# Digital Realty — facility pages (Next.js, __NEXT_DATA__ Drupal node) + press releases
id: digitalrealty
name: Digital Realty
domain: www.digitalrealty.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © Digital Realty Trust, Inc."
attribution: "Digital Realty"
homepage: https://www.digitalrealty.com/data-centers
notes: >
  Open site, plain HTTP works with the bot UA. Only the English sitemap is crawled (the index lists 6 other locales
  that mirror the same facilities). Facility URLs: /data-centers/<region>/<metro>/<code>; metro pages are indexes.
  No RSS feed exists; press releases are classified from the sitemap (/about/newsroom/press-releases/<id>/<slug>).
  robots.txt disallows /taxonomy/ and /node/ only.
coverage: { global: true, operators: ["Digital Realty"] }
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: ["https://www.digitalrealty.com/en-sitemap.xml"]
  include:
    - "^https://www\\.digitalrealty\\.com/data-centers/"
    - "^https://www\\.digitalrealty\\.com/about/newsroom/press-releases/"
  exclude: ["\\?", "/taxonomy/", "/node/"]
  classify:
    - { pattern: "/data-centers/[a-z-]+/[a-z-]+/[a-z0-9-]+$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-centers/[a-z-]+/[a-z-]+$", group: index, pageType: facility_index, priority: 20 }
    - { pattern: "/data-centers/[a-z-]+$", group: index, pageType: facility_index, priority: 10 }
    - { pattern: "/about/newsroom/press-releases/\\d+/", group: newsroom, pageType: press_release, priority: 70 }
    - { pattern: "/about/newsroom/press-releases/?$", group: index, pageType: news_index, priority: 10 }
  maxUrlsPerRun: 2000
schedule:
  facility_pages: weekly
  newsroom: daily
  index: monthly
  sitemap: weekly
extractors:
  facility:
    # selected by URL only (classifyPage() may mislabel deep facility URLs; the URL is authoritative)
    match: ["/data-centers/[a-z-]+/[a-z-]+/[a-z0-9-]+$"]
    kind: facility
    parser: digitalrealty_facility_v1
defaults:
  operatorName: Digital Realty
  operatorKind: colocation
  facilityType: colocation
parserVersion: v1