I'm sorry, Mr. Bond

[no_slogan]

...but your code has been broken. Below is a simple known plaintext attack which "nearly" recovers the encryption key, given a block of encrypted data and a few known plaintext bytes. It's generally not difficult to get ahold of the known plaintexts, especially if you're encrypting files like Word docs that have a fixed header.

You say your program was "based on the ABC1 idea." I don't know how close the resemblence is, but I'd like to point out that changing a cipher even slightly can destroy its security.

Here's the sample output. Notice that, for each key byte, the correct value is one of the given possibilities.

Secret key is 'Shaken'
Guessing key bytes
key byte 0
'S' : 10
key byte 1
'#' : 9
'h' : 10
key byte 2
'/' : 10
'3' : 9
'F' : 10
'N' : 9
'a' : 11
key byte 3
'k' : 11
'o' : 9
key byte 4
'F' : 9
'[' : 9
'e' : 11
'o' : 9
key byte 5
'a' : 9
'n' : 10

And here's the code. It needs to know the block size and the length of the encryption key. That information might not be available, but it's easy enough to find by brute force.

#!/usr/bin/perl

use strict;
use warnings;

# this information is known to the attacker
my ($ciphertext, $known_plaintext, $keylen);

{ # secret stuff in here
  my $key = "Shaken";
  print "Secret key is '$key'\n";
  $keylen = length($key);
  open F, $0 or die; # read test data from this script - why not?
  my $plaintext;
  read F, $plaintext, 1024 or die;
  close F;
  $known_plaintext = substr($plaintext, 0, 8);
  $ciphertext = encrypt_blk($plaintext, $key);
} # secret stuff in here

print "Guessing key bytes\n";
my $plain_bits  = length($known_plaintext)*8;
my $cipher_bits = length($ciphertext)*8 - 1;
foreach my $keypos (0 .. $keylen-1) {
  print "key byte $keypos\n";
  my $first = $keypos > 1 ? $keypos - 1 : $keylen + $keypos - 1;
  foreach my $keybyte (0x20 .. 0x7e) {
    my $keyval = ($keybyte >= 0x30 && $keybyte <= 0x39)
        ? $keybyte-0x2f : $keybyte+1;
    my $hits = 0;
    my $trials = 0;
    for (my $bitpos = $first; $bitpos < $plain_bits; $bitpos += $keyle
+n) {
      my $pos = ($bitpos * $keyval) % $cipher_bits;
      $trials++;
      $hits++ if vec($known_plaintext,$bitpos,1) == vec($ciphertext,$p
+os,1);
    }
    my $thresh = int($trials * .8);
    print "'",chr($keybyte),"' : $hits\n" if $hits > $thresh;
  }
}

sub encrypt_blk {
  my ($inblk, $key) = @_;
  my @key = map { (/\d/ ? $_ : ord) + 1 } split(//, $key);
  my $blkbytes = length($inblk);
  my $blkbits  = $blkbytes * 8;
  my $outblk   = "\0" x $blkbytes;
  my $used     = "\0" x $blkbytes;
  my $keypos   = 0;
  my $newpos;
  for (my $blkpos = 0; $blkpos < $blkbits; $blkpos++) {
    ++$keypos >= @key and $keypos = 0;
    $newpos = ($key[$keypos] * $blkpos) % ($blkbits - 1);
    while (vec($used, $newpos, 1)) {
      ++$newpos >= $blkbits and $newpos = 0;
    }
    vec($used,   $newpos, 1) = 1;
    vec($outblk, $newpos, 1) = vec($inblk, $blkpos, 1);
  }
  return $outblk;
} # encrypt_blk

Update: The DMCA can byte me. Show me where this cipher is used to "effectively control access to a copyrighted work." Cryptography is perfectly legal, and I'm getting tired of hearing from you fear-mongers.

[gmpassos]

Cool! I liked what you made, and you are right, documents like word, pdf, have the same header, and using this information we can use brute force to get the headers, and geting the headers we can get the rest. But if we use another block size (not 1024, the default) and a key with 20+ digits will be very difficult. To make a better cipher we can just add something to mask the headers too, or encrypt again the data!

The James Bond cipher are not based in the source of the ABC1, just in the idea to move bits.

But remember, I made this for fun, this is the only cipher that I made! I never used this for real, because I don't have top secret things! Every thing that I made is for Open Source. Why I will encrypt what I will publish for the world!?

"The hardest door to breack is the door that we don't need to open".

And "no_slogan", DMCA can't kill our liberty of expression. Nice code! And Cryptography is legal, because without it internet can't exist.

"The creativity is the expression of the liberty".

[no_slogan]

I like your attitude, so I gave you a ++. Some people get really touchy about their home-grown ciphers. It might be a good idea to include a comment on your thoughts about the cipher's strength, if you're going to be distributing it.

...a key with 20+ digits will be very difficult.

The central problem with your cipher is that there is very little diffusion of key information, which allows me to attack the key bytes one at a time. Adding more key bytes only results in a linear increase in the difficulty of attack. In a properly diffused cipher, it would increase exponentially. Of course, more key bytes means more diffusion required... take a look at CAST5 or AES sometime. There's a lot to be learned by looking at existing ciphers, even though they appear very confusing.