Removing unneeded pages from F941

batch-process-f941.sh

#!/bin/bash

INPUT=~/Downloads/f941
DONE=$INPUT/done
OUTPUT=$INPUT/output

mkdir -p $DONE
mkdir -p $OUTPUT

PDFS=$INPUT/*.pdf
FILES=($INPUT/*.pdf)
COUNT=${#FILES[@]}
NTH=0

function process_pdf {
  pdftk "$1" cat ${@:3} output "$OUTPUT/$2"
  if [ $? == 0 ]
  then
    mv "$1" "$DONE"
  else
    echo "!!!!!!!!!! ERRORED"
  fi
}

for f in $PDFS
do
  FILENAME=$(basename "$f")
  NUMBER_OF_PAGES=$(pdfinfo "$f" | grep Pages | awk '{print $2}')

  if [ $NUMBER_OF_PAGES == 4 ] || [ $NUMBER_OF_PAGES == 5 ]
  then
    process_pdf "$f" "$FILENAME" 1 2
  elif [ $NUMBER_OF_PAGES == 6 ]
  then
    process_pdf "$f" "$FILENAME" 1 2 5
  else
    echo "!!!!!!!!!! SKIPPED: File $f has weird number of pages! ($NUMBER_OF_PAGES)"
  fi

  ((++NTH))
  echo "processed $NTH/$COUNT files: $FILENAME"
done

echo "DONE! Check your directory for any skipped files!"