corrade-vassal – Blame information for rev 1

Subversion Repositories:
Rev:
Rev Author Line No. Line
1 vero 1 /*
2 * Copyright (c) 2001-2003, David Janssens
3 * Copyright (c) 2002-2003, Yannick Verschueren
4 * Copyright (c) 2003-2005, Francois Devaux and Antonin Descampe
5 * Copyright (c) 2005, Herve Drolon, FreeImage Team
6 * Copyright (c) 2002-2005, Communications and remote sensing Laboratory, Universite catholique de Louvain, Belgium
7 * Copyright (c) 2005-2006, Dept. of Electronic and Information Engineering, Universita' degli Studi di Perugia, Italy
8 * All rights reserved.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS `AS IS'
20 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
23 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31  
32 #ifdef USE_JPWL
33  
34 /**
35 @file rs.c
36 @brief Functions used to compute the Reed-Solomon parity and check of byte arrays
37  
38 */
39  
40 /**
41 * Reed-Solomon coding and decoding
42 * Phil Karn (karn@ka9q.ampr.org) September 1996
43 *
44 * This file is derived from the program "new_rs_erasures.c" by Robert
45 * Morelos-Zaragoza (robert@spectra.eng.hawaii.edu) and Hari Thirumoorthy
46 * (harit@spectra.eng.hawaii.edu), Aug 1995
47 *
48 * I've made changes to improve performance, clean up the code and make it
49 * easier to follow. Data is now passed to the encoding and decoding functions
50 * through arguments rather than in global arrays. The decode function returns
51 * the number of corrected symbols, or -1 if the word is uncorrectable.
52 *
53 * This code supports a symbol size from 2 bits up to 16 bits,
54 * implying a block size of 3 2-bit symbols (6 bits) up to 65535
55 * 16-bit symbols (1,048,560 bits). The code parameters are set in rs.h.
56 *
57 * Note that if symbols larger than 8 bits are used, the type of each
58 * data array element switches from unsigned char to unsigned int. The
59 * caller must ensure that elements larger than the symbol range are
60 * not passed to the encoder or decoder.
61 *
62 */
63 #include <stdio.h>
64 #include <stdlib.h>
65 #include "rs.h"
66  
67 /* This defines the type used to store an element of the Galois Field
68 * used by the code. Make sure this is something larger than a char if
69 * if anything larger than GF(256) is used.
70 *
71 * Note: unsigned char will work up to GF(256) but int seems to run
72 * faster on the Pentium.
73 */
74 typedef int gf;
75  
76 /* KK = number of information symbols */
77 static int KK;
78  
79 /* Primitive polynomials - see Lin & Costello, Appendix A,
80 * and Lee & Messerschmitt, p. 453.
81 */
82 #if(MM == 2)/* Admittedly silly */
83 int Pp[MM+1] = { 1, 1, 1 };
84  
85 #elif(MM == 3)
86 /* 1 + x + x^3 */
87 int Pp[MM+1] = { 1, 1, 0, 1 };
88  
89 #elif(MM == 4)
90 /* 1 + x + x^4 */
91 int Pp[MM+1] = { 1, 1, 0, 0, 1 };
92  
93 #elif(MM == 5)
94 /* 1 + x^2 + x^5 */
95 int Pp[MM+1] = { 1, 0, 1, 0, 0, 1 };
96  
97 #elif(MM == 6)
98 /* 1 + x + x^6 */
99 int Pp[MM+1] = { 1, 1, 0, 0, 0, 0, 1 };
100  
101 #elif(MM == 7)
102 /* 1 + x^3 + x^7 */
103 int Pp[MM+1] = { 1, 0, 0, 1, 0, 0, 0, 1 };
104  
105 #elif(MM == 8)
106 /* 1+x^2+x^3+x^4+x^8 */
107 int Pp[MM+1] = { 1, 0, 1, 1, 1, 0, 0, 0, 1 };
108  
109 #elif(MM == 9)
110 /* 1+x^4+x^9 */
111 int Pp[MM+1] = { 1, 0, 0, 0, 1, 0, 0, 0, 0, 1 };
112  
113 #elif(MM == 10)
114 /* 1+x^3+x^10 */
115 int Pp[MM+1] = { 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1 };
116  
117 #elif(MM == 11)
118 /* 1+x^2+x^11 */
119 int Pp[MM+1] = { 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1 };
120  
121 #elif(MM == 12)
122 /* 1+x+x^4+x^6+x^12 */
123 int Pp[MM+1] = { 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1 };
124  
125 #elif(MM == 13)
126 /* 1+x+x^3+x^4+x^13 */
127 int Pp[MM+1] = { 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1 };
128  
129 #elif(MM == 14)
130 /* 1+x+x^6+x^10+x^14 */
131 int Pp[MM+1] = { 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1 };
132  
133 #elif(MM == 15)
134 /* 1+x+x^15 */
135 int Pp[MM+1] = { 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 };
136  
137 #elif(MM == 16)
138 /* 1+x+x^3+x^12+x^16 */
139 int Pp[MM+1] = { 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1 };
140  
141 #else
142 #error "MM must be in range 2-16"
143 #endif
144  
145 /* Alpha exponent for the first root of the generator polynomial */
146 #define B0 0 /* Different from the default 1 */
147  
148 /* index->polynomial form conversion table */
149 gf Alpha_to[NN + 1];
150  
151 /* Polynomial->index form conversion table */
152 gf Index_of[NN + 1];
153  
154 /* No legal value in index form represents zero, so
155 * we need a special value for this purpose
156 */
157 #define A0 (NN)
158  
159 /* Generator polynomial g(x)
160 * Degree of g(x) = 2*TT
161 * has roots @**B0, @**(B0+1), ... ,@^(B0+2*TT-1)
162 */
163 /*gf Gg[NN - KK + 1];*/
164 gf Gg[NN - 1];
165  
166 /* Compute x % NN, where NN is 2**MM - 1,
167 * without a slow divide
168 */
169 static /*inline*/ gf
170 modnn(int x)
171 {
172 while (x >= NN) {
173 x -= NN;
174 x = (x >> MM) + (x & NN);
175 }
176 return x;
177 }
178  
179 /*#define min(a,b) ((a) < (b) ? (a) : (b))*/
180  
181 #define CLEAR(a,n) {\
182 int ci;\
183 for(ci=(n)-1;ci >=0;ci--)\
184 (a)[ci] = 0;\
185 }
186  
187 #define COPY(a,b,n) {\
188 int ci;\
189 for(ci=(n)-1;ci >=0;ci--)\
190 (a)[ci] = (b)[ci];\
191 }
192 #define COPYDOWN(a,b,n) {\
193 int ci;\
194 for(ci=(n)-1;ci >=0;ci--)\
195 (a)[ci] = (b)[ci];\
196 }
197  
198 void init_rs(int k)
199 {
200 KK = k;
201 if (KK >= NN) {
202 printf("KK must be less than 2**MM - 1\n");
203 exit(1);
204 }
205  
206 generate_gf();
207 gen_poly();
208 }
209  
210 /* generate GF(2**m) from the irreducible polynomial p(X) in p[0]..p[m]
211 lookup tables: index->polynomial form alpha_to[] contains j=alpha**i;
212 polynomial form -> index form index_of[j=alpha**i] = i
213 alpha=2 is the primitive element of GF(2**m)
214 HARI's COMMENT: (4/13/94) alpha_to[] can be used as follows:
215 Let @ represent the primitive element commonly called "alpha" that
216 is the root of the primitive polynomial p(x). Then in GF(2^m), for any
217  
218 @^i = a(0) + a(1) @ + a(2) @^2 + ... + a(m-1) @^(m-1)
219 where the binary vector (a(0),a(1),a(2),...,a(m-1)) is the representation
220 of the integer "alpha_to[i]" with a(0) being the LSB and a(m-1) the MSB. Thus for
221 example the polynomial representation of @^5 would be given by the binary
222 representation of the integer "alpha_to[5]".
223 Similarily, index_of[] can be used as follows:
224 As above, let @ represent the primitive element of GF(2^m) that is
225 the root of the primitive polynomial p(x). In order to find the power
226 of @ (alpha) that has the polynomial representation
227 a(0) + a(1) @ + a(2) @^2 + ... + a(m-1) @^(m-1)
228 we consider the integer "i" whose binary representation with a(0) being LSB
229 and a(m-1) MSB is (a(0),a(1),...,a(m-1)) and locate the entry
230 "index_of[i]". Now, @^index_of[i] is that element whose polynomial
231 representation is (a(0),a(1),a(2),...,a(m-1)).
232 NOTE:
233 The element alpha_to[2^m-1] = 0 always signifying that the
234 representation of "@^infinity" = 0 is (0,0,0,...,0).
235 Similarily, the element index_of[0] = A0 always signifying
236 that the power of alpha which has the polynomial representation
237 (0,0,...,0) is "infinity".
238  
239 */
240  
241 void
242 generate_gf(void)
243 {
244 register int i, mask;
245  
246 mask = 1;
247 Alpha_to[MM] = 0;
248 for (i = 0; i < MM; i++) {
249 Alpha_to[i] = mask;
250 Index_of[Alpha_to[i]] = i;
251 /* If Pp[i] == 1 then, term @^i occurs in poly-repr of @^MM */
252 if (Pp[i] != 0)
253 Alpha_to[MM] ^= mask; /* Bit-wise EXOR operation */
254 mask <<= 1; /* single left-shift */
255 }
256 Index_of[Alpha_to[MM]] = MM;
257 /*
258 * Have obtained poly-repr of @^MM. Poly-repr of @^(i+1) is given by
259 * poly-repr of @^i shifted left one-bit and accounting for any @^MM
260 * term that may occur when poly-repr of @^i is shifted.
261 */
262 mask >>= 1;
263 for (i = MM + 1; i < NN; i++) {
264 if (Alpha_to[i - 1] >= mask)
265 Alpha_to[i] = Alpha_to[MM] ^ ((Alpha_to[i - 1] ^ mask) << 1);
266 else
267 Alpha_to[i] = Alpha_to[i - 1] << 1;
268 Index_of[Alpha_to[i]] = i;
269 }
270 Index_of[0] = A0;
271 Alpha_to[NN] = 0;
272 }
273  
274  
275 /*
276 * Obtain the generator polynomial of the TT-error correcting, length
277 * NN=(2**MM -1) Reed Solomon code from the product of (X+@**(B0+i)), i = 0,
278 * ... ,(2*TT-1)
279 *
280 * Examples:
281 *
282 * If B0 = 1, TT = 1. deg(g(x)) = 2*TT = 2.
283 * g(x) = (x+@) (x+@**2)
284 *
285 * If B0 = 0, TT = 2. deg(g(x)) = 2*TT = 4.
286 * g(x) = (x+1) (x+@) (x+@**2) (x+@**3)
287 */
288 void
289 gen_poly(void)
290 {
291 register int i, j;
292  
293 Gg[0] = Alpha_to[B0];
294 Gg[1] = 1; /* g(x) = (X+@**B0) initially */
295 for (i = 2; i <= NN - KK; i++) {
296 Gg[i] = 1;
297 /*
298 * Below multiply (Gg[0]+Gg[1]*x + ... +Gg[i]x^i) by
299 * (@**(B0+i-1) + x)
300 */
301 for (j = i - 1; j > 0; j--)
302 if (Gg[j] != 0)
303 Gg[j] = Gg[j - 1] ^ Alpha_to[modnn((Index_of[Gg[j]]) + B0 + i - 1)];
304 else
305 Gg[j] = Gg[j - 1];
306 /* Gg[0] can never be zero */
307 Gg[0] = Alpha_to[modnn((Index_of[Gg[0]]) + B0 + i - 1)];
308 }
309 /* convert Gg[] to index form for quicker encoding */
310 for (i = 0; i <= NN - KK; i++)
311 Gg[i] = Index_of[Gg[i]];
312 }
313  
314  
315 /*
316 * take the string of symbols in data[i], i=0..(k-1) and encode
317 * systematically to produce NN-KK parity symbols in bb[0]..bb[NN-KK-1] data[]
318 * is input and bb[] is output in polynomial form. Encoding is done by using
319 * a feedback shift register with appropriate connections specified by the
320 * elements of Gg[], which was generated above. Codeword is c(X) =
321 * data(X)*X**(NN-KK)+ b(X)
322 */
323 int
324 encode_rs(dtype *data, dtype *bb)
325 {
326 register int i, j;
327 gf feedback;
328  
329 CLEAR(bb,NN-KK);
330 for (i = KK - 1; i >= 0; i--) {
331 #if (MM != 8)
332 if(data[i] > NN)
333 return -1; /* Illegal symbol */
334 #endif
335 feedback = Index_of[data[i] ^ bb[NN - KK - 1]];
336 if (feedback != A0) { /* feedback term is non-zero */
337 for (j = NN - KK - 1; j > 0; j--)
338 if (Gg[j] != A0)
339 bb[j] = bb[j - 1] ^ Alpha_to[modnn(Gg[j] + feedback)];
340 else
341 bb[j] = bb[j - 1];
342 bb[0] = Alpha_to[modnn(Gg[0] + feedback)];
343 } else { /* feedback term is zero. encoder becomes a
344 * single-byte shifter */
345 for (j = NN - KK - 1; j > 0; j--)
346 bb[j] = bb[j - 1];
347 bb[0] = 0;
348 }
349 }
350 return 0;
351 }
352  
353 /*
354 * Performs ERRORS+ERASURES decoding of RS codes. If decoding is successful,
355 * writes the codeword into data[] itself. Otherwise data[] is unaltered.
356 *
357 * Return number of symbols corrected, or -1 if codeword is illegal
358 * or uncorrectable.
359 *
360 * First "no_eras" erasures are declared by the calling program. Then, the
361 * maximum # of errors correctable is t_after_eras = floor((NN-KK-no_eras)/2).
362 * If the number of channel errors is not greater than "t_after_eras" the
363 * transmitted codeword will be recovered. Details of algorithm can be found
364 * in R. Blahut's "Theory ... of Error-Correcting Codes".
365 */
366 int
367 eras_dec_rs(dtype *data, int *eras_pos, int no_eras)
368 {
369 int deg_lambda, el, deg_omega;
370 int i, j, r;
371 gf u,q,tmp,num1,num2,den,discr_r;
372 gf recd[NN];
373 /* Err+Eras Locator poly and syndrome poly */
374 /*gf lambda[NN-KK + 1], s[NN-KK + 1];
375 gf b[NN-KK + 1], t[NN-KK + 1], omega[NN-KK + 1];
376 gf root[NN-KK], reg[NN-KK + 1], loc[NN-KK];*/
377 gf lambda[NN + 1], s[NN + 1];
378 gf b[NN + 1], t[NN + 1], omega[NN + 1];
379 gf root[NN], reg[NN + 1], loc[NN];
380 int syn_error, count;
381  
382 /* data[] is in polynomial form, copy and convert to index form */
383 for (i = NN-1; i >= 0; i--){
384 #if (MM != 8)
385 if(data[i] > NN)
386 return -1; /* Illegal symbol */
387 #endif
388 recd[i] = Index_of[data[i]];
389 }
390 /* first form the syndromes; i.e., evaluate recd(x) at roots of g(x)
391 * namely @**(B0+i), i = 0, ... ,(NN-KK-1)
392 */
393 syn_error = 0;
394 for (i = 1; i <= NN-KK; i++) {
395 tmp = 0;
396 for (j = 0; j < NN; j++)
397 if (recd[j] != A0) /* recd[j] in index form */
398 tmp ^= Alpha_to[modnn(recd[j] + (B0+i-1)*j)];
399 syn_error |= tmp; /* set flag if non-zero syndrome =>
400 * error */
401 /* store syndrome in index form */
402 s[i] = Index_of[tmp];
403 }
404 if (!syn_error) {
405 /*
406 * if syndrome is zero, data[] is a codeword and there are no
407 * errors to correct. So return data[] unmodified
408 */
409 return 0;
410 }
411 CLEAR(&lambda[1],NN-KK);
412 lambda[0] = 1;
413 if (no_eras > 0) {
414 /* Init lambda to be the erasure locator polynomial */
415 lambda[1] = Alpha_to[eras_pos[0]];
416 for (i = 1; i < no_eras; i++) {
417 u = eras_pos[i];
418 for (j = i+1; j > 0; j--) {
419 tmp = Index_of[lambda[j - 1]];
420 if(tmp != A0)
421 lambda[j] ^= Alpha_to[modnn(u + tmp)];
422 }
423 }
424 #ifdef ERASURE_DEBUG
425 /* find roots of the erasure location polynomial */
426 for(i=1;i<=no_eras;i++)
427 reg[i] = Index_of[lambda[i]];
428 count = 0;
429 for (i = 1; i <= NN; i++) {
430 q = 1;
431 for (j = 1; j <= no_eras; j++)
432 if (reg[j] != A0) {
433 reg[j] = modnn(reg[j] + j);
434 q ^= Alpha_to[reg[j]];
435 }
436 if (!q) {
437 /* store root and error location
438 * number indices
439 */
440 root[count] = i;
441 loc[count] = NN - i;
442 count++;
443 }
444 }
445 if (count != no_eras) {
446 printf("\n lambda(x) is WRONG\n");
447 return -1;
448 }
449 #ifndef NO_PRINT
450 printf("\n Erasure positions as determined by roots of Eras Loc Poly:\n");
451 for (i = 0; i < count; i++)
452 printf("%d ", loc[i]);
453 printf("\n");
454 #endif
455 #endif
456 }
457 for(i=0;i<NN-KK+1;i++)
458 b[i] = Index_of[lambda[i]];
459  
460 /*
461 * Begin Berlekamp-Massey algorithm to determine error+erasure
462 * locator polynomial
463 */
464 r = no_eras;
465 el = no_eras;
466 while (++r <= NN-KK) { /* r is the step number */
467 /* Compute discrepancy at the r-th step in poly-form */
468 discr_r = 0;
469 for (i = 0; i < r; i++){
470 if ((lambda[i] != 0) && (s[r - i] != A0)) {
471 discr_r ^= Alpha_to[modnn(Index_of[lambda[i]] + s[r - i])];
472 }
473 }
474 discr_r = Index_of[discr_r]; /* Index form */
475 if (discr_r == A0) {
476 /* 2 lines below: B(x) <-- x*B(x) */
477 COPYDOWN(&b[1],b,NN-KK);
478 b[0] = A0;
479 } else {
480 /* 7 lines below: T(x) <-- lambda(x) - discr_r*x*b(x) */
481 t[0] = lambda[0];
482 for (i = 0 ; i < NN-KK; i++) {
483 if(b[i] != A0)
484 t[i+1] = lambda[i+1] ^ Alpha_to[modnn(discr_r + b[i])];
485 else
486 t[i+1] = lambda[i+1];
487 }
488 if (2 * el <= r + no_eras - 1) {
489 el = r + no_eras - el;
490 /*
491 * 2 lines below: B(x) <-- inv(discr_r) *
492 * lambda(x)
493 */
494 for (i = 0; i <= NN-KK; i++)
495 b[i] = (lambda[i] == 0) ? A0 : modnn(Index_of[lambda[i]] - discr_r + NN);
496 } else {
497 /* 2 lines below: B(x) <-- x*B(x) */
498 COPYDOWN(&b[1],b,NN-KK);
499 b[0] = A0;
500 }
501 COPY(lambda,t,NN-KK+1);
502 }
503 }
504  
505 /* Convert lambda to index form and compute deg(lambda(x)) */
506 deg_lambda = 0;
507 for(i=0;i<NN-KK+1;i++){
508 lambda[i] = Index_of[lambda[i]];
509 if(lambda[i] != A0)
510 deg_lambda = i;
511 }
512 /*
513 * Find roots of the error+erasure locator polynomial. By Chien
514 * Search
515 */
516 COPY(&reg[1],&lambda[1],NN-KK);
517 count = 0; /* Number of roots of lambda(x) */
518 for (i = 1; i <= NN; i++) {
519 q = 1;
520 for (j = deg_lambda; j > 0; j--)
521 if (reg[j] != A0) {
522 reg[j] = modnn(reg[j] + j);
523 q ^= Alpha_to[reg[j]];
524 }
525 if (!q) {
526 /* store root (index-form) and error location number */
527 root[count] = i;
528 loc[count] = NN - i;
529 count++;
530 }
531 }
532  
533 #ifdef DEBUG
534 printf("\n Final error positions:\t");
535 for (i = 0; i < count; i++)
536 printf("%d ", loc[i]);
537 printf("\n");
538 #endif
539 if (deg_lambda != count) {
540 /*
541 * deg(lambda) unequal to number of roots => uncorrectable
542 * error detected
543 */
544 return -1;
545 }
546 /*
547 * Compute err+eras evaluator poly omega(x) = s(x)*lambda(x) (modulo
548 * x**(NN-KK)). in index form. Also find deg(omega).
549 */
550 deg_omega = 0;
551 for (i = 0; i < NN-KK;i++){
552 tmp = 0;
553 j = (deg_lambda < i) ? deg_lambda : i;
554 for(;j >= 0; j--){
555 if ((s[i + 1 - j] != A0) && (lambda[j] != A0))
556 tmp ^= Alpha_to[modnn(s[i + 1 - j] + lambda[j])];
557 }
558 if(tmp != 0)
559 deg_omega = i;
560 omega[i] = Index_of[tmp];
561 }
562 omega[NN-KK] = A0;
563  
564 /*
565 * Compute error values in poly-form. num1 = omega(inv(X(l))), num2 =
566 * inv(X(l))**(B0-1) and den = lambda_pr(inv(X(l))) all in poly-form
567 */
568 for (j = count-1; j >=0; j--) {
569 num1 = 0;
570 for (i = deg_omega; i >= 0; i--) {
571 if (omega[i] != A0)
572 num1 ^= Alpha_to[modnn(omega[i] + i * root[j])];
573 }
574 num2 = Alpha_to[modnn(root[j] * (B0 - 1) + NN)];
575 den = 0;
576  
577 /* lambda[i+1] for i even is the formal derivative lambda_pr of lambda[i] */
578 for (i = min(deg_lambda,NN-KK-1) & ~1; i >= 0; i -=2) {
579 if(lambda[i+1] != A0)
580 den ^= Alpha_to[modnn(lambda[i+1] + i * root[j])];
581 }
582 if (den == 0) {
583 #ifdef DEBUG
584 printf("\n ERROR: denominator = 0\n");
585 #endif
586 return -1;
587 }
588 /* Apply error to data */
589 if (num1 != 0) {
590 data[loc[j]] ^= Alpha_to[modnn(Index_of[num1] + Index_of[num2] + NN - Index_of[den])];
591 }
592 }
593 return count;
594 }
595  
596  
597 #endif /* USE_JPWL */