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 # --------------------------------------------------------------------------------
183 my $create_source_map;
197 my %descriptor_code = (
198 item_type => 'substr($ldr,6,1)',
199 item_form => '(substr($ldr,6,1) =~ /^(?:f|g|i|m|o|p|r)$/) ? substr($oo8,29,1) : substr($oo8,23,1)',
200 bib_level => 'substr($ldr,7,1)',
201 control_type => 'substr($ldr,8,1)',
202 char_encoding => 'substr($ldr,9,1)',
203 enc_level => 'substr($ldr,17,1)',
204 cat_form => 'substr($ldr,18,1)',
205 pub_status => 'substr($ldr,5,1)',
206 item_lang => 'substr($oo8,35,3)',
207 audience => 'substr($oo8,22,1)',
217 if ($self->api_name =~ /no_map/o) {
221 $in_xact = $self->method_lookup( 'open-ils.storage.transaction.current')
223 $begin = $self->method_lookup( 'open-ils.storage.transaction.begin')
225 $commit = $self->method_lookup( 'open-ils.storage.transaction.commit')
227 $rollback = $self->method_lookup( 'open-ils.storage.transaction.rollback')
229 $sm_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.search.source')
231 $mr_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.search.fingerprint')
233 $mr_update = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.batch.update')
235 $mr_create = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.create')
237 $create_source_map = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.batch.create')
238 unless ($create_source_map);
239 $lookup = $self->method_lookup('open-ils.storage.direct.biblio.record_entry.batch.retrieve')
241 $update_entry = $self->method_lookup('open-ils.storage.direct.biblio.record_entry.batch.update')
242 unless ($update_entry);
243 $rm_old_sm = $self->method_lookup( 'open-ils.storage.direct.metabib.metarecord_source_map.mass_delete')
245 $rm_old_rd = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.mass_delete')
247 $rm_old_fr = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.mass_delete')
249 $rm_old_tr = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.mass_delete')
251 $rm_old_ar = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.mass_delete')
253 $rm_old_sr = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.mass_delete')
255 $rm_old_kr = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.mass_delete')
257 $rd_create = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.batch.create')
259 $fr_create = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.batch.create')
261 $$create{title} = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.batch.create')
262 unless ($$create{title});
263 $$create{author} = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.batch.create')
264 unless ($$create{author});
265 $$create{subject} = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.batch.create')
266 unless ($$create{subject});
267 $$create{keyword} = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.batch.create')
268 unless ($$create{keyword});
271 my ($outer_xact) = $in_xact->run;
273 unless ($outer_xact) {
274 my ($r) = $begin->run($client);
275 unless (defined $r and $r) {
277 throw OpenSRF::EX::PANIC ("Couldn't BEGIN transaction!")
281 throw OpenSRF::EX::PANIC ("WoRM Couldn't BEGIN transaction!")
291 for my $entry ( $lookup->run(@docids) ) {
292 # step -1: grab the doc from storage
293 next unless ($entry);
295 my $xml = $entry->marc;
296 my $docid = $entry->id;
297 my $marcdoc = $parser->parse_string($xml);
298 my $modsdoc = $mods_sheet->transform($marcdoc);
300 my $mods = $modsdoc->documentElement;
301 $mods->setNamespace( "http://www.loc.gov/mods/", "mods", 1 );
303 $entry->fingerprint( fingerprint_mods( $mods ) );
304 push @entry_list, $entry;
307 my ($mr) = $mr_lookup->run( $entry->fingerprint );
309 $mr = new Fieldmapper::metabib::metarecord;
310 $mr->fingerprint( $entry->fingerprint );
311 $mr->master_record( $entry->id );
312 my ($new_mr) = $mr_create->run($mr);
314 unless (defined $mr) {
315 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord.create!")
323 my $sm = new Fieldmapper::metabib::metarecord_source_map;
324 $sm->metarecord( $mr->id );
325 $sm->source( $entry->id );
326 push @source_maps, $sm;
329 my $ldr = $marcdoc->documentElement->getChildrenByTagName('leader')->pop->textContent;
330 my $oo8 = $marcdoc->documentElement->findvalue('//*[local-name()="controlfield" and @tag="008"]');
332 my $rd_obj = Fieldmapper::metabib::record_descriptor->new;
333 for my $rd_field ( keys %descriptor_code ) {
334 $rd_obj->$rd_field( eval "$descriptor_code{$rd_field};" );
336 $rd_obj->record( $docid );
337 push @rd_list, $rd_obj;
339 push @mods_data, { $docid => $self->modsdoc_to_values( $mods ) };
341 # step 2: build the KOHA rows
342 my @tmp_list = _marcxml_to_full_rows( $marcdoc );
343 $_->record( $docid ) for (@tmp_list);
344 push @ns_list, @tmp_list;
348 last unless ($self->api_name =~ /batch$/o);
351 $rm_old_rd->run( { record => \@docids } );
352 $rm_old_fr->run( { record => \@docids } );
353 $rm_old_sm->run( { source => \@docids } ) unless ($no_map);
354 $rm_old_tr->run( { source => \@docids } );
355 $rm_old_ar->run( { source => \@docids } );
356 $rm_old_sr->run( { source => \@docids } );
357 $rm_old_kr->run( { source => \@docids } );
360 my ($sm) = $create_source_map->run(@source_maps);
361 unless (defined $sm) {
362 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord_source_map.batch.create!")
364 my ($mr) = $mr_update->run(@mr_list);
365 unless (defined $mr) {
366 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord.batch.update!")
370 my ($re) = $update_entry->run(@entry_list);
371 unless (defined $re) {
372 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.biblio.record_entry.batch.update!")
375 my ($rd) = $rd_create->run(@rd_list);
376 unless (defined $rd) {
377 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.record_descriptor.batch.create!")
380 my ($fr) = $fr_create->run(@ns_list);
381 unless (defined $fr) {
382 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.full_rec.batch.create!")
385 # step 5: insert the new metadata
386 for my $class ( qw/title author subject keyword/ ) {
388 for my $doc ( @mods_data ) {
389 my ($did) = keys %$doc;
390 my ($data) = values %$doc;
392 my $fm_constructor = "Fieldmapper::metabib::${class}_field_entry";
393 for my $row ( keys %{ $$data{$class} } ) {
394 next unless (exists $$data{$class}{$row});
395 next unless ($$data{$class}{$row}{value});
396 my $fm_obj = $fm_constructor->new;
397 $fm_obj->value( $$data{$class}{$row}{value} );
398 $fm_obj->field( $$data{$class}{$row}{field_id} );
399 $fm_obj->source( $did );
400 $log->debug("$class entry: ".$fm_obj->source." => ".$fm_obj->field." : ".$fm_obj->value, DEBUG);
402 push @md_list, $fm_obj;
406 my ($cr) = $$create{$class}->run(@md_list);
407 unless (defined $cr) {
408 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.${class}_field_entry.batch.create!")
412 unless ($outer_xact) {
413 my ($c) = $commit->run;
414 unless (defined $c and $c) {
416 throw OpenSRF::EX::PANIC ("Couldn't COMMIT changes!")
422 __PACKAGE__->register_method(
423 api_name => "open-ils.worm.wormize",
428 __PACKAGE__->register_method(
429 api_name => "open-ils.worm.wormize.no_map",
434 __PACKAGE__->register_method(
435 api_name => "open-ils.worm.wormize.batch",
440 __PACKAGE__->register_method(
441 api_name => "open-ils.worm.wormize.no_map.batch",
448 # --------------------------------------------------------------------------------
451 sub _marcxml_to_full_rows {
457 my $root = $marcxml->documentElement;
459 for my $tagline ( @{$root->getChildrenByTagName("leader")} ) {
460 next unless $tagline;
462 my $ns = new Fieldmapper::metabib::full_rec;
465 my $val = $tagline->textContent;
466 $val =~ s/(\pM)//gso;
472 for my $tagline ( @{$root->getChildrenByTagName("controlfield")} ) {
473 next unless $tagline;
475 my $ns = new Fieldmapper::metabib::full_rec;
477 $ns->tag( $tagline->getAttribute( "tag" ) );
478 my $val = $tagline->textContent;
479 $val =~ s/(\pM)//gso;
485 for my $tagline ( @{$root->getChildrenByTagName("datafield")} ) {
486 next unless $tagline;
488 my $tag = $tagline->getAttribute( "tag" );
489 my $ind1 = $tagline->getAttribute( "ind1" );
490 my $ind2 = $tagline->getAttribute( "ind2" );
492 for my $data ( $tagline->childNodes ) {
495 my $ns = new Fieldmapper::metabib::full_rec;
500 $ns->subfield( $data->getAttribute( "code" ) );
501 my $val = $data->textContent;
502 $val =~ s/(\pM)//gso;
503 $ns->value( lc($val) );
511 sub _get_field_value {
513 my( $root, $xpath ) = @_;
517 # grab the set of matching nodes
518 my @nodes = $root->findnodes( $xpath );
519 for my $value (@nodes) {
521 # grab all children of the node
522 my @children = $value->childNodes();
523 for my $child (@children) {
525 # add the childs content to the growing buffer
526 my $content = quotemeta($child->textContent);
527 next if ($string =~ /$content/); # uniquify the values
528 $string .= $child->textContent . " ";
531 $string .= $value->textContent . " ";
534 $string =~ s/(\pM)//gso;
539 sub modsdoc_to_values {
540 my( $self, $mods ) = @_;
542 for my $class (keys %$xpathset) {
543 $data->{$class} = {};
544 for my $type (keys %{$xpathset->{$class}}) {
545 $data->{$class}->{$type} = {};
546 $data->{$class}->{$type}->{value} = _get_field_value( $mods, $xpathset->{$class}->{$type}->{xpath} );
547 $data->{$class}->{$type}->{field_id} = $xpathset->{$class}->{$type}->{id};