7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.04; |
use WebPAC::Parser 0.07; |
11 |
use WebPAC::Input 0.13; |
use WebPAC::Input 0.13; |
12 |
use WebPAC::Store 0.11; |
use WebPAC::Store 0.11; |
13 |
use WebPAC::Normalize 0.11; |
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; |
16 |
use WebPAC::Output::MARC; |
use WebPAC::Output::MARC; |
67 |
|
|
68 |
turn on extra validation of imput records, see L<WebPAC::Validation> |
turn on extra validation of imput records, see L<WebPAC::Validation> |
69 |
|
|
|
=item --marc-normalize conf/normalize/mapping.pl |
|
|
|
|
|
This option specifies normalisation file for MARC creation |
|
|
|
|
|
=item --marc-output out/marc/test.marc |
|
|
|
|
|
Optional path to output file |
|
|
|
|
70 |
=item --marc-lint |
=item --marc-lint |
71 |
|
|
72 |
By default turned on if C<--marc-normalize> is used. You can disable lint |
By default turned on if normalisation file has C<marc*> directives. You can disable lint |
73 |
messages with C<--no-marc-lint>. |
messages with C<--no-marc-lint>. |
74 |
|
|
75 |
=item --marc-dump |
=item --marc-dump |
102 |
my $only_filter; |
my $only_filter; |
103 |
my $stats = 0; |
my $stats = 0; |
104 |
my $validate_path; |
my $validate_path; |
|
my ($marc_normalize, $marc_output); |
|
105 |
my $marc_lint = 1; |
my $marc_lint = 1; |
106 |
my $marc_dump = 0; |
my $marc_dump = 0; |
107 |
my $parallel = 0; |
my $parallel = 0; |
120 |
"debug+" => \$debug, |
"debug+" => \$debug, |
121 |
"stats" => \$stats, |
"stats" => \$stats, |
122 |
"validate=s" => \$validate_path, |
"validate=s" => \$validate_path, |
|
"marc-normalize=s" => \$marc_normalize, |
|
|
"marc-output=s" => \$marc_output, |
|
123 |
"marc-lint!" => \$marc_lint, |
"marc-lint!" => \$marc_lint, |
124 |
"marc-dump!" => \$marc_dump, |
"marc-dump!" => \$marc_dump, |
125 |
"parallel=i" => \$parallel, |
"parallel=i" => \$parallel, |
160 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
161 |
} |
} |
162 |
|
|
|
# disable indexing when creating marc |
|
|
$use_indexer = undef if ($marc_normalize); |
|
|
|
|
163 |
# parse normalize files and create source files for lookup and normalization |
# parse normalize files and create source files for lookup and normalization |
164 |
|
|
165 |
my $parser = new WebPAC::Parser( config => $config ); |
my $parser = new WebPAC::Parser( config => $config ); |
340 |
delete( $input->{lookup} ); |
delete( $input->{lookup} ); |
341 |
} |
} |
342 |
|
|
|
my $lookup; |
|
343 |
my $lookup_coderef; |
my $lookup_coderef; |
344 |
|
|
345 |
if (@lookups) { |
if (@lookups) { |
350 |
my $rec = shift || die "need rec!"; |
my $rec = shift || die "need rec!"; |
351 |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
352 |
|
|
353 |
|
$store->save_row( |
354 |
|
database => $database, |
355 |
|
input => $input_name, |
356 |
|
id => $mfn, |
357 |
|
row => $rec, |
358 |
|
); |
359 |
|
|
360 |
WebPAC::Normalize::data_structure( |
WebPAC::Normalize::data_structure( |
361 |
row => $rec, |
row => $rec, |
362 |
rules => $rules, |
rules => $rules, |
|
lookup => $lookup, |
|
363 |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
364 |
); |
); |
365 |
|
|
366 |
warn "current lookup = ", dump($lookup) if ($lookup); |
#warn "current lookup: ", dump(WebPAC::Normalize::_get_lookup()); |
367 |
}; |
}; |
368 |
|
|
369 |
WebPAC::Normalize::_set_lookup( undef ); |
WebPAC::Normalize::_set_lookup( undef ); |
372 |
|
|
373 |
}; |
}; |
374 |
|
|
375 |
|
my $lookup_jar; |
376 |
|
|
377 |
my $maxmfn = $input_db->open( |
my $maxmfn = $input_db->open( |
378 |
path => $input->{path}, |
path => $input->{path}, |
379 |
code_page => $input->{encoding}, # database encoding |
code_page => $input->{encoding}, # database encoding |
380 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
381 |
|
lookup => $lookup_jar, |
382 |
%{ $input }, |
%{ $input }, |
383 |
); |
); |
384 |
|
|
407 |
$log->info("Generating report file $path"); |
$log->info("Generating report file $path"); |
408 |
} |
} |
409 |
|
|
410 |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
my $marc; |
411 |
@{ $input->{normalize} } : ( $input->{normalize} ); |
if ($parser->generate_marc( $database, $input_name )) { |
412 |
|
$marc = new WebPAC::Output::MARC( |
413 |
if ($marc_normalize) { |
path => "out/marc/${database}-${input_name}.marc", |
|
@norm_array = ( { |
|
|
path => $marc_normalize, |
|
|
output => $marc_output || "out/marc/${database}-${input_name}.marc", |
|
|
} ); |
|
|
} |
|
|
|
|
|
foreach my $normalize (@norm_array) { |
|
|
|
|
|
my $normalize_path = $normalize->{path} || $log->logdie("can't find normalize path in config"); |
|
|
|
|
|
$log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i ); |
|
|
|
|
|
my $rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!"; |
|
|
|
|
|
$log->info("Using $normalize_path for normalization..."); |
|
|
|
|
|
my $marc = new WebPAC::Output::MARC( |
|
|
path => $normalize->{output}, |
|
414 |
lint => $marc_lint, |
lint => $marc_lint, |
415 |
dump => $marc_dump, |
dump => $marc_dump, |
416 |
) if ($normalize->{output}); |
); |
417 |
|
} |
|
# reset position in database |
|
|
$input_db->seek(1); |
|
|
|
|
|
# generate name of config key for indexer (strip everything after -) |
|
|
my $indexer_config = $use_indexer; |
|
|
$indexer_config =~ s/^(\w+)-?.*$/$1/g if ($indexer_config); |
|
|
|
|
|
foreach my $pos ( 0 ... $input_db->size ) { |
|
|
|
|
|
my $row = $input_db->fetch || next; |
|
|
|
|
|
my $mfn = $row->{'000'}->[0]; |
|
418 |
|
|
419 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
my $rules = $parser->normalize_rules($database,$input_name) || $log->logdie("no normalize rules found for $database/$input_name"); |
420 |
$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos"); |
$log->debug("parsed normalize rules:\n$rules"); |
|
$mfn = $pos; |
|
|
push @{ $row->{'000'} }, $pos; |
|
|
} |
|
421 |
|
|
422 |
|
# reset position in database |
423 |
|
$input_db->seek(1); |
424 |
|
|
425 |
if ($validate) { |
# generate name of config key for indexer (strip everything after -) |
426 |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
my $indexer_config = $use_indexer; |
427 |
$log->error( "MFN $mfn validation error:\n", |
$indexer_config =~ s/^(\w+)-?.*$/$1/g if ($indexer_config); |
428 |
$validate->report_error( $errors ) |
|
429 |
|
my $lookup_hash; |
430 |
|
my $depends = $parser->depends($database,$input_name); |
431 |
|
|
432 |
|
if ($depends) { |
433 |
|
$log->debug("$database/$input_name depends on: ", dump($depends)) if ($depends); |
434 |
|
$log->logdie("parser->depends didn't return HASH") unless (ref($depends) eq 'HASH'); |
435 |
|
|
436 |
|
foreach my $db (keys %$depends) { |
437 |
|
foreach my $i (keys %{$depends->{$db}}) { |
438 |
|
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
439 |
|
$log->debug("loading lookup $db/$i"); |
440 |
|
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
441 |
|
database => $db, |
442 |
|
input => $i, |
443 |
|
key => $k, |
444 |
); |
); |
445 |
} |
} |
446 |
} |
} |
447 |
|
} |
448 |
|
|
449 |
my $ds = WebPAC::Normalize::data_structure( |
$log->debug("lookup_hash = ", dump( $lookup_hash )); |
450 |
row => $row, |
} |
|
rules => $rules, |
|
|
lookup => $lookup ? $lookup->lookup_hash : undef, |
|
|
config => create_ds_config( $db_config, $database, $input, $mfn ), |
|
|
marc_encoding => 'utf-8', |
|
|
); |
|
451 |
|
|
|
$store->save_ds( |
|
|
database => $database, |
|
|
input => $input_name, |
|
|
id => $mfn, |
|
|
ds => $ds, |
|
|
) if ($ds && !$stats); |
|
452 |
|
|
453 |
$indexer->add( |
foreach my $pos ( 0 ... $input_db->size ) { |
|
id => "${input_name}/${mfn}", |
|
|
ds => $ds, |
|
|
type => $config->get($indexer_config)->{type}, |
|
|
) if ($indexer && $ds); |
|
|
|
|
|
if ($marc) { |
|
|
my $i = 0; |
|
|
|
|
|
while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) { |
|
|
$marc->add( |
|
|
id => $mfn . ( $i ? "/$i" : '' ), |
|
|
fields => $fields, |
|
|
leader => WebPAC::Normalize::marc_leader(), |
|
|
row => $row, |
|
|
); |
|
|
$i++; |
|
|
} |
|
454 |
|
|
455 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
my $row = $input_db->fetch || next; |
456 |
} |
|
457 |
|
my $mfn = $row->{'000'}->[0]; |
458 |
|
|
459 |
$total_rows++; |
if (! $mfn || $mfn !~ m#^\d+$#) { |
460 |
|
$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos"); |
461 |
|
$mfn = $pos; |
462 |
|
push @{ $row->{'000'} }, $pos; |
463 |
} |
} |
464 |
|
|
465 |
|
|
466 |
if ($validate) { |
if ($validate) { |
467 |
my $errors = $validate->report; |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
468 |
if ($errors) { |
$log->error( "MFN $mfn validation error:\n", |
469 |
$log->info("validation errors:\n$errors\n" ); |
$validate->report_error( $errors ) |
470 |
print $report_fh "$errors\n" if ($report_fh); |
); |
471 |
} |
} |
472 |
} |
} |
473 |
|
|
474 |
if ($stats) { |
my $ds = WebPAC::Normalize::data_structure( |
475 |
my $s = $input_db->stats; |
row => $row, |
476 |
$log->info("statistics of fields usage:\n$s"); |
rules => $rules, |
477 |
print $report_fh "Statistics of fields usage:\n$s" if ($report_fh); |
lookup => $lookup_hash, |
478 |
|
config => create_ds_config( $db_config, $database, $input, $mfn ), |
479 |
|
marc_encoding => 'utf-8', |
480 |
|
load_row_coderef => sub { |
481 |
|
my ($database,$input,$mfn) = @_; |
482 |
|
return $store->load_row( |
483 |
|
database => $database, |
484 |
|
input => $input, |
485 |
|
id => $mfn, |
486 |
|
); |
487 |
|
}, |
488 |
|
); |
489 |
|
|
490 |
|
$log->debug("ds = ",dump($ds)); |
491 |
|
|
492 |
|
$store->save_ds( |
493 |
|
database => $database, |
494 |
|
input => $input_name, |
495 |
|
id => $mfn, |
496 |
|
ds => $ds, |
497 |
|
) if ($ds && !$stats); |
498 |
|
|
499 |
|
$indexer->add( |
500 |
|
id => "${input_name}/${mfn}", |
501 |
|
ds => $ds, |
502 |
|
type => $config->get($indexer_config)->{type}, |
503 |
|
) if ($indexer && $ds); |
504 |
|
|
505 |
|
if ($marc) { |
506 |
|
my $i = 0; |
507 |
|
|
508 |
|
while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) { |
509 |
|
$marc->add( |
510 |
|
id => $mfn . ( $i ? "/$i" : '' ), |
511 |
|
fields => $fields, |
512 |
|
leader => WebPAC::Normalize::marc_leader(), |
513 |
|
row => $row, |
514 |
|
); |
515 |
|
$i++; |
516 |
|
} |
517 |
|
|
518 |
|
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
519 |
} |
} |
520 |
|
|
521 |
# close MARC file |
$total_rows++; |
522 |
$marc->finish if ($marc); |
} |
523 |
|
|
524 |
|
if ($validate) { |
525 |
|
my $errors = $validate->report; |
526 |
|
if ($errors) { |
527 |
|
$log->info("validation errors:\n$errors\n" ); |
528 |
|
print $report_fh "$errors\n" if ($report_fh); |
529 |
|
} |
530 |
|
} |
531 |
|
|
532 |
# close report |
if ($stats) { |
533 |
close($report_fh) if ($report_fh) |
my $s = $input_db->stats; |
534 |
|
$log->info("statistics of fields usage:\n$s"); |
535 |
|
print $report_fh "Statistics of fields usage:\n$s" if ($report_fh); |
536 |
} |
} |
537 |
|
|
538 |
|
# close MARC file |
539 |
|
$marc->finish if ($marc); |
540 |
|
|
541 |
|
# close report |
542 |
|
close($report_fh) if ($report_fh) |
543 |
|
|
544 |
} |
} |
545 |
|
|
546 |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |