spb/datacenterindex

Public

338.7 MBsize

maindefault branch

3 h agolast push

HTML 53.9% TypeScript 44.5% JavaScript 0.6% SQL 0.5%

1.3 KB · 37 lines yaml

Raw Blame History

1# Prince William County, Virginia — News — generated connector definition (see docs/connectors/prince-william-county.md)2id: prince-william-county3name: "Prince William County, Virginia \u2014 News"4domain: www.pwcva.gov5kind: government6mode: hybrid7priority: 28license: "Public record (Prince William County Government)"9attribution: "Prince William County, Virginia (pwcva.gov)"10homepage: https://www.pwcva.gov/11notes: "Drupal news index, no working RSS (403). robots.txt answers 403 \u2192 treated as no rules."12coverage: { countries: ["US"] }13fetch:14  level: 115  maxLevel: 216  rpm: 1017  concurrency: 118  respectRobots: true19  maxCreditsPerRun: 020discovery:21  seeds:22    - { url: "/news", group: seed, pageType: news_index }23  follow: ["^https://www\\.pwcva\\.gov/news/[a-z0-9-]+$"]24  include: ["/news/[a-z0-9-]+$"]25  classify:26    - { pattern: "/news/[a-z0-9-]+$", group: newsroom, pageType: press_release, priority: 70 }27schedule:28  newsroom: daily29  rss: daily30  seed: daily31extractors:32  news:33    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]34    kind: news_event35    parser: news_article_v136    params: { keepText: true }37