synth-ai 0.2.4.dev3__py3-none-any.whl → 0.2.4.dev5__py3-none-any.whl

synth_ai/environments/examples/red/engine_helpers/reward_library/exploration_rewards.py

@@ -0,0 +1,330 @@
+"""
+Exploration & Discovery Reward Components
+
+Rewards for map exploration, item discovery, and world interaction.
+"""
+
+from synth_ai.environments.environment.rewards.core import RewardComponent
+from typing import Dict, Any, Set
+
+
+class NewAreaDiscoveryReward(RewardComponent):
+    """Reward for entering a new map/area for the first time - +10 points"""
+
+    def __init__(self):
+        self.discovered_areas: Set[int] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        current_map = state["map_id"]
+        if current_map not in self.discovered_areas:
+            self.discovered_areas.add(current_map)
+            return 10.0
+        return 0.0
+
+
+class AreaCompletionReward(RewardComponent):
+    """Reward for visiting all accessible tiles in an area - +5 points"""
+
+    def __init__(self):
+        self.area_tiles: Dict[int, Set[tuple]] = {}
+        self.completed_areas: Set[int] = set()
+        # These would be loaded from game data in a real implementation
+        self.area_tile_counts = {
+            0: 25,  # Pallet Town
+            1: 15,  # House interior
+            3: 20,  # Oak's Lab
+            # Add more as needed
+        }
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        current_map = state["map_id"]
+        if current_map in self.completed_areas:
+            return 0.0
+
+        # Track tiles visited in this area
+        if current_map not in self.area_tiles:
+            self.area_tiles[current_map] = set()
+
+        tile = (state["player_x"], state["player_y"])
+        self.area_tiles[current_map].add(tile)
+
+        # Check if area is complete
+        required_tiles = self.area_tile_counts.get(current_map, 999)
+        if len(self.area_tiles[current_map]) >= required_tiles:
+            self.completed_areas.add(current_map)
+            return 5.0
+        return 0.0
+
+
+class RouteCompletionReward(RewardComponent):
+    """Reward for fully exploring a route - +15 points"""
+
+    def __init__(self):
+        self.completed_routes: Set[int] = set()
+        self.route_progress: Dict[int, Dict[str, bool]] = {}
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        current_map = state["map_id"]
+
+        # Check if this is a route (routes typically have IDs in a certain range)
+        if current_map < 10 or current_map in self.completed_routes:
+            return 0.0
+
+        # Track route exploration elements
+        if current_map not in self.route_progress:
+            self.route_progress[current_map] = {
+                "all_grass_visited": False,
+                "all_trainers_fought": False,
+                "all_items_found": False,
+            }
+
+        # This is simplified - would need actual game data
+        # For now, just reward first full exploration
+        if current_map not in self.completed_routes:
+            self.completed_routes.add(current_map)
+            return 15.0
+        return 0.0
+
+
+class BuildingEntryReward(RewardComponent):
+    """Reward for entering buildings - +3 points"""
+
+    def __init__(self):
+        self.buildings_entered: Set[int] = set()
+        # Building map IDs (would be loaded from game data)
+        self.building_maps = {3, 4, 5, 6, 7, 8, 9, 10}  # Example building IDs
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        current_map = state["map_id"]
+        prev_map = action.get("prev_map_id", -1)
+
+        # Entering a building for the first time
+        if (
+            current_map in self.building_maps
+            and current_map not in self.buildings_entered
+            and prev_map != current_map
+        ):
+            self.buildings_entered.add(current_map)
+            return 3.0
+        return 0.0
+
+
+class HiddenAreaDiscoveryReward(RewardComponent):
+    """Reward for finding secret areas - +20 points"""
+
+    def __init__(self):
+        self.hidden_areas_found: Set[int] = set()
+        # Hidden area map IDs (would be loaded from game data)
+        self.hidden_areas = {50, 51, 52}  # Example hidden area IDs
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        current_map = state["map_id"]
+
+        if current_map in self.hidden_areas and current_map not in self.hidden_areas_found:
+            self.hidden_areas_found.add(current_map)
+            return 20.0
+        return 0.0
+
+
+class HiddenItemFoundReward(RewardComponent):
+    """Reward for finding hidden items - +5 points"""
+
+    def __init__(self):
+        self.hidden_items_found: Set[tuple] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        # Detect item acquisition at specific locations
+        prev_inventory_count = len(action.get("prev_inventory", []))
+        current_inventory_count = len(state.get("inventory", []))
+
+        if current_inventory_count > prev_inventory_count:
+            # Item was acquired
+            item_location = (state["player_x"], state["player_y"], state["map_id"])
+            if item_location not in self.hidden_items_found:
+                # Check if this is a hidden item location (would use game data)
+                # For now, use heuristic based on position
+                if self._is_hidden_item_location(item_location):
+                    self.hidden_items_found.add(item_location)
+                    return 5.0
+        return 0.0
+
+    def _is_hidden_item_location(self, location: tuple) -> bool:
+        """Check if location is known to have hidden items"""
+        # This would be loaded from game data
+        hidden_locations = {
+            (3, 5, 0),  # Example hidden item in Pallet Town
+            (7, 2, 1),  # Example hidden item in route
+        }
+        return location in hidden_locations
+
+
+class FirstItemOfTypeReward(RewardComponent):
+    """Reward for finding first item of each type - +10 points"""
+
+    def __init__(self):
+        self.item_types_found: Set[str] = set()
+        # Item type mappings (would be loaded from game data)
+        self.item_types = {
+            1: "pokeball",
+            2: "pokeball",
+            3: "pokeball",
+            10: "potion",
+            11: "potion",
+            12: "potion",
+            20: "tm",
+            21: "tm",
+            22: "tm",
+            # Add more mappings
+        }
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        prev_inventory = action.get("prev_inventory", [])
+        current_inventory = state.get("inventory", [])
+
+        # Check for new items
+        prev_item_ids = {item.get("item_id") for item in prev_inventory}
+        current_item_ids = {item.get("item_id") for item in current_inventory}
+        new_items = current_item_ids - prev_item_ids
+
+        total_reward = 0.0
+        for item_id in new_items:
+            item_type = self.item_types.get(item_id, "unknown")
+            if item_type not in self.item_types_found and item_type != "unknown":
+                self.item_types_found.add(item_type)
+                total_reward += 10.0
+
+        return total_reward
+
+
+class RareItemDiscoveryReward(RewardComponent):
+    """Reward for finding rare items - +25 points"""
+
+    def __init__(self):
+        self.rare_items_found: Set[int] = set()
+        # Rare item IDs (would be loaded from game data)
+        self.rare_items = {1, 50, 100}  # Master Ball, rare TMs, etc.
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        prev_inventory = action.get("prev_inventory", [])
+        current_inventory = state.get("inventory", [])
+
+        # Check for new rare items
+        prev_item_ids = {item.get("item_id") for item in prev_inventory}
+        current_item_ids = {item.get("item_id") for item in current_inventory}
+        new_items = current_item_ids - prev_item_ids
+
+        total_reward = 0.0
+        for item_id in new_items:
+            if item_id in self.rare_items and item_id not in self.rare_items_found:
+                self.rare_items_found.add(item_id)
+                total_reward += 25.0
+
+        return total_reward
+
+
+class KeyItemAcquisitionReward(RewardComponent):
+    """Reward for obtaining story-critical items - +30 points"""
+
+    def __init__(self):
+        self.key_items_obtained: Set[int] = set()
+        # Key item IDs (would be loaded from game data)
+        self.key_items = {200, 201, 202}  # Pokedex, Town Map, etc.
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        prev_inventory = action.get("prev_inventory", [])
+        current_inventory = state.get("inventory", [])
+
+        # Check for new key items
+        prev_item_ids = {item.get("item_id") for item in prev_inventory}
+        current_item_ids = {item.get("item_id") for item in current_inventory}
+        new_items = current_item_ids - prev_item_ids
+
+        total_reward = 0.0
+        for item_id in new_items:
+            if item_id in self.key_items and item_id not in self.key_items_obtained:
+                self.key_items_obtained.add(item_id)
+                total_reward += 30.0
+
+        return total_reward
+
+
+class FirstWarpUsageReward(RewardComponent):
+    """Reward for using doors, cave entrances, etc. for first time - +5 points"""
+
+    def __init__(self):
+        self.warp_types_used: Set[str] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        prev_map = action.get("prev_map_id", -1)
+        current_map = state["map_id"]
+
+        if prev_map != current_map:
+            # Determine warp type based on map transition
+            warp_type = self._get_warp_type(prev_map, current_map)
+            if warp_type and warp_type not in self.warp_types_used:
+                self.warp_types_used.add(warp_type)
+                return 5.0
+        return 0.0
+
+    def _get_warp_type(self, prev_map: int, current_map: int) -> str:
+        """Determine type of warp based on map transition"""
+        # This would use game data to classify warps
+        if prev_map == 0 and current_map > 0:
+            return "door"
+        elif prev_map > 10 and current_map > 10:
+            return "cave"
+        elif abs(prev_map - current_map) == 1:
+            return "route_transition"
+        return ""
+
+
+class PCUsageReward(RewardComponent):
+    """Reward for first time using Pokemon PC storage - +10 points"""
+
+    def __init__(self):
+        self.pc_used = False
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        if self.pc_used:
+            return 0.0
+
+        # Detect PC usage (would need menu state tracking)
+        # Simplified: assume PC is used if in Pokemon Center and menu state changes
+        if state["map_id"] in [4, 8, 12] and state.get("menu_state", 0) > 0:  # Pokemon Centers
+            # This is a placeholder - would need actual PC detection
+            if not action.get("prev_menu_state", 0):
+                self.pc_used = True
+                return 10.0
+        return 0.0
+
+
+class VendingMachineReward(RewardComponent):
+    """Reward for discovering and using vending machines - +5 points"""
+
+    def __init__(self):
+        self.vending_machines_used: Set[tuple] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        # Detect vending machine usage
+        prev_money = action.get("prev_money", 0)
+        current_money = state.get("money", 0)
+        prev_inventory_count = len(action.get("prev_inventory", []))
+        current_inventory_count = len(state.get("inventory", []))
+
+        # Money decreased and items increased = purchase
+        if current_money < prev_money and current_inventory_count > prev_inventory_count:
+            # Check if at vending machine location
+            vending_location = (state["player_x"], state["player_y"], state["map_id"])
+            if self._is_vending_machine_location(vending_location):
+                if vending_location not in self.vending_machines_used:
+                    self.vending_machines_used.add(vending_location)
+                    return 5.0
+        return 0.0
+
+    def _is_vending_machine_location(self, location: tuple) -> bool:
+        """Check if location has a vending machine"""
+        # This would be loaded from game data
+        vending_locations = {
+            (5, 3, 15),  # Example vending machine location
+        }
+        return location in vending_locations

synth_ai/environments/examples/red/engine_helpers/reward_library/novelty_rewards.py

@@ -0,0 +1,120 @@
+"""
+Novelty & Exploration Bonus Reward Components
+
+Rewards for first-time experiences and curiosity.
+"""
+
+from synth_ai.environments.environment.rewards.core import RewardComponent
+from typing import Dict, Any, Set
+
+
+class FirstBattleReward(RewardComponent):
+    """Reward for engaging in first battle - +20 points"""
+
+    def __init__(self):
+        self.first_battle = False
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        if self.first_battle:
+            return 0.0
+
+        prev_in_battle = action.get("prev_in_battle", False)
+        current_in_battle = state["in_battle"]
+
+        if not prev_in_battle and current_in_battle:
+            self.first_battle = True
+            return 20.0
+        return 0.0
+
+
+class FirstPokemonCenterVisitReward(RewardComponent):
+    """Reward for first healing - +15 points"""
+
+    def __init__(self):
+        self.first_heal = False
+        self.pokemon_center_maps = {4, 8, 12, 16}  # Pokemon Center map IDs
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        if self.first_heal:
+            return 0.0
+
+        if state["map_id"] in self.pokemon_center_maps:
+            # Check for HP restoration (simplified)
+            party = state.get("party", [])
+            for pokemon in party:
+                if pokemon.get("hp_current", 0) == pokemon.get("hp_max", 0):
+                    self.first_heal = True
+                    return 15.0
+        return 0.0
+
+
+class FirstPokemartPurchaseReward(RewardComponent):
+    """Reward for first item purchase - +10 points"""
+
+    def __init__(self):
+        self.first_purchase = False
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        if self.first_purchase:
+            return 0.0
+
+        prev_money = action.get("prev_money", 0)
+        current_money = state.get("money", 0)
+
+        if current_money < prev_money and prev_money > 0:
+            self.first_purchase = True
+            return 10.0
+        return 0.0
+
+
+class FirstSaveReward(RewardComponent):
+    """Reward for saving the game - +5 points"""
+
+    def __init__(self):
+        self.first_save = False
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        if self.first_save:
+            return 0.0
+
+        # This would need save detection
+        if state.get("game_saved", False):
+            self.first_save = True
+            return 5.0
+        return 0.0
+
+
+class MenuExplorationReward(RewardComponent):
+    """Reward for opening and exploring different menus - +3 points"""
+
+    def __init__(self):
+        self.menus_explored: Set[str] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        # This would need menu state tracking
+        # Placeholder implementation
+        return 0.0
+
+
+class ButtonDiscoveryReward(RewardComponent):
+    """Reward for discovering START menu, SELECT uses - +5 points"""
+
+    def __init__(self):
+        self.buttons_discovered: Set[str] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        # This would need button usage tracking
+        # Placeholder implementation
+        return 0.0
+
+
+class FeatureDiscoveryReward(RewardComponent):
+    """Reward for discovering PC, daycare, etc. - +10 points"""
+
+    def __init__(self):
+        self.features_discovered: Set[str] = set()
+
+    async def score(self, state: Dict[str, Any], action: Dict[str, Any]) -> float:
+        # This would need feature usage detection
+        # Placeholder implementation
+        return 0.0