--- trunk/run.pl 2007/09/03 15:26:46 887 +++ trunk/run.pl 2007/11/12 12:22:08 1041 @@ -133,7 +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, @@ -191,7 +191,13 @@ # parse normalize files and create source files for lookup and normalization -my $parser = new WebPAC::Parser( config => $config ); +my ($only_database,$only_input) = split(m#/#, $only_filter) if $only_filter; + +my $parser = new WebPAC::Parser( + config => $config, + only_database => $only_database, + only_input => $only_input, +); my $total_rows = 0; my $start_t = time(); @@ -215,7 +221,6 @@ 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); if ($parallel) { @@ -255,10 +260,7 @@ } elsif ($use_indexer eq 'kinosearch') { - # open KinoSearch - require 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"); @@ -302,18 +304,34 @@ debug => $debug, }); + # # prepare output # - my @outputs; - if (defined( $db_config->{output} )) { - my $module = $db_config->{output}->{module} || $log->logdie("need module in output section of $database"); + 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 @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"; - my $out = new $module->new( $db_config->{output} ); + + # 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 @outputs, $out; + + push @output_modules, $out; } @@ -321,14 +339,10 @@ # now, iterate through input formats # - my @inputs; - if (ref($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) { @@ -338,9 +352,16 @@ 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); @@ -500,6 +521,16 @@ } + # 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 = $input_db->fetch || next; @@ -532,6 +563,7 @@ marc_encoding => 'utf-8', load_row_coderef => sub { my ($database,$input,$mfn) = @_; +#warn "### load_row($database,$input,$mfn) from data_structure\n"; return $store->load_row( database => $database, input => $input, @@ -540,29 +572,39 @@ }, ); - $log->debug("ds = ", sub { dump($ds) }) if ($ds); + $log->debug("ds = ", sub { dump($ds) }); + + if ( $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); + $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; + + 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!") unless $marc; + } if ($marc) { my $i = 0; - while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) { + while (my $fields = WebPAC::Normalize::MARC::_get_marc_fields( fetch_next => 1 ) ) { $marc->add( id => $mfn . ( $i ? "/$i" : '' ), fields => $fields, - leader => WebPAC::Normalize::_get_marc_leader(), + leader => WebPAC::Normalize::MARC::_get_marc_leader(), row => $row, ); $i++; @@ -571,10 +613,6 @@ $log->info("Created $i instances of MFN $mfn\n") if ($i > 1); } - foreach my $out ( @outputs ) { - $out->add( $mfn, $ds ) if $out->can('add'); - } - } if ($validate) { @@ -605,7 +643,7 @@ eval { $indexer->finish } if ($indexer && $indexer->can('finish')); - foreach my $out ( @outputs ) { + foreach my $out ( @output_modules ) { $out->finish if $out->can('finish'); }