--- EPrints/index.pl	2007/06/29 18:46:45	13
+++ EPrints/index.pl	2007/07/02 12:55:49	18
@@ -5,11 +5,14 @@
 use Data::Dump qw/dump/;
 
 use EPrints qw/_x/;
-
-use lib '/home/dpavlin/stem-hr/';
-use StemHR;
+use KinoSearch::Simple;
 
 my $debug = shift @ARGV;
+my $use = {
+	score => 1,
+	stem => 1,
+	slogovi => 1,
+};
 
 my $dbh = EPrints->dbh;
 my $sth = $dbh->prepare(qq{
@@ -17,6 +20,8 @@
 	archive_title.eprintid as id,
 	title
 FROM archive_title 
+WHERE
+	lang = 'hr'
 }) || die $dbh->errstr();
 $sth->execute() || die $sth->errstr();
 
@@ -36,28 +41,67 @@
 
 my $total = 0;
 
+my $kino = KinoSearch::Simple->new(
+	path => 'kinoindex/',
+	language => 'ru',
+);
+
 while (my $row = $sth->fetchrow_hashref ) {
-	EPrints->id( $row->{id} );
-	my ( $title, $keywords, $abstract ) = (
-		_x( $row->{title} ),
-		EPrints->lookup( 'keywords' ),
-		EPrints->lookup( 'abstract' )
-	);
-	my @body = split( /\W*\s+\W*/, "$title $title $title $keywords $keywords $abstract" );
-	my $body = '';
-	foreach my $word ( @body ) {
-		$body .= StemHR->stem( $word ) . ' ';
+	my $id = $row->{id};
+	EPrints->id( $id );
+
+	my $parts = {
+		title => [ _x( $row->{title} ), 3 ],
+		keywords => [ EPrints->lookup( 'keywords' ), 2 ],
+		abstract => [ EPrints->lookup( 'abstract' ), 1 ],
+#		content => [ EPrints->fulltext_content, 1 ],
+	};
+
+	my $skip = 0;
+	foreach my $part ( qw/title keywords abstract/ ) {
+		if ( ! $parts->{$part}->[0] ) {
+			warn "skipped $id doesn't have required part $part\n";
+			$skip = 1;
+			last;
+		}
 	}
+	next if $skip;
+
+	my $body = '';
 
-	$body .= EPrints::slogovi( "$title $keywords $abstract" );
+	foreach my $part ( qw/title keywords abstract content/ ) {
+		my $content = $parts->{$part}->[0] || next;
 
-	warn "body: $body\n" if $debug;
+		if ( $use->{slogovi} ) {
+			$body .= ' ' . EPrints->slogovi( $content );
+		}
+
+		if ( $use->{stem} ) {
+			my $stem = EPrints->stem( $content );
+			warn "stem of '$content' didn't return anything\n" unless $stem;
+			$content = $stem;
+		}
+
+
+		if ( $use->{score} ) {
+			map { $body .= "$content " } 1 .. $parts->{$part}->[1];
+		} else {
+			$body .= "$content ";
+		}
 
-	$body .= EPrints->fulltext_content;
+		warn ">>> $body <<<\n" if $debug;
+	}
 
-	$indexer->index( $row->{id}, join(" ", @body, $body ) );
+	$indexer->index( $row->{id}, $body );
+	$kino->add_doc({
+		id => $id,
+		title => $parts->{title},
+		keywords => $parts->{keywords},
+		abstract => $parts->{abstract},
+	});
+	
 	$total++;
-	print STDERR _x( $row->{id}, " ", $row->{title} ), "\n";
+	print STDERR "$total: ", $row->{id}, " ", _x( $row->{title} ), " - ", length($body), " bytes\n";
 }