gen/
.clang-format
570 B
.gitignore
30 B
.gitsigners
112 B
LICENSE
1.1 KiB
Makefile
911 B
README
1.8 KiB
ast.c
5.0 KiB
ast.h
15.1 KiB
desugar.c
23.1 KiB
desugar.h
286 B
gen.c
108.5 KiB
gen.h
4.9 KiB
io.c
1.1 KiB
io.h
444 B
limits.h
1.3 KiB
module.c
10.0 KiB
module.h
2.2 KiB
options.c
1.4 KiB
options.h
472 B
parser.c
68.3 KiB
parser.h
942 B
radiance.c
3.7 KiB
ralloc.c
2.0 KiB
ralloc.h
1.1 KiB
resolver.c
109.7 KiB
resolver.h
5.6 KiB
riscv.c
12.0 KiB
riscv.h
12.0 KiB
scanner.c
10.2 KiB
scanner.h
3.2 KiB
strings.c
2.6 KiB
strings.h
407 B
symtab.c
5.7 KiB
symtab.h
4.6 KiB
types.h
1.0 KiB
util.h
1.5 KiB
parser.c
raw
| 1 | #include <stdarg.h> |
| 2 | #include <stdio.h> |
| 3 | #include <stdlib.h> |
| 4 | #include <string.h> |
| 5 | |
| 6 | #include "ast.h" |
| 7 | #include "io.h" |
| 8 | #include "limits.h" |
| 9 | #include "parser.h" |
| 10 | #include "scanner.h" |
| 11 | #include "strings.h" |
| 12 | |
| 13 | #define error(...) __error(__VA_ARGS__, NULL) |
| 14 | |
| 15 | static node_t *parse_expr(parser_t *p); |
| 16 | static node_t *parse_stmt_or_block(parser_t *p); |
| 17 | static node_t *parse_cond(parser_t *p); |
| 18 | static node_t *parse_if(parser_t *p); |
| 19 | static node_t *parse_if_let(parser_t *p); |
| 20 | static node_t *parse_if_case(parser_t *p); |
| 21 | static node_t *parse_block(parser_t *p); |
| 22 | static node_t *parse_stmt(parser_t *p); |
| 23 | static node_t *parse_type(parser_t *p); |
| 24 | static node_t *parse_union(parser_t *p, node_t *attrs); |
| 25 | static node_t *parse_record(parser_t *p, node_t *attrs); |
| 26 | static node_t *parse_record_type(parser_t *p); |
| 27 | static node_t *parse_record_lit(parser_t *p, node_t *type_name); |
| 28 | static node_t *parse_postfix(parser_t *p, node_t *expr); |
| 29 | static node_t *parse_as_cast(parser_t *p, node_t *expr); |
| 30 | static node_t *parse_name_type_value(parser_t *p, nodeclass_t cls); |
| 31 | static node_t *parse_static(parser_t *p); |
| 32 | static node_t *parse_ident(parser_t *p, const char *error); |
| 33 | static node_t *parse_ident_or_placeholder(parser_t *p, const char *error); |
| 34 | static node_t *parse_scope_segment(parser_t *p, const char *error); |
| 35 | static node_t *parse_label(parser_t *p, const char *error); |
| 36 | static node_t *parse_assignment(parser_t *p, node_t *lval); |
| 37 | static node_t *parse_fn_call_arg(parser_t *p); |
| 38 | static node_t *parse_match(parser_t *p); |
| 39 | static node_t *parse_match_case(parser_t *p); |
| 40 | static node_t *parse_builtin(parser_t *p); |
| 41 | static node_t *parse_throw(parser_t *p); |
| 42 | static node_t *parse_try(parser_t *p, bool panic, bool optional); |
| 43 | static node_t *parse_panic(parser_t *p); |
| 44 | static bool token_is_stmt_terminator(tokenclass_t cls); |
| 45 | static bool stmt_requires_semicolon(const node_t *stmt); |
| 46 | static bool consume_statement_separator( |
| 47 | parser_t *p, node_t *stmt, bool require |
| 48 | ); |
| 49 | |
| 50 | /* Initialize parser. */ |
| 51 | void parser_init(parser_t *p) { |
| 52 | p->root = NULL; |
| 53 | p->errors = 0; |
| 54 | p->nnodes = 0; |
| 55 | p->nptrs = 0; |
| 56 | p->context = PARSE_CTX_NORMAL; |
| 57 | } |
| 58 | |
| 59 | /* Report an error with optional format string. */ |
| 60 | static void __error(parser_t *p, const char *fmt, ...) { |
| 61 | va_list ap; |
| 62 | va_start(ap, fmt); |
| 63 | |
| 64 | location_t loc = scanner_get_location(&p->scanner, p->current.position); |
| 65 | fprintf(stderr, "%s:%u:%u: error: ", loc.file, loc.line, loc.col); |
| 66 | vfprintf(stderr, fmt, ap); |
| 67 | fprintf(stderr, "\n"); |
| 68 | va_end(ap); |
| 69 | |
| 70 | p->errors++; |
| 71 | } |
| 72 | |
| 73 | /* Check that the current token is equal to the given type. */ |
| 74 | static bool check(parser_t *p, tokenclass_t cls) { |
| 75 | return p->current.cls == cls; |
| 76 | } |
| 77 | |
| 78 | /* Advance the parser by one token. */ |
| 79 | static void advance(parser_t *p) { |
| 80 | p->previous = p->current; |
| 81 | p->current = scanner_next(&p->scanner); |
| 82 | } |
| 83 | |
| 84 | /* Like `check`, but also advances the parser if it matches. */ |
| 85 | static bool consume(parser_t *p, tokenclass_t cls) { |
| 86 | if (check(p, cls)) { |
| 87 | advance(p); |
| 88 | return true; |
| 89 | } |
| 90 | return false; |
| 91 | } |
| 92 | |
| 93 | /* Like `consume`, but report an error if it doesn't match. */ |
| 94 | __nodiscard static bool expect( |
| 95 | parser_t *p, tokenclass_t cls, const char *message |
| 96 | ) { |
| 97 | if (consume(p, cls)) { |
| 98 | return true; |
| 99 | } |
| 100 | error(p, message); |
| 101 | |
| 102 | return false; |
| 103 | } |
| 104 | |
| 105 | /* Allocate a new AST node. */ |
| 106 | static node_t *node(parser_t *p, nodeclass_t cls) { |
| 107 | if (p->nnodes >= MAX_NODES) { |
| 108 | abort(); |
| 109 | } |
| 110 | node_t *n = &p->nodes[p->nnodes++]; |
| 111 | n->cls = cls; |
| 112 | n->type = NULL; |
| 113 | n->sym = NULL; |
| 114 | n->offset = p->current.position; |
| 115 | n->length = p->current.length; |
| 116 | n->file = p->scanner.file; |
| 117 | |
| 118 | return n; |
| 119 | } |
| 120 | |
| 121 | /* Parse a type annotation. |
| 122 | * Eg. `i32` or `[i32; 12]` */ |
| 123 | static node_t *parse_type(parser_t *p) { |
| 124 | /* Parse optional types. */ |
| 125 | if (p->current.cls == T_QUESTION) { |
| 126 | node_t *opt = node(p, NODE_TYPE); |
| 127 | advance(p); /* Consume `?`. */ |
| 128 | |
| 129 | node_t *elem_type = parse_type(p); |
| 130 | if (!elem_type) |
| 131 | return NULL; |
| 132 | |
| 133 | opt->val.type.tclass = TYPE_OPT; |
| 134 | opt->val.type.elem_type = elem_type; |
| 135 | |
| 136 | return opt; |
| 137 | } |
| 138 | |
| 139 | /* Parse pointer types and slice types. */ |
| 140 | if (p->current.cls == T_STAR) { |
| 141 | advance(p); /* Consume `*`. */ |
| 142 | |
| 143 | /* Consume `mut` */ |
| 144 | bool mut = consume(p, T_MUT); |
| 145 | |
| 146 | /* Parse slice types like `*[i32]` or `*mut [i32]` */ |
| 147 | if (p->current.cls == T_LBRACKET) { |
| 148 | node_t *slice = node(p, NODE_TYPE); |
| 149 | advance(p); /* Consume `[`. */ |
| 150 | |
| 151 | node_t *elem_type = parse_type(p); |
| 152 | if (!elem_type) |
| 153 | return NULL; |
| 154 | |
| 155 | if (!expect(p, T_RBRACKET, "expected `]` after slice element type")) |
| 156 | return NULL; |
| 157 | |
| 158 | slice->val.type.tclass = TYPE_SLICE; |
| 159 | slice->val.type.elem_type = elem_type; |
| 160 | slice->val.type.info.slice.mut = mut; |
| 161 | |
| 162 | return slice; |
| 163 | } |
| 164 | |
| 165 | /* Otherwise it's a pointer type like `*i32` or `*mut i32` */ |
| 166 | node_t *ptr = node(p, NODE_TYPE); |
| 167 | node_t *elem_type = parse_type(p); |
| 168 | if (!elem_type) |
| 169 | return NULL; |
| 170 | |
| 171 | ptr->val.type.tclass = TYPE_PTR; |
| 172 | ptr->val.type.elem_type = elem_type; |
| 173 | ptr->val.type.info.ptr.mut = mut; |
| 174 | |
| 175 | return ptr; |
| 176 | } |
| 177 | |
| 178 | /* Parse array types. */ |
| 179 | if (p->current.cls == T_LBRACKET) { |
| 180 | advance(p); /* Consume `[`. */ |
| 181 | |
| 182 | /* Get the element type. */ |
| 183 | node_t *elem_type = parse_type(p); |
| 184 | if (!elem_type) |
| 185 | return NULL; |
| 186 | |
| 187 | /* Expect a semicolon separator. */ |
| 188 | if (!expect(p, T_SEMICOLON, "expected `;` in array type")) |
| 189 | return NULL; |
| 190 | |
| 191 | /* Parse the array length. */ |
| 192 | node_t *length = parse_expr(p); |
| 193 | if (!length) { |
| 194 | error(p, "expected array size expression"); |
| 195 | return NULL; |
| 196 | } |
| 197 | /* Expect the closing bracket */ |
| 198 | if (!expect(p, T_RBRACKET, "expected `]` after array size")) |
| 199 | return NULL; |
| 200 | |
| 201 | node_t *ary = node(p, NODE_TYPE); |
| 202 | ary->val.type.tclass = TYPE_ARRAY; |
| 203 | ary->val.type.elem_type = elem_type; |
| 204 | ary->val.type.info.array.length = length; |
| 205 | |
| 206 | return ary; |
| 207 | } |
| 208 | |
| 209 | /* Type identifiers are treated differently, as a concrete type cannot |
| 210 | * yet be assigned. */ |
| 211 | if (p->current.cls == T_IDENT || p->current.cls == T_SUPER) { |
| 212 | node_t *path = |
| 213 | parse_scope_segment(p, "expected type identifier or `super`"); |
| 214 | if (!path) |
| 215 | return NULL; |
| 216 | |
| 217 | while (consume(p, T_COLON_COLON)) { |
| 218 | node_t *next = |
| 219 | parse_scope_segment(p, "expected identifier name after `::`"); |
| 220 | if (!next) |
| 221 | return NULL; |
| 222 | |
| 223 | node_t *scope = node(p, NODE_SCOPE); |
| 224 | scope->val.access.lval = path; |
| 225 | scope->val.access.rval = next; |
| 226 | path = scope; |
| 227 | } |
| 228 | return path; |
| 229 | } |
| 230 | node_t *n = node(p, NODE_TYPE); |
| 231 | |
| 232 | switch (p->current.cls) { |
| 233 | case T_I8: |
| 234 | advance(p); |
| 235 | n->val.type.tclass = TYPE_I8; |
| 236 | return n; |
| 237 | case T_I16: |
| 238 | advance(p); |
| 239 | n->val.type.tclass = TYPE_I16; |
| 240 | return n; |
| 241 | case T_I32: |
| 242 | advance(p); |
| 243 | n->val.type.tclass = TYPE_I32; |
| 244 | return n; |
| 245 | case T_U8: |
| 246 | advance(p); |
| 247 | n->val.type.tclass = TYPE_U8; |
| 248 | return n; |
| 249 | case T_U16: |
| 250 | advance(p); |
| 251 | n->val.type.tclass = TYPE_U16; |
| 252 | return n; |
| 253 | case T_U32: |
| 254 | advance(p); |
| 255 | n->val.type.tclass = TYPE_U32; |
| 256 | return n; |
| 257 | case T_BOOL: |
| 258 | advance(p); |
| 259 | n->val.type.tclass = TYPE_BOOL; |
| 260 | return n; |
| 261 | case T_VOID: |
| 262 | advance(p); |
| 263 | n->val.type.tclass = TYPE_VOID; |
| 264 | return n; |
| 265 | case T_OPAQUE: |
| 266 | advance(p); |
| 267 | n->val.type.tclass = TYPE_OPAQUE; |
| 268 | return n; |
| 269 | case T_FN: { |
| 270 | advance(p); /* consume `fn` */ |
| 271 | |
| 272 | if (!expect(p, T_LPAREN, "expected `(` after `fn`")) |
| 273 | return NULL; |
| 274 | |
| 275 | n->val.type.tclass = TYPE_FN; |
| 276 | n->val.type.info.fn.params = nodespan_alloc(p, MAX_FN_PARAMS); |
| 277 | n->val.type.info.fn.ret = NULL; |
| 278 | n->val.type.info.fn.throws = nodespan_alloc(p, MAX_FN_THROWS); |
| 279 | |
| 280 | /* Parse parameter types */ |
| 281 | if (!check(p, T_RPAREN)) { |
| 282 | node_t *param = NULL; |
| 283 | |
| 284 | do { |
| 285 | if (n->val.type.info.fn.params.len >= MAX_FN_PARAMS) { |
| 286 | error(p, "too many function pointer parameters"); |
| 287 | return NULL; |
| 288 | } |
| 289 | if (!(param = parse_type(p))) { |
| 290 | return NULL; |
| 291 | } |
| 292 | nodespan_push(p, &n->val.type.info.fn.params, param); |
| 293 | } while (consume(p, T_COMMA)); |
| 294 | } |
| 295 | if (!expect( |
| 296 | p, T_RPAREN, "expected `)` after function pointer parameters" |
| 297 | )) |
| 298 | return NULL; |
| 299 | |
| 300 | /* Parse return type */ |
| 301 | if (consume(p, T_ARROW)) { |
| 302 | if (!(n->val.type.info.fn.ret = parse_type(p))) { |
| 303 | return NULL; |
| 304 | } |
| 305 | } |
| 306 | |
| 307 | if (consume(p, T_THROWS)) { |
| 308 | if (!expect(p, T_LPAREN, "expected `(` after `throws`")) |
| 309 | return NULL; |
| 310 | |
| 311 | if (!check(p, T_RPAREN)) { |
| 312 | do { |
| 313 | if (n->val.type.info.fn.throws.len >= MAX_FN_THROWS) { |
| 314 | error(p, "maximum number of thrown types exceeded"); |
| 315 | return NULL; |
| 316 | } |
| 317 | |
| 318 | node_t *thrown = parse_type(p); |
| 319 | if (!thrown) |
| 320 | return NULL; |
| 321 | |
| 322 | nodespan_push(p, &n->val.type.info.fn.throws, thrown); |
| 323 | } while (consume(p, T_COMMA)); |
| 324 | } |
| 325 | |
| 326 | if (!expect(p, T_RPAREN, "expected `)` after throws clause")) |
| 327 | return NULL; |
| 328 | } |
| 329 | return n; |
| 330 | } |
| 331 | default: |
| 332 | error(p, "expected type annotation, eg. `i32`, `bool`, etc."); |
| 333 | return NULL; |
| 334 | } |
| 335 | } |
| 336 | |
| 337 | /* Parse primary expressions. */ |
| 338 | static node_t *parse_array_literal(parser_t *p) { |
| 339 | node_t *n = NULL; |
| 340 | |
| 341 | if (check(p, T_RBRACKET)) { /* Empty array `[]` */ |
| 342 | n = node(p, NODE_ARRAY_LIT); |
| 343 | n->val.array_lit.elems = (nodespan_t){ 0 }; |
| 344 | } else { |
| 345 | node_t *expr = parse_expr(p); |
| 346 | if (!expr) |
| 347 | return NULL; |
| 348 | |
| 349 | /* Check if this is a repeat array [value; count] */ |
| 350 | if (consume(p, T_SEMICOLON)) { |
| 351 | n = node(p, NODE_ARRAY_REPEAT_LIT); |
| 352 | n->val.array_repeat_lit.value = expr; |
| 353 | n->val.array_repeat_lit.count = parse_expr(p); |
| 354 | |
| 355 | if (!n->val.array_repeat_lit.count) |
| 356 | return NULL; |
| 357 | } else { |
| 358 | /* Regular array literal [a, b, ...] */ |
| 359 | n = node(p, NODE_ARRAY_LIT); |
| 360 | n->val.array_lit.elems = (nodespan_t){ 0 }; |
| 361 | nodespan_push(p, &n->val.array_lit.elems, expr); |
| 362 | |
| 363 | /* Continue parsing remaining elements */ |
| 364 | while (consume(p, T_COMMA) && !check(p, T_RBRACKET)) { |
| 365 | node_t *elem = parse_expr(p); |
| 366 | if (!elem) |
| 367 | return NULL; |
| 368 | |
| 369 | nodespan_push(p, &n->val.array_lit.elems, elem); |
| 370 | } |
| 371 | } |
| 372 | } |
| 373 | if (!expect(p, T_RBRACKET, "expected `]` after array elements")) |
| 374 | return NULL; |
| 375 | |
| 376 | return n; |
| 377 | } |
| 378 | |
| 379 | static node_t *parse_builtin(parser_t *p) { |
| 380 | node_t *n = node(p, NODE_BUILTIN); |
| 381 | |
| 382 | /* Token is @identifier, skip the '@' to get the name. */ |
| 383 | const char *name = p->current.start + 1; |
| 384 | usize length = p->current.length - 1; |
| 385 | |
| 386 | advance(p); /* consume `@identifier` */ |
| 387 | |
| 388 | builtin_kind_t kind; |
| 389 | |
| 390 | if (!strncmp(name, "sizeOf", 6)) { |
| 391 | kind = BUILTIN_SIZE_OF; |
| 392 | } else if (!strncmp(name, "alignOf", 7)) { |
| 393 | kind = BUILTIN_ALIGN_OF; |
| 394 | } else if (!strncmp(name, "sliceOf", 7)) { |
| 395 | kind = BUILTIN_SLICE_OF; |
| 396 | } else { |
| 397 | error(p, "unknown builtin `@%.*s`", (int)length, name); |
| 398 | return NULL; |
| 399 | } |
| 400 | if (!expect(p, T_LPAREN, "expected `(` after builtin name")) |
| 401 | return NULL; |
| 402 | |
| 403 | n->val.builtin.kind = kind; |
| 404 | n->val.builtin.args = (nodespan_t){ 0 }; |
| 405 | |
| 406 | /* @sliceOf takes two expression arguments: @sliceOf(ptr, len) */ |
| 407 | if (kind == BUILTIN_SLICE_OF) { |
| 408 | parse_ctx_t prev = p->context; |
| 409 | p->context = PARSE_CTX_NORMAL; |
| 410 | |
| 411 | node_t *ptr_expr = parse_expr(p); |
| 412 | if (!ptr_expr) |
| 413 | return NULL; |
| 414 | nodespan_push(p, &n->val.builtin.args, ptr_expr); |
| 415 | |
| 416 | if (!expect( |
| 417 | p, T_COMMA, "expected `,` after first argument to @sliceOf" |
| 418 | )) |
| 419 | return NULL; |
| 420 | |
| 421 | node_t *len_expr = parse_expr(p); |
| 422 | if (!len_expr) |
| 423 | return NULL; |
| 424 | nodespan_push(p, &n->val.builtin.args, len_expr); |
| 425 | |
| 426 | p->context = prev; |
| 427 | } else { |
| 428 | /* @sizeOf and @alignOf take type arguments only. */ |
| 429 | node_t *type_arg = parse_type(p); |
| 430 | if (!type_arg) |
| 431 | return NULL; |
| 432 | nodespan_push(p, &n->val.builtin.args, type_arg); |
| 433 | } |
| 434 | |
| 435 | if (!expect(p, T_RPAREN, "expected `)` after builtin argument")) |
| 436 | return NULL; |
| 437 | |
| 438 | return n; |
| 439 | } |
| 440 | |
| 441 | static node_t *parse_primary(parser_t *p) { |
| 442 | node_t *n; |
| 443 | |
| 444 | switch (p->current.cls) { |
| 445 | case T_LBRACKET: /* Array literal [a, b, c] */ |
| 446 | advance(p); |
| 447 | return parse_array_literal(p); |
| 448 | |
| 449 | case T_NOT: /* Unary not operator */ |
| 450 | n = node(p, NODE_UNOP); |
| 451 | n->val.unop.op = OP_NOT; |
| 452 | advance(p); |
| 453 | |
| 454 | if (!(n->val.unop.expr = parse_primary(p))) |
| 455 | return NULL; |
| 456 | |
| 457 | return n; |
| 458 | |
| 459 | case T_RECORD: { |
| 460 | advance(p); /* consume `record` */ |
| 461 | |
| 462 | node_t *rtype = parse_record_type(p); |
| 463 | if (!rtype) |
| 464 | return NULL; |
| 465 | |
| 466 | if (p->context == PARSE_CTX_NORMAL && consume(p, T_LBRACE)) { |
| 467 | return parse_record_lit(p, rtype); |
| 468 | } |
| 469 | if (p->context == PARSE_CTX_NORMAL) { |
| 470 | error(p, "expected `{` after anonymous record type"); |
| 471 | return NULL; |
| 472 | } |
| 473 | return rtype; |
| 474 | } |
| 475 | |
| 476 | case T_LBRACE: |
| 477 | if (p->context == PARSE_CTX_CONDITION) { |
| 478 | error(p, "unexpected `{` in this context"); |
| 479 | return NULL; |
| 480 | } |
| 481 | advance(p); /* consume `{` */ |
| 482 | |
| 483 | return parse_record_lit(p, NULL); |
| 484 | |
| 485 | case T_MINUS: /* Unary negation operator */ |
| 486 | n = node(p, NODE_UNOP); |
| 487 | n->val.unop.op = OP_NEG; |
| 488 | advance(p); |
| 489 | |
| 490 | if (!(n->val.unop.expr = parse_primary(p))) |
| 491 | return NULL; |
| 492 | |
| 493 | return n; |
| 494 | |
| 495 | case T_TILDE: /* Bitwise NOT operator */ |
| 496 | n = node(p, NODE_UNOP); |
| 497 | n->val.unop.op = OP_BNOT; |
| 498 | advance(p); |
| 499 | |
| 500 | if (!(n->val.unop.expr = parse_primary(p))) |
| 501 | return NULL; |
| 502 | |
| 503 | return n; |
| 504 | |
| 505 | case T_AMP: |
| 506 | n = node(p, NODE_REF); |
| 507 | advance(p); |
| 508 | |
| 509 | n->val.ref.mut = consume(p, T_MUT); |
| 510 | |
| 511 | if (!(n->val.ref.target = parse_primary(p))) |
| 512 | return NULL; |
| 513 | |
| 514 | return n; |
| 515 | |
| 516 | case T_STAR: |
| 517 | n = node(p, NODE_UNOP); |
| 518 | advance(p); |
| 519 | |
| 520 | n->val.unop.op = OP_DEREF; |
| 521 | if (!(n->val.unop.expr = parse_primary(p))) |
| 522 | return NULL; |
| 523 | |
| 524 | return n; |
| 525 | |
| 526 | case T_NUMBER: |
| 527 | n = node(p, NODE_NUMBER); |
| 528 | advance(p); |
| 529 | |
| 530 | n->val.number.text = p->previous.start; |
| 531 | n->val.number.text_len = p->previous.length; |
| 532 | |
| 533 | if (check(p, T_DOT_DOT)) { |
| 534 | return parse_postfix(p, n); |
| 535 | } |
| 536 | return n; |
| 537 | |
| 538 | case T_CHAR: |
| 539 | n = node(p, NODE_CHAR); |
| 540 | advance(p); |
| 541 | |
| 542 | if (p->previous.start[1] == '\\') { |
| 543 | switch (p->previous.start[2]) { |
| 544 | case 'n': |
| 545 | n->val.char_lit = '\n'; |
| 546 | break; |
| 547 | case 't': |
| 548 | n->val.char_lit = '\t'; |
| 549 | break; |
| 550 | case 'r': |
| 551 | n->val.char_lit = '\r'; |
| 552 | break; |
| 553 | case '\'': |
| 554 | n->val.char_lit = '\''; |
| 555 | break; |
| 556 | case '\\': |
| 557 | n->val.char_lit = '\\'; |
| 558 | break; |
| 559 | default: |
| 560 | abort(); |
| 561 | } |
| 562 | } else { |
| 563 | n->val.char_lit = p->previous.start[1]; |
| 564 | } |
| 565 | if (check(p, T_DOT_DOT)) { |
| 566 | return parse_postfix(p, n); |
| 567 | } |
| 568 | return n; |
| 569 | |
| 570 | case T_STRING: { |
| 571 | n = node(p, NODE_STRING); |
| 572 | advance(p); |
| 573 | |
| 574 | /* Account for quotes. */ |
| 575 | const char *data = p->previous.start + 1; |
| 576 | usize len = p->previous.length - 2; |
| 577 | |
| 578 | /* Intern string. This escapes the string properly and |
| 579 | * NULL-terminates it. */ |
| 580 | n->val.string_lit.data = strings_alloc_len(data, len); |
| 581 | n->val.string_lit.length = strlen(n->val.string_lit.data); |
| 582 | |
| 583 | return n; |
| 584 | } |
| 585 | |
| 586 | case T_AT_IDENT: |
| 587 | return parse_builtin(p); |
| 588 | |
| 589 | case T_SUPER: |
| 590 | n = node(p, NODE_SUPER); |
| 591 | advance(p); |
| 592 | |
| 593 | if (check(p, T_COLON_COLON)) { |
| 594 | return parse_postfix(p, n); |
| 595 | } |
| 596 | return n; |
| 597 | |
| 598 | case T_IDENT: |
| 599 | n = node(p, NODE_IDENT); |
| 600 | n->val.ident.name = p->current.start; |
| 601 | n->val.ident.length = p->current.length; |
| 602 | |
| 603 | advance(p); |
| 604 | |
| 605 | /* Check for record initializer, eg. `{ x: 1, y: 2 }` */ |
| 606 | if (p->context == PARSE_CTX_NORMAL && consume(p, T_LBRACE)) { |
| 607 | return parse_record_lit(p, n); |
| 608 | } |
| 609 | |
| 610 | /* Check for field access or array indexing. */ |
| 611 | if (check(p, T_DOT) || check(p, T_LBRACKET) || |
| 612 | check(p, T_COLON_COLON) || check(p, T_LPAREN) || |
| 613 | check(p, T_DOT_DOT)) { |
| 614 | return parse_postfix(p, n); |
| 615 | } |
| 616 | return n; |
| 617 | |
| 618 | case T_LPAREN: |
| 619 | advance(p); |
| 620 | |
| 621 | /* Inside parentheses, we are in a normal parsing context */ |
| 622 | parse_ctx_t prev = p->context; |
| 623 | p->context = PARSE_CTX_NORMAL; |
| 624 | n = parse_expr(p); |
| 625 | p->context = prev; |
| 626 | |
| 627 | if (!expect(p, T_RPAREN, "expected closing `)` after expression")) |
| 628 | return NULL; |
| 629 | |
| 630 | /* Check for field access or array indexing. */ |
| 631 | if (check(p, T_DOT) || check(p, T_LBRACKET) || check(p, T_DOT_DOT)) { |
| 632 | return parse_postfix(p, n); |
| 633 | } |
| 634 | return n; |
| 635 | |
| 636 | case T_TRUE: |
| 637 | n = node(p, NODE_BOOL); |
| 638 | n->val.bool_lit = true; |
| 639 | advance(p); |
| 640 | |
| 641 | return n; |
| 642 | |
| 643 | case T_FALSE: |
| 644 | n = node(p, NODE_BOOL); |
| 645 | n->val.bool_lit = false; |
| 646 | advance(p); |
| 647 | |
| 648 | return n; |
| 649 | |
| 650 | case T_NIL: |
| 651 | n = node(p, NODE_NIL); |
| 652 | advance(p); |
| 653 | |
| 654 | return n; |
| 655 | |
| 656 | case T_UNDEF: |
| 657 | n = node(p, NODE_UNDEF); |
| 658 | advance(p); |
| 659 | |
| 660 | return n; |
| 661 | |
| 662 | case T_UNDERSCORE: |
| 663 | n = node(p, NODE_PLACEHOLDER); |
| 664 | advance(p); |
| 665 | |
| 666 | return n; |
| 667 | |
| 668 | case T_TRY: { |
| 669 | advance(p); |
| 670 | |
| 671 | bool panic = consume(p, T_BANG); |
| 672 | bool optional = consume(p, T_QUESTION); |
| 673 | |
| 674 | node_t *expr = parse_try(p, panic, optional); |
| 675 | if (!expr) |
| 676 | return NULL; |
| 677 | |
| 678 | if (check(p, T_DOT) || check(p, T_LBRACKET) || |
| 679 | check(p, T_COLON_COLON) || check(p, T_LPAREN) || |
| 680 | check(p, T_DOT_DOT)) { |
| 681 | return parse_postfix(p, expr); |
| 682 | } |
| 683 | return expr; |
| 684 | } |
| 685 | |
| 686 | default: |
| 687 | error( |
| 688 | p, |
| 689 | "expected expression, got `%.*s`", |
| 690 | p->current.length, |
| 691 | p->current.start |
| 692 | ); |
| 693 | return NULL; |
| 694 | } |
| 695 | } |
| 696 | |
| 697 | /* Parse binary expressions with precedence climbing. */ |
| 698 | static node_t *parse_binary(parser_t *p, node_t *left, int precedence) { |
| 699 | /* Operator precedence table. */ |
| 700 | static const struct { |
| 701 | tokenclass_t tok; |
| 702 | binop_t op; |
| 703 | int prec; |
| 704 | } ops[] = { |
| 705 | /* Arithmetic operators (higher precedence). */ |
| 706 | { T_PLUS, OP_ADD, 6 }, |
| 707 | { T_MINUS, OP_SUB, 6 }, |
| 708 | { T_STAR, OP_MUL, 7 }, |
| 709 | { T_SLASH, OP_DIV, 7 }, |
| 710 | { T_PERCENT, OP_MOD, 7 }, |
| 711 | /* Shift operators. */ |
| 712 | { T_LSHIFT, OP_SHL, 5 }, |
| 713 | { T_RSHIFT, OP_SHR, 5 }, |
| 714 | /* Bitwise operators. */ |
| 715 | { T_AMP, OP_BAND, 4 }, |
| 716 | { T_CARET, OP_XOR, 3 }, |
| 717 | { T_PIPE, OP_BOR, 2 }, |
| 718 | /* Comparison operators. */ |
| 719 | { T_EQ_EQ, OP_EQ, 1 }, |
| 720 | { T_BANG_EQ, OP_NE, 1 }, |
| 721 | { T_LT, OP_LT, 1 }, |
| 722 | { T_GT, OP_GT, 1 }, |
| 723 | { T_LT_EQ, OP_LE, 1 }, |
| 724 | { T_GT_EQ, OP_GE, 1 }, |
| 725 | /* Logical operators (lowest precedence). */ |
| 726 | { T_AND, OP_AND, 0 }, |
| 727 | { T_OR, OP_OR, 0 }, |
| 728 | }; |
| 729 | |
| 730 | for (;;) { |
| 731 | int next = -1; |
| 732 | binop_t op; |
| 733 | |
| 734 | /* Find matching operator and its precedence. */ |
| 735 | for (usize i = 0; i < sizeof(ops) / sizeof(ops[0]); i++) { |
| 736 | if (check(p, ops[i].tok) && ops[i].prec > precedence) { |
| 737 | if (next == -1 || ops[i].prec < next) { |
| 738 | next = ops[i].prec; |
| 739 | op = ops[i].op; |
| 740 | } |
| 741 | } |
| 742 | } |
| 743 | if (next == -1) |
| 744 | break; |
| 745 | |
| 746 | /* Consume the operator token. */ |
| 747 | advance(p); |
| 748 | |
| 749 | /* Parse the right operand. */ |
| 750 | node_t *right = parse_primary(p); |
| 751 | |
| 752 | if (!right) |
| 753 | return NULL; |
| 754 | |
| 755 | /* Handle `as` casts on the right operand */ |
| 756 | while (check(p, T_AS)) { |
| 757 | right = parse_as_cast(p, right); |
| 758 | if (!right) |
| 759 | return NULL; |
| 760 | } |
| 761 | /* Look for higher precedence operators. */ |
| 762 | for (usize i = 0; i < sizeof(ops) / sizeof(ops[0]); i++) { |
| 763 | if (check(p, ops[i].tok) && ops[i].prec > next) { |
| 764 | right = parse_binary(p, right, next); |
| 765 | break; |
| 766 | } |
| 767 | } |
| 768 | |
| 769 | /* Build binary expression node. */ |
| 770 | node_t *binop = node(p, NODE_BINOP); |
| 771 | binop->offset = left->offset; |
| 772 | binop->length = right->offset + right->length - left->offset; |
| 773 | binop->val.binop.op = op; |
| 774 | binop->val.binop.left = left; |
| 775 | binop->val.binop.right = right; |
| 776 | left = binop; |
| 777 | } |
| 778 | return left; |
| 779 | } |
| 780 | |
| 781 | /* Parse an `if let` statement. |
| 782 | * Syntax: if let x in (expr) { ... } else { ... } |
| 783 | */ |
| 784 | static node_t *parse_if_let(parser_t *p) { |
| 785 | /* Consume 'let' */ |
| 786 | if (!expect(p, T_LET, "expected 'let'")) |
| 787 | return NULL; |
| 788 | |
| 789 | /* Check for `if let case` syntax. */ |
| 790 | if (check(p, T_CASE)) { |
| 791 | return parse_if_case(p); |
| 792 | } |
| 793 | node_t *n = node(p, NODE_IF_LET); |
| 794 | |
| 795 | /* Parse identifier or placeholder */ |
| 796 | if (consume(p, T_UNDERSCORE)) { |
| 797 | n->val.if_let_stmt.var = node(p, NODE_PLACEHOLDER); |
| 798 | } else if (expect(p, T_IDENT, "expected identifier or '_' after 'let'")) { |
| 799 | n->val.if_let_stmt.var = node(p, NODE_IDENT); |
| 800 | n->val.if_let_stmt.var->val.ident.name = p->previous.start; |
| 801 | n->val.if_let_stmt.var->val.ident.length = p->previous.length; |
| 802 | } else { |
| 803 | return NULL; |
| 804 | } |
| 805 | n->val.if_let_stmt.guard = NULL; |
| 806 | |
| 807 | /* Expect '=' */ |
| 808 | if (!expect(p, T_EQ, "expected `=` after identifier")) |
| 809 | return NULL; |
| 810 | |
| 811 | /* Parse expression yielding an optional. */ |
| 812 | n->val.if_let_stmt.expr = parse_cond(p); |
| 813 | if (!n->val.if_let_stmt.expr) |
| 814 | return NULL; |
| 815 | |
| 816 | /* Optional boolean guard. */ |
| 817 | if (consume(p, T_SEMICOLON)) { |
| 818 | n->val.if_let_stmt.guard = parse_cond(p); |
| 819 | } |
| 820 | /* Parse the 'then' branch */ |
| 821 | n->val.if_let_stmt.lbranch = parse_block(p); |
| 822 | if (!n->val.if_let_stmt.lbranch) |
| 823 | return NULL; |
| 824 | |
| 825 | /* Parse optional 'else' branch */ |
| 826 | if (consume(p, T_ELSE)) { |
| 827 | /* Check for `else if` construct. */ |
| 828 | if (check(p, T_IF)) { |
| 829 | advance(p); /* Consume the 'if' token. */ |
| 830 | |
| 831 | /* Create a block to hold the nested if statement. */ |
| 832 | node_t *block = node(p, NODE_BLOCK); |
| 833 | block->val.block.stmts = (nodespan_t){ 0 }; |
| 834 | |
| 835 | node_t *nested_if = parse_if(p); |
| 836 | |
| 837 | if (!nested_if) |
| 838 | return NULL; |
| 839 | |
| 840 | /* Add the nested if as a statement in the block. */ |
| 841 | nodespan_push(p, &block->val.block.stmts, nested_if); |
| 842 | /* Set the block as the else branch. */ |
| 843 | n->val.if_let_stmt.rbranch = block; |
| 844 | } else { |
| 845 | /* Regular else clause. */ |
| 846 | n->val.if_let_stmt.rbranch = parse_block(p); |
| 847 | } |
| 848 | } else { |
| 849 | n->val.if_let_stmt.rbranch = NULL; |
| 850 | } |
| 851 | |
| 852 | return n; |
| 853 | } |
| 854 | |
| 855 | /* Parse an `if let case` statement. Called after 'let' has been consumed. */ |
| 856 | static node_t *parse_if_case(parser_t *p) { |
| 857 | node_t *n = node(p, NODE_IF_CASE); |
| 858 | |
| 859 | if (!expect(p, T_CASE, "expected 'case'")) |
| 860 | return NULL; |
| 861 | |
| 862 | parse_ctx_t pctx = p->context; |
| 863 | p->context = PARSE_CTX_NORMAL; |
| 864 | node_t *pattern = parse_primary(p); |
| 865 | p->context = pctx; |
| 866 | |
| 867 | if (!pattern) |
| 868 | return NULL; |
| 869 | |
| 870 | n->val.if_case_stmt.pattern = pattern; |
| 871 | |
| 872 | if (!expect(p, T_EQ, "expected `=` after pattern")) |
| 873 | return NULL; |
| 874 | |
| 875 | n->val.if_case_stmt.expr = parse_cond(p); |
| 876 | if (!n->val.if_case_stmt.expr) |
| 877 | return NULL; |
| 878 | |
| 879 | n->val.if_case_stmt.guard = NULL; |
| 880 | |
| 881 | if (consume(p, T_SEMICOLON)) { |
| 882 | n->val.if_case_stmt.guard = parse_cond(p); |
| 883 | if (!n->val.if_case_stmt.guard) |
| 884 | return NULL; |
| 885 | } |
| 886 | n->val.if_case_stmt.lbranch = parse_block(p); |
| 887 | if (!n->val.if_case_stmt.lbranch) |
| 888 | return NULL; |
| 889 | |
| 890 | if (consume(p, T_ELSE)) { |
| 891 | if (check(p, T_IF)) { |
| 892 | advance(p); |
| 893 | |
| 894 | node_t *block = node(p, NODE_BLOCK); |
| 895 | block->val.block.stmts = (nodespan_t){ 0 }; |
| 896 | |
| 897 | node_t *nested_if = parse_if(p); |
| 898 | |
| 899 | if (!nested_if) |
| 900 | return NULL; |
| 901 | |
| 902 | nodespan_push(p, &block->val.block.stmts, nested_if); |
| 903 | n->val.if_case_stmt.rbranch = block; |
| 904 | } else { |
| 905 | n->val.if_case_stmt.rbranch = parse_block(p); |
| 906 | if (!n->val.if_case_stmt.rbranch) |
| 907 | return NULL; |
| 908 | } |
| 909 | } else { |
| 910 | n->val.if_case_stmt.rbranch = NULL; |
| 911 | } |
| 912 | return n; |
| 913 | } |
| 914 | |
| 915 | /* Parse a `let case` statement: |
| 916 | * `let case PATTERN = EXPR [; GUARD] else { ... };` */ |
| 917 | static node_t *parse_let_case(parser_t *p) { |
| 918 | node_t *n = node(p, NODE_GUARD_CASE); |
| 919 | usize start = p->previous.position; |
| 920 | |
| 921 | parse_ctx_t pctx = p->context; |
| 922 | p->context = PARSE_CTX_NORMAL; |
| 923 | node_t *pattern = parse_primary(p); |
| 924 | p->context = pctx; |
| 925 | |
| 926 | if (!pattern) |
| 927 | return NULL; |
| 928 | |
| 929 | n->val.guard_case_stmt.pattern = pattern; |
| 930 | |
| 931 | if (!expect(p, T_EQ, "expected `=` after pattern")) |
| 932 | return NULL; |
| 933 | if (!(n->val.guard_case_stmt.expr = parse_cond(p))) |
| 934 | return NULL; |
| 935 | |
| 936 | n->val.guard_case_stmt.guard = NULL; |
| 937 | |
| 938 | if (consume(p, T_IF)) { |
| 939 | if (!(n->val.guard_case_stmt.guard = parse_cond(p))) |
| 940 | return NULL; |
| 941 | } |
| 942 | if (!expect(p, T_ELSE, "expected `else` after pattern")) |
| 943 | return NULL; |
| 944 | |
| 945 | if (!(n->val.guard_case_stmt.rbranch = parse_stmt_or_block(p))) |
| 946 | return NULL; |
| 947 | |
| 948 | n->offset = start; |
| 949 | n->length = p->previous.position + p->previous.length - start; |
| 950 | |
| 951 | return n; |
| 952 | } |
| 953 | |
| 954 | /* Parse an `if` expression, with optional `else` or `else if` clauses. |
| 955 | * `else if` is desugared into a nested if inside a block. */ |
| 956 | static node_t *parse_if(parser_t *p) { |
| 957 | /* Check for `if let` or `if let case` syntax */ |
| 958 | if (check(p, T_LET)) { |
| 959 | return parse_if_let(p); |
| 960 | } |
| 961 | /* Regular if statement */ |
| 962 | node_t *n = node(p, NODE_IF); |
| 963 | |
| 964 | n->val.if_stmt.cond = parse_cond(p); |
| 965 | if (!n->val.if_stmt.cond) |
| 966 | return NULL; |
| 967 | |
| 968 | n->val.if_stmt.lbranch = parse_block(p); |
| 969 | if (!n->val.if_stmt.lbranch) |
| 970 | return NULL; |
| 971 | |
| 972 | if (consume(p, T_ELSE)) { |
| 973 | /* Check for `else if` construct. */ |
| 974 | if (check(p, T_IF)) { |
| 975 | advance(p); /* Consume the 'if' token. */ |
| 976 | |
| 977 | /* Create a block to hold the nested if statement. */ |
| 978 | node_t *block = node(p, NODE_BLOCK); |
| 979 | block->val.block.stmts = (nodespan_t){ 0 }; |
| 980 | |
| 981 | node_t *nested_if = parse_if(p); |
| 982 | |
| 983 | if (!nested_if) |
| 984 | return NULL; |
| 985 | |
| 986 | /* Add the nested if as a statement in the block. */ |
| 987 | nodespan_push(p, &block->val.block.stmts, nested_if); |
| 988 | /* Set the block as the else branch. */ |
| 989 | n->val.if_stmt.rbranch = block; |
| 990 | } else { |
| 991 | /* Regular else clause. */ |
| 992 | n->val.if_stmt.rbranch = parse_block(p); |
| 993 | } |
| 994 | } else { |
| 995 | n->val.if_stmt.rbranch = NULL; |
| 996 | } |
| 997 | return n; |
| 998 | } |
| 999 | |
| 1000 | /* Parse a match statement. */ |
| 1001 | static node_t *parse_match(parser_t *p) { |
| 1002 | node_t *n = node(p, NODE_MATCH); |
| 1003 | n->val.match_stmt.cases = (nodespan_t){ 0 }; |
| 1004 | |
| 1005 | /* Parse the expression to match on */ |
| 1006 | if (!(n->val.match_stmt.expr = parse_cond(p))) |
| 1007 | return NULL; |
| 1008 | if (!expect(p, T_LBRACE, "expected '{' before match cases")) |
| 1009 | return NULL; |
| 1010 | |
| 1011 | /* Parse cases until we reach the end of the match block */ |
| 1012 | while (!check(p, T_RBRACE) && !check(p, T_EOF)) { |
| 1013 | node_t *case_node = parse_match_case(p); |
| 1014 | if (!case_node) |
| 1015 | return NULL; |
| 1016 | |
| 1017 | if (!nodespan_push(p, &n->val.match_stmt.cases, case_node)) { |
| 1018 | error(p, "too many cases in match statement"); |
| 1019 | return NULL; |
| 1020 | } |
| 1021 | |
| 1022 | /* Consume the comma separating cases if present */ |
| 1023 | bool consumed = consume(p, T_COMMA); |
| 1024 | (void)consumed; |
| 1025 | } |
| 1026 | if (!expect(p, T_RBRACE, "expected '}' after match cases")) |
| 1027 | return NULL; |
| 1028 | |
| 1029 | return n; |
| 1030 | } |
| 1031 | |
| 1032 | /* Parse a single match case. */ |
| 1033 | static node_t *parse_match_case(parser_t *p) { |
| 1034 | node_t *n = node(p, NODE_MATCH_CASE); |
| 1035 | n->val.match_case.patterns = (nodespan_t){ 0 }; |
| 1036 | n->val.match_case.guard = NULL; |
| 1037 | |
| 1038 | if (check(p, T_ELSE)) { |
| 1039 | /* For the 'else' case, we use zero patterns |
| 1040 | * to indicate the else case */ |
| 1041 | advance(p); |
| 1042 | } else { |
| 1043 | if (!expect(p, T_CASE, "expected 'case' at start of match case")) |
| 1044 | return NULL; |
| 1045 | |
| 1046 | /* Parse one or more comma-separated patterns */ |
| 1047 | do { |
| 1048 | parse_ctx_t pctx = p->context; |
| 1049 | p->context = PARSE_CTX_NORMAL; |
| 1050 | node_t *pattern = parse_primary(p); |
| 1051 | p->context = pctx; |
| 1052 | |
| 1053 | if (!pattern) { |
| 1054 | return NULL; |
| 1055 | } |
| 1056 | /* Add pattern to the case */ |
| 1057 | if (!nodespan_push(p, &n->val.match_case.patterns, pattern)) { |
| 1058 | error(p, "too many patterns in case statement"); |
| 1059 | return NULL; |
| 1060 | } |
| 1061 | } while (consume(p, T_COMMA)); /* Continue if there's a comma */ |
| 1062 | |
| 1063 | if (consume(p, T_IF)) { |
| 1064 | if (!(n->val.match_case.guard = parse_cond(p))) |
| 1065 | return NULL; |
| 1066 | } |
| 1067 | } |
| 1068 | if (!expect(p, T_FAT_ARROW, "expected `=>` after case pattern")) |
| 1069 | return NULL; |
| 1070 | |
| 1071 | n->val.match_case.body = parse_stmt(p); |
| 1072 | if (!n->val.match_case.body) |
| 1073 | return NULL; |
| 1074 | |
| 1075 | return n; |
| 1076 | } |
| 1077 | |
| 1078 | /* Parse a `log` statement. */ |
| 1079 | /* Parse a record declaration. */ |
| 1080 | static node_t *parse_record(parser_t *p, node_t *attrs) { |
| 1081 | node_t *n = node(p, NODE_RECORD); |
| 1082 | n->val.record_decl.attribs = attrs; |
| 1083 | n->val.record_decl.fields = (nodespan_t){ 0 }; |
| 1084 | n->val.record_decl.tuple = false; |
| 1085 | n->val.record_decl.name = parse_ident(p, "expected record name"); |
| 1086 | |
| 1087 | if (!n->val.record_decl.name) |
| 1088 | return NULL; |
| 1089 | |
| 1090 | if (consume(p, T_LPAREN)) { |
| 1091 | n->val.record_decl.tuple = true; |
| 1092 | |
| 1093 | if (!check(p, T_RPAREN)) { |
| 1094 | do { |
| 1095 | node_t *field = node(p, NODE_RECORD_FIELD); |
| 1096 | field->val.var.ident = NULL; /* No field name for tuples */ |
| 1097 | field->val.var.type = parse_type(p); |
| 1098 | field->val.var.value = NULL; |
| 1099 | field->val.var.align = NULL; |
| 1100 | |
| 1101 | if (!field->val.var.type) |
| 1102 | return NULL; |
| 1103 | |
| 1104 | if (!nodespan_push(p, &n->val.record_decl.fields, field)) { |
| 1105 | error(p, "too many record fields"); |
| 1106 | return NULL; |
| 1107 | } |
| 1108 | } while (consume(p, T_COMMA) && !check(p, T_RPAREN)); |
| 1109 | } |
| 1110 | if (!expect(p, T_RPAREN, "expected `)` after record fields")) |
| 1111 | return NULL; |
| 1112 | |
| 1113 | /* Unlabeled records must end with semicolon */ |
| 1114 | if (!expect(p, T_SEMICOLON, "expected `;` after record declaration")) |
| 1115 | return NULL; |
| 1116 | } else { |
| 1117 | /* Record with named fields */ |
| 1118 | if (!expect(p, T_LBRACE, "expected `{` before record body")) |
| 1119 | return NULL; |
| 1120 | |
| 1121 | node_t *field; |
| 1122 | do { |
| 1123 | if (!(field = parse_name_type_value(p, NODE_RECORD_FIELD))) |
| 1124 | return NULL; |
| 1125 | if (!nodespan_push(p, &n->val.record_decl.fields, field)) { |
| 1126 | error(p, "too many record fields"); |
| 1127 | return NULL; |
| 1128 | } |
| 1129 | } while (consume(p, T_COMMA) && !check(p, T_RBRACE)); |
| 1130 | |
| 1131 | if (!expect(p, T_RBRACE, "expected `}`")) |
| 1132 | return NULL; |
| 1133 | } |
| 1134 | return n; |
| 1135 | } |
| 1136 | |
| 1137 | static node_t *parse_record_type(parser_t *p) { |
| 1138 | node_t *n = node(p, NODE_RECORD_TYPE); |
| 1139 | n->val.record_type.fields = (nodespan_t){ 0 }; |
| 1140 | |
| 1141 | if (!expect(p, T_LBRACE, "expected `{` after `record`")) |
| 1142 | return NULL; |
| 1143 | |
| 1144 | if (!check(p, T_RBRACE)) { |
| 1145 | do { |
| 1146 | node_t *field = parse_name_type_value(p, NODE_RECORD_FIELD); |
| 1147 | if (!field) |
| 1148 | return NULL; |
| 1149 | if (field->val.var.value) { |
| 1150 | error(p, "anonymous record fields cannot have initializers"); |
| 1151 | return NULL; |
| 1152 | } |
| 1153 | if (!nodespan_push(p, &n->val.record_type.fields, field)) { |
| 1154 | error(p, "too many record fields"); |
| 1155 | return NULL; |
| 1156 | } |
| 1157 | } while (consume(p, T_COMMA) && !check(p, T_RBRACE)); |
| 1158 | } |
| 1159 | |
| 1160 | if (!expect(p, T_RBRACE, "expected `}` after record fields")) |
| 1161 | return NULL; |
| 1162 | |
| 1163 | return n; |
| 1164 | } |
| 1165 | |
| 1166 | /* Parse a single record literal field (labeled or shorthand). */ |
| 1167 | static node_t *parse_record_lit_field(parser_t *p) { |
| 1168 | node_t *n = node(p, NODE_RECORD_LIT_FIELD); |
| 1169 | usize start = p->current.position; |
| 1170 | |
| 1171 | record_lit_field_t *field = &n->val.record_lit_field; |
| 1172 | |
| 1173 | /* Field must start with an identifier. */ |
| 1174 | node_t *name = parse_ident(p, "expected field name"); |
| 1175 | if (!name) |
| 1176 | return NULL; |
| 1177 | |
| 1178 | if (consume(p, T_COLON)) { |
| 1179 | /* Labeled field: `name: value` */ |
| 1180 | field->name = name; |
| 1181 | field->value = parse_expr(p); |
| 1182 | if (!field->value) |
| 1183 | return NULL; |
| 1184 | } else { |
| 1185 | /* Shorthand syntax: `{ x }` is equivalent to `{ x: x }` */ |
| 1186 | field->name = name; |
| 1187 | field->value = name; |
| 1188 | } |
| 1189 | n->offset = start; |
| 1190 | n->length = p->previous.position + p->previous.length - start; |
| 1191 | |
| 1192 | return n; |
| 1193 | } |
| 1194 | |
| 1195 | /* Parse a record literal expression (e.g., Point { x: 1, y: 2 }) |
| 1196 | * Also handles pattern syntax: Variant { .. } to discard all fields */ |
| 1197 | static node_t *parse_record_lit(parser_t *p, node_t *type_name) { |
| 1198 | node_t *n = node(p, NODE_RECORD_LIT); |
| 1199 | n->val.record_lit.type = type_name; |
| 1200 | n->val.record_lit.fields = (nodespan_t){ 0 }; |
| 1201 | n->val.record_lit.etc = false; |
| 1202 | |
| 1203 | do { |
| 1204 | /* Check for `..` to discard remaining fields. */ |
| 1205 | if (consume(p, T_DOT_DOT)) { |
| 1206 | n->val.record_lit.etc = true; |
| 1207 | break; |
| 1208 | } |
| 1209 | node_t *field = parse_record_lit_field(p); |
| 1210 | if (!field) |
| 1211 | return NULL; |
| 1212 | |
| 1213 | if (!nodespan_push(p, &n->val.record_lit.fields, field)) { |
| 1214 | error(p, "too many record fields"); |
| 1215 | return NULL; |
| 1216 | } |
| 1217 | |
| 1218 | } while (consume(p, T_COMMA) && !check(p, T_RBRACE)); |
| 1219 | |
| 1220 | if (!expect(p, T_RBRACE, "expected '}' to end record literal")) |
| 1221 | return NULL; |
| 1222 | |
| 1223 | return n; |
| 1224 | } |
| 1225 | |
| 1226 | /* Parse a union declaration. |
| 1227 | * Eg. `union Color { Red, Green, Blue = 5 }` */ |
| 1228 | static node_t *parse_union(parser_t *p, node_t *attrs) { |
| 1229 | node_t *n = node(p, NODE_UNION); |
| 1230 | n->val.union_decl.attribs = attrs; |
| 1231 | n->val.union_decl.variants = (nodespan_t){ 0 }; |
| 1232 | n->val.union_decl.name = parse_ident(p, "expected union name"); |
| 1233 | |
| 1234 | if (!n->val.union_decl.name) |
| 1235 | return NULL; |
| 1236 | |
| 1237 | /* Parse union body with { ... } */ |
| 1238 | if (!expect(p, T_LBRACE, "expected `{` before union body")) |
| 1239 | return NULL; |
| 1240 | |
| 1241 | /* Parse union variants. */ |
| 1242 | if (!check(p, T_RBRACE)) { |
| 1243 | do { |
| 1244 | /* Allow optional `case` keyword before variant name. */ |
| 1245 | consume(p, T_CASE); |
| 1246 | |
| 1247 | /* Parse variant name. */ |
| 1248 | node_t *variant_name = parse_ident(p, "expected variant name"); |
| 1249 | if (!variant_name) |
| 1250 | return NULL; |
| 1251 | |
| 1252 | node_t *v = node(p, NODE_UNION_VARIANT); |
| 1253 | union_variant_t *variant = &v->val.union_variant; |
| 1254 | |
| 1255 | variant->name = variant_name; |
| 1256 | variant->type = NULL; |
| 1257 | variant->value_expr = NULL; |
| 1258 | |
| 1259 | if (consume(p, T_LPAREN)) { |
| 1260 | /* Tuple-like variant: Foo(Type) */ |
| 1261 | node_t *payload = parse_type(p); |
| 1262 | if (!payload) |
| 1263 | return NULL; |
| 1264 | variant->type = payload; |
| 1265 | if (!expect(p, T_RPAREN, "expected `)` after variant type")) |
| 1266 | return NULL; |
| 1267 | } else if (check(p, T_LBRACE)) { |
| 1268 | /* Struct-like variant: Bar { x: i32, y: i32 } */ |
| 1269 | node_t *payload = parse_record_type(p); |
| 1270 | if (!payload) |
| 1271 | return NULL; |
| 1272 | variant->type = payload; |
| 1273 | } else { |
| 1274 | /* Check for explicit value assignment. */ |
| 1275 | if (consume(p, T_EQ)) { |
| 1276 | if (!expect( |
| 1277 | p, T_NUMBER, "expected integer literal after `=`" |
| 1278 | )) |
| 1279 | return NULL; |
| 1280 | |
| 1281 | token_t literal_tok = p->previous; |
| 1282 | node_t *literal = node(p, NODE_NUMBER); |
| 1283 | |
| 1284 | literal->offset = literal_tok.position; |
| 1285 | literal->length = literal_tok.length; |
| 1286 | literal->val.number.text = literal_tok.start; |
| 1287 | literal->val.number.text_len = literal_tok.length; |
| 1288 | literal->val.number.value = (imm_t){ 0 }; |
| 1289 | |
| 1290 | variant->value_expr = literal; |
| 1291 | } else { |
| 1292 | /* Auto-assign value. */ |
| 1293 | } |
| 1294 | } |
| 1295 | /* Add variant to declaration node. */ |
| 1296 | if (!nodespan_push(p, &n->val.union_decl.variants, v)) { |
| 1297 | error(p, "too many union variants"); |
| 1298 | return NULL; |
| 1299 | } |
| 1300 | /* Allow trailing comma. */ |
| 1301 | } while (consume(p, T_COMMA) && !check(p, T_RBRACE)); |
| 1302 | } |
| 1303 | if (!expect(p, T_RBRACE, "expected `}`")) |
| 1304 | return NULL; |
| 1305 | |
| 1306 | return n; |
| 1307 | } |
| 1308 | |
| 1309 | /* Parse a code block or an expression */ |
| 1310 | static node_t *parse_stmt_or_block(parser_t *p) { |
| 1311 | if (check(p, T_LBRACE)) { |
| 1312 | return parse_block(p); |
| 1313 | } |
| 1314 | node_t *stmt = parse_stmt(p); |
| 1315 | node_t *blk = node(p, NODE_BLOCK); |
| 1316 | blk->val.block.stmts = (nodespan_t){ 0 }; |
| 1317 | nodespan_push(p, &blk->val.block.stmts, stmt); |
| 1318 | |
| 1319 | return blk; |
| 1320 | } |
| 1321 | |
| 1322 | /* Parse a code block, enclosed by `{}`. */ |
| 1323 | static node_t *parse_block(parser_t *p) { |
| 1324 | if (!expect(p, T_LBRACE, "expected '{' before block")) { |
| 1325 | return NULL; |
| 1326 | } |
| 1327 | node_t *n = node(p, NODE_BLOCK); |
| 1328 | node_t *stmt; |
| 1329 | |
| 1330 | /* Parse statements. */ |
| 1331 | n->val.block.stmts = (nodespan_t){ 0 }; |
| 1332 | while (!check(p, T_RBRACE) && !check(p, T_EOF)) { |
| 1333 | usize start = p->current.position; |
| 1334 | |
| 1335 | if (!(stmt = parse_stmt(p))) |
| 1336 | return NULL; |
| 1337 | |
| 1338 | if (!consume_statement_separator(p, stmt, true)) |
| 1339 | return NULL; |
| 1340 | |
| 1341 | stmt->offset = start; |
| 1342 | stmt->length = p->current.position - start; |
| 1343 | |
| 1344 | if (!nodespan_push(p, &n->val.block.stmts, stmt)) { |
| 1345 | error(p, "too many statements in block"); |
| 1346 | return NULL; |
| 1347 | } |
| 1348 | } |
| 1349 | |
| 1350 | if (!expect(p, T_RBRACE, "expected matching '}' after block")) |
| 1351 | return NULL; |
| 1352 | |
| 1353 | return n; |
| 1354 | } |
| 1355 | |
| 1356 | /* Parse an expression. */ |
| 1357 | static node_t *parse_expr(parser_t *p) { |
| 1358 | node_t *lval; |
| 1359 | |
| 1360 | if ((lval = parse_primary(p)) == NULL) |
| 1361 | return NULL; |
| 1362 | |
| 1363 | /* Handle `as` casts before binary operators (higher precedence than |
| 1364 | * binary ops, lower than unary) */ |
| 1365 | while (check(p, T_AS)) { |
| 1366 | lval = parse_as_cast(p, lval); |
| 1367 | if (!lval) |
| 1368 | return NULL; |
| 1369 | } |
| 1370 | lval = parse_binary(p, lval, -1); |
| 1371 | |
| 1372 | return lval; |
| 1373 | } |
| 1374 | |
| 1375 | /* Parse an assignment statement. */ |
| 1376 | static node_t *parse_assignment(parser_t *p, node_t *lval) { |
| 1377 | /* We've already verified this is an assignment. */ |
| 1378 | if (lval->cls != NODE_IDENT && lval->cls != NODE_ACCESS && |
| 1379 | lval->cls != NODE_ARRAY_INDEX && |
| 1380 | !(lval->cls == NODE_UNOP && lval->val.unop.op == OP_DEREF)) { |
| 1381 | error( |
| 1382 | p, |
| 1383 | "can't assign to `%.*s`", |
| 1384 | lval->length, |
| 1385 | &p->scanner.source[lval->offset] |
| 1386 | ); |
| 1387 | return NULL; |
| 1388 | } |
| 1389 | node_t *rval; |
| 1390 | |
| 1391 | if (!(rval = parse_expr(p))) |
| 1392 | return NULL; |
| 1393 | |
| 1394 | node_t *assign = node(p, NODE_ASSIGN); |
| 1395 | assign->val.assign.lval = lval; |
| 1396 | assign->val.assign.rval = rval; |
| 1397 | |
| 1398 | return assign; |
| 1399 | } |
| 1400 | |
| 1401 | /* Parse a condition. */ |
| 1402 | static node_t *parse_cond(parser_t *p) { |
| 1403 | parse_ctx_t prev = p->context; |
| 1404 | p->context = PARSE_CTX_CONDITION; |
| 1405 | |
| 1406 | node_t *cond = parse_expr(p); |
| 1407 | if (!cond) { |
| 1408 | p->context = prev; |
| 1409 | return NULL; |
| 1410 | } |
| 1411 | p->context = prev; |
| 1412 | |
| 1413 | return cond; |
| 1414 | } |
| 1415 | |
| 1416 | static bool token_is_stmt_terminator(tokenclass_t cls) { |
| 1417 | switch (cls) { |
| 1418 | case T_SEMICOLON: |
| 1419 | case T_RBRACE: |
| 1420 | case T_COMMA: |
| 1421 | case T_CASE: |
| 1422 | case T_ELSE: |
| 1423 | case T_EOF: |
| 1424 | return true; |
| 1425 | default: |
| 1426 | return false; |
| 1427 | } |
| 1428 | } |
| 1429 | |
| 1430 | static bool stmt_requires_semicolon(const node_t *stmt) { |
| 1431 | switch (stmt->cls) { |
| 1432 | case NODE_IF: |
| 1433 | case NODE_IF_LET: |
| 1434 | case NODE_IF_CASE: |
| 1435 | case NODE_WHILE: |
| 1436 | case NODE_WHILE_LET: |
| 1437 | case NODE_LOOP: |
| 1438 | case NODE_FOR: |
| 1439 | case NODE_MATCH: |
| 1440 | case NODE_BLOCK: |
| 1441 | case NODE_FN: |
| 1442 | case NODE_RECORD: |
| 1443 | case NODE_UNION: |
| 1444 | return false; |
| 1445 | default: |
| 1446 | return true; |
| 1447 | } |
| 1448 | } |
| 1449 | |
| 1450 | static bool consume_statement_separator( |
| 1451 | parser_t *p, node_t *stmt, bool require |
| 1452 | ) { |
| 1453 | if (stmt_requires_semicolon(stmt)) { |
| 1454 | return expect(p, T_SEMICOLON, "expected `;` after statement"); |
| 1455 | } |
| 1456 | if (require) |
| 1457 | consume(p, T_SEMICOLON); |
| 1458 | return true; |
| 1459 | } |
| 1460 | |
| 1461 | /* Parse a `return` statement. */ |
| 1462 | static node_t *parse_return(parser_t *p) { |
| 1463 | node_t *n = node(p, NODE_RETURN); |
| 1464 | |
| 1465 | if (!token_is_stmt_terminator(p->current.cls)) { |
| 1466 | n->val.return_stmt.value = parse_expr(p); |
| 1467 | if (!n->val.return_stmt.value) |
| 1468 | return NULL; |
| 1469 | } else { |
| 1470 | n->val.return_stmt.value = NULL; /* Return void. */ |
| 1471 | } |
| 1472 | |
| 1473 | return n; |
| 1474 | } |
| 1475 | |
| 1476 | static node_t *parse_throw(parser_t *p) { |
| 1477 | node_t *n = node(p, NODE_THROW); |
| 1478 | |
| 1479 | if (!(n->val.throw_stmt.expr = parse_expr(p))) |
| 1480 | return NULL; |
| 1481 | |
| 1482 | return n; |
| 1483 | } |
| 1484 | |
| 1485 | /* Parse a `break` statement. */ |
| 1486 | static node_t *parse_break(parser_t *p) { |
| 1487 | node_t *n = node(p, NODE_BREAK); |
| 1488 | |
| 1489 | return n; |
| 1490 | } |
| 1491 | |
| 1492 | /* Parse a `for` statement. */ |
| 1493 | static node_t *parse_for(parser_t *p) { |
| 1494 | node_t *n = node(p, NODE_FOR); |
| 1495 | n->val.for_stmt.rbranch = NULL; |
| 1496 | n->val.for_stmt.idx = NULL; |
| 1497 | |
| 1498 | /* Parse the loop variable name or placeholder */ |
| 1499 | if (!(n->val.for_stmt.var = |
| 1500 | parse_ident_or_placeholder(p, "expected identifier or '_'"))) |
| 1501 | return NULL; |
| 1502 | |
| 1503 | /* Check for optional index variable: `for x, i in xs` */ |
| 1504 | if (consume(p, T_COMMA)) { |
| 1505 | /* Parse the index variable name or placeholder */ |
| 1506 | if (!(n->val.for_stmt.idx = parse_ident_or_placeholder( |
| 1507 | p, "expected index identifier or '_' after comma" |
| 1508 | ))) |
| 1509 | return NULL; |
| 1510 | } |
| 1511 | |
| 1512 | if (!expect(p, T_IN, "expected `in`")) |
| 1513 | return NULL; |
| 1514 | |
| 1515 | if (!(n->val.for_stmt.iter = parse_cond(p))) |
| 1516 | return NULL; |
| 1517 | |
| 1518 | if (!(n->val.for_stmt.body = parse_block(p))) |
| 1519 | return NULL; |
| 1520 | |
| 1521 | /* Parse optional `else` clause */ |
| 1522 | if (consume(p, T_ELSE)) { |
| 1523 | if (!(n->val.for_stmt.rbranch = parse_block(p))) |
| 1524 | return NULL; |
| 1525 | } |
| 1526 | return n; |
| 1527 | } |
| 1528 | |
| 1529 | /* Parse a `while let` statement. */ |
| 1530 | static node_t *parse_while_let(parser_t *p) { |
| 1531 | if (!expect(p, T_LET, "expected `let`")) |
| 1532 | return NULL; |
| 1533 | |
| 1534 | node_t *n = node(p, NODE_WHILE_LET); |
| 1535 | |
| 1536 | /* Parse identifier or placeholder */ |
| 1537 | if (consume(p, T_UNDERSCORE)) { |
| 1538 | n->val.while_let_stmt.var = node(p, NODE_PLACEHOLDER); |
| 1539 | } else if (expect(p, T_IDENT, "expected identifier or '_' after `let`")) { |
| 1540 | n->val.while_let_stmt.var = node(p, NODE_IDENT); |
| 1541 | n->val.while_let_stmt.var->val.ident.name = p->previous.start; |
| 1542 | n->val.while_let_stmt.var->val.ident.length = p->previous.length; |
| 1543 | } else { |
| 1544 | return NULL; |
| 1545 | } |
| 1546 | n->val.while_let_stmt.guard = NULL; |
| 1547 | n->val.while_let_stmt.rbranch = NULL; |
| 1548 | |
| 1549 | if (!expect(p, T_EQ, "expected `=` after identifier")) |
| 1550 | return NULL; |
| 1551 | |
| 1552 | /* Parse expression yielding an optional. */ |
| 1553 | n->val.while_let_stmt.expr = parse_cond(p); |
| 1554 | if (!n->val.while_let_stmt.expr) |
| 1555 | return NULL; |
| 1556 | |
| 1557 | /* Optional guard condition after semicolon. */ |
| 1558 | if (consume(p, T_SEMICOLON)) { |
| 1559 | if (!(n->val.while_let_stmt.guard = parse_cond(p))) |
| 1560 | return NULL; |
| 1561 | } |
| 1562 | |
| 1563 | /* Parse the loop body and optional 'else' branch */ |
| 1564 | if (!(n->val.while_let_stmt.body = parse_block(p))) |
| 1565 | return NULL; |
| 1566 | if (consume(p, T_ELSE)) { |
| 1567 | if (!(n->val.while_let_stmt.rbranch = parse_block(p))) |
| 1568 | return NULL; |
| 1569 | } |
| 1570 | return n; |
| 1571 | } |
| 1572 | |
| 1573 | /* Parse a `while` statement. */ |
| 1574 | static node_t *parse_while(parser_t *p) { |
| 1575 | /* Check for `while let` syntax */ |
| 1576 | if (check(p, T_LET)) { |
| 1577 | return parse_while_let(p); |
| 1578 | } |
| 1579 | node_t *n = node(p, NODE_WHILE); |
| 1580 | n->val.while_stmt.rbranch = NULL; |
| 1581 | |
| 1582 | if (!(n->val.while_stmt.cond = parse_cond(p))) |
| 1583 | return NULL; |
| 1584 | if (!(n->val.while_stmt.body = parse_block(p))) |
| 1585 | return NULL; |
| 1586 | |
| 1587 | /* Parse optional else clause */ |
| 1588 | if (consume(p, T_ELSE)) { |
| 1589 | if (!(n->val.while_stmt.rbranch = parse_block(p))) |
| 1590 | return NULL; |
| 1591 | } |
| 1592 | return n; |
| 1593 | } |
| 1594 | |
| 1595 | /* Parse a `loop` statement. */ |
| 1596 | static node_t *parse_loop(parser_t *p) { |
| 1597 | node_t *n = node(p, NODE_LOOP); |
| 1598 | |
| 1599 | if (!(n->val.loop_stmt.body = parse_block(p))) |
| 1600 | return NULL; |
| 1601 | |
| 1602 | return n; |
| 1603 | } |
| 1604 | |
| 1605 | static node_t *parse_try(parser_t *p, bool panic, bool optional) { |
| 1606 | node_t *n = node(p, NODE_TRY); |
| 1607 | |
| 1608 | n->val.try_expr.expr = NULL; |
| 1609 | n->val.try_expr.catch_expr = NULL; |
| 1610 | n->val.try_expr.handlers = nodespan_alloc(p, MAX_TRY_CATCHES); |
| 1611 | n->val.try_expr.panic = panic; |
| 1612 | n->val.try_expr.optional = optional; |
| 1613 | |
| 1614 | if (!(n->val.try_expr.expr = parse_primary(p))) |
| 1615 | return NULL; |
| 1616 | |
| 1617 | /* Parse catch clause: `catch { ... }` or `catch e { ... }` */ |
| 1618 | if (consume(p, T_CATCH)) { |
| 1619 | node_t *catch_node = node(p, NODE_CATCH); |
| 1620 | catch_node->val.catch_clause.binding = NULL; |
| 1621 | catch_node->val.catch_clause.body = NULL; |
| 1622 | catch_node->val.catch_clause.scope = NULL; |
| 1623 | |
| 1624 | /* Check for error binding: `catch e { ... }` */ |
| 1625 | if (check(p, T_IDENT)) { |
| 1626 | node_t *binding = node(p, NODE_IDENT); |
| 1627 | binding->val.ident.name = p->current.start; |
| 1628 | binding->val.ident.length = p->current.length; |
| 1629 | catch_node->val.catch_clause.binding = binding; |
| 1630 | advance(p); |
| 1631 | } |
| 1632 | |
| 1633 | if (!check(p, T_LBRACE)) { |
| 1634 | error(p, "expected `{` after `catch`"); |
| 1635 | return NULL; |
| 1636 | } |
| 1637 | if (!(catch_node->val.catch_clause.body = parse_block(p))) |
| 1638 | return NULL; |
| 1639 | |
| 1640 | n->val.try_expr.catch_expr = catch_node; |
| 1641 | } |
| 1642 | return n; |
| 1643 | } |
| 1644 | |
| 1645 | static node_t *parse_panic(parser_t *p) { |
| 1646 | node_t *panic = node(p, NODE_PANIC); |
| 1647 | |
| 1648 | /* `panic { "Something's wrong!" }` */ |
| 1649 | if (consume(p, T_LBRACE)) { |
| 1650 | node_t *expr = parse_expr(p); |
| 1651 | if (!(panic->val.panic_stmt.message = expr)) |
| 1652 | return NULL; |
| 1653 | if (!expect(p, T_RBRACE, "expected closing `}` after expression")) |
| 1654 | return NULL; |
| 1655 | |
| 1656 | return panic; |
| 1657 | } |
| 1658 | |
| 1659 | if (token_is_stmt_terminator(p->current.cls)) { |
| 1660 | panic->val.panic_stmt.message = NULL; |
| 1661 | return panic; |
| 1662 | } |
| 1663 | |
| 1664 | node_t *expr = parse_expr(p); |
| 1665 | if (!(panic->val.panic_stmt.message = expr)) |
| 1666 | return NULL; |
| 1667 | |
| 1668 | return panic; |
| 1669 | } |
| 1670 | |
| 1671 | /* Parse a name, type, and optional value. |
| 1672 | * |
| 1673 | * Used for record field declarations, variable declarations, and record field |
| 1674 | * initializations. */ |
| 1675 | static node_t *parse_name_type_value(parser_t *p, nodeclass_t cls) { |
| 1676 | node_t *n = node(p, cls); |
| 1677 | usize start = p->current.position; |
| 1678 | node_t *type = NULL; |
| 1679 | bool is_typed = false; |
| 1680 | |
| 1681 | n->val.var.ident = |
| 1682 | parse_ident_or_placeholder(p, "expected identifier or '_'"); |
| 1683 | if (!n->val.var.ident) |
| 1684 | return NULL; |
| 1685 | |
| 1686 | if (cls == NODE_VAR) { |
| 1687 | /* Type annotation is optional for variable declarations. */ |
| 1688 | if (consume(p, T_COLON)) |
| 1689 | is_typed = true; |
| 1690 | } else { |
| 1691 | if (!expect(p, T_COLON, "expected `:` after identifier")) |
| 1692 | return NULL; |
| 1693 | is_typed = true; |
| 1694 | } |
| 1695 | |
| 1696 | if (is_typed) { |
| 1697 | type = parse_type(p); |
| 1698 | if (!type) |
| 1699 | return NULL; |
| 1700 | |
| 1701 | if (cls == NODE_VAR) { |
| 1702 | n->val.var.align = NULL; |
| 1703 | |
| 1704 | if (consume(p, T_ALIGN)) { |
| 1705 | if (!expect(p, T_LPAREN, "expected `(` after `align`")) |
| 1706 | return NULL; |
| 1707 | |
| 1708 | n->val.var.align = node(p, NODE_ALIGN); |
| 1709 | n->val.var.align->val.align = parse_expr(p); |
| 1710 | |
| 1711 | if (!expect(p, T_RPAREN, "expected `)` after expression")) |
| 1712 | return NULL; |
| 1713 | } |
| 1714 | } |
| 1715 | } else if (cls == NODE_VAR) { |
| 1716 | n->val.var.align = NULL; |
| 1717 | } |
| 1718 | n->val.var.type = type; |
| 1719 | n->val.var.value = NULL; |
| 1720 | |
| 1721 | /* Parse the optional value. */ |
| 1722 | if (consume(p, T_EQ)) { |
| 1723 | node_t *value = parse_expr(p); |
| 1724 | if (!value) |
| 1725 | return NULL; |
| 1726 | n->val.var.value = value; |
| 1727 | } |
| 1728 | /* Set the node location. */ |
| 1729 | n->offset = start; |
| 1730 | n->length = p->previous.position + p->previous.length - start; |
| 1731 | |
| 1732 | return n; |
| 1733 | } |
| 1734 | |
| 1735 | /* Parse a variable declaration. */ |
| 1736 | static node_t *parse_var(parser_t *p, bool mutable) { |
| 1737 | node_t *var = parse_name_type_value(p, NODE_VAR); |
| 1738 | |
| 1739 | if (!var) |
| 1740 | return NULL; |
| 1741 | |
| 1742 | var->val.var.mutable = mutable; |
| 1743 | |
| 1744 | /* Parse optional `else` clause. */ |
| 1745 | if (consume(p, T_ELSE)) { |
| 1746 | if (mutable) { |
| 1747 | error(p, "let-else bindings cannot be mutable"); |
| 1748 | return NULL; |
| 1749 | } |
| 1750 | if (!var->val.var.value) { |
| 1751 | error(p, "let-else requires an initializer"); |
| 1752 | return NULL; |
| 1753 | } |
| 1754 | node_t *rbranch = parse_stmt_or_block(p); |
| 1755 | if (!rbranch) |
| 1756 | return NULL; |
| 1757 | |
| 1758 | var->cls = NODE_GUARD_LET; |
| 1759 | var->val.guard_let_stmt.var = var->val.var.ident; |
| 1760 | var->val.guard_let_stmt.expr = var->val.var.value; |
| 1761 | var->val.guard_let_stmt.rbranch = rbranch; |
| 1762 | var->length = p->previous.position + p->previous.length - var->offset; |
| 1763 | |
| 1764 | return var; |
| 1765 | } |
| 1766 | var->length = p->previous.position + p->previous.length - var->offset; |
| 1767 | |
| 1768 | return var; |
| 1769 | } |
| 1770 | |
| 1771 | /* Parse a static variable declaration. */ |
| 1772 | static node_t *parse_static(parser_t *p) { |
| 1773 | node_t *n = node(p, NODE_STATIC); |
| 1774 | usize start = p->previous.position; |
| 1775 | |
| 1776 | node_t *ident = parse_label(p, "expected identifier in static declaration"); |
| 1777 | if (!ident) |
| 1778 | return NULL; |
| 1779 | |
| 1780 | node_t *type = parse_type(p); |
| 1781 | if (!type) |
| 1782 | return NULL; |
| 1783 | |
| 1784 | if (!expect(p, T_EQ, "expected `=` in static declaration")) |
| 1785 | return NULL; |
| 1786 | |
| 1787 | node_t *value = parse_expr(p); |
| 1788 | if (!value) |
| 1789 | return NULL; |
| 1790 | |
| 1791 | n->val.static_decl.ident = ident; |
| 1792 | n->val.static_decl.type = type; |
| 1793 | n->val.static_decl.value = value; |
| 1794 | n->offset = start; |
| 1795 | n->length = p->previous.position + p->previous.length - start; |
| 1796 | |
| 1797 | return n; |
| 1798 | } |
| 1799 | |
| 1800 | /* Parse a constant declaration. */ |
| 1801 | static node_t *parse_const(parser_t *p) { |
| 1802 | node_t *var = parse_name_type_value(p, NODE_CONST); |
| 1803 | |
| 1804 | if (!var) |
| 1805 | return NULL; |
| 1806 | |
| 1807 | return var; |
| 1808 | } |
| 1809 | |
| 1810 | /* Parse a module use declaration. */ |
| 1811 | static node_t *parse_use(parser_t *p, node_t *attrs) { |
| 1812 | usize start = p->current.position; |
| 1813 | |
| 1814 | /* Parse the first identifier in the path. */ |
| 1815 | node_t *path = parse_scope_segment(p, "expected module name after 'use'"); |
| 1816 | if (!path) |
| 1817 | return NULL; |
| 1818 | |
| 1819 | /* Track if this is a wildcard import. */ |
| 1820 | bool wildcard = false; |
| 1821 | |
| 1822 | /* Continue parsing the dotted path if present. */ |
| 1823 | while (consume(p, T_COLON_COLON)) { |
| 1824 | /* Check for wildcard import (e.g., `use foo::*`) */ |
| 1825 | if (consume(p, T_STAR)) { |
| 1826 | wildcard = true; |
| 1827 | break; |
| 1828 | } |
| 1829 | |
| 1830 | node_t *n = node(p, NODE_SCOPE); |
| 1831 | n->val.access.lval = path; |
| 1832 | |
| 1833 | /* Parse the sub-module name. */ |
| 1834 | node_t *mod = |
| 1835 | parse_scope_segment(p, "expected identifier or '*' after '::'"); |
| 1836 | if (!mod) |
| 1837 | return NULL; |
| 1838 | |
| 1839 | n->val.access.rval = mod; |
| 1840 | path = n; |
| 1841 | } |
| 1842 | |
| 1843 | /* Create a use node and wrap the path. */ |
| 1844 | node_t *use_node = node(p, NODE_USE); |
| 1845 | use_node->val.use_decl.path = path; |
| 1846 | use_node->val.use_decl.attribs = attrs; |
| 1847 | use_node->val.use_decl.wildcard = wildcard; |
| 1848 | |
| 1849 | /* Set position information. */ |
| 1850 | use_node->offset = start; |
| 1851 | use_node->length = p->previous.position + p->previous.length - start; |
| 1852 | |
| 1853 | return use_node; |
| 1854 | } |
| 1855 | |
| 1856 | /* Parse a module declaration. */ |
| 1857 | static node_t *parse_mod(parser_t *p, node_t *attrs) { |
| 1858 | usize start = p->current.position; |
| 1859 | |
| 1860 | node_t *ident = parse_ident(p, "expected module name after 'mod'"); |
| 1861 | if (!ident) |
| 1862 | return NULL; |
| 1863 | node_t *mod_node = node(p, NODE_MOD); |
| 1864 | mod_node->val.mod_decl.ident = ident; |
| 1865 | mod_node->val.mod_decl.attribs = attrs; |
| 1866 | |
| 1867 | mod_node->offset = start; |
| 1868 | mod_node->length = p->previous.position + p->previous.length - start; |
| 1869 | |
| 1870 | return mod_node; |
| 1871 | } |
| 1872 | |
| 1873 | /* Parse a function parameter. */ |
| 1874 | static node_t *parse_fn_param(parser_t *p) { |
| 1875 | /* Create parameter node. */ |
| 1876 | node_t *param = node(p, NODE_PARAM); |
| 1877 | node_t *name = parse_label(p, "expected parameter name"); |
| 1878 | if (!name) |
| 1879 | return NULL; |
| 1880 | |
| 1881 | param->val.param.ident = name; |
| 1882 | |
| 1883 | /* Parse and store parameter type. */ |
| 1884 | if (!(param->val.param.type = parse_type(p))) |
| 1885 | return NULL; |
| 1886 | |
| 1887 | return param; |
| 1888 | } |
| 1889 | |
| 1890 | static node_t *parse_module_body(parser_t *p) { |
| 1891 | node_t *mod = node(p, NODE_MOD_BODY); |
| 1892 | mod->val.block.stmts = (nodespan_t){ 0 }; |
| 1893 | |
| 1894 | while (!check(p, T_EOF)) { |
| 1895 | node_t *stmt; |
| 1896 | usize start = p->current.position; |
| 1897 | |
| 1898 | if (!(stmt = parse_stmt(p))) |
| 1899 | return NULL; |
| 1900 | |
| 1901 | if (!consume_statement_separator(p, stmt, true)) |
| 1902 | return NULL; |
| 1903 | |
| 1904 | stmt->offset = start; |
| 1905 | stmt->length = p->current.position - start; |
| 1906 | |
| 1907 | if (!nodespan_push(p, &mod->val.block.stmts, stmt)) { |
| 1908 | error(p, "too many statements in module"); |
| 1909 | return NULL; |
| 1910 | } |
| 1911 | } |
| 1912 | return mod; |
| 1913 | } |
| 1914 | |
| 1915 | /* Parse a function definition. */ |
| 1916 | static node_t *parse_fn(parser_t *p, node_t *attrs) { |
| 1917 | node_t *n = node(p, NODE_FN); |
| 1918 | node_t *param = NULL; |
| 1919 | |
| 1920 | /* Parse the function name. */ |
| 1921 | node_t *name = parse_ident(p, "expected function name"); |
| 1922 | if (!name) |
| 1923 | return NULL; |
| 1924 | |
| 1925 | n->val.fn_decl.ident = name; |
| 1926 | n->val.fn_decl.params = nodespan_alloc(p, MAX_FN_PARAMS); |
| 1927 | n->val.fn_decl.throws = nodespan_alloc(p, MAX_FN_THROWS); |
| 1928 | n->val.fn_decl.attribs = attrs; |
| 1929 | n->val.fn_decl.body = NULL; |
| 1930 | |
| 1931 | /* Check if it's an extern function */ |
| 1932 | bool is_extern = (attrs && attrs->val.attrib & ATTRIB_EXTERN); |
| 1933 | |
| 1934 | if (!expect(p, T_LPAREN, "expected `(` after function name")) |
| 1935 | return NULL; |
| 1936 | |
| 1937 | /* Parse parameters with types */ |
| 1938 | if (!check(p, T_RPAREN)) { |
| 1939 | do { |
| 1940 | if (n->val.fn_decl.params.len >= MAX_FN_PARAMS) { |
| 1941 | error( |
| 1942 | p, |
| 1943 | "maximum number of function parameters (%d) exceeded", |
| 1944 | MAX_FN_PARAMS |
| 1945 | ); |
| 1946 | return NULL; |
| 1947 | } |
| 1948 | if (!(param = parse_fn_param(p))) { |
| 1949 | return NULL; |
| 1950 | } |
| 1951 | node_fn_add_param(p, n, param); |
| 1952 | |
| 1953 | } while (consume(p, T_COMMA)); |
| 1954 | } |
| 1955 | if (!expect(p, T_RPAREN, "expected matching `)` after parameters list")) |
| 1956 | return NULL; |
| 1957 | |
| 1958 | if (consume(p, T_ARROW)) { |
| 1959 | if (!(n->val.fn_decl.return_type = parse_type(p))) { |
| 1960 | return NULL; |
| 1961 | } |
| 1962 | } else { |
| 1963 | n->val.fn_decl.return_type = NULL; |
| 1964 | } |
| 1965 | if (consume(p, T_THROWS)) { |
| 1966 | if (!expect(p, T_LPAREN, "expected `(` after `throws`")) |
| 1967 | return NULL; |
| 1968 | |
| 1969 | if (!check(p, T_RPAREN)) { |
| 1970 | do { |
| 1971 | if (n->val.fn_decl.throws.len >= MAX_FN_THROWS) { |
| 1972 | error(p, "maximum number of thrown types exceeded"); |
| 1973 | return NULL; |
| 1974 | } |
| 1975 | node_t *thrown = parse_type(p); |
| 1976 | if (!thrown) |
| 1977 | return NULL; |
| 1978 | |
| 1979 | nodespan_push(p, &n->val.fn_decl.throws, thrown); |
| 1980 | } while (consume(p, T_COMMA)); |
| 1981 | } |
| 1982 | if (!expect(p, T_RPAREN, "expected `)` after throws clause")) |
| 1983 | return NULL; |
| 1984 | } |
| 1985 | |
| 1986 | /* For extern functions, expect semicolon instead of body */ |
| 1987 | if (is_extern) { |
| 1988 | if (!expect( |
| 1989 | p, T_SEMICOLON, "expected `;` after extern function declaration" |
| 1990 | )) |
| 1991 | return NULL; |
| 1992 | } else { |
| 1993 | if (!(n->val.fn_decl.body = parse_block(p))) |
| 1994 | return NULL; |
| 1995 | } |
| 1996 | return n; |
| 1997 | } |
| 1998 | |
| 1999 | /* Try to parse an annotation like `@default`. |
| 2000 | * Returns true if a known annotation was found and consumed. |
| 2001 | * Returns false if not an annotation (e.g. @sizeOf) - tokens not consumed. */ |
| 2002 | static bool try_parse_annotation(parser_t *p, attrib_t *attrs) { |
| 2003 | if (!check(p, T_AT_IDENT)) |
| 2004 | return false; |
| 2005 | |
| 2006 | /* Token is @identifier, skip the '@' to get the name. */ |
| 2007 | const char *name = p->current.start + 1; |
| 2008 | usize length = p->current.length - 1; |
| 2009 | |
| 2010 | if (length == 7 && !strncmp(name, "default", 7)) { |
| 2011 | advance(p); /* Consume `@default`. */ |
| 2012 | *attrs |= ATTRIB_DEFAULT; |
| 2013 | return true; |
| 2014 | } |
| 2015 | if (length == 4 && !strncmp(name, "test", 4)) { |
| 2016 | advance(p); /* Consume `@test`. */ |
| 2017 | *attrs |= ATTRIB_TEST; |
| 2018 | return true; |
| 2019 | } |
| 2020 | if (length == 9 && !strncmp(name, "intrinsic", 9)) { |
| 2021 | advance(p); /* Consume `@intrinsic`. */ |
| 2022 | *attrs |= ATTRIB_INTRINSIC; |
| 2023 | return true; |
| 2024 | } |
| 2025 | /* Not a known annotation - leave for parse_builtin to handle. */ |
| 2026 | return false; |
| 2027 | } |
| 2028 | |
| 2029 | /* Parse statement attributes. */ |
| 2030 | static node_t *parse_attribs(parser_t *p) { |
| 2031 | node_t *n = NULL; |
| 2032 | attrib_t attrs = ATTRIB_NONE; |
| 2033 | |
| 2034 | for (;;) { |
| 2035 | if (consume(p, T_PUB)) { |
| 2036 | if (attrs & ATTRIB_PUB) { |
| 2037 | error(p, "duplicate `pub` attribute"); |
| 2038 | return NULL; |
| 2039 | } |
| 2040 | attrs |= ATTRIB_PUB; |
| 2041 | } else if (try_parse_annotation(p, &attrs)) { |
| 2042 | /* Annotation was consumed, continue. */ |
| 2043 | } else if (consume(p, T_EXTERN)) { |
| 2044 | if (attrs & ATTRIB_EXTERN) { |
| 2045 | error(p, "duplicate `extern` attribute"); |
| 2046 | return NULL; |
| 2047 | } |
| 2048 | attrs |= ATTRIB_EXTERN; |
| 2049 | } else { |
| 2050 | break; |
| 2051 | } |
| 2052 | } |
| 2053 | |
| 2054 | if (attrs != ATTRIB_NONE) { |
| 2055 | n = node(p, NODE_ATTRIBUTE); |
| 2056 | n->val.attrib = attrs; |
| 2057 | } |
| 2058 | return n; |
| 2059 | } |
| 2060 | |
| 2061 | /* Parse a statement. */ |
| 2062 | static node_t *parse_stmt(parser_t *p) { |
| 2063 | /* Parse any attributes that come before the statement. */ |
| 2064 | node_t *attrs = parse_attribs(p); |
| 2065 | |
| 2066 | if (attrs) { |
| 2067 | switch (p->current.cls) { |
| 2068 | case T_FN: |
| 2069 | case T_UNION: |
| 2070 | case T_RECORD: |
| 2071 | case T_MOD: |
| 2072 | case T_CONST: |
| 2073 | case T_USE: |
| 2074 | break; |
| 2075 | default: |
| 2076 | error(p, "attributes are not allowed in this context"); |
| 2077 | return NULL; |
| 2078 | } |
| 2079 | |
| 2080 | /* Verify extern is only used with functions */ |
| 2081 | if ((attrs->val.attrib & ATTRIB_EXTERN) && p->current.cls != T_FN) { |
| 2082 | error( |
| 2083 | p, "extern attribute is only allowed on function declarations" |
| 2084 | ); |
| 2085 | return NULL; |
| 2086 | } |
| 2087 | } |
| 2088 | |
| 2089 | switch (p->current.cls) { |
| 2090 | case T_LBRACE: |
| 2091 | return parse_block(p); |
| 2092 | case T_LET: |
| 2093 | advance(p); |
| 2094 | if (consume(p, T_CASE)) { |
| 2095 | return parse_let_case(p); |
| 2096 | } |
| 2097 | if (consume(p, T_MUT)) { |
| 2098 | return parse_var(p, true); |
| 2099 | } |
| 2100 | return parse_var(p, false); |
| 2101 | case T_STATIC: |
| 2102 | advance(p); |
| 2103 | return parse_static(p); |
| 2104 | case T_CONST: |
| 2105 | advance(p); |
| 2106 | return parse_const(p); |
| 2107 | case T_USE: |
| 2108 | advance(p); |
| 2109 | return parse_use(p, attrs); |
| 2110 | case T_MOD: |
| 2111 | advance(p); |
| 2112 | return parse_mod(p, attrs); |
| 2113 | case T_RETURN: |
| 2114 | advance(p); |
| 2115 | return parse_return(p); |
| 2116 | case T_THROW: |
| 2117 | advance(p); |
| 2118 | return parse_throw(p); |
| 2119 | case T_BREAK: |
| 2120 | advance(p); |
| 2121 | return parse_break(p); |
| 2122 | case T_WHILE: |
| 2123 | advance(p); |
| 2124 | return parse_while(p); |
| 2125 | case T_FOR: |
| 2126 | advance(p); |
| 2127 | return parse_for(p); |
| 2128 | case T_LOOP: |
| 2129 | advance(p); |
| 2130 | return parse_loop(p); |
| 2131 | case T_IF: |
| 2132 | advance(p); |
| 2133 | return parse_if(p); |
| 2134 | case T_MATCH: |
| 2135 | advance(p); |
| 2136 | return parse_match(p); |
| 2137 | case T_FN: |
| 2138 | advance(p); |
| 2139 | return parse_fn(p, attrs); |
| 2140 | case T_UNION: |
| 2141 | advance(p); |
| 2142 | return parse_union(p, attrs); |
| 2143 | case T_RECORD: |
| 2144 | advance(p); |
| 2145 | return parse_record(p, attrs); |
| 2146 | case T_PANIC: |
| 2147 | advance(p); |
| 2148 | return parse_panic(p); |
| 2149 | default: |
| 2150 | break; |
| 2151 | } |
| 2152 | /* Parse an expression as a statement or an assignment statement. */ |
| 2153 | node_t *expr; |
| 2154 | |
| 2155 | if ((expr = parse_expr(p)) == NULL) |
| 2156 | return NULL; |
| 2157 | |
| 2158 | /* If we see an equals sign, this is an assignment statement */ |
| 2159 | if (consume(p, T_EQ)) { |
| 2160 | return parse_assignment(p, expr); |
| 2161 | } |
| 2162 | |
| 2163 | /* Create an expression statement node. */ |
| 2164 | node_t *stmt = node(p, NODE_EXPR_STMT); |
| 2165 | stmt->val.expr_stmt = expr; |
| 2166 | |
| 2167 | return stmt; |
| 2168 | } |
| 2169 | |
| 2170 | /* Parse a function argument, which may have an optional label. */ |
| 2171 | static node_t *parse_fn_call_arg(parser_t *p) { |
| 2172 | usize start = p->current.position; |
| 2173 | node_t *arg = node(p, NODE_CALL_ARG); |
| 2174 | |
| 2175 | /* Parse the expression first */ |
| 2176 | node_t *expr = parse_expr(p); |
| 2177 | if (!expr) |
| 2178 | return NULL; |
| 2179 | |
| 2180 | /* Check if this was an identifier followed by a colon |
| 2181 | * (making it a label), or the complete expression. */ |
| 2182 | if (expr->cls == NODE_IDENT && consume(p, T_COLON)) { |
| 2183 | /* It's a label, parse the actual value expression */ |
| 2184 | arg->val.call_arg.label = expr; |
| 2185 | |
| 2186 | if (!(arg->val.call_arg.expr = parse_expr(p))) { |
| 2187 | return NULL; |
| 2188 | } |
| 2189 | } else { |
| 2190 | arg->val.call_arg.label = NULL; |
| 2191 | arg->val.call_arg.expr = expr; |
| 2192 | } |
| 2193 | arg->offset = start; |
| 2194 | arg->length = p->previous.position + p->previous.length - start; |
| 2195 | |
| 2196 | return arg; |
| 2197 | } |
| 2198 | |
| 2199 | /* Parse an identifier. */ |
| 2200 | static node_t *parse_ident(parser_t *p, const char *error) { |
| 2201 | if (!expect(p, T_IDENT, error)) |
| 2202 | return NULL; |
| 2203 | |
| 2204 | node_t *ident = node(p, NODE_IDENT); |
| 2205 | |
| 2206 | ident->val.ident.name = p->previous.start; |
| 2207 | ident->val.ident.length = p->previous.length; |
| 2208 | |
| 2209 | return ident; |
| 2210 | } |
| 2211 | |
| 2212 | /* Parse either an identifier or a placeholder ('_'). */ |
| 2213 | static node_t *parse_ident_or_placeholder(parser_t *p, const char *error) { |
| 2214 | if (consume(p, T_UNDERSCORE)) { |
| 2215 | return node(p, NODE_PLACEHOLDER); |
| 2216 | } |
| 2217 | return parse_ident(p, error); |
| 2218 | } |
| 2219 | |
| 2220 | /* Parse a label. |
| 2221 | * Returns an identifier node. Expects IDENT followed by COLON. */ |
| 2222 | static node_t *parse_label(parser_t *p, const char *error) { |
| 2223 | if (!expect(p, T_IDENT, error)) |
| 2224 | return NULL; |
| 2225 | |
| 2226 | node_t *ident = node(p, NODE_IDENT); |
| 2227 | |
| 2228 | ident->val.ident.name = p->previous.start; |
| 2229 | ident->val.ident.length = p->previous.length; |
| 2230 | |
| 2231 | if (!expect(p, T_COLON, "expected ':' after identifier")) |
| 2232 | return NULL; |
| 2233 | |
| 2234 | return ident; |
| 2235 | } |
| 2236 | |
| 2237 | static node_t *parse_scope_segment(parser_t *p, const char *error) { |
| 2238 | if (check(p, T_SUPER)) { |
| 2239 | node_t *super_node = node(p, NODE_SUPER); |
| 2240 | advance(p); |
| 2241 | return super_node; |
| 2242 | } |
| 2243 | return parse_ident(p, error); |
| 2244 | } |
| 2245 | |
| 2246 | static node_t *parse_as_cast(parser_t *p, node_t *expr) { |
| 2247 | if (!consume(p, T_AS)) |
| 2248 | return NULL; |
| 2249 | |
| 2250 | node_t *as = node(p, NODE_AS); |
| 2251 | as->val.as_expr.expr = expr; |
| 2252 | |
| 2253 | /* Parse the target type */ |
| 2254 | node_t *typ = parse_type(p); |
| 2255 | if (!typ) |
| 2256 | return NULL; |
| 2257 | |
| 2258 | as->val.as_expr.type = typ; |
| 2259 | as->offset = expr->offset; |
| 2260 | as->length = p->current.position - as->offset; |
| 2261 | |
| 2262 | return as; |
| 2263 | } |
| 2264 | |
| 2265 | /* Parse postfix expressions (field access and array indexing). |
| 2266 | * |
| 2267 | * This function handles both field access (expr.field) and array indexing |
| 2268 | * (expr[index]) in a unified way, enabling arbitrarily complex nested |
| 2269 | * expressions like `x.y.z[1].w[2][3].q`. |
| 2270 | */ |
| 2271 | static node_t *parse_postfix(parser_t *p, node_t *expr) { |
| 2272 | node_t *result = expr; |
| 2273 | |
| 2274 | for (;;) { |
| 2275 | if (consume(p, T_DOT)) { /* Field access. */ |
| 2276 | node_t *n = node(p, NODE_ACCESS); |
| 2277 | n->val.access.lval = result; |
| 2278 | |
| 2279 | node_t *field = parse_ident(p, "expected field name after `.`"); |
| 2280 | if (!field) |
| 2281 | return NULL; |
| 2282 | |
| 2283 | field->val.ident.name = p->previous.start; |
| 2284 | field->val.ident.length = p->previous.length; |
| 2285 | n->val.access.rval = field; |
| 2286 | |
| 2287 | result = n; |
| 2288 | } else if (consume(p, T_DOT_DOT)) { |
| 2289 | node_t *range = node(p, NODE_RANGE); |
| 2290 | range->val.range.start = result; |
| 2291 | range->val.range.end = NULL; |
| 2292 | |
| 2293 | /* Check if there's a right-hand side for the range. */ |
| 2294 | if (!check(p, T_RBRACKET) && !check(p, T_SEMICOLON) && |
| 2295 | !check(p, T_COMMA) && !check(p, T_RPAREN) && |
| 2296 | !check(p, T_LBRACE)) { |
| 2297 | if (!(range->val.range.end = parse_expr(p))) { |
| 2298 | return NULL; |
| 2299 | } |
| 2300 | } |
| 2301 | result = range; |
| 2302 | } else if (consume(p, T_COLON_COLON)) { /* Scope access */ |
| 2303 | node_t *ident = |
| 2304 | parse_scope_segment(p, "expected identifier name after `::`"); |
| 2305 | if (!ident) |
| 2306 | return NULL; |
| 2307 | |
| 2308 | node_t *n = node(p, NODE_SCOPE); |
| 2309 | n->val.access.lval = result; |
| 2310 | n->val.access.rval = ident; |
| 2311 | |
| 2312 | result = n; |
| 2313 | } else if (consume(p, T_LBRACKET)) { /* Array indexing or slicing. */ |
| 2314 | node_t *expr = NULL; |
| 2315 | |
| 2316 | if (consume(p, T_DOT_DOT)) { |
| 2317 | /* Either `..` or `..n` */ |
| 2318 | /* Create range node with NULL start and end. */ |
| 2319 | expr = node(p, NODE_RANGE); |
| 2320 | expr->val.range.start = NULL; |
| 2321 | expr->val.range.end = NULL; |
| 2322 | |
| 2323 | if (!check(p, T_RBRACKET)) { |
| 2324 | if (!(expr->val.range.end = parse_expr(p))) { |
| 2325 | return NULL; |
| 2326 | } |
| 2327 | } |
| 2328 | } else { |
| 2329 | /* Either `n`, `n..` or `n..m` */ |
| 2330 | node_t *index = parse_expr(p); |
| 2331 | if (!index) |
| 2332 | return NULL; |
| 2333 | |
| 2334 | expr = index; |
| 2335 | } |
| 2336 | /* Create array index node with the index expression */ |
| 2337 | node_t *n = node(p, NODE_ARRAY_INDEX); |
| 2338 | n->val.access.lval = result; |
| 2339 | n->val.access.rval = expr; |
| 2340 | |
| 2341 | n->offset = result->offset; |
| 2342 | n->length = result->length; |
| 2343 | |
| 2344 | /* Expect closing bracket */ |
| 2345 | if (!expect(p, T_RBRACKET, "expected `]` after array index")) |
| 2346 | return NULL; |
| 2347 | |
| 2348 | result = n; |
| 2349 | } else if (consume(p, T_LPAREN)) { /* Parse function call. */ |
| 2350 | node_t *call = node(p, NODE_CALL); |
| 2351 | call->val.call.callee = result; |
| 2352 | call->val.call.args = nodespan_alloc(p, MAX_FN_PARAMS); |
| 2353 | |
| 2354 | node_t *arg = NULL; |
| 2355 | if (!check(p, T_RPAREN)) { |
| 2356 | do { |
| 2357 | if (!(arg = parse_fn_call_arg(p))) { |
| 2358 | return NULL; |
| 2359 | } |
| 2360 | nodespan_push(p, &call->val.call.args, arg); |
| 2361 | } while (consume(p, T_COMMA)); |
| 2362 | } |
| 2363 | if (!expect(p, T_RPAREN, "expected `)` after function arguments")) |
| 2364 | return NULL; |
| 2365 | |
| 2366 | result = call; |
| 2367 | } else if (p->context == PARSE_CTX_NORMAL && |
| 2368 | result->cls == NODE_SCOPE && check(p, T_LBRACE)) { |
| 2369 | /* Record literal after scope access: `Union::Variant { ... }`. */ |
| 2370 | advance(p); /* consume `{` */ |
| 2371 | |
| 2372 | node_t *literal = parse_record_lit(p, result); |
| 2373 | if (!literal) |
| 2374 | return NULL; |
| 2375 | |
| 2376 | result = literal; |
| 2377 | } else { |
| 2378 | /* No postfix operators to try. */ |
| 2379 | break; |
| 2380 | } |
| 2381 | } |
| 2382 | return result; |
| 2383 | } |
| 2384 | |
| 2385 | /* Parse a complete program, return the root of the AST, or `NULL` |
| 2386 | * if parsing failed. */ |
| 2387 | node_t *parser_parse(parser_t *p) { |
| 2388 | p->current = scanner_next(&p->scanner); |
| 2389 | |
| 2390 | /* Create a top-level module. */ |
| 2391 | node_t *root = parse_module_body(p); |
| 2392 | if (!root) |
| 2393 | return NULL; |
| 2394 | |
| 2395 | if (!expect(p, T_EOF, "expected end-of-file")) |
| 2396 | return NULL; |
| 2397 | |
| 2398 | root->length = (usize)(p->scanner.cursor - p->scanner.source); |
| 2399 | |
| 2400 | return (p->root = root); |
| 2401 | } |