--- trunk/run.pl 2006/09/29 20:18:39 731 +++ trunk/run.pl 2007/10/30 20:32:16 916 @@ -7,12 +7,12 @@ 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 0.15; +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; @@ -21,6 +21,7 @@ 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 @@ -31,7 +32,7 @@ B -Options: +=head1 OPTIONS =over 4 @@ -67,17 +68,22 @@ turn on extra validation of imput records, see L -=item --marc-normalize conf/normalize/mapping.pl +You can use special variables C<$database> and $C<$input> in this parametar +like C<--validate 'conf/validate/$database-$input'> to construct filename -This option specifies normalisation file for MARC creation +=item --validate-delimiters path/to/validate_delimiters_file -=item --marc-output out/marc/test.marc +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. -Optional path to output file +=item --marc-generate + +Generate MARC file. This will automatically be on if file contains C 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 directives. You can disable lint messages with C<--no-marc-lint>. =item --marc-dump @@ -110,12 +116,14 @@ 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 $log = _new WebPAC::Common()->_get_logger(); @@ -129,15 +137,20 @@ "debug+" => \$debug, "stats" => \$stats, "validate=s" => \$validate_path, - "marc-normalize=s" => \$marc_normalize, - "marc-output=s" => \$marc_output, + "validate-delimiters=s" => \$validate_delimiters_path, + "marc-generate!" => \$marc_generate, "marc-lint!" => \$marc_lint, "marc-dump!" => \$marc_dump, "parallel=i" => \$parallel, "only-links!" => \$only_links, "merge" => \$merge, + "help" => \$help, ); +$marc_generate = 0 if ( $validate_delimiters_path ); + +pod2usage(-verbose => 2) if ($help); + my $config = new WebPAC::Config( path => $config_path ); #print "config = ",dump($config) if ($debug); @@ -146,6 +159,12 @@ $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'; @@ -156,24 +175,20 @@ $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("option --stats disables update of indexing engine..."); + $log->debug("disabled indexing for stats collection"); $use_indexer = undef; -} else { +} elsif ( $use_indexer ) { $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 ); @@ -197,7 +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); @@ -212,7 +228,7 @@ } my $indexer; - if ($use_indexer) { + if ($use_indexer && $parser->have_rules( 'search', $database )) { my $cfg_name = $use_indexer; $cfg_name =~ s/\-.*$//; @@ -228,27 +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); } @@ -282,23 +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 @@ -346,6 +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); @@ -366,11 +383,11 @@ 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 ); @@ -387,12 +404,30 @@ 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( @@ -412,12 +447,17 @@ 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 $marc; - if ($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, ); @@ -443,17 +483,19 @@ 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 ) }); } @@ -461,9 +503,11 @@ my $row = $input_db->fetch || next; + $total_rows++; + 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'} }, $pos; @@ -471,11 +515,12 @@ if ($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( @@ -484,9 +529,9 @@ 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, @@ -494,6 +539,8 @@ }, ); + $log->debug("ds = ", sub { dump($ds) }) if ($ds); + $store->save_ds( database => $database, input => $input_name, @@ -514,7 +561,7 @@ $marc->add( id => $mfn . ( $i ? "/$i" : '' ), fields => $fields, - leader => WebPAC::Normalize::marc_leader(), + leader => WebPAC::Normalize::_get_marc_leader(), row => $row, ); $i++; @@ -523,7 +570,10 @@ $log->info("Created $i instances of MFN $mfn\n") if ($i > 1); } - $total_rows++; + foreach my $out ( @outputs ) { + $out->add( $mfn, $ds ) if $out->can('add'); + } + } if ($validate) { @@ -532,6 +582,11 @@ $log->info("validation errors:\n$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) { @@ -544,12 +599,15 @@ $marc->finish if ($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]", @@ -572,6 +630,9 @@ $log->info("all parallel processes finished"); } +# save new delimiters if needed +$validate->save_delimiters_templates if ( $validate_delimiters_path ); + # # handle links or merge after indexing #