/[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 424 - (hide annotations)
Sat Sep 11 22:32:06 2004 UTC (15 years, 2 months ago) by dpavlin
File MIME type: text/plain
File size: 6550 byte(s)
refactore tree output into data-driven one (and recursive)

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 424 if(0) { # XXX
60    
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 420 $log->info("lookup hash: ",Dumper($webpac->{'lookup'}));
154    
155 dpavlin 419 #
156     # tree dump implementation which shouldn't be here :-)
157     #
158    
159     my $file = 'out/browse.html';
160     open(HTML, "> $file") || $log->logdie("can't open '$file'");
161    
162     print HTML qq{
163     <html>
164     <head>
165     <title>Browse</title>
166     <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/>
167 dpavlin 423 <script type="text/javascript" src="bfilter/hide.js"></script>
168 dpavlin 419 </head>
169     <body>
170     <ul>
171     };
172    
173 dpavlin 424 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
174 dpavlin 419
175 dpavlin 424 my @tree = ({
176     # level 0
177     code_arr => sub { sort keys %{$l} },
178     filter_code => sub {
179     my $t = shift;
180     return $t if ($t =~ s/root://);
181     },
182     lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
183     lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
184     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
185     have_children => sub { $l->{"a:".$_[0]."::" } },
186     child_code => sub { return $_[0] },
187     style => 'display: none',
188     },{
189     # 1
190     code_arr => sub { @{$l->{"a:".$_[0]."::"}} },
191     filter_code => sub { shift }, # nop
192     lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
193     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
194     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
195     have_children => sub { $l->{"a:".$_[1].":" } },
196     child_code => sub { return $_[1] },
197     style => 'display: none',
198     },{
199     # 2
200     code_arr => sub { @{$l->{"a:".$_[0].":"}} },
201     filter_code => sub { shift },
202     lookup_v900 => sub { shift },
203     lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
204     lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
205     have_children => sub { 0 },
206     child_code => sub { 0 },
207 dpavlin 419
208 dpavlin 424 });
209 dpavlin 419
210 dpavlin 424 unroll(0,'');
211 dpavlin 419
212 dpavlin 424 $log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99"));
213 dpavlin 419
214 dpavlin 424 sub unroll {
215     my ($level,$start_code) = @_;
216 dpavlin 419
217 dpavlin 424 $log->logconfess("need level") unless (defined($level));
218 dpavlin 419
219 dpavlin 424 # all levels passed?
220     return if (! defined($tree[$level]));
221 dpavlin 419
222 dpavlin 424 $log->debug("unroll level $level, start code $start_code");
223 dpavlin 419
224 dpavlin 424 foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) {
225 dpavlin 419
226 dpavlin 424 if ($code = $tree[$level]->{'filter_code'}->($code)) {
227    
228     $log->debug("# $level filter passed code $code");
229 dpavlin 419
230 dpavlin 424 my $v900 = $tree[$level]->{'lookup_v900'}->($code) || $log->logdie("can't lookup_v900 '$code'");
231     $log->debug("# $level lookup_v900($code) = $v900");
232     my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->logdie("can't lookup_term '$v900'");
233     $log->debug("# $level lookup_term($code,$v900) = $term");
234     my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->logdie("can't lookup_mfn '$v900'");
235     $log->debug("# $level lookup_mfn($code,$v900) = $mfn");
236 dpavlin 419
237 dpavlin 424 $log->debug("$code -> $v900 : $term [$mfn]");
238 dpavlin 419
239 dpavlin 424 print HTML " " x $level .
240     qq{<li><a href="#mfn$mfn" onClick="return hide_show('mfn$mfn');">$term</a>&nbsp;<a href="thes/$mfn.html">&raquo;</a></li>\n} if (-e "out/thes/$mfn.html");
241 dpavlin 419
242 dpavlin 424 unless ($tree[$level]->{'have_children'}->($code,$v900)) {
243     $log->warn("can't find children of $code at level $level");
244     next;
245 dpavlin 419 }
246 dpavlin 424
247     print HTML " " x $level .
248     qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}.
249     ($tree[$level]->{'style'} ? ' style="'.$tree[$level]->{'style'}.'"' : '').
250     qq{>\n};
251 dpavlin 419
252 dpavlin 424 unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900));
253    
254     print HTML " " x $level . qq{</ul>\n};
255    
256 dpavlin 419 }
257     }
258     }
259    
260     print HTML qq{
261     </ul>
262     </body>
263     </html>
264     };
265    
266 dpavlin 390 $index->close;
267    
268 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