/[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 429 - (hide annotations)
Sun Sep 12 17:21:47 2004 UTC (19 years, 6 months ago) by dpavlin
File MIME type: text/plain
File size: 10521 byte(s)
don't use body on load, use style in html (speed up first access to browse
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     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 372 # debug => 1,
22 dpavlin 423 # low_mem => 1,
23 dpavlin 352 ) || die;
24    
25 dpavlin 421 my $log = $webpac->_get_logger() || die "can't get logger";
26 dpavlin 373
27 dpavlin 421 $log->debug("creating WebPAC::jsFind object");
28    
29 dpavlin 390 my $index = new WebPAC::jsFind(
30     index_path => './out/index',
31 dpavlin 413 keys => 10,
32 dpavlin 390 ) || die;
33    
34 dpavlin 410 my $thes;
35    
36 dpavlin 10 $|=1;
37 dpavlin 9
38 dpavlin 352 my $maxmfn = $webpac->open_isis(
39 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
40 dpavlin 352 lookup => [
41 dpavlin 348 { 'key' => 'd:v900', 'val' => 'v250^a' },
42 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
43     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
44     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
45     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
46 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
47 dpavlin 419 # tree structure
48     { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
49     { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
50 dpavlin 352 ],
51 dpavlin 54 );
52 dpavlin 3
53 dpavlin 421 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
54    
55 dpavlin 372 $log->info("rows: $maxmfn");
56 dpavlin 170
57 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
58    
59 dpavlin 425 if(1) { # XXX
60 dpavlin 424
61 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
62 dpavlin 358
63 dpavlin 370 my @ds = $webpac->data_structure($rec);
64 dpavlin 366
65 dpavlin 411 if (0 && $log->is_debug) {
66 dpavlin 372 $log->debug("rec = ",Dumper($rec));
67     $log->debug("ds = ",Dumper(\@ds));
68     }
69    
70 dpavlin 374 next if (! @ds);
71 dpavlin 366
72 dpavlin 421 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
73 dpavlin 374
74     if ($filename) {
75 dpavlin 411 $webpac->output_file(
76     file => $filename,
77 dpavlin 374 template => 'html.tt',
78     data => \@ds,
79 dpavlin 398 headline => $webpac->{'headline'},
80 dpavlin 374 );
81     } else {
82     print $webpac->output(
83     template => 'text.tt',
84     data => \@ds,
85 dpavlin 398 headline => $webpac->{'headline'},
86 dpavlin 374 );
87     }
88    
89 dpavlin 398 my $headline = $webpac->{'headline'};
90 dpavlin 390
91 dpavlin 392 my $f = $filename;
92     $f =~ s!out/!!;
93    
94 dpavlin 390 # save into index
95     foreach my $ds (@ds) {
96     next if (! $ds->{'swish'});
97    
98     $index->insert(
99     index_name => $ds->{'tag'},
100 dpavlin 392 path => $f,
101 dpavlin 390 headline => $headline,
102     words => join(" ",@{$ds->{'swish'}})
103     );
104     }
105    
106 dpavlin 410 # save into sorted index (thesaurus)
107     foreach my $ds (@ds) {
108     next if (! $ds->{'index'});
109    
110     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
111    
112 dpavlin 415 foreach my $h (@{$ds->{'index'}}) {
113     $thes->{$ds->{'tag'}}->insert(
114     path => $f,
115     headline => $h,
116     );
117     }
118 dpavlin 410 }
119    
120 dpavlin 390 # print Dumper(\@ds);
121    
122 dpavlin 3 }
123    
124 dpavlin 411 foreach my $t (keys %{$thes}) {
125    
126     my @e = $thes->{$t}->elements;
127     if (! @e) {
128     $log->logwarn("no elements in sorted index $t?");
129     next;
130     }
131    
132 dpavlin 415 my $file = "./out/bfilter/$t.txt";
133     $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
134 dpavlin 411
135     $webpac->output_file(
136 dpavlin 415 file => $file,
137 dpavlin 411 template => 'index.tt',
138     data => \@e,
139     index_name => $t,
140     );
141     }
142    
143     if (0 && $log->is_debug) {
144 dpavlin 372 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
145     $log->debug("data hash: ",Dumper($webpac->{'data'}));
146 dpavlin 410 foreach my $t (keys %{$thes}) {
147     $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
148     }
149 dpavlin 372 }
150 dpavlin 390
151 dpavlin 424 } # XXX if(0)
152    
153 dpavlin 428 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
154 dpavlin 420
155 dpavlin 419 #
156     # tree dump implementation which shouldn't be here :-)
157     #
158    
159 dpavlin 428
160 dpavlin 419 my $file = 'out/browse.html';
161 dpavlin 428 my $js_url = 'tree-ids.js';
162    
163     $log->info("creating '$file' for tree html");
164 dpavlin 419 open(HTML, "> $file") || $log->logdie("can't open '$file'");
165    
166     print HTML qq{
167     <html>
168     <head>
169     <title>Browse</title>
170     <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/>
171 dpavlin 423 <script type="text/javascript" src="bfilter/hide.js"></script>
172 dpavlin 428 <script type="text/javascript" src="$js_url"></script>
173 dpavlin 419 </head>
174 dpavlin 428
175 dpavlin 429 <!--
176 dpavlin 428 <body onLoad="show_hide_display(show,hide);">
177 dpavlin 429 -->
178     <body>
179 dpavlin 428
180     <div style="float: right; width: 10em;">
181     Folding:
182     <a href="#" onClick="show_hide_display(show,hide);">default</a>
183     <a href="#" onClick="show_display(show); show_display(hide);">all</a>
184     <a href="#" onClick="hide_display(hide); hide_display(show);">none</a>
185     <a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length);">debug</a>
186     </div>
187    
188    
189 dpavlin 419 <ul>
190     };
191    
192 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
193 dpavlin 419
194 dpavlin 424 my @tree = ({
195     # level 0
196     code_arr => sub { sort keys %{$l} },
197     filter_code => sub {
198     my $t = shift;
199     return $t if ($t =~ s/root://);
200     },
201     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
202     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
203     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
204 dpavlin 426 have_children => sub { defined($l->{"a:".$_[0]."::"}) },
205 dpavlin 424 child_code => sub { return $_[0] },
206     style => 'display: none',
207     },{
208     # 1
209     code_arr => sub { @{$l->{"a:".$_[0]."::"}} },
210     filter_code => sub { shift }, # nop
211     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
212     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
213     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
214 dpavlin 426 have_children => sub { defined($l->{"a:".$_[1].":"}) },
215 dpavlin 424 child_code => sub { return $_[1] },
216     style => 'display: none',
217     },{
218     # 2
219     code_arr => sub { @{$l->{"a:".$_[0].":"}} },
220     filter_code => sub { shift },
221     lookup_v900 => sub { shift },
222     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
223     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
224 dpavlin 426 have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) },
225 dpavlin 425 child_code => sub { return "a:".$_[2].":".$_[1] },
226     style => 'display: none',
227     },{
228     # 3 u¾i pojam
229     code_arr => sub { @{$l->{$_[0]}} },
230     filter_code => sub { shift },
231 dpavlin 426 lookup_v900 => sub {
232     my ($c,$p) = @_;
233     $p =~ s/^a:(..:....):.*$/$1/;
234     return "a:".$p.":".$c;
235     },
236     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
237     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
238     have_children => sub { defined($l->{$_[1]}) },
239     child_code => sub { return $_[1] },
240     },{
241     # 4
242     code_arr => sub { @{$l->{$_[0]}} },
243     filter_code => sub { shift },
244     lookup_v900 => sub {
245     my ($c,$p) = @_;
246     $p =~ s/^a:(..:....):.*$/$1/;
247     return "a:".$p.":".$c;
248     },
249     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
250     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
251 dpavlin 427 have_children => sub { defined($l->{$_[1]}) },
252     child_code => sub { return $_[1] },
253     },{
254     # 5
255     code_arr => sub { @{$l->{$_[0]}} },
256     filter_code => sub { shift },
257     lookup_v900 => sub {
258     my ($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:".$_[0]}} },
264     have_children => sub { defined($l->{$_[1]}) },
265     child_code => sub { return $_[1] },
266     },{
267     # 6
268     code_arr => sub { @{$l->{$_[0]}} },
269     filter_code => sub { shift },
270     lookup_v900 => sub {
271     my ($c,$p) = @_;
272     $p =~ s/^a:(..:....):.*$/$1/;
273     return "a:".$p.":".$c;
274     },
275     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
276     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
277     have_children => sub { defined($l->{$_[1]}) },
278     child_code => sub { return $_[1] },
279     },{
280     # 7
281     code_arr => sub { @{$l->{$_[0]}} },
282     filter_code => sub { shift },
283     lookup_v900 => sub {
284     my ($c,$p) = @_;
285     $p =~ s/^a:(..:....):.*$/$1/;
286     return "a:".$p.":".$c;
287     },
288     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
289     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
290     have_children => sub { defined($l->{$_[1]}) },
291     child_code => sub { return $_[1] },
292     },{
293     # 8
294     code_arr => sub { @{$l->{$_[0]}} },
295     filter_code => sub { shift },
296     lookup_v900 => sub {
297     my ($c,$p) = @_;
298     $p =~ s/^a:(..:....):.*$/$1/;
299     return "a:".$p.":".$c;
300     },
301     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
302     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
303 dpavlin 426 # have_children => sub { defined($l->{$_[1]}) },
304     # child_code => sub { return $_[1] },
305 dpavlin 424 have_children => sub { 0 },
306     child_code => sub { 0 },
307     });
308 dpavlin 419
309 dpavlin 428 my @show_ids;
310     my @hide_ids;
311    
312 dpavlin 424 unroll(0,'');
313 dpavlin 419
314 dpavlin 424 $log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99"));
315 dpavlin 419
316 dpavlin 424 sub unroll {
317     my ($level,$start_code) = @_;
318 dpavlin 419
319 dpavlin 424 $log->logconfess("need level") unless (defined($level));
320 dpavlin 419
321 dpavlin 424 # all levels passed?
322     return if (! defined($tree[$level]));
323 dpavlin 419
324 dpavlin 424 $log->debug("unroll level $level, start code $start_code");
325 dpavlin 419
326 dpavlin 424 foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) {
327 dpavlin 419
328 dpavlin 424 if ($code = $tree[$level]->{'filter_code'}->($code)) {
329    
330     $log->debug("# $level filter passed code $code");
331 dpavlin 419
332 dpavlin 426 my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)");
333     $log->debug("# $level lookup_v900($code,$start_code) = $v900");
334    
335     my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)");
336 dpavlin 424 $log->debug("# $level lookup_term($code,$v900) = $term");
337 dpavlin 426
338     my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)");
339 dpavlin 424 $log->debug("# $level lookup_mfn($code,$v900) = $mfn");
340 dpavlin 419
341 dpavlin 424 $log->debug("$code -> $v900 : $term [$mfn]");
342 dpavlin 419
343 dpavlin 426 my ($link_start,$link_end) = ('','');
344 dpavlin 425
345     my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code);
346 dpavlin 426 if ($have_children) {
347 dpavlin 428 ($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>});
348 dpavlin 426 } else {
349     $log->debug("# $level doesn't have_children($code,$v900,$start_code)");
350     }
351 dpavlin 419
352 dpavlin 425 my $mfn_link = "thes/$mfn.html";
353     if (-e "out/$mfn_link") {
354     print HTML " " x $level .
355     qq{<li>${link_start}${term}${link_end}}.
356     qq{&nbsp;<a href="$mfn_link">&raquo;</a></li>\n};
357     } else {
358     $log->warn("file 'out/$mfn_link' doesn't exist, skipping");
359     }
360    
361     unless ($have_children) {
362 dpavlin 424 next;
363 dpavlin 419 }
364 dpavlin 428 my $style = $tree[$level]->{'style'};
365    
366 dpavlin 424 print HTML " " x $level .
367     qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}.
368 dpavlin 429 ($style ? ' style="'.$style.'"' : '').
369 dpavlin 424 qq{>\n};
370 dpavlin 419
371 dpavlin 428 if ($style) {
372     if ($style =~ m/display\s*:\s*none/i) {
373     push @hide_ids, "mfn$mfn";
374     } else {
375     push @show_ids, "mfn$mfn";
376     }
377     } else {
378     # default: show
379     push @show_ids, "mfn$mfn";
380     }
381    
382 dpavlin 425 unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code));
383 dpavlin 424
384     print HTML " " x $level . qq{</ul>\n};
385    
386 dpavlin 419 }
387     }
388     }
389    
390     print HTML qq{
391     </ul>
392     </body>
393     </html>
394     };
395    
396 dpavlin 425 close(HTML);
397    
398 dpavlin 428
399     my $js_file = "out/$js_url";
400     $log->info("creating '$js_file' with arrays of shown and hidden ids");
401     open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!");
402     print JS "var show = ['",join("','",@show_ids),"'];\n";
403     print JS "var hide = ['",join("','",@hide_ids),"'];\n";
404     close(JS);
405    
406     $log->info("closing index");
407 dpavlin 390 $index->close;
408    
409 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