Mercurial > hg-old > index.cgi
annotate src/expr.c @ 40:d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
| author | lost |
|---|---|
| date | Sat, 03 Jan 2009 19:41:39 +0000 |
| parents | efa19ec69df9 |
| children | 7eafdb3a8074 |
| rev | line source |
|---|---|
|
13
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
1 /* |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
2 expr.c |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
3 Copyright © 2008 William Astle |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
4 |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
5 This file is part of LWASM. |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
6 |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
7 LWASM is free software: you can redistribute it and/or modify it under the |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
8 terms of the GNU General Public License as published by the Free Software |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
9 Foundation, either version 3 of the License, or (at your option) any later |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
10 version. |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
11 |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
12 This program is distributed in the hope that it will be useful, but WITHOUT |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
13 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
15 more details. |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
16 |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
17 You should have received a copy of the GNU General Public License along with |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
18 this program. If not, see <http://www.gnu.org/licenses/>. |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
19 */ |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
20 |
|
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
21 /* |
| 18 | 22 This file contains the actual expression evaluator |
|
13
05d4115b4860
Started work on new expression evaluator system and major code re-work for next release
lost
parents:
diff
changeset
|
23 */ |
| 14 | 24 |
| 25 #define __expr_c_seen__ | |
| 15 | 26 |
| 18 | 27 #include <ctype.h> |
| 15 | 28 #include <stdlib.h> |
| 18 | 29 #include <string.h> |
| 15 | 30 |
| 14 | 31 #include "expr.h" |
| 17 | 32 #include "util.h" |
|
40
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
33 #include "lwasm.h" |
| 39 | 34 |
| 17 | 35 lwasm_expr_stack_t *lwasm_expr_stack_create(void) |
| 36 { | |
| 37 lwasm_expr_stack_t *s; | |
| 38 | |
| 39 s = lwasm_alloc(sizeof(lwasm_expr_stack_t)); | |
| 40 s -> head = NULL; | |
| 41 s -> tail = NULL; | |
| 42 return s; | |
| 43 } | |
| 15 | 44 |
| 17 | 45 void lwasm_expr_stack_free(lwasm_expr_stack_t *s) |
| 46 { | |
| 47 while (s -> head) | |
| 48 { | |
| 49 s -> tail = s -> head; | |
| 50 s -> head = s -> head -> next; | |
| 51 lwasm_expr_term_free(s -> tail -> term); | |
| 52 lwasm_free(s -> tail); | |
| 53 } | |
| 54 lwasm_free(s); | |
| 55 } | |
| 14 | 56 |
| 17 | 57 void lwasm_expr_term_free(lwasm_expr_term_t *t) |
| 58 { | |
| 59 if (t) | |
| 60 { | |
| 61 if (t -> symbol) | |
| 62 lwasm_free(t -> symbol); | |
| 63 lwasm_free(t); | |
| 64 } | |
| 65 } | |
| 66 | |
| 67 lwasm_expr_term_t *lwasm_expr_term_create_oper(int oper) | |
| 68 { | |
| 69 lwasm_expr_term_t *t; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
70 |
| 39 | 71 debug_message(10, "Creating operator term: %d", oper); |
| 17 | 72 |
| 73 t = lwasm_alloc(sizeof(lwasm_expr_term_t)); | |
| 74 t -> term_type = LWASM_TERM_OPER; | |
| 75 t -> value = oper; | |
| 76 return t; | |
| 77 } | |
| 15 | 78 |
| 17 | 79 lwasm_expr_term_t *lwasm_expr_term_create_int(int val) |
| 14 | 80 { |
| 17 | 81 lwasm_expr_term_t *t; |
| 39 | 82 debug_message(10, "Creating integer term: %d", val); |
| 17 | 83 |
| 84 t = lwasm_alloc(sizeof(lwasm_expr_term_t)); | |
| 85 t -> term_type = LWASM_TERM_INT; | |
| 86 t -> value = val; | |
| 87 return t; | |
| 88 } | |
| 89 | |
| 90 lwasm_expr_term_t *lwasm_expr_term_create_sym(char *sym) | |
| 91 { | |
| 92 lwasm_expr_term_t *t; | |
| 14 | 93 |
| 39 | 94 debug_message(10, "Creating symbol term: %s", sym); |
| 95 | |
| 17 | 96 t = lwasm_alloc(sizeof(lwasm_expr_term_t)); |
| 97 t -> term_type = LWASM_TERM_SYM; | |
| 98 t -> symbol = lwasm_strdup(sym); | |
| 99 return t; | |
| 100 } | |
| 15 | 101 |
| 17 | 102 lwasm_expr_term_t *lwasm_expr_term_dup(lwasm_expr_term_t *t) |
| 103 { | |
| 104 switch (t -> term_type) | |
| 15 | 105 { |
| 17 | 106 case LWASM_TERM_INT: |
| 107 return lwasm_expr_term_create_int(t -> value); | |
| 108 | |
| 109 case LWASM_TERM_OPER: | |
| 110 return lwasm_expr_term_create_oper(t -> value); | |
| 111 | |
| 112 case LWASM_TERM_SYM: | |
| 113 return lwasm_expr_term_create_sym(t -> symbol); | |
| 114 | |
| 115 default: | |
| 39 | 116 debug_message(0, "lwasm_expr_term_dup(): invalid term type %d", t -> term_type); |
| 17 | 117 exit(1); |
| 118 } | |
| 119 // can't get here | |
| 120 } | |
| 121 | |
| 122 void lwasm_expr_stack_push(lwasm_expr_stack_t *s, lwasm_expr_term_t *t) | |
| 123 { | |
| 124 lwasm_expr_stack_node_t *n; | |
| 125 | |
| 126 if (!s) | |
| 127 { | |
| 39 | 128 debug_message(0, "lwasm_expr_stack_push(): invalid stack pointer"); |
| 17 | 129 exit(1); |
| 15 | 130 } |
| 131 | |
| 17 | 132 n = lwasm_alloc(sizeof(lwasm_expr_stack_node_t)); |
| 133 n -> next = NULL; | |
| 134 n -> prev = s -> tail; | |
| 135 n -> term = lwasm_expr_term_dup(t); | |
| 136 | |
| 137 if (s -> head) | |
| 138 { | |
| 139 s -> tail -> next = n; | |
| 140 s -> tail = n; | |
| 141 } | |
| 142 else | |
| 15 | 143 { |
| 17 | 144 s -> head = n; |
| 145 s -> tail = n; | |
| 146 } | |
| 147 } | |
| 148 | |
| 149 lwasm_expr_term_t *lwasm_expr_stack_pop(lwasm_expr_stack_t *s) | |
| 150 { | |
| 151 lwasm_expr_term_t *t; | |
| 152 lwasm_expr_stack_node_t *n; | |
| 153 | |
| 154 if (!(s -> tail)) | |
| 155 return NULL; | |
| 156 | |
| 157 n = s -> tail; | |
| 158 s -> tail = n -> prev; | |
| 159 if (!(n -> prev)) | |
| 160 { | |
| 161 s -> head = NULL; | |
| 15 | 162 } |
| 14 | 163 |
| 17 | 164 t = n -> term; |
| 165 n -> term = NULL; | |
| 166 | |
| 167 lwasm_free(n); | |
| 168 | |
| 169 return t; | |
| 14 | 170 } |
| 18 | 171 |
| 172 // the following two functions are co-routines which actually parse | |
| 173 // an infix expression onto the expression stack, each returns -1 | |
| 174 // if an error is encountered | |
| 175 | |
| 176 /* | |
| 177 parse a term and push it onto the stack | |
| 178 | |
| 179 this function handles unary prefix operators (-, +, .not., .com.) | |
| 180 as well as () | |
| 181 */ | |
| 182 int lwasm_expr_parse_term(lwasm_expr_stack_t *s, const char **p) | |
| 183 { | |
| 184 lwasm_expr_term_t *t; | |
| 39 | 185 debug_message(2, "Expression string %s", *p); |
| 186 | |
| 18 | 187 eval_next: |
| 188 if (**p == '(') | |
| 189 { | |
| 39 | 190 debug_message(3, "Starting paren"); |
| 18 | 191 (*p)++; |
| 192 lwasm_expr_parse_expr(s, p, 0); | |
| 193 if (**p != ')') | |
| 194 return -1; | |
| 195 (*p)++; | |
| 196 return 0; | |
| 197 } | |
| 198 | |
| 199 if (**p == '+') | |
| 200 { | |
| 39 | 201 debug_message(3, "Unary +"); |
| 18 | 202 (*p)++; |
| 203 goto eval_next; | |
| 204 } | |
| 205 | |
| 206 if (**p == '-') | |
| 207 { | |
| 208 // parse expression following "-" | |
| 209 (*p)++; | |
| 210 if (lwasm_expr_parse_expr(s, p, 200) < 0) | |
| 211 return -1; | |
| 212 t = lwasm_expr_term_create_oper(LWASM_OPER_NEG); | |
| 213 lwasm_expr_stack_push(s, t); | |
| 214 lwasm_expr_term_free(t); | |
| 215 return 0; | |
| 216 } | |
| 217 | |
| 23 | 218 if (**p == '^') |
| 219 { | |
| 220 // parse expression following "^" | |
| 221 (*p)++; | |
| 222 if (lwasm_expr_parse_expr(s, p, 200) < 0) | |
| 223 return -1; | |
| 224 t = lwasm_expr_term_create_oper(LWASM_OPER_COM); | |
| 225 lwasm_expr_stack_push(s, t); | |
| 226 lwasm_expr_term_free(t); | |
| 227 return 0; | |
| 228 } | |
| 229 | |
| 18 | 230 /* |
| 231 we have an actual term here so evaluate it | |
| 232 | |
| 233 it could be one of the following: | |
| 234 | |
| 235 1. a decimal constant | |
| 236 2. a hexadecimal constant | |
| 237 3. an octal constant | |
| 238 4. a binary constant | |
| 239 5. a symbol reference | |
| 240 6. the "current" instruction address (*) | |
| 241 7. the "current" data address (.) | |
| 242 8. a "back reference" (<) | |
| 243 9. a "forward reference" (>) | |
| 244 | |
| 245 items 6 through 9 are stored as symbol references | |
| 246 | |
| 247 (a . followed by a . or a alpha char or number is a symbol) | |
| 248 */ | |
| 249 if (**p == '*' | |
| 250 || ( | |
| 251 **p == '.' | |
| 252 && (*p)[1] != '.' | |
| 253 && !((*p)[1] >= 'A' && (*p)[1] <= 'Z') | |
| 254 && !((*p)[1] >= 'a' && (*p)[1] <= 'z') | |
| 255 && !((*p)[1] >= '0' && (*p)[1] <= '9') | |
| 256 ) | |
| 257 || **p == '<' | |
| 258 || **p == '>') | |
| 259 { | |
| 260 char tstr[2]; | |
| 261 tstr[0] = **p; | |
| 262 tstr[1] = '\0'; | |
| 263 t = lwasm_expr_term_create_sym(tstr); | |
| 264 lwasm_expr_stack_push(s, t); | |
| 265 lwasm_expr_term_free(t); | |
| 266 (*p)++; | |
| 267 return 0; | |
| 268 } | |
| 269 | |
| 270 /* | |
| 271 - a symbol will be a string of characters introduced by a letter, ".", | |
| 272 "_" but NOT a number | |
| 273 - a decimal constant will consist of only digits, optionally prefixed | |
| 274 with "&" | |
| 275 - a binary constant will consist of only 0s and 1s either prefixed with % | |
| 276 or suffixed with "B" | |
| 277 - a hex constant will consist of 0-9A-F either prefixed with $ or | |
| 278 suffixed with "H"; a hex number starting with A-F must be prefixed | |
| 279 with $ or start with 0 and end with H | |
| 280 - an octal constant will consist of 0-7 either prefixed with @ or | |
| 281 suffixed with "O" or "Q" | |
| 282 - an ascii constant will be a single character prefixed with a ' | |
| 283 - a double ascii constant will be two characters prefixed with a " | |
| 284 | |
| 285 */ | |
| 286 if (**p == '"') | |
| 287 { | |
| 288 // double ascii constant | |
| 289 int val; | |
| 290 (*p)++; | |
| 291 if (!**p) | |
| 292 return -1; | |
| 293 if (!*((*p)+1)) | |
| 294 return -1; | |
| 295 val = **p << 8 | *((*p) + 1); | |
| 296 (*p) += 2; | |
| 297 t = lwasm_expr_term_create_int(val); | |
| 298 lwasm_expr_stack_push(s, t); | |
| 299 lwasm_expr_term_free(t); | |
| 300 return 0; | |
| 301 } | |
| 302 else if (**p == '\'') | |
| 303 { | |
| 304 // single ascii constant | |
| 305 int val; | |
| 306 (*p)++; | |
| 307 if (!**p) | |
| 308 return -1; | |
| 309 val = **p; | |
| 310 (*p)++; | |
| 311 t = lwasm_expr_term_create_int(val); | |
| 312 lwasm_expr_stack_push(s, t); | |
| 313 lwasm_expr_term_free(t); | |
| 314 } | |
| 315 else if (**p == '&') | |
| 316 { | |
| 317 // decimal constant | |
| 318 int val = 0; | |
| 319 | |
| 320 (*p)++; | |
| 321 while (strchr("0123456789", **p)) | |
| 322 { | |
| 323 val = val * 10 + (**p - '0'); | |
| 324 (*p)++; | |
| 325 } | |
| 326 t = lwasm_expr_term_create_int(val); | |
| 327 lwasm_expr_stack_push(s, t); | |
| 328 lwasm_expr_term_free(t); | |
| 329 return 0; | |
| 330 } | |
| 331 else if (**p == '%') | |
| 332 { | |
| 333 // binary constant | |
| 334 int val = 0; | |
| 335 | |
| 336 (*p)++; | |
| 337 while (**p == '0' || **p == '1') | |
| 338 { | |
| 339 val = val * 2 + (**p - '0'); | |
| 340 (*p)++; | |
| 341 } | |
| 342 t = lwasm_expr_term_create_int(val); | |
| 343 lwasm_expr_stack_push(s, t); | |
| 344 lwasm_expr_term_free(t); | |
| 345 return 0; | |
| 346 } | |
| 347 else if (**p == '$') | |
| 348 { | |
| 349 // hexadecimal constant | |
| 350 int val = 0, val2; | |
| 351 | |
| 352 (*p)++; | |
| 353 while (strchr("0123456789ABCDEFabcdef", **p)) | |
| 354 { | |
| 355 val2 = toupper(**p) - '0'; | |
| 356 if (val2 > 9) | |
| 357 val2 -= 7; | |
| 358 val = val * 16 + val2; | |
| 359 (*p)++; | |
| 360 } | |
| 361 t = lwasm_expr_term_create_int(val); | |
| 362 lwasm_expr_stack_push(s, t); | |
| 363 lwasm_expr_term_free(t); | |
| 364 return 0; | |
| 365 } | |
|
40
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
366 // an @ followed by a digit is an octal number |
|
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
367 // but if it's followed by anything else, it is a symbol |
|
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
368 else if (**p == '@' && isdigit(*(*p + 1))) |
| 18 | 369 { |
| 370 // octal constant | |
| 371 int val = 0; | |
| 372 | |
| 373 (*p)++; | |
| 374 while (strchr("01234567", **p)) | |
| 375 { | |
| 376 val = val * 8 + (**p - '0'); | |
| 377 (*p)++; | |
| 378 } | |
| 379 t = lwasm_expr_term_create_int(val); | |
| 380 lwasm_expr_stack_push(s, t); | |
| 381 lwasm_expr_term_free(t); | |
| 382 return 0; | |
| 383 } | |
| 384 | |
| 385 // symbol or bare decimal or suffix identified constant here | |
| 386 // all numbers will start with a digit at this point | |
| 387 if (**p < '0' || **p > '9') | |
| 388 { | |
| 389 int l = 0; | |
| 390 char *sb; | |
| 391 | |
| 392 // evaluate a symbol here | |
| 393 static const char *symchars = "_.$@\\abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"; | |
| 394 while (strchr(symchars, (*p)[l])) | |
| 395 l++; | |
| 396 | |
| 397 if (l == 0) | |
| 398 return -1; | |
| 399 | |
| 400 sb = lwasm_alloc(l + 1); | |
| 401 sb[l] = '\0'; | |
| 402 memcpy(sb, *p, l); | |
| 403 t = lwasm_expr_term_create_sym(sb); | |
| 404 lwasm_expr_stack_push(s, t); | |
| 405 lwasm_expr_term_free(t); | |
| 406 lwasm_free(sb); | |
| 407 (*p) += l; | |
| 408 return 0; | |
| 409 } | |
| 410 | |
| 411 if (!**p) | |
| 412 return -1; | |
| 413 | |
| 414 // evaluate a suffix based constant | |
| 415 { | |
| 416 int decval = 0, binval = 0, hexval = 0, octval = 0; | |
| 417 int valtype = 15; // 1 = bin, 2 = oct, 4 = dec, 8 = hex | |
| 418 int bindone = 0; | |
| 419 int val; | |
| 420 int dval; | |
| 421 | |
| 422 while (1) | |
| 423 { | |
| 424 if (!**p || !strchr("0123456789ABCDEFabcdefqhoQHO", **p)) | |
| 425 { | |
| 426 // we can legally have bin or decimal here | |
| 427 if (bindone) | |
| 428 { | |
| 429 // we just finished a binary value | |
| 430 val = binval; | |
| 431 break; | |
| 432 } | |
| 433 else if (valtype & 4) | |
| 434 { | |
| 435 // otherwise we must be decimal (if we're still allowed one) | |
| 436 val = decval; | |
| 39 | 437 debug_message(3, "End of decimal value"); |
| 18 | 438 break; |
| 439 } | |
| 440 else | |
| 441 { | |
| 442 // bad value | |
| 443 return -1; | |
| 444 } | |
| 445 } | |
| 446 | |
| 447 dval = toupper(**p); | |
| 448 (*p)++; | |
| 449 | |
| 450 if (bindone) | |
| 451 { | |
| 452 // any characters past "B" means it is not binary | |
| 453 bindone = 0; | |
| 454 valtype &= 14; | |
| 455 } | |
| 456 | |
| 457 switch (dval) | |
| 458 { | |
| 459 case 'Q': | |
| 460 case 'O': | |
| 461 if (valtype & 2) | |
| 462 { | |
| 463 val = octval; | |
| 464 valtype = -1; | |
| 465 break; | |
| 466 } | |
| 467 else | |
| 468 { | |
| 469 // not a valid octal value | |
| 470 return -1; | |
| 471 } | |
| 472 /* can't get here */ | |
| 473 | |
| 474 case 'H': | |
| 475 if (valtype & 8) | |
| 476 { | |
| 477 val = hexval; | |
| 478 valtype = -1; | |
| 479 break; | |
| 480 } | |
| 481 else | |
| 482 { | |
| 483 // not a valid hex number | |
| 484 return -1; | |
| 485 } | |
| 486 /* can't get here */ | |
| 487 | |
| 488 case 'B': | |
| 489 // this is a bit of a sticky one since B is a legit hex | |
| 490 // digit so this may or may not be the end of the number | |
| 491 // so we fall through to the digit case | |
| 492 | |
| 493 if (valtype & 1) | |
| 494 { | |
| 495 // could still be binary | |
| 496 bindone = 1; | |
| 497 valtype = 9; // hex and binary | |
| 498 } | |
| 499 /* fall through intentional */ | |
| 500 | |
| 501 default: | |
| 502 // digit | |
| 503 dval -= '0'; | |
| 504 if (dval > 9) | |
| 505 dval -= 7; | |
| 39 | 506 debug_message(3, "Got digit: %d", dval); |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
507 // if (dval > 1) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
508 // valtype &= 14; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
509 // if (dval > 7) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
510 // valtype &= 12; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
511 // if (dval > 9) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
512 // valtype &= 8; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
513 |
| 18 | 514 if (valtype & 8) |
| 515 { | |
| 516 hexval = hexval * 16 + dval; | |
| 517 } | |
| 518 if (valtype & 4) | |
| 519 { | |
| 520 if (dval > 9) | |
| 521 valtype &= 11; | |
| 522 else | |
| 523 decval = decval * 10 + dval; | |
| 524 } | |
| 525 if (valtype & 2) | |
| 526 { | |
| 527 if (dval > 7) | |
| 528 valtype &= 13; | |
| 529 else | |
| 530 octval = octval * 8 + dval; | |
| 531 } | |
| 532 if (valtype & 1) | |
| 533 { | |
| 534 if (dval > 1) | |
| 535 valtype &= 14; | |
| 536 else | |
| 537 binval = binval * 2 + dval; | |
| 538 } | |
| 539 } | |
| 540 // break out if we have a return value | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
541 if (valtype == -1) |
| 18 | 542 break; |
| 543 // return if no more valid possibilities! | |
| 544 if (valtype == 0) | |
| 545 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
546 val = decval; // in case we fall through |
| 18 | 547 } |
| 548 | |
| 549 // we get here when we have a value to return | |
| 550 t = lwasm_expr_term_create_int(val); | |
| 551 lwasm_expr_stack_push(s, t); | |
| 552 lwasm_expr_term_free(t); | |
| 553 return 0; | |
| 554 } | |
| 555 /* can't get here */ | |
| 556 } | |
| 557 | |
| 558 // parse an expression and push the result onto the stack | |
| 559 // if an operator of lower precedence than the value of "prec" is found, | |
| 560 int lwasm_expr_parse_expr(lwasm_expr_stack_t *s, const char **p, int prec) | |
| 561 { | |
| 562 static const struct operinfo | |
| 563 { | |
| 564 int opernum; | |
| 565 char *operstr; | |
| 566 int operprec; | |
| 567 } operators[] = | |
| 568 { | |
| 569 { LWASM_OPER_PLUS, "+", 100 }, | |
| 570 { LWASM_OPER_MINUS, "-", 100 }, | |
| 571 { LWASM_OPER_TIMES, "*", 150 }, | |
| 572 { LWASM_OPER_DIVIDE, "/", 150 }, | |
| 573 { LWASM_OPER_MOD, "%", 150 }, | |
| 574 { LWASM_OPER_INTDIV, "\\", 150 }, | |
| 575 | |
| 576 { LWASM_OPER_NONE, "", 0 } | |
| 577 }; | |
| 578 int opern, i; | |
| 579 lwasm_expr_term_t *operterm; | |
| 580 | |
| 581 // return if we are at the end of the expression or a subexpression | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
582 if (!**p || isspace(**p) || **p == ')' || **p == ',') |
| 18 | 583 return 0; |
| 584 | |
| 585 if (lwasm_expr_parse_term(s, p) < 0) | |
| 586 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
587 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
588 eval_next: |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
589 if (!**p || isspace(**p) || **p == ')' || **p == ',') |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
590 return 0; |
| 18 | 591 |
| 592 // expecting an operator here | |
| 593 for (opern = 0; operators[opern].opernum != LWASM_OPER_NONE; opern++) | |
| 594 { | |
| 595 for (i = 0; (*p)[i] && operators[opern].operstr[i] && (*p[i] == operators[opern].operstr[i]); i++) | |
| 596 /* do nothing */ ; | |
| 597 if (operators[opern].operstr[i] == '\0') | |
| 598 break; | |
| 599 } | |
| 600 if (operators[opern].opernum == LWASM_OPER_NONE) | |
| 601 { | |
| 602 // unrecognized operator | |
| 603 return -1; | |
| 604 } | |
| 605 | |
| 606 // the operator number in question is in opern; i is the length of the | |
| 607 // operator string | |
| 608 | |
| 609 // logic: | |
| 610 // if the precedence of this operation is <= to the "prec" flag, | |
| 611 // we simply return without advancing the input pointer; the operator | |
| 612 // will be evaluated again in the enclosing function call | |
| 613 if (operators[opern].operprec <= prec) | |
| 614 return 0; | |
| 615 | |
| 616 // logic: | |
| 617 // we have a higher precedence operator here so we will advance the | |
| 618 // input pointer to the next term and let the expression evaluator | |
| 619 // loose on it after which time we will push our operator onto the | |
| 620 // stack and then go on with the expression evaluation | |
| 621 (*p) += i; // advance input pointer | |
| 622 | |
| 623 // evaluate next expression(s) of higher precedence | |
| 624 if (lwasm_expr_parse_expr(s, p, operators[opern].operprec) < 0) | |
| 625 return -1; | |
| 626 | |
| 627 operterm = lwasm_expr_term_create_oper(operators[opern].opernum); | |
| 628 lwasm_expr_stack_push(s, operterm); | |
| 629 lwasm_expr_term_free(operterm); | |
| 630 | |
| 631 // return if we are at the end of the expression or a subexpression | |
| 632 if (!**p || isspace(**p) || **p == ')') | |
| 633 return 0; | |
| 634 | |
| 635 // continue evaluating | |
| 636 goto eval_next; | |
| 637 } | |
| 638 | |
| 639 /* | |
| 640 actually evaluate an expression | |
| 641 | |
| 642 This happens in two stages. The first stage merely parses the expression into | |
| 643 a lwasm_expr_stack_t * which is then evaluated as much as possible before the | |
| 644 result is returned. | |
| 645 | |
| 646 Returns NULL on a parse error or otherwise invalid expression. *outp will | |
| 647 contain the pointer to the next character after the expression if and only | |
| 648 if there is no error. In the case of an error, *outp is undefined. | |
| 649 */ | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
650 lwasm_expr_stack_t *lwasm_expr_eval(const char *inp, const char **outp, int (*sfunc)(char *sym, void *state, int *val), void *state) |
| 18 | 651 { |
| 652 lwasm_expr_stack_t *s; | |
| 653 const char *p; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
654 int rval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
655 |
| 18 | 656 // actually parse the expression |
| 657 p = inp; | |
| 658 s = lwasm_expr_stack_create(); | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
659 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
660 rval = lwasm_expr_parse_expr(s, &p, 0); |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
661 if (rval < 0) |
| 18 | 662 goto cleanup_error; |
| 663 | |
| 664 // save end of expression | |
| 665 if (outp) | |
| 666 (*outp) = p; | |
| 667 | |
| 668 // return potentially partial expression | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
669 if (lwasm_expr_reval(s, sfunc, state) < 0) |
| 18 | 670 goto cleanup_error; |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
671 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
672 if (lwasm_expr_is_constant(s)) |
| 39 | 673 debug_message(3, "Constant expression evaluates to: %d", lwasm_expr_get_value(s)); |
| 18 | 674 |
| 675 return s; | |
| 676 | |
| 677 cleanup_error: | |
| 678 lwasm_expr_stack_free(s); | |
| 679 return NULL; | |
| 680 } | |
| 681 | |
| 682 /* | |
| 683 take an expression stack s and scan for operations that can be completed | |
| 684 | |
| 685 return -1 on error, 0 on no error | |
| 686 | |
| 687 possible errors are: division by zero or unknown operator | |
| 688 | |
| 689 theory of operation: | |
| 690 | |
| 691 scan the stack for an operator which has two constants preceding it (binary) | |
| 692 or 1 constant preceding it (unary) and if found, perform the calculation | |
| 693 and replace the operator and its operands with the result | |
| 694 | |
| 695 repeat the scan until no futher simplications are found or if there are no | |
| 696 further operators or only a single term remains | |
| 697 | |
| 698 */ | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
699 int lwasm_expr_reval(lwasm_expr_stack_t *s, int (*sfunc)(char *sym, void *state, int *val), void *state) |
| 18 | 700 { |
| 701 lwasm_expr_stack_node_t *n; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
702 int sval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
703 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
704 // resolve symbols |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
705 // symbols that do not resolve to a constant are left alone |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
706 for (n = s -> head; n; n = n -> next) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
707 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
708 if (n -> term -> term_type == LWASM_TERM_SYM) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
709 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
710 if (sfunc(n -> term -> symbol, state, &sval) == 0) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
711 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
712 n -> term -> term_type = LWASM_TERM_INT; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
713 n -> term -> value = sval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
714 lwasm_free(n -> term -> symbol); |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
715 n -> term -> symbol = NULL; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
716 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
717 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
718 } |
| 18 | 719 |
| 720 next_iter: | |
| 721 // a single term | |
| 722 if (s -> head == s -> tail) | |
| 723 return 0; | |
| 724 | |
| 725 // search for an operator | |
| 726 for (n = s -> head; n; n = n -> next) | |
| 727 { | |
| 728 if (n -> term -> term_type == LWASM_TERM_OPER) | |
| 729 { | |
| 730 if (n -> term -> value == LWASM_OPER_NEG | |
| 731 || n -> term -> value == LWASM_OPER_COM | |
| 732 ) | |
| 733 { | |
| 734 // unary operator | |
| 735 if (n -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 736 { | |
| 737 // a unary operator we can resolve | |
| 738 // we do the op then remove the term "n" is pointing at | |
| 739 if (n -> term -> value == LWASM_OPER_NEG) | |
| 740 { | |
| 741 n -> prev -> term -> value = -(n -> prev -> term -> value); | |
| 742 } | |
| 743 else if (n -> term -> value == LWASM_OPER_COM) | |
| 744 { | |
| 745 n -> prev -> term -> value = ~(n -> prev -> term -> value); | |
| 746 } | |
| 747 n -> prev -> next = n -> next; | |
| 748 if (n -> next) | |
| 749 n -> next -> prev = n -> prev; | |
| 750 else | |
| 751 s -> tail = n -> prev; | |
| 752 | |
| 753 lwasm_expr_term_free(n -> term); | |
| 754 lwasm_free(n); | |
| 755 break; | |
| 756 } | |
| 757 } | |
| 758 else | |
| 759 { | |
| 760 // binary operator | |
| 761 if (n -> prev && n -> prev -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT && n -> prev -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 762 { | |
| 763 // a binary operator we can resolve | |
| 764 switch (n -> term -> value) | |
| 765 { | |
| 766 case LWASM_OPER_PLUS: | |
| 767 n -> prev -> prev -> term -> value += n -> prev -> term -> value; | |
| 768 break; | |
| 769 | |
| 770 case LWASM_OPER_MINUS: | |
| 771 n -> prev -> prev -> term -> value -= n -> prev -> term -> value; | |
| 772 break; | |
| 773 | |
| 774 case LWASM_OPER_TIMES: | |
| 775 n -> prev -> prev -> term -> value *= n -> prev -> term -> value; | |
| 776 break; | |
| 777 | |
| 778 case LWASM_OPER_DIVIDE: | |
| 779 if (n -> prev -> term -> value == 0) | |
| 780 return -1; | |
| 781 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 782 break; | |
| 783 | |
| 784 case LWASM_OPER_MOD: | |
| 785 if (n -> prev -> term -> value == 0) | |
| 786 return -1; | |
| 787 n -> prev -> prev -> term -> value %= n -> prev -> term -> value; | |
| 788 break; | |
| 789 | |
| 790 case LWASM_OPER_INTDIV: | |
| 791 if (n -> prev -> term -> value == 0) | |
| 792 return -1; | |
| 793 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 794 break; | |
| 795 | |
| 796 case LWASM_OPER_BWAND: | |
| 797 n -> prev -> prev -> term -> value &= n -> prev -> term -> value; | |
| 798 break; | |
| 799 | |
| 800 case LWASM_OPER_BWOR: | |
| 801 n -> prev -> prev -> term -> value |= n -> prev -> term -> value; | |
| 802 break; | |
| 803 | |
| 804 case LWASM_OPER_BWXOR: | |
| 805 n -> prev -> prev -> term -> value ^= n -> prev -> term -> value; | |
| 806 break; | |
| 807 | |
| 808 case LWASM_OPER_AND: | |
| 809 n -> prev -> prev -> term -> value = (n -> prev -> term -> value && n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 810 break; | |
| 811 | |
| 812 case LWASM_OPER_OR: | |
| 813 n -> prev -> prev -> term -> value = (n -> prev -> term -> value || n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 814 break; | |
| 815 | |
| 816 default: | |
| 817 // return error if unknown operator! | |
| 818 return -1; | |
| 819 } | |
| 820 | |
| 821 // now remove the two unneeded entries from the stack | |
| 822 n -> prev -> prev -> next = n -> next; | |
| 823 if (n -> next) | |
| 824 n -> next -> prev = n -> prev -> prev; | |
| 825 else | |
| 826 s -> tail = n -> prev -> prev; | |
| 827 | |
| 828 lwasm_expr_term_free(n -> term); | |
| 829 lwasm_expr_term_free(n -> prev -> term); | |
| 830 lwasm_free(n -> prev); | |
| 831 lwasm_free(n); | |
| 832 break; | |
| 833 } | |
| 834 } | |
| 835 } | |
| 836 } | |
| 837 // note for the terminally confused about dynamic memory and pointers: | |
| 838 // n will not be NULL even after the lwasm_free calls above so | |
| 839 // this test will still work (n will be a dangling pointer) | |
| 840 // (n will only be NULL if we didn't find any operators to simplify) | |
| 841 if (n) | |
| 842 goto next_iter; | |
| 843 | |
| 844 return 0; | |
| 845 } |