--- trunk/run.pl	2006/02/26 23:21:50	416
+++ trunk/run.pl	2006/05/14 22:24:18	504
@@ -12,12 +12,13 @@
 use WebPAC::Input 0.03;
 use WebPAC::Store 0.03;
 use WebPAC::Normalize::XML;
+use WebPAC::Normalize::Set;
 use WebPAC::Output::TT;
-use WebPAC::Output::Estraier '0.10';
 use YAML qw/LoadFile/;
 use Getopt::Long;
 use File::Path;
 use Time::HiRes qw/time/;
+use File::Slurp;
 
 =head1 NAME
 
@@ -41,14 +42,19 @@
 
 remove database and Hyper Estraier index before indexing
 
-=item --one=database_name
+=item --only=database_name
 
-reindex just single database
+reindex just single database (legacy name is --one)
 
 =item --config conf/config.yml
 
 path to YAML configuration file
 
+=item --force-set
+
+force conversion C<normalize->path> in C<config.yml> from
+C<.xml> to C<.pl>
+
 =back
 
 =cut
@@ -59,15 +65,18 @@
 my $clean = 0;
 my $config = 'conf/config.yml';
 my $debug = 0;
-my $one_db_name;
+my $only_db_name;
+my $force_set = 0;
 
 GetOptions(
 	"limit=i" => \$limit,
 	"offset=i" => \$offset,
 	"clean" => \$clean,
-	"one=s" => \$one_db_name,
+	"one=s" => \$only_db_name,
+	"only=s" => \$only_db_name,
 	"config" => \$config,
 	"debug" => \$debug,
+	"force-set" => \$force_set,
 );
 
 $config = LoadFile($config);
@@ -76,25 +85,43 @@
 
 die "no databases in config file!\n" unless ($config->{databases});
 
+my $log = _new WebPAC::Common()->_get_logger();
+
+my $use_indexer = $config->{use_indexer} || 'hyperestraier';
+$log->info("using $use_indexer indexing engine...");
+
 my $total_rows = 0;
 my $start_t = time();
 
 while (my ($database, $db_config) = each %{ $config->{databases} }) {
 
-	next if ($one_db_name && $database !~ m/$one_db_name/i);
+	next if ($only_db_name && $database !~ m/$only_db_name/i);
 
-	my $log = _new WebPAC::Common()->_get_logger();
+	my $indexer;
 
-	#
-	# open Hyper Estraier database
-	#
+	my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
+	$indexer_config->{database} = $database;
+	$indexer_config->{clean} = $clean;
+	$indexer_config->{label} = $db_config->{name};
+
+	if ($use_indexer eq 'hyperestraier') {
+
+		# open Hyper Estraier database
+		use WebPAC::Output::Estraier '0.10';
+		$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
+	
+	} elsif ($use_indexer eq 'kinosearch') {
+
+		# open KinoSearch
+		use WebPAC::Output::KinoSearch;
+		$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
+		$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
 
-	my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration");
-	$est_config->{database} = $database;
-	$est_config->{clean} = $clean;
-	$est_config->{label} = $db_config->{name};
+	} else {
+		$log->logdie("unknown use_indexer: $use_indexer");
+	}
 
-	my $est = new WebPAC::Output::Estraier( %{ $est_config } );
+	$log->logide("can't continue without valid indexer") unless ($indexer);
 
 	#
 	# now WebPAC::Store
@@ -170,18 +197,34 @@
 			prefix => $input->{name},
 		);
 
+		my $rules;
 		my $normalize_path = $input->{normalize}->{path};
 
+		if ($force_set) {
+			my $new_norm_path = $normalize_path;
+			$new_norm_path =~ s/\.xml$/.pl/;
+			if (-e $new_norm_path) {
+				$log->debug("--force-set replaced $normalize_path with $new_norm_path");
+				$normalize_path = $new_norm_path;
+			} else {
+				$log->debug("--force-set failed on $new_norm_path, fallback to $normalize_path");
+			}
+		}
+
 		if ($normalize_path =~ m/\.xml$/i) {
 			$n->open(
 				tag => $input->{normalize}->{tag},
-				xml_file => $input->{normalize}->{path},
+				xml_file => $normalize_path,
 			);
 		} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) {
 			$n->open_yaml(
 				path => $normalize_path,
 				tag => $input->{normalize}->{tag},
 			);
+		} elsif ($normalize_path =~ m/\.(?:pl)$/i) {
+			$n = undef;
+			$log->info("using WebPAC::Normalize::Set to process $normalize_path");
+			$rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
 		}
 
 		foreach my $pos ( 0 ... $input_db->size ) {
@@ -196,12 +239,17 @@
 				push @{ $row->{'000'} }, $pos;
 			}
 
-			my $ds = $n->data_structure($row);
+			my $ds = $n ? $n->data_structure($row) :
+				WebPAC::Normalize::Set::data_structure(
+					row => $row,
+					rules => $rules,
+					lookup => $lookup->lookup_hash,
+				);
 
-			$est->add(
+			$indexer->add(
 				id => $input->{name} . "/" . $mfn,
 				ds => $ds,
-				type => $config->{hyperestraier}->{type},
+				type => $config->{$use_indexer}->{type},
 			);
 
 			$total_rows++;
@@ -209,6 +257,8 @@
 
 	};
 
+	eval { $indexer->finish } if ($indexer->can('finish'));
+
 	my $dt = time() - $start_t;
 	$log->info("$total_rows records indexed in " .
 		sprintf("%.2f sec [%.2f rec/sec]",
@@ -221,12 +271,16 @@
 	#
 	if (ref($db_config->{links}) eq 'ARRAY') {
 		foreach my $link (@{ $db_config->{links} }) {
-			$log->info("adding link $database -> $link->{to} [$link->{credit}]");
-			$est->add_link(
-				from => $database,
-				to => $link->{to},
-				credit => $link->{credit},
-			);
+			if ($use_indexer eq 'hyperestraier') {
+				$log->info("adding link $database -> $link->{to} [$link->{credit}]");
+				$indexer->add_link(
+					from => $database,
+					to => $link->{to},
+					credit => $link->{credit},
+				);
+			} else {
+				$log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
+			}
 		}
 	}