We Keep Coding

Order matters with class metadata in Genson - Is there a work-around?

I'm using Genson to serialize + deserialize json in my android app into polymorphic objects. The JSON is coming from a variety of sources though and I can't guarantee that the #class metadata will be the first line item in the json. Walking through the Genson code and writing test cases it looks like the #class metadata has to be the first entry in the dictionary.
Has anyone had luck working around this constraint? Is it time to switch to something else, and if so, what?
public class Message {
Payload payload;
// getters & setters
}
public abstract class Payload {
//
}
public class Notification1 extends Payload {
String text;
// getters & setters
}
public class Notification2 extends Payload {
String otherText
// getters & setters
}
String correctOrder = {"#class":"Message","payload":{"#class":"Notification1","text":"Text"}}
String modifiedOrder = {"#class":"Message","payload":{"text":"Text", "#class":"Notification1"}}
Genson g = Genson.Builder()
.addAlias("Notification1", Notification1.class)
.addAlias("Notification2", Notification2.class)
.useRuntimeType(true)
.useClassMetadata(true)
.useMetadata(true)
.useFields(false)
.useIndentation(false)
.create();
g.deserialize(correctOrder, Message.class) // This works
g.deserialize(modifiedOrder, Message.class) // This barfs with the error: com.owlike.genson.JsonBindingException: Could not deserialize to type class com.ol.communication.messages.Message

Indeed the order matters. This was choosed on purpose, see the remarks in the user guide.
If we allow the #class property anywhere in the json object, then we will have to first deserialize all the json object (and its sub properties obj/arr etc) to an intermediary data structure and then to the correct type.
This would incur additional memory overhead and less speed but greater flexibility, true.
A solution would be to mark classes that are polymorphic (annotation/config in the builder), for whom Genson would search/produce the #class property in the stream. This would allow to have this overhead only for the polymorphic objects in the stream.
At the moment it is not implemented, but I opened an issue. It will come in a future release.
Outside of the technical aspects, I don't think you should have polymorphic logic (or any other fancy stuff) when you are dealing with multiple external API. I mean this kind of features is library specific, so if you don't use the same tool on both sides you can run into troubles. Usually people have a layer that will be used to communicate with the APIs and map the data to YOUR model. If you don't own the code on both ends, I think this would be a good solution on the long term.

Instantiate generic based on property file setting

I'm building a data driven test system. I have done this before in XML but json is giving me some interesting issues.
For each request and response type json, I have a setting in my script where I specify a pojo type. This type is instantiated to a class object thats passed to jackson to marshal the json into a usable pojo. so its like this:
"responseType": "java.util.List",
eventually gets pumped to
Class<?> reponseType = null;
try {
if (d.shouldPass) {
reponseType = Class.forName(d.responseType);
}
} catch (ClassNotFoundException e) {
throw new RequestResponseTypeInvalid(testName);
}
and I have usable class info to use in jackson. My problem is I need to do this:
"responseType": "java.util.List<foo>",
otherwise complex json types parse as hashmaps instead of pojo's. I suppose I can get creative and put something in to go from hashmap to pojo if I need to but I was wondering if there was any straight forward way to do this.
I suppose another way is to implement a factory class where I could say list_foo in the property file and have the factory class map that to an actual class object. That wouldn't be very hard but not as easy as just using the property.
thanks

You can't do this in the way that you're hoping, I'm afraid. Generics are a compile-time thing only, and can't be used in this way at runtime, because of type erasure.
The best you could do would be to have some list_foo properties, and map these explicitly to List<Foo> and so in in your code. But you can't do it by reflection.

Is there a way to create the bean class from a json response

Converting JSON to Java
The above question is with reference to what has been described on the above thread. There are so many API(s) which provide the flexibility to return responses either in XML or JSON. **I would like to know if there is a way to automatically construct the java bean corresponding to a JSON response. **

lets say you get an object like
[
{
"name":"Java 6 Greatest Hits",
"Author":"Jim Bob Jones",
"price":10.25
},
{
"name":"How to raise a goat",
"Author":"Sir Paxton",
"price":55.97
},
{
"name":"Snow - It is cold",
"Author":"Dr. White",
"price":9.99
}
]
And you want a class like
public class Book{
private String author;
private String name;
private Number price
}
with getters and setters
One option is to use a service like JSONGen, which will create that class. You need to use it first, and include the generated code in your project.
Another option could be dynamically generate the class using javassist or CGLib, but that class would be useless unless you use reflection to access its members, so even if it would be a class, it will behave like a really annoying Map. In no way will be better that simple using JSONObject

seems a simple Message Type Entity not meet you requirement ?
if you want convert a json to an existed and known java bean class,
many lib can do so, like
http://json-lib.sourceforge.net/apidocs/net/sf/json/class-use/JSONObject.html
JSONObject.toBean(JSONObject jsonObject, Class beanClass)
Creates a bean from a JSONObject, with a specific target class.
btw, if you are communicating with restful webservice, org.springframework.web.client.RestTemplate will help you get direct bean result
insteadof json.
if class does not exists, you need program with java reflect mechanism.
try use CGLIB ,http://cglib.sourceforge.net/, dynamic create some class like BeanMap. i wrote a simple sample,
but be ware, opearting class byte is hard and you may meet strange trouble with JVM . Strongly not encourage to do so.
public static BeanMap generateBean(JSONObject json) {
BeanGenerator generator = new BeanGenerator();
Iterator keys = json.keys();
while (keys.hasNext()) {
Object key = keys.next();
Object value = json.get(key);
Class keyClass = guessValueClass(value);
generator.addProperty(key.toString(), keyClass);
}
Object result = generator.create();
BeanMap bean = BeanMap.create(result);
keys = json.keys();
while (keys.hasNext()) {
Object key = keys.next();
Object value = json.get(key);
bean.put(key, value);
}
return bean;
}
/**
* TODO fix guess
*/
static Class guessValueClass(Object value) {
try {
Integer.parseInt(value.toString());
return Integer.class;
} catch (NumberFormatException e1) {
}
try {
Double.parseDouble(value.toString());
return Double.class;
} catch (NumberFormatException e1) {
}
return String.class;
}

I believe the main issue here is that the JSON response lacks type information and last time I checked :-) in Java you need to declare the type of a class property. So some heuristics will be needed to infer the type form the value in the JSON response.
For a related question here in SO have a look at: Generate Java class from JSON?

Yes check out http://flexjson.sourceforge.net

If you're wanting to generate Java classes from JSON, perhaps you could try Jackson. It provides a lot of JSON-related functionality, including the ability to generate bytecode from arbitrary JSON. See this blog post for details.

If you're using Jackson (the most popular library there), try
https://bitbucket.org/astav/jsontojava/wiki/Home
Its open source and anyone should be able to contribute.
Summary
A JsonToJava source class file generator that deduces the schema based on supplied sample json data and generates the necessary java data structures.
It encourages teams to think in Json first, before writing actual code.
Features
Can generate classes for an arbitrarily complex hierarchy (recursively)
Can read your existing Java classes and if it can deserialize into those structures, will do so
Will prompt for user input when ambiguous cases exist

JSON to Objects in java?

Does anyone know if there is the ability to generate objects for JSON data? I know there are generic JSON object libraries, but I am looking for more specific - similar to how jaxb can convert SOAP definitions or XSDs into an object model. I know there would need to be some sort of JSON definition file (which I do not know if that concept even exists within JSON), but I feel like that would be a lot more beneficial. Think:
Generic case:
genericJsonObect.get("name");
Specific case:
specificJsonObject.getName();

Jackson and XStream have the ability to map json to POJOs.

Do you want the .java source file to be generated for you? Or to map exiting java beans to JSON objects?
If the former, there is no such a library ( that I'm aware of ) if the later, Google GSON is exactly what you need.
From the samples:
class BagOfPrimitives {
public int value1 = 1;
private String value2 = "abc";
private transient int value3 = 3;
BagOfPrimitives() {
// no-args constructor
}
}
(Serialization)
BagOfPrimitives obj = new BagOfPrimitives();
Gson gson = new Gson();
String json = gson.toJson(obj);
System.out.println( json );
Prints
{"value1":1,"value2":"abc"}
( Deserialization )
BagOfPrimitives obj2 = gson.fromJson(json, BagOfPrimitives.class);
System.out.println( obj2.value1 ) ; // value1 is 1

I think the Jackson data mapper can do what you need. It can serialize/deserialize a real Java object into a Json tree.
But others API should also work :
Sojo
FlexJSON
Gson

I am not familiar of such code generation project, although I am sure many Java JSON library projects would be interested in having such thing. Main issue is that there is good Schema language for JSON that would allow code generation; JSON Schema only works for validation.
However: one possibility you could consider is to just use JAXB to generate beans, and then use Jackson to use those beans. It has support for JAXB annotations so you would be able to work with JSON and beans generated.

I have found this site very useful.
http://jsongen.byingtondesign.com/ and have used it in our projects.

Use class name as root key for JSON Jackson serialization

Suppose I have a pojo:
import org.codehaus.jackson.map.*;
public class MyPojo {
int id;
public int getId()
{ return this.id; }
public void setId(int id)
{ this.id = id; }
public static void main(String[] args) throws Exception {
MyPojo mp = new MyPojo();
mp.setId(4);
ObjectMapper mapper = new ObjectMapper();
mapper.configure(SerializationConfig.Feature.WRAP_ROOT_VALUE, true);
System.out.println(mapper.getSerializationConfig().isEnabled(SerializationConfig.Feature.WRAP_ROOT_VALUE));
System.out.println(mapper.writeValueAsString(mp));
}
}
When I serialize using the Jackson ObjectMapper, I just get
true
{"id":4}
but I want
true
{"MyPojo":{"id":4}}
I've searched all over, Jacksons documentation is really unorganized and mostly out of date.

By adding the jackson annotation #JsonTypeInfo in class level you can have the expected output. i just added no-changes in your class.
package com.test.jackson;
import org.codehaus.jackson.map.ObjectMapper;
import org.codehaus.jackson.map.SerializationConfig;
import com.fasterxml.jackson.annotation.JsonTypeInfo;
import com.fasterxml.jackson.annotation.JsonTypeInfo.As;
import com.fasterxml.jackson.annotation.JsonTypeInfo.Id;
#JsonTypeInfo(include=As.WRAPPER_OBJECT, use=Id.NAME)
public class MyPojo {
// Remain same as you have
}
output:
{
"MyPojo": {
"id": 4
}
}

I'm not using jackson, but searching I found this configuration that seems to be what you want: WRAP_ROOT_VALUE
Feature that can be enabled to make root value (usually JSON Object but can be any type) wrapped within a single property JSON object, where key as the "root name", as determined by annotation introspector (esp. for JAXB that uses #XmlRootElement.name) or fallback (non-qualified class name). Feature is mostly intended for JAXB compatibility.
Default setting is false, meaning root
value is not wrapped.
So that you can configure mapper:
objectMapper.configure(SerializationConfig.Feature.WRAP_ROOT_VALUE, true);
I hope it helps you...

Below is a way to achieve this
Map<String, MyPojo> singletonMap = Collections.singletonMap("mypojo", mp);
System.out.println(mapper.writeValueAsString(singletonMap));
Output
{ "mypojo" : { "id" : 4}}
Here the advantage is that we can give our on name for the root key of json object. By the above code, mypojo will be the root key. This approach will be most useful when we use java script template like Mustache.js for iteration of json objects

To achieve this you need to use the JsonTypeInfo annotation on your class and in particular WRAPPER_OBJECT
#JsonTypeName("foo")
#JsonTypeInfo(include = JsonTypeInfo.As.WRAPPER_OBJECT ,use = JsonTypeInfo.Id.NAME)
public class Bar(){
)

There is also a nice annotation for this:
#JsonRootName(value = "my_pojo")
public class MyPojo{
...
}
will generate:
{
"my_pojo" : {...}
}

How about simplest possible solution; just use a wrapper class like:
class Wrapper {
public MyPojo MyPojo;
}
and wrapping/unwrapping in your code?
Beyond this, it would help to know WHY you would like additional json object entry like this? I know this is done by libs that emulate json via xml api (because of impedance between xml and json, due to conversion from xml to json), but for pure json solutions it is usually not needed.
Is it to allow you do figure out what actual type is?
If so, perhaps you could consider enabled polymorphic type information, to let Jackson handle it automatically? (see 1.5 release notes, entry for PTH, for details).

there is another way i used and that worked for me.
I am working with a third party jar, so i have no control for annotations.
So i had to write through bit of hack.
Override: org.codehaus.jackson.map.ser.BeanSerializerFactory.findBeanProperties(SerializationConfig, BasicBeanDescription)
Add your property as below
List<BeanPropertyWriter> props = super.findBeanProperties(config, beanDesc);
BeanPropertyWriter bpw = null;
try {
Class cc = beanDesc.getType().getRawClass();
Method m = cc.getMethod("getClass", null);
bpw = new BeanPropertyWriter("$className", null, null, m, null,true, null);
} catch (SecurityException e) {
// TODO
} catch (NoSuchMethodException e) {
// TODO
}
props.add(bpw);
return props;
This way i get more control and can do other kind of filters too.

#JsonTypeInfo(include=As.WRAPPER_OBJECT, use=Id.NAME)
This annotation works perfectly, as suggested by Arun Prakash. I was trying to get json in this form:
{"Rowset":{"ROW":{"receiptno":"881604199388936","status":"SUCCESS"}}}
but getting like this:
{"ROW":{"receiptno":"881604199388936","status":"SUCCESS"}}
Now that annotation resolved my problem.

I would be interested in hearing the OP's solution for this. I'm having similar issues where my RESTful web service is serializing objects as either XML or JSON for clients. The Javascript clients need to know the wrapping type so that can parse it. Coupling the type to a URI pattern is not an option.
Thanks.
Edit: I noticed that Spring MappingJacksonJsonMarshaller adds the wrapping class when marshalling, so I stepped through the code in debug and noticed that Spring passes in a HashMap with a single key-value pair such that the key is the wrapping name and the value is the object. So, I extended JacksonJaxbJsonProvider, override the writeTo() method and added the following:
HashMap<String, Object> map = new HashMap<String, Object>();
map.put(value.getClass().getSimpleName(), value);
super.writeTo(map, type, genericType, annotations, mediaType, httpHeaders,entityStream);
It's a bit of a hack, but it works nicely.

use withRootName.
objectMapper.writer().withRootName(MyPojo.class.getName());

I have found through experience that it is a good idea for all JSON to include both the backend type (as a string) and the component type used to render it in the front end (if using something like angular or Vue).
The justification for doing this is so that you can process various types with a single set of code.
In vue, for example, having the name of the UI component in the data allows you, among other things, to have a screen rendering a list of children of different types using only a single tag in the parent template.
<component :is="child.componentType"/>.
For backend systems and web services - I prefer to use a single web service processor class that provides logging, auditing and exception handling for all web services by looking up the appropriate processor class based on the incoming payload. That makes the implementation of all my web services look exactly the same (about 3 lines of code), and I get detailed event logging through the lifecycle of the call without writing any per service code to do so.
Having the type wrapping the JSON makes it self documenting. If all you see are the properties, you have no idea what you are looking at until you find the corresponding end point.
If you want to write data driven software, being able to identify what you are processing is a basic requirement.