9 |
use Text::Iconv; |
use Text::Iconv; |
10 |
use Data::Pageset; |
use Data::Pageset; |
11 |
|
|
|
|
|
12 |
sub get_snippet { |
sub get_snippet { |
13 |
my $context_chars = 100; |
|
14 |
my $max_desc = 16384; # 16k to filter |
# maximum length of context in characters |
15 |
|
my $cc = 50; |
16 |
|
|
17 |
my $desc = shift || return ''; |
my $desc = shift || return ''; |
18 |
$desc = substr($desc,0,$max_desc) if (length($desc) > $max_desc); |
$desc = e($desc); |
|
# test if $desc contains any of our query words |
|
|
my @snips; |
|
19 |
|
|
20 |
|
# sort words from longer to shorter (for hilighting later) |
21 |
|
my @words = sort { length($b) <=> length($a) } @_; |
22 |
|
|
23 |
|
# colors to highlite |
24 |
my @colors = qw{#ffff66 #a0ffff #99ff99 #ff9999 #ff66ff}; |
my @colors = qw{#ffff66 #a0ffff #99ff99 #ff9999 #ff66ff}; |
25 |
|
|
26 |
my $i = 0; |
# construct regex |
27 |
|
my $re = qq/^(.*?\\b)(/ . join('|', @words) . qq/)/; |
28 |
|
|
29 |
|
my $ellip = ' ... '; |
30 |
|
my $snippet = ''; |
31 |
|
|
32 |
|
#print "<ul>"; |
33 |
|
|
34 |
|
while ($desc =~ s/$re//si) { |
35 |
|
my ($foo, $match) = ($1,$2); |
36 |
|
|
37 |
|
#print "<br>desc: <small>$desc</small>\n"; |
38 |
|
#print "<br>foo: <small>$foo<b>$match</b></small>\n"; |
39 |
|
|
40 |
|
if (length($foo) < $cc * 2) { |
41 |
|
$snippet .= $foo . $match; |
42 |
|
} else { |
43 |
|
|
44 |
|
if ($foo =~ m/^(.{0,$cc})(\s.*?\s|\s|)?(.{0,$cc})$/) { |
45 |
|
|
46 |
|
# print "<li><small>$snippet</small><br> |
47 |
|
# ",length($1),": <i>$1</i><br> |
48 |
|
# ",length($2),": <span style=\"color:grey\">$2</span><br> |
49 |
|
# ",length($3),": <i>$3</i><br> |
50 |
|
# <b>$match</b>\n"; |
51 |
|
|
52 |
|
if ($snippet) { |
53 |
|
$snippet .= $1 . $ellip . $3 . $match; |
54 |
|
} else { |
55 |
|
$snippet = $ellip . $3 . $match ; |
56 |
|
} |
57 |
|
|
58 |
|
} else { |
59 |
|
# print "<li> <big>SKIP</big> $foo\n"; |
60 |
|
print STDERR "this shouldn't happen!\n"; |
61 |
|
} |
62 |
|
|
|
for my $q (@_) { |
|
|
if ($desc =~ m/(.*?)(\Q$q\E)(.*)/si) { |
|
|
my $bef = $1; |
|
|
my $qm = $2; |
|
|
my $af = $3; |
|
|
$bef = substr $bef, -$context_chars; |
|
|
$af = substr $af, 0, $context_chars; |
|
|
|
|
|
# no partial words... |
|
|
$af =~ s,^\S+\s+|\s+\S+$,,gs; |
|
|
$bef =~ s,^\S+\s+|\s+\S+$,,gs; |
|
|
|
|
|
push(@snips, "$bef <span style=\"background:".$colors[$i]."; color:black;\">$qm</span> $af"); |
|
|
$i++; |
|
|
$i = 0 if ($i > $#colors); |
|
63 |
} |
} |
64 |
|
|
65 |
} |
} |
66 |
my $ellip = ' ... '; |
#print "</ul>"; |
67 |
my $snippet = $ellip. join($ellip, @snips) . $ellip if (@snips); |
|
68 |
|
# color offset |
69 |
|
my $i = 0; |
70 |
|
|
71 |
|
foreach my $w (@words) { |
72 |
|
$snippet =~ s,(\b\Q$w\E),<span style="background: $colors[$i]; color:black;">$1</span>,gsi; |
73 |
|
$i++; |
74 |
|
$i = 0 if ($i > $#colors); |
75 |
|
} |
76 |
|
|
77 |
|
$snippet .= $ellip if ($snippet); |
78 |
|
|
79 |
return $snippet; |
return $snippet; |
80 |
} |
} |
81 |
|
|
141 |
my %path_label; |
my %path_label; |
142 |
my @path_name; |
my @path_name; |
143 |
foreach (@{$config->{paths}->{path}}) { |
foreach (@{$config->{paths}->{path}}) { |
|
|
|
|
print STDERR "##: $_->{limit}",x($_->{content}),"\n"; |
|
144 |
push @path_name,x($_->{limit}); |
push @path_name,x($_->{limit}); |
145 |
$path_label{$_->{limit}} = x($_->{content}); |
$path_label{$_->{limit}} = x($_->{content}); |
146 |
} |
} |
173 |
my $s; |
my $s; |
174 |
# re-write query from +/- to and/and not |
# re-write query from +/- to and/and not |
175 |
|
|
176 |
|
my @s_elem; |
177 |
|
|
178 |
my $search = param('search'); |
my $search = param('search'); |
179 |
my $s_phrase = ""; |
|
180 |
|
# strip spaces |
181 |
|
$search =~ s/^\s+//; |
182 |
|
$search =~ s/\s+$//; |
183 |
|
# fixup search string |
184 |
|
$search=~tr/¹ð¾èæ©Ð®ÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
185 |
|
$search=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
186 |
|
|
187 |
|
# extract phrases and put them first |
188 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
189 |
$s .= "$1 "; |
$s .= "$1 "; |
190 |
|
push @s_elem, $1; |
191 |
} |
} |
|
$search =~ s/^\s+//; |
|
|
$search =~ s/\s+$//; |
|
192 |
|
|
193 |
my %words; |
my %words; |
194 |
|
|
207 |
} else { |
} else { |
208 |
$s.="$2* "; |
$s.="$2* "; |
209 |
} |
} |
210 |
|
push @s_elem, $2; |
211 |
} else { |
} else { |
212 |
if (@spellings && !param('no_affix')) { |
if (@spellings && !param('no_affix')) { |
213 |
my $w = $_; $w =~ s/[\*\s]+//g; |
my $w = $_; $w =~ s/[\*\s]+//g; |
|
#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
|
214 |
my $or=""; |
my $or=""; |
215 |
foreach my $spelling_alt (@spellings) { |
foreach my $spelling_alt (@spellings) { |
216 |
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) "; |
219 |
} else { |
} else { |
220 |
$s.="$_* "; |
$s.="$_* "; |
221 |
} |
} |
222 |
|
push @s_elem, $_; |
223 |
} |
} |
224 |
} |
} |
225 |
|
|
226 |
# fixup search string |
# fix multiple stars |
|
$s=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
|
|
$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
|
227 |
$s=~s/\*\*+/*/g; |
$s=~s/\*\*+/*/g; |
228 |
|
|
229 |
# limit to some path |
# limit to some path |
234 |
# default format for output |
# default format for output |
235 |
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n"; |
236 |
|
|
|
# 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> |
|
|
}; |
|
|
|
|
237 |
if (@properties) { |
if (@properties) { |
238 |
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
$hit_fmt = x($config->{hit}) if (! param('no_properties')); |
239 |
$params{properties} = \@properties; |
$params{properties} = \@properties; |
270 |
} |
} |
271 |
|
|
272 |
my %path2title; |
my %path2title; |
|
use Data::Dumper; |
|
273 |
foreach my $p (@{$config->{path2title}->{path}}) { |
foreach my $p (@{$config->{path2title}->{path}}) { |
274 |
$path2title{$p->{dir}} = $p->{content}; |
$path2title{$p->{dir}} = $p->{content}; |
275 |
} |
} |
276 |
|
|
277 |
|
# output start of table |
278 |
|
print qq{ |
279 |
|
<table border="0"> |
280 |
|
}; |
281 |
|
# html before and after each hit |
282 |
|
my $tr_pre = qq{ |
283 |
|
<tr><td> |
284 |
|
}; |
285 |
|
my $tr_post = qq{ |
286 |
|
</td></tr> |
287 |
|
}; |
288 |
|
|
289 |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
for(my $i=$pager->first; $i<=$pager->last; $i++) { |
290 |
|
|
291 |
my $result = $results->NextResult; |
my $result = $results->NextResult; |
297 |
if ($prop =~ m/swishdescription/) { |
if ($prop =~ m/swishdescription/) { |
298 |
my $tmp = get_snippet( |
my $tmp = get_snippet( |
299 |
$result->Property($prop), |
$result->Property($prop), |
300 |
split(/\s+/,$search) |
@s_elem, |
301 |
); |
); |
302 |
|
|
303 |
push @arr, $tmp; |
push @arr, $tmp; |