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';
our $debug = 0;

=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 $debug;

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

        }

        # 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	dpavlin	1130	our $debug = 0;
24	dpavlin	1124
25			=head1 SYNOPSIS
26
27			Open file in Ovid citation export fromat
28
29			my $input = new WebPAC::Input::Ovid(
30			path => '/path/to/ovid-cites.txt',
31			);
32
33			=head1 FUNCTIONS
34
35			=head2 new
36
37			Returns new low-level input API object
38
39			my $input = new WebPAC::Input::Ovid(
40			path => '/path/to/ebsco.txt',
41			filter => sub {
42			my ($l,$field_nr) = @_;
43			# do something with $l which is line of input file
44			return $l;
45			},
46			}
47
48			Options:
49
50			=over 4
51
52			=item path
53
54			path to Ovid export file
55
56			=back
57
58			=cut
59
60			sub new {
61			my $class = shift;
62			my $self = {@_};
63			bless($self, $class);
64
65			my $arg = {@_};
66
67			my $log = $self->_get_logger();
68
69			open( my $fh, '<', $arg->{path} ) \|\| $log->logconfess("can't open $arg->{path}: $!");
70
71			$log->info("reading '$arg->{path}'");
72
73			my $rec;
74			my $size = 0;
75			my $tag;
76
77			$self->{_rec} = [];
78
79			while( my $line = <$fh> ) {
80			$line =~ s{[\r\n]+$}{};
81			next if $line eq '';
82
83	dpavlin	1130	warn "<< $line\n" if $debug;
84	dpavlin	1124
85			if ( $line =~ m/^<(\d+)>$/ ) {
86	dpavlin	1126	push @{ $self->{_rec} }, $rec if $rec;
87	dpavlin	1130	warn "## rec = ",dump( $rec ),$/ if $debug;
88	dpavlin	1124	my $expect_rec = $#{ $self->{_rec} } + 2;
89	dpavlin	1130	warn "wrong Ovid record number: $1 != $expect_rec" if $debug && $1 != $expect_rec;
90	dpavlin	1126	$rec = { '000' => [ $1 ] };
91	dpavlin	1124	} elsif ( $line =~ /^(\w.+)/ ) {
92			$tag = $1;
93	dpavlin	1130	warn "++ $tag\n" if $debug;
94	dpavlin	1124	} elsif ( $line =~ /^\s\s(.+)/ ) {
95			my $v = $1;
96			$v =~ s{[\s\.]+$}{};
97	dpavlin	1126	$rec->{$tag} = [ $v ];
98	dpavlin	1124	} else {
99	dpavlin	1130	warn "### skip: '$line'\n" if $debug;
100	dpavlin	1124	}
101
102			}
103
104			# save last rec
105			push @{ $self->{_rec} }, $rec if $rec;
106
107			$log->debug("loaded ", $self->size, " records");
108
109			$self ? return $self : return undef;
110			}
111
112			=head2 fetch_rec
113
114			Return record with ID C<$mfn> from database
115
116			my $rec = $input->fetch_rec( $mfn, $filter_coderef );
117
118			=cut
119
120			sub fetch_rec {
121			my $self = shift;
122
123			my ( $mfn, $filter_coderef ) = @_;
124
125			return $self->{_rec}->[$mfn-1];
126			}
127
128
129			=head2 size
130
131			Return number of records in database
132
133			my $size = $input->size;
134
135			=cut
136
137			sub size {
138			my $self = shift;
139	dpavlin	1126	return $#{ $self->{_rec} } + 1;
140	dpavlin	1124	}
141
142			=head1 AUTHOR
143
144			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
145
146			=head1 COPYRIGHT & LICENSE
147
148			Copyright 2008 Dobrica Pavlinusic, All Rights Reserved.
149
150			This program is free software; you can redistribute it and/or modify it
151			under the same terms as Perl itself.
152
153			=cut
154
155			1; # End of WebPAC::Input::Ovid