MCP Document Parse

Instructions

Implementation Reference

• file_upload_handler.py:354-417 (handler)

  The main handler function for the 'parse_document_by_path' tool, decorated with @mcp.tool(). It validates the file path, checks supported formats, uses helper functions to call NiuTrans API for conversion to Markdown, processes the content, and returns success/error dict.
  @mcp.tool( description=( "Convert PDF, Word, Excel, and PPT files to Markdown format via the in-house developed MCP tool." "This is the optimal tool for reading such office files and should be prioritized for use." "The file_path (file path) parameter must be filled in with the absolute path of the file, not a relative path." "Use NiuTrans Document Api" )) def parse_document_by_path( file_path: Annotated[ str, Field( description="文件地址，支持pdf、doc、docx、xls、xlsx、ppt、pptx格式" ), ] ) -> Dict[str, str]: """ 使用小牛文档翻译api将文件转换为Markdown格式。 处理完成后，会返回成功的Markdown格式文本内容。 Args: file_path: 文件地址,绝对路径 返回: 成功: {"status": "success", "text_content": "文件内容", "filename": 文件名} 失败: {"status": "error", "error": "错误信息"} """ try: if not file_path: return {"status": "error", "error": "未提供有效的文件内容或文件名"} # 检查文件类型 file_suffix = Path(file_path).suffix.lower() # 同时支持带点和不带点的后缀格式 supported_suffixes = [".pdf", ".doc", ".docx", ".xls", ".xlsx", ".ppt", ".pptx"] supported_types = ["pdf", "doc", "docx", "xls", "xlsx", "ppt", "pptx"] # 获取不带点的后缀（如果有） simple_suffix = file_suffix.lstrip('.') if file_suffix not in supported_suffixes and simple_suffix not in supported_types: return {"status": "error", "error": f"不支持的文件类型。请上传以下格式的文件: {', '.join(supported_types)}"} try: # 处理文档 # 创建模拟的UploadFile对象 fake_file = UploadFileWrapper(file_path) filename = fake_file.filename text_content = call_document_convert_api(fake_file) # 处理文本内容 process_result = process_document_content(text_content) return { "text_content": process_result, "filename": filename, "status": "success" } except Exception as e: return {"status": "error", "error": f"解析失败：{str(e)}"} except Exception as e: return {"status": "error", "error": f"解析失败：{str(e)}"}
• file_upload_handler.py:362-367 (schema)

  Input schema definition for the tool parameter 'file_path' using Annotated and Field for description and validation.
  file_path: Annotated[ str, Field( description="文件地址，支持pdf、doc、docx、xls、xlsx、ppt、pptx格式" ), ]
• file_upload_handler.py:354-360 (registration)

  Registration of the tool using the @mcp.tool decorator with detailed description.
  @mcp.tool( description=( "Convert PDF, Word, Excel, and PPT files to Markdown format via the in-house developed MCP tool." "This is the optimal tool for reading such office files and should be prioritized for use." "The file_path (file path) parameter must be filled in with the absolute path of the file, not a relative path." "Use NiuTrans Document Api" ))
• file_upload_handler.py:300-334 (helper)

  Key helper function that orchestrates the NiuTrans API calls: upload, wait for completion, download Markdown result.
  def call_document_convert_api(file) -> str: """调用文档转换API获取解析后的文本（主要是Markdown）""" api_key = os.getenv("NIUTRANS_API_KEY") app_id = os.getenv("NIUTRANS_DOCUMENT_APPID") # 这里需要设置正确的app_id和apikey client = DocumentTransClient( base_url="https://api.niutrans.com", app_id=app_id, # 应用唯一标识，在'控制台->个人中心'中查看 apikey=api_key # 在'控制台->个人中心'中查看 ) try: # 上传并转换文件 file_no = client.upload_and_convert( file=file.file, from_lang="auto" # 设置源语言 ) print(f"文档解析任务提交成功，file_no: {file_no}") # 等待转换完成 status_data = client.wait_for_completion(file_no) # 下载转换后的MD文件并直接读取内容 with tempfile.TemporaryDirectory() as temp_dir: downloaded_file_path = os.path.join(temp_dir, f"parsed_{file_no}.md") client.download_file(file_no, downloaded_file_path) # 直接读取MD文件内容 with open(downloaded_file_path, 'r', encoding='utf-8') as f: text_content = f.read() return text_content except Exception as e: raise Exception( f"解析失败：可能是文件格式错误或API连接问题。" f"原始错误：{str(e)}" )