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 |
|
# maximum length of context in characters |
15 |
|
my $cc = 50; |
16 |
|
|
17 |
my $desc = shift || return ''; |
my $desc = shift || return ''; |
18 |
$desc = e($desc); |
$desc = e($desc); |
19 |
|
|
20 |
# test if $desc contains any of our query words |
# sort words from longer to shorter (for hilighting later) |
21 |
my @snips; |
my @words = sort { length($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 |
# construct regex |
# construct regex |
27 |
my $re = qq/^.*?(.{$context_chars}?)(\Q/ . join("|",@_) . |
my $re = qq/^(.*?\\b)(/ . join('|', @words) . qq/)/; |
28 |
qq/\E)(.{$context_chars})/; |
|
29 |
|
my $ellip = ' ... '; |
30 |
|
my $snippet = ''; |
31 |
|
|
32 |
|
#print "<ul>"; |
33 |
|
|
34 |
while ($desc =~ s/$re//si) { |
while ($desc =~ s/$re//si) { |
35 |
my ($bef,$qm,$af) = ($1, $2, $3); |
my ($foo, $match) = ($1,$2); |
|
|
|
|
# no partial words... |
|
|
$bef =~ s,^\S+\s+|\s+\S+$,,gs; |
|
|
$af =~ s,^\S+\s+|\s+\S+$,,gs; |
|
36 |
|
|
37 |
push @snips, "$bef $qm $af"; |
#print "<br>desc: <small>$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 |
|
|
63 |
|
} |
64 |
|
|
|
my $ellip = ' ... '; |
|
|
my $snippet; |
|
|
|
|
|
if (@snips) { |
|
|
$snippet = $ellip. join($ellip, @snips) . $ellip; |
|
|
} else { |
|
|
return ''; |
|
65 |
} |
} |
66 |
|
#print "</ul>"; |
67 |
|
|
68 |
# color offset |
# color offset |
69 |
my $i = 0; |
my $i = 0; |
70 |
|
|
71 |
foreach my $w (@_) { |
foreach my $w (@words) { |
72 |
$snippet =~ s,(\Q$w\E),<span style="background: $colors[$i]; color:black;">$1</span>,gsi; |
$snippet =~ s,(\b\Q$w\E),<span style="background: $colors[$i]; color:black;">$1</span>,gsi; |
73 |
$i++; |
$i++; |
74 |
$i = 0 if ($i > $#colors); |
$i = 0 if ($i > $#colors); |
75 |
} |
} |
76 |
|
|
77 |
|
$snippet .= $ellip if ($snippet); |
78 |
|
|
79 |
return $snippet; |
return $snippet; |
80 |
} |
} |
81 |
|
|
181 |
$search =~ s/^\s+//; |
$search =~ s/^\s+//; |
182 |
$search =~ s/\s+$//; |
$search =~ s/\s+$//; |
183 |
# fixup search string |
# fixup search string |
184 |
$search=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
$search=~tr/¹ð¾èæ©Ð®ÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2 |
185 |
$search=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
$search=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/; |
186 |
|
|
187 |
# extract phrases and put them first |
# extract phrases and put them first |