added header_first to WebPAC::Input::CSV

[webpac2] / run.pl

diff --git a/run.pl b/run.pl
index f87ebdb..cd940c9 100755 (executable)
--- a/run.pl
+++ b/run.pl
@@ -22,7 +22,7 @@ use File::Slurp;
 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
 
@@ -86,6 +86,10 @@ You can use this option as C<--no-marc-generate> to disable MARC generation.
 By default turned on if normalisation file has C<marc*> directives. You can disable lint
 messages with C<--no-marc-lint>.
 
+=item --marc-xml
+
+Creeate MARCXML file (this can be quite large)
+
 =item --marc-dump
 
 Force dump or input and marc record for debugging.
@@ -122,8 +126,9 @@ my $stats = 0;
 my $validate_path;
 my $validate_delimiters_path;
 my $marc_generate = 1;
-my $marc_lint = 1;
+my $marc_lint = 0;
 my $marc_dump = 0;
+my $marc_xml = 0;
 my $parallel = 0;
 my $only_links = 0;
 my $merge = 0;
@@ -146,6 +151,7 @@ GetOptions(
        "marc-generate!" => \$marc_generate,
        "marc-lint!" => \$marc_lint,
        "marc-dump!" => \$marc_dump,
+       "marcxml!" => \$marc_xml,
        "parallel=i" => \$parallel,
        "only-links!" => \$only_links,
        "merge" => \$merge,
@@ -383,7 +389,7 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                        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 +423,8 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                        $base =~ s{/[^/]+$}{};
                        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" );
@@ -429,7 +435,7 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                my $input_db = new WebPAC::Input(
                        module => $input_module,
                        limit => $limit || $input->{limit},
-                       offset => $offset,
+                       offset => $offset || $input->{offset},
                        recode => $input->{recode},
                        stats => $stats,
                        modify_records => $input->{modify_records},
@@ -535,14 +541,18 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                        mkpath $out_marc unless -e $out_marc;
 
                        $marc = new WebPAC::Output::MARC(
-                               path => "$out_marc/${database}-${input_name}.marc",
+                               path => "$out_marc/${database}-${input_name}",
                                lint => $marc_lint,
                                dump => $marc_dump,
+                               marcxml => $marc_xml,
                        );
                }
 
                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 +613,9 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                                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 +626,11 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                                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 +647,33 @@ foreach my $database ( sort keys %{ $config->databases } ) {
                                        },
                                );
 
-                               $log->debug("ds = ", sub { dump($ds) });
-
-                               if ( $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;
+                       } elsif ( $input_db->input_module->can('normalize') ) {
+                               $ds = $input_db->input_module->normalize( $mfn );
+                       }
 
-                                       foreach my $out ( @output_modules ) {
-                                               $out->add( $mfn, $ds ) if $out->can('add');
-                                       }
+                       if ( $ds ) {
+                               $log->debug("ds = ", sub { dump($ds) });
 
-                               } else {
-                                       $log->warn("record $pos didn't produce any output after normalization rules!") unless $marc;
+                               $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;