Get Youtube download URLs

index.js

var request     = require('./request')
var vm          = require('vm')
var parseString = require('xml2js').parseString

var url = 'https://www.youtube.com/watch?v=si81bIoZRJQ'
var html = ''

request(url).then(function(body) {
  html = body
  var player = body.match(/html5player-([\w\d\-]+)\\\/html5player(.*?).js/g)
  var playerUrl = 'https://s.ytimg.com/yts/jsbin/' + player[0].replace('\\', '')
  return request(playerUrl)
}).then(function(player) {

  // get the function name
  var name = /\.sig\|\|([a-zA-Z0-9$]+)\(/g.exec(player)[1]
  var regex = 'function ' + name + '\(.+){.*?};'
  var functionContent = new RegExp(regex, 'g').exec(player)[0]

  var helperObj = /;([A-Za-z0-9]+?)\./.exec(functionContent)[1]

  var helperRegex = 'var ' + helperObj + '={.*?};'
  var helperFunction = new RegExp(helperRegex, 'g').exec(player)[0]

  var dashmpdUrl = /\"dashmpd\":\"(.*?)\"/g.exec(html)[1]
  var signature = /\\\/s\\\/([a-fA-F0-9\.]+)/g.exec(dashmpdUrl)[1]
  signature = signature ? signature : ''
  
  var fn = helperFunction + functionContent
          + 'var signature = ' + name
          + '(\'' + signature + '\')'

  // eval is evil, so we use a sandbox
  var sandbox = { signature: signature }
  vm.runInNewContext(fn, sandbox)

  dashmpdUrl = dashmpdUrl
    .replace(/\/s\\\/([a-fA-F0-9\.]+)/, '\/signature\\\/' + sandbox.signature)
    .replace(/\\/g, '')

  return request(dashmpdUrl)
}).then(function(xml) {
  parseString(xml, function (err, result) {
    var mimeTypes = result.MPD.Period[0].AdaptationSet
    var formatList = []

    mimeTypes.forEach(function(mimeType) {
      mimeType.Representation.forEach(function(format) {
        formatList.push({
          mimeType: mimeType.$.mimeType,
          itag: format.$.id,
          codecs: format.$.codecs,
          audioSamplingRate: format.$.audioSamplingRate || null,
          url: format.BaseURL[0]._,
          clen: format.BaseURL[0].$['yt:contentLength'],
          width: format.$.width || null,
          height: format.$.height || null,
        })
      })
    })

    console.log(formatList)
  })
})

package.json

{
  "main": "index.js",
  "dependencies": {
    "bluebird": "~2.10.1",
    "xml2js": "~0.4.12"
  }
}

request.js

var http    = require('http')
var https   = require('https')
var parse   = require('url').parse
var Promise = require('bluebird')

var httpLibs = { 'http:': http, 'https:': https }

module.exports = function(url) {
  var parsed = parse(url)
  var httpLib = httpLibs[parsed.protocol]
  return new Promise(function(resolve, reject) {

    if (!httpLib) {
      var err = new Error('Invalid URL: ' + url)
      reject(err)
      return
    }

    var req = httpLib.get(parsed)
    req.on('response', function(res) {
      if (res.statusCode !== 200) {
        reject(new Error('status code ' + res.statusCode));
        return
      }

      var body = ''
      res.setEncoding('utf8')
      res.on('data', function(chunk) {
        body += chunk
      })
      res.on('end', function() {
        resolve(body)
      })
    })
    req.on('error', reject)
  })
}