/[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 421 - (hide annotations)
Fri Sep 10 22:24:42 2004 UTC (15 years, 1 month ago) by dpavlin
File MIME type: text/plain
File size: 5188 byte(s)
low_mem option for desktop class-machines

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 421 # limit_mfn => 100,
21 dpavlin 372 # debug => 1,
22 dpavlin 421 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 421 $log->debug("data: ",Dumper($webpac->{'data'}));
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 420 $log->info("lookup hash: ",Dumper($webpac->{'lookup'}));
152    
153 dpavlin 419 #
154     # tree dump implementation which shouldn't be here :-)
155     #
156    
157     my $file = 'out/browse.html';
158     open(HTML, "> $file") || $log->logdie("can't open '$file'");
159    
160     print HTML qq{
161     <html>
162     <head>
163     <title>Browse</title>
164     <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/>
165     </head>
166     <body>
167     <ul>
168     };
169    
170     foreach my $code (sort keys %{$webpac->{'lookup'}}) {
171    
172     my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
173    
174     if ($code =~ m/^root:/) {
175     my $v900 = shift @{ $l->{$code} } || $log->logconfess("can't lookup '$code'");
176     $code =~ s/^root:// || die;
177    
178     my $term = shift @{ $l->{"d:${v900}"} } || die;
179     my $mfn = shift @{ $l->{"900_mfn:${v900}"} } || die;
180    
181 dpavlin 421 $log->debug("$code -> $v900 : $term [$mfn]");
182 dpavlin 419
183     print HTML qq{ <li><a href="thes/$mfn.html">$term</a></li>\n} if (-e "out/thes/$mfn.html");
184    
185 dpavlin 420 unless ($l->{"a:${code}::"}) {
186     $log->warn("can't find 'a:${code}::'");
187     next;
188     }
189 dpavlin 419
190     print HTML qq{ <ul>\n};
191    
192     foreach my $l2_v900 (@{ $l->{"a:${code}::"} }) {
193    
194     my $l2_code = shift @{ $l->{"code:${l2_v900}"} } || die;
195     my $l2_term = shift @{ $l->{"d:$l2_v900"} } || die;
196     my $l2_mfn = shift @{ $l->{"900_mfn:${l2_v900}"} } || die;
197    
198 dpavlin 421 $log->debug("$l2_code -> $l2_v900 : $l2_term [$l2_mfn]");
199 dpavlin 419
200     print HTML qq{ <li><a href="thes/$l2_mfn.html">$l2_term</a></li>\n} if (-e "out/thes/$l2_mfn.html");
201 dpavlin 420
202     unless ($l->{"a:${l2_code}:"}) {
203     $log->warn("can't find 'a:${l2_code}:'");
204     next;
205     }
206 dpavlin 419
207     print HTML qq{ <ul>\n};
208    
209     foreach my $l3_v900 (@{ $l->{"a:${l2_code}:"} }) {
210    
211     my $l3_term = shift @{ $l->{"d:$l3_v900"} } || die;
212     my $l3_mfn = shift @{ $l->{"900_mfn:${l3_v900}"} } || die;
213    
214 dpavlin 421 $log->debug(" -> $l3_v900 : $l3_term [$l3_mfn]");
215 dpavlin 419
216     print HTML qq{ <li><a href="thes/$l3_mfn.html">$l3_term</a></li>\n} if (-e "out/thes/$l3_mfn.html");
217     }
218    
219     print HTML qq{ </ul>\n};
220     }
221     print HTML qq{ </ul>\n};
222     }
223     }
224    
225     print HTML qq{
226     </ul>
227     </body>
228     </html>
229     };
230    
231 dpavlin 390 $index->close;
232    

Properties

Name Value
cvs2svn:cvs-rev 1.64
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26