Get Cinderella Girls

scraping.rb

require 'open-uri'
require 'Oga'

base_uri = "http://imas.gamedbs.jp"

def open_uri(uri)
  html = open(uri) do |f|
    f.read
  end

  Oga.parse_html(html)
end

begin
  # トップページ読み込み
  doc = open_uri(base_uri + "/cg")

  # 詳細ページリンクを取る
  detail_uris = []
  doc.xpath('//*[@id="result"]/ul/li/a').each do |node|
    detail_uris.push(node.get('href'))
  end

  # 詳細ページ読み込み
  detail_uris.each do |detail_uri|

    idle_doc = open_uri(base_uri + detail_uri)

    # 個別のカードページURLを取得
    card_uris = []
    idle_doc.xpath('//*[@id="gallery"]/div[5]/div/div/div/a').each do |node|
      card_uris.push(node.get('href'))
    end

    # 個別のカードページ読み込み
    idle_info = []
    card_uris.each_with_index do |card_uri, i|
      if i % 2 == 0
        card_doc = open_uri("http:" + card_uri)

        for n in 1..2 do
          tmp = {}
          card_doc.xpath('//*[@id="cards"]/ul/li[' + n.to_s + ']/h2').map{|n|
            info = n.text.strip.gsub(" ", "").split("\n")
            # rank = info[0]
            # name = info[1]
            tmp["rank"] = info[0]
            tmp["name"] = info[1]
          }
          card_doc.xpath('//*[@id="cards"]/ul/li[' + n.to_s + ']/div/div[1]/ul/li[1]/a/img').map{|n|
            tmp["img"] = n.attributes[1].value
          }
          idle_info.push(tmp)
        end
      end
      p "idle_info"
      p idle_info
    end
  end
rescue Interrupt
  p 'QUIT!'
end