Recent studies show that developers encounter difficulties when evolving microservices architectures, especially from a code and data management perspective. Specifically, they struggle to recover a high-level view, although essential for understanding and maintaining complex interactions across various microservices codebases and their databases. This paper introduces an approach for systematically analyzing data access code fragments in complex applications composed of multiple microservices and distributed across several codebases. By combining heuristic-based code analysis with natural language processing, the approach holistically identify, extracts, interprets, and documents the interactions between these microservices and their databases. The resulting report is designed to support software evolution tasks, such as locating or visualizing linked data access code fragments. A preliminary evaluation on 5 JavaScript microservices architectures with MongoDB and Redis provides emerging results.