| 1 |
dpavlin |
8 |
#!/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 |
|
|
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
| 12 |
|
|
my $config=XMLin(undef, |
| 13 |
|
|
# keyattr => { label => "value" }, |
| 14 |
|
|
forcecontent => 0, |
| 15 |
|
|
); |
| 16 |
|
|
|
| 17 |
dpavlin |
18 |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
| 18 |
|
|
sub x { |
| 19 |
|
|
return $from_utf8->convert($_[0]); |
| 20 |
|
|
} |
| 21 |
|
|
|
| 22 |
dpavlin |
29 |
# Escape <, >, & and ", and to produce valid XML |
| 23 |
|
|
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"'); |
| 24 |
|
|
my $escape_re = join '|' => keys %escape; |
| 25 |
|
|
sub e { |
| 26 |
|
|
my $out; |
| 27 |
|
|
foreach my $v (@_) { |
| 28 |
|
|
$v =~ s/($escape_re)/$escape{$1}/g; |
| 29 |
|
|
$out .= $v; |
| 30 |
|
|
} |
| 31 |
|
|
return $out; |
| 32 |
|
|
} |
| 33 |
dpavlin |
8 |
|
| 34 |
dpavlin |
39 |
my @spellings; |
| 35 |
dpavlin |
8 |
# FIX: doesn't work very well |
| 36 |
dpavlin |
16 |
if ($config->{findaffix}) { |
| 37 |
dpavlin |
39 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
| 38 |
dpavlin |
18 |
$spelling_alt->load_findaffix(x($config->{findaffix})); |
| 39 |
dpavlin |
39 |
push @spellings,$spelling_alt; |
| 40 |
dpavlin |
8 |
} |
| 41 |
dpavlin |
16 |
if ($config->{affix}) { |
| 42 |
dpavlin |
39 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
| 43 |
dpavlin |
18 |
$spelling_alt->load_affix(x($config->{affix})); |
| 44 |
dpavlin |
39 |
push @spellings,$spelling_alt; |
| 45 |
dpavlin |
16 |
} |
| 46 |
dpavlin |
8 |
|
| 47 |
|
|
my $hits=0; |
| 48 |
dpavlin |
18 |
my $max_hits=x($config->{max_hits}); |
| 49 |
dpavlin |
8 |
|
| 50 |
|
|
my %labels; |
| 51 |
|
|
foreach (@{$config->{labels}->{label}}) { |
| 52 |
dpavlin |
18 |
$labels{$_->{value}} = x($_->{content}); |
| 53 |
dpavlin |
8 |
} |
| 54 |
|
|
|
| 55 |
dpavlin |
23 |
my $path = param('path'); # limit to this path |
| 56 |
|
|
my %path_label; |
| 57 |
|
|
my @path_name; |
| 58 |
|
|
foreach (@{$config->{paths}->{path}}) { |
| 59 |
dpavlin |
39 |
|
| 60 |
|
|
print STDERR "##: $_->{limit}",x($_->{content}),"\n"; |
| 61 |
dpavlin |
23 |
push @path_name,x($_->{limit}); |
| 62 |
|
|
$path_label{$_->{limit}} = x($_->{content}); |
| 63 |
|
|
} |
| 64 |
|
|
|
| 65 |
dpavlin |
32 |
my @properties = split(/\s+/,x($config->{properties})); |
| 66 |
|
|
|
| 67 |
dpavlin |
16 |
if ($config->{charset}) { |
| 68 |
dpavlin |
18 |
print header(-charset=>x($config->{charset})); |
| 69 |
dpavlin |
16 |
} else { |
| 70 |
|
|
print header; |
| 71 |
|
|
} |
| 72 |
dpavlin |
18 |
print start_html(-title=>x($config->{title})),start_form; |
| 73 |
|
|
print x($config->{text}->{search}); |
| 74 |
dpavlin |
16 |
print popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits); |
| 75 |
dpavlin |
18 |
print x($config->{text}->{documents}); |
| 76 |
dpavlin |
16 |
print textfield('search'); |
| 77 |
dpavlin |
18 |
print submit(-value=> x($config->{text}->{submit})); |
| 78 |
dpavlin |
39 |
print br,checkbox(-name=>'no_affix', -checked=>0, -label=>x($config->{text}->{no_spell})) if (@spellings); |
| 79 |
dpavlin |
32 |
print checkbox(-name=>'no_properties', -checked=>0, -label=>($config->{text}->{no_properties})) if (@properties); |
| 80 |
dpavlin |
23 |
if (@path_name) { |
| 81 |
|
|
print br,x($config->{text}->{limit}); |
| 82 |
|
|
print popup_menu(-name=>'path',-values=>\@path_name,-labels=>\%path_label,-default=>$path); |
| 83 |
|
|
} |
| 84 |
dpavlin |
8 |
print end_form,hr; |
| 85 |
|
|
|
| 86 |
|
|
if (param('search')) { |
| 87 |
|
|
|
| 88 |
|
|
my $s; |
| 89 |
|
|
# re-write query from +/- to and/and not |
| 90 |
|
|
|
| 91 |
|
|
my $search = param('search'); |
| 92 |
|
|
my $s_phrase = ""; |
| 93 |
|
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
| 94 |
|
|
$s .= "$1 "; |
| 95 |
|
|
} |
| 96 |
|
|
$search =~ s/^\s+//; |
| 97 |
|
|
$search =~ s/\s+$//; |
| 98 |
|
|
|
| 99 |
|
|
foreach (split(/\s+/,$search)) { |
| 100 |
|
|
if (m/^([+-])(\S+)/) { |
| 101 |
|
|
$s.= ($s) ? "and " : ""; |
| 102 |
|
|
$s.="not " if ($1 eq "-"); |
| 103 |
dpavlin |
39 |
if (@spellings && !param('no_affix')) { |
| 104 |
dpavlin |
8 |
my $w = $2; $w =~ s/[\*\s]+//g; |
| 105 |
|
|
$w =~ s/^(['"]*)([^'"]+)(['"]*)/$2/; |
| 106 |
dpavlin |
39 |
my $or=""; |
| 107 |
|
|
foreach my $spelling_alt (@spellings) { |
| 108 |
|
|
$s.="$or$1(".join("* or ",$spelling_alt->alternatives($w))."*)$3 "; |
| 109 |
|
|
$or = "or "; |
| 110 |
|
|
} |
| 111 |
dpavlin |
8 |
} else { |
| 112 |
|
|
$s.="$2* "; |
| 113 |
|
|
} |
| 114 |
|
|
} else { |
| 115 |
dpavlin |
39 |
if (@spellings && !param('no_affix')) { |
| 116 |
dpavlin |
8 |
my $w = $_; $w =~ s/[\*\s]+//g; |
| 117 |
|
|
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
| 118 |
dpavlin |
39 |
my $or=""; |
| 119 |
|
|
foreach my $spelling_alt (@spellings) { |
| 120 |
|
|
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
| 121 |
|
|
$or = "or "; |
| 122 |
|
|
} |
| 123 |
dpavlin |
8 |
} else { |
| 124 |
|
|
$s.="$_* "; |
| 125 |
|
|
} |
| 126 |
|
|
} |
| 127 |
|
|
} |
| 128 |
|
|
|
| 129 |
|
|
# fixup search string |
| 130 |
|
|
$s=~tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
| 131 |
|
|
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
| 132 |
|
|
$s=~s/\*\*+/*/g; |
| 133 |
|
|
|
| 134 |
dpavlin |
23 |
# limit to some path |
| 135 |
dpavlin |
39 |
$s = "swishdocpath=(\"*$path*\") and $s" if ($path); |
| 136 |
dpavlin |
23 |
|
| 137 |
dpavlin |
22 |
my %params; # optional parametars for swish |
| 138 |
|
|
|
| 139 |
dpavlin |
32 |
# default format for output |
| 140 |
|
|
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
| 141 |
dpavlin |
21 |
|
| 142 |
dpavlin |
32 |
if (@properties) { |
| 143 |
|
|
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
| 144 |
dpavlin |
39 |
$params{properties} = \@properties; |
| 145 |
dpavlin |
32 |
} else { |
| 146 |
|
|
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
| 147 |
|
|
} |
| 148 |
|
|
|
| 149 |
dpavlin |
8 |
my $sh = SWISH->connect('Fork', |
| 150 |
dpavlin |
18 |
prog => x($config->{prog}), |
| 151 |
|
|
indexes => x($config->{index}), |
| 152 |
dpavlin |
8 |
results => sub { |
| 153 |
|
|
my ($sh,$hit) = @_; |
| 154 |
|
|
|
| 155 |
dpavlin |
18 |
if ($config->{url}) { |
| 156 |
dpavlin |
32 |
printf ($hit_fmt ,"http://".virtual_host().x($config->{url}).$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties)); |
| 157 |
dpavlin |
18 |
} else { |
| 158 |
dpavlin |
29 |
printf ($hit_fmt ,$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties) ); |
| 159 |
dpavlin |
8 |
|
| 160 |
dpavlin |
18 |
} |
| 161 |
|
|
|
| 162 |
dpavlin |
8 |
# print $_[1]->as_string,"<br>\n"; |
| 163 |
|
|
# my @fields = $hit->field_names; |
| 164 |
|
|
# print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
| 165 |
|
|
}, |
| 166 |
|
|
maxhits => param('max_hits') || $max_hits, |
| 167 |
dpavlin |
22 |
\%params, |
| 168 |
dpavlin |
8 |
); |
| 169 |
|
|
|
| 170 |
|
|
die $SWISH::errstr unless $sh; |
| 171 |
|
|
|
| 172 |
|
|
|
| 173 |
|
|
$hits = $sh->query($s); |
| 174 |
|
|
|
| 175 |
dpavlin |
35 |
if ($hits && $hits > 0) { |
| 176 |
dpavlin |
16 |
print p,hr; |
| 177 |
dpavlin |
18 |
printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s); |
| 178 |
dpavlin |
8 |
} else { |
| 179 |
dpavlin |
16 |
print p; |
| 180 |
dpavlin |
18 |
printf (x($config->{text}->{no_hits}),$s,$sh->errstr); |
| 181 |
dpavlin |
8 |
} |
| 182 |
|
|
} else { |
| 183 |
dpavlin |
18 |
print p(x($config->{text}->{footer})); |
| 184 |
dpavlin |
8 |
} |
Parent Directory
| 
Revision Log