Ruby library to generate webvtt formatted transcripts from an array of timestamped words.

webvtt.rb

module Webvtt
  DEFAULT_MAX_CHAR_LENGTH = 40

  class Cue
    attr_accessor :start, :end, :words
    def initialize(words)
      self.words = words
    end

    def to_s
      "#{title}\n#{text}"
    end

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

    def text
      self.words.map{|w| w['word']}.join(' ')
    end

    def start_time
      Time.at(self.words.first['starts_at']).strftime("00:%M:%S.%L") #not using %H
    end

    def end_time
      Time.at(self.words.last['ends_at']).strftime("00:%M:%S.%L")
    end

  end


  class Vtt
    attr_accessor :words, :max_line_length

    def initialize(words, options={})
      self.words = words
      self.max_line_length = options[:max_line_length] || DEFAULT_MAX_CHAR_LENGTH
    end

    def cues
      @cues ||= begin
        all_cues = []
        current_cue = []
        line_length = 0
        words.each do |word|
          if (line_length + ' '.length + word['word'].length <= self.max_line_length)
            current_cue << word
            line_length = line_length + word['word'].length
          else
            all_cues << Cue.new(current_cue)
            current_cue = [word]
            line_length = 0
          end
        end
        all_cues << Cue.new(current_cue) unless current_cue.empty?
        all_cues
      end
    end


    def to_s
      @vtt ||= begin
        lines = []
        lines << "WEBVTT\n"

        cues.each_with_index do |cue, i|
          lines << (i+1).to_s
          lines << "#{cue.to_s}\n"
        end
        lines.join("\n")
      end
    end
  end

end