I got sick of re-writing the crawl login again and again, and I don't like using the attribute CSS selectors [attr=value]

So I wrote this.


require 'rubygems'
require 'open-uri'
require 'net/http'
require 'hpricot'

class Rule
  attr_accessor :tag, :callback
  def initialize(tag, &callback)
    @tag = tag
    @callback = callback
    @preconditions = []
  end
  
  def all_ok?(tag)
    @preconditions.each do |condition|
      return false unless condition.call(tag)
    end  
    true
  end

  def with(condition, &callback)
    @preconditions << condition
    self.callback = callback   
    self
  end
  
  def with(attribute, value, &callback)
    @preconditions << lambda {|tag| tag.attributes[attribute.to_s].include?(value)}
    self.callback = callback   
    self
  end
end

class Farser
  def initialize(source)
    @source = Hpricot(open(source))
    @rules = []
  end
  
  def find(tag, &callback)
    rule = Rule.new(tag, &callback)
    @rules << rule 
    rule   
  end
  
  def farse
    @rules.each do |rule|
      @source.search(rule.tag) do |found|
        next unless rule.all_ok?(found)
        rule.callback.call(found)
      end
    end    
  end  
  
  def self.fetch(from, to)
    uri = URI.parse(from)
    return if File.exists?(to)
    File.open(to, "wb") do |f|
      Net::HTTP.start(uri.host) { |http|
        resp = http.get(uri.path)
        f.write(resp.body)
      }
    end
  end
end


The usage is pretty simple


farser = Farser.new("http://something.com")
farser.find("a").with(:href, 'index').with(:name, 'someone') do |index_link| 
end


No error handling and super duper method chaining.

I love it!