| 362 | ) |
| 363 | |
| 364 | def _visit_pattern(self, pattern: str, name: str) -> str: |
| 365 | assert pattern.startswith("^") and pattern.endswith("$"), ( |
| 366 | 'Pattern must start with "^" and end with "$"' |
| 367 | ) |
| 368 | pattern = pattern[1:-1] |
| 369 | sub_rule_ids: Dict[str, str] = {} |
| 370 | index = 0 |
| 371 | length = len(pattern) |
| 372 | |
| 373 | def to_rule(item: Tuple[str, bool]) -> str: |
| 374 | text, is_literal = item |
| 375 | return f'"{text}"' if is_literal else text |
| 376 | |
| 377 | def transform() -> Tuple[str, bool]: |
| 378 | nonlocal index |
| 379 | start = index |
| 380 | sequence: List[Tuple[str, bool]] = [] |
| 381 | |
| 382 | def get_dot() -> str: |
| 383 | rule = self.DOTALL if self._dotall else self.DOT |
| 384 | return self._add_rule("dot", rule) |
| 385 | |
| 386 | def join_sequence() -> Tuple[str, bool]: |
| 387 | if len(sequence) == 1: |
| 388 | return sequence[0] |
| 389 | return (" ".join(to_rule(item) for item in sequence), False) |
| 390 | |
| 391 | while index < length: |
| 392 | char = pattern[index] |
| 393 | if char == ".": |
| 394 | sequence.append((get_dot(), False)) |
| 395 | index += 1 |
| 396 | elif char == "(": |
| 397 | index += 1 |
| 398 | if index < length: |
| 399 | assert pattern[index] != "?", ( |
| 400 | f'Unsupported pattern syntax "{pattern[index]}" at index {index} of /{pattern}/' |
| 401 | ) |
| 402 | sequence.append((f"({to_rule(transform())})", False)) |
| 403 | elif char == ")": |
| 404 | index += 1 |
| 405 | assert start > 0 and pattern[start - 1] == "(", ( |
| 406 | f"Unbalanced parentheses; start = {start}, index = {index}, pattern = {pattern}" |
| 407 | ) |
| 408 | return join_sequence() |
| 409 | elif char == "[": |
| 410 | square_brackets = char |
| 411 | index += 1 |
| 412 | while index < length and pattern[index] != "]": |
| 413 | if pattern[index] == "\\": |
| 414 | square_brackets += pattern[index : index + 2] |
| 415 | index += 2 |
| 416 | else: |
| 417 | square_brackets += pattern[index] |
| 418 | index += 1 |
| 419 | assert index < length, ( |
| 420 | f"Unbalanced square brackets; start = {start}, index = {index}, pattern = {pattern}" |
| 421 | ) |