Search-Based Generation of Complex Inputs with FANDANGO

Summary

Generating complex, semantically valid test inputs remains a central challenge in search-based software testing. FANDANGO addresses this challenge by combining context-free grammars with fully specified Python constraints and using evolutionary search to satisfy them. Compared with the symbolic state of the art, FANDANGO achieves speedups of up to three orders of magnitude while maintaining 100% input validity and equal or higher grammar coverage.FANDANGO is under active development: recent work explores whitebox fuzzing and stateful fuzzing through interaction grammars that model multi-turn protocol behavior. With more than 100,000 PyPI downloads, adoption in industry (including Volkswagen and Bosch), and coverage in the science press, FANDANGO has established itself as a practical input generator.