Selenium爬取百度黑人高清图片

# 导包
from time import sleep
from selenium import webdriver
from selenium.webdriver.common.by import By
import re
import time
import requests

# 实例化浏览器
driver = webdriver.Chrome()

# 打开网址
driver.get("http://www.baidu.com")
# driver.get('https://www.selenium.dev/selenium/web/web-form.html')
# 窗口最大化
driver.maximize_window()
driver.implicitly_wait(0.5)
text_box = driver.find_element(by=By.ID, value='kw')
submit_button = driver.find_element(by=By.ID, value="su")
text_box.send_keys("黑人")  # 搜索黑人
submit_button.click()  # 点击百度查找按钮
driver.implicitly_wait(3)

black_name = driver.find_element(by=By.CLASS_NAME, value="foot-more_3ukcZ")  # 找到所有黑人图片
driver.implicitly_wait(1)
black_name.click()  # 点击进去

driver.implicitly_wait(1)
black_gq = driver.find_element(by=By.XPATH, value='//*[@id="typeFilter"]/div[2]/ul/li[2]')  # 找到所有黑人图片
black_gq.click()

time.sleep(2)
head = 'https://image.baidu.com'
result = re.findall(r'"imgbox-border"><a href="(.*?)" target="_blank"', driver.page_source)
print(len(result), result)

#下滑窗口下載所需的数据量
# js = "var q=document.documentElement.scrollTop=10000"  # documentElement表示获取根节点元素
# while (len(result) < 100):  # 下滑窗口下載所需的数据量
#     driver.execute_script(js)
#     time.sleep(2)
#     result = re.findall(r'"imgbox-border"><a href="(.*?)" target="_blank"', driver.page_source)
#     print(len(result), result)

#获取高清图片的链接
https = []
count = 1
for i in result:
    temp = head + i
    temp = temp.replace("amp;", "")
    print(count, temp)
    https.append(temp)
    count += 1

#进入高清图片链接获取图片下载链接
photo=[]
for i in range(len(https)):
    driver.get(https[i])
    time.sleep(0.5)
    temp=re.findall(r';" src="(.*?)" width=',driver.page_source)
    photo.append(temp)
print(len(photo),photo)
photo = list(filter(None, photo))#去除为空的数据


#对图片进行下载
for i in range(len(photo)):
    if photo[i][0]==None:
        continue
    print(i,photo[i])
    temp=requests.get(photo[i][0])
    path_img = "E:/Python代码/网络爬虫/image/图{}.jpg".format(i)
    temp.raise_for_status()
    with open(path_img,'wb') as f:
        f.write(temp.content)
        f.close()
        print(path_img,"保存成功！")
input('wait...')