23 |
use Storable qw/dclone/; |
use Storable qw/dclone/; |
24 |
use Pod::Usage qw/pod2usage/; |
use Pod::Usage qw/pod2usage/; |
25 |
|
|
|
use Proc::Queue size => 1; |
|
26 |
use POSIX ":sys_wait_h"; # imports WNOHANG |
use POSIX ":sys_wait_h"; # imports WNOHANG |
27 |
|
|
28 |
=head1 NAME |
=head1 NAME |
206 |
my @links; |
my @links; |
207 |
|
|
208 |
if ($parallel) { |
if ($parallel) { |
209 |
|
eval 'use Proc::Queue size => 1;'; |
210 |
|
die $@ if $@; |
211 |
$log->info("Using $parallel processes for speedup"); |
$log->info("Using $parallel processes for speedup"); |
212 |
Proc::Queue::size($parallel); |
Proc::Queue::size($parallel); |
213 |
} |
} |
329 |
|
|
330 |
# add database to arugemnts for output filter |
# add database to arugemnts for output filter |
331 |
$output->{database} = $database; |
$output->{database} = $database; |
332 |
|
$output->{clean} = $clean; |
333 |
|
|
334 |
$log->debug("calling $module->new(",dump( $output ),")"); |
$log->debug("calling $module->new(",dump( $output ),")"); |
335 |
my $out = new $module->new( $output ); |
my $out = new $module->new( $output ); |
350 |
$log->info("database $database doesn't have inputs defined"); |
$log->info("database $database doesn't have inputs defined"); |
351 |
} ); |
} ); |
352 |
|
|
353 |
|
if ( -e 'out/debug' ) { # FIXME flag? |
354 |
|
my $out; |
355 |
|
foreach my $i ( @inputs ) { |
356 |
|
warn dump( $i ); |
357 |
|
next unless defined $i->{normalize}; |
358 |
|
warn dump( $i->{normalize} ); |
359 |
|
foreach my $normalize ( @{ $i->{normalize} } ) { |
360 |
|
my $path = $normalize->{path}; |
361 |
|
$out .= qq/\n##\n## $path\n##\n\n/; |
362 |
|
$out .= read_file( $path ); |
363 |
|
} |
364 |
|
} |
365 |
|
my $all = "out/debug/all-normalize.pl"; |
366 |
|
write_file( $all, $out ); |
367 |
|
warn "### all normalize for this input saved to: $all"; |
368 |
|
}; |
369 |
|
|
370 |
foreach my $input (@inputs) { |
foreach my $input (@inputs) { |
371 |
|
|
372 |
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)); |
400 |
|
|
401 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
402 |
module => $input_module, |
module => $input_module, |
|
encoding => $config->webpac('webpac_encoding'), |
|
403 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
404 |
offset => $offset, |
offset => $offset, |
405 |
recode => $input->{recode}, |
recode => $input->{recode}, |
444 |
|
|
445 |
my $maxmfn = $input_db->open( |
my $maxmfn = $input_db->open( |
446 |
path => $input->{path}, |
path => $input->{path}, |
447 |
code_page => $input->{encoding}, # database encoding |
input_encoding => $input->{encoding}, # database encoding |
448 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
449 |
lookup => $lookup_jar, |
lookup => $lookup_jar, |
450 |
%{ $input }, |
%{ $input }, |
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; |
|
); |
|
611 |
|
|
612 |
$log->debug("ds = ", sub { dump($ds) }); |
$indexer->add( |
613 |
|
id => "${input_name}/${mfn}", |
614 |
|
ds => $ds, |
615 |
|
type => $config->get($indexer_config)->{type}, |
616 |
|
) if $indexer; |
617 |
|
|
618 |
if ( $ds ) { |
foreach my $out ( @output_modules ) { |
619 |
|
$out->add( $mfn, $ds ) if $out->can('add'); |
620 |
$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; |
|
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) { |