WebPAC/Input/ISI.pm

package WebPAC::Input::ISI;

use warnings;
use strict;

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

use Data::Dump qw/dump/;

=head1 NAME

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

=cut

our $VERSION = '0.03';

=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 $full_cr = shift;
                my @v = split(/, /, $full_cr);
                my $f = { full => $full_cr };
                foreach ( qw/author year reference volume page doi/ ) {
                        if ( my $tmp = shift @v ) {
                                $f->{$_} = $tmp;
                        }
                }
                if ( $f->{author} =~ /^\*(.+)/ ) {
                        delete $f->{author};
                        $f->{institution} = $1;
                }
                $f->{doi} =~ s{DOI\s+}{} if $f->{doi}; # strip DOI prefix
                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;

        $self->{size} = 0;
        my $max_size;
        $max_size = ( $self->{offset} || 0 ) + $self->{limit} if $self->{limit};

        warn "# max_size: $max_size";

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

                my $v;

                if ( $line =~ /^(\S\S)\s(.+)$/ ) {
                                $tag = $1;
                                $v = $2;
                } elsif ( $line =~ /^\s{3}(.+)$/ ) {
                                $v = $1;
                                if ( $tag eq 'CR' && $v =~ m{DOI$} ) {
                                        my $doi = <$fh>;
                                        chomp($doi);
                                        $doi =~ s{^\s{3}}{ } || die "can't find DOI in: $doi";
                                        $v .= $doi;
                                }
                } elsif ( $line eq 'ER' ) {
                        # join tags
                        foreach ( qw/AB DE ID TI SO RP SC FU FX PA JI/ ) {
                                $rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
                        }
                        # split on ;
                        foreach ( qw/ID SC DE/ ) {
                                $rec->{$_} = [ split(/;\s/, $rec->{$_}) ] if defined $rec->{$_};
                        }
                        $rec->{'000'} = [ ++$self->{size} ];
                        push @{ $self->{_rec} }, $rec;

                        last if $max_size && $self->{size} == $max_size;

                        $rec = {};
                        $line = <$fh>;
                        chomp $line;
                        $log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' );
                } elsif ( $line eq 'EF' ) {
                        last;
                } elsif ( $line =~ m{^(\S\S)\s*$} ) {
                        warn "# $arg->{path} +$. empty |$line|\n";
                } else {
                        $log->logdie("can't parse +$. $arg->{path} |$line|");
                }

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

                        $log->debug("$tag: ", sub { dump( $v ) });
                        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->{size};
}

=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	dpavlin	898	package WebPAC::Input::ISI;
2
3			use warnings;
4			use strict;
5
6			use WebPAC::Input;
7			use base qw/WebPAC::Common/;
8
9	dpavlin	904	use Data::Dump qw/dump/;
10
11	dpavlin	898	=head1 NAME
12
13			WebPAC::Input::ISI - support for ISI Export Format
14
15			=cut
16
17	dpavlin	1209	our $VERSION = '0.03';
18	dpavlin	898
19			=head1 SYNOPSIS
20
21			Open file in ISI export fromat
22
23			my $input = new WebPAC::Input::ISI(
24			path => '/path/to/ISI/records.txt',
25			);
26
27			=head1 FUNCTIONS
28
29			=head2 new
30
31			Returns new low-level input API object
32
33			my $input = new WebPAC::Input::ISI(
34			path => '/path/to/ISI/records.txt'
35			filter => sub {
36			my ($l,$field_nr) = @_;
37			# do something with $l which is line of input file
38			return $l;
39			},
40			}
41
42			Options:
43
44			=over 4
45
46			=item path
47
48			path to ISI export file
49
50			=back
51
52			=cut
53
54	dpavlin	901	my $subfields = {
55			'CR' => sub {
56	dpavlin	1194	my $full_cr = shift;
57			my @v = split(/, /, $full_cr);
58			my $f = { full => $full_cr };
59	dpavlin	1209	foreach ( qw/author year reference volume page doi/ ) {
60	dpavlin	901	if ( my $tmp = shift @v ) {
61			$f->{$_} = $tmp;
62			}
63			}
64			if ( $f->{author} =~ /^\*(.+)/ ) {
65			delete $f->{author};
66			$f->{institution} = $1;
67			}
68	dpavlin	1209	$f->{doi} =~ s{DOI\s+}{} if $f->{doi}; # strip DOI prefix
69	dpavlin	901	return $f;
70			},
71			};
72
73	dpavlin	898	sub new {
74			my $class = shift;
75			my $self = {@_};
76			bless($self, $class);
77
78			my $arg = {@_};
79
80			my $log = $self->_get_logger();
81
82			open( my $fh, '<', $arg->{path} ) \|\| $log->logconfess("can't open $arg->{path}: $!");
83
84			my ( $format, $version );
85
86			my $line = <$fh>;
87			chomp($line);
88			if ( $line =~ /^FN\s(.+)$/) {
89			$format = $1;
90			} else {
91			$log->logdie("first line of $arg->{path} has to be FN, but is: $line");
92			}
93
94			$line = <$fh>;
95			chomp($line);
96			if ( $line =~ /^VR\s(.+)$/) {
97			$version = $1;
98			} else {
99			$log->logdie("second line of $arg->{path} has to be VN, but is: $line");
100			}
101
102			$log->info("opening $format $version database '$arg->{path}'");
103
104	dpavlin	899	my $tag;
105			my $rec;
106
107	dpavlin	1077	$self->{size} = 0;
108	dpavlin	1287	my $max_size;
109			$max_size = ( $self->{offset} \|\| 0 ) + $self->{limit} if $self->{limit};
110	dpavlin	1077
111	dpavlin	1287	warn "# max_size: $max_size";
112
113	dpavlin	899	while( $line = <$fh> ) {
114			chomp($line);
115
116			my $v;
117
118			if ( $line =~ /^(\S\S)\s(.+)$/ ) {
119			$tag = $1;
120			$v = $2;
121			} elsif ( $line =~ /^\s{3}(.+)$/ ) {
122			$v = $1;
123	dpavlin	1209	if ( $tag eq 'CR' && $v =~ m{DOI$} ) {
124			my $doi = <$fh>;
125			chomp($doi);
126			$doi =~ s{^\s{3}}{ } \|\| die "can't find DOI in: $doi";
127			$v .= $doi;
128			}
129	dpavlin	899	} elsif ( $line eq 'ER' ) {
130	dpavlin	901	# join tags
131	dpavlin	1288	foreach ( qw/AB DE ID TI SO RP SC FU FX PA JI/ ) {
132	dpavlin	901	$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
133			}
134	dpavlin	1287	# split on ;
135			foreach ( qw/ID SC DE/ ) {
136			$rec->{$_} = [ split(/;\s/, $rec->{$_}) ] if defined $rec->{$_};
137			}
138	dpavlin	1077	$rec->{'000'} = [ ++$self->{size} ];
139	dpavlin	899	push @{ $self->{_rec} }, $rec;
140	dpavlin	1287
141			last if $max_size && $self->{size} == $max_size;
142
143	dpavlin	899	$rec = {};
144			$line = <$fh>;
145			chomp $line;
146			$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' );
147			} elsif ( $line eq 'EF' ) {
148			last;
149	dpavlin	1302	} elsif ( $line =~ m{^(\S\S)\s*$} ) {
150			warn "# $arg->{path} +$. empty \|$line\|\n";
151	dpavlin	899	} else {
152	dpavlin	1302	$log->logdie("can't parse +$. $arg->{path} \|$line\|");
153	dpavlin	899	}
154
155	dpavlin	904	if ( defined $v ) {
156			$v = $subfields->{$tag}->($v) if defined $subfields->{$tag};
157	dpavlin	901
158	dpavlin	904	$log->debug("$tag: ", sub { dump( $v ) });
159			push @{ $rec->{$tag} }, $v;
160			}
161	dpavlin	899
162			}
163
164			$log->debug("loaded ", $self->size, " records");
165
166	dpavlin	898	$self ? return $self : return undef;
167			}
168
169			=head2 fetch_rec
170
171			Return record with ID C<$mfn> from database
172
173	dpavlin	899	my $rec = $input->fetch_rec( $mfn, $filter_coderef );
174	dpavlin	898
175			=cut
176
177			sub fetch_rec {
178			my $self = shift;
179
180	dpavlin	899	my ( $mfn, $filter_coderef ) = @_;
181	dpavlin	898
182	dpavlin	899	return $self->{_rec}->[$mfn-1];
183	dpavlin	898	}
184
185
186			=head2 size
187
188			Return number of records in database
189
190			my $size = $input->size;
191
192			=cut
193
194			sub size {
195			my $self = shift;
196	dpavlin	1077	return $self->{size};
197	dpavlin	898	}
198
199	dpavlin	900	=head1 SEE ALSO
200
201			L<http://isibasic.com/help/helpprn.html> is only sane source of document format which Google could find...
202	dpavlin	901
203	dpavlin	898	=head1 AUTHOR
204
205			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
206
207			=head1 COPYRIGHT & LICENSE
208
209			Copyright 2007 Dobrica Pavlinusic, All Rights Reserved.
210
211			This program is free software; you can redistribute it and/or modify it
212			under the same terms as Perl itself.
213
214			=cut
215
216			1; # End of WebPAC::Input::ISI