1 |
#!/usr/bin/perl -w |
2 |
|
3 |
use strict; |
4 |
use CGI qw/:standard -no_xhtml/; |
5 |
use CGI::Carp qw(fatalsToBrowser); |
6 |
use SWISH; |
7 |
use Unicode::String qw(utf8 utf16); |
8 |
use Lingua::Spelling::Alternative; |
9 |
require Unicode::Map8; |
10 |
use GDBM_File; |
11 |
|
12 |
my $dir='/home/dpavlin/nn'; |
13 |
my $prog='/usr/bin/swish-e'; |
14 |
my $url='http://www.nn.hr/clanci/sluzbeno/'; |
15 |
|
16 |
my $hits=0; |
17 |
my $max_hits=100; |
18 |
|
19 |
my %labels = (100=>' 100', 200=>' 200', 500=>' 500', 0=>'neograničeno'); |
20 |
my %index = ('title'=>'naslovu', 'text'=>'tekstu'); |
21 |
|
22 |
my %god_lables; |
23 |
|
24 |
my %brzakona; |
25 |
tie %brzakona, 'GDBM_File', "$dir/swish/brzakona.gdbm", &GDBM_READER, 0640 || die "tie: $!"; |
26 |
foreach (sort keys %brzakona) { |
27 |
$god_lables{$_} = sprintf("%-8s (%d zakona)",$_,$brzakona{$_}); |
28 |
} |
29 |
untie %brzakona; |
30 |
|
31 |
my %file2title; |
32 |
tie %file2title, 'GDBM_File', "$dir/swish/file2title.gdbm", &GDBM_READER, 0640 || die "tie: $!"; |
33 |
|
34 |
my $hr = new Lingua::Spelling::Alternative( DEBUG => 0 ); |
35 |
$hr->load_findaffix("$dir/prvih_50.txt"); |
36 |
|
37 |
print header(-charset=>'iso-8859-2'),start_html(-title=>'Narodne Novine pretrazivanje',-lang=>'hr'),start_form(-name=>'search_form'); |
38 |
print "Potra¾i ",popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits)," zakona sa riječima: ",textfield('search'); |
39 |
print " u ",popup_menu(-name=>'index',-values=>[ sort keys %index ],-labels=>\%index,-default=>'title')," zakona "; |
40 |
print submit(-value=>'prika¾i'); |
41 |
print br,checkbox(-name=>'god_limit', -checked=>0, -label=>"ograniči pretra¾ivanje samo na godinu "),popup_menu(-name=>'god',-values=>[sort keys %god_lables],-labels=>\%god_lables,-onChange=>'this.form.god_limit.checked=true'); |
42 |
print end_form,hr; |
43 |
|
44 |
if (param('search')) { |
45 |
|
46 |
my $s; |
47 |
# re-write query from +/- to and/and not |
48 |
foreach (split(/\s+/,param('search'))) { |
49 |
if (m/^([+-])(\S+)/) { |
50 |
$s.= ($s) ? "and " : ""; |
51 |
$s.="not " if ($1 eq "-"); |
52 |
$s.="(".join("* or ",$hr->alternatives($2)).") "; |
53 |
} else { |
54 |
$s .= "(".join("* or ",$hr->alternatives($_)).") "; |
55 |
} |
56 |
} |
57 |
$s=~tr/ščęŠČĘ/¹š¾čę©Š®ČĘ/; # 1250 -> iso8859-2 |
58 |
$s=~tr/¹©šŠčČęĘ¾®/sSdDcCcCzZ/; |
59 |
|
60 |
my $l2_map = Unicode::Map8->new("ISO-8859-2") || die; |
61 |
my $us = Unicode::String->new(); |
62 |
|
63 |
my $sw_q; |
64 |
my $sh; |
65 |
|
66 |
if (param('index') eq 'title') { |
67 |
|
68 |
$sh = SWISH->connect('Fork', |
69 |
prog => $prog, |
70 |
indexes => "$dir/swish/nn.index", |
71 |
properties => [qw/god br nr/], |
72 |
results => sub { |
73 |
my ($sh,$hit) = @_; |
74 |
|
75 |
$us->utf8($hit->swishtitle); |
76 |
|
77 |
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"; |
78 |
|
79 |
# print $_[1]->as_string,"<br>\n"; |
80 |
# my @fields = $hit->field_names; |
81 |
# print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
82 |
}, |
83 |
maxhits => param('max_hits') || $max_hits, |
84 |
); |
85 |
|
86 |
die $SWISH::errstr unless $sh; |
87 |
|
88 |
$sw_q = "naslov_czs=($s)"; |
89 |
if (param('god_limit')) { |
90 |
$sw_q .= " and god=".int(param('god')) if (param('god')); |
91 |
} |
92 |
|
93 |
} else { |
94 |
# search in full text |
95 |
|
96 |
$sh = SWISH->connect('Fork', |
97 |
prog => $prog, |
98 |
indexes => "$dir/swish/sluzbeno.index", |
99 |
results => sub { |
100 |
my ($sh,$hit) = @_; |
101 |
|
102 |
my $path = $hit->swishdocpath; |
103 |
if ($file2title{$path}) { |
104 |
my ($god,$br,$nr,undef,$naslov) = split(/ /,$file2title{$path},5); |
105 |
print "<tt><a href=\"$url",$path,"\">NN$god/$br</a> $nr</tt> $naslov [",$hit->swishrank,"]<br>\n"; |
106 |
|
107 |
} else { |
108 |
print "<!-- error! can't find $path -->\n"; |
109 |
} |
110 |
|
111 |
}, |
112 |
maxhits => param('max_hits') || $max_hits, |
113 |
); |
114 |
|
115 |
die $SWISH::errstr unless $sh; |
116 |
|
117 |
$sw_q = $s; |
118 |
if (param('god_limit')) { |
119 |
$sw_q .= " and swishdocpath=".int(param('god')) if (param('god')); |
120 |
} |
121 |
|
122 |
} |
123 |
|
124 |
print "<!-- swish query: $sw_q -->"; |
125 |
|
126 |
$hits = $sh->query($sw_q); |
127 |
|
128 |
if ($hits > 0) { |
129 |
print p,hr,"Prikazujem $hits zakona"; |
130 |
print " iz godine ",param('god') if (param('god_limit')); |
131 |
print " (maks. ",param('max_hits') || $max_hits,")... <small>($s)</small>"; |
132 |
} else { |
133 |
print p,"Nije našen niti jedan zakon... <small>($s, ",$sh->errstr,")</small>"; |
134 |
} |
135 |
} else { |
136 |
my $dir=$0; |
137 |
$dir=~s,(^.*?)/[^/]+$,$1,g; |
138 |
open(HELP, "$dir/nn-help.html") || die "can't open '$dir/nn-help.html'"; |
139 |
while(<HELP>) { |
140 |
print; |
141 |
} |
142 |
close(HELP); |
143 |
} |
144 |
|
145 |
untie %file2title; |