vortjo.py

#!/usr/bin/env python
import requests
import lxml.etree
import tqdm
import math
import heapq
import numpy
import random

xml_text = requests.get("http://www.remush.be/tezauro/PIV.xml").content
xml_text = xml_text.replace(b"encoding=\"UTF-16\"", b"encoding=\"UTF-8\"")
words = [
    word.text.strip()
    for word in lxml.etree.fromstring(xml_text)
]
print(f"{len(words)} esperanto words")

wordle_words = [word for word in words if len(word) == 5 and word == word.lower()]

print(f"{len(wordle_words)} wordle esperanto words")

def wordle_filter(words, green_letters, yellow_letters, gray_letters):
    return [
        word
        for word in words
        if all([
                all(letter == word[i]  for (i, letter) in green_letters),
                all(letter != word[i]  for (i, letter) in yellow_letters),
                all(letter in word     for (_, letter) in yellow_letters),
                all(letter not in word for letter in gray_letters),
        ])
    ]

def wordle_entropy(words, guess):
    entropy = 0
    for letter in set(guess):
        selected_words = sum(letter in word for word in words)
        p = selected_words / len(words)
        entropy += -p * (math.log(p) / math.log(2))
    return entropy

def wordle_best_guesses(words, n, quiet = False):
    return heapq.nlargest(
        min(n, len(words)),
        tqdm.tqdm(words, disable=quiet),
        key=lambda guess: wordle_entropy(words, guess),
    )

candidates = wordle_words
for guess in wordle_best_guesses(candidates, 20):
    print(guess, wordle_entropy(wordle_words, guess))

# If you want a hint, fill in the letters of your wordle game here.
candidates = wordle_filter(
    wordle_words,
    green_letters=((0, "l"), (1, "a"), (2, "ŭ"), (4, "o")),
    yellow_letters=(),
    gray_letters="erniutrkt",
)
for guess in wordle_best_guesses(candidates, 20):
    print(guess, wordle_entropy(wordle_words, guess))

# These functions are for simulating an entire game.
def wordle_grade_guess(guess, answer):
    if len(guess) != len(answer):
        raise RuntimeError(f"Guess must be {len(answer)} long")
    green = []
    yellow = []
    gray = []
    for i, letter in enumerate(guess):
        if letter == answer[i]:
            green.append((i, letter))
        elif letter in answer:
            yellow.append((i, letter))
        else:
            gray.append(letter)
    return green, yellow, gray

def wordle_play(wordle_words):
    answer = random.choice(wordle_words)
    candidates = wordle_words
    guess = None
    guesses = 0
    green = set()
    yellow = set()
    gray = set()
    while guess != answer:
        guess = wordle_best_guesses(candidates, 1, quiet=True)[0]
        this_green, this_yellow, this_gray = wordle_grade_guess(guess, answer)
        green |= set(this_green)
        yellow |= set(this_yellow)
        gray |= set(this_gray)
        # print(len(green), len(yellow), len(gray), guess)
        candidates = wordle_filter(wordle_words, green, yellow, gray)
        guesses += 1
    return guesses