前言
代码直接运行即可
此处是用的selenium模拟,主要记录的难点是如何 模拟人工滑速
具体原理和利用到的东西都有做注释,逻辑完整,小白还是可以尝试理解的
一、正常简单模拟滑动
目标网址:点击
import base64
import time
import ddddocr
from selenium import webdriver
from selenium.webdriver.common.by import Bydef text_dis(bg, fg):slide = ddddocr.DdddOcr(det=False, ocr=False)with open(bg, 'rb') as f:target_bytes = f.read()with open(fg, 'rb') as f:background_bytes = f.read()res = slide.slide_comparison(target_bytes, background_bytes) # 计算出距离return res.get('target')[0]def get_slide():options = webdriver.ChromeOptions()# 对于老版本的浏览器不行options.add_argument('--disable-blink-features=AutomationControlled')driver = webdriver.Chrome(chrome_options=options)driver.maximize_window()driver.get('https://www.geetest.com/demo/slide-bind.html')driver.find_element(By.ID, 'username').send_keys('13697028751') # 输入框输入账号和密码driver.find_element(By.ID, 'password').send_keys('19961227ai')time.sleep(2)driver.find_element(By.CSS_SELECTOR, 'div.btn').click() # 点击登录time.sleep(2)img_src = driver.execute_script('return document.getElementsByClassName("geetest_canvas_bg geetest_absolute")[0].toDataURL("image/png");')print(img_src)im_base64 = img_src.split(',')[1]im_bytes = base64.b64decode(im_base64)print(im_base64)print(im_bytes)print('-----------------------------------')with open('./bg.png', 'wb') as f:f.write(im_bytes)temp = driver.execute_script("return document.getElementsByClassName('geetest_canvas_fullbg geetest_fade geetest_absolute')[0].toDataURL('image/png');")print(temp)temp_base64 = temp.split(',')[1]temp_bytes = base64.b64decode(temp_base64)with open('./temp.png', 'wb') as f:f.write(temp_bytes)distance = text_dis('bg.png', 'temp.png') # 计算出距离print('距离:', distance)# 拖动滑块slide = driver.find_element(By.CSS_SELECTOR, 'div.geetest_slider_button')action_chains = webdriver.ActionChains(driver)# 点击,准备拖拽action_chains.click_and_hold(slide) # 鼠标左键点击但不释放action_chains.pause(0.2)action_chains.move_by_offset(distance - 10, 0)action_chains.pause(0.8)action_chains.move_by_offset(10, 0)action_chains.pause(1.4)action_chains.move_by_offset(-10, 0)action_chains.release()action_chains.perform()time.sleep(20)get_slide()
扩展
selenium常用模拟操作
1、行为控制
perform --- 执行所有准备好的Action
reset_actions --- 清空所有准备好的Action # 该方法在 selenium 3.141.0版本不生效
pause --- 设置Action之间的动作时间间隔2、鼠标操作
click --- 鼠标左键点击(可以指定或不指定元素对象)
click_and_hold --- 鼠标左键点击但不释放(可以指定或不指定元素对象)
release --- 释放鼠标点击动作(可以指定或不指定在目标元素对象上释放)
context_click --- 鼠标右键点击(可以指定或不指定元素对象)
double_click --- 鼠标左键双击(可以指定或不指定元素对象)
drag_and_drop --- 鼠标左键在两个元素之间拖拽
drag_and_drop_by_offset --- 鼠标左键拖拽元素到目标偏移位置
move_by_offset --- 鼠标移动指定偏移
move_to_element --- 鼠标移动到指定元素
move_to_element_with_offset --- 鼠标移动到指定元素的指定偏移位置
二、应对滑速检测代码
目标网址:点击
#!/usr/bin/env python
# -*- coding:utf-8 -*-
import json
import random
import re
import time
import cv2
import base64
import os
import numpy as np
from PIL import Image
from selenium import webdriver
from selenium.webdriver import ActionChains
from selenium.webdriver import ChromeOptions
from selenium.webdriver.common.by import By
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.support.wait import WebDriverWaitdef get_random_float(min, max, digits=4):return round(random.uniform(min, max), 4)def base64_to_image(base64_code, img_name):dir_path = re.sub(r'/([a-z]|_|-)*.(png|jp(e)?g)$', '', img_name)if not os.path.exists(dir_path):os.makedirs(dir_path)img_data = base64.b64decode(base64_code)file = open(img_name, 'wb')file.write(img_data)file.close()return img_nameclass JD_Register(object):# jd 注册页面验证码def __init__(self, url, username, pwd=''):super(JD_Register, self).__init__()self.url = url # 实际地址options = ChromeOptions()options.add_experimental_option('excludeSwitches', ['enable-automation'])self.driver = webdriver.Chrome(options=options)self.wait = WebDriverWait(self.driver, 10)self.username = username # 账户信息self.password = pwdself.target_path = "./target_reg.png"self.template_path = "./template_reg.png"self.zoom = 1 # 网页图片缩放def open_url(self, url=None):self.driver.maximize_window()self.driver.get(url if url else self.url)def main(self):self.open_url() # 打开网页self.loginOn() # 登录打开验证码self._crack_slider() # 模拟人滑动验证码def loginOn(self):print("------------------------ 进度1:填写账号~")self.driver.find_element(By.XPATH, '/html/body/div[4]/div[2]/div/div[2]/button').click() # 同意time.sleep(2)self.driver.find_element(By.ID, 'form-phone').send_keys('13697028751') # 输入框输入账号和密码time.sleep(2)self.driver.find_element(By.CLASS_NAME, 'form-item-getcode').click() # 点击登录time.sleep(2)def _crack_slider(self):pic_success = self._get_pic() # 获取图片if pic_success:# 模板匹配target = cv2.imread(self.target_path) # 查看完整图片template = cv2.imread(self.template_path) # 查看缺少图片distance = self._match_templet(target, template) # 计算缺口最大距离tracks = self._get_tracks(distance * self.zoom) # 拖拽轨迹计算self._slider_action(tracks) # 模拟人工移动滑块# 判断登录is_go_on = input('是否继续测试?y:是 其它:退出')if is_go_on and is_go_on.lower() == 'y':print("开始下一次尝试")return self._crack_slider()else:return Falsedef _get_pic(self):"""下载图片到本地"""# print("查找缺口图片")time.sleep(1)target = self.wait.until(EC.presence_of_element_located((By.XPATH, '//*[@id="slideAuthCode"]/div/div[1]/div[2]/div[1]/img')))template = self.wait.until(EC.presence_of_element_located((By.XPATH, '//*[@id="slideAuthCode"]/div/div[1]/div[2]/div[2]/img')))if target and template:print("------------------------ 进度2:下载滑块验证码图片")target_base64 = target.get_attribute('src')template_base64 = template.get_attribute('src')target_base64_str = re.sub(r'data:[a-z]*/[a-z]*;base64,', '', target_base64)template_base64_str = re.sub(r'data:[a-z]*/[a-z]*;base64,', '', template_base64)base64_to_image(target_base64_str, self.target_path)base64_to_image(template_base64_str, self.template_path)time.sleep(1)local_img = Image.open(self.target_path) # 打开图片size_loc = local_img.size # 获取图片大小尺寸self.zoom = 364 / int(size_loc[0]) # 对比判断图片是否一致,有无放大过print("计算缩放比例 zoom = %f" % round(self.zoom, 4))return Trueelse:print("未找到缺口图片")return Falsedef _match_templet(self, img_target, img_template):"""模板匹配(用于寻找缺口):param img_target: 带有缺口的背景图:param img_template: 缺口的滑块图:return: 缺口所在的位置的x轴距离"""print("------------------------ 进度3:图片缺口模板匹配")# 滑块图片处理tpl = self.__handle_slider_img(img_template) # 误差来源就在于滑块的背景图为白色blurred = cv2.GaussianBlur(img_target, (3, 3), 0) # 图片高斯滤波gray = cv2.cvtColor(blurred, cv2.COLOR_BGR2GRAY) # 图片灰度化width, height = tpl.shape[:2]result = cv2.matchTemplate(gray, tpl, cv2.TM_CCOEFF_NORMED) # 灰度化模板匹配 使用灰度化图片print("result = {}".format(len(np.where(result >= 0.5)[0])))# 查找数组中匹配的最大值min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(result)left_up = max_locright_down = (left_up[0] + height, left_up[1] + width) # 通过获取最远距离cv2.rectangle(img_target, left_up, right_down, (7, 279, 151), 2)print("验证码位移距离为:%d" % left_up[0])return left_up[0]def __handle_slider_img(self, image):"""对滑块进行二值化处理:param image: cv类型的图片对象"""kernel = np.ones((8, 8), np.uint8) # 去滑块的前景噪声内核gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # 灰度化# 灰化背景width, heigth = gray.shapefor h in range(heigth):for w in range(width):if gray[w, h] == 0:gray[w, h] = 96# 排除背景binary = cv2.inRange(gray, 96, 96)res = cv2.morphologyEx(binary, cv2.MORPH_OPEN, kernel) # 开运算去除白色噪点return resdef _get_cookie(self):cookie_items = self.driver.get_cookies()ck_dict = {}for cookie in cookie_items:ck_dict[cookie['name']] = cookie['value']print("cookie = %s" % ck_dict)self._save_to_file(json.dumps(ck_dict, separators=(',', ':'), ensure_ascii=False))self.driver.quit()def _save_to_file(self, str_data):file = Nonetry:file = open("../static/temp/cookie.txt", "w")file.write(str_data)except:print("保存cookie异常")finally:if file:file.close()# ---- 拖拽轨迹计算 start ----def _get_tracks(self, distance):"""根据偏移量获取移动轨迹3:param distance: 偏移量:return: 移动轨迹"""print('------------------------ 进度4:计算图片拖拽轨迹')track = []mid1 = round(distance * random.uniform(0.1, 0.2))mid2 = round(distance * random.uniform(0.65, 0.76))mid3 = round(distance * random.uniform(0.84, 0.88))# 设置初始位置、初始速度、时间间隔current, v, t = 0, 0, 0.2distance = round(distance)# 四段加速度while current < distance:if current < mid1:a = random.randint(10, 15)elif current < mid2:a = random.randint(30, 40)elif current < mid3:a = -70else:a = random.randint(-25, -18)v0 = v # 初速度 v0v = v0 + a * t # 当前速度 v = v0 + atv = v if v >= 0 else 0move = v0 * t + 1 / 2 * a * (t ** 2)move = round(move if move >= 0 else 1)current += move # 当前位移track.append(move) # 加入轨迹# 超出范围back_tracks = []out_range = distance - currentprint("当前= {}, 距离= {}, 超出范围 = {}".format(current, distance, out_range))if out_range < -8:sub = int(out_range + 8)back_tracks = [-1, sub, -3, -1, -1, -1, -1]elif out_range < -2:sub = int(out_range + 3)back_tracks = [-1, -1, sub]print("向前轨道= {}, 返回轨道={}".format(track, back_tracks))return {'forward_tracks': track, 'back_tracks': back_tracks}# ---- 移动滑块 start ----def _slider_action(self, tracks):print("------------------------ 进度5:模拟人工移动滑块")# 点击滑块slider = self.wait.until(EC.element_to_be_clickable((By.XPATH, '//*[@id="slideAuthCode"]/div/div[2]/div[3]')))if slider:ActionChains(self.driver).click_and_hold(slider).perform() # 鼠标左键点击不释放,保持滑动状态# 正向滑动for track in tracks['forward_tracks']:yoffset_random = random.uniform(-2, 4)ActionChains(self.driver).move_by_offset(xoffset=track, yoffset=yoffset_random).perform()time.sleep(random.uniform(0.06, 0.5))# 反向滑动for back_tracks in tracks['back_tracks']:yoffset_random = random.uniform(-2, 2)ActionChains(self.driver).move_by_offset(xoffset=back_tracks, yoffset=yoffset_random).perform()# 抖动ActionChains(self.driver).move_by_offset(xoffset=get_random_float(0, -1.67), yoffset=get_random_float(-1, 1)).perform()ActionChains(self.driver).move_by_offset(xoffset=get_random_float(0, 1.67), yoffset=get_random_float(-1, 1)).perform()time.sleep(get_random_float(0.2, 0.6))ActionChains(self.driver).release().perform()print("滑块移动成功")return Trueelse:print("未找到滑块")return Falseif __name__ == '__main__':c = JD_Register(url='https://reg.jd.com/p/regPage', username='13965216565')c.main()