caption_shifter.rb

#!/usr/bin/env ruby
require 'time'

if ARGV.size < 2
  fail "Please supply shift modifier\n" +
    "USAGE: ./caption_shifter <caption-input> <shift-modifier>\n" +
    "EXAMPLE: ./caption_shifter html.vtt 2"
elsif ARGV.size < 1
  fail "Please supply caption file\n" +
    "USAGE: ./caption_shifter <caption-input> <shift-modifier>\n" +
    "EXAMPLE: ./caption_shifter html.vtt 2"
end

caption_file = ARGV[0]
shift_modifier = ARGV[1]
output_file = "#{File.basename(caption_file, File.extname(caption_file))}_shifted.vtt"

puts "caption_file: #{caption_file}, modifier: #{shift_modifier}"

module CaptionShifter

  def self.parse(lines)
    captions = []
    grouped_lines = []
    lines.shift while lines.first.empty?
    lines.each_with_index do |line, index|
      line.strip!
      if line.empty? && grouped_lines.length > 0
        captions << Caption.parse(grouped_lines)
        grouped_lines = []
        next
      end

      unless line.empty?
        grouped_lines << line
      end

      if grouped_lines.length > 0 && lines.length == index + 1
        captions << Caption.parse(grouped_lines)
      end
    end
    captions
  end

  Caption = Struct.new(:id, :start_time, :end_time, :caption) do
    TIME_FORMAT = "%H:%M:%S,000"

    def self.parse(lines)
      lines.shift while lines.first.empty?
      id = lines[0].chomp
      start_time, end_time = lines[1].chomp.split(" --> ")
      caption = lines[2..-1].map(&:chomp).join(" ")
      new(id, start_time, end_time, caption)
    end

    def to_s
      [id, time_range, caption].join("\n")
    end

    def time_range
      "#{start_time} --> #{end_time}"
    end

    def shift!(modifier)
      modifier = modifier.to_i
      _start_time = Time.parse(start_time.gsub(/,\d+$/, ''))
      _end_time = Time.parse(end_time.gsub(/,\d+$/, ''))
      unless [_start_time.hour, _start_time.min, _start_time.sec].all? {|time| time == 0}
        self.start_time = (_start_time + modifier).strftime(TIME_FORMAT)
      end
      unless [_end_time.hour, _end_time.min, _end_time.sec].all? {|time| time == 0}
        self.end_time = (_end_time + modifier).strftime(TIME_FORMAT)
      end
    end
  end
end


captions = CaptionShifter.parse(File.readlines(caption_file))
captions.each { |caption| caption.shift!(shift_modifier) }

File.open(output_file, 'w+') { |file| file.write(captions.join("\n\n") + "\n\n") }

puts "wrote #{captions.length} captions to #{output_file}"