# CyrusOne — facility pages (HubSpot CMS) + press-release RSS
id: cyrusone
name: CyrusOne
domain: www.cyrusone.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © CyrusOne LLC"
attribution: "CyrusOne"
homepage: https://www.cyrusone.com/data-centers
notes: >
  Open site (HubSpot). Facility URLs: /data-centers/(north-america|emea|apac)/<slug>; pages publish city/country,
  facility code(s), "N MW Total IT Capacity", technical space and sustainability items — no street address and no
  coordinates (geo left null). Press releases: /resources/press-releases/rss.xml.
  robots.txt disallows HubSpot preview/preference paths only.
coverage: { global: true, operators: ["CyrusOne"] }
fetch:
  level: 1
  maxLevel: 4
  rpm: 20
  concurrency: 2
  respectRobots: true
  maxCreditsPerRun: 40
  maxCreditsPerDay: 120   # premium credits per UTC day for this connector (all runs, all workers)
discovery:
  sitemap: true
  rss: ["/resources/press-releases/rss.xml"]
  include: ["/data-centers/", "/resources/press-releases/"]
  exclude: ["\\?", "#", "back-up-generator-schedule", "/_hcms/", "/hs/"]
  classify:
    - { pattern: "/data-centers/(north-america|emea|apac)/[a-z0-9-]+$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-centers/?$", group: index, pageType: facility_index, priority: 10 }
    - { pattern: "/resources/press-releases/[a-z0-9._%-]+$", group: newsroom, pageType: press_release, priority: 70 }
    - { pattern: "/resources/press-releases/?$", group: index, pageType: news_index, priority: 10 }
  maxUrlsPerRun: 1000
schedule:
  facility_pages: weekly
  newsroom: daily
  index: monthly
  sitemap: weekly
  rss: daily
extractors:
  facility:
    # selected by URL only (classifyPage() may mislabel deep facility URLs; the URL is authoritative)
    match: ["/data-centers/(north-america|emea|apac)/[a-z0-9-]+$"]
    kind: facility
    parser: cyrusone_facility_v1
defaults:
  operatorName: CyrusOne
  operatorKind: colocation
  facilityType: wholesale
parserVersion: v1