def _evaluate(self,
x, #
out,
*args,
**kwargs):
"""
This method iterate over an Individual, execute the refactoring operation sequentially,
and compute quality attributes for the refactored version of the program, as objectives of the search
params:
x (Individual): x is an instance of Individual (i.e., a list of refactoring operations)
"""
# Stage 0: Git restore
logger.debug("Executing git restore.")
git_restore(config.PROJECT_PATH)
update_understand_database(config.UDB_PATH)
# Stage 1: Execute all refactoring operations in the sequence x
logger.debug(f"Reached Individual with Size {len(x[0])}")
for refactoring_operation in x[0]:
refactoring_operation.do_refactoring()
# Update Understand DB
update_understand_database(config.UDB_PATH)
# Stage 2: Computing quality attributes
obj = DesignQualityAttributes(udb_path=config.UDB_PATH)
o1 = obj.average_sum
del obj
o2 = testability_main(config.UDB_PATH)
o3 = modularity_main(config.UDB_PATH)
logger.info(f"QMOOD AVG Score: {o1}")
logger.info(f"Testability Score: {o2}")
logger.info(f"Modularity Score: {o3}")
# Stage 3: Marshal objectives into vector
out["F"] = np.array([-1 * o1, -1 * o2, -1 * o3], dtype=float)
def _evaluate(self,
x, #
out,
*args,
**kwargs):
"""
This method iterate over a population, execute the refactoring operations in each individual sequentially,
and compute quality attributes for the refactored version of the program, as objectives of the search
Args:
x (Population): x is a matrix where each row is an individual, and each column a variable. \
We have one variable of type list (Individual) ==> x.shape = (len(Population), 1)
"""
objective_values = []
for k, individual_ in enumerate(x):
# Stage 0: Git restore
logger.debug("Executing git restore.")
git_restore(config.PROJECT_PATH)
logger.debug("Updating understand database after git restore.")
update_understand_database(config.UDB_PATH)
# Stage 1: Execute all refactoring operations in the sequence x
logger.debug(f"Reached Individual with Size {len(individual_[0])}")
for refactoring_operation in individual_[0]:
refactoring_operation.do_refactoring()
# Update Understand DB
logger.debug(f"Updating understand database after {refactoring_operation.name}.")
update_understand_database(config.UDB_PATH)
# Stage 2:
arr = Array('d', range(self.n_obj_virtual))
if self.evaluate_in_parallel:
# Stage 2 (parallel mood): Computing quality attributes
p1 = Process(target=calc_qmood_objectives, args=(arr,))
p2 = Process(target=calc_testability_objective, args=(config.UDB_PATH, arr,))
p3 = Process(target=calc_modularity_objective, args=(config.UDB_PATH, arr,))
p1.start(), p2.start(), p3.start()
p1.join(), p2.join(), p3.join()
o1 = sum([i for i in arr[:6]]) / 6.
o2 = arr[6]
o3 = arr[7]
else:
# Stage 2 (sequential mood): Computing quality attributes
qmood_quality_attributes = DesignQualityAttributes(udb_path=config.UDB_PATH)
o1 = qmood_quality_attributes.average_sum
o2 = testability_main(config.UDB_PATH, initial_value=config.CURRENT_METRICS.get("TEST", 1.0))
o3 = modularity_main(config.UDB_PATH, initial_value=config.CURRENT_METRICS.get("MODULE", 1.0))
del qmood_quality_attributes
# Stage 3: Marshal objectives into vector
objective_values.append([-1 * o1, -1 * o2, -1 * o3])
logger.info(f"Objective values for individual {k}: {[-1 * o1, -1 * o2, -1 * o3]}")
# Stage 4: Marshal all objectives into out dictionary
out['F'] = np.array(objective_values, dtype=float)
def _evaluate(
self,
x, #
out,
*args,
**kwargs):
"""
This method iterate over an Individual, execute the refactoring operation sequentially,
and compute quality attributes for the refactored version of the program, as objectives of the search
params:
x (Individual): x is an instance of Individual (i.e., a list of refactoring operations)
"""
# Git restore`
logger.debug("Executing git restore.")
git_restore(config.PROJECT_PATH)
update_understand_database(config.UDB_PATH)
# Stage 1: Execute all refactoring operations in the sequence x
logger.debug(f"Reached Individual with Size {len(x[0])}")
for refactoring_operation in x[0]:
refactoring_operation.do_refactoring()
# Update Understand DB
update_understand_database(config.UDB_PATH)
# Stage 2: Computing quality attributes
qmood = DesignQualityAttributes(udb_path=config.UDB_PATH)
o1 = qmood.reusability
o2 = qmood.understandability
o3 = qmood.flexibility
o4 = qmood.functionality
o5 = qmood.effectiveness
o6 = qmood.extendability
del qmood
o7 = testability_main(config.UDB_PATH)
o8 = modularity_main(config.UDB_PATH)
logger.info(f"Reusability Score: {o1}")
logger.info(f"Understandability Score: {o2}")
logger.info(f"Flexibility Score: {o3}")
logger.info(f"Functionality Score: {o4}")
logger.info(f"Effectiveness Score: {o5}")
logger.info(f"Extendability Score: {o6}")
logger.info(f"Testability Score: {o7}")
logger.info(f"Modularity Score: {o8}")
# Stage 3: Marshal objectives into vector
out["F"] = np.array([
-1 * o1,
-1 * o2,
-1 * o3,
-1 * o4,
-1 * o5,
-1 * o6,
-1 * o7,
-1 * o8,
],
dtype=float)
rand_pop = rand_init.generate_population()
score = DesignQualityAttributes(udb_path=udb_path).reusability
k = 0
limit = len(rand_pop)
best_answer = None
print("Reusability before starting the algorithm:", score)
# Random search for improving reusability
while k < MAX_ITERATIONS and limit > 0:
print("Iteration No.", k)
index = random.randint(0, limit - 1)
limit -= 1
print(f"index is {index}")
individual = rand_pop.pop(index)
print(f"len(individual) is {len(individual)}")
# git restore prev changes
git_restore(project_dir)
# execute individual
for refactoring, params in individual:
print(params)
try:
refactoring(**params)
except Exception as e:
print(e)
# update understand database
update_understand_database(udb_path)
# calculate objective
obj_score = DesignQualityAttributes(udb_path=udb_path).reusability
print("Objective Score", obj_score)
if obj_score < score:
score = obj_score
best_answer = individual
def _evaluate(self, x, out, *args, **kwargs):
"""
This method iterate over a population, execute the refactoring operations in each individual sequentially,
and compute quality attributes for the refactored version of the program, as objectives of the search.
By default, elementwise_evaluation is set to False, which implies the _evaluate retrieves a set of solutions.
Args:
x (Population): x is a matrix where each row is an individual, and each column a variable.\
We have one variable of type list (Individual) ==> x.shape = (len(Population), 1)
"""
objective_values = []
for k, individual_ in enumerate(x):
# Stage 0: Git restore
logger.debug("Executing git restore.")
git_restore(config.PROJECT_PATH)
logger.debug("Updating understand database after git restore.")
update_understand_database(config.UDB_PATH)
# Stage 1: Execute all refactoring operations in the sequence x
logger.debug(f"Reached an Individual with size {len(individual_[0])}")
for refactoring_operation in individual_[0]:
res = refactoring_operation.do_refactoring()
# Update Understand DB
logger.debug(f"Updating understand database after {refactoring_operation.name}.")
update_understand_database(config.UDB_PATH)
# Stage 2:
arr = Array('d', range(self.n_obj))
if self.evaluate_in_parallel:
# Stage 2 (parallel mood): Computing quality attributes
p1 = Process(target=calc_qmood_objectives, args=(arr,))
if self.n_obj == 8:
p2 = Process(target=calc_testability_objective, args=(config.UDB_PATH, arr,))
p3 = Process(target=calc_modularity_objective, args=(config.UDB_PATH, arr,))
p1.start(), p2.start(), p3.start()
p1.join(), p2.join(), p3.join()
else:
p1.start()
p1.join()
else:
# Stage 2 (sequential mood): Computing quality attributes
qmood_quality_attributes = DesignQualityAttributes(udb_path=config.UDB_PATH)
arr[0] = qmood_quality_attributes.reusability
arr[1] = qmood_quality_attributes.understandability
arr[2] = qmood_quality_attributes.flexibility
arr[3] = qmood_quality_attributes.functionality
arr[4] = qmood_quality_attributes.effectiveness
arr[5] = qmood_quality_attributes.extendability
if self.n_obj == 8:
arr[6] = testability_main(config.UDB_PATH, initial_value=config.CURRENT_METRICS.get("TEST", 1.0))
arr[7] = modularity_main(config.UDB_PATH, initial_value=config.CURRENT_METRICS.get("MODULE", 1.0))
if self.verbose_design_metrics:
design_metrics = {
"DSC": [qmood_quality_attributes.DSC],
"NOH": [qmood_quality_attributes.NOH],
"ANA": [qmood_quality_attributes.ANA],
"MOA": [qmood_quality_attributes.MOA],
"DAM": [qmood_quality_attributes.DAM],
"CAMC": [qmood_quality_attributes.CAMC],
"CIS": [qmood_quality_attributes.CIS],
"NOM": [qmood_quality_attributes.NOM],
"DCC": [qmood_quality_attributes.DCC],
"MFA": [qmood_quality_attributes.MFA],
"NOP": [qmood_quality_attributes.NOP]
}
self.log_design_metrics(design_metrics)
del qmood_quality_attributes
# Stage 3: Marshal objectives into vector
objective_values.append([-1 * i for i in arr])
logger.info(f"Objective values for individual {k}: {[i for i in arr]}")
# Stage 4: Marshal all objectives into out dictionary
out['F'] = np.array(objective_values, dtype=float)