TWEG

A Twitter Steganography built with Golang and based on the great work of Steg Of The Dump. Not ready for production.

How it works

Encoding

Given this :

• Tweet : A koala arrives in the great forest of Wumpalumpa
• Hidden message : alpaga
• Alphabet : [space]abcdefghijklmnopqrstuvwxyz123456789'0.:/\\%-_?&
• Alphabet Bit Length : 6

Alphabet Bit Length is 6 because the length of the alphabet is 48, which is 110000 in binary, its length is 6.

Step 1 : generate binary representation of the hidden message

First letter is a, its position in the alphabet is 1. So its binary reprsentation is 000001.
Second letter is l, its position in the alphabet is 12. So its binary reprsentation is 001100.
Third letter is p, its position in the alphabet is 16. So its binary reprsentation is 010000.
Forth letter is a, its position in the alphabet is 1. So its binary reprsentation is 000001.
Fifth letter is g, its position in the alphabet is 7. So its binary reprsentation is 000111.
Sixth letter is a, its position in the alphabet is 1. So its binary reprsentation is 000001.
A space is added (I do not know why..), its position in the alphabet is 0. So its binary reprsentation is 000000.
Binary of hidden message is 000001001100010000000001000111000001000000.

Then, it needs to ensure that the binary is divisible by the alphabet bitlength, which is 6. To do so, it adds as many 0 as needed at the end.

Binary of hidden message is finally 0000010011000100000000010001110000010000000000.

Step 2 : hide the message in the tweet

First character

First character of the tweet is A, it has 3 homoglyphs, that means 4 possibilities. The binary of 4 is 100, which has a length of 3. It is substracting 1 so we get 2. That means we can use 2 bit to represent the 4 possibilities of the character A.

So, we can get the first 2 bit of the binary representation of the hidden message. Which are 00. This is 0 in decimal. That means that this character is left intact.

Final tweet starts with A. The first 2 bit of the binary representation of the hidden message are removed.

Second character

Second character of the tweet is (space), it has 15 homoglyphs, that means 16 possibilities. The binary of 16 is 10000, which has a length of 5. It is substracting 1 so we get 4. That means we can use 4 bit to represent the 16 possibilities of the character .

So, we can get the next 4 bit of the binary representation of the hidden message. Which are 0001. This is 1 in decimal. It is subtracting 1 and searching the hexadecimal code of the homoglyph at this position, the hexadecimal code of the homoglyph of at position 2 is 2000. So the character is U+2000, which is a homoplygh of a space.

Final tweet is now A . The first 4 bit of the binary representation of the hidden message are removed.

Again and again

Repeat the step.
The final result is A kｏａla arrivｅs іn the great forest of Wumpalumpa.

Decoding

Encoded message : A kｏａla arrivｅs іn the great forest of Wumpalumpa

Step 0 : homoglyphs lookup table

In order to decode the message, a lookup dictionnary is built, like this :

{
  "A": "00",
  "B": "00",
  "C": "00",
  "D": "0",
  "a": "0",
  "b": "0",
  "c": "00",
  "d": "0"
}

Step 1 : generate binary representation

First character is A, its lookup is 00.
Second character is  , its lookup is 0001.
Third character is k, its lookup is 0.
Forth character is ｏ, its lookup is 01.
Fifth character is ａ, its lookup is 1.
Sixth character is l, its lookup is 0.
Etc..

The binary representation is : 000001001100010000000001000111000001000000.

Then, it needs to ensure that the binary is divisible by the alphabet bitlength, which is 6. To do so, it adds as many 0 as needed at the end.

The binary representation is finally 0000010011000100000000010001110000010000000000.

Step 2 : decode the hidden message

The binary representation is cut into blocs of the length of the alphabet bitlength, which is 6. Then these blocs are converted to characters. Which means :

• 000001 : a
• 001100 : l
• 010000 : p
• 000001 : a
• 000111 : g
• 000001 : a

The secret message is alpaga.

How to use it

The library can be used like this.

Disclaimer : It is not production ready !

import "github.com/fallais/tweg"

t := tweg.NewTweg()
result, err := t.Encode("xxxxxxxxxxx", "xxxxxxxxxxxx")
if err != nil {
  logrus.Errorln(err)
  return
}
logrus.Infoln("Result is :", result)