# LangChain創(chuàng)建ReAct智能體的典型示例
from langchain.agents import Tool, AgentExecutor, create_react_agent
from langchain.tools.ddg_search import DuckDuckGoSearchRun
from langchain_openai import ChatOpenAI

# 定義可調(diào)用的工具（此處為DuckDuckGo搜索）
search_tool = DuckDuckGoSearchRun()
tools = [
    Tool(
        name="Search",
        func=search_tool.run,
        descriptinotallow="用于搜索互聯(lián)網(wǎng)獲取最新信息"
    )
]

# 初始化語言模型
llm = ChatOpenAI(model="gpt-4")

# 創(chuàng)建ReAct模式智能體
agent = create_react_agent(llm, tools, "你是一個(gè) helpful 的AI助手。")
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

# 執(zhí)行智能體任務(wù)
response = agent_executor.invoke({"input": "AI智能體框架的最新發(fā)展有哪些？"})
print(response["output"])

# AutoGen實(shí)現(xiàn)多智能體協(xié)作的代碼示例
import autogen

# 配置LLM（此處使用GPT-4）
llm_config = {
    "config_list": [{"model": "gpt-4", "api_key": "your-api-key"}]
}

# 創(chuàng)建助手智能體
assistant = autogen.AssistantAgent(
    name="assistant",
    llm_cnotallow=llm_config,
    system_message="你是一個(gè) helpful 的AI助手。"
)

# 創(chuàng)建用戶代理智能體（支持人類介入）
user_proxy = autogen.UserProxyAgent(
    name="user_proxy",
    human_input_mode="TERMINATE",  # 任務(wù)完成時(shí)自動(dòng)終止
    max_consecutive_auto_reply=10,
    is_termination_msg=lambda x: x.get("content", "").rstrip().endswith("TERMINATE"),
    code_execution_cnotallow={"work_dir": "coding"}
)

# 啟動(dòng)智能體對(duì)話
user_proxy.initiate_chat(
    assistant,
    message="編寫一個(gè)計(jì)算斐波那契數(shù)列的Python函數(shù)。"
)

# CrewAI構(gòu)建智能體團(tuán)隊(duì)的示例代碼
from crewai import Agent, Task, Crew
from langchain_openai import ChatOpenAI

# 初始化語言模型
llm = ChatOpenAI(model="gpt-4")

# 定義研究員角色智能體
researcher = Agent(
    role="研究分析師",
    goal="發(fā)現(xiàn)并分析AI技術(shù)的最新趨勢(shì)",
    backstory="你是AI研究專家，對(duì)新興趨勢(shì)有敏銳洞察力",
    verbose=True,
    llm=llm
)

# 定義技術(shù)作家角色智能體
writer = Agent(
    role="技術(shù)作家",
    goal="基于研究結(jié)果創(chuàng)建全面報(bào)告",
    backstory="你是熟練的技術(shù)作家，能清晰解釋復(fù)雜概念",
    verbose=True,
    llm=llm
)

# 定義研究任務(wù)（依賴于后續(xù)的寫作任務(wù)）
research_task = Task(
    descriptinotallow="研究AI智能體框架的最新發(fā)展",
    expected_output="當(dāng)前AI智能體框架的全面分析",
    agent=researcher
)

# 定義寫作任務(wù)（依賴于研究任務(wù)的結(jié)果）
writing_task = Task(
    descriptinotallow="基于研究撰寫AI框架詳細(xì)報(bào)告",
    expected_output="結(jié)構(gòu)良好的AI框架報(bào)告",
    agent=writer,
    cnotallow=[research_task]
)

# 創(chuàng)建智能體團(tuán)隊(duì)并執(zhí)行任務(wù)
crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    verbose=True
)
result = crew.kickoff()
print(result)

# Semantic Kernel創(chuàng)建語義函數(shù)的示例
import semantic_kernel as sk
from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion

# 初始化內(nèi)核
kernel = sk.Kernel()

# 配置OpenAI服務(wù)
api_key = "your-api-key"
model = "gpt-4"
kernel.add_chat_service("chat_completion", OpenAIChatCompletion(model, api_key))

# 通過自然語言定義語義函數(shù)
prompt = """
生成一個(gè)關(guān)于{{$input}}的創(chuàng)意故事。
故事應(yīng)引人入勝，約100字。
"""

# 注冊(cè)語義函數(shù)
story_function = kernel.create_semantic_function(prompt, max_tokens=500)

# 執(zhí)行函數(shù)
result = story_function("一個(gè)學(xué)習(xí)繪畫的機(jī)器人")
print(result)

# 創(chuàng)建簡單智能體并規(guī)劃任務(wù)
from semantic_kernel.planning import ActionPlanner
planner = ActionPlanner(kernel)
plan = await planner.create_plan("寫一首關(guān)于人工智能的詩")
result = await plan.invoke()
print(result)

# LangGraph定義狀態(tài)圖工作流的示例
from typing import TypedDict, Annotated, Sequence
from langgraph.graph import StateGraph, END
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage, AIMessage

# 定義工作流狀態(tài)結(jié)構(gòu)
class AgentState(TypedDict):
    messages: Annotated[Sequence[HumanMessage | AIMessage], "對(duì)話中的消息"]
    next_step: Annotated[str, "下一步操作"]

# 初始化語言模型
llm = ChatOpenAI(model="gpt-4")

# 定義工作流節(jié)點(diǎn)（研究、分析、總結(jié)）
def research(state: AgentState) -> AgentState:
    messages = state["messages"]
    response = llm.invoke(messages + [HumanMessage(cnotallow="徹底研究這個(gè)主題。")])
    return {"messages": state["messages"] + [response], "next_step": "analyze"}

def analyze(state: AgentState) -> AgentState:
    messages = state["messages"]
    response = llm.invoke(messages + [HumanMessage(cnotallow="分析研究結(jié)果。")])
    return {"messages": state["messages"] + [response], "next_step": "conclude"}

def conclude(state: AgentState) -> AgentState:
    messages = state["messages"]
    response = llm.invoke(messages + [HumanMessage(cnotallow="基于分析提供結(jié)論。")])
    return {"messages": state["messages"] + [response], "next_step": "end"}

# 創(chuàng)建狀態(tài)圖并定義流程
workflow = StateGraph(AgentState)
workflow.add_node("research", research)
workflow.add_node("analyze", analyze)
workflow.add_node("conclude", conclude)
workflow.add_edge("research", "analyze")
workflow.add_edge("analyze", "conclude")
workflow.add_edge("conclude", END)
workflow.set_entry_point("research")

# 編譯并執(zhí)行工作流
agent = workflow.compile()
result = agent.invoke({
    "messages": [HumanMessage(cnotallow="告訴我關(guān)于AI智能體框架的信息")],
    "next_step": "research"
})

# 輸出最終結(jié)果
for message in result["messages"]:
    print(f"{message.type}: {message.content}\n")

# LlamaIndex創(chuàng)建數(shù)據(jù)驅(qū)動(dòng)智能體的示例
from llama_index.core.agent import FunctionCallingAgentWorker
from llama_index.core.tools import FunctionTool
from llama_index.llms.openai import OpenAI

# 定義文檔搜索工具函數(shù)
def search_documents(query: str) -> str:
    """在文檔數(shù)據(jù)庫中搜索信息。"""
    # 實(shí)際應(yīng)用中會(huì)查詢文檔存儲(chǔ)系統(tǒng)
    return f"以下是關(guān)于{query}的搜索結(jié)果：..."

# 創(chuàng)建函數(shù)工具
search_tool = FunctionTool.from_defaults(
    name="search_documents",
    fn=search_documents,
    descriptinotallow="在文檔數(shù)據(jù)庫中搜索信息"
)

# 初始化語言模型
llm = OpenAI(model="gpt-4")

# 創(chuàng)建智能體（支持函數(shù)調(diào)用）
agent = FunctionCallingAgentWorker.from_tools(
    [search_tool],
    llm=llm,
    verbose=True
)

# 執(zhí)行智能體任務(wù)
response = agent.chat("查找關(guān)于AI智能體框架的信息")
print(response)

# Atomic Agents實(shí)現(xiàn)原子化組件的示例
from pydantic import BaseModel, Field
from typing import List
import os

# 定義輸入輸出數(shù)據(jù)模型
class ResearchQuery(BaseModel):
    topic: str = Field(descriptinotallow="需要研究的主題")
    depth: int = Field(descriptinotallow="研究深度（1-5級(jí)）")

class ResearchResult(BaseModel):
    findings: List[str] = Field(descriptinotallow="研究的關(guān)鍵發(fā)現(xiàn)")
    sources: List[str] = Field(descriptinotallow="信息來源")

# 定義原子化研究組件
class ResearchAgent:
    def __init__(self, api_key: str):
        self.api_key = api_key

    def process(self, input_data: ResearchQuery) -> ResearchResult:
        print(f"正在深度{input_data.depth}研究{input_data.topic}")

        # 模擬研究結(jié)果
        findings = [
            f"關(guān)于{input_data.topic}的發(fā)現(xiàn)1",
            f"關(guān)于{input_data.topic}的發(fā)現(xiàn)2",
            f"關(guān)于{input_data.topic}的發(fā)現(xiàn)3"
        ]

        sources = ["https://github.com/...", "https://docs.ai/..."]
        return ResearchResult(findings=findings, sources=sources)

# 使用示例
if __name__ == "__main__":
    agent = ResearchAgent(api_key=os.environ.get("OPENAI_API_KEY", "default-key"))
    query = ResearchQuery(topic="AI智能體框架", depth=3)
    result = agent.process(query)

    print("\n研究發(fā)現(xiàn)：")

# Rasa項(xiàng)目結(jié)構(gòu)與核心代碼示例
# domain.yml - 定義智能體的領(lǐng)域知識(shí)
"""
version: "3.1"
intents:
  - greet         # 問候意圖
  - goodbye       # 告別意圖
  - ask_framework # 詢問框架意圖
responses:
  utter_greet:
    - text: "你好！今天需要了解AI框架的哪些方面？"
  utter_about_frameworks:
    - text: "主流AI智能體框架包括LangChain、AutoGen等，你想了解哪一個(gè)？"
entities:
  - framework_name  # 框架名稱實(shí)體
slots:
  framework_name:
    type: text
    mappings:
    - type: from_entity
      entity: framework_name
"""

# data/nlu.yml - 自然語言理解訓(xùn)練數(shù)據(jù)
"""
nlu:
- intent: greet
  examples: |
    - 嘿
    - 你好
    - 早上好
- intent: ask_framework
  examples: |
    - 給我講講AI框架
    - 最好的AI智能體框架有哪些
    - [LangChain](framework_name)的工作原理是什么
"""

# actions/actions.py - 自定義動(dòng)作實(shí)現(xiàn)
from rasa_sdk import Action, Tracker
from rasa_sdk.executor import CollectingDispatcher

class ActionFrameworkInfo(Action):
    def name(self) -> str:
        return "action_framework_info"

    def run(self, dispatcher, tracker, domain):
        framework = tracker.get_slot("framework_name")
        if framework.lower() == "langchain":
            dispatcher.utter_message(text="LangChain是用于構(gòu)建LLM應(yīng)用的開源框架...")
        # 其他框架的處理邏輯
        return []

# MetaGPT模擬軟件開發(fā)團(tuán)隊(duì)的示例
from metagpt.roles import ProjectManager, ProductManager, Architect, Engineer
from metagpt.team import Team
import asyncio

async def main():
    # 定義項(xiàng)目需求
    requirement = "開發(fā)一個(gè)允許用戶搜索和比較AI智能體框架的Web應(yīng)用"

    # 創(chuàng)建不同角色的智能體
    product_manager = ProductManager()
    project_manager = ProjectManager()
    architect = Architect()
    engineer = Engineer()

    # 組建開發(fā)團(tuán)隊(duì)
    team = Team(
        name="AI框架探索團(tuán)隊(duì)",
        members=[product_manager, project_manager, architect, engineer]
    )

    # 啟動(dòng)團(tuán)隊(duì)工作
    await team.run(requirement)
    # 輸出結(jié)果將包括PRD、設(shè)計(jì)文檔、代碼和測(cè)試用例

# Camel-AI實(shí)現(xiàn)智能體對(duì)話的示例
from camel.agents import ChatAgent
from camel.messages import BaseMessage
from camel.typing import ModelType
import asyncio

async def main():
    # 創(chuàng)建用戶智能體（需要數(shù)據(jù)分析幫助）
    user_agent = ChatAgent(
        model_type=ModelType.GPT_4,
        system_message="你是需要幫助分析AI框架數(shù)據(jù)的用戶。"
    )

    # 創(chuàng)建助手智能體（數(shù)據(jù)分析專家）
    assistant_agent = ChatAgent(
        model_type=ModelType.GPT_4,
        system_message="你是專注于數(shù)據(jù)分析和AI框架的AI助手。"
    )

    # 發(fā)起對(duì)話
    initial_message = BaseMessage.make_user_message(
        role_name="User",
        cnotallow="我需要比較不同AI智能體框架的特性，你能幫我分析嗎？"
    )

    # 多輪對(duì)話模擬
    assistant_response = await assistant_agent.step(initial_message)
    print(f"助手: {assistant_response.content}\n")

    for _ in range(3):
        user_response = await user_agent.step(assistant_response)
        print(f"用戶: {user_response.content}\n")
        assistant_response = await assistant_agent.step(user_response)
        print(f"助手: {assistant_response.content}\n")