|
| 1 | +#include <stdio.h> |
| 2 | +#include <stdlib.h> |
| 3 | +#include <stdint.h> |
| 4 | +#include <string.h> |
| 5 | +#include <openssl/sha.h> |
| 6 | +#include <openssl/evp.h> |
| 7 | +#include <omp.h> |
| 8 | +#include <unistd.h> |
| 9 | +#include <time.h> |
| 10 | + |
| 11 | +#define CHUNK_SIZE 8192 |
| 12 | + |
| 13 | +typedef struct { |
| 14 | + int r, c; |
| 15 | + int weight; |
| 16 | +} CoordWeight; |
| 17 | + |
| 18 | +// Forward declaration of to_grid |
| 19 | +void to_grid(uint8_t byte, char grid[3][3]); |
| 20 | +// Precompute weighted patterns for all 256 byte values |
| 21 | +uint8_t precomputed_coords[256][8][2]; |
| 22 | + |
| 23 | +// Precompute weighted patterns |
| 24 | +void precompute_weighted_patterns() { |
| 25 | + for (int byte = 0; byte < 256; byte++) { |
| 26 | + char grid[3][3]; |
| 27 | + to_grid(byte, grid); |
| 28 | + |
| 29 | + CoordWeight coords[9]; |
| 30 | + int count = 0; |
| 31 | + |
| 32 | + int position_bias[3][3] = { |
| 33 | + {3, 2, 3}, |
| 34 | + {2, 4, 2}, |
| 35 | + {3, 2, 3} |
| 36 | + }; |
| 37 | + |
| 38 | + for (int r = 0; r < 3; r++) { |
| 39 | + for (int c = 0; c < 3; c++) { |
| 40 | + if (grid[r][c] != '\0') { |
| 41 | + coords[count].r = r; |
| 42 | + coords[count].c = c; |
| 43 | + coords[count].weight = (grid[r][c] == '1' ? 10 : 5) + position_bias[r][c]; |
| 44 | + count++; |
| 45 | + } |
| 46 | + } |
| 47 | + } |
| 48 | + |
| 49 | + // Sort by weight descending |
| 50 | + for (int i = 0; i < count - 1; i++) { |
| 51 | + for (int j = i + 1; j < count; j++) { |
| 52 | + if (coords[j].weight > coords[i].weight) { |
| 53 | + CoordWeight temp = coords[i]; |
| 54 | + coords[i] = coords[j]; |
| 55 | + coords[j] = temp; |
| 56 | + } |
| 57 | + } |
| 58 | + } |
| 59 | + |
| 60 | + for (int i = 0; i < 8; i++) { |
| 61 | + precomputed_coords[byte][i][0] = (i < count) ? coords[i].r : -1; // Prevent out of bounds |
| 62 | + precomputed_coords[byte][i][1] = (i < count) ? coords[i].c : -1; // Prevent out of bounds |
| 63 | + } |
| 64 | + } |
| 65 | +} |
| 66 | + |
| 67 | +// Converts a byte to a 3x3 grid with 1 blank (None) |
| 68 | +void to_grid(uint8_t byte, char grid[3][3]) { |
| 69 | + char bits[9] = {0}; |
| 70 | + for (int i = 0; i < 8; i++) { |
| 71 | + bits[i] = ((byte >> (7 - i)) & 1) + '0'; |
| 72 | + } |
| 73 | + bits[8] = '\0'; |
| 74 | + |
| 75 | + int k = 0; |
| 76 | + for (int r = 0; r < 3; r++) { |
| 77 | + for (int c = 0; c < 3; c++) { |
| 78 | + if (k < 8) |
| 79 | + grid[r][c] = bits[k++]; |
| 80 | + else |
| 81 | + grid[r][c] = '\0'; // blank |
| 82 | + } |
| 83 | + } |
| 84 | +} |
| 85 | + |
| 86 | +// Flattens the grid using the precomputed weighted order |
| 87 | +void flatten(char grid[3][3], char* out, uint8_t byte) { |
| 88 | + int idx = 0; |
| 89 | + for (int i = 0; i < 8; i++) { |
| 90 | + int r = precomputed_coords[byte][i][0]; |
| 91 | + int c = precomputed_coords[byte][i][1]; |
| 92 | + if (r >= 0 && r < 3 && c >= 0 && c < 3 && grid[r][c] != '\0') { |
| 93 | + out[idx++] = grid[r][c]; |
| 94 | + } |
| 95 | + } |
| 96 | + out[idx] = '\0'; |
| 97 | +} |
| 98 | + |
| 99 | +// Processes a byte with precomputed weighted read pattern |
| 100 | +void process_byte(uint8_t byte, char* out) { |
| 101 | + char grid[3][3]; |
| 102 | + to_grid(byte, grid); |
| 103 | + |
| 104 | + flatten(grid, out, byte); |
| 105 | +} |
| 106 | + |
| 107 | +void directional_hash_file(const char* filename, int bits, int chunk_size, int max_workers) { |
| 108 | + FILE* file = fopen(filename, "rb"); |
| 109 | + if (!file) { |
| 110 | + perror("Failed to open file"); |
| 111 | + return; |
| 112 | + } |
| 113 | + |
| 114 | + EVP_MD_CTX* ctx = EVP_MD_CTX_new(); |
| 115 | + const EVP_MD* md = NULL; |
| 116 | + |
| 117 | + switch (bits) { |
| 118 | + case 256: |
| 119 | + md = EVP_sha256(); break; |
| 120 | + case 512: |
| 121 | + md = EVP_sha512(); break; |
| 122 | + case 1024: |
| 123 | + case 2048: |
| 124 | + md = EVP_shake256(); break; |
| 125 | + default: |
| 126 | + fprintf(stderr, "Unsupported bit size: %d\n", bits); |
| 127 | + fclose(file); |
| 128 | + return; |
| 129 | + } |
| 130 | + |
| 131 | + EVP_DigestInit_ex(ctx, md, NULL); |
| 132 | + |
| 133 | + uint8_t buffer[chunk_size]; |
| 134 | + size_t read; |
| 135 | + char* bin = malloc(33 * chunk_size * sizeof(char)); |
| 136 | + size_t bin_index = 0; |
| 137 | + |
| 138 | + if (!bin) { |
| 139 | + perror("Failed to allocate memory for bin"); |
| 140 | + EVP_MD_CTX_free(ctx); |
| 141 | + fclose(file); |
| 142 | + return; |
| 143 | + } |
| 144 | + |
| 145 | + while ((read = fread(buffer, 1, chunk_size, file)) > 0) { |
| 146 | + // Allocate thread-local buffers outside parallel region |
| 147 | + char** processed_array = malloc(read * sizeof(char*)); |
| 148 | + |
| 149 | +#pragma omp parallel for num_threads(max_workers) |
| 150 | + for (size_t j = 0; j < read; j++) { |
| 151 | + processed_array[j] = malloc(1024); // allocate per thread |
| 152 | + processed_array[j][0] = '\0'; |
| 153 | + process_byte(buffer[j], processed_array[j]); |
| 154 | + } |
| 155 | + |
| 156 | + bin[0] = '\0'; |
| 157 | + for (size_t j = 0; j < read; j++) { |
| 158 | + // Use safer string copy method |
| 159 | + strcat(bin, processed_array[j]); |
| 160 | + free(processed_array[j]); |
| 161 | + } |
| 162 | + free(processed_array); |
| 163 | + |
| 164 | + size_t len = strlen(bin); |
| 165 | + size_t byte_len = (len + 7) / 8; |
| 166 | + uint8_t* byte_buf = malloc(byte_len); |
| 167 | + memset(byte_buf, 0, byte_len); |
| 168 | + |
| 169 | + for (size_t i = 0; i < len; i++) { |
| 170 | + if (bin[i] == '1') { |
| 171 | + byte_buf[i / 8] |= 1 << (7 - (i % 8)); |
| 172 | + } |
| 173 | + } |
| 174 | + |
| 175 | + EVP_DigestUpdate(ctx, byte_buf, byte_len); |
| 176 | + free(byte_buf); |
| 177 | + bin[0] = '\0'; |
| 178 | + } |
| 179 | + |
| 180 | + size_t hash_output_size = bits / 8; |
| 181 | + unsigned char hash[EVP_MAX_MD_SIZE * 4]; |
| 182 | + |
| 183 | + if (bits == 1024 || bits == 2048) { |
| 184 | + EVP_DigestFinalXOF(ctx, hash, hash_output_size); |
| 185 | + } else { |
| 186 | + unsigned int hash_len = 0; |
| 187 | + EVP_DigestFinal_ex(ctx, hash, &hash_len); |
| 188 | + hash_output_size = (size_t)hash_len; |
| 189 | + } |
| 190 | + |
| 191 | + for (size_t i = 0; i < hash_output_size; i++) { |
| 192 | + printf("%02x", hash[i]); |
| 193 | + } |
| 194 | + printf("\n"); |
| 195 | + |
| 196 | + free(bin); |
| 197 | + EVP_MD_CTX_free(ctx); |
| 198 | + fclose(file); |
| 199 | +} |
| 200 | + |
| 201 | +int main(int argc, char* argv[]) { |
| 202 | + if (argc < 2) { |
| 203 | + fprintf(stderr, "Usage: %s <file> [bits=256|512|1024|2048] [chunk_size=8192] [max_workers=4] [--time]\n", argv[0]); |
| 204 | + return 1; |
| 205 | + } |
| 206 | + |
| 207 | + int bits = 256; |
| 208 | + int chunk_size = 8192; |
| 209 | + int max_workers = 4; |
| 210 | + int time_flag = 0; |
| 211 | + |
| 212 | + for (int i = 1; i < argc; i++) { |
| 213 | + if (strcmp(argv[i], "--time") == 0) { |
| 214 | + time_flag = 1; |
| 215 | + } |
| 216 | + } |
| 217 | + |
| 218 | + if (argc > 2) bits = atoi(argv[2]); |
| 219 | + if (argc > 3) chunk_size = atoi(argv[3]); |
| 220 | + if (argc > 4) max_workers = atoi(argv[4]); |
| 221 | + |
| 222 | + // Precompute all weight orders |
| 223 | + precompute_weighted_patterns(); |
| 224 | + |
| 225 | + struct timespec start, end; |
| 226 | + if (time_flag) { |
| 227 | + clock_gettime(CLOCK_MONOTONIC, &start); |
| 228 | + } |
| 229 | + |
| 230 | + directional_hash_file(argv[1], bits, chunk_size, max_workers); |
| 231 | + |
| 232 | + if (time_flag) { |
| 233 | + clock_gettime(CLOCK_MONOTONIC, &end); |
| 234 | + double elapsed = (end.tv_sec - start.tv_sec) + (end.tv_nsec - start.tv_nsec) / 1e9; |
| 235 | + printf("Time elapsed: %.6f seconds\n", elapsed); |
| 236 | + } |
| 237 | + return 0; |
| 238 | +} |
0 commit comments