spb/datacenterindex

1# Data Centre Magazine — generated connector definition (see docs/connectors/datacentremagazine.md)2id: datacentremagazine3name: "Data Centre Magazine"4domain: datacentremagazine.com5kind: news6mode: hybrid7priority: 38license: "Headlines, dates, facts and links only; \u00a9 publisher"9attribution: "Data Centre Magazine (datacentremagazine.com)"10homepage: https://datacentremagazine.com/11notes: "BizClik title; no RSS, the /news and /articles indexes are server-rendered and open, but article pages sit behind a Cloudflare challenge (403 at L1 and L2): the runtime escalates to Scrapfly (L4). Keep the credit budget small."12coverage: { global: true }13fetch:14  level: 115  maxLevel: 416  renderJs: false        # static article HTML; ASP only (≈1 credit/page instead of ≈6)17  maxCreditsPerDay: 60   # shared Redis cap across workers18  rpm: 619  concurrency: 120  respectRobots: true21  maxCreditsPerRun: 4022discovery:23  seeds:24    - { url: "/news", group: seed, pageType: news_index }25    - { url: "/articles", group: seed, pageType: news_index }26  follow: ["^https://datacentremagazine\\.com/news/[a-z0-9-]+$"]27  include: ["/news/[a-z0-9-]+$"]28  exclude: ["/executives?/"]29  classify:30    - { pattern: "/news/[a-z0-9-]+$", group: newsroom, pageType: press_release, priority: 70 }31schedule:32  newsroom: 6h33  rss: 6h34  seed: 6h35extractors:36  news:37    pageTypes: [press_release, project_announcement, expansion, acquisition, construction_update, planning_document, power_infrastructure, utility_announcement, closure, cloud_region, financial_disclosure, facility_page, operator_page, unknown]38    kind: news_event39    parser: news_article_v140    params: { keepText: false }41