8 |
use Lingua::Spelling::Alternative; |
use Lingua::Spelling::Alternative; |
9 |
use Text::Iconv; |
use Text::Iconv; |
10 |
use Data::Pageset; |
use Data::Pageset; |
11 |
|
use FormatResult; |
12 |
|
|
13 |
# for pager |
# for pager |
14 |
my $pages_per_set = 20; |
my $pages_per_set = 20; |
17 |
my $config=XMLin(undef, |
my $config=XMLin(undef, |
18 |
# keyattr => { label => "value" }, |
# keyattr => { label => "value" }, |
19 |
forcecontent => 0, |
forcecontent => 0, |
20 |
|
ForceArray => [ 'path' ], |
21 |
); |
); |
22 |
|
|
23 |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset}); |
24 |
sub x { |
sub x { |
25 |
|
return if (! defined $_[0]); |
26 |
return $from_utf8->convert($_[0]); |
return $from_utf8->convert($_[0]); |
27 |
} |
} |
28 |
|
|
42 |
# FIX: doesn't work very well |
# FIX: doesn't work very well |
43 |
if ($config->{findaffix}) { |
if ($config->{findaffix}) { |
44 |
foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) { |
foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) { |
45 |
|
next if (! -f $findaffix); |
46 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
my $spelling_alt = new Lingua::Spelling::Alternative; |
47 |
$spelling_alt->load_findaffix($findaffix); |
$spelling_alt->load_findaffix($findaffix); |
48 |
push @spellings,$spelling_alt; |
push @spellings,$spelling_alt; |
50 |
} |
} |
51 |
if ($config->{affix}) { |
if ($config->{affix}) { |
52 |
foreach my $affix (split(/[, ]+/,x($config->{affix}))) { |
foreach my $affix (split(/[, ]+/,x($config->{affix}))) { |
53 |
|
next if (! -f $affix); |
54 |
my $spelling_alt = new Lingua::Spelling::Alternative; |
my $spelling_alt = new Lingua::Spelling::Alternative; |
55 |
$spelling_alt->load_affix($affix); |
$spelling_alt->load_affix($affix); |
56 |
push @spellings,$spelling_alt; |
push @spellings,$spelling_alt; |
66 |
$labels{$_->{value}} = x($_->{content}); |
$labels{$_->{value}} = x($_->{content}); |
67 |
} |
} |
68 |
|
|
69 |
my $path = param('path'); # limit to this path |
my $path; |
70 |
|
# limit to this path |
71 |
|
$path .= '"'.join('*" or "',param('path')).'*"' if (param('path')); |
72 |
my %path_label; |
my %path_label; |
73 |
my @path_name; |
my @path_name; |
74 |
foreach (@{$config->{paths}->{path}}) { |
foreach (@{$config->{paths}->{path}}) { |
|
|
|
|
print STDERR "##: $_->{limit}",x($_->{content}),"\n"; |
|
75 |
push @path_name,x($_->{limit}); |
push @path_name,x($_->{limit}); |
76 |
$path_label{$_->{limit}} = x($_->{content}); |
$path_label{$_->{limit}} = x($_->{content}); |
77 |
} |
} |
78 |
|
|
79 |
my @properties = split(/\s+/,x($config->{properties})); |
my @properties = split(/\s+/,x($config->{properties})) if ($config->{properties}); |
80 |
|
|
81 |
if ($config->{charset}) { |
if ($config->{charset}) { |
82 |
print header(-charset=>x($config->{charset})); |
print header(-charset=>x($config->{charset})); |
94 |
if (@path_name) { |
if (@path_name) { |
95 |
print br,x($config->{text}->{limit}); |
print br,x($config->{text}->{limit}); |
96 |
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); |
97 |
|
} elsif (param('path')) { |
98 |
|
print hidden(-name=>'path',-values=>param('path')); |
99 |
} |
} |
100 |
print end_form,hr; |
print end_form,hr; |
101 |
|
|
104 |
my $s; |
my $s; |
105 |
# re-write query from +/- to and/and not |
# re-write query from +/- to and/and not |
106 |
|
|
107 |
|
my @s_highlite; |
108 |
|
|
109 |
my $search = param('search'); |
my $search = param('search'); |
110 |
my $s_phrase = ""; |
|
111 |
|
# strip spaces |
112 |
|
$search =~ s/^\s+//; |
113 |
|
$search =~ s/\s+$//; |
114 |
|
# fixup search string |
115 |
|
$search=~tr/¹ð¾èæ©Ð®ÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
116 |
|
$search=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
117 |
|
|
118 |
|
# extract phrases and put them first |
119 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
120 |
$s .= "$1 "; |
$s .= "$1 "; |
121 |
|
push @s_highlite, $1; |
122 |
} |
} |
|
$search =~ s/^\s+//; |
|
|
$search =~ s/\s+$//; |
|
123 |
|
|
124 |
my %words; |
my %words; |
125 |
|
|
138 |
} else { |
} else { |
139 |
$s.="$2* "; |
$s.="$2* "; |
140 |
} |
} |
141 |
|
push @s_highlite, $2 if ($1 ne "-"); |
142 |
} else { |
} else { |
143 |
if (@spellings && !param('no_affix')) { |
if (@spellings && !param('no_affix')) { |
144 |
my $w = $_; $w =~ s/[\*\s]+//g; |
my $w = $_; $w =~ s/[\*\s]+//g; |
|
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
|
145 |
my $or=""; |
my $or=""; |
146 |
foreach my $spelling_alt (@spellings) { |
foreach my $spelling_alt (@spellings) { |
147 |
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
150 |
} else { |
} else { |
151 |
$s.="$_* "; |
$s.="$_* "; |
152 |
} |
} |
153 |
|
push @s_highlite, $_; |
154 |
} |
} |
155 |
} |
} |
156 |
|
|
157 |
# fixup search string |
# fix multiple stars |
|
$s=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
|
|
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
|
158 |
$s=~s/\*\*+/*/g; |
$s=~s/\*\*+/*/g; |
159 |
|
|
160 |
# limit to some path |
# limit to some path |
161 |
$s = "swishdocpath=(\"*$path*\") and $s" if ($path); |
$s = "swishdocpath=($path) and $s" if ($path); |
162 |
|
|
163 |
my %params; # optional parametars for swish |
my %params; # optional parametars for swish |
164 |
|
|
165 |
# default format for output |
# default format for output |
166 |
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
167 |
|
|
|
# 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> |
|
|
}; |
|
|
|
|
168 |
if (@properties) { |
if (@properties) { |
169 |
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
170 |
$params{properties} = \@properties; |
$params{properties} = \@properties; |
172 |
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
$hit_fmt = x($config->{hit}) if (x($config->{hit})); |
173 |
} |
} |
174 |
|
|
|
# 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); |
|
|
# } |
|
|
|
|
175 |
my $swish = SWISH::API->new($config->{index}); |
my $swish = SWISH::API->new($config->{index}); |
|
|
|
176 |
$swish->AbortLastError if $swish->Error; |
$swish->AbortLastError if $swish->Error; |
|
|
|
177 |
my $results = $swish->Query($s); |
my $results = $swish->Query($s); |
|
|
|
178 |
my $hits = $results->Hits; |
my $hits = $results->Hits; |
179 |
|
|
180 |
|
|
|
|
|
181 |
# build pager |
# build pager |
182 |
my $current_page = param('page') || 1; |
my $current_page = param('page') || 1; |
183 |
|
|
200 |
printf (x($config->{text}->{hits}),$i,$results->Hits,$s); |
printf (x($config->{text}->{hits}),$i,$results->Hits,$s); |
201 |
} |
} |
202 |
|
|
203 |
|
my %path2title; |
204 |
|
foreach my $p (@{$config->{path2title}->{path}}) { |
205 |
|
$path2title{$p->{dir}} = $p->{content}; |
206 |
|
} |
207 |
|
|
208 |
|
# output start of table |
209 |
|
print qq{ |
210 |
|
<table border="0"> |
211 |
|
}; |
212 |
|
# html before and after each hit |
213 |
|
my $tr_pre = qq{ |
214 |
|
<tr><td> |
215 |
|
}; |
216 |
|
my $tr_post = qq{ |
217 |
|
</td></tr> |
218 |
|
}; |
219 |
|
|
220 |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
221 |
|
|
223 |
last if (! $result); |
last if (! $result); |
224 |
|
|
225 |
my @arr; |
my @arr; |
226 |
|
|
227 |
foreach my $prop (@properties) { |
foreach my $prop (@properties) { |
228 |
if ($prop =~ m/swishdescription/) { |
if ($prop =~ m/swishdescription/) { |
229 |
my $tmp = $result->Property($prop); |
my $tmp = FormatResult::get_snippet( |
230 |
$tmp =~ s/<[^>]+>//g; |
e($result->Property($prop)), |
231 |
|
@s_highlite, |
232 |
|
); |
233 |
|
|
234 |
push @arr, $tmp; |
push @arr, $tmp; |
235 |
} else { |
} else { |
236 |
push @arr, $result->Property($prop); |
push @arr, $result->Property($prop); |
237 |
} |
} |
238 |
} |
} |
239 |
|
|
240 |
|
my $title = e($result->Property("swishtitle")) || 'untitled'; |
241 |
|
my $rank = $result->Property("swishrank"); |
242 |
|
my $host = $result->Property("swishdocpath"); |
243 |
|
$host = "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath") if ($config->{url}); |
244 |
|
|
245 |
|
foreach my $p (keys %path2title) { |
246 |
|
if ($host =~ m/$p/i) { |
247 |
|
$title =~ s/$path2title{$p}\s*[:-]+\s*//; |
248 |
|
$title = $path2title{$p}." :: ".$title; |
249 |
|
last; |
250 |
|
} |
251 |
|
} |
252 |
|
|
253 |
print $tr_pre,$i,". "; |
print $tr_pre,$i,". "; |
254 |
if ($config->{url}) { |
# print collection name which is not link |
255 |
printf($hit_fmt, "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath"), |
if ($title =~ s/^(.+? :: )//) { |
256 |
e($result->Property("swishtitle")) || 'untitled', |
print $1; |
|
$result->Property("swishrank"), |
|
|
@arr); |
|
|
} else { |
|
|
printf($hit_fmt, $result->Property("swishdocpath"), |
|
|
e($result->Property("swishtitle")) || 'untitled', |
|
|
$result->Property("swishrank"), |
|
|
@arr); |
|
257 |
} |
} |
258 |
|
|
259 |
|
printf($hit_fmt, $host, $title || 'untitled', $rank, @arr); |
260 |
print $tr_post; |
print $tr_post; |
261 |
|
|
262 |
} |
} |
266 |
|
|
267 |
my $nav_fmt=qq{ <a href="%s">%s</a> }; |
my $nav_fmt=qq{ <a href="%s">%s</a> }; |
268 |
|
|
269 |
|
if ($pager->current_page() > $pager->first_page) { |
270 |
|
param('page', $pager->current_page - 1); |
271 |
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'<<'); |
272 |
|
} |
273 |
|
|
274 |
if ($pager->previous_set) { |
if ($pager->previous_set) { |
275 |
param('page', $pager->previous_set); |
param('page', $pager->previous_set); |
276 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'<<'); |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..'); |
277 |
} |
} |
278 |
|
|
279 |
|
|
280 |
foreach my $p (@{$pager->pages_in_set()}) { |
foreach my $p (@{$pager->pages_in_set()}) { |
281 |
|
next if ($p < 0); |
282 |
# for (my $p=$pager->previous_set; $p <= $pager->next_set; $p++) { |
# for (my $p=$pager->previous_set; $p <= $pager->next_set; $p++) { |
283 |
if($p == $pager->current_page()) { |
if($p == $pager->current_page()) { |
284 |
$nav_html .= "<b>$p</b> "; |
$nav_html .= "<b>$p</b> "; |
290 |
|
|
291 |
if ($pager->next_set) { |
if ($pager->next_set) { |
292 |
param('page', $pager->next_set); |
param('page', $pager->next_set); |
293 |
|
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..'); |
294 |
|
} |
295 |
|
|
296 |
|
if ($pager->current_page() < $pager->last_page) { |
297 |
|
param('page', $pager->current_page + 1); |
298 |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'>>'); |
$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'>>'); |
299 |
} |
} |
300 |
|
|
301 |
|
if ($config->{text}->{pages}) { |
302 |
|
$nav_html = x($config->{text}->{pages})." ".$nav_html; |
303 |
|
} |
304 |
|
|
305 |
# end html table |
# end html table |
306 |
print qq{ |
print qq{ |
307 |
<tr><td> |
<tr><td> |
308 |
Pages: $nav_html |
$nav_html |
309 |
</td></tr> |
</td></tr> |
310 |
</table> |
</table> |
311 |
}; |
}; |