id: africadatacentres
name: Africa Data Centres
domain: www.africadatacentres.com
kind: operator
mode: hybrid
priority: 2
license: "Publicly accessible operator pages; factual data only; © Africa Data Centres"
attribution: "Africa Data Centres"
homepage: https://www.africadatacentres.com/
notes: "One page per site at the site root (/capetown/, /lagos/, /midrand/, /nairobi/, /samrand-2/) with code (CPT1), current MW, planned MW upon completion and IT space. Seeds are explicit; the facility extractor is restricted to these paths so news posts never yield facilities."
coverage: { operators: ["Africa Data Centres"], countries: ["ZA", "NG", "KE"] }
fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }
discovery:
  sitemap: ["https://www.africadatacentres.com/post-sitemap.xml"]
  seeds: [{ url: "/capetown/", group: facility_pages, pageType: facility_page }, { url: "/lagos/", group: facility_pages, pageType: facility_page }, { url: "/midrand/", group: facility_pages, pageType: facility_page }, { url: "/nairobi/", group: facility_pages, pageType: facility_page }, { url: "/samrand-2/", group: facility_pages, pageType: facility_page }]
  include: ["/(capetown|lagos|midrand|nairobi|samrand-2)/$", "africadatacentres\\.com/[a-z0-9-]+/$"]
  exclude: ["\\?", "#"]
  classify:
    - { pattern: "/(capetown|lagos|midrand|nairobi|samrand-2)/$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "africadatacentres\\.com/[a-z0-9-]+/$", group: newsroom, pageType: press_release, priority: 40 }
schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly }
extractors:
  facility: { match: ["/(capetown|lagos|midrand|nairobi|samrand-2)/$"], kind: facility, parser: adc_facility_v1 }
defaults: { operatorName: Africa Data Centres, facilityType: colocation }