戦コレDB のデータを CSV に落とす。

senkoredb2csv.rb

#!/usr/bin/env ruby
# coding: utf-8

require 'active_support/all'
require 'open-uri'
require 'nokogiri'
require 'cgi'
require 'ir_b'

def parse(no, body)
  doc = Nokogiri::HTML(body)

  text = doc.text

  sub, name = doc.to_s.scan(/<\/span>(?:<br>)?(?:\[(.*)?\])?(.*?)<span/)[2]
  return unless name

  name = CGI.unescapeHTML(name)

  type = doc.search('span')[1].text rescue nil
  type_a = { '炎' => 'R', '水' => 'B', '風' => 'G' }[type]
  num  = text.scan(/No\.(\d+)/)[0][0].to_i
  gender = text.scan(/♂|♀/)[0]
  rare = text.scan(/S?(?:ﾚｱ|ﾉｰﾏﾙ)\+?/)[0]
  rare = rare.sub(/ﾚｱ/,"R").sub(/ﾉｰﾏﾙ/,"N") unless rare.empty?
  magic = text.scan(/魔力(\d+)/).join
  atk, matk = text.scan(/攻撃力(\d+)⇒(\d*)/)[0]
  deff, mdeff = text.scan(/防御力(\d+)⇒(\d*)/)[0]
  skill = doc.at("div.skill span a").try(:text)

  [no, type, num, gender, rare, sub, name, magic, atk, deff, matk, mdeff, skill, type_a]
end

def direct(no)
  $stderr.puts "getting http://enedo.velvet.jp/senkore/sk_data.php?no=#{no}"
  body = open("http://enedo.velvet.jp/senkore/sk_data.php?no=#{no}", 'r:cp932'){|f| f.read.encode("utf-8") }
  parse(no, body)
end


def output_csv(from, to, out = $stdout)
  to ||= from
  (from..to).each_with_index do |no, idx|
    sleep(rand(5) + 5) unless idx == 0
    result = direct(no)
    out.puts result.join(',') if result
  end
end

def main(args)
  case args.length
  when 0
    from = 1
    to   = 890
  when 1
    from = args[0].to_i 
    to   = nil
  else
    from = args[0].to_i
    to   = args[1].to_i
  end

  from = 1    if from.zero?
  to   = from if to.nil? || to.zero?

  output_csv(from, to)
end

main(ARGV)