Generating i-frame and byterange m3u8 playlist from .ts video file for usage in HLS

playlist_builder.rb

module  PlaylistBuilder
  SEG_DURATION, PKT_SIZE, PKT_POS, PKT_TIME = 0,1,2,3

  def self.extract_iframes_data(video_filepath)
    raise "#{video_filepath} does not exists!" unless File.exists?(video_filepath)
    iframes_data = []
    cmd = "ffprobe -show_frames -select_streams v -of compact -show_entries packet=pts_time,codec_type,pos:frame=pict_type,pkt_pts_time,pkt_size,pkt_pos -i #{video_filepath.shellescape}"
    frames_and_packets = nil
    r = Benchmark.measure('') do
      frames_and_packets = `#{cmd}`.split("\n")
    end
    puts r

    iframes = frames_and_packets.grep(/pict_type=I/)
    next_pkt_pts_time = nil
    iframes.each_with_index do |iframe, index|
      pkt_pts_time = next_pkt_pts_time || iframe.match(/.*pkt_pts_time=([0-9]*.[0-9]*).*/)[1]
      if index >= iframes.count - 1
        next_pkt_pts_time = frames_and_packets.last.match(/.*pkt_pts_time=([0-9]*.[0-9]*).*/)[1]
      else
        next_pkt_pts_time = iframes[index + 1].match(/.*pkt_pts_time=([0-9]*.[0-9]*).*/)[1]
      end

      pkt_size = iframe.match(/.*pkt_size=([0-9]*).*/)[1]
      pkt_pos = iframe.match(/.*pkt_pos=([0-9]*).*/)[1]
      segment_duration = (next_pkt_pts_time.to_f - pkt_pts_time.to_f).round(5)

      iframes_data << [segment_duration, pkt_size.to_i, pkt_pos.to_i, pkt_pts_time.to_f.round(5)]
    end
    packets_pos = frames_and_packets.map{|p| p.match(/^packet.*pos=([0-9]*).*/).try{|m| m[1].to_i}}.compact
    packets_pos_reversed = {}
    packets_pos.each_with_index do |pos, i|
      packets_pos_reversed[pos] = i
    end

    iframes_data.each do |iframe_data|
      iframe_size = nil
      if packets_pos_reversed[iframe_data[PKT_POS]] && packets_pos[packets_pos_reversed[iframe_data[PKT_POS]] + 1]
        # https://github.com/pbs/iframe-playlist-generator/blob/master/iframeplaylistgenerator/generator.py
        # "We compared the output of our library to Apple's example streams, and were off by 188 bytes
        # for each I-frame byte-range."
        # biomancer: Even with this fix result is sometimes off by extra 188 (188*2 total), for random chunks, 
        # I have not found any correlation with other packet parameters
        iframe_size = packets_pos[packets_pos_reversed[iframe_data[PKT_POS]] + 1] - iframe_data[PKT_POS] + 188
      end
      iframe_data[PKT_SIZE] = iframe_size if iframe_size
    end

    iframes_data
  end

  def self.build_byterange_playlist(iframes_data, filesize, options = {})
    options[:target_duration] ||= 10
    playlist = <<-EOS_TOKEN
#EXTM3U
#EXT-X-TARGETDURATION:#{options[:target_duration]}
#EXT-X-VERSION:4
#EXT-X-MEDIA-SEQUENCE:0
#EXT-X-PLAYLIST-TYPE:VOD
    EOS_TOKEN

    last_iframe = [0,0,0,0]
    iframes_data.each_with_index do |iframe_data, index|
      duration_from_last_iframe = iframe_data[PKT_TIME] - last_iframe[PKT_TIME]
      # playlist += "\n#DEBUG: duration_from_last_iframe: #{duration_from_last_iframe}\n"
      if !iframes_data[index+1]
        #last part
        if duration_from_last_iframe + iframe_data[SEG_DURATION] > options[:target_duration]
          playlist += byterange_chunk(
              duration_from_last_iframe.round(5),
              iframe_data[PKT_POS] - last_iframe[PKT_POS],
              last_iframe[PKT_POS],
              options[:url_filename]
          )
          last_iframe = iframe_data
          duration_from_last_iframe = 0
        end

        playlist += byterange_chunk(
            (duration_from_last_iframe + iframe_data[SEG_DURATION]).round(5),
            filesize - last_iframe[PKT_POS],
            last_iframe[PKT_POS],
            options[:url_filename]
        )
      elsif iframes_data[index+1][PKT_TIME] - last_iframe[PKT_TIME] >= options[:target_duration]
        #first or mid part
        playlist += byterange_chunk(
            duration_from_last_iframe.round(5),
            iframe_data[PKT_POS] - last_iframe[PKT_POS],
            last_iframe[PKT_POS],
            options[:url_filename]
        )
        last_iframe = iframe_data
      end
    end
    playlist += "#EXT-X-ENDLIST\n"
  end

  def self.build_iframe_playlist(iframes_data, options = {})
    options[:target_duration] ||= iframes_data.map(&:first).max.ceil

    playlist = <<-EOS_TOKEN
#EXTM3U
#EXT-X-TARGETDURATION:#{options[:target_duration]}
#EXT-X-VERSION:4
#EXT-X-MEDIA-SEQUENCE:0
#EXT-X-PLAYLIST-TYPE:VOD
#EXT-X-I-FRAMES-ONLY
    EOS_TOKEN

    iframes_data.each do |iframe_data|
      playlist += byterange_chunk(
          iframe_data[SEG_DURATION],
          iframe_data[PKT_SIZE],
          iframe_data[PKT_POS],
          options[:url_filename]
      )
    end
    playlist += "#EXT-X-ENDLIST\n"
  end

  def self.byterange_chunk(duration, size, pos, file)
    <<-EOS_TOKEN
#EXTINF:#{duration},
#EXT-X-BYTERANGE:#{size}@#{pos}
#{file}
    EOS_TOKEN
  end

  def self.build_byterange_playlist_from_file(video_filepath, output_filepath, options = {})
    iframes_data = extract_iframes_data(video_filepath)
    options[:url_filename] ||= File.basename video_filepath
    playlist = build_byterange_playlist(iframes_data, File.size(video_filepath), options)
    File.open(output_filepath, 'w') {|f| f.write(playlist) }
  end

  def self.build_iframe_playlist_from_file(video_filepath, output_filepath, options = {})
    iframes_data = extract_iframes_data(video_filepath)
    options[:url_filename] ||= File.basename video_filepath
    playlist = build_iframe_playlist(iframes_data, options)
    File.open(output_filepath, 'w') {|f| f.write(playlist) }
  end
end

Usage example:

PlaylistBuilder.build_byterange_playlist_from_file('/home/me/somevideo.ts', '/home/me/byterange_playlist.m3u8', {target_duration: 7})

PlaylistBuilder.build_iframe_playlist_from_file('/home/me/somevideo.ts', '/home/me/iframe_playlist.m3u8', {target_duration: 7})

how about getting validity of an online file ? uri needed.. is that possible ?