--- trunk/lib/WebPAC/Output/KinoSearch.pm	2006/04/17 16:50:53	434
+++ trunk/lib/WebPAC/Output/KinoSearch.pm	2006/08/25 16:20:21	620
@@ -9,6 +9,7 @@
 use KinoSearch::Analysis::PolyAnalyzer;
 use Encode qw/from_to/;
 use Data::Dumper;
+use Storable;
 
 =head1 NAME
 
@@ -16,11 +17,11 @@
 
 =head1 VERSION
 
-Version 0.01
+Version 0.03
 
 =cut
 
-our $VERSION = '0.01';
+our $VERSION = '0.03';
 
 =head1 SYNOPSIS
 
@@ -89,7 +90,11 @@
 
 	$self->{encoding} ||= 'ISO-8859-2';
 
-	$log->info("using index $self->{index_path} with encoding $self->{encoding}");
+	$self->{index_path} .= '/' . $self->{database};
+
+	$self->{clean} = 1 if (! -e $self->{index_path} . '/segments');
+
+	$log->info("using", $self->{clean} ? ' new' : '', " index $self->{index_path} with encoding $self->{encoding}");
 
 	my $analyzer = KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );
 
@@ -99,6 +104,18 @@
 		analyzer => $analyzer,
 	);
 
+	my $fields_path = $self->{index_path} . '/fields.storable';
+	$fields_path =~ s#//#/#g;
+	if (-e $fields_path) {
+		$self->{fields} = retrieve($fields_path) ||
+			$log->warn("can't open $fields_path: $!");
+	} else {
+		$log->error("This will be dummy run since no fields statistics are found!");
+		$log->error("You will have to re-run indexing to get search results!");
+		$self->{dummy_run} = 1;
+	}
+	$self->{fields_path} = $fields_path;
+
 	foreach my $f (@{ $self->{fields} }) {
 		$self->{invindex}->spec_field( 
 			name  => $f,
@@ -155,13 +172,18 @@
 
 	my $doc = $self->{invindex}->new_doc( $uri ) || $log->logdie("can't create new_doc( $uri )");
 
-	sub add_value($$$) {
-		my ($doc,$n,$v) = @_;
+	sub _add_value($$$$$) {
+		my ($self,$log,$doc,$n,$v) = @_;
+		return unless ($v);
+
+		$self->{value_usage}->{$n}++;
+		return if ($self->{dummy_run});
+
 		eval { $doc->set_value($n, $self->convert($v) ) };
 		$log->warn("can't insert: $n = $v") if ($@);
 	}
 
-	add_value($doc, 'uri', $uri);
+	_add_value($self,$log,$doc, 'uri', $uri);
 
 	$log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
 
@@ -183,11 +205,11 @@
 		$vals = $self->convert( $vals ) or
 			$log->logdie("can't convert '$vals' to UTF-8");
 
-		add_value($doc, $tag, $vals );
+		_add_value($self, $log, $doc, $tag, $vals );
 	}
 
 	if (my $text = $args->{'text'}) {
-		add_value($doc, 'bodytext', $text );
+		_add_value($self, $log, $doc, 'bodytext', $text );
 	}
 
 	#$log->debug("adding ", sub { $doc->dump_draft } );
@@ -207,8 +229,20 @@
 sub finish {
 	my $self = shift;
 
-	$self->_get_logger()->info("finish index writing to disk");
+	my $log = $self->_get_logger();
+
+	$log->info("finish index writing to disk");
 	$self->{invindex}->finish;
+
+	$log->info("writing value usage file");
+
+	# add fields from last run
+	map { $self->{value_usage}->{$_}++ } @{ $self->{fields} };
+
+	my @fields = keys %{ $self->{value_usage} };
+	store \@fields, $self->{fields_path} ||
+		$log->warn("can't write $self->{fields_path}: $!");
+
 }
 
 =head2 convert