Python script to determine the optimal starting guess for Wordle using Shannon information entropy.

wordle_github.py

# This Python script calculates the Shannon information entropy
# for each initial guess in Wordle. Words with high entropy are
# good starting guesses for Wordle.


import numpy as np
import requests
from collections import Counter


def get_data():
    """
    This function retrieves the Wordle wordlist from a pastebin site.
    It returns the list of all five-letter words which are allowed as guesses,
    and a shorter list of common five-letter words, from which the secret
    word of the day is selected.
    """
    url = 'https://paste.ee/d/4zigF/0'
    lines = requests.get(url).text.split('\n')[1:-1]
    words = [line.split('\t')[2] for line in lines]
    targets = words[:words.index('aahed')]
    return words, targets


def get_hints(guess, target):
    """
    This function returns the hints resulting from a guess and a secret (target) word.
    The hints are encoded as a five-digit number from 00000 to 22222, where 0 means that
    the letter is not in the target, 1 means that the letter is in the target but in the
    wrong position, and 2 means that the letter is in the target and in the correct position.
    """
    guess = guess.lower()
    target = target.lower()
    length = len(target)
    hints = 0
    matched = [False] * length
    for i in range(length):
        if guess[i] == target[i]:
            hints += 2 * 10 ** (length - i - 1)
            matched[i] = True
    for i in range(length):
        if guess[i] != target[i]:
            for j in range(length):
                if guess[i] == target[j] and not matched[j]:
                    hints += 10 ** (length - i - 1)
                    matched[j] = True
                    break
    return hints


def entropy(guess, targets):
    """
    Calculates the Shannon information entropy for a guess, given a list of target words.
    """
    counter = Counter(get_hints(guess, target) for target in targets)
    x = np.array(list(counter.values()))
    p = x / np.sum(x)
    return -np.sum(p * np.log2(p))


def main():
    words, targets = get_data()
    print("Most informative guess - common words only:")
    print(*max((entropy(guess, targets), guess) for guess in targets))
    print("Most informative guess - all words:")
    print(*max((entropy(guess, targets), guess) for guess in words))


if __name__ == '__main__':
    main()