Extract text from copy protected PDFs with qpdf

decrypt_pdf.sh

#!/bin/bash

# Function to decrypt and rename a PDF file
decrypt_and_rename() {
  input_file="$1"
  output_file="${input_file%.pdf}_decrypted.pdf"
  qpdf --decrypt "$input_file" "$output_file"
}

# Prompt the user to enter the folder path
read -p "Enter the folder path to search for PDF files: " search_folder

# Check if the folder exists
if [ ! -d "$search_folder" ]; then
  echo "Error: The specified folder does not exist."
  exit 1
fi

# Use 'find' to locate all PDF files in subdirectories
find "$search_folder" -type f -name '*.pdf' | while read -r pdf_file; do
  echo "Processing: $pdf_file"
  
  # Check if the PDF file has restrictions for text extraction
  encryption_output=$(qpdf --show-encryption "$pdf_file" 2>&1)
  if echo "$encryption_output" | grep -q "extract for any purpose: not allowed"; then
    # Decrypt and rename the PDF file
    decrypt_and_rename "$pdf_file"
    echo "Decrypted and renamed to: ${pdf_file%.pdf}_decrypted.pdf"
  else
    echo "Already decrypted or not encrypted for extraction: $pdf_file"
  fi
done

echo "All PDF files decrypted and renamed."