--- trunk/run.pl	2006/08/23 10:04:08	611
+++ trunk/run.pl	2006/09/13 17:32:49	675
@@ -8,11 +8,11 @@
 
 use WebPAC::Common 0.02;
 use WebPAC::Lookup 0.03;
-use WebPAC::Input 0.07;
+use WebPAC::Input 0.11;
 use WebPAC::Store 0.03;
 use WebPAC::Normalize 0.11;
 use WebPAC::Output::TT;
-use WebPAC::Validate;
+use WebPAC::Validate 0.06;
 use WebPAC::Output::MARC;
 use YAML qw/LoadFile/;
 use Getopt::Long;
@@ -60,8 +60,8 @@
 
 =item --stats
 
-disable indexing and dump statistics about field and subfield
-usage for each input
+disable indexing, modify_* in configuration and dump statistics about field
+and subfield usage for each input
 
 =item --validate path/to/validation_file
 
@@ -208,7 +208,11 @@
 
 	my $indexer;
 	if ($use_indexer) {
-		my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
+
+		my $cfg_name = $use_indexer;
+		$cfg_name =~ s/\-.*$//;
+
+		my $indexer_config = $config->{$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};
@@ -222,6 +226,12 @@
 			use WebPAC::Output::Estraier '0.10';
 			$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
 		
+		} elsif ($use_indexer eq 'hyperestraier-native') {
+
+			# open Hyper Estraier database
+			use WebPAC::Output::EstraierNative;
+			$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } );
+
 		} elsif ($use_indexer eq 'kinosearch') {
 
 			# open KinoSearch
@@ -323,6 +333,12 @@
 			$input->{lookup} ? "lookup '$input->{lookup}'" : ""
 		);
 
+		if ($stats) {
+			# disable modification of records if --stats is in use
+			delete($input->{modify_records});
+			delete($input->{modify_file});
+		}
+
 		my $input_db = new WebPAC::Input(
 			module => $input_module,
 			encoding => $config->{webpac}->{webpac_encoding},
@@ -335,6 +351,7 @@
 			recode => $input->{recode},
 			stats => $stats,
 			modify_records => $input->{modify_records},
+			modify_file => $input->{modify_file},
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);
 
@@ -344,6 +361,16 @@
 			%{ $input },
 		);
 
+		my $report_fh;
+		if ($stats || $validate) {
+			my $path = "out/report/" . $database . '-' . $input->{name} . '.txt';
+			open($report_fh, '>', $path) || $log->logdie("can't open $path: $!");
+
+			print $report_fh "Report for database '$database' input '$input->{name}' records ",
+				$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n";
+			$log->info("Generating report file $path");
+		}
+
 		my @norm_array = ref($input->{normalize}) eq 'ARRAY' ?
 			@{ $input->{normalize} } : ( $input->{normalize} );
 
@@ -373,6 +400,10 @@
 			# 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;
@@ -387,8 +418,11 @@
 
 
 				if ($validate) {
-					my @errors = $validate->validate_errors( $row );
-					$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
+					if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) {
+						$log->error( "MFN $mfn validation error:\n",
+							$validate->report_error( $errors )
+						);
+					}
 				}
 
 				my $ds_config = dclone($db_config);
@@ -419,7 +453,7 @@
 				$indexer->add(
 					id => $input->{name} . "/" . $mfn,
 					ds => $ds,
-					type => $config->{$use_indexer}->{type},
+					type => $config->{$indexer_config}->{type},
 				) if ($indexer && $ds);
 
 				if ($marc) {
@@ -441,11 +475,25 @@
 				$total_rows++;
 			}
 
-			$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats);
+			if ($validate) {
+				my $errors = $validate->report;
+				if ($errors) {
+					$log->info("validation errors:\n$errors\n" );
+					print $report_fh "$errors\n" if ($report_fh);
+				}
+			}
+
+			if ($stats) {
+				my $s = $input_db->stats;
+				$log->info("statistics of fields usage:\n$s");
+				print $report_fh "Statistics of fields usage:\n$s" if ($report_fh);
+			}
 
 			# close MARC file
 			$marc->finish if ($marc);
 
+			# close report
+			close($report_fh) if ($report_fh)
 		}
 
 	}