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 |
## remove accented characters |
65 |
my $string = shift || return; |
my $string = shift || return; |
66 |
$string = unac_string($code_page,$string); |
$string = unac_string($code_page,$string); |
67 |
$string =~ tr/ðÐ/dD/; |
$string =~ tr/ðÐ/dD/; |
68 |
|
$string = unac_entities($string); |
69 |
return $string; |
return $string; |
70 |
} |
} |
71 |
sub unac_2 { |
sub unac_2 { |
76 |
$string =~ tr/ðÐ/dD/; |
$string =~ tr/ðÐ/dD/; |
77 |
$string = $pr . $string; |
$string = $pr . $string; |
78 |
} |
} |
79 |
|
$string = unac_entities($string); |
80 |
return lc($string); |
return lc($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 $thes; |
128 |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
129 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
130 |
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
131 |
|
{ 'key' => 'crovoc:v900', 'val' => 'filter{CROVOC}v800' }, |
132 |
], |
], |
133 |
); |
); |
134 |
|
|
179 |
|
|
180 |
# strip all non word characters from beginning or end |
# strip all non word characters from beginning or end |
181 |
# of word |
# of word |
182 |
my $words = join(" ",@{$ds->{'swish'}}); |
my $words = unac(join(" ",@{$ds->{'swish'}})); |
183 |
$words =~ s/^\W+//; |
$words =~ s/^\W+//; |
184 |
$words =~ s/\W*\s+\W*/ /g; |
$words =~ s/\W*\s+\W*/ /g; |
185 |
$words =~ s/\W+$//; |
$words =~ s/\W+$//; |
196 |
#path => $f, |
#path => $f, |
197 |
path => $webpac->mfn, |
path => $webpac->mfn, |
198 |
headline => $h, |
headline => $h, |
199 |
words => unac($words), |
words => $words, |
200 |
); |
); |
201 |
} |
} |
202 |
|
|
277 |
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
278 |
}, |
}, |
279 |
style => 'display: none', |
style => 'display: none', |
280 |
|
iframe => 1, |
281 |
},{ |
},{ |
282 |
# 1 |
# 1 |
283 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
295 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
296 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
297 |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
298 |
#style => 'display: none', |
style => 'display: none', |
299 |
},{ |
},{ |
300 |
# 3 u¾i pojam |
# 3 u¾i pojam |
301 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
359 |
|
|
360 |
my $tree = new WebPAC::Tree( |
my $tree = new WebPAC::Tree( |
361 |
tree => \@tree, |
tree => \@tree, |
362 |
|
detail_url => sub { |
363 |
|
my $mfn = shift; |
364 |
|
my $url = "thes/${mfn}.html"; |
365 |
|
return $url if (-e "./out/$url"); |
366 |
|
}, |
367 |
|
nodes_dir => 'nodes', |
368 |
); |
); |
369 |
|
|
370 |
$tree->output( |
$tree->output( |
371 |
dir => './out', |
dir => './out', |
372 |
html => 'browse.html', |
html => 'browse.html', |
373 |
template => './output_template/tree.tt', |
template_dir => './output_template/', |
374 |
|
template_tree => 'tree.tt', |
375 |
|
template_node => 'node.tt', |
376 |
js => 'tree-ids.js', |
js => 'tree-ids.js', |
377 |
); |
); |
378 |
|
|
379 |
$tree->output( |
$tree->output( |
380 |
dir => './eurovoc', |
dir => './eurovoc', |
381 |
html => 'hijerarhija.html', |
html => 'hijerarhija.html', |
382 |
template => './output_template/hijerarhija.tt', |
template_dir => './output_template/', |
383 |
|
template_tree => 'hijerarhija.tt', |
384 |
|
template_node => 'hijerarhija_node.tt', |
385 |
js => 'tree-ids.js', |
js => 'tree-ids.js', |
386 |
); |
); |
387 |
|
|