added header_first to WebPAC::Input::CSV

[webpac2] / run.pl

diff --git a/run.pl b/run.pl
index 09261f0..cd940c9 100755 (executable)
--- a/run.pl
+++ b/run.pl
@@ -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});
 
@@ -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", $input_db->input_module->can('normalize') ? " using normalize from input module" : '') 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");