swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 1 | #include "as.h" |
| 2 | #include "../cpu.h" |
| 3 | #include "../instructions.h" |
| 4 | #include "../mnemonics.h" |
swissChili | ca0d2e2 | 2020-08-16 15:09:25 -0700 | [diff] [blame] | 5 | #include "map.h" |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 6 | |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 7 | #include <string.h> |
| 8 | #include <ctype.h> |
| 9 | #include <stdbool.h> |
| 10 | |
| 11 | enum |
| 12 | { |
| 13 | ARG_16, /* Absolute 16 bit argument */ |
| 14 | ARG_8, /* Absolute 8 bit argument */ |
| 15 | ARG_8REL, /* Relative 8 bit argument */ |
| 16 | ARG_REL, /* Relative label */ |
| 17 | ARG_ABS, /* Absolute label */ |
| 18 | ARG_IMP, /* Implied argument */ |
| 19 | }; |
| 20 | |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 21 | #define ERR "\033[31m" |
| 22 | #define RESET "\033[0m" |
| 23 | |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 24 | typedef struct |
| 25 | { |
| 26 | uint8_t opcode; |
| 27 | uint8_t arg_type; |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 28 | uint16_t line; |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 29 | union |
| 30 | { |
| 31 | char label[32]; |
| 32 | uint16_t long_arg; |
| 33 | uint8_t byte_arg; |
| 34 | int8_t rel_arg; |
| 35 | }; |
| 36 | } inst_t; |
| 37 | |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 38 | // Normal strtok() counts 2 seperators as one, ie asdf\n\njlk is 2 lines. |
| 39 | // This functions counts that as 3 lines, and will return an empty line in between. |
| 40 | char *strtok_fix(char *string, const char *token) |
| 41 | { |
| 42 | static char *pos; |
| 43 | if (string) |
| 44 | pos = string; |
| 45 | else |
| 46 | string = pos; |
| 47 | |
| 48 | if (*pos == 0) |
| 49 | return NULL; |
| 50 | |
| 51 | for (; *string; string++) |
| 52 | { |
| 53 | for (int i = 0; i < strlen(token); i++) |
| 54 | { |
| 55 | if (*string == token[i]) |
| 56 | { |
| 57 | *string = 0; |
| 58 | char *old_pos = pos; |
| 59 | pos = string + 1; |
| 60 | |
| 61 | return old_pos; |
| 62 | } |
| 63 | } |
| 64 | } |
| 65 | return pos; |
| 66 | } |
| 67 | |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 68 | void print_inst(inst_t *arg) |
| 69 | { |
| 70 | char *arg_types = |
| 71 | "16 8 8RELREL ABS IMP "; |
| 72 | |
| 73 | printf("\033[33mInst: %.4s $%x ", arg_types + arg->arg_type * 4, arg->opcode); |
| 74 | |
| 75 | switch (arg->arg_type) |
| 76 | { |
| 77 | case ARG_16: |
| 78 | printf("%x", arg->long_arg); |
| 79 | break; |
| 80 | case ARG_8: |
| 81 | printf("%x", arg->byte_arg); |
| 82 | break; |
| 83 | case ARG_8REL: |
| 84 | printf("%d", arg->rel_arg); |
| 85 | break; |
| 86 | case ARG_REL: |
| 87 | case ARG_ABS: |
| 88 | printf("%s", arg->label); |
| 89 | break; |
| 90 | } |
| 91 | |
| 92 | printf("\033[0m\n"); |
| 93 | } |
| 94 | |
| 95 | bool is_ident(char c) |
| 96 | { |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 97 | return c && (isalpha(c) || isdigit(c) |
| 98 | || c == '_' || c == '-' |
| 99 | || c == '$' || c == '.'); |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 100 | } |
| 101 | |
| 102 | uint32_t skip_ws(char **code) |
| 103 | { |
| 104 | uint32_t len = 0; |
| 105 | |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 106 | for (; **code == ' ' || **code == '\t'; (*code)++, len++) |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 107 | {} |
| 108 | |
| 109 | return len; |
| 110 | } |
| 111 | |
| 112 | uint32_t skip_to_eol(char **code) |
| 113 | { |
| 114 | uint32_t len = 0; |
| 115 | |
| 116 | for (; **code && **code != '\n'; (*code)++, len++) |
| 117 | {} |
| 118 | |
| 119 | if (**code) |
| 120 | (*code)++; |
| 121 | |
| 122 | return len; |
| 123 | } |
| 124 | |
| 125 | char *parse_label_name(char **code) |
| 126 | { |
| 127 | char *start = *code; |
| 128 | for (; is_ident(**code); (*code)++) |
| 129 | {} |
| 130 | |
| 131 | if (start == *code) |
| 132 | return false; |
| 133 | |
| 134 | **code = 0; |
| 135 | return start; |
| 136 | } |
| 137 | |
| 138 | char *parse_label(char **code) |
| 139 | { |
| 140 | char *start = *code; |
| 141 | |
| 142 | for (; is_ident(**code); (*code)++) |
| 143 | {} |
| 144 | |
| 145 | skip_ws(code); |
| 146 | |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 147 | if (*code != start && **code == ':') |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 148 | { |
| 149 | **code = 0; |
| 150 | (*code)++; |
| 151 | return start; |
| 152 | } |
| 153 | |
| 154 | *code = start; |
| 155 | |
| 156 | return NULL; |
| 157 | } |
| 158 | |
| 159 | char *parse_inst(char **code) |
| 160 | { |
| 161 | char *start = *code; |
| 162 | |
| 163 | for (; isalpha(**code); (*code)++) |
| 164 | {} |
| 165 | |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 166 | if (start == *code) |
| 167 | return NULL; |
| 168 | |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 169 | // If code is incremented when it points to \0, it will wrap to the next line |
| 170 | // returned by strtok(), which causes a bug where instructions followed immediately |
| 171 | // by a newline and no arguments causes the next instruction to parse the entire |
| 172 | // program as its argument (not good) |
| 173 | if (**code) |
| 174 | { |
| 175 | **code = 0; |
| 176 | (*code)++; |
| 177 | } |
| 178 | |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 179 | return start; |
| 180 | } |
| 181 | |
| 182 | bool is_eol(char c) |
| 183 | { |
| 184 | return c == ';' || |
| 185 | c == '\n' || |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 186 | c == '\0'; |
| 187 | } |
| 188 | |
| 189 | bool skip(char **code, const char *p) |
| 190 | { |
| 191 | for (; *p && *p == **code; p++, (*code)++) |
| 192 | {} |
| 193 | |
| 194 | if (!*p) |
| 195 | return true; |
| 196 | return false; |
| 197 | } |
| 198 | |
| 199 | bool parse_num(char **code, uint64_t *num) |
| 200 | { |
| 201 | char *start = *code; |
| 202 | int base = 10; |
| 203 | if (**code == '$') |
| 204 | { |
| 205 | base = 16; |
| 206 | (*code)++; |
| 207 | } |
| 208 | |
| 209 | skip_ws(code); |
| 210 | |
| 211 | char *endptr = *code; |
| 212 | int64_t val = strtol(*code, &endptr, base); |
| 213 | |
| 214 | if (*code == endptr) |
| 215 | { |
| 216 | *code = start; |
| 217 | return false; |
| 218 | } |
| 219 | *num = val; |
| 220 | *code = endptr; |
| 221 | return true; |
| 222 | } |
| 223 | |
| 224 | bool parse_num_max(char **code, uint64_t *num, uint64_t max) |
| 225 | { |
| 226 | uint64_t n; |
| 227 | if (parse_num(code, &n)) |
| 228 | { |
| 229 | if (n > max) |
| 230 | return false; |
| 231 | |
| 232 | *num = n; |
| 233 | return true; |
| 234 | } |
| 235 | else return false; |
| 236 | } |
| 237 | |
| 238 | bool parse_u8(char **code, uint8_t *num) |
| 239 | { |
| 240 | uint64_t n; |
| 241 | if (!parse_num_max(code, &n, 0xFF)) |
| 242 | return false; |
| 243 | |
| 244 | *num = n & 0xFF; |
| 245 | return true; |
| 246 | } |
| 247 | |
| 248 | bool parse_u16(char **code, uint16_t *num) |
| 249 | { |
| 250 | uint64_t n; |
| 251 | if (!parse_num_max(code, &n, 0xFFFF)) |
| 252 | return false; |
| 253 | |
| 254 | *num = n & 0xFFFF; |
| 255 | return true; |
| 256 | } |
| 257 | |
| 258 | bool ws_end(char **code) |
| 259 | { |
| 260 | skip_ws(code); |
| 261 | return is_eol(**code); |
| 262 | } |
| 263 | |
| 264 | bool parse_arg(char *code, int am, inst_t *inst) |
| 265 | { |
| 266 | skip_ws(&code); |
| 267 | |
| 268 | uint16_t num; |
| 269 | uint8_t num8; |
| 270 | char *lbl; |
| 271 | |
| 272 | switch (am) |
| 273 | { |
| 274 | case AM_ACC: |
| 275 | case AM_IMP: |
| 276 | printf("Trying AM_IMP on '%.8s'\n", code); |
| 277 | skip_ws(&code); |
| 278 | if (is_eol(*code)) |
| 279 | { |
| 280 | inst->arg_type = ARG_IMP; |
| 281 | return ws_end(&code); |
| 282 | } |
| 283 | break; |
| 284 | |
| 285 | case AM_IMM: |
| 286 | printf("Trying AM_IMM on '%.8s'\n", code); |
| 287 | if (!skip(&code, "#")) |
| 288 | return false; |
| 289 | skip_ws(&code); |
| 290 | case AM_ZP: |
| 291 | if (parse_u8(&code, &num8)) |
| 292 | { |
| 293 | inst->arg_type = ARG_8; |
| 294 | inst->byte_arg = num8; |
| 295 | |
| 296 | return ws_end(&code); |
| 297 | } |
| 298 | break; |
| 299 | |
| 300 | case AM_ABS: |
| 301 | if (parse_u16(&code, &num)) |
| 302 | { |
| 303 | inst->arg_type = ARG_16; |
| 304 | inst->long_arg = num; |
| 305 | return true; |
| 306 | } |
| 307 | else if ((lbl = parse_label_name(&code))) |
| 308 | { |
| 309 | inst->arg_type = ARG_ABS; |
| 310 | strncpy(inst->label, lbl, 32); |
| 311 | return true; |
| 312 | } |
| 313 | break; |
| 314 | |
| 315 | case AM_REL: |
| 316 | if (parse_u8(&code, &num8)) |
| 317 | { |
| 318 | inst->arg_type = ARG_8REL; |
| 319 | inst->rel_arg = num; |
| 320 | return ws_end(&code); |
| 321 | } |
| 322 | else if ((lbl = parse_label_name(&code))) |
| 323 | { |
| 324 | inst->arg_type = ARG_REL; |
| 325 | strncpy(inst->label, lbl, 32); |
| 326 | return ws_end(&code); |
| 327 | } |
| 328 | break; |
| 329 | |
| 330 | case AM_IND: |
| 331 | if (!skip(&code,"(")) |
| 332 | return false; |
| 333 | |
| 334 | if (!parse_u16(&code, &num)) |
| 335 | return false; |
| 336 | |
| 337 | if (!skip(&code, ")")) |
| 338 | return false; |
| 339 | |
| 340 | inst->arg_type = ARG_16; |
| 341 | inst->long_arg = num; |
| 342 | return true; |
| 343 | |
| 344 | case AM_AX: |
| 345 | case AM_ZPX: |
| 346 | case AM_AY: |
| 347 | case AM_ZPY: |
| 348 | if (am == AM_AX || am == AM_AY) |
| 349 | { |
| 350 | if (!parse_u16(&code, &num)) |
| 351 | return false; |
| 352 | inst->arg_type = ARG_16; |
| 353 | inst->long_arg = num; |
| 354 | } |
| 355 | else |
| 356 | { |
| 357 | if (!parse_u8(&code, &num8)) |
| 358 | return false; |
| 359 | inst->arg_type = ARG_8; |
| 360 | inst->byte_arg = num8; |
| 361 | } |
| 362 | if (!skip(&code, ",")) |
| 363 | return false; |
| 364 | |
| 365 | skip_ws(&code); |
| 366 | |
| 367 | if (tolower(*code) != (am == AM_AY || am == AM_ZPY ? 'y' : 'x')) |
| 368 | return false; |
| 369 | |
| 370 | return ws_end(&code); |
| 371 | |
| 372 | case AM_ZIX: |
| 373 | if (!skip(&code, "(")) |
| 374 | break; |
| 375 | skip_ws(&code); |
| 376 | if (!parse_u8(&code, &num8)) |
| 377 | break; |
| 378 | skip_ws(&code); |
| 379 | if (!skip(&code, ",")) |
| 380 | break; |
| 381 | skip_ws(&code); |
| 382 | if (tolower(*code) != 'x') |
| 383 | return false; |
| 384 | skip_ws(&code); |
| 385 | |
| 386 | if (!skip(&code, ")")) |
| 387 | break; |
| 388 | |
| 389 | inst->arg_type = ARG_8; |
| 390 | inst->byte_arg = num8; |
| 391 | return ws_end(&code); |
| 392 | |
| 393 | case AM_ZIY: |
| 394 | if (!skip(&code, "(")) |
| 395 | break; |
| 396 | skip_ws(&code); |
| 397 | if (!parse_u8(&code, &num8)) |
| 398 | break; |
| 399 | skip_ws(&code); |
| 400 | if (!skip(&code, ")")) |
| 401 | break; |
| 402 | skip_ws(&code); |
| 403 | if (!skip(&code, ",")) |
| 404 | break; |
| 405 | skip_ws(&code); |
| 406 | if (tolower(*code) != 'x') |
| 407 | break; |
| 408 | |
| 409 | inst->arg_type = ARG_8; |
| 410 | inst->byte_arg = num8; |
| 411 | return ws_end(&code); |
| 412 | } |
| 413 | return false; |
| 414 | } |
| 415 | |
| 416 | uint32_t assemble(char *code, FILE *out) |
| 417 | { |
swissChili | ca0d2e2 | 2020-08-16 15:09:25 -0700 | [diff] [blame] | 418 | uintptr_t num_insts = 0, |
| 419 | pc = 0x600; |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 420 | uint32_t line_no = 1; |
swissChili | ca0d2e2 | 2020-08-16 15:09:25 -0700 | [diff] [blame] | 421 | map_t *labels = new_map(); |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 422 | char *line, |
| 423 | *orig_line, |
| 424 | *line_start; |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 425 | |
| 426 | printf("Assembling File\n"); |
| 427 | printf("%s\n", code); |
| 428 | |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 429 | orig_line = strtok_fix(code, "\n"); |
| 430 | |
| 431 | while (orig_line) |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 432 | { |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 433 | line = strdup(orig_line); |
| 434 | line_start = line; |
| 435 | |
| 436 | if (*line == 0) |
| 437 | goto end_of_line; |
| 438 | printf("line %d: \033[36m%.12s\033[0m\n", line_no, line); |
| 439 | |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 440 | skip_ws(&line); |
| 441 | |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 442 | if (is_eol(*line)) |
| 443 | { |
| 444 | printf("skip_ws() brought us to EOL\n"); |
| 445 | goto end_of_line; |
| 446 | } |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 447 | |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 448 | char *label = parse_label(&line); |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 449 | skip_ws(&code); |
| 450 | if (is_eol(*line)) |
| 451 | goto end_of_line; |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 452 | char *mn = parse_inst(&line); |
| 453 | printf(" skipping %d ", skip_ws(&line)); |
| 454 | //printf("\033[33m%s\033[0m\n", line); |
| 455 | |
| 456 | bool no_argument = false; |
| 457 | printf("eol is %c ($%x)\n", *line, *line); |
| 458 | if (is_eol(*line)) |
| 459 | { |
| 460 | no_argument = true; |
| 461 | printf("... no argument\n"); |
| 462 | } |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 463 | int32_t mnemonic = -1; |
| 464 | |
| 465 | if (label) |
| 466 | { |
swissChili | ca0d2e2 | 2020-08-16 15:09:25 -0700 | [diff] [blame] | 467 | map_set(labels, label, (void *)pc); |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 468 | printf("Set label %s at $%lx\n", label, pc); |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 469 | } |
| 470 | |
| 471 | if (mn) |
| 472 | { |
| 473 | #define MN(a) if (!strcasecmp(mn, #a)) \ |
swissChili | ca0d2e2 | 2020-08-16 15:09:25 -0700 | [diff] [blame] | 474 | { \ |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 475 | mnemonic = a; \ |
swissChili | ca0d2e2 | 2020-08-16 15:09:25 -0700 | [diff] [blame] | 476 | } \ |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 477 | else |
| 478 | |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 479 | MNEMONICS |
| 480 | { |
| 481 | printf(ERR "Could not parse instruction on line %d\n%s\n" RESET, line_no, orig_line); |
| 482 | goto cleanup; |
| 483 | } |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 484 | #undef MN |
| 485 | |
| 486 | printf("Got instruction %s %d\n", mn, mnemonic); |
| 487 | |
| 488 | inst_t arg; |
| 489 | // printf("Parsing '%s'\n", line); |
| 490 | #define INST(_mn, am, op, len) \ |
swissChili | a4f49b5 | 2020-08-16 17:35:37 -0700 | [diff] [blame] | 491 | if ((no_argument && (_mn == AM_IMP || _mn == AM_ACC)) \ |
| 492 | || (mnemonic == _mn && parse_arg(line, am, &arg))) \ |
| 493 | { \ |
| 494 | arg.opcode = op; \ |
| 495 | pc += len; \ |
| 496 | print_inst(&arg); \ |
| 497 | } \ |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 498 | else |
| 499 | |
| 500 | INSTRUCTIONS |
| 501 | { |
| 502 | printf("\033[31mCould not be parsed: %s '%s'\033[0m\n", mn, line); |
| 503 | } |
| 504 | #undef INST |
| 505 | } |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 506 | end_of_line: |
| 507 | line_no++; |
| 508 | printf("Line is %d\n", line_no); |
| 509 | orig_line = strtok_fix(NULL, "\n"); |
| 510 | free(line_start); |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 511 | } |
| 512 | |
swissChili | 7acb4ce | 2020-08-16 20:16:10 -0700 | [diff] [blame^] | 513 | cleanup: |
swissChili | 97b5d8b | 2020-08-15 20:00:54 -0700 | [diff] [blame] | 514 | free_map(labels); |
| 515 | |
| 516 | return num_insts; |
| 517 | } |