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; |
47 |
}, |
}, |
48 |
}; |
}; |
49 |
|
|
50 |
|
## remove accented characters |
51 |
|
# |
52 |
|
sub unac { |
53 |
|
my $string = shift || return; |
54 |
|
$string = unac_string($code_page,$string); |
55 |
|
$string =~ tr/ðÐ/dD/; |
56 |
|
return $string; |
57 |
|
} |
58 |
|
sub unac_2 { |
59 |
|
my $string = shift || return; |
60 |
|
if (length($string) > 2) { |
61 |
|
my $pr = substr($string,0,2); |
62 |
|
$string = unac_string($code_page,substr($string,2)); |
63 |
|
$string =~ tr/ðÐ/dD/; |
64 |
|
$string = $pr . $string; |
65 |
|
} |
66 |
|
$string =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; |
67 |
|
$string =~ s/ð/d/g; |
68 |
|
$string =~ s/&E[tT][hH];/D/g; |
69 |
|
$string =~ s/&(\w\w)lig;/$1/gi; |
70 |
|
return lc($string); |
71 |
|
} |
72 |
|
|
73 |
# create WebPAC object |
# create WebPAC object |
74 |
# |
# |
75 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
174 |
#path => $f, |
#path => $f, |
175 |
path => $webpac->mfn, |
path => $webpac->mfn, |
176 |
headline => $h, |
headline => $h, |
177 |
words => $words, |
words => unac($words), |
178 |
); |
); |
179 |
} |
} |
180 |
|
|
186 |
|
|
187 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
188 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
189 |
headline => $h, |
sort_by => unac_2($h), |
190 |
mfn => $webpac->mfn, |
mfn => $webpac->mfn, |
191 |
|
headline => $h, |
192 |
); |
); |
193 |
} |
} |
194 |
} |
} |
226 |
|
|
227 |
} # XXX if(0) |
} # XXX if(0) |
228 |
|
|
229 |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
230 |
|
|
231 |
$log->info("creating tree"); |
$log->info("creating tree"); |
232 |
|
|
335 |
}); |
}); |
336 |
|
|
337 |
my $tree = new WebPAC::Tree( |
my $tree = new WebPAC::Tree( |
338 |
dir => './out', |
tree => \@tree, |
339 |
html => 'browse.html', |
); |
340 |
|
|
341 |
|
$tree->output( |
342 |
|
dir => './out', |
343 |
|
html => 'browse.html', |
344 |
template => './output_template/tree.tt', |
template => './output_template/tree.tt', |
345 |
js => 'tree-ids.js', |
js => 'tree-ids.js', |
|
tree => \@tree, |
|
346 |
); |
); |
347 |
|
|
348 |
|
$tree->output( |
349 |
|
dir => './eurovoc', |
350 |
|
html => 'hijerarhija.html', |
351 |
|
template => './output_template/hijerarhija.tt', |
352 |
|
js => 'tree-ids.js', |
353 |
|
); |
354 |
|
|
355 |
|
|
356 |
$log->info("closing index"); |
$log->info("closing index"); |
357 |
$index->close; |
$index->close; |
358 |
|
|