Compiler.c 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459
  1. #include <setjmp.h>
  2. #include <stdarg.h>
  3. #include <stdio.h>
  4. #include "Compiler.h"
  5. #include "FunctionMap.h"
  6. #include "Operation.h"
  7. #include "StringIntMap.h"
  8. #include "Tokenizer.h"
  9. #define ERROR_LENGTH 256
  10. #define RETURN_BUFFER 16
  11. static jmp_buf errorJump;
  12. static char error[ERROR_LENGTH] = {'\0'};
  13. static ByteCode* code;
  14. static int16 line = 1;
  15. static int varIndex = 0;
  16. static StringIntMap vars[2];
  17. static FunctionMap functions;
  18. static int returns[RETURN_BUFFER];
  19. static int returnIndex = 0;
  20. static int returnState = 0;
  21. static void cError(const char* format, ...) {
  22. va_list args;
  23. va_start(args, format);
  24. vsnprintf(error, ERROR_LENGTH, format, args);
  25. va_end(args);
  26. longjmp(errorJump, 0);
  27. }
  28. static int cAddVar(const char* var) {
  29. int index = vars[varIndex].entries;
  30. simAdd(vars + varIndex, var, &index);
  31. return index;
  32. }
  33. static void cUnexpectedToken(Token t) {
  34. cError("unexpected token on line %d: %s", line, tGetTokenName(t));
  35. }
  36. static void cAddOperation(Operation token) {
  37. unsigned char c = token;
  38. bcAddBytes(code, &c, 1);
  39. }
  40. static int cReserveInt() {
  41. return bcReserveBytes(code, sizeof(int));
  42. }
  43. static void cSetInt(int p, int i) {
  44. bcSetBytes(code, p, &i, sizeof(int));
  45. }
  46. static void cAddInt(int i) {
  47. bcAddBytes(code, &i, sizeof(int));
  48. }
  49. static void cAddInt16(int16 i) {
  50. bcAddBytes(code, &i, sizeof(int16));
  51. }
  52. static void cAddFloat(float f) {
  53. bcAddBytes(code, &f, sizeof(float));
  54. }
  55. static int cAddPush(int offset) {
  56. cAddOperation(OP_PUSH_VARS);
  57. int p = cReserveInt();
  58. cAddInt(offset);
  59. return p;
  60. }
  61. static void cAddPop(int p, int vars) {
  62. cAddOperation(OP_POP_VARS);
  63. cAddInt(vars);
  64. cSetInt(p, vars);
  65. }
  66. static Token cReadTokenAndLine() {
  67. Token t = tReadToken();
  68. if(tReadInt16(&line)) {
  69. return t;
  70. }
  71. return T_END;
  72. }
  73. static void cConsumeToken(Token wanted) {
  74. Token t = cReadTokenAndLine();
  75. if(wanted != t) {
  76. cError("unexpected token on line %d: expected '%s' got '%s'", line, tGetTokenName(wanted), tGetTokenName(t));
  77. }
  78. }
  79. static bool cConsumeTokenIf(Token t) {
  80. if(tPeekToken() == t) {
  81. cReadTokenAndLine();
  82. return true;
  83. }
  84. return false;
  85. }
  86. static void cConstantInt() {
  87. int value;
  88. if(!tReadInt(&value)) {
  89. cError("int token without an int on line %d", line);
  90. }
  91. cAddOperation(OP_PUSH_INT);
  92. cAddInt(value);
  93. }
  94. static void cConstantFloat() {
  95. float value;
  96. if(!tReadFloat(&value)) {
  97. cError("float token without a float on line %d", line);
  98. }
  99. cAddOperation(OP_PUSH_FLOAT);
  100. cAddFloat(value);
  101. }
  102. static const char* cReadString() {
  103. const char* literal = tReadString();
  104. if(literal == NULL) {
  105. cError("literal without string on line %d", line);
  106. }
  107. return literal;
  108. }
  109. static void cGetVar(const char* var) {
  110. cAddOperation(OP_GET);
  111. cAddInt(cAddVar(var));
  112. }
  113. static void cExpression();
  114. static int cCallFunctionArguments() {
  115. int arguments = 0;
  116. while(!cConsumeTokenIf(T_CLOSE_BRACKET)) {
  117. arguments++;
  118. cExpression();
  119. if(cConsumeTokenIf(T_COMMA) && tPeekToken() == T_CLOSE_BRACKET) {
  120. cUnexpectedToken(tPeekToken());
  121. }
  122. }
  123. return arguments;
  124. }
  125. static void cCallFunction(const char* literal, bool noReturn) {
  126. cAddOperation(OP_PUSH_INT);
  127. cAddInt(0);
  128. int arguments = cCallFunctionArguments();
  129. Function* f = fmSearch(&functions, literal, arguments);
  130. cAddOperation(OP_GOSUB);
  131. if(f == NULL) {
  132. fmEnqueue(&functions, literal, arguments, line, cReserveInt(), noReturn);
  133. cAddInt(arguments);
  134. cAddOperation(OP_NOTHING);
  135. } else {
  136. if(!noReturn && !f->returns) {
  137. cError("function '%s' needs a return value on line %d", f->name, line);
  138. }
  139. cAddInt(f->address);
  140. cAddInt(arguments);
  141. if(f->returns && noReturn) {
  142. cAddOperation(OP_POP);
  143. }
  144. }
  145. }
  146. static void cLiteral() {
  147. const char* literal = cReadString();
  148. if(cConsumeTokenIf(T_OPEN_BRACKET)) {
  149. cCallFunction(literal, false);
  150. } else {
  151. cGetVar(literal);
  152. }
  153. }
  154. static void cPrimary() {
  155. Token t = cReadTokenAndLine();
  156. switch(t) {
  157. case T_INT: cConstantInt(); break;
  158. case T_FLOAT: cConstantFloat(); break;
  159. case T_NULL: cAddOperation(OP_PUSH_NULL); break;
  160. case T_TRUE: cAddOperation(OP_PUSH_TRUE); break;
  161. case T_FALSE: cAddOperation(OP_PUSH_FALSE); break;
  162. case T_OPEN_BRACKET:
  163. cExpression();
  164. cConsumeToken(T_CLOSE_BRACKET);
  165. break;
  166. case T_LITERAL: cLiteral(); break;
  167. default: cUnexpectedToken(t); break;
  168. }
  169. }
  170. static void cMul() {
  171. cPrimary();
  172. while(cConsumeTokenIf(T_MUL)) {
  173. cPrimary();
  174. cAddOperation(OP_MUL);
  175. }
  176. }
  177. static void cAdd() {
  178. cMul();
  179. while(true) {
  180. if(cConsumeTokenIf(T_ADD)) {
  181. cMul();
  182. cAddOperation(OP_ADD);
  183. } else if(cConsumeTokenIf(T_SUB)) {
  184. cMul();
  185. cAddOperation(OP_SUB);
  186. } else {
  187. break;
  188. }
  189. }
  190. }
  191. static void cComparison() {
  192. cAdd();
  193. while(true) {
  194. if(cConsumeTokenIf(T_LESS)) {
  195. cAdd();
  196. cAddOperation(OP_LESS);
  197. } else if(cConsumeTokenIf(T_LESS_EQUAL)) {
  198. cAdd();
  199. cAddOperation(OP_GREATER);
  200. cAddOperation(OP_NOT);
  201. } else if(cConsumeTokenIf(T_GREATER)) {
  202. cAdd();
  203. cAddOperation(OP_GREATER);
  204. } else if(cConsumeTokenIf(T_GREATER_EQUAL)) {
  205. cAdd();
  206. cAddOperation(OP_LESS);
  207. cAddOperation(OP_NOT);
  208. } else if(cConsumeTokenIf(T_EQUAL)) {
  209. cAdd();
  210. cAddOperation(OP_EQUAL);
  211. } else if(cConsumeTokenIf(T_NOT_EQUAL)) {
  212. cAdd();
  213. cAddOperation(OP_EQUAL);
  214. cAddOperation(OP_NOT);
  215. } else {
  216. break;
  217. }
  218. }
  219. }
  220. static void cExpression() {
  221. cComparison();
  222. }
  223. static void cSetVar(const char* literal) {
  224. cExpression();
  225. cConsumeToken(T_SEMICOLON);
  226. cAddOperation(OP_SET);
  227. cAddInt(cAddVar(literal));
  228. }
  229. static void cLineLiteral() {
  230. const char* literal = cReadString();
  231. Token t = cReadTokenAndLine();
  232. switch(t) {
  233. case T_SET: cSetVar(literal); break;
  234. case T_OPEN_BRACKET:
  235. cCallFunction(literal, true);
  236. cConsumeToken(T_SEMICOLON);
  237. break;
  238. default: cUnexpectedToken(t);
  239. }
  240. }
  241. static int cFunctionArguments() {
  242. int arguments = 0;
  243. while(!cConsumeTokenIf(T_CLOSE_BRACKET)) {
  244. cConsumeToken(T_LITERAL);
  245. arguments++;
  246. cAddVar(cReadString());
  247. if(cConsumeTokenIf(T_COMMA) && tPeekToken() != T_LITERAL) {
  248. cUnexpectedToken(tPeekToken());
  249. }
  250. }
  251. return arguments;
  252. }
  253. static void cLine(Token t);
  254. static void cConsumeBody() {
  255. cConsumeToken(T_OPEN_CURVED_BRACKET);
  256. int oldLine = line;
  257. while(!cConsumeTokenIf(T_CLOSE_CURVED_BRACKET)) {
  258. Token t = cReadTokenAndLine();
  259. if(t == T_END) {
  260. cError("unexpected end of file: non closed curved bracket on line %d", oldLine);
  261. }
  262. cLine(t);
  263. }
  264. }
  265. static void cLinkReturns() {
  266. for(int i = 0; i < returnIndex; i++) {
  267. cSetInt(returns[i], vars[1].entries);
  268. }
  269. returnIndex = 0;
  270. }
  271. static void cFunctionBody(const char* name, int arguments) {
  272. int oldLine = line;
  273. cAddOperation(OP_GOTO);
  274. int gotoIndex = cReserveInt();
  275. int address = code->length;
  276. returnState = 0;
  277. int p = cAddPush(arguments);
  278. cConsumeBody();
  279. cAddPop(p, vars[1].entries);
  280. cLinkReturns();
  281. if(!fmAdd(&functions, name, arguments, address, returnState == 2)) {
  282. cError("function registered twice on line %d", oldLine);
  283. }
  284. cAddOperation(OP_RETURN);
  285. cSetInt(gotoIndex, code->length);
  286. }
  287. static void cFunction() {
  288. if(varIndex == 1) {
  289. cError("function inside function on line %d", line);
  290. }
  291. cConsumeToken(T_LITERAL);
  292. const char* name = cReadString();
  293. cConsumeToken(T_OPEN_BRACKET);
  294. varIndex = 1;
  295. vars[1].entries = 0;
  296. cFunctionBody(name, cFunctionArguments());
  297. varIndex = 0;
  298. }
  299. static void cAddReturn() {
  300. cAddOperation(OP_POP_VARS);
  301. returns[returnIndex++] = cReserveInt(vars);
  302. cAddOperation(OP_RETURN);
  303. }
  304. static void cReturn() {
  305. if(varIndex == 0) {
  306. cError("return without a function on line %d", line);
  307. } else if(returnIndex >= RETURN_BUFFER) {
  308. cError("too much returns in function around line %d", line);
  309. }
  310. if(cConsumeTokenIf(T_SEMICOLON)) {
  311. if(returnState == 2) {
  312. cError("mixed return type on line %d", line);
  313. }
  314. returnState = 1;
  315. cAddReturn();
  316. } else {
  317. if(returnState == 1) {
  318. cError("mixed return type on line %d", line);
  319. }
  320. returnState = 2;
  321. cExpression();
  322. cAddOperation(OP_SET_RETURN);
  323. cAddReturn();
  324. cConsumeToken(T_SEMICOLON);
  325. }
  326. }
  327. static void cPrint() {
  328. cExpression();
  329. cConsumeToken(T_SEMICOLON);
  330. cAddOperation(OP_PRINT);
  331. }
  332. static void cIf() {
  333. cConsumeToken(T_OPEN_BRACKET);
  334. cExpression();
  335. cConsumeToken(T_CLOSE_BRACKET);
  336. cAddOperation(OP_IF_GOTO);
  337. int p = cReserveInt();
  338. cConsumeBody();
  339. cSetInt(p, code->length);
  340. }
  341. static void cLine(Token t) {
  342. cAddOperation(OP_LINE);
  343. cAddInt16(line);
  344. switch(t) {
  345. case T_PRINT: cPrint(); break;
  346. case T_LITERAL: cLineLiteral(); break;
  347. case T_FUNCTION: cFunction(); break;
  348. case T_RETURN: cReturn(); break;
  349. case T_IF: cIf(); break;
  350. default: cUnexpectedToken(t);
  351. }
  352. }
  353. static void cForEachLine() {
  354. Token t = cReadTokenAndLine();
  355. while(t != T_END) {
  356. cLine(t);
  357. t = cReadTokenAndLine();
  358. }
  359. }
  360. static void cLinkQueuedFunctions() {
  361. for(int i = 0; i < functions.queueEntries; i++) {
  362. Function* f = fmSearch(&functions, functions.queue[i].name, functions.queue[i].arguments);
  363. if(f == NULL) {
  364. cError("unknown function on line %d", functions.queue[i].line);
  365. } else if(!functions.queue[i].noReturn && !f->returns) {
  366. cError("function '%s' needs a return value on line %d", f->name, functions.queue[i].line);
  367. }
  368. cSetInt(functions.queue[i].reserved, f->address);
  369. if(functions.queue[i].noReturn && f->returns) {
  370. code->code[functions.queue[i].reserved + sizeof(int) * 2] = OP_POP;
  371. }
  372. }
  373. }
  374. static void cAllocAndCompile() {
  375. varIndex = 0;
  376. returnIndex = 0;
  377. simInit(vars);
  378. simInit(vars + 1);
  379. fmInit(&functions);
  380. if(!setjmp(errorJump)) {
  381. int p = cAddPush(0);
  382. cForEachLine();
  383. cAddPop(p, vars[varIndex].entries);
  384. cLinkQueuedFunctions();
  385. }
  386. fmDelete(&functions);
  387. simDelete(vars + 1);
  388. simDelete(vars);
  389. }
  390. ByteCode* cCompile() {
  391. error[0] = '\0';
  392. code = bcInit();
  393. cAllocAndCompile();
  394. if(error[0] != '\0') {
  395. bcDelete(code);
  396. return NULL;
  397. }
  398. return code;
  399. }
  400. const char* cGetError() {
  401. return error;
  402. }