/[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 413 - (hide annotations)
Tue Sep 7 18:02:06 2004 UTC (15 years ago) by dpavlin
File MIME type: text/plain
File size: 2736 byte(s)
create index files for bfilter

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 413 # limit_mfn => 500,
21 dpavlin 372 # debug => 1,
22 dpavlin 352 ) || die;
23    
24 dpavlin 373 my $log = $webpac->_get_logger();
25    
26 dpavlin 390 my $index = new WebPAC::jsFind(
27     index_path => './out/index',
28 dpavlin 413 keys => 10,
29 dpavlin 390 ) || die;
30    
31 dpavlin 410 my $thes;
32    
33 dpavlin 10 $|=1;
34 dpavlin 9
35 dpavlin 352 my $maxmfn = $webpac->open_isis(
36 dpavlin 357 filename => shift @ARGV || '/data/hidra/THS/THS',
37 dpavlin 352 lookup => [
38 dpavlin 348 { 'key' => 'd:v900', 'val' => 'v250^a' },
39 dpavlin 352 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
40     # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
41     # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
42     { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
43 dpavlin 374 { 'key' => '900_mfn:v900', 'val' => 'v000' },
44 dpavlin 352 ],
45 dpavlin 54 );
46 dpavlin 3
47 dpavlin 372 $log->info("rows: $maxmfn");
48 dpavlin 170
49 dpavlin 363 $webpac->open_import_xml(type => 'isis_hidra_ths');
50    
51 dpavlin 362 while (my $rec = $webpac->fetch_rec) {
52 dpavlin 358
53 dpavlin 370 my @ds = $webpac->data_structure($rec);
54 dpavlin 366
55 dpavlin 411 if (0 && $log->is_debug) {
56 dpavlin 372 $log->debug("rec = ",Dumper($rec));
57     $log->debug("ds = ",Dumper(\@ds));
58     }
59    
60 dpavlin 374 next if (! @ds);
61 dpavlin 366
62 dpavlin 374 my $filename = $webpac->{'current_filename'};
63    
64     if ($filename) {
65 dpavlin 411 $webpac->output_file(
66     file => $filename,
67 dpavlin 374 template => 'html.tt',
68     data => \@ds,
69 dpavlin 398 headline => $webpac->{'headline'},
70 dpavlin 374 );
71     } else {
72     print $webpac->output(
73     template => 'text.tt',
74     data => \@ds,
75 dpavlin 398 headline => $webpac->{'headline'},
76 dpavlin 374 );
77     }
78    
79 dpavlin 398 my $headline = $webpac->{'headline'};
80 dpavlin 390
81 dpavlin 392 my $f = $filename;
82     $f =~ s!out/!!;
83    
84 dpavlin 390 # save into index
85     foreach my $ds (@ds) {
86     next if (! $ds->{'swish'});
87    
88     $index->insert(
89     index_name => $ds->{'tag'},
90 dpavlin 392 path => $f,
91 dpavlin 390 headline => $headline,
92     words => join(" ",@{$ds->{'swish'}})
93     );
94     }
95    
96 dpavlin 410 # save into sorted index (thesaurus)
97     foreach my $ds (@ds) {
98     next if (! $ds->{'index'});
99    
100     $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
101    
102     $thes->{$ds->{'tag'}}->insert(
103     path => $f,
104 dpavlin 413 headline => join(" ",@{$ds->{'index'}}),
105 dpavlin 410 );
106     }
107    
108 dpavlin 390 # print Dumper(\@ds);
109    
110 dpavlin 3 }
111    
112 dpavlin 411 foreach my $t (keys %{$thes}) {
113    
114     my @e = $thes->{$t}->elements;
115     if (! @e) {
116     $log->logwarn("no elements in sorted index $t?");
117     next;
118     }
119    
120 dpavlin 413 $log->info("saving sorted index $t [".scalar(@e)." elements]");
121 dpavlin 411
122     $webpac->output_file(
123 dpavlin 413 file => "./out/bfilter/$t.txt",
124 dpavlin 411 template => 'index.tt',
125     data => \@e,
126     index_name => $t,
127     );
128     }
129    
130     if (0 && $log->is_debug) {
131 dpavlin 372 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
132     $log->debug("data hash: ",Dumper($webpac->{'data'}));
133 dpavlin 410 foreach my $t (keys %{$thes}) {
134     $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
135     }
136 dpavlin 372 }
137 dpavlin 390
138     $index->close;
139    

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.26