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 |
dpavlin |
59 |
use SWISH::API; |
7 |
dpavlin |
8 |
use XML::Simple; |
8 |
|
|
use Lingua::Spelling::Alternative; |
9 |
|
|
use Text::Iconv; |
10 |
dpavlin |
59 |
use Data::Pageset; |
11 |
dpavlin |
8 |
|
12 |
dpavlin |
59 |
# for pager |
13 |
|
|
my $pages_per_set = 20; |
14 |
|
|
|
15 |
dpavlin |
8 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
16 |
|
|
my $config=XMLin(undef, |
17 |
|
|
# keyattr => { label => "value" }, |
18 |
|
|
forcecontent => 0, |
19 |
|
|
); |
20 |
|
|
|
21 |
dpavlin |
18 |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
22 |
|
|
sub x { |
23 |
|
|
return $from_utf8->convert($_[0]); |
24 |
|
|
} |
25 |
|
|
|
26 |
dpavlin |
29 |
# Escape <, >, & and ", and to produce valid XML |
27 |
|
|
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"'); |
28 |
|
|
my $escape_re = join '|' => keys %escape; |
29 |
|
|
sub e { |
30 |
|
|
my $out; |
31 |
|
|
foreach my $v (@_) { |
32 |
|
|
$v =~ s/($escape_re)/$escape{$1}/g; |
33 |
|
|
$out .= $v; |
34 |
|
|
} |
35 |
|
|
return $out; |
36 |
|
|
} |
37 |
dpavlin |
8 |
|
38 |
dpavlin |
39 |
my @spellings; |
39 |
dpavlin |
8 |
# FIX: doesn't work very well |
40 |
dpavlin |
16 |
if ($config->{findaffix}) { |
41 |
dpavlin |
41 |
foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) { |
42 |
|
|
my $spelling_alt = new Lingua::Spelling::Alternative; |
43 |
|
|
$spelling_alt->load_findaffix($findaffix); |
44 |
|
|
push @spellings,$spelling_alt; |
45 |
|
|
} |
46 |
dpavlin |
8 |
} |
47 |
dpavlin |
16 |
if ($config->{affix}) { |
48 |
dpavlin |
41 |
foreach my $affix (split(/[, ]+/,x($config->{affix}))) { |
49 |
|
|
my $spelling_alt = new Lingua::Spelling::Alternative; |
50 |
|
|
$spelling_alt->load_affix($affix); |
51 |
|
|
push @spellings,$spelling_alt; |
52 |
|
|
} |
53 |
dpavlin |
16 |
} |
54 |
dpavlin |
8 |
|
55 |
|
|
my $hits=0; |
56 |
dpavlin |
59 |
my $max_hits=param('max_hits') || x($config->{max_hits}); |
57 |
dpavlin |
8 |
|
58 |
|
|
my %labels; |
59 |
|
|
foreach (@{$config->{labels}->{label}}) { |
60 |
dpavlin |
59 |
next if (! $_->{value}); # skip unlimited (0) |
61 |
dpavlin |
18 |
$labels{$_->{value}} = x($_->{content}); |
62 |
dpavlin |
8 |
} |
63 |
|
|
|
64 |
dpavlin |
23 |
my $path = param('path'); # limit to this path |
65 |
|
|
my %path_label; |
66 |
|
|
my @path_name; |
67 |
|
|
foreach (@{$config->{paths}->{path}}) { |
68 |
dpavlin |
39 |
|
69 |
|
|
print STDERR "##: $_->{limit}",x($_->{content}),"\n"; |
70 |
dpavlin |
23 |
push @path_name,x($_->{limit}); |
71 |
|
|
$path_label{$_->{limit}} = x($_->{content}); |
72 |
|
|
} |
73 |
|
|
|
74 |
dpavlin |
32 |
my @properties = split(/\s+/,x($config->{properties})); |
75 |
|
|
|
76 |
dpavlin |
16 |
if ($config->{charset}) { |
77 |
dpavlin |
18 |
print header(-charset=>x($config->{charset})); |
78 |
dpavlin |
16 |
} else { |
79 |
|
|
print header; |
80 |
|
|
} |
81 |
dpavlin |
18 |
print start_html(-title=>x($config->{title})),start_form; |
82 |
|
|
print x($config->{text}->{search}); |
83 |
dpavlin |
16 |
print popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits); |
84 |
dpavlin |
18 |
print x($config->{text}->{documents}); |
85 |
dpavlin |
16 |
print textfield('search'); |
86 |
dpavlin |
18 |
print submit(-value=> x($config->{text}->{submit})); |
87 |
dpavlin |
39 |
print br,checkbox(-name=>'no_affix', -checked=>0, -label=>x($config->{text}->{no_spell})) if (@spellings); |
88 |
dpavlin |
32 |
print checkbox(-name=>'no_properties', -checked=>0, -label=>($config->{text}->{no_properties})) if (@properties); |
89 |
dpavlin |
23 |
if (@path_name) { |
90 |
|
|
print br,x($config->{text}->{limit}); |
91 |
|
|
print popup_menu(-name=>'path',-values=>\@path_name,-labels=>\%path_label,-default=>$path); |
92 |
|
|
} |
93 |
dpavlin |
8 |
print end_form,hr; |
94 |
|
|
|
95 |
|
|
if (param('search')) { |
96 |
|
|
|
97 |
|
|
my $s; |
98 |
|
|
# re-write query from +/- to and/and not |
99 |
|
|
|
100 |
|
|
my $search = param('search'); |
101 |
|
|
my $s_phrase = ""; |
102 |
|
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
103 |
|
|
$s .= "$1 "; |
104 |
|
|
} |
105 |
|
|
$search =~ s/^\s+//; |
106 |
|
|
$search =~ s/\s+$//; |
107 |
|
|
|
108 |
dpavlin |
41 |
my %words; |
109 |
|
|
|
110 |
dpavlin |
8 |
foreach (split(/\s+/,$search)) { |
111 |
|
|
if (m/^([+-])(\S+)/) { |
112 |
|
|
$s.= ($s) ? "and " : ""; |
113 |
|
|
$s.="not " if ($1 eq "-"); |
114 |
dpavlin |
39 |
if (@spellings && !param('no_affix')) { |
115 |
dpavlin |
8 |
my $w = $2; $w =~ s/[\*\s]+//g; |
116 |
|
|
$w =~ s/^(['"]*)([^'"]+)(['"]*)/$2/; |
117 |
dpavlin |
39 |
my $or=""; |
118 |
|
|
foreach my $spelling_alt (@spellings) { |
119 |
|
|
$s.="$or$1(".join("* or ",$spelling_alt->alternatives($w))."*)$3 "; |
120 |
|
|
$or = "or "; |
121 |
|
|
} |
122 |
dpavlin |
8 |
} else { |
123 |
|
|
$s.="$2* "; |
124 |
|
|
} |
125 |
|
|
} else { |
126 |
dpavlin |
39 |
if (@spellings && !param('no_affix')) { |
127 |
dpavlin |
8 |
my $w = $_; $w =~ s/[\*\s]+//g; |
128 |
|
|
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
129 |
dpavlin |
39 |
my $or=""; |
130 |
|
|
foreach my $spelling_alt (@spellings) { |
131 |
|
|
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
132 |
|
|
$or = "or "; |
133 |
|
|
} |
134 |
dpavlin |
8 |
} else { |
135 |
|
|
$s.="$_* "; |
136 |
|
|
} |
137 |
|
|
} |
138 |
|
|
} |
139 |
|
|
|
140 |
|
|
# fixup search string |
141 |
|
|
$s=~tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
142 |
|
|
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
143 |
|
|
$s=~s/\*\*+/*/g; |
144 |
|
|
|
145 |
dpavlin |
23 |
# limit to some path |
146 |
dpavlin |
39 |
$s = "swishdocpath=(\"*$path*\") and $s" if ($path); |
147 |
dpavlin |
23 |
|
148 |
dpavlin |
22 |
my %params; # optional parametars for swish |
149 |
|
|
|
150 |
dpavlin |
32 |
# default format for output |
151 |
|
|
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
152 |
dpavlin |
21 |
|
153 |
dpavlin |
59 |
# output start of table |
154 |
|
|
print qq{ |
155 |
|
|
<table border="0"> |
156 |
|
|
}; |
157 |
|
|
# html before and after each hit |
158 |
|
|
my $tr_pre = qq{ |
159 |
|
|
<tr><td> |
160 |
|
|
}; |
161 |
|
|
my $tr_post = qq{ |
162 |
|
|
</td></tr> |
163 |
|
|
}; |
164 |
|
|
|
165 |
dpavlin |
32 |
if (@properties) { |
166 |
|
|
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
167 |
dpavlin |
39 |
$params{properties} = \@properties; |
168 |
dpavlin |
32 |
} else { |
169 |
|
|
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
170 |
|
|
} |
171 |
|
|
|
172 |
dpavlin |
59 |
# my $sh = SWISH->connect('Fork', |
173 |
|
|
# prog => x($config->{prog}), |
174 |
|
|
# indexes => x($config->{index}), |
175 |
|
|
# results => sub { |
176 |
|
|
# my ($sh,$hit) = @_; |
177 |
|
|
# |
178 |
|
|
# if ($config->{url}) { |
179 |
|
|
# printf ($hit_fmt ,"http://".virtual_host().x($config->{url}).$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties)); |
180 |
|
|
# } else { |
181 |
|
|
# printf ($hit_fmt ,$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties) ); |
182 |
|
|
# |
183 |
|
|
# } |
184 |
|
|
# |
185 |
|
|
## print $_[1]->as_string,"<br>\n"; |
186 |
|
|
## my @fields = $hit->field_names; |
187 |
|
|
## print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
188 |
|
|
# }, |
189 |
|
|
# maxhits => param('max_hits') || $max_hits, |
190 |
|
|
# \%params, |
191 |
|
|
# ); |
192 |
|
|
# |
193 |
|
|
# die $SWISH::errstr unless $sh; |
194 |
|
|
# |
195 |
|
|
# $hits = $sh->query($s); |
196 |
|
|
# |
197 |
|
|
# if ($hits && $hits > 0) { |
198 |
|
|
# print p,hr; |
199 |
|
|
# printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s); |
200 |
|
|
# } else { |
201 |
|
|
# print p; |
202 |
|
|
# printf (x($config->{text}->{no_hits}),$s,$sh->errstr); |
203 |
|
|
# } |
204 |
|
|
# if ($hits && $hits > 0) { |
205 |
|
|
# print p,hr; |
206 |
|
|
# printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s); |
207 |
|
|
# } else { |
208 |
|
|
# print p; |
209 |
|
|
# printf (x($config->{text}->{no_hits}),$s,$sh->errstr); |
210 |
|
|
# } |
211 |
|
|
|
212 |
|
|
my $swish = SWISH::API->new($config->{index}); |
213 |
|
|
|
214 |
|
|
$swish->AbortLastError if $swish->Error; |
215 |
|
|
|
216 |
|
|
my $results = $swish->Query($s); |
217 |
|
|
|
218 |
|
|
my $hits = $results->Hits; |
219 |
|
|
|
220 |
|
|
|
221 |
|
|
|
222 |
|
|
# build pager |
223 |
|
|
my $current_page = param('page') || 1; |
224 |
|
|
|
225 |
|
|
my $pager = Data::Pageset->new({ |
226 |
|
|
'total_entries' => $hits, |
227 |
|
|
'entries_per_page' => $max_hits, |
228 |
|
|
'current_page' => $current_page, |
229 |
|
|
'pages_per_set' => $pages_per_set, |
230 |
|
|
}); |
231 |
|
|
|
232 |
|
|
$results->SeekResult( $pager->first - 1 ); |
233 |
|
|
|
234 |
|
|
# get number of entries on this page |
235 |
|
|
my $i = $pager->entries_on_this_page; |
236 |
|
|
|
237 |
|
|
# print number of hits or error message |
238 |
|
|
if ( !$hits ) { |
239 |
|
|
printf (x($config->{text}->{no_hits}),$s,$swish->ErrorString); |
240 |
|
|
} else { |
241 |
|
|
printf (x($config->{text}->{hits}),$i,$results->Hits,$s); |
242 |
dpavlin |
57 |
} |
243 |
|
|
|
244 |
dpavlin |
8 |
|
245 |
dpavlin |
59 |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
246 |
|
|
|
247 |
|
|
my $result = $results->NextResult; |
248 |
|
|
last if (! $result); |
249 |
|
|
|
250 |
|
|
my @arr; |
251 |
|
|
foreach my $prop (@properties) { |
252 |
|
|
if ($prop =~ m/swishdescription/) { |
253 |
|
|
my $tmp = $result->Property($prop); |
254 |
|
|
$tmp =~ s/<[^>]+>//g; |
255 |
|
|
push @arr, $tmp; |
256 |
dpavlin |
18 |
} else { |
257 |
dpavlin |
59 |
push @arr, $result->Property($prop); |
258 |
dpavlin |
18 |
} |
259 |
dpavlin |
59 |
} |
260 |
dpavlin |
18 |
|
261 |
dpavlin |
59 |
print $tr_pre,$i,". "; |
262 |
|
|
if ($config->{url}) { |
263 |
|
|
printf($hit_fmt, "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath"), |
264 |
|
|
e($result->Property("swishtitle")) || 'untitled', |
265 |
|
|
$result->Property("swishrank"), |
266 |
|
|
@arr); |
267 |
|
|
} else { |
268 |
|
|
printf($hit_fmt, $result->Property("swishdocpath"), |
269 |
|
|
e($result->Property("swishtitle")) || 'untitled', |
270 |
|
|
$result->Property("swishrank"), |
271 |
|
|
@arr); |
272 |
|
|
} |
273 |
|
|
print $tr_post; |
274 |
dpavlin |
8 |
|
275 |
dpavlin |
59 |
} |
276 |
dpavlin |
8 |
|
277 |
dpavlin |
59 |
# pager navigation |
278 |
|
|
my $nav_html; |
279 |
dpavlin |
8 |
|
280 |
dpavlin |
59 |
my $nav_fmt=qq{ <a href="%s">%s</a> }; |
281 |
dpavlin |
8 |
|
282 |
dpavlin |
59 |
if ($pager->previous_set) { |
283 |
|
|
param('page', $pager->previous_set); |
284 |
|
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'<<'); |
285 |
dpavlin |
8 |
} |
286 |
dpavlin |
59 |
|
287 |
|
|
|
288 |
|
|
foreach my $p (@{$pager->pages_in_set()}) { |
289 |
dpavlin |
60 |
next if ($p < 0); |
290 |
dpavlin |
59 |
# for (my $p=$pager->previous_set; $p <= $pager->next_set; $p++) { |
291 |
|
|
if($p == $pager->current_page()) { |
292 |
|
|
$nav_html .= "<b>$p</b> "; |
293 |
|
|
} else { |
294 |
|
|
param('page', $p); |
295 |
|
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),$p); |
296 |
|
|
} |
297 |
|
|
} |
298 |
|
|
|
299 |
|
|
if ($pager->next_set) { |
300 |
|
|
param('page', $pager->next_set); |
301 |
|
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'>>'); |
302 |
|
|
} |
303 |
|
|
|
304 |
|
|
# end html table |
305 |
|
|
print qq{ |
306 |
|
|
<tr><td> |
307 |
|
|
Pages: $nav_html |
308 |
|
|
</td></tr> |
309 |
|
|
</table> |
310 |
|
|
}; |
311 |
|
|
|
312 |
|
|
|
313 |
|
|
|
314 |
dpavlin |
8 |
} else { |
315 |
dpavlin |
18 |
print p(x($config->{text}->{footer})); |
316 |
dpavlin |
8 |
} |