/libpsn00b/lzp/compress.c

https://github.com/Lameguy64/PSn00bSDK · C · 489 lines · 323 code · 158 blank · 8 comment · 102 complexity · 34a0b3be5e34a58e50ae954501bc1ff5 MD5 · raw file 

    

  1. // Based on ilia muraviev's CRUSH compressor program which falls under public domain
    2. 

  2. 

  3. #include <string.h>
    4. 

  4. #if LZP_USE_MALLOC == TRUE
    5. 

  5. #include <stdlib.h>
    6. 

  6. #include <malloc.h>
    7. 

  7. #endif
    8. 

  8. 

  9. #include "lzconfig.h"
    10. 

  10. #include "bit.h"
    11. 

  11. #include "lzp.h"
    12. 

  12. 

  13. 

  14. // Internal structure for hash table allocation sizes
    15. 

  15. #if LZP_NO_COMPRESS == FALSE
    16. 

  16. 

  17. struct {
    18. 

  18. 	short WindowSize;	// Window size (17 - 23)
    19. 

  19. 	short Hash1Size;	// Hash 1 table size (10 - 21)
    20. 

  20. 	short Hash2Size;	// Hash 2 table size (12 - 24)
    21. 

  21. } lzHashParam = {
    22. 

  22. 	LZP_WINDOW_SIZE,
    23. 

  23.     LZP_HASH1_SIZE,
    24. 

  24.     LZP_HASH2_SIZE
    25. 

  25. };
    26. 

  26. 

  27. #endif
    28. 

  28. 

  29. 

  30. // Defines and macros for lz77 compression/decompression (don't touch)
    31. 

  31. #define W_BITS		lzHashParam.WindowSize
    32. 

  32. #define HASH1_BITS	lzHashParam.Hash1Size
    33. 

  33. #define HASH2_BITS	lzHashParam.Hash2Size
    34. 

  34. 

  35. #define W_SIZE		(1<<W_BITS)
    36. 

  36. #define W_MASK		(W_SIZE-1)
    37. 

  37. #define SLOT_BITS	4
    38. 

  38. #define NUM_SLOTS	(1<<SLOT_BITS)
    39. 

  39. 

  40. #define A_BITS		2 // 1 xx
    41. 

  41. #define B_BITS		2 // 01 xx
    42. 

  42. #define C_BITS		2 // 001 xx
    43. 

  43. #define D_BITS		3 // 0001 xxx
    44. 

  44. #define E_BITS		5 // 00001 xxxxx
    45. 

  45. #define F_BITS		9 // 00000 xxxxxxxxx
    46. 

  46. #define A			(1<<A_BITS)
    47. 

  47. #define B			((1<<B_BITS)+A)
    48. 

  48. #define C			((1<<C_BITS)+B)
    49. 

  49. #define D			((1<<D_BITS)+C)
    50. 

  50. #define E			((1<<E_BITS)+D)
    51. 

  51. #define F			((1<<F_BITS)+E)
    52. 

  52. #define MIN_MATCH	3
    53. 

  53. #define MAX_MATCH	((F-1)+MIN_MATCH)
    54. 

  54. 

  55. #define BUF_SIZE	(1<<26)
    56. 

  56. #define TOO_FAR		(1<<16)
    57. 

  57. 

  58. #define HASH1_LEN	MIN_MATCH
    59. 

  59. #define HASH2_LEN	(MIN_MATCH+1)
    60. 

  60. #define HASH1_SIZE	(1<<HASH1_BITS)
    61. 

  61. #define HASH2_SIZE	(1<<HASH2_BITS)
    62. 

  62. #define HASH1_MASK	(HASH1_SIZE-1)
    63. 

  63. #define HASH2_MASK	(HASH2_SIZE-1)
    64. 

  64. #define HASH1_SHIFT	((HASH1_BITS+(HASH1_LEN-1))/HASH1_LEN)
    65. 

  65. #define HASH2_SHIFT	((HASH2_BITS+(HASH2_LEN-1))/HASH2_LEN)
    66. 

  66. 

  67. 

  68. // LZ77
    69. 

  69. //
    70. 

  70. 

  71. #if LZP_NO_COMPRESS == FALSE
    72. 

  72. 

  73. int update_hash1(int h, int c) {
    74. 

  74. 

  75. 	return(((h<<HASH1_SHIFT)+c)&HASH1_MASK);
    76. 

  76. 

  77. }
    78. 

  78. 

  79. int update_hash2(int h, int c) {
    80. 

  80. 

  81. 	return(((h<<HASH2_SHIFT)+c)&HASH2_MASK);
    82. 

  82. 

  83. }
    84. 

  84. 

  85. int get_min(int a, int b) {
    86. 

  86. 

  87. 	return(a<b?a:b);
    88. 

  88. 

  89. }
    90. 

  90. 

  91. int get_max(int a, int b) {
    92. 

  92. 

  93. 	return(a>b?a:b);
    94. 

  94. 

  95. }
    96. 

  96. 

  97. int get_penalty(int a, int b) {
    98. 

  98. 

  99. 	int p=0;
    100. 

  100. 

  101. 	while(a > b) {
    102. 

  102. 		a >>= 3;
    103. 

  103. 		++p;
    104. 

  104. 	}
    105. 

  105. 

  106. 	return(p);
    107. 

  107. 

  108. }
    109. 

  109. 

  110. int lzCompress(void* outBuff, void* inBuff, int inSize, int level) {
    111. 

  111. 

  112. 	#if LZP_USE_MALLOC == FALSE
    113. 

  113. 	int head[HASH1_SIZE+HASH2_SIZE];
    114. 

  114. 	int prev[W_SIZE];
    115. 

  115. 	#else
    116. 

  116. 	int* head = malloc(4*(HASH1_SIZE+HASH2_SIZE));
    117. 

  117. 	int* prev = malloc(4*W_SIZE);
    118. 

  118. 	#endif
    119. 

  119. 

  120. 

  121. 	int max_chain[] = {4, 256, 1<<12};
    122. 

  122. 

  123. 	int i,s;
    124. 

  124. 	int h1=0;
    125. 

  125. 	int h2=0;
    126. 

  126. 	int p=0;
    127. 

  127. 

  128. 	int len;
    129. 

  129. 	int offset;
    130. 

  130. 

  131. 	int max_match;
    132. 

  132. 	int limit;
    133. 

  133. 

  134. 	int chain_len;
    135. 

  135. 	int next_p;
    136. 

  136. 	int max_lazy;
    137. 

  137. 	int log;
    138. 

  138. 

  139. 

  140. 	inPtr = (unsigned char*)inBuff;
    141. 

  141. 	outPtr = (unsigned char*)outBuff;
    142. 

  142. 	outBytes = 0;
    143. 

  143. 

  144. 

  145. 	for (i=0; i<HASH1_SIZE+HASH2_SIZE; ++i)
    146. 

  146. 		head[i] = -1;
    147. 

  147. 

  148. 	for (i=0; i<HASH1_LEN; ++i)
    149. 

  149. 		h1=update_hash1(h1, inPtr[i]);
    150. 

  150. 

  151. 	for (i=0; i<HASH2_LEN; ++i)
    152. 

  152. 		h2=update_hash2(h2, inPtr[i]);
    153. 

  153. 

  154. 	init_bits();
    155. 

  155. 

  156. 	// Put window size value so that the compressed data will be independent of the compression settings
    157. 

  157. 	put_bits(5, lzHashParam.WindowSize);
    158. 

  158. 

  159. 	while(p < inSize) {
    160. 

  160. 

  161. 		len = MIN_MATCH-1;
    162. 

  162. 		offset = W_SIZE;
    163. 

  163. 

  164. 		max_match = get_min(MAX_MATCH, inSize-p);
    165. 

  165. 		limit = get_max(p-W_SIZE, 0);
    166. 

  166. 

  167. 		if (head[h1] >= limit) {
    168. 

  168. 

  169. 			s = head[h1];
    170. 

  170. 

  171. 			if (inPtr[s] == inPtr[p]) {
    172. 

  172. 

  173. 				i = 0;
    174. 

  174. 

  175. 				while(++i < max_match) {
    176. 

  176. 					if (inPtr[s+i] != inPtr[p+i])
    177. 

  177. 						break;
    178. 

  178. 				}
    179. 

  179. 

  180. 				if (i > len) {
    181. 

  181. 					len = i;
    182. 

  182. 					offset = p-s;
    183. 

  183. 				}
    184. 

  184. 

  185. 			}
    186. 

  186. 

  187. 		}
    188. 

  188. 

  189. 		if (len < MAX_MATCH) {
    190. 

  190. 

  191. 			chain_len = max_chain[level];
    192. 

  192. 			s = head[h2+HASH1_SIZE];
    193. 

  193. 

  194. 			while((chain_len-- != 0) && (s >= limit)) {
    195. 

  195. 

  196. 				if ((inPtr[s+len] == inPtr[p+len]) && (inPtr[s] == inPtr[p])) {
    197. 

  197. 

  198. 					i = 0;
    199. 

  199. 

  200. 					while(++i < max_match) {
    201. 

  201. 						if (inPtr[s+i] != inPtr[p+i])
    202. 

  202. 							break;
    203. 

  203. 					}
    204. 

  204. 

  205. 					if (i > len+get_penalty((p-s)>>4, offset)) {
    206. 

  206. 						len = i;
    207. 

  207. 						offset = p-s;
    208. 

  208. 					}
    209. 

  209. 

  210. 					if (i == max_match)
    211. 

  211. 						break;
    212. 

  212. 

  213. 				}
    214. 

  214. 

  215. 				s=prev[s&W_MASK];
    216. 

  216. 

  217. 			}
    218. 

  218. 

  219. 		}
    220. 

  220. 

  221. 		if ((len == MIN_MATCH) && (offset > TOO_FAR))
    222. 

  222. 			len=0;
    223. 

  223. 

  224. 		if ((level >= 2) && (len >= MIN_MATCH) && (len < max_match)) {
    225. 

  225. 

  226. 			next_p = p+1;
    227. 

  227. 			max_lazy = get_min(len+4, max_match);
    228. 

  228. 

  229. 			chain_len = max_chain[level];
    230. 

  230. 			s = head[update_hash2(h2, inPtr[next_p+(HASH2_LEN-1)])+HASH1_SIZE];
    231. 

  231. 

  232. 			while((chain_len-- != 0) && (s >= limit)) {
    233. 

  233. 

  234. 				if ((inPtr[s+len] == inPtr[next_p+len]) && (inPtr[s] == inPtr[next_p])) {
    235. 

  235. 

  236. 					i = 0;
    237. 

  237. 

  238. 					while(++i < max_lazy) {
    239. 

  239. 						if (inPtr[s+i] != inPtr[next_p+i])
    240. 

  240. 							break;
    241. 

  241. 					}
    242. 

  242. 

  243. 					if (i > len+get_penalty(next_p-s, offset)) {
    244. 

  244. 						len = 0;
    245. 

  245. 						break;
    246. 

  246. 					}
    247. 

  247. 

  248. 					if (i == max_lazy)
    249. 

  249. 						break;
    250. 

  250. 

  251. 				}
    252. 

  252. 

  253. 				s = prev[s&W_MASK];
    254. 

  254. 

  255. 			}
    256. 

  256. 

  257. 		}
    258. 

  258. 

  259. 

  260. 		if (len >= MIN_MATCH) { // Match
    261. 

  261. 

  262. 			put_bits(1, 1);
    263. 

  263. 

  264. 			i = len-MIN_MATCH;
    265. 

  265. 

  266. 			if (i < A) {
    267. 

  267. 				put_bits(1, 1); // 1
    268. 

  268. 				put_bits(A_BITS, i);
    269. 

  269. 			} else if (i < B) {
    270. 

  270. 				put_bits(2, 1<<1); // 01
    271. 

  271. 				put_bits(B_BITS, i-A);
    272. 

  272. 			} else if (i < C) {
    273. 

  273. 				put_bits(3, 1<<2); // 001
    274. 

  274. 				put_bits(C_BITS, i-B);
    275. 

  275. 			} else if (i < D) {
    276. 

  276. 				put_bits(4, 1<<3); // 0001
    277. 

  277. 				put_bits(D_BITS, i-C);
    278. 

  278. 			} else if (i < E) {
    279. 

  279. 				put_bits(5, 1<<4); // 00001
    280. 

  280. 				put_bits(E_BITS, i-D);
    281. 

  281. 			} else {
    282. 

  282. 				put_bits(5, 0); // 00000
    283. 

  283. 				put_bits(F_BITS, i-E);
    284. 

  284. 			}
    285. 

  285. 

  286. 			--offset;
    287. 

  287. 			log = W_BITS-NUM_SLOTS;
    288. 

  288. 

  289. 			while(offset >= (2<<log))
    290. 

  290. 				++log;
    291. 

  291. 

  292. 			put_bits(SLOT_BITS, log-(W_BITS-NUM_SLOTS));
    293. 

  293. 

  294. 			if (log>(W_BITS-NUM_SLOTS))
    295. 

  295. 				put_bits(log, offset-(1<<log));
    296. 

  296. 			else
    297. 

  297. 				put_bits(W_BITS-(NUM_SLOTS-1), offset);
    298. 

  298. 

  299. 		} else { // Literal
    300. 

  300. 

  301. 			len = 1;
    302. 

  302. 			put_bits(9, inPtr[p]<<1); // 0 xxxxxxxx
    303. 

  303. 

  304. 		}
    305. 

  305. 

  306. 		while(len-- != 0) { // Insert new strings
    307. 

  307. 

  308. 			head[h1] = p;
    309. 

  309. 			prev[p&W_MASK] = head[h2+HASH1_SIZE];
    310. 

  310. 			head[h2+HASH1_SIZE] = p;
    311. 

  311. 

  312. 			++p;
    313. 

  313. 

  314. 			h1 = update_hash1(h1, inPtr[p+(HASH1_LEN-1)]);
    315. 

  315. 			h2 = update_hash2(h2, inPtr[p+(HASH2_LEN-1)]);
    316. 

  316. 

  317. 		}
    318. 

  318. 

  319. 	}
    320. 

  320. 

  321. 	flush_bits();
    322. 

  322. 

  323. 	#if LZP_USE_MALLOC == TRUE
    324. 

  324. 	free(head);
    325. 

  325. 	free(prev);
    326. 

  326. 	#endif
    327. 

  327. 

  328. 	return(outBytes);
    329. 

  329. 

  330. }
    331. 

  331. 

  332. void lzSetHashSizes(int window, int hash1, int hash2) {
    333. 

  333. 

  334. 	lzHashParam.WindowSize = window;
    335. 

  335. 	lzHashParam.Hash1Size = hash1;
    336. 

  336. 	lzHashParam.Hash2Size = hash2;
    337. 

  337. 

  338. }
    339. 

  339. 

  340. void lzResetHashSizes() {
    341. 

  341. 

  342. 	lzHashParam.WindowSize	= LZP_WINDOW_SIZE;
    343. 

  343. 	lzHashParam.Hash1Size	= LZP_HASH1_SIZE;
    344. 

  344. 	lzHashParam.Hash2Size	= LZP_HASH2_SIZE;
    345. 

  345. 

  346. }
    347. 

  347. 

  348. #endif // LZP_NO_COMPRESS
    349. 

  349. 

  350. int lzDecompress(void* outBuff, void* inBuff, int inSize) {
    351. 

  351. 

  352. 	int p=0;
    353. 

  353. 	int len;
    354. 

  354. 	int log;
    355. 

  355. 	int s;
    356. 

  356. 	int windowSize;
    357. 

  357. 

  358.     inPtr = (unsigned char*)inBuff;
    359. 

  359.     outPtr = (unsigned char*)outBuff;
    360. 

  360.     inBytes = 0;
    361. 

  361.     outBytes = 0;
    362. 

  362. 

  363. 	init_bits();
    364. 

  364. 

  365. 	// Get window size value
    366. 

  366. 	windowSize = get_bits(5);
    367. 

  367. 

  368. 	while(inBytes < inSize) {
    369. 

  369. 

  370. 		if (get_bits(1)) {
    371. 

  371. 

  372. 			if (get_bits(1))
    373. 

  373. 				len = get_bits(A_BITS);
    374. 

  374. 			else if (get_bits(1))
    375. 

  375. 				len = get_bits(B_BITS)+A;
    376. 

  376. 			else if (get_bits(1))
    377. 

  377. 				len = get_bits(C_BITS)+B;
    378. 

  378. 			else if (get_bits(1))
    379. 

  379. 				len = get_bits(D_BITS)+C;
    380. 

  380. 			else if (get_bits(1))
    381. 

  381. 				len = get_bits(E_BITS)+D;
    382. 

  382. 			else
    383. 

  383. 				len = get_bits(F_BITS)+E;
    384. 

  384. 

  385. 			log = get_bits(SLOT_BITS)+(windowSize-NUM_SLOTS);
    386. 

  386. 

  387. 			s =~ (log>(windowSize-NUM_SLOTS) ? get_bits(log)+(1<<log) : get_bits(windowSize-(NUM_SLOTS-1)))+p;
    388. 

  388. 

  389. 			if (s < 0)
    390. 

  390. 				return(LZP_ERR_DECOMPRESS);
    391. 

  391. 

  392. 			outPtr[p++] = outPtr[s++];
    393. 

  393. 			outPtr[p++] = outPtr[s++];
    394. 

  394. 			outPtr[p++] = outPtr[s++];
    395. 

  395. 

  396. 			while(len-- != 0)
    397. 

  397. 				outPtr[p++] = outPtr[s++];
    398. 

  398. 

  399. 		} else {
    400. 

  400. 

  401. 			outPtr[p++] = get_bits(8);
    402. 

  402. 

  403. 		}
    404. 

  404. 

  405. 	}
    406. 

  406. 

  407. 	return(p);
    408. 

  408. 

  409. }
    410. 

  410. 

  411. int lzDecompressLen(void* outBuff, int outSize, void* inBuff, int inSize) {
    412. 

  412. 

  413. 	int p=0;
    414. 

  414. 	int len;
    415. 

  415. 	int log;
    416. 

  416. 	int s;
    417. 

  417. 	int windowSize;
    418. 

  418. 

  419.     inPtr = (unsigned char*)inBuff;
    420. 

  420.     outPtr = (unsigned char*)outBuff;
    421. 

  421.     inBytes = 0;
    422. 

  422.     outBytes = 0;
    423. 

  423. 

  424. 	init_bits();
    425. 

  425. 

  426. 	// Get window size value
    427. 

  427. 	windowSize = get_bits(5);
    428. 

  428. 

  429. 	while(inBytes < inSize) {
    430. 

  430. 

  431. 		if (get_bits(1)) {
    432. 

  432. 

  433. 			if (get_bits(1))
    434. 

  434. 				len = get_bits(A_BITS);
    435. 

  435. 			else if (get_bits(1))
    436. 

  436. 				len = get_bits(B_BITS)+A;
    437. 

  437. 			else if (get_bits(1))
    438. 

  438. 				len = get_bits(C_BITS)+B;
    439. 

  439. 			else if (get_bits(1))
    440. 

  440. 				len = get_bits(D_BITS)+C;
    441. 

  441. 			else if (get_bits(1))
    442. 

  442. 				len = get_bits(E_BITS)+D;
    443. 

  443. 			else
    444. 

  444. 				len = get_bits(F_BITS)+E;
    445. 

  445. 

  446. 			log = get_bits(SLOT_BITS)+(windowSize-NUM_SLOTS);
    447. 

  447. 

  448. 			s =~ (log>(windowSize-NUM_SLOTS) ? get_bits(log)+(1<<log) : get_bits(windowSize-(NUM_SLOTS-1)))+p;
    449. 

  449. 

  450. 			if (s < 0)
    451. 

  451. 				return(LZP_ERR_DECOMPRESS);
    452. 

  452. 

  453. 			outPtr[p++] = outPtr[s++];
    454. 

  454. 			if (p >= outSize)
    455. 

  455. 				break;
    456. 

  456. 

  457. 			outPtr[p++] = outPtr[s++];
    458. 

  458. 			if (p >= outSize)
    459. 

  459. 				break;
    460. 

  460. 

  461. 			outPtr[p++] = outPtr[s++];
    462. 

  462. 			if (p >= outSize)
    463. 

  463. 				break;
    464. 

  464. 

  465. 			while(len-- != 0) {
    466. 

  466. 

  467. 				outPtr[p++] = outPtr[s++];
    468. 

  468. 				if (p >= outSize)
    469. 

  469. 					break;
    470. 

  470. 

  471. 			}
    472. 

  472. 

  473. 			if (p >= outSize)
    474. 

  474. 				break;
    475. 

  475. 

  476. 		} else {
    477. 

  477. 

  478. 			outPtr[p++] = get_bits(8);
    479. 

  479. 

  480. 		}
    481. 

  481. 

  482. 		if (p >= outSize)
    483. 

  483. 			break;
    484. 

  484. 

  485. 	}
    486. 

  486. 

  487. 	return(p);
    488. 

  488. 

  489. }
    490.