index_of_coincidence.py

from tabulate import tabulate

from stats_collector import StatsCollector
import matplotlib.pyplot as plt

desired_ioc_value = 0.065
random_ioc_value = 0.038


class IndexOfCoincidence:
    def __init__(self, cipher_statistics: StatsCollector):
        self.cipher_statistics = cipher_statistics
        self.ioc_map = {}

    def find_length(self):
        for m in range(1, 14):
            substrings = self.substrings(m)
            average_ioc = sum(self.calculate_ioc(y) for y in substrings) / m
            self.ioc_map[m] = average_ioc
        print(tabulate(self.ioc_map.items(), headers="firstrow", tablefmt="rounded_grid"))
        self.plot()

    def substrings(self, m):
        return [
            "".join(
                self.cipher_statistics.data[j]
                for j in range(i, self.cipher_statistics.total_chars, m)
            )
            for i in range(m)
        ]

    @staticmethod
    def calculate_ioc(data):
        print(data)
        data_stats = StatsCollector(data)
        data_stats.print_stats()
        f = sum(v * v for v in data_stats.freq_dict.values())
        return f / (data_stats.total_chars * data_stats.total_chars)

    def plot(self):
        plt.bar(self.ioc_map.keys(), self.ioc_map.values())
        plt.axhline(y=desired_ioc_value, color='b', linestyle='-', label='Desired IoC')
        plt.axhline(y=random_ioc_value, color='r', linestyle='-', label='Random IoC')
        plt.errorbar(self.ioc_map.keys(), self.ioc_map.values(),
                     yerr=[abs(self.ioc_map[i] - desired_ioc_value) for i in self.ioc_map], fmt='.', ecolor='orange',
                     color='yellow', capsize=5)
        plt.legend()
        plt.xlabel("Key Length")
        plt.ylabel("IoC")
        plt.title("IoC Values for Random Key Length")
        plt.show()