Are you sure about the flat-file approach? That way you always have to read/write the whole thing when editing a single item like:

#!/usr/bin/perl -i

#first arg is the item_number, second the text-file with replacement
#dbase file on STDIN, edited file on STDOUT

my $item_number = $ARGV[0];
open IN, $ARGV[1] or die "Could not open $ARGV[1]: $!";
my $replacement;
{
  local $/ = undef;
  $replacement = <IN>;
}

{ 
  local $/ = '::';
  while( <> ) {
    $_ = $replacement if $. == $item_number;
    print;
    print '::' if not eof;
  }
}
[download]

When the dbase gets large, the overhead may become problematic. Consider a RDBM like postgres or a good DB like BerkeleyDB.

Cheers,

Jeroen
"We are not alone"(FZ)