r1393@llin: dpavlin | 2007-10-31 00:52:52 +0100

[webpac2] / run.pl

diff --git a/run.pl b/run.pl
index c324c65..03de8b3 100755 (executable)
--- a/run.pl
+++ b/run.pl
@@ -9,7 +9,7 @@ use lib './lib';
 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;
@@ -185,7 +185,7 @@ $stats ||= $validate;
 if ($stats) {
        $log->debug("disabled indexing for stats collection");
        $use_indexer = undef;
-} else {
+} elsif ( $use_indexer ) {
        $log->info("using $use_indexer indexing engine...");
 }
 
@@ -244,27 +244,24 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                if ($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);
        }
 
 
@@ -298,36 +295,60 @@ foreach my $database ( sort keys %{ $config->databases } ) {
        #
        # now WebPAC::Store
        #
-       my $abs_path = abs_path($0);
-       $abs_path =~ s#/[^/]*$#/#;
+       my $store = new WebPAC::Store({
+               debug => $debug,
+       });
+
+       sub iterate_over {
+               my ( $what, $error ) = @_;
+               my @result;
+               if ( ref( $what ) eq 'ARRAY' ) {
+                       @result = @{ $what };
+               } elsif ($db_config->{input}) {
+                       @result =  ( $what );
+               } else {
+                       $error->() if ref($error) eq 'CODE';
+               }
+               return @result;
+       }
+
+       #
+       # prepare output
+       #
+       my @outputs = iterate_over( $db_config->{output}, sub {
+               $log->error("Database $database doesn't have any outputs defined. Do you want to remove it from configuration?" );
+       } );
 
-       my $db_path = $config->webpac('db_path');
+       my @output_modules;
 
-       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");
-       }
+       foreach my $output ( @outputs ) {
 
-       my $store = new WebPAC::Store(
-               path => $db_path,
-               debug => $debug,
-       );
+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";
+       
+               $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 = iterate_over( $db_config->{input}, sub {
                $log->info("database $database doesn't have inputs defined");
-       }
+       } );
 
        foreach my $input (@inputs) {
 
@@ -362,6 +383,7 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                        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);
 
@@ -506,7 +528,7 @@ foreach my $database ( sort keys %{ $config->databases } ) {
 
                        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;
@@ -568,6 +590,11 @@ foreach my $database ( sort keys %{ $config->databases } ) {
 
                                $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');
+                       }
+
                }
 
                if ($validate) {
@@ -598,6 +625,10 @@ foreach my $database ( sort keys %{ $config->databases } ) {
 
        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]",