Script for the PDF OCR workflow described in https://ryanfb.github.io/etc/2014/11/13/command_line_ocr_on_mac_os_x.html. See also: https://github.com/fritz-hh/OCRmyPDF

ocrpdf.sh

#!/bin/bash

PREFIX=$(basename "$1" .pdf)
if [ ! -z "$TESSERACT_FLAGS" ]; then
  echo "Picked up TESSERACT_FLAGS: $TESSERACT_FLAGS"
fi
echo "Prefix is: $PREFIX"
echo "Converting to TIFF..."
if command -v parallel >/dev/null 2>&1; then
  LAST_PAGE=$(($(pdfinfo "$1"|grep '^Pages:'|awk '{print $2}') - 1))
  parallel --bar -u convert -density 300 "$1"\[\{1\}\] -type Grayscale -compress lzw -background white +matte -depth 32 "${PREFIX}_page_%05d.tif" ::: $(seq 0 $LAST_PAGE)
else
  convert -density 300 "$1" -type Grayscale -compress lzw -background white +matte -depth 32 "${PREFIX}_page_%05d.tif"
fi
echo "Performing OCR..."
if command -v parallel >/dev/null 2>&1; then
  parallel --bar -u --retries 3 "tesseract $TESSERACT_FLAGS {} {.} pdf 2>/dev/null" ::: "${PREFIX}"_page_*.tif
else
  for i in "${PREFIX}"_page_*.tif; do
    echo $i
    tesseract $TESSERACT_FLAGS "$i" "$(basename "$i" .tif)" pdf 2>/dev/null
  done
fi
echo "Verifying output and converting missing files back without OCR..."
for i in "${PREFIX}"_page_*.pdf; do
  if ! pdfinfo "$i" > /dev/null; then
    rm -v $i
    convert -density 300 $(basename $i .pdf).tif $i
  fi
done
echo "Combining output to ${PREFIX}-OCR.pdf..."
if command -v pdftk >/dev/null 2>&1; then
  pdftk "${PREFIX}"_page_*.pdf cat output "${PREFIX}-OCR.pdf"
else
  gs -q -dNOPAUSE -dBATCH -dProvideUnicode -sDEVICE=pdfwrite -sOutputFile="${PREFIX}-OCR.pdf" "${PREFIX}"_page_*.pdf >/dev/null 2>&1
fi
echo "Cleaning up..."
rm "${PREFIX}"_page_*.tif "${PREFIX}"_page_*.pdf "${PREFIX}"_page_*.txt