# CoreSite (American Tower) — facility pages with schema.org DataCenter JSON-LD + news RSS
id: coresite
name: CoreSite
domain: www.coresite.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © CoreSite, an American Tower company"
attribution: "CoreSite"
homepage: https://www.coresite.com/data-centers/locations
notes: >
  Open HubSpot site, single sitemap.xml. Facility URLs: /data-center/<code>-<city>-<st> (29 sites) with a
  schema.org "DataCenter" JSON-LD carrying PostalAddress + GeoCoordinates; spec tabs give square footage,
  certifications, utility provider; cloud on-ramps and peering exchanges are named in the text.
  Metro pages /data-center-locations/<metro> are indexes. RSS: /news/rss.xml.
  robots.txt disallows HubSpot preview/preference paths only.
coverage: { countries: ["US"], operators: ["CoreSite"] }
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: ["/news/rss.xml"]
  include: ["/data-center/", "/data-center-locations/", "/news/"]
  exclude: ["\\?", "/news/?$", "/_hcms/", "/hs/"]
  classify:
    - { pattern: "/data-center/[a-z]{2}\\d{1,2}-[a-z0-9-]+$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-center-locations/[a-z0-9-]+$", group: index, pageType: facility_index, priority: 20 }
    - { pattern: "/news/[a-z0-9-]+$", group: newsroom, pageType: press_release, priority: 70 }
  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-center/[a-z]{2}\\d{1,2}-[a-z0-9-]+$"]
    kind: facility
    parser: coresite_facility_v1
defaults:
  operatorName: CoreSite
  operatorKind: colocation
  facilityType: colocation
parserVersion: v1