Saving, loading, and other languages

Save and load

@enum Mood happy unhappy mixed
@ai function mood(review::String)::Mood
    "How does the customer feel about what they bought?"
end
taught = with_demos(mood, ["Broke in a day." => unhappy, "Love it!" => happy])

dir = joinpath(mktempdir(), "mood")
FunctAI.save(dir, taught)
readdir(dir)

1-element Vector{String}:
 "functai.json"

functai.json holds the signature, the instruction, the worked examples, the layout and the settings, with the fingerprints of the requests it renders. It is indented JSON, to read, diff and keep in git.

loaded = FunctAI.load(dir; types = (result = Mood,))
version(loaded) == version(taught)

true

Loading checks, before any call, that the function sends exactly what it sent when it was saved; a difference refuses with LoadRefused (saved-differs) rather than running another function under this one's name. A folder holds JSON shapes, not Julia types, so without types a choice comes back as String, a record as a NamedTuple; types gives fields their Julia types back, when the shapes agree.

Across languages

A function saved in Python, TypeScript or R loads here and sends the same bytes; one saved here loads there. What only the saving language can run is refused, with the reason:

For the same reason, a Julia function with code of its own or tools can't be saved yet: the folder carries no Julia code.

python_mood = FunctAI.load("saved/mood")        # saved by functai.save in Python
python_mood("Arrived late but fine.")