/[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 432 - (hide annotations)
Sun Sep 12 20:36:33 2004 UTC (15 years ago) by dpavlin
File MIME type: text/plain
File size: 11132 byte(s)
navigation

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 1
13 dpavlin 352 use lib './lib';
14 dpavlin 354 use WebPAC;
15 dpavlin 390 use WebPAC::jsFind;
16 dpavlin 410 use WebPAC::Index;
17 dpavlin 348
18 dpavlin 354 my $webpac = new WebPAC(
19 dpavlin 352 code_page => 'ISO-8859-2',
20 dpavlin 423 limit_mfn => 500,
21 dpavlin 431 start_mfn => 6000,
22 dpavlin 372 # debug => 1,
23 dpavlin 423 # low_mem => 1,
24 dpavlin 352 ) || die;
25    
26 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
27 dpavlin 373
28 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
29    
30 dpavlin 390 my $index = new WebPAC::jsFind(
31     index_path => './out/index',
32 dpavlin 413 keys => 10,
33 dpavlin 390 ) || die;
34    
35 dpavlin 410 my $thes;
36    
37 dpavlin 10 $|=1;
38 dpavlin 9
39 dpavlin 352 my $maxmfn = $webpac->open_isis(
40 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
41 dpavlin 352 lookup => [
42 dpavlin 348 { 'key' => 'd:v900', 'val' => 'v250^a' },
43 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
44     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
45     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
46     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
47 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
48 dpavlin 419 # tree structure
49     { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
50     { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
51 dpavlin 352 ],
52 dpavlin 54 );
53 dpavlin 3
54 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
55    
56 dpavlin 372 $log->info("rows: $maxmfn");
57 dpavlin 170
58 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
59    
60 dpavlin 425 if(1) { # XXX
61 dpavlin 424
62 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
63 dpavlin 358
64 dpavlin 370 my @ds = $webpac->data_structure($rec);
65 dpavlin 366
66 dpavlin 411 if (0 && $log->is_debug) {
67 dpavlin 372 $log->debug("rec = ",Dumper($rec));
68     $log->debug("ds = ",Dumper(\@ds));
69     }
70    
71 dpavlin 374 next if (! @ds);
72 dpavlin 366
73 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
74 dpavlin 374
75     if ($filename) {
76 dpavlin 411 $webpac->output_file(
77     file => $filename,
78 dpavlin 374 template => 'html.tt',
79     data => \@ds,
80 dpavlin 398 headline => $webpac->{'headline'},
81 dpavlin 374 );
82     } else {
83     print $webpac->output(
84     template => 'text.tt',
85     data => \@ds,
86 dpavlin 398 headline => $webpac->{'headline'},
87 dpavlin 374 );
88     }
89    
90 dpavlin 398 my $headline = $webpac->{'headline'};
91 dpavlin 390
92 dpavlin 392 my $f = $filename;
93     $f =~ s!out/!!;
94    
95 dpavlin 390 # save into index
96     foreach my $ds (@ds) {
97     next if (! $ds->{'swish'});
98    
99     $index->insert(
100     index_name => $ds->{'tag'},
101 dpavlin 392 path => $f,
102 dpavlin 390 headline => $headline,
103     words => join(" ",@{$ds->{'swish'}})
104     );
105     }
106    
107 dpavlin 410 # save into sorted index (thesaurus)
108     foreach my $ds (@ds) {
109     next if (! $ds->{'index'});
110    
111     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
112    
113 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
114     $thes->{$ds->{'tag'}}->insert(
115     path => $f,
116     headline => $h,
117     );
118     }
119 dpavlin 410 }
120    
121 dpavlin 390 # print Dumper(\@ds);
122    
123 dpavlin 3 }
124    
125 dpavlin 411 foreach my $t (keys %{$thes}) {
126    
127     my @e = $thes->{$t}->elements;
128     if (! @e) {
129     $log->logwarn("no elements in sorted index $t?");
130     next;
131     }
132    
133 dpavlin 415 my $file = "./out/bfilter/$t.txt";
134     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
135 dpavlin 411
136     $webpac->output_file(
137 dpavlin 415 file => $file,
138 dpavlin 411 template => 'index.tt',
139     data => \@e,
140     index_name => $t,
141     );
142     }
143    
144     if (0 && $log->is_debug) {
145 dpavlin 372 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
146     $log->debug("data hash: ",Dumper($webpac->{'data'}));
147 dpavlin 410 foreach my $t (keys %{$thes}) {
148     $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
149     }
150 dpavlin 372 }
151 dpavlin 390
152 dpavlin 424 } # XXX if(0)
153    
154 dpavlin 428 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
155 dpavlin 420
156 dpavlin 419 #
157     # tree dump implementation which shouldn't be here :-)
158     #
159    
160 dpavlin 428
161 dpavlin 419 my $file = 'out/browse.html';
162 dpavlin 428 my $js_url = 'tree-ids.js';
163    
164     $log->info("creating '$file' for tree html");
165 dpavlin 419 open(HTML, "> $file") || $log->logdie("can't open '$file'");
166    
167     print HTML qq{
168     <html>
169     <head>
170     <title>Browse</title>
171     <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/>
172 dpavlin 423 <script type="text/javascript" src="bfilter/hide.js"></script>
173 dpavlin 428 <script type="text/javascript" src="$js_url"></script>
174 dpavlin 419 </head>
175 dpavlin 428
176 dpavlin 429 <!--
177 dpavlin 428 <body onLoad="show_hide_display(show,hide);">
178 dpavlin 429 -->
179     <body>
180 dpavlin 428
181 dpavlin 432 <span style="background: #e0e0e0;">
182     <a href="search.html">search</a> |
183     <a href="thesaurus.html">thesarus</a> |
184     <a href="browse.html"><b>browse</b></a>
185     </span>
186    
187 dpavlin 430 <div style="float: right; width: 10em; text-align: center; margin: 0.5m; background: #e0e0e0; border: 1px dashed #c0c0c0; z-index: 1;">
188 dpavlin 428 Folding:
189 dpavlin 430 <br/>
190     <a href="#" onClick="back_display();">&laquo;</a>&nbsp;<a href="#" onClick="apply_changed_display();">default</a>&nbsp;<a href="#" onClick="forward_display();">&raquo;</a>
191     <br/>
192     <a href="#" onClick="show_hide_display(show,hide);">reset</a>
193 dpavlin 428 <a href="#" onClick="show_display(show); show_display(hide);">all</a>
194     <a href="#" onClick="hide_display(hide); hide_display(show);">none</a>
195 dpavlin 430 <br/>
196     <a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length+', changed:'+changed_display_ids.length+', positin: '+changed_display_pos);">debug</a>
197 dpavlin 432
198     <a href="#" onClick="outline_display('mfn6030');">outline</a>
199 dpavlin 428 </div>
200    
201 dpavlin 419 <ul>
202     };
203    
204 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
205 dpavlin 419
206 dpavlin 424 my @tree = ({
207     # level 0
208     code_arr => sub { sort keys %{$l} },
209     filter_code => sub {
210     my $t = shift;
211     return $t if ($t =~ s/root://);
212     },
213     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
214     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
215     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
216 dpavlin 426 have_children => sub { defined($l->{"a:".$_[0]."::"}) },
217 dpavlin 424 child_code => sub { return $_[0] },
218     style => 'display: none',
219     },{
220     # 1
221     code_arr => sub { @{$l->{"a:".$_[0]."::"}} },
222     filter_code => sub { shift }, # nop
223     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
224     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
225     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
226 dpavlin 426 have_children => sub { defined($l->{"a:".$_[1].":"}) },
227 dpavlin 424 child_code => sub { return $_[1] },
228     style => 'display: none',
229     },{
230     # 2
231     code_arr => sub { @{$l->{"a:".$_[0].":"}} },
232     filter_code => sub { shift },
233     lookup_v900 => sub { shift },
234     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
235     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
236 dpavlin 426 have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) },
237 dpavlin 425 child_code => sub { return "a:".$_[2].":".$_[1] },
238     style => 'display: none',
239     },{
240     # 3 u¾i pojam
241     code_arr => sub { @{$l->{$_[0]}} },
242     filter_code => sub { shift },
243 dpavlin 426 lookup_v900 => sub {
244     my ($c,$p) = @_;
245     $p =~ s/^a:(..:....):.*$/$1/;
246     return "a:".$p.":".$c;
247     },
248     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
249     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
250     have_children => sub { defined($l->{$_[1]}) },
251     child_code => sub { return $_[1] },
252     },{
253     # 4
254     code_arr => sub { @{$l->{$_[0]}} },
255     filter_code => sub { shift },
256     lookup_v900 => sub {
257     my ($c,$p) = @_;
258     $p =~ s/^a:(..:....):.*$/$1/;
259     return "a:".$p.":".$c;
260     },
261     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
262     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
263 dpavlin 427 have_children => sub { defined($l->{$_[1]}) },
264     child_code => sub { return $_[1] },
265     },{
266     # 5
267     code_arr => sub { @{$l->{$_[0]}} },
268     filter_code => sub { shift },
269     lookup_v900 => sub {
270     my ($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:".$_[0]}} },
276     have_children => sub { defined($l->{$_[1]}) },
277     child_code => sub { return $_[1] },
278     },{
279     # 6
280     code_arr => sub { @{$l->{$_[0]}} },
281     filter_code => sub { shift },
282     lookup_v900 => sub {
283     my ($c,$p) = @_;
284     $p =~ s/^a:(..:....):.*$/$1/;
285     return "a:".$p.":".$c;
286     },
287     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
288     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
289     have_children => sub { defined($l->{$_[1]}) },
290     child_code => sub { return $_[1] },
291     },{
292     # 7
293     code_arr => sub { @{$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:".$_[0]}} },
302     have_children => sub { defined($l->{$_[1]}) },
303     child_code => sub { return $_[1] },
304     },{
305     # 8
306     code_arr => sub { @{$l->{$_[0]}} },
307     filter_code => sub { shift },
308     lookup_v900 => sub {
309     my ($c,$p) = @_;
310     $p =~ s/^a:(..:....):.*$/$1/;
311     return "a:".$p.":".$c;
312     },
313     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
314     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
315 dpavlin 426 # have_children => sub { defined($l->{$_[1]}) },
316     # child_code => sub { return $_[1] },
317 dpavlin 424 have_children => sub { 0 },
318     child_code => sub { 0 },
319     });
320 dpavlin 419
321 dpavlin 428 my @show_ids;
322     my @hide_ids;
323    
324 dpavlin 424 unroll(0,'');
325 dpavlin 419
326 dpavlin 424 $log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99"));
327 dpavlin 419
328 dpavlin 424 sub unroll {
329     my ($level,$start_code) = @_;
330 dpavlin 419
331 dpavlin 424 $log->logconfess("need level") unless (defined($level));
332 dpavlin 419
333 dpavlin 424 # all levels passed?
334     return if (! defined($tree[$level]));
335 dpavlin 419
336 dpavlin 424 $log->debug("unroll level $level, start code $start_code");
337 dpavlin 419
338 dpavlin 424 foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) {
339 dpavlin 419
340 dpavlin 424 if ($code = $tree[$level]->{'filter_code'}->($code)) {
341    
342     $log->debug("# $level filter passed code $code");
343 dpavlin 419
344 dpavlin 426 my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)");
345     $log->debug("# $level lookup_v900($code,$start_code) = $v900");
346    
347     my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)");
348 dpavlin 424 $log->debug("# $level lookup_term($code,$v900) = $term");
349 dpavlin 426
350     my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)");
351 dpavlin 424 $log->debug("# $level lookup_mfn($code,$v900) = $mfn");
352 dpavlin 419
353 dpavlin 424 $log->debug("$code -> $v900 : $term [$mfn]");
354 dpavlin 419
355 dpavlin 426 my ($link_start,$link_end) = ('','');
356 dpavlin 425
357     my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code);
358 dpavlin 426 if ($have_children) {
359 dpavlin 428 ($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>});
360 dpavlin 426 } else {
361     $log->debug("# $level doesn't have_children($code,$v900,$start_code)");
362     }
363 dpavlin 419
364 dpavlin 425 my $mfn_link = "thes/$mfn.html";
365     if (-e "out/$mfn_link") {
366     print HTML " " x $level .
367     qq{<li>${link_start}${term}${link_end}}.
368     qq{&nbsp;<a href="$mfn_link">&raquo;</a></li>\n};
369     } else {
370     $log->warn("file 'out/$mfn_link' doesn't exist, skipping");
371     }
372    
373     unless ($have_children) {
374 dpavlin 424 next;
375 dpavlin 419 }
376 dpavlin 428 my $style = $tree[$level]->{'style'};
377    
378 dpavlin 424 print HTML " " x $level .
379     qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}.
380 dpavlin 429 ($style ? ' style="'.$style.'"' : '').
381 dpavlin 424 qq{>\n};
382 dpavlin 419
383 dpavlin 428 if ($style) {
384     if ($style =~ m/display\s*:\s*none/i) {
385     push @hide_ids, "mfn$mfn";
386     } else {
387     push @show_ids, "mfn$mfn";
388     }
389     } else {
390     # default: show
391     push @show_ids, "mfn$mfn";
392     }
393    
394 dpavlin 425 unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code));
395 dpavlin 424
396     print HTML " " x $level . qq{</ul>\n};
397    
398 dpavlin 419 }
399     }
400     }
401    
402     print HTML qq{
403     </ul>
404     </body>
405     </html>
406     };
407    
408 dpavlin 425 close(HTML);
409    
410 dpavlin 428
411     my $js_file = "out/$js_url";
412     $log->info("creating '$js_file' with arrays of shown and hidden ids");
413     open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!");
414     print JS "var show = ['",join("','",@show_ids),"'];\n";
415     print JS "var hide = ['",join("','",@hide_ids),"'];\n";
416     close(JS);
417    
418     $log->info("closing index");
419 dpavlin 390 $index->close;
420    
421 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