dp自动化登陆之hCaptcha 验证码

顾一大人

已于 2025-05-12 15:53:28 修改

阅读量189

点赞数 1

文章标签： python 自动化爬虫

于 2025-05-12 14:29:49 首次发布

本文链接：https://blog.csdn.net/qq_41828603/article/details/147893988

版权

hCaptcha 是一种常见的验证码服务，用于区分人类用户和自动化程序。由于其基于图像识别和行为分析，下面介绍如何使用自动化点击验证码完成登陆。
思路：登陆目标网站触发验证码，截图并发给打码平台返回坐标，模拟人工点击完成验证

1、注册识别 hCaptcha API，获取授权

在这里插入图片描述

import requests
def get_rect(base64_list, title_list):
	url = "https://api.acedata.cloud/captcha/recognition/hcaptcha"
	headers = {
	    "accept": "application/json",
	    "authorization": "授权码",
	    "content-type": "application/json"
	}
	payload = {
	      "queries": base64_list,
	      "question": ",".join(title_list)
	}
	response = requests.post(url, json=payload, headers=headers).json()
	return response

2、登陆目标网站，触发网站验证码

在这里插入图片描述

co = ChromiumOptions().headless(False).auto_port()
co.set_argument('--enable-translate')
page = ChromiumPage(co)
co.incognito()
page.get('https://www.ebay.com/sh/ord/?filter=status%3AALL_ORDERS%2Ctimerange%3APREVIOUSMONTH')
if page.title != "Security Measure":
    login(page, username_str, password_str)

3、将验证码截图获取图片并转为base64编码

def land():
	page.get_screenshot(path='tmp', name='pic.jpg', full_page=True)
	rangle = (74, 10, 474, 610)  # 左、上、右、下
	img = Image.open(r'tmp/pic.jpg')
	cropped_img = img.crop(rangle)
	# 保存裁剪后的图片
	cropped_img.save(r'tmp/cropped_image.png')
	title = (79, 15, 464, 130)
	img.crop(title).save(r"tmp/title.png")
	reader = easyocr.Reader(['ch_sim', 'en'])
    result = reader.readtext(img)
    title_list = []
    for detection in result:
        print(detection[1])
        title_list.append(detection[1])
	img1 = (79, 135, 209, 260)
	img.crop(img1).save(r"tmp/img1.png")
	img2 = (209, 135, 339, 260)
	img.crop(img2).save(r"tmp/img2.png")
	img3 = (339, 135, 469,260)
	img.crop(img3).save(r"tmp/img3.png")
	img4 = (79, 265, 209, 395)
	img.crop(img4).save(r"tmp/img4.png")
	img5 = (209, 265, 339, 395)
	img.crop(img5).save(r"tmp/img5.png")
	img6 = (339, 265, 469, 395)
	img.crop(img6).save(r"tmp/img6.png")
	img7 = (79, 395, 209, 525)
	img.crop(img7).save(r"tmp/img7.png")
	img8 = (209, 395, 339, 525)
	img.crop(img8).save(r"tmp/img8.png")
	img9 = (339, 395, 469, 525)
	img.crop(img9).save(r"tmp/img9.png")
	base64_list = []
	for i in range(1, 10):
	    with  open(f'tmp/img{i}.png', "rb") as img_file:
	        encoded_string = base64.b64encode(img_file.read()).decode('utf-8')
	        base64_list.append(encoded_string)
	return base64_list,title_list

4、将图片列表和标题传入图片识别api返回目标坐标，模拟完成点击

res = get_rect(base64_list, 'tmp/title.png')
objects = res["solution"]["objects"]
img_list = [img1, img2, img3, img4, img5, img6, img7, img8, img9]
print("1111", objects)
for i in range(9):
    if objects[i]:
        target = ((int(img_list[i][0]) + int(img_list[i][2])) / 2, (int(img_list[i][1]) + int(img_list[i][3])) / 2)
        page.actions.move_to(target).click()
        time.sleep(1)