ドクペ価格botを支える技術

amzn-drpepper-price-bot.rb

# encoding: utf-8
require "open-uri"
require "rubygems"
require "nokogiri"
require "kconv"
require "twitter"

# load OAuth keys and tokens
load File.expand_path(File.dirname(__FILE__) + '/key.rb')

# log file for price comparing
PRICE_LOG = File.expand_path(File.dirname(__FILE__) + "/pricelog")

# Dr.Peppers to retrieve price
target_drinks =
  {"500mlペット×24" => {"uri" => "http://www.amazon.co.jp/dp/B001U7651A/", "amt" => 500, "pcs" => 24},
   "350ml缶×24" => {"uri" => "http://www.amazon.co.jp/dp/B001DDOIBU/", "amt" => 350, "pcs" => 24}}

def trim_price (target_css_selector)
  target_css_selector.gsub(/[[:blank:][:space:]]/, '').gsub(/<[a-zA-Z="\/]*>/, '').gsub(/([0-9,]{256})/, '\1')
end

# confirm whether the specification of goods page of amazons is not changed
def test_amazon_html
  tester = "http://www.amazon.co.jp/dp/4101098018/" # ボッコちゃん
  html = open(tester).read
  page = Nokogiri::HTML(html, nil, 'mbcs')
  # has
  (!!(trim_price(page.css('.priceLarge').to_html().encode('utf-8')) =~ /[0-9]+/))
end

# retrieve prices and send tweet
if test_amazon_html then
  # detect current time
  time = Time.now
  # format current minute of time to two digits
  tweet_text = "#{time.month}/#{time.day} #{time.hour}:%02d現在のドクペ価格は，"%[time.min]
  # open price log file
  price_log = File.open(PRICE_LOG, "r:UTF-8")
  old_price = {}
  # parse log file
  price_log.each do |line|
    old_price.merge!({line.match(/(^.+),/)[1] => Integer(line.match(/,([0-9]+)$/)[1])})
  end
  price_log.close
  price_log = File.open(PRICE_LOG, "w:UTF-8")
  price_log.write("")
  # {drink_name => old_price}
  target_drinks.each { |name, prop|
    html = open(prop["uri"]).read
    page = Nokogiri::HTML(html, nil, 'mbcs')
    price = trim_price(page.css('.priceLarge').to_html().encode('utf-8'))
    real_price = Integer(price.gsub(/[^0-9]/, ""))
    # write price into log here to compare transition of price
    price_log.puts "#{name},#{real_price}"
    # compare transition of price
    price_comparison = ""
    if (old_price[name] == real_price) then
      price_comparison = "±￥0"
    elsif (old_price[name] > real_price) then
      price_comparison = "↓￥#{old_price[name] - real_price}"
    else
      price_comparison = "↑￥#{real_price - old_price[name]}"
    end
    tweet_text += "\n#{name}が #{price}（#{price_comparison}，￥#{real_price / prop["pcs"]}/本，￥%.3f/ml）"%[(real_price + 0.0) / (prop["pcs"] * prop["amt"] + 0.0)]
  }
  price_log.close()
else
  print "Error: scraping failed\n"
  tweet_text = "@polamjag Error: price not found in Amazon's page"
end

Twitter.update(tweet_text)

# output log
log_time = Time.now()
print "Sent at #{log_time.year}-#{log_time.month}-#{log_time.day} #{log_time.hour}:#{log_time.min}: #{tweet_text}\n"

key.rb

require 'twitter'

# add your own key below
Twitter.configure do |config|
  config.consumer_key       = ""
  config.consumer_secret    = ""
  config.oauth_token        = ""
  config.oauth_token_secret = ""
end