spb/datacenterindex

1# Example connector definition (disabled: file extension not .yaml). Copy to <id>.yaml.2id: example-operator3name: Example Operator4domain: example.com5kind: operator            # operator | government | filing | utility | dataset | community | secondary | news | cloud_provider | registry6mode: hybrid7priority: 2               # 1 best … 58license: "Publicly accessible operator pages; facts only, attribution kept"9attribution: "Example Operator"10coverage: { global: true }11fetch:12  level: 1                # start direct13  maxLevel: 4             # allow escalation to Scrapfly when blocked14  rpm: 2015  concurrency: 216  respectRobots: true17  maxCreditsPerRun: 15018discovery:19  sitemap: true20  include: ["/data-centers/", "/newsroom/"]21  exclude: ["/login", "\\?page="]22  rss: ["/newsroom/rss.xml"]23  classify:24    - { pattern: "/data-centers/[a-z0-9-]+/[a-z0-9-]+$", group: facility_pages, pageType: facility_page, priority: 60 }25    - { pattern: "/newsroom/", group: newsroom, pageType: press_release, priority: 70 }26schedule:27  facility_pages: weekly28  newsroom: daily29  sitemap: weekly30extractors:31  facility:32    pageTypes: [facility_page]33    kind: facility34    key: "example:{code}"35    fields:36      name: "h1"37      code: { selector: "h1", regex: "\\b([A-Z]{2}\\d{1,2})\\b" }38      address: { jsonld: "Place", jsonPath: "address.streetAddress" }39      city: { jsonld: "Place", jsonPath: "address.addressLocality" }40      countryIso2: { jsonld: "Place", jsonPath: "address.addressCountry", transform: country }41      lat: { jsonld: "Place", jsonPath: "geo.latitude", transform: float }42      lng: { jsonld: "Place", jsonPath: "geo.longitude", transform: float }43      totalPowerMw: { selector: ".specs", regex: "(\\d+(?:\\.\\d+)?\\s*MW)", transform: mw }44  news:45    pageTypes: [press_release, project_announcement, expansion, acquisition]46    kind: news_event47    parser: article_v148defaults:49  operatorName: Example Operator50  facilityType: colocation51