2 |
|
|
3 |
use strict; |
use strict; |
4 |
use CGI qw/:standard -no_xhtml/; |
use CGI qw/:standard -no_xhtml/; |
|
use Text::Query; |
|
5 |
use CGI::Carp qw(fatalsToBrowser); |
use CGI::Carp qw(fatalsToBrowser); |
6 |
use SWISH; |
use SWISH; |
7 |
|
use Unicode::String qw(utf8 utf16); |
8 |
|
require Unicode::Map8; |
9 |
|
use GDBM_File; |
10 |
|
|
11 |
my $dir='/home/dpavlin/nn/swish'; |
my $dir='/home/dpavlin/nn/swish'; |
12 |
|
my $prog='/usr/bin/swish-e'; |
13 |
|
|
14 |
my $hits=0; |
my $hits=0; |
15 |
|
my $max_hits=100; |
16 |
|
|
17 |
|
my %labels = (100=>' 100', 200=>' 200', 500=>' 500', 0=>'neograničeno'); |
18 |
|
|
19 |
|
my %brs_labels; |
20 |
|
|
21 |
|
my %tie; |
22 |
|
tie %tie, 'GDBM_File', "$dir/brzakona.gdbm", &GDBM_READER, 0640 || die "tie: $!"; |
23 |
|
foreach (sort keys %tie) { |
24 |
|
$brs_labels{$_} = sprintf("%-8s (%d zakona)",$_,$tie{$_}); |
25 |
|
} |
26 |
|
untie %tie; |
27 |
|
|
28 |
print header(-charset=>'iso-8859-2'),start_html(-title=>'NN pretrazivanje',-lang=>'hr'),start_form; |
print header(-charset=>'iso-8859-2'),start_html(-title=>'NN pretrazivanje',-lang=>'hr'),start_form; |
29 |
print "Prikažim zakone sa riječima: ",textfield('search'); |
print "Potraži ",popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits)," zakona sa riječima: ",textfield('search'); |
30 |
print submit(-value=>'prikaži'); |
print submit(-value=>'prikaži'); |
31 |
|
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); |
32 |
print end_form,hr; |
print end_form,hr; |
33 |
|
|
34 |
if (param('search')) { |
if (param('search')) { |
35 |
|
|
36 |
my $s=param('search'); |
my $s; |
37 |
|
# re-write query from +/- to and/and not |
38 |
|
foreach (split(/\s+/,param('search'))) { |
39 |
|
if (m/^([+-])(\S+)/) { |
40 |
|
$s.= ($s) ? "and " : ""; |
41 |
|
$s.="not " if ($1 eq "-"); |
42 |
|
$s.="$2* "; |
43 |
|
} else { |
44 |
|
$s .= "$_* "; |
45 |
|
} |
46 |
|
} |
47 |
$s=~tr/đčćĐČĆ/šđžčćŠĐŽČĆ/; # 1250 -> iso8859-2 |
$s=~tr/đčćĐČĆ/šđžčćŠĐŽČĆ/; # 1250 -> iso8859-2 |
48 |
$s=~tr/šŠđĐčČćĆžŽ/sSdDcCcCzZ/; |
$s=~tr/šŠđĐčČćĆžŽ/sSdDcCcCzZ/; |
49 |
|
|
50 |
|
my $l2_map = Unicode::Map8->new("ISO-8859-2") || die; |
51 |
|
my $us = Unicode::String->new(); |
52 |
|
|
53 |
my $sh = SWISH->connect('Fork', |
my $sh = SWISH->connect('Fork', |
54 |
prog => "$dir/swish-e", |
prog => $prog, |
55 |
indexes => "$dir/nn.index", |
indexes => "$dir/nn.index", |
56 |
|
properties => [qw/god br nr/], |
57 |
results => sub { |
results => sub { |
58 |
my ($sh, $hit) = @_; |
my ($sh,$hit) = @_; |
59 |
|
|
60 |
print $hit->swishreccount, '<a href="',$hit->swishdocpath,"'>",$hit->swishtitle,"</a> [",$hit->swishrank,"]<br>\n"; |
$us->utf8($hit->swishtitle); |
61 |
|
|
62 |
# print $_[1]->as_string,"\n"; |
print "<a href=\"",$hit->swishdocpath,"\"><tt>NN",$hit->god,"/",$hit->br,"</a> ",$hit->nr," </tt>",$l2_map->to8($us->utf16),"</a> [",$hit->swishrank,"]<br>\n"; |
63 |
|
|
64 |
|
# print $_[1]->as_string,"<br>\n"; |
65 |
# my @fields = $hit->field_names; |
# my @fields = $hit->field_names; |
66 |
# print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
# print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
67 |
}, |
}, |
68 |
|
maxhits => param('max_hits') || $max_hits, |
69 |
); |
); |
70 |
|
|
71 |
die $SWISH::errstr unless $sh; |
die $SWISH::errstr unless $sh; |
72 |
|
|
73 |
$hits = $sh->query("naslov_czs=($s)"); |
my $sw_q = "naslov_czs=($s)"; |
74 |
|
|
75 |
|
if (param('br_limit')) { |
76 |
|
my ($god,$br) = split(/\//,param('br')); |
77 |
|
$sw_q .= " and god=".int($god) if ($god); |
78 |
|
$sw_q .= " and br=".int($br) if ($br); |
79 |
|
} |
80 |
|
|
81 |
|
print "<!-- swish query: $sw_q -->"; |
82 |
|
|
83 |
|
$hits = $sh->query($sw_q); |
84 |
|
|
85 |
if ($hits > 0) { |
if ($hits > 0) { |
86 |
print p,hr,"Nađeno je $hits zakona..."; |
print p,hr,"Prikazujem $hits zakona"; |
87 |
|
print " iz broja ",param('br') if (param('br_limit')); |
88 |
|
print " (maks. ",param('max_hits') || $max_hits,")... <small>($s)</small>"; |
89 |
} else { |
} else { |
90 |
print p,"Nije nađen niti jedan zakon... (",$sh->errstr,")"; |
print p,"Nije nađen niti jedan zakon... <small>($s, ",$sh->errstr,")</small>"; |
91 |
} |
} |
92 |
} else { |
} else { |
93 |
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('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>'); |
94 |
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("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."); |
95 |
} |
} |