Prevent writing default attribute values JAXB - java

I am trying to write the class in order to write an element with attributes in JAXB. In this XML there are some default values whether they be Strings, ints, or custom class types.
The following cut down example:
#XmlAccessorType(XmlAccessType.NONE)
#XmlRootElement(name = "FIELD")
public class TestLayoutNode
{
// I want to not write this to the xml when it is 0
#XmlAttribute(name = "num")
private int number;
// I want to not write this when it is "default"
#XmlAttribute(name = "str")
private String str;
}
As per JAXB Avoid saving default values
I know if I want to not write the String I can modify the getters/setters to write null and read in the default value if it reads in null.
However, with the int I am not sure what to do as it will always have the value 0 unless it is specifically changed.
Is there a nicer way to do this? I could change the internal data types to String and then cast it whenever it is needed but that's a bit messy.

You could do the following by changing the fields to be the object types by default null values do not appear in the XML representation) and putting some logic in the getters:
#XmlAccessorType(XmlAccessType.NONE)
#XmlRootElement(name = "FIELD")
public class TestLayoutNode
{
#XmlAttribute(name = "num")
private Integer number;
#XmlAttribute
private String str;
public int getNumber() {
if(null == number) {
return 0;
} else {
return number;
}
}
public void setNumber(int number) {
this.number = number;
}
public String getStr() {
if(null == str) {
return "default";
} else {
return str;
}
}
public void setStr(String str) {
this.str = str;
}
}
Allowing the Property to be Unset
If you want to allow the set operation to return a property to its default state then you need to add logic in the set method.
public void setNumber(int number) {
if(0 == number) {
this.number = null;
} else {
this.number = number;
}
}
Alternatively you could offer an unset method:
public void unsetNumber() {
this.number = null;
}
Allowing a Set to null
If you want to allow the str property to be set to null so that the get method will return null and not "default" then you can maintain a flag to track if it has been set:
private strSet = false;
public String getStr() {
if(null == str && !strSet) {
return "default";
} else {
return str;
}
}
public void setStr(String str) {
this.str = str;
this.strSet = true;
}
UPDATE
Blaise, don't you think that the solution is pretty verbose?
Yes
I mean that such use case should be probably supported by framework.
For example using annotation like #DefaultValue.
How JAXB Supports Default Values Today
If a node is absent from the XML then a set is not performed on the corresponding field/property in the Java Object. This means whatever value you have initialized the property to be is still there. On a marshal since the value is populated it will be marshalled out.
What is Really Being Asked For
What is really being asked for is to not marshal the field/property when it has the default value. In this way you want the marshal behaviour to be the same for null and default values. This introduces some problems to be solved:
How do you now marshal null to XML? By default is it still marshalled as a missing node?
Does a mechanism need to be provided to distinguish between the property being the default value (not present in the XML) and having been set to the same value as the default (present in the XML)?
What Are People Doing Today?
Generally for this use case people would just change the int property to Integer and have null be the default. I haven't encountered someone asking for this behaviour for a String before.

Use Integer instead of primitive int. Replace all primitive types with their object counterparts, then you can use NULL.
As per the string default value, use and modify the getter
import javax.xml.bind.JAXBContext;
import javax.xml.bind.JAXBException;
import javax.xml.bind.Marshaller;
import javax.xml.bind.annotation.XmlAttribute;
import javax.xml.bind.annotation.XmlElement;
import javax.xml.bind.annotation.XmlRootElement;
#XmlRootElement(name = "FIELD")
public class NullAttrs {
private Integer number;
private String str;
public void setNumber(Integer number) {
this.number = number;
}
#XmlAttribute(name = "num")
public Integer getNumber() {
return number;
}
public void setStr(String str) {
this.str = str;
}
#XmlAttribute(name = "str")
public String getStr() {
if (str != null && str.equalsIgnoreCase("default"))
return null;
else if (str == null)
return "default";
else
return str;
}
public static void main(String[] args) throws JAXBException {
JAXBContext jc = JAXBContext.newInstance(NullAttrs.class);
NullAttrs root = new NullAttrs();
root.setNumber(null);
root.setStr("default");
Marshaller marshaller = jc.createMarshaller();
marshaller.setProperty(Marshaller.JAXB_FORMATTED_OUTPUT, true);
marshaller.marshal(root, System.out);
}
}
Result in this case would be, empty FIELD:
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<FIELD/>

You can change to Integer
private Integer number;
Then the value of the object will be null when not instantiated.

Though it's not as terse as one would wish, one can create XmlAdapters to avoid marshalling the default values.
The use case is like this:
#XmlRootElement(name = "FIELD")
public class TestLayoutNode
{
#XmlAttribute(name = "num")
#XmlJavaTypeAdapter(value = IntegerZero.class, type = int.class)
public int number;
#XmlAttribute(name = "str")
#XmlJavaTypeAdapter(StringDefault.class)
public String str = "default";
}
And here are adapters.
IntegerZero:
public class IntegerZero extends DefaultValue<Integer>
{
public Integer defaultValue() { return 0; }
}
StringDefault:
public class StringDefault extends DefaultValue<String>
{
public String defaultValue() { return "default"; }
}
DefaultValueAdapter:
public class DefaultValue<T> extends XmlAdapter<T, T>
{
public T defaultValue() { return null; }
public T marshal(T value) throws Exception
{
return (value == null) || value.equals(defaultValue()) ? null : value;
}
public T unmarshal(T value) throws Exception
{
return value;
}
}
With small number of different default values this approach works well.

I find the solution using custom getters/setters or adapters annoyingly verbose, so I went for a different solution: a marshaller that checks values and nulls them out if they are at default.
import java.lang.reflect.Field;
import java.util.Collection;
import java.util.Set;
import javax.xml.bind.JAXBException;
import javax.xml.bind.Marshaller;
import javax.xml.bind.PropertyException;
import javax.xml.bind.helpers.AbstractMarshallerImpl;
import javax.xml.transform.Result;
import com.google.common.collect.ImmutableSet;
class MyJaxbMarshaller extends AbstractMarshallerImpl {
/** See https://docs.oracle.com/cd/E13222_01/wls/docs103/webserv/data_types.html#wp221620 */
private static final Set<String> SUPPORTED_BASIC_TYPES = ImmutableSet.of(
"boolean", "java.lang.Boolean", "byte", "java.lang.Byte", "double", "java.lang.Double",
"float", "java.lang.Float", "long", "java.lang.Long", "int", "java.lang.Integer",
"javax.activation.DataHandler", "java.awt.Image", "java.lang.String",
"java.math.BigInteger", "java.math.BigDecimal", "java.net.URI", "java.util.Calendar",
"java.util.Date", "java.util.UUID", "javax.xml.datatype.XMLGregorianCalendar",
"javax.xml.datatype.Duration", "javax.xml.namespace.QName",
"javax.xml.transform.Source", "short", "java.lang.Short");
private final Marshaller delegate;
MyJaxbMarshaller(Marshaller delegate) {
this.delegate = delegate;
}
#Override
public void setProperty(String name, Object value) throws PropertyException {
super.setProperty(name, value);
delegate.setProperty(name, value);
}
#Override
public void marshal(Object jaxbElement, Result result) throws JAXBException {
try {
delegate.marshal(clearDefaults(jaxbElement), result);
} catch (ReflectiveOperationException ex) {
throw new JAXBException(ex);
}
}
private Object clearDefaults(Object element) throws ReflectiveOperationException {
if (element instanceof Collection) {
return clearDefaultsFromCollection((Collection<?>) element);
}
Class<?> clazz = element.getClass();
if (isSupportedBasicType(clazz)) {
return element;
}
Object adjusted = clazz.getConstructor().newInstance();
for (Field field : clazz.getDeclaredFields()) {
field.setAccessible(true);
copyOrRemove(field, element, adjusted);
}
return adjusted;
}
private Object clearDefaultsFromCollection(Collection<?> collection)
throws ReflectiveOperationException {
#SuppressWarnings("unchecked")
Collection<Object> result = collection.getClass().getConstructor().newInstance();
for (Object element : collection) {
result.add(clearDefaults(element));
}
return result;
}
private static boolean isSupportedBasicType(Class<?> clazz) {
return SUPPORTED_BASIC_TYPES.contains(clazz.getName());
}
private void copyOrRemove(Field field, Object element, Object adjusted)
throws ReflectiveOperationException {
Object value = field.get(element);
if (value != null) {
if (value.equals(field.get(adjusted))) {
value = null;
} else {
value = clearDefaults(value);
}
}
field.set(adjusted, value);
}
}
This works with classes like
#XmlRootElement
public class Foo {
#XmlAttribute public Integer intAttr = 0;
#XmlAttribute public String strAttr = "default";
}
You can make this more flexible if you want, e.g. you can use an annotation to mark attributes you want to omit when they're at default, or extend the class to be aware of things like #XmlTransient or method accessors (neither of which is an issue in my project right now).
The price you pay for the simplicity of your binding classes is that the marshaller is going to create a deep copy of the object you're about to marshal, and make lots of comparisons to defaults to determine what to null out. So if runtime performance is an issue for you, this might be a no-go.

Related

How to get the field returned by a method invocation

I am writing a custom SonarQube rule for java, where I want to check an object is created with an argument having a specific annotation.
the file i am testing against
class MyClass {
public void doSomething() {
final var v = new Dto();
new MyObject(v.value1()); // Compliant since value1 has #MyAnnotation
new MyObject(v.value2()); // Noncompliant
}
public static class MyObject {
private final String value;
public MyObject(String value) {
this.value = value;
}
}
#Target(ElementType.FIELD)
#Retention(RetentionPolicy.RUNTIME)
public #interface MyAnnotation {
}
public static class Dto {
#MyAnnotation
private String value1;
private String value2;
public String value1() {
return this.value1;
}
public String value2() {
return this.value2;
}
}
}
the check
public class MyObjectCheck extends IssuableSubscriptionVisitor {
#Override
public List<Kind> nodesToVisit() {
return Collections.singletonList(Kind.NEW_CLASS);
}
#Override
public void visitNode(Tree tree) {
NewClassTree ctor = (NewClassTree) tree;
if(!ctor.identifier().symbolType().name().contains("MyObject")) { //to change
return;
}
if(ctor.arguments().size() == 1) {
final ExpressionTree expressionTree = ctor.arguments().get(0);
if(expressionTree.is(Kind.METHOD_INVOCATION)) {
MethodInvocationTree methodInvocation = (MethodInvocationTree) expressionTree;
}
}
}
}
from the methodInvocation, I can manage to call methodSelect to have a MethodInvocationTree but then I can't figure how to go to the field returned by the method.
I had to make concession where I consider the class of the method invoked being a POJO or a java record. This way I was able to fetch the linked field and annotations :
String methodName = methodInvocationTree.symbol().name();
final Symbol.TypeSymbol methodClass = (Symbol.TypeSymbol) methodInvocationTree.symbol().owner();
final List<SymbolMetadata.AnnotationInstance> annotations = methodClass.lookupSymbols(methodName).iterator().next().metadata().annotations();

Spring Custom Annotation Validation with multiple field

A little greedy question here, hope this one could also help others who want to know more about annotation validation
I am currently studying Spring, and for now, I am planning to try out the customize annotated validation.
I have searched a lot and now I know there are mainly two kinds of validations, one is used for the controller, and the other is the annotation method using #Valid
So here's my scenario:
Suppose I have two or more fields which can be null when they are ALL NULL.
But only when one of those fields contains any value except an empty string, those fields are required to have input. And I had two ideas but didn't know how to implement them correctly.
Here's the Class Example:
public class Subscriber {
private String name;
private String email;
private Integer age;
private String phone;
private Gender gender;
private Date birthday;
private Date confirmBirthday;
private String birthdayMessage;
private Boolean receiveNewsletter;
//Getter and Setter
}
Suppose I want that the birthday and confirmBirthday field need to be both null or the oppose, I may want to annotate them using one annotation for each of them and looks like this:
public class Subscriber {
private String name;
private String email;
private Integer age;
private String phone;
private Gender gender;
#NotNullIf(fieldName="confirmBirthday")
private Date birthday;
#NotNullIf(fieldName="birthday")
private Date confirmBirthday;
private String birthdayMessage;
private Boolean receiveNewsletter;
//Getter and Setter
}
So i do need to create the validation Annotation like this:
#Documented
#Constraint(validatedBy = NotNullIfConstraintValidator.class)
#Retention(RetentionPolicy.RUNTIME)
#Target({ ElementType.METHOD, ElementType.FIELD })
public #interface NotNullIf {
String fieldName();
String message() default "{NotNullIf.message}";
Class<?>[] group() default {};
Class<? extends Payload>[] payload() default {};
}
And After that i will need to create the Validator itself:
public class NotNullIfConstraintValidator implements ConstraintValidator<NotNullIf, String>{
private String fieldName;
public void initialize(NotNullIf constraintAnnotation) {
fieldName = constraintAnnotation.fieldName();
}
public boolean isValid(String value, ConstraintValidatorContext context) {
if(value == null) {
return true;
};
//TODO Validation
return false;
}
}
So how can it be achievable?
For another idea using the same Class as an example which said that i want birthday, confirmBirthday and birthdayMessdage can only be null or the oppose at the same time.
I may require to use the class annotated validation this time for cross-field validation.
Here's how i suppose to annotate the class:
#NotNullIf(fieldName={"birthday", "confirmBirthday", "birthdayMessage"})
public class Subscriber {
//Those field same as the above one
}
So when one of that field is not null, the rest of them also needs to be entered on the client size.
Is it Possible?
I have read this article: How to access a field which is described in annotation property
But I still confusing on how the annotation validation works from those elements I listed above.
Maybe I need some detail explanation on that code or even worse I may need some basic concept inspection.
Please Help!
For this you can use a type level annotation only because a field level annotation has no access to other fields!
I did something similar to allow a choice validation (exactly one of a number of properties has to be not null). In your case the #AllOrNone annotation (or whatever name you prefer) would need an array of field names and you will get the whole object of the annotated type to the validator:
#Target(ElementType.TYPE)
#Retention(RUNTIME)
#Documented
#Constraint(validatedBy = AllOrNoneValidator.class)
public #interface AllOrNone {
String[] value();
String message() default "{AllOrNone.message}";
Class<?>[] groups() default {};
Class<? extends Payload>[] payload() default {};
}
public class AllOrNoneValidator implements ConstraintValidator<AllOrNone, Object> {
private static final SpelExpressionParser PARSER = new SpelExpressionParser();
private String[] fields;
#Override
public void initialize(AllOrNone constraintAnnotation) {
fields = constraintAnnotation.value();
}
#Override
public boolean isValid(Object value, ConstraintValidatorContext context) {
long notNull = Stream.of(fields)
.map(field -> PARSER.parseExpression(field).getValue(value))
.filter(Objects::nonNull)
.count();
return notNull == 0 || notNull == fields.length;
}
}
(As you said you use Spring I used SpEL to allow even nested fields access)
Now you can annotate your Subscriber type:
#AllOrNone({"birthday", "confirmBirthday"})
public class Subscriber {
private String name;
private String email;
private Integer age;
private String phone;
private Gender gender;
private Date birthday;
private Date confirmBirthday;
private String birthdayMessage;
private Boolean receiveNewsletter;
}
Consider adding compile-time validation for the field names. For example, in #Arne answer the strings "birthday" and "confirmBirthday" are not guaranteed to match actual field names at compile time. If you want to add that functionality, here's an example from my code for a slightly different example that assumes there are exactly two fields. The purpose is to assert that two fields are ordered... For example, it could be used for "beginDate" and "endDate".
public class OrderedValidator extends AbstractProcessor implements ConstraintValidator<Ordered, Object>
{
private String field1;
private String field2;
private Messager messager;
public void initialize(Ordered constraintAnnotation)
{
this.field1 = constraintAnnotation.field1();
this.field2 = constraintAnnotation.field2();
}
#Override
public synchronized void init(ProcessingEnvironment processingEnv)
{
super.init(processingEnv);
messager = processingEnv.getMessager();
}
#SuppressWarnings("unchecked")
public boolean isValid(Object value, ConstraintValidatorContext context)
{
Object field1Value = new BeanWrapperImpl(value).getPropertyValue(field1);
Object field2Value = new BeanWrapperImpl(value).getPropertyValue(field2);
boolean valid = true;
if (field1Value != null && field2Value != null)
{
if (field1Value.getClass().equals(field2Value.getClass()))
{
valid = ((Comparable) field1Value).compareTo((Comparable) field2Value) <= 0;
}
}
return valid;
}
#Override
public boolean process(Set<? extends TypeElement> annotations, RoundEnvironment roundEnv)
{
for (Element annotatedElement : roundEnv.getElementsAnnotatedWith(Ordered.class))
{
if (annotatedElement.getKind() != ElementKind.CLASS)
{
messager.printMessage(Diagnostic.Kind.ERROR, "Only classes can be annotated with " + Ordered.class.getSimpleName());
return true;
}
TypeElement typeElement = (TypeElement) annotatedElement;
List<? extends Element> elements = typeElement.getEnclosedElements();
boolean field1Found = false;
boolean field2Found = false;
for (Element e : elements)
{
if (e.getKind() == ElementKind.FIELD && field1 != null && field1.equals(e.getSimpleName()))
{
field1Found = true;
}
else if (e.getKind() == ElementKind.FIELD && field2 != null && field2.equals(e.getSimpleName()))
{
field2Found = true;
}
}
if (field1 != null && !field1Found)
{
messager.printMessage(Diagnostic.Kind.ERROR, "Could not find field named " + field1);
return true;
}
if (field2 != null && !field2Found)
{
messager.printMessage(Diagnostic.Kind.ERROR, "Could not find field named " + field2);
return true;
}
}
return false;
}
}

JACKSON JSON deserialization issue with overrloaded setter

From several reasons we have to make for developers convenience be able to set the one reference via overloaded setters ( this due it is modelled as oneOf attribute).
I would expect that depending on the JSON schema the polymorphic (oneOf) property would have the deserialized reference to object of FooType or BarType, ....
depending on the JSON schema.
I was hoping since FooType , BarType follow bean convention they can be easily determined like it happens for JacksonFeature in JAXRS ....
In my dummy test it seems to not work as described below :
ObjectMapper mapper = new ObjectMapper();
mapper.enable(SerializationFeature.INDENT_OUTPUT);
mapper.setSerializationInclusion(JsonInclude.Include.NON_NULL);
mapper.configure(SerializationFeature.WRITE_DATES_AS_TIMESTAMPS, false);
mapper.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);
mapper.configure(DeserializationFeature.READ_ENUMS_USING_TO_STRING, true);
mapper.configure(SerializationFeature.WRITE_ENUMS_USING_TO_STRING, true);
mapper.readValue(SCHEMA, SimplePojo.class);
The issue is that the mapper crashes, and JSON schema(SCH1) can not be deserialized to POJO
The JSON schema (SCH1)
{
"dummy" : {
"bar" : "bar",
"baz" : 10
},
"other" : {
"foo" : "hi there"
},
"simple" : "simple"
}
The sub element types dummy, other look like :
public class BarType {
private String bar;
private Number baz;
public String getBar() {
return bar;
}
public void setBar(String bar) {
this.bar = bar;
}
public Number getBaz() {
return baz;
}
public void setBaz(Number baz) {
this.baz = baz;
}
and
public class FooType {
private Object foo;
public Object getFoo() {
return foo;
}
public void setFoo(Object foo) {
this.foo = foo;
}
The top level POJO ( i skipped some part )
public class SimplePojo {
private String simpleField;
private Object dummyField;
private Object otherField;
public String getSimple() {
return simpleField;
}
public void setSimple(String simple) {
this.simpleField = simple;
}
...
public void setDummy(final FooType dummy) {
this.dummyField = dummy;
}
public void setDummy(final BarType dummy) {
this.dummyField = dummy;
}
public void setDummy(final String dummy) {
this.dummyField = dummy;
}
the issue is that i can not deserialize correctly the schema (SCH1), instead I receive the :
com.fasterxml.jackson.databind.JsonMappingException: Conflicting setter definitions for property "dummy": com.hybris.api.poc.SimplePojo#setDummy(1 params) vs com.hybris.api.poc.SimplePojo#setDummy(1 params)
I was trying to use the #JsonCreator, and #JsonDeserialize but no luck it seems that i can not have two (non primitive) override setters
#JsonDeserialize( builder = FooType.FooTypeBuilder.class)
#JsonCreator
public void setDummy(final FooType dummy) {
this.dummyField = dummy;
}
/**
* Type specific setter for #dummy;
*
* #param dummy a reference to be set for #dummy
*/
#JsonDeserialize( builder = BarType.BarTypeBuilder.class )
#JsonCreator
public void setDummy(final BarType dummy) {
this.dummyField = dummy;
}
Can you hint me where I should the solution or am i breaking some principal concept ?
I do think you are breaking some principal concept there. For this type of scenario, having a base abstract class with JsonTypeInfo and JsonSubTypes annotations to describe your sub-object would probably be preferred. If you absolutely need the ability to set the three types via setDummy(...), would this work for you?
Replace:
public void setDummy(final FooType dummy) {
this.dummyField = dummy;
}
public void setDummy(final BarType dummy) {
this.dummyField = dummy;
}
public void setDummy(final String dummy) {
this.dummyField = dummy;
}
With:
#JsonDeserialize( using = DummyDeserializer.class )
public void setDummy(final Object dummy) {
// If you really need to restrict to the three types, throw exception here
if (! (dummy instanceof FooType || dummy instanceof BarType || dummy instanceof String) ) {
throw new Exception("Cannot setDummy dummy!");
}
this.dummyField = dummy;
}
This would require you to do the deserialization manually for all three classes in your DummyBuilder, but should solve your multi-setters problem. I've not tried to implement this, but think it works.
No, without inheritance structure Jackson has no way of automatically determining intended type during deserialization. If they did share the same base type, you could use #JsonTypeInfo to indicate how type id is included (usually as a property); and have a single setter (or creator property).
Otherwise you can not have conflicting setters (i.e. more than one with types that are not related to each other by sub-typing).

Esper: Grammatically defining a type with sub types?

I have to define the class below in ESPER so I'm able to reference the sub-types and internal arrays. I have to do it pragmatically. I don't care how:
UPDATE: The complete class:
public class IoTEntityEvent implements java.io.Serializable {
private IoTProperty[] Properties;
private String About;
IoTEntityEvent (){
this.About = null;
this.Properties = null;
}
public String getAbout() {
return About;
}
public void setAbout( String value){
this.About = value;
}
public void setProperties(int index, IoTProperty value) {
Properties[index] = value;
}
public IoTProperty getProperties(int index) {
return Properties[index];
}
public void setProperties( IoTProperty[] value) {
Properties = value;
}
public IoTProperty[] getProperties() {
return Properties;
}
}
This is the sub-class:
public class IoTProperty implements java.io.Serializable {
private Map<String,String>[] IoTStateObservation =null;
private String About = null;
IoTProperty (){
this.About = null;
this.IoTStateObservation = null;
}
public String getAbout() {
return About;
}
public void setAbout(String value) {
About = value;
}
public Map<String,String>[] getIoTStateObservation() {
return IoTStateObservation;
}
public void setIoTStateObservation( Map<String,String>[] value) {
IoTStateObservation = value;
}
public Map<String,String> getIoTStateObservation(int index) {
return IoTStateObservation[index];
}
public void setIoTStateObservation(int index, Map<String,String> value) {
IoTStateObservation[0] = value;
}
}
I tried like this :
eventNames[0] = "About";
eventType[0] = String.class;
eventNames[1] = "Properties";
eventType[1] = IoTProperty[].class;
epService.getEPAdministrator().getConfiguration().addEventType("type", eventNames, eventType);
This works but I can't access the sub-types. I also tried to define the sub type in similar manner. Can someone can explain how I suppose to do it?
What do you mean with "This works but I can't access the sub-types."
Tried like "select Properties[0].whatever" from type?
According to the Esper documentation:
Plain-old Java object events are object instances that expose event properties through JavaBeans-style getter methods. Events classes or interfaces do not have to be fully compliant to the JavaBean specification; however for the Esper engine to obtain event properties, the required JavaBean getter methods must be present or an accessor-style and accessor-methods may be defined via configuration.
In short, you need to create the JavaBean getters and setters in order to access your private members.
Thank you for the help. I found out how and is as following:
epService.getEPAdministrator().getConfiguration().addEventType("type",IoTEntityEvent.class);
Then the event should be send like this without any casting:
IoTValue[] va= {new IoTValue("0.62","2014-06-09T18:08:40.968Z","2014-06-09T18:08:40.968Z")};
IoTProperty[] pr = {new IoTProperty(va,"property")};
IoTEntityEvent event = new IoTEntityEvent(pr,"Entity");
epService.getEPRuntime().sendEvent(event);

Providing custom value serialization for enums via JAXB

For a project I'm working on, we have a lot of enums in use. The model object itself is composed from a lot of tiny classes; this model we then serialize to our DB as XML via JAXB. Now, we want to be able to serialize our enum values using the return of a particular method in the enum; that is given:
public enum Qualifier {
FOO("1E", "Foo type document"),
BAR("2", "Bar object");
private String code, description;
public Qualifier(String code, String description) {
this.code = code;
this.description = description;
}
public String getCode() {
return this.code;
}
public String getDescription() {
return this.description;
}
}
etc. etc. Currently, when serialized to XML, we get something like:
<qualifier>FOO</qualifier>
which is how JAXB handles it. However, we need the value to be the return of getCode(), and a whole lot of our enums do follow that convention (with a corresponding static method for lookup via code), so that the above XML fragment looks like:
<qualifier>1E</qualifier>
instead. We can annotate it with #XmlEnum and #XmlEnumValue, but that's too tedious -- some enums have up to 30 enumerated values, and hand-editing it is not good. We're also thinking of using a custom serializer instead, but I'd like to avoid going that route for now (but if that's the way to go, then I have no problem with it).
Any ideas how?
Try using the XmlAdapter mechanism for this. You create an XmlAdapter subclass for each enum type, and which knows how to marshal/unmarshal the enum to and from XML.
You then associate the adapter with the property, e.g.
public class QualifierAdapter extends XmlAdapter<String, Qualifier> {
public String marshal(Qualifier qualifier) {
return qualifier.getCode();
}
public Qualifier unmarshal(String val) {
return Qualifier.getFromCode(val); // I assume you have a way of doing this
}
}
and then in the model classes:
#XmlJavaTypeAdapter(QualifierAdapter.class)
private Qualifier qualifier;
You can also declare this at the package level, inside a file called package-info.java in the same package as your model classes, using the rather idiosyncratic package annotations:
#javax.xml.bind.annotation.adapters.XmlJavaTypeAdapters({
#javax.xml.bind.annotation.adapters.XmlJavaTypeAdapter(
type=Qualifier.class, value=QualifierAdapter.class
)
})
package com.xyz;
Found this question while looking for something else but I read your comment about something more generic. Heres what I have been using to convert upper case enum types to camel case. I am going to use your enum type but put my adapter on it. As you can see you dont need to reference every instance of Qualifier but just annotate the enum itself.
The CamelCaseEnumAdapter can take any enum however the enum class must be passed to it therefore you need to have a class extend it, I just use a private static class inside the enum itself.
Enum:
#XmlJavaTypeAdapter(Qualifier.Adapter.class)
public enum Qualifier {
FOO("1E", "Foo type document"),
BAR("2", "Bar object");
private String code, description;
public Qualifier(String code, String description) {
this.code = code;
this.description = description;
}
public String getCode() {
return this.code;
}
public String getDescription() {
return this.description;
}
private static class Adapter extends CamelCaseEnumAdapter<Qualifier> {
public Adapter() {
super(Qualifier.class, FOO);
}
}
}
Adapter
public abstract class CamelCaseEnumAdapter<E extends Enum> extends XmlAdapter<String, E>{
private Class<E> clazz;
private E defaultValue;
public CamelCaseEnumAdapter(Class<E> clazz) {
this(clazz, null);
}
public CamelCaseEnumAdapter(Class<E> clazz, E defaultValue) {
this.clazz = clazz;
this.defaultValue = defaultValue;
}
#Override
#SuppressWarnings("unchecked")
public E unmarshal(String v) throws Exception {
if(v == null || v.isEmpty())
return defaultValue;
return (E) Enum.valueOf(clazz, v.replaceAll("([a-z])([A-Z])", "$1_$2").toUpperCase());
}
#Override
public String marshal(E v) throws Exception {
if(v == defaultValue)
return null;
return toCamelCase(v.name());
}
private String toCamelCase(String s){
String[] parts = s.split("_");
String camelCaseString = "";
for (String part : parts){
if(camelCaseString.isEmpty())
camelCaseString = camelCaseString + part.toLowerCase();
else
camelCaseString = camelCaseString + toProperCase(part);
}
return camelCaseString;
}
private String toProperCase(String s) {
return s.substring(0, 1).toUpperCase() +
s.substring(1).toLowerCase();
}
}

Categories