9 |
use Text::Iconv; |
use Text::Iconv; |
10 |
use Data::Pageset; |
use Data::Pageset; |
11 |
|
|
12 |
|
|
13 |
|
sub get_snippet { |
14 |
|
my $context_chars = 100; |
15 |
|
|
16 |
|
my $desc = shift || return ''; |
17 |
|
$desc = e($desc); |
18 |
|
|
19 |
|
# test if $desc contains any of our query words |
20 |
|
my @snips; |
21 |
|
|
22 |
|
my @colors = qw{#ffff66 #a0ffff #99ff99 #ff9999 #ff66ff}; |
23 |
|
|
24 |
|
# construct regex |
25 |
|
my $re = qq/^.*?(.{$context_chars}?)(\Q/ . join("|",@_) . |
26 |
|
qq/\E)(.{$context_chars})/; |
27 |
|
|
28 |
|
while ($desc =~ s/$re//si) { |
29 |
|
my ($bef,$qm,$af) = ($1, $2, $3); |
30 |
|
|
31 |
|
# no partial words... |
32 |
|
$bef =~ s,^\S+\s+|\s+\S+$,,gs; |
33 |
|
$af =~ s,^\S+\s+|\s+\S+$,,gs; |
34 |
|
|
35 |
|
push @snips, "$bef $qm $af"; |
36 |
|
} |
37 |
|
|
38 |
|
my $ellip = ' ... '; |
39 |
|
my $snippet; |
40 |
|
|
41 |
|
if (@snips) { |
42 |
|
$snippet = $ellip. join($ellip, @snips) . $ellip; |
43 |
|
} else { |
44 |
|
return ''; |
45 |
|
} |
46 |
|
|
47 |
|
# color offset |
48 |
|
my $i = 0; |
49 |
|
|
50 |
|
foreach my $w (@_) { |
51 |
|
$snippet =~ s,(\Q$w\E),<span style="background: $colors[$i]; color:black;">$1</span>,gsi; |
52 |
|
$i++; |
53 |
|
$i = 0 if ($i > $#colors); |
54 |
|
} |
55 |
|
|
56 |
|
return $snippet; |
57 |
|
} |
58 |
|
|
59 |
# for pager |
# for pager |
60 |
my $pages_per_set = 20; |
my $pages_per_set = 20; |
61 |
|
|
63 |
my $config=XMLin(undef, |
my $config=XMLin(undef, |
64 |
# keyattr => { label => "value" }, |
# keyattr => { label => "value" }, |
65 |
forcecontent => 0, |
forcecontent => 0, |
66 |
|
ForceArray => [ 'path' ], |
67 |
); |
); |
68 |
|
|
69 |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
88 |
# FIX: doesn't work very well |
# FIX: doesn't work very well |
89 |
if ($config->{findaffix}) { |
if ($config->{findaffix}) { |
90 |
foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) { |
foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) { |
91 |
|
next if (! -f $findaffix); |
92 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
my $spelling_alt = new Lingua::Spelling::Alternative; |
93 |
$spelling_alt->load_findaffix($findaffix); |
$spelling_alt->load_findaffix($findaffix); |
94 |
push @spellings,$spelling_alt; |
push @spellings,$spelling_alt; |
96 |
} |
} |
97 |
if ($config->{affix}) { |
if ($config->{affix}) { |
98 |
foreach my $affix (split(/[, ]+/,x($config->{affix}))) { |
foreach my $affix (split(/[, ]+/,x($config->{affix}))) { |
99 |
|
next if (! -f $affix); |
100 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
my $spelling_alt = new Lingua::Spelling::Alternative; |
101 |
$spelling_alt->load_affix($affix); |
$spelling_alt->load_affix($affix); |
102 |
push @spellings,$spelling_alt; |
push @spellings,$spelling_alt; |
112 |
$labels{$_->{value}} = x($_->{content}); |
$labels{$_->{value}} = x($_->{content}); |
113 |
} |
} |
114 |
|
|
115 |
my $path = param('path'); # limit to this path |
my $path; |
116 |
|
# limit to this path |
117 |
|
$path .= '"'.join('*" or "',param('path')).'*"' if (param('path')); |
118 |
my %path_label; |
my %path_label; |
119 |
my @path_name; |
my @path_name; |
120 |
foreach (@{$config->{paths}->{path}}) { |
foreach (@{$config->{paths}->{path}}) { |
|
|
|
|
print STDERR "##: $_->{limit}",x($_->{content}),"\n"; |
|
121 |
push @path_name,x($_->{limit}); |
push @path_name,x($_->{limit}); |
122 |
$path_label{$_->{limit}} = x($_->{content}); |
$path_label{$_->{limit}} = x($_->{content}); |
123 |
} |
} |
140 |
if (@path_name) { |
if (@path_name) { |
141 |
print br,x($config->{text}->{limit}); |
print br,x($config->{text}->{limit}); |
142 |
print popup_menu(-name=>'path',-values=>\@path_name,-labels=>\%path_label,-default=>$path); |
print popup_menu(-name=>'path',-values=>\@path_name,-labels=>\%path_label,-default=>$path); |
143 |
|
} elsif (param('path')) { |
144 |
|
print hidden(-name=>'path',-values=>param('path')); |
145 |
} |
} |
146 |
print end_form,hr; |
print end_form,hr; |
147 |
|
|
150 |
my $s; |
my $s; |
151 |
# re-write query from +/- to and/and not |
# re-write query from +/- to and/and not |
152 |
|
|
153 |
|
my @s_elem; |
154 |
|
|
155 |
my $search = param('search'); |
my $search = param('search'); |
156 |
my $s_phrase = ""; |
|
157 |
|
# strip spaces |
158 |
|
$search =~ s/^\s+//; |
159 |
|
$search =~ s/\s+$//; |
160 |
|
# fixup search string |
161 |
|
$search=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
162 |
|
$search=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
163 |
|
|
164 |
|
# extract phrases and put them first |
165 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
166 |
$s .= "$1 "; |
$s .= "$1 "; |
167 |
|
push @s_elem, $1; |
168 |
} |
} |
|
$search =~ s/^\s+//; |
|
|
$search =~ s/\s+$//; |
|
169 |
|
|
170 |
my %words; |
my %words; |
171 |
|
|
184 |
} else { |
} else { |
185 |
$s.="$2* "; |
$s.="$2* "; |
186 |
} |
} |
187 |
|
push @s_elem, $2; |
188 |
} else { |
} else { |
189 |
if (@spellings && !param('no_affix')) { |
if (@spellings && !param('no_affix')) { |
190 |
my $w = $_; $w =~ s/[\*\s]+//g; |
my $w = $_; $w =~ s/[\*\s]+//g; |
|
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
|
191 |
my $or=""; |
my $or=""; |
192 |
foreach my $spelling_alt (@spellings) { |
foreach my $spelling_alt (@spellings) { |
193 |
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
196 |
} else { |
} else { |
197 |
$s.="$_* "; |
$s.="$_* "; |
198 |
} |
} |
199 |
|
push @s_elem, $_; |
200 |
} |
} |
201 |
} |
} |
202 |
|
|
203 |
# fixup search string |
# fix multiple stars |
|
$s=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
|
|
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
|
204 |
$s=~s/\*\*+/*/g; |
$s=~s/\*\*+/*/g; |
205 |
|
|
206 |
# limit to some path |
# limit to some path |
207 |
$s = "swishdocpath=(\"*$path*\") and $s" if ($path); |
$s = "swishdocpath=($path) and $s" if ($path); |
208 |
|
|
209 |
my %params; # optional parametars for swish |
my %params; # optional parametars for swish |
210 |
|
|
211 |
# default format for output |
# default format for output |
212 |
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
213 |
|
|
|
# output start of table |
|
|
print qq{ |
|
|
<table border="0"> |
|
|
}; |
|
|
# html before and after each hit |
|
|
my $tr_pre = qq{ |
|
|
<tr><td> |
|
|
}; |
|
|
my $tr_post = qq{ |
|
|
</td></tr> |
|
|
}; |
|
|
|
|
214 |
if (@properties) { |
if (@properties) { |
215 |
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
216 |
$params{properties} = \@properties; |
$params{properties} = \@properties; |
218 |
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
219 |
} |
} |
220 |
|
|
|
# my $sh = SWISH->connect('Fork', |
|
|
# prog => x($config->{prog}), |
|
|
# indexes => x($config->{index}), |
|
|
# results => sub { |
|
|
# my ($sh,$hit) = @_; |
|
|
# |
|
|
# if ($config->{url}) { |
|
|
# printf ($hit_fmt ,"http://".virtual_host().x($config->{url}).$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties)); |
|
|
# } else { |
|
|
# printf ($hit_fmt ,$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties) ); |
|
|
# |
|
|
# } |
|
|
# |
|
|
## print $_[1]->as_string,"<br>\n"; |
|
|
## my @fields = $hit->field_names; |
|
|
## print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields; |
|
|
# }, |
|
|
# maxhits => param('max_hits') || $max_hits, |
|
|
# \%params, |
|
|
# ); |
|
|
# |
|
|
# die $SWISH::errstr unless $sh; |
|
|
# |
|
|
# $hits = $sh->query($s); |
|
|
# |
|
|
# if ($hits && $hits > 0) { |
|
|
# print p,hr; |
|
|
# printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s); |
|
|
# } else { |
|
|
# print p; |
|
|
# printf (x($config->{text}->{no_hits}),$s,$sh->errstr); |
|
|
# } |
|
|
# if ($hits && $hits > 0) { |
|
|
# print p,hr; |
|
|
# printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s); |
|
|
# } else { |
|
|
# print p; |
|
|
# printf (x($config->{text}->{no_hits}),$s,$sh->errstr); |
|
|
# } |
|
|
|
|
221 |
my $swish = SWISH::API->new($config->{index}); |
my $swish = SWISH::API->new($config->{index}); |
|
|
|
222 |
$swish->AbortLastError if $swish->Error; |
$swish->AbortLastError if $swish->Error; |
|
|
|
223 |
my $results = $swish->Query($s); |
my $results = $swish->Query($s); |
|
|
|
224 |
my $hits = $results->Hits; |
my $hits = $results->Hits; |
225 |
|
|
226 |
|
|
|
|
|
227 |
# build pager |
# build pager |
228 |
my $current_page = param('page') || 1; |
my $current_page = param('page') || 1; |
229 |
|
|
246 |
printf (x($config->{text}->{hits}),$i,$results->Hits,$s); |
printf (x($config->{text}->{hits}),$i,$results->Hits,$s); |
247 |
} |
} |
248 |
|
|
249 |
|
my %path2title; |
250 |
|
foreach my $p (@{$config->{path2title}->{path}}) { |
251 |
|
$path2title{$p->{dir}} = $p->{content}; |
252 |
|
} |
253 |
|
|
254 |
|
# output start of table |
255 |
|
print qq{ |
256 |
|
<table border="0"> |
257 |
|
}; |
258 |
|
# html before and after each hit |
259 |
|
my $tr_pre = qq{ |
260 |
|
<tr><td> |
261 |
|
}; |
262 |
|
my $tr_post = qq{ |
263 |
|
</td></tr> |
264 |
|
}; |
265 |
|
|
266 |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
267 |
|
|
269 |
last if (! $result); |
last if (! $result); |
270 |
|
|
271 |
my @arr; |
my @arr; |
272 |
|
|
273 |
foreach my $prop (@properties) { |
foreach my $prop (@properties) { |
274 |
if ($prop =~ m/swishdescription/) { |
if ($prop =~ m/swishdescription/) { |
275 |
my $tmp = $result->Property($prop); |
my $tmp = get_snippet( |
276 |
$tmp =~ s/<[^>]+>//g; |
$result->Property($prop), |
277 |
|
@s_elem, |
278 |
|
); |
279 |
|
|
280 |
push @arr, $tmp; |
push @arr, $tmp; |
281 |
} else { |
} else { |
282 |
push @arr, $result->Property($prop); |
push @arr, $result->Property($prop); |
287 |
my $rank = $result->Property("swishrank"); |
my $rank = $result->Property("swishrank"); |
288 |
my $host = $result->Property("swishdocpath"); |
my $host = $result->Property("swishdocpath"); |
289 |
$host = "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath") if ($config->{url}); |
$host = "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath") if ($config->{url}); |
290 |
|
|
291 |
|
foreach my $p (keys %path2title) { |
292 |
|
if ($host =~ m/$p/i) { |
293 |
|
$title =~ s/$path2title{$p}\s*[:-]+\s*//; |
294 |
|
$title = $path2title{$p}." :: ".$title; |
295 |
|
last; |
296 |
|
} |
297 |
|
} |
298 |
|
|
299 |
print $tr_pre,$i,". "; |
print $tr_pre,$i,". "; |
300 |
# print collection name which is not link |
# print collection name which is not link |
301 |
if ($title =~ s/^(.+? :: )//) { |
if ($title =~ s/^(.+? :: )//) { |
302 |
print $1; |
print $1; |
303 |
} |
} |
304 |
|
|
305 |
printf($hit_fmt, $host, $title || 'untitled', $rank, @arr); |
printf($hit_fmt, $host, $title || 'untitled', $rank, @arr); |
306 |
print $tr_post; |
print $tr_post; |
307 |
|
|
312 |
|
|
313 |
my $nav_fmt=qq{ <a href="%s">%s</a> }; |
my $nav_fmt=qq{ <a href="%s">%s</a> }; |
314 |
|
|
315 |
|
if ($pager->current_page() > $pager->first_page) { |
316 |
|
param('page', $pager->current_page - 1); |
317 |
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'<<'); |
318 |
|
} |
319 |
|
|
320 |
if ($pager->previous_set) { |
if ($pager->previous_set) { |
321 |
param('page', $pager->previous_set); |
param('page', $pager->previous_set); |
322 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'<<'); |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..'); |
323 |
} |
} |
324 |
|
|
325 |
|
|
336 |
|
|
337 |
if ($pager->next_set) { |
if ($pager->next_set) { |
338 |
param('page', $pager->next_set); |
param('page', $pager->next_set); |
339 |
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..'); |
340 |
|
} |
341 |
|
|
342 |
|
if ($pager->current_page() < $pager->last_page) { |
343 |
|
param('page', $pager->current_page + 1); |
344 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'>>'); |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'>>'); |
345 |
} |
} |
346 |
|
|
347 |
|
if ($config->{text}->{pages}) { |
348 |
|
$nav_html = x($config->{text}->{pages})." ".$nav_html; |
349 |
|
} |
350 |
|
|
351 |
# end html table |
# end html table |
352 |
print qq{ |
print qq{ |
353 |
<tr><td> |
<tr><td> |
354 |
Pages: $nav_html |
$nav_html |
355 |
</td></tr> |
</td></tr> |
356 |
</table> |
</table> |
357 |
}; |
}; |