download hls segments from m3u8

hlsFetcher.py

#!/usr/bin/env python3
# Http Live Streaming -- fetcher/downloader
# A simple script to download ts segments/m3u8 files from given url, including variant streams from master m3u8
# require: python3

import sys
import os
import urllib.request

TEST_URL = 'https://devimages.apple.com.edgekey.net/streaming/examples/bipbop_4x3/bipbop_4x3_variant.m3u8'


def isValidUrl(url: str) -> bool:
    if url == '':
        print('Invalid URL: empty url')
        return False
    elif not (url.startswith('http') or url.startswith('https')):
        print('Invalid URL: require \'http/https\' url')
        return False
    elif os.path.splitext(url)[1].lower() != '.m3u8':
        print('Invalid URL: not hls source')
        return False
    else:
        return True


def readDataFromUrl(url: str) -> bytes:
    with urllib.request.urlopen(url) as response:
        data = response.read()
    return data


def writeFile(path: str, filename: str, data: bytes) -> None:
    fullPath = os.path.join(path, filename)
    full_dir_name = os.path.dirname(fullPath)
    # create directory first, if it doesn't exist
    os.makedirs(full_dir_name, exist_ok=True)
    with open(fullPath, 'wb') as file:
        file.write(data)
    return None


def parseM3U8(baseDir: str, baseUrl: str, data: bytes) -> None:
    for line in data.splitlines():
        line = line.strip()
        extension = os.path.splitext(line)[1]
        if extension.lower() == b'.ts' or extension.lower() == b'.aac':
            tsUrl = baseUrl + '/' + line.decode()
            print('downloading ', tsUrl)
            tsData = readDataFromUrl(tsUrl)
            writeFile(baseDir, line.decode(), tsData)
        elif extension.lower() == b'.m3u8':
            simpleUrl = baseUrl + '/' + line.decode()
            binDir = os.path.join(baseDir, simpleUrl.split('/')[-2])
            m3u8Name = os.path.basename(simpleUrl)
            print('In master m3u8, processing ', simpleUrl)
            if not os.path.exists(binDir):
                try:
                    os.mkdir(binDir)
                except Exception as e:
                    print(e, ' Create ', binDir, ' failed, exit.')
                    return
            m3u8Data = readDataFromUrl(simpleUrl)
            writeFile(binDir, m3u8Name, m3u8Data)
            parseM3U8(binDir, os.path.dirname(simpleUrl), m3u8Data)


def fetchData(url: str) -> bool:
    curPath = os.path.abspath(os.curdir)
    baseUrl = os.path.dirname(url)
    m3u8Name = os.path.basename(url)
    binDir = os.path.join(curPath, url.split('/')[-2])
    print(curPath, baseUrl, binDir)
    if not os.path.exists(binDir):
        try:
            os.mkdir(binDir)
        except Exception as e:
            print(e, ' Create ', binDir, ' failed, exit.')
            return False
    m3u8Data = readDataFromUrl(url)
    writeFile(binDir, m3u8Name, m3u8Data)
    parseM3U8(binDir, baseUrl, m3u8Data)
    return True


if __name__ == "__main__":
    if len(sys.argv) != 2:
        print('Invalid arguments: require 1 parameter, but you gave ', len(sys.argv) - 1)
        exit(-1)
    url = sys.argv[1]
    #url = TEST_URL
    if isValidUrl(url):
        fetchData(url)
        print('Done')
    else:
        exit(-1)

This still misses URIs in EXT-X-MEDIA tags, e.g. 105560_audio_1_stereo_128000.m3u8 in

#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio_high",NAME="english stereo",LANGUAGE="en",AUTOSELECT=YES,URI="105560_audio_1_stereo_128000.m3u8"