gistfile1.txt

const fs = require('fs');

const cheerio = require('cheerio');
const request = require('request-promise-native');

function sanitizeName (name) {
  name = name.replace(/\s+\(Disc \d+\)/i, '');
  name = name.replace(' (Europe)', '');
  name = name.replace(' (Japan)', '');
  name = name.replace(' (NTSC)', '');
  name = name.replace(' (PAL)', '');
  name = name.replace(' (PAL, NTSC)', '');
  return name;
}

async function getRedumpList () {
  var games = {};
  var page = 1;

  while (1) {
    var url = 'http://redump.org/discs/system/dc/?page=' + page;
    var html = await request(url);
    var $ = cheerio.load(html);

    var rows = $('table.games tbody tr[class!="th"]');

    if (!rows.length) {
      break;
    }

    rows.each(function (i, el) {
      var columns = $(el).children('td');

      var name = sanitizeName(columns.eq(1).children('a').contents()[0].data);
      var version = columns.eq(3).text();
      var serials = columns.eq(6).text().split(/[,|]/).map(function (x) {
        return x.trim();
      });

      if (!games[name]) {
        games[name] = [];
      }

      for (var serial of serials) {
        games[name].push({
          version: version,
          serial: serial
        });
      }
    });

    page++;
  }

  return games;
}

(async function () {
  var redump = await getRedumpList();
  console.log(redump);
}())