Correct typo according to http://norvig.com/spell-correct.html

typo.py

#%%
letter = 'abcdefghijklmnopqrstuvwxyz'

import requests, re, collections

# fetch the big.txt to calculate every words probability.
big_text = requests.get("http://norvig.com/big.txt").text

#use regex to match all words and then count their frequency
statistic = collections.Counter(re.findall("\w+", big_text))

#%%

def edit1(word):
    """
    candidate words which edit distance is 1 
    """
    split = [(word[:i], word[i:]) for i in range(len(word)+1)]
    delete_one = [L + R[1:] for L, R in split if len(R) > 0]
    add_one = [L + w + R for L, R in split for w in letter]
    replace_one = [L + w + R[1:] for L, R in split if len(R) > 0 for w in letter]
    transcode = [L + R[1] + R[0] + R[2:] for L, R in split if len(R) > 2]
    return delete_one + add_one + replace_one + transcode

def P(word):
    """
    Calculate every words frequency
    """
    return statistic[word] / sum(statistic.values())

def edit2(word):
    """
    edit distance is 2, edit2(word) = edit1(edit1(word))
    """
    return (w for e in edit1(word) for w in edit1(e))

def known(words):
    """
    filter words if not exist in big.txt
    """
    return list(filter(lambda w: w in statistic, words))

def candidates(word):
    """
    1、if word exist, it is the corret spelling.
    2、edit distance 1 exist word
    3、edit distance 2 exist word
    4、unknown word and edit distance 1 & 2 both does not exist, return the word
    """
    # print(set(known(edit1(word))))
    # print(set(known(edit2(word))))
    return (known([word]) or known(edit1(word)) or known(edit2(word)) or [word])

def correction(word):
    """
    In all candidate just choose the highest probability word.
    """
    return max(candidates(word), key=P)


#%%
print(correction('worlw'))