7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
10 |
|
use locale; |
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; |
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.*$/ #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 |
|
$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/đĐ/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/đĐ/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( |
97 |
start_mfn => $start_mfn, |
start_mfn => $start_mfn, |
98 |
debug => $debug, |
debug => $debug, |
99 |
low_mem => $low_mem, |
low_mem => $low_mem, |
100 |
|
filter => $filter, |
101 |
) || die; |
) || die; |
102 |
|
|
103 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
my $log = $webpac->_get_logger() || die "can't get logger"; |
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; |
116 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
117 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
118 |
lookup => [ |
lookup => [ |
119 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' }, |
120 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
121 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
122 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
123 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
124 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
125 |
# tree structure |
# tree structure |
126 |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
127 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
128 |
|
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
129 |
], |
], |
130 |
); |
); |
131 |
|
|
174 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
175 |
next if (! $ds->{'swish'}); |
next if (! $ds->{'swish'}); |
176 |
|
|
177 |
|
# strip all non word characters from beginning or end |
178 |
|
# of word |
179 |
|
my $words = join(" ",@{$ds->{'swish'}}); |
180 |
|
$words =~ s/^\W+//; |
181 |
|
$words =~ s/\W*\s+\W*/ /g; |
182 |
|
$words =~ s/\W+$//; |
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 |
headline => $headline, |
path => $webpac->mfn, |
195 |
words => join(" ",@{$ds->{'swish'}}) |
headline => $h, |
196 |
|
words => unac($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"); |
$log->info("creating tree"); |
251 |
|
|
267 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
268 |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
269 |
have_children_at_level => sub { |
have_children_at_level => sub { |
270 |
my $l2 = shift; |
return unless (defined($l->{"code:".$_[1]})); |
271 |
my $l1 = substr($l2,0,1); |
my $code = shift @{$l->{"code:".$_[1]}}; |
272 |
return(9, "a:$l1:$l2:" ) if ($l->{"a:$l1:$l2"}); |
print STDERR "## $_[1] -> $code\n"; |
273 |
|
return undef unless($code); |
274 |
|
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
275 |
}, |
}, |
276 |
style => 'display: none', |
style => 'display: none', |
277 |
},{ |
},{ |
287 |
# 2 |
# 2 |
288 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
289 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
290 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
291 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
292 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
293 |
have_children => sub { return $l->{"a:".$_[2].":".$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
294 |
#style => 'display: none', |
#style => 'display: none', |
295 |
},{ |
},{ |
296 |
# 3 uži pojam |
# 3 uži pojam |
297 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
298 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
299 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
300 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
301 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
302 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
303 |
},{ |
},{ |
304 |
# 4 |
# 4 |
305 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
306 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
307 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
308 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
309 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
310 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
311 |
},{ |
},{ |
312 |
# 5 |
# 5 |
313 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
314 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
315 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
316 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
317 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
318 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
319 |
},{ |
},{ |
320 |
# 6 |
# 6 |
321 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
322 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
323 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
324 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
325 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
326 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
327 |
},{ |
},{ |
328 |
# 7 |
# 7 |
329 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
330 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
331 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
332 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
333 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
334 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
335 |
},{ |
},{ |
336 |
# 8 |
# 8 |
337 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
338 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
339 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
340 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
341 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
342 |
# have_children => sub { return $l->{$_[1]} }, |
# have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
# child_code => sub { return $_[1] }, |
|
343 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
344 |
},{ |
},{ |
345 |
# 9 - level which is never reached except explicitly |
# 9 - level which is never reached except explicitly |
346 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
347 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
348 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(.+:.+):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
349 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
350 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
351 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
352 |
have_children_at_level => sub { defined($l->{$_[1]}) && return (9,$_[1]) }, |
have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) }, |
353 |
},{ |
},{ |
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 |
|
|