Main dispatch: drains pending queue, then routes each byte via BYTE_CLASS / SINGLE_TOK lookups. */
(&mut self)
| 343 | |
| 344 | /* Main dispatch: drains pending queue, then routes each byte via BYTE_CLASS / SINGLE_TOK lookups. */ |
| 345 | pub fn next_token(&mut self) -> Option<(TokenType, usize, usize, usize)> { |
| 346 | if let Some(tok) = self.pending.pop() { |
| 347 | return Some(tok); |
| 348 | } |
| 349 | |
| 350 | self.skip_whitespace(); |
| 351 | if self.pos >= self.src.len() { return None; } |
| 352 | |
| 353 | let start = self.pos; |
| 354 | let line_at_start = self.line; |
| 355 | let b = self.src[self.pos]; |
| 356 | let cl = BYTE_CLASS[b as usize]; |
| 357 | |
| 358 | // Newline |
| 359 | if b == b'\n' { |
| 360 | self.pos += 1; |
| 361 | self.handle_newline(start); |
| 362 | return self.pending.pop(); |
| 363 | } |
| 364 | |
| 365 | if b == b'#' { // inside an f-string interpolation, '#' is part of the format spec, skip silently |
| 366 | if !self.fstring_stack.is_empty() { |
| 367 | self.pos += 1; |
| 368 | return self.next_token(); |
| 369 | } |
| 370 | while self.pos < self.src.len() && self.src[self.pos] != b'\n' { |
| 371 | self.pos += 1; |
| 372 | } |
| 373 | return Some((TokenType::Comment, line_at_start, start, self.pos)); |
| 374 | } |
| 375 | |
| 376 | // Identifier / keyword / string-prefix / f-string-prefix |
| 377 | if cl & ID_START != 0 { |
| 378 | self.pos += if b >= 0x80 { utf8_char_len(b) } else { 1 }; |
| 379 | self.scan_id_rest(); |
| 380 | let slice = &self.src[start..self.pos]; |
| 381 | |
| 382 | if is_fstring_prefix(slice) |
| 383 | && let Some(&q) = self.src.get(self.pos) |
| 384 | && (q == b'"' || q == b'\'') |
| 385 | { |
| 386 | let pe = self.pos; |
| 387 | self.start_fstring(start, pe); |
| 388 | return self.pending.pop(); |
| 389 | } |
| 390 | |
| 391 | if is_string_prefix(slice) |
| 392 | && let Some(&q) = self.src.get(self.pos) |
| 393 | && (q == b'"' || q == b'\'') |
| 394 | { |
| 395 | let q_start = self.pos; |
| 396 | self.pos += 1; |
| 397 | self.scan_string(q, q_start); |
| 398 | return Some((TokenType::String, line_at_start, start, self.pos)); |
| 399 | } |
| 400 | |
| 401 | /* Bytes literal (b/B, optional r/R for raw): scans like a string, the parser decodes bytes-specific escapes later. */ |
| 402 | if is_bytes_prefix(slice) |
no test coverage detected