I have a question about algorithm:
How to find all characters in a string whose appearance is greater than a specific number, say 2 for example efficiently?
Counting sort will be extremely efficient for one-byte encodings, border case is two-byte encodings. For wider encodings it is not so efficient, but counting array may be replaced with hash table.
EDIT: By the way, that is too general solution, doing only counting phase and outputting results on the fly will be more than enough.