11 |
use WebPAC::Input 0.16; |
use WebPAC::Input 0.16; |
12 |
use WebPAC::Store 0.15; |
use WebPAC::Store 0.15; |
13 |
use WebPAC::Normalize 0.22; |
use WebPAC::Normalize 0.22; |
14 |
use WebPAC::Output::TT; |
#use WebPAC::Output::TT; |
15 |
use WebPAC::Validate 0.11; |
use WebPAC::Validate 0.11; |
16 |
use WebPAC::Output::MARC; |
use WebPAC::Output::MARC; |
17 |
use WebPAC::Config; |
use WebPAC::Config; |
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/; |
use Pod::Usage qw/pod2usage/; |
25 |
|
use LWP::Simple qw//; |
26 |
|
|
|
use Proc::Queue size => 1; |
|
27 |
use POSIX ":sys_wait_h"; # imports WNOHANG |
use POSIX ":sys_wait_h"; # imports WNOHANG |
28 |
|
|
29 |
=head1 NAME |
=head1 NAME |
103 |
|
|
104 |
Create merged index of databases which have links |
Create merged index of databases which have links |
105 |
|
|
106 |
|
=item --mirror http://www.example.com |
107 |
|
|
108 |
|
Tries to download input path files from mirror URI |
109 |
|
|
110 |
=back |
=back |
111 |
|
|
112 |
=cut |
=cut |
127 |
my $parallel = 0; |
my $parallel = 0; |
128 |
my $only_links = 0; |
my $only_links = 0; |
129 |
my $merge = 0; |
my $merge = 0; |
130 |
|
my $mirror; |
131 |
my $help; |
my $help; |
132 |
|
|
133 |
my $log = _new WebPAC::Common()->_get_logger(); |
my $log = _new WebPAC::Common()->_get_logger(); |
149 |
"parallel=i" => \$parallel, |
"parallel=i" => \$parallel, |
150 |
"only-links!" => \$only_links, |
"only-links!" => \$only_links, |
151 |
"merge" => \$merge, |
"merge" => \$merge, |
152 |
|
"mirror=s" => \$mirror, |
153 |
"help" => \$help, |
"help" => \$help, |
154 |
); |
); |
155 |
|
|
213 |
my @links; |
my @links; |
214 |
|
|
215 |
if ($parallel) { |
if ($parallel) { |
216 |
|
eval 'use Proc::Queue size => 1;'; |
217 |
|
die $@ if $@; |
218 |
$log->info("Using $parallel processes for speedup"); |
$log->info("Using $parallel processes for speedup"); |
219 |
Proc::Queue::size($parallel); |
Proc::Queue::size($parallel); |
220 |
} |
} |
357 |
$log->info("database $database doesn't have inputs defined"); |
$log->info("database $database doesn't have inputs defined"); |
358 |
} ); |
} ); |
359 |
|
|
360 |
if ( -e 'out/debug' ) { # fixme flag? |
if ( -e 'out/debug' ) { # FIXME flag? |
361 |
my $out; |
my $out; |
362 |
foreach my $i ( @inputs ) { |
foreach my $i ( @inputs ) { |
363 |
warn dump( $i ); |
warn dump( $i ); |
378 |
|
|
379 |
my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input)); |
my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input)); |
380 |
|
|
381 |
|
if ( $input->{skip} ) { |
382 |
|
$log->info("skip $input_name"); |
383 |
|
next; |
384 |
|
} |
385 |
|
|
386 |
next if ($only_input && ($input_name !~ m#$only_input#i && $input->{type} !~ m#$only_input#i)); |
next if ($only_input && ($input_name !~ m#$only_input#i && $input->{type} !~ m#$only_input#i)); |
387 |
|
|
388 |
my $type = lc($input->{type}); |
my $type = lc($input->{type}); |
410 |
delete($input->{modify_file}); |
delete($input->{modify_file}); |
411 |
} |
} |
412 |
|
|
413 |
|
if ( $mirror ) { |
414 |
|
my $path = $input->{path} || die "no input path in ",dump( $input ); |
415 |
|
|
416 |
|
my $base = $path; |
417 |
|
$base =~ s{/[^/]+$}{}; |
418 |
|
mkpath $base unless -e $base; |
419 |
|
|
420 |
|
my $rc = LWP::Simple::mirror( "$mirror/$path", $path ); |
421 |
|
if (LWP::Simple::is_error( $rc )) { |
422 |
|
die "can't mirror $mirror/$path -> $path [$rc]"; |
423 |
|
} else { |
424 |
|
$log->info( "mirror ", $path, " [$rc] ", -s $path, " bytes" ); |
425 |
|
} |
426 |
|
|
427 |
|
} |
428 |
|
|
429 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
430 |
module => $input_module, |
module => $input_module, |
|
encoding => $config->webpac('webpac_encoding'), |
|
431 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
432 |
offset => $offset, |
offset => $offset, |
433 |
recode => $input->{recode}, |
recode => $input->{recode}, |
472 |
|
|
473 |
my $maxmfn = $input_db->open( |
my $maxmfn = $input_db->open( |
474 |
path => $input->{path}, |
path => $input->{path}, |
475 |
code_page => $input->{encoding}, # database encoding |
input_encoding => $input->{encoding}, # database encoding |
476 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
477 |
lookup => $lookup_jar, |
lookup => $lookup_jar, |
478 |
%{ $input }, |
%{ $input }, |
513 |
|
|
514 |
my $report_fh; |
my $report_fh; |
515 |
if ($stats || $validate) { |
if ($stats || $validate) { |
516 |
my $path = "out/report/${database}-${input_name}.txt"; |
my $out_report = 'out/report'; # FIXME move to config |
517 |
|
mkpath $out_report unless -e $out_report; |
518 |
|
my $path = "$out_report/${database}-${input_name}.txt"; |
519 |
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
open($report_fh, '>', $path) || $log->logdie("can't open $path: $!"); |
520 |
|
|
521 |
print $report_fh "Report for database '$database' input '$input_name' records ", |
print $report_fh "Report for database '$database' input '$input_name' records ", |
530 |
|
|
531 |
my $marc; |
my $marc; |
532 |
if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) { |
if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) { |
533 |
|
|
534 |
|
my $out_marc = 'out/marc'; # FIXME move to config |
535 |
|
mkpath $out_marc unless -e $out_marc; |
536 |
|
|
537 |
$marc = new WebPAC::Output::MARC( |
$marc = new WebPAC::Output::MARC( |
538 |
path => "out/marc/${database}-${input_name}.marc", |
path => "$out_marc/${database}-${input_name}.marc", |
539 |
lint => $marc_lint, |
lint => $marc_lint, |
540 |
dump => $marc_dump, |
dump => $marc_dump, |
541 |
); |
); |
603 |
push @{ $row->{'000'} }, $pos; |
push @{ $row->{'000'} }, $pos; |
604 |
} |
} |
605 |
|
|
606 |
|
foreach my $out ( @output_modules ) { |
607 |
|
$out->add_row( $mfn, $row ) if $out->can('add_row'); |
608 |
|
} |
609 |
|
|
610 |
if ($validate) { |
if ($validate) { |
611 |
if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { |
if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { |
705 |
close($report_fh) if ($report_fh); |
close($report_fh) if ($report_fh); |
706 |
} |
} |
707 |
|
|
708 |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
$indexer->finish if $indexer && $indexer->can('finish'); |
709 |
|
|
710 |
foreach my $out ( @output_modules ) { |
foreach my $out ( @output_modules ) { |
711 |
$out->finish if $out->can('finish'); |
$out->finish if $out->can('finish'); |