Ruby script to generate .ics files for the Staatstheater Nürnberg schedule

schedule.rb

require "net/http"
require "nokogiri"
require "date"
require "json"
require "digest/sha1"

class Schedule
  attr_reader :genre

  def initialize
    @genre = :all
  end

  def genre=(genre)
    if GENRE_KEYS.keys.include?(genre)
      @genre = genre
    else
      raise ArgumentError, "unknown genre `#{genre}'"
    end
  end

  def months
    resp = get("/index.php?page=spielplan")
    doc = Nokogiri::HTML(resp.body)
    doc.css("select[name='filter[month]'] option").map { |node| node["value"].strip }
  end

  def events
    [].tap do |entries|
      hashes = []
      months.map do |month|
        entries_for_month(month).each do |entry|
          unless hashes.include?(entry.fetch(:hash))
            hashes << entry.fetch(:hash)
            entries << entry
          end
        end
      end
    end
  end

  private

  def entries_for_month(month)
    resp = post("/index.php?page=spielplan", "filter[day]=1&filter[month]=#{month}&filter[genre]=#{genre_key}&filter[production]=0")
    resp = get("/index.php?page=spielplan")
    doc = Nokogiri::HTML(resp.body)
    doc.css(".entry[title]").map { |entry| parse_entry(entry) }
  end

  GENRE_KEYS = {
    all: "0",
    opera: "3,11,17,13,9,21",
    ballet: "5",
    concert: "6,12,16,22,47",
    play: "4",
    guest: "18,15,54,60",
  }

  def genre_key
    GENRE_KEYS.fetch(@genre)
  end

  DEFAULT_DURATION = 2 * 60 * 60

  def parse_entry(entry)
    {}.tap do |hash|
      if entry.attr("title") =~ /\A(\d{1,2})\.(\d{1,2})\.(\d{4}) (\d{1,2}):(\d{2})/
        year, month, day, hour, minute = $3.to_i, $2.to_i, $1.to_i, $4.to_i, $5.to_i
      else
        raise "could not parse date (#{entry.attr("title").inspect})"
      end

      find_first(entry, ".time") do |text|
        case text.strip
        when /^(\d{1,2}):(\d{2}) Uhr, (.*)$/
          hash[:start] = Time.mktime(year, month, day, $1.to_i, $2.to_i)
          hash[:end] = hash[:start] + DEFAULT_DURATION
          hash[:location] = $3.strip
        when /^(\d{1,2}):(\d{2}) - (\d{1,2}):(\d{2}) Uhr, (.*)$/
          hash[:start] = Time.mktime(year, month, day, $1.to_i, $2.to_i)
          hash[:end] = Time.mktime(year, month, day, $3.to_i, $4.to_i)
          hash[:location] = $5.strip
        end
      end

      hash[:start] ||= Time.mktime(year, month, day, hour, minute)
      hash[:end] ||= hash[:start] + DEFAULT_DURATION

      find_first(entry, ".title") do |text, node|
        if node["onclick"]
          matches = node["onclick"].match(/document.location.href='(.*)';/)
          hash[:url] = "http://www.staatstheater-nuernberg.de/#{matches[1]}"
        end
        hash[:title] = text.strip
      end

      hash[:hash] = hash_entry(hash)
    end
  end

  def hash_entry(entry)
    Digest::SHA1.hexdigest(entry.to_json)
  end

  def find_first(node, selector)
    nodes = node.css(selector)
    if nodes.any?
      node = nodes.first
      yield node.text, node
    else
      raise "no match"
    end
  end

  def http
    @http ||= Net::HTTP.new("www.staatstheater-nuernberg.de")
  end

  def get(path, initheader = {}, dest = nil)
    initheader["Cookie"] ||= @cookie if @cookie
    resp = http.get(path, initheader, dest)
    @cookie = resp["Set-Cookie"] if resp.key?("Set-Cookie")
    resp
  end

  def post(path, data, initheader = {}, dest = nil)
    initheader["Cookie"] ||= @cookie if @cookie
    resp = http.post(path, data, initheader, dest)
    @cookie = resp["Set-Cookie"] if resp.key?("Set-Cookie")
    resp
  end
end

class Calendar
  def self.calendar_for_events(events)
    "".tap do |cal|
      cal << "BEGIN:VCALENDAR\n"
      cal << "VERSION:2.0\n"
      cal << "PRODID:https://thcyron.de/staatstheater-nuernberg-ical/\n"
      events.each do |event|
        cal << "BEGIN:VEVENT\n"
        cal << "UID:staatstheater-nuernberg-ical+#{event.fetch(:hash)}@thcyron.de\n"
        if event.has_key?(:location)
          cal << "LOCATION:#{escape_newlines(event.fetch(:location))}\n"
        end
        cal << "SUMMARY:#{escape_newlines(event.fetch(:title))}\n"
        cal << "DTSTART:#{event.fetch(:start).utc.strftime("%Y%m%dT%H%M%SZ")}\n"
        cal << "DTEND:#{event.fetch(:end).utc.strftime("%Y%m%dT%H%M%SZ")}\n"
        cal << "URL:#{event.fetch(:url)}\n" if event.has_key?(:url)
        cal << "END:VEVENT\n"
      end
      cal << "END:VCALENDAR\n"
    end
  end

  def self.escape_newlines(s)
    s.gsub("\n", "\\n").gsub("\r", "")
  end
end

if $0 == __FILE__
  ENV["TZ"] = "Europe/Berlin"
  sched = Schedule.new
  sched.genre = ARGV.first.to_sym if ARGV.size > 0 
  puts Calendar.calendar_for_events(sched.events)
end