SPB Git forge
38commits 1branches 0releases
338.7 MBsize
maindefault branch
3 h agolast push
HTML 53.9% TypeScript 44.5% JavaScript 0.6% SQL 0.5%
1.4 KB · 24 lines yaml
Raw Blame History
1id: nextdc2name: NEXTDC3domain: www.nextdc.com4kind: operator5mode: hybrid6priority: 17license: "Publicly accessible operator pages; factual data only; © NEXTDC Limited"8attribution: "NEXTDC"9homepage: https://www.nextdc.com/data-centres10notes: "Facility pages /data-centres/<metro>-data-centres/<code>-<city> publish IT capacity, technical space, rack capacity, Tier and certifications. robots.txt allows / (HubSpot preview paths excluded)."11coverage: { operators: ["NEXTDC"], countries: ["AU", "NZ", "MY", "JP"] }12fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }13discovery:14  sitemap: ["https://www.nextdc.com/sitemap.xml"]15  include: ["/data-centres/[a-z-]+-data-centres/[a-z]{1,2}\\d{1,2}-[a-z-]+$", "/news/", "/media-releases?/", "/newsroom/"]16  exclude: ["\\?", "#"]17  classify:18    - { pattern: "/data-centres/[a-z-]+-data-centres/[a-z]{1,2}\\d{1,2}-[a-z-]+$", group: facility_pages, pageType: facility_page, priority: 60 }19    - { pattern: "/(news|media-releases?|newsroom)/.", group: newsroom, pageType: press_release, priority: 50 }20schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly }21extractors:22  facility: { match: ["/data-centres/[a-z-]+-data-centres/[a-z]{1,2}\\d{1,2}-[a-z-]+$"], kind: facility, parser: nextdc_facility_v1 }23defaults: { operatorName: NEXTDC, facilityType: colocation, countryIso2: AU }24