trunk/scripts/dump_fastmarc.pl

#!/usr/bin/perl -w

use strict;
use lib 'lib';

use MARC::Fast;
use Getopt::Std;
use Data::Dump qw/dump/;

=head1 NAME

dump_fastmarc.pl - display MARC records

=head2 USAGE

  dump_fastmarc.pl /path/to/dump.marc

=head2 OPTIONS

=over 16

=item -o offset

dump records starting with C<offset>

=item -l limit

dump just C<limit> records

=item -h

dump result of C<to_hash> on record

=item -d

turn debugging output on

=item -t

dump tsv file for TokyoCabinet import

=back

=cut

my %opt;
getopts('do:l:ht', \%opt);

my $file = shift @ARGV || die "usage: $0 [-o offset] [-l limit] [-h] [-d] file.marc\n";

my $marc = new MARC::Fast(
        marcdb => $file,
        debug => $opt{d},
);


my $min = 1;
my $max = $marc->count;

if (my $mfn = $opt{n}) {
        $min = $max = $mfn;
        print STDERR "Dumping $mfn only\n";
} elsif (my $limit = $opt{l}) {
        print STDERR "$file has $max records, using first $limit\n";
        $max = $limit;
} else {
        print STDERR "$file has $max records...\n";
}

for my $mfn ($min .. $max) {
        my $rec = $marc->fetch($mfn) || next;
        warn "rec is ",dump($rec) if ($opt{d});
        if ( $opt{t} ) {
                print "rec\t$mfn\tleader\t", $marc->last_leader, "\t";
                my $ascii = $marc->to_ascii($mfn);
                $ascii =~ s{\n}{\t}gs;
                print "$ascii\n";
        } else {
                print "REC $mfn\n";
                print $marc->last_leader,"\n";
                print $marc->to_ascii($mfn),"\n";
        }
        warn "hash is ",dump($marc->to_hash($mfn, include_subfields => 1)) if ($opt{h});
}
1	dpavlin	1	#!/usr/bin/perl -w
2
3			use strict;
4	dpavlin	37	use lib 'lib';
5	dpavlin	1
6			use MARC::Fast;
7	dpavlin	4	use Getopt::Std;
8	dpavlin	26	use Data::Dump qw/dump/;
9	dpavlin	1
10	dpavlin	7	=head1 NAME
11
12			dump_fastmarc.pl - display MARC records
13
14			=head2 USAGE
15
16			dump_fastmarc.pl /path/to/dump.marc
17
18			=head2 OPTIONS
19
20			=over 16
21
22	dpavlin	10	=item -o offset
23	dpavlin	7
24	dpavlin	10	dump records starting with C<offset>
25	dpavlin	7
26			=item -l limit
27
28	dpavlin	10	dump just C<limit> records
29	dpavlin	7
30			=item -h
31
32			dump result of C<to_hash> on record
33
34			=item -d
35
36			turn debugging output on
37
38	dpavlin	46	=item -t
39
40			dump tsv file for TokyoCabinet import
41
42	dpavlin	7	=back
43
44			=cut
45
46	dpavlin	4	my %opt;
47	dpavlin	46	getopts('do:l:ht', \%opt);
48	dpavlin	1
49	dpavlin	10	my $file = shift @ARGV \|\| die "usage: $0 [-o offset] [-l limit] [-h] [-d] file.marc\n";
50	dpavlin	4
51	dpavlin	1	my $marc = new MARC::Fast(
52			marcdb => $file,
53	dpavlin	7	debug => $opt{d},
54	dpavlin	1	);
55
56
57	dpavlin	4	my $min = 1;
58			my $max = $marc->count;
59
60	dpavlin	7	if (my $mfn = $opt{n}) {
61	dpavlin	4	$min = $max = $mfn;
62	dpavlin	6	print STDERR "Dumping $mfn only\n";
63	dpavlin	7	} elsif (my $limit = $opt{l}) {
64			print STDERR "$file has $max records, using first $limit\n";
65			$max = $limit;
66	dpavlin	6	} else {
67			print STDERR "$file has $max records...\n";
68	dpavlin	4	}
69
70			for my $mfn ($min .. $max) {
71	dpavlin	1	my $rec = $marc->fetch($mfn) \|\| next;
72	dpavlin	46	warn "rec is ",dump($rec) if ($opt{d});
73			if ( $opt{t} ) {
74			print "rec\t$mfn\tleader\t", $marc->last_leader, "\t";
75			my $ascii = $marc->to_ascii($mfn);
76			$ascii =~ s{\n}{\t}gs;
77			print "$ascii\n";
78			} else {
79			print "REC $mfn\n";
80			print $marc->last_leader,"\n";
81			print $marc->to_ascii($mfn),"\n";
82			}
83			warn "hash is ",dump($marc->to_hash($mfn, include_subfields => 1)) if ($opt{h});
84	dpavlin	1	}