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; |
32 |
"low_mem!" => \$low_mem, |
"low_mem!" => \$low_mem, |
33 |
); |
); |
34 |
|
|
35 |
|
my $filter = { |
36 |
|
'CROVOC' => sub { |
37 |
|
my $tmp = shift || return; |
38 |
|
return undef unless ($tmp =~ s/\s*CROVOC.*$/ */); |
39 |
|
return $tmp; |
40 |
|
}, |
41 |
|
'CROVOC_tree' => sub { |
42 |
|
my $tmp = shift || return; |
43 |
|
$tmp =~ s/\s*CROVOC.*$/ */; |
44 |
|
$tmp =~ s/\s*EUROVOC.*//; |
45 |
|
return $tmp; |
46 |
|
}, |
47 |
|
}; |
48 |
|
|
49 |
# create WebPAC object |
# create WebPAC object |
50 |
# |
# |
51 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
54 |
start_mfn => $start_mfn, |
start_mfn => $start_mfn, |
55 |
debug => $debug, |
debug => $debug, |
56 |
low_mem => $low_mem, |
low_mem => $low_mem, |
57 |
|
filter => $filter, |
58 |
) || die; |
) || die; |
59 |
|
|
60 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
my $log = $webpac->_get_logger() || die "can't get logger"; |
73 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
74 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
75 |
lookup => [ |
lookup => [ |
76 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' }, |
77 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
78 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
79 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
80 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
81 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
82 |
# tree structure |
# tree structure |
83 |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
84 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
85 |
|
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
86 |
], |
], |
87 |
); |
); |
88 |
|
|
131 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
132 |
next if (! $ds->{'swish'}); |
next if (! $ds->{'swish'}); |
133 |
|
|
134 |
|
# strip all non word characters from beginning or end |
135 |
|
# of word |
136 |
|
my $words = join(" ",@{$ds->{'swish'}}); |
137 |
|
$words =~ s/^\W+//; |
138 |
|
$words =~ s/\W*\s+\W*/ /g; |
139 |
|
$words =~ s/\W+$//; |
140 |
|
|
141 |
|
# first try to generate headline for this entry from index |
142 |
|
my $h = shift @{$ds->{'index'}}; |
143 |
|
# then, from display |
144 |
|
$h ||= shift @{$ds->{'display'}}; |
145 |
|
# and as last resport, fallback to headline |
146 |
|
$h ||= $headline; |
147 |
|
|
148 |
$index->insert( |
$index->insert( |
149 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
150 |
path => $f, |
#path => $f, |
151 |
headline => $headline, |
path => $webpac->mfn, |
152 |
words => join(" ",@{$ds->{'swish'}}) |
headline => $h, |
153 |
|
words => $words, |
154 |
); |
); |
155 |
} |
} |
156 |
|
|
158 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
159 |
next if (! $ds->{'index'}); |
next if (! $ds->{'index'}); |
160 |
|
|
161 |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); |
162 |
|
|
163 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
164 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
223 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
224 |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
225 |
have_children_at_level => sub { |
have_children_at_level => sub { |
226 |
my $l2 = shift; |
return unless (defined($l->{"code:".$_[1]})); |
227 |
my $l1 = substr($l2,0,1); |
my $code = shift @{$l->{"code:".$_[1]}}; |
228 |
return(9, "a:$l1:$l2:" ) if ($l->{"a:$l1:$l2"}); |
print STDERR "## $_[1] -> $code\n"; |
229 |
|
return undef unless($code); |
230 |
|
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
231 |
}, |
}, |
232 |
style => 'display: none', |
style => 'display: none', |
233 |
},{ |
},{ |
243 |
# 2 |
# 2 |
244 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
245 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
246 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
247 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
248 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
249 |
have_children => sub { return $l->{"a:".$_[2].":".$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
250 |
#style => 'display: none', |
#style => 'display: none', |
251 |
},{ |
},{ |
252 |
# 3 uži pojam |
# 3 uži pojam |
253 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
254 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
255 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
256 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
257 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
258 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
259 |
},{ |
},{ |
260 |
# 4 |
# 4 |
261 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
262 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
263 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
264 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
265 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
266 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
267 |
},{ |
},{ |
268 |
# 5 |
# 5 |
269 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
270 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
271 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
272 |
lookup_term => sub { shift @{$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 { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
275 |
},{ |
},{ |
276 |
# 6 |
# 6 |
277 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
278 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
279 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
280 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
281 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
282 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
283 |
},{ |
},{ |
284 |
# 7 |
# 7 |
285 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
286 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
287 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
288 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
289 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
290 |
have_children => sub { return $l->{$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
291 |
},{ |
},{ |
292 |
# 8 |
# 8 |
293 |
code_arr => sub { @{$_[0]} }, |
code_arr => sub { @{$_[0]} }, |
294 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
295 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
296 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
297 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
298 |
# have_children => sub { return $l->{$_[1]} }, |
# have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
# child_code => sub { return $_[1] }, |
|
299 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
300 |
},{ |
},{ |
301 |
# 9 - level which is never reached except explicitly |
# 9 - level which is never reached except explicitly |
302 |
code_arr => sub { @{$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:".$_[0]}} }, |
307 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
308 |
have_children_at_level => sub { defined($l->{$_[1]}) && return (9,$_[1]) }, |
have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) }, |
309 |
},{ |
},{ |
310 |
}); |
}); |
311 |
|
|