B Melborp
---------

This is Problem B in reverse.

The look and say description of a sequence of digits is defined in 
problem B.  If n is a positive integer, let las(n) be the result
of describing the ordinary decimal representation of n (with no
leading 0's).

For example, las(122344111) = 1122132431 and las(1111111111) = 101.
Note that las(10^112 - 1) = las(1199) = 1129, so las is not a one-to-one
function.  In this question, you will be given a number m and you must
find the length of the smallest positive integer n such that las(n)=m.
(Here, the length of n just means the number of digits needed to
represent it in decimal.)

For example, if m = 1129, 1199 is the smallest number n such that 
las(n) = 1129, so you should output 4 (the length of 1129).

Input
-----

Each line will contain a positive integer m.  The end of the input
will be indicated by a value m = 0, which should not be processed.
Each m will be less than 10^17.

Output
------

For each input value m, output the length of the smallest positive
integer n such that las(n) = m.  If there is no such n, print "impossible".
Each output should appear on a new line.

Sample Input
------------

1129
3131
7
1010
0

Sample Output
-------------

4
313
impossible
impossible