使用Ruby和Nokogiri模拟爬虫导出RSS种子的实例详解

# encoding: utf-8
require 'thread'
require 'nokogiri'
require 'open-uri'
require 'rss/maker'
 
$result=Queue.new
def extract_readme_header(no,name,url)
  frame = Nokogiri::HTML(open(url))
  return unless frame
  readme=$url+frame.css('frame')[1]['src']
  return unless readme
  open(readme) do |f|
    doc = Nokogiri::HTML(f.read)
    text=doc.css(div#content div#filecontents p)[0..4].map { |c| c.content }.join( ).strip
    return if text.length==0
    if text !~ /(rails)|(activ_)/i
      puts ========= #{no} #{name} : #{text[0..50]}
      date = f.last_modified
      $result << [no,name,readme,date,text]
    end
  end
rescue
  puts $!.to_s
end
 
def make_rss(items)
  RSS::Maker.make(2.0) do |m|
    m.channel.title = GtitHub recently updated projects
    m.channel.link = http://localhost
    m.channel.description = GitHub recently updated projects
    m.items.do_sort = true
    items.each do |no,name,url,date,descr|
      i = m.items.new_item
      i.title = name
      i.link = url
      i.description=descr
      i.date = date
    end
  end
end
 
############################## M A I N ########################
 
############# Scan list of recent project
 
lth=[]
$url=http://rdoc.info
puts get url #{$url}...
doc = Nokogiri::HTML(open($url))
doc.css('ul.libraries')[1].css('li').each_with_index do |li,i|
  aname =li.css('a').first
  name=aname.content
  purl=$url+aname['href']
  lth << Thread.new(i,name,purl) { |j,n,u| extract_readme_header(j,n,u)  }
end
 
################ wait all readme are read
 
lth.each { |th| th.join() }
 
################ dequeue results and sort them by date descending
 
result=[]
result << $result.shift while $result.size>0
result.sort!  { |a,b| a[0] <=> b[0] }
 
 
################ format results in rss
 
File.open(RubyFeeds.rss,w) do |file|
  file.write make_rss(result)
end