--- trunk/search/nn-swish.cgi	2002/06/19 18:04:12	70
+++ trunk/search/nn-swish.cgi	2003/09/28 02:19:59	91
@@ -2,49 +2,139 @@
 
 use strict;
 use CGI qw/:standard -no_xhtml/;
-use Text::Query;
 use CGI::Carp qw(fatalsToBrowser);
 use SWISH;
-
-my $dir='/home/dpavlin/nn/swish';
+use Unicode::String qw(utf8 utf16);
+use Lingua::Spelling::Alternative;
+require Unicode::Map8;
+use GDBM_File;
+
+my $dir='/home/dpavlin/nn';
+my $prog='/usr/bin/swish-e';
+my $url='http://www.nn.hr/clanci/sluzbeno/';
 
 my $hits=0;
+my $max_hits=100;
+
+my %labels = (100=>' 100', 200=>' 200', 500=>' 500', 0=>'neograničeno');
+my %index = ('title'=>'naslovu', 'text'=>'tekstu');
+
+my %brs_labels;
+
+my %brzakona;
+tie %brzakona, 'GDBM_File', "$dir/swish/brzakona.gdbm", &GDBM_READER, 0640 || die "tie: $!";
+foreach (sort keys %brzakona) {
+	$brs_labels{$_} = sprintf("%-8s (%d zakona)",$_,$brzakona{$_});
+}
+untie %brzakona;
+
+my %file2title;
+tie %file2title, 'GDBM_File', "$dir/swish/file2title.gdbm", &GDBM_READER, 0640 || die "tie: $!";
+
+my $hr = new Lingua::Spelling::Alternative( DEBUG => 0 );
+$hr->load_findaffix("$dir/prvih_50.txt");
 
 print header(-charset=>'iso-8859-2'),start_html(-title=>'NN pretrazivanje',-lang=>'hr'),start_form;
-print "Prikažim 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 " u ",popup_menu(-name=>'index',-values=>[ sort keys %index ],-labels=>\%index,-default=>'title')," zakona ";
 print submit(-value=>'prikaži');
+print br,checkbox(-name=>'br_limit', -checked=>0, -label=>"ograniči pretraživanje samo na godinu "),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 "-");
+			$s.="(".join("* or ",$hr->alternatives($2)).") ";
+		} else {
+			$s .= "(".join("* or ",$hr->alternatives($_)).") ";
+		}
+	}
 	$s=~tr/šđžčćŠĐŽČĆ/šđžčćŠĐŽČĆ/;	# 1250 -> iso8859-2
 	$s=~tr/šŠđĐčČćĆžŽ/sSdDcCcCzZ/;
 
-	my $sh = SWISH->connect('Fork',
-		prog     => "$dir/swish-e",
-		indexes  => "$dir/nn.index",
-		results  => sub {
-			my ($sh, $hit) = @_;
-
-			print $hit->swishreccount, '<a href="',$hit->swishdocpath,"'>",$hit->swishtitle,"</a> [",$hit->swishrank,"]<br>\n";
-
-#			print $_[1]->as_string,"\n";
-#			my @fields = $hit->field_names;
-#			print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields;
-		},
-	);
+	my $l2_map = Unicode::Map8->new("ISO-8859-2") || die;
+	my $us = Unicode::String->new();
+
+	my $sw_q;
+	my $sh;
 
-	die $SWISH::errstr unless $sh;
+	if (param('index') eq 'title') {
 
-	$hits = $sh->query("naslov_czs=($s)");
+		$sh = SWISH->connect('Fork',
+			prog     => $prog,
+			indexes  => "$dir/swish/nn.index",
+			properties  => [qw/god br nr/],
+			results  => sub {
+				my ($sh,$hit) = @_;
+
+				$us->utf8($hit->swishtitle);
+
+				print "<tt><a href=\"$url",$hit->swishdocpath,"\">NN",$hit->god,"/",$hit->br,"</a> ",$hit->nr," </tt>",$l2_map->to8($us->utf16)," [",$hit->swishrank,"]<br>\n";
+
+	#			print $_[1]->as_string,"<br>\n";
+	#			my @fields = $hit->field_names;
+	#			print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields;
+			},
+			maxhits => param('max_hits') || $max_hits,
+		);
+
+		die $SWISH::errstr unless $sh;
+
+		$sw_q = "naslov_czs=($s)";
+		if (param('br_limit')) {
+			$sw_q .= " and god=".int(param('br')) if (param('br'));
+		}
+
+	} else {
+		# search in full text
+
+		$sh = SWISH->connect('Fork',
+			prog     => $prog,
+			indexes  => "$dir/swish/sluzbeno.index",
+			results  => sub {
+				my ($sh,$hit) = @_;
+
+				my $path = $hit->swishdocpath;
+				if ($file2title{$path}) {
+					my ($god,$br,$nr,undef,$naslov) = split(/ /,$file2title{$path},5);
+					print "<tt><a href=\"$url",$path,"\">NN$god/$br</a> $nr</tt> $naslov [",$hit->swishrank,"]<br>\n";
+
+				} else {
+					print "<!-- error! can't find $path -->\n";
+				}
+
+			},
+			maxhits => param('max_hits') || $max_hits,
+		);
+
+		die $SWISH::errstr unless $sh;
+
+		$sw_q = $s;
+		if (param('br_limit')) {
+			$sw_q .= " and swishdocpath=".int(param('br')) if (param('br'));
+		}
+
+	}
+
+	print "<!-- swish query: $sw_q -->";
+
+	$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') || $max_hits,")... <small>($s)</small>";
 	} else {
-		print p,"Nije nađen niti jedan zakon... (",$sh->errstr,")";
+		print p,"Nije nađen niti jedan zakon... <small>($s, ",$sh->errstr,")</small>";
 	}
 } else {
-	print p('U jednostavnom pretraživanju pretraživač pronalazi sve zakone u kojima se pojavljuje <b>bilo koja</b> od tih riječi. Da bi našli sve zakone u kojima se pojavljuju <b>sve upisane riječi</b>, upišite ispred svake riječi znak plus (+). Npr: <tt>+zakon +kava</tt>',br,'Ako ispred riječi upišete minus (-) neće se prikazivati zakoni koji imaju takvu riječ. Npr. <tt>+kava +zakon -dopunama</tt>');
-	print p("Možete pročitati i <a href=\"http://www.rot13.org/~dpavlin/nn.html\">članak</a> o tome kako je ovaj pretraživač napravljen i zašto.");
+	print p('Kod pretraživanja pretraživač pronalazi sve zakone u kojima se pojavljuju <b>sve upisanje riječi</b>.',br,'Ako ispred riječi upišete minus (-) neće se prikazivati zakoni koji imaju takvu riječ. Npr. <tt>+kava +zakon -dopunama</tt>');
+	print p("Možete pročitati i <a href=\"http://www.rot13.org/~dpavlin/nn.html\">članak</a> o tome kako je ovaj pretraživač napravljen i zašto."),p("Ovo je <i>nova i brža verzija pretraživača</i> o kojoj <a href=\"http://www.rot13.org/~dpavlin/nn.html#buducnost\">više možete saznati</a> iz članka."),p("Novosti od rujna 2003.: pretraživanje po godinama i po punom tekstu zakona!");
 }
+
+untie %file2title;