logo
CanyamAI summaries for academic research
language
MobileMobile
  • Phone FeatureWeChat Mini Program
EnglishEnglish
noticeNotice
BlogBlog
delete
arrow
Return

JSimpo: Structural Deobfuscation of JavaScript Programs

delete2025-08-18
delete0
PRE
AI
T
Tianyu Chen
李
李丁 (Ding Li)
Y
Ying Zhang
T
Tao Xie
DOI：10.1145/3714460delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
JavaScript (JS) obfuscation is now prevalent among popular websites and introduces challenges for malware detection and code review. Given an obfuscated JS program, existing deobfuscation techniques aim to recover the original JS program. However, these techniques overlook structural obfuscation (e.g., control-flow flattening), which causes deobfuscation to have a sub-optimal success rate. To address these challenges, in this article, we propose the first approach of structural deobfuscation named JSimpo for JS programs with two novel techniques: slice symbolic execution and dynamic code execution. We implement our JSimpo approach and evaluate it on 2,000 JS programs from the top 100 JS projects on GitHub. The evaluation results show that JSimpo can effectively conduct structural deobfuscation, boosting the average structural similarity to 78.41% (from 39.33%) between obfuscated programs and their original programs, whereas the best of the state-of-the-art/practice deobfuscators can achieve only 62.64%. The results also show JSimpo's generalization ability over programs obfuscated by various obfuscators. Additionally, JSimpo preserves the semantics of deobfuscated programs by passing all test cases that obfuscated programs have passed.
Keywords:
JavaScript obfuscation
structural deobfuscation
slice symbolic execution
dynamic code execution
program similarity

Journal

A
ACM Transactions on Software Engineering and Methodology
IF:
6.2
Papers:
1.2K
Citations:
3.4K

Organization

No organization information available