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); |
328 |
|
|
329 |
# add database to arugemnts for output filter |
# add database to arugemnts for output filter |
330 |
$output->{database} = $database; |
$output->{database} = $database; |
331 |
|
$output->{clean} = $clean; |
332 |
|
|
333 |
$log->debug("calling $module->new(",dump( $output ),")"); |
$log->debug("calling $module->new(",dump( $output ),")"); |
334 |
my $out = new $module->new( $output ); |
my $out = new $module->new( $output ); |
335 |
$out->init; |
if ( $out->init ) { |
336 |
|
push @output_modules, $out; |
337 |
push @output_modules, $out; |
} else { |
338 |
|
$log->warn("SKIPPED $module"); |
339 |
|
} |
340 |
} |
} |
341 |
|
|
342 |
|
|
349 |
$log->info("database $database doesn't have inputs defined"); |
$log->info("database $database doesn't have inputs defined"); |
350 |
} ); |
} ); |
351 |
|
|
352 |
|
if ( -e 'out/debug' ) { # fixme flag? |
353 |
|
my $out; |
354 |
|
foreach my $i ( @inputs ) { |
355 |
|
warn dump( $i ); |
356 |
|
next unless defined $i->{normalize}; |
357 |
|
warn dump( $i->{normalize} ); |
358 |
|
foreach my $normalize ( @{ $i->{normalize} } ) { |
359 |
|
my $path = $normalize->{path}; |
360 |
|
$out .= qq/\n##\n## $path\n##\n\n/; |
361 |
|
$out .= read_file( $path ); |
362 |
|
} |
363 |
|
} |
364 |
|
my $all = "out/debug/all-normalize.pl"; |
365 |
|
write_file( $all, $out ); |
366 |
|
warn "### all normalize for this input saved to: $all"; |
367 |
|
}; |
368 |
|
|
369 |
foreach my $input (@inputs) { |
foreach my $input (@inputs) { |
370 |
|
|
371 |
my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input)); |
my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input)); |
387 |
|
|
388 |
my @lookups = $parser->have_lookup_create($database, $input); |
my @lookups = $parser->have_lookup_create($database, $input); |
389 |
|
|
390 |
$log->info("working on input '$input_name' in $input->{path} [type: $input->{type}] using $input_module", |
$log->info("working on $database/$input_name with $input_module on $input->{path}", |
391 |
@lookups ? " creating lookups: ".join(", ", @lookups) : "" |
@lookups ? " creating lookups: ".join(", ", @lookups) : "" |
392 |
); |
); |
393 |
|
|
507 |
); |
); |
508 |
} |
} |
509 |
|
|
510 |
my $rules = $parser->normalize_rules($database,$input_name) || $log->logdie("no normalize rules found for $database/$input_name"); |
my $rules = $parser->normalize_rules($database,$input_name); |
511 |
|
$log->logwarn("no normalize rules for $database/$input_name") unless $rules; |
512 |
|
|
513 |
$log->debug("parsed normalize rules:\n$rules"); |
$log->debug("parsed normalize rules:\n$rules"); |
514 |
|
|
515 |
# reset position in database |
# reset position in database |
579 |
next; # validation doesn't create any output |
next; # validation doesn't create any output |
580 |
} |
} |
581 |
|
|
582 |
my $ds = WebPAC::Normalize::data_structure( |
if ($rules) { |
583 |
row => $row, |
|
584 |
rules => $rules, |
my $ds = WebPAC::Normalize::data_structure( |
585 |
lookup => $lookup_hash, |
row => $row, |
586 |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
rules => $rules, |
587 |
marc_encoding => 'utf-8', |
lookup => $lookup_hash, |
588 |
load_row_coderef => sub { |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
589 |
my ($database,$input,$mfn) = @_; |
marc_encoding => 'utf-8', |
590 |
|
load_row_coderef => sub { |
591 |
|
my ($database,$input,$mfn) = @_; |
592 |
#warn "### load_row($database,$input,$mfn) from data_structure\n"; |
#warn "### load_row($database,$input,$mfn) from data_structure\n"; |
593 |
return $store->load_row( |
return $store->load_row( |
594 |
|
database => $database, |
595 |
|
input => $input, |
596 |
|
id => $mfn, |
597 |
|
); |
598 |
|
}, |
599 |
|
); |
600 |
|
|
601 |
|
$log->debug("ds = ", sub { dump($ds) }); |
602 |
|
|
603 |
|
if ( $ds ) { |
604 |
|
|
605 |
|
$store->save_ds( |
606 |
database => $database, |
database => $database, |
607 |
input => $input, |
input => $input_name, |
608 |
id => $mfn, |
id => $mfn, |
609 |
); |
ds => $ds, |
610 |
}, |
) if !$stats; |
|
); |
|
|
|
|
|
$log->debug("ds = ", sub { dump($ds) }); |
|
611 |
|
|
612 |
if ( $ds ) { |
$indexer->add( |
613 |
|
id => "${input_name}/${mfn}", |
614 |
|
ds => $ds, |
615 |
|
type => $config->get($indexer_config)->{type}, |
616 |
|
) if $indexer; |
617 |
|
|
618 |
$store->save_ds( |
foreach my $out ( @output_modules ) { |
619 |
database => $database, |
$out->add( $mfn, $ds ) if $out->can('add'); |
620 |
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; |
|
621 |
|
|
622 |
foreach my $out ( @output_modules ) { |
} else { |
623 |
$out->add( $mfn, $ds ) if $out->can('add'); |
$log->warn("record $pos didn't produce any output after normalization rules!") unless $marc; |
624 |
} |
} |
|
|
|
|
} else { |
|
|
$log->warn("record $pos didn't produce any output after normalization rules!") unless $marc; |
|
625 |
} |
} |
626 |
|
|
627 |
if ($marc) { |
if ($marc) { |