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/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 { |
47 |
|
my $tmp = shift || return; |
48 |
|
$tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/; |
49 |
|
$tmp =~ s/\s*EUROVOC.*//; |
50 |
|
$tmp =~ s/\s*HIDRA.*//; |
51 |
|
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/đĐ/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/đĐ/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; |
118 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
119 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
120 |
lookup => [ |
lookup => [ |
121 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' }, |
122 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
123 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
124 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
183 |
$words =~ s/\W*\s+\W*/ /g; |
$words =~ s/\W*\s+\W*/ /g; |
184 |
$words =~ s/\W+$//; |
$words =~ s/\W+$//; |
185 |
|
|
186 |
|
# first try to generate headline for this entry from index |
187 |
|
my $h = $ds->{'index'}->[0]; |
188 |
|
# then, from display |
189 |
|
$h ||= $ds->{'display'}->[0]; |
190 |
|
# and as last resport, fallback to headline |
191 |
|
$h ||= $headline; |
192 |
|
|
193 |
$index->insert( |
$index->insert( |
194 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
195 |
#path => $f, |
#path => $f, |
196 |
path => $webpac->mfn, |
path => $webpac->mfn, |
197 |
headline => $headline, |
headline => $h, |
198 |
words => $words, |
words => unac($words), |
199 |
); |
); |
200 |
} |
} |
201 |
|
|
203 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
204 |
next if (! $ds->{'index'}); |
next if (! $ds->{'index'}); |
205 |
|
|
206 |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); |
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 |
} |
} |
247 |
|
|
248 |
} # XXX if(0) |
} # XXX if(0) |
249 |
|
|
250 |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
251 |
|
|
252 |
$log->info("creating tree"); |
$log->info("creating tree"); |
253 |
|
|
285 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
286 |
have_children => sub { return $l->{"a:".$_[1].":"} }, |
have_children => sub { return $l->{"a:".$_[1].":"} }, |
287 |
style => 'display: none', |
style => 'display: none', |
288 |
|
iframe => 1, |
289 |
},{ |
},{ |
290 |
# 2 |
# 2 |
291 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
294 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
295 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
296 |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
297 |
#style => 'display: none', |
style => 'display: none', |
298 |
},{ |
},{ |
299 |
# 3 uži pojam |
# 3 uži pojam |
300 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
357 |
}); |
}); |
358 |
|
|
359 |
my $tree = new WebPAC::Tree( |
my $tree = new WebPAC::Tree( |
|
dir => './out', |
|
|
html => 'browse.html', |
|
|
template => './output_template/tree.tt', |
|
|
js => 'tree-ids.js', |
|
360 |
tree => \@tree, |
tree => \@tree, |
361 |
|
detail_url => sub { |
362 |
|
my $mfn = shift; |
363 |
|
my $url = "thes/${mfn}.html"; |
364 |
|
return $url if (-e "./out/$url"); |
365 |
|
}, |
366 |
|
nodes_dir => 'nodes', |
367 |
); |
); |
368 |
|
|
369 |
|
$tree->output( |
370 |
|
dir => './out', |
371 |
|
html => 'browse.html', |
372 |
|
template_dir => './output_template/', |
373 |
|
template_tree => 'tree.tt', |
374 |
|
template_node => 'node.tt', |
375 |
|
js => 'tree-ids.js', |
376 |
|
); |
377 |
|
|
378 |
|
$tree->output( |
379 |
|
dir => './eurovoc', |
380 |
|
html => 'hijerarhija.html', |
381 |
|
template_dir => './output_template/', |
382 |
|
template_tree => 'hijerarhija.tt', |
383 |
|
template_node => 'hijerarhija_node.tt', |
384 |
|
js => 'tree-ids.js', |
385 |
|
); |
386 |
|
|
387 |
|
|
388 |
$log->info("closing index"); |
$log->info("closing index"); |
389 |
$index->close; |
$index->close; |
390 |
|
|