1 package OpenILS::Application::Ingest;
2 use OpenILS::Application;
3 use base qw/OpenILS::Application/;
5 use Unicode::Normalize;
6 use OpenSRF::EX qw/:try/;
8 use OpenSRF::AppSession;
9 use OpenSRF::Utils::SettingsClient;
10 use OpenSRF::Utils::Logger qw/:level/;
12 use OpenILS::Utils::ScriptRunner;
13 use OpenILS::Utils::Fieldmapper;
14 use OpenSRF::Utils::JSON;
16 use OpenILS::Utils::Fieldmapper;
20 use Time::HiRes qw(time);
22 our %supported_formats = (
23 mods32 => {ns => 'http://www.loc.gov/mods/v3'},
24 mods3 => {ns => 'http://www.loc.gov/mods/v3'},
25 mods => {ns => 'http://www.loc.gov/mods/'},
26 marcxml => {ns => 'http://www.loc.gov/MARC21/slim'},
27 srw_dc => {ns => 'info:srw/schema/1/dc-schema'},
28 oai_dc => {ns => 'http://www.openarchives.org/OAI/2.0/oai_dc/'},
29 rdf_dc => {ns => 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'},
30 atom => {ns => 'http://www.w3.org/2005/Atom'},
31 rss091 => {ns => 'http://my.netscape.com/rdf/simple/0.9/'},
35 rss10 => {ns => 'http://purl.org/rss/1.0/'},
36 rss11 => {ns => 'http://purl.org/net/rss1.1#'},
41 my $log = 'OpenSRF::Utils::Logger';
43 my $parser = XML::LibXML->new();
44 my $xslt = XML::LibXSLT->new();
54 unless (keys %$xpathset) {
55 $log->debug("Running post_init", DEBUG);
57 my $xsldir = OpenSRF::Utils::SettingsClient->new->config_value(dirs => 'xsl');
59 unless ($supported_formats{mods}{xslt}) {
60 $log->debug("Loading MODS XSLT", DEBUG);
61 my $xslt_doc = $parser->parse_file( $xsldir . "/MARC21slim2MODS.xsl");
62 $supported_formats{mods}{xslt} = $xslt->parse_stylesheet( $xslt_doc );
65 unless ($supported_formats{mods3}{xslt}) {
66 $log->debug("Loading MODS v3 XSLT", DEBUG);
67 my $xslt_doc = $parser->parse_file( $xsldir . "/MARC21slim2MODS3.xsl");
68 $supported_formats{mods3}{xslt} = $xslt->parse_stylesheet( $xslt_doc );
71 unless ($supported_formats{mods32}{xslt}) {
72 $log->debug("Loading MODS v32 XSLT", DEBUG);
73 my $xslt_doc = $parser->parse_file( $xsldir . "/MARC21slim2MODS32.xsl");
74 $supported_formats{mods32}{xslt} = $xslt->parse_stylesheet( $xslt_doc );
77 my $req = OpenSRF::AppSession
78 ->create('open-ils.cstore')
80 # XXX testing new metabib field use for faceting
81 #->request( 'open-ils.cstore.direct.config.metabib_field.search.atomic', { id => { '!=' => undef } } )
82 ->request( 'open-ils.cstore.direct.config.metabib_field.search.atomic', { search_field => 't' } )
86 if (ref $req and @$req) {
88 $xpathset->{ $f->field_class }->{ $f->name }->{xpath} = $f->xpath;
89 $xpathset->{ $f->field_class }->{ $f->name }->{id} = $f->id;
90 $xpathset->{ $f->field_class }->{ $f->name }->{format} = $f->format;
91 $log->debug("Loaded XPath from DB: ".$f->field_class." => ".$f->name." : ".$f->xpath, DEBUG);
102 $stuff = NFD($stuff);
104 $stuff = NFC($stuff);
107 $stuff =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
111 # --------------------------------------------------------------------------------
114 package OpenILS::Application::Ingest::Biblio;
115 use base qw/OpenILS::Application::Ingest/;
116 use Unicode::Normalize;
118 sub rw_biblio_ingest_single_object {
123 my ($blob) = $self->method_lookup("open-ils.ingest.full.biblio.object.readonly")->run($bib);
124 return undef unless ($blob);
126 $bib->fingerprint( $blob->{fingerprint}->{fingerprint} );
127 $bib->quality( $blob->{fingerprint}->{quality} );
129 my $cstore = OpenSRF::AppSession->connect('open-ils.cstore');
131 my $xact = $cstore->request('open-ils.cstore.transaction.begin')->gather(1);
133 # update full_rec stuff ...
134 my $tmp = $cstore->request(
135 'open-ils.cstore.direct.metabib.full_rec.id_list.atomic',
136 { record => $bib->id }
139 $cstore->request( 'open-ils.cstore.direct.metabib.full_rec.delete' => $_ )->gather(1) for (@$tmp);
140 $cstore->request( 'open-ils.cstore.direct.metabib.full_rec.create' => $_ )->gather(1) for (@{ $blob->{full_rec} });
142 # update rec_descriptor stuff ...
143 $tmp = $cstore->request(
144 'open-ils.cstore.direct.metabib.record_descriptor.id_list.atomic',
145 { record => $bib->id }
148 $cstore->request( 'open-ils.cstore.direct.metabib.record_descriptor.delete' => $_ )->gather(1) for (@$tmp);
149 $cstore->request( 'open-ils.cstore.direct.metabib.record_descriptor.create' => $blob->{descriptor} )->gather(1);
151 # deal with classed fields...
152 for my $class ( qw/title author subject keyword series/ ) {
153 $tmp = $cstore->request(
154 "open-ils.cstore.direct.metabib.${class}_field_entry.id_list.atomic",
155 { source => $bib->id }
158 $cstore->request( "open-ils.cstore.direct.metabib.${class}_field_entry.delete" => $_ )->gather(1) for (@$tmp);
160 for my $obj ( @{ $blob->{field_entries} } ) {
161 my $class = $obj->class_name;
162 $class =~ s/^Fieldmapper:://o;
164 $cstore->request( "open-ils.cstore.direct.$class.create" => $obj )->gather(1);
169 $tmp = $cstore->request(
170 'open-ils.cstore.direct.metabib.metarecord_source_map.search.atomic',
171 { source => $bib->id }
174 $cstore->request( 'open-ils.cstore.direct.metabib.metarecord_source_map.delete' => $_->id )->gather(1) for (@$tmp);
177 my $old_mrs = $cstore->request(
178 'open-ils.cstore.direct.metabib.metarecord.search.atomic' => { id => [map { $_->metarecord } @$tmp] }
179 )->gather(1) if (@$tmp);
181 $old_mrs = [] if (!ref($old_mrs));
184 for my $m (@$old_mrs) {
185 if ($m->fingerprint eq $bib->fingerprint) {
188 my $others = $cstore->request(
189 'open-ils.cstore.direct.metabib.metarecord_source_map.id_list.atomic' => { metarecord => $m->id }
194 'open-ils.cstore.direct.metabib.metarecord.delete' => $m->id
204 # Get the matchin MR, if any.
205 $mr = $cstore->request(
206 'open-ils.cstore.direct.metabib.metarecord.search',
207 { fingerprint => $bib->fingerprint }
210 $holds = $cstore->request(
211 'open-ils.cstore.direct.action.hold_request.search.atomic',
212 { hold_type => 'M', target => [ map { $_->id } grep { $_->isdeleted } @$old_mrs ] }
213 )->gather(1) if (@$old_mrs);
216 for my $h (@$holds) {
218 $cstore->request( 'open-ils.cstore.direct.action.hold_request.update' => $h )->gather(1);
225 $mr = new Fieldmapper::metabib::metarecord;
226 $mr->fingerprint( $bib->fingerprint );
227 $mr->master_record( $bib->id );
230 "open-ils.cstore.direct.metabib.metarecord.create",
231 $mr => { quiet => 'true' }
235 for my $h (grep { !$_->ischanged } @$holds) {
237 $cstore->request( 'open-ils.cstore.direct.action.hold_request.update' => $h )->gather(1);
240 my $mrm = $cstore->request(
241 'open-ils.cstore.direct.metabib.metarecord_source_map.search.atomic',
242 { metarecord => $mr->id }
246 my $best = $cstore->request(
247 "open-ils.cstore.direct.biblio.record_entry.search",
248 { id => [ map { $_->source } @$mrm ] },
249 { 'select' => { bre => [ qw/id quality/ ] },
250 order_by => { bre => "quality desc" },
255 if ($best->quality > $bib->quality) {
256 $mr->master_record($best->id);
258 $mr->master_record($bib->id);
261 $mr->master_record($bib->id);
266 $cstore->request( 'open-ils.cstore.direct.metabib.metarecord.update' => $mr )->gather(1);
269 my $mrm = new Fieldmapper::metabib::metarecord_source_map;
270 $mrm->source($bib->id);
271 $mrm->metarecord($mr->id);
273 $cstore->request( 'open-ils.cstore.direct.metabib.metarecord_source_map.create' => $mrm )->gather(1);
274 $cstore->request( 'open-ils.cstore.direct.biblio.record_entry.update' => $bib )->gather(1);
276 $cstore->request( 'open-ils.cstore.transaction.commit' )->gather(1) || return undef;;
280 __PACKAGE__->register_method(
281 api_name => "open-ils.ingest.full.biblio.object",
282 method => "rw_biblio_ingest_single_object",
287 sub rw_biblio_ingest_single_record {
292 OpenILS::Application::Ingest->post_init();
293 my $cstore = OpenSRF::AppSession->connect( 'open-ils.cstore' );
294 $cstore->request('open-ils.cstore.transaction.begin')->gather(1);
296 my $r = $cstore->request( 'open-ils.cstore.direct.biblio.record_entry.retrieve' => $rec )->gather(1);
298 $cstore->request('open-ils.cstore.transaction.rollback')->gather(1);
301 return undef unless ($r and @$r);
303 return ($self->method_lookup("open-ils.ingest.full.biblio.object")->run($r))[0];
305 __PACKAGE__->register_method(
306 api_name => "open-ils.ingest.full.biblio.record",
307 method => "rw_biblio_ingest_single_record",
312 sub rw_biblio_ingest_record_list {
315 my @rec = ref($_[0]) ? @{ $_[0] } : @_ ;
317 OpenILS::Application::Ingest->post_init();
318 my $cstore = OpenSRF::AppSession->connect( 'open-ils.cstore' );
319 $cstore->request('open-ils.cstore.transaction.begin')->gather(1);
321 my $r = $cstore->request( 'open-ils.cstore.direct.biblio.record_entry.search.atomic' => { id => $rec } )->gather(1);
323 $cstore->request('open-ils.cstore.transaction.rollback')->gather(1);
326 return undef unless ($r and @$r);
329 $count += ($self->method_lookup("open-ils.ingest.full.biblio.object")->run($_))[0] for (@$r);
333 __PACKAGE__->register_method(
334 api_name => "open-ils.ingest.full.biblio.record_list",
335 method => "rw_biblio_ingest_record_list",
340 sub ro_biblio_ingest_single_object {
344 my $xml = OpenILS::Application::Ingest::entityize($bib->marc);
346 my $document = $parser->parse_string($xml);
348 my @mfr = $self->method_lookup("open-ils.ingest.flat_marc.biblio.xml")->run($document);
349 my @mXfe = $self->method_lookup("open-ils.ingest.extract.field_entry.all.xml")->run($document);
350 my ($fp) = $self->method_lookup("open-ils.ingest.fingerprint.xml")->run($xml);
351 my ($rd) = $self->method_lookup("open-ils.ingest.descriptor.xml")->run($xml);
353 $_->source($bib->id) for (@mXfe);
354 $_->record($bib->id) for (@mfr);
355 $rd->record($bib->id) if ($rd);
357 return { full_rec => \@mfr, field_entries => \@mXfe, fingerprint => $fp, descriptor => $rd };
359 __PACKAGE__->register_method(
360 api_name => "open-ils.ingest.full.biblio.object.readonly",
361 method => "ro_biblio_ingest_single_object",
366 sub ro_biblio_ingest_single_xml {
369 my $xml = OpenILS::Application::Ingest::entityize(shift);
371 my $document = $parser->parse_string($xml);
373 my @mfr = $self->method_lookup("open-ils.ingest.flat_marc.biblio.xml")->run($document);
374 my @mXfe = $self->method_lookup("open-ils.ingest.extract.field_entry.all.xml")->run($document);
375 my ($fp) = $self->method_lookup("open-ils.ingest.fingerprint.xml")->run($xml);
376 my ($rd) = $self->method_lookup("open-ils.ingest.descriptor.xml")->run($xml);
378 return { full_rec => \@mfr, field_entries => \@mXfe, fingerprint => $fp, descriptor => $rd };
380 __PACKAGE__->register_method(
381 api_name => "open-ils.ingest.full.biblio.xml.readonly",
382 method => "ro_biblio_ingest_single_xml",
387 sub ro_biblio_ingest_single_record {
392 OpenILS::Application::Ingest->post_init();
393 my $r = OpenSRF::AppSession
394 ->create('open-ils.cstore')
395 ->request( 'open-ils.cstore.direct.biblio.record_entry.retrieve' => $rec )
398 return undef unless ($r and @$r);
400 my ($res) = $self->method_lookup("open-ils.ingest.full.biblio.xml.readonly")->run($r->marc);
402 $_->source($rec) for (@{$res->{field_entries}});
403 $_->record($rec) for (@{$res->{full_rec}});
404 $res->{descriptor}->record($rec);
408 __PACKAGE__->register_method(
409 api_name => "open-ils.ingest.full.biblio.record.readonly",
410 method => "ro_biblio_ingest_single_record",
415 sub ro_biblio_ingest_stream_record {
419 OpenILS::Application::Ingest->post_init();
421 my $ses = OpenSRF::AppSession->create('open-ils.cstore');
423 while (my ($resp) = $client->recv( count => 1, timeout => 5 )) {
425 my $rec = $resp->content;
426 last unless (defined $rec);
428 $log->debug("Running open-ils.ingest.full.biblio.record.readonly ...");
429 my ($res) = $self->method_lookup("open-ils.ingest.full.biblio.record.readonly")->run($rec);
431 $_->source($rec) for (@{$res->{field_entries}});
432 $_->record($rec) for (@{$res->{full_rec}});
434 $client->respond( $res );
439 __PACKAGE__->register_method(
440 api_name => "open-ils.ingest.full.biblio.record_stream.readonly",
441 method => "ro_biblio_ingest_stream_record",
446 sub ro_biblio_ingest_stream_xml {
450 OpenILS::Application::Ingest->post_init();
452 my $ses = OpenSRF::AppSession->create('open-ils.cstore');
454 while (my ($resp) = $client->recv( count => 1, timeout => 5 )) {
456 my $xml = $resp->content;
457 last unless (defined $xml);
459 $log->debug("Running open-ils.ingest.full.biblio.xml.readonly ...");
460 my ($res) = $self->method_lookup("open-ils.ingest.full.biblio.xml.readonly")->run($xml);
462 $client->respond( $res );
467 __PACKAGE__->register_method(
468 api_name => "open-ils.ingest.full.biblio.xml_stream.readonly",
469 method => "ro_biblio_ingest_stream_xml",
474 sub rw_biblio_ingest_stream_import {
478 OpenILS::Application::Ingest->post_init();
480 my $ses = OpenSRF::AppSession->create('open-ils.cstore');
482 while (my ($resp) = $client->recv( count => 1, timeout => 5 )) {
484 my $bib = $resp->content;
485 last unless (defined $bib);
487 $log->debug("Running open-ils.ingest.full.biblio.xml.readonly ...");
488 my ($res) = $self->method_lookup("open-ils.ingest.full.biblio.xml.readonly")->run($bib->marc);
490 $_->source($bib->id) for (@{$res->{field_entries}});
491 $_->record($bib->id) for (@{$res->{full_rec}});
493 $client->respond( $res );
498 __PACKAGE__->register_method(
499 api_name => "open-ils.ingest.full.biblio.bib_stream.import",
500 method => "rw_biblio_ingest_stream_import",
506 # --------------------------------------------------------------------------------
509 package OpenILS::Application::Ingest::Authority;
510 use base qw/OpenILS::Application::Ingest/;
511 use Unicode::Normalize;
513 sub ro_authority_ingest_single_object {
517 my $xml = OpenILS::Application::Ingest::entityize($bib->marc);
519 my $document = $parser->parse_string($xml);
521 my @mfr = $self->method_lookup("open-ils.ingest.flat_marc.authority.xml")->run($document);
523 $_->record($bib->id) for (@mfr);
525 return { full_rec => \@mfr };
527 __PACKAGE__->register_method(
528 api_name => "open-ils.ingest.full.authority.object.readonly",
529 method => "ro_authority_ingest_single_object",
534 sub ro_authority_ingest_single_xml {
537 my $xml = OpenILS::Application::Ingest::entityize(shift);
539 my $document = $parser->parse_string($xml);
541 my @mfr = $self->method_lookup("open-ils.ingest.flat_marc.authority.xml")->run($document);
543 return { full_rec => \@mfr };
545 __PACKAGE__->register_method(
546 api_name => "open-ils.ingest.full.authority.xml.readonly",
547 method => "ro_authority_ingest_single_xml",
552 sub ro_authority_ingest_single_record {
557 OpenILS::Application::Ingest->post_init();
558 my $r = OpenSRF::AppSession
559 ->create('open-ils.cstore')
560 ->request( 'open-ils.cstore.direct.authority.record_entry.retrieve' => $rec )
563 return undef unless ($r and @$r);
565 my ($res) = $self->method_lookup("open-ils.ingest.full.authority.xml.readonly")->run($r->marc);
567 $_->record($rec) for (@{$res->{full_rec}});
568 $res->{descriptor}->record($rec);
572 __PACKAGE__->register_method(
573 api_name => "open-ils.ingest.full.authority.record.readonly",
574 method => "ro_authority_ingest_single_record",
579 sub ro_authority_ingest_stream_record {
583 OpenILS::Application::Ingest->post_init();
585 my $ses = OpenSRF::AppSession->create('open-ils.cstore');
587 while (my ($resp) = $client->recv( count => 1, timeout => 5 )) {
589 my $rec = $resp->content;
590 last unless (defined $rec);
592 $log->debug("Running open-ils.ingest.full.authority.record.readonly ...");
593 my ($res) = $self->method_lookup("open-ils.ingest.full.authority.record.readonly")->run($rec);
595 $_->record($rec) for (@{$res->{full_rec}});
597 $client->respond( $res );
602 __PACKAGE__->register_method(
603 api_name => "open-ils.ingest.full.authority.record_stream.readonly",
604 method => "ro_authority_ingest_stream_record",
609 sub ro_authority_ingest_stream_xml {
613 OpenILS::Application::Ingest->post_init();
615 my $ses = OpenSRF::AppSession->create('open-ils.cstore');
617 while (my ($resp) = $client->recv( count => 1, timeout => 5 )) {
619 my $xml = $resp->content;
620 last unless (defined $xml);
622 $log->debug("Running open-ils.ingest.full.authority.xml.readonly ...");
623 my ($res) = $self->method_lookup("open-ils.ingest.full.authority.xml.readonly")->run($xml);
625 $client->respond( $res );
630 __PACKAGE__->register_method(
631 api_name => "open-ils.ingest.full.authority.xml_stream.readonly",
632 method => "ro_authority_ingest_stream_xml",
637 sub rw_authority_ingest_stream_import {
641 OpenILS::Application::Ingest->post_init();
643 my $ses = OpenSRF::AppSession->create('open-ils.cstore');
645 while (my ($resp) = $client->recv( count => 1, timeout => 5 )) {
647 my $bib = $resp->content;
648 last unless (defined $bib);
650 $log->debug("Running open-ils.ingest.full.authority.xml.readonly ...");
651 my ($res) = $self->method_lookup("open-ils.ingest.full.authority.xml.readonly")->run($bib->marc);
653 $_->record($bib->id) for (@{$res->{full_rec}});
655 $client->respond( $res );
660 __PACKAGE__->register_method(
661 api_name => "open-ils.ingest.full.authority.bib_stream.import",
662 method => "rw_authority_ingest_stream_import",
668 # --------------------------------------------------------------------------------
669 # MARC index extraction
671 package OpenILS::Application::Ingest::XPATH;
672 use base qw/OpenILS::Application::Ingest/;
673 use Unicode::Normalize;
675 # give this an XML documentElement and an XPATH expression
676 sub xpath_to_string {
680 my $ns_prefix = shift;
683 $xml->setNamespace( $ns_uri, $ns_prefix, 1 ) if ($ns_uri && $ns_prefix);
687 # grab the set of matching nodes
688 my @nodes = $xml->findnodes( $xpath );
689 for my $value (@nodes) {
691 # grab all children of the node
692 my @children = $value->childNodes();
693 for my $child (@children) {
695 # add the childs content to the growing buffer
696 my $content = quotemeta($child->textContent);
697 next if ($unique && $string =~ /$content/); # uniquify the values
698 $string .= $child->textContent . " ";
701 $string .= $value->textContent . " ";
705 $string =~ s/(\d{4})-(\d{4})/$1 $2/sgo;
710 sub class_index_string_xml {
716 OpenILS::Application::Ingest->post_init();
717 $xml = $parser->parse_string(OpenILS::Application::Ingest::entityize($xml)) unless (ref $xml);
721 for my $class (@classes) {
722 my $class_constructor = "Fieldmapper::metabib::${class}_field_entry";
723 for my $type ( keys %{ $xpathset->{$class} } ) {
725 my $def = $xpathset->{$class}->{$type};
726 my $sf = $OpenILS::Application::Ingest::supported_formats{$def->{format}};
731 $document = $transform_cache{$def->{format}} || $sf->{xslt}->transform($xml);
732 $transform_cache{$def->{format}} = $document;
735 my $value = xpath_to_string(
736 $document->documentElement => $def->{xpath},
737 $sf->{ns} => $def->{format},
743 $value = NFD($value);
744 $value =~ s/\pM+//sgo;
745 $value =~ s/\pC+//sgo;
746 $value =~ s/\W+$//sgo;
748 $value =~ s/\b\.+\b//sgo;
751 my $fm = $class_constructor->new;
752 $fm->value( $value );
753 $fm->field( $xpathset->{$class}->{$type}->{id} );
754 $client->respond($fm);
759 __PACKAGE__->register_method(
760 api_name => "open-ils.ingest.field_entry.class.xml",
761 method => "class_index_string_xml",
767 sub class_index_string_record {
773 OpenILS::Application::Ingest->post_init();
774 my $r = OpenSRF::AppSession
775 ->create('open-ils.cstore')
776 ->request( 'open-ils.cstore.direct.authority.record_entry.retrieve' => $rec )
779 return undef unless ($r and @$r);
781 for my $fm ($self->method_lookup("open-ils.ingest.field_entry.class.xml")->run($r->marc, @classes)) {
783 $client->respond($fm);
787 __PACKAGE__->register_method(
788 api_name => "open-ils.ingest.field_entry.class.record",
789 method => "class_index_string_record",
795 sub all_index_string_xml {
800 for my $fm ($self->method_lookup("open-ils.ingest.field_entry.class.xml")->run($xml, keys(%$xpathset))) {
801 $client->respond($fm);
805 __PACKAGE__->register_method(
806 api_name => "open-ils.ingest.extract.field_entry.all.xml",
807 method => "all_index_string_xml",
813 sub all_index_string_record {
818 OpenILS::Application::Ingest->post_init();
819 my $r = OpenSRF::AppSession
820 ->create('open-ils.cstore')
821 ->request( 'open-ils.cstore.direct.biblio.record_entry.retrieve' => $rec )
824 return undef unless ($r and @$r);
826 for my $fm ($self->method_lookup("open-ils.ingest.field_entry.class.xml")->run($r->marc, keys(%$xpathset))) {
828 $client->respond($fm);
832 __PACKAGE__->register_method(
833 api_name => "open-ils.ingest.extract.field_entry.all.record",
834 method => "all_index_string_record",
840 # --------------------------------------------------------------------------------
843 package OpenILS::Application::Ingest::FlatMARC;
844 use base qw/OpenILS::Application::Ingest/;
845 use Unicode::Normalize;
848 sub _marcxml_to_full_rows {
851 my $xmltype = shift || 'metabib';
853 my $type = "Fieldmapper::${xmltype}::full_rec";
857 my ($root) = $marcxml->findnodes('//*[local-name()="record"]');
859 for my $tagline ( @{$root->getChildrenByTagName("leader")} ) {
860 next unless $tagline;
865 my $val = $tagline->textContent;
875 for my $tagline ( @{$root->getChildrenByTagName("controlfield")} ) {
876 next unless $tagline;
880 $ns->tag( $tagline->getAttribute( "tag" ) );
881 my $val = $tagline->textContent;
891 for my $tagline ( @{$root->getChildrenByTagName("datafield")} ) {
892 next unless $tagline;
894 my $tag = $tagline->getAttribute( "tag" );
895 my $ind1 = $tagline->getAttribute( "ind1" );
896 my $ind2 = $tagline->getAttribute( "ind2" );
898 for my $data ( @{$tagline->getChildrenByTagName('subfield')} ) {
906 $ns->subfield( $data->getAttribute( "code" ) );
907 my $val = $data->textContent;
912 $val =~ s/(\d{4})-(\d{4})/$1 $2/sgo;
913 $ns->value( lc($val) );
919 $log->debug("Returning ".scalar(@ns_list)." Fieldmapper nodes from $xmltype xml");
928 $log->debug("processing [$xml]");
930 $xml = $parser->parse_string(OpenILS::Application::Ingest::entityize($xml)) unless (ref $xml);
932 my $type = 'metabib';
933 $type = 'authority' if ($self->api_name =~ /authority/o);
935 OpenILS::Application::Ingest->post_init();
937 $client->respond($_) for (_marcxml_to_full_rows($xml, $type));
940 __PACKAGE__->register_method(
941 api_name => "open-ils.ingest.flat_marc.authority.xml",
942 method => "flat_marc_xml",
947 __PACKAGE__->register_method(
948 api_name => "open-ils.ingest.flat_marc.biblio.xml",
949 method => "flat_marc_xml",
955 sub flat_marc_record {
961 $type = 'authority' if ($self->api_name =~ /authority/o);
963 OpenILS::Application::Ingest->post_init();
964 my $r = OpenSRF::AppSession
965 ->create('open-ils.cstore')
966 ->request( "open-ils.cstore.direct.${type}.record_entry.retrieve" => $rec )
970 return undef unless ($r and $r->marc);
972 my @rows = $self->method_lookup("open-ils.ingest.flat_marc.$type.xml")->run($r->marc);
973 for my $row (@rows) {
974 $client->respond($row);
975 $log->debug(OpenSRF::Utils::JSON->perl2JSON($row), DEBUG);
979 __PACKAGE__->register_method(
980 api_name => "open-ils.ingest.flat_marc.biblio.record_entry",
981 method => "flat_marc_record",
986 __PACKAGE__->register_method(
987 api_name => "open-ils.ingest.flat_marc.authority.record_entry",
988 method => "flat_marc_record",
994 # --------------------------------------------------------------------------------
997 package OpenILS::Application::Ingest::Biblio::Fingerprint;
998 use base qw/OpenILS::Application::Ingest/;
999 use Unicode::Normalize;
1000 use OpenSRF::EX qw/:try/;
1002 sub biblio_fingerprint_record {
1007 OpenILS::Application::Ingest->post_init();
1009 my $r = OpenSRF::AppSession
1010 ->create('open-ils.cstore')
1011 ->request( 'open-ils.cstore.direct.biblio.record_entry.retrieve' => $rec )
1014 return undef unless ($r and $r->marc);
1016 my ($fp) = $self->method_lookup('open-ils.ingest.fingerprint.xml')->run($r->marc);
1017 $log->debug("Returning [$fp] as fingerprint for record $rec", INFO);
1018 $fp->{quality} = int($fp->{quality});
1021 __PACKAGE__->register_method(
1022 api_name => "open-ils.ingest.fingerprint.record",
1023 method => "biblio_fingerprint_record",
1029 sub biblio_fingerprint {
1032 my $xml = OpenILS::Application::Ingest::entityize(shift);
1034 $log->internal("Got MARC [$xml]");
1037 my @pfx = ( "apps", "open-ils.ingest","app_settings" );
1038 my $conf = OpenSRF::Utils::SettingsClient->new;
1040 my $libs = $conf->config_value(@pfx, 'script_path');
1041 my $script_file = $conf->config_value(@pfx, 'scripts', 'biblio_fingerprint');
1042 my $script_libs = (ref($libs)) ? $libs : [$libs];
1044 $log->debug("Loading script $script_file for biblio fingerprinting...");
1046 $fp_script = new OpenILS::Utils::ScriptRunner
1047 ( file => $script_file,
1048 paths => $script_libs,
1049 reset_count => 100 );
1052 $fp_script->insert('environment' => {marc => $xml} => 1);
1054 my $res = $fp_script->run || ($log->error( "Fingerprint script died! $@" ) && return undef);
1055 $log->debug("Script for biblio fingerprinting completed successfully...");
1059 __PACKAGE__->register_method(
1060 api_name => "open-ils.ingest.fingerprint.xml",
1061 method => "biblio_fingerprint",
1067 sub biblio_descriptor {
1070 my $xml = OpenILS::Application::Ingest::entityize(shift);
1072 $log->internal("Got MARC [$xml]");
1075 my @pfx = ( "apps", "open-ils.ingest","app_settings" );
1076 my $conf = OpenSRF::Utils::SettingsClient->new;
1078 my $libs = $conf->config_value(@pfx, 'script_path');
1079 my $script_file = $conf->config_value(@pfx, 'scripts', 'biblio_descriptor');
1080 my $script_libs = (ref($libs)) ? $libs : [$libs];
1082 $log->debug("Loading script $script_file for biblio descriptor extraction...");
1084 $rd_script = new OpenILS::Utils::ScriptRunner
1085 ( file => $script_file,
1086 paths => $script_libs,
1087 reset_count => 100 );
1090 $log->debug("Setting up environment for descriptor extraction script...");
1091 $rd_script->insert('environment.marc' => $xml => 1);
1092 $log->debug("Environment building complete...");
1094 my $res = $rd_script->run || ($log->error( "Descriptor script died! $@" ) && return undef);
1095 $log->debug("Script for biblio descriptor extraction completed successfully");
1097 my $d1 = $res->date1;
1098 if ($d1 && $d1 ne ' ') {
1103 my $d2 = $res->date2;
1104 if ($d2 && $d2 ne ' ') {
1111 __PACKAGE__->register_method(
1112 api_name => "open-ils.ingest.descriptor.xml",
1113 method => "biblio_descriptor",