refactor: 重构目录结构 - 简化层级

backend/app/subgraphs/news_analysis/__init__.py

@@ -0,0 +1,46 @@
+"""
+资讯子图 - 完善版
+News Analysis Subgraph Module - Complete
+"""
+
+from .state import (
+    NewsAnalysisState,
+    NewsAction,
+    NewsItem,
+    NewsSource
+)
+from .graph import build_news_analysis_subgraph
+from .nodes import (
+    parse_intent,
+    query_news,
+    analyze_url,
+    extract_keywords,
+    generate_report,
+    format_result,
+    should_continue
+)
+from .api_client import news_api, NewsAPIClient
+
+__all__ = [
+    # State
+    "NewsAnalysisState",
+    "NewsAction",
+    "NewsItem",
+    "NewsSource",
+    
+    # Graph
+    "build_news_analysis_subgraph",
+    
+    # Nodes
+    "parse_intent",
+    "query_news",
+    "analyze_url",
+    "extract_keywords",
+    "generate_report",
+    "format_result",
+    "should_continue",
+    
+    # API
+    "news_api",
+    "NewsAPIClient"
+]

backend/app/subgraphs/news_analysis/api_client.py

@@ -0,0 +1,196 @@
+"""
+资讯子图API调用工具
+News Analysis API Client
+支持 async 和真实数据库缓存
+"""
+
+from typing import Dict, Any, Optional, List
+import random
+from datetime import datetime
+from dataclasses import dataclass
+
+
+@dataclass
+class NewsAPIClient:
+    """
+    资讯API客户端 - 可扩展支持多种API和数据库缓存
+    """
+    
+    # 可以配置多个API（如 NewsAPI, 今日头条, 百度新闻等）
+    newsapi_key: Optional[str] = None
+    
+    # 数据库 Repository（可选，用于缓存新闻）
+    news_repository: Optional[Any] = None
+    
+    async def query_news_db(self, user_id: str, keyword: str) -> Optional[List[Dict[str, Any]]]:
+        """从数据库缓存查询新闻"""
+        if not self.news_repository:
+            return None
+        try:
+            entities = await self.news_repository.search_by_keywords(user_id, keyword)
+            if entities:
+                return [
+                    {
+                        "title": e.title,
+                        "source": e.source,
+                        "summary": e.content,
+                        "keywords": e.keywords.split(",") if e.keywords else [],
+                        "author": "",
+                        "published_at": e.created_at
+                    }
+                    for e in entities
+                ]
+        except Exception as e:
+            print(f"从数据库查询新闻失败：{e}")
+        return None
+    
+    async def cache_news_db(self, user_id: str, news: Dict[str, Any]):
+        """把新闻缓存到数据库"""
+        if not self.news_repository:
+            return
+        try:
+            from ...db.models import NewsEntity
+            entity = NewsEntity(
+                user_id=user_id,
+                title=news.get("title", ""),
+                content=news.get("summary", ""),
+                url=news.get("url", ""),
+                source=news.get("source", ""),
+                keywords=",".join(news.get("keywords", []))
+            )
+            await self.news_repository.insert(entity)
+        except Exception as e:
+            print(f"缓存新闻到数据库失败：{e}")
+    
+    def query_news_mock(self, query: str) -> List[Dict[str, Any]]:
+        """
+        模拟查询资讯 - 目前用于演示
+        """
+        # 模拟资讯数据库
+        mock_news = [
+            {
+                "title": "OpenAI发布GPT-5：智能再升级",
+                "source": "Tech News",
+                "summary": "最新消息，OpenAI刚刚发布了GPT-5模型，智能水平再次取得重大突破...",
+                "keywords": ["AI", "GPT-5", "OpenAI"],
+                "author": "AI Team",
+                "published_at": datetime.now().isoformat()
+            },
+            {
+                "title": "大模型在医疗领域的应用",
+                "source": "Health Tech",
+                "summary": "大模型AI技术正在医疗领域展现巨大潜力，从辅助诊断到药物研发...",
+                "keywords": ["医疗", "大模型", "应用"],
+                "author": "Medical Team",
+                "published_at": datetime.now().isoformat()
+            },
+            {
+                "title": "2026年AI行业发展趋势报告",
+                "source": "Business Daily",
+                "summary": "最新行业报告显示，AI行业将继续保持高速增长，企业数字化转型加速...",
+                "keywords": ["趋势", "AI", "商业"],
+                "author": "Business Team",
+                "published_at": datetime.now().isoformat()
+            }
+        ]
+        
+        # 根据查询词简单过滤
+        results = []
+        query_lower = query.lower()
+        
+        for news in mock_news:
+            if (query_lower in news["title"].lower() or 
+                query_lower in news["summary"].lower() or 
+                any(keyword.lower() in query_lower for keyword in news["keywords"])):
+                results.append(news)
+        
+        # 如果没有匹配到，返回前两条
+        if not results:
+            results = mock_news[:2]
+        
+        return results
+    
+    def analyze_url_mock(self, url: str) -> Dict[str, Any]:
+        """
+        模拟URL分析 - 目前用于演示
+        """
+        return {
+            "title": f"分析结果：{url}",
+            "source": "URL Analyzer",
+            "summary": "已完成对该URL的内容分析，包含文章摘要和情感倾向判断...",
+            "keywords": ["News", "Analysis", url.split("/")[-1] if url else "unknown"]
+        }
+    
+    def extract_keywords_mock(self, text: str) -> List[str]:
+        """
+        模拟关键词提取 - 目前用于演示
+        """
+        # 简单的关键词提取模拟
+        common_keywords = ["AI", "大模型", "应用场景", "行业趋势", "创新", "技术"]
+        result = []
+        
+        for keyword in common_keywords:
+            if keyword.lower() in text.lower():
+                result.append(keyword)
+        
+        # 如果没找到，返回默认关键词
+        if not result:
+            result = ["AI", "大模型", "应用场景", "行业趋势"]
+        
+        return result
+    
+    def generate_report_mock(self, query: str) -> str:
+        """
+        模拟报告生成 - 目前用于演示
+        """
+        report = f"""═══════════════════════════════════════════
+📊 资讯分析报告
+═══════════════════════════════════════════
+
+主题：{query}
+
+📋 摘要：
+这是一份关于 {query} 的资讯分析综合报告，包含最新行业动态和趋势分析。
+
+🔍 主要发现：
+1. AI技术持续快速发展
+2. 大模型应用场景不断拓展
+3. 行业数字化转型加速
+
+🏷️ 关键词：
+- AI
+- 大模型
+- 数字化转型
+- 创新
+
+═══════════════════════════════════════════
+💡 建议：继续关注行业动态，把握发展机遇！
+"""
+        return report
+    
+    # ========== 统一入口（优先查缓存） ==========
+    async def query_news(self, user_id: str = "default", query: str = "", use_cache: bool = True) -> List[Dict[str, Any]]:
+        """查询新闻（统一入口，优先查数据库缓存）"""
+        # 1. 先查数据库缓存
+        if use_cache:
+            cached = await self.query_news_db(user_id, query)
+            if cached:
+                return cached
+        
+        # 2. 查第三方 API（暂未实现）
+        # api_result = await self.query_news_api(query)
+        # if api_result:
+        #     for news in api_result:
+        #         await self.cache_news_db(user_id, news)
+        #     return api_result
+        
+        # 3. 用模拟数据（兜底）
+        mock_result = self.query_news_mock(query)
+        if use_cache:
+            for news in mock_result:
+                await self.cache_news_db(user_id, news)
+        return mock_result
+
+
+# 单例实例（模拟模式，保持向后兼容）
+news_api = NewsAPIClient()

backend/app/subgraphs/news_analysis/graph.py

@@ -0,0 +1,63 @@
+"""
+资讯子图构建器
+News Analysis Subgraph Builder
+"""
+
+from app.main_graph.graph import StateGraph, START, END
+
+from .state import NewsAnalysisState
+from .nodes import (
+    parse_intent,
+    query_news,
+    analyze_url,
+    extract_keywords,
+    generate_report,
+    format_result,
+    should_continue
+)
+
+
+def build_news_analysis_subgraph() -> StateGraph:
+    """
+    构建资讯子图
+    
+    Returns:
+        配置好的 StateGraph
+    """
+    # 创建图
+    graph = StateGraph(NewsAnalysisState)
+    
+    # 添加节点
+    graph.add_node("parse_intent", parse_intent)
+    graph.add_node("query_news", query_news)
+    graph.add_node("analyze_url", analyze_url)
+    graph.add_node("extract_keywords", extract_keywords)
+    graph.add_node("generate_report", generate_report)
+    graph.add_node("format_result", format_result)
+    
+    # 添加边
+    # 从START开始
+    graph.add_edge(START, "parse_intent")
+    
+    # 从parse_intent根据条件路由
+    graph.add_conditional_edges(
+        "parse_intent",
+        should_continue,
+        {
+            "query_news": "query_news",
+            "analyze_url": "analyze_url",
+            "extract_keywords": "extract_keywords",
+            "generate_report": "generate_report",
+        }
+    )
+    
+    # 从各个操作节点到format_result
+    graph.add_edge("query_news", "format_result")
+    graph.add_edge("analyze_url", "format_result")
+    graph.add_edge("extract_keywords", "format_result")
+    graph.add_edge("generate_report", "format_result")
+    
+    # 最终到END
+    graph.add_edge("format_result", END)
+    
+    return graph

backend/app/subgraphs/news_analysis/nodes.py

@@ -0,0 +1,185 @@
+"""
+资讯子图节点 - 使用公共工具版本
+News Analysis Subgraph Nodes - Using Common Tools
+"""
+
+from typing import Dict, Any
+from datetime import datetime
+
+# 公共工具
+from ..common import MarkdownFormatter
+
+from .state import (
+    NewsAnalysisState,
+    NewsAction,
+    NewsItem,
+    NewsSource
+)
+from .api_client import news_api
+
+
+def parse_intent(state: NewsAnalysisState) -> NewsAnalysisState:
+    """
+    解析用户意图节点
+    确定用户想做什么操作
+    """
+    query_lower = state.user_query.lower()
+    
+    if any(keyword in query_lower for keyword in ["资讯", "新闻", "news", "report"]):
+        state.action = NewsAction.QUERY_NEWS
+    elif any(keyword in query_lower for keyword in ["分析", "analyze", "url", "链接"]):
+        state.action = NewsAction.ANALYZE_URL
+    elif any(keyword in query_lower for keyword in ["关键词", "keyword", "提取"]):
+        state.action = NewsAction.EXTRACT_KEYWORDS
+    elif any(keyword in query_lower for keyword in ["报告", "生成", "generate"]):
+        state.action = NewsAction.GENERATE_REPORT
+    else:
+        state.action = NewsAction.QUERY_NEWS
+    
+    return state
+
+
+def query_news(state: NewsAnalysisState) -> NewsAnalysisState:
+    """
+    查询资讯节点
+    """
+    state.current_phase = "executing"
+    
+    # 使用 API 客户端
+    news_items = news_api.query_news(state.user_query)
+    state.news_items = news_items
+    
+    return state
+
+
+def analyze_url(state: NewsAnalysisState) -> NewsAnalysisState:
+    """
+    分析 URL 节点
+    """
+    state.current_phase = "executing"
+    
+    # 从用户输入中提取 URL（简单处理）
+    query = state.user_query
+    url = query
+    for keyword in ["分析", "analyze", "url", "链接"]:
+        url = url.replace(keyword, "").strip()
+    
+    if not url:
+        url = "https://example.com/news/article"
+    
+    state.custom_urls = [url]
+    
+    # 使用 API 客户端
+    analysis = news_api.analyze_url(url)
+    state.analysis = analysis
+    
+    return state
+
+
+def extract_keywords(state: NewsAnalysisState) -> NewsAnalysisState:
+    """
+    提取关键词节点
+    """
+    state.current_phase = "executing"
+    
+    # 使用 API 客户端
+    keywords = news_api.extract_keywords(state.user_query)
+    state.extracted_keywords = keywords
+    
+    return state
+
+
+def generate_report(state: NewsAnalysisState) -> NewsAnalysisState:
+    """
+    生成报告节点
+    """
+    state.current_phase = "executing"
+    
+    # 使用 API 客户端
+    report = news_api.generate_report(state.user_query)
+    state.report_content = report
+    
+    return state
+
+
+def format_result(state: NewsAnalysisState) -> NewsAnalysisState:
+    """
+    格式化结果节点（使用公共工具）
+    """
+    state.current_phase = "formatting"
+    
+    md = MarkdownFormatter()
+    output_lines = []
+    
+    output_lines.append("┌───────────────────────────────────┐")
+    output_lines.append("│          📰 资讯助手               │")
+    output_lines.append("└───────────────────────────────────┘")
+    output_lines.append("")
+    
+    if state.action == NewsAction.QUERY_NEWS and state.news_items:
+        output_lines.append(md.heading("📰 最新资讯", 2))
+        output_lines.append("")
+        
+        for item in state.news_items:
+            output_lines.append(md.heading(item.title, 3))
+            output_lines.append(f"> 来源: {item.source.value}")
+            output_lines.append(f"> 时间: {item.published_at.strftime('%Y-%m-%d %H:%M')}")
+            if item.summary:
+                output_lines.append("")
+                output_lines.append(item.summary)
+            if item.url:
+                output_lines.append(f"🔗 链接: {md.link(item.title, item.url)}")
+            output_lines.append("")
+    
+    elif state.action == NewsAction.EXTRACT_KEYWORDS and state.extracted_keywords:
+        output_lines.append(md.heading("🏷️ 提取的关键词", 2))
+        output_lines.append("")
+        keywords_data = [
+            {"关键词": k, "权重": f"{w:.2f}"}
+            for k, w in state.extracted_keywords.items()
+        ]
+        output_lines.append(md.table(keywords_data))
+    
+    elif state.action == NewsAction.GENERATE_REPORT and state.report_content:
+        output_lines.append(md.heading("📊 分析报告", 2))
+        output_lines.append("")
+        output_lines.append(state.report_content)
+    
+    elif state.action == NewsAction.ANALYZE_URL and state.analysis:
+        output_lines.append(md.heading("🔍 URL 分析", 2))
+        output_lines.append("")
+        output_lines.append(f"> URL: {state.custom_urls[0]}")
+        output_lines.append("")
+        output_lines.append(state.analysis)
+    
+    else:
+        output_lines.append(md.heading("✨ 操作完成", 2))
+        output_lines.append("您的请求已处理。")
+    
+    # 页脚提示
+    output_lines.append("")
+    output_lines.append("---")
+    output_lines.append("💡 提示：您可以继续查询资讯、提取关键词或者生成报告！")
+    
+    state.final_result = "\n".join(output_lines)
+    state.success = True
+    state.current_phase = "completed"
+
+    return state
+
+
+def should_continue(state: NewsAnalysisState) -> str:
+    """
+    条件路由函数：根据 action 决定下一个节点
+    """
+    action = state.action
+    if action == NewsAction.QUERY:
+        return "query_news"
+    elif action == NewsAction.ANALYZE_URL:
+        return "analyze_url"
+    elif action == NewsAction.EXTRACT_KEYWORDS:
+        return "extract_keywords"
+    elif action == NewsAction.GENERATE_REPORT:
+        return "generate_report"
+    else:
+        return "format_result"

backend/app/subgraphs/news_analysis/state.py

@@ -0,0 +1,89 @@
+"""
+资讯子图状态定义
+News Analysis Subgraph State Definition
+"""
+
+from enum import Enum, auto
+from typing import Optional, Dict, List, Any
+from dataclasses import dataclass, field
+
+
+class NewsAction(Enum):
+    """资讯操作类型"""
+    NONE = auto()
+    QUERY_NEWS = auto()  # 查询资讯
+    ANALYZE_URL = auto()  # 分析资讯
+    GENERATE_REPORT = auto()  # 生成报告
+    FETCH_FROM_SOURCES = auto()  # 从指定源获取
+    EXTRACT_KEYWORDS = auto()  # 提取关键词
+
+
+@dataclass
+class NewsItem:
+    """资讯条目"""
+    title: str = ""
+    url: str = ""
+    source: str = ""
+    content: str = ""
+    author: str = ""
+    published_at: Optional[str] = None
+    summary: str = ""
+    keywords: List[str] = field(default_factory=list)
+    sentiment: float = 0.0  # 情感分析得分
+    metadata: Dict[str, Any] = field(default_factory=dict)
+
+
+@dataclass
+class NewsSource:
+    """资讯源"""
+    name: str = ""
+    url: str = ""
+    type: str = ""  # rss, website, api
+    enabled: bool = True
+    last_fetched_at: Optional[str] = None
+    metadata: Dict[str, Any] = field(default_factory=dict)
+
+
+@dataclass
+class NewsAnalysisState:
+    """资讯子图状态"""
+    # ========== 输入 ==========
+    user_query: str = ""  # 用户查询
+    user_id: str = ""  # 用户ID
+    
+    # 操作控制
+    action: NewsAction = NewsAction.NONE
+    action_params: Dict[str, Any] = field(default_factory=dict)
+    
+    # 源配置
+    use_follow_list: bool = False
+    custom_urls: List[str] = field(default_factory=list)
+    
+    # ========== 执行过程 ==========
+    current_phase: str = "init"  # init, fetching, analyzing, done
+    current_source_index: int = 0
+    primary_fetched: bool = False
+    
+    # 源列表
+    sources: List[NewsSource] = field(default_factory=list)
+    
+    # 资讯条目
+    news_items: List[NewsItem] = field(default_factory=list)
+    
+    # 关键词
+    extracted_keywords: List[str] = field(default_factory=list)
+    
+    # 报告
+    report_content: str = ""
+    
+    # ========== 结果 ==========
+    success: bool = False
+    error_message: str = ""
+    final_result: str = ""
+    result_data: Dict[str, Any] = field(default_factory=dict)
+    
+    # ========== 元数据 ==========
+    start_time: Optional[str] = None
+    end_time: Optional[str] = None
+    duration: float = 0.0
+    debug_info: Dict[str, Any] = field(default_factory=dict)