trunk/vector/index.pl

#!/usr/bin/perl -wT

use Search::ContextGraph;
use GDBM_File;
use Term::ProgressBar;

my $gdbm_file = '../swish/file2title.gdbm';
my $graph_file = 'stored.cng';

my %gdbm_file;
tie %gdbm_file, 'GDBM_File', $gdbm_file, &GDBM_READER, 0640;

my $cg = Search::ContextGraph->new();

my @docs = keys %gdbm_file;

my $i = 0;
my $max = $#docs;

my $progress = Term::ProgressBar->new ({
        name => 'Indexing',
        ETA => 'linear',
        count => $max,
});

my $next_update = 0;

foreach (@docs) {
        $cg->add( $_, [ split(/\s+/,$gdbm_file{$_}) ] );

        $i++;
        $next_update = $progress->update($i) if ($i > $next_update);
}

$progress->message("Saving to '$graph_file'");
$cg->store( $graph_file );

untie %gdbm_file;
1	#!/usr/bin/perl -wT
2
3	use Search::ContextGraph;
4	use GDBM_File;
5	use Term::ProgressBar;
6
7	my $gdbm_file = '../swish/file2title.gdbm';
8	my $graph_file = 'stored.cng';
9
10	my %gdbm_file;
11	tie %gdbm_file, 'GDBM_File', $gdbm_file, &GDBM_READER, 0640;
12
13	my $cg = Search::ContextGraph->new();
14
15	my @docs = keys %gdbm_file;
16
17	my $i = 0;
18	my $max = $#docs;
19
20	my $progress = Term::ProgressBar->new ({
21	name => 'Indexing',
22	ETA => 'linear',
23	count => $max,
24	});
25
26	my $next_update = 0;
27
28	foreach (@docs) {
29	$cg->add( $_, [ split(/\s+/,$gdbm_file{$_}) ] );
30
31	$i++;
32	$next_update = $progress->update($i) if ($i > $next_update);
33	}
34
35	$progress->message("Saving to '$graph_file'");
36	$cg->store( $graph_file );
37
38	untie %gdbm_file;