--- trunk/run.pl	2006/07/04 11:36:12	581
+++ trunk/run.pl	2006/08/01 13:59:47	606
@@ -7,8 +7,8 @@
 use lib './lib';
 
 use WebPAC::Common 0.02;
-use WebPAC::Lookup;
-use WebPAC::Input 0.03;
+use WebPAC::Lookup 0.03;
+use WebPAC::Input 0.07;
 use WebPAC::Store 0.03;
 use WebPAC::Normalize 0.11;
 use WebPAC::Output::TT;
@@ -20,6 +20,10 @@
 use Time::HiRes qw/time/;
 use File::Slurp;
 use Data::Dump qw/dump/;
+use Storable qw/dclone/;
+
+use Proc::Queue size => 1;
+use POSIX ":sys_wait_h"; # imports WNOHANG
 
 =head1 NAME
 
@@ -80,6 +84,11 @@
 
 Force dump or input and marc record for debugging.
 
+=item --parallel 4
+
+Run databases in parallel (aproximatly same as number of processors in
+machine if you want to use full load)
+
 =back
 
 =cut
@@ -97,6 +106,8 @@
 my $marc_lint = 1;
 my $marc_dump = 0;
 
+my $parallel = 0;
+
 GetOptions(
 	"limit=i" => \$limit,
 	"offset=i" => \$offset,
@@ -111,6 +122,7 @@
 	"marc-output=s" => \$marc_output,
 	"marc-lint!" => \$marc_lint,
 	"marc-dump!" => \$marc_dump,
+	"parallel=i" => \$parallel,
 );
 
 $config = LoadFile($config);
@@ -144,11 +156,25 @@
 my @links;
 my $indexer;
 
+if ($parallel) {
+	$log->info("Using $parallel processes for speedup");
+	Proc::Queue::size($parallel);
+}
+
 while (my ($database, $db_config) = each %{ $config->{databases} }) {
 
 	my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
 	next if ($only_database && $database !~ m/$only_database/i);
 
+	if ($parallel) {
+		my $f=fork;
+		if(defined ($f) and $f==0) {
+			$log->info("Created processes $$ for speedup");
+		} else {
+			next;
+		}
+	}
+
 	if ($use_indexer) {
 		my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
 		$indexer_config->{database} = $database;
@@ -221,9 +247,13 @@
 
 		die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
 
-		my $lookup = new WebPAC::Lookup(
-			lookup_file => $input->{lookup},
-		) if ($input->{lookup});
+		my $lookup;
+		if ($input->{lookup}) {
+			$lookup = new WebPAC::Lookup(
+				lookup_file => $input->{lookup},
+			);
+			delete( $input->{lookup} );
+		}
 
 		my $input_module = $config->{webpac}->{inputs}->{$type};
 
@@ -233,12 +263,16 @@
 
 		my $input_db = new WebPAC::Input(
 			module => $input_module,
-			code_page => $config->{webpac}->{webpac_encoding},
+			encoding => $config->{webpac}->{webpac_encoding},
 			limit => $limit || $input->{limit},
 			offset => $offset,
-			lookup => $lookup,
+			lookup_coderef => sub {
+				my $rec = shift || return;
+				$lookup->add( $rec );
+			},
 			recode => $input->{recode},
 			stats => $stats,
+			modify_records => $input->{modify_records},
 		);
 		$log->logdie("can't create input using $input_module") unless ($input);
 
@@ -295,10 +329,22 @@
 					$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
 				}
 
+				my $ds_config = dclone($db_config);
+
+				# default values -> database key
+				$ds_config->{_} = $database;
+
+				# current mfn
+				$ds_config->{_mfn} = $mfn;
+
+				# attach current input
+				$ds_config->{input} = $input;
+
 				my $ds = WebPAC::Normalize::data_structure(
 					row => $row,
 					rules => $rules,
 					lookup => $lookup ? $lookup->lookup_hash : undef,
+					config => $ds_config,
 					marc_encoding => 'utf-8',
 				);
 
@@ -369,9 +415,22 @@
 		}
 	}
 
+	# end forked process
+	if ($parallel) {
+		$log->info("parallel process $$ finished");
+		exit(0);
+	}
+
+}
+
+if ($parallel) {
+	# wait all children to finish
+	sleep(1) while wait != -1;
+	$log->info("all parallel processes finished");
 }
 
 foreach my $link (@links) {
 	$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
 	$indexer->add_link( %{ $link } );
 }
+