evaluate

#!/usr/bin/env python3
import argparse
import re
from subprocess import PIPE, run

import matplotlib.pyplot as plt
import numpy as np
import seaborn as sns


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("-n", help="number of times to run", type=int, default=3)
    parser.add_argument("-b", help="evaluate BentCigarFunction", action="store_true")
    parser.add_argument("-k", help="evaluate KatsuuraEvaluation", action="store_true")
    parser.add_argument("-s", help="evaluate SchaffersEvaluation", action="store_true")
    parser.add_argument("-a", help="test all functions", action="store_true")
    parser.add_argument("-p", "--plot", help="make a boxplot of the scores", action="store_true")
    args = parser.parse_args()

    functions = ["BentCigarFunction", "KatsuuraEvaluation", "SchaffersEvaluation"]

    function = None
    if args.b:
        function = functions[0]
    elif args.k:
        function = functions[1]
    elif args.s:
        function = functions[2]

    if function is not None:
        score = get_score(function, args.n)
        print("Average of {}: {}".format(args.n, score))
        return

    if args.a:
        function = functions
        bent = get_score(functions[0], args.n)
        katsuura = get_score(functions[1], args.n)
        schaffers = get_score(functions[2], args.n)
        results = [bent, katsuura, schaffers]
        data = [x[0] for x in results]
        scores = [x[1] for x in results]
        errors = [x[2] for x in results]
        print("BentCigarFunction   - average of {}: {} ± {}".format(args.n, scores[0], errors[0]))
        print("KatsuuraEvaluation  - average of {}: {} ± {}".format(args.n, scores[1], errors[1]))
        print("SchaffersEvaluation - average of {}: {} ± {}".format(args.n, scores[2], errors[2]))

        if args.plot:
            ax = sns.boxplot(x=["BentCigar", "Katsuura", "Schaffers"], y=data)
            sns.set_style("whitegrid")
            plt.xlabel("Function")
            plt.ylabel("Score")
            plt.savefig("results_n{}.png".format(args.n), dpi=400)

        return

    # no flags given
    print("Defaulting to BentCigarFunction...")
    score = get_score(functions[0], args.n)
    print("Average of {}: {}".format(args.n, score))

def get_score(function: str, n: int):
        scores = []
        for i in range(n):
            output = run("make run FUNC={}".format(function).split(), stdout=PIPE).stdout.decode("utf-8")
            print(output)
            start = output.find("Score: ")
            score = re.match("\d+\.\d+", output[start + 7:])
            score = float(score.group())
            scores.append(score)
        return np.array(scores), sum(scores) / n, np.std(scores)


if __name__ == "__main__":
    main()