question about the nature of DBM ties

[abez]

Have you tried it without saving to file?

On Tue, 24 Sep 2002, nkuipers wrote:

> Hi all,
> 
> When you tie a data structure to an external file, is populating that 
> structure for large input updating directly into the file or is it crowding 
> more and more stuff into memory which then gets dumped into the file, or what?
>  In other words, does tying in this manner free up more RAM?  The bottleneck 
> in my code is the unique function but this function is necessary.  All in all 
> the code works perfectly but takes too long.
> 
> #!/usr/bin/perl
> 
> use strict;
> use warnings;
> use DB_File;
> 
> my $infile = shift;
> my $wordsize = 10;
> my %clusters; #key=>value = 'id_string' => 'DNA_string'
> my %k_strings; #key=>value =(ie.) 'ACGTGGTCAC' => [id_string1, id_string2,...]
> 
> tie(%k_strings, "DB_File", "index.tmp") or die "Can't open filename: $!";
> 
> %k_strings = &build_index(\%clusters);
> 
> untie %k_strings;
> 
> sub build_index {
> 	my $clusters_hashref = shift;
> 	my %k_hash;
> 	while ( (my $id, my $sequence) = each %$clusters_hashref ) {
> 		my $tmp = $sequence;
> 		while ( length($tmp) >= $wordsize ) {
> 			my $kstring = substr($tmp, 0, $wordsize);
> 			if ( exists $k_hash{$kstring} ) {
> 				push @{ $k_hash{$kstring} }, $id
> 				if unique(\@{ $k_hash{$kstring} }, \$id)
> 			} else { $k_hash{$kstring} = [ $id ] }
> 			$tmp =~ s/^\w//;
> 		}
> 	}
> 	return %k_hash;
> }
> 
> sub unique {
> 	my ($array_ref, $id_ref) = @_;
> 	my $flag = 0;
> 	for (@$array_ref) {
> 		if ( $_ eq $$id_ref ) {
> 			$flag = 1;
> 			last;
> 		}
> 	}
> 	$flag == 1 ?  return 0 : (return 1);
> }
> 
> __END__
> 

-- 
ABeZ------------ ------- ------ - ---------- -- ------------
http://www.indexdirect.com/abez/ Abram Hindle (abez at abez.ca)
---- ------- ----------- ----------- - - ------ --------ABeZ