script_with_chop.sh

#!/bin/bash
FILE=./input.doc
if test -f "$FILE"; then
  lowriter --convert-to pdf:writer_pdf_Export input.doc
fi

mkdir image
pdftoppm input.pdf image/page -png

# DUPLICATE
#page=1
#for image_file in ./image/*.png; do
#  additional_zeros=""
#  if [ $page -lt 10 ]
#  then
#    additional_zeros="00"
#  elif [ $page -lt 100 ]
#  then
#    additional_zeros="0"
#  fi
#  cp $image_file "./image/"$additional_zeros$page"A.png"
#  mv $image_file "./image/"$additional_zeros$page"B.png"
#  page=$((page+1))
#done

# CROP NORTH
#for image_file in ./image/*.png; do
#  convert $image_file -gravity North -chop 0x474 $image_file
#done

# CROP SOUTH
#for image_file in ./image/*.png; do
#  convert $image_file -gravity South -chop 0x426 $image_file
#done

# CROP EAST
#for image_file in ./image/*A.png; do
#  convert $image_file -gravity East -chop 825x0 $image_file
#done

# CROP WEST
#for image_file in ./image/*B.png; do
#  convert $image_file -gravity West -chop 825x0 $image_file
#done

mkdir text
page=1
for image_file in ./image/*.png; do
  additional_zeros=""
  if [ $page -lt 10 ]
  then
    additional_zeros="00"
  elif [ $page -lt 100 ]
  then
    additional_zeros="0"
  fi
  tesseract $image_file "./text/"$additional_zeros$page -l spa
  page=$((page+1))
done


for text_file in ./text/*.txt; do
  cat $text_file >> output.txt
done