Not a kata issue.

Your code isn't failing on "", it's failing on a different input. Look more closely... you fail to account for this input in your code.

Please review: C++ Translation

Reread the description. You get "wrong" input, yes, but you have to correct that input to get the result :)

It will shuffled up to digit. Not only numbers.

Yeah, same.

let testEncodeOuput="000111000111000111000000000000111111111111111111000111111111000000111111000111111111000111000000000000111000000000000111000000111000000111000111"
let myEncodeOuput="000111000111000111000000000000111111111111111111000111111111000000111111000111111111000111000000000000111000000000000111000000111000000111000111"
let theirDecodeInput="000111000111000111000010000000111111111111011111000111111111000000111111000111101111000111000000000000111000000000000111000000111000000111000111"
in last test case they are giving wrong input but expecting same result as input of encode function .Here my encode output and their are same,but in decode function they are giving different input than output of encode funtion.

findNumber(1, 21, "2198765123416171890101112131415")
from this i got [20] instead of [12,21] ,more over i can see 21 and 12 both are present in the string .How this gonna work?

may be test's are different now,that's why your solution won't pass all the test's.

The answer is 4:16, but what you return is 4:12.
Your solution works fine now, so this is not a problem for you anymore, I think.

You're probably right, I haven't checked it out anymore. But now I can see the "author thinking" when I expand the "Test Cases" tab ;)

I don't think so, it still occasionally fails.

I think I managed to find the way of thinking of the author of this executioner. See my final solution (Python) ;)