Mercurial > hg-old > index.cgi
annotate src/expr.c @ 41:7eafdb3a8074
Fixed symbol parsing to prevent string overflow
| author | lost |
|---|---|
| date | Sat, 03 Jan 2009 19:50:30 +0000 |
| parents | d2cee0c335e7 |
| children | 2330b88f9600 |
| 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 == '$') | |
| 348 { | |
| 349 // hexadecimal constant | |
| 350 int val = 0, val2; | |
| 351 | |
| 352 (*p)++; | |
| 353 while (strchr("0123456789ABCDEFabcdef", **p)) | |
| 354 { | |
| 355 val2 = toupper(**p) - '0'; | |
| 356 if (val2 > 9) | |
| 357 val2 -= 7; | |
| 358 val = val * 16 + val2; | |
| 359 (*p)++; | |
| 360 } | |
| 361 t = lwasm_expr_term_create_int(val); | |
| 362 lwasm_expr_stack_push(s, t); | |
| 363 lwasm_expr_term_free(t); | |
| 364 return 0; | |
| 365 } | |
|
40
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
366 // an @ followed by a digit is an octal number |
|
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
367 // but if it's followed by anything else, it is a symbol |
|
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
368 else if (**p == '@' && isdigit(*(*p + 1))) |
| 18 | 369 { |
| 370 // octal constant | |
| 371 int val = 0; | |
| 372 | |
| 373 (*p)++; | |
| 374 while (strchr("01234567", **p)) | |
| 375 { | |
| 376 val = val * 8 + (**p - '0'); | |
| 377 (*p)++; | |
| 378 } | |
| 379 t = lwasm_expr_term_create_int(val); | |
| 380 lwasm_expr_stack_push(s, t); | |
| 381 lwasm_expr_term_free(t); | |
| 382 return 0; | |
| 383 } | |
| 384 | |
| 385 // symbol or bare decimal or suffix identified constant here | |
| 386 // all numbers will start with a digit at this point | |
| 387 if (**p < '0' || **p > '9') | |
| 388 { | |
| 389 int l = 0; | |
| 390 char *sb; | |
| 391 | |
| 392 // evaluate a symbol here | |
| 41 | 393 static const char *symchars = "_.$@?abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"; |
| 394 while ((*p)[l] && strchr(symchars, (*p)[l])) | |
| 18 | 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); | |
| 41 | 406 (*p) += l; |
| 407 debug_message(3, "Symbol: '%s'; (%s)", sb, *p); | |
| 18 | 408 lwasm_free(sb); |
| 409 return 0; | |
| 410 } | |
| 411 | |
| 412 if (!**p) | |
| 413 return -1; | |
| 414 | |
| 415 // evaluate a suffix based constant | |
| 416 { | |
| 417 int decval = 0, binval = 0, hexval = 0, octval = 0; | |
| 418 int valtype = 15; // 1 = bin, 2 = oct, 4 = dec, 8 = hex | |
| 419 int bindone = 0; | |
| 420 int val; | |
| 421 int dval; | |
| 422 | |
| 423 while (1) | |
| 424 { | |
| 425 if (!**p || !strchr("0123456789ABCDEFabcdefqhoQHO", **p)) | |
| 426 { | |
| 427 // we can legally have bin or decimal here | |
| 428 if (bindone) | |
| 429 { | |
| 430 // we just finished a binary value | |
| 431 val = binval; | |
| 432 break; | |
| 433 } | |
| 434 else if (valtype & 4) | |
| 435 { | |
| 436 // otherwise we must be decimal (if we're still allowed one) | |
| 437 val = decval; | |
| 39 | 438 debug_message(3, "End of decimal value"); |
| 18 | 439 break; |
| 440 } | |
| 441 else | |
| 442 { | |
| 443 // bad value | |
| 444 return -1; | |
| 445 } | |
| 446 } | |
| 447 | |
| 448 dval = toupper(**p); | |
| 449 (*p)++; | |
| 450 | |
| 451 if (bindone) | |
| 452 { | |
| 453 // any characters past "B" means it is not binary | |
| 454 bindone = 0; | |
| 455 valtype &= 14; | |
| 456 } | |
| 457 | |
| 458 switch (dval) | |
| 459 { | |
| 460 case 'Q': | |
| 461 case 'O': | |
| 462 if (valtype & 2) | |
| 463 { | |
| 464 val = octval; | |
| 465 valtype = -1; | |
| 466 break; | |
| 467 } | |
| 468 else | |
| 469 { | |
| 470 // not a valid octal value | |
| 471 return -1; | |
| 472 } | |
| 473 /* can't get here */ | |
| 474 | |
| 475 case 'H': | |
| 476 if (valtype & 8) | |
| 477 { | |
| 478 val = hexval; | |
| 479 valtype = -1; | |
| 480 break; | |
| 481 } | |
| 482 else | |
| 483 { | |
| 484 // not a valid hex number | |
| 485 return -1; | |
| 486 } | |
| 487 /* can't get here */ | |
| 488 | |
| 489 case 'B': | |
| 490 // this is a bit of a sticky one since B is a legit hex | |
| 491 // digit so this may or may not be the end of the number | |
| 492 // so we fall through to the digit case | |
| 493 | |
| 494 if (valtype & 1) | |
| 495 { | |
| 496 // could still be binary | |
| 497 bindone = 1; | |
| 498 valtype = 9; // hex and binary | |
| 499 } | |
| 500 /* fall through intentional */ | |
| 501 | |
| 502 default: | |
| 503 // digit | |
| 504 dval -= '0'; | |
| 505 if (dval > 9) | |
| 506 dval -= 7; | |
| 39 | 507 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
|
508 // if (dval > 1) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
509 // valtype &= 14; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
510 // if (dval > 7) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
511 // valtype &= 12; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
512 // if (dval > 9) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
513 // valtype &= 8; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
514 |
| 18 | 515 if (valtype & 8) |
| 516 { | |
| 517 hexval = hexval * 16 + dval; | |
| 518 } | |
| 519 if (valtype & 4) | |
| 520 { | |
| 521 if (dval > 9) | |
| 522 valtype &= 11; | |
| 523 else | |
| 524 decval = decval * 10 + dval; | |
| 525 } | |
| 526 if (valtype & 2) | |
| 527 { | |
| 528 if (dval > 7) | |
| 529 valtype &= 13; | |
| 530 else | |
| 531 octval = octval * 8 + dval; | |
| 532 } | |
| 533 if (valtype & 1) | |
| 534 { | |
| 535 if (dval > 1) | |
| 536 valtype &= 14; | |
| 537 else | |
| 538 binval = binval * 2 + dval; | |
| 539 } | |
| 540 } | |
| 541 // 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
|
542 if (valtype == -1) |
| 18 | 543 break; |
| 544 // return if no more valid possibilities! | |
| 545 if (valtype == 0) | |
| 546 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
547 val = decval; // in case we fall through |
| 18 | 548 } |
| 549 | |
| 550 // we get here when we have a value to return | |
| 551 t = lwasm_expr_term_create_int(val); | |
| 552 lwasm_expr_stack_push(s, t); | |
| 553 lwasm_expr_term_free(t); | |
| 554 return 0; | |
| 555 } | |
| 556 /* can't get here */ | |
| 557 } | |
| 558 | |
| 559 // parse an expression and push the result onto the stack | |
| 560 // if an operator of lower precedence than the value of "prec" is found, | |
| 561 int lwasm_expr_parse_expr(lwasm_expr_stack_t *s, const char **p, int prec) | |
| 562 { | |
| 563 static const struct operinfo | |
| 564 { | |
| 565 int opernum; | |
| 566 char *operstr; | |
| 567 int operprec; | |
| 568 } operators[] = | |
| 569 { | |
| 570 { LWASM_OPER_PLUS, "+", 100 }, | |
| 571 { LWASM_OPER_MINUS, "-", 100 }, | |
| 572 { LWASM_OPER_TIMES, "*", 150 }, | |
| 573 { LWASM_OPER_DIVIDE, "/", 150 }, | |
| 574 { LWASM_OPER_MOD, "%", 150 }, | |
| 575 { LWASM_OPER_INTDIV, "\\", 150 }, | |
| 576 | |
| 577 { LWASM_OPER_NONE, "", 0 } | |
| 578 }; | |
| 579 int opern, i; | |
| 580 lwasm_expr_term_t *operterm; | |
| 581 | |
| 582 // 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
|
583 if (!**p || isspace(**p) || **p == ')' || **p == ',') |
| 18 | 584 return 0; |
| 585 | |
| 586 if (lwasm_expr_parse_term(s, p) < 0) | |
| 587 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
588 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
589 eval_next: |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
590 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
|
591 return 0; |
| 18 | 592 |
| 593 // expecting an operator here | |
| 594 for (opern = 0; operators[opern].opernum != LWASM_OPER_NONE; opern++) | |
| 595 { | |
| 596 for (i = 0; (*p)[i] && operators[opern].operstr[i] && (*p[i] == operators[opern].operstr[i]); i++) | |
| 597 /* do nothing */ ; | |
| 598 if (operators[opern].operstr[i] == '\0') | |
| 599 break; | |
| 600 } | |
| 601 if (operators[opern].opernum == LWASM_OPER_NONE) | |
| 602 { | |
| 603 // unrecognized operator | |
| 604 return -1; | |
| 605 } | |
| 606 | |
| 607 // the operator number in question is in opern; i is the length of the | |
| 608 // operator string | |
| 609 | |
| 610 // logic: | |
| 611 // if the precedence of this operation is <= to the "prec" flag, | |
| 612 // we simply return without advancing the input pointer; the operator | |
| 613 // will be evaluated again in the enclosing function call | |
| 614 if (operators[opern].operprec <= prec) | |
| 615 return 0; | |
| 616 | |
| 617 // logic: | |
| 618 // we have a higher precedence operator here so we will advance the | |
| 619 // input pointer to the next term and let the expression evaluator | |
| 620 // loose on it after which time we will push our operator onto the | |
| 621 // stack and then go on with the expression evaluation | |
| 622 (*p) += i; // advance input pointer | |
| 623 | |
| 624 // evaluate next expression(s) of higher precedence | |
| 625 if (lwasm_expr_parse_expr(s, p, operators[opern].operprec) < 0) | |
| 626 return -1; | |
| 627 | |
| 628 operterm = lwasm_expr_term_create_oper(operators[opern].opernum); | |
| 629 lwasm_expr_stack_push(s, operterm); | |
| 630 lwasm_expr_term_free(operterm); | |
| 631 | |
| 632 // return if we are at the end of the expression or a subexpression | |
| 633 if (!**p || isspace(**p) || **p == ')') | |
| 634 return 0; | |
| 635 | |
| 636 // continue evaluating | |
| 637 goto eval_next; | |
| 638 } | |
| 639 | |
| 640 /* | |
| 641 actually evaluate an expression | |
| 642 | |
| 643 This happens in two stages. The first stage merely parses the expression into | |
| 644 a lwasm_expr_stack_t * which is then evaluated as much as possible before the | |
| 645 result is returned. | |
| 646 | |
| 647 Returns NULL on a parse error or otherwise invalid expression. *outp will | |
| 648 contain the pointer to the next character after the expression if and only | |
| 649 if there is no error. In the case of an error, *outp is undefined. | |
| 650 */ | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
651 lwasm_expr_stack_t *lwasm_expr_eval(const char *inp, const char **outp, int (*sfunc)(char *sym, void *state, int *val), void *state) |
| 18 | 652 { |
| 653 lwasm_expr_stack_t *s; | |
| 654 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
|
655 int rval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
656 |
| 18 | 657 // actually parse the expression |
| 658 p = inp; | |
| 659 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
|
660 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
661 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
|
662 if (rval < 0) |
| 18 | 663 goto cleanup_error; |
| 664 | |
| 665 // save end of expression | |
| 666 if (outp) | |
| 667 (*outp) = p; | |
| 668 | |
| 669 // 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
|
670 if (lwasm_expr_reval(s, sfunc, state) < 0) |
| 18 | 671 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
|
672 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
673 if (lwasm_expr_is_constant(s)) |
| 39 | 674 debug_message(3, "Constant expression evaluates to: %d", lwasm_expr_get_value(s)); |
| 18 | 675 |
| 676 return s; | |
| 677 | |
| 678 cleanup_error: | |
| 679 lwasm_expr_stack_free(s); | |
| 680 return NULL; | |
| 681 } | |
| 682 | |
| 683 /* | |
| 684 take an expression stack s and scan for operations that can be completed | |
| 685 | |
| 686 return -1 on error, 0 on no error | |
| 687 | |
| 688 possible errors are: division by zero or unknown operator | |
| 689 | |
| 690 theory of operation: | |
| 691 | |
| 692 scan the stack for an operator which has two constants preceding it (binary) | |
| 693 or 1 constant preceding it (unary) and if found, perform the calculation | |
| 694 and replace the operator and its operands with the result | |
| 695 | |
| 696 repeat the scan until no futher simplications are found or if there are no | |
| 697 further operators or only a single term remains | |
| 698 | |
| 699 */ | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
700 int lwasm_expr_reval(lwasm_expr_stack_t *s, int (*sfunc)(char *sym, void *state, int *val), void *state) |
| 18 | 701 { |
| 702 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
|
703 int sval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
704 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
705 // resolve symbols |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
706 // 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
|
707 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
|
708 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
709 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
|
710 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
711 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
|
712 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
713 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
|
714 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
|
715 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
|
716 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
|
717 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
718 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
719 } |
| 18 | 720 |
| 721 next_iter: | |
| 722 // a single term | |
| 723 if (s -> head == s -> tail) | |
| 724 return 0; | |
| 725 | |
| 726 // search for an operator | |
| 727 for (n = s -> head; n; n = n -> next) | |
| 728 { | |
| 729 if (n -> term -> term_type == LWASM_TERM_OPER) | |
| 730 { | |
| 731 if (n -> term -> value == LWASM_OPER_NEG | |
| 732 || n -> term -> value == LWASM_OPER_COM | |
| 733 ) | |
| 734 { | |
| 735 // unary operator | |
| 736 if (n -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 737 { | |
| 738 // a unary operator we can resolve | |
| 739 // we do the op then remove the term "n" is pointing at | |
| 740 if (n -> term -> value == LWASM_OPER_NEG) | |
| 741 { | |
| 742 n -> prev -> term -> value = -(n -> prev -> term -> value); | |
| 743 } | |
| 744 else if (n -> term -> value == LWASM_OPER_COM) | |
| 745 { | |
| 746 n -> prev -> term -> value = ~(n -> prev -> term -> value); | |
| 747 } | |
| 748 n -> prev -> next = n -> next; | |
| 749 if (n -> next) | |
| 750 n -> next -> prev = n -> prev; | |
| 751 else | |
| 752 s -> tail = n -> prev; | |
| 753 | |
| 754 lwasm_expr_term_free(n -> term); | |
| 755 lwasm_free(n); | |
| 756 break; | |
| 757 } | |
| 758 } | |
| 759 else | |
| 760 { | |
| 761 // binary operator | |
| 762 if (n -> prev && n -> prev -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT && n -> prev -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 763 { | |
| 764 // a binary operator we can resolve | |
| 765 switch (n -> term -> value) | |
| 766 { | |
| 767 case LWASM_OPER_PLUS: | |
| 768 n -> prev -> prev -> term -> value += n -> prev -> term -> value; | |
| 769 break; | |
| 770 | |
| 771 case LWASM_OPER_MINUS: | |
| 772 n -> prev -> prev -> term -> value -= n -> prev -> term -> value; | |
| 773 break; | |
| 774 | |
| 775 case LWASM_OPER_TIMES: | |
| 776 n -> prev -> prev -> term -> value *= n -> prev -> term -> value; | |
| 777 break; | |
| 778 | |
| 779 case LWASM_OPER_DIVIDE: | |
| 780 if (n -> prev -> term -> value == 0) | |
| 781 return -1; | |
| 782 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 783 break; | |
| 784 | |
| 785 case LWASM_OPER_MOD: | |
| 786 if (n -> prev -> term -> value == 0) | |
| 787 return -1; | |
| 788 n -> prev -> prev -> term -> value %= n -> prev -> term -> value; | |
| 789 break; | |
| 790 | |
| 791 case LWASM_OPER_INTDIV: | |
| 792 if (n -> prev -> term -> value == 0) | |
| 793 return -1; | |
| 794 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 795 break; | |
| 796 | |
| 797 case LWASM_OPER_BWAND: | |
| 798 n -> prev -> prev -> term -> value &= n -> prev -> term -> value; | |
| 799 break; | |
| 800 | |
| 801 case LWASM_OPER_BWOR: | |
| 802 n -> prev -> prev -> term -> value |= n -> prev -> term -> value; | |
| 803 break; | |
| 804 | |
| 805 case LWASM_OPER_BWXOR: | |
| 806 n -> prev -> prev -> term -> value ^= n -> prev -> term -> value; | |
| 807 break; | |
| 808 | |
| 809 case LWASM_OPER_AND: | |
| 810 n -> prev -> prev -> term -> value = (n -> prev -> term -> value && n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 811 break; | |
| 812 | |
| 813 case LWASM_OPER_OR: | |
| 814 n -> prev -> prev -> term -> value = (n -> prev -> term -> value || n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 815 break; | |
| 816 | |
| 817 default: | |
| 818 // return error if unknown operator! | |
| 819 return -1; | |
| 820 } | |
| 821 | |
| 822 // now remove the two unneeded entries from the stack | |
| 823 n -> prev -> prev -> next = n -> next; | |
| 824 if (n -> next) | |
| 825 n -> next -> prev = n -> prev -> prev; | |
| 826 else | |
| 827 s -> tail = n -> prev -> prev; | |
| 828 | |
| 829 lwasm_expr_term_free(n -> term); | |
| 830 lwasm_expr_term_free(n -> prev -> term); | |
| 831 lwasm_free(n -> prev); | |
| 832 lwasm_free(n); | |
| 833 break; | |
| 834 } | |
| 835 } | |
| 836 } | |
| 837 } | |
| 838 // note for the terminally confused about dynamic memory and pointers: | |
| 839 // n will not be NULL even after the lwasm_free calls above so | |
| 840 // this test will still work (n will be a dangling pointer) | |
| 841 // (n will only be NULL if we didn't find any operators to simplify) | |
| 842 if (n) | |
| 843 goto next_iter; | |
| 844 | |
| 845 return 0; | |
| 846 } |