/[webpac]/trunk2/all2all.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk2/all2all.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 579 - (hide annotations)
Mon Nov 1 22:15:41 2004 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 10103 byte(s)
make iframes after first level of tree

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 dpavlin 490 use locale;
11 dpavlin 454 use YAML;
12 dpavlin 352 use Carp;
13 dpavlin 437 use Getopt::Long;
14 dpavlin 546 use Text::Unaccent 1.02;
15 dpavlin 1
16 dpavlin 352 use lib './lib';
17 dpavlin 354 use WebPAC;
18 dpavlin 390 use WebPAC::jsFind;
19 dpavlin 410 use WebPAC::Index;
20 dpavlin 441 use WebPAC::Tree;
21 dpavlin 348
22 dpavlin 437 # options which can be changed via command line
23     #
24     my $code_page = 'ISO-8859-2';
25     my ($limit_mfn, $start_mfn, $debug, $low_mem);
26     my $index_path = './out/index';
27    
28     my $result = GetOptions(
29     "code_page=s" => \$code_page,
30     "limit_mfn=i" => \$limit_mfn,
31     "start_mfn=i" => \$start_mfn,
32     "debug!" => \$debug,
33     "low_mem!" => \$low_mem,
34     );
35    
36 dpavlin 500 my $filter = {
37     'CROVOC' => sub {
38     my $tmp = shift || return;
39 dpavlin 566 $tmp =~ s/\s*HIDRA.*//;
40 dpavlin 560 return undef unless ($tmp =~ s/\s*CROVOC.*$/ #C#/);
41     # remove repeating stars
42     # FIXME this should be fixed at right place, not here!
43     $tmp =~ s/(\s*#C#)+/ #C#/g;
44 dpavlin 500 return $tmp;
45     },
46 dpavlin 501 'CROVOC_tree' => sub {
47     my $tmp = shift || return;
48 dpavlin 561 $tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/;
49 dpavlin 503 $tmp =~ s/\s*EUROVOC.*//;
50 dpavlin 566 $tmp =~ s/\s*HIDRA.*//;
51 dpavlin 501 return $tmp;
52     },
53 dpavlin 560 # TT filter
54     'CROVOC_img' => sub {
55     my $tmp = shift;
56     $tmp =~ s/\s*#C#\s*/ <img src="..\/img\/crovoc.png" border="0">/gis;
57 dpavlin 562 $tmp =~ s/"img\/crovoc.png"/"..\/img\/crovoc.png"/gis;
58 dpavlin 560 return $tmp;
59     }
60 dpavlin 500 };
61    
62 dpavlin 546 ## remove accented characters
63     #
64     sub unac {
65     my $string = shift || return;
66     $string = unac_string($code_page,$string);
67     $string =~ tr/ðÐ/dD/;
68 dpavlin 550 $string = unac_entities($string);
69 dpavlin 546 return $string;
70     }
71     sub unac_2 {
72     my $string = shift || return;
73     if (length($string) > 2) {
74     my $pr = substr($string,0,2);
75     $string = unac_string($code_page,substr($string,2));
76     $string =~ tr/ðÐ/dD/;
77     $string = $pr . $string;
78     }
79 dpavlin 550 $string = unac_entities($string);
80 dpavlin 546 return lc($string);
81     }
82    
83 dpavlin 550 sub unac_entities {
84     my $ent = shift || return;
85    
86     $ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi;
87     $ent =~ s/&eth;/d/g;
88     $ent =~ s/&E[tT][hH];/D/g;
89     $ent =~ s/&(\w\w)lig;/$1/gi;
90    
91     return $ent;
92     }
93    
94 dpavlin 437 # create WebPAC object
95     #
96 dpavlin 354 my $webpac = new WebPAC(
97 dpavlin 437 code_page => $code_page,
98     limit_mfn => $limit_mfn,
99     start_mfn => $start_mfn,
100     debug => $debug,
101     low_mem => $low_mem,
102 dpavlin 500 filter => $filter,
103 dpavlin 352 ) || die;
104    
105 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
106 dpavlin 373
107 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
108    
109 dpavlin 390 my $index = new WebPAC::jsFind(
110 dpavlin 437 index_path => $index_path,
111 dpavlin 560 keys => 62,
112 dpavlin 390 ) || die;
113    
114 dpavlin 410 my $thes;
115    
116 dpavlin 10 $|=1;
117 dpavlin 9
118 dpavlin 352 my $maxmfn = $webpac->open_isis(
119 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
120 dpavlin 352 lookup => [
121 dpavlin 501 { 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' },
122 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
123     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
124     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
125     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
126 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
127 dpavlin 419 # tree structure
128 dpavlin 471 { 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
129 dpavlin 419 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
130 dpavlin 460 { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
131 dpavlin 352 ],
132 dpavlin 54 );
133 dpavlin 3
134 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
135    
136 dpavlin 372 $log->info("rows: $maxmfn");
137 dpavlin 170
138 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
139    
140 dpavlin 425 if(1) { # XXX
141 dpavlin 424
142 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
143 dpavlin 358
144 dpavlin 370 my @ds = $webpac->data_structure($rec);
145 dpavlin 366
146 dpavlin 411 if (0 && $log->is_debug) {
147 dpavlin 454 $log->debug("rec = ",Dump($rec));
148     $log->debug("ds = ",Dump(\@ds));
149 dpavlin 372 }
150    
151 dpavlin 374 next if (! @ds);
152 dpavlin 366
153 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
154 dpavlin 374
155     if ($filename) {
156 dpavlin 411 $webpac->output_file(
157     file => $filename,
158 dpavlin 374 template => 'html.tt',
159     data => \@ds,
160 dpavlin 398 headline => $webpac->{'headline'},
161 dpavlin 374 );
162     } else {
163     print $webpac->output(
164     template => 'text.tt',
165     data => \@ds,
166 dpavlin 398 headline => $webpac->{'headline'},
167 dpavlin 374 );
168     }
169    
170 dpavlin 398 my $headline = $webpac->{'headline'};
171 dpavlin 390
172 dpavlin 392 my $f = $filename;
173     $f =~ s!out/!!;
174    
175 dpavlin 390 # save into index
176     foreach my $ds (@ds) {
177     next if (! $ds->{'swish'});
178    
179 dpavlin 466 # strip all non word characters from beginning or end
180     # of word
181     my $words = join(" ",@{$ds->{'swish'}});
182     $words =~ s/^\W+//;
183     $words =~ s/\W*\s+\W*/ /g;
184     $words =~ s/\W+$//;
185    
186 dpavlin 504 # first try to generate headline for this entry from index
187 dpavlin 520 my $h = $ds->{'index'}->[0];
188 dpavlin 504 # then, from display
189 dpavlin 520 $h ||= $ds->{'display'}->[0];
190 dpavlin 504 # and as last resport, fallback to headline
191     $h ||= $headline;
192    
193 dpavlin 390 $index->insert(
194     index_name => $ds->{'tag'},
195 dpavlin 466 #path => $f,
196     path => $webpac->mfn,
197 dpavlin 504 headline => $h,
198 dpavlin 546 words => unac($words),
199 dpavlin 390 );
200     }
201    
202 dpavlin 410 # save into sorted index (thesaurus)
203     foreach my $ds (@ds) {
204     next if (! $ds->{'index'});
205    
206 dpavlin 511 $thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} );
207 dpavlin 410
208 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
209     $thes->{$ds->{'tag'}}->insert(
210 dpavlin 546 sort_by => unac_2($h),
211     mfn => $webpac->mfn,
212 dpavlin 415 headline => $h,
213     );
214     }
215 dpavlin 410 }
216    
217 dpavlin 454 # print Dump(\@ds);
218 dpavlin 390
219 dpavlin 3 }
220    
221 dpavlin 411 foreach my $t (keys %{$thes}) {
222    
223     my @e = $thes->{$t}->elements;
224     if (! @e) {
225     $log->logwarn("no elements in sorted index $t?");
226     next;
227     }
228    
229 dpavlin 415 my $file = "./out/bfilter/$t.txt";
230     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
231 dpavlin 411
232     $webpac->output_file(
233 dpavlin 415 file => $file,
234 dpavlin 411 template => 'index.tt',
235     data => \@e,
236     index_name => $t,
237     );
238     }
239    
240     if (0 && $log->is_debug) {
241 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
242     $log->debug("data hash: ",Dump($webpac->{'data'}));
243 dpavlin 410 foreach my $t (keys %{$thes}) {
244 dpavlin 454 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
245 dpavlin 410 }
246 dpavlin 372 }
247 dpavlin 390
248 dpavlin 424 } # XXX if(0)
249    
250 dpavlin 530 #$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
251 dpavlin 420
252 dpavlin 441 $log->info("creating tree");
253    
254 dpavlin 419 #
255 dpavlin 441 # define tree structure
256 dpavlin 419 #
257    
258 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
259 dpavlin 419
260 dpavlin 424 my @tree = ({
261     # level 0
262     code_arr => sub { sort keys %{$l} },
263     filter_code => sub {
264     my $t = shift;
265     return $t if ($t =~ s/root://);
266     },
267     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
268     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
269     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
270 dpavlin 454 have_children => sub { return $l->{"a:".$_[0]."::"} },
271     have_children_at_level => sub {
272 dpavlin 471 return unless (defined($l->{"code:".$_[1]}));
273     my $code = shift @{$l->{"code:".$_[1]}};
274     print STDERR "## $_[1] -> $code\n";
275     return undef unless($code);
276     return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
277 dpavlin 454 },
278 dpavlin 424 style => 'display: none',
279 dpavlin 579 iframe => 1,
280 dpavlin 424 },{
281     # 1
282 dpavlin 454 code_arr => sub { @{$_[0]} },
283 dpavlin 424 filter_code => sub { shift }, # nop
284     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
285     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
286     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
287 dpavlin 454 have_children => sub { return $l->{"a:".$_[1].":"} },
288 dpavlin 424 style => 'display: none',
289     },{
290     # 2
291 dpavlin 454 code_arr => sub { @{$_[0]} },
292 dpavlin 424 filter_code => sub { shift },
293 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
294 dpavlin 424 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
295     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
296 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
297 dpavlin 572 style => 'display: none',
298 dpavlin 425 },{
299     # 3 u¾i pojam
300 dpavlin 454 code_arr => sub { @{$_[0]} },
301 dpavlin 425 filter_code => sub { shift },
302 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
303 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
304     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
305 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
306 dpavlin 426 },{
307     # 4
308 dpavlin 454 code_arr => sub { @{$_[0]} },
309 dpavlin 426 filter_code => sub { shift },
310 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
311 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
312     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
313 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
314 dpavlin 427 },{
315     # 5
316 dpavlin 454 code_arr => sub { @{$_[0]} },
317 dpavlin 427 filter_code => sub { shift },
318 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
319 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
320     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
321 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
322 dpavlin 427 },{
323     # 6
324 dpavlin 454 code_arr => sub { @{$_[0]} },
325 dpavlin 427 filter_code => sub { shift },
326 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
327 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
328     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
329 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
330 dpavlin 427 },{
331     # 7
332 dpavlin 454 code_arr => sub { @{$_[0]} },
333 dpavlin 427 filter_code => sub { shift },
334 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
335 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
336     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
337 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
338 dpavlin 427 },{
339     # 8
340 dpavlin 454 code_arr => sub { @{$_[0]} },
341 dpavlin 427 filter_code => sub { shift },
342 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
343 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
344     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
345 dpavlin 460 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
346 dpavlin 424 have_children => sub { 0 },
347 dpavlin 454 },{
348     # 9 - level which is never reached except explicitly
349 dpavlin 471 code_arr => sub { @{$_[0]} },
350 dpavlin 454 filter_code => sub { shift },
351 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
352 dpavlin 454 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
353     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
354     have_children => sub { 0 },
355 dpavlin 471 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
356 dpavlin 454 },{
357 dpavlin 424 });
358 dpavlin 419
359 dpavlin 441 my $tree = new WebPAC::Tree(
360 dpavlin 530 tree => \@tree,
361 dpavlin 573 detail_url => sub {
362     my $mfn = shift;
363 dpavlin 574 my $url = "thes/${mfn}.html";
364     return $url if (-e "./out/$url");
365 dpavlin 573 },
366 dpavlin 575 nodes_dir => 'nodes',
367 dpavlin 530 );
368    
369     $tree->output(
370     dir => './out',
371     html => 'browse.html',
372 dpavlin 572 template_dir => './output_template/',
373     template_tree => 'tree.tt',
374     template_node => 'node.tt',
375 dpavlin 441 js => 'tree-ids.js',
376     );
377 dpavlin 428
378 dpavlin 530 $tree->output(
379     dir => './eurovoc',
380     html => 'hijerarhija.html',
381 dpavlin 572 template_dir => './output_template/',
382     template_tree => 'hijerarhija.tt',
383     template_node => 'hijerarhija_node.tt',
384 dpavlin 530 js => 'tree-ids.js',
385     );
386    
387    
388 dpavlin 428 $log->info("closing index");
389 dpavlin 390 $index->close;
390    
391 dpavlin 422 $log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'}));

Properties

Name Value
cvs2svn:cvs-rev 1.64
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26