/[webpac2]/trunk/run.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /trunk/run.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 540 - (show annotations)
Thu Jun 29 15:29:41 2006 UTC (17 years, 9 months ago) by dpavlin
File MIME type: text/plain
File size: 8081 byte(s)
 r726@llin:  dpavlin | 2006-06-29 17:31:13 +0200
 add marc21 to normalize and create MARC file from those data [2.22]

1 #!/usr/bin/perl -w
2
3 use strict;
4
5 use Cwd qw/abs_path/;
6 use File::Temp qw/tempdir/;
7 use Data::Dumper;
8 use lib './lib';
9
10 use WebPAC::Common 0.02;
11 use WebPAC::Lookup;
12 use WebPAC::Input 0.03;
13 use WebPAC::Store 0.03;
14 use WebPAC::Normalize;
15 use WebPAC::Output::TT;
16 use WebPAC::Validate;
17 use YAML qw/LoadFile/;
18 use Getopt::Long;
19 use File::Path;
20 use Time::HiRes qw/time/;
21 use File::Slurp;
22 use MARC::Record;
23
24 =head1 NAME
25
26 run.pl - start WebPAC indexing
27
28 B<this command will probably go away. Don't get used to it!>
29
30 Options:
31
32 =over 4
33
34 =item --offset 42
35
36 start loading (all) databases at offset 42
37
38 =item --limit 100
39
40 limit loading to 100 records
41
42 =item --clean
43
44 remove database and Hyper Estraier index before indexing
45
46 =item --only=database_name/input_filter
47
48 reindex just single database (legacy name is --one)
49
50 C</input_filter> is optional part which can be C<name>
51 or C<type> from input
52
53 =item --config conf/config.yml
54
55 path to YAML configuration file
56
57 =item --stats
58
59 disable indexing and dump statistics about field and subfield
60 usage for each input
61
62 =item --validate path/to/validation_file
63
64 turn on extra validation of imput records, see L<WebPAC::Validation>
65
66 =back
67
68 =cut
69
70 my $offset;
71 my $limit;
72
73 my $clean = 0;
74 my $config = 'conf/config.yml';
75 my $debug = 0;
76 my $only_filter;
77 my $stats = 0;
78 my $validate_path;
79
80 GetOptions(
81 "limit=i" => \$limit,
82 "offset=i" => \$offset,
83 "clean" => \$clean,
84 "one=s" => \$only_filter,
85 "only=s" => \$only_filter,
86 "config" => \$config,
87 "debug" => \$debug,
88 "stats" => \$stats,
89 "validate=s" => \$validate_path,
90 );
91
92 $config = LoadFile($config);
93
94 print "config = ",Dumper($config) if ($debug);
95
96 die "no databases in config file!\n" unless ($config->{databases});
97
98 my $log = _new WebPAC::Common()->_get_logger();
99 $log->info( "-" x 79 );
100
101 my $validate;
102 $validate = new WebPAC::Validate(
103 path => $validate_path,
104 ) if ($validate_path);
105
106 my $use_indexer = $config->{use_indexer} || 'hyperestraier';
107 if ($stats) {
108 $log->debug("option --stats disables update of indexing engine...");
109 $use_indexer = undef;
110 } else {
111 $log->info("using $use_indexer indexing engine...");
112 }
113
114 my $total_rows = 0;
115 my $start_t = time();
116
117 my @links;
118 my $indexer;
119
120 while (my ($database, $db_config) = each %{ $config->{databases} }) {
121
122 my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
123 next if ($only_database && $database !~ m/$only_database/i);
124
125 if ($use_indexer) {
126 my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
127 $indexer_config->{database} = $database;
128 $indexer_config->{clean} = $clean;
129 $indexer_config->{label} = $db_config->{name};
130
131 if ($use_indexer eq 'hyperestraier') {
132
133 # open Hyper Estraier database
134 use WebPAC::Output::Estraier '0.10';
135 $indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
136
137 } elsif ($use_indexer eq 'kinosearch') {
138
139 # open KinoSearch
140 use WebPAC::Output::KinoSearch;
141 $indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
142 $indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
143
144 } else {
145 $log->logdie("unknown use_indexer: $use_indexer");
146 }
147
148 $log->logide("can't continue without valid indexer") unless ($indexer);
149 }
150
151
152 #
153 # now WebPAC::Store
154 #
155 my $abs_path = abs_path($0);
156 $abs_path =~ s#/[^/]*$#/#;
157
158 my $db_path = $config->{webpac}->{db_path} . '/' . $database;
159
160 if ($clean) {
161 $log->info("creating new database '$database' in $db_path");
162 rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
163 } else {
164 $log->info("working on database '$database' in $db_path");
165 }
166
167 my $db = new WebPAC::Store(
168 path => $db_path,
169 database => $database,
170 debug => $debug,
171 );
172
173
174 #
175 # now, iterate through input formats
176 #
177
178 my @inputs;
179 if (ref($db_config->{input}) eq 'ARRAY') {
180 @inputs = @{ $db_config->{input} };
181 } elsif ($db_config->{input}) {
182 push @inputs, $db_config->{input};
183 } else {
184 $log->info("database $database doesn't have inputs defined");
185 }
186
187 my @supported_inputs = keys %{ $config->{webpac}->{inputs} };
188
189 foreach my $input (@inputs) {
190
191 next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));
192
193 my $type = lc($input->{type});
194
195 die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
196
197 my $lookup = new WebPAC::Lookup(
198 lookup_file => $input->{lookup},
199 ) if ($input->{lookup});
200
201 my $input_module = $config->{webpac}->{inputs}->{$type};
202
203 $log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module",
204 $input->{lookup} ? "lookup '$input->{lookup}'" : ""
205 );
206
207 my $input_db = new WebPAC::Input(
208 module => $input_module,
209 code_page => $config->{webpac}->{webpac_encoding},
210 limit => $limit || $input->{limit},
211 offset => $offset,
212 lookup => $lookup,
213 recode => $input->{recode},
214 stats => $stats,
215 );
216 $log->logdie("can't create input using $input_module") unless ($input);
217
218 my $maxmfn = $input_db->open(
219 path => $input->{path},
220 code_page => $input->{encoding}, # database encoding
221 %{ $input },
222 );
223
224 my @norm_array = ref($input->{normalize}) eq 'ARRAY' ?
225 @{ $input->{normalize} } : ( $input->{normalize} );
226
227 foreach my $normalize (@norm_array) {
228
229 my $rules;
230 my $normalize_path = $normalize->{path} || $log->logdie("can't find normalize path in config");
231
232 $log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i );
233
234 my $rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
235
236 $log->info("Using $normalize_path for normalization...");
237
238 my $marc_fh;
239 if (my $path = $normalize->{marc21}) {
240 open($marc_fh, '>', $path) ||
241 $log->logdie("can't open MARC output $path: $!");
242
243 $log->info("Creating MARC export file $path\n");
244 }
245
246 # reset position in database
247 $input_db->seek(1);
248
249 foreach my $pos ( 0 ... $input_db->size ) {
250
251 my $row = $input_db->fetch || next;
252
253 my $mfn = $row->{'000'}->[0];
254
255 if (! $mfn || $mfn !~ m#^\d+$#) {
256 $log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
257 $mfn = $pos;
258 push @{ $row->{'000'} }, $pos;
259 }
260
261
262 if ($validate) {
263 my @errors = $validate->validate_errors( $row );
264 $log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
265 }
266
267
268 my $ds = WebPAC::Normalize::data_structure(
269 row => $row,
270 rules => $rules,
271 lookup => $lookup ? $lookup->lookup_hash : undef,
272 );
273
274 $db->save_ds(
275 id => $mfn,
276 ds => $ds,
277 prefix => $input->{name},
278 ) if ($ds && !$stats);
279
280 $indexer->add(
281 id => $input->{name} . "/" . $mfn,
282 ds => $ds,
283 type => $config->{$use_indexer}->{type},
284 ) if ($indexer && $ds);
285
286 if ($marc_fh) {
287 my $marc = new MARC::Record;
288 $marc->add_fields( WebPAC::Normalize::_get_marc21_fields() );
289 print $marc_fh $marc->as_usmarc;
290 }
291
292 $total_rows++;
293 }
294
295 $log->info("statistics of fields usage:\n", $input_db->stats) if ($stats);
296
297 # close MARC file
298 close($marc_fh) if ($marc_fh);
299
300 }
301
302 }
303
304 eval { $indexer->finish } if ($indexer && $indexer->can('finish'));
305
306 my $dt = time() - $start_t;
307 $log->info("$total_rows records ", $indexer ? "indexed " : "",
308 sprintf("in %.2f sec [%.2f rec/sec]",
309 $dt, ($total_rows / $dt)
310 )
311 );
312
313 #
314 # add Hyper Estraier links to other databases
315 #
316 if (ref($db_config->{links}) eq 'ARRAY') {
317 foreach my $link (@{ $db_config->{links} }) {
318 if ($use_indexer eq 'hyperestraier') {
319 $log->info("saving link $database -> $link->{to} [$link->{credit}]");
320 push @links, {
321 from => $database,
322 to => $link->{to},
323 credit => $link->{credit},
324 };
325 } else {
326 $log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
327 }
328 }
329 }
330
331 }
332
333 foreach my $link (@links) {
334 $log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
335 $indexer->add_link( %{ $link } );
336 }

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26