2023-03-02 20:16:26 +01:00
|
|
|
import json
|
|
|
|
from enum import Enum
|
|
|
|
from typing import List, Optional
|
2023-03-02 21:13:25 +01:00
|
|
|
import more_itertools
|
2023-03-11 20:00:56 +01:00
|
|
|
from variants import variant_id, variant_name
|
2023-03-02 20:16:26 +01:00
|
|
|
|
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
BASE62 = "abcdefghijklmnopqrstuvwxyz0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ";
|
2023-03-03 11:17:31 +01:00
|
|
|
COLORS = 'rygbpt'
|
2023-03-02 20:16:26 +01:00
|
|
|
|
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
# Helper method, iterate over chunks of length n in a string
|
2023-03-02 20:38:48 +01:00
|
|
|
def chunks(s: str, n: int):
|
|
|
|
for i in range(0, len(s), n):
|
|
|
|
yield s[i:i+n]
|
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
|
2023-03-02 20:16:26 +01:00
|
|
|
class DeckCard():
|
|
|
|
def __init__(self, suitIndex: int, rank: int):
|
|
|
|
self.suitIndex: int = suitIndex
|
|
|
|
self.rank: int = rank
|
|
|
|
|
2023-03-02 21:13:25 +01:00
|
|
|
@staticmethod
|
|
|
|
def from_json(deck_card):
|
|
|
|
return DeckCard(**deck_card)
|
|
|
|
|
2023-03-02 20:16:26 +01:00
|
|
|
def __eq__(self, other):
|
|
|
|
return self.suitIndex == other.suitIndex and self.rank == other.rank
|
|
|
|
|
|
|
|
def __repr__(self):
|
|
|
|
return COLORS[self.suitIndex] + str(self.rank)
|
|
|
|
|
|
|
|
|
|
|
|
class ActionType(Enum):
|
|
|
|
Play = 0
|
|
|
|
Discard = 1
|
|
|
|
ColorClue = 2
|
|
|
|
RankClue = 3
|
|
|
|
EndGame = 4
|
2023-03-14 09:12:46 +01:00
|
|
|
VoteTerminate = 5 ## hack: online, this is encoded as a 10
|
2023-03-02 20:16:26 +01:00
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
|
2023-03-02 20:16:26 +01:00
|
|
|
class Action():
|
|
|
|
def __init__(self, type_: ActionType, target: int, value: Optional[int] = None):
|
|
|
|
self.type = type_
|
|
|
|
self.target = target
|
|
|
|
self.value = value
|
|
|
|
|
2023-03-02 21:13:25 +01:00
|
|
|
@staticmethod
|
|
|
|
def from_json(action):
|
|
|
|
return Action(
|
|
|
|
ActionType(action['type']),
|
|
|
|
int(action['target']),
|
|
|
|
action.get('value', None)
|
|
|
|
)
|
|
|
|
|
2023-03-02 20:38:48 +01:00
|
|
|
def __repr__(self):
|
|
|
|
match self.type:
|
|
|
|
case ActionType.Play:
|
|
|
|
return "Play card {}".format(self.target)
|
|
|
|
case ActionType.Discard:
|
|
|
|
return "Discard card {}".format(self.target)
|
|
|
|
case ActionType.ColorClue:
|
|
|
|
return "Clue color {} to player {}".format(self.value, self.target)
|
2023-03-02 22:15:47 +01:00
|
|
|
case ActionType.RankClue:
|
2023-03-02 20:38:48 +01:00
|
|
|
return "Clue rank {} to player {}".format(self.value, self.target)
|
|
|
|
case ActionType.EndGame:
|
|
|
|
return "Player {} ends the game (code {})".format(self.target, self.value)
|
2023-03-02 22:15:47 +01:00
|
|
|
case ActionType.VoteTerminate:
|
2023-03-03 11:17:31 +01:00
|
|
|
return "Players vote to terminate the game (code {})".format(self.value)
|
2023-03-02 22:15:47 +01:00
|
|
|
return "Undefined action"
|
2023-03-02 20:16:26 +01:00
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
|
2023-03-14 09:12:46 +01:00
|
|
|
def compress_actions(actions: List[Action], game_id=None) -> str:
|
2023-03-02 20:16:26 +01:00
|
|
|
minType = 0
|
|
|
|
maxType = 0
|
|
|
|
if len(actions) != 0:
|
|
|
|
minType = min(map(lambda a: a.type.value, actions))
|
|
|
|
maxType = max(map(lambda a: a.type.value, actions))
|
|
|
|
typeRange = maxType - minType + 1
|
|
|
|
def compress_action(action):
|
2023-03-03 11:17:31 +01:00
|
|
|
## We encode action values with +1 to differentiate
|
|
|
|
# null (encoded 0) and 0 (encoded 1)
|
|
|
|
value = 0 if action.value is None else action.value + 1
|
2023-03-14 09:12:46 +01:00
|
|
|
if action.type == ActionType.VoteTerminate:
|
|
|
|
value = 0
|
|
|
|
with open('vote_terminate_actions.txt', 'a') as f:
|
|
|
|
f.write('target: {}, value: {}, game_id: {}\n'.format(action.target, action.value, game_id))
|
2023-03-02 20:16:26 +01:00
|
|
|
a = BASE62[typeRange * value + (action.type.value - minType)]
|
|
|
|
b = BASE62[action.target]
|
|
|
|
return a + b
|
|
|
|
out = str(minType) + str(maxType)
|
|
|
|
out += ''.join(map(compress_action, actions))
|
|
|
|
return out
|
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
|
2023-03-02 20:38:48 +01:00
|
|
|
def decompress_actions(actions_str: str) -> List[Action]:
|
|
|
|
try:
|
|
|
|
minType = int(actions_str[0])
|
|
|
|
maxType = int(actions_str[1])
|
|
|
|
except ValueError:
|
2023-03-02 21:13:25 +01:00
|
|
|
raise ValueError("invalid action string: invalid min/max range")
|
2023-03-02 20:38:48 +01:00
|
|
|
assert(maxType >= minType)
|
2023-03-02 21:13:25 +01:00
|
|
|
if not len(actions_str) % 2 == 0:
|
|
|
|
raise ValueError("Invalid length of action str")
|
2023-03-02 20:38:48 +01:00
|
|
|
typeRange = maxType - minType + 1
|
|
|
|
def decompress_action(action):
|
2023-03-02 21:27:00 +01:00
|
|
|
actionType = ActionType((BASE62.index(action[0]) % typeRange) + minType)
|
2023-03-02 20:38:48 +01:00
|
|
|
value = None
|
|
|
|
if actionType not in [actionType.Play, actionType.Discard]:
|
2023-03-03 11:17:31 +01:00
|
|
|
## We encode values with +1 to differentiate null (encoded 0) and 0 (encoded 1)
|
|
|
|
value = BASE62.index(action[0]) // typeRange - 1
|
|
|
|
if value == -1:
|
|
|
|
value = None
|
2023-03-02 20:38:48 +01:00
|
|
|
target = BASE62.index(action[1])
|
|
|
|
return Action(actionType, target, value)
|
|
|
|
return [decompress_action(a) for a in chunks(actions_str[2:], 2)]
|
|
|
|
|
2023-03-02 20:16:26 +01:00
|
|
|
|
|
|
|
def compress_deck(deck: List[DeckCard]) -> str:
|
|
|
|
assert(len(deck) != 0)
|
|
|
|
minRank = min(map(lambda c: c.rank, deck))
|
|
|
|
maxRank = max(map(lambda c: c.rank, deck))
|
|
|
|
rankRange = maxRank - minRank + 1
|
|
|
|
def compress_card(card):
|
|
|
|
return BASE62[rankRange * card.suitIndex + (card.rank - minRank)]
|
|
|
|
out = str(minRank) + str(maxRank)
|
|
|
|
out += ''.join(map(compress_card, deck))
|
|
|
|
return out
|
|
|
|
|
|
|
|
|
|
|
|
def decompress_deck(deck_str: str) -> List[DeckCard]:
|
|
|
|
assert(len(deck_str) >= 2)
|
|
|
|
minRank = int(deck_str[0])
|
|
|
|
maxRank = int(deck_str[1])
|
|
|
|
assert(maxRank >= minRank)
|
|
|
|
rankRange = maxRank - minRank + 1
|
|
|
|
def decompress_card(card_char):
|
|
|
|
index = BASE62.index(card_char)
|
|
|
|
suitIndex = index // rankRange
|
|
|
|
rank = index % rankRange + minRank
|
|
|
|
return DeckCard(suitIndex, rank)
|
|
|
|
return [decompress_card(c) for c in deck_str[2:]]
|
|
|
|
|
|
|
|
|
2023-03-02 21:13:25 +01:00
|
|
|
def compressJSONGame(game_json: dict) -> str:
|
|
|
|
out = ""
|
|
|
|
num_players = len(game_json.get('players', []))
|
|
|
|
num_players = game_json.get('num_players', num_players)
|
|
|
|
if not 2 <= num_players:
|
|
|
|
raise ValueError("Invalid JSON game: could not parse num players")
|
|
|
|
out = "{}".format(num_players)
|
|
|
|
try:
|
|
|
|
deck = game_json['deck']
|
|
|
|
except:
|
|
|
|
raise KeyError("JSON game contains no deck")
|
|
|
|
assert(len(deck) > 0)
|
|
|
|
if type(deck[0]) != DeckCard:
|
|
|
|
try:
|
|
|
|
deck = [DeckCard.from_json(card) for card in deck]
|
|
|
|
except:
|
|
|
|
raise ValueError("invalid jSON format: could not convert to deck cards")
|
|
|
|
# now, deck is in the correct form
|
|
|
|
out += compress_deck(deck)
|
|
|
|
out += "," # first part finished
|
|
|
|
actions = game_json.get('actions', [])
|
|
|
|
if len(actions) == 0:
|
|
|
|
print("WARNING: action array is empty")
|
|
|
|
else:
|
|
|
|
if type(actions[0]) != Action:
|
|
|
|
try:
|
|
|
|
actions = [Action.from_json(action) for action in actions]
|
|
|
|
except:
|
|
|
|
raise ValueError("invalid JSON format: could not convert to actions")
|
|
|
|
out += compress_actions(actions)
|
|
|
|
out += ","
|
|
|
|
variant = game_json.get("variant", "No Variant")
|
|
|
|
out += str(variant_id(variant))
|
|
|
|
return ''.join(more_itertools.intersperse("-", out, 20))
|
2023-03-02 20:16:26 +01:00
|
|
|
|
2023-03-02 21:28:37 +01:00
|
|
|
|
2023-03-02 21:27:00 +01:00
|
|
|
def decompressJSONGame(game_str: str)->dict:
|
|
|
|
game = {}
|
|
|
|
game_str = game_str.replace("-", "")
|
|
|
|
try:
|
|
|
|
[players_deck, actions, variant_id] = game_str.split(",")
|
|
|
|
except:
|
|
|
|
raise ValueError("Invalid format of compressed string!")
|
|
|
|
game['players'] = ["Alice", "Bob", "Cathy", "Donald", "Emily"][:int(players_deck[0])]
|
|
|
|
game['deck'] = decompress_deck(players_deck[1:])
|
|
|
|
game['actions'] = decompress_actions(actions)
|
|
|
|
game['options'] = {
|
|
|
|
"variant": variant_name(int(variant_id))
|
|
|
|
}
|
|
|
|
return game
|
2023-03-13 17:21:25 +01:00
|
|
|
|
|
|
|
def link(game_json: dict) -> str:
|
|
|
|
compressed = compressJSONGame(game_json)
|
|
|
|
return "https://hanab.live/replay-json/{}".format(compressed)
|