skill中InputFileNode的输入无法通过api正常访问

[GuoMin]

api访问参考文档https://dataelem.feishu.cn/wiki/CpXNw6ZTjiKimjkv66Wcw0EWnHc

import requests
from typing import Optional

BASE_API_URL = "https://bisheng.dataelem.com/api/v1/process"
FLOW_ID = "e4917a84-fad7-4d5d-85fd-bf2ec3c42c26"
TWEAKS = {
    "CombineDocsChain-Pud2p": {},
    "RetrievalQA-qH6Mk": {},
    "OpenAIProxyEmbedding-yvld7": {},
    "PyPDFLoader-SIeEa": {},
    "InputFileNode-jdLkB": {},
    "Milvus-T3kRH": {"collection_name": "col_333"},
    "ChatOpenAI-qVQL6": {}
}


def upload_file():
    server = "https://bisheng.dataelem.com"
    url = server + '/api/v1/knowledge/upload'
    headers = {}
    files = {'file': open('/Users/sara/Desktop/20231201-223053.jpeg', 'rb')}
    res = requests.post(url, headers=headers, files=files)
    file_path = res.json()['data'].get('file_path', '')
    return file_path


def run_flow(inputs: dict, flow_id: str, file_path: str, tweaks: Optional[dict] = None) -> dict:
    api_url = f"{BASE_API_URL}/{flow_id}"
    payload = {"inputs": inputs}

    if tweaks:
        payload["tweaks"] = tweaks

    payload["inputs"]["file_path"] = file_path  # Add the file_path to the inputs
    
    if "InputFileNode-jdLkB" in tweaks:
        tweaks["InputFileNode-jdLkB"]["file_path"] = file_path
    response = requests.post(api_url, json=payload)
    return response.json()


# Get the file path from api_call
file_path = upload_file()

# Setup any tweaks you want to apply to the flow
inputs = {"query": "总结下文档内容", "id": "RetrievalQA-qH6Mk"}
print(run_flow(inputs, flow_id=FLOW_ID, file_path=file_path, tweaks=TWEAKS))

skill示例图如下

整体为一个pdf内容提取的小skill，在使用过程中，前端由于自动关闭了websocket，导致需要在聊天框输入任意内容触发返回

bisheng后端版本：0.2.3

观察日志与网页端的区别为 没有触发pdf的内容解析导致整个流程出错，进而导致api调用失败

[GuoMin]

0.2.6版本已经修复了