AppleStreamDump

README.md

AppleStreamDump

Dumps apple HLS stream format - including alternative audio tracks and all subtitles

Dependencies

• m3u8
• ffmpy (and ffmpeg)
• pycountry

Caveats

• mov_text format is used for webvtt conversion - thus output container should support mov_text (e.g., mp4)

main.py

import m3u8
import sys
import ffmpy
import pycountry


def parse_main_m3u8(main_m3u8_url):
    main_playlist = m3u8.load(main_m3u8_url)

    video_playlist_uri = get_main_video_stream(main_playlist)
    audio_playlist_media_list = get_audio_streams(main_playlist)
    sub_playlist_media_list = get_sub_streams(main_playlist)

    return video_playlist_uri, audio_playlist_media_list, sub_playlist_media_list


def get_main_video_stream(main_playlist):
    bandwidth = -1
    for playlist in main_playlist.playlists:
        if playlist.stream_info.bandwidth > bandwidth:
            bandwidth = playlist.stream_info.bandwidth

    for playlist in main_playlist.playlists:
        if playlist.stream_info.bandwidth == bandwidth:
            return playlist.uri


def get_audio_streams(main_playlist):
    to_return = list()
    for media in main_playlist.media:
        if media.type == "AUDIO":
            to_return.append(media)

    return to_return


def get_sub_streams(main_playlist):
    to_return = list()
    for media in main_playlist.media:
        if media.type == "SUBTITLES":
            to_return.append(media)

    return to_return


def main():
    if len(sys.argv) != 3:
        print("Usage: ", sys.argv[0], " (stream file name) (output_file_name)")
        exit(-1)

    video_m3u8_url, audio_playlist_media_list, sub_playlist_media_list = parse_main_m3u8(sys.argv[1])

    metadata_string = ""
    for idx, audio_media in enumerate(audio_playlist_media_list):
        metadata_string = metadata_string + \
                          " -metadata:s:a:" + str(idx) + ' name="' + audio_media.name + '"' + \
                          " -metadata:s:a:" + str(idx) + ' language="' + \
                          pycountry.languages.get(alpha_2=audio_media.language).alpha_3 + '"'

    for idx, sub_media in enumerate(sub_playlist_media_list):
        metadata_string = metadata_string + \
                          " -metadata:s:s:" + str(idx) + ' name="' + sub_media.name + '"' +\
                          " -metadata:s:s:" + str(idx) + ' language="' + \
                          pycountry.languages.get(alpha_2=sub_media.language).alpha_3 + '"'

    map_string = ""

    # Hard coded video stream count
    stream_count = 1
    for i in range(0, len(audio_playlist_media_list)):
        stream_count += 1
    for i in range(0, len(sub_playlist_media_list)):
        stream_count += 1

    for i in range(0, stream_count):
        map_string = map_string + " -map " + str(i)

    stream_input_dict = dict()
    stream_input_dict[video_m3u8_url] = None
    for audio_media in audio_playlist_media_list:
        stream_input_dict[audio_media.uri] = None
    for sub_media in sub_playlist_media_list:
        stream_input_dict[sub_media.uri] = None

    ffmpy.FFmpeg(
        inputs=stream_input_dict,
        outputs={sys.argv[2]: map_string + ' -c:a copy -c:v copy -c:s mov_text' + metadata_string}
    ).run()


if __name__ == '__main__':
    main()