210406课 多线程/多进程 编程
如何实现并发(并行)编程?
-
并行:真正的同时进行,需多核CPU支持
-
并发:通过快速的切换制造出并行的效果
线程:操作系统分配CPU基本单位,最小的执行单元
进程:操作系统分配内存的基本单位,是程序在一组数据上的活动
通常我们启动一个程序,就启动了一个(或多个)进程,一个进程中又可以包含多个线程
单线程:只有主线程,只有唯一的一个执行线索
# 例子:单线程爬取360的美女图片
import time
import requests
from selenium.webdriver import Chrome
def dowload_image(image_url):
resp = requests.get(image_url)
if resp.status_code == 200:
filename = image_url[image_url.rfind('/') + 1:]
try:
with open(f'image360/{filename}', 'wb') as file:
for data in resp.iter_content(512):
file.write(data)
except OSError:
pass
def main():
browser = Chrome()
browser.get('https://image.so.com/z?ch=beauty')
# 让浏览器执行javascript代码
for _ in range(20):
browser.execute_script('window.scrollTo(0, window.scrollY + 1500)')
time.sleep(0.5)
# 隐式等待
browser.implicitly_wait(10)
image_tags = browser.find_elements_by_css_selector('span.img > img')
start = time.time()
for image_tag in image_tags:
image_url = image_tag.get_attribute('src')
dowload_image(image_url)
end = time.time()
print(f'{end - start:.3f}秒')
browser.close()
if __name__ == '__main__':
main()
多线程:处理I/O密集型任务
(爬虫)
–> GIL(全局解释器锁)
'''
多线程
'''
import random
import time
from concurrent.futures.thread import ThreadPoolExecutor
from threading import Thread
'''
装饰器:用一个函数去装饰另外一个函数或者类,为其提供额外的功能
专门作用于横切关注功能(跟正常业务逻辑没有必然影响的功能)
装饰器实现了设计模式(前人总结的可重复使用的设计理念)的代理模式
'''
# 装饰器的参数是被装饰的函数,返回值是带有装饰功能的函数
def record_time(func):
# 带有装饰功能的函数
def wrapper(*args, **kwargs):
start = time.time()
result = func(*args, **kwargs)
end = time.time()
print(f'{func.__name__}执行时间:{end - start:.3f}秒')
return result
return wrapper
@record_time
def download(filename):
print(f'开始下载{filename}.')
time.sleep(random.randint(5, 8))
print(f'{filename}下载完成.')
@record_time
def upload(filename):
print(f'开始上传{filename}.')
time.sleep(random.randint(5, 8))
print(f'{filename}上传完成.')
# 用法一:
# @record_time
# def main():
# # 通过调用Thread类的构造器就可以创建线程对象
# # 通过指定target和args参数告诉线程启动后要执行的任务以及对应的参数
# t1 = Thread(target=download, args=('爬虫从入门到入狱',))
# t2 = Thread(target=download, args=('MYSQL从删库到跑路',))
# t3 = Thread(target=upload, args=('数据分析从精通到放弃',))
# # 通过调用线程对象的start()方法启动线程
# t1.start()
# t2.start()
# t3.start()
# # 通过调用线程对象的join()方法等待线程结束
# # 让主线程等待t1,t2,t3对应的线程执行结束
# t1.join()
# t2.join()
# t3.join()
# 用法二: 线程池
@record_time
def main():
with ThreadPoolExecutor(max_workers=32) as pool:
# 调用线程池对象的submit方法将任务交给线程池中的线程去执行
# submit方法会返回将来线程的执行结果
f1 = pool.submit(download, '爬虫从入门到入狱')
f2 = pool.submit(download, 'MYSQL从删库到跑路')
f3 = pool.submit(upload, '数据分析从精通到放弃')
# f1.result()
# f2.result()
# f3.result()
if __name__ == '__main__':
main()
# 例子:多线程爬取360的美女图片
import time
from concurrent.futures.thread import ThreadPoolExecutor
import requests
from selenium.webdriver import Chrome
def dowload_image(image_url):
resp = requests.get(image_url)
if resp.status_code == 200:
filename = image_url[image_url.rfind('/') + 1:]
try:
with open(f'image360/{filename}', 'wb') as file:
for data in resp.iter_content(512):
file.write(data)
except OSError:
pass
def main():
browser = Chrome()
browser.get('https://image.so.com/z?ch=beauty')
# 让浏览器执行javascript代码
for _ in range(20):
browser.execute_script('window.scrollTo(0, window.scrollY + 1500)')
time.sleep(0.5)
# 隐式等待
browser.implicitly_wait(10)
image_tags = browser.find_elements_by_css_selector('span.img > img')
with ThreadPoolExecutor(max_workers=32) as pool:
start = time.time()
for image_tag in image_tags:
image_url = image_tag.get_attribute('src')
pool.submit(dowload_image, image_url)
end = time.time()
print(f'下载耗时:{end - start:.3f}秒')
browser.close()
if __name__ == '__main__':
main()
多进程:处理计算密集型任务
音视频编解码,文件压缩,机器学习算法
--> 一般在GPU显卡上运行
# 例子:多进程爬取360的美女图片
import time
from concurrent.futures.process import ProcessPoolExecutor
import requests
from selenium.webdriver import Chrome
def dowload_image(image_url):
resp = requests.get(image_url)
if resp.status_code == 200:
filename = image_url[image_url.rfind('/') + 1:]
try:
with open(f'image360/{filename}', 'wb') as file:
for data in resp.iter_content(512):
file.write(data)
except OSError:
pass
def main():
browser = Chrome()
browser.get('https://image.so.com/z?ch=beauty')
for _ in range(20):
# 让浏览器执行JavaScript代码
browser.execute_script('window.scrollTo(0, window.scrollY + 1500)')
time.sleep(0.5)
# 隐式等待
# 在使用find_xxx方法获取页面元素时,如果元素还没有出现,可以等待指定的时间,
# 如果等待超时就引发异常,如果在指定的时间内获取到元素就返回对应的元素
browser.implicitly_wait(10)
# # 显示等待 ---> 创建一个等待对象
# wait = WebDriverWait(browser, 10)
# # 设置等待条件
# wait.until(expected_conditions.presence_of_element_located(
# (By.CSS_SELECTOR, 'span.img > img')
# ))
image_tags = browser.find_elements_by_css_selector('span.img > img')
# print(len(image_tags))
with ProcessPoolExecutor(max_workers=32) as pool:
start = time.time()
for image_tag in image_tags:
image_url = image_tag.get_attribute('src')
pool.submit(dowload_image, image_url)
end = time.time()
print(f'下载耗时: {end - start:.3f}秒')
browser.close()
if __name__ == '__main__':
main()
异步编程:适合处理I/O密集型任务
--> 通过提高CPU利用率达到协作式并发效果
-->比多线程更高效,由于内容复杂,择日再写
-
easyocr 光学文字识别
-
用于识别验证码
-
安装方式: 在终端上输入 pip install easyocr
-