Year 2 compilers coureswork

main.c 11KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431
  1. #include <stdio.h>
  2. #include <stdlib.h>
  3. #include <ctype.h>
  4. #include <string.h>
  5. #include "nodes.h"
  6. #include "C.tab.h"
  7. #include "types.h"
  8. #include "list.h"
  9. extern TOKEN* lookup_token(char*);
  10. TOKEN* gen_tmp(void) {
  11. char tmp[10];
  12. static char num = 0;
  13. snprintf(tmp, 10, "$t%d", num++);
  14. return lookup_token(tmp);
  15. }
  16. TOKEN* gen_label(void) {
  17. char tmp[10];
  18. static char label = 0;
  19. snprintf(tmp, 10, "L%d", label++);
  20. return lookup_token(tmp);
  21. }
  22. char *named(int t)
  23. {
  24. static char b[100];
  25. if (isgraph(t) || t==' ') {
  26. sprintf(b, "%c", t);
  27. return b;
  28. }
  29. switch (t) {
  30. default: return "???";
  31. case IDENTIFIER:
  32. return "id";
  33. case CONSTANT:
  34. return "constant";
  35. case STRING_LITERAL:
  36. return "string";
  37. case LE_OP:
  38. return "<=";
  39. case GE_OP:
  40. return ">=";
  41. case EQ_OP:
  42. return "==";
  43. case NE_OP:
  44. return "!=";
  45. case EXTERN:
  46. return "extern";
  47. case AUTO:
  48. return "auto";
  49. case INT:
  50. return "int";
  51. case VOID:
  52. return "void";
  53. case APPLY:
  54. return "apply";
  55. case LEAF:
  56. return "leaf";
  57. case IF:
  58. return "if";
  59. case ELSE:
  60. return "else";
  61. case WHILE:
  62. return "while";
  63. case CONTINUE:
  64. return "continue";
  65. case BREAK:
  66. return "break";
  67. case RETURN:
  68. return "return";
  69. }
  70. }
  71. void print_leaf(NODE *tree, int level)
  72. {
  73. TOKEN *t = (TOKEN *)tree;
  74. int i;
  75. for (i=0; i<level; i++) putchar(' ');
  76. if (t->type == CONSTANT) printf("%d\n", t->value);
  77. else if (t->type == STRING_LITERAL) printf("\"%s\"\n", t->lexeme);
  78. else if (t) puts(t->lexeme);
  79. }
  80. void print_tree0(NODE *tree, int level)
  81. {
  82. int i;
  83. if (tree==NULL) return;
  84. if (tree->type==LEAF) {
  85. print_leaf(tree->left, level);
  86. }
  87. else {
  88. for(i=0; i<level; i++) putchar(' ');
  89. printf("%s\n", named(tree->type));
  90. print_tree0(tree->left, level+2);
  91. print_tree0(tree->right, level+2);
  92. }
  93. }
  94. void print_tree(NODE *tree)
  95. {
  96. print_tree0(tree, 0);
  97. }
  98. void new_tac(TACS* tacs, TOKEN* dst, TOKEN* src, TOKEN* tgt, int op) {
  99. if (tacs->list == NULL) {
  100. TAC_T* tac;
  101. tac = create_tac(op, src, tgt, dst);
  102. tacs->list = new_tac_list(tac, NULL);
  103. } else {
  104. add_tac(tacs->list, op, src, tgt, dst);
  105. }
  106. }
  107. char* stok(TOKEN* tok) {
  108. return tok->lexeme;
  109. }
  110. void print_tac(TLIST* tac_list) {
  111. TLIST *ptr = tac_list;
  112. while (ptr != NULL) {
  113. int op = ptr->elem->op;
  114. if (op == '=') printf("%s := %s\n", stok(ptr->elem->dst), stok(ptr->elem->src));
  115. if (op=='+' || op=='-' || op=='*' || op=='/' || op =='%') printf("%s := %s %c %s\n", stok(ptr->elem->dst), stok(ptr->elem->src), op, stok(ptr->elem->tgt));
  116. if (op == RETURN) printf("ret %s\n", stok(ptr->elem->src));
  117. if (op == 'D') printf("\nfunc %s\n", stok(ptr->elem->dst));
  118. if (op == '>' || op == '<') printf("if %s %c %s then %s\n", stok(ptr->elem->src), op, stok(ptr->elem->tgt), stok(ptr->elem->dst));
  119. if (op == GE_OP) printf("if %s >= %s then %s\n", stok(ptr->elem->src), stok(ptr->elem->tgt), stok(ptr->elem->dst));
  120. if (op == LE_OP) printf("if %s <= %s then %s\n", stok(ptr->elem->src), stok(ptr->elem->tgt), stok(ptr->elem->dst));
  121. if (op == EQ_OP) printf("if %s == %s then %s\n", stok(ptr->elem->src), stok(ptr->elem->tgt), stok(ptr->elem->dst));
  122. if (op == NE_OP) printf("if %s != %s then %s\n", stok(ptr->elem->src), stok(ptr->elem->tgt), stok(ptr->elem->dst));
  123. if (op == IDENTIFIER) printf("%s\n", stok(ptr->elem->dst));
  124. if (op == VOID) printf("goto %s\n", stok(ptr->elem->dst));
  125. if (op == AUTO) printf("param %s\n", stok(ptr->elem->dst));
  126. if (op == APPLY) printf("call %s, %s\n", stok(ptr->elem->dst), stok(ptr->elem->tgt));
  127. if (op == 278) printf("arg %s\n", stok(ptr->elem->dst));
  128. if (op == 279) printf("end %s\n", stok(ptr->elem->dst));
  129. ptr = ptr->next;
  130. }
  131. }
  132. int invert_op(int op) {
  133. if (op == LE_OP) {
  134. return '>';
  135. } else if (op == GE_OP) {
  136. return '<';
  137. } else if (op == '>') {
  138. return LE_OP;
  139. } else if (op == '<') {
  140. return GE_OP;
  141. } else if (op == EQ_OP) {
  142. return NE_OP;
  143. } else if (op == NE_OP) {
  144. return EQ_OP;
  145. } else {
  146. printf("Unknown operator: %c", op);
  147. exit(1);
  148. }
  149. }
  150. int count_args(NODE *tree) {
  151. NODE *tmp = tree;
  152. int a = 1;
  153. while (tmp != NULL) {
  154. if (tmp->type != ',') {
  155. break;
  156. }
  157. tmp = tmp->left;
  158. a++;
  159. }
  160. return a;
  161. }
  162. void populate_arg_list(NODE *tree, TOKEN** arr, int arr_len) {
  163. if (tree->type == '~') {
  164. for (int i = 0; i < arr_len; i++) {
  165. if (arr[i] == NULL) {
  166. TOKEN* name = (TOKEN*)tree->right->left;
  167. arr[i] = name;
  168. break;
  169. }
  170. }
  171. } else {
  172. populate_arg_list(tree->left, arr, arr_len);
  173. populate_arg_list(tree->right, arr, arr_len);
  174. }
  175. }
  176. TOKEN** get_argument_list(NODE *tree) {
  177. int num_args = count_args(tree->left->right->right);
  178. TOKEN** arr = (TOKEN**) malloc(sizeof(TOKEN*) * num_args);
  179. populate_arg_list(tree->left->right->right, arr, num_args);
  180. return arr;
  181. }
  182. void populate_val_list(NODE* tree, NODE** arr, int num_args) {
  183. if (tree->type == ',') {
  184. populate_val_list(tree->left, arr, num_args);
  185. populate_val_list(tree->right, arr, num_args);
  186. } else {
  187. for (int i = 0; i < num_args; i++) {
  188. if (arr[i] == NULL) {
  189. arr[i] = tree;
  190. break;
  191. }
  192. }
  193. }
  194. }
  195. NODE** get_val_list(NODE *tree) {
  196. int num_args = count_args(tree);
  197. NODE** arr = (NODE**) malloc(sizeof(NODE*) * num_args);
  198. populate_val_list(tree, arr, num_args);
  199. return arr;
  200. }
  201. TOKEN* build_tac(NODE *tree, TACS* tac) {
  202. if (tree == NULL) return NULL;
  203. int type = tree->type;
  204. if (type == LEAF) {
  205. return (TOKEN *) tree->left;
  206. } else if (type == '=') {
  207. TOKEN* dst = (TOKEN*) tree->left->left;
  208. TOKEN* result = build_tac(tree->right, tac);
  209. new_tac(tac, dst, result, NULL, tree->type);
  210. } else if (type=='+' || type=='-' || type=='*' || type=='/' || type =='%') {
  211. TOKEN* dst = gen_tmp();
  212. TOKEN* src = build_tac(tree->left, tac);
  213. TOKEN* tgt = build_tac(tree->right, tac);
  214. new_tac(tac, dst, src, tgt, tree->type);
  215. return dst;
  216. } else if (type == IF) {
  217. int op = invert_op(tree->left->type);
  218. TOKEN* lhs = build_tac(tree->left->left, tac);
  219. TOKEN* rhs = build_tac(tree->left->right, tac);
  220. TOKEN* label = gen_label();
  221. new_tac(tac, label, lhs, rhs, op);
  222. if (tree->right->type == ELSE) {
  223. TOKEN* cont = gen_label();
  224. build_tac(tree->right->left, tac);
  225. new_tac(tac, cont, NULL, NULL, VOID);
  226. new_tac(tac, label, NULL, NULL, IDENTIFIER);
  227. build_tac(tree->right->right, tac);
  228. new_tac(tac, cont, NULL, NULL, IDENTIFIER);
  229. } else {
  230. build_tac(tree->right, tac);
  231. new_tac(tac, label, NULL, NULL, IDENTIFIER);
  232. }
  233. } else if (type == RETURN) {
  234. TOKEN* src = build_tac(tree->left, tac);
  235. new_tac(tac, NULL, src, NULL, RETURN);
  236. } else if (type == 'D') {
  237. new_tac(tac, (TOKEN*) tree->left->right->left->left, NULL, NULL, 'D');
  238. if (tree->left->right->right != NULL) {
  239. TOKEN** arg_list = get_argument_list(tree);
  240. int arg_count = count_args(tree->left->right->right);
  241. for (int i = 0; i < arg_count; i++) {
  242. new_tac(tac, arg_list[i], NULL, NULL, AUTO);
  243. }
  244. }
  245. build_tac(tree->right, tac);
  246. new_tac(tac, (TOKEN*) tree->left->right->left->left, NULL, NULL, 279);
  247. } else if (type == APPLY) {
  248. TOKEN* ret_val = gen_tmp();
  249. if (tree->right != NULL) {
  250. int num_args;
  251. NODE **val_list;
  252. num_args = count_args(tree->right);
  253. val_list = get_val_list(tree->right);
  254. for (int i = 0; i < num_args; i++) {
  255. TOKEN* arg_token = build_tac(val_list[i], tac);
  256. new_tac(tac, arg_token, NULL, NULL, 278);
  257. }
  258. }
  259. new_tac(tac, (TOKEN*) tree->left->left, NULL, ret_val, APPLY);
  260. return ret_val;
  261. } else {
  262. build_tac(tree->left, tac);
  263. build_tac(tree->right, tac);
  264. return NULL;
  265. }
  266. return NULL;
  267. }
  268. int tmp_regs[10] = {8, 9, 10, 11, 12, 13, 14, 15, 24, 25};
  269. int svd_regs[8] = {16, 17, 18, 19, 20, 21, 22, 23};
  270. int tmp_cntr = 0;
  271. VARLOC** vartab;
  272. int is_imm(TOKEN *t) {
  273. if (t->type==CONSTANT) return 1;
  274. else return 0;
  275. }
  276. int find_var_idx(TOKEN* var) {
  277. int i;
  278. for (i = 0; i < 8; i++) {
  279. if (vartab[i]->var == var) {
  280. return vartab[i]->loc_idx;
  281. }
  282. }
  283. return -1;
  284. }
  285. int store_var(TOKEN* name) {
  286. int i;
  287. for (i = 0; i < 8; i++) {
  288. if (vartab[i]->var == NULL) {
  289. vartab[i]->var = name;
  290. vartab[i]->loc_idx = svd_regs[i];
  291. return vartab[i]->loc_idx;
  292. } else if (name == vartab[i]->var) {
  293. vartab[i]->loc_idx = svd_regs[i];
  294. return svd_regs[i];
  295. }
  296. }
  297. printf("bugger\n");
  298. exit(1);
  299. }
  300. void init_reg() {
  301. int i;
  302. vartab = (VARLOC**) malloc(8 * sizeof(VARLOC*));
  303. for (i = 0; i < 8; i++) {
  304. vartab[i] = (VARLOC*) malloc(sizeof(VARLOC));
  305. vartab[i]->var = NULL;
  306. vartab[i]->loc_idx = -1;
  307. }
  308. }
  309. int alloc_register(TOKEN* val) {
  310. int idx;
  311. int imm = is_imm(val);
  312. if (imm) {
  313. if (val->value <= 32767 && val->value >= -32767) {
  314. printf("ori $%d, $0, %d\n", tmp_regs[tmp_cntr], val->value);
  315. idx = tmp_regs[tmp_cntr++];
  316. } else {
  317. printf("li $%d, %d\n", tmp_regs[tmp_cntr], val->value);
  318. idx = tmp_regs[tmp_cntr++];
  319. }
  320. } else {
  321. idx = find_var_idx(val);
  322. }
  323. if (tmp_cntr == 10) tmp_cntr = 0;
  324. return idx;
  325. }
  326. void compile_tac(TLIST* tacs) {
  327. printf("=====\n");
  328. init_reg();
  329. while(tacs != NULL) {
  330. TAC_T* t = tacs->elem;
  331. if (t->op == '=') {
  332. int src_idx = alloc_register(t->src);
  333. printf("or $%d, $0, $%d\n", store_var(t->dst), src_idx);
  334. }
  335. if (t->op == 'D') {
  336. printf("%s:\n", t->dst->lexeme);
  337. }
  338. if (t->op == '+') {
  339. int tgt_idx = alloc_register(t->tgt);
  340. int src_idx = alloc_register(t->src);
  341. printf("add $%d, $%d, $%d\n", store_var(t->dst), src_idx, tgt_idx);
  342. }
  343. if (t->op == '-') {
  344. int tgt_idx = alloc_register(t->tgt);
  345. int src_idx = alloc_register(t->src);
  346. printf("sub $%d, $%d, $%d\n", store_var(t->dst), src_idx, tgt_idx);
  347. }
  348. if (t->op == '*') {
  349. int tgt_idx = alloc_register(t->tgt);
  350. int src_idx = alloc_register(t->src);
  351. printf("mult $%d, $%d\n", src_idx, tgt_idx);
  352. printf("mflo $%d\n", store_var(t->dst));
  353. }
  354. if (t->op == '/') {
  355. int tgt_idx = alloc_register(t->tgt);
  356. int src_idx = alloc_register(t->src);
  357. printf("div $%d, $%d\n", src_idx, tgt_idx);
  358. printf("mflo $%d\n", store_var(t->dst));
  359. }
  360. if (t->op == '%') {
  361. int tgt_idx = alloc_register(t->tgt);
  362. int src_idx = alloc_register(t->src);
  363. printf("div $%d, $%d\n", src_idx, tgt_idx);
  364. printf("mfhi $%d\n", store_var(t->dst));
  365. }
  366. if (t->op == RETURN) {
  367. int src_idx = alloc_register(t->src);
  368. printf("or $v1, $0, $%d\n", src_idx);
  369. }
  370. tacs = tacs->next;
  371. }
  372. printf("ori $v0, 10\n");
  373. printf("syscall\n");
  374. }
  375. void interpret_tree(NODE *tree) {
  376. TACS* gen_tac = (TACS*) malloc(sizeof(TACS));
  377. build_tac(tree, gen_tac);
  378. print_tac(gen_tac->list);
  379. compile_tac(gen_tac->list);
  380. }
  381. extern int yydebug;
  382. #ifdef __APPLE__
  383. extern NODE* yyparse(void);
  384. #endif
  385. extern NODE* ans;
  386. extern void init_symbtable(void);
  387. int main(int argc, char** argv)
  388. {
  389. NODE* tree;
  390. if (argc>1 && strcmp(argv[1],"-d")==0) yydebug = 1;
  391. init_symbtable();
  392. printf("--C COMPILER\n");
  393. yyparse();
  394. tree = ans;
  395. printf("parse finished\n");
  396. print_tree(tree);
  397. interpret_tree(tree);
  398. return 0;
  399. }