test_langid.js

var walk = require('walk')
var fs = require('fs')
var detectLang = require('lang-detector')
var supported = [
  'javascript', 'c', 'c++', 'python', 'java', 'html', 'css', 'ruby', 'go', 'php'
]
var ext2Lang = {
  'csharp': 'C#', 'gcc': 'C', 'gpp': 'C++', 'ghc': 'Haskell', 'jruby': 'Ruby',
  'python3': 'Python', 'hack': 'PHP', 'yarv': 'Ruby', 'C-sharp': 'C#'
}
var count = 0.0
var correct = 0

// Or path to benchmark data (https://github.com/nbraud/benchmarksgame/tree/master/bench)
var walker = walk.walk('./test', { followLinks: false })

walker.on('file', function (root, stat, next) {
  var path = root + '/' + stat.name
  var lang = path.split('.').slice(-1)[0]
  if (lang in ext2Lang) {
    lang = ext2Lang[lang]
  }
  if (supported.indexOf(lang.toLowerCase()) > 0) {
    fs.readFile(path, {
      encoding: 'utf8'
    }, function (err, code) {
      if (err) {
        throw err
      }
      count += 1
      var out = detectLang(code)
      if (out.toLowerCase() === lang.toLowerCase()) {
        correct += 1
      } else {
        console.log(out, lang, path)
      }
    })
  }
  next()
})

walker.on('end', function () {
  console.log(count, correct)
})