Wikipedia API

wikipedia_api.rb

class Wiki
  def self.parse(lang, page)
    begin
      json = Yajl::Parser.parse open("http://#{lang}.wikipedia.org/w/api.php?maxlag=5&redirects=1&format=json&disablepp=1&prop=text%7Cdisplaytitle%7Crevid&action=parse&section=0&page=#{CGI.escape(page)}").read
      doc = Nokogiri::HTML(json['parse']['text']['*'])

      # article message boxes
      doc.css('.ambox').remove
      doc.css('.bandeau').remove
      doc.css('.bandeau-portail').remove
      doc.css('.homonymie').remove

      # error messages
      doc.css('.error').remove

      # infoboxes
      doc.css('.infobox').remove
      doc.css('.infobox_v2').remove
      doc.css('.thumb.tright').remove

      doc.at_css('body').inner_html.to_s.sub(%r{<p><br></p>}, '')
    rescue SocketError
      ''
    end
  end
end