I have this aspect that is called on every method annotated with #LogActivity annotation.
#Aspect
#Component
public class EndpointAspect {
#Around("#annotation(LogActivity)")
public Object logActivity(ProceedingJoinPoint joinPoint) throws Throwable {
Object result = joinPoint.proceed();
// I want to access message here. see below
return result;
}
}
And this is one annotated method:
public class contrlr {
#LogActivity
#GetRequest("/home")
public String aMethod() {
String message = "some RUNTIME-GENERATED message"; // I want to send it to aspect
return "home.html"
}
}
I want to pass the message to aspect. note that message may not be hard-coded constant. How to do that?
Related
I had written an aspect as part of an application using Spring AOP/AspectJ annotations similar to below aspect:
#Aspect
#Component
public class LoggingAspect {
#Around("#annotation(loggable)")
public Object log(final ProceedingJoinPoint joinPoint, final Loggable loggable) throws Throwable {
//log method arguments
try {
Object returnValue = joinPoint.proceed();
// log return value
return returnValue;
} catch (Exception ex) {
// publish exception metrics to some other system
throw ex;
}
}
}
Now I want to use this same aspect in another project, but this project uses Guice instead of Spring.
I was reading about Guice AOP which requires aspect to implement the MethodInterceptor interface and thus I will need to implement the below method:
Object invoke(MethodInvocation methodInvocation) throws Throwable;
What I was thinking was to modify the already existent aspect to implement the MethodInterceptor and internally call the log method. Something like below:
#Aspect
#Component
public class LoggingAspect implements MethodInterceptor {
#Override
public Object invoke(MethodInvocation methodInvocation) throws Throwable {
// call already defined log method, but that method expects a ProceedingJoinPoint, however
// I get MethodInvocation as input parameter in this method
}
// already defined log method
#Around("#annotation(loggable)")
public Object log(final ProceedingJoinPoint joinPoint, final Loggable loggable) throws Throwable {
......
.....
}
But due to incompatible type between two methods, I am unable to proceed.
Is there a way I can reuse the existing code instead of writing a brand new aspect with duplicate code to support Guice?
If I understand correctly, you want to reverse the control flow, which can be done with callbacks.
#Aspect
#Component
class LoggingAspect implements MethodInterceptor {
#Around("#annotation(loggable)")
public Object log(final ProceedingJoinPoint joinPoint, final Loggable loggable) throws Throwable {
return log(joinPoint::getArgs, () -> joinPoint.proceed(joinPoint.getArgs()));
}
#Override
public Object invoke(MethodInvocation methodInvocation) throws Throwable {
return log(methodInvocation::getArguments, methodInvocation::proceed);
}
public Object log(Supplier<Object[]> arguments, Supplier<Object[]> proceed) {
Object[] args = arguments.get();
//log method arguments
try {
Object returnValue = proceed.get();
// log return value
return returnValue;
} catch (Exception ex) {
// publish exception metrics to some other system
throw ex;
}
}
}
BTW do you intentionally catch only Exception and not Throwable? Errors would not be logged.
Given the following basic domain model:
abstract class BaseData { ... }
class DataA extends BaseData { ... }
class DataB extends BaseData { ... }
I want to write a Spring MVC controller endpoint thus ...
#PostMapping(path="/{typeOfData}", ...)
ResponseEntity<Void> postData(#RequestBody BaseData baseData) { ... }
The required concrete type of baseData can be inferred from the typeOfData in the path.
This allows me to have a single method that can handle multiple URLs with different body payloads. I would have a concrete type for each payload but I don't want to have to create multiple controller methods that all do the same thing (albeit each would do very little).
The challenge that I am facing is how to "inform" the deserialization process so that the correct concrete type is instantiated.
I can think of two ways to do this.
First use a custom HttpMessageConverter ...
#Bean
HttpMessageConverter httpMessageConverter() {
return new MappingJackson2HttpMessageConverter() {
#Override
public Object read(final Type type, final Class<?> contextClass, final HttpInputMessage inputMessage)
throws IOException, HttpMessageNotReadableException {
// TODO How can I set this dynamically ?
final Type subType = DataA.class;
return super.read(subType, contextClass, inputMessage);
}
};
}
... which gives me the challenge to determine the subType based on the HttpInputMessage. Possibly I could use a Filter to set a custom header earlier when the URL is available to me, or I could use a ThreadLocal also set via a Filter. Neither sounds ideal to me.
My second approach would be to again use a Filter and this time wrap the incoming payload in an outer object which would then provide the type in a way that enables Jackson to do the work via #JsonTypeInfo. At the moment this is probably my preferred approach.
I have investigated HandlerMethodArgumentResolver but if I try to register a custom one it is registered AFTER the RequestResponseBodyMethodProcessor and that class takes priority.
Hmm, so after typing all of that out I had a quick check of something in the RequestResponseBodyMethodProcessor before posting the question and found another avenue to explore, which worked neatly.
Excuse the #Configuration / #RestController / WebMvcConfigurer mash-up and public fields, all for brevity. Here's what worked for me and achieved exactly what I wanted:
#Configuration
#RestController
#RequestMapping("/dummy")
public class DummyController implements WebMvcConfigurer {
#Target(ElementType.PARAMETER)
#Retention(RetentionPolicy.RUNTIME)
#Documented
#interface BaseData {}
public static class AbstractBaseData {}
public static class DataA extends AbstractBaseData {
public String a;
}
public static class DataB extends AbstractBaseData {
public String b;
}
private final MappingJackson2HttpMessageConverter converter;
DummyController(final MappingJackson2HttpMessageConverter converter) {
this.converter = converter;
}
#Override
public void addArgumentResolvers(List<HandlerMethodArgumentResolver> resolvers) {
resolvers.add(
new RequestResponseBodyMethodProcessor(Collections.singletonList(converter)) {
#Override
public boolean supportsParameter(MethodParameter parameter) {
return parameter.hasParameterAnnotation(BaseData.class)
&& parameter.getParameterType() == AbstractBaseData.class;
}
#Override
protected <T> Object readWithMessageConverters(
NativeWebRequest webRequest, MethodParameter parameter, Type paramType)
throws IOException, HttpMediaTypeNotSupportedException,
HttpMessageNotReadableException {
final String uri =
webRequest.getNativeRequest(HttpServletRequest.class).getRequestURI();
return super.readWithMessageConverters(
webRequest, parameter, determineActualType(webRequest, uri));
}
private Type determineActualType(NativeWebRequest webRequest, String uri) {
if (uri.endsWith("data-a")) {
return DataA.class;
} else if (uri.endsWith("data-b")) {
return DataB.class;
}
throw new HttpMessageNotReadableException(
"Unable to determine actual type for request URI",
new ServletServerHttpRequest(
webRequest.getNativeRequest(HttpServletRequest.class)));
}
});
}
#PostMapping(
path = "/{type}",
consumes = MediaType.APPLICATION_JSON_VALUE,
produces = MediaType.APPLICATION_JSON_VALUE)
ResponseEntity<? extends AbstractBaseData> post(#BaseData AbstractBaseData baseData) {
return ResponseEntity.ok(baseData);
}
}
The key to this is that I stopped using #RequestBody because that is what was preventing me overriding the built-in behaviour. By using #BaseData instead I get a HandlerMethodArgumentResolver that uniquely supports the parameter.
Other than that it was a case of assembling the two objects that already did what I needed, so autowire a MappingJackson2HttpMessageConverter and instantiate a RequestResponseBodyMethodProcessor with that one converter. Then pick the right method to override so that I could control what parameter type was used at a point that I had access to the URI.
Quick test. Given the following payload for both requests ...
{
"a": "A",
"b": "B"
}
POST http://localhost:8081/dummy/data-a
... gives a response of ...
{
"a": "A"
}
POST http://localhost:8081/dummy/data-b
... gives a response of ...
{
"b": "B"
}
In our real-world example this means that we will be able to write one method each that supports the POST / PUT. We need to build the objects and configure the validation possibly - or alternatively if we use OpenAPI 3.0 which we are investigating we could generate the model and validate without writing any further code ... but that's a separate task ;)
Say I have an annotation (#RequiresAccount) introduced in another library and I'm using it in my project, is there a way to conditionally apply it to a method, e.g. apply it when the customer is from website A and not apply when customer is from website B?
I've taken a look and the only possibility I've found was, creating a wrapper-Annotation:
#Aspect
#Component
public class RequiresAccountWrapperAspect {
#Autowired
private HttpServletRequest request;
private RequiresAccountAspect requiresAccountAspect = new RequiresAccountAspect();
#Around("#annotation(com.example.demo.components.RequiresAccountWrapper)")
public Object checkIfRequiresAccount(ProceedingJoinPoint joinPoint) throws Throwable {
String requestURL = request.getRequestURL().toString();
if (requestURL.startsWith("http://localhost")) {
requiresAccountAspect.checkAccount(joinPoint);
}
return joinPoint.proceed();
}
}
So everywhere you've used your RequiresAccount annotation, you can use this wrapper instead. For example:
#GetMapping("/test")
#RequiresAccountWrapper
public String h() {
return "test";
}
As you can see I'm creating a new instance of the aspect. I don't know if you have access to the Aspect-class itself but if you have you can then call the method in it and pass the joinPoint. To find the URL from the request you can inject the HttpServletRequest.
I'm trying to enrich the SLF4J MDC on each request with the user's ID. The problem is that the ID can be passed in many ways, sometimes as a path parameter, sometimes in the body, and sometimes injected by a custom ValueFactoryProvider that first decrypts it.
If I could somehow access all the injected (i.e. already deserialized) parameter values, I could handle all these cases easily.
E.g.
For a resource such as:
#GET
//#Encrypted params are injected by a custom ValueFactoryProvider
public Something getSomething(#Encrypted("userId") String userId) {
return ...;
}
#POST
public Something getSomething(#RequestBody RequestWithUserId requestWithUserId) {
return ...;
}
I could have a filter such as:
public class MdcFilter implements ContainerRequestFilter, ContainerResponseFilter {
#Context
private ResourceInfo resourceInfo;
#Override
public void filter(ContainerRequestContext requestContext) throws IOException {
Method theMethod = resourceInfo.getResourceMethod();
for (Parameter parameter : theMethod.getParameters()) {
//Deal with the #Encrypted case
if (parameter.isAnnotationPresent(Encrypted.class) && parameter.getAnnotation(Encrypted.class).value().equals("userId")) {
MDC.put("userId", somehowGetTheValue());
}
//Deal with the #RequestBody case
if (parameter.isAnnotationPresent(RequestBody.class) && parameter.getType().equals(RequestWithUserId.class)) {
MDC.put("userId", ((RequestWithUserId)somehowGetTheValue()).getUserId());
}
... //other possibilities
}
}
#Override
public void filter(ContainerRequestContext requestContext, ContainerResponseContext responseContext) throws IOException {
MDC.clear();
}
}
But I don't see a way to implement somehowGetTheValue either from a ContainerRequestFilter an interceptor or anything else...
Jersey uses HK2 under the hood for dependency injection. And HK2 has AOP support. One option for your use case would be use this AOP support. All you need to do is implement a MethodInterceptor and an InterceptionService. In the MethodInterceptor, you can get all the arguments from the MethodInvocation and you can get parameter annotation from the Method
class MyMethodInteceptor implements MethodInterceptor {
#Override
public Object invoke(MethodInvocation invocation) throws Throwable {
Method method = invocation.getMethod();
Object[] args = invocation.getArguments();
// do your logging or whatever with the args.
// invoke method and get return value.
Object returnValue = invocation.proceed();
// if you want to do something with the return
// value before returning it, you can.
return returnValue;
}
}
To use the interceptor, you configure the InterceptionService.
public class MyInterceptionService implements InterceptionService {
private final static MethodInterceptor METHOD_INTERCEPTOR
= new MyMethodInterceptor();
private final static List<MethodInterceptor> METHOD_LIST
= Collections.singletonList(METHOD_INTERCEPTOR);
#Override
public Filter getDescriptorFilter() {
return BuilderHelper.allFilter();
}
#Override
public List<MethodInterceptor> getMethodInterceptors(Method method) {
// you implement shouldIntercept
if (shouldIntercept(method)) {
return METHOD_LIST;
}
return null;
}
#Override
public List<ConstructorInterceptor> getConstructorInterceptors(Constructor<?> constructor) {
return null;
}
}
You determine which method should be intercepted in the getMethodInterceptors() method. If the method should be intercepted, then return a list of interceptors, otherwise return null. A common way of handling this is to create a custom annotation and just annotate the method. The in the above method, just check
if (method.isAnnotationPresent(YourAnno.class)) {
return METHOD_LIST;
}
To make it all work, you just need to register the InteceptionService with HK2. You can do that in an AbstractBinder, which is what is used in a Jersey app to configure your DI.
ResourceConfig config = new ResourceConfig();
config.register(new AbstractBinder() {
#Override
protected void configure() {
bind(MyInterceptionService.class)
.to(InterceptionService.class)
.in(Singleton.class);
}
});
You can see a complete example in this GitHub repo. There is also an official example in the HK2 site. Just see "AOP support" the link at the top of the post.
You can get it like this
StringWriter stringWriter = new StringWriter();
IOUtils.copy(new InputStreamReader(requestContext.getEntityStream()), stringWriter);
System.out.println(stringWriter.toString());// String representation of the payload
requestContext.setEntityInputStream(new ByteArrayInputStream(requestEntity));
Basically the idea is to copy the stream and do any processing and then set the stream back. Because if you don't do that, then in your controller method you would get null, becuase the stream was already read.
Is there a way to create a custom, or use an existing, annotation to trigger code to run when the annotated method is called? Preferably, I would like to use Spring libraries.
For example:
#SendEmail("templateName")
public void doSomething() {
log.info("Something is happening");
}
public void sendEmail(String templateName) {
// This method is called everytime doSomething() is called
log.info("Sending email using template " + templateName);
}
#Component
#Aspect
public class Mail {
#After("execution (#com.yourdirectoryofyourcustomAnnotation.SendMail * *(..))")
public void sendEmail(JointPoint jp){
// it will send a mail after every method which tagged by your annotation
}
}