我正在使用Windows 10和python3。我的问题是使用硒铬驱动程序。我已经写了一些代码,但是出错了。当我被禁止时,我需要轮换代理。请帮助您解决我的代码
# -*- coding: utf-8 -*-
from selenium import webdriver
from time import sleep
import random
# import urls as urls
urls=[
'https://angel.co/_tom_c_',
'https://angel.co/2lr',
'https://angel.co/aamerbdullah10',
'https://angel.co/aaronbatalion',
'https://angel.co/aaron-bertinetti',
'https://angel.co/aaron-schumm-1',
'https://angel.co/aarrieta',
'https://angel.co/abenmeir-me-com',
'https://angel.co/abhas2002',
'https://angel.co/abrams',
'https://angel.co/adam-gerstein',
'https://angel.co/adam-rogas',
'https://angel.co/adam-rothenberg',
'https://angel.co/adaugelli',
'https://angel.co/adil-wali',]
out = r'E:\\out_html\\'
PROXY = [
'196.202.153.67:3128',
'177.125.243.12:3128',
'43.245.216.190:8080',
'45.221.77.82:8080',
'202.62.12.50:80',
]
chrome_options = webdriver.ChromeOptions()
chrome_options.add_argument(f'--proxy-server={random.choice(PROXY)}')
driver = webdriver.Chrome(chrome_options=chrome_options)
driver.delete_all_cookies()
x = 0
for link in urls.urls:
driver.get(link)
while "Sorry for the interruption..." in driver.page_source or 'No internet' in driver.page_source or "Please complete the security check to access" in driver.page_source:
driver.delete_all_cookies()
if "Sorry for the interruption..." in driver.page_source or 'No internet' in driver.page_source or "Please complete the security check to access" in driver.page_source:
if "Sorry for the interruption..." in driver.page_source:
print('\n')
print('Sorry for the interruption')
if 'No internet' in driver.page_source:
print('\n')
print('Sorry for No internet')
if "Please complete the security check to access" in driver.page_source:
print('\n')
print('Please complete the security check to access')
# sleep(120)
# driver.execute_script("window.stop();")
# rr = input('type r and hit enter to start again ! : ')
# if rr =='r' or '' in rr:
# driver.refresh()
chrome_options = webdriver.ChromeOptions()
chrome_options.add_argument(f'--proxy-server={random.choice(PROXY)}')
driver = webdriver.Chrome(chrome_options=chrome_options)
driver.delete_all_cookies()
driver.get(link)
x+=1
nn = link.split('/')[-1]
with open(out+f'{nn} {x}.html', 'wb') as f:
f.write(driver.page_source.encode("utf-8"))
print('\n')
print(f'{nn} {x}.html')
driver.delete_all_cookies()
driver.quit()
print('done click to end')
input()
我希望这会循环所有网址并随机获取代理列表 <------------------------------------------------- ------------------------>