id: stream name: Stream Data Centers domain: www.streamdatacenters.com kind: operator mode: hybrid priority: 2 license: "Publicly accessible operator pages; factual data only; © Stream Data Centers" attribution: "Stream Data Centers" homepage: https://www.streamdatacenters.com/locations/ notes: "Site pages /locations// publish city/state, product type, leasing status, critical MW, square feet and acreage. Metro roll-up pages yield no record." coverage: { operators: ["Stream Data Centers"], countries: ["US"] } fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 } discovery: sitemap: ["https://www.streamdatacenters.com/datacenters-sitemap.xml", "https://www.streamdatacenters.com/post-sitemap.xml"] include: ["/locations/[a-z0-9-]+/$", "/news/", "/press-releases?/", "/articles/[a-z0-9-]+/$"] exclude: ["\\?", "#"] classify: - { pattern: "/locations/[a-z0-9-]+/$", group: facility_pages, pageType: facility_page, priority: 60 } - { pattern: "/(news|press-releases?|articles)/.", group: newsroom, pageType: press_release, priority: 50 } schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly } extractors: facility: { match: ["/locations/[a-z0-9-]+/$"], kind: facility, parser: stream_facility_v1 } defaults: { operatorName: Stream Data Centers, facilityType: wholesale }