我正在使用Spring MVC 4.1为一个安静的Web服务添加速率限制。
我创建了一个@RateLimited
注释,可以应用于控制器方法。 Spring AOP方面拦截对这些方法的调用,如果请求太多则抛出异常:
@Aspect
@Component
@Order(Ordered.HIGHEST_PRECEDENCE)
public class RateLimitingAspect {
@Autowired
private RateLimitService rateLimitService;
@Before("execution(* com.example..*.*(.., javax.servlet.ServletRequest+, ..)) " +
"&& @annotation(com.example.RateLimited)")
public void wait(JoinPoint jp) throws Throwable {
ServletRequest request =
Arrays
.stream(jp.getArgs())
.filter(Objects::nonNull)
.filter(arg -> ServletRequest.class.isAssignableFrom(arg.getClass()))
.map(ServletRequest.class::cast)
.findFirst()
.get();
String ip = request.getRemoteAddr();
int secondsToWait = rateLimitService.secondsUntilNextAllowedAttempt(ip);
if (secondsToWait > 0) {
throw new TooManyRequestsException(secondsToWait);
}
}
这一切都很有效,除非@RateLimited
控制器方法的参数标记为@Valid
,例如:
@RateLimited
@RequestMapping(method = RequestMethod.POST)
public HttpEntity<?> createAccount(
HttpServletRequest request,
@Valid @RequestBody CreateAccountRequestDto dto) {
...
}
问题:如果验证失败,验证程序将抛出由MethodArgumentNotValidException
处理的@ExceptionHandler
,它会向客户端返回错误响应,从不触发我的@Before
因此绕过速率限制。
如何以优先于参数验证的方式拦截此类网络请求?
我曾想过使用Spring Interceptors或普通的servlet过滤器,但它们是通过简单的url-patterns映射的,我需要通过GET / POST / PUT /等来区分。
答案 0 :(得分:3)
我最终放弃了尝试寻找AOP解决方案并改为创建一个Spring Interceptor。拦截器preHandle
所有请求和监视处理程序为@RateLimited
的请求。
@Component
public class RateLimitingInterceptor extends HandlerInterceptorAdapter {
@Autowired
private final RateLimitService rateLimitService;
@Override
public boolean preHandle(HttpServletRequest request, HttpServletResponse response, Object handler) throws Exception {
if (HandlerMethod.class.isAssignableFrom(handler.getClass())) {
rateLimit(request, (HandlerMethod)handler);
}
return super.preHandle(request, response, handler);
}
private void rateLimit(HttpServletRequest request, HandlerMethod handlerMethod) throws TooManyRequestsException {
if (handlerMethod.getMethodAnnotation(RateLimited.class) != null) {
String ip = request.getRemoteAddr();
int secondsToWait = rateLimitService.secondsUntilNextAllowedInvocation(ip);
if (secondsToWait > 0) {
throw new TooManyRequestsException(secondsToWait);
} else {
rateLimitService.recordInvocation(ip);
}
}
}
}
答案 1 :(得分:0)
看看你是否可以为@@ AfterThrowing建议实现类似的逻辑,它也会有类似的切入点。