7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.07; |
use WebPAC::Parser 0.08; |
11 |
use WebPAC::Input 0.13; |
use WebPAC::Input 0.16; |
12 |
use WebPAC::Store 0.11; |
use WebPAC::Store 0.14; |
13 |
use WebPAC::Normalize 0.22; |
use WebPAC::Normalize 0.22; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate 0.06; |
use WebPAC::Validate 0.06; |
135 |
|
|
136 |
$log->info( "-" x 79 ); |
$log->info( "-" x 79 ); |
137 |
|
|
138 |
|
my $log_file = 'log'; |
139 |
|
|
140 |
|
if (-e $log_file ) { # && -s $log_file > 5 * 1024 * 1024) { |
141 |
|
$log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'"); |
142 |
|
rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!"); |
143 |
|
} |
144 |
|
|
145 |
my $estcmd_fh; |
my $estcmd_fh; |
146 |
my $estcmd_path = './estcmd-merge.sh'; |
my $estcmd_path = './estcmd-merge.sh'; |
159 |
|
|
160 |
|
|
161 |
my $use_indexer = $config->use_indexer; |
my $use_indexer = $config->use_indexer; |
162 |
|
$stats ||= $validate; |
163 |
if ($stats) { |
if ($stats) { |
164 |
$log->debug("option --stats disables update of indexing engine..."); |
$log->debug("disabled indexing for stats collection"); |
165 |
$use_indexer = undef; |
$use_indexer = undef; |
166 |
} else { |
} else { |
167 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
205 |
} |
} |
206 |
|
|
207 |
my $indexer; |
my $indexer; |
208 |
if ($use_indexer) { |
if ($use_indexer && $parser->have_rules( 'search', $database )) { |
209 |
|
|
210 |
my $cfg_name = $use_indexer; |
my $cfg_name = $use_indexer; |
211 |
$cfg_name =~ s/\-.*$//; |
$cfg_name =~ s/\-.*$//; |
357 |
my $rec = shift || die "need rec!"; |
my $rec = shift || die "need rec!"; |
358 |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
359 |
|
|
|
$store->save_row( |
|
|
database => $database, |
|
|
input => $input_name, |
|
|
id => $mfn, |
|
|
row => $rec, |
|
|
); |
|
|
|
|
360 |
WebPAC::Normalize::data_structure( |
WebPAC::Normalize::data_structure( |
361 |
row => $rec, |
row => $rec, |
362 |
rules => $rules, |
rules => $rules, |
380 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
381 |
lookup => $lookup_jar, |
lookup => $lookup_jar, |
382 |
%{ $input }, |
%{ $input }, |
383 |
|
load_row => sub { |
384 |
|
my $a = shift; |
385 |
|
return $store->load_row( |
386 |
|
database => $database, |
387 |
|
input => $input_name, |
388 |
|
id => $a->{id}, |
389 |
|
); |
390 |
|
}, |
391 |
|
save_row => sub { |
392 |
|
my $a = shift; |
393 |
|
return $store->save_row( |
394 |
|
database => $database, |
395 |
|
input => $input_name, |
396 |
|
id => $a->{id}, |
397 |
|
row => $a->{row}, |
398 |
|
); |
399 |
|
}, |
400 |
|
|
401 |
); |
); |
402 |
|
|
403 |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
404 |
|
|
405 |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
406 |
$log->debug("created following lookups: ", dump( $lookup_data )); |
$log->debug("created following lookups: ", sub { dump( $lookup_data ) } ); |
407 |
|
|
408 |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
409 |
$store->save_lookup( |
$store->save_lookup( |
426 |
} |
} |
427 |
|
|
428 |
my $marc; |
my $marc; |
429 |
if ($parser->generate_marc( $database, $input_name )) { |
if ($parser->have_rules( 'marc', $database, $input_name )) { |
430 |
$marc = new WebPAC::Output::MARC( |
$marc = new WebPAC::Output::MARC( |
431 |
path => "out/marc/${database}-${input_name}.marc", |
path => "out/marc/${database}-${input_name}.marc", |
432 |
lint => $marc_lint, |
lint => $marc_lint, |
454 |
foreach my $db (keys %$depends) { |
foreach my $db (keys %$depends) { |
455 |
foreach my $i (keys %{$depends->{$db}}) { |
foreach my $i (keys %{$depends->{$db}}) { |
456 |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
457 |
|
my $t = time(); |
458 |
$log->debug("loading lookup $db/$i"); |
$log->debug("loading lookup $db/$i"); |
459 |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
460 |
database => $db, |
database => $db, |
461 |
input => $i, |
input => $i, |
462 |
key => $k, |
key => $k, |
463 |
); |
); |
464 |
|
$log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t)); |
465 |
} |
} |
466 |
} |
} |
467 |
} |
} |
468 |
|
|
469 |
$log->debug("lookup_hash = ", dump( $lookup_hash )); |
$log->debug("lookup_hash = ", sub { dump( $lookup_hash ) }); |
470 |
} |
} |
471 |
|
|
472 |
|
|
474 |
|
|
475 |
my $row = $input_db->fetch || next; |
my $row = $input_db->fetch || next; |
476 |
|
|
477 |
|
$total_rows++; |
478 |
|
|
479 |
my $mfn = $row->{'000'}->[0]; |
my $mfn = $row->{'000'}->[0]; |
480 |
|
|
481 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
if (! $mfn || $mfn !~ m#^\d+$#) { |
486 |
|
|
487 |
|
|
488 |
if ($validate) { |
if ($validate) { |
489 |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { |
490 |
$log->error( "MFN $mfn validation error:\n", |
$log->error( "MFN $mfn validation error:\n", |
491 |
$validate->report_error( $errors ) |
$validate->report_error( $errors ) |
492 |
); |
); |
493 |
} |
} |
494 |
|
next; # validation doesn't create any output |
495 |
} |
} |
496 |
|
|
497 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
510 |
}, |
}, |
511 |
); |
); |
512 |
|
|
513 |
$log->debug("ds = ",dump($ds)); |
$log->debug("ds = ", sub { dump($ds) }) if ($ds); |
514 |
|
|
515 |
$store->save_ds( |
$store->save_ds( |
516 |
database => $database, |
database => $database, |
540 |
|
|
541 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
542 |
} |
} |
|
|
|
|
$total_rows++; |
|
543 |
} |
} |
544 |
|
|
545 |
if ($validate) { |
if ($validate) { |