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 |
81 |
|
13 |
|
|
sub get_snippet { |
14 |
|
|
my $context_chars = 100; |
15 |
dpavlin |
82 |
my $max_desc = 16384; # 16k to filter |
16 |
dpavlin |
81 |
|
17 |
|
|
my $desc = shift || return ''; |
18 |
dpavlin |
82 |
$desc = substr($desc,0,$max_desc) if (length($desc) > $max_desc); |
19 |
dpavlin |
83 |
$desc = e($desc); |
20 |
dpavlin |
81 |
# test if $desc contains any of our query words |
21 |
|
|
my @snips; |
22 |
|
|
|
23 |
|
|
my @colors = qw{#ffff66 #a0ffff #99ff99 #ff9999 #ff66ff}; |
24 |
|
|
|
25 |
|
|
my $i = 0; |
26 |
|
|
|
27 |
|
|
for my $q (@_) { |
28 |
|
|
if ($desc =~ m/(.*?)(\Q$q\E)(.*)/si) { |
29 |
|
|
my $bef = $1; |
30 |
|
|
my $qm = $2; |
31 |
|
|
my $af = $3; |
32 |
|
|
$bef = substr $bef, -$context_chars; |
33 |
|
|
$af = substr $af, 0, $context_chars; |
34 |
|
|
|
35 |
|
|
# no partial words... |
36 |
|
|
$af =~ s,^\S+\s+|\s+\S+$,,gs; |
37 |
|
|
$bef =~ s,^\S+\s+|\s+\S+$,,gs; |
38 |
|
|
|
39 |
dpavlin |
82 |
push(@snips, "$bef <span style=\"background:".$colors[$i]."; color:black;\">$qm</span> $af"); |
40 |
dpavlin |
81 |
$i++; |
41 |
|
|
$i = 0 if ($i > $#colors); |
42 |
|
|
} |
43 |
|
|
} |
44 |
|
|
my $ellip = ' ... '; |
45 |
dpavlin |
82 |
my $snippet = $ellip. join($ellip, @snips) . $ellip if (@snips); |
46 |
dpavlin |
81 |
|
47 |
|
|
return $snippet; |
48 |
|
|
} |
49 |
|
|
|
50 |
dpavlin |
59 |
# for pager |
51 |
|
|
my $pages_per_set = 20; |
52 |
|
|
|
53 |
dpavlin |
8 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
54 |
|
|
my $config=XMLin(undef, |
55 |
|
|
# keyattr => { label => "value" }, |
56 |
|
|
forcecontent => 0, |
57 |
dpavlin |
80 |
ForceArray => [ 'path' ], |
58 |
dpavlin |
8 |
); |
59 |
|
|
|
60 |
dpavlin |
18 |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
61 |
|
|
sub x { |
62 |
dpavlin |
62 |
return if (! defined $_[0]); |
63 |
dpavlin |
18 |
return $from_utf8->convert($_[0]); |
64 |
|
|
} |
65 |
|
|
|
66 |
dpavlin |
29 |
# Escape <, >, & and ", and to produce valid XML |
67 |
|
|
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"'); |
68 |
|
|
my $escape_re = join '|' => keys %escape; |
69 |
|
|
sub e { |
70 |
|
|
my $out; |
71 |
|
|
foreach my $v (@_) { |
72 |
|
|
$v =~ s/($escape_re)/$escape{$1}/g; |
73 |
|
|
$out .= $v; |
74 |
|
|
} |
75 |
|
|
return $out; |
76 |
|
|
} |
77 |
dpavlin |
8 |
|
78 |
dpavlin |
39 |
my @spellings; |
79 |
dpavlin |
8 |
# FIX: doesn't work very well |
80 |
dpavlin |
16 |
if ($config->{findaffix}) { |
81 |
dpavlin |
41 |
foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) { |
82 |
dpavlin |
80 |
next if (! -f $findaffix); |
83 |
dpavlin |
41 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
84 |
|
|
$spelling_alt->load_findaffix($findaffix); |
85 |
|
|
push @spellings,$spelling_alt; |
86 |
|
|
} |
87 |
dpavlin |
8 |
} |
88 |
dpavlin |
16 |
if ($config->{affix}) { |
89 |
dpavlin |
41 |
foreach my $affix (split(/[, ]+/,x($config->{affix}))) { |
90 |
dpavlin |
80 |
next if (! -f $affix); |
91 |
dpavlin |
41 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
92 |
|
|
$spelling_alt->load_affix($affix); |
93 |
|
|
push @spellings,$spelling_alt; |
94 |
|
|
} |
95 |
dpavlin |
16 |
} |
96 |
dpavlin |
8 |
|
97 |
|
|
my $hits=0; |
98 |
dpavlin |
59 |
my $max_hits=param('max_hits') || x($config->{max_hits}); |
99 |
dpavlin |
8 |
|
100 |
|
|
my %labels; |
101 |
|
|
foreach (@{$config->{labels}->{label}}) { |
102 |
dpavlin |
59 |
next if (! $_->{value}); # skip unlimited (0) |
103 |
dpavlin |
18 |
$labels{$_->{value}} = x($_->{content}); |
104 |
dpavlin |
8 |
} |
105 |
|
|
|
106 |
dpavlin |
81 |
my $path; |
107 |
|
|
# limit to this path |
108 |
|
|
$path .= '"'.join('*" or "',param('path')).'*"' if (param('path')); |
109 |
dpavlin |
23 |
my %path_label; |
110 |
|
|
my @path_name; |
111 |
|
|
foreach (@{$config->{paths}->{path}}) { |
112 |
dpavlin |
39 |
|
113 |
|
|
print STDERR "##: $_->{limit}",x($_->{content}),"\n"; |
114 |
dpavlin |
23 |
push @path_name,x($_->{limit}); |
115 |
|
|
$path_label{$_->{limit}} = x($_->{content}); |
116 |
|
|
} |
117 |
|
|
|
118 |
dpavlin |
62 |
my @properties = split(/\s+/,x($config->{properties})) if ($config->{properties}); |
119 |
dpavlin |
32 |
|
120 |
dpavlin |
16 |
if ($config->{charset}) { |
121 |
dpavlin |
18 |
print header(-charset=>x($config->{charset})); |
122 |
dpavlin |
16 |
} else { |
123 |
|
|
print header; |
124 |
|
|
} |
125 |
dpavlin |
18 |
print start_html(-title=>x($config->{title})),start_form; |
126 |
|
|
print x($config->{text}->{search}); |
127 |
dpavlin |
16 |
print popup_menu(-name=>'max_hits',-values=>[ sort keys %labels ],-labels=>\%labels,-default=>$max_hits); |
128 |
dpavlin |
18 |
print x($config->{text}->{documents}); |
129 |
dpavlin |
16 |
print textfield('search'); |
130 |
dpavlin |
18 |
print submit(-value=> x($config->{text}->{submit})); |
131 |
dpavlin |
39 |
print br,checkbox(-name=>'no_affix', -checked=>0, -label=>x($config->{text}->{no_spell})) if (@spellings); |
132 |
dpavlin |
32 |
print checkbox(-name=>'no_properties', -checked=>0, -label=>($config->{text}->{no_properties})) if (@properties); |
133 |
dpavlin |
23 |
if (@path_name) { |
134 |
|
|
print br,x($config->{text}->{limit}); |
135 |
|
|
print popup_menu(-name=>'path',-values=>\@path_name,-labels=>\%path_label,-default=>$path); |
136 |
dpavlin |
81 |
} elsif (param('path')) { |
137 |
|
|
print hidden(-name=>'path',-values=>param('path')); |
138 |
dpavlin |
23 |
} |
139 |
dpavlin |
8 |
print end_form,hr; |
140 |
|
|
|
141 |
|
|
if (param('search')) { |
142 |
|
|
|
143 |
|
|
my $s; |
144 |
|
|
# re-write query from +/- to and/and not |
145 |
|
|
|
146 |
|
|
my $search = param('search'); |
147 |
|
|
my $s_phrase = ""; |
148 |
|
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
149 |
|
|
$s .= "$1 "; |
150 |
|
|
} |
151 |
|
|
$search =~ s/^\s+//; |
152 |
|
|
$search =~ s/\s+$//; |
153 |
|
|
|
154 |
dpavlin |
41 |
my %words; |
155 |
|
|
|
156 |
dpavlin |
8 |
foreach (split(/\s+/,$search)) { |
157 |
|
|
if (m/^([+-])(\S+)/) { |
158 |
|
|
$s.= ($s) ? "and " : ""; |
159 |
|
|
$s.="not " if ($1 eq "-"); |
160 |
dpavlin |
39 |
if (@spellings && !param('no_affix')) { |
161 |
dpavlin |
8 |
my $w = $2; $w =~ s/[\*\s]+//g; |
162 |
|
|
$w =~ s/^(['"]*)([^'"]+)(['"]*)/$2/; |
163 |
dpavlin |
39 |
my $or=""; |
164 |
|
|
foreach my $spelling_alt (@spellings) { |
165 |
|
|
$s.="$or$1(".join("* or ",$spelling_alt->alternatives($w))."*)$3 "; |
166 |
|
|
$or = "or "; |
167 |
|
|
} |
168 |
dpavlin |
8 |
} else { |
169 |
|
|
$s.="$2* "; |
170 |
|
|
} |
171 |
|
|
} else { |
172 |
dpavlin |
39 |
if (@spellings && !param('no_affix')) { |
173 |
dpavlin |
8 |
my $w = $_; $w =~ s/[\*\s]+//g; |
174 |
|
|
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
175 |
dpavlin |
39 |
my $or=""; |
176 |
|
|
foreach my $spelling_alt (@spellings) { |
177 |
|
|
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
178 |
|
|
$or = "or "; |
179 |
|
|
} |
180 |
dpavlin |
8 |
} else { |
181 |
|
|
$s.="$_* "; |
182 |
|
|
} |
183 |
|
|
} |
184 |
|
|
} |
185 |
|
|
|
186 |
|
|
# fixup search string |
187 |
|
|
$s=~tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
188 |
|
|
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
189 |
|
|
$s=~s/\*\*+/*/g; |
190 |
|
|
|
191 |
dpavlin |
23 |
# limit to some path |
192 |
dpavlin |
81 |
$s = "swishdocpath=($path) and $s" if ($path); |
193 |
dpavlin |
23 |
|
194 |
dpavlin |
22 |
my %params; # optional parametars for swish |
195 |
|
|
|
196 |
dpavlin |
32 |
# default format for output |
197 |
|
|
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
198 |
dpavlin |
21 |
|
199 |
dpavlin |
32 |
if (@properties) { |
200 |
|
|
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
201 |
dpavlin |
39 |
$params{properties} = \@properties; |
202 |
dpavlin |
32 |
} else { |
203 |
|
|
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
204 |
|
|
} |
205 |
|
|
|
206 |
dpavlin |
59 |
my $swish = SWISH::API->new($config->{index}); |
207 |
|
|
$swish->AbortLastError if $swish->Error; |
208 |
|
|
my $results = $swish->Query($s); |
209 |
|
|
my $hits = $results->Hits; |
210 |
|
|
|
211 |
|
|
|
212 |
|
|
# build pager |
213 |
|
|
my $current_page = param('page') || 1; |
214 |
|
|
|
215 |
|
|
my $pager = Data::Pageset->new({ |
216 |
|
|
'total_entries' => $hits, |
217 |
|
|
'entries_per_page' => $max_hits, |
218 |
|
|
'current_page' => $current_page, |
219 |
|
|
'pages_per_set' => $pages_per_set, |
220 |
|
|
}); |
221 |
|
|
|
222 |
|
|
$results->SeekResult( $pager->first - 1 ); |
223 |
|
|
|
224 |
|
|
# get number of entries on this page |
225 |
|
|
my $i = $pager->entries_on_this_page; |
226 |
|
|
|
227 |
|
|
# print number of hits or error message |
228 |
|
|
if ( !$hits ) { |
229 |
|
|
printf (x($config->{text}->{no_hits}),$s,$swish->ErrorString); |
230 |
|
|
} else { |
231 |
|
|
printf (x($config->{text}->{hits}),$i,$results->Hits,$s); |
232 |
dpavlin |
57 |
} |
233 |
|
|
|
234 |
dpavlin |
80 |
my %path2title; |
235 |
|
|
foreach my $p (@{$config->{path2title}->{path}}) { |
236 |
|
|
$path2title{$p->{dir}} = $p->{content}; |
237 |
|
|
} |
238 |
dpavlin |
8 |
|
239 |
dpavlin |
83 |
# output start of table |
240 |
|
|
print qq{ |
241 |
|
|
<table border="0"> |
242 |
|
|
}; |
243 |
|
|
# html before and after each hit |
244 |
|
|
my $tr_pre = qq{ |
245 |
|
|
<tr><td> |
246 |
|
|
}; |
247 |
|
|
my $tr_post = qq{ |
248 |
|
|
</td></tr> |
249 |
|
|
}; |
250 |
|
|
|
251 |
dpavlin |
59 |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
252 |
|
|
|
253 |
|
|
my $result = $results->NextResult; |
254 |
|
|
last if (! $result); |
255 |
|
|
|
256 |
|
|
my @arr; |
257 |
dpavlin |
81 |
|
258 |
dpavlin |
59 |
foreach my $prop (@properties) { |
259 |
|
|
if ($prop =~ m/swishdescription/) { |
260 |
dpavlin |
81 |
my $tmp = get_snippet( |
261 |
|
|
$result->Property($prop), |
262 |
|
|
split(/\s+/,$search) |
263 |
|
|
); |
264 |
|
|
|
265 |
dpavlin |
59 |
push @arr, $tmp; |
266 |
dpavlin |
18 |
} else { |
267 |
dpavlin |
59 |
push @arr, $result->Property($prop); |
268 |
dpavlin |
18 |
} |
269 |
dpavlin |
59 |
} |
270 |
dpavlin |
18 |
|
271 |
dpavlin |
73 |
my $title = e($result->Property("swishtitle")) || 'untitled'; |
272 |
|
|
my $rank = $result->Property("swishrank"); |
273 |
|
|
my $host = $result->Property("swishdocpath"); |
274 |
|
|
$host = "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath") if ($config->{url}); |
275 |
dpavlin |
80 |
|
276 |
|
|
foreach my $p (keys %path2title) { |
277 |
|
|
if ($host =~ m/$p/i) { |
278 |
|
|
$title =~ s/$path2title{$p}\s*[:-]+\s*//; |
279 |
|
|
$title = $path2title{$p}." :: ".$title; |
280 |
|
|
last; |
281 |
|
|
} |
282 |
|
|
} |
283 |
|
|
|
284 |
dpavlin |
59 |
print $tr_pre,$i,". "; |
285 |
dpavlin |
73 |
# print collection name which is not link |
286 |
|
|
if ($title =~ s/^(.+? :: )//) { |
287 |
|
|
print $1; |
288 |
dpavlin |
59 |
} |
289 |
dpavlin |
80 |
|
290 |
dpavlin |
73 |
printf($hit_fmt, $host, $title || 'untitled', $rank, @arr); |
291 |
dpavlin |
59 |
print $tr_post; |
292 |
dpavlin |
8 |
|
293 |
dpavlin |
59 |
} |
294 |
dpavlin |
8 |
|
295 |
dpavlin |
59 |
# pager navigation |
296 |
|
|
my $nav_html; |
297 |
dpavlin |
8 |
|
298 |
dpavlin |
59 |
my $nav_fmt=qq{ <a href="%s">%s</a> }; |
299 |
dpavlin |
8 |
|
300 |
dpavlin |
75 |
if ($pager->current_page() > $pager->first_page) { |
301 |
|
|
param('page', $pager->current_page - 1); |
302 |
|
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'<<'); |
303 |
|
|
} |
304 |
|
|
|
305 |
dpavlin |
59 |
if ($pager->previous_set) { |
306 |
|
|
param('page', $pager->previous_set); |
307 |
dpavlin |
75 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..'); |
308 |
dpavlin |
8 |
} |
309 |
dpavlin |
59 |
|
310 |
|
|
|
311 |
|
|
foreach my $p (@{$pager->pages_in_set()}) { |
312 |
dpavlin |
60 |
next if ($p < 0); |
313 |
dpavlin |
59 |
# for (my $p=$pager->previous_set; $p <= $pager->next_set; $p++) { |
314 |
|
|
if($p == $pager->current_page()) { |
315 |
|
|
$nav_html .= "<b>$p</b> "; |
316 |
|
|
} else { |
317 |
|
|
param('page', $p); |
318 |
|
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),$p); |
319 |
|
|
} |
320 |
|
|
} |
321 |
|
|
|
322 |
|
|
if ($pager->next_set) { |
323 |
|
|
param('page', $pager->next_set); |
324 |
dpavlin |
75 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..'); |
325 |
|
|
} |
326 |
|
|
|
327 |
|
|
if ($pager->current_page() < $pager->last_page) { |
328 |
|
|
param('page', $pager->current_page + 1); |
329 |
dpavlin |
59 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'>>'); |
330 |
|
|
} |
331 |
|
|
|
332 |
dpavlin |
76 |
if ($config->{text}->{pages}) { |
333 |
|
|
$nav_html = x($config->{text}->{pages})." ".$nav_html; |
334 |
|
|
} |
335 |
|
|
|
336 |
dpavlin |
59 |
# end html table |
337 |
|
|
print qq{ |
338 |
|
|
<tr><td> |
339 |
dpavlin |
76 |
$nav_html |
340 |
dpavlin |
59 |
</td></tr> |
341 |
|
|
</table> |
342 |
|
|
}; |
343 |
|
|
|
344 |
|
|
|
345 |
|
|
|
346 |
dpavlin |
8 |
} else { |
347 |
dpavlin |
18 |
print p(x($config->{text}->{footer})); |
348 |
dpavlin |
8 |
} |