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 XML::Simple; |
8 |
use Lingua::Spelling::Alternative; |
9 |
use Text::Iconv; |
10 |
|
11 |
# output charset |
12 |
my $CHARSET='ISO-8859-2'; |
13 |
|
14 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
15 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
16 |
|
17 |
my $config=XMLin(undef, |
18 |
# keyattr => { label => "value" }, |
19 |
forcecontent => 0, |
20 |
); |
21 |
|
22 |
use Data::Dumper; |
23 |
#print Dumper($config); |
24 |
|
25 |
my $spelling_alt; |
26 |
# FIX: doesn't work very well |
27 |
if ($config->{findaffix}) { |
28 |
$spelling_alt = new Lingua::Spelling::Alternative; |
29 |
$spelling_alt->load_findaffix($config->{affix}); |
30 |
} |
31 |
if ($config->{affix}) { |
32 |
$spelling_alt = new Lingua::Spelling::Alternative; |
33 |
$spelling_alt->load_affix($config->{affix}); |
34 |
} |
35 |
|
36 |
my $hits=0; |
37 |
my $max_hits=$config->{max_hits}; |
38 |
|
39 |
my %labels; |
40 |
foreach (@{$config->{labels}->{label}}) { |
41 |
$labels{$_->{value}} = $from_utf8->convert($_->{content}); |
42 |
} |
43 |
|
44 |
if ($config->{charset}) { |
45 |
print header(-charset=>$config->{charset}); |
46 |
} else { |
47 |
print header; |
48 |
} |
49 |
print start_html(-title=>$config->{title}),start_form; |
50 |
print $config->{text}->{search}; |
51 |
print popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits); |
52 |
print $config->{text}->{documents}; |
53 |
print textfield('search'); |
54 |
print submit(-value=> $config->{text}->{submit}); |
55 |
print checkbox(-name=>'no_affix', -checked=>0, -label=>$config->{text}->{no_spell}) if ($spelling_alt); |
56 |
print end_form,hr; |
57 |
|
58 |
if (param('search')) { |
59 |
|
60 |
my $s; |
61 |
# re-write query from +/- to and/and not |
62 |
|
63 |
my $search = param('search'); |
64 |
my $s_phrase = ""; |
65 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
66 |
$s .= "$1 "; |
67 |
} |
68 |
$search =~ s/^\s+//; |
69 |
$search =~ s/\s+$//; |
70 |
|
71 |
foreach (split(/\s+/,$search)) { |
72 |
if (m/^([+-])(\S+)/) { |
73 |
$s.= ($s) ? "and " : ""; |
74 |
$s.="not " if ($1 eq "-"); |
75 |
if ($spelling_alt && !param('no_affix')) { |
76 |
my $w = $2; $w =~ s/[\*\s]+//g; |
77 |
$w =~ s/^(['"]*)([^'"]+)(['"]*)/$2/; |
78 |
$s.="$1(".join("* or ",$spelling_alt->alternatives($w))."*)$3 "; |
79 |
} else { |
80 |
$s.="$2* "; |
81 |
} |
82 |
} else { |
83 |
if ($spelling_alt && !param('no_affix')) { |
84 |
my $w = $_; $w =~ s/[\*\s]+//g; |
85 |
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
86 |
$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
87 |
} else { |
88 |
$s.="$_* "; |
89 |
} |
90 |
} |
91 |
} |
92 |
|
93 |
# fixup search string |
94 |
$s=~tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
95 |
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
96 |
$s=~s/\*\*+/*/g; |
97 |
|
98 |
my $sh = SWISH->connect('Fork', |
99 |
prog => $config->{prog}, |
100 |
indexes => $config->{index}, |
101 |
# properties => [qw/god br nr/], |
102 |
results => sub { |
103 |
my ($sh,$hit) = @_; |
104 |
|
105 |
printf ("<a href=\"%s\">%s</a> [%s]<br>\n","http://".virtual_host().$config->{url}.$hit->swishdocpath,$hit->swishtitle || 'untitled',$hit->swishrank); |
106 |
|
107 |
# print $_[1]->as_string,"<br>\n"; |
108 |
# my @fields = $hit->field_names; |
109 |
# print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
110 |
}, |
111 |
maxhits => param('max_hits') || $max_hits, |
112 |
); |
113 |
|
114 |
die $SWISH::errstr unless $sh; |
115 |
|
116 |
|
117 |
$hits = $sh->query($s); |
118 |
|
119 |
if ($hits > 0) { |
120 |
print p,hr; |
121 |
printf ($config->{text}->{hits},$hits,param('max_hits') || $max_hits,$s); |
122 |
} else { |
123 |
print p; |
124 |
printf ($config->{text}->{no_hits},$s,$sh->errstr); |
125 |
} |
126 |
} else { |
127 |
print p($config->{text}->{footer}); |
128 |
} |