在spring boot项目中已经包含有json序列化的框架,具体在包com.fasterxml.jackson.annotation中,建议看看详细源码。
但在项目应用上还是会有一些坑会出现的,举个例子:
在一个复杂的业务模型中包含有200个字段,在查询列表时只查询其中某20个字段,在查询详情中需要把所有字段都查询出来。
一般情况下,如果是开始做一个新功能,那么我们的设计应该类似是这样的:
model
---- QueryModel ,包含20个字段,响应查询列表结果
---- DetailModel extend POJO , 包含所有字段,响应查询实体结果
entity
---- POJO,包含所有字段
但维护从来都是一件蛋疼的事情,200个字段是迭代出来的,他们的逻辑是这样的:
entity
---- POJO,包含所有字段,响应查询列表和查询实体结果
这时候会发现一切蛋疼的原因就是直接把pojo拿来当model用了,导致所有参数和结果无法拓展。为什么会这么蛋疼呢?原因不重要,如何解决才重要。
方案1:代码解耦,改造成model和entity分离
此方案的好处是一劳永逸,后续的拓展也比较轻松,弊端也显而易见,会发费许多时间去理解业务重构业务,而且一个稳定的系统一般不会尝试大范围的改造,万一改后出现一堆bug呢?
方案2:把不输出到页面的字段忽略掉
这时候用到注解 @JsonIgnore,该注解既可以作用在字段上也可以作用在方法上(另外两种先不说),可以看看源码和注释:
package com.fasterxml.jackson.annotation; import java.lang.annotation.ElementType; import java.lang.annotation.Retention; import java.lang.annotation.RetentionPolicy; import java.lang.annotation.Target; /** * Marker annotation that indicates that the logical property that * the accessor (field, getter/setter method or Creator parameter * [of {@link JsonCreator}-annotated constructor or factory method]) * is to be ignored by introspection-based * serialization and deserialization functionality. *<p> * Annotation only needs to be added to one of the accessors (often * getter method, but may be setter, field or creator parameter), * if the complete removal of the property is desired. * However: if only particular accessor is to be ignored (for example, * when ignoring one of potentially conflicting setter methods), * this can be done by annotating other not-to-be-ignored accessors * with {@link JsonProperty} (or its equivalents). This is considered * so-called "split property" case and allows definitions of * "read-only" (read from input into POJO) and "write-only" (write * in output but ignore on output) *<br> * NOTE! As Jackson 2.6, there is a new and improved way to define * `read-only` and `write-only` properties, using * {@link JsonProperty#access()} annotation: this is recommended over * use of separate <code>JsonIgnore</code> and {@link JsonProperty} * annotations. *<p> * For example, a "getter" method that would otherwise denote * a property (like, say, "getValue" to suggest property "value") * to serialize, would be ignored and no such property would * be output unless another annotation defines alternative method to use. *<p> * When ignoring the whole property, the default behavior if encountering * such property in input is to ignore it without exception; but if there * is a {@link JsonAnySetter} it will be called instead. Either way, * no exception will be thrown. *<p> * Annotation is usually used just a like a marker annotation, that * is, without explicitly defining ‘value‘ argument (which defaults * to <code>true</code>): but argument can be explicitly defined. * This can be done to override an existing `JsonIgnore` by explicitly * defining one with ‘false‘ argument: either in a sub-class, or by * using "mix-in annotations". */ @Target({ElementType.ANNOTATION_TYPE, ElementType.METHOD, ElementType.CONSTRUCTOR, ElementType.FIELD}) @Retention(RetentionPolicy.RUNTIME) @JacksonAnnotation public @interface JsonIgnore { /** * Optional argument that defines whether this annotation is active * or not. The only use for value ‘false‘ if for overriding purposes * (which is not needed often); most likely it is needed for use * with "mix-in annotations" (aka "annotation overrides"). * For most cases, however, default value of "true" is just fine * and should be omitted. */ boolean value() default true; }
因此在使用的时候需要注意,如果只需要在查询的时候忽略,在保存的时候不忽略,那么需要在getter方法上注解@JsonIgnore,在setter方法上注解 @JsonProperty,(这时候如果使用的lombok就很尴尬了),举个例子:
public class TestEntity{private String name; /** * getter */ @JsonIgnore public String getName() { return name; } /** * setter */ @JsonProperty public void setName(String name) { this.name= name; } }
然后在sql上从select *改成select 指定字段,解决查询列表问题。
然而,在查询单个实体详情的时候这些被忽略的字段也无法传到前端了,这是一条乍一看很理想的死胡同。
但从sql的角度上可以看出select 指定字段的时候,在反序列化到pojo的时候其他字段是没有值的,那么可以把方案换一下,让有值得字段序列化出去,没有值的不序列化
方案3:让有值得字段序列化出去,没有值的不序列化
把sql上从select *改成select 指定字段
在pojo里用上@JsonInclude(JsonInclude.Include.NON_NULL),
此时既满足查询列表,又满足查询详情,解决问题。
原文地址:https://www.cnblogs.com/zqyx/p/10221244.html