# Switch — campus pages (The Core, The Citadel, The Keep, The Pyramid) + WordPress news feed
id: switch
name: Switch
domain: www.switch.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © Switch, Inc."
attribution: "Switch"
homepage: https://www.switch.com/
notes: >
  Open WordPress site (Yoast sitemaps page-sitemap.xml / post-sitemap.xml, RSS /feed/). The four campus pages
  (/core/, /citadel/, /keep/, /pyramid/) render their body client-side (empty <article> in HTML), so only the
  server-rendered <title> ("Switch – The Core Campus, Las Vegas, Nevada") and meta description ("Tier 5 Platinum …
  up to 2.4 million sqft and 315 MW") are extracted: name, city/state, tier, planned MW ("up to"), sq ft.
  No address or coordinates. Market pages (/las-vegas/, /tahoe-reno/, /atlanta/, /grand-rapids/, /austin/) are
  indexes. Press releases are root-level WordPress posts (from post-sitemap + RSS); marketing pages are excluded
  explicitly and the news fallback drops anything not data-center related. robots.txt has no disallows.
coverage: { countries: ["US"], operators: ["Switch"] }
fetch:
  level: 1
  maxLevel: 4
  rpm: 20
  concurrency: 2
  respectRobots: true
  maxCreditsPerRun: 30
  maxCreditsPerDay: 120   # premium credits per UTC day for this connector (all runs, all workers)
discovery:
  sitemap: ["/page-sitemap.xml", "/post-sitemap.xml"]
  rss: ["/feed/"]
  include: []
  exclude:
    - "\\?"
    - "^https://www\\.switch\\.com/$"
    - "^https://www\\.switch\\.com/[a-z0-9-]+/[a-z0-9-]+/"   # two-level marketing paths (careers, partnerportal, sustainable-data-centers…); posts are single-level
    - "/(ebc|brochure|l|videos|reports|resources|contact[a-z0-9-]*|gdpr|compliance|privacy[a-z0-9-]*|application|switch-[a-z0-9-]*|switchstack|inventrepreneur|wireless|telecom-meeting-request|referral-partner-questionnaire|partner-lead-registration-request|non-badged-access|bgp-communities|photos-slider|slider-options|sponsorship-terms-and-conditions|core-providers|the-core-cooperative|patent-video|green|wp-content|feed|tag|category|author|page)/"
  classify:
    - { pattern: "^https://www\\.switch\\.com/(core|citadel|keep|pyramid|rock)/$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "^https://www\\.switch\\.com/(las-vegas|tahoe-reno|atlanta|grand-rapids|austin|locations-[a-z0-9-]+)/$", group: index, pageType: facility_index, priority: 20 }
    - { pattern: "^https://www\\.switch\\.com/[a-z0-9-]+/$", group: newsroom, pageType: press_release, priority: 40 }
  maxUrlsPerRun: 600
schedule:
  facility_pages: weekly
  newsroom: daily
  index: monthly
  sitemap: weekly
  rss: daily
extractors:
  campus:
    # selected by URL only (classifyPage() may mislabel deep facility URLs; the URL is authoritative)
    match: ["^https://www\\.switch\\.com/(core|citadel|keep|pyramid|rock)/$"]
    kind: facility
    parser: switch_campus_v1
defaults:
  operatorName: Switch
  operatorKind: colocation
  facilityType: colocation
parserVersion: v1