"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,
my $config = new WebPAC::Config( path => $config_path );
+WebPAC::Normalize::_debug( $debug - 1 ) if $debug > 1;
+
#print "config = ",dump($config) if ($debug);
die "no databases in config file!\n" unless ($config->databases);
# 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();
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) {
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 {
+ 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?" );
} );
foreach my $output ( @outputs ) {
-warn '## output = ',dump( $output );
+#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";
-
+
+ # add database to arugemnts for output filter
$output->{database} = $database;
$log->debug("calling $module->new(",dump( $output ),")");
#
- my @inputs = iterate_over( $db_config->{input}, sub {
+ my @inputs = force_array( $db_config->{input}, sub {
$log->info("database $database doesn't have inputs defined");
} );
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);
}
+ # 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 = $input_db->fetch || next;
marc_encoding => 'utf-8',
load_row_coderef => sub {
my ($database,$input,$mfn) = @_;
+#warn "### load_row($database,$input,$mfn) from data_structure\n";
return $store->load_row(
database => $database,
input => $input,
},
);
- $log->debug("ds = ", sub { dump($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);
+ $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');
+ }
- $indexer->add(
- id => "${input_name}/${mfn}",
- ds => $ds,
- type => $config->get($indexer_config)->{type},
- ) if ($indexer && $ds);
+ } 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++;
$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) {