1 package OpenILS::Application::Storage::WORM;
2 use base qw/OpenILS::Application::Storage/;
3 use strict; use warnings;
5 use OpenSRF::EX qw/:try/;
7 use OpenSRF::Utils::Logger qw/:level/;
8 my $log = 'OpenSRF::Utils::Logger';
10 use OpenILS::Utils::FlatXML;
11 use OpenILS::Utils::Fieldmapper;
16 use Time::HiRes qw(time);
18 my $xml_util = OpenILS::Utils::FlatXML->new();
20 my $parser = XML::LibXML->new();
21 my $xslt = XML::LibXSLT->new();
22 my $xslt_doc = $parser->parse_file( "/home/miker/cvs/OpenILS/app_server/stylesheets/MARC21slim2MODS.xsl" );
23 #my $xslt_doc = $parser->parse_file( "/pines/cvs/ILS/Open-ILS/xsl/MARC21slim2MODS.xsl" );
24 my $mods_sheet = $xslt->parse_stylesheet( $xslt_doc );
31 my $meth = __PACKAGE__->method_lookup('open-ils.storage.direct.config.metabib_field.all');
32 for my $f ($meth->run) {
33 $xpathset->{ $f->field_class }->{ $f->name }->{xpath} = $f->xpath;
34 $xpathset->{ $f->field_class }->{ $f->name }->{id} = $f->id;
35 $log->debug("Loaded XPath from DB: ".$f->field_class." => ".$f->name." : ".$f->xpath, DEBUG);
39 # --------------------------------------------------------------------------------
43 '//mods:mods/mods:typeOfResource[text()="text"]' => [
46 '//mods:mods/mods:titleInfo[mods:title and (@type="uniform")]',
47 '//mods:mods/mods:titleInfo[mods:title and (@type="translated")]',
48 '//mods:mods/mods:titleInfo[mods:title and (@type="alternative")]',
49 '//mods:mods/mods:titleInfo[mods:title and not(@type)]',
55 $text =~ s/^\s*(.+)\s*$/$1/sgo;
56 $text =~ s/\b(?:the|an?)\b//sgo;
57 $text =~ s/\[.[^\]]+\]//sgo;
58 $text =~ s/\s*[;\/\.]*$//sgo;
65 '//mods:mods/mods:name[mods:role/mods:text/text()="creator" and @type="personal"]/mods:namePart',
66 '//mods:mods/mods:name[mods:role/mods:text/text()="creator"]/mods:namePart',
72 $text =~ s/^\s*(.+)\s*$/$1/sgo;
73 $text =~ s/,?\s+.*$//sgo;
80 '//mods:mods/mods:relatedItem[@type!="host"]' => [
83 '//mods:mods/mods:relatedItem/mods:titleInfo[mods:title and (@type="uniform")]',
84 '//mods:mods/mods:relatedItem/mods:titleInfo[mods:title and (@type="translated")]',
85 '//mods:mods/mods:relatedItem/mods:titleInfo[mods:title and (@type="alternative")]',
86 '//mods:mods/mods:relatedItem/mods:titleInfo[mods:title and not(@type)]',
87 '//mods:mods/mods:titleInfo[mods:title and (@type="uniform")]',
88 '//mods:mods/mods:titleInfo[mods:title and (@type="translated")]',
89 '//mods:mods/mods:titleInfo[mods:title and (@type="alternative")]',
90 '//mods:mods/mods:titleInfo[mods:title and not(@type)]',
96 $text =~ s/^\s*(.+)\s*$/$1/sgo;
97 $text =~ s/\b(?:the|an?)\b//sgo;
98 $text =~ s/\[.[^\]]+\]//sgo;
99 $text =~ s/\s*[;\/\.]*$//sgo;
100 $text =~ s/\pM+//gso;
106 '//mods:mods/mods:relatedItem/mods:name[mods:role/mods:text/text()="creator" and @type="personal"]/mods:namePart',
107 '//mods:mods/mods:relatedItem/mods:name[mods:role/mods:text/text()="creator"]/mods:namePart',
108 '//mods:mods/mods:name[mods:role/mods:text/text()="creator" and @type="personal"]/mods:namePart',
109 '//mods:mods/mods:name[mods:role/mods:text/text()="creator"]/mods:namePart',
114 $text =~ s/\s+/ /sgo;
115 $text =~ s/^\s*(.+)\s*$/$1/sgo;
116 $text =~ s/,?\s+.*$//sgo;
117 $text =~ s/\pM+//gso;
125 push @fp_mods_xpath, '//mods:mods/mods:titleInfo' => $fp_mods_xpath[1];
127 sub fingerprint_mods {
134 while ( my $match_xpath = $fp_mods_xpath[$match_index] ) {
135 if ( my @nodes = $mods->findnodes( $match_xpath ) ) {
137 my $block_name_index = 0;
138 my $block_value_index = 1;
139 my $block = $fp_mods_xpath[$block_index];
140 while ( my $part = $$block[$block_value_index] ) {
142 for my $xpath ( @{ $part->{xpath} } ) {
143 $text = $mods->findvalue( $xpath );
147 $log->debug("Found fingerprint text using $$block[$block_name_index] : [$text]", DEBUG);
154 $block_name_index += 2;
155 $block_value_index += 2;
159 $fp_string =~ s/\W+//gso;
160 $log->debug("Fingerprint is [$fp_string]", INFO);;
172 # --------------------------------------------------------------------------------
181 my $create_source_map;
195 my %descriptor_code = (
196 item_type => 'substr($ldr,6,1)',
197 item_form => '(substr($ldr,6,1) =~ /^(?:f|g|i|m|o|p|r)$/) ? substr($oo8,29,1) : substr($oo8,23,1)',
198 bib_level => 'substr($ldr,7,1)',
199 control_type => 'substr($ldr,8,1)',
200 char_encoding => 'substr($ldr,9,1)',
201 enc_level => 'substr($ldr,17,1)',
202 cat_form => 'substr($ldr,18,1)',
203 pub_status => 'substr($ldr,5,1)',
204 item_lang => 'substr($oo8,35,3)',
205 audience => 'substr($oo8,22,1)',
214 $begin = $self->method_lookup( 'open-ils.storage.transaction.begin')
216 $commit = $self->method_lookup( 'open-ils.storage.transaction.commit')
218 $rollback = $self->method_lookup( 'open-ils.storage.transaction.rollback')
220 $sm_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.search.source')
222 $mr_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.search.fingerprint')
224 $mr_create = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.create')
226 $create_source_map = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.batch.create')
227 unless ($create_source_map);
228 $lookup = $self->method_lookup('open-ils.storage.direct.biblio.record_entry.batch.retrieve')
230 $update_entry = $self->method_lookup('open-ils.storage.direct.biblio.record_entry.batch.update')
231 unless ($update_entry);
232 $rm_old_sm = $self->method_lookup( 'open-ils.storage.direct.metabib.metarecord_source_map.mass_delete')
234 $rm_old_rd = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.mass_delete')
236 $rm_old_fr = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.mass_delete')
238 $rm_old_tr = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.mass_delete')
240 $rm_old_ar = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.mass_delete')
242 $rm_old_sr = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.mass_delete')
244 $rm_old_kr = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.mass_delete')
246 $rd_create = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.batch.create')
248 $fr_create = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.batch.create')
250 $$create{title} = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.batch.create')
251 unless ($$create{title});
252 $$create{author} = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.batch.create')
253 unless ($$create{author});
254 $$create{subject} = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.batch.create')
255 unless ($$create{subject});
256 $$create{keyword} = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.batch.create')
257 unless ($$create{keyword});
261 my ($r) = $begin->run($client);
262 unless (defined $r and $r) {
264 throw OpenSRF::EX::PANIC ("Couldn't BEGIN transaction!")
267 throw OpenSRF::EX::PANIC ("WoRM Couldn't BEGIN transaction!")
276 for my $entry ( $lookup->run(@docids) ) {
277 # step -1: grab the doc from storage
278 next unless ($entry);
280 my $xml = $entry->marc;
281 my $docid = $entry->id;
282 my $marcdoc = $parser->parse_string($xml);
283 my $modsdoc = $mods_sheet->transform($marcdoc);
285 my $mods = $modsdoc->documentElement;
286 $mods->setNamespace( "http://www.loc.gov/mods/", "mods", 1 );
288 $entry->fingerprint( fingerprint_mods( $mods ) );
289 push @entry_list, $entry;
291 my ($mr) = $mr_lookup->run( $entry->fingerprint );
293 $mr = new Fieldmapper::metabib::metarecord;
294 $mr->fingerprint( $entry->fingerprint );
295 $mr->master_record( $entry->id );
296 my ($new_mr) = $mr_create->run($mr);
298 unless (defined $mr) {
299 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord.create!")
305 my $sm = new Fieldmapper::metabib::metarecord_source_map;
306 $sm->metarecord( $mr->id );
307 $sm->source( $entry->id );
308 push @source_maps, $sm;
310 my $ldr = $marcdoc->documentElement->getChildrenByTagName('leader')->pop->textContent;
311 my $oo8 = $marcdoc->documentElement->findvalue('//*[local-name()="controlfield" and @tag="008"]');
313 my $rd_obj = Fieldmapper::metabib::record_descriptor->new;
314 for my $rd_field ( keys %descriptor_code ) {
315 $rd_obj->$rd_field( eval "$descriptor_code{$rd_field};" );
317 $rd_obj->record( $docid );
318 push @rd_list, $rd_obj;
320 push @mods_data, { $docid => $self->modsdoc_to_values( $mods ) };
322 # step 2: build the KOHA rows
323 my @tmp_list = _marcxml_to_full_rows( $marcdoc );
324 $_->record( $docid ) for (@tmp_list);
325 push @ns_list, @tmp_list;
329 last unless ($self->api_name =~ /batch$/o);
332 $rm_old_rd->run( { record => \@docids } );
333 $rm_old_fr->run( { record => \@docids } );
334 $rm_old_sm->run( { source => \@docids } );
335 $rm_old_tr->run( { source => \@docids } );
336 $rm_old_ar->run( { source => \@docids } );
337 $rm_old_sr->run( { source => \@docids } );
338 $rm_old_kr->run( { source => \@docids } );
340 my ($sm) = $create_source_map->run(@source_maps);
341 unless (defined $sm) {
342 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord_source_map.batch.create!")
345 my ($re) = $update_entry->run(@entry_list);
346 unless (defined $re) {
347 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.biblio.record_entry.batch.update!")
350 my ($rd) = $rd_create->run(@rd_list);
351 unless (defined $rd) {
352 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.record_descriptor.batch.create!")
355 my ($fr) = $fr_create->run(@ns_list);
356 unless (defined $fr) {
357 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.full_rec.batch.create!")
360 # step 5: insert the new metadata
361 for my $class ( qw/title author subject keyword/ ) {
363 for my $doc ( @mods_data ) {
364 my ($did) = keys %$doc;
365 my ($data) = values %$doc;
367 my $fm_constructor = "Fieldmapper::metabib::${class}_field_entry";
368 for my $row ( keys %{ $$data{$class} } ) {
369 next unless (exists $$data{$class}{$row});
370 next unless ($$data{$class}{$row}{value});
371 my $fm_obj = $fm_constructor->new;
372 $fm_obj->value( $$data{$class}{$row}{value} );
373 $fm_obj->field( $$data{$class}{$row}{field_id} );
374 $fm_obj->source( $did );
375 $log->debug("$class entry: ".$fm_obj->source." => ".$fm_obj->field." : ".$fm_obj->value, DEBUG);
377 push @md_list, $fm_obj;
381 my ($cr) = $$create{$class}->run(@md_list);
382 unless (defined $cr) {
383 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.${class}_field_entry.batch.create!")
387 my ($c) = $commit->run;
388 unless (defined $c and $c) {
390 throw OpenSRF::EX::PANIC ("Couldn't COMMIT changes!")
395 __PACKAGE__->register_method(
396 api_name => "open-ils.worm.wormize",
401 __PACKAGE__->register_method(
402 api_name => "open-ils.worm.wormize.batch",
409 # --------------------------------------------------------------------------------
412 sub _marcxml_to_full_rows {
418 my $root = $marcxml->documentElement;
420 for my $tagline ( @{$root->getChildrenByTagName("leader")} ) {
421 next unless $tagline;
423 my $ns = new Fieldmapper::metabib::full_rec;
426 my $val = $tagline->textContent;
427 $val =~ s/(\pM)//gso;
433 for my $tagline ( @{$root->getChildrenByTagName("controlfield")} ) {
434 next unless $tagline;
436 my $ns = new Fieldmapper::metabib::full_rec;
438 $ns->tag( $tagline->getAttribute( "tag" ) );
439 my $val = $tagline->textContent;
440 $val =~ s/(\pM)//gso;
446 for my $tagline ( @{$root->getChildrenByTagName("datafield")} ) {
447 next unless $tagline;
449 my $tag = $tagline->getAttribute( "tag" );
450 my $ind1 = $tagline->getAttribute( "ind1" );
451 my $ind2 = $tagline->getAttribute( "ind2" );
453 for my $data ( $tagline->childNodes ) {
456 my $ns = new Fieldmapper::metabib::full_rec;
461 $ns->subfield( $data->getAttribute( "code" ) );
462 my $val = $data->textContent;
463 $val =~ s/(\pM)//gso;
464 $ns->value( lc($val) );
472 sub _get_field_value {
474 my( $root, $xpath ) = @_;
478 # grab the set of matching nodes
479 my @nodes = $root->findnodes( $xpath );
480 for my $value (@nodes) {
482 # grab all children of the node
483 my @children = $value->childNodes();
484 for my $child (@children) {
486 # add the childs content to the growing buffer
487 my $content = quotemeta($child->textContent);
488 next if ($string =~ /$content/); # uniquify the values
489 $string .= $child->textContent . " ";
492 $string .= $value->textContent . " ";
495 $string =~ s/(\pM)//gso;
500 sub modsdoc_to_values {
501 my( $self, $mods ) = @_;
503 for my $class (keys %$xpathset) {
504 $data->{$class} = {};
505 for my $type (keys %{$xpathset->{$class}}) {
506 $data->{$class}->{$type} = {};
507 $data->{$class}->{$type}->{value} = _get_field_value( $mods, $xpathset->{$class}->{$type}->{xpath} );
508 $data->{$class}->{$type}->{field_id} = $xpathset->{$class}->{$type}->{id};