112 lines
3.6 KiB
Python
112 lines
3.6 KiB
Python
from gensim.models import KeyedVectors
|
|
from colorama import Fore, Back, Style
|
|
import argparse
|
|
import readline
|
|
import random
|
|
import time
|
|
|
|
|
|
def random_word(model, k=5, dist=100):
|
|
base_words = [
|
|
model.index_to_key[random.randint(0, len(model))]
|
|
for _ in range(k)
|
|
]
|
|
|
|
complete_list = base_words.copy()
|
|
for word in base_words:
|
|
complete_list += [i[0] for i in model.most_similar(word, topn=dist)]
|
|
|
|
rk_words = model.rank_by_centrality(complete_list)
|
|
return rk_words[random.randint(0,5)][1]
|
|
|
|
|
|
def cemantix(model, word=None):
|
|
while word is None or len(word) < 5 or '-' in word or '_' in word:
|
|
word = random_word(model, k=1, dist=1) # augment numbers to try a "smooth selection"
|
|
|
|
nearest = [word]+[i[0] for i in model.most_similar(word, topn=1000)]
|
|
guesses = [] # guess, temp, rank
|
|
def get_rank(guess):
|
|
if guess not in nearest:
|
|
return None
|
|
return 1000 - nearest.index(guess)
|
|
|
|
def formatted_status(guesses, last=None):
|
|
text = ""
|
|
for w, temp, rank in guesses:
|
|
if rank is not None:
|
|
text += Back.RED+Fore.BLACK+Style.BRIGHT+str(rank)+Style.RESET_ALL
|
|
text += "\t"
|
|
if temp >= 75:
|
|
text += Fore.BLUE
|
|
text += str(temp) + Style.RESET_ALL + "\t"
|
|
if last == w:
|
|
text += Style.BRIGHT+Back.WHITE+Fore.BLACK
|
|
text += w + Style.RESET_ALL+"\n"
|
|
return text[:-1]
|
|
|
|
def tried(word, guessed):
|
|
return word in [i[0] for i in guessed]
|
|
|
|
def interpret_command(cmd, guesses):
|
|
match cmd:
|
|
case "clear":
|
|
guesses = [g for g in guesses if g[1] <= 75.]
|
|
case "help":
|
|
best_rk = max([rk for _, _, rk in guesses if rk is not None]+[749])
|
|
print("Maybe try "+Back.YELLOW+Fore.BLACK+nearest[999-best_rk]+Style.RESET_ALL)
|
|
case _:
|
|
print(Fore.RED+"Unknown command"+Style.RESET_ALL)
|
|
|
|
return guesses
|
|
|
|
|
|
while True:
|
|
try:
|
|
guess = input(Style.BRIGHT+"Your guess > "+Style.RESET_ALL).strip()
|
|
readline.add_history(guess)
|
|
if guess.endswith("()"):
|
|
guesses = interpret_command(guess[:-2], guesses)
|
|
continue
|
|
except (EOFError, KeyboardInterrupt):
|
|
print("The word was "+Style.BRIGHT+word+Style.RESET_ALL)
|
|
print("Goodbye!")
|
|
return -1
|
|
try:
|
|
dist = round(round(model.distance(word, guess), 4)*100, 2)
|
|
except KeyError:
|
|
print(Fore.RED+"Key not present"+Style.RESET_ALL)
|
|
continue
|
|
|
|
if not tried(guess, guesses):
|
|
guesses.append((guess, dist, get_rank(guess)))
|
|
guesses.sort(key=lambda x:-x[1])
|
|
print(chr(27) + "[2J")
|
|
print(formatted_status(guesses, last=guess))
|
|
if guess == word:
|
|
time.sleep(1)
|
|
print(Fore.GREEN+"Correct!"+Style.RESET_ALL+f" {len(guesses)} tries.")
|
|
return len(guesses)
|
|
|
|
|
|
|
|
|
|
def main():
|
|
parser = argparse.ArgumentParser(description="local infinite cemantix")
|
|
parser.add_argument("-w", "--word", dest="word", action="store",
|
|
help="Specify goal word")
|
|
parser.add_argument("-m", "--model", dest="model", action="store",
|
|
default="frWac_non_lem_no_postag_no_phrase_200_skip_cut100.bin",
|
|
help="Specify model to use")
|
|
args = parser.parse_args()
|
|
|
|
model = KeyedVectors.load_word2vec_format(
|
|
args.model,
|
|
binary=True,
|
|
unicode_errors="ignore"
|
|
)
|
|
cemantix(model, word=args.word)
|
|
|
|
if __name__ == "__main__":
|
|
main()
|