id: kaodata
name: Kao Data
domain: www.kaodata.com
kind: operator
mode: hybrid
priority: 2
license: "Publicly accessible operator pages; factual data only; © Kao Data"
attribution: "Kao Data"
homepage: https://www.kaodata.com/locations/
notes: "Campus pages /locations/<campus>/ publish the campus address, full-build IT load, acreage and a list of buildings 'KLON-01 (8.8MW Operational)' → campus record + one record per building with its own MW and lifecycle status. Cookie-consent scripts trip the blocked heuristic → maxLevel 2."
coverage: { operators: ["Kao Data"], countries: ["GB"] }
fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }
discovery:
  sitemap: ["https://kaodata.com/wp-sitemap-posts-page-1.xml", "https://kaodata.com/wp-sitemap-posts-news-1.xml"]
  include: ["/locations/[a-z-]+/$", "/news/."]
  exclude: ["\\?", "#"]
  classify:
    - { pattern: "/locations/[a-z-]+/$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/news/.", group: newsroom, pageType: press_release, priority: 50 }
schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly }
extractors:
  facility: { match: ["/locations/[a-z-]+/$"], kind: facility, parser: kaodata_facility_v1 }
defaults: { operatorName: Kao Data, facilityType: wholesale, countryIso2: GB }