--- trunk/run.pl	2006/05/16 15:23:12	516
+++ trunk/run.pl	2006/06/26 16:39:51	536
@@ -11,8 +11,7 @@
 use WebPAC::Lookup;
 use WebPAC::Input 0.03;
 use WebPAC::Store 0.03;
-use WebPAC::Normalize::XML;
-use WebPAC::Normalize::Set;
+use WebPAC::Normalize;
 use WebPAC::Output::TT;
 use WebPAC::Validate;
 use YAML qw/LoadFile/;
@@ -54,11 +53,6 @@
 
 path to YAML configuration file
 
-=item --force-set
-
-force conversion C<< normalize->path >> in C<config.yml> from
-C<.xml> to C<.pl>
-
 =item --stats
 
 disable indexing and dump statistics about field and subfield
@@ -79,7 +73,6 @@
 my $config = 'conf/config.yml';
 my $debug = 0;
 my $only_filter;
-my $force_set = 0;
 my $stats = 0;
 my $validate_path;
 
@@ -91,7 +84,6 @@
 	"only=s" => \$only_filter,
 	"config" => \$config,
 	"debug" => \$debug,
-	"force-set" => \$force_set,
 	"stats" => \$stats,
 	"validate=s" => \$validate_path,
 );
@@ -121,13 +113,14 @@
 my $total_rows = 0;
 my $start_t = time();
 
+my @links;
+my $indexer;
+
 while (my ($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);
 
-	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;
@@ -202,11 +195,13 @@
 
 		my $lookup = new WebPAC::Lookup(
 			lookup_file => $input->{lookup},
-		);
+		) if ($input->{lookup});
 
 		my $input_module = $config->{webpac}->{inputs}->{$type};
 
-		$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module lookup '$input->{lookup}'");
+		$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module",
+			$input->{lookup} ? "lookup '$input->{lookup}'" : ""
+		);
 
 		my $input_db = new WebPAC::Input(
 			module => $input_module,
@@ -222,45 +217,15 @@
 		my $maxmfn = $input_db->open(
 			path => $input->{path},
 			code_page => $input->{encoding},	# database encoding
-		);
-
-		my $n = new WebPAC::Normalize::XML(
-		#	filter => { 'foo' => sub { shift } },
-			db => $db,
-			lookup_regex => $lookup->regex,
-			lookup => $lookup,
-			prefix => $input->{name},
+			%{ $input },
 		);
 
 		my $rules;
 		my $normalize_path = $input->{normalize}->{path};
 
-		if ($force_set) {
-			my $new_norm_path = $normalize_path;
-			$new_norm_path =~ s/\.xml$/.pl/;
-			if (-e $new_norm_path) {
-				$log->debug("--force-set replaced $normalize_path with $new_norm_path");
-				$normalize_path = $new_norm_path;
-			} else {
-				$log->debug("--force-set failed on $new_norm_path, fallback to $normalize_path");
-			}
-		}
+		$log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i );
 
-		if ($normalize_path =~ m/\.xml$/i) {
-			$n->open(
-				tag => $input->{normalize}->{tag},
-				xml_file => $normalize_path,
-			);
-		} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) {
-			$n->open_yaml(
-				path => $normalize_path,
-				tag => $input->{normalize}->{tag},
-			);
-		} elsif ($normalize_path =~ m/\.(?:pl)$/i) {
-			$n = undef;
-			$log->info("using WebPAC::Normalize::Set to process $normalize_path");
-			$rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
-		}
+		my $rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
 
 		foreach my $pos ( 0 ... $input_db->size ) {
 
@@ -281,22 +246,17 @@
 			}
 
 				
-			my $ds;
-			if ($n) {
-				$ds = $n->data_structure($row);
-			} else {
-				$ds = WebPAC::Normalize::Set::data_structure(
-					row => $row,
-					rules => $rules,
-					lookup => $lookup->lookup_hash,
-				);
-
-				$db->save_ds(
-					id => $mfn,
-					ds => $ds,
-					prefix => $input->{name},
-				) if ($ds && !$stats);
-			}
+			my $ds = WebPAC::Normalize::data_structure(
+				row => $row,
+				rules => $rules,
+				lookup => $lookup ? $lookup->lookup_hash : undef,
+			);
+
+			$db->save_ds(
+				id => $mfn,
+				ds => $ds,
+				prefix => $input->{name},
+			) if ($ds && !$stats);
 
 			$indexer->add(
 				id => $input->{name} . "/" . $mfn,
@@ -326,12 +286,12 @@
 	if (ref($db_config->{links}) eq 'ARRAY') {
 		foreach my $link (@{ $db_config->{links} }) {
 			if ($use_indexer eq 'hyperestraier') {
-				$log->info("adding link $database -> $link->{to} [$link->{credit}]");
-				$indexer->add_link(
+				$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}]");
 			}
@@ -340,3 +300,7 @@
 
 }
 
+foreach my $link (@links) {
+	$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
+	$indexer->add_link( %{ $link } );
+}