しょぼいカレンダーからスタッフとキャストをパースして保存する

syoboi.rb

# coding: utf-8
require 'nokogiri'
require 'open-uri'

module My
  class SyoboiParser
    attr_accessor :anime_number
    def initialize(anime_number)
      @anime_number = anime_number
    end

    def get_html()
      @doc = Nokogiri::HTML(open("http://cal.syoboi.jp/tid/#{@anime_number}"))
    end

    def add_hash(person_name, hash)
      hash[person_name] = 1.0
    end

    def pick_staffandcast(hash)
      hash["title"] = @doc.search("//div[@id='main']//h1").to_html.gsub(/<\/?[^>]*>/, ",").split(",")[1]
      striped_elem = Array.new
      @doc.search("//table[@class='section staff']//a").each do |elem|

        striped_elem = elem.to_html.gsub(/<\/?[^>]*>/, "")
        next if striped_elem.length < 1
        puts "add_hash ok." if add_hash(striped_elem, hash)
      end
      @doc.search("//table[@class='section cast']//a").each do |elem|
        striped_elem = elem.to_html.gsub(/<\/?[^>]*>/, "")
        next if striped_elem.length < 1
        if add_hash(striped_elem, hash)
        end
      end
    end
  end
end


anime_number = 1
syoboi = My::SyoboiParser.new(anime_number)

anime_hash = Array.new

while syoboi.get_html() do
  puts anime_number
  anime_hash[anime_number] = Hash.new(0)
  syoboi.pick_staffandcast(anime_hash[anime_number])
  p anime_hash[anime_number]

  # $B%k!<%W$9$k2s?t(B(=$B$7$g$\$$%+%l%s%@!<$KEPO?$5$l$F$$$kHVAH?t(B)
  # ($B%O!<%I%3!<%G%#%s%0$9$.$k$,!";~4V$,L5$$$N$G!D(B)
  howmany = 2842
  break if anime_number >= howmany
  anime_number += 1
  syoboi = My::SyoboiParser.new(anime_number)
end

all_list=[]
for anime_hash_each in anime_hash.each
  if anime_hash_each.is_a?(Hash)
    for word in anime_hash_each.keys
      all_list << word
    end
  end
end
all_list_uniq = all_list.uniq

for word in all_list_uniq.each
  open("data.txt","a"){|f| f.print "#{word}\t"}
end

for anime_hash_each in anime_hash.each
  if anime_hash_each.is_a?(Hash)
    open("data.txt","a"){|f| f.puts }
    for word in all_list_uniq.each
      open("data.txt","a"){|f| f.print "#{anime_hash_each[word]}\t"}
    end
  end
end