大家好，我是小小明。

最近看到几个视频网站的地址依然是m3u8格式，不禁有了使用python进行下载的想法，虽然下载m3u8格式视频的工具很多，但如果我们自行编码就能应对更多的情况。

关于m3u8的基础知识可以参考：Python实时下载B站直播间视频（M3U8视频流）

下面我们将使用Python下载m3u8格式的加密离线视频流。

游览器抓包过滤能够获取该影片的m3u8播放地址：

首先，测试一下该地址：

import m3u8

headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36"
}

playlist = m3u8.load(
    uri='https://vod8.wenshibaowenbei.com/20210628/g4yNLlI7/index.m3u8', headers=headers)
playlist.data

{'media_sequence': None,
 'is_variant': True,
 'is_endlist': False,
 'is_i_frames_only': False,
 'is_independent_segments': False,
 'playlist_type': None,
 'playlists': [{'uri': '/20210628/g4yNLlI7/1000kb/hls/index.m3u8',
   'stream_info': {'program_id': 1,
    'bandwidth': 1000000,
    'resolution': '1280x720'}}],
 'segments': [],
 'iframe_playlists': [],
 'media': [],
 'keys': [],
 'rendition_reports': [],
 'skip': {},
 'part_inf': {},
 'session_data': [],
 'session_keys': []}

从结果看到，这是一个嵌套的地址。

所以写个方法解析真实地址：

import m3u8

headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36"
}


def get_real_url(url):
    playlist = m3u8.load(uri=url, headers=headers)
    return playlist.playlists[0].absolute_uri


real_url = get_real_url(
    'https://vod8.wenshibaowenbei.com/20210628/g4yNLlI7/index.m3u8')
real_url

'https://vod8.wenshibaowenbei.com/20210628/g4yNLlI7/1000kb/hls/index.m3u8'

解析真实地址的加密key：

playlist = m3u8.load(uri=real_url, headers=headers)
key = playlist.keys[-1]
print(key.uri, key.method, key.iv)

https://ts8.hhmm0.com:9999/20210628/g4yNLlI7/1000kb/hls/key.key AES-128 None

可以看到密钥下载地址和加密类型。

使用request下载密钥：

import requests

r = requests.get(playlist.keys[0].uri, headers=headers)
key = r.content
key

b'7ec5143edebbc899'

可以单线程直接下载视频：

import time

n = len(playlist.segments)
size = 0
start = time.time()
for i, seg in enumerate(playlist.segments, 1):
    r = requests.get(seg.absolute_uri, headers=headers)
    data = r.content
    data = AESDecrypt(data, key=key, iv=key)
    size += len(data)
    with open("reusult.mp4", "ab") as f:
        f.write(data)
    print(f"\r下载进度({i}/{n})，已下载：{size/1024/1024:.2f}MB，下载已耗时：{time.time()-start:.2f}s", end=" ")

下载进度(1435/1435)，已下载：424.69MB，下载已耗时：850s

单线程下载，好处是不会产生多余的文件，缺点是速度太慢了，一个视频下载了10多分钟。

下面我们整理一下完整的代码：

单线程视频下载的完整代码

import time
from Crypto.Util.Padding import pad
from Crypto.Cipher import AES
import requests
import m3u8

headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36"
}


def get_real_url(url):
    playlist = m3u8.load(uri=url, headers=headers)
    return playlist.playlists[0].absolute_uri


def AESDecrypt(cipher_text, key, iv):
    cipher_text = pad(data_to_pad=cipher_text, block_size=AES.block_size)
    aes = AES.new(key=key, mode=AES.MODE_CBC, iv=key)
    cipher_text = aes.decrypt(cipher_text)
    return cipher_text


def download_m3u8_video(url, save_name):
    real_url = get_real_url(url)
    playlist = m3u8.load(uri=real_url, headers=headers)
    key = requests.get(playlist.keys[-1].uri, headers=headers).content

    n = len(playlist.segments)
    size = 0
    start = time.time()
    for i, seg in enumerate(playlist.segments, 1):
        r = requests.get(seg.absolute_uri, headers=headers)
        data = r.content
        data = AESDecrypt(data, key=key, iv=key)
        size += len(data)
        with open(save_name, "ab" if i != 1 else "wb") as f:
            f.write(data)
        print(
            f"\r下载进度({i}/{n})，已下载：{size/1024/1024:.2f}MB，下载已耗时：{time.time()-start:.2f}s", end=" ")


download_m3u8_video('https://vod8.wenshibaowenbei.com/20210628/g4yNLlI7/index.m3u8', '走进家门.mp4')

多线程下载改造

对于多线程，由于下载的文件可能出现间断，所以我们不能直接追加到目标视频中，可以先下载下来，最后统一合并并删除。

先创建ts视频下载的方法：

import os
import requests


def download_ts(url, key, i):
    r = requests.get(url, headers=headers)
    data = r.content
    data = AESDecrypt(data, key=key, iv=key)
    with open(f"tmp/{i:0>5d}.ts", "ab") as f:
        f.write(data)
    print(f"\r{i:0>5d}.ts已下载", end="  ")


if not os.path.exists("tmp"):
    os.mkdir('tmp')

任意下载一个片段测试一下：

import requests
import m3u8


def get_real_url(url):
    playlist = m3u8.load(uri=url, headers=headers)
    return playlist.playlists[0].absolute_uri


headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36"
}
real_url = get_real_url(
    'https://vod8.wenshibaowenbei.com/20210628/g4yNLlI7/index.m3u8')
playlist = m3u8.load(uri=real_url, headers=headers)
key = requests.get(playlist.keys[-1].uri, headers=headers).content

download_ts(playlist.segments[0].absolute_uri, key, 1)

00001.ts已下载  

检查该片段可以正常播放。

然后执行以下方法即可10个线程同时一起下载：

from concurrent.futures import ThreadPoolExecutor

with ThreadPoolExecutor(max_workers=10) as pool:
    for i, seg in enumerate(playlist.segments):
        pool.submit(download_ts, seg.absolute_uri, key, i)

经过一分20秒左右的时间，所有视频流已经全部下载完毕，比单线程的速度快了不止10倍。

最后我们实现文件的合并和ts临时文件清除：

import glob

with open('video.mp4', 'wb') as fw:
    files = glob.glob('tmp/*.ts')
    for file in files:
        with open(file, 'rb') as fr:
            fw.write(fr.read())
            print(f'\r{file}已合并!总数:{len(files)}', end="     ")
        os.remove(file)

执行后，已经在1秒左右时间合并并清除完临时文件。

多线程下载的完整代码

import glob
from concurrent.futures import ThreadPoolExecutor
import m3u8
import os
import requests
from Crypto.Util.Padding import pad
from Crypto.Cipher import AES
import requests

headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36"
}


def download_ts(url, key, i):
    r = requests.get(url, headers=headers)
    data = r.content
    data = AESDecrypt(data, key=key, iv=key)
    with open(f"tmp/{i:0>5d}.ts", "ab") as f:
        f.write(data)
    print(f"\r{i:0>5d}.ts已下载", end="  ")


def get_real_url(url):
    playlist = m3u8.load(uri=url, headers=headers)
    return playlist.playlists[0].absolute_uri


def AESDecrypt(cipher_text, key, iv):
    cipher_text = pad(data_to_pad=cipher_text, block_size=AES.block_size)
    aes = AES.new(key=key, mode=AES.MODE_CBC, iv=key)
    cipher_text = aes.decrypt(cipher_text)
    return cipher_text


def download_m3u8_video(url, save_name, max_workers=10):
    if not os.path.exists("tmp"):
        os.mkdir('tmp')

    real_url = get_real_url(url)
    playlist = m3u8.load(uri=real_url, headers=headers)
    key = requests.get(playlist.keys[-1].uri, headers=headers).content

    with ThreadPoolExecutor(max_workers=max_workers) as pool:
        for i, seg in enumerate(playlist.segments):
            pool.submit(download_ts, seg.absolute_uri, key, i)

    with open(save_name, 'wb') as fw:
        files = glob.glob('tmp/*.ts')
        for file in files:
            with open(file, 'rb') as fr:
                fw.write(fr.read())
                print(f'\r{file}已合并!总数:{len(files)}', end="     ")
            os.remove(file)


download_m3u8_video('https://vod8.wenshibaowenbei.com/20210628/g4yNLlI7/index.m3u8', '走进家门.mp4')