# Runtime smoke test (disabled: extension is not .yaml). Rename to digitalrealty-smoke.yaml to exercise the
# crawl runtime end to end against live pages:  pnpm dci run digitalrealty-smoke --dry-run --limit 2
# Sitemap discovery + generic classification only; no extractor claims beyond what the pages state.
id: digitalrealty-smoke
name: Digital Realty (runtime smoke)
domain: www.digitalrealty.com
kind: operator
mode: sitemap
priority: 2
license: "Publicly accessible operator pages; facts only, attribution kept"
attribution: "Digital Realty"
homepage: https://www.digitalrealty.com
coverage: { global: true }
fetch:
  level: 1
  maxLevel: 2          # never spend premium credits in a smoke test
  rpm: 12
  concurrency: 2
  respectRobots: true
  maxCreditsPerRun: 0
discovery:
  sitemap: true
  include: ["/data-centers/", "/news/", "/press-releases/"]
  exclude: ["\\?", "/login", "/careers"]
  classify:
    - { pattern: "/data-centers/[a-z0-9-]+/[a-z0-9-]+/?$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-centers/[a-z0-9-]+/?$", group: facility_index, pageType: facility_index, priority: 40 }
    - { pattern: "/(news|press-releases)/", group: newsroom, pageType: press_release, priority: 70 }
  maxUrlsPerRun: 500
schedule:
  facility_pages: weekly
  facility_index: weekly
  newsroom: daily
  sitemap: weekly
extractors:
  facility:
    pageTypes: [facility_page]
    kind: facility
    key: "digitalrealty:{url}"
    fields:
      name: "h1"
      address: { jsonld: "Place", jsonPath: "address.streetAddress" }
      city: { jsonld: "Place", jsonPath: "address.addressLocality" }
      countryIso2: { jsonld: "Place", jsonPath: "address.addressCountry", transform: country }
      lat: { jsonld: "Place", jsonPath: "geo.latitude", transform: float }
      lng: { jsonld: "Place", jsonPath: "geo.longitude", transform: float }
defaults:
  operatorName: Digital Realty
  facilityType: colocation