CEDEC 2017 のセッション情報を取得する

cedec2017-session-scraper.rb

#!/usr/bin/env ruby

require 'open-uri'
require 'nokogiri'
require 'json'

sessions = []

%w(
  http://cedec.cesa.or.jp/2017/session/schedule_0830/
  http://cedec.cesa.or.jp/2017/session/schedule_0831/
  http://cedec.cesa.or.jp/2017/session/schedule_0901/
).each do |url|
  doc = Nokogiri::HTML(open(url))
  uri = URI.parse(url)
  date = doc.title[%r{^\d+/\d+}]

  doc.search('ul.schedule_session_index > li').each do |li|
    session = { date: date }
    %i(ss_time_start ss_time_end room_number ss_style ss_title schedule_speaker).each do |klass|
      session[klass] = li.at(".#{klass}").inner_text.strip.gsub("\u3000", ' ').gsub(/\s{2,}/, ' ')
    end
    genres = []
    translation = false

    li.search('.ss_spec > img').each do |img|
      case img[:src]
      when %r{/genre_icon/}
        genres << File.basename(img[:src]).sub(/_.*$/, '')
      when %r{/trans_sim_ja}
        translation = true
      end
    end

    session[:url] = (uri + li.at('.ss_title a')[:href]).to_s
    session[:genres] = genres
    session[:translation] = translation
    session[:type] = li.at('.ss_ippr_icon > img')&.attr(:title)
    session[:timeshift] = li.at('.timeshift-icon > img')[:title].end_with?('あり')
    session[:ss_style].delete!('［］')

    sessions << session
  end
end

puts JSON.generate(sessions)