使用 Jackson 将 JSON 数组解析为 Java 类

Question

我有一个相当复杂的 JSON 文件，我需要将其解析为多个 Java 类。

JSON的结构如下：

{
  "dataBlock": [
    {
      "headingName": {
        "name": "Operational Information",
        "position": "1",
        "attributes": [
          {
            "name": "Brand",
            "position": "1",
            "value": [
              "A",
              "B"
            ]
          },
          {
            "name": "Data Model Id",
            "position": "2",
            "value": "000001"
          }
        ]
      }
    },
    {
      "headingName": {
        "name": "CRA",
        "position": "6",
        "attributes": [
          {
            "name": "Company",
            "position": "1",
            "value": "private_limited_company"
          },
          {
            "name": "Address",
            "position": "3",
            "value": {
              "line1": "AAA",
              "line2": "BBB",
              "line3": "CCC",
              "line4": "DDD",
              "postalCode": "AB XYZ",
              "countryCode": "GBR"
            }
          }
        ]
      }
    }
  ]
}

我希望将每个“headingName”节点存储到一个单独的 Java Class （该结构将来可能会因数据块数组中的各个类别/标题名称而有所不同，例如我的回复是：

@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
@JsonIgnoreProperties(ignoreUnknown = true)
public class MyResponse {
    private OperationalInformation operationalInformation;
    private CRA cra;
    ...
}

我该如何做到这一点？

Answer 1

我的解决方案涉及两个阶段：

反序列化

将 json 反序列化为与输入 json 具有相同层次结构和结构的对象，这些对象包含 String 和 object 属性。 这组类可以支持任何和所有标题名称和属性

@Data
public abstract class JsonCommonProperties {
    protected String name;
    protected String position;
}

@Data
@ToString(callSuper = true)
public class JsonAttribute<T> extends JsonCommonProperties {
    private T value;
}

@Data
@ToString(callSuper = true)
public class JsonHeading extends JsonCommonProperties {
    private List<JsonAttribute<?>> attributes;
}

public class JsonResponse {
    
    public List<HeadingName> dataBlock;
    
    public static class HeadingName {
        public JsonHeading headingName;
    }
}

这组类可用于使用ObjectMapper.readValue()进行直接反序列化。

构建我的响应

将上述 object 层次结构转换为 MyResponse，使用反射和 3rd 方库来构建对象并从字符串名称填充实例变量。

我为所有标题使用了一个常见的超级 class。 下面指定了类集

@Getter
@Setter
@ToString
public abstract class Heading {
    private String name;

    public Heading(String name) {
        setName(name);
    }
}

@Getter
@Setter
@ToString(callSuper = true)
class CRA extends Heading {
    private String company;
    private Map<String, String> address;

    public CRA (String name) {
        super(name);
    }
}

@Getter
@Setter
@ToString(callSuper = true)
class OperationalInformation extends Heading {
    private List<String> brand;
    private String dataModelId;

    public OperationalInformation (String name) {
        super(name);
    }
}

@Data
public class MyResponse {
    private OperationalInformation operationalInformation;
    private CRA cra;
}

挑战和解决方案是

1. convert heading name to type of heading: I used apache common StringUtils and apache text CaseUtils to convert json property to class name, and then used reflection to instantiate an instance

2. 从JsonAttribute填充标题的实例变量：我使用Jodd Util库（更好的替代 apache BeanUtils ）从字符串名称动态调用设置器。

我把上面的解决方案放在一个标题厂class

public class HeadingFactory {

    /**
     * instantiate concrete Heading from specified JsonHeading
     */
    @SuppressWarnings("unchecked")
    public static Heading getHeading(JsonHeading jsonHeading) throws ReflectiveOperationException {
        Class<Heading> concreteHeadingClass =
            (Class<Heading>)Class.forName(getMyResponsePackage() + "." + getHeadingType(jsonHeading));
        Heading heading = concreteHeadingClass.getConstructor(String.class).newInstance(jsonHeading.getName());
        populateHeading(heading, jsonHeading);
        return heading;
    }

    /**
     * return class name of concrete Heading (type of instance var in MyResponse)
     * from specified JsonHeading
     */
    public static String getHeadingType(JsonHeading jsonHeading) {
        // handle all uppercase heading, like CRA
        if (StringUtils.isAllUpperCase(jsonHeading.getName())) return jsonHeading.getName();
        // handle separate word heading, like Operational Information
        return CaseUtils.toCamelCase(jsonHeading.getName(), true, ' ');
    }

    private static String getMyResponsePackage() {
        return StringUtils.substringBeforeLast(MyResponse.class.getName(), ".");
    }

    // use jodd util to dynamically call setter methods of heading
    // and pass values of jsonAttributes of jsonHeading
    private static void populateHeading(Heading heading, JsonHeading jsonHeading) {
        jsonHeading.getAttributes().forEach(jsonAttribute ->
            BeanUtil.pojo.setProperty(heading, jsonAttribute.getPropertyName(), jsonAttribute.getValue())
        );
    }
}

使用相同的技术和库填充 MyResponse 的实例变量。 这是我的主要方法

public static void main(String[] args) {
    ObjectMapper mapper = new ObjectMapper();
    try (BufferedReader br = Files.newBufferedReader(Paths.get("C://temp/json.json"))) {
        JsonResponse jsonResponse = mapper.readValue(br, JsonResponse.class);
        MyResponse myResponse = new MyResponse();
        for (JsonResponse.HeadingName headingName : jsonResponse.dataBlock) {
            Heading heading = HeadingFactory.getHeading(headingName.headingName);
            String myResponseProperty = headingName.headingName.getPropertyName();
            BeanUtil.pojo.setProperty(myResponse, myResponseProperty, heading);
        }
        System.out.println(myResponse);

    } catch (Exception e) {
        e.printStackTrace();
    }
}

Answer 2

在反序列化为 POJO 之前，使用 JSON 查询语言转换 JSON 结构。 一条Josson查询语句就可以完成这项工作。

dataBlock.map(headingName.name.camelCase()::headingName.field(name:)).mergeObjects()

https://github.com/octomix/josson

演示

Josson josson = Josson.fromJsonString(
    "{" +
    "  \"dataBlock\": [ {" +
    "    \"headingName\": {" +
    "      \"name\": \"Operational Information\"," +
    "      \"position\": \"1\"," +
    "      \"attributes\": [ {" +
    "          \"name\": \"Brand\"," +
    "          \"position\": \"1\"," +
    "          \"value\": [\"A\",\"B\"]" +
    "      }," +
    "      {" +
    "        \"name\": \"Data Model Id\"," +
    "        \"position\": \"2\"," +
    "        \"value\": \"000001\"" +
    "      } ]" +
    "    }" +
    "  }," +
    "  {" +
    "    \"headingName\": {" +
    "      \"name\": \"CRA\"," +
    "      \"position\": \"6\"," +
    "      \"attributes\": [ {" +
    "        \"name\": \"Company\"," +
    "        \"position\": \"1\"," +
    "        \"value\": \"private_limited_company\"" +
    "      }," +
    "      {" +
    "        \"name\": \"Address\"," +
    "        \"position\": \"3\"," +
    "        \"value\": {" +
    "          \"line1\": \"AAA\"," +
    "          \"line2\": \"BBB\"," +
    "          \"line3\": \"CCC\"," +
    "          \"line4\": \"DDD\"," +
    "          \"postalCode\": \"AB XYZ\"," +
    "          \"countryCode\": \"GBR\"" +
    "        }" +
    "      } ]" +
    "    }" +
    "  } ]" +
    "}");
JsonNode node = josson.getNode(
    "dataBlock.map(headingName.name.camelCase()::headingName.field(name:)).mergeObjects()");
System.out.println(node.toPrettyString());

Output

{
  "operationalInformation" : {
    "position" : "1",
    "attributes" : [ {
      "name" : "Brand",
      "position" : "1",
      "value" : [ "A", "B" ]
    }, {
      "name" : "Data Model Id",
      "position" : "2",
      "value" : "000001"
    } ]
  },
  "cra" : {
    "position" : "6",
    "attributes" : [ {
      "name" : "Company",
      "position" : "1",
      "value" : "private_limited_company"
    }, {
      "name" : "Address",
      "position" : "3",
      "value" : {
        "line1" : "AAA",
        "line2" : "BBB",
        "line3" : "CCC",
        "line4" : "DDD",
        "postalCode" : "AB XYZ",
        "countryCode" : "GBR"
      }
    } ]
  }
}