| Sapphire.java |
1 /**
2 * Distribution License:
3 * JSword is free software; you can redistribute it and/or modify it under
4 * the terms of the GNU Lesser General Public License, version 2.1 or later
5 * as published by the Free Software Foundation. This program is distributed
6 * in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even
7 * the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
8 * See the GNU Lesser General Public License for more details.
9 *
10 * The License is available on the internet at:
11 * http://www.gnu.org/copyleft/lgpl.html
12 * or by writing to:
13 * Free Software Foundation, Inc.
14 * 59 Temple Place - Suite 330
15 * Boston, MA 02111-1307, USA
16 *
17 * © CrossWire Bible Society, 2005 - 2016
18 *
19 */
20 package org.crosswire.common.crypt;
21
22 /**
23 * The Sapphire II Stream Cipher is a port of Sword's C++ implementation of
24 * Michael Paul Johnson's 2 January 1995 public domain cipher. Below is the
25 * documentation that he originally provided for it. It has been converted to
26 * JavaDoc and the C++ fragment has been removed.
27 *
28 * <h1>THE SAPPHIRE II STREAM CIPHER</h1>
29 *
30 * <p>
31 * The Sapphire II Stream Cipher is designed to have the following properties:
32 * </p>
33 * <ul>
34 *
35 * <li>Be useful for generation of cryptographic check values as well as
36 * protecting message privacy.</li>
37 *
38 * <li>Accept a variable length key.</li>
39 *
40 * <li>Strong enough to justify <i>at least</i> a 64 bit key for balanced
41 * security.</li>
42 *
43 * <li>Small enough to be built into other applications with several keys active
44 * at once.</li>
45 *
46 * <li>Key setup fast enough to support frequent key change operations but slow
47 * enough to discourage brute force attack on the key.</li>
48 *
49 * <li>Fast enough to not significantly impact file read & write operations on
50 * most current platforms.</li>
51 *
52 * <li>Portable among common computers and efficient in C, C++, and Pascal.</li>
53 *
54 * <li>Byte oriented.</li>
55 *
56 * <li>Include both ciphertext and plain text feedback (for both optimal data
57 * hiding and value in creation of cryptographic check values).</li>
58 *
59 * <li>Acceptable performance as a pure pseudorandom number generator without
60 * providing a data stream for encryption or decryption.</li>
61 *
62 * <li>Allow <i>limited</i> key reuse without serious security degradation.</li>
63 * </ul>
64 *
65 * <h2>HISTORY AND RELATED CIPHERS</h2>
66 *
67 * <p>
68 * The Sapphire Stream Cipher is very similar to a cipher I started work on in
69 * November 1993. It is also similar in some respects to the alledged RC-4 that
70 * was posted to sci.crypt recently. Both operate on the principle of a mutating
71 * permutation vector. Alledged RC-4 doesn't include any feedback of ciphertext
72 * or plain text, however. This makes it more vulnerable to a known plain text
73 * attack, and useless for creation of cryptographic check values. On the other
74 * hand, alledged RC-4 is faster.
75 * </p>
76 *
77 * <p>
78 * The Sapphire Stream Cipher is used in the shareware product Quicrypt, which
79 * is available at ftp://ftp.csn.net/mpj/qcrypt10.zip and on the Colorado
80 * Catacombs BBS (303-772-1062). There are two versions of Quicrypt: the
81 * exportable version (with a session key limited to 32 bits but with strong
82 * user keys allowed) and the commercial North American version (with a session
83 * key of 128 bits). A variant of the Sapphire Stream Cipher is also used in the
84 * shareware program Atbash, which has no weakened exportable version.
85 * </p>
86 *
87 * <p>
88 * The Sapphire II Stream Cipher is a modification of the Sapphire Stream Cipher
89 * designed to be much more resistant to adaptive chosen plaintext attacks (with
90 * reorigination of the cipher allowed). The Sapphire II Stream Cipher is used
91 * in an encryption utility called ATBASH2.
92 * </p>
93 *
94 *
95 * <h2>OVERVIEW</h2>
96 *
97 * <p>
98 * The Sapphire Stream Cipher is based on a state machine. The state consists of
99 * 5 index values and a permutation vector. The permutation vector is simply an
100 * array containing a permutation of the numbers from 0 through 255. Four of the
101 * bytes in the permutation vector are moved to new locations (which may be the
102 * same as the old location) for every byte output. The output byte is a
103 * nonlinear function of all 5 of the index values and 8 of the bytes in the
104 * permutation vector, thus frustrating attempts to solve for the state
105 * variables based on past output. On initialization, the permutation vector
106 * (called the cards array in the source code below) is shuffled based on the
107 * user key. This shuffling is done in a way that is designed to minimize the
108 * bias in the destinations of the bytes in the array. The biggest advantage in
109 * this method is not in the elimination of the bias, per se, but in slowing
110 * down the process slightly to make brute force attack more expensive.
111 * Eliminating the bias (relative to that exhibited by RC-4) is nice, but this
112 * advantage is probably of minimal cryptographic value. The index variables are
113 * set (somewhat arbitrarily) to the permutation vector elements at locations 1,
114 * 3, 5, 7, and a key dependent value (rsum) left over from the shuffling of the
115 * permutation vector (cards array).
116 * </p>
117 *
118 *
119 * <h2>KEY SETUP</h2>
120 *
121 * <p>
122 * Key setup (illustrated by the function initialize(), below) consists of three
123 * parts:
124 * </p>
125 * <ol>
126 * <li>Initialize the index variables.</li>
127 * <li>Set the permutation vector to a known state (a simple counting sequence).
128 * </li>
129 * <li>Starting at the end of the vector, swap each element of the permutation
130 * vector with an element indexed somewhere from 0 to the current index (chosen
131 * by the function keyrand()).</li>
132 * </ol>
133 * <p>
134 * The keyrand() function returns a value between 0 and some maximum number
135 * based on the user's key, the current state of the permutation vector, and an
136 * index running sum called rsum. Note that the length of the key is used in
137 * keyrand(), too, so that a key like "abcd" will not result in the same
138 * permutation as a key like "abcdabcd".
139 * </p>
140 *
141 *
142 * <h2>ENCRYPTION</h2>
143 *
144 * <p>
145 * Each encryption involves updating the index values, moving (up to) 4 bytes
146 * around in the permutation vector, selecting an output byte, and adding the
147 * output byte bitwise modulo-2 (exclusive-or) to the plain text byte to produce
148 * the cipher text byte. The index values are incremented by different rules.
149 * The index called rotor just increases by one (modulo 256) each time. Ratchet
150 * increases by the value in the permutation vector pointed to by rotor.
151 * Avalanche increases by the value in the permutation vector pointed to by
152 * another byte in the permutation vector pointed to by the last cipher text
153 * byte. The last plain text and the last cipher text bytes are also kept as
154 * index variables. See the function called encrypt(), below for details.
155 * </p>
156 *
157 *
158 * <h2>PSUEDORANDOM BYTE GENERATION</h2>
159 *
160 * <p>
161 * If you want to generate random numbers without encrypting any particular
162 * ciphertext, simply encrypt 0. There is still plenty of complexity left in the
163 * system to ensure unpredictability (if the key is not known) of the output
164 * stream when this simplification is made.
165 * </p>
166 *
167 *
168 * <h2>DECRYPTION</h2>
169 *
170 * <p>
171 * Decryption is the same as encryption, except for the obvious swapping of the
172 * assignments to last_plain and last_cipher and the return value. See the
173 * function decrypt(), below.
174 * </p>
175 *
176 *
177 * <h2>C++ SOURCE CODE FRAGMENT</h2>
178 *
179 * <p>
180 * The original implimentation of this cipher was in Object Oriented Pascal, but
181 * C++ is available for more platforms.
182 * </p>
183 *
184 * <h2>GENERATION OF CRYPTOGRAPHIC CHECK VALUES (HASH VALUES)</h2>
185 *
186 * <p>
187 * For a fast way to generate a cryptographic check value (also called a hash or
188 * message integrity check value) of a message of arbitrary length:
189 * </p>
190 * <ol>
191 * <li>Initialize either with a key (for a keyed hash value) or call hash_init
192 * with no key (for a public hash value).</li>
193 *
194 * <li>Encrypt all of the bytes of the message or file to be hashed. The results
195 * of the encryption need not be kept for the hash generation process.
196 * (Optionally decrypt encrypted text, here).</li>
197 *
198 * <li>Call hash_final, which will further "stir" the permutation vector by
199 * encrypting the values from 255 down to 0, then report the results of
200 * encrypting 20 zeroes.</li>
201 * </ol>
202 *
203 * <h2>SECURITY ANALYSIS</h2>
204 *
205 * <p>
206 * There are several security issues to be considered. Some are easier to
207 * analyze than others. The following includes more "hand waving" than
208 * mathematical proofs, and looks more like it was written by an engineer than a
209 * mathematician. The reader is invited to improve upon or refute the following,
210 * as appropriate.
211 * </p>
212 *
213 *
214 * <h2>KEY LENGTH</h2>
215 *
216 * <p>
217 * There are really two kinds of user keys to consider: (1) random binary keys,
218 * and (2) pass phrases. Analysis of random binary keys is fairly straight
219 * forward. Pass phrases tend to have much less entropy per byte, but the
220 * analysis made for random binary keys applies to the entropy in the pass
221 * phrase. The length limit of the key (255 bytes) is adequate to allow a pass
222 * phrase with enough entropy to be considered strong.
223 * </p>
224 *
225 * <p>
226 * To be real generous to a cryptanalyst, assume dedicated Sapphire Stream
227 * Cipher cracking hardware. The constant portion of the key scheduling can be
228 * done in one cycle. That leaves at least 256 cycles to do the swapping
229 * (probably more, because of the intricacies of keyrand(), but we'll ignore
230 * that, too, for now). Assume a machine clock of about 256 MegaHertz (fairly
231 * generous). That comes to about one key tried per microsecond. On average, you
232 * only have to try half of the keys. Also assume that trying the key to see if
233 * it works can be pipelined, so that it doesn't add time to the estimate. Based
234 * on these assumptions (reasonable for major governments), and rounding to two
235 * significant digits, the following key length versus cracking time estimates
236 * result:
237 * </p>
238 *
239 * <pre>
240 * Key length, bits Time to crack
241 * ---------------- -------------
242 * 32 35 minutes (exportable in qcrypt)
243 * 33 1.2 hours (not exportable in qcrypt)
244 * 40 6.4 days
245 * 56 1,100 years (kind of like DES's key)
246 * 64 290,000 years (good enough for most things)
247 * 80 19 billion years (kind of like Skipjack's key)
248 * 128 5.4E24 years (good enough for the clinically paranoid)
249 * </pre>
250 *
251 * <p>
252 * Naturally, the above estimates can vary by several orders of magnitude based
253 * on what you assume for attacker's hardware, budget, and motivation.
254 * </p>
255 *
256 * <p>
257 * In the range listed above, the probability of spare keys (two keys resulting
258 * in the same initial permutation vector) is small enough to ignore. The proof
259 * is left to the reader.
260 * </p>
261 *
262 *
263 * <h2>INTERNAL STATE SPACE</h2>
264 *
265 * <p>
266 * For a stream cipher, internal state space should be at least as big as the
267 * number of possible keys to be considered strong. The state associated with
268 * the permutation vector alone (256!) constitutes overkill.
269 * </p>
270 *
271 *
272 * <h2>PREDICTABILITY OF THE STATE</h2>
273 *
274 * <p>
275 * If you have a history of stream output from initialization (or equivalently,
276 * previous known plaintext and ciphertext), then rotor, last_plain, and
277 * last_cipher are known to an attacker. The other two index values, flipper and
278 * avalanche, cannot be solved for without knowing the contents of parts of the
279 * permutation vector that change with each byte encrypted. Solving for the
280 * contents of the permutation vector by keeping track of the possible positions
281 * of the index variables and possible contents of the permutation vector at
282 * each byte position is not possible, since more variables than known values
283 * are generated at each iteration. Indeed, fewer index variables and swaps
284 * could be used to achieve security, here, if it were not for the hash
285 * requirements.
286 * </p>
287 *
288 *
289 * <h2>CRYPTOGRAPHIC CHECK VALUE</h2>
290 *
291 * <p>
292 * The change in state altered with each byte encrypted contributes to an
293 * avalanche of generated check values that is radically different after a
294 * sequence of at least 64 bytes have been encrypted. The suggested way to
295 * create a cryptographic check value is to encrypt all of the bytes of a
296 * message, then encrypt a sequence of bytes counting down from 255 to 0. A
297 * single bit change in a message causes a radical change in the check value
298 * generated (about half of the bits change). This is an essential feature of a
299 * cryptographic check value.
300 * </p>
301 *
302 * <p>
303 * Another good property of a cryptographic check value is that it is too hard
304 * to compute a message that results in a certain check value. In this case, we
305 * assume the attacker knows the key and the contents of a message that has the
306 * desired check value, and wants to compute a bogus message having the same
307 * check value. There are two obvious ways to do this attack. One is to solve
308 * for a sequence that will restore the state of the permutation vector and
309 * indices back to what it was before the alteration. The other one is the
310 * so-called "birthday" attack that is to cryptographic hash functions what
311 * brute force is to key search.
312 * </p>
313 *
314 * <p>
315 * To generate a sequence that restores the state of the cipher to what it was
316 * before the alteration probably requires at least 256 bytes, since the index
317 * "rotor" marches steadily on its cycle, one by one. The values to do this
318 * cannot easily be computed, due to the nonlinearity of the feedback, so there
319 * would probably have to be lots of trial and error involved. In practical
320 * applications, this would leave a gaping block of binary garbage in the middle
321 * of a document, and would be quite obvious, so this is not a practical attack,
322 * even if you could figure out how to do it (and I haven't). If anyone has a
323 * method to solve for such a block of data, though, I would be most interested
324 * in finding out what it is. Please email me at
325 * <m.p.johnson@ieee.org> if you find one.
326 * </p>
327 *
328 * <p>
329 * The "birthday" attack just uses the birthday paradox to find a message that
330 * has the same check value. With a 20 byte check value, you would have to find
331 * at least 80 bits to change in the text such that they wouldn't be noticed (a
332 * plausible situation), then try the combinations until one matches. 2 to the
333 * 80th power is a big number, so this isn't practical either. If this number
334 * isn't big enough, you are free to generate a longer check value with this
335 * algorithm. Someone who likes 16 byte keys might prefer 32 byte check values
336 * for similar stringth.
337 * </p>
338 *
339 *
340 * <h2>ADAPTIVE CHOSEN PLAIN TEXT ATTACKS</h2>
341 *
342 * <p>
343 * Let us give the attacker a keyed black box that accepts any input and
344 * provides the corresponding output. Let us also provide a signal to the black
345 * box that causes it to reoriginate (revert to its initial keyed state) at the
346 * attacker's will. Let us also be really generous and provide a free copy of
347 * the black box, identical in all respects except that the key is not provided
348 * and it is not locked, so the array can be manipulated directly.
349 * </p>
350 *
351 * <p>
352 * Since each byte encrypted only modifies at most 5 of the 256 bytes in the
353 * permutation vector, and it is possible to find different sequences of two
354 * bytes that leave the five index variables the same, it is possible for the
355 * attacker to find sets of chosen plain texts that differ in two bytes, but
356 * which have cipher texts that are the same for several of the subsequent
357 * bytes. Modeling indicates that as many as ten of the following bytes
358 * (although not necessarily the next ten bytes) might match. This information
359 * would be useful in determining the structure of the Sapphire Stream Cipher
360 * based on a captured, keyed black box. This means that it would not be a good
361 * substitute for the Skipjack algorithm in the EES, but we assume that the
362 * attacker already knows the algorithm, anyway. This departure from the
363 * statistics expected from an ideal stream cipher with feedback doesn't seem to
364 * be useful for determining any key bytes or permutation vector bytes, but it
365 * is the reason why post-conditioning is required when computing a
366 * cryptographic hash with the Sapphire Stream Cipher. Thanks to Bryan G.
367 * Olson's <olson@umbc.edu> continued attacks on the Sapphire Stream
368 * Cipher, I have come up with the Sapphire II Stream Cipher. Thanks again to
369 * Bryan for his valuable help.
370 * </p>
371 *
372 * <p>
373 * Bryan Olson's "differential" attack of the original Sapphire Stream Cipher
374 * relies on both of these facts:
375 * </p>
376 *
377 * <ol>
378 * <li>By continual reorigination of a black box containing a keyed version of
379 * the Sapphire Stream Cipher, it is possible to find a set of input strings
380 * that differ only in the first two (or possibly three) bytes that have
381 * identical output after the first three (or possibly four) bytes. The output
382 * suffixes so obtained will not contain the values of the permutation vector
383 * bytes that <i>differ</i> because of the different initial bytes encrypted.</li>
384 *
385 * <li>Because the five index values are initialized to constants that are known
386 * by the attacker, most of the locations of the "missing" bytes noted in the
387 * above paragraph are known to the attacker (except for those indexed by the
388 * ratchet index variable for encryptions after the first byte).</li>
389 * </ol>
390 *
391 * <p>
392 * I have not yet figured out if Bryan's attack on the original Sapphire Stream
393 * Cipher had complexity of more or less than the design strength goal of 2^64
394 * encryptions, but some conservative estimations I made showed that it could
395 * possibly come in significantly less than that. (I would probably have to
396 * develop a full practical attack to accurately estimate the complexity more
397 * accurately, and I have limited time for that). Fortunately, there is a way to
398 * frustrate this type of attack without fully developing it.
399 * </p>
400 *
401 * <p>
402 * Denial of condition 1 above by increased alteration of the state variables is
403 * too costly, at least using the methods I tried. For example, doubling the
404 * number of index variables and the number of permutation vector items
405 * referenced in the output function of the stream cipher provides only doubles
406 * the cost of getting the data in item 1, above. This is bad crypto-economics.
407 * A better way is to change the output function such that the stream cipher
408 * output byte is a combination of two permutation vector bytes instead of one.
409 * That means that all possible output values can occur in the differential
410 * sequences of item 1, above.
411 * </p>
412 *
413 * <p>
414 * Denial of condition 2 above, is simpler. By making the initial values of the
415 * five index variables dependent on the key, Bryan's differential attack is
416 * defeated, since the attacker has no idea which elements of the permutation
417 * vector were different between data sets, and exhaustive search is too
418 * expensive.
419 * </p>
420 *
421 *
422 * <h2>OTHER HOLES</h2>
423 *
424 * <p>
425 * Are there any? Take you best shot and let me know if you see any. I offer no
426 * challenge text with this algorithm, but you are free to use it without
427 * royalties to me if it is any good.
428 * </p>
429 *
430 *
431 * <h2>CURRENT STATUS</h2>
432 *
433 * <p>
434 * This is a new (to the public) cipher, and an even newer approach to
435 * cryptographic hash generation. Take your best shot at it, and please let me
436 * know if you find any weaknesses (proven or suspected) in it. Use it with
437 * caution, but it still looks like it fills a need for reasonably strong
438 * cryptography with limited resources.
439 * </p>
440 *
441 *
442 * <h2>LEGAL STUFF</h2>
443 *
444 * <p>
445 * The intention of this document is to share some research results on an
446 * informal basis. You may freely use the algorithm and code listed above as far
447 * as I'm concerned, as long as you don't sue me for anything, but there may be
448 * other restrictions that I am not aware of to your using it. The C++ code
449 * fragment above is just intended to illustrate the algorithm being discussed,
450 * and is not a complete application. I understand this document to be
451 * Constitutionally protected publication, and not a munition, but don't blame
452 * me if it explodes or has toxic side effects.
453 * </p>
454 *
455 * <pre>
456 * ___________________________________________________________
457 * | |
458 * |\ /| | | Michael Paul Johnson Colorado Catacombs BBS 303-772-1062 |
459 * | \/ |o| | PO Box 1151, Longmont CO 80502-1151 USA John 3:16-17 |
460 * | | | / _ | mpj@csn.org aka mpj@netcom.com m.p.johnson@ieee.org |
461 * | |||/ /_\ | ftp://ftp.csn.net/mpj/README.MPJ CIS: 71331,2332 |
462 * | |||\ ( | ftp://ftp.netcom.com/pub/mp/mpj/README -. --- ----- .... |
463 * | ||| \ \_/ | PGPprint=F2 5E A1 C1 A6 CF EF 71 12 1F 91 92 6A ED AE A9 |
464 * |___________________________________________________________|
465 * </pre>
466 *
467 * Regarding this port to Java and not the original code, the following license
468 * applies:
469 *
470 * @see gnu.lgpl.License The GNU Lesser General Public License for details.
471 * @author Michael Paul Johnson [ kahunapule at mpj dot cx] Original code
472 * @author unascribed Sword's C++ implementation
473 * @author DM Smith Java port from Sword's C++ implementation
474 */
475 public class Sapphire {
476
477 /**
478 * Construct a Sapphire Stream Cipher from a key, possibly null or empty.
479 *
480 * @param aKey the cipher key
481 */
482 public Sapphire(byte[] aKey) {
483 byte[] key = aKey;
484 if (key == null) {
485 key = new byte[0];
486 }
487 cards = new int[256];
488 if (key.length > 0) {
489 initialize(key);
490 } else {
491 hashInit();
492 }
493 }
494
495 /**
496 * Decipher a single byte, presumably the next.
497 *
498 * @param b
499 * the next byte to decipher
500 * @return the enciphered byte
501 */
502 public byte cipher(byte b) {
503 // Picture a single enigma rotor with 256 positions, rewired
504 // on the fly by card-shuffling.
505
506 // This cipher is a variant of one invented and written
507 // by Michael Paul Johnson in November, 1993.
508
509 // Shuffle the deck a little more.
510
511 // Convert from a byte to an int, but prevent sign extension.
512 // So -16 becomes 240
513 int bVal = b & 0xFF;
514 ratchet += cards[rotor++];
515 // Keep ratchet and rotor in the range of 0-255
516 // The C++ code relied upon overflow of an unsigned char
517 ratchet &= 0xFF;
518 rotor &= 0xFF;
519 int swaptemp = cards[lastCipher];
520 cards[lastCipher] = cards[ratchet];
521 cards[ratchet] = cards[lastPlain];
522 cards[lastPlain] = cards[rotor];
523 cards[rotor] = swaptemp;
524 avalanche += cards[swaptemp];
525 // Keep avalanche in the range of 0-255
526 avalanche &= 0xFF;
527
528 // Output one byte from the state in such a way as to make it
529 // very hard to figure out which one you are looking at.
530 lastPlain = bVal ^ cards[(cards[ratchet] + cards[rotor]) & 0xFF] ^ cards[cards[(cards[lastPlain] + cards[lastCipher] + cards[avalanche]) & 0xFF]];
531
532 lastCipher = bVal;
533
534 // Convert back to a byte
535 // E.g. 240 becomes -16
536 return (byte) lastPlain;
537 }
538
539 /**
540 * Destroy the key and state information in RAM.
541 */
542 public void burn() {
543 // Destroy the key and state information in RAM.
544 for (int i = 0; i < 256; i++) {
545 cards[i] = 0;
546 }
547 rotor = 0;
548 ratchet = 0;
549 avalanche = 0;
550 lastPlain = 0;
551 lastCipher = 0;
552 }
553
554 /**
555 * @param hash the destination
556 */
557 public void hashFinal(byte[] hash) { // Destination
558 for (int i = 255; i >= 0; i--) {
559 cipher((byte) i);
560 }
561 for (int i = 0; i < hash.length; i++) {
562 hash[i] = cipher((byte) 0);
563 }
564 }
565
566 /**
567 * Initializes the cards array to be deterministically random based upon the
568 * key.
569 * <p>
570 * Key size may be up to 256 bytes. Pass phrases may be used directly, with
571 * longer length compensating for the low entropy expected in such keys.
572 * Alternatively, shorter keys hashed from a pass phrase or generated
573 * randomly may be used. For random keys, lengths of from 4 to 16 bytes are
574 * recommended, depending on how secure you want this to be.
575 * </p>
576 *
577 * @param key
578 * used to initialize the cipher engine.
579 */
580 private void initialize(byte[] key) {
581
582 // Start with cards all in order, one of each.
583 for (int i = 0; i < 256; i++) {
584 cards[i] = i;
585 }
586
587 // Swap the card at each position with some other card.
588 int swaptemp;
589 int toswap = 0;
590 keypos = 0; // Start with first byte of user key.
591 rsum = 0;
592 for (int i = 255; i >= 0; i--) {
593 toswap = keyrand(i, key);
594 swaptemp = cards[i];
595 cards[i] = cards[toswap];
596 cards[toswap] = swaptemp;
597 }
598
599 // Initialize the indices and data dependencies.
600 // Indices are set to different values instead of all 0
601 // to reduce what is known about the state of the cards
602 // when the first byte is emitted.
603 rotor = cards[1];
604 ratchet = cards[3];
605 avalanche = cards[5];
606 lastPlain = cards[7];
607 lastCipher = cards[rsum];
608
609 // ensure that these have no useful values to those that snoop
610 toswap = 0;
611 swaptemp = toswap;
612 rsum = swaptemp;
613 keypos = rsum;
614 }
615
616 /**
617 * Initialize non-keyed hash computation.
618 */
619 private void hashInit() {
620
621 // Initialize the indices and data dependencies.
622 rotor = 1;
623 ratchet = 3;
624 avalanche = 5;
625 lastPlain = 7;
626 lastCipher = 11;
627
628 // Start with cards all in inverse order.
629
630 int j = 255;
631 for (int i = 0; i < 256; i++) {
632 cards[i] = j--;
633 }
634 }
635
636 private int keyrand(int limit, byte[] key) {
637 int u; // Value from 0 to limit to return.
638
639 if (limit == 0) {
640 return 0; // Avoid divide by zero error.
641 }
642
643 int retryLimiter = 0; // No infinite loops allowed.
644
645 // Fill mask with enough bits to cover the desired range.
646 int mask = 1;
647 while (mask < limit) {
648 mask = (mask << 1) + 1;
649 }
650
651 do {
652 // Convert a byte from the key to an int, but prevent sign
653 // extension.
654 // So -16 becomes 240
655 // Also keep rsum in the range of 0-255
656 // The C++ code relied upon overflow of an unsigned char
657 rsum = (cards[rsum] + (key[keypos++] & 0xFF)) & 0xFF;
658
659 if (keypos >= key.length) {
660 keypos = 0; // Recycle the user key.
661 rsum += key.length; // key "aaaa" != key "aaaaaaaa"
662 rsum &= 0xFF;
663 }
664
665 u = mask & rsum;
666
667 if (++retryLimiter > 11) {
668 u %= limit; // Prevent very rare long loops.
669 }
670 } while (u > limit);
671 return u;
672 }
673
674 private int[] cards;
675 private int rotor;
676 private int ratchet;
677 private int avalanche;
678 private int lastPlain;
679 private int lastCipher;
680 private int keypos;
681 private int rsum;
682 }
683