--- trunk2/all2all.pl 2004/06/17 17:25:12 370 +++ trunk2/all2all.pl 2004/09/08 15:30:07 415 @@ -12,13 +12,24 @@ use lib './lib'; use WebPAC; +use WebPAC::jsFind; +use WebPAC::Index; my $webpac = new WebPAC( code_page => 'ISO-8859-2', limit_mfn => 500, - debug => 1, +# debug => 1, ) || die; +my $log = $webpac->_get_logger(); + +my $index = new WebPAC::jsFind( + index_path => './out/index', + keys => 10, +) || die; + +my $thes; + $|=1; my $maxmfn = $webpac->open_isis( @@ -29,10 +40,11 @@ # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, + { 'key' => '900_mfn:v900', 'val' => 'v000' }, ], ); -print "rows: $maxmfn\n\n"; +$log->info("rows: $maxmfn"); $webpac->open_import_xml(type => 'isis_hidra_ths'); @@ -40,12 +52,91 @@ my @ds = $webpac->data_structure($rec); - print $webpac->output( - template => 'text.tt', - data => \@ds, - ) if (@ds); + if (0 && $log->is_debug) { + $log->debug("rec = ",Dumper($rec)); + $log->debug("ds = ",Dumper(\@ds)); + } + + next if (! @ds); + + my $filename = $webpac->{'current_filename'}; + + if ($filename) { + $webpac->output_file( + file => $filename, + template => 'html.tt', + data => \@ds, + headline => $webpac->{'headline'}, + ); + } else { + print $webpac->output( + template => 'text.tt', + data => \@ds, + headline => $webpac->{'headline'}, + ); + } + + my $headline = $webpac->{'headline'}; + + my $f = $filename; + $f =~ s!out/!!; + + # save into index + foreach my $ds (@ds) { + next if (! $ds->{'swish'}); + + $index->insert( + index_name => $ds->{'tag'}, + path => $f, + headline => $headline, + words => join(" ",@{$ds->{'swish'}}) + ); + } + + # 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 => $h, + ); + } + } + +# print Dumper(\@ds); } -print "## lookup ",Dumper($webpac->{'lookup'}); -print "## data ",Dumper($webpac->{'data'}); +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; +