--- trunk2/all2all.pl 2004/06/16 16:50:30 362 +++ trunk2/all2all.pl 2004/07/21 16:58:27 392 @@ -12,15 +12,22 @@ use lib './lib'; use WebPAC; +use WebPAC::jsFind; my $webpac = new WebPAC( code_page => 'ISO-8859-2', - limit_mfn => 500, + limit_mfn => 100, +# debug => 1, +) || die; + +my $log = $webpac->_get_logger(); + +my $index = new WebPAC::jsFind( + index_path => './out/index', ) || die; $|=1; -print "reading database\n"; my $maxmfn = $webpac->open_isis( filename => shift @ARGV || '/data/hidra/THS/THS', lookup => [ @@ -29,32 +36,72 @@ # { '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"); -while (my $rec = $webpac->fetch_rec) { +$webpac->open_import_xml(type => 'isis_hidra_ths'); - print "-- ",$webpac->parse($rec,'v250^a / [v562^4] v562^a'),"\n"; +while (my $rec = $webpac->fetch_rec) { - print " ", - $webpac->parse($rec,'eval{v901^a eq "Deskriptor"}v250^a / [v562^4] v562^a'), - $webpac->parse($rec,'eval{v901^a ne "Deskriptor"}250a [251]'), - "\n"; + my @ds = $webpac->data_structure($rec); - my @t = $webpac->fill_in($rec,'v553^1;;v553^a'); - print " Uži pojam: ",join("\t\n",@t),"\n" if (@t); + if ($log->is_debug) { + $log->debug("rec = ",Dumper($rec)); + $log->debug("ds = ",Dumper(\@ds)); + } + + next if (! @ds); + + my $filename = $webpac->{'current_filename'}; + + if ($filename) { + open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); + print OUT $webpac->output( + template => 'html.tt', + data => \@ds, + ); + close(OUT); + } else { + print $webpac->output( + template => 'text.tt', + data => \@ds, + ); + } + + my $headline; + foreach my $ds (@ds) { + if ($ds->{'tag'} eq 'headline') { + $headline = join(" ",@{$ds->{'display'}}); + last; + } + } + + 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'}}) + ); + } - @t = $webpac->fill_in($rec,'eval{v901^a eq "Područje"}[a:v251::];;[d:[a:v251::]]'); - print " Područje: ",join("\t\n",@t),"\n" if (@t); +# print Dumper(\@ds); - @t = $webpac->fill_in($rec,'eval{v901^a eq "Mikrotezaurus"}[a:v561^4:v251:];;[d:[a:v561^4:v251:]]'); - print " Mikrotezaurus: ",join("\t\n",@t),"\n" if (@t); +} - @t = $webpac->fill_in($rec,'eval{v901^a eq "Deskriptor"}[a:v561^4:v562^4:v900];;[d:[a:v561^4:v562^4:v900]]'); - print " Deskriptor: ",join("\t\n",@t),"\n" if (@t); +if ($log->is_debug) { + $log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); + $log->debug("data hash: ",Dumper($webpac->{'data'})); } -print "## lookup ",Dumper($webpac->{'lookup'}); -print "## data ",Dumper($webpac->{'data'}); +$index->close; +