This is a minimal wrapper around the jbarlow83/ocrmypdf container which
adds an inotify-based directory watcher and automatically runs OCR processing
on all incoming files in the input directory.
Think of this as a simplified version of cmccambridge/ocrmypdf-auto,
where ocrallmypdf is using the existing tools incron and task-spooler
to achieve a similar effect.
Use your container runtime of choice and run:
docker run -d \
-v /path/to/input:/input \
-v /path/to/output:/output \
ghcr.io/ansemjo/ocrallmypdf
Existing files in the input directory are not processed. The processing only
triggers on CLOSE_WRITE and MOVED_TO events.
The container can be customized with a number of environment variables:
| env | description | default |
|---|---|---|
INPUT |
watched input directory inside of the container | /input |
OUTPUT |
output directory for processed file in the container, be careful not to create an inotify loop by using the same directories! | /output |
OCR_OPTIONS |
options passed to ocrmypdf command |
--clean --deskew --output-type pdfa --skip-text |
OCR_LANGUAGE |
language used with tesseract (-l flag of ocrmypdf), missing languages will be installed on startup |
deu+eng |
REMOVE_ORIGINAL |
truthy value whether to remove original input files | yes |
JOBS |
number of concurrent jobs in task spooler, using too many at once may lead to ocr timeouts, since ocrmypdf uses all cores per process by default |
1 |