I'd like to use the MvcUriComponentsBuilder::fromMethodCall method to build URLs from my controllers. I normally have a String return type (which returns the view name) and a Model instance as method parameter in my controller methods like:
#Controller
public class MyController {
#RequestMapping("/foo")
public String foo(Model uiModel) {
uiModel.addAttribute("pi", 3.1415);
return "fooView";
}
}
I try to generate a URL e.g. like:
String url = MvcUriComponentsBuilder.fromMethodCall(on(MyController.class).foo(null)).build().toUriString();
This leads to this exception:
org.springframework.web.util.NestedServletException: Request processing failed; nested exception is java.lang.IllegalArgumentException: Cannot subclass final class class java.lang.String
at org.springframework.web.servlet.FrameworkServlet.processRequest(FrameworkServlet.java:978) ~[spring-webmvc-4.1.4.RELEASE.jar:4.1.4.RELEASE]
This happens because the String return type wants to get proxied, but can't as a final class.
What's a way to overcome this? I'd like to keep the String as a return type and get the Model as input from a parameter in my controller methods because IMHO it's way easier than handling a ModelAndView instance in every controller method.
fromMethodCall uses CGLIB proxy in the process which is why you run into the issue. This article details why.https://github.com/spring-projects/spring-hateoas/issues/155. Try using fromMethodName if you want to maintain the String return types.
MvcUriComponentsBuilder.fromMethodName(MyController.class, "foo", new Object()).build();
Consider changing the signature of the method to return Spring's ModelAndView vs. returning String. For example:
#Controller
public class MyController {
#RequestMapping("/foo")
public ModelAndView foo() {
return new ModelAndView("fooView", "pi", 3.1415);
}
}
With this refactored signature, the corresponding fromMethodCall invocation would look like this:
UriComponents uri = fromMethodCall(on(MyController.class).foo()).build();
Related
I am writing an API using Spring MVC and I am coming up with a problem allowing apps written in different languages to consume my API.
It turns out that the "Ruby users" like to have their params named in snake_case and our "Java users" like to have their param names in camel_case.
Is it possible to create my methods that allow param names to be named multiple ways, but mapped to the same method variable?
For instance... If I have a method that accepts a number of variables, of them there is mapped to a postal code. Could I write my method with a #RequestParam that accepts BOTH "postal_code" and "postalCode" and maps it to the same variable?
Neither JAX-RS #QueryParam nor Spring #RequestParam support your requirement i.e., mapping multiple request parameter names to the same variable.
I recommend not to do this as it will be very hard to support because of the confusion like which parameter is coming from which client.
But if you really wanted to handle this ((because you can't change the URL coming from 3rd parties, agreed long back), then the alternative is to make use of HandlerMethodArgumentResolver which helps in passing our own request argument (like #MyRequestParam) to the controller method like as shown in the below code:
Controller class:
#Controller
public class MyController {
#RequestMapping(value="/xyz")
public void train1(#MyRequestParam String postcode) {//custom method argument injected
//Add your code here
}
}
MyRequestParam :
#Retention(RetentionPolicy.RUNTIME)
#Target(ElementType.PARAMETER)
public #interface MyRequestParam {
}
HandlerMethodArgumentResolver Impl class:
public class MyRequestParamWebArgumentResolver implements HandlerMethodArgumentResolver {
#Override
public Object resolveArgument(MethodParameter parameter,
ModelAndViewContainer mavContainer,
NativeWebRequest webRequest,
WebDataBinderFactory binderFactory) {
MyRequestParam myRequestParam =
parameter.getParameterAnnotation(MyRequestParam.class);
if(myRequestParam != null) {
HttpServletRequest request =
(HttpServletRequest) webRequest.getNativeRequest();
String myParamValueToBeSentToController = "";
//set the value from request.getParameter("postal_code")
//or request.getParameter("postalCode")
return myParamValueToBeSentToController;
}
return null;
}
#Override
public boolean supportsParameter(MethodParameter parameter) {
return (parameter.getParameterAnnotation(MyRequestParam.class) != null);
}
}
WebMvcConfigurerAdapter class:
#Configuration
class WebMvcContext extends WebMvcConfigurerAdapter {
#Override
public void addArgumentResolvers(List<HandlerMethodArgumentResolver> argumentResolvers) {
argumentResolvers.add(new MyRequestParamWebArgumentResolver());
}
}
I think what you want to do is not allowed by Spring framework with the annotation RequestParam.
But if you can change the code or say to your third party to modify the calls i would suggest you 2 options
Option 1:
Use the #PathVariable property
#RequestMapping(value = "/postalcode/{postalCode}", method = RequestMethod.GET)
public ModelAndView yourMethod(#PathVariable("postalCode") String postalCode) {
//...your code
Here does not matter if the are calling your URL as:
http://domain/app/postalcode/E1-2ES
http://domain/app/postalcode/23580
Option 2:
Create 2 methods in your controller and use the same service
#RequestMapping(value = "/postalcode", method = RequestMethod.GET, params={"postalCode"})
public ModelAndView yourMethod(#RequestParam("postalCode") String postalCode) {
//...call the service
#RequestMapping(value = "/postalcode", method = RequestMethod.GET, params={"postal_code"})
public ModelAndView yourMethodClient2(#RequestParam("postal_code") String postalCode) {
//...call the service
If is possible, I would suggest you option 1 is much more scalable
My controller. Note the custom #Exists annotation:
#RestController
public class ClientApiController {
#RequestMapping(path = "/{client}/someaction", method = RequestMethod.GET)
String handleRequest(#Exists Client client) {
// ...
}
}
The Exists annotation:
/**
* Indicates that a controller request mapping method parametet should not be
* null. This is meant to be used on model types to indicate a required entity.
*/
#Target(ElementType.PARAMETER)
#Retention(RetentionPolicy.RUNTIME)
#Documented
public #interface Exists {}
The converter which converts the String from the path variable into a Client instance:
#Component
public class StringToClient implements Converter<String, Client> {
#Autowired
private ClientDAO clientDAO;
#Override
public Client convert(String source) {
return clientDAO.getClientById(source);
}
}
The ResourceNotFoundException exception used to trigger a 404
#ResponseStatus(value = HttpStatus.NOT_FOUND)
public class ResourceNotFoundException extends RuntimeException {
}
My controller method receives the converted Client as desired. If the client id used in the URL matches a client, everything works fine. If the id doesn't match, the client parameter is null empty (uses default constructor) in the handle() controller method.
What I can't get to work now is declarative checking that the Client is not null (i.e. that the id refers to an existing client). If it's null, a ResourceNotFoundException should be thrown. Checking whether the argument is null in the method body and throwing my custom ResourceNotFoundException is easy to do, but repetitive (like this one does). Also, this declarative approach should work for all model classes implementing the interface ModelWithId so it can be used for multiple model types.
I've searched the Spring documentation and I haven't found how to achieve this. I need to insert some processing somewhere after type conversion and before the controller's handleRequest method.
I'm using Spring Boot 1.3.3
After type conversion and before the controller's method there is a validation. You can implement custom validator and raise exception in it. Add new validator to DataBinder, and mark method's parameter as #Validated:
#RestController
public class ClientApiController {
#InitBinder
public void initBinder(DataBinder binder){
binder.addValidators(new Validator() {
#Override
public boolean supports(Class<?> aClass) {
return aClass==Client.class;
}
#Override
public void validate(Object o, Errors errors) {
Client client = (Client)o;
if(client.getId()==null) throw new ResourceNotFoundException();
}
});
}
#RequestMapping(path = "/{client}/someaction", method = RequestMethod.GET)
String handleRequest(#Validated #Exists Client client) {
// ...
}
#RequestMapping(path = "/{client}/anotheraction", method = RequestMethod.GET)
String handleAnotherRequest(#Validated #Exists Client client) {
// ...
}
}
Of course, you can declare validator as separate class, and use it repeatedly in other controllers. Actually, you can raise exception right in your converter, but there is possibility, that you'll need the conversion without exception in other places of your application.
I have a GET request in the format below
http://www.example.com/companies?filters=%7B%22q%22%3A%22aaa%22%7D
After decode it is
filters={"q":"aaa"}
I have created an Object named Filters as below
public class Filters {
private String q;
//getter setter....
}
and in my controller
#RequestMapping(method = RequestMethod.GET)
public List<CompanyDTO> getCompanies(Filters filters) {
filters.getQ();
//do things
}
However, the filters.getQ() is null.
Am I doing something incorrect here?
You need to associate the request parameter to the method argument. Add #RequestParam to your method i.e.
#RequestMapping(method = RequestMethod.GET)
public List<CompanyDTO> getCompanies(#RequestParam(value="filters") Filters filters) {
filters.getQ();
//do things
}
Instead of #RequestParam, use #RequestBody
Instead of String filters=%7B%22q%22%3A%22aaa%22%7D, pass JSON object as parameter http://www.example.com/companies?filters={"q":"aaa"}
I have many controllers in my Spring MVC web application and there is a param mandatoryParam let's say which has to be present in all the requests to the web application.
Now I want to make that param-value available to all the methods in my web-layer and service-layer. How can I handle this scenario effectively?
Currently I am handling it in this way:
... controllerMethod(#RequestParam String mandatoryParam, ...)
and, then passing this param to service layer by calling it's method
#ControllerAdvice("net.myproject.mypackage")
public class MyControllerAdvice {
#ModelAttribute
public void myMethod(#RequestParam String mandatoryParam) {
// Use your mandatoryParam
}
}
myMethod() will be called for every request to any controller in the net.myproject.mypackage package. (Before Spring 4.0, you could not define a package. #ControllerAdvice applied to all controllers).
See the Spring Reference for more details on #ModelAttribute methods.
Thanks Alexey for leading the way.
His solution is:
Add a #ControllerAdvice triggering for all controllers, or selected ones
This #ControllerAdvice has a #PathVariable (for a "/path/{variable}" URL) or a #RequestParam (for a "?variable=..." in URL) to get the ID from the request (worth mentioning both annotations to avoid blind-"copy/past bug", true story ;-) )
This #ControllerAdvice then populates a model attribute with the data fetched from database (for instance)
The controllers with uses #ModelAttribute as method parameters to retrieve the data from the current request's model
I'd like to add a warning and a more complete example:
Warning: see JavaDoc for ModelAttribute.name() if no name is provided to the #ModelAttribute annotation (better to not clutter the code):
The default model attribute name is inferred from the declared
attribute type (i.e. the method parameter type or method return type),
based on the non-qualified class name:
e.g. "orderAddress" for class "mypackage.OrderAddress",
or "orderAddressList" for "List<mypackage.OrderAddress>".
The complete example:
#ControllerAdvice
public class ParentInjector {
#ModelAttribute
public void injectParent(#PathVariable long parentId, Model model) {
model.addAttribute("parentDTO", new ParentDTO(parentId, "A faked parent"));
}
}
#RestController
#RequestMapping("/api/parents/{parentId:[0-9]+}/childs")
public class ChildResource {
#GetMapping("/{childId:[0-9]+}")
public ChildDTO getOne(#ModelAttribute ParentDTO parent, long childId) {
return new ChildDTO(parent, childId, "A faked child");
}
}
To continue about the warning, requests are declaring the parameter "#ModelAttribute ParentDTO parent": the name of the model attribute is not the variable name ("parent"), nor the original "parentId", but the classname with first letter lowerified: "parentDTO", so we have to be careful to use model.addAttribute("parentDTO"...)
Edit: a simpler, less-error-prone, and more complete example:
#Target(ElementType.TYPE)
#Retention(RetentionPolicy.RUNTIME)
#Documented
#RestController
public #interface ProjectDependantRestController {
/**
* The value may indicate a suggestion for a logical component name,
* to be turned into a Spring bean in case of an autodetected component.
*
* #return the suggested component name, if any
*/
String value() default "";
}
#ControllerAdvice(annotations = ParentDependantRestController.class)
public class ParentInjector {
#ModelAttribute
public ParentDTO injectParent(#PathVariable long parentId) {
return new ParentDTO(parentId, "A faked parent");
}
}
#ParentDependantRestController
#RequestMapping("/api/parents/{parentId:[0-9]+}/childs")
public class ChildResource {
#GetMapping("/{childId:[0-9]+}")
public ChildDTO getOne(#ModelAttribute ParentDTO parent, long childId) {
return new ChildDTO(parent, childId, "A faked child");
}
}
I am using Spring 3.0.6, and have noted a few (apparently) contradictory statements regarding what happens when a controller method declares a void return type (or returns a null). Consider the following :
#Controller
#RequestMapping(value="admin/*")
public class AdminController {
#RequestMapping
public ResponseEntity<String> hello() {
System.out.println("hellooooooo");
}
}
This takes the view name as the method name as stated here. But as stated in the accepted answer here, the view name is derived from url (not the method name). The method in question was :
#Controller
#RequestMapping("/form")
public class FormController {
#RequestMapping(method=RequestMethod.POST)
public String processSubmit(#Valid FormBean form,
BindingResult result,
WebRequest webRequest,
HttpSession session, Model model) {
if (result.hasErrors()) {
return null;
} else {
session.setAttribute("form", form);
return "redirect:/form";
}
}
}
skaffman's answer in What does it mean when Spring MVC #Controller returns null view name? is the correct one, because code can not lie.
So according to the docs and to the code: the view name is derived from the url (not from the method name).
I think that the other question (Spring MVC #RequestMapping … using method name as action value?) (and its answer) are a bit missleading, because there the request url and the method name are the same.