⚡ Bash script that automates Exiftool to feed back metadata from a Google Photos Takeout Export's JSON sidecar format into the original media files 💡

migrate-takeout.sh

#!/bin/bash

# This is a small helper script that utilizes Exiftool to feed back metadata from a Google Photos Takeout Export's JSON sidecar format
# into the original media files

# Usage: ./migrate-takeout -i <input_dir> -o <output_dir> -e <exiftool_path>

# Tells Bash to stop on errors if variables are unset
set -eu

# Parse command line options
while getopts i:o:e: opt ; do
	case "$opt" in
		i) input_dir=$OPTARG ;;
		o) output_dir=$OPTARG ;;
		e) exiftool_path=$OPTARG ;;
		*) echo -e "\033[0;33mInvalid options specified. Usage: ./migrate-takeout -i <input_dir> -o <output_dir> -e <exiftool_path>\033[0m" ; exit 2 ;;
	esac
done
shift $((OPTIND-1))

echo -e "\033[0;36mRunning Google Photos migration script on input \"$input_dir\" and output \"$output_dir\".\033[0m"

# Check if input dir exists
if [ ! -d "$input_dir" ]; then
	echo -e "\033[0;33mInput dir \"$input_dir\" does not exist, exiting.\033[0m"
	exit
fi

# Convert relative paths to absolute
input_dir=$(realpath "$input_dir")
output_dir=$(realpath "$output_dir")
exiftool_path=$(realpath "$exiftool_path")

# Create the output directory (and ignore if it already exists)
mkdir -p "$output_dir"

# Iterate recursively over all media files in the target directory
find "$input_dir" -type f -iname "*.jpg" -o -iname "*.png" -o -iname "*.mp4" -o -iname "*.mov" | while read -r file;
do
	# Figure out a few variables based on the file name
	media_file="$file"
	extension="${file##*.}"
	file_name=$(basename "$media_file")
	base_name="${file_name%.*}" # Remove extension
	out_file="${output_dir}/${base_name}.$extension"

	# Figure out the path of the JSON sidecar. Since JSON sidecar names can be cropped
	# (e.g. Screenshot_2022-12-01-06-44-03-257_com.snapchat.jpg => Screenshot_2022-12-01-06-44-03-257_com.snapcha.json)
	# First, lets check if the file is already present without any changes 
	json_sidecar="${file_name}.supplemental-metadata.json"
	json_file=$(find "$input_dir" -type f -name "$(basename "$json_sidecar")" | head -n 1)

	# If JSON file not found yet, apply progressive trimming to handle truncation
	if [ ! -f "$json_file" ]; then
		temp_name="${file_name}.supplemental-metadata"
		while [ -z "$json_file" ] && [ ${#temp_name} -gt 0 ]; do
			json_file=$(find "$input_dir" -type f -name "$(basename "$temp_name").json" | head -n 1)
			temp_name="${temp_name:0:-1}" # Remove last character from temp_name if no match
		done
	fi

	# Check if file exists now
	if [ ! -f "$json_file" ]; then
		echo -e "\033[0;33mFile \"$media_file\" has no JSON sidecar, ignoring.\033[0m"
		continue # Skip if file is still not existing
	fi

	# Retrieve a few params from JSON sidecar (will return nothing if value is null or zero)
	latitude=$(jq '.geoData.latitude | select( . != null and . != 0 )' "$json_file")
	longitude=$(jq '.geoData.longitude | select( . != null and . != 0 )' "$json_file")
	altitude=$(jq '.geoData.altitude | select( . != null and . != 0 )' "$json_file")
	google_photos_origin=$(jq '.googlePhotosOrigin.mobileUpload.deviceType | select( . != null )' "$json_file" | tr -d '"')

	echo -e "\033[0;36mProcessing $media_file => $out_file\033[0m"

	# A few Exiftool parameters explained:
	# 	-q				quiet processing
	#	-dateFormat		how to format dates (%s means as a string, which uses internal UNIX timestamp conversion then)
	#	-tagsFromFile	instead of loading EXIF metadata tags from the input media file, load tags from this file instead (JSON is supported)
	#	-o				store the output in the following path

	# The "stringified" arguments are all EXIF tags that are written. Access to the tags loaded from -tagsFromFile is allowed here, which gives us access to
	# stuff like "PhotoTakenTimeTimestamp" or "CreationTimeTimestamp" which happen to be part of the JSON. The rest is fed in from jq.
	exiftool_common_options=( 
		-q
		-dateFormat "%s"
		-tagsFromFile "$json_file"
	)

	exiftool_date_options=( 
		"-AllDates<PhotoTakenTimeTimestamp"
		"-XMP-Exif:DateTimeOriginal<PhotoTakenTimeTimestamp"
		"-PNG:CreationTime<PhotoTakenTimeTimestamp"
		"-QuickTime:TrackCreateDate<PhotoTakenTimeTimestamp"
		"-QuickTime:TrackModifyDate<PhotoTakenTimeTimestamp"
		"-QuickTime:MediaCreateDate<PhotoTakenTimeTimestamp"
		"-QuickTime:MediaModifyDate<PhotoTakenTimeTimestamp"
		"-DateTimeOriginal<PhotoTakenTimeTimestamp"
		"-FileCreateDate<PhotoTakenTimeTimestamp"
		"-FileModifyDate<PhotoTakenTimeTimestamp"
	)

	exiftool_gps_options=( 
		"-GpsLatitude*=$latitude"
		"-GpsLongitude*=$longitude"
		"-GpsAltitude*=$altitude"
		-GPSCoordinates="$latitude, $longitude, $altitude"
	)

	exiftool_keyword_options=(
		-Keywords+="$google_photos_origin" # This is mostly used for photos
	)

	# Test if GPS coordinates were found
	if [ -z "$latitude" ] || [ -z "$longitude" ]
	then
		# Run command line with just dates, no GPS data
		$exiftool_path "${exiftool_common_options[@]}" "${exiftool_date_options[@]}" "${exiftool_keyword_options[@]}" -o "$out_file" "$media_file"
	else
		# Run command line with GPS data
		$exiftool_path "${exiftool_common_options[@]}" "${exiftool_date_options[@]}" "${exiftool_gps_options[@]}" "${exiftool_keyword_options[@]}" -o "$out_file" "$media_file"
	fi
done