Contents

module Scrapzirra
  class Scrap
    attr_accessor :url

    def initialize url, options = {}
      @url = url
      @doc = Nokogiri::HTML(options[:doc]) if options[:doc]
    end

    def doc
      @doc || load_doc
    end

    def load_doc
      require 'nokogiri'
      require 'open-uri'

      @doc = Nokogiri::HTML(open(@url).read)
      @doc.encoding = 'utf-8'
      @doc
    rescue
      @doc = Nokogiri::HTML::Document.parse('No data')
    end

    def selector_content css
      doc.css(css).collect{|a| a.inner_html }.inject(:+)
    end
  end

end

Version data entries

2 entries across 2 versions & 1 rubygems

Version	Path
scrapzirra-0.0.6	lib/scrapzirra/scrap.rb
scrapzirra-0.0.5	lib/scrapzirra/scrap.rb