/[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 591 - (hide annotations)
Sat Nov 6 22:46:30 2004 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 10163 byte(s)
added crovoc simbols (using lookup)

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 591 { 'key' => 'crovoc:v900', 'val' => 'filter{CROVOC}v800' },
132 dpavlin 352 ],
133 dpavlin 54 );
134 dpavlin 3
135 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
136    
137 dpavlin 372 $log->info("rows: $maxmfn");
138 dpavlin 170
139 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
140    
141 dpavlin 425 if(1) { # XXX
142 dpavlin 424
143 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
144 dpavlin 358
145 dpavlin 370 my @ds = $webpac->data_structure($rec);
146 dpavlin 366
147 dpavlin 411 if (0 && $log->is_debug) {
148 dpavlin 454 $log->debug("rec = ",Dump($rec));
149     $log->debug("ds = ",Dump(\@ds));
150 dpavlin 372 }
151    
152 dpavlin 374 next if (! @ds);
153 dpavlin 366
154 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
155 dpavlin 374
156     if ($filename) {
157 dpavlin 411 $webpac->output_file(
158     file => $filename,
159 dpavlin 374 template => 'html.tt',
160     data => \@ds,
161 dpavlin 398 headline => $webpac->{'headline'},
162 dpavlin 374 );
163     } else {
164     print $webpac->output(
165     template => 'text.tt',
166     data => \@ds,
167 dpavlin 398 headline => $webpac->{'headline'},
168 dpavlin 374 );
169     }
170    
171 dpavlin 398 my $headline = $webpac->{'headline'};
172 dpavlin 390
173 dpavlin 392 my $f = $filename;
174     $f =~ s!out/!!;
175    
176 dpavlin 390 # save into index
177     foreach my $ds (@ds) {
178     next if (! $ds->{'swish'});
179    
180 dpavlin 466 # strip all non word characters from beginning or end
181     # of word
182 dpavlin 587 my $words = unac(join(" ",@{$ds->{'swish'}}));
183 dpavlin 466 $words =~ s/^\W+//;
184     $words =~ s/\W*\s+\W*/ /g;
185     $words =~ s/\W+$//;
186    
187 dpavlin 504 # first try to generate headline for this entry from index
188 dpavlin 520 my $h = $ds->{'index'}->[0];
189 dpavlin 504 # then, from display
190 dpavlin 520 $h ||= $ds->{'display'}->[0];
191 dpavlin 504 # and as last resport, fallback to headline
192     $h ||= $headline;
193    
194 dpavlin 390 $index->insert(
195     index_name => $ds->{'tag'},
196 dpavlin 466 #path => $f,
197     path => $webpac->mfn,
198 dpavlin 504 headline => $h,
199 dpavlin 587 words => $words,
200 dpavlin 390 );
201     }
202    
203 dpavlin 410 # save into sorted index (thesaurus)
204     foreach my $ds (@ds) {
205     next if (! $ds->{'index'});
206    
207 dpavlin 511 $thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} );
208 dpavlin 410
209 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
210     $thes->{$ds->{'tag'}}->insert(
211 dpavlin 546 sort_by => unac_2($h),
212     mfn => $webpac->mfn,
213 dpavlin 415 headline => $h,
214     );
215     }
216 dpavlin 410 }
217    
218 dpavlin 454 # print Dump(\@ds);
219 dpavlin 390
220 dpavlin 3 }
221    
222 dpavlin 411 foreach my $t (keys %{$thes}) {
223    
224     my @e = $thes->{$t}->elements;
225     if (! @e) {
226     $log->logwarn("no elements in sorted index $t?");
227     next;
228     }
229    
230 dpavlin 415 my $file = "./out/bfilter/$t.txt";
231     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
232 dpavlin 411
233     $webpac->output_file(
234 dpavlin 415 file => $file,
235 dpavlin 411 template => 'index.tt',
236     data => \@e,
237     index_name => $t,
238     );
239     }
240    
241     if (0 && $log->is_debug) {
242 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
243     $log->debug("data hash: ",Dump($webpac->{'data'}));
244 dpavlin 410 foreach my $t (keys %{$thes}) {
245 dpavlin 454 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
246 dpavlin 410 }
247 dpavlin 372 }
248 dpavlin 390
249 dpavlin 424 } # XXX if(0)
250    
251 dpavlin 530 #$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
252 dpavlin 420
253 dpavlin 441 $log->info("creating tree");
254    
255 dpavlin 419 #
256 dpavlin 441 # define tree structure
257 dpavlin 419 #
258    
259 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
260 dpavlin 419
261 dpavlin 424 my @tree = ({
262     # level 0
263     code_arr => sub { sort keys %{$l} },
264     filter_code => sub {
265     my $t = shift;
266     return $t if ($t =~ s/root://);
267     },
268     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
269     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
270     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
271 dpavlin 454 have_children => sub { return $l->{"a:".$_[0]."::"} },
272     have_children_at_level => sub {
273 dpavlin 471 return unless (defined($l->{"code:".$_[1]}));
274     my $code = shift @{$l->{"code:".$_[1]}};
275     print STDERR "## $_[1] -> $code\n";
276     return undef unless($code);
277     return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
278 dpavlin 454 },
279 dpavlin 424 style => 'display: none',
280 dpavlin 579 iframe => 1,
281 dpavlin 424 },{
282     # 1
283 dpavlin 454 code_arr => sub { @{$_[0]} },
284 dpavlin 424 filter_code => sub { shift }, # nop
285     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
286     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
287     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
288 dpavlin 454 have_children => sub { return $l->{"a:".$_[1].":"} },
289 dpavlin 424 style => 'display: none',
290     },{
291     # 2
292 dpavlin 454 code_arr => sub { @{$_[0]} },
293 dpavlin 424 filter_code => sub { shift },
294 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
295 dpavlin 424 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
296     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
297 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
298 dpavlin 572 style => 'display: none',
299 dpavlin 425 },{
300     # 3 u¾i pojam
301 dpavlin 454 code_arr => sub { @{$_[0]} },
302 dpavlin 425 filter_code => sub { shift },
303 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
304 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
305     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
306 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
307 dpavlin 426 },{
308     # 4
309 dpavlin 454 code_arr => sub { @{$_[0]} },
310 dpavlin 426 filter_code => sub { shift },
311 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
312 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
313     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
314 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
315 dpavlin 427 },{
316     # 5
317 dpavlin 454 code_arr => sub { @{$_[0]} },
318 dpavlin 427 filter_code => sub { shift },
319 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
320 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
321     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
322 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
323 dpavlin 427 },{
324     # 6
325 dpavlin 454 code_arr => sub { @{$_[0]} },
326 dpavlin 427 filter_code => sub { shift },
327 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
328 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
329     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
330 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
331 dpavlin 427 },{
332     # 7
333 dpavlin 454 code_arr => sub { @{$_[0]} },
334 dpavlin 427 filter_code => sub { shift },
335 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
336 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
337     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
338 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
339 dpavlin 427 },{
340     # 8
341 dpavlin 454 code_arr => sub { @{$_[0]} },
342 dpavlin 427 filter_code => sub { shift },
343 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
344 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
345     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
346 dpavlin 460 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
347 dpavlin 424 have_children => sub { 0 },
348 dpavlin 454 },{
349     # 9 - level which is never reached except explicitly
350 dpavlin 471 code_arr => sub { @{$_[0]} },
351 dpavlin 454 filter_code => sub { shift },
352 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
353 dpavlin 454 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
354     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
355     have_children => sub { 0 },
356 dpavlin 471 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
357 dpavlin 454 },{
358 dpavlin 424 });
359 dpavlin 419
360 dpavlin 441 my $tree = new WebPAC::Tree(
361 dpavlin 530 tree => \@tree,
362 dpavlin 573 detail_url => sub {
363     my $mfn = shift;
364 dpavlin 574 my $url = "thes/${mfn}.html";
365     return $url if (-e "./out/$url");
366 dpavlin 573 },
367 dpavlin 575 nodes_dir => 'nodes',
368 dpavlin 530 );
369    
370     $tree->output(
371     dir => './out',
372     html => 'browse.html',
373 dpavlin 572 template_dir => './output_template/',
374     template_tree => 'tree.tt',
375     template_node => 'node.tt',
376 dpavlin 441 js => 'tree-ids.js',
377     );
378 dpavlin 428
379 dpavlin 530 $tree->output(
380     dir => './eurovoc',
381     html => 'hijerarhija.html',
382 dpavlin 572 template_dir => './output_template/',
383     template_tree => 'hijerarhija.tt',
384     template_node => 'hijerarhija_node.tt',
385 dpavlin 530 js => 'tree-ids.js',
386     );
387    
388    
389 dpavlin 428 $log->info("closing index");
390 dpavlin 390 $index->close;
391    
392 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