12 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.03; |
13 |
use WebPAC::Normalize 0.11; |
use WebPAC::Normalize 0.11; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate; |
use WebPAC::Validate 0.06; |
16 |
use WebPAC::Output::MARC; |
use WebPAC::Output::MARC; |
17 |
use YAML qw/LoadFile/; |
use YAML qw/LoadFile/; |
18 |
use Getopt::Long; |
use Getopt::Long; |
60 |
|
|
61 |
=item --stats |
=item --stats |
62 |
|
|
63 |
disable indexing and dump statistics about field and subfield |
disable indexing, modify_* in configuration and dump statistics about field |
64 |
usage for each input |
and subfield usage for each input |
65 |
|
|
66 |
=item --validate path/to/validation_file |
=item --validate path/to/validation_file |
67 |
|
|
232 |
use WebPAC::Output::EstraierNative; |
use WebPAC::Output::EstraierNative; |
233 |
$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } ); |
$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } ); |
234 |
|
|
|
$use_indexer = 'hyperestraier'; |
|
|
|
|
235 |
} elsif ($use_indexer eq 'kinosearch') { |
} elsif ($use_indexer eq 'kinosearch') { |
236 |
|
|
237 |
# open KinoSearch |
# open KinoSearch |
333 |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
334 |
); |
); |
335 |
|
|
336 |
|
if ($stats) { |
337 |
|
# disable modification of records if --stats is in use |
338 |
|
delete($input->{modify_records}); |
339 |
|
delete($input->{modify_file}); |
340 |
|
} |
341 |
|
|
342 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
343 |
module => $input_module, |
module => $input_module, |
344 |
encoding => $config->{webpac}->{webpac_encoding}, |
encoding => $config->{webpac}->{webpac_encoding}, |
351 |
recode => $input->{recode}, |
recode => $input->{recode}, |
352 |
stats => $stats, |
stats => $stats, |
353 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
354 |
|
modify_file => $input->{modify_file}, |
355 |
); |
); |
356 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
357 |
|
|
361 |
%{ $input }, |
%{ $input }, |
362 |
); |
); |
363 |
|
|
364 |
|
my $report_fh; |
365 |
|
if ($stats || $validate) { |
366 |
|
my $path = "out/report/" . $database . '-' . $input->{name} . '.txt'; |
367 |
|
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
368 |
|
|
369 |
|
print $report_fh "Report for database '$database' input '$input->{name}' records ", |
370 |
|
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
371 |
|
$log->info("Generating report file $path"); |
372 |
|
} |
373 |
|
|
374 |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
375 |
@{ $input->{normalize} } : ( $input->{normalize} ); |
@{ $input->{normalize} } : ( $input->{normalize} ); |
376 |
|
|
400 |
# reset position in database |
# reset position in database |
401 |
$input_db->seek(1); |
$input_db->seek(1); |
402 |
|
|
403 |
|
# generate name of config key for indexer (strip everything after -) |
404 |
|
my $indexer_config = $use_indexer; |
405 |
|
$indexer_config =~ s/^(\w+)-?.*$/$1/g if ($indexer_config); |
406 |
|
|
407 |
foreach my $pos ( 0 ... $input_db->size ) { |
foreach my $pos ( 0 ... $input_db->size ) { |
408 |
|
|
409 |
my $row = $input_db->fetch || next; |
my $row = $input_db->fetch || next; |
418 |
|
|
419 |
|
|
420 |
if ($validate) { |
if ($validate) { |
421 |
my @errors = $validate->validate_errors( $row ); |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
422 |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
$log->error( "MFN $mfn validation error:\n", |
423 |
|
$validate->report_error( $errors ) |
424 |
|
); |
425 |
|
} |
426 |
} |
} |
427 |
|
|
428 |
my $ds_config = dclone($db_config); |
my $ds_config = dclone($db_config); |
453 |
$indexer->add( |
$indexer->add( |
454 |
id => $input->{name} . "/" . $mfn, |
id => $input->{name} . "/" . $mfn, |
455 |
ds => $ds, |
ds => $ds, |
456 |
type => $config->{$use_indexer}->{type}, |
type => $config->{$indexer_config}->{type}, |
457 |
) if ($indexer && $ds); |
) if ($indexer && $ds); |
458 |
|
|
459 |
if ($marc) { |
if ($marc) { |
475 |
$total_rows++; |
$total_rows++; |
476 |
} |
} |
477 |
|
|
478 |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
if ($validate) { |
479 |
|
my $errors = $validate->report; |
480 |
|
if ($errors) { |
481 |
|
$log->info("validation errors:\n$errors\n" ); |
482 |
|
print $report_fh "$errors\n" if ($report_fh); |
483 |
|
} |
484 |
|
} |
485 |
|
|
486 |
|
if ($stats) { |
487 |
|
my $s = $input_db->stats; |
488 |
|
$log->info("statistics of fields usage:\n$s"); |
489 |
|
print $report_fh "Statistics of fields usage:\n$s" if ($report_fh); |
490 |
|
} |
491 |
|
|
492 |
# close MARC file |
# close MARC file |
493 |
$marc->finish if ($marc); |
$marc->finish if ($marc); |
494 |
|
|
495 |
|
# close report |
496 |
|
close($report_fh) if ($report_fh) |
497 |
} |
} |
498 |
|
|
499 |
} |
} |