6 |
use WebPAC::Input; |
use WebPAC::Input; |
7 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
8 |
|
|
9 |
|
use Data::Dump qw/dump/; |
10 |
|
|
11 |
=head1 NAME |
=head1 NAME |
12 |
|
|
13 |
WebPAC::Input::ISI - support for ISI Export Format |
WebPAC::Input::ISI - support for ISI Export Format |
14 |
|
|
|
=head1 VERSION |
|
|
|
|
|
Version 0.01 |
|
|
|
|
15 |
=cut |
=cut |
16 |
|
|
17 |
our $VERSION = '0.01'; |
our $VERSION = '0.03'; |
|
|
|
18 |
|
|
19 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
20 |
|
|
53 |
|
|
54 |
my $subfields = { |
my $subfields = { |
55 |
'CR' => sub { |
'CR' => sub { |
56 |
my @v = split(/, /, shift); |
my $full_cr = shift; |
57 |
my $f; |
my @v = split(/, /, $full_cr); |
58 |
foreach ( qw/author year reference volume page/ ) { |
my $f = { full => $full_cr }; |
59 |
|
foreach ( qw/author year reference volume page doi/ ) { |
60 |
if ( my $tmp = shift @v ) { |
if ( my $tmp = shift @v ) { |
61 |
$f->{$_} = $tmp; |
$f->{$_} = $tmp; |
62 |
} |
} |
65 |
delete $f->{author}; |
delete $f->{author}; |
66 |
$f->{institution} = $1; |
$f->{institution} = $1; |
67 |
} |
} |
68 |
|
$f->{doi} =~ s{DOI\s+}{} if $f->{doi}; # strip DOI prefix |
69 |
return $f; |
return $f; |
70 |
}, |
}, |
71 |
}; |
}; |
104 |
my $tag; |
my $tag; |
105 |
my $rec; |
my $rec; |
106 |
|
|
107 |
|
my $offset = $self->{offset} || 0; |
108 |
|
my $limit = $self->{limit} || 0; |
109 |
|
|
110 |
|
my $file_pos = 0; |
111 |
|
my $end_pos = 0; |
112 |
|
$end_pos = $offset + $limit if $limit; |
113 |
|
|
114 |
|
$self->{_rec} = []; |
115 |
|
|
116 |
|
warn "# offset: $offset limit: $limit end: $end_pos"; |
117 |
|
|
118 |
while( $line = <$fh> ) { |
while( $line = <$fh> ) { |
119 |
chomp($line); |
chomp($line); |
|
|
|
120 |
my $v; |
my $v; |
121 |
|
|
122 |
if ( $line =~ /^(\S\S)\s(.+)$/ ) { |
if ( $line eq 'EF' ) { |
123 |
$tag = $1; |
last; |
|
$v = $2; |
|
|
} elsif ( $line =~ /^\s{3}(.+)$/ ) { |
|
|
$v = $1; |
|
124 |
} elsif ( $line eq 'ER' ) { |
} elsif ( $line eq 'ER' ) { |
125 |
# join tags |
$file_pos++; |
126 |
foreach ( qw/AB/ ) { |
last if $end_pos && $file_pos > $end_pos; |
127 |
$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_}; |
|
128 |
|
if ( ! $offset || $file_pos > $offset ) { |
129 |
|
|
130 |
|
# join tags |
131 |
|
foreach ( qw/AB DE ID TI SO RP SC FU FX PA JI/ ) { |
132 |
|
$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_}; |
133 |
|
} |
134 |
|
# split on ; |
135 |
|
foreach ( qw/ID SC DE/ ) { |
136 |
|
$rec->{$_} = [ split(/;\s/, $rec->{$_}) ] if defined $rec->{$_}; |
137 |
|
} |
138 |
|
$rec->{'000'} = [ $file_pos ]; |
139 |
|
push @{ $self->{_rec} }, $rec; |
140 |
} |
} |
141 |
push @{ $self->{_rec} }, $rec; |
|
142 |
$rec = {}; |
$rec = {}; |
143 |
$line = <$fh>; |
$line = <$fh>; |
144 |
chomp $line; |
chomp $line; |
145 |
$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' ); |
$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' ); |
146 |
} elsif ( $line eq 'EF' ) { |
} elsif ( $offset && $file_pos < $offset ) { |
147 |
last; |
next; |
148 |
|
} elsif ( $line =~ /^(\S\S)\s(.+)$/ ) { |
149 |
|
$tag = $1; |
150 |
|
$v = $2; |
151 |
|
} elsif ( $line =~ /^\s{3}(.+)$/ ) { |
152 |
|
$v = $1; |
153 |
|
if ( $tag eq 'CR' && $v =~ m{DOI$} ) { |
154 |
|
my $doi = <$fh>; |
155 |
|
chomp($doi); |
156 |
|
$doi =~ s{^\s{3}}{ } || die "can't find DOI in: $doi"; |
157 |
|
$v .= $doi; |
158 |
|
} |
159 |
|
} elsif ( $line =~ m{^(\S\S)\s*$} ) { |
160 |
|
warn "# $arg->{path} +$. empty |$line|\n"; |
161 |
} else { |
} else { |
162 |
$log->logdie("can't parse +$. $arg->{path} : $line"); |
$log->logdie("can't parse +$. $arg->{path} |$line|"); |
163 |
} |
} |
164 |
|
|
165 |
$v = $subfields->{$tag}->($v) if defined $subfields->{$tag}; |
if ( defined $v ) { |
166 |
|
$v = $subfields->{$tag}->($v) if defined $subfields->{$tag}; |
167 |
|
|
168 |
push @{ $rec->{$tag} }, $v; |
$log->debug("$tag: ", sub { dump( $v ) }); |
169 |
|
push @{ $rec->{$tag} }, $v; |
170 |
|
} |
171 |
|
|
172 |
} |
} |
|
|
|
173 |
$log->debug("loaded ", $self->size, " records"); |
$log->debug("loaded ", $self->size, " records"); |
174 |
|
|
175 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
188 |
|
|
189 |
my ( $mfn, $filter_coderef ) = @_; |
my ( $mfn, $filter_coderef ) = @_; |
190 |
|
|
191 |
|
$mfn -= $self->{offset} if $self->{offset}; |
192 |
return $self->{_rec}->[$mfn-1]; |
return $self->{_rec}->[$mfn-1]; |
193 |
} |
} |
194 |
|
|
203 |
|
|
204 |
sub size { |
sub size { |
205 |
my $self = shift; |
my $self = shift; |
206 |
return $#{$self->{_rec}} + 1; |
$#{ $self->{_rec} } + 1; |
207 |
} |
} |
208 |
|
|
209 |
=head1 SEE ALSO |
=head1 SEE ALSO |