pdf2ocr

#!/bin/bash

if [ -z "$1" ]
  then
    echo "No argument supplied"
    exit 1
fi

if [[ -r $1 ]]; then
    echo "$1 exist"
    file_path=$(realpath -e "$1")
else
    echo "$1 can't be read"
    exit 1
fi

tempdir=$(mktemp -d)
current_dir=$PWD

echo "Created tempdir to work in: $tempdir"
echo "Converting $file_path"

cd $tempdir

pdftoppm -r 300 "$file_path" temp

digits_last_out_file=$(ls -1 | awk 'END{print}' | sed -n 's/temp-\(.*\)\.ppm/\1/p')
size=${#digits_last_out_file}

unpaper -v temp-%0${size}d.ppm utemp-%0${size}d.ppm

out_filename=$(echo "$file_path" | sed 's/.pdf/_ocr/'i )

echo "$(find utemp-*.ppm)"  | tesseract -l deu - "$out_filename" pdf 

cd $current_dir

rm -r $tempdir

echo "done"