11 |
use YAML; |
use YAML; |
12 |
use Carp; |
use Carp; |
13 |
use Getopt::Long; |
use Getopt::Long; |
14 |
|
use Text::Unaccent 1.02; |
15 |
|
|
16 |
use lib './lib'; |
use lib './lib'; |
17 |
use WebPAC; |
use WebPAC; |
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.*$/ */); |
$tmp =~ s/\s*HIDRA.*//; |
40 |
|
return undef unless ($tmp =~ s/\s*CROVOC.*$/ #C#/); |
41 |
|
# remove repeating stars |
42 |
|
# FIXME this should be fixed at right place, not here! |
43 |
|
$tmp =~ s/(\s*#C#)+/ #C#/g; |
44 |
return $tmp; |
return $tmp; |
45 |
}, |
}, |
46 |
'CROVOC_tree' => sub { |
'CROVOC_tree' => sub { |
47 |
my $tmp = shift || return; |
my $tmp = shift || return; |
48 |
$tmp =~ s/\s*CROVOC.*$/ */; |
$tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/; |
49 |
$tmp =~ s/\s*EUROVOC.*//; |
$tmp =~ s/\s*EUROVOC.*//; |
50 |
|
$tmp =~ s/\s*HIDRA.*//; |
51 |
return $tmp; |
return $tmp; |
52 |
}, |
}, |
53 |
|
# TT filter |
54 |
|
'CROVOC_img' => sub { |
55 |
|
my $tmp = shift; |
56 |
|
$tmp =~ s/\s*#C#\s*/ <img src="..\/img\/crovoc.png" border="0">/gis; |
57 |
|
$tmp =~ s/"img\/crovoc.png"/"..\/img\/crovoc.png"/gis; |
58 |
|
return $tmp; |
59 |
|
} |
60 |
}; |
}; |
61 |
|
|
62 |
|
## remove accented characters |
63 |
|
# |
64 |
|
sub unac { |
65 |
|
my $string = shift || return; |
66 |
|
$string = unac_string($code_page,$string); |
67 |
|
$string =~ tr/ðÐ/dD/; |
68 |
|
$string = unac_entities($string); |
69 |
|
return $string; |
70 |
|
} |
71 |
|
sub unac_2 { |
72 |
|
my $string = shift || return; |
73 |
|
if (length($string) > 2) { |
74 |
|
my $pr = substr($string,0,2); |
75 |
|
$string = unac_string($code_page,substr($string,2)); |
76 |
|
$string =~ tr/ðÐ/dD/; |
77 |
|
$string = $pr . $string; |
78 |
|
} |
79 |
|
$string = unac_entities($string); |
80 |
|
return lc($string); |
81 |
|
} |
82 |
|
|
83 |
|
sub unac_entities { |
84 |
|
my $ent = shift || return; |
85 |
|
|
86 |
|
$ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; |
87 |
|
$ent =~ s/ð/d/g; |
88 |
|
$ent =~ s/&E[tT][hH];/D/g; |
89 |
|
$ent =~ s/&(\w\w)lig;/$1/gi; |
90 |
|
|
91 |
|
return $ent; |
92 |
|
} |
93 |
|
|
94 |
# create WebPAC object |
# create WebPAC object |
95 |
# |
# |
96 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
108 |
|
|
109 |
my $index = new WebPAC::jsFind( |
my $index = new WebPAC::jsFind( |
110 |
index_path => $index_path, |
index_path => $index_path, |
111 |
keys => 10, |
keys => 62, |
112 |
) || die; |
) || die; |
113 |
|
|
114 |
my $thes; |
my $thes; |
195 |
#path => $f, |
#path => $f, |
196 |
path => $webpac->mfn, |
path => $webpac->mfn, |
197 |
headline => $h, |
headline => $h, |
198 |
words => $words, |
words => unac($words), |
199 |
); |
); |
200 |
} |
} |
201 |
|
|
207 |
|
|
208 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
209 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
210 |
headline => $h, |
sort_by => unac_2($h), |
211 |
mfn => $webpac->mfn, |
mfn => $webpac->mfn, |
212 |
|
headline => $h, |
213 |
); |
); |
214 |
} |
} |
215 |
} |
} |