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.

parse.c 15KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561
  1. #include "parse/parse.h"
  2. #include "trace.h"
  3. #include "gen/gen.h"
  4. static int tok_is_end(struct l2_token *tok) {
  5. enum l2_token_kind kind = l2_token_get_kind(tok);
  6. return
  7. kind == L2_TOK_CLOSE_BRACE || kind == L2_TOK_CLOSE_BRACKET ||
  8. kind == L2_TOK_CLOSE_PAREN || kind == L2_TOK_EOF ||
  9. kind == L2_TOK_EOL;
  10. }
  11. static int tok_is_infix(struct l2_token *tok) {
  12. if (l2_token_get_kind(tok) != L2_TOK_IDENT) return 0;
  13. char *str;
  14. if (l2_token_is_small(tok)) {
  15. str = tok->v.strbuf;
  16. } else {
  17. str = tok->v.str;
  18. }
  19. return
  20. (str[0] == '$' && str[1] != '\0') ||
  21. strcmp(str, "+") == 0 ||
  22. strcmp(str, "-") == 0 ||
  23. strcmp(str, "*") == 0 ||
  24. strcmp(str, "/") == 0 ||
  25. strcmp(str, "==") == 0 ||
  26. strcmp(str, "!=") == 0 ||
  27. strcmp(str, "<") == 0 ||
  28. strcmp(str, "<=") == 0 ||
  29. strcmp(str, ">") == 0 ||
  30. strcmp(str, ">=") == 0;
  31. }
  32. static int parse_expression(
  33. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err);
  34. static int parse_arg_level_expression(
  35. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err);
  36. static int parse_object_literal(
  37. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  38. l2_trace_scope("object literal");
  39. // '{' and EOL already skipped by parse_object_or_function_literal
  40. l2_gen_namespace(gen);
  41. while (1) {
  42. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  43. if (l2_token_get_kind(tok) == L2_TOK_CLOSE_BRACE) {
  44. l2_lexer_consume(lexer); // '}'
  45. break;
  46. } else if (l2_token_get_kind(tok) != L2_TOK_IDENT) {
  47. l2_parse_err(err, tok, "In object literal: Expected identifier, got %s",
  48. l2_token_get_name(tok));
  49. return -1;
  50. }
  51. l2_trace("key: '%s'", tok->v.str);
  52. struct l2_token_value key = l2_token_extract_val(tok);
  53. l2_lexer_consume(lexer); // ident
  54. tok = l2_lexer_peek(lexer, 1);
  55. if (l2_token_get_kind(tok) != L2_TOK_COLON) {
  56. if (!(key.flags & L2_TOK_SMALL)) free(key.str);
  57. l2_parse_err(err, tok, "In object literal: Expected ':', got %s",
  58. l2_token_get_name(tok));
  59. return -1;
  60. }
  61. l2_lexer_consume(lexer); // ':'
  62. if (parse_expression(lexer, gen, err) < 0) {
  63. if (!(key.flags & L2_TOK_SMALL)) free(key.str);
  64. return -1;
  65. }
  66. if (key.flags & L2_TOK_SMALL) {
  67. l2_gen_namespace_set_copy(gen, key.strbuf);
  68. } else {
  69. l2_gen_namespace_set_copy(gen, key.str);
  70. }
  71. l2_gen_discard(gen);
  72. tok = l2_lexer_peek(lexer, 1);
  73. if (
  74. l2_token_get_kind(tok) != L2_TOK_EOL &&
  75. l2_token_get_kind(tok) != L2_TOK_CLOSE_BRACE) {
  76. l2_parse_err(err, tok, "In object literal: Expected EOL or '}', got %s",
  77. l2_token_get_name(tok));
  78. return -1;
  79. }
  80. if (l2_token_get_kind(tok) == L2_TOK_EOL) {
  81. l2_lexer_consume(lexer); // EOL
  82. }
  83. }
  84. return 0;
  85. }
  86. static int parse_function_literal_impl(
  87. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  88. l2_trace_scope("function literal");
  89. // '{' and EOL already skipped by parse_object_or_function_literal
  90. // The arguments array will be at the top of the stack
  91. char *ident = "$";
  92. l2_gen_stack_frame_set_copy(gen, ident);
  93. int first = 1;
  94. while (1) {
  95. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_CLOSE_BRACE) {
  96. l2_lexer_consume(lexer); // '}'
  97. break;
  98. }
  99. if (!first) {
  100. l2_gen_discard(gen);
  101. }
  102. l2_trace_scope("function literal expression");
  103. if (parse_expression(lexer, gen, err) < 0) {
  104. return -1;
  105. }
  106. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  107. first = 0;
  108. }
  109. // All functions must put _something_ on the stack
  110. if (first) {
  111. l2_gen_none(gen);
  112. }
  113. l2_gen_ret(gen);
  114. return 0;
  115. }
  116. static int parse_function_literal(
  117. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  118. l2_gen_flush(gen);
  119. struct l2_io_writer *prev_writer = gen->writer.w;
  120. // Generate the function to a buffer in memory
  121. struct l2_io_mem_writer w = {0};
  122. w.w.write = l2_io_mem_write;
  123. gen->writer.w = &w.w;
  124. // Generates five bytes; RJMP, then 4 byte counter
  125. l2_gen_rjmp_placeholder(gen);
  126. l2_word pos = gen->pos;
  127. // Generate the function body itself
  128. int ret = parse_function_literal_impl(lexer, gen, err);
  129. l2_gen_flush(gen);
  130. gen->writer.w = prev_writer;
  131. if (ret < 0) {
  132. free(w.mem);
  133. return -1;
  134. }
  135. unsigned char *ops = w.mem;
  136. l2_word opcount = w.len - 5;
  137. // Write the jump distance (little endian)
  138. ops[1] = (opcount >> 0) & 0xff;
  139. ops[2] = (opcount >> 8) & 0xff;
  140. ops[3] = (opcount >> 16) & 0xff;
  141. ops[4] = (opcount >> 24) & 0xff;
  142. l2_bufio_put_n(&gen->writer, ops, w.len);
  143. free(w.mem);
  144. l2_gen_function(gen, pos);
  145. return 0;
  146. }
  147. static int parse_object_or_function_literal(
  148. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  149. l2_trace_scope("object or function literal");
  150. l2_lexer_consume(lexer); // '{'
  151. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  152. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  153. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  154. if (l2_token_get_kind(tok) == L2_TOK_CLOSE_BRACE) {
  155. l2_trace_scope("empty object literal");
  156. l2_lexer_consume(lexer); // '}'
  157. l2_gen_namespace(gen);
  158. } else if (
  159. l2_token_get_kind(tok) == L2_TOK_IDENT &&
  160. l2_token_get_kind(tok2) == L2_TOK_COLON) {
  161. if (parse_object_literal(lexer, gen, err) < 0) {
  162. return -1;
  163. }
  164. } else {
  165. if (parse_function_literal(lexer, gen, err) < 0) {
  166. return -1;
  167. }
  168. }
  169. return 0;
  170. }
  171. static int parse_array_literal(
  172. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  173. l2_trace_scope("array literal");
  174. l2_lexer_consume(lexer); // '['
  175. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  176. int count = 0;
  177. while (1) {
  178. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_CLOSE_BRACKET) {
  179. l2_lexer_consume(lexer); // ']'
  180. break;
  181. }
  182. count += 1;
  183. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  184. return -1;
  185. }
  186. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  187. }
  188. l2_gen_array(gen, count);
  189. return 0;
  190. }
  191. static int parse_arg_level_expression_base(
  192. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  193. l2_trace_scope("arg level expression base");
  194. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  195. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  196. if (l2_token_get_kind(tok) == L2_TOK_OPEN_PAREN) {
  197. l2_trace_scope("group expr");
  198. l2_lexer_consume(lexer); // '('
  199. if (parse_expression(lexer, gen, err) < 0) {
  200. return -1;
  201. }
  202. tok = l2_lexer_peek(lexer, 1);
  203. if (l2_token_get_kind(tok) != L2_TOK_CLOSE_PAREN) {
  204. l2_parse_err(err, tok, "Expected ')', got %s",
  205. l2_token_get_name(tok));
  206. return -1;
  207. }
  208. l2_lexer_consume(lexer); // ')'
  209. } else if (l2_token_get_kind(tok) == L2_TOK_IDENT) {
  210. l2_trace_scope("ident");
  211. l2_trace("ident '%s'", tok->v.str);
  212. struct l2_token_value ident = l2_token_extract_val(tok);
  213. l2_lexer_consume(lexer); // ident
  214. if (ident.flags & L2_TOK_SMALL) {
  215. l2_gen_stack_frame_lookup_copy(gen, ident.strbuf);
  216. } else {
  217. l2_gen_stack_frame_lookup(gen, &ident.str);
  218. }
  219. } else if (l2_token_get_kind(tok) == L2_TOK_NUMBER) {
  220. l2_trace_scope("number literal");
  221. l2_trace("number %g", tok->v.num);
  222. double number = tok->v.num;
  223. l2_lexer_consume(lexer); // number
  224. l2_gen_number(gen, number);
  225. } else if (l2_token_get_kind(tok) == L2_TOK_STRING) {
  226. l2_trace_scope("string literal");
  227. l2_trace("string '%s'", tok->v.str);
  228. struct l2_token_value str = l2_token_extract_val(tok);
  229. l2_lexer_consume(lexer); // string
  230. if (str.flags & L2_TOK_SMALL) {
  231. l2_gen_string_copy(gen, str.strbuf);
  232. } else {
  233. l2_gen_string(gen, &str.str);
  234. }
  235. } else if (
  236. l2_token_get_kind(tok) == L2_TOK_QUOT &&
  237. l2_token_get_kind(tok2) == L2_TOK_IDENT) {
  238. l2_trace_scope("atom literal");
  239. l2_trace("atom '%s'", tok->v.str);
  240. struct l2_token_value ident = l2_token_extract_val(tok2);
  241. l2_lexer_consume(lexer); // "'"
  242. l2_lexer_consume(lexer); // ident
  243. if (ident.flags & L2_TOK_SMALL) {
  244. l2_gen_atom_copy(gen, ident.strbuf);
  245. } else {
  246. l2_gen_atom(gen, &ident.str);
  247. }
  248. } else if (l2_token_get_kind(tok) == L2_TOK_OPEN_BRACE) {
  249. if (parse_object_or_function_literal(lexer, gen, err) < 0) {
  250. return -1;
  251. }
  252. } else if (l2_token_get_kind(tok) == L2_TOK_OPEN_BRACKET) {
  253. if (parse_array_literal(lexer, gen, err) < 0) {
  254. return -1;
  255. }
  256. } else {
  257. l2_parse_err(err, tok, "Unexpected token %s",
  258. l2_token_get_name(tok));
  259. return -1;
  260. }
  261. return 0;
  262. }
  263. static int parse_arg_level_expression(
  264. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  265. l2_trace_scope("arg level expression");
  266. if (parse_arg_level_expression_base(lexer, gen, err) < 0) {
  267. return -1;
  268. }
  269. while (1) {
  270. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  271. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  272. struct l2_token *tok3 = l2_lexer_peek(lexer, 3);
  273. if (
  274. l2_token_get_kind(tok) == L2_TOK_OPEN_PAREN &&
  275. l2_token_get_kind(tok2) == L2_TOK_CLOSE_PAREN) {
  276. l2_trace_scope("niladic func call");
  277. l2_lexer_consume(lexer); // '('
  278. l2_lexer_consume(lexer); // ')'
  279. l2_gen_func_call(gen, 0);
  280. } else if (
  281. l2_token_get_kind(tok) == L2_TOK_PERIOD &&
  282. l2_token_get_kind(tok2) == L2_TOK_IDENT &&
  283. l2_token_get_kind(tok3) == L2_TOK_EQUALS) {
  284. l2_trace_scope("namespace assign");
  285. l2_trace("ident '%s'", tok2->v.str);
  286. struct l2_token_value ident = l2_token_extract_val(tok2);
  287. l2_lexer_consume(lexer); // '.'
  288. l2_lexer_consume(lexer); // ident
  289. l2_lexer_consume(lexer); // '='
  290. if (parse_expression(lexer, gen, err) < 0) {
  291. if (!(ident.flags & L2_TOK_SMALL)) free(ident.str);
  292. return -1;
  293. }
  294. if (ident.flags & L2_TOK_SMALL) {
  295. l2_gen_namespace_set_copy(gen, ident.strbuf);
  296. } else {
  297. l2_gen_namespace_set(gen, &ident.str);
  298. }
  299. l2_gen_swap_discard(gen);
  300. } else if (
  301. l2_token_get_kind(tok) == L2_TOK_PERIOD &&
  302. l2_token_get_kind(tok2) == L2_TOK_IDENT) {
  303. l2_trace_scope("namespace lookup");
  304. l2_trace("ident '%s'", tok2->v.str);
  305. struct l2_token_value ident = l2_token_extract_val(tok2);
  306. l2_lexer_consume(lexer); // '.'
  307. l2_lexer_consume(lexer); // ident
  308. if (ident.flags & L2_TOK_SMALL) {
  309. l2_gen_namespace_lookup_copy(gen, ident.strbuf);
  310. } else {
  311. l2_gen_namespace_lookup(gen, &ident.str);
  312. }
  313. } else if (
  314. l2_token_get_kind(tok) == L2_TOK_DOT_NUMBER &&
  315. l2_token_get_kind(tok2) == L2_TOK_EQUALS) {
  316. l2_trace_scope("direct array assign");
  317. int number = tok->v.integer;
  318. l2_lexer_consume(lexer); // dot-number
  319. l2_lexer_consume(lexer); // '='
  320. if (parse_expression(lexer, gen, err) < 0) {
  321. return -1;
  322. }
  323. l2_gen_array_set(gen, number);
  324. l2_gen_swap_discard(gen);
  325. } else if (l2_token_get_kind(tok) == L2_TOK_DOT_NUMBER) {
  326. l2_trace_scope("direct array lookup");
  327. int number = tok->v.integer;
  328. l2_lexer_consume(lexer); // dot-number
  329. l2_gen_array_lookup(gen, number);
  330. } else if (
  331. l2_token_get_kind(tok) == L2_TOK_PERIOD &&
  332. l2_token_get_kind(tok2) == L2_TOK_OPEN_PAREN) {
  333. l2_trace_scope("dynamic lookup");
  334. l2_lexer_consume(lexer); // '.'
  335. l2_lexer_consume(lexer); // '('
  336. if (parse_expression(lexer, gen, err) < 0) {
  337. return -1;
  338. }
  339. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) != L2_TOK_CLOSE_PAREN) {
  340. l2_parse_err(err, tok, "Expected '(', got %s",
  341. l2_token_get_name(tok));
  342. return -1;
  343. }
  344. l2_lexer_consume(lexer); // ')'
  345. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_EQUALS) {
  346. l2_lexer_consume(lexer); // '='
  347. if (parse_expression(lexer, gen, err) < 0) {
  348. return -1;
  349. }
  350. l2_gen_dynamic_set(gen);
  351. } else {
  352. l2_gen_dynamic_lookup(gen);
  353. }
  354. } else {
  355. break;
  356. }
  357. }
  358. return 0;
  359. }
  360. static int parse_func_call_after_base(
  361. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  362. l2_trace_scope("func call after base");
  363. size_t argc = 0;
  364. do {
  365. if (tok_is_infix(l2_lexer_peek(lexer, 1))) {
  366. do {
  367. // We already have one value (the lhs) on the stack,
  368. // so we need to parse the operator, then the rhs
  369. // Operator
  370. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  371. return -1;
  372. }
  373. // RHS
  374. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  375. return -1;
  376. }
  377. l2_gen_func_call_infix(gen);
  378. } while (tok_is_infix(l2_lexer_peek(lexer, 1)));
  379. // If this was the "first argument", this wasn't a function call
  380. // after all, it was just a (series of?) infix calls.
  381. if (argc == 0) {
  382. return 0;
  383. }
  384. // Don't increment argc here, because after an infix, we have
  385. // neither added nor removed an arguemnt, just transformed one
  386. } else {
  387. l2_trace_scope("func call param");
  388. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  389. return -1;
  390. }
  391. argc += 1;
  392. }
  393. } while (!tok_is_end(l2_lexer_peek(lexer, 1)));
  394. // The 'argc' previous expressions were arguments, the one before that was the function
  395. l2_gen_func_call(gen, argc);
  396. return 0;
  397. }
  398. static int parse_expression(
  399. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  400. l2_trace_scope("expression");
  401. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  402. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  403. if (
  404. l2_token_get_kind(tok) == L2_TOK_IDENT &&
  405. l2_token_get_kind(tok2) == L2_TOK_COLON_EQ) {
  406. l2_trace_scope("assign expression");
  407. l2_trace("ident '%s'", tok->v.str);
  408. struct l2_token_value ident = l2_token_extract_val(tok);
  409. l2_lexer_consume(lexer); // ident
  410. l2_lexer_consume(lexer); // :=
  411. if (parse_expression(lexer, gen, err) < 0) {
  412. if (!(ident.flags & L2_TOK_SMALL)) free(ident.str);
  413. return -1;
  414. }
  415. if (ident.flags & L2_TOK_SMALL) {
  416. l2_gen_stack_frame_set_copy(gen, ident.strbuf);
  417. } else {
  418. l2_gen_stack_frame_set(gen, &ident.str);
  419. }
  420. } else if (
  421. l2_token_get_kind(tok) == L2_TOK_IDENT &&
  422. l2_token_get_kind(tok2) == L2_TOK_EQUALS) {
  423. l2_trace_scope("replacement assign expression");
  424. l2_trace("ident '%s'", tok->v.str);
  425. struct l2_token_value ident = l2_token_extract_val(tok);
  426. l2_lexer_consume(lexer); // ident
  427. l2_lexer_consume(lexer); // =
  428. if (parse_expression(lexer, gen, err) < 0) {
  429. if (!(ident.flags & L2_TOK_SMALL)) free(ident.str);
  430. return -1;
  431. }
  432. if (ident.flags & L2_TOK_SMALL) {
  433. l2_gen_stack_frame_replace_copy(gen, ident.strbuf);
  434. } else {
  435. l2_gen_stack_frame_replace(gen, &ident.str);
  436. }
  437. } else {
  438. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  439. return -1;
  440. }
  441. if (!tok_is_end(l2_lexer_peek(lexer, 1))) {
  442. if (parse_func_call_after_base(lexer, gen, err) < 0) {
  443. return -1;
  444. }
  445. }
  446. }
  447. return 0;
  448. }
  449. int l2_parse_program(
  450. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  451. l2_trace_scope("program");
  452. while (1) {
  453. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  454. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_EOF) {
  455. break;
  456. }
  457. if (parse_expression(lexer, gen, err) < 0) {
  458. l2_gen_halt(gen);
  459. l2_gen_flush(gen);
  460. return -1;
  461. }
  462. l2_gen_discard(gen);
  463. }
  464. l2_gen_halt(gen);
  465. l2_gen_flush(gen);
  466. return 0;
  467. }