wdenk | e84ec90 | 2005-05-05 00:04:14 +0000 | [diff] [blame] | 1 | /* |
| 2 | * JFFS2 -- Journalling Flash File System, Version 2. |
| 3 | * |
| 4 | * Copyright (C) 2004 Patrik Kluba, |
| 5 | * University of Szeged, Hungary |
| 6 | * |
| 7 | * For licensing information, see the file 'LICENCE' in the |
| 8 | * jffs2 directory. |
| 9 | * |
| 10 | * $Id: compr_lzari.c,v 1.3 2004/06/23 16:34:39 havasi Exp $ |
| 11 | * |
| 12 | */ |
| 13 | |
| 14 | /* |
| 15 | Lempel-Ziv-Arithmetic coding compression module for jffs2 |
| 16 | Based on the LZARI source included in LDS (lossless datacompression sources) |
| 17 | */ |
| 18 | |
| 19 | /* -*- Mode: C; indent-tabs-mode: t; c-basic-offset: 4; tab-width: 4 -*- */ |
| 20 | |
| 21 | /* |
| 22 | Original copyright follows: |
| 23 | |
| 24 | ************************************************************** |
| 25 | LZARI.C -- A Data Compression Program |
| 26 | (tab = 4 spaces) |
| 27 | ************************************************************** |
| 28 | 4/7/1989 Haruhiko Okumura |
| 29 | Use, distribute, and modify this program freely. |
| 30 | Please send me your improved versions. |
| 31 | PC-VAN SCIENCE |
| 32 | NIFTY-Serve PAF01022 |
| 33 | CompuServe 74050,1022 |
| 34 | ************************************************************** |
| 35 | |
| 36 | LZARI.C (c)1989 by Haruyasu Yoshizaki, Haruhiko Okumura, and Kenji Rikitake. |
| 37 | All rights reserved. Permission granted for non-commercial use. |
| 38 | |
| 39 | */ |
| 40 | |
| 41 | /* |
| 42 | |
| 43 | 2004-02-18 pajko <pajko(AT)halom(DOT)u-szeged(DOT)hu> |
| 44 | Removed unused variables and fixed no return value |
| 45 | |
| 46 | 2004-02-16 pajko <pajko(AT)halom(DOT)u-szeged(DOT)hu> |
| 47 | Initial release |
| 48 | |
| 49 | */ |
| 50 | |
| 51 | |
| 52 | #include <config.h> |
Jon Loeliger | 80eb47c | 2007-07-09 17:56:50 -0500 | [diff] [blame^] | 53 | #if defined(CONFIG_CMD_JFFS2) && defined(CONFIG_JFFS2_LZO_LZARI) |
wdenk | e84ec90 | 2005-05-05 00:04:14 +0000 | [diff] [blame] | 54 | |
| 55 | #include <linux/stddef.h> |
| 56 | #include <jffs2/jffs2.h> |
| 57 | |
| 58 | |
| 59 | #define N 4096 /* size of ring buffer */ |
| 60 | #define F 60 /* upper limit for match_length */ |
| 61 | #define THRESHOLD 2 /* encode string into position and length |
| 62 | if match_length is greater than this */ |
| 63 | #define NIL N /* index for root of binary search trees */ |
| 64 | |
| 65 | static unsigned char |
| 66 | text_buf[N + F - 1]; /* ring buffer of size N, |
| 67 | with extra F-1 bytes to facilitate string comparison */ |
| 68 | |
| 69 | /********** Arithmetic Compression **********/ |
| 70 | |
| 71 | /* If you are not familiar with arithmetic compression, you should read |
| 72 | I. E. Witten, R. M. Neal, and J. G. Cleary, |
| 73 | Communications of the ACM, Vol. 30, pp. 520-540 (1987), |
| 74 | from which much have been borrowed. */ |
| 75 | |
| 76 | #define M 15 |
| 77 | |
| 78 | /* Q1 (= 2 to the M) must be sufficiently large, but not so |
| 79 | large as the unsigned long 4 * Q1 * (Q1 - 1) overflows. */ |
| 80 | |
| 81 | #define Q1 (1UL << M) |
| 82 | #define Q2 (2 * Q1) |
| 83 | #define Q3 (3 * Q1) |
| 84 | #define Q4 (4 * Q1) |
| 85 | #define MAX_CUM (Q1 - 1) |
| 86 | |
| 87 | #define N_CHAR (256 - THRESHOLD + F) |
| 88 | /* character code = 0, 1, ..., N_CHAR - 1 */ |
| 89 | |
| 90 | static unsigned long char_to_sym[N_CHAR], sym_to_char[N_CHAR + 1]; |
| 91 | static unsigned long |
| 92 | sym_freq[N_CHAR + 1], /* frequency for symbols */ |
| 93 | sym_cum[N_CHAR + 1], /* cumulative freq for symbols */ |
| 94 | position_cum[N + 1]; /* cumulative freq for positions */ |
| 95 | |
| 96 | static void StartModel(void) /* Initialize model */ |
| 97 | { |
| 98 | unsigned long ch, sym, i; |
| 99 | |
| 100 | sym_cum[N_CHAR] = 0; |
| 101 | for (sym = N_CHAR; sym >= 1; sym--) { |
| 102 | ch = sym - 1; |
| 103 | char_to_sym[ch] = sym; sym_to_char[sym] = ch; |
| 104 | sym_freq[sym] = 1; |
| 105 | sym_cum[sym - 1] = sym_cum[sym] + sym_freq[sym]; |
| 106 | } |
| 107 | sym_freq[0] = 0; /* sentinel (!= sym_freq[1]) */ |
| 108 | position_cum[N] = 0; |
| 109 | for (i = N; i >= 1; i--) |
| 110 | position_cum[i - 1] = position_cum[i] + 10000 / (i + 200); |
| 111 | /* empirical distribution function (quite tentative) */ |
| 112 | /* Please devise a better mechanism! */ |
| 113 | } |
| 114 | |
| 115 | static void UpdateModel(unsigned long sym) |
| 116 | { |
| 117 | unsigned long c, ch_i, ch_sym; |
| 118 | unsigned long i; |
| 119 | if (sym_cum[0] >= MAX_CUM) { |
| 120 | c = 0; |
| 121 | for (i = N_CHAR; i > 0; i--) { |
| 122 | sym_cum[i] = c; |
| 123 | c += (sym_freq[i] = (sym_freq[i] + 1) >> 1); |
| 124 | } |
| 125 | sym_cum[0] = c; |
| 126 | } |
| 127 | for (i = sym; sym_freq[i] == sym_freq[i - 1]; i--) ; |
| 128 | if (i < sym) { |
| 129 | ch_i = sym_to_char[i]; ch_sym = sym_to_char[sym]; |
| 130 | sym_to_char[i] = ch_sym; sym_to_char[sym] = ch_i; |
| 131 | char_to_sym[ch_i] = sym; char_to_sym[ch_sym] = i; |
| 132 | } |
| 133 | sym_freq[i]++; |
| 134 | while (--i > 0) sym_cum[i]++; |
| 135 | sym_cum[0]++; |
| 136 | } |
| 137 | |
| 138 | static unsigned long BinarySearchSym(unsigned long x) |
| 139 | /* 1 if x >= sym_cum[1], |
| 140 | N_CHAR if sym_cum[N_CHAR] > x, |
| 141 | i such that sym_cum[i - 1] > x >= sym_cum[i] otherwise */ |
| 142 | { |
| 143 | unsigned long i, j, k; |
| 144 | |
| 145 | i = 1; j = N_CHAR; |
| 146 | while (i < j) { |
| 147 | k = (i + j) / 2; |
| 148 | if (sym_cum[k] > x) i = k + 1; else j = k; |
| 149 | } |
| 150 | return i; |
| 151 | } |
| 152 | |
| 153 | unsigned long BinarySearchPos(unsigned long x) |
| 154 | /* 0 if x >= position_cum[1], |
| 155 | N - 1 if position_cum[N] > x, |
| 156 | i such that position_cum[i] > x >= position_cum[i + 1] otherwise */ |
| 157 | { |
| 158 | unsigned long i, j, k; |
| 159 | |
| 160 | i = 1; j = N; |
| 161 | while (i < j) { |
| 162 | k = (i + j) / 2; |
| 163 | if (position_cum[k] > x) i = k + 1; else j = k; |
| 164 | } |
| 165 | return i - 1; |
| 166 | } |
| 167 | |
| 168 | static int Decode(unsigned char *srcbuf, unsigned char *dstbuf, unsigned long srclen, |
| 169 | unsigned long dstlen) /* Just the reverse of Encode(). */ |
| 170 | { |
| 171 | unsigned long i, r, j, k, c, range, sym; |
| 172 | unsigned char *ip, *op; |
| 173 | unsigned char *srcend = srcbuf + srclen; |
| 174 | unsigned char *dstend = dstbuf + dstlen; |
| 175 | unsigned char buffer = 0; |
| 176 | unsigned char mask = 0; |
| 177 | unsigned long low = 0; |
| 178 | unsigned long high = Q4; |
| 179 | unsigned long value = 0; |
| 180 | |
| 181 | ip = srcbuf; |
| 182 | op = dstbuf; |
| 183 | for (i = 0; i < M + 2; i++) { |
| 184 | value *= 2; |
| 185 | if ((mask >>= 1) == 0) { |
| 186 | buffer = (ip >= srcend) ? 0 : *(ip++); |
| 187 | mask = 128; |
| 188 | } |
| 189 | value += ((buffer & mask) != 0); |
| 190 | } |
| 191 | |
| 192 | StartModel(); |
| 193 | for (i = 0; i < N - F; i++) text_buf[i] = ' '; |
| 194 | r = N - F; |
| 195 | |
| 196 | while (op < dstend) { |
| 197 | range = high - low; |
| 198 | sym = BinarySearchSym((unsigned long) |
| 199 | (((value - low + 1) * sym_cum[0] - 1) / range)); |
| 200 | high = low + (range * sym_cum[sym - 1]) / sym_cum[0]; |
| 201 | low += (range * sym_cum[sym ]) / sym_cum[0]; |
| 202 | for ( ; ; ) { |
| 203 | if (low >= Q2) { |
| 204 | value -= Q2; low -= Q2; high -= Q2; |
| 205 | } else if (low >= Q1 && high <= Q3) { |
| 206 | value -= Q1; low -= Q1; high -= Q1; |
| 207 | } else if (high > Q2) break; |
| 208 | low += low; high += high; |
| 209 | value *= 2; |
| 210 | if ((mask >>= 1) == 0) { |
| 211 | buffer = (ip >= srcend) ? 0 : *(ip++); |
| 212 | mask = 128; |
| 213 | } |
| 214 | value += ((buffer & mask) != 0); |
| 215 | } |
| 216 | c = sym_to_char[sym]; |
| 217 | UpdateModel(sym); |
| 218 | if (c < 256) { |
| 219 | if (op >= dstend) return -1; |
| 220 | *(op++) = c; |
| 221 | text_buf[r++] = c; |
| 222 | r &= (N - 1); |
| 223 | } else { |
| 224 | j = c - 255 + THRESHOLD; |
| 225 | range = high - low; |
| 226 | i = BinarySearchPos((unsigned long) |
| 227 | (((value - low + 1) * position_cum[0] - 1) / range)); |
| 228 | high = low + (range * position_cum[i ]) / position_cum[0]; |
| 229 | low += (range * position_cum[i + 1]) / position_cum[0]; |
| 230 | for ( ; ; ) { |
| 231 | if (low >= Q2) { |
| 232 | value -= Q2; low -= Q2; high -= Q2; |
| 233 | } else if (low >= Q1 && high <= Q3) { |
| 234 | value -= Q1; low -= Q1; high -= Q1; |
| 235 | } else if (high > Q2) break; |
| 236 | low += low; high += high; |
| 237 | value *= 2; |
| 238 | if ((mask >>= 1) == 0) { |
| 239 | buffer = (ip >= srcend) ? 0 : *(ip++); |
| 240 | mask = 128; |
| 241 | } |
| 242 | value += ((buffer & mask) != 0); |
| 243 | } |
| 244 | i = (r - i - 1) & (N - 1); |
| 245 | for (k = 0; k < j; k++) { |
| 246 | c = text_buf[(i + k) & (N - 1)]; |
| 247 | if (op >= dstend) return -1; |
| 248 | *(op++) = c; |
| 249 | text_buf[r++] = c; |
| 250 | r &= (N - 1); |
| 251 | } |
| 252 | } |
| 253 | } |
| 254 | return 0; |
| 255 | } |
| 256 | |
| 257 | int lzari_decompress(unsigned char *data_in, unsigned char *cpage_out, |
| 258 | u32 srclen, u32 destlen) |
| 259 | { |
| 260 | return Decode(data_in, cpage_out, srclen, destlen); |
| 261 | } |
Jon Loeliger | 80eb47c | 2007-07-09 17:56:50 -0500 | [diff] [blame^] | 262 | #endif |