Fetches list of top 1000 companies using node, the jsdom module, and coffee-script.

top1000.coffee

jsdom = require 'jsdom'

dataz = []
listings = 1000

http_host = "http://www.stormscape.com"
path = "/inspiration/website-lists/global1000/"

# Build up the list of url's to visit.
urls = for page in [0..9]
  file = "index.php?pageNum_list=#{ page }&totalRows_list=1000&display=names"
  "#{ http_host }#{ path }#{ file }"

for url in urls
  jsdom.env
    html: url
    scripts:
      'https://raw.github.com/ded/qwery/master/qwery.min.js'
    done: (errors, window) ->
      q = window.qwery

      # Two list comprehensions, and a concat... Also, skip the first tr... Its
      # the table header.
      dataz.push (td.textContent for td in (q 'td', tr)) for tr in (q "table.type1 tr")[1..]

      # Have we reached the record requirement?
      if dataz.length is listings
        # Quick little sort on the company ranking.
        dataz.sort (a,b) ->
          +a[0] - +b[0]

        # Put the headers back in.
        console.log "Ranking, Company"
        for row in dataz
          console.log "#{ row[0] }, #{ row[1] }"

top1000.js

(function() {
  var dataz, file, http_host, jsdom, listings, page, path, url, urls, _i, _len;

  jsdom = require('jsdom');

  dataz = [];

  listings = 1000;

  http_host = "http://www.stormscape.com";

  path = "/inspiration/website-lists/global1000/";

  urls = (function() {
    var _results;
    _results = [];
    for (page = 0; page <= 9; page++) {
      file = "index.php?pageNum_list=" + page + "&totalRows_list=1000&display=names";
      _results.push("" + http_host + path + file);
    }
    return _results;
  })();

  for (_i = 0, _len = urls.length; _i < _len; _i++) {
    url = urls[_i];
    jsdom.env({
      html: url,
      scripts: 'https://raw.github.com/ded/qwery/master/qwery.min.js',
      done: function(errors, window) {
        var q, row, td, tr, _j, _k, _len2, _len3, _ref, _results;
        q = window.qwery;
        _ref = (q("table.type1 tr")).slice(1);
        for (_j = 0, _len2 = _ref.length; _j < _len2; _j++) {
          tr = _ref[_j];
          dataz.push((function() {
            var _k, _len3, _ref2, _results;
            _ref2 = q('td', tr);
            _results = [];
            for (_k = 0, _len3 = _ref2.length; _k < _len3; _k++) {
              td = _ref2[_k];
              _results.push(td.textContent);
            }
            return _results;
          })());
        }
        if (dataz.length === listings) {
          dataz.sort(function(a, b) {
            return +a[0] - +b[0];
          });
          console.log("Ranking, Company");
          _results = [];
          for (_k = 0, _len3 = dataz.length; _k < _len3; _k++) {
            row = dataz[_k];
            _results.push(console.log("" + row[0] + ", " + row[1]));
          }
          return _results;
        }
      }
    });
  }

}).call(this);