Given two strings of equal length that contain digits and some placeholder value, I want a fast way to determine if the pair are 'compatible'.

Eg. Given (where underscore represents the placeholder but this could be substituted by any other non-digit value that helped the algorithm)

_8__3__19
48____7__
[download]

These two strings are compatible because no single digit appears in both strings except where it appears in the same position (8)

Whereas these two would be incompatible

_8__3__19
4_8___7__
[download]

because the digit 8 appears in both, but at a different position. And these two are incompatible

_8__3__19
48_____7_
[download]

because the second last digit in both strings contains a different value.

I know how to do this with a loop and substr and a hash, but it is rather slow and I have a set N of M sets of these strings and I want to produce a set P, where each element of P is a string that is combines one string from as many of the N sets of strings as are compatible. Given the combinatorial nature of the problem, I need to make it as quick as possible.

I keep thinking that this can be done using string-wise boolean operations, but I cannot see how?

If there is a better way of structuring the data (eg. arrays of char rather than strings) that makes the process quicker or easier that's good to.