/[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 562 - (show annotations)
Sat Oct 30 23:56:57 2004 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 9780 byte(s)
fix patch to icon image

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