--- trunk/run.pl	2007/05/24 12:44:45	837
+++ trunk/run.pl	2007/11/02 14:28:43	969
@@ -9,7 +9,7 @@
 use WebPAC::Common 0.02;
 use WebPAC::Parser 0.08;
 use WebPAC::Input 0.16;
-use WebPAC::Store 0.14;
+use WebPAC::Store 0.15;
 use WebPAC::Normalize 0.22;
 use WebPAC::Output::TT;
 use WebPAC::Validate 0.11;
@@ -68,6 +68,9 @@
 
 turn on extra validation of imput records, see L<WebPAC::Validation>
 
+You can use special variables C<$database> and $C<$input> in this parametar
+like C<--validate 'conf/validate/$database-$input'> to construct filename
+
 =item --validate-delimiters path/to/validate_delimiters_file
 
 this option is used with C<--validate> to turn on extra validation of
@@ -130,7 +133,7 @@
 	"clean" => \$clean,
 	"one=s" => \$only_filter,
 	"only=s" => \$only_filter,
-	"config" => \$config_path,
+	"config=s" => \$config_path,
 	"debug+" => \$debug,
 	"stats" => \$stats,
 	"validate=s" => \$validate_path,
@@ -144,6 +147,8 @@
 	"help" => \$help,
 );
 
+$marc_generate = 0 if ( $validate_delimiters_path );
+
 pod2usage(-verbose => 2) if ($help);
 
 my $config = new WebPAC::Config( path => $config_path );
@@ -172,9 +177,7 @@
 
 my $validate;
 $validate = new WebPAC::Validate(
-	path => $validate_path,
 	delimiters => $config->webpac('delimiters'),
-	delimiters_path => $validate_delimiters_path,
 ) if ($validate_path || $validate_delimiters_path);
 
 my $use_indexer = $config->use_indexer;
@@ -182,7 +185,7 @@
 if ($stats) {
 	$log->debug("disabled indexing for stats collection");
 	$use_indexer = undef;
-} else {
+} elsif ( $use_indexer ) {
 	$log->info("using $use_indexer indexing engine...");
 }
 
@@ -209,7 +212,8 @@
 	return $c;
 }
 
-while (my ($database, $db_config) = each %{ $config->databases }) {
+foreach my $database ( sort keys %{ $config->databases } ) {
+	my $db_config = $config->databases->{$database};
 
 	my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
 	next if ($only_database && $database !~ m/$only_database/i);
@@ -240,27 +244,24 @@
 		if ($use_indexer eq 'hyperestraier') {
 
 			# open Hyper Estraier database
-			use WebPAC::Output::Estraier '0.10';
+			require WebPAC::Output::Estraier;
 			$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
 		
 		} elsif ($use_indexer eq 'hyperestraier-native') {
 
 			# open Hyper Estraier database
-			use WebPAC::Output::EstraierNative;
+			require WebPAC::Output::EstraierNative;
 			$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } );
 
 		} elsif ($use_indexer eq 'kinosearch') {
 
-			# open KinoSearch
-			use WebPAC::Output::KinoSearch;
-			$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
-			$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
+			die "no longer supported";
 
 		} else {
 			$log->logdie("unknown use_indexer: $use_indexer");
 		}
 
-		$log->logide("can't continue without valid indexer") unless ($indexer);
+		$log->logdie("can't continue without valid indexer") unless ($indexer);
 	}
 
 
@@ -294,36 +295,49 @@
 	#
 	# now WebPAC::Store
 	#
-	my $abs_path = abs_path($0);
-	$abs_path =~ s#/[^/]*$#/#;
+	my $store = new WebPAC::Store({
+		debug => $debug,
+	});
 
-	my $db_path = $config->webpac('db_path');
 
-	if ($clean) {
-		$log->info("creating new database '$database' in $db_path");
-		rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
-	} else {
-		$log->info("working on database '$database' in $db_path");
-	}
+	#
+	# prepare output
+	#
+	my @outputs = force_array( $db_config->{output}, sub {
+		$log->error("Database $database doesn't have any outputs defined. Do you want to remove it from configuration?" );
+	} );
 
-	my $store = new WebPAC::Store(
-		path => $db_path,
-		debug => $debug,
-	);
+	my @output_modules;
+
+	foreach my $output ( @outputs ) {
+
+#warn '## output = ',dump( $output );
+
+		my $module = $output->{module} || $log->logdie("need module in output section of $database");
+		$module = 'WebPAC::Output::' . $module unless $module =~ m/::/;
+	
+		$log->debug("loading output module $module");
+		eval "require $module";
+
+		# add database to arugemnts for output filter
+		$output->{database} = $database;
+
+		$log->debug("calling $module->new(",dump( $output ),")");
+		my $out = new $module->new( $output );
+		$out->init;
+
+		push @output_modules, $out;
+	}
 
 
 	#
 	# now, iterate through input formats
 	#
 
-	my @inputs;
-	if (ref($db_config->{input}) eq 'ARRAY') {
-		@inputs = @{ $db_config->{input} };
-	} elsif ($db_config->{input}) {
-		push @inputs, $db_config->{input};
-	} else {
+
+	my @inputs = force_array( $db_config->{input}, sub {
 		$log->info("database $database doesn't have inputs defined");
-	}
+	} );
 
 	foreach my $input (@inputs) {
 
@@ -333,9 +347,16 @@
 
 		my $type = lc($input->{type});
 
-		die "I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!\n" unless (grep(/$type/, $config->webpac('inputs')));
+		# FIXME check if input module exists
+		my $input_module = $input->{module};
 
-		my $input_module = $config->webpac('inputs')->{$type};
+		if ( ! $input_module ) {
+			if ( grep(/$type/, $config->webpac('inputs')) ) {
+				$input_module = $config->webpac('inputs')->{$type};
+			} else {
+				$log->logdie("I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!" );
+			}
+		}
 
 		my @lookups = $parser->have_lookup_create($database, $input);
 
@@ -358,6 +379,7 @@
 			stats => $stats,
 			modify_records => $input->{modify_records},
 			modify_file => $input->{modify_file},
+			input_config => $input,
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);
 
@@ -442,6 +464,11 @@
 			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");
+
+			if ( $validate ) {
+				$validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ) if ( $validate_path );
+				$validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ) if ( $validate_delimiters_path );
+			}
 		}
 
 		my $marc;
@@ -489,6 +516,16 @@
 		}
 
 
+		# setup input name for all output filters
+		foreach my $out ( @output_modules ) {
+			if ( $out->can('input') ) {
+				$out->input( $input_name );
+			} else {
+				$log->warn("output filter ",ref($out)," doesn't support input name");
+			}
+		}
+
+
 		foreach my $pos ( 0 ... $input_db->size ) {
 
 			my $row = $input_db->fetch || next;
@@ -497,7 +534,7 @@
 
 			my $mfn = $row->{'000'}->[0];
 
-			if (! $mfn || $mfn !~ m#^\d+$#) {
+			if (! $mfn || $mfn !~ m{^\d+$}) {
 				$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
 				$mfn = $pos;
 				push @{ $row->{'000'} }, $pos;
@@ -513,7 +550,7 @@
 				next;	# validation doesn't create any output
 			}
 
-			my $ds = WebPAC::Normalize::data_structure(
+			if ( my $ds = WebPAC::Normalize::data_structure(
 				row => $row,
 				rules => $rules,
 				lookup => $lookup_hash,
@@ -527,38 +564,47 @@
 						id => $mfn,
 					);
 				},
-			);
+			) ) { # if
 
-			$log->debug("ds = ", sub { dump($ds) }) if ($ds);
+				$log->debug("ds = ", sub { dump($ds) });
 
-			$store->save_ds(
-				database => $database,
-				input => $input_name,
-				id => $mfn,
-				ds => $ds,
-			) if ($ds && !$stats);
-
-			$indexer->add(
-				id => "${input_name}/${mfn}",
-				ds => $ds,
-				type => $config->get($indexer_config)->{type},
-			) if ($indexer && $ds);
-
-			if ($marc) {
-				my $i = 0;
-
-				while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) {
-					$marc->add(
-						id => $mfn . ( $i ? "/$i" : '' ),
-						fields => $fields,
-						leader => WebPAC::Normalize::_get_marc_leader(),
-						row => $row,
-					);
-					$i++;
+				$store->save_ds(
+					database => $database,
+					input => $input_name,
+					id => $mfn,
+					ds => $ds,
+				) if !$stats;
+
+				$indexer->add(
+					id => "${input_name}/${mfn}",
+					ds => $ds,
+					type => $config->get($indexer_config)->{type},
+				) if $indexer;
+
+				if ($marc) {
+					my $i = 0;
+
+					while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) {
+						$marc->add(
+							id => $mfn . ( $i ? "/$i" : '' ),
+							fields => $fields,
+							leader => WebPAC::Normalize::_get_marc_leader(),
+							row => $row,
+						);
+						$i++;
+					}
+
+					$log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
 				}
 
-				$log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
+				foreach my $out ( @output_modules ) {
+					$out->add( $mfn, $ds ) if $out->can('add');
+				}
+
+			} else {
+				$log->warn("record $pos didn't produce any output after normalization rules!");
 			}
+
 		}
 
 		if ($validate) {
@@ -568,7 +614,10 @@
 				print $report_fh "$errors\n" if ($report_fh);
 			}
 
-			print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1 ), "\n\n";
+			print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1, current_input => 1 ), "\n\n";
+
+			# must be last thing that touches $validate for this input
+			$validate->reset;
 		}
 
 		if ($stats) {
@@ -581,12 +630,15 @@
 		$marc->finish if ($marc);
 
 		# close report
-		close($report_fh) if ($report_fh)
-
+		close($report_fh) if ($report_fh);
 	}
 
 	eval { $indexer->finish } if ($indexer && $indexer->can('finish'));
 
+	foreach my $out ( @output_modules ) {
+		$out->finish if $out->can('finish');
+	}
+
 	my $dt = time() - $start_t;
 	$log->info("$total_rows records ", $indexer ? "indexed " : "",
 		sprintf("in %.2f sec [%.2f rec/sec]",
@@ -610,7 +662,7 @@
 }
 
 # save new delimiters if needed
-$validate->save_delimiters_templates;
+$validate->save_delimiters_templates if ( $validate_delimiters_path );
 
 #
 # handle links or merge after indexing