vdiff

#!/bin/bash

# Function to display progress bar
show_progress() {
    local progress=$1
    local total=$2
    local bar_width=75  # Set progress bar width to 75 characters

    # Calculate the percentage completed as an integer
    local percent=$(awk "BEGIN {printf \"%d\", ($progress/$total)*100}")
    
    # Calculate the number of hashes for the progress bar
    local num_hashes=$(awk "BEGIN {printf \"%d\", ($progress/$total)*$bar_width}")
    
    # Ensure num_hashes is at least 1
    if [ "$num_hashes" -lt 1 ]; then
        num_hashes=1
    fi
    
    # Create the progress bar
    local bar=$(printf "%.0s#" {1..5000} | cut -c1-$num_hashes)

    # Print the progress bar and percentage as an integer
    printf "\r[%-${bar_width}s] %s%%" "$bar" "$percent"
}

# Check if both PDFs are provided as arguments
if [ "$#" -lt 3 ]; then
    echo "Usage: $0 <input_pdf1> <input_pdf2> <output_pdf> [--use-cache]"
    exit 1
fi

# Assign input arguments to variables
pdf1=$1
pdf2=$2
output_pdf=$3

# Check if the use-cache flag is set
use_cache=false
if [ "$4" == "--use-cache" ]; then
    use_cache=true
    echo "Cache mode enabled. Cache will be preserved."
fi

# Create cache directory if it doesn't exist
mkdir -p cache

# Get the total number of pages in each PDF using pdfinfo
total_pages_pdf1=$(pdfinfo "$pdf1" | grep Pages | awk '{print $2}')
total_pages_pdf2=$(pdfinfo "$pdf2" | grep Pages | awk '{print $2}')

# Ensure both PDFs have the same number of pages
if [ "$total_pages_pdf1" -ne "$total_pages_pdf2" ]; then
    echo "Error: The two PDFs do not have the same number of pages."
    exit 1
fi

page_count=$total_pages_pdf1

# Output the page count
echo "Page count: $page_count"

# Function to reset progress
reset_progress() {
    echo 0 > "$progress_file"
}

# Progress file to track completed tasks
progress_file=$(mktemp)

# Update progress after each task
update_progress() {
    local total=$1
    local progress=$(cat "$progress_file")
    progress=$((progress + 1))
    echo "$progress" > "$progress_file"
    show_progress "$progress" "$total"
}

# Convert PDF1 and PDF2 to greyscale page by page
echo "Converting to greyscale..."
reset_progress  # Reset progress before starting

for ((i=1; i<=$page_count; i++)); do
    pdf1_grey="cache/pdf1_page_$(printf '%03d' $i).png"
    pdf2_grey="cache/pdf2_page_$(printf '%03d' $i).png"

    if [ "$use_cache" == false ] || [ ! -f "$pdf1_grey" ]; then
        magick -density 300 "$pdf1[$((i-1))]" -colorspace Gray "$pdf1_grey"
    fi

    if [ "$use_cache" == false ] || [ ! -f "$pdf2_grey" ]; then
        magick -density 300 "$pdf2[$((i-1))]" -colorspace Gray "$pdf2_grey"
    fi

    update_progress "$page_count"
done
echo ""

# Generate red-tinted images for PDF1
echo "Generating red-tinted images for PDF1..."
reset_progress  # Reset progress before starting

magick -size 1x256 gradient:white-red "cache/red_gradient.png"
for ((i=1; i<=$page_count; i++)); do
    img="cache/pdf1_page_$(printf '%03d' $i).png"
    red_img="cache/red_page_$(printf '%03d' $i).png"
    if [ "$use_cache" == false ] || [ ! -f "$red_img" ]; then
        magick "$img" -negate cache/red_gradient.png -clut "$red_img"
    fi
    update_progress "$page_count"
done
echo ""

# Generate blue-tinted images for PDF2
echo "Generating blue-tinted images for PDF2..."
reset_progress  # Reset progress before starting

magick -size 1x256 gradient:white-blue "cache/blue_gradient.png"
for ((i=1; i<=$page_count; i++)); do
    img="cache/pdf2_page_$(printf '%03d' $i).png"
    blue_img="cache/blue_page_$(printf '%03d' $i).png"
    if [ "$use_cache" == false ] || [ ! -f "$blue_img" ]; then
        magick "$img" -negate cache/blue_gradient.png -clut "$blue_img"
    fi
    update_progress "$page_count"
done
echo ""

# Combine and process the images, store everything in the cache folder
echo "Processing pages and combining outputs..."
reset_progress  # Reset progress before starting

for ((i=1; i<=$page_count; i++)); do
    grey_page1="cache/pdf1_page_$(printf '%03d' $i).png"
    grey_page2="cache/pdf2_page_$(printf '%03d' $i).png"
    red_page="cache/red_page_$(printf '%03d' $i).png"
    blue_page="cache/blue_page_$(printf '%03d' $i).png"
    
    final_page="cache/final_page_$(printf '%03d' $i).png"

    if [ "$use_cache" == true ] && [ -f "$final_page" ]; then
        update_progress "$page_count"  # Increment progress even if the file already exists (cache hit)
        continue
    fi

    diff_page="cache/diff_page_$(printf '%03d' $i).png"
    mask_page="cache/mask_page_$(printf '%03d' $i).png"
    similar_page="cache/similar_page_$(printf '%03d' $i).png"
    blended_page="cache/blended_page_$(printf '%03d' $i).png"

    # Compute the difference between the two images to find similarities
    magick "$grey_page1" "$grey_page2" -compose Difference -composite "$diff_page"
    
    # Threshold to highlight identical areas
    magick "$diff_page" -threshold 0 -negate "$mask_page"

    # Mask the original image to show only similarities
    magick "$grey_page1" "$mask_page" -compose CopyOpacity -composite "$similar_page"

    # Blend red and blue images at 50%
    magick "$red_page" "$blue_page" -compose Blend -define compose:args=50,50 -composite "$blended_page"

    # Combine the blended red/blue image with the greyscale similarities using Darken
    magick "$blended_page" "$similar_page" -compose Darken -composite "$final_page"

    update_progress "$page_count"
done
echo ""

# Combine all final images into a single output PDF
echo "Combining final images into $output_pdf..."
magick $(ls cache/final_page_*.png | sort) "$output_pdf"

# Compress the final PDF using Ghostscript with PDF compatibility level 1.5 and higher compression (/screen or /ebook)
compressed_output="compressed_output.pdf"
echo "Compressing final PDF for higher compression..."
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.5 -dPDFSETTINGS=/ebook -dNOPAUSE -dQUIET -dBATCH -sOutputFile="$compressed_output" "$output_pdf"

# Move the compressed PDF to overwrite the original output PDF
mv "$compressed_output" "$output_pdf"

# Clean up intermediate images
echo "Cleaning up..."

# Remove the cache folder unless --use-cache flag is present
if [ "$use_cache" == false ]; then
    echo "Removing cache folder..."
    rm -rf cache
fi

# Clean up progress file
rm "$progress_file"

echo "Done! The output is saved as $output"