use lib './lib';
use WebPAC::Common 0.02;
-use WebPAC::Parser 0.04;
-use WebPAC::Input 0.13;
-use WebPAC::Store 0.11;
-use WebPAC::Normalize 0.21;
+use WebPAC::Parser 0.08;
+use WebPAC::Input 0.16;
+use WebPAC::Store 0.14;
+use WebPAC::Normalize 0.22;
use WebPAC::Output::TT;
-use WebPAC::Validate 0.06;
+use WebPAC::Validate 0.11;
use WebPAC::Output::MARC;
use WebPAC::Config;
use Getopt::Long;
use File::Slurp;
use Data::Dump qw/dump/;
use Storable qw/dclone/;
+use Pod::Usage qw/pod2usage/;
use Proc::Queue size => 1;
use POSIX ":sys_wait_h"; # imports WNOHANG
B<this command will probably go away. Don't get used to it!>
-Options:
+=head1 OPTIONS
=over 4
turn on extra validation of imput records, see L<WebPAC::Validation>
-=item --marc-normalize conf/normalize/mapping.pl
+=item --validate-delimiters path/to/validate_delimiters_file
-This option specifies normalisation file for MARC creation
+this option is used with C<--validate> to turn on extra validation of
+delimiters. If file is non existant, it will be created on first run.
-=item --marc-output out/marc/test.marc
+=item --marc-generate
-Optional path to output file
+Generate MARC file. This will automatically be on if file contains C<marc*> directives.
+You can use this option as C<--no-marc-generate> to disable MARC generation.
=item --marc-lint
-By default turned on if C<--marc-normalize> is used. You can disable lint
+By default turned on if normalisation file has C<marc*> directives. You can disable lint
messages with C<--no-marc-lint>.
=item --marc-dump
my $only_filter;
my $stats = 0;
my $validate_path;
-my ($marc_normalize, $marc_output);
+my $validate_delimiters_path;
+my $marc_generate = 1;
my $marc_lint = 1;
my $marc_dump = 0;
my $parallel = 0;
my $only_links = 0;
my $merge = 0;
+my $help;
my $log = _new WebPAC::Common()->_get_logger();
"debug+" => \$debug,
"stats" => \$stats,
"validate=s" => \$validate_path,
- "marc-normalize=s" => \$marc_normalize,
- "marc-output=s" => \$marc_output,
+ "validate-delimiters=s" => \$validate_delimiters_path,
+ "marc-generate!" => \$marc_generate,
"marc-lint!" => \$marc_lint,
"marc-dump!" => \$marc_dump,
"parallel=i" => \$parallel,
"only-links!" => \$only_links,
"merge" => \$merge,
+ "help" => \$help,
);
+pod2usage(-verbose => 2) if ($help);
+
my $config = new WebPAC::Config( path => $config_path );
#print "config = ",dump($config) if ($debug);
$log->info( "-" x 79 );
+my $log_file = 'log';
+
+if (-e $log_file ) { # && -s $log_file > 5 * 1024 * 1024) {
+ $log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'");
+ rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!");
+}
my $estcmd_fh;
my $estcmd_path = './estcmd-merge.sh';
$log->info("created merge batch file $estcmd_path");
}
-
my $validate;
$validate = new WebPAC::Validate(
path => $validate_path,
-) if ($validate_path);
-
+ delimiters => $config->webpac('delimiters'),
+ delimiters_path => $validate_delimiters_path,
+) if ($validate_path || $validate_delimiters_path);
my $use_indexer = $config->use_indexer;
+$stats ||= $validate;
if ($stats) {
- $log->debug("option --stats disables update of indexing engine...");
+ $log->debug("disabled indexing for stats collection");
$use_indexer = undef;
} else {
$log->info("using $use_indexer indexing engine...");
}
-# disable indexing when creating marc
-$use_indexer = undef if ($marc_normalize);
-
# parse normalize files and create source files for lookup and normalization
my $parser = new WebPAC::Parser( config => $config );
}
my $indexer;
- if ($use_indexer) {
+ if ($use_indexer && $parser->have_rules( 'search', $database )) {
my $cfg_name = $use_indexer;
$cfg_name =~ s/\-.*$//;
WebPAC::Normalize::data_structure(
row => $rec,
- rules => "_debug(2); $rules _debug(0)",
+ rules => $rules,
config => create_ds_config( $db_config, $database, $input, $mfn ),
);
- warn "current lookup: ", dump(WebPAC::Normalize::_get_lookup());
+ #warn "current lookup: ", dump(WebPAC::Normalize::_get_lookup());
};
WebPAC::Normalize::_set_lookup( undef );
lookup_coderef => $lookup_coderef,
lookup => $lookup_jar,
%{ $input },
+ load_row => sub {
+ my $a = shift;
+ return $store->load_row(
+ database => $database,
+ input => $input_name,
+ id => $a->{id},
+ );
+ },
+ save_row => sub {
+ my $a = shift;
+ return $store->save_row(
+ database => $database,
+ input => $input_name,
+ id => $a->{id},
+ row => $a->{row},
+ );
+ },
+
);
my $lookup_data = WebPAC::Normalize::_get_lookup();
if (defined( $lookup_data->{$database}->{$input_name} )) {
- $log->debug("created following lookups: ", dump( $lookup_data ));
+ $log->debug("created following lookups: ", sub { dump( $lookup_data ) } );
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) {
$store->save_lookup(
}
my $marc;
- if ($marc_normalize) {
+ if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) {
$marc = new WebPAC::Output::MARC(
- path => $marc_output || "out/marc/${database}-${input_name}.marc",
+ path => "out/marc/${database}-${input_name}.marc",
lint => $marc_lint,
dump => $marc_dump,
);
foreach my $db (keys %$depends) {
foreach my $i (keys %{$depends->{$db}}) {
foreach my $k (keys %{$depends->{$db}->{$i}}) {
+ my $t = time();
$log->debug("loading lookup $db/$i");
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup(
database => $db,
input => $i,
key => $k,
);
+ $log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t));
}
}
}
- $log->debug("lookup_hash = ", dump( $lookup_hash ));
+ $log->debug("lookup_hash = ", sub { dump( $lookup_hash ) });
}
my $row = $input_db->fetch || next;
+ $total_rows++;
+
my $mfn = $row->{'000'}->[0];
if (! $mfn || $mfn !~ m#^\d+$#) {
if ($validate) {
- if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) {
+ if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) {
$log->error( "MFN $mfn validation error:\n",
$validate->report_error( $errors )
);
}
+ next; # validation doesn't create any output
}
my $ds = WebPAC::Normalize::data_structure(
lookup => $lookup_hash,
config => create_ds_config( $db_config, $database, $input, $mfn ),
marc_encoding => 'utf-8',
- load_ds_coderef => sub {
+ load_row_coderef => sub {
my ($database,$input,$mfn) = @_;
- return $store->load_ds(
+ return $store->load_row(
database => $database,
input => $input,
id => $mfn,
},
);
+ $log->debug("ds = ", sub { dump($ds) }) if ($ds);
+
$store->save_ds(
database => $database,
input => $input_name,
$marc->add(
id => $mfn . ( $i ? "/$i" : '' ),
fields => $fields,
- leader => WebPAC::Normalize::marc_leader(),
+ leader => WebPAC::Normalize::_get_marc_leader(),
row => $row,
);
$i++;
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
}
-
- $total_rows++;
}
if ($validate) {
$log->info("validation errors:\n$errors\n" );
print $report_fh "$errors\n" if ($report_fh);
}
+
+ print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1 ), "\n\n";
}
if ($stats) {
$log->info("all parallel processes finished");
}
+# save new delimiters if needed
+$validate->save_delimiters_templates;
+
#
# handle links or merge after indexing
#