I had a crashed hard disk and was forced to recover data because many photos and movies were contained without any backup.

My steps to recover my data:

• Having no longer any access to hard drive I thought the PCB controller could be the problem

• So I bought a second hand hard disk having same hardware specification as my crashed device

• Replaced defect PCB controller

• Booted with a LIVE Linux CD …​

• Was able to access my hard drive again - but with errors and no file system

• Started a full dump / restore of my hard disk with ddrescue:
  ddrescue /dev/sdb ./full-sdb-backup.img ./full-sdb-rescue.log

• After some time even the "new" PCB controller crashed, but at least an image was available with some data.

• I used photorec, which is part of testdisk suite, to restore photos from former created image.
  (Simply call photorec on command line)

I was able to recover a big amount of data - not all, but most data. Thanx a lot to all people involved developing formerly mentioned tools!

Unfortunetely the resulting file structure contains all recovered files in a random structure and also random file names. Only the modification time stamp was recovered as well:

I was only interested in some specific files - mostly in JPEG. Also sorting all these files manually (>60.000 files) was no option. So I wrote this software.

To keep this simple this software is MIT license.

Just execute ./gradlew fatJar

This will build pdb.jar inside build/libs which contains all dependencies.

java -jar pdb.jar -s ${photoRecOutputDir} -t ${pdbTargetDir}

It simply tries to get origin creation time from files (JPEG meta information). If not possible the "last modified" file time stamp is used as a fallback. It also makes top level sub directories which have upppercased filending name, so garbage can be easily destroyed by a folder delete.

File names are in format yyyy-MM-dd_HH-mm-ss with origin file name ending.

Here an example output: