gpt4 book ai didi

java - 使用 fastxml.jackson 将 JSON 解析为 Java 记录

转载 作者:行者123 更新时间:2023-12-03 07:59:24 25 4
gpt4 key购买 nike

根据设计,Java 记录不能从另一个对象继承(请参阅 Why Java records do not support inheritance? )。所以我想知道实现以下目标的最佳方法是什么。

鉴于我的 JSON 数据包含具有一些通用数据 + 唯一数据的对象。例如,类型、宽度和高度适用于所有形状,但根据类型,它们可以具有附加字段:

{
"name": "testDrawing",
"shapes": [
{
"type": "shapeA",
"width": 100,
"height": 200,
"label": "test"
},
{
"type": "shapeB",
"width": 100,
"height": 200,
"length": 300
},
{
"type": "shapeC",
"width": 100,
"height": 200,
"url": "www.test.be",
"color": "#FF2233"
}
]
}

在“传统”Java 中,您可以这样做

BaseShape with width and height
ShapeA extends BaseShape with label
ShapeB extends BaseShape with length
ShapeC extends BaseShape with URL and color

但我有点固执,真的很想用记录。

我的解决方案现在如下所示:

  • 没有基础形状
  • 公共(public)字段在所有记录中重复
@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record Drawing(
@JsonProperty("name")
String name,

@JsonProperty("shapes")
@JsonDeserialize(using = TestDeserializer.class)
List<Object> shapes // I don't like the Objects here...
) {
}

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record ShapeA (
@JsonProperty("type") String type,
@JsonProperty("width") Integer width,
@JsonProperty("height") Integer height,
@JsonProperty("label") String label
) {
}

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record ShapeB(
@JsonProperty("type") String type,
@JsonProperty("width") Integer width,
@JsonProperty("height") Integer height,
@JsonProperty("length") Integer length
) {
}

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record ShapeC(
@JsonProperty("type") String type,
@JsonProperty("width") Integer width,
@JsonProperty("height") Integer height,
@JsonProperty("url") String url,
@JsonProperty("color") String color
) {
}

我不喜欢重复的代码,这是一种不好的做法...但最终我可以使用这个辅助类来加载它:

public class TestDeserializer extends JsonDeserializer {

ObjectMapper mapper = new ObjectMapper();

@Override
public List<Object> deserialize(JsonParser jsonParser, DeserializationContext deserializationContext) throws IOException {
List<Object> rt = new ArrayList<>();

JsonNode node = jsonParser.getCodec().readTree(jsonParser);

if (node instanceof ArrayNode array) {
for (Iterator<JsonNode> it = array.elements(); it.hasNext(); ) {
JsonNode childNode = it.next();
rt.add(getShape(childNode));
}
} else {
rt.add(getShape(node));
}

return rt;
}

private Object getShape(JsonNode node) {
var type = node.get("type").asText();
switch (type) {
case "shapeA":
return mapper.convertValue(node, ShapeA.class);
case "shapeB":
return mapper.convertValue(node, ShapeB.class);
case "shapeC":
return mapper.convertValue(node, ShapeC.class);
default:
throw new IllegalArgumentException("Shape could not be parsed");
}
}
}

事实证明这个测试工作正常:

@Test
void fromJsonToJson() throws IOException, JSONException {
File f = new File(this.getClass().getResource("/test.json").getFile());
String jsonFromFile = Files.readString(f.toPath());

ObjectMapper mapper = new ObjectMapper();
Drawing drawing = mapper.readValue(jsonFromFile, Drawing.class);
String jsonFromObject = mapper.writeValueAsString(drawing);

System.out.println("Original:\n" + jsonFromFile.replace("\n", "").replace(" ", ""));
System.out.println("Generated:\n" + jsonFromObject);

assertAll(
//() -> assertEquals(jsonFromFile, jsonFromObject),
() -> assertEquals("testDrawing", drawing.name()),
() -> assertTrue(drawing.shapes().get(0) instanceof ShapeA),
() -> assertTrue(drawing.shapes().get(1) instanceof ShapeB),
() -> assertTrue(drawing.shapes().get(2) instanceof ShapeC)
);
}

使用 Jackson 库和 Java Records 实现这一目标的最佳方法是什么?

额外的旁注:我还需要能够以与原始格式相同的格式写回 JSON。

最佳答案

记录不能继承,因为它们旨在成为一个可靠的契约,但它们可以实现一个接口(interface)。因此,您可以使用 Jackson 2.12 或更高版本使用 JasonSubTypes 执行类似的操作:

型号

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record Drawing(
String name,
List<BaseShape> shapes
) { }

// added benefit of interface here is it reminds you to have the default fields
@JsonTypeInfo(use = JsonTypeInfo.Id.DEDUCTION)
@JsonSubTypes({
@JsonSubTypes.Type(ShapeA.class),
@JsonSubTypes.Type(ShapeB.class),
@JsonSubTypes.Type(ShapeC.class)
})
public interface BaseShape {
Integer width();
Integer height();
}

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record ShapeA (
Integer width,
Integer height,
String label
) implements BaseShape { }

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record ShapeB(
Integer width,
Integer height,
Integer length
) implements BaseShape { }

@JsonIgnoreProperties(ignoreUnknown = true)
@JsonInclude(JsonInclude.Include.NON_NULL)
public record ShapeC(
Integer width,
Integer height,
String url,
String color
) implements BaseShape { }

测试类

@Slf4j
class DemoTest {

private ObjectMapper objectMapper = ObjectMapperBuilder.getObjectMapper();

@Test
void test() throws JsonProcessingException {
final String testString = objectMapper
.writerWithDefaultPrettyPrinter()
.writeValueAsString(
new Drawing(
"happy",
List.of(
new ShapeA(1, 1, "happyShape"),
new ShapeB(2, 2, 3),
new ShapeC(2, 2, "www.shape.com/shape", "blue"
)
)
)
);

log.info("From model to string {}", testString);

Drawing drawing = objectMapper.readValue(testString, Drawing.class);

log.info(
"Captured types {}",
drawing
.shapes()
.stream()
.map(s -> s.getClass().getName())
.collect(Collectors.toSet())
);

log.info(
"From string back to model then again to string {}",
objectMapper
.writerWithDefaultPrettyPrinter()
.writeValueAsString(drawing)
);
}

}

这是测试日志输出:

17:06:41.293 [Test worker] INFO com.demo.DemoTest - From model to string {
"name" : "happy",
"shapes" : [ {
"width" : 1,
"height" : 1,
"label" : "happyShape"
}, {
"width" : 2,
"height" : 2,
"length" : 3
}, {
"width" : 2,
"height" : 2,
"url" : "www.shape.com/shape",
"color" : "blue"
} ]
}
17:06:41.353 [Test worker] INFO com.demo.DemoTest - Captured types [com.demo.DemoTest$ShapeB, com.demo.DemoTest$ShapeA, com.demo.DemoTest$ShapeC]
17:06:41.354 [Test worker] INFO com.demo.DemoTest - From string back to model then again to string {
"name" : "happy",
"shapes" : [ {
"width" : 1,
"height" : 1,
"label" : "happyShape"
}, {
"width" : 2,
"height" : 2,
"length" : 3
}, {
"width" : 2,
"height" : 2,
"url" : "www.shape.com/shape",
"color" : "blue"
} ]
}

请注意,您可以将 type 字段添加为 @JsonSubTypes.Type 注释的 name 属性,但这无论有或没有都有效只要您的记录中的字段永远不会完全相同,就可以作为鉴别器。

您可以阅读有关JsonSubtypes的更多信息here .

关于java - 使用 fastxml.jackson 将 JSON 解析为 Java 记录,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/74890806/

25 4 0
Copyright 2021 - 2024 cfsdn All Rights Reserved 蜀ICP备2022000587号
广告合作:1813099741@qq.com 6ren.com