Mercurial > hg-old > index.cgi
annotate src/expr.c @ 61:73423b66e511
Fixed error with single char ascii constants
| author | lost |
|---|---|
| date | Mon, 05 Jan 2009 00:57:14 +0000 |
| parents | 804d7465e0f9 |
| children | 2fe5fd7d65a3 |
| 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 { | |
| 44 | 61 if (t -> term_type == LWASM_TERM_SYM) |
| 17 | 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: |
| 61 | 188 if (!**p || isspace(**p) || **p == ')' || **p == ']') |
| 189 return -1; | |
| 18 | 190 if (**p == '(') |
| 191 { | |
| 39 | 192 debug_message(3, "Starting paren"); |
| 18 | 193 (*p)++; |
| 194 lwasm_expr_parse_expr(s, p, 0); | |
| 195 if (**p != ')') | |
| 196 return -1; | |
| 197 (*p)++; | |
| 198 return 0; | |
| 199 } | |
| 200 | |
| 201 if (**p == '+') | |
| 202 { | |
| 39 | 203 debug_message(3, "Unary +"); |
| 18 | 204 (*p)++; |
| 205 goto eval_next; | |
| 206 } | |
| 207 | |
| 208 if (**p == '-') | |
| 209 { | |
| 210 // parse expression following "-" | |
| 211 (*p)++; | |
| 212 if (lwasm_expr_parse_expr(s, p, 200) < 0) | |
| 213 return -1; | |
| 214 t = lwasm_expr_term_create_oper(LWASM_OPER_NEG); | |
| 215 lwasm_expr_stack_push(s, t); | |
| 216 lwasm_expr_term_free(t); | |
| 217 return 0; | |
| 218 } | |
| 219 | |
| 23 | 220 if (**p == '^') |
| 221 { | |
| 222 // parse expression following "^" | |
| 223 (*p)++; | |
| 224 if (lwasm_expr_parse_expr(s, p, 200) < 0) | |
| 225 return -1; | |
| 226 t = lwasm_expr_term_create_oper(LWASM_OPER_COM); | |
| 227 lwasm_expr_stack_push(s, t); | |
| 228 lwasm_expr_term_free(t); | |
| 229 return 0; | |
| 230 } | |
| 231 | |
| 18 | 232 /* |
| 233 we have an actual term here so evaluate it | |
| 234 | |
| 235 it could be one of the following: | |
| 236 | |
| 237 1. a decimal constant | |
| 238 2. a hexadecimal constant | |
| 239 3. an octal constant | |
| 240 4. a binary constant | |
| 241 5. a symbol reference | |
| 242 6. the "current" instruction address (*) | |
| 243 7. the "current" data address (.) | |
| 244 8. a "back reference" (<) | |
| 245 9. a "forward reference" (>) | |
| 246 | |
| 247 items 6 through 9 are stored as symbol references | |
| 248 | |
| 249 (a . followed by a . or a alpha char or number is a symbol) | |
| 250 */ | |
| 251 if (**p == '*' | |
| 252 || ( | |
| 253 **p == '.' | |
| 254 && (*p)[1] != '.' | |
| 255 && !((*p)[1] >= 'A' && (*p)[1] <= 'Z') | |
| 256 && !((*p)[1] >= 'a' && (*p)[1] <= 'z') | |
| 257 && !((*p)[1] >= '0' && (*p)[1] <= '9') | |
| 258 ) | |
| 259 || **p == '<' | |
| 260 || **p == '>') | |
| 261 { | |
| 262 char tstr[2]; | |
| 263 tstr[0] = **p; | |
| 264 tstr[1] = '\0'; | |
| 265 t = lwasm_expr_term_create_sym(tstr); | |
| 266 lwasm_expr_stack_push(s, t); | |
| 267 lwasm_expr_term_free(t); | |
| 268 (*p)++; | |
| 269 return 0; | |
| 270 } | |
| 271 | |
| 272 /* | |
| 273 - a symbol will be a string of characters introduced by a letter, ".", | |
| 274 "_" but NOT a number | |
| 275 - a decimal constant will consist of only digits, optionally prefixed | |
| 276 with "&" | |
| 277 - a binary constant will consist of only 0s and 1s either prefixed with % | |
| 278 or suffixed with "B" | |
| 279 - a hex constant will consist of 0-9A-F either prefixed with $ or | |
| 280 suffixed with "H"; a hex number starting with A-F must be prefixed | |
| 281 with $ or start with 0 and end with H | |
| 282 - an octal constant will consist of 0-7 either prefixed with @ or | |
| 283 suffixed with "O" or "Q" | |
| 284 - an ascii constant will be a single character prefixed with a ' | |
| 285 - a double ascii constant will be two characters prefixed with a " | |
| 286 | |
| 287 */ | |
| 288 if (**p == '"') | |
| 289 { | |
| 290 // double ascii constant | |
| 291 int val; | |
| 292 (*p)++; | |
| 293 if (!**p) | |
| 294 return -1; | |
| 295 if (!*((*p)+1)) | |
| 296 return -1; | |
| 297 val = **p << 8 | *((*p) + 1); | |
| 298 (*p) += 2; | |
| 299 t = lwasm_expr_term_create_int(val); | |
| 300 lwasm_expr_stack_push(s, t); | |
| 301 lwasm_expr_term_free(t); | |
| 302 return 0; | |
| 303 } | |
| 304 else if (**p == '\'') | |
| 305 { | |
| 306 // single ascii constant | |
| 307 int val; | |
| 308 (*p)++; | |
| 61 | 309 debug_message(3, "Single ascii character constant '%c'", **p); |
| 18 | 310 if (!**p) |
| 311 return -1; | |
| 312 val = **p; | |
| 313 (*p)++; | |
| 314 t = lwasm_expr_term_create_int(val); | |
| 315 lwasm_expr_stack_push(s, t); | |
| 316 lwasm_expr_term_free(t); | |
| 61 | 317 return 0; |
| 18 | 318 } |
| 319 else if (**p == '&') | |
| 320 { | |
| 321 // decimal constant | |
| 322 int val = 0; | |
| 323 | |
| 324 (*p)++; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
325 while (**p && strchr("0123456789", **p)) |
| 18 | 326 { |
| 327 val = val * 10 + (**p - '0'); | |
| 328 (*p)++; | |
| 329 } | |
| 330 t = lwasm_expr_term_create_int(val); | |
| 331 lwasm_expr_stack_push(s, t); | |
| 332 lwasm_expr_term_free(t); | |
| 333 return 0; | |
| 334 } | |
| 335 else if (**p == '%') | |
| 336 { | |
| 337 // binary constant | |
| 338 int val = 0; | |
| 339 | |
| 340 (*p)++; | |
| 341 while (**p == '0' || **p == '1') | |
| 342 { | |
| 343 val = val * 2 + (**p - '0'); | |
| 344 (*p)++; | |
| 345 } | |
| 346 t = lwasm_expr_term_create_int(val); | |
| 347 lwasm_expr_stack_push(s, t); | |
| 348 lwasm_expr_term_free(t); | |
| 349 return 0; | |
| 350 } | |
| 351 else if (**p == '$') | |
| 352 { | |
| 353 // hexadecimal constant | |
| 354 int val = 0, val2; | |
| 355 | |
| 356 (*p)++; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
357 debug_message(3, "Found prefix hex constant: %s", *p); |
|
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
358 while (**p && strchr("0123456789ABCDEFabcdef", **p)) |
| 18 | 359 { |
| 360 val2 = toupper(**p) - '0'; | |
| 361 if (val2 > 9) | |
| 362 val2 -= 7; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
363 debug_message(3, "Got char: %c (%d)", **p, val2); |
| 18 | 364 val = val * 16 + val2; |
| 365 (*p)++; | |
| 366 } | |
| 367 t = lwasm_expr_term_create_int(val); | |
| 368 lwasm_expr_stack_push(s, t); | |
| 369 lwasm_expr_term_free(t); | |
| 370 return 0; | |
| 371 } | |
|
40
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
372 // 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
|
373 // 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
|
374 else if (**p == '@' && isdigit(*(*p + 1))) |
| 18 | 375 { |
| 376 // octal constant | |
| 377 int val = 0; | |
| 378 | |
| 379 (*p)++; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
380 while (**p && strchr("01234567", **p)) |
| 18 | 381 { |
| 382 val = val * 8 + (**p - '0'); | |
| 383 (*p)++; | |
| 384 } | |
| 385 t = lwasm_expr_term_create_int(val); | |
| 386 lwasm_expr_stack_push(s, t); | |
| 387 lwasm_expr_term_free(t); | |
| 388 return 0; | |
| 389 } | |
| 390 | |
| 391 // symbol or bare decimal or suffix identified constant here | |
| 392 // all numbers will start with a digit at this point | |
| 393 if (**p < '0' || **p > '9') | |
| 394 { | |
| 395 int l = 0; | |
| 396 char *sb; | |
| 397 | |
| 398 // evaluate a symbol here | |
| 41 | 399 static const char *symchars = "_.$@?abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"; |
| 400 while ((*p)[l] && strchr(symchars, (*p)[l])) | |
| 18 | 401 l++; |
| 402 | |
| 403 if (l == 0) | |
| 404 return -1; | |
| 405 | |
| 406 sb = lwasm_alloc(l + 1); | |
| 407 sb[l] = '\0'; | |
| 408 memcpy(sb, *p, l); | |
| 409 t = lwasm_expr_term_create_sym(sb); | |
| 410 lwasm_expr_stack_push(s, t); | |
| 411 lwasm_expr_term_free(t); | |
| 41 | 412 (*p) += l; |
| 413 debug_message(3, "Symbol: '%s'; (%s)", sb, *p); | |
| 18 | 414 lwasm_free(sb); |
| 415 return 0; | |
| 416 } | |
| 417 | |
| 418 if (!**p) | |
| 419 return -1; | |
| 420 | |
| 421 // evaluate a suffix based constant | |
| 422 { | |
| 423 int decval = 0, binval = 0, hexval = 0, octval = 0; | |
| 424 int valtype = 15; // 1 = bin, 2 = oct, 4 = dec, 8 = hex | |
| 425 int bindone = 0; | |
| 426 int val; | |
| 427 int dval; | |
| 428 | |
| 429 while (1) | |
| 430 { | |
| 431 if (!**p || !strchr("0123456789ABCDEFabcdefqhoQHO", **p)) | |
| 432 { | |
| 433 // we can legally have bin or decimal here | |
| 434 if (bindone) | |
| 435 { | |
| 436 // we just finished a binary value | |
| 437 val = binval; | |
| 438 break; | |
| 439 } | |
| 440 else if (valtype & 4) | |
| 441 { | |
| 442 // otherwise we must be decimal (if we're still allowed one) | |
| 443 val = decval; | |
| 39 | 444 debug_message(3, "End of decimal value"); |
| 18 | 445 break; |
| 446 } | |
| 447 else | |
| 448 { | |
| 449 // bad value | |
| 450 return -1; | |
| 451 } | |
| 452 } | |
| 453 | |
| 454 dval = toupper(**p); | |
| 455 (*p)++; | |
| 456 | |
| 457 if (bindone) | |
| 458 { | |
| 459 // any characters past "B" means it is not binary | |
| 460 bindone = 0; | |
| 461 valtype &= 14; | |
| 462 } | |
| 463 | |
| 464 switch (dval) | |
| 465 { | |
| 466 case 'Q': | |
| 467 case 'O': | |
| 468 if (valtype & 2) | |
| 469 { | |
| 470 val = octval; | |
| 471 valtype = -1; | |
| 472 break; | |
| 473 } | |
| 474 else | |
| 475 { | |
| 476 // not a valid octal value | |
| 477 return -1; | |
| 478 } | |
| 479 /* can't get here */ | |
| 480 | |
| 481 case 'H': | |
| 482 if (valtype & 8) | |
| 483 { | |
| 484 val = hexval; | |
| 485 valtype = -1; | |
| 486 break; | |
| 487 } | |
| 488 else | |
| 489 { | |
| 490 // not a valid hex number | |
| 491 return -1; | |
| 492 } | |
| 493 /* can't get here */ | |
| 494 | |
| 495 case 'B': | |
| 496 // this is a bit of a sticky one since B is a legit hex | |
| 497 // digit so this may or may not be the end of the number | |
| 498 // so we fall through to the digit case | |
| 499 | |
| 500 if (valtype & 1) | |
| 501 { | |
| 502 // could still be binary | |
| 503 bindone = 1; | |
| 504 valtype = 9; // hex and binary | |
| 505 } | |
| 506 /* fall through intentional */ | |
| 507 | |
| 508 default: | |
| 509 // digit | |
| 510 dval -= '0'; | |
| 511 if (dval > 9) | |
| 512 dval -= 7; | |
| 39 | 513 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
|
514 // if (dval > 1) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
515 // valtype &= 14; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
516 // if (dval > 7) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
517 // valtype &= 12; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
518 // if (dval > 9) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
519 // valtype &= 8; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
520 |
| 18 | 521 if (valtype & 8) |
| 522 { | |
| 523 hexval = hexval * 16 + dval; | |
| 524 } | |
| 525 if (valtype & 4) | |
| 526 { | |
| 527 if (dval > 9) | |
| 528 valtype &= 11; | |
| 529 else | |
| 530 decval = decval * 10 + dval; | |
| 531 } | |
| 532 if (valtype & 2) | |
| 533 { | |
| 534 if (dval > 7) | |
| 535 valtype &= 13; | |
| 536 else | |
| 537 octval = octval * 8 + dval; | |
| 538 } | |
| 539 if (valtype & 1) | |
| 540 { | |
| 541 if (dval > 1) | |
| 542 valtype &= 14; | |
| 543 else | |
| 544 binval = binval * 2 + dval; | |
| 545 } | |
| 546 } | |
| 547 // 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
|
548 if (valtype == -1) |
| 18 | 549 break; |
| 550 // return if no more valid possibilities! | |
| 551 if (valtype == 0) | |
| 552 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
553 val = decval; // in case we fall through |
| 18 | 554 } |
| 555 | |
| 556 // we get here when we have a value to return | |
| 557 t = lwasm_expr_term_create_int(val); | |
| 558 lwasm_expr_stack_push(s, t); | |
| 559 lwasm_expr_term_free(t); | |
| 560 return 0; | |
| 561 } | |
| 562 /* can't get here */ | |
| 563 } | |
| 564 | |
| 565 // parse an expression and push the result onto the stack | |
| 566 // if an operator of lower precedence than the value of "prec" is found, | |
| 567 int lwasm_expr_parse_expr(lwasm_expr_stack_t *s, const char **p, int prec) | |
| 568 { | |
| 569 static const struct operinfo | |
| 570 { | |
| 571 int opernum; | |
| 572 char *operstr; | |
| 573 int operprec; | |
| 574 } operators[] = | |
| 575 { | |
| 576 { LWASM_OPER_PLUS, "+", 100 }, | |
| 577 { LWASM_OPER_MINUS, "-", 100 }, | |
| 578 { LWASM_OPER_TIMES, "*", 150 }, | |
| 579 { LWASM_OPER_DIVIDE, "/", 150 }, | |
| 580 { LWASM_OPER_MOD, "%", 150 }, | |
| 581 { LWASM_OPER_INTDIV, "\\", 150 }, | |
| 582 | |
| 583 { LWASM_OPER_NONE, "", 0 } | |
| 584 }; | |
| 585 int opern, i; | |
| 586 lwasm_expr_term_t *operterm; | |
| 587 | |
| 588 // return if we are at the end of the expression or a subexpression | |
| 61 | 589 if (!**p || isspace(**p) || **p == ')' || **p == ',' || **p == ']') |
| 18 | 590 return 0; |
| 591 | |
| 592 if (lwasm_expr_parse_term(s, p) < 0) | |
| 593 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
594 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
595 eval_next: |
| 61 | 596 if (!**p || isspace(**p) || **p == ')' || **p == ',' || **p == ']') |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
597 return 0; |
| 18 | 598 |
| 599 // expecting an operator here | |
| 600 for (opern = 0; operators[opern].opernum != LWASM_OPER_NONE; opern++) | |
| 601 { | |
| 602 for (i = 0; (*p)[i] && operators[opern].operstr[i] && (*p[i] == operators[opern].operstr[i]); i++) | |
| 603 /* do nothing */ ; | |
| 604 if (operators[opern].operstr[i] == '\0') | |
| 605 break; | |
| 606 } | |
| 607 if (operators[opern].opernum == LWASM_OPER_NONE) | |
| 608 { | |
| 609 // unrecognized operator | |
| 610 return -1; | |
| 611 } | |
| 612 | |
| 613 // the operator number in question is in opern; i is the length of the | |
| 614 // operator string | |
| 615 | |
| 616 // logic: | |
| 617 // if the precedence of this operation is <= to the "prec" flag, | |
| 618 // we simply return without advancing the input pointer; the operator | |
| 619 // will be evaluated again in the enclosing function call | |
| 620 if (operators[opern].operprec <= prec) | |
| 621 return 0; | |
| 622 | |
| 623 // logic: | |
| 624 // we have a higher precedence operator here so we will advance the | |
| 625 // input pointer to the next term and let the expression evaluator | |
| 626 // loose on it after which time we will push our operator onto the | |
| 627 // stack and then go on with the expression evaluation | |
| 628 (*p) += i; // advance input pointer | |
| 629 | |
| 630 // evaluate next expression(s) of higher precedence | |
| 631 if (lwasm_expr_parse_expr(s, p, operators[opern].operprec) < 0) | |
| 632 return -1; | |
| 633 | |
| 634 operterm = lwasm_expr_term_create_oper(operators[opern].opernum); | |
| 635 lwasm_expr_stack_push(s, operterm); | |
| 636 lwasm_expr_term_free(operterm); | |
| 637 | |
| 638 // return if we are at the end of the expression or a subexpression | |
| 639 if (!**p || isspace(**p) || **p == ')') | |
| 640 return 0; | |
| 641 | |
| 642 // continue evaluating | |
| 643 goto eval_next; | |
| 644 } | |
| 645 | |
| 646 /* | |
| 647 actually evaluate an expression | |
| 648 | |
| 649 This happens in two stages. The first stage merely parses the expression into | |
| 650 a lwasm_expr_stack_t * which is then evaluated as much as possible before the | |
| 651 result is returned. | |
| 652 | |
| 653 Returns NULL on a parse error or otherwise invalid expression. *outp will | |
| 654 contain the pointer to the next character after the expression if and only | |
| 655 if there is no error. In the case of an error, *outp is undefined. | |
| 656 */ | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
657 lwasm_expr_stack_t *lwasm_expr_eval(const char *inp, const char **outp, int (*sfunc)(char *sym, void *state, int *val), void *state) |
| 18 | 658 { |
| 659 lwasm_expr_stack_t *s; | |
| 660 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
|
661 int rval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
662 |
| 18 | 663 // actually parse the expression |
| 664 p = inp; | |
| 665 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
|
666 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
667 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
|
668 if (rval < 0) |
| 18 | 669 goto cleanup_error; |
| 670 | |
| 671 // save end of expression | |
| 672 if (outp) | |
| 673 (*outp) = p; | |
| 674 | |
| 675 // 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
|
676 if (lwasm_expr_reval(s, sfunc, state) < 0) |
| 18 | 677 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
|
678 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
679 if (lwasm_expr_is_constant(s)) |
| 39 | 680 debug_message(3, "Constant expression evaluates to: %d", lwasm_expr_get_value(s)); |
| 18 | 681 |
| 682 return s; | |
| 683 | |
| 684 cleanup_error: | |
| 685 lwasm_expr_stack_free(s); | |
| 686 return NULL; | |
| 687 } | |
| 688 | |
| 689 /* | |
| 690 take an expression stack s and scan for operations that can be completed | |
| 691 | |
| 692 return -1 on error, 0 on no error | |
| 693 | |
| 694 possible errors are: division by zero or unknown operator | |
| 695 | |
| 696 theory of operation: | |
| 697 | |
| 698 scan the stack for an operator which has two constants preceding it (binary) | |
| 699 or 1 constant preceding it (unary) and if found, perform the calculation | |
| 700 and replace the operator and its operands with the result | |
| 701 | |
| 702 repeat the scan until no futher simplications are found or if there are no | |
| 703 further operators or only a single term remains | |
| 704 | |
| 705 */ | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
706 int lwasm_expr_reval(lwasm_expr_stack_t *s, int (*sfunc)(char *sym, void *state, int *val), void *state) |
| 18 | 707 { |
| 708 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
|
709 int sval; |
|
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 // resolve symbols |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
712 // 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
|
713 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
|
714 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
715 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
|
716 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
717 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
|
718 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
719 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
|
720 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
|
721 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
|
722 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
|
723 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
724 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
725 } |
| 18 | 726 |
| 727 next_iter: | |
| 728 // a single term | |
| 729 if (s -> head == s -> tail) | |
| 730 return 0; | |
| 731 | |
| 732 // search for an operator | |
| 733 for (n = s -> head; n; n = n -> next) | |
| 734 { | |
| 735 if (n -> term -> term_type == LWASM_TERM_OPER) | |
| 736 { | |
| 737 if (n -> term -> value == LWASM_OPER_NEG | |
| 738 || n -> term -> value == LWASM_OPER_COM | |
| 739 ) | |
| 740 { | |
| 741 // unary operator | |
| 742 if (n -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 743 { | |
| 744 // a unary operator we can resolve | |
| 745 // we do the op then remove the term "n" is pointing at | |
| 746 if (n -> term -> value == LWASM_OPER_NEG) | |
| 747 { | |
| 748 n -> prev -> term -> value = -(n -> prev -> term -> value); | |
| 749 } | |
| 750 else if (n -> term -> value == LWASM_OPER_COM) | |
| 751 { | |
| 752 n -> prev -> term -> value = ~(n -> prev -> term -> value); | |
| 753 } | |
| 754 n -> prev -> next = n -> next; | |
| 755 if (n -> next) | |
| 756 n -> next -> prev = n -> prev; | |
| 757 else | |
| 758 s -> tail = n -> prev; | |
| 759 | |
| 760 lwasm_expr_term_free(n -> term); | |
| 761 lwasm_free(n); | |
| 762 break; | |
| 763 } | |
| 764 } | |
| 765 else | |
| 766 { | |
| 767 // binary operator | |
| 768 if (n -> prev && n -> prev -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT && n -> prev -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 769 { | |
| 770 // a binary operator we can resolve | |
| 771 switch (n -> term -> value) | |
| 772 { | |
| 773 case LWASM_OPER_PLUS: | |
| 774 n -> prev -> prev -> term -> value += n -> prev -> term -> value; | |
| 775 break; | |
| 776 | |
| 777 case LWASM_OPER_MINUS: | |
| 778 n -> prev -> prev -> term -> value -= n -> prev -> term -> value; | |
| 779 break; | |
| 780 | |
| 781 case LWASM_OPER_TIMES: | |
| 782 n -> prev -> prev -> term -> value *= n -> prev -> term -> value; | |
| 783 break; | |
| 784 | |
| 785 case LWASM_OPER_DIVIDE: | |
| 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_MOD: | |
| 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_INTDIV: | |
| 798 if (n -> prev -> term -> value == 0) | |
| 799 return -1; | |
| 800 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 801 break; | |
| 802 | |
| 803 case LWASM_OPER_BWAND: | |
| 804 n -> prev -> prev -> term -> value &= n -> prev -> term -> value; | |
| 805 break; | |
| 806 | |
| 807 case LWASM_OPER_BWOR: | |
| 808 n -> prev -> prev -> term -> value |= n -> prev -> term -> value; | |
| 809 break; | |
| 810 | |
| 811 case LWASM_OPER_BWXOR: | |
| 812 n -> prev -> prev -> term -> value ^= n -> prev -> term -> value; | |
| 813 break; | |
| 814 | |
| 815 case LWASM_OPER_AND: | |
| 816 n -> prev -> prev -> term -> value = (n -> prev -> term -> value && n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 817 break; | |
| 818 | |
| 819 case LWASM_OPER_OR: | |
| 820 n -> prev -> prev -> term -> value = (n -> prev -> term -> value || n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 821 break; | |
| 822 | |
| 823 default: | |
| 824 // return error if unknown operator! | |
| 825 return -1; | |
| 826 } | |
| 827 | |
| 828 // now remove the two unneeded entries from the stack | |
| 829 n -> prev -> prev -> next = n -> next; | |
| 830 if (n -> next) | |
| 831 n -> next -> prev = n -> prev -> prev; | |
| 832 else | |
| 833 s -> tail = n -> prev -> prev; | |
| 834 | |
| 835 lwasm_expr_term_free(n -> term); | |
| 836 lwasm_expr_term_free(n -> prev -> term); | |
| 837 lwasm_free(n -> prev); | |
| 838 lwasm_free(n); | |
| 839 break; | |
| 840 } | |
| 841 } | |
| 842 } | |
| 843 } | |
| 844 // note for the terminally confused about dynamic memory and pointers: | |
| 845 // n will not be NULL even after the lwasm_free calls above so | |
| 846 // this test will still work (n will be a dangling pointer) | |
| 847 // (n will only be NULL if we didn't find any operators to simplify) | |
| 848 if (n) | |
| 849 goto next_iter; | |
| 850 | |
| 851 return 0; | |
| 852 } |