Method and apparatus for generating a refactored code
View Patent ↗Methods and apparatuses are provided for code refactoring. The method includes acquiring a code and identifying, using processing circuitry and based on a Markov decision process model, a refactoring sequence. The refactoring sequence includes a plurality of refactoring steps to be applied to the code. Further, the method includes refactoring, by the processing circuitry, the code according to the refactoring sequence.
1. A method for generating a refactored code comprising:
acquiring a code;
identifying, using processing circuitry and based on a Markov decision process model, a refactoring sequence, wherein the refactoring sequence includes a plurality of refactoring steps to be applied to the code;
wherein the Markov decision process model includes:
identifying a plurality of states, each state representing a potential state of the code;
identifying a plurality of actions, wherein each action corresponds to a refactoring step;
determining transition probabilities using enumeration techniques wherein each transition probability represents the probability that the code transition from a current state to a subsequent state by taking an action from the plurality of actions that cause the transition to the subsequent state;
determining a set of rewards, wherein each reward is associated with each of the transition probabilities, wherein determining each reward includes applying:
r
(
s
k
,
a
k
,
s
k
+
1
)
=
∑
i
=
1
L
w
i
·
v
i
where r is the reward for transitioning from state s k , to state s k+1 by taking action a k , w i is the weight of each utility function, wherein the utility function is based on software metrics and bad smell attributes;
determining a refactoring policy that maximize a sum of rewards as a function of the plurality of states, the plurality of actions, the transition probabilities, and the set of rewards, wherein determining the refactoring policy includes applying:
π
*
=
arg
max
π
E
[
∑
k
=
0
K
r
(
s
k
,
a
k
,
s
k
+
1
)
|
π
]
where π is the refactoring policy, r is the reward for transitioning from state s k to state s k+1 by taking action a k , is the expected sum of rewards, and K is a predetermined number; and
applying, using the processing circuitry, the plurality of refactoring steps according to the refactoring sequence to generate a refactored code.
2. The method of claim 1 , wherein the reward is a function of software metrics and bad smells attributes.
3. The method of claim 2 , wherein the software metrics and bad smells attributes have predetermined weights.
4. The method of claim 2 , wherein the predetermined weights are determined using a genetic algorithm.
5. The method of claim 2 , wherein the software metrics include internal and external attributes.
6. The method of claim 5 , wherein the internal attributes include at least one of number of children, coupling between objects, weighted methods per class, lines of code, and lack of cohesion on methods.
7. The method of claim 5 , wherein the external attributes include at least one of maintainability, usability, efficiency, and reliability.
8. The method of claim 1 , wherein the code is acquired from an external device.
9. The method of claim 1 , wherein the refactoring sequence is transmitted to an external device using communication circuitry.
10. An apparatus for generating a refactored code comprising:
processing circuitry configured to
acquire a code,
identify, based on a Markov decision process model, a refactoring sequence, wherein the refactoring sequence includes a plurality of refactoring steps to be applied to the code,
wherein the Markov decision process model includes:
identify a plurality of states, each state representing a potential state of the code,
identify a plurality of actions, wherein each action corresponds to a refactoring step,
determine transition probabilities using enumeration techniques wherein each transition probability represents the probability that the code transition from a current state to a subsequent state by taking an action from the plurality of actions that cause the transition to the subsequent state,
determine a set of rewards, wherein each reward is associated with each of the transition probabilities, wherein determining each reward includes applying:
r
(
s
k
,
a
k
,
s
k
+
1
)
=
∑
i
=
1
L
w
i
·
v
i
where r is the reward for transitioning from state s k to state s k+1 by taking action a k , w i is the weight of each utility function, wherein the utility function is based on software metrics and bad smell attributes
determine a refactoring policy that maximize a sum of rewards as a function of the plurality of states, the plurality of actions, the transition probabilities, and the set of rewards, wherein determining the refactoring policy includes applying:
π
*
=
arg
max
π
E
[
∑
k
=
0
K
r
(
s
k
,
a
k
,
s
k
+
1
)
|
π
]
where π is the refactoring policy, r is the reward for transitioning from state s k , to state s k+1 by taking action a k , E is the expected sum of rewards, and K is a predetermined number, and
apply the plurality of refactoring steps according to the refactoring sequence to generate a refactored code.
11. The apparatus of claim 10 , wherein the reward is a function of software metrics and bad smells attributes.