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.*$/ */); |
return undef unless ($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 |
61 |
|
# |
62 |
|
sub unac { |
63 |
|
my $string = shift || return; |
64 |
|
$string = unac_string($code_page,$string); |
65 |
|
$string =~ tr/ðÐ/dD/; |
66 |
|
$string = unac_entities($string); |
67 |
|
return $string; |
68 |
|
} |
69 |
|
sub unac_2 { |
70 |
|
my $string = shift || return; |
71 |
|
if (length($string) > 2) { |
72 |
|
my $pr = substr($string,0,2); |
73 |
|
$string = unac_string($code_page,substr($string,2)); |
74 |
|
$string =~ tr/ðÐ/dD/; |
75 |
|
$string = $pr . $string; |
76 |
|
} |
77 |
|
$string = unac_entities($string); |
78 |
|
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; |
181 |
$words =~ s/\W*\s+\W*/ /g; |
$words =~ s/\W*\s+\W*/ /g; |
182 |
$words =~ s/\W+$//; |
$words =~ s/\W+$//; |
183 |
|
|
184 |
|
# first try to generate headline for this entry from index |
185 |
|
my $h = $ds->{'index'}->[0]; |
186 |
|
# then, from display |
187 |
|
$h ||= $ds->{'display'}->[0]; |
188 |
|
# and as last resport, fallback to headline |
189 |
|
$h ||= $headline; |
190 |
|
|
191 |
$index->insert( |
$index->insert( |
192 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
193 |
#path => $f, |
#path => $f, |
194 |
path => $webpac->mfn, |
path => $webpac->mfn, |
195 |
headline => $headline, |
headline => $h, |
196 |
words => $words, |
words => unac($words), |
197 |
); |
); |
198 |
} |
} |
199 |
|
|
201 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
202 |
next if (! $ds->{'index'}); |
next if (! $ds->{'index'}); |
203 |
|
|
204 |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); |
205 |
|
|
206 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
207 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
208 |
headline => $h, |
sort_by => unac_2($h), |
209 |
mfn => $webpac->mfn, |
mfn => $webpac->mfn, |
210 |
|
headline => $h, |
211 |
); |
); |
212 |
} |
} |
213 |
} |
} |
245 |
|
|
246 |
} # XXX if(0) |
} # XXX if(0) |
247 |
|
|
248 |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
249 |
|
|
250 |
$log->info("creating tree"); |
$log->info("creating tree"); |
251 |
|
|
354 |
}); |
}); |
355 |
|
|
356 |
my $tree = new WebPAC::Tree( |
my $tree = new WebPAC::Tree( |
357 |
dir => './out', |
tree => \@tree, |
358 |
html => 'browse.html', |
); |
359 |
|
|
360 |
|
$tree->output( |
361 |
|
dir => './out', |
362 |
|
html => 'browse.html', |
363 |
template => './output_template/tree.tt', |
template => './output_template/tree.tt', |
364 |
js => 'tree-ids.js', |
js => 'tree-ids.js', |
|
tree => \@tree, |
|
365 |
); |
); |
366 |
|
|
367 |
|
$tree->output( |
368 |
|
dir => './eurovoc', |
369 |
|
html => 'hijerarhija.html', |
370 |
|
template => './output_template/hijerarhija.tt', |
371 |
|
js => 'tree-ids.js', |
372 |
|
); |
373 |
|
|
374 |
|
|
375 |
$log->info("closing index"); |
$log->info("closing index"); |
376 |
$index->close; |
$index->close; |
377 |
|
|