PyLingual - A Python Decompilation Framework for Evolving Python Versions

Learn about a groundbreaking research presentation from Black Hat that introduces PyLingual, an innovative Python decompilation framework designed to handle evolving Python versions. Explore how this 25-minute talk combines Natural Language Processing (NLP) techniques with Programming Language theory to create an adaptive decompiler that automatically accommodates new language features and bytecode specification changes. Discover the framework's three major contributions: a scalable decompilation system supporting Python versions 3.6-3.12, a verification framework for perfect decompilation, and a free online service that has successfully reverse engineered over 5,000 PYC binaries. Gain insights from a collaborative research effort between The University of Texas at Dallas, KAIST, AWS, and Flatiron Health that addresses the challenges of maintaining Python decompilation tools in response to the language's growing use in malware development and its constantly evolving bytecode specifications.