| 354 | } |
| 355 | |
| 356 | int cbm_lex(const char *input, cbm_lex_result_t *out) { |
| 357 | memset(out, 0, sizeof(*out)); |
| 358 | if (!input) { |
| 359 | return CBM_NOT_FOUND; |
| 360 | } |
| 361 | |
| 362 | int len = (int)strlen(input); |
| 363 | int i = 0; |
| 364 | |
| 365 | while (i < len) { |
| 366 | if (lex_skip_whitespace_comments(input, len, &i)) { |
| 367 | continue; |
| 368 | } |
| 369 | |
| 370 | char c = input[i]; |
| 371 | |
| 372 | /* String literals */ |
| 373 | if (c == '"' || c == '\'') { |
| 374 | char quote = c; |
| 375 | i++; |
| 376 | lex_string_literal(input, len, &i, quote, out); |
| 377 | continue; |
| 378 | } |
| 379 | |
| 380 | /* Numbers — stop before ".." (DOTDOT operator) */ |
| 381 | if (lex_try_number(input, len, &i, out)) { |
| 382 | continue; |
| 383 | } |
| 384 | |
| 385 | /* Identifiers / keywords */ |
| 386 | if (lex_try_ident(input, len, &i, out)) { |
| 387 | continue; |
| 388 | } |
| 389 | |
| 390 | /* Two-character tokens */ |
| 391 | { |
| 392 | bool found_two = lex_try_two_char(input, len, &i, out); |
| 393 | if (found_two) { |
| 394 | continue; |
| 395 | } |
| 396 | } |
| 397 | |
| 398 | /* Single-character tokens */ |
| 399 | cbm_token_type_t stype = lex_single_char(c); |
| 400 | |
| 401 | if (stype != TOK_EOF) { |
| 402 | char buf[PAIR_LEN] = {c, '\0'}; |
| 403 | lex_push(out, stype, buf, i); |
| 404 | i++; |
| 405 | continue; |
| 406 | } |
| 407 | |
| 408 | /* Unknown character — skip */ |
| 409 | i++; |
| 410 | } |
| 411 | |
| 412 | /* Add EOF */ |
| 413 | lex_push(out, TOK_EOF, "", i); |
no test coverage detected