# ax_gfina 工程配置
# =====================================================================

# 产业/行业配置 ----------------------------------------------------------
industries:
  - id: semiconductor
    name: 半导体
    keywords: ["芯片", "晶圆", "EDA", "光刻", "封测", "存储器"]
  - id: new_energy
    name: 新能源
    keywords: ["锂电", "光伏", "风电", "储能", "氢能", "充电桩"]
  - id: ai
    name: 人工智能
    keywords: ["大模型", "AI芯片", "AIGC", "智能驾驶", "机器人"]
  - id: biopharma
    name: 生物医药
    keywords: ["创新药", "疫苗", "CXO", "基因", "医疗器械"]

# 数据源配置 ------------------------------------------------------------
sources:
  # RSS 订阅源
  rss:
    enabled: true
    feeds:
      - name: 36氪
        url: "https://36kr.com/feed"
      - name: 虎嗅
        url: "https://rsshub.app/huxiu/article"
      - name: 钛媒体
        url: "https://rsshub.app/tmtpost"
    # 抓取间隔(秒)
    interval: 3600

  # API 接口源
  api:
    enabled: true
    endpoints: []
    # - name: example_api
    #   url: "https://api.example.com/news"
    #   api_key: ""
    #   params:
    #     page_size: 50

  # 网页爬取源
  web:
    enabled: false
    sites: []

# 输出配置 --------------------------------------------------------------
output:
  # 数据存储目录
  data_dir: "./data"
  raw_dir: "./data/raw"
  processed_dir: "./data/processed"
  # HTML 报告输出目录
  report_dir: "./output"
  # 报告文件名 (支持 strftime 格式)
  report_name: "report_%Y%m%d.html"

# 分析配置 --------------------------------------------------------------
analysis:
  # 时间窗口(天): 只统计最近 N 天的动态
  time_window_days: 7
  # 热点关键词 top N
  top_keywords: 20
  # 趋势图分桶数
  trend_bins: 7

# 可视化配置 ------------------------------------------------------------
visualize:
  # 图表主题
  theme: "walden"
  # 每页展示条目数
  page_size: 20
