Break it

dhromed

There's some encryption mechanism X used by my macro tool, but since the cipher's readily undoable to produce the original string, it can't be that difficult. It's certainly no hash. The docs are silent on the method used, but maybe a little crowdsourcing will help to break it.

Some input strings and the resultant output:

lol zort            0732CD5635C756BA5C
smurfigheden    9F52D325D453DD37EC131BEC18
h                    F018
hhh                C870B471
hurk                A751AA719F

Output digits are always in the range [0-F], and each input character is encoded as two digits, plus an additional two.

Anyone up for a little crackin'? Hell, for all I know it's ROT-X, with X being the first hex digits, or the value of the previous two hex digits, or whatever. I'm going to hack away at this later.

PJH

Does the same plaintext always produce the same output?

dhromed

Ah, good question.

(...)

nope, it does not.

I suppose the extra two digits are a random seed for something.

here's "hurk" a few more times:

96BF59A750

BD67B04FF9

658C6E8CB4

C66E8F6E9A

PJH

How many variations are there for a single character, e.g. 'a'. Ditto for 'b'.

dhromed

Here's a ton of consecutive encodings for "a". It's a lot of clickwork. :) Couple of repeats. Given a single input, it seems that the same digits always follow another set.

D275
C265
57F9
A447
2CCC
D374
50F0
B95A
0325
8DAF
F417
2CCC
0828
F013
F718
97B9
56F6
8DAF
7292
092B
B95A
B152
7494
EC0F
CF70
193B
7292
7A9A
46E6
DE01
BC5F
DB7C
EB0C
6888
23C5
AA4D
1838
6B8D
F619
1E3E
E90A
94B4
EE11
50F0
2CCC
8DAF
40E0
5BFD
2FD1
D275
4CEC
698B
1232
FE21
7696
E102
AA4D
48E8
CA6D
47E9
6D8F
7696
B152
B85B
C768
7799
0B2D
FD1E
4EEE
1739

spamcourt

Can we find / download the tool you are using?

dhromed

Sure. I thought it'd be sliiiightly more fun it that would remain a surprise, but here. Make a new macro, and encrypt the output.

spamcourt

@dhromed said:

Sure. I thought it'd be sliiiightly more fun it that would remain a surprise, but here. Make a new macro, and encrypt the output.

you are right but today I'm in mood to try my super reversing skills. Also, having the source might help. Thanks!

too_many_usernames

It helps if you sort the data - some patterns are immediately visible by looking at sequential leading bytes. It's not a straight add or XOR, but there is some obvious behavior:

0325	6888	C265
0828	698B	C768
092B	6B8D	CA6D
0B2D	6D8F	CF70
1232	7292	D275
1739	7292	D275
1838	7494	D374
193B	7696	DB7C
1E3E	7696	DE01
23C5	7799	E102
2CCC	7A9A	E90A
2CCC	8DAF	EB0C
2CCC	8DAF	EC0F
2FD1	8DAF	EE11
40E0	94B4	F013
46E6	97B9	F417
47E9	A447	F619
48E8	AA4D	F718
4CEC	AA4D	FD1E
4EEE	B152	FE21
50F0	B152	
50F0	B85B	
56F6	B95A	
57F9	B95A	
5BFD	BC5F

We know that 'a' is 0x61, so we have things like:

n8 <op> 61 = m8
n9 <op> 61 = mB</op></op>

Ah, I just noticed: if the first byte is xy where y is even, the new value always ends in that same y; if y is odd, the next byte ends with y+2.

The high nybble is more tricky. We have things like:
0 -> 2, 1->3, 2->C, 2->D, 4->E, 5->F, 6->8. So if this is straight math there are carries involved.

Anyway, this is interesting, but that's about all the time I have for now...

Scarlet_Manuka

@too_many_usernames said:

Ah, I just noticed: if the first byte is xy where y is even, the new value
always ends in that same y; if y is odd, the next byte ends with y+2.

You only looked at the first half or so of the dataset, didn't you? The pattern is different when x is in A-F.

flop

@Scarlet Manuka said:

@too_many_usernames said:
Ah, I just noticed: if the first byte is xy where y is even, the new value always ends in that same y; if y is odd, the next byte ends with y+2.
You only looked at the first half or so of the dataset, didn't you? The pattern is different when x is in A-F.

I remember having seen software for boolean expression simplification somewhere ...just give all known values of input and (single) output bits, and let it derive the boolean expressions for that. Perhaps there's some simple pattern, like right shifts and so on.

fatbull

@dhromed said:

I thought it'd be sliiiightly more fun it that would remain a surprise, but here.

http://allchars.svn.sourceforge.net/viewvc/allchars/trunk/src/Encryp.pas?view=markup#l99
http://allchars.svn.sourceforge.net/viewvc/allchars/trunk/src/MacrosF.pas?view=markup#l449