/[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

Contents of /trunk2/all2all.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 591 - (show 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 #!/usr/bin/perl -w
2
3 =head1 NAME
4
5 all2all.pl - basic script for all WebPAC needs
6
7 =cut
8
9 use strict;
10 use locale;
11 use YAML;
12 use Carp;
13 use Getopt::Long;
14 use Text::Unaccent 1.02;
15
16 use lib './lib';
17 use WebPAC;
18 use WebPAC::jsFind;
19 use WebPAC::Index;
20 use WebPAC::Tree;
21
22 # 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 my $filter = {
37 'CROVOC' => sub {
38 my $tmp = shift || return;
39 $tmp =~ s/\s*HIDRA.*//;
40 return undef unless ($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 return $tmp;
45 },
46 'CROVOC_tree' => sub {
47 my $tmp = shift || return;
48 $tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/;
49 $tmp =~ s/\s*EUROVOC.*//;
50 $tmp =~ s/\s*HIDRA.*//;
51 return $tmp;
52 },
53 # 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 $tmp =~ s/"img\/crovoc.png"/"..\/img\/crovoc.png"/gis;
58 return $tmp;
59 }
60 };
61
62 ## 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 $string = unac_entities($string);
69 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 $string = unac_entities($string);
80 return lc($string);
81 }
82
83 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 # create WebPAC object
95 #
96 my $webpac = new WebPAC(
97 code_page => $code_page,
98 limit_mfn => $limit_mfn,
99 start_mfn => $start_mfn,
100 debug => $debug,
101 low_mem => $low_mem,
102 filter => $filter,
103 ) || die;
104
105 my $log = $webpac->_get_logger() || die "can't get logger";
106
107 $log->debug("creating WebPAC::jsFind object");
108
109 my $index = new WebPAC::jsFind(
110 index_path => $index_path,
111 keys => 62,
112 ) || die;
113
114 my $thes;
115
116 $|=1;
117
118 my $maxmfn = $webpac->open_isis(
119 filename => shift @ARGV || '/data/hidra/THS/THS',
120 lookup => [
121 { 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' },
122 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
123 # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
124 # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
125 { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
126 { 'key' => '900_mfn:v900', 'val' => 'v000' },
127 # tree structure
128 { 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
129 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
130 { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
131 { 'key' => 'crovoc:v900', 'val' => 'filter{CROVOC}v800' },
132 ],
133 );
134
135 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
136
137 $log->info("rows: $maxmfn");
138
139 $webpac->open_import_xml(type => 'isis_hidra_ths');
140
141 if(1) { # XXX
142
143 while (my $rec = $webpac->fetch_rec) {
144
145 my @ds = $webpac->data_structure($rec);
146
147 if (0 && $log->is_debug) {
148 $log->debug("rec = ",Dump($rec));
149 $log->debug("ds = ",Dump(\@ds));
150 }
151
152 next if (! @ds);
153
154 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
155
156 if ($filename) {
157 $webpac->output_file(
158 file => $filename,
159 template => 'html.tt',
160 data => \@ds,
161 headline => $webpac->{'headline'},
162 );
163 } else {
164 print $webpac->output(
165 template => 'text.tt',
166 data => \@ds,
167 headline => $webpac->{'headline'},
168 );
169 }
170
171 my $headline = $webpac->{'headline'};
172
173 my $f = $filename;
174 $f =~ s!out/!!;
175
176 # save into index
177 foreach my $ds (@ds) {
178 next if (! $ds->{'swish'});
179
180 # strip all non word characters from beginning or end
181 # of word
182 my $words = unac(join(" ",@{$ds->{'swish'}}));
183 $words =~ s/^\W+//;
184 $words =~ s/\W*\s+\W*/ /g;
185 $words =~ s/\W+$//;
186
187 # first try to generate headline for this entry from index
188 my $h = $ds->{'index'}->[0];
189 # then, from display
190 $h ||= $ds->{'display'}->[0];
191 # and as last resport, fallback to headline
192 $h ||= $headline;
193
194 $index->insert(
195 index_name => $ds->{'tag'},
196 #path => $f,
197 path => $webpac->mfn,
198 headline => $h,
199 words => $words,
200 );
201 }
202
203 # save into sorted index (thesaurus)
204 foreach my $ds (@ds) {
205 next if (! $ds->{'index'});
206
207 $thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} );
208
209 foreach my $h (@{$ds->{'index'}}) {
210 $thes->{$ds->{'tag'}}->insert(
211 sort_by => unac_2($h),
212 mfn => $webpac->mfn,
213 headline => $h,
214 );
215 }
216 }
217
218 # print Dump(\@ds);
219
220 }
221
222 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 my $file = "./out/bfilter/$t.txt";
231 $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
232
233 $webpac->output_file(
234 file => $file,
235 template => 'index.tt',
236 data => \@e,
237 index_name => $t,
238 );
239 }
240
241 if (0 && $log->is_debug) {
242 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
243 $log->debug("data hash: ",Dump($webpac->{'data'}));
244 foreach my $t (keys %{$thes}) {
245 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
246 }
247 }
248
249 } # XXX if(0)
250
251 #$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
252
253 $log->info("creating tree");
254
255 #
256 # define tree structure
257 #
258
259 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
260
261 my @tree = ({
262 # level 0
263 code_arr => sub { sort keys %{$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 have_children => sub { return $l->{"a:".$_[0]."::"} },
272 have_children_at_level => sub {
273 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 },
279 style => 'display: none',
280 iframe => 1,
281 },{
282 # 1
283 code_arr => sub { @{$_[0]} },
284 filter_code => sub { shift }, # nop
285 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
286 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
287 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
288 have_children => sub { return $l->{"a:".$_[1].":"} },
289 style => 'display: none',
290 },{
291 # 2
292 code_arr => sub { @{$_[0]} },
293 filter_code => sub { shift },
294 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
295 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
296 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
297 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
298 style => 'display: none',
299 },{
300 # 3 u¾i pojam
301 code_arr => sub { @{$_[0]} },
302 filter_code => sub { shift },
303 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
304 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
305 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
306 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
307 },{
308 # 4
309 code_arr => sub { @{$_[0]} },
310 filter_code => sub { shift },
311 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
312 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
313 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
314 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
315 },{
316 # 5
317 code_arr => sub { @{$_[0]} },
318 filter_code => sub { shift },
319 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
320 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
321 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
322 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
323 },{
324 # 6
325 code_arr => sub { @{$_[0]} },
326 filter_code => sub { shift },
327 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
328 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
329 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
330 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
331 },{
332 # 7
333 code_arr => sub { @{$_[0]} },
334 filter_code => sub { shift },
335 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
336 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
337 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
338 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
339 },{
340 # 8
341 code_arr => sub { @{$_[0]} },
342 filter_code => sub { shift },
343 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
344 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
345 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
346 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
347 have_children => sub { 0 },
348 },{
349 # 9 - level which is never reached except explicitly
350 code_arr => sub { @{$_[0]} },
351 filter_code => sub { shift },
352 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
353 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
354 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
355 have_children => sub { 0 },
356 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
357 },{
358 });
359
360 my $tree = new WebPAC::Tree(
361 tree => \@tree,
362 detail_url => sub {
363 my $mfn = shift;
364 my $url = "thes/${mfn}.html";
365 return $url if (-e "./out/$url");
366 },
367 nodes_dir => 'nodes',
368 );
369
370 $tree->output(
371 dir => './out',
372 html => 'browse.html',
373 template_dir => './output_template/',
374 template_tree => 'tree.tt',
375 template_node => 'node.tt',
376 js => 'tree-ids.js',
377 );
378
379 $tree->output(
380 dir => './eurovoc',
381 html => 'hijerarhija.html',
382 template_dir => './output_template/',
383 template_tree => 'hijerarhija.tt',
384 template_node => 'hijerarhija_node.tt',
385 js => 'tree-ids.js',
386 );
387
388
389 $log->info("closing index");
390 $index->close;
391
392 $log->info("elapsed time: ",$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