路飞学城—Python爬虫实战密训班 第二章
一、Selenium基础
Selenium是一个第三方模块，可以完全模拟用户在浏览器上操作（相当于在浏览器上点点点）。
1.安装
- pip install selenium
2.优缺点
- 无需查看和确定请求头请求体等数据细节，直接模拟人点击浏览器的行为
- 效率不高
3.依赖驱动：
- Firefox
https://github.com/mozilla/geckodriver/releases
- Chrome
http://chromedriver.storage.googleapis.com/index.html
4.与selenium相关的基本操作
from selenium import webdriver # 配置驱动
#驱动一定要自己下载并放在一个目录，否则会出错 option = webdriver.ChromeOptions()
driver = webdriver.Chrome('/Users/wupeiqi/drivers/chromedriver', chrome_options=option) # 1. 控制浏览器打开指定页面
driver.get("https://dig.chouti.com/all/hot/recent/1") # 2. 找到登录按钮
btn_login = driver.find_element_by_xpath('//*[@id="login-link-a"]')
# 3. 点击按钮
btn_login.click() # 4. 找到手机标签
input_user = driver.find_element_by_xpath('//*[@id="mobile"]')
# 5. 找到密码标签
input_pwd = driver.find_element_by_xpath('//*[@id="mbpwd"]') # 6. 输入用户名
input_user.send_keys('13121758648')
# 7. 输入密码
input_pwd.send_keys('woshiniba') # 8. 点击登录按钮
input_submit = driver.find_element_by_xpath(
'//*[@id="footer-band"]/div[5]/div/div/div[1]/div[2]/div[4]/div[2]/div/span[1]')
input_submit.click() print(driver.get_cookies()) # # 9. 点击跳转
# news = driver.find_element_by_xpath('//*[@id="newsContent20646261"]/div[1]/a[1]')
# # news.click()
# driver.execute_script("arguments[0].click();", news) # 10.管理浏览器
# driver.close()
二、破解滑动验证码
WuSir为我们带来的精彩的讲解，从__main__的主函数调用开始，先讲了图片的截取和距离的测算，接下来分析了怎么模拟人类行为的滑动过程，通过速度和加速度的空值实现，而且会故意制造匹配之后的小幅振动行为，最后点击确定就可以破解该验证码，重点是像素的选择和速度的调节，感谢！
from selenium import webdriver
from selenium.webdriver import ActionChains
from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.support.wait import WebDriverWait
import os
import shutil
from PIL import Image
import time def get_snap(driver):
driver.save_screenshot('full_snap.png')
page_snap_obj = Image.open('full_snap.png') return page_snap_obj def get_image(driver):
img = driver.find_element_by_class_name('geetest_canvas_img')
time.sleep(2)
location = img.location
size = img.size left = location['x']
top = location['y']
right = left + size['width']
bottom = top + size['height'] page_snap_obj = get_snap(driver) image_obj = page_snap_obj.crop((left * 2, top * 2, right * 2, bottom * 2))
# image_obj.show()
with open('code.png', 'wb') as f:
image_obj.save(f, format='png')
return image_obj def get_distance(image1, image2):
# start = 0
# threhold = 70
# for i in range(start, image1.size[0]):
# for j in range(0, image1.size[1]):
# rgb1 = image1.load()[i, j]
# rgb2 = image2.load()[i, j]
# res1 = abs(rgb1[0] - rgb2[0])
# res2 = abs(rgb1[1] - rgb2[1])
# res3 = abs(rgb1[2] - rgb2[2])
# # print(res1,res2,res3)
# if not (res1 < threhold and res2 < threhold and res3 < threhold):
# print(111111, i, j)
# return i - 13
# print(2222, i, j)
# return i - 13
start = 0
threhold = 70
v = []
for i in range(start, image1.size[0]):
for j in range(0, image1.size[1]):
rgb1 = image1.load()[i, j]
rgb2 = image2.load()[i, j]
res1 = abs(rgb1[0] - rgb2[0])
res2 = abs(rgb1[1] - rgb2[1])
res3 = abs(rgb1[2] - rgb2[2]) if not (res1 < threhold and res2 < threhold and res3 < threhold):
print(i)
if i not in v:
v.append(i) stop = 0
for i in range(0, len(v)):
val = i + v[0]
if v[i] != val:
stop = v[i]
break width = stop - v[0]
print(stop, v[0], width)
return width def get_tracks(distance):
import random
exceed_distance = random.randint(0, 5)
distance += exceed_distance # 先滑过一点，最后再反着滑动回来
v = 0
t = 0.2
forward_tracks = [] current = 0
mid = distance * 3 / 5
while current < distance:
if current < mid:
a = random.randint(1, 3)
else:
a = random.randint(1, 3)
a = -a
s = v * t + 0.5 * a * (t ** 2)
v = v + a * t
current += s
forward_tracks.append(round(s)) # 反着滑动到准确位置
v = 0
t = 0.2
back_tracks = [] current = 0
mid = distance * 4 / 5
while abs(current) < exceed_distance:
if current < mid:
a = random.randint(1, 3)
else:
a = random.randint(-3, -5)
a = -a
s = -v * t - 0.5 * a * (t ** 2)
v = v + a * t
current += s
back_tracks.append(round(s))
return {'forward_tracks': forward_tracks, 'back_tracks': list(reversed(back_tracks))} def crack(driver): # 破解滑动认证
# 1、点击按钮，得到没有缺口的图片
button = driver.find_element_by_xpath('//*[@id="embed-captcha"]/div/div[2]/div[1]/div[3]')
button.click() # 2、获取没有缺口的图片
image1 = get_image(driver) # 3、点击滑动按钮，得到有缺口的图片
button = driver.find_element_by_class_name('geetest_slider_button')
button.click() # 4、获取有缺口的图片
image2 = get_image(driver) # 5、对比两种图片的像素点，找出位移
distance = get_distance(image1, image2)
print(distance)
#
# 6、模拟人的行为习惯，根据总位移得到行为轨迹
tracks = get_tracks(int(distance / 2)) # 7、按照行动轨迹先正向滑动，后反滑动
button = driver.find_element_by_class_name('geetest_slider_button')
ActionChains(driver).click_and_hold(button).perform() # 正常人类总是自信满满地开始正向滑动，自信地表现是疯狂加速
for track in tracks['forward_tracks']:
ActionChains(driver).move_by_offset(xoffset=track, yoffset=0).perform() # 结果傻逼了，正常的人类停顿了一下，回过神来发现，卧槽，滑过了,然后开始反向滑动
time.sleep(0.5)
for back_track in tracks['back_tracks']:
ActionChains(driver).move_by_offset(xoffset=back_track, yoffset=0).perform()
#
# # 小范围震荡一下，进一步迷惑极验后台，这一步可以极大地提高成功率
ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform()
ActionChains(driver).move_by_offset(xoffset=-3, yoffset=0).perform() # # 成功后，骚包人类总喜欢默默地欣赏一下自己拼图的成果，然后恋恋不舍地松开那只脏手
time.sleep(0.5)
ActionChains(driver).release().perform() def login_luffy(username, password):
driver = webdriver.Chrome('/Users/wupeiqi/drivers/chromedriver')
driver.set_window_size(960, 800)
try:
# 1、输入账号密码回车
driver.implicitly_wait(3)
driver.get('https://www.luffycity.com/login') input_username = driver.find_element_by_xpath('//*[@id="router-view"]/div/div/div[2]/div[2]/input[1]')
input_pwd = driver.find_element_by_xpath('//*[@id="router-view"]/div/div/div[2]/div[2]/input[2]') input_username.send_keys(username)
input_pwd.send_keys(password) # 2、破解滑动认证
crack(driver) time.sleep(10) # 睡时间长一点，确定登录成功
finally:
pass
# driver.close() if __name__ == '__main__':
login_luffy(username='wupeiqi', password='123123123')
三：总结
前半段的直播都是由咸湿的，哦不对、是亲切的Alex老师为我们分享了关于职场方面的一些东西，尤其是咸湿的，哦不对、是亲切的Alex老师用他曾经的经历来讲述这些东西，这些经验和思想，听完后对大家讨论得都很热烈，挺受启发的。
通过学习selenium模块，使得部分对于很复杂的爬虫，用selenium做起来还是比较方便的。但如果使用selenium模块的话，对于爬虫程序可以说基本毫无性能可言，一般的解决方案可以通过selenium + 其它模块一起配合使用来相互弥补。最后，WuSir通过selenium 和 PIL模块一起配合使用，破解了极验的滑动验证码、但此方式有个大问题，只能处理简单的图片，对于复杂的图片命中率会不高，面对更加复杂的验证码只能通过打码平台来解决了。