Configured ObjectMapper not used in spring-boot-webflux - spring-boot

I have mixins configured in my objectmapperbuilder config, using the regular spring web controller, the data outputted according to the mixins.
However using webflux, a controller with a method returning a Flow or Mono have the data serialized like if the objectmapper a default one.
How to get webflux to enforce an objectmapper configuration to be used ?
sample config:
#Bean
JavaTimeModule javatimeModule(){
return new JavaTimeModule();
}
#Bean
Jackson2ObjectMapperBuilderCustomizer jackson2ObjectMapperBuilderCustomizer(){
return jacksonObjectMapperBuilder -> jacksonObjectMapperBuilder.featuresToEnable(SerializationFeature.WRITE_DATES_AS_TIMESTAMPS)
.mixIn(MyClass.class, MyClassMixin.class);
}

I actually found my solution by stepping through the init code:
#Configuration
public class Config {
#Bean
JavaTimeModule javatimeModule(){
return new JavaTimeModule();
}
#Bean
Jackson2ObjectMapperBuilderCustomizer jackson2ObjectMapperBuilderCustomizer(){
return jacksonObjectMapperBuilder -> jacksonObjectMapperBuilder.featuresToEnable(SerializationFeature.WRITE_DATES_AS_TIMESTAMPS)
.mixIn(MyClass.class, MyClassMixin.class);
}
#Bean
Jackson2JsonEncoder jackson2JsonEncoder(ObjectMapper mapper){
return new Jackson2JsonEncoder(mapper);
}
#Bean
Jackson2JsonDecoder jackson2JsonDecoder(ObjectMapper mapper){
return new Jackson2JsonDecoder(mapper);
}
#Bean
WebFluxConfigurer webFluxConfigurer(Jackson2JsonEncoder encoder, Jackson2JsonDecoder decoder){
return new WebFluxConfigurer() {
#Override
public void configureHttpMessageCodecs(ServerCodecConfigurer configurer) {
configurer.defaultCodecs().jackson2JsonEncoder(encoder);
configurer.defaultCodecs().jackson2JsonDecoder(decoder);
}
};
}
}

I translated the solution of #Alberto Galiana to Java and injected the configured Objectmapper for convenience, so you avoid having to do multiple configurations:
#Configuration
#RequiredArgsConstructor
public class WebFluxConfig implements WebFluxConfigurer {
private final ObjectMapper objectMapper;
public void configureHttpMessageCodecs(ServerCodecConfigurer configurer) {
configurer.defaultCodecs().jackson2JsonEncoder(
new Jackson2JsonEncoder(objectMapper)
);
configurer.defaultCodecs().jackson2JsonDecoder(
new Jackson2JsonDecoder(objectMapper)
);
}
}

Just implement WebFluxConfigurer and override method configureHttpMessageCodecs
Sample code for Spring Boot 2 + Kotlin
#Configuration
#EnableWebFlux
class WebConfiguration : WebFluxConfigurer {
override fun configureHttpMessageCodecs(configurer: ServerCodecConfigurer) {
configurer.defaultCodecs().jackson2JsonEncoder(Jackson2JsonEncoder(ObjectMapper()
.setSerializationInclusion(JsonInclude.Include.NON_EMPTY)))
configurer.defaultCodecs().jackson2JsonDecoder(Jackson2JsonDecoder(ObjectMapper()
.enable(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES)))
}
}
Make sure all your data classes to be encoded/decoded have all its properties annotated with #JsonProperty even if property name is equal in class and json data
data class MyClass(
#NotNull
#JsonProperty("id")
val id: String,
#NotNull
#JsonProperty("my_name")
val name: String)

In my case, I was trying to use a customized ObjectMapper while inheriting all of the behavior from my app's default WebClient.
I found that I had to use WebClient.Builder.codecs. When I used WebClient.Builder.exchangeStrategies, the provided overrides were ignored. Not sure if this behavior is something specific to using WebClient.mutate, but this is the only solution I found that worked.
WebClient customizedWebClient = webClient.mutate()
.codecs(clientCodecConfigurer ->
clientCodecConfigurer.defaultCodecs()
.jackson2JsonDecoder(new Jackson2JsonDecoder(customObjectMapper)))
.build();

I have tried all the different solutions (#Primary #Bean for ObjectMapper, configureHttpMessageCodecs(), etc.). What worked for me at the end was specifying a MIME type. Here's an example:
#Configuration
class WebConfig: WebFluxConfigurer {
override fun configureHttpMessageCodecs(configurer: ServerCodecConfigurer) {
val encoder = Jackson2JsonEncoder(objectMapper, MimeTypeUtils.APPLICATION_JSON)
val decoder = Jackson2JsonDecoder(objectMapper, MimeTypeUtils.APPLICATION_JSON)
configurer.defaultCodecs().jackson2JsonEncoder(encoder)
configurer.defaultCodecs().jackson2JsonDecoder(decoder)
}
}

Related

SpringBoot Rest Controller Supporting both xml and json with JAXB Pojos + Custom Http Message Converters + Swagger-UI

Getting following Swagger error when I introduced custom message converters
Unable to render this definition
The provided definition does not specify a valid version field.
Please indicate a valid Swagger or OpenAPI version field. Supported version fields are swagger: "2.0" and those that match openapi: 3.0.n (for example, openapi: 3.0.0).
The error is misleading, I guess its getting confused with the converters. Any Ideas to get it fixed?
I followed - swagger-ui not working with custom XML ObjectMapper (no luck)
Background:
I have generated pojos from xsd(s) through xjc. And I have a rest endpoint which needs to support both xml and json for request/response
We got it working by following [spring documentation][1] section: 22.16.12 Message Converters
Here is what I added in MyConfig
#Configuration
#EnableWebMvc
public class MyConfig implements WebMvcConfigurer {
#Override
public void configureMessageConverters(List<HttpMessageConverter<?>> converters) {
converters.add(new MappingJackson2HttpMessageConverter(objectMapper()));
converters.add(new MappingJackson2XmlHttpMessageConverter(xmlMapper()));
}
#Bean
#Primary
public ObjectMapper objectMapper() {
return new Jackson2ObjectMapperBuilder()
.modulesToInstall(new JaxbAnnotationModule())
.build();
}
#Bean
public XmlMapper xmlMapper() {
return new Jackson2ObjectMapperBuilder()
.indentOutput(true)
.defaultUseWrapper(false)
.serializationInclusion(JsonInclude.Include.NON_EMPTY)
.modulesToInstall(new JaxbAnnotationModule())
.createXmlMapper(true)
.build();
}
}
and my controller
// all the open api annotations //
#RequestMapping(value = "/run",
produces = {"application/json", "application/xml"},
consumes = {"application/json", "application/xml"},
method = RequestMethod.POST)
public ResponseEntity<MyResponse> run(#RequestBody MyRequest request) {
Ok its about which objectMapper. I had to clean up and remove '#primary' which made the trick in MyConfig
#Bean
public MappingJackson2HttpMessageConverter mappingJackson2HttpMessageConverter() {
return new MappingJackson2HttpMessageConverter(new Jackson2ObjectMapperBuilder()
.modulesToInstall(new JaxbAnnotationModule())
.build());
}
#Bean
public MappingJackson2XmlHttpMessageConverter mappingJackson2XmlHttpMessageConverter() {
return new MappingJackson2XmlHttpMessageConverter(new Jackson2ObjectMapperBuilder()
.indentOutput(true)
.defaultUseWrapper(false)
.serializationInclusion(JsonInclude.Include.NON_EMPTY)
.modulesToInstall(new JaxbAnnotationModule())
.createXmlMapper(true)
.build());
}

Spring Data Rest: #Autowire in Custom JsonDeserializer

I am trying to autowire a component into a custom JsonDeserializer but cannot get it right even with the following suggestions I found:
Autowiring in JsonDeserializer: SpringBeanAutowiringSupport vs HandlerInstantiator
Right way to write JSON deserializer in Spring or extend it
How to customise the Jackson JSON mapper implicitly used by Spring Boot?
Spring Boot Autowiring of JsonDeserializer in Integration test
My final goal is to accept URLs to resources in different microservices and store only the ID of the resource locally. But I don't want to just extract the ID from the URL but also verify that the rest of the URL is correct.
I have tried many things and lost track a bit of what I tried but I believe I tried everything mentioned in the links above. I created tons of beans for SpringHandlerInstantiator, Jackson2ObjectMapperBuilder, MappingJackson2HttpMessageConverter, RestTemplate and others and also tried with setting the SpringHandlerInstantiator in RepositoryRestConfigurer#configureJacksonObjectMapper.
I am using Spring Boot 2.1.6.RELEASE which makes me think something might have changed since some of the linked threads are quite old.
Here's my last attempt:
#Configuration
public class JacksonConfig {
#Bean
public HandlerInstantiator handlerInstantiator(ApplicationContext applicationContext) {
return new SpringHandlerInstantiator(applicationContext.getAutowireCapableBeanFactory());
}
}
#Configuration
public class RestConfiguration implements RepositoryRestConfigurer {
#Autowired
private Validator validator;
#Autowired
private HandlerInstantiator handlerInstantiator;
#Override
public void configureValidatingRepositoryEventListener(ValidatingRepositoryEventListener validatingListener) {
validatingListener.addValidator("beforeCreate", validator);
validatingListener.addValidator("beforeSave", validator);
}
#Override
public void configureJacksonObjectMapper(ObjectMapper objectMapper) {
objectMapper.setHandlerInstantiator(handlerInstantiator);
}
}
#Component
public class RestResourceURLSerializer extends JsonDeserializer<Long> {
#Autowired
private MyConfig config;
#Override
public Long deserialize(JsonParser p, DeserializationContext ctxt) throws IOException, JsonProcessingException {
ServiceConfig serviceConfig = config.getServices().get("identity");
URI serviceUri = serviceConfig.getExternalUrl();
String servicePath = serviceUri.getPath();
URL givenUrl = p.readValueAs(URL.class);
String givenPath = givenUrl.getPath();
if (servicePath.equals(givenPath)) {
return Long.parseLong(givenPath.substring(givenPath.lastIndexOf('/') + 1));
}
return null;
}
}
I keep getting a NullPointerException POSTing something to the API endpoint that is deserialized with the JsonDeserializer above.
I was able to solve a similar problem by marking my deserializer constructor accept a parameter (and therefore removing the empty constructor) and marking constructor as #Autowired.
public class MyDeserializer extends JsonDeserializer<MyEntity> {
private final MyBean bean;
// no default constructor
#Autowired
public MyDeserializer(MyBean bean){
this.bean = bean
}
...
}
#JsonDeserialize(using = MyDeserializer.class)
public class MyEntity{...}
My entity is marked with annotation #JsonDeserialize so I don't have to explicitly register it with ObjectMapper.

Jackson - configure override for collections via Jackson2ObjectMapperBuilderCustomizer

I am customizing treatment of collections in my Jackson's object mapper in my Spring Boot config by constructing a new mapper like so
#Configuration
public class Config {
#Autowired(required = true)
public void objectMapper(ObjectMapper mapper) {
mapper.configOverride(Collection.class).setInclude(JsonInclude.Value.construct(JsonInclude.Include.NON_EMPTY, null));
mapper.configOverride(List.class).setInclude(JsonInclude.Value.construct(JsonInclude.Include.NON_EMPTY, null));
mapper.configOverride(Map.class).setInclude(JsonInclude.Value.construct(JsonInclude.Include.NON_EMPTY, null));
}
While this works, I understand that a more elegant approach is to use Jackson2ObjectMapperBuilderCustomizer :
#Bean
public Jackson2ObjectMapperBuilderCustomizer customizeJackson2ObjectMapper() {
return new Jackson2ObjectMapperBuilderCustomizer() {
#Override
public void customize(Jackson2ObjectMapperBuilder builder) {
builder
.indentOutput(true)
.someOtherMethod(...)
}
};
}
How do I implement ObjectMapper collection tweaks above via Jackson2ObjectMapperBuilder ?
You can use a simple Module defined locally, like in this other use case. The SetupContext also has a configOverride() method, just like the ObjectMapper itself.
No idea ? I'm interested to do the same just to add :
mapper.configOverride(Map.Entry.class).setFormat(forShape(Shape.OBJECT));
Because #JsonFormat(shape = JsonFormat.Shape.OBJECT) doesn't work well ( https://github.com/FasterXML/jackson-databind/issues/1419 ) and but after Jackson 2.5 it is the only solution (but requires 2.9.x) to restore the previous behavior without writing a custom serializer.

Upgrade from Spring Boot 1.3 to Spring Boot 1.4 and Pageable is not working as expected.

I am converting an existing Spring Boot application from 1.3.6 to 1.4.1. I would like to have a default page size for repository and controller responses of 25. I am not getting the expected behavior in either case. For repository methods I am getting a page size of 20. For controllers I am getting 0 for the page size.
I added a new configuration class to define the default page size. I found this code snippet in another article. The debug message does get printed out.
#Configuration
public class RestConfigurationAdapter extends WebMvcConfigurerAdapter {
private static final int DEFAULT_PAGE_SIZE = 25;
#Override
public void addArgumentResolvers(List<HandlerMethodArgumentResolver> argumentResolvers) {
System.out.println("DEBUG: AddArguments----");
PageableHandlerMethodArgumentResolver resolver = new PageableHandlerMethodArgumentResolver();
resolver.setFallbackPageable(new PageRequest(0, DEFAULT_PAGE_SIZE));
argumentResolvers.add(resolver);
super.addArgumentResolvers(argumentResolvers);
}
}
In a custom controller I would like to have a default pageable populated with a size of 25. However the pageable object is null in this controller. In 1.3.x the pageable object worked as expected.
public class BatchManagerController
{
#Autowired
private BatchRepository batchRepository;
#Autowired
private PagedResourcesAssembler pagedResourcesAssembler;
#Transactional(readOnly = true)
#RequestMapping(value = "/search/managerBatchView", method = RequestMethod.GET, produces = MediaType.APPLICATION_JSON_VALUE)
#PreAuthorize("hasRole(T(com.nextgearcapital.tms.api.util.AuthorityEnum).MANAGER)")
public ResponseEntity<?> getManagerBatchListView(BatchListSearchRequest requestDTO, Pageable pageable, PersistentEntityResourceAssembler myAssembler)
{
System.out.println("DEBUG1:---------- " + pageable);
Page<Batch> batchPage = batchRepository.findBatchesForManager(requestDTO, pageable);
PagedResources<VaultResource> pagedResources = pagedResourcesAssembler.toResource(batchPage, myAssembler);
return new ResponseEntity<>(pagedResources, HttpStatus.OK);
}
}
When calling SDR Repository methods with a pageable parameter, the parameter works correctly, but it has a default page size of 20, rather than 25.
I would appreciate any help and advise in getting the correct configuration for pagination.
You probably have 2 solutions
Register the PageableHandlerMethodArgumentResolver as an #Bean which will disable the auto configuration for Spring Data Web.
Create a BeanPostProcessor to do additional configuration on the existing PageableHandlerMethodArgumentResolver.
Using #Bean
#Configuration
public class RestConfigurationAdapter extends WebMvcConfigurerAdapter {
private static final int DEFAULT_PAGE_SIZE = 25;
#Bean
public PageableHandlerMethodArgumentResolver pageableResolver() {
PageableHandlerMethodArgumentResolver resolver = new PageableHandlerMethodArgumentResolver();
resolver.setFallbackPageable(new PageRequest(0, DEFAULT_PAGE_SIZE));
return resolver;
}
#Override
public void addArgumentResolvers(List<HandlerMethodArgumentResolver> argumentResolvers) {
System.out.println("DEBUG: AddArguments----");
argumentResolvers.add(pageableResolver());
}
}
Drawback is that it will disable the autoconfiguration for Spring Data Web, so you might miss some things.
Using a BeanPostProcessor.
#Bean
public BeanPostProcessor pageableProcessor() {
private static final int DEFAULT_PAGE_SIZE = 25;
return new BeanPostProcessor() {
public Object postProcessBeforeInitialization(Object bean, String beanName) throws BeansException {
if (bean instanceof PageableHandlerMethodArgumentResolver) {
((PageableHandlerMethodArgumentResolver) bean).setFallbackPageable(new PageRequest(0, DEFAULT_PAGE_SIZE));
}
return bean;
}
public Object postProcessAfterInitialization(Object bean, String beanName) throws BeansException {
return bean;
}
}
}
Drawback is that it is a little more complex as registering your own PageableHandlerMethodArgumentResolver instance as a bean. Advantage however is that you can simply use this to add additional configuration to existing beans and leave the auto configuration in tact.
Starting in spring-data-commons version 2.0, there is are 2 new classes that might make this kind of thing easier:
SortHandlerMethodArgumentResolverCustomizer
PageableHandlerMethodArgumentResolverCustomizer
Unfortunately that's not the version that ships with the current version (1.5.9) of Spring Boot, so replace at your own risk.
#Bean
PageableHandlerMethodArgumentResolverCustomizer pagingCustomizer() {
// p is PageableHandlerMethodArgumentResolver
return p -> p.setMaxPageSize(25);
}
In this case, one would probably call resolveArgument to manipulate it.
That said, I'm not sure spring-data-rest would use that config. There is a HateoasPageableHandlerMethodArgumentResolver which seems more likely that source of what I would think SDR would use. If that's the case, the BeanPostProcessor #M. Deinum suggested is probably your best option.
Spring Data Web Support

Spring java config: bean config after component scan

I have the following configuration:
#Configuration
#ComponentScan("com.xyz.svc")
public class SvcConfig {
#Autowired private Filter filter1;
#Autowired private Filter filter2;
#Autowired private Filter filter3;
#Bean
public List<Filter> filters() {
// Filters are added in the desired order of execution
return ImmutableList.of(
filter1,
filter2,
filter3);
}
}
When leadFilters() method is run all the components that it depends on (ie filter1, filter2, filter3) are null. Basically, these components are registered through #ComponentScan. The problem is leadFilters() method is getting executed before #ComponentScan.
How do I make this work?
Basically, you can't, reliably. A #Configuration class is a #Component that is meant to register bean definitions through #Bean annotated methods. If a request for a bean (handled through a #Bean method) comes in before the BeanPostProcessor that handles #Autowired, then you will see the behavior you are describing.
Note that the following will cause you problems as Spring won't know which to inject.
#Autowired
private Filter filter1;
#Autowired
private Filter filter2;
#Autowired
private Filter filter3;
Assuming this was just an example, you could refactor so that instead of having #Component classes for these filters, you instead declare #Bean methods for them.
#Bean
public Filter filter1() {
return new FilterImpl1();
}
#Bean
public Filter filter2() {
return new FilterImpl2();
}
#Bean
public Filter filter3() {
return new FilterImpl3();
}
You can then use these beans in your other #Bean method
#Bean
public List<Filter> filters() {
// Filters are added in the desired order of execution
return ImmutableList.of(
filter1(),
filter2(),
filter3());
}

Resources