An effective long string searching algorithm towards component security testing

In the execution of method invocation sequences to test component security, abnormal or normal information is generated and recorded in a monitor log. By searching abnormal information from monitor log, the exceptions that the component has can be determined. To facilitate the searching process, string searching methods could be employed. However, current approaches are not effective enough to search long pattern string. In order to mine the specific information with less number of matches, we proposed an improved Sunday string searching algorithm in this paper. Unlike Sunday algorithm which does not make use of the already matched characters, the proposed approach presents two ideas — utilizing and recycling these characters. We take advantage of all matched characters in main string, if they are still in the matchable interval compared with pattern string, to increase the distance that pattern string moves backwards. Experimental analysis shows that, compared to Sunday algorithm, our method could greatly reduce the matching times, if the scale of character set constituting both main string and pattern string is small, or if the length of pattern string is long. Also, the proposed approach can improve the search effectiveness for abnormal information in component security testing.