2023-05-09 11:47:47 +02:00
|
|
|
|
from math import log as log
|
|
|
|
|
from numpy import zeros as zeros
|
|
|
|
|
from math import fabs as fabs
|
|
|
|
|
from math import floor as floor
|
|
|
|
|
from math import sqrt as sqrt
|
|
|
|
|
from scipy.special import erfc as erfc
|
|
|
|
|
from scipy.special import gammaincc as gammaincc
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TotOnline:
|
2023-05-15 15:40:10 +02:00
|
|
|
|
|
2023-05-09 11:47:47 +02:00
|
|
|
|
@staticmethod
|
2023-05-15 15:40:10 +02:00
|
|
|
|
def total_failure_test(binary_data: bytes, pattern_length=10):
|
|
|
|
|
length_of_binary_data = len(binary_data) * 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Convert bytes to binary string representation
|
|
|
|
|
binary_data_str = ''
|
|
|
|
|
for byte in binary_data:
|
|
|
|
|
binary_data_str += format(byte, '08b')
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
# Augment the n-bit sequence to create n overlapping m-bit sequences by appending m-1 bits
|
|
|
|
|
# from the beginning of the sequence to the end of the sequence.
|
2023-05-15 15:40:10 +02:00
|
|
|
|
binary_data_str += binary_data_str[:pattern_length + 1]
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
# Get max length one patterns for m, m-1, m-2
|
2023-05-15 15:40:10 +02:00
|
|
|
|
max_pattern = '1' * (pattern_length + 2)
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
# Keep track of each pattern's frequency (how often it appears)
|
2023-05-15 15:40:10 +02:00
|
|
|
|
vobs_01 = zeros(int(max_pattern[0:pattern_length], 2) + 1)
|
|
|
|
|
vobs_02 = zeros(int(max_pattern[0:pattern_length + 1], 2) + 1)
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
for i in range(length_of_binary_data):
|
|
|
|
|
# Work out what pattern is observed
|
2023-05-15 15:40:10 +02:00
|
|
|
|
vobs_01[int(binary_data_str[i:i + pattern_length], 2)] += 1
|
|
|
|
|
vobs_02[int(binary_data_str[i:i + pattern_length + 1], 2)] += 1
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
# Calculate the test statistics and p values
|
|
|
|
|
vObs = [vobs_01, vobs_02]
|
|
|
|
|
|
|
|
|
|
sums = zeros(2)
|
|
|
|
|
for i in range(2):
|
|
|
|
|
for j in range(len(vObs[i])):
|
|
|
|
|
if vObs[i][j] > 0:
|
|
|
|
|
sums[i] += vObs[i][j] * log(vObs[i][j] / length_of_binary_data)
|
|
|
|
|
sums /= length_of_binary_data
|
|
|
|
|
ape = sums[0] - sums[1]
|
|
|
|
|
|
|
|
|
|
xObs = 2.0 * length_of_binary_data * (log(2) - ape)
|
|
|
|
|
|
|
|
|
|
p_value = gammaincc(pow(2, pattern_length - 1), xObs / 2.0)
|
|
|
|
|
|
|
|
|
|
return p_value, (p_value >= 0.01)
|
|
|
|
|
|
|
|
|
|
@staticmethod
|
2023-05-15 15:40:10 +02:00
|
|
|
|
def monobit_test(binary_data: bytes):
|
|
|
|
|
length_of_bit_string = len(binary_data) * 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Variable for S(n)
|
2023-05-09 11:47:47 +02:00
|
|
|
|
count = 0
|
2023-05-15 15:40:10 +02:00
|
|
|
|
|
|
|
|
|
# Iterate each bit in the byte sequence and compute for S(n)
|
|
|
|
|
for byte in binary_data:
|
|
|
|
|
for bit in range(8):
|
|
|
|
|
if (byte >> (7 - bit)) & 1 == 0:
|
|
|
|
|
# If bit is 0, then -1 from the S(n)
|
|
|
|
|
count -= 1
|
|
|
|
|
else:
|
|
|
|
|
# If bit is 1, then +1 to the S(n)
|
|
|
|
|
count += 1
|
|
|
|
|
|
|
|
|
|
# Compute the test statistic
|
2023-05-09 11:47:47 +02:00
|
|
|
|
sObs = count / sqrt(length_of_bit_string)
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Compute p-Value
|
2023-05-09 11:47:47 +02:00
|
|
|
|
p_value = erfc(fabs(sObs) / sqrt(2))
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# return a p_value and randomness result
|
2023-05-09 11:47:47 +02:00
|
|
|
|
return p_value, (p_value >= 0.01)
|
|
|
|
|
|
|
|
|
|
@staticmethod
|
2023-05-15 15:40:10 +02:00
|
|
|
|
def block_frequency_test(binary_data: bytes, block_size=128):
|
|
|
|
|
length_of_bit_string = len(binary_data) * 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
if length_of_bit_string < block_size:
|
|
|
|
|
block_size = length_of_bit_string
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Compute the number of blocks based on the input given. Discard the remainder
|
|
|
|
|
number_of_blocks = length_of_bit_string // block_size
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
if number_of_blocks == 1:
|
|
|
|
|
# For block size M=1, this test degenerates to test 1, the Frequency (Monobit) test.
|
|
|
|
|
return TotOnline.monobit_test(binary_data[0:block_size])
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Initialize variables
|
2023-05-09 11:47:47 +02:00
|
|
|
|
proportion_sum = 0.0
|
|
|
|
|
|
|
|
|
|
# Create a for loop to process each block
|
|
|
|
|
for counter in range(number_of_blocks):
|
|
|
|
|
# Partition the input sequence and get the data for block
|
2023-05-15 15:40:10 +02:00
|
|
|
|
block_start = counter * block_size // 8
|
|
|
|
|
block_end = block_start + block_size // 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
block_data = binary_data[block_start:block_end]
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Determine the proportion πi of ones in each M-bit
|
2023-05-09 11:47:47 +02:00
|
|
|
|
one_count = 0
|
2023-05-15 15:40:10 +02:00
|
|
|
|
for byte in block_data:
|
|
|
|
|
for bit in range(8):
|
|
|
|
|
if (byte >> (7 - bit)) & 1 == 1:
|
|
|
|
|
one_count += 1
|
|
|
|
|
# Compute π
|
|
|
|
|
pi = one_count / (block_size * 8) * 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
# Compute Σ(πi -½)^2.
|
|
|
|
|
proportion_sum += pow(pi - 0.5, 2.0)
|
|
|
|
|
|
|
|
|
|
# Compute 4M Σ(πi -½)^2.
|
|
|
|
|
result = 4.0 * block_size * proportion_sum
|
|
|
|
|
|
|
|
|
|
# Compute P-Value
|
|
|
|
|
p_value = gammaincc(number_of_blocks / 2, result / 2)
|
|
|
|
|
|
|
|
|
|
return p_value, (p_value >= 0.01)
|
|
|
|
|
|
|
|
|
|
@staticmethod
|
2023-05-15 15:40:10 +02:00
|
|
|
|
def run_test(binary_data: bytes):
|
2023-05-09 11:47:47 +02:00
|
|
|
|
vObs = 0
|
2023-05-15 15:40:10 +02:00
|
|
|
|
length_of_binary_data = len(binary_data) * 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
# Predefined tau = 2 / sqrt(n)
|
|
|
|
|
tau = 2 / sqrt(length_of_binary_data)
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Step 1 - Compute the pre-test proportion π of ones in the input sequence: π = Σjεj / n
|
|
|
|
|
one_count = 0
|
|
|
|
|
for byte in binary_data:
|
|
|
|
|
for bit in range(8):
|
|
|
|
|
if (byte >> (7 - bit)) & 1 == 1:
|
|
|
|
|
one_count += 1
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
pi = one_count / length_of_binary_data
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
# Step 2 - If it can be shown that the absolute value of (π - 0.5) is greater than or equal to tau,
|
2023-05-09 11:47:47 +02:00
|
|
|
|
# then the run test need not be performed.
|
|
|
|
|
if abs(pi - 0.5) >= tau:
|
|
|
|
|
return 0.0000
|
|
|
|
|
else:
|
|
|
|
|
# Step 3 - Compute vObs
|
2023-05-15 15:40:10 +02:00
|
|
|
|
for i in range(1, length_of_binary_data):
|
|
|
|
|
if ((binary_data[i // 8] >> (7 - (i % 8))) & 1) != ((binary_data[(i - 1) // 8] >> (7 - ((i - 1) % 8))) & 1):
|
2023-05-09 11:47:47 +02:00
|
|
|
|
vObs += 1
|
|
|
|
|
vObs += 1
|
|
|
|
|
|
|
|
|
|
# Step 4 - Compute p_value = erfc((|vObs − 2nπ * (1−π)|)/(2 * sqrt(2n) * π * (1−π)))
|
|
|
|
|
p_value = erfc(abs(vObs - (2 * length_of_binary_data * pi * (1 - pi))) / (2 * sqrt(2 * length_of_binary_data) * pi * (1 - pi)))
|
|
|
|
|
|
|
|
|
|
return p_value, (p_value > 0.01)
|
|
|
|
|
|
|
|
|
|
@staticmethod
|
2023-05-15 15:40:10 +02:00
|
|
|
|
def longest_one_block_test(binary_data: bytes):
|
|
|
|
|
length_of_binary_data = len(binary_data) * 8
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
if length_of_binary_data < 128:
|
|
|
|
|
return 0.00000, 'Error: Not enough data to run this test'
|
|
|
|
|
elif length_of_binary_data < 6272:
|
|
|
|
|
k = 3
|
|
|
|
|
m = 8
|
|
|
|
|
v_values = [1, 2, 3, 4]
|
|
|
|
|
pi_values = [0.2148, 0.3672, 0.2305, 0.1875]
|
|
|
|
|
elif length_of_binary_data < 750000:
|
|
|
|
|
k = 5
|
|
|
|
|
m = 128
|
|
|
|
|
v_values = [4, 5, 6, 7, 8, 9]
|
|
|
|
|
pi_values = [0.1174, 0.2430, 0.2493, 0.1752, 0.1027, 0.1124]
|
|
|
|
|
else:
|
|
|
|
|
k = 6
|
|
|
|
|
m = 10000
|
|
|
|
|
v_values = [10, 11, 12, 13, 14, 15, 16]
|
|
|
|
|
pi_values = [0.0882, 0.2092, 0.2483, 0.1933, 0.1208, 0.0675, 0.0727]
|
|
|
|
|
|
|
|
|
|
number_of_blocks = floor(length_of_binary_data / m)
|
|
|
|
|
block_start = 0
|
|
|
|
|
block_end = m
|
|
|
|
|
xObs = 0
|
|
|
|
|
frequencies = zeros(k + 1)
|
|
|
|
|
|
|
|
|
|
for count in range(number_of_blocks):
|
2023-05-15 15:40:10 +02:00
|
|
|
|
block_data = binary_data[block_start // 8:block_end // 8]
|
2023-05-09 11:47:47 +02:00
|
|
|
|
max_run_count = 0
|
|
|
|
|
run_count = 0
|
|
|
|
|
|
2023-05-15 15:40:10 +02:00
|
|
|
|
for byte in block_data:
|
|
|
|
|
for bit in range(7, -1, -1):
|
|
|
|
|
if (byte >> bit) & 1 == 1:
|
|
|
|
|
run_count += 1
|
|
|
|
|
max_run_count = max(max_run_count, run_count)
|
|
|
|
|
else:
|
|
|
|
|
max_run_count = max(max_run_count, run_count)
|
|
|
|
|
run_count = 0
|
2023-05-09 11:47:47 +02:00
|
|
|
|
|
|
|
|
|
max(max_run_count, run_count)
|
|
|
|
|
|
|
|
|
|
if max_run_count < v_values[0]:
|
|
|
|
|
frequencies[0] += 1
|
|
|
|
|
for j in range(k):
|
|
|
|
|
if max_run_count == v_values[j]:
|
|
|
|
|
frequencies[j] += 1
|
|
|
|
|
if max_run_count > v_values[k - 1]:
|
|
|
|
|
frequencies[k] += 1
|
|
|
|
|
|
|
|
|
|
block_start += m
|
|
|
|
|
block_end += m
|
|
|
|
|
for count in range(len(frequencies)):
|
|
|
|
|
xObs += pow((frequencies[count] - (number_of_blocks * pi_values[count])), 2.0) / (
|
|
|
|
|
number_of_blocks * pi_values[count])
|
|
|
|
|
p_value = gammaincc(float(k / 2), float(xObs / 2))
|
|
|
|
|
|
|
|
|
|
return p_value, (p_value > 0.01)
|
2023-05-15 15:40:10 +02:00
|
|
|
|
|