r730@llin: dpavlin | 2006-06-29 21:33:48 +0200

 use MARC::Record 2.0 to support utf-8 encoding in MARC
 http://marcpm.sourceforge.net/

git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@541 07558da8-63fa-0310-ba24-9fe276d99e06

diff --git a/Makefile.PL b/Makefile.PL
index f812691..ed806d5 100644 (file)
--- a/Makefile.PL
+++ b/Makefile.PL
@@ -29,7 +29,7 @@ WriteMakefile(
        'Search::Estraier' => 0.06,
        'List::Util' => 0,
        'Spreadsheet::ParseExcel', => 0,
-       'MARC::Record' => 0,
+       'MARC::Record' => 2.0,
     },
     dist                => { COMPRESS => 'gzip -9f', SUFFIX => 'gz', },
     clean               => { FILES => 'WebPAC-* pod2html Makefile tags' },

diff --git a/lib/WebPAC/Normalize.pm b/lib/WebPAC/Normalize.pm
index ea31b37..459bb7a 100644 (file)
--- a/lib/WebPAC/Normalize.pm
+++ b/lib/WebPAC/Normalize.pm
@@ -17,6 +17,7 @@ use strict;
 
 #use base qw/WebPAC::Common/;
 use Data::Dumper;
+use Encode qw/from_to/;
 
 =head1 NAME
 
@@ -24,11 +25,11 @@ WebPAC::Normalize - describe normalisaton rules using sets
 
 =head1 VERSION
 
-Version 0.05
+Version 0.06
 
 =cut
 
-our $VERSION = '0.05';
+our $VERSION = '0.06';
 
 =head1 SYNOPSIS
 
@@ -57,6 +58,7 @@ Return data structure
        lookup => $lookup->lookup_hash,
        row => $row,
        rules => $normalize_pl_config,
+       marc_encoding => 'utf-8',
   );
 
 Options C<lookup>, C<row>, C<rules> and C<log> are mandatory while all
@@ -78,8 +80,7 @@ sub data_structure {
        no strict 'subs';
        _set_lookup( $arg->{lookup} );
        _set_rec( $arg->{row} );
-       _clean_ds();
-
+       _clean_ds( %{ $arg } );
        eval "$arg->{rules}";
        die "error evaling $arg->{rules}: $@\n" if ($@);
 
@@ -110,6 +111,7 @@ Return hash formatted as data structure
 
 my $out;
 my $marc21;
+my $marc_encoding;
 
 sub _get_ds {
        return $out;
@@ -124,8 +126,10 @@ Clean data structure hash for next record
 =cut
 
 sub _clean_ds {
+       my $a = {@_};
        $out = undef;
        $marc21 = undef;
+       $marc_encoding = $a->{marc_encoding};
 }
 
 =head2 _set_lookup
@@ -222,7 +226,10 @@ sub marc21 {
 
        my $sf = shift or die "marc21 needs subfield";
 
-       foreach my $v (@_) {
+       foreach (@_) {
+               my $v = $_;             # make var read-write for Encode
+               next unless (defined($v) && $v !~ /^\s+$/);
+               from_to($v, 'iso-8859-2', $marc_encoding) if ($marc_encoding);
                push @{ $marc21 }, [ $f, ' ', ' ', $sf => $v ];
        }
 }

diff --git a/run.pl b/run.pl
index 64e8095..6bb258f 100755 (executable)
--- a/run.pl
+++ b/run.pl
@@ -19,7 +19,7 @@ use Getopt::Long;
 use File::Path;
 use Time::HiRes qw/time/;
 use File::Slurp;
-use MARC::Record;
+use MARC::Record 2.0;  # need 2.0 for utf-8 encoding see marcpm.sf.net
 
 =head1 NAME
 
@@ -269,6 +269,7 @@ while (my ($database, $db_config) = each %{ $config->{databases} }) {
                                        row => $row,
                                        rules => $rules,
                                        lookup => $lookup ? $lookup->lookup_hash : undef,
+                                       marc_encoding => 'utf-8',
                                );
 
                                $db->save_ds(
@@ -285,6 +286,7 @@ while (my ($database, $db_config) = each %{ $config->{databases} }) {
 
                                if ($marc_fh) {
                                        my $marc = new MARC::Record;
+                                       $marc->encoding( 'utf-8' );
                                        $marc->add_fields( WebPAC::Normalize::_get_marc21_fields() );
                                        print $marc_fh $marc->as_usmarc;
                                }