--- trunk2/all2all.pl 2004/07/21 16:58:27 392 +++ trunk2/all2all.pl 2004/09/05 22:22:37 411 @@ -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; @@ -24,8 +25,11 @@ my $index = new WebPAC::jsFind( index_path => './out/index', + keys => 80, ) || die; +my $thes; + $|=1; my $maxmfn = $webpac->open_isis( @@ -48,7 +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)); } @@ -58,26 +62,21 @@ 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', data => \@ds, + headline => $webpac->{'headline'}, ); } - my $headline; - foreach my $ds (@ds) { - if ($ds->{'tag'} eq 'headline') { - $headline = join(" ",@{$ds->{'display'}}); - last; - } - } + my $headline = $webpac->{'headline'}; my $f = $filename; $f =~ s!out/!!; @@ -94,13 +93,46 @@ ); } + # save into sorted index (thesaurus) + foreach my $ds (@ds) { + next if (! $ds->{'index'}); + + $thes->{$ds->{'tag'}} ||= new WebPAC::Index; + + $thes->{$ds->{'tag'}}->insert( + path => $f, + headline => $headline, + ); + } + # print Dumper(\@ds); } -if ($log->is_debug) { +foreach my $t (keys %{$thes}) { + + my @e = $thes->{$t}->elements; + if (! @e) { + $log->logwarn("no elements in sorted index $t?"); + next; + } + + $log->debug("saving sorted index $t [".scalar(@e)." elements]"); + + $webpac->output_file( + file => "./out/thes_$t.html", + 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;