app_simplified.py

import gradio as gr
import requests
import json
import os
from huggingface_hub import InferenceClient

# ========== 配置两个 MCP 服务 ==========
MCP_SERVICES = {
    "financial": {
        "name": "SEC Financial Reports",
        "url": "https://jc321-easyreportdatamcp.hf.space/mcp",
        "tools": [
            {
                "type": "function",
                "function": {
                    "name": "advanced_search_company",
                    "description": "Search for US listed companies by name or ticker to get CIK and basic info",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "company_input": {"type": "string", "description": "Company name or ticker (e.g., 'Apple', 'AAPL')"}
                        },
                        "required": ["company_input"]
                    }
                }
            },
            {
                "type": "function",
                "function": {
                    "name": "get_latest_financial_data",
                    "description": "Get latest financial data (revenue, net income, EPS, etc.) for a company",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "cik": {"type": "string", "description": "10-digit CIK number"}
                        },
                        "required": ["cik"]
                    }
                }
            },
            {
                "type": "function",
                "function": {
                    "name": "extract_financial_metrics",
                    "description": "Get multi-year financial trends (3 or 5 years)",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "cik": {"type": "string", "description": "10-digit CIK number"},
                            "years": {"type": "integer", "enum": [3, 5]}
                        },
                        "required": ["cik", "years"]
                    }
                }
            }
        ]
    },
    "market": {
        "name": "Market & Stock Data",
        "url": "https://jc321-marketandstockmcp.hf.space/mcp",
        "tools": []  # 需要获取实际工具列表
    }
}

# 合并所有工具
ALL_TOOLS = []
TOOL_ROUTING = {}  # tool_name -> mcp_url 的映射

for service_key, service in MCP_SERVICES.items():
    for tool in service["tools"]:
        tool_name = tool["function"]["name"]
        ALL_TOOLS.append(tool)
        TOOL_ROUTING[tool_name] = service["url"]

# ========== 初始化 LLM 客户端 ==========
hf_token = os.environ.get("HF_TOKEN") or os.environ.get("HUGGING_FACE_HUB_TOKEN")
client = InferenceClient(api_key=hf_token) if hf_token else InferenceClient()
print(f"✅ LLM client initialized (Qwen/Qwen2.5-72B-Instruct:novita)")

# ========== 系统提示词 ==========
SYSTEM_PROMPT = """You are an intelligent financial and market analysis assistant with access to real-time data.

You have access to two powerful data sources:
1. **SEC Financial Reports** - Get official financial data for US-listed companies (revenue, earnings, cash flow, etc.)
2. **Market & Stock Data** - Get real-time market data, stock prices, and market analysis

When users ask about:
- Company financials, earnings, revenue → Use SEC Financial Reports tools
- Stock prices, market trends, trading data → Use Market & Stock Data tools

Be conversational, insightful, and provide data-driven analysis. Automatically fetch data when needed."""

# ========== 核心函数：调用 MCP 工具 ==========
def call_mcp_tool(tool_name, arguments):
    """调用 MCP 工具"""
    mcp_url = TOOL_ROUTING.get(tool_name)
    if not mcp_url:
        return {"error": f"Unknown tool: {tool_name}"}
    
    try:
        response = requests.post(
            mcp_url,
            json={
                "jsonrpc": "2.0",
                "method": "tools/call",
                "params": {"name": tool_name, "arguments": arguments},
                "id": 1
            },
            headers={"Content-Type": "application/json"},
            timeout=60
        )
        
        if response.status_code == 200:
            return response.json()
        else:
            return {"error": f"HTTP {response.status_code}", "detail": response.text[:200]}
    except Exception as e:
        return {"error": str(e)}

# ========== 核心函数：AI 助手 ==========
def chatbot_response(message, history):
    """AI 助手主函数"""
    try:
        # 构建消息历史
        messages = [{"role": "system", "content": SYSTEM_PROMPT}]
        
        # 添加对话历史（最近5轮）
        if history:
            for item in history[-5:]:
                if isinstance(item, dict):
                    messages.append(item)
                elif isinstance(item, (list, tuple)) and len(item) == 2:
                    user_msg, assistant_msg = item
                    messages.append({"role": "user", "content": user_msg})
                    messages.append({"role": "assistant", "content": assistant_msg})
        
        messages.append({"role": "user", "content": message})
        
        # LLM 调用循环（支持多轮工具调用）
        tool_calls_log = []
        max_iterations = 5
        
        for iteration in range(max_iterations):
            # 调用 LLM
            response = client.chat_completion(
                messages=messages,
                model="Qwen/Qwen2.5-72B-Instruct:novita",
                tools=ALL_TOOLS,
                max_tokens=3000,
                temperature=0.7,
                tool_choice="auto"
            )
            
            choice = response.choices[0]
            
            # 检查是否有工具调用
            if choice.message.tool_calls:
                messages.append(choice.message)
                
                for tool_call in choice.message.tool_calls:
                    tool_name = tool_call.function.name
                    tool_args = json.loads(tool_call.function.arguments)
                    
                    # 记录工具调用
                    tool_calls_log.append({"name": tool_name, "arguments": tool_args})
                    
                    # 调用 MCP 工具
                    tool_result = call_mcp_tool(tool_name, tool_args)
                    
                    # 添加工具结果到消息
                    messages.append({
                        "role": "tool",
                        "name": tool_name,
                        "content": json.dumps(tool_result),
                        "tool_call_id": tool_call.id
                    })
                
                continue  # 继续下一轮
            else:
                # 无工具调用，返回最终答案
                response_text = choice.message.content
                break
        
        # 构建最终响应
        final_response = ""
        
        # 显示模型信息
        final_response += f"<div style='padding: 8px; background: #e3f2fd; border-left: 3px solid #2196f3; margin-bottom: 10px; font-size: 0.9em;'>🤖 <strong>Model:</strong> Qwen/Qwen2.5-72B-Instruct:novita</div>\n\n"
        
        # 显示工具调用日志
        if tool_calls_log:
            final_response += "**🛠️ MCP Tools Used:**\n\n"
            for i, tool_call in enumerate(tool_calls_log, 1):
                final_response += f"{i}. `{tool_call['name']}` - {json.dumps(tool_call['arguments'])}\n"
            final_response += "\n---\n\n"
        
        final_response += response_text
        
        return final_response
        
    except Exception as e:
        return f"❌ Error: {str(e)}"

# ========== Gradio 界面 ==========
with gr.Blocks(title="Financial & Market AI Assistant") as demo:
    gr.Markdown("# 🤖 Financial & Market AI Assistant")
    
    gr.Markdown("""
    <div style='padding: 15px; background: #d4edda; border-left: 4px solid #28a745; margin: 10px 0; border-radius: 4px;'>
        <strong>✅ AI Powered by:</strong> Qwen/Qwen2.5-72B-Instruct:novita
        <br>
        <strong>📊 Data Sources:</strong> SEC Financial Reports + Market & Stock Data
    </div>
    """)
    
    chat = gr.ChatInterface(
        fn=chatbot_response,
        examples=[
            "What's Apple's latest revenue and profit?",
            "Show me NVIDIA's 3-year financial trends",
            "How is Tesla's stock performing today?",
            "Compare Microsoft's earnings with its stock price",
            "Analyze Amazon's cash flow and market cap",
        ],
        title="💬 AI Assistant",
        description="Ask me about company financials, stock prices, or market trends. I'll automatically fetch the data you need!"
    )

# 启动应用
if __name__ == "__main__":
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        show_error=True,
        ssr_mode=False
    )