/[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 566 - (show annotations)
Sun Oct 31 18:43:33 2004 UTC (19 years, 5 months ago) by dpavlin
File MIME type: text/plain
File size: 9832 byte(s)
remove HIDRA mark

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 ],
132 );
133
134 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
135
136 $log->info("rows: $maxmfn");
137
138 $webpac->open_import_xml(type => 'isis_hidra_ths');
139
140 if(1) { # XXX
141
142 while (my $rec = $webpac->fetch_rec) {
143
144 my @ds = $webpac->data_structure($rec);
145
146 if (0 && $log->is_debug) {
147 $log->debug("rec = ",Dump($rec));
148 $log->debug("ds = ",Dump(\@ds));
149 }
150
151 next if (! @ds);
152
153 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
154
155 if ($filename) {
156 $webpac->output_file(
157 file => $filename,
158 template => 'html.tt',
159 data => \@ds,
160 headline => $webpac->{'headline'},
161 );
162 } else {
163 print $webpac->output(
164 template => 'text.tt',
165 data => \@ds,
166 headline => $webpac->{'headline'},
167 );
168 }
169
170 my $headline = $webpac->{'headline'};
171
172 my $f = $filename;
173 $f =~ s!out/!!;
174
175 # save into index
176 foreach my $ds (@ds) {
177 next if (! $ds->{'swish'});
178
179 # 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 # first try to generate headline for this entry from index
187 my $h = $ds->{'index'}->[0];
188 # then, from display
189 $h ||= $ds->{'display'}->[0];
190 # and as last resport, fallback to headline
191 $h ||= $headline;
192
193 $index->insert(
194 index_name => $ds->{'tag'},
195 #path => $f,
196 path => $webpac->mfn,
197 headline => $h,
198 words => unac($words),
199 );
200 }
201
202 # save into sorted index (thesaurus)
203 foreach my $ds (@ds) {
204 next if (! $ds->{'index'});
205
206 $thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} );
207
208 foreach my $h (@{$ds->{'index'}}) {
209 $thes->{$ds->{'tag'}}->insert(
210 sort_by => unac_2($h),
211 mfn => $webpac->mfn,
212 headline => $h,
213 );
214 }
215 }
216
217 # print Dump(\@ds);
218
219 }
220
221 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 my $file = "./out/bfilter/$t.txt";
230 $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
231
232 $webpac->output_file(
233 file => $file,
234 template => 'index.tt',
235 data => \@e,
236 index_name => $t,
237 );
238 }
239
240 if (0 && $log->is_debug) {
241 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
242 $log->debug("data hash: ",Dump($webpac->{'data'}));
243 foreach my $t (keys %{$thes}) {
244 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
245 }
246 }
247
248 } # XXX if(0)
249
250 #$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
251
252 $log->info("creating tree");
253
254 #
255 # define tree structure
256 #
257
258 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
259
260 my @tree = ({
261 # level 0
262 code_arr => sub { sort keys %{$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 have_children => sub { return $l->{"a:".$_[0]."::"} },
271 have_children_at_level => sub {
272 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 },
278 style => 'display: none',
279 },{
280 # 1
281 code_arr => sub { @{$_[0]} },
282 filter_code => sub { shift }, # nop
283 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
284 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
285 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
286 have_children => sub { return $l->{"a:".$_[1].":"} },
287 style => 'display: none',
288 },{
289 # 2
290 code_arr => sub { @{$_[0]} },
291 filter_code => sub { shift },
292 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
293 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
294 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
295 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
296 #style => 'display: none',
297 },{
298 # 3 u¾i pojam
299 code_arr => sub { @{$_[0]} },
300 filter_code => sub { shift },
301 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
302 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
303 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
304 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
305 },{
306 # 4
307 code_arr => sub { @{$_[0]} },
308 filter_code => sub { shift },
309 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
310 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
311 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
312 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
313 },{
314 # 5
315 code_arr => sub { @{$_[0]} },
316 filter_code => sub { shift },
317 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
318 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
319 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
320 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
321 },{
322 # 6
323 code_arr => sub { @{$_[0]} },
324 filter_code => sub { shift },
325 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
326 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
327 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
328 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
329 },{
330 # 7
331 code_arr => sub { @{$_[0]} },
332 filter_code => sub { shift },
333 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
334 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
335 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
336 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
337 },{
338 # 8
339 code_arr => sub { @{$_[0]} },
340 filter_code => sub { shift },
341 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
342 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
343 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
344 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
345 have_children => sub { 0 },
346 },{
347 # 9 - level which is never reached except explicitly
348 code_arr => sub { @{$_[0]} },
349 filter_code => sub { shift },
350 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
351 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
352 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
353 have_children => sub { 0 },
354 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
355 },{
356 });
357
358 my $tree = new WebPAC::Tree(
359 tree => \@tree,
360 );
361
362 $tree->output(
363 dir => './out',
364 html => 'browse.html',
365 template => './output_template/tree.tt',
366 js => 'tree-ids.js',
367 );
368
369 $tree->output(
370 dir => './eurovoc',
371 html => 'hijerarhija.html',
372 template => './output_template/hijerarhija.tt',
373 js => 'tree-ids.js',
374 );
375
376
377 $log->info("closing index");
378 $index->close;
379
380 $log->info("elapsed time: ",$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