36 |
my $filter = { |
my $filter = { |
37 |
'CROVOC' => sub { |
'CROVOC' => sub { |
38 |
my $tmp = shift || return; |
my $tmp = shift || return; |
39 |
return undef unless ($tmp =~ s/\s*CROVOC.*$/ */); |
return undef unless ($tmp =~ s/\s*CROVOC.*$/ #C#/); |
40 |
|
# remove repeating stars |
41 |
|
# FIXME this should be fixed at right place, not here! |
42 |
|
$tmp =~ s/(\s*#C#)+/ #C#/g; |
43 |
return $tmp; |
return $tmp; |
44 |
}, |
}, |
45 |
'CROVOC_tree' => sub { |
'CROVOC_tree' => sub { |
46 |
my $tmp = shift || return; |
my $tmp = shift || return; |
47 |
$tmp =~ s/\s*CROVOC.*$/ */; |
$tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/; |
48 |
$tmp =~ s/\s*EUROVOC.*//; |
$tmp =~ s/\s*EUROVOC.*//; |
49 |
return $tmp; |
return $tmp; |
50 |
}, |
}, |
51 |
|
# TT filter |
52 |
|
'CROVOC_img' => sub { |
53 |
|
my $tmp = shift; |
54 |
|
$tmp =~ s/\s*#C#\s*/ <img src="..\/img\/crovoc.png" border="0">/gis; |
55 |
|
$tmp =~ s/"img\/crovoc.png"/"..\/img\/crovoc.png"/gis; |
56 |
|
return $tmp; |
57 |
|
} |
58 |
}; |
}; |
59 |
|
|
60 |
## remove accented characters |
## remove accented characters |
63 |
my $string = shift || return; |
my $string = shift || return; |
64 |
$string = unac_string($code_page,$string); |
$string = unac_string($code_page,$string); |
65 |
$string =~ tr/ðÐ/dD/; |
$string =~ tr/ðÐ/dD/; |
66 |
|
$string = unac_entities($string); |
67 |
return $string; |
return $string; |
68 |
} |
} |
69 |
sub unac_2 { |
sub unac_2 { |
74 |
$string =~ tr/ðÐ/dD/; |
$string =~ tr/ðÐ/dD/; |
75 |
$string = $pr . $string; |
$string = $pr . $string; |
76 |
} |
} |
77 |
|
$string = unac_entities($string); |
78 |
return lc($string); |
return lc($string); |
79 |
} |
} |
80 |
|
|
81 |
|
sub unac_entities { |
82 |
|
my $ent = shift || return; |
83 |
|
|
84 |
|
$ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; |
85 |
|
$ent =~ s/ð/d/g; |
86 |
|
$ent =~ s/&E[tT][hH];/D/g; |
87 |
|
$ent =~ s/&(\w\w)lig;/$1/gi; |
88 |
|
|
89 |
|
return $ent; |
90 |
|
} |
91 |
|
|
92 |
# create WebPAC object |
# create WebPAC object |
93 |
# |
# |
94 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
106 |
|
|
107 |
my $index = new WebPAC::jsFind( |
my $index = new WebPAC::jsFind( |
108 |
index_path => $index_path, |
index_path => $index_path, |
109 |
keys => 10, |
keys => 62, |
110 |
) || die; |
) || die; |
111 |
|
|
112 |
my $thes; |
my $thes; |