--- trunk/run.pl 2007/05/27 11:14:40 851 +++ trunk/run.pl 2007/10/30 20:32:16 916 @@ -9,7 +9,7 @@ use WebPAC::Common 0.02; use WebPAC::Parser 0.08; use WebPAC::Input 0.16; -use WebPAC::Store 0.14; +use WebPAC::Store 0.15; use WebPAC::Normalize 0.22; use WebPAC::Output::TT; use WebPAC::Validate 0.11; @@ -68,6 +68,9 @@ 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 @@ -174,9 +177,7 @@ my $validate; $validate = new WebPAC::Validate( - path => $validate_path, delimiters => $config->webpac('delimiters'), - delimiters_path => $validate_delimiters_path, ) if ($validate_path || $validate_delimiters_path); my $use_indexer = $config->use_indexer; @@ -184,7 +185,7 @@ if ($stats) { $log->debug("disabled indexing for stats collection"); $use_indexer = undef; -} else { +} elsif ( $use_indexer ) { $log->info("using $use_indexer indexing engine..."); } @@ -211,7 +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); @@ -242,27 +244,24 @@ 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; - $indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path}); - $indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } ); + die "no longer supported"; } else { $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); } @@ -296,23 +295,26 @@ # # now WebPAC::Store # - my $abs_path = abs_path($0); - $abs_path =~ s#/[^/]*$#/#; - - my $db_path = $config->webpac('db_path'); + my $store = new WebPAC::Store({ + debug => $debug, + }); - if ($clean) { - $log->info("creating new database '$database' in $db_path"); - rmtree( $db_path ) || $log->warn("can't remove $db_path: $!"); - } else { - $log->info("working on database '$database' in $db_path"); + # + # 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"; + $db_config->{output}->{database} = $database; + $log->debug("calling $module->new(",dump( $db_config->{output} ),")"); + my $out = new $module->new( $db_config->{output} ); + $out->init; + push @outputs, $out; } - my $store = new WebPAC::Store( - path => $db_path, - debug => $debug, - ); - # # now, iterate through input formats @@ -360,6 +362,7 @@ 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); @@ -444,6 +447,11 @@ 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; @@ -499,7 +507,7 @@ 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; @@ -561,6 +569,11 @@ $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) { @@ -591,6 +604,10 @@ 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 ", $indexer ? "indexed " : "", sprintf("in %.2f sec [%.2f rec/sec]", @@ -614,7 +631,7 @@ } # save new delimiters if needed -$validate->save_delimiters_templates if ( $validate ); +$validate->save_delimiters_templates if ( $validate_delimiters_path ); # # handle links or merge after indexing