Parse Hacker News front page.

hn_parsing.rb

  class Update < R '/update'
    def get

      articles = []
      doc = Hpricot(open('http://news.ycombinator.com/'))
      (doc/'.subtext/..').each do |subtext|
        article = subtext.previous_node
        articles << {
          :rank => 
            article.at('.title').inner_html.strip,
          :title => 
            article.at('.title/a').inner_html.strip,
          :link => 
            article.at('.title/a')[:href],
          :comments => 
            Article.normalize_url(
              subtext.at('a:last')[:href]),
          :comment_count => 
            subtext.at('a:last').inner_html[/\d+/].to_i,
          :author => 
            subtext.at('a').inner_html.strip,
          :points => 
            subtext.at('span').inner_html[/\d+/].to_i
        }
      end

      if TOP_COMMENT
        # [snip]
      end
 
      # GET Embedly Pro Preview metadata
      urls = articles.collect do |a| 
        Article.normalize_url(a[:link])
      end.reject do |a| 
          Preview.key_exists? a
      end

      if urls.size > 0
        api = ::Embedly::API.new :key => EMBEDLY_KEY
        api.preview(:urls => urls, :maxwidth => 200).
            each_with_index do |preview, i|
          Preview.save_preview urls[i], preview
        end
      end

      Article.delete_all
      Article.create articles

      redirect R(Index)
    end
  end