The Research On Adaptive Random Testing For Grammarware

As a class of system or program that is used to describe and solve problems by grammar,the grammarware is distinct from other programs in software testing.The test case of grammarware is a sentence that meets the grammar rules,and the way to obtain it is sentence generation algorithms.In addition,the cost of grammarwaretesting is usually high,it is expected to use a small number of high-quality test cases to reveal lots of errors.Therefore,it is important to generate sentence with high error-detection rate for grammarware testing.In current sentence generation algorithms,the sentence set generated by the rule-coverage method have low error detection capability in testing,and random generationmethod is an important supplement.Although random generation method can make up the deficiency of rule-coverage method,there are also shortcomings about containing the similar or even the same test case which leads to low test efficiency.For this reason,this paper introduces an adaptive random testingmethod to improve the quality of random generation sentences.The experiments are used to verify the supplemental effect of random generation method for rule-coverage method.Aiming at the loop problem in random generation method,the maximum number of parse steps is presented.The algorithm framework of sentence adaptive random generation is proposed,and the sentence distance definition is deeply discussed.According to the characteristics of grammarware inputs domain,this paper puts forward several distance hypothesis from string,production state and tree structure.In addition,the application of sentence enumeration in adaptive random testing is analyzed.Experiments show that the edit distance from the production tree is defined as the distance of sentences,which has theoretical basisis in accordance with the judgment of intuition,and is well performed in each experiment and costs less time than others.The edit distance of the production tree is defined as the distance of the sentence,and the generated sentence cases are generally with high quality,which can effectively improve the efficiency of grammarware testing.