Automated subtitle file renamer

subrenamer.rb

#!/usr/bin/env ruby
# License: BSD-0-clause

VIDEO_EXTS = %w(.mkv .avi .mp4)
SUBTITLE_EXTS = %w(.ssa .ass .srt)

if ARGV.empty?
  puts "Automatic subtitle file renamer by [email protected]"
  puts "Usage: #{$0} *.ssa *.mkv"
  puts "   or: DRY_RUN=1 #{$0} *.ssa *.mkv"
  puts "Known extensions: #{VIDEO_EXTS.join ' '} #{SUBTITLE_EXTS.join ' '}"
  exit 0
end

videos    = ARGV.select { |f| VIDEO_EXTS.include? File.extname(f) }
subtitles = ARGV.select { |f| SUBTITLE_EXTS.include? File.extname(f) }

if videos.size + subtitles.size != ARGV.size
  puts "Unsupported files: #{ARGV - videos - subtitles}"
  exit 1
end

def common_prefix(strings)
  prefix = strings.first
  old_prefix = prefix
  strings.each do |string|
    until string.start_with? prefix
      prefix = prefix.chop
      if prefix.empty?
        puts "No shared prefix between #{old_prefix.inspect} and #{string.inspect}"
        exit 1
      end
      old_prefix = prefix
    end
  end
  prefix.sub(/\d+$/, '')
end

video_prefix    = common_prefix(videos)
subtitle_prefix = common_prefix(subtitles)

puts "Video prefix:    #{video_prefix.inspect}"
puts "Subtitle prefix: #{subtitle_prefix.inspect}"

def split(prefix, string)
  if string =~ /^(^#{Regexp.escape prefix})(\d+)(.+)$/
    [$1, $2, $3]
  else
    puts "Cannot extract episode number from #{string.inspect}"
    exit 1
  end
end

subtitles_by_episode = {}
subtitles.each do |subtitle|
  prefix, episode, suffix = split(subtitle_prefix, subtitle)
  if subtitles_by_episode.include? episode
    puts "Duplicate subtitle for episode #{episode}"
    exit 1
  end
  subtitles_by_episode[episode] = [prefix, suffix]
end

subtitle_map = {}
videos.each do |video|
  video_prefix, episode, video_suffix = split(video_prefix, File.basename(video, ".*"))
  unless subtitles_by_episode.include? episode
    puts "Episode #{episode} does not have a subtitle track"
    exit 1
  end

  subtitle_prefix, subtitle_suffix = subtitles_by_episode[episode]
  subtitle_map["#{subtitle_prefix}#{episode}#{subtitle_suffix}"] =
    "#{video_prefix}#{episode}#{video_suffix}#{File.extname subtitle_suffix}"
end

non_unique_subtitles = (subtitle_map.values - subtitle_map.values.uniq).uniq
if non_unique_subtitles.any?
  puts "Renamed subtitles are non-unique: #{non_unique_subtitles.inspect}"
  exit 1
end

subtitle_map.each do |from, to|
  puts "#{from.inspect} => #{to.inspect}"
  if File.exists? to
    puts "Renamed subtitle file already exists: #{to.inspect}"
    exit 1
  end
end

unless ENV.include? 'DRY_RUN'
  subtitle_map.each do |from, to|
    File.rename from, to
  end
end