def get_legal_actions(self):
self.play_treasures()
coins = self.coins
cards_purchasable = dominion_utils.get_purchaseable_cards(self, coins, self.kingdom)
return cards_purchasable
def buy_phase(self):
assert self.num_buys >= 0
if self.num_buys == 0:
# "None" is not pass - to skip a buy would be Card(-1)
return None
self.play_treasures()
purchaseable_cards = dominion_utils.get_purchaseable_cards(
self, self.coins, self.kingdom)
max_action = None
max_action_val = float("-inf")
for e in purchaseable_cards:
state_features = dominion_utils.state_features(
self, self.kingdom, e)
action_val = self.model(state_features)
print(e.name, action_val.item())
if action_val > max_action_val:
max_action = e
max_action_val = action_val
assert max_action is not None
# max_action = dominion_utils.force_buy(9, self, max_action)
dominion_utils.buy_card(self, max_action, self.kingdom)
return max_action
def gain_card_up_to(self, limit):
# todo use buy policy to choose which to gain
# todo eventually make this its own policy? or at least update q values based on this choice
# todo make this a dominion_util function we can call in both dominion_agent.py and agent.py
gainable = dominion_utils.get_purchaseable_cards(self, limit, self.kingdom, True)
card_to_gain = random.choice(gainable)
dominion_utils.gain_card(self, card_to_gain, self.kingdom)
def buy_phase(self):
self.play_treasures()
coins = self.coins
cards_purchasable = dominion_utils.get_purchaseable_cards(
self, coins, self.kingdom)
bought_card = random.choice(cards_purchasable)
dominion_utils.buy_card(self, bought_card, self.kingdom)
def get_purchaseable_cards():
global kingdom, person, agent
person.play_treasures()
purchaseable_cards = dominion_utils.get_purchaseable_cards(
person, person.coins, kingdom)
purchaseable_cards_data = dominion_utils.serialize_cards(
purchaseable_cards)
data = {
"purchaseable": purchaseable_cards_data,
"in_play": dominion_utils.serialize_cards(person.in_play),
"hand": dominion_utils.serialize_cards(person.hand),
"num_coins": person.coins
}
resp = Response(json.dumps(data), status=200, mimetype='application/json')
return resp
def __init__(self, type, person, props=None):
self.gain_card_limit = None
self.gainable_cards = []
self.new_handsize = None
self.reminaing = 0
self.type = type
if type == 1:
# Gain card up to
self.gain_card_limit = props["gain_card_limit"]
self.gainable_cards = dominion_utils.get_purchaseable_cards(
person, self.gain_card_limit, person.kingdom, True)
self.remaining = 1
# todo add None option
elif type == 2:
# Discard down to
self.new_handsize = props["handsize"]
self.remaining = len(person.hand) - self.new_handsize
def gain_card_up_to(self, limit):
gainable = dominion_utils.get_purchaseable_cards(
self, limit, self.kingdom, True)
card_to_gain = random.choice(gainable)
dominion_utils.gain_card(self, card_to_gain, self.kingdom)
