WebPAC/Input/Ovid.pm

package WebPAC::Input::Ovid;

use warnings;
use strict;

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

use Data::Dump qw/dump/;

=head1 NAME

WebPAC::Input::Ovid - support for Ovid citation export

=head1 VERSION

Version 0.01

=cut

our $VERSION = '0.01';

=head1 SYNOPSIS

Open file in Ovid citation export fromat

 my $input = new WebPAC::Input::Ovid(
        path => '/path/to/ovid-cites.txt',
 );

=head1 FUNCTIONS

=head2 new

Returns new low-level input API object

  my $input = new WebPAC::Input::Ovid(
        path => '/path/to/ebsco.txt',
        filter => sub {
                my ($l,$field_nr) = @_;
                # do something with $l which is line of input file
                return $l;
        },
  }

Options:

=over 4

=item path

path to Ovid export file

=back

=cut

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

        my $arg = {@_};

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

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

        $log->info("reading '$arg->{path}'");

        my $rec;
        my $size = 0;
        my $tag;

        $self->{_rec} = [];

        while( my $line = <$fh> ) {
                $line =~ s{[\r\n]+$}{};
                next if $line eq '';

                warn "<< $line\n";

                if ( $line =~ m/^<(\d+)>$/ ) {
                        push @{ $self->{_rec} }, $rec if $rec;
                        warn "## rec = ",dump( $rec ),$/;
                        my $expect_rec = $#{ $self->{_rec} } + 2;
                        warn "wrong Ovid record number: $1 != $expect_rec" unless $1 == $expect_rec;
                        $rec = { '000' => [ $1 ] };
                } elsif ( $line =~ /^(\w.+)/ ) {
                        $tag = $1;
                        warn "++ $tag\n";
                } elsif ( $line =~ /^\s\s(.+)/ ) {
                        my $v = $1;
                        $v =~ s{[\s\.]+$}{};
                        $rec->{$tag} = [ $v ];
                } else {
                        warn "### skip: '$line'\n";
                }

        }

        # save last rec
        push @{ $self->{_rec} }, $rec if $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 = shift;

        my ( $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->{_rec} } + 1;
}

=head1 AUTHOR

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

=head1 COPYRIGHT & LICENSE

Copyright 2008 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::Ovid
1	dpavlin	1124	package WebPAC::Input::Ovid;
2
3			use warnings;
4			use strict;
5
6			use lib 'lib';
7			use WebPAC::Input;
8			use base qw/WebPAC::Common/;
9
10			use Data::Dump qw/dump/;
11
12			=head1 NAME
13
14			WebPAC::Input::Ovid - support for Ovid citation export
15
16			=head1 VERSION
17
18			Version 0.01
19
20			=cut
21
22			our $VERSION = '0.01';
23
24			=head1 SYNOPSIS
25
26			Open file in Ovid citation export fromat
27
28			my $input = new WebPAC::Input::Ovid(
29			path => '/path/to/ovid-cites.txt',
30			);
31
32			=head1 FUNCTIONS
33
34			=head2 new
35
36			Returns new low-level input API object
37
38			my $input = new WebPAC::Input::Ovid(
39			path => '/path/to/ebsco.txt',
40			filter => sub {
41			my ($l,$field_nr) = @_;
42			# do something with $l which is line of input file
43			return $l;
44			},
45			}
46
47			Options:
48
49			=over 4
50
51			=item path
52
53			path to Ovid export file
54
55			=back
56
57			=cut
58
59			sub new {
60			my $class = shift;
61			my $self = {@_};
62			bless($self, $class);
63
64			my $arg = {@_};
65
66			my $log = $self->_get_logger();
67
68			open( my $fh, '<', $arg->{path} ) \|\| $log->logconfess("can't open $arg->{path}: $!");
69
70			$log->info("reading '$arg->{path}'");
71
72			my $rec;
73			my $size = 0;
74			my $tag;
75
76			$self->{_rec} = [];
77
78			while( my $line = <$fh> ) {
79			$line =~ s{[\r\n]+$}{};
80			next if $line eq '';
81
82			warn "<< $line\n";
83
84			if ( $line =~ m/^<(\d+)>$/ ) {
85	dpavlin	1126	push @{ $self->{_rec} }, $rec if $rec;
86			warn "## rec = ",dump( $rec ),$/;
87	dpavlin	1124	my $expect_rec = $#{ $self->{_rec} } + 2;
88			warn "wrong Ovid record number: $1 != $expect_rec" unless $1 == $expect_rec;
89	dpavlin	1126	$rec = { '000' => [ $1 ] };
90	dpavlin	1124	} elsif ( $line =~ /^(\w.+)/ ) {
91			$tag = $1;
92			warn "++ $tag\n";
93			} elsif ( $line =~ /^\s\s(.+)/ ) {
94			my $v = $1;
95			$v =~ s{[\s\.]+$}{};
96	dpavlin	1126	$rec->{$tag} = [ $v ];
97	dpavlin	1124	} else {
98			warn "### skip: '$line'\n";
99			}
100
101			}
102
103			# save last rec
104			push @{ $self->{_rec} }, $rec if $rec;
105
106			$log->debug("loaded ", $self->size, " records");
107
108			$self ? return $self : return undef;
109			}
110
111			=head2 fetch_rec
112
113			Return record with ID C<$mfn> from database
114
115			my $rec = $input->fetch_rec( $mfn, $filter_coderef );
116
117			=cut
118
119			sub fetch_rec {
120			my $self = shift;
121
122			my ( $mfn, $filter_coderef ) = @_;
123
124			return $self->{_rec}->[$mfn-1];
125			}
126
127
128			=head2 size
129
130			Return number of records in database
131
132			my $size = $input->size;
133
134			=cut
135
136			sub size {
137			my $self = shift;
138	dpavlin	1126	return $#{ $self->{_rec} } + 1;
139	dpavlin	1124	}
140
141			=head1 AUTHOR
142
143			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
144
145			=head1 COPYRIGHT & LICENSE
146
147			Copyright 2008 Dobrica Pavlinusic, All Rights Reserved.
148
149			This program is free software; you can redistribute it and/or modify it
150			under the same terms as Perl itself.
151
152			=cut
153
154			1; # End of WebPAC::Input::Ovid