Recover files by extension from a directory

recover-files.js

var fs      = require('fs');
var path    = require('path');
var mkdirp  = require('mkdirp');
var rimraf  = require('rimraf');
var through = require('through2');


var SOURCE_PATH   = '/media/roberthurst/CE543276543260FF';
var DEST_PATH     = '/media/roberthurst/478FC40638130E7A/recovered-files';
var MANIFEST_FILE = '/media/roberthurst/478FC40638130E7A/recovered-files.txt';
var DOCUMENT_EXT  = ['doc', 'docx', 'wpd', 'wps', 'csv', 'pps', 'ppt', 'xlr', 'xls', 'xlsx', 'indd', 'pdf', 'accdb'];
var IMAGE_EXT     = ['bmp', 'gif', 'jpg', 'png', 'psd', 'tif'];
var AUDIO_EXT     = ['aif', 'iff', 'm3u', 'm4a', 'mid', 'mp3', 'mpa', 'ra', 'wav', 'wma', 'ai', 'drw', 'eps', 'ps', 'svg'];
var VIDEO_EXT     = ['3g2', '3gp', 'asf', 'asx', 'avi', 'flv', 'mov', 'mp4', 'mpg', 'rm', 'swf', 'vob', 'wmv'];
var SPECIAL_EXT   = [];

// gather the extensions whitelist
var extensions = [].concat(DOCUMENT_EXT, IMAGE_EXT, AUDIO_EXT, VIDEO_EXT, SPECIAL_EXT);
extensions = extensions.filter(function(extension) {
  return extensions.indexOf(extension) === extensions.lastIndexOf(extension);
});

// create the write stream
var pathStream    = through();
var newLineStream = through(function(chunk, _, cb) { cb(null, chunk + '\n'); });
var writeStream   = fs.createWriteStream(MANIFEST_FILE);

// link the streams
pathStream.pipe(newLineStream).pipe(writeStream);

// create the working path pointers
var workingPathIndexPointers = [];
var workingPathIndexes       = [];

// load the target path to kick things off
fs.readdir(SOURCE_PATH, function(err, index) {
  if (err) { throw err; }

  // add the base path data to the working path data
  workingPathIndexPointers.unshift(0);
  workingPathIndexes.unshift(index.map(function(filename) {
    return path.join(SOURCE_PATH, filename);
  }));

  // clear the destination path
  rimraf(path.join(DEST_PATH, '*'), function(err) {
    if (err) { throw err; }

    // begin searching the target
    (function rec() {
      var workingPathIndexPointer = workingPathIndexPointers[0];
      var workingPathIndex        = workingPathIndexes[0];
      var workingPath             = workingPathIndex[workingPathIndexPointer];

      // advance the pointer
      if (workingPathIndexPointer + 1 < workingPathIndex.length) {
        workingPathIndexPointers[0] += 1;
      } else {
        workingPathIndexPointers.shift();
        workingPathIndexes.shift();

        if (workingPathIndexes.length === 0) {
          return writeStream.end();
        }

        return rec();
      }

      // check the file type
      fs.lstat(workingPath, function(err, stat) {
        if (err) { throw err; }

        // if a file then push it into the stream
        if (stat.isFile() && extensions.indexOf(path.extname(workingPath).slice(1)) !== -1) {
          var filePath  = workingPath.slice(SOURCE_PATH.length);
          var writePath = path.join(DEST_PATH, filePath);

          return mkdirp(path.dirname(writePath), function(err) {
            if (err) { throw err; }

            var readStream  = fs.createReadStream(workingPath);
            var writeStream = fs.createWriteStream(path.join(DEST_PATH, filePath));

            pathStream.write(filePath);
            readStream.pipe(writeStream).on('finish', function() {
              console.log(filePath);
              rec();
            });
          });
        }

        // if a directory then dive into it
        if (stat.isDirectory()) {
          return fs.readdir(workingPath, function(err, index) {
            if (err) { throw err; }

            workingPathIndexPointers.unshift(0);
            workingPathIndexes.unshift(index.map(function(filename) {
              return path.join(workingPath, filename);
            }));

            rec();
          });
        }

        rec();
      });
    })();
  });
});