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 |
$tmp =~ s/\s*CROVOC.*$/ */; |
$tmp =~ s/\s*CROVOC.*$/ #C#/; |
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 |
|
return $tmp; |
56 |
|
} |
57 |
}; |
}; |
58 |
|
|
59 |
## remove accented characters |
## remove accented characters |
62 |
my $string = shift || return; |
my $string = shift || return; |
63 |
$string = unac_string($code_page,$string); |
$string = unac_string($code_page,$string); |
64 |
$string =~ tr/ðÐ/dD/; |
$string =~ tr/ðÐ/dD/; |
65 |
|
$string = unac_entities($string); |
66 |
return $string; |
return $string; |
67 |
} |
} |
68 |
sub unac_2 { |
sub unac_2 { |
73 |
$string =~ tr/ðÐ/dD/; |
$string =~ tr/ðÐ/dD/; |
74 |
$string = $pr . $string; |
$string = $pr . $string; |
75 |
} |
} |
76 |
|
$string = unac_entities($string); |
77 |
return lc($string); |
return lc($string); |
78 |
} |
} |
79 |
|
|
80 |
|
sub unac_entities { |
81 |
|
my $ent = shift || return; |
82 |
|
|
83 |
|
$ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; |
84 |
|
$ent =~ s/ð/d/g; |
85 |
|
$ent =~ s/&E[tT][hH];/D/g; |
86 |
|
$ent =~ s/&(\w\w)lig;/$1/gi; |
87 |
|
|
88 |
|
return $ent; |
89 |
|
} |
90 |
|
|
91 |
# create WebPAC object |
# create WebPAC object |
92 |
# |
# |
93 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
105 |
|
|
106 |
my $index = new WebPAC::jsFind( |
my $index = new WebPAC::jsFind( |
107 |
index_path => $index_path, |
index_path => $index_path, |
108 |
keys => 10, |
keys => 62, |
109 |
) || die; |
) || die; |
110 |
|
|
111 |
my $thes; |
my $thes; |