// 这些查询会自动触发 Tavily 搜索
"分析当前市场状况"
"最新的苹果股价分析"
"今天的标普500指数表现"
"最近的美联储利率决定"

// 在 context.marketContext 中会包含：
{
  currentDate: "2024-12-19",
  marketConditions: "volatile",
  interestRateEnvironment: "rising",
  economicIndicators: { ... },
  realTimeData: {
    searchQuery: "分析当前市场状况 market analysis trends 2024",
    directAnswer: "当前市场呈现波动性上升趋势...",
    sources: [
      {
        title: "Market Analysis Today",
        url: "https://bloomberg.com/...",
        snippet: "Markets showed increased volatility..."
      }
    ],
    timestamp: "2024-12-19T10:30:00Z"
  },
  dataSource: "tavily_search",
  lastUpdated: "2024-12-19T10:30:00Z"
}

// 用户查询
const userQuery = "分析当前科技股市场趋势和苹果公司最新财报影响";
 
// 工作流执行过程：
// 1. Intent Analysis: 识别为 market_analysis
// 2. Context Analysis: 检测到 "当前"、"最新" 关键词，触发 Tavily 搜索
//    搜索查询: "科技股市场趋势苹果公司财报 market analysis trends 2024"
// 3. Routing Decision: 选择复杂分析流程
// 4. Execution Plan: 包含 market_agent 和 research_agent
// 5. Agent Execution: 
//    - market_agent 获取实时市场数据
//    - research_agent 搜索苹果财报分析
// 6. Report Generation: 整合所有数据生成报告

// 用户查询
const userQuery = "评估我的投资组合在当前市场环境下的风险";
 
// 工作流执行过程：
// 1. Intent Analysis: 识别为 portfolio_analysis + risk_control
// 2. Context Analysis: 检测到 "当前市场环境"，触发搜索
//    搜索查询: "投资组合风险当前市场环境 portfolio market conditions investment 2024"
// 3. Agent Execution:
//    - portfolio_agent: 分析组合配置
//    - risk_agent: 计算风险指标
//    - market_agent: 获取当前市场风险因子

# .env.local
TAVILY_API_KEY="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"
COMPOSIO_API_KEY="your-composio-api-key"

// 在你的工具配置中启用 Tavily
import { ToolConfig } from '@/lib/tools/composio-config';
 
const toolConfig: ToolConfig = {
  enabled: true,
  apps: ['tavily'], // 添加 tavily 到应用列表
  maxToolRoundtrips: 3,
  useSmartToolSelection: true
};

// 使用现有的工作流，无需修改
import { financialOrchestrationGraph } from '@/lib/agents/financial-orchestration-workflow';
 
// 执行工作流
const result = await financialOrchestrationGraph.invoke({
  messages: [],
  userQuery: "分析最新市场趋势",
  userId: "user-123"
});
 
// 检查是否使用了网络搜索
if (result.context?.marketContext?.dataSource === 'tavily_search') {
  console.log('使用了实时市场数据');
  console.log('搜索结果:', result.context.marketContext.realTimeData);
}

// 实时关键词
const realTimeKeywords = [
  'latest', 'current', 'today', 'recent', 'now', 'update',
  'price', 'market', 'news', 'earnings', 'fed', 'rate',
  'inflation', 'gdp', 'unemployment', 'stock', 'index'
];
 
// 需要实时数据的意图类别
const realTimeCategories = [
  'market_analysis',
  'portfolio_analysis', 
  'risk_control',
  'pricing_automation'
];

// Market Agent: 总是尝试获取实时市场数据
// Research Agent: 基于执行步骤描述进行搜索

const financialDomains = [
  "bloomberg.com",      // 彭博财经
  "reuters.com",        // 路透社
  "marketwatch.com",    // 市场观察
  "yahoo.com",          // 雅虎财经
  "cnbc.com",           // CNBC
  "wsj.com",            // 华尔街日报
  "ft.com",             // 金融时报
  "sec.gov",            // 美国证监会
  "federalreserve.gov"  // 美联储
];

// 在节点执行过程中查看日志
console.log('📊 Context Analysis Node - Starting');
console.log('🔍 Fetching real-time market data using Tavily');
console.log('✅ Tavily search completed: Found 5 results');

# 启动 LangGraph Studio
npx @langchain/langgraph-cli dev
 
# 访问 Studio UI
# https://smith.langchain.com/studio?baseUrl=http://localhost:2024

// 检查工作流状态中的搜索结果
const state = await graph.getState();
console.log('Web search performed:', state.webSearchPerformed);
console.log('Search results:', state.webSearchResults);

// 实现搜索结果缓存（5分钟）
const searchCache = new Map();
const CACHE_DURATION = 5 * 60 * 1000; // 5 minutes
 
function getCachedResult(query: string) {
  const cached = searchCache.get(query);
  if (cached && Date.now() - cached.timestamp < CACHE_DURATION) {
    return cached.result;
  }
  return null;
}

// 在 Agent Execution 中并行执行多个搜索
const [marketData, researchData] = await Promise.all([
  fetchMarketData(),
  fetchResearchData(description)
]);

try {
  const marketData = await fetchRealTimeMarketData(userQuery, intent);
  // 使用实时数据
} catch (error) {
  console.log(`Failed to fetch real-time data: ${error}`);
  // 使用基础数据继续执行
}

// 设置搜索超时
const searchWithTimeout = Promise.race([
  searchTool.callback(params),
  new Promise((_, reject) => 
    setTimeout(() => reject(new Error('Search timeout')), 10000)
  )
]);

// Automatic topic selection based on intent
const topic = this.getSearchTopic(intent);
 
const searchResult = await searchTool.callback({
  query: searchQuery,
  topic: topic, // Dynamically selected
  searchDepth: "basic",
  includeImages: false,
  includeAnswer: true,
  maxResults: 5,
  includeDomains: [
    "bloomberg.com",
    "reuters.com", 
    "marketwatch.com"
  ]
});

const searchResult = await searchTool.callback({
  query: "stock market today S&P 500 Dow Jones financial markets",
  topic: "finance",
  searchDepth: "basic",
  includeImages: false,
  includeAnswer: true,
  maxResults: 5,
  includeDomains: [
    "bloomberg.com",
    "reuters.com",
    "marketwatch.com",
    "yahoo.com",
    "cnbc.com"
  ]
});

// Technology research
const techResult = await searchTool.callback({
  query: "artificial intelligence fintech applications 2024",
  topic: "technology",
  searchDepth: "advanced",
  includeImages: false,
  includeAnswer: true,
  maxResults: 8
});
 
// Financial research  
const financeResult = await searchTool.callback({
  query: searchQuery,
  topic: "finance", 
  searchDepth: "advanced",
  includeImages: false,
  includeAnswer: true,
  maxResults: 8,
  includeDomains: [
    "bloomberg.com",
    "reuters.com",
    "ft.com",
    "wsj.com",
    "sec.gov"
  ]
});

import { performIntelligentSearch } from '../lib/tools/tavily-example';
 
const result = await performIntelligentSearch(
  "blockchain adoption in banking",
  "market_analysis", // Intent category
  {
    searchDepth: "advanced",
    maxResults: 10,
    includeDomains: ["coindesk.com", "bloomberg.com"]
  }
);
 
// Result includes selected topic and reasoning
console.log(result.selectedTopic); // "technology"
console.log(result.searchMetadata.topicReasoning);

import { TOPIC_USAGE_PATTERNS } from '../lib/tools/tavily-example';
 
// Use predefined patterns
const pattern = TOPIC_USAGE_PATTERNS.techResearch;
 
const result = await searchTool.callback({
  query: "AI in financial services",
  topic: pattern.topic, // "technology"
  searchDepth: pattern.searchDepth, // "advanced"
  includeDomains: pattern.domains,
  maxResults: 8
});

try {
  const result = await searchTool.callback({
    query: "market analysis",
    topic: "finance",
    // ... other parameters
  });
  
  if (result.success) {
    // Use search results
  } else {
    // Fallback to base analysis
  }
} catch (error) {
  // Log error and continue with fallback
  console.log(`Search failed: ${error.message}`);
}

npm run test:tavily