webapp/app/services/llm_service.py

import aiohttp
import asyncio
import json
import os
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import select
from app.core.config import settings
from app.schemas.llm import ParseUserInputsPayload, ParseUserInputsOut
from app.models.schema import VasSchema

# --- 配置区 ---
# 请换成你新生成的 API Key
API_KEY = settings.openai_api_key
API_URL = "https://api.openai.com/v1/chat/completions"


class LlmService:

    async def handle_parse(db: AsyncSession, payload: ParseUserInputsPayload):
        stmt = select(VasSchema).where(VasSchema.id == payload.schema_id)
        obj = (await db.execute(stmt)).scalar_one_or_none()
        if not obj:
            raise NotFoundError("Schema not exist")
        parsed_obj = await LlmService.parse_data_async(payload.input_raw_str, obj.schema_content)
        out = ParseUserInputsOut(parsed_obj=parsed_obj)
        return out
    
    @staticmethod
    async def parse_data_async(user_text: str, json_schema: dict):
        """
        [异步版本] 调用 LLM 解析数据
        """
        headers = {
            "Authorization": f"Bearer {API_KEY}",
            "Content-Type": "application/json"
        }

        # 构造 Prompt
        system_instruction = "You are a specialized data extraction API. Output valid JSON only."
        
        user_prompt = f"""
        Extract data from the text strictly based on the provided JSON Schema.
        
        [JSON Schema]
        {json.dumps(json_schema)}
        
        [User Text]
        {user_text}
        """

        payload = {
            "model": "gpt-4o",  # 或 gpt-3.5-turbo
            "messages": [
                {"role": "system", "content": system_instruction},
                {"role": "user", "content": user_prompt}
            ],
            "temperature": 0,
            "response_format": {"type": "json_object"} # 强制 JSON 模式
        }

  
        async with aiohttp.ClientSession() as session:
            async with session.post(API_URL, headers=headers, json=payload, timeout=30) as response:
                
                # 1. 检查 HTTP 状态码
                if response.status != 200:
                    error_text = await response.text()
                    return {"error": f"HTTP {response.status}", "detail": error_text}
                
                # 2. 获取响应体
                result = await response.json()
                
                # 3. 提取并解析内容
                content_str = result['choices'][0]['message']['content']
                return json.loads(content_str)

# --- 测试运行 ---
if __name__ == "__main__":
    # 定义你的 Schema
    my_schema = {
        "type": "object",
        "properties": {
            "full_name": {"type": "string"},
            "budget": {"type": "integer"},
            "items": {"type": "array", "items": {"type": "string"}}
        },
        "required": ["full_name", "budget"]
    }

    # 模拟用户输入
    user_input = "我是张三，打算花2000块钱买个耳机和键盘。"
    
    print("正在解析...")
    result = parse_data_api(user_input, my_schema)
    print(json.dumps(result, ensure_ascii=False, indent=2))