fuzzy_matcher.py

import jellyfish

def fuzzy_set(string_list, similarity_ratio):
    '''
    Input  : List of strings
    Output : Set of strings. All similar kinds of strings are merged together
             and the most
    '''
    string_list = list(set(string_list))
    string_list.sort()

    flag = 0

    for i in string_list:
        flag = 0
        f = string_list.index(i) + 1
        for j in string_list:
            if jellyfish.jaro_winkler(i, j) >= similarity_ratio and string_list.index(i) != string_list.index(j):
                flag = 1
                break
        if flag == 1:
            string_list.pop(string_list.index(i))

    return set(string_list)