8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.08; |
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 0.11; |
16 |
use WebPAC::Output::MARC; |
use WebPAC::Output::MARC; |
17 |
use WebPAC::Config; |
use WebPAC::Config; |
18 |
use Getopt::Long; |
use Getopt::Long; |
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 |
|
|
68 |
|
|
69 |
turn on extra validation of imput records, see L<WebPAC::Validation> |
turn on extra validation of imput records, see L<WebPAC::Validation> |
70 |
|
|
71 |
|
=item --marc-generate |
72 |
|
|
73 |
|
Generate MARC file. This will automatically be on if file contains C<marc*> directives. |
74 |
|
You can use this option as C<--no-marc-generate> to disable MARC generation. |
75 |
|
|
76 |
=item --marc-lint |
=item --marc-lint |
77 |
|
|
78 |
By default turned on if normalisation file has C<marc*> directives. You can disable lint |
By default turned on if normalisation file has C<marc*> directives. You can disable lint |
108 |
my $only_filter; |
my $only_filter; |
109 |
my $stats = 0; |
my $stats = 0; |
110 |
my $validate_path; |
my $validate_path; |
111 |
|
my $marc_generate = 1; |
112 |
my $marc_lint = 1; |
my $marc_lint = 1; |
113 |
my $marc_dump = 0; |
my $marc_dump = 0; |
114 |
my $parallel = 0; |
my $parallel = 0; |
115 |
my $only_links = 0; |
my $only_links = 0; |
116 |
my $merge = 0; |
my $merge = 0; |
117 |
|
my $help; |
118 |
|
|
119 |
my $log = _new WebPAC::Common()->_get_logger(); |
my $log = _new WebPAC::Common()->_get_logger(); |
120 |
|
|
128 |
"debug+" => \$debug, |
"debug+" => \$debug, |
129 |
"stats" => \$stats, |
"stats" => \$stats, |
130 |
"validate=s" => \$validate_path, |
"validate=s" => \$validate_path, |
131 |
|
"marc-generate!" => \$marc_generate, |
132 |
"marc-lint!" => \$marc_lint, |
"marc-lint!" => \$marc_lint, |
133 |
"marc-dump!" => \$marc_dump, |
"marc-dump!" => \$marc_dump, |
134 |
"parallel=i" => \$parallel, |
"parallel=i" => \$parallel, |
135 |
"only-links!" => \$only_links, |
"only-links!" => \$only_links, |
136 |
"merge" => \$merge, |
"merge" => \$merge, |
137 |
|
"help" => \$help, |
138 |
); |
); |
139 |
|
|
140 |
|
pod2usage(-verbose => 2) if ($help); |
141 |
|
|
142 |
my $config = new WebPAC::Config( path => $config_path ); |
my $config = new WebPAC::Config( path => $config_path ); |
143 |
|
|
144 |
#print "config = ",dump($config) if ($debug); |
#print "config = ",dump($config) if ($debug); |
147 |
|
|
148 |
$log->info( "-" x 79 ); |
$log->info( "-" x 79 ); |
149 |
|
|
150 |
|
my $log_file = 'log'; |
151 |
|
|
152 |
|
if (-e $log_file ) { # && -s $log_file > 5 * 1024 * 1024) { |
153 |
|
$log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'"); |
154 |
|
rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!"); |
155 |
|
} |
156 |
|
|
157 |
my $estcmd_fh; |
my $estcmd_fh; |
158 |
my $estcmd_path = './estcmd-merge.sh'; |
my $estcmd_path = './estcmd-merge.sh'; |
167 |
my $validate; |
my $validate; |
168 |
$validate = new WebPAC::Validate( |
$validate = new WebPAC::Validate( |
169 |
path => $validate_path, |
path => $validate_path, |
170 |
|
delimiters => $config->webpac('delimiters'), |
171 |
) if ($validate_path); |
) if ($validate_path); |
172 |
|
|
173 |
|
|
174 |
my $use_indexer = $config->use_indexer; |
my $use_indexer = $config->use_indexer; |
175 |
|
$stats ||= $validate; |
176 |
if ($stats) { |
if ($stats) { |
177 |
$log->debug("option --stats disables update of indexing engine..."); |
$log->debug("disabled indexing for stats collection"); |
178 |
$use_indexer = undef; |
$use_indexer = undef; |
179 |
} else { |
} else { |
180 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
370 |
my $rec = shift || die "need rec!"; |
my $rec = shift || die "need rec!"; |
371 |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
372 |
|
|
|
$store->save_row( |
|
|
database => $database, |
|
|
input => $input_name, |
|
|
id => $mfn, |
|
|
row => $rec, |
|
|
); |
|
|
|
|
373 |
WebPAC::Normalize::data_structure( |
WebPAC::Normalize::data_structure( |
374 |
row => $rec, |
row => $rec, |
375 |
rules => $rules, |
rules => $rules, |
393 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
394 |
lookup => $lookup_jar, |
lookup => $lookup_jar, |
395 |
%{ $input }, |
%{ $input }, |
396 |
|
load_row => sub { |
397 |
|
my $a = shift; |
398 |
|
return $store->load_row( |
399 |
|
database => $database, |
400 |
|
input => $input_name, |
401 |
|
id => $a->{id}, |
402 |
|
); |
403 |
|
}, |
404 |
|
save_row => sub { |
405 |
|
my $a = shift; |
406 |
|
return $store->save_row( |
407 |
|
database => $database, |
408 |
|
input => $input_name, |
409 |
|
id => $a->{id}, |
410 |
|
row => $a->{row}, |
411 |
|
); |
412 |
|
}, |
413 |
|
|
414 |
); |
); |
415 |
|
|
416 |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
417 |
|
|
418 |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
419 |
$log->debug("created following lookups: ", dump( $lookup_data )); |
$log->debug("created following lookups: ", sub { dump( $lookup_data ) } ); |
420 |
|
|
421 |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
422 |
$store->save_lookup( |
$store->save_lookup( |
439 |
} |
} |
440 |
|
|
441 |
my $marc; |
my $marc; |
442 |
if ($parser->have_rules( 'marc', $database, $input_name )) { |
if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) { |
443 |
$marc = new WebPAC::Output::MARC( |
$marc = new WebPAC::Output::MARC( |
444 |
path => "out/marc/${database}-${input_name}.marc", |
path => "out/marc/${database}-${input_name}.marc", |
445 |
lint => $marc_lint, |
lint => $marc_lint, |
467 |
foreach my $db (keys %$depends) { |
foreach my $db (keys %$depends) { |
468 |
foreach my $i (keys %{$depends->{$db}}) { |
foreach my $i (keys %{$depends->{$db}}) { |
469 |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
470 |
|
my $t = time(); |
471 |
$log->debug("loading lookup $db/$i"); |
$log->debug("loading lookup $db/$i"); |
472 |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
473 |
database => $db, |
database => $db, |
474 |
input => $i, |
input => $i, |
475 |
key => $k, |
key => $k, |
476 |
); |
); |
477 |
|
$log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t)); |
478 |
} |
} |
479 |
} |
} |
480 |
} |
} |
481 |
|
|
482 |
$log->debug("lookup_hash = ", dump( $lookup_hash )); |
$log->debug("lookup_hash = ", sub { dump( $lookup_hash ) }); |
483 |
} |
} |
484 |
|
|
485 |
|
|
487 |
|
|
488 |
my $row = $input_db->fetch || next; |
my $row = $input_db->fetch || next; |
489 |
|
|
490 |
|
$total_rows++; |
491 |
|
|
492 |
my $mfn = $row->{'000'}->[0]; |
my $mfn = $row->{'000'}->[0]; |
493 |
|
|
494 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
if (! $mfn || $mfn !~ m#^\d+$#) { |
499 |
|
|
500 |
|
|
501 |
if ($validate) { |
if ($validate) { |
502 |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { |
503 |
$log->error( "MFN $mfn validation error:\n", |
$log->error( "MFN $mfn validation error:\n", |
504 |
$validate->report_error( $errors ) |
$validate->report_error( $errors ) |
505 |
); |
); |
506 |
} |
} |
507 |
|
next; # validation doesn't create any output |
508 |
} |
} |
509 |
|
|
510 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
523 |
}, |
}, |
524 |
); |
); |
525 |
|
|
526 |
$log->debug("ds = ",dump($ds)) if ($ds); |
$log->debug("ds = ", sub { dump($ds) }) if ($ds); |
527 |
|
|
528 |
$store->save_ds( |
$store->save_ds( |
529 |
database => $database, |
database => $database, |
545 |
$marc->add( |
$marc->add( |
546 |
id => $mfn . ( $i ? "/$i" : '' ), |
id => $mfn . ( $i ? "/$i" : '' ), |
547 |
fields => $fields, |
fields => $fields, |
548 |
leader => WebPAC::Normalize::marc_leader(), |
leader => WebPAC::Normalize::_get_marc_leader(), |
549 |
row => $row, |
row => $row, |
550 |
); |
); |
551 |
$i++; |
$i++; |
553 |
|
|
554 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
555 |
} |
} |
|
|
|
|
$total_rows++; |
|
556 |
} |
} |
557 |
|
|
558 |
if ($validate) { |
if ($validate) { |
561 |
$log->info("validation errors:\n$errors\n" ); |
$log->info("validation errors:\n$errors\n" ); |
562 |
print $report_fh "$errors\n" if ($report_fh); |
print $report_fh "$errors\n" if ($report_fh); |
563 |
} |
} |
564 |
|
|
565 |
|
print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1 ), "\n\n"; |
566 |
} |
} |
567 |
|
|
568 |
if ($stats) { |
if ($stats) { |