id: cdc
name: CDC Data Centres
domain: cdc.com
kind: operator
mode: hybrid
priority: 2
license: "Publicly accessible operator pages; factual data only; © CDC Data Centres"
attribution: "CDC Data Centres"
homepage: https://cdc.com/locations/
notes: "City pages /locations/<city>/ list campuses with 'Current capacity (operating) NN MW' or 'Total capacity NN MW' (under construction) → one campus-level record each. cdcdatacentres.com.au redirects here."
coverage: { operators: ["CDC Data Centres"], countries: ["AU", "NZ"] }
fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }
discovery:
  sitemap: true
  seeds: [{ url: "/locations/auckland/", group: facility_pages, pageType: facility_page }, { url: "/locations/canberra/", group: facility_pages, pageType: facility_page }, { url: "/locations/melbourne/", group: facility_pages, pageType: facility_page }, { url: "/locations/sydney/", group: facility_pages, pageType: facility_page }, { url: "/locations/perth/", group: facility_pages, pageType: facility_page }, { url: "/locations/maddington/", group: facility_pages, pageType: facility_page }]
  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: cdc_campus_v1 }
defaults: { operatorName: CDC Data Centres, facilityType: hyperscale }