# Convert integer to balanced dozenal

Create a function which takes an integer and returns an array of integers representing the number in balanced base twelve. Specifically, the representation should follow these rules regarding 6 & -6 (from the above link):

1. If the last digit of a number could be written with either 6 or ϑ, we use the one whose sign is opposite the sign of the first digit. For example, we write decimal 30 as dozenal , but decimal -30 as as dozenal ε6, not ζϑ.
2. If a 6 or ϑ occurs before the last digit of a number, we use the one whose sign is opposite the next digit after it. For example, we write decimal 71 as dozenal , but decimal 73 as as dozenal 1ϑ1, not 61.
3. If a 6 or ϑ is followed by a 0, we keep looking to the right until we can apply one of the first two rules. For example, we write decimal -72 as dozenal τ60, not ϑ0.

Otherwise, the only restrictions are to use only digits in [-6..6], and for the output, when interpreted as big-endian base twelve, to be equivalent to the input.

Example of use:

f(215) => [1,6,-1]
f(216) => [2,-6,0]
f(-82) => [-1,5,2]
f(0)   => 


Winner is the shortest, with upvotes being the tiebreaker.

## J, 74 72 characters

Here's my J implementation (f =. not included in character count):

f =. **}.^:(*@#*0={.)@([:((12-~[),0>:@{.@][]}]),@.(6>[)/&.|.0,12&#.^:_1)@|


And some tests (each result is boxed, and underscore means negative):

   NB. Integers [-10..10]:
(<@f"0)i:10
┌────┬────┬────┬────┬────┬──┬──┬──┬──┬──┬─┬─┬─┬─┬─┬─┬────┬────┬────┬────┬────┐
│_1 2│_1 3│_1 4│_1 5│_1 6│_5│_4│_3│_2│_1│0│1│2│3│4│5│1 _6│1 _5│1 _4│1 _3│1 _2│
└────┴────┴────┴────┴────┴──┴──┴──┴──┴──┴─┴─┴─┴─┴─┴─┴────┴────┴────┴────┴────┘
NB. Some boundary values:
(<@f"0)215 216 217 71 72 73
┌──────┬──────┬──────┬────┬──────┬──────┐
│1 6 _1│2 _6 0│2 _6 1│6 _1│1 _6 0│1 _6 1│
└──────┴──────┴──────┴────┴──────┴──────┘


Interestingly, the rules for 6 have no special place in the function. They work that way just by handling all digits >= 6 together and only working on positive numbers.

I made another version, which doesn't rely on #.^:_1, but it's longer:

**_2&([:}.^:(0={.)+/\)@(((<.@%&12,(1,-&12)(0,])@.(6>])@(12&|))@{.,}.)^:(0~:{.)^:_)@(,&0)@|


### GolfScript, 6953 49 characters

[.0<\{.12:^%.3$5+>{^-\^+\}*@1$.0<@if@^/.}do;;]-1%


GolfScript expression which takes top of the stack and replaces it by the result. You can play with the code online.

• Working on a different solution which should be much shorter - but unfortunately doesn't work for several corner cases... Dec 31, 2013 at 18:42

# APL(NARS), 71 chars, 142 bytes

{⍵=0:,0⋄⍵{⍵=0:⍬⋄((m+⍺×m=0)∇⌊12÷⍨⍵+q),m←t-q←12×(6<t)+(⍺>0)×6=t←12∣⍵}⍨×⍵}


test:

  z←{⍵=0:,0⋄⍵{⍵=0:⍬⋄((m+⍺×m=0)∇⌊12÷⍨⍵+q),m←t-q←12×(6<t)+(⍺>0)×6=t←12∣⍵}⍨×⍵}
⎕fmt z¨215 216 ¯82 0 30 ¯30 71 72 73 ¯72
┌10──────────────────────────────────────────────────────────────────────────────────────┐
│┌3──────┐ ┌3──────┐ ┌3──────┐ ┌1─┐ ┌2────┐ ┌2────┐ ┌2────┐ ┌3──────┐ ┌3──────┐ ┌3──────┐│
││ 1 6 ¯1│ │ 2 ¯6 0│ │ ¯1 5 2│ │ 0│ │ 3 ¯6│ │ ¯3 6│ │ 6 ¯1│ │ 1 ¯6 0│ │ 1 ¯6 1│ │ ¯1 6 0││
│└~──────┘ └~──────┘ └~──────┘ └~─┘ └~────┘ └~────┘ └~────┘ └~──────┘ └~──────┘ └~──────┘2
└∊───────────────────────────────────────────────────────────────────────────────────────┘