Script to process SSRN articles for enlarged printing or viewing on e-reader

process_ssrn_pdf.sh

#!/bin/bash

# Function to parse arguments
parse_arguments() {
  while getopts ":m:" opt; do
    case $opt in
      m)
        margin="$OPTARG"
        ;;
      \?)
        echo "Invalid option: -$OPTARG"
        exit 1
        ;;
      :)
        echo "Option -$OPTARG requires an argument."
        exit 1
        ;;
    esac
  done
  shift $((OPTIND - 1))

  # Default margin
  if [ -z "$margin" ]; then
    margin=15
  fi

  input_file="$1"

  if [ -z "$input_file" ]; then
    echo "Input file is required."
    exit 1
  fi
}

# Parse arguments
parse_arguments "$@"

# Temporary file names
TEMP1=TEMPORARY01.pdf
TEMP2=TEMPORARY02.pdf
TEMP3=TEMPORARY03.pdf
TEMP4=TEMPORARY04.pdf

# Function to handle errors
handle_error() {
  echo "An error occurred: $1"
  rm -f "$TEMP1" "$TEMP2" "$TEMP3" "$TEMP4"
  exit 1
}

# Uncompress PDF
qpdf --stream-data=uncompress "$input_file" "$TEMP1" || handle_error "qpdf failed"

# Remove "Electronic copy available at:" text
sed 's/Electronic copy available at: .*//' < "$TEMP1" > "$TEMP2" || handle_error "sed failed"

# Optimize PDF
mutool clean "$TEMP2" "$TEMP3" || handle_error "mutool clean failed"

# Compress PDF
qpdf --compress-streams=y --object-streams=generate "$TEMP3" "$TEMP4" || handle_error "qpdf failed"

# Crop PDF
pdfcrop --margins "$margin" "$TEMP4" "${input_file%.*}_processed.${input_file##*.}" || handle_error "pdfcrop failed"

# Remove temporary files
rm -f "$TEMP1" "$TEMP2" "$TEMP3" "$TEMP4"

echo "Processing completed successfully. Output file: ${input_file%.*}_processed.${input_file##*.}"