custom_bot.py

"""
custom_bot.py
standard_bot.py reimplemented as an import from a clashroyalebuildabot install
"""

import random
import time

from clashroyalebuildabot.bot import Bot
from custom_action import CustomAction
from clashroyalebuildabot.data.constants import DISPLAY_WIDTH, SCREENSHOT_WIDTH, DISPLAY_HEIGHT, SCREENSHOT_HEIGHT


class CustomBot(Bot):
    def __init__(self, card_names, debug=False):
        card_names = ('minions', 'archers', 'arrows', 'giant', 'minipekka', 'fireball', 'knight', 'musketeer')
        preset_deck = {'minions', 'archers', 'arrows', 'giant', 'minipekka', 'fireball', 'knight', 'musketeer'}
        if set(card_names) != preset_deck:
            raise ValueError(f'You must use the preset deck with cards {preset_deck} for CustomBot')
        super().__init__(card_names, CustomAction, debug=debug)

    def _preprocess(self):
        """
        Perform preprocessing on the state

        Estimate the tile of each unit to be the bottom of their bounding box
        """
        for side in ['ally', 'enemy']:
            for k, v in self.state['units'][side].items():
                for unit in v['positions']:
                    bbox = unit['bounding_box']
                    bbox[0] *= DISPLAY_WIDTH / SCREENSHOT_WIDTH
                    bbox[1] *= DISPLAY_HEIGHT / SCREENSHOT_HEIGHT
                    bbox[2] *= DISPLAY_WIDTH / SCREENSHOT_WIDTH
                    bbox[3] *= DISPLAY_HEIGHT / SCREENSHOT_HEIGHT
                    bbox_bottom = [((bbox[0] + bbox[2]) / 2), bbox[3]]
                    unit['tile_xy'] = self._get_nearest_tile(*bbox_bottom)

    def run(self):
        print("Custom Bot is now running...")
        while True:
            # Set the state of the game
            self.set_state()
            # Obtain a list of playable actions
            actions = self.get_actions()
            if actions:
                # Shuffle the actions (because action scores might be the same)
                random.shuffle(actions)
                # Preprocessing
                self._preprocess()
                # Get the best action
                action = max(actions, key=lambda x: x.calculate_score(self.state))
                # Skip the action if it doesn't score high enough
                if action.score[0] == 0:
                    continue
                # Play the best action
                self.play_action(action)
                # Log the result
                print(f'Playing {action} with score {action.score} and sleeping for 1 second')
                time.sleep(1.0)