gistfile1.rb

# RSS采集并自动发帖到某网站

require 'mechanize'
require 'simple-rss'

puts "loading exists guid list..."
# 读取已存在的guids
old_guids = []
guid_file = File.open('guids.dat',"a+")
guid_file.each_line { |line| old_guids << line.gsub("\n",'') }


# 登陆到 网站
puts "login to website..."
agent = WWW::Mechanize.new

page = agent.get("http://renjian.com/dd/index")
form = page.forms.first
form.username = 'user'
form.password = '123123'
# 登陆，并进入首页
page_home = form.submit

# 取得最新的Feed
feed_urls = [
  'http://github.com/feeds/mangos/commits/mangos/master',
  'http://www.javaeye.com/rss/ask',
  'http://www.javaeye.com/rss/forum',
  'http://news.163.com/special/00011K6L/rss_newstop.xml',
  'http://news.163.com/special/00011K6L/rss_gn.xml',
  'http://news.163.com/special/00011K6L/rss_gj.xml',
  'http://news.163.com/special/00011K6L/rss_sh.xml',
  'http://news.163.com/special/00011K6L/rss_war.xml',
  'http://news.163.com/special/00011K6L/historyrss.xml',
  'http://sports.163.com/special/00051K7F/rss_sportslq.xml',
  'http://sports.163.com/special/00051K7F/rss_sportsyc.xml',
  'http://sports.163.com/special/00051K7F/rss_sportsyj.xml',
  'http://news.cnblogs.com/rss',
  'http://www.cnblogs.com/rss',
  'http://www.cnbeta.com/backend.php',
  'http://feeds.digg.com/digg/container/technology/popular.rss',
  'http://feeds.digg.com/digg/topic/programming/popular.rss',
  'http://rss.sina.com.cn/news/marquee/ddt.xml'
]

feed_urls.each do |url|
  puts "read from #{url}"
  begin
    feed = SimpleRSS::parse open(url)
  rescue
    feed = nil
  end
  if feed != nil
    feeds = []
    
    feed.items.each do |item|     
      # 判断以前有没有发过       
      if old_guids.index(item.guid) == nil
        feeds << {  
          "guid" => item.guid,
          "title" => item.title,
          "link" => item.link,
          "summary" => item.content == nil ? item.summary : item.content,
          "author" => item.author,
        }
      end
    end
    
    if feeds.length > 0 
      puts "save to exist guid list..."
      new_guids = ''
      feeds.each { |feed| new_guids += "#{feed['guid']}\n"}
      guid_file.puts new_guids
    
      puts "There have #{feeds.length} feeds will be send."
        
    
      # 分享表单
      form_share = page_home.forms[1]
      feeds.each do |feed|
        form_share.link_type = "LINK"
        author = feed['author'] == '' ? "" : "#{feed['author']}:"
        form_share.text = "#{author}#{feed['title']}"
        form_share.link_url = feed['link']
        form_share.original_url = feed['link']
        form_share.link_des = feed['summary']
        begin
          form_share.submit        
          puts "#{feed['title']} submited."
        rescue
          puts "errored."
        end
      end   
    end
  end
end