Mercurial > hg-old > index.cgi
annotate src/expr.c @ 159:71561c12b20b
Updated docs to reflect new cescapes pragma and discuss implicit assumption of the bss section flag for sections named bss and .bss
| author | lost |
|---|---|
| date | Sat, 31 Jan 2009 06:32:27 +0000 |
| parents | 718998b673ee |
| children |
| 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 | |
|
91
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
90 lwasm_expr_term_t *lwasm_expr_term_create_secbase(void) |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
91 { |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
92 lwasm_expr_term_t *t; |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
93 debug_message(10, "Creating section base term"); |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
94 |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
95 t = lwasm_alloc(sizeof(lwasm_expr_term_t)); |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
96 t -> term_type = LWASM_TERM_SECBASE; |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
97 return t; |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
98 } |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
99 |
| 17 | 100 lwasm_expr_term_t *lwasm_expr_term_create_sym(char *sym) |
| 101 { | |
| 102 lwasm_expr_term_t *t; | |
| 14 | 103 |
| 39 | 104 debug_message(10, "Creating symbol term: %s", sym); |
| 105 | |
| 17 | 106 t = lwasm_alloc(sizeof(lwasm_expr_term_t)); |
| 107 t -> term_type = LWASM_TERM_SYM; | |
| 108 t -> symbol = lwasm_strdup(sym); | |
| 109 return t; | |
| 110 } | |
| 15 | 111 |
| 17 | 112 lwasm_expr_term_t *lwasm_expr_term_dup(lwasm_expr_term_t *t) |
| 113 { | |
| 114 switch (t -> term_type) | |
| 15 | 115 { |
| 17 | 116 case LWASM_TERM_INT: |
| 117 return lwasm_expr_term_create_int(t -> value); | |
| 118 | |
| 119 case LWASM_TERM_OPER: | |
| 120 return lwasm_expr_term_create_oper(t -> value); | |
| 121 | |
| 122 case LWASM_TERM_SYM: | |
| 123 return lwasm_expr_term_create_sym(t -> symbol); | |
|
91
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
124 |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
125 case LWASM_TERM_SECBASE: |
|
718998b673ee
Added incomplete references to object output and added support for section base terms in expression handler
lost
parents:
76
diff
changeset
|
126 return lwasm_expr_term_create_secbase(); |
| 17 | 127 |
| 128 default: | |
| 39 | 129 debug_message(0, "lwasm_expr_term_dup(): invalid term type %d", t -> term_type); |
| 17 | 130 exit(1); |
| 131 } | |
| 132 // can't get here | |
| 133 } | |
| 134 | |
| 135 void lwasm_expr_stack_push(lwasm_expr_stack_t *s, lwasm_expr_term_t *t) | |
| 136 { | |
| 137 lwasm_expr_stack_node_t *n; | |
| 138 | |
| 139 if (!s) | |
| 140 { | |
| 39 | 141 debug_message(0, "lwasm_expr_stack_push(): invalid stack pointer"); |
| 17 | 142 exit(1); |
| 15 | 143 } |
| 144 | |
| 17 | 145 n = lwasm_alloc(sizeof(lwasm_expr_stack_node_t)); |
| 146 n -> next = NULL; | |
| 147 n -> prev = s -> tail; | |
| 148 n -> term = lwasm_expr_term_dup(t); | |
| 149 | |
| 150 if (s -> head) | |
| 151 { | |
| 152 s -> tail -> next = n; | |
| 153 s -> tail = n; | |
| 154 } | |
| 155 else | |
| 15 | 156 { |
| 17 | 157 s -> head = n; |
| 158 s -> tail = n; | |
| 159 } | |
| 160 } | |
| 161 | |
| 162 lwasm_expr_term_t *lwasm_expr_stack_pop(lwasm_expr_stack_t *s) | |
| 163 { | |
| 164 lwasm_expr_term_t *t; | |
| 165 lwasm_expr_stack_node_t *n; | |
| 166 | |
| 167 if (!(s -> tail)) | |
| 168 return NULL; | |
| 169 | |
| 170 n = s -> tail; | |
| 171 s -> tail = n -> prev; | |
| 172 if (!(n -> prev)) | |
| 173 { | |
| 174 s -> head = NULL; | |
| 15 | 175 } |
| 14 | 176 |
| 17 | 177 t = n -> term; |
| 178 n -> term = NULL; | |
| 179 | |
| 180 lwasm_free(n); | |
| 181 | |
| 182 return t; | |
| 14 | 183 } |
| 18 | 184 |
| 185 // the following two functions are co-routines which actually parse | |
| 186 // an infix expression onto the expression stack, each returns -1 | |
| 187 // if an error is encountered | |
| 188 | |
| 189 /* | |
| 190 parse a term and push it onto the stack | |
| 191 | |
| 192 this function handles unary prefix operators (-, +, .not., .com.) | |
| 193 as well as () | |
| 194 */ | |
| 195 int lwasm_expr_parse_term(lwasm_expr_stack_t *s, const char **p) | |
| 196 { | |
| 197 lwasm_expr_term_t *t; | |
| 39 | 198 debug_message(2, "Expression string %s", *p); |
| 199 | |
| 18 | 200 eval_next: |
| 61 | 201 if (!**p || isspace(**p) || **p == ')' || **p == ']') |
| 202 return -1; | |
| 18 | 203 if (**p == '(') |
| 204 { | |
| 39 | 205 debug_message(3, "Starting paren"); |
| 18 | 206 (*p)++; |
| 207 lwasm_expr_parse_expr(s, p, 0); | |
| 208 if (**p != ')') | |
| 209 return -1; | |
| 210 (*p)++; | |
| 211 return 0; | |
| 212 } | |
| 213 | |
| 214 if (**p == '+') | |
| 215 { | |
| 39 | 216 debug_message(3, "Unary +"); |
| 18 | 217 (*p)++; |
| 218 goto eval_next; | |
| 219 } | |
| 220 | |
| 221 if (**p == '-') | |
| 222 { | |
| 223 // parse expression following "-" | |
| 224 (*p)++; | |
| 225 if (lwasm_expr_parse_expr(s, p, 200) < 0) | |
| 226 return -1; | |
| 227 t = lwasm_expr_term_create_oper(LWASM_OPER_NEG); | |
| 228 lwasm_expr_stack_push(s, t); | |
| 229 lwasm_expr_term_free(t); | |
| 230 return 0; | |
| 231 } | |
| 232 | |
| 23 | 233 if (**p == '^') |
| 234 { | |
| 235 // parse expression following "^" | |
| 236 (*p)++; | |
| 237 if (lwasm_expr_parse_expr(s, p, 200) < 0) | |
| 238 return -1; | |
| 239 t = lwasm_expr_term_create_oper(LWASM_OPER_COM); | |
| 240 lwasm_expr_stack_push(s, t); | |
| 241 lwasm_expr_term_free(t); | |
| 242 return 0; | |
| 243 } | |
| 244 | |
| 18 | 245 /* |
| 246 we have an actual term here so evaluate it | |
| 247 | |
| 248 it could be one of the following: | |
| 249 | |
| 250 1. a decimal constant | |
| 251 2. a hexadecimal constant | |
| 252 3. an octal constant | |
| 253 4. a binary constant | |
| 254 5. a symbol reference | |
| 255 6. the "current" instruction address (*) | |
| 256 7. the "current" data address (.) | |
| 257 8. a "back reference" (<) | |
| 258 9. a "forward reference" (>) | |
| 259 | |
| 260 items 6 through 9 are stored as symbol references | |
| 261 | |
| 262 (a . followed by a . or a alpha char or number is a symbol) | |
| 263 */ | |
| 264 if (**p == '*' | |
| 265 || ( | |
| 266 **p == '.' | |
| 267 && (*p)[1] != '.' | |
| 268 && !((*p)[1] >= 'A' && (*p)[1] <= 'Z') | |
| 269 && !((*p)[1] >= 'a' && (*p)[1] <= 'z') | |
| 270 && !((*p)[1] >= '0' && (*p)[1] <= '9') | |
| 271 ) | |
| 272 || **p == '<' | |
| 273 || **p == '>') | |
| 274 { | |
| 275 char tstr[2]; | |
| 276 tstr[0] = **p; | |
| 277 tstr[1] = '\0'; | |
| 278 t = lwasm_expr_term_create_sym(tstr); | |
| 279 lwasm_expr_stack_push(s, t); | |
| 280 lwasm_expr_term_free(t); | |
| 281 (*p)++; | |
| 282 return 0; | |
| 283 } | |
| 284 | |
| 285 /* | |
| 286 - a symbol will be a string of characters introduced by a letter, ".", | |
| 287 "_" but NOT a number | |
| 288 - a decimal constant will consist of only digits, optionally prefixed | |
| 289 with "&" | |
| 290 - a binary constant will consist of only 0s and 1s either prefixed with % | |
| 291 or suffixed with "B" | |
| 292 - a hex constant will consist of 0-9A-F either prefixed with $ or | |
| 293 suffixed with "H"; a hex number starting with A-F must be prefixed | |
| 294 with $ or start with 0 and end with H | |
| 295 - an octal constant will consist of 0-7 either prefixed with @ or | |
| 296 suffixed with "O" or "Q" | |
| 297 - an ascii constant will be a single character prefixed with a ' | |
| 298 - a double ascii constant will be two characters prefixed with a " | |
| 299 | |
| 300 */ | |
| 301 if (**p == '"') | |
| 302 { | |
| 303 // double ascii constant | |
| 304 int val; | |
| 305 (*p)++; | |
| 306 if (!**p) | |
| 307 return -1; | |
| 308 if (!*((*p)+1)) | |
| 309 return -1; | |
| 310 val = **p << 8 | *((*p) + 1); | |
| 311 (*p) += 2; | |
| 312 t = lwasm_expr_term_create_int(val); | |
| 313 lwasm_expr_stack_push(s, t); | |
| 314 lwasm_expr_term_free(t); | |
| 315 return 0; | |
| 316 } | |
| 317 else if (**p == '\'') | |
| 318 { | |
| 319 // single ascii constant | |
| 320 int val; | |
| 321 (*p)++; | |
| 61 | 322 debug_message(3, "Single ascii character constant '%c'", **p); |
| 18 | 323 if (!**p) |
| 324 return -1; | |
| 325 val = **p; | |
| 326 (*p)++; | |
| 327 t = lwasm_expr_term_create_int(val); | |
| 328 lwasm_expr_stack_push(s, t); | |
| 329 lwasm_expr_term_free(t); | |
| 61 | 330 return 0; |
| 18 | 331 } |
| 332 else if (**p == '&') | |
| 333 { | |
| 334 // decimal constant | |
| 335 int val = 0; | |
| 336 | |
| 337 (*p)++; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
338 while (**p && strchr("0123456789", **p)) |
| 18 | 339 { |
| 340 val = val * 10 + (**p - '0'); | |
| 341 (*p)++; | |
| 342 } | |
| 343 t = lwasm_expr_term_create_int(val); | |
| 344 lwasm_expr_stack_push(s, t); | |
| 345 lwasm_expr_term_free(t); | |
| 346 return 0; | |
| 347 } | |
| 348 else if (**p == '%') | |
| 349 { | |
| 350 // binary constant | |
| 351 int val = 0; | |
| 352 | |
| 353 (*p)++; | |
| 354 while (**p == '0' || **p == '1') | |
| 355 { | |
| 356 val = val * 2 + (**p - '0'); | |
| 357 (*p)++; | |
| 358 } | |
| 359 t = lwasm_expr_term_create_int(val); | |
| 360 lwasm_expr_stack_push(s, t); | |
| 361 lwasm_expr_term_free(t); | |
| 362 return 0; | |
| 363 } | |
| 364 else if (**p == '$') | |
| 365 { | |
| 366 // hexadecimal constant | |
| 367 int val = 0, val2; | |
| 368 | |
| 369 (*p)++; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
370 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
|
371 while (**p && strchr("0123456789ABCDEFabcdef", **p)) |
| 18 | 372 { |
| 373 val2 = toupper(**p) - '0'; | |
| 374 if (val2 > 9) | |
| 375 val2 -= 7; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
376 debug_message(3, "Got char: %c (%d)", **p, val2); |
| 18 | 377 val = val * 16 + val2; |
| 378 (*p)++; | |
| 379 } | |
| 380 t = lwasm_expr_term_create_int(val); | |
| 381 lwasm_expr_stack_push(s, t); | |
| 382 lwasm_expr_term_free(t); | |
| 383 return 0; | |
| 384 } | |
|
40
d2cee0c335e7
adjusted symbol rules to accept symbols starting with @ but not @<digit>
lost
parents:
39
diff
changeset
|
385 // 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
|
386 // 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
|
387 else if (**p == '@' && isdigit(*(*p + 1))) |
| 18 | 388 { |
| 389 // octal constant | |
| 390 int val = 0; | |
| 391 | |
| 392 (*p)++; | |
|
47
804d7465e0f9
Implemented ORG and fixed problems with constants using $, &, or @ to specify base
lost
parents:
44
diff
changeset
|
393 while (**p && strchr("01234567", **p)) |
| 18 | 394 { |
| 395 val = val * 8 + (**p - '0'); | |
| 396 (*p)++; | |
| 397 } | |
| 398 t = lwasm_expr_term_create_int(val); | |
| 399 lwasm_expr_stack_push(s, t); | |
| 400 lwasm_expr_term_free(t); | |
| 401 return 0; | |
| 402 } | |
| 403 | |
| 404 // symbol or bare decimal or suffix identified constant here | |
| 405 // all numbers will start with a digit at this point | |
| 406 if (**p < '0' || **p > '9') | |
| 407 { | |
| 408 int l = 0; | |
| 409 char *sb; | |
| 410 | |
| 411 // evaluate a symbol here | |
| 41 | 412 static const char *symchars = "_.$@?abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"; |
| 413 while ((*p)[l] && strchr(symchars, (*p)[l])) | |
| 18 | 414 l++; |
| 415 | |
| 416 if (l == 0) | |
| 417 return -1; | |
| 418 | |
| 419 sb = lwasm_alloc(l + 1); | |
| 420 sb[l] = '\0'; | |
| 421 memcpy(sb, *p, l); | |
| 422 t = lwasm_expr_term_create_sym(sb); | |
| 423 lwasm_expr_stack_push(s, t); | |
| 424 lwasm_expr_term_free(t); | |
| 41 | 425 (*p) += l; |
| 426 debug_message(3, "Symbol: '%s'; (%s)", sb, *p); | |
| 18 | 427 lwasm_free(sb); |
| 428 return 0; | |
| 429 } | |
| 430 | |
| 431 if (!**p) | |
| 432 return -1; | |
| 433 | |
| 434 // evaluate a suffix based constant | |
| 435 { | |
| 436 int decval = 0, binval = 0, hexval = 0, octval = 0; | |
| 437 int valtype = 15; // 1 = bin, 2 = oct, 4 = dec, 8 = hex | |
| 438 int bindone = 0; | |
| 439 int val; | |
| 440 int dval; | |
| 441 | |
| 442 while (1) | |
| 443 { | |
| 444 if (!**p || !strchr("0123456789ABCDEFabcdefqhoQHO", **p)) | |
| 445 { | |
| 446 // we can legally have bin or decimal here | |
| 447 if (bindone) | |
| 448 { | |
| 449 // we just finished a binary value | |
| 450 val = binval; | |
| 451 break; | |
| 452 } | |
| 453 else if (valtype & 4) | |
| 454 { | |
| 455 // otherwise we must be decimal (if we're still allowed one) | |
| 456 val = decval; | |
| 39 | 457 debug_message(3, "End of decimal value"); |
| 18 | 458 break; |
| 459 } | |
| 460 else | |
| 461 { | |
| 462 // bad value | |
| 463 return -1; | |
| 464 } | |
| 465 } | |
| 466 | |
| 467 dval = toupper(**p); | |
| 468 (*p)++; | |
| 469 | |
| 470 if (bindone) | |
| 471 { | |
| 472 // any characters past "B" means it is not binary | |
| 473 bindone = 0; | |
| 474 valtype &= 14; | |
| 475 } | |
| 476 | |
| 477 switch (dval) | |
| 478 { | |
| 479 case 'Q': | |
| 480 case 'O': | |
| 481 if (valtype & 2) | |
| 482 { | |
| 483 val = octval; | |
| 484 valtype = -1; | |
| 485 break; | |
| 486 } | |
| 487 else | |
| 488 { | |
| 489 // not a valid octal value | |
| 490 return -1; | |
| 491 } | |
| 492 /* can't get here */ | |
| 493 | |
| 494 case 'H': | |
| 495 if (valtype & 8) | |
| 496 { | |
| 497 val = hexval; | |
| 498 valtype = -1; | |
| 499 break; | |
| 500 } | |
| 501 else | |
| 502 { | |
| 503 // not a valid hex number | |
| 504 return -1; | |
| 505 } | |
| 506 /* can't get here */ | |
| 507 | |
| 508 case 'B': | |
| 509 // this is a bit of a sticky one since B is a legit hex | |
| 510 // digit so this may or may not be the end of the number | |
| 511 // so we fall through to the digit case | |
| 512 | |
| 513 if (valtype & 1) | |
| 514 { | |
| 515 // could still be binary | |
| 516 bindone = 1; | |
| 517 valtype = 9; // hex and binary | |
| 518 } | |
| 519 /* fall through intentional */ | |
| 520 | |
| 521 default: | |
| 522 // digit | |
| 523 dval -= '0'; | |
| 524 if (dval > 9) | |
| 525 dval -= 7; | |
| 39 | 526 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
|
527 // if (dval > 1) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
528 // valtype &= 14; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
529 // if (dval > 7) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
530 // valtype &= 12; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
531 // if (dval > 9) |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
532 // valtype &= 8; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
533 |
| 18 | 534 if (valtype & 8) |
| 535 { | |
| 536 hexval = hexval * 16 + dval; | |
| 537 } | |
| 538 if (valtype & 4) | |
| 539 { | |
| 540 if (dval > 9) | |
| 541 valtype &= 11; | |
| 542 else | |
| 543 decval = decval * 10 + dval; | |
| 544 } | |
| 545 if (valtype & 2) | |
| 546 { | |
| 547 if (dval > 7) | |
| 548 valtype &= 13; | |
| 549 else | |
| 550 octval = octval * 8 + dval; | |
| 551 } | |
| 552 if (valtype & 1) | |
| 553 { | |
| 554 if (dval > 1) | |
| 555 valtype &= 14; | |
| 556 else | |
| 557 binval = binval * 2 + dval; | |
| 558 } | |
| 559 } | |
| 560 // 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
|
561 if (valtype == -1) |
| 18 | 562 break; |
| 563 // return if no more valid possibilities! | |
| 564 if (valtype == 0) | |
| 565 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
566 val = decval; // in case we fall through |
| 18 | 567 } |
| 568 | |
| 569 // we get here when we have a value to return | |
| 570 t = lwasm_expr_term_create_int(val); | |
| 571 lwasm_expr_stack_push(s, t); | |
| 572 lwasm_expr_term_free(t); | |
| 573 return 0; | |
| 574 } | |
| 575 /* can't get here */ | |
| 576 } | |
| 577 | |
| 578 // parse an expression and push the result onto the stack | |
| 579 // if an operator of lower precedence than the value of "prec" is found, | |
| 580 int lwasm_expr_parse_expr(lwasm_expr_stack_t *s, const char **p, int prec) | |
| 581 { | |
| 582 static const struct operinfo | |
| 583 { | |
| 584 int opernum; | |
| 585 char *operstr; | |
| 586 int operprec; | |
| 587 } operators[] = | |
| 588 { | |
| 589 { LWASM_OPER_PLUS, "+", 100 }, | |
| 590 { LWASM_OPER_MINUS, "-", 100 }, | |
| 591 { LWASM_OPER_TIMES, "*", 150 }, | |
| 592 { LWASM_OPER_DIVIDE, "/", 150 }, | |
| 593 { LWASM_OPER_MOD, "%", 150 }, | |
| 594 { LWASM_OPER_INTDIV, "\\", 150 }, | |
| 595 | |
| 596 { LWASM_OPER_NONE, "", 0 } | |
| 597 }; | |
| 598 int opern, i; | |
| 599 lwasm_expr_term_t *operterm; | |
| 600 | |
| 601 // return if we are at the end of the expression or a subexpression | |
| 61 | 602 if (!**p || isspace(**p) || **p == ')' || **p == ',' || **p == ']') |
| 18 | 603 return 0; |
| 604 | |
| 605 if (lwasm_expr_parse_term(s, p) < 0) | |
| 606 return -1; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
607 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
608 eval_next: |
| 61 | 609 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
|
610 return 0; |
| 18 | 611 |
| 612 // expecting an operator here | |
| 613 for (opern = 0; operators[opern].opernum != LWASM_OPER_NONE; opern++) | |
| 614 { | |
| 615 for (i = 0; (*p)[i] && operators[opern].operstr[i] && (*p[i] == operators[opern].operstr[i]); i++) | |
| 616 /* do nothing */ ; | |
| 617 if (operators[opern].operstr[i] == '\0') | |
| 618 break; | |
| 619 } | |
| 620 if (operators[opern].opernum == LWASM_OPER_NONE) | |
| 621 { | |
| 622 // unrecognized operator | |
| 623 return -1; | |
| 624 } | |
| 625 | |
| 626 // the operator number in question is in opern; i is the length of the | |
| 627 // operator string | |
| 628 | |
| 629 // logic: | |
| 630 // if the precedence of this operation is <= to the "prec" flag, | |
| 631 // we simply return without advancing the input pointer; the operator | |
| 632 // will be evaluated again in the enclosing function call | |
| 633 if (operators[opern].operprec <= prec) | |
| 634 return 0; | |
| 635 | |
| 636 // logic: | |
| 637 // we have a higher precedence operator here so we will advance the | |
| 638 // input pointer to the next term and let the expression evaluator | |
| 639 // loose on it after which time we will push our operator onto the | |
| 640 // stack and then go on with the expression evaluation | |
| 641 (*p) += i; // advance input pointer | |
| 642 | |
| 643 // evaluate next expression(s) of higher precedence | |
| 644 if (lwasm_expr_parse_expr(s, p, operators[opern].operprec) < 0) | |
| 645 return -1; | |
| 646 | |
| 647 operterm = lwasm_expr_term_create_oper(operators[opern].opernum); | |
| 648 lwasm_expr_stack_push(s, operterm); | |
| 649 lwasm_expr_term_free(operterm); | |
| 650 | |
| 651 // return if we are at the end of the expression or a subexpression | |
| 652 if (!**p || isspace(**p) || **p == ')') | |
| 653 return 0; | |
| 654 | |
| 655 // continue evaluating | |
| 656 goto eval_next; | |
| 657 } | |
| 658 | |
| 659 /* | |
| 660 actually evaluate an expression | |
| 661 | |
| 662 This happens in two stages. The first stage merely parses the expression into | |
| 663 a lwasm_expr_stack_t * which is then evaluated as much as possible before the | |
| 664 result is returned. | |
| 665 | |
| 666 Returns NULL on a parse error or otherwise invalid expression. *outp will | |
| 667 contain the pointer to the next character after the expression if and only | |
| 668 if there is no error. In the case of an error, *outp is undefined. | |
| 669 */ | |
| 76 | 670 lwasm_expr_stack_t *lwasm_expr_eval(const char *inp, const char **outp, lwasm_expr_stack_t *(*sfunc)(char *sym, void *state), void *state) |
| 18 | 671 { |
| 672 lwasm_expr_stack_t *s; | |
| 673 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
|
674 int rval; |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
675 |
| 18 | 676 // actually parse the expression |
| 677 p = inp; | |
| 678 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
|
679 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
680 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
|
681 if (rval < 0) |
| 18 | 682 goto cleanup_error; |
| 683 | |
| 684 // save end of expression | |
| 685 if (outp) | |
| 686 (*outp) = p; | |
| 687 | |
| 688 // 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
|
689 if (lwasm_expr_reval(s, sfunc, state) < 0) |
| 18 | 690 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
|
691 |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
692 if (lwasm_expr_is_constant(s)) |
| 39 | 693 debug_message(3, "Constant expression evaluates to: %d", lwasm_expr_get_value(s)); |
| 18 | 694 |
| 695 return s; | |
| 696 | |
| 697 cleanup_error: | |
| 698 lwasm_expr_stack_free(s); | |
| 699 return NULL; | |
| 700 } | |
| 701 | |
| 702 /* | |
| 703 take an expression stack s and scan for operations that can be completed | |
| 704 | |
| 705 return -1 on error, 0 on no error | |
| 706 | |
| 707 possible errors are: division by zero or unknown operator | |
| 708 | |
| 709 theory of operation: | |
| 710 | |
| 711 scan the stack for an operator which has two constants preceding it (binary) | |
| 712 or 1 constant preceding it (unary) and if found, perform the calculation | |
| 713 and replace the operator and its operands with the result | |
| 714 | |
| 715 repeat the scan until no futher simplications are found or if there are no | |
| 716 further operators or only a single term remains | |
| 717 | |
| 718 */ | |
| 76 | 719 int lwasm_expr_reval(lwasm_expr_stack_t *s, lwasm_expr_stack_t *(*sfunc)(char *sym, void *state), void *state) |
| 18 | 720 { |
| 76 | 721 lwasm_expr_stack_node_t *n, *n2; |
| 722 lwasm_expr_stack_t *ss; | |
| 723 int c; | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
724 |
| 76 | 725 next_iter_sym: |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
726 // resolve symbols |
| 76 | 727 // symbols that do not resolve to an expression are left alone |
| 728 for (c = 0, n = s -> head; n; n = n -> next) | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
729 { |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
730 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
|
731 { |
| 76 | 732 ss = sfunc(n -> term -> symbol, state); |
| 733 if (ss) | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
734 { |
| 76 | 735 c++; |
| 736 // splice in the result stack | |
| 737 if (n -> prev) | |
| 738 { | |
| 739 n -> prev -> next = ss -> head; | |
| 740 } | |
| 741 else | |
| 742 { | |
| 743 s -> head = ss -> head; | |
| 744 } | |
| 745 ss -> head -> prev = n -> prev; | |
| 746 ss -> tail -> next = n -> next; | |
| 747 if (n -> next) | |
| 748 { | |
| 749 n -> next -> prev = ss -> tail; | |
| 750 } | |
| 751 else | |
| 752 { | |
| 753 s -> tail = ss -> tail; | |
| 754 } | |
| 755 lwasm_expr_term_free(n -> term); | |
| 756 lwasm_free(n); | |
| 757 n = ss -> tail; | |
| 758 | |
| 759 ss -> head = NULL; | |
| 760 ss -> tail = NULL; | |
| 761 lwasm_expr_stack_free(ss); | |
|
37
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
762 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
763 } |
|
538e15927776
Added symbol handling to expression subsystem; adpated instruction handlers to the new scheme; misc fixes
lost
parents:
23
diff
changeset
|
764 } |
| 76 | 765 if (c) |
| 766 goto next_iter_sym; | |
| 18 | 767 |
| 768 next_iter: | |
| 769 // a single term | |
| 770 if (s -> head == s -> tail) | |
| 771 return 0; | |
| 772 | |
| 773 // search for an operator | |
| 774 for (n = s -> head; n; n = n -> next) | |
| 775 { | |
| 776 if (n -> term -> term_type == LWASM_TERM_OPER) | |
| 777 { | |
| 778 if (n -> term -> value == LWASM_OPER_NEG | |
| 779 || n -> term -> value == LWASM_OPER_COM | |
| 780 ) | |
| 781 { | |
| 782 // unary operator | |
| 783 if (n -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 784 { | |
| 785 // a unary operator we can resolve | |
| 786 // we do the op then remove the term "n" is pointing at | |
| 787 if (n -> term -> value == LWASM_OPER_NEG) | |
| 788 { | |
| 789 n -> prev -> term -> value = -(n -> prev -> term -> value); | |
| 790 } | |
| 791 else if (n -> term -> value == LWASM_OPER_COM) | |
| 792 { | |
| 793 n -> prev -> term -> value = ~(n -> prev -> term -> value); | |
| 794 } | |
| 795 n -> prev -> next = n -> next; | |
| 796 if (n -> next) | |
| 797 n -> next -> prev = n -> prev; | |
| 798 else | |
| 799 s -> tail = n -> prev; | |
| 800 | |
| 801 lwasm_expr_term_free(n -> term); | |
| 802 lwasm_free(n); | |
| 803 break; | |
| 804 } | |
| 805 } | |
| 806 else | |
| 807 { | |
| 808 // binary operator | |
| 809 if (n -> prev && n -> prev -> prev && n -> prev -> term -> term_type == LWASM_TERM_INT && n -> prev -> prev -> term -> term_type == LWASM_TERM_INT) | |
| 810 { | |
| 811 // a binary operator we can resolve | |
| 812 switch (n -> term -> value) | |
| 813 { | |
| 814 case LWASM_OPER_PLUS: | |
| 815 n -> prev -> prev -> term -> value += n -> prev -> term -> value; | |
| 816 break; | |
| 817 | |
| 818 case LWASM_OPER_MINUS: | |
| 819 n -> prev -> prev -> term -> value -= n -> prev -> term -> value; | |
| 820 break; | |
| 821 | |
| 822 case LWASM_OPER_TIMES: | |
| 823 n -> prev -> prev -> term -> value *= n -> prev -> term -> value; | |
| 824 break; | |
| 825 | |
| 826 case LWASM_OPER_DIVIDE: | |
| 827 if (n -> prev -> term -> value == 0) | |
| 828 return -1; | |
| 829 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 830 break; | |
| 831 | |
| 832 case LWASM_OPER_MOD: | |
| 833 if (n -> prev -> term -> value == 0) | |
| 834 return -1; | |
| 835 n -> prev -> prev -> term -> value %= n -> prev -> term -> value; | |
| 836 break; | |
| 837 | |
| 838 case LWASM_OPER_INTDIV: | |
| 839 if (n -> prev -> term -> value == 0) | |
| 840 return -1; | |
| 841 n -> prev -> prev -> term -> value /= n -> prev -> term -> value; | |
| 842 break; | |
| 843 | |
| 844 case LWASM_OPER_BWAND: | |
| 845 n -> prev -> prev -> term -> value &= n -> prev -> term -> value; | |
| 846 break; | |
| 847 | |
| 848 case LWASM_OPER_BWOR: | |
| 849 n -> prev -> prev -> term -> value |= n -> prev -> term -> value; | |
| 850 break; | |
| 851 | |
| 852 case LWASM_OPER_BWXOR: | |
| 853 n -> prev -> prev -> term -> value ^= n -> prev -> term -> value; | |
| 854 break; | |
| 855 | |
| 856 case LWASM_OPER_AND: | |
| 857 n -> prev -> prev -> term -> value = (n -> prev -> term -> value && n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 858 break; | |
| 859 | |
| 860 case LWASM_OPER_OR: | |
| 861 n -> prev -> prev -> term -> value = (n -> prev -> term -> value || n -> prev -> prev -> term -> value) ? 1 : 0; | |
| 862 break; | |
| 863 | |
| 864 default: | |
| 865 // return error if unknown operator! | |
| 866 return -1; | |
| 867 } | |
| 868 | |
| 869 // now remove the two unneeded entries from the stack | |
| 870 n -> prev -> prev -> next = n -> next; | |
| 871 if (n -> next) | |
| 872 n -> next -> prev = n -> prev -> prev; | |
| 873 else | |
| 874 s -> tail = n -> prev -> prev; | |
| 875 | |
| 876 lwasm_expr_term_free(n -> term); | |
| 877 lwasm_expr_term_free(n -> prev -> term); | |
| 878 lwasm_free(n -> prev); | |
| 879 lwasm_free(n); | |
| 880 break; | |
| 881 } | |
| 882 } | |
| 883 } | |
| 884 } | |
| 885 // note for the terminally confused about dynamic memory and pointers: | |
| 886 // n will not be NULL even after the lwasm_free calls above so | |
| 887 // this test will still work (n will be a dangling pointer) | |
| 888 // (n will only be NULL if we didn't find any operators to simplify) | |
| 889 if (n) | |
| 890 goto next_iter; | |
| 891 | |
| 892 return 0; | |
| 893 } |