--- trunk/lib/WebPAC/Input/ISI.pm	2007/10/10 19:01:57	899
+++ trunk/lib/WebPAC/Input/ISI.pm	2009/05/30 14:21:58	1209
@@ -6,18 +6,15 @@
 use WebPAC::Input;
 use base qw/WebPAC::Common/;
 
+use Data::Dump qw/dump/;
+
 =head1 NAME
 
 WebPAC::Input::ISI - support for ISI Export Format
 
-=head1 VERSION
-
-Version 0.00
-
 =cut
 
-our $VERSION = '0.00';
-
+our $VERSION = '0.03';
 
 =head1 SYNOPSIS
 
@@ -54,6 +51,25 @@
 
 =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 = {@_};
@@ -88,6 +104,8 @@
 	my $tag;
 	my $rec;
 
+	$self->{size} = 0;
+
 	while( $line = <$fh> ) {
 		chomp($line);
 
@@ -98,7 +116,18 @@
 				$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/ ) {
+				$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
+			}
+			$rec->{'000'} = [ ++$self->{size} ];
 			push @{ $self->{_rec} }, $rec;
 			$rec = {};
 			$line = <$fh>;
@@ -110,7 +139,12 @@
 			$log->logdie("can't parse +$. $arg->{path} : $line");
 		}
 
-		push @{ $rec->{$tag} }, $v;
+		if ( defined $v ) {
+			$v = $subfields->{$tag}->($v) if defined $subfields->{$tag};
+
+			$log->debug("$tag: ", sub { dump( $v ) });
+			push @{ $rec->{$tag} }, $v;
+		}
 
 	}
 
@@ -146,9 +180,13 @@
 
 sub size {
 	my $self = shift;
-	return $#{$self->{_rec}} + 1;
+	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> >>