WORDLE

Filtering 5-letter words may involve some not-so-complicated shell programming. e.g. something along these lines:

while IFS= read -r word; do
    if [ "$(echo -n "$word" | wc -m)" = 5 ] ; then
        echo $word >> "$OUT"
    fi
done < <(cat "$IN")

This is not a complete script, just an idea. Some additional filtering may have to be applied too.

As for the amount, anything less than a few thousands wouldn't be particularly interesting. There shouldn't be any performance problems even with tens of thousands of words.

I decided to keep the pulley menu hidden while the game is being played to keep the page non-flickable (vertical drag could interfere with typing) and make sure that you don't accidentally start a new game. Settings are placed to the flip side of the main screen (rather than invoked from the pulley menu) for pretty much the same reason.

I'll consider making this behavior configurable.

Integrated, thanks!

That's your chance to expand your vocabulary)) I noticed that too, but (almost) every time I googled such words, it was turning out that these words do exists. They are just... strange. Often obsolete.

If you feel that some answers are really really really strange, I suggest that you create an issue on github (or list them here) and I'll consider moving those over to the extended dictionary (the words that can be used for checking but are never proposed for guessing) or even drop them completely. Actually, I already retired a few words from the main dictionary to the extended one.

Speaking of the words that you mentioned, I think хабар is perfectly fine, насып is obsolete but still fine, and I agree that кодак deserves to be moved out of the main dictionary.

I have written a small README about adding new dictionaries.

I have pushed the tool and a README.

Do you mean UI translation or the dictionary? Yes, you can help with both, if you like.

UI translations are here.

The dictionaries are here. Those get installed to /usr/share/harbour-wordle/data directory on the device. The program scans that directory at startup and populates the list of languages, meaning that you don't even have to recompile the app in order to test your language. Just creating /usr/share/harbour-wordle/data/cz directory on the device and copying files there would do the trick.

To make a dictionary, you basically need two lists of 5-letter words. One (called words) contains the words to be guessed, and the second one (xwords) is the list of the additional words that can be used for checking. The first list should contain proper, respectable words which are expected be used in public writing, literature etc. The second one allows jargonism, slang, even swearing (why not?).

I'll push the tool which I used for preparing the dictionaries. That tool converts input into a single byte encoding if necessary (e.g. utf-8 => windows-1250), removes spaces, sorts the words, removes duplicates and writes the words (or xwords) file in the form expected by the program. It's done in a slightly complicated way in order to optimize word search at run time (straight mmap + binary search).

There's also language.json but that's the most trivial part. Its format is obvious.

For English it's 2315 words to guess + 10657 additional words that can be used for checking. Those have been pulled out of the original source (i.e. the JS file which the browser loads from the website mentioned in the description).