news-cnbc.yaml

vectara:
  corpus_id: 154
  customer_id: 1526022105
  reindex: false

crawling:
  crawler_type: rss

rss_crawler:
  source: cnbc
  rss_pages: [
    "https://www.cnbc.com/id/100003114/device/rss/rss.html", "https://www.cnbc.com/id/105545213/device/rss/rss.html",
    "https://www.cnbc.com/id/10000664/device/rss/rss.html", "https://www.cnbc.com/id/10000744/device/rss/rss.html",
    "https://www.cnbc.com/id/10000108/device/rss/rss.html", "https://www.cnbc.com/id/10000115/device/rss/rss.html",
    "https://www.cnbc.com/id/10000113/device/rss/rss.html", "https://www.cnbc.com/id/10000114/device/rss/rss.html", 
    "https://www.cnbc.com/id/10000116/device/rss/rss.html", "https://www.cnbc.com/id/19836768/device/rss/rss.html", 
    "https://www.cnbc.com/id/10000110/device/rss/rss.html", "https://www.cnbc.com/id/10000739/device/rss/rss.html", 
    "https://www.cnbc.com/id/10000117/device/rss/rss.html"
  ]
  days_past: 30
  delay: 1
  extraction: playwright           # pdf or playwright