# Vantage Data Centers — campus pages (WordPress) + news feed
id: vantage
name: Vantage Data Centers
domain: vantage-dc.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © Vantage Data Centers"
attribution: "Vantage Data Centers"
homepage: https://vantage-dc.com/data-center-locations/
notes: >
  www.vantage-dc.com redirects to vantage-dc.com. Yoast sitemaps: data_center-sitemap.xml (campus pages, EN + /fr/
  duplicates) and news-sitemap.xml; RSS at /feed/. Campus pages publish critical IT load, acreage, square footage,
  the campus street address and "City, ST". The stylised map marker (data-lat/data-lng) is unreliable (Ashburn II
  sits ~90 km off) → coordinates deliberately not extracted. French pages excluded.
  robots.txt disallows /management/ only.
coverage: { global: true, operators: ["Vantage Data Centers"] }
fetch:
  level: 1
  maxLevel: 4
  rpm: 20
  concurrency: 2
  respectRobots: true
  maxCreditsPerRun: 40
  maxCreditsPerDay: 120   # premium credits per UTC day for this connector (all runs, all workers)
discovery:
  sitemap: ["/data_center-sitemap.xml", "/news-sitemap.xml"]
  rss: ["/feed/"]
  include: ["/data-center-locations/", "/data-center/", "/news/"]
  exclude: ["\\?", "/fr/", "/management/", "/news/?$", "/data-center-locations/?$"]
  classify:
    - { pattern: "/data-center-locations/(north-america|emea|apac)/[a-z0-9-]+/?$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-center/[a-z0-9-]+/?$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-center-locations/[a-z-]+/?$", group: index, pageType: facility_index, priority: 10 }
    - { pattern: "/news/[a-z0-9._%-]+/?$", group: newsroom, pageType: press_release, priority: 70 }
  maxUrlsPerRun: 1000
schedule:
  facility_pages: weekly
  newsroom: daily
  index: monthly
  sitemap: weekly
  rss: daily
extractors:
  facility:
    # selected by URL only (classifyPage() may mislabel deep facility URLs; the URL is authoritative)
    match: ["/data-center-locations/(north-america|emea|apac)/[a-z0-9-]+/?$", "/data-center/[a-z0-9-]+/?$"]
    kind: facility
    parser: vantage_facility_v1
defaults:
  operatorName: Vantage Data Centers
  operatorKind: developer
  facilityType: wholesale
parserVersion: v1