Canon MF421DW - watch and process incoming scans (detect/remove blank pages, deskew, OCR)

pdf-ocr-process.sh

#!/bin/bash

file=$2
basename=${file%%.*}
source_path=$1

if [[ $source_path == */User2/ ]] ;
then
	target_path=/mnt/s/Files/Docs/Scans/User2/
else
	target_path=/mnt/s/Files/Docs/Scans/User1/
fi

work_path=`mktemp -d`
[[ "${work_path}" != */ ]] && work_path="${work_path}/"
rm -f ${work_path}*
cd ${work_path}

convert -quiet -units PixelsPerInch -density 300 ${source_path}${file} ${work_path}${basename}-%02d.png

for filename in ${work_path}*.png; do
	[ -f "$filename" ] || continue
	white=`convert -quiet ${filename} -format "%[mean]" info:`
	if [ ${white%.*} -gt 65400 ]; then
	        mkdir -p ${target_path}EmptyPages/
		mv ${filename} ${target_path}EmptyPages/
	else
		convert -quiet ${filename} -deskew 40 +repage ${filename}-deskewed.png
	fi
done

if [ ! -z "$(ls -A ${work_path})" ]; then
	convert -quiet *-deskewed.png -compress zip ${work_path}${basename}.tif
	tesseract ${work_path}${basename}.tif ${target_path}${basename} -l deu --psm 1 --oem 1 pdf
	rm -rf ${work_path}
	if [ -s ${target_path}${basename}.pdf ]; then
		rm -f ${source_path}${file}
	fi
fi

pdf-ocr-watch.sh

#!/bin/bash

inotifywait -m /mnt/s/Files/Docs/Scans/Incoming/User1 /mnt/s/Files/Docs/Scans/Incoming/User2 -e create -e moved_to |
	while read path action file; do
		if ! [[ $file == cifs_accesibility_check* ]] ;
		then
			echo $file detected
			loopcount=0
			check="initial state";
			while [[ $check != "" ]] ; do
				echo -n .
				loopcount=$[$loopcount +1]
				sleep 0.5
				check=`identify -quiet -format "" ${path}${file} 2>&1`
				if [[ loopcount -gt 200 ]] ;
				then
					break
				fi
			done
			echo ""
			if [[ loopcount -gt 200 ]] ;
			then
				echo Error: ${check}
				echo -e "\a"
				echo -e "\a"
				echo -e "\a"
			else
				filesize=$(stat -c%s "${path}${file}")
				echo processing... ${filesize}
				bash /mnt/c/Programs/Scripts/pdf-ocr-process.sh $path $file
				echo -e "\a"
			fi
		fi
	done

setup.sh

#!/bin/bash

# requires latest tesseract-ocr (4+)
sudo add-apt-repository ppa:alex-p/tesseract-ocr
sudo apt-get update
sudo apt-get install tesseract-ocr tesseract-ocr-deu
sudo apt-get install imagemagick
sudo apt-get install inotify-tools

start-watcher.sh

#!/bin/bash

nohup bash -c '/mnt/c/Programs/Scripts/pdf-ocr-watch.sh' &> /var/log/pdf-ocr.log&