import os
import operator
import numpy as np
from binarytree import tree,Node                

def entropy_single(p):
    return p*np.log2(1/p)


def entropy_of_list(data,letter_number):
    H = 0
    for letter in data:
        (letter_name, letter_frequency) = letter
        probability_letter = letter_frequency/letter_number
        H = H + entropy_single(probability_letter)
    return H

def preprocess_sequence(sequences):
    letter = set(sequences)
    data = {i:sequences.count(i) for i in letter}
    letter_number = len(data)
    data_sort  = sorted(data.items(), key=operator.itemgetter(1), reverse=True)
    
    return data_sort,letter_number

#Input 
sequences = "Hello thanh, good!"
data_sort,letter_number = preprocess_sequence(sequences)

H = entropy_of_list(data_sort,letter_number)
print(H)