近期参加的项目，主管丢给我一个任务，说要支持滑动验证码。我身为50岁的软件攻城师，当时正背着双手，好像一个受训的保安似的，中规中矩地参加每日站会，心想滑动验证码在今时今日已经是标配了，司空见惯，想必网上一搜一大把，岂非手到擒来。so easy，妈妈再也不用担心我的工作与学习。
孰料在网上寻寻觅觅点点击击，结果就是凄凄惨惨戚戚。好像提的最多的就是AJ-Captcha，但居然貌似下线了，文档打不开，demo也不见。还有一个声称可能是最好的滑动验证码，但好像很复杂，并且日本少女漫画风，跟我有代沟。有一个貌似跟Ant Design有点关联的组件，叫Wetrial的，好像还比较符合我的要求。但它只有前端，没有给出后端实现，并且它的前端好像也用不了。
但是，这个Wetrial.SliderCaptcha阐述了从后端获得的数据，仿佛制订了一个滑动验证码的接口标准。加上我在搜索过程中，看到的一些具体提示，有了一些思路。考虑到这个滑动验证，不仅要给自己的web端使用，还要开放给开发手机APP的外包人员调用，因此需要可控、便利、清晰，决定自己搞一个。
一、思路
1、背景图片和拼图图片都从后端，以base64的方式返回给前端
2、一起返回给前端的是一个json对象，包括背景和拼图内容、尺寸、token。token的作用是验证时即销毁，避免重放攻击，即每张背景图只验证一次
3、准备多张相同尺寸，不同内容的背景图，每次随机选一张
4、拼图从背景图中抠，抠后的坑填上白色，然后采集背景图的颜色，生成噪点加入这个坑。为的是避免机器容易识别这个白坑。
在chapGPT的指导下，历时一天，终于搞了个demo。效果如下
滑动验证
二、后端
后端就2个接口，一个供数据下载，一个供验证。
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.core.io.Resource;
import org.springframework.core.io.support.PathMatchingResourcePatternResolver;
import org.springframework.data.redis.core.StringRedisTemplate;
import org.springframework.data.redis.core.ValueOperations;
import org.springframework.web.bind.annotation.*;import javax.annotation.PostConstruct;
import javax.imageio.ImageIO;
import javax.servlet.http.HttpServletRequest;
import java.awt.*;
import java.awt.image.BufferedImage;
import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.io.InputStream;
import java.util.*;
import java.util.concurrent.TimeUnit;@RestController
public class CaptchaController {@Autowiredprivate StringRedisTemplate redisTemplate;private String[] images;int puzzlePieceWidth = 40;int puzzlePieceHeight = 40;@PostConstructpublic void init() throws IOException {PathMatchingResourcePatternResolver resolver = new PathMatchingResourcePatternResolver();Resource[] resources = resolver.getResources("classpath:/images/*.jpg"); // 修改为 *.jpgimages = new String[resources.length];for (int i = 0; i < resources.length; i++) {images[i] = resources[i].getURI().toString();}}@GetMapping("/slideCaptcha")public Map<String, Object> getCaptcha() throws IOException {Map<String, Object> response = new HashMap<>();// 生成唯一的 tokenString token = UUID.randomUUID().toString();// 随机选择背景图像BufferedImage backgroundImage = getBgImg();// 生成拼图块的随机位置int puzzlePieceLeft = (int) (Math.random() * (backgroundImage.getWidth() - puzzlePieceWidth));int puzzlePieceTop = (int) (Math.random() * (backgroundImage.getHeight() - puzzlePieceHeight));// 创建拼图块BufferedImage puzzlePieceImage = new BufferedImage(puzzlePieceWidth, puzzlePieceHeight, BufferedImage.TYPE_INT_ARGB);Graphics2D puzzleG = puzzlePieceImage.createGraphics();puzzleG.drawImage(backgroundImage, 0, 0, puzzlePieceWidth, puzzlePieceHeight, puzzlePieceLeft, puzzlePieceTop, puzzlePieceLeft + puzzlePieceWidth, puzzlePieceTop + puzzlePieceHeight, null);puzzleG.dispose();// 在背景图像上掩盖拼图块setMask(backgroundImage, puzzlePieceLeft, puzzlePieceTop);// 将图像转换为 Base64ByteArrayOutputStream baos = new ByteArrayOutputStream();ImageIO.write(backgroundImage, "jpg", baos); // 保持为 "jpg"String backgroundImageBase64 = Base64.getEncoder().encodeToString(baos.toByteArray());baos.reset();ImageIO.write(puzzlePieceImage, "png", baos); // 保持为 "png" 以支持透明度String puzzlePieceBase64 = Base64.getEncoder().encodeToString(baos.toByteArray());// 缓存 token 和位置ValueOperations<String, String> ops = redisTemplate.opsForValue();ops.set(token, String.valueOf(puzzlePieceLeft), 5, TimeUnit.MINUTES);response.put("backgroundImage", backgroundImageBase64);response.put("puzzlePiece", puzzlePieceBase64);response.put("token", token);//response.put("puzzlePieceLeft", puzzlePieceLeft);//response.put("puzzlePieceTop", puzzlePieceTop);response.put("backgroundWidth", backgroundImage.getWidth());response.put("backgroundHeight", backgroundImage.getHeight());response.put("puzzlePieceWidth", puzzlePieceWidth);response.put("puzzlePieceHeight", puzzlePieceHeight);return response;}@PostMapping("/slideVerify")public Map<String, Object> verifyCaptcha(HttpServletRequest request, @RequestBody Map<String, Object> map) {Map<String, Object> response = new HashMap<>();String token = (String) map.get("token");int position = (Integer) map.get("position");ValueOperations<String, String> ops = redisTemplate.opsForValue();String correctPositionStr = ops.get(token);if (correctPositionStr != null) {int correctPosition = Integer.parseInt(correctPositionStr);if (Math.abs(position - correctPosition) < 10) {response.put("success", true);} else {response.put("success", false);}redisTemplate.delete(token);} else {response.put("success", false);}return response;}private BufferedImage getBgImg() throws IOException {String selectedImage = images[(int) (Math.random() * images.length)];PathMatchingResourcePatternResolver resolver = new PathMatchingResourcePatternResolver();Resource resource = resolver.getResource(selectedImage);InputStream inputStream = resource.getInputStream();return ImageIO.read(inputStream);}private void setMask(BufferedImage backgroundImage, int puzzlePieceLeft, int puzzlePieceTop) {Graphics2D g = backgroundImage.createGraphics();g.setComposite(AlphaComposite.Src);g.setColor(Color.WHITE); // 使用白色填充g.fillRect(puzzlePieceLeft, puzzlePieceTop, puzzlePieceWidth, puzzlePieceHeight);// 从整幅背景图像采集颜色Color[][] sampledColors = new Color[backgroundImage.getWidth()][backgroundImage.getHeight()];for (int x = 0; x < backgroundImage.getWidth(); x++) {for (int y = 0; y < backgroundImage.getHeight(); y++) {sampledColors[x][y] = new Color(backgroundImage.getRGB(x, y));}}for (int i = puzzlePieceLeft; i < puzzlePieceLeft + puzzlePieceWidth; i++) {for (int j = puzzlePieceTop; j < puzzlePieceTop + puzzlePieceHeight; j++) {// 获取背景区域的颜色Color noiseColor = sampledColors[(int) (Math.random() * i)][(int) (Math.random() * j)];// 绘制扰乱元素g.setColor(noiseColor);g.fillRect(i, j, 1, 1); // 绘制单个像素点，覆盖原始的白色矩形}}g.dispose();}
}
三、前端
demo使用经典的html + js + css来编写。注意请求后台的接口路径采用了nginx进行转发，避免浏览器的跨域限制.
<!DOCTYPE html>
<html lang="en">
<head><meta charset="UTF-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><title>Captcha Verification</title><style>.captcha-container {position: relative;width: 367px;height: 267px;margin: 50px auto;border: 1px solid #ddd;background-color: #f3f3f3;}.background-image {position: absolute;top: 0;left: 0;width: 100%;height: 100%;}.puzzle-piece {position: absolute;width: 40px;height: 40px;cursor: move;box-shadow: 0 0 10px rgba(0, 0, 0, 0.3); /* 添加阴影效果 */}.slider-container {width: 400px;margin: 20px auto;text-align: center;display: flex;align-items: center;justify-content: center;}.slider {width: 100%;-webkit-appearance: none; /* 去除默认样式 */appearance: none;height: 10px; /* 设置滑道高度 */background: #ddd; /* 滑道背景色 */border-radius: 5px; /* 圆角 */outline: none; /* 去除聚焦时的外边框 */transition: background .2s; /* 过渡效果 */}.slider::-webkit-slider-thumb {-webkit-appearance: none; /* 去除默认样式 */appearance: none;width: 20px; /* 滑块宽度 */height: 20px; /* 滑块高度 */background: #4CAF50; /* 滑块背景色 */border-radius: 50%; /* 圆形 */cursor: pointer; /* 光标样式 */box-shadow: 0 0 5px rgba(0, 0, 0, 0.5); /* 滑块阴影效果 */}.refresh-btn {margin-left: 10px;padding: 8px 16px;cursor: pointer;background-color: #4CAF50;color: white;border: none;border-radius: 4px;font-size: 14px;}</style><!-- Font Awesome CSS --><link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.15.3/css/all.min.css">
</head>
<body><div class="captcha-container"><img id="backgroundImage" class="background-image" src="" alt="Background Image"><div id="puzzlePiece" class="puzzle-piece"></div></div><div class="slider-container"><input type="range" min="0" max="327" value="0" class="slider" id="slider"><button class="refresh-btn" id="refreshBtn"><i class="fas fa-sync-alt"></i></button></div><script>document.addEventListener('DOMContentLoaded', function() {let slider = document.getElementById('slider');let puzzlePiece = document.getElementById('puzzlePiece');let token = '';function loadCaptcha() {fetch('/api/slideCaptcha') // 替换为你的后端接口地址.then(response => response.json()).then(data => {document.getElementById('backgroundImage').src = 'data:image/jpeg;base64,' + data.backgroundImage;puzzlePiece.style.backgroundImage = 'url(data:image/jpeg;base64,' + data.puzzlePiece + ')';puzzlePiece.style.top = data.puzzlePieceTop + 'px';puzzlePiece.style.left = '0px';token = data.token;slider.value = 0;}).catch(error => console.error('Error fetching captcha:', error));}let refreshBtn = document.getElementById('refreshBtn');refreshBtn.addEventListener('click', function() {loadCaptcha();});slider.addEventListener('input', function() {puzzlePiece.style.left = slider.value + 'px';});slider.addEventListener('change', function() {fetch('/api/slideVerify', { // 替换为你的后端验证接口地址method: 'POST',headers: {'Content-Type': 'application/json',},body: JSON.stringify({token: token,position: parseInt(slider.value)}),}).then(response => response.json()).then(data => {if (data.success) {alert(':-) 验证成功！');} else {alert('验证失败，请重试！');}loadCaptcha();}).catch(error => console.error('Error verifying captcha:', error));});loadCaptcha();});</script>
</body>
</html>四、小结
俄国10月革命一声炮响，送来了美国的chatGPT。chatGPT吧，已经成了我的老师和工人。上面那些代码，都是我提要求，然后chatGPT生成的，甚至包括注释。我只修改了极少的地方。功能的确强大。但它其实又还不够智能，一些算法我一下子能看出问题，需要重重复复地提要求，每次它都说：明白了。它输入了海量的资料，知识渊博，各种编程语法更是精通，提交代码给它审查找问题，最是合适不过。它一般也能按要求给出初始代码，但有时总是差那么点意思。最讨厌的，是问它一些社科历史类的问题，经常一本正经地胡说八道。
这不是我想要的生活。
参考文章：
SlideCaptcha - 滑动验证码
滑块验证 - 使用AJ-Captcha插件【超简单.jpg】
TIANAI-CAPTCHA