C4/Record.pm

   1 package C4::Record;
   2 #
   3 # Copyright 2006 (C) LibLime
   4 # Joshua Ferraro <jmf@liblime.com>
   5 #
   6 # This file is part of Koha.
   7 #
   8 # Koha is free software; you can redistribute it and/or modify it under the
   9 # terms of the GNU General Public License as published by the Free Software
  10 # Foundation; either version 2 of the License, or (at your option) any later
  11 # version.
  12 #
  13 # Koha is distributed in the hope that it will be useful, but WITHOUT ANY
  14 # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  15 # A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
  16 #
  17 # You should have received a copy of the GNU General Public License along with
  18 # Koha; if not, write to the Free Software Foundation, Inc., 59 Temple Place,
  19 # Suite 330, Boston, MA  02111-1307 USA
  20 #
  21 #
  22 use strict;# use warnings; #FIXME: turn off warnings before release
  23
  24 # please specify in which methods a given module is used
  25 use MARC::Record; # marc2marcxml, marcxml2marc, html2marc, changeEncoding
  26 use MARC::File::XML; # marc2marcxml, marcxml2marc, html2marcxml, changeEncoding
  27 use MARC::Crosswalk::DublinCore; # marc2dcxml
  28 use Biblio::EndnoteStyle;
  29 use Unicode::Normalize; # _entity_encode
  30 use XML::LibXSLT;
  31 use XML::LibXML;
  32
  33 use vars qw($VERSION @ISA @EXPORT);
  34
  35 # set the version for version checking
  36 $VERSION = 3.00;
  37
  38 @ISA = qw(Exporter);
  39
  40 # only export API methods
  41
  42 @EXPORT = qw(
  43   &marc2endnote
  44   &marc2marc
  45   &marc2marcxml
  46   &marcxml2marc
  47   &marc2dcxml
  48   &marc2modsxml
  49
  50   &html2marcxml
  51   &html2marc
  52   &changeEncoding
  53 );
  54
  55 =head1 NAME
  56
  57 C4::Record - MARC, MARCXML, DC, MODS, XML, etc. Record Management Functions and API
  58
  59 =head1 SYNOPSIS
  60
  61 New in Koha 3.x. This module handles all record-related management functions.
  62
  63 =head1 API (EXPORTED FUNCTIONS)
  64
  65 =head2 marc2marc - Convert from one flavour of ISO-2709 to another
  66
  67 =over 4
  68
  69 my ($error,$newmarc) = marc2marc($marc,$to_flavour,$from_flavour,$encoding);
  70
  71 Returns an ISO-2709 scalar
  72
  73 =back
  74
  75 =cut
  76
  77 sub marc2marc {
  78         my ($marc,$to_flavour,$from_flavour,$encoding) = @_;
  79         my $error = "Feature not yet implemented\n";
  80         return ($error,$marc);
  81 }
  82
  83 =head2 marc2marcxml - Convert from ISO-2709 to MARCXML
  84
  85 =over 4
  86
  87 my ($error,$marcxml) = marc2marcxml($marc,$encoding,$flavour);
  88
  89 Returns a MARCXML scalar
  90
  91 =over 2
  92
  93 C<$marc> - an ISO-2709 scalar or MARC::Record object
  94
  95 C<$encoding> - UTF-8 or MARC-8 [UTF-8]
  96
  97 C<$flavour> - MARC21 or UNIMARC
  98
  99 C<$dont_entity_encode> - a flag that instructs marc2marcxml not to entity encode the xml before returning (optional)
 100
 101 =back
 102
 103 =back
 104
 105 =cut
 106
 107 sub marc2marcxml {
 108         my ($marc,$encoding,$flavour,$dont_entity_encode) = @_;
 109         my $error; # the error string
 110         my $marcxml; # the final MARCXML scalar
 111
 112         # test if it's already a MARC::Record object, if not, make it one
 113         my $marc_record_obj;
 114         if ($marc =~ /^MARC::Record/) { # it's already a MARC::Record object
 115                 $marc_record_obj = $marc;
 116         } else { # it's not a MARC::Record object, make it one
 117                 eval { $marc_record_obj = MARC::Record->new_from_usmarc($marc) }; # handle exceptions
 118
 119                 # conversion to MARC::Record object failed, populate $error
 120                 if ($@) { $error .="\nCreation of MARC::Record object failed: ".$MARC::File::ERROR };
 121         }
 122         # only proceed if no errors so far
 123         unless ($error) {
 124
 125                 # check the record for warnings
 126                 my @warnings = $marc_record_obj->warnings();
 127                 if (@warnings) {
 128                         warn "\nWarnings encountered while processing ISO-2709 record with title \"".$marc_record_obj->title()."\":\n";
 129                         foreach my $warn (@warnings) { warn "\t".$warn };
 130                 }
 131                 unless($encoding) {$encoding = "UTF-8"}; # set default encoding
 132                 unless($flavour) {$flavour = C4::Context->preference("marcflavour")}; # set default MARC flavour
 133
 134                 # attempt to convert the record to MARCXML
 135                 eval { $marcxml = $marc_record_obj->as_xml_record($flavour) }; #handle exceptions
 136
 137                 # record creation failed, populate $error
 138                 if ($@) {
 139                         $error .= "Creation of MARCXML failed:".$MARC::File::ERROR;
 140                         $error .= "Additional information:\n";
 141                         my @warnings = $@->warnings();
 142                         foreach my $warn (@warnings) { $error.=$warn."\n" };
 143
 144                 # record creation was successful
 145         } else {
 146
 147                         # check the record for warning flags again (warnings() will be cleared already if there was an error, see above block
 148                         @warnings = $marc_record_obj->warnings();
 149                         if (@warnings) {
 150                                 warn "\nWarnings encountered while processing ISO-2709 record with title \"".$marc_record_obj->title()."\":\n";
 151                                 foreach my $warn (@warnings) { warn "\t".$warn };
 152                         }
 153                 }
 154
 155                 # only proceed if no errors so far
 156                 unless ($error) {
 157
 158                         # entity encode the XML unless instructed not to
 159                 unless ($dont_entity_encode) {
 160                         my ($marcxml_entity_encoded) = _entity_encode($marcxml);
 161                         $marcxml = $marcxml_entity_encoded;
 162                 }
 163                 }
 164         }
 165         # return result to calling program
 166         return ($error,$marcxml);
 167 }
 168
 169 =head2 marcxml2marc - Convert from MARCXML to ISO-2709
 170
 171 =over 4
 172
 173 my ($error,$marc) = marcxml2marc($marcxml,$encoding,$flavour);
 174
 175 Returns an ISO-2709 scalar
 176
 177 =over 2
 178
 179 C<$marcxml> - a MARCXML record
 180
 181 C<$encoding> - UTF-8 or MARC-8 [UTF-8]
 182
 183 C<$flavour> - MARC21 or UNIMARC
 184
 185 =back
 186
 187 =back
 188
 189 =cut
 190
 191 sub marcxml2marc {
 192     my ($marcxml,$encoding,$flavour) = @_;
 193         my $error; # the error string
 194         my $marc; # the final ISO-2709 scalar
 195         unless($encoding) {$encoding = "UTF-8"}; # set the default encoding
 196         unless($flavour) {$flavour = C4::Context->preference("marcflavour")}; # set the default MARC flavour
 197
 198         # attempt to do the conversion
 199         eval { $marc = MARC::Record->new_from_xml($marcxml,$encoding,$flavour) }; # handle exceptions
 200
 201         # record creation failed, populate $error
 202         if ($@) {$error .="\nCreation of MARCXML Record failed: ".$@;
 203                 $error.=$MARC::File::ERROR if ($MARC::File::ERROR);
 204                 };
 205         # return result to calling program
 206         return ($error,$marc);
 207 }
 208
 209 =head2 marc2dcxml - Convert from ISO-2709 to Dublin Core
 210
 211 =over 4
 212
 213 my ($error,$dcxml) = marc2dcxml($marc,$qualified);
 214
 215 Returns a DublinCore::Record object, will eventually return a Dublin Core scalar
 216
 217 FIXME: should return actual XML, not just an object
 218
 219 =over 2
 220
 221 C<$marc> - an ISO-2709 scalar or MARC::Record object
 222
 223 C<$qualified> - specify whether qualified Dublin Core should be used in the input or output [0]
 224
 225 =back
 226
 227 =back
 228
 229 =cut
 230
 231 sub marc2dcxml {
 232         my ($marc,$qualified) = @_;
 233         my $error;
 234     # test if it's already a MARC::Record object, if not, make it one
 235     my $marc_record_obj;
 236     if ($marc =~ /^MARC::Record/) { # it's already a MARC::Record object
 237         $marc_record_obj = $marc;
 238     } else { # it's not a MARC::Record object, make it one
 239                 eval { $marc_record_obj = MARC::Record->new_from_usmarc($marc) }; # handle exceptions
 240
 241                 # conversion to MARC::Record object failed, populate $error
 242                 if ($@) {
 243                         $error .="\nCreation of MARC::Record object failed: ".$MARC::File::ERROR;
 244                 }
 245         }
 246         my $crosswalk = MARC::Crosswalk::DublinCore->new;
 247         if ($qualified) {
 248                 $crosswalk = MARC::Crosswalk::DublinCore->new( qualified => 1 );
 249         }
 250         my $dcxml = $crosswalk->as_dublincore($marc_record_obj);
 251         my $dcxmlfinal = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
 252         $dcxmlfinal .= "<metadata
 253   xmlns=\"http://example.org/myapp/\"
 254   xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"
 255   xsi:schemaLocation=\"http://example.org/myapp/ http://example.org/myapp/schema.xsd\"
 256   xmlns:dc=\"http://purl.org/dc/elements/1.1/\"
 257   xmlns:dcterms=\"http://purl.org/dc/terms/\">";
 258
 259         foreach my $element ( $dcxml->elements() ) {
 260                 $dcxmlfinal.="<"."dc:".$element->name().">".$element->content()."</"."dc:".$element->name().">\n";
 261     }
 262         $dcxmlfinal .= "\n</metadata>";
 263         return ($error,$dcxmlfinal);
 264 }
 265 =head2 marc2modsxml - Convert from ISO-2709 to MODS
 266
 267 =over 4
 268
 269 my ($error,$modsxml) = marc2modsxml($marc);
 270
 271 Returns a MODS scalar
 272
 273 =back
 274
 275 =cut
 276
 277 sub marc2modsxml {
 278         my ($marc) = @_;
 279         # grab the XML, run it through our stylesheet, push it out to the browser
 280         my $xmlrecord = marc2marcxml($marc);
 281         my $xslfile = C4::Context->config('intrahtdocs')."/prog/en/xslt/MARC21slim2MODS3-1.xsl";
 282         my $parser = XML::LibXML->new();
 283         my $xslt = XML::LibXSLT->new();
 284         my $source = $parser->parse_string($xmlrecord);
 285         my $style_doc = $parser->parse_file($xslfile);
 286         my $stylesheet = $xslt->parse_stylesheet($style_doc);
 287         my $results = $stylesheet->transform($source);
 288         my $newxmlrecord = $stylesheet->output_string($results);
 289         return ($newxmlrecord);
 290 }
 291
 292 sub marc2endnote {
 293     my ($marc) = @_;
 294         my $marc_rec_obj =  MARC::Record->new_from_usmarc($marc);
 295         my $f260 = $marc_rec_obj->field('260');
 296         my $f260a = $f260->subfield('a') if $f260;
 297     my $f710 = $marc_rec_obj->field('710');
 298     my $f710a = $f710->subfield('a') if $f710;
 299         my $f500 = $marc_rec_obj->field('500');
 300         my $abstract = $f500->subfield('a') if $f500;
 301         my $fields = {
 302                 DB => C4::Context->preference("LibraryName"),
 303                 Title => $marc_rec_obj->title(),
 304                 Author => $marc_rec_obj->author(),
 305                 Publisher => $f710a,
 306                 City => $f260a,
 307                 Year => $marc_rec_obj->publication_date,
 308                 Abstract => $abstract,
 309         };
 310         my $endnote;
 311         my $style = new Biblio::EndnoteStyle();
 312         my $template;
 313         $template.= "DB - DB\n" if C4::Context->preference("LibraryName");
 314         $template.="T1 - Title\n" if $marc_rec_obj->title();
 315         $template.="A1 - Author\n" if $marc_rec_obj->author();
 316         $template.="PB - Publisher\n" if  $f710a;
 317         $template.="CY - City\n" if $f260a;
 318         $template.="Y1 - Year\n" if $marc_rec_obj->publication_date;
 319         $template.="AB - Abstract\n" if $abstract;
 320         my ($text, $errmsg) = $style->format($template, $fields);
 321         return ($text);
 322
 323 }
 324
 325
 326 =head2 html2marcxml
 327
 328 =over 4
 329
 330 my ($error,$marcxml) = html2marcxml($tags,$subfields,$values,$indicator,$ind_tag);
 331
 332 Returns a MARCXML scalar
 333
 334 this is used in addbiblio.pl and additem.pl to build the MARCXML record from
 335 the form submission.
 336
 337 FIXME: this could use some better code documentation
 338
 339 =back
 340
 341 =cut
 342
 343 sub html2marcxml {
 344     my ($tags,$subfields,$values,$indicator,$ind_tag) = @_;
 345         my $error;
 346         # add the header info
 347     my $marcxml= MARC::File::XML::header(C4::Context->preference('TemplateEncoding'),C4::Context->preference('marcflavour'));
 348
 349         # some flags used to figure out where in the record we are
 350     my $prevvalue;
 351     my $prevtag=-1;
 352     my $first=1;
 353     my $j = -1;
 354
 355         # handle characters that would cause the parser to choke FIXME: is there a more elegant solution?
 356     for (my $i=0;$i<=@$tags;$i++){
 357                 @$values[$i] =~ s/&/&amp;/g;
 358                 @$values[$i] =~ s/</&lt;/g;
 359                 @$values[$i] =~ s/>/&gt;/g;
 360                 @$values[$i] =~ s/"/&quot;/g;
 361                 @$values[$i] =~ s/'/&apos;/g;
 362
 363                 if ((@$tags[$i] ne $prevtag)){
 364                         $j++ unless (@$tags[$i] eq "");
 365                         #warn "IND:".substr(@$indicator[$j],0,1).substr(@$indicator[$j],1,1)." ".@$tags[$i];
 366                         if (!$first){
 367                                 $marcxml.="</datafield>\n";
 368                                 if ((@$tags[$i] > 10) && (@$values[$i] ne "")){
 369                         my $ind1 = substr(@$indicator[$j],0,1);
 370                                         my $ind2 = substr(@$indicator[$j],1,1);
 371                                         $marcxml.="<datafield tag=\"@$tags[$i]\" ind1=\"$ind1\" ind2=\"$ind2\">\n";
 372                                         $marcxml.="<subfield code=\"@$subfields[$i]\">@$values[$i]</subfield>\n";
 373                                         $first=0;
 374                                 } else {
 375                                         $first=1;
 376                                 }
 377                         } else {
 378                                 if (@$values[$i] ne "") {
 379                                         # handle the leader
 380                                         if (@$tags[$i] eq "000") {
 381                                                 $marcxml.="<leader>@$values[$i]</leader>\n";
 382                                                 $first=1;
 383                                         # rest of the fixed fields
 384                                         } elsif (@$tags[$i] lt '010') { # don't compare numerically 010 == 8
 385                                                 $marcxml.="<controlfield tag=\"@$tags[$i]\">@$values[$i]</controlfield>\n";
 386                                                 $first=1;
 387                                         } else {
 388                                                 my $ind1 = substr(@$indicator[$j],0,1);
 389                                                 my $ind2 = substr(@$indicator[$j],1,1);
 390                                                 $marcxml.="<datafield tag=\"@$tags[$i]\" ind1=\"$ind1\" ind2=\"$ind2\">\n";
 391                                                 $marcxml.="<subfield code=\"@$subfields[$i]\">@$values[$i]</subfield>\n";
 392                                                 $first=0;
 393                                         }
 394                                 }
 395                         }
 396                 } else { # @$tags[$i] eq $prevtag
 397                         if (@$values[$i] eq "") {
 398                         } else {
 399                                 if ($first){
 400                                         my $ind1 = substr(@$indicator[$j],0,1);
 401                                         my $ind2 = substr(@$indicator[$j],1,1);
 402                                         $marcxml.="<datafield tag=\"@$tags[$i]\" ind1=\"$ind1\" ind2=\"$ind2\">\n";
 403                                         $first=0;
 404                                 }
 405                                 $marcxml.="<subfield code=\"@$subfields[$i]\">@$values[$i]</subfield>\n";
 406                         }
 407                 }
 408                 $prevtag = @$tags[$i];
 409         }
 410         $marcxml.= MARC::File::XML::footer();
 411         #warn $marcxml;
 412         return ($error,$marcxml);
 413 }
 414
 415 =head2 html2marc
 416
 417 =over 4
 418
 419 Probably best to avoid using this ... it has some rather striking problems:
 420
 421 =over 2
 422
 423 * saves blank subfields
 424
 425 * subfield order is hardcoded to always start with 'a' for repeatable tags (because it is hardcoded in the addfield routine).
 426
 427 * only possible to specify one set of indicators for each set of tags (ie, one for all the 650s). (because they were stored in a hash with the tag as the key).
 428
 429 * the underlying routines didn't support subfield reordering or subfield repeatability.
 430
 431 =back
 432
 433 I've left it in here because it could be useful if someone took the time to fix it. -- kados
 434
 435 =back
 436
 437 =cut
 438
 439 sub html2marc {
 440     my ($dbh,$rtags,$rsubfields,$rvalues,%indicators) = @_;
 441     my $prevtag = -1;
 442     my $record = MARC::Record->new();
 443 #   my %subfieldlist=();
 444     my $prevvalue; # if tag <10
 445     my $field; # if tag >=10
 446     for (my $i=0; $i< @$rtags; $i++) {
 447         # rebuild MARC::Record
 448 #           warn "0=>".@$rtags[$i].@$rsubfields[$i]." = ".@$rvalues[$i].": ";
 449         if (@$rtags[$i] ne $prevtag) {
 450             if ($prevtag < 10) {
 451                 if ($prevvalue) {
 452                     if (($prevtag ne '000') && ($prevvalue ne "")) {
 453                         $record->add_fields((sprintf "%03s",$prevtag),$prevvalue);
 454                     } elsif ($prevvalue ne ""){
 455                         $record->leader($prevvalue);
 456                     }
 457                 }
 458             } else {
 459                 if (($field) && ($field ne "")) {
 460                     $record->add_fields($field);
 461                 }
 462             }
 463             $indicators{@$rtags[$i]}.='  ';
 464                 # skip blank tags, I hope this works
 465                 if (@$rtags[$i] eq ''){
 466                 $prevtag = @$rtags[$i];
 467                 undef $field;
 468                 next;
 469             }
 470             if (@$rtags[$i] <10) {
 471                 $prevvalue= @$rvalues[$i];
 472                 undef $field;
 473             } else {
 474                 undef $prevvalue;
 475                 if (@$rvalues[$i] eq "") {
 476                 undef $field;
 477                 } else {
 478                 $field = MARC::Field->new( (sprintf "%03s",@$rtags[$i]), substr($indicators{@$rtags[$i]},0,1),substr($indicators{@$rtags[$i]},1,1), @$rsubfields[$i] => @$rvalues[$i]);
 479                 }
 480 #           warn "1=>".@$rtags[$i].@$rsubfields[$i]." = ".@$rvalues[$i].": ".$field->as_formatted;
 481             }
 482             $prevtag = @$rtags[$i];
 483         } else {
 484             if (@$rtags[$i] <10) {
 485                 $prevvalue=@$rvalues[$i];
 486             } else {
 487                 if (length(@$rvalues[$i])>0) {
 488                     $field->add_subfields(@$rsubfields[$i] => @$rvalues[$i]);
 489 #           warn "2=>".@$rtags[$i].@$rsubfields[$i]." = ".@$rvalues[$i].": ".$field->as_formatted;
 490                 }
 491             }
 492             $prevtag= @$rtags[$i];
 493         }
 494     }
 495     #}
 496     # the last has not been included inside the loop... do it now !
 497     #use Data::Dumper;
 498     #warn Dumper($field->{_subfields});
 499     $record->add_fields($field) if (($field) && $field ne "");
 500     #warn "HTML2MARC=".$record->as_formatted;
 501     return $record;
 502 }
 503
 504 =head2 changeEncoding - Change the encoding of a record
 505
 506 =over 4
 507
 508 my ($error, $newrecord) = changeEncoding($record,$format,$flavour,$to_encoding,$from_encoding);
 509
 510 Changes the encoding of a record
 511
 512 =over 2
 513
 514 C<$record> - the record itself can be in ISO-2709, a MARC::Record object, or MARCXML for now (required)
 515
 516 C<$format> - MARC or MARCXML (required)
 517
 518 C<$flavour> - MARC21 or UNIMARC, if MARC21, it will change the leader (optional) [defaults to Koha system preference]
 519
 520 C<$to_encoding> - the encoding you want the record to end up in (optional) [UTF-8]
 521
 522 C<$from_encoding> - the encoding the record is currently in (optional, it will probably be able to tell unless there's a problem with the record)
 523
 524 =back
 525
 526 FIXME: the from_encoding doesn't work yet
 527
 528 FIXME: better handling for UNIMARC, it should allow management of 100 field
 529
 530 FIXME: shouldn't have to convert to and from xml/marc just to change encoding someone needs to re-write MARC::Record's 'encoding' method to actually alter the encoding rather than just changing the leader
 531
 532 =back
 533
 534 =cut
 535
 536 sub changeEncoding {
 537         my ($record,$format,$flavour,$to_encoding,$from_encoding) = @_;
 538         my $newrecord;
 539         my $error;
 540         unless($flavour) {$flavour = C4::Context->preference("marcflavour")};
 541         unless($to_encoding) {$to_encoding = "UTF-8"};
 542
 543         # ISO-2709 Record (MARC21 or UNIMARC)
 544         if (lc($format) =~ /^marc$/o) {
 545                 # if we're converting encoding of an ISO2709 file, we need to roundtrip through XML
 546                 #       because MARC::Record doesn't directly provide us with an encoding method
 547                 #       It's definitely less than idea and should be fixed eventually - kados
 548                 my $marcxml; # temporary storage of MARCXML scalar
 549                 ($error,$marcxml) = marc2marcxml($record,$to_encoding,$flavour);
 550                 unless ($error) {
 551                         ($error,$newrecord) = marcxml2marc($marcxml,$to_encoding,$flavour);
 552                 }
 553
 554         # MARCXML Record
 555         } elsif (lc($format) =~ /^marcxml$/o) { # MARCXML Record
 556                 my $marc;
 557                 ($error,$marc) = marcxml2marc($record,$to_encoding,$flavour);
 558                 unless ($error) {
 559                         ($error,$newrecord) = marc2marcxml($record,$to_encoding,$flavour);
 560                 }
 561         } else {
 562                 $error.="Unsupported record format:".$format;
 563         }
 564         return ($error,$newrecord);
 565 }
 566
 567 =head1 INTERNAL FUNCTIONS
 568
 569 =head2 _entity_encode - Entity-encode an array of strings
 570
 571 =over 4
 572
 573 my ($entity_encoded_string) = _entity_encode($string);
 574
 575 or
 576
 577 my (@entity_encoded_strings) = _entity_encode(@strings);
 578
 579 Entity-encode an array of strings
 580
 581 =back
 582
 583 =cut
 584
 585 sub _entity_encode {
 586         my @strings = @_;
 587         my @strings_entity_encoded;
 588         foreach my $string (@strings) {
 589                 my $nfc_string = NFC($string);
 590                 $nfc_string =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
 591                 push @strings_entity_encoded, $nfc_string;
 592         }
 593         return @strings_entity_encoded;
 594 }
 595
 596 END { }       # module clean-up code here (global destructor)
 597 1;
 598 __END__
 599
 600 =head1 AUTHOR
 601
 602 Joshua Ferraro <jmf@liblime.com>
 603
 604 =head1 MODIFICATIONS
 605
 606
 607 =cut