--- trunk/run.pl	2006/10/25 21:01:51	765
+++ trunk/run.pl	2007/05/27 10:50:39	850
@@ -8,11 +8,11 @@
 
 use WebPAC::Common 0.02;
 use WebPAC::Parser 0.08;
-use WebPAC::Input 0.14;
+use WebPAC::Input 0.16;
 use WebPAC::Store 0.14;
 use WebPAC::Normalize 0.22;
 use WebPAC::Output::TT;
-use WebPAC::Validate 0.06;
+use WebPAC::Validate 0.11;
 use WebPAC::Output::MARC;
 use WebPAC::Config;
 use Getopt::Long;
@@ -21,6 +21,7 @@
 use File::Slurp;
 use Data::Dump qw/dump/;
 use Storable qw/dclone/;
+use Pod::Usage qw/pod2usage/;
 
 use Proc::Queue size => 1;
 use POSIX ":sys_wait_h"; # imports WNOHANG
@@ -31,7 +32,7 @@
 
 B<this command will probably go away. Don't get used to it!>
 
-Options:
+=head1 OPTIONS
 
 =over 4
 
@@ -67,6 +68,16 @@
 
 turn on extra validation of imput records, see L<WebPAC::Validation>
 
+=item --validate-delimiters path/to/validate_delimiters_file
+
+this option is used with C<--validate> to turn on extra validation of
+delimiters. If file is non existant, it will be created on first run.
+
+=item --marc-generate
+
+Generate MARC file. This will automatically be on if file contains C<marc*> directives.
+You can use this option as C<--no-marc-generate> to disable MARC generation.
+
 =item --marc-lint
 
 By default turned on if normalisation file has C<marc*> directives. You can disable lint
@@ -102,11 +113,14 @@
 my $only_filter;
 my $stats = 0;
 my $validate_path;
+my $validate_delimiters_path;
+my $marc_generate = 1;
 my $marc_lint = 1;
 my $marc_dump = 0;
 my $parallel = 0;
 my $only_links = 0;
 my $merge = 0;
+my $help;
 
 my $log = _new WebPAC::Common()->_get_logger();
 
@@ -120,13 +134,18 @@
 	"debug+" => \$debug,
 	"stats" => \$stats,
 	"validate=s" => \$validate_path,
+	"validate-delimiters=s" => \$validate_delimiters_path,
+	"marc-generate!" => \$marc_generate,
 	"marc-lint!" => \$marc_lint,
 	"marc-dump!" => \$marc_dump,
 	"parallel=i" => \$parallel,
 	"only-links!" => \$only_links,
 	"merge" => \$merge,
+	"help" => \$help,
 );
 
+pod2usage(-verbose => 2) if ($help);
+
 my $config = new WebPAC::Config( path => $config_path );
 
 #print "config = ",dump($config) if ($debug);
@@ -135,6 +154,12 @@
 
 $log->info( "-" x 79 );
 
+my $log_file = 'log';
+
+if (-e $log_file ) {	# && -s $log_file > 5 * 1024 * 1024) {
+	$log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'");
+	rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!");
+}
 
 my $estcmd_fh;
 my $estcmd_path = './estcmd-merge.sh';
@@ -145,16 +170,17 @@
 	$log->info("created merge batch file $estcmd_path");
 }
 
-
 my $validate;
 $validate = new WebPAC::Validate(
 	path => $validate_path,
-) if ($validate_path);
-
+	delimiters => $config->webpac('delimiters'),
+	delimiters_path => $validate_delimiters_path,
+) if ($validate_path || $validate_delimiters_path);
 
 my $use_indexer = $config->use_indexer;
+$stats ||= $validate;
 if ($stats) {
-	$log->debug("option --stats disables update of indexing engine...");
+	$log->debug("disabled indexing for stats collection");
 	$use_indexer = undef;
 } else {
 	$log->info("using $use_indexer indexing engine...");
@@ -419,7 +445,7 @@
 		}
 
 		my $marc;
-		if ($parser->have_rules( 'marc', $database, $input_name )) {
+		if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) {
 			$marc = new WebPAC::Output::MARC(
 				path => "out/marc/${database}-${input_name}.marc",
 				lint => $marc_lint,
@@ -467,6 +493,8 @@
 
 			my $row = $input_db->fetch || next;
 
+			$total_rows++;
+
 			my $mfn = $row->{'000'}->[0];
 
 			if (! $mfn || $mfn !~ m#^\d+$#) {
@@ -477,11 +505,12 @@
 
 
 			if ($validate) {
-				if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) {
+				if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) {
 					$log->error( "MFN $mfn validation error:\n",
 						$validate->report_error( $errors )
 					);
 				}
+				next;	# validation doesn't create any output
 			}
 
 			my $ds = WebPAC::Normalize::data_structure(
@@ -522,7 +551,7 @@
 					$marc->add(
 						id => $mfn . ( $i ? "/$i" : '' ),
 						fields => $fields,
-						leader => WebPAC::Normalize::marc_leader(),
+						leader => WebPAC::Normalize::_get_marc_leader(),
 						row => $row,
 					);
 					$i++;
@@ -530,8 +559,6 @@
 
 				$log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
 			}
-
-			$total_rows++;
 		}
 
 		if ($validate) {
@@ -540,6 +567,8 @@
 				$log->info("validation errors:\n$errors\n" );
 				print $report_fh "$errors\n" if ($report_fh);
 			}
+
+			print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1 ), "\n\n";
 		}
 
 		if ($stats) {
@@ -552,8 +581,10 @@
 		$marc->finish if ($marc);
 
 		# close report
-		close($report_fh) if ($report_fh)
+		close($report_fh) if ($report_fh);
 
+		# reset validate stats
+		$validate->reset if ( $validate );
 	}
 
 	eval { $indexer->finish } if ($indexer && $indexer->can('finish'));
@@ -580,6 +611,9 @@
 	$log->info("all parallel processes finished");
 }
 
+# save new delimiters if needed
+$validate->save_delimiters_templates if ( $validate );
+
 #
 # handle links or merge after indexing
 #