7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.07; |
use WebPAC::Parser 0.08; |
11 |
use WebPAC::Input 0.13; |
use WebPAC::Input 0.16; |
12 |
use WebPAC::Store 0.11; |
use WebPAC::Store 0.14; |
13 |
use WebPAC::Normalize 0.22; |
use WebPAC::Normalize 0.22; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate 0.06; |
use WebPAC::Validate 0.06; |
21 |
use File::Slurp; |
use File::Slurp; |
22 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
23 |
use Storable qw/dclone/; |
use Storable qw/dclone/; |
24 |
|
use Pod::Usage qw/pod2usage/; |
25 |
|
|
26 |
use Proc::Queue size => 1; |
use Proc::Queue size => 1; |
27 |
use POSIX ":sys_wait_h"; # imports WNOHANG |
use POSIX ":sys_wait_h"; # imports WNOHANG |
32 |
|
|
33 |
B<this command will probably go away. Don't get used to it!> |
B<this command will probably go away. Don't get used to it!> |
34 |
|
|
35 |
Options: |
=head1 OPTIONS |
36 |
|
|
37 |
=over 4 |
=over 4 |
38 |
|
|
108 |
my $parallel = 0; |
my $parallel = 0; |
109 |
my $only_links = 0; |
my $only_links = 0; |
110 |
my $merge = 0; |
my $merge = 0; |
111 |
|
my $help; |
112 |
|
|
113 |
my $log = _new WebPAC::Common()->_get_logger(); |
my $log = _new WebPAC::Common()->_get_logger(); |
114 |
|
|
127 |
"parallel=i" => \$parallel, |
"parallel=i" => \$parallel, |
128 |
"only-links!" => \$only_links, |
"only-links!" => \$only_links, |
129 |
"merge" => \$merge, |
"merge" => \$merge, |
130 |
|
"help" => \$help, |
131 |
); |
); |
132 |
|
|
133 |
|
pod2usage(-verbose => 2) if ($help); |
134 |
|
|
135 |
my $config = new WebPAC::Config( path => $config_path ); |
my $config = new WebPAC::Config( path => $config_path ); |
136 |
|
|
137 |
#print "config = ",dump($config) if ($debug); |
#print "config = ",dump($config) if ($debug); |
140 |
|
|
141 |
$log->info( "-" x 79 ); |
$log->info( "-" x 79 ); |
142 |
|
|
143 |
|
my $log_file = 'log'; |
144 |
|
|
145 |
|
if (-e $log_file ) { # && -s $log_file > 5 * 1024 * 1024) { |
146 |
|
$log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'"); |
147 |
|
rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!"); |
148 |
|
} |
149 |
|
|
150 |
my $estcmd_fh; |
my $estcmd_fh; |
151 |
my $estcmd_path = './estcmd-merge.sh'; |
my $estcmd_path = './estcmd-merge.sh'; |
164 |
|
|
165 |
|
|
166 |
my $use_indexer = $config->use_indexer; |
my $use_indexer = $config->use_indexer; |
167 |
|
$stats ||= $validate; |
168 |
if ($stats) { |
if ($stats) { |
169 |
$log->debug("option --stats disables update of indexing engine..."); |
$log->debug("disabled indexing for stats collection"); |
170 |
$use_indexer = undef; |
$use_indexer = undef; |
171 |
} else { |
} else { |
172 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
210 |
} |
} |
211 |
|
|
212 |
my $indexer; |
my $indexer; |
213 |
if ($use_indexer) { |
if ($use_indexer && $parser->have_rules( 'search', $database )) { |
214 |
|
|
215 |
my $cfg_name = $use_indexer; |
my $cfg_name = $use_indexer; |
216 |
$cfg_name =~ s/\-.*$//; |
$cfg_name =~ s/\-.*$//; |
362 |
my $rec = shift || die "need rec!"; |
my $rec = shift || die "need rec!"; |
363 |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
364 |
|
|
|
$store->save_row( |
|
|
database => $database, |
|
|
input => $input_name, |
|
|
id => $mfn, |
|
|
row => $rec, |
|
|
); |
|
|
|
|
365 |
WebPAC::Normalize::data_structure( |
WebPAC::Normalize::data_structure( |
366 |
row => $rec, |
row => $rec, |
367 |
rules => $rules, |
rules => $rules, |
385 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
386 |
lookup => $lookup_jar, |
lookup => $lookup_jar, |
387 |
%{ $input }, |
%{ $input }, |
388 |
|
load_row => sub { |
389 |
|
my $a = shift; |
390 |
|
return $store->load_row( |
391 |
|
database => $database, |
392 |
|
input => $input_name, |
393 |
|
id => $a->{id}, |
394 |
|
); |
395 |
|
}, |
396 |
|
save_row => sub { |
397 |
|
my $a = shift; |
398 |
|
return $store->save_row( |
399 |
|
database => $database, |
400 |
|
input => $input_name, |
401 |
|
id => $a->{id}, |
402 |
|
row => $a->{row}, |
403 |
|
); |
404 |
|
}, |
405 |
|
|
406 |
); |
); |
407 |
|
|
408 |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
409 |
|
|
410 |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
411 |
$log->debug("created following lookups: ", dump( $lookup_data )); |
$log->debug("created following lookups: ", sub { dump( $lookup_data ) } ); |
412 |
|
|
413 |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
414 |
$store->save_lookup( |
$store->save_lookup( |
431 |
} |
} |
432 |
|
|
433 |
my $marc; |
my $marc; |
434 |
if ($parser->generate_marc( $database, $input_name )) { |
if ($parser->have_rules( 'marc', $database, $input_name )) { |
435 |
$marc = new WebPAC::Output::MARC( |
$marc = new WebPAC::Output::MARC( |
436 |
path => "out/marc/${database}-${input_name}.marc", |
path => "out/marc/${database}-${input_name}.marc", |
437 |
lint => $marc_lint, |
lint => $marc_lint, |
459 |
foreach my $db (keys %$depends) { |
foreach my $db (keys %$depends) { |
460 |
foreach my $i (keys %{$depends->{$db}}) { |
foreach my $i (keys %{$depends->{$db}}) { |
461 |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
462 |
|
my $t = time(); |
463 |
$log->debug("loading lookup $db/$i"); |
$log->debug("loading lookup $db/$i"); |
464 |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
465 |
database => $db, |
database => $db, |
466 |
input => $i, |
input => $i, |
467 |
key => $k, |
key => $k, |
468 |
); |
); |
469 |
|
$log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t)); |
470 |
} |
} |
471 |
} |
} |
472 |
} |
} |
473 |
|
|
474 |
$log->debug("lookup_hash = ", dump( $lookup_hash )); |
$log->debug("lookup_hash = ", sub { dump( $lookup_hash ) }); |
475 |
} |
} |
476 |
|
|
477 |
|
|
479 |
|
|
480 |
my $row = $input_db->fetch || next; |
my $row = $input_db->fetch || next; |
481 |
|
|
482 |
|
$total_rows++; |
483 |
|
|
484 |
my $mfn = $row->{'000'}->[0]; |
my $mfn = $row->{'000'}->[0]; |
485 |
|
|
486 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
if (! $mfn || $mfn !~ m#^\d+$#) { |
491 |
|
|
492 |
|
|
493 |
if ($validate) { |
if ($validate) { |
494 |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { |
495 |
$log->error( "MFN $mfn validation error:\n", |
$log->error( "MFN $mfn validation error:\n", |
496 |
$validate->report_error( $errors ) |
$validate->report_error( $errors ) |
497 |
); |
); |
498 |
} |
} |
499 |
|
next; # validation doesn't create any output |
500 |
} |
} |
501 |
|
|
502 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
515 |
}, |
}, |
516 |
); |
); |
517 |
|
|
518 |
$log->debug("ds = ",dump($ds)); |
$log->debug("ds = ", sub { dump($ds) }) if ($ds); |
519 |
|
|
520 |
$store->save_ds( |
$store->save_ds( |
521 |
database => $database, |
database => $database, |
545 |
|
|
546 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
547 |
} |
} |
|
|
|
|
$total_rows++; |
|
548 |
} |
} |
549 |
|
|
550 |
if ($validate) { |
if ($validate) { |