京东自动登录过滑块、选图形验证码

环境准备

1. 安装Python:

   • 确保已安装Python 3.7及以上版本。

   • 您可以通过命令 python --version 检查Python版本。

2. 安装依赖库:

   • 安装 playwright, ddddocr, 和 opencv-python 库。

   • 使用命令行运行以下命令:

     1 pip install playwright ddddocr opencv-python

3. 下载Playwright浏览器数据:

   • 根据您的操作系统下载对应的Playwright浏览器数据。

   • 运行命令:

     1 python -m playwright install chromium

代码结构解析

1. 导入必要的模块

1 import asyncio
2 import os
3 import json
4 import cv2
5 import ddddocr
6 import requests
7 from playwright.async_api import async_playwright
8 import random

2. 初始化ddddocr实例

在代码的开头，我们初始化了ddddocr的实例ocr，用于后续的OCR识别。

# 传参获得已初始化的ddddocr实例
ocr = None

3. 支持的颜色和形状

我们定义了支持的颜色和形状类型，用于处理颜色和形状验证码。

# 支持的形状类型
supported_types = [
    "三角形",
    "正方形",
    "长方形",
    "五角星",
    "六边形",
    "圆形",
    "梯形",
    "圆环",
]

# 定义了支持的每种颜色的 HSV 范围
supported_colors = {
    "紫色": ([125, 50, 50], [145, 255, 255]),
    "灰色": ([0, 0, 50], [180, 50, 255]),
    "粉色": ([160, 50, 50], [180, 255, 255]),
    "蓝色": ([100, 50, 50], [130, 255, 255]),
    "绿色": ([40, 50, 50], [80, 255, 255]),
    "橙色": ([10, 50, 50], [25, 255, 255]),
    "黄色": ([25, 50, 50], [35, 255, 255]),
    "红色": ([0, 50, 50], [10, 255, 255]),
}

4. 初始化浏览器

在main函数中，我们初始化了浏览器的路径，并根据操作系统选择合适的浏览器路径。

async def main(workList, uid, oocr):
    global ocr
    ocr = oocr

    async def init_chrome():
        if platform.system() == "Windows":
            # Windows系统的浏览器路径初始化
            chrome_dir = os.path.join(
                os.environ["USERPROFILE"],
                "AppData",
                "Local",
                "pyppeteer",
                "pyppeteer",
                "local-chromium",
                "588429",
                "chrome-win32",
            )
            chrome_exe = os.path.join(chrome_dir, "chrome.exe")
            if os.path.exists(chrome_exe):
                return chrome_exe
            else:
                # 下载并解压浏览器
                print("貌似第一次使用，未找到chrome，正在下载chrome浏览器....")
                chromeurl = "https://mirrors.huaweicloud.com/chromium-browser-snapshots/Win_x64/588429/chrome-win32.zip"
                target_file = "chrome-win.zip"
                await download_file(chromeurl, target_file)
                with zipfile.ZipFile(target_file, "r") as zip_ref:
                    zip_ref.extractall(chrome_dir)
                os.remove(target_file)
                return chrome_exe
        elif platform.system() == "Linux":
            # Linux系统的浏览器路径初始化
            chrome_path = os.path.expanduser(
                "~/.local/share/pyppeteer/local-chromium/1181205/chrome-linux/chrome"
            )
            download_path = os.path.expanduser(
                "~/.local/share/pyppeteer/local-chromium/1181205/"
            )
            if os.path.isfile(chrome_path):
                return chrome_path
            else:
                print("貌似第一次使用，未找到chrome，正在下载chrome浏览器....")
                download_url = "https://mirrors.huaweicloud.com/chromium-browser-snapshots/Linux_x64/884014/chrome-linux.zip"
                target_file = os.path.join(
                    download_path, "chrome-linux.zip"
                )
                await download_file(download_url, target_file)
                with zipfile.ZipFile(target_file, "r") as zip_ref:
                    zip_ref.extractall(download_path)
                os.remove(target_file)
                os.chmod(chrome_path, 0o755)
                return chrome_path
        elif platform.system() == "Darwin":
            return "mac"
        else:
            return "unknown"

5. 处理验证码

滑块验证码

在verification函数中，我们通过OpenCV计算滑块的滑动距离，并模拟滑动操作。

async def verification(page):
    print("开始过滑块")

    async def get_distance():
        # 使用OpenCV计算滑动距离
        img = cv2.imread("image.png", 0)
        template = cv2.imread("template.png", 0)
        img = cv2.GaussianBlur(img, (5, 5), 0)
        template = cv2.GaussianBlur(template, (5, 5), 0)
        bg_edge = cv2.Canny(img, 100, 200)
        cut_edge = cv2.Canny(template, 100, 200)
        img = cv2.cvtColor(bg_edge, cv2.COLOR_GRAY2RGB)
        template = cv2.cvtColor(cut_edge, cv2.COLOR_GRAY2RGB)
        res = cv2.matchTemplate(img, template, cv2.TM_CCOEFF_NORMED)
        value = cv2.minMaxLoc(res)[3][0]
        distance = value + random.uniform(5, 15)
        print(f"计算出的滑动距离: {distance}")
        return distance

    # 下载并处理验证码图片
    await page.waitForSelector("#cpc_img", {'timeout': 10000})
    image_src = await page.Jeval("#cpc_img", 'el => el.getAttribute("src")')
    request.urlretrieve(image_src, "image.png")
    width = await page.evaluate('() => document.getElementById("cpc_img").clientWidth')
    height = await page.evaluate('() => document.getElementById("cpc_img").clientHeight')
    image = Image.open("image.png")
    resized_image = image.resize((width, height))
    resized_image.save("image.png")

    template_src = await page.Jeval("#small_img", 'el => el.getAttribute("src")')
    request.urlretrieve(template_src, "template.png")
    width = await page.evaluate('() => document.getElementById("small_img").clientWidth')
    height = await page.evaluate('() => document.getElementById("small_img").clientHeight')
    image = Image.open("template.png")
    resized_image = image.resize((width, height))
    resized_image.save("template.png")

    el = await page.querySelector("#captcha_modal > div > div.captcha_footer > div > img")
    box = await el.boundingBox()
    distance = await get_distance()

    # 模拟滑动操作
    await page.mouse.move(box["x"] + 10, box["y"] + 10)
    await page.mouse.down()
    steps = 30
    for i in range(steps):
        progress = i / steps
        x = box["x"] + 10 + (distance * (progress ** 2)) + random.uniform(-2, 2)
        y = box["y"] + 10 + random.uniform(-3, 3)
        await page.mouse.move(x, y, {'steps': 1})
        await asyncio.sleep(random.uniform(0.03, 0.07))
    await page.mouse.up()

    print("过滑块结束")

形状验证码

在verification_shape函数中，我们通过ddddocr识别验证码中的文字，并结合OpenCV检测形状和颜色。

async def verification_shape(page):
    print("开始过颜色、形状验证")

    # 初始化文字检测器
    det = ddddocr.DdddOcr(det=True)

    def get_shape_location_by_type(img_path, type: str):
        # 通过OpenCV检测形状
        img = cv2.imread(img_path)
        imgGray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
        imgBlur = cv2.GaussianBlur(imgGray, (5, 5), 1)
        imgCanny = cv2.Canny(imgBlur, 60, 60)
        contours, hierarchy = cv2.findContours(
            imgCanny, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE
        )
        for obj in contours:
            perimeter = cv2.arcLength(obj, True)
            approx = cv2.approxPolyDP(obj, 0.02 * perimeter, True)
            CornerNum = len(approx)
            x, y, w, h = cv2.boundingRect(approx)

            if CornerNum == 3:
                obj_type = "三角形"
            elif CornerNum == 4:
                if w == h:
                    obj_type = "正方形"
                else:
                    obj_type = "长方形"
            elif CornerNum == 6:
                obj_type = "六边形"
            elif CornerNum == 8:
                obj_type = "圆形"
            elif CornerNum == 20:
                obj_type = "五角星"
            else:
                obj_type = "未知"

            if obj_type == type:

6. 获取Cookie

在登录成功后，我们通过getCookie函数获取页面的Cookie信息，特别是pt_key和pt_pin，这些是后续操作的重要凭证。

async def getCookie(page):
    cookies = await page.cookies()
    pt_key = ""
    pt_pin = ""
    for cookie in cookies:
        if cookie["name"] == "pt_key":
            pt_key = cookie["value"]
        elif cookie["name"] == "pt_pin":
            pt_pin = cookie["value"]
    ck = f"pt_key={pt_key};pt_pin={pt_pin};"
    print(f"登录成功 {ck}")
    return ck

7. 发送短信验证码

在sendSMS和sendSMSDirectly函数中，我们模拟用户点击发送短信验证码按钮，并处理可能出现的验证码弹窗。

async def sendSMS(page):
    async def preSendSMS(page):
        print("进行发送验证码前置操作")
        await page.waitForXPath(
            '//*[@id="app"]/div/div[2]/div[2]/span/a'
        )
        await page.waitFor(random.randint(1, 3) * 1000)
        elements = await page.xpath(
            '//*[@id="app"]/div/div[2]/div[2]/span/a'
        )
        await elements[0].click()
        await page.waitForXPath(
            '//*[@id="app"]/div/div[2]/div[2]/button'
        )
        await page.waitFor(random.randint(1, 3) * 1000)
        elements = await page.xpath(
            '//*[@id="app"]/div/div[2]/div[2]/button'
        )
        await elements[0].click()
        await page.waitFor(3000)

    await preSendSMS(page)
    print("开始发送验证码")

    try:
        while True:
            if await page.xpath('//*[@id="captcha_modal"]/div/div[3]/div'):
                await verification(page)

            elif await page.xpath('//*[@id="captcha_modal"]/div/div[3]/button'):
                await verification_shape(page)

            else:
                break

            await page.waitFor(3000)

    except Exception as e:
        raise e

8. 输入短信验证码

在typeSMScode函数中，我们等待用户输入短信验证码，并将其填入表单中。

async def typeSMScode(page, workList, uid):
    print("开始输入验证码")

    async def get_verification_code(workList, uid):
        print("开始从全局变量获取验证码")
        retry = 60
        while not workList[uid].SMS_CODE and not retry < 0:
            await asyncio.sleep(1)
            retry -= 1
        if retry < 0:
            workList[uid].status = "error"
            workList[uid].msg = "输入短信验证码超时"
            return

        workList[uid].status = "pending"
        return workList[uid].SMS_CODE

    await page.waitForXPath('//*[@id="app"]/div/div[2]/div[2]/div/input')
    code = await get_verification_code(workList, uid)
    if not code:
        return

    workList[uid].status = "pending"
    workList[uid].msg = "正在通过短信验证"
    input_elements = await page.xpath('//*[@id="app"]/div/div[2]/div[2]/div/input')

    try:
        if input_elements:
            input_value = await input_elements[0].getProperty("value")
            if input_value:
                print("清除验证码输入框中已有的验证码")
                await page.evaluate(
                    '(element) => element.value = ""', input_elements[0]
                )

    except Exception as e:
        print("typeSMScode" + str(e))

    await input_elements[0].type(code)
    await page.waitForXPath('//*[@id="app"]/div/div[2]/a[1]')
    await page.waitFor(random.randint(1, 3) * 1000)
    elements = await page.xpath('//*[@id="app"]/div/div[2]/a[1]')
    await elements[0].click()
    await page.waitFor(random.randint(2, 3) * 1000)

9. 错误处理和状态更新

在整个登录流程中，我们通过多个函数检查登录状态，处理错误情况，并更新任务状态。

async def isWrongAccountOrPassword(page, verify=False):
    try:
        element = await page.xpath('//*[@id="app"]/div/div[5]')
        if element:
            text = await page.evaluate(
                "(element) => element.textContent", element[0]
            )
            if text == "账号或密码不正确":
                if verify == True:
                    return True
                await asyncio.sleep(2)
                return await isWrongAccountOrPassword(page, verify=True)
        return False
    except Exception as e:
        print("isWrongAccountOrPassword " + str(e))
        return False

10. 任务管理和资源清理

在deleteSession函数中，我们管理任务状态，并在任务完成后清理资源。

async def deleteSession(workList, uid):
    s = workList.get(uid, "")
    if s:
        await asyncio.sleep(60)
        del workList[uid]

11. 主函数

在main函数中，我们初始化浏览器，启动登录流程，并在完成后清理临时文件。

async def main(workList, uid, oocr):
    global ocr
    ocr = oocr

    async def init_chrome():
        # 浏览器初始化逻辑，如前所述

    chromium_path = await init_chrome()
    headless = platform.system() != "Windows"
    await logon_main(chromium_path, workList, uid, headless)
    os.remove("image.png") if os.path.exists("image.png") else None
    os.remove("template.png") if os.path.exists("template.png") else None
    os.remove("shape_image.png") if os.path.exists("shape_image.png") else None
    os.remove("rgba_word_img.png") if os.path.exists("rgba_word_img.png") else None
    os.remove("rgb_word_img.png") if os.path.exists("rgb_word_img.png") else None
    print("登录完成")
    await asyncio.sleep(10)