--- trunk/run.pl	2005/12/18 23:34:24	291
+++ trunk/run.pl	2005/12/19 21:26:04	301
@@ -15,13 +15,57 @@
 use WebPAC::Output::TT;
 use WebPAC::Output::Estraier 0.05;
 use YAML qw/LoadFile/;
-use LWP::Simple;
+use Getopt::Long;
+use File::Path;
 
-my $limit = shift @ARGV;
+=head1 NAME
 
-my $config = LoadFile('conf/config.yml');
+run.pl - start WebPAC indexing
 
-print "config = ",Dumper($config);
+B<this command will probably go away. Don't get used to it!>
+
+Options:
+
+=over 4
+
+=item --offset 42
+
+start loading (all) databases at offset 42
+
+=item --limit 100
+
+limit loading to 100 records
+
+=item --clean
+
+remove database and Hyper Estraier index before indexing
+
+=item --config conf/config.yml
+
+path to YAML configuration file
+
+=back
+
+=cut
+
+my $offset;
+my $limit;
+
+my $clean = 0;
+my $config = 'conf/config.yml';
+my $debug = 0;
+
+GetOptions(
+	"limit=i" => \$limit,
+	"offset=i" => \$offset,
+	"clean" => \$clean,
+	"config" => \$config,
+	"debug" => \$debug,
+);
+
+$config = LoadFile($config);
+
+print "config = ",Dumper($config) if ($debug);
 
 die "no databases in config file!\n" unless ($config->{databases});
 
@@ -38,12 +82,16 @@
 	my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration");
 	$est_config->{database} = $database;
 
-	$log->info("using HyperEstraier URL $est_config->{masterurl}");
-
 	my $est = new WebPAC::Output::Estraier(
 		%{ $est_config },
 	);
 
+	if ($clean) {
+		$log->warn("creating new empty index $database");
+		$est->master( action => 'nodedel', name => $database );
+		$est->master( action => 'nodeadd', name => $database, label => $database );
+	}
+
 	#
 	# now WebPAC::Store
 	#
@@ -52,12 +100,17 @@
 
 	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
 
-	$log->info("working on $database in $db_path");
+	if ($clean) {
+		$log->info("creating new database $database in $db_path");
+		rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
+	} else {
+		$log->info("working on $database in $db_path");
+	}
 
 	my $db = new WebPAC::Store(
 		path => $db_path,
 		database => $database,
-		debug => 1,
+		debug => $debug,
 	);
 
 
@@ -93,7 +146,8 @@
 		my $input_db = new WebPAC::Input(
 			module => $input_module,
 			code_page => $config->{webpac}->{webpac_encoding},
-			limit => $input->{limit},
+			limit => $limit || $input->{limit},
+			offset => $offset,
 			lookup => $lookup,
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);