Problem

You have a file in one DBM format, but another program expects input in a different DBM format.

Solution

Reads the keys and values from the initial DBM file and writes them to a new file in the different DBM format as in Example 14.2 .

#!/usr/bin/perl -w
# db2gdbm: converts DB to GDBM

use strict;

use DB_File;
use GDBM_File;

unless (@ARGV == 2) {
    die "usage: db2gdbm infile outfile\n";
}

my ($infile, $outfile) = @ARGV;                     
my (%db_in, %db_out);                               

# open the files
tie(%db_in, 'DB_File', $infile)
    or die "Can't tie $infile: $!";
tie(%db_out, 'GDBM_File', $outfile, GDBM_WRCREAT, 0666)
    or die "Can't tie $outfile: $!";

# copy (don't use %db_out = %db_in because it's slow on big databases)
while (my($k, $v) = each %db_in) {
    $db_out{$k} = $v;
}

# these unties happen automatically at program exit
untie %db_in;
untie %db_out;

Call the program as:

% db2gdbm /tmp/users.db /tmp/users.gdbm

Discussion

When multiple types of DBM file are used in the same program, you have to use tie , not the dbmopen interface. That's because with dbmopen you can only use one database format, which is why its use is deprecated.

Copying hashes by simple assignment, as in %new = %old , works on DBM files. However, it loads everything into memory first as a list, which doesn't matter with small hashes, but can be prohibitively expensive in the case of DBM files. For database hashes, use each to iterate through them instead.

See Also

The documentation for the standard modules GDBM_File, NDBM_File, SDBM_File, DB_File, also in Chapter 7 of Programming Perl ; Recipe 14.1