/[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 490 - (hide annotations)
Sat Oct 9 21:44:25 2004 UTC (19 years, 6 months ago) by dpavlin
File MIME type: text/plain
File size: 7980 byte(s)
use locale so that \W (non-word character) won't eat local characters

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 1
15 dpavlin 352 use lib './lib';
16 dpavlin 354 use WebPAC;
17 dpavlin 390 use WebPAC::jsFind;
18 dpavlin 410 use WebPAC::Index;
19 dpavlin 441 use WebPAC::Tree;
20 dpavlin 348
21 dpavlin 437 # options which can be changed via command line
22     #
23     my $code_page = 'ISO-8859-2';
24     my ($limit_mfn, $start_mfn, $debug, $low_mem);
25     my $index_path = './out/index';
26    
27     my $result = GetOptions(
28     "code_page=s" => \$code_page,
29     "limit_mfn=i" => \$limit_mfn,
30     "start_mfn=i" => \$start_mfn,
31     "debug!" => \$debug,
32     "low_mem!" => \$low_mem,
33     );
34    
35     # create WebPAC object
36     #
37 dpavlin 354 my $webpac = new WebPAC(
38 dpavlin 437 code_page => $code_page,
39     limit_mfn => $limit_mfn,
40     start_mfn => $start_mfn,
41     debug => $debug,
42     low_mem => $low_mem,
43 dpavlin 352 ) || die;
44    
45 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
46 dpavlin 373
47 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
48    
49 dpavlin 390 my $index = new WebPAC::jsFind(
50 dpavlin 437 index_path => $index_path,
51 dpavlin 413 keys => 10,
52 dpavlin 390 ) || die;
53    
54 dpavlin 410 my $thes;
55    
56 dpavlin 10 $|=1;
57 dpavlin 9
58 dpavlin 352 my $maxmfn = $webpac->open_isis(
59 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
60 dpavlin 352 lookup => [
61 dpavlin 348 { 'key' => 'd:v900', 'val' => 'v250^a' },
62 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
63     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
64     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
65     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
66 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
67 dpavlin 419 # tree structure
68 dpavlin 471 { 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
69 dpavlin 419 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
70 dpavlin 460 { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
71 dpavlin 352 ],
72 dpavlin 54 );
73 dpavlin 3
74 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
75    
76 dpavlin 372 $log->info("rows: $maxmfn");
77 dpavlin 170
78 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
79    
80 dpavlin 425 if(1) { # XXX
81 dpavlin 424
82 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
83 dpavlin 358
84 dpavlin 370 my @ds = $webpac->data_structure($rec);
85 dpavlin 366
86 dpavlin 411 if (0 && $log->is_debug) {
87 dpavlin 454 $log->debug("rec = ",Dump($rec));
88     $log->debug("ds = ",Dump(\@ds));
89 dpavlin 372 }
90    
91 dpavlin 374 next if (! @ds);
92 dpavlin 366
93 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
94 dpavlin 374
95     if ($filename) {
96 dpavlin 411 $webpac->output_file(
97     file => $filename,
98 dpavlin 374 template => 'html.tt',
99     data => \@ds,
100 dpavlin 398 headline => $webpac->{'headline'},
101 dpavlin 374 );
102     } else {
103     print $webpac->output(
104     template => 'text.tt',
105     data => \@ds,
106 dpavlin 398 headline => $webpac->{'headline'},
107 dpavlin 374 );
108     }
109    
110 dpavlin 398 my $headline = $webpac->{'headline'};
111 dpavlin 390
112 dpavlin 392 my $f = $filename;
113     $f =~ s!out/!!;
114    
115 dpavlin 390 # save into index
116     foreach my $ds (@ds) {
117     next if (! $ds->{'swish'});
118    
119 dpavlin 466 # strip all non word characters from beginning or end
120     # of word
121     my $words = join(" ",@{$ds->{'swish'}});
122     $words =~ s/^\W+//;
123     $words =~ s/\W*\s+\W*/ /g;
124     $words =~ s/\W+$//;
125    
126 dpavlin 390 $index->insert(
127     index_name => $ds->{'tag'},
128 dpavlin 466 #path => $f,
129     path => $webpac->mfn,
130 dpavlin 390 headline => $headline,
131 dpavlin 466 words => $words,
132 dpavlin 390 );
133     }
134    
135 dpavlin 410 # save into sorted index (thesaurus)
136     foreach my $ds (@ds) {
137     next if (! $ds->{'index'});
138    
139     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
140    
141 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
142     $thes->{$ds->{'tag'}}->insert(
143     headline => $h,
144 dpavlin 448 mfn => $webpac->mfn,
145 dpavlin 415 );
146     }
147 dpavlin 410 }
148    
149 dpavlin 454 # print Dump(\@ds);
150 dpavlin 390
151 dpavlin 3 }
152    
153 dpavlin 411 foreach my $t (keys %{$thes}) {
154    
155     my @e = $thes->{$t}->elements;
156     if (! @e) {
157     $log->logwarn("no elements in sorted index $t?");
158     next;
159     }
160    
161 dpavlin 415 my $file = "./out/bfilter/$t.txt";
162     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
163 dpavlin 411
164     $webpac->output_file(
165 dpavlin 415 file => $file,
166 dpavlin 411 template => 'index.tt',
167     data => \@e,
168     index_name => $t,
169     );
170     }
171    
172     if (0 && $log->is_debug) {
173 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
174     $log->debug("data hash: ",Dump($webpac->{'data'}));
175 dpavlin 410 foreach my $t (keys %{$thes}) {
176 dpavlin 454 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
177 dpavlin 410 }
178 dpavlin 372 }
179 dpavlin 390
180 dpavlin 424 } # XXX if(0)
181    
182 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
183 dpavlin 420
184 dpavlin 441 $log->info("creating tree");
185    
186 dpavlin 419 #
187 dpavlin 441 # define tree structure
188 dpavlin 419 #
189    
190 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
191 dpavlin 419
192 dpavlin 424 my @tree = ({
193     # level 0
194     code_arr => sub { sort keys %{$l} },
195     filter_code => sub {
196     my $t = shift;
197     return $t if ($t =~ s/root://);
198     },
199     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
200     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
201     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
202 dpavlin 454 have_children => sub { return $l->{"a:".$_[0]."::"} },
203     have_children_at_level => sub {
204 dpavlin 471 return unless (defined($l->{"code:".$_[1]}));
205     my $code = shift @{$l->{"code:".$_[1]}};
206     print STDERR "## $_[1] -> $code\n";
207     return undef unless($code);
208     return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
209 dpavlin 454 },
210 dpavlin 424 style => 'display: none',
211     },{
212     # 1
213 dpavlin 454 code_arr => sub { @{$_[0]} },
214 dpavlin 424 filter_code => sub { shift }, # nop
215     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
216     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
217     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
218 dpavlin 454 have_children => sub { return $l->{"a:".$_[1].":"} },
219 dpavlin 424 style => 'display: none',
220     },{
221     # 2
222 dpavlin 454 code_arr => sub { @{$_[0]} },
223 dpavlin 424 filter_code => sub { shift },
224 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
225 dpavlin 424 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
226     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
227 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
228 dpavlin 454 #style => 'display: none',
229 dpavlin 425 },{
230     # 3 u¾i pojam
231 dpavlin 454 code_arr => sub { @{$_[0]} },
232 dpavlin 425 filter_code => sub { shift },
233 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
234 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
235     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
236 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
237 dpavlin 426 },{
238     # 4
239 dpavlin 454 code_arr => sub { @{$_[0]} },
240 dpavlin 426 filter_code => sub { shift },
241 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
242 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
243     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
244 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
245 dpavlin 427 },{
246     # 5
247 dpavlin 454 code_arr => sub { @{$_[0]} },
248 dpavlin 427 filter_code => sub { shift },
249 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
250 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
251     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
252 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
253 dpavlin 427 },{
254     # 6
255 dpavlin 454 code_arr => sub { @{$_[0]} },
256 dpavlin 427 filter_code => sub { shift },
257 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
258 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
259     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
260 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
261 dpavlin 427 },{
262     # 7
263 dpavlin 454 code_arr => sub { @{$_[0]} },
264 dpavlin 427 filter_code => sub { shift },
265 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
266 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
267     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
268 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
269 dpavlin 427 },{
270     # 8
271 dpavlin 454 code_arr => sub { @{$_[0]} },
272 dpavlin 427 filter_code => sub { shift },
273 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
274 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
275     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
276 dpavlin 460 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
277 dpavlin 424 have_children => sub { 0 },
278 dpavlin 454 },{
279     # 9 - level which is never reached except explicitly
280 dpavlin 471 code_arr => sub { @{$_[0]} },
281 dpavlin 454 filter_code => sub { shift },
282 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
283 dpavlin 454 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
284     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
285     have_children => sub { 0 },
286 dpavlin 471 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
287 dpavlin 454 },{
288 dpavlin 424 });
289 dpavlin 419
290 dpavlin 441 my $tree = new WebPAC::Tree(
291     dir => './out',
292     html => 'browse.html',
293     template => './output_template/tree.tt',
294     js => 'tree-ids.js',
295     tree => \@tree,
296     );
297 dpavlin 428
298     $log->info("closing index");
299 dpavlin 390 $index->close;
300    
301 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