| In general, software test is a procedure of program execution, which is to find and correct the hidden faults with as little as possible time and labor. Unit test, as the lowest level activity in software test, can discover and solve the hidden program faults by fault localization technology, which reduces the test cost and improves software qualities. Fault localization technology is divided into manual and automatic styles. Automatic fault localization can solve the disadvantages caused by manual style such as blindness, low efficiency and localization accuracy, large workload and so on. So it has important significance to research automatic fault localization.Automatic fault localization generally takes the way of carrying out tested program using a set of test cases, then calculating suspicious degree of statements according to the execution results, and finally determining the fault position. Focusing on this technology research, this paper does following works.(1) As for the execution paths redundancy problem, this paper proposes an execution paths reduction technology based on similarity measure, which deletes repeated paths and successful paths least similar with failed paths to make up a suspicious space and provide the basis for calculation.(2) To solve the problem that how to calculate suspicious degree, this paper provides a method which first calculates the path edges suspicious degree and then path nodes’. Since this method considers the fault transmission between path nodes, it makes more accurate suspicious degree compared with the method only calculating nodes suspicious degree.(3) This paper realizes above automatic fault localization technology in Code Test System. At the same time, the author performs specific experiments on10tested programs combining this technology and test generation and fault detection. The experiment verifies that it’s an effective modification for relevant software fault localization strategies. |
PDF Full Text Request