A script to OCR a pdf file

ocrpdf

#!/usr/bin/env sh

set -e

if [ ! -f "$1" ]; then
  echo "Input file doesn't exist"
  exit 1
fi
if [ ! "$2" ]; then
  echo "Must provide output file"
  exit 1
fi

input="$(realpath "$1")"
output="$(realpath "$2")"

shift 2

tmpdir="$(mktemp -d)"
cd "$tmpdir"

cleanup() {
  cd /
  rm -rf "$tmpdir"
}

trap cleanup 2

pdftoppm "$input" img -png

find . -name 'img-*.png' | sort >fileslist

ln -s "$output" out.pdf

touch out.pdf

tesseract fileslist out $@ pdf

cleanup