# Example connector definition (disabled: file extension not .yaml). Copy to <id>.yaml.
id: example-operator
name: Example Operator
domain: example.com
kind: operator            # operator | government | filing | utility | dataset | community | secondary | news | cloud_provider | registry
mode: hybrid
priority: 2               # 1 best … 5
license: "Publicly accessible operator pages; facts only, attribution kept"
attribution: "Example Operator"
coverage: { global: true }
fetch:
  level: 1                # start direct
  maxLevel: 4             # allow escalation to Scrapfly when blocked
  rpm: 20
  concurrency: 2
  respectRobots: true
  maxCreditsPerRun: 150
discovery:
  sitemap: true
  include: ["/data-centers/", "/newsroom/"]
  exclude: ["/login", "\\?page="]
  rss: ["/newsroom/rss.xml"]
  classify:
    - { pattern: "/data-centers/[a-z0-9-]+/[a-z0-9-]+$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/newsroom/", group: newsroom, pageType: press_release, priority: 70 }
schedule:
  facility_pages: weekly
  newsroom: daily
  sitemap: weekly
extractors:
  facility:
    pageTypes: [facility_page]
    kind: facility
    key: "example:{code}"
    fields:
      name: "h1"
      code: { selector: "h1", regex: "\\b([A-Z]{2}\\d{1,2})\\b" }
      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 }
      totalPowerMw: { selector: ".specs", regex: "(\\d+(?:\\.\\d+)?\\s*MW)", transform: mw }
  news:
    pageTypes: [press_release, project_announcement, expansion, acquisition]
    kind: news_event
    parser: article_v1
defaults:
  operatorName: Example Operator
  facilityType: colocation