/[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 454 - (hide annotations)
Wed Sep 15 22:36:23 2004 UTC (19 years, 7 months ago) by dpavlin
File MIME type: text/plain
File size: 7768 byte(s)
new simplier implementation of WebPAC::Tree with support for recursion
(still buggy)

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 454 use YAML;
11 dpavlin 352 use Carp;
12 dpavlin 437 use Getopt::Long;
13 dpavlin 1
14 dpavlin 352 use lib './lib';
15 dpavlin 354 use WebPAC;
16 dpavlin 390 use WebPAC::jsFind;
17 dpavlin 410 use WebPAC::Index;
18 dpavlin 441 use WebPAC::Tree;
19 dpavlin 348
20 dpavlin 437 # options which can be changed via command line
21     #
22     my $code_page = 'ISO-8859-2';
23     my ($limit_mfn, $start_mfn, $debug, $low_mem);
24     my $index_path = './out/index';
25    
26     my $result = GetOptions(
27     "code_page=s" => \$code_page,
28     "limit_mfn=i" => \$limit_mfn,
29     "start_mfn=i" => \$start_mfn,
30     "debug!" => \$debug,
31     "low_mem!" => \$low_mem,
32     );
33    
34     # create WebPAC object
35     #
36 dpavlin 354 my $webpac = new WebPAC(
37 dpavlin 437 code_page => $code_page,
38     limit_mfn => $limit_mfn,
39     start_mfn => $start_mfn,
40     debug => $debug,
41     low_mem => $low_mem,
42 dpavlin 352 ) || die;
43    
44 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
45 dpavlin 373
46 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
47    
48 dpavlin 390 my $index = new WebPAC::jsFind(
49 dpavlin 437 index_path => $index_path,
50 dpavlin 413 keys => 10,
51 dpavlin 390 ) || die;
52    
53 dpavlin 410 my $thes;
54    
55 dpavlin 10 $|=1;
56 dpavlin 9
57 dpavlin 352 my $maxmfn = $webpac->open_isis(
58 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
59 dpavlin 352 lookup => [
60 dpavlin 348 { 'key' => 'd:v900', 'val' => 'v250^a' },
61 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
62     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
63     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
64     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
65 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
66 dpavlin 419 # tree structure
67     { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
68     { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
69 dpavlin 352 ],
70 dpavlin 54 );
71 dpavlin 3
72 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
73    
74 dpavlin 372 $log->info("rows: $maxmfn");
75 dpavlin 170
76 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
77    
78 dpavlin 425 if(1) { # XXX
79 dpavlin 424
80 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
81 dpavlin 358
82 dpavlin 370 my @ds = $webpac->data_structure($rec);
83 dpavlin 366
84 dpavlin 411 if (0 && $log->is_debug) {
85 dpavlin 454 $log->debug("rec = ",Dump($rec));
86     $log->debug("ds = ",Dump(\@ds));
87 dpavlin 372 }
88    
89 dpavlin 374 next if (! @ds);
90 dpavlin 366
91 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
92 dpavlin 374
93     if ($filename) {
94 dpavlin 411 $webpac->output_file(
95     file => $filename,
96 dpavlin 374 template => 'html.tt',
97     data => \@ds,
98 dpavlin 398 headline => $webpac->{'headline'},
99 dpavlin 374 );
100     } else {
101     print $webpac->output(
102     template => 'text.tt',
103     data => \@ds,
104 dpavlin 398 headline => $webpac->{'headline'},
105 dpavlin 374 );
106     }
107    
108 dpavlin 398 my $headline = $webpac->{'headline'};
109 dpavlin 390
110 dpavlin 392 my $f = $filename;
111     $f =~ s!out/!!;
112    
113 dpavlin 390 # save into index
114     foreach my $ds (@ds) {
115     next if (! $ds->{'swish'});
116    
117     $index->insert(
118     index_name => $ds->{'tag'},
119 dpavlin 392 path => $f,
120 dpavlin 390 headline => $headline,
121     words => join(" ",@{$ds->{'swish'}})
122     );
123     }
124    
125 dpavlin 410 # save into sorted index (thesaurus)
126     foreach my $ds (@ds) {
127     next if (! $ds->{'index'});
128    
129     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
130    
131 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
132     $thes->{$ds->{'tag'}}->insert(
133     headline => $h,
134 dpavlin 448 mfn => $webpac->mfn,
135 dpavlin 415 );
136     }
137 dpavlin 410 }
138    
139 dpavlin 454 # print Dump(\@ds);
140 dpavlin 390
141 dpavlin 3 }
142    
143 dpavlin 411 foreach my $t (keys %{$thes}) {
144    
145     my @e = $thes->{$t}->elements;
146     if (! @e) {
147     $log->logwarn("no elements in sorted index $t?");
148     next;
149     }
150    
151 dpavlin 415 my $file = "./out/bfilter/$t.txt";
152     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
153 dpavlin 411
154     $webpac->output_file(
155 dpavlin 415 file => $file,
156 dpavlin 411 template => 'index.tt',
157     data => \@e,
158     index_name => $t,
159     );
160     }
161    
162     if (0 && $log->is_debug) {
163 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
164     $log->debug("data hash: ",Dump($webpac->{'data'}));
165 dpavlin 410 foreach my $t (keys %{$thes}) {
166 dpavlin 454 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
167 dpavlin 410 }
168 dpavlin 372 }
169 dpavlin 390
170 dpavlin 424 } # XXX if(0)
171    
172 dpavlin 454 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
173 dpavlin 420
174 dpavlin 441 $log->info("creating tree");
175    
176 dpavlin 419 #
177 dpavlin 441 # define tree structure
178 dpavlin 419 #
179    
180 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
181 dpavlin 419
182 dpavlin 424 my @tree = ({
183     # level 0
184     code_arr => sub { sort keys %{$l} },
185     filter_code => sub {
186     my $t = shift;
187     return $t if ($t =~ s/root://);
188     },
189     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
190     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
191     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
192 dpavlin 454 have_children => sub { return $l->{"a:".$_[0]."::"} },
193     have_children_at_level => sub {
194     my $l2 = shift;
195     my $l1 = substr($l2,0,1);
196     return(9, "a:$l1:$l2:" ) if ($l->{"a:$l1:$l2"});
197     },
198 dpavlin 424 style => 'display: none',
199     },{
200     # 1
201 dpavlin 454 code_arr => sub { @{$_[0]} },
202 dpavlin 424 filter_code => sub { shift }, # nop
203     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
204     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
205     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
206 dpavlin 454 have_children => sub { return $l->{"a:".$_[1].":"} },
207 dpavlin 424 style => 'display: none',
208     },{
209     # 2
210 dpavlin 454 code_arr => sub { @{$_[0]} },
211 dpavlin 424 filter_code => sub { shift },
212     lookup_v900 => sub { shift },
213     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
214     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
215 dpavlin 454 have_children => sub { return $l->{"a:".$_[2].":".$_[1]} },
216     #style => 'display: none',
217 dpavlin 425 },{
218     # 3 u¾i pojam
219 dpavlin 454 code_arr => sub { @{$_[0]} },
220 dpavlin 425 filter_code => sub { shift },
221 dpavlin 426 lookup_v900 => sub {
222     my ($c,$p) = @_;
223     $p =~ s/^a:(..:....):.*$/$1/;
224     return "a:".$p.":".$c;
225     },
226     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
227     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
228 dpavlin 454 have_children => sub { return $l->{$_[1]} },
229 dpavlin 426 },{
230     # 4
231 dpavlin 454 code_arr => sub { @{$_[0]} },
232 dpavlin 426 filter_code => sub { shift },
233     lookup_v900 => sub {
234     my ($c,$p) = @_;
235     $p =~ s/^a:(..:....):.*$/$1/;
236     return "a:".$p.":".$c;
237     },
238     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
239     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
240 dpavlin 454 have_children => sub { return $l->{$_[1]} },
241 dpavlin 427 },{
242     # 5
243 dpavlin 454 code_arr => sub { @{$_[0]} },
244 dpavlin 427 filter_code => sub { shift },
245     lookup_v900 => sub {
246     my ($c,$p) = @_;
247     $p =~ s/^a:(..:....):.*$/$1/;
248     return "a:".$p.":".$c;
249     },
250     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
251     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
252 dpavlin 454 have_children => sub { return $l->{$_[1]} },
253 dpavlin 427 },{
254     # 6
255 dpavlin 454 code_arr => sub { @{$_[0]} },
256 dpavlin 427 filter_code => sub { shift },
257     lookup_v900 => sub {
258     my ($c,$p) = @_;
259     $p =~ s/^a:(..:....):.*$/$1/;
260     return "a:".$p.":".$c;
261     },
262     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
263     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
264 dpavlin 454 have_children => sub { return $l->{$_[1]} },
265 dpavlin 427 },{
266     # 7
267 dpavlin 454 code_arr => sub { @{$_[0]} },
268 dpavlin 427 filter_code => sub { shift },
269     lookup_v900 => sub {
270     my ($c,$p) = @_;
271     $p =~ s/^a:(..:....):.*$/$1/;
272     return "a:".$p.":".$c;
273     },
274     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
275     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
276 dpavlin 454 have_children => sub { return $l->{$_[1]} },
277 dpavlin 427 },{
278     # 8
279 dpavlin 454 code_arr => sub { @{$_[0]} },
280 dpavlin 427 filter_code => sub { shift },
281     lookup_v900 => sub {
282     my ($c,$p) = @_;
283     $p =~ s/^a:(..:....):.*$/$1/;
284     return "a:".$p.":".$c;
285     },
286     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
287     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
288 dpavlin 454 # have_children => sub { return $l->{$_[1]} },
289 dpavlin 426 # child_code => sub { return $_[1] },
290 dpavlin 424 have_children => sub { 0 },
291 dpavlin 454 },{
292     # 9 - level which is never reached except explicitly
293     code_arr => sub { @{$l->{$_[0]}} },
294     filter_code => sub { shift },
295     lookup_v900 => sub {
296     my ($c,$p) = @_;
297     $p =~ s/^a:(.+:.+):.*$/$1/;
298     return "a:".$p.":".$c;
299     },
300     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
301     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
302     have_children => sub { 0 },
303     have_children_at_level => sub { defined($l->{$_[1]}) && return (9,$_[1]) },
304     },{
305 dpavlin 424 });
306 dpavlin 419
307 dpavlin 441 my $tree = new WebPAC::Tree(
308     dir => './out',
309     html => 'browse.html',
310     template => './output_template/tree.tt',
311     js => 'tree-ids.js',
312     tree => \@tree,
313     );
314 dpavlin 428
315     $log->info("closing index");
316 dpavlin 390 $index->close;
317    
318 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