8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Lookup 0.03; |
use WebPAC::Lookup 0.03; |
11 |
use WebPAC::Input 0.10; |
use WebPAC::Input 0.11; |
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 |
|
|
208 |
|
|
209 |
my $indexer; |
my $indexer; |
210 |
if ($use_indexer) { |
if ($use_indexer) { |
211 |
my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration"); |
|
212 |
|
my $cfg_name = $use_indexer; |
213 |
|
$cfg_name =~ s/\-.*$//; |
214 |
|
|
215 |
|
my $indexer_config = $config->{$cfg_name} || $log->logdie("can't find '$cfg_name' part in confguration"); |
216 |
$indexer_config->{database} = $database; |
$indexer_config->{database} = $database; |
217 |
$indexer_config->{clean} = $clean; |
$indexer_config->{clean} = $clean; |
218 |
$indexer_config->{label} = $db_config->{name}; |
$indexer_config->{label} = $db_config->{name}; |
226 |
use WebPAC::Output::Estraier '0.10'; |
use WebPAC::Output::Estraier '0.10'; |
227 |
$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } ); |
$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } ); |
228 |
|
|
229 |
|
} elsif ($use_indexer eq 'hyperestraier-native') { |
230 |
|
|
231 |
|
# open Hyper Estraier database |
232 |
|
use WebPAC::Output::EstraierNative; |
233 |
|
$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } ); |
234 |
|
|
235 |
|
$use_indexer = 'hyperestraier'; |
236 |
|
|
237 |
} elsif ($use_indexer eq 'kinosearch') { |
} elsif ($use_indexer eq 'kinosearch') { |
238 |
|
|
239 |
# open KinoSearch |
# open KinoSearch |
335 |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
336 |
); |
); |
337 |
|
|
338 |
|
if ($stats) { |
339 |
|
# disable modification of records if --stats is in use |
340 |
|
delete($input->{modify_records}); |
341 |
|
delete($input->{modify_file}); |
342 |
|
} |
343 |
|
|
344 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
345 |
module => $input_module, |
module => $input_module, |
346 |
encoding => $config->{webpac}->{webpac_encoding}, |
encoding => $config->{webpac}->{webpac_encoding}, |
353 |
recode => $input->{recode}, |
recode => $input->{recode}, |
354 |
stats => $stats, |
stats => $stats, |
355 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
356 |
|
modify_file => $input->{modify_file}, |
357 |
); |
); |
358 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
359 |
|
|
406 |
|
|
407 |
|
|
408 |
if ($validate) { |
if ($validate) { |
409 |
my @errors = $validate->validate_errors( $row ); |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
410 |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
$log->error( "MFN $mfn validation error:\n", |
411 |
|
dump( $errors ) |
412 |
|
); |
413 |
|
} |
414 |
} |
} |
415 |
|
|
416 |
my $ds_config = dclone($db_config); |
my $ds_config = dclone($db_config); |
463 |
$total_rows++; |
$total_rows++; |
464 |
} |
} |
465 |
|
|
466 |
|
if ($validate && defined($validate->all_errors)) { |
467 |
|
my $validate_errors = $validate->all_errors; |
468 |
|
|
469 |
|
$log->info("validation errors:\n", dump( $validate_errors ) ); |
470 |
|
} |
471 |
|
|
472 |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
473 |
|
|
474 |
# close MARC file |
# close MARC file |