9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.08; |
use WebPAC::Parser 0.08; |
11 |
use WebPAC::Input 0.16; |
use WebPAC::Input 0.16; |
12 |
use WebPAC::Store 0.14; |
use WebPAC::Store 0.15; |
13 |
use WebPAC::Normalize 0.22; |
use WebPAC::Normalize 0.22; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate 0.11; |
use WebPAC::Validate 0.11; |
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 |
die "no longer supported"; |
|
use WebPAC::Output::KinoSearch; |
|
|
$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path}); |
|
|
$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } ); |
|
259 |
|
|
260 |
} else { |
} else { |
261 |
$log->logdie("unknown use_indexer: $use_indexer"); |
$log->logdie("unknown use_indexer: $use_indexer"); |
262 |
} |
} |
263 |
|
|
264 |
$log->logide("can't continue without valid indexer") unless ($indexer); |
$log->logdie("can't continue without valid indexer") unless ($indexer); |
265 |
} |
} |
266 |
|
|
267 |
|
|
295 |
# |
# |
296 |
# now WebPAC::Store |
# now WebPAC::Store |
297 |
# |
# |
298 |
my $abs_path = abs_path($0); |
my $store = new WebPAC::Store({ |
299 |
$abs_path =~ s#/[^/]*$#/#; |
debug => $debug, |
300 |
|
}); |
|
my $db_path = $config->webpac('db_path'); |
|
301 |
|
|
302 |
if ($clean) { |
# |
303 |
$log->info("creating new database '$database' in $db_path"); |
# prepare output |
304 |
rmtree( $db_path ) || $log->warn("can't remove $db_path: $!"); |
# |
305 |
} else { |
my @outputs; |
306 |
$log->info("working on database '$database' in $db_path"); |
if (defined( $db_config->{output} )) { |
307 |
|
my $module = $db_config->{output}->{module} || $log->logdie("need module in output section of $database"); |
308 |
|
$module = 'WebPAC::Output::' . $module unless $module =~ m/::/; |
309 |
|
$log->debug("loading output module $module"); |
310 |
|
eval "require $module"; |
311 |
|
$db_config->{output}->{database} = $database; |
312 |
|
$log->debug("calling $module->new(",dump( $db_config->{output} ),")"); |
313 |
|
my $out = new $module->new( $db_config->{output} ); |
314 |
|
$out->init; |
315 |
|
push @outputs, $out; |
316 |
} |
} |
317 |
|
|
|
my $store = new WebPAC::Store( |
|
|
path => $db_path, |
|
|
debug => $debug, |
|
|
); |
|
|
|
|
318 |
|
|
319 |
# |
# |
320 |
# now, iterate through input formats |
# now, iterate through input formats |
362 |
stats => $stats, |
stats => $stats, |
363 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
364 |
modify_file => $input->{modify_file}, |
modify_file => $input->{modify_file}, |
365 |
|
input_config => $input, |
366 |
); |
); |
367 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
368 |
|
|
507 |
|
|
508 |
my $mfn = $row->{'000'}->[0]; |
my $mfn = $row->{'000'}->[0]; |
509 |
|
|
510 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
if (! $mfn || $mfn !~ m{^\d+$}) { |
511 |
$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"); |
512 |
$mfn = $pos; |
$mfn = $pos; |
513 |
push @{ $row->{'000'} }, $pos; |
push @{ $row->{'000'} }, $pos; |
569 |
|
|
570 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
571 |
} |
} |
572 |
|
|
573 |
|
foreach my $out ( @outputs ) { |
574 |
|
$out->add( $mfn, $ds ) if $out->can('add'); |
575 |
|
} |
576 |
|
|
577 |
} |
} |
578 |
|
|
579 |
if ($validate) { |
if ($validate) { |
604 |
|
|
605 |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
606 |
|
|
607 |
|
foreach my $out ( @outputs ) { |
608 |
|
$out->finish if $out->can('finish'); |
609 |
|
} |
610 |
|
|
611 |
my $dt = time() - $start_t; |
my $dt = time() - $start_t; |
612 |
$log->info("$total_rows records ", $indexer ? "indexed " : "", |
$log->info("$total_rows records ", $indexer ? "indexed " : "", |
613 |
sprintf("in %.2f sec [%.2f rec/sec]", |
sprintf("in %.2f sec [%.2f rec/sec]", |
631 |
} |
} |
632 |
|
|
633 |
# save new delimiters if needed |
# save new delimiters if needed |
634 |
$validate->save_delimiters_templates if ( $validate ); |
$validate->save_delimiters_templates if ( $validate_delimiters_path ); |
635 |
|
|
636 |
# |
# |
637 |
# handle links or merge after indexing |
# handle links or merge after indexing |