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 OpenILS::Utils::FlatXML;
8 use OpenILS::Utils::Fieldmapper;
13 use Time::HiRes qw(time);
15 my $xml_util = OpenILS::Utils::FlatXML->new();
17 my $parser = XML::LibXML->new();
18 my $xslt = XML::LibXSLT->new();
19 my $xslt_doc = $parser->parse_file( "/home/miker/cvs/OpenILS/app_server/stylesheets/MARC21slim2MODS.xsl" );
20 #my $xslt_doc = $parser->parse_file( "/pines/cvs/ILS/Open-ILS/xsl/MARC21slim2MODS.xsl" );
21 my $mods_sheet = $xslt->parse_stylesheet( $xslt_doc );
28 # __PACKAGE__->method_lookup('i.do.not.exist');
35 # get me from the database
41 "//mods:mods/mods:titleInfo[mods:title and (\@type='abreviated')]",
44 "//mods:mods/mods:titleInfo[mods:title and (\@type='translated')]",
47 "//mods:mods/mods:titleInfo[mods:title and (\@type='uniform')]",
50 "//mods:mods/mods:titleInfo[mods:title and not (\@type)]",
56 "//mods:mods/mods:name[\@type='corporate']/mods:namePart".
57 "[../mods:role/mods:text[text()='creator']][1]",
60 "//mods:mods/mods:name[\@type='personal']/mods:namePart".
61 "[../mods:role/mods:text[text()='creator']][1]",
64 "//mods:mods/mods:name[\@type='conference']/mods:namePart".
65 "[../mods:role/mods:text[text()='creator']][1]",
68 "//mods:mods/mods:name[\@type='personal']/mods:namePart",
74 "//mods:mods/mods:subject/mods:geographic",
77 "//mods:mods/mods:subject/mods:name",
80 "//mods:mods/mods:subject/mods:temporal",
83 "//mods:mods/mods:subject/mods:topic",
86 "//mods:mods/mods:genre",
90 keyword => { keyword => "//mods:mods/*[not(local-name()='originInfo')]", },
95 # --------------------------------------------------------------------------------
117 $begin = $self->method_lookup( 'open-ils.storage.transaction.begin')
119 $commit = $self->method_lookup( 'open-ils.storage.transaction.commit')
121 $rollback = $self->method_lookup( 'open-ils.storage.transaction.rollback')
123 $lookup = $self->method_lookup('open-ils.storage.biblio.record_marc.batch.retrieve')
125 $rm_old_fr = $self->method_lookup( 'open-ils.storage.metabib.full_rec.mass_delete')
127 $rm_old_tr = $self->method_lookup( 'open-ils.storage.metabib.title_field_entry.mass_delete')
129 $rm_old_ar = $self->method_lookup( 'open-ils.storage.metabib.author_field_entry.mass_delete')
131 $rm_old_sr = $self->method_lookup( 'open-ils.storage.metabib.subject_field_entry.mass_delete')
133 $rm_old_kr = $self->method_lookup( 'open-ils.storage.metabib.keyword_field_entry.mass_delete')
135 $fr_create = $self->method_lookup( 'open-ils.storage.metabib.full_rec.batch.create')
137 $$create{title} = $self->method_lookup( 'open-ils.storage.metabib.title_field_entry.batch.create')
138 unless ($$create{title});
139 $$create{author} = $self->method_lookup( 'open-ils.storage.metabib.author_field_entry.batch.create')
140 unless ($$create{author});
141 $$create{subject} = $self->method_lookup( 'open-ils.storage.metabib.subject_field_entry.batch.create')
142 unless ($$create{subject});
143 $$create{keyword} = $self->method_lookup( 'open-ils.storage.metabib.keyword_field_entry.batch.create')
144 unless ($$create{keyword});
148 my ($r) = $begin->run($client);
149 unless (defined $r and $r) {
151 throw OpenSRF::EX::PANIC ("Couldn't BEGIN transaction!")
154 throw OpenSRF::EX::PANIC ("WoRM Couldn't BEGIN transaction!")
160 for my $marc ( $lookup->run(@docids) ) {
161 # step -1: grab the doc from storage
164 my $xml = $marc->marc;
165 my $docid = $marc->id;
166 my $marcdoc = $parser->parse_string($xml);
168 push @mods_data, { $docid => $self->modsdoc_to_values( $mods_sheet->transform($marcdoc) ) };
170 # step 2: build the KOHA rows
171 my @tmp_list = _marcxml_to_full_rows( $marcdoc );
172 $_->record( $docid ) for (@tmp_list);
173 push @ns_list, @tmp_list;
177 last unless ($self->api_name =~ /batch$/o);
180 $rm_old_fr->run( { record => \@docids } );
181 $rm_old_tr->run( { source => \@docids } );
182 $rm_old_ar->run( { source => \@docids } );
183 $rm_old_sr->run( { source => \@docids } );
184 $rm_old_kr->run( { source => \@docids } );
186 my ($fr) = $fr_create->run(@ns_list);
187 unless (defined $fr) {
188 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.metabib.full_rec.batch.create!")
191 # step 5: insert the new metadata
192 for my $class ( qw/title author subject keyword/ ) {
194 for my $doc ( @mods_data ) {
195 my ($did) = keys %$doc;
196 my ($data) = values %$doc;
198 my $fm_constructor = "Fieldmapper::metabib::${class}_field_entry";
199 for my $row ( keys %{ $$data{$class} } ) {
200 next unless (exists $$data{$class}{$row});
201 next unless ($$data{$class}{$row});
202 my $fm_obj = $fm_constructor->new;
203 $fm_obj->value( $$data{$class}{$row} );
204 $fm_obj->source( $did );
206 # XXX This needs to be a real thing once the xpath is in the DB
209 push @md_list, $fm_obj;
213 my ($cr) = $$create{$class}->run(@md_list);
214 unless (defined $cr) {
215 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.metabib.${class}_field_entry.batch.create!")
219 my ($c) = $commit->run;
220 unless (defined $c and $c) {
222 throw OpenSRF::EX::PANIC ("Couldn't COMMIT changes!")
227 __PACKAGE__->register_method(
228 api_name => "open-ils.worm.wormize",
233 __PACKAGE__->register_method(
234 api_name => "open-ils.worm.wormize.batch",
241 # --------------------------------------------------------------------------------
244 sub _marcxml_to_full_rows {
250 my $root = $marcxml->documentElement;
252 for my $tagline ( @{$root->getChildrenByTagName("leader")} ) {
253 next unless $tagline;
255 my $ns = new Fieldmapper::metabib::full_rec;
258 my $val = $tagline->textContent;
259 $val =~ s/(\pM)//gso;
265 for my $tagline ( @{$root->getChildrenByTagName("controlfield")} ) {
266 next unless $tagline;
268 my $ns = new Fieldmapper::metabib::full_rec;
270 $ns->tag( $tagline->getAttribute( "tag" ) );
271 my $val = $tagline->textContent;
272 $val =~ s/(\pM)//gso;
278 for my $tagline ( @{$root->getChildrenByTagName("datafield")} ) {
279 next unless $tagline;
281 for my $data ( @{$tagline->getChildrenByTagName("subfield")} ) {
282 next unless $tagline;
284 my $ns = new Fieldmapper::metabib::full_rec;
286 $ns->tag( $tagline->getAttribute( "tag" ) );
287 $ns->ind1( $tagline->getAttribute( "ind1" ) );
288 $ns->ind2( $tagline->getAttribute( "ind2" ) );
289 $ns->subfield( $data->getAttribute( "code" ) );
290 my $val = $data->textContent;
291 $val =~ s/(\pM)//gso;
292 $ns->value( lc($val) );
300 sub _get_field_value {
302 my( $mods, $xpath ) = @_;
305 my $root = $mods->documentElement;
306 $root->setNamespace( "http://www.loc.gov/mods/", "mods", 1 );
308 # grab the set of matching nodes
309 my @nodes = $root->findnodes( $xpath );
310 for my $value (@nodes) {
312 # grab all children of the node
313 my @children = $value->childNodes();
314 for my $child (@children) {
316 # add the childs content to the growing buffer
317 my $content = quotemeta($child->textContent);
318 next if ($string =~ /$content/); # uniquify the values
319 $string .= $child->textContent . " ";
322 $string .= $value->textContent . " ";
325 $string =~ s/(\pM)//gso;
330 sub modsdoc_to_values {
331 my( $self, $mods ) = @_;
333 for my $class (keys %$xpathset) {
334 $data->{$class} = {};
335 for my $type(keys %{$xpathset->{$class}}) {
336 my $value = _get_field_value( $mods, $xpathset->{$class}->{$type} );
337 $data->{$class}->{$type} = $value;