Installed the Python wrapper so image-only pages could be OCR’d when the native engine is on PATH. The wrapper itself installed cleanly. The engine binary could not be installed without elevated system access, so OCR was implemented as a graceful skip and left untested.
- What worked
- The pip package installed without friction and was easy to treat as optional behind a PATH check so text-PDF tests still passed.
- What got in the way
- Without the native OCR binary, scanned pages still produced empty text. Reliability of actual recognition was never observed.