Open-ILS/src/perlmods/OpenILS/Utils/ModsParser.pm

   1 package OpenILS::Utils::ModsParser;
   2 use strict; use warnings;
   3
   4 use OpenSRF::EX qw/:try/;
   5 use XML::LibXML;
   6 use XML::LibXSLT;
   7 use Time::HiRes qw(time);
   8 use OpenILS::Utils::Fieldmapper;
   9 use Data::Dumper;
  10
  11 my $parser              = XML::LibXML->new();
  12 my $xslt                        = XML::LibXSLT->new();
  13 my $xslt_doc    = $parser->parse_file(
  14                 "/pines/cvs/ILS/Open-ILS/xsl/MARC21slim2MODS3.xsl" );
  15 my $mods_sheet = $xslt->parse_stylesheet( $xslt_doc );
  16
  17 # ----------------------------------------------------------------------------------------
  18 # XPATH for extracting info from a MODS doc
  19 my $isbn_xpath                  = "//mods:mods/mods:identifier[\@type='isbn']";
  20 my $resource_xpath      = "//mods:mods/mods:typeOfResource";
  21 my $pub_xpath                   = "//mods:mods/mods:originInfo//mods:dateIssued[\@encoding='marc']|" .
  22                                                                 "//mods:mods/mods:originInfo//mods:dateIssued[1]";
  23 my $tcn_xpath                   = "//mods:mods/mods:recordInfo/mods:recordIdentifier";
  24 my $publisher_xpath     = "//mods:mods/mods:originInfo//mods:publisher[1]";
  25 my $edition_xpath               = "//mods:mods/mods:originInfo//mods:edition[1]";
  26 my $abstract_xpath      = "//mods:mods/mods:abstract";
  27 my $toc_xpath                   = "";
  28 my $related_xpath               = "";
  29 my $online_loc_xpath = "(//mods:location/mods:url|//mods:location/mods:url/\@displayLabel)";
  30
  31 my $xpathset = {
  32
  33         title => {
  34                 abbreviated =>
  35                         "//mods:mods/mods:titleInfo[mods:title and (\@type='abreviated')]",
  36                 translated =>
  37                         "//mods:mods/mods:titleInfo[mods:title and (\@type='translated')]",
  38                 uniform =>
  39                         "//mods:mods/mods:titleInfo[mods:title and (\@type='uniform')]",
  40                 proper =>
  41                         "//mods:mods/mods:titleInfo[mods:title and not (\@type)]",
  42         },
  43
  44         author => {
  45                 corporate =>
  46                         "//mods:mods/mods:name[\@type='corporate']/*[local-name()='namePart']".
  47                                 "[../mods:role/mods:text[text()='creator']][1]",
  48                 personal =>
  49                         "//mods:mods/mods:name[\@type='personal']/*[local-name()='namePart']".
  50                                 "[../mods:role/mods:text[text()='creator']][1]",
  51                 conference =>
  52                         "//mods:mods/mods:name[\@type='conference']/*[local-name()='namePart']".
  53                                 "[../mods:role/mods:text[text()='creator']][1]",
  54                 other =>
  55                         "//mods:mods/mods:name[\@type='personal']/*[local-name()='namePart']",
  56         },
  57
  58         subject => {
  59
  60                 topic =>
  61                         "//mods:mods/mods:subject/*[local-name()!='geographicCode']/parent::mods:subject",
  62
  63 #               geographic =>
  64 #                       "//mods:mods/*[local-name()='subject']/*[local-name()='geographic']",
  65 #               name =>
  66 #                       "//mods:mods/*[local-name()='subject']/*[local-name()='name']",
  67 #               temporal =>
  68 #                       "//mods:mods/*[local-name()='subject']/*[local-name()='temporal']",
  69 #               topic =>
  70 #                       "//mods:mods/*[local-name()='subject']/*[local-name()='topic']",
  71         },
  72         #keyword => { keyword => "//mods:mods/*[not(local-name()='originInfo')]", },
  73
  74         series => {
  75                 series => "//mods:mods/mods:relatedItem[\@type='series']/mods:titleInfo"
  76         }
  77 };
  78 # ----------------------------------------------------------------------------------------
  79
  80
  81
  82 sub new { return bless( {}, shift() ); }
  83
  84 sub get_field_value {
  85
  86         my( $self, $mods, $xpath ) = @_;
  87
  88         my @string;
  89         my $root = $mods->documentElement;
  90         $root->setNamespace( "http://www.loc.gov/mods/v3", "mods", 1 );
  91
  92         # grab the set of matching nodes
  93         my @nodes = $root->findnodes( $xpath );
  94         for my $value (@nodes) {
  95
  96                 # grab all children of the node
  97                 my @children = $value->childNodes();
  98                 my @child_text;
  99                 for my $child (@children) {
 100                         next unless( $child->nodeType != 3 );
 101
 102                         if($child->childNodes) {
 103                                 my @a;
 104                                 for my $c (@{$child->childNodes}){
 105                                         push @a, $c->textContent;
 106                                 }
 107                                 push(@child_text, join(' ', @a));
 108
 109                         } else {
 110                                 push(@child_text, $child->textContent);
 111                         }
 112
 113                 }
 114                 if(@child_text) {
 115                         push(@string, \@child_text);
 116                 }
 117
 118                 if( !@child_text  ) {
 119                         push(@string, $value->textContent );
 120                 }
 121         }
 122         return @string;
 123 }
 124
 125 =head
 126 sub _modsdoc_to_values {
 127         my( $self, $mods ) = @_;
 128         my $data = {};
 129         for my $class (keys %$xpathset) {
 130                 $data->{$class} = {};
 131                 for my $type(keys %{$xpathset->{$class}}) {
 132                         my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
 133                         if( $class eq "subject" ) {
 134                                 push( @{$data->{$class}->{$type}},  @value );
 135                         } else {
 136                                 $data->{$class}->{$type} = $value[0];
 137                         }
 138                 }
 139         }
 140         return $data;
 141 }
 142 =cut
 143
 144 sub modsdoc_to_values {
 145         my( $self, $mods ) = @_;
 146         my $data = {};
 147
 148         {
 149                 my $class = "subject";
 150                 $data->{$class} = {};
 151                 for my $type(keys %{$xpathset->{$class}}) {
 152                         my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
 153                         for my $arr (@value) {
 154                                 push( @{$data->{$class}->{$type}},  $arr);
 155                         }
 156                 }
 157         }
 158
 159         {
 160                 my $class = "title";
 161                 $data->{$class} = {};
 162                 for my $type(keys %{$xpathset->{$class}}) {
 163                         my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
 164                         for my $arr (@value) {
 165                                 if( ref($arr) ) {
 166                                         $data->{$class}->{$type} = join(" ", @$arr);
 167                                 } else {
 168                                         $data->{$class}->{$type} = $arr;
 169                                 }
 170                         }
 171                 }
 172         }
 173
 174         {
 175                 my $class = "author";
 176                 $data->{$class} = {};
 177                 for my $type(keys %{$xpathset->{$class}}) {
 178                         my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
 179                         $data->{$class}->{$type} = $value[0];
 180                 }
 181         }
 182
 183         {
 184                 my $class = "series";
 185                 $data->{$class} = {};
 186                 for my $type(keys %{$xpathset->{$class}}) {
 187                         my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
 188                         for my $arr (@value) {
 189                                 if( ref($arr) ) {
 190                                         push(@{$data->{$class}->{$type}}, join(" ", @$arr));
 191                                 } else {
 192                                         push( @{$data->{$class}->{$type}}, $arr );
 193                                 }
 194                         }
 195                 }
 196
 197         }
 198
 199         return $data;
 200 }
 201
 202
 203
 204
 205 # ---------------------------------------------------------------------------
 206 # Grabs the data 'we want' from the MODS doc and returns it in hash form
 207 # ---------------------------------------------------------------------------
 208 sub mods_values_to_mods_slim {
 209         my( $self, $modsperl ) = @_;
 210
 211         my $title = "";
 212         my $author = "";
 213         my $subject = [];
 214         my $series      = [];
 215
 216         my $tmp = $modsperl->{title};
 217
 218
 219         if(!$tmp) { $title = ""; }
 220         else {
 221                 ($title = $tmp->{proper}) ||
 222                 ($title = $tmp->{translated}) ||
 223                 ($title = $tmp->{abbreviated}) ||
 224                 ($title = $tmp->{uniform});
 225         }
 226
 227         $tmp = $modsperl->{author};
 228         if(!$tmp) { $author = ""; }
 229         else {
 230                 ($author = $tmp->{personal}) ||
 231                 ($author = $tmp->{other}) ||
 232                 ($author = $tmp->{corporate}) ||
 233                 ($author = $tmp->{conference});
 234         }
 235
 236         $tmp = $modsperl->{subject};
 237         if(!$tmp) { $subject = []; }
 238         else {
 239                 for my $key( keys %{$tmp}) {
 240                         push(@$subject, @{$tmp->{$key}}) if ($tmp->{$key});
 241                 }
 242         }
 243
 244         $tmp = $modsperl->{'series'};
 245         if(!$tmp) { $series = []; }
 246         else { $series = $tmp->{'series'}; }
 247
 248
 249         return { series => $series, title => $title,
 250                         author => $author, subject => $subject };
 251 }
 252
 253
 254
 255 # ---------------------------------------------------------------------------
 256 # Initializes a MARC -> Unified MODS batch process
 257 # ---------------------------------------------------------------------------
 258
 259 sub start_mods_batch {
 260
 261         my( $self, $master_doc ) = @_;
 262
 263         my $xmldoc = $parser->parse_string($master_doc);
 264         my $mods = $mods_sheet->transform($xmldoc);
 265
 266         warn "-" x 100 . "\n";
 267         warn "MODS " . $mods->toString(1) . "\n";
 268         warn "-" x 100 . "\n";
 269
 270         $self->{master_doc} = $self->modsdoc_to_values( $mods );
 271         $self->{master_doc} = $self->mods_values_to_mods_slim( $self->{master_doc} );
 272
 273         ($self->{master_doc}->{isbn}) =
 274                 $self->get_field_value( $mods, $isbn_xpath );
 275
 276         $self->{master_doc}->{type_of_resource} =
 277                 [ $self->get_field_value( $mods, $resource_xpath ) ];
 278
 279         ($self->{master_doc}->{tcn}) =
 280                 $self->get_field_value( $mods, $tcn_xpath );
 281
 282         ($self->{master_doc}->{pubdate}) =
 283                 $self->get_field_value( $mods, $pub_xpath );
 284
 285         ($self->{master_doc}->{publisher}) =
 286                 $self->get_field_value( $mods, $publisher_xpath );
 287
 288         ($self->{master_doc}->{edition}) =
 289                 $self->get_field_value( $mods, $edition_xpath );
 290
 291
 292
 293 # ------------------------------
 294         # holds an array of [ link, title, link, title, ... ]
 295         $self->{master_doc}->{online_loc} = [];
 296         push(@{$self->{master_doc}->{online_loc}},
 297                 $self->get_field_value( $mods, $online_loc_xpath ));
 298
 299         ($self->{master_doc}->{synopsis}) =
 300                 $self->get_field_value( $mods, $abstract_xpath );
 301
 302 }
 303
 304
 305
 306 # ---------------------------------------------------------------------------
 307 # Takes a MARCXML string and adds it to the growing MODS doc
 308 # ---------------------------------------------------------------------------
 309 sub push_mods_batch {
 310         my( $self, $marcxml ) = @_;
 311
 312         my $xmldoc = $parser->parse_string($marcxml);
 313         my $mods = $mods_sheet->transform($xmldoc);
 314
 315         my $xmlperl = $self->modsdoc_to_values( $mods );
 316         $xmlperl = $self->mods_values_to_mods_slim( $xmlperl );
 317
 318         for my $subject( @{$xmlperl->{subject}} ) {
 319                 push @{$self->{master_doc}->{subject}}, $subject;
 320         }
 321
 322         push( @{$self->{master_doc}->{type_of_resource}},
 323                 $self->get_field_value( $mods, $resource_xpath ));
 324
 325         if(!($self->{master_doc}->{isbn}) ) {
 326                 ($self->{master_doc}->{isbn}) =
 327                         $self->get_field_value( $mods, $isbn_xpath );
 328         }
 329 }
 330
 331
 332 # ---------------------------------------------------------------------------
 333 # Completes a MARC -> Unified MODS batch process and returns the perl hash
 334 # ---------------------------------------------------------------------------
 335 sub init_virtual_record {
 336         my $record = new Fieldmapper::metabib::virtual_record;
 337         $record->subject([]);
 338         $record->types_of_resource([]);
 339         $record->call_numbers([]);
 340         return $record;
 341 }
 342
 343 sub finish_mods_batch {
 344         my $self = shift;
 345         my $perl = $self->{master_doc};
 346         my $record = init_virtual_record();
 347
 348         # turn the hash into a fieldmapper object
 349         (my $title = $perl->{title}) =~ s/\[.*?\]//og;
 350         (my $author = $perl->{author}) =~ s/\(.*?\)//og;
 351
 352         my @series;
 353         for my $s (@{$perl->{series}}) {
 354                 push @series, (split( /\s*;/, $s ))[0];
 355         }
 356
 357         # uniquify the types of resource
 358         my $rtypes = $perl->{type_of_resource};
 359         my %hash = map { ($_ => 1) } @$rtypes;
 360         $rtypes = [ keys %hash ];
 361
 362         $record->title($title);
 363         $record->author($author);
 364
 365         $record->doc_id($perl->{doc_id});
 366         $record->isbn($perl->{isbn});
 367         $record->pubdate($perl->{pubdate});
 368         $record->publisher($perl->{publisher});
 369         $record->tcn($perl->{tcn});
 370
 371         $record->edition($perl->{edition});
 372
 373         $record->subject($perl->{subject});
 374         $record->types_of_resource($rtypes);
 375         $record->series(\@series);
 376
 377         $record->online_loc($perl->{online_loc});
 378         $record->synopsis($perl->{synopsis});
 379
 380         $self->{master_doc} = undef;
 381         return $record;
 382 }
 383
 384
 385