--- trunk/run.pl	2006/08/01 13:59:47	606
+++ trunk/run.pl	2006/08/23 10:04:08	611
@@ -89,6 +89,14 @@
 Run databases in parallel (aproximatly same as number of processors in
 machine if you want to use full load)
 
+=item --only-links
+
+Create just links
+
+=item --merge
+
+Create merged index of databases which have links
+
 =back
 
 =cut
@@ -105,8 +113,19 @@
 my ($marc_normalize, $marc_output);
 my $marc_lint = 1;
 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,
@@ -123,22 +142,35 @@
 	"marc-lint!" => \$marc_lint,
 	"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...");
@@ -154,7 +186,6 @@
 my $start_t = time();
 
 my @links;
-my $indexer;
 
 if ($parallel) {
 	$log->info("Using $parallel processes for speedup");
@@ -175,12 +206,16 @@
 		}
 	}
 
+	my $indexer;
 	if ($use_indexer) {
 		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};
 
+		# force clean if database has links
+		$indexer_config->{clean} = 1 if ($db_config->{links});
+
 		if ($use_indexer eq 'hyperestraier') {
 
 			# open Hyper Estraier database
@@ -203,6 +238,33 @@
 
 
 	#
+	# store Hyper Estraier links to other databases
+	#
+	if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
+		foreach my $link (@{ $db_config->{links} }) {
+			if ($use_indexer eq 'hyperestraier') {
+				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}]");
+			}
+		}
+	}
+	next if ($only_links);
+
+
+	#
 	# now WebPAC::Store
 	#
 	my $abs_path = abs_path($0);
@@ -397,23 +459,6 @@
 		)
 	);
 
-	#
-	# add Hyper Estraier links to other databases
-	#
-	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, {
-					from => $database,
-					to => $link->{to},
-					credit => $link->{credit},
-				};
-			} else {
-				$log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
-			}
-		}
-	}
 
 	# end forked process
 	if ($parallel) {
@@ -429,8 +474,18 @@
 	$log->info("all parallel processes finished");
 }
 
-foreach my $link (@links) {
-	$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
-	$indexer->add_link( %{ $link } );
+#
+# handle links or merge after indexing
+#
+
+if ($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->();
+	}
 }
-