#!/usr/bin/env ruby #You might want to change this ENV["RAILS_ENV"] ||= "production" require File.dirname(__FILE__) + "/../../config/environment" $running = true; Signal.trap("TERM") do $running = false end ### require "rubygems" require "scrapi" #### France 2 #### fr2_extrait = Scraper.define do process "a.etxtnoir11", :titre=>:text, :texte=>"@onmouseover" process "a.etxtgris10", :debut=>:text result :titre, :texte, :debut end fr2 = Scraper.define do array :extraits process "table tr", :extraits=>fr2_extrait result :extraits end fr2_extraits = fr2.scrape(URI.parse("http://jt.france2.fr/20h/html_generated/iframe_0.php?start=0")) fr2_extraits.each do |extr| ActiveRecord::Base.logger << extr.debut.slice(0,8).gsub(/[^\d]/,':') ActiveRecord::Base.logger << "\n" ActiveRecord::Base.logger << CGI::unescapeHTML(extr.titre) ActiveRecord::Base.logger << "\n" ActiveRecord::Base.logger << CGI::unescapeHTML(extr.texte).slice(15..-3).gsub("\\'","'").gsub("\\n","\n") ActiveRecord::Base.logger << "\n" end ################## ActiveRecord::Base.logger << "\nDONE\n"