pega as top stories do google news e lista as notícias paralelas aplicando outra cor às palavras já ditas.

parallelnews.rb

#encoding: utf-8
class FrontController < ApplicationController

  def list
    coder    = HTMLEntities.new
    @entries = []
    url      = "http://news.google.com/?output=rss"
    link     = "http://ajax.googleapis.com/ajax/services/feed/load?v=1.0&num=10&q="+url
    result   = ActiveSupport::JSON.decode(open(link).read)
    max    = 0
    count = 0
    # pega as noticias do google news
    result["responseData"]["feed"]["entries"].each do |entry|
      count    +=1
      e        = entry["title"].split(" - ")
      title    = e[0]
      source   = e[1]

      related  = entry["content"].scan(/<b>all\s(.+)\snews/i)[0][0].to_s || nil
      more_url = nil
      # pega a url das related news
      if related
        more     = entry["content"].scan(/(http:\/\/news.google.com\/news\/more(.+)topic=h)/i)
        more_url = more[0][0].gsub("more", "story").gsub(/&amp;/i, "&")
      end

      related_page   = coder.decode(URI.parse(more_url).read)
      # extrai snippets, heads, sources e datas usando expressões regulares
      snippets       = related_page.scan(/<div class="snippet">(.*?)(\s<b>...<\/b>\s)?<\/div>/)
      heads          = related_page.scan(/<span class="titletext">(.*?)(<b>...<\/b>)?<\/span>/)
      sources        = related_page.scan(/<span class="source source-pref .+?">(.*?)<\/span>/)
      dates          = related_page.scan(/<span class="date ">(.*?)<\/span><\/div>/)

      max = (snippets.size>max) ? snippets.size : max
      clean_snippets = []
      count          = 0
      # decodifica html, limpa e junta dados num conjunto de clean_snippets 
      snippets.each do |snippet|
        source = sources[count] ? sources[count][0].to_s : ""
        text  = snippet[0].to_s + (snippet[1] ? " ..." : "")
        clean_snippets << {:text=>text, :title=>heads[count][0].to_s, :source=>source, :date=>dates[count][0].to_s, :count=>count+=1}
      end

      # aplica cor as palavras já ditas no texto do snippet
      lighted_snippets= lowlight(clean_snippets, :text)
      # e nos títulos
      lighted_snippets= lowlight(lighted_snippets, :title)
      @entries << {:max=>(max+3), :snippets=>lighted_snippets}
    end
  end

  def lowlight(clean_snippets, text)
      parallel_snippets = []
      snippets = clean_snippets.dup
      snippets.each do |snippet|
        # quebra texto em conjunto de palavras
        break_text = snippet[text].split(" ")
        # para cada palavra
        break_text.each do |word|          
          # nos outros snippets
          snippets.each do |other|
            # se no outro snippets tiver essa palavra
            if (other[text].split(" ").index(word) && snippet[:count]!=other[:count])
              # aplica esse estilo html: said => cinza.
              word.gsub!(word, '<span class="said">'+word+'</span>')
              break
            end
          end
        end
        # junta as palavras novamente
        snippet[text] = break_text.join(" ")
        parallel_snippets << snippet
      end
      parallel_snippets
  end

end

=> http://parallelnews.net