import pandas as pd import numpy as np from IPython.display import clear_output

word_bank_series = pd.read_csv('words-new.txt',names=['words']).squeeze()

class Wordle: def __init__(self,letters_series): self.series = letters_series def word(self,chosen_word): self.chosen_word = chosen_word self.series = self.series.loc[~self.series.str.contains(self.chosen_word,regex=False)].reset_index(drop=True) def filter_green(self,green_letters=None): self.existing = list(filter(None,green_letters)) self.green_letters = green_letters if any(green_letters): regex_filter = ['[a-z]' if letter is None else letter for letter in green_letters] regex_filter = ''.join(regex_filter) self.series = self.series.loc[self.series.str.contains(regex_filter,regex=True)].reset_index(drop=True) def filter_yellow(self,yellow_letters=None): if any(yellow_letters): self.yellow = list(filter(None,yellow_letters)) self.existing.extend(self.yellow) #First filter: keep words with 'yellow letters' regex_filter = [f'(?=.*{letter})' for letter in yellow_letters if letter is not None] regex_filter = ''.join(regex_filter) self.series = self.series.loc[self.series.str.contains(regex_filter,regex=True)].reset_index(drop=True) #Second filter: drop words with 'yellow letters' in their place. regex_filter = ''.join(['[a-z]' if letter is None else f'[a-z](?<!{letter})' for letter in yellow_letters]) self.series = self.series.loc[self.series.str.contains(regex_filter,regex=True)].reset_index(drop=True) #Third filter: if a yellow letter is a green letter, keep only words with both. yellow_green_letters = [letter for letter in yellow_letters if letter in self.green_letters and letter is not None] if any(yellow_green_letters): regex_filter = ''.join([f'(?=.*[{letter}].*[{letter}])' for letter in yellow_green_letters]) self.series = self.series.loc[self.series.str.contains(regex_filter,regex=True)].reset_index(drop=True) def filter_gray(self,gray_letters=None): if any(gray_letters): self.gray_letters = gray_letters #Drop letters that are gray, but skip gray letters that are also green or yellow letters. regex_filter = [letter for letter in gray_letters if letter is not None and letter not in self.existing] if regex_filter: regex_filter = '|'.join(regex_filter) self.series = self.series.loc[~self.series.str.contains(regex_filter,regex=True)].reset_index(drop=True) #Drop gray letters that are also green letters but are in a different position gray_green_letters_list = set(filter(None,gray_letters)).intersection(list(filter(None,self.green_letters))) if gray_green_letters_list: pre_regex = '|'.join(gray_green_letters_list) regex_filter = ''.join([f'[a-z](?<!{pre_regex})' if letter is None else f'[{letter}]' for letter in self.green_letters]) self.series = self.series.loc[self.series.str.contains(regex_filter,regex=True)].reset_index(drop=True) def show_by_incidence(self): if self.series.size > 5 and any(self.existing): word_letters_df = self.series.str.split("",expand=True).drop(labels=[0,6],axis=1) columns = [index+1 for index,letter in list(enumerate(self.gray_letters)) if letter is None] words_incidence = word_letters_df.value_counts(subset = columns).iloc[0:1] incidence_combination = words_incidence.index.map(list) pre_regex_incidence = list(map(lambda letter: "(?=.*{})".format(letter),incidence_combination[0])) regex_incidence = "|".join([''.join(row) for row in pre_regex_incidence]) best_options = self.series.loc[self.series.str.contains(regex_incidence,regex=True)].reset_index(drop=True) print(best_options[0:5]) else: print(self.series)

class Answers(Wordle): def __init__(self,letter_series): super().__init__(letter_series) @property def player_input(self): print(f'Word: {self.word}') print(f'Color places: {self.color_position}') @player_input.setter def player_input(self,word_and_color = None): self.word, color_position = word_and_color while len(self.word) != 5: print('Sorry, the word must have five letters.') self.word = input('Word: ') self.color_position = color_position.split(',') while len(self.color_position) != 5: print('Sorry, you must specify five color positions.') color_position = input('Color places: ') self.color_position = color_position.split(',') def __format_input(self,word,color_position): self.word_letters = list(self.word) self.color_letter = list(zip(self.color_position,self.word_letters)) self.green_letters = [letter if color == 'green' else None for color,letter in self.color_letter] self.yellow_letters = [letter if color == 'yellow' else None for color,letter in self.color_letter] self.gray_letters = [letter if color == 'gray' else None for color,letter in self.color_letter] def analyze_letters(self): self.__format_input(self.word,self.color_position) super().word(self.word) super().filter_green(self.green_letters) super().filter_yellow(self.yellow_letters) super().filter_gray(self.gray_letters) if self.series.empty: print('Sorry, it appears the word is not in the dataset.') else: super().show_by_incidence()

#Number of chances a player has to find the right word. count = 0 #Reset the Series with the answers. answers = Answers(word_bank_series)

chosen_word = input('Word: ') #If you won, enter "win" as the input. while count < 6 and chosen_word != 'win': color_position = input('Color places:') answers.player_input = (chosen_word,color_position) answers.player_input answers.analyze_letters() chosen_word = input('Word: ') count += 1