future之与中文理解起来其实挺微妙，不过这与生活中大家熟知的期物在底层逻辑上是一致的，future英文词义中就有期货的意思，都是封存一个东西，平常你该干嘛就干嘛，可以不用去理会，在未来的某个时候去看结果就行，只是python中那个物是对象而已。而关键词是延迟，异步。
思考题：添加异常处理
def download_all(sites):
    with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
        to_do = {}
        for site in sites:
            future = executor.submit(download_one, site)
            to_do[future]=site
            
        for future in concurrent.futures.as_completed(to_do):
            try：
                res=future.result()
            except request.exceptions.HTTPError as e:
                e_msg=‘HTTP erro’
            except request.exceptions.ConnectionError as e:
                e_msg=‘Connection erro’
            else:
                e_msg=''
            if e_msg:
                site=to_do[future]
                Print(‘Error is {} from {}’.format(e_msg,site))

需要加异常的应该就只有一个地方：requests.get()发送网页请求的时候。其它地方不涉及IO。也不涉及数据类型变化，不用做数据类型判断。
由于不能访问wiki，所以网页改了成了国内的。
-- ps: 和0.2s比起来太慢了。

# -*- encoding -*-
'''
py 3.6
sulime
'''
import concurrent.futures
import threading
import requests
import time

now = lambda: time.perf_counter()

def download_one(url):
    try:
        req = requests.get(url)
        req.raise_for_status()
        print('Read {} from {}'.format(len(req.content), url))
    except:
        print('404')

def download_all(sites):
    with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
        executor.map(download_one, sites)

def main():
    sites = [
            'https://www.baidu.com/',
            'https://pypi.org/',
            'https://www.sina.com.cn/',
            'https://www.163.com/',
            'https://news.qq.com/',
            'http://www.ifeng.com/',
            'http://www.ce.cn/',
            'https://news.baidu.com/',
            'http://www.people.com.cn/',
            'http://www.ce.cn/',
            'https://news.163.com/',
            'http://news.sohu.com/'
            ]
    start = now()
    download_all(sites)
    print('Download {} sites in {} s'.format(len(sites), now() - start))

if __name__ == '__main__':
    main()

# Read 2443 from https://www.baidu.com/
# Read 6216 from https://news.qq.com/
# Read 699004 from https://www.163.com/
# Read 250164 from http://www.ifeng.com/
# Read 579572 from https://www.sina.com.cn/
# Read 107530 from http://www.ce.cn/
# Read 165901 from http://www.people.com.cn/
# Read 107530 from http://www.ce.cn/
# Read 210816 from https://news.163.com/
# Read 74060 from https://news.baidu.com/
# Read 174553 from http://news.sohu.com/
# Read 19492 from https://pypi.org/
# Download 12 sites in 2.8500169346527673 s
# [Finished in 3.6s]

老师好，看到文中为了使用.as_complete()作的修改似乎做了重复的工作，我对比了使用.as_complete()和.submit()后直接result()，得到的是相同的结果。
-- 问1：这里所做的修改只是为了展示.as_complete的功能么？我查看了文档也没想明白。
-- 问2：.as_complete()可能会在什么场景下使用得比较多？

eg.2.`.submit()`后直接`.result()`
    
    with concurrent.futures.ThreadPoolExecutor(max_workers=10) as executor:
        for site in sites:
            future = executor.submit(download_one, site)
            print(future.result())

学习到的知识点：
1. 并发和并行的区别，大佬通俗易懂的方式让我更深刻的体会到了程序到底是如何跑在多核机器上的
2. python中 Futures 特性，第一次接触到这个模块，待后续继续加深了解；
3. Python 中之所以同一时刻只允许一个线程运行，大佬解释了这是因为全局解释器锁的存在，而全局解释器锁又是为了解决 race condition而引入的，这个也从另一方面验证了我之前学习到的，python中多线程是无法利用多核的；
但是多线程无法利用多核也并不是一无是处，就像大佬在文中聊到的，多线程主要的适用场景就是 有IO延迟的场景，因为一个线程遇到IO延迟，它占用的全局解释器锁就会释放，而另一个线程即可以拿到锁开始执行； 这种在IO延迟场景中的并发，高效也是显而易见的；

并发是交替执行，同时只有一个任务执行。用于高I/O程序。并行是多个进程同时运行，用于CPU-heavy程序。原来Python可以做并行运算的，建立线程池的时候不指定线程数量就行了。以前写并发程序比较少。
思考题，只写异常处理的地方吧。
# 处理requests异常
    except ConnectionError as err:
        print(err)
    except HTTPError as err:
        print(err)
    except Timeout as err:
        print(err)
    # 处理futures异常
    except TooManyRedirects as err:
        print(err)
    except CancelledError as err:
        print(err)
    except TimeoutError as err:
        print(err)
    except BrokenExecutor as err:
        print(err)
    except:
        print("发生错误")

课程的练习代码: https://github.com/zwdnet/PythonPractice

简单的加了个异常处理，请老师指正
#多线程实现网站内容下载演示
import concurrent.futures
import requests
import threading
import time

def download_one(url):
    resp = requests.get(url)
    print('Read {} from {}'.format(len(resp.content), url))
    
    #模拟下载任务出错
    print(1/0)

def download_all(sites):
    try:
        #以下是用futures对象并发下载
        with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
            to_do = []
            for site in sites:
                future = executor.submit(download_one, site)
                to_do.append(future)

            for future in concurrent.futures.as_completed(to_do):
                future.result()
                    
    except Exception as e:
        print('运行错误：',e)