--- trunk/run.pl	2006/05/15 17:49:01	511
+++ trunk/run.pl	2006/05/22 19:34:45	529
@@ -14,6 +14,7 @@
 use WebPAC::Normalize::XML;
 use WebPAC::Normalize::Set;
 use WebPAC::Output::TT;
+use WebPAC::Validate;
 use YAML qw/LoadFile/;
 use Getopt::Long;
 use File::Path;
@@ -63,6 +64,10 @@
 disable indexing and dump statistics about field and subfield
 usage for each input
 
+=item --validate path/to/validation_file
+
+turn on extra validation of imput records, see L<WebPAC::Validation>
+
 =back
 
 =cut
@@ -76,6 +81,7 @@
 my $only_filter;
 my $force_set = 0;
 my $stats = 0;
+my $validate_path;
 
 GetOptions(
 	"limit=i" => \$limit,
@@ -87,6 +93,7 @@
 	"debug" => \$debug,
 	"force-set" => \$force_set,
 	"stats" => \$stats,
+	"validate=s" => \$validate_path,
 );
 
 $config = LoadFile($config);
@@ -98,6 +105,11 @@
 my $log = _new WebPAC::Common()->_get_logger();
 $log->info( "-" x 79 );
 
+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...");
@@ -109,13 +121,14 @@
 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#/#, $only_filter);
+	my ($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;
@@ -182,7 +195,7 @@
 
 	foreach my $input (@inputs) {
 
-		next if ($only_input && $input->{name} =~ m#$only_input#i || $input->{type} =~ m#$only_input#i);
+		next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));
 
 		my $type = lc($input->{type});
 
@@ -190,11 +203,13 @@
 
 		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,
@@ -210,12 +225,13 @@
 		my $maxmfn = $input_db->open(
 			path => $input->{path},
 			code_page => $input->{encoding},	# database encoding
+			%{ $input },
 		);
 
 		my $n = new WebPAC::Normalize::XML(
 		#	filter => { 'foo' => sub { shift } },
 			db => $db,
-			lookup_regex => $lookup->regex,
+			lookup_regex => $lookup ? $lookup->regex : undef,
 			lookup => $lookup,
 			prefix => $input->{name},
 		);
@@ -262,6 +278,12 @@
 				push @{ $row->{'000'} }, $pos;
 			}
 
+
+			if ($validate) {
+				my @errors = $validate->validate_errors( $row );
+				$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
+			}
+
 				
 			my $ds;
 			if ($n) {
@@ -270,7 +292,7 @@
 				$ds = WebPAC::Normalize::Set::data_structure(
 					row => $row,
 					rules => $rules,
-					lookup => $lookup->lookup_hash,
+					lookup => $lookup ? $lookup->lookup_hash : undef,
 				);
 
 				$db->save_ds(
@@ -308,12 +330,12 @@
 	if (ref($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}]");
 			}
@@ -322,3 +344,7 @@
 
 }
 
+foreach my $link (@links) {
+	$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
+	$indexer->add_link( %{ $link } );
+}