7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
10 |
use Data::Dumper; |
use locale; |
11 |
|
use YAML; |
12 |
use Carp; |
use Carp; |
13 |
|
use Getopt::Long; |
14 |
|
use Text::Unaccent 1.02; |
15 |
|
|
16 |
use lib './lib'; |
use lib './lib'; |
17 |
use WebPAC; |
use WebPAC; |
18 |
use WebPAC::jsFind; |
use WebPAC::jsFind; |
19 |
use WebPAC::Index; |
use WebPAC::Index; |
20 |
|
use WebPAC::Tree; |
21 |
|
|
22 |
|
# options which can be changed via command line |
23 |
|
# |
24 |
|
my $code_page = 'ISO-8859-2'; |
25 |
|
my ($limit_mfn, $start_mfn, $debug, $low_mem); |
26 |
|
my $index_path = './out/index'; |
27 |
|
|
28 |
|
my $result = GetOptions( |
29 |
|
"code_page=s" => \$code_page, |
30 |
|
"limit_mfn=i" => \$limit_mfn, |
31 |
|
"start_mfn=i" => \$start_mfn, |
32 |
|
"debug!" => \$debug, |
33 |
|
"low_mem!" => \$low_mem, |
34 |
|
); |
35 |
|
|
36 |
|
my $filter = { |
37 |
|
'CROVOC' => sub { |
38 |
|
my $tmp = shift || return; |
39 |
|
return undef unless ($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; |
44 |
|
}, |
45 |
|
'CROVOC_tree' => sub { |
46 |
|
my $tmp = shift || return; |
47 |
|
$tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/; |
48 |
|
$tmp =~ s/\s*EUROVOC.*//; |
49 |
|
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 |
|
return $tmp; |
56 |
|
} |
57 |
|
}; |
58 |
|
|
59 |
|
## remove accented characters |
60 |
|
# |
61 |
|
sub unac { |
62 |
|
my $string = shift || return; |
63 |
|
$string = unac_string($code_page,$string); |
64 |
|
$string =~ tr/đĐ/dD/; |
65 |
|
$string = unac_entities($string); |
66 |
|
return $string; |
67 |
|
} |
68 |
|
sub unac_2 { |
69 |
|
my $string = shift || return; |
70 |
|
if (length($string) > 2) { |
71 |
|
my $pr = substr($string,0,2); |
72 |
|
$string = unac_string($code_page,substr($string,2)); |
73 |
|
$string =~ tr/đĐ/dD/; |
74 |
|
$string = $pr . $string; |
75 |
|
} |
76 |
|
$string = unac_entities($string); |
77 |
|
return lc($string); |
78 |
|
} |
79 |
|
|
80 |
|
sub unac_entities { |
81 |
|
my $ent = shift || return; |
82 |
|
|
83 |
|
$ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; |
84 |
|
$ent =~ s/ð/d/g; |
85 |
|
$ent =~ s/&E[tT][hH];/D/g; |
86 |
|
$ent =~ s/&(\w\w)lig;/$1/gi; |
87 |
|
|
88 |
|
return $ent; |
89 |
|
} |
90 |
|
|
91 |
|
# create WebPAC object |
92 |
|
# |
93 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
94 |
code_page => 'ISO-8859-2', |
code_page => $code_page, |
95 |
limit_mfn => 500, |
limit_mfn => $limit_mfn, |
96 |
# debug => 1, |
start_mfn => $start_mfn, |
97 |
# low_mem => 1, |
debug => $debug, |
98 |
|
low_mem => $low_mem, |
99 |
|
filter => $filter, |
100 |
) || die; |
) || die; |
101 |
|
|
102 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
my $log = $webpac->_get_logger() || die "can't get logger"; |
104 |
$log->debug("creating WebPAC::jsFind object"); |
$log->debug("creating WebPAC::jsFind object"); |
105 |
|
|
106 |
my $index = new WebPAC::jsFind( |
my $index = new WebPAC::jsFind( |
107 |
index_path => './out/index', |
index_path => $index_path, |
108 |
keys => 10, |
keys => 62, |
109 |
) || die; |
) || die; |
110 |
|
|
111 |
my $thes; |
my $thes; |
115 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
116 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
117 |
lookup => [ |
lookup => [ |
118 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' }, |
119 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
120 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
121 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
122 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
123 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
124 |
# tree structure |
# tree structure |
125 |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
126 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
127 |
|
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
128 |
], |
], |
129 |
); |
); |
130 |
|
|
141 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
142 |
|
|
143 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
144 |
$log->debug("rec = ",Dumper($rec)); |
$log->debug("rec = ",Dump($rec)); |
145 |
$log->debug("ds = ",Dumper(\@ds)); |
$log->debug("ds = ",Dump(\@ds)); |
146 |
} |
} |
147 |
|
|
148 |
next if (! @ds); |
next if (! @ds); |
173 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
174 |
next if (! $ds->{'swish'}); |
next if (! $ds->{'swish'}); |
175 |
|
|
176 |
|
# strip all non word characters from beginning or end |
177 |
|
# of word |
178 |
|
my $words = join(" ",@{$ds->{'swish'}}); |
179 |
|
$words =~ s/^\W+//; |
180 |
|
$words =~ s/\W*\s+\W*/ /g; |
181 |
|
$words =~ s/\W+$//; |
182 |
|
|
183 |
|
# first try to generate headline for this entry from index |
184 |
|
my $h = $ds->{'index'}->[0]; |
185 |
|
# then, from display |
186 |
|
$h ||= $ds->{'display'}->[0]; |
187 |
|
# and as last resport, fallback to headline |
188 |
|
$h ||= $headline; |
189 |
|
|
190 |
$index->insert( |
$index->insert( |
191 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
192 |
path => $f, |
#path => $f, |
193 |
headline => $headline, |
path => $webpac->mfn, |
194 |
words => join(" ",@{$ds->{'swish'}}) |
headline => $h, |
195 |
|
words => unac($words), |
196 |
); |
); |
197 |
} |
} |
198 |
|
|
200 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
201 |
next if (! $ds->{'index'}); |
next if (! $ds->{'index'}); |
202 |
|
|
203 |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); |
204 |
|
|
205 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
206 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
207 |
path => $f, |
sort_by => unac_2($h), |
208 |
|
mfn => $webpac->mfn, |
209 |
headline => $h, |
headline => $h, |
210 |
); |
); |
211 |
} |
} |
212 |
} |
} |
213 |
|
|
214 |
# print Dumper(\@ds); |
# print Dump(\@ds); |
215 |
|
|
216 |
} |
} |
217 |
|
|
235 |
} |
} |
236 |
|
|
237 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
238 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
239 |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
$log->debug("data hash: ",Dump($webpac->{'data'})); |
240 |
foreach my $t (keys %{$thes}) { |
foreach my $t (keys %{$thes}) { |
241 |
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
$log->debug("thesaurus $t hash: ",Dump($thes->{$t})); |
242 |
} |
} |
243 |
} |
} |
244 |
|
|
245 |
} # XXX if(0) |
} # XXX if(0) |
246 |
|
|
247 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
248 |
|
|
249 |
|
$log->info("creating tree"); |
250 |
|
|
251 |
# |
# |
252 |
# tree dump implementation which shouldn't be here :-) |
# define tree structure |
253 |
# |
# |
254 |
|
|
|
|
|
|
my $file = 'out/browse.html'; |
|
|
my $js_url = 'tree-ids.js'; |
|
|
|
|
|
$log->info("creating '$file' for tree html"); |
|
|
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
|
|
|
|
|
print HTML qq{ |
|
|
<html> |
|
|
<head> |
|
|
<title>Browse</title> |
|
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
|
|
<script type="text/javascript" src="bfilter/hide.js"></script> |
|
|
<script type="text/javascript" src="$js_url"></script> |
|
|
</head> |
|
|
|
|
|
<body onLoad="show_hide_display(show,hide);"> |
|
|
|
|
|
<div style="float: right; width: 10em;"> |
|
|
Folding: |
|
|
<a href="#" onClick="show_hide_display(show,hide);">default</a> |
|
|
<a href="#" onClick="show_display(show); show_display(hide);">all</a> |
|
|
<a href="#" onClick="hide_display(hide); hide_display(show);">none</a> |
|
|
<a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length);">debug</a> |
|
|
</div> |
|
|
|
|
|
|
|
|
<ul> |
|
|
}; |
|
|
|
|
255 |
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
256 |
|
|
257 |
my @tree = ({ |
my @tree = ({ |
264 |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
265 |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
266 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
267 |
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
268 |
child_code => sub { return $_[0] }, |
have_children_at_level => sub { |
269 |
|
return unless (defined($l->{"code:".$_[1]})); |
270 |
|
my $code = shift @{$l->{"code:".$_[1]}}; |
271 |
|
print STDERR "## $_[1] -> $code\n"; |
272 |
|
return undef unless($code); |
273 |
|
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
274 |
|
}, |
275 |
style => 'display: none', |
style => 'display: none', |
276 |
},{ |
},{ |
277 |
# 1 |
# 1 |
278 |
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
code_arr => sub { @{$_[0]} }, |
279 |
filter_code => sub { shift }, # nop |
filter_code => sub { shift }, # nop |
280 |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
281 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
282 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
283 |
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
have_children => sub { return $l->{"a:".$_[1].":"} }, |
|
child_code => sub { return $_[1] }, |
|
284 |
style => 'display: none', |
style => 'display: none', |
285 |
},{ |
},{ |
286 |
# 2 |
# 2 |
287 |
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
code_arr => sub { @{$_[0]} }, |
288 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
289 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
290 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
291 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
292 |
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
293 |
child_code => sub { return "a:".$_[2].":".$_[1] }, |
#style => 'display: none', |
|
style => 'display: none', |
|
294 |
},{ |
},{ |
295 |
# 3 uži pojam |
# 3 uži pojam |
296 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
297 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
298 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
299 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
300 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
301 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
302 |
},{ |
},{ |
303 |
# 4 |
# 4 |
304 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
305 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
306 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
307 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
308 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
309 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
310 |
},{ |
},{ |
311 |
# 5 |
# 5 |
312 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
313 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
314 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
315 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
316 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
317 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
318 |
},{ |
},{ |
319 |
# 6 |
# 6 |
320 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
321 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
322 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
323 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
324 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
325 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
326 |
},{ |
},{ |
327 |
# 7 |
# 7 |
328 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
329 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
330 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
331 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
332 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
333 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
334 |
},{ |
},{ |
335 |
# 8 |
# 8 |
336 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
337 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
338 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
339 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
340 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
341 |
# have_children => sub { defined($l->{$_[1]}) }, |
# have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
# child_code => sub { return $_[1] }, |
|
342 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
343 |
child_code => sub { 0 }, |
},{ |
344 |
|
# 9 - level which is never reached except explicitly |
345 |
|
code_arr => sub { @{$_[0]} }, |
346 |
|
filter_code => sub { shift }, |
347 |
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
348 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
349 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
350 |
|
have_children => sub { 0 }, |
351 |
|
have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) }, |
352 |
|
},{ |
353 |
}); |
}); |
354 |
|
|
355 |
my @show_ids; |
my $tree = new WebPAC::Tree( |
356 |
my @hide_ids; |
tree => \@tree, |
357 |
|
); |
|
unroll(0,''); |
|
|
|
|
|
$log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99")); |
|
|
|
|
|
sub unroll { |
|
|
my ($level,$start_code) = @_; |
|
|
|
|
|
$log->logconfess("need level") unless (defined($level)); |
|
|
|
|
|
# all levels passed? |
|
|
return if (! defined($tree[$level])); |
|
|
|
|
|
$log->debug("unroll level $level, start code $start_code"); |
|
|
|
|
|
foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) { |
|
|
|
|
|
if ($code = $tree[$level]->{'filter_code'}->($code)) { |
|
|
|
|
|
$log->debug("# $level filter passed code $code"); |
|
|
|
|
|
my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)"); |
|
|
$log->debug("# $level lookup_v900($code,$start_code) = $v900"); |
|
|
|
|
|
my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)"); |
|
|
$log->debug("# $level lookup_term($code,$v900) = $term"); |
|
|
|
|
|
my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)"); |
|
|
$log->debug("# $level lookup_mfn($code,$v900) = $mfn"); |
|
|
|
|
|
$log->debug("$code -> $v900 : $term [$mfn]"); |
|
|
|
|
|
my ($link_start,$link_end) = ('',''); |
|
|
|
|
|
my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code); |
|
|
if ($have_children) { |
|
|
($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>}); |
|
|
} else { |
|
|
$log->debug("# $level doesn't have_children($code,$v900,$start_code)"); |
|
|
} |
|
|
|
|
|
my $mfn_link = "thes/$mfn.html"; |
|
|
if (-e "out/$mfn_link") { |
|
|
print HTML " " x $level . |
|
|
qq{<li>${link_start}${term}${link_end}}. |
|
|
qq{ <a href="$mfn_link">»</a></li>\n}; |
|
|
} else { |
|
|
$log->warn("file 'out/$mfn_link' doesn't exist, skipping"); |
|
|
} |
|
|
|
|
|
unless ($have_children) { |
|
|
next; |
|
|
} |
|
|
my $style = $tree[$level]->{'style'}; |
|
|
|
|
|
print HTML " " x $level . |
|
|
qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}. |
|
|
($style ? ' style="$style"' : ""). |
|
|
qq{>\n}; |
|
|
|
|
|
if ($style) { |
|
|
if ($style =~ m/display\s*:\s*none/i) { |
|
|
push @hide_ids, "mfn$mfn"; |
|
|
} else { |
|
|
push @show_ids, "mfn$mfn"; |
|
|
} |
|
|
} else { |
|
|
# default: show |
|
|
push @show_ids, "mfn$mfn"; |
|
|
} |
|
|
|
|
|
unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code)); |
|
|
|
|
|
print HTML " " x $level . qq{</ul>\n}; |
|
|
|
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
|
print HTML qq{ |
|
|
</ul> |
|
|
</body> |
|
|
</html> |
|
|
}; |
|
358 |
|
|
359 |
close(HTML); |
$tree->output( |
360 |
|
dir => './out', |
361 |
|
html => 'browse.html', |
362 |
|
template => './output_template/tree.tt', |
363 |
|
js => 'tree-ids.js', |
364 |
|
); |
365 |
|
|
366 |
|
$tree->output( |
367 |
|
dir => './eurovoc', |
368 |
|
html => 'hijerarhija.html', |
369 |
|
template => './output_template/hijerarhija.tt', |
370 |
|
js => 'tree-ids.js', |
371 |
|
); |
372 |
|
|
|
my $js_file = "out/$js_url"; |
|
|
$log->info("creating '$js_file' with arrays of shown and hidden ids"); |
|
|
open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!"); |
|
|
print JS "var show = ['",join("','",@show_ids),"'];\n"; |
|
|
print JS "var hide = ['",join("','",@hide_ids),"'];\n"; |
|
|
close(JS); |
|
373 |
|
|
374 |
$log->info("closing index"); |
$log->info("closing index"); |
375 |
$index->close; |
$index->close; |