id: nextdc
name: NEXTDC
domain: www.nextdc.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © NEXTDC Limited"
attribution: "NEXTDC"
homepage: https://www.nextdc.com/data-centres
notes: "Facility pages /data-centres/<metro>-data-centres/<code>-<city> publish IT capacity, technical space, rack capacity, Tier and certifications. robots.txt allows / (HubSpot preview paths excluded)."
coverage: { operators: ["NEXTDC"], countries: ["AU", "NZ", "MY", "JP"] }
fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }
discovery:
  sitemap: ["https://www.nextdc.com/sitemap.xml"]
  include: ["/data-centres/[a-z-]+-data-centres/[a-z]{1,2}\\d{1,2}-[a-z-]+$", "/news/", "/media-releases?/", "/newsroom/"]
  exclude: ["\\?", "#"]
  classify:
    - { pattern: "/data-centres/[a-z-]+-data-centres/[a-z]{1,2}\\d{1,2}-[a-z-]+$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/(news|media-releases?|newsroom)/.", group: newsroom, pageType: press_release, priority: 50 }
schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly }
extractors:
  facility: { match: ["/data-centres/[a-z-]+-data-centres/[a-z]{1,2}\\d{1,2}-[a-z-]+$"], kind: facility, parser: nextdc_facility_v1 }
defaults: { operatorName: NEXTDC, facilityType: colocation, countryIso2: AU }