Rock Paper Scissors command line

rock-paper-scissors.py

# inspiration for this code came from here https://daniel.lawrence.lu/programming/rps/
import random

def selectBest(s):
  return [i for i in range(len(s)) if max(s) == s[i]][0]

def selectBestDict(s):
  ew = {i:s[beatedBy[beatedBy[i]]] - s[beatedBy[i]] for i in s.keys()};
  return [i for i in ew.keys() if max(ew.values()) == ew[i]][0]

vals = 'R P S'.split(' ')
msg_win = "You win this round."
msg_lose = "You lose this round."
msg_tie = "Tie! You both picked "
help = "Enter R for rock, P for paper, S for scissors, H for help, or Q to quit."

history = []
moves = ["R","P","S"]
beatedBy = {"R":"P", "P":"S", "S":"R"}
result = {"R":{"R":0, "P":-1, "S":1}, "P":{"R":1, "P":0, "S":-1}, "S":{"R":-1, "P":1, "S":0}}

alpha = 0.01

# setup
markov_orders = [0,1,2,3,4,5,6]
historyCount = {}
M = 1 + 6 * len(markov_orders)

weight = [1] * M
decay = [0.85] * M

score = [0] * M
scoreboard = {"player": 0, "ai": 0}
selected = [0] * M
move = [random.choice(moves) for i in range(M)]
last = None

def loop():
    global weight
    global last

    print("Rock Paper Scissors")
    print("-------------------")

    while 1:

        # get input
        userval = input("Enter Your Move: ")
        userval = userval.upper()


        if userval == 'Q':
            print("Final Score: player %s to ai %s" % (scoreboard["player"], scoreboard["ai"]))
            if scoreboard["player"] > scoreboard["ai"]:
                print("You win!")
            elif scoreboard["player"] < scoreboard["ai"]:
                print("You Lose!")
            else:
                print("You tied.")
            break
        elif userval == 'H':
            print(help)
        elif userval in vals:
            output, weight = compute_move(userval, history, score, weight)
            print(f"AI guessed......{output}")
            scoreboard["ai"] += result[output][userval]
            scoreboard["player"] += result[userval][output]
            print(scoreboard)

        else:
            print("%s is not a valid command." % userval)

def compute_move(user_input, history, score, weight):
    global last
    score = [decay[i] * score[i] + weight[i] * result[move[i]][user_input] for i in range(M)]
    weight = [weight[i] + alpha * result[move[i]][user_input] for i in range(M)]

    index = 0
    # random optimal
    move[index] = random.choice(moves)
    # adjust random optimal score to zero
    score[index] = 0
    index += 1

    first_meta_index = index

    # markov with meta strategies
    for m in markov_orders:
        if len(history) > m:
            key = tuple(history[-m - 1:-1])
            if not (key in historyCount):
                historyCount[key] = [{"R": 0, "P": 0, "S": 0}, {"R": 0, "P": 0, "S": 0}]
            historyCount[key][0][history[-1][0]] += 1
            historyCount[key][1][history[-1][1]] += 1

    for m in markov_orders:
        if len(history) >= m:
            key = tuple(history[-m:])
            if key in historyCount:
                move[index] = selectBestDict(historyCount[key][0])
                move[index + 3] = selectBestDict(historyCount[key][1])
            else:
                move[index] = random.choice(moves)
                move[index + 3] = random.choice(moves)
        else:
            move[index] = random.choice(moves)
            move[index + 3] = random.choice(moves)
        index += 6

    # set other meta strategies
    for i in range(first_meta_index, M, 3):
        move[i + 1] = beatedBy[move[i]]
        move[i + 2] = beatedBy[move[i + 1]]

    best = selectBest(score)
    selected[best] += 1
    output = move[best]
    last = output
    history += [(last, user_input)]

    return output, weight

if __name__ == '__main__':
    loop()