lightroom-recover.rb

#!/usr/bin/env ruby
#
# Recreate Lightroom catalog by replacing corrupt files with recovered copies
#
# Keeps the directory structure and creates the new catalog in a new path,
# which means you'll need at least 3x the size of your photos of free space.
#
# Dependencies:
#  - ruby
#  - exiftool
#
# See: https://ptrchm.com/blog/how-i-nearly-lost-all-my-photos/

require 'yaml'
require 'fileutils'

# Replace these:
CORRUPT_CATALOG_PATH = '/Volumes/Photos Piotr'
RECOVERED_FILES_PATH = '/Volumes/home/foo'
NEW_CATALOG_PATH     = '/Volumes/Photos LR'

CATALOG_INDEX_CACHE_PATH  = './catalog_index.yml'
RECOVERY_INDEX_CACHE_PATH = './recovery_index.yml'

# 1. Scan existing catalog and create index of: file_path => timestamp
#    This can take a while, depending on the size of the catalog
unless File.exists?(CATALOG_INDEX_CACHE_PATH)
  catalog_index = {}

  puts
  puts "1. Scanning existing catalog..."
  Dir["#{CORRUPT_CATALOG_PATH}/**/*"].each do |filename|
    next unless File.file?(filename)

    puts "\t#{filename}"

    timestamp = `exiftool -s -s -s -ModifyDate '#{filename}'`.strip
    puts "\t\t#{timestamp}"

    catalog_index[filename] = timestamp
  end
  puts "\n\tDone.\n\n"

  File.open(CATALOG_INDEX_CACHE_PATH, "w") { |file| file.write(catalog_index.to_yaml) }
end

# 2. Scan recovered files and create index of: timestamp => file_path
#    NOTE: timestamps for XMP sidecar files are the same as ARW files,
#          so multiple files can be assigned to 1 timestamp
unless File.exists?(RECOVERY_INDEX_CACHE_PATH)
  recovery_index = {}

  puts "2. Scanning recovered files..."
  Dir["#{RECOVERED_FILES_PATH}/**/*"].each do |filename|
    next unless File.file?(filename)

    puts "\tFile: #{filename}"

    timestamp = `exiftool -s -s -s -ModifyDate '#{filename}'`.strip
    puts "\t\t#{timestamp}"

    next if timestamp.empty?

    recovery_index[timestamp] ||= []
    recovery_index[timestamp] << filename
  end
  puts "\n\tDone.\n\n"

  File.open(RECOVERY_INDEX_CACHE_PATH, "w") { |file| file.write(recovery_index.to_yaml) }
end

# 3. Recreate catalog in a new destination
catalog_index  = YAML.load(File.read(CATALOG_INDEX_CACHE_PATH))
recovery_index = YAML.load(File.read(RECOVERY_INDEX_CACHE_PATH))
missing_files  = []

catalog_index.each do |filename, timestamp|
  next if timestamp.empty?

  recovered = false

  print "Recovering file: #{filename} | #{timestamp}... "

  path     = File.dirname(filename)
  basename = File.basename(filename)
  extname  = File.extname(filename)
  new_path = path.gsub(CORRUPT_CATALOG_PATH, NEW_CATALOG_PATH)

  # Recreate directory in a new path
  FileUtils.mkdir_p(new_path)

  # Find the corresponding recovered file and copy it to the new catalog
  if recovery_index[timestamp]
    # Compare extensions in case multiple files are stored under the same timestamp
    recovery_index[timestamp].each do |recovered_filename|
      next if extname.downcase != File.extname(recovered_filename).downcase

      FileUtils.cp(recovered_filename, "#{new_path}/#{basename}")
      recovered = true
      print "OK"
      break
    end
  end

  puts

  unless recovered
    puts "\tFile not recovered...\n\n"
    missing_files << filename
  end
end

puts "ALL DONE!"
puts "\tFiles to recover: #{catalog_index.keys.count}"
puts "\tMissing files: #{missing_files.count}"

missing_files.each do |missing|
  puts "\t\t#{missing}"
end