--- search/nn-swish.cgi 2002/06/20 07:57:05 1.4 +++ search/nn-swish.cgi 2003/09/09 08:20:51 1.13 @@ -2,56 +2,94 @@ use strict; use CGI qw/:standard -no_xhtml/; -use Text::Query; use CGI::Carp qw(fatalsToBrowser); use SWISH; +use Unicode::String qw(utf8 utf16); require Unicode::Map8; +use GDBM_File; my $dir='/home/dpavlin/nn/swish'; +my $prog='/usr/bin/swish-e'; my $hits=0; +my $max_hits=100; + +my %labels = (100=>' 100', 200=>' 200', 500=>' 500', 0=>'neograničeno'); + +my %brs_labels; + +my %tie; +tie %tie, 'GDBM_File', "$dir/brzakona.gdbm", &GDBM_READER, 0640 || die "tie: $!"; +foreach (sort keys %tie) { + $brs_labels{$_} = sprintf("%-8s (%d zakona)",$_,$tie{$_}); +} +untie %tie; print header(-charset=>'iso-8859-2'),start_html(-title=>'NN pretrazivanje',-lang=>'hr'),start_form; -print "Potraži zakone sa riječima: ",textfield('search'); +print "Potraži ",popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits)," zakona sa riječima: ",textfield('search'); print submit(-value=>'prikaži'); +print br,checkbox(-name=>'br_limit', -checked=>0, -label=>"ograniči pretraživanje samo na broj "),popup_menu(-name=>'br',-values=>[sort keys %brs_labels],-labels=>\%brs_labels); print end_form,hr; if (param('search')) { - my $s=param('search'); + my $s; + # re-write query from +/- to and/and not + foreach (split(/\s+/,param('search'))) { + if (m/^([+-])(\S+)/) { + $s.= ($s) ? "and " : ""; + $s.="not " if ($1 eq "-"); + $s.="$2* "; + } else { + $s .= "$_* "; + } + } $s=~tr/šđžčćŠĐŽČĆ/šđžčćŠĐŽČĆ/; # 1250 -> iso8859-2 $s=~tr/šŠđĐčČćĆžŽ/sSdDcCcCzZ/; my $l2_map = Unicode::Map8->new("ISO-8859-2") || die; - #my $utf8_map = Unicode::Map8->new("utf8") || die; + my $us = Unicode::String->new(); my $sh = SWISH->connect('Fork', - prog => "$dir/swish-e", + prog => $prog, indexes => "$dir/nn.index", properties => [qw/god br nr/], results => sub { my ($sh,$hit) = @_; - my $us=$hit->swishtitle; + $us->utf8($hit->swishtitle); - print "swishdocpath,"\">NN",$hit->god,"/",$hit->br," ",$hit->nr," ",$hit->swishtitle," [",$hit->swishrank,"]
\n"; + print "swishdocpath,"\">NN",$hit->god,"/",$hit->br," ",$hit->nr," ",$l2_map->to8($us->utf16)," [",$hit->swishrank,"]
\n"; # print $_[1]->as_string,"
\n"; # my @fields = $hit->field_names; # print "Field '$_' = '", $hit->$_, "'
\n" for sort @fields; }, + maxhits => param('max_hits') || $max_hits, ); die $SWISH::errstr unless $sh; - $hits = $sh->query("naslov_czs=($s)"); + my $sw_q = "naslov_czs=($s)"; + + if (param('br_limit')) { + my ($god,$br) = split(/\//,param('br')); + $sw_q .= " and god=".int($god) if ($god); + $sw_q .= " and br=".int($br) if ($br); + } + + print ""; + + $hits = $sh->query($sw_q); if ($hits > 0) { - print p,hr,"Nađeno je $hits zakona..."; + print p,hr,"Prikazujem $hits zakona"; + print " iz broja ",param('br') if (param('br_limit')); + print " (maks. ",param('max_hits') || $max_hits,")... ($s)"; } else { - print p,"Nije nađen niti jedan zakon... (",$sh->errstr,")"; + print p,"Nije nađen niti jedan zakon... ($s, ",$sh->errstr,")"; } } else { print p('Kod pretraživanja pretraživač pronalazi sve zakone u kojima se pojavljuju sve upisanje riječi.',br,'Ako ispred riječi upišete minus (-) neće se prikazivati zakoni koji imaju takvu riječ. Npr. +kava +zakon -dopunama'); - print p("Možete pročitati i članak o tome kako je ovaj pretraživač napravljen i zašto."); + print p("Možete pročitati i članak o tome kako je ovaj pretraživač napravljen i zašto."),p("Ovo je nova i brža verzija pretraživača o kojoj više možete saznati iz članka."); }