Quick and dirty implementation of HK flight data parsing

hk_flights.rb

#!/usr/bin/env ruby

require 'curb'
require 'nokogiri'
require 'date'

CURRENT_DATE = Date.today.strftime("%Y-%-m-%d")
link = "http://www.hongkongairport.com/flightinfo/eng/real_arrinfo.do?fromDate<wbr>=#{CURRENT_DATE}"
USER_AGENT = "Mac / Firefox 29: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:29.0) Gecko/20100101 Firefox/29.0"

h = Curl::Easy.new(link) do |http|
  http.headers["User-Agent"] = USER_AGENT
  http.follow_location = true
  http.enable_cookies = true
  http.cookiefile = "cookies-2015Mar.txt"
  http.cookiejar = "cookies-2015Mar.txt"
  http.verbose = false
end

h.perform

html = Nokogiri::HTML(h.body_str) do |config|
  config.noent.noblanks.noerror
end

html.search("tr[@mr='true']").remove
rows = html.xpath('//table[@id="list"]/tr')

flight_records = rows.collect do |row|
  if row.attr("date")
    date = Date.strptime(row.attr("date").split(",").last, "%F")
  end

  {}.tap do |flight_record|
    [
      ['td[1]/text()', :time],
      ['td[2]/text()', :flight],
      ['td[3]/text()', :origin],
      ['td[4]/text()', :airline],
      ['td[5]/text()', :hall],
      ['td[6]/text()', :status],
    ].each do |xpath, option|
      flight_record[option] = row.at_xpath(xpath).to_s.strip
    end

    flight_record[:date] = date
  end
end

puts flight_records.group_by { |record| record[:date] }