scrapers/riesa-efau-kalender/scrape.rb

#!/usr/bin/env ruby
# coding: utf-8

require 'open-uri'
require 'nokogiri'
require 'erb'

MONTHS = [
  "Januar", "Februar", "März", "April",
  "Mai", "Juni", "Juli", "August",
  "September", "Oktober", "November", "Dezember",
]

def fmt_time t
  t.strftime "%Y%m%dT%H%M%S"
end

class Event
  attr_accessor :name, :location, :link, :image
  attr_reader :date

  def date=(s)
    if s =~ /(\d{1,2})\. (.+?) (\d{4}), (\d\d):(\d\d) Uhr/
      year = $3.to_i
      month = MONTHS.index($2) + 1
      day = $1.to_i
      hour = $4.to_i
      minute = $5.to_i
      @date = Time::local year, month, day, hour, minute, 0
    else
      raise "Invalid date: #{s.inspect}"
    end
  end
end

events = []

url = "https://riesa-efau.de/index.php?id=87"
doc = Nokogiri::HTML URI.open(url)
doc.css(".tx-cal-controller .vevent").each do |vevent|
  ev = Event::new
  ev.name = vevent.css("h3").children[0].text
  ev.location = vevent.css(".location").text
                  .sub(/ *\.\.\.[^\.]*$/, "")
  ev.date = vevent.css(".datum").text
  ev.link = URI.join url, vevent.css(".url").attr('href').value
  ev.image = URI.join url, vevent.css(".images img").attr('src').value
  events << ev
end

events.sort_by! { |ev| ev.date }

ical = ERB::new <<~EOF
  BEGIN:VCALENDAR
  VERSION:2.0
  METHOD:PUBLISH
  X-WR-TIMEZONE;VALUE=TEXT:Europe/Berlin
  <% events.each  do |ev| %>
  BEGIN:VEVENT
  METHOD:PUBLISH
  CLASS:PUBLIC
  UID:<%= ev.link %>
  DTSTART:<%= fmt_time(ev.date) %>
  DTEND:<%= fmt_time(ev.date + 2 * 3600) %>
  SUMMARY:<%= ev.name %>
  LOCATION:<%= ev.location %>
  URL:<%= ev.link %>
  ATTACH;FMTTYPE=image/jpeg:<%= ev.image %>
  END:VEVENT
  <% end  %>
  END:VCALENDAR
EOF

puts ical.result