/[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 437 - (hide annotations)
Mon Sep 13 15:32:55 2004 UTC (19 years, 6 months ago) by dpavlin
File MIME type: text/plain
File size: 11513 byte(s)
command-line options

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     use Data::Dumper;
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 348
19 dpavlin 437 # options which can be changed via command line
20     #
21     my $code_page = 'ISO-8859-2';
22     my ($limit_mfn, $start_mfn, $debug, $low_mem);
23     my $index_path = './out/index';
24    
25     my $result = GetOptions(
26     "code_page=s" => \$code_page,
27     "limit_mfn=i" => \$limit_mfn,
28     "start_mfn=i" => \$start_mfn,
29     "debug!" => \$debug,
30     "low_mem!" => \$low_mem,
31     );
32    
33     # create WebPAC object
34     #
35 dpavlin 354 my $webpac = new WebPAC(
36 dpavlin 437 code_page => $code_page,
37     limit_mfn => $limit_mfn,
38     start_mfn => $start_mfn,
39     debug => $debug,
40     low_mem => $low_mem,
41 dpavlin 352 ) || die;
42    
43 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
44 dpavlin 373
45 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
46    
47 dpavlin 390 my $index = new WebPAC::jsFind(
48 dpavlin 437 index_path => $index_path,
49 dpavlin 413 keys => 10,
50 dpavlin 390 ) || die;
51    
52 dpavlin 410 my $thes;
53    
54 dpavlin 10 $|=1;
55 dpavlin 9
56 dpavlin 352 my $maxmfn = $webpac->open_isis(
57 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
58 dpavlin 352 lookup => [
59 dpavlin 348 { 'key' => 'd:v900', 'val' => 'v250^a' },
60 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
61     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
62     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
63     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
64 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
65 dpavlin 419 # tree structure
66     { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
67     { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
68 dpavlin 352 ],
69 dpavlin 54 );
70 dpavlin 3
71 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
72    
73 dpavlin 372 $log->info("rows: $maxmfn");
74 dpavlin 170
75 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
76    
77 dpavlin 425 if(1) { # XXX
78 dpavlin 424
79 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
80 dpavlin 358
81 dpavlin 370 my @ds = $webpac->data_structure($rec);
82 dpavlin 366
83 dpavlin 411 if (0 && $log->is_debug) {
84 dpavlin 372 $log->debug("rec = ",Dumper($rec));
85     $log->debug("ds = ",Dumper(\@ds));
86     }
87    
88 dpavlin 374 next if (! @ds);
89 dpavlin 366
90 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
91 dpavlin 374
92     if ($filename) {
93 dpavlin 411 $webpac->output_file(
94     file => $filename,
95 dpavlin 374 template => 'html.tt',
96     data => \@ds,
97 dpavlin 398 headline => $webpac->{'headline'},
98 dpavlin 374 );
99     } else {
100     print $webpac->output(
101     template => 'text.tt',
102     data => \@ds,
103 dpavlin 398 headline => $webpac->{'headline'},
104 dpavlin 374 );
105     }
106    
107 dpavlin 398 my $headline = $webpac->{'headline'};
108 dpavlin 390
109 dpavlin 392 my $f = $filename;
110     $f =~ s!out/!!;
111    
112 dpavlin 390 # save into index
113     foreach my $ds (@ds) {
114     next if (! $ds->{'swish'});
115    
116     $index->insert(
117     index_name => $ds->{'tag'},
118 dpavlin 392 path => $f,
119 dpavlin 390 headline => $headline,
120     words => join(" ",@{$ds->{'swish'}})
121     );
122     }
123    
124 dpavlin 410 # save into sorted index (thesaurus)
125     foreach my $ds (@ds) {
126     next if (! $ds->{'index'});
127    
128     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
129    
130 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
131     $thes->{$ds->{'tag'}}->insert(
132     path => $f,
133     headline => $h,
134     );
135     }
136 dpavlin 410 }
137    
138 dpavlin 390 # print Dumper(\@ds);
139    
140 dpavlin 3 }
141    
142 dpavlin 411 foreach my $t (keys %{$thes}) {
143    
144     my @e = $thes->{$t}->elements;
145     if (! @e) {
146     $log->logwarn("no elements in sorted index $t?");
147     next;
148     }
149    
150 dpavlin 415 my $file = "./out/bfilter/$t.txt";
151     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
152 dpavlin 411
153     $webpac->output_file(
154 dpavlin 415 file => $file,
155 dpavlin 411 template => 'index.tt',
156     data => \@e,
157     index_name => $t,
158     );
159     }
160    
161     if (0 && $log->is_debug) {
162 dpavlin 372 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
163     $log->debug("data hash: ",Dumper($webpac->{'data'}));
164 dpavlin 410 foreach my $t (keys %{$thes}) {
165     $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
166     }
167 dpavlin 372 }
168 dpavlin 390
169 dpavlin 424 } # XXX if(0)
170    
171 dpavlin 428 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
172 dpavlin 420
173 dpavlin 419 #
174     # tree dump implementation which shouldn't be here :-)
175     #
176    
177 dpavlin 428
178 dpavlin 419 my $file = 'out/browse.html';
179 dpavlin 428 my $js_url = 'tree-ids.js';
180    
181     $log->info("creating '$file' for tree html");
182 dpavlin 419 open(HTML, "> $file") || $log->logdie("can't open '$file'");
183    
184     print HTML qq{
185     <html>
186     <head>
187     <title>Browse</title>
188     <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/>
189 dpavlin 423 <script type="text/javascript" src="bfilter/hide.js"></script>
190 dpavlin 428 <script type="text/javascript" src="$js_url"></script>
191 dpavlin 419 </head>
192 dpavlin 428
193 dpavlin 429 <!--
194 dpavlin 428 <body onLoad="show_hide_display(show,hide);">
195 dpavlin 429 -->
196     <body>
197 dpavlin 428
198 dpavlin 432 <span style="background: #e0e0e0;">
199     <a href="search.html">search</a> |
200     <a href="thesaurus.html">thesarus</a> |
201     <a href="browse.html"><b>browse</b></a>
202     </span>
203    
204 dpavlin 430 <div style="float: right; width: 10em; text-align: center; margin: 0.5m; background: #e0e0e0; border: 1px dashed #c0c0c0; z-index: 1;">
205 dpavlin 428 Folding:
206 dpavlin 430 <br/>
207 dpavlin 433 <a href="#" onClick="back_display();">&laquo;</a>&nbsp;<a href="#" onClick="default_display();">default</a>&nbsp;<a href="#" onClick="forward_display();">&raquo;</a>
208 dpavlin 430 <br/>
209     <a href="#" onClick="show_hide_display(show,hide);">reset</a>
210 dpavlin 428 <a href="#" onClick="show_display(show); show_display(hide);">all</a>
211     <a href="#" onClick="hide_display(hide); hide_display(show);">none</a>
212 dpavlin 430 <br/>
213     <a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length+', changed:'+changed_display_ids.length+', positin: '+changed_display_pos);">debug</a>
214 dpavlin 432
215     <a href="#" onClick="outline_display('mfn6030');">outline</a>
216 dpavlin 428 </div>
217    
218 dpavlin 419 <ul>
219     };
220    
221 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
222 dpavlin 419
223 dpavlin 424 my @tree = ({
224     # level 0
225     code_arr => sub { sort keys %{$l} },
226     filter_code => sub {
227     my $t = shift;
228     return $t if ($t =~ s/root://);
229     },
230     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
231     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
232     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
233 dpavlin 426 have_children => sub { defined($l->{"a:".$_[0]."::"}) },
234 dpavlin 424 child_code => sub { return $_[0] },
235     style => 'display: none',
236     },{
237     # 1
238     code_arr => sub { @{$l->{"a:".$_[0]."::"}} },
239     filter_code => sub { shift }, # nop
240     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
241     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
242     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
243 dpavlin 426 have_children => sub { defined($l->{"a:".$_[1].":"}) },
244 dpavlin 424 child_code => sub { return $_[1] },
245     style => 'display: none',
246     },{
247     # 2
248     code_arr => sub { @{$l->{"a:".$_[0].":"}} },
249     filter_code => sub { shift },
250     lookup_v900 => sub { shift },
251     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
252     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
253 dpavlin 426 have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) },
254 dpavlin 425 child_code => sub { return "a:".$_[2].":".$_[1] },
255     style => 'display: none',
256     },{
257     # 3 u¾i pojam
258     code_arr => sub { @{$l->{$_[0]}} },
259     filter_code => sub { shift },
260 dpavlin 426 lookup_v900 => sub {
261     my ($c,$p) = @_;
262     $p =~ s/^a:(..:....):.*$/$1/;
263     return "a:".$p.":".$c;
264     },
265     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
266     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
267     have_children => sub { defined($l->{$_[1]}) },
268     child_code => sub { return $_[1] },
269     },{
270     # 4
271     code_arr => sub { @{$l->{$_[0]}} },
272     filter_code => sub { shift },
273     lookup_v900 => sub {
274     my ($c,$p) = @_;
275     $p =~ s/^a:(..:....):.*$/$1/;
276     return "a:".$p.":".$c;
277     },
278     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
279     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
280 dpavlin 427 have_children => sub { defined($l->{$_[1]}) },
281     child_code => sub { return $_[1] },
282     },{
283     # 5
284     code_arr => sub { @{$l->{$_[0]}} },
285     filter_code => sub { shift },
286     lookup_v900 => sub {
287     my ($c,$p) = @_;
288     $p =~ s/^a:(..:....):.*$/$1/;
289     return "a:".$p.":".$c;
290     },
291     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
292     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
293     have_children => sub { defined($l->{$_[1]}) },
294     child_code => sub { return $_[1] },
295     },{
296     # 6
297     code_arr => sub { @{$l->{$_[0]}} },
298     filter_code => sub { shift },
299     lookup_v900 => sub {
300     my ($c,$p) = @_;
301     $p =~ s/^a:(..:....):.*$/$1/;
302     return "a:".$p.":".$c;
303     },
304     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
305     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
306     have_children => sub { defined($l->{$_[1]}) },
307     child_code => sub { return $_[1] },
308     },{
309     # 7
310     code_arr => sub { @{$l->{$_[0]}} },
311     filter_code => sub { shift },
312     lookup_v900 => sub {
313     my ($c,$p) = @_;
314     $p =~ s/^a:(..:....):.*$/$1/;
315     return "a:".$p.":".$c;
316     },
317     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
318     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
319     have_children => sub { defined($l->{$_[1]}) },
320     child_code => sub { return $_[1] },
321     },{
322     # 8
323     code_arr => sub { @{$l->{$_[0]}} },
324     filter_code => sub { shift },
325     lookup_v900 => sub {
326     my ($c,$p) = @_;
327     $p =~ s/^a:(..:....):.*$/$1/;
328     return "a:".$p.":".$c;
329     },
330     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
331     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
332 dpavlin 426 # have_children => sub { defined($l->{$_[1]}) },
333     # child_code => sub { return $_[1] },
334 dpavlin 424 have_children => sub { 0 },
335     child_code => sub { 0 },
336     });
337 dpavlin 419
338 dpavlin 428 my @show_ids;
339     my @hide_ids;
340    
341 dpavlin 424 unroll(0,'');
342 dpavlin 419
343 dpavlin 424 $log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99"));
344 dpavlin 419
345 dpavlin 424 sub unroll {
346     my ($level,$start_code) = @_;
347 dpavlin 419
348 dpavlin 424 $log->logconfess("need level") unless (defined($level));
349 dpavlin 419
350 dpavlin 424 # all levels passed?
351     return if (! defined($tree[$level]));
352 dpavlin 419
353 dpavlin 424 $log->debug("unroll level $level, start code $start_code");
354 dpavlin 419
355 dpavlin 424 foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) {
356 dpavlin 419
357 dpavlin 424 if ($code = $tree[$level]->{'filter_code'}->($code)) {
358    
359     $log->debug("# $level filter passed code $code");
360 dpavlin 419
361 dpavlin 426 my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)");
362     $log->debug("# $level lookup_v900($code,$start_code) = $v900");
363    
364     my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)");
365 dpavlin 424 $log->debug("# $level lookup_term($code,$v900) = $term");
366 dpavlin 426
367     my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)");
368 dpavlin 424 $log->debug("# $level lookup_mfn($code,$v900) = $mfn");
369 dpavlin 419
370 dpavlin 424 $log->debug("$code -> $v900 : $term [$mfn]");
371 dpavlin 419
372 dpavlin 426 my ($link_start,$link_end) = ('','');
373 dpavlin 425
374     my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code);
375 dpavlin 426 if ($have_children) {
376 dpavlin 433 ($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('id$mfn');">},qq{</a>});
377 dpavlin 426 } else {
378     $log->debug("# $level doesn't have_children($code,$v900,$start_code)");
379     }
380 dpavlin 419
381 dpavlin 425 my $mfn_link = "thes/$mfn.html";
382     if (-e "out/$mfn_link") {
383     print HTML " " x $level .
384     qq{<li>${link_start}${term}${link_end}}.
385     qq{&nbsp;<a href="$mfn_link">&raquo;</a></li>\n};
386     } else {
387     $log->warn("file 'out/$mfn_link' doesn't exist, skipping");
388     }
389    
390     unless ($have_children) {
391 dpavlin 424 next;
392 dpavlin 419 }
393 dpavlin 428 my $style = $tree[$level]->{'style'};
394    
395 dpavlin 424 print HTML " " x $level .
396 dpavlin 433 qq{<a name="mfn$mfn"></a>\n <ul id="id$mfn"}.
397 dpavlin 429 ($style ? ' style="'.$style.'"' : '').
398 dpavlin 424 qq{>\n};
399 dpavlin 419
400 dpavlin 428 if ($style) {
401     if ($style =~ m/display\s*:\s*none/i) {
402 dpavlin 433 push @hide_ids, "id$mfn";
403 dpavlin 428 } else {
404 dpavlin 433 push @show_ids, "id$mfn";
405 dpavlin 428 }
406     } else {
407     # default: show
408 dpavlin 433 push @show_ids, "id$mfn";
409 dpavlin 428 }
410    
411 dpavlin 425 unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code));
412 dpavlin 424
413     print HTML " " x $level . qq{</ul>\n};
414    
415 dpavlin 419 }
416     }
417     }
418    
419     print HTML qq{
420     </ul>
421     </body>
422     </html>
423     };
424    
425 dpavlin 425 close(HTML);
426    
427 dpavlin 428
428     my $js_file = "out/$js_url";
429     $log->info("creating '$js_file' with arrays of shown and hidden ids");
430     open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!");
431     print JS "var show = ['",join("','",@show_ids),"'];\n";
432     print JS "var hide = ['",join("','",@hide_ids),"'];\n";
433     close(JS);
434    
435     $log->info("closing index");
436 dpavlin 390 $index->close;
437    
438 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