日本雑誌協会の印刷部数公表データからコミック誌の部数を拾ってきてCSV出力するスクリプトです。

j-magazine-scraper.rb

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

uri_template = 'https://www.j-magazine.or.jp/user/printed/index/%d/%d'

result = []
periods = []
genre_list = [
  14, # 少年向けコミック誌
  15, # 男性向けコミック誌
  29, # 少女向けコミック誌
  30, # 女性向けコミック誌
]

(1..41).map {|period|
  genre_list.each do |genre|
    doc = Nokogiri::HTML(open(uri_template % [period, genre]))
    if periods.empty?
      doc.search('#period_cd_top option').each do |option|
        periods[option.attribute('value').value.to_i] = option.inner_html
      end
    end

    doc.search("#content table tr").each do |tr|
      tr.search("td.MagDataTab_MagNa a").each do |anchor|
        title = anchor.inner_html
        link = anchor.attribute('href').value
        count = tr.search('td.MagDataTab_count')[0].inner_html

        result.push({period: period, title: title, link: link, count: count})
      end
    end
  end
}

box = []
titles = ['期間']
result.each do |r|
  i = titles.index(r[:title])
  unless i
    titles.push(r[:title])
    i = titles.index(r[:title])
  end

  box[r[:period]] = [] unless box[r[:period]]
  box[r[:period]][i] = r[:count]
  box[r[:period]][0] = periods[r[:period].to_i] unless box[r[:period]][0]
end
box[0] = titles

box.each do |b|
  print b.to_csv
end