parse_hss_audio.py

import requests
from xml.etree import ElementTree
import sys
from urllib.parse import urlparse
import os

# Input is a HTTP Smooth streaming manifest URL (root XML tag is SmoothStreamingMedia)
# This script takes a XML manifest as input and output to stdout audio stream raw data
# Use like this: python3 parse_hss_audio.py http://host.com/stream.isml/Manifest

manifestUrl = urlparse(sys.argv[1])
baseUrl = manifestUrl.scheme + "://" + manifestUrl.netloc + os.path.dirname(manifestUrl.path)

lastAudioTime = 0

while True:
    response = requests.get(manifestUrl.geturl())

    root = ElementTree.fromstring(response.text)
    timeScale = root.get('TimeScale')
    audioStreamUrl = root.find("./StreamIndex[@Type='audio']").get("Url")
    audioBitrates = [ e.get("Bitrate") for e in root.findall("./StreamIndex[@Type='audio']/QualityLevel") ]
    audioFirstTimestamp = int(root.find("./StreamIndex[@Type='audio']/c[1][@t]").get("t"))
    audioDurations = [ int(e.get("d")) for e in root.findall("./StreamIndex[@Type='audio']/c") ]
    audioTimes = [audioFirstTimestamp]
    for duration in audioDurations:
        audioTimes.append(audioTimes[-1] + duration)

    fragmentUrl = baseUrl + "/" + audioStreamUrl
    bitrate = audioBitrates[0]

    for audioTime in audioTimes:
        if lastAudioTime >= audioTime:
            continue
        lastAudioTime = audioTime
        dataRequest = requests.get(fragmentUrl.format(**{'bitrate':bitrate, 'Bitrate':bitrate, 'start time':audioTime, 'start_time':audioTime}))
        sys.stdout.buffer.write(dataRequest.content)