ruby爬虫

一个快速得到回家火车票的方法

Update(2010/02/07): 火车票已经买到，不过不是通过这种方式搞到的，是自己排队加托人才搞到的，票贩子太猖獗了。

就是抓抓抓抓抓抓抓抓抓抓抓抓抓抓抓抓……..

require 'rubygems'
require 'hpricot'
require 'open-uri'
require 'active_support'
require 'action_mailer'
$KCODE = 'u' 
require 'jcode'
require 'tlsmail' #need install tlsmail gem to support TLS connect
Net::SMTP.enable_tls(OpenSSL::SSL::VERIFY_NONE)

ActionMailer::Base.smtp_settings = { 
  :address => 'smtp.gmail.com',     #default: localhost
  :port => '25',                    #default: 25
  :user_name => 'xxxx@gmail.com',#login name
  :password => 'xxxx',        #login password
  :authentication => :login         #:plain, :login or :cram_md5
}
class SimpleMailer < ActionMailer::Base
  def simple_message(recipient, mail_subject, mail_body)
    from 'weekface@gmail.com'
    recipients recipient
    subject mail_subject
    body mail_body
  end 
end

url = "http://shenghuo.google.cn/shenghuo/search?a_y0=9&a_n0=%E7%81%AB%E8%BD%A6%E7%A5%A8&view=Table&a_n1=%E5%A7%8B%E5%8F%91%E7%AB%99&a_y1=1&a_o1=0&a_v1=%E5%8C%97%E4%BA%AC&a_n2=%E5%88%B0%E8%BE%BE%E7%AB%99&a_y2=1&a_o2=0&a_v2=%E9%82%AF%E9%83%B8"
doc = Hpricot(open(url))

arivals = []
doc.search("tr.cssf").each do |ht|
  match = ht.to_s.match(/(\d)张 发车日期:2010-02-(\d\d)/m)
  match_time = ht.to_s.match(/<td>\s?(\d+)分钟前/)
  if match_time
    if match && match[1].to_i > 1 &&
        (match[2].to_i == 11 || match[2].to_i == 12) &&
        match_time[1].to_i <= 60
      a = ht.search("td>a")
      SimpleMailer.deliver_simple_message 'weekface@gmail.com', a.first.inner_html, a.first.attributes['href']
    end
  end
end