185 |
if ($stats) { |
if ($stats) { |
186 |
$log->debug("disabled indexing for stats collection"); |
$log->debug("disabled indexing for stats collection"); |
187 |
$use_indexer = undef; |
$use_indexer = undef; |
188 |
} else { |
} elsif ( $use_indexer ) { |
189 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
190 |
} |
} |
191 |
|
|
244 |
if ($use_indexer eq 'hyperestraier') { |
if ($use_indexer eq 'hyperestraier') { |
245 |
|
|
246 |
# open Hyper Estraier database |
# open Hyper Estraier database |
247 |
use WebPAC::Output::Estraier '0.10'; |
require WebPAC::Output::Estraier; |
248 |
$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } ); |
$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } ); |
249 |
|
|
250 |
} elsif ($use_indexer eq 'hyperestraier-native') { |
} elsif ($use_indexer eq 'hyperestraier-native') { |
251 |
|
|
252 |
# open Hyper Estraier database |
# open Hyper Estraier database |
253 |
use WebPAC::Output::EstraierNative; |
require WebPAC::Output::EstraierNative; |
254 |
$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } ); |
$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } ); |
255 |
|
|
256 |
} elsif ($use_indexer eq 'kinosearch') { |
} elsif ($use_indexer eq 'kinosearch') { |
257 |
|
|
258 |
# open KinoSearch |
# open KinoSearch |
259 |
use WebPAC::Output::KinoSearch; |
require WebPAC::Output::KinoSearch; |
260 |
$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path}); |
$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path}); |
261 |
$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } ); |
$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } ); |
262 |
|
|
264 |
$log->logdie("unknown use_indexer: $use_indexer"); |
$log->logdie("unknown use_indexer: $use_indexer"); |
265 |
} |
} |
266 |
|
|
267 |
$log->logide("can't continue without valid indexer") unless ($indexer); |
$log->logdie("can't continue without valid indexer") unless ($indexer); |
268 |
} |
} |
269 |
|
|
270 |
|
|
299 |
# now WebPAC::Store |
# now WebPAC::Store |
300 |
# |
# |
301 |
my $abs_path = abs_path($0); |
my $abs_path = abs_path($0); |
302 |
$abs_path =~ s#/[^/]*$#/#; |
$abs_path =~ s#/[^/]*$#/#; # |
303 |
|
|
304 |
my $db_path = $config->webpac('db_path'); |
my $db_path = $config->webpac('db_path'); |
305 |
|
|
315 |
debug => $debug, |
debug => $debug, |
316 |
); |
); |
317 |
|
|
318 |
|
# |
319 |
|
# prepare output |
320 |
|
# |
321 |
|
my @outputs; |
322 |
|
if (defined( $db_config->{output} )) { |
323 |
|
my $module = $db_config->{output}->{module} || $log->logdie("need module in output section of $database"); |
324 |
|
$module = 'WebPAC::Output::' . $module unless $module =~ m/::/; |
325 |
|
$log->debug("loading output module $module"); |
326 |
|
eval "require $module"; |
327 |
|
my $out = new $module->new( $db_config->{output} ); |
328 |
|
$out->init; |
329 |
|
push @outputs, $out; |
330 |
|
} |
331 |
|
|
332 |
|
|
333 |
# |
# |
334 |
# now, iterate through input formats |
# now, iterate through input formats |
376 |
stats => $stats, |
stats => $stats, |
377 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
378 |
modify_file => $input->{modify_file}, |
modify_file => $input->{modify_file}, |
379 |
|
input_config => $input, |
380 |
); |
); |
381 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
382 |
|
|
463 |
$log->info("Generating report file $path"); |
$log->info("Generating report file $path"); |
464 |
|
|
465 |
if ( $validate ) { |
if ( $validate ) { |
466 |
$validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ); |
$validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ) if ( $validate_path ); |
467 |
$validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ); |
$validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ) if ( $validate_delimiters_path ); |
468 |
} |
} |
469 |
} |
} |
470 |
|
|
521 |
|
|
522 |
my $mfn = $row->{'000'}->[0]; |
my $mfn = $row->{'000'}->[0]; |
523 |
|
|
524 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
if (! $mfn || $mfn !~ m{^\d+$}) { |
525 |
$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos"); |
$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos"); |
526 |
$mfn = $pos; |
$mfn = $pos; |
527 |
push @{ $row->{'000'} }, $pos; |
push @{ $row->{'000'} }, $pos; |
583 |
|
|
584 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
585 |
} |
} |
586 |
|
|
587 |
|
foreach my $out ( @outputs ) { |
588 |
|
$out->add( $mfn, $ds ) if $out->can('add'); |
589 |
|
} |
590 |
|
|
591 |
} |
} |
592 |
|
|
593 |
if ($validate) { |
if ($validate) { |
618 |
|
|
619 |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
620 |
|
|
621 |
|
foreach my $out ( @outputs ) { |
622 |
|
$out->finish if $out->can('finish'); |
623 |
|
} |
624 |
|
|
625 |
my $dt = time() - $start_t; |
my $dt = time() - $start_t; |
626 |
$log->info("$total_rows records ", $indexer ? "indexed " : "", |
$log->info("$total_rows records ", $indexer ? "indexed " : "", |
627 |
sprintf("in %.2f sec [%.2f rec/sec]", |
sprintf("in %.2f sec [%.2f rec/sec]", |
645 |
} |
} |
646 |
|
|
647 |
# save new delimiters if needed |
# save new delimiters if needed |
648 |
$validate->save_delimiters_templates if ( $validate ); |
$validate->save_delimiters_templates if ( $validate_delimiters_path ); |
649 |
|
|
650 |
# |
# |
651 |
# handle links or merge after indexing |
# handle links or merge after indexing |