/[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 431 - (hide annotations)
Sun Sep 12 20:31:34 2004 UTC (14 years, 10 months ago) by dpavlin
File MIME type: text/plain
File size: 10897 byte(s)
new option: start_mfn

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 430 <div style="float: right; width: 10em; text-align: center; margin: 0.5m; background: #e0e0e0; border: 1px dashed #c0c0c0; z-index: 1;">
182 dpavlin 428 Folding:
183 dpavlin 430 <br/>
184     <a href="#" onClick="back_display();">&laquo;</a>&nbsp;<a href="#" onClick="apply_changed_display();">default</a>&nbsp;<a href="#" onClick="forward_display();">&raquo;</a>
185     <br/>
186     <a href="#" onClick="show_hide_display(show,hide);">reset</a>
187 dpavlin 428 <a href="#" onClick="show_display(show); show_display(hide);">all</a>
188     <a href="#" onClick="hide_display(hide); hide_display(show);">none</a>
189 dpavlin 430 <br/>
190     <a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length+', changed:'+changed_display_ids.length+', positin: '+changed_display_pos);">debug</a>
191 dpavlin 428 </div>
192    
193 dpavlin 419 <ul>
194     };
195    
196 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
197 dpavlin 419
198 dpavlin 424 my @tree = ({
199     # level 0
200     code_arr => sub { sort keys %{$l} },
201     filter_code => sub {
202     my $t = shift;
203     return $t if ($t =~ s/root://);
204     },
205     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
206     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
207     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
208 dpavlin 426 have_children => sub { defined($l->{"a:".$_[0]."::"}) },
209 dpavlin 424 child_code => sub { return $_[0] },
210     style => 'display: none',
211     },{
212     # 1
213     code_arr => sub { @{$l->{"a:".$_[0]."::"}} },
214     filter_code => sub { shift }, # nop
215     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
216     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
217     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
218 dpavlin 426 have_children => sub { defined($l->{"a:".$_[1].":"}) },
219 dpavlin 424 child_code => sub { return $_[1] },
220     style => 'display: none',
221     },{
222     # 2
223     code_arr => sub { @{$l->{"a:".$_[0].":"}} },
224     filter_code => sub { shift },
225     lookup_v900 => sub { shift },
226     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
227     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
228 dpavlin 426 have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) },
229 dpavlin 425 child_code => sub { return "a:".$_[2].":".$_[1] },
230     style => 'display: none',
231     },{
232     # 3 u¾i pojam
233     code_arr => sub { @{$l->{$_[0]}} },
234     filter_code => sub { shift },
235 dpavlin 426 lookup_v900 => sub {
236     my ($c,$p) = @_;
237     $p =~ s/^a:(..:....):.*$/$1/;
238     return "a:".$p.":".$c;
239     },
240     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
241     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
242     have_children => sub { defined($l->{$_[1]}) },
243     child_code => sub { return $_[1] },
244     },{
245     # 4
246     code_arr => sub { @{$l->{$_[0]}} },
247     filter_code => sub { shift },
248     lookup_v900 => sub {
249     my ($c,$p) = @_;
250     $p =~ s/^a:(..:....):.*$/$1/;
251     return "a:".$p.":".$c;
252     },
253     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
254     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
255 dpavlin 427 have_children => sub { defined($l->{$_[1]}) },
256     child_code => sub { return $_[1] },
257     },{
258     # 5
259     code_arr => sub { @{$l->{$_[0]}} },
260     filter_code => sub { shift },
261     lookup_v900 => sub {
262     my ($c,$p) = @_;
263     $p =~ s/^a:(..:....):.*$/$1/;
264     return "a:".$p.":".$c;
265     },
266     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
267     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
268     have_children => sub { defined($l->{$_[1]}) },
269     child_code => sub { return $_[1] },
270     },{
271     # 6
272     code_arr => sub { @{$l->{$_[0]}} },
273     filter_code => sub { shift },
274     lookup_v900 => sub {
275     my ($c,$p) = @_;
276     $p =~ s/^a:(..:....):.*$/$1/;
277     return "a:".$p.":".$c;
278     },
279     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
280     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
281     have_children => sub { defined($l->{$_[1]}) },
282     child_code => sub { return $_[1] },
283     },{
284     # 7
285     code_arr => sub { @{$l->{$_[0]}} },
286     filter_code => sub { shift },
287     lookup_v900 => sub {
288     my ($c,$p) = @_;
289     $p =~ s/^a:(..:....):.*$/$1/;
290     return "a:".$p.":".$c;
291     },
292     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
293     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
294     have_children => sub { defined($l->{$_[1]}) },
295     child_code => sub { return $_[1] },
296     },{
297     # 8
298     code_arr => sub { @{$l->{$_[0]}} },
299     filter_code => sub { shift },
300     lookup_v900 => sub {
301     my ($c,$p) = @_;
302     $p =~ s/^a:(..:....):.*$/$1/;
303     return "a:".$p.":".$c;
304     },
305     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
306     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
307 dpavlin 426 # have_children => sub { defined($l->{$_[1]}) },
308     # child_code => sub { return $_[1] },
309 dpavlin 424 have_children => sub { 0 },
310     child_code => sub { 0 },
311     });
312 dpavlin 419
313 dpavlin 428 my @show_ids;
314     my @hide_ids;
315    
316 dpavlin 424 unroll(0,'');
317 dpavlin 419
318 dpavlin 424 $log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99"));
319 dpavlin 419
320 dpavlin 424 sub unroll {
321     my ($level,$start_code) = @_;
322 dpavlin 419
323 dpavlin 424 $log->logconfess("need level") unless (defined($level));
324 dpavlin 419
325 dpavlin 424 # all levels passed?
326     return if (! defined($tree[$level]));
327 dpavlin 419
328 dpavlin 424 $log->debug("unroll level $level, start code $start_code");
329 dpavlin 419
330 dpavlin 424 foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) {
331 dpavlin 419
332 dpavlin 424 if ($code = $tree[$level]->{'filter_code'}->($code)) {
333    
334     $log->debug("# $level filter passed code $code");
335 dpavlin 419
336 dpavlin 426 my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)");
337     $log->debug("# $level lookup_v900($code,$start_code) = $v900");
338    
339     my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)");
340 dpavlin 424 $log->debug("# $level lookup_term($code,$v900) = $term");
341 dpavlin 426
342     my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)");
343 dpavlin 424 $log->debug("# $level lookup_mfn($code,$v900) = $mfn");
344 dpavlin 419
345 dpavlin 424 $log->debug("$code -> $v900 : $term [$mfn]");
346 dpavlin 419
347 dpavlin 426 my ($link_start,$link_end) = ('','');
348 dpavlin 425
349     my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code);
350 dpavlin 426 if ($have_children) {
351 dpavlin 428 ($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>});
352 dpavlin 426 } else {
353     $log->debug("# $level doesn't have_children($code,$v900,$start_code)");
354     }
355 dpavlin 419
356 dpavlin 425 my $mfn_link = "thes/$mfn.html";
357     if (-e "out/$mfn_link") {
358     print HTML " " x $level .
359     qq{<li>${link_start}${term}${link_end}}.
360     qq{&nbsp;<a href="$mfn_link">&raquo;</a></li>\n};
361     } else {
362     $log->warn("file 'out/$mfn_link' doesn't exist, skipping");
363     }
364    
365     unless ($have_children) {
366 dpavlin 424 next;
367 dpavlin 419 }
368 dpavlin 428 my $style = $tree[$level]->{'style'};
369    
370 dpavlin 424 print HTML " " x $level .
371     qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}.
372 dpavlin 429 ($style ? ' style="'.$style.'"' : '').
373 dpavlin 424 qq{>\n};
374 dpavlin 419
375 dpavlin 428 if ($style) {
376     if ($style =~ m/display\s*:\s*none/i) {
377     push @hide_ids, "mfn$mfn";
378     } else {
379     push @show_ids, "mfn$mfn";
380     }
381     } else {
382     # default: show
383     push @show_ids, "mfn$mfn";
384     }
385    
386 dpavlin 425 unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code));
387 dpavlin 424
388     print HTML " " x $level . qq{</ul>\n};
389    
390 dpavlin 419 }
391     }
392     }
393    
394     print HTML qq{
395     </ul>
396     </body>
397     </html>
398     };
399    
400 dpavlin 425 close(HTML);
401    
402 dpavlin 428
403     my $js_file = "out/$js_url";
404     $log->info("creating '$js_file' with arrays of shown and hidden ids");
405     open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!");
406     print JS "var show = ['",join("','",@show_ids),"'];\n";
407     print JS "var hide = ['",join("','",@hide_ids),"'];\n";
408     close(JS);
409    
410     $log->info("closing index");
411 dpavlin 390 $index->close;
412    
413 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