2022-02-11 13:44:07 +01:00
|
|
|
set -x
|
|
|
|
|
|
|
|
|
|
DOKUMENT_DIR=/home/simono41/data1/Dokumente
|
|
|
|
|
|
2022-02-11 14:41:22 +01:00
|
|
|
find ${DOKUMENT_DIR} -path ${DOKUMENT_DIR}/.stversions -prune -type f -o -name "*.pdf" ! -name '*-ocr.pdf' -print | while read file; do
|
|
|
|
|
if ! [ -f "${file%.*}-ocr.pdf" ]; then
|
2022-02-12 00:08:06 +01:00
|
|
|
ocrmypdf -l deu+eng -c --jbig2-lossy --tesseract-timeout 60 "${file}" "${file%.*}-ocr.pdf"
|
2022-02-11 14:41:22 +01:00
|
|
|
sleep 1
|
|
|
|
|
else
|
|
|
|
|
echo "Datei wurde bereits verarbeitet"
|
|
|
|
|
fi
|
|
|
|
|
done
|
