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 14KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556
  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. }
  28. static int parse_expression(
  29. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err);
  30. static int parse_arg_level_expression(
  31. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err);
  32. static int parse_object_literal(
  33. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  34. l2_trace_scope("object literal");
  35. // '{' and EOL already skipped by parse_object_or_function_literal
  36. l2_gen_namespace(gen);
  37. while (1) {
  38. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  39. if (l2_token_get_kind(tok) == L2_TOK_CLOSE_BRACE) {
  40. l2_lexer_consume(lexer); // '}'
  41. break;
  42. } else if (l2_token_get_kind(tok) != L2_TOK_IDENT) {
  43. l2_parse_err(err, tok, "In object literal: Expected identifier, got %s",
  44. l2_token_get_name(tok));
  45. return -1;
  46. }
  47. l2_trace("key: '%s'", tok->v.str);
  48. struct l2_token_value key = l2_token_extract_val(tok);
  49. l2_lexer_consume(lexer); // ident
  50. tok = l2_lexer_peek(lexer, 1);
  51. if (l2_token_get_kind(tok) != L2_TOK_COLON) {
  52. if (!(key.flags & L2_TOK_SMALL)) free(key.str);
  53. l2_parse_err(err, tok, "In object literal: Expected ':', got %s",
  54. l2_token_get_name(tok));
  55. return -1;
  56. }
  57. l2_lexer_consume(lexer); // ':'
  58. if (parse_expression(lexer, gen, err) < 0) {
  59. if (!(key.flags & L2_TOK_SMALL)) free(key.str);
  60. return -1;
  61. }
  62. if (key.flags & L2_TOK_SMALL) {
  63. l2_gen_namespace_set_copy(gen, key.strbuf);
  64. } else {
  65. l2_gen_namespace_set_copy(gen, key.str);
  66. }
  67. l2_gen_discard(gen);
  68. tok = l2_lexer_peek(lexer, 1);
  69. if (
  70. l2_token_get_kind(tok) != L2_TOK_EOL &&
  71. l2_token_get_kind(tok) != L2_TOK_CLOSE_BRACE) {
  72. l2_parse_err(err, tok, "In object literal: Expected EOL or '}', got %s",
  73. l2_token_get_name(tok));
  74. return -1;
  75. }
  76. if (l2_token_get_kind(tok) == L2_TOK_EOL) {
  77. l2_lexer_consume(lexer); // EOL
  78. }
  79. }
  80. return 0;
  81. }
  82. static int parse_function_literal_impl(
  83. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  84. l2_trace_scope("function literal");
  85. // '{' and EOL already skipped by parse_object_or_function_literal
  86. // The arguments array will be at the top of the stack
  87. char *ident = "$";
  88. l2_gen_stack_frame_set_copy(gen, ident);
  89. int first = 1;
  90. while (1) {
  91. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_CLOSE_BRACE) {
  92. l2_lexer_consume(lexer); // '}'
  93. break;
  94. }
  95. if (!first) {
  96. l2_gen_discard(gen);
  97. }
  98. l2_trace_scope("function literal expression");
  99. if (parse_expression(lexer, gen, err) < 0) {
  100. return -1;
  101. }
  102. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  103. first = 0;
  104. }
  105. // All functions must put _something_ on the stack
  106. if (first) {
  107. l2_gen_none(gen);
  108. }
  109. l2_gen_ret(gen);
  110. return 0;
  111. }
  112. static int parse_function_literal(
  113. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  114. l2_gen_flush(gen);
  115. struct l2_io_writer *prev_writer = gen->writer.w;
  116. // Generate the function to a buffer in memory
  117. struct l2_io_mem_writer w = {0};
  118. w.w.write = l2_io_mem_write;
  119. gen->writer.w = &w.w;
  120. // Generates two words; RJMP, 0
  121. l2_gen_rjmp(gen, 0);
  122. l2_word pos = gen->pos;
  123. // Generate the function body itself
  124. int ret = parse_function_literal_impl(lexer, gen, err);
  125. l2_gen_flush(gen);
  126. gen->writer.w = prev_writer;
  127. if (ret < 0) {
  128. free(w.mem);
  129. return -1;
  130. }
  131. l2_word *ops = w.mem;
  132. l2_word opcount = w.len / sizeof(l2_word);
  133. // Due to the earlier gen_rjmp, the second word will be the argument to RJMP.
  134. // Need to set it properly to skip the function body.
  135. // The '- 2' is because we don't skip the RJMP, <count> sequence.
  136. ops[1] = opcount - 2;
  137. l2_bufio_put_n(&gen->writer, ops, opcount * sizeof(l2_word));
  138. free(w.mem);
  139. l2_gen_function(gen, pos);
  140. return 0;
  141. }
  142. static int parse_object_or_function_literal(
  143. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  144. l2_trace_scope("object or function literal");
  145. l2_lexer_consume(lexer); // '{'
  146. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  147. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  148. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  149. if (l2_token_get_kind(tok) == L2_TOK_CLOSE_BRACE) {
  150. l2_trace_scope("empty object literal");
  151. l2_lexer_consume(lexer); // '}'
  152. l2_gen_namespace(gen);
  153. } else if (
  154. l2_token_get_kind(tok) == L2_TOK_IDENT &&
  155. l2_token_get_kind(tok2) == L2_TOK_COLON) {
  156. if (parse_object_literal(lexer, gen, err) < 0) {
  157. return -1;
  158. }
  159. } else {
  160. if (parse_function_literal(lexer, gen, err) < 0) {
  161. return -1;
  162. }
  163. }
  164. return 0;
  165. }
  166. static int parse_array_literal(
  167. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  168. l2_trace_scope("array literal");
  169. l2_lexer_consume(lexer); // '['
  170. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  171. int count = 0;
  172. while (1) {
  173. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_CLOSE_BRACKET) {
  174. l2_lexer_consume(lexer); // ']'
  175. break;
  176. }
  177. count += 1;
  178. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  179. return -1;
  180. }
  181. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  182. }
  183. l2_gen_array(gen, count);
  184. return 0;
  185. }
  186. static int parse_arg_level_expression_base(
  187. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  188. l2_trace_scope("arg level expression base");
  189. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  190. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  191. if (l2_token_get_kind(tok) == L2_TOK_OPEN_PAREN) {
  192. l2_trace_scope("group expr");
  193. l2_lexer_consume(lexer); // '('
  194. if (parse_expression(lexer, gen, err) < 0) {
  195. return -1;
  196. }
  197. tok = l2_lexer_peek(lexer, 1);
  198. if (l2_token_get_kind(tok) != L2_TOK_CLOSE_PAREN) {
  199. l2_parse_err(err, tok, "Expected '(', got %s",
  200. l2_token_get_name(tok));
  201. return -1;
  202. }
  203. l2_lexer_consume(lexer); // ')'
  204. } else if (l2_token_get_kind(tok) == L2_TOK_IDENT) {
  205. l2_trace_scope("ident");
  206. l2_trace("ident '%s'", tok->v.str);
  207. struct l2_token_value ident = l2_token_extract_val(tok);
  208. l2_lexer_consume(lexer); // ident
  209. if (ident.flags & L2_TOK_SMALL) {
  210. l2_gen_stack_frame_lookup_copy(gen, ident.strbuf);
  211. } else {
  212. l2_gen_stack_frame_lookup(gen, &ident.str);
  213. }
  214. } else if (l2_token_get_kind(tok) == L2_TOK_NUMBER) {
  215. l2_trace_scope("number literal");
  216. l2_trace("number %g", tok->v.num);
  217. double number = tok->v.num;
  218. l2_lexer_consume(lexer); // number
  219. l2_gen_number(gen, number);
  220. } else if (l2_token_get_kind(tok) == L2_TOK_STRING) {
  221. l2_trace_scope("string literal");
  222. l2_trace("string '%s'", tok->v.str);
  223. struct l2_token_value str = l2_token_extract_val(tok);
  224. l2_lexer_consume(lexer); // string
  225. if (str.flags & L2_TOK_SMALL) {
  226. l2_gen_string_copy(gen, str.strbuf);
  227. } else {
  228. l2_gen_string(gen, &str.str);
  229. }
  230. } else if (
  231. l2_token_get_kind(tok) == L2_TOK_QUOT &&
  232. l2_token_get_kind(tok2) == L2_TOK_IDENT) {
  233. l2_trace_scope("atom literal");
  234. l2_trace("atom '%s'", tok->v.str);
  235. struct l2_token_value ident = l2_token_extract_val(tok2);
  236. l2_lexer_consume(lexer); // "'"
  237. l2_lexer_consume(lexer); // ident
  238. if (ident.flags & L2_TOK_SMALL) {
  239. l2_gen_atom_copy(gen, ident.strbuf);
  240. } else {
  241. l2_gen_atom(gen, &ident.str);
  242. }
  243. } else if (l2_token_get_kind(tok) == L2_TOK_OPEN_BRACE) {
  244. if (parse_object_or_function_literal(lexer, gen, err) < 0) {
  245. return -1;
  246. }
  247. } else if (l2_token_get_kind(tok) == L2_TOK_OPEN_BRACKET) {
  248. if (parse_array_literal(lexer, gen, err) < 0) {
  249. return -1;
  250. }
  251. } else {
  252. l2_parse_err(err, tok, "Unexpected token %s",
  253. l2_token_get_name(tok));
  254. return -1;
  255. }
  256. return 0;
  257. }
  258. static int parse_arg_level_expression(
  259. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  260. l2_trace_scope("arg level expression");
  261. if (parse_arg_level_expression_base(lexer, gen, err) < 0) {
  262. return -1;
  263. }
  264. while (1) {
  265. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  266. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  267. struct l2_token *tok3 = l2_lexer_peek(lexer, 3);
  268. if (
  269. l2_token_get_kind(tok) == L2_TOK_OPEN_PAREN &&
  270. l2_token_get_kind(tok2) == L2_TOK_CLOSE_PAREN) {
  271. l2_trace_scope("niladic func call");
  272. l2_lexer_consume(lexer); // '('
  273. l2_lexer_consume(lexer); // ')'
  274. l2_gen_func_call(gen, 0);
  275. } else if (
  276. l2_token_get_kind(tok) == L2_TOK_PERIOD &&
  277. l2_token_get_kind(tok2) == L2_TOK_IDENT &&
  278. l2_token_get_kind(tok3) == L2_TOK_EQUALS) {
  279. l2_trace_scope("namespace assign");
  280. l2_trace("ident '%s'", tok2->v.str);
  281. struct l2_token_value ident = l2_token_extract_val(tok2);
  282. l2_lexer_consume(lexer); // '.'
  283. l2_lexer_consume(lexer); // ident
  284. l2_lexer_consume(lexer); // '='
  285. if (parse_expression(lexer, gen, err) < 0) {
  286. if (!(ident.flags & L2_TOK_SMALL)) free(ident.str);
  287. return -1;
  288. }
  289. if (ident.flags & L2_TOK_SMALL) {
  290. l2_gen_namespace_set_copy(gen, ident.strbuf);
  291. } else {
  292. l2_gen_namespace_set(gen, &ident.str);
  293. }
  294. l2_gen_swap_discard(gen);
  295. } else if (
  296. l2_token_get_kind(tok) == L2_TOK_PERIOD &&
  297. l2_token_get_kind(tok2) == L2_TOK_IDENT) {
  298. l2_trace_scope("namespace lookup");
  299. l2_trace("ident '%s'", tok2->v.str);
  300. struct l2_token_value ident = l2_token_extract_val(tok2);
  301. l2_lexer_consume(lexer); // '.'
  302. l2_lexer_consume(lexer); // ident
  303. if (ident.flags & L2_TOK_SMALL) {
  304. l2_gen_namespace_lookup_copy(gen, ident.strbuf);
  305. } else {
  306. l2_gen_namespace_lookup(gen, &ident.str);
  307. }
  308. } else if (
  309. l2_token_get_kind(tok) == L2_TOK_DOT_NUMBER &&
  310. l2_token_get_kind(tok2) == L2_TOK_EQUALS) {
  311. l2_trace_scope("direct array assign");
  312. int number = tok->v.integer;
  313. l2_lexer_consume(lexer); // dot-number
  314. l2_lexer_consume(lexer); // '='
  315. if (parse_expression(lexer, gen, err) < 0) {
  316. return -1;
  317. }
  318. l2_gen_array_set(gen, number);
  319. l2_gen_swap_discard(gen);
  320. } else if (l2_token_get_kind(tok) == L2_TOK_DOT_NUMBER) {
  321. l2_trace_scope("direct array lookup");
  322. int number = tok->v.integer;
  323. l2_lexer_consume(lexer); // dot-number
  324. l2_gen_array_lookup(gen, number);
  325. } else if (
  326. l2_token_get_kind(tok) == L2_TOK_PERIOD &&
  327. l2_token_get_kind(tok2) == L2_TOK_OPEN_PAREN) {
  328. l2_trace_scope("dynamic lookup");
  329. l2_lexer_consume(lexer); // '.'
  330. l2_lexer_consume(lexer); // '('
  331. if (parse_expression(lexer, gen, err) < 0) {
  332. return -1;
  333. }
  334. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) != L2_TOK_CLOSE_PAREN) {
  335. l2_parse_err(err, tok, "Expected '(', got %s",
  336. l2_token_get_name(tok));
  337. return -1;
  338. }
  339. l2_lexer_consume(lexer); // ')'
  340. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_EQUALS) {
  341. l2_lexer_consume(lexer); // '='
  342. if (parse_expression(lexer, gen, err) < 0) {
  343. return -1;
  344. }
  345. l2_gen_dynamic_set(gen);
  346. } else {
  347. l2_gen_dynamic_lookup(gen);
  348. }
  349. } else {
  350. break;
  351. }
  352. }
  353. return 0;
  354. }
  355. static int parse_func_call_after_base(
  356. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  357. l2_trace_scope("func call after base");
  358. size_t argc = 0;
  359. do {
  360. if (tok_is_infix(l2_lexer_peek(lexer, 1))) {
  361. do {
  362. // We already have one value (the lhs) on the stack,
  363. // so we need to parse the operator, then the rhs
  364. // Operator
  365. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  366. return -1;
  367. }
  368. // RHS
  369. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  370. return -1;
  371. }
  372. l2_gen_func_call_infix(gen);
  373. } while (tok_is_infix(l2_lexer_peek(lexer, 1)));
  374. // If this was the "first argument", this wasn't a function call
  375. // after all, it was just a (series of?) infix calls.
  376. if (argc == 0) {
  377. return 0;
  378. }
  379. // Don't increment argc here, because after an infix, we have
  380. // neither added nor removed an arguemnt, just transformed one
  381. } else {
  382. l2_trace_scope("func call param");
  383. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  384. return -1;
  385. }
  386. argc += 1;
  387. }
  388. } while (!tok_is_end(l2_lexer_peek(lexer, 1)));
  389. // The 'argc' previous expressions were arguments, the one before that was the function
  390. l2_gen_func_call(gen, argc);
  391. return 0;
  392. }
  393. static int parse_expression(
  394. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  395. l2_trace_scope("expression");
  396. struct l2_token *tok = l2_lexer_peek(lexer, 1);
  397. struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
  398. if (
  399. l2_token_get_kind(tok) == L2_TOK_IDENT &&
  400. l2_token_get_kind(tok2) == L2_TOK_COLON_EQ) {
  401. l2_trace_scope("assign expression");
  402. l2_trace("ident '%s'", tok->v.str);
  403. struct l2_token_value ident = l2_token_extract_val(tok);
  404. l2_lexer_consume(lexer); // ident
  405. l2_lexer_consume(lexer); // :=
  406. if (parse_expression(lexer, gen, err) < 0) {
  407. if (!(ident.flags & L2_TOK_SMALL)) free(ident.str);
  408. return -1;
  409. }
  410. if (ident.flags & L2_TOK_SMALL) {
  411. l2_gen_stack_frame_set_copy(gen, ident.strbuf);
  412. } else {
  413. l2_gen_stack_frame_set(gen, &ident.str);
  414. }
  415. } else if (
  416. l2_token_get_kind(tok) == L2_TOK_IDENT &&
  417. l2_token_get_kind(tok2) == L2_TOK_EQUALS) {
  418. l2_trace_scope("replacement assign expression");
  419. l2_trace("ident '%s'", tok->v.str);
  420. struct l2_token_value ident = l2_token_extract_val(tok);
  421. l2_lexer_consume(lexer); // ident
  422. l2_lexer_consume(lexer); // =
  423. if (parse_expression(lexer, gen, err) < 0) {
  424. if (!(ident.flags & L2_TOK_SMALL)) free(ident.str);
  425. return -1;
  426. }
  427. if (ident.flags & L2_TOK_SMALL) {
  428. l2_gen_stack_frame_replace_copy(gen, ident.strbuf);
  429. } else {
  430. l2_gen_stack_frame_replace(gen, &ident.str);
  431. }
  432. } else {
  433. if (parse_arg_level_expression(lexer, gen, err) < 0) {
  434. return -1;
  435. }
  436. if (!tok_is_end(l2_lexer_peek(lexer, 1))) {
  437. if (parse_func_call_after_base(lexer, gen, err) < 0) {
  438. return -1;
  439. }
  440. }
  441. }
  442. return 0;
  443. }
  444. int l2_parse_program(
  445. struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
  446. l2_trace_scope("program");
  447. while (1) {
  448. l2_lexer_skip_opt(lexer, L2_TOK_EOL);
  449. if (l2_token_get_kind(l2_lexer_peek(lexer, 1)) == L2_TOK_EOF) {
  450. break;
  451. }
  452. if (parse_expression(lexer, gen, err) < 0) {
  453. l2_gen_halt(gen);
  454. l2_gen_flush(gen);
  455. return -1;
  456. }
  457. l2_gen_discard(gen);
  458. }
  459. l2_gen_halt(gen);
  460. l2_gen_flush(gen);
  461. return 0;
  462. }