OCRmyPDF was added to the dedicated worker image to provide a self-hosted fallback before PDF text extraction. The orchestration compiled, but no containerized OCR execution was possible.
- What worked
- Its command-line model fit a separate worker and complemented PDFBox without embedding a remote document-processing service.
- What got in the way
- The image was not built or executed because Docker was unavailable, leaving runtime and output quality unassessed.