--- trunk/run.pl	2006/05/14 22:24:18	504
+++ trunk/run.pl	2006/05/15 13:32:18	508
@@ -52,9 +52,13 @@
 
 =item --force-set
 
-force conversion C<normalize->path> in C<config.yml> from
+force conversion C<< normalize->path >> in C<config.yml> from
 C<.xml> to C<.pl>
 
+=item --stats
+
+dump statistics about used fields and subfields in each input
+
 =back
 
 =cut
@@ -67,6 +71,7 @@
 my $debug = 0;
 my $only_db_name;
 my $force_set = 0;
+my $stats = 0;
 
 GetOptions(
 	"limit=i" => \$limit,
@@ -77,6 +82,7 @@
 	"config" => \$config,
 	"debug" => \$debug,
 	"force-set" => \$force_set,
+	"stats" => \$stats,
 );
 
 $config = LoadFile($config);
@@ -181,6 +187,7 @@
 			offset => $offset,
 			lookup => $lookup,
 			recode => $input->{recode},
+			stats => $stats,
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);
 
@@ -239,13 +246,24 @@
 				push @{ $row->{'000'} }, $pos;
 			}
 
-			my $ds = $n ? $n->data_structure($row) :
-				WebPAC::Normalize::Set::data_structure(
+				
+			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);
+			}
+
 			$indexer->add(
 				id => $input->{name} . "/" . $mfn,
 				ds => $ds,
@@ -255,6 +273,8 @@
 			$total_rows++;
 		}
 
+		$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats);
+
 	};
 
 	eval { $indexer->finish } if ($indexer->can('finish'));