website-stalker.yaml

---
sites:
  - url:
      - https://www.haw-hamburg.de/cyberangriff/
      - https://www.haw-hamburg.de/cyberangriff/datenleak/

      - https://www.haw-hamburg.de/hochschule/technik-und-informatik/studium-und-lehre/fakultaetsservicebuero/

      # Quellen des Kalenderbots
      - https://www.haw-hamburg.de/hochschule/technik-und-informatik/departments/informatik/studium/aktuelle-veranstaltungen/
      - https://www.haw-hamburg.de/en/study/degree-courses-a-z/study-courses-in-detail/course/courses/show/information-engineering/Studierende/
      - https://www.haw-hamburg.de/hochschule/technik-und-informatik/departments/informations-und-elektrotechnik/studium/studienorganisation/studienplaene/
      - https://www.haw-hamburg.de/studium/studiengaenge-a-z/studiengaenge-detail/course/courses/show/elektrotechnik-und-informationstechnik/Studierende/
    editors:
      - css_select: '#main, .last_change'
      - css_remove: img, .side-bar, .dots
      - html_markdownify

  - url:
      - https://www.haw-hamburg.de/detail/news/news/show/informationen-aus-dem-department-informatik-zum-angriff-auf-die-it/
    editors:
      - css_select: .article
      - css_remove: img, .dots
      - html_markdownify

  - url: https://userdoc.informatik.haw-hamburg.de/doku.php?id=start
    editors:
      - css_select: .page
      - css_remove: script, img, .dw__toc
      - html_markdownify
  - url:
      # Quellen des Kalenderbots
      - https://userdoc.informatik.haw-hamburg.de/doku.php?id=stundenplan:ics_public&do=media&ns=stundenplan
      - https://userdoc.informatik.haw-hamburg.de/doku.php?id=stundenplan:ics_public
    editors:
      - css_select: .page
      - css_remove: script, img, .dw__toc
      - html_prettify

  # Way too noisy because of duplicated mensa entries on the page (they should finally get their data in check)
  # - url: https://www.stwhh.de/gastronomie/mensen-cafes-weiteres
  #   editors:
  #     - css_select: "#page-content .tx-epwerkbuilding-element-wrapper"
  #     - css_remove: svg, img, .tx-eptemplate-werk-button
  #     - css_sort: # Sort the contents of the wrapper. This assumes that the headline stuff ends up at the top
  #         selector: .tx-epwerkbuilding-element-wrapper > div
  #     - css_sort: # Sort all the wrappers ideally by their headline and then whatever follows
  #         selector: .tx-epwerkbuilding-element-wrapper
  #         sort_by:
  #           - html_textify
  #     - regex_replace:
  #         pattern: <div.+tx-epwerkbuilding-list-headline[^>]+>[\s\n]*(.+)[\s\n]*</div>
  #         replace: <h3>$1</h3>
  #     - html_markdownify