def test_initialization(self):
parameters = [3, 3, 3, 3]
constraints = {((0, 0), (2, 0)), ((0, 1), (1, 1), (2, 0)), ((0, 2), (3, 2))}
t_value = 2
solver = Solver(parameters, constraints)
combination_matrix = CombinationMatrix(parameters, t_value)
solver.clean_hash_table(combination_matrix, t_value)
cit = Cit(parameters, t_value, constraints)
self.assertEqual(combination_matrix, cit.combination_matrix, "The initialization of cit algorithm is wrong")
def __init__(self, input_data, t_value, constraints):
"""
Creation of CombinationMatrix from user input
:param input_data: parameters from user
:param t_value: size of one combination
:param constraints: constraints of combinations
"""
self.data = input_data
self.t_value = t_value
# CombinationMatrix creation
self.combination_matrix = CombinationMatrix(input_data, t_value)
# Creation of solver and simplification of constraints
self.solver = Solver(input_data, constraints)
# Combinations which do not match to the constraints are disabled
self.solver.clean_hash_table(self.combination_matrix, t_value)
self.final_matrix = []
self.__throbber = Throbber()
def test_solver_constraints_without_simplification(self):
"""
Test that, solver do not delete important constraints
"""
parameters = [3, 3, 3, 3]
constraints = {((0, 0), (2, 0)), ((0, 1), (1, 1), (2, 0)),
((0, 2), (3, 2))}
solver = Solver(parameters, constraints)
self.assertEqual(solver.constraints, constraints,
"solver deleted some important constraints")
def test_solver_constraints_with_simplification_and_secrete(self):
"""
Test that, Minimum forbidden tuple algorithm can find and simplify constraints
"""
parameters = [3, 3, 3, 3]
constraints = {((0, 0), (1, 0)), ((0, 0), (1, 2)), ((0, 1), (3, 0)),
((0, 2), (3, 0)), ((1, 1), (3, 0))}
solver = Solver(parameters, constraints)
expectation = {((0, 0), (1, 0)), ((0, 0), (1, 2)), ((3, 0), )}
self.assertEqual(solver.constraints, expectation,
"solver can not compute and simplify constraints")
def test_solver_without_secrete_constraints(self):
"""
Test that, solver didn't change constraints if there isn't any secret constraint
"""
parameters = [3, 3, 3]
constraints = {((0, 0), (2, 0)), ((0, 1), (1, 1)), ((1, 0), (2, 0))}
solver = Solver(parameters, constraints)
self.assertEqual(
solver.constraints, constraints,
"solver change constraints without secret "
"constraint")
def test_simplify_constraints_constraints_without_simplification(self):
"""
Test that, function do not delete important constraints
"""
parameters = [3, 3, 3, 3]
constraints = {((0, 0), (2, 0)), ((0, 1), (1, 1), (2, 0)),
((0, 2), (3, 2))}
solver = Solver([], [])
solver.data = parameters
solver.constraints = constraints
solver.read_constraints()
solver.simplify_constraints()
self.assertEqual(
solver.constraints, constraints,
"simplify_constraints deleted some important constraints")
def test_compute_constraints_without_secret_constraint(self):
"""
Test that, function shouldn't change constraints if there isn't any secret constraint
"""
parameters = [3, 3, 3]
constraints = {((0, 0), (2, 0)), ((0, 1), (1, 1)), ((1, 0), (2, 0))}
solver = Solver([], [])
solver.data = parameters
solver.constraints = constraints
solver.read_constraints()
solver.compute_constraints()
self.assertEqual(
solver.constraints, constraints,
"compute_constraints change constraints without secret "
"constraint")
def test_compute_constraints_with_one_secrete_constraint(self):
"""
Test that, function should find new constraint
"""
parameters = [3, 3, 3, 3]
constraints = {((0, 0), (2, 0)), ((0, 1), (1, 1), (2, 0)),
((0, 2), (3, 2))}
solver = Solver([], [])
solver.data = parameters
solver.constraints = constraints
solver.read_constraints()
solver.compute_constraints()
expectation = {((0, 0), (2, 0)), ((0, 1), (1, 1), (2, 0)),
((0, 2), (3, 2)), ((1, 1), (2, 0), (3, 2))}
self.assertEqual(solver.constraints, expectation,
"compute_constraints didn't find secret constraints")
def test_compute_constraints_new_constraint_with_more_than_one_value_for_one_parameter(
self):
"""
Test that, function shouldn't change constraints. It founds new constraint,
but the constraint has one parameter with two values. This means that it isn't secreted constraint.
"""
parameters = [3, 3, 3]
constraints = {((0, 0), (1, 0)), ((0, 1), (1, 1)), ((1, 2), (2, 0))}
solver = Solver([], [])
solver.data = parameters
solver.constraints = constraints
solver.read_constraints()
solver.compute_constraints()
self.assertEqual(
solver.constraints, constraints,
"compute_constraints change constraints without secret "
"constraint")
def test_compute_constraints_detect_invalid_constraints(self):
"""
Test that, function should raise an exception, if there is invalid_constraints
"""
parameters = [3, 3, 3]
constraints = {((0, 0), ), ((0, 1), ), ((0, 2), )}
solver = Solver([], [])
solver.data = parameters
solver.constraints = constraints
solver.read_constraints()
self.assertRaises(ValueError, solver.compute_constraints)
class Cit:
def __init__(self, input_data, t_value, constraints):
"""
Creation of CombinationMatrix from user input
:param input_data: parameters from user
:param t_value: size of one combination
:param constraints: constraints of combinations
"""
self.data = input_data
self.t_value = t_value
# CombinationMatrix creation
self.combination_matrix = CombinationMatrix(input_data, t_value)
# Creation of solver and simplification of constraints
self.solver = Solver(input_data, constraints)
# Combinations which do not match to the constraints are disabled
self.solver.clean_hash_table(self.combination_matrix, t_value)
self.final_matrix = []
self.__throbber = Throbber()
def final_matrix_init(self):
"""
Creation of the first solution. This solution is the start of searching
for the best solution
:return: solution matrix (list(list))
"""
self.final_matrix = [self.create_random_row_with_constraints()]
self.combination_matrix.cover_solution_row(self.final_matrix[0])
while self.combination_matrix.total_uncovered != 0:
if self.combination_matrix.total_uncovered < self.combination_matrix.total_covered_more_than_ones:
new_row = self.compute_row()
else:
new_row = self.compute_row_using_hamming_distance()
self.combination_matrix.cover_solution_row(new_row)
self.final_matrix.append(new_row)
return self.final_matrix
def compute(self):
"""
Searching for the best solution. It creates one solution and from that,
it tries to create smaller solution. This searching process is limited
by ITERATIONS_SIZE. When ITERATIONS_SIZE is 0 the last found solution is
the best solution.
:return: The best solution
"""
self.final_matrix = self.final_matrix_init()
matrix = [x[:] for x in self.final_matrix]
iterations = ITERATIONS_SIZE
step_size = 1
deleted_rows = []
while step_size != 0:
for i in range(step_size):
delete_row = matrix.pop(random.randint(0, len(matrix) - 1))
self.combination_matrix.uncover_solution_row(delete_row)
deleted_rows.append(delete_row)
LOG.debug("I'm trying solution with size %s and %s iterations",
len(matrix), iterations)
matrix, is_better_solution = self.find_better_solution(iterations, matrix)
if is_better_solution:
self.final_matrix = matrix[:]
deleted_rows = []
step_size *= 2
LOG.debug("-----solution with size %s was found-----\n",
len(matrix))
iterations = ITERATIONS_SIZE
else:
LOG.debug("-----solution with size %s was not found-----\n",
len(matrix))
for i in range(step_size):
self.combination_matrix.cover_solution_row(deleted_rows[i])
matrix.append(deleted_rows[i])
if step_size > 1:
step_size = 1
else:
step_size = 0
return self.final_matrix
def find_better_solution(self, counter, matrix):
"""
Changing the matrix to cover all combinations
:param counter: maximum number of changes in the matrix
:param matrix: matrix to be changed
:return: new matrix and is changes have been successful?
"""
while self.combination_matrix.total_uncovered != 0:
LOG.debug(self.__throbber.render(), extra={"skip_newline": True})
solution, row_index, _ = self.use_random_algorithm(matrix)
if len(solution) != 0:
self.combination_matrix.uncover_solution_row(matrix[row_index])
self.combination_matrix.cover_solution_row(solution)
matrix[row_index] = solution
if counter == 0:
return matrix, False
counter -= 1
return matrix, True
def use_random_algorithm(self, matrix):
"""
Applies one of these algorithms to the matrix.
It chooses algorithm by random in proportion 1:1:8
:param matrix: matrix to be changed
:return: new row of matrix, index of row inside matrix and parameters which has been changed
"""
switch = random.randint(0, 9)
if switch == 0:
solution, row_index, parameters = self.change_one_value(matrix)
elif switch == 1:
solution, row_index, parameters = self.change_one_column(matrix)
else:
solution, row_index, parameters = self.cover_missing_combination(matrix)
return solution, row_index, parameters
def compute_row(self):
"""
Computation of one row which covers most of combinations
:return: new solution row
"""
is_valid_row = False
while not is_valid_row:
possible_parameters = list(self.combination_matrix.uncovered_rows)
row = [-1] * len(self.data)
while len(possible_parameters) != 0:
# finding uncovered combination
combination_parameters_index = random.randint(0, len(possible_parameters) - 1)
combination_parameters = possible_parameters[combination_parameters_index]
del possible_parameters[combination_parameters_index]
combination_row = self.combination_matrix.get_row(combination_parameters)
possible_combinations = list(combination_row.get_all_uncovered_combinations())
combination_index = random.randint(0, len(possible_combinations) - 1)
combination = possible_combinations[combination_index]
is_parameter_used = False
# Are parameters already used in row?
for i in combination_parameters:
if row[i] != -1:
is_parameter_used = True
break
if is_parameter_used:
continue
row_copy = row.copy()
# Is combination matches the constraints?
for index, parameter in enumerate(combination_parameters):
row_copy[parameter] = combination[index]
if self.combination_matrix.is_valid_solution(row_copy):
row = row_copy
# Filling in of free spaces inside the row
for index, r in enumerate(row):
if r == -1:
is_valid = False
while not is_valid:
row[index] = random.randint(0, self.data[index] - 1)
is_valid = self.combination_matrix.is_valid_solution(row)
is_valid_row = self.combination_matrix.is_valid_solution(row)
return row
def cover_missing_combination(self, matrix):
"""
Randomly finds one missing combination. This combination puts into each
row of the matrix. The row with the best coverage is the solution
:param matrix: matrix to be changed
:return: solution, index of solution inside matrix and parameters which has been changed
"""
parameters, combination = self.get_missing_combination_random()
best_uncover = float("inf")
best_solution = []
best_row_index = 0
for row_index in range(len(matrix)):
solution = [x for x in matrix[row_index]]
for index, item in enumerate(parameters):
solution[item] = combination[index]
if self.combination_matrix.is_valid_combination(solution, parameters):
self.combination_matrix.uncover_combination(matrix[row_index], parameters)
self.combination_matrix.cover_combination(solution, parameters)
if self.combination_matrix.total_uncovered < best_uncover:
best_uncover = self.combination_matrix.total_uncovered
best_solution = solution
best_row_index = row_index
self.combination_matrix.uncover_combination(solution, parameters)
self.combination_matrix.cover_combination(matrix[row_index], parameters)
if best_uncover == 0:
break
return best_solution, best_row_index, parameters
def get_missing_combination_random(self):
"""
Randomly finds one missing combination.
:return: parameter of combination and values of combination
"""
possible_parameters = list(self.combination_matrix.uncovered_rows)
combination_parameters_index = random.randint(0, len(possible_parameters) - 1)
combination_parameters = possible_parameters[combination_parameters_index]
combination_row = self.combination_matrix.get_row(combination_parameters)
possible_combinations = list(combination_row.get_all_uncovered_combinations())
combination_index = random.randint(0, len(possible_combinations) - 1)
combination = possible_combinations[combination_index]
return combination_parameters, combination
def change_one_column(self, matrix):
"""
Randomly choose one column of the matrix. In each cell of this column
changes value. The row with the best coverage is the solution.
:param matrix: matrix to be changed
:return: solution, index of solution inside matrix and parameters which has been changed
"""
column_index = random.randint(0, len(self.data) - 1)
best_uncover = float("inf")
best_solution = []
best_row_index = 0
for row_index in range(len(matrix)):
try:
solution, row_index, parameters = self.change_one_value(matrix, row_index, column_index)
except ValueError:
continue
self.combination_matrix.uncover_combination(matrix[row_index], parameters)
self.combination_matrix.cover_combination(solution, parameters)
if self.combination_matrix.total_uncovered < best_uncover:
best_uncover = self.combination_matrix.total_uncovered
best_solution = solution
best_row_index = row_index
self.combination_matrix.uncover_combination(solution, parameters)
self.combination_matrix.cover_combination(matrix[row_index], parameters)
if best_uncover == 0:
break
return best_solution, best_row_index, [column_index]
def change_one_value(self, matrix, row_index=None, column_index=None):
"""
Change one cell inside the matrix
:param matrix: matrix to be changed
:param row_index: row inside matrix. If it's None it is chosen randomly
:param column_index: column inside matrix. If it's None it is chosen randomly
:return: solution, index of solution inside matrix and parameters which has been changed
"""
is_cell_chosen = True
if row_index is None:
is_cell_chosen = False
row_index = random.randint(0, len(matrix) - 1)
row = [x for x in matrix[row_index]]
if column_index is None:
is_cell_chosen = False
column_index = random.randint(0, len(row) - 1)
possible_numbers = list(range(0, row[column_index])) + list(
range(row[column_index] + 1, self.data[column_index]))
row[column_index] = random.choice(possible_numbers)
while not self.combination_matrix.is_valid_combination(row, [column_index]):
possible_numbers.remove(row[column_index])
if len(possible_numbers) == 0:
if is_cell_chosen:
raise ValueError("Selected cell can't be changed")
column_index = random.randint(0, len(row) - 1)
row_index = random.randint(0, len(matrix) - 1)
row = [x for x in matrix[row_index]]
possible_numbers = list(range(0, row[column_index])) + list(
range(row[column_index] + 1, self.data[column_index]))
row[column_index] = random.choice(possible_numbers)
return row, row_index, [column_index]
def compute_row_using_hamming_distance(self):
"""
:return: row with the biggest hamming distance from final matrix
"""
row_1 = self.create_random_row_with_constraints()
row_2 = self.create_random_row_with_constraints()
if self.compute_hamming_distance(row_1) >= self.compute_hamming_distance(row_2):
return row_1
else:
return row_2
def compute_hamming_distance(self, row):
"""
:return: hamming distance of row from final matrix
"""
distance = 0
for final_row in self.final_matrix:
for index, cell in enumerate(final_row):
if row[index] != cell:
distance += 1
return distance
def create_random_row_with_constraints(self):
"""
Create a new test-case random row, and the row meets the constraints.
:return: new random row
:rtype: list
"""
data_size = len(self.data)
row = [-1]*data_size
for parameter in random.sample(range(data_size), data_size):
possible_values = self.solver.get_possible_values(row, parameter)
value_choice = random.choice(possible_values)
row[parameter] = value_choice
return row