Software Bug Reproduction Based On Genetic Algorithm And Seeding Strategy

With its advantages of high efficiency,reliability,and wide coverage,the search algorithm has been widely respected in the field of software automation testing,especially in the aspect of bug reproduction.However,the current search-based software bug reproduction method has strict requirements on bug information,and its versatility is poor,and the existing method only designs the fitness function from the two aspects of bug trigger statement coverage and bug type,and the bug reproduction is insufficient.Guidance is not conducive to automatically generating test cases that can effectively reproduct bugs.In addition,most of the current search-based bug reproduction methods do not consider the parameter generation problem in the test method sequence generation process,and the parameter values are random values,resulting in high cost of generating test data for reproducting bugs,which further affects the reliability of software bug reproduction.effect and efficiency.For this reason,this paper proposes a software bug reproduction method based on the bootstrap genetic algorithm and seeding strategy,using the bug stack trace as the source of bug information,based on the bug stack trace and the bootstrap genetic algorithm to generate test case evolution,to realize automatic bug reproduction In addition,the method in this paper designs and optimizes the fitness function from the three aspects of bug trigger sentence coverage,bug type and stack trace similarity,so as to effectively identify potential excellent individuals,improve computational efficiency,and accelerate the convergence speed of the algorithm.Furthermore,most of the parameter values of the test sequence in the existing bug reproduct method are randomly generated,which cannot cover some statements that require specific parameters to reach.In this regard,this paper proposes a parameter value generation method based on the seeding strategy.From the static analysis of the source program bytecode and the dynamic analysis of the individual running trajectory,the possible values of the parameters in the test sequence are collected,and the seeding strategy is used to optimize The generation of the initial population and the genetic operation of the population optimize the test case generation process for bug reproduction and improve the effect of software bug reproduction.In order to verify the effectiveness of the method in this paper,this paper collected 105 real software bugs from five large-scale open source software.Whether the seeding strategy can improve the effect of bug reproduction is tested to verify the effectiveness of the method in this paper.The results prove that,compared with the existing software bug reproduction methods,the method in this paper can reproduce more bugs on the real software bug data set,and the reproduction effect has been significantly improved.