WebPAC/Input/TSV.pm

package WebPAC::Input::TSV;

use warnings;
use strict;

use WebPAC::Input;
use base qw/WebPAC::Common/;

use Encode;
use Data::Dump qw/dump/;

=head1 NAME

WebPAC::Input::TSV - tab separated values

=cut

=head1 FUNCTIONS

=head2 new

  my $input = new WebPAC::Input::TSV(
        path => '/path/to/records.tsv',
  );

=back

Default encoding of input file is C<utf-8>

=cut

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $arg = {@_};

        my $log = $self->_get_logger();

        open( my $fh, '<:raw', $arg->{path} ) || $log->logconfess("can't open $arg->{path}: $!");

        $self->{size} = 0;

        while ( my $line = <$fh> ) {

                my $rec;
                $rec->{'000'} = [ ++$self->{size} ];

                my $col = 'A';
                foreach my $v ( split(/\t/,$line) ) {
                        next if $v eq '\N';
                        $rec->{ $col++ } = Encode::decode_utf8( $v );
                }

                push @{ $self->{_rec} }, $rec;

        };

        $log->debug("loaded ", $self->size, " records");

        $self ? return $self : return undef;
}

=head2 fetch_rec

Return record with ID C<$mfn> from database

  my $rec = $input->fetch_rec( $mfn, $filter_coderef );

=cut

sub fetch_rec {
        my ( $self, $mfn, $filter_coderef ) = @_;

        return $self->{_rec}->[$mfn-1];
}


=head2 size

Return number of records in database

  my $size = $input->size;

=cut

sub size {
        my $self = shift;
        return $self->{size};
}

=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2010 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Input::TSV
1	package WebPAC::Input::TSV;
2
3	use warnings;
4	use strict;
5
6	use WebPAC::Input;
7	use base qw/WebPAC::Common/;
8
9	use Encode;
10	use Data::Dump qw/dump/;
11
12	=head1 NAME
13
14	WebPAC::Input::TSV - tab separated values
15
16	=cut
17
18	=head1 FUNCTIONS
19
20	=head2 new
21
22	my $input = new WebPAC::Input::TSV(
23	path => '/path/to/records.tsv',
24	);
25
26	=back
27
28	Default encoding of input file is C<utf-8>
29
30	=cut
31
32	sub new {
33	my $class = shift;
34	my $self = {@_};
35	bless($self, $class);
36
37	my $arg = {@_};
38
39	my $log = $self->_get_logger();
40
41	open( my $fh, '<:raw', $arg->{path} ) \|\| $log->logconfess("can't open $arg->{path}: $!");
42
43	$self->{size} = 0;
44
45	while ( my $line = <$fh> ) {
46
47	my $rec;
48	$rec->{'000'} = [ ++$self->{size} ];
49
50	my $col = 'A';
51	foreach my $v ( split(/\t/,$line) ) {
52	next if $v eq '\N';
53	$rec->{ $col++ } = Encode::decode_utf8( $v );
54	}
55
56	push @{ $self->{_rec} }, $rec;
57
58	};
59
60	$log->debug("loaded ", $self->size, " records");
61
62	$self ? return $self : return undef;
63	}
64
65	=head2 fetch_rec
66
67	Return record with ID C<$mfn> from database
68
69	my $rec = $input->fetch_rec( $mfn, $filter_coderef );
70
71	=cut
72
73	sub fetch_rec {
74	my ( $self, $mfn, $filter_coderef ) = @_;
75
76	return $self->{_rec}->[$mfn-1];
77	}
78
79
80	=head2 size
81
82	Return number of records in database
83
84	my $size = $input->size;
85
86	=cut
87
88	sub size {
89	my $self = shift;
90	return $self->{size};
91	}
92
93	=head1 AUTHOR
94
95	Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
96
97	=head1 COPYRIGHT & LICENSE
98
99	Copyright 2010 Dobrica Pavlinusic, All Rights Reserved.
100
101	This program is free software; you can redistribute it and/or modify it
102	under the same terms as Perl itself.
103
104	=cut
105
106	1; # End of WebPAC::Input::TSV