Constraint-Driven Fuzzing at Scale with FANDANGO

Summary

By combining grammars with semantic constraints, language-based testing enables the generation of complex, valid test inputs. We present FANDANGO, a test generator that replaces symbolic solving with genetic algorithms, allowing constraints to be written directly in Python rather than in a limited function-solver library. Compared to the state-of-the-art, this approach delivers input generation speeds one to three orders of magnitude higher while maintaining near-perfect validity. FANDANGO is easy to use and supports a wide range of input formats. This paper emphasizes the public release: packaging, command-line, and Python APIs, reproducible examples, and early adoption in teaching and industry. A screencast is available at: https://youtu.be/7W9CaHEC1bM