static void encode(VMPM *vmpm) { Arithcoder *ac; Arithmodel *am; Arithmodel *bin_am; Arithmodel **low_ams; unsigned int *symbol_to_index; int i, n, match_found; unsigned int j; //debug_message_fn("()\n"); if (!vmpm->outfile) { debug_message_fnc("outfile is NULL.\n"); return; } ac = arithcoder_arith_create(); arithcoder_encode_init(ac, vmpm->outfile); if (vmpm->nlowbits < 8) { am = arithmodel_order_zero_create(); arithmodel_encode_init(am, ac); arithmodel_order_zero_set_update_escape_freq(am, update_escape_freq); bin_am = arithmodel_order_zero_create(); arithmodel_encode_init(bin_am, ac); { Arithmodel_order_zero *am_oz = (Arithmodel_order_zero *)am; am_oz->bin_am = bin_am; am_oz->escape_encoded_with_rle = 1; } match_found = 0; for (i = vmpm->I; i >= 1; i--) { int nsymbols = 0; for (j = 0; j < vmpm->token_index[i]; j++) { Token *t = vmpm->token[i][j]; Token_value tv = t->value - 1; if (nsymbols == tv) { nsymbols++; } else { match_found++; break; } } if (match_found) { stat_message(vmpm, "Match found at Level %d\n", i); break; } } fprintf(vmpm->outfile, "%c", i); if (match_found) { for (; i >= 1; i--) { int nsymbols = 0; stat_message(vmpm, "Level %d (%d tokens, %d distinct): ", i, vmpm->token_index[i], vmpm->newtoken[i] - 1); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); /* Send the number of distinct symbols. */ arithmodel_encode_cbt(bin_am, vmpm->newtoken[i] - 1, vmpm->token_index[i], 0, 1); arithmodel_order_zero_reset(am, 0, vmpm->newtoken[i]); for (j = 0; j < vmpm->token_index[i]; j++) { Token *t = vmpm->token[i][j]; Token_value tv = t->value - 1; if (nsymbols == tv) { stat_message(vmpm, "e "); nsymbols++; } else { stat_message(vmpm, "%d ", tv); } arithmodel_encode(am, tv); } stat_message(vmpm, "\n"); stat_message(vmpm, "Level %d: %ld bytes\n", i, ftell(vmpm->outfile)); } } if ((symbol_to_index = malloc(vmpm->alphabetsize * sizeof(unsigned int))) == NULL) memory_error(NULL, MEMORY_ERROR); memset(symbol_to_index, 255, vmpm->alphabetsize * sizeof(unsigned int)); n = 0; arithmodel_order_zero_reset(am, 0, vmpm->alphabetsize - 1); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); stat_message(vmpm, "Level 0 (%d tokens): ", vmpm->token_index[0]); for (j = 0; j < vmpm->token_index[0]; j++) { if (symbol_to_index[(int)vmpm->token[0][j]] == (unsigned int)-1) { stat_message(vmpm, "e "); arithmodel_encode(am, n); symbol_to_index[(int)vmpm->token[0][j]] = n++; arithmodel_encode_bits(bin_am, (int)vmpm->token[0][j], vmpm->bits_per_symbol, 0, 1); } else { stat_message(vmpm, "%d ", symbol_to_index[(int)vmpm->token[0][j]]); arithmodel_encode(am, symbol_to_index[(int)vmpm->token[0][j]]); } } stat_message(vmpm, "\n"); free(symbol_to_index); arithmodel_encode_final(bin_am); arithmodel_encode_final(am); arithmodel_destroy(bin_am); arithmodel_destroy(am); } stat_message(vmpm, "Higher part: %ld bytes (%s)\n", ftell(vmpm->outfile), vmpm->outfilepath); if ((low_ams = calloc(1 << (8 - vmpm->nlowbits), sizeof(Arithmodel *))) == NULL) memory_error(NULL, MEMORY_ERROR); for (i = 0; i < (1 << (8 - vmpm->nlowbits)); i++) { low_ams[i] = arithmodel_order_zero_create(); arithmodel_encode_init(low_ams[i], ac); arithmodel_order_zero_reset(low_ams[i], 0, 0); for (j = 0; j < (1 << vmpm->nlowbits); j++) arithmodel_install_symbol(low_ams[i], 1); } for (i = 0; i < vmpm->buffer_low_size; i++) arithmodel_encode(low_ams[vmpm->buffer_high[i]], vmpm->buffer_low[i]); for (i = 0; i < (1 << (8 - vmpm->nlowbits)); i++) arithmodel_encode_final(low_ams[i]); for (i = 0; i < (1 << (8 - vmpm->nlowbits)); i++) arithmodel_destroy(low_ams[i]); arithcoder_encode_final(ac); arithcoder_destroy(ac); free(low_ams); }
static void encode(VMPM *vmpm) { Arithcoder *ac; Arithmodel *am; Arithmodel *bin_am; unsigned int j, nsymbols, *symbol_to_index; int i, match_found; //debug_message_fn("()\n"); if (!vmpm->outfile) { debug_message_fnc("outfile is NULL.\n"); return; } ac = arithcoder_arith_create(); arithcoder_encode_init(ac, vmpm->outfile); am = arithmodel_order_zero_create(); arithmodel_encode_init(am, ac); arithmodel_order_zero_set_update_escape_freq(am, update_escape_freq); bin_am = arithmodel_order_zero_create(); arithmodel_encode_init(bin_am, ac); match_found = 0; for (i = vmpm->I; i >= 1; i--) { nsymbols = 0; for (j = 0; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; if (nsymbols == tv) { nsymbols++; } else { match_found++; break; } } if (match_found) { stat_message(vmpm, "Match found at Level %d\n", i); break; } } fputc(i, vmpm->outfile); if (match_found) { for (; i >= 1; i--) { int jj = -1; nsymbols = 0; stat_message(vmpm, "Level %d (%d tokens, %d distinct): ", i, vmpm->token_index[i], vmpm->newtoken[i] - 1); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); /* Send the number of distinct symbols. */ arithmodel_encode_cbt(bin_am, vmpm->newtoken[i] - 1, vmpm->token_index[i], 0, 1); arithmodel_order_zero_reset(am, 0, vmpm->newtoken[i]); arithmodel_order_zero_reset(bin_am, 0, 0); /* The first token of each level must be t_0. */ if (vmpm->token[i][0]->value != 1) generic_error((char *)"Invalid token value.\n", INVALID_TOKEN_VALUE_ERROR); /* Hence, we don't need to encode it. */ stat_message(vmpm, "e "); arithmodel_install_symbol(am, 1); nsymbols++; for (j = 1; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; if (nsymbols == tv) { if (nsymbols > vmpm->newtoken[i] - 1) generic_error((char *)"Internal error\n", INTERNAL_ERROR); stat_message(vmpm, "e "); nsymbols++; arithmodel_encode(am, tv); if (nsymbols == vmpm->newtoken[i] - 1 && j < vmpm->token_index[i] - 1) { /* All escapes are emitted. */ jj = j; arithmodel_order_zero_uninstall_escape(am); debug_message_fnc("Level %d: Escape is uninstalled: %d/%d.\n", i, j, vmpm->token_index[i] - 1); } } else { stat_message(vmpm, "%d ", tv); arithmodel_encode(am, tv); } } stat_message(vmpm, "\n"); if (jj != -1) stat_message(vmpm, "Level %d: Escape is uninstalled at %d\n", i, jj); stat_message(vmpm, "Level %d: %ld bytes\n", i, ftell(vmpm->outfile)); } } if ((symbol_to_index = malloc(vmpm->alphabetsize * sizeof(unsigned int))) == NULL) memory_error(NULL, MEMORY_ERROR); memset(symbol_to_index, 255, vmpm->alphabetsize * sizeof(unsigned int)); nsymbols = 0; arithmodel_order_zero_reset(am, 0, vmpm->alphabetsize - 1); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); stat_message(vmpm, "Level 0 (%d tokens): ", vmpm->token_index[0]); for (j = 0; j < vmpm->token_index[0]; j++) { if (symbol_to_index[(int)vmpm->token[0][j]] == (unsigned int)-1) { if (nsymbols > vmpm->alphabetsize) generic_error((char *)"Internal error\n", INTERNAL_ERROR); stat_message(vmpm, "e "); arithmodel_encode(am, nsymbols); symbol_to_index[(int)vmpm->token[0][j]] = nsymbols++; arithmodel_encode_bits(bin_am, (int)vmpm->token[0][j], vmpm->bits_per_symbol, 0, 1); if (nsymbols == vmpm->alphabetsize) { arithmodel_order_zero_uninstall_escape(am); debug_message_fnc("Level 0: Escape is uninstalled: %d/%d.\n", j, vmpm->token_index[0] - 1); } } else { stat_message(vmpm, "%d ", symbol_to_index[(int)vmpm->token[0][j]]); arithmodel_encode(am, symbol_to_index[(int)vmpm->token[0][j]]); } } stat_message(vmpm, "\n"); free(symbol_to_index); arithmodel_encode_final(bin_am); arithmodel_encode_final(am); arithcoder_encode_final(ac); arithmodel_destroy(bin_am); arithmodel_destroy(am); arithcoder_destroy(ac); }
static void encode(VMPM *vmpm) { Arithcoder *ac; Arithmodel *am; Arithmodel *bin_am; unsigned int j, nsymbols, *symbol_to_index; int i, match_found; //debug_message_fn("()\n"); if (!vmpm->outfile) { debug_message_fnc("outfile is NULL.\n"); return; } ac = arithcoder_arith_create(); arithcoder_encode_init(ac, vmpm->outfile); am = arithmodel_order_zero_create(); arithmodel_encode_init(am, ac); arithmodel_order_zero_set_update_escape_freq(am, update_escape_freq); bin_am = arithmodel_order_zero_create(); arithmodel_encode_init(bin_am, ac); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); match_found = 0; for (i = vmpm->I; i >= 1; i--) { nsymbols = 0; for (j = 0; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; if (nsymbols == tv) { nsymbols++; } else { match_found++; break; } } if (match_found) { stat_message(vmpm, "Match found at Level %d\n", i); break; } } fputc(i, vmpm->outfile); if (match_found) { for (; i >= 1; i--) { stat_message(vmpm, "Level %d (%d tokens, %d distinct): ", i, vmpm->token_index[i], vmpm->newtoken[i] - 1); /* Encode escape symbols at once. */ arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); if (vmpm->token[i][0]->value != 1) generic_error((char *)"Invalid token value.\n", INVALID_TOKEN_VALUE_ERROR); /* Hence, we don't need to encode it. */ stat_message(vmpm, "e "); nsymbols = 1; #if 0 /* It's simpler to encode simultaneously. */ for (j = 1; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; if (nsymbols == tv) { stat_message(vmpm, "e "); nsymbols++; arithmodel_encode(bin_am, 1); } else { stat_message(vmpm, "%d ", tv); arithmodel_encode(bin_am, 0); arithmodel_encode_cbt(bin_am, tv, nsymbols - 1, 0, 1); } } #else /* It's better to encode separately. */ for (j = 1; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; if (nsymbols == tv) { stat_message(vmpm, "e "); nsymbols++; arithmodel_encode(bin_am, 1); } else { stat_message(vmpm, "%d ", tv); arithmodel_encode(bin_am, 0); } } /* Then encode non-escape symbols. */ arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); nsymbols = 1; for (j = 0; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; if (nsymbols == tv) { nsymbols++; } else { arithmodel_encode_cbt(bin_am, tv, nsymbols - 1, 0, 1); } } #endif stat_message(vmpm, "\n"); stat_message(vmpm, "Level %d: %ld bytes\n", i, ftell(vmpm->outfile)); } } if ((symbol_to_index = malloc(vmpm->alphabetsize * sizeof(unsigned int))) == NULL) memory_error(NULL, MEMORY_ERROR); memset(symbol_to_index, 255, vmpm->alphabetsize * sizeof(unsigned int)); nsymbols = 0; arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); arithmodel_order_zero_reset(am, 0, vmpm->alphabetsize - 1); stat_message(vmpm, "Level 0 (%d tokens): ", vmpm->token_index[0]); for (j = 0; j < vmpm->token_index[0]; j++) { if (symbol_to_index[(int)vmpm->token[0][j]] == (unsigned int)-1) { stat_message(vmpm, "e "); arithmodel_encode(am, nsymbols); symbol_to_index[(int)vmpm->token[0][j]] = nsymbols++; arithmodel_encode_bits(bin_am, (int)vmpm->token[0][j], vmpm->bits_per_symbol, 0, 1); } else { stat_message(vmpm, "%d ", symbol_to_index[(int)vmpm->token[0][j]]); arithmodel_encode(am, symbol_to_index[(int)vmpm->token[0][j]]); } } stat_message(vmpm, "\n"); free(symbol_to_index); arithmodel_encode_final(bin_am); arithmodel_encode_final(am); arithcoder_encode_final(ac); arithmodel_destroy(bin_am); arithmodel_destroy(am); arithcoder_destroy(ac); }
static void encode(VMPM *vmpm) { Arithcoder *ac; Arithmodel *char_am; Arithmodel *am; Arithmodel *bin_am; unsigned int j, nsymbols, *symbol_to_index; int i, match_found; //debug_message_fn("()\n"); if (!vmpm->outfile) { debug_message_fnc("outfile is NULL.\n"); return; } ac = arithcoder_arith_create(); arithcoder_encode_init(ac, vmpm->outfile); char_am = arithmodel_order_zero_create(); arithmodel_encode_init(char_am, ac); arithmodel_order_zero_set_update_escape_freq(char_am, update_escape_freq); am = arithmodel_order_zero_create(); arithmodel_encode_init(am, ac); bin_am = arithmodel_order_zero_create(); arithmodel_encode_init(bin_am, ac); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); match_found = 0; for (i = vmpm->I; i >= 1; i--) { nsymbols = 0; for (j = 0; j < vmpm->token_index[i]; j++) { Token *t = vmpm->token[i][j]; Token_value tv = t->value - 1; if (nsymbols == tv) { nsymbols++; } else { match_found++; break; } } if (match_found) { stat_message(vmpm, "Match found at Level %d\n", i); break; } } fputc(i, vmpm->outfile); if (match_found) { for (; i >= 1; i--) { nsymbols = 0; stat_message(vmpm, "Level %d (%d tokens, %d distinct): ", i, vmpm->token_index[i], vmpm->newtoken[i] - 1); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); /* Send the number of distinct symbols. */ arithmodel_encode_cbt(bin_am, vmpm->newtoken[i] - 1, vmpm->token_index[i], 0, 1); arithmodel_order_zero_reset(am, 0, vmpm->newtoken[i]); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); /* The first token of each level must be t_0. */ if (vmpm->token[i][0]->value != 1) generic_error((char *)"Invalid token value.\n", INVALID_TOKEN_VALUE_ERROR); /* Hence, we don't need to encode it. */ stat_message(vmpm, "e "); nsymbols = 1; arithmodel_order_zero_reset(am, 0, 0); arithmodel_install_symbol(am, 1); for (j = 1; j < vmpm->token_index[i]; j++) { Token_value tv = vmpm->token[i][j]->value - 1; Arithmodel_order_zero *bin_am_oz = (Arithmodel_order_zero *)bin_am; /* calculate and assign the escape symbol's probability */ bin_am_oz->freq[0] = vmpm->token_index[i] - (vmpm->newtoken[i] - 1) + nsymbols - (j + 1) + 1; bin_am_oz->freq[1] = vmpm->token_index[i] - (j + 1) + 1; //stat_message(vmpm, "freq[0] = %d, freq[1] = %d, nsymbols %d.\n", bin_am_oz->freq[0], bin_am_oz->freq[1], nsymbols); if (nsymbols == tv) { stat_message(vmpm, "e "); nsymbols++; if (bin_am_oz->freq[0] > 0 && bin_am_oz->freq[1] > bin_am_oz->freq[0]) arithmodel_encode(bin_am, 1); else if (bin_am_oz->freq[0] != 0) { stat_message(vmpm, "BUG(1)\n"); } arithmodel_install_symbol(am, 1); } else { stat_message(vmpm, "%d ", tv); if (bin_am_oz->freq[0] > 0 && bin_am_oz->freq[1] > bin_am_oz->freq[0]) arithmodel_encode(bin_am, 0); else if (bin_am_oz->freq[0] == 0) { stat_message(vmpm, "BUG(2)\n"); } arithmodel_encode(am, tv); } } stat_message(vmpm, "\n"); stat_message(vmpm, "Level %d: %ld bytes\n", i, ftell(vmpm->outfile)); } } if ((symbol_to_index = malloc(vmpm->alphabetsize * sizeof(unsigned int))) == NULL) memory_error(NULL, MEMORY_ERROR); memset(symbol_to_index, 255, vmpm->alphabetsize * sizeof(unsigned int)); nsymbols = 0; arithmodel_order_zero_reset(char_am, 0, vmpm->alphabetsize - 1); arithmodel_order_zero_reset(bin_am, 0, 0); arithmodel_install_symbol(bin_am, 1); arithmodel_install_symbol(bin_am, 1); stat_message(vmpm, "Level 0 (%d tokens): ", vmpm->token_index[0]); for (j = 0; j < vmpm->token_index[0]; j++) { if (symbol_to_index[(int)vmpm->token[0][j]] == (unsigned int)-1) { stat_message(vmpm, "e "); arithmodel_encode(char_am, nsymbols); symbol_to_index[(int)vmpm->token[0][j]] = nsymbols++; arithmodel_encode_bits(bin_am, (int)vmpm->token[0][j], vmpm->bits_per_symbol, 0, 1); } else { stat_message(vmpm, "%d ", symbol_to_index[(int)vmpm->token[0][j]]); arithmodel_encode(char_am, symbol_to_index[(int)vmpm->token[0][j]]); } } stat_message(vmpm, "\n"); free(symbol_to_index); arithmodel_encode_final(bin_am); arithmodel_encode_final(char_am); arithcoder_encode_final(ac); arithmodel_destroy(bin_am); arithmodel_destroy(char_am); arithcoder_destroy(ac); }