--- trunk/run.pl 2006/11/03 19:41:28 768 +++ trunk/run.pl 2006/12/13 10:08:27 790 @@ -8,7 +8,7 @@ use WebPAC::Common 0.02; use WebPAC::Parser 0.08; -use WebPAC::Input 0.14; +use WebPAC::Input 0.16; use WebPAC::Store 0.14; use WebPAC::Normalize 0.22; use WebPAC::Output::TT; @@ -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 @@ -107,6 +108,7 @@ my $parallel = 0; my $only_links = 0; my $merge = 0; +my $help; my $log = _new WebPAC::Common()->_get_logger(); @@ -125,8 +127,11 @@ "parallel=i" => \$parallel, "only-links!" => \$only_links, "merge" => \$merge, + "help" => \$help, ); +pod2usage(-verbose => 2) if ($help); + my $config = new WebPAC::Config( path => $config_path ); #print "config = ",dump($config) if ($debug); @@ -135,6 +140,12 @@ $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'; @@ -153,8 +164,9 @@ 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..."); @@ -467,6 +479,8 @@ my $row = $input_db->fetch || next; + $total_rows++; + my $mfn = $row->{'000'}->[0]; if (! $mfn || $mfn !~ m#^\d+$#) { @@ -477,11 +491,12 @@ if ($validate) { - if ( my $errors = $validate->validate_rec( $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( @@ -530,8 +545,6 @@ $log->info("Created $i instances of MFN $mfn\n") if ($i > 1); } - - $total_rows++; } if ($validate) {