--- trunk/run.pl	2005/12/18 22:16:39	288
+++ trunk/run.pl	2006/03/22 00:18:56	423
@@ -13,22 +13,77 @@
 use WebPAC::Store 0.03;
 use WebPAC::Normalize::XML;
 use WebPAC::Output::TT;
-use WebPAC::Output::Estraier 0.05;
+use WebPAC::Output::Estraier '0.10';
 use YAML qw/LoadFile/;
-use LWP::Simple;
+use Getopt::Long;
+use File::Path;
+use Time::HiRes qw/time/;
 
-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 --only=database_name
+
+reindex just single database (legacy name is --one)
+
+=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;
+my $only_db_name;
+
+GetOptions(
+	"limit=i" => \$limit,
+	"offset=i" => \$offset,
+	"clean" => \$clean,
+	"one=s" => \$only_db_name,
+	"only=s" => \$only_db_name,
+	"config" => \$config,
+	"debug" => \$debug,
+);
+
+$config = LoadFile($config);
+
+print "config = ",Dumper($config) if ($debug);
 
 die "no databases in config file!\n" unless ($config->{databases});
 
 my $total_rows = 0;
+my $start_t = time();
 
 while (my ($database, $db_config) = each %{ $config->{databases} }) {
 
+	next if ($only_db_name && $database !~ m/$only_db_name/i);
+
 	my $log = _new WebPAC::Common()->_get_logger();
 
 	#
@@ -37,12 +92,10 @@
 
 	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};
 
-	$log->info("using HyperEstraier URL $est_config->{masterurl}");
-
-	my $est = new WebPAC::Output::Estraier(
-		%{ $est_config },
-	);
+	my $est = new WebPAC::Output::Estraier( %{ $est_config } );
 
 	#
 	# now WebPAC::Store
@@ -52,12 +105,17 @@
 
 	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->debug("working on $database in $db_path");
+	}
 
 	my $db = new WebPAC::Store(
 		path => $db_path,
 		database => $database,
-		debug => 1,
+		debug => $debug,
 	);
 
 
@@ -88,13 +146,15 @@
 
 		my $input_module = $config->{webpac}->{inputs}->{$type};
 
-		$log->info("working on input $input->{path} [$input->{type}] using $input_module");
+		$log->info("working on input '$input->{path}' [$input->{type}] using $input_module lookup '$input->{lookup}'");
 
 		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,
+			recode => $input->{recode},
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);
 
@@ -125,16 +185,22 @@
 			);
 		}
 
-		for ( 0 ... $input_db->size ) {
+		foreach my $pos ( 0 ... $input_db->size ) {
 
 			my $row = $input_db->fetch || next;
 
-			my $mfn = $row->{'000'}->[0] || die "can't find MFN";
+			my $mfn = $row->{'000'}->[0];
+
+			if (! $mfn || $mfn !~ m#^\d+$#) {
+				$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
+				$mfn = $pos;
+				push @{ $row->{'000'} }, $pos;
+			}
 
 			my $ds = $n->data_structure($row);
 
 			$est->add(
-				id => $input->{name} . "#" . $mfn,
+				id => $input->{name} . "/" . $mfn,
 				ds => $ds,
 				type => $config->{hyperestraier}->{type},
 			);
@@ -144,7 +210,12 @@
 
 	};
 
-	$log->info("$total_rows records indexed");
+	my $dt = time() - $start_t;
+	$log->info("$total_rows records indexed in " .
+		sprintf("%.2f sec [%.2f rec/sec]",
+			$dt, ($total_rows / $dt)
+		)
+	);
 
 	#
 	# add Hyper Estraier links to other databases