/[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 501 - (hide annotations)
Sun Oct 10 11:25:10 2004 UTC (14 years, 9 months ago) by dpavlin
File MIME type: text/plain
File size: 8292 byte(s)
implement CROVOC marker for tree

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 dpavlin 500 my $filter = {
36     'CROVOC' => sub {
37     my $tmp = shift || return;
38 dpavlin 501 return undef unless ($tmp =~ s/\s+CROVOC.*$/ */);
39 dpavlin 500 return $tmp;
40     },
41 dpavlin 501 'CROVOC_tree' => sub {
42     my $tmp = shift || return;
43     $tmp =~ s/\s+CROVOC.*$/ */;
44     $tmp =~ s/\s+EUROVOC.*//;
45     return $tmp;
46     },
47 dpavlin 500 };
48    
49 dpavlin 437 # create WebPAC object
50     #
51 dpavlin 354 my $webpac = new WebPAC(
52 dpavlin 437 code_page => $code_page,
53     limit_mfn => $limit_mfn,
54     start_mfn => $start_mfn,
55     debug => $debug,
56     low_mem => $low_mem,
57 dpavlin 500 filter => $filter,
58 dpavlin 352 ) || die;
59    
60 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
61 dpavlin 373
62 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
63    
64 dpavlin 390 my $index = new WebPAC::jsFind(
65 dpavlin 437 index_path => $index_path,
66 dpavlin 413 keys => 10,
67 dpavlin 390 ) || die;
68    
69 dpavlin 410 my $thes;
70    
71 dpavlin 10 $|=1;
72 dpavlin 9
73 dpavlin 352 my $maxmfn = $webpac->open_isis(
74 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
75 dpavlin 352 lookup => [
76 dpavlin 501 { 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' },
77 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
78     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
79     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
80     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
81 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
82 dpavlin 419 # tree structure
83 dpavlin 471 { 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
84 dpavlin 419 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
85 dpavlin 460 { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
86 dpavlin 352 ],
87 dpavlin 54 );
88 dpavlin 3
89 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
90    
91 dpavlin 372 $log->info("rows: $maxmfn");
92 dpavlin 170
93 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
94    
95 dpavlin 425 if(1) { # XXX
96 dpavlin 424
97 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
98 dpavlin 358
99 dpavlin 370 my @ds = $webpac->data_structure($rec);
100 dpavlin 366
101 dpavlin 411 if (0 && $log->is_debug) {
102 dpavlin 454 $log->debug("rec = ",Dump($rec));
103     $log->debug("ds = ",Dump(\@ds));
104 dpavlin 372 }
105    
106 dpavlin 374 next if (! @ds);
107 dpavlin 366
108 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
109 dpavlin 374
110     if ($filename) {
111 dpavlin 411 $webpac->output_file(
112     file => $filename,
113 dpavlin 374 template => 'html.tt',
114     data => \@ds,
115 dpavlin 398 headline => $webpac->{'headline'},
116 dpavlin 374 );
117     } else {
118     print $webpac->output(
119     template => 'text.tt',
120     data => \@ds,
121 dpavlin 398 headline => $webpac->{'headline'},
122 dpavlin 374 );
123     }
124    
125 dpavlin 398 my $headline = $webpac->{'headline'};
126 dpavlin 390
127 dpavlin 392 my $f = $filename;
128     $f =~ s!out/!!;
129    
130 dpavlin 390 # save into index
131     foreach my $ds (@ds) {
132     next if (! $ds->{'swish'});
133    
134 dpavlin 466 # strip all non word characters from beginning or end
135     # of word
136     my $words = join(" ",@{$ds->{'swish'}});
137     $words =~ s/^\W+//;
138     $words =~ s/\W*\s+\W*/ /g;
139     $words =~ s/\W+$//;
140    
141 dpavlin 390 $index->insert(
142     index_name => $ds->{'tag'},
143 dpavlin 466 #path => $f,
144     path => $webpac->mfn,
145 dpavlin 390 headline => $headline,
146 dpavlin 466 words => $words,
147 dpavlin 390 );
148     }
149    
150 dpavlin 410 # save into sorted index (thesaurus)
151     foreach my $ds (@ds) {
152     next if (! $ds->{'index'});
153    
154     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
155    
156 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
157     $thes->{$ds->{'tag'}}->insert(
158     headline => $h,
159 dpavlin 448 mfn => $webpac->mfn,
160 dpavlin 415 );
161     }
162 dpavlin 410 }
163    
164 dpavlin 454 # print Dump(\@ds);
165 dpavlin 390
166 dpavlin 3 }
167    
168 dpavlin 411 foreach my $t (keys %{$thes}) {
169    
170     my @e = $thes->{$t}->elements;
171     if (! @e) {
172     $log->logwarn("no elements in sorted index $t?");
173     next;
174     }
175    
176 dpavlin 415 my $file = "./out/bfilter/$t.txt";
177     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
178 dpavlin 411
179     $webpac->output_file(
180 dpavlin 415 file => $file,
181 dpavlin 411 template => 'index.tt',
182     data => \@e,
183     index_name => $t,
184     );
185     }
186    
187     if (0 && $log->is_debug) {
188 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
189     $log->debug("data hash: ",Dump($webpac->{'data'}));
190 dpavlin 410 foreach my $t (keys %{$thes}) {
191 dpavlin 454 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
192 dpavlin 410 }
193 dpavlin 372 }
194 dpavlin 390
195 dpavlin 424 } # XXX if(0)
196    
197 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
198 dpavlin 420
199 dpavlin 441 $log->info("creating tree");
200    
201 dpavlin 419 #
202 dpavlin 441 # define tree structure
203 dpavlin 419 #
204    
205 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
206 dpavlin 419
207 dpavlin 424 my @tree = ({
208     # level 0
209     code_arr => sub { sort keys %{$l} },
210     filter_code => sub {
211     my $t = shift;
212     return $t if ($t =~ s/root://);
213     },
214     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
215     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
216     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
217 dpavlin 454 have_children => sub { return $l->{"a:".$_[0]."::"} },
218     have_children_at_level => sub {
219 dpavlin 471 return unless (defined($l->{"code:".$_[1]}));
220     my $code = shift @{$l->{"code:".$_[1]}};
221     print STDERR "## $_[1] -> $code\n";
222     return undef unless($code);
223     return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
224 dpavlin 454 },
225 dpavlin 424 style => 'display: none',
226     },{
227     # 1
228 dpavlin 454 code_arr => sub { @{$_[0]} },
229 dpavlin 424 filter_code => sub { shift }, # nop
230     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
231     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
232     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
233 dpavlin 454 have_children => sub { return $l->{"a:".$_[1].":"} },
234 dpavlin 424 style => 'display: none',
235     },{
236     # 2
237 dpavlin 454 code_arr => sub { @{$_[0]} },
238 dpavlin 424 filter_code => sub { shift },
239 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
240 dpavlin 424 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
241     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
242 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
243 dpavlin 454 #style => 'display: none',
244 dpavlin 425 },{
245     # 3 u¾i pojam
246 dpavlin 454 code_arr => sub { @{$_[0]} },
247 dpavlin 425 filter_code => sub { shift },
248 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
249 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
250     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
251 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
252 dpavlin 426 },{
253     # 4
254 dpavlin 454 code_arr => sub { @{$_[0]} },
255 dpavlin 426 filter_code => sub { shift },
256 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
257 dpavlin 426 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
258     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
259 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
260 dpavlin 427 },{
261     # 5
262 dpavlin 454 code_arr => sub { @{$_[0]} },
263 dpavlin 427 filter_code => sub { shift },
264 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
265 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
266     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
267 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
268 dpavlin 427 },{
269     # 6
270 dpavlin 454 code_arr => sub { @{$_[0]} },
271 dpavlin 427 filter_code => sub { shift },
272 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
273 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
274     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
275 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
276 dpavlin 427 },{
277     # 7
278 dpavlin 454 code_arr => sub { @{$_[0]} },
279 dpavlin 427 filter_code => sub { shift },
280 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
281 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
282     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
283 dpavlin 460 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
284 dpavlin 427 },{
285     # 8
286 dpavlin 454 code_arr => sub { @{$_[0]} },
287 dpavlin 427 filter_code => sub { shift },
288 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
289 dpavlin 427 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
290     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
291 dpavlin 460 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
292 dpavlin 424 have_children => sub { 0 },
293 dpavlin 454 },{
294     # 9 - level which is never reached except explicitly
295 dpavlin 471 code_arr => sub { @{$_[0]} },
296 dpavlin 454 filter_code => sub { shift },
297 dpavlin 460 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
298 dpavlin 454 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
299     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
300     have_children => sub { 0 },
301 dpavlin 471 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
302 dpavlin 454 },{
303 dpavlin 424 });
304 dpavlin 419
305 dpavlin 441 my $tree = new WebPAC::Tree(
306     dir => './out',
307     html => 'browse.html',
308     template => './output_template/tree.tt',
309     js => 'tree-ids.js',
310     tree => \@tree,
311     );
312 dpavlin 428
313     $log->info("closing index");
314 dpavlin 390 $index->close;
315    
316 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