版权声明:青春看似荒唐,没人会选择投降。 https://blog.csdn.net/sdzhr/article/details/82943349
哔哩哔哩(极验)滑块验证码破解
使用selenium、PIL以及OpenCV来获取图片缺口位置以及模拟鼠标拖动效果
目标页面网址:https://passport.bilibili.com/login
from selenium import webdriver
import requests
import time
from selenium.webdriver.common.action_chains import ActionChains
import re
from PIL import Image
import io
import cv2
import numpy as np
def get_img(div_list):
position_list = []
for div in div_list:
style = div.get_attribute('style')
position_tuple = re.search(r'background-position:\s+-?(\d+)px\s-?(\d+)px', style).groups(1)
position_list.append((int(position_tuple[0]), int(position_tuple[1])))
style = div_list[0].get_attribute('style')
img_url = re.search(r'url\(\"(.*?)\"\);', style).group(1)
return position_list, img_url
def restore_img(img_data, position_list):
img = Image.open(io.BytesIO(img_data))
new_img = Image.new('RGB', (260, 116))
i = 0
j = 0
for position in position_list:
if position[1] == 58:
region = img.crop((position[0], 58, position[0] + 10, 58 + 58))
new_img.paste(region, (i, 0, i + 10, 58))
i += 10
else:
region = img.crop((position[0], 0, position[0] + 10, 58))
new_img.paste(region, (j, 58, j + 10, 58 + 58))
j += 10
return new_img
# 第一种方法是遍历带缺口的和不带缺口的两张图片的所有像素点寻找不同之处
def get_move_location(full_img, cut_img):
for i in range(260):
for j in range(116):
full_pixel = full_img.getpixel((i, j))
cut_pixel = cut_img.getpixel((i, j))
for index in range(3):
if abs(full_pixel[index] - cut_pixel[index]) >= 50:
return i - 5
# 第二种方法是使用OpenCV寻找滑块图片在带缺口的图片中的位置
def get_move_location2(filename1, filename2):
target = cv2.imread(filename1)
template = cv2.imread(filename2)
temp = 'temp.jpg'
targ = 'targ.jpg'
cv2.imwrite(temp, template)
cv2.imwrite(targ, target)
target = cv2.imread(targ)
target = abs(255 - target)
cv2.imwrite(targ, target)
target = cv2.imread(targ)
template = cv2.imread(temp)
result = cv2.matchTemplate(target, template, cv2.TM_CCOEFF_NORMED)
x, y = np.unravel_index(result.argmax(), result.shape)
return y
# 模拟人手的滑动轨迹
def get_tracks(distance):
v = 0
t = 0.3
tracks = []
current = 0
mid = distance * 5 / 6
while current < distance:
if current < mid:
a = 2
else:
a = -3
v0 = v
s = v0 * t + 0.5 * a * (t ** 2)
current += s
tracks.append(round(s))
v = v0 + a * t
if current > distance:
tracks.append(distance - current)
return tracks
def move(location):
element = driver.find_element_by_xpath('//div[@class="gt_slider_knob gt_show"]')
ActionChains(driver).click_and_hold(element).perform()
tracks = get_tracks(location)
for track in tracks:
ActionChains(driver).move_by_offset(track, 0).perform()
time.sleep(0.8)
ActionChains(driver).release(element).perform()
if __name__ == "__main__":
driver = webdriver.Chrome()
driver.get('https://passport.bilibili.com/login')
time.sleep(1)
full_img_div_list = driver.find_elements_by_class_name('gt_cut_fullbg_slice')
cut_img_div_list = driver.find_elements_by_class_name('gt_cut_bg_slice')
full_img_position_list, full_img_url = get_img(full_img_div_list)
cut_bg_img_position_list, cut_bg_img_url = get_img(cut_img_div_list)
full_img = restore_img(requests.get(full_img_url).content, full_img_position_list)
cut_bg_img = restore_img(requests.get(cut_bg_img_url).content, cut_bg_img_position_list)
x = get_move_location(full_img, cut_bg_img)
# cut_style = driver.find_element_by_xpath('//div[@class="gt_slice gt_show"]').get_attribute('style')
# cut_url = re.search(r'url\(\"(.*?)\"\);', cut_style).group(1)
# cut = Image.open(io.BytesIO(requests.get(cut_url).content))
# cut = cut.convert('L')
# cut_bg = cut_bg_img.convert('L')
# cut.save("cut.png")
# cut_bg.save("cut_bg.png")
# x = get_move_location2("cut.png", "cut_bg.png")
print("滑块偏移量:{}px".format(x))
# print(location)
move(x)
time.sleep(5)
driver.close()