spb/datacenterindex

1# Runtime smoke test (disabled: extension is not .yaml). Rename to digitalrealty-smoke.yaml to exercise the2# crawl runtime end to end against live pages:  pnpm dci run digitalrealty-smoke --dry-run --limit 23# Sitemap discovery + generic classification only; no extractor claims beyond what the pages state.4id: digitalrealty-smoke5name: Digital Realty (runtime smoke)6domain: www.digitalrealty.com7kind: operator8mode: sitemap9priority: 210license: "Publicly accessible operator pages; facts only, attribution kept"11attribution: "Digital Realty"12homepage: https://www.digitalrealty.com13coverage: { global: true }14fetch:15  level: 116  maxLevel: 2          # never spend premium credits in a smoke test17  rpm: 1218  concurrency: 219  respectRobots: true20  maxCreditsPerRun: 021discovery:22  sitemap: true23  include: ["/data-centers/", "/news/", "/press-releases/"]24  exclude: ["\\?", "/login", "/careers"]25  classify:26    - { pattern: "/data-centers/[a-z0-9-]+/[a-z0-9-]+/?$", group: facility_pages, pageType: facility_page, priority: 60 }27    - { pattern: "/data-centers/[a-z0-9-]+/?$", group: facility_index, pageType: facility_index, priority: 40 }28    - { pattern: "/(news|press-releases)/", group: newsroom, pageType: press_release, priority: 70 }29  maxUrlsPerRun: 50030schedule:31  facility_pages: weekly32  facility_index: weekly33  newsroom: daily34  sitemap: weekly35extractors:36  facility:37    pageTypes: [facility_page]38    kind: facility39    key: "digitalrealty:{url}"40    fields:41      name: "h1"42      address: { jsonld: "Place", jsonPath: "address.streetAddress" }43      city: { jsonld: "Place", jsonPath: "address.addressLocality" }44      countryIso2: { jsonld: "Place", jsonPath: "address.addressCountry", transform: country }45      lat: { jsonld: "Place", jsonPath: "geo.latitude", transform: float }46      lng: { jsonld: "Place", jsonPath: "geo.longitude", transform: float }47defaults:48  operatorName: Digital Realty49  facilityType: colocation50