Extract photos/video from a Google Takeout tarball

extract-google-photos

#!/usr/bin/env bash

set -euo pipefail

if [[ $# -ne 2 ]] || [[ ! -f "$1" ]] || [[ ! -d "$2" ]]; then
  echo "usage: extract-google-photos [tarfile] [dest-dir]"
  exit 1
fi

if ! command -v exiftool > /dev/null || ! command -v gawk > /dev/null; then
  echo "Please install exiftool and gawk"
  exit 1
fi

[[ "$1" == /* ]] && tarfile="$1" || tarfile="$(pwd)/$1"
[[ "$2" == /* ]] && dest="$2" || dest="$(pwd)/$2"
tmpdir=$(mktemp -d)

echo "Working dir ${tmpdir}"
cd $tmpdir
tar -xzf "$tarfile"

find . -type f -regex '.+\(\(\.jpg\)\|\(\.JPG\)\|\(\.mp4\)\|\(\.m4v\)\)' | while read src_file; do
  file_name=$(basename "$src_file")
  [[ "$file_name" == .* ]] && continue

  subdir=$(exiftool -T -createdate "$src_file" | gawk -F: '/^[0-9]{4}:[0-9]{2}/ { printf "%s/%s", $1, $2 }')
  [[ -z "$subdir" ]] && subdir="Unknown"
  dest_dir="${dest}/${subdir}"
  dest_file="${dest_dir}/${file_name}"

  [[ ! -d "$dest_dir" ]] && mkdir -p "$dest_dir"

  if [[ -f "$dest_file" ]]; then
    [[ $(md5sum "$src_file") == $(md5sum "$dest_file") ]] && continue
    while [[ -f "$dest_file" ]]; do
      dup_file_name=$(basename "$dest_file")
      dest_file="${dest_dir}/2${dup_file_name}"
    done
  fi

  mv "$src_file" "$dest_file"
done

cd ..
rm -r $tmpdir