--- trunk2/all2all.pl 2004/09/05 17:54:51 407 +++ trunk2/all2all.pl 2004/09/08 15:30:07 415 @@ -13,10 +13,11 @@ use lib './lib'; use WebPAC; use WebPAC::jsFind; +use WebPAC::Index; my $webpac = new WebPAC( code_page => 'ISO-8859-2', -# limit_mfn => 100, + limit_mfn => 500, # debug => 1, ) || die; @@ -24,9 +25,11 @@ my $index = new WebPAC::jsFind( index_path => './out/index', - keys => 80, + keys => 10, ) || die; +my $thes; + $|=1; my $maxmfn = $webpac->open_isis( @@ -49,7 +52,7 @@ my @ds = $webpac->data_structure($rec); - if ($log->is_debug) { + if (0 && $log->is_debug) { $log->debug("rec = ",Dumper($rec)); $log->debug("ds = ",Dumper(\@ds)); } @@ -59,13 +62,12 @@ my $filename = $webpac->{'current_filename'}; if ($filename) { - open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); - print OUT $webpac->output( + $webpac->output_file( + file => $filename, template => 'html.tt', data => \@ds, headline => $webpac->{'headline'}, ); - close(OUT); } else { print $webpac->output( template => 'text.tt', @@ -91,13 +93,49 @@ ); } + # save into sorted index (thesaurus) + foreach my $ds (@ds) { + next if (! $ds->{'index'}); + + $thes->{$ds->{'tag'}} ||= new WebPAC::Index; + + foreach my $h (@{$ds->{'index'}}) { + $thes->{$ds->{'tag'}}->insert( + path => $f, + headline => $h, + ); + } + } + # print Dumper(\@ds); } -if ($log->is_debug) { +foreach my $t (keys %{$thes}) { + + my @e = $thes->{$t}->elements; + if (! @e) { + $log->logwarn("no elements in sorted index $t?"); + next; + } + + my $file = "./out/bfilter/$t.txt"; + $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); + + $webpac->output_file( + file => $file, + template => 'index.tt', + data => \@e, + index_name => $t, + ); +} + +if (0 && $log->is_debug) { $log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); $log->debug("data hash: ",Dumper($webpac->{'data'})); + foreach my $t (keys %{$thes}) { + $log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); + } } $index->close;