WebPAC/Input/ISI.pm

package WebPAC::Input::ISI;

use warnings;
use strict;

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

=head1 NAME

WebPAC::Input::ISI - support for ISI Export Format

=head1 VERSION

Version 0.01

=cut

our $VERSION = '0.01';


=head1 SYNOPSIS

Open file in ISI export fromat

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

=head1 FUNCTIONS

=head2 new

Returns new low-level input API object

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

Options:

=over 4

=item path

path to ISI export file

=back

=cut

my $subfields = {
        'CR' => sub {
                my @v = split(/, /, shift);
                my $f;
                foreach ( qw/author year reference volume page/ ) {
                        if ( my $tmp = shift @v ) {
                                $f->{$_} = $tmp;
                        }
                }
                if ( $f->{author} =~ /^\*(.+)/ ) {
                        delete $f->{author};
                        $f->{institution} = $1;
                }
                return $f;
        },
};

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}: $!");

        my ( $format, $version );

        my $line = <$fh>;
        chomp($line);
        if ( $line =~ /^FN\s(.+)$/) {
                $format = $1;
        } else {
                $log->logdie("first line of $arg->{path} has to be FN, but is: $line");
        }

        $line = <$fh>;
        chomp($line);
        if ( $line =~ /^VR\s(.+)$/) {
                $version = $1;
        } else {
                $log->logdie("second line of $arg->{path} has to be VN, but is: $line");
        }

        $log->info("opening $format $version database '$arg->{path}'");

        my $tag;
        my $rec;

        while( $line = <$fh> ) {
                chomp($line);

                my $v;

                if ( $line =~ /^(\S\S)\s(.+)$/ ) {
                                $tag = $1;
                                $v = $2;
                } elsif ( $line =~ /^\s{3}(.+)$/ ) {
                                $v = $1;
                } elsif ( $line eq 'ER' ) {
                        # join tags
                        foreach ( qw/AB DE ID TI/ ) {
                                $rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
                        }
                        push @{ $self->{_rec} }, $rec;
                        $rec = {};
                        $line = <$fh>;
                        chomp $line;
                        $log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' );
                } elsif ( $line eq 'EF' ) {
                        last;
                } else {
                        $log->logdie("can't parse +$. $arg->{path} : $line");
                }

                $v = $subfields->{$tag}->($v) if defined $subfields->{$tag};

                push @{ $rec->{$tag} }, $v;

        }

        $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 SEE ALSO

L<http://isibasic.com/help/helpprn.html> is only sane source of document format which Google could find...
 
=head1 AUTHOR

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

=head1 COPYRIGHT & LICENSE

Copyright 2007 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::ISI
1	package WebPAC::Input::ISI;
2
3	use warnings;
4	use strict;
5
6	use WebPAC::Input;
7	use base qw/WebPAC::Common/;
8
9	=head1 NAME
10
11	WebPAC::Input::ISI - support for ISI Export Format
12
13	=head1 VERSION
14
15	Version 0.01
16
17	=cut
18
19	our $VERSION = '0.01';
20
21
22	=head1 SYNOPSIS
23
24	Open file in ISI export fromat
25
26	my $input = new WebPAC::Input::ISI(
27	path => '/path/to/ISI/records.txt',
28	);
29
30	=head1 FUNCTIONS
31
32	=head2 new
33
34	Returns new low-level input API object
35
36	my $input = new WebPAC::Input::ISI(
37	path => '/path/to/ISI/records.txt'
38	filter => sub {
39	my ($l,$field_nr) = @_;
40	# do something with $l which is line of input file
41	return $l;
42	},
43	}
44
45	Options:
46
47	=over 4
48
49	=item path
50
51	path to ISI export file
52
53	=back
54
55	=cut
56
57	my $subfields = {
58	'CR' => sub {
59	my @v = split(/, /, shift);
60	my $f;
61	foreach ( qw/author year reference volume page/ ) {
62	if ( my $tmp = shift @v ) {
63	$f->{$_} = $tmp;
64	}
65	}
66	if ( $f->{author} =~ /^\*(.+)/ ) {
67	delete $f->{author};
68	$f->{institution} = $1;
69	}
70	return $f;
71	},
72	};
73
74	sub new {
75	my $class = shift;
76	my $self = {@_};
77	bless($self, $class);
78
79	my $arg = {@_};
80
81	my $log = $self->_get_logger();
82
83	open( my $fh, '<', $arg->{path} ) \|\| $log->logconfess("can't open $arg->{path}: $!");
84
85	my ( $format, $version );
86
87	my $line = <$fh>;
88	chomp($line);
89	if ( $line =~ /^FN\s(.+)$/) {
90	$format = $1;
91	} else {
92	$log->logdie("first line of $arg->{path} has to be FN, but is: $line");
93	}
94
95	$line = <$fh>;
96	chomp($line);
97	if ( $line =~ /^VR\s(.+)$/) {
98	$version = $1;
99	} else {
100	$log->logdie("second line of $arg->{path} has to be VN, but is: $line");
101	}
102
103	$log->info("opening $format $version database '$arg->{path}'");
104
105	my $tag;
106	my $rec;
107
108	while( $line = <$fh> ) {
109	chomp($line);
110
111	my $v;
112
113	if ( $line =~ /^(\S\S)\s(.+)$/ ) {
114	$tag = $1;
115	$v = $2;
116	} elsif ( $line =~ /^\s{3}(.+)$/ ) {
117	$v = $1;
118	} elsif ( $line eq 'ER' ) {
119	# join tags
120	foreach ( qw/AB DE ID TI/ ) {
121	$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
122	}
123	push @{ $self->{_rec} }, $rec;
124	$rec = {};
125	$line = <$fh>;
126	chomp $line;
127	$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' );
128	} elsif ( $line eq 'EF' ) {
129	last;
130	} else {
131	$log->logdie("can't parse +$. $arg->{path} : $line");
132	}
133
134	$v = $subfields->{$tag}->($v) if defined $subfields->{$tag};
135
136	push @{ $rec->{$tag} }, $v;
137
138	}
139
140	$log->debug("loaded ", $self->size, " records");
141
142	$self ? return $self : return undef;
143	}
144
145	=head2 fetch_rec
146
147	Return record with ID C<$mfn> from database
148
149	my $rec = $input->fetch_rec( $mfn, $filter_coderef );
150
151	=cut
152
153	sub fetch_rec {
154	my $self = shift;
155
156	my ( $mfn, $filter_coderef ) = @_;
157
158	return $self->{_rec}->[$mfn-1];
159	}
160
161
162	=head2 size
163
164	Return number of records in database
165
166	my $size = $input->size;
167
168	=cut
169
170	sub size {
171	my $self = shift;
172	return $#{$self->{_rec}} + 1;
173	}
174
175	=head1 SEE ALSO
176
177	L<http://isibasic.com/help/helpprn.html> is only sane source of document format which Google could find...
178
179	=head1 AUTHOR
180
181	Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
182
183	=head1 COPYRIGHT & LICENSE
184
185	Copyright 2007 Dobrica Pavlinusic, All Rights Reserved.
186
187	This program is free software; you can redistribute it and/or modify it
188	under the same terms as Perl itself.
189
190	=cut
191
192	1; # End of WebPAC::Input::ISI