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)',
215 if ($self->api_name =~ /no_map/o) {
219 $begin = $self->method_lookup( 'open-ils.storage.transaction.begin')
221 $commit = $self->method_lookup( 'open-ils.storage.transaction.commit')
223 $rollback = $self->method_lookup( 'open-ils.storage.transaction.rollback')
225 $sm_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.search.source')
227 $mr_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.search.fingerprint')
229 $mr_create = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.create')
231 $create_source_map = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.batch.create')
232 unless ($create_source_map);
233 $lookup = $self->method_lookup('open-ils.storage.direct.biblio.record_entry.batch.retrieve')
235 $update_entry = $self->method_lookup('open-ils.storage.direct.biblio.record_entry.batch.update')
236 unless ($update_entry);
237 $rm_old_sm = $self->method_lookup( 'open-ils.storage.direct.metabib.metarecord_source_map.mass_delete')
239 $rm_old_rd = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.mass_delete')
241 $rm_old_fr = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.mass_delete')
243 $rm_old_tr = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.mass_delete')
245 $rm_old_ar = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.mass_delete')
247 $rm_old_sr = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.mass_delete')
249 $rm_old_kr = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.mass_delete')
251 $rd_create = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.batch.create')
253 $fr_create = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.batch.create')
255 $$create{title} = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.batch.create')
256 unless ($$create{title});
257 $$create{author} = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.batch.create')
258 unless ($$create{author});
259 $$create{subject} = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.batch.create')
260 unless ($$create{subject});
261 $$create{keyword} = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.batch.create')
262 unless ($$create{keyword});
266 my ($r) = $begin->run($client);
267 unless (defined $r and $r) {
269 throw OpenSRF::EX::PANIC ("Couldn't BEGIN transaction!")
272 throw OpenSRF::EX::PANIC ("WoRM Couldn't BEGIN transaction!")
281 for my $entry ( $lookup->run(@docids) ) {
282 # step -1: grab the doc from storage
283 next unless ($entry);
285 my $xml = $entry->marc;
286 my $docid = $entry->id;
287 my $marcdoc = $parser->parse_string($xml);
288 my $modsdoc = $mods_sheet->transform($marcdoc);
290 my $mods = $modsdoc->documentElement;
291 $mods->setNamespace( "http://www.loc.gov/mods/", "mods", 1 );
293 $entry->fingerprint( fingerprint_mods( $mods ) );
294 push @entry_list, $entry;
297 my ($mr) = $mr_lookup->run( $entry->fingerprint );
299 $mr = new Fieldmapper::metabib::metarecord;
300 $mr->fingerprint( $entry->fingerprint );
301 $mr->master_record( $entry->id );
302 my ($new_mr) = $mr_create->run($mr);
304 unless (defined $mr) {
305 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord.create!")
311 my $sm = new Fieldmapper::metabib::metarecord_source_map;
312 $sm->metarecord( $mr->id );
313 $sm->source( $entry->id );
314 push @source_maps, $sm;
317 my $ldr = $marcdoc->documentElement->getChildrenByTagName('leader')->pop->textContent;
318 my $oo8 = $marcdoc->documentElement->findvalue('//*[local-name()="controlfield" and @tag="008"]');
320 my $rd_obj = Fieldmapper::metabib::record_descriptor->new;
321 for my $rd_field ( keys %descriptor_code ) {
322 $rd_obj->$rd_field( eval "$descriptor_code{$rd_field};" );
324 $rd_obj->record( $docid );
325 push @rd_list, $rd_obj;
327 push @mods_data, { $docid => $self->modsdoc_to_values( $mods ) };
329 # step 2: build the KOHA rows
330 my @tmp_list = _marcxml_to_full_rows( $marcdoc );
331 $_->record( $docid ) for (@tmp_list);
332 push @ns_list, @tmp_list;
336 last unless ($self->api_name =~ /batch$/o);
339 $rm_old_rd->run( { record => \@docids } );
340 $rm_old_fr->run( { record => \@docids } );
341 $rm_old_sm->run( { source => \@docids } ) unless ($no_map);
342 $rm_old_tr->run( { source => \@docids } );
343 $rm_old_ar->run( { source => \@docids } );
344 $rm_old_sr->run( { source => \@docids } );
345 $rm_old_kr->run( { source => \@docids } );
348 my ($sm) = $create_source_map->run(@source_maps);
349 unless (defined $sm) {
350 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord_source_map.batch.create!")
354 my ($re) = $update_entry->run(@entry_list);
355 unless (defined $re) {
356 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.biblio.record_entry.batch.update!")
359 my ($rd) = $rd_create->run(@rd_list);
360 unless (defined $rd) {
361 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.record_descriptor.batch.create!")
364 my ($fr) = $fr_create->run(@ns_list);
365 unless (defined $fr) {
366 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.full_rec.batch.create!")
369 # step 5: insert the new metadata
370 for my $class ( qw/title author subject keyword/ ) {
372 for my $doc ( @mods_data ) {
373 my ($did) = keys %$doc;
374 my ($data) = values %$doc;
376 my $fm_constructor = "Fieldmapper::metabib::${class}_field_entry";
377 for my $row ( keys %{ $$data{$class} } ) {
378 next unless (exists $$data{$class}{$row});
379 next unless ($$data{$class}{$row}{value});
380 my $fm_obj = $fm_constructor->new;
381 $fm_obj->value( $$data{$class}{$row}{value} );
382 $fm_obj->field( $$data{$class}{$row}{field_id} );
383 $fm_obj->source( $did );
384 $log->debug("$class entry: ".$fm_obj->source." => ".$fm_obj->field." : ".$fm_obj->value, DEBUG);
386 push @md_list, $fm_obj;
390 my ($cr) = $$create{$class}->run(@md_list);
391 unless (defined $cr) {
392 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.${class}_field_entry.batch.create!")
396 my ($c) = $commit->run;
397 unless (defined $c and $c) {
399 throw OpenSRF::EX::PANIC ("Couldn't COMMIT changes!")
404 __PACKAGE__->register_method(
405 api_name => "open-ils.worm.wormize",
410 __PACKAGE__->register_method(
411 api_name => "open-ils.worm.wormize.no_map",
416 __PACKAGE__->register_method(
417 api_name => "open-ils.worm.wormize.batch",
422 __PACKAGE__->register_method(
423 api_name => "open-ils.worm.wormize.no_map.batch",
430 # --------------------------------------------------------------------------------
433 sub _marcxml_to_full_rows {
439 my $root = $marcxml->documentElement;
441 for my $tagline ( @{$root->getChildrenByTagName("leader")} ) {
442 next unless $tagline;
444 my $ns = new Fieldmapper::metabib::full_rec;
447 my $val = $tagline->textContent;
448 $val =~ s/(\pM)//gso;
454 for my $tagline ( @{$root->getChildrenByTagName("controlfield")} ) {
455 next unless $tagline;
457 my $ns = new Fieldmapper::metabib::full_rec;
459 $ns->tag( $tagline->getAttribute( "tag" ) );
460 my $val = $tagline->textContent;
461 $val =~ s/(\pM)//gso;
467 for my $tagline ( @{$root->getChildrenByTagName("datafield")} ) {
468 next unless $tagline;
470 my $tag = $tagline->getAttribute( "tag" );
471 my $ind1 = $tagline->getAttribute( "ind1" );
472 my $ind2 = $tagline->getAttribute( "ind2" );
474 for my $data ( $tagline->childNodes ) {
477 my $ns = new Fieldmapper::metabib::full_rec;
482 $ns->subfield( $data->getAttribute( "code" ) );
483 my $val = $data->textContent;
484 $val =~ s/(\pM)//gso;
485 $ns->value( lc($val) );
493 sub _get_field_value {
495 my( $root, $xpath ) = @_;
499 # grab the set of matching nodes
500 my @nodes = $root->findnodes( $xpath );
501 for my $value (@nodes) {
503 # grab all children of the node
504 my @children = $value->childNodes();
505 for my $child (@children) {
507 # add the childs content to the growing buffer
508 my $content = quotemeta($child->textContent);
509 next if ($string =~ /$content/); # uniquify the values
510 $string .= $child->textContent . " ";
513 $string .= $value->textContent . " ";
516 $string =~ s/(\pM)//gso;
521 sub modsdoc_to_values {
522 my( $self, $mods ) = @_;
524 for my $class (keys %$xpathset) {
525 $data->{$class} = {};
526 for my $type (keys %{$xpathset->{$class}}) {
527 $data->{$class}->{$type} = {};
528 $data->{$class}->{$type}->{value} = _get_field_value( $mods, $xpathset->{$class}->{$type}->{xpath} );
529 $data->{$class}->{$type}->{field_id} = $xpathset->{$class}->{$type}->{id};