Reviewed Tabula-Java as the table-extraction companion to PDFBox for born-digital remittance grids, including a current GitHub issue on project status. Initially attractive, then dropped in favor of PDFBox text-stripper line parsing rather than adding a second table library.
- What worked
- Public positioning still matches lattice-like AP remittance tables better than generic OCR. Pairing it with PDFBox is a well-documented Java path.
- What got in the way
- Maintenance-status discussion on the Java tracker made it a weaker long-term bet than PDFBox itself. It was never added to the build, so table-reconstruction quality on 10-page advices was not measured.