解析文档，支持单个路径、路径列表或文件 meta 列表
(self, **state)
| 82 | self._audio_service = audio_service |
| 83 | |
| 84 | def process(self, **state) -> Dict[str, Any]: |
| 85 | """解析文档，支持单个路径、路径列表或文件 meta 列表""" |
| 86 | doc_input = state.get(self.input_key) |
| 87 | if not doc_input: |
| 88 | logger.warning(f"DocumentNode: 未找到文档路径 (key={self.input_key})") |
| 89 | return {self.get_output_key(): ""} |
| 90 | |
| 91 | # 支持多种输入格式: |
| 92 | # str — 单个文件路径 (File 类型) |
| 93 | # list[str] — 文件路径列表 |
| 94 | # list[dict] — 文件 meta 列表 (Files 类型: [{file_path, original_name, ...}]) |
| 95 | if isinstance(doc_input, list): |
| 96 | file_infos = self._normalize_file_inputs(doc_input) |
| 97 | results = [self._parse_file(info["file_path"]) for info in file_infos if info.get("file_path")] |
| 98 | combined = "\n\n---\n\n".join(r for r in results if r) |
| 99 | return {self.get_output_key(): combined} |
| 100 | else: |
| 101 | file_info = self._normalize_file_input(doc_input) |
| 102 | content = self._parse_file(file_info["file_path"]) if file_info.get("file_path") else "" |
| 103 | return {self.get_output_key(): content} |
| 104 | |
| 105 | async def async_execute(self, state: dict, context) -> Dict[str, Any]: |
| 106 | """解析文档；配置 ASR 时音频文件走异步转写。""" |
no test coverage detected