133 |
"clean" => \$clean, |
"clean" => \$clean, |
134 |
"one=s" => \$only_filter, |
"one=s" => \$only_filter, |
135 |
"only=s" => \$only_filter, |
"only=s" => \$only_filter, |
136 |
"config" => \$config_path, |
"config=s" => \$config_path, |
137 |
"debug+" => \$debug, |
"debug+" => \$debug, |
138 |
"stats" => \$stats, |
"stats" => \$stats, |
139 |
"validate=s" => \$validate_path, |
"validate=s" => \$validate_path, |
153 |
|
|
154 |
my $config = new WebPAC::Config( path => $config_path ); |
my $config = new WebPAC::Config( path => $config_path ); |
155 |
|
|
156 |
|
WebPAC::Normalize::_debug( $debug - 1 ) if $debug > 1; |
157 |
|
|
158 |
#print "config = ",dump($config) if ($debug); |
#print "config = ",dump($config) if ($debug); |
159 |
|
|
160 |
die "no databases in config file!\n" unless ($config->databases); |
die "no databases in config file!\n" unless ($config->databases); |
193 |
|
|
194 |
# parse normalize files and create source files for lookup and normalization |
# parse normalize files and create source files for lookup and normalization |
195 |
|
|
196 |
my $parser = new WebPAC::Parser( config => $config ); |
my ($only_database,$only_input) = split(m#/#, $only_filter) if $only_filter; |
197 |
|
|
198 |
|
my $parser = new WebPAC::Parser( |
199 |
|
config => $config, |
200 |
|
only_database => $only_database, |
201 |
|
only_input => $only_input, |
202 |
|
); |
203 |
|
|
204 |
my $total_rows = 0; |
my $total_rows = 0; |
205 |
my $start_t = time(); |
my $start_t = time(); |
223 |
foreach my $database ( sort keys %{ $config->databases } ) { |
foreach my $database ( sort keys %{ $config->databases } ) { |
224 |
my $db_config = $config->databases->{$database}; |
my $db_config = $config->databases->{$database}; |
225 |
|
|
|
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
|
226 |
next if ($only_database && $database !~ m/$only_database/i); |
next if ($only_database && $database !~ m/$only_database/i); |
227 |
|
|
228 |
if ($parallel) { |
if ($parallel) { |
318 |
|
|
319 |
foreach my $output ( @outputs ) { |
foreach my $output ( @outputs ) { |
320 |
|
|
321 |
warn '## output = ',dump( $output ); |
#warn '## output = ',dump( $output ); |
322 |
|
|
323 |
my $module = $output->{module} || $log->logdie("need module in output section of $database"); |
my $module = $output->{module} || $log->logdie("need module in output section of $database"); |
324 |
$module = 'WebPAC::Output::' . $module unless $module =~ m/::/; |
$module = 'WebPAC::Output::' . $module unless $module =~ m/::/; |
354 |
|
|
355 |
my $type = lc($input->{type}); |
my $type = lc($input->{type}); |
356 |
|
|
357 |
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 |
358 |
|
my $input_module = $input->{module}; |
359 |
|
|
360 |
my $input_module = $config->webpac('inputs')->{$type}; |
if ( ! $input_module ) { |
361 |
|
if ( grep(/$type/, $config->webpac('inputs')) ) { |
362 |
|
$input_module = $config->webpac('inputs')->{$type}; |
363 |
|
} else { |
364 |
|
$log->logdie("I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!" ); |
365 |
|
} |
366 |
|
} |
367 |
|
|
368 |
my @lookups = $parser->have_lookup_create($database, $input); |
my @lookups = $parser->have_lookup_create($database, $input); |
369 |
|
|
565 |
marc_encoding => 'utf-8', |
marc_encoding => 'utf-8', |
566 |
load_row_coderef => sub { |
load_row_coderef => sub { |
567 |
my ($database,$input,$mfn) = @_; |
my ($database,$input,$mfn) = @_; |
568 |
|
#warn "### load_row($database,$input,$mfn) from data_structure\n"; |
569 |
return $store->load_row( |
return $store->load_row( |
570 |
database => $database, |
database => $database, |
571 |
input => $input, |
input => $input, |
574 |
}, |
}, |
575 |
); |
); |
576 |
|
|
577 |
$log->debug("ds = ", sub { dump($ds) }) if ($ds); |
$log->debug("ds = ", sub { dump($ds) }); |
578 |
|
|
579 |
|
if ( $ds ) { |
580 |
|
|
581 |
|
$store->save_ds( |
582 |
|
database => $database, |
583 |
|
input => $input_name, |
584 |
|
id => $mfn, |
585 |
|
ds => $ds, |
586 |
|
) if !$stats; |
587 |
|
|
588 |
|
$indexer->add( |
589 |
|
id => "${input_name}/${mfn}", |
590 |
|
ds => $ds, |
591 |
|
type => $config->get($indexer_config)->{type}, |
592 |
|
) if $indexer; |
593 |
|
|
594 |
|
foreach my $out ( @output_modules ) { |
595 |
|
$out->add( $mfn, $ds ) if $out->can('add'); |
596 |
|
} |
597 |
|
|
598 |
$store->save_ds( |
} else { |
599 |
database => $database, |
$log->warn("record $pos didn't produce any output after normalization rules!") unless $marc; |
600 |
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 && $ds); |
|
601 |
|
|
602 |
if ($marc) { |
if ($marc) { |
603 |
my $i = 0; |
my $i = 0; |
604 |
|
|
605 |
while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) { |
while (my $fields = WebPAC::Normalize::MARC::_get_marc_fields( fetch_next => 1 ) ) { |
606 |
$marc->add( |
$marc->add( |
607 |
id => $mfn . ( $i ? "/$i" : '' ), |
id => $mfn . ( $i ? "/$i" : '' ), |
608 |
fields => $fields, |
fields => $fields, |
609 |
leader => WebPAC::Normalize::_get_marc_leader(), |
leader => WebPAC::Normalize::MARC::_get_marc_leader(), |
610 |
row => $row, |
row => $row, |
611 |
); |
); |
612 |
$i++; |
$i++; |
615 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
616 |
} |
} |
617 |
|
|
|
foreach my $out ( @output_modules ) { |
|
|
$out->add( $mfn, $ds ) if $out->can('add'); |
|
|
} |
|
|
|
|
618 |
} |
} |
619 |
|
|
620 |
if ($validate) { |
if ($validate) { |