This project was a quick experiment to develop a console application which would encrypt or decrypt the name of all files and directories recursively starting from the specified path.
It had three versions, addressing various issues I came across.
The very first version encrypted the file names with AES and then stored it with Base64 encoding. While theoretically it fulfilled its purpose, if you randomly found the folder, you would've known something's up when you're greeted with a bunch of garbage file names:
. . ├───documents ├───eWVhaCwgc3VyZQ │ ├───taxes_2006.pdf --> │ ├───d2hlbiBpIHdhcyAxMw │ └───taxes_2007.pdf │ └───aSB3YXMgaGlkaW5n └───something else └───bXkgdGF4ZXMsIGZvciBzdXJlIQ
In order to combat this, I've decided to go down on another route, and use wordlists to generate the file names. While the algorithm wasn't as efficient as it could be looking at the code now, it was able to take the file name, use modulo arithmetic to convert the base 26 name to base size of wordlist, it then arranged it into various predefined patterns.
.vimrc would turn into
data_something_from_wordlist_0215.bin, which looked less shady. This, however, had the issue of overinflating the file names in some cases, hitting the upper limit of 255 characters per file name.
After I was done with the renaming, even though I encrypted the extension as well, I found that most applications would still open the files. While it was clear from the beginning that I'm only concentrating on the file names, avoiding encrypting the whole content, as it would be a time-consuming implementation for the use-case when you quickly need to access randomly hidden files.
With some research, I was able to find out it was enough to overwrite the first few bytes in some of the formats. For example, the starting
The last version of the application, besides renaming the files, it opened them up, and applied against their first four bytes.