Designed to recover files from partly unreadable disks, diskettes or other media.

The idea is based on two types of errors (input/output error and path/file access error). Routine reads file by 10kb blocks and when hardware involves errors, the size of block decrease tenfold and so on until the moment when block of data will be read without any errors, or it reach minimum size of block, then program skips one byte and starts to read second one. After successfully reading data, block size increase tenfold again and again to reach maximum size - 10kb.