This research introduces CORAL, a neuro-symbolic framework that translates entire code repositories between programming languages while maintaining functional correctness.

• Combines repository-level awareness with symbolic validation techniques
• Achieves up to 93% success rate on real-world repositories
• Outperforms existing approaches by handling complex project structures and dependencies
• Implements a practical test-and-repair approach to ensure translation correctness

For engineering teams, this breakthrough enables efficient migration between programming languages while preserving application functionality—potentially saving thousands of development hours in legacy system modernization.

Repository-Level Compositional Code Translation and Validation