Function main

Python/requests/pdf2file.py:76–121

(file)

def main(file):
    # 上传文件并等待解析完成
    upload_data = preupload()
    print(upload_data)
    url, uid = upload_data["url"], upload_data["uid"]

    put_file(file, url)

    for _ in range(100):
        status_data = get_status(uid)
        status = status_data.get("status")
        if status == "success":
            print("Save result to result.json")
            with open("result.json", "w") as f:
                json.dump(status_data["result"], f)
            break
        elif status == "failed":
            print(status_data)
            raise Exception(f"parse failed: {status_data.get('detail')}")
        elif status == "processing":
            print(status_data)
            print(f"progress: {status_data.get('progress')}")
            time.sleep(3)
    else:
        raise Exception(f"Fails to deal with uid: {uid} after 100 retries")

    # 导出文件
    print("Start exporting file...")
    export_file(uid, "docx", "normal")  # 可以根据需要修改格式和公式模式

    for _ in range(100):
        result_data = get_export_result(uid)
        status = result_data.get("status")
        if status == "success":
            file_url = result_data["url"]
            output_path = "output.docx"  # 根据实际格式修改扩展名
            print(f"Downloading file to {output_path}")
            download_file(file_url, output_path)
            return
        elif status == "failed":
            print(result_data)
            raise Exception("Export failed")
        elif status == "processing":
            print("Export processing...")
            time.sleep(3)
    raise Exception(f"Export timeout with uid: {uid} after 100 retries")


if __name__ == "__main__":