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; |
use Getopt::Long; |
14 |
|
use Text::Unaccent 1.02; |
15 |
|
|
16 |
use lib './lib'; |
use lib './lib'; |
17 |
use WebPAC; |
use WebPAC; |
33 |
"low_mem!" => \$low_mem, |
"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.*$/ */); |
40 |
|
return $tmp; |
41 |
|
}, |
42 |
|
'CROVOC_tree' => sub { |
43 |
|
my $tmp = shift || return; |
44 |
|
$tmp =~ s/\s*CROVOC.*$/ */; |
45 |
|
$tmp =~ s/\s*EUROVOC.*//; |
46 |
|
return $tmp; |
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/đĐ/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/đĐ/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( |
78 |
start_mfn => $start_mfn, |
start_mfn => $start_mfn, |
79 |
debug => $debug, |
debug => $debug, |
80 |
low_mem => $low_mem, |
low_mem => $low_mem, |
81 |
|
filter => $filter, |
82 |
) || die; |
) || die; |
83 |
|
|
84 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
my $log = $webpac->_get_logger() || die "can't get logger"; |
97 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
98 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
99 |
lookup => [ |
lookup => [ |
100 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' }, |
101 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
102 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
103 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
104 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
105 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
106 |
# tree structure |
# tree structure |
107 |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
108 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
109 |
|
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
110 |
], |
], |
111 |
); |
); |
112 |
|
|
123 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
124 |
|
|
125 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
126 |
$log->debug("rec = ",Dumper($rec)); |
$log->debug("rec = ",Dump($rec)); |
127 |
$log->debug("ds = ",Dumper(\@ds)); |
$log->debug("ds = ",Dump(\@ds)); |
128 |
} |
} |
129 |
|
|
130 |
next if (! @ds); |
next if (! @ds); |
155 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
156 |
next if (! $ds->{'swish'}); |
next if (! $ds->{'swish'}); |
157 |
|
|
158 |
|
# strip all non word characters from beginning or end |
159 |
|
# of word |
160 |
|
my $words = join(" ",@{$ds->{'swish'}}); |
161 |
|
$words =~ s/^\W+//; |
162 |
|
$words =~ s/\W*\s+\W*/ /g; |
163 |
|
$words =~ s/\W+$//; |
164 |
|
|
165 |
|
# first try to generate headline for this entry from index |
166 |
|
my $h = $ds->{'index'}->[0]; |
167 |
|
# then, from display |
168 |
|
$h ||= $ds->{'display'}->[0]; |
169 |
|
# and as last resport, fallback to headline |
170 |
|
$h ||= $headline; |
171 |
|
|
172 |
$index->insert( |
$index->insert( |
173 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
174 |
path => $f, |
#path => $f, |
175 |
headline => $headline, |
path => $webpac->mfn, |
176 |
words => join(" ",@{$ds->{'swish'}}) |
headline => $h, |
177 |
|
words => unac($words), |
178 |
); |
); |
179 |
} |
} |
180 |
|
|
182 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
183 |
next if (! $ds->{'index'}); |
next if (! $ds->{'index'}); |
184 |
|
|
185 |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); |
186 |
|
|
187 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
188 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
189 |
path => $f, |
sort_by => unac_2($h), |
190 |
|
mfn => $webpac->mfn, |
191 |
headline => $h, |
headline => $h, |
192 |
); |
); |
193 |
} |
} |
194 |
} |
} |
195 |
|
|
196 |
# print Dumper(\@ds); |
# print Dump(\@ds); |
197 |
|
|
198 |
} |
} |
199 |
|
|
217 |
} |
} |
218 |
|
|
219 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
220 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
221 |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
$log->debug("data hash: ",Dump($webpac->{'data'})); |
222 |
foreach my $t (keys %{$thes}) { |
foreach my $t (keys %{$thes}) { |
223 |
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
$log->debug("thesaurus $t hash: ",Dump($thes->{$t})); |
224 |
} |
} |
225 |
} |
} |
226 |
|
|
227 |
} # XXX if(0) |
} # XXX if(0) |
228 |
|
|
229 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
230 |
|
|
231 |
$log->info("creating tree"); |
$log->info("creating tree"); |
232 |
|
|
246 |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
247 |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
248 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
249 |
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
250 |
child_code => sub { return $_[0] }, |
have_children_at_level => sub { |
251 |
|
return unless (defined($l->{"code:".$_[1]})); |
252 |
|
my $code = shift @{$l->{"code:".$_[1]}}; |
253 |
|
print STDERR "## $_[1] -> $code\n"; |
254 |
|
return undef unless($code); |
255 |
|
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
256 |
|
}, |
257 |
style => 'display: none', |
style => 'display: none', |
258 |
},{ |
},{ |
259 |
# 1 |
# 1 |
260 |
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
code_arr => sub { @{$_[0]} }, |
261 |
filter_code => sub { shift }, # nop |
filter_code => sub { shift }, # nop |
262 |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
263 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
264 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
265 |
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
have_children => sub { return $l->{"a:".$_[1].":"} }, |
|
child_code => sub { return $_[1] }, |
|
266 |
style => 'display: none', |
style => 'display: none', |
267 |
},{ |
},{ |
268 |
# 2 |
# 2 |
269 |
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
code_arr => sub { @{$_[0]} }, |
270 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
271 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
272 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
273 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
274 |
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
275 |
child_code => sub { return "a:".$_[2].":".$_[1] }, |
#style => 'display: none', |
|
style => 'display: none', |
|
276 |
},{ |
},{ |
277 |
# 3 uži pojam |
# 3 uži pojam |
278 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
279 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
280 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
281 |
lookup_term => sub { shift @{$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->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
284 |
},{ |
},{ |
285 |
# 4 |
# 4 |
286 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
287 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
288 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
289 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
290 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
291 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
292 |
},{ |
},{ |
293 |
# 5 |
# 5 |
294 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
295 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
296 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
297 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
298 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
299 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
300 |
},{ |
},{ |
301 |
# 6 |
# 6 |
302 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
303 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
304 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
305 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
306 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
307 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
308 |
},{ |
},{ |
309 |
# 7 |
# 7 |
310 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
311 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
312 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
313 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
314 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
315 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
316 |
},{ |
},{ |
317 |
# 8 |
# 8 |
318 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
319 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
320 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
321 |
my ($c,$p) = @_; |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
322 |
$p =~ s/^a:(..:....):.*$/$1/; |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
323 |
return "a:".$p.":".$c; |
# have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
324 |
}, |
have_children => sub { 0 }, |
325 |
|
},{ |
326 |
|
# 9 - level which is never reached except explicitly |
327 |
|
code_arr => sub { @{$_[0]} }, |
328 |
|
filter_code => sub { shift }, |
329 |
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
330 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
331 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
|
# have_children => sub { defined($l->{$_[1]}) }, |
|
|
# child_code => sub { return $_[1] }, |
|
332 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
333 |
child_code => sub { 0 }, |
have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) }, |
334 |
|
},{ |
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 |
|
|