#!/usr/bin/perl -w
# Generate a base file with some random MD5, then generate 16 more files
# that each share one of the MD5 bytes with the base file.  Because
# there is no single MD5 byte that uniquely identifies the original
# file, digest for this file will require at least two MD5 bytes.

use strict;
use Digest::MD5 'md5';

# Generate base file
my $data = (rand . "\n");
my $base_digest = md5($data);

my $file;
open $file, ">collision_base.txt" or die $!;
print $file $data;
close $file;

# Generate extra files until we found a one byte collision at each
# offset with the base file.
my %collisions = ();
for(my $i = 0; (scalar keys %collisions) < length($base_digest); $i++)
{
   $data = "$i\n";
   my $digest = md5($data);
   for(my $j = 0; $j < length($base_digest); $j++)
   {
      next if exists $collisions{$j};
      if( substr($digest, $j, 1) eq substr($base_digest, $j, 1) )
      {
         open $file, ">collision_$j.txt" or die $!;
         print $file $data;
         close $file;
         $collisions{$j} = 1;
         print "$i: ", (scalar keys %collisions), " collisions\n";
      }
   }
}