This code provides a utility to help solving of Wordle problems. The daily Wordle challenge can be found here: https://www.powerlanguage.co.uk/wordle/
The usage of the utility is as follows:
java -jar wordle-solver.jar <pattern> [exclusions]
The pattern uses the following elements:
- '-' for an unknown letter
- uppercase for a correct letter in the right place
- lowercase for a letter that's in the word but not at this location
Examples:
----- : No guesses
S---- : Letter 's' present in the word at the correct position
--e-- : Letter 'e' present in the word but not at this position
The exclusions are a list of letters that have been tried and found to not be present in the word. Example:
arf
Therefore, a complete example is:
java -jar wordle-solver.jar S-e-- arf
The output of the commands is a list of possible words ordered by probability:
stile: 0.2883618312189741 (1.0, 0.11858797573083288, 0.12465526751241036, 0.07225592939878654, 0.12630998345284059)
stole: 0.288030888030888 (1.0, 0.11858797573083288, 0.12300055157198014, 0.07225592939878654, 0.12630998345284059)
stine: 0.2867071152785438 (1.0, 0.11858797573083288, 0.12465526751241036, 0.06398234969663541, 0.12630998345284059)
stone: 0.28637617209045774 (1.0, 0.11858797573083288, 0.12300055157198014, 0.06398234969663541, 0.12630998345284059)
The first figure after the word is the average probability of each letter being correct.
This list of additional figures is the propbability of each individual character in the word being correct.
The dictionary filter uses the following algorithm:
First it filters the dictionary down to only words that:
- have the correct number of letters
- have the letters at known positions present
It then further filters the first set of word into a second set that only contains those words that:
- do not contain any of the excluded letters
- contain all the mis-located letters in any of the other locations
- on early passes (< 6 excluded letters) words with repeated letters are also excluded as this increases the chances of early letter matches.
The result is then two sets of words:
- One with all the words that could occur with just the known letters
- One with all the words that could occur given all the other constraints taken into account
For each of these two sets of words an analysis is then generated that looks at the probability of each letter occurring at the given location in the set of words.
The candidate generator then takes the 5 most common letters at each location in the second set of words, generates every possible combination of these letters and restricts them to only words present in this set of words. These are the guess candidates.
Finally, the solution sorts these words based on the probabilities of each letter occurring at that position in the wider first set of words.
The words are then listed in descending order of probability.