HorseList from netkeiba

scrape.rb

require 'rubygems'
require 'nokogiri'
require 'kconv'

f = File.open(ARGV[0])
doc = Nokogiri::HTML(f.read.toutf8, nil, 'utf-8')
f.close
doc.search("//table[@class='nk_tb_common race_table_01']//tr[position()>1]").each{|tr|
  row = Array.new
  row.push tr.search("td[2]").text.chomp
  row.push tr.search("td[2] > a").attribute("href").value
  #doc.search("//p[@class='db_prof_top_kakikomi_num']//a").each{|a|
  #  puts a.text.tosjis
  #}
  row.push tr.search("td[3]").text.chomp
  row.push tr.search("td[6] > a").text.chomp
  row.push tr.search("td[7] > a").text.chomp
  row.push tr.search("td[8] > a").text.chomp
  row.push tr.search("td[9] > a").text.chomp
  row.push tr.search("td[10] > a").text.chomp
  row.push tr.search("td[11] > a").text.chomp
  puts row.join(",")
}