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" |
| 5 | |
| 6 | #include <collect/map.h> |
| 7 | #include <collect/vector.h> |
| 8 | #include <string.h> |
| 9 | #include <ctype.h> |
| 10 | #include <stdbool.h> |
| 11 | |
| 12 | enum |
| 13 | { |
| 14 | ARG_16, /* Absolute 16 bit argument */ |
| 15 | ARG_8, /* Absolute 8 bit argument */ |
| 16 | ARG_8REL, /* Relative 8 bit argument */ |
| 17 | ARG_REL, /* Relative label */ |
| 18 | ARG_ABS, /* Absolute label */ |
| 19 | ARG_IMP, /* Implied argument */ |
| 20 | }; |
| 21 | |
| 22 | typedef struct |
| 23 | { |
| 24 | uint8_t opcode; |
| 25 | uint8_t arg_type; |
| 26 | union |
| 27 | { |
| 28 | char label[32]; |
| 29 | uint16_t long_arg; |
| 30 | uint8_t byte_arg; |
| 31 | int8_t rel_arg; |
| 32 | }; |
| 33 | } inst_t; |
| 34 | |
| 35 | void print_inst(inst_t *arg) |
| 36 | { |
| 37 | char *arg_types = |
| 38 | "16 8 8RELREL ABS IMP "; |
| 39 | |
| 40 | printf("\033[33mInst: %.4s $%x ", arg_types + arg->arg_type * 4, arg->opcode); |
| 41 | |
| 42 | switch (arg->arg_type) |
| 43 | { |
| 44 | case ARG_16: |
| 45 | printf("%x", arg->long_arg); |
| 46 | break; |
| 47 | case ARG_8: |
| 48 | printf("%x", arg->byte_arg); |
| 49 | break; |
| 50 | case ARG_8REL: |
| 51 | printf("%d", arg->rel_arg); |
| 52 | break; |
| 53 | case ARG_REL: |
| 54 | case ARG_ABS: |
| 55 | printf("%s", arg->label); |
| 56 | break; |
| 57 | } |
| 58 | |
| 59 | printf("\033[0m\n"); |
| 60 | } |
| 61 | |
| 62 | bool is_ident(char c) |
| 63 | { |
| 64 | return c && (isalpha(c) || isdigit(c)); |
| 65 | } |
| 66 | |
| 67 | uint32_t skip_ws(char **code) |
| 68 | { |
| 69 | uint32_t len = 0; |
| 70 | |
| 71 | for (; isspace(**code); (*code)++, len++) |
| 72 | {} |
| 73 | |
| 74 | return len; |
| 75 | } |
| 76 | |
| 77 | uint32_t skip_to_eol(char **code) |
| 78 | { |
| 79 | uint32_t len = 0; |
| 80 | |
| 81 | for (; **code && **code != '\n'; (*code)++, len++) |
| 82 | {} |
| 83 | |
| 84 | if (**code) |
| 85 | (*code)++; |
| 86 | |
| 87 | return len; |
| 88 | } |
| 89 | |
| 90 | char *parse_label_name(char **code) |
| 91 | { |
| 92 | char *start = *code; |
| 93 | for (; is_ident(**code); (*code)++) |
| 94 | {} |
| 95 | |
| 96 | if (start == *code) |
| 97 | return false; |
| 98 | |
| 99 | **code = 0; |
| 100 | return start; |
| 101 | } |
| 102 | |
| 103 | char *parse_label(char **code) |
| 104 | { |
| 105 | char *start = *code; |
| 106 | |
| 107 | for (; is_ident(**code); (*code)++) |
| 108 | {} |
| 109 | |
| 110 | skip_ws(code); |
| 111 | |
| 112 | if (**code == ':') |
| 113 | { |
| 114 | **code = 0; |
| 115 | (*code)++; |
| 116 | return start; |
| 117 | } |
| 118 | |
| 119 | *code = start; |
| 120 | |
| 121 | return NULL; |
| 122 | } |
| 123 | |
| 124 | char *parse_inst(char **code) |
| 125 | { |
| 126 | char *start = *code; |
| 127 | |
| 128 | for (; isalpha(**code); (*code)++) |
| 129 | {} |
| 130 | |
| 131 | **code = 0; |
| 132 | |
| 133 | if (start == *code) |
| 134 | return NULL; |
| 135 | |
| 136 | (*code)++; |
| 137 | return start; |
| 138 | } |
| 139 | |
| 140 | bool is_eol(char c) |
| 141 | { |
| 142 | return c == ';' || |
| 143 | c == '\n' || |
| 144 | c == '\r' || |
| 145 | c == '\0'; |
| 146 | } |
| 147 | |
| 148 | bool skip(char **code, const char *p) |
| 149 | { |
| 150 | for (; *p && *p == **code; p++, (*code)++) |
| 151 | {} |
| 152 | |
| 153 | if (!*p) |
| 154 | return true; |
| 155 | return false; |
| 156 | } |
| 157 | |
| 158 | bool parse_num(char **code, uint64_t *num) |
| 159 | { |
| 160 | char *start = *code; |
| 161 | int base = 10; |
| 162 | if (**code == '$') |
| 163 | { |
| 164 | base = 16; |
| 165 | (*code)++; |
| 166 | } |
| 167 | |
| 168 | skip_ws(code); |
| 169 | |
| 170 | char *endptr = *code; |
| 171 | int64_t val = strtol(*code, &endptr, base); |
| 172 | |
| 173 | if (*code == endptr) |
| 174 | { |
| 175 | *code = start; |
| 176 | return false; |
| 177 | } |
| 178 | *num = val; |
| 179 | *code = endptr; |
| 180 | return true; |
| 181 | } |
| 182 | |
| 183 | bool parse_num_max(char **code, uint64_t *num, uint64_t max) |
| 184 | { |
| 185 | uint64_t n; |
| 186 | if (parse_num(code, &n)) |
| 187 | { |
| 188 | if (n > max) |
| 189 | return false; |
| 190 | |
| 191 | *num = n; |
| 192 | return true; |
| 193 | } |
| 194 | else return false; |
| 195 | } |
| 196 | |
| 197 | bool parse_u8(char **code, uint8_t *num) |
| 198 | { |
| 199 | uint64_t n; |
| 200 | if (!parse_num_max(code, &n, 0xFF)) |
| 201 | return false; |
| 202 | |
| 203 | *num = n & 0xFF; |
| 204 | return true; |
| 205 | } |
| 206 | |
| 207 | bool parse_u16(char **code, uint16_t *num) |
| 208 | { |
| 209 | uint64_t n; |
| 210 | if (!parse_num_max(code, &n, 0xFFFF)) |
| 211 | return false; |
| 212 | |
| 213 | *num = n & 0xFFFF; |
| 214 | return true; |
| 215 | } |
| 216 | |
| 217 | bool ws_end(char **code) |
| 218 | { |
| 219 | skip_ws(code); |
| 220 | return is_eol(**code); |
| 221 | } |
| 222 | |
| 223 | bool parse_arg(char *code, int am, inst_t *inst) |
| 224 | { |
| 225 | skip_ws(&code); |
| 226 | |
| 227 | uint16_t num; |
| 228 | uint8_t num8; |
| 229 | char *lbl; |
| 230 | |
| 231 | switch (am) |
| 232 | { |
| 233 | case AM_ACC: |
| 234 | case AM_IMP: |
| 235 | printf("Trying AM_IMP on '%.8s'\n", code); |
| 236 | skip_ws(&code); |
| 237 | if (is_eol(*code)) |
| 238 | { |
| 239 | inst->arg_type = ARG_IMP; |
| 240 | return ws_end(&code); |
| 241 | } |
| 242 | break; |
| 243 | |
| 244 | case AM_IMM: |
| 245 | printf("Trying AM_IMM on '%.8s'\n", code); |
| 246 | if (!skip(&code, "#")) |
| 247 | return false; |
| 248 | skip_ws(&code); |
| 249 | case AM_ZP: |
| 250 | if (parse_u8(&code, &num8)) |
| 251 | { |
| 252 | inst->arg_type = ARG_8; |
| 253 | inst->byte_arg = num8; |
| 254 | |
| 255 | return ws_end(&code); |
| 256 | } |
| 257 | break; |
| 258 | |
| 259 | case AM_ABS: |
| 260 | if (parse_u16(&code, &num)) |
| 261 | { |
| 262 | inst->arg_type = ARG_16; |
| 263 | inst->long_arg = num; |
| 264 | return true; |
| 265 | } |
| 266 | else if ((lbl = parse_label_name(&code))) |
| 267 | { |
| 268 | inst->arg_type = ARG_ABS; |
| 269 | strncpy(inst->label, lbl, 32); |
| 270 | return true; |
| 271 | } |
| 272 | break; |
| 273 | |
| 274 | case AM_REL: |
| 275 | if (parse_u8(&code, &num8)) |
| 276 | { |
| 277 | inst->arg_type = ARG_8REL; |
| 278 | inst->rel_arg = num; |
| 279 | return ws_end(&code); |
| 280 | } |
| 281 | else if ((lbl = parse_label_name(&code))) |
| 282 | { |
| 283 | inst->arg_type = ARG_REL; |
| 284 | strncpy(inst->label, lbl, 32); |
| 285 | return ws_end(&code); |
| 286 | } |
| 287 | break; |
| 288 | |
| 289 | case AM_IND: |
| 290 | if (!skip(&code,"(")) |
| 291 | return false; |
| 292 | |
| 293 | if (!parse_u16(&code, &num)) |
| 294 | return false; |
| 295 | |
| 296 | if (!skip(&code, ")")) |
| 297 | return false; |
| 298 | |
| 299 | inst->arg_type = ARG_16; |
| 300 | inst->long_arg = num; |
| 301 | return true; |
| 302 | |
| 303 | case AM_AX: |
| 304 | case AM_ZPX: |
| 305 | case AM_AY: |
| 306 | case AM_ZPY: |
| 307 | if (am == AM_AX || am == AM_AY) |
| 308 | { |
| 309 | if (!parse_u16(&code, &num)) |
| 310 | return false; |
| 311 | inst->arg_type = ARG_16; |
| 312 | inst->long_arg = num; |
| 313 | } |
| 314 | else |
| 315 | { |
| 316 | if (!parse_u8(&code, &num8)) |
| 317 | return false; |
| 318 | inst->arg_type = ARG_8; |
| 319 | inst->byte_arg = num8; |
| 320 | } |
| 321 | if (!skip(&code, ",")) |
| 322 | return false; |
| 323 | |
| 324 | skip_ws(&code); |
| 325 | |
| 326 | if (tolower(*code) != (am == AM_AY || am == AM_ZPY ? 'y' : 'x')) |
| 327 | return false; |
| 328 | |
| 329 | return ws_end(&code); |
| 330 | |
| 331 | case AM_ZIX: |
| 332 | if (!skip(&code, "(")) |
| 333 | break; |
| 334 | skip_ws(&code); |
| 335 | if (!parse_u8(&code, &num8)) |
| 336 | break; |
| 337 | skip_ws(&code); |
| 338 | if (!skip(&code, ",")) |
| 339 | break; |
| 340 | skip_ws(&code); |
| 341 | if (tolower(*code) != 'x') |
| 342 | return false; |
| 343 | skip_ws(&code); |
| 344 | |
| 345 | if (!skip(&code, ")")) |
| 346 | break; |
| 347 | |
| 348 | inst->arg_type = ARG_8; |
| 349 | inst->byte_arg = num8; |
| 350 | return ws_end(&code); |
| 351 | |
| 352 | case AM_ZIY: |
| 353 | if (!skip(&code, "(")) |
| 354 | break; |
| 355 | skip_ws(&code); |
| 356 | if (!parse_u8(&code, &num8)) |
| 357 | break; |
| 358 | skip_ws(&code); |
| 359 | if (!skip(&code, ")")) |
| 360 | break; |
| 361 | skip_ws(&code); |
| 362 | if (!skip(&code, ",")) |
| 363 | break; |
| 364 | skip_ws(&code); |
| 365 | if (tolower(*code) != 'x') |
| 366 | break; |
| 367 | |
| 368 | inst->arg_type = ARG_8; |
| 369 | inst->byte_arg = num8; |
| 370 | return ws_end(&code); |
| 371 | } |
| 372 | return false; |
| 373 | } |
| 374 | |
| 375 | uint32_t assemble(char *code, FILE *out) |
| 376 | { |
| 377 | uintptr_t num_insts = 0; |
| 378 | uint32_t line_no = 1; |
| 379 | map *labels = new_map(); |
| 380 | vector *insts = new_vector(); |
| 381 | char *line; |
| 382 | |
| 383 | printf("Assembling File\n"); |
| 384 | printf("%s\n", code); |
| 385 | |
| 386 | line = strtok(code, "\r\n"); |
| 387 | |
| 388 | while (line) |
| 389 | { |
| 390 | skip_ws(&line); |
| 391 | |
| 392 | printf("\033[36m%.9s\033[0m\n", line); |
| 393 | |
| 394 | char *label = parse_label(&line), |
| 395 | *mn = parse_inst(&line); |
| 396 | int32_t mnemonic = -1; |
| 397 | |
| 398 | if (label) |
| 399 | { |
| 400 | map_set(labels, label, (void *)num_insts); |
| 401 | printf("Set label %s at %lu\n", label, num_insts); |
| 402 | } |
| 403 | |
| 404 | if (mn) |
| 405 | { |
| 406 | #define MN(a) if (!strcasecmp(mn, #a)) \ |
| 407 | mnemonic = a; \ |
| 408 | else |
| 409 | |
| 410 | MNEMONICS; |
| 411 | #undef MN |
| 412 | |
| 413 | printf("Got instruction %s %d\n", mn, mnemonic); |
| 414 | |
| 415 | inst_t arg; |
| 416 | // printf("Parsing '%s'\n", line); |
| 417 | #define INST(_mn, am, op, len) \ |
| 418 | if (mnemonic == _mn && parse_arg(line, am, &arg)) \ |
| 419 | { \ |
| 420 | arg.opcode = op; \ |
| 421 | print_inst(&arg); \ |
| 422 | } \ |
| 423 | else |
| 424 | |
| 425 | INSTRUCTIONS |
| 426 | { |
| 427 | printf("\033[31mCould not be parsed: %s '%s'\033[0m\n", mn, line); |
| 428 | } |
| 429 | #undef INST |
| 430 | } |
| 431 | |
| 432 | num_insts++; |
| 433 | line = strtok(NULL, "\r\n"); |
| 434 | } |
| 435 | |
| 436 | free_map(labels); |
| 437 | |
| 438 | return num_insts; |
| 439 | } |