--- trunk/run.pl 2009/05/18 18:51:40 1184 +++ trunk/run.pl 2009/07/22 09:23:36 1246 @@ -22,7 +22,7 @@ use Data::Dump qw/dump/; use Storable qw/dclone/; use Pod::Usage qw/pod2usage/; -use LWP::Simple; +use LWP::Simple qw//; use POSIX ":sys_wait_h"; # imports WNOHANG @@ -383,7 +383,7 @@ next; } - next if ($only_input && ($input_name !~ m#$only_input#i && $input->{type} !~ m#$only_input#i)); + next if defined $only_input && $input_name !~ m#$only_input#i; my $type = lc($input->{type}); @@ -417,8 +417,8 @@ $base =~ s{/[^/]+$}{}; mkpath $base unless -e $base; - my $rc = mirror( "$mirror/$path", $path ); - if (is_error( $rc )) { + my $rc = LWP::Simple::mirror( "$mirror/$path", $path ); + if (LWP::Simple::is_error( $rc )) { die "can't mirror $mirror/$path -> $path [$rc]"; } else { $log->info( "mirror ", $path, " [$rc] ", -s $path, " bytes" ); @@ -542,7 +542,10 @@ } my $rules = $parser->normalize_rules($database,$input_name); - $log->logwarn("no normalize rules for $database/$input_name") unless $rules; + if ( ! $rules ) { + $log->logwarn("no normalize rules for $database/$input_name", $input_db->input_module->can('normalize') ? " using normalize from input module" : ''); + next; + } $log->debug("parsed normalize rules:\n$rules"); @@ -603,6 +606,9 @@ push @{ $row->{'000'} }, $pos; } + foreach my $out ( @output_modules ) { + $out->add_row( $mfn, $row ) if $out->can('add_row'); + } if ($validate) { if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { @@ -613,9 +619,11 @@ next; # validation doesn't create any output } + my $ds; + if ($rules) { - my $ds = WebPAC::Normalize::data_structure( + $ds = WebPAC::Normalize::data_structure( row => $row, rules => $rules, lookup => $lookup_hash, @@ -632,32 +640,33 @@ }, ); - $log->debug("ds = ", sub { dump($ds) }); - - if ( $ds ) { + } elsif ( $input_db->input_module->can('normalize') ) { + $ds = $input_db->input_module->normalize( $mfn ); + } - $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 ( $ds ) { + $log->debug("ds = ", sub { dump($ds) }); - foreach my $out ( @output_modules ) { - $out->add( $mfn, $ds ) if $out->can('add'); - } + $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; - } else { - $log->warn("record $pos didn't produce any output after normalization rules!") unless $marc; + 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;