8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.04; |
use WebPAC::Parser 0.04; |
11 |
use WebPAC::Lookup 0.03; |
use WebPAC::Input 0.13; |
12 |
use WebPAC::Input 0.11; |
use WebPAC::Store 0.10; |
|
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 0.06; |
use WebPAC::Validate 0.06; |
188 |
Proc::Queue::size($parallel); |
Proc::Queue::size($parallel); |
189 |
} |
} |
190 |
|
|
191 |
|
sub create_ds_config { |
192 |
|
my ($db_config, $database, $input, $mfn) = @_; |
193 |
|
my $c = dclone( $db_config ); |
194 |
|
$c->{_} = $database || $log->logconfess("need database"); |
195 |
|
$c->{_mfn} = $mfn || $log->logconfess("need mfn"); |
196 |
|
$c->{input} = $input || $log->logconfess("need input"); |
197 |
|
return $c; |
198 |
|
} |
199 |
|
|
200 |
while (my ($database, $db_config) = each %{ $config->databases }) { |
while (my ($database, $db_config) = each %{ $config->databases }) { |
201 |
|
|
202 |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
294 |
$log->info("working on database '$database' in $db_path"); |
$log->info("working on database '$database' in $db_path"); |
295 |
} |
} |
296 |
|
|
297 |
my $db = new WebPAC::Store( |
my $store = new WebPAC::Store( |
298 |
path => $db_path, |
path => $db_path, |
299 |
database => $database, |
database => $database, |
300 |
debug => $debug, |
debug => $debug, |
316 |
|
|
317 |
foreach my $input (@inputs) { |
foreach my $input (@inputs) { |
318 |
|
|
319 |
next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i)); |
my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input)); |
320 |
|
|
321 |
|
next if ($only_input && ($input_name !~ m#$only_input#i && $input->{type} !~ m#$only_input#i)); |
322 |
|
|
323 |
my $type = lc($input->{type}); |
my $type = lc($input->{type}); |
324 |
|
|
326 |
|
|
327 |
my $input_module = $config->webpac('inputs')->{$type}; |
my $input_module = $config->webpac('inputs')->{$type}; |
328 |
|
|
329 |
$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module", |
my @lookups = $parser->have_lookup_create($database, $input); |
330 |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
|
331 |
|
$log->info("working on input '$input_name' in $input->{path} [type: $input->{type}] using $input_module", |
332 |
|
@lookups ? " creating lookups: ".join(", ", @lookups) : "" |
333 |
); |
); |
334 |
|
|
335 |
if ($stats) { |
if ($stats) { |
338 |
delete($input->{modify_file}); |
delete($input->{modify_file}); |
339 |
} |
} |
340 |
|
|
|
warn "parser->depends = ", dump( $parser->{depends} ); |
|
|
warn "depends on: ", dump( $parser->depends($database, $input->{name}) ); |
|
|
warn "lookup_create_rules = ", dump( $parser->lookup_create_rules($database, $input->{name}) ); |
|
|
warn "parser->_lookup_create = ", dump( $parser->{_lookup_create} ); |
|
|
|
|
|
my $lookup; |
|
|
|
|
341 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
342 |
module => $input_module, |
module => $input_module, |
343 |
encoding => $config->webpac('webpac_encoding'), |
encoding => $config->webpac('webpac_encoding'), |
344 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
345 |
offset => $offset, |
offset => $offset, |
|
lookup_coderef => sub { |
|
|
my $rec = shift || return; |
|
|
$lookup->add( $rec ); |
|
|
}, |
|
346 |
recode => $input->{recode}, |
recode => $input->{recode}, |
347 |
stats => $stats, |
stats => $stats, |
348 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
350 |
); |
); |
351 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
352 |
|
|
353 |
|
if (defined( $input->{lookup} )) { |
354 |
|
$log->warn("$database/$input_name has depriciated lookup definition, removing it..."); |
355 |
|
delete( $input->{lookup} ); |
356 |
|
} |
357 |
|
|
358 |
|
my $lookup; |
359 |
|
my $lookup_coderef; |
360 |
|
|
361 |
|
if (@lookups) { |
362 |
|
|
363 |
|
my $rules = $parser->lookup_create_rules($database, $input) || $log->logdie("no rules found for $database/$input"); |
364 |
|
|
365 |
|
$lookup_coderef = sub { |
366 |
|
my $rec = shift || die "need rec!"; |
367 |
|
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
368 |
|
|
369 |
|
WebPAC::Normalize::data_structure( |
370 |
|
row => $rec, |
371 |
|
rules => $rules, |
372 |
|
lookup => $lookup, |
373 |
|
config => create_ds_config( $db_config, $database, $input, $mfn ), |
374 |
|
); |
375 |
|
|
376 |
|
warn "current lookup = ", dump($lookup) if ($lookup); |
377 |
|
}; |
378 |
|
|
379 |
|
WebPAC::Normalize::_set_lookup( undef ); |
380 |
|
|
381 |
|
$log->debug("created lookup_coderef using:\n$rules"); |
382 |
|
|
383 |
|
}; |
384 |
|
|
385 |
my $maxmfn = $input_db->open( |
my $maxmfn = $input_db->open( |
386 |
path => $input->{path}, |
path => $input->{path}, |
387 |
code_page => $input->{encoding}, # database encoding |
code_page => $input->{encoding}, # database encoding |
388 |
|
lookup_coderef => $lookup_coderef, |
389 |
%{ $input }, |
%{ $input }, |
390 |
); |
); |
391 |
|
|
392 |
|
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
393 |
|
|
394 |
|
$log->debug("created following lookups: ", dump( $lookup_data )); |
395 |
|
|
396 |
|
foreach my $key (keys %$lookup_data) { |
397 |
|
$store->save_lookup( $database, $input_name, $key, $lookup_data->{$key} ); |
398 |
|
} |
399 |
|
|
400 |
my $report_fh; |
my $report_fh; |
401 |
if ($stats || $validate) { |
if ($stats || $validate) { |
402 |
my $path = "out/report/" . $database . '-' . $input->{name} . '.txt'; |
my $path = "out/report/${database}-${input_name}.txt"; |
403 |
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
404 |
|
|
405 |
print $report_fh "Report for database '$database' input '$input->{name}' records ", |
print $report_fh "Report for database '$database' input '$input_name' records ", |
406 |
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
407 |
$log->info("Generating report file $path"); |
$log->info("Generating report file $path"); |
408 |
} |
} |
413 |
if ($marc_normalize) { |
if ($marc_normalize) { |
414 |
@norm_array = ( { |
@norm_array = ( { |
415 |
path => $marc_normalize, |
path => $marc_normalize, |
416 |
output => $marc_output || 'out/marc/' . $database . '-' . $input->{name} . '.marc', |
output => $marc_output || "out/marc/${database}-${input_name}.marc", |
417 |
} ); |
} ); |
418 |
} |
} |
419 |
|
|
461 |
} |
} |
462 |
} |
} |
463 |
|
|
|
my $ds_config = dclone($db_config); |
|
|
|
|
|
# default values -> database key |
|
|
$ds_config->{_} = $database; |
|
|
|
|
|
# current mfn |
|
|
$ds_config->{_mfn} = $mfn; |
|
|
|
|
|
# attach current input |
|
|
$ds_config->{input} = $input; |
|
|
|
|
464 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
465 |
row => $row, |
row => $row, |
466 |
rules => $rules, |
rules => $rules, |
467 |
lookup => $lookup ? $lookup->lookup_hash : undef, |
lookup => $lookup ? $lookup->lookup_hash : undef, |
468 |
config => $ds_config, |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
469 |
marc_encoding => 'utf-8', |
marc_encoding => 'utf-8', |
470 |
); |
); |
471 |
|
|
472 |
$db->save_ds( |
$store->save_ds( |
473 |
id => $mfn, |
id => $mfn, |
474 |
ds => $ds, |
ds => $ds, |
475 |
prefix => $input->{name}, |
prefix => $input_name, |
476 |
) if ($ds && !$stats); |
) if ($ds && !$stats); |
477 |
|
|
478 |
$indexer->add( |
$indexer->add( |
479 |
id => $input->{name} . "/" . $mfn, |
id => "${input_name}/${mfn}", |
480 |
ds => $ds, |
ds => $ds, |
481 |
type => $config->get($indexer_config)->{type}, |
type => $config->get($indexer_config)->{type}, |
482 |
) if ($indexer && $ds); |
) if ($indexer && $ds); |