Base64 Encoding

Base64 is encoding scheme and NOT encryption.

I have to keep remind myself.

I am sure, being a programmer, you have come across base64 encoding many time in your career.

Same with me, and I always wanted to know the secret behind it.

When you are in the middle of high delivery project, you don’t get time to research.

Today is the best time, Sunday raining morning. Good excuse to stay indoor infront of machine.

I have too many things on my plate already i.e. finish the manuscript for upcoming book on DBIx::Class.

Also thinking of submitting a talk for London Perl Workshop 2026. I have few topics in my mind.

I will have to choose one, soon otherwise I will miss the train. I want to keep the tradition of giving a talk at every LPW event.

Coming back to the main topic, Base64 Encoding.

Here is the simple introduction, using the tool, base64:

$ echo -n "Super_Secret_Password" | base64
U3VwZXJfU2VjcmV0X1Bhc3N3b3Jk

So, the base64 encoding of text “Super_Secret_Password” is U3VwZXJfU2VjcmV0X1Bhc3N3b3Jk.

How about reverse engineering?

We can use the same tool to decode the encoded string.

$ base64 -d <<< U3VwZXJfU2VjcmV0X1Bhc3N3b3Jk
Super_Secret_Password

It looks simple, isn’t it?

Let’s figure out the magic behind the scene.

For this, we will use tool, xxd. It creates a hexadecimal dump of a given file or standard input.

$ echo -n abc | xxd -b
00000000: 01100001 01100010 01100011                             abc

So we got 3 letters, 8 bits each as below:

a: 0 1 1 0 0 0 0 1
b: 0 1 1 0 0 0 1 0
c: 0 1 1 0 0 0 1 1

First this gets converted into 4 groups, 6 bits each.

Group 1: 0 1 1 0 0 0
Group 2: 0 1 0 1 1 0
Group 3: 0 0 1 0 0 1
Group 4: 1 0 0 0 1 1

Having done that, now we need to lookup this table for encoding.

Binary	Value	Char	Binary	Value	Char	Binary	Value	Char	Binary	Value	Char
000000	0	A	010000	16	Q	100000	32	g	110000	48	w
000001	1	B	010001	17	R	100001	33	h	110001	49	x
000010	2	C	010010	18	S	100010	34	i	110010	50	y
000011	3	D	010011	19	T	100011	35	j	110011	51	z
000100	4	E	010100	20	U	100100	36	k	110100	52	0
000101	5	F	010101	21	V	100101	37	l	110101	53	1
000110	6	G	010110	22	W	100110	38	m	110110	54	2
000111	7	H	010111	23	X	100111	39	n	110111	55	3
001000	8	I	011000	24	Y	101000	40	o	111000	56	4
001001	9	J	011001	25	Z	101001	41	p	111001	57	5
001010	10	K	011010	26	a	101010	42	q	111010	58	6
001011	11	L	011011	27	b	101011	43	r	111011	59	7
001100	12	M	011100	28	c	101100	44	s	111100	60	8
001101	13	N	011101	29	d	101101	45	t	111101	61	9
001110	14	O	011110	30	e	101110	46	u	111110	62	+
001111	15	P	011111	31	f	101111	47	v	111111	63	/

Using the above encoding table, replace the each group with the equivalent alphabet.

Group 1: 0 1 1 0 0 0 => Y
Group 2: 0 1 0 1 1 0 => W
Group 3: 0 0 1 0 0 1 => J
Group 4: 1 0 0 0 1 1 => j

Double check, the above manual encoding with this command:

$ echo -n abc | base64
YWJj

It wasn’t difficult to decode the magic.

Hang on, there is still one gotcha.

Let me show you another example as below:

$ echo -n ab | base64
YWI=

Where did this = appear in the output?

It’s not part of the encoding table above.

This is what I was talking about earlier.

First find the hexadecimal form the text:

$ echo -n ab | xxd -b
00000000: 01100001 01100010                                      ab

Let’s get encode the text manually first.

So far, we know base64 works with 24 bits at a time.

Group 1: _ _ _ _ _ _  => []
Group 2: _ _ _ _ _ _  => []
Group 3: _ _ _ _ _ _  => []
Group 4: _ _ _ _ _ _  => []

So the above dump looks like this now:

a: 0 1 1 0 0 0 0 1
b: 0 1 1 0 0 0 1 0

Total 16 bits and in the group it would look like this with the help of encoding table.

Group 1: 0 1 1 0 0 0 => [Y]
Group 2: 0 1 0 1 1 0 => [W]
Group 3: 0 0 1 0 _ _ => [ ]
Group 4: _ _ _ _ _ _ => [ ]

As we see Group 3 is missing 2 bits and Group 4 is completely empty.

For missing bits, we fill it with 0 and empty group gets special character =.

Group 1: 0 1 1 0 0 0 => [Y]
Group 2: 0 1 0 1 1 0 => [W]
Group 3: 0 0 1 0 0 0 => [I]
Group 4: _ _ _ _ _ _ => [=]

Let’s prove the manual encoding.

$ echo -n ab | base64
YWI=

In the next post, I will share the common usecase of base64 encoding.

Happy Hacking !!!