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; |
|
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; |
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); |
324 |
|
|
325 |
my $input_module = $config->webpac('inputs')->{$type}; |
my $input_module = $config->webpac('inputs')->{$type}; |
326 |
|
|
327 |
|
my @lookups = $parser->have_lookup_create($database, $input); |
328 |
|
|
329 |
$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module", |
$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module", |
330 |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
@lookups ? " creating lookups: ".join(", ", @lookups) : "" |
331 |
); |
); |
332 |
|
|
333 |
if ($stats) { |
if ($stats) { |
336 |
delete($input->{modify_file}); |
delete($input->{modify_file}); |
337 |
} |
} |
338 |
|
|
|
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; |
|
|
|
|
339 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
340 |
module => $input_module, |
module => $input_module, |
341 |
encoding => $config->webpac('webpac_encoding'), |
encoding => $config->webpac('webpac_encoding'), |
342 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
343 |
offset => $offset, |
offset => $offset, |
|
lookup_coderef => sub { |
|
|
my $rec = shift || return; |
|
|
$lookup->add( $rec ); |
|
|
}, |
|
344 |
recode => $input->{recode}, |
recode => $input->{recode}, |
345 |
stats => $stats, |
stats => $stats, |
346 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
348 |
); |
); |
349 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
350 |
|
|
351 |
|
if (defined( $input->{lookup} )) { |
352 |
|
$log->warn("$database/", $input->{name}, " has depriciated lookup definition, removing it..."); |
353 |
|
delete( $input->{lookup} ); |
354 |
|
} |
355 |
|
|
356 |
|
my $lookup; |
357 |
|
my $lookup_coderef; |
358 |
|
|
359 |
|
if (@lookups) { |
360 |
|
|
361 |
|
my $rules = $parser->lookup_create_rules($database, $input) || $log->logdie("no rules found for $database/$input"); |
362 |
|
|
363 |
|
$lookup_coderef = sub { |
364 |
|
my $rec = shift || die "need rec!"; |
365 |
|
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
366 |
|
|
367 |
|
WebPAC::Normalize::data_structure( |
368 |
|
row => $rec, |
369 |
|
rules => $rules, |
370 |
|
lookup => $lookup, |
371 |
|
config => create_ds_config( $db_config, $database, $input, $mfn ), |
372 |
|
); |
373 |
|
|
374 |
|
warn "current lookup = ", dump($lookup) if ($lookup); |
375 |
|
}; |
376 |
|
|
377 |
|
WebPAC::Normalize::_set_lookup( undef ); |
378 |
|
|
379 |
|
$log->debug("created lookup_coderef using:\n$rules"); |
380 |
|
|
381 |
|
}; |
382 |
|
|
383 |
my $maxmfn = $input_db->open( |
my $maxmfn = $input_db->open( |
384 |
path => $input->{path}, |
path => $input->{path}, |
385 |
code_page => $input->{encoding}, # database encoding |
code_page => $input->{encoding}, # database encoding |
386 |
|
lookup_coderef => $lookup_coderef, |
387 |
%{ $input }, |
%{ $input }, |
388 |
); |
); |
389 |
|
|
390 |
|
$log->debug("created following lookups: ", dump( WebPAC::Normalize::_get_lookup() ) ); |
391 |
|
|
392 |
my $report_fh; |
my $report_fh; |
393 |
if ($stats || $validate) { |
if ($stats || $validate) { |
394 |
my $path = "out/report/" . $database . '-' . $input->{name} . '.txt'; |
my $path = "out/report/" . $database . '-' . $input->{name} . '.txt'; |
453 |
} |
} |
454 |
} |
} |
455 |
|
|
|
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; |
|
|
|
|
456 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
457 |
row => $row, |
row => $row, |
458 |
rules => $rules, |
rules => $rules, |
459 |
lookup => $lookup ? $lookup->lookup_hash : undef, |
lookup => $lookup ? $lookup->lookup_hash : undef, |
460 |
config => $ds_config, |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
461 |
marc_encoding => 'utf-8', |
marc_encoding => 'utf-8', |
462 |
); |
); |
463 |
|
|