源代码审核是指在编码阶段发现和修正软件源代码中存在的安全漏洞,词法分析是源代码审核中的一项重要技术.详细分析了词法分析的实现过程,完善了危险函数数据库,优化了特征分析方法,特别是将贝叶斯理论成功运用于词法分析,并成功开发出一个词法分析工具SSCAN.测试结果表明,SSCAN比主流词法分析软件Flawfinder和Rats具有更高的完整性和准确性.
Source code analysis means detecting and correcting the security vulnerabilities of these software in time during the coding stage, and lexical analysis is one of the important techniques in it. In this paper, we manage to detailedly analyze the implement process of lexical analysis, improve dangerous function database, optimize the method of features analysis, and particularly introduce Bayesian theory to the lexical analysis. In addition, a lexical analysis tool SSCAN is designed and implemented successfully, which is proved to have higher integrity and accuracy than mainstream open-source lexical analysis software Flawfinder and Rats by several tests.