1 |
dpavlin |
1 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
dpavlin |
348 |
=head1 NAME |
4 |
|
|
|
5 |
dpavlin |
354 |
all2all.pl - basic script for all WebPAC needs |
6 |
dpavlin |
348 |
|
7 |
|
|
=cut |
8 |
|
|
|
9 |
dpavlin |
1 |
use strict; |
10 |
|
|
use Data::Dumper; |
11 |
dpavlin |
352 |
use Carp; |
12 |
dpavlin |
1 |
|
13 |
dpavlin |
352 |
use lib './lib'; |
14 |
dpavlin |
354 |
use WebPAC; |
15 |
dpavlin |
390 |
use WebPAC::jsFind; |
16 |
dpavlin |
410 |
use WebPAC::Index; |
17 |
dpavlin |
348 |
|
18 |
dpavlin |
354 |
my $webpac = new WebPAC( |
19 |
dpavlin |
352 |
code_page => 'ISO-8859-2', |
20 |
dpavlin |
423 |
limit_mfn => 500, |
21 |
dpavlin |
372 |
# debug => 1, |
22 |
dpavlin |
423 |
# low_mem => 1, |
23 |
dpavlin |
352 |
) || die; |
24 |
|
|
|
25 |
dpavlin |
421 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
26 |
dpavlin |
373 |
|
27 |
dpavlin |
421 |
$log->debug("creating WebPAC::jsFind object"); |
28 |
|
|
|
29 |
dpavlin |
390 |
my $index = new WebPAC::jsFind( |
30 |
|
|
index_path => './out/index', |
31 |
dpavlin |
413 |
keys => 10, |
32 |
dpavlin |
390 |
) || die; |
33 |
|
|
|
34 |
dpavlin |
410 |
my $thes; |
35 |
|
|
|
36 |
dpavlin |
10 |
$|=1; |
37 |
dpavlin |
9 |
|
38 |
dpavlin |
352 |
my $maxmfn = $webpac->open_isis( |
39 |
dpavlin |
357 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
40 |
dpavlin |
352 |
lookup => [ |
41 |
dpavlin |
348 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
42 |
dpavlin |
352 |
# { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
43 |
|
|
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
44 |
|
|
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
45 |
|
|
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
46 |
dpavlin |
374 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
47 |
dpavlin |
419 |
# tree structure |
48 |
|
|
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
49 |
|
|
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
50 |
dpavlin |
352 |
], |
51 |
dpavlin |
54 |
); |
52 |
dpavlin |
3 |
|
53 |
dpavlin |
421 |
$log->debug("isis file ",$webpac->{'isis_filename'}," opened"); |
54 |
|
|
|
55 |
dpavlin |
372 |
$log->info("rows: $maxmfn"); |
56 |
dpavlin |
170 |
|
57 |
dpavlin |
363 |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
58 |
|
|
|
59 |
dpavlin |
425 |
if(1) { # XXX |
60 |
dpavlin |
424 |
|
61 |
dpavlin |
362 |
while (my $rec = $webpac->fetch_rec) { |
62 |
dpavlin |
358 |
|
63 |
dpavlin |
370 |
my @ds = $webpac->data_structure($rec); |
64 |
dpavlin |
366 |
|
65 |
dpavlin |
411 |
if (0 && $log->is_debug) { |
66 |
dpavlin |
372 |
$log->debug("rec = ",Dumper($rec)); |
67 |
|
|
$log->debug("ds = ",Dumper(\@ds)); |
68 |
|
|
} |
69 |
|
|
|
70 |
dpavlin |
374 |
next if (! @ds); |
71 |
dpavlin |
366 |
|
72 |
dpavlin |
421 |
my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object"); |
73 |
dpavlin |
374 |
|
74 |
|
|
if ($filename) { |
75 |
dpavlin |
411 |
$webpac->output_file( |
76 |
|
|
file => $filename, |
77 |
dpavlin |
374 |
template => 'html.tt', |
78 |
|
|
data => \@ds, |
79 |
dpavlin |
398 |
headline => $webpac->{'headline'}, |
80 |
dpavlin |
374 |
); |
81 |
|
|
} else { |
82 |
|
|
print $webpac->output( |
83 |
|
|
template => 'text.tt', |
84 |
|
|
data => \@ds, |
85 |
dpavlin |
398 |
headline => $webpac->{'headline'}, |
86 |
dpavlin |
374 |
); |
87 |
|
|
} |
88 |
|
|
|
89 |
dpavlin |
398 |
my $headline = $webpac->{'headline'}; |
90 |
dpavlin |
390 |
|
91 |
dpavlin |
392 |
my $f = $filename; |
92 |
|
|
$f =~ s!out/!!; |
93 |
|
|
|
94 |
dpavlin |
390 |
# save into index |
95 |
|
|
foreach my $ds (@ds) { |
96 |
|
|
next if (! $ds->{'swish'}); |
97 |
|
|
|
98 |
|
|
$index->insert( |
99 |
|
|
index_name => $ds->{'tag'}, |
100 |
dpavlin |
392 |
path => $f, |
101 |
dpavlin |
390 |
headline => $headline, |
102 |
|
|
words => join(" ",@{$ds->{'swish'}}) |
103 |
|
|
); |
104 |
|
|
} |
105 |
|
|
|
106 |
dpavlin |
410 |
# save into sorted index (thesaurus) |
107 |
|
|
foreach my $ds (@ds) { |
108 |
|
|
next if (! $ds->{'index'}); |
109 |
|
|
|
110 |
|
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
111 |
|
|
|
112 |
dpavlin |
415 |
foreach my $h (@{$ds->{'index'}}) { |
113 |
|
|
$thes->{$ds->{'tag'}}->insert( |
114 |
|
|
path => $f, |
115 |
|
|
headline => $h, |
116 |
|
|
); |
117 |
|
|
} |
118 |
dpavlin |
410 |
} |
119 |
|
|
|
120 |
dpavlin |
390 |
# print Dumper(\@ds); |
121 |
|
|
|
122 |
dpavlin |
3 |
} |
123 |
|
|
|
124 |
dpavlin |
411 |
foreach my $t (keys %{$thes}) { |
125 |
|
|
|
126 |
|
|
my @e = $thes->{$t}->elements; |
127 |
|
|
if (! @e) { |
128 |
|
|
$log->logwarn("no elements in sorted index $t?"); |
129 |
|
|
next; |
130 |
|
|
} |
131 |
|
|
|
132 |
dpavlin |
415 |
my $file = "./out/bfilter/$t.txt"; |
133 |
|
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
134 |
dpavlin |
411 |
|
135 |
|
|
$webpac->output_file( |
136 |
dpavlin |
415 |
file => $file, |
137 |
dpavlin |
411 |
template => 'index.tt', |
138 |
|
|
data => \@e, |
139 |
|
|
index_name => $t, |
140 |
|
|
); |
141 |
|
|
} |
142 |
|
|
|
143 |
|
|
if (0 && $log->is_debug) { |
144 |
dpavlin |
372 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
145 |
|
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
146 |
dpavlin |
410 |
foreach my $t (keys %{$thes}) { |
147 |
|
|
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
148 |
|
|
} |
149 |
dpavlin |
372 |
} |
150 |
dpavlin |
390 |
|
151 |
dpavlin |
424 |
} # XXX if(0) |
152 |
|
|
|
153 |
dpavlin |
428 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
154 |
dpavlin |
420 |
|
155 |
dpavlin |
419 |
# |
156 |
|
|
# tree dump implementation which shouldn't be here :-) |
157 |
|
|
# |
158 |
|
|
|
159 |
dpavlin |
428 |
|
160 |
dpavlin |
419 |
my $file = 'out/browse.html'; |
161 |
dpavlin |
428 |
my $js_url = 'tree-ids.js'; |
162 |
|
|
|
163 |
|
|
$log->info("creating '$file' for tree html"); |
164 |
dpavlin |
419 |
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
165 |
|
|
|
166 |
|
|
print HTML qq{ |
167 |
|
|
<html> |
168 |
|
|
<head> |
169 |
|
|
<title>Browse</title> |
170 |
|
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
171 |
dpavlin |
423 |
<script type="text/javascript" src="bfilter/hide.js"></script> |
172 |
dpavlin |
428 |
<script type="text/javascript" src="$js_url"></script> |
173 |
dpavlin |
419 |
</head> |
174 |
dpavlin |
428 |
|
175 |
|
|
<body onLoad="show_hide_display(show,hide);"> |
176 |
|
|
|
177 |
|
|
<div style="float: right; width: 10em;"> |
178 |
|
|
Folding: |
179 |
|
|
<a href="#" onClick="show_hide_display(show,hide);">default</a> |
180 |
|
|
<a href="#" onClick="show_display(show); show_display(hide);">all</a> |
181 |
|
|
<a href="#" onClick="hide_display(hide); hide_display(show);">none</a> |
182 |
|
|
<a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length);">debug</a> |
183 |
|
|
</div> |
184 |
|
|
|
185 |
|
|
|
186 |
dpavlin |
419 |
<ul> |
187 |
|
|
}; |
188 |
|
|
|
189 |
dpavlin |
424 |
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
190 |
dpavlin |
419 |
|
191 |
dpavlin |
424 |
my @tree = ({ |
192 |
|
|
# level 0 |
193 |
|
|
code_arr => sub { sort keys %{$l} }, |
194 |
|
|
filter_code => sub { |
195 |
|
|
my $t = shift; |
196 |
|
|
return $t if ($t =~ s/root://); |
197 |
|
|
}, |
198 |
|
|
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
199 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
200 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
201 |
dpavlin |
426 |
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
202 |
dpavlin |
424 |
child_code => sub { return $_[0] }, |
203 |
|
|
style => 'display: none', |
204 |
|
|
},{ |
205 |
|
|
# 1 |
206 |
|
|
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
207 |
|
|
filter_code => sub { shift }, # nop |
208 |
|
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
209 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
210 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
211 |
dpavlin |
426 |
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
212 |
dpavlin |
424 |
child_code => sub { return $_[1] }, |
213 |
|
|
style => 'display: none', |
214 |
|
|
},{ |
215 |
|
|
# 2 |
216 |
|
|
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
217 |
|
|
filter_code => sub { shift }, |
218 |
|
|
lookup_v900 => sub { shift }, |
219 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
220 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
221 |
dpavlin |
426 |
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
222 |
dpavlin |
425 |
child_code => sub { return "a:".$_[2].":".$_[1] }, |
223 |
|
|
style => 'display: none', |
224 |
|
|
},{ |
225 |
|
|
# 3 u¾i pojam |
226 |
|
|
code_arr => sub { @{$l->{$_[0]}} }, |
227 |
|
|
filter_code => sub { shift }, |
228 |
dpavlin |
426 |
lookup_v900 => sub { |
229 |
|
|
my ($c,$p) = @_; |
230 |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
231 |
|
|
return "a:".$p.":".$c; |
232 |
|
|
}, |
233 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
234 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
235 |
|
|
have_children => sub { defined($l->{$_[1]}) }, |
236 |
|
|
child_code => sub { return $_[1] }, |
237 |
|
|
},{ |
238 |
|
|
# 4 |
239 |
|
|
code_arr => sub { @{$l->{$_[0]}} }, |
240 |
|
|
filter_code => sub { shift }, |
241 |
|
|
lookup_v900 => sub { |
242 |
|
|
my ($c,$p) = @_; |
243 |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
244 |
|
|
return "a:".$p.":".$c; |
245 |
|
|
}, |
246 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
247 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
248 |
dpavlin |
427 |
have_children => sub { defined($l->{$_[1]}) }, |
249 |
|
|
child_code => sub { return $_[1] }, |
250 |
|
|
},{ |
251 |
|
|
# 5 |
252 |
|
|
code_arr => sub { @{$l->{$_[0]}} }, |
253 |
|
|
filter_code => sub { shift }, |
254 |
|
|
lookup_v900 => sub { |
255 |
|
|
my ($c,$p) = @_; |
256 |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
257 |
|
|
return "a:".$p.":".$c; |
258 |
|
|
}, |
259 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
260 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
261 |
|
|
have_children => sub { defined($l->{$_[1]}) }, |
262 |
|
|
child_code => sub { return $_[1] }, |
263 |
|
|
},{ |
264 |
|
|
# 6 |
265 |
|
|
code_arr => sub { @{$l->{$_[0]}} }, |
266 |
|
|
filter_code => sub { shift }, |
267 |
|
|
lookup_v900 => sub { |
268 |
|
|
my ($c,$p) = @_; |
269 |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
270 |
|
|
return "a:".$p.":".$c; |
271 |
|
|
}, |
272 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
273 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
274 |
|
|
have_children => sub { defined($l->{$_[1]}) }, |
275 |
|
|
child_code => sub { return $_[1] }, |
276 |
|
|
},{ |
277 |
|
|
# 7 |
278 |
|
|
code_arr => sub { @{$l->{$_[0]}} }, |
279 |
|
|
filter_code => sub { shift }, |
280 |
|
|
lookup_v900 => sub { |
281 |
|
|
my ($c,$p) = @_; |
282 |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
283 |
|
|
return "a:".$p.":".$c; |
284 |
|
|
}, |
285 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
286 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
287 |
|
|
have_children => sub { defined($l->{$_[1]}) }, |
288 |
|
|
child_code => sub { return $_[1] }, |
289 |
|
|
},{ |
290 |
|
|
# 8 |
291 |
|
|
code_arr => sub { @{$l->{$_[0]}} }, |
292 |
|
|
filter_code => sub { shift }, |
293 |
|
|
lookup_v900 => sub { |
294 |
|
|
my ($c,$p) = @_; |
295 |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
296 |
|
|
return "a:".$p.":".$c; |
297 |
|
|
}, |
298 |
|
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
299 |
|
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
300 |
dpavlin |
426 |
# have_children => sub { defined($l->{$_[1]}) }, |
301 |
|
|
# child_code => sub { return $_[1] }, |
302 |
dpavlin |
424 |
have_children => sub { 0 }, |
303 |
|
|
child_code => sub { 0 }, |
304 |
|
|
}); |
305 |
dpavlin |
419 |
|
306 |
dpavlin |
428 |
my @show_ids; |
307 |
|
|
my @hide_ids; |
308 |
|
|
|
309 |
dpavlin |
424 |
unroll(0,''); |
310 |
dpavlin |
419 |
|
311 |
dpavlin |
424 |
$log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99")); |
312 |
dpavlin |
419 |
|
313 |
dpavlin |
424 |
sub unroll { |
314 |
|
|
my ($level,$start_code) = @_; |
315 |
dpavlin |
419 |
|
316 |
dpavlin |
424 |
$log->logconfess("need level") unless (defined($level)); |
317 |
dpavlin |
419 |
|
318 |
dpavlin |
424 |
# all levels passed? |
319 |
|
|
return if (! defined($tree[$level])); |
320 |
dpavlin |
419 |
|
321 |
dpavlin |
424 |
$log->debug("unroll level $level, start code $start_code"); |
322 |
dpavlin |
419 |
|
323 |
dpavlin |
424 |
foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) { |
324 |
dpavlin |
419 |
|
325 |
dpavlin |
424 |
if ($code = $tree[$level]->{'filter_code'}->($code)) { |
326 |
|
|
|
327 |
|
|
$log->debug("# $level filter passed code $code"); |
328 |
dpavlin |
419 |
|
329 |
dpavlin |
426 |
my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)"); |
330 |
|
|
$log->debug("# $level lookup_v900($code,$start_code) = $v900"); |
331 |
|
|
|
332 |
|
|
my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)"); |
333 |
dpavlin |
424 |
$log->debug("# $level lookup_term($code,$v900) = $term"); |
334 |
dpavlin |
426 |
|
335 |
|
|
my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)"); |
336 |
dpavlin |
424 |
$log->debug("# $level lookup_mfn($code,$v900) = $mfn"); |
337 |
dpavlin |
419 |
|
338 |
dpavlin |
424 |
$log->debug("$code -> $v900 : $term [$mfn]"); |
339 |
dpavlin |
419 |
|
340 |
dpavlin |
426 |
my ($link_start,$link_end) = ('',''); |
341 |
dpavlin |
425 |
|
342 |
|
|
my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code); |
343 |
dpavlin |
426 |
if ($have_children) { |
344 |
dpavlin |
428 |
($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>}); |
345 |
dpavlin |
426 |
} else { |
346 |
|
|
$log->debug("# $level doesn't have_children($code,$v900,$start_code)"); |
347 |
|
|
} |
348 |
dpavlin |
419 |
|
349 |
dpavlin |
425 |
my $mfn_link = "thes/$mfn.html"; |
350 |
|
|
if (-e "out/$mfn_link") { |
351 |
|
|
print HTML " " x $level . |
352 |
|
|
qq{<li>${link_start}${term}${link_end}}. |
353 |
|
|
qq{ <a href="$mfn_link">»</a></li>\n}; |
354 |
|
|
} else { |
355 |
|
|
$log->warn("file 'out/$mfn_link' doesn't exist, skipping"); |
356 |
|
|
} |
357 |
|
|
|
358 |
|
|
unless ($have_children) { |
359 |
dpavlin |
424 |
next; |
360 |
dpavlin |
419 |
} |
361 |
dpavlin |
428 |
my $style = $tree[$level]->{'style'}; |
362 |
|
|
|
363 |
dpavlin |
424 |
print HTML " " x $level . |
364 |
|
|
qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}. |
365 |
dpavlin |
428 |
($style ? ' style="$style"' : ""). |
366 |
dpavlin |
424 |
qq{>\n}; |
367 |
dpavlin |
419 |
|
368 |
dpavlin |
428 |
if ($style) { |
369 |
|
|
if ($style =~ m/display\s*:\s*none/i) { |
370 |
|
|
push @hide_ids, "mfn$mfn"; |
371 |
|
|
} else { |
372 |
|
|
push @show_ids, "mfn$mfn"; |
373 |
|
|
} |
374 |
|
|
} else { |
375 |
|
|
# default: show |
376 |
|
|
push @show_ids, "mfn$mfn"; |
377 |
|
|
} |
378 |
|
|
|
379 |
dpavlin |
425 |
unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code)); |
380 |
dpavlin |
424 |
|
381 |
|
|
print HTML " " x $level . qq{</ul>\n}; |
382 |
|
|
|
383 |
dpavlin |
419 |
} |
384 |
|
|
} |
385 |
|
|
} |
386 |
|
|
|
387 |
|
|
print HTML qq{ |
388 |
|
|
</ul> |
389 |
|
|
</body> |
390 |
|
|
</html> |
391 |
|
|
}; |
392 |
|
|
|
393 |
dpavlin |
425 |
close(HTML); |
394 |
|
|
|
395 |
dpavlin |
428 |
|
396 |
|
|
my $js_file = "out/$js_url"; |
397 |
|
|
$log->info("creating '$js_file' with arrays of shown and hidden ids"); |
398 |
|
|
open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!"); |
399 |
|
|
print JS "var show = ['",join("','",@show_ids),"'];\n"; |
400 |
|
|
print JS "var hide = ['",join("','",@hide_ids),"'];\n"; |
401 |
|
|
close(JS); |
402 |
|
|
|
403 |
|
|
$log->info("closing index"); |
404 |
dpavlin |
390 |
$index->close; |
405 |
|
|
|
406 |
dpavlin |
422 |
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |