/[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 561 - (show annotations)
Sat Oct 30 23:24:54 2004 UTC (19 years, 5 months ago) by dpavlin
File MIME type: text/plain
File size: 9723 byte(s)
make icons in tree also

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