Outputs chunks for a specific section of a Twitch VOD

twitch_vod_clipper.py

import requests
import m3u8
import sys
import re

CHUNK_RE = "(.+\.ts)\?start_offset=(\d+)&end_offset=(\d+)"
VOD_API = "https://api.twitch.tv/api/vods/{}/access_token"
INDEX_API = "http://usher.twitch.tv/vod/{}"


def get_vod_index(vod_id):
    # Get access code
    url = VOD_API.format(vod_id)
    r = requests.get(url)
    data = r.json()


    # Fetch vod index
    url = INDEX_API.format(vod_id)
    payload = {'nauth': data['token'], 'nauthsig': data['sig']}
    r = requests.get(url, params=payload)

    m = m3u8.loads(r.content)
    url = m.playlists[0].uri
    return m3u8.load(url)


def extract_clip(index, start_time, end_time):
    # Get the piece we need
    position = 0
    pieces = []

    for seg in index.segments:
        # Add duration of current segment
        position += seg.duration

        # Check if we have gotten to the start of the clip
        if position < start_time:
            continue

        # Extract clip name and byte range
        p = re.match(CHUNK_RE, seg.absolute_uri)
        filename, start_byte, end_byte = p.groups()

        # If we have a new file, add it tot he list
        if not pieces or pieces[-1][0] != filename:
            pieces.append([filename, start_byte, end_byte])
        else: # Else, update the end byte
            pieces[-1][2] = end_byte

        # Check if we have reached the end of clip
        if position > end_time:
            break

    return pieces


if __name__ == '__main__':
    vod_id = sys.argv[1]
    start = int(sys.argv[2])
    end = int(sys.argv[3])

    index = get_vod_index(vod_id)
    pieces = extract_clip(index, start, end)

    # Output all the chunks we need
    for part in pieces:
        print "{}?start_offset={}&end_offset={}".format(*part)