Create a searchable PDF from images or a raster PDF with OCR

#!/bin/bash
echo Creating a searchable PDF for $1
fullname=$(basename "$1")
name=${fullname%.*}

dpi="300" #Predefined definition is 300dpi
language="ita" #Predefined language is italian

if [[ -z $1 ]]; then
echo "Usage:"
echo "searchable-pdf-ocr.sh originalfile.pdf eng 300"
echo "searchable-pdf-ocr.sh image1.png image2.png image3.png"
echo "When you specify a pdf, you can also set the language and resolution (in dpi). When you specify a list of images, you cannot set language or dots-per-inch by command line."
echo "WARNING: This script does not work if your filenames contain spaces."
echo "This script has been written by Luca Tringali - TRINGALINVENT [at] libero.it"
exit 0
fi

if [[ ! -z $2 ]]; then
language="$2"
fi

if [[ ! -z $3 ]]; then
dpi="$3"
fi

echo "Language and dpi:"
echo $language
echo $dpi

if [[ "$1" =~ .*\.(pdf|eps) ]]; then
#or just [[ "$1" == *.pdf  ]]
echo "Extracting pages"
gs -dNOPAUSE -q -r$dpix$dpi -sDEVICE=tiff32nc -dBATCH -sOutputFile="$name-%04d.tmppage.tiff" "$fullname"
images=( *.tmppage.tiff )
else
images=$@
fi

for f in ${images[@]}; do
  echo $f
  if [[ ! "$f" =~ .*\.(tiff) ]]; then
   echo "Converting to TIFF..."
   convert "$f" -background white -flatten +matte "${f%.*}.tiff"
  fi

  echo "Tesseract OCR..."
  tesseract -l $language -psm 3 "${f%.*}.tiff" "${f%.*}ocr" pdf 1>/dev/null 2>&1

  rm "${f%.*}.tiff"
  mv ${f%.*}ocr.pdf $f.tmp.pdf
done

echo "Merging all pages into a single PDF..."
gs -dCompatibilityLevel=1.4 -dNOPAUSE -dQUIET -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -sOutputFile=${name}.searchable.pdf *.tmp.pdf

rm *.tmp.pdf
echo "Created $name.searchable.pdf"