32-bit block cipher based on Skipjack

3 years after

NAME Crypt::Skip32 - 32-bit block cipher based on Skipjack

SYNOPSIS use Crypt::Skip32;

  my $cipher     = new Crypt::Skip32 $key;
  my $ciphertext = $cipher->encrypt($plaintext);
  my $plaintext  = $cipher->decrypt($ciphertext);
  my $blocksize  = $cipher->blocksize;
  my $keysize    = $cipher->keysize;

DESCRIPTION SKIP32 is a 80-bit key, 32-bit block cipher based on Skipjack. The Perl code for the algorithm is a direct translation from C to Perl of skip32.c by Greg Rose found here:

This cipher can be handy for scrambling small (32-bit) values when you
would like to obscure them while keeping the encrypted output size small
(also only 32 bits).

One example where Crypt::Skip32 has been useful: You have numeric
database record ids which increment sequentially. You would like to use
them in URLs, but you don't want to make it obvious how many X's you
have in the database by putting the ids directly in the URLs.

You can use Crypt::Skip32 to scramble ids and put the resulting 32-bit
value in URLs (perhaps as 8 hex digits or some other shorter encoding).
When a user requests a URL, you can unscramble the id to retrieve the
object from the database.

Warning: A 32-bit value can only go a little over 4 billion (American).
Plan ahead if what you need to encrypt might eventually go over this

FUNCTIONS new my $cipher = new Crypt::Skip32 $key;

    Creates a new Crypt::Skip32 block cipher object, using $key, where
    $key is a key of "keysize" bytes (10).

      my $ciphertext = $cipher->encrypt($plaintext);

    Encrypt $plaintext and return the $ciphertext. The $plaintext must
    be of "blocksize" bytes (4).

    See the EXAMPLE below for hints on how to take a plain integer,
    encrypt it, and encode it for use in URLs and other non-binary

      my $plaintext = $cipher->decrypt($ciphertext);

    Decrypt $ciphertext and return the $plaintext. The $ciphertext must
    be of "blocksize" bytes (4).

      my $blocksize = $cipher->blocksize;
      my $blocksize = Crypt::Skip32->blocksize;

    Returns the size (in bytes) of the block cipher. This is always 4
    bytes (for 32 bits).

      my $keysize = $cipher->keysize;
      my $keysize = Crypt::Skip32->keysize;

    Returns the size (in bytes) of the key. This is always 10 bytes (for
    80 bits).

NOTES If Crypt::Skip32::XS is installed, this module will use it and the constructor will return an object of that type, though the interface is identical. You can stick with the pure Perl version by setting the CRYPT_SKIP32_PP environment variable before using this module.

If reporting a bug, please try to determine (if possible) if it is this
module or the XS one, and report it to the corresponding maintainer.

EXAMPLE This sample code demonstrates how Crypt::Skip32 can be used to encrypt unsigned integers and encode them for use in web URLs, form values, and other places where short encrypted text might be useful.

  use Crypt::Skip32;

  # Create a cipher. Change the long hex string to your secret key.
  my $key         = pack("H20", "112233445566778899AA");
  my $cipher      = new Crypt::Skip32 $key; # Always 10 bytes!

  # Encrypt an unsigned integer (under 2^32) into an 8-digit hex string.
  my $number      = 3493209676;
  my $plaintext   = pack("N", $number);
  my $ciphertext  = $cipher->encrypt($plaintext); # Always 4 bytes!
  my $cipherhex   = unpack("H8", $ciphertext);
  print "$number encrypted and converted to hex: $cipherhex\n";

  # Decrypt an encrypted, hexified unsigned integer.
  my $ciphertext2 = pack("H8", $cipherhex);
  my $plaintext2  = $cipher->decrypt($ciphertext2); # Always 4 bytes!
  my $number2     = unpack("N", $plaintext2);
  print "$cipherhex converted back and decrypted: $number2\n";

The above code generates the output:

  3493209676 encrypted and converted to hex: 6da27100
  6da27100 converted back and decrypted: 3493209676

CAVEATS This initial alpha Perl implementation of Crypt::Skip32 has not been extentively reviewed by cryptographic experts, nor has it been tested extensively on many different platforms. It is recommended that this code not be used for applications which require a high level of security.

Reviewers and testers welcomed.

Though this module has been coded to follow a Crypt::CBC usable
interface, it is not intended for use in encrypting long chunks of text.
For those purposes, it is suggested you use another high quality, proven
cipher with a longer block size.

INSTALLATION If your Linux distro does not have a prepared package for this module, then the preferred method for installation is directly from the CPAN using a command like:

    sudo cpan Crypt::Skip32

SOURCE The source for this module is being maintained on github:

Forks and patches will be reviewed, but please be aware that the
targeted functionality of this particular module is very narrow.

Feel free to build other abstractions on top of this module if you want
to make it easier to use or to create a particular application for its

BUGS Problems and feature requests can be submitted through the github "issues" link:

A gentle reminder sent directly to the author (below) may also help
increase awareness and attention.

SEE ALSO The original SKIP32 implementation in C by Greg Rose:

The 80-bit key, 64-bit block Skipjack cipher created by the NSA (Perl
code maintained by Julius C. Duque): Crypt::Skipjack


AUTHOR Perl code maintained by Eric Hammond [email protected]

Original SKIP32 C code written 1999-04-27 by Greg Rose, based on an
implementation of the Skipjack algorithm written by Panu Rissanen.

COPYRIGHT AND LICENSE Copyright (C) 2007-2011 Eric Hammond [email protected]

This library is free software; you can redistribute it and/or modify it
under the same terms as Perl itself, either Perl version 5.8.8 or, at
your option, any later version of Perl 5 you may have available.

The original C version of SKIP32 by Greg Rose (see below) is explicitly
"not copyright, no rights reserved". Even so, permission was requested
and granted to make a Perl version available on the CPAN.

ORIGINAL C SOURCE /* SKIP32 -- 32 bit block cipher based on SKIPJACK. Written by Greg Rose, QUALCOMM Australia, 1999/04/27.

    In common: F-table, G-permutation, key schedule.
    Different: 24 round feistel structure.
    Based on:  Unoptimized test implementation of SKIPJACK algorithm
               Panu Rissanen <[email protected]>

    SKIPJACK and KEA Algorithm Specifications
    Version 2.0
    29 May 1998

    Not copyright, no rights reserved.

 typedef unsigned char   BYTE; /* 8 bits */
 typedef unsigned short  WORD; /* 16 bits */

 const BYTE ftable[256] = {

 g(BYTE *key, int k, WORD w)
     BYTE g1, g2, g3, g4, g5, g6;

     g1 = (w>>8)&0xff;
     g2 = w&0xff;

     g3 = ftable[g2 ^ key[(4*k)%10]] ^ g1;
     g4 = ftable[g3 ^ key[(4*k+1)%10]] ^ g2;
     g5 = ftable[g4 ^ key[(4*k+2)%10]] ^ g3;
     g6 = ftable[g5 ^ key[(4*k+3)%10]] ^ g4;

     return ((g5<<8) + g6);

 skip32(BYTE key[10], BYTE buf[4], int encrypt)
     int         k; /* round number */
     int         i; /* round counter */
     int         kstep;
     WORD        wl, wr;

     /* sort out direction */
     if (encrypt)
         kstep = 1, k = 0;
         kstep = -1, k = 23;

     /* pack into words */
     wl = (buf[0] << 8) + buf[1];
     wr = (buf[2] << 8) + buf[3];

     /* 24 feistel rounds, doubled up */
     for (i = 0; i < 24/2; ++i) {
         wr ^= g(key, k, wl) ^ k;
         k += kstep;
         wl ^= g(key, k, wr) ^ k;
         k += kstep;

     /* implicitly swap halves while unpacking */
     buf[0] = wr >> 8;   buf[1] = wr & 0xFF;
     buf[2] = wl >> 8;   buf[3] = wl & 0xFF;

 #include <stdio.h>
 int main(int ac, char *av[])
     BYTE        in[4] = { 0x33,0x22,0x11,0x00 };
     BYTE        key[10] = { 0x00,0x99,0x88,0x77,0x66,0x55,0x44,0x33,0x22,0x11 };
     int         i, encrypt;
     int         bt;

     if (ac == 1) {
         skip32(key, in, 1);
         printf("%02x%02x%02x%02x\n", in[0], in[1], in[2], in[3]);
         if (in[0] != 0x81 || in[1] != 0x9d || in[2] != 0x5f || in[3] != 0x1f) {
             printf("819d5f1f is the answer! Didn't encrypt correctly!\n");
             return 1;
         skip32(key, in, 0);
         if (in[0] != 0x33 || in[1] != 0x22 || in[2] != 0x11 || in[3] != 0x00) {
             printf("%02x%02x%02x%02x\n", in[0], in[1], in[2], in[3]);
             printf("33221100 is the answer! Didn't decrypt correctly!\n");
             return 1;
     else if (ac != 4) {
         fprintf(stderr, "usage: %s e/d kkkkkkkkkkkkkkkkkkkk dddddddd\n", av[0]);
         return 1;
     else {
         encrypt = av[1][0] == 'e';
         for (i = 0; i < 10; ++i) {
             sscanf(&av[2][i*2], "%02x", &bt);
             key[i] = bt;
         for (i = 0; i < 4; ++i) {
             sscanf(&av[3][i*2], "%02x", &bt);
             in[i] = bt;
         skip32(key, in, encrypt);
         printf("%02x%02x%02x%02x\n", in[0], in[1], in[2], in[3]);
     return 0;

Related Repositories



Drop-in replacement for Crypt::Skip32 ...