Researchers developed an automated approach for generating inputs to improve grammar mining from existing parsers. The method uses Dynamic Symbolic Execution with iterative expansion and a three-phase approach to overcome limitations when analyzing structured input parsers. Testing on eleven benchmark applications showed the technique achieves high precision and recall while uncovering edge cases typically missed by existing grammar miners like Mimid. This enables software engineers to automatically reconstruct specifications from legacy parsers without manual input generation.
Table of contents
Abstract21 Impressions