I need to implement a custom converter in spring data elastic-search. I need to concatenate something to the text when saving and retrieving. I have seen a similar question is here. And in the answer is says it is implemented now and available. But I didn't found any example how to implement it. So anybody know how to do this?
You can find examples in the test code of the library.
You have to create a converter:
class FooConverter implements PropertyValueConverter {
public static final String PREFIX = "foo-";
#Override
public Object write(Object value) {
return PREFIX + value.toString();
}
#Override
public Object read(Object value) {
String valueString = value.toString();
if (valueString.startsWith(PREFIX)) {
return valueString.substring(PREFIX.length());
} else {
return value;
}
}
}
and then register it for the property of your entity class:
#Document(indexName = "foo")
class Entity {
#Field(type = FieldType.Text)
#ValueConverter(FooConverter.class)
private String someField;
// ...
}
Related
I'm struggling with mappings some field. I looked for an answer but couldn't find anything solving my case. Let's cut to the chase.
I have my document class
#Doucment
public class DocumentClass {
#Field(type = FieldType.Nested)
private EmployeeId employeeId;
}
An EmployeeId is wrapper for my uuid identifier. This object has nothing but just getters and setters and jackson annotations. The thing is that object extends some base class so such objects like EmployeeId can inherit this object. This super class has field id and this causes the problem. When I post some data to elasticsearch then it looks like this:
{
"employeeId": {
"id": "someUUID"
}
}
But I want to map this to be like:
{
"employeeId": "someUUID"
}
I wonder if there is a way to flatten this object.
If I get it right, you want to convert your EmployeeId class to a String and back. You have 2 possibilities to do that:
Using a property converter
If you only want to convert an EmployeeId in this entity and might keep it as it is in another, you should use a property converter that is only registered for this property:
import org.springframework.data.elasticsearch.core.mapping.PropertyValueConverter;
public class EmployeeIdConverter implements PropertyValueConverter {
#Override
public Object write(Object value) {
return value instanceof EmployeeId employeeId ? employeeId.getId() : value.toString();
}
#Override
public Object read(Object value) {
return new EmployeeId(value.toString());
}
}
This converter must be registered on the property, notice that the field type is set to Keyword as it probably should not be analysed:
import org.springframework.data.elasticsearch.annotations.ValueConverter;
#Document
public class DocumentClass {
#Field(type = FieldType.Keyword)
#ValueConverter(EmployeeIdConverter.class)
private EmployeeId employeeId;
}
Using a global converter
If you are using this EmployeeId at several places you might want register globally 2 converters for the two conversion directions:
#WritingConverter
public class EmployeeIdToString implements Converter<EmployeeId, String>{
#Nullable
#Override
public String convert(EmployeeId employeeId) {
return employeeId.getId();
}
}
#ReadingConverter
public class StringToEmployeeId implements Converter<String, EmployeeId>{
#Nullable
#Override
public EmployeeId convert(String id) {
return new EmployeeId(id);
}
}
To register these, you need to provide a custom client configuration (see the documentation):
#Configuration
public class MyClientConfig extends ElasticsearchConfiguration {
#Override
public ClientConfiguration clientConfiguration() {
return ClientConfiguration.builder()
.connectedTo("localhost:9200")
.build();
}
#Override
public ElasticsearchCustomConversions elasticsearchCustomConversions() {
Collection<Converter<?, ?>> converters = new ArrayList<>();
converters.add(new EmployeeIdToString());
converters.add(new StringToEmployeeId());
return new ElasticsearchCustomConversions(converters);
}
}
In this case, only the field type needs to be adjusted
#Document
public class DocumentClass {
#Field(type = FieldType.Keyword)
private EmployeeId employeeId;
}
I am guessing you have an Id field which is a string.
You need to put a #JsonValue annotation on that field to make jackson serialize it the way you want.
Field annotated by JsonValue will be used to serialize your pojo into json.
If your Id field is private, then add the annotation on the getter of that field.
https://fasterxml.github.io/jackson-annotations/javadoc/2.8/com/fasterxml/jackson/annotation/JsonValue.html
Is it possible to use Jersey with Moxy to/from Json and Java 8 Optionals?
How to configure it?
You can declare following class:
public class OptionalAdapter<T> extends XmlAdapter<T, Optional<T>> {
#Override
public Optional<T> unmarshal(T value) throws Exception {
return Optional.ofNullable(value);
}
#Override
public T marshal(Optional<T> value) throws Exception {
return value.orElse(null);
}
}
And use like this:
#XmlRootElement
public class SampleRequest {
#XmlElement(type = Integer.class)
#XmlJavaTypeAdapter(value = OptionalAdapter.class)
private Optional<Integer> id;
#XmlElement(type = String.class)
#XmlJavaTypeAdapter(value = OptionalAdapter.class)
private Optional<String> text;
/* ... */
}
Or declare in package-info.java and remove #XmlJavaTypeAdapter from POJOs:
#XmlAccessorType(XmlAccessType.FIELD)
#XmlJavaTypeAdapters({
#XmlJavaTypeAdapter(type = Optional.class, value = OptionalAdapter.class)
})
But here are some drawbacks:
Adapter above can only work with simple types like Integer, String, etc. that can be parsed by MOXY by default.
You have to specify #XmlElement(type = Integer.class) explicitly to tell the parser type are working with, otherwise null values would be passed to adapter's unmarshal method.
You miss the opportunity of using adapters for custom types, e.g. custom adapter for java.util.Date class based on some date format string. To overcome this you'll need to create adapter something like class OptionalDateAdapter<String> extends XmlAdapter<String, Optional<Date>>.
Also using Optional on field is not recommended, see this discussion for details.
Taking into account all the above, I would suggest just using Optional as return type for your POJOs:
#XmlRootElement
public class SampleRequest {
#XmlElement
private Integer id;
public Optional<Integer> getId() {
return Optional.ofNullable(id);
}
public void setId(Integer id) {
this.id = id;
}
}
In our web app create and update forms have a size validation. For instance:
#Size(min = 4, max = 20)
private String mobile;
As seen the field is not required. But at the front-end user wants to clear field. Then form validation fails because of length restriction. Incoming data is an empty string instead of null. So minimum length validation restricts the input.
Therefore I start to search a solution to convert empty strings to null values. I found a #InitBinder and StringTrimmerEditor solution but our system uses #ResponseBody approach. So It doesn't fit.
Adding #JsonDeserialize(using = CustomTrimDeserializer.class) annotation or writing a custom setter for every string field is not DRY solution.
I just want to add app wide custom deserializer for String fields.
I finally examine the JsonComponentModule class and noticed spring is looking for the JsonComponent annotation for deserializer registration.
This is a one file spring boot project for solution
#RestController
#SpringBootApplication
public class CheckNullApplication {
public static void main(String[] args) {
SpringApplication.run(CheckNullApplication.class, args);
}
#PostMapping("/check-null")
public boolean checkNull(#RequestBody final HelloForm form) {
return form.getName() == null;
}
public static class HelloForm {
private String name;
public String getName() { return name; }
public void setName(final String name) { this.name = name;}
}
#JsonComponent
public static class StringTrimmerDeserializer extends JsonDeserializer<String> {
#Override
public String deserialize(final JsonParser p, final DeserializationContext ctxt)
throws IOException, JsonProcessingException {
String result = StringDeserializer.instance.deserialize(p, ctxt);
if (result != null) {
result = result.trim();
if (StringUtils.isEmpty(result)) {
return null;
}
}
return result;
}
}
}
Instead of adding #JsonDeserialize annotation you may want to just register your custom deserializer via Module (for example, SimpleModule), and it will apply to all String valued properties. Something like:
SimpleModule module = new SimpleModule(...);
module.addDeserializer(String.class, new CustomTrimDeserializer());
mapper.registerModule(module);
Create a class as following and annotate with #JsonComponent. Spring boot will pick that up as a component.
import com.fasterxml.jackson.databind.deser.std.StringDeserializer;
#JsonComponent
public class WhitSpaceTrimmerDeserializer extends StringDeserializer {
#Override
public String deserialize(JsonParser p, DeserializationContext ctxt) throws IOException {
final String value = super.deserialize(p, ctxt);
return value!=null?value.trim():null;
}
Bear with my English. I have a simple json,
{
"Hint2": "L"
}
this is the POJO that works.
public class Hints {
#SerializedName("Hint2")
#Expose
private String Hint2;
public void setHint1(Object Hint2) {
this.Hint2 = (Hint2);
}
}
i want to change it to
public class Hints {
#SerializedName("Hint2")
#Expose
public final ObservableField<String> Hint2 = new ObservableField<>();
public void setHint2(String Hint2) {
this.Hint2.set(Hint2);
}
}
both class has same setter method, same #SerializedName annotation tag. only type of Hint2 object is changed. but the latter one throws exception shown below
Caused by: com.google.gson.JsonSyntaxException: java.lang.IllegalStateException: Expected BEGIN_OBJECT but was STRING at..
so i believe deserialization depends on what kind of variable "Hint2" is.
Is there a way to make it work with ObservableField rather than using String?
The reason i'm trying this is android binding library, which supports binding objects directly to xml files. and the ObservableField automatically updates UI when corresponding value in POJO is changed.
Update:
gson design document has this
Using fields vs getters to indicate Json elements
Some Json libraries use the getters of a type to deduce the Json elements. We chose to use all fields (up the inheritance hierarchy) that are not transient, static, or synthetic. We did this because not all classes are written with suitably named getters. Moreover, getXXX or isXXX might be semantic rather than indicating properties.
However, there are good arguments to support properties as well. We intend to enhance Gson in a latter version to support properties as an alternate mapping for indicating Json fields. For now, Gson is fields-based.
so this indicates that Gson is fields-based. this pretty much answers my question but still waiting if anyone has someway around this.
I came across the same requirements, and resolved it finally, here are the steps:
create the class GsonUtils:
public class GsonUtils {
// code following
//...
}
following code are in this class
write a customized serializer & deserializer:
private static class ObservableFieldSerializerDeserializer implements JsonSerializer>, JsonDeserializer> {
#Override
public JsonElement serialize(ObservableField src, Type typeOfSrc, JsonSerializationContext context) {
return context.serialize(src.get());
}
#Override
public ObservableField deserialize(JsonElement json, Type typeOfT, JsonDeserializationContext context) throws JsonParseException {
final Type type = ((ParameterizedType) typeOfT).getActualTypeArguments()[0];
return new ObservableField((T) GsonUtils.getGson().fromJson(json, type));
}
}
you need to register ObservableField types to Gson:
private static GsonBuilder createGsonBuilder() {
final GsonBuilder gsonBuilder = new GsonBuilder();
gsonBuilder.registerTypeAdapter(new TypeToken<ObservableField<String>>(){}.getType(), new ObservableFieldSerializerDeserializer<String>());
...// register more types which are wrapped by ObservableFields
return gsonBuilder;
}
create the Gson which is used by the deserializer
private static final Gson sGson = createGson();
private static Gson createGson() {
return createGsonBuilder().create();
}
// this is used by the deserializer
public static Gson getGson() {
return sGson;
}
that's all, hope it helps
I just ran into what I think is the same issue, and here is a JUnit4 test showing how I solved it with Jackson for a POJO, but of course String would work as well.
public class ObservableDeserializationTest {
private static class ObservableDeserializer extends JsonDeserializer<ObservableField> implements ContextualDeserializer {
private Class<?> mTargetClass;
#Override
public JsonDeserializer<?> createContextual(DeserializationContext ctxt, BeanProperty property) throws JsonMappingException {
mTargetClass = property.getType().containedType(0).getRawClass();
return this;
}
#Override
public ObservableField deserialize(JsonParser p, DeserializationContext ctxt) throws IOException, JsonProcessingException {
ObservableField result = new ObservableField();
result.set(p.readValueAs(mTargetClass));
return result;
}
}
private static class SomePojo {
public String id;
public String name;
}
private static class ObservableTestClass {
#JsonDeserialize(using = ObservableDeserializer.class)
public ObservableField<SomePojo> testObj = new ObservableField<>();
}
#Test
public void DeserializingAnObservableObjectShouldSetValueCorrectly() {
ObservableTestClass tc = null;
try {
tc = new ObjectMapper().readValue("{\"testObj\":{\"name\":\"TestName\",\"id\":\"TestId\"}}", ObservableTestClass.class);
} catch (IOException e) {
e.printStackTrace();
}
Assert.assertEquals("TestName", tc.testObj.get().name);
Assert.assertEquals("TestId", tc.testObj.get().id);
}
}
The key is the ContextualDeserializer interface that allows extracting the contained class type. Jackson provides several options for registering a custom deserializer, so this is but one way of doing it. Also, it would probably be a good idea to override getNullValue as well in the deserializer if you would use this for real.
We are using JPA Entities to get the database rows and then when we transfer that to the external, we want to use disconnected object (DTO) which are simple beans annotated with JAX-B.
We use a mapper and its code looks like this:
public BillDTO map(BillEntity source, BillDTO target) {
BeanUtils.copyProperties(source, target);
return target;
}
But when the code is running we get an error like this:
java.lang.IllegalArgumentException: argument type mismatch
Note this is the Spring implementation of the BeanUtils:
import org.springframework.beans.BeanUtils
And the naming of the properties are identical (with their getter/setter).
Anybody knows why the error happens?
And how to use a fast way instead just copying properties one by one?
This example working well. Here String property is copied to enum property:
Entity:
public class A {
private String valueFrom;
public String getValue() {
return valueFrom;
}
public void setValue(String value) {
this.valueFrom = value;
}
}
DTO (En is enumeration):
public class B {
private En valueTo;
public void setValue(String def) {
this.valueTo = En.valueOf(def);
}
public void setEnumValue(En enumVal) {
this.valueTo = enumVal;
}
}
As for your GitHub example, problem in class B in getter should be:
public String getValue()
Example:
public String getValue() {
return value.toString();
}