amazon_sales_rank.rb

#!/usr/bin/env ruby
# -*- coding: utf-8 -*-
require 'rubygems'
require 'hpricot'
require 'open-uri'
require 'nkf'
$KCODE = "u" if VERSION < "1.9.0"

class AmazonItem
  attr_accessor :asin
  def initialize(asin)
    @asin = asin
  end

  def url
    "http://amazon.jp/dp/%s" % @asin
  end

  def title
    @title ||= Hpricot(content).search("#btAsinTitle").inner_text
  end

  def rank
    return @rank if @rank 
    rank = Hpricot(content).search("#SalesRank").inner_text.scan(/([\d,]*)位/u)[0][0]
    @rank = Integer(rank.tr(",","")) rescue nil
  end

  private
  def content
    @doc ||= NKF.nkf("-w", open(url).read)
  end
end

def write_sales_rank_to(out, asin)
  item = AmazonItem.new(asin)
  [out, $stdout].each do |o|
    o.write("#{item.asin}\t#{Time.now.to_i}\t#{item.rank}\n")
  end
end

TSV_BASE = File.expand_path("amazon.co.jp/sales_rank", ENV["HOME"])
def crowl_sales_rank(asin, fname)
  path = "#{TSV_BASE}/#{fname}.tsv"
  File.open(path, 'a') do |out|
    write_sales_rank_to(out, asin)
  end
end

entries = {
  "4774127280" => "lifehack-press",
  "4873113202" => "java2ruby",
  "4274066940" => "agile_practice",
  "4873113660" => "interface_oriented_design",
}

entries.each do |asin, fname|
  crowl_sales_rank(asin, fname)
end