Split battle/ into data/ and engine/ components

engine/battle/ai/move.asm

@@ -0,0 +1,221 @@
+AIChooseMove: ; 440ce
+; Score each move in EnemyMonMoves starting from Buffer1. Lower is better.
+; Pick the move with the lowest score.
+
+; Wildmons attack at random.
+	ld a, [wBattleMode]
+	dec a
+	ret z
+
+	ld a, [wLinkMode]
+	and a
+	ret nz
+
+; No use picking a move if there's no choice.
+	farcall CheckEnemyLockedIn
+	ret nz
+
+
+; The default score is 20. Unusable moves are given a score of 80.
+	ld a, 20
+	ld hl, Buffer1
+	ld [hli], a
+	ld [hli], a
+	ld [hli], a
+	ld [hl], a
+
+; Don't pick disabled moves.
+	ld a, [EnemyDisabledMove]
+	and a
+	jr z, .CheckPP
+
+	ld hl, EnemyMonMoves
+	ld c, 0
+.CheckDisabledMove:
+	cp [hl]
+	jr z, .ScoreDisabledMove
+	inc c
+	inc hl
+	jr .CheckDisabledMove
+.ScoreDisabledMove:
+	ld hl, Buffer1
+	ld b, 0
+	add hl, bc
+	ld [hl], 80
+
+; Don't pick moves with 0 PP.
+.CheckPP:
+	ld hl, Buffer1 - 1
+	ld de, EnemyMonPP
+	ld b, 0
+.CheckMovePP:
+	inc b
+	ld a, b
+	cp EnemyMonMovesEnd - EnemyMonMoves + 1
+	jr z, .ApplyLayers
+	inc hl
+	ld a, [de]
+	inc de
+	and $3f
+	jr nz, .CheckMovePP
+	ld [hl], 80
+	jr .CheckMovePP
+
+
+; Apply AI scoring layers depending on the trainer class.
+.ApplyLayers:
+	ld hl, TrainerClassAttributes + TRNATTR_AI_MOVE_WEIGHTS
+
+	; If we have a battle in BattleTower just load the Attributes of the first TrainerClass (Falkner)
+	; so we have always the same AI, regardless of the loaded class of trainer
+	ld a, [InBattleTowerBattle]
+	bit 0, a
+	jr nz, .battle_tower_skip
+
+	ld a, [TrainerClass]
+	dec a
+	ld bc, 7 ; Trainer2AI - Trainer1AI
+	call AddNTimes
+
+.battle_tower_skip
+	lb bc, CHECK_FLAG, 0
+	push bc
+	push hl
+
+.CheckLayer:
+	pop hl
+	pop bc
+
+	ld a, c
+	cp 16 ; up to 16 scoring layers
+	jr z, .DecrementScores
+
+	push bc
+	ld d, BANK(TrainerClassAttributes)
+	predef FlagPredef
+	ld d, c
+	pop bc
+
+	inc c
+	push bc
+	push hl
+
+	ld a, d
+	and a
+	jr z, .CheckLayer
+
+	ld hl, AIScoringPointers
+	dec c
+	ld b, 0
+	add hl, bc
+	add hl, bc
+	ld a, [hli]
+	ld h, [hl]
+	ld l, a
+	ld a, BANK(AIScoring)
+	call FarCall_hl
+
+	jr .CheckLayer
+
+; Decrement the scores of all moves one by one until one reaches 0.
+.DecrementScores:
+	ld hl, Buffer1
+	ld de, EnemyMonMoves
+	ld c, EnemyMonMovesEnd - EnemyMonMoves
+
+.DecrementNextScore:
+	; If the enemy has no moves, this will infinite.
+	ld a, [de]
+	inc de
+	and a
+	jr z, .DecrementScores
+
+	; We are done whenever a score reaches 0
+	dec [hl]
+	jr z, .PickLowestScoreMoves
+
+	; If we just decremented the fourth move's score, go back to the first move
+	inc hl
+	dec c
+	jr z, .DecrementScores
+
+	jr .DecrementNextScore
+
+; In order to avoid bias towards the moves located first in memory, increment the scores
+; that were decremented one more time than the rest (in case there was a tie).
+; This means that the minimum score will be 1.
+.PickLowestScoreMoves:
+	ld a, c
+
+.move_loop
+	inc [hl]
+	dec hl
+	inc a
+	cp NUM_MOVES + 1
+	jr nz, .move_loop
+
+	ld hl, Buffer1
+	ld de, EnemyMonMoves
+	ld c, NUM_MOVES
+
+; Give a score of 0 to a blank move
+.loop2
+	ld a, [de]
+	and a
+	jr nz, .skip_load
+	ld [hl], a
+
+; Disregard the move if its score is not 1
+.skip_load
+	ld a, [hl]
+	dec a
+	jr z, .keep
+	xor a
+	ld [hli], a
+	jr .after_toss
+
+.keep
+	ld a, [de]
+	ld [hli], a
+.after_toss
+	inc de
+	dec c
+	jr nz, .loop2
+
+; Randomly choose one of the moves with a score of 1
+.ChooseMove:
+	ld hl, Buffer1
+	call Random
+	and 3
+	ld c, a
+	ld b, 0
+	add hl, bc
+	ld a, [hl]
+	and a
+	jr z, .ChooseMove
+
+	ld [CurEnemyMove], a
+	ld a, c
+	ld [CurEnemyMoveNum], a
+	ret
+; 441af
+
+
+AIScoringPointers: ; 441af
+	dw AI_Basic
+	dw AI_Setup
+	dw AI_Types
+	dw AI_Offensive
+	dw AI_Smart
+	dw AI_Opportunist
+	dw AI_Aggressive
+	dw AI_Cautious
+	dw AI_Status
+	dw AI_Risky
+	dw AI_None
+	dw AI_None
+	dw AI_None
+	dw AI_None
+	dw AI_None
+	dw AI_None
+; 441cf