在spring boot專案中已經包含有json序列化的框架，具體在包com.fasterxml.jackson.annotation中，建議看看詳細原始碼。

但在專案應用上還是會有一些坑會出現的，舉個例子：

在一個複雜的業務模型中包含有200個欄位，在查詢列表時只查詢其中某20個欄位，在查詢詳情中需要把所有欄位都查詢出來。

一般情況下，如果是開始做一個新功能，那麼我們的設計應該類似是這樣的：

model

     ---- QueryModel ，包含20個欄位，響應查詢列表結果

     ---- DetailModel extend POJO ， 包含所有欄位，響應查詢實體結果

entity

     ---- POJO，包含所有欄位

但維護從來都是一件蛋疼的事情，200個欄位是迭代出來的，他們的邏輯是這樣的：

 

entity

     ---- POJO，包含所有欄位，響應查詢列表和查詢實體結果

這時候會發現一切蛋疼的原因就是直接把pojo拿來當model用了，導致所有引數和結果無法拓展。為什麼會這麼蛋疼呢？原因不重要，如何解決才重要。

方案1：程式碼解耦，改造成model和entity分離

此方案的好處是一勞永逸，後續的拓展也比較輕鬆，弊端也顯而易見，會發費許多時間去理解業務重構業務，而且一個穩定的系統一般不會嘗試大範圍的改造，萬一改後出現一堆bug呢？

方案2：把不輸出到頁面的欄位忽略掉

這時候用到註解 @JsonIgnore，該註解既可以作用在欄位上也可以作用在方法上（另外兩種先不說），可以看看原始碼和註釋：

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;
}

因此在使用的時候需要注意，如果只需要在查詢的時候忽略，在儲存的時候不忽略，那麼需要在getter方法上註解@JsonIgnore，在setter方法上註解 @JsonProperty，（這時候如果使用的lombok就很尷尬了），舉個例子：

public class TestEntity{private String name;

    /**
     * getter
     */
    @JsonIgnore
    public String getName() {
        return name;
    }

    /**
     * setter
     */
    @JsonProperty
    public void setName(String name) {
        this.name= name;
    }
}

然後在sql上從select *改成select 指定欄位，解決查詢列表問題。

然而，在查詢單個實體詳情的時候這些被忽略的欄位也無法傳到前端了，這是一條乍一看很理想的死衚衕。

但從sql的角度上可以看出select 指定欄位的時候，在反序列化到pojo的時候其他欄位是沒有值的，那麼可以把方案換一下，讓有值得欄位序列化出去，沒有值的不序列化

方案3：讓有值得欄位序列化出去，沒有值的不序列化

把sql上從select *改成select 指定欄位

在pojo裡用上@JsonInclude(JsonInclude.Include.NON_NULL)，

此時既滿足查詢列表，又滿足查詢詳情，解決問題。