Monday, November 18, 2019

Jakarta Microprofile REST Client in Eclipse

Today we are going to implement a simple REST client for an Eclipse RCP application. Now with Jakarta @ Eclipse and all these nice Microprofile implementations this should be a piece of cake, right? Now lets see...

Source code for this tutorial is available on github as a single zip archive, as a Team Project Set or you can browse the files online.

Step 1: Dependencies

The Eclipse Microprofile REST Client repository is a good place to get started. It points to several implementations (at the bottom of the readme). Unfortunately these implementations do not host any kind of p2 sites which we could use directly. So our next stop is Eclipse Orbit, but same situation there. This means we need to collect our dependencies manually.

For my example I used RESTEasy, simply as it was the only one I could get working within reasonable time. To fetch dependencies, download the latest version of RESTEasy. As the RESTEasy download package does not contain the REST client API, we need to fetch that from another source. I found it in the Apache CXF project, so download the latest version too. If you know a better source, please let me know in the comments.

Now create a new Plug-in from Existing JAR Archives. Click on Add External... and add all jars from resteasy-jaxrs-x.y.z.Final/lib/*.jar. Further add apache-cxf-x.y.z/lib/jakarta.ws.rs-api-x.y.z.jar.
This plug-in now contains all dependencies we need for our client. Unfortunately also a lot of other stuff we probably do not need, but we leave the cleanup for later.

Step 2: Define the REST service

For our example we will build a client for the Petstore Service, which can be used for testing purposes. Further it provides a swagger interface to test the REST calls online. I recommend to check out the API and play with some commands online and with curl.

Lets write a simple client for the store with its 4 commands. The simplest seems to be the inventory command, so we will start there. Create a new Java interface:
package com.codeandme.restclient.resteasy;

import javax.ws.rs.GET;
import javax.ws.rs.Path;
import javax.ws.rs.Produces;
import javax.ws.rs.core.MediaType;
import javax.ws.rs.core.Response;

public interface IStoreService {

 @GET
 @Path("/v2/store/inventory")
 @Produces(MediaType.APPLICATION_JSON)
 Response getInventory();
}
Everything necessary for RESTEasy is provided via annotations:

  • @Path defines the path for the command of the REST service
  • @GET defines that we have to use a GET command (there exist annotations for POST, DELETE, PUT)
  • @Produces finally defines the type of data we do get in response from the server.
Step 3: Create an instance of the service

Create a new class StoreServiceFactory:
package com.codeandme.restclient.resteasy;

import java.net.URI;
import java.net.URISyntaxException;

import org.jboss.resteasy.client.jaxrs.ResteasyClient;
import org.jboss.resteasy.client.jaxrs.ResteasyWebTarget;
import org.jboss.resteasy.client.jaxrs.internal.ResteasyClientBuilderImpl;

public class StoreServiceFactory {

 public static IStoreService createStoreService() throws URISyntaxException {
  ResteasyClient client = new ResteasyClientBuilderImpl().build();
  ResteasyWebTarget target = client.target(new URI("https://petstore.swagger.io/"));
  return target.proxy(IStoreService.class);
 }
}

This is the programmatic way to create a client instance. There also exists another method called CDI, which I did not try out in Eclipse.

The service is ready and usable, so give it a try. The result object returned does contain some valuable information:

  • getStatus() provides the HTTP response status. 200 is expected for a successful getInventory()
  • getEntity() provides an InputStream which contains the JSON encoded response data from the server
Step 4: Response decoding

Our response is encoded as JSON collection of properties. In Java terms this basically reflects to a Map<String, String>. Instead of decoding the data manually, we let the framework do it for us:

Change the IStoreService to:

 Map<String, String> getInventory();
Anything else is done by the framework. Now how easy was that?

Step 5: POST request

To place an order we need order parameters. Best we encapsulate them in a dedicated Order class. From the definition of the order REST call we can see that we need following class properties: id, petId, quantity, shipDate, status, complete. Add these parameters as fields to the Order class and create getters/setters for them.

Now we can extend our IStoreService with the fileOrder() call:


@Path("/v2/store")
public interface IStoreService {

 @GET
 @Path("inventory")
 @Produces(MediaType.APPLICATION_JSON)
 Map<String, String> getInventory();

 @POST
 @Path("order")
 @Consumes(MediaType.APPLICATION_JSON)
 void fileOrder(Order order);
}

The Order automatically gets encoded as JSON object. No need for us to do the coding manually!

As parts of the path are the same for both calls, I moved the common component to the class level.

Step 6: Path parameters

To fetch an order we need to put the orderId in the request path. Coding of such parameters is put in curly braces. The parameter on the java call then gets annotated so the framework knows which parameter value to put into the path:

 @GET
 @Path("order/{orderId}")
 @Produces(MediaType.APPLICATION_JSON)
 Order getOrder(@PathParam("orderId") int orderId);

Again the framework takes care of the decoding of the JSON data.

Step 7: DELETE an Order

Deleting needs the orderId as before:

 @DELETE
 @Path("order/{orderId}")
 void deleteOrder(@PathParam("orderId") int orderId);

The REST API does not provide a useful JSON response to the delete call. One option is to leave the response type to void. In case the command fails, an exception will be thrown (eg when the orderId is not found and the server returns 404).

Another option is to set the return type to javax.ws.rs.core.Response. Now we do get everything the server sends back and no execption is thrown anymore. Sometimes we might only be interested in the status code. This can be fetched when setting the return type to Response.Status. Again, no exception will be thrown on a 404.

Optional: Only have required RESTEasy dependencies

Looking at all these jars I could not figure out a good way to get rid of the ones unused by the REST client. So I provided unit tests for all my calls and then removed dependencies step by step until I found the minimal set of required jars.



1 comment:

  1. The article is so appealing. You should read this article before choosing the Big data consulting services you want to learn.

    ReplyDelete