A regression has affected the ability to OCR certain image types.
tesserocr, which ingest-file uses for OCR, is displaying a surprising piece of behaviour which Aleph users have also noticed - they could no longer OCR JPEG images. This is due to the fact that the pre-compiled binaries aren't compiled with jpeg support anymore, nor support for a few other file formats.
This PR forces ingest-file to build tesserocr instead of using the binary wheel, and adds a JPEG test that can catch the regression.
This PR also introduces the ingestors clear-cache command, which takes a prefix and can delete all ingest cache entries.
A regression has affected the ability to OCR certain image types.
tesserocr, whichingest-fileuses for OCR, is displaying a surprising piece of behaviour which Aleph users have also noticed - they could no longer OCR JPEG images. This is due to the fact that the pre-compiled binaries aren't compiled with jpeg support anymore, nor support for a few other file formats.This PR forces
ingest-fileto buildtesserocrinstead of using the binary wheel, and adds a JPEG test that can catch the regression.This PR also introduces the
ingestors clear-cachecommand, which takes a prefix and can delete all ingest cache entries.