refactor: api.lua

[S1M0N38]

The Lua file containing the endpoints implementation is starting to get a bit out of hand. When the Lua API is somewhat stable and all the main endpoints have been implemented, I'd like to refactor to a similar structure like in tests/lua/endpoints.

What do you think, @stirby ?

[S1M0N38]

We have also this block repeated a lot of times:

API.pending_requests["name"] = {
  condition = utils.COMPLETION_CONDITIONS["name"][args.action or ""],
  action = function()
    local game_state = utils.get_game_state()
    API.send_response(game_state)
  end,
  args = args,
}

that we can enclose in a separate function (or directly in the API update).

But first, we need to clean up spurious completion conditions like the one in the shop for buying a card.

[stirby]

Absolutely agreed. Decomposing into endpoints/ would be a large improvement.

The shop function is going to be dense one way or another... I wouldn't be opposed to making distinct function calls for each of the shop actions as part of this work.

[S1M0N38]

When refactoring the api.lua, we also need to double-check:

• endpoint functions' logic (compare to the ones in button_callbacks.lua)
• waiting condition. For now, we've used simple conditions (compare to the ones in button_callbacks.lua)

[S1M0N38]

I've noticed that for the rearrange endpoints we are using:

rearrange_consumeables({consumeables = [...]})
rearrange_jokers({jokers = [...]})
rearrange_hand({card = [...]})

maybe could be simplified to

rearrange_consumeables({cards = [...]})
rearrange_jokers({cards = [...]})
rearrange_hand({cards = [...]})

so that cards is always a list of int (with card pos in all the endpoints)

We can also port other endpoints that are using index to card. e.g.

buy_card({index = ...})

to

buy_card({card = ...})

so that card is always an int (with card pos in all the endpoints).

---

Alternatively, we can have the same endpoint, with different params, i.e.

rearrange({consumeables = [...]})
rearrange({jokers = [...]})
rearrange({hand = [...]})

[S1M0N38]

Right now we are implementing a defer mechanism for callbacks:

• api.lua : wait for new game state for returning it
• log.lua : wait for new game state for logging it

Maybe we can leverage Balatro event manager system

[S1M0N38]

After the refactoring we must update the docs as well.

[S1M0N38]

See #82. Make use of the highlight_cards endpoint in play_hand_or_discard (and other endpoint that need to select cards).