PDF table extraction in enterprise systems is an architectural problem, not just a library choice. Stream parsing works well for clean text PDFs but breaks under layout drift, wrapped rows, and mixed ...