2 |
|
|
3 |
=head1 NAME |
=head1 NAME |
4 |
|
|
5 |
webpac.pl - basic script for all WebPac needs |
all2all.pl - basic script for all WebPAC needs |
6 |
|
|
7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
|
use OpenIsis; |
|
|
use Text::Iconv; |
|
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
|
use Carp; |
12 |
|
|
13 |
my $cp = Text::Iconv->new("852", "iso-8859-2"); |
use lib './lib'; |
14 |
|
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
use WebPAC::Index; |
17 |
|
|
18 |
|
my $webpac = new WebPAC( |
19 |
|
code_page => 'ISO-8859-2', |
20 |
|
limit_mfn => 500, |
21 |
|
start_mfn => 6000, |
22 |
|
# debug => 1, |
23 |
|
# low_mem => 1, |
24 |
|
) || die; |
25 |
|
|
26 |
|
my $log = $webpac->_get_logger() || die "can't get logger"; |
27 |
|
|
28 |
|
$log->debug("creating WebPAC::jsFind object"); |
29 |
|
|
30 |
|
my $index = new WebPAC::jsFind( |
31 |
|
index_path => './out/index', |
32 |
|
keys => 10, |
33 |
|
) || die; |
34 |
|
|
35 |
|
my $thes; |
36 |
|
|
37 |
$|=1; |
$|=1; |
38 |
|
|
39 |
my $db = OpenIsis::open( shift @ARGV || '/data/hidra/THS-500/THS' ); |
my $maxmfn = $webpac->open_isis( |
40 |
my $maxmfn = OpenIsis::maxRowid( $db ) || 1; |
filename => shift @ARGV || '/data/hidra/THS/THS', |
41 |
|
lookup => [ |
42 |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
43 |
|
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
44 |
|
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
45 |
|
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
46 |
|
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
47 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
48 |
|
# tree structure |
49 |
|
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
50 |
|
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
51 |
|
], |
52 |
|
); |
53 |
|
|
54 |
print "rows: $maxmfn\n\n"; |
$log->debug("isis file ",$webpac->{'isis_filename'}," opened"); |
55 |
|
|
56 |
my $data; |
$log->info("rows: $maxmfn"); |
57 |
|
|
58 |
my @lookup = ( |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
|
|
{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
|
|
{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'ma:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
|
|
{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'da:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
|
|
); |
|
59 |
|
|
60 |
sub fill_in { |
if(1) { # XXX |
|
my $tmp = shift || return; |
|
|
my $data = shift || return; |
|
|
my $i = shift || 0; |
|
|
|
|
|
my $found = 0; |
|
|
|
|
|
sub get_sf { |
|
|
my ($found,$data,$f,$sf,$i) = @_; |
|
|
if ($$data->{$f} && $$data->{$f}->[$i]->{$sf}) { |
|
|
$$found++; |
|
|
return $$data->{$f}->[$i]->{$sf}; |
|
|
} else { |
|
|
return ''; |
|
|
} |
|
|
} |
|
61 |
|
|
62 |
sub get_nosf { |
while (my $rec = $webpac->fetch_rec) { |
63 |
my ($found,$data,$f,$i) = @_; |
|
64 |
if ($$data->{$f} && $$data->{$f}->[$i]) { |
my @ds = $webpac->data_structure($rec); |
65 |
$$found++; |
|
66 |
return $$data->{$f}->[$i]; |
if (0 && $log->is_debug) { |
67 |
} else { |
$log->debug("rec = ",Dumper($rec)); |
68 |
return ''; |
$log->debug("ds = ",Dumper(\@ds)); |
|
} |
|
69 |
} |
} |
70 |
|
|
71 |
$tmp =~ s/v(\d+)\^(\w)/get_sf(\$found,\$data,$1,$2,$i)/ges; |
next if (! @ds); |
|
$tmp =~ s/v(\d+)/get_nosf(\$found,\$data,$1,$i)/ges; |
|
72 |
|
|
73 |
if ($found) { |
my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object"); |
74 |
return $tmp; |
|
75 |
|
if ($filename) { |
76 |
|
$webpac->output_file( |
77 |
|
file => $filename, |
78 |
|
template => 'html.tt', |
79 |
|
data => \@ds, |
80 |
|
headline => $webpac->{'headline'}, |
81 |
|
); |
82 |
} else { |
} else { |
83 |
return; |
print $webpac->output( |
84 |
|
template => 'text.tt', |
85 |
|
data => \@ds, |
86 |
|
headline => $webpac->{'headline'}, |
87 |
|
); |
88 |
} |
} |
|
} |
|
89 |
|
|
90 |
print "reading database...\n"; |
my $headline = $webpac->{'headline'}; |
91 |
|
|
92 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
my $f = $filename; |
93 |
|
$f =~ s!out/!!; |
94 |
|
|
95 |
my $row = OpenIsis::read( $db, $mfn ); |
# save into index |
96 |
foreach my $k (keys %{$row}) { |
foreach my $ds (@ds) { |
97 |
if ($k ne "mfn") { |
next if (! $ds->{'swish'}); |
98 |
foreach my $l (@{$row->{$k}}) { |
|
99 |
$l = $cp->convert($l); |
$index->insert( |
100 |
# has subfields? |
index_name => $ds->{'tag'}, |
101 |
my $val; |
path => $f, |
102 |
if ($l =~ m/\^/) { |
headline => $headline, |
103 |
foreach my $t (split(/\^/,$l)) { |
words => join(" ",@{$ds->{'swish'}}) |
104 |
next if (! $t); |
); |
105 |
$val->{substr($t,0,1)} = substr($t,1); |
} |
|
} |
|
|
} else { |
|
|
$val = $l; |
|
|
} |
|
106 |
|
|
107 |
push @{$data->{$mfn}->{$k}}, $val; |
# save into sorted index (thesaurus) |
108 |
} |
foreach my $ds (@ds) { |
109 |
|
next if (! $ds->{'index'}); |
110 |
|
|
111 |
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
112 |
|
|
113 |
|
foreach my $h (@{$ds->{'index'}}) { |
114 |
|
$thes->{$ds->{'tag'}}->insert( |
115 |
|
path => $f, |
116 |
|
headline => $h, |
117 |
|
); |
118 |
} |
} |
119 |
|
} |
120 |
|
|
121 |
|
# print Dumper(\@ds); |
122 |
|
|
123 |
|
} |
124 |
|
|
125 |
|
foreach my $t (keys %{$thes}) { |
126 |
|
|
127 |
|
my @e = $thes->{$t}->elements; |
128 |
|
if (! @e) { |
129 |
|
$log->logwarn("no elements in sorted index $t?"); |
130 |
|
next; |
131 |
|
} |
132 |
|
|
133 |
|
my $file = "./out/bfilter/$t.txt"; |
134 |
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
135 |
|
|
136 |
|
$webpac->output_file( |
137 |
|
file => $file, |
138 |
|
template => 'index.tt', |
139 |
|
data => \@e, |
140 |
|
index_name => $t, |
141 |
|
); |
142 |
|
} |
143 |
|
|
144 |
|
if (0 && $log->is_debug) { |
145 |
|
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
146 |
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
147 |
|
foreach my $t (keys %{$thes}) { |
148 |
|
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
149 |
} |
} |
150 |
|
} |
151 |
|
|
152 |
|
} # XXX if(0) |
153 |
|
|
154 |
|
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
155 |
|
|
156 |
|
# |
157 |
|
# tree dump implementation which shouldn't be here :-) |
158 |
|
# |
159 |
|
|
160 |
|
|
161 |
|
my $file = 'out/browse.html'; |
162 |
|
my $js_url = 'tree-ids.js'; |
163 |
|
|
164 |
|
$log->info("creating '$file' for tree html"); |
165 |
|
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
166 |
|
|
167 |
|
print HTML qq{ |
168 |
|
<html> |
169 |
|
<head> |
170 |
|
<title>Browse</title> |
171 |
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
172 |
|
<script type="text/javascript" src="bfilter/hide.js"></script> |
173 |
|
<script type="text/javascript" src="$js_url"></script> |
174 |
|
</head> |
175 |
|
|
176 |
|
<!-- |
177 |
|
<body onLoad="show_hide_display(show,hide);"> |
178 |
|
--> |
179 |
|
<body> |
180 |
|
|
181 |
|
<div style="float: right; width: 10em; text-align: center; margin: 0.5m; background: #e0e0e0; border: 1px dashed #c0c0c0; z-index: 1;"> |
182 |
|
Folding: |
183 |
|
<br/> |
184 |
|
<a href="#" onClick="back_display();">«</a> <a href="#" onClick="apply_changed_display();">default</a> <a href="#" onClick="forward_display();">»</a> |
185 |
|
<br/> |
186 |
|
<a href="#" onClick="show_hide_display(show,hide);">reset</a> |
187 |
|
<a href="#" onClick="show_display(show); show_display(hide);">all</a> |
188 |
|
<a href="#" onClick="hide_display(hide); hide_display(show);">none</a> |
189 |
|
<br/> |
190 |
|
<a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length+', changed:'+changed_display_ids.length+', positin: '+changed_display_pos);">debug</a> |
191 |
|
</div> |
192 |
|
|
193 |
|
<ul> |
194 |
|
}; |
195 |
|
|
196 |
|
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
197 |
|
|
198 |
|
my @tree = ({ |
199 |
|
# level 0 |
200 |
|
code_arr => sub { sort keys %{$l} }, |
201 |
|
filter_code => sub { |
202 |
|
my $t = shift; |
203 |
|
return $t if ($t =~ s/root://); |
204 |
|
}, |
205 |
|
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
206 |
|
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
207 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
208 |
|
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
209 |
|
child_code => sub { return $_[0] }, |
210 |
|
style => 'display: none', |
211 |
|
},{ |
212 |
|
# 1 |
213 |
|
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
214 |
|
filter_code => sub { shift }, # nop |
215 |
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
216 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
217 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
218 |
|
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
219 |
|
child_code => sub { return $_[1] }, |
220 |
|
style => 'display: none', |
221 |
|
},{ |
222 |
|
# 2 |
223 |
|
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
224 |
|
filter_code => sub { shift }, |
225 |
|
lookup_v900 => sub { shift }, |
226 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
227 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
228 |
|
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
229 |
|
child_code => sub { return "a:".$_[2].":".$_[1] }, |
230 |
|
style => 'display: none', |
231 |
|
},{ |
232 |
|
# 3 uži pojam |
233 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
234 |
|
filter_code => sub { shift }, |
235 |
|
lookup_v900 => sub { |
236 |
|
my ($c,$p) = @_; |
237 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
238 |
|
return "a:".$p.":".$c; |
239 |
|
}, |
240 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
241 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
242 |
|
have_children => sub { defined($l->{$_[1]}) }, |
243 |
|
child_code => sub { return $_[1] }, |
244 |
|
},{ |
245 |
|
# 4 |
246 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
247 |
|
filter_code => sub { shift }, |
248 |
|
lookup_v900 => sub { |
249 |
|
my ($c,$p) = @_; |
250 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
251 |
|
return "a:".$p.":".$c; |
252 |
|
}, |
253 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
254 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
255 |
|
have_children => sub { defined($l->{$_[1]}) }, |
256 |
|
child_code => sub { return $_[1] }, |
257 |
|
},{ |
258 |
|
# 5 |
259 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
260 |
|
filter_code => sub { shift }, |
261 |
|
lookup_v900 => sub { |
262 |
|
my ($c,$p) = @_; |
263 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
264 |
|
return "a:".$p.":".$c; |
265 |
|
}, |
266 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
267 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
268 |
|
have_children => sub { defined($l->{$_[1]}) }, |
269 |
|
child_code => sub { return $_[1] }, |
270 |
|
},{ |
271 |
|
# 6 |
272 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
273 |
|
filter_code => sub { shift }, |
274 |
|
lookup_v900 => sub { |
275 |
|
my ($c,$p) = @_; |
276 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
277 |
|
return "a:".$p.":".$c; |
278 |
|
}, |
279 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
280 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
281 |
|
have_children => sub { defined($l->{$_[1]}) }, |
282 |
|
child_code => sub { return $_[1] }, |
283 |
|
},{ |
284 |
|
# 7 |
285 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
286 |
|
filter_code => sub { shift }, |
287 |
|
lookup_v900 => sub { |
288 |
|
my ($c,$p) = @_; |
289 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
290 |
|
return "a:".$p.":".$c; |
291 |
|
}, |
292 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
293 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
294 |
|
have_children => sub { defined($l->{$_[1]}) }, |
295 |
|
child_code => sub { return $_[1] }, |
296 |
|
},{ |
297 |
|
# 8 |
298 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
299 |
|
filter_code => sub { shift }, |
300 |
|
lookup_v900 => sub { |
301 |
|
my ($c,$p) = @_; |
302 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
303 |
|
return "a:".$p.":".$c; |
304 |
|
}, |
305 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
306 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
307 |
|
# have_children => sub { defined($l->{$_[1]}) }, |
308 |
|
# child_code => sub { return $_[1] }, |
309 |
|
have_children => sub { 0 }, |
310 |
|
child_code => sub { 0 }, |
311 |
|
}); |
312 |
|
|
313 |
|
my @show_ids; |
314 |
|
my @hide_ids; |
315 |
|
|
316 |
|
unroll(0,''); |
317 |
|
|
318 |
|
$log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99")); |
319 |
|
|
320 |
|
sub unroll { |
321 |
|
my ($level,$start_code) = @_; |
322 |
|
|
323 |
|
$log->logconfess("need level") unless (defined($level)); |
324 |
|
|
325 |
|
# all levels passed? |
326 |
|
return if (! defined($tree[$level])); |
327 |
|
|
328 |
|
$log->debug("unroll level $level, start code $start_code"); |
329 |
|
|
330 |
|
foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) { |
331 |
|
|
332 |
|
if ($code = $tree[$level]->{'filter_code'}->($code)) { |
333 |
|
|
334 |
|
$log->debug("# $level filter passed code $code"); |
335 |
|
|
336 |
|
my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)"); |
337 |
|
$log->debug("# $level lookup_v900($code,$start_code) = $v900"); |
338 |
|
|
339 |
|
my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)"); |
340 |
|
$log->debug("# $level lookup_term($code,$v900) = $term"); |
341 |
|
|
342 |
|
my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)"); |
343 |
|
$log->debug("# $level lookup_mfn($code,$v900) = $mfn"); |
344 |
|
|
345 |
|
$log->debug("$code -> $v900 : $term [$mfn]"); |
346 |
|
|
347 |
|
my ($link_start,$link_end) = ('',''); |
348 |
|
|
349 |
|
my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code); |
350 |
|
if ($have_children) { |
351 |
|
($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>}); |
352 |
|
} else { |
353 |
|
$log->debug("# $level doesn't have_children($code,$v900,$start_code)"); |
354 |
|
} |
355 |
|
|
356 |
|
my $mfn_link = "thes/$mfn.html"; |
357 |
|
if (-e "out/$mfn_link") { |
358 |
|
print HTML " " x $level . |
359 |
|
qq{<li>${link_start}${term}${link_end}}. |
360 |
|
qq{ <a href="$mfn_link">»</a></li>\n}; |
361 |
|
} else { |
362 |
|
$log->warn("file 'out/$mfn_link' doesn't exist, skipping"); |
363 |
|
} |
364 |
|
|
365 |
foreach my $i (@lookup) { |
unless ($have_children) { |
366 |
my $d = $data->{$mfn}; |
next; |
|
if ($i->{'eval'}) { |
|
|
my $eval = fill_in($i->{'eval'},$d); |
|
|
my $key = fill_in($i->{'key'}, $d); |
|
|
my $val = fill_in($i->{'val'}, $d); |
|
|
if ($key && $val && eval $eval) { |
|
|
push @{$data->{lookup}->{$key}}, $val; |
|
367 |
} |
} |
368 |
} else { |
my $style = $tree[$level]->{'style'}; |
369 |
my $key = fill_in($i->{'key'}, $d); |
|
370 |
my $val = fill_in($i->{'val'}, $d); |
print HTML " " x $level . |
371 |
if ($key && $val) { |
qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}. |
372 |
push @{$data->{lookup}->{$key}}, $val; |
($style ? ' style="'.$style.'"' : ''). |
373 |
|
qq{>\n}; |
374 |
|
|
375 |
|
if ($style) { |
376 |
|
if ($style =~ m/display\s*:\s*none/i) { |
377 |
|
push @hide_ids, "mfn$mfn"; |
378 |
|
} else { |
379 |
|
push @show_ids, "mfn$mfn"; |
380 |
|
} |
381 |
|
} else { |
382 |
|
# default: show |
383 |
|
push @show_ids, "mfn$mfn"; |
384 |
} |
} |
385 |
|
|
386 |
|
unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code)); |
387 |
|
|
388 |
|
print HTML " " x $level . qq{</ul>\n}; |
389 |
|
|
390 |
} |
} |
391 |
} |
} |
|
|
|
392 |
} |
} |
|
print "\n"; |
|
393 |
|
|
394 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
print HTML qq{ |
395 |
my $d = $data->{$mfn}; |
</ul> |
396 |
print $d->{'250'}->[0]->{'a'},"\n"; |
</body> |
397 |
my $t = fill_in('v553^1;;v553^a', $d); |
</html> |
398 |
print " Uži pojam: $t\n" if ($t); |
}; |
399 |
$t = fill_in('[a:v251];;[d:[a:v251]]', $d); |
|
400 |
print " Uži pojam: $t\n" if ($t); |
close(HTML); |
401 |
} |
|
402 |
|
|
403 |
|
my $js_file = "out/$js_url"; |
404 |
|
$log->info("creating '$js_file' with arrays of shown and hidden ids"); |
405 |
|
open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!"); |
406 |
|
print JS "var show = ['",join("','",@show_ids),"'];\n"; |
407 |
|
print JS "var hide = ['",join("','",@hide_ids),"'];\n"; |
408 |
|
close(JS); |
409 |
|
|
410 |
|
$log->info("closing index"); |
411 |
|
$index->close; |
412 |
|
|
413 |
print Dumper($data); |
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |