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 $debug = 0;
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 ) if $debug;

                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" if $debug;
        }

}
1	#!/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 $debug = 0;
13	my @fields;
14
15	sub parse_subfields {
16	my $text = shift;
17	if ( $text =~ /^_([a-z])(.+)$/ ) {
18	return ( $1, $2 );
19	} else {
20	$text =~ s/^ // \|\| warn "no space in front of: '$text'\n";
21	return $text;
22	}
23	}
24
25	my $leader;
26
27	while(<>) {
28	chomp;
29	$_ = decode('utf-8', $_);
30
31	if ( /^LDR (.+)$/ ) {
32	$leader = $1;
33	} elsif ( /^(\d\d\d) (.)(.) (.+)$/ ) {
34	if ( $1 < 10 ) {
35	push @fields, [ $1, parse_subfields($4) ];
36	} else {
37	push @fields, [ $1, $2, $3, parse_subfields($4) ];
38	}
39	} elsif ( /^\s{7}_([a-z])(.+)$/ ) {
40	push @{ $fields[ $#fields ] }, $1, $2;
41	} elsif ( /^$/ ) {
42	warn dump( @fields ) if $debug;
43
44	my $marc = new MARC::Record;
45	$marc->encoding( 'utf-8' );
46	$marc->leader( $leader );
47	$marc->add_fields( @fields );
48
49	warn $marc->as_formatted, $/, $/;
50	print $marc->as_usmarc;
51
52	@fields = ();
53	} else {
54	warn "IGNORED: $_\n" if $debug;
55	}
56
57	}