JSFox: integrating static and dynamic type analysis of JavaScript programs

Tian Huat Tan, Yinxing Xue, Manman Chen, Shuang Liu, Yi Yu, Jun Sun

Abstract: JavaScript is a dynamic programming language that has been widely used nowadays. The dynamism has become a hindrance of type analysis for JavaScript. Existing works use either static or dynamic type analysis to infer variable types for JavaScript. Static type analysis of JavaScript is difficult since it is hard to predict the behavior of the language without execution. Dynamic type analysis is usually incomplete as it might not cover all paths of a JavaScript program. In this work, we propose jsFox, a browser-agnostic approach that provides integrated type analysis, based on both static and dynamic type analysis, which enables us to gain the merits of both types of analysis. We have made use of the integrated type analysis for finding type issues that could potentially lead to erroneous results. jsFox discovers 23 type issues in existing benchmark suites and real-world Web applications.