Currently fallback languages/locale only looks up a language/locale if a specific language/locale doesn't exist.
Ideally we'd rather have a way to look up possible missing entries, and parse through those as well from other languages.

Currently lexicon_text() only handles strings that are within the text entries. This slight addition will allow you to pass a string as the first argument for Lexicon to modify, allowing you to parse and replace text with the appropriate %S and %var%.

Looking for towards v3.

Allows Lexicon to reference variables from other objects/structs.

i.e. %score% will default to any structs passed as an argument. If none of the structs contain "score", then it'll fall back to the object that called lexicon_text and then global if the object doesn't have it either.

Lexicon will also allow passing in direct references, skipping multiple checks.

%score% - References as per order (structs -> self -> global)

%object_index#score% - References the object provided

%self#score% - References self

%global#score% - References the global space.

%foo#score% - References as per order (structs -> self -> global), but treats foo as a struct and retrieves its variable.

This will also possibly allow additional strings returned to be passed through.

Currently if you provide foo.bar to lexicon_text and the text entry doesn't exist, it'll return foo.bar.
This would allow you to alter the value that Lexicon returns instead for invalid text entries.

See https://icu4c-demos.unicode.org/icu-bin/locexp

Been considering supporting this

{
    "greetings": [
        "Hi!",
        "Howdy!",
        "How do you do?"
    ]
}

---> "greetings0", "greetings1", "greetings2"

Frankly something I should've had reintroduced into 2.0.0 but I briefly went away from it to experiment on other concepts.

Allows to define a seperator string that Lexicon will skip over, rather than adding it as a text entry.

The idea is to extract all of the necessary unicode ranges that a font needs to render the correct characters for any piece of text.

Hi, great project, really useful... I think I found a cache bug in lexicon_text where it doesn't generate the correct cache key when you have LEXICON_CACHE_ARG_THRESHOLD (or more arguments).
It includes _textEntry in the cache key twice because it's not iterating the arguments correctly. The below 2 changes seem to work, hope they make sense:

for(var _i = 1; _i <= _count; ++_i) {       //change to <=
	if (_i > argument_count) break;
	_args[_i-1] = argument[_i];           //change to argument[_1];