--- trunk/run.pl	2006/09/11 20:26:01	672
+++ trunk/run.pl	2006/09/25 12:51:17	699
@@ -7,6 +7,7 @@
 use lib './lib';
 
 use WebPAC::Common 0.02;
+use WebPAC::Parser 0.04;
 use WebPAC::Lookup 0.03;
 use WebPAC::Input 0.11;
 use WebPAC::Store 0.03;
@@ -14,7 +15,7 @@
 use WebPAC::Output::TT;
 use WebPAC::Validate 0.06;
 use WebPAC::Output::MARC;
-use YAML qw/LoadFile/;
+use WebPAC::Config;
 use Getopt::Long;
 use File::Path;
 use Time::HiRes qw/time/;
@@ -105,7 +106,7 @@
 my $limit;
 
 my $clean = 0;
-my $config = 'conf/config.yml';
+my $config_path;
 my $debug = 0;
 my $only_filter;
 my $stats = 0;
@@ -119,21 +120,13 @@
 
 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,
 	"offset=i" => \$offset,
 	"clean" => \$clean,
 	"one=s" => \$only_filter,
 	"only=s" => \$only_filter,
-	"config" => \$config,
+	"config" => \$config_path,
 	"debug+" => \$debug,
 	"stats" => \$stats,
 	"validate=s" => \$validate_path,
@@ -146,11 +139,11 @@
 	"merge" => \$merge,
 );
 
-$config = LoadFile($config);
+my $config = new WebPAC::Config( path => $config_path );
 
 #print "config = ",dump($config) if ($debug);
 
-die "no databases in config file!\n" unless ($config->{databases});
+die "no databases in config file!\n" unless ($config->databases);
 
 $log->info( "-" x 79 );
 
@@ -171,7 +164,7 @@
 ) if ($validate_path);
 
 
-my $use_indexer = $config->{use_indexer} || 'hyperestraier';
+my $use_indexer = $config->use_indexer;
 if ($stats) {
 	$log->debug("option --stats disables update of indexing engine...");
 	$use_indexer = undef;
@@ -182,6 +175,10 @@
 # disable indexing when creating marc
 $use_indexer = undef if ($marc_normalize);
 
+# parse normalize files and create source files for lookup and normalization
+
+my $parser = new WebPAC::Parser( config => $config );
+
 my $total_rows = 0;
 my $start_t = time();
 
@@ -192,7 +189,7 @@
 	Proc::Queue::size($parallel);
 }
 
-while (my ($database, $db_config) = each %{ $config->{databases} }) {
+while (my ($database, $db_config) = each %{ $config->databases }) {
 
 	my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
 	next if ($only_database && $database !~ m/$only_database/i);
@@ -212,7 +209,7 @@
 		my $cfg_name = $use_indexer;
 		$cfg_name =~ s/\-.*$//;
 
-		my $indexer_config = $config->{$cfg_name} || $log->logdie("can't find '$cfg_name' part in confguration");
+		my $indexer_config = $config->get( $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};
@@ -232,8 +229,6 @@
 			use WebPAC::Output::EstraierNative;
 			$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } );
 
-			$use_indexer = 'hyperestraier';
-
 		} elsif ($use_indexer eq 'kinosearch') {
 
 			# open KinoSearch
@@ -282,7 +277,7 @@
 	my $abs_path = abs_path($0);
 	$abs_path =~ s#/[^/]*$#/#;
 
-	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
+	my $db_path = $config->get('webpac')->{db_path} . '/' . $database;
 
 	if ($clean) {
 		$log->info("creating new database '$database' in $db_path");
@@ -311,25 +306,15 @@
 		$log->info("database $database doesn't have inputs defined");
 	}
 
-	my @supported_inputs = keys %{ $config->{webpac}->{inputs} };
-
 	foreach my $input (@inputs) {
 
 		next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));
 
 		my $type = lc($input->{type});
 
-		die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
+		die "I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!\n" unless (grep(/$type/, $config->webpac('inputs')));
 
-		my $lookup;
-		if ($input->{lookup}) {
-			$lookup = new WebPAC::Lookup(
-				lookup_file => $input->{lookup},
-			);
-			delete( $input->{lookup} );
-		}
-
-		my $input_module = $config->{webpac}->{inputs}->{$type};
+		my $input_module = $config->webpac('inputs')->{$type};
 
 		$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module",
 			$input->{lookup} ? "lookup '$input->{lookup}'" : ""
@@ -341,9 +326,16 @@
 			delete($input->{modify_file});
 		}
 
+		warn "parser->depends = ", dump( $parser->{depends} );
+		warn "depends on: ", dump( $parser->depends($database, $input->{name}) );
+		warn "lookup_create_rules = ", dump( $parser->lookup_create_rules($database, $input->{name}) );
+		warn "parser->_lookup_create = ", dump( $parser->{_lookup_create} );
+
+		my $lookup;
+
 		my $input_db = new WebPAC::Input(
 			module => $input_module,
-			encoding => $config->{webpac}->{webpac_encoding},
+			encoding => $config->webpac('webpac_encoding'),
 			limit => $limit || $input->{limit},
 			offset => $offset,
 			lookup_coderef => sub {
@@ -402,6 +394,10 @@
 			# reset position in database
 			$input_db->seek(1);
 
+			# generate name of config key for indexer (strip everything after -)
+			my $indexer_config = $use_indexer;
+			$indexer_config =~ s/^(\w+)-?.*$/$1/g if ($indexer_config);
+
 			foreach my $pos ( 0 ... $input_db->size ) {
 
 				my $row = $input_db->fetch || next;
@@ -451,7 +447,7 @@
 				$indexer->add(
 					id => $input->{name} . "/" . $mfn,
 					ds => $ds,
-					type => $config->{$use_indexer}->{type},
+					type => $config->get($indexer_config)->{type},
 				) if ($indexer && $ds);
 
 				if ($marc) {