--- trunk/run.pl	2006/08/01 14:15:50	607
+++ trunk/run.pl	2006/09/06 19:51:32	638
@@ -8,7 +8,7 @@
 
 use WebPAC::Common 0.02;
 use WebPAC::Lookup 0.03;
-use WebPAC::Input 0.07;
+use WebPAC::Input 0.11;
 use WebPAC::Store 0.03;
 use WebPAC::Normalize 0.11;
 use WebPAC::Output::TT;
@@ -60,8 +60,8 @@
 
 =item --stats
 
-disable indexing and dump statistics about field and subfield
-usage for each input
+disable indexing, modify_* in configuration and dump statistics about field
+and subfield usage for each input
 
 =item --validate path/to/validation_file
 
@@ -93,6 +93,10 @@
 
 Create just links
 
+=item --merge
+
+Create merged index of databases which have links
+
 =back
 
 =cut
@@ -111,6 +115,17 @@
 my $marc_dump = 0;
 my $parallel = 0;
 my $only_links = 0;
+my $merge = 0;
+
+my $log = _new WebPAC::Common()->_get_logger();
+
+my $hostname = `hostname`;
+chomp($hostname);
+$hostname =~ s/\..+$//;
+if (-e "conf/$hostname.yml") {
+	$config = "conf/$hostname.yml";
+	$log->info("using host configuration file: $config");
+}
 
 GetOptions(
 	"limit=i" => \$limit,
@@ -128,22 +143,34 @@
 	"marc-dump!" => \$marc_dump,
 	"parallel=i" => \$parallel,
 	"only-links!" => \$only_links,
+	"merge" => \$merge,
 );
 
 $config = LoadFile($config);
 
-print "config = ",dump($config) if ($debug);
+#print "config = ",dump($config) if ($debug);
 
 die "no databases in config file!\n" unless ($config->{databases});
 
-my $log = _new WebPAC::Common()->_get_logger();
 $log->info( "-" x 79 );
 
+
+my $estcmd_fh;
+my $estcmd_path = './estcmd-merge.sh';
+if ($merge) {
+	open($estcmd_fh, '>', $estcmd_path) || $log->logdie("can't open $estcmd_path: $!");
+	print $estcmd_fh 'cd /data/estraier/_node/ || exit 1',$/;
+	print $estcmd_fh 'sudo /etc/init.d/hyperestraier stop',$/;
+	$log->info("created merge batch file $estcmd_path");
+}
+
+
 my $validate;
 $validate = new WebPAC::Validate(
 	path => $validate_path,
 ) if ($validate_path);
 
+
 my $use_indexer = $config->{use_indexer} || 'hyperestraier';
 if ($stats) {
 	$log->debug("option --stats disables update of indexing engine...");
@@ -181,17 +208,32 @@
 
 	my $indexer;
 	if ($use_indexer) {
-		my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
+
+		my $cfg_name = $use_indexer;
+		$cfg_name =~ s/\-.*$//;
+
+		my $indexer_config = $config->{$cfg_name} || $log->logdie("can't find '$cfg_name' part in confguration");
 		$indexer_config->{database} = $database;
 		$indexer_config->{clean} = $clean;
 		$indexer_config->{label} = $db_config->{name};
 
+		# force clean if database has links
+		$indexer_config->{clean} = 1 if ($db_config->{links});
+
 		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 'hyperestraier-native') {
+
+			# open Hyper Estraier database
+			use WebPAC::Output::EstraierNative;
+			$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } );
+
+			$use_indexer = 'hyperestraier';
+
 		} elsif ($use_indexer eq 'kinosearch') {
 
 			# open KinoSearch
@@ -213,15 +255,19 @@
 	if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
 		foreach my $link (@{ $db_config->{links} }) {
 			if ($use_indexer eq 'hyperestraier') {
-				$log->info("saving link $database -> $link->{to} [$link->{credit}]");
-				push @links, sub {
-					$log->info("adding link $database -> $link->{to} [$link->{credit}]");
-					$indexer->add_link(
-						from => $database,
-						to => $link->{to},
-						credit => $link->{credit},
-					);
-				};
+				if ($merge) {
+					print $estcmd_fh 'sudo -u www-data estcmd merge ' . $database . ' ' . $link->{to},$/;
+				} else {
+					$log->info("saving link $database -> $link->{to} [$link->{credit}]");
+					push @links, sub {
+						$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}]");
 			}
@@ -289,6 +335,12 @@
 			$input->{lookup} ? "lookup '$input->{lookup}'" : ""
 		);
 
+		if ($stats) {
+			# disable modification of records if --stats is in use
+			delete($input->{modify_records});
+			delete($input->{modify_file});
+		}
+
 		my $input_db = new WebPAC::Input(
 			module => $input_module,
 			encoding => $config->{webpac}->{webpac_encoding},
@@ -301,6 +353,7 @@
 			recode => $input->{recode},
 			stats => $stats,
 			modify_records => $input->{modify_records},
+			modify_file => $input->{modify_file},
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);
 
@@ -440,9 +493,18 @@
 	$log->info("all parallel processes finished");
 }
 
-
-foreach my $link (@links) {
-	$log->logdie("coderef in link ", Dumper($link), " is ", ref($link), " and not CODE") unless (ref($link) eq 'CODE');
-	$link->();
+#
+# handle links or merge after indexing
+#
+
+if ($merge) {
+	print $estcmd_fh 'sudo /etc/init.d/hyperestraier start',$/;
+	close($estcmd_fh);
+	chmod 0700, $estcmd_path || $log->warn("can't chmod 0700 $estcmd_path: $!");
+	system $estcmd_path;
+} else {
+	foreach my $link (@links) {
+		$log->logdie("coderef in link ", Dumper($link), " is ", ref($link), " and not CODE") unless (ref($link) eq 'CODE');
+		$link->();
+	}
 }
-