Compiler.c 7.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326
  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 MAX_BYTES (1024 * 1024)
  10. #define ERROR_LENGTH 256
  11. #define VARS 256
  12. static jmp_buf errorJump;
  13. static char error[ERROR_LENGTH] = {'\0'};
  14. static ByteCode* code;
  15. static int16 line = 1;
  16. static int varIndex = 0;
  17. static StringIntMap vars[2];
  18. static FunctionMap functions;
  19. static void cError(const char* format, ...) {
  20. va_list args;
  21. va_start(args, format);
  22. vsnprintf(error, ERROR_LENGTH, format, args);
  23. va_end(args);
  24. longjmp(errorJump, 0);
  25. }
  26. static int cAddVar(const char* var) {
  27. int index = vars[varIndex].entries;
  28. simAdd(vars + varIndex, var, &index);
  29. return index;
  30. }
  31. static void cUnexpectedToken(Token t) {
  32. cError("unexpected token on line %d: %s", line, tGetTokenName(t));
  33. }
  34. static void cAddOperation(Operation token) {
  35. unsigned char c = token;
  36. bcAddBytes(code, &c, 1);
  37. }
  38. static int cReserveInt() {
  39. return bcReserveBytes(code, sizeof(int));
  40. }
  41. static void cSetInt(int p, int i) {
  42. bcSetBytes(code, p, &i, sizeof(int));
  43. }
  44. static void cAddInt(int i) {
  45. bcAddBytes(code, &i, sizeof(int));
  46. }
  47. static void cAddInt16(int16 i) {
  48. bcAddBytes(code, &i, sizeof(int16));
  49. }
  50. static void cAddFloat(float f) {
  51. bcAddBytes(code, &f, sizeof(float));
  52. }
  53. static int cAddPush(int offset) {
  54. cAddOperation(OP_PUSH);
  55. int p = cReserveInt();
  56. cAddInt(offset);
  57. return p;
  58. }
  59. static void cAddPop(int p, int vars) {
  60. cAddOperation(OP_POP);
  61. cAddInt(vars);
  62. cSetInt(p, vars);
  63. }
  64. static Token cReadTokenAndLine() {
  65. Token t = tReadToken();
  66. if(tReadInt16(&line)) {
  67. return t;
  68. }
  69. return T_END;
  70. }
  71. static void cConsumeToken(Token wanted) {
  72. Token t = cReadTokenAndLine();
  73. if(wanted != t) {
  74. cError("unexpected token on line %d: expected '%s' got '%s'", line, tGetTokenName(wanted), tGetTokenName(t));
  75. }
  76. }
  77. static bool cConsumeTokenIf(Token t) {
  78. if(tPeekToken() == t) {
  79. cReadTokenAndLine();
  80. return true;
  81. }
  82. return false;
  83. }
  84. static void cConstantInt() {
  85. int value;
  86. if(!tReadInt(&value)) {
  87. cError("int token without an int on line %d", line);
  88. }
  89. cAddOperation(OP_PUSH_INT);
  90. cAddInt(value);
  91. }
  92. static void cConstantFloat() {
  93. float value;
  94. if(!tReadFloat(&value)) {
  95. cError("float token without a float on line %d", line);
  96. }
  97. cAddOperation(OP_PUSH_FLOAT);
  98. cAddFloat(value);
  99. }
  100. static const char* cReadString() {
  101. const char* literal = tReadString();
  102. if(literal == NULL) {
  103. cError("literal without string on line %d", line);
  104. }
  105. return literal;
  106. }
  107. static void cGetVar() {
  108. cAddOperation(OP_GET);
  109. cAddInt(cAddVar(cReadString()));
  110. }
  111. static void cExpression();
  112. static void cPrimary() {
  113. Token t = cReadTokenAndLine();
  114. switch(t) {
  115. case T_INT: cConstantInt(); break;
  116. case T_FLOAT: cConstantFloat(); break;
  117. case T_NULL: cAddOperation(OP_PUSH_NULL); break;
  118. case T_TRUE: cAddOperation(OP_PUSH_TRUE); break;
  119. case T_FALSE: cAddOperation(OP_PUSH_FALSE); break;
  120. case T_OPEN_BRACKET:
  121. cExpression();
  122. cConsumeToken(T_CLOSE_BRACKET);
  123. break;
  124. case T_LITERAL: cGetVar(); break;
  125. default: cUnexpectedToken(t); break;
  126. }
  127. }
  128. static void cMul() {
  129. cPrimary();
  130. while(cConsumeTokenIf(T_MUL)) {
  131. cPrimary();
  132. cAddOperation(OP_MUL);
  133. }
  134. }
  135. static void cAdd() {
  136. cMul();
  137. while(cConsumeTokenIf(T_ADD)) {
  138. cMul();
  139. cAddOperation(OP_ADD);
  140. }
  141. }
  142. static void cExpression() {
  143. cAdd();
  144. }
  145. static void cSetVar(const char* literal) {
  146. cExpression();
  147. cConsumeToken(T_SEMICOLON);
  148. cAddOperation(OP_SET);
  149. cAddInt(cAddVar(literal));
  150. }
  151. static int cCallFunctionArguments() {
  152. int arguments = 0;
  153. while(!cConsumeTokenIf(T_CLOSE_BRACKET)) {
  154. arguments++;
  155. cExpression();
  156. if(cConsumeTokenIf(T_COMMA) && tPeekToken() == T_CLOSE_BRACKET) {
  157. cUnexpectedToken(tPeekToken());
  158. }
  159. }
  160. return arguments;
  161. }
  162. static void cCallFunction(const char* literal) {
  163. cAddOperation(OP_PUSH_INT);
  164. cAddInt(0);
  165. int arguments = cCallFunctionArguments();
  166. int address = fmSearchAddress(&functions, literal, arguments);
  167. if(address == -1) {
  168. cError("unknown function on line %d", line);
  169. }
  170. cAddOperation(OP_GOSUB);
  171. cAddInt(address);
  172. cAddInt(arguments);
  173. cConsumeToken(T_SEMICOLON);
  174. }
  175. static void cLiteral() {
  176. const char* literal = cReadString();
  177. Token t = cReadTokenAndLine();
  178. switch(t) {
  179. case T_SET: cSetVar(literal); break;
  180. case T_OPEN_BRACKET: cCallFunction(literal); break;
  181. default: cUnexpectedToken(t);
  182. }
  183. }
  184. static int cFunctionArguments() {
  185. int arguments = 0;
  186. while(!cConsumeTokenIf(T_CLOSE_BRACKET)) {
  187. cConsumeToken(T_LITERAL);
  188. arguments++;
  189. cAddVar(cReadString());
  190. if(cConsumeTokenIf(T_COMMA) && tPeekToken() != T_LITERAL) {
  191. cUnexpectedToken(tPeekToken());
  192. }
  193. }
  194. return arguments;
  195. }
  196. static void cLine(Token t);
  197. static void cFunctionInnerBody(int arguments) {
  198. int p = cAddPush(arguments);
  199. int oldLine = line;
  200. while(!cConsumeTokenIf(T_CLOSE_CURVED_BRACKET)) {
  201. Token t = cReadTokenAndLine();
  202. if(t == T_END) {
  203. cError("unexpected end of file: function not closed on line %d", oldLine);
  204. }
  205. cLine(t);
  206. }
  207. cAddPop(p, vars[1].entries);
  208. }
  209. static void cFunctionBody(const char* name, int arguments) {
  210. cConsumeToken(T_OPEN_CURVED_BRACKET);
  211. cAddOperation(OP_GOTO);
  212. int gotoIndex = cReserveInt();
  213. if(!fmAdd(&functions, name, arguments, code->length)) {
  214. cError("function registered twice on line %d", line);
  215. }
  216. cFunctionInnerBody(arguments);
  217. cAddOperation(OP_RETURN);
  218. cSetInt(gotoIndex, code->length);
  219. }
  220. static void cFunction() {
  221. if(varIndex == 1) {
  222. cError("function inside function on line %d", line);
  223. }
  224. cConsumeToken(T_LITERAL);
  225. const char* name = cReadString();
  226. cConsumeToken(T_OPEN_BRACKET);
  227. varIndex = 1;
  228. vars[1].entries = 0;
  229. cFunctionBody(name, cFunctionArguments());
  230. varIndex = 0;
  231. }
  232. static void cPrint() {
  233. cExpression();
  234. cConsumeToken(T_SEMICOLON);
  235. cAddOperation(OP_PRINT);
  236. }
  237. static void cLine(Token t) {
  238. cAddOperation(OP_LINE);
  239. cAddInt16(line);
  240. switch(t) {
  241. case T_PRINT: cPrint(); break;
  242. case T_LITERAL: cLiteral(); break;
  243. case T_FUNCTION: cFunction(); break;
  244. default: cUnexpectedToken(t);
  245. }
  246. }
  247. static void cForEachLine() {
  248. Token t = cReadTokenAndLine();
  249. while(t != T_END) {
  250. cLine(t);
  251. t = cReadTokenAndLine();
  252. }
  253. }
  254. static void cAllocAndCompile() {
  255. varIndex = 0;
  256. simInit(vars);
  257. simInit(vars + 1);
  258. fmInit(&functions);
  259. if(!setjmp(errorJump)) {
  260. int p = cAddPush(0);
  261. cForEachLine();
  262. cAddPop(p, vars[varIndex].entries);
  263. }
  264. fmDelete(&functions);
  265. simDelete(vars + 1);
  266. simDelete(vars);
  267. }
  268. ByteCode* cCompile() {
  269. error[0] = '\0';
  270. code = bcInit();
  271. cAllocAndCompile();
  272. if(error[0] != '\0') {
  273. bcDelete(code);
  274. return NULL;
  275. }
  276. return code;
  277. }
  278. const char* cGetError() {
  279. return error;
  280. }