
casties
parents:
diff changeset
63 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
64 # internal variables
30497c6a3eca Initial revision
casties
parents:
diff changeset
65 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
66
30497c6a3eca Initial revision
casties
parents:
diff changeset
67 # all files in the document directory tree
30497c6a3eca Initial revision
casties
parents:
diff changeset
68 my %files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
69 # all directories in the document directory tree
30497c6a3eca Initial revision
casties
parents:
diff changeset
70 my %dirs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
71 # number of errors
30497c6a3eca Initial revision
casties
parents:
diff changeset
72 my $errcnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
73 # number of warnings
30497c6a3eca Initial revision
casties
parents:
diff changeset
74 my $warncnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
75
30497c6a3eca Initial revision
casties
parents:
diff changeset
76 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
77 # subroutines
30497c6a3eca Initial revision
casties
parents:
diff changeset
78 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
79
30497c6a3eca Initial revision
casties
parents:
diff changeset
80 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
81 # fs_read_files($realdir, $docdir, \%files, \%dirs)
30497c6a3eca Initial revision
casties
parents:
diff changeset
82 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
83 # reads all files and directories below $realdir and puts the
30497c6a3eca Initial revision
casties
parents:
diff changeset
84 # files in %files and directories in %dirs
30497c6a3eca Initial revision
casties
parents:
diff changeset
85 # $docdir is only for recursion, it should be empty when called
30497c6a3eca Initial revision
casties
parents:
diff changeset
86 # from outside
30497c6a3eca Initial revision
casties
parents:
diff changeset
87 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
88 sub fs_read_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
89 my ($directory, $docdir, $files, $dirs) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
90 my $cnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
91
30497c6a3eca Initial revision
casties
parents:
diff changeset
92 if (! opendir DIR, $directory) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
93 return 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
94 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
95 my @dirfiles = readdir DIR;
30497c6a3eca Initial revision
casties
parents:
diff changeset
96 foreach my $fn (@dirfiles) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
97 # ignore names starting with a dot
30497c6a3eca Initial revision
casties
parents:
diff changeset
98 next if ($fn =~ /^\./);
30497c6a3eca Initial revision
casties
parents:
diff changeset
99 # ignore other silly files
30497c6a3eca Initial revision
casties
parents:
diff changeset
100 next if ($junk_files{$fn});
30497c6a3eca Initial revision
casties
parents:
diff changeset
101
30497c6a3eca Initial revision
casties
parents:
diff changeset
102 $cnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
103 my $f = "$directory/$fn";
30497c6a3eca Initial revision
casties
parents:
diff changeset
104 my $docf = ($docdir) ? "$docdir/$fn" : $fn;
30497c6a3eca Initial revision
casties
parents:
diff changeset
105 #logger("fs_file: \"$f\"");
30497c6a3eca Initial revision
casties
parents:
diff changeset
106 if (-f $f) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
107 #logger(" is file");
30497c6a3eca Initial revision
casties
parents:
diff changeset
108 my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
30497c6a3eca Initial revision
casties
parents:
diff changeset
109 $atime,$mtime,$ctime,$blksize,$blocks)
30497c6a3eca Initial revision
casties
parents:
diff changeset
110 = stat(_);
30497c6a3eca Initial revision
casties
parents:
diff changeset
111 $$files{$docf} = [$fn, $size, $mtime];
30497c6a3eca Initial revision
casties
parents:
diff changeset
112 } elsif (-d _) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
113 #logger(" is dir");
30497c6a3eca Initial revision
casties
parents:
diff changeset
114 $$dirs{$docf} = $fn;
30497c6a3eca Initial revision
casties
parents:
diff changeset
115 # recurse into directory
30497c6a3eca Initial revision
casties
parents:
diff changeset
116 $cnt += fs_read_files($f, $docf, $files, $dirs);
30497c6a3eca Initial revision
casties
parents:
diff changeset
117 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
118 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
119 return $cnt;
30497c6a3eca Initial revision
casties
parents:
diff changeset
120 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
121
30497c6a3eca Initial revision
casties
parents:
diff changeset
122
30497c6a3eca Initial revision
casties
parents:
diff changeset
123 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
124 # check_dirs($rootnode, \%dirs)
30497c6a3eca Initial revision
casties
parents:
diff changeset
125 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
126 # reads all dir elements under $rootnode and compares with the directory list
30497c6a3eca Initial revision
casties
parents:
diff changeset
127 # in %dirs
30497c6a3eca Initial revision
casties
parents:
diff changeset
128 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
129 sub check_dirs {
30497c6a3eca Initial revision
casties
parents:
diff changeset
130 my ($root, $fsdirs) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
131
30497c6a3eca Initial revision
casties
parents:
diff changeset
132 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
133 # iterate all dir tags
30497c6a3eca Initial revision
casties
parents:
diff changeset
134 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
135 my @dirnodes = $root->findnodes('child::dir');
30497c6a3eca Initial revision
casties
parents:
diff changeset
136 my %okdirs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
137 foreach my $dirnode (@dirnodes) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
138 my $dirname = sstrip($dirnode->find('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
139 my $dirpath = sstrip($dirnode->find('child::path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
140 my $description = sstrip($dirnode->find('child::description'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
141 # name must be valid
30497c6a3eca Initial revision
casties
parents:
diff changeset
142 if (! valid_dir_name($dirname)) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
143 logger("ERROR", "directory name ($dirpath) $dirname invalid!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
144 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
145 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
146 # description can be present
30497c6a3eca Initial revision
casties
parents:
diff changeset
147 if (! $description) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
148 logger("WARNING", "description for directory $dirname (in $dirpath/) missing!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
149 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
150 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
151 # check with dirs on filesystem
30497c6a3eca Initial revision
casties
parents:
diff changeset
152 my $fn;
30497c6a3eca Initial revision
casties
parents:
diff changeset
153 if ($dirpath) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
154 $fn = "$dirpath/$dirname";
30497c6a3eca Initial revision
casties
parents:
diff changeset
155 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
156 $fn = "$dirname";
30497c6a3eca Initial revision
casties
parents:
diff changeset
157 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
158 #logger("dir: \"$dirname\", \"$dirpath\"");
30497c6a3eca Initial revision
casties
parents:
diff changeset
159 if ($$fsdirs{$fn}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
160 #logger(" OK ($$fsdirs{$fn})");
30497c6a3eca Initial revision
casties
parents:
diff changeset
161 $okdirs{$fn} = $dirname;
30497c6a3eca Initial revision
casties
parents:
diff changeset
162 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
163 logger("ERROR", "directory $dirname (in $dirpath/) missing on disk!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
164 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
165 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
166 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
167 #logger("dirs: ", (scalar keys %$fsdirs), " vs ", (scalar keys %okdirs), "");
30497c6a3eca Initial revision
casties
parents:
diff changeset
168 if ((scalar keys %$fsdirs) != (scalar keys %okdirs)) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
169 # number of dir tags and dirs don't match
30497c6a3eca Initial revision
casties
parents:
diff changeset
170 # iterate through all dirs
30497c6a3eca Initial revision
casties
parents:
diff changeset
171 foreach my $f (sort keys %$fsdirs) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
172 # was this dir missing?
30497c6a3eca Initial revision
casties
parents:
diff changeset
173 if (! $okdirs{$f}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
174 my ($name, $path) = split_file_path($f);
30497c6a3eca Initial revision
casties
parents:
diff changeset
175 # name must be valid
30497c6a3eca Initial revision
casties
parents:
diff changeset
176 if (! valid_dir_name($name)) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
177 $path = "." unless ($path);
30497c6a3eca Initial revision
casties
parents:
diff changeset
178 logger("ERROR", "directory name $name (in $path/) invalid!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
179 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
180 next;
30497c6a3eca Initial revision
casties
parents:
diff changeset
181 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
182 if ($fix_dirs) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
183 # add missing dir tag
30497c6a3eca Initial revision
casties
parents:
diff changeset
184 my $dir_node = $root->addNewChild($namespace, "dir");
30497c6a3eca Initial revision
casties
parents:
diff changeset
185 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
186 # add name
30497c6a3eca Initial revision
casties
parents:
diff changeset
187 my $name_node = $dir_node->addNewChild($namespace, "name");
30497c6a3eca Initial revision
casties
parents:
diff changeset
188 $name_node->appendTextNode($name);
30497c6a3eca Initial revision
casties
parents:
diff changeset
189 # add path
30497c6a3eca Initial revision
casties
parents:
diff changeset
190 if ($path) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
191 my $path_node = $dir_node->addNewChild($namespace, "path");
30497c6a3eca Initial revision
casties
parents:
diff changeset
192 $path_node->appendTextNode($path);
30497c6a3eca Initial revision
casties
parents:
diff changeset
193 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
194 logger("INFO", "directory $f to be added to index file!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
195 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
196 logger("ERROR", "directory $f missing in index file!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
197 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
198 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
199 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
200 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
201 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
202 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
203
30497c6a3eca Initial revision
casties
parents:
diff changeset
204
30497c6a3eca Initial revision
casties
parents:
diff changeset
205 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
206 # check_files($rootnode, \%files)
30497c6a3eca Initial revision
casties
parents:
diff changeset
207 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
208 # reads all file elements under $rootnode and compares with the file list
30497c6a3eca Initial revision
casties
parents:
diff changeset
209 # in %files
30497c6a3eca Initial revision
casties
parents:
diff changeset
210 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
211 sub check_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
212 my ($root, $fsfiles) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
213
30497c6a3eca Initial revision
casties
parents:
diff changeset
214 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
215 # iterate all file tags
30497c6a3eca Initial revision
casties
parents:
diff changeset
216 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
217 my @filenodes = $root->findnodes('child::file');
30497c6a3eca Initial revision
casties
parents:
diff changeset
218 my %okfiles;
30497c6a3eca Initial revision
casties
parents:
diff changeset
219 foreach my $filenode (@filenodes) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
220 my $filename = sstrip($filenode->find('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
221 my $filepath = sstrip($filenode->find('child::path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
222 my $filesize = sstrip($filenode->find('child::size'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
223 my $filedate = sstrip($filenode->find('child::date'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
224 # name must be valid
30497c6a3eca Initial revision
casties
parents:
diff changeset
225 if (! valid_file_name($filename)) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
226 logger("ERROR", "file name ($filepath)$filename invalid!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
227 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
228 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
229 my $fn = ($filepath) ? "$filepath/$filename" : "$filename";
30497c6a3eca Initial revision
casties
parents:
diff changeset
230 #logger("file: \"$filename\", \"$filepath\"");
30497c6a3eca Initial revision
casties
parents:
diff changeset
231 if ($$fsfiles{$fn}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
232 #logger(" OK ($$fsfiles{$fn})");
30497c6a3eca Initial revision
casties
parents:
diff changeset
233 $okfiles{$fn} = $filename;
30497c6a3eca Initial revision
casties
parents:
diff changeset
234 # check file size and date
30497c6a3eca Initial revision
casties
parents:
diff changeset
235 if ($filesize) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
236 if ($filesize != $$fsfiles{$fn}->[1]) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
237 logger("WARNING", "size of file $fn changed: $filesize to $$fsfiles{$fn}->[1]");
30497c6a3eca Initial revision
casties
parents:
diff changeset
238 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
239 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
240 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
241 # file date
30497c6a3eca Initial revision
casties
parents:
diff changeset
242 if ($filedate) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
243 if ($filedate ne stime($$fsfiles{$fn}->[2])) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
244 logger("WARNING", "date of file $fn changed: $filedate to ", stime($$fsfiles{$fn}->[2]), "");
30497c6a3eca Initial revision
casties
parents:
diff changeset
245 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
246 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
247 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
248 # update file size and date
30497c6a3eca Initial revision
casties
parents:
diff changeset
249 if ($fix_fs_meta) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
250 # delete size and date
30497c6a3eca Initial revision
casties
parents:
diff changeset
251 foreach my $n ($filenode->findnodes('child::size')) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
252 $filenode->removeChild($n);
30497c6a3eca Initial revision
casties
parents:
diff changeset
253 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
254 foreach my $n ($filenode->findnodes('child::date')) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
255 $filenode->removeChild($n);
30497c6a3eca Initial revision
casties
parents:
diff changeset
256 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
257 # add new size and date
30497c6a3eca Initial revision
casties
parents:
diff changeset
258 my $node = $filenode->addNewChild($namespace, "size");
30497c6a3eca Initial revision
casties
parents:
diff changeset
259 $node->appendTextNode($$fsfiles{$fn}->[1]);
30497c6a3eca Initial revision
casties
parents:
diff changeset
260 $node = $filenode->addNewChild($namespace, "date");
30497c6a3eca Initial revision
casties
parents:
diff changeset
261 $node->appendTextNode(stime($$fsfiles{$fn}->[2]));
30497c6a3eca Initial revision
casties
parents:
diff changeset
262 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
263 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
264 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
265 logger("ERROR", "file $filename (in $filepath/) missing on disk!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
266 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
267 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
268 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
269 #logger("files: ", (scalar keys %$fsfiles), " vs ", (scalar keys %okfiles), "");
30497c6a3eca Initial revision
casties
parents:
diff changeset
270 if ((scalar keys %$fsfiles) != (scalar keys %okfiles)) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
271 # number of file tags and files don't match
30497c6a3eca Initial revision
casties
parents:
diff changeset
272 # iterate through all files
30497c6a3eca Initial revision
casties
parents:
diff changeset
273 foreach my $f (sort keys %$fsfiles) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
274 my ($name, $path) = split_file_path($f);
30497c6a3eca Initial revision
casties
parents:
diff changeset
275 # was this file missing?
30497c6a3eca Initial revision
casties
parents:
diff changeset
276 if (! $okfiles{$f}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
277 # is an ignoreable file?
30497c6a3eca Initial revision
casties
parents:
diff changeset
278 if ($index_files{$name}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
279 next;
30497c6a3eca Initial revision
casties
parents:
diff changeset
280 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
281 # name must be valid
30497c6a3eca Initial revision
casties
parents:
diff changeset
282 if (! valid_dir_name($name)) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
283 $path = "." unless ($path);
30497c6a3eca Initial revision
casties
parents:
diff changeset
284 logger("ERROR", "file name $name (in $path/) invalid!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
285 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
286 next;
30497c6a3eca Initial revision
casties
parents:
diff changeset
287 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
288 if ($fix_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
289 # add missing file tag
30497c6a3eca Initial revision
casties
parents:
diff changeset
290 my $file_node = $root->addNewChild($namespace, "file");
30497c6a3eca Initial revision
casties
parents:
diff changeset
291 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
292 # add name
30497c6a3eca Initial revision
casties
parents:
diff changeset
293 my $name_node = $file_node->addNewChild($namespace, "name");
30497c6a3eca Initial revision
casties
parents:
diff changeset
294 $name_node->appendTextNode($name);
30497c6a3eca Initial revision
casties
parents:
diff changeset
295 # add path
30497c6a3eca Initial revision
casties
parents:
diff changeset
296 if ($path) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
297 my $path_node = $file_node->addNewChild($namespace, "path");
30497c6a3eca Initial revision
casties
parents:
diff changeset
298 $path_node->appendTextNode($path);
30497c6a3eca Initial revision
casties
parents:
diff changeset
299 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
300 # add size
30497c6a3eca Initial revision
casties
parents:
diff changeset
301 my $size_node = $file_node->addNewChild($namespace, "size");
30497c6a3eca Initial revision
casties
parents:
diff changeset
302 $size_node->appendTextNode($$fsfiles{$f}->[1]);
30497c6a3eca Initial revision
casties
parents:
diff changeset
303 # add date
30497c6a3eca Initial revision
casties
parents:
diff changeset
304 my $date_node = $file_node->addNewChild($namespace, "date");
30497c6a3eca Initial revision
casties
parents:
diff changeset
305 $date_node->appendTextNode(stime($$fsfiles{$f}->[2]));
30497c6a3eca Initial revision
casties
parents:
diff changeset
306 logger("INFO", "file $f to be added to index file!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
307 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
308 logger("ERROR", "file $f missing in index file!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
309 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
310 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
311 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
312 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
313 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
314 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
315
30497c6a3eca Initial revision
casties
parents:
diff changeset
316 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
317 # check_resource_meta($rootnode)
30497c6a3eca Initial revision
casties
parents:
diff changeset
318 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
319 # checks general resource meta information
30497c6a3eca Initial revision
casties
parents:
diff changeset
320 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
321 sub check_resource_meta {
30497c6a3eca Initial revision
casties
parents:
diff changeset
322 my ($rootnode) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
323
30497c6a3eca Initial revision
casties
parents:
diff changeset
324 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
325 # description
30497c6a3eca Initial revision
casties
parents:
diff changeset
326 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
327 my $description = $rootnode->findvalue('child::description');
30497c6a3eca Initial revision
casties
parents:
diff changeset
328 if (! $description) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
329 logger("ERROR", "resource description element missing!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
330 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
331 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
332 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
333 # name
30497c6a3eca Initial revision
casties
parents:
diff changeset
334 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
335 my $name = sstrip($rootnode->findvalue('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
336 if ($name) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
337 my ($dirname, $dirpath) = split_file_path($docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
338 if ($dirname ne $name) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
339 logger("ERROR", "resource name element '$name' does not match directory name '$dirname'!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
340 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
341 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
342 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
343 logger("ERROR", "resource name element missing!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
344 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
345 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
346 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
347 # archive path
30497c6a3eca Initial revision
casties
parents:
diff changeset
348 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
349 my $realpath;
30497c6a3eca Initial revision
casties
parents:
diff changeset
350 # get real path
30497c6a3eca Initial revision
casties
parents:
diff changeset
351 if ($docdir =~ /^\//) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
352 # docdir is absolute
30497c6a3eca Initial revision
casties
parents:
diff changeset
353 $realpath = $docdir;
30497c6a3eca Initial revision
casties
parents:
diff changeset
354 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
355 # docdir is relative -- try with the shell
30497c6a3eca Initial revision
casties
parents:
diff changeset
356 if (open PWDCMD, "cd $docdir ; pwd|") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
357 $realpath = <PWDCMD>;
30497c6a3eca Initial revision
casties
parents:
diff changeset
358 chomp $realpath;
30497c6a3eca Initial revision
casties
parents:
diff changeset
359 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
360 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
361 if (! $realpath) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
362 logger("ERROR", "unable to check real archive path!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
363 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
364 return;
30497c6a3eca Initial revision
casties
parents:
diff changeset
365 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
366 # get archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
367 my $archnode = ($rootnode->find('child::archive-path'))->get_node(1);
30497c6a3eca Initial revision
casties
parents:
diff changeset
368 if ($archnode) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
369 my $arch = sstrip($archnode->textContent);
30497c6a3eca Initial revision
casties
parents:
diff changeset
370 if ($arch ne $realpath) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
371 logger("WARNING", "incorrect archive-path '$arch' will be changed to '$realpath'!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
372 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
373 # correct archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
374 $archnode->removeChildNodes;
30497c6a3eca Initial revision
casties
parents:
diff changeset
375 $archnode->appendTextNode($realpath);
30497c6a3eca Initial revision
casties
parents:
diff changeset
376 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
377 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
378 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
379 # add archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
380 $archnode = $rootnode->addNewChild($namespace, "archive-path");
30497c6a3eca Initial revision
casties
parents:
diff changeset
381 $archnode->appendTextNode($realpath);
30497c6a3eca Initial revision
casties
parents:
diff changeset
382 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
383 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
384
30497c6a3eca Initial revision
casties
parents:
diff changeset
385 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
386
30497c6a3eca Initial revision
casties
parents:
diff changeset
387
30497c6a3eca Initial revision
casties
parents:
diff changeset
388
30497c6a3eca Initial revision
casties
parents:
diff changeset
389 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
390 # main
30497c6a3eca Initial revision
casties
parents:
diff changeset
391 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
392
30497c6a3eca Initial revision
casties
parents:
diff changeset
393 logger("INFO", "metacheck $version");
30497c6a3eca Initial revision
casties
parents:
diff changeset
394
30497c6a3eca Initial revision
casties
parents:
diff changeset
395 my ($document, $rootnode) = read_xml($metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
396
30497c6a3eca Initial revision
casties
parents:
diff changeset
397 check_resource_meta($rootnode);
30497c6a3eca Initial revision
casties
parents:
diff changeset
398
30497c6a3eca Initial revision
casties
parents:
diff changeset
399 my $fnum = fs_read_files($docdir, "", \%files, \%dirs);
30497c6a3eca Initial revision
casties
parents:
diff changeset
400 logger("INFO", "$fnum files on FS");
30497c6a3eca Initial revision
casties
parents:
diff changeset
401 #foreach (keys %files) {logger(" file ($_): $files{$_}");}
30497c6a3eca Initial revision
casties
parents:
diff changeset
402
30497c6a3eca Initial revision
casties
parents:
diff changeset
403 check_files($rootnode, \%files);
30497c6a3eca Initial revision
casties
parents:
diff changeset
404 check_dirs($rootnode, \%dirs);
30497c6a3eca Initial revision
casties
parents:
diff changeset
405
30497c6a3eca Initial revision
casties
parents:
diff changeset
406 logger("INFO", "$warncnt warnings");
30497c6a3eca Initial revision
casties
parents:
diff changeset
407 logger("INFO", "$errcnt errors");
30497c6a3eca Initial revision
casties
parents:
diff changeset
408 if ($errcnt > 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
409 logger("ABORT", "there were errors!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
410 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
411 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
412 if ($fix_xml) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
413 write_xml($document, $metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
414 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
415 logger("DONE", "index file checked successfully!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
416 }