runner.js

var system = require('system'), fs=require('fs'), webpage=require('webpage');
var base_url = "http://www2.menu.com.do/#!/";
var base_output = "snapshots/";
var content = fs.read('businesses.json');
var content_obj = JSON.parse(content);
var businesses = content_obj.result;
var pages = [
  { url: '',                  output: 'index.html' },
  { url: 'intersticial',      output: 'intersticial.html' },
  { url: 's/a:*',             output: 's/a:_.html' },
  { url: 's/sw:0',            output: 's/sw:0.html' },
  { url: 's/sw:a',            output: 's/sw:a.html' },
  { url: 's/sw:b',            output: 's/sw:b.html' },
  { url: 's/sw:c',            output: 's/sw:c.html' },
  { url: 's/sw:d',            output: 's/sw:d.html' },
  { url: 's/sw:e',            output: 's/sw:e.html' },
  { url: 's/sw:f',            output: 's/sw:f.html' },
  { url: 's/sw:g',            output: 's/sw:g.html' },
  { url: 's/sw:h',            output: 's/sw:h.html' },
  { url: 's/sw:i',            output: 's/sw:i.html' },
  { url: 's/sw:j',            output: 's/sw:j.html' },
  { url: 's/sw:k',            output: 's/sw:k.html' },
  { url: 's/sw:l',            output: 's/sw:l.html' },
  { url: 's/sw:m',            output: 's/sw:m.html' },
  { url: 's/sw:n',            output: 's/sw:n.html' },
  { url: 's/sw:o',            output: 's/sw:o.html' },
  { url: 's/sw:p',            output: 's/sw:p.html' },
  { url: 's/sw:q',            output: 's/sw:q.html' },
  { url: 's/sw:r',            output: 's/sw:r.html' },
  { url: 's/sw:s',            output: 's/sw:s.html' },
  { url: 's/sw:t',            output: 's/sw:t.html' },
  { url: 's/sw:u',            output: 's/sw:u.html' },
  { url: 's/sw:v',            output: 's/sw:v.html' },
  { url: 's/sw:w',            output: 's/sw:w.html' },
  { url: 's/sw:x',            output: 's/sw:x.html' },
  { url: 's/sw:y',            output: 's/sw:y.html' },
  { url: 's/sw:z',            output: 's/sw:z.html' },
  { url: 'tipos-de-comida',   output: 'tipos-de-comida.html' },
  { url: 'menus-del-dia/',    output: 'menus-del-dia.html' },
  { url: 'actividades/',      output: 'actividades.html' },
  { url: 'ofertas/',          output: 'ofertas.html' },
  { url: 's/c:catering',      output: 's/c:catering.html' },
  { url: 's/c:panaderia,reposteria', output: 's/c:panaderia,reposteria.html' },
  { url: 'terminos',          output: 'terminos.html' },
  { url: 'privacidad',        output: 'privacidad.html' },
  { url: 'contacto',          output: 'contacto.html' },
];

function handle_page(url, output){
  url = base_url + url;
  output = base_output + output;
  console.log('Fetching ' + url);
  var page = webpage.create();
  page.onResourceRequested = function(requestData, request) {
    switch (requestData['Content-Type']) {
        case 'text/css':
        case 'image/jpeg':
        case 'image/png':
          request.abort();
          break;
        case 'text/html':
          if ((/afr\.php/gi).test(requestData.url)) {
            request.abort();
          }
        break;
    }

  };
  page.open(url, function (status) {
    if (status != 'success') {
      console.log('Failed. Status:' + status);
      setTimeout(next_page, 0);
    } else {
      console.log('Success.');
      console.log('Waiting for page to be ready...');
      var delay, checker = (function() {
        var html = page.evaluate(function () {
          var body = document.getElementsByTagName('body')[0];
          if(body.getAttribute('data-status') == 'ready') {
            console.log('Ready.');
            window.$("script[async]").remove();
            var scripts=window.$("body script");
            for (var i=0,len=scripts.length;i<len;i++) {
              if (scripts[i] && scripts[i].src && (/map/gi).test(scripts[i].src)) {
                window.$(scripts[i]).remove();
              }
            }
            return document.getElementsByTagName('html')[0].outerHTML;
          }
        });
        if(html) {
          clearTimeout(delay);
          console.log('Writing to ' + output);
          fs.write(output, html, 'w');
          console.log('Done.');
          setTimeout(next_page, 0);
        }
      });
      delay = setInterval(checker, 100);
      }
  });
}

function next_business() {
  var business = businesses.shift();
  var pages_len = pages.length;
  if (!business) {
    console.log("Done.");
    console.log("Pages: " + pages_len);
    setTimeout(next_page, 0);
    return;
  }

  console.log("Business: " + business.Business.name + " (" + business.Business.vanity_name + ")");
  if (business.Business.is_hidden || business.Business.deleted) {

    console.log('+' + (pages.length - pages_len) + ' pages');
    setTimeout(next_business, 0);
    return;
  }
  // Profile
  pages.push({
    url: 'n/' + business.Business.vanity_name + '/',
    output: 'n/' + business.Business.vanity_name + '.html',
  });
  // Menu
  if (!business.Business.menu_hidden) {
    pages.push({
      url: 'n/' + business.Business.vanity_name + '/menu',
      output: 'n/' + business.Business.vanity_name + '/menu.html',
    });
  }
  //
  if (!business.Business.venues_hidden) {
    pages.push({
      url: 'n/' + business.Business.vanity_name + '/mapa',
      output: 'n/' + business.Business.vanity_name + '/mapa.html',
    });
  }
  if (!business.Business.offers_hidden) {
    pages.push({
      url: 'n/' + business.Business.vanity_name + '/ofertas/',
      output: 'n/' + business.Business.vanity_name + '/ofertas.html',
    });
  }
  if (!business.Business.activities_hidden) {
    pages.push({
      url: 'n/' + business.Business.vanity_name + '/actividades/',
      output: 'n/' + business.Business.vanity_name + '/actividades.html',
    });
  }
  if (!business.Business.comments_hidden) {
    pages.push({
      url: 'n/' + business.Business.vanity_name + '/comentarios',
      output: 'n/' + business.Business.vanity_name + '/comentarios.html',
    });
  }
  console.log('+' + (pages.length - pages_len) + ' pages');
  setTimeout(next_business, 0);
}

next_business();

function next_page() {
  var current_page = pages.shift();

  if (!current_page) {
    console.log('Done.');
    phantom.exit();
  } else {
    handle_page(current_page.url, current_page.output);
  }

}