Open-ILS/src/perlmods/OpenILS/Application/Storage/WORM.pm

   1 package OpenILS::Application::Storage::WORM;
   2 use base qw/OpenILS::Application::Storage/;
   3 use strict; use warnings;
   4
   5 use OpenSRF::EX qw/:try/;
   6
   7 use OpenSRF::Utils::Logger qw/:level/;
   8 my $log = 'OpenSRF::Utils::Logger';
   9
  10 use OpenILS::Utils::FlatXML;
  11 use OpenILS::Utils::Fieldmapper;
  12 use JSON;
  13
  14 use XML::LibXML;
  15 use XML::LibXSLT;
  16 use Time::HiRes qw(time);
  17
  18 my $xml_util    = OpenILS::Utils::FlatXML->new();
  19
  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 );
  25
  26 use open qw/:utf8/;
  27
  28 my $xpathset = {};
  29
  30 sub child_init {
  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);
  36         }
  37 }
  38
  39 # --------------------------------------------------------------------------------
  40 # Fingerprinting
  41
  42 my @fp_mods_xpath = (
  43         '//mods:mods/mods:typeOfResource[text()="text"]' => [
  44                         title   => {
  45                                         xpath   => [
  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)]',
  50                                         ],
  51                                         fixup   => '
  52                                                         do {
  53                                                                 $text = lc($text);
  54                                                                 $text =~ s/\s+/ /sgo;
  55                                                                 $text =~ s/^\s*(.+)\s*$/$1/sgo;
  56                                                                 $text =~ s/\b(?:the|an?)\b//sgo;
  57                                                                 $text =~ s/\[.[^\]]+\]//sgo;
  58                                                                 $text =~ s/\s*[;\/\.]*$//sgo;
  59                                                                 $text =~ s/\pM+//gso;
  60                                                         };
  61                                         ',
  62                         },
  63                         author  => {
  64                                         xpath   => [
  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',
  67                                         ],
  68                                         fixup   => '
  69                                                         do {
  70                                                                 $text = lc($text);
  71                                                                 $text =~ s/\s+/ /sgo;
  72                                                                 $text =~ s/^\s*(.+)\s*$/$1/sgo;
  73                                                                 $text =~ s/,?\s+.*$//sgo;
  74                                                                 $text =~ s/\pM+//gso;
  75                                                         };
  76                                         ',
  77                         },
  78         ],
  79
  80         '//mods:mods/mods:relatedItem[@type!="host"]' => [
  81                         title   => {
  82                                         xpath   => [
  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)]',
  91                                         ],
  92                                         fixup   => '
  93                                                         do {
  94                                                                 $text = lc($text);
  95                                                                 $text =~ s/\s+/ /sgo;
  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;
 101                                                         };
 102                                         ',
 103                         },
 104                         author  => {
 105                                         xpath   => [
 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',
 110                                         ],
 111                                         fixup   => '
 112                                                         do {
 113                                                                 $text = lc($text);
 114                                                                 $text =~ s/\s+/ /sgo;
 115                                                                 $text =~ s/^\s*(.+)\s*$/$1/sgo;
 116                                                                 $text =~ s/,?\s+.*$//sgo;
 117                                                                 $text =~ s/\pM+//gso;
 118                                                         };
 119                                         ',
 120                         },
 121         ],
 122
 123 );
 124
 125 push @fp_mods_xpath, '//mods:mods/mods:titleInfo' => $fp_mods_xpath[1];
 126
 127 sub fingerprint_mods {
 128         my $mods = shift;
 129
 130         my $fp_string = '';
 131
 132         my $match_index = 0;
 133         my $block_index = 1;
 134         while ( my $match_xpath = $fp_mods_xpath[$match_index] ) {
 135                 if ( my @nodes = $mods->findnodes( $match_xpath ) ) {
 136
 137                         my $block_name_index = 0;
 138                         my $block_value_index = 1;
 139                         my $block = $fp_mods_xpath[$block_index];
 140                         while ( my $part = $$block[$block_value_index] ) {
 141                                 my $text;
 142                                 for my $xpath ( @{ $part->{xpath} } ) {
 143                                         $text = $mods->findvalue( $xpath );
 144                                         last if ($text);
 145                                 }
 146
 147                                 $log->debug("Found fingerprint text using $$block[$block_name_index] : [$text]", DEBUG);
 148
 149                                 if ($text) {
 150                                         eval $$part{fixup};
 151                                         $fp_string .= $text;
 152                                 }
 153
 154                                 $block_name_index += 2;
 155                                 $block_value_index += 2;
 156                         }
 157                 }
 158                 if ($fp_string) {
 159                         $fp_string =~ s/\W+//gso;
 160                         $log->debug("Fingerprint is [$fp_string]", INFO);;
 161                         return $fp_string;
 162                 }
 163
 164                 $match_index += 2;
 165                 $block_index += 2;
 166         }
 167         return undef;
 168 }
 169
 170
 171
 172 # --------------------------------------------------------------------------------
 173
 174 my $begin;
 175 my $commit;
 176 my $rollback;
 177 my $lookup;
 178 my $update_entry;
 179 my $mr_lookup;
 180 my $mr_create;
 181 my $create_source_map;
 182 my $sm_lookup;
 183 my $rm_old_rd;
 184 my $rm_old_sm;
 185 my $rm_old_fr;
 186 my $rm_old_tr;
 187 my $rm_old_ar;
 188 my $rm_old_sr;
 189 my $rm_old_kr;
 190
 191 my $fr_create;
 192 my $rd_create;
 193 my $create = {};
 194
 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)',
 206 );
 207
 208 sub wormize {
 209
 210         my $self = shift;
 211         my $client = shift;
 212         my @docids = @_;
 213
 214         my $no_map = 0;
 215         if ($self->api_name =~ /no_map/o) {
 216                 $no_map = 1;
 217         }
 218
 219         $begin = $self->method_lookup( 'open-ils.storage.transaction.begin')
 220                 unless ($begin);
 221         $commit = $self->method_lookup( 'open-ils.storage.transaction.commit')
 222                 unless ($commit);
 223         $rollback = $self->method_lookup( 'open-ils.storage.transaction.rollback')
 224                 unless ($rollback);
 225         $sm_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord_source_map.search.source')
 226                 unless ($sm_lookup);
 227         $mr_lookup = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.search.fingerprint')
 228                 unless ($mr_lookup);
 229         $mr_create = $self->method_lookup('open-ils.storage.direct.metabib.metarecord.create')
 230                 unless ($mr_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')
 234                 unless ($lookup);
 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')
 238                 unless ($rm_old_sm);
 239         $rm_old_rd = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.mass_delete')
 240                 unless ($rm_old_rd);
 241         $rm_old_fr = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.mass_delete')
 242                 unless ($rm_old_fr);
 243         $rm_old_tr = $self->method_lookup( 'open-ils.storage.direct.metabib.title_field_entry.mass_delete')
 244                 unless ($rm_old_tr);
 245         $rm_old_ar = $self->method_lookup( 'open-ils.storage.direct.metabib.author_field_entry.mass_delete')
 246                 unless ($rm_old_ar);
 247         $rm_old_sr = $self->method_lookup( 'open-ils.storage.direct.metabib.subject_field_entry.mass_delete')
 248                 unless ($rm_old_sr);
 249         $rm_old_kr = $self->method_lookup( 'open-ils.storage.direct.metabib.keyword_field_entry.mass_delete')
 250                 unless ($rm_old_kr);
 251         $rd_create = $self->method_lookup( 'open-ils.storage.direct.metabib.record_descriptor.batch.create')
 252                 unless ($rd_create);
 253         $fr_create = $self->method_lookup( 'open-ils.storage.direct.metabib.full_rec.batch.create')
 254                 unless ($fr_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});
 263
 264
 265         try {
 266                 my ($r) = $begin->run($client);
 267                 unless (defined $r and $r) {
 268                         $rollback->run;
 269                         throw OpenSRF::EX::PANIC ("Couldn't BEGIN transaction!")
 270                 }
 271         } catch Error with {
 272                 throw OpenSRF::EX::PANIC ("WoRM Couldn't BEGIN transaction!")
 273         };
 274
 275         my @source_maps;
 276         my @entry_list;
 277         my @rd_list;
 278         my @ns_list;
 279         my @mods_data;
 280         my $ret = 0;
 281         for my $entry ( $lookup->run(@docids) ) {
 282                 # step -1: grab the doc from storage
 283                 next unless ($entry);
 284
 285                 my $xml = $entry->marc;
 286                 my $docid = $entry->id;
 287                 my $marcdoc = $parser->parse_string($xml);
 288                 my $modsdoc = $mods_sheet->transform($marcdoc);
 289
 290                 my $mods = $modsdoc->documentElement;
 291                 $mods->setNamespace( "http://www.loc.gov/mods/", "mods", 1 );
 292
 293                 $entry->fingerprint( fingerprint_mods( $mods ) );
 294                 push @entry_list, $entry;
 295
 296                 unless ($no_map) {
 297                         my ($mr) = $mr_lookup->run( $entry->fingerprint );
 298                         if (!@$mr) {
 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);
 303                                 $mr->id($new_mr);
 304                                 unless (defined $mr) {
 305                                         throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord.create!")
 306                                 }
 307                         } else {
 308                                 $mr = $$mr[0];
 309                         }
 310
 311                         my $sm = new Fieldmapper::metabib::metarecord_source_map;
 312                         $sm->metarecord( $mr->id );
 313                         $sm->source( $entry->id );
 314                         push @source_maps, $sm;
 315                 }
 316
 317                 my $ldr = $marcdoc->documentElement->getChildrenByTagName('leader')->pop->textContent;
 318                 my $oo8 = $marcdoc->documentElement->findvalue('//*[local-name()="controlfield" and @tag="008"]');
 319
 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};" );
 323                 }
 324                 $rd_obj->record( $docid );
 325                 push @rd_list, $rd_obj;
 326
 327                 push @mods_data, { $docid => $self->modsdoc_to_values( $mods ) };
 328
 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;
 333
 334                 $ret++;
 335
 336                 last unless ($self->api_name =~ /batch$/o);
 337         }
 338
 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         $rm_old_kr->run( { source => \@docids } );
 346
 347         unless ($no_map) {
 348                 my ($sm) = $create_source_map->run(@source_maps);
 349                 unless (defined $sm) {
 350                         throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.metarecord_source_map.batch.create!")
 351                 }
 352         }
 353
 354         my ($re) = $update_entry->run(@entry_list);
 355         unless (defined $re) {
 356                 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.biblio.record_entry.batch.update!")
 357         }
 358
 359         my ($rd) = $rd_create->run(@rd_list);
 360         unless (defined $rd) {
 361                 throw OpenSRF::EX::PANIC ("Couldn't run open-ils.storage.direct.metabib.record_descriptor.batch.create!")
 362         }
 363
 364         my ($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!")
 367         }
 368
 369         # step 5: insert the new metadata
 370         for my $class ( qw/title author subject keyword/ ) {
 371                 my @md_list = ();
 372                 for my $doc ( @mods_data ) {
 373                         my ($did) = keys %$doc;
 374                         my ($data) = values %$doc;
 375
 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( $$data{$class}{$row}{field_id} );
 383                                 $fm_obj->source( $did );
 384                                 $log->debug("$class entry: ".$fm_obj->source." => ".$fm_obj->field." : ".$fm_obj->value, DEBUG);
 385
 386                                 push @md_list, $fm_obj;
 387                         }
 388                 }
 389
 390                 my ($cr) = $$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!")
 393                 }
 394         }
 395
 396         my ($c) = $commit->run;
 397         unless (defined $c and $c) {
 398                 $rollback->run;
 399                 throw OpenSRF::EX::PANIC ("Couldn't COMMIT changes!")
 400         }
 401
 402         return $ret;
 403 }
 404 __PACKAGE__->register_method(
 405         api_name        => "open-ils.worm.wormize",
 406         method          => "wormize",
 407         api_level       => 1,
 408         argc            => 1,
 409 );
 410 __PACKAGE__->register_method(
 411         api_name        => "open-ils.worm.wormize.no_map",
 412         method          => "wormize",
 413         api_level       => 1,
 414         argc            => 1,
 415 );
 416 __PACKAGE__->register_method(
 417         api_name        => "open-ils.worm.wormize.batch",
 418         method          => "wormize",
 419         api_level       => 1,
 420         argc            => 1,
 421 );
 422 __PACKAGE__->register_method(
 423         api_name        => "open-ils.worm.wormize.no_map.batch",
 424         method          => "wormize",
 425         api_level       => 1,
 426         argc            => 1,
 427 );
 428
 429
 430 # --------------------------------------------------------------------------------
 431
 432
 433 sub _marcxml_to_full_rows {
 434
 435         my $marcxml = shift;
 436
 437         my @ns_list;
 438
 439         my $root = $marcxml->documentElement;
 440
 441         for my $tagline ( @{$root->getChildrenByTagName("leader")} ) {
 442                 next unless $tagline;
 443
 444                 my $ns = new Fieldmapper::metabib::full_rec;
 445
 446                 $ns->tag( 'LDR' );
 447                 my $val = $tagline->textContent;
 448                 $val =~ s/(\pM)//gso;
 449                 $ns->value( $val );
 450
 451                 push @ns_list, $ns;
 452         }
 453
 454         for my $tagline ( @{$root->getChildrenByTagName("controlfield")} ) {
 455                 next unless $tagline;
 456
 457                 my $ns = new Fieldmapper::metabib::full_rec;
 458
 459                 $ns->tag( $tagline->getAttribute( "tag" ) );
 460                 my $val = $tagline->textContent;
 461                 $val =~ s/(\pM)//gso;
 462                 $ns->value( $val );
 463
 464                 push @ns_list, $ns;
 465         }
 466
 467         for my $tagline ( @{$root->getChildrenByTagName("datafield")} ) {
 468                 next unless $tagline;
 469
 470                 my $tag = $tagline->getAttribute( "tag" );
 471                 my $ind1 = $tagline->getAttribute( "ind1" );
 472                 my $ind2 = $tagline->getAttribute( "ind2" );
 473
 474                 for my $data ( $tagline->childNodes ) {
 475                         next unless $data;
 476
 477                         my $ns = new Fieldmapper::metabib::full_rec;
 478
 479                         $ns->tag( $tag );
 480                         $ns->ind1( $ind1 );
 481                         $ns->ind2( $ind2 );
 482                         $ns->subfield( $data->getAttribute( "code" ) );
 483                         my $val = $data->textContent;
 484                         $val =~ s/(\pM)//gso;
 485                         $ns->value( lc($val) );
 486
 487                         push @ns_list, $ns;
 488                 }
 489         }
 490         return @ns_list;
 491 }
 492
 493 sub _get_field_value {
 494
 495         my( $root, $xpath ) = @_;
 496
 497         my $string = "";
 498
 499         # grab the set of matching nodes
 500         my @nodes = $root->findnodes( $xpath );
 501         for my $value (@nodes) {
 502
 503                 # grab all children of the node
 504                 my @children = $value->childNodes();
 505                 for my $child (@children) {
 506
 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 . " ";
 511                 }
 512                 if( ! @children ) {
 513                         $string .= $value->textContent . " ";
 514                 }
 515         }
 516         $string =~ s/(\pM)//gso;
 517         return lc($string);
 518 }
 519
 520
 521 sub modsdoc_to_values {
 522         my( $self, $mods ) = @_;
 523         my $data = {};
 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};
 530                 }
 531         }
 532         return $data;
 533 }
 534
 535
 536 1;
 537
 538