大模型应用开发：从调用API到构建智能产品

from langchain.text_splitter import RecursiveCharacterTextSplitter

# 文档分块：按语义边界切分
text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=1000,           # 每块大小
    chunk_overlap=200,         # 重叠部分，保持上下文连贯
    separators=["\n\n", "\n", "。", "！", "？", "；", " ", ""]
)
chunks = text_splitter.split_text(document)

from openai import OpenAI

client = OpenAI()

def rag_answer(question, retrieved_docs):
    """基于检索结果生成答案"""
    context = "\n\n".join(retrieved_docs)
    
    prompt = f"""请根据以下参考资料回答问题。
如果参考资料中没有相关信息，请直接说"抱歉，我没有找到相关信息"。

【参考资料】
{context}

【问题】
{question}

【回答】"""

    response = client.chat.completions.create(
        model="gpt-4",
        messages=[{"role": "user", "content": prompt}],
        temperature=0.3
    )
    return response.choices[0].message.content

import json

def get_weather(city: str) -> str:
    """获取城市天气信息"""
    # 实际实现中调用天气API
    return f"{city}：晴，25°C"

def search_news(topic: str, limit: int = 3) -> str:
    """搜索新闻"""
    # 实际实现中调用新闻API
    return f"关于{topic}的最新新闻：..."

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "获取指定城市的当前天气",
            "parameters": {
                "type": "object",
                "properties": {
                    "city": {"type": "string", "description": "城市名称"}
                },
                "required": ["city"]
            }
        }
    },
    {
        "type": "function",
        "function": {
            "name": "search_news",
            "description": "搜索最新新闻",
            "parameters": {
                "type": "object",
                "properties": {
                    "topic": {"type": "string", "description": "新闻主题"},
                    "limit": {"type": "integer", "description": "返回条数"}
                },
                "required": ["topic"]
            }
        }
    }
]

def agent_loop(user_request):
    messages = [{"role": "user", "content": user_request}]
    max_iterations = 10
    
    for _ in range(max_iterations):
        # 第一步：模型思考并决定调用哪个工具
        response = client.chat.completions.create(
            model="gpt-4",
            messages=messages,
            tools=tools,
            tool_choice="auto"
        )
        
        message = response.choices[0].message
        messages.append(message)
        
        # 第二步：如果没有工具调用请求，说明任务完成
        if not message.tool_calls:
            return message.content
        
        # 第三步：执行工具调用
        for tool_call in message.tool_calls:
            func_name = tool_call.function.name
            args = json.loads(tool_call.function.arguments)
            
            if func_name == "get_weather":
                result = get_weather(args["city"])
            elif func_name == "search_news":
                result = search_news(args["topic"], args.get("limit", 3))
            else:
                result = f"未知工具: {func_name}"
            
            # 将工具结果加入对话
            messages.append({
                "role": "tool",
                "tool_call_id": tool_call.id,
                "content": result
            })
    
    return "任务执行超时，请简化请求后重试"

# docker-compose.yml
version: "3.8"
services:
  chromadb:
    image: ghcr.io/chroma-core/chroma:latest
    ports:
      - "8000:8000"
    environment:
      - IS_PERSISTENT=TRUE
    volumes:
      - ./chroma_data:/chroma/chroma
    command: uvicorn chromadb.app:app --reload --workers 1 --host 0.0.0.0 --port 8000