转载请标明出处:http://fuliang.iteye.com/blog/1122008
经常需要从网上抓取一些需要的内容做成语料,供分类使用。所以需要一个灵活的抓取、抽取程序-自己的DSL来做这件事,这样每次只需要写几行代码就能得到需要的内容。比如我比较希望以下几行代码就能把我的博客的内容给抓下来:
crawler = Crawler.new
1.upto(10) do |pn|
urls = []
crawler.fetch "http://fuliang.iteye.com/?page=#{pn}" do |page|
page.css("div.blog_title > h3 > a").each do |node|
urls << "http://fuliang.iteye.com#{node.attributes['href']}"
end
end
urls.each do |url|
crawler.fetch url do |page|
page.xpath(:title => '//*[@id="main"]/div/div[2]/h3/a',:content => '//*[@id="blog_content"]').each do |entry|
printf("%s\t%s\n",entry[:title].text.gsub(/\s+/,""),entry[:content].text.gsub(/\s+/,""))
end
end
end
break
end
我们先创建一个Crawler对象,然后按照我博客的列表页分页特征,得到第pn页的url是
http://fuliang.iteye.com/?page=#{pn},当然有可能有复杂的规则,构建列表页的url列表,然后遍历。crawler只有一个fetch方法,就可以把页面fetch下来,然后得到这个页面在块中处理。这个页面可以直接根据xpath、css来得到需要抽取的内容,还可以一次抽取一个记录,只需要向xpath,css方法中传递一个字段到xpath/css的
hash,然后得到对应的记录的hash。
按照上面的描述,我们先编写一个简单的Crawler,为了防止被封我们使用了几个代理:
class Crawler
def initialize
@proxies = 1.upto(6).collect{|index| "http://l-crwl#{index}:1080"}
end
def fetch(url)
yield Page.new( Nokogiri::HTML(open(url,fetch_options)) )
end
private
def rand_proxy
@proxies[(rand * 6).to_i]
end
def fetch_options
user_agent = "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.2) Gecko/20061201 Firefox/2.0.0.2 (Ubuntu-feisty)"
fetch_options = {
"User-Agent" => user_agent,
"proxy" => rand_proxy
}
end
end
然后我们定义Page类,动态定义了css和xpath的方法,我们直接代理给Nokogiri
的css、xpath,让它来做事情,我们收集一下抽取的结果,一些就ok了:
class Page
def initialize(html)
@html = html
end
class_eval do
[:css,:xpath].each do |extract_by|
define_method extract_by do |arg,&block|
if arg.is_a? String then
if block.nil? then
@html.send(extract_by,arg)
else
block.call(@html.send(extract_by,arg))
end
elsif arg.is_a? Hash then
extract_raw = arg.collect{|key,value| [key, @html.send(extract_by,value)]}
data = extract_raw.collect do |key, vals|
([key] * vals.size).zip(vals)
end
result = data[0].zip(data[1]).collect{|e| Hash[ * e.flatten ]}
if block.nil? then
result
else
block.call(result)
end
else
raise ArgumentError.new('Argument type must String or Hash type')
end
end
end
end
end
整个的代码:
#!/usr/bin/env ruby
require 'rubygems'
require 'nokogiri'
require 'open-uri'
class Crawler
def initialize
@proxies = 1.upto(6).collect{|index| "http://l-crwl#{index}:1080"}
end
def fetch(url)
yield Page.new( Nokogiri::HTML(open(url,fetch_options)) )
end
private
def rand_proxy
@proxies[(rand * 6).to_i]
end
def fetch_options
user_agent = "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.2) Gecko/20061201 Firefox/2.0.0.2 (Ubuntu-feisty)"
fetch_options = {
"User-Agent" => user_agent,
"proxy" => rand_proxy
}
end
end
class Page
def initialize(html)
@html = html
end
class_eval do
[:css,:xpath].each do |extract_by|
define_method extract_by do |arg,&block|
if arg.is_a? String then
if block.nil? then
@html.send(extract_by,arg)
else
block.call(@html.send(extract_by,arg))
end
elsif arg.is_a? Hash then
extract_raw = arg.collect{|key,value| [key, @html.send(extract_by,value)]}
data = extract_raw.collect do |key, vals|
([key] * vals.size).zip(vals)
end
result = data[0].zip(data[1]).collect{|e| Hash[ * e.flatten ]}
if block.nil? then
result
else
block.call(result)
end
else
raise ArgumentError.new('Argument type must String or Hash type')
end
end
end
end
end
crawler = Crawler.new
1.upto(10) do |pn|
urls = []
crawler.fetch "http://fuliang.iteye.com/?page=#{pn}" do |page|
page.css("div.blog_title > h3 > a").each do |node|
urls << "http://fuliang.iteye.com#{node.attributes['href']}"
end
end
urls.each do |url|
crawler.fetch url do |page|
page.xpath(:title => '//*[@id="main"]/div/div[2]/h3/a',:content => '//*[@id="blog_content"]').each do |entry|
printf("%s\t%s\n",entry[:title].text.gsub(/\s+/,""),entry[:content].text.gsub(/\s+/,""))
end
end
end
break
end