Wordle Solver

Find it difficult to solve Wordle? Here are some tools that help you:

• Find all 5-letter possible words that follow a given pattern

  from solver import words_like
  
  print(words_like("_IGHT"))
  print(words_like("C___S"))

• Narrow down all possible words given your previous attempt results

  from solver import possible_words, possible_words_simple
  
  # A simple way: provide all your previous attempts and results
  print(possible_words_simple(
      trials=["slate", "mealy"],
      colors=["BYGBY", "BGGYG"]
  ))
  
  # Alternatively, provide all yellows, greens and blacks (letter and position)
  print(possible_words(
      yellows=["L2", "E5"],
      greens=["A3", "E2", "Y5"],
      blacks=("S", "T", "M")
  ))

• Suggest your next attempt given your previous attempt results

  from solver import suggestion
  
  print(suggestion(
      yellows=["L2", "E5"],
      greens=["A3", "E2", "Y5"],
      blacks=("S", "T", "M"),
      attempts=['slate', 'mealy']
  ))

• Solve Wordle games automatically

  from solver import solver, test_solver_single, test_solver_all
  
  # input the result each time (e.g. 'YBGBY') and get suggestion
  solver()
  
  # show how to solve one word
  test_solver_single('leafy')
  
  # evaluate on all 2.5k Wordle words 
  test_solver_all()

  test_solver_single works this way:

  round 1, pred 'slate', got ⬛️🟨🟩⬛️🟨
  round 2, pred 'mealy', got ⬛️🟩🟩🟨🟩
  round 3, pred 'leaky', got 🟩🟩🟩⬛️🟩
  round 4, pred 'leafy', got 🟩🟩🟩🟩🟩
  

• Play Wordle

  python play_wordle.py

  It works this way:

  your guess: broad
  ⬛️⬛️🟨⬛️⬛️. your guess: joker
  ⬛️🟩⬛️🟩⬛️. your guess: month
  🟨🟩🟨⬛️⬛️. your guess: women
  🟩🟩🟩🟩🟩. congratulations!
  

• Play Absurdle -- A tough version, strongly recommend! The correct answer changes according to your prediction

  python play_absurdle.py

  It works this way:

   ⬛️⬛️⬛️⬛️⬛️. your guess: mouse
   ⬛️🟩⬛️⬛️⬛️. your guess: hobby
   🟨🟩⬛️⬛️⬛️. your guess: horny
   🟨🟩🟩⬛️⬛️. your guess: forth
   ⬛️🟩🟩🟨🟩. your guess: torch
   🟩🟩🟩🟩🟩. congratulations!
  

Notes

suggestion is based on a simple greedy strategy: computes score of all words based on freq (The frequency that one letter appears at one position), and select the word with highest score.

It is a rather simple method which can still be improved, for example we sometimes choose a word with unused characters to narrow down possible answers.

test_solver_all now gives an average attempts of 3.9 times (maximum 9 times). Alternative methods on Github can be no greater than maxmimum 6 times.