id: telehouse
name: Telehouse
domain: www.telehouse.net
kind: operator
mode: hybrid
priority: 2
license: "Publicly accessible operator pages; factual data only; © Telehouse"
attribution: "Telehouse"
homepage: https://www.telehouse.net/data-centre-services/
notes: "Facility pages /data-centre-services/<country>/<city>/<facility>/ publish specifications (power, cooling, carriers) but no MW/address; name, city and country come from the page/URL."
coverage: { operators: ["Telehouse"], countries: ["GB", "FR", "DE", "CA"] }
fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }
discovery:
  sitemap: ["https://www.telehouse.net/page-sitemap.xml", "https://www.telehouse.net/post-sitemap.xml"]
  include: ["/data-centre-services/[a-z]+/[a-z-]+/[a-z0-9-]+/$", "/news/", "/press-releases?/"]
  exclude: ["\\?", "#"]
  classify:
    - { pattern: "/data-centre-services/[a-z]+/[a-z-]+/[a-z0-9-]+/$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/(news|press-releases?)/.", group: newsroom, pageType: press_release, priority: 50 }
schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly }
extractors:
  facility: { match: ["/data-centre-services/[a-z]+/[a-z-]+/[a-z0-9-]+/$"], kind: facility, parser: telehouse_facility_v1 }
defaults: { operatorName: Telehouse, facilityType: colocation }