An emulator, assembler, and disassembler for the Sega Game Gear
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 

561 lines
15 KiB

  1. /* Copyright (C) 2014-2015 Ben Kurtovic <ben.kurtovic@gmail.com>
  2. Released under the terms of the MIT License. See LICENSE for details. */
  3. #include <limits.h>
  4. #include <stdlib.h>
  5. #include <string.h>
  6. #include "parse_util.h"
  7. #include "directives.h"
  8. #include "../logging.h"
  9. #include "../util.h"
  10. #define MAX_REGION_SIZE 32
  11. #define LCASE(c) ((c >= 'A' && c <= 'Z') ? (c + 'a' - 'A') : c)
  12. #define DIRECTIVE_PARSE_FUNC(name, type) \
  13. bool dparse_##name(type *result, const ASMLine *line, const char *directive)
  14. /*
  15. All public functions in this file follow the same return conventions:
  16. - Return true on success and false on failure.
  17. - *result is only modified on success.
  18. */
  19. /*
  20. Adjust *ap_info for an indirect argument, like (hl) or (ix+*).
  21. ap_info->size must be > 2 to begin with, and will always be > 0 on success.
  22. *ap_info is not modified on failure.
  23. */
  24. static bool adjust_for_indirection(ASMArgParseInfo *ap_info)
  25. {
  26. const char *arg = ap_info->arg;
  27. ssize_t size = ap_info->size;
  28. if (arg[0] != '(' || arg[size - 1] != ')')
  29. return false;
  30. arg++;
  31. size -= 2;
  32. if (arg[0] == ' ') {
  33. arg++;
  34. if (--size <= 0)
  35. return false;
  36. }
  37. if (arg[size - 1] == ' ') {
  38. if (--size <= 0)
  39. return false;
  40. }
  41. ap_info->arg = arg;
  42. ap_info->size = size;
  43. return true;
  44. }
  45. /*
  46. Read in a boolean value and store it in *result.
  47. */
  48. bool parse_bool(bool *result, const char *arg, ssize_t size)
  49. {
  50. switch (size) {
  51. case 1: // 0, 1
  52. if (*arg == '0' || *arg == '1')
  53. return (*result = *arg - '0'), true;
  54. return false;
  55. case 2: // on
  56. if (!strncmp(arg, "on", 2))
  57. return (*result = true), true;
  58. return false;
  59. case 3: // off
  60. if (!strncmp(arg, "off", 3))
  61. return (*result = false), true;
  62. return false;
  63. case 4: // true
  64. if (!strncmp(arg, "true", 4))
  65. return (*result = true), true;
  66. return false;
  67. case 5: // false
  68. if (!strncmp(arg, "false", 5))
  69. return (*result = false), true;
  70. return false;
  71. }
  72. return false;
  73. }
  74. /*
  75. Read in a 32-bit integer and store it in *result.
  76. */
  77. bool parse_uint32_t(uint32_t *result, const char *arg, ssize_t size)
  78. {
  79. if (size <= 0)
  80. return false;
  81. const char *end = arg + size;
  82. uint64_t value = 0;
  83. if (*arg == '$') {
  84. arg++;
  85. if (arg == end)
  86. return false;
  87. while (arg < end) {
  88. if (*arg >= '0' && *arg <= '9')
  89. value = (value * 0x10) + (*arg - '0');
  90. else if (*arg >= 'a' && *arg <= 'f')
  91. value = (value * 0x10) + 0xA + (*arg - 'a');
  92. else
  93. return false;
  94. if (value > UINT32_MAX)
  95. return false;
  96. arg++;
  97. }
  98. }
  99. else {
  100. while (arg < end) {
  101. if (*arg < '0' || *arg > '9')
  102. return false;
  103. value = (value * 10) + (*arg - '0');
  104. if (value > UINT32_MAX)
  105. return false;
  106. arg++;
  107. }
  108. }
  109. *result = value;
  110. return true;
  111. }
  112. /*
  113. Read in a string, possibly with escape sequences, and store it in *result.
  114. *length is also updated to the size of the string, which is *not*
  115. null-terminated. *result must be free()'d when finished.
  116. */
  117. bool parse_string(char **result, size_t *length, const char *arg, ssize_t size)
  118. {
  119. if (size < 2 || arg[0] != '"' || arg[size - 1] != '"')
  120. return false;
  121. ssize_t i, slashes = 0;
  122. for (i = 1; i < size; i++) {
  123. if (arg[i] == '"' && (slashes % 2) == 0)
  124. break;
  125. // TODO: parse escape codes here
  126. if (arg[i] == '\\')
  127. slashes++;
  128. else
  129. slashes = 0;
  130. }
  131. if (i != size - 1) // Junk present after closing quote
  132. return false;
  133. *length = size - 2;
  134. *result = malloc(sizeof(char) * (*length));
  135. if (!*result)
  136. OUT_OF_MEMORY()
  137. memcpy(*result, arg + 1, *length);
  138. return true;
  139. }
  140. /*
  141. Read in a space-separated sequence of bytes and store it in *result.
  142. *length is also updated to the number of bytes in the array. *result must
  143. be free()'d when finished.
  144. */
  145. bool parse_bytes(uint8_t **result, size_t *length, const char *arg, ssize_t size)
  146. {
  147. if (size <= 0)
  148. return false;
  149. const char *end = arg + size;
  150. uint8_t *bytes = NULL;
  151. size_t nbytes = 0;
  152. while (arg < end) {
  153. const char *start = arg;
  154. while (arg != end && *arg != ' ' && *arg != ',')
  155. arg++;
  156. uint32_t temp;
  157. if (!parse_uint32_t(&temp, start, arg - start) || temp > UINT8_MAX) {
  158. free(bytes);
  159. return false;
  160. }
  161. nbytes++;
  162. bytes = realloc(bytes, sizeof(uint8_t) * nbytes);
  163. if (!bytes)
  164. OUT_OF_MEMORY()
  165. bytes[nbytes - 1] = temp;
  166. if (arg < end - 1 && *arg == ',' && *(arg + 1) == ' ')
  167. arg += 2;
  168. else if (arg++ >= end)
  169. break;
  170. }
  171. *result = bytes;
  172. *length = nbytes;
  173. return true;
  174. }
  175. /*
  176. Read in a register argument and store it in *result.
  177. */
  178. bool argparse_register(ASMArgRegister *result, ASMArgParseInfo ai)
  179. {
  180. if (ai.size < 1 || ai.size > 3)
  181. return false;
  182. char buf[3] = {'\0'};
  183. switch (ai.size) {
  184. case 3: buf[2] = LCASE(ai.arg[2]);
  185. case 2: buf[1] = LCASE(ai.arg[1]);
  186. case 1: buf[0] = LCASE(ai.arg[0]);
  187. }
  188. switch (ai.size) {
  189. case 1:
  190. switch (buf[0]) {
  191. case 'a': return (*result = REG_A), true;
  192. case 'f': return (*result = REG_F), true;
  193. case 'b': return (*result = REG_B), true;
  194. case 'c': return (*result = REG_C), true;
  195. case 'd': return (*result = REG_D), true;
  196. case 'e': return (*result = REG_E), true;
  197. case 'h': return (*result = REG_H), true;
  198. case 'l': return (*result = REG_L), true;
  199. case 'i': return (*result = REG_I), true;
  200. case 'r': return (*result = REG_R), true;
  201. }
  202. return false;
  203. case 2:
  204. switch ((buf[0] << 8) + buf[1]) {
  205. case 0x6166: return (*result = REG_AF), true;
  206. case 0x6263: return (*result = REG_BC), true;
  207. case 0x6465: return (*result = REG_DE), true;
  208. case 0x686C: return (*result = REG_HL), true;
  209. case 0x6978: return (*result = REG_IX), true;
  210. case 0x6979: return (*result = REG_IY), true;
  211. case 0x7063: return (*result = REG_PC), true;
  212. case 0x7370: return (*result = REG_SP), true;
  213. }
  214. return false;
  215. case 3:
  216. switch ((buf[0] << 16) + (buf[1] << 8) + buf[2]) {
  217. case 0x616627: return (*result = REG_AF_), true;
  218. case 0x697868: return (*result = REG_IXH), true;
  219. case 0x69786C: return (*result = REG_IXL), true;
  220. case 0x697968: return (*result = REG_IYH), true;
  221. case 0x69796C: return (*result = REG_IYL), true;
  222. }
  223. return false;
  224. }
  225. return false;
  226. }
  227. /*
  228. Read in a condition argument and store it in *result.
  229. */
  230. bool argparse_condition(ASMArgCondition *result, ASMArgParseInfo ai)
  231. {
  232. if (ai.size < 1 || ai.size > 2)
  233. return false;
  234. char buf[2] = {'\0'};
  235. switch (ai.size) {
  236. case 2: buf[1] = LCASE(ai.arg[1]);
  237. case 1: buf[0] = LCASE(ai.arg[0]);
  238. }
  239. switch (ai.size) {
  240. case 1:
  241. switch (buf[0]) {
  242. case 'n': return (*result = COND_N), true;
  243. case 'c': return (*result = COND_C), true;
  244. case 'p': return (*result = COND_P), true;
  245. case 'm': return (*result = COND_M), true;
  246. }
  247. return false;
  248. case 2:
  249. switch ((buf[0] << 8) + buf[1]) {
  250. case 0x6E7A: return (*result = COND_NZ), true;
  251. case 0x6E63: return (*result = COND_NC), true;
  252. case 0x706F: return (*result = COND_PO), true;
  253. case 0x7065: return (*result = COND_PE), true;
  254. }
  255. return false;
  256. }
  257. return false;
  258. }
  259. /*
  260. Read in an immediate argument and store it in *result.
  261. */
  262. bool argparse_immediate(ASMArgImmediate *result, ASMArgParseInfo ai)
  263. {
  264. if (ai.size <= 0)
  265. return false;
  266. // TODO: lookup here for definition table
  267. bool negative = false;
  268. ssize_t i = 0;
  269. while (ai.arg[i] == '-' || ai.arg[i] == '+' || ai.arg[i] == ' ') {
  270. if (ai.arg[i] == '-')
  271. negative = !negative;
  272. if (++i >= ai.size)
  273. return false;
  274. }
  275. uint32_t uval;
  276. if (!parse_uint32_t(&uval, ai.arg + i, ai.size - i) || uval > UINT16_MAX)
  277. return false;
  278. int32_t sval = negative ? -uval : uval;
  279. if (sval < INT16_MIN)
  280. return false;
  281. result->uval = uval;
  282. result->sval = sval;
  283. result->mask = 0;
  284. if (sval < 0) {
  285. if (sval >= INT8_MIN)
  286. result->mask |= IMM_S8;
  287. if (sval >= INT8_MIN + 2)
  288. result->mask |= IMM_REL;
  289. } else {
  290. result->mask = IMM_U16;
  291. if (uval <= UINT8_MAX)
  292. result->mask |= IMM_U8;
  293. if (uval <= INT8_MAX)
  294. result->mask |= IMM_S8;
  295. if (uval <= INT8_MAX + 2)
  296. result->mask |= IMM_REL;
  297. if (uval <= 7)
  298. result->mask |= IMM_BIT;
  299. if (!(uval & ~0x38))
  300. result->mask |= IMM_RST;
  301. if (uval <= 2)
  302. result->mask |= IMM_IM;
  303. }
  304. return true;
  305. }
  306. /*
  307. Read in an indirect argument and store it in *result.
  308. */
  309. bool argparse_indirect(ASMArgIndirect *result, ASMArgParseInfo ai)
  310. {
  311. if (ai.size < 3 || !adjust_for_indirection(&ai))
  312. return false;
  313. ASMArgRegister reg;
  314. ASMArgImmediate imm;
  315. if (argparse_register(&reg, ai)) {
  316. if (reg == REG_BC || reg == REG_DE || reg == REG_HL) {
  317. result->type = AT_REGISTER;
  318. result->addr.reg = reg;
  319. return true;
  320. }
  321. } else if (argparse_immediate(&imm, ai)) {
  322. if (imm.mask & IMM_U16) {
  323. result->type = AT_IMMEDIATE;
  324. result->addr.imm = imm;
  325. return true;
  326. }
  327. }
  328. return false;
  329. }
  330. /*
  331. Read in an indexed argument and store it in *result.
  332. */
  333. bool argparse_indexed(ASMArgIndexed *result, ASMArgParseInfo ai)
  334. {
  335. if (ai.size < 4 || !adjust_for_indirection(&ai) || ai.size < 2)
  336. return false;
  337. ASMArgRegister reg;
  338. if (ai.arg[0] != 'i')
  339. return false;
  340. if (ai.arg[1] == 'x')
  341. reg = REG_IX;
  342. else if (ai.arg[1] == 'y')
  343. reg = REG_IY;
  344. else
  345. return false;
  346. ai.arg += 2;
  347. ai.size -= 2;
  348. if (ai.size > 0 && ai.arg[0] == ' ') {
  349. ai.arg++;
  350. ai.size--;
  351. }
  352. if (ai.size > 0) {
  353. ASMArgImmediate imm;
  354. if (!argparse_immediate(&imm, ai) || !(imm.mask & IMM_S8))
  355. return false;
  356. result->offset = imm.sval;
  357. } else {
  358. result->offset = 0;
  359. }
  360. result->reg = reg;
  361. return true;
  362. }
  363. /*
  364. Read in a label argument and store it in *result.
  365. */
  366. bool argparse_label(ASMArgLabel *result, ASMArgParseInfo ai)
  367. {
  368. if (ai.size <= 0 || ai.size >= MAX_SYMBOL_SIZE)
  369. return false;
  370. // TODO: check for deftable
  371. for (const char *i = ai.arg; i < ai.arg + ai.size; i++) {
  372. char c = *i;
  373. if (!((c >= 'a' && c <= 'z') || c == '_' || c == '.' ||
  374. (i != ai.arg && c >= '0' && c <= '9')))
  375. return false;
  376. }
  377. strncpy(result->text, ai.arg, ai.size);
  378. result->text[ai.size] = '\0';
  379. return true;
  380. }
  381. /*
  382. Read in a boolean argument from the given line and store it in *result.
  383. */
  384. DIRECTIVE_PARSE_FUNC(bool, bool)
  385. {
  386. size_t offset = DIRECTIVE_OFFSET(line, directive) + 1;
  387. return parse_bool(result, line->data + offset, line->length - offset);
  388. }
  389. /*
  390. Read in a 32-bit int argument from the given line and store it in *result.
  391. */
  392. DIRECTIVE_PARSE_FUNC(uint32_t, uint32_t)
  393. {
  394. size_t offset = DIRECTIVE_OFFSET(line, directive) + 1;
  395. return parse_uint32_t(result, line->data + offset, line->length - offset);
  396. }
  397. /*
  398. Read in a 16-bit int argument from the given line and store it in *result.
  399. */
  400. DIRECTIVE_PARSE_FUNC(uint16_t, uint16_t)
  401. {
  402. uint32_t value;
  403. if (dparse_uint32_t(&value, line, directive) && value <= UINT16_MAX)
  404. return (*result = value), true;
  405. return false;
  406. }
  407. /*
  408. Read in an 8-bit int argument from the given line and store it in *result.
  409. */
  410. DIRECTIVE_PARSE_FUNC(uint8_t, uint8_t)
  411. {
  412. uint32_t value;
  413. if (dparse_uint32_t(&value, line, directive) && value <= UINT8_MAX)
  414. return (*result = value), true;
  415. return false;
  416. }
  417. /*
  418. Parse a ROM size string in an ASMLine and store it in *result.
  419. */
  420. DIRECTIVE_PARSE_FUNC(rom_size, uint32_t)
  421. {
  422. const char *arg = line->data + DIRECTIVE_OFFSET(line, directive) + 1;
  423. const char *end = line->data + line->length - 1;
  424. if (end - arg < 5)
  425. return false;
  426. if (*(arg++) != '"' || *(end--) != '"')
  427. return false;
  428. if (*end != 'B' && *end != 'b')
  429. return false;
  430. end--;
  431. uint32_t factor;
  432. if (*end == 'K' || *end == 'k')
  433. factor = 1 << 10;
  434. else if (*end == 'M' || *end == 'm')
  435. factor = 1 << 20;
  436. else
  437. return false;
  438. end--;
  439. if (*end != ' ')
  440. return false;
  441. uint32_t value = 0;
  442. while (arg < end) {
  443. if (*arg < '0' || *arg > '9')
  444. return false;
  445. value = (value * 10) + (*arg - '0');
  446. if (value > UINT16_MAX)
  447. return false;
  448. arg++;
  449. }
  450. *result = value * factor;
  451. return true;
  452. }
  453. /*
  454. Parse a region code string in an ASMLine and store it in *result.
  455. */
  456. DIRECTIVE_PARSE_FUNC(region_string, uint8_t)
  457. {
  458. char buffer[MAX_REGION_SIZE];
  459. size_t offset = DIRECTIVE_OFFSET(line, directive) + 1;
  460. const char *arg = line->data + offset;
  461. ssize_t len = line->length - offset;
  462. if (len <= 2 || len >= MAX_REGION_SIZE + 2) // Account for double quotes
  463. return false;
  464. if (arg[0] != '"' || arg[len - 1] != '"')
  465. return false;
  466. strncpy(buffer, arg + 1, len - 2);
  467. buffer[len - 2] = '\0';
  468. uint8_t code = region_string_to_code(buffer);
  469. if (code)
  470. return (*result = code), true;
  471. return false;
  472. }
  473. /*
  474. Parse a size code in an ASMLine and store it in *result.
  475. */
  476. DIRECTIVE_PARSE_FUNC(size_code, uint8_t)
  477. {
  478. uint32_t bytes;
  479. if (!dparse_uint32_t(&bytes, line, directive)) {
  480. if (!dparse_rom_size(&bytes, line, directive))
  481. return false;
  482. }
  483. uint8_t code = size_bytes_to_code(bytes);
  484. if (code != INVALID_SIZE_CODE)
  485. return (*result = code), true;
  486. return false;
  487. }