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

[webpac2] / lib / WebPAC / Normalize.pm

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 ];
        }
 }