Protect your files by splitting their souls.
A horcrux is a file that contains a fraction of the information of another file, similar to how horcruxes in Harry Potter contain split pieces of a soul to protect it. It is not possible to tell if a file is a horcrux of another file or not, and the horcrux itself does not contain any information about other horcruxes to the same file or any information about even if either of those things exist. The information that each horcrux holds is essentially random bytes, so none of the data from the original file can be recovered. Even 99 out of 100 horcruxes combined will not yield any information about the original file; all horcruxes must be present to merge to the original file. The data a horcrux holds is only meaningful if all the other pieces are present.
At it's core, encryption and decryption calculations revolve around the use of XOR. When a file is split, securely random bytes are generated in a way such that when they are combined together via the use of XOR it will yield the decrypted file.
For example, assume the file passwords.txt is going to be split into 1.hcx,
2.hcx, and 3.hcx. For each byte in passwords.txt, two random bytes are
generated and placed into 1.hcx and 2.hcx respectively. The byte for 3.hcx
is calculated by XORing all random bytes with the original byte
In pseudocode:
third_byte = original_byte XOR first_rand_byte XOR second_rand_byte
To merge files together, the process is even easier. To recover the original
file from it's 'horcruxes', the program goes through every byte in each horcrux
and XORs them all together. For example, to recover the file passwords.txt
from its horcruxes 1.hcx, 2.hcx, and 3.hcx, the first byte of
passwords.txt would be equal to the first byte of 1.hcx XOR 2.hcx XOR
3.hcx.
In pseudocode:
original_byte = byte_from_1hcx XOR byte_from_2hcx XOR byte_from_3hcx
Since horcruxes are generated with purely random numbers, there is an astronomically low chance that a significantly large portion of the original data will be in one of the split components. To put into perspective how low of a chance this is of occuring, a file with three bytes has a 1 in 16,777,216 chance of all three bytes not being encrypted and a 1 in 65,281 chance of two of the bytes not being encrypted. As more bytes are added to the file the chances of data being leaked from the original exponentially decreases. In order to verify that there isn't a significant amount of information leaked from the original, a comparison algorithm was added. This checks multiple files, and returns a percentage of the similarity between the files.
In pseudocode, the percentage of the similarity is calculated with the following:
percentage = number_of_identical_bytes / total_number_of_bytes
To create a horcrux that is compatible with multiple files, A 'locket' with the same size as the largest input file is generated. After this locket is generated, each input file is individually merged with it. If a file is shorter than the locket, then the extra bytes of it are not used for merging (This is the default behavior for merging in general). The output files will each be horcruxes that can be merged with the locket to recover it's original file. Of course, both the locket and the horcruxes and be split or merged even more.
usage: horcrux [-h] [-L LOCKET] [-o [OUTPUT ...]] [-c CHUNK_SIZE] [-p PARTS] [-n NOISE] {split,merge,locket,compare} [inputs ...]
Protect your files by splitting their souls.
positional arguments:
{split,merge,locket,compare}
Action to perform.
inputs
options:
-h, --help show this help message and exit
-L LOCKET, --locket LOCKET
Filename of locket horcrux.
-o [OUTPUT ...], --output [OUTPUT ...]
Names of output files.
-c CHUNK_SIZE, --chunk-size CHUNK_SIZE
1M Default. How many bytes to load into RAM per file. Use K, M, G for more units or no unit for bytes.
-p PARTS, --parts PARTS
How many parts to split into.
-n NOISE, --noise NOISE
Default: 0. Max amount of extra bytes to generate for horcruxes after splitting. Use K, M, G for more units or no unit for bytes.
Note that on the above usage section, [inputs ...] is listed after all the
command flags. This is due to how the argparse library prints usage. When
using the horcrux command, inputs must be specified before flags.
On windows, the program can be run with the python or python3 command. This
can also be done on linux, or horcrux.py can be placed in a folder in the PATH
variable with execution permissions and used directly as a command.
To split a file into any number of parts, use the following syntax:
horcrux split <fileToSplit> <part1> <part2> [<part3> ...]
The file will be split into however many output files are specified. So if five filenames are specified, then it will be split into five parts. (NOTE: At least two output files are needed to split).
Note that for merging, ALL parts of a file are required for merging. If a file is split into 100 horcruxes and one horcrux is missing, the original data CANNOT be recovered.
The name or file extension of the horcruxes (output files) does not matter, they will appear as corrupted files either way.
To merge horcruxes into their original file, use the following syntax:
horcrux merge <part1> <part2> [<part3> ...] <outputFile>
The horcruxes will be merged into a single file whether or not they will created a decrypted file. If the horcruxes vary in size, the merging algorithm will use the size of the smallest horcrux. Since there's no way to confirm if a file is a horcrux, the merge algorithm can be applied to any list of files.
Different Files can share horcruxes that hold the information to both of them.
For example, the horcrux a.hcx merged with the horcrux k.hcx may yield an
decrypted file, and the completely different horcrux b.hcx merged with the
same horcrux k.hcx may yield a completely different decrypted file.
This can be done by generated a special horcrux that is used in the calculations for splitting multiple differnet files to get their other halves. While this horcrux is able to decrypt multiple things, it still is fundamentally the same and indistinguishable from a regular horcrux.
To split multiple files at once with a shared horcrux, use the following syntax:
horcrux locket <file1> [<file2> ...] --locket <locketFileName>
If you want to specify specific output names, use the following:
horcrux locket <file1> [<file2> ...] --output [output1 output2 ...] --locket <locketFileName>
To Encrypt the files a.txt, b.txt, and c.txt with a locket locket.hcx:
horcrux locket a.txt b.txt c.txt --locket locket.hcx
This will generate the horcruxes: a.txt.hcx, b.txt.hcx, and c.txt.hcx.
The file extension of the horcruxes does not matter. To use specific outupts instead of auto generated ones the above example would be changed into:
horcrux locket a.txt b.txt c.txt --locket locket.hcx --output Ahorcrux.bin Bhorcrux.bin
Note that in the above example an output name for c.txt is not specified, so
the name c.txt.hcx will be used.
To view the similarity between two files, use the following syntax:
horcrux compare <fileToCompare> <file1> [<file2> ...]
The above command will show the percent similarity between fileToCompare and
all the other files. Note that all other files won't be compared with each
other, only fileToCompare and the other files.