huffman.c (11948B)
1 /* Copyright 2013 Google Inc. All Rights Reserved. 2 3 Distributed under MIT license. 4 See file LICENSE for detail or copy at https://opensource.org/licenses/MIT 5 */ 6 7 /* Utilities for building Huffman decoding tables. */ 8 9 #include "huffman.h" 10 11 #include "../common/constants.h" 12 #include "../common/platform.h" 13 14 #if defined(__cplusplus) || defined(c_plusplus) 15 extern "C" { 16 #endif 17 18 #define BROTLI_REVERSE_BITS_MAX 8 19 20 #if defined(BROTLI_RBIT) 21 #define BROTLI_REVERSE_BITS_BASE \ 22 ((sizeof(brotli_reg_t) << 3) - BROTLI_REVERSE_BITS_MAX) 23 #else 24 #define BROTLI_REVERSE_BITS_BASE 0 25 static BROTLI_MODEL("small") 26 uint8_t kReverseBits[1 << BROTLI_REVERSE_BITS_MAX] = { 27 0x00, 0x80, 0x40, 0xC0, 0x20, 0xA0, 0x60, 0xE0, 28 0x10, 0x90, 0x50, 0xD0, 0x30, 0xB0, 0x70, 0xF0, 29 0x08, 0x88, 0x48, 0xC8, 0x28, 0xA8, 0x68, 0xE8, 30 0x18, 0x98, 0x58, 0xD8, 0x38, 0xB8, 0x78, 0xF8, 31 0x04, 0x84, 0x44, 0xC4, 0x24, 0xA4, 0x64, 0xE4, 32 0x14, 0x94, 0x54, 0xD4, 0x34, 0xB4, 0x74, 0xF4, 33 0x0C, 0x8C, 0x4C, 0xCC, 0x2C, 0xAC, 0x6C, 0xEC, 34 0x1C, 0x9C, 0x5C, 0xDC, 0x3C, 0xBC, 0x7C, 0xFC, 35 0x02, 0x82, 0x42, 0xC2, 0x22, 0xA2, 0x62, 0xE2, 36 0x12, 0x92, 0x52, 0xD2, 0x32, 0xB2, 0x72, 0xF2, 37 0x0A, 0x8A, 0x4A, 0xCA, 0x2A, 0xAA, 0x6A, 0xEA, 38 0x1A, 0x9A, 0x5A, 0xDA, 0x3A, 0xBA, 0x7A, 0xFA, 39 0x06, 0x86, 0x46, 0xC6, 0x26, 0xA6, 0x66, 0xE6, 40 0x16, 0x96, 0x56, 0xD6, 0x36, 0xB6, 0x76, 0xF6, 41 0x0E, 0x8E, 0x4E, 0xCE, 0x2E, 0xAE, 0x6E, 0xEE, 42 0x1E, 0x9E, 0x5E, 0xDE, 0x3E, 0xBE, 0x7E, 0xFE, 43 0x01, 0x81, 0x41, 0xC1, 0x21, 0xA1, 0x61, 0xE1, 44 0x11, 0x91, 0x51, 0xD1, 0x31, 0xB1, 0x71, 0xF1, 45 0x09, 0x89, 0x49, 0xC9, 0x29, 0xA9, 0x69, 0xE9, 46 0x19, 0x99, 0x59, 0xD9, 0x39, 0xB9, 0x79, 0xF9, 47 0x05, 0x85, 0x45, 0xC5, 0x25, 0xA5, 0x65, 0xE5, 48 0x15, 0x95, 0x55, 0xD5, 0x35, 0xB5, 0x75, 0xF5, 49 0x0D, 0x8D, 0x4D, 0xCD, 0x2D, 0xAD, 0x6D, 0xED, 50 0x1D, 0x9D, 0x5D, 0xDD, 0x3D, 0xBD, 0x7D, 0xFD, 51 0x03, 0x83, 0x43, 0xC3, 0x23, 0xA3, 0x63, 0xE3, 52 0x13, 0x93, 0x53, 0xD3, 0x33, 0xB3, 0x73, 0xF3, 53 0x0B, 0x8B, 0x4B, 0xCB, 0x2B, 0xAB, 0x6B, 0xEB, 54 0x1B, 0x9B, 0x5B, 0xDB, 0x3B, 0xBB, 0x7B, 0xFB, 55 0x07, 0x87, 0x47, 0xC7, 0x27, 0xA7, 0x67, 0xE7, 56 0x17, 0x97, 0x57, 0xD7, 0x37, 0xB7, 0x77, 0xF7, 57 0x0F, 0x8F, 0x4F, 0xCF, 0x2F, 0xAF, 0x6F, 0xEF, 58 0x1F, 0x9F, 0x5F, 0xDF, 0x3F, 0xBF, 0x7F, 0xFF 59 }; 60 #endif /* BROTLI_RBIT */ 61 62 #define BROTLI_REVERSE_BITS_LOWEST \ 63 ((brotli_reg_t)1 << (BROTLI_REVERSE_BITS_MAX - 1 + BROTLI_REVERSE_BITS_BASE)) 64 65 /* Returns reverse(num >> BROTLI_REVERSE_BITS_BASE, BROTLI_REVERSE_BITS_MAX), 66 where reverse(value, len) is the bit-wise reversal of the len least 67 significant bits of value. */ 68 static BROTLI_INLINE brotli_reg_t BrotliReverseBits(brotli_reg_t num) { 69 #if defined(BROTLI_RBIT) 70 return BROTLI_RBIT(num); 71 #else 72 return kReverseBits[num]; 73 #endif 74 } 75 76 /* Stores code in table[0], table[step], table[2*step], ..., table[end] */ 77 /* Assumes that end is an integer multiple of step */ 78 static BROTLI_INLINE void ReplicateValue(HuffmanCode* table, 79 int step, int end, 80 HuffmanCode code) { 81 do { 82 end -= step; 83 table[end] = code; 84 } while (end > 0); 85 } 86 87 /* Returns the table width of the next 2nd level table. |count| is the histogram 88 of bit lengths for the remaining symbols, |len| is the code length of the 89 next processed symbol. */ 90 static BROTLI_INLINE int NextTableBitSize(const uint16_t* const count, 91 int len, int root_bits) { 92 int left = 1 << (len - root_bits); 93 while (len < BROTLI_HUFFMAN_MAX_CODE_LENGTH) { 94 left -= count[len]; 95 if (left <= 0) break; 96 ++len; 97 left <<= 1; 98 } 99 return len - root_bits; 100 } 101 102 void BrotliBuildCodeLengthsHuffmanTable(HuffmanCode* table, 103 const uint8_t* const code_lengths, 104 uint16_t* count) { 105 HuffmanCode code; /* current table entry */ 106 int symbol; /* symbol index in original or sorted table */ 107 brotli_reg_t key; /* prefix code */ 108 brotli_reg_t key_step; /* prefix code addend */ 109 int step; /* step size to replicate values in current table */ 110 int table_size; /* size of current table */ 111 int sorted[BROTLI_CODE_LENGTH_CODES]; /* symbols sorted by code length */ 112 /* offsets in sorted table for each length */ 113 int offset[BROTLI_HUFFMAN_MAX_CODE_LENGTH_CODE_LENGTH + 1]; 114 int bits; 115 int bits_count; 116 BROTLI_DCHECK(BROTLI_HUFFMAN_MAX_CODE_LENGTH_CODE_LENGTH <= 117 BROTLI_REVERSE_BITS_MAX); 118 BROTLI_DCHECK(BROTLI_HUFFMAN_MAX_CODE_LENGTH_CODE_LENGTH == 5); 119 120 /* Generate offsets into sorted symbol table by code length. */ 121 symbol = -1; 122 bits = 1; 123 /* BROTLI_HUFFMAN_MAX_CODE_LENGTH_CODE_LENGTH == 5 */ 124 BROTLI_REPEAT_5({ 125 symbol += count[bits]; 126 offset[bits] = symbol; 127 bits++; 128 }); 129 /* Symbols with code length 0 are placed after all other symbols. */ 130 offset[0] = BROTLI_CODE_LENGTH_CODES - 1; 131 132 /* Sort symbols by length, by symbol order within each length. */ 133 symbol = BROTLI_CODE_LENGTH_CODES; 134 do { 135 BROTLI_REPEAT_6({ 136 symbol--; 137 sorted[offset[code_lengths[symbol]]--] = symbol; 138 }); 139 } while (symbol != 0); 140 141 table_size = 1 << BROTLI_HUFFMAN_MAX_CODE_LENGTH_CODE_LENGTH; 142 143 /* Special case: all symbols but one have 0 code length. */ 144 if (offset[0] == 0) { 145 code = ConstructHuffmanCode(0, (uint16_t)sorted[0]); 146 for (key = 0; key < (brotli_reg_t)table_size; ++key) { 147 table[key] = code; 148 } 149 return; 150 } 151 152 /* Fill in table. */ 153 key = 0; 154 key_step = BROTLI_REVERSE_BITS_LOWEST; 155 symbol = 0; 156 bits = 1; 157 step = 2; 158 do { 159 for (bits_count = count[bits]; bits_count != 0; --bits_count) { 160 code = ConstructHuffmanCode((uint8_t)bits, (uint16_t)sorted[symbol++]); 161 ReplicateValue(&table[BrotliReverseBits(key)], step, table_size, code); 162 key += key_step; 163 } 164 step <<= 1; 165 key_step >>= 1; 166 } while (++bits <= BROTLI_HUFFMAN_MAX_CODE_LENGTH_CODE_LENGTH); 167 } 168 169 uint32_t BrotliBuildHuffmanTable(HuffmanCode* root_table, 170 int root_bits, 171 const uint16_t* const symbol_lists, 172 uint16_t* count) { 173 HuffmanCode code; /* current table entry */ 174 HuffmanCode* table; /* next available space in table */ 175 int len; /* current code length */ 176 int symbol; /* symbol index in original or sorted table */ 177 brotli_reg_t key; /* prefix code */ 178 brotli_reg_t key_step; /* prefix code addend */ 179 brotli_reg_t sub_key; /* 2nd level table prefix code */ 180 brotli_reg_t sub_key_step; /* 2nd level table prefix code addend */ 181 int step; /* step size to replicate values in current table */ 182 int table_bits; /* key length of current table */ 183 int table_size; /* size of current table */ 184 int total_size; /* sum of root table size and 2nd level table sizes */ 185 int max_length = -1; 186 int bits; 187 int bits_count; 188 189 BROTLI_DCHECK(root_bits <= BROTLI_REVERSE_BITS_MAX); 190 BROTLI_DCHECK(BROTLI_HUFFMAN_MAX_CODE_LENGTH - root_bits <= 191 BROTLI_REVERSE_BITS_MAX); 192 193 while (symbol_lists[max_length] == 0xFFFF) max_length--; 194 max_length += BROTLI_HUFFMAN_MAX_CODE_LENGTH + 1; 195 196 table = root_table; 197 table_bits = root_bits; 198 table_size = 1 << table_bits; 199 total_size = table_size; 200 201 /* Fill in the root table. Reduce the table size to if possible, 202 and create the repetitions by memcpy. */ 203 if (table_bits > max_length) { 204 table_bits = max_length; 205 table_size = 1 << table_bits; 206 } 207 key = 0; 208 key_step = BROTLI_REVERSE_BITS_LOWEST; 209 bits = 1; 210 step = 2; 211 do { 212 symbol = bits - (BROTLI_HUFFMAN_MAX_CODE_LENGTH + 1); 213 for (bits_count = count[bits]; bits_count != 0; --bits_count) { 214 symbol = symbol_lists[symbol]; 215 code = ConstructHuffmanCode((uint8_t)bits, (uint16_t)symbol); 216 ReplicateValue(&table[BrotliReverseBits(key)], step, table_size, code); 217 key += key_step; 218 } 219 step <<= 1; 220 key_step >>= 1; 221 } while (++bits <= table_bits); 222 223 /* If root_bits != table_bits then replicate to fill the remaining slots. */ 224 while (total_size != table_size) { 225 memcpy(&table[table_size], &table[0], 226 (size_t)table_size * sizeof(table[0])); 227 table_size <<= 1; 228 } 229 230 /* Fill in 2nd level tables and add pointers to root table. */ 231 key_step = BROTLI_REVERSE_BITS_LOWEST >> (root_bits - 1); 232 sub_key = (BROTLI_REVERSE_BITS_LOWEST << 1); 233 sub_key_step = BROTLI_REVERSE_BITS_LOWEST; 234 for (len = root_bits + 1, step = 2; len <= max_length; ++len) { 235 symbol = len - (BROTLI_HUFFMAN_MAX_CODE_LENGTH + 1); 236 for (; count[len] != 0; --count[len]) { 237 if (sub_key == (BROTLI_REVERSE_BITS_LOWEST << 1U)) { 238 table += table_size; 239 table_bits = NextTableBitSize(count, len, root_bits); 240 table_size = 1 << table_bits; 241 total_size += table_size; 242 sub_key = BrotliReverseBits(key); 243 key += key_step; 244 root_table[sub_key] = ConstructHuffmanCode( 245 (uint8_t)(table_bits + root_bits), 246 (uint16_t)(((size_t)(table - root_table)) - sub_key)); 247 sub_key = 0; 248 } 249 symbol = symbol_lists[symbol]; 250 code = ConstructHuffmanCode((uint8_t)(len - root_bits), (uint16_t)symbol); 251 ReplicateValue( 252 &table[BrotliReverseBits(sub_key)], step, table_size, code); 253 sub_key += sub_key_step; 254 } 255 step <<= 1; 256 sub_key_step >>= 1; 257 } 258 return (uint32_t)total_size; 259 } 260 261 uint32_t BrotliBuildSimpleHuffmanTable(HuffmanCode* table, 262 int root_bits, 263 uint16_t* val, 264 uint32_t num_symbols) { 265 uint32_t table_size = 1; 266 const uint32_t goal_size = 1U << root_bits; 267 switch (num_symbols) { 268 case 0: 269 table[0] = ConstructHuffmanCode(0, val[0]); 270 break; 271 case 1: 272 if (val[1] > val[0]) { 273 table[0] = ConstructHuffmanCode(1, val[0]); 274 table[1] = ConstructHuffmanCode(1, val[1]); 275 } else { 276 table[0] = ConstructHuffmanCode(1, val[1]); 277 table[1] = ConstructHuffmanCode(1, val[0]); 278 } 279 table_size = 2; 280 break; 281 case 2: 282 table[0] = ConstructHuffmanCode(1, val[0]); 283 table[2] = ConstructHuffmanCode(1, val[0]); 284 if (val[2] > val[1]) { 285 table[1] = ConstructHuffmanCode(2, val[1]); 286 table[3] = ConstructHuffmanCode(2, val[2]); 287 } else { 288 table[1] = ConstructHuffmanCode(2, val[2]); 289 table[3] = ConstructHuffmanCode(2, val[1]); 290 } 291 table_size = 4; 292 break; 293 case 3: { 294 int i, k; 295 for (i = 0; i < 3; ++i) { 296 for (k = i + 1; k < 4; ++k) { 297 if (val[k] < val[i]) { 298 uint16_t t = val[k]; 299 val[k] = val[i]; 300 val[i] = t; 301 } 302 } 303 } 304 table[0] = ConstructHuffmanCode(2, val[0]); 305 table[2] = ConstructHuffmanCode(2, val[1]); 306 table[1] = ConstructHuffmanCode(2, val[2]); 307 table[3] = ConstructHuffmanCode(2, val[3]); 308 table_size = 4; 309 break; 310 } 311 case 4: { 312 if (val[3] < val[2]) { 313 uint16_t t = val[3]; 314 val[3] = val[2]; 315 val[2] = t; 316 } 317 table[0] = ConstructHuffmanCode(1, val[0]); 318 table[1] = ConstructHuffmanCode(2, val[1]); 319 table[2] = ConstructHuffmanCode(1, val[0]); 320 table[3] = ConstructHuffmanCode(3, val[2]); 321 table[4] = ConstructHuffmanCode(1, val[0]); 322 table[5] = ConstructHuffmanCode(2, val[1]); 323 table[6] = ConstructHuffmanCode(1, val[0]); 324 table[7] = ConstructHuffmanCode(3, val[3]); 325 table_size = 8; 326 break; 327 } 328 } 329 while (table_size != goal_size) { 330 memcpy(&table[table_size], &table[0], 331 (size_t)table_size * sizeof(table[0])); 332 table_size <<= 1; 333 } 334 return goal_size; 335 } 336 337 #if defined(__cplusplus) || defined(c_plusplus) 338 } /* extern "C" */ 339 #endif