import os
import sys
import requests
sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from utils.enums import ChatRole, LLMRole
from utils.config import Config
from models.message import MessageRepository
def parse_msgs(msgHistory):
"""
解析前端返回消息记录，转换为提供给大模型的格式
"""
results = []
for msg in msgHistory:
role = LLMRole.ASSISTANT.value if msg["role"] == "AI" else LLMRole.USER.value
results.append({"role": role, "content": msg["content"]})
return results
async def save_chat_history(
agent_id,
conversation_id,
user_id,
role: ChatRole,
content,
is_abort=False,
access_token: str = None,
session_id_xx="sessionId",
chanelId="PC",
parent_id=None,
):
"""
保存对话记录
"""
role = "USER" if role == ChatRole.USER.value else "AI"
repo = MessageRepository()
message_id = await repo.add_message(conversation_id, role, content)
return message_id
async def update_chat_history(
message_id,
content,
access_token: str = None,
):
"""
更新对话记录
"""
repo = MessageRepository()
await repo.update_message(message_id, content)
# print(f"{response.text}")
async def get_message_history(
conversation_id, top_num, add_one, convert_type, access_token
):
"""
获取消息记录
"""
repo = MessageRepository()
data = await repo.get_messages(conversation_id)
if isinstance(data, list):
if add_one == 1 or add_one == "1":
# 倒序
data.reverse()
data = data[0 : len(data) - 1]
# print(data)
# 转换为llm对话格式的消息记录
if convert_type == "llm":
data = parse_msgs(data)
# 转换为str格式的消息记录，直接加入到提示词
elif convert_type == "str":
msgs = ""
for d in data:
msgs += f"{d['msgType']}: {d['msg']}\n"
data = msgs
# print(f"{response.json()}")
return data
