trunk/bin/dump2marc.pl

#!/usr/bin/perl

# Convert marcdump output back to marc file

use warnings;
use strict;

use MARC::Record;
use Data::Dump qw(dump);
use Encode;

my @fields;

sub parse_subfields {
        my $text = shift;
        if ( $text =~ /^_([a-z])(.+)$/ ) {
                return ( $1, $2 );
        } else {
                $text =~ s/^ // || warn "no space in front of: '$text'\n";
                return $text;
        }
}

my $leader;

while(<>) {
        chomp;
        $_ = decode('utf-8', $_);

        if ( /^LDR (.+)$/ ) {
                $leader = $1;
        } elsif ( /^(\d\d\d) (.)(.) (.+)$/ ) {
                if ( $1 < 10 ) {
                        push @fields, [ $1, parse_subfields($4) ];
                } else {
                        push @fields, [ $1, $2, $3, parse_subfields($4) ];
                }
        } elsif ( /^\s{7}_([a-z])(.+)$/ ) {
                push @{ $fields[ $#fields ] }, $1, $2;
        } elsif ( /^$/ ) {
                warn dump( @fields );

                my $marc = new MARC::Record;
                $marc->encoding( 'utf-8' );
                $marc->leader( $leader );
                $marc->add_fields( @fields );

                warn $marc->as_formatted;
                print $marc->as_usmarc;

                @fields = ();
        } else {
                warn "IGNORED: $_\n";
        }

}
1	dpavlin	1316	#!/usr/bin/perl
2
3			# Convert marcdump output back to marc file
4
5			use warnings;
6			use strict;
7
8			use MARC::Record;
9			use Data::Dump qw(dump);
10			use Encode;
11
12			my @fields;
13
14			sub parse_subfields {
15			my $text = shift;
16			if ( $text =~ /^_([a-z])(.+)$/ ) {
17			return ( $1, $2 );
18			} else {
19			$text =~ s/^ // \|\| warn "no space in front of: '$text'\n";
20			return $text;
21			}
22			}
23
24			my $leader;
25
26			while(<>) {
27			chomp;
28			$_ = decode('utf-8', $_);
29
30			if ( /^LDR (.+)$/ ) {
31			$leader = $1;
32			} elsif ( /^(\d\d\d) (.)(.) (.+)$/ ) {
33			if ( $1 < 10 ) {
34			push @fields, [ $1, parse_subfields($4) ];
35			} else {
36			push @fields, [ $1, $2, $3, parse_subfields($4) ];
37			}
38			} elsif ( /^\s{7}_([a-z])(.+)$/ ) {
39			push @{ $fields[ $#fields ] }, $1, $2;
40			} elsif ( /^$/ ) {
41			warn dump( @fields );
42
43			my $marc = new MARC::Record;
44			$marc->encoding( 'utf-8' );
45			$marc->leader( $leader );
46			$marc->add_fields( @fields );
47
48			warn $marc->as_formatted;
49			print $marc->as_usmarc;
50
51			@fields = ();
52			} else {
53			warn "IGNORED: $_\n";
54			}
55
56			}