前言

目前创新互联建站已为上1000+的企业提供了网站建设、域名、虚拟空间、绵阳服务器托管、企业网站设计、龙南网站维护等服务，公司将坚持客户导向、应用为本的策略，正道将秉承"和谐、参与、激情"的文化，与客户和合作伙伴齐心协力一起成长，共同发展。

简单学习过网络爬虫,只是之前都是照着书上做并发,大概能理解,却还是无法自己用到自己项目中,这里自己研究实现一个网页嗅探HTML5播放控件中基于m3u8协议ts格式视频资源的项目,并未考虑过复杂情况,毕竟只是练练手.

源码

# coding=utf-8
import asyncio
import multiprocessing
import os
import re
import time
from math import floor
from multiprocessing import Manager
import aiohttp
import requests
from lxml import html
import threading
from src.my_lib import retry
from src.my_lib import time_statistics


class M3U8Download:
 _path = "./resource\\" # 本地文件路径
 _url_seed = None # 资源所在链接前缀
 _target_url = {} # 资源任务目标字典
 _mode = ""
 _headers = {"User-agent": "Mozilla/5.0"} # 浏览器代理
 _target_num = 100

 def __init__(self):
 self._ml = Manager().list() # 进程通信列表
 if not os.path.exists(self._path): # 检测本地目录存在否
  os.makedirs(self._path)
 exec_str = r'chcp 65001'
 os.system(exec_str) # 先切换utf-8输出,防止控制台乱码

 def sniffing(self, url):
 self._url = url
 print("开始嗅探...")
 try:
  r = requests.get(self._url) # 访问嗅探网址,获取网页信息
 except:
  print("嗅探失败,网址不正确")
  os.system("pause")
 else:
  tree = html.fromstring(r.content)
  try:
  source_url = tree.xpath('//video//source/@src')[0] # 嗅探资源控制文件链接,这里只针对一个资源控制文件
  # self._url_seed = re.split("/\w+\.m3u8", source_url)[0] # 从资源控制文件链接解析域名
  except:
  print("嗅探失败,未发现资源")
  os.system("pause")
  else:
  self.analysis(source_url)

 def analysis(self, source_url):
 try:
  self._url_seed = re.split("/\w+\.m3u8", source_url)[0] # 从资源控制文件链接解析域名
  with requests.get(source_url) as r: # 访问资源控制文件,获得资源信息
  src = re.split("\n*#.+\n", r.text) # 解析资源信息
  for sub_src in src: # 将资源地址储存到任务字典
   if sub_src:
   self._target_url[sub_src] = self._url_seed + "/" + sub_src
 except Exception as e:
  print("资源无法成功解析", e)
  os.system("pause")
 else:
  self._target_num = len(self._target_url)
  print("sniffing success!!!,found", self._target_num, "url.")
  self._mode = input(
  "1:-> 单进程(Low B)\n2:-> 多进程+多线程(网速开始biubiu飞起!)\n3:-> 多进程+协程(最先进的并发!!!)\n")
  if self._mode == "1":
  for path, url in self._target_url.items():
   self._download(path, url)
  elif self._mode == "2" or self._mode == "3":
  self._multiprocessing()

 def _multiprocessing(self, processing_num=4): # 多进程,多线程
 target_list = {} # 进程任务字典,储存每个进程分配的任务
 pool = multiprocessing.Pool(processes=processing_num) # 开启进程池
 i = 0 # 任务分配标识
 for path, url in self._target_url.items(): # 分配进程任务
  target_list[path] = url
  i += 1
  if i % 10 == 0 or i == len(self._target_url): # 每个进程分配十个任务
  if self._mode == "2":
   pool.apply_async(self._sub_multithreading, kwds=target_list) # 使用多线程驱动方法
  else:
   pool.apply_async(self._sub_coroutine, kwds=target_list) # 使用协程驱动方法
  target_list = {}
 pool.close() # join函数等待所有子进程结束
 pool.join() # 调用join之前，先调用close函数，否则会出错。执行完close后不会有新的进程加入到pool
 while True:
  if self._judge_over():
  self._combine()
  break

 def _sub_multithreading(self, **kwargs):
 for path, url in kwargs.items(): # 根据进程任务开启线程
  t = threading.Thread(target=self._download, args=(path, url,))
  t.start()

 @retry()
 def _download(self, path, url): # 同步下载方法
 with requests.get(url, headers=self._headers) as r:
  if r.status_code == 200:
  with open(self._path + path, "wb")as file:
   file.write(r.content)
  self._ml.append(0) # 每成功一个就往进程通信列表增加一个值
  percent = '%.2f' % (len(self._ml) / self._target_num * 100)
  print(len(self._ml), ": ", path, "->OK", "\tcomplete:", percent, "%") # 显示下载进度
  else:
  print(path, r.status_code, r.reason)

 def _sub_coroutine(self, **kwargs):
 tasks = []
 for path, url in kwargs.items(): # 根据进程任务创建协程任务列表
  tasks.append(asyncio.ensure_future(self._async_download(path, url)))
 loop = asyncio.get_event_loop() # 创建异步事件循环
 loop.run_until_complete(asyncio.wait(tasks)) # 注册任务列表

 async def _async_download(self, path, url): # 异步下载方法
 async with aiohttp.ClientSession() as session:
  async with session.get(url, headers=self._headers) as resp:
  try:
   assert resp.status == 200, "E" # 断言状态码为200,否则抛异常,触发重试装饰器
   with open(self._path + path, "wb")as file:
   file.write(await resp.read())
  except Exception as e:
   print(e)
  else:
   self._ml.append(0) # 每成功一个就往进程通信列表增加一个值
   percent = '%.2f' % (len(self._ml) / self._target_num * 100)
   print(len(self._ml), ": ", path, "->OK", "\tcomplete:", percent, "%") # 显示下载进度

 def _combine(self): # 组合资源方法
 try:
  print("开始组合资源...")
  identification = str(floor(time.time()))
  exec_str = r'copy /b "' + self._path + r'*.ts" "' + self._path + 'video' + identification + '.mp4"'
  os.system(exec_str) # 使用cmd命令将资源整合
  exec_str = r'del "' + self._path + r'*.ts"'
  os.system(exec_str) # 删除原来的文件
 except:
  print("资源组合失败")
 else:
  print("资源组合成功!")

 def _judge_over(self): # 判断是否全部下载完成
 if len(self._ml) == len(self._target_url):
  return True
 return False


@time_statistics
def app():
 multiprocessing.freeze_support()
 url = input("输入嗅探网址:\n")
 m3u8 = M3U8Download()
 m3u8.sniffing(url)
 # m3u8.analysis(url)


if __name__ == "__main__":
 app()

名称栏目：python爬取基于m3u8协议的ts文件并合并-创新互联
本文来源：http://csdahua.cn/article/ddcooc.html