this takes a directory of work items (in this case, PDF's in need of OCR) and processes them, skipping the old items. This particular variant uses a file to feed the selection process (allowing a quick distribution of processing over multiple nodes. (poor mans cluster) pypdfocr worked great on this one. I've written this construct a few bazillio…

process_new.sh

#!/bin/bash
DIR=/tmp/

FILES=$DIR/*.pdf

cd $DIR

#for FILE in $FILES; do
while read FILE; do
for 
#  echo ">>"

#  echo "FILE=" $FILE ""
  [[ $FILE =~ ([0-9]+).pdf ]]
  SHORTNAME=${BASH_REMATCH[1]}
  OCR_NAME=$DIR$SHORTNAME"_ocr.pdf"
  echo "OCR_NAME=" $OCR_NAME ""

  if [ -f "$OCR_NAME" ]; then
    echo "File $OCR_NAME exists."
  else
    echo "Processing $OCR_NAME file..."
    # DO NEW WORK HERE
  fi
done <$1