源码分析springboot自定义jackson序列化，默认null值个性化处理返回值

Posted 2021-04-05 JAVA 开发微指南

最近项目要实现一种需求，对于后端返回给前端的json格式的一种规范，不允许缺少字段和字段值都为null，所以琢磨了一下如何进行将springboot的Jackson序列化自定义一下，先看看如何实现，再去看源码

第一步：写配置类

@Configurationpublic class WebConfiguration extends WebMvcConfigurationSupport {@Override protected void extendMessageConverters(List<HttpMessageConverter<?>> converters) { converters.stream().filter(c -> c instanceof MappingJackson2HttpMessageConverter) .map(c ->(MappingJackson2HttpMessageConverter)c) .forEach(c->{ ObjectMapper mapper = c.getObjectMapper(); // 为mapper注册一个带有SerializerModifier的Factory，此modifier主要做的事情为：当序列化类型为array，list、set时，当值为空时，序列化成[] mapper.setSerializerFactory(mapper.getSerializerFactory().withSerializerModifier(new MyBeanSerializerModifier())); c.setObjectMapper(mapper); }); }}

第二步：编写值为null时的自定义序列化

/** * @title: MyBeanSerializerModifier * @Author junyu * 旧巷里有一个穿着白衬衫笑起来如太阳般温暖我的少年。 * 记忆里有一个穿着连衣裙哭起来如孩子般讨人喜的女孩。 * 他说，哪年树弯了腰，人见了老，桃花落了白发梢，他讲的笑话她还会笑，那便是好。 * 她说，哪年国改了号，坟长了草，地府过了奈何桥，她回头看时他还在瞧，就不算糟。 * @Date: 2020/9/12 16:44 * @Version 1.0 */public class MyBeanSerializerModifier extends BeanSerializerModifier {
 private MyNullStringJsonSerializer myNullStringJsonSerializer; private MyNullArrayJsonSerializer MyNullArrayJsonSerializer; private MyNullObjectJsonSerializer MyNullObjectJsonSerializer; private MyNullJsonSerializer myNullJsonSerializer;
 public MyBeanSerializerModifier(){ myNullStringJsonSerializer = new MyNullStringJsonSerializer(); MyNullArrayJsonSerializer = new MyNullArrayJsonSerializer(); MyNullObjectJsonSerializer = new MyNullObjectJsonSerializer(); myNullJsonSerializer = new MyNullJsonSerializer(); }
 @Override public List<BeanPropertyWriter> changeProperties(SerializationConfig config, BeanDescription beanDesc, List<BeanPropertyWriter> beanProperties) { // 循环所有的beanPropertyWriter beanProperties.forEach(writer ->{ // 判断字段的类型 if (isArrayType(writer)) { //给writer注册一个自己的nullSerializer writer.assignNullSerializer(MyNullArrayJsonSerializer); } else if (isObjectType(writer)) { writer.assignNullSerializer(MyNullObjectJsonSerializer); } else if (isStringType(writer)) { writer.assignNullSerializer(myNullStringJsonSerializer); } else if (isPrimitiveType(writer)) { writer.assignNullSerializer(myNullJsonSerializer); } }); return beanProperties; }
 // 判断是否是boolean类型 private boolean isPrimitiveType(BeanPropertyWriter writer) { Class<?> clazz = writer.getType().getRawClass(); return clazz.isPrimitive(); }
 // 判断是否是string类型 private boolean isStringType(BeanPropertyWriter writer) { Class<?> clazz = writer.getType().getRawClass(); return clazz.equals(String.class); }
 // 判断是否是对象类型 private boolean isObjectType(BeanPropertyWriter writer) { Class<?> clazz = writer.getType().getRawClass(); return !clazz.isPrimitive() && !clazz.equals(String.class) && clazz.isAssignableFrom(Object.class); } // 判断是否是集合类型 protected boolean isArrayType(BeanPropertyWriter writer) { Class<?> clazz = writer.getType().getRawClass(); return clazz.isArray() || clazz.equals(List.class) || clazz.equals(Set.class); }
 class MyNullJsonSerializer extends JsonSerializer<Object>{
 @Override public void serialize(Object value, JsonGenerator jgen, SerializerProvider serializers) throws IOException { if (value == null) { jgen.writeNull(); } } }

 class MyNullStringJsonSerializer extends JsonSerializer<Object>{
 @Override public void serialize(Object value, JsonGenerator jgen, SerializerProvider serializers) throws IOException { if (value == null) { jgen.writeString(StringUtils.EMPTY); } } }
 class MyNullArrayJsonSerializer extends JsonSerializer<Object>{
 @Override public void serialize(Object value, JsonGenerator jgen, SerializerProvider serializers) throws IOException { if (value == null) { jgen.writeStartArray(); jgen.writeEndArray(); } } }
 class MyNullObjectJsonSerializer extends JsonSerializer<Object>{
 @Override public void serialize(Object value, JsonGenerator jgen, SerializerProvider serializers) throws IOException { if (value == null) { jgen.writeStartObject(); jgen.writeEndObject(); } } }
}

这样基本配置就完事了，现在可以试试效果了，自己定义一个bean用来返回，定义一个简单的controller去接受访问就行了，博主就不进行写这两个类了。返回结果如下

这是我的项目需求需要实现的，大家可以根据的自己的需求去改写MyBeanSerializerModifier这个类。还有另一种实现方式：不继承

@Configurationpublic class WebConfiguration {
@Bean public MappingJackson2HttpMessageConverter mappingJackson2HttpMessageConverter(){ MappingJackson2HttpMessageConverter mappingJackson2HttpMessageConverter = new MappingJackson2HttpMessageConverter(); ObjectMapper mapper = mappingJackson2HttpMessageConverter.getObjectMapper(); mapper.setSerializerFactory(mapper.getSerializerFactory().withSerializerModifier(new MyBeanSerializerModifier())); mappingJackson2HttpMessageConverter.setObjectMapper(mapper); return mappingJackson2HttpMessageConverter; }
}

这种方法也是可以设置成功的，主要是不是继承了WebMvcConfigurationSupport类，毕竟这个类有很多可以自定义的方法，用起来顺手而已。

　　第一个问题：为什么继承WebMvcConfigurationSupport后，要重写extendMessageConverters方法；

　　第二个问题：为什么继承WebMvcConfigurationSupport后，再去生成@Bean的MappingJackson2HttpMessageConverter，却不生效；

　　第三个问题：为什么不继承WebMvcConfigurationSupport时，生成@Bean的MappingJackson2HttpMessageConverter是生效的；

　　这几个问题，都需要我们进入源码观察，废活不多说，我们来进入源码的世界。解决问题之前必须搞清楚在哪里进行了序列化。

　　第一步：我们要弄清楚在哪里进行的Jackson序列化，看这里https://www.processon.com/embed/5f5c6464f346fb7afd55448b，从返回请求开始的序列化基本流程就在这里了，虽然图有点low，但是清楚的记录的每一步，我们主要看一下下面的源码

/* /********************************************************** /* Field serialization methods /********************************************************** */ //序列化每一个字段 protected void serializeFields(Object bean, JsonGenerator gen, SerializerProvider provider) throws IOException{ final BeanPropertyWriter[] props; if (_filteredProps != null && provider.getActiveView() != null) { props = _filteredProps; } else { props = _props; } int i = 0; try { for (final int len = props.length; i < len; ++i) { BeanPropertyWriter prop = props[i]; if (prop != null) { // can have nulls in filtered list //关键就在这一步进行的序列化，而为什么BeanPropertyWriter是数组，我们一会解释 prop.serializeAsField(bean, gen, provider); } } if (_anyGetterWriter != null) { _anyGetterWriter.getAndSerialize(bean, gen, provider); } } catch (Exception e) { String name = (i == props.length) ? "[anySetter]" : props[i].getName(); wrapAndThrow(provider, e, bean, name); } catch (StackOverflowError e) { // 04-Sep-2009, tatu: Dealing with this is tricky, since we don't have many // stack frames to spare... just one or two; can't make many calls.
 // 10-Dec-2015, tatu: and due to above, avoid "from" method, call ctor directly: //JsonMappingException mapE = JsonMappingException.from(gen, "Infinite recursion (StackOverflowError)", e); JsonMappingException mapE = new JsonMappingException(gen, "Infinite recursion (StackOverflowError)", e);
 String name = (i == props.length) ? "[anySetter]" : props[i].getName(); mapE.prependPath(new JsonMappingException.Reference(bean, name)); throw mapE; } }

既然已经找到了在哪里要进行序列化，那我们看看是如何实现的：

/** * Method called to access property that this bean stands for, from within * given bean, and to serialize it as a JSON Object field using appropriate * serializer. */ @Override public void serializeAsField(Object bean, JsonGenerator gen, SerializerProvider prov) throws Exception { // inlined 'get()' final Object value = (_accessorMethod == null) ? _field.get(bean) : _accessorMethod.invoke(bean, (Object[]) null);
 // Null handling is bit different, check that first if (value == null) { //看到这里大家应该就知道null值是如何进行序列化 的了，如果不配置的话，默认是返回null //因为_nullSerializer是有默认值的，大家看一看这个类的初始化 //那我们要是改一下_nullSerializer的这个默认类，让每一个字段调用我们自己的_nullSerializer不就可以了吗， //yes、我们就这么干 if (_nullSerializer != null) { gen.writeFieldName(_name); _nullSerializer.serialize(null, gen, prov); } return; } // then find serializer to use JsonSerializer<Object> ser = _serializer; if (ser == null) { Class<?> cls = value.getClass(); PropertySerializerMap m = _dynamicSerializers; ser = m.serializerFor(cls); if (ser == null) { ser = _findAndAddDynamic(m, cls, prov); } } // and then see if we must suppress certain values (default, empty) if (_suppressableValue != null) { if (MARKER_FOR_EMPTY == _suppressableValue) { if (ser.isEmpty(prov, value)) { return; } } else if (_suppressableValue.equals(value)) { return; } } // For non-nulls: simple check for direct cycles if (value == bean) { // three choices: exception; handled by call; or pass-through if (_handleSelfReference(bean, gen, prov, ser)) { return; } } gen.writeFieldName(_name); if (_typeSerializer == null) { ser.serialize(value, gen, prov); } else { ser.serializeWithType(value, gen, prov, _typeSerializer); } }

那我们来解决第一个问题：为什么继承WebMvcConfigurationSupport后，要重写extendMessageConverters方法？

　　不知道大家记得不记得我们请求过来的时候，如果我们配置类集成了WebMvcConfigurationSupport类，dispatchservlet处理handle请求的ha，其实就是RequestMappingHandlerAdapter类，这个类是在WebMvcConfigurationSupport配置的，看源码：

@Bean public RequestMappingHandlerAdapter requestMappingHandlerAdapter( @Qualifier("mvcContentNegotiationManager") ContentNegotiationManager contentNegotiationManager, @Qualifier("mvcConversionService") FormattingConversionService conversionService, @Qualifier("mvcValidator") Validator validator) {
 RequestMappingHandlerAdapter adapter = createRequestMappingHandlerAdapter(); adapter.setContentNegotiationManager(contentNegotiationManager); adapter.setMessageConverters(getMessageConverters()); adapter.setWebBindingInitializer(getConfigurableWebBindingInitializer(conversionService, validator)); adapter.setCustomArgumentResolvers(getArgumentResolvers()); adapter.setCustomReturnValueHandlers(getReturnValueHandlers());
 if (jackson2Present) { adapter.setRequestBodyAdvice(Collections.singletonList(new JsonViewRequestBodyAdvice())); adapter.setResponseBodyAdvice(Collections.singletonList(new JsonViewResponseBodyAdvice())); }
 AsyncSupportConfigurer configurer = new AsyncSupportConfigurer(); configureAsyncSupport(configurer); if (configurer.getTaskExecutor() != null) { adapter.setTaskExecutor(configurer.getTaskExecutor()); } if (configurer.getTimeout() != null) { adapter.setAsyncRequestTimeout(configurer.getTimeout()); } adapter.setCallableInterceptors(configurer.getCallableInterceptors()); adapter.setDeferredResultInterceptors(configurer.getDeferredResultInterceptors());
 return adapter; }

adapter.setMessageConverters(getMessageConverters());当大家看到这个方法的时候，应该就会想到我们的默认jackson转换器：MappingJackson2HttpMessageConverter,我们看看这个getMessageConverters()有什么幺蛾子：

protected final List<HttpMessageConverter<?>> getMessageConverters() { if (this.messageConverters == null) { this.messageConverters = new ArrayList<>(); configureMessageConverters(this.messageConverters); if (this.messageConverters.isEmpty()) { addDefaultHttpMessageConverters(this.messageConverters); } extendMessageConverters(this.messageConverters); } return this.messageConverters; }

protected final void addDefaultHttpMessageConverters(List<HttpMessageConverter<?>> messageConverters) {
 //这些都不用管，跟我们的需求没啥作用，我们只看关键的部分，在下面 messageConverters.add(new ByteArrayHttpMessageConverter()); messageConverters.add(new StringHttpMessageConverter()); messageConverters.add(new ResourceHttpMessageConverter()); messageConverters.add(new ResourceRegionHttpMessageConverter()); try { messageConverters.add(new SourceHttpMessageConverter<>()); } catch (Throwable ex) { // Ignore when no TransformerFactory implementation is available... } messageConverters.add(new AllEncompassingFormHttpMessageConverter());
 if (romePresent) { messageConverters.add(new AtomFeedHttpMessageConverter()); messageConverters.add(new RssChannelHttpMessageConverter()); }
 if (jackson2XmlPresent) { Jackson2ObjectMapperBuilder builder = Jackson2ObjectMapperBuilder.xml(); if (this.applicationContext != null) { builder.applicationContext(this.applicationContext); } messageConverters.add(new MappingJackson2XmlHttpMessageConverter(builder.build())); } else if (jaxb2Present) { messageConverters.add(new Jaxb2RootElementHttpMessageConverter()); }
 if (jackson2Present) { Jackson2ObjectMapperBuilder builder = Jackson2ObjectMapperBuilder.json(); if (this.applicationContext != null) { builder.applicationContext(this.applicationContext); } //解析我们返回值的转换器就是在这里生成的 messageConverters.add(new MappingJackson2HttpMessageConverter(builder.build())); } else if (gsonPresent) { messageConverters.add(new GsonHttpMessageConverter()); } else if (jsonbPresent) { messageConverters.add(new JsonbHttpMessageConverter()); }
 if (jackson2SmilePresent) { Jackson2ObjectMapperBuilder builder = Jackson2ObjectMapperBuilder.smile(); if (this.applicationContext != null) { builder.applicationContext(this.applicationContext); } messageConverters.add(new MappingJackson2SmileHttpMessageConverter(builder.build())); } if (jackson2CborPresent) { Jackson2ObjectMapperBuilder builder = Jackson2ObjectMapperBuilder.cbor(); if (this.applicationContext != null) { builder.applicationContext(this.applicationContext); } messageConverters.add(new MappingJackson2CborHttpMessageConverter(builder.build())); } }

我们的MappingJackson2HttpMessageConverter类就是这里初始化的，初始化的时候默认的_nullSerializer也会被初始化，大家肯定说这已经初始化完了，该咋办，大家应该看到了extendMessageConverters(this.messageConverters);这个方法就是用来重写实现的了，这回知道我们继承WebMvcConfigurationSupport后，为什么要重写extendMessageConverters，我们的配置类遍历已经获取到的convert，然后对我们想要的转换器进行修改添加，那修改完了，是在哪里起作用的呢，我们再来看一看源码：

在序列化之前有一些方法是可以进行修改操作的，在调用writeWithMessageConverters方法的时候：

protected <T> void writeWithMessageConverters(@Nullable T value, MethodParameter returnType, ServletServerHttpRequest inputMessage, ServletServerHttpResponse outputMessage) throws IOException, HttpMediaTypeNotAcceptableException, HttpMessageNotWritableException {
 .......
 MediaType selectedMediaType = null; MediaType contentType = outputMessage.getHeaders().getContentType(); boolean isContentTypePreset = contentType != null && contentType.isConcrete(); if (isContentTypePreset) { if (logger.isDebugEnabled()) { logger.debug("Found 'Content-Type:" + contentType + "' in response"); } selectedMediaType = contentType; } else { HttpServletRequest request = inputMessage.getServletRequest(); List<MediaType> acceptableTypes = getAcceptableMediaTypes(request); //这里进行自定义操作修改MappingJackson2HttpMessageConverter List<MediaType> producibleTypes = getProducibleMediaTypes(request, valueType, targetType);
 .......
 if (selectedMediaType != null) { selectedMediaType = selectedMediaType.removeQualityValue(); //这这里进行选择我们的MappingJackson2HttpMessageConverter去自定义序列化 for (HttpMessageConverter<?> converter : this.messageConverters) { GenericHttpMessageConverter genericConverter = (converter instanceof GenericHttpMessageConverter ? (GenericHttpMessageConverter<?>) converter : null); if (genericConverter != null ? ((GenericHttpMessageConverter) converter).canWrite(targetType, valueType, selectedMediaType) : converter.canWrite(valueType, selectedMediaType)) { body = getAdvice().beforeBodyWrite(body, returnType, selectedMediaType, (Class<? extends HttpMessageConverter<?>>) converter.getClass(), inputMessage, outputMessage); if (body != null) { Object theBody = body; LogFormatUtils.traceDebug(logger, traceOn -> "Writing [" + LogFormatUtils.formatValue(theBody, !traceOn) + "]"); addContentDispositionHeader(inputMessage, outputMessage); if (genericConverter != null) { genericConverter.write(body, targetType, selectedMediaType, outputMessage); } else { ((HttpMessageConverter) converter).write(body, selectedMediaType, outputMessage); } } else { if (logger.isDebugEnabled()) { logger.debug("Nothing to write: null body"); } } return; } } }
 if (body != null) { Set<MediaType> producibleMediaTypes = (Set<MediaType>) inputMessage.getServletRequest() .getAttribute(HandlerMapping.PRODUCIBLE_MEDIA_TYPES_ATTRIBUTE);
 if (isContentTypePreset || !CollectionUtils.isEmpty(producibleMediaTypes)) { throw new HttpMessageNotWritableException( "No converter for [" + valueType + "] with preset Content-Type '" + contentType + "'"); } throw new HttpMediaTypeNotAcceptableException(this.allSupportedMediaTypes); } }

我们一直追踪getProducibleMediaTypes后，最终发现会调用BeanSerializerFactory的constructBeanOrAddOnSerializer，就是这里进行修改操作的。

protected JsonSerializer<Object> constructBeanOrAddOnSerializer(SerializerProvider prov, JavaType type, BeanDescription beanDesc, boolean staticTyping) throws JsonMappingException { // 13-Oct-2010, tatu: quick sanity check: never try to create bean serializer for plain Object // 05-Jul-2012, tatu: ... but we should be able to just return "unknown type" serializer, right? if (beanDesc.getBeanClass() == Object.class) { return prov.getUnknownTypeSerializer(Object.class);// throw new IllegalArgumentException("Cannot create bean serializer for Object.class"); } final SerializationConfig config = prov.getConfig(); BeanSerializerBuilder builder = constructBeanSerializerBuilder(beanDesc); builder.setConfig(config);
 // First: any detectable (auto-detect, annotations) properties to serialize? List<BeanPropertyWriter> props = findBeanProperties(prov, beanDesc, builder); if (props == null) { props = new ArrayList<BeanPropertyWriter>(); } else { props = removeOverlappingTypeIds(prov, beanDesc, builder, props); }
 // [databind#638]: Allow injection of "virtual" properties: prov.getAnnotationIntrospector().findAndAddVirtualProperties(config, beanDesc.getClassInfo(), props);
 // [JACKSON-440] Need to allow modification bean properties to serialize: if (_factoryConfig.hasSerializerModifiers()) { for (BeanSerializerModifier mod : _factoryConfig.serializerModifiers()) { props = mod.changeProperties(config, beanDesc, props); } }.......//此处省略}

大家看一下props = mod.changeProperties(config, beanDesc, props);我们在配置类里面可是为我们的MappingJackson2HttpMessageConverter配置了 withSerializerModifier方法，并且设置了我们的MyBeanSerializerModifier并且继承BeanSerializerModifier重写了 changeProperties，所以会调用我们的changeProperties方法，进行修改null值的序列化类，我们也返回了一个list类型的BeanPropertyWriter，所以知道为什么那个BeanPropertyWriter在解析时，会是个数组类型的了吧，因为不同字段解析是不一样的。

剩下的解释一下为什么单独配置并设置实例化@bean的MappingJackson2HttpMessageConverter也是好使的呢，大家可以看一下JacksonHttpMessageConvertersConfiguration类的源码，里面明确写了@ConditionalOnMissingBean注解，如果我们自己进行配置了，这个注入后就是一个备胎，以我们的为准，这个不多说

　　我们再来解析一下第二个问题：为什么继承WebMvcConfigurationSupport后，再去生成@Bean的MappingJackson2HttpMessageConverter，却不生效，这需要跟第三个问题一起解决：为什么不继承WebMvcConfigurationSupport时，生成@Bean的MappingJackson2HttpMessageConverter是生效的。

　　我们知道当我们继承WebMvcConfigurationSupport后，有一个配置会自动失效，就是自动注入的一个mvc配置，可以看看@SpringBootApplication注解里面有个@EnableAutoConfiguration注解，会引入一个AutoConfigurationImportSelector类。这个类就会扫描org.springframework.boot:spring-boot-autoconfigure下的spring.factories文件，这里面有一个我们默认的mvn配置也是继承了WebMvcConfigurationSupport，叫WebMvcAutoConfiguration，我们来看一下源码：

@Configuration(proxyBeanMethods = false)@ConditionalOnWebApplication(type = Type.SERVLET)@ConditionalOnClass({ Servlet.class, DispatcherServlet.class, WebMvcConfigurer.class })//注意此处有一个ConditionalOnMissingBean注解，所以如果我们自己继承后，就相当于已经存在WebMvcConfigurationSupport类，//就会走我们自己的配置类，此配置会失效@ConditionalOnMissingBean(WebMvcConfigurationSupport.class)@AutoConfigureOrder(Ordered.HIGHEST_PRECEDENCE + 10)@AutoConfigureAfter({ DispatcherServletAutoConfiguration.class, TaskExecutionAutoConfiguration.class, ValidationAutoConfiguration.class })public class WebMvcAutoConfiguration { .....
 @Configuration(proxyBeanMethods = false) public static class EnableWebMvcConfiguration extends DelegatingWebMvcConfiguration implements ResourceLoaderAware {
 private final ResourceProperties resourceProperties;
 private final WebMvcProperties mvcProperties;
 private final ListableBeanFactory beanFactory;
 private final WebMvcRegistrations mvcRegistrations;
 private ResourceLoader resourceLoader;
 public EnableWebMvcConfiguration(ResourceProperties resourceProperties, ObjectProvider<WebMvcProperties> mvcPropertiesProvider, ObjectProvider<WebMvcRegistrations> mvcRegistrationsProvider, ListableBeanFactory beanFactory) { this.resourceProperties = resourceProperties; this.mvcProperties = mvcPropertiesProvider.getIfAvailable(); this.mvcRegistrations = mvcRegistrationsProvider.getIfUnique(); this.beanFactory = beanFactory; } //如果我们不继承的话，处理请求的RequestMappingHandlerAdapter就会在这里生成 //会调用DelegatingWebMvcConfiguration里面的 requestMappingHandlerAdapter方法， @Bean @Override public RequestMappingHandlerAdapter requestMappingHandlerAdapter( @Qualifier("mvcContentNegotiationManager") ContentNegotiationManager contentNegotiationManager, @Qualifier("mvcConversionService") FormattingConversionService conversionService, @Qualifier("mvcValidator") Validator validator) { RequestMappingHandlerAdapter adapter = super.requestMappingHandlerAdapter(contentNegotiationManager, conversionService, validator); adapter.setIgnoreDefaultModelOnRedirect( this.mvcProperties == null || this.mvcProperties.isIgnoreDefaultModelOnRedirect()); return adapter; }
.....}

不知道大家是否还记得getMessageConverters()方法里面添加默认messageConverters的时候之前，会调用一个configureMessageConverters(this.messageConverters);方法，我们的DelegatingWebMvcConfiguration 就已经重写了这个方法，所以我们如果不继承WebMvcConfigurationSupport，将会把我们的@bean形式存在的MappingJackson2HttpMessageConverter将会被扫描到

@Override protected void configureMessageConverters(List<HttpMessageConverter<?>> converters) { this.configurers.configureMessageConverters(converters); }
//会添加我们的convert@Override public void configureMessageConverters(List<HttpMessageConverter<?>> converters) { for (WebMvcConfigurer delegate : this.delegates) { delegate.configureMessageConverters(converters); } }

现在我们配置的自定义jackson序列化已经生效了，但是，你仔细看我的流程图会发现，其实调用序列化的时候走的是RequestResponseBodyMethodProcessor的handleReturnValue方法

@Override public void handleReturnValue(@Nullable Object returnValue, MethodParameter returnType, ModelAndViewContainer mavContainer, NativeWebRequest webRequest) throws IOException, HttpMediaTypeNotAcceptableException, HttpMessageNotWritableException {
 mavContainer.setRequestHandled(true); ServletServerHttpRequest inputMessage = createInputMessage(webRequest); ServletServerHttpResponse outputMessage = createOutputMessage(webRequest);
 // Try even with null return value. ResponseBodyAdvice could get involved. //这里进入序列化流程 writeWithMessageConverters(returnValue, returnType, inputMessage, outputMessage); }

最后在序列化的时候也是从这个类或则父类里面的一个属性：messageConverters

public abstract class AbstractMessageConverterMethodArgumentResolver implements HandlerMethodArgumentResolver {
 private static final Set<HttpMethod> SUPPORTED_METHODS = EnumSet.of(HttpMethod.POST, HttpMethod.PUT, HttpMethod.PATCH);
 private static final Object NO_VALUE = new Object();

 protected final Log logger = LogFactory.getLog(getClass()); //这个属性取值的 protected final List<HttpMessageConverter<?>> messageConverters;
 protected final List<MediaType> allSupportedMediaTypes;
 private final RequestResponseBodyAdviceChain advice;
...}

于是，小伙伴们就疑惑了，这我们自定义的在RequestMappingHandlerAdapter里面呢，跟这个类也没关系啊，属性是咋设置进来的呢？我们再看看RequestMappingHandlerAdapter的源码，你会发现，RequestMappingHandlerAdapter这个类实现了InitializingBean类，也就说明，创建RequestMappingHandlerAdapter的时候会调用afterPropertiesSet方法，至于为啥，看源码吧：（不是主要流程）

 //在createBean的时候会调用这个方法，看看是否实现了InitializingBean  protected void invokeInitMethods(String beanName, final Object bean, @Nullable RootBeanDefinition mbd)              throws Throwable {            boolean isInitializingBean = (bean instanceof InitializingBean);          if (isInitializingBean && (mbd == null || !mbd.isExternallyManagedInitMethod("afterPropertiesSet"))) {              if (logger.isTraceEnabled()) {                  logger.trace("Invoking afterPropertiesSet() on bean with name '" + beanName + "'");              }             if (System.getSecurityManager() != null) {                 try {                     AccessController.doPrivileged((PrivilegedExceptionAction<Object>) () -> {                         ((InitializingBean) bean).afterPropertiesSet();                         return null;                     }, getAccessControlContext());                 }                 catch (PrivilegedActionException pae) {                     throw pae.getException();                 }             }             else {                 //在这里进行调用的，                 ((InitializingBean) bean).afterPropertiesSet();             }         }          if (mbd != null && bean.getClass() != NullBean.class) {             String initMethodName = mbd.getInitMethodName();             if (StringUtils.hasLength(initMethodName) &&                     !(isInitializingBean && "afterPropertiesSet".equals(initMethodName)) &&                     !mbd.isExternallyManagedInitMethod(initMethodName)) {                 invokeCustomInitMethod(beanName, bean, mbd);             } }     }

那我们看看RequestMappingHandlerAdapter的afterPropertiesSet方法都干了些啥吧。

@Override public void afterPropertiesSet() { // Do this first, it may add ResponseBody advice beans initControllerAdviceCache();
 if (this.argumentResolvers == null) { List<HandlerMethodArgumentResolver> resolvers = getDefaultArgumentResolvers(); this.argumentResolvers = new HandlerMethodArgumentResolverComposite().addResolvers(resolvers); } if (this.initBinderArgumentResolvers == null) { List<HandlerMethodArgumentResolver> resolvers = getDefaultInitBinderArgumentResolvers(); this.initBinderArgumentResolvers = new HandlerMethodArgumentResolverComposite().addResolvers(resolvers); } if (this.returnValueHandlers == null) { //是在这里生成的类 List<HandlerMethodReturnValueHandler> handlers = getDefaultReturnValueHandlers(); this.returnValueHandlers = new HandlerMethodReturnValueHandlerComposite().addHandlers(handlers); } }

private List<HandlerMethodReturnValueHandler> getDefaultReturnValueHandlers() { List<HandlerMethodReturnValueHandler> handlers = new ArrayList<>();
 // Single-purpose return value types handlers.add(new ModelAndViewMethodReturnValueHandler()); handlers.add(new ModelMethodProcessor()); handlers.add(new ViewMethodReturnValueHandler()); handlers.add(new ResponseBodyEmitterReturnValueHandler(getMessageConverters(), this.reactiveAdapterRegistry, this.taskExecutor, this.contentNegotiationManager)); handlers.add(new StreamingResponseBodyReturnValueHandler()); handlers.add(new HttpEntityMethodProcessor(getMessageConverters(), this.contentNegotiationManager, this.requestResponseBodyAdvice)); handlers.add(new HttpHeadersReturnValueHandler()); handlers.add(new CallableMethodReturnValueHandler()); handlers.add(new DeferredResultMethodReturnValueHandler()); handlers.add(new AsyncTaskMethodReturnValueHandler(this.beanFactory));
 // Annotation-based return value types handlers.add(new ModelAttributeMethodProcessor(false)); //看到这个类了吗？生成的时候将RequestMappingHandlerAdapter里面的转换器设置进去了 handlers.add(new RequestResponseBodyMethodProcessor(getMessageConverters(), this.contentNegotiationManager, this.requestResponseBodyAdvice));
 // Multi-purpose return value types handlers.add(new ViewNameMethodReturnValueHandler()); handlers.add(new MapMethodProcessor());
 // Custom return value types if (getCustomReturnValueHandlers() != null) { handlers.addAll(getCustomReturnValueHandlers()); }
 // Catch-all if (!CollectionUtils.isEmpty(getModelAndViewResolvers())) { handlers.add(new ModelAndViewResolverMethodReturnValueHandler(getModelAndViewResolvers())); } else { handlers.add(new ModelAttributeMethodProcessor(true)); }
 return handlers; }

讲到这里，不知道大家理解了多少，这些都是博主遇到需求后，自己问自己的问题，自己通过源码回答问题的，也希望大家能理解源码。还有一篇源码文章在路上：为什么我们的项目里出现两个配置类继承WebMvcConfigurationSupport时，只有一个会生效。我在网上找了半天都是说结果的，没有人分析源码到底是为啥，博主准备讲解一下，到希望可以帮 大家 ！

JAVA 开发微指南 发起了一个读者讨论 为什么我们的项目里出现两个配置类继承WebMvcConfigurationSupport时，只有一个会生效