7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Lookup 0.03; |
use WebPAC::Parser 0.04; |
11 |
use WebPAC::Input 0.11; |
use WebPAC::Input 0.13; |
12 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.11; |
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; |
174 |
# disable indexing when creating marc |
# disable indexing when creating marc |
175 |
$use_indexer = undef if ($marc_normalize); |
$use_indexer = undef if ($marc_normalize); |
176 |
|
|
177 |
|
# parse normalize files and create source files for lookup and normalization |
178 |
|
|
179 |
|
my $parser = new WebPAC::Parser( config => $config ); |
180 |
|
|
181 |
my $total_rows = 0; |
my $total_rows = 0; |
182 |
my $start_t = time(); |
my $start_t = time(); |
183 |
|
|
188 |
Proc::Queue::size($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); |
285 |
my $abs_path = abs_path($0); |
my $abs_path = abs_path($0); |
286 |
$abs_path =~ s#/[^/]*$#/#; |
$abs_path =~ s#/[^/]*$#/#; |
287 |
|
|
288 |
my $db_path = $config->get('webpac')->{db_path} . '/' . $database; |
my $db_path = $config->webpac('db_path'); |
289 |
|
|
290 |
if ($clean) { |
if ($clean) { |
291 |
$log->info("creating new database '$database' in $db_path"); |
$log->info("creating new database '$database' in $db_path"); |
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, |
|
database => $database, |
|
299 |
debug => $debug, |
debug => $debug, |
300 |
); |
); |
301 |
|
|
315 |
|
|
316 |
foreach my $input (@inputs) { |
foreach my $input (@inputs) { |
317 |
|
|
318 |
next if ($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)); |
319 |
|
|
320 |
|
next if ($only_input && ($input_name !~ m#$only_input#i && $input->{type} !~ m#$only_input#i)); |
321 |
|
|
322 |
my $type = lc($input->{type}); |
my $type = lc($input->{type}); |
323 |
|
|
324 |
die "I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!\n" unless (grep(/$type/, $config->webpac('inputs'))); |
die "I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!\n" unless (grep(/$type/, $config->webpac('inputs'))); |
325 |
|
|
|
my $lookup; |
|
|
if ($input->{lookup}) { |
|
|
$lookup = new WebPAC::Lookup( |
|
|
lookup_file => $input->{lookup}, |
|
|
); |
|
|
delete( $input->{lookup} ); |
|
|
} |
|
|
|
|
326 |
my $input_module = $config->webpac('inputs')->{$type}; |
my $input_module = $config->webpac('inputs')->{$type}; |
327 |
|
|
328 |
$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module", |
my @lookups = $parser->have_lookup_create($database, $input); |
329 |
$input->{lookup} ? "lookup '$input->{lookup}'" : "" |
|
330 |
|
$log->info("working on input '$input_name' in $input->{path} [type: $input->{type}] using $input_module", |
331 |
|
@lookups ? " creating lookups: ".join(", ", @lookups) : "" |
332 |
); |
); |
333 |
|
|
334 |
if ($stats) { |
if ($stats) { |
342 |
encoding => $config->webpac('webpac_encoding'), |
encoding => $config->webpac('webpac_encoding'), |
343 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
344 |
offset => $offset, |
offset => $offset, |
|
lookup_coderef => sub { |
|
|
my $rec = shift || return; |
|
|
$lookup->add( $rec ); |
|
|
}, |
|
345 |
recode => $input->{recode}, |
recode => $input->{recode}, |
346 |
stats => $stats, |
stats => $stats, |
347 |
modify_records => $input->{modify_records}, |
modify_records => $input->{modify_records}, |
349 |
); |
); |
350 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
351 |
|
|
352 |
|
if (defined( $input->{lookup} )) { |
353 |
|
$log->warn("$database/$input_name has depriciated lookup definition, removing it..."); |
354 |
|
delete( $input->{lookup} ); |
355 |
|
} |
356 |
|
|
357 |
|
my $lookup; |
358 |
|
my $lookup_coderef; |
359 |
|
|
360 |
|
if (@lookups) { |
361 |
|
|
362 |
|
my $rules = $parser->lookup_create_rules($database, $input) || $log->logdie("no rules found for $database/$input"); |
363 |
|
|
364 |
|
$lookup_coderef = sub { |
365 |
|
my $rec = shift || die "need rec!"; |
366 |
|
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
367 |
|
|
368 |
|
WebPAC::Normalize::data_structure( |
369 |
|
row => $rec, |
370 |
|
rules => $rules, |
371 |
|
lookup => $lookup, |
372 |
|
config => create_ds_config( $db_config, $database, $input, $mfn ), |
373 |
|
); |
374 |
|
|
375 |
|
warn "current lookup = ", dump($lookup) if ($lookup); |
376 |
|
}; |
377 |
|
|
378 |
|
WebPAC::Normalize::_set_lookup( undef ); |
379 |
|
|
380 |
|
$log->debug("created lookup_coderef using:\n$rules"); |
381 |
|
|
382 |
|
}; |
383 |
|
|
384 |
my $maxmfn = $input_db->open( |
my $maxmfn = $input_db->open( |
385 |
path => $input->{path}, |
path => $input->{path}, |
386 |
code_page => $input->{encoding}, # database encoding |
code_page => $input->{encoding}, # database encoding |
387 |
|
lookup_coderef => $lookup_coderef, |
388 |
%{ $input }, |
%{ $input }, |
389 |
); |
); |
390 |
|
|
391 |
|
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
392 |
|
|
393 |
|
if (defined( $lookup_data->{$database}->{$input_name} )) { |
394 |
|
$log->debug("created following lookups: ", dump( $lookup_data )); |
395 |
|
|
396 |
|
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
397 |
|
$store->save_lookup( |
398 |
|
database => $database, |
399 |
|
input => $input_name, |
400 |
|
key => $key, |
401 |
|
data => $lookup_data->{$database}->{$input_name}->{$key}, |
402 |
|
); |
403 |
|
} |
404 |
|
} |
405 |
|
|
406 |
my $report_fh; |
my $report_fh; |
407 |
if ($stats || $validate) { |
if ($stats || $validate) { |
408 |
my $path = "out/report/" . $database . '-' . $input->{name} . '.txt'; |
my $path = "out/report/${database}-${input_name}.txt"; |
409 |
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
410 |
|
|
411 |
print $report_fh "Report for database '$database' input '$input->{name}' records ", |
print $report_fh "Report for database '$database' input '$input_name' records ", |
412 |
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
413 |
$log->info("Generating report file $path"); |
$log->info("Generating report file $path"); |
414 |
} |
} |
419 |
if ($marc_normalize) { |
if ($marc_normalize) { |
420 |
@norm_array = ( { |
@norm_array = ( { |
421 |
path => $marc_normalize, |
path => $marc_normalize, |
422 |
output => $marc_output || 'out/marc/' . $database . '-' . $input->{name} . '.marc', |
output => $marc_output || "out/marc/${database}-${input_name}.marc", |
423 |
} ); |
} ); |
424 |
} |
} |
425 |
|
|
467 |
} |
} |
468 |
} |
} |
469 |
|
|
|
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; |
|
|
|
|
470 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
471 |
row => $row, |
row => $row, |
472 |
rules => $rules, |
rules => $rules, |
473 |
lookup => $lookup ? $lookup->lookup_hash : undef, |
lookup => $lookup ? $lookup->lookup_hash : undef, |
474 |
config => $ds_config, |
config => create_ds_config( $db_config, $database, $input, $mfn ), |
475 |
marc_encoding => 'utf-8', |
marc_encoding => 'utf-8', |
476 |
); |
); |
477 |
|
|
478 |
$db->save_ds( |
$store->save_ds( |
479 |
|
database => $database, |
480 |
|
input => $input_name, |
481 |
id => $mfn, |
id => $mfn, |
482 |
ds => $ds, |
ds => $ds, |
|
prefix => $input->{name}, |
|
483 |
) if ($ds && !$stats); |
) if ($ds && !$stats); |
484 |
|
|
485 |
$indexer->add( |
$indexer->add( |
486 |
id => $input->{name} . "/" . $mfn, |
id => "${input_name}/${mfn}", |
487 |
ds => $ds, |
ds => $ds, |
488 |
type => $config->get($indexer_config)->{type}, |
type => $config->get($indexer_config)->{type}, |
489 |
) if ($indexer && $ds); |
) if ($indexer && $ds); |