super-auto-pets-board-game/internal/ai/shop.go

package ai

import (
	"math/rand/v2"
	"slices"

	"github.com/greyson/super-auto-pets-board-game/internal/game"
)

// deckHasPet reports whether a hypothetical deck still contains at least one
// pet. A deck of only food (apples) can never field a fighter, so it is an
// automatic loss — the bot must never voluntarily sell or trade its way there.
func deckHasPet(deck []game.Card) bool {
	for _, c := range deck {
		if c.IsPet() {
			return true
		}
	}
	return false
}

// applyTemplateShopEffects mirrors the engine's shop-time triggers on a
// hypothetical deck: buying an Otter really does come with an apple, and the
// bot should value that. Only deck-changing effects matter here (coin
// refunds don't alter the deck being scored).
func (cx *ctx) applyTemplateShopEffects(deck []game.Card, c game.Card, trigger game.EffectTrigger) []game.Card {
	for _, e := range c.Effects {
		if e.Trigger != trigger || cx.v.Round < e.MinRound {
			continue
		}
		switch e.Action {
		case game.ActionGainApple:
			n := max(e.Count, 1)
			switch e.Per {
			case game.PerShopFaintPets:
				n *= countShopFaintPets(cx.v.ShopRow)
			case game.PerBuysThisRound:
				// This buy will bump the counter, so count it (Blue-Ringed Octopus).
				n *= cx.me.BuysThisRound + 1
			}
			for range n {
				deck = append(deck, cx.simApple())
			}
		case game.ActionRevealForApples:
			// Cockatoo: the bot would reveal its highest-power other pet.
			best := 0
			for _, d := range deck {
				if d.IsPet() && d.ID != c.ID && d.Power > best {
					best = d.Power
				}
			}
			for range best {
				deck = append(deck, cx.simApple())
			}
		case game.ActionRevealTierForApples:
			// Quetzalcoatl: a fixed apple reward if a low-tier pet can be revealed.
			eligible := false
			for _, d := range deck {
				if d.IsPet() && d.ID != c.ID && (e.Cap <= 0 || d.Tier <= e.Cap) {
					eligible = true
					break
				}
			}
			if eligible {
				for range max(e.Count, 1) {
					deck = append(deck, cx.simApple())
				}
			}
		case game.ActionDoubleApples:
			apples := 0
			for _, dc := range deck {
				if dc.Food == game.FoodApple {
					apples++
				}
			}
			for range apples {
				deck = append(deck, cx.simApple())
			}
		case game.ActionApplesInPlay:
			// Golden pack (Hercules Beetle, sold): in-play apples aren't in the
			// deck, but they buff the front pet, so approximate them as deck
			// apples for scoring. MinRound is already gated above.
			for range max(e.Count, 1) {
				deck = append(deck, cx.simApple())
			}
		case game.ActionBuyTopFree:
			// Golden pack (Stoat): grabs an unknown top-of-deck card; approximate
			// with a sampled card of the current tier.
			pool := cx.unseenPool(cx.v.Round)
			if len(pool) == 0 {
				pool = game.TierContentsForPacks(cx.v.Packs, cx.v.Round)
			}
			if len(pool) > 0 {
				rc := pool[rand.IntN(len(pool))]
				rc.ID = cx.nextSimID()
				deck = append(deck, rc)
			}
		case game.ActionSetAside:
			// Golden pack (Avocado): the just-bought token is set aside, not
			// kept in the deck being scored.
			if i := slices.IndexFunc(deck, func(d game.Card) bool { return d.ID == c.ID }); i >= 0 {
				deck = slices.Delete(deck, i, i+1)
			}
		case game.ActionUpgradeNextTier:
			// Unicorn pack (Water of Youth): discard this food and the lowest-value
			// pet, then gain a sampled top card of the next tier for free.
			nextTier := cx.v.Round + 1
			if nextTier > game.MaxRounds {
				continue
			}
			if i := slices.IndexFunc(deck, func(d game.Card) bool { return d.ID == c.ID }); i >= 0 {
				deck = slices.Delete(deck, i, i+1)
			}
			worst, worstVal := -1, 0.0
			for i, d := range deck {
				if !d.IsPet() {
					continue
				}
				if v := keepValue(d); worst < 0 || v < worstVal {
					worst, worstVal = i, v
				}
			}
			if worst < 0 {
				continue // no pet to sacrifice; the food is wasted
			}
			deck = slices.Delete(deck, worst, worst+1)
			pool := cx.unseenPool(nextTier)
			if len(pool) == 0 {
				pool = game.TierContentsForPacks(cx.v.Packs, nextTier)
			}
			if len(pool) > 0 {
				rc := pool[rand.IntN(len(pool))]
				rc.ID = cx.nextSimID()
				deck = append(deck, rc)
				deck = cx.applyTemplateShopEffects(deck, rc, game.TriggerBuy)
			}
		}
	}
	return deck
}

// countShopFaintPets counts pets in the shop row with a Faint effect (mirrors
// the engine's Opossum payout).
func countShopFaintPets(row []game.Card) int {
	n := 0
	for _, c := range row {
		if c.ID == "" || !c.IsPet() {
			continue
		}
		for _, e := range c.Effects {
			if e.Trigger == game.TriggerFaint {
				n++
				break
			}
		}
	}
	return n
}

func (cx *ctx) simApple() game.Card {
	return game.Card{
		ID:        cx.nextSimID(),
		Kind:      game.KindFood,
		Name:      "Apple",
		Food:      game.FoodApple,
		Temporary: true,
	}
}

// previewSellDown applies the sell-down a pass would eventually force onto a
// hypothetical deck: while over the pet limit, the lowest-value pet is sold
// for an apple. This lets the bot buy a sixth pet on purpose, knowing what
// it will cost.
func (cx *ctx) previewSellDown(deck []game.Card) []game.Card {
	for {
		pets := 0
		worst, worstVal := -1, 0.0
		for i, c := range deck {
			if !c.IsPet() {
				continue
			}
			pets++
			if v := keepValue(c); worst < 0 || v < worstVal {
				worst, worstVal = i, v
			}
		}
		if pets <= cx.v.MaxPets || worst < 0 {
			return deck
		}
		sold := deck[worst]
		deck = slices.Delete(deck, worst, worst+1)
		deck = append(deck, cx.simApple())
		deck = cx.applyTemplateShopEffects(deck, sold, game.TriggerSell)
	}
}

// score fills in every candidate's score: a weighted blend of the estimated
// next-battle win chance (deck arranged by the book ordering — the full
// ordering search happens later, at arrange time) and the deck's future
// value. All candidates face the same opponent guesses.
func (b *Bot) score(cx *ctx, cands []candidate) {
	oppSamples, simsPer := b.budget()
	oppDecks := cx.oppArrangements(oppSamples)
	alpha := immediateWeight(cx.v)
	for i := range cands {
		total := 0.0
		for _, deck := range cands[i].decks {
			imm := cx.winProb(heuristicOrder(deck, 0), oppDecks, simsPer)
			fut := normFuture(deckValue(deck, cx.v.Round, cx.v.MaxRounds))
			total += alpha*imm + (1-alpha)*fut
		}
		cands[i].score = total/float64(len(cands[i].decks)) + cands[i].bias
	}
}

// decideShop picks one shop action: buy a row card, sell some own cards,
// trade in a suit triple, or pass. Only buying costs gold; selling and
// trading are free, and passing (the only way to end the round's shopping)
// is legal only at or under the pet limit.
func (b *Bot) decideShop(v *game.View, mem *Memory) *Action {
	cx := newCtx(v, mem)
	deck := cx.me.Deck
	var cands []candidate

	// Passing ends the bot's shopping for the round; it is the baseline every
	// other option must beat, nudged down while unspent coins remain because
	// spending them is usually right. Illegal over the pet limit — the sell
	// and trade candidates below always exist then, so the bot works its way
	// back under.
	if cx.me.PetCount <= v.MaxPets {
		bias := 0.0
		if cx.me.Coins > 0 {
			bias = -0.02
		}
		cands = append(cands, candidate{
			act:   &Action{Type: "pass"},
			decks: [][]game.Card{slices.Clone(deck)},
			bias:  bias,
		})
	}

	if cx.me.Coins > 0 {
		for i, c := range v.ShopRow {
			if c.ID == "" {
				continue
			}
			nd := append(slices.Clone(deck), c)
			nd = cx.applyTemplateShopEffects(nd, c, game.TriggerBuy)
			nd = cx.previewSellDown(nd)
			cands = append(cands, candidate{
				act:   &Action{Type: "buy", Row: i},
				decks: [][]game.Card{nd},
			})
		}
	}

	// Golden pack: buying by discarding an Avocado yields the same deck as a
	// coin buy, so it's only preferred when coins are scarce — a small negative
	// bias keeps the token in reserve otherwise.
	if cx.me.Avocados > 0 {
		for i, c := range v.ShopRow {
			if c.ID == "" {
				continue
			}
			nd := append(slices.Clone(deck), c)
			nd = cx.applyTemplateShopEffects(nd, c, game.TriggerBuy)
			nd = cx.previewSellDown(nd)
			cands = append(cands, candidate{
				act:   &Action{Type: "buyAvocado", Row: i},
				decks: [][]game.Card{nd},
				bias:  -0.05,
			})
		}
	}

	// Sell candidates: the worst 1, 2, or 3 keepers. Selling is free and
	// takes any number of cards, so bulk-dumping junk before a battle is one
	// action. Temporary cards are excluded — selling an apple for an apple
	// does nothing.
	sellable := slices.Clone(deck)
	sellable = slices.DeleteFunc(sellable, func(c game.Card) bool { return c.Temporary })
	slices.SortStableFunc(sellable, func(a, b game.Card) int {
		av, bv := keepValue(a), keepValue(b)
		switch {
		case av < bv:
			return -1
		case av > bv:
			return 1
		}
		return 0
	})
	// Selling a pet permanently trades a body for a temporary apple — the
	// apple's one-battle buff shows up in the rollout, but it vanishes next
	// round, so the persistent value of the pet is simply destroyed. The
	// future-value term barely registers this (normFuture squashes small deck
	// swings), which once let the bot treat "sell my worst pet" as free and
	// bleed its board down to a single pet over successive shop turns. This
	// explicit penalty prices the loss back in: it scales with the persistent
	// worth of the pets sold and with how much the future still matters
	// (1-alpha), so it bites hardest early and fades to nothing in the final
	// round, where selling for a decisive last battle is a legitimate play the
	// rollout can judge on its own.
	futureWeight := 1 - immediateWeight(v)
	for k := 1; k <= min(3, len(sellable)); k++ {
		ids := make([]string, 0, k)
		nd := slices.Clone(deck)
		petValueSold := 0.0
		for _, s := range sellable[:k] {
			ids = append(ids, s.ID)
			if s.IsPet() {
				petValueSold += keepValue(s)
			}
			idx := slices.IndexFunc(nd, func(c game.Card) bool { return c.ID == s.ID })
			nd = slices.Delete(nd, idx, idx+1)
			nd = append(nd, cx.simApple())
			nd = cx.applyTemplateShopEffects(nd, s, game.TriggerSell)
		}
		// Never dump the last pet: an all-food deck loses on sight, so this
		// candidate is off the table no matter how the rollouts score.
		if !deckHasPet(nd) {
			continue
		}
		cands = append(cands, candidate{
			act:   &Action{Type: "sell", Cards: ids},
			decks: [][]game.Card{nd},
			bias:  -sellPetPenalty * futureWeight * petValueSold,
		})
	}

	// Trade candidates: for each suit with three or more pets, trade the
	// three lowest-value ones. The reward card is unknown (top two of the
	// next tier's deck), so each trade is scored across several sampled
	// rewards.
	if v.Round < v.MaxRounds && v.Round < len(v.DeckCounts) && v.DeckCounts[v.Round] >= 2 {
		bySuit := map[game.Suit][]game.Card{}
		for _, c := range deck {
			if c.IsPet() && c.Suit != "" {
				bySuit[c.Suit] = append(bySuit[c.Suit], c)
			}
		}
		for _, pets := range bySuit {
			if len(pets) < game.TradeInCount {
				continue
			}
			slices.SortStableFunc(pets, func(a, b game.Card) int {
				av, bv := keepValue(a), keepValue(b)
				switch {
				case av < bv:
					return -1
				case av > bv:
					return 1
				}
				return 0
			})
			trio := pets[:game.TradeInCount]
			base := slices.Clone(deck)
			ids := make([]string, 0, game.TradeInCount)
			for _, t := range trio {
				ids = append(ids, t.ID)
				idx := slices.IndexFunc(base, func(c game.Card) bool { return c.ID == t.ID })
				base = slices.Delete(base, idx, idx+1)
				base = cx.applyTemplateShopEffects(base, t, game.TriggerTriple)
			}
			// Never bet the whole board on the trade: the reward is the top of
			// the next tier's deck, which can be a food card (or nothing, if the
			// deck is spent), so trading away the last pets risks an all-food,
			// auto-losing deck. Mirror the sell guard and skip such a trade.
			if !deckHasPet(base) {
				continue
			}
			pool := cx.unseenPool(v.Round + 1)
			if len(pool) == 0 {
				pool = game.TierContentsForPacks(v.Packs, v.Round+1)
			}
			var decks [][]game.Card
			for range 3 {
				reward := pool[rand.IntN(len(pool))]
				reward.ID = cx.nextSimID()
				nd := append(slices.Clone(base), reward)
				nd = cx.applyTemplateShopEffects(nd, reward, game.TriggerBuy)
				nd = cx.previewSellDown(nd)
				decks = append(decks, nd)
			}
			cands = append(cands, candidate{
				act:   &Action{Type: "trade", Cards: ids},
				decks: decks,
			})
		}
	}

	b.score(cx, cands)
	return b.pick(cands).act
}

// decideTradeChoose resolves the bot's own pending trade: score keeping
// either revealed card and pick.
func (b *Bot) decideTradeChoose(v *game.View, mem *Memory) *Action {
	cx := newCtx(v, mem)
	var cands []candidate
	for pick, c := range v.Pending.Options {
		nd := append(slices.Clone(cx.me.Deck), c)
		nd = cx.applyTemplateShopEffects(nd, c, game.TriggerBuy)
		nd = cx.previewSellDown(nd)
		cands = append(cands, candidate{
			act:   &Action{Type: "tradeChoose", Pick: pick},
			decks: [][]game.Card{nd},
		})
	}
	b.score(cx, cands)
	return b.pick(cands).act
}