I ve written some many years ago dynamic Huffman algorithm to compress and decompress data. It is mainly targeted to data with some symbols occuring more often than the rest (e.g. having some data file consisted of 3 different symbols and their total