3 |
use strict; |
use strict; |
4 |
use CGI qw/:standard -no_xhtml/; |
use CGI qw/:standard -no_xhtml/; |
5 |
use CGI::Carp qw(fatalsToBrowser); |
use CGI::Carp qw(fatalsToBrowser); |
6 |
use SWISH; |
use SWISH::API; |
|
use Unicode::String qw(utf8 utf16); |
|
7 |
use Lingua::Spelling::Alternative; |
use Lingua::Spelling::Alternative; |
8 |
require Unicode::Map8; |
use Encode qw/from_to/; |
9 |
use GDBM_File; |
use GDBM_File; |
10 |
use lib '/data/swish/html'; |
use lib '/data/swish/html'; |
11 |
use FormatResult; |
use FormatResult; |
12 |
|
|
13 |
my $dir='/home/dpavlin/nn'; |
my $dir='/home/dpavlin/nn'; |
14 |
my $prog='/usr/bin/swish-e'; |
#my $prog='/usr/bin/swish-e'; |
15 |
my $url='http://www.nn.hr/clanci/sluzbeno/'; |
my $url='http://www.nn.hr/clanci/sluzbeno/'; |
16 |
|
|
|
my $hits=0; |
|
17 |
my $max_hits=100; |
my $max_hits=100; |
18 |
|
|
19 |
my %labels = (100=>' 100', 200=>' 200', 500=>' 500', 0=>'neogranièeno'); |
my %labels = (100=>' 100', 200=>' 200', 500=>' 500', 0=>'neogranièeno'); |
64 |
$s=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
$s=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
65 |
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
66 |
|
|
67 |
my $l2_map = Unicode::Map8->new("ISO-8859-2") || die; |
my $swish_query; |
|
my $us = Unicode::String->new(); |
|
|
|
|
|
my $sw_q; |
|
|
my $sh; |
|
68 |
|
|
69 |
if (defined(param('max_hits'))) { |
if (defined(param('max_hits'))) { |
70 |
$max_hits = param('max_hits'); |
$max_hits = param('max_hits'); |
71 |
} |
} |
72 |
|
|
73 |
if (param('index') eq 'title') { |
my ( $index_path, $result_coderef ); |
|
|
|
|
$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); |
|
74 |
|
|
75 |
my $naslov = $l2_map->to8($us->utf16); |
if (param('index') eq 'title') { |
|
$naslov = FormatResult::highlite_words(\$naslov, \@hl_words); |
|
|
print "<tt><a href=\"$url",$hit->swishdocpath,"\">NN",$hit->god,"/",$hit->br,"</a> ",$hit->nr," </tt> $naslov [",$hit->swishrank,"]<br>\n"; |
|
|
# print $_[1]->as_string,"<br>\n"; |
|
|
# my @fields = $hit->field_names; |
|
|
# print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
|
|
}, |
|
|
maxhits => $max_hits, |
|
|
); |
|
76 |
|
|
77 |
die $SWISH::errstr unless $sh; |
$result_coderef = sub { |
78 |
|
my $hit = shift || die; |
79 |
|
my $naslov = $hit->property('swishtitle'); |
80 |
|
from_to($naslov, 'utf-8', 'iso-8859-2'); |
81 |
|
$naslov = FormatResult::highlite_words(\$naslov, \@hl_words); |
82 |
|
print "<tt><a href=\"$url",$hit->property('swishdocpath'),"\">NN",$hit->property('god'),"/",$hit->property('br'),"</a> ",$hit->property('nr')," </tt> $naslov [",$hit->property('swishrank'),"]<br>\n"; |
83 |
|
}; |
84 |
|
|
85 |
$sw_q = "naslov_czs=($s)"; |
$index_path = "$dir/swish/nn.index"; |
86 |
|
$swish_query = "naslov_czs=($s)"; |
87 |
if (param('god_limit')) { |
if (param('god_limit')) { |
88 |
$sw_q .= " and god=".int(param('god')) if (param('god')); |
$swish_query .= " and god=".int(param('god')) if (param('god')); |
89 |
} |
} |
90 |
|
|
91 |
} else { |
} else { |
92 |
# search in full text |
# search in full text |
93 |
|
|
94 |
$sh = SWISH->connect('Fork', |
$result_coderef = sub { |
95 |
prog => $prog, |
my $hit = shift || die; |
|
indexes => "$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); |
|
|
$naslov = FormatResult::highlite_words(\$naslov, \@hl_words); |
|
|
print "<tt><a href=\"$url",$path,"\">NN$god/$br</a> $nr</tt> $naslov [",$hit->swishrank,"]<br>\n"; |
|
|
|
|
|
} else { |
|
|
print "<!-- error! can't find $path -->\n"; |
|
|
} |
|
|
|
|
|
}, |
|
|
maxhits => $max_hits, |
|
|
); |
|
96 |
|
|
97 |
die $SWISH::errstr unless $sh; |
my $path = $hit->property('swishdocpath'); |
98 |
|
if ($file2title{$path}) { |
99 |
|
my ($god,$br,$nr,undef,$naslov) = split(/ /,$file2title{$path},5); |
100 |
|
$naslov = FormatResult::highlite_words(\$naslov, \@hl_words); |
101 |
|
from_to($naslov, 'utf-8', 'iso-8859-2'); |
102 |
|
print "<tt><a href=\"$url",$path,"\">NN$god/$br</a> $nr</tt> $naslov [",$hit->property('swishrank'),"]<br>\n"; |
103 |
|
|
104 |
$sw_q = $s; |
} else { |
105 |
|
print "<!-- error! can't find $path -->\n"; |
106 |
|
} |
107 |
|
|
108 |
|
}; |
109 |
|
|
110 |
|
$index_path = "$dir/swish/sluzbeno.index"; |
111 |
|
$swish_query = $s; |
112 |
if (param('god_limit')) { |
if (param('god_limit')) { |
113 |
$sw_q .= " and swishdocpath=".int(param('god')) if (param('god')); |
$swish_query .= " and swishdocpath=".int(param('god')) if (param('god')); |
114 |
} |
} |
115 |
|
|
116 |
} |
} |
117 |
|
|
118 |
print "<!-- swish query: $sw_q -->"; |
print "<!-- swish query: $swish_query -->"; |
119 |
|
|
120 |
|
my $swish = SWISH::API->new( $index_path ); |
121 |
|
$swish->abort_last_error if $swish->Error; |
122 |
|
|
123 |
$hits = $sh->query($sw_q); |
my $results = $swish->query($swish_query); |
124 |
|
|
125 |
|
my $hits = $results->hits; |
126 |
|
|
127 |
if ($hits > 0) { |
if ($hits > 0) { |
128 |
print p,hr,"Prikazujem $hits zakona"; |
my $i = 1; |
129 |
|
while ( my $hit = $results->next_result ) { |
130 |
|
$result_coderef->( $hit ); |
131 |
|
last if $i++ == $max_hits; |
132 |
|
} |
133 |
|
|
134 |
|
print p,hr,"Pronaðeno $hits zakona"; |
135 |
print " iz godine ",param('god') if (param('god_limit')); |
print " iz godine ",param('god') if (param('god_limit')); |
136 |
print " (maks. $max_hits)... " if ($max_hits); |
print " (prikazano prvih $max_hits)... " if ($max_hits); |
137 |
print " <small>[$s]</small>"; |
print " <small>[$s]</small>"; |
138 |
} else { |
} else { |
139 |
print p,"Nije naðen niti jedan zakon... <small>[$s, ",$sh->errstr,"]</small>"; |
print p,"Nije naðen niti jedan zakon... <small>[$s, ",$swish->error_string,"]</small>"; |
140 |
} |
} |
141 |
} else { |
} else { |
142 |
my $dir=$0; |
my $dir=$0; |