--- trunk/run.pl 2007/04/01 21:47:48 814 +++ trunk/run.pl 2007/08/23 20:57:00 884 @@ -12,7 +12,7 @@ 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; @@ -68,6 +68,14 @@ turn on extra validation of imput records, see L +You can use special variables C<$database> and $C<$input> in this parametar +like C<--validate 'conf/validate/$database-$input'> to construct filename + +=item --validate-delimiters path/to/validate_delimiters_file + +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-generate Generate MARC file. This will automatically be on if file contains C directives. @@ -108,6 +116,7 @@ my $only_filter; my $stats = 0; my $validate_path; +my $validate_delimiters_path; my $marc_generate = 1; my $marc_lint = 1; my $marc_dump = 0; @@ -128,6 +137,7 @@ "debug+" => \$debug, "stats" => \$stats, "validate=s" => \$validate_path, + "validate-delimiters=s" => \$validate_delimiters_path, "marc-generate!" => \$marc_generate, "marc-lint!" => \$marc_lint, "marc-dump!" => \$marc_dump, @@ -137,6 +147,8 @@ "help" => \$help, ); +$marc_generate = 0 if ( $validate_delimiters_path ); + pod2usage(-verbose => 2) if ($help); my $config = new WebPAC::Config( path => $config_path ); @@ -163,19 +175,17 @@ $log->info("created merge batch file $estcmd_path"); } - my $validate; $validate = new WebPAC::Validate( - path => $validate_path, -) if ($validate_path); - + delimiters => $config->webpac('delimiters'), +) if ($validate_path || $validate_delimiters_path); my $use_indexer = $config->use_indexer; $stats ||= $validate; if ($stats) { $log->debug("disabled indexing for stats collection"); $use_indexer = undef; -} else { +} elsif ( $use_indexer ) { $log->info("using $use_indexer indexing engine..."); } @@ -202,7 +212,8 @@ return $c; } -while (my ($database, $db_config) = each %{ $config->databases }) { +foreach my $database ( sort keys %{ $config->databases } ) { + my $db_config = $config->databases->{$database}; my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); next if ($only_database && $database !~ m/$only_database/i); @@ -233,19 +244,19 @@ if ($use_indexer eq 'hyperestraier') { # open Hyper Estraier database - use WebPAC::Output::Estraier '0.10'; + require WebPAC::Output::Estraier; $indexer = new WebPAC::Output::Estraier( %{ $indexer_config } ); } elsif ($use_indexer eq 'hyperestraier-native') { # open Hyper Estraier database - use WebPAC::Output::EstraierNative; + require WebPAC::Output::EstraierNative; $indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } ); } elsif ($use_indexer eq 'kinosearch') { # open KinoSearch - use WebPAC::Output::KinoSearch; + require WebPAC::Output::KinoSearch; $indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path}); $indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } ); @@ -253,7 +264,7 @@ $log->logdie("unknown use_indexer: $use_indexer"); } - $log->logide("can't continue without valid indexer") unless ($indexer); + $log->logdie("can't continue without valid indexer") unless ($indexer); } @@ -288,7 +299,7 @@ # now WebPAC::Store # my $abs_path = abs_path($0); - $abs_path =~ s#/[^/]*$#/#; + $abs_path =~ s#/[^/]*$#/#; # my $db_path = $config->webpac('db_path'); @@ -304,6 +315,20 @@ debug => $debug, ); + # + # prepare output + # + my @outputs; + if (defined( $db_config->{output} )) { + my $module = $db_config->{output}->{module} || $log->logdie("need module in output section of $database"); + $module = 'WebPAC::Output::' . $module unless $module =~ m/::/; + $log->debug("loading output module $module"); + eval "require $module"; + my $out = new $module->new( $db_config->{output} ); + $out->init; + push @outputs, $out; + } + # # now, iterate through input formats @@ -351,6 +376,7 @@ stats => $stats, modify_records => $input->{modify_records}, modify_file => $input->{modify_file}, + input_config => $input, ); $log->logdie("can't create input using $input_module") unless ($input); @@ -435,6 +461,11 @@ print $report_fh "Report for database '$database' input '$input_name' records ", $offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; $log->info("Generating report file $path"); + + if ( $validate ) { + $validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ) if ( $validate_path ); + $validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ) if ( $validate_delimiters_path ); + } } my $marc; @@ -490,7 +521,7 @@ my $mfn = $row->{'000'}->[0]; - if (! $mfn || $mfn !~ m#^\d+$#) { + if (! $mfn || $mfn !~ m{^\d+$}) { $log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos"); $mfn = $pos; push @{ $row->{'000'} }, $pos; @@ -552,6 +583,11 @@ $log->info("Created $i instances of MFN $mfn\n") if ($i > 1); } + + foreach my $out ( @outputs ) { + $out->add( $mfn, $ds ) if $out->can('add'); + } + } if ($validate) { @@ -560,6 +596,11 @@ $log->info("validation errors:\n$errors\n" ); print $report_fh "$errors\n" if ($report_fh); } + + print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1, current_input => 1 ), "\n\n"; + + # must be last thing that touches $validate for this input + $validate->reset; } if ($stats) { @@ -572,12 +613,15 @@ $marc->finish if ($marc); # close report - close($report_fh) if ($report_fh) - + close($report_fh) if ($report_fh); } eval { $indexer->finish } if ($indexer && $indexer->can('finish')); + foreach my $out ( @outputs ) { + $out->finish if $out->can('finish'); + } + my $dt = time() - $start_t; $log->info("$total_rows records ", $indexer ? "indexed " : "", sprintf("in %.2f sec [%.2f rec/sec]", @@ -600,6 +644,9 @@ $log->info("all parallel processes finished"); } +# save new delimiters if needed +$validate->save_delimiters_templates if ( $validate_delimiters_path ); + # # handle links or merge after indexing #