C4/Record.pm

   1 package C4::Record;
   2 #
   3 # Copyright 2006 (C) LibLime
   4 # Parts copyright 2010 BibLibre
   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
  18 # with Koha; if not, write to the Free Software Foundation, Inc.,
  19 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  20 #
  21 #
  22 use strict;
  23 #use warnings; FIXME - Bug 2505
  24
  25 # please specify in which methods a given module is used
  26 use MARC::Record; # marc2marcxml, marcxml2marc, changeEncoding
  27 use MARC::File::XML; # marc2marcxml, marcxml2marc, changeEncoding
  28 use MARC::Crosswalk::DublinCore; # marc2dcxml
  29 use Biblio::EndnoteStyle;
  30 use Unicode::Normalize; # _entity_encode
  31 use XML::LibXSLT;
  32 use XML::LibXML;
  33 use C4::Biblio; #marc2bibtex
  34 use C4::Csv; #marc2csv
  35 use C4::Koha; #marc2csv
  36 use YAML; #marcrecords2csv
  37 use Text::CSV::Encoded; #marc2csv
  38
  39 use vars qw($VERSION @ISA @EXPORT);
  40
  41 # set the version for version checking
  42 $VERSION = 3.07.00.049;
  43
  44 @ISA = qw(Exporter);
  45
  46 # only export API methods
  47
  48 @EXPORT = qw(
  49   &marc2endnote
  50   &marc2marc
  51   &marc2marcxml
  52   &marcxml2marc
  53   &marc2dcxml
  54   &marc2modsxml
  55   &marc2madsxml
  56   &marc2bibtex
  57   &marc2csv
  58   &changeEncoding
  59 );
  60
  61 =head1 NAME
  62
  63 C4::Record - MARC, MARCXML, DC, MODS, XML, etc. Record Management Functions and API
  64
  65 =head1 SYNOPSIS
  66
  67 New in Koha 3.x. This module handles all record-related management functions.
  68
  69 =head1 API (EXPORTED FUNCTIONS)
  70
  71 =head2 marc2marc - Convert from one flavour of ISO-2709 to another
  72
  73   my ($error,$newmarc) = marc2marc($marc,$to_flavour,$from_flavour,$encoding);
  74
  75 Returns an ISO-2709 scalar
  76
  77 =cut
  78
  79 sub marc2marc {
  80         my ($marc,$to_flavour,$from_flavour,$encoding) = @_;
  81         my $error;
  82     if ($to_flavour =~ m/marcstd/) {
  83         my $marc_record_obj;
  84         if ($marc =~ /^MARC::Record/) { # it's already a MARC::Record object
  85             $marc_record_obj = $marc;
  86         } else { # it's not a MARC::Record object, make it one
  87             eval { $marc_record_obj = MARC::Record->new_from_usmarc($marc) }; # handle exceptions
  88
  89 # conversion to MARC::Record object failed, populate $error
  90                 if ($@) { $error .="\nCreation of MARC::Record object failed: ".$MARC::File::ERROR };
  91         }
  92         unless ($error) {
  93             my @privatefields;
  94             foreach my $field ($marc_record_obj->fields()) {
  95                 if ($field->tag() =~ m/9/ && ($field->tag() != '490' || C4::Context->preference("marcflavour") eq 'UNIMARC')) {
  96                     push @privatefields, $field;
  97                 } elsif (! ($field->is_control_field())) {
  98                     $field->delete_subfield(code => '9') if ($field->subfield('9'));
  99                 }
 100             }
 101             $marc_record_obj->delete_field($_) for @privatefields;
 102             $marc = $marc_record_obj->as_usmarc();
 103         }
 104     } else {
 105         $error = "Feature not yet implemented\n";
 106     }
 107         return ($error,$marc);
 108 }
 109
 110 =head2 marc2marcxml - Convert from ISO-2709 to MARCXML
 111
 112   my ($error,$marcxml) = marc2marcxml($marc,$encoding,$flavour);
 113
 114 Returns a MARCXML scalar
 115
 116 C<$marc> - an ISO-2709 scalar or MARC::Record object
 117
 118 C<$encoding> - UTF-8 or MARC-8 [UTF-8]
 119
 120 C<$flavour> - MARC21 or UNIMARC
 121
 122 C<$dont_entity_encode> - a flag that instructs marc2marcxml not to entity encode the xml before returning (optional)
 123
 124 =cut
 125
 126 sub marc2marcxml {
 127         my ($marc,$encoding,$flavour,$dont_entity_encode) = @_;
 128         my $error; # the error string
 129         my $marcxml; # the final MARCXML scalar
 130
 131         # test if it's already a MARC::Record object, if not, make it one
 132         my $marc_record_obj;
 133         if ($marc =~ /^MARC::Record/) { # it's already a MARC::Record object
 134                 $marc_record_obj = $marc;
 135         } else { # it's not a MARC::Record object, make it one
 136                 eval { $marc_record_obj = MARC::Record->new_from_usmarc($marc) }; # handle exceptions
 137
 138                 # conversion to MARC::Record object failed, populate $error
 139                 if ($@) { $error .="\nCreation of MARC::Record object failed: ".$MARC::File::ERROR };
 140         }
 141         # only proceed if no errors so far
 142         unless ($error) {
 143
 144                 # check the record for warnings
 145                 my @warnings = $marc_record_obj->warnings();
 146                 if (@warnings) {
 147                         warn "\nWarnings encountered while processing ISO-2709 record with title \"".$marc_record_obj->title()."\":\n";
 148                         foreach my $warn (@warnings) { warn "\t".$warn };
 149                 }
 150                 unless($encoding) {$encoding = "UTF-8"}; # set default encoding
 151                 unless($flavour) {$flavour = C4::Context->preference("marcflavour")}; # set default MARC flavour
 152
 153                 # attempt to convert the record to MARCXML
 154                 eval { $marcxml = $marc_record_obj->as_xml_record($flavour) }; #handle exceptions
 155
 156                 # record creation failed, populate $error
 157                 if ($@) {
 158                         $error .= "Creation of MARCXML failed:".$MARC::File::ERROR;
 159                         $error .= "Additional information:\n";
 160                         my @warnings = $@->warnings();
 161                         foreach my $warn (@warnings) { $error.=$warn."\n" };
 162
 163                 # record creation was successful
 164         } else {
 165
 166                         # check the record for warning flags again (warnings() will be cleared already if there was an error, see above block
 167                         @warnings = $marc_record_obj->warnings();
 168                         if (@warnings) {
 169                                 warn "\nWarnings encountered while processing ISO-2709 record with title \"".$marc_record_obj->title()."\":\n";
 170                                 foreach my $warn (@warnings) { warn "\t".$warn };
 171                         }
 172                 }
 173
 174                 # only proceed if no errors so far
 175                 unless ($error) {
 176
 177                         # entity encode the XML unless instructed not to
 178                 unless ($dont_entity_encode) {
 179                         my ($marcxml_entity_encoded) = _entity_encode($marcxml);
 180                         $marcxml = $marcxml_entity_encoded;
 181                 }
 182                 }
 183         }
 184         # return result to calling program
 185         return ($error,$marcxml);
 186 }
 187
 188 =head2 marcxml2marc - Convert from MARCXML to ISO-2709
 189
 190   my ($error,$marc) = marcxml2marc($marcxml,$encoding,$flavour);
 191
 192 Returns an ISO-2709 scalar
 193
 194 C<$marcxml> - a MARCXML record
 195
 196 C<$encoding> - UTF-8 or MARC-8 [UTF-8]
 197
 198 C<$flavour> - MARC21 or UNIMARC
 199
 200 =cut
 201
 202 sub marcxml2marc {
 203     my ($marcxml,$encoding,$flavour) = @_;
 204         my $error; # the error string
 205         my $marc; # the final ISO-2709 scalar
 206         unless($encoding) {$encoding = "UTF-8"}; # set the default encoding
 207         unless($flavour) {$flavour = C4::Context->preference("marcflavour")}; # set the default MARC flavour
 208
 209         # attempt to do the conversion
 210         eval { $marc = MARC::Record->new_from_xml($marcxml,$encoding,$flavour) }; # handle exceptions
 211
 212         # record creation failed, populate $error
 213         if ($@) {$error .="\nCreation of MARCXML Record failed: ".$@;
 214                 $error.=$MARC::File::ERROR if ($MARC::File::ERROR);
 215                 };
 216         # return result to calling program
 217         return ($error,$marc);
 218 }
 219
 220 =head2 marc2dcxml - Convert from ISO-2709 to Dublin Core
 221
 222   my ($error,$dcxml) = marc2dcxml($marc,$qualified);
 223
 224 Returns a DublinCore::Record object, will eventually return a Dublin Core scalar
 225
 226 FIXME: should return actual XML, not just an object
 227
 228 C<$marc> - an ISO-2709 scalar or MARC::Record object
 229
 230 C<$qualified> - specify whether qualified Dublin Core should be used in the input or output [0]
 231
 232 =cut
 233
 234 sub marc2dcxml {
 235         my ($marc,$qualified) = @_;
 236         my $error;
 237     # test if it's already a MARC::Record object, if not, make it one
 238     my $marc_record_obj;
 239     if ($marc =~ /^MARC::Record/) { # it's already a MARC::Record object
 240         $marc_record_obj = $marc;
 241     } else { # it's not a MARC::Record object, make it one
 242                 eval { $marc_record_obj = MARC::Record->new_from_usmarc($marc) }; # handle exceptions
 243
 244                 # conversion to MARC::Record object failed, populate $error
 245                 if ($@) {
 246                         $error .="\nCreation of MARC::Record object failed: ".$MARC::File::ERROR;
 247                 }
 248         }
 249         my $crosswalk = MARC::Crosswalk::DublinCore->new;
 250         if ($qualified) {
 251                 $crosswalk = MARC::Crosswalk::DublinCore->new( qualified => 1 );
 252         }
 253         my $dcxml = $crosswalk->as_dublincore($marc_record_obj);
 254         my $dcxmlfinal = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
 255         $dcxmlfinal .= "<metadata
 256   xmlns=\"http://example.org/myapp/\"
 257   xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"
 258   xsi:schemaLocation=\"http://example.org/myapp/ http://example.org/myapp/schema.xsd\"
 259   xmlns:dc=\"http://purl.org/dc/elements/1.1/\"
 260   xmlns:dcterms=\"http://purl.org/dc/terms/\">";
 261
 262         foreach my $element ( $dcxml->elements() ) {
 263                 $dcxmlfinal.="<"."dc:".$element->name().">".$element->content()."</"."dc:".$element->name().">\n";
 264     }
 265         $dcxmlfinal .= "\n</metadata>";
 266         return ($error,$dcxmlfinal);
 267 }
 268
 269 =head2 marc2modsxml - Convert from ISO-2709 to MODS
 270
 271   my $modsxml = marc2modsxml($marc);
 272
 273 Returns a MODS scalar
 274
 275 =cut
 276
 277 sub marc2modsxml {
 278     my ($marc) = @_;
 279     return _transformWithStylesheet($marc, "/prog/en/xslt/MARC21slim2MODS3-1.xsl");
 280 }
 281
 282 =head2 marc2madsxml - Convert from ISO-2709 to MADS
 283
 284   my $madsxml = marc2madsxml($marc);
 285
 286 Returns a MADS scalar
 287
 288 =cut
 289
 290 sub marc2madsxml {
 291     my ($marc) = @_;
 292     return _transformWithStylesheet($marc, "/prog/en/xslt/MARC21slim2MADS.xsl");
 293 }
 294
 295 =head2 _transformWithStylesheet - Transform a MARC record with a stylesheet
 296
 297     my $xml = _transformWithStylesheet($marc, $stylesheet)
 298
 299 Returns the XML scalar result of the transformation. $stylesheet should
 300 contain the path to a stylesheet under intrahtdocs.
 301
 302 =cut
 303
 304 sub _transformWithStylesheet {
 305     my ($marc, $stylesheet) = @_;
 306     # grab the XML, run it through our stylesheet, push it out to the browser
 307     my $xmlrecord = marc2marcxml($marc);
 308     my $xslfile = C4::Context->config('intrahtdocs') . $stylesheet;
 309     my $parser = XML::LibXML->new();
 310     my $xslt = XML::LibXSLT->new();
 311     my $source = $parser->parse_string($xmlrecord);
 312     my $style_doc = $parser->parse_file($xslfile);
 313     my $stylesheet = $xslt->parse_stylesheet($style_doc);
 314     my $results = $stylesheet->transform($source);
 315     my $newxmlrecord = $stylesheet->output_string($results);
 316     return ($newxmlrecord);
 317 }
 318
 319 sub marc2endnote {
 320     my ($marc) = @_;
 321         my $marc_rec_obj =  MARC::Record->new_from_usmarc($marc);
 322     my ( $abstract, $f260a, $f710a );
 323     my $f260 = $marc_rec_obj->field('260');
 324     if ($f260) {
 325         $f260a = $f260->subfield('a') if $f260;
 326     }
 327     my $f710 = $marc_rec_obj->field('710');
 328     if ($f710) {
 329         $f710a = $f710->subfield('a');
 330     }
 331     my $f500 = $marc_rec_obj->field('500');
 332     if ($f500) {
 333         $abstract = $f500->subfield('a');
 334     }
 335         my $fields = {
 336                 DB => C4::Context->preference("LibraryName"),
 337                 Title => $marc_rec_obj->title(),
 338                 Author => $marc_rec_obj->author(),
 339                 Publisher => $f710a,
 340                 City => $f260a,
 341                 Year => $marc_rec_obj->publication_date,
 342                 Abstract => $abstract,
 343         };
 344         my $endnote;
 345         my $style = new Biblio::EndnoteStyle();
 346         my $template;
 347         $template.= "DB - DB\n" if C4::Context->preference("LibraryName");
 348         $template.="T1 - Title\n" if $marc_rec_obj->title();
 349         $template.="A1 - Author\n" if $marc_rec_obj->author();
 350         $template.="PB - Publisher\n" if  $f710a;
 351         $template.="CY - City\n" if $f260a;
 352         $template.="Y1 - Year\n" if $marc_rec_obj->publication_date;
 353         $template.="AB - Abstract\n" if $abstract;
 354         my ($text, $errmsg) = $style->format($template, $fields);
 355         return ($text);
 356
 357 }
 358
 359 =head2 marc2csv - Convert several records from UNIMARC to CSV
 360
 361   my ($csv) = marc2csv($biblios, $csvprofileid, $itemnumbers);
 362
 363 Pre and postprocessing can be done through a YAML file
 364
 365 Returns a CSV scalar
 366
 367 C<$biblio> - a list of biblionumbers
 368
 369 C<$csvprofileid> - the id of the CSV profile to use for the export (see export_format.export_format_id and the GetCsvProfiles function in C4::Csv)
 370
 371 C<$itemnumbers> - a list of itemnumbers to export
 372
 373 =cut
 374
 375 sub marc2csv {
 376     my ($biblios, $id, $itemnumbers) = @_;
 377     my $output;
 378     my $csv = Text::CSV::Encoded->new();
 379
 380     # Getting yaml file
 381     my $configfile = "../tools/csv-profiles/$id.yaml";
 382     my ($preprocess, $postprocess, $fieldprocessing);
 383     if (-e $configfile){
 384         ($preprocess,$postprocess, $fieldprocessing) = YAML::LoadFile($configfile);
 385     }
 386
 387     # Preprocessing
 388     eval $preprocess if ($preprocess);
 389
 390     my $firstpass = 1;
 391     if ( $itemnumbers ) {
 392         for my $itemnumber ( @$itemnumbers) {
 393             my $biblionumber = GetBiblionumberFromItemnumber $itemnumber;
 394             $output .= marcrecord2csv( $biblionumber, $id, $firstpass, $csv, $fieldprocessing, [$itemnumber] );
 395             $firstpass = 0;
 396         }
 397     } else {
 398         foreach my $biblio (@$biblios) {
 399             $output .= marcrecord2csv( $biblio, $id, $firstpass, $csv, $fieldprocessing );
 400             $firstpass = 0;
 401         }
 402     }
 403
 404     # Postprocessing
 405     eval $postprocess if ($postprocess);
 406
 407     return $output;
 408 }
 409
 410 =head2 marcrecord2csv - Convert a single record from UNIMARC to CSV
 411
 412   my ($csv) = marcrecord2csv($biblio, $csvprofileid, $header);
 413
 414 Returns a CSV scalar
 415
 416 C<$biblio> - a biblionumber
 417
 418 C<$csvprofileid> - the id of the CSV profile to use for the export (see export_format.export_format_id and the GetCsvProfiles function in C4::Csv)
 419
 420 C<$header> - true if the headers are to be printed (typically at first pass)
 421
 422 C<$csv> - an already initialised Text::CSV object
 423
 424 C<$fieldprocessing>
 425
 426 C<$itemnumbers> a list of itemnumbers to export
 427
 428 =cut
 429
 430
 431 sub marcrecord2csv {
 432     my ($biblio, $id, $header, $csv, $fieldprocessing, $itemnumbers) = @_;
 433     my $output;
 434
 435     # Getting the record
 436     my $record = GetMarcBiblio($biblio);
 437     next unless $record;
 438     C4::Biblio::EmbedItemsInMarcBiblio( $record, $biblio, $itemnumbers );
 439     # Getting the framework
 440     my $frameworkcode = GetFrameworkCode($biblio);
 441
 442     # Getting information about the csv profile
 443     my $profile = GetCsvProfile($id);
 444
 445     # Getting output encoding
 446     my $encoding          = $profile->{encoding} || 'utf8';
 447     # Getting separators
 448     my $csvseparator      = $profile->{csv_separator}      || ',';
 449     my $fieldseparator    = $profile->{field_separator}    || '#';
 450     my $subfieldseparator = $profile->{subfield_separator} || '|';
 451
 452     # TODO: Be more generic (in case we have to handle other protected chars or more separators)
 453     if ($csvseparator eq '\t') { $csvseparator = "\t" }
 454     if ($fieldseparator eq '\t') { $fieldseparator = "\t" }
 455     if ($subfieldseparator eq '\t') { $subfieldseparator = "\t" }
 456     if ($csvseparator eq '\n') { $csvseparator = "\n" }
 457     if ($fieldseparator eq '\n') { $fieldseparator = "\n" }
 458     if ($subfieldseparator eq '\n') { $subfieldseparator = "\n" }
 459
 460     $csv = $csv->encoding_out($encoding) ;
 461     $csv->sep_char($csvseparator);
 462
 463     # Getting the marcfields
 464     my $marcfieldslist = $profile->{marcfields};
 465
 466     # Getting the marcfields as an array
 467     my @marcfieldsarray = split('\|', $marcfieldslist);
 468
 469    # Separating the marcfields from the user-supplied headers
 470     my @marcfields;
 471     foreach (@marcfieldsarray) {
 472         my @result = split('=', $_);
 473         if (scalar(@result) == 2) {
 474            push @marcfields, { header => $result[0], field => $result[1] };
 475         } else {
 476            push @marcfields, { field => $result[0] }
 477         }
 478     }
 479
 480     # If we have to insert the headers
 481     if ($header) {
 482         my @marcfieldsheaders;
 483         my $dbh   = C4::Context->dbh;
 484
 485         # For each field or subfield
 486         foreach (@marcfields) {
 487
 488             my $field = $_->{field};
 489         # Remove any blank char that might have unintentionally insered into the tag name
 490         $field =~ s/\s+//g;
 491
 492             # If we have a user-supplied header, we use it
 493             if (exists $_->{header}) {
 494                     push @marcfieldsheaders, $_->{header};
 495             } else {
 496                 # If not, we get the matching tag name from koha
 497                 if (index($field, '$') > 0) {
 498                     my ($fieldtag, $subfieldtag) = split('\$', $field);
 499                     my $query = "SELECT liblibrarian FROM marc_subfield_structure WHERE tagfield=? AND tagsubfield=?";
 500                     my $sth = $dbh->prepare($query);
 501                     $sth->execute($fieldtag, $subfieldtag);
 502                     my @results = $sth->fetchrow_array();
 503                     push @marcfieldsheaders, $results[0];
 504                 } else {
 505                     my $query = "SELECT liblibrarian FROM marc_tag_structure WHERE tagfield=?";
 506                     my $sth = $dbh->prepare($query);
 507                     $sth->execute($field);
 508                     my @results = $sth->fetchrow_array();
 509                     push @marcfieldsheaders, $results[0];
 510                 }
 511             }
 512         }
 513         $csv->combine(@marcfieldsheaders);
 514         $output = $csv->string() . "\n";
 515     }
 516
 517     # For each marcfield to export
 518     my @fieldstab;
 519     foreach (@marcfields) {
 520         my $marcfield = $_->{field};
 521         # If it is a subfield
 522         if (index($marcfield, '$') > 0) {
 523             my ($fieldtag, $subfieldtag) = split('\$', $marcfield);
 524             my @fields = $record->field($fieldtag);
 525             my @tmpfields;
 526
 527             # For each field
 528             foreach my $field (@fields) {
 529
 530                 # We take every matching subfield
 531                 my @subfields = $field->subfield($subfieldtag);
 532                 foreach my $subfield (@subfields) {
 533
 534                     # Getting authorised value
 535                     my $authvalues = GetKohaAuthorisedValuesFromField($fieldtag, $subfieldtag, $frameworkcode, undef);
 536                     push @tmpfields, (defined $authvalues->{$subfield}) ? $authvalues->{$subfield} : $subfield;
 537                 }
 538             }
 539             push (@fieldstab, join($subfieldseparator, @tmpfields));
 540         # Or a field
 541         } else {
 542             my @fields = ($record->field($marcfield));
 543             my $authvalues = GetKohaAuthorisedValuesFromField($marcfield, undef, $frameworkcode, undef);
 544
 545         my @valuesarray;
 546         foreach (@fields) {
 547             my $value;
 548
 549             # If it is a control field
 550             if ($_->is_control_field) {
 551                 $value = defined $authvalues->{$_->as_string} ? $authvalues->{$_->as_string} : $_->as_string;
 552             } else {
 553                 # If it is a field, we gather all subfields, joined by the subfield separator
 554                 my @subvaluesarray;
 555                 my @subfields = $_->subfields;
 556                 foreach my $subfield (@subfields) {
 557                     push (@subvaluesarray, defined $authvalues->{$subfield->[1]} ? $authvalues->{$subfield->[1]} : $subfield->[1]);
 558                 }
 559                 $value = join ($subfieldseparator, @subvaluesarray);
 560             }
 561
 562             # Field processing
 563             eval $fieldprocessing if ($fieldprocessing);
 564
 565             push @valuesarray, $value;
 566         }
 567             push (@fieldstab, join($fieldseparator, @valuesarray));
 568          }
 569     };
 570
 571     $csv->combine(@fieldstab);
 572     $output .= $csv->string() . "\n";
 573
 574     return $output;
 575
 576 }
 577
 578
 579 =head2 changeEncoding - Change the encoding of a record
 580
 581   my ($error, $newrecord) = changeEncoding($record,$format,$flavour,$to_encoding,$from_encoding);
 582
 583 Changes the encoding of a record
 584
 585 C<$record> - the record itself can be in ISO-2709, a MARC::Record object, or MARCXML for now (required)
 586
 587 C<$format> - MARC or MARCXML (required)
 588
 589 C<$flavour> - MARC21 or UNIMARC, if MARC21, it will change the leader (optional) [defaults to Koha system preference]
 590
 591 C<$to_encoding> - the encoding you want the record to end up in (optional) [UTF-8]
 592
 593 C<$from_encoding> - the encoding the record is currently in (optional, it will probably be able to tell unless there's a problem with the record)
 594
 595 FIXME: the from_encoding doesn't work yet
 596
 597 FIXME: better handling for UNIMARC, it should allow management of 100 field
 598
 599 FIXME: shouldn't have to convert to and from xml/marc just to change encoding someone needs to re-write MARC::Record's 'encoding' method to actually alter the encoding rather than just changing the leader
 600
 601 =cut
 602
 603 sub changeEncoding {
 604         my ($record,$format,$flavour,$to_encoding,$from_encoding) = @_;
 605         my $newrecord;
 606         my $error;
 607         unless($flavour) {$flavour = C4::Context->preference("marcflavour")};
 608         unless($to_encoding) {$to_encoding = "UTF-8"};
 609
 610         # ISO-2709 Record (MARC21 or UNIMARC)
 611         if (lc($format) =~ /^marc$/o) {
 612                 # if we're converting encoding of an ISO2709 file, we need to roundtrip through XML
 613                 #       because MARC::Record doesn't directly provide us with an encoding method
 614                 #       It's definitely less than idea and should be fixed eventually - kados
 615                 my $marcxml; # temporary storage of MARCXML scalar
 616                 ($error,$marcxml) = marc2marcxml($record,$to_encoding,$flavour);
 617                 unless ($error) {
 618                         ($error,$newrecord) = marcxml2marc($marcxml,$to_encoding,$flavour);
 619                 }
 620
 621         # MARCXML Record
 622         } elsif (lc($format) =~ /^marcxml$/o) { # MARCXML Record
 623                 my $marc;
 624                 ($error,$marc) = marcxml2marc($record,$to_encoding,$flavour);
 625                 unless ($error) {
 626                         ($error,$newrecord) = marc2marcxml($record,$to_encoding,$flavour);
 627                 }
 628         } else {
 629                 $error.="Unsupported record format:".$format;
 630         }
 631         return ($error,$newrecord);
 632 }
 633
 634 =head2 marc2bibtex - Convert from MARC21 and UNIMARC to BibTex
 635
 636   my ($bibtex) = marc2bibtex($record, $id);
 637
 638 Returns a BibTex scalar
 639
 640 C<$record> - a MARC::Record object
 641
 642 C<$id> - an id for the BibTex record (might be the biblionumber)
 643
 644 =cut
 645
 646
 647 sub marc2bibtex {
 648     my ($record, $id) = @_;
 649     my $tex;
 650     my $marcflavour = C4::Context->preference("marcflavour");
 651
 652     # Authors
 653     my $author;
 654     my @texauthors;
 655     my ( $mintag, $maxtag, $fields_filter );
 656     if ( $marcflavour eq "UNIMARC" ) {
 657         $mintag        = "700";
 658         $maxtag        = "712";
 659         $fields_filter = '7..';
 660     }
 661     else {
 662         $mintag        = "700";
 663         $maxtag        = "720";
 664         $fields_filter = '7..';
 665     }
 666     foreach my $field ( $record->field($fields_filter) ) {
 667         next unless $field->tag() >= $mintag && $field->tag() <= $maxtag;
 668         # author formatted surname, firstname
 669         my $texauthor = '';
 670         if ( $marcflavour eq "UNIMARC" ) {
 671             $texauthor = join ', ',
 672               ( $field->subfield('a'), $field->subfield('b') );
 673         }
 674         else {
 675             $texauthor = $field->subfield('a');
 676         }
 677         push @texauthors, $texauthor if $texauthor;
 678     }
 679     $author = join ' and ', @texauthors;
 680
 681     # Defining the conversion hash according to the marcflavour
 682     my %bh;
 683     if ( $marcflavour eq "UNIMARC" ) {
 684
 685         # FIXME, TODO : handle repeatable fields
 686         # TODO : handle more types of documents
 687
 688         # Unimarc to bibtex hash
 689         %bh = (
 690
 691             # Mandatory
 692             author    => $author,
 693             title     => $record->subfield("200", "a") || "",
 694             editor    => $record->subfield("210", "g") || "",
 695             publisher => $record->subfield("210", "c") || "",
 696             year      => $record->subfield("210", "d") || $record->subfield("210", "h") || "",
 697
 698             # Optional
 699             volume  =>  $record->subfield("200", "v") || "",
 700             series  =>  $record->subfield("225", "a") || "",
 701             address =>  $record->subfield("210", "a") || "",
 702             edition =>  $record->subfield("205", "a") || "",
 703             note    =>  $record->subfield("300", "a") || "",
 704             url     =>  $record->subfield("856", "u") || ""
 705         );
 706     } else {
 707
 708         # Marc21 to bibtex hash
 709         %bh = (
 710
 711             # Mandatory
 712             author    => $author,
 713             title     => $record->subfield("245", "a") || "",
 714             editor    => $record->subfield("260", "f") || "",
 715             publisher => $record->subfield("260", "b") || "",
 716             year      => $record->subfield("260", "c") || $record->subfield("260", "g") || "",
 717
 718             # Optional
 719             # unimarc to marc21 specification says not to convert 200$v to marc21
 720             series  =>  $record->subfield("490", "a") || "",
 721             address =>  $record->subfield("260", "a") || "",
 722             edition =>  $record->subfield("250", "a") || "",
 723             note    =>  $record->subfield("500", "a") || "",
 724             url     =>  $record->subfield("856", "u") || ""
 725         );
 726     }
 727
 728     $tex .= "\@book{";
 729     $tex .= join(",\n", $id, map { $bh{$_} ? qq(\t$_ = "$bh{$_}") : () } keys %bh);
 730     $tex .= "\n}\n";
 731
 732     return $tex;
 733 }
 734
 735
 736 =head1 INTERNAL FUNCTIONS
 737
 738 =head2 _entity_encode - Entity-encode an array of strings
 739
 740   my ($entity_encoded_string) = _entity_encode($string);
 741
 742 or
 743
 744   my (@entity_encoded_strings) = _entity_encode(@strings);
 745
 746 Entity-encode an array of strings
 747
 748 =cut
 749
 750 sub _entity_encode {
 751         my @strings = @_;
 752         my @strings_entity_encoded;
 753         foreach my $string (@strings) {
 754                 my $nfc_string = NFC($string);
 755                 $nfc_string =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
 756                 push @strings_entity_encoded, $nfc_string;
 757         }
 758         return @strings_entity_encoded;
 759 }
 760
 761 END { }       # module clean-up code here (global destructor)
 762 1;
 763 __END__
 764
 765 =head1 AUTHOR
 766
 767 Joshua Ferraro <jmf@liblime.com>
 768
 769 =head1 MODIFICATIONS
 770
 771
 772 =cut