Photo Copy Script

copy_media.sh

#!/bin/bash
# Usage: process_media_parallel.sh /path/to/source /path/to/destination
# Requirements: ImageMagick v7 (for the "magick" command)

# copies all mov/mp4/pdfs as is	
# copies all images (png/jpg/tiff/heic) to jpgs with max size of 3mb	
# copies images from source nested folders into top level destination folder, with src foldername in filename	
# creates txt file reporting all files observed but not copied	
# Iterated on w GPT o1	

# ps run this to make macs stop generating thumbnails for network storage folders of images, cant view 	
# folders wout rainbow spin w default thumbnail gen.	
# `defaults write com.apple.desktopservices DSDontWriteNetworkStores -bool true`	
# `killall Finder`

if [ "$#" -ne 2 ]; then
  echo "Usage: $0 source_directory destination_directory"
  exit 1
fi

src="$1"
dest="$2"

if [ ! -d "$src" ]; then
  echo "Source directory does not exist."
  exit 1
fi

mkdir -p "$dest"

# File to log unprocessed files
unprocessed="$dest/unprocessed_files.txt"
: > "$unprocessed"  # initialize log

if ! command -v magick >/dev/null 2>&1; then
  echo "Error: ImageMagick 'magick' command not found. Please install ImageMagick v7 or later."
  exit 1
fi

# Maximum number of parallel jobs
max_jobs=16

preserve_metadata() {
  local src="$1"
  local dest="$2"

  touch -r "$src" "$dest"

  # Copy all metadata using exiftool
  if command -v exiftool >/dev/null 2>&1; then
    exiftool -overwrite_original -TagsFromFile "$src" -all:all "$dest"
  fi

  # Optionally, preserve the creation date using SetFile (requires Xcode Command Line Tools)
  if command -v SetFile >/dev/null 2>&1; then
    local birth creation_date
    birth=$(stat -f %B "$src")
    creation_date=$(date -r "$birth" "+%m/%d/%Y %H:%M:%S")
    SetFile -d "$creation_date" "$dest"
  fi
}

process_file() {
  local file="$1"

  local ext="${file##*.}"
  local ext_lower
  ext_lower=$(echo "$ext" | tr '[:upper:]' '[:lower:]')
  local relpath dir newname folderprefix output

  # Compute relative path from the source and build the new name.
  relpath="${file#$src/}"
  dir=$(dirname "$relpath")
  if [ "$dir" = "." ]; then
    newname="$(basename "$file")"
  else
    folderprefix=$(echo "$dir" | tr '/' '_')
    newname="${folderprefix}_$(basename "$file")"
  fi

  case "$ext_lower" in
    jpg|jpeg|png|tif|tiff|heic|heif|gif)
      output="$dest/${newname%.*}.jpg"
      echo "Processing image: $file -> $output"
      
      # Determine max file size based on extension.
      ext="${file##*.}"
      if [[ "$ext" =~ ^(tif|tiff)$ ]]; then
        max_size=7000000 # 7mb
      else
        max_size=3000000 # 3mb
      fi

      magick "$file" -define jpeg:extent="$max_size" "$output"
      preserve_metadata "$file" "$output"
      ;;
    mov|mp4|pdf)
      # Copy video/pdf files as-is.
      output="$dest/$newname"
      echo "Copying video or pdf: $file -> $output"
      cp "$file" "$output"
      ;;
    *)
      echo "Skipping file (unprocessed format): $file"
      echo "$file" >> "$unprocessed"
      ;;
  esac
}

# Export variables for subshell access
export src dest unprocessed

# Process files in parallel using process substitution.
while IFS= read -r -d '' file; do
  process_file "$file" &
  # echo "$file" &
  # Limit to $max_jobs concurrent jobs.
  while [ "$(jobs -rp | wc -l)" -ge "$max_jobs" ]; do
    sleep 0.2
  done
  # macOS creates hidden AppleDouble files (prefixed with ._) to store extended
  # attributes and resource fork data. When you run find, it lists both the 
  # actual file and its corresponding ._ metadata file, which is why you might 
  # see duplicates.
done < <(find "$src" \( -type d \( -name "originals" -o -name "thumbs" \) -prune \) -o -type f ! -name '._*' -print0)


wait  # Wait for all background jobs to finish

echo "Processing complete. Unprocessed files are listed in: $unprocessed"

echo "Files in src:"
find $src \( -type d \( -name "originals" -o -name "thumbs" \) -prune \) -o -type f | wc -l

echo "Files in destination:"
find $dest -type f | wc -l

# ddrescue /dev/sr0 /mnt/user/mitchinson_primary/ArchivalStuff/gma_feb_25/cds/9_cd.iso ~/9_cd.log