so-net XMLTV exporter

so-net-xmltv.rb

#!/usr/bin/env ruby
# coding: utf-8
require 'date'
require 'open-uri'
require 'nokogiri'

class Station
  attr_accessor :id, :name, :channel

  def xmltv_id
    id + ".tv.so-net.ne.jp"
  end
end

class Program
  attr_accessor :title, :subtitle, :summary, :start, :stop, :station, :genres, :replay, :new

  def url
    URI("https://tv.so-net.ne.jp/schedule/#{station.id}#{start.strftime("%Y%m%d%H%M")}.action")
  end
end

class Genre
  attr_reader :text

  TRANSLATION = {
    "劇場/公演" => nil,
    "趣味/教育" => "Educational",
    "情報/ワイドショー" => nil,
    "映画" => "Movie",
    "福祉" => nil,
    "音楽" => "Music",
    "アニメ/特撮" => "Animated",
    "その他" => "Other",
    "ドラマ" => "Drama",
    "ニュース/報道" => "News",
    "スポーツ" => "Sports",
    "バラエティー" => "Talk Show",
    "ドキュメンタリー/教養" => "Documentary",
  }

  def initialize(text)
    @text = text
  end

  def translated
    TRANSLATION[@text]
  end
end

class Fetcher
  def initialize
    @area = ENV['AREA_ID'] || '23' # Tokyo
    @stations = {}
    @previous = {}
    @programs = []
    @genre    = {}
  end

  def dates
    (0..7).map do |i|
      Date.today + i
    end
  end

  def run
    dates.each do |date|
      fetch "https://tv.so-net.ne.jp/chart/#{@area}.action?head=#{date.strftime('%Y%m%d')}0000&span=24"
    end

    fixup_last_program

    puts emit_tv_xml
  end

  def fixup_last_program
    @programs.each do |program|
      unless program.stop
        warn "Fetching #{program.url}"
        doc = Nokogiri::HTML(open(program.url))
        minutes = /（(\d+)分）/
                  .match( doc.css(".mainColumn dl.basicTxt dd").text )[1].to_i
        program.stop = program.start + minutes * 60
      end
    end
  end

  def fetch(url)
    warn "Fetching #{url}"
    doc = Nokogiri::HTML(open(URI(url)))

    doc.css('.cell-station.cell-top').each do |cell|
      if cell.attr('id') =~ /^cell-station-top-(\d+)/
        id = $1
        name = cell.attr('title')
        channel = /^\s*(\d+)\./.match(cell.css('a:not(.inlineArrow)').text)[1]

        station = Station.new
        station.id = id
        station.name = name
        station.channel = channel

        @stations[station.id] ||= station
      end
    end

    doc.css('#chart-footer-genres li label').each do |label|
      id = label.css('input').first.attr('value')
      @genre[id] ||= Genre.new(label.text.strip)
    end

    doc.css('.cell-schedule').each do |cell|
      a = cell.css('.td-schedule a.schedule-link').first or next
      match = /\/schedule\/(?<station>\d{6})(?<time>\d{12})/
              .match(a.attr('href'))

      genres = cell.attr('class').scan(/cell-genre-(\d+)/).flatten.map {|id| @genre[id] }.compact

      program = Program.new
      program.summary = cell.css('.schedule-summary').text.strip
      program.start = Time.strptime(match[:time], '%Y%m%d%H%M')
      program.station = @stations[match[:station]]
      program.genres = genres
      
      title = a.text.strip
      title.gsub! /(?:\[(?:字|解|デ|(再)|二|双)\])+(.*)$/ do
        if $1
          program.replay = true
        end

        if $2
          program.subtitle = $2
        end
        ''
      end

      title.sub! /^\[新\]/ do
        program.new = true
        ''
      end

      title.gsub! /(?:【|「)(.+?)(?:】|」)$/ do
        program.subtitle = $1
        ''
      end

      program.title = title

      if prev = @previous[match[:station]]
        prev.stop = program.start
      end
      
      @previous[match[:station]] = program
      @programs << program
    end
  end

  def emit_tv_xml
    builder = Nokogiri::XML::Builder.new(encoding: "UTF-8") do |xml|
      xml.doc.create_internal_subset('tv', nil, "xmltv.dtd")
      
      xml.tv(:"source-info-url" => "https://tv.so-net.ne.jp/", :"source-info-name" => "So-Net TV") {
        @stations.each do |id, station|
          xml.channel(id: station.xmltv_id) {
            xml.send(:"display-name", station.name)
            xml.send(:"display-name", station.channel + " " + station.name)
            xml.send(:"display-name", station.channel)
          }
        end

        @programs.each do |program|
          next unless program.stop

          xml.programme(
            start: program.start.strftime("%Y%m%d%H%M%S %z"),
            stop: program.stop.strftime("%Y%m%d%H%M%S %z"),
            channel: program.station.xmltv_id,
          ) {
            xml.title(lang: "ja") { xml.text program.title }
            if program.subtitle
              xml.send(:"sub-title", lang: "ja") { xml.text program.subtitle }
            end
            xml.desc(lang: "ja") { xml.text program.summary }
            program.genres.each do |genre|
              xml.category(lang: "ja") { xml.text genre.text }
              if genre.translated
                xml.category(lang: "en") { xml.text genre.translated }
              end
            end
            if program.replay
              xml.send :"previously-shown"
            end
            if program.new
              xml.new
            end
          }
        end
      }
    end

    builder.to_xml
  end
end

ENV['TZ'] = 'Asia/Tokyo'

Fetcher.new.run