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 0.07; |
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 |
|
|
117 |
my $only_links = 0; |
my $only_links = 0; |
118 |
my $merge = 0; |
my $merge = 0; |
119 |
|
|
120 |
|
my $log = _new WebPAC::Common()->_get_logger(); |
121 |
|
|
122 |
|
my $hostname = `hostname`; |
123 |
|
chomp($hostname); |
124 |
|
$hostname =~ s/\..+$//; |
125 |
|
if (-e "conf/$hostname.yml") { |
126 |
|
$config = "conf/$hostname.yml"; |
127 |
|
$log->info("using host configuration file: $config"); |
128 |
|
} |
129 |
|
|
130 |
GetOptions( |
GetOptions( |
131 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
132 |
"offset=i" => \$offset, |
"offset=i" => \$offset, |
148 |
|
|
149 |
$config = LoadFile($config); |
$config = LoadFile($config); |
150 |
|
|
151 |
print "config = ",dump($config) if ($debug); |
#print "config = ",dump($config) if ($debug); |
152 |
|
|
153 |
die "no databases in config file!\n" unless ($config->{databases}); |
die "no databases in config file!\n" unless ($config->{databases}); |
154 |
|
|
|
my $log = _new WebPAC::Common()->_get_logger(); |
|
155 |
$log->info( "-" x 79 ); |
$log->info( "-" x 79 ); |
156 |
|
|
157 |
|
|
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 |
|
|
363 |
%{ $input }, |
%{ $input }, |
364 |
); |
); |
365 |
|
|
366 |
|
my $report_fh; |
367 |
|
if ($stats || $validate) { |
368 |
|
my $path = "out/report/" . $database . '-' . $input->{name} . '.txt'; |
369 |
|
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
370 |
|
|
371 |
|
print $report_fh "Report for database '$database' input '$input->{name}' records ", |
372 |
|
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
373 |
|
$log->info("Generating report file $path"); |
374 |
|
} |
375 |
|
|
376 |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
377 |
@{ $input->{normalize} } : ( $input->{normalize} ); |
@{ $input->{normalize} } : ( $input->{normalize} ); |
378 |
|
|
416 |
|
|
417 |
|
|
418 |
if ($validate) { |
if ($validate) { |
419 |
my @errors = $validate->validate_errors( $row ); |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
420 |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
$log->error( "MFN $mfn validation error:\n", |
421 |
|
$validate->report_error( $errors ) |
422 |
|
); |
423 |
|
} |
424 |
} |
} |
425 |
|
|
426 |
my $ds_config = dclone($db_config); |
my $ds_config = dclone($db_config); |
473 |
$total_rows++; |
$total_rows++; |
474 |
} |
} |
475 |
|
|
476 |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
if ($validate) { |
477 |
|
my $errors = $validate->report; |
478 |
|
if ($errors) { |
479 |
|
$log->info("validation errors:\n$errors\n" ); |
480 |
|
print $report_fh "$errors\n" if ($report_fh); |
481 |
|
} |
482 |
|
} |
483 |
|
|
484 |
|
if ($stats) { |
485 |
|
my $s = $input_db->stats; |
486 |
|
$log->info("statistics of fields usage:\n$s"); |
487 |
|
print $report_fh "Statistics of fields usage:\n$s" if ($report_fh); |
488 |
|
} |
489 |
|
|
490 |
# close MARC file |
# close MARC file |
491 |
$marc->finish if ($marc); |
$marc->finish if ($marc); |
492 |
|
|
493 |
|
# close report |
494 |
|
close($report_fh) || $log->logdie("can't close report: $!"); |
495 |
} |
} |
496 |
|
|
497 |
} |
} |